Formulas for IPhO 7. sin(α ± β) = sin α cos β ± cos α sin β 3. Read carefully the recommendations in the 5.
3. Read carefully the recommendations in the 5.
Version: November 11, 2017
I Mathematics
1. Taylor series (truncate for approximations):
X α+β
F (x) = F (x0 ) + F (n) (x0 )(x − x0 )n /n!
Special case — linear approximation:
F (x) ≈ F (x0 ) + F 0 (x0 )(x − x0 )
Some examples for |x|  1:
sin x ≈ x, cos x ≈ 1 − x2 /2, ex ≈ 1 + x
ln(1 + x) ≈ x, (1 + x)n ≈ 1 + nx
2. Perturbation method: find the solution itera-
tively using the solution to the ”non-perturbed”
(directly solvable) problem as the 0th approxi-
mation; corrections for the next approximation
are calculated on the basis on the previous one.
3. Solution of the linear differential equation
with constant coefficients ay 00 + by 0 + cy = 0:
y = A exp(λ1 x) + B exp(λ2 x),
where λ1,2 is the solution of the characteristic
equation aλ2 + bλ + c = 0 if λ1 6= λ2 . If the
solution of the characteristic equation is com-
plex, while a, b and c are real numbers, then Special case of the substitution method:
λ1,2 = γ ± iω and
Z ~v =
γx f (ax + b)dx = F (ax + b)/a.
y = Ce sin(ωx + ϕ0 ).
4. Complex numbers
z = a + bi = |z|eiϕ , z̄ = a − ib = |z|e−iϕ
b xn+1 = xn − f (xn )/f 0 (xn ).
|z|2 = z z̄ = a2 + b2 , ϕ = arg z = arcsin
|z|
Rez = (z + z̄)/2, Imz = (z − z̄)/2
|z1 z2 | = |z1 ||z2 |, arg z1 z2 = arg z1 + arg z2
eiϕ = cos ϕ + i sin ϕ
iϕ −iϕ iϕ −iϕ
cos ϕ = e +e 2 , sin ϕ = e −e2i
5. Cross and dot products of vectors are dis- entiate/integrate each component; alternatively
tributive: a(b + c) = ab + ac.
~ ~
~a · b = b · ~a = ax bx + ay by + . . . = ab cos ϕ
|~a × ~b| = ab sin ϕ; ~a × ~b = −~b × ~a ⊥ ~a, ~b
~a × ~b = (ay bz − by az )~ex + (az bx − bz ax )~ey + . . .
~a × [~b × ~c] = ~b(~a · ~c) − ~c(~a · ~b).
Mixed prod. (volume of parallelep. def. by 3 dimensions; b) test simple special cases (two
vec.):
(~a, ~b, ~c) ≡ (~a · [~b × ~c]) = ([~a × ~b] · ~c) = (~b, ~c, ~a). or ∞); c) verify the plausibility of solution’s
6. Cosine and sine laws:
c2 = a2 + b2 − 2ab cos ϕ
a/ sin α = b/ sin β = 2R
Non-inertial reference frames:
cos(α ± β) = cos α cos β ∓ sin α sin β
tan(α ± β) = (tan α + tan β)/(1 ∓ tan α tan β)
2α 2α
cos2 α = 1+cos 2 , sin2 α = 1−cos 2
cos(α+β)+cos(α−β)
cos α cos β = 2 , ...
cos α + cos β = 2(cos 2 + cos α−β 2 ), . . . .
8. An angle inscribed in a circle is half of the
central angle that subtends the same arc on
the circle. Conclusions: hypotenuse of a right
triangle is the diameter of its circumcircle; if
the angles of a quadrilateral are supplementary,
it is a cyclic quadrilateral.
9. Taking derivatives:
(f g)0 = f g 0 + f 0 g, f [g(x)]0 = f 0 [g(x)]g 0
(sin x)0 = cos x, (cos x)0 = − sin x
x 0
(e ) = ex , (ln x)0 = 1/x, (xn )0 = nxn−1
) eqns. for force, 1 eq. for torque. 1 (2) eq. for
(arctan x)0 = 1/(1 + x2p
0 0
(arcsin x) = −(arccos x) = 1/ 1 − x2
10. Integration: the formulas are the same as
for derivatives, but with swapped left-hand-side
and rhs. (inverse operation!), e.g.
Z force application lines coincide; for 3 points, the
xn dx = xn+1 /(n + 1).
11. Numerical methods. Newton’s iterative
method for finding roots f (x) = 0:
Trapezoidal rule for approximate integration:
Z b
b−a
f (x)dx ≈ [f (x0 ) + 2f (x1 ) + . . .
a 2n
+2f (xn−1 ) + f (xn )]
12. Derivatives and integrals of vectors: differ-
differentiate by applying the triangle rule for the
difference of two infinitesimally close vectors.
II General recommendations
1. Check all formulas for veracity: a) examine
parameters are equal, one param. tends to 0
qualitative behaviour.
2. If there is an extraordinary coincidence in
the problem text (e.g. two things are equal)
then the key to the solution might be there.
problem’s text. Pay attention to the problem’s ~v2 = ~v0 + ~v1 , ~a2 = ~a0 + ~a1 + ω 2 R~ + ~aCor
formulation — insignificant details may carry Note: ~aCor ⊥ ~v1 , ω ~ ; ~aCor = 0 if ~v1 = 0.
vital information. If you have solved for some
time unsuccessfully, then read the text again —
6∗ . Ballistic problem: reachable region
y ≤ v02 /(2g) − gx2 /2v02 .
perhaps you misunderstood the problem.
For an optimal ballistic trajectory, initial and
4. Postpone long and time-consuming math- final velocities are perpendicular.
ematical calculations to the very end (when
7. For finding fastest paths, Fermat’s and Huy-
everything else is done) while writing down all
gens’s principles can be used.
the initial equations which need to be simplified.
8. To find a vector (velocity, acceleration), it
5. If the problem seems to be hopelessly diffi-
cult, it has usually a very simple solution (and a is enough to find its direction and a projection
to a single (possibly inclined) axes.
simple answer). This is valid only for Olympiad
problems, which are definitely solvable.
6. In experiments a) sketch the experimental IV Mechanics
scheme even if you don’t have time for measure-
1. For a 2D equilibrium of a rigid body: 2
ments; b) think, how to increase the precision of
the results; c) write down (as a table) all your
force can be substituted with 1 (2) for torque.
direct measurements.
Torque is often better — “boring” forces can
be eliminated by a proper choice of origin. If
III Kinematics forces are applied only to 2 points, the (net)
1. For a point or for a translational motion of 3 lines meet at a single point.
a rigid body (integral → area under a graph):
Z Z 2. Normal force and friction force can be com-
d~
x bined into a single force, applied to the contact
, ~
x= ~v dt (x = vx dt etc.)
dt point under angle arctan µ with respect to the
Z normal force.
d~v d2 ~
x
~a = = 2 , ~v = ~adt 3. Newton’s 2nd law for transl. and rot. motion:
Z dt Z dt Z
vx F
~ = m~a, M ~ = I~ ε (M ~ = ~r × F~ ).
t= vx−1 dx = a−1
x dvx , x = dvx ~
ax For 2D geometry M and ~ ε are essentially scalars
If a = Const., then previous integrals can be and M = F l = Ft r, where l is the arm of a force.
found easily, e.g. 4. Generalized coordinates. Let the system’s
x = v0 t + at2 /2 = (v 2 − v02 )/2a. state be defined by a single parameter ξ and
time derivative ξ˙ so that the pot. energy
2. Rotational motion — analogous to the trans- its
Π = Π(ξ) and kin. en. K = µξ˙2 /2; then
lational one: ω = dϕ/dt, ε = dω/dt;
µξ¨ = −dΠ(ξ)/dξ. (Hence for transl. motion:
~a = ~τ dv/dt + ~nv 2 /R force is the derivative of pot. en.)
system consists of mass points mi :
3. Curvilinear motion — same as point 1, but 5. If the X X X
vectors are to be replaced by linear velocities, ~rc = mi~ri / mj , P~ = mi~vi
accelerations and path lengths.
L
~ =
X
mi~ri × ~vi , K =
X
mi vi2 /2
4. Motion of a rigid body. a) vA cos α = Z
vB cos β; ~vA , ~vB — velocities of pts. A and X
B; α, β — angles formed by ~vA , ~vB with line Iz = mi (x2i + yi2 ) = (x2 + y 2 )dm.
AB. b) The instantaneous center of rotation (6= 6. In a frame where the mass center’s velocity
center of curvature of material pt. trajectories!) is ~v (index c denotes quantities rel. to the mass
c
can be found as the intersection pt. of perpendic- center):
ulars to ~vA and ~vB , or (if ~vA , ~vB ⊥ AB) as the
intersection pt. of AB with the line connecting L
~ =L ~ c + MΣ R ~ c × ~vc , K = Kc + MΣ vc2 /2
endpoints of ~vA and ~vB . P
~ =P ~c + MΣ~vc
7. Steiner’s theorem is analogous (b — distance equal projections to the contact plane; f) if slid- 2. Eq. of motion
P for a system of coupled oscil- 2. Snell’s law:
of the mass center from rot. axis): I = Ic + mb2 . ing doesn’t stop, the momentum delivered from lators: ẍi = j aij xj . sin α1 / sin α2 = n2 /n1 = v1 /v2 .
8. With P~ and L ~ from pt. 5, Newton’s 2nd law: one body to the other forms angle arctan µ with 3. A system of N coupled oscillators has N
the normal of the contact plane. 3. If refraction index changes continuously,
F
~Σ = dP ~ /dt, M ~
~ Σ = dL/dt different eigenmodes when all the oscillators
17. Every motion of a rigid body can be repre- then we imaginarily divide the media into lay-
oscillate with the same frequency ωi , xj =
ers of constant n and apply Snell’s law. Light
9∗ . Additionally to pt. 5, the mom. of inertia sented as a rotation around the instantaneous xj0 sin(ωi t + ϕij ), and N eigenfrequencies ωi
ray can travel along a layer of constant n, if
rel. to the z-axis through the mass center can center of rotation C (in terms of velocities of (which can be multiple, ωi = ωj ). General solu-
P the requirement of total internal reflection is
be found as Iz0 = i,j mi mj [(xi − xj )2 + (yi − the body points). NB! Distance of a body point tion (with 2N integration constants Xi and φi )
P from C 6= to the radius of curvature of the marginally satisfied, n0 = n/r (where r is the
yj )2 ]/2MΣ . is a superposition
trajectory of P .
X of all the eigenmotions : curvature radius).
10. of inertia rel. to the origin θ = xj = Xi xj0 sin(ωi t + ϕij + φi ) 4. If refraction index depends only on z, the
2 18. Tension in a string: for a massive hang-
P Mom.
mi~ri is useful for calculating Iz of 2D bodies i
ing string, tension’s horizontal component is photon’s mom. px , py , and en. are conserved:
or bodies with central symmetry using 2θ = 4. If a system described with a generalized
Ix + Iy + Iz . constant and vertical changes according to the coordinate ξ (cf IV-2) and K = µξ˙2 /2 has an kx , ky = Const., |~k|/n = Const.
string’s mass underneath. Pressure force (per equilibrium state at ξ = 0, for small oscilla-
11. Physical pendulum with a reduced length unit length) of a string resting on a smooth 5. The thin lens equation (pay attention to
˜
l: tions Π(ξ) ≈ κξ 2 /2 [where κ = Π00 (0)] so that signs):
surface is determined by its radius of curvature ω 2 = κ/µ.
ω 2 (l) = g/(l + I/ml), and tension: N = T /R. Analogy: surface ten- 1/a + 1/b = 1/f ≡ D.
q
ω(l) = ω(˜
l − l) = g/˜
l, ˜
l = l + I/ml sion pressure p = 2σ/R; to derive, study the 5. The phase of a wave at pt. x, t is ϕ = Newton’s eq. (x1 , x2 — distances of the
6.
pressure force along the diameter. kx − ωt + ϕ0 , where k = 2π/λ is a wave vec- image and the object from the focal planes):
12. Coefficients for the momenta of inertia: tor. The value at x, t is a0 cos ϕ = <a0 eiϕ . The
x1 x2 = f 2 .
cylinder 12 , solid sphere 25 , thin spherical shell
19. Liquid surface takes equipot. shape (ne- phase velocity is vf = νλ = ω/k and group
2 1 1 1
glecting σ); in incompr. liquid, p = p0 − w, velocity v = dω/dk.
g
7. Parallax method for finding the position of
3 , rod 12 (rel. to endpoint 3 ), square 6 . w—vol. dens. of pot. en. (for a gas, see X-6). an image: find such a pos. for a pencil’s tip
6. For linear waves (electromagn. w., small- that it wouldn’t shift with resp. to the image
13. Often applicable conservation laws: 20. Bernoulli law for incompr. fluid: amplit. sound- and water w.) any pulse can be
energy (elastic bodies, no friction), 1 2 when moving perpendicularly the position of
p + ρv + ρϕ = const; considered as a superpos. of sinusoidal waves; a
your eye.
momentum (no net external force; can hold only 2 standing w. is the sum of two identical counter-
along one axis), in homog. field, the gravit. potential ϕ = gh. propagating w.: 8. Geometrical constructions for finding the
angular momentum (no net ext. torque, e.g. the For gas of specific heat cp [J/kg], paths of light rays through lenses:
ei(kx−ωt) + ei(−kx−ωt) = 2e−iωt cos kx.
arms of ext. forces are 0 (can be written rel. to 1 2 a) ray passing the lens center does not refract;
2 or 3 pts., then substitutes conservation of lin. v + cp T = const. 7. Speed b) ray k to the optical axis passes through the
2 p of sound in a gas p p
mom.). cs = (∂p/∂ρ)adiab = γp/ρ = v̄ γ/3. focus;
14. Additional forces in non-inertial frames of 21∗ . Adiabatic invariant: if the relative change 8. Speed of sound in elastic material cs = c) after refr., initially k rays meet at the focal
~ of the parameters of an oscillating system is p plane;
ref.: inertial force −m~a, centrifugal force mω 2 R E/ρ.
∗ ~ small during one period, the area of the loop d) image of a plane is a plane; these two planes
and Coriolis force 2m~v × Ω (better to avoid it; 9.
drawn on the phase plane (ie. in p-x coordi- √ Sp. of shallow (h  λ) water waves: v = meet at the plane of the lens.
being ⊥ to the velocity, it does not create any gh.
nates) is conserved with a very high accuracy.
work). 1+v /c 9. Luminous flux Φ [unit: lumen (lm)] mea-
22. For studying stability use a) principle of 10. Doppler’s effect: ν = ν0 1−ukk /css . sures the energy of light (emitted, passing a
15. Tilted coordinates: for a motion on an
inclined plane, it is often practical to align axes
minimum potential energy or b) principle of 11. Huygens’ principle: wavefront can be con- contour, etc), weighted according to the sensi-
small virtual displacement. structed step by step, placing an imaginary tivity of an eye. Luminous intensity [candela
along and ⊥ to the plane; gravit. acceleration
has then both x- and y- components. Axes may 23∗ . Virial theorem for finite movement: wave source in every point of previous wave (cd)] is the luminous flux (emitted by a source)
also be oblique (not ⊥ to each other), but then a) If F ∝ |~r|, then hKi = hΠi (time averages); front. Results are curves separated by distance per solid angle: I = Φ/Ω. Illuminance [lux (lx)]
with ~v = vx~ex + vy ~ey , vx 6= to the x-projection b) If F ∝ |~r|−2 , then 2 hKi = − hΠi. ∆x = cs ∆t, where ∆t is time step and cs is theis the luminous flux (falling onto a surface) per
of ~v . 24. Tsiolkovsky rocket equation ∆v = u ln M . unit area: E = Φ/S.
velocity in given point. Waves travel perpendic-
m
16. Collision of 2 bodies: conserved are a) net ular to wavefront. 10. Gauss theorem for luminous flux: the flux
momentum, b) net angular mom., c) angular through a closed surface surrounding
V Oscillations and waves sources of intensity Ii is Φ = 4π
P the point
Ii ; single-
mom. of one of the bodies with respect to the VI Geometrical optics.
impact point, d) total energy (for elastic colli- 1. Damped oscillator: source-case: at a distance r, E = I/r2 .
Photometry.
sions); in case of friction, kin. en. is conserved ẍ + 2γ ẋ + ω02 x = 0 (γ < ω0 ). 11. An experimental hint: if a grease stain on
only along the axis ⊥ to the friction force. Also: 1. Fermat’s principle: waves path from point a paper is as bright as the surrounding paper,
Solution of this equation is (cf. I.2.):
e) if the sliding stops during the impact, the p A to point B is such that the wave travels the then the paper is equally illuminated from both
final velocities of the contact points will have x = x0 e−γt sin(t ω02 − γ 2 − ϕ0 ). least time. sides.
 VII Wave optics 12. Malus’ law: for linearly polarized light 9. E = −dΦ/dt = −d(LI)/dt, Φ = BS. inside a ball (d = 3), cylinder (d = 2) or layer
I = I0 cos2 ϕ, where ϕ is the angle between the 10. Nonlinear elements: graphical method — (d = 1) of homogeneous ρ or ~j:
1. Diffraction — method based on Huygens’ polarization planes. find the solution in U -I coordinates as an inter-
principle: if obstacles cut the wavefront into E
~ = ρ ~r; B ~ = 1 ~j × ~r
13. λ/4-plate: phase shift π/2 between linearly section point of a nonlinear curve and a line rep- dε0 dµ0
fragment(s), the wavefront can be divided into
polarized components. resenting Ohm/Kirchoff laws. In case of many (~r — radius vector from the centre).
small pieces which serve as imaginary point-like
14. Brewster’s angle: reflected and refracted intersection points study stability — some solu- 7. Long solenoid: inside B = Inµµ0 , outside
light sources; the wave amplitude at the observ.
rays are ⊥; reflected ray is completely polarized; tions are usually unstable.
site will be the sum over the contributions of 0; flux Φ = N BS and (with n = Nl ) induc-
these sources. incidence angle tan ϕB = n. 11. Make use of short- and long-time limits. tance L = Φ/I = V n2 µµ0 . Short solenoid
15. Diffr. with optical elements: no need to For tobservation  τRC or τLR , quasiequilibrium 0Ω
2. Two slit interference (the slit width Bk = Inµµ4π (Ω — solid angle).
calculate optical path lengths through lenses, is reached: IC ≈ 0 (wire is “broken” near C)
d  a, λ): angles of maxima ϕmax = 8. Measuring magnetic field with a small
and EL ≈ 0 (L is effectively short-circuited). For R E coil
arcsin(nλ/a), n ∈ Z; I ∝ cos2 (k a2 sin ϕ), where prisms etc.: work simply with images. Particu- and ballistic galvanometer: q = R dt =
k = 2π/λ. lar conclusion: biprism gives the same diffr. as tobservation  τRC or τLR , the charge leakage
N S∆B/R.
a double slit. of C and current drop in L are small, ∆Q  Q
3. Single slit: angles of minima ϕmin = ∗ and ∆I  I: C is “short-circuited” and L is
16 . Optical fibres: Mach-Zehnder interfer-
9. Potential energy ofZ a system of charges:
arcsin(nλ/d), n ∈ Z, n 6= 0. NB! the central “broken”.
X qi qj 1
2 d ometer is analogous to a double-slit diffraction; Π=k = ϕ(~r)dq, dq = ρ(~r)dV.
maximum is double-wide. I ∝ sin (k 2 sin ϕ)/ϕ. rij 2
circular resonator — to Fabry-Pérot interferom- 12. If L 6= 0, then I(t) is a continuous function. i>j
4. Diffraction grating: the main maxima are eter; Bragg filters work similarly to the X-ray 13. Through a superconducting contour, mag-
the same as in pt. 2, the width of the main case. Single-mode fibres: ∆n/n ≈ 1 (λ/d)2 . 10. Force between parts of a uniformly charged
2 netic flux Φ = Const. In particular, with no sphere or cylindrical surface: substitute force
maxima — the same as for pt. 3 with d being
external B, LI = Const. due to charges with force due to hydrostatic
the net grating length. Spectral resolving power
λ VIII Circuits 14. Mutual inductance: magnetic flux through pressure.
∆λ = nN , where n is the order number of the
main max. and N — the number of slits. a contour Φ1 = L1 I1 + L12 I2 (I2 — current in 11. If all the charges are at the distance R (eg.
1. U = IR, P = U I a second
λ √ contour). Theorems: L12 = L21 ≡ M ; at the center of an inhomogeneously charged
5. Resolving power of a spectral device: ∆λ = X X −1
L Rseries = Ri , Rk−1 = Ri M ≤ L1 L2 . sphere or ring), ϕ = kQ/r.
λ , where L is the optical path difference be-
tween the shortest and longest beams. 2. Kirchoff’s 12. To find the net charge (or potential) in-
λ
Xlaws: X
IX Electromagnetism duced by external charges, use the superpos.
6. Resolving power of a prism: ∆λ = a dn
dλ . I = 0, U =0 pr.: “smear” the charges to make the problem
7. Angular distance when two pts. are resolved node contour 1. F = kq1 q2 /r2 , Π = kq1 q2 /r — Kepler’s symmetric.
in an ideal telescope (lens): ϕ = 1.22λ/d. For 3. To reduce the number of eqns. for pt 2: laws are applicable (Ch. XII). 13. Conductor shields charges and electric
that angle, the center of one point falls onto the method of node potentials; method of loop cur- H
2. Gauss’s law: Bd ~ S~ = 0, fields, eg. charge distribution inside a hollow
first diffr. min. of the other point. rents; equivalent circuits (any 3-terminals ⇒ ∆ I I sphere cannot be seen from outside (it seems
8. Bragg theory: a set of k ion planes of a crys- or Y ; 2-term. with emf ⇒ r and E in series). ~ S
εε0 Ed ~ = Q, ~g dS
~ = −4πGM. as if there is a conducting ball carrying a total
tal reflects X-rays if 2a sin α = kλ; a — distance 4. Resistance of infinite chain: use self- charge Q)
between neighb. planes, α — glancing angle. similarity; resistance between neighbour nodes 14. Capacitances: C = εε0 S/d (plane),
9. Reflection from optically denser dielectric of infinite grid: generalized method of electrical
3.ICirculation theorem
I I
~
Bdl 4πεε0 r(sphere), 2πεε0 l(ln R/r)−1 (coaxial).
media: phase shift π. Semi-transparent thin images. Ed
~ ~l = 0 (= Φ̇), = I, ~g d~l = 0. 15. Dipole moment:
µµ0
filmS also introduce phase shifts. 5. AC: apply pts. 1–4 while substituting R X
d~e = ~
qi~ri = ~lq, d~µ = I S.
10. Fabry-Pérot interferometer: two k semi- with Z: 4. Magnetic field caused by current element:
transp. mirrors with large reflectivity r (1−r  ZR = R, ZC = 1/iωC, ZL = iωL; ~ 16. Energy and torque of a dipole:
ν 2a dB~ = µµ0 I dl × ~er ;
1). Resolving power ∆ν ≈ λ(1−r) . Trans- ϕ = arg Z, Ueff = |Z|I 4π r2 W = −d~ · E ~
~ (B), M~ = d~ × E ~
~ (B).
mission spectrum can be found by introduc- Xeff 2
0I
P = |U ||I| cos(arg Z) = Ii Ri . hence, at the center of circular I: B = µ2r 17. Dipole field: ϕ = kd~ · ~er /r2 ; E, B ∝ r−3 .
ing 5 plane waves (for left- and rightwards-
propagating waves before the device, in the dev. 6. Characteristic ~ F
5. F~ = e(~v × B~ + E), ~
~ = I~ × Bl. 18. Forces acting on a dipole: F = (E ~ d~e )0 ,
√ times: τRC = RC, τLR =
and after the dev.) and tailoring these at the L/R, ωLC = 1/ LC. Relaxation to stationary 0
6. From the Gauss’s and circulation laws: F = (B ~ d~µ ) ; interaction between 2 dipoles:
region boundaries. σ Iµ0
current distribution exponential, ∝ e−t/τ . charged wire: E = 2πε 0r
, DC: B = 2πr ; F ∝ r−4 .
σ
11. Coherent electromagnetic waves: electric 7. Energy conservation for electric circuits: charged surface E = 2ε0 , current sheet B = 19. Point charge as a magn. dipole: dµ ∝ Φ ∝
fields are added; vector diagram can be used, ∆W + Q = U q, where q is charge which has µ0 j 2
v⊥ /B is an adiab. inv (see IV-20).
2 ;
angle between vectors p is the phase shift; NB! dis- crossed a potential drop U ; work of emf is inside a sphere (or infinite cylindrical surface) 20. Electric and magnetic images: grounded
persion: n = n(ω) = ε(ω). Energy flux den- A = Eq. of homogeneous surface charge E = 0, inside a (superconducting for magnets) planes act as mir-
sity (en. per unit area and time): I = nc ε0 E 2 . 8. WC = CU 2 /2, WL = LI 2 /2. cylindrical surface current k to the axes B = 0, rors. Field of a grounded (or isolated) sphere
can be found as a field of one (or two) fictive 2. Internal energy of one mole U = 2i RT . 2. Interference: as in wave optics. 7. Properties of an ellipse: l1 + l2 = 2a (l1 , l2
charge(s) inside the sphere. The field in a planar 3. Volume of one mole at standard cond. is 3. Uncertainty (as a math. theorem): — distances to the foci), α1 = α2 (light from one
waveguide (slit between metallic plates) can be 22,4 l. h̄ h̄ 1 focus is reflected to the other), S = πab.
obtained as a superposition of electromagnetic ∆p∆x ≥ , ∆E∆t ≥ , ∆ω∆t ≥ . 8. A circle and an ellipse with a focus at the
4. Adiabatic processes: slow as compared to 2 2 2
plane waves. For qualitative estimates by non-smooth shapes, circle’s center can touch each other only at the
sound speed, no heat exchange: pV γ = Const.
21. Ball’s (cylinder’s) polarization in homo- (and T V γ−1 = Const.). h serves better (∆p∆x ≈ h etc). longer axis.
geneous (electric) field: superpos. of homoge- 4. Spectra: hν = En − Em ; width of spectral 9∗ . Runge-Lenz vector (the ellipticity vector):
neously charged (+ρ and −ρ) balls (cylinders),
5. γ = cp /cv = (i + 2)/i.
6. Boltzmann’s distribution: lines is related to lifetime: Γτ ≈ h̄. L
~ × ~v
d ∝ E. ~
ε= + ~er = Const.
ρ = ρ0 e−µgh/RT = ρ0 e−U/kT . 5. Oscillator’s (eg. molecule) en. levels (with GM m
22. Eddy currents: power dissipation density = (n+ 12 )hν0 . For many
∼ B 2 v 2 /ρ; momentum given during a single
eigenfrequency ν0 ): EnP
Maxwell’s distribution (how many molecules
7. eigenfrequencies: E = i hni νi . XIII Theory of relativity
2
pass: F τ ∼ B 2 a3 d/ρ (where d — thickness; a
have speed v) ∝ e−mv /2kT .
— size). 6. Tunnelling effect: barrier Γ with width l 1. Lorentz transforms (rotation of 4D
easily penetrable, if Γτ ≈ h̄, where τ = space-time of Minkowski geometry), γ =
23. Inside a superconductor and for fast pro-
8. Atm. pressure: if ∆p  p,pthen ∆p = ρg∆h. is p
l/ Γ/m. p
9. p = 13 mnv̄2 = nkT , v̄ = 3kT /m,
cesses inside a conductor B = 0 and thus I = 0 1/ 1 − v 2 /c2 :
ν = vnS. 2
(current flows in surface layer — skin effect). 7. Bohr’s model: En ∝ −1/n . In a (clas- x0 = γ(x − vt), y 0 = y, t0 = γ(t − vx/c2 )
24. Charge in homog. magnetic field B~ = B~ez10. Carnot’s cycle: 2 adiabats, 2 isotherms. sically calculated) circular orbit, there is an
η = (T1 − T2 )/T1 ; derive using S-T -coordinates. integer number of wavelengths λ = h/mv. p0x = γ(px − mv), m0 = γ(m − px v/c2 )
moves along a cycloid with drift speed v =
1
E/B = F/eB; generalized mom. is conserved 11. Heat pump, inverse Carnot: η = T1T−T 2
. 8. Compton effect — if photon is scattered 2. Length of2 4-vector:
p0x = mvx − Byq, p0y = mvy + Bxq, from an electron, photon’s ∆λ = λC (1 − cos θ). s = c 2 t 2 − x2 − y 2 − z 2
12. Entropy: dS = dQ/T . 2 2
as well as gen. angular mom. L0 = L + 12 Bqr2 . m0 c2 = m2 c2 − p2x − p2y − p2z
13. I law of thermodynamics: δU = δQ + δA 9. Photoeffect: A + mv /2 = hν (A - work
25. MHD generator (a — length along the of exit for electrons). I-U -graph: photocurrent
14. II law of thermodynamics: ∆S ≥ 0 (and starts at the counter-voltage U = −(hν − A)/e, 3. Adding velocities:
~
direction of E): ηreal ≤ ηCarnot ). w = (u + v)/(1 + uv/c2 ).
saturates for large forward voltages.
E = vBa, r = ρa/bc. 4. Doppler effect:
26. Hysteresis: S-shaped curve (loop) in B-
15. GasZwork (look also p. 10) 10. Stefan-Boltzmann: P = σT 4 . p
i ν 0 = ν0 (1 − v/c)/(1 + v/c).
H-coordinates (for a coil with core also B-I- A= pdV, adiabatic: A = ∆(pV )
2 5. Minkowski space can be made Euclidean
coord.): the loop area gives the thermal energy P XII Kepler laws
dissipation density per one cycle). 16. Dalton’s law: p = pi . if time is imaginary (t → ict). Then, for rot.
1. F = GM m/r2 , Π = −GM m/r. angle ϕ, tan ϕ = v/ic. Express sin ϕ, and cos ϕ
27. Fields in matter: D ~ = εε0 E ~ = ε0 E 17. Boiling: pressure of saturated vapour
~ +P ~,
via tan ϕ, and apply the Euclidean geometry
~ pv = p0 ; at the interface betw. 2 liquids: 2. Gravitational interaction of 2 point masses
where P is dielectric polarization vector (vol-
pv1 + pv2 = p0 . (Kepler’s I law): trajectory of each of them is an formulae.
ume density of dipole moment); H ~
~ = B/µµ 0 = parabola or hyperbola, with a focus at 6. Shortening of length: l0 = l0 /γ.
~ ~ ~ 18. Heat flux P = kS∆T /l (k — thermal ellipse,
B/µ0 − J, where J is magnetization vector (vol- the center of mass of the system. Derive from 7. Lengthening of time: t0 = t0 γ.
ume density of magnetic moment). conductivity); analogy to DC circuits (P corre-
R.-L. v. (pt 9). 8. Simultaneity is relative, ∆t = −γv∆x/c2 .
sponds to I, ∆T to U , k to 1/ρ).
28. In an interface between two substances Et , R 3. Kepler’s II law (conserv. of angular mom.):
Dn (= εEt ), Ht (= Bt /µ) and Bn are continu-19. Heat capacity: Q = c(T )dT . Solids: for for a point mass in a central force field, radius 9. F~ = d~p/dt [= dtd (m~v), where m = m0 γ].
ous. low temperatures, c ∝ T 3 ; for high T , c = 3N k, vector covers equal areas in equal times. 10. Ultrarelativistic
p approximation:
p v ≈ c,
where N — number of ions in crystal lattice.
29. Energy density: W = 12 (εε0 E 2 + B 2 /µµ0 ). p ≈ mc, 1 − v 2 /c2 ≈ 2(1 − v/c).
4. Kepler’s III law: for two point masses at
20. Surface tension:
30. For µ  1, fieldlines of B are attracted to elliptic orbits in r−2 -force field, revolution peri- 11∗ . Lorentz tr. for E-B: B ~0 = B ~0 = E
~ || , E ~ || ,
|| ||
the ferromagnetic (acts as a potential hole, cf. U = Sσ, F = lσ, p = 2σ/R. ods relate as the longer semiaxes to the power
3 0 0
~
pt. 28). 21. Stefan-Boltzmann law (gray body): P = of 2 : E~⊥ = γ(E ~ ⊥ +~v × B~ ⊥ ), B
~⊥ = γ(B ~ ⊥ −~v × E⊥ ).
c2
31. Current density ~j = ne~v = σE~ = E/ρ. ~ εσT 4 . T12 /T22 = a31 /a32 .
32. Lenz’s law: system responds so as to op- 22. Wien’s law: fmax = AkB T /h (A ≈ 2.8), 5. Full energy (K + Π) of a body in a gravity
pose to changes. λmax = hc/A0 kB T (A0 ≈ 5) field:
∗
E = −GM m/2a. marks an advanced material.
6. For small ellipticities ε = d/a  1, trajec- Corrections/suggestions ⇒ kalda@ioc.ee.
X Thermodynamics XI Quantum mechanics
tories can be considered as having a circular Composed by J. Kalda, translated by U. Visk
1. pV = mµ RT 1. p~ = h̄~k (|~p| = h/λ), E = h̄ω = hν. shapes, with shifted foci. and J.K.
 1. INTRODUCTION
PROBLEMS ON KINEMATICS of reference, some velocity or its component (or acceleration
Jaan Kalda or its component) vanishes or two velocities are equal. Once
a suitable frame of reference has been found, we may change
Translation partially by Taavi Pungas back into the laboratory frame and transform the now known
Version: 29th November 2017 velocities-accelerations using the rule of adding velocities (ac-
celerations). NB! the accelerations can be added in the same
1 INTRODUCTION way as velocities only if the frame’s motion is translational (i.e.
it does not rotate).
For a majority of physics problems, solving can be reduced to
using a relatively small number of ideas (this also applies to pr 1. On a river coast, there is a port; when a barge passed
other disciplines, e.g. mathematics). In order to become good the port, a motor boat departed from the port to a village at
at problem solving, one must learn these ideas. However, it is the distance s1 = 15 km downstream. It reached its destination
not enough if you only know the ideas: you also need to learn after t = 45 min, turned around, and started immediately mov-
how to recognize which ideas are to be used for a given prob- ing back towards the starting point. At the distance s2 = 9 km
lem With experience it becomes clear that usually problems from the village, it met the barge. What is the speed of the
actually contain hints about which ideas need to be used. river water, and what is the speed of the boat with respect to
This text attempts to summarise the main ideas en- the water? Note that the barge did not move with respect to
countered in solving kinematics problems (though, some of the water.
these ideas are more universal, and can be applied to some
Here, the motion takes place relative to the water, which
problems of other fields of physics). For each idea, there are
gives us a hint: let us try solving the problem when using the
one or several illustrative problems. First you should try to
water frame of reference. If we look at things closer, it be-
solve the problems while keeping in mind those ideas which
comes clear that this is, indeed, a good choice: in that frame,
are suggested for the given problem. If this turns out to be
the speed of the boat is constant, and barge is at rest, i.e. the
too difficult, you can look at the hints — for each problem,
motion of the bodies is much simpler than in the coastal frame
rather detailed hints are given in the respective section. It is
of reference.
intentional that there are no full solutions: just reading the
solutions and agreeing to what is written is not the best way pr 2. Two planes fly at the same height with speeds v1 =
of polishing your problem solving skills. However, there is a 800 km/h and v2 = 600 km/h, respectively. The planes ap-
section of answers — you can check if your results are correct. proach each other; at a certain moment of time, the plane tra-
There are also revision problems for which there are no sugges- jectories are perpendicular to each other and both planes are at
tions provided in the text: it is your task to figure out which the distance a = 20 km from the intersection points of their tra-
ideas can be used (there are still hints). jectories. Find the minimal distance between the planes during
Problems are classified as being simple , normal , and their flight assuming their velocities will remain constant.
difficult (the problem numbers are coloured according to this
colour code). Please keep in mind that difficulty levels are relat-
ive and individual categories: some problem marked as difficult
may be simple for you, and vice versa. As a rule of thumb, a
problem has been classified as a simple one if it makes use of
only one idea (unless it is a really tricky idea), and a difficult
one if the solution involves three or more ideas.
It is assumed that the reader is familiar with the concepts
of speed, velocity and acceleration, radian as the measure for
angles, angular speed and angular acceleration, trigonometric
functions and quadratic equations. In few places, derivatives
and differentials are used, so a basic understanding of these con-
cepts is also advisable (however, one can skip the appropriate The idea 1 advises us that we should look for a frame where
sections during the first reading). some bodies are at rest; that would be the frame of one of the
planes. However, here we have a two-dimensional motion, so
the velocities need to be added and subtracted vectorially.
2 VELOCITIES
def. 1: A scalar quantity is a quantity which can be fully
idea 1: Choose the most appropriate frame of reference. You
described by a single numerical value only; a vector quantity is
can choose several ones, and switch between them as needed.
a quantity which needs to be described by a magnitude (also
Potentially useful frames are where:
referred to as modulus or length), and a direction. The sum
⋆ some bodies are at rest;
of two vectors ⃗a and ⃗b is defined so that if the vectors are in-
⋆ some projections of velocities vanish;
terpreted as displacements (the modulus of a vector gives the
⋆ motion is symmetric.
distance, and its direction — the direction of the displacement)
It is recommended to investigate process in all potentially use- then the vector ⃗a + ⃗b corresponds to the net displacement as
ful frames of reference. As mentioned above, in a good frame a result of two sequentially performed displacements ⃗a and ⃗b.
— page 1 —
 2. VELOCITIES
This corresponds to the triangle rule of addition, see figure.
Subtraction is defined as the reverse operation of addition: if v
⃗a + ⃗b = ⃗c then ⃗a = ⃗c − ⃗b.
a
~ + b~
a

~
a
b~
The idea 4 can be used again in the following problem:
b~ ~ − b~
a
~
a pr 4. Balloons with constant ascending velocity can be used
to investigate wind velocities at various heights. The given
After having been introduced the concept of vectors, we can graph of elevation angle against time was obtained by observing
also fix our terminology. a such balloon. The balloon was released at distance L = 1 km
from the point of observation and it seemed to be rising dir-
def. 2: Velocity is a vectorial quantity which can be defined ectly upwards. Knowing that wind velocity near the ground
by the projections to the axes √ ⃗v = (vx , vy , vz ); speed is the was zero, find the balloon’s height at time t = 7 min after its
modulus of a vector, v = |⃗v | = vx2 + vy2 + vz2 . Similarly, dis- start and wind velocity at this height.
placement is a vector pointing from the starting point of a body α/deg
60
to its final position; travelled distance is the sum of the moduli
of all the elementary displacements (the curve length).
40
For vectorial addition, there are two options. First, we can
select two axes, for instance x and y, and work with the respect-
ive projections of the velocity vectors. So, if our frame moves 20
with the velocity ⃗u and the velocity of a body in that frame is
⃗v then its velocity in the lab frame is w ⃗ = ⃗u + ⃗v , which can be
found via projections wx = vx + ux and wy = vy + uy . Altern- 0 2 4 6 8 t /min
atively, we can approach geometrically and apply the triangle
In order to answer to the first question here, we need also the
rule of addition, see above.
following idea.
Once we have chosen the reference frame of one of the
planes, the problem 2 can be solved by using the following idea 5: If a graph of y versus x is given, quite often some tan-
dy
idea. gent line and its slope dx turn out to be useful. In such cases,
unless it is obvious, you have to show that the derivative dy is
idea 2: For problems involving addition of vectors (velocities, related to a physical quantity z relevant to the solution ofdxthe
forces), the problems can be often reduced to the application of
problem. To this end, you need to express z in terms of small
simple geometrical facts, such as (a) the shortest path from a
(infinitesimal) increments dx and dy, and manipulate mathem-
point to a line (or plane) is perpendicular to the line (plane); (b)
atically until these increments enter the expression only via the
among such triangles ABC which have two fixed side lengths dy
ratio dx .
|BC| = a and |AC| = b < a, the triangle of largest ∠ABC has
∠BAC = 90◦ . To be mathematically correct, there are two options. First, in
simpler cases, you can say that the increments are infinitely
The next problem requires the application of several ideas
small (infinitesimal), and denote these via differentials dx and
and because of that, it is classified as a difficult problem. When
dy. In more complex cases it may be more convenient to start
switching between reference frames, the following ideas will be
with small but finite increments ∆x and ∆y, make your calcula-
useful.
tions while keeping only the leading terms (e.g. for ∆x+∆x·∆y,
idea 3: Try to reveal hidden symmetries, and make the prob- the second term is a product of two small quantities and can
lem into a symmetric one. be neglected as compared with the first one), and finally go to
the limit of infinitely small increments, ∆x → dx, ∆y → dy.
In order to answer the second question, we need one more
idea 4: It is possible to figure out everything about a velocity
idea.
or acceleration once we know one of its components and the
direction of the vector. idea 6: There are calculations which cannot be done in a
generic case, but are relatively easy for certain special values
Mathematicians’ way of stating it is that a right-angled triangle of the parameters. If some unusual coincidence stands out in
is determined by one angle and one of its sides. For example, the problem (in this case the slope of the tangent is zero at the
if we know that velocity is at angle α to the horizontal and its given time) then it is highly probable that this circumstance
horizontal component is w then its modulus is w/ sin α. has to be used.
pr 3. One of two rings with radius r is at rest and the other idea 7: If friction affects the motion then usually the most ap-
moves at velocity v towards the first one. Find how the velocity
propriate frame of reference is that of the environment causing
of the upper point of intersection depends on a, the distance
the friction.
between two rings’ centres.
— page 2 —
 2. VELOCITIES
pr 5. A white piece of chalk is thrown onto a black hori- It would be a very difficult task (requiring a numerical integ-
zontal board moving at constant velocity. Initially, the chalk’s ration of a differential equation) to estimate the trajectory of
velocity was perpendicular to the board’s direction of motion. the ball subject to a turbulent air drag. However, this is not
What is the shape of the chalk’s trace on the board? what you need to do, because the air drag is not described by a
formula for the drag force, but instead, by the final departure
To solve the next problem, in addition to the previous idea we from the corresponding free-fall-trajectory.
also need to use 2, which can be rephrased in a slightly more So, with the help of idea 9 we conclude that the air drag
general (but less specific) way: some minima and maxima can cannot be neglected here. Once we have understood that, it
be found without taking any derivatives, in fact the solution becomes evident that we need to apply the idea 7. However,
without a derivative can turn out to be much simpler. For this even when equipped with this knowledge, you might run into
problem, an even more narrowed down formulation would be mathematical difficulties as there is no direct way of express-
the following. ing the flight time t in terms of the given quantities. Instead,
idea 8: If one of two vectors is constant and the direction of you are advised to write down an equation containing t as an
the other is fixed then the modulus of their sum is minimal if unknown, and then to solve it.
they form a right-angled triangle. idea 10: It is often useful first to write down an equation (or
a system of equations) containing the required quantity as an
pr 6. A block is pushed onto a conveyor belt. The belt unknown, instead of trying to express it directly (sometimes
is moving at velocity v0 = 1 m/s, the block’s initial velocity it is necessary to include additional unknowns that later get
u0 = 2 m/s is perpendicular to the belt’s velocity. During its eliminated).
subsequent motion, what is the minimum velocity of the block
Furthermore, unlike the problems we had thus far, this problem
with respect to the ground? The coefficient of friction is large
deals with a 3-dimensional geometry, which makes it difficult
enough to prevent the block from falling off the belt.
to draw sketches on a 2-dimensional sheet of paper. Thus we
The next problem is slightly unusual, specific comments will need one more simple idea.
be given after the problem. To tackle such situation one can
idea 11: It is difficult to analyse three-dimensional motion
give seemingly trivial but very often an overlooked advice.
as a whole, so whenever possible, it should be reduced to two
idea 9: Read carefully the problem text, try to understand dimensions (projecting on a plane, looking at planes of inter-
the meaning of every statement, don’t make hasty assumptions section).
by yourself.
The next problem illustrates
For a well-written problem, there are no redundant sentences.
idea 12: An elastic collision is analysed most conveniently in
Things become more troublesome if that is not the case. Some-
the centre of mass frame of the process.
times the problem author wants to educate you more than just
by giving you the very problem, and tells you many things Let us derive from this idea a ready-to-use recipe when a ball
(such as historical background) which are definitely interesting collides with a moving wall. First, since the wall is heavy, the
but unrelated to the solution of the problem. It is OK if you system’s centre of mass coincides with that of the wall, hence
are solving the problem as an exercise at home and you have we’ll use the wall’s frame. In the frame of the centre of mass, if
plenty of time. However, you need to develop skills of parsing the collision is elastic and there is no friction then due to the en-
fast through such paragraphs at competitions under time pres- ergy and momentum conservation, the bodies will depart with
sure: you need to make sure that there are really no important the same speed as they approached, i.e. the normal component
hints hidden inside. of the ball’s velocity is reversed. If we apply the addition of
velocities twice (when we move to the wall’s frame, and when
pr 7. After being kicked by a footballer, a ball started to we switch back to the lab frame), we arrive at the following
fly straight towards the goal at velocity v = 25 m/s making conclusion.
an angle α = arccos 0.8 with the horizontal. Due to side wind
blowing at u = 10 m/s perpendicular the initial velocity of the idea 13: For an elastic bouncing of a ball from a wall which
ball, the ball had deviated from its initial course by s = 2 m by moves with a velocity ⃗u in the direction of the surface normal,
the time it reached the plane of the goal. Find the time that the normal component ⃗vn of the ball’s velocity ⃗v is increased
′
it took the ball to reach the plane of the goal, if the goal was by 2⃗u, i.e. ⃗vn = −⃗vn + 2⃗u.
situated at distance L = 32 m from the footballer. For this problem we must also remember
A typical problem gives all the parameter values describing a fact 1: Angle between velocity vectors depends on the frame
system and then asks about its behaviour. Here, the system of reference!
might seem to be over-described: why do we need the value of s,
couldn’t we just use the initial velocity to determine the flight pr 8. A tennis ball falls at velocity v onto a heavy racket
L
time to deduce t = v cos α ? Such a question might arise, first and bounces back elastically. What does the racket’s velocity
of all, because you are used to ignoring air friction. However, u have to be to make the ball bounce back at a right angle to
no-one mentioned that you can neglect it here! Furthermore, it its initial trajectory and not start spinning if it did not spin
is even evident that the air drag cannot be neglected, because before the bounce? What is the angle β between ⃗u and the
otherwise the ball would not depart from its free-fall trajectory! normal of the racket’s plane, if the corresponding angle for ⃗v is
— page 3 —
 3. ACCELERATIONS, DISPLACEMENTS
α? axis, the direction being given by the screw rule (if the screw is
rotated in the same way as the body, the vector points in the
If we keep in mind the idea 9 and read the text carefully, we
direction of the screw movement).
notice that the racket is heavy so that we can use the idea 13.
Also, pay attention that the ball will not rotate after the colli- idea 17: When switching between rotating frames of refer-
sion — this is important for finding the parallel (to the racket’s ence, angular velocities are to be added in the same as transla-
plane) component of the velocity. tional velocities in the case of translationally moving frames of
Earlier we mentioned that vectors can be dealt with either reference. NB! This remains valid even if the angular velocit-
geometrically (e.g. by applying the triangle rule for a sum of ies are not parallel (although non-small rotation angles can be
vectors and solving a trigonometrical problem), or algebraically added only as long as the rotation axis remains unchanged).
using projections. Quite often, geometrical approach provides
shorter solutions, but not always; this observation leads us to This idea is illustrated by a relatively simple problem below.
the following recommendation. pr 9. Vertical mirror with two reflecting surfaces (front and
idea 14: For vectorial calculations, prefer geometrical ap- back) rotates around a vertical axis as shown in figure, with
proach, but if it seems unreasonable (e.g. some of the conditions angular speed ω. There is an unmoving point source of light S
are formulated through the projections of the vectors) switch at a distance a from the rotation axis. Find the speed of the
to the algebraic approach and write expressions down in terms image of the point source as a function of time.
of components. ω
For the algebraic approach, optimal choice of axes is very a
important. “Optimal” means that the conditions are written
S
in the simplest possible way. Sometimes it may happen that
the most useful coordinate axes are not even at right angles.
For the problem 8, geometrical solution turns out to be sim-
3 ACCELERATIONS, DISPLACEMENTS
pler, but more difficult to come up with. This is quite typical:
algebraic approach leads to a brute-force-solution when it is Thus far we dealt with instantaneous or constant velocities, and
clear from the beginning what you need to do, but the calcula- in few cases we applied a simple formula s = vt for displace-
tions are mathematically long. Still, there are no fundamental ments. In general, when the velocity ⃗v is not constant, the dis-
difficulties and you just need to execute it. As long as the placement is found as the curve under the graph of the velocity
mathematical part will not be unreasonably long or leading to as a function of time. For instance, the displacement along
fundamental difficulties (such as unsolvable equations), brute x-coordinate ∆x is surface area under the graph vx = vx (t);
∫
force approach is still OK: figuring out an elegant solution can mathematically we can write it via integral ∆x = vx (t)dt.
also take some time. You don’t need to know more about integrals right now, just
Typically, the geometrical solutions of physics problems rep- that it represents surface areas under graphs.
resent very simple geometrical tasks and hence, finding these idea 18: Calculation of many physical quantities can be re-
shorter-than-algebraic solutions is also quite easy. In this case, duced (sometimes not in an obvious way) to the calculation of
however, the geometrical task turns out to be quite a tricky surface areas under a graph (i.e. to an integral). In particular:
problem. While the idea 14 suggests that the algebraic ap- distance is the area under a v − t curve (velocity-time), velocity
proach is good for problem 8 (the no-rotation-requirement gives the area below an a − t curve etc.
us a condition for the parallel component of the velocity), it is
recommended that you try both methods here. In both cases Note that drawing a graph is not always absolutely necessary (if
you need one more mathematical idea. you are skilled with integrals, formulae can be derived analytic-
ally, without drawing graphs), but doing it helps to imagine the
idea 15: Two vectors ⃗a = (ax , ay , az ) and ⃗b = (bx , by , bz ) process. Visualisation of this kind is always beneficial, it sim-
are perpendicular if their scalar product is zero, ax bx + ay by + plifies finding the solution and reduces the chances of making
az bz = 0. (This assumes that the axes x, y and z are perpen- mistakes.
dicular to each other.)
pr 10. A particle starts from the origin of coordinates; the
idea 16: For trigonometric problems involving right triangles figure shows its velocity as a function of time. What is its
keep in mind that the circumcentre of a right triangle is at the maximum shift from the origin?
centre of the hypotenuse, hence the median drawn from the v (m/s)
right angle divides the triangle into two isosceles triangles, and 6
the right angle into the angles equal to the acute angles of the 4
triangle. 2
t (s)
Idea 1 told us to make use of switching between different 0
frames of reference. This idea can be also used when dealing 5 10 15
-2
with rotational motion.
def. 3: Angular velocity ω
⃗ equals by modulus to the rotation The next problem is much more difficult, although it is also
angle (in radians) per unit time, and is parallel to the rotation reduced to finding a surface area; due to difficulty, the full
— page 4 —
 3. ACCELERATIONS, DISPLACEMENTS
solution (apart from replotting the graph and numerical calcu- In particular, if a problem involves two or more free-falling
lations) is given under “hints”. bodies then using a free-falling frame simplifies calculations
significantly.
pr 11. The acceleration of a boat depends on its speed as
shown in graph. The boat is given initial speed v0 = 4 m/s. It should be emphasized that if the frame rotates, the formula
What is the total distance travelled until the boat will almost for acceleration obtains additional terms.
come to rest? pr 12. Two smooth slides lie within the same vertical plane
a (m/s2) and make angles α to the horizontal (see the figure). At some
moment, two small balls are released from points A and B and
they start sliding down. It took time t1 for the first ball that
0.6
started from point A to reach the ground; for the second one
the time of descent was t2 . At what time was distance between
the balls the smallest?
0.4 A
B
0.2
α α
1 2 3 v (m/s)
idea 20: Sometimes, it is possible to separate two- or three-
The idea 18 can be also used to derive basic formulae for dimensional motion of a body into independent motions in per-
displacement in the case of a motion with constant accelera- pendicular directions: (a) motion along x is independent from
tion. Suppose that a body has initial speed v0 and moves with the motion along y for 2D geometry; (b) motion along x is in-
constant acceleration a; we want to know to which distance it dependent from the motion along y, which is independent from
travels by the moment of time t. The surface area under the the motion along z; (c) motion along x is independent from the
graph is a right trapezoid (see figure), with surface area equal motion in y − z-plane. In particular, this can be done for fric-
to the product of the median v0 + 12 at, and the height t, i.e. tionless collisions from a plane2 : if the axis x lies in the plane,
s = v0 t + at2 /2. and y is perpendicular to it, you can study separately motion
v0+at v along x and motion along y.
The simplest application of this idea is provided by a two-
v0 dimensional motion of a body in an homogeneous gravity field,
t which is studied in every textbook on kinematics: horizontal
t and vertical motions are decoupled, because vertical acceler-
Alternatively, if we are given the initial and final velocities (v0 ation g does not depend on the horizontal coordinate x and
and v1 ) instead of the travel time, the median of the trapezoid horizontal velocity vx , and body moves with a constant speed,
is expressed as 21 (v0 + v1 ), and the height as t = (v1 − v0 )/a. and horizontal acceleration (0) does not depend on the vertical
This leads us to s = (v0 + v1 )(v1 − v0 )/2a = (v12 − v02 )/2a. If coordinate x and vertical velocity vy . As a result, we obtain
we rewrite it as as = 21 (v12 − v02 ), we can call it the energy x = v0x t and y = v0y t − gt2 /2 (where v0x and v0y are the re-
conservation law for unit mass if the free fall acceleration is a. spective initial velocity components); the respective trajectory
in x − y-plane is a parabola which we obtain if we eliminate t
fact 2: If a body moves with initial speed v0 , final speed v1 from the second equation by substituting t = x/v0x .
and constant acceleration a during time t, the distance trav-
elled fact 3: Free fall trajectory of the centre of mass a body in
1 v − v0
2 2
homogeneous gravity field g is a parabola, parametrically given
s = v0 t + at2 = 1 .
2 2a as x = v0x t and y = v0y t − gt2 /2.
The next problem can be solved in various ways, but the Let us discuss in more details how to apply the idea 20 to
simplest solution involves the following idea. frictionless interactions (collisions or sliding) of a body with a
idea 19: Sometimes it is useful to change into a non-inertial plane. If the plane is inclined, we need to take the axes to be
frame of reference: velocities are added just in the usual way, inclined as well; then, the gravitational acceleration will have
⃗vlab = ⃗vrel +⃗vfr , where ⃗vlab is the velocity in the lab frame, ⃗vrel — a non-zero component along both axes, i.e. motion will have
velocity in the moving frame, and ⃗vfr — the speed of that point acceleration in both directions.
of the moving frame where the body is at the given moment. fact 4: Free fall problems can be also analysed when using
If the frame moves translationally (without rotations) then the inclined system of axis (this might be useful because of idea
accelerations can be added in the same way, ⃗alab = ⃗arel + ⃗afr 1 . 20); then, the free fall acceleration is decomposed into two re-
1 This can be easily seen if we relate the respective radius vectors as ⃗
rlab = ⃗
rrel + ⃗
rfr and take twice the time derivative: the derivative of a sum is
the sum of derivatives, even if we deal with vectors (this can be understood if we work with projections, e.g. the addition rule for the x-components
of the accelerations can be obtained by taking twice the time derivative of the equality relating the x-coordinates, xlab = xrel + xfr ).
2 As long as there is no other mechanism (such as the Lorenz force) which couples the motions in different directions

— page 5 —
 3. ACCELERATIONS, DISPLACEMENTS
spective perpendicular components, ⃗g = ⃗gx +⃗gy with gx = sin α of the two surfaces and y-axis lays on the inclined surface, x, y,
and gy = cos α, α being the angle between the surface and the and z-motions are all separated; at the impact, vz goes to zero,
horizon. and due to the absence of friction, vx and vy are preserved.
During a collision, as there is no friction force, vx (parallel to In this problem, however, the transition from one surface
the surface velocity component) does not change, i.e. it does to the other is smooth: around the line separating the two
not “notice” that there was a collision: in order to analyse the flat surfaces, there is a narrow region where the surface has a
evolution of the x-coordinate, we can completely forget about curvature. Within this narrow region, the motion in y- and
the changes of the y-coordinate (and vice versa). z-directions cannot be separated from each other, and we need
one more idea.
If the surface is curved, generally such a separation is no
longer possible. Indeed, previously x was independent of y be- idea 21: If a force is perpendicular to the direction of motion
cause the dependence of the acceleration on the coordinates is (normal force when sliding along a curved surface, tension in a
introduced by the normal force, which is a function of y only, rope when a moving body is attached to an unstretchable rope
and has no x-component. If the surface is curved, it is im- fixed at the other end, force on a charge in magnetic field) then
possible to have the x-axis to be everywhere parallel to the the velocity vector can only turn, its modulus will not change.4
surface: the acceleration due to the normal force has both non-
vanishing x and y components, and depends both on x and y pr 15. Three turtles are initially situated in the corners of
coordinates. However, in the case of side surfaces of cylinders, an equilateral triangle at distances 1 m from one another. They
prisms and other generalized cylinders3 , it is still possible to move at constant velocity 10 cm/s in such a way that the first
find one axis x which is everywhere parallel to the surface and always heading towards the second, the second towards the
hence, motion along x can be separated from the motion in third and the third towards the first. After what time will they
y − z-plane. meet?

pr 13. An elastic ball is released above an inclined plane Two approaches are possible here: first, we may go into the
(inclination angle α) at distance d from the plane. What is frame of reference rotating with the turtles, in which case we
the distance between the first bouncing point and the second? apply the following idea.
Collisions occur without friction. idea 22: Sometimes even a reference frame undergoing very
complex motion can be useful.
The next problem makes also use of the idea 20; however, one
more idea is needed, see below. Alternatively, we can use
idea 23: Instead of calculating physical velocities, it is some-
pr 14. A puck slides onto an icy inclined plane with inclin-
times wise to look at the rate of change of some distance, the
ation angle α. The angle between the plane’s edge and the
ratio of two lengths, etc.
puck’s initial velocity v0 = 10 m/s is β = 60◦ . The trace left
by the puck on the plane is given in the figure (this is only The following problem requires integration5 , so it can be
a part of the trajectory). Find α under the assumption that skipped by those who are not familiar with it.
friction can be neglected and that transition onto the slope was pr 16. An ant is moving along a rubber band at velocity
smooth. v = 1 cm/s. One end of the rubber band (the one from which
the ant started) is fixed to a wall, the other (initially at dis-
tance L = 1 m from the wall) is pulled at u = 1 m/s. Will the
ant reach the other end of the band? If yes then when will it
β happen?
α
Here we need to apply the
2.5m

idea 24: For some problems, optimal choice of parametriza-

tion can simplify mathematical calculations significantly. An
incomplete list of options: Cartesian, polar, cylindrical, and
2.5m spherical coordinates; travel distance; Lagrangian coordinates
(i.e. for fluids flow using the initial coordinate of a fluid particle
The last sentence here is very important: if the transition is instead of its current coordinate); relative position of a particle
sharp, the puck approaches the inclined plane by sliding along according to a certain ranking scheme, etc.
the horizontal and collides with it — either elastically in which Here, the problem itself contains a hint about which type of
case it jumps up, or plastically. In particular, if the collision parametrization is to be used. It is clear that the Cartesian
is perfectly plastic then that part of the kinetic energy which coordinate of the ant is not good: it does not reflect the pro-
is associated with the motion along the surface normal of the gress of the ant in advancing along the rubber band. In order
inclined plane is lost. More specifically, if we introduce perpen- to describe such a progress, we can use the relative position on
dicular coordinates so that the x-axis is along the contact line the band: which fraction k of the rubber is left behind; the ant
3 Surfaceswith constant cross-sections.
4 Thisis the energy conservation law ∫using the fact that forces perpendicular to the velocity will not perform work.
5 You may find helpful to know that dx
ax+b
= a−1 ln(ax + b) + C

— page 6 —
 4. OPTIMAL TRAJECTORIES
starts with k = 0, and k = 1 corresponds to the ant reaching light beam path. Finally, in the case of reflections, we need to
the end of the band. The parameter k is essentially a Lag- compare only those paths which include similar reflections. So,
rangian coordinate: it equals to the initial coordinate of the the path s5 is faster than the path s4 , but the former does not
current rubber point in the units of the initial rubber length. involve reflections and cannot be included into the set of ref-
erence paths. Among those paths which include one reflection
4 OPTIMAL TRAJECTORIES from the mirror and represent a small variation of the path s4
(such as the path s6 ), the path s4 is the fastest one and hence,
Majority of the kinematical optimal trajectory problems fall represents a valid light beam path.
into two categories: the problems of finding the trajectories of
If light can travel from one point to the other along several
shortest travel time, and the problems of finding the smallest
different paths (e.g. from some point through a lens to the
initial speeds of a free-falling body.
optical image of that point) then time along all these paths is
idea 25: In those kinematics problems where velocities in vari- exactly the same.
ous environments are given and the quickest way from point A In order to apply the idea 25 to kinematics problems, we
to point B is asked, Fermat’s principle (formulated for geomet- often need the Snell’s law.
rical optics) can be of help.
fact 5: Let a point A be situated in a medium where the light
Namely, if we have a configuration of bodies with different in- propagation speed is v1 , and point B — in a medium where the
dices of refraction, and if a ray of light originating from point speed is v2 . Then, the light propagates from A to B according
A passes through point B then the actual path of the ray is to the Snell’s law: it refracts at the interface so that the angle
the quickest way for light to reach point B from point A (as a between the surface normal and the path forms angles α1 and
reminder, if the index of refraction of some environment is n α2 (see figure) satisfying equality sin α1 / sin α2 = v1 /v2 .
then the travelling speed of light is c/n). Therefore time along
path s1 or s2 is longer than along s0 , see figure. B
α2

A s2 s0
s1
n1 n2 B
We must clarify that the Fermat’ principle applies to a local
minimum: the travel time along the path s0 needs to be shorter
than for any other path which departs from the path s0 but
remains in its immediate neighbourhood. Furthermore, it is re-
quired that for small path variations, the travel time variations
remain also small. The following figure clarifies in which cases
the Fermat’ principle is applicable.
mirror
s6 s5
s4
glass
A plate s2 B l
s0
s1
s3
As the propagation speed in the glass plate is smaller than in
the air, the global minimum of the travel time is achieved for
the path s1 . However, such arbitrarily small variations of the
path s1 which go thorough the glass plate have a non-small
change in the travel time, hence s1 is not a valid light beam
path (the path can be deformed downwards, e.g. into s3 , but up-
wards deformations incur a jump in travel time due to passing
through the glass plate). Next, the path s0 provides a good
local minimum: the travel time along s0 is smaller than along
any small variation of the path s0 , and if the path variation is
small, the time variation remains also small. Hence, the Fer-
mat’ principle can be applied: the path s0 provides a valid
— page 7 —
A
α1
pr 17. A boy lives on the shore OP of a bay M OP (see the
figure). Two shores of the bay make an angle α. The boy’s
house is situated at point A at distance h from the shore and
√
h2 + l2 from point O. The boy wants to go fishing to the
shore OM . At what distance x from point O should be the
fishing spot, so that it would take as little time as possible to
get there from the house? How long is this time? The boy
moves at velocity v on the ground and at velocity u < v when
using a boat.
M P
h
A
α
O
Here we can use a small addition to the last idea: if the quickest
way to a plane (in a 3-d problem) or to a line (in 2-d) is asked
then this plane or line can be substituted with a point very far
(at infinity) in the perpendicular direction to it. The reason for
that is quite simple: it takes the same amount of time to reach
any point on the plane (line) from that very-very distant point.
If we think about this in terms of geometrical optics then it
means that a set of light rays normal to the surface falls onto
the plane (line).
pr 18. A boy is situated at point A in a river, at distance
a from the riverbank. He can swim at speed u or run at speed
 4. OPTIMAL TRAJECTORIES
v > u on the shore; water flows in the river at velocity w > u. fronts, the rays of light can be found as such curves which are
The boy wants to reach the point C upstream on the riverbank everywhere perpendicular to the wavefronts.
with minimal time. At what distance x from point B aligned As a simple demonstration of how the Huygens principle
with point A should he get out of the water? can be applied for calculations, let us express the angle of the
so-called Mach cone in terms of wave speed c and wave source
w speed u. If the wave source moves faster than the waves, it
A gives rise to what is known as the Cherenkov radiation, see be-
u low6 . Let us consider a boat moving along a straight line and
a construct the Huygens wavefront as discussed earlier, see the
C v B x figure below. We have drawn a series of circles corresponding to
disturbances created my the moving source along its trajectory
Here we have two options: first, to use a brute force approach at a series of moments of time. The envelope of the wave-
and express the travel time t as a function of x, and then equate fronts is a straight line, because the ratio of a circle’s radius
dt is proportional to the distance of its centre from the current
dx = 0. The second option is to apply the methods of geomet-
rical optics. However, notice that in the lab frame, the speed in position O of the wave source. The angle ∠P OQ is referred
water depends on the direction of swimming, and in the water’s to as half of the Mach cone angle — it is called cone because
frame, the starting and destination points are moving. in three-dimensional geometry, the circles become spheres and
the envelope of the wave fronts becomes a cone. It is easy to
warning: Fermat’s principle can be applied only if velocities see that sin ∠P OQ = cT = c .
vT v
are the same in all directions and initial and final points are at
rest.
Now we have two sub-options. First, we can try to modify the
problem so that while the answer remains the same, the Fer-
mat’s principle becomes applicable; try to do this. The second
option is to use the Huygens’ method of building wavefronts;
let us consider this approach in more details
idea 26: When studying a reversible process, sometimes it is Continuing with the problem 18 (with the reversed velo-
easier to analyse the reverse process. cities as discussed above), we need to build “wavefronts” as
Notice that in the case of the problem 18, the process can be the sets of farthest points which the boy can reach for a given
reversed: if we make all the velocities opposite then the river moment of time T once departing from point C and starting
flows from right to left, the boy starts running from point C, swimming at an arbitrary intermediate moment of time t < T .
and wants to reach the point A in the river as fast as pos- The construction of such a wavefront is depicted in the figure
sible. Obviously, if a certain forward-process-trajectory is the below.
fastest among all the alternatives then the same applies to the
w
respective reverse process.
A
u
idea 27: For the fastest path problems in kinematics, the
approach based on the Huygens principle can be used.
uT u(t - T )
C vt - w(T - t)
For wave propagation problems, according to the Huygens prin- wT vT
ciple, wave fronts can be constructed step-by-step, by putting a
Here, the green circle corresponds to the set of farthest points
series of fictitious light sources at a previous wave front. Then,
the boy can reach if he starts swimming immediately, and the
after a short time period t, around each fictitious light source
cyan dot depicts his position if he continues running along the
a circular wavefront of radius ct is formed (where c stands for
coast; the bold black line shows the overall wavefront. If we let
the speed of light); the overall new wavefront is the envelope
the wavefront evolve, it propagates towards the point A and
of all the small wavefronts, see figure (orange dots are the first
reaches it at a certain moment T . Our procedure essentially
generation Huygens sources, and the orange circles are the re-
tests all the swimming strategies and hence, T equals to the
spective wavefronts; red lines are the overall wavefronts, and
shortest travel time; what is left to do, is to trace back, which
brown colour corresponds to the second generation Huygens
Huygens sources created that part of the wavefront which met
sources).
the point A, i.e. what would be the optimal trajectory of the
boy.
wave source idea 28: Questions involving optimal ball-throwing can be of-
ten reduced to the ballistic range problem: a cannon can shoot
projectiles with a fixed launching speed; in which range can
In the case of light waves, once we have the pattern of wave- the targets be hit? Therefore, it is useful to know the answer:
6 Classically, Cherenkov radiation is used for radiation created by superluminal charges: in dielectric medium, the speed of light waves is reduced

n times, where n is the coefficient of refraction, and relativistic particles can move faster than that; however, sonic booms (shock wave caused by a
supersonic flight) and waves behind fast boats are caused by the same physical phenomenon.

— page 8 —
 4. OPTIMAL TRAJECTORIES
the targets should be within a paraboloidal region, and the can- Here a strategy which comes to mind is compiling an equa-
non is at its focus; this paraboloid is the envelope of all the tion for the launching angle α required for hitting a target at
possible projectile trajectories, see figure (red curve is the en- the coordinates (x, z); if there are solutions to this equation,
velope, the cyan curve — trajectories for 45◦ -launching angles, the target lies within the region R, and if there are no solu-
green curves — trajectories for 0◦ -launching angles). tions then it lies outside. Then, formula describing the region
comes from the condition (inequality) which needs to be sat-
isfied for the existence of solutions. Furthermore, it is quite
easy to figure out that if the point T = (x, z) lies inside the re-
gion R, there should be actually two solutions for the launching
angle. Indeed, the target can be hit both at the rising leg of the
projectile’s trajectory, and at the descending leg of it7 So, we
expect that the equation has two solutions (T is within R), one
solution (T is at the boundary of R), or no solutions (T is out-
In order to be able to use the full potential of the idea 28, the
side of R). Such a behaviour is consistent with the quadratic
following simple facts need to be kept in mind.
equation, so we can hope that with a good parametrization (c.f.
fact 6: The black and cyan curves (in the figure above) idea 24), we obtain a quadratic equation. To summarize, let
represent the optimal trajectories for hitting targets at the red us formulate
parabola: for these trajectories, the projectile’s launching speed
idea 29: If it is asked to find the region in which a solution
at the origin is minimal.
exists to a certain problem then the boundary of this region can
Indeed, for a target inside the region R, the region can be made often be found as a curve for which some discriminant vanishes.
smaller by reducing the launching speed in such a way that the
The solution of problem 19 can be also used to derive a simple
target would still remain inside the shooting range.
particular conclusion which we formulate as a fact.
fact 7: When a target is shot with the smallest possible
launching speed, the trajectory and the shooting range bound-
fact 8: If the target is at the same level as the canon, the op-
timal shooting angle (corresponding to the smallest launching
ary (corresponding to the launching speed) are tangent to each
speed) is 45◦ .
other at the target’s position.
If you trust what has been stated above (or have proved Indeed, from the solution of problem 19 we have a quadratic
it once and now want to use the fact), it is easy to figure equation for the shooting angle, where we can put z = 0; the
out the parameters of the parabola: first, the trajectory for required result is immediately obtained if we use the fact that
90◦ -launching angle meets the tip of the parabola, and the red for optimal shooting, the discriminant of the equation is zero.
curve needs to have the same shape as the green curve, because
the green one represents the optimal trajectory for targets at pr 20. Under the assumptions of the problem 19, and know-
very low altitudes beneath the horizon. ing that the boundary of the region R is a parabola, show that
Trust, but verify: let us solve the following problem. the cannon is at the focus of the parabola.

pr 19. A cannon is situated in the origin of coordinate axes While there is an extremely simple solution to this problem, it
and can give initial velocity v0 to a projectile, the shooting dir- might not be easy to come up with it, because we need
ection can be chosen at will. What is the region of space R
that the projectile can reach? idea 30: For several problems of kinematics, geometrical solu-
tions making use of the properties of a parabola are possible;
This question is an example of a class of problems that seem
typically, such solutions are considerably shorter than the al-
easy, but the solution can get very long if brute force is ap-
ternatives.
plied. This can lead to mistakes or giving up on the problem
altogether.
fact 9: Each parabola has a focus, the properties of which
warning: If equations get tediously long then it is the right are most easily expressed in terms of geometrical optics: if
time to pause and think whether there can be an alternative
the parabola reflects light, all those rays which are parallel
way to reach the answer. If one exists, it pays to quit and try
to the symmetry axis are reflected to the focus (dashed lines
out the other path and see if it is shorter.
in the figure); due to the Fermat’ principle, this means also
In such cases, before actual calculations, you should outline a that for each point on the parabola the distance to the focus
strategy for tackling the problem: you should see in your mind plus the distance to the infinitely distant light source are equal,
a “road”, a sequence of calculations which you hope you are l1 + h1 + d1 = 2h0 + d0 = . . .; since d0 = d1 = . . ., we have
able to perform, and which, if successful, lead to the answer. 2h0 = l1 + h1 = l2 = l3 + h3 (see the figure).
7 This is somewhat simplifying statement; to be more rigorous, we need to consider the two intersection points of the projectile’s trajectory with a

horizontal line z = z0 , and how these intersections x = x1 and x = x2 (with x2 > x1 ) move when the launching angle is changed from 90◦ to 0◦ : for
α = 90◦ , they are both at x = 0, and start moving to larger values of x with increasing α. For a certain value of α, x2 reaches its maximal value xmax
and starts decreasing; the two solutions merge and disappear when the trajectory is tangent to the line z = z0 . Hence, during this process, each point
on the segment 0 < x < xmax is passed exactly two times by one of the intersection points (either x1 or x2 ), and the target at that point (x, z0 ) can
be hit by the two corresponding values of α.

— page 9 —
 5. RIGID BODIES, HINGES AND ROPES
make use of mathematical observations, formulated below as a
facts.
h0

l1
h1 fact 10: If a function f (x) has a minimum or maximum at
l2

h3<0
x = x0 then for small departures ∆x from x0 , the variation of
F l3 d2
d0 d3 the function value ∆f = f (x+∆x)−f (x) remains quadratically
d1
small8 and can be often neglected.

fact 11: As it follows from the vector addition rule (c.f. def.
pr 21. What is the minimum initial velocity that has to be 1), if the difference of two vectors of equal moduli is very small,
given to a stone in order to throw it across a sloped roof? The it is almost perpendicular to each of them.
roof has width b, its two edges have heights a and c.
Try to solve the problem 22 using this alternative approach, as
b well!

5 RIGID BODIES, HINGES AND ROPES

a
pr 23. A rigid lump has been squeezed between two plates,
c one of which is moving at velocity v1 and the other at v2 . At
the given moment, velocities are horizontal and the contact
points of the lump and plates are aligned. In the figure, mark
all points of the lump with velocity modulus equal to v1 or v2 .
This problem has two solutions, a brute force one, and a short
but tricky one. Both solutions, however, start in the same way, v1
by using
idea 31: To find a minimum (or a maximum), we have to
vary free parameters (in this case the throwing point and the
angle) by infinitesimally small increments and see what hap-
pens to the quantity of interest. If it increases for all allowed v2
variations, we have found a minimum.
You are supposed to use this method for showing that the stone This question is entirely based on
touches both edges of the roof. idea 33: The motion of a rigid body can always be considered
Both solutions share also the next step, reducing the prob- as rotation about an instantaneous axis of rotation (for 2D geo-
lem to the case c = 0. This reduction would be very useful metry, about a rotation centre) with a certain angular speed.
because then we know the optimal throwing point — the right
edge of the roof. For this step you need
~v1 ~r
idea 32: For a free-fall of a body, there is an integral of mo-
tion (quantity which is conserved), − gh, where v is an
1 2 ~v
2v
instantaneous speed, and h is the current height of the body. O

This, of course, is a particular form of the energy conservation ~v2

law, which will be considered in many more details in the other
sections (mainly in “Mechanics”).
Next we need to apply the idea 28; while in the case of the
brute force approach, this is a straightforward mathematical
application of the idea, the trickier solution makes use of the
idea 30 (and of the fact 9).
Let us consider one more problem which illustrates the us-
age of the idea 30.
pr 22. A target is shoot with the smallest possible launch-
ing speed; show that the launching velocity is perpendicular to
the terminal velocity (i.e. at the target).
This centre of rotation can be reconstructed if
(a) we know the directions of velocities of two points and these
directions are not parallel - it is where perpendicular lines
drawn from these points intersect;
(b) we know the velocities of two points, and the vectors are
parallel and perpendicular to the line connecting these
points - we find the intersection of the line connecting
the points and the line connecting the tips of velocity
vectors (see the figure)

Note that this problem can be also solved without using Note that it is also possible that the velocities of two points
the idea 30. Then, instead, you should use the idea 19 by con- are equal (by modulus and by direction), in which case the ro-
sidering the relative motion of two projectiles which are shoot tation centre is at infinity, i.e. the body moves translationally.
simultaneously at slightly different angles (still, very close to All the other combinations not covered by options (a) and (b)
the optimal angle) and with the same speed. You also need to are impossible for a rigid body.
8 This is the consequence of the Taylor expansion, to be discussed in more details in other booklets.

— page 10 —
 5. RIGID BODIES, HINGES AND ROPES
Once the centre of rotation has been found then the velocity Note that in the figure, the increments are exaggerated, ac-
of any point can be found as a vector perpendicular to the line tually these are very small. It is easy to see from the fig-
drawn from the centre of rotation with modulus proportional ure that for small angle approximation, ∆vr = ω∆t · v; with
to distance r from the centre of rotation (see the figure), ac- this we recover the expression for the radial (centripetal) ac-
cording to the formula v = ωr, where ω is the instantaneous celeration ar = ωv: we just need to divide the equality
angular speed. by ∆t. Meanwhile, due to the small angle approximation,
The last formula can be generalized: ⃗ t | ≈ |⃗v ′ | − |⃗v | = ω ′ r − ωr = ∆ωr = ε∆tr. Hence, at the
|∆v
|∆vt |
fact 12: if a vector ⃗a rotates with an angular speed ω ⃗ so that limit of infinitesimal increments, at = ∆t = εr. This leads
the rotation axis is perpendicular to ⃗a then the time derivative us to
d⃗
a
dt is perpendicular to ⃗a and | d⃗
dt | = ωa
a
fact 13: If a rigid body is rotating about a fixed axis then
This formula is derived similarly to the fact 11: we need to the acceleration of any of its points has two components: cent-
apply the definition 1 and consider a small time increment ripetal acceleration ω 2 r = v 2 /r directed towards the axis of
∆t. According to the definition of vector derivatives, d⃗ a
dt equals rotation and a component perpendicular to it, the tangential
∆⃗a
to ∆t at the limit of infinitesimally small increments ∆⃗a and acceleration ϵr. NB! the formula cannot be applied if we deal
∆t. The increment ∆⃗a is calculated using the definition 1: with an instantaneous rotation axis11 (more precisely, the for-
this is the base of the isosceles triangle, the equal sides of mula can be used if the acceleration of the instantaneous ro-
which are formed by the initial and final positions of the vector tation axis is zero). The formula v 2 /r can be also used for
⃗a. Since the vector rotates at the angular speed ω, the apex the perpendicular to the motion acceleration of a point along
angle (in radians) is ω∆t; therefore,
 ∆⃗a  the base length equals to a curved trajectory; then, r is the curvature radius of the tra-
 
2a tan(ω∆t/2) ≈ aω∆t Thus, ∆t = ∆t = aω 10 . Note that
9 aω∆t
jectory.
the formula can be rewritten via a vector product as d⃗ a
⃗ ×⃗a.
dt = ω
Within this booklet, vector products will not be needed; how-
ever, it is still useful to know that a vector product of two vec-
tors ⃗a and ⃗b equals by modulus to the surface area S = ab sin α pr 24. Cycloid is a curve which can be defined as a traject-
of the parallelepiped built on the vectors ⃗a and ⃗b (α is the angle ory of a point marked on the rim of a rolling wheel or radius
between ⃗a and ⃗b); the vector is perpendicular to both ⃗a and ⃗b, R. Determine the curvature radius of such curve at its highest
the direction being given by the screw rule (rotate the screw point.
from the first vector to the second one).
Now we can apply the fact 12 to a rotating velocity vec-
tor: let a point rotate with constant angular speed ω around
the origin so that its distance to the origin remains constant. idea 34: For kinematics problems, often the tangential accel-
 d⃗r 
According to the formula we obtain v =   = ωr and eration is not known initially, but the rotation speed is known
 v dt
a =  d⃗  2
dt = ωv = ω r. The last equation can be also writ-
and hence, the centripetal acceleration can be easily calculated
ten for the acceleration vector as ⃗a = −ω 2⃗r; the minus sign using the fact 13. You may be able figure out the direction of
means that the acceleration is directed towards the origin; be- the acceleration using other arguments, in which case you can
cause of this it is called centripetal acceleration. If the angular recover the whole acceleration by using the idea 4.
velocity is not constant, the point will obtain also a tangential
acceleration (tangent to the trajectory).
The next problem will illustrate this idea. However, the
def. 4: Angular acceleration is defined as the time derivative following idea will be also useful.
d⃗
ω
of the angular velocity, ⃗ε = dt .
Let us consider the case when the rotation axis remains fixed; idea 35: Since the distance between any two points is fixed in
then, both ⃗ε and ω
⃗ have a fixed direction and hence, are essen- a rigid body, the projections of velocities of both points on the
tially scalar quantities, so that the vector signs can be dropped. line connecting them are equal. NB! the respective projections
In order to derive an expression for the tangential acceleration, of accelerations are not necessarily equal due to the centripetal
let us consider a small time increment ∆t; this corresponds to acceleration; when dealing with accelerations you need to use
a small angular speed increment ∆ω, as well as to a small ve- the fact 13 in one point’s frame of reference, instead.
locity increment ∆⃗v . Let us decompose the velocity increment
into radial and tangential components, ∆⃗v = ∆⃗vr + ∆⃗vt . Now,
the velocity vector changes both its direction and length, see
pr 25. A hinged structure consists of two links of length
figure below.
2l. One of its ends is attached to a wall, the other is moving
~v ’ at distance 3l from the wall at constant vertical velocity v0 .
ω∆t Find the acceleration of the hinge connecting the links when a)
∆v
~r the link closer to the wall is horizontal b) the velocity of the
~v ∆v
~t
connection point is zero.
9 Herewe have used approximate calculation for small angles: sin α ≈ tan α ≈ α, where α ≪ 1 is measured in radians.
10 We have replaced approximate equality by strict equality because the tangent is taken from an infinitely small angle
11 For instantaneous axis, r would not be constant, which makes the derivation void.

— page 11 —
 6.
3l
2l
2l
v fixed length then one way of calculating velocities and acceler-
Now, let us continue with rolling spool problems.
pr 26. Some thread has been wound around a cylinder, the 6 MISCELLANEOUS TOPICS
other end of the thread is fixed to a wall. The cylinder lies
on a horizontal surface that is being pulled with horizontal ve- idea 39: Some kinematics problems are based on the con-
locity v (perpendicular to the axis of the cylinder). Find the
velocity of the cylinder’s axis as a function of α, the angle that
thread makes with the vertical. The cylinder rolls on the sur-
face without slipping.
MISCELLANEOUS TOPICS
v and the angle between the floor and the rod is α?
Here the velocity can be found using the idea 33 and the ac-
celeration — using the idea 34; however, please solve it now
using the following idea (its full solution is given in the section
of hints).
idea 38: If parts of a system are connected by some ties with
ations is to write out this relation in terms of coordinates and
take the time derivative (twice for acceleration) of the whole
expression.
tinuity law12 : if there is a stationary flow of something and we
consider a certain region of space, as much as flows in, flows
also out.
Let us illustrate the continuity law with the following problem.

α pr 28. A single-lane road is full of cars which move at the

speed v = 90 km/h. The average distance between neighbour-
ing cars is such that a standing observer would measure a time
lapse τ = 2 s from the head of the first car until the head of
v the second car. One car has to stop due to malfunctioning and
a queue of standing cars starts forming behind it. Find the
This problem can be solved using the idea 33. However it can speed u with which the queue length is growing if the average
be also solved using the following idea (you are encouraged to distance between the centres of neighbouring cars in the queue
try both methods). and the spool is rolling along some surface. is l = 6 m?
In these cases, we need
Pay attention that the flow of cars is not stationary in the lab
idea 36: Problems involving spools and rope (un)wounding frame of reference; hence, first you need to use the idea 1 to
can be typically solved by writing down the “rope balance” make it stationary.
equation, i.e. by relating the rope unwinding rate to the velo-
city of the spool.
idea 40: In the case of problems involving moving objects
and moving medium (air, water), if the moving objects leave
When writing down such equations, it is useful to notice that as trails, and you are asked to analyse a sketch (a photo), pay
long as the unwound rope does not rotate, the rate of unwind- attention to the fact that if the objects met, the meeting point
ing equals to ΩR, where Ω is the spool’s angular speed, and R corresponds to an intersection point of the trails.
— its radius; if the unwound rope rotates, idea 17 can be used
to conclude that now the spool’s angular speed Ω needs to be pr 29. Figure below is copied from an aerophoto: there are
substituted with the difference of two angular velocities. two trains (depicted in red) which both travel with the speed of
In order to relate the rope unwinding rate to the motion of v = 50 km/h along a railway (grey dashed line). Their engines
the spool, the following very general idea (with a wide applic- emit fume, the trails of which are depicted by black lines. De-
ation scope) is helpful. termine the direction and speed of wind (express the direction
idea 37: Draw two very close (infinitesimally close) states of of wind as a clock-wise rotation angle from north). You may
the system and examine the change in the quantity of interest draw lines and measure distances using a ruler in the figure.
(in this case, the length of the rope).
When doing that, we should not forget that the change was
infinitesimal, so that we can simplify our calculations (e.g. two
N
subsequent states of the rope can be considered parallel). Note
that we have actually already used this method when deriving
the fact 13.
S
pr 27. One end of a rod of length l is supported by the
floor and the other leans against a vertical wall. What is the idea 41: Sometimes it is useful to include a time axis in addi-
velocity u and acceleration a of the lower end at the moment tion to spatial coordinates and analyse graphs even if the prob-
when the upper end is sliding downwards at a constant velocity lem’s text does not mention time dependence explicitly. Thus
12 The continuity law plays an important role in physics, in general.

— page 12 —
 6. MISCELLANEOUS TOPICS
we’ll be studying two-dimensional graphs for one-dimensional idea 43: For some problems, the main difficulty is understand-
problems and three-dimensional graphs for a planar motion. ing, what is going on; once you understand, the calculations are
typically quite easy. Keep your mind calm: in physics compet-
Try to apply this idea to the following problem. Keep in mind
itions, you are not asked to do something impossible: act as a
the following stereometric facts: 3 points always lie in a plane;
detective by investigating step-by-step what is going on, and
a straight line and one point determine a plane (unless the
narrowing down possibilities by the exclusion method.
point lies on the line).

pr 30. Points A, B and C lie on a straight line, so that B is

situated between A and C. Three bodies a, b and c start from And here is a test case for the detective inside you!
these points, moving at constant (but different) velocities. It
is known that i) if c was missing, a and b would collide and ii)
if b was missing, a and c would collide and it would happen pr 32. This photo of a rotating propeller of a plane is taken
before than in i). Would b and c collide if a was missing? with a phone camera. For such a camera, the image is scanned
line-by-line: at first, the leftmost column of image pixels is
In mathematics, for geometrical problems, there is a tech- read, followed by the second column of pixels, etc.
nique called auxiliary constructions: the solution of the prob- i) in which direction does the propeller rotate as seen by the
lem can be significantly simplified if you draw an additional photographer (clockwise, or counterclockwise)?
line(s). While in physics, such constructions are less often used, ii) How many blades does the propeller have?
but in some cases auxiliary constructions are still useful (in iii) How many rotations does the propeller make in one minute
fact, idea 41 makes use of such construction). The following if the total scanning time of this image was 1 seconds?
8
idea represents a recipe for one more auxiliary construction.
idea 42: Sometimes it is useful to consider instead of a single
particle, a fictitious ensemble of auxiliary particles.
While the following problem can be solved via a brute force
approach, introducing an ensemble of auxiliary particles (re-
leased all over the wheel at the same time) simplifies your cal-
culations.

pr 31. A wheel with radius R is situated at height R from

the ground and is rotating at angular velocity Ω. At some point
A, a drop of water separates from the wheel and reaches the
ground at point B situated directly below the wheel’s axle (see
the figure). Find the falling time of the drop and the location
of point A (i.e. angle α).

a α R The first part of the next problem can be also solved using
the idea 43. The second part, however, becomes mathematic-
ally challenging; it would be much easier to solve it using the
following idea.

B idea 44: Wave propagation problems can be often conveni-

In addition to the previous ones, there is an idea that helps
solving this problem. However, it is very non-standard, i.e.
there are almost no other problems where to use this idea. To
apply the idea elsewhere, it has to be formulated in a very gen-
eral form: More specifically for this question: imagine that at
the same time with the given drop, small droplets separated
from all the other points of the wheel as well. It is clear that
in a freely falling frame of reference, this set of droplets forms
a circle at all times; it should not be too difficult to find the
radius of that circle as a function of time. The first droplet
touching the ground is the one that we are interested in. We
can express time from the equation describing the condition of
touching (the height of the lowest point of the circle becomes
zero) - idea 10.
13 Sinusoidal waves are studied in more details in the booklet “Wave optics”.
ently analysed using the wave vectors: using orthogonal co-
ordinates (x, y, z), and time t, a sinusoidal wave a(x, y, z, t) can
be represented as sin(⃗k⃗r − ωt), where the scalar product of the
wave vector k and radius vector ⃗r equals to ⃗k⃗r = kx x+ky y+kz z.
Here ω is the angular frequency of the wave and the wave’s
propagation speed v = ω/k 13 .
pr 33. i) There are two combs which are depicted in the
figure below. The figure renders the proportions of the combs
correctly; the scale of the figure is unknown. The grey comb
moves with a speed of v = 1 cm/s (the direction of its motion is
shown in the figure); the black comb is at rest. Find the speed
and direction of motion of the dark stripes.

— page 13 —
 7. CONCLUSION
dark stripe are those mathematical skills with which you are assumed to
be familiar with: when not certain which solving route to take,
this can help you narrowing down the options16 . While extra
skills are sometimes useful, these can also easily lead you to a
wrong (mathematically unnecessarily complex or even unsolv-
able) path; therefore, apply your extra skills only if you can see
1 cm/s in your mind the whole path to the answer, i.e. when you are
ii) Now, we have the same situation (and the same question) as sure that you will not run into mathematical difficulties.
before, except that the black comb is under a small angle α ≪ 1 Note that such nontrivial conservation laws can include all
with respect to the grey comb (for numerics, use α = 0.1 rad). these quantities which enter the differential equations, except
for those derivatives for which these equations are written. For
How to apply the idea 44 to the problem 33? Pay attention instance, here, the equations express velocities in terms of co-
to the following. Instead of a moving comb, we can take a ordinates, so velocities cannot be included. If the equations
wave a(x, y, t) = sin(⃗k⃗r − ωt); then, centres of the teeth corres- were expressing accelerations in terms of velocities and coordin-
pond to points where a(x, y, t) = 1. If we have two overlapping ates then both velocities and coordinates could enter the law
combs, the “transparent regions” where the centres of the teeth (but not the accelerations). The conservation law is typically
of both combs are at the same position can be found as the proved by showing that its time derivative is zero. For instance,
points where a1 (x, y, t)a2 (x, y, t) = 1, and the regions where if our conservation law would be a sum of two distances, we
the teeth are almost at the same position can be found as the would need to show that the changing rates of those two dis-
points where a1 (x, y, t)a2 (x, y, t) ≈ 1. Now we have a product tances are equal by modulus and have opposite signs.
of two sinusoidal waves, so we can say that the moiré pattern
(this is how these dark stripes are called) is due to a nonlinear
interaction of waves. Further, in order to answer the question 7 CONCLUSION
about what is the propagation speed of the moiré pattern, we
Some of the ideas presented here (in particular, 1, 4, 5, 10, 14,
need to decompose this product of sinusoids into a sum of si-
18, 19, 31, 37) are more universal than others. In any case,
nusoids (for a sinusoid, we can calculate the speed as the ratio
they are all worth remembering. Note that it is always useful
of the angular frequency and the modulus of the wave vector).
to summarise the idea(s) in your mind after finding a solution
idea 45: Well-known conservation laws are the ones of en- to a new problem; from time-to-time you may be able to come
ergy, momentum, and angular momentum14 . However, some- up with some entirely new ideas! For further practice, here are
times additional quantities can be conserved (then, of course, some additional problems based on aforementioned ideas.
you need to show that it is conserved), which makes otherwise
mathematically very difficult problems easily solvable. pr 35. A boy is swimming in a fast-flowing river of width
L; the water speed is u = 2 m/s, and the boy can swim with
What can be hints that a non-trivial conservation law is valid speed v = 1 m/s. While being at point A near one coast, he
for a problem? Well, if you understand clearly that the diffi- wants to reach such a point B near the other coast which is
culty is mathematical (you understand what is going on, and directly across the river (AB ⃗ ⊥ ⃗u). Since the river is too fast,
are able to write down the equations) and what you have at he cannot avoid being carried downstream to a certain distance
hand are differential equation(s)15 then that might be the case. a from point B; what is the smallest possible value of a which
For instance, let us consider the following problem. he can achieve for the optimally chosen swimming direction?
pr 34. A dog is chasing a fox running at constant velocity
v along a straight line. The modulus of the dog’s velocity is pr 36. Rings O and O′ are slipping freely along vertical
constant and also equal to v, but the vector ⃗v is always direc- fixed rails AB and A′ B ′ (see the figure); the distance between
ted towards the fox. When the dog noticed the fox and started the rails is b. Some unstretchable rope has been tied to ring O
chasing, the distance between them was L and at the first mo- and pulled through ring O′ . The other end of the rope is fixed
ment, their velocity vectors formed a right angle. What is the to point A′ . At the moment when ∠AOO′ = α, the ring O′ is
minimal distance between them during the chase? moving downwards at a constant velocity v. Find the velocity
Here we can express the velocity components of the dog in and acceleration of the ring O at the same moment.
A A
terms of its coordinates (using also the coordinates of the fox,
which are known functions of time); these are differential equa-
tions. Since you are not supposed to be able to solve differ- O
ential equations for this booklet, this should not be the way
to go (furthermore, unless properly parametrized, c.f. idea 24, O
these equations are actually not easy to solve). Note that at B
B
physics competitions, you should always keep in mind, what
14 These are typically not applicable to the kinematics problem.
15 These are equations relating derivatives to coordinates (typically, velocities or accelerations as functions of coordinates and/or velocities and which
define the evolution of the system.
16 Of course, such arguments are cannot be used for real-life problems.

— page 14 —
 7. CONCLUSION

pr 37. A small ball is moving at velocity v0 along a smooth pr 42. A hinged structure consists of rhombi with side
horizontal surface and falls into a cylindrical vertical well at lengths l, 2l and 3l (see the figure). Point A3 is moving at
point A, after which it starts bouncing elastically against the constant horizontal velocity v0 . Find the velocities of points
wall and the smooth horizontal bottom. The well has depth A1 , A2 and B2 at the moment when all angles of the structure
H and radius R; the angle between ⃗v0 and the well’s diameter are equal to 90◦ . Also, find the acceleration of point B2 .
drawn through point A is α. What condition between R, H, B3
v0 and α must be satisfied for the ball to exit from the well B2
again? Rotation of the ball can be neglected. B1
~v0
A1 A2 A3 v0
α
A
R

pr 38. A ball A lies on a wedge with angle α. It is also

tied to an unstretchable string, the other end of which is at-
tached to a vertical wall at point B (see the figure). What will
be the trajectory of the ball? What is its acceleration if the
acceleration of the wedge is a?
pr 43. Two motorboats simultaneously depart from two har-
bours (A and B) at distance l from one another, velocities of
the boats are v1 and v2 , respectively. The angles between their
velocities and the line connecting A and B are α and β, re-
spectively. What is the minimum distance between the boats?

pr 44. A heavy disk of radius R is rolling downwards, un-

winding two strings in the process. The strings are attached to
a the ceiling and always remain under tension during the motion.
What was the magnitude of velocity of the disk’s centre when
α its angular velocity was ω and the angle between the strings
was α?

pr 39. A dog is chasing a fox running at constant velocity

v1 along a straight line. The modulus of the dog’s velocity is
α
constant and equal to v2 , but the vector ⃗v is always directed
ω
towards the fox. The distance between the animals was l at the
moment when their velocity vectors were perpendicular. What
was the acceleration of the dog at that moment?

pr 40. A spinner having the shape of a cone (height h, ra- pr 45. Two boards have been placed at right angles to one
another. Their line of touching is horizontal and one of them
dius r) is moving along a smooth table and spinning rapidly.
(A) makes an angle α to the horizontal. An elastic ball is re-
What does its translational velocity v have to be in order to
leased at a point at distance a from plane A and b from B.
avoid bumping against the edge of the table when it gets there?
r On the average, how many times does the ball bounce against
wall B for each time it bounces against wall A? Collisions are
absolutely elastic.
v b
h
a
B
A
α

pr 41. A uniform rope has been manufactured from an ex- pr 46. One end of a string of a negligible mass has been at-
plosive material, combustion travels along the rope at velocity tached to the side of a cylinder, not far from the ground. The
v. The velocity of a shock wave in the air is c, with c < v. Along cylinder itself has been fixed on smooth slippery horizontal sur-
which curve should the rope be laid out to make the shock wave face, with its axis vertical. The string has been wound k times
reach a given point at the same time from all points of the rope? around the cylinder. The free end of the string has been tied
(Finding a quantitative formula for the shape requires solving to a block, which is given a horizontal velocity v directed along
a very simple differential equation.) the radius vector drawn from the cylinder’s axis. After what
— page 15 —

time will the string be fully wound around the cylinder again,
this time the other way round? [This problem leads to a very
simple differential equation; if you don’t know how to solve it,
2
dl
the following equality can be helpful: l dt = 12 d(ldt ) .]
v
r
pr 47. A heavy box is being pulled using two tractors. One
of these has velocity v1 , the other v2 , the angle between velo-
cities is α. What is the velocity of the box, if we assume that
the ropes are parallel to velocity vectors?
v2
v1
α simultaneously, and travel with a constant speed v. The figure
pr 48. A boy is running on a large field of ice with velocity
v = 5 m/s toward the north. The coefficient of friction between
his feet and the ice is µ = 0.1. Assume as a simplification that
the reaction force between the boy and the ice stays constant
(in reality it varies with every push, but the assumption is jus-
tified by the fact that the value averaged over one step stays
constant).
i) What is the minimum time necessary for him to change his
moving direction to point towards the east so that the final
speed is also v = 5 m/s?
i) What is the shape of the optimal trajectory called?
pr 49. A ball thrown with an initial speed v0 moves in a
homogeneous gravitational field of strength g; neglect the air
drag. The throwing point can be freely selected on the ground
level z = 0, and the launching angle can be adjusted as needed;
the aim is to hit the topmost point of a spherical building of
radius R (see fig.) with as small as possible initial speed v0
(prior hitting the target, bouncing off the roof is not allowed).
Sketch qualitatively the shape of the optimal trajectory of the
ball. What is the minimal launching speed vmin needed to hit
the topmost point of a spherical building of radius R?
z
sp h eri
din
ca l C
l b ui x
pr 50. The figure represents a photo which was taken using
a very long exposure time (camera was pointing directly down).
What you can see is a trace of a blue lamp which burned con-
tinuously, but also flashed periodically with a red light (after
each t = 0.1 s. The lamp was fixed to the surface of a solid
disk, at the distance a = 4.5 cm from its symmetry axis. The
— page 16 —
7. CONCLUSION
axis was vertical, and the disk slid and rotated freely on a ho-
rizontal smooth ice surface. What was the speed of the centre
of the disk? You can take measurements from the figure using
a ruler.
pr 51. There is a capital O and three cities A, B and C, con-
nected with the capital via roads 1, 2, and 3 as shown in the left
figure. Each road has length 2a. Two cars travel from one city
to another: they depart from their respective starting points
on the right depicts the increasing rate of the distance between
the cars (negative values means that the distance decreases) as
measured by the GPS devices of the cars. The turns are taken
by the cars so fast that the GPS devices will not record the
behaviour during these periods.
i) Which cities were the starting and destination points of the
cars? Motivate your answer.
ii) What is the surface area between the vdist -graph and the
t-axis for the interval from t = 0 to t = a/v?
iii) Now let us consider a case when three cars (denoted by A,
B, and C) depart simultaneously from their cities (A, B, and
C, respectively) towards the capital; all the cars travel with a
constant speed v. Sketch the graphs for the distance changing
rate for the following car pairs: A − B and B − C.
iv) Suppose that now the GPS-devices are good enough to re-
cord the periods of taking the turns. Sketch a new appropriate
graph for the pair of cars B − C. The curvature of the turns is
small enough so that the cars can still keep theirs speed v.
A B vdist
road 1
a
a
road 2
a 2a
90o a O a 90o 0 v vt
90o
a
road 3 - v0
90o - v0 2
a
pr 52. Consider two rings with radius r as depicted in the
figure: the blue ring is at rest, and the yellow ring rotates
around the point O (which is one of the intersection points
of the two rings) with a constant angular speed ω. Find the
minimal and maximal speeds vmin and vmax of the other inter-
section point of the two rings.
 7. CONCLUSION

pr 53. A moving bicycle is photographed using a relatively

long exposure time with a motionless camera. As a result, the
bike appears smudged on the photo; however, certain points of
certain spokes appear sharp. Determine the shape of the curve pr 56. On a wide river, two boats move with constant velo-
upon which the sharp points of spokes lie. cities. The velocity of the water in the river is constant across
the whole area depicted in the figure, and parallel to the coast-
line. The figure is based on a photo which was taken from air,
pr 54. A lamp is attached to the edge of a disk, which the camera being directed straight down. The positions of the
moves (slides) rotating on ice. The lamp emits light pulses: boats are marked with a square and a triangle, and the posi-
the duration of each pulse is negligible, the interval between tions of litter fallen from the boats — with pentagrams. One
two pulses is τ = 100 ms. The first pulse is of orange light, of the boats departed from the point A; it is known that the
the next one is blue, followed by red, green, yellow, and again boats did meet with each other at a certain moment. From
orange (the process starts repeating periodically). The motion which coastal point did the other boat depart? Solve the prob-
of the disk is photographed using so long exposure time that lem using geometrical constructions.
exactly four pulses are recorded on the photo (see figure). Due
to the shortness of the pulses and small size of the lamp, each
pulse corresponds to a coloured dot on the photo. The colors
of the dots are provided with the corresponding lettering: o —
orange, b — blue, r —red, g — green, and y — yellow). The
friction forces acting on the disk can be neglected.

i) Mark on the figure by numbers (1–4) the order of the pulses

(dots). Motivate your answer. What can be said about the
value of the exposure time?
A
ii) Using the provided figure, find the radius of the disk R, the
velocity of the center of the disk v and the angular velocity pr 57. Let us consider the merger of two traffic lanes A
ω (it is known that ω < 30 rad/s). The scale of the figure is and B into a single lane C, see figure. During a rush hour, all
provided by the image of a line of length l = 10 cm; the lanes are filled with cars; the average distance between the
cars can be assumed to be equal for all the lanes. The lengths
of the lanes A, B, and C is respectively equal to LA = 1 km,
g
LB = 3 km, and LC = 2 km. The average speed of cars on the
lane A is vA = 3 km/h; and the travel time of a car on the
lane B is tB = 36 min. How long will it take for a car to travel
y
10 cm from the beginning of the lane A to the end of the lane C?
b A A
r
C C
B B

pr 55. The photo depicts a jet of water, together with back-

ground grid. The pitch of the grid equals to the diameter of pr 58. On a windless rainy day, a standing man gets wet
the jet at the exit from the horizontal pipe. The water flow during t = 2 min; if he runs with the speed v2 = 18 km/h, he
rate is constant in time, and if a vessel of volume V = 150 cm3 gets wet during t2 = 0.5 min. How long will it take for him to
is used to collect the outflowing water, it is filled during the get wet when walking with the speed v1 = 6 km/h? Assume
time period of t = 5 min. Find the diameter of the jet at the that the body shape of a man can be approximated (a) with
exit of the pipe. a vertical rectangular prism; (b) with a sphere. Here “getting
— page 17 —
 7. CONCLUSION
wet” is defined as receiving a certain amount of water. the effective wavelength grows in time: at the beginning, the
wavelength is of the order of the circle radius, but later, it tends
pr 59. A photographer took a photo of a waterfall. Due to towards a value which is of the order of the water depth h. It
the reflected sunlight, the droplet were seen as speckles. Owing turns out that at the beginning, the wavecrest moves with ac-
to the fast falling speed of the droplets, the speckles created celeration a = g/π, where g is the free fall acceleration; later, it
√
bright stripes on the photo. When the camera was in a normal tends towards the value v∞ = hg. Based on this knowledge,
“landscape” position, the length of the stripes was l1 = 120 estimate the depth of the pond h assuming that it is constant
pixels; when the camera was rotated around the optical axis of everywhere; express your answer in terms of the length scale L
the lens by 180◦ into a “head-down” position, the length of the provided in the figure. You can take measurements from the
stripes was l2 = 200 pixels. What was the length of the stripes figure using a ruler.
when the camera was in a “portrait” position, i.e. rotated by
90◦ ? Assume that the exposure time was equal in all three pr 61. A motorboat approaches a straight coastline perpen-
cases. If there are several possibilities, give all the possible dicularly, and at a distance L starts turning back by drawing a
answers. half-circle of radius R; later it departs perpendicularly to the
coastline. The speed of the boat is constant and equal to v,
Hint. The main components of the camera are the lens which
the water wave speed can be assumed to be constant and equal
creates an image on the sensor, and the shutter. The purpose
to u (with u < v). How long will it take for the waves of the
of the shutter is to limit the time during which the sensor is ex-
wake behind the boat to reach the coast (as measured from the
posed to the light to a short (and appropriate) period of time:
moment when the boat starts turning)?
normally, it covers the sensor, and the image created by the
lens falls on the sensor only when the shutter is opened. The
shutter is made of two curtains: at the beginning, the first cur-
pr 62. On a wide river, a motorboat moves with a con-
tain covers the sensor; when a photo is taken, it moves down stant speed v = 7 m/s from village A to village B over the
with a certain speed v opening the sensor; once the sensor has river. When answering the following questions you may take
been opened for the required period of time, the second curtain measurements from the figure below depicting waves behind
moves down with the same speed v, covering again the sensor. the boat. What is the water speed in the river, and what is
In order to achieve very short exposure times, both curtains the water depth h? Note: wave speed in shallow water17 is
√ 2
move together, creating a narrow moving slit through which w = gh, where g = 9.81 m/s .
the light can pass through to reach the sensor.

pr 60. If a stone is thrown into a pond, a circular wave is

created which expands in time. The following figure depicts
the propagation of such a wave: different circles correspond A B
to the position of the wave crest at different moments of time;
the underlying snapshots have been taken with a regular (but
unknown) interval.

L
pr 63. Provided sketches (a) and (b) are made on the basis
of satellite images, preserving proportions. They represent
tractors, together with their smoke trails. The tractors were
moving along the roads in the direction indicated by the ar-
rows. The velocity of all the tractors was v0 = 30 km/h. For
sketch (a), the direction of wind is indicated by another arrow.
When solving the problem, you may draw lines and measure
distances using a ruler.
i) Using the provided sketch, find the wind speed for case (a).
ii) Using the provided sketch, find the wind speed for case (b).

(a)

(b)

Note that the wave speed depends on the wavelength, and here,
17 when water depth is noticeably smaller than the wavelength

— page 18 —

pr 64. The following snapshot (a larger version is on an
extra sheet) depicts two balls that were thrown simultaneously
and with the same initial speed, but in different directions from 3.
point P . What was the initial speed? Use g = 9.8 m/s2 .
1st ball
g 2nd ball
P apply the idea 4.
throwing 5. In the boards frame of reference, there is only horizontal
point
10m
pr 65. A boat travelled from its home port to an island
at the distance of l = 4 km directly towards south. Its tra-
jectory consisted of three straight segments the directions of
which were not recorded. During each of the segments, the boat
maintained a constant velocity; however, a different speed was
maintained for diffe- rent segments. During the travel time,
wind speed and direction was measured from the boat. The
travel time on the first segment was t1 = 3 min, the measured
wind speed was v1 = 15 m/s and the wind blew directly from
east. The travel time on the first segment was t2 = 1.5 min, the
measured wind speed was v2 = 10 m/s and the wind blew dir- 7.
ectly from southeast. The travel time on the first segment was
t3 = 1.5 min, the measured wind speed was v3 = 5 m/s and the
wind blew directly from southwest. What was the wind speed?
It is known that the wind speed and direction were constant 8.
during all the travel time.
pr 66. There is a long chute of constant slope angle along
which balls can slide frictionlessly (the chute is narrow so that
the motion of the balls is essentially one-dimensional). Let
there be N identical perfectly elastic balls sliding on that chute.
The total number of pair-wise collisions between the balls in the
chute depends on their initial velocities and positions. What
is the largest possible number of collisions? (If you don’t know
how will move two absolutely elastic balls after a collision, check
the hints section.)
8 HINTS
1. In the water’s frame of reference, it is clear that departing
from the barge and returning to it took exactly the same
amount of time.
2. In the frame of the red plane, the blue plane moves along
a line s which forms an angle α = arctan 34 with the hori-
zontal dashed line in the figure. The distance of the red
plane from this line is most conveniently found considering
— page 19 —
8. HINTS
two similar right triangles, the larger of which is formed
by the line s, and the two dashed lines in the figure.
Use the frame which moves with the speed v2 to find the
horizontal component of the intersection point’s velocity.
Now, in the lab frame, we know the direction of this ve-
locity, as well as the horizontal projection; apply the idea
4.
4. Since the ascending speed is constant, it is sufficient to
calculate it for a single position of the balloon — when its
height is small and hence, the horizontal velocity is almost
zero; apply the idea 5. For t = 7 min, the angular ascend-
ing speed is zero, hence the balloon needs to move along
the line connecting the balloon and the observation point;
force (the friction force) has a constant direction, antipar-
allel to the velocity.
6. According to the idea 7, we use the conveyor’s frame, but
as we are asked about the speed in lab frame, we need
to switch back to the lab frame. In the conveyor’s frame,
the velocity vector becomes shorter while preserving the
direction, i.e. can be represented as w ⃗ = kw ⃗ 0 , where its
initial value w⃗ 0 = ⃗v0 − ⃗u0 and the factor k takes numerical
values from 0 to 1. Hence, the velocity in the lab frame
⃗v = ⃗u0 + k w⃗ 0 : this is a vector connecting the right angle
of the right triangle defined by its catheti ⃗u0 and ⃗v0 with a
point on the hypotenuse; the specific position of this point
depends on the value of the factor k (which is a function
of time).
Express the lateral displacement of the ball as the sum of
two components: lateral displacement in the air’s frame of
reference (a trigonometrical task; this does not depend on
t), and the lateral displacement of the moving frame.
Algebraic approach: take one of the axes (say x) to be
perpendicular to the racket’s plane and the other one
(y) parallel to it. Absence of rotation means that the y-
components of the ball’s and racket’s velocities are equal,
uy = vy , and there is no parallel force acting on the ball,
hence vy′ = vy . Using idea 13 we find that vx′ = −vx + 2ux .
Applying idea 15 to the vectors ⃗v and ⃗v ′ gives us an equa-
tion for finding ux ; apply Pythagoras’ theorem obtain |⃗u|.
To find angle β, express tan β = uy /ux .
Geometric approach: draw a right trapezoid as follows:
we decompose ⃗v into parallel and perpendicular compon-
ents, ⃗v = ⃗vx + ⃗vy ; let us mark points A, B and C so that
⃗ = ⃗vx and BC
AB ⃗ = ⃗vy (then, AC ⃗ = ⃗v ). Next we mark
points D, E and F so that CD = ⃗vy′ = ⃗vy , DE
⃗ ⃗ = −⃗vx ,
⃗ = 2⃗ux ; then, CF ⃗ = ⃗vy′ − ⃗vx + 2⃗ux ≡ ⃗v ′ and
and EF
AF = 2⃗vy + 2⃗ux ≡ 2⃗u. Due to the problem conditions,
⃗
∠ACF = 90◦ . Let us also mark point G as the centre of
AF ; then, GC is both the median of the right trapezoid
ABDF (and hence, parallel to AB and the x-axis), and
the median of the triangle ACF . What is left to do, is ex-
pressing the hypotenuse of △ACF in terms of v = |AC|,
and apply the idea 16.
 8. HINTS
9. Use the frame where the mirror is at rest: the source S the rule of velocity addition, for each of the turtles we
rotates with angular velocity −ω. Now go back to the lab need to use the local velocity vector of the rotating frame
frame and find the angular velocity of the image in that at the location of the particular turtle. According to the
frame. second method, at each moment, let us project the velocit-
10. The area under the graph, from t = 0 until the given mo- ies of two turtles on the straight line connecting them —
ment, gives the displacement, and regions below the t-axis that way we can find the rate of decrease in the distance
make negative contributions. Hence, if a certain moment between two turtles.
t corresponds to a maximal displacement then v(t) = 0 16. Write down the relationship between small increments:
(otherwise, the displacement could be increased somewhat dk = v · dt/(L + ut); the answer is obtained upon integra-
by making t slightly smaller or larger, depending on the tion of the left- and right-hand parts of this equality.
sign of v(t). (Alternatively, we can say that extrema cor- 17. Apparently the boy will need to arrive to the coast OM
respond to zero derivative, and the derivative of ∆x is at the right angle, c.f. idea 2. Now we can apply the idea
v.) The same arguments lead us to the conclusion that 25 together with the fact 5 to the refraction of his traject-
for slightly smaller t-values we must have positive v, and ory at the coast OP to find the angle at which he needs to
for slightly larger t-values — negative v. Hence, the can- arrive to the coast OP (the answers are expressed straight-
didates are t = 4.7 s, t = 7 s, t = 12.5 s, and t = 18.3 s. forwardly in terms of this angle).
Calculate the surface areas to see, which of them maxim-
18. First approach: use the water’s frame of reference, because
izes the displacement.
∑ then the swimming speed is independent of the swimming
11. Let us divide the displacement into small pieces, s = ∆s, direction. In that frame, the boy moves on the coast with
where ∆s = v∆t. If the function v(t) were known, the the speed equal to v + w. It is clear that the boy needs to
last formula would have been completed our task, because
∑ start immediately swimming, i.e. it doesn’t matter weather
v(t)∆t is the sum of rectangles making up the area un- the point A is moving or motionless in the new frame. It
der the v − t-graph . However, the acceleration is given is also clear that if we have found the fastest way of reach-
to us as a function of v, hence we need to substitute ∆t ing the point C, the same trajectory would give us the
with ∆v. While trying to do that, we can introduce the fastest way of reaching any other upstream point on the
acceleration (which is given to as a function of v): same coast; in particular, we can take a point C ′ which
∆t ∆v ∆v
∆t = ∆v · = = . moves together with the water (is motionless in the water’s
∆v ∆v/∆t a frame), and the optimal trajectory would still remain the
This result serves us perfectly well: ∫
∑ ∑v same. With these modifications, we have a problem where
v
s= v∆t = ∆v → dv, we can apply the Fermat’ principle; the resulting geomet-
a a(v)
rical optics problem is essentially the problem of finding
i.e. the displacement equals to the surface area under a
v the angle of total internal reflection. For the second ap-
graph which depicts a(v) as a function of v.
proach, once the front meets the point A, the front forms
12. Use the frame of one of the sliding balls — according to a cathetus of a right triangle AP Q, where Q is the point
the acceleration addition rule, the other one moves with a where the front meets the riverbank, and P is the position
constant horizontal acceleration. Apply the idea 2 to find of that Huygens source on the riverbank which creates the
the position where the distance is minimal. Express the circular wave meeting the point A. Notice that the point
answer in terms of the distance AB; apply the idea 10 for P is the point where the boy needs to start swimming in
finding the distance AB. the water’s frame of reference, and is displaced by wT from
13. If the axis x is parallel to the plane (and points down- the corresponding point in the lab frame.
wards), the ball performs a free fall along x-axis, the accel- 19. Once you write down trajectory parametrically, x = x(t)
eration being equal g sin α; if the axis y is perpendicular to and z = z(t), time t can be eliminated; as a result, you
the plane, the ball bounces along the y-axis up and down, obtain an equation relating x, y, and the shooting angle α
with the free fall acceleration g cos α. to each other, which we consider as an equation for find-
14. Use perpendicular coordinates so that the x-axis is along ing angle α. The angle enters into this equation via two
the contact line of the two surfaces and y-axis lays on the terms, one containing tan α, and the other — cos−2 α. In
inclined surface; then, motion in the x-direction is inde- order to solve such equations, one possibility is to express
pendent of the motion in y − z-plane. Due to the idea all trigonometric functions via a single one. It is possible
21, the speed remains constant throughout the transition to express tan α via cos α, but that involves a square root,
from one plane to the other. Use fact 3 in conjunction which is inconvenient. Meanwhile, cos−2 α can be nicely
with the provided graph to figure out the value of gy , the expressed in terms of tan α, and so, this is the way to go;
projection of the free fall acceleration to the y-axis. Now, as a result, you’ll obtain a quadratic equation for tan α.
2
since we know the full free fall acceleration g = 9.8 m/s , 20. Use the fact that all vertical rays are reflected by the range
the relationship gy = g sin α allows us to find the angle α. boundary so that they will pass through the focus. It is
15. According to the first method, we use the frame which enough to find the intersection point of two rays. Take one
co-rotates with the turtles, so that in the new frame, the ray at x = 0, and the other one such that it will hit the
turtles move radially towards the centre; when applying range boundary at the level z = 0 (we know the tangent
— page 20 —

of the range boundary at that point due to the fact 8).
21. Step 1, proof by contradiction: if the trajectory doesn’t
touch neither of the edges, the throwing speed can obvi-
ously be reduced slightly while keeping the angle constant.
If it touches only one edge, let it be the farther edge, the
boy can step slightly forward so that now the trajectory
doesn’t touch neither of the edges. Step 2: from the law
v(z)2 − 2gz = cosnt, we conclude that the speed at the
height z = c [denoted as v(c)] is a uniquely defined growing
function of the speed at the ground level v(0): if the speed
v(0) is minimal then v(c) is also minimal. Step 3, brute
force approach: find the minimal throwing speed from the
right edge F of the roof by √ requiring that the other edge
with coordinates P = (a−c, b2 − (a − c)2 belongs to the
parabola found in problem 19. Geometric approach (idea
30: according to the idea 28 (keep in mind the facts 6 and
7), when throwing optimally from the point F to point P ,
the point P belongs to the envelope-parabola (which separ-
ates the region where the targets can be hit with the given
speed, see idea 28), and F is its focus; let the respective
minimal throwing speed be u [recall that this is the speed
at the altitude of the point F , i.e. u = v(c)]. According
to the idea 28, when throwing from F , any point of the
envelope-parabola can be reached with the same speed u
if we adjust the throwing angle accordingly; in order to
reach the tip Q of the envelope-parabola we need to throw
straight up. Now it would be a trivial task to find the
throwing speed u from the energy conservation law if the
height h = |F Q| of the envelope-parabola were known. To
obtain h, we use the property of a parabola (see fact 9):
the sum of the distance of a point from the focus and the
point’s height is constant: b + (a − b) = h + h.
22. Due to the idea 28, together with facts 6 ,7, and 9, a
vertical ray directed at the target is reflected by the pro-
jectile’s trajectory to the focus, i.e. to the cannon. When
making use of the idea 26 we see that this projectile’s tra-
jectory is also optimal for shooting the cannon’s position
from the location of the target; hence, the projectile’s tra-
jectory reflects a vertical ray directed to the cannon to-
wards the target. If we combine these two observations we
see that a vertical ray directed to the cannon is rotated
after two reflections from the trajectory by 180◦ , which
means that the reflecting surfaces must have been perpen-
dicular to each other (showing this mathematically is left
to the reader as a simple geometrical task).
For the alternative solution, we consider the motion of two
projectiles of initial velocities ⃗v and ⃗v ′ = ⃗v + ∆⃗v as sugges-
ted above, i.e. with |∆⃗v | ≪ |⃗v |, ∆⃗v ⊥ ⃗v and |⃗v | ≈ |⃗v ′ |. In
the free-falling frame, they depart at a constant velocity
∆⃗v , the relative velocity which was given at the beginning.
Hence, at the destination, the displacement vector between
the projectiles ∆⃗r = t∆⃗v (t being the flight time) which is 27. Let us take y to be the coordinate of the upper end
perpendicular to the initial velocity ⃗v . On the other hand,
we can apply the fact 10 by considering the dependence of
the x-coordinate of the projectile at the target’s altitude
as a function of the launching angle. Hence, we conclude
18 More precisely, the closest distance will be quadratically small; meanwhile, the displacement vector t∆⃗v is linearly small, i.e. much larger.
— page 21 —
8. HINTS
that while the first projectile hits the target exactly, the
other one must also come very close to it (at the target’s
level, ⃗v gives the optimal shooting angle, and hence, ⃗v ′ is
nearly optimal): if one of them is at the target, the other
one will be there very soon18 . Recall that when the first
projectile is at the target, the other one is displaced by
t∆⃗v from it, i.e. its current velocity ⃗u ∥ t∆⃗v ⊥ ⃗v .
23. Notice that all those points which have the same speed
lie on the same distance from the instantaneous rotation
center.
24. Equate the acceleration calculated in two inertial frames:
in the lab frame, and in the frame of the wheel’s centre.
25. a) First we use the idea 35: when applying it to the left
rod we conclude that the joint’s velocity is vertical, and
when applying the idea to the right rod we conclude that
the modulus is v. Now we can apply the fact 13 to cal-
culate the horizontal projection of the joint’s acceleration.
In order to use the idea 34, we need to know the direction
of the acceleration, as well. This knowledge is obtained
if we switch to the frame moving up with constant speed
v: the joint’s speed is zero and hence, the centripetal ac-
celeration is zero. b) Notice that in the frame moving
up with constant speed v, the question b) is the same as
mirror-reflected question a).
26. First method: find the instantaneous centre of rotation by
figuring out the direction of velocities of for two points of
the cylinder: first, point A where the rope meets the cylin-
der (notice that the velocity of A as a rope’s point equals
to the velocity of A as a cylinder’s point, and apply the
idea 35 to the rope), and point B where the cylinder meets
the plate (what is the vertical component of the velocity
of the cylinder’s point B?).
Second method: make a drawing with two close positions
of the cylinder and rope: let us mark on the left position
of the rope a point P where the rope meets the cylinder,
and on its right position — point P ′ which is at the same
height as point P . Let us denote the point where the rope
is fixed to the wall by Q. Then, the rope segment QP ′
consists of a straight segment QP ′′ and a curved segment
P ′′ P ′ . However, since the displacement of the cylinder
P P ′ is small, the length of the curve QP ′′ P ′ has almost
the same length as the straight line QP ′ . While the actual
unwound length |QP | − |QP ′′ | is contributed by two rota-
tions (rotation of the cylinder and rotation of the rope),
the length difference |QP | − |QP ′ | is contributed only by
the cylinders rotation (the point P ′ is at the same relative
position on the cylinder as the point P ), and is, hence,
equal to ωR∆t. On the other hand, we can express this
length trigonometrically in terms of the cylinder’s displace-
ment P P ′ .
and x that of the lower end. Then the rod has length
l2 = x2 + y 2 ; l is constant, so its derivative must be zero.
Let us take the time derivative of the whole expression,
using the chain rule of differentiation known from math-
 8. HINTS
ematics: 0 = xẋ + y ẏ = xu + yv (a dot on top of a symbol can be counted per one stripe period.
means its time derivative). From that, we can express ii) The product of two waves can be expanded as a =
u = −vy/x = −v tan α. sin(⃗k1⃗r − ωt) sin(⃗k2⃗r) = 21 {cos[(⃗k1 − ⃗k2 )⃗r − ωt] − cos[(⃗k1 +
⃗k2 )⃗r − ωt]}. The bands where a ≈ 1 are where the “slow”
28. Go to the frame which moves with speed u and where
boundary between the queue and moving cars is station- sinusoid is close to one, cos[(⃗k1 − ⃗k2 )⃗r − ωt] ≈ 1; this is a
ary; equate the flux of cars (how many cars is passing in a sinusoidal wave which moves with speed u = ω/|⃗k1 − ⃗k2 |.
unit time) in the region of queue to its value in the region 34. Let us consider the evolution of the vector ⃗r pointing from
of moving cars. the dog to the fox; calculate the changing rates of the mod-
ulus |⃗r|, and of rx , the projection of ⃗r to the x-axis (taken
29. The intersection point P of two trails corresponds to the
parallel to the fox’s velocity). Is it possible to make such
moment when the heads of the trains met: it was carried
a linear combination of rx and |⃗r| that its time derivative
by wind to where it currently is. So, based on the speeds of
would be zero? If yes, we would obtain a new conserva-
trains we find their meeting point Q; since the speeds are
tion law which could be used to obtain immediately the
equal, this is the middle point of the segment AB connect-
answer.
ing the current positions of the train heads. The segment
AB was covered during the given time interval at the speed 35. Use the vector addition rule to draw a rectangle of velocit-
2v = 100 km/h; this value can be used as a scale for find- ies ⃗u + ⃗v = w,
⃗ where w ⃗ is the velocity of the boy relative
ing the wind speed based on the length of segment QP to the coast. Apply the idea 2: if we fix the position of
(you need to measure |AB| and |P Q|). the vector ⃗u and draw the possible triangles for different
swimming directions, we’ll see that the possible positions
30. First we conclude, based on the two collisions, that all
of the endpoint of ⃗v lie on a circle. Now it is not difficult
the bodies move on the same plane, henceforth the (x, y)-
to conclude that the optimal swimming corresponds to ⃗v
plane. According to the idea 41, we plot the trajectories
being tangent to the circle.
of the bodies in a 3D plot (as lines x = xa (t), y = ya (t);
x = xb (t), y = yb (t); x = xc (t), y = yc (t). Collisions cor- 36. Introduce coordinates x (the vertical position of the ring
respond to intersections of these lines, and intersection of O) and y (the vertical position of the ring O′ ), with the
two lines means that the two lines are coplanar (lie on a origins at A and A′ , respectively. Also introduce the rope
single plane). length L (to be eliminated later from the answer). Relate
these quantities to each other via Pythagorean theorem
31. In the free-falling frame, all the particles move with con-
(keep things squared to avoid square roots for easier ma-
stant velocities; each particle had initial velocity equal to
nipulation) and apply the idea 38.
the wheel’s velocity at the releasing point, i.e. tangential
to the wheel and equal by modulus to ΩR. Hence the en- 37. Apply idea 20: vertical motion is independent from the
semble of particle expands as a circle, the radius of which motion in the horizontal plane. The ball can escape the
can be calculated from the Pythagorean theorem. In the well if it hits the upper rim of the well. This will happen if
lab frame, the centre of the circle performs a free fall. A the period of vertical motion relates to the time between
droplet reaching the point A corresponds to the expanding two collisions with the well’s walls for the horizontal mo-
circle touching the ground. tion as a rational number (ratio of two integers).
38. Use the idea 1: switch to the frame of the wedge; determ-
32. Useful observations: each column of pixels is obtained very
ine there the acceleration of the ball; apply the rule for
fast, essentially simultaneously; so, each column of pixels
the addition of accelerations to find the acceleration of
represents a vertical cut of the real object at the respective
the ball in the lab frame. Once knowing the acceleration
position for a certain moment of time. However, different
and initial velocity, the trajectory can be also found easily.
columns correspond to different time. Each vertical curved
shape on the photo is a sequence of vertical cuts and hence 39. Apply the fact 12 to calculate the acceleration — the time
must correspond to the same blade of the propeller. At the derivative of the velocity vector. The angular speed of the
upper half of the photo, the scanning ling moves towards rotation of the velocity vector can be found using the idea
the motion of blades, and none of the blades is missed: 37.
if we number the blades by 1, 2 and 3 then the blade se- 40. Apply the idea 1: use the cone’s (free-falling) frame where
quence at the upper half must be 1, 2, 3, 1, 2, . . . . In lower the table’s corner moves upwards with the acceleration g.
half of the photo, the scanning line and blades move in the 41. Apply the idea 24 (work with polar coordinates r and φ).
same direction, and the blades are faster, hence here the Express the squared curve length increment dl2 via the
sequence is 3, 2, 1, 3 . . . . The scanning line moves appar- squared increments dφ2 and dr2 (use the Pythagorean the-
ently with a constant speed, hence the horizontal position orem); and relate dr2 to dl2 via the propagation speeds.
on the photo can be used to measure the time. Note that differential equations in the form kdx = xdy
33. i) While the grey combs moves by half of the teeth pitch, (where k is a constant) can be solved by separating vari-
a dark stripe moves to where currently there is a white ables, i.e. bringing all the x-s and y-s to the respective
stripe, i.e. by half of the dark stripe distance. Hence, the sides of equality (here, k dx
x = dy), and integrating left-
stripe speed is as many times faster as many grey teeth hand right-hand-sides of the equality (which leads here to
19 It is nicer to keep ln(x/x0 ) instead of ln(x) − ln(x0 ) as otherwise it would be difficult to check if the dimensionalities of the expressions are OK.

— page 22 —

k ln(x/x0 ) = y, where x0 is a constant emerging when
taking the indefinite integrals19 .
42. Denoting the diagonal length via d, relate the distance of
A3 from the wall to d; express the time derivative of d
in terms of v (c.f. idea 38 on Pg 12), and use this result
to find the speeds of A1 and A2 . Use the idea 1 (switch
to the frame of the centre of the largest rhombus) to find
the direction of the acceleration of B2 . Apply the idea 34
to deduce the modulus of the acceleration: switch to the
frame where A2 (or A3 ) is at rest and the point B3 moves
along a circle, and determine the centripetal acceleration
— the projection of the whole acceleration to the leg A2 B3 .
43. Apply the ideas 1 (use the frame of one of the boats), and
the idea 2. You need the angle between the line AB and
the relative velocity; note that the tangent of that angle
can be expressed quite easily.
44. Apply the idea 33: you can find the direction of velocities
for those two points of disc where the ropes are tangent to
the disc. Indeed, keep in mind the idea 35 and notice that
the upper ends of the ropes have zero velocities (where in
contact, the rope and disc points have equal velocities).
45. Apply the idea 20: the motion along each of the boards are
independent. Use the fact 4 (the free fall acceleration com-
ponents will be g cos α and g sin α); calculate the jumping
periods for each of the motions.
46. Apply the idea 21 to conclude that the speed of the block
remains constant. Express the angular speed ω of the un-
wound part of the rope in terms of its current length l; note
that ω is also equal to the angular speed of the point P
where the rope is tangent to the cylinder; relate this speed
dl
to the rate at which the rope is unwound, dt . Apply the
2
provided formula to conclude that d(ldt ) remains constant
during unwinding and winding. Don’t forget that there
is also a period when the rope is fully unwound and the
block draws a semicircle.
47. Apply idea 35 to determine the projections of the box’s
velocity ⃗v to the directions of the ropes. Using the idea 16
one can conclude that the quadrilateral formed by the vec-
tor ⃗v as a diagonal and the projections of ⃗v as its two sides
is a cyclic one whereas ⃗v is the diameter of its circumcircle.
Apply the cosine theorem to determine the length of the
other diagonal, and the sine theorem to determine |⃗v |, the 50. It can be seen that the red dots are repeated at the same
diameter of the circumcircle.
48. First approach: apply the idea 1: use the inertial frame
which moves with the initial velocity of the boy; while the
acceleration of the boy is fixed by modulus, the direction
can be adjusted as needed. Since in this frame, the ini-
tial velocity is zero, the optimizations task becomes quite
trivial. Second approach: study the evolution of the boy’s
velocity vector ⃗v in (vx , vy )-plane: acceleration ⃗a = d⃗
v
dt is
constant by modulus, hence the endpoint of ⃗v moves with
a constant speed |⃗a| from its initial position (0, v) to the 51. i) v
final position (v, 0).
49. There are at least three different solutions; the simplest
one is based on the idea 30, and is quite similar to the geo-
metric solution of the problem 21. All the solutions start
— page 23 —
8. HINTS
in the same way as the ones of problem 30 by (a) showing
that the optimal trajectory needs to touch the sphere at
a certain point P before hitting the sphere at its top T ,
and (b) applying the idea 26 to consider throwing from T
instead of throwing from an unknown point at the ground
level.
The first method makes use of the geometric properties of
the envelope parabola (see ideas 28 and 30, and facts 6, 7
and 9): |T P | + |P A| = 2|T Q| + R, where P A is a vertical
line, A lies at the same height as the sphere’s centre O,
and Q is the tip of the envelope parabola. As the shoot-
ing speed is defined by |T Q| (see the hints of problem 21
for more details), we only need calculate |T P | and |P A|.
Since T P is a line coming from the focus, it is reflected
by the parabola at P to a vertical line; since parabola is
tangent to the sphere, it is also reflected by the sphere to
a vertical line which means that ∠T P O = ∠OP A. Since
T O ∥ P A we can conclude that ∠T OP = ∠OP A, and as
△T OP is isosceles, ∠OT P = ∠T P O; together with equal-
ity ∠T P O + ∠T OP + ∠OT P = 180◦ we conclude that
∠OT P = 60◦ , which gives us immediately |T P | = R and
|P A| = R2 .
The second method makes use of the expression for the
envelope parabola (idea 28) with a focus at T : write down
equation for finding the intersection points of the sphere
and the parabola — this will be a biquadratic equation.
According to the fact 7, in the optimal case the parabolic
trajectory, envelope parabola, and sphere are tangent to
each other at the same point; hence, at this optimal case
there are exactly two symmetric solutions to the biquad-
ratic equation (if the launching speed is smaller than op-
timal then there are 4 solutions, and if it is larger then
there are no solutions). Apply idea 29 to find the speed.
The third solution makes use of the result of the problem
22 and idea 24: if we use φ = ∠T OP as the parameter, we
know that from the point T , the optimal shooting angle
must be equal to 90◦ − φ. Hence, we can write down the
conditions that the trajectory goes through P and the fi-
nal, and initial and final velocities are perpendicular (idea
15). This gives us three equations containing φ, touching
time t, and initial speed v as unknowns. It appears that
this system of equations simplifies nicely.
places of the blue cycloid for each of its periods, and
there are three dots per period; according to the idea 6,
we’ll make use of this fact (in generic case it considerably
longer calculations would be needed) to conclude that the
disk’s rotation period is three times longer than the flash-
ing period, i.e. 0.3 s. The distance travelled corresponds to
the period length of ca 4 grid units; the value of a grid unit
can be found by noticing that the cycloid height equals to
2a.
dist is always negative, hence the overall, the distance
decreased, hence the cars must have started from different
cities. O could not have been the starting point of a car
because in that case the final distance would have been the
same or larger than at the beginning. Now we just need
 8. HINTS
to consider three possibilities, out of which two are easily 57. Apply the idea 39: consider the number of cars passing a
excluded based on the behaviour for 0 < t < av . given point in unit time (the car frequency), and equate
ii) Use the idea 18. the sum of car frequencies of the two merging lanes with
iii) and iv) Calculate relative velocity according to the vec- the frequency for the lane C; this will give you the speed
tor subtraction rule and project it to the line connecting on the lane C (the rest of the calculations are straightfor-
the cars for different positions on their path. ward).
52. Approach the same way as in the case of problem 3: use 58. Apply the idea 39: introduce the density ρ of water in air;
the frame which rotates with the angular speed ω2 to obtain then, the water flux (mass per unit time) is given by Sρv,
the angular speed of the line OP , P being the intersection where S is the cross-sectional area, and v is the speed of
point. Consider isosceles triangle QOP (where Q is the the man relative to the rain droplets. For a rectangular
centre of the blue ring) to obtain the angular speed of the man we calculate the total flux in two parts — water fall-
line QP . ing on the horizontal surface due to the vertical velocity of
53. Note that for a spike, such a point P appears sharp for the rain droplets, and water falling on the vertical surface
which velocity is parallel to the spike. Apply the idea 33: due to the motion of the man. For a spherical man, we
the instantaneous rotation centre is the lowest point G of need to calculate the relative speed by using the triangle
the wheel (as it is contact with the ground and has thus rule for adding the velocities. In either case we obtain
zero speed). Thus, GP must be perpendicular to the spike; a system of equations from which the surface areas and
recall now the idea 16. wind’s speed can be eliminated.

54. i) Use the fact that orange dot is missing; the exposure 59. The length of the trails is defined by the time interval
time must be appropriate for capturing exactly four dots.
ii) Use the idea 1: in the frame of the disk’s centre, the
⃗
displacement vector d between neighbouring flashes has
always the same modulus d = 2R sin(ωτ /2), and neigh-
bouring displacement vectors are always rotated by the
same angle ωτ . In the lab frame, additional constant dis-
placement vector ⃗v τ is to be added due to the translational
′
motion of the frame: d⃗ = d⃗ + ⃗v τ . Because of that, if we
bring all the displacement vectors to such positions that
their starting points coincide, the endpoints will lie on a
circle of radius d. So, we redraw the displacement vectors
⃗ rg,
br, ⃗ and gy, ⃗ and draw the circumcircle of the triangle
formed by the endpoints of the vectors; from that figure
we can measure both the rotation angle ωτ (for finding
during which the droplet’s image remains within the gap
between the curtains. This, in turn, is inversely propor-
tional to that component of the image’s relative velocity
which is perpendicular to the curtain’s edge. In one case,
the velocity of the curtains ⃗v and the velocity of the
droplet’s image ⃗u are parallel, in the other case — anti-
parallel, and in the third case — perpendicular. In the
antiparallel case there are two possibilities: we don’t know
which is faster, the curtain or the image. While formally
we do have three unknown quantities, v, u, and the gap’s
width d, these enter the equations only in two combina-
tions, v/d and u/d, i.e. we have essentially only two un-
known parameters. Therefore, the expressions for l1 and
l 2 can be solved with respect to v/d and u/d, which are

ω), constant displacement a = vτ (for finding v), and the further used to calculate the value of l3 .
circle’s radius d = 2R sin(ωτ /2) (for finding R). 60. For small circles, the radii should form a sequence rn =
g 2 2
2
55. For the parabolic shape y = kx of the jet, the factor k can 2π n τ (with√n = 1, 2, . . .); for larger rings we should have
rn+1 − rn = hgτ . By taking measurements from the fig-
be determined from the figure; this relates the unknown g 2 2 √
grid unit d, the initial speed v of the jet, and g = 9.81 m/s 2 ure, we can determine both 2π n τ and hgτ as a product
to each other. The idea 39 relates the flow rate π4 d2 v to the of L and a certain numerical value; this gives us two equa-
vessel filling rate; we have two equations and two unknown tions from where we can eliminate τ and express h.
parameters, so the system of equations can be solved. 61. Apply the idea 27 and construct the wave front behind the
56. Using the idea 1 we can conclude that all the litter must lie boat. Note that locally, the front propagates perpendicu-
on the same line with that boat from which these fell: this larly to itself, so that the first to arrive to the coast is that
allows us to conclude, which litter corresponds to which part of the wave front which was initially (i.e. at the point
boat. Using the idea 40 we mark the point where the boats of creation behind the boat) parallel to the coast. We can
met — the intersection point of two trails. The distance of also conclude by using the Huygens principle that close
the boats from that point is proportional to the speed of to the boat, the wavefront forms angle arcsin uv with the
the boats. The distance of A from the point where the trail boat’s trajectory.
of the boat intersects with the coastline gives the distance 62. Apply the idea 1: use the water’s frame of reference. We
carried by the water flow (the frame displacement); from assume that the water moves as a whole, across the whole
that moment when the boats met, the frame displacement depth, hence in the water’s frame, the wave speed is the
was smaller, and can be found geometrically from similar same for all the propagation directions. Thus we can con-
triangles formed by the following lines: the trail, the line clude that in the water’s frame, the boat’s velocity is par-
connecting the boat with its starting point, coastline, and allel to the bisector of the angle formed by the waves. We
a parallel line to the coast, drawn through the intersec- know that in lab frame, the boat moves parallel to the
tion point of the trails. Analogous construction of similar line AB, hence we can deduce the water speed from the
triangles for the other boat will complete the task. triangle of velocities using the known value of the boat’s
— page 24 —
 9. ANSWERS
√
speed. From the Huygens principle we know that the angle 3. u = v/2 1 − (a/2r)2 .
between the bisector and wave front is arcsin wv , hence we 4. Ascending velocity 4.85 m/s; h = 2000 m; wind velocity
can measure the angle to find w and calculate the depth u = 2.8 m/s.
h = w2 /g.
5. Straight line.
63. i) Use the idea 1: in the air’s frame, the trail is parallel to √
6. 2/ 5 m/s
the relative velocity of the tractor. Hence, ⃗vtractor − ⃗vwind
s L
is parallel to the trail; these two vectors form a triangle 7. t = u + v cos α = 1.8 s
which can be easily constructed (we know the directions 8. u = v/2 cos α; β = 180◦ − 2α
of its two sides, and one length), from where the length of 9. v(t) ≡ 2ωa.
⃗vwind can be measured.
10. 18.75 m
ii) Following the idea 40 we’d like to make use of the in-
tersection point of the trails. However, the tractors didn’t 11. 39 m
√
meet. So, we need to draw one more trail which would 12. t = (t21 − t22 )/2
have been observed if the tractors were meeting at the
13. 8d tan α
road crossing (shift the left tractor together with its trail
◦
appropriately). Then, the wind speed can be related im- 14. α = arcsin 0.5 = 30
mediately to the distance of the trails intersection point 15. 6.7 s
from the road crossing (you just need to compare it with 16. e100 − 1 seconds
the distance of the tractors from the crossing).
17. x = cos α(l − h tan β) and t = h cos v
β
+ l sin α
u , where
64. The main idea is to consider motion in a free-falling frame. β = arcsin(v sin α/u), if tan β < l/h; otherwise x = 0 and
√
More specifically, in the free-falling frame, the balls move t = h2 + l2 /v.
with constant velocities, hence are at the base vertices B ( w ) ( )
18. x = a u cos − tan α , where α = arcsin u
w+v .
and C of an isosceles triangle. In that frame, the throwing α
2
point is the top vertex A of the isosceles triangle. The 19. z ≤ v0 − gx2
2g 2v02
current position of the point A in the lab frame can be
20.
found by constructing the isosceles triangle. The point A √
has fallen in the lab frame from point P with free fall ac- 21. vmin = g(a + b + c)
celeration, and the falling time can be found by measuring 22.
the falling distance |AP |. The velocities can be found by
23. Concentric arcs of radii l1 and l2 , where l1 and l2 are the
measuring the flight distance |AB| (or |AC|) in the falling
distances of the instantaneous rotation centre from the top and
frame.
bottom plates, respectively.
65. Consider the motion of the boat in the air’s frame of refer- 24. r = 4R
ence. More specifically, based on wind data, calculate the 2
√ 2
√
displacement vector (i.e. the displacement in east-west and 25. a1 = v0 / 3l; a2 = v0 / 3l
the displacement in north-south directions); knowing the 26. v0 = v/(1 + sin α)
displacement vector in lab frame, calculate the displace- 27. u = −v tan α, a = v23 .
l cos α
ment vector of air, and based on that — its speed.
28. u = vτ /l−1 ≈ 3.4 m/s.
v

66. To begin with, you need to know that (as it follows from
29. u ≈ 15 km/h ≈ 4.2 m/s, α ≈ 27◦ .
the energy and momentum conservation laws) if two ab-
solutely elastic balls of equal mass collide centrally while 30. yes √
( )
moving along the same line, they will exchange their velo- 31. t = 2 R 1 + RΩ2 and α = arctan(Ωt).
g g
cities: the ball A departs with the initial velocity of the
ball B and vice versa. 32. counterclockwise; 3; 15 Hz
Next, there are two main ideas which need to be applied: 33. 7 cm/s; √ 1−2
cm/s
≈ 5.7 cm/s.
α +1/49
first, consider the motion of balls in a free-sliding frame
34. l/2.
where all the balls move with constant speeds. The second √
one is the idea 41: add a time axis and study the graphs. 35. L 3.
( ) 2
You’ll notice that the x−t graph consists of N intersecting 36. v0 = v cos1 α − 1 , a = vb tan3 α.
√
lines with the intersection points corresponding to colli- 37. nv0 2H/g = mR cos α with integer n and m
sions. The number of intersections is found as the number
38. a = 2a0 sin(α/2)
of different possibilities of picking 2 lines out of the set of
N lines. 39. a = v1 v2 /l
√
40. v ≥ r2 g/2h
√
9 ANSWERS 41. a logarithmic spiral ( vc )2 − 1 ln rr0 = φ.
√ √
1. vr = 4 km/h, vb = 16 km/h. 42. v0 /6, v0 /2, v0 5/6, 2v02 /36l
2. 4 km. 43. l sin ϕ, where tan ϕ = |v1 sin β −v2 sin α|/|v2 cos β −v1 cos α|
— page 25 —
 9. ANSWERS
44. ωR/ cos(α/2)
√
45. a tan α/b
46. 2π 2 kr(2k + 1)/v
√
47. v12 + v22 − 2v1 v2 cos α/ sin α
√
48. 2v/µg; a parabola.
√
49. 4.5gR
50. 20 cm/s.
√
51. i) from A and C to the capital; ii) a(2 − 2); iii) for A-B: 0
√
until turning point, −2v henceforth; for B-C: constantly −v 2;
iv) speed drops briefly down to −2v.
52. vmin = vmax = ωr.
53. A circle (ring) touching the ground and passing through
the wheel’s centre.
54. blue pulse was the first one; 300 ms < T < 500 ms;
v ≈ 65 cm/s, ω ≈ 23 rad/s, R ≈ 5 cm
55. We define k ≈ 0.014 via the shape of the jet y = kx2 using
( )
the units of the figure grid; then d = 32V 2 k/π 2 t2 g ≈ 1 mm.
56. Let a line s1 connect the triangle (henceforth T ) with two
lower stars, and a line s2 — the square (S) with the remaining
star; draw a horizontal line s3 through the intersection point
of s1 and s2 ; mark the intersection point Q of the lines s3 and
T A; the starting point B of the other boat is the intersection
point of the upper coast with the line SQ.
57. 35 min
√
58. 1 min; 1.5 min ≈ 73 s.
59. 150 or 600 pixels.
60. h ≈ 3.2L.
√
61. R
v arccos uv + L
u − R u−2 − v −2 .
62. v ≈ 1.8 m/s; h ≈ 2.0 m
63. i) vwind ≈ 13 km/h; ii) vwind ≈ 21 km/h
64. v ≈ 20 m/s
65. ≈ 12 m/s
N (N −1)
66. 2 .

— page 26 —

PROBLEMS ON MECHANICS 1
Jaan Kalda Version: 1.2β , 11th November 2017
Partially translated by: S. Ainsaar, T. Pungas, S. Zavjalov
1 INTRODUCTION
This booklet is a sequel to a similar collection of problems on
kinematics and has two main parts: Section 3 — Statics and
Section 4 — Dynamics; Section 5 contains revision problems.
The main aim of this collection of problems is to present the
most important solving ideas; using these, one can solve most
(> 95%) of olympiad problems on mechanics. Usually a prob-
lem is stated first, and is followed by some relevant ideas and
suggestions (letter ‘K’ in front of the number of an idea refers
to the correspondingly numbered idea in the kinematics book-
let; cross-linking works if the kinematics booklet is stored in
the same folder as the mechanics one). The answers to the
problems are listed at the end of the booklet (Section 7). They
are preceded by quite detailed hints (Section 6), but it is recom-
mended that you use the hints only as a last resort, after your
very best efforts at tackling a problem fail (still, once you have
solved a problem successfully by yourself, it is useful to check
if your approach was the same as suggested by the hints).
The guiding principle of this booklet argues that almost all
olympiad problems are “variations” on a specific set of topics
— the solutions follow from corresponding solution ideas. Usu-
ally it is not very hard to recognize the right idea for a given
problem, having studied enough solution ideas. Discovering all
the necessary ideas during the actual solving would certainly
show much more creativity and offer a greater joy, but the skill
of conceiving ideas is unfortunately difficult (or even imprac-
ticable) to learn or teach. Moreover, it may take a long time
to reach a new idea, and those relying on trying it during an
olympiad would be in disadvantage in comparison to those who
have mastered the ideas.
In science as a whole, solution ideas play a similar role as
in olympiads: most scientific papers apply and combine known
ideas for solving new (or worse, old) problems, at best develop-
ing and generalising the ideas. Genuinely new good ideas occur
extremely rarely and many of them are later known as master-
pieces of science. However, as the whole repertoire of scientific
ideas encompasses immensely more than mere mechanics, it
is not so easy to remember and utilise them in right places.
The respective skill is highly valued; an especial achievement
would be employing a well-known idea in an unconventional
(unexpected, novel) situation.
In addition to ideas, the booklet also presents “facts” and
“methods”. The distinction is largely arbitrary, some ideas
could have been called methods or facts and vice versa; at-
tempt has been made to pursue the following categorization.
Facts are fundamental or particular findings, the knowledge
of which can be useful or necessary for problem solving, but
which are not formulated as ready recipes. While in theory,
all problems can be solved starting from the first principles
(the fundamental “facts”), but typically such a “brute force”
approach leads to long and sometimes unrealistically complex
calculations; the “ideas” are recipes of how to solve problems
1 As compared with v. 1.0, introductory theory sections are added
2 I.Newton 1687
3 which will also serve as the mass unit
4 or a cylinder of 39.17 mm height and diameter, made of platinum-iridium alloy Pt-10Ir (the official SI definition).
— page 1 —
1. INTRODUCTION
more easily. The “methods” are powerful “ideas” of particu-
larly wide applicability.
Several sources have been used for the problems: Estonian
olympiads’ regional and national rounds, Estonian-Finnish
Olympiads, International Physics Olympiads, journal “Kvant”,
Russian and Soviet Union’s olympiads; some problems have
been modified (either easier or tougher), some are “folklore”
(origins unknown).
Similarly to the kinematics booklet, problems are classified
as being simple , normal , and difficult : the problem num-
bers are coloured according to this colour code (keep in mind
that difficulty is a subjective category!).
Finally, don’t despair if there are some things (or some sec-
tions) which you are not able to understand for the time being:
just forward to the next topic or next problem; you can return
to those parts which you didn’t understand later.
2 FIRST LAWS — THEORETICAL BASIS
Those who are familiar with the basic laws of mechanics can
skip this Section (though, you can still read it, you may get
some new insight), and turn to Section 3. In fact, it is ex-
pected that majority of readers can do this because almost
all the physics courses start with mechanics, and it is unlikely
that someone is drawn to such a booklet aiming to develop
advanced problem-solving skills without any prior experience
in physics. However, attempts have been made to keep this
series of study guides self-contained; this explains the inclusion
of the current chapter. Still, the presentation in this Section is
highly compressed and in some places involves such mathemat-
ical formalism which may seem intimidating for beginners (e.g.
∑
usage of the summation symbol and differentials), therefore
it is not an easy reading. If you find this section to be too
difficult to start with, take a high-school mechanics textbook
and turn here to the Section “Statics”.
2.1 Postulates of classical mechanics
Classical mechanics, the topic of this booklet, is a science based
entirely on the three Newton’s laws2 , formulated here as “facts”.
fact 1: (Newton’s 1st law.) While the motion of bodies de-
pends on the reference frame (e.g. a body which moves with a
constant velocity in one frame moves with an acceleration in
another frame if the relative acceleration of the frames is non-
zero), there exist so-called inertial reference frames where the
facts 2–5 are valid for all the bodies.
fact 2: (Newton’s 2nd law.) In an inertial frame of refer-
ence, a non-zero acceleration ⃗a of a body is always caused by
an external influence; each body can be characterized by an
inertial mass m (in what follows the adjective “inertial” will
be dropped), and each influence can be characterized by a vec-
torial quantity F⃗ , henceforth referred to as the force, so that
equality F⃗ = m⃗a is valid for any influence-body pair.
 2.2 Basic rules derived from the postulates
Please note that once we establish an etalon for the mass3 , e.g. 1–5 as simple as possible, let us assume that the fact 2 is valid
define 1 kg as the mass of one cubic decimeter of water4 , the for so-called point masses, i.e. for very small bodies the dimen-
fact 2 serves us also as the definitions of the mass of a body, sions of which are much smaller than the characteristic travel
and of the magnitude of a force. Indeed, if we have a fixed ref- distances; then, the position of a point mass is described by
erence force which is (a) guaranteed to have always the same a single point which has unambiguously defined velocity and
magnitude, and (b) can be applied to an arbitrary body (e.g. acceleration. We can generalize the fact 2 to real finite-sized-
a spring deformed by a given amount) then we can define the bodies by dividing these fictitiously into tiny pieces, each of
mass of any other body in kilograms numerically equal to the which can be treated as a point mass.
ratio of its acceleration to the acceleration of the etalon when To begin with, one can derive (see appendix 1) the generic
the both bodies are subject to the reference force. Newton’s formulation of the Newton’s 2nd law.
2nd law is valid if this is a self-consistent definition, i.e. if the fact 6: (Momentum conservation law/generalized Newton’s
obtained mass is independent of what reference force was used. 2nd law.) For the net momentum P ⃗ = ∑ mi⃗vi of a system of
i
Similarly, the magnitude of any force in Newtons (denoted as point masses5 ,
N ≡ kg m/s2 ) can be defined to be equal to the product of d ⃗
P = F,⃗ (1)
the mass and the acceleration of a body subject to that force; dt
Newton’s 2 law is valid and this definition is self-consistent where F is the net force (the sum of external forces) acting on
nd ⃗
if the result is independent of which test body was used. the system. In particular, the net momentum is conserved
To sum up: the Newton’s 2 law F = m⃗a serves us both P
nd ⃗ ( ⃗ =const) if F ⃗ = 0.
as the definition of the mass of a body (assuming that we have By substituting P ⃗ = ∑ mi⃗vi = ∑ mi d⃗ri = d ∑ mi⃗ri
i i dt dt i
chosen a mass etalon), and the force of an interaction; the law (where ⃗ri denotes the position vector of the i-th point mass),
ensures that these are self-consistent definitions: the mass of we can rewrite Eq (1) as
a body and the magnitude of a force are independent of the d2 r⃗C
M = F,
⃗
measurement procedure. dt2
where ∑
fact 3: Forces are additive as vector quantities: if there are mi⃗ri
⃗rC = ∑i (2)
⃗
many forces Fi (i = 1 . . . n) acting on a body of mass m then i mi
∑ is called the centre of mass. This result clarifies: in the case
the fact 2 remains valid with F⃗ = i F⃗i .
∑ of macroscopic bodies, the fact 2 remains valid if we use the
The vector sum i F⃗i can be calculated using either the tri- acceleration of the centre of mass.
angle/parallelogram rule, or component-wise arithmetic addi- According to the Newton’s 2nd law, once we know how the
∑
tion: Fx = i F⃗ix , where an index x denotes the x-component interaction forces between bodies depend on the inter-body dis-
(projection onto the x-axis) of a vector; similar expressions can tances and on the velocities, we can (in theory) calculate how
be written for the y- and z-axis. the system will evolve in time (such systems are refereed to as
fact 4: Masses are additive as scalar quantities: if a body determinstic systems). Indeed, we know the accelerations of all
the bodies, and hence, we can determine the velocities and pos-
is made up of smaller parts of masses mj (j = 1 . . . m) then
itions after a small time increment: if the time increment ∆t is
the total mass of the compound body equals to the sum of the
∑ small enough, the changes in accelerations ∆⃗a can be neglected,
masses of its components, m = j mj .
which means that the new velocity for the i-th body will be
⃗vi′ = ⃗vi + ⃗ai ∆t, and the new position vector ⃗ri′ = ⃗ri + ⃗vi ∆t; the
fact 5: (Newton’s 3rd law.) If a body A exerts a force F⃗ on whole temporal dependences ⃗vi (t) and ⃗ri (t) (with i = 1 . . . N
a body B then the body B exerts simultaneously the body A
where N is the number of bodies) can be obtained by advancing
with an equal in modulus and antiparallel force −F⃗ .
in time step-by-step. In mathematical terms, this is a numer-
ical integration of a system of ordinary differential equations:
2.2 Basic rules derived from the postulates the second time derivatives of coordinates ẍi , ÿi , and z̈i are
The facts 1–5 can be considered to be the postulates of classical expressed in terms of the coordinates xi , yi , zi , and the first
(Newtonian) mechanics, confirmed by experiments. All the sub- derivatives ẋi , ẏi , żi 6 . While in principle, these calculations
sequent “facts”, theorems, etc. can be derived mathematically can be always made, at least numerically and assuming that
using these postulates. we have enough computational power, in practice the mathem-
Thus far we have used a vague concept of the acceleration of atical task can be very difficult7 . Apart from the facts 1–5,
a body. Everything is fine as long as a body moves translation- the Newtonian mechanics is a collection of recipes for easier
ally, i.e. so that all its points have the same acceleration vector. solution of these differential equations.
However, if a body has a considerable size and rotates then Among such recipes, finding and applying conservation laws
different points have different accelerations, so that we need to has a central role. This is because according to what has been
clarify, the acceleration of which point needs to be used. In said above, the evolution of mechanical systems is described
order to overcome this difficulty and keep our set of postulates by a system of differential equations, and each conservation
5 such as rigid bodies in which case the point masses are bound by inter-molecular forces together into a macroscopic body
6 Here we assume that the force depends only on the coordinates and velocities of the bodies; with the exception of the Abraham-Lorentz force
(accounting for the cyclotron radiation), this is always satisfied.
7 Also, there is the issue of possibly chaotic behavour when in many-body-systems, small differences in initial conditions lead to exponentially

growing differences — in the same way as it is impossible to put a sharp pencil vertically standig on its tip onto a flat surface.

— page 2 —
 2.2 Basic rules derived from the postulates
law reduces the order of that system by one; this makes the constant, hence the sum of all the forces, as well as the sum
mathematical task much simpler. The conservation laws can of all the torques acting on a body must be zero; this applies
be derived mathematically from the Newton’s laws; while it is also to any fictitious part of a body. While there are statics
definitely useful to know how it is done, majority of mechanics problems which study deformable bodies (which change shape
problems can be solved without being familiar with this pro- when forces are applied to it), there is an important idealiza-
cedure. Because of that, the conservation laws are derived in tion of rigid body: a body which preserves its shape under any
Appendices 1,2, and 3; here we just provide the formulations. (not-too-large) forces.
We have already dealt with the momentum conservation law While for the Newton’s 2nd law [Eq. (1)], and for the static
(see fact 6), so we can proceed to the next one. force balance condition, it doesn’t matter where the force is
applied to, in the case of angular momentum [Eq. (3)] and for
fact 7: (Angular momentum conservation law.) For the net
angular momentum L ⃗ = ∑ mi⃗ri × ⃗vi of a system of bodies, the static torque balance condition, it becomes important. In
i
classical mechanics, the forces are divided into contact forces
d ⃗ ⃗
L=T, (3) which are applied at the contact point of two bodies (elasti-
dt
where city forces in its various forms, such as normal and friction
∑
T =
⃗ ⃗ri × Fi
⃗ forces, see below), and body forces which are applied to every
i point of the solid body (such as gravity and electrostatic forces).
⃗
is the net torque acting on the system; here Fi stands for the The application point of contact forces is obviously the contact
net force acting on the i-th point mass. In particular, the net point; in the case of body forces, the torque can be calculated
angular momentum of the system is conserved if T⃗ = 0. by dividing the entire body (system of bodies) into so small
parts (point masses) and by integrating the torques applied to
Eq. (3) is derived in appendix 2, and can be considered to be
each of these. It is easy to see that with the total body force
the Newton’s 2nd law for rotational motion of bodies.
(i.e. the sum of all the body forces applied to different parts
In three-dimensional geometry, calculating the vector of the body) F ⃗ and total torque T⃗ applied to a body, one can
products to determine the net torque and angular momentum always find such a radius vector ⃗r that T⃗ = ⃗r × F, ⃗ i.e. although
may be quite difficult. Luckily, most of the Olympiad prob- the body forces are applied to each point of the body, the net
lems involve two-dimensional geometry: velocities, momenta, effect is as if the net force F ⃗ were to applied to a certain ef-
and radius vectors lie in the x − y-plane, and vector products fective application centre; in some cases, there are simple rules
(torques and angular momenta) are parallel to the z-axis, i.e. for finding such effective application centres, e.g. in the case
we can consider L ⃗ = ⃗r × p⃗ and T⃗ = ⃗r × F⃗ as scalars, char-
of an homogeneous gravity field, it appears to be the centre of
acterized by their projection to the z-axis (in what follows de- mass.8 .
noted as L and T , respectively). According to the definition At the microscopic level of quantum mechanics, such a di-
of the vector product, the sign of such torque is positive if the vision of forces becomes meaningless, because on the one hand,
rotation from the vector ⃗r to the vector F⃗ corresponds to a fields which mediate body forces are also material things and
clockwise motion, and negative otherwise. Thus we can write in this sense, all the forces are contact forces. On the other
T = |⃗r| × |F⃗ | sin α, where α is the angle between the radius hand, classical contact forces are also mediated at the micro-
vector and the force and can be either positive (rotation from scopic level via fields so that in a certain sense, all the forces
⃗r to F⃗ is clockwise) or negative. We can introduce the arm of are body forces. Regardless, at the macroscopic level such a
the force h = |⃗r| sin α (see figure), in which case division still remains helpful.
T = |F⃗ |h;
fact 8: (Energy conservation law; for more details, see ap-
similarly we can use the tangential component of the force
pendix 3.) If we define the kinetic energy for a system of point
Ft = |F⃗ | sin α and obtain
masses (or translationally moving rigid bodies) as
T = |⃗r|Ft . 1∑
K= mi⃗vi2 ,
Similar procedure can be applied to the angular momenta: 2 i
L = |⃗r||⃗p| sin α = h|⃗ p| = |⃗r|pt . and the total work done by all the forces during infinitely small
displacements d⃗ri of the point masses as
∑
area = dW = F⃗i · d⃗ri
O torque
origin α ~
Ft i
(axis)
r~ then the change of the kinetic energy equals to the total work
F~
h done by all the forces,
force application point dK = dW ;
⃗
here, Fi denotes the total force acting on the i-th point mass.
The discipline of statics studies equilibria of bodies, i.e. con- The work done by so-called conservative forces depends only on
ditions when there is an inertial frame of reference where a the initial and final states of the system (i.e. on the positions of
body remains motionless. It is clear that both the momentum the point masses), and not along which trajectories the point
and angular momentum of a body at equilibrium needs to be masses moved. This means that the work done by conservative
8 Similarly, if two bodies make contact over a finite-sized area (rather than at just few contact points), we would need to find the total torque by

integrating over the contact area, and one can always find the effective application point of these forces
9 dΠ ≡ Π(⃗ r1 + d⃗r1 , ⃗ r1 , . . .) − Π(⃗
r2 + d⃗ r1 , ⃗
r2 , . . .)

— page 3 —
 2.3 Basic forces
forces can be expressed as the decrease of a certain function of From the Newton’s 3rd law we know that each force is
state Π(⃗r1 , ⃗r2 . . .) which is referred to as the potential energy; caused by some other body: a gravity force acting on a body
for infinitesimal displacements we can write dWcons = −dΠ9 . A needs to be caused by a body B. We also know that the
Therefore, if we define the full mechanical energy as E = K +Π gravity force is caused by and is proportional to the mass of a
then body, and apparently this applies both to the body A, and body
dE = dW ′ , B. Hence, the force needs to be proportional to the product of
′
where dW stands for the work done by the non-conservative masses, F = cmA mB , where the coefficient of proportionality c
forces; if there are no such forces then dE = 0, hence can be a function of distance. It appears that c is inversely pro-
E = Π + K = const. (4) portional to the squared distance, c = G/r2 ; let us consider this
as an experimental finding. Here G ≈ 6.67 × 10−11 m3 kg−1 s−2
Here few comments are needed. First, while the momentum
is called the gravitational constant. Now it is easy to guess
of a body is the momentum of its centre of mass, the kinetic
that this force must be parallel to the only preferred direction
energy of a composite body is not just the kinetic energy of its
for a system of two point masses, the line connecting the two
centre of mass: the kinetic energy in the frame of the centre of
points. This is indeed the case; furthermore, the gravitational
mass (“CM-frame”) needs to be added, as well,
1∑
force appears to be an attractive force.
1 2
K = M vC + mi ⃗u2i ,
2 2 i fact 9: The gravitational force acting on the i-th point mass
where where ⃗ui = ⃗vi − ⃗vC is the velocity of the i-th point mass due to the j-th point mass can be expressed as
in the CM-frame 10 . How to calculate the kinetic energy in Gmi mj
F⃗i = r̂ij 2 , (5)
the frame of the centre of mass for rotating rigid bodies will be rij
discussed somewhat later. where r̂ij = (⃗rj −⃗ri )/|⃗rj −⃗ri | stands for the unit vector pointing
Finally, let us notice that forces depending on the velocities from the i-th body to the j-th body and rij = |⃗rj − ⃗ri |11 . The
(e.g. friction forces) and/or on time (e.g. normal force exerted presence of a third body does affect the validity of this law,
by a moving wall), cannot be conservative because the work i.e. the superposition principle holds: total gravitational force
done by such forces along a path depends clearly on how fast can be found by summing up the contributions from all the
the bodies move. An exception is provided by those velocity- gravitating bodies according to Eq. (5).12 Eq. (5) remains also
dependent forces which are always perpendicular to the velo- valid when the gravitationally interacting bodies have spheric-
city (e.g. for the Lorentz force and normal force) and for which ally symmetric mass distribution — in that case, ⃗ri and ⃗rj point
dW = F⃗ · d⃗r = F⃗ · ⃗v dt ≡ 0. to the respective centres of symmetry (which coincide with the
centres of mass) 13 NB! in the case of arbitrarily shaped bodies,
2.3 Basic forces using the centres of mass would be incorrect; the force needs
Gravity. Now, let us consider the case of a gravity field in to be calculated by dividing the bodies into point masses and
more details; it can be described by the free fall acceleration taking integral.
vector ⃗g . From the “Kinematics” we know that then all the
For the gravitational pull of the Earth, we can typically ap-
bodies move with the acceleration ⃗g ; then, according to the
proximate rij with the radius of the Earth RE , so that
Newton’s 2nd law, this should be caused by a force
GME
F⃗ = m⃗g . F⃗ = m⃗g , ⃗g = ẑ 2 ≈ ẑ · 9.81 m/s2 , (6)
RE
This is called the gravity force. The fact that in a given gravity where ẑ stands for a downwards pointing unit vector and M
E
field, the gravity force is proportional to the mass of a body is denotes the Earth’s mass.
to be considered as an experimental finding. Let us recall that
Note that the force due to a homogeneous gravity field ⃗g is
the mass is introduced via the Newton’s 2nd law and describes
applied effectively to a centre of mass of a body, regardless of
the inertia of a body, i.e. the capability of a body to retain its
its shape. Indeed, the torque of the gravity force is calculated
velocity; because of this we can call it also the inertial mass.
then as ( )
Here, however, the mass enters a completely different law: the
∑ ∑
gravity force is proportional to the mass. It is easy to imagine T = ⃗ri × ⃗g mi = ⃗ri mi × ⃗g = ⃗rC × ⃗g M,
that the gravity force is defined by another characteristic of
∑ i i
a body, let us call it the gravitational mass, unrelated to the where M = i mi .
inertial mass. Experiments show that the gravitational mass Gravity force is a conservative force because for any pair
is always equal to the inertial mass and thus we can drop the of point masses, the force is directed along the line connecting
adjectives “gravitational” and “inertial”. As a matter of fact, these point masses and depends only on the distance between
the equivalence of inertial and gravitational mass has a great these (cf. appendix 3). The work done by a gravity force
significance in physics and represents the main postulate and F⃗ = m⃗g due to a homogeneous gravity field can be expressed
cornerstone of the general theory of relativity. as dA = m⃗g · d⃗r, hence Π = −m⃗g · d⃗r; upon integration we
∑ ∑
10 Indeed, we divide the body into point masses and write K = 1 mi (⃗vi − ⃗vC + ⃗vC )2 = 21 ui + ⃗vC )2 ; now we can open the braces and
mi (⃗
1
∑ 2 1
∑ 2
∑ 1 2
2 ∑i ∑ i
factorize ⃗vC : 2 (⃗
u + ⃗vC ) = 2
i i i
mi ⃗
ui + ⃗vC i
ui + 2 ⃗vC
mi ⃗ i
mi . Here, i
mi ⃗
ui is the total momentum in the CM-frame, which is zero
according to the definition of the centre of mass.
11 I. Newton 1687
12 The superposition principle corresponds to the linearity of the non-relativistic equations of the gravity field and can be treated as an experimentally

verified postulate.
13 In the booklet of electromagnetism, this property will be derived from Eq. (5) using the superposition principle.

— page 4 —
 2.3 Basic forces
obtain |⃗a|; however, then the stiffness (the proportionality coefficient
Π = −m⃗g · (⃗r − ⃗r0 ), k) depends also on â, and force is not necessarily antiparallel
where ⃗r0 is the vector pointing to an arbitrarily chosen refer- to the displacement16 .
ence point. Now, keeping in mind that the energy is additive, Elasticity forces are conservative, potential energy can be
∫ ∫
we can write an expression for the gravitational potential en- found by simple integration Π = − F⃗ · d⃗a = kada = 12 ka2 .
ergy of N bodies:∑ ∑
Π = −⃗g · mi (⃗ri − ⃗r0 ) = −⃗g · (⃗rC − ⃗r0 ) mi . (7) fact 12: Under the assumptions of fact 9, the potential energy
i
Gravitational energy of two point masses can be calculated sim- of an elastically deformable body is given by
1
ilarly by integration; for two point masses, it is usually conveni- Π = ka2 . (10)
2
ent to take the reference configuration (for which the potential
It is quite clear that if we take, for instance, a rubber band
energy is zero) such that the distance between the two bodies is
of length l and stiffness k, making it twice longer will reduce its
infinite. For small displacements, the work done by the gravity
stiffness by a factor of two. Indeed, one can divide the longer
forces acting on the both bodies
m1 m2 m1 m2 band fictitiously into two halves, each of length l, which means
dW = G 2 r̂12 · (d⃗r1 − d⃗r2 ) = −G 2 dr12
r12 r12 that if we apply now the same force F to the endpoints of the
[here we took into account that d⃗r1 − d⃗r2 = d(⃗r1 − ⃗r2 ), and long band, both halves will be deformed by x = F/k and hence,
r̂12 · d(⃗r1 − ⃗r2 ) = −dr 12 ]. So,
∫ r12 the entire band is deformed by x′ = x + x = 2F/k so that the
Π=
m1 m2
G 2 dr = −G
m1 m2
. stiffness k ′ = F/x′ = k/2. Similarly, making the band twice
∞ r r12 thicker will increase the stiffness by a factor of two because
If there are more than two interacting bodies, we can use the we can consider the thicker band as being made of two paral-
superposition principle to find∑ lel thinner bands. NB! this does not apply to the springs and
mi mj
Π = −G ; (8) bending deformation: while making a spring twice longer will,
i>j
rij
of course, still decrease k by a factor of two, making it from a
note that the interaction energy of any pair of bodies needs to
twice thicker wire will increase the stiffness more than two-fold:
be counted only once, hence we sum over pairs with i > j.
by the same bending angle the thicker wire will be deformed
fact 10: Potential energy of two gravitationally interacting more than a thinner wire. This paragraph can be summarized
spherically symmetric bodies is given by Eq. (8); in the case as the following fact.
of homogeneous downwards-directed gravity field of strength
fact 13: For a band of elastic material of length L and cross-
g, the change of potential energy of a body of mass m is
sectional area A, the stiffness
∆Π = g∆h, where h is the change of height.
k = AY /L, (11)
Elasticity. Similarly to gravity forces, elasticity forces can where Y denotes the so-called Young modulus of the material17 .
be met literally at each our step. While microscopically, all
elasticity forces can be explained (at least in principle) in terms This equality makes it possible to give an alternative formu-
of electrostatic interactions using quantum mechanics, macro- lation of the Hooke’s law. To that end, let us introduce the
scopically it can take different forms. First of all, there is the concepts of the strain, which is defined as the relative deform-
Hooke’s law which describes elasticity forces for deformable ation ε = a/L, and stress which is defined as the elasticity
bodies (e.g. a rubber band or a spring); there are also normal force per unit area, σ = F/A. Then, the Hooke’s law can be
force and dry friction force which seemingly have nothing to do rewritten as
with elasticity, but in reality, both normal force and dry fric- σ = εY.
tion force have microscopically the same origin as the Hooke’s One can also introduce the concept of energy density of a
law. deformed material, the ratio of the potential energy and the
1 2
volume, w = Π/(LA) = 2 Y ε .
fact 11: (Hooke’s law14 ) If the deformation of an elastically Now we can also address the question, which deformations
deformable body is not too large, the deformed body exerts a
can be considered as “small enough” so that the Hooke’s law
force which (a) is antiparallel to the deformation vector ⃗a and
remains valid. One might think that we need to have ε ≪ 1,
(b) by modulus is proportional to the deformation, i.e.
but typically this is a too loose requirement: majority of elastic
F⃗ = −k⃗a. (9) materials break far before deformations ε ∼ 1 are reached. Typ-
This law is valid for small deformations of all elastic ma- ically, the Hooke’s law starts failing when deformations are so
terials including rubber bands, springs, etc, as long as the de- large that the material is close to breaking. Also, for such large
formation is not too large, and the body deformation includes deformations, the material is no longer fully elastic: when the
only stretching (or compression), and does not involve bend- force is removed, the material does not fully restore its ini-
ing or shear15 . If bending and shear deformations are involved, tial shape and some residual deformation will remain; such
with a fixed deformation direction (described by the unit vec- deformations are referred to as plastic deformations. There are
tor â ≡ ⃗a/|⃗a|), the force modulus remains to be proportional to also materials (which can be referred to as plastic materials or
14 R. Hooke 1660
15 Deformation of springs, in fact, does involve bending, but the Hooke’s law remains in that case nevertheless valid.
16 A proper description of the elasticity forces when bending and shearing are involved requires tensorial description and is beyond the scope of this

booklet
17 L. Euler 1727, G. Riccati 1782

— page 5 —

plastically deformable materials) which deform plastically over
a very wide range of strain values before breaking into pieces.
In such cases, the Hooke’s law remains valid only for extremely
small strains by which the deformation still remains elastic.
Note that there are hyperelastic materials for which the de-
formation can be very large, ε > 1; then, indeed, the condition
ε ≪ 1 is required for the applicability of the Hooke’s law.
Majority of the materials have rather large values of Y
which means that unless we have really long and thin threads
or wires, moderate forces will cause only a really minute and
un-noticeable deformations. This is typically the case for wires,
ropes, rods, and solid surfaces. In those cases, while the geo-
metrical effect of the deformations can be neglected, such a de-
formation will be formed that the elasticity force compensates
any applied external force. If we are dealing with a solid sur-
face, such an elasticity force is called the normal force; in the
case of rods, wires and ropes, we call it the tension force.
Unless otherwise emphasized, it is assumed that the tension
force is parallel to the rod, wire or rope. In the case of a rope
or a thin wire, this is essentially always the case: there is no
possibility of having an elastic shear or bending because the
rope is typically very “soft” with respect to such deformations:
if we try to create a perpendicular elasticity force by apply-
ing a peperpendicular external force, the rope will be bended
without giving rise to any noticeable force. In the case of a rod,
this is not true: if we apply an external perpendicular force, the
rod resists elastically to the attempts of bending and creates a
perpendicular component of the tension force. Still, if all the
external forces applied to a rod are parallel to it, according to
the Newton’s 3rd law, the tension force will be also parallel to
the rod.
In the case of a stretched string (rod, rope, etc), we can
divide it fictitiously into two parts. Then, at the division point
P , the two pieces attract each other with a certain elasticity
force. The direction of this force depends on which part of
the string is considered, but due to the Newton’s 3rd law, the
modulus of the force remains the same. The force with which
the two fictitious parts of the rope interact with each other at
the point P describes the state of the rope at that point, and
is referred to as the tension. So, we’ll distinguish the force F⃗
which is applied to the endpoint of a rope, and the tension T ,
which is defined for any point of the rope and describes the
state of it; note that when external forces are applied only to
the endpoints of a rope at equilibrium, |F⃗ | = T (this follows
from the force balance condition for any fictitious part of the
rope).
fact 14: Tension is an elasticity force in a linear construction
element such as string (rod, wire, etc)18 . For a non-stretchable
string19 , if it is being pulled (or pushed, which can happen in
the case of a rod) the tension adjusts itself to the externally
applied force to prevent stretching. If the mass of a thread or
rope can be neglected, the tension is constant along it. In a
freely bending rope, the tension force at a point P is parallel
to the tangent drawn to the rope at point P .
18 In the bulk of three-dimensional elastic bodies, instead of tension, the concept of stress is used; the respective description is mathematically more
complicated, e.g. the stress is a tensor quantity.
19 More precisely, a string made of a material with a very large value of Young modulus Y .
20 The friction laws were developed by L. da Vinci 1493, G. Amontons 1699, and C.A. Coulomb 1785.
— page 6 —
3. STATICS
fact 15: Normal force is the perpendicular to the surface
component of an elasticity force at the surface of a rigid (non-
deformable) body with which the rigid body acts on a con-
tacting body; it adjusts itself to the externally applied force
preventing thereby the rigid body from being deformed.
Note that if the externally applied force is not perpendicu-
lar to the rigid body surface then due to the Newton’s 3rd law,
the elasticity force will have both perpendicular to the surface
and tangential (parallel to the surface) components. The latter
manifests itself at the contact points of two solid bodies as the
friction force. More accurately, the friction force is the force at
the contact point of two bodies due to the interaction of the mo-
lecules of one body with the molecules of the other body when
the bodies try to slip one over the other; the surface molecules
are kept at their place due to the elasticity forces inside each of
the bodies; these elasticity forces are caused by the (typically
unnoticeably small) shear deformation of the bodies.
In the case of solid bodies, if the external tangential forces
are not too large as compared with the normal forces, the bod-
ies will not slip, and the friction force adjusts itself so as to
compensate the external tangential forces; this is called the
static friction force.
fact 16: (Amontons’ law.) The maximal static friction force
at the contact area of two bodies Fmax = µs N , where N is the
normal force at that contact area and µs is a constant depend-
ent on the two contacting materials, referred to as the static
coefficient of friction; it may also depend on the temperature,
humidity, etc. Thus, Fmax is independent of the contact area.
fact 17: (Coulomb’s friction law.) When two bodies move
with respect to each other, the friction force at the contact
area of these bodies F = µk N , where N is the normal force
at that contact area and µk is a constant dependent on the
two contacting materials, referred to as the kinetic coefficient
of friction; it does depend slightly on the slipping speed, but
this dependence is weak and typically ignored.
In the case of Olympiad problems, most often it is assumed
that µs = µk , but sometimes these are taken to be different,
with µs > µk .
While the friction laws are very simple and have been for-
mulated a long time ago20 , deriving it from the microscopic
(molecular) dynamics turns out to be a very difficult task (there
are still papers being published on that topic in research pa-
pers, c.f. M.H. Müser et al., Phys. Rev. Lett. 86, 1295 (2000),
and O.M. Braun et al., Phys. Rev. Lett. 110, 085503.
3 STATICS
When solving problems on statics, one can always use stand-
ard brute-force-approach: equations (1) and (3) tells us that
for each body at equilibrium, F⃗ = 0 and T⃗ = 0. So, for each
solid body, we have a force balance condition, and a torque
balance condition. According to the standard procedure, these
equations are to be projected onto x-, y- and z-axis yielding us
 3. STATICS
six equations (assuming 3-dimensional geometry); if there are force balance condition to the perpendicular of OO′ . It should
N interacting bodies, the overall number of equations is 6N . be emphasized that at least one torques equation needs to be
For a correctly posed problem, we need to have also 6N un- left into your set of equations: the “trading” of force equations
known parameters so that we could solve this set of algebraic for torques equations works because a rotation around O by
equations. The description of the procedure sounds simple, but a small angle dφ followed by a rotation around O′ by −dφ
solving so many equations might be fairly difficult. In the case results in a translational motion by |OO′ |dφ, but there is no
of 2-dimensional geometry, the number of equations is reduced such sequence of translational motions which could result in a
twice (while the number of force balance equations comes down rotational motion.
to two, all torques will be perpendicular to the plane, so there
idea 3: Using torque balance conditions is usually more effi-
is only one torque equation), but even with only two bodies,
cient than using force balance conditions because for any force
we have still 6 equations.
balance condition, we can eliminate only one force21 by project-
Luckily, there are tricks which can help us reducing the
ing the condition to the perpendicular of that force; meanwhile,
number of equations! Usually the main ingenuity lies in
if we choose the pivot point as the intersection point of the two
idea 1: choose optimal axes to zero as many projections of lines along which two unparallel forces are applied, both forces
forces as possible. It is especially good to zero the projections disappear from the equation.
of the forces we do not know and are not interested in,
for instance, the reaction force between two bodies or the pr 1. An end of a light wire rod is bent into a hoop of radius
tensile force in a string (or a rod). To zero as many forces r. The straight part of the rod has length l; a ball of mass M
as possible it is worthwhile to note that a) the axes may not be is attached to the other end of the rod. The pendulum thus
perpendicular; b) if the system consists of several bodies, then formed is hung by the hoop onto a revolving shaft. The coeffi-
a different set of axes may be chosen for each body. cient of friction between the shaft and the hoop is µ. Find the
equilibrium angle between the rod and the vertical.
idea 2: for the torques equation it is wise to choose such µ
a pivot point that zeroes as many moment arms as possible. ω
Again it is especially beneficial to zero the torques of “uninter- r
esting” forces.
For example, if we choose the pivot to be at the contact point of
two bodies, then the moment arms of the friction force between

l
the bodies and of their reaction force are both zero.
As mentioned above, for a two-dimensional system, we M
can write two equations per body for the forces (x- and y-
components) and one equation (per body) for the torques. We
This problem is classified as a difficult one because most people
could increase the number of equations either by using more
who try to solve it have difficulties in drawing a qualitatively
than two projections for force balance equations, or more than
correct sketch. What really helps making a correct sketch is
one pivot point (“axis” of rotation) for the torque balance.
relying on the idea 2. Mathematical simplifications are further
However,
offered by
fact 18: the maximum number of linearly independent equa- fact 19: on an inclined surface, slipping will start when the
tions (describing force and torque balance) equals the number slope angle α fulfills tan α = µ.
of degrees of freedom of the body (three in the two-dimensional
case, as the body can rotate in a plane and shift along the x- pr 2. On an incline with slope angle α there lies a cylinder
and y-axis, and six in the three-dimensional case). with mass M , its axis being horizontal. A small block with
So, if we write down two force balance conditions with two mass m is placed inside it. The coefficient of friction between
torque balance conditions then one of the four equations would the block and the cylinder is µ; the incline is nonslippery. What
always be a redundant consequence of the three others. is the maximum slope angle α for the cylinder to stay at rest?
Equations (1) and (3) seem to tell us that for 2-dimensional The block is much smaller than the radius of the cylinder.
geometry, we should use one torques equation and two force M
equations; however, each force equation can be “traded for”
one torques equation. So, apart for the “canonical” set of two
force equations and one torque equation, we can use one force
m
equation with two torque equations (with two different pivot
α
points), and we can also use three torques equations with three
different pivot points which must not lie on a single line. Indeed, Here we can again use fact 19 and idea 2 if we add
∑ −−→
let us have two torque balance conditions, i OPi × F⃗i = 0 and
∑ −− ′
→ idea 4: sometimes it is useful to consider a system of two
i O Pi × Fi = 0 where Pi is the application point of the force (or more) bodies as one whole and write the equations for the
⃗
F⃗i . Once we subtract one equation from the other, we obtain
∑ −−→′ ⃗ −−→ ∑ ⃗ forces and/or the torques for the whole system.
i OO × Fi = OO′ × Fi = 0, which is the projection of the
i
21 unless there are parallel forces

— page 7 —
 3. STATICS
Then, the net force (or torque) acting on the compond body pr 5. A block rests on an inclined surface with slope angle
is the sum of external forces (torques) acting on the constitu- α. The surface moves with a horizontal acceleration a which
ents. Our calculations are simplified because the internal forces lies in the same vertical plane as a normal vector to the surface.
(torques) between the different parts of the compound body can Determine the values of the coefficient of friction µ that allow
be ignored (due to Newton’s 3rd they cancel each other out). the block to remain still.
For the last problem, it is useful to assemble a compound body „
from the cylinder and the block. a
fi
pr 3. Three identical rods are connected by hinges to each
other, the outmost ones are hinged to a ceiling at points A and
Here we are helped by the very universal
B. The distance between these points is twice the length of a
rod. A weight of mass m is hanged onto hinge C. At least how idea 7: many problems become very easy in a non-inertial
strong a force onto hinge D is necessary to keep the system translationally moving reference frame.
stationary with the rod CD horizontal? To clarify: in a translationally moving reference frame we can
A B re-establish Newton’s laws by imagining that each body is ad-
ditionally acted on by an inertial force −m⃗a where ⃗a is the
C DF acceleration of the frame of reference and m is the mass of
m
a given body. Indeed, we have learned in kinematics that for
Again we can use idea 2. The work is also aided by translational motion of a reference frame, the accelerations can
fact 20: if forces are applied only to two endpoints of a rod be added, cf. idea K-19; so, in a moving frame, all the bodies
obtain additional acceleration −⃗a, as if there was an additional
and the fixture(s) of the rod at its endpoint(s) is (are) not rigid
force F⃗ = −m⃗a acting on a body of mass m.
(the rod rests freely on its supports or is attached to a string
or a hinge), then the tension force in the rod is directed along Note that that due to the equivalence of the inertial and
the rod. heavy mass (cf. “Gravity”, Section 2.3) the inertial foce is
totally analogous to the gravitational force22 . Because of that,
Indeed, at either endpoints, the applied net external force F⃗ we can use
must point along the rod, as its torque with respect to the other
endpoint must be zero. Further, according to the Newton’s 3rd idea 8: the net of the inertial and gravitational forces is
law, the external force F⃗ must be met by an equal and oppos- usable as an effective gravitational force.
ite force exerted by the rod, which is the tension force T⃗ , so
F⃗ = −T⃗ . pr 6. A cylinder with radius R spins around its axis with an
Some ideas are very universal, especially the mathematical angular speed ω. On its inner surface there lies a small block;
ones. the coefficient of friction between the block and the inner sur-
idea K-2 some extrema are easier to nd without using deriv-
face of the cylinder is µ. Find the values of ω for which the
atives,
block does not slip (stays still with respect to the cylinder).
Consider the cases where (a) the axis of the cylinder is hori-
for example, the shortest path from a point to a plane is per- zontal; (b) the axis is inclined by angle α with respect to the
pendicular to it. horizon.
ω
pr 4. What is the minimum force needed to dislodge a block
α
of mass m resting on an inclined plane of slope angle α, if the
coefficient of friction is µ? Investigate the cases when a) α = 0;
b) 0 < α < arctan µ.
m F idea 9: a rotating frame of reference may be used by adding
a centrifugal force mω 2 R⃗ (with ω being the angular speed of
fi the frame and R being a vector drawn from the axis of rota-
tion to the point in question) and Coriolis force. The latter is
idea 5: force balance can sometimes be resolved vectorially unimportant (a) for a body standing still or moving in parallel
without projecting anything onto axes. to the axis of rotation in a rotating frame of reference (in this
case the Coriolis force is zero); (b) for energy conservation (in
Fact 19, or rather its following generalisation, turns out to be this case the Coriolis force is perpendicular to the velocity and,
of use: thus, does not change the energy).
idea 6: if a body is on the verge of slipping (or already slip- Warning: in this idea, the axis of rotation must be actual, not
ping), then the sum of the friction force and the reaction force
instantaneous. Expressions for the centrifugal force and Cori-
is angled by arctan µ from the surface normal.
olis force are derived in appendix 4.
This idea can be used fairly often, for instance in the next For the problem 6, recall also idea K-2b and idea 6; for part
problem. (b), add
22 Their equivalence is the cornerstone of the theory of general relativity (more specifically, it assumes the inertial and gravitational forces to be
indistinguishable in any local measurement).

— page 8 —
 3. STATICS
idea K-11: in case of three-dimensional geometry, consider two- quantity as a superposition of two regions with opposite signs
dimensional sections. It is especially good if all interesting of the same quantity.
objects (for example, force vectors) lie on one section. The
This quantity can be mass density (like in this case), charge or
orientation and location of the sections may change in time.
current density, some force field etc. Often this trick can be
combined with
pr 7. A cart has two cylindrical wheels connected by a
weightless horizontal rod using weightless spokes and friction- idea 12: Make the problem as symmetric as possible.
less axis as shown in the figure. Each of the wheels is made This goal can be reached by applying idea 11, but also by us-
of a homogeneous disc of radius R, and has a cylidrical hole ing appropriate reference frames, dividing the process of solv-
of radius R/2 drilled coaxially at the distance R/3 from the ing into several phases (where some phases use symmetric geo-
centre of the wheel. The wheels are turned so that the holes metry), etc.
point towards each other, and the cart is put into motion on
a horizontal floor. What is the critical speed v by which the pr 8. A hollow cylinder with mass m and radius R stands
wheels start jumping? on a horizontal surface with its smooth flat end in contact the
surface everywhere. A thread has been wound around it and
its free end is pulled with velocity v in parallel to the thread.
Find the speed of the cylinder. Consider two cases: (a) the
coefficient of friction between the surface and the cylinder is
zero everywhere except for a thin straight band (much thinner
than the radius of the cylinder) with a coefficient of friction
This problem is somewhat similar to the previous one, and of µ, the band is parallel to the thread and its distance to the
we would be able to solve it using those ideas which we have thread a < 2R (the figure shows a top-down view); (b) the
already studied. Indeed, if we consider the process in a frame coefficient of friction is µ everywhere. Hint: any planar motion
co-moving with the cart, we can just apply Newton’s 2nd law of a rigid body can be viewed as rotation around an instant
to the centripetal acceleration of the wheel’s centre of mass. centre of rotation, i.e. the velocity vector of any point of the
However, let us solve it using few more ideas. body is the same as if the instant centre were the real axis of
rotation.
idea 10: Gravity force (or a fictitious force which is propor- v
tional to the mass of a body) can be considered to be applied

a
to the centre of mass of a body only in the following cases:
(a) the effective gravity field is homogeneous; „
(b) the body has a spherically symmetric mass distribution;
(c) the effective gravity field is proportional to the radius vec-
tor, e.g. centrifugal force field if the motion is constrained to
the plane perpendicular to the frame’s axis of rotation.
In all the other cases, it may happen by coincidence that the
This is quite a hard problem. It is useful to note
gravity force is still applied to the centre of mass, but typically
it is not. For instance, the Coriolis force can be considered to idea 13: if a body has to move with a constant velocity, then
be applied to the centre of mass only if the body is not rotating the problem is about statics.
(as seen from the rotating frame). Also remember ideas 1 and 2. The latter can be replaced with
The part (a) of this claim has been motivated in the paragraph its consequence,
following the idea 9; parts (b) and (c) will be motivated in the idea 14: if a body in equilibrium is acted on by three forces
booklet of electromagnetism (electrostatic and non-relativistic at three separate points, then their lines of action intersect at
gravitational fields obey similar laws). In order to prove that one point (note that the intersection point can be infinitely far
the part (d) is valid, we need to show that total force and total — lines intersecting at infinity means that the lines are parallel
torque exerted by the gravitational have the same magnitudes to each other). If there are only two points of action, then the
what would be obtained if the body were a point mass at the po- corresponding lines coincide.
sition of the centre of mass. So, using the attraction(repulsion)
This very useful idea follows directly from the torque balance
centre as the origin and assuming ⃗g = k⃗r, let us express the
∑ condition if the intersection point of two lines of action is taken
total torque as T⃗ = ⃗ri kmi⃗ri ×⃗ri ≡ 0; the same result would
for the pivot point (with two arms and the total torque being
be obtained for a centre of mass as the gravity force would
equal to zero, the third arm must be also equal to zero).
have a zero arm. Further, let us express the total force as
F⃗ = ∑ kmi⃗ri = kM (∑ mi⃗ri /M ) = kM⃗rC ; here M = ∑ mi is Another useful fact is
the total mass of the body. fact 21: the friction force acting on a given point is always
There are two more ideas which can be used here, antiparallel to the velocity of the point in the frame of reference
of the body causing the friction.
idea 11: In order to achieve a more symmetric configura-
tion or to make the situation simpler in some other way, it is From time to time some mathematical tricks are also of use;
sometimes useful to represent a region with zero value of some here it is the property of inscribed angles, and more specific-
— page 9 —
 3. STATICS
ally the particular case of the Thales’ theorem (among geomet- string.
rical theorems, this is probably the most useful one for solving
physics problems),
fact 22: a right angle is subtended by a semicircle (in general:
an inscribed angle in radians equals half of the ratio between
its arc-length and radius). m
The property of inscribed angles is also useful in the next
This problem is the easiest to solve using the method of
problem, if we add (somewhat trivial)
virtual displacement.
idea 15: in stable equilibrium the potential energy of a body
is minimum.
method 1: Imagine that we are able to change the length of
the string or rod the tension in which is searched for by an in-

pr 9. A light wire is bent into a right angle and a heavy nitesimal amount ∆x. Equating the work T ∆x by the change
ball is attached to the bend. The wire is placed onto supports ∆Π of the potential energy, we get T = ∆Π/∆x.
with height difference h and horizontal distance a. Find the Generalisation: if some additional external forces F⃗i (i =
position of the wire in its equilibrium. Express the position as 1, 2, . . .) act on the system with the displacements of their
the angle between the bisector of the right angle and the ver- points of action being δ⃗xi , while the interesting string or rod
tical. Neglect any friction between the wire and the supports; undergoes a virtual lengthening of ∆x, then
∑
the supports have little grooves keeping all motion in the plane T = (∆Π − δ⃗xi · F⃗i )/∆x.
of the wire and the figure. i

α The method can also be used for finding some other forces
h

than tension (for example, in problems about pulleys): by ima-

ginarily shifting the point of action of the unknown force one
a
can find the projection of this force onto the direction of the
virtual displacement.

pr 10. A rod with length l is hinged to a ceiling with height
h < l. Underneath, a board is being dragged on the floor. The
rod is meant to block the movement the board in one direction
while allowing it move in the opposite direction. What condi-
tion should be fulfilled for it to do its job? The coefficient of
friction is µ1 between the board and the rod, and µ2 between
the board and the floor.
pr 12. A rope with mass m is hung from the ceiling by its
both ends and a weight with mass M is attached to its centre.
The tangent to the rope at its either end forms angle α with
the ceiling. What is the angle β between the tangents to the
rope at the weight?
α α
m
β

M
fi
Let us recall the fact 14: The tension in a freely hanging
„2 „1
string is directed along the tangent to the string. In addition,
we can employ
Let’s remember fact 6: if the relative sliding between two
idea 17: for hanging ropes, membranes etc. it is usefult to
bodies has a known direction then the direction of the sum of
consider a piece of rope separately and think about the com-
the friction and reaction force vectors is always uniquely de-
ponentwise balance of forces acting onto it.
termined by the coefficient of friction. If a force makes one of
the bodies move in such a way that the reaction force grows In fact, here we do not need the idea as a whole, but, rather,
then they jam: the larger the forces we try to drag the bodies its consequence,
with, the larger friction and reaction forces restrain them. fact 23: the horizontal component of the tension in a massive
idea 16: Friction can block movement. In such a case, all rope is constant.
forces become negligible except for the friction force, reaction Using the idea 17 and fact 23, it is relatively easy to show that
force and the externally applied force that tries to make the the following approximation is valid.
system move, because gravitational (and such) forces are fixed,
but the said forces become the larger the harder we push or
idea 18: If the weight of a hanging part of a rope is much less
than its tension then the curvature of the rope is small and its
pull.
horizontal mass distribution can quite accurately be regarded
pr 11. Four long and four half as long rods are hinged to as constant.
each other forming three identical rhombi. One end of the con- This allows us to write down the condition of torque balance
traption is hinged to a ceiling, the other one is attached to a for the hanging portion of the rope (as we know the horizontal
weight of mass m. The hinge next to the weight is connected coordinate of its centre of mass). The next problem illustrates
to the hinge above by a string. Find the tension force in the that approach.
— page 10 —
 3. STATICS

pr 13. A boy is dragging a rope with length L = 50 m along idea 21: The equilibrium x = y = 0 of a system having
a horizontal ground with a coefficient of friction of µ = 0.6,
holding an end of the rope at height H = 1 m from the ground.
What is the length l of the part of the rope not touching the
ground?
pr 14. A light rod with length l is hinged in such a way
that the hinge folds in one plane only. The hinge is spun with
angular speed ω around a vertical axis. A small ball is fixed to
the other end of the rod. (a) Find the angular speeds for which
the vertical orientation is stable. (b) The ball is now attached
to another hinge and, in turn, to another identical rod (see the
figure below); the upper hinge is spun in the same way. What
is now the condition of stability for the vertical orientation?
ω ω
l
l
two degrees of freedom is stable if (and only if23 ) the potential
energy Π(x, y) as a function of two variables has a local min-
imum at x = y = 0, i.e. for any pair of values x, y within a
small neigbourhood of the equilibrium point (0, 0), inequality
Π(x, y) > Π(0, 0) must hold.
pr 15. If a beam with square cross-section and very low dens-
ity is placed in water, it will turn one pair of its long opposite
faces horizontal. This orientation, however, becomes unstable
as we increase its density. Find the critical density when this
transition occurs. The density of water is ρw = 1000 kg/m3 .
idea 22: The torque acting on a body placed into a liquid
is equal to torque from buoyancy, if we take the latter force to
be acting on the centre of the mass of the displaced liquid.

The validity of the idea 22 can be seen if we imagine that the

l

a) b) displaced volume is, again, filled with the liquid, and the body
itself is removed. Then, of course, the re-filled volume is at
Here the following idea is to be used. equilibrium (as it is a part of the resting liquid). This means
idea 19: For analysing stability of an equilibrium, there are that the torque of the buoyancy force must be balancing out
two options. the torque due to the weight of the re-filled volume; the weight
First, presume that the system deviates a little from the equi- of the re-filled volume is applied to its centre of mass, and ac-
librium, either by a small displacement ∆x or by a small angle cording to the idea 14, the buoyancy force must be therefore
∆φ, and find the direction of the appearing force or torque — also acting along the line drawn through the centre of mass.
whether it is towards the equilibrium or away from it. Apart from the idea 22, solution of the problem 15 can be
Second, express the change of total potential energy in terms of simplified by using the ideas 11 and 12.
the small displacement to see if it has a minimum or maximum
(for a system at equilibrium, its potential energy must have an pr 16. A hemispherical container is placed upside down on
extremum); minimum corresponds to stability, and maximum a smooth horizontal surface. Through a small hole at the bot-
— to instability (for a motivation and generalization of this tom of the container, water is then poured in. Exactly when
method, see appendix 5). the container gets full, water starts leaking from between the
table and the edge of the container. Find the mass of the con-
NB! compute approximately: when working with forces tainer if water has density ρ and radius of the hemisphere is
(torques), it is almost always enough to keep only those terms R.
which are linear in the deviation; when working with potential
energy, quadratic approximation is to be used. M
̺ R
It is extremely important in physics to be able to apply lin-
ear, quadratic, and sometimes also higher order approxima-
tions, which is based on
idea 23: If water starts flowing out from under an upside
idea 20: Taylor series: down container, normal force must have vanished between the
∆x2
f (x + ∆x) ≈ f (x) + f ′ (x)∆x + f ′′ (x) + ..., table and the edge of the container. Therefore force acting on
2 the system container+liquid from the table is equal solely to
2 2
for instace: sin φ ≈ tan φ ≈ φ; cos φ ≈ 1 − x2 ; ex ≈ 1 + x + x2 , force from hydrostatic pressure.
2
(1 + x)a ≈ 1 + ax + a(a−1)2 x2 , ln 1 + x ≈ 1 + x − x2 . Ana-
logous approach can be use for multivariable expressions, e.g. The latter is given by pS, where p is pressure of the liquid near
(x + ∆x)(y + ∆y) ≈ xy + x∆y + y∆x. Consider using such the tabletop and S is area of the container’s open side.
approximations wherever initial data suggest some parameter
to be small. pr 17. A block is situated on a slope with angle α, the
coefficient of friction between them is µ > tan α. The slope
The case (b) is substantially more difficult as the system is rapidly driven back and forth in a way that its velocity vec-
has two degrees of freedom (for example, the deviation angles tor ⃗u is parallel to both the slope and the horizontal and has
∆φ1 and ∆φ2 of the rods). Although idea 19 is generalisable constant modulus v; the direction of ⃗u reverses abruptly after
for more than one degrees of freedom, apparently it is easier to each time interval τ . What will be the average velocity w of
start from idea 15. the block’s motion? Assume that gτ ≪ v.
23 We assume that apart from the energy, there are no other conserved quantities for this system.

— page 11 —
 3. STATICS
y vu t pr 19. A horizontal platform rotates around a vertical axis
−v τ at angular velocity ω. A disk with radius R can freely rotate
u and move up and down along a slippery vertical axle situated
x α z at distance d > R from the platform’s axis. The disk is pressed
against the rotating platform due to gravity, the coefficient of
friction between them is µ. Find the angular velocity acquired
idea 24: If the system changes at high frequency, then it by the disk. Assume that pressure is distributed evenly over
is often pratical to use time-averaged values ⟨X⟩ instead of the entire base of the disk.
detailed calculations. In more complicated situations a high-
d
frequency component X̃ might have to be included (so that R r
X = ⟨X⟩ + X̃).
ω
method 2: (perturbation method) If the impact of some force
on a body's motion can be assumed to be small, then solve the
problem in two (or more) phases: rst nd motion of the body
idea 26: If we transform into a rotating frame of reference,
in the absence of that force (so-called zeroth approximation);
then we can add angular velocities about instantaneous axes of
then pretend that the body is moving just as found in the rst
rotation in the same way as we usually add velocities.
phase, but there is this small force acting on it. Look what cor-
Thus ω ⃗3 = ω⃗ 1 +⃗ω2 , where ω
⃗ 1 is angular velocity of the reference
rection (so-called rst correction) has to be made to the zeroth
frame, ω⃗ 2 angular velocity of the body in the rotating frame of
approximation due to that force.
reference and ω ⃗ 3 that in the stationary frame. In this question,
In this particular case, the choice of zeroth approximation we can use fact 21, ideas 2, 9, 13, K-7, and also
needs some explanation. The condition gτ ≪ v implies that idea K-33 Arbitrary motion of a rigid body can be considered
within one period, the block’s velocity cannot change much. as rotation about an instantaneous centre of rotation (in terms
Therefore if the block is initially slipping downwards at some of velocity vectors of the body).
velocity w and we investigate a short enough time interval, then
we can take the block’s velocity to be constant in zeroth approx- method 3: (dierential calculus) Divide the object into in-
imation, so that it is moving in a straight line. We can then nitesimally small bits or the process into innitesimally short
move on to phase two and find the average value of frictional periods (if necessary, combine this with idea 20).
force, based on the motion obtained in phase one.
For problem 17, recall also a lesson from kinematics, Within an infinitesimal bit (period), quantities changing in
space (time) can be taken constant (in our case, that quant-
idea K-7 If friction aects the motion then usually the most ap-
ity is the direction of frictional force vector). If necessary (see
propriate frame of reference is that of the environment causing
the next question), these quantities may be summed over all
the friction.
bits — this is called integration.
pr 18. Let us investigate the extent to which an iron de- pr 20. A waxing machine consists of a heavy disk with mass
posit can influence water level. Consider an iron deposit at M densely covered with short bristles on one side, so that if
the bottom of the ocean at depth h = 2 km. To simplify our it lies on the floor, then its weight is evenly distributed over
analysis, let us assume that it is a spherical volume with ra- a circular area with radius R. An electrical motor makes the
dius 1 km with density greater from the surrounding rock by disk rotate at angular velocity ω, the user compensates for the
∆ρ = 1000 kg/m3 . Presume that this sphere touches the bot- torque from frictional forces by a long handle. The same handle
tom of the ocean with its top, i.e. that its centre is situated at can be used to push the machine back and forth along the floor.
depth r + h. By how much is the water level directly above the With what force does the machine have to be pushed to make
iron deposit different from the average water level? it move at velocity v? Assume that angular velocity of the disk
is large, ωR ≫ v, and that the force needed to compensate for
h

the torque can be neglected. The coefficient of friction between

r
the bristles and the floor is µ.
iron
deposit Here we need fact 21, ideas K-33, 11, and additionally
idea 27: Try to determine the region of space where forces
idea 25: The surface of a liquid in equilibrium takes an equi- (or torques etc) cancel at pairs of points.
potential shape, i.e. energies of its constituent particles are the These pairs of points are often symmetrically located. Idea 12
same at every point of the surface. is relevant as well.
If this was not the case, the potential energy of the liquid could pr 21. A hexagonal pencil lies on a slope with inclination
be decreased by allowing some particles on the surface to flow angle α; the angle between the pencil’s axis and the line of
along the surface to where their potential energy is smaller (cf. intersection of the slope and the horizontal is φ. Under what
idea 15). Recall also the fact 10. condition will the pencil not roll down?
— page 12 —
 4. DYNAMICS
Let us note that this statement is in accordance with fact 18
that gives the number of available equations (there can be no
ϕ more unknowns than equations). In this particular case, we
α are dealing with effectively one-dimensional geometry with no
horizontal forces, but the body could rotate (in absence of the
wires). Thus we have two degrees of freedom, corresponding
idea 28: When solving three-dimensional problems, some- to vertical and rotational motion. Since the wires are identical,
times calculating coordinates in appropriately chosen axes and they must have the same stiffness as well; the word “wire” hints
applying formulae of spatial rotations can be of use. For the at large stiffness, i.e. deformations (and the inclination angle
rotation around z-axis by angle φ, x′ = x cos φ − y sin φ and of the bar) are small.
y ′ = y cos φ + x sin φ.

What (which vector) could be expressed in terms of its compon- 4 DYNAMICS

ents in our case? The only promising option is the small shift
A large proportion of dynamics problems consist of finding the
vector of centre of mass when its starts to move; ultimately we
acceleration of some system of bodies, or finding the forces act-
are only interested in its vertical component.
ing upon the bodies. There are several possible approaches for
pr 22. A slippery cylinder with radius R has been tilted solving such problems, here we consider three of them.
to make an angle α between its axis and the horizontal. A
method 4: For each body, we nd all the forces acting on it,
string with length L has been attached to the highest point P 24
including normal forces and frictional forces , and write out
of some cross-section of the cylinder, the other end of it is tied nd
Newton's 2 law in terms of components (i.e. by projecting
to a weight with mass m. The string takes its equilibrium pos-
the equation on x, y , and possibly z -axes). NB! Select the
ition, how long (l) is the part not touching the cylinder? The
directions of the axes carefully, cf. idea 1. In some cases, it
weight is shifted from its equilibrium position in such a way
may be possible (and more convenient) to abstain from using
that the shift vector is parallel to the vertical plane including
projections and work with vectorial equalities.
the cylinder’s axis; what is the period of small oscillations?
Keep in mind that for a correctly posed problem, it should be
P possible to write as many linearly independent equations as
there are unknowns (following idea 1 may help to reduce that
number). The guideline for figuring out how many equations
l can be found remains the same as in the case of statics problem,
α
see idea 18 (for time being we consider problems where bodies
do not rotate, so we need to count only the translational de-
grees of freedom). If the number of equations and the number
idea 29: Unfolding a part of the surface of a three- of unknowns don’t match, it is either an ill-posed problem, or
dimensional object and looking at the thereby flattened surface
you need to make additional physical assumptions (like in the
can assist in solving problems, among other things it helps to
case of problem 23).
find shortest distances.
pr 24. A block with mass M lies on a slippery horizontal
pr 23. A uniform bar with mass m and length l hangs on surface. On top of it there is another block with mass m which
four identical light wires. The wires have been attached to the in turn is attached to an identical block by a string. The string
bar at distances 3l from one another and are vertical, whereas has been pulled across a pulley situated at the corner of the
the bar is horizontal. Initially, tensions are the same in all big block and the second small block is hanging vertically. Ini-
wires, T0 = mg/4. Find tensions after one of the outermost tially, the system is held at rest. Find the acceleration of the
wires has been cut. big block immediately after the system is released. You may
neglect friction, as well as masses of the string and the pulley.
m
l/3
m

M
l

idea 30: If more fixing elements (rods, strings, etc) than the This question can be successfully solved using method 4,
necessary minimum have been used to keep a body in static but we need three more ideas.
equilibrium (i.e. more than the number of degrees of freedom) idea 31: If a body is initially at rest, then its shift vector
and fixing elements are absolutely rigid, then tensions in the is parallel to the force acting on it (and its acceleration) right
elements cannot be determined. In order to make it possible, after the start of its motion.
the elements have to be considered elastic (able to deform);
recall the fact 13. idea 32: If bodies are connected by a rope or a rod or per-
24 It is convenient to make a sketch and draw all the force vectors from their application points.

— page 13 —
 4. DYNAMICS
haps a pulley or one is supported by the other, then there is M
a linear25 arithmetic relation between the bodies’ shifts (and
m
velocities, accelerations) that describes the fact that length of
α α
the string (rod, etc.) is constant.

The relation for shifts is usually the easiest to find; if the mo-
tion is along a straight line, this relation can be differentiated
The final method is based on using generalised coordinates and
once or twice with respect to time, to obtain the relation for
originates from analytical mechanics. There, it is known as Lag-
velocities or accelerations. For curved motion care needs to be
rangian formalism and is introduced using relatively advanced
taken. In the case of velocities and for rigid bodies, we result in
mathematical apparatus (partial derivatives, variational ana-
the idea K-35 (for two points of a rigid body, the projections of
lysis), but for most problems, its simplified version outlined
velocities of both points on the line connecting them are equal).
below will suffice. More detailed discussion of the Lagrangian
In the case of accelerations, situation is more complicated as
formalism is provided in appendix 6.
there will be also centripetal accelerations. However, if the mo-
tion starts from the rest, for a very short time period t we can method 6: Let us call ξ a generalised coordinate if the en-

assume that the acceleration is constant and the centripetal tire state of a system can be described by this single number.

acceleration v 2 /r is negligible (the speed v is still very small), Say we need to nd the acceleration ξ¨ of coordinate ξ. If we
2
hence there is a simple linear relation ⃗s = t2 ⃗a between the shift can express the potential energy Π of the system as a func-

⃗s and the acceleration ⃗a of a certain point of a certain body; tion Π(ξ) of ξ and the kinetic energy in the form ˙2
K = Mξ /2
M
2
then the factor t2 cancels out from the linear relation for the where coecient is a combination of masses of the bodies

shifts, so that it carries directly over to the accelerations. (and perhaps of moments of inertia), then

ξ¨ = −Π′ (ξ)/M.
idea 33: If a light rope is thrown over an ideal pulley (both Here, a dot denotes differentiation w.r.t. time and dash
of negligible mass) then tension in the rope has the same mag- w.r.t. coordinate ξ. Indeed, due to conservation of energy
nitude to either side of the pulley; if the rope bends at the Π(ξ) + Mξ˙2 /2 =Const. Differentiating that w.r.t. time and
pulley, there is a normal force between the pulley and the rope using the chain rule, we obtain Π′ (ξ)ξ˙ + Mξ˙ξ¨ = 0. We reach
which can be found as the vectorial sum of the tension forces. the aforementioned formula after dividing through by ξ.˙

pr 26. A small block with mass m lies on a wedge with

Indeed, consider that piece of rope which is in contact with the
angle α and mass M . The block is attached to a rope pulled
pulley; its mass can be neglected, hence the inertial term in the
over a pulley attached to the tip of the wedge and fixed to a
Newton’s 2nd law can be also neglected, hence the normal force
horizontal wall (see the figure). Find the acceleration of the
acting on the rope must be equal and opposite to the vectorial
wedge. All surfaces are slippery (there is no friction).
sum of the tension forces.

method 5: Otherwise the same as method 4, but motion M m

is investigated in a non-inertial frame of reference (see idea 7)
a=?
where one of the bodies is at rest. α

Method 5 is useful in many questions concerning wedges: it is
difficult to write out the condition for an object to stay on the
wedge in the laboratory frame. Applying idea 32 is also often
easier in the wedge’s frame of reference than in the laboratory
frame. Don’t forget that the body defining the reference frame
is at rest: we have one or more equations expressing its static
equilibrium.
Full solution of this problem is given in the hints’ section
to illustrate method 6
pr 27. A wedge with mass M and acute angles α1 and α2
lies on a horizontal surface. A string has been drawn across
a pulley situated at the top of the wedge, its ends are tied to
blocks with masses m1 and m2 . What will be the acceleration
of the wedge? There is no friction anywhere.

pr 25. A wedge has been made out of a very light and slip-
pery material. Its upper surface consists of two slopes making
an angle α with the horizontal and inclined towards one an-
other. The block is situated on a horizontal plane; a ball with
mass m lies at the bottom of the hole on its upper surface. An-
other ball with mass M is placed higher than the first ball and
the system is released. On what condition will the small ball
with mass m start slipping upwards along the slope? Friction
can be neglected.
25 It is linear in shifts but may contain coefficients which are expressed in terms of nonlinear functions, e.g. trigonometric functions of angles.
m1 m2
α1 M α2
It may seem that there is more than one degree of freedom
in this question: the wedge can move and the string can shift
w.r.t. the wedge. However, we are saved by
idea 34: If x-components of the sum of external forces and
of centre of mass velocity are both zero, then the x-coordinate
of the centre of mass remains constant.

— page 14 —
 4. DYNAMICS
We can use this circumstance to reduce the effective number idea 38: Project Newton’s 2nd law on the axis perpendicular
of degrees of freedom. In our particular case, the system con- to an unwanted vector, e.g. an unknown force or the tangential
sists of two components and thus the shift of component can component of acceleration.
be expressed by that of the other.
We can easily find the cylinder’s velocity (and thus the radial
idea 35: The x-coordinate of the centre of mass of a system component of acceleration) if we use
of bodies is ∑ ∑ idea 39: If energy is conserved (or its change can be calcu-
XC = xi mi / mi , lated from work done etc), write it out immediately. Energy is
where mi denotes mass of the i-th component and xi the co- conserved if there is no dissipation (friction, inelastic collisions
ordinate of its centre of mass. The formula can be rewritten in etc) and external forces acting on the system are static (e.g. a
∫ ∫
integral form, XC = xdm/ dm, where dm = ρ(x, y, z)dV is stationary inclined plane);
differential of mass.
forces changing in time (force acting on a moving point, moving
inclined plane) change energy as well. Idea 32 helps to write
pr 28. Two slippery horizontal surfaces form a step. A block
out conservation of energy (relation between bodies’ velocities!).
with the same height as the step is pushed near the step, and a
To answer the second question, we need
cylinder with radius r is placed on the gap. Both the cylinder
and the block have mass m. Find the normal force N between idea 40: Normal force vanishes at the moment when a body
the cylinder and the step at the moment when distance between detaches from a surface.
√
the block and the step is 2r. Initially, the block and the step Also, review idea 32 for horizontal components of accelerations.
were very close together and all bodies were at rest. Friction
is zero everywhere. Will the cylinder first separate from the pr 29. Light wheels with radius R are attached to a heavy
block or the step? axle. The system rolls along a horizontal surface which sud-
denly turns into a slope with angle α. For which angles α will
r

m the wheels move without lifting off, i.e. touch the surface at all
times? Mass of the wheels can be neglected. The axle is par-
allel to the boundary between horizontal and sloped surfaces
m √ − and has velocity v.
2r

v
It is easy to end up with very complicated expressions when m
solving this problem, this may lead to mistakes. Therefore it
is wise to plan the solution carefully before writing down any
equations. α

idea 36: Newton’s laws are mostly used to find acceleration

from force, but sometimes it is clever to find force from accel- idea 41: To answer the question whether a body lifts off,
eration. we have to find the point on the non-lifting-off trajectory with
But how to find acceleration(s) in that case? It is entirely pos- smallest normal force.
sible if we use method 6, but this path leads to long expressions. If normal force has to be negative at that point, then the body
A tactical suggestion: if you see that the solution is getting very lifts off; the critical value is zero — compare with idea 40).
complicated technically, take a break and think if there is an Also, review ideas 1, 39 and K29.
easier way. There is a “coincidence” in this particular problem:
straight lines drawn from the sphere’s centre to points of touch- pr 30. A block with mass M lies on a horizontal slippery
ing are perpendicular; can this perhaps help? It turns out that surface and also touches a vertical wall. In the upper surface
it does. of the block, there is a cavity with the shape of a half-cylinder
idea 37: Pay attention to special cases and use simplifications with radius r. A small pellet with mass m is released at the
that they give rise to! upper edge of the cavity, on the side closer to the wall. What
is the maximum velocity of the block during its subsequent
Let us remind what we learned in kinematics:
motion? Friction can be neglected.
idea K-34 In case of motion along a curve, the radial compon-
m

ent (perpendicular to the trajectory) of a point's acceleration

r
v 2 /R is determined by velocity v and radius of curvature R; the
component along the trajectory is linear acceleration (equal to
M
εR in case of rotational motion, ε is angular acceleration).

The centre of mass of the cylinder undergoes rotational motion,

method 6 is necessary to find angular acceleration — but we
idea 42: Conservation law can hold only during some period
hoped to refrain from using it. An improvement on idea 1 helps
of time.
us out:
— page 15 —

idea 43: Momentum is conserved if the sum of external
forces is zero; sometimes momentum is conserved only along
one axis.
You will also need idea 39.
idea 44: Velocity is maximal (or minimal) when accelera- the force on the axle immediately after the motion begins. The
tion (and net force) is zero (since 0 = dv
dt = a); shift is extremal
when velocity is zero. Possible other pairs: electrical charge
(capacitor’s voltage)-current, current-inductive emf, etc.
4. DYNAMICS
pr 32. An inextensible rough thread with mass per unit
length ρ and length L is thrown over a pulley such that the
length of one hanging end is l. The pulley is comprised of a
hoop of mass m and radius R attached to a horizontal axle
by light spokes. The initially motionless system is let go. Find
friction between the pulley and the axle is negligible.
R

pr 31. A light rod with length 3l is attached to the ceiling

by two strings with equal lengths. Two balls with masses m
and M are fixed to the rod, the distance between them and
their distances from the ends of the rod are all equal to l. Find
the tension in the second string right after the first has been
cut. Why not proceed as follows: to find the force, we will use idea
36; the acceleration of the system will be found using Method
6. To apply idea 36 most handily, let us employ
idea 48: Newton’s 2nd law can be written as F⃗ = M⃗aC ,
m M
where ⃗aC is the acceleration of the centre of mass.
l l l
This idea is best utilised when a part of the system’s mass is mo-
tionless and only a relatively small mass is moved about (just
There are several good solutions for this problem, all of which
like in this case: the only difference after a small period of time
share applying idea 36 and the need to find the angular accel-
is that a short length of thread is “lost” at one end and “gained”
eration of the rod. Firstly, angular acceleration of the rod can
at the other end). Obviously idea 34 will be useful here, and
be found from method 6 by choosing angle of rotation φ to
idea 11 will save you some effort. Bear in mind that in this
be the generalised coordinate. Secondly, we may use Newton’s
case we are not interested in the centre of mass coordinate per
2nd law for rotational motion: we find the torque on the rod
se, but only in its change as a function of time; therefore in the
about the point of attachment of the second string and equate
expression for this coordinate we can omit the terms that are
it to Iε with angular acceleration ε and moment of inertia
independent of time: their time derivatives will vanish. The
I = ml2 + 4M l2 . More generally,
time-dependent part of the centre of mass coordinate should
idea 45: When a body is rotating around the axis s, the net be expressed using the same coordinate that we will use with
torque it experiences is M = Iε (not to be confused with the Method 6 (since Method 6 will produce its second derivative
body’s mass), where I is its moment of inertia with respect to with respect to time). A technical bit of advice may help: a
∑ ∫ ∫
the axis s, I = mi ri2 = r2 · dm = r2 ρ · dV and ri is the vector is specified by (a) its magnitude and direction; (b) its
distance of i-th particle from the axis s (the sum is evaluated projections onto coordinate axes in a given coordinate system;
over all particles of the body). Kinetic energy is K = 12 Iω 2 .
Once the angular acceleration is found, in order to apply the idea 49: sometimes it is easier to compute the components
idea 36 it may be helpful to use of a vector, even if we are interested in its magnitude only.
idea 46: The more general and sometimes indispensable Above all, this applies when the direction of the vector is
⃗
form of Newton’s 2nd law is F⃗ = ddtP , where P⃗ is the net mo- neither known nor apparent. In this instance, we should find
mentum of the system and F⃗ is the sum of external forces Fx and Fy in a suitable coordinate system.
⃗ = dL ⃗
acting on the system. An analogous formula is M dt , where
⃗ is the net angular momentum of the system (with respect to
L pr 33. A thread is thrown over a pulley. At its both ends
a given point) and M⃗ is the sum of external torques. there are two blocks with equal masses. Initially the two blocks
are at the same height. One of them is instantaneously given a
In our case this last method is fruitful when applied both to
small horizontal velocity v. Which of the two blocks will reach
forces and torques.
higher during the subsequent motion? The pulley’s mass is
Another solution method is to consider the rod and the negligible.
balls as three different (interacting) bodies. Then the balls’ ac-
celerations can be found as per idea 32; one can also employ

idea 47: Net force and torque acting on very light bodies
(compared to other bodies) are zero.
v
Clearly if this were not true, a non-zero force would generate
an infinite acceleration for a massless body.
— page 16 —
 4. DYNAMICS
This problem is really tough, because the key to the solution cause of the momentum conservation law. The conservation of
is a very specific and rarely used energy means that the sides of the triangle satisfy Pythagore’s
idea 50: If the centre of mass of a system cannot move, then theorem. A special case of this result is (see the problem after
the net force acting on it is zero. next)

Here the centre of mass can move about a little bit, but in the fact 25: When an elastic ball undergoes a central collision
longer term (averaged over one period of the pendulum-like with another identical stationary ball, then the first ball stops
motion of the kicked block — cf. idea 24) it is motionless: the and the second gains the velocity of the first ball.
blocks have the same mass and if one of them rises, then in
the expression for the centre of mass this will be compensated pr 36. An absolutely elastic and slippery billiard ball is mov-
by the descent of the other block. This is also true for the ing with velocity v toward two motionless identical balls. The
horizontal coordinate of the centre of mass, but it is enough to motionless balls are touching and their centres lie on a straight
consider the vertical coordinate only to solve the problem. Let line that is perpendicular to the incoming ball’s velocity vector.
us also bring up the rather obvious The moving ball is directed exactly toward the touching point
of the two balls. Which velocity will the incoming ball have
fact 24: the tension in a weightless thread thrown over a
after the collisions? Consider two scenarios: (a) the incoming
weightless pulley or pulled along a frictionless surface is the
ball hits exactly in the middle between the balls; (b) its tra-
same everywhere.
jectory is a little bit off and it hits one of the stationary balls
The solution algorithm is then as follows: we write down New- v
ton’s 2nd law for (a) the system made out of two blocks and marginally earlier.
(b) one block; we average both equations and use the equality
apparent from (a) to find the average tension in the thread, To answer the first question, it is necessary to use
which we then substitute into equation (b). Based on idea 24, idea 53: collisions (and other many-body interactions, like
we partition the tension in the thread into the average and the the motion of balls connected by threads or springs) are easier
high-frequency component and use idea 20. to treated in the centre of mass system, because in that system
pr 34. A system of blocks sits on a smooth surface, as shown the momentum conservation is the easiest to write down (the
in the figure. The coefficient of friction between the blocks is net momentum is zero).
µ, while that between the blocks and the surface is µ = 0. Also, do not forget idea 39! For the second question, let us use
m m
F idea 54: if a force acting on a body during a known time does
M M
not change direction, then the transferred momentum has the
x same direction as the force.
The bottom right block is being pulled by a force F . Find the
accelerations of all blocks. pr 37. n absolutely elastic beads are sliding along the fric-
tionless wire. What is the maximum possible number of col-
idea 51: When bodies are connected by frictional forces, then lisions? The sizes of the beads are negligible, and so is the
to answer some questions fully one needs to consider all pos- probability that more than two beads will collide at the same
sible combinations of there being relative slipping between all time.
possible touching surfaces.
For example, if we are to assume that there is no slipping idea 55: Representing the process visually, e.g. with a graph,
between two touching bodies, then they could be treated as tends to be great help.
a whole. Then one should find the frictional force Fh between
Here is an auxiliary question: what would the elastic collision
the bodies and determine when the assumption holds, or when
of two balls on an x − t diagram look like?
is Fh less that the maximum static friction force µN .

pr 35. A billiard ball hits another stationary billiard ball. pr 38. A plank of length L and mass M is lying on a smooth
horisontal surface; on its one end lies a small block of mass m.
At which collection of points could the stationary ball be posi-
The coefficient of friction between the block and the plank is µ.
tioned such that it would be possible to achieve the situation
What is the minimal velocity v that needs to be imparted to
where both balls will fall into two (different) pockets on the
the plank with a quick shove such that during the subsequent
table? The collisions are perfectly elastic, the balls are per-
motion the block would slide the whole length of the board
fectly slippery (hence the rotation of the balls is negligible).
and then would fall off the plank? The size of the block is
idea 52: If an absolutely elastic ball hits another motion- negligible.
less identical ball and the rotation (rolling) of the balls can be µ= 0 m µ L v
M
ignored, then upon impact there will be a right angle between
the velocity vectors of the two balls.
This problem has two more or less equivalent solutions. First,
To prove this, note that the three velocity vectors (velocity be- we could solve it using idea 7. Second, we could use ideas 39
fore and the two velocities after the impact) form a triangle be- and 53, further employing
— page 17 —
 4. DYNAMICS
idea 56: if a body slides along a level surface, then the en- law!). On the other hand, for tap B the laminar flow is not
ergy that gets converted to heat is equal to the product of the preserved; there will be eddies and loss of energy. We could
friction force and the length of the sliding track. nonetheless work with momentum: we write the expression for
Indeed, the friction force has a constant magnitude and, as seen the pressure exerted on the liquid by the walls of the barrel
in the reference frame of the support, it is always parallel to (generally the pressures exerted by the left and the right hand
displacement. side walls of the barrel cancel each other out, but there remains
an uncompensated pressure p = ρgH exerted to the left of the
pr 39. The given figure has been produced off a stroboscopic cross-section of the tap S).
photograph and it depicts the collision of two balls of equal dia-
meters but different masses. The arrow notes the direction of pr 41. Sand is transported to the construction site using a
motion of one of the balls before the impact. Find the ratio conveyor belt. The length of the belt is l, the angle with respect
of the masses of the two balls and show what the direction of to the horizontal is α; the belt is driven by the lower pulley with
motion for the second ball was before the impact. radius R, powered externally. The sand is put onto the belt at
a constant rate µ (kg/s). What is the minimal required torque
needed to transport the sand? What is the velocity of the belt
at that torque? The coefficient of friction is large enough for
the sand grains to stop moving immediately after hitting the
belt; take the initial velocity of the sand grains to be zero.
R
µ l

α
idea 57: sometimes it is beneficial to treat momenta as
vectors, treating their vectorial sums and differences using tri-
angle or parallelogram rules (this is also true of other vectorial fact 28: To make anything move — bodies or a flow (e.g. of
quantities: displacements, velocities, accelerations, forces etc.) sand) — force needs to be exerted.
To be more specific: when two bodies interact, the vector of For this problem, idea 58 and methode 3 will come in handy
the impulse is equal to the vectorial difference of their two mo- in addition to
menta. Cf. idea 5.
idea 59: (the condition for continuity) for a stationary flow
fact 26: In a stroboscopic photograph, the vector from one the flux of matter (the quantity of stuff crossing the cross-
position of the body to the next is proportional to its velocity
section of the flow per unite time) is constant and is independ-
(vector).
ent of the cross-section: σv = Const [σ(x) is the matter density
fact 27: (Newton’s 3rd law) if two bodies have interacted, per unit distance and v(x) — the velocity of the flow].
the changes of momenta of the two bodies are equal and oppos- For a flow of incompressible (constant density) liquid in a pipe,
ite. such a density is σ = ρS and therefore vS = Const. For a
region of space where the flow is discharged — a sink — the
pr 40. There are two barrels (A and B) whose taps have mass increases: dm dt = σv — this equation, too, could be called
different design, see figure. The tap is opened, the height of the condition for continuity.
the water surface from the tap is H. What velocity does the
water stream leave the barrels with? pr 42. A ductile blob of clay falls against the floor from
the height h and starts sliding. What is the velocity of the
A B blob at the very beginning of sliding if the coefficient of fric-
H H tion between the floor and the blob is µ? The initial horizontal
velocity of the blob was u.

idea 60: If during an impact against a hard wall there is

always sliding, then the ratio of the impulses imparted along
idea 58: If it seems that it is possible to solve a problem and perpendicular to the wall is µ.
using both energy and momentum conservation, then at least ∫
Indeed, ∆p⊥ = N (t)dt (integrated over the duration of the
one of these is not actually conserved! ∫ ∫
impact) and ∆p∥ = µN (t)dt = µ N (t)dt.
It could not be otherwise: the answers are, after all, different.
It pays to be attentive here. While designing the tap A, there pr 43. A boy is dragging a sled by the rope behind him as
was a clear attempt to preserve the laminarity of the flow: en- he slowly ascends a hill. What is the work that the boy does
ergy is conserved. However, if, motivated by method 3, we to transport the sled to the tip of the hill if its height is h and
were to write down the momentum given to the stream by the the horizontal distance from the foot of the hill to its tip is a?
air pressure during an infinitesimal time dt — pSdt (where S Assume that the rope is always parallel to the tangent of the
is the tap’s area of cross-section), we would see that, owing to hill’s slope, and that the coefficient of friction between the sled
the flow of water, p ̸= ρg (cf. dynamical pressure, Bernoulli’s and the snow is µ.
— page 18 —
 4. DYNAMICS
axis of interest using the parallel-axis (Steiner) theorem; (b) ap-

h
ply idea 61 to calculate kinetic energy or angular momentum
a (in which case it is only enough to know the moment of inertia
with respect to the centre of mass).
fact 29: if the exact shape of a certain surface or a time
pr 45. A rod of mass M and length 2l is sliding on ice. The
dependence is not given, then you have to deal with the general
velocity of the centre of mass of the rod is v, the rod’s angular
case: prove that the proposition is true for an arbitrary shape.
velocity is ω. At the instant when the centre of mass velocity
Clearly, to apply the fact 29, one will need idea 3. is perpendicular to the rod itself, it hits a motionless post with
an end. What is the velocity of the centre of mass of the rod
pr 44. An empty cylinder with mass M is rolling without after the impact if (a) the impact is perfectly inelastic (the end
slipping along a slanted surface, whose angle of inclination is
that hits the post stops moving); (b) the impact is perfectly
α = 45◦ . On its inner surface can slide freely a small block
elastic.
of mass m = M/2. What is the angle β between the normal
2l
to the slanted surface and the straight line segment connecting M ω
the centre of the cylinder and the block? v
M
In case of an absolutely elastic collision one equation follows
from energy conservation; if the collision is inelastic, then an-
other condition arises: that of a motionless end of the rod. Still,
β
m we have two variables. The second equation arises from
α
idea 64: if a body collides with something, then its angular
momentum is conserved with respect to the point of impact.
Clearly the simplest solution is based on idea 6, but one needs
to calculate the kinetic energy of a rolling cylinder. Indeed, during the impact the body’s motion is affected by the
normal and frictional forces, but both are applied through the
idea 61: K = Kc + MΣ vc2 /2, where Kc is the kinetic energy
point of impact: their lever arm is zero. If a body is moving
as seen in the centre of mass frame and MΣ — is the net mass
in a gravitational or similar field, then in the longer term the
of the system. Analogously: P⃗ = MΣ⃗vc (since P⃗c ≡ 0) and the
⃗ = Lc + ⃗rc × P⃗ . Parallel-axis (Steiner) angular momentum with respect to the point of impact may be-
angular momentum L
gin to change, but immediately before and after the collision it
theorem holds: I = I0 + MΣ a2 , where I is the moment of iner-
is nonetheless the same (gravity is not too strong as opposed to
tia with respect to an axis s and I0 — that with respect to an
the normal forces that are strong yet short-lived; even though
axis through the centre of mass (parallel to s) while a is the
gravity’s lever arm is non-zero, it cannot change the angular
distance between these two axes.
momentum in an instant).
We will have to compute angular momentum already in the
next problem, so let us clarify things a little. pr 46. If one hits something rigid — e.g. a lamppost —
idea 62: Angular momentum∑is additive. Dividing the sys- with a bat, the hand holding the bat may get stung (hurt) as
tem into point-like masses, L ⃗ = ⃗ i , where for i-th point-like
L long as the impact misses the so-called centre of percussion of
mass L⃗ i = ⃗ri × p⃗i (generally) or Li = hi pi = ri pti (motion in the bat (and hits either below or above such a centre). De-
a plane), hi = ri sin αi is the lever arm and pti = pi sin α — is termine the position of the centre of percussion for a bat of
the tangential component of the momentum). Kinetic energy, uniform density. You may assume that during an impact the
momentum etc. are also additive. bat is rotating around its holding hand.

If in a three-dimensional space the angular momentum is a vec-
tor, for a motion in a plane this vector is perpendicular to the
plane and is therefore effectively a scalar (and thus one can
abandon cross products). It is often handy to combine ideas
61 and 62: we do not divide the system into particles but, in-
∑
stead, into rigid bodies (L = Li ), we compute the moment
of inertia Li of each body according to idea 61: the moment
of inertia of the centre of mass plus the moment of inertia as
measured in the centre of mass frame.
idea 63: Here are moments of inertia for a few bodies,
with respect to the centre of mass. A rod of length of l:
1 2 2 2 2 2
12 M l , solid sphere: 5 M R , spherical shell: 3 M R , cylinder:
1 2
2 M R , square with side length a, axis perpendicular to its
plane: 16 M a2 .
If the the rotation axis does not go through the centre of mass,
then one can (a) find the moment of inertia with respect to the
— page 19 —
method 7: Convert a real-life problem into the formal lan-
guage of physics and math  in other words, create a model.
Phrased like that, it may seem that the method is rather point-
less. However, converting and interpreting real-life scenarios —
modelling the problem — is one of the most challenging and in-
teresting aspects of physics. It is interesting because it supplies
more creative freedom than solving an existing model using
well-established ideas. Still, this freedom has limits: the model
has to describe the reality as best as possible, the approxim-
ations have to make sense and it is desirable that the model
were solvable either mentally or with aid of a computer. For a
given problem, there is not much freedom left and the business
is simplified: there clear hints as to sensible assumptions. Let
us begin translating: “A rigid rod of length l and uniform dens-
ity is rotating around one end with the angular velocity ω, the
rotation axis is perpendicular to the rod. At a distance x from

the axis there is a motionless post that is parallel to the axis
of rotation. The rod hits the post.” Now we encounter the first
obstacle: is the impact elastic or inelastic? This is not brought
up in the text of the problem. Let us leave it for now: maybe
we can get somewhere even without the corresponding assump-
tion (it turns out that this is the case). Now we encounter the
central question: what does it mean for the hand “not to get
stung”? We know it hurts when something hits our hand — if
this something gets an impulse from the hand during a short
period of time (the impact), as this implies a large force. The
hand is stationary, so the hand-held end of the bat should come
to halt without receiving any impulse from the hand. Thus our
interpretation of the problem is complete: “Following the im-
pact, the rotation is reversed, 0 ≥ ω ′ ≥ −ω; during the impact
the axis of rotation imparts no impulse on the rod. Find x.”
The penultimate sentence hints at the usage of idea 64.
pr 47. A massive cylinder of radius R and mass M is lying
on the floor. A narrow groove of depth a has been chiselled
along the circumference of the cylinder. A thread has been
wrapped around the groove and is now being pulled by its free
end, held horizontally, with a force F . The cylinder is posi-
tioned such that the thread is being freed from below the cyl-
inder. With what acceleration will the cylinder start moving?
The friction between the floor and the cylinder is large enough
for there to be no slipping.
4. DYNAMICS
Indeed, the points where the normal force and the gravity are
applied are on the same straight line with the forces themselves
and their sum is zero, meaning that their net torque is also zero;
the force of friction is lying in the plane of the surface, and so
its lever arm with respect to an axis in the same plane is zero.
pr 49. A “spring-dumbbell” comprises two balls of mass m
that are connected with a spring of stiffness k. Two such dumb-
bells are sliding toward one another, the velocity of either is v0 .
At some point the distance between them is L (see fig.). After
which time is the distance between them equal to L again? The
collisions are perfectly elastic.
v0 v0
L
idea 67: If a system consisting of elastic bodies, connected by
springs, threads etc., interacts with other bodies, then the dur-
ation of impact of the elastic bodies is significantly smaller than
the characteristic times of other processes. The whole process
can then be divided into simpler stages: an almost instantan-
eous collision of elastic bodies (that could be considered free,
as e.g. the spring exerts an insignificant force compared to that
exerted in an elastic collision) and the subsequent (or preced-
ent, or in between the collisions) slow process: the oscillations
of the spring etc.

M R a Note: this is a rather general idea, division into simpler steps

F can be useful if rapid (almost instantaneous) processes can oc-
⊙F
cur in a dynamical system; see next problem for an example
(also recall idea 53)
There are multiple ways to tackle this problem, but let us use
the following idea. pr 50. Small grains of sand are sliding without friction along
idea 65: The relation Iε = M is clearly valid only if the cylindrical trough of radius R (see fig.). The inclination angle
a
centre of rotation is motionless; however, it turns out that it of the trough is α. All grains have initial velocity zero and start
also holds when the instantaneous axis of rotation is moving near point A (but not necessarily at the point A itself). What
translationally such that the distance of the body’s centre of should be the length of the trough such that all grains would
mass from the axis does not change (eg when rolling a cyl- exit it at the point B (i.e. exactly at the bottom of the trough)?
indrical or spherical object).
L
To prove this idea, recall idea 6: kinetic energy appears when A
1 2
work is done, K = 2 Iω = M φ (φ is the angle of rotation of
the body, ω = dφ/dt). If the moment of inertia with respect to B α
the instantaneous axis of rotation I does not depend on time,
then dK/dt = 12 Idω 2 /dt = Iωε = dM φ/dt = M ω, which gives
Iε = M . idea 68: If the motion of a spread collection of particles
could be divided into oscillation in a known direction and an
pr 48. A ball is rolling along a horizontal floor in the region oscillation-free motion (so motion perpendicular to the oscilla-
x < 0 with velocity ⃗v0 = (vx0 , vy0 ). In the region x > 0 there tion), then the particles are focussed at certain points: where
is a conveyor belt that moves with velocity ⃗u = (0, u) (parallel the oscillation phase of all particles is either zero or is an integer
to its edge x = 0). Find the velocity of the ball ⃗v = (vx , vy ) multiple of 2π.
with respect to the belt after it has rolled onto the belt. The
surface of the conveyor belt is rough (the ball does not slip)
and is level with the floor. pr 51. A coat hanger made of wire with a non-uniform
density distribution is oscillating with a small amplitude in the
idea 66: For cylindrical or spherical bodies rolling or slipping plane of the figure. In the first two cases the longer side of the
on a horizontal surface, the angular momentum is conserved triangle is horizontal. In all three cases the periods of oscilla-
with respect to an arbitrary axis lying in the plane of the sur- tion are equal. Find the position of the centre of mass and the
face. period of oscillation.
— page 20 —
 4. DYNAMICS
the effective mass of the body increases by αρ0 V . In the prob-
lem above, the constant α for the spherical body can be found
using the conditions given in the first half of the problem.
In case (B), if we assume that the velocity of the body is con-
10cm stant, we find K = 12 v 2 ρ0 (αSvt), where S is the cross-sectional
42cm
area of the body and αS is the cross-sectional area of the turbu-
lent ‘tail’. This α, again, characterizes the body. From here, it
is easy to find F v = dK α 3 α 2
dt = 2 v ρ0 S, which gives F = 2 v ρ0 S.
Background info: A finite-size rigid body that oscillates
around a fixed axis is known as the physical pendulum. Its pr 53. A stream of water falls against a trough’s bottom
frequency of small oscillations is easy to derive from the re- with velocity v and splits into smaller streams going to the left
lation I φ̈ = −mglφ, where I is the moment of inertia with and to the right. Find the velocities of both streams if the
respect to the axis of oscillation and l is the distance of the incoming stream was inclined at an angle α to the trough (and
centre of mass from that axis: ω −2 = I/mgl = I0 /mgl + l/g the resultant streams). What is the ratio of amounts of water
(here we employ the parallel-axis/Steiner theorem, see idea 61). carried per unit time in the two outgoing streams?
The reduced length of the physical pendulum is the distance
˜l = l + I0 /ml such that the frequency of oscillation of a math-
ematical pendulum of that length is the same as for the given α
physical pendulum.
idea 69: If we draw a straight line of length ˜l such that it This is a rather hard problem. Let us first state a few ideas
passes through the centre of mass and one of its ends is by
and facts.
the axis of rotation, then if we move the rotation axis to the
other end of the segment (and let the body reach a stable equi- idea 71: For liquid flow, Bernoulli’s (i.e. energy conservation)
librium), then the new frequency of oscillation is the same as law is often helpful: p + ρgh + 21 ρv 2 = Const, where p is the
before. Conclusion: the set of points where the axis of rotation static pressure, h is the height of the considered point and v is
could be placed without changing the frequency of oscillation, the velocity of the flow at that point.
consists of two concentric circles around the centre of mass.
fact 30: Inside the liquid close to its free surface the static
Proof: the formula above could be rewritten as a quadratic pressure is equal to the external pressure.
equation to find the length l corresponding to the given fre-
quency ω (i.e. to the given reduced length ˜l = g/ω 2 ): To solve the second half of the problem, the following is needed:
l2 − l˜l + I0 /m = 0. According to Vieta’s formulae, the solutions
l1 and l2 satisfy l1 + l2 = l, so that l1 and l2 = ˜l − l1 result in idea 72: Idea 46 can be generalized in a way that would hold
the same frequency of oscillations. for open systems (certain amounts of matter enter and leave
⃗ ⃗ P in − Φ
⃗ P out , where Φ
⃗ P in and Φ
⃗ P out
pr 52. A metallic sphere of radius 2 mm and density ρ = the system): F⃗ = dt + Φ
dP

3000 kg/m3 is moving in water, falling freely with the accelera- are the entering and the outgoing fluxes of momentum (in other
tion a0 = 0,57g. The water density is ρ0 = 1000 kg/m3 . With words, the net momentum of the matter entering and leaving
what acceleration would a spherical bubble of radius 1 mm rise the system, respectively).
in the water? Consider the flow to be laminar in both cases; The momentum flux of the flowing liquid could be calculated as
neglect friction. the product of momentum volume density ρ⃗v with the flow rate
(volume of liquid entering/leaving the system per unit time).
idea 70: If a body moves in a liquid, the fluid will also move. What is the open system we should be considering in this
(A) If the flow is laminar (no eddies), only the liquid adjacent case? Clearly, a system that would allow relating the incoming
to the body will move; (B) if the flow is turbulent, there will flow rate µ (kg/s) to the outgoing fluxes (µl ja µr ) using the
be a turbulent ‘tail’ behind the body. In either case the charac- formula above: a small imaginary region of space that would
teristic velocity of the moving liquid is the same as the velocity include the region where the stream splits into two.
of the body.
fact 31: If we can ignore viscosity, the component of the force
Using method 6 we find that in the case (A) the kinetic energy exerted by the stream bed (including the ‘walls’ limiting the
of the system K = 12 v 2 (m + αρ0 V ), where the constant α is a flow) on the flow that is parallel to these walls is zero.
number that characterizes the geometry of the body that cor-
respond to the extent of the region of the liquid that will move pr 54. Find the velocity of propagation of small waves
(compared to the volume of the body itself). This expression in shallow water. The water is considered shallow if the
is obtained by noticing that the characteristic speed of the li- wavelength is considerably larger than the depth of the wa-
quid around the body is v, and the characteristic size of the ter H. Thanks to this we can assume that along a vertical
region where the liquid moves (the speed is not much smaller cross-section the horizontal velocity of all particles vh is the
than v) is estimated as the size of the body itself. If a body is same and that the horizontal velocity of water particles is sig-
acted on by a force F , then the power produced by this force nificantly smaller than the vertical velocity. The smallness of
is P = F v = dK dt = va(m + αρ0 V ). Thus F = a(m + αρ0 V ): the waves means that their height is significantly smaller than
— page 21 —
 5. REVISION PROBLEMS
the depth of the water. This allows us to assume that the ho- down the stick, from top to bottom. How should the bug’s
rizontal velocity of the water particles is significantly smaller acceleration depend on its distance from the top endpoint of
than the wave velocity, u. the stick? The bug’s mass is m, the length of the stick is l, the
angle between the floor and the stick is α and the stick’s mass
idea 73: A standard method for finding the velocity of is negligible; both the floor and the wall are slippery (µ = 0).
propagation (or another characteristic) of a wave (or another How long will it take the bug to reach the bottom of the stick
structure with persistent shape) is to choose a reference system having started at the top (from rest)?
where the wave is at rest. In this frame, (a) continuity (idea 59)
x
and (b) energy conservation (e.g. in the form of Bernoulli’s law)
hold. In certain cases energy conservation law can be replaced l
by the balance of forces. a

(An alternative approach is to linearise and solve a system of α

coupled partial differential equations.)

pr 55. A small sphere with mass m = 1 g is moving along a
smooth surface, sliding back and forth and colliding elastically
with a wall and a block. The mass of the rectangular block
is M = 1 kg, the initial velocity of the sphere is v0 = 10 m/s.
What is the velocity of the sphere at the instant when the dis-
tance between the sphere and the wall has doubled as compared
with the initial distance? By how many times will the average
force (averaged over time) exerted by the sphere on the wall
have changed?
pr 58. A wedge with the angle α at the tip is lying on the
horizontal floor. There is a hole with smooth walls in the ceil-
ing. A rod has been inserted snugly into that hole, and it can
move up and down without friction, while its axis is fixed to
be vertical. The rod is supported against the wedge; the only
point with friction is the contact point of the wedge and the
rod: the friction coefficient there is µ. For which values of µ is
it possible to push the wedge through, behind the rod, by only
applying a sufficiently large horizontal force?

idea 74: If a similar oscillatory motion takes place, for which

the parameters of the system change slowly (compared to the F
µ
period of oscillation), then the so-called adiabatic invariant I is α
conserved: it is the area enclosed by the closed contour traced
by the trajectory of the system on the so-called phase diagram
(where the coordinates are the spatial coordinate x and mo- pr 59. Sometimes a contraption is used to hang pictures etc.
mentum px ). on the wall, whose model will be presented below. Against a
fixed vertical surface is an immovable tilted plane, where the
Let us be more precise here. The closed contour is produced as angle between the surface and the plane is α. There is a gap
a parametric curve (the so-called phase trajectory) x(t), px (t) between the surface and the plane, where a thin plate could
if we trace the motion of the system during one full period T . be fit. The plate is positioned tightly against the vertical sur-
The phase trajectory is normally drawn with an arrow that in- face; the coefficient of friction between them can be considered
dicated the direction of motion. The adiabatic invariant is not equal to zero. In the space between the plate and the plane a
exactly and perfectly conserved, but the precision with which cylinder of mass m can move freely, its axis being horizontal
it is conserved grows if the ratio τ /T grows, where τ is the and parallel to all considered surfaces. The cylinder rests on
characteristic time of change of the system’s parameters. the plate and the plane and the coefficients of friction on those
Adiabatic invariant plays an instrumental role in physics: two surfaces are, respectively, µ and µ . For which values of
1 2
from the adiabatic law in gases (compare the result of the pre- the friction coefficients the plate will assuredly not fall down
vious problem with the adiabatic expansion law for an ideal regardless of its weight?
gas with one degree of freedom!) and is applicable even in
quantum mechanics (the number of quanta in the system —
m µ2
e.g. photons — is conserved if the parameters of the system
are varied slowly).
α µ1
5 REVISION PROBLEMS
pr 56. A straight homogeneous rod is being externally sup- µ=0
ported against a vertical wall such that the angle between the F
wall and the rod is α < 90◦ . For which values of α can the rod
remain stationary when thus supported? Consider two scen- pr 60. On top of a cylinder with a horisontal axis a plank is
arios: a) the wall is slippery and the floor is rough with the placed, whose length is l and thickness is h. For which radius
friction coefficient µ ; b) the floor is slippery and the wall is R of the cylinder the horizontal position of the plank is stable?
rough with the friction coefficient µ.
l
pr 57. A light stick rests with one end against a vertical R
wall and another on a horizontal floor. A bug wants to crawl
— page 22 —
 5.
pr 61. A vessel in the shape of a cylinder, whose height
equals its radius R and whose cavity is half-spherical, is filled
to the brim with water, turned upside down and positioned on
a horizontal surface. The radius of the half-spherical cavity is
also R and there is a little hole in the vessel’s bottom. From
below the edges of the freely lying vessel some water leaks out.
How high will the remaining layer of water be, if the mass of
the vessel is m and the water density is ρ? If necessary, use
the formula for the volume of a slice of a sphere (see Fig.):
V = πH 2 (R − H/3).
h H V
pr 62. A vertical cylindrical vessel with radius R is rotating
around its axis with the angular velocity ω. By how much does spherical surface. Find the height at which it will lose contact
the water surface height at the axis differ from the height next with the surface. The sphere is held in place and its radius is
to the vessel’s edges? R; there is no friction.
pr 63. A block with mass M is on a slippery horizontal sur-
face. A thread extends over one of its corners. The thread is
attached to the wall at its one end and to a little block of mass
m, which is inclined by an angle α with respect to the vertical,
at the other. Initially the thread is stretched and the blocks
are held in place. Then the blocks are released. For which ratio
of masses will the angle α remain unchanged throughout the
subsequent motion?
M α m
pr 64. Two slippery (µ = 0) wedge-shaped inclined surfaces
with equal tilt angles α are positioned such that their sides are
parallel, the inclines are facing each other and there is a little
gap in between (see fig.). On top of the surfaces are positioned
a cylinder and a wedge-shaped block, whereas they are resting
one against the other and one of the block’s sides is horizontal.
The masses are, respectively, m and M . What accelerations
will the cylinder and the block move with? Find the reaction
force between them.
m M
α α
pr 65. Three little cylinders are connected with weightless
rods, where there is a hinge near the middle cylinder, so that
the angle between the rods can change freely. Initially this
angle is a right angle. Two of the cylinders have mass m, an-
other one at the side has the mass 4m. Find the acceleration
of the heavier cylinder immediately after the motion begins.
Ignore friction.
— page 23 —
REVISION PROBLEMS
m
o
90
a=?
m 4m
pr 66. A slippery rod is positioned at an angle α with re-
spect to the horizon. A little ring of mass m can slide along
the rod, to which a long thread is attached. A small sphere
of size M is attached to the thread. Initially the ring is held
motionless, and the thread hangs vertically. Then the ring is
released. What is the acceleration of the sphere immediately
after that?
m
α
M
pr 67. A block begins sliding at the uppermost point of a
pr 68. The length of a weightless rod is 2l. A small sphere
of mass m is fixed at a distance x = l from its upper end. The
rod rests with its one end against the wall and the other against
the floor. The end that rests on the floor is being moved with
a constant velocity v away from the wall. a) Find the force
with which the sphere affects the rod at the moment, when the
angle between the wall and the rod is α = 45◦ ; (b) what is the
answer if x ̸= l?
α x
m 2l
v
pr 69. A light rod with length l is connected to the hori-
zontal surface with a hinge; a small sphere of mass m is connec-
ted to the end of the rod. Initially the rod is vertical and the
sphere rests against the block of mass M . The system is left
to freely move and after a certain time the block loses contact
with the surface of the block — at the moment when the rod
forms an angle α = π/6 with the horizontal. Find the ratio of
masses M/m and the velocity u of the block at the moment of
separation.
m
M
l
pr 70. At a distance l from the edge of the table lies a block
that is connected with a thread to another exact same block.
The length of the thread is 2l and it is extended around the
pulley sitting at the edge of the table. The other block is held
above the table such that the string is under tension. Then
the second block is released. What happens first: does the first
block reach the pulley or does the second one hit the table?
 5. REVISION PROBLEMS
l l A
l

H
l
B

pr 71. A cylindrical ice hockey puck with a uniform thick- pr 76.
ness and density is given an angular velocity ω and a transla-
tional velocity u. What trajectory will the puck follow if the
ice is equally slippery everywhere? In which case will it slide
farther: when ω = 0 or when ω ̸= 0, assuming that in both
cases u is the same?
A stick with uniform density rests with one end
against the ground and with the other against the wall. Ini-
tially it was vertical and began sliding from rest such that all
of the subsequent motion takes place in a plane that is perpen-
dicular to the intersection line of the floor and the wall. What
was the angle between the stick and the wall at the moment
when the stick lost contact with the wall? Ignore friction.

pr 72. A little sphere of mass M hangs at the end of a very
long thread; to that sphere is, with a weightless rod, attached
another little sphere of mass m. The length of the rod is l.
Initially the system is in equilibrium. What horizontal velocity
needs to be given to the bottom sphere for it to ascend the
same height with the upper sphere? The sizes of the spheres
are negligible compared to the length of the rod.
M straight and its linear density is constant.
l
pr 77. A log with mass M is sliding along the ice while ro-
tating. The velocity of the log’s centre of mass is v, its angular
velocity is ω. At the moment when the log is perpendicular
to the velocity of its centre of mass, the log hits a stationery
puck with mass m. For which ratio of the masses M/m is the
situation, where the log stays in place while the puck slides
away, possible? The collisions are perfectly elastic. The log is
M ω
v m

v m
pr 78. A ball falls down from height h, initially the ball’s
horizontal velocity was v0 and it wasn’t rotating. a) Find the
velocity and the angular velocity of the ball after the following
pr 73. A block of mass m lies on a slippery horizontal sur-
collision against the floor: the ball’s deformation against the
face. On top of it lies another block of mass m, and on top of
floor was absolutely elastic, yet there was friction at the con-
that — another block of mass m. A thread that connects the
tact surface such that the part of the ball that was in contact
first and the third block has been extended around a weight-
with the floor stopped. b) Answer the same question with the
less pulley. The threads are horizontal and the pulley is being
assumption that the velocities of the surfaces in contact never
pulled by a force F . What is the acceleration of the second
homogenized and that throughout the collision there was fric-
block? The coefficient of friction between the blocks is µ.
tion with coefficient µ.
m F
m a=?
m
pr 79. A ball is rolling down an inclined plane. Find the
ball’s acceleration. The plane is inclined at an angle α, the
coefficient of friction between the ball and the plane is µ.
pr 74. A boy with mass m wants to push another boy stand-
ing on the ice, whose mass M is bigger that his own. To that
end, he speeds up, runs toward the other boy and pushed him pr 80. A hoop of mass M and radius r stands on a slippery
for as long as they can stand up. What is the maximal distance horizontal surface. There is a thin slippery tunnel inside the
by which it is possible to push in this fashion? The maximal hoop, along which a tiny block of mass m can slide. Initially
velocity of a run is v, the coefficient of friction between both all the bodies are at rest and the block is at the hoop’s up-
boys and the ice is µ. permost point. Find the velocity and the acceleration of the
hoop’s central point at the moment when the angle between
the imaginary line connecting the hoop’s central point and the
pr 75. A uniform rod with length l is attached with a weight- block’s position and the vertical is φ.
less thread (whose length is also l) to the ceiling at point A.
The bottom end of the rod rests on the slippery floor at point
B, which is exactly below point A. The length of AB is H,
O
l < H < 2l. The rod begins to slide from rest; find the max-
imal speed of its centre during subsequent motion. Also, find ϕ
r

the acceleration of the rod’s centre and tension in the thread

A
at that moment when the rod’s speed is maximal if the rod’s
mass is m.
— page 24 —
 5. REVISION PROBLEMS

pr 81. A block with mass m = 10 g is put on a board that whose height is c, whose base is a right triangle with legs a and
has been made such that, when sliding to the left, the coeffi- b and which is made out of material with density ρ; b) an body
cient of friction µ1 = 0,3, while when sliding to the right it of an arbitrary shape with volume V and density ρ.
is µ2 = 0,5. The board is repeatedly moved left-right accord- p1
p
cs
ing to the graph v(t) (see fig.). The graph is periodic with p0

b
a

c
period T = 0,01 s; the velocity v of the board is considered x
v
1 m/s pr 86. A dumbbell consisting of two elastic spheres connec-
ted with a thin steel rod is moving parallel to its axis with
a velocity v toward another exact same spheres. Find the ve-
T/2 T t locity of the dumbbell after a central collision. Is the kinetic
energy of the system conserved?
v0

positive when directed to the right. Using the graph, find the appendix 1: Momentum conservation law.
average velocity that the block will move with. Let us consider a system of N point masses (bodies), and
let us represent the force acting on the i-th point as a sum,
∑ ⃗
pr 82. A water turbine consists of a large number of paddles F⃗i = j Fij + Fi ,
⃗ where F⃗ij is the force exerted on the i-th
that could be considered as light flat boards with length l, that body due to the j -th body
26
and F
⃗i is an external force, i.e.
are at one end attached to a rotating axis. The paddles’ free the net force exerted by such bodies which are not part of the
ends are positioned on the surface of an imaginary cylinder given system. Then, the Newton's 2nd law for the i-th body is
that is coaxial with the turbine’s axis. A stream of water with written as
d ∑
velocity v and flow rate µ (kg/s) is directed on the turbine mi ⃗vi = F⃗ij + F
⃗i .
such that it only hits the edges of the paddles. Find the max- dt j
imum possible usable power that could be extracted with such If we sum this equality over the index i, we obtain at the left-

a turbine. hand-side
∑ d d ∑ d
µ v mi ⃗vi = mi⃗vi = P⃗ ,
dt dt i dt
∑ i
where P⃗ = i mi⃗vi is called the momentum of the system of
bodies. Here we have kept in mind that F⃗ii = 0, and made use
of the additivity of dierentiation: derivative of a sum is the
l

sum of derivatives. The internal forces at the right-hand-side

cancel out:  
∑ ∑ ∑ ∑
pr 83. A flat board is inclined at an angle α to the vertical.  F⃗ij  = F⃗ij = (F⃗ij + F⃗ji ) = 0.
One of its ends is in the water, the other one is outside the i j i,j i>j
water. The board is moving with velocity v with respect to its Here we rst represented the sum as being taken over all the
normal. What is the velocity of the water stream directed up index ij pairs, and then grouped the terms with symmetric in-
∑
the board? dices (ij and ji) together ( i>j means that the sum is taken
over all such ij -pairs where i > j ); nally, we use the Newton's
u

3rd law to conclude that F⃗ij + F⃗ji = 0. Upon introducing the

⃗ =∑ F
v
net external force as F ⃗
i i , we obtain
d ⃗
P = F.⃗
dt
The last equality is essentially a generalization of the Newton's

pr 84. A motor-driven wagon is used to transport a load 2nd law to a system of bodies. In particular, if there are no

horizontally by a distance L. The load is attached to the side of external forces, F

⃗ = 0, and the momentum P⃗ is conserved.

the wagon by a cable of length l. Half of the time the wagon is Notice that if there are no external forces, the equations of
uniformly accelerated, the other half — uniformly decelerated. motion (equations which dene how the system will evolve),
Find the values of the acceleration a such that, upon reaching i.e. equations expressing the Newton's 2nd law, obey transla-
the destination, the load will be hanging down motionlessly. tional symmetry : we can translationally displace the reference
You can assume that a ≪ g. frame by a vector ⃗a without any change to the equations of
motion. Indeed, the new vectors pointing to the positions of
pr 85. A shockwave could be considered as a discontinuous the bodies (the radius vectors ) are expressed in terms of the
jump of the air pressure from value p0 to p1 , propagating with old ones as ⃗ri′ = ⃗ri − ⃗a. The internal forces F⃗ij depend only
speed cs . Find the speed which will be obtained, when influ- on the relative placement of the bodies, i.e. on the vectors
enced by the shockwave, (a) a wedge-shaped block: a prism ⃗ri′ − ⃗rj′ = (⃗ri − ⃗a) − (⃗rj − ⃗a) = ⃗ri − ⃗rj which are expressed
26 Due ⃗ii = −F
to the fact 5, F ⃗ii , hence F
⃗ii = 0: a body cannot exert a force on itself

— page 25 —
 5. REVISION PROBLEMS
in terms of the new coordinates exactly in the same way as in appendix 3: Energy conservation law.
terms of the old coordinates. In order to derive the energy conservation law, let us consider

The discipline of analytical mechanics shows that each sym-

the time derivative of the kinetic energy of a system of bodies,
∑ ∑
metry of the equations of motion containing a parameter
dened as K =
1
2 vi2 = 12 j mi⃗vi · ⃗vi (here we wrote vi2
j mi⃗
(which can take arbitrarily small values) gives rise to a conser-
as a dot product of ⃗vi with itself ). For dierentiation of dot
27 products, the ordinary rule for the derivative of a product ap-
vation law . Here we have actually three independent para-
meters, the components of the displacement vector ax , ay , and
plies: (⃗ a ·⃗b)′ = ⃗a′ ·⃗b+⃗a ·⃗b′ ; since the dot product is commutative
az ; because of that we have three conserved quantities  the
a · b = ⃗b · ⃗a), we obtain (⃗a · ⃗a)′ = 2⃗a · ⃗a′ . So,
(i.e. ⃗ ⃗
d ∑ d⃗vi ∑
respective components of the momentum vector P⃗ . K= mi · ⃗vi = F⃗i · ⃗vi ,
dt j
dt j

appendix 2: Angular momentum conservation law. which can be rewritten for the dierential of the kinetic energy
∑ ⃗
Similarly to the momentum conservation law, we consider a as dK = j Fi · ⃗
vi dt, where the force acting on the i-th body

N can be expressed as the sum of internal and external forces,

system of bodies, with the same designations. Then, we ∑
can take the time derivative of the expression of the angular F⃗i = j F⃗ij + F
⃗i ⃗vi dt = d⃗ri . Then
and
∑ ∑
momentum of the i-th body:( ) dK = F⃗ij · d⃗ri + F
⃗i · d⃗ri .
d d⃗ri d⃗vi i,j i
mi⃗ri × ⃗vi = mi × ⃗vi + ⃗ri × .
dt dt dt
′
Here, F⃗ij · d⃗ri is called the work done by the force F⃗ij . If the
Here we have applied the product dierentiation rule (ab) = ⃗ij
forces F
a b+ab which is still valid in vector algebra: (⃗a · b) = ⃗a · b+⃗a ·⃗b′
′ ′ ⃗ ′ ′ ⃗

a × ⃗b)′ = ⃗a′ × ⃗b + ⃗a × ⃗b′ (NB! we need to keep the or-

and (⃗ (a) depend only on the coordinates ⃗ri and do not depend on

der of the vectors since the cross product is anticommutative

, the velocities ⃗vi and on time t;
⃗a × ⃗b = −⃗b × ⃗a)
Notice that
d
dt ⃗
ri = ⃗
v i and ⃗
v i × ⃗
v i = 0 , hence (b) there exists such function Π ≡ Π(⃗r1 , ⃗r2 , . . . ⃗rN ) (hence-
the rst term in right-hand-side drops out. Further, let us sum forth referred to as the potential energy ) that by any in-
our rst equality over the index i, and substitute the remaining nitesimal (i.e. innitely small) displacements d⃗ri of the
nd
 2 law, 
terms at the right-hand-side using the Newton's bodies, the total work done by all the internal forces F⃗ij
d⃗vi ∑
⃗i ⇒ mi⃗ri × d⃗vi = ⃗ri × 
∑
⃗i 
and external forces F
⃗i equals to the opposite of the total
mi = F⃗ij + F F⃗ij + F
dt dt dierential 29 of the potential energy, i.e.
 
j j
to obtain ∑ ∑
d ∑ ∑ −dΠ(⃗r1 , ⃗r2 , . . . ⃗rN ) = F
⃗i + F⃗ij  · d⃗ri ;
L= ⃗ri × F⃗ij + ⃗ri × F
⃗i .
i j
dt i,j i
Now, let us notice that due to the Newton's 3rd law F⃗ij = −F⃗ji ; then for any displacement of the system, dK = −dΠ and hence,
all the macroscopic non-relativistic forces between two point d(K +Π) = 0, i.e. the total energy E = K +Π = const; such in-
masses take place either at the contact point when these
⃗ij and external forces F
ternal forces F ⃗i which satisfy the above
two point masses touch each other (elasticity force, friction listed conditions are referred to as conservative forces.
force), or is parallel to the line connecting these points (electro- Note that the condition (b) above is equivalent to saying that
28
static force, gravitational force) In either case, we can write the work done by the forces depends only on the initial and
⃗rj = ⃗ri + k F⃗ji ; if we multiply this equality by F⃗ji , we obtain nal states of the system (i.e. on the positions of the point
⃗ri × F⃗ij = ⃗rj × F⃗ij = −⃗rj × F⃗ji , hence the internal torques cancel masses), and not along which trajectories the point masses
pair-wise out from the sum; what remains is the net external moved. Mathematically, this condition can be also rewritten
∑
torque T⃗ = i ⃗ri × F
⃗i : 30
using partial derivatives ,
d ⃗ ⃗
L=T. ∂Π(⃗r1 , ⃗r2 , . . . ⃗rN ) ∑
dt − = Fijx + Fix ; (12)
∂xi
This can be considered as the generalization of the Newton's j

II law to the rotational motion of a system of bodies; if the right-hand-side of this condition is the x-component of the

torque of external forces is zero (T

⃗ = 0) then we end up with total force acting on the i-th point mass (xi stands for the

the conservation of angular momentum, L

⃗ =const. x-coordinate of the i-th particle); similar conditions need to be
valid also for they - and z -components.
Note that within the discipline of analytical mechanics, the an-
gular momentum conservation can be derived from rotational An important case when the forces are conservative is the case

symmetry of the full energy of a mechanical system (while we of central force elds: the internal force between two point

rotate the reference frame by an angle α around an axis which masses is parallel to the line connecting these point masses

goes through the origin, the expression for the full energy needs and depends by modulus only on the distance,

to remain unchanged). F⃗ij = (⃗ri − ⃗rj )fij (|⃗ri − ⃗rj |),

27 This is the content of the Noether’s theorem (E. Noether 1918).
28 At non-relativistic speeds, the Lorentz force acting between two moving charges is much smaller than the electrostatic force and is therefore a
relativistic effect; still, the Lorentz force can lead to situations with seeming violation of the conservation of angular momentum, e.g. in the case of a
moving charge at the centre of a ring current.
29 The total differential is defined as dΠ(⃗ r1 , ⃗ rN ) ≡ Π(⃗
r2 , . . . ⃗ r1 + d⃗
r1 , ⃗
r2 + d⃗
r2 , . . . ⃗ rN ) − Π(⃗
rN + d⃗ r1 , ⃗
r2 , . . . ⃗
rN )
30 Partial derivative of a function: while taking derivative with respect to a given variable, all the other variables are assumed to be constant; for
∂f (x,y)
instance, ∂x
denotes the derivative of f (x, y) with respect to x while y is considered to be a constant.

— page 26 —
 5. REVISION PROBLEMS
d⃗
v
and the external force acting upon i-th point mass has a similar Let us recall that
dt is the acceleration of the point P as seen
d⃗
u
property with respect to a reference point at the ⃗ri0 , in the lab frame of reference, and
dτ is the same as seen in the
Fi = (⃗ri − ⃗ri0 )fi (|⃗ri − ⃗ri0 |);
⃗ rotating frame of reference. Now, if P is a point mass m, and

note that due to Newton's 3

rd
law, fij (r) = fji (r). Then, with there is an external force F⃗ acting on P, then ⃗
F = m d⃗
v
dt and

positive values of fij and fi corresponding to repulsion and hence,

d⃗u ⃗ × ⃗um + Ω2⃗rm,
negative values to attraction, the potential energy is given by
∑ ∑ m = F⃗ − 2Ω
dτ
Π= gij (|⃗ri − ⃗rj |) + gi (⃗ri − ⃗ri0 ), where i.e. in the rotating system of reference, the body behaves as if
i<j≤N i≤N ⃗ × ⃗um,
−2Ω
∫ r ∫ r
there were additional forces: the Coriolis force and
2
gij (r) = − fij (r′ )r′ dr′ , gi (r) = − fi (r′ )r′ dr′ (13) the centrifugal force Ω ⃗rm.

(the lower bound of these integrals can be arbitrary). It is not

appendix 5: Stablity and conservation laws.
too dicult to check that with such potential energy, equality
It is well-known that a system is stable at the minimum of
(12) is satised, indeed, for all values of i.
its potential energy. But why? Why is a minimum dierent
If there are also some non-conservative external forces present from a maximum? In the case of the Fermat' principle, there
then we can separate conservative and non-conservative forces, is a clear dierence: there is no longest optical path between
F ⃗′ + F
⃗i = F ⃗ ′′ , leading to
i i ∑ two points  the ray could just go zig-zag , but there is
d(K + Π) = F⃗i′′ d⃗ri , denitely one which is the shortest!
i
⃗ ′′ denotes the sum of all the non-conservative forces The reason is simple  at an equilibrium state, the kinetic
where F i
energy has always minimum (as long as masses are positive).
acting on the i-th point mass.
What we actually do need for a stability is a conditional ex-
tremum of one conserved quantity (such as the net en-
appendix 4: Centrifugal force and Coriolis force. ergy), under the assumption that the other conserved
Consider a system of reference, which rotates around the origin
quantities are kept constant (unconditional extremum is
O with an angular velocity ⃗
Ω (the vector denes the rotation
OK, too). Consider the motion of a body along x-axis and let
axis according to the corkscrew rule). Consider a point P,
us describe it on the phase plane, with coordinates x and p (the
which is motionless in the rotating system, and let us denote
−−→ momentum). The overall energy is E = U (x) + p2 /2m. Now,
⃗r = OP . In the lab system of reference, the point P moves
if we depict this energy as a surface in 3-dimensional space
with velocity v = rΩ, and when studying the direction of the
with coordinates x, p and E, the point describing the state of
v = d⃗
velocity ⃗
r ⃗ r. Now, if the point P
dt , one can see that ⃗
v = Ω×⃗
the system will move along the intersection line of that surface
d⃗
r
moves in the rotating frame of reference with velocity ⃗ u = dτ
with a horizontal plane E = Const. At the minimum of U (x),
(let us use τ to measure the time in the rotating system), then
with p = 0, this intersection line would be just a single point,
this additional velocity needs to be added to what would have
because this is the lowest point of that surface. The near-by
been for a motionless point:
trajectories will be obtained if we ascend the horizontal plane
d⃗r d⃗r ⃗
= + Ω × ⃗r. a little, E = Emin + ε, so that it no longer just touches the
dt dτ
So, we can conclude that the time-derivatives of vectors in ro- surface, but cuts a tiny ellips from it. All the points of that

tating and lab frames of reference are related via equality trajectory (the ellips) are close to the equilibrium point, so the
d d ⃗ ×. state is, indeed, stable.
= +Ω
dt dτ
This is written in the form of an operator, which means that
we can write any vector (e.g. ⃗r or ⃗v ) rightwards of all the three
terms. In particular, we can apply this formula to the right-
and left-hand-sides of the equality ⃗ × ⃗r:
⃗v = ⃗u + Ω
( )( )
d⃗v d ⃗ ⃗ × ⃗r
= + Ω× ⃗u + Ω
dt dτ
d⃗u ⃗ d(Ω × ⃗r) ⃗ ( ⃗ )
= + Ω × ⃗u + + Ω × Ω × ⃗r .
dτ dτ
Here we need to bear in mind that when taking derivatives of
vectors and products of vectors, all the well-known rules can
a × ⃗b) = d⃗ ⃗ a × d⃗b and It appears that a system can be stable also because of a con-
dt × b + ⃗
d a
be applied; in particular,
dt (⃗ dt
⃗
⃗b) = d⃗a · ⃗b + ⃗a · db . We also need the rule for the double ditional maximum of the net energy: while an unconditional
d
dt (⃗
a · dt dt
extremum of the kinetic energy can only be a minimum, things
cross product, ⃗a × (⃗b × ⃗c) = ⃗b(⃗a · ⃗c) − ⃗c(⃗a · ⃗b); you can memorize
are dierent for conditional extrema. Perhaps the simplest ex-
this equality by keeping in mind that the double product is a
ample is the rotation of a rigid body. Let us consider a rectan-
linear combination of the vectors from the inner braces, and
gular brick with length a, width b, and thickness c (a > b > c).
that the sign '+' comes with the vector from the middle posi-
⃗
dΩ d⃗
r Let Ic be its moment of inertia for the axis passing its centre
tion. And so, bearing in mind that
dτ = 0 and dτ = ⃗u, and
of mass and perpendicular to the (a, b)-plane; Ib and Ia are
assuming that ⃗ ⃗ ⃗
r ⊥ Ω ⇒ ⃗r · Ω = 0, we obtain
dened in a similar way. For a generic case, the moment of
d⃗v d⃗u ⃗ × ⃗u − Ω2⃗r.
= + 2Ω inertia I will depend on the orientation of the rotation axis,
dt dτ
— page 27 —
 5. REVISION PROBLEMS
but it is quite clear that Ic ≥ I ≥ Ia (it can be shown eas- as small as possible. This integral, referred to as the action, is
ily once you learn how to use tensor calculations). Now, let dened via the full potential and kinetic energies of the system,
us throw the brick rotating into air and study the motion in denoted as V and T , respectively; V depends on the coordin-
a frame which moves together with the centre of mass of the ates, V = V (qi ), i ∈ [1, n], and T also on the changing rate of
brick (in that frame, we can ignore gravity). There are two coordinates q̇i :
conserved quantities: angular momentum L, and rotation en-
∫ τ′
2
ergy K = L /2I . We see that for a xed L, the system has
S= L[qi (t), q̇i (t), t]dt, (14)
τ
minimal energy for I = Ic (axis is parallel to the shortest where
edge of the brick), and maximal energy for I = Ia (axis is L(qi , q̇i , t) = T (qi , q̇i , t) − V (qi , t) (15)
parallel to the longest edge of the brick). You can easily is called theLagrangian of the system, and the postulate itself
check experimentally that both ways of rotation are, indeed, as the principle of least action
31

stable! Meanwhile, if the axis is parallel to the third edge,

Using the methods of variational analysis, one can show that
the rotation is unstable. This phenomenon is demonstrated 32
the integral S has an extremum if
in a video made by NASA on the International Space Station,
d ∂L ∂L
https://mix.msfc.nasa.gov/abstracts.php?p=3873. = ; (16)
dt ∂ q̇i ∂qi
Well, actually the rotation with the minimal energy is still a here
∂L
means that we take derivative of the Lagrangian
∂ q̇i
little bit more stable than that of with the maximal energy; L(qi (t), q̇i , t) with respect to only one of its 2n + 1 variables, q̇i ,
the reason is in dissipation. If we try to represent the motion while considering all the other variables to be constant.
∂L
∂qi is
of the system in the phase space (as described above), a bowl- d
dened analogously. Meanwhile,
dt denotes taking a full time
shaped energy surface (as shown in the gure above) would be
derivative, i.e. we take into account that L depends on time
substituted by a hill-shaped one; at the equilibrium, the phase
both explicitly through its last argument t, as well as impli-
trajectory is contracted into a point  the point where the
citly since the quantities qi and q̇i are also functions of time.
top the hill is touching a horizontal plane E = Emax . Due
Please note that Eq. (16) is valid for every i, so that we have
to dissipation, the energy will decrease, E = Emax − ε, and
a system of n equations. From the principle of least action to
the phase trajectory would be a slowly winding-out spiral. So,
Equation (16), there is only one mathematical step, so we can
while you are probably used to know that dissipation draws a
say that analytical mechanics basically postulates the Equation
system towards a stable state, here it is vice versa, it draws the
(16).
system away from the stable state! This is what is known as
dissipative instability. Which way is better: the historical way of postulating the New-
ton's laws, or postulating Eq. 16? Both approaches have strong
and weak points. While the classical approach is built up step-
appendix 6: Lagrangian formalism. by-step, from immediate experimental ndings, the approach
In our approach to mechanics, we postulated the Newton's laws;
of analytical mechanics takes the least action principle out of
based on that, we derived energy conservation law which is
thin air. Meanwhile, Eq. (16) gives us a very universal and
valid for conservative forces, and using energy conservation law,
powerful tool for theoretical analysis (the usage of which is not
we arrived at the method of generalized coordinates.
limited to mechanics): as soon as we have an expression for the
In analytical mechanics, the order is opposite. First, we postu- Lagrangian, we can write down the evolution equation describ-
late that any mechanical system has a certain potential energy, ing how the system will evolve. However, it should be kept in
and a certain kinetic energy, both of which are additive; we mind that only in the case of classical mechanics, L = T −V,
also establish a formula for kinetic energy of point masses, and and one should keep vigilance even in the case of classical mech-
for potential energies of point mass interactions depending on anics (see below).
the type of interaction (this is done similarly to how we estab-
As a matter of fact, the least action principle can be introduced
lished rules for calculating forces for dierent interaction types
more naturally (not our of thin air) using quantum mechanics.
in chapter 2).
Indeed, if we consider a point mass as a quantum-mechanical
Second, let us consider a mechanical system obeying n degrees probability wave then using quasi-classical approximation, we
of freedom, i.e. in order to specify uniquely the state of the can express the phase of the wave as
∫ ∫
system, we need n parameters. We postulate that if this sys-
φ= (⃗k · d⃗r − ωdt) = ~−1 p · d⃗r − Edt);
(⃗ (17)
tem evolves from one state described by a set of coordinates
qi , i ∈ [1, n] at time t = τ [this state corresponds to a point here p⃗ E  the energy of the particle.
is the momentum and
in the n + 1-dimensional conguration space with coordinates r = ⃗v dt and ⃗v · p⃗ = 2T then we can
If we keep in mind that d⃗
∫
(q1 , q2 , . . . , qn , t)] to another state qi′ at t = τ ′ then the evolu- further write φ = ~ [2T − (T + V )]dt = ~S . So, the action S
tion of the system in time takes place along such a path qi (t) gives us directly the phase of the wave. The waves add up con-
(a curved line connecting the initial and nal states in the con- structively if they arrive at the same phase, and many waves
guration space) that makes the value of a certain integral S coming along dierent paths arrive almost at the same phase
31 Also as the principle of a stationary action.
32 More precisely, functional — a scalar quantity which depends on which function(s) qi (t) we have.
33 We can see rainbow exactly due to the same reason: the rainbow arc can be seen because the angle by which a light beam is deflected after a

reflection inside a spherical water droplet has an extremum (as a function of the aim parameter); hence, near the deflection angle extremum, a wide
range of aim parameters corresponds to a narrow range of deflection angles.

— page 28 —
 5.
if these paths are close to the path of least action. It should
be noted that exactly the same phenomenon happens in the
case of light propagation, and can be summarized as the Fer-
33
mat' principle. We can say that according to the Huygens
principle for wave propagation, the amplitude of the probabil-
ity wave can be found as the sum over the contributions from
all the possible ray tracing paths; however, majority of these
contributions cancel out due to opposite phases, and only the
contribution of the optimal path (and its immediate neigh-
bourhood) is left intact; optimal means corresponding to an
extremum (which appears to be a minimum) of the action. So,
we can say that a point mass moves along the trajectory of
least action.
REVISION PROBLEMS
action. In this conguration space, the Lagrangian needs to
account for the inter-molecular interaction energies, as well.
While the expression for the inter-molecular interaction ener-
gies may be fairly complicated, as long as we are interested only
in the macroscopic dynamics, we just need to x the distances.
The distances can be xed with a simplied Lagrangian: we
say that the inter-molecular interaction energy is zero, if the
distance between two molecules equals to what it should be,
and becomes very large otherwise. Due to the inter-molecular
distances being xed, the state of this system of molecules can
be fully described by six generalized coordinates; this means
that all the trajectories in the 3N + 1-dimensional congura-
tion space are constrained into a six-dimensional hypersurface

Now, let us check if the least action principle is in agreement

M (points A and B also need to lie on that hypersurface). We

with the Newton's laws. To this end, let us consider a system of

know that the trajectory σ minimizes the action between A
point masses mi , i ∈ [1, n], and use the ordinary Euclidian co- and B in the 3N + 1-dimensional conguration space; the hy-
ordinates: let ⃗ri point to the position of i-th point mass. Then persurface M is a part of that space, so it surely minimizes
the action between A and B in the hypersurface M, as well.
we dene (postulate) the Lagrangian as
1∑ Therefore, Eq. (16) must remain valid when we use six general-
L= mi⃗vi2 − Π(⃗r1 , ⃗r2 , . . .);
2 i,j ized coordinates to describe the state of a rigid body. Similar

d⃗
r argumentation works not only for a rigid body, but for any con-
here we have denoted ⃗vi =
dt and assumed that all the inter-
straints xing relative positions of the parts of a system (and
action forces are conservative: Π(⃗
r1 , ⃗r2 , . . .) denotes the total
thereby reducing the number of degrees of freedom).
potential energy as a function of coordinates of all the particles.
Then, if we apply Eq. 16 to this Lagrangian, and keep in mind From the discussions of the previous paragraph we can derive
∂L
that
∂vix = mi vix (where index x denotes a projection of a also an important rule: if we write Lagrangian using gener-
vector to the x-axis), we obtain alized coordinates for a system with intrinsic constraints, the
d ∂Π(⃗r1 , ⃗r2 , . . . ⃗rN ) number of coordinates should be as small as possible. For in-
mi vix = − .
dt ∂xi stance, if we have a rigid body, we should use six coordinates
According to Eq. (12), what we have at the right-hand-side is
and not seven, because the value of the seventh coordinate can
the x-component of the force acting on the i-th particle (obvi-
be derived from the rst six (with the seventh coordinate, we
ously a similar expression is obtain for y - and z -components).
would need to add additional term to the Lagrangian xing the
So, we conclude that if written for Euclidian coordinates, equa-
value of the seventh coordinate).
tion (16) is equivalent to the Newton's laws. Meanwhile, equa-
tion (16) being satised is equivalent to the least action prin- So, we have now two alternative options: we can use the Lag-

ciple being valid. Now, let us notice that the least action prin- rangian equation (16), and we can use the method 6 in which

ciple is formulated independently of the coordinate system: if case we derive the equation of motion from the energy conserva-

a certain trajectory ⃗ri = ⃗ri (t) has minimal action in Euclidian tion law. These two approaches are fairly similar: in both cases

coordinates ⃗ri then it remains being minimal even if expressed we need to express the kinetic and potential energies in terms

in terms of the generalized coordinates qi = qi (t). Since the tra- of generalized coordinates and time derivatives of these. How-

jectory has an extremal action in terms of generalized coordin- ever, there are also certain dierences: in one case, we derive

ates qi then (according to the results of the variational analysis), the equation of motion directly from the energy conservation

the Lagrangian equation (16) must be also valid when the gen-
34
eralized coordinates qi are used. This completes our proof
that Newton's laws and Eq. (16) are equivalent.
Although we kind of completed the proof, we need to make a
comment regarding the cases when the number of degrees of
freedom is recuced due to various constraints. As an example,
let us consider a rigid body made of N molecules; this set of
molecules has 3N degrees of freedom. However, the relative dis-
tances between molecules are xed by molecular forces, so that
there are only six degrees of freedom left: three numbers x the
position of the centre of mass, and the orientation of the body is
xed by three angular coordinates. Previously we have proved
the least action principle for a set of point masses (molecules),
so we know that our system evolves in the 3N + 1-dimensional
conguration space along such a curve σ connecting the start-
ing point A with the destination point B which minimizes the
34 From which we dropped the mathematical piece using variational analysis
law; in the other case we consider the dierence of these two
energies and apply a formula which we can either consider to
be postulated, or derived from the Newton's laws in a fairly
complicated way.
Which way is better? To begin with, it should be emphasized
that while Eq. (16) can be always used, the method 6 based
on energy conservation law can be applied only in those cases
when there is a single degree of freedom, i.e. the state of the
system can be described with only one generalized coordinate.
Indeed, upon taking time derivative of the energy conservation
law, we obtain one dierential equation, but we need as many
equations as there are unknown functions (degrees of freedom).
However, for a majority of Olympiad problems, this condition
is satised (keep in mind that each additional conserved quant-
ity, e.g. momentum, reduces the eective number of degrees of
freedom by one).

— page 29 —
 5. REVISION PROBLEMS
So, let us compare these two methods when we have one gen- state of the system. If we use these two coordinates with the
eralized coordinate q, and let us assume that the energies corresponding Lagrangian then everything will be correct: the
do not depend explicitly on time. In the case of Newtonian action
∫ t2 [ ]
mechanics, kinetic energy is proportional to squared speed, mẋ2 mφ̇2 (a + x)2 1 2
+ − kx dt
2 M(q)q̇ . Then, the energy
1 2 4 4 2
so we may assume that T = t1

2 M(q)q̇ + V (q) = E , hence

1 2 is minimized by the true trajectory if we compare the traject-
conservation law states that
′ ′
2 M (q)q̇ + M(q)q̇ q̈ + V (q)q̇ = 0 and
1 3 ories connecting the initial state x1 , φ1 and the nal state x2 ,
1 φ2 . Now, however, we have dropped the variable φ, and if we
M(q)q̈ = − M′ (q)q̇ 2 − V ′ (q). drop the condition for the initial and nal angles, many more
2
Meanwhile, the Lagrangian is expressed as L = M(q) −
1 2
q̇ trajectories will connect the initial state x1 with the nal state
2
′
d ∂L
V (q); then, with dt ∂ q̇ = d
dt M(q)q̇ = M (q)q̇ 2
+M (q)q̈ , Eq. (16) x2 : the true trajectory does no longer need to be the one with
is rewritten as the smallest action. An important lesson from this analysis is
1 ′
M ′ (q)q̇ 2 + M (q)q̈ = M (q)q̇ 2 − V ′ (q). that don't use Eq. (16) if you reduce the number of de-
2 grees of freedom by making use of a constraint (a con-
It is easy to see that we obtained in both cases the same
servation law) which involves time derivatives of the
equation, and that mathematically, diculty level was almost
coordinates, because by xing the value of a time derivative
the same. However, we needed to memorize Eq. (16), which
we do not x the value of the coordinate itself. If you have such
makes the method based on the energy conservation law slightly
conservation laws and manage to nd so many constraints that
easier.
you can bring the number of coordinates down to one, go ahead
Before we make any nal conclusion, let us consider a system
and use the method 6; otherwise keep the original number of
of two balls of mass m, connected with a spring of length a coordinates and apply Eq. (16).
and stiness k, rotating with angular momentum ⃗
L (which is
perpendicular to the spring) in weightlessness. Here, it seems Finally, let us also emphasize that the Lagrangian is given by

that we have two degrees of freedom (the angle and length of the dierence of kinetic and potential energies only in the case

the spring), but an additional (to the energy) conservation law of classical mechanics; in other cases, the rst task is to gure

(of angular momentum) reduces the eective number of degrees out the expression for the Lagrangian. How to do it? Basically

of freedom down to one. Let us use the deformation x of the there are two options. Assuming we know already the equa-

spring as the generalized coordinate. Then, tion of motion in Euclidian coordinates xi , we can do it by
2
mẋ 2
L 1 trial and error nding such L(xi , ẋi , t) that Eq. (16) becomes
T = + 2
, Π = kx2 . identical to the equation of motion. Note that the original
4 m(a + x) 2
The remarkable thing here is that the kinetic energy depends equation of motion does not need to have origins in physics.

now not only on ẋ, but also on x; in eect, the second term However, once we have found the corresponding Lagrangian,

of the kinetic energy behaves as a potential one, and can be we can interpret it physically: for instance, if the Lagrangian

combined into an eective potential energy in the expression obeys translational symmetry, we can use the Noether's the-
35
for the full energy. Following the method 6, we obtain orem to nd a conserved quantity and call it momentum . In
2 2 electromagnetism, we'll use this method to derive generalized
1 2L 4L k
mẍẋ − 3
ẋ + kxẋ = 0 ⇒ ẍ = 2 3
− 2 x. momentum of a charged particle in magnetic eld.
2 m(a + x) m (a + x) m
Further, let us try to obtain the same result using the Lag-
The second option works if we study a system which can be con-
rangian (NB! This will be wrong!):
sidered quantum-mechanically; let us illustrate this by consid-
2 2
mẋ L 1
L= + 2
− kx2 , ering a relativistic point mass. We know that the least action
4 m(a + x) 2
principle in mechanics corresponds to the Huygens principle
hence
[see Eq. (17)] and hence, the action must be the phase of the
1 2L2 4L2 k
mẍ = − 3
− kx ⇒ ẍ = − 2 − 2 x. quantum-mechanical probability wave, multiplied by ~  in
2 m(a + x) m (a + x)3 m
that case the classical action would be the small-speed-limit of
This is not the same result as before  the rst term in right
hand side has a dierent sign! So, what went wrong? The rst
the relativistic one. So, with m denoting the relativistic mass

result is clearly the correct one as the total energy is clearly

andm0 the rest mass of a particle,
∫ ∫ ∫
conserved here. What went wrong is that by making use of
S = (⃗
p ·⃗v −E)dt = (mv −T −V )dt = [m(v 2 −c2 )−V ]dt,
2
the angular momentum conservation law to reduce the number
of coordinates we changed the starting and ending points in hence
√
the conguration space. As we proved above, the least action v2
principle [and hence, Eq. (16)] is valid if we don't use conser-
L = m(v − c ) − V = −m0
2 2
1− − V.
c2
vation laws to reduce the number of degrees of freedom, and
It is easy to verify that if we put this Lagrangian into Eq. (16),
all the conservation laws themselves are to be considered as
by keeping in mind that V = V (x, y, z) and ⃗v = (ẋ, ẏ, ż), we
the consequence of Eq. (16). In this case, the original number
obtain relativistic Newton's 2
nd law. As we can see, there is no
of degrees of freedom was two: we can use the deformation
kinetic energy now included into the Lagrangian.
x and the rotation angle φ of the spring to describe fully the

35 As ∂L d
a matter of fact, in the case of translational symmetry we don’t even need to use the Noether’s theorem: ∂qi
= 0, hence p
dt i
= 0, where
∂L
pi = ∂ q̇i
is the i-th component of the momentum.

— page 30 —
 6. HINTS
6 HINTS tion force must be zero. Use the rotating frame of a wheel;
apply the ideas 12 and11 to substitute one asymmetric body
1. Write out the balance of torques for the contact point O of
(the cylinder with a hole) with two symmetric bodies, a hole-
the hoop and the shaft. What is the angle that the tangent
less cylinder, and a superimposed cylinder of negative density;
to the shaft at point O forms with the horizon (given that the
further use ideas 9 and 10 to draw the gravity and centrifugal
wire slips on the shaft)?
forces; keep in mind that the rod can provide any horizontal
2. Write down the equation for the torques for the cylinder & force, but cannot exert any vertical force.
block system with respect to the contact point of the cylinder
8. Based on the idea 14, on which line does the intersection
and the inclined plane. What angle with respect to the hori-
point of the frictional forces have to lie? What can be said
zon is formed by the tangent to the cylinder constructed at the
about the two angles formed by the frictional force vectors and
position of the little block?
the thread’s direction. Given the Idea no. 1 (the axis is per-
3. According to the idea 4, consider the system “rod CD + the pendicular with the tension in the thread)? Now combine the
mass m” as a whole; there are four forces acting on it: m⃗g , F⃗ , two conclusions above. Where is the intersection point of the
and the tension forces of the rods, T⃗AC and T⃗BD . The tension friction force vectors? What is the direction of the cylinder’s
forces are the ones which we don’t know and don’t want to velocity vectors at the points where the cylinder rests on the
know. According to the idea 2, these will drop out from the rough band? Where is the cylinder’s instantaneous rotation
balance of torques acting on the rod CD with respect to the axis (see how to find it in the kinematics brochure)? What is
intersection point of AC and BD. Indeed, due to the fact 20, the velocity vector of the cylinder’s centre point? (b) Will the
the tension force in the rod AC is parallel to AC; the same equilibrium condition found above be violated if the surface is
applies to the rod BD. Now, what must be the torque of force uniformly rough?
F ? For what direction of the force will this torque be achieved
9. Draw a circle whose diameter is the straight line connect-
with the minimum magnitude?
ing the points of support. Use Fact no. 22: which curve can
4. The vector sum of the forces F⃗ and m⃗g has to compensate the ball move along? Where is the bottom-most point of this
the sum of the friction and the normal force f⃗ = N ⃗ + F⃗h , i.e.
curve?
has to be at an angle arctan µ with respect to the normal to the
plane. Let us draw the force triangle m⃗g + f⃗ + F⃗ = 0: the vec- 10. Consider the torques acting on the rod with respect to the
tor m⃗g can be drawn immediately (its direction and magnitude hinge. For which angle α will the net force of the normal and
are known), the direction of f⃗ can be noted by a straight line frictional forces push the rod harder against the board?
passing through the terminal point of m⃗g . F⃗ has to connect 11. By how much will the block descend if the thread is exten-
that straight line to the initial point of m⃗g . For which direction ded by δ?
is its magnitude minimal? 12. Let’s assume that the horizontal component of the tension
5. Go to the reference frame of the inclined surface (invoke in the rope is Tx . What is the vertical component of the ten-
Ideas 7 and 8) and use the same method as for problem 4 sion next to the ceiling? Next to the weight? Write down the
(⃗a + ⃗g functions as the effective gravity ⃗ge ). condition for the balance of the forces acting on a) the weight
6. Use a rotating reference frame associated with the cylin- and b) the system of weight & rope (cf. Idea no. 4).
der (where the block is at rest, and the centrifugal force f⃗t is 13. Seeing as H ≪ L, clearly the curvature of the rope is small,
constant and pointing downwards). (a) The terminal point of and the angle between the tangent to the rope and horizon re-
the net force of gravity and centrifugal force is moving on a mains everywhere small. From the horizontal force balance
circle and has to be equal to the net force f⃗ of the normal and for the rope, express the horizontal component of the tension
frictional forces. What is the maximum allowed angle between force Tx as a function of the length l (note that while Tx re-
the vectors f⃗t and f⃗ so that there be no slipping? For which mains constant over the entire hanging segment of the rope,
direction of m⃗g is the angle between the vectors f⃗t and f⃗ max- we’ll need its value at the point P separating the hanging and
imal? (b) There are still only three forces; as long as there is lying segments). Write down the balance of torques acting on
an equilibrium, these three vectors must form a triangle and the hanging piece of the rope with respect to the holding hand
hence, must lay on the same plane. According to the idea K- (according to what has been mentioned above, the arm of the
11, we’ll depict the force balance in this plane, i.e. in the plane gravity force can be approximated as l/2). As a result, you
defined by the vectors ⃗g and f⃗t . The approach used in part (a) should obtain a quadratic equation for the length l.
can still be used, but the terminal point of f⃗t + m⃗g draws only 14. Use Idea 9: change into the reference frame of the rotating
an arc of a full circle. Determine the central angle of that arc. hinge. a) Following the idea 19, write down the condition of
Depending on the arc length, it may happen that the maximal torque balance with respect to the hinge (Idea no. 2) for a small
angle between the surface normal (= the direction of f⃗t ) and deviation angle φ. Which generates a bigger torque, m⃗g or the
f⃗ is achieved at one of the endpoints of the arc. centrifugal force? (Note that alternatively, the idea 21 can be
7. Notice that while rolling at constant speed, the centre of also used to approach this problem). b) Following the idea 21,
mass of the whole cart moves also with a constant speed, i.e. express the net potential energy for the small deviation angles
there should be no horizontal forces acting on the cart. Also, φ1 and φ2 using the energy of the centrifugal force (which re-
each of the cylinders rotates with a constant angular speed, sembles elastic force!) and the gravitational force; according to
hence there should be no torque acting on it, hence the fric- the idea 20, keep only the quadratic terms. You should obtain
— page 31 —

a quadratic polynomial of two variables, φ1 and φ2 . The equi-
librium φ1 = φ2 = 0 is stable if it corresponds to the potential
energy minimum, i,e, if the polynomial yields positive values
for any departure from the equilibrium point; this condition
leads to two inequalities. First, upon considering φ2 = 0 (with
φ2 ̸= 0) we conclude that the multiplier of φ21 has to be posit-
ive. Second, for any φ2 ̸= 0, the polynomial should be strictly
positive, i.e. if we equate this expression to zero and consider it
as a quadratic equation for φ1 , there should be no real-valued
roots, which means that the discriminant should be negative.
15. Apply the ideas 19 and 22 for such a angular position of
the beam, for which the magnitude of the buoyant force doesn’t
change (i.e. by assuming a balance of vertical forces). From
idea no. 2, take the centre of mass for the pivot point. While
computing the torque of the buoyant force, use Ideas 11, 12: if a
certain region has no displaced water, the displaced water dens-
ity is zero, but it can be represented as overlapping negative
and positive mass densities: 0 = ρw +(−ρw ). The cross-section
of the underwater part of the beam could be represented as a
superposition of a rectangle and two symmetrically positioned
narrow triangles (one of them of negative mass).
16. The container & water system is affected by the gravity
and the normal reaction force of the horizontal surface on the
liquid. Since we know the pressure of the liquid at the base of
the container, we can express the mass of the container from
the vertical condition for equilibrium.
17. To compute the first correction using the perturbation
method we use the Fact 51 and the reference system of the
block sliding down uniformly and rectilinearly: knowing the
magnitude and the direction of the frictional force we can find
its component in w ⃗ and ⃗u direction. The sign of the latter flips
after half a period, and so it cancels out upon averaging.
18. Let us choose the origin of the vertical x-axis to be a point
on the surface of the ocean very far from the iron deposit. For
the zero reference point of the Earth’s gravitational potential
we shall choose x = 0 (i.e. φearth = gx), for that of the iron
deposit we shall take a point at infinity. Then, for the points
on the ocean’s surface very far from the iron deposit, the grav-
itational potential is zero. It remains to find an expression for
the potential above the iron deposit as a function of x (using
the principle of superposition) and equate it to zero.
19. Let us employ the reference frame of the platform. Let us
the consider the balance of torques with respect to the axis of
the small disk (then the lever arm of the force exerted by that
axis is zero). Let us divide the disk into little pieces of equal
size. The frictional forces acting on the pieces are equal by mag-
nitude and are directed along the linear velocities of the points
of the disk (in the chosen reference frame). Since the motion of
the disk can be represented as a rotation around an instantan-
eous axis, then concentric circles of frictional force vectors are
formed (centred at the instantaneous rotation axis). Clearly,
the net torque of these vectors with respect to the disk’s axis
is the smaller, the smaller is the circles’ curvature (i.e. the
farther the instantaneous rotation axis is): the torque is zero
when the instantaneous rotation axis is at infinity and the con-
centric circles become parallel straight lines. An instantaneous
rotation axis at infinity means that the motion is translational,
— page 32 —
6. HINTS
ω3 = 0 (since the linear velocity v = ω3 r of a given point is
finite, but r = ∞).
20. The instantaneous axis of rotation is at a distance r = v/ω
from the disk’s axis. Let’s use the same imaginary slicing as
in the previous problem. Now compute the component of the
net force in the direction of motion. Notice that the frictional
forces on the points that are symmetrical with respect to the
instantaneous rotation axis balance each other across a whole
circular region of radius R − r. The non-balanced region is un-
fortunately shaped for calculation. Let us imagine extending
the "balanced" region up to R (the dashed circle in the fig-
ure). The part of this extended balanced region, where there
is no actual rotating disk underneath (the dark gray crescent
in the figure), could be represented as a superposition of the
two disks, one rotating clockwise and the other – anticlock-
wise. In that case the clockwise component partakes in the
balancing, whereas the anticlockwise component remains un-
balanced. To sum up, two thin crescent-shaped regions remain
unbalanced: one corresponds to the the real disk (light gray in
the figure), the other — to a disk rotating anticlockwise (dark
gray); normal to ⃗v , the width of these regions is everywhere
equal to r. The net force is the easiest to find by integrating
across the crescent-shaped regions using the polar coordinate
φ: |dF⃗ | = A · dS, where dS is the area of the surface element;
∫ ∫ 2π
dFx = A cos φdS = B cos2 φdφ, Fx = dFx = B 0 cos2 φdφ.
What are the values of the constants A and B?
r
O
21. Consider the unit vector ⃗τ directed along the infinitesimal
displacement vector of the centre of the mass at the instant
when the pencil begins moving. Let’s express its coordinates
in the Cartesian axes (x, y, z), where x is parallel to the pen-
cil and the (x, y)-plane is parallel to the inclined slope. Using
the spatial rotations formulae we represent it in the new co-
ordinates (x′ , y ′ , z), which are rotated with respect to (x, y, z)
around the z-axis by an angle φ (so that the axis x′ is hori-
zontal). Using the spatial rotations formulae we express the
vector’s ⃗τ vertical coordinate z ′ in the (x′ , y ′ , z ′ ) coordinate
axes, which is obtained from the axes (x′ , y ′ , z) by rotating
about the x′ by the angle α.
22. The string connects the two points with the shortest dis-
tance along the cylinder’s side; when unfolded, the cylinder
is a rectangle. Consider the vertical plane touching the sur-
face of the cylinder that includes the hanging portion of the
string. This plane and the cylinder touch along a straight line
s. If you imagine unfolding the cylinder, the angle between the
string and the straight line s is equal to the cylinder’s inclin-
ation angle α. Given this, l is easy to find. When the weight
oscillates, the trace of the string still stays straight on the unfol-
ded cylinder. Therefore the length of the hanging string (and
thus the weight’s potential energy) do not depend in any os-
cillatory state on whether the surface of the cylinder is truly
cylindrical or is unfolded into a planar vertical surface (as long
as the spatial orientation of the axis s is preserved).
 6. HINTS
23. Write down the two equations describing the balance of the second time, the block’s velocity is maximal (why?).
force and torques, and then another one that describes the lin- 31. Let’s apply Idea no. 46 for P⃗ : the system’s net momentum
ear relation between the elongations of the string: T1 − T2 = is P = ωlm + 2ωlM , net force F = (m + M )g − T . The
T2 − T3 . same using rotational considerations: with respect to the left-
24. Initially only the vertical forces affect the hanging block, most ball’s initial position, the angular momentum is l(2ωl)M
therefore the initial displacement vector is also vertical. If the (velocity is 2ωl, the velocity’s lever arm — l); net torque is
acceleration of the large block is a1 , that of the block on top of (T + M g)l. Now, for the formula given in Idea no. 46 we need
it — a2 and that of the hanging block — a3 , then a1 + a2 = a3 the angular acceleration ε = ω̇. Let’s find it using Method no.
holds. Now we can write down Newton’s 2nd law for each body. 6: Π = lφ(m + 2M ), K = 12 φ̇2 l2 (m + 4M ). Another solution
The fourth and the final unknown is the tension in the string. route: the ratio of accelerations is 1:2; there are four unknowns
25. Go to the reference frame of the wedge-block. In the bor- (two normal forces, acceleration and string tension); equations:
derline case, the force of inertia’s and gravity’s net force on the three force balances (for either ball and the rod) and one torque
ball m is normal to the left slope (so that the ball stay at rest balance (wrt the left endpoint of the rod).
there). Consider the net forces acting on the balls. Their com- 32. Method no. 6: for the generalized coordinate ξ we can use
ponents normal to the surface they rest on are F⃗⊥1 and F⃗⊥2 . the displacement of the thread’s endpoint. Ideas no. 34,12: the
These are equal to the normal forces N ⃗ 1 and N⃗ 2 acting on the change of the system’s CM y-coordinate is ξρh/M (h — the dif-
balls and therefore have to have equal magnitudes (F⊥1 = F⊥2 ) ference in the heights of the thread’s endpoints, M — the net
to ensure that the force balance is achieved horizontally for the mass of the system; assume that ξ ≪ h). For the x-coordinate
wedge-block. it’s 2ξρR/M .
26. Let’s take the displacement ξ of the wedge as coordinate 33. ⟨T (1 + cos α)⟩ = 2mg, T = ⟨T ⟩ + T̃ , where |T̃ | ≪ T . Based
⟨ ⟩
describing the system’s position. If the wedge moves by ξ, then on the Idea no. 20 we ignore the tiniest term T̃ α2 and note
⟨ 2⟩
the block moves the same amount with respect to the wedge, be- that α > 0.
cause the rope is unstretchable, and the kinetic energy changes 34. We have to consider two options: either all the bodies move
by Π = mgξ sin α. The velocity of the wedge is ξ˙ and that of together, or the rightmost large block moves separately. Why
the block is 2ξ˙ sin α2 (found by adding velocities, where the two cannot the situations occur where (a) all three components
vectors ξ˙ are at an angle α), therefore the net kinetic energy move separately, or (b) the left large block moves separately?
K = 21 ξ˙2 (M + 4m sin2 α2 ). Then we find Π′ (ξ) = mg sin α and
M = M + 4m sin2 α2 ; their sum gives the answer. 35. After the collision the ball’s trajectories are orthogonal
crossing straight lines; the angle with respect to the initial tra-
27. Again, let’s take the wedge’s displacement as the coordin- jectory is determined by how much the collision was off-centre.
ate ξ; if the displacement of the block along the surface of
the wedge is η, then the centre of mass being at rest gives 36. For slightly non-central motion: what will be the direc-
η(m1 cos α1 + m2 cos α2 ) = (M + m1 + m2 )ξ. From here one tion of momentum of the ball that was first to be hit? Now
can extract η as a function of ξ, but to keep the formulae brief apply the Idea no. 52 again. Central motion: express the ve-
it’s better not to substitute this expression everywhere. The locities after the collision via the horizontal component of the
kinetic energies of the block can be found as sums of horizontal momentum px that has been transferred to one of the balls.
[ 21 mi (ξ˙ − η̇ cos αi )2 ] and vertical [ 12 mi (η̇ sin αi )2 ] energies. What is the transferred vertical component py ? Energy conser-
vation provides us an equation to find py (it is convenient to
28. When writing down energy conservation, note that the express the energy as p2 /2m).
block’s velocity is twice the cylinder’s velocity horizontal com-
ponent and that the latter is equal to the vertical component, 37. The graph looks like n intersecting straight lines; the inter-
too (why?). Project Newton’s 2nd law onto the axis that passes section point of a pair of straight lines corresponds to a collision
through the top corner of the step and the cylinder’s centre: of two balls (the graph of either ball’s motion is a jagged line;
this axis is perpendicular both to the normal force between at a collision point the angles of the two jagged lines touch one
the block and the cylinder and to the cylinder’s tangential ac- another so that it looks as if the two straight lines intersect).
mv Mv
celeration. Second question: the ratio of two normal forces 38. Initial velocities in the centre of mass: m+M , m+M , final
is constant (why? what is it equal to? Hint: compare the velocities are zero; friction does work: µmgL.
horizontal accelerations of the cylinder and the block and re- 39. Based on the figure we immediately obtain (to within a
member Newton’s 2nd law), therefore they will be equal to zero multiplicative constant) the magnitudes and directions of the
at the same instant. momenta, but not which momentum is which ball’s. It is ne-
29. By projecting Newton’s 2nd law on the axis in the direction cessary to find out where the ball marked with an arrow will
of the normal force we see that the normal force is the smallest proceed after the collision. Fact no. 27 will help choose from
at the bottommost point of the trajectory’s arch-shaped part. the three options.
(There, the centripetal acceleration is the largest, gravitational
40. Energy: in time dt the distribution of the liquid will change:
force’s component along the axis is the smallest). there is still some water at the centre, but a certain mass dm
30. The energy of the "pellet & block" system is always con- has been displaced from above to the level of the tap (and then
served; momentum will only start to be conserved once the through the tap), so the change in the system’s potential en-
pellet passes the bottommost point. When it arrives there for ergy is gH · dm. Momentum: the water in the barrel obtains
— page 33 —
 6. HINTS
the total momentum ρgHS ·dt from the walls. This momentum must accommodate at least two points out of the three, while
is passed on to the stream of water with the mass ρSv · dt. the circles’ centre (the hanger’s centre of mass) must lie in-
41. Energy is not conserved: the grains of sand slip and exper- side the region bounded by the hanger’s wires on its symmetry
ience friction. In time dt the sand landing on the conveyor belt axis. There is only one pair of circles that satisfies all these
receives momentum dp = v · dm = vµ · dt from the belt: the conditions. Computing the radii l1 and l2 of the circles using
force between the freshly fallen sand and the belt is F1 = dp/dt. trigonometry we determine the reduced length of the pendulum
The sand already lying on the belt experiences the gravitational l1 + l2 and, using that, the oscillation period.
force mg which is compensated by the component of the fric- 52. The effective mass of the moving water can be found using
tion parallel to the belt, F2 = mg cos α, where m = σL is the the acceleration of the falling ball. For the rising bubble the ef-
mass of the sand on the belt and σv = µ. The minimization fective mass is exactly the same, the mass of the gas, compared
has to be done over v. to that, is negligibly small.
√
42. During the collision ∆p⊥ = 2gh. 53. The water stream could be mentally divided into two parts:
the leftmost stream will turn to the left upon touching the
43. Consider a short section of the path along the hill with
trough, the rightmost — to the right. Thus, two imaginary
length dl. In addition to the change in the potential energy
’water tubes’ form. In either tube the static pressure is equal
work is done to overcome friction, dAh = µmg tan α · dl. WE
to the external pressure (since there is the liquid’s outer sur-
find dAh = C · dx, where C is a constant. Summing over all
face in the vicinity): according to Bernoulli’s law, the velo-
such little path increments dl we find Ah = C∆x.
city of the liquid cannot change. Based on the conservation of
44. The kinetic energy K = m 2 2
2 ẋ + M ẋ , where x is the dis- momentum horizontally, the momentum flows of the left- and
placement along the slanted surface; Π = (M + m) sin α. Hav- right-flowing streams have to add up to the original stream’s
ing found the acceleration a we change into a reference frame momentum flow’s horizontal component. Note that due to con-
(of the cylinder) moving with acceleration a (Ideas no. 7 and tinuity, µ = µv + µp .
8), where the block is being displaced along the effective accel-
54. Due to continuity (u + v)(H + h) = Hu Const, where
eration due to gravity — as low as possible.
h = h(x) is the height of the water at point x and v = v(x)
45. According to the Ideas no. 61 and 62, the angular mo- is the velocity. We can write down Bernoulli’s law for an
mentum of the rod before the collision is L0 = M lv − 13 M l2 ω; imaginary ’tube’ near the surface (the region between the
after the collision L1 = M lv ′ − 13 M l2 ω ′ ; L1 = L2 . The expres- free surface and the stream lines not far from the surface):
sion for energy is K = 21 M v 2 + 16 M l2 ω 2 . The condition for 1 ρ(u + v)2 + ρg(H + h) = 1 ρu2 + ρgH = Const. We can ignore
2 2
being at the end: v ′ + lω ′ = 0 (we consider ω to be positive if that small second order terms (which include the factors v 2 or
the rotation is in the direction marked in the figure). vh)
46. The angular momentum with respect to the impact point 55. The phase trajectory is a horizontal rectangle with sides L
before the collision: mv(x − 2l ) − I0 ω, where v = ω 2l and and 2mv, where L is the distance from the block to the wall;
1
I0 = 12 ml2 . the adiabatic invariant is thus 4Lmv.
47. The instantaneous rotation axis passes the contact point of 56. Consider the balance of torques. For the net force vec-
the cylinder and the floor; its distance from the centre of mass tors of the normal and frictional forces, when you extend them,
does not change, so we can use Idea no. 65; I = 32 mR2 . their crossing point must be above the centre of mass.
48. Let us direct the z axis upward (this will fix the signs of the 57. Let’s write down Newton’s 2nd law for rotational motion
angular momenta). The final moment of inertia with respect with respect to the crossing point of the normal forces: the an-
to the x-axis is − 75 mvy R − muR and with respect to the y-axis gular momentum of the bug is L = mvl sin α cos α, the speed of
is 75 mvx R. change of this angular momentum will be equal to the torque
49. Immediately after the first collision the centres of masses due to gravity acting on the bug (the other forces’ lever arms
of both dumbbells are at rest, the velocities of the colliding are zero). When computing the period, note that the accel-
balls reverse direction, the non-colliding balls’ velocities don’t eration is negative and proportional to the distance from the
change. Both dumbbells act like pendula and complete half bottom endpoint, i.e. we are dealing with harmonic oscilla-
an oscillation period, after which the second collision occurs – tions.
analogous to the first one. 58. The blocking occurs if the net force of normal and frictional
50. The grains of sand perform harmonic oscillations in the forces pulls the rod downwards.
plane perpendicular to the cylinder’s axis — like a mathemat- 59. Once the blocking occurs we can ignore all the forces
ical pendulum of length l = R in the gravitational field g cos α;
apart from normal and frictional ones. Suppose it has occurred.
along the axis there is uniform acceleration (a = g sin α). Fo-Then the net frictional and normal forces acting from the left
cussing occurs if the time to cross the trough along its axis is
and from the right have to balance each other both as forces
an integer multiple of the oscillation’s half-period. and torques, i.e. lie on the same straight line and have equal
51. Observing the equilibrium position we conclude that the magnitudes. Thus we obtain the angle between the surface
centre of mass lies on the symmetry axis of the hanger. The normal and the net force of friction and normal force.
three suspension points must be located on the two concentric 60. Consider the direction of the torque acting on the plank
circles mentioned by Idea no. 69. Therefore one of the circles with respect to the point of contact, when the plank has turned
— page 34 —
 6. HINTS
by an angle φ: the contact point shifts by Rφ, the horizontal 70. Using Newton’s 2nd law investigate whither the system’s
coordinate of the centre of mass shifts by the distance h2 φ from centre of mass will move — to the left or to the right (if the
the original position of the contact point. centre of mass had not move, then the both events would have
happened at the same time).
61. The only force from the surface on the system vessel &
2
water is equal to the hydrostatic pressure ρghπR ; it balances 71. To answer the first part: show that the force perpendicular
the gravitational force (m + ρV )g. Note that H = R − h. to velocity is zero (use Method no. 3 and Idea no. 27). To
answer the second part use Method no. 3 and idea 56.
62. The gravitational potential of the centrifugal force is 12 ω 2 r2 ,
where r is the distance from the rotation axis. 72. Due to the length of the thread there are no horizontal
forces, i.e. the horizontal component of momentum is con-
63. Assume the reference frame of the large block (which moves
served, and so is the energy. From the two corresponding equa-
with acceleration a). Where does the effective gravity (the net
tion the limiting velocity v = v0 can be found, for which the
force of the gravity and the force of inertia) have to be dir-
bottom sphere ascends exactly to the height of the top one.
ected? What is a? With which acceleration does the little
Note that at that point its vertical velocity is zero, cf. Idea no.
block fall in this reference frame? What is the tension T of the
44.
thread? Having answers to these questions we can write down
the equilibrium condition for the large block ma = T (1−sin α). 73. Use Idea no. 51. Options: all block keep together;
everything slides; the top one slides and and the bottom two
64. Let us use the displacement of the sphere (down the in- stay together (why is it not possible that the top two keep
clined surface) as the generalized coordinate ξ. What is the together and the bottom one slides?).
displacement of the sphere (up the other inclined surface)?
Evidently Π = (m − M )gξ sin α. The normal force between 74. Which conservation law acts when the two boys collide
the two bodies can be found by projecting Newton’s 2nd law (during a limited time of collision) — do we consider the colli-
onto the inclined surface’s direction. sion absolutely elastic or inelastic (can momentum be lost and
where? If it is inelastic, where does the energy go?), see Idea
65. Let the displacement of the large cylinder be ξ, the ho- no. 58? After the collision: the common acceleration of the
rizontal displacement of the middle and the leftmost cylinder, two boys is constant, knowing the initial and final velocities
respectively, x and y. What is the relationship between them finding the distance becomes an easy kinematics problem.
given that the centre of mass is at rest? What is the rela-
75. Prove that for a vertical thread the velocity v is max-
tionship between them given that the length of the rods does
imal (by applying Idea no. 44 for the rotation angle of the
not change? From the two equations thus obtained we can ex-
rod show that its angular velocity is zero in that position; use
press x and y via ξ. If we assume the displacement to be tiny,
Idea no. 61). Then it only remains to apply energy conser-
what is the relationship between the vertical displacement z of
vation (remember that ω = 0). For the acceleration a, let
the middle cylinder and the horizontal projection of the rod’s
us use idea 44 and notice that horizontal acceleration of the
length, ξ − x? Knowing these results, applying Method no. 6
centre must be zero; this follows from the Newton’s 2nd law for
is straightforward.
the horizontal motion (there are no horizontal forces at that
66. Where is the small displacement ξ of the sphere directed moment). Further, notice that the vertical coordinate of the
(see Idea no. 31)? What is the displacement of the ring ex- centre of mass is arithmetic average of the coordinates of the
pressed via ξ? Use Method no. 6. endpoints, xO = 12 (xA + xB ); upon taking time derivative we
67. Use Idea no. 40 along with energy conservation by pro- obtain ẋO = 2 and ẍO ≡ a = 2 (keep in mind that xB is con-
ẋA ẍA

jecting the force and the acceleration in the Newton’s 2nd law stant). Hence, the acceleration of O can be found as half of the
radially. vertical acceleration of the rod’s upper end A; this is the radial,
i.e. centripetal component of the acceleration of point A on its
68. Let us use some ideas from kinematics to find the accel-
circular motion around the hanging point. Finally, for the ten-
eration of the sphere (K1, K29 and K2: by changing into the
sion force T we have now one equation T + N − mg = ma, but
reference frame moving with velocity v we find the component
we need still another one. To obtain that, we need to consider
of the sphere’s acceleration along the rod and by noticing that
angular motion of the rod in the frame of O. At the given mo-
the horizontal acceleration of the sphere is zero, we obtain, us-
ment of time, our new frame moves translationally to right with
ing trigonometry, the magnitude of the acceleration). Now use
speed v, and with upwards acceleration a. The torque with re-
Newton’s 2nd law.
spect to the centre of mass is caused only by T and N (gravity
69. Using the velocity v of the sphere we can express the velo- and inertial forces have zero arm). So we can relate T − N to
city of the block at the moment being investigated (bearing in the angular acceleration of the rod via Newton’s 2nd for angu-
mind that their horizontal velocities are equal). Using Idea no. lar motion. To find angular acceleration, let us notice that in
40 we find that the block’s (and thus the sphere’s) horizontal the laboratory frame, the speeds of A and O are equal; indeed,
acceleration is zero; by using Newton’s 2nd law for the sphere both velocities are horizontal (vertical velocity of A is zero be-
and the horizontal direction we conclude that the tension in cause it is the lowest point of its trajectory, and ẋO = ẋ2A ), and
the rod is also zero. From the energy conservation law we ex- using the kinematics idea 35 (projections of the velocities to the
press v 2 and from Newton’s 2nd law for the sphere and the axis direction of the rod must be equal) we can conclude that the
directed along the rod we obtain an equation wherein hides the velocities must be also equal. Thus, in the moving frame, the
solution. speed of point A is zero and so is its centripetal acceleration.
— page 35 —
 7. ANSWERS
Hence, the acceleration must be perpendicular to the rod; we two paddles and never hits a paddle). This allows to com-
know the vertical component 2a of this acceleration and using pute the momentum imparted to the paddle per unit time (i.e.
trigonometry can deduce its modulus. With this acceleration, the force), which ends up being proportional to the difference:
we can find the angular acceleration of the rod. F ∝ (v − u)2 . From there, it is not very hard to find the
76. Find the instantaneous rotation axis (make sure that its maximum of the power F u.
distance from the centre of mass is 12 ). Prove that the centre of 83. In the reference frame of the board the problem is equival-
mass moves along a circle centred at the corned of the wall and ent to the problem no. 52.
the floor, whereas the polar coordinate of the centre of mass 84. Go into the (accelerated)
on that circle is the same as the angle φ between the wall and √ reference frame of the wagon,
where the effective gravity a2 + g 2 is at a small angle with
the stick. Express the kinetic energy as a function of the deriv- respect to the vertical. The load will oscillate yet remain mo-
ative φ̇ of the generalized coordinate φ using the parallel-axis tionless at the end if the cable is vertical at the stopping mo-
(Steiner’s) theorem and express the energy conservation law as ment and the load’s velocity is zero. It is possible when the
ω 2 = f (φ); using Method no. 6 we obtain ε = ω̇ = 12 f ′ (φ). corresponding position is the maximal deviation during the os-
When the normal force against the wall reaches zero, the accel- cillation. Therefore the oscillation amplitude has to be the
eration of the centre of mass is vertical: present this condition same both during the acceleration and deceleration, so that
using the tangential and radial accelerations of the centre of even when the deceleration begins the cable has to be vertical.
mass on its circular orbit ( 2l ε and 2l ω 2 respectively) and use it In that case, how are the acceleration time and the oscillation
as an equation to find φ. period related?
77. Based on Idea no. 64 we find that ω = 6v/l. Using energy 85. If the shockwave is at the point where the intersection area
and momentum conservation we eliminate the puck’s velocity of its wavefront and the considered body is S, then what is the
after the collision and express the mass ratio. force acting on the body? Let us assume that the body stays
78. The forces along the normal to the surface are elastic forces, (almost) at the same place as the shockwave passes it. Then the
so the energy in vertical direction is conserved during the colli- momentum imparted during the time dt can be found using the
sion: after the collision the corresponding velocity component cross-sectional area S and the distance dx = cs · dt covered by
is the same as before. To find the other two unknowns, the the wavefront. Note that S · dx is the volume element. Finally
horizontal and angular velocities, we can obtain one equation we sum over all imparted momenta.
using Idea no. 64. The second equation arises from (a) the 86. The rod will act like a spring (since the rod is thin and
condition that the velocity of the ball’s surface is zero at the made out of steel, while steel is elastic). After the left sphere
contact point (no sliding; (b) the equation arising from 60). has collided with the stationary sphere, the latter will acquire
79. Using the idea 51 we investigate the sliding and rolling velocity v0 and the former will stay at rest. Then the dumb-
regimes. In the latter case the quickest way to find the answer bell, as a system of spheres and springs, will begin oscillating
is to use Idea no. 65. around its centre of mass. What is the velocity of the centre
of mass? Convince yourself that after half a period the single
80. The velocity can be found from the conservation laws for
sphere is already far enough that the left sphere is not going
energy and momentum (note that the hoop is moving transla-
to collide with it again. The oscillations of the dumbbell will
tionally). To find the acceleration it is convenient to use the
decay little by little — so some energy will be lost there.
non-inertial reference frame of the hoop, where the centripetal
acceleration of the block is easily found. The condition for the
radial balance of the block gives the normal force between the 7 ANSWERS
block and the hoop (don’t forget the force of inertia!); the hori- 1. arcsin √
Rµ
.
zontal balance condition for the hoop provides an equation for (R+l) µ2 +1

finding the acceleration. 2. arcsin Mm+m

√ µ2 .
µ +1
81. Let us assume the block’s velocity to be approximately 3. mg/2.
constant. For a certain time tl the base slides to the left with √
4. a) µmg/ 1 + µ2 ; b) mg sin(arctan µ − α).
respect to the block and the momentum imparted by the fric-
|g sin α−a cos α|
tional force at that time is also directed to the left. During the 5. µ ≥ g cos α+a sin α , if g + a tan α > 0.
remaining time tr the base slides to the right with respective √
6. a) ω 2 R ≥√g 1 + µ−2 ;
momentum directed to the right as well. The equilibrium con- b) ω 2 R ≥ g 1 + µ−2 , if µ < cot α and
dition is that the two momenta have equal magnitudes; hence ω 2 R ≥ g(cos α + µ−1 sin α) if µ > cot α
we ding the equilibrium value of tl /tr . From the graph we find √
the velocity for which that ratio has the needed value. 7. v = 3 gR

82. As the water flows against the paddles it obtain the same 8. v/2.
vertical velocity u as the paddles themselves. You should also 9. tan 2α = h/a
keep in mind that in the paddle’s frame of reference, the speed √
10. µ1 ≥ l2 − h2 /h
of water is decreased v−u v times as compared with speed in
the lab frame; the mass flow rate is decreased by the same 11. 3mg
m
factor (in lab frame, some water is “lost”, i.e. remains between 12. 2 arctan[(1 + M ) cot α]
— page 36 —
 √ 7. ANSWERS
√
13. 2HLµ + µ2 H 2 − µH ≈
2HLµ − µH ≈ 7.2 m. 55. 5 m/s.
√
14. a) ω < g/l ; b) ω < (2 − 2)g/l
2 2
56. (a) tan ≤ 2µ ; (a) impossible.
√
15. 2 (1 −
1
3−1/2 )ρv ≈ 211 kg/m3 57. g(1 − xl ) sin−1 α; π2 l sin α/g
π 3
16. 3 ρR 58. µ < cot α.
√
17. v/ µ2 cot2 α − 1 59. µ1 < tan α2 and µ2 < tan α2 .
18. 4 3
3 πGr ∆ρ/g(r + h) ≈ 0.95 cm 60. R > h/2
√
19. −ω 61. 3 3m/πρ
20. µmgv/ωR 62. ω 2 R2 /2g
(1−sin α)2
21. cos φ tan α < tan 30◦ 63. M/m = .
√ sin α
22. L − πR/2 cos α; 2π L/g 64. 2mM
M +m g tan α
1 1 7
23. 12 mg, 3 mg, 12 mg 65. g/9.
24. mg/(2M + m) m+M 2
66. g m+M sin2 α sin α.
25. m < M cos 2α. 67. 2/3R
√
26. mg sin α/[M + 2m(1 − cos α)] = 68. m[g − v 2 (2l − x)/ 2l2 ]
mg sin α/[M + 4m sin2 α2 ]. √
69. M/m = 4, u = gl/8.
1 sin α1 −m2 sin α2 )(m1 cos α1 +m2 cos α2 )
27. g (m(m
1 +m2 +M )(m1 +m2 )−(m1 cos α1 +m2 cos α2 )
2.
70. The first one arrives first
√
28. mg(5 2 − 4)/6); Simultaneously. 71. A straight line; if ω ̸= 0
√
29. cos α ≥ 13 (2 + v 2 /gR) 72. 2gl(1 + m/M )
√
30. 2 Mm+m 2gR
F
73. 3m F
, if mµg F
< 6; 4m + 12 µg, if
F F
31. mM g/(m + 4M ) 6 < mµg < 10; 3µg, if mµg > 10
32. Fx = 2Raρ, Fy = (m + ρL)g − ρ(L − πR − 2l)a, where 74. m2 v 2 /2(M 2 − m2 )µg
a = ρg(L − πR − 2l)/(m + ρL). √
75. v = (l − H 2 )g, a = 2 (1 −
g H
2l ), T = mg
4 (3 − H
2l + l
6H ).
33. The one that had not been pushed.
76. arccos 23 ≈ 48◦ 12′
34. If F ≤ 2µmg 2m+Mm+M
: a1 = a2 = 1 F
2 M +m ; otherwise
a1 = M − µg M , a2 = µg 2m+M
F m m
. 77. M/m = 4.
√
78. (a) ω = 5v0 /7R, vx = 5v0 /7, vy = 2gh;
35. On a half-circle. √
(b) vy = 2gh, vx = v0 − 2µvy ,
36. (a) v/5; (b) v/4. √
ω = 5 2ghµ/R.
37. n(n − 1)/2
√ 79. 5
7 g sin α, if µ > 2
7 tan α, otherwise g sin α − µg cos α
38. 2µgL(1 + m √
M) 2gr 1+cos φ
80. m+M m sin2 φ+M m cos φ;
39. 3.5; was coming from below right.
m2 cos φ(1+cos φ)
√ √ gm sin 2φ 1
m sin2 φ+M [ 2 + (m sin2 φ+M )(m+M ) ]
40. A: 2gh ; gh.
√ √ 81. 0.6 m/s
41. 2Rµ gl sin α, gl sin α.
√ 82. 4
27 µv
2
42. u − µ 2gh.
83. v/ cos α
43. mg(h + µa).
84. n−2 Lg/4π 2 l, n = 1, 2, . . .
44. arctan 52 ≈ 21◦ 48′ .
85. (a),(b) (p1 − p0 )V /mcs .
45. (a) (3v − ωl)/4; (b) (v − ωl)/2.
86. 12 v0 ; no, a fraction goes into the longitudinal oscillations of
46. At a distance 2l/3 from the holding hand, where l is the
the rod and then (as the oscillations die) into heat
length of the bat.
2 F a
47. 3M R
48. (vx0 , vy0 − 75 u)
√
49. L/v0 + π m/2k
1 2
50. 2 π (n + 12 )2 R tan α
51. 1.03 s
52. 2.0 g
α
53. v1 = v2 = v; cot2 2
√
54. gH.
— page 37 —
 1. INTRODUCTION
Thermodynamics where ∆U is the change of the internal (thermal) energy, ∆Q
Jaan Kalda Version: 11th November 2017 — the amount of heat given to the system, and
∆W = p∆V
1 INTRODUCTION is the mechanical work done by the system while expanding; p
This booklet follows the structure of the study guide “Mech- stands for the pressure at the system’s external boundary, and
anics”. Some thermodynamical concepts are inherently rather ∆V denotes the change of the system’s volume.
complex; meanwhile, understanding all the details of that com- This formula (the thermodynamical energy conservation law)
plexity are not necessary for a successful solving of Olympiad is referred to as the 1st law of thermodynamics (1LTD). All
problems. Because of that, the material is divided into two the three terms of this law are signed quantities: if the system
categories: (a) basic topics, learning of which is sufficient for gives away some heat then ∆Q < 0; if the volume contracts
solving a majority of the Olympiad problems, and (b) an ad- then the work done by the system is negative, ∆W < 0 (in-
vanced material, which is primarily aimed for deepening the stead of W , one can also use the work done by external forces
understanding of the origins of thermodynamics and is given ∆We in which case ∆We = −∆W ).
in a small shrift or moved into Appendices. The expression for the work here assumes that the relative
Thermodynamics differs from the other branches of physics changes of the system properties are small, so that the pressure
in that its laws are statistical and hence, are not absolutely p can be considered constant; thus, it would be more correct to
strict: they are valid only with a certain accuracy and certain substitute the ∆-symbols (which are typically used to designate
probability. For macroscopic systems, these statistical laws are a finite change) with differentials, or to write
∫
based on averaging over many atoms and molecules; owing to
W = p dV.
that, the relative statistical fluctuations of average quantities
are very small, so that the accuracy of the statistical laws is This formula implies that the work can
very high. be found as the surface area under p(V )-
This claim can be also stated mathematically. Mathematical statist- graph and can be derived as follows. Let
ics teaches us that if the value of a certain random quantity is obtained us consider a small surface area ∆Ai of
independently N times, the standard deviation of the respective average the interface separating the system under
√
value (averaged over these N realizations) is N times smaller than the study from the rest of the space, and let
standard deviation of a single measurement. For instance, the root-mean- us denote with ∆xi that component of its displacement which
square (rms) of the speeds of N molecules (v̄) has a standard deviation δv̄ is perpendicular to the surface element ∆Ai ; we can consider so
√
which is N times smaller than that of a single molecule, δv. The latter short time interval that the displacement ∆xi remains as small
has the same order of magnitude as the rms speed itself, δv ∼ v̄. Hence, as needed. The pressure force acting on the surface element
the relative magnitude of fluctuations of the rms speed of N molecules can ∆Fi = p∆Ai , and its work ∆Wi = ∆Fi ∆xi = p∆Vi , where
√
be estimated as δv̄/v̄ ∼ 1/ N . To get an idea about typical numerical ∆Vi = ∆Ai ∆x1 is the increase of the system’s volume under
values, let us consider a gas inside a volume of one cubic centimetres. Un- the surface element ∆Ai . In order to find the total work, we
der normal conditions, one mole occupies a volume of 22.4 litres, and one need to sum over the entire interface,
∑ ∑
cubic centimetre contains N ≈ 6 × 1023 22400 1
≈ 3 · 1019 molecules, hence ∆W = i p∆Vi = p i ∆Vi = p∆V .
the relative statistical fluctuations of the thermodynamical quantities are For an infinitesimal volume increment dV , it is rewritten as
√
of the order of 1/ N ≈ 5 · 10−9 , i.e. really negligible. dW = p dV and can be integrated over the whole process to
∫
Within the framework of school physics and for a majority yield W = p dV .
of the Olympiad problems, the statistical nature of the thermo- In order to be able to solve problems, we also need the
dynamics remains unnoticed, because the formulae of thermo- concept of temperature, which is intuitively clear for everyone,
dynamics and molecular kinetics can be applied exactly in the but a correct definition of which is quite complicated. Classical
same way as all the other physical formulae. (non-digital) thermometers measure the temperature using the
thermal expansion of liquids: warmer liquid takes more space
than a colder one. However, this cannot be used as a good
2 Heat and temperature
definition of temperature: different liquids expand at a differ-
def. 1: Heat is the energy of the thermal motion of a system. ent rate, and most importantly, this approach would be applic-
able only to moderate temperatures (at high temperatures, all
Note that the term “thermal motion” stands for a random mo-
the matter is in the plasma state, and at temperatures close
tion of the microscopic particles making up the bodies (most
to the absolute zero, the few substances which remain in the
typically atoms and molecules). These particles are so small
liquid state undergo phase transitions).
and move so fast that typically, the thermal motion cannot be
We know that if the temperature outside is low, we feel
directly seen.
cold: our body gives away some heat, and according to the
Since the overall energy of any closed system is conserved,
definition 1, the kinetic energy of our body molecules becomes
the following very useful conclusion can be drawn directly from
lower. Inversely, if the temperature is high, it’s hot: our body
this definition (formulated as a “fact”).
cannot give away as much heat as is produced via physiological
fact 1: the full energy (thermal+mechanical etc.) of a closed processes, and the kinetic energy of our body molecules starts
system is conserved: rising. Therefore we can start with a qualitative definition of
∆U = ∆Q − ∆W, the temperature:
— page 1 —
 2. HEAT AND TEMPERATURE
def. 2: Temperature is a quantity which characterizes the dir- bodies: one small (or moderately-sized) and one huge, which is
ection of heat exchange between two bodies, both of which have assumed to be in a thermal contact with the small one, i.e. the
reached a thermal equilibrium: when brought into contact, the two bodies can exchange energy. For instance, the small body
heat flows from a body of a higher temperature to the body of can be a single atom of a monomolecular gas, and the heat
a lower temperature (equal temperatures correspond to a zero reservoir can be the rest of the gas. The large one is referred
heat flux). to as the heat bath and is needed to ensure that when the small
body obtains or gives away due to random fluctuations some
This definition is usable (not self-contradictory) only if we have
of its thermal energy, the temperature would remain constant.
the following property. Consider an arbitrary set of three bod-
Therefore, it needs to be much large (have much larger heat
ies A, B, and C; suppose that when A and B are brought into
capacitance) than the body which we analyse.
contact, heat flows from A to B, and when B and C are brought
into contact, heat flows from B to C. Then we can be sure that Let the energy levels of the body states be denoted by Ei
when A and C will be brought into contact, heat will flow from (e.g. for a single, almost free atom, Ei = p2i /2m, where pi is
A to C. Experimental observations confirm this property (so the momentum of the atom in its i-th state).
we can use def. 2), and actually even a more generic property fact 3: It can be shown using mathematical statistics (see
which is known as the second law of thermodynamics (2LTD): Appendix 1) that the probabilities of the body for being in the
i-th or j-th state satisfy the Boltzmann’s law:
fact 2: Whichever tricks are used (heat engines, Maxwell’s pi
= e−β(Ei −Ej ) .
demons1 , etc), if no external work is done, heat can flow only pj
from a body of a higher temperature to a body of a lower tem- The constant β depends on (and describes) the state of the heat
perature (i.e. the direction of heat flow cannot be reversed as reservoir, i.e. on the temperature; it is easy to see that smaller
compared to what is observed in the case of a direct contact of values of β correspond to higher mean energies of the body.
two bodies). Therefore, the temperature of the reservoir can be defined as
Ei − Ej
Within the theory of classical thermodynamics, 2LTD is a T̃ = 1/kβ = ,
k ln(pj /pi )
postulate following from experimental data; within the stat-
where k is a constant which could be freely chosen, but is taken
istical thermodynamics, the methods of mathematical statist-
equal to be kB ≈ 1.38 × 10−23 J/K to ensure agreement with
ics are used to show that for processes involving macroscopic
def. 3.
bodies (i.e. made of a large number of molecules), outcomes
violating 2LTD will have negligible probabilities (the proof is To sum up, the Boltzmann’s law
mathematically complicated and will not be provided here). p ∝ e−E/kB T
Note that the thermal energies of two bodies of equal tem- serves as the definition of temperature which coincides with the
perature are not necessarily equal; however, for a fixed body, Kelvin’s temperature scale; here ∝ denotes proportionality and
thermal energy is a monotonously increasing function of its E is the system’s energy. The only difference with the Kelvin’s
temperature2 . scale is that with the statistical temperature definition, one
Now we have a qualitative definition of temperature, but we can also have negative temperatures — assuming that the body
still lack a quantitative one. For many problems it is enough and heat reservoir have finite number of quantum-mechanical
to know and understand the simplified and classical one: states, and the net energy of the reservoir is so high that higher
def. 3: Temperature is a measure of a body at thermal energy states are more probable than the lower ones.
(quasi)equilibrium: if two bodies of different temperatures are def. 4: Each body is characterized by its thermal capacit-
brought into contact, heat flows from the higher temperature ance C which shows how much heat needs to be given to raise
body to the lower temperature one. For a given body, tem- the temperature by one degree: C = dQ/dT .
perature is a monotonously increasing function of its internal
heat energy. Kelvin’s temperature scale is defined so that zero Note that this definition, when written for finite increments
heat energy corresponds to T = 0 K, ice melts by atmospheric as C = ∆Q/∆T , assumes that there are no phase transitions
pressure at 273.15 K, and water boils at 373.15 K. (such as melting or boiling) within the considered temperature
range, because phase transitions incur additional heat absorp-
This definition is clearly insufficient for designing thermo- tion or heat release.
meters suitable for very low temperatures. Apart from this
definition, there is also the Kelvin’s one which will be discussed def. 5: In order to bring a substance from one phase to an-
in section 4. The best and most generic definition is based on other (typically from solid phase to liquid or from liquid phase
the statistical approach to thermodynamics (in which case the to gaseous phase), certain amount of heat needs to be given
2LTD is derived from the laws of statistics); let us consider this to the substance, which is proportional to the mass m of the
in more details. Quantum mechanics tells us that a body (a substance: Q = λm. The factor λ is referred to as the latent
set of particles which are bound by forces into a finite region heat (e.g. latent heat of evaporation or latent heat of melting).
of space) is characterized by a countable set of its stationary Typically, for moderate temperature increments near room
states — the states where the total energy has a well-defined temperature, the heat capacitance can be assumed to be con-
value. For statistical thermodynamics we need to consider two stant; in that case, we can write ∆Q = C∆T . In the case
1 See http://en.wikipedia.org/wiki/Maxwell%27s_demon
2 This can be shown using 2LTD and a construction involving Carnot cycles (ideal heat engines).

— page 2 —
 2. HEAT AND TEMPERATURE
of larger temperature increments, this is no longer valid and What was the mass of the ice if the heating power of the kettle
∫
integration is needed: ∆Q = CdT . P = 1 kW. The latent heat of melting for ice L = 335 kJ/kg,
the heat capacitance of water c = 4.2 kJ/kg · K. The room
fact 4: At low temperatures, the heat capacitance of crystal
temperature T1 = 20 ◦ C.
materials is proportional to the third power of temperature,
C ∝ T 3. T (°C)
80
If the body which is being heated expands, the supplied heat
is partially converted into mechanical work. Therefore, the
60
heat capacitance depends on the conditions under which the
body is held; most often, heat capacitance at constant volume
CV and the heat capacitance at constant pressure Cp are used. 40

fact 5: If a body is heated at a constant volume, no expan-

sion work is done, so that according to the 1LTD, CV can be 20
0 50 100 150 200 t (s)
used to find the change of the internal energy of the thermal
motion of molecules: dU = CV dT and
∫ T In the case of this problem, we can also formulate a fact,
U= CV (T ′ )dT ′ . and a rather simple idea.
0

This integration makes use of the fact that at T = 0, there is fact 6: The heat exchange rate (i.e. the heat flux, measured in
no thermal motion of molecules, and hence, internal energy is Watts) between a body and the environment is a function of the
zero. In particular, for crystal materials at low temperatures temperatures of the body and of the environment, T1 and T2 ,
when C = AT 3 integration yields U = 14 T 4 . respectively; for a small tempera difference ∆T ≡ T1 − T2 ,
In the case of liquids and crystals, the volume change is the heat flux is proportional to ∆T ; this is referred to
typically small so that the expansion work is negligibly small as the Fourier’s law. For larger temperature differences, the
and Cp ≈ CV ; however, this not valid for gases, in which case dependence becomes nonlinear, because (a) heat conductivity
cp = CV + R (this will be discussed later in more details). of the materials may depend on the temperature, (b) the heat
flux due to heat radiation is a non-linear function of T1 and
pr 1. [IPhO-1996] A thermally insulated piece of metal is T2 (however, it can be still linearized for small values of ∆T );
heated under atmospheric pressure by an electric current so (c) large temperature differences may cause convection of air
that it receives electric energy at a constant power P . This and fluids which will enhance heat flux in a nonlinear way. Pay
leads to an increase of the absolute temperature T of the metal attention that the heat flux goes to zero for T1 = T2 !
with time t as follows:
T (t) = T0 [1 + a(t − t0 )]1/4 idea 2: Heat exchange rate of a body with the environment is
Here a, t0 and T0 are constants. Determine the heat capacity a function of the temperatures; hence, for processes during
Cp (T ) of the metal (temperature dependent in the temperature which the temperature change remains small, the heat
range of the experiment). exchange rate can be assumed to be constant. In partic-
ular, if a body has reached a thermal equilibrium while being
The solution of this problem is rather straightforward: it is heated with a power P , it has reached a temperature T by
0 0
enough to apply the definition 4 and make some mathematical which the heat loss power equals exactly to P : P (T ) = P ;
0 HL 0 0
manipulations to eliminate time t. We can still try to formulate hence, if the heater is suddenly switched off, the initial cooling
an appropriate “idea” as follows. power of the body (due to the heat loss to the environment) is
idea 1: The body temperature T and the net heating power P0 .
P are related via equality P ≡ dQ dT
dt = C dt . Therefore, the Returning to the problem 2, the graph allows us to determ-
tangent to the T (t) curve is proportional to the net power and ine, how long time-delay τ in the heating process has been
inversely proportional to the heat capacitance; many problems caused by the piece of ice: this is the time needed to melt and
can be solved by using this observation. heat the meltwater up to the current water temperature. The
In the case of the problem 1, this idea is to be used algebraic- value of the ice-heating-time τ would allow us to determine the
ally. In those cases when a T (t)-dependence is provided in a mass of ice from the 1LTD if the net heating power (P minus
graph, it tells us that as long as the heat capacitance of a body the energy loss to the environment) were known. The aver-
remains approximately constant, we can find how many times age net power received by the kettle during that period when
the net heating power is changed by determining the slopes of the ice is being melted and the meltwater being heated can be
the tangent at two different points of the T (t)-graph. Let this determined from the graph using the ideas 2 and 1.
be illustrated with the following problem. The particular case mentioned at the end of the idea 2 is
illustrated with the following problem.
pr 2. [EstPhO-2004] Water is being heated in an electric pr 3. [EstPhO-2001] The filament of a halogen lamp has
kettle. At a certain moment of time, a piece of ice at tem- length l = 5.0 cm and is made of tungsten. At the working tem-
perature T0 = 0 ◦ C was put into the kettle. Figure depicts perature of the lamp T0 = 3200 ◦ C, the density of the tungsten
the dependence of the water temperature as a function of time. ρ = 18 200 kg/m3 , the specific heat c = 235 J/(K · kg), and the
— page 3 —
 2. HEAT AND TEMPERATURE
resistivity ρel = 9.95 × 10−7 Ω · m. When a voltage of rectan- the figure. The cross-sectional area of the wire S = 1 mm2 and
gular waveform as shown in graph is applied to the leads of the length l = 4 cm. Find the heat flux through the wire if one
lamp, the temperature of the filament will reach the nominal end of the wire is kept at the temperature T1 = 100 ◦ C and the
value T0 (strictly speaking, this is the average temperature; the other — at T2 = 0 ◦ C.
interior regions of the filament are slightly hotter). However, (K·m/W)

due to voltage oscillations, there are small oscillations of the

filament’s temperature; find the amplitude of these oscillations
0.2
∆T .

0.1

The usage of the linear dependence mentioned in the fact 6

x(cm)
is illustrated by the following problem.
0 1 2 3 4
pr 4. A small house is being heated by an electric heater Second half of this problem makes use of a rather generic
of power P = 1 kW which maintains the interior temperature
method.
t1 = 19 ◦ C; the outside temperature t0 = 0 ◦ C. A man enters
the house, upon which the room temperature starts rising and idea 4: Calculation of many physical quantities can be re-
achieves a new value t2 = 20 ◦ C. Find the “heating power” of duced (sometimes not in an obvious way) to the calculation of
the man. surface areas under a graph (i.e. to an integral).
More specifically, if a system is described by a parameter x
Here the task of figuring out that the Fourier’s law can be
(which can be time, coordinate, velocity, etc.) and a quantity
used (because the temperature difference is much smaller than ∑
A can be expressed as A = i Fi ∆x, where ∆x is a small in-
the temperatures in Kelvins) is left to those who solve the
terval of the parameter x, the sum is taken over all the small
problem. Be aware that for some loosely formulated problems,
intervals, and Fi is a function of x (but not a function of A)
the Fourier’s law is assumed to be used even when its applic- ∫
then at the limit ∆x → 0 we can write A = F (x)dx, i.e. A is
ability is actually questionable (for instance for temperatures
the surface area under the graph F (x).
T1 = 273 K and T2 = 373 K).
In order to illustrate this method, let us consider the fol-
idea 3: The problems on heat exchange rate are analogous to lowing mechanical problem. For a boat in water, the friction
the problems on electrical circuits. There are following match- force is given as a function of speed F (v), depicted in a graph
ing pairs: temperatures correspond to voltages, heat energy - (it has a crossover from a linear function at small speeds to a
to a charge, heat fluxes — to currents, a thermal resistance parabola for large speeds). You are asked to find, how far will
(proportionality coefficient between the heat flux and ∆T , c.f. travel a boat of a given mass m and initial speed v asymptotic-
idea 6) — to an electrical resistance, heat capacitance — to ally (i.e. upon waiting for a very long time). Let us divide the
∑
electrical capacitance. The both Kirchoff’s law remain valid for displacement into small pieces, s = ∆s, where ∆s = v∆t.
heat exchange processes: while the counterpart of the voltage If the function v(t) were known, the last formula would have
law is quite useless due to triviality (along a closed circuit, the been completed our task. However, the force is given to us as
sum of temperature increments is zero), the counterpart of the a function of v, hence we need to substitute ∆t with ∆v. Force
current law is useful and can be referred to as the continuity is directly related to the acceleration, a(v) = dv dt = F (v)/m,
of thermal fluxes: if the bodies of a system have reached sta- which gives us a hint that we should try substituting ∆t via
tionary temperatures, for each body, the sum of heat fluxes ∆v while introducing the acceleration:
entering it equals to zero. ∆t ∆v ∆v
∆t = ∆v · = = .
∆v ∆v/∆t a
pr 5. [EstFin-2003] Thermal resistivity is a characteristic of This result ∑
serves us perfectly well:
∑v ∑ vm ∫
vm
a material and is defined as the coefficient of proportionality s= v∆t = ∆v = ∆v → dv,
dT
between the temperature gradient dx and the heat flux dens- a F (v) F (v)
ity (here we assume that the heat flux is parallel to the x-axis). i.e. the displacement
vm
equals to the surface area under a graph
Note that the unit of the heat flux density is W/m2 , hence the which depicts F (v) as a function of v.
unit of the thermal resistivity is K · m/W. The next problem makes also use of the continuity of heat
(i) A microprocessor of power dissipation P = 90 W is cooled flux; however, it also requires the knowledge of the Stefan-
using a flowing water; the thermal contact between the pro- Boltzmann law for heat radiation.
cessor and the water is established via a copper plate of fact 7: For “grey’ bodies (which absorb a constant frac-
thickness d = 5 mm and surface area s = 100 mm2 . De- tion of the incident electromagnetic radiation, independently
termine the temperature difference between the microprocessor of its wavelength), the heat radiation flux density (i.e. radi-
and the flowing water. Thermal resistivity of copper ρ = ated energy per unit area) w = εσT 4 , where T is the tem-
2.6 mm · K/W. perature of the body, ε ∈ [0, 1] — the absorption factor, and
(ii) A wire is made of different alloys so that its thermal res- σ = 5.67 × 10−8 W · m−2 · K−4 — the Stefan-Boltzmann con-
istivity depends on the coordinate along the wire as shown in stant.
— page 4 —
 3. GASES
The factor ε shows, which fraction of the incident light is ab- Th Tl
sorbed; for a perfectly black surface ε = 1; for a perfectly white
surface ε = 0. This formula is derived using the assumption that the ra-
diating body is in a thermal equilibrium with the electromagnetic waves in
the adjacent vacuum; a knowledge of quantum mechanics and application
of the Boltzmann’s distribution are required, as well as certain mathem-
This problem leads us to a system of N linear equations.
atical skills.
In general, solving such a system of equations is not an easy
task; this particular system, however, can be easily solved. It
pr 6. [IPhO-1992] A satellite represents a homogeneous should not be surprising, because long mathematical calcula-
sphere of diameter D = 1 m; you may assume that all the tions are not compatible with the format of physics Olympi-
parts of the satellite have the same temperature. The satel- ads; we can formulate an appropriate recommendation — as a
lite orbits around Earth (but is not in its shade). The Sun “non-physical” idea.
can be considered to be an absolutely black body with the
idea 5: As a rule, the problems of physics Olympiads do not
surface temperature T⊙ = 6000 K, the radius of the Sun
require long mathematical calculations. If you obtain a long
R⊙ = 6.96 × 108 m, the orbital radius of the Earth around
or difficult system of equations then most likely one of the fol-
the Sun L = 1.5 × 1011 m. Find the temperature of the satel-
lowing is valid: (a) you have followed a non-optimal (i.e. more
lite assuming that it is coated with a perfectly grey paint (the
complicated than necessary) solving method; (b) the method is
absorption factor is independent of the wavelength of the elec-
good, and the system of equations is only seemingly complex,
tromagnetic radiation). You may neglect the heat radiation of
e.g. almost all the unknowns can be eliminated simultaneously
the Earth.
using the symmetries of the equations.
For this problem, the following fact is to be used (those who
are “afraid” of the surface integral can skip to the paragraph pr 8. [EstOpen-2014] Consider a black cube which is made
next to the fact). from a perfectly heat-conducting material. A parallel beam of
light with intensity I (W/m2 ) falls onto this cube. The equi-
fact 8: In a stationary state, the Gauss theorem is valid for librium temperature T of the cube depends on its orientation;
heat fluxes (this is completely analogous to the Gauss theorem find the minimal and maximal values of T (T
H min and Tmax ,
for the electric and magnetic fields): w ⃗ · dS
⃗ = P , where the
respectively).
integral is taken over a closed surface, P is the heat power
released inside the surface, the heat flux density vector w⃗ is This problem is physically quite straightforward, but full
parallel to the propagation direction of the heat energy, the (rigourous) solution requires mathetical skills in the field of
vector dS⃗ is parallel to the surface normal, and is equal by vector calculus.
modulus to the surface area of a small (infinitesimal) surface idea 6: In the case of a homogeneous vector field F⃗ (x, y, z) ≡
∫
element. F⃗0 , its flux Φ = S F⃗ · dS⃗ trhough a (possibly curved) surface
S can be found as the dot product of a certain effective flat
This is essentially a mathematical formulation of the continuity ∫
surface element S ⃗eff = dS, ⃗ Φ=S ⃗eff · F⃗0 .
law for the heat flux. In the case of a spherically symmetric S

geometry (with an isotropic spherical heat source), the surface

integral simplifies to the product of the heat flux density with 3 Gases
the surface area of a sphere:
In statistical physics (thermodynamics), it is relatively easy
2
4πR w = P, (i.e. difficult, but in many cases still possible) to make calcula-
tions if everything is very regular, for instance in the case of
where w = w(R) is the heat flux density at the distance R from crystalline solids. It is even easier to make calculations when
the centre of the heat source. everything is chaotic, for instance in the case of gases and plas-
The next problem completes the topic of heat flux continu- mas, because then the physical quantities can be averaged. The
ity. most difficult to analyse are media where order and disorder
exist together - such as fluids, granular media, solids near phase
transition, etc. Because of that, high school physics deals only
pr 7. [IPhO-1996] Perfectly black hot surface is kept at a with the chaotic case — with gases.
constant temperature Th . There is another perfectly black cold A good model which describes the reality quite well is the
surface which is parallel to the first one, and is kept at a con- model of ideal gases.
stant temperature Tl ; there is a vacuum in the space between def. 6: An ideal gas is made up of molecules of negligible
the plates. In order to reduce the heat flux between the plates, size (as compared with the average intermolecular distances)
a screen is used, which is made of N parallel thermally insu- which behave as elastic balls (monoatomic gas) or spring-ball-
lated perfectly black plates (in the figure, N = 2). Such a systems (polyatomic gas), and move randomly colliding with
screen is placed between the hot and cold plates, parallel to each other and with the walls of the container. It is assumed
them. By which factor x will the heat flux be reduced once a that there are no other forces between the molecules than the
thermal equilibrium is reached? The edge effects caused by the contact forces (for instance, electrostatic interactions need to
finite size of the plates can be neglected. be negligible as electrostatic forces act over a distance).
— page 5 —

Let n be the number density of the molecules (number
per unit volume), and f (vx ) — their distribution function
over vx (the x-component of the velocity), defined so that
∆n = nf (vx )∆vx gives us the number density of such mo-
lecules for which the x-component of the velocity lies within the
interval [vx ; vx + ∆vx ] (here, ∆vx denotes a small velocity in-
crement). According to the Boltzmann’s distribution, f (vx ) =
Ae−mvx /2kT (which in this form is referred to as the Maxwell’s
2
distribution) 3 , where the factor A is such that the area under
∫∞
the f (vx )-graph is unity, i.e. A = 1/ −∞ e−mvx /2kT dvx .
2
3. GASES
found as F = 2muN/t = nmu2 A. Now, let us recall that in
reality, molecules can have different speeds, so that we need to
average the result, F = nm⟨u2 ⟩A, where angular braces denote
averaging. Finally, the pressure p = F/A = is found as
p = nm⟨u2 ⟩.
So, the pressure is expressed in terms of the mean square of
the x-component of the velocity. This expression can be also re-
⟨ ⟩
written in terms of the average net speed v 2 as p = 13 nm⟨v 2 ⟩.
Indeed, v 2 = vx2 + vy2 + vz2 (where vx ≡ u) and therefore,

⟨v 2 ⟩ = 3⟨u2 ⟩ (we have taken into account that due to stat-

The integral in the last expression can be taken using two techniques.
istical isotropy, in average all the propagation directions of the
The first (and the simpler) technique is to change integration variable by
√ molecules are equivalent, hence ⟨vx2 ⟩ = ⟨vy2 ⟩ = ⟨vz2 ⟩).
substituting x = vx /vT , where vT = 2kT /m, this allows us to get rid of
Using the Maxwell’s distribution one can express the aver-
the parameter v0 in the integrand. Indeed, since v0 constant (independent ∫
age values in terms of integrals: ⟨vx2 ⟩ = vx2 f (vx )dvx . This in-
of the integration variable), dvx = vT dx and
∫∞ ∫∞ tegral can be taken (see appendix 2), resulting in ⟨vx2 ⟩ = kT /m,
2 2
e−mvx /2kT dvx = vT e−x dx. i.e. the average kinetic energy related to the motion along the
−∞ −∞
A second trick (which we don’t discuss here) needs to applied to show that x-axis
∫ ∞ −x2 √ m 1
−∞
e dx = π, resulting in ⟨ vx2 ⟩ = kT.
2 √ 2 2
f (vx ) = e−mvx /2kT /vT π. Using this expression we obtain
p = nm⟨u2 ⟩ = nkT.
m
fact 9: Velocity distribution of molecules in a gas is given by Once we substitute n = N/V = (NA is the Avogadro µ NA /V
number, µ is the molar mass of the gas, m is the total mass of
the Maxwell’s distribution,
N √ the gas, and V is its volume), we obtain pV = m m
µ Na kT = µ RT,
−vx
2 2
dN = √ e /vT
dvx vT = 2kT /m, where R = kNA is called the gas constant. This law is known
vT π
where N is the total number of molecules, and vx is the x- as the ideal gas law; sometimes it is convenient to express it in
component of the molecules. In the isotropic 3-dimensional terms of the gas density ρ as pµ = ρRT. Let us sum up our
case, it can be also written for the speed, results.
4N 2 −v2 /vT2 fact 10: Ideal gas state is described by the law
dN = √ v e dv.
vT3π
pV = νRT,
In order to derive the last equation, we assumed uncorrelated motion in
which can be also expressed as
x, y, and z-directions so that the individual distributions can be multi-
( )3 2 2 3 2 p = nkT or pµ = ρRT.
plied: dN = N
√ e−(vx +vy +vz )/vT dvx dvy dvz , and integrated over
vT π
∫
a spherical layer in vx − vy − vz -space, dvx dvy dvz = 4πv 2 dv. method 1: Various estimation problems related to gases can

Let us find an expression for the pressure. We do this in be done in the same way as we derived the ideal gas law, i.e.

two parts: first we make calculations assuming that all the mo-
lecules approach the wall with the same speed u (let the x-axis
be perpendicular to the wall; then u is the x-component of the
molecule’s velocity), and then we generalize the result to the
realistic case of different speeds. Pressure is defined as force per
unit area, p = F/A; the force is found as the momentum given
to the wall per unit time, F = 2N mu/t, where N is the num-
ber of molecules colliding with the wall during a time period
t. The coefficient two reflects the fact that the average speed
with which the molecules depart from the wall after collision
equals to the approaching speed4 , hence each molecule receives
momentum 2mu from the wall. During the time period t, only
those molecules will reach the wall which are within the near-
wall-layer of thickness ut and move towards the wall. Thus, the
total number of collisions is found as the number of molecules
within the layer of volume V = AvT t, reduced by a factor of
two (only half of the molecules move in the direction of the
wall): N = 12 nV ≈ Aut. Thus, the force exerted to the wall is
3 Strictly speaking, the Boltzmann’s distribution is valid for the average occupancy of a quantum-mechanical state; here we make use of the fact
using the molecular kinetic approach.
Exact calculations require often integration (averaging using
the Maxwell’s distribution). However, in the case of Olympiad
problems, it is typically enough to make only estimates, and
there is no need to use the Maxwell’s distribution. This is due
to two reasons: (a) taking complex integrals does not test the
knowledge of physics, (b) in many cases even seemingly exact
calculations are exact only for the so-called vacuum approx-
imation when the mean free path (the distance travelled by
a molecule between two subsequent collisions with other mo-
lecules is much larger than the characteristic dimensions of the
system), and only approximate otherwise.
pr 9. In vacuum and weightlessness, at the bottom of a
cylindrical vessel (a cup), there is a layer of solid substance
of molar mass µ. This substance sublimes slowly (evaporates
from the solid phase into gaseous phase) and pushes thereby
the vessel to the opposite direction. Estimate the terminal

that for an elastic ball in a box, the quantum-mechanical levels are evenly distributed over the values of vx .
4 Strictly speaking, for a single collision, the departure speed is random and depends on the momentary speed of that wall’s molecule with which

it collides. However, when averaged over many collisions and assuming that the gas and the wall have equal temperatures, the average departure
velocity is such as if there were an elastic collision with a flat wall

— page 6 —

speed of the vessel. The mass of the vessel M , and the initial
mass of the substance m ≪ M ; the temperature of the ves-
sel is T ; the process can be assumed to be isothermal (cooling
due to evaporation and heat radiation remains negligible). The
cross-sectional area of the vessel is A.
pr 10. Natural uranium consists of mainly two isotopes,
U238 and U235 , whereas the relative concentration of the latter
is 0,7%. Uranium is “enriched” (i.e. the concentration of U235
is increased) by implementing a multi-stage process, where at
each stage, evaporated chemical compound UF6 is led through
a porous wall. The porous wall can be considered to be a thin
film having microscopic holes in it (the size of the holes is smal-
ler than the mean free path of the molecules, but larger than
the dimension of the molecules How many stages is needed to
increase the U235 content to 1,4%? The molar mass of fluoride
(F) is 19 g/mol.
pr 11. [PhysCup-2012] Determine or estimate the net heat
flux density P between two parallel plates at distance L from
each other, which are at temperatures T1 and T2 , respectively.
The space between the plates is filled with a monoatomic gas
of molar density n and of molar mass M . You may use the
following approximations:
(i) the gas density is so low that the mean free path λ ≫ L;
(ii) T1 ≫ T2 ;
(iii) when gas molecules bounce from the plates, they obtain
the temperature of the respective plates (for instance, this will
happen if they are absorbed/bound for a short time by the
molecules of the plate, and then released back into the space
between the plates);
(iv) you may neglect the black body radiation.
Note that “Estimate” means that the numeric prefactor of your
expression does not need to be accurate.
The internal energy U of a monomolecular gas (i.e.
the net kinetic energy of the molecules) can be expressed
in terms of the temperature using the above obtained result
mv̄x2 = kT . Indeed, a single atom essentially cannot rotate due
to its small dimensions5 . Therefore, U is found as the product
of the energy of a single molecule with the number of molecules
N,
U =Nm 2 2 2
2 (v̄x + v̄y + v̄z ).
Due to statistical isotropy v̄x = v̄y = v̄z2 = kT ; substitut-
2 2
ing N = νNA (where ν stands for the number of moles) and
NA k = R we obtain
U = 32 νRT .
In the case of diatomic and polyatomic molecules, the mo-
lecules obey also rotational energy Ur = 12 N (Ix ω̄x2 +Iy ω̄y2 +Iz ω̄z2 ,
where Ix , Iy and Iz denote momenta of inertia with respect to
different axes, and ωx , ωy andωz — the respective angular ve-
locities. Using the Boltzmann’s distribution for the rotational
motion around the x-axis, p ∝ e−Ix ωx /2kT , we obtain an expres-
2
sion which is very similar to the one which we had for trans-
lational motion (just vx is substituted by ωx and m — with
Ix ). Because of that, the result is also the same, Ur = 32 νRT .
However, one should keep in mind that if we have a diatomic
5 The reasons are quantum-mechanical, and to a certain will be discussed later.
— page 7 —
3. GASES
molecule (or polyatomic linear molecule, such as CO2 ), it can-
not rotate around the axis passing through its atoms (let it
be the z-axis). Therefore, rotation will take place only around
the x- and y-axis; the rotational energy is reduced respectively:
Ur = νRT .
Let us notice that each of the so-called degrees of freedom
( translational motion along the x-axis, rotation around the
x-axis, translational motion along the y-axis, etc) contribute
to the overall internal energy always the same amount 21 νRT
( 12 kT if we speak about the average energy of a single molecule).
Because of that, the internal energy of a gas is conveniently ex-
pressed in terms of the number of degrees of freedom i as
U = 2i νRT.
For a monoatomic gas i = 3, for a diatomic gas (and a linear
polyatomic gas) i = 5; for all other cases i = 6. In the case of a
mixture of gases (such as air), the effective number of degrees
of freedom may turn out to be fractional.
Using the last expression we can easily derive an expression
for the internal energy of a gas by constant volume. If the
volume remains constant, there is no mechanical work done
∫
(A = pdV = 0), hence all the heat given to the gas goes
to the increase of the internal energy, CV = dU/dT = 2i νR.
Particularly simple expression is obtained for the molar heat
capacitance, cV = CV /ν:
cV = 2i R.
If we are dealing with an isobaric process (at a constant pres-
sure) then the internal energy change can be, of course, calcu-
lated using the previous result, but according to the 1LTD,
part of the heat goes to the work performed by the gas,
∆A = p∆V = νR∆T . Hence, the heat given to the gas
∆Q = ∆U + ∆A = i+2 2 νR∆T , i.e. the molar heat capacitance
for isobaric processes
cp = i+2
2 R.
Rule “ 21 kT per each degree of freedom” is valid, however,
only for moderate temperatures. When temperature is large
enough then molecules will start oscillating similarly to balls
which are connected with springs, and oscillation energy needs
to be included, as well. In that case one can say that the os-
cillatory degrees of freedom are excited. For relative light gas
molecules, such as the components of air, the critical temperat-
ure at which the oscillations are excited is well above the room
temperature. Meanwhile, heavy gas molecules (e.g. Br2 ) oscil-
late already at the room temperatures. It should be emphas-
ized that the 12 kT -rule is not valid for the oscillatory degrees of
freedom. It appears that each excited oscillatory degree
of freedom contributes full kT to the heat energy of a mo-
lecule. Thus, each oscillatory degree of freedom increases the
effective number of degrees of freedom by 2, e.g. for i = 5+2 = 7
for Br2 at room temperature.
The reason why oscillatory degrees of freedom contribute twice as
much as the translational and rotational degrees of freedom lies in the
structure of quantum mechanical energy levels. For translational and ro-
tational energy, the energy level is a quadratic function of the level order
number n, En = h2 n2 /8L2 m; here, L stands for the length of the box
where our particle is placed. Meanwhile, for oscillatory motion, energy
is a linear function of the order number, En = ~ω(n + 1
2
), where ω is
 3. GASES
the natural frequency. Therefore, the average oscillation energy can be masses we need already j = 3N coordinates. In order to de-
calculated using the Boltzmann’s distribution as scribe the spatial placement of a solid body, 6 coordinates are
∑∞ −~ωn/kT
∑∞ −~ωn/kT needed: apart from the coordinates of a certain point of the
Ē = 0
~ωne / 0 e .
body (e.g. the centre of mass), orientation needs to be described
The sums can be taken to find Ē = kT .
by additional 3 coordinates (two angles describing the direction
On the other hand, at very low temperatures, even the ro-
of an axis of the body, and one angle describing rotation around
tational degrees of freedom of polyatomic molecules may not
that axis). In the case of a linear molecule, last coordinate, the
be excited; this means that their effective number of degrees
rotation angle, is not needed, therefore j = 5.
of freedom is 3. In order to determine if one or another degree
of freedom is excited, it is necessary to compare the energy If a system consists of N point masses, the motion of which
level difference ∆E between the ground level and the first ex- is limited by certain restrictions (e.g. the distance between two
cited level with the average energy of a degree of freedom kT : point masses is fixed with a rod, or the angle between the lines
the respective degree of freedom starts to become excited at connecting a certain point A with two neighbouring points has
T ∼ E/k. a fixed value), the number of degrees of freedom can be found
Indeed, let us consider two energy levels E0 and E1 . According to
using the formula j = 3N − r, where r on is the number of re-
the Boltzmann’s distribution, p1 /p0 = e−(E1 −E0 )/kT . If E1 − E0 ≫ kT
strictions. In the case of molecules, the bonds are the chemical
then the probability of being at the excited state is exponentially small
bonds which also serve as the springs and can give rise to oscil-
and hence, the related average energy increase is exponentially small, as
lations; in that case, r is the number of oscillatory degrees of
well.
freedom. Knowing that for a linear molecule, the total number
Now it becomes also clear why the molecules of heavy gases start os-
of degrees of freedom j = 5, and for planar or three-dimensional
cillating at lower temperatures than light gases. The natural frequencies
molecules j = 6, we can use this formula to find the number
√ of oscillatory degrees of freedom as r = 3N − j. For instance,
ω ≈ k/m of heavy molecules are lower than that of light molecules
(the stiffness of a chemical bond as a “spring” has always the same or-
CO 2 molecule is linear: N = 3 and j = 5, hence r = 4. Note

der of magnitude, the main difference comes from the effective mass m).
that in the theory of coupled oscillators, it is shown that the
This means that the energy level difference ∆E = ~ω is correspondingly
number of oscillatory degrees of freedom equals to the number
smaller, leading to a lower excitation temperature.
of natural modes and natural frequencies.
Thus, the effective number of degrees of freedom depends on Let us consider one problem on the topic of degrees of free-
temperature; furthermore, for such a temperature range where dom.
the effective number is changing from one value to another, the pr 12. [EstAcadPhO-2006] Toy “Supermag” makes it pos-
system behaviour is not well described by an integer value of sible to construct, among others, polyhedrons — e.g. tetrahed-
i. Because of that, instead of i, often molar heat capacities cV rons, cubes, and many irregular polyhedrons, where the edges
and cp are used. If the behaviour of a gas is well described of the polyhedron are made of magnetic bars, which are connec-
by the model of ideal gas, it is enough to give just the value ted at the vertices with the help of steel balls. The steel balls
of cV as cp can then be found from the equality cp = cV + R. fix the endpoint of a steel bar to itself firmly, but the angle
Conversely, if for a certain gas cp − cV ̸= R then it does not be- between magnetic bars meeting at a steel ball can be changed
have as an ideal gas, and relationship pV = νRT does not hold with a little effort. It appears that a tetrahedron is a rigid
well. Meanwhile, it is well possible that for an ideal gas, 2cV /R construction, but a cube can be easily deformed. Prove the
is not an integer: fractional values just show that the current following theorem: a convex polyhedron is rigid then and only
temperature is close to the excitation temperature of a certain then if all the faces of the polyhedron are triangles.
degree of freedom. Let us make an intermediate summary.
In order to solve this problem, the formula j = 3N − s is
fact 11: For an ideal gas, used, but one more formula is needed — the Euler’s formula
cV = 2i R; cp = cV + R; ∆U = νcV ∆T, for graphs (including polyhedrons) relating the number of ver-
where i is the number of excited degrees of freedom (oscillat- tices N , faces f , and edges e via equality N + f = e + 2.Euler’s
ory degrees of freedom are to be added with double weight). theorem is provided here without proof (which is actually not difficult and
A certain degree of freedom is excited when kT is larger than is based on mathematical induction). In order to solve problem 12,
the energy level difference of the quantum mechanical states it is necessary to prove that if all the faces are triangular, the
corresponding to the given degree of freedom. As a rule of number of degrees of remaining freedom j = 3N − s equals to
thumb, near room temperature for a monoatomic gas i = 3, that of a rigid body. If the number of degrees of freedom were
for diatomic gas i = 5, and for polyatomic gases i = 6. larger, it would be needed to use more parameters than in the
case of a rigid body to describe its state, i.e. it cannot be rigid.
Thus far we used the concept of the degrees of freedom
without proper definition; let us discuss this notion in more As we have learned, atoms and ions connected in molecules
details. with chemical bounds can oscillate as spring-block systems.
This applies not only to molecules, but also to crystal materials.
def. 7: The number of degrees of freedom shows, how many Similarly to heavy gases, in the case of many crystal materials,
parameters (generalized coordinates) is needed to describe com- all the oscillatory degrees of freedom are excited. In the case
pletely the state of a system. of N -ion(atom)-crystal substance, there are 3N degrees of free-
For instance, in order to describe a point mass in three- dom; almost all these degrees of freedom (except for 6 degrees
dimensional space, j = 3 coordinates are needed; for N point of freedom of a solid body) are oscillatory. Since N is very
— page 8 —

large, we can forget about that 6, so that the internal energy
can be expressed as U = 3νRT . Therefore, molar heat capa-
citance cV = 3R. In the case of light ions, it may still happen
that the highest-frequency natural modes have ω > kT /~ and
therefore are not excited yet. In that case, the internal energy
is smaller than 3νRT , and the effective number of degrees of
freedom per single ion (atom) is an increasing function of tem-
perature, approaching asymptotically the limit value i=6 (as
all degrees of freedom are oscillatory, the number of spatial co-
ordinates is multiplied by two). Let us also recall that at low
temperatures it starts as a cubic function of T . Let us derive the
cV ∝ T 3 law for low temperatures via rough estimates (for a diamond,
this is valid even for room temperatures). Each excited oscillatory degree
of freedom carries energy kT ; thus we need to estimate the number of
excited oscillations, i.e. number of natural modes with ω < kT /~. Nat-
ural oscillation modes of a crystal can be considered as standing sound
waves. Sound wave frequency is expressed as ω = cs κ, where cs is sound
speed, and κ = 2π/λ — a wave vector. Thus, all standing waves with
wave vector λkT /~cs and wavelength λ> hcs /kT will be excited. Let the
crystal have a cubic shape of side length L. At the free boundary of the
crystal, there can be no strain, hence for the displacement vectors of the
molecules, there must be antinodes at the boundary. Therefore, L must be
a multiple of the half-wavelength: Lκx = mx π, where κx is the projection
of the wave vector to the x-axis, and mx is the number of nodes along
x. For the smallest excited wavelength, λs = hcs /kT and mx = 2L/λs .
Note that all those standing waves which have smaller number of nodes
are also excited. Similarly, along y- and z-axes, for exited oscillations the
number of nodes varies between 1 and 2L/λs . The number of different
standing waves is found as the number of different triplets (mx , my , mz ),
which is equal to (2L/λs )3 . Then, the internal energy is found as
U ≈ M kT ≈ kT (2LkT /hcs )3 ∝ T 4 .
Therefore, the internal energy is proportional to T 4 , and the heat capacit-
ance CV = dU
dT
∝ T 3.
Now, let us turn to the problems which are not related
to the molecular kinematic approach, and instead, require the
knowledge of the ideal gas law and its internal energy. Many
Olympiad problems are based merely on those two equations;
in the case of the next problem, the only “trick” is that un-
like for ordinary isohoric, isobaric, etc. processes, it is not the
amount of gas which is conserved, but the pressure, hence the
number of moles will change according to νT = Const.
pr 13. [EstPhO-2003] An aerostat (an air balloon made of
unstretchable skin) has volume V0 , is filled with hydrogen, and
hovers stably at a certain height where the external pressure is
p, and the air temperature is Tair . Due to sunlight, the aero-
stat is heated up to a temperature T1 . As a result, part of the
air exits the balloon via a valve. The valve is made so that
the if the interior pressure exceeds the exterior one, the excess
gas is released; the valve never lets gas enter the balloon from
outside. Now, the sunlight is shaded by a cloud: the balloon’s
temperature drops, and volume becomes smaller. How much
ballast (which mass m) needs to be thrown out to keep the
aerostat at the same height (where the pressure is still equal to
p)? Both air and hydrogen can be treated as ideal gases, the
molar masses are µH2 and µair , respectively.
The next problem deals with a more complicated energy
transfer process; the following idea will be useful.
— page 9 —
3. GASES
idea 7: If two reservoirs are connected via a narrow pipe,
narrow valve, etc, then a quasi-equilibrium is possible: while
the temperatures of the two reservoirs to the other sides of the
pipe are different, the pressures are equal. Thus, there is a
mechanical equilibrium, but no thermal equilibrium (it takes
much longer time to reach the latter).
pr 14. Consider a balloon which has thick rigid walls and
from which all the air has been pumped out. Now, the valve of
the balloon is slightly opened, and the balloon is slowly filled
with the air from outside. Find the temperature of the air
inside the balloon once the air flow has stopped (since a mech-
anical equilibrium has been reached). The room temperature
is T , the balloon walls have low heat conductance and heat ca-
pacitance so that heat flux through the walls can be neglected.
The first step towards the solution of this problem is under-
standing to what is spent the work done by external air p∆V :
how is it converted to heat and how is the external volume
change ∆V related to the volume of the cavity V . Let us also
notice that presumably the size of the valve opening is larger
than the mean free path length, so that a molecular kinetic
approach is not needed, and instead, macroscopic approach is
to be applied: a narrow (but still macroscopic) stream of air
enters the cavity thorough the valve.
Adiabatic processes
Typical reply to the question of what are adiabatic processes
tends to be that these are so fast processes with a gas that there
is essentially no heat exchange with the gas (during the given
time period, the heat transferred conductively and radiatively
to the gas remains much smaller than the internal energy of
the gas).
However, this is only half of the truth, and actually the less
important half. In order to understand this, let us consider the
following simple problem. Thermally isolated vessel is divided
into two halves via a wall. Let one of the halves contain a gas
at a pressure p, and let the other one be empty (i.e. contain
vacuum at a negligibly small pressure). The wall separating
the two halves is removed instantaneously; what is the pres-
sure of the gas inside the vessel upon achieving mechanical and
thermal equilibrium?
Since the wall is removed instantaneously, the gas cannot
perform any mechanical work (the wall does not move in the
direction of the force). There is no thermal exchange to the
environment and so according to the 1LTD the internal en-
ergy remains constant, hence the temperature does not change.
Meanwhile, the volume is increased twice and due to the ideal
gas law, this means that the pressure is decreased twice. So, we
dealing with an isothermal process, and not with an adiabatic
one!
As a matter of fact, the main requirement for having an
adiabatic process is that the process needs to be sufficiently
slow — the characteristic time of the process needs to be much
longer than the period of natural oscillation of the system. In
the case of gases, the natural oscillation modes are standing
waves; the period of the slowest mode is cs /2L, where L is the
vessel length and the factor two accounts for the fact that the
 3. GASES
wavelength of the longest standing wave has wavelength equal the universe grows two times? What is the adiabatic exponent
to 2L; cs denotes the speed of sound. Thus, v/L ≪ cs /2L, for a photon gas?
hence v ≪ cs , where v is the speed of the container walls (or
For the adiabatic processes with gases, it is also re-
of a piston).
quired that the heat exchange rate of the given gas is negligibly
def. 8: Adiabatic process is a process by which the system small (as compared with the internal energy change): ∆Q = 0.
parameters change so slowly that the characteristic time of If we recall the adiabatic motion of a particle between two
changing is much longer than the period of the slowest mode walls, we see that the no-heat-exchange requirement is very
of natural oscillations; also, there should be no dissipative pro- much consistent with that adiabatic process: the particle did
cesses (where mechanical energy is converted to heat), e.g. fric- not receive any external energy apart from the energy received
tion. In the case of gases, this means that the speed of the from the moving wall (heat flux corresponds to an external en-
container walls needs to be much smaller than the speed of ergy source for the kinetic energy of the molecules), and there
sound, and also there should be no external heat supply. were no energy losses due to fiction or other processes. So, the
adiabatic process with a gas can be arbitrarily slow, in prin-
Adiabatic processes in such a wider-than-thermodynamical
ciple, but in practice, for too slow processes the heat exchange
sense play important role also in quantum mechanics, electro-
can no longer be ignored.
dynamics, etc.
Before we derive the adiabatic law for gases, let us note
In the course of analytical mechanics it is shown that for
that at Olympiads, this derivation is typically not needed, it
adiabatic processes on periodically oscillating systems, the so-
suffices just to no the law.
called adiabatic invariant is conserved; this is the surface area of
To begin with, we apply the 1LTD to an adiabatic process
the loop drawn by the oscillator in the phase plane (the curve in
H on one mole of gas, dU = pdV , where dU = cV dT , i.e.
x−px -plane px dx, where x is the coordinate and px — the cor-
responding momentum), c.f. the booklet of mechanics. While a cV dT = −pdV .
general proof is complicated, a problem of the mechanics book- Note that even though the volume is not constant, the internal
let dealt with a particle bouncing back and forth between two energy change is expressed in terms of cV . This is a standard
slowly moving plates; it was shown that Lv = Const, v being method: with constant volume, there is no mechanical work
the particle’s speed and L — the distance between the walls. done and hence, all the provided heat cV dT goes to the change
Now, let us consider one-dimensional gas: there are many of the internal energy dU .
particles bouncing along the x-axis between the plates. From idea 9: Internal energy change can be calculated as
the molecular kinetic theory, the pressure of such a gas p = dU = cV dT.
nmv̄ 2 . The number density of the particles is, of course, in-
We want to derive relationship between the pressure and
versely proportional to the “volume ”, which is in the one-
volume, so we need to eliminate here the temperature; to that
dimensional case just the distance between the plates L, i.e.
end, we use the ideal gas law, RT = pV , hence
n ∝ 1/L. Due to the adiabatic invariant, the speed of each
of the molecules is inversely proportional to L and hence, the RdT = pdV + V dp.
average speed follows the same law, v̄ ∝ 1/L. Thus, p ∝ 1/L3 , Here we applied the differentiation rule d(pV ) = pdV + V dp.
or alternatively, pL3 = Const. This is our adiabatic law for the Now we can easily eliminate dT from these two equations to
one-dimensional gas: pV γ = Const, with V = L and γ = 3. obtain
What is even more important, is that in quantum mech- pdV (cV + R) + cV V dp = 0.
anics, the surface area in phase space is related to the quasi- Let us notice that cV + R = cp , and introduce the adiabatic
classical quantization rule, for n-th quantum-mechanical en- index defined as
H
ergy level, px dx = n~. Now, if the potential changes slowly γ = cp /cV .
in time, the adiabatic invariant is conserved, hence the system This allows us to rewrite the last equation as
dp
will remain at the same quantum-mechanical energy level (with γ dV
V + p = 0.
∫
the same order number) as it was at the beginning (exceptions If we integrate this equality, the first term yields γ dV =
γ
∫ dp V
are possible only in the case of a “separatrix-crossing”, e.g. if γ ln V = ln V , and the second one — p = ln p, so that
the “potential well” becomes so narrow that the given bound ln V γ + ln p = ln pV γ = Const, hence
state disappears, and the particle formerly bound to the well pV γ = Const.
becomes free). Now, let us return to the one-dimensional gas discussed be-
idea 8: During adiabatic processes, particles remain at their fore; it has one degree of freedom, thus cV = 12 R and cp = 32 R
initial quantum-mechanical levels. When combining this obser- so that γ = 3, which is in a complete agreement with our res-
vation with the Boltzmann’s law, it is often possible to find the ult for the adiabatic invariant. Let us make an intermediate
new temperature. summary.
fact 12: Adiabatic law, which is valid simultaneously with
pr 15. Expansion of our Universe can be considered as an the ideal gas law pV /T = Const, states that
adiabatic processes for the photons of the microwave back- pV γ = Const; kus; γ = cp /cV .
ground radiation: the wavelength of the photons grows pro- Combining this with the gas law we have also equalities
portionally to the linear size of the Universe. How many times
T V γ−1 = Const; and; T γ ∝ pγ−1 .
will change the temperature of the radiation, if the linear size of
— page 10 —
 3. GASES
idea 10: For adiabatic processes, the work done can be cal- significant, it would be necessary to use here the Boltzmann’s
culated as the change of internal energy (since there is no heat distribution ρ ∝ e−µgh/RT . While for µgh ≪ RT , this is ex-
exchange): pression can be simplified using linear term of the Taylor’s
∫
i
∆A = ν 2 R∆T = pdV . expansion ex ≈ 1 + x, the same approximate result can be
obtained (arguably) more easily using the following idea.
For real-life problems when the applicable assumptions are
not explicitly stated, it is important to understand, when the idea 13: If the height differences are small and there is no
heat exchange rate can be neglected (then the process is adia- need to apply the Boltzmann’s distribution, the air pressure
batic), and when the heat exchange is so fast that the process change can be found as the air column pressure ∆p = ρgh,
is essentially isothermal. Here it is useful to compare charac- where ρ can be considered in the first approximation to be
teristic time scales of thermal relaxation time τth (the time independent of height.
during which the temperature differences decrease by a factor Using the result of this problem, it becomes also clear, why
of e), and the characteristic time of the process itself (e.g. the mountain peaks are snowy — for instance 3 km would corres-
time during which the system is brought from its initial state to pond to the temperature difference of 30 degrees. In reality,
the final state). It is useful to know that the thermal relaxation the temperature difference remains somewhat lower since the
time depends on the system size (linear dimension) L, being air which ascends along the mountain slopes is being heated due
proportional to L2 ; in the case of air under normal conditions, to heat conductance from the mountain surface (see above).
for L ≈ 1 cm, τth ≈ 1 s.
During calm night weather, the adiabatic atmosphere may
Next we proceed to the discussion of atmospheric processes.
not be observed. This is because the Earth’s surface gives away
heat via radiation and cools down; lower air masses are also
idea 11: When air masses are in vertical motion, typically cooled via heat conduction. Let us note that heat radiation
during day time, but also during cyclonal activity and strong of the air itself is negligible: being transparent, it doesn’t ab-
winds, the vertical motion of air incurs adiabatic expansion. sorb radiation; due to the 2LTD, absorption and radiation are
This results in the so-called adiabatic atmosphere, where the symmetric phenomena and therefore clean air cannot radiate.
temperature falls with the height h, Lower cold air layers have higher density and therefore, a very
T ∝p
γ γ−1
, p = p0 − ρgh; stable air stratification is formed: there is almost no vertical
here the expression for p is valid for moderate height differences air motion. Such a phenomenon is called “inversion”. In the
when the variation of the air density ρ can be neglected. case of inversion, all the waste gases remain near the surface,
giving rise to high air pollution.
Indeed, air masses are very large and hence, even though the
Bernoulli equation, 12 ρv 2 + ρgh + p = Const (where ρ is
air may rise slowly, the thermal relaxation time is even slower,
the density, v — speed, p — pressure, g — free fall accelera-
so that the process is adiabatic. During daytime, the vertical
tion, and h — height) is a well-known equality expressing the
motion of air is caused by sunlight heating the ground, which
fact that in the case of a stationary flow, the energy flux en-
in its term, heats the air via thermal conductivity (note that
tering any volume must be equal to the energy flux exiting
as air is transparent, it doesn’t absorb light, and hence, direct
that volume: otherwise, the total amount of energy inside that
heating by sunlight is negligible). The heated air has smaller
volume would start growing, contradicting the assumption of
density than the colder air at higher altitudes, and starts rising.
stationarity. Bernoulli equation, however, is valid only in the
At the ground level, void cannot be created, so there must be
case of incompressible flows. As we have seen above, atmo-
also descending air masses. So, air moves up and down, result-
spheric air flows are typically adiabatic, in which case the en-
ing in a fairly good vertical mixing, and adiabatic atmosphere.
ergy balance equation needs to take into account the change of
What does it mean in terms of temperature drop at high alti-
internal energy of the gas, and the work done by expansion.
tudes, we’ll learn through the following problem.
idea 14: For problems set on stationary gas flow, two conser-
pr 16. What is the temperature at the top of a hill of height vation laws are to be used. First, along streamlines
H = 200 m if at the bottom of the valley t0 = 20 ◦ C? You can
v2
assume that when air masses move along the slope of the hill, + gh + cp T = const,
2
air expansion and/or contraction can be considered to be adia-
where cp = Cp /M is the specific heat of the gas per unit mass
batic. The air pressure at the valley level p0 = 105 Pa, air
3 by constant pressure (Cp denotes the molar specific heat, and
density ρ = 1,29 kg/m and adiabatic index γ = 1, 4.
M — the molar mass). This equality reflects energy conser-
This problem can be solved very fast if we make use of the fact vation. Alternatively, if the flow speed is much smaller than
that here, the relative pressure change is very small. the sonic speed and density variations along streamlines remain
idea 12: In the case of equalities involving products, e.g. small, ρ ≪ 1, the original Bernoulli law
∆ρ

aα bβ = Const, if the relative changes of variables are small 1 2

ρv + ρgh + p = const
(∆a ≪ a, ∆b ≪ b) then it is convenient to use approximate 2
calculation based on logarithmic differentiation, can be also used.
da db ∆a ∆b Second,
0 = d ln(aα bβ ) = α +β ⇒ α = −β .
a b a b ρvA = const,
If the hill height would be very large, so that air density dif- where A is the cross-sectional surface area of a fictitious tube
ference between the top and bottom of the mountain would be formed by streamlines; this reflects mass conservation as ρv
— page 11 —
 3. GASES
is equal to the mass flux density (mass passing a unit cross- pr 18. [IPhO-2012] In the fig. below, a cross-section of an
sectional area per unit time). aircraft wing is depicted together with streamlines of the air
It should be emphasized that in the case of gas flows when ρ flow around the wing, as seen in the wing’s reference frame. As-
2
is not constant, v2 +gh+cp T = const holds exactly; meanwhile, sume that (a) the air flow is purely two-dimensional (i.e. that
the velocity vectors of air lie in the figure plane); (b) the stream-
ρ ≪ 1.
the Bernoulli law is valid only at the limit ∆ρ
The fact that the Bernoulli law can be used at the limit line pattern is independent of the aircraft speed; (c) there is
no wind; (d) the dynamic pressure is much smaller than the
ρ ≪ 1 is actually non-trivial. Indeed, the Bernoulli law is
∆ρ

derived from energy balance: work done by pressure is conver- atmospheric pressure p0 = 1.0 × 10 Pa. You can use a ruler
5

ted into kinetic and/or potential energy. However, when gas to take measurements from the fig. on the answer sheet.
is compressible, there is also a conversion into internal energy
due to compression.

Consider a displacement of a certain amount of gas inside a

fictitious tube made from streamlines as shown in figure: let
us assume that before displacement, this volume of gas occu-
pied the region shaded into blue and grey. As the result of
displacement, now it occupies the region shaded into grey and
red. There is no change in the state of the grey region, but a) If the aircraft’s ground speed is v0 = 100 m/s, what is the
gas from blue region has been moved into the red region. The speed of the air vP at the point P (marked in fig.) with respect
to the ground?
change in kinetic energy is m2 (vr − vb ), where vb and vr stand
2 2

for the gas speeds at the blue and red positions, respectively.
Energy balance is written as
m water droplets is created behind the wing. The droplets emerge
W = (vr2 − vb2 ) + ∆U,
2 at a certain point Q. Mark the point Q in fig. on the answer
where ∆U is the change of the internal energy, and the work
done by external pressure
W = pV − (p + ∆p)(V + ∆V ).
Now, consider a longer displacement of gas — such that the
gas of the blue region has covered the whole distance from blue
to red, and occupies now the volume shaded into red. Dur-
ing that displacement, the process on the parcel was adiabatic,
∫ unperturbed air Ta = 293 K and psb = 2.46 kPa at Tb = 294 K.
so ∆U = − pdV ; assuming ∆p p ≪ 1 (which is equivalent
to requiring ∆ρ ρ ≪ 1), the integral can be approximated as
∆U = −p∆V . Returning to the energy balance equation, we
can write
m 2
(v − vb2 ) = V ∆p + ∆V ∆p,
2 r
which gives us the Bernoulli law if we neglect the quadratically
small term ∆V ∆p and assume that the densities at the blue
and red positions are approximately equal, m V ≈ V +∆V . This
m
is valid if the relative change in the flow speed is much larger
than the relative change in density, which is true if the flow
speed is much smaller than the sonic speed.
When we derived here the Bernoulli law, we used a trick
which can be also formulated as an idea.
idea 15: If a column of liquid or gas, a rod or a rope (and so
on) moves parallel to itself so that the new state and old state
mostly overlap, and there are only small regions where the state
is has been changed (e.g. the red and blue regions in the figure
above), the change of the total energy (momentum, angular
momentum, etc.) can be calculated by considering only those
regions where the state has been changed.
pr 17. Prove that for a stationary gas flow, along stream-
v2
lines 2 + gh + cp T = const.
— page 12 —
b) In the case of high relative humidity, as the ground speed
of the aircraft increases over a critical value vcrit , a stream of
sheet. Explain qualitatively (using formulae and as few text as
possible) how you determined its position.
c) Estimate the critical speed vcrit using the following data: re-
lative humidity of the air is r = 90%, specific heat of air at
constant pressure cp = 1.00 × 103 J/kg · K, pressure of satur-
ated water vapour: psa = 2.31 kPa at the temperature of the
Idea 14 is also useful when deriving sound speed cs in a gas.
Then, however, we need one more idea.
idea 16: In order to find the propagation speed of a pulse in
homogeneous medium, for instance gas or stretched rope, it is
convenient to use a moving reference frame where the pulse is
at rest.
Let us consider a sound wave which propagates in the direc-
tion of x-axis; then, the air density ρ = ρ(x − cs t). Following
the idea 16, we consider a frame which moves with speed cs ,
with coordinate axis x′ = x − cs t. In that frame, the density
perturbation remains constant in time, ρ = ρ(x′ ). This means
that we can use idea 14, so that we obtain two equations:
(ρ0 + ∆ρ)(v + cs ) = ρ0 cs ,
1 1
(v + cs )2 + cp T = c2s + cp T0 ,
2 2
where v ≪ cs is the speed of gas in laboratory frame. In sound
wave, the density perturbations are typically small so that we
can assume ∆ρ ≪ ρ. Upon opening the braces and neglecting
quadratically small terms (e.g. v∆ρ and v 2 ), we obtain
ρ0 v + cs ∆ρ = 0, cs v + cp ∆T = 0.
If the wavelength is long enough (much longer than the mean
free path length of the gas molecules), the gas flow is adiabatic,
so that we can relate ∆T to ∆ρ via adiabatic law T ρ1−γ =

T0 + (1 − γ) ρ0 = 0, hence
∆ρ
const. Using the idea 12, we obtain ∆T
∆T = (γ − 1) Tρ00 ∆ρ. With this, our equations can be rewritten
as
T0
ρ0 v + cs ∆ρ = 0, ρ0 cs v + cp (γ − 1) ∆ρ = 0,
ρ0
C −C
where cp (γ − 1) = M −1 Cp pCV V = M −1 Cp CRV = γ M R
. This
is a set of two linear equations with two unknowns, v and ∆ρ
which can have non-zero solutions only if the determinant is
zero (the equations are linearly dependent), i.e.
RT0
c2s = γ .
M difference by ∆U0 . Now we can draw two conclusions.
When calculating cs , we have ignored one important fact:
in the case of straight streamlines (what we have in the case of
planar wavefronts/one-dimensional wave propagation), apart
from energy and mass conservation, we have also the conserva-
tion of linear momentum of the gas.
pr 19. Show that in the case of gas flow along straight
streamlines, due to Newton’s II law for the gas parcels,
ρv 2 + p = const.
It should be emphasized that this last law is valid only in the
case of straight streamlines: otherwise, the pressure force by
which the walls of the fictitious tube (made from streamlines)
act on the gas parcels would also contribute to the acceleration
along streamlines.
When we calculated cs , we used three equations: energy
conservation law, 21 v 2 + cp T = const, mass conservation ρv =
const, and adiabatic law ρ1−γ T = const. We could have substi-
tuted the adiabatic law with the momentum conservation law
ρv 2 + p = const, and still we would have obtained the same
result. This means that if the amplitude of the density per-
turbations is small, the four equation are linearly dependent,
and the adiabatic law can be derived from the remaining three:
Thus, as long as the amplitude of the wave is small, the sound
propagation is adiabatic. On the other hand, for large amp-
litudes this is no longer the case: the three conservation laws
(know as the Rankine-Hugoniot conditions) may (and will) be
in contradiction with the adiabatic law. Therefore, in the case
of large-amplitude shock waves, the process on gas is not adia-
batic, and part of mechanical energy is converted into heat.
pr 20. In the case of a shock wave, a high-pressure region
propagates in gas with a speed which is somewhat larger than
the sound speed. Consider a shock wave in which air density
grows exactly two times; how many times is such a shock wave
faster than the sound speed?
Interface between liquid and vapours
The liquid (and crystal) particles (atoms, molecules, ions) are
bound to each other via chemical bonds — forces which de-
pend on the distance between particles in a complicated way.
By origin, these forces have electrostatic nature: the inter-
play of quantum-mechanical laws and electrostatic interactions
between orbital electrons and atomic nuclei results in the force
between particles being a complex function of the positions of
interacting particles. At small distances between the atomic
nuclei, the force becomes repulsive (positive charges of nuclei
repel), at moderate distances the force is attractive (this keeps
— page 13 —
3. GASES
the particles together in the liquid or crystal phase), and at
large distances the force decays faster than the electrostatic
law of r−2 . Attraction at moderate distances means that the
potential energy (with respect to an infinite distance) for a typ-
ical near-neighbour-pair is negative; let the average potential
energy per particle be −U0 . For particles inside the liquid (or
crystal) phase, the number of near neighbours is larger than for
the particles at the surface. Respectively, the average poten-
tial energy of the inter-particle forces per particle at the surface
is also smaller by modulus than U0 ; let us denote the energy
First, increasing the surface area of a liquid will increase
the number of particles at the surface, and hence, the overall
energy of molecular forces. Naturally, the number of particles
adjacent to the surface is proportional to the surface area A,
and so is the overall energy W associated with the particles be-
ing at the surface instead of being inside the bulk of the liquid:
W = σA, where the coefficient σ characterizes the molecular
forces of the liquid, and is called the surface tension. This
phenomenon will be discussed in the next section.
Second, in order to pull out a particle from the liquid phase,
it needs to obtain energy U0 (this energy is analogous to the
work function in the Einstein’s equation of photoelectric ef-
fect). Thus far we have learned that in thermodynamical sys-
tems, particles can be in higher- and lower-energy-states, which
means that the liquid molecules can obtain also, due to random
interactions with other molecules, energy exceeding U0 , and as
a result, “jump out” of the liquid phase. This process (mo-
lecules jumping out of the liquid) gives rise to evaporation: the
number of molecules in liquid phase will decrease. However,
there is also the reverse process, condensation at the liquid sur-
face. Indeed, if the gaseous phase above the liquid surface con-
tains the molecules of the liquid, the vapour molecules (which
are in a random motion) can hit the liquid surface and get
“stuck” to it.
For a molecule, the probability of having an additional en-
ergy exceeding U0 is proportional to e−U0 /kT ; the evaporation
rate is proportional to that probability and hence, increases
very rapidly with the temperature. Condensation rate, on
the other hand, is less sensitive to the temperature
√ (is propor-
tional to the speed of the molecules, i.e. T ), and depends
mostly on the concentration of the vapour molecules in the
gaseous phase. Obviously, if the evaporation rate exceeds the
condensation rate, the amount of liquid is decreasing, and vice
versa. If we take a certain amount of liquid and seal it tightly
into a bottle, the two processes reach an equilibrium: the con-
centration of the vapour molecules in the gaseous phase will
reach such a value that the evaporation rate equals to the con-
densation rate.
def. 9: Vapour with such a concentration which leads to the
condensation rate being equal to the evaporation rate at the
given temperature is called the saturation vapour.
Most often, the amount of vapours is measured as the partial
pressure caused by the given type of molecules in the gaseous
phase. This method is valid owing to the Dalton’s law.
fact 13: Dalton’s law states that pressure is an additive
quantity: the total pressure exerted by a gas equals to the
 3. GASES
sum of partial pressures: p = PA + PB + . . ., where pA is the the case of extremely clean air, when there are only very few
pressure due to the molecules of type A, pA — that of molecules condensation centres, for a certain period of time r > 100% is
of type B, etc. possible, such a gas is called an oversaturated vapour.
This law follows directly from the principles of molecular kin- Now, let us discuss in more details evaporation at r < 100%.
etic theory: pressure is the momentum of molecules transferred In the case of water vapours in air, air is most often relatively
to the container walls per unit time and unit surface area, and dry, with r ranging typically from 40% to 80%. This is the
according to the laws of mechanics, momentum is an additive case because free water surface forms only a small fraction
quantity (the momentum of an entire system is the sum of the of all the surfaces: solar heating reduces r by increasing the
momenta of all its parts). saturation pressure, but the vapour pressure remains almost
constant since there no vapour sources (in the form of liquid
Returning to the vapours, the saturation vapour is typically
water) are available. Even relatively close to water surfaces, r
characterised via its pressure, the saturation vapour pressure.
will remain moderate: there is not enough time for establishing
a thermal equilibrium, evaporation is relatively slow and mov-
def. 10: The relative humidity is defined as the ratio of the ing air brings new dry air masses into contact with the water
vapour pressure to the saturation vapour pressure at the given surface.
temperature of the gas.
However, at a narrow layer near water surface, the air moves
There is another process which is very similar to evapora- slowly due to friction against the water surface; let the thick-
tion: sublimation. In the case of sublimation, the only dif- ness of this (almost) stagnant layer of air be δ (note that δ
ference is that the molecules go straight from solid phase to depends on the wind speed: strong wind makes the stagnant
gaseous phase; examples for sublimation include water at tem- layer narrower. The air molecules in the stagnant layer pre-
peratures below 0 ◦ C, and iodine at room temperatures. In the vent the water vapour molecules leaving into the bulk of the
case of sublimation, the notions of saturation pressure, vapour air: water molecules hit often the air molecules and move ran-
pressure, and relative humidity remain intact. domly back and forth; as a result, their departure speed from
The following problem tests understanding of the concept the water surface is slow as compared with the thermal speed.
of saturation pressure. If the mean free path λ of the vapour molecules is much smaller
than the thickness of that surface layer (which is normally the
pr 21. In vacuum and weightlessness, at the bottom of a cyl- case) then we can consider separately a thin layer of air which
indrical vessel (a cup), there is a layer of solid substance. This is directly above the water surface, with the thickness of few λ
substance sublimes slowly (evaporates from the solid phase into (this layer is the lowest part of the stagnant layer). Within that
gaseous phase) and pushes thereby the vessel to the opposite layer, let us consider two processes: first, molecules jumping
direction. The mass of the vessel is M , and the initial mass of from the water into the thin layer and then returning to the
the substance m ≪ M . The cross-sectional area of the vessel liquid phase after performing few collisions with air molecules;
is A, and the pressure of saturated vapours of the substance at second, vapour molecules leaving the stagnant layer of thick-
the temperature T is p0 . What is the acceleration of the vessel? ness δ. Since λ ≪ δ, the characteristic time scale of the first
Provide answer for two cases: (a) the mean free path length process is much smaller than that of the second process. If one
λ of the molecules at the saturation pressure is much smaller process is much faster than the other, a quasi-equilibrium is
than the length of the vessel, and (b) much larger than that. reached for the faster process. Therefore, inside the thin layer
If the current vapour pressure is smaller than the saturation of thickness λ, thermal quasi-equilibrium is reached between
vapour pressure (i.e. relative humidity is smaller than 100%), the water molecules in the liquid phase and vapour phase: the
evaporation at a liquid surface dominates over condensation, water molecules jump out from the liquid phase almost as often
and the mass of liquid is slowly decreasing. On the other hand, as the vapour molecules hit the surface and get stuck into the
if the relative humidity is larger than r = 100%, condensation liquid. By the definition of the saturation pressure, this means
dominates over evaporation. However, unlike evaporation, con- that inside the thin layer, the vapour pressure equals to the
densation can take place not only at the free liquid surface, but saturation pressure (at the water surface temperature).
also start forming droplets. Still, the vapour molecules cannot If there were no collisions with the air molecules, i.e. if the
start forming a droplet at an empty space as in the case of mean free path of the molecules were very large and those mo-
a very small droplet (of the size of few molecules), the liquid lecules which jump out of the surface would never return, the
phase is not yet formed, and molecular attraction forces are not evaporation rate could be derived from the saturation vapour
strong enough to keep molecules together (the binding energy is pressure. Indeed the full saturation pressure can be divided
smaller than that of a liquid phase, U0 ). Therefore, in order to into two components: the momentum per unit time and unit
start forming a droplet, a condensation centre is needed: it may area, (a) exchanged by those molecules which jump out of the
be dust particle or tiny salt crystal hovering in air. So, around water, with those molecules which remain in the liquid-phase,
each such condensation centre, a tiny droplet is formed; this is and (b) transferred by those vapour molecules which hit the
a mist. If there is a sufficiently large number of condensation water surface and get “stuck” into it, to the liquid water. At
centres, the tiny droplets “suck away” all the excess vapour, equilibrium, both processes have equal intensities, hence both
leading to r = 100%. If the number of condensation centres is contribute half of the saturation pressure. If the second com-
smaller, each of the droplets will need to absorb the vapours ponent were missing, the vapour pressure would be reduced to
from a larger volume, so the process will take longer time. In the half of the full saturation pressure, and the evaporation
— page 14 —

intensity (number of molecules leaving per unit time and unit
area) could be expressed in terms of the saturation pressure,
temperature, and molar mass using the kinetic theory of gases.
In reality, however, as long as we are not dealing with sub-
limation in vacuum, the vapour molecules cannot leave freely
the boundary layer, and many molecules end up colliding again
with the water surface, and condensing at it. Then, the effect-
ive evaporation rate is defined by the interplay of two processes:
molecules (a) jumping out of the water surface, and (b) diffus-
ing through the boundary layer. As discussed earlier, directly
above the water surface, the water and vapours are at equi-
librium, i.e. r = 100%. At the top of the stagnant layer, air
motion brings fresh air with relative humidity r equal to that
of the bulk of the air masses. The thickness of the stagnant
layer depends mainly on the wind speed; narrower layer means
higher vapour concentration gradient, and hence, larger flux
of vapour molecules, i.e. larger evaporation speed. Therefore,
moist things dry faster if the relative humidity of air is low,
and if there is a strong wind.
idea 17: If the dependence between two physical quantities y
and x is given as a graph where y is plotted versus x [y = f (x)],
and you are asked to find, which value of x (or f ) is taken, and
from theoretical considerations one can derive another func-
tional dependence between f and x, for instance y = ax + b,
then the solution can be found as the intersection point of two
graphs, y = f (x) and y = ax + b.
Thus, it suffices to draw a line y = ax + b onto the (provided
to you) graph y = f (x), and read out the intersection point co-
ordinates. If the theoretical dependence is not linear, plotting
efforts are increased: either you redraw the graph y = f (x)
using other coordinates — such that the theoretical depend-
ence would be given by a straight line — or plot onto the given
graph y = f (x) the curve corresponding to the theoretical de-
pendence. In generic case, there is no need for the theoretical
dependence to be a linear function; however, linear depend-
ences are not too rare. For instance, in the case of electrical
circuits, Kirchoff’s laws ensure linearity, c.f. idea 22 from the
booklet of electrical circuits.
pr 22. [EstPhO-1997] One method for finding relative hu- d) A laundry is drying so that there is no direct sunlight falling
midity is based on taking the readings of a dry and a wet
thermometer. The wet thermometer has a wet piece of cloth
wrapped around its sensor. For the method to yield an accurate
result, a wind must blow onto the wet sensor (you can create
it using a fan). Throughout this problem, you can use the
graph of the pressure of saturated water vapours as a function
of temperature.
When wind blows onto the wet sensor, there will be two pro-
cesses influencing the sensor temperature: first, wind removes
the high-humidity air from the neighbourhood of the wet cloth
surface so that the water can evaporate; second, it increases
the heat flux from the warm air towards the colder sensor by
reducing the thickness of the stagnant air layer where the tem-
perature drop is localized. For question a) you may neglect
the effect of the heat flux (i.e. assume that the heat conductiv-
ity of air is very small). In what follows, assume that the air
temperature T0 = 20 ◦ C.
— page 15 —
3. GASES
a) Find the temperature difference between the wet and dry
thermometers if the relative humidity r = 90%.
3
pk(kPa)
2
1
T C
0
0 5 10 15 20 25
b) If we assume that the wind blows with a fixed speed, then
the conductive heat flux Qc (measured in watts) is proportional
to the difference between the temperature of the air T0 and that
of the wet cloth T , Qc = a(T0 − T ). The coefficient of propor-
tionality a depends on the shape and size of the wet cloth, as
well as on the wind speed. Apart from the heat flux due to
conductivity, there is also heat flux due to evaporation: evap-
orating molecules take away heat according to the latent heat
of evaporation. This evaporation heat flux Qe is proportional
to the evaporation rate, which in its turn is proportional to the
difference of the saturation pressure (at the temperature of the
wet cloth), ps (T ), and the pressure of vapours in the surround-
ing air, pa . Thus, Qe = b[ps (T ) − pa ], where the coefficient b
depends on the same things as the coefficient a.
However, it appears that the ratio a/b is a quantity which can
be assumed to be constant for a wide range of conditions, as-
suming that the saturation pressure is much smaller than the
atmospheric pressure. It does depend slowly on the air pres-
sure and temperature, but for any reasonable temperatures and
air pressures at the sea level it can be taken to be equal to
a/b = 65 Pa/K. Taken into account what has been mentioned
above, what would be the reading of the wet thermometer if
the air is absolutely dry (i.e. r = 0)?
c) Derive as simple as possible expression for finding the dif-
ference between the readings of the wet and dry thermometers
within the relative accuracy of 10% which could be applied for
T0 = 20 ◦ C and for the humidity range 80% < r < 100%.
onto it. In one case, the humidity is 95%, in the other case —
80%; all the other conditions are exactly the same. How many
times faster will dry the laundry in the second case?
fact 14: Liquid will start boiling if the condition ps (T ) > patm
is satisfied.
Conversely, if ps (T ) < patm , the external pressure will not
allow expansion of a bubble: larger external pressure would al-
ways compress the bubble back to its initial tiny size (defined
by the impurity size, or the number of non-soluble molecules
of some other gas inside it).
Only extremely clean liquids can be almost free of evapora-
tion centres, in which case it is possible to have what is called
overheated liquids: while ps (T ) is slightly larger than patm , due
to the absence of evaporation centres inside the liquid, evap-
oration can take place only at the surface. Even though this
process is now relatively fast (no wind is needed to accelerate

evaporation as the vapours can now freely push away the air),
there is no evaporation from the bulk in the form of bubbles, i.e.
there is no boiling. This is very similar to the oversaturated va-
pours discussed earlier. Furthermore, under similar conditions
(of very clean liquids) it is possible to have overcooled liquids
which stay in the liquid phase below the melting point. For
instance, a very clean water can stay liquid at few degrees be-
low 0 ◦ C. Both for overheated and overcooled liquids, shaking
can initiate a very fast phase transition: for overheated liquid,
the excess heat is released via latent heat during partial evap-
oration, and in the case of overcooled liquid, part of the liquid
solidifies releasing latent heat and raising the temperature to
the melting point value.
It should be noted that overcooling can happen also with
rain droplets, which leads to the phenomenon known as glaze
frost: as soon as a droplet hits a solid surface (road, electrical
wires, etc), part of the water solidifies at that solid surface.
1,5
1,0
0,5
100 120 140 160 180 200 T C
pr 23. Geysers can be considered as large underground reser- The masses of the liquids A and B were initially the same, each
voirs which are filled with ground water and are being heated
by hot walls. Such a reservoir is connected with the earth sur-
face via a narrow and deep channel, which in the geyser’s rest
state (inactivity period) is filled with relatively cold water (the
channel’s walls are cold and do not heat the water inside the
channel). The geyser becomes active when the water in the
reservoir starts boiling; during the activity period, the channel
connecting the reservoir with the ground surface is filled with
water vapours. The fresh water supply rate of the reservoir is
so slow that the inflow can be neglected during the (relatively
very short) activity period; meanwhile, it is so fast that the en-
tire reservoir and the entire channel (up to the ground surface)
becomes filled with the water during the (relatively very long)
inactivity period.
Let the height difference between the endpoints of the chan-
nel be h = 90 m. The latent heat of evaporation for water
λ = 2.26 × 106 J/kg; its specific heat c = 4.2 × 103 J/kg · K.
The dependence of the saturated water vapour pressure on tem-
perature is depicted in the graph. Find how large fraction of
the water in the reservoir is lost during a single activity period.
If we have an interface between two liquids, the facts 13
and 14 need to be combined: into the bubbles at the interface
of two liquids, the molecules of both liquids can enter in the va-
pour phase. According to the Dalton’s law, the pressure inside
the bubble will be equal to p1s + p2s , where p1s and p2s are the
saturation pressures of the first and second liquid at the given
— page 16 —
3. GASES
temperature, respectively.
fact 15: At the interface of two liquids, boiling can start at
considerably lower temperatures than in both liquids, separ-
ately: boiling will start when the condition p1s + p2s > patm is
satisfied.
pr 24. [IPhO-1989] Consider two liquids A and B insol-
uble in each other. The pressures pi (i = A, B) of their sat-
urated vapours obey, to a good approximation, the formula
ln(pi /p0 ) = ai /T +bi , where p0 denotes the normal atmospheric
pressure, T — the absolute temperature of the vapour, and ai
and bi (i = A, B) — certain constants depending on the liquid.
The values of the ratio pi /p0 , i = A, B, for the liquids A and
B at the temperature 40◦ C and 90◦ C are given in the table
below.
T (◦ C) pA /p0 pB /p0
40 0,248 0,07278
90 1,476 0,6918
a) Determine the boiling temperatures of the liquids A and B
under the pressure p0 .
The liquids A and B were poured into a vessel in which the
layers shown in figure were formed. The surface of the liquid
B has been covered with a thin layer of a non-volatile liquid C,
which is insoluble in the liquids A and B and vice versa, thereby
preventing any free evaporation from the upper surface of the
liquid B, The ratio of molecular masses of the liquids A and B
(in the gaseous phase) is g = MA /MB = 8
equal to m = 100g. The heights of the layers of the liquids in
the vessel and the densities of the liquids are small enough to
make the assumption that the pressure in any point in the ves-
sel is practically equal to the normal atmospheric pressure p0 .
The system of liquids in the vessel is slowly, but continuously
and uniformly, heated. It was established that the temperature
t of the liquids changed with time τ as shown schematically in
the figure. Determine the temperatures t1 and t2 correspond-
ing to the horizontal parts of the diagram and the masses of
the liquids A and B at the time τ1 . The temperatures should
be rounded to the nearest degree (in ◦ C) and the masses of the
liquids should be determined to one-tenth of gram.
REMARK: Assume that the vapours of the liquids, to a good
approximation,
(1) obey the Dalton law stating that the pressure of a mixture
of gases is equal to the sum of the partial pressures of the gases
forming the mixture and
(2) can be treated as perfect gases up to the pressures corres-
ponding to the saturated vapours.
p0 t t2
C
B
t1
A

Surface tension
As we saw in the previous chapter, the molecules of a substance
in the liquid phase are being attracted by the other liquid mo-
lecules and therefore have a certain negative potential energy
with respect to infinity. Notice that those liquid molecules
which are directly at the liquid-gas interface are being attrac-
ted by other molecules only from the side of the liquid phase.
Therefore, for the molecules directly at the surface, as com-
pared with the molecules in the bulk of the liquid, the number
of attracting neighbours is smaller and respectively, the negat-
ive potential energy is also smaller by modulus. This missing
negative energy can be interpreted as a positive energy of the
surface, which is proportional to the number of molecules at
the surface, which is in its turn proportional to the surface
area of the liquid,
U = Sσ,
where the coefficient of proportionality σ is called the surface
tension. At room temperature, it takes typically values from
17 mN/m (diethyl ether) to 73 mN/m (water); and 480 mN/m
for mercury. In the case of water, very small concentrations of
substances known as surfactants can lower the surface tension
two- or three-fold.
Now, let us consider a rectangular shape a×b of a liquid sur-
face, and a process where we increase the side length a by ∆a.
By doing so we increase the surface area by b∆a, and hence,
the surface energy by σb∆a. As the energy was increased, some
work must have been done. In order to increase the side length
a, we needed to pull one edge of length b by ∆a. Let us assume
that in order to do so, a force F was needed. Then, from the
energy balance we can equate the work done F ∆a with the
energy increase σb∆a; hence,
F = σa,
i.e. σ is the force per unit length. To sum up, similarly to the
tension in rope, we can say that if we make an imaginary cut
line of length L on the surface, the two halves of the surface
pull each with force F = σL.
Note that surface tension exists not only at a free liquid sur-
face, i.e. liquid-air interface, but also at liquid-liquid interfaces
between two insoluble liquids, liquid-solid interfaces, solid-gas
interfaces, and solid-solid interfaces. However, unless we have
nano-scale (or smaller) objects in which case the surface-area-
to-bulk-volume ratio (and along with it, the relative import-
ance of the surface energy) is anomalously large, surface ten-
sion doesn’t have noticeable effects at solid-solid and solid-air
interfaces due to the fact that the size and shape of these sur-
faces is fixed by the shape of the solid bodies. As an exception,
the energy of a solid-air interfaces becomes important if it is ad-
jacent to a solid-liquid and solid-air interface. Let us consider
this case in more details.
Let us consider the contact line of three substances, a gas
(e.g. air), a liquid (e.g. water), and a solid (e.g. a glass where
the water is kept). Thus we have liquid-gas, solid-gas, and
solid-liquid interfaces; let the respective surface tensions be σl-g ,
σs-g and σs-l . We study the equilibrium for the position of the
contact line of the three substances. Solid surface cannot be
deformed, so this contact line can move only along the solid sur-
face. Now, suppose that the length of the contact line is L and
it moved towards the liquid phase from its initial position by
— page 17 —
3. GASES
a small distance |AB| = a, see figure depicting a cross-section
of the system. Then the solid-gas surface energy was increased
by σs-g aL, and the solid-liquid surface energy was decreased
by σs-g aL. Suppose that the liquid-air and solid-liquid surfaces
form angle α, see figure. Note that the figure is drawn assum-
ing that the overall volume of the liquid is so large that the
displaced liquid volume ABC flowing rightwards of the point
P does not incur any significant changes in the liquid surface
shape in that region. Then, the displacement of contact line
will reduce the liquid-air surface energy by σl-g aL cos α. Indeed,
from the right triangle ABC, the liquid-gas surface length is de-
creased by |AD| = |AB| cos α (we ignore the length difference
|CD| − |CB| which has second-order smallness in the small
parameter |AB|/|AC|).
C
AD
B
Now we can conclude using the energy balance that it is en-
ergetically favourable for the contact line to move rightwards if
σl-g cos α + σs-l > σs-g , and leftwards, if the opposite inequality
holds; the equilibrium is possible only if
σl-g cos α + σs-l = σs-g .
Now we can express for the so-called contact angle α
σs-g − σs-l
cos α = .
σl-g
In the case of normal liquids and surfaces, the contact angle
is larger than 0 and smaller than π; cases with α < π/2 are
referred to as hydrophilic or wetting, and cases with α > π/2
— hydrophobic or non-wetting. In the case of perfect wetting,
cos α = 1. In the case of even smaller values of σl-g we would ob-
tain cos α > 1, which is clearly impossible; instead, inequality
σl-g + σs-l < σs-g means that the gas-solid surface tension σs-g
is so large and the surface tensions of the liquid is so small that
it is energetically useful for a drop of liquid to disperse over
the whole solid surface so that it will be covered completely
with a thin layer of liquid. If such a liquid is kept in a vessel
with vertical walls, it would be even energetically favourable
for the liquid to “climb” up along the walls: in the case of ex-
tremely thin liquid layer, gravitational potential energy would
be negligible. However, the flow rate of the liquid in an ex-
tremely thin layer will be extremely slow due to viscous drag.
Because of that, such behaviour can be observed only if there
is no viscous drag, in the case of superfluidity (helium at very
low temperatures).
If there were interfaces with σl-g + σs-g ≤ σs-l , it would be
energetically favourable to keep a narrow air gap between the
liquid and solid: liquid would not be sticking to the surface,
and liquid droplets would be able to move essentially without
a drag along the surface. While there are no such surfaces,
there are interfaces for which cos α is fairly close to −1. For in-
stance, the contact angle of mercury on most surfaces is around
140 ◦ .
pr 25. For some natural materials (e.g. lotus leafs), and for
materials with modern nano-technological coatings, the con-
tact angle of liquids is increased due to the micro-structure of

the surface: there are microscopical “whiskers” which keep the
droplets on the tips of the whiskers and reduce thereby the
contact area between the liquid and the solid material. If the
apparent contact area of a droplet with such a surface is A,
and the real contact area due to the microscopic whiskers is
rA with r = 0.006, what is the contact angle? Assume that
without “whiskers”, the contact angle would be α0 = 110 ◦ .
Note that the energy minimum condition can be also in-
terpreted as the force balance condition for the line where
the three phases meet (point A in figure): forces σl-g , σs-l ,
and σs-g (per unit length) pull the liquid molecules at the
three-phase-contact-line along the liquid-gas, solid-liquid, and
solid-gas interfaces, respectively. Hence, then the equality
σl-g cos α + σs-l = σs-g represents the equilibrium condition for
the horizontal direction (along which the line could move if the
area covered by the liquid were to contract or expand).
Let us sum up what we have learned.
fact 16: Interfaces between phases carry energy; each inter-
face is characterized by the surface tension coefficient σ which
gives the energy per unit area. As a direct consequence, σ gives
also the force per unit length of a fictitious cut of the interface
(known as the surface tension or capillary force).
Now, let us study the capillary pressure which is the gauge
pressure due to spherical liquid-air interface of radius r, such
as one would have in the case of a bubble inside a liquid, or a
small droplet hovering in the air. Let us divide the bubble into
two halves by a plane passing through its centre, and consider
the force balance for one of the halves. To begin with, we need
a new idea.
idea 19: Consider (infinitesimally) small volumes (line seg-
idea 18: It is often useful to consider a force balance condi- ments, etc.) and write down applicable force balances and/or
tion for fictitiously separated part of a liquid involving gravity conservation laws for these volumes to derive either a relation-
force, surface tension force, and force due to hydrostatic pres- ship between physical quantities, or differential equations de-
sure.
Let us consider, for instance, an air
bubble inside a liquid, and let us cut it
fictitiously into two equal halves by a flat
surface. The body for which we write the
force balance consists of the air inside the
half-bubble, and the hemispherical liquid-
air interface around it. The forces acting on this body are:
the capillary force across the cut line pulling towards the other
half, F1 = 2πσr; the force due to the hydrostatic pressure pin
inside the bubble, F2 = πr2 pin , exerted by the other half of the
bubble and pushing the two halves apart from each other; the
force F3 due to the hydrostatic pressure pout outside the sphere
acting in the same direction as the capillary force. Note that
pout acts onto a curved surface so that the calculation of the
resultant force would require integration pout x̂ · dS,
⃗ where x̂ is
a unit vector along the axis of the hemisphere, and dS ⃗ is an
infinitesimal surface element (its modulus being equal to the
corresponding surface area, and pointing parallel to the sur-
face normal). However, if such an hemisphere were surrounded
both from the curved and flat sides by the same hydrostatic
pressure (cf. a half of a watermelon subject to the air pressure
in atmosphere), the body would obviously remain still at rest,
i.e. the pressure forces to both sides would be equal. There-
— page 18 —
3. GASES
fore F3 = πr2 pout . And so, the equilibrium condition for the
hemisphere is written as πr2 pout + 2πσr = πr2 pin , hence
pin − pout = 2σ/r.
In the case of a cylindrical surface, we can use a completely
similar approach to obtain pin − pout = σ/r.
However, cylindrical geometry can
x
be analysed also using a different
approach: consider the force bal- F F
ance for a very small surface ele-
ment (in figure, α ≪ 1) in the perpendicular direction to the
surface (the x-axis in figure). The surface tension force F ,
tangent to the surface, is applied to the edges of the surface
element, yielding Fx = − 21 σl sin α ≈ − 12 σlα (l is the length of
the surface element perpendicularly to the figure plane). The
area of the surface element ∆A = lrα, and hence, the force
balance is written as (pin − pout )lRα = 2Fx = σlα, yielding
immediately pin − pout = σ/r. Notice also that in the figure
above, the liquid-gas interface can be thought to be a rope of
tension T , in which case we obtain the normal force per unit
length of the rope n = T /R.
Finally, the last paragraph illustrates a very important and
universal method in physics, differential calculus approach. Dif-
ferential calculus has many flavours: for instance, in mathemat-
ics, we speak most often about taking derivatives and integrals,
and solving differential equations. In physics, it is a very useful
skill to know how to show that one or another physical quant-
ity is a certain integral (cf. idea 2 from the electrical circuits
booklet). Let us try to give a specific recipe — summary about
what was done in the previous paragraph.
scribing a physical quantity as a function of other quantities.
Make use of the smallness of these quantities: drop anything
which has higher order of smallness (e.g. squared small quant-
ities).
Let us also sum up the capillary pressure results.
fact 17: The gauge pressure due to capillary forces across
a curved interface is ∆p = 2σ/r in spherical geometry, and
∆p = σ/r in cylindrical geometry.
It appears that these expressions can be generalized to arbit-
rary shapes of the interface; here we just provide the result,
pin − pout = σ(r1−1 + r2−1 ),
where r1 and r2 are the curvature radii of two curves at their
crossing point P ; the two curves are formed as the intersection
lines of the air-liquid interface with two planes, assuming that
all the three surfaces are mutually perpendicular at the point
P . It can be shown that while the individual curvatures r1
and r2 depend on the orientation of the two plains, the sum
σ(r1−1 + r2−1 ) remains invariant when the planes are rotated
around the normal drawn through the point P to the air-liquid
interface.
In particular, the volume of a meniscus, the mass of a
droplet falling from a water tube, etc. can be found using the
idea 18, i.e. from the force balance condition for the surface ten-
 3. GASES
sion force and the gravity force. The following problem serves
as an illustration here.

pr 26. [EstPhO-1995] Measuring cylinder has volume V =

100 ml and height h = 20 cm. There are marker lines drawn on
the cylinder after each 1ml. Suppose the cylinder is used for
measuring water volume, and the reading is taken according to
the lowest point of the curved water surface. How large mistake
is made due to the fact that the water surface is curved, and One can spot two issues which can invalidate the solution of
not flat? Surface tension coefficient for water σ = 0.073 N/m; the problem 28 for δ & 0.5: (a) at the droplet’s upper rim
Assume that water wets perfectly the walls of the cylinder. (where it is connected to the tap), the droplet walls are no
longer vertical (hence, the force balance needs to include the
cosine of the angle between a vertical line and the surface tan-
gent); (b) the pressure force due to the gauge pressure inside
pr 27. [EstAcadPhO-2003] Two coaxial the droplet needs to be calculated more precisely (curvature
rings of radius R = 10 cm are placed to a dis- radii and hence, the gauge pressure change along the height of
tance L from each other. There is a soap film the droplet). Furthermore, it is not clear, which part of the
connecting the two rings as shown in figure. De- droplet separates from the tap when the droplet starts falling:
rive a differential equation describing the shape r(z) of the film, it is apparently somewhat less than the whole volume depicted
where r is the radial distance of the film from the symmetry in the figure. The figure above indicates the relative correc-
axis, as the function of the distance z along the axis. Show tion to the droplet’s mass ε due the above mentioned effects
that cosh x is one of its solutions. When the distance between (for δ = 0.5 and δ = 1, it is assumed that what separates
rings is slowly increased, at a certain critical distance L0 , the as a falling droplet is the part beneath the narrowest part of
soap film breaks. Find L0 . the droplet). Parameter κ shows the relative difference of the
gauge pressures at the bottom and at the top of the droplet;
It is tempting to apply the idea 18 to the droplets falling red line shows the level of zero gauge pressure.
from a tap, straw, or syringe, to relate the mass of a droplet to The case δ = 3.218 (also shown in figure) yields the largest
the internal diameter of the tap. droplet (with all the volume beneath the rim of the tap be-
ing taken into account); it is also the droplet which is hanging
from a horizontal ceiling when water vapours condense on it.
When the tap radius is further increased, the droplets become
pr 28. Consider droplets falling from a syringe as the A a
even flatter, and at δ ≈ 3.83, the droplet’s height becomes zero.
syringe is kept vertically an pointing downwards; it is pressed
When the tap radius is larger than 3.83λ, the tap can no longer
slowly so that liquid drops are falling from the tip of its needle
contain water: due to what is known as the Rayleigh-Taylor in-
(the needle’s tip is flat, i.e. is cut perpendicularly to its axis).
stability, water flows out.
The surface tension of the liquid is σ, its density is ρ, free
fall acceleration
√ is g, and internal diameter of the needle d idea 20: Equilibrium states of a system can be found as its
(d ≪ σ/ρg) Find the mass m of a falling droplet (provide minimal total energy state; this is valid not only for mechanical
a correction to your answer by taking into account the small systems, but also for systems involving thermodynamic (elec-
pressure force due to the gauge pressure inside the droplet). tromagnetic etc.) phenomena.

pr 29. [EstPhO-1995] Liquid is poured onto a horizontal

totally non-wetting surface where it forms a pool (a layer of
This approach is, indeed, valid if the internal diameter d
liquid). Find the thickness of the layer if the density of the
of the tap/syringe is small. However, if that is not the case,
liquid is ρ and the surface tension coefficient is σ. What would
the problem becomes much more complicated. Based on the
be the thickness if it were a partially wetting liquid, with the
static force balance for horizontal layers of a growing droplet, a
angle between the air-liquid and liquid-solid interfaces being
differential equation can be composed to determine the shape
equal to α (0 < α < π)?
of the droplet. When liquid is being injected into the droplet,
it is growing, and its shape is changing; at a certain moment,
maximal size is achieved: with larger droplet volumes, for the
pr 30. Find the height of the meniscus in the case of prob-
lem 26, i.e. the height difference between the highest and lowest
given diameter of the tap, there are no solutions of the differ-
points of the air-liquid interface. Use the data of problem 26;
ential equation. This is the moment when the droplet falls. In
the liquid density is ρ.
the figure below, the shapes of falling droplets are depicted for
different syringe (tap)
√ radii r (indicated in dimensionless units, Finally, regarding instabilities due to surface tension: any
δ = r/λ, where λ = σ/ρg is the characteristic length scale at liquid tries to take the shape of minimal energy, which in the
which capillary and hydrostatic pressures have the same order case of weightlessness is a sphere. Because of that, any other
of magnitude, σ/λ = ρgλ). shapes, such as cylindrical or flat (soap films) are unstable. In
— page 19 —
 4. ENTROPY AND CARNOT CYCLE
the case of a cylindrical shape, the instability evolves fairly fast; in the case of positive temperatures, these two are identical.
consider the following problem. However, discussion of the Kelvin’s temperature scale serves
us as a useful exercise, and it provides additional insight to
pr 31. [Seagull-2014] Due to instability (know as thermodynamics.
the Plateau-Rayleigh instability), a tap water stream The Kelvin’s definition of temperature is based on the
breaks into droplets at a certain height. This process Carnot’ cycle which will be discussed later in more details; here
can be modelled by the instability of a long water cylin- it is enough to know its definition.
der in weightlessness. Let the diameter of the cylinder
be d = 1 mm; estimate the time period T during which def. 12: Carnot’ cycle is a reversible process with a gas which
the amplitude of the most unstable perturbations will has four stages:
increase by a factor of e ≈ 2.718. The surface-tension (i) the gas expands isothermally (i.e. at a constant temperat-
2
of water γ = 72 g/s , and the density ρ = 1 g/cm . 3 ure) while receiving slowly heat Q1 from a heat reservoir6 at a
temperature T1 ,
Due to the same reason, falling droplets don’t have the
(ii) it expands adiabatically (i.e. slowly without receiving or
shape of a droplet, because a sharp tip would imply a huge
giving away any heat) and is thereby cooled (cf. fact 12) down
capillary pressure beneath the tip, and would push the fluid
to a temperature T2 ;
immediately towards the regions of smaller pressure (larger
(iii) the gas is brought to a thermal contact with another heat
curvature radius). Instead, the falling droplets take a shape
reservoir of temperature T2 ; the gas gives away heat Q2 to the
which (due to air drag) is similar to the shape of small droplets
reservoir and contracts therefore isothermally;
lying on a hydrophobic surface: slightly ellipsoidal. Similarly,
(iv) the gas is compressed adiabatically until it reaches the tem-
thin films are unstable, but in the case of very thin films (soap
perature T1 .
films), the characteristic time for the growth of instabilities is
very large. Soap films are therefore fairly stable, and can be Note that thus far we don’t have quantitative definitions of
met often in problems; in that case, don’t forget the factor “2” temperature, but we know that the temperatures T1 and T2
for pressures, forces and energies reflecting the fact that a film are different, T1 > T2 : based on the fact 2 we know that if
has two interfaces: inner and outer ones! the reservoirs were brought into contact, there would be a heat
flow from the first one to the second one.
Since the pressure of the gas falls when cooling down (see
4 Entropy and Carnot cycle
Section 3), the net mechanical work performed by the gas dur-
The classical theory of thermodynamics is built around the ing the whole cycle is positive (contribution of the expansion
∫
concept of reversible processes. stage A1 = exp pdV > 0 dominates over that of the contrac-
∫
def. 11: Reversible processes are processes for which the sys- tion stage, A2 = contr pdV < 0). Hence, due to the 1LTD,
tem under consideration is always at a thermodynamical quasi- Q1 = Q2 + A1 + A2 > Q2 . So, the system works as a heat en-
equilibrium (those parts of the system which are in a thermal gine: the heat difference goes to mechanical work W = Q1 −Q2 ;
contact must have almost the same temperature), and there the ratio η ≡ Q1 = 1 − Q1 is called the efficiency of the heat
W Q2

should be no dissipation (i.e. conversion of mechanical energy engine.

into heat) inside the system.
So, the system can have two or more thermally isolated parts
of different temperatures, but each of such parts should be
thermally isolated (there should be no heat flux between them).
As soon as heat is allowed to flow from one part to another, the
temperatures of these parts must be almost equal: in that case,
the direction of the heat flow can be reversed by changing the
temperatures only by a very small amount. For a perfectly re-
versible process, the temperature difference should be infinitely
small, so that the temperature change required for a process re-
versal would be infinitely small. However, such a process would
be also infinitely slow, because the heat flow rate between the
bodies would be also infinitely small. Thus, real processes can
be reasonably close to being reversible, but never perfectly re-
versible. Notice that for a reversed process, all the quantities
(exchanged heat, work) obtain opposite signs.
As mentioned before, the best definition of temperature is
based on statistical thermodynamics; this, however, is based
on quantum mechanics, and in pre-quantum-mechanical era, a
different definition, the Kelvin’s one was used. Since we are
already equipped with the statistical definition of temperature,
there is strictly speaking no need to discuss the Kelvin’s one:
6 heat reservoir — so large thermally isolated body that receiving or giving away some heat will not cause any noticeable changes in its temperature.
def. 13: Heat engine is a device which converts heat energy
into a mechanical work using a temperature difference between
the heating and cooling bodies; heat pump makes use of mech-
anical work to “pump” heat energy from a body with lower
temperature to a body with higher temperature. Ideal heat
engine is based on Carnot’ cycle; ideal heat pump makes use of
the reversed Carnot cycle (all the stages are reversed: contrac-
tion is substituted with expansion, cooling is substituted with
heating, etc).
pr 32. Show that for a Carnot’ cycle, the efficiency can
depend only on the temperatures of the heating and cooling
bodies and does not depend on which gas is used as the work-
ing gas.
The solution here is based on the 2LTD, which we formulate
as an idea.
idea 21: The 2LTD can be used to prove by contradiction im-
possibility or nonexistence of various things. To that end, it is
necessary to show that when assuming that the opposite is true,
one can construct a scheme by which energy is transferred from
a body of lower temperature to a body with a higher temper-

— page 20 —
 4. ENTROPY AND CARNOT CYCLE
ature while no net work is done. These constructions usually opposite is not true. The effect can be enhanced by overlaying
make use of ideal heat engines and/or heat pumps. an absorbing and reflecting layers and turning the reflecting
The solution of problem 32 can be directly used to show layer outside. Then, while total transmittance is equal from
more generic (and very useful) statement, which we formulate both sides, the reflectance from outside is larger because from
as a fact. outside, reflected light does not pass through the absorbing
layer.
fact 18: Any reversible heat engine (not necessarily based on
Carnot’ cycle) which takes heat from a heat bath at a single idea 22: The fact (No. 18) that all reversible heat engines
fixed temperature T1 , and gives away heat to another heat bath have the same efficiency ηC = 1 − T1 can be used to solve a
T2

at another fixed temperature T2 , must have (and does have) ex- series of problems.
actly the same efficiency as a Carnot’ cycle, ηC = 1 − TT21 . For First of all, this idea applies to all the problems which deal with
non-ideal (non-reversible) heat engines, the efficiency can be heat engines if these engines involve heat baths with exactly
only smaller than ηC and never larger than that. two different temperatures. In particular, one should keep in
This efficiency expression follows directly from the following mind that devices based on the thermoelectric effect (see fact
temperature definition. According to the Problem 32, the ratio 20) can theoretically be reversible (of course, ohmic dissipation
Q2 /Q1 is a quantity which depends only on the temperatures in such devices is irreversible and needs thus to be excluded
of the heat reservoirs. This fact can be used to quantify the from energy balance). In some cases, it is possible to solve a
temperatures by defining the ratio of the temperatures of two problem (derive an equation) by making a thought experiment
bodies as with a suitably designed hypothetical heat engine, see the next
T1 /T2 = Q1 /Q2 , problem.
where Q1 and Q2 are the heat amounts received and given
pr 35. Derive the so-called Clausius-Clapeyron equality re-
away, respectively, by a fictitious Carnot’ cycle using these two
lating the temperature derivative of the saturation pressures of
bodies as the heating and cooling heat reservoirs. This is the
a substance, dpdT , to the latent heat of evaporation λ, temper-
s

Kelvin’s temperature definition; the temperature unit 1 K is

ature T , saturation pressure ps , and molar mass µ.
defined by postulating that the water’s triple point7 temper-
ature is T0 = 273.16 K. This value for T0 is chosen so that By how much will change the pressure of saturated vapours of
◦
at the atmospheric pressure, the difference of the boiling and water if the temperature is decreased from T0 = 100.0 C down
◦
melting temperatures of water would be equal to 100 K. Note to T1 = 99.9 C? Atmospheric pressure P0 = 1.0 × 10 Pa, lat-
5

that using this definition, the Carnot’ cycle efficiency can be ent heat of evaporation for water L = 2260 kJ/kg. Hint: con-
rewritten as ηC = 1 − Q Q2
= 1 − TT21 . sider Carnot cycle where work is being done by water vapours,
1
and both cooling and heating reservoirs are made of water, at
pr 33. Show that fact 18 is a consequence of the fact 2. temperatures T0 and T1 , respectively.

As mentioned, in statistical thermodynamics it is shown

pr 36. Note that the Clausius-Clapeyron equality can be
that with the statistical definition of temperature (from the −U/kB T
Boltzmann’s law) T̃ , in the case of a Carnot cycle Q1 /Q2 = written as a Boltzmann’s law, pS = p0 e ; express U in
T̃1 /T̃2 . This means that the Kelvin’s definition and the defini- terms of the parameters listed in problem 35, and interpret it
tion from statistical mechanics are identical; equality of units physically.
is achieved if the constant k in the Boltzmann’s law is taken
def. 14: Classical (unrelated to quantum mechanics) entropy
equal to kB ≈ 1.38 × 10−23 J/K.
S is defined only via increments (similarly to potential energy):
pr 34. [IPhO-1992] A manufacturer advertises a special ∆S = ∆Q/T,
paint in the following way: “This paint will reflect more than where ∆Q is the heat given to the system, and T — the tem-
90% of all incoming radiation (both visible light and infrared) perature of the system.
but it will radiate at all frequencies (visible light and infrared)
Conclusion: adiabatic process is an isoentropic process.
as a black body, thus removing lots of heat from the satellite.
While in classical thermodynamics, only entropy increments are
Thus the paint will help keep the satellite as cool as possible.”
defined, in statistical thermodynamics which is based on quantum mech-
Can such paint exist? Why or why not?
anics, absolute values of entropy are well-defined; roughly speaking, it
A consequence of this problem is that radiation and absorp- is S = kB ln N , where N is the number of thermally excited quantum-
tion properties of a material must be identical throughout the mechanical states; more precisely S = −kB ⟨ln pi ⟩i , where pi is the prob-
entire spectrum. It can be similarly shown that a partially ability of the i-the quantum-mechanical state, and angular braces denote
reflecting material must have equal transmittance from both averaging over all the possible states. Using mathematical statistics, one
sides. It may seem that dark window glasses are more trans- can show that this definition yields ∆S = ∆Q/T , in agreement with the
parent when looking from inside of a darkly lit room, but this classical definition.
is a mere illusion: when looking from outside, a small fraction It is quite easy to see that for a closed system undergoing re-
of reflected abundant outside light can easily dominate over versible processes, the entropy is conserved. Indeed, since heat
the transmitted part of the light coming from inside, but the exchange between its different parts takes place at the same
7 Triple point of a substance — such a combination of temperature and pressure that the solid, liquid and gaseous states are all in thermal
equilibrium with each other.

— page 21 —
 4. ENTROPY AND CARNOT CYCLE
temperature then by exchanging a certain amount of heat, the and length of the rectangle — ∆S. According to the defini-
two parts obtain equal by modulus and opposite entropy incre- tion of entropy, the heat received from the heating reservoir
∫B
ments. However, in the case of irreversible processes, according ∆Q1 = A T dS = T1 ∆S is the surface area of the large
to the 2LTD (idea 2) that part of the system which gives away rectangle; similarly, the heat given to the cooling reservoir
∫D
heat (−∆Q < 0) has larger temperature than that part which ∆Q2 = C T dS = T2 ∆S is the surface area of the dark-grey
receives the heat, T1 > T2 . Hence, the total entropy change for rectangle. According to the energy conservation law, the work
H
the entire system ∆S = − ∆Q ∆Q
T1 + T2 > 0 is positive. done ∆W = ABCD T dS = ∆Q1 − ∆Q2 = ∆S(T1 − T2 ) is
the surface area of the light grey rectangle. According to the
fact 19: The entropy of a closed system remains constant definition of efficiency,
in reversible processes and grows in irreversible ones. ∆W T1 − T2 T2
ηC = = =1− .
∆Q1 T1 T1
Being equipped with the statistical definition of entropy, we can say
Reversed Carnot’ cycle has a counter-clock-wise motion in
that if a system evolves by itself, the total number of excited quantum-
the S − T -diagram: all contractions become expansions (and
mechanical states can only grow. This can be formally shown using math-
vice versa), heat flow direction is reversed, and mechanical work
ematical statistics. The number of excited states can be also considered as
becomes negative, i.e. a work needs to be done to keep the
a parameter describing the degree of order: more occupied states implies
process going on. Such a reversed heat engine can be used
less order. So, the degree of order of a closed system is can only increase.
for two purposes: for heat pumps and fridges. In a fridge,
Yo can use this as an excuse if your room becomes messy.
a working gas takes heat from the interior of the fridge at the
Finally, notice that the facts 19 and 18 are also often re- inside-temperature T2 , flows through the ribs at the back of the
ferred to as the 2LTD, which is OK since these are equivalent fridge and gives there more heat away at the ribs’ temperature
to fact 2: an entropy reduction would directly mean that heat T1 > T2 ; the electrical engine of the fridge keeps the process
has been transferred from a lower temperature body to a higher going on. The fridge efficiency (often called the coefficient of
temperature one. performance, or COP) is characterized by the ratio of the heat
We have shown already equivalence between the fact 2 and Q2 which is taken away from the inside, and work W done by
19. A hypothetical heat engine with η > ηC is referred to as the engine (consumed electrical power),
perpetuum mobile of second kind, and it is as impossible as the Q2 T2
ηfridge = = .
perpetuum mobile of the first kind (the one which violates en- W T1 − T2
ergy conservation law). However, people tend to trust the laws From this expression we can deduce that the efficiency of a
of statistics less than the other laws of nature (else, who would fridge can be both larger and smaller than one, but it is very
buy lottery tickets!). This has resulted in a large number of difficult to achieve extremely low temperatures, because the
failed attempts at creating perpetuum mobile of second kind, efficiency goes down together with the temperature inside the
but also in really interesting paradoxes. Perhaps the most fam- fridge.
ous one is the Maxwell’s demon: a nano-scaled guy who sits at Heat pump can be considered as a fridge, which has its hot
a gate between two parts of a vessel and who opens the gate ribs inside our living rooms, and which takes heat from the
when a fast molecule is approaching, and keeps the gate closed colder heat reservoirs outside the house (air or ground). Now,
otherwise. Apparently, the temperature at the other side of what is useful to us is the heat Q1 received by our living room,
the vessel would start rising, violating 2LTD. The resolution of so that the efficiency of a heat pump
Q1 T1
the paradox is that the guy needs to obtain information about ηHP = = ,
W T1 − T2
approaching molecules and needs to probe these by sending, for which is always larger than one: it works always more efficiently
instance, photons. However, photons scattered from molecules than an electrical radiator.
would contribute to an increase of entropy. Finally, let us discuss the thermoelectric effect. To begin
Now, let us consider the forward and reversed Carnot’ cycles with, let us recall problem 36: for a molecule to be able to
in more details. leave the liquid phase, a certain energy U needs to be supplied.
This means that effectively, the liquid phase molecules obey po-
idea 23: The processes involving Carnot’ cycle are typically
tential energy −U with respect to vapour phase molecules; this
most conveniently studied using a S −T -diagram, because then,
potential energy is referred to as the chemical potential, and is
the process has a rectangular shape.
equal to the change of total energy when one particle is ad-
ded to a system (final energy minus the particle’s and system’s
initial energies). Similarly to molecules in liquid phase, elec-
trons in metals (semiconductors, dielectrics) are also described
by chemical potential, often referred to as the Fermi level EF .
The Fermi level of a material depends on temperature; if the
temperature of a wire changes along its length, the Fermi level
will also change along its length. As a result, electrons move to-
wards lower Fermi levels (smaller potential energy), creating a
While we have already derived the efficiency of a Carnot’ surplus of charges and an electric field. Finally, an equilibrium
cycle starting from the Kelvin’s definition of temperature, let is reached: electrostatic potential compensates Fermi level dif-
us do it, once again, starting from the definition of entropy ference. So, at equilibrium, there is a changing potential along
and using the S − T -diagram. Let the S-axis be horizontal, the wire. Let us summarize.
— page 22 —
 4. ENTROPY AND CARNOT CYCLE
fact 20: If a wire is being heated from its one end, and cooled pr 37. A thermocouple is being used as a battery: one
at the other end so that the temperatures of its endpoints are of its solder joints is at the room temperature T1 = 20 ◦ C,
Th and Tc , respectively, then there will be a voltage the other is kept inside a glass of water with ice, at temper-
V = S(Th − Tc ) ature T2 = 0 ◦ C. The output leads are connected to a res-
between the endpoints, where the Seebeck coefficient S takes istor R = 10Ω, connected in series with an ammeter which
different for different materials. This is known as the Seebeck shows that current I = 10 mA. How much ice is melted during
effect. t = 10 h if the glass with ice has very good thermal insulation
and room temperature remains constant? Latent heat of melt-
A thermocouple is a device made of wires of two different ing for ice λ = 330 kJ/kg. Assume the energy transfer processes
materials which are selected so that the difference of Seebeck in the thermocouple to be reversible.
coefficients would be as large as possible. A wire of material A
is connected at its two endpoints, the junctions, to the wires of appendix 1: Motivation of the Boltzmann’s law
material B; if the junctions are kept at different temperature, The reason why an energy level occupation probability vanishes
an electromotive force is created at the output terminals of the exponentially with energy lies in the fact that in the case of a
thermocouple. Thermocouples which are used for generation heat reservoir which is supposed to consist of a large number of
of electrical energy are called thermoelectric generators. Ther- subsystems, the number of quantum-mechanical states grows
mocouples take and release heat at the junctions, and perform exponentially with the total energy Etot . Let us consider sim-
electrical work, so they operate as heat engines. If we were to plifyingly a body which has several equispaced energy levels,
reverse such a heat engine, we would need to supply a current E0 , E1 = E0 + E , . . . En = E0 + nE . Further, let the heat
into it from an external source. Then we would expect that reservoir be made of a large number N ≫ n identical particles,
one of the junctions would release heat, and the other would each of which has two energy levels, 0 (ground level) and E
absorb heat; this is, indeed, what will happen, and is referred (the excited state), and let the total number of particles at the
to as the Peltier effect. The direction of current defines, which excited state be M (with M, N − M ≫ n). Now, the body is
of the junctions will absorb heat. brought to a contact with the reservoir; let us assume that its
For majority of metals, S takes values remaining less than energy becomes equal to nE . Due to the energy conservation
10 µV/K, and for certain metal alloys up to 30 µV/K (70 µV/K law, this will decrease the number of excited particles inside
for bismuth). What matters in the case of a thermocouple, is the heat reservoir by n. Thus, the number of excited particles
the difference of Seebeck coefficients of the two wire materials, is M − n; the number of dierent states of heat reservoir satis-
and in the case of chromel and constantan pair, the difference fying this condition equals to the number of dierent ways of
at the room temperature is 62 µV/K. For semiconductor ma- selecting M − n particles from the set of N particles, given by
( )
terials, S can reach much higher values. N N!
= .
Now we can ask, how close to being reversible is the thermo- M −n (M − n)!(N − M + n)!
electric effect. For a system to be reversible, all its parts need For the ground level of the body, the number of the heat reser-
to have reached a state very close to a thermal equilibrium. voir states is similarly ( given
) by
In the case of a thermocouple, however, we have a wire which N N!
= .
connects cold and warm junctions: the temperature difference M M !(N − M )!
Since all these states are equally probable, the ratio of prob-
creates a heat flux through the wire, which is very irreversible.
abilities for the n-th excited state and the ground state of the
For the process to be reversible, the irreversible heat flux
body is given by the ratio of the number of states,
κA
Φir = (Th − Tc ) pn N! M !(N − M )!
l = =
(where κ is the heat conductivity, A — cross-sectional area, and p 0 (M − n)!(N − M + n)! N! ( )n
(M − n + 1)(M − n + 2) . . . M M
l — length) needs to be small as compared with the reversible ≈ .
(N − M + 1)(N − M + 2) . . . (N − M + n) N −M
heat flux which is spent on producing the electrical power, M
Here,
N −M is a number which characterizes the state of the
P V 2 Th S 2 (Th − Tc )Th
heat reservoir; we can introduce the parameter β = − ln(
M −
1 N
Φr = = = . E
ηC R Th − Tc ρl/A
1), which leads us to the Boltzmann's law
So, the process can be considered to be reversible if pn
= e−βEn .
Φr S 2 Th p0
= ≫ 1.
Φir κρ This proof  is probably not too satisfying for a mathematically
This dimensionless parameter is called the merit factor and oriented reader, because we made very limiting assumptions for
denoted by the particles of the heat reservoir. For a more generic proof, one
S2T can introduce the energy level density g(E) of the particles
ZT ≡ ;
κρ making up the heat reservoir, express the energy level density
currently the materials with highest merit factor reach values
of the entire reservoir as the N -th convolution of g(E) with
ZT ≈ 2.5, but theoretically there is no upper limit for ZT . It
itself, calculate it in the Fourier space (because the Fourier
appears that in the case of ZT ≈ 2.5, the maximal efficiency
transform of a convolution is just the product of the Fourier
of a thermocouple is 30% of the Carnot’ cycle efficiency. It is a
transforms), take the inverse Fourier transform and estimate it
challenge for material sciences to create materials with higher
asymptotically (for N → ∞) using the saddle point method 
merit factors which could be used for making compact thermo-
these techniques are well beyond the scope of this booklet.
electric generators.
— page 23 —
 4.
Revision problems
pr 38. [EstPhO-2007] Certain room is being heated using a
heating device the output power P (T ) of which depends on the
room temperature T as shown in figure. If the outside temper-
ature is T1 then the room temperature will reach value T2 (see
figure). Which room temperature will be reached if the outside
temperature is T3 ? (find the solution graphically using the fig-
ure). The heat exchange loss rate of the room is proportional
to the difference of the inside and outside temperatures.
P (T )
ENTROPY AND CARNOT CYCLE
a piston can move without friction up and
down. The cylinder is divided into two com-
partments by a freely moving weightless wall
which conducts slowly heat, see figure. Ini-
tially the gas temperatures are equal, and
the volume of hydrogen is 3 times smaller than that of helium.
Helium receives a certain amount of heat, because of which the
piston moves up by d1 = 5 cm. After waiting for some longer
period of time, an additional displacement of the piston was ob-
served. By how much and in what direction the piston moved
additionally? Gases can be considered to be ideal. Molar heat
capacitance at constant pressure is CP H2 = 7/2R for hydrogen,
and CP He = 5/2R for helium.

pr 42. [EstFin-2006] According to the wide-spread belief,

it is useful to keep window open when drying laundry even if
the relative humidity outside is 100%, because the temperat-
ure of the incoming air rises and thereby the relative humidity
drops. Let us analyse, do these arguments hold, when heating
is switched off.
Suppose that inside a room, the volume of air V1 = 20 m3
from inside at the temperature t1 = 25 ◦ C is mixed with the
volume of air V2 = 10 m3 from outside at the temperature
0 T1 T2 T3 T t2 = 1 ◦ C. The specific heat of the air (by fixed pressure)
cp = 1005 J/(kg · K) can be assumed to be constant for the
pr 39. [EstPhO-2000] The nominal voltage of a light bulb is given temperature range; the heat exchange with the medium
V0 = 26 V and nominal current I0 = 0.12 A. Cold tungsten fila- can be neglected. For the time being, you may neglect the
ment of this light bulb has resistance R0 = 24 Ω. Estimate the possibility of (partial) condensation of the vapour.
length l and the diameter d of this filament. Also, at which tem- a) Prove that the total volume of the air will not change i.e.
perature T is this filament supposed to emit light (when work- that the volume of the mixed air V = V1 + V2 .
ing in nominal regime)? The resistivity of tungsten at room
b) What is the temperature of the mixed air T ?
temperature ρ0 = 5.3 × 10−8 Ω · m. For metals, the resistivity
can be taken to be proportional to the temperature in Kelvins.
Treat tungsten as a perfectly grey body which absorbs k = 0.3
of the incident radiation at any wavelength. Stefan-Boltzmann
constant σ = 5.67 × 10−8 W/(m2 · K4 ). Neglect thermal expan- 2,0
sion of the filament.

pr 40. [EstPhO-2006] Due to cold weather, heating system 1,5

was broken and temperature in a room started decreasing. A
fan heater was quickly bought and switched on. During all
that period, the room temperature changed in time as shown in
graph. Which room temperature will be reached in long term?
1,0
Outside temperature remained constant. Note that due to cer-
tain construction elements, the heat exchange rate between the
room and outside environment was a nonlinear function of the t (0C)
temperature difference. 0 5 10 15 20 25
24

20 c) The graph below shows the dependence of the saturated

16
vapour density for water as a function of temperature. Before
mixing, both the interior and exterior air had relative humidity
T C

12
of r0 = 100%. What is the relative humidity r of the mixed air
8
(if it happens to increase then assume that an oversaturated
4
vapour with r > 100% is formed)?
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 d) If you happened to obtain r > 100% then the oversatur-
ated vapour breaks down into a fog which contains tiny water
droplets. In that case, what is the mass m of the condensed
pr 41. In a thermally isolated cylinder, water (i.e. the total mass of the water droplets)? Air dens-
— page 24 —
 4. ENTROPY AND CARNOT CYCLE
ity ρ0 = 1.189 kg/m3 ; latent heat of vaporization for water hand pump. A manometer shows that the excess pressure
q = 2500 kJ/kg. (the difference between the pressures inside and outside) is
p1 = 2 × 105 Pa, the air pressure p0 = 1 × 105 Pa. Find the
pr 43. [IPhO-1999] A cylindrical vessel, with its axis ver- temperature of the air which enters the tire through the valve.
tical, contains a molecular gas at thermodynamic equilibrium. The pump has two valves. One of them lets gas into the pump
The upper base of the cylinder can be displaced freely and is from the atmosphere; it opens as soon as the pressure inside the
made out of a glass plate; let’s assume that there is no gas pump becomes lower than p0 , and closes as soon as it becomes
leakage and that the friction between glass plate and cylinder larger than p0 . The other lets gas from the pump into the tire;
walls is just sufficient to damp oscillations but doesn’t involve it opens as soon as the pressure inside the pump becomes lar-
any significant loss of energy with respect to the other energies ger than the pressure inside the tire, and closes as soon as this
involved. Initially the gas temperature is equal to that of the inequality no longer holds. Room temperature T0 = 20 ◦ C
surrounding environment. The gas can be considered as perfect
within a good approximation. Let’s assume that the cylinder pr 45. [EstPhO-2002] The main components of a car engine
walls (including the bases) have a very low thermal conductivity are: a cylinder, a piston which moves inside it, and a valve for
and capacity, and therefore the heat transfer between gas and letting gases in and out of the cylinder.
environment is very slow, and can be neglected in the solution piston
of this problem. Through the glass plate we send into the cylin- valve
cylinder
der the light emitted by a constant power laser; this radiation is
easily transmitted by air and glass but is completely absorbed The working cycle of the engine consists of the following stages:
by the gas inside the vessel. By absorbing this radiation the I. Gas entry: the piston moves from the rightmost position to
molecules reach excited states, where they quickly emit infrared left; fresh air comes in through the valve and fills the cylinder.
radiation returning in steps to the molecular ground state; this II. Pressure increase: the valve closes, and the piston moves
infrared radiation, however, is further absorbed by other mo- back to the rightmost position; the air is compressed adiabat-
lecules and is reflected by the vessel walls, including the glass ically.
plate. The energy absorbed from the laser is therefore trans- III. Work: Fuel is injected into the cylinder, and is ignited. You
ferred in a very short time into thermal movement (molecular may assume that the fuel burns instantaneously. Gas starts ex-
chaos) and thereafter stays in the gas for a sufficiently long panding and pushes the piston to the leftmost position.
time. We observe that the glass plate moves upwards; after a IV. Gas disposal: valve is opened, piston moves to right and
certain irradiation time we switch the laser off and we measure the gas is pushed out from the cylinder. The process starts
this displacement. periodically repeating.
a) Using the data below and - if necessary - those on the sheet a) Depict the entire cycle in p − V -diagram.
with physical constants, compute the temperature and the pres- b) Find the efficiency η.
sure of the gas after the irradiation. Data. Adiabatic index for air γ = 1.4. The compression factor
b) Compute the mechanical work carried out by the gas as a is defined as the ratio of the largest and smallest volumes of
consequence of the radiation absorption. the cylinder, k = Vl /Vs ; here k = 10. Neglect friction when
c) Compute the radiant energy absorbed during the irradi- the piston moves. The number of fuel molecules is much smal-
ation. ler than the number of air molecules inside the cylinder. The
d)Compute the power emitted by the laser that is absorbed by air is to be considered as an ideal gas, one mole of which has
the gas, and the corresponding number of photons (and thus internal energy equal to U = cV T , where cV is its molar heat
of elementary absorption processes) per unit time. capacitance by constant volume.
e) Compute the efficiency of the conversion process of optical N(103hours)
energy into a change of mechanical potential energy of the glass
plate. 8

Thereafter the cylinder axis is slowly rotated by 90 ◦ , bringing 7

it into a horizontal direction. The heat exchanges between gas 6
and vessel can still be neglected.
5
f) State whether the pressure and/or the temperature of the
gas change as a consequence of such a rotation, and - if that is 4
the case what is its/their new value. 3

Data: Room pressure: p0 = 101.3 kPa; Room temperature: 2

T0 = 20.0 ◦ C; Inner diameter of the cylinder: 2r = 100 mm; 1
Mass of the glass plate: m = 800 g; Quantity of gas within the T(oC)
vessel: n = 0.100 mol; Molar specific heat at constant volume -40 -30 -20 -10 0 10 20 30
of the gas: cV = 20.8 J/(mol · K); Emission wavelength of the pr 46. [EstPhO-2002] Archive storage rooms need to main-
laser: l = 514 nm; Irradiation time: : ∆t = 10.0 s; Displace- tain constant temperature T0 = 15 ◦ C throughout the year.
ment of the movable plate after irradiation: ∆s = 30.0 mm. Find the annual electricity expenses needed to maintain that
temperature if the heat exchange rate throughout the walls
pr 44. [EstPhO-1998] A tire is being pumped using a of the building P = C∆T , where C = 200 W/K and ∆T is
— page 25 —
 4. ENTROPY AND CARNOT CYCLE
the temperature difference inside the storage room and outside metal plate cold air
the building. In order to regulate the temperature inside the
warm air
room, heat pump is used; the heat pump can also operate as
an air conditioner. If the device is used as an air conditioner b) Attached is a plot of the heat exchange rate P of the wire of
(for cooling the room), the efficiency ηc = 10 (usually referred an electric heater as a function of temperature (assuming the
◦
to as coefficient of performance, COP; this is the ratio of the room temperature is T0 = 20 C). The operating temperature
◦
cooling power to the consumed electrical power); if used as an of the wire is T1 = 800 C. The heater is switched off; find the
heat pump (for heating), the efficiency ηh = 6 (COP; the heat- time after which the temperature of the wire will drop down to
◦
ing power and electrical energy consumption rate ratio). The T2 = 100 C. The heat capacitance of the wire is C = 10 J/K.
attached N − T -graph depicts how many hours in year (N ) P(W )
the outside air temperature was higher than T . The electrical
300
energy cost is c = 0.1 EUR/kWh.

200
pr 47. [EstPhO-2001] Using the assumptions and numer-
ical data of problem 3, and the density of tungsten at room
temperature ρ = 19 250 kg/m3 , answer the following questions. 100

a) A smaller-than-nominal direct voltage U0 = 3 V is applied o

200 400 600T( C )
to the leads of the halogen bulb. How long does it take for the
tungsten filament to reach temperature T1 = 40 ◦ C starting
from the room temperature Tr = 20 ◦ C?
pr 49. [EstFin-2004] Consider a passive cooling system de-
picted in figure. Cold air (at normal conditions: p0 = 105 Pa,
b) Accidentally, a too large DC voltage U1 = 120 V is applied
T0 = 293 K) flows over the heat sink of a chip of power dissip-
to the leads of the lamp. How long does it take to reach the
ation P = 100 W, into a vertical pipe of length L = 1 m and
melting temperature T2 = 3410 ◦ C of the tungsten? Heat losses
cross-sectional area S = 25 cm2 . After passing the pipe, air
can be neglected, as well as the temperature dependence of the
enters the ambient room. Assume that the air inside the pipe
tungsten’s density and specific heat; use the graph depicting
becomes well mixed; neglect the viscous and turbulent friction
the resistivity of tungsten as a function of temperature.
of air inside the pipe and heat sink. Air can be considered as
an ideal gas with adiabatic exponent γ = 1.4 and molar mass
µ = 29 g/mol; gas constant R = 8.31 J/(K · mol)

pr 48. [EstFin-2005] a) Consider a simplified model of the

air ventilation system of a house using a passive heat exchanger.
The exchanger consists of a metal plate of length x and width
y and thickness d dividing the air channel into two halves, one
for incoming cold air, and another for outgoing warm air. Both
channels have constant thickness h, air flow velocity is v see
Figure. Thermal conductance of the metal is σ (the heat flux
through a unit area of the plate per unit time, assuming that
the temperature drops by one degree per unit thickness of the
plate). Specific heat capacity of the air by constant pressure is
cp , air density is ρ (neglect its temperature dependence). You
may assume that the air is turbulently mixed in the channel,
so that the incoming and outgoing air temperatures Tin and
Tout depend only on the coordinate x (the x-axes is taken par-
allel to the flow velocity), i.e. Tin ≡ Tin (x) and Tout ≡ Tout (x).
Assuming that the inside and outside temperatures are T0 and
T1 , respectively, what is the temperature T2 of the incoming
air at the entrance to the room?
— page 26 —
a) Express heat capacitance at constant pressure cp via quant-
ities γ and R.
b) Find a relationship between the outflowing air density ρ and
temperature T (the relationship may contain also the paramet-
ers defined above).
c) Find a relationship between the air flow velocity in the pipe
v and outflowing air density ρ (the relationship may contain
also the parameters defined above).
d) Express the power dissipation P in terms of the air flow ve-
locity v, the outflowing air temperature T , and density ρ (the
relationship may contain also the parameters defined above).
e) What is the temperature T of the outflowing air? In your
calculations, you may use approximation T − T0 ≪ T0 .
pr 50. [EstPhO-2000] Glycerol is stored in a tightly closed
vessel of volume V = 1 l; inside glycerol, there is an air bubble
 4. ENTROPY AND CARNOT CYCLE
of volume w = 1 ml. At temperature T0 = 20 ◦ C, the pressure c) What temperature T2 is measured at the ridge of the moun-
inside the vessel is p0 = 1 atm. The linear expansion coeffi- tain range?
cient of the vessel material is very small and the vessel walls
d) Determine the height of the water column (precipitation
are thick. The volumetric expansion coefficient of the glycerol
level) precipitated by the air stream in 3 hours, assuming a
α = 5.1 × 10−4 K−1 . Air can be treated as an ideal gas.
homogeneous rainfall between points M1 and M2 .
a) Find the dependence of the pressure inside the bottle as a
function of temperature. e) What temperature T3 is measured in the back of the moun-
b) At which temperature and for which physical reason the tain range at station M3 ? Discuss the state of the atmosphere
obtained equality will no longer be valid? at station M3 in comparison with that at station M0 .
Hints and Data. The atmosphere is to be dealt with as an
pr 51. [EstPhO-2003] Juice bottle is being pasteurised at ideal gas. Influences of the water vapour on the specific heat
temperature t1 = 80 ◦ C so that a light cap lies freely on the capacity and the atmospheric density are to be neglected; the
bottle rim (excess gas can exit freely from the bottle, but out- same applies to the temperature dependence of the specific lat-
side air cannot come in). Then the cap is tightly fixed so that ent heat of vaporisation. The temperatures are to be determ-
air can no longer enter and exit from the bottle; the cap is so ined to an accuracy of 1 K, the height of the cloud ceiling to an
rigid that its deformation due to excess pressure can be neg- accuracy of 10 m, and the precipitation level to an accuracy of
lected. The bottle is cooled down to the room temperature 1 mm. Specific heat capacity of the atmosphere in the pertain-
t2 = 20 ◦ C. What is the pressure under the bottle cap? As- ing temperature range: cp = 1005 J/(kg · K). Air density for at
sume that the physical properties of the juice are identical to the station M0 (i.e. for p0 and T0 ) is ρ0 = 1,189 Kg/m3 . Specific
that of water; the pressure of saturated water vapours is de- latent heat of vaporisation of the water within the volume of the
picted as a function of temperature in a graph. Atmospheric cloud: qV = 2500 kJ/kg; free fall acceleration g = 9,81 m/s2 .
pressure P0 = 1.01 × 105 Pa. Adiabatic index for both wet and dry air γ = cp /cV = 1,4.
105
P(Pa)
pr 53. In a science fiction novel, the following situation is
4.104
described. There is an emergency on a spaceship, and an astro-
2.104 naut got by an accident to a distance of L = 100 m from the
104 spaceship. He has a cup with solidified water (ice) and uses the
sublimation (evaporation) of ice to return to the spaceship. Es-
4.103 timate, how realistic is this method. You may assume that the
2.103
sublimation takes place at a constant temperature T = 272 K,
by which the pressure of saturated vapours is P = 550 Pa. Es-
103
t( oC) timate the dimensions of the cup and the mass of the water by
0 20 40 60 80 100 yourself.

pr 52. [IPhO-1987] Moist air is streaming adiabatically

pr 54. [EstFin-2009] Lord Rayleigh had in 1891 a lecture
across a mountain range as indicated in the figure. Equal atmo-
about taking photos of physical processes. Among others, he
spheric pressures of p0 = 100 kPa are measured at meteorolo-
showed a photo of a soap film, which is falling apart (see figure).
gical stations M0 and M3 and a pressure of p2 = 70 kPa at sta-
Instead of a flash, he used an electric spark (well, nowadays the
tion M2 . The temperature of the air at M0 is t0 = +20◦ C. As
flashes are also based on electric sparks). Estimate, how pre-
the air is ascending, cloud formation sets in at p1 = 84,5 KPa.
cise must have been the timing, i.e. estimate the time for a
Consider a quantity of moist air ascending the mountain with a
soap film to fall apart. Let the thickness of the soap film be
mass of 2000 kg over each square meter. This moist air reaches
h = 1 µm, the ring diameter D = 10 cm and the surface tension
the mountain ridge (station M2 ) after 1500 seconds. During
σ = 0.025 N/m.
that rise an amount of m = 2,45 g of water per kilogram of air
is precipitated as rain. Hint: you may use a model, according
to which the already broken part of the
M soap film gathers into a single front and
moves all together towards the still pre-
served part of the film.
M
pr 55. Consider two soap bubbles which have “stuck” to-
M M gether. Now three soap film parts can be distinguished: one
separates the interior of the first bubble from the outside air —
a) Determine temperature T1 at M1 where the cloud ceiling this has curvature radius R; second separates the interior of the
forms second bubble from the outside air — this has curvature radius
2R; third separates the interiors of the two bubbles. Find the
b) What is the height h1 (at M1 ) above station M0 of the cloud surface area of the third soap film part.
ceiling assuming a linear decrease of atmospheric density?
— page 27 —
 4. ENTROPY AND CARNOT CYCLE

pr 56. [EstFin-2013] Sun-rays are focused with a lens of Let us model the Hadley circu-
diameter d = 10 cm and focal length of f = 7 cm to the lation simplifyingly as a heat
black side of a thin plate. One side of the plate is perfectly engine shown in the schematic
black, and the other side is perfectly white. Angular dia- below. For simplicity, let us as-
meter of the Sun is α = 32′ and its intensity on the surface sume that when air moves from
of the Earth is I = 1000 W/m2 , Stefan-Boltzmann constant A to B, it expands adiabatic-
σ = 5.670 × 10−8 W/(m2 K4 ). (i) Find the temperature of the ally, and when it moves from D
heated point of the plate. (ii) Using thermodynamic arguments, to E, it contracts adiabatically.
estimate the maximal diameter-to-focal-length ratio of a lens. Further, let us assume that the
processes with air when it moves from B to D, and from E to
A are isothermal; on segment E-A it receives heat from the
surface, on segment B-C — receives heat from vapour condens-
pr 57. [EstFin-2008] There is wet wood burning in a fire- ation, and on segment C-D — gives heat away due to radiation.
place on the ground. Seven meters above ground, the smoke
a) Given that atmospheric pressure at a vertical level owes its
is at a temperature of t7 = 40 ◦ C. Disregard the exchange
origin to the weight of the air above that level, and knowing
of heat with the surrounding air and assume that the atmo-
that pA = 1000 hPa and pD = 225 hPa, order the pressures pA ,
spheric pressure at the ground is constant in time and equal to
pB , pC , pD , pE , respectively at the points A, B, C, D, E by a
p0 = 1000 hPa; the air temperature t0 = 20 ◦ C is independent
series of inequalities.
of height 8 . Assume that the smoke represents an ideal gas
of a molar mass µ = 29 g/mol (i.e. equal to the molar mass b) Let the temperature next to the surface and at the top of the
of the air), and of a molar specific heat at constant volume atmosphere be TH and TC respectively. Given that the pres-
CV = 2.5R; universal gas constant R = 8.31 J/kg · K. How sure difference between points A and E is 20 hPa, calculate TC
high will the smoke column rise? for TH = 300 K. Note that the ratio of molar gas constant
(R) to molar heat capacity at constant pressure (cp ) for air is
κ = 2/7.
c) Calculate the pressure pB .
pr 58. [IPhO-2014] A bubble of radius r = 5.00 cm is a
soap film of thickness h = 10.0 µm containing a diatomic ideal d) For an air mass moving once around the winter Hadley cir-
gas. It is placed in vacuum. The soap film has surface tension culation, using the molar gas constant, R, and the quantities
σ = 4.00 × 10−2 N/m and density ρ = 1.10 g/cm3 . defined above, obtain expressions for the net work done per
a) Find a formula for the molar heat capacity of the gas in the unit mole Wnet ignoring surface friction and for the heat loss
bubble, for a process in which the gas is heated so slowly that per unit mole Qloss at the top of the atmosphere.
the bubble remains in mechanical equilibrium. Evaluate your e) What is the value of the ideal thermodynamic efficiency εi
answer. for the winter Hadley circulation?
b) Find a formula for the frequency ω of small radial oscilla- f) Prove that the actual thermodynamic efficiency ε for the
tions of the bubble and evaluate it under the assumption that winter Hadley circulation is always smaller than εi .
the heat capacity of the soap film is much greater than the heat
capacity of the gas in the bubble. Assume that the interior of g) Which of the following statements best explains why ε is less
the bubble reaches thermal equilibrium much faster than the than the ideal value? Tick the correct answer(s). There can be
period of oscillations; also, the total mass of the gas is much more than one correct answer.
smaller than the mass of the soap film. Neglect the possibility (I) We have ignored work done against surface friction.
of soap film’s evaporation. (II) Condensation occurs at a temperature lower than the tem-
perature of the heat source.
(III) There is irreversible evaporation of water at the surface.
(IV) The ideal efficiency is applicable only when there is no
pr 59. [APhO-2014] Let us consider the so-called Hadley phase change of water
circulation: this is a large-scale circulation of air masses, which
rise at a certain geographical latitude (close to the equator) pr 60. [EstFin-2008] A microcalorimeter is a thin circular
from lower atmospheric layers (from near-surface regions) to silicon nitride membrane, thermally isolated from the surround-
higher altitudes. During this rising phase the air cools adiabat- ings, except that it is thermally connected to the wafer by four
ically which leads to a condensation of air vapours, cloud form- thin and narrow thermal bridges (see Figure). The microcalor-
ation and rain. The cool air moves at higher atmospheric layers imeter is equipped with a small heater in the middle of the
in north-south direction (while radiating heat slowly away into membrane and a similar structure on the edge of the membrane
space) until it reaches higher geographical latitudes, where it working as a thermometer. This micro calorimeter is used to
descends to Earth’s surface while heating adiabatically. Finally, study the thermal properties of nanoscale Ti disks (light tiny
the warm air moves along Earth’s surface to its starting point dots in Fig). The thermal power of the heater depends sinusoid-
while being heated by surface (which is hot due to sunlight). ally on time, P = P0 cos(ωt) (negative power implies a with-
8 Actually, during day time, this is not the case: air temperature decreases with height. However, during evening and night, due to heat radiation,
the lower layers of air cool more rapidly than upper layers, and it may easily happen that the temperature is roughly independent of height.

— page 28 —
 4. ENTROPY AND CARNOT CYCLE
drawal of heat). The circular frequency ω is sufficiently low, b) Estimate, how long time does it take for a full evaporization
so that for any moment of time t, the temperature of the mi- of the liquid nitrogen (the vapour escapes through an over pres-
crocalorimeter T (t) can be considered constant across its entire sure valve). For the liquid nitrogen, density ρ = 810 g/l and
surface, and the temperature profile along the thermal bridges latent heat of vaporization λ = 199 kJ/kg). NB! If you were
can be considered linear. The wafer, to which the bridges are unable to find P (for question a), express the evaporization time
connected, is large and thick enough, so that its temperature symbolically (i.e. using the symbol P ).
T0 can be considered to be constant all the time. Each of the
four bridges have length L and cross sectional area of S; the pr 62. [EstFin-2011] Let us study how a vacuum can be cre-
thermal conductance of them is κ. Thermal conductance is ated inside a bulb by pumping. Let the volume of the bulb be
defined as the heat flux (measured in Watts) per surface area, V , and the pump consist in a piston moving inside a cylinder
assuming that the temperature drop is 1 ◦ C per 1 m. The heat of volume αV , where α ≪ 1. The pumping cycles starts with
capacity of the microcalorimeter (with Ti-disks) is C. piston being pulled up; when the pressure inside the cylinder
becomes smaller than inside the bulb, a valve VA (connecting
the cylinder and the bulb) opens and remains open as long as
the piston moves up. When piston is released, it starts moving
down, at that moment, the valve VA closes. As long as the
valve VA is open, the pressures of the bulb and the cylinder
can be considered as equal to each other. When the piston
moves down, the pressure in the cylinder increases adiabatic-
ally until becoming equal to the outside pressure p0 = 105 Pa;
at that moment, another valve VB opens letting the gas out of
the cylinder. When the piston reaches the bottommost posi-
tion, there is no residual air left inside the cylinder. Now, the
piston is ready for being lift up: the valve VB closes and VA
a) Find the thermal resistance R between the microcalorimeter opens, marking the beginning of the next pumping cycle. The
and the wafer (i.e. the ratio of the temperature difference and air inside the bulb can be considered isothermal, with the tem-
heat flux). perature being equal to the surrounding temperature T0 . The
For questions (ii) and (iii), use quantity R, without substitut- adiabatic exponent of air γ = cp /cV = 1.4.
ing it via the answer of question (a). a) How many pumping cycles N needs to be done to reduce
b) Write down the heat balance equation for the microcalor- the pressure inside bulb from p = p0 down to p = βp0 , where
imeter and find the temperature of the microcalorimeter as β ≪ 1?
a function of time T (t) [you may seek it in the form T = b) What is the net mechanical work done during such a pump-
T0 + ∆T sin(ωt + ϕ)]. ing (covering all the N cycles)?
c) In order to study the thermal properties of the Ti-nanodisks, c) What is the temperature of the air released from the cylinder
the amplitude of the sinusoidal oscillations of T (t) should to the surroundings at the end of the pumping process (when
change by as large as possible value, as a response to a small the pressure inside the bulb has become equal to βp0 )?
change of C (which is caused by the Ti-disks). Find the optimal d) According to the above described pumping scheme, there is
circular frequency ω0 . a considerable loss of mechanical work during the period when
d) We have assumed that the temperature profile along the the piston is released and moves down. Such a loss can be
bridges is linear, i.e. their heat capacity can be neglected. For avoided if there is another pump, which moves in an opposite
high frequencies ω & ωc , this is not the case. Estimate the crit- phase: the force due to outside air pressure pushing the piston
ical frequency ωc in terms of κ, l, specific heat c and density ρ down can be transmitted to the other pump for lifting the pis-
of the bridge material. ton up. What is the net mechanical work done when such a
pumping scheme is used?
pr 61. [EstFin-2010] In order to study the thermal prop-
erties of a thermos bottle, let us model it as two concentric pr 63. [EstFin-2011] Consider a heat sink in the form of a
spherical vessels, with radii R1 = 7 cm and R2 = 10 cm. The copper plate of a constant thickness (much smaller than the
gap between the walls of the vessels contains vacuum (hence, diameter d of the plate). An electronic component is fixed to
the heat conductivity can be neglected). the plate, and a temperature sensor is fixed to the plate at
a) Find the radiative heat flux (i.e. transmitted heat per unit some distance from that component. You may assume that
time) between the walls of the bottle, assuming that the ambi- the heat flux (i.e. power per unit area) from the plate to the
ent temperature is T2 = 293 K and the inner sphere is filled surrounding air is proportional to the difference of the plate
with liquid nitrogen at the boiling temperature T1 = 77 K. temperature at the given point (the coefficient of proportional-
The emissivities of all the surfaces are equal to that of stain- ity is constant over the entire plate, including the site of the
less steel: ε = 0.1. Remark: The emitted heat flux per electronic component).
unit area is given Stefan-Boltzmann’s law P = εσT 4 , where a) The electronic component has been dissipating energy with
σ = 5.67 × 10−8 W/m2 K4 (assuming that ε is independent of a constant power of P = 35 W for a long time, and the aver-
the wavelength. age plate temperature has stabilized at the value T0 = 49 ◦ C.
— page 29 —
 4. ENTROPY AND CARNOT CYCLE
C
Now, the component is switched off, and the average plate tem- 3R v
perature starts dropping; it takes τ = 10 s to reach the value
T1 = 48 ◦ C. Determine the heat capacity C (units J/◦ C) of 2R
the plate. The capacities of the electronic component and the
temperature sensor are negligible. R
b) Now, the electronic component has been switched off for
a long time; at the moment t = 0, a certain amount of heat T(K)
Q is dissipated at it during a very short time. In the Figure
0 100 200 300
and Table, the temperature is given as a function of time, as a) What is the total heat energy of such a cube at the initial
recorded by the sensor. Determine the dissipated heat amount temperature T0 ?
Q. b) Now, the cube has 5 faces painted in white (reflects all rel-
evant wavelengths) and one face painted in black (absorbs all
these waves). The cube is surrounded by vacuum at a very low
temperature (near absolute zero); there is no gravity field. Ini-
tially, the cube is at rest; as it cools down due to heat radiation,
it starts slowly moving. Estimate its terminal speed v1 .
c) At very low temperatures, the heat capacitance of aluminium
is proportional to T 3 , where T is its temperature. Which func-
tional dependence f (t) describes the temperature as a function
of time [T = A · f (Bt), where A and B are constants] for such
very low temperatures under the assumptions of the previous
question?
t (s) 0 20 30 100 200 d) Now, the cube has 5 faces covered with a thermal insulation
300
T (◦ C) 20.0 20.0 20.4 32.9 41.6 layer (you may neglect heat transfer through these faces). One
42.2
t (s) 400 600 800 1000 1200 face is left uncovered. The cube is surrounded by hydrogen at-
1400
T (◦ C) 39.9 33.4 27.9 24.4 22.3 mosphere at a very low temperature (molar mass of hydrogen
21.2
molecules MH = 2 g/mol). The cube starts cooling down due
to heat transfer to the surrounding gas; you may neglect the
pr 64. [Seagull-2016] What would be the temperature in the heat radiation. Initially, the cube is at rest; as it cools down,
middle of a celestial body of the size of our Moon in Kelvins it starts slowly moving. Estimate the order of magnitude of its
if the following assumptions can be made. The celestial body terminal speed v2 . Assume that the surrounding gas is sparse,
consists of an homogeneous solid material of average density so that the mean free path of the molecules is much larger than
3 and heat conductivity k = 3 W/m · K, and has the
g
ρ = 3 cm a. Assume that v2 ≪ cs where cs is the speed sound in the
shape of a sphere of radius R = 2000 km. It radiates heat atmosphere surrounding the cube.
as a perfectly black body, i.e. the heat flux density at its sur-
face w = σT 4 , where σ = 5.67 × 10−8 J/s · m2 · K4 and T is pr 67. [PhysCup] As compared to ordinary light bulbs, light
its surface temperature. There is no external heat flux fall- emitting diodes (LED) provide very high lighting efficiency.
ing onto the surface of this celestial body. Due to nuclear The reason is that the spectral energy distribution of ordin-
decay of various isotopes (mostly uranium-238 and its decay ary lamps is close to black body radiation, in which case one
products), the power density released in the crust material can say that the photons are in thermal equilibrium with the
P = 7 × 10−12 W/kg. black body. Then, the total energy radiated by a black body
per unit area, unit time, and unit frequency interval is given
pr 65. [EstFin-2011] Calculate the thermal efficiency of by Planck’s law
an ideal-gas cycle consisting of two isotherms at temperatures 2πh ν3
I = 2 hν/kT ,
T1 and T2, and two isochores joining them. (An isochore is c e −1
−34
where ν is the frequency, h = 6.626 × 10 J · s — the Planck
a constant-volume process.) The engine is constructed so that
constant, c = 2.997 × 108 m/s — the speed of light, k =
the heat released during the cooling isochore is used for feeding
1.38 × 10−23 J · K−1 — the Boltzmann constant, and T — the
the heating isochore
temperature; note that ∫
∞
Idν = σT 4 ,
pr 66. [EstFin-2014] Consider a cube of side length a = 0
1 cm, made of aluminium (density ρ = 2.7 g/cm3 , molar mass where σ = 5.678 × 10−8 W · m−2 · K−4 is the Stefan-Boltzmann
MA = 23 g/mol). The heat capacitance of one mole of alu- constant. With a black body radiation, a lot of energy is wasted
minium is given as a function of temperature in the graph be- by radiating non-visible light. Meanwhile, LED-s can be con-
low. The speed of light c = 3 × 108 m/s, universal gas constant structed so that they radiate almost only visible light.
R = 8.31 J/(kg · K). The initial temperature of the cube is In recent experiments 9 , it has been reported that such LED-s
T0 = 300 K. have been constructed which have efficiency higher that 100%.
9 P. Santhanam et al, Thermoelectrically Pumped Light-Emitting Diodes Operating above Unity Efficiency, Phys. Rev. Lett. 108, 097403 (2012)

— page 30 —
 4. ENTROPY AND CARNOT CYCLE
Here the efficiency is defined as the ratio of the radiated light lower, and assume that this slab expands adiabatically so that
energy to the consumed electrical energy. its temperature drops to the temperature of the surrounding
air.
Based on reasonable approximations, find what is the theoret-
a) How is the fractional change in temperature dT /T related
ically highest possible efficiency of a LED assuming that:
to dp/p, the fractional change in pressure?
(a) the LED has a heat sink which is kept at the room temper-
b) Express the pressure difference dp in terms of dz, the change
ature T0 = 293 K (via a fast enough heat exchange with the
in height.
surrounding medium);
c) What is the resulting temperature at the top of the building?
(b) the LED emits light at wavelengths smaller than λ0 =
Data: Boltzmann constant: kB = 1.38 × 1023 J · K−1 , Mass of
700 nm
a nitrogen molecule: m = 4.65 × 10−26 kg, Gravitational accel-
(b) the surface area of the light-emitting part of the LED is
eration: g = 9.80 m · s−2 .
S = 1 mm2 ;
(c) the light emission power of the LED is P = 1 µW.

pr 68. [EstOpen-2010] A liquid rocket engine is schematic-
ally shown in the figure below. Burning products are formed
inside the combustion chamber at a high pressure and tem-
perature, and expand adiabatically while leaving the chamber
through the nozzle. If the nozzle is correctly constructed (the
neck diameter needs to correspond to the burning rate), ex-
haust gases continue adiabatic expansion after passing through
the throat, up to the exit from the nozzle. As a result, a major
part of the heat energy is converted into kinetic energy of the
exhaust gases at the exit of the nozzle. The temperature of the
exhaust gases inside the combustion chamber is T0 ; at the exit,
the cross-sectional area of the nozzle is S, temperature of the
exhaust gases is T1 , and pressure is p1 . Find the thrust force
F assuming that T0 ≫ T1 , the effect of atmospheric pressure
on the thrust can be neglected, and the kinetic energy of the
gases inside the combustion chamber is negligible (as compared
to the thermal energy). The heat capacitance of one mole of
the exhaust gases by constant pressure is cV = 52 R, where R is
the universal gas constant.
pr 71. [IPhO-2006] An egg, taken directly from the fridge
at temperatureT0 = 4◦ C, is dropped into a pot with water that
is kept boiling at temperature T1 .
a) How large is the amount of energy U that is needed to get
the egg coagulated?
b) How large is the heat flow J that is flowing into the egg?
c) How large is the heat power P transferred to the egg?
d) For how long do you need to cook the egg so that it is
hard-boiled?
Hint: You may use the simplified form of Fouriers Law J =
κ∆T /∆r, where ∆T is the temperature difference associated
with ∆r, the typical length scale of the problem. The heat flow
J is in units of W · m−2 .
Data: Mass density of the egg: µ = 10 × 103 kg · m−3 , specific
heat capacity of the egg: C = 4.2 J · K−1 g−1 , radius of the egg:
R = 2.5 cm, coagulation temperature of albumen (egg protein):
Tc = 65 ◦ C, heat transport coefficient: κ = 0.64 W · K−1 m−1
(assumed to be the same for liquid and solid albumen)

pr 72. [EstPhO-2000] A tightly closed vessel with thick and

solid walls is filled with glycerin. Inside the glycerin, there is a
bubble of air of volume w = 1 ml. At temperature T0 = 20 ◦ C,
the pressure in the vessel is p0 = 1 atm. The total volume of
the vessel V = 1 l linear thermal expansion coefficient of the
vessel is negligibly small. The volumetric thermal expansion
pr 69. During a day time, air masses in atmosphere are usu- coefficient of glycerin α = 5.1 × 10−4 K−1. Express the pres-
ally in a turbulent (random) motion, up and down, which leads sure inside the vessel as a function of temperature; at which
to what is referred to as adiabatic atmosphere: when a parcel of temperatures and why your expression becomes invalid?
air (a fictitious volume) is raising up, it is expanding and cool-
ing adiabatically; in adiabatic atmosphere, the surrounding air
is in mechanical and thermodynamical equilibrium with the
raising air parcel (has exactly the same pressure and temper- pr 73. [WoPhO-2012] Tornado is a violent vortex (rotating
ature). Find, how depends air temperature T on height h in column) of air connecting the base of cumulonimbus10 cloud
adiabatic atmosphere. At the ground level (h = 0), T = T0 ; ex- and the ground. A distinct feature of the tornado is its funnel-
press your answer in terms of the adiabatic index of air γ, molar like core or condensation funnel (Region II) which is made of
mass µ, universal gas constant R, and free fall acceleration g. small water droplets that condense as they are sucked into the
core as shown in Fig. 1(b). This region is defined by the core
pr 70. [IPhO-2006] At the bottom of a 1000 m high sky- radius rC (z) which generally increases with altitude forming
scraper, the outside temperature is Tbot = 30 ◦ C. The objective the signature funnel-shape of the tornado. Region I is the re-
is to estimate the outside temperature Ttop at the top. Con- gion outside tornado core. In our simplified model, region I and
sider a thin slab of air (ideal nitrogen gas with adiabatic coef- II have different velocity distribution profile as we will explore
ficient γ = 7/5) rising slowly to height z where the pressure is later.
10 Cumulonimbus cloud is a towering vertical cloud that is very tall, dense, and involved in thunderstorms and other rainy weather.

— page 31 —
 4.
In what follows, you may use the following data:
gravitational acceleration g = 9.8 m/s2 ,
air density ρair = 1.2 kg/m3 ,
molar mass of dry air Mair = 0.029 kg/mol,
universal gas constant R = 8.314 J/(mol · K),
pressure at point A (see figure) P0 = 1.0 × 105 Pa,
temperature at point A (see figure) T0 = 15 ◦ C,
heat capacity ratio of air Cp/Cv = 1.4.
A) The calm weather. We will investigate the atmospheric
pressure of the troposphere (the lowest part of the atmosphere)
where most of the weather phenomena including tornado oc-
curs. Let us start from a calm weather location at point A far
away from the tornado.
i) Assuming ideal gas law, constant gravity acceleration and a
constant temperature T0 . Show that the atmospheric pressure
as a function of altitude z is P (z) = P0 e−αz ; express α in terms
of the constants listed above.
ii) For simplicity we neglect the dependence of air density, ρair
on z. Derive the pressure as a function of altitude, P (z).
iii) Using your result in (ii) calculate the pressure at point B
on the base of the cumulonimbus! (use h = 1 km)
B)The shape. Inside the tornado’s core the water vapour con-
denses into liquid droplets as the air spirals into the core form-
ing condensation funnel. The water vapour condenses when the
temperature drops below dew point. The temperature drop is
caused by a pressure drop so the points where the water vapour
starts to condense lay on a surface of equal pressure, called
isobar boundary layer shown as red boundary in figure. This
is the boundary between region I and II. For now we only con-
sider region I. Consider a point G (see figure) very close to the
ground (z ≈ 0) located at radial distance r = rG from the axis
(point C). At that point, the wind speed vG can be treated
as the ground rotation speed of the tornado. We further make
the following assumptions: the tornado is stationary (only has
rotation and no translation); the radial velocity of the wind is
much smaller than the tangential speed v = v(r) which will be
assumed to be independent of altitude z (depends only on r);
turbulence very close to the ground can be ignored; air mass
— page 32 —
ENTROPY AND CARNOT CYCLE
density (ρair ) can be assumed to be constant.
i) Show that in both region I and II,
dP v2
= ρair .
dr r
ii) In region I, calculate the wind velocity v as a function of r,
in terms of vG and rG .
iii) Estimate the air speed vG .
iv) Derive the shape of the condensation funnel, i.e. find the
function rC (z), where rC denotes the radial distance of a point
C at the boundary layer (see figure). Sketch this tornado shape
in dimensionless quantities z/h vs. r/rG , where h is the height
defined in figure.
v) Most tornadoes look like funnel (the radius is larger at higher
altitude) while some are more uniform in diameter, like a pipe.
Which type has the higher ground rotation speed vG ?
C) The core. Now we will consider both region I and II.
i) In region II (r < rC ) the tornado core behaves as rigid body,
derive expression for the (tangential) speed v(r) in this region.
ii) Calculate the pressure at the center of the tornado (point
D, at the same altitude as point G).
iii) Estimate the temperature TD at the center of the tornado.
iv) Based on your finding in (iii) suggest, what could be a pos-
sible source of tornado’s tremendous energy.
D) Windows. The differential pressure near a tornado is
thought to cause poorly ventilated houses to “explode” even
though the tornado is only passing at a distance. Therefore
some people suggest that the windows have to be opened to let
the pressure in the house equilibrates with outside. Then, how-
ever, wind and debris and can freely enter the house. Consider
a house with all windows and openings closed with a at roof of
dimension (width × length × thickness) 15 m × 15 m × 0.1 m
and mass density ρroof = 800 kg/m3 . The tornado is coming
fast and passing at a distance d = 2rG away from the house.
i) What is the lift-force-to-weight ratio for the roof?
ii) Shall you open or close the windows?
pr 74. [IPhO-2004] A rubber balloon filled with helium
gas goes up high into the sky where the pressure and tem-
perature decrease with height. In the following questions, as-
sume that the shape of the balloon remains spherical regard-
less of the payload, and neglect the payload volume. Also
assume that the temperature of the helium gas inside of the
balloon is always the same as that of the ambient air, and
treat all gases as ideal gases. The universal gas constant is
R = 8.31 J/(mol · K) and the molar masses of helium and air
are MH = 4.00 × 10−3 kg/mol and MA = 28.9 × 10−3 kg/mol,
respectively. The gravitational acceleration is g = 9.8 m/s2 .
A-i) Let the pressure of the ambient air be P and the temper-
ature be T . The pressure inside of the balloon is higher than
that of outside due to the surface tension of the balloon. The
balloon contains n moles of helium gas and the pressure inside
is P + ∆P . Find the buoyant force FB acting on the balloon
as a function of P and ∆P .
ii) On a particular summer day in Korea, the air temperat-
ure T at the height z from the sea level was found to be
T (z) = T0 (1 − z/z0 ) in the range of 0 < z < 15 km with
z0 = 49 km and T0 = 303 K. The pressure and density at the
sea level were P0 = 1 atm = 1.01 × 105 Pa and ρ0 = 1.16 kg/m3 ,
 4. ENTROPY AND CARNOT CYCLE
respectively. For this height range, the pressure takes the form
1,6
P (z) = P0 (1 − z/z0 )η . Express η in terms of z0 , ρ0 , P0 , and
g, and find its numerical value to the two significant digits.
Treat the gravitational acceleration as a constant, independent 1,2
of height.
B) When a rubber balloon of spherical shape with un-stretched 0,8
radius 0r is inflated to a sphere of radius r0 (r ≥ r0 ), the balloon
surface contains extra elastic energy due to the stretching. In 0,4
a simplistic theory, the elastic energy at constant temperature
T can be expressed by
U = 4πr02 κRT (2λ2 + λ−4 − 3, 0 2 4 6 8 10

whereλ ≡ r/r0 (≥ 1) is the size-inflation ratio and κ is a con- In many applications it is necessary to keep the satellite as
stant in units of mol/m2 . cool as possible. To cool the satellite, engineers use a reflect-
i) Express ∆P in terms of parameters given in Eq. above, and ive coating that reflects light above a cut-off frequency but
sketch ∆P as a function of λ = r/r0 . does not prevent heat radiation at lower frequency from escap-
ii) The constant κ can be determined from the amount of ing. Assume that this (sharp) cut-off frequency corresponds to
the gas needed to inflate the balloon. At T0 = 303 K and hf /k = 1200 K.
P0 = 1 atm, an un-stretched balloon (λ = 1) contains n0 = 12.5 Estimate now, what is the temperature of the satellite.
moles. It takes n = 3.6n0 = 45 moles in total to inflate the
Hint: exact calculation is not required; because of that, in-
balloon to λ = 1.5 at the same T0 and P0 . Express the bal-
stead of precise and complicated integrations use approximate
loon parameter a, defined as a = κ/κ0 , in terms of n, n0 , and
calculations where possible. The value of the following integral
λ, where κ0 = r0 P0 /4RT0 . Evaluate a to the two significant
digits.
is known: ∫
η 3 dη π4
∞ η = .
C) A balloon is prepared as in (C-ii) at the sea level (inflated 0 e −1 15
Function η /(e − 1) maximum is near η ≈ 2.82. For small
3 η
to λ = 1.5 with n = 3,6 = 45 moles of helium gas at T0 = 303 K
values of η, approximation eη ≈ 1 + η can be used.
and P0 = 1 atm). The total mass including gas, balloon itself,
and other payloads is MT = 1.12 kg. Now let the balloon rise
from the sea level. pr 76. [IPhO-200011 ] Ideal heat engine receives heat from
i) Suppose that the balloon eventually stops at the height zf body A and gives it away to body B; both bodies have equal
where the buoyant force balances the total weight. Find zf and heat capacitance C. Initial temperature of body A was T1 and
the inflation ratio λf at that height. Give the answers in two that of body B — T2 . What is the maximal amount of work
significant digits. Assume there are no drift effect and no gas which can be done by such a heat engine?
leakage during the upward flight.
pr 77. Both solder joints of a thermocouple exchange heat
with the surrounding air at the rate Pe = κ∆T , where ∆T is the
pr 75. [IPhO-1992] This is a continuation of the prob-
difference between the temperature of a solder joint and that of
lem 6; here we repeat the numerical data: the satellite is a
the air; the constant κ = 30 µW/K. The thermocouple sensit-
sphere of radius 1 m all points of which have the same tem-
ivity α is defined as the proportionality coefficient between the
perature. The satellite is near Earth, but is not shaded by
thermocouple voltage and the temperature difference between
it; The surface temperature of the Sun T⊙ = 6000 K, and its
the solder joints; in this case, α = 50 µV/K. The thermocouple
radius R⊙ = 6.96 × 108 m, The distance between the Earth
is fed with electric current I = 10 mA; find the temperature of
and the Sun L = 1.5 × 1011 m Stefan-Boltzmann constant
the colder solder joint if the air temperature ta = 20◦ C. As-
σ = 5.67 × 10−8 W · m−2 · K−4 .
sume the energy transfer processes in the thermocouple to be
The blackbody radiation spectrum u(T, f ) of a body at tem- reversible.
perature T obeys Planck’s radiation law
η 3 dη pr 78. [EstPhO-2010] A disc of radius r = 1 cm is made
u(T, f )df = A η
e −1 from dry ice (carbon dioxide in solid phase), and is pressed
where udf is the energy density of the electromagnetic radi- against a plate which is made from a material of high heat
ation in a frequency interval (f, f + df ), and η = hf /kT ; con- conductivity. The disc is pressed against the plate with force
stant A can be expressed in terms of fundamental constants c F = 100 N, which is applied to the centre of the disc and is
(the speed of light), k (Boltzmann’s constant), and h (Planck’s almost (but not exactly) perpendicular to the plate. When the
4
constant): A = 8πk
c3 h3 . The blackbody spectrum, integrated over plate temperature is low, the friction force between the disk
all frequencies f and directions of emission, gives the total radi- and plate keeps the disk at rest. However, when the plate tem-
∫
ated power per unit area P = u(T, f )df = σT 4 as expressed perature rises above a critical value T0 , the friction force disap-
in the Stefan-Boltzmann law given above. The figure shows pears almost completely, and the disc starts sliding along the
the normalized spectrum u(T, f )/AT 4 as a function of η. plate. Find T0 . Air pressure is 101 kPa, triple point pressure
11 Voted out by the International Board from the final text

— page 33 —
 4. ENTROPY AND CARNOT CYCLE
of carbon dioxide is pt = 5.1 × 105 kPa, saturation pressure of
carbon dioxide is plotted against temperature in figure below.
pr 79. [EstPhO-2009] Room is being heated by an electrical
radiator of power P = 1 kW. Outside temperature is t0 = 0 ◦ C,
the room temperature has been stabilized at t1 = 20 ◦ C. A
ventilation window is opened, supplying now a fresh air at the
volume rate v = 20 l/s. What will be the new room temperat-
ure? Air can be considered to be an ideal bimolecular gas.
pr 80. [IPhO-1997] On a given day, the air is dry and
has a density ρ = 1.250 kg/m3 . The next day the humid-
ity has increased and the air is 2% by mass water vapour.
The pressure and temperature are the same as the day before.
What is the air density ρ′ now? Mean molar mass of dry air:
Ma = 28.8 g/mol; molar mass of water: Mw = 18 g/mol. As-
sume ideal-gas behaviour.
pr 81. [IPhO-2010] Chimney. Gaseous products of burn-
ing are released into the atmosphere of temperature Tair
through a high chimney of cross-section A and height h (see
figure a). The solid matter is burned in the furnace which is
at temperature Tsmoke . The volume of gases produced per unit
time in the furnace is B. Assume that:
⋄ the velocity of the gases in the furnace is negligibly small;
⋄ the average molar mass of the gases produced in the fur- height of 195 m, and a radius 5 m. The collector is circular
nace (smoke) equals to that of the air, and these gases can with diameter of 244 m. The specific heat of the air under
be treated as ideal;
⋄ chimney is not very high (not higher than a hundred
meters);
⋄ smoke temperature can be assumed to be constant through- is typically 150 W/m during a sunny day.
out the entire length of the chimney.
— page 34 —
a) What is the minimal height of the chimney needed for an
efficient withdrawal of the gases? Express answer in terms of
B, A, Tair , g = 9.81 m/s2 , and ∆T ≡ Tsmoke − Tair . In all
subsequent tasks assume that this minimal height is the height
of the chimney.
b) Assume that two chimneys are built to serve exactly the
same purpose. Their cross sections are identical, but are de-
signed to work in different parts of the world: one in cold re-
gions, designed to work at an average atmospheric temperature
of −30 ◦ C and the other in warm regions, designed to work at
an average atmospheric temperature of 30 ◦ C. The temperat-
ure of the furnace is 400 ◦ C. It was calculated that the height
of the chimney designed to work in cold regions is 100 m. How
high is the other chimney?
c) How does the velocity of the gases vary along the height of
the chimney? Make a sketch/graph assuming that the chimney
cross-section does not change along the height.
d) How does the pressure of the gases vary along the height of
the chimney?
Solar power plant. The chimney operation principle can be
used to construct a particular kind of solar power plant, solar
chimney. The idea is illustrated in figure b. The Sun heats the
air underneath the collector of area S with an open periphery
to allow the undisturbed inflow of air. As the heated air rises
through the chimney (red arrows), new cold air enters the col-
lector from its surrounding (blue arrows) enabling a continuous
flow of air through the power plant. The flow of air through
the chimney powers a turbine, resulting in the production of
electrical energy. The energy of solar radiation per unit time
per unit of horizontal area of the collector is G. Assume that
all that energy can be used to heat the air in the collector (the
mass heat capacity of the air is c, and one can neglect its de-
pendence on the air temperature). We define the efficiency of
the solar chimney as the ratio of the kinetic energy of the gas
flow and the solar energy absorbed in heating of the air prior
to its entry into the chimney.
e) What is the efficiency of the solar chimney power plant?
f) Make a graph showing how the efficiency of the chimney
changes with its height.
The prototype chimney built in Manzanares, Spain, had a
typical operational conditions of the prototype solar chimney
is 1012 J/kg · K, the density of the hot air is about 0.9 kg/m3 ,
and the typical temperature of the atmosphere Tair = 295 K.
In Manzanares, the solar power per unit of horizontal surface
2
g) What is the efficiency of the prototype power plant? Write
 4. ENTROPY AND CARNOT CYCLE
down the numerical estimate. b) The mean free path λ is the average distance covered by a
h) How much power could be produced in the prototype power moving particle in a gas between two consecutive collisions and
plant? this can be expressed by the following equality:
i) How much energy could the power plant produce during a λ = (σn)−1 ,
typical sunny day? where σ is the effective cross sectional area; σ = 0.2 nm2 for
j) How large is the rise in the air temperature as it enters the the collisions of an hydrogen atom with the air molecules (with
chimney (warm air) from the surrounding (cold air)? Write the molecular composition corresponding to exobase), and n is the
general formula and evaluate it for the prototype chimney. number of molecules per unit volume. We know that at an
k) What is the mass flow rate of air through the system? altitude of h0 = 250 km, the pressure is p0 = 21 µPa, and tem-
perature can be assumed to be constant between the exobase
pr 82. Air (a mixture of oxygen and nitrogen) is stored and the reference height h0 . Determine the mean free path for
in a closed container equipped with a piston on one end at the hydrogen atoms at the altitude h0 .
a temperature of T = 77.4 K. The total amount of the gas c) Determine the exobase height hEB .
mixture occupies volume V0 = 1.00 l and its initial pressure is
d) Atmospheric escape. Particles in the exobase with enough
p0 = 0.500 atm. With the help of the piston the gas mixture
outwards velocity will escape gravitational attraction. Assum-
is slowly compressed at constant temperature. Using plausible
ing a Maxwellian distribution, determine the probability that a
assumptions, plot the pressure of the system as a function of
hydrogen atom has a velocity greater than the escape velocity
its volume until one tenth of the initial volume, if the ratio
in the exobase.
of the number of moles of oxygen to the number of moles of
nitrogen is nO2 : nN2 = 21 : 79. Find the pressure and volume e) Determine the hydrogen atoms flux (number of particle per
at distinctive points of these isothermal curves. unit area and per unit time) that will escape the atmosphere,
You can use the following data: boiling point of liquid nitro- knowing that the concentration of hydrogen atoms in the exo-
gen at p1 = 1.00 atm: TN2 = 77.4 K; boiling point of liquid base is nH = 1 × 1011 m−3 . You may assume that the hydrogen
oxygen at p1 = 1.00 atm: TO2 = 90.2 K; heat of vaporization is in a monoatomic state: when atmospheric molecules reach
of oxygen: λO2 = 213 J/g (can be assumed to be constant for a certain height lower than hEB , they split into atoms due to
TN2 ≤ T ≤ TO2 ). Neglect solubility of gaseous nitrogen in solar radiation.
liquid oxygen and vice versa. The following calculations will show hat thermal atmospheric
escape cannot be the single processes that explains why some
pr 83. [WoPhOSR-2013] Thermal atmospheric escape is a gases are present in the atmosphere and some others are not.
process in which small gas molecules reach speeds high enough f) Knowing that the current average molar mass of the at-
to escape the gravitational field of the Earth and reach outer mosphere is MAtm = 29 gr/mol, atmospheric pressure P0 =
space. This process, known as Jeans escape, is believed to have 1 × 105 Pa, and a fraction of χH = SI5.5e − 7 of the atmo-
been responsible for the loss of water from Venus and Mars at- spheric molecules are hydrogen molecules, estimate the number
mospheres, due to their lower escape velocity. NH of hydrogen atoms in the Earth’s atmosphere. Neglect the
presence of other hydrogen-containing molecules.
In what follows we assume that the distribution of the mo-
lecules of the atmosphere is given by the Maxwellian distribu- g) Let us make a robust model and assume that the concentra-
tion. Thus at any temperature there can always be some mo- tion of hydrogen atoms in the exobase remains constant over
lecules whose velocity is greater than the escape velocity. A mo- time. Find out how much time would it take for half of the
lecule located in the lower part of the atmosphere would not be hydrogen atoms to escape the Earth’s atmosphere.
able to escape to outer space even though its velocity is greater h) Now let us consider helium atoms; currently, their concen-
than the limit velocity because it would soon collide with other tration in the exobase is nHe = 2.5 × 1012 m−3 , and they make
molecules, losing a big part of its energy. In order to escape, χHe = 5 × 10−6 of the atmospheric molecules. Let us use now
these molecules need to be at a certain height: such that density a more realistic model and assume that their concentration in
is so low that their probability of colliding is negligible. The re- exobase is proportional to their concentration in the lower at-
gion in the atmosphere where this condition is satisfied is called mosphere.
exosphere and its lower boundary, which separates the dense
i) Suggest a reason, why there is currently still some hydrogen
zone from the exosphere, is called exobase. You may assume
in the Earth’s atmosphere.
that the temperature near the exobase is roughly T = 1000 K;
universal gas constant R = 8.31 J/K · mol, free fall acceleration
pr 84. [IPhO-2012] Let us model the formation of a star as
g ≈ 9.5 m/s2 , Avogadro number NA = 6.02 × 1023 1/mol.
follows. A spherical cloud of sparse interstellar gas, initially at
a) Exobase height. Exobase is defined as the height above which rest, starts to collapse due to its own gravity. The initial radius
a radially outward moving particle will suffer less than one of the ball is r0 and the mass — m. The temperature of the
backscattering collision on average. This means that the mean surroundings (much sparser than the gas) and the initial tem-
free path has to be equal to the scale height H, which is defined perature of the gas is uniformly T0 . The gas may be assumed
as such an height increment which leads to the atmosphere’s to be ideal. The average molar mass of the gas is µ and its
density dropping e times. What is the average molar mass of adiabatic index is γ> 34 . Assume that G mµ r0 ≫ RT0 , where R
the Earth’s atmosphere at the exobase if H = 60 km? is the gas constant and G — the gravity constant.
— page 35 —
 4. ENTROPY AND CARNOT CYCLE
a) During much of the collapse, the gas is so transparent that shrinking speed Ṙ as
any heat generated is immediately radiated away, i.e. the ball Ek = 2πρ0 R3 Ṙ2 .
stays in a thermodynamic equilibrium with its surroundings.
a) Find the minimal radius Rm and maximal temperature Tm
How many times (n) does the pressure increase while the ra-
of the bubble during the shrinking process.
dius is halved (r1 = 0.5r0 )? Assume that the gas density stays
uniform. b) Find the radius Rf of the bubble at the moment tf when
the bubble’s shrinking speed is maximal.
b) At some radius r3 ≪ r0 , the gas becomes dense enough to
be opaque to the heat radiation. Calculate the amount of heat c) Most of the heat radiation is emitted between t = tf and
Q radiated away during the collapse from the radius r0 down the moment tm when the minimal radius is reached. Estimate
to r3 . Assume that the gas density stays uniform12 the total duration of the radiation period.

c) For radii smaller than r3 you may neglect heat radiation. d) For our model to remain valid, the heat loss Q due to radi-
Determine how the temperature T of the ball depends on its ation must be much smaller than the internal energy of the gas
radius r< r3 . U ; let us require that Q ≤ 15 U . This means that if we consider
the entrapped gas as a sphere of radius R emitting thermal ra-
d) Eventually we cannot neglect the effect of the pressure on
diation at emissivity a, the emissivity should be not too large,
the dynamics of the gas and the collapse stops at r = r4 (with
a ≤ a0 . Estimate the value of a0 . Stefan-Boltzmann constant
r4 ≪ r3 ). However, the radiation can still be neglected and
σ = 5.678 × 10−8 W · m−2 · K−4 .
the temperature is not yet high enough to ignite nuclear fu-
sion. The pressure of such a protostar is not uniform anymore, pr 86. [APhO-2007] a) For a metal at thermal equilibrium,
but rough estimates with inaccurate numerical prefactors can the average energy of the conduction electrons makes a certain
still be done. Estimate the final radius r4 and the respective contribution to the specific heat. According to the classical
temperature T4 . physics the conduction electrons in metals constitute free elec-
tron gas trapped into the volume of the metal, and this gas
pr 85. [APhO-2010]13 Sonoluminescence is a phenomenon can be considered to be ideal. Find the average contribution
when strong sound waves put small bubbles inside a liquid (e.g. of each electron to the specific heat at constant volume (cV )
water) into oscillatory motion; collapsing bubbles may heat the using the classical model.
entrapped gas so much that it will emit black body radiation
b) Experimentally it has been shown that the specific heat
in visible range of wavelengths.
of the conduction electrons at constant volume in metals de-
Throughout the problem, you may use the following model. pends on temperature, and the experimental value at room
All times t, the bubble of radius R = R(t) is spherical and temperature is about two orders of magnitude lower than its
its centre remains stationary in the water of density ρ0 = classical counterpart. This is because the electrons obey the
1.0 × 103 kg/m3 . The entrapped gas density ρ pressure P , quantum statistics rather than classical statistics. According
and temperature T are always uniform inside the bubble as to the quantum theory, for a metallic material the density of
its size diminishes. The liquid containing the bubble is as- states of conduction electrons (the number of electronic states
sumed to be isotropic, nonviscous, incompressible, and very per unit volume and per unit energy) is proportional to the
much larger in extent than the bubble. This means that we square root of electron energy E, then the number of states
can consider a fictitious water sphere of radius r ≫ R centred dE within energy range for a metal of volume V can be writ-
around the bubble, and it will contract isotropically during ten as √
the contraction of the bubble. Heat exchange between the dS = CV EdE,
gas and the surrounding water can be neglected. All effects where C is the normalization constant, determined by the total
due to gravity and surface tension are neglected so that for number of electrons of the system. The probability that the
any t, the pressure inside the liquid is a function of r only: state of energy E is occupied by electron is called the Fermi
P = P (r) for r > R. You may assume that the ambient pres- distribution,
sure P0 = P (r → ∞) = 1.01 × 105 Pa and the water temperat- [ ( )]−1
E − EF
ure T0 = 300 K remain constant in time. Initially, the bubble f (E) = 1 + exp ,
kB T
has radius Ri = 7R0 , entrapped gas temperature T = T0 , where kB = 1.381 × 10−23 J/K is the Boltzmann constant
and the surrounding water is motionless (i.e. dR dt = 0); here and T is the absolute temperature, while EF is called Fermi
R0 = 5.00 µm denotes the equilibrium radius of the bubble at level. Usually at room temperature EF is about several eVs
which the entrapped gas would have pressure P0 with T = T0 . for metallic materials (1 eV = 1.602 × 10−19 J) and therefore
The ratio of specific heat at constant pressure to that at con- kB T ≪ EF , in which case the Fermi distribution behaves as
stant volume for the gas is γ = 5/3. shown in figure by red curve; you are allowed to approximate
You may find it useful to know that the kinetic energy of the it by a piece-wise linear function as depicted by blue line. Ad-
surrounding water can be expressed in terms of the bubble’s ditionally, you may neglect the dependence of Fermi level on
12 This is valid as long as the gravitational energy dominates over the heat energy; in the original IPhO problem, the students needed to conclude
this by themselves, based on the dynamics of the gas cloud. It should be emphasized that such a model with a ballistic contraction stage deviates from
the standard model of protostar collapse in which case a gas cloud contracts due to radiative cooling while maintaining a quasi-equilibrium between
gravitational and pressure forces.
13 Subtasks are cut from this version: you have a freedom of bypassing unnecessarily long mathematical calculations, but you need to demonstrate

more independence.

— page 36 —
 4. ENTROPY AND CARNOT CYCLE
temperature so that its value can be expressed in terms of the Hints
total number of electrons by putting T = 0.
Express the average contribution of each electron to the spe-
cific heat at constant volume in terms of kB , EF , and T using
the quantum mechanical model. Give a qualitative explanation
for the deviation of the classical result from that of quantum
theory.
pr 87. [EstFin-2014]14 Let us consider a system of N inde-
pendent magnetic dipoles (spins) in a magnetic field B and tem-
perature T . Each spin has two states: spin up, and spin down;
the x-component of its dipole moment takes values +µB ≡ 2m e~
and −µB , respectively (x is a fixed axis).
a) What is the average value of the total energy Es of the spin
system as a function of B and T ?
b) At the high temperature limit T ≫ αBm
k , what is the heat
capacity C of such a spin system?
pr 88. [EstOpen-2001] In order to store ν = 1 mol of ideal 5. Follow the idea 3; the thermal resistance of the copper
gas at temperatur T0 , a vessel made of steel (density ρ) is
used. Assume that the vessel’s walls are much thinner than the
vessel’s linear dimensions. According to the specification, the 22. Recall the idea 7: for dynamical processes, at first, a mech-
mechanical stress in the walls (force per cross-sectional area)
must be everywhere (and for any direction) not larger than σ0 .
What is the minimal mass of the vessel?
appendix 2: Finding mean kinetic energy from
Maxwell’s law Using substitution rule,
∫ ∞ −v2 /v2 ∫ ∞ −x2
−∞
e x 0 dvx = v0 −∞ e dx.
Integration is essentially taking a sum, and we know that a
derivative of a sum is the sum of the derivatives. Hence, we
can take derivative from the both sides of this equality by v0 ;
left-hand-side yields
d
∫ ∞ −v2 /v2 ∫ ∞ d −v2 /v2
dv0 −∞ e
0 dv
x
x = −∞ dv0 e x 0 dvx =
∫∞
2v0−3 −∞ vx2 e−vx /v0 dvx ,
2 2
1. Express the constant heating power as P = dQ dt and sub-
stitute dQ = Cp dT ; this allows you to find Cp as a func-
tion of time t; eliminate t using the provided dependence
T = T (t).
2. Some hints are already given after the problem; determ-
ine the net heating power at the average ice-heating-
temperature T2 ≈ 72 ◦ C by comparing the tangents of the
T (t)-curve at T = T1 (c.f. the fact 6) and T = T2 .
3. In average, electrical heating power equals to the cooling
power due to heat loss. The temperature oscillations are
small, hence (according to the idea 2) the cooling power
is almost constant; meanwhile, heating power oscillates
between a maximal value U12 /R and zero according to a
rectangular waveform. Here, the filament’s resistance R
can be expressed in terms of the filament’s length l and
cross-sectional area S (the value of the latter is not given
but we may hope that it will cancel out from the final an-
swer). Now we can express the thermal power drawn to
(from) the filament for both of the half-periods, and find
the temperature oscillations amplitude from the 1LTD.
4. Based on the fact 6, determine how many percents has in-
creased the thermal flux to the environment, and use this
to conclude, how many percents has increased the heat
production rate inside the house.
plate is calculated in the same way as an electrical resist-
ance, RT = ρd/s.
anical equilibrium is reached, which means the equality
of pressures; the other equilibria (e.g. thermal) will be
reached later (if ever within a reasonable time frame). In
particular, this means that if there is evaporation from
a water surface, and because of that, close to the water
surface, there is an higher concentration of water vapours,
then there must be a lower concentration of air molecules.
Indeed, while due to mechanical equilibrium, the total
pressure must remain equal to the atmospheric one, it also
equals to the sum of the vapour pressure and air pressure
13. Equivalently can be said that the air pressure equals
to the atmospheric pressure minus the vapour pressure.
If the saturation pressure ps (T ) becomes larger than the

atmospheric pressure patm then mechanical equilibrium is

and the right-hand-side yields
∫ ∞ −x2 ∫∞ no longer possible: as we learned earlier, very close to
dx = −∞ e−x dx.
2
d
dv0 v0 −∞ e the water surface, there is thermal quasi-equilibrium and
This means that r = 100% and hence, in that layer, the vapour pressure
∫ ∞ 2 −v2 /v2 ∫ ∞ −x2
v e x 0 dvx = 2 v0 −∞ e 1 3
dx. pv = ps (T ). The total pressure p in that layer is sum of
−∞ x
−vx /v0
∫ ∞ the vapour pressure and air pressure, hence p ≥ pv > patm .
/ −∞ e−vx /v0 dvx . So we can
2 2 2 2
Let us recall that f (vx ) = e
Therefore, the vapours at the liquid surface will have larger
rewrite the expression for the root-mean-square velocity as
∫∞ ∫∞ pressure than the atmospheric one, and the surrounding
v̄x2 = −∞ vx2 e−vx /v0 dvx / −∞ e−vx /v0 dvx = 21 v02 .
2 2 2 2
air will be pushed away. Furthermore, if there were a small
√
Now, if we substitute back v0 = 2kB T /m (which corresponds bubble inside the liquid, it would also have higher pressure
to the Maxwell's distribution), we obtain nally of vapours inside than the pressure of the surroundings,
v̄x2 = kB T. hence the bubble would start growing. It should be noted
that there are always either microscopic bubbles or other
14 Text here is somewhat modified

— page 37 —
 4. ENTROPY AND CARNOT CYCLE
impurities inside the liquid which can serve as evaporation Answers
centres.
1. 4P T 3 /aT04
25. Apply the same approach what was used to derive expres- 2. m = Q/(L + c∆T ) ≈ 28 g (Q = 500 W · 37 s)
sion for the contact angle in terms of surface tensions, but U12 T
keep in mind that solid-liquid contact area is now smaller. 3. ∆T = 8ρe ρcl2 ≈ 34 K
26. We have a meniscus as shown in figure, and separate a 4. P ′ = P tt21 −t1 ◦
−t0 ≈ 53 C
∫
fraction of water (depicted in grey) by a fictitious hori- 5. ∆T = P ρd/s ≈ 11.7 K; Q = (T1 − T2 )S/ ρdx ≈ 18 mW
zontal plane passing the flat bottom of the meniscus, and √
6. T = T⊙ R⊙ /2L ≈ 290 K
consider the force balance for the grey volume. At the sep-
aration plane inside the liquid, the hydrostatic pressure 7. N +1
√
equals to the atmospheric one. Indeed, at the bottom of 8. Tmin = (I/σ)1/4 , Tmax = ( 3I/σ)1/4
the meniscus, through the flat water-air interface, there is √
no capillary pressure, hence the hydrostatic gauge pressure 9. v≈M m RT
µ

must be zero; inside the liquid, the hydrostatic pressure is 10. N ≈ 162
√
a function of height only, so the pressure remains equal
11. Φ ≈ 32 nRT1 RT2
M
to the atmospheric one through the horizontal plane. So,
the volume depicted by grey in the figure is surrounded by 13. ∆m = ( T 1 − 1 p0 V0
T1 ) R (µair − µH )
air
atmospheric pressure, i.e. there is no extra net force acting 14. T = 1.4T0 .
on it due to pressure.
4
15. is reduced by 2 times; γ = 3
ρ0 T0 g∆h γ−1
16. ∆T = p0 γ = 2.1 celsius
18. a) vP ≈ 23 m/s; b) where the distance between the lines
29. Notice that the volume of the water is fixed: if the layer is the smallest; c) vcrit ≈ 23 m/s,
√
thickness were small, the gravitational energy would be 20. 2.5
small, but the surface energy would be large (the opposite
21. (a) a = pM
0S
; (b) a = p2M0S

also holds). Express the total energy for a fixed amount

of liquid. 22. a) ∆T ≈ 1.5 ◦ C

30. Use the fact 17 and idea 19 to conclude that the shape of b) ∆T ≈ 6.5 ◦ C
the cross-section of the meniscus is identical to the cross- c) ∆T = 12.1(1 − r)
section of the pool of liquid laying on the desk (cf. problem d) ≈ 4
29). Indeed, the surface curvature depends only on the sur- T −T100 ◦ C
face tension and gauge pressure, which can be matched in 23. κ = λ c = 0.154
the case of the meniscus and a pool as shown in figure. 24. a) TA = 350 K TB = 100 K
b) t1 ≈ 67 ◦ C t2 = 100 ◦ C mA
mB ≈ 22
◦
25. 175
√
2π Vπ
26. ∆V = ρ h = 5.79 ml3
27. 13.3 cm
√
32. Assuming the opposite, consider two heat engines with dif- 29. h = 2σ ρg (1 + cos α)
ferent efficiencies. The Carnot’ cycle is reversible, so we’ll √
σ
use the engine of higher efficiency as a heat pump (with 30. h = ρg
reversed Carnot’ cycle): it takes heat from a body of lower T0 −T1 P0 µL
35. dp λµ
dT = ps RT 2 ; ∆P =
s
T02 R ≈ 350 Pa
temperature and gives it to a body of higher temperature;
a net mechanical work needs to be done to operate the 36. U = L/µNA — exit work, negative potential energy of a
heat pump. Show that if the work done by the heat en- molecule in the liquid phase.
gine of lower efficiency is used to propel the heat pump of 37. m = I 2 RtT2 ≈ 1.5 g
(T1 −T2 )λ
higher efficiency, the net effect of the entire system would
be in violation to 2LTD. 38. T3 ≈ 1.4T3 ≈ 3.5T1
39. T = I0VR0 0 T0 = 2650 K
46. For the cooling period, during the number of days ∆N √
for which the temperature stayed in the (small) range −5
d = 3 π4V 2 R kσT 4 = 1.5 × 10
0 I0 ρ 0
m
between T and T + ∆T , the heat loss is Pc = (T −
0

2
T0 )C∆N · 3600s /h. (T − T0 )∆N is a horizontal narrow l = R4ρ
0d π
0
= 0.0566 m
rectangular region between the graph and the vertical line 41. d = 1 d
2 11 1
T = T0 , hence the total heat loss is related to the area
between the graph and the line T = T0 . 42. b) T = VV11 +VV22 ≈ 16.5 ◦ C; c) r ≈ 1.22; d) m ≈ 7.5 g.
T1 + T2

— page 38 —
 4. ENTROPY AND CARNOT CYCLE
∆s(p0 πr 2 +mg) εσ4πR12 (T24 −T14 )
43. a) T2 = T0 + nR = 322 K 61. a)P = Qdi − Qdo = 1+(1−ε)R12 /R22
≈ 1.78 W;
p = p0 + mg
= 102.32 kPa b) τ = 43 πρR3 λµ/P ≈ 36 h.
πr 2
b) W = (mg + p0 πr2 )∆s = 24.1 J 62. a) N = − lnαβ ;
( ) b) A ≈ N p0 V α = p0 V | ln β|;
c) Q = ∆(p0 πr2 + mg) cRV + 1 = 84 J
c) T = T0 β γ −1 ;
1
( )
d) P = cRV + 1 ∆s 2
∆t (p0 πr + mg) = 8.4 W d) A = 2p0 V .
λ
ṅ = P hc = 2.2 × 1019 s−1 63. a) C ≈ T0P−T
τ
= 350 J/◦ C;
1
e) η = ( p0 πr 2
1)
= 2.8 × 10 −3
b) Q = CTc ≈ 46 kJ.
1+ mg (1+ cRV )
( ) γ−1 64. 4689 K
p0 γ
f) T3 = T2 = 321 K 65. 1 − T1
T2 .
p
γ−1
44. T1 = 3 γ T0 ≈ 400 K 66. a) Q = qν ≈ 546 J;
3 c ≈ 0.22 mm/s (an approximate answer, e.g.
Q
b) v = 3ρa
45. η = 1 − k 1−γ = 0.602
v ≈ 0.67 mm/s. is also OK);
46. a = 290 EU R −Bt
c) T = A · e√ .;
47. a) ρ20 ρel cl2 (T1 − Tr )/U02 = 25 ms d) v ∼ Q MH
≈ 180 m/s.
∫ ρa3 RT
b) ρ20 cl2 U0−2 ρel dT = 1.5 ms 67. 135%.
xσT0 +ρhcp vdT1
48. a) T2 = xσ+ρhcp vd . 68. F = 7Sp1 T0 /T1 .
∫ T2
b) t = −C T1
dT
P (T ) ≈ 120 s 69. T = T0 − (1 − γ −1 ) µg
R
γ
49. a) cp = γ−1 R 70. a) dT
T = (1 − γ −1 ) dp ◦
p ; b) dp = − kB T dz; c) Ttop = 20.6 C.
mgp

p0 M
b) ρ = RT 71. a) U = µ 43 πR3 C(Tc − T0 ) = 16 768 J;
c) ∆ρgL = ρ0 v 2 (2∆ρgL = ρ0 v 2 is also OK, cf. Mechanics b) J = κ(T1 − T0 )/R = 2458 W · m−2 ;
Pr. 29.) c) P = 4πR2 J = 19.3 W; d) τ = U/P = 869 s (these are
[ ( )2 ] estimates, so different numerical factors are acceptable).
P gL γ−1 3
1
d) T = T0 1 + gL Sp0 γ ≈ 322 K w0 ◦
72. p = p0 TT0 w0 −V α(T −T0 ) ; invalid if T > 22 C.
w
50. a) p = p0 w−V α(T T
−T0 ) T0 73. Ai) α = gMair /RT0 ; Aii) P = P0 − ρgz; Aiii) 88 kPa
b) Tmax = T0 + w
= 22.0 ◦ C Bii) v = vG rG√/r; Biii) 141 m/s;
Vα
Biv) r = rG / 1 − 2gz/vG 2 ; Bv) Pipe-like ones.
51. p ≈ 56 kPa Ci) v = vG r/rG ; Cii) 76 kPa;
( )1− γ1
52. a) T1 = T0 p1
= 279 K Ciii) between 5 ◦ C (at G) and −7 ◦ C (neglecting latent
p0
heat of condensation); Civ) heat released by condensation;
2 p0 −p1
b) h1 = gρ0 1+ p1 T0 = 1410 m Di) 3.8; Dii) a close call.
p T 0 1
( )1− γ1 P
74. Ai) FB = MA ng P +∆P ; Aii) γ = ρ0 z0 g/P0 = 5.5;
c) T2 = T1 p2
+ qV m
= 271 K ( )
p1 cp Bi) ∆P = r0 λ − λ−7 ; Bii) a = 0.110;
4κRT −1

d) h = 35 mm Ci) zf = 11 km, λf = 2.1.

( )1− γ1 √
75. T⊙ = R⊙ 2L ≈ 289 K
e) T3 = T2 pp23 = 300 K √ √
( √ ) √ 76. Wmax = C( T1 − T2 )2
2
53. L = 12 MmpS RT + t − 1 2RT m RT
µ pS µ M 2µ 77. T2 = T0 κ
κ+Iα ≈ 288 K
t ≈ 160 s
78. T0 ≈ 212 K
54. t ∼ 1
100 s
√ 79. 13.2 ◦ C
√
56. T = d I
σ; D/F > 2 80. 1.2352 kg/m3
αf
√
B 2 1 Tair
57. 1900 m. 81. a) h ≥ A2 2g ∆T ; b) 45 m; c) v = 2gh T∆T is constant;
√ air
58. a) C = 4R; b) ω = r−1 8σ/ρh d) p = p0 − (ρair − ρsmoke )gh − ρsmoke gz; e) η = cTgh ;
air
f) linear; g) 0.64%; h) 45 kW; i) 360 kWh with 8 h;
59. a) pE > pA > pD > pB > pC ; b) 195 K; c) 220 hPa; d) ( )1/3
Wnet = R(TH − TC ) ln(pE /pA ), Qloss = RTC ln(pD /pC ); G2 S 2 T
j) ∆T = A2 c2 ρ2 air 2gh
≈ 9.1K; w = 760 kg/s
hot
e) εi = 0.35; g) II & III.
82. V1 = 0.472 l, p1 = 1.06 atm; V2 = 0.395 l, p2 = 1.22 atm,
60. a) R = L/4κS; ( √ ) Vf = 0.100 l, pf = 1.22 atm
P0 cos ωt+arcsin(Cω/ C 2 ω 2 +R−2 )
b) T = T0 + √ ;
C 2 ω 2 +R−2 83. a) 14 g/mol; b) 3.3 km; c) 420 km; d) 2.6 × 10−3 ;
√
c) ω = 1/ 2CR; e) 7.5 × 1011 1/m2 s; f) 1.2 × 1038 ; g) 4500 years;
d) ωc ≈ κ/cρL2 . h) 9.6 × 1011 years; i) supply as H2 O from oceans.
— page 39 —
 4. ENTROPY AND CARNOT CYCLE
3mRT0
( )3γ−3
84. a) 8; b) µ ln rr03 ; c) T0 rr3 ;
( ) 3γ−4
1 ( ) 3γ−3
RT0 r3 4−3γ
d) r4 ≈ r3 RT 0 r3
µmG , T4 ≈ T0 µmG

85. a) Rm = 2.31 µm, Tm = 6.86 × 104 K; b) Rf = 2.98 µm;

c) τ ∼ 3 × 10−9 s; d) a0 ∼ 0.01
2
86. a) 32 kB T ; b) k3 B T /EF , as the temperature grows, energy
of a majority of the electrons will remain unchanged.
87. a)−N µB B tanh(µB B/kB T ) b) N (µB B)2 /kB T .
88. m = 1.5RT νρ/σ (proof of minimality required).

— page 40 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
ELECTRICAL CIRCUITS chemical energy (in the case of magnetohydrodynamic generat-
Jaan Kalda Version: 3rd December 2017 ors and inductors/inductor based dynamos, the nature of emf.
New: idea 49, fact 8, appendix 9, and problems 89, 106–111; updated ideas 47,52 is somewhat different but for the time being, the details are not
1 Circuits with resistors, batteries, ammeters important: practical application of the Kirchoff’s laws remains
the same). With the mountain-hiking-analogy, an electromot-
and voltmeters
ive force can be considered as a ski-elevator which lifts you
The fundamental physics of circuits of resistors, batteries, upwards and performs a certain work on you each time you
ammeters and voltmeters is really simple, and is essentially use it.
covered with just four laws: the two Kirchoff’s laws, Ohm’s While the mountain-hiking-analogy works only for the
law and Joule’s law1 — formulated here as “facts”. First, the Kirhoff’s voltage law, the channel-network-analogy can be ex-
Kirchoff’s laws: tended to all the direct current phenomena. More specifically,
we consider a closed system of water channels; in a channel, the
fact 1: The sum of electrical currents flowing into a node2
water flows only downhill, but there are also pumps which raise
of a circuit is zero.
the water uphill. Then, there are the following matching pairs:
Mathematically, (a) electrical charge Q — mass of water m; (b) electrical current
∑
Iν = 0, I in a wire, defined as the charge flow rate Q/t, where Q is the
wires connected to the i-th node charge flowing through a cross-section of a wire during a time
where Iiν stands for the current in the ν-th wire. This assumes interval t — mass flow rate of water µ in a channel, defined as
that Iν is taken with a ‘+’ sign if it flows into the i-th node m/t, where m is the water mass flowing through a cross-section
and with a ‘−’ sign otherwise. We can also say that the sum of of a channel during a time interval t; (c) a battery of electro-
in-flowing currents equals to the sum of out-flowing currents. motive force E which performs work EQ on charge Q (which
Since the electrical current is defined as the charge flowing passes through the battery) — a pump which pumps water up-
through a wire’s cross-section per unit time, this law is essen- hill, to an height h, and performs work hm on a pumped water
tially the continuity law for electrical charges, combined with mass m4 . Then, obviously, for a closed loop of channels and
the fact that typically, the capacitance of any wire and any cir- pumps, the total pumping height (i.e. the sum of the contribu-
cuit node is negligible 3 (hence, the charges residing on the tions of all the pumps) equals to the total downhill descending
nodes and wires can be neglected). height in channels (i.e. to the sum of downhill displacements of
For those who are not yet developed good intuition with all the channels).
electrical currents, the analogy with water flow in branching Next, the Ohm’s law:
rivers or water pipes might be useful: the sum of the water fact 3: Typically, the voltage V between the input- and out-
fluxes (measured in cubic meters of water per second) equals put leads (also referred to as the ports)5 of a piece of electrically
to the water flux in the main stream. Note that the continuity conducting material can be considered to be proportional to the
law plays an important role for many physical processes (with current I through it; the coefficient of proportionality
gas- fluid or granular flows, but also e.g. for traffic fluxes).
R = V /I
fact 2: Along a closed loop of an electrical circuit, the sum is referred to as its resistance, and the circuit elements of a
of voltage drops on the circuit elements (resistors, diodes, ca- non-negligible resistance are called resistors.
pacitors, etc) equals to the sum of the electromotive forces (of
batteries and inductors). Let us try to interpret this using the pipe-flow analogy. Con-
sider a straight pipe connecting two water reservoirs at different
Mathematically, height. Let us assume that the drag force F between a unit
∑
Vν = 0, volume of the flowing water and the pipe’s walls is proportional
wires forming a closed loop to the speed v of the flow6 : F = kv. Then, the water speed
where the voltage drop on ν-th wire is taken with ‘+’ sign if is established by the balance between the drag F = kv and
the voltage of the destination node is lower than that of the pressure ρw gh, where h is the height difference, ρw — the wa-
departure node, unless the wire includes an electromotive force ter density, and g — the free fall acceleration. Therefore, v
(emf.): the voltage drop caused by an emf. is taken with the will be proportional to h, which, according to the analogy, cor-
opposite sign. responds to the voltage. Now, let us recall that the current
This law simply states that electrostatic field is a potential I corresponds to the water flux, which equals to the product
field; using a mountain-hiking-analogy, if you walk so that you of the water speed v and pipe’s cross-sectional area S, and is
end at the same point from where you started, you ascended therefore also proportional to h (the counterpart of the voltage
exactly as many meters as you descended. The electromotive V ). Such a proportionality is exactly what is stated by the
force of a battery performs work on charge carriers by using Ohm’s law.
1 G.
Kirchoff 1845, G.S. Ohm 1827, and J.P. Joule 1841, respectively
2 node
(=vertex in graph theory) — a point where different wires meet
3 In the case of very fast or high-frequency processes, this approximation is not valid; then, an equivalent circuit can be used, with ideal wiring and

equivalent capacitors and inductors representing the capacitances and inductances of the real wires.
4 Here we have put the free fall acceleration g = 1 which can be done if an appropriate system of units is used
5 Ports (input- and output leads) — points where the current can enter and exit; often just the endpoints of a wire.
6 This is valid for sufficiently thin pipes for which viscous drag dominates over the turbulent one

— page 1 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
For water flow in a narrow pipe, the drag force is propor- idea 1: If a circuit can be presented as a combination of
tional to the flow speed and to the pipe length l, i.e. k = κl. series- and parallel connections (see below for an algorithm),
For ordinary pipes, the drag force (and hence, the coefficient the system of Kirchoff’s equations becomes decoupled, and
κ) depends also on the diameter of the pipe. However, let us there is no need to write the full system of equations. Instead,
assume that κ is constant (this would correspond to the case the following rules can be applied. (A) For parallel connections,
when we fill the pipe with a granular material, e.g. coarse sand). the net conductance (inverse resistance) is the sum of conduct-
Pursuing the analogy, the resistance R = V /I corresponds to ances and the net current splits in proportions proportional to
the ratio of the height difference h, and water flux. According the conductances:
to the discussions above, this is proportional to the pipe length 1 ∑ 1 Rpar I
= ; Ii = .
l, and inversely proportional to the cross-sectional area S of Rpar i
R i Ri
the pipe (because for a fixed v, the flux is proportional to S). (B) For series connections, the net resistance is the sum of res-
Hence, we arrive at the following fact. istances and the voltage is distributed proportionally to the
fact 4: The electrical resistance of a wire (of a length l and resistances: ∑ Ri V
cross-sectional area S) Rser = Ri ; Vi = .
i
Rser
R = ρl/S,
Algorithmically, the procedure of applying the idea 1 can
where ρ is called the electrical resistivity of the wire material
be formulated as follows. If two or more resistors are connected
(σ = 1/ρ is called the conductivity).
between the same pair of nodes A and B, substitute these with
The proportionality law between V and I fails actually an equivalent resistance according to the formula for Rpar ; if
quite often: for instance, in the case of light bulbs, the de- two or more resistors form a branch-less chain connecting nodes
pendence between the voltage and current is nonlinear. Even A and B, substitute this chain with an equivalent resistance ac-
then, the ratio V /I is referred to as the resistance. In the case cording to the formula for Rser ; remove all the “dangling ends”
of a non-linear V − I dependence, the resistance just depends (parts of the circuit which are connected to the rest of the cir-
on the voltage; the derivative dVdt is referred to as the differen- cuit only via a single wire); repeat the process iteratively. The
tial resistance. If a circuit element is referred to as a resistor, process will stop if (A) only one equivalent resistance remains,
its resistance is assumed to be constant. or (B) if a bridge is formed (i.e. for a set of four nodes, five or
Finally, the Joule’s law: more pairs are connected via resistors).
fact 5: The power dissipated on a circuit element These rules and formulae can be easily derived using the
Kirchoff’s laws. Indeed, all the resistors connected in paral-
P = IV,
lel between A and B have the same lead voltage VAB ; hence,
where V is the voltage on its leads, and I — the current through the currents I = V /R are proportional to the conduct-
i AB i
it. Alternatively, bearing in mind that ances. This gives rise to total current between A and B
∑ ∑ ∑
i 1/Ri , which leads us
I = Q/t, I = i Ii = i VAB /Ri = VAB
where Q is a charge flowing through the element and t is a time to the above given formula for Rpar ≡ VAB /I. All the res-
interval, we can say that the current performs work istors connected in series between A and B have the same
current IAB passing through, so that the voltage on each of
A = QV.
them Vi = IAB Ri , i.e. the voltage is proportional to the res-
∑ ∑
Using the analogy of the rivers (channels), the power of a water- istance. The total voltage is V = i Vi = IAB i Ri , hence
∑
fall’s power plant is given by the gravitational potential energy Rser ≡ VAB /I = i Ri . Finally, regarding the removal of the
released per unit time, which is proportional to the product of “dangling ends”: due to the Kirchoff’s laws, the sum of currents
the waterfall’s height and the water flow rate. entering a subset of a circuit needs to be zero; if there is only
Using mathematical induction, it is not too difficult to show one wire connecting a circuit’s subset to the rest of the circuit,
the following fact. its current needs to be zero, hence it does not affect the current
fact 6: If all the resistors and battery voltages (the electro- distribution and its presence can be ignored.
motive forces) are known, and currents of the wires are con- The following problem illustrates the idea 1 in its simplest
sidered as unknown variables then the Kirchoff’s laws and the form (for a series connection), together with the fact 4.
Ohm’s law form a closed set of linear equations which can be
pr 1. A uniform wire of cross-sectional area A0 = 1 mm2 had
solved to find all the currents and voltages in the circuit (i.e.
a millimetre scale marked on it: an array of streaks with inter-
the solution is unique).
streak distance a0 = 1 mm covered the entire length of the wire.
This fact itself can be sometimes useful: if you are able to The wire was stretched in a non-uniform way, so that the inter-
“guess” the solution, it will suffice to show that all the Kirchoff’s streak distance a is now a function of the distance l from one
equations are satisfied (there is no need to derive the answer end of the wire (as measured after the stretching), see figure.
systematically). The new length of the wire is L = 4 m. Using the graph, determ-
Typically, the number of unknown variables in the ine the electrical resistance R of the stretched wire assuming
Kirchoff’s equations (and hence, the number of equations) is that the resistivity of the wire material is ρ = 1.0 × 10−6 Ω m.
large, and solving can be tedious. In order to make calcula- During the stretching, the density of the wire material remains
tions easier, several tricks and techniques can be applied. constant.
— page 2 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
The next problem serves as another simple example of the
idea 1.

pr 2. In the figure, R1 /R2 = 4. If we add a lamp as shown

if figure, current through R1 will increase by ∆I = 0.1 A. Find
the current through the lamp.

However, for this problem we need one more idea, which is

very universal, not limited to electricity. It can be solved in a long way, and in a short way. For the
idea 2: Many physical quantities can be expressed as integrals long solution, another very generic idea is used.
of other quantities — these can be found as surface areas under idea 3: If it seems that there are too few data provided in
graphs. In order to figure out, which surface area is needed, the the problem text, just assume the “missing” data to be known
following technique can be used. Divide the parameter range (here, for instance, the lead voltage U and the resistance R1 );
(for the problem above, the parameter l) into small intervals; if if everything goes well, the “missing” data will cancel out from
each interval makes an additive contribution to the given phys- the answer.
ical quantity (here, the resistance R), express this contribution
For the short solution, a useful modification of the Kirchoff’s
in terms of the interval width and other relevant parameters;
laws can be applied.
design such graph axes that this contribution is proportional
to the surface area of a thin rectangular region in the graph. idea 4: Kirchoff’s laws are not valid only for the currents and
Then, once we sum up the contributions of all the intervals and voltages, but also for voltage increments ∆Vi = Vi (after) −
tend the interval widths to zero, the physical quantity of our Vi (before) and current increments ∆Ii = Ii (after) − Ii (before).
interest will be expressed in terms of a graph area. Sometimes the circuit is drawn so that it is not very easy
For the problem above, each wire segment of length ∆l will to understand, does it break down into parallel- and series con-
contribute a resistance ∆R = ρ∆l/A to the overall resistance nections or not. In that case, the following idea is to be used.
R; these wire pieces are connected in series, so the resistances
can be just added up. The wire volume remains constant, idea 5: Redraw a circuit so that its structure becomes as
Aa = A0 a0 , hence A = A0 a0 /a so that clear and simple as possible: contract plain wires (which con-
ρ
∆R = a · ∆l. nect a pair of leads) into a single point and if possible, em-
A0 a0
Note that A0 a0 is a constant (does no depend on l), and a · ∆l
ρ phasize the structure of parallel- and series connections. Bear
is the surface area of the blue rectangle in the a − l, see fig. be- in mind that if several leads are all connected together with a
low. The sum of all these rectangles (the grey and blue region plain wire, the wiring can be arbitrarily rearranged (as long as
in figure) approximates the surface area of the region between the relevant leads remain connected), for instance as shown in
the a(l)-curve and the l-axis, and at the limit ∆l → 0 becomes figure below. Indeed, one can say that the effect of a wire is
equal to that area. Such infinitesimally small increments are equalling the voltages on two leads, and in the case of several
called differentials and are denoted by the prefix d (substitutes leads, it doesn’t matter in which order the lead voltages are
the prefix ∆ which we used for finite increments); the sum over made equal.
all the infinitesimally small intervals is denoted by the integral
∫ ∑
sign . So, we can say that R ≈ A0ρa0 a · ∆l (where the sign
∑
denotes summing over all the intervals), and at the limit of
This idea can be illustrated with a task from the 27th IPhO,
infinitesimal increments dl we obtain
∫ 4 m equality
ρ see below.
R= a · dl,
∫ 4m A0 a0 0 pr 3. Determine the resistance between the leads of the cir-
where 0 a · dl is the surface area under the a(l) graph. cuit in figure.

[IPhO-1996] For complex circuits, it is easy to make mistakes

while simplifying the circuit; typically, this happens when the
remote nodes are connected with wires. To avoid mistakes,
the following technique can be applied. Label all the resist-
ors, e.g. with letters; if there is more than one battery, label
the batteries, as well. Label also all the nodes, so that all the
nodes connected with a plain wire bear the same label, and
those which have no direct wire connection have different la-
— page 3 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
bel. Then, start redrawing the circuit by marking (on a sheet
of paper) one node and drawing all those resistors which are
connected to it. Next, select another lead of one of the drawn
resistors or batteries, mark the respective nodes and draw the
resistors which are attached to that node; repeat the process
until the entire circuit is redrawn.
As an example, let us consider the last problem. We mark
the nodes and resistors as shown in figure. Note that due to
the wire connections, the node symbols appear in two different
places.
We start with drawing the node ‘A’,
see the figure. Since the node ‘A’ is
directly connected to the resistors ‘1’,
‘2’ and ‘3’, we draw these resistors at-
tached to the node ‘A’ as shown in figure. The other ends of
the resistors ‘1’ and ‘2’ are fixed to the node ‘C’, hence we can
connect the respective wires and designate the connection point
by ‘C’. Further, the other end of the resistor ‘3’ is connected to
the node ‘B’, so we draw a wire and mark its end with ‘B’. Now,
by noticing that the resistors ‘4’ and ‘5’ connect the nodes ‘B’
and ‘C’, it is easy to complete the circuit.
In the case of non-trivial circuit-redrawing tasks, it is highly
recommended to use this technique of denoting resistors and
nodes with letters and numbers (you don’t want to make a
mistake in redrawing!).
idea 6: If a bridge connection involves only an ideal ammeter
(of zero resistance) or an ideal voltmeter (of infinite resistance),
the bridge connection is only seemingly there, and can be es-
sentially removed (for voltmeter) or short-circuited (for am-
meter). Similarly, it can be removed if there is no current in
the bridge connection due to symmetry. Once the simplified
circuit is solved, it may be necessary to return to the original
(non-simplified) circuit: in the case of an ammeter in a bridge
connection, its current can be found from the Kirchoff’s current
law (written for the currents entering the node to which the am-
meter is attached to); in the case of a voltmeter, its voltage can
be found as the voltage difference between the nodes to which
it is attached using the Kirchoff’s voltage law and the voltages
of the relevant resistors.
In order to illustrate this idea, let us consider the following
problems.
pr 4. Determine the resistance between the leads of the cir-
cuit in figure.
pr 5. Determine the reading of the ammeter in figure.
— page 4 —
idea 7: If there are non-ideal ammeters, voltmeters, batter-
ies or current sources included into a circuit then the following
rules can be applied: (a) non-ideal battery of internal resistance
r can be represented as a series connection of an ideal battery
(of zero internal resistance) and a resistance r; (b) non-ideal
current source of internal resistance r can be represented as
a parallel connection of an ideal current source (of infinite in-
ternal resistance) and a resistance r; (c) non-ideal voltmeter
can be represented as a parallel connection of an ideal volt-
meter (of infinite resistance) and a resistance R; (d) non-ideal
ammeter can be represented as a series connection of an ideal
ammeter (of zero resistance) and a resistance R. NB! A non-
ideal ammeter is not a faulty ammeter: it still shows the true
current through itself; similarly, a non-ideal voltmeter shows
the true voltage on its leads.
Regarding the typical values of the internal resistances of real
ammeters and voltmeters, the following guideline can be used.
The most common digital voltmeters have internal resist-
ance of 10 MΩ, but cheaper ones can have also R = 1 MΩ, and
the expensive ones can reach a gigaohm range; typically, the
internal resistance is independent of the measuring range. For
analogue voltmeters, the resistance does depend on the selec-
ted measuring range Vmax , and can be determined by knowing
the so called full-scale deflection current (FSDC). Essentially,
an analogue voltmeter is a galvanometer (device which has a
needle which deflects proportionally to the current through it),
connected in series with such a resistance that with the max-
imal voltage Vmax applied, the current will be equal to the
FSDC. So, if IFSDC = 100 µA, and the 10-volt range is selected
then the resistance r = 10 V/100 µA = 100 kΩ. Typical values
of the FSDC are in the range from 25 µA to 1 mA.
Digital ammeters measure internally voltage on a small
resistor (shunt) and translate the result into corresponding
amperage; depending on the selected range of currents, dif-
ferent shunt is used; the voltage drop on the shunt is called
the burden voltage, and the maximal burden voltage (MBV)
VMBV can be used to determine the resistance; for instance,
for the 20-mA range and VMBV = 300 mV, the resistance is
300 mV/20 mA = 15 Ω. Typical values for VMBV range from
100 mV to 1 V. An analogue ammeter is essentially a galvano-
meter connected in parallel with a small resistor (shunt); the
shunt controls which fraction of the net current goes through
the galvanometer and ensures that the voltage on the galvano-
meter does not exceed the full-scale deflection voltage (FSDV).
The shunt resistance can be determined in the same way as in
the case of a digital ammeter: here, FSDV plays the role of the
MBV.
In the case of theoretical Olympiad problems, voltmeters
and ammeters are usually assumed to be ideal, unless other-
wise noted. However, there is an exception to this rule: if the
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
problem conditions contradict the assumption of ideality, you Note that ∆-connection is also called “triangular”, and Y -
need to abandon it. Please note that in the case of theoret- connection — a “star”. So, the idea is to substitute either a
ical problems, it is not wise to make assumptions regarding the triangular connection with a star connection or vice versa so
values of the internal resistances of non-ideal ammeters and that the resulting circuit is simpler to analyse than the ori-
voltmeters: quite often, the authors of the problems do not ginal one. While doing so, all the three lead potentials need
check how real the numerical values of the resistances are. to remain unchanged. Let us consider the simplest case when
all the three resistors are equal: for a ∆-connection R, and
pr 6. Two identical voltmeters and two ammeters are con-
for a Y -connection — R. Then, the inter-lead resistance of
nected to a battery as shown in figure. The readings of the
the Y -connection is 2R (two resistors in series), and for the
devices is as follows: ammeter A1 — I1 = 200 µA, voltmeter
∆-connection — 23 r (2r in parallel with r). Therefore, there is
V1 = 100 V, and voltmeter V2 = 2 V. What is the reading of
matching between these circuits if 2R = 23 r, hence r = 3R: the
ammeter A2 ? Estimate, how realistic are those internal res-
∆-connection needs to have thrice as large resistances as in the
istances which can be determined from these data; if there is
case of a Y -connection. This rule — if forgotten — can easily
something strange, is it possible to “fix” the problem by chan-
derived whenever needed.
ging the circuit so that the solution would remain intact?
In the generic case of non-equal resistances, the Y − ∆-
substitution formulae are derived by solving the system of three
equations stating pair-wise equality of the inter-lead resistances
rAB , rBC , and rCA ; the result is as follows: for a ∆-to-Y -
substitution
RAB RAC
RA = , (1)
RAB + RAC + RBC
and analogously for RB and RC (the indices are to be substi-
idea 8: Sometimes it is convenient to consider the Kirchoff’s tuted cyclically); for a Y -to-∆-substitution,
current law for a whole region and not just for a single circuit 1
· 1
1
node: the sum of currents entering the region equals to the = 1 RB 1 RC 1 , (2)
RBC RA + RB + RC
sum of outgoing currents.
and analogously for RBC and RCA .
This idea can be illustrated with the following problem. It would consume quite a lot of time to derive these formu-
pr 7. [EstPhO-2003] 15 identical voltmeters and 15 non- lae during an Olympiad, so it is better to remember them. Re-
identical ammeters are connected to a battery as shown in fig- membering is actually not that difficult; first, let us talk about
ure. The reading of the first voltmeter is V = 9 V, the readings ∆-to-y-substitution which is typically more useful than the re-
of the first two ammeters are I1 = 2.9 mA and I2 = 2.6 mA. verse one (there are exceptions) as it removes a loop from the
What is the sum of the readings of all the other voltmeters? circuit — loops can create bridge connections and are difficult
to analyse. Even if that is not the case, ∆-to-y-substitution
tends to reduce the number of parallel connections (the reverse
substitution tends to increase it), leading to simpler calcula-
tions as typically, resistances and not conductances are given.
The denominator of the formula is very simple - just the sum
In some cases, the bridge connection is real and cannot be of all the resistances. The nominator is also simple, a product
removed. In the case of Olympiad problems, this is happens of two resistances, we just need to be able to figure out which
very seldom, because in that case, the difficulties are actually of the three resistances is to be excluded from the product.
only mathematical: it is needed to solve the linear system of This, however, can be easily figured out from the symmetry
Kirchoff equations. There are several methods which simplify considerations: so, for a resistor attached to the node B in Y -
this mathematical task which are presented in what follows. connection, we exclude the resistor at the opposing side AC of
idea 9: Any circuit which consists only of resistors and has the ∆-connection.
three ports is equivalent to a ∆- or a Y -connection of three If we really need a y-to-∆-substitution, the formula can
appropriately chosen resistors7 In particular, a Y -connection be also easily deduced from the structure of the ∆-to-y-
can be substituted by a ∆-connection and vice versa8 . substitution: just the resistances need to be changed to con-
ductances.
Note that the idea 9 cannot be generalized to circuits of
resistors with n ports with n > 3.9 An exception is the case
when all the pair resistances are equal (to R), in which case the
circuit is equivalent to a star connection of n resistors, each of
resistance R/2 (though it is still not equivalent to a n-gon-
connection of equal resistances, because for a polygon, close
node pairs have smaller resistance than remote node pairs).
7 The proof is provided in Appendix 1 on pg. 14.
8 A.E. Kennelly, 1899.
9 Indeed, there are 1 n(n − 1) different lead pairs, which can all have different resistances; for a generic case, the respective 1 n(n − 1) equivalence
2 2
equations cannot be solved with respect to the n resistances of a star (or a polygon) connection as long as 12 n(n − 1) > n, i.e. n > 3.

— page 5 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
As an illustration, let us consider the following problem. orange wire (see figure) and calculate the resistance between
the leads: the parallel connections of 1-ohm resistors and the
pr 8. Determine the current through the battery. parallel connections of 2-ohm resistors are connected in series,
so that r = ( 12 + 1)Ω = 1.5 Ω. Finally, we return the red part of
the circuit to its place for the equivalent circuit at right (keep-
ing in mind that E = 0 and r = 1.5 Ω): the current through
the 3-V-battery is I = 3 V/1.5 Ω = 2 A.

idea 10: Any circuit which consists only of resistors and

batteries and has two ports A and B is equivalent to a series
connection of a battery and a resistance (the Thvenin’s the-
orem)10 . The electromotive force E of the battery can be found
as the voltage difference between the leads A and B when there
is no load connected externally between these leads (this is be- The following fact can be quite easily derived, but knowing
cause the original and the substitution circuits must behave it will can save some time during an Olympiad.
identically when there is no load).
fact 7: For drawing the maximal power from a battery, the
The resistance r (the internal resistance of the battery) can be load’s resistance needs to be equal to that of the internal res-
found as E/I0 , where I0 is the current which would flow in a istance of the battery.
wire short-circuiting the two leads (this is because the original
Indeed, the load current I = E/(R + r), where R is the resist-
and the substitution circuits must behave identically when the
ance of the load. Hence, the power dissipation at the load
leads are connected by a wire). Alternatively, r can be found
can be found as P = RI 2 = E 2 R/(R + r)2 . Let us no-
as the resistance between the leads A and B when there is no
tice that instead of P , it would be easier to analyse 1/P , be-
external load, and all the ideal internal electromotive forces are
cause then the expression will break down into three additive
substituted by wires (this is because the original and the substi-
terms: P1 = E −2 r( Rr + Rr + 2). If P is maximal then P1 is
tution circuits must have identical increase of the lead voltage
minimal; we need to minimize this expression over the values
when there is a certain increase of the lead current, and an
of R. Upon taking derivative with respect to R we obtain
ideal battery and a piece of wire have identically a zero voltage d 1 −2
response to an increase of the current). dR P = E (1 − r/R2 ) = 0, hence R = r. (Alternatively, it
would have been possible to apply the fact that the sum of a
As an illustration, let us consider the following problems. number x and its reciprocal x1 has minimum for x = 1, hence
r = 1.) So, Pmax = E /4r.
R 2
pr 9. Determine the current through the batteries.
pr 10. Determine the maximal power which can be dissip-
ated on a load connected to the leads of the circuit in figure
(the power depends on the resistance of the load, you need to
find the maximum of this dependence).

In order to make the application of the idea 10 more trans-

parent, let us solve the first part of the last problem, and find
the current through the 3-V-battery. In figure below, the black
and blue part of the circuit will be substituted by a battery
of electromotive force E and internal resistance r (see figure,
idea 11: Sometimes it is convenient to deal with constant
section on right). To begin with, we assume that there is no
current sources — instead of batteries (and sometimes, a cur-
load, i.e. the part drawn in red is missing. Then, the blue bat-
rent source is already present). A battery with electromotive
tery creates currents 4 V/2 Ω = 2 A and 4 V/4 Ω = 1 A in the
force E and internal resistance r is equivalent to a constant
left and right loops, respectively. Consequently, the voltage
current source with I = E/r which is connected parallel to the
drops on the resistances at the bottom of the figure (1 Ω and
shunt resistance r11 .
1 Ω) are equal to 2 A · 1 Ω = 2 V and 1 A · 2 Ω = 2 V, respect-
ively. Hence, the lead voltage is 2 V − 2 V = 0 V, i.e. E = 0. Constant current source is a device which generates a constant
Next, we calculate the internal resistance r of the equivalent current I regardless of which load is connected to the output
battery. To this end, we substitute the blue battery with the leads — as long as the load resistance is non-infinite. The
10 Formulated by H. Helmholtz in 1853 and L. Th�venin in 1883; for a proof, note that the behaviour of a two-lead circuit is defined by the relationship
between the lead voltage V and lead current I; owing to the linearity of Kirchoff’s and Ohm’s laws, this relationship is always linear, V = a − Ib. This
can be always matched with a battery of electromotive E and internal resistance r, for which V = E − Ir.
11 If we apply this equvalence to the Th�venin’s theorem, we obtain what is called the Norton’s theorem [E.L. Norton (1926), H.F. Mayer (1926)].

— page 6 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
validity of this theorem can be easily verified: it suffices to Let us illustrate the method and the concept of linearly inde-
check that for the same lead voltages, the lead currents are pendent loops using the following problem.
also equal. Suppose that a battery (of electromotive force E
pr 13. Determine the resistance between the output leads
and internal resistance r) has lead voltage V ; then, the voltage
of the circuit using the method of loop currents.
on its internal resistance is E − V and hence, the lead cur-
rent Ibattery = (E − V )/r. If the same voltage is applied
to a constant current source (of constant current I = E/r),
the shunt current will be V /r, i.e. the total current will be
Ic-source = I − V /r = E/r − V /r = (E − V )/r. Indeed,
Ibattery = Ic-source for any lead voltage V , hence this battery
and this current source behave identically. This problem can be solved using the idea 9 — and this is
The next problem illustrates the idea 11 (although it can possibly the simplest solution. However, here we provide its
be also solved using the idea 10). solution using the idea 13. To begin with, we need one more
idea.
pr 11. n batteries with electromotive forces Ei and internal
resistances ri (with i = 1, 2, . . . , n) are connected in parallel. idea 14: If the task is to find the resistance of a circuit
What is the effective electromotive force and the internal res- between two leads, it is often useful to assume that either
istance of such a system of batteries? a voltage V is applied to the leads, or a current I is driven
through these leads. Then we need to find the missing quant-
idea 12: The Kirchoff’s equations and the Ohm’s law are ity (I or V , respectively), and calculate R = V /I.
linear (each term includes only a first power of a current or a And so, we assume that a current I is driven through the cir-
voltage), hence the superposition principle is valid. More spe- cuit. Let us have a look on possible shapes of loop currents on
cifically, suppose we have a circuit which includes only resistors, the figure below.
n ideal batteries and m ideal current sources. Then the current
in the j-th wire can be found as
∑
n+m
Ij = Ij (k),
k=1
where Ij (k) is the current in that wire when only the k-th bat-
tery (or current source) is included into the circuit (all the other Let the blue loop be denoted by i1 , red — by i2 , green —
batteries are short-circuited and all the other current sources by i3 , and violet — by i4 . If we take the red and blue loop
are removed by cutting off a connection wire). currents with equal amplitude, they cancel out in the segment
passing through the 4-ohm resistance, hence their sum will be
pr 12. [EstPhO-2012] In the figure below, the batteries are equivalent to the green loop current. Therefore the green loop
current is linearly dependent on red and blue loop currents: out
ideal, R1 = R2 = R3 = R4 = R and E1 = E2 = E. Find the
of the three loop currents, only two can be kept as unknowns
currents in the resistors (i.e. I1 , I2 , I3 and I4 , expressed via R
(if we were keeping all the three loop currents, the number of
and E).
unknowns would be larger than the number of equations). It
doesn’t matter which pair of loop currents will be selected; let
us opt for i1 and i2 . However, with just the red and blue loops,
we cannot obtain any current through the input leads, which
means that the system of loop current is not yet closed: we
need a loop passing through the input leads. Any shape of
such a loop would do; let us use the one depicted by the violet
curve (it can be thought to be closed via the external battery).
Let us note that i4 needs to be equal to I — to the current
driven through the circuit.
(Note that this problem can be also solved using the idea 19.)
Now we have a full set of loop currents, i1 , i2 , and i4 = I,
idea 13: The number of unknowns and the number of lin- and we need to write down the Kirchoff’s laws for the voltages.
ear equations can be reduced by using the method of loop The current through the 3 Ω-resistor is i1 + I, so its voltage is
currents, in which case the first set of Kirchoff’s equations is V3 = 3 Ω(i1 + I); similarly V4 = 4 Ω(i1 − i2 ) (the minus corres-
automatically satisfied. The first step is selecting a full set of ponds to the fact that the currents i1 and i2 are antiparallel
linearly independent loops l1 , l2 . . . ln (the concept of linear de- in this resistor), and V1 = 1 Ωi1 . Please note that the signs
pendences is explained below); the second step is assigning to of these voltages have been taken corresponding to the blue
the loops respective currents I1 , I2 , . . . In , and expressing the loop current: positive voltage value means that when moving
currents in resistors via these loop currents. The final steps along the blue loop, the voltage decreases. According to the
is expressing the second set of Kirchoff’s equations in terms of Kirchoff’s laws, upon performing a full loop, the voltage drop
resistors’ currents using the Ohm’s law, and solving this set of needs to be zero:
equations with respect to the loop currents. 0 = V3 +V4 +V1 ⇒ 3(i1 +I)+4(i1 −i2 )+i1 = 8i1 −4i2 +3I = 0.
— page 7 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
We can write down analogous equation for the red loop’s in the case of planar circuits (i.e. circuits which can be drawn
voltage drop: on a paper so that the wires don’t intersect and meet only at
i2 (2 + 4 + 5) − 4i1 + 2I = 0. the nodes), this step is not needed if we use a slight modific-
From the first Eq., 4i1 = 2i2 − 2 I, substituting it into the ation of the loop current method which will be referred to as
3

second Eq. leads us to 9i2 + 72 I = 0, hence i2 = − 18 7

I and the streamfunction method.
i1 = − 4 ( 9 + 2 )I = − 72 I. Thus, V3 = (1 − 72 )I · 3 Ω =
1 7 3 41 41 The concept of streamfunction ψ(x, y) can be used for two-
72 Ω · I, and V2 = 18 Ω · I; the total voltage on the circuit
93 22 dimensional incompressible flows in which case the streamlines
is V = V2 + V3 = 181 72 Ω · I, which means that the resistance follow the lines of constant value ψ(x, y) = const, and the flow
181
R = V /I = 72 Ω. flux 12 between two isolines equals to the difference between the
This has been quite a lot of algebraic work, and we would respective values of ψ. In the case of a two-dimensional fluid
like to be sure that we didn’t do any mistakes. The ab- flow, the flow flux is the surface area which flows through a
sence of mistakes can be easily checked by calculating V = cross-section within a unit time; in the case of electrical cur-
V1 + V5 : we need to get the same result! Let us note that rent, the flow flux is the total electrical current which flows
V1 = −i1 · 1 Ω = 72 41
Ω · I and V5 = −i2 · 5 Ω = 35
18 Ω · I; therefore,
through a cross-section. In the case of planar circuits, current
R = ( 72 + 18 )Ω = 181
41 35
72 Ω. flows only along the wires and hence, the streamfunction is con-
stant between the wires, and jumps at the position of wires13 .
idea 15: The number of unknowns and the number of linear
equations can be reduced by using the method of poten- idea 16: For planar circuits, instead of the method of loop
tials, in which case the second set of Kirchoff’s equations is currents, the method of streamfunction can be used. Each
automatically satisfied. The first step is assigning to each node face (the empty area between wires) of the circuit is assigned a
(connection point of wires) a potential φn (where the index n streamfunction value: i-th face is assigned a value ψi which is
refers to the n-th node). The second step is expressing the first to be found using the Kirchoff’s voltage laws. The current in a
set of Kirchoff’s equations in terms of the potentials using the wire separating i-th and j-th face is found as Iij = ψi − ψj ; the
Ohm’s law, and solving the obtained system of equations. sign of Iij here is chosen so that if we move along the direction
of Iij , the i-th face remains to our left hand.
pr 14. Solve the previous problem using the method of Now we can also draw an important conclusion regarding
potentials. the total number of linearly independent loops for planar cir-
Similarly to what we did before, we assume that the circuit cuits:
leads are attached to a battery. The reference level for po- fact 8: The number of degrees of freedom (and hence, the
tential can be chosen arbitrarily, and thus it is convenient to maximal number of linearly independent current loops) for the
equate the potential of one output lead to zero (let it be the current distribution in a planar circuit equals to the number of
left one); then the second lead’s potential equals to the battery faces in the corresponding graph (excluding the infinite face).
voltage V . There are two more nodes on the circuit, let the
Indeed, the current distribution can be fully described by the
respective potentials be φ1 (the upper one), and φ2 . The cur-
streamfunction values at the faces, and the infinite face can be
rent to the upper node from the right wire I3 = (V − φ1 )/3Ω;
postulated to have ψ = 0, so the number of degrees of freedom
the current from the upper node to the left wire I2 = φ1 /2Ω;
equals to the number of finite faces of the graph.
the current from the upper node downwards I4 = (φ1 −φ2 )/4Ω.
In order to illustrate the idea 16, let us consider, again, the
According to the Kirchoff’s law for currents, I3 = I2 +I4 , hence
problem 13; the unknown values of streamfunction (ψ1 and ψ2 )
(V − φ1 )/3Ω = φ1 /2Ω + (φ1 − φ2 )/4Ω ⇒
are marked together with the corresponding currents in the fig-
13φ1 − 3φ2 = 4V.
ure below (as compared to problem 13, the resistance values
Similarly, for the lower node, I4 + I1 = I5 , where I1 =
are changed). The set of Kirchoff’s voltage laws will be very
(V − φ2 )/1Ω and I5 = φ2 /5Ω. This leads us to
similar to what we obtained with the method of loop currents,
(φ1 − φ2 )/4Ω + (V − φ2 )/1Ω = φ2 /5Ω ⇒
so we skip that part of the solution.
−5φ1 + 29φ2 = 20V.
88
Solving this linear system of equations results in φ1 = 181 V and
140
φ1 = 181 V ; total current can be calculated using the Kirchoff’s
law for the leftmost node, I = I2 + I5 = 44+28 181 V /Ω, hence
R = V /I = 181 72 Ω. The control of this result can be done by
calculating the total current on the basis of the rightmost node.
This example shows that the difficulty level of the both idea 17: Due to the symmetry of the Kirchoff’s voltage law
methods (c.f. ideas 13 and 15) is approximately the same, so and current law, there is a duality between electrical currents
the choice is typically based on personal preferences. In the and voltages14 which means that we can interchange voltages
case of loop currents, selecting a good set of linearly independ- and currents, and we’ll obtain, as a result, a very similar prob-
ent loops may seem as an additional step in the solution, but lem. It works out most conveniently in the case of planar cir-
12 For fluid flow, the fow flux is the amount of fluid carried through a cross-section per unit time; in the case of electric current, it is just the total

current.
13 This statement is valid for non-planar circuits, as well, but then there will be faces for which the streamfunction value is not a free parameter (is

defined by the streamfunction values of the neighbouring faces) so that the application of the idea 16 would be less straightforward
14 A. Russell 1904.

— page 8 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
cuits in which case voltage values transfer into streamfunction cuit so that on each new circuit segment, there is a resistor the
values, and vice versa; the circuit itself is transferred into its “resistance” of which is equal to the conductance of the corres-
dual circuit (see below). Most often, we transfer one circuit ponding old circuit segment (if the old circuit segment had a
problem just into another circuit problem, but in the case of battery, use now a current source). Calculate the “resistance”
self-dual circuits (when the circuit is identical to its dual cir- of this dual circuit and take reciprocal to obtain the resistance
cuit), the symmetry may prove to be very useful15 . of the original circuit.
Note that the bridge connection considered above is topolo-
Let us apply the concept of duality to the bridge-connection
gically self-dual, because its dual circuit has the same structure
drawn above; the dual circuit is obtained by putting one node
— is made of five “resistors” forming a bridge. Furthermore, the
inside each face of the original circuit, and connecting the new
numerical values of the resistors are such that if we multiply
nodes with wires so that each old wire is crossed by exactly one
all the “resistances” by the same factor (4 Ω2 ), the dual circuit
new wire, see below.
becomes identical to the original circuit; this is the property
which makes it a self-dual circuit.

pr 15. Find the resistance of the bridge connection illus-

trating the idea 16; use the idea 17 and the self-duality of this
bridge connection.
The simple bridge connection considered above is self-dual,
For our original circuit, we had a battery which kept the
because its dual circuit is also a bridge connection of the same
voltage between the two ports equal to E, and in our new cir-
type.
cuit, we have a current source which keeps the streamfunction
difference between the top and bottom nodes equal to I. When idea 18: Infinite periodic chains of electronic components
using the node potential method with our old circuit, we had (resistors, capacitors etc) can be studied by making use of the
each node ascribed a potential; now we have each node ascribed self-similarity of the chain: removal of the first period does not
a streamfunction value. For the old circuit, the unknown po- change its properties.
tentials φi were found from the Kirchoff’s current laws written
for each node; for j-th
∑node pr 16. [IPhO-196716 ] Determine the resistance of the infin-
(φi − φj )/Rij = 0, ite periodic circuit
i R1
where Rij is the resistance between the i-th and j-th nodes,
and the sum is taken over all those nodes which are directly
R2
connected to the j-th node. For the new circuit, the unknown
streamfunction values ψi are to be found from the Kirchoff’s
voltage laws written ∑ for each new node; for j-th node
According to the idea 18, we “cut off” the first period of the
(ψi − ψj )Rij = 0,
infinite chain (painted in orange in the figure below); the re-
i
where Rij is the resistance on that wire (of the old circuit) maining part (blue) is equivalent to the original circuit of (yet
which is intersected by a new circuit segment connecting the unknown) resistance R. Because of that, we can write equality
i-th and j-th nodes, and the sum is taken over all those nodes RR2
R = R1 + ,
which are directly connected to the j-th node. We would ob- R + R2
which can be solved with respect to R.
tain exactly the same set of equations if we were considering
the new circuit as a usual resistor network with resistances be-
ing equal to the conductances of the old circuit (so that in the
figure above, the “resistance” of the “resistor” a is 1 Ω−1 , the
“resistance” of the “resistor” b is 0.25 Ω−1 , etc). This procedure
assumes also that the new “voltage” applied between the top This idea can be combined with other ideas — for the next
and bottom nodes of the new circuit is I, and the new total problem, together with the idea 10.
“current” (the sum of “currents” in “resistors” a and b) equals
to the voltage difference E between the two ports of the old pr 17. Determine the electromotive force and internal res-
circuit. Therefore, the “resistance” R⋆ of our dual circuit is istance of the following system of batteries.
R
expressed as
I 1 E
R⋆ = ≡ ,
E R r
where R is the total resistance of the old circuit (everything is
OK with the dimensionalities as both R⋆ and the component-
resistors of our dual circuit are measured in Ω−1 ).
To sum up, the procedure of using dual circuits to calculate idea 19: As soon as you detect a symmetry in a problem,
the resistance of a given circuit is as follows: build a dual cir- try exploit it.
15 L.A. Zadeh 1951
16 At the IPhO-1967, all resistors were equal to r

— page 9 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
The next problem can be solved exploiting its symmetry, in is I/2, which corresponds to the voltage V = RI/2. Therefore,
conjunction with the idea 14. the resistance r = V /I = R/2.
It appears that such a symmetrization technique can be also
pr 18. Determine the resistance between opposing corners applied to finite lattices, see the next problem.
of a cube, the edges of which are made of wire, see figure; the
resistance of one edge is 1 Ω pr 21. Determine the resistance between two neighbouring
vertices of a dodecahedron (see figure), the edges of which are
made of wire; the resistance of each edge is R.

Sometimes it is convenient to use this idea in conjunction idea 23: Sometimes the problem symmetrization can be
with specific algorithms how to reduce a circuit to a combina- achieved by introducing fictitious negative resistances: there
tion of parallel and series connections. is no problem with applying Kirchoff’s laws to negative resist-
ances.17 In particular, R and −R in parallel correspond to an
idea 20: Node-merging method: if two nodes have equal infinite resistance, and in series — to a zero resistance.
potential (e.g. due to symmetry), they can be short-circuited.
pr 22. Determine the resistance between two neighbouring
idea 21: Edge-splitting method: a resistor between nodes vertices A and B of a dodecahedron, the edges of which are
A and B can be represented as a parallel connection of two made of wire; the resistance of each edge is R, except for the
resistors, and the node A can be split into two nodes, if the edge connecting the vertices A and B, which is cut off.
potentials of the new nodes A′ and A′′ will be equal.
idea 24: If there are nonlinear resistors included into a cir-
These ideas are illustrated with the following problem.
cuit which are characterized with a nonlinear current-voltage
pr 19. An hexagon ABCDEF with six “spokes” (connect- dependence I(V ) then the current through the nonlinear ele-
ing its centre O with the vertices) is made of 12 pieces of ment can be found graphically: I–V -dependence can be also
wire, each having a electrical resistance R. Find the resistance expressed using the Kirchoff’s laws, in simpler cases this will
between the vertices A and O using methods 20 and 21. be a linear law V = U0 − Ir. Then, the solution will be the
intersection point of the two curves, U0 − Ir and I(V ).
idea 22: Non-symmetric problems can be sometimes conver- Solutions (intersection points) in the negative differential res-
ted into symmetric ones using superposition principle. istance range (where Rdiff ≡ dV
dI < 0) can be unstable; stability
analysis requires knowledge about inductors, and so we post-
pr 20. Determine the resistance between two neighbouring pone it accordingly.
vertices A and B of an infinite square lattice assuming that the fact 9: If there is more than one stable solution then the
edges of the lattice are made of wire, and the resistance of each question of which solution is actually realized is resolved based
edge is R. on the history (e.g. if the voltage applied to the circuit has
This problem does not possess enough symmetry to be solved been increased or decreased) because internally, nonlinear ele-
immediately: if we drive a current I into the vertex A and draw ments obey inertia (for instance, the density of charge carriers
it out from the vertex B, the geometry of the problem would can change fast, but not instantaneously) and it will not jump
possesses only a mirror symmetry, which is not sufficient for from one equilibrium state to another without a good reason
concluding how the current I is distributed between the four (such as a loss of stability or disappearance of the current solu-
wires connected to the input vertex. However, it is possible to tion branch).
construct a rotationally symmetric problem: suppose that the Let us illustrate the idea 24 on the basis of a tunnel diode
current I is driven into a vertex A and taken out symmetric- connected via a resistor to a battery of variable electromotive
ally at infinitely remote edges of the mesh. Then it is clear that force E.
the current I is distributed equally between the four outgoing
wires: the current in each of them is I/4. Similarly, we can
drive the current in a rotationally symmetric way at infinity,
and draw it out from the vertex B. The superposition of these
two symmetric configurations provides exactly what we need:
the current is driven into A and drawn out from B; at the
infinitely remote edges, the current cancels out. In the wire
connecting A and B, the both superposition components have
the same direction and are equal to I/4, hence the net current
17 Care should be taken only with oscillatory circuits which include also inductors and capacitors: positive resistance corresponds to a dissipation
(decay of oscillations), negative resistance can cause instability (growth of oscillations).

— page 10 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
If E is small, there is only one intersection point (the leftmost
blue dot in figure); if E is increased, the intersection point
moves up, and even though at a certain moment, there are
more than one intersection points, the real current and voltage
correspond to the leftmost intersection point as a continuous
evolution of the original solution. When E is further increased,
at a certain moment, this solution disappears and the solu-
tion is forced to jump rightwards as shown in figure by blue
almost horizontal arrow. Now, if E starts decreasing, the inter-
section point depicting the solution moves continuously down
and during the period when there are three intersection points,
the rightmost one will correspond to the real solution. If E is idea 26: In the case of a small variation of the voltage
further decreased, that intersection point disappears, and the Ṽ ≡ V − V0 on a nonlinear element, and a small current vari-
solution jumps back to the only remaining intersection point. ation I ≡ I − I0 through it, one can linearize the V − I curve as
˜
˜ where Rdiff = dV is referred to as the differential
Ṽ = Rdiff I, dI
The phenomenon when the system state depends on its his-
resistance. Here, V0 and I0 are the unperturbed (equilibrium)
tory is called hysteresis. Hysteresis will typically appear if the
values of the voltage and current. Then, the total voltage on
system can have more than one internal states; a simple ex- ˜ Now, if we write down
the nonlinear element V = V0 + Rdiff I.
ample is provided by the following problem. ˜ in
the Kirhoff’s voltage law in terms of the current variation I,
˜ we
addition to the “Ohm’s law” for the voltage variation Rdiff I,
pr 23. [EstOPhC-2009] Element X in the circuit below has have additional constant term V0 which can be inerpreted as
a resistance RX which depends on the voltage VX on it: for an effective electromotive force. On any linear resistor R, the
VX ≤ 1 V, RX = 1 Ω, and for VX > 1 V, RX = 2 Ω. Three such voltage is also a sum of a constant term I0 R and the variation
elements are connected with an ideal ammeter as shown below; term RI. ˜ All the unperturbed constant terms together must
the voltage on the leads of the circuit varies in time as shown cancel out from the Kirchoff’s voltage law because V0 and I0
in the graph. Plot the reading of the ammeter as a function of were assumed to be valid solutions of the Kirchoff’s laws. In-
time. deed, if we put all the perturbations equal to zero then I˜ = 0
and Ṽ = 0 should provide a solution to the Kirchoff’s laws,
hence all the constant must cancel out.
To sum up, instead of studying voltages and currents, we study
the perturbations Ṽ and I˜ of these quantities; the effective cir-
cuit describing the perturbed values is obtained by removing all
the unperturbed voltage and current sources (such as batteries
of constant electromotive force), and by substituting nonlinear
This problem is otherwise quite simple, but for certain
elements with their differential resistances. NB! The differen-
voltages, the state of circuit’s components will depend on the
tial resistances of a nonlinear element depends on the current;
history. Here a typical mistake is solving the problem cor-
we need to use its unperturbed value.
rectly for the first 10 seconds, and then assuming a mirror-
symmetrical graph for the current. How to avoid such mis- The usefulness of this idea is demonstrated by the following
takes? The first and the best way is to always avoid rushed problem.
extrapolations (in the given case — mirror-extrapolation of the
first 10 seconds to the next 10 seconds). Another way to figure
pr 25. [EstFin-200318 ] In the figure below, the circuit of a
simple tunnel-diode-based amplifier is given. Find the ampli-
out that things are not as simple as they seem is formulated as
fication factor for small-amplitude input signals using the fol-
an idea.
lowing values: R = 10 Ω, E = 0.25 V.
idea 25: Try to think, what was the reasoning of the author
of the problem. In particular, if an Olympiad problem has
seemingly similar questions, there is typically some essential
difference. (As an exception, this is not a physical idea.)

In the given case, would it have been interesting to ask about

the next 10 seconds if you can obtain the result by a simple
mirror-extrapolation?
Returning to the idea 24, a simple illustration is provided
by the next problem. idea 27: It is possible to obtain upper and lower limits for
the resistance of a circuit using the following theorems.
pr 24. Find the current in the circuit given below; the I(V ) (I) For an arbitrary circuit which consists of resistors and has
dependence of the diode is shown in graph. two leads, A and B, if a current I is driven into the lead A
18 Only a part of the full problem

— page 11 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
and out from the lead B, the current distributes between the resistance for reverse current), you need to consider separately
resistors of the circuit so as to minimize the overall power dis- two cases: (a) assume that there is a forward current and the
sipation. In other words, the power dissipation of the actual diode is open, hence it can be substituted by a wire; (b) assume
current distribution is always smaller as compared to any ficti- that there is a reverse current, and hence, it can be “cut off”.
tious current distribution satisfying only the Kirchoff’s law of Depending on the problem, it may be apparent, which option
currents. 19 is to be used, or you may need to use the calculation results to
21
(II) For the same circuit, if there is a voltage drop V between verify, which assumption was valid .
the leads A and B, the voltage distributes between the nodes of
the circuit so as to minimize the overall power dissipation. In pr 28. How many times will change the power dissipation
other words, the power dissipation of the actual voltage distri- in the resistor A when the polarity of the battery is reversed?
bution is always smaller as compared to any fictitious voltage All the resistors have equal resistance. Diodes are ideal.
distribution violating the Kirchoff’s law of currents. 20
Particular conclusions of these theorems are: cutting off a wire
will increase the resistance, and short-circuiting a wire will de-
crease the resistance. Indeed, if we cut a wire, we disable the
respective current and this leads to what can be considered as
a fictitious current distribution, which has an increased overall
power dissipation I 2 R, and hence, an increased net resistance idea 29: Non-ideal diodes which are approximated with
R. Similarly, short-circuiting makes it possible for the currentan idealized V − I curve with a non-zero opening voltage Vc
to jump between the nodes — something which was impossible (there is no current for V < Vc , and for any forward current,
originally and violates the Kirchoff’s laws of currents for the V = Vc ) ca be also handled according to the idea 28; the only
original circuit. Hence, the power dissipation in the modified difference is that for forward currents, diode needs to be re-
circuit V 2 /R is increased, and consequently, the resistance R placed by a battery with emf. E = Vc . Additionally, the power
is decreased. dissipation on the diode is calculated in the same way as the
pr 26. There is an octagon all diagonals of which are res- work done by a battery: dissipation power is Vc I, and the dis-
istors of equal resistance R; the sides of the octagon are made sipated heat — Vc ∆Q, where ∆Q is the charge passing through
of an insulating material. Find lower and upper bounds for the the diode.
resistance between two neighbouring nodes of such an octagon. Note that this idea can be made even more general: if we ap-
The solution here is as follows. First, we cut off several resist- proximate a certain nonlinear V − I curve with a curve which
ances, and leave only those which are shown in the left figure consists of n pieces of straight line segments (piece-wise linear
below. The resistance of the left circuit is 2R R graph) then we need to consider separately n cases; for each
4 = 2 . Further, we
short-circuit six nodes as shown in the right figure; the resist- case, the non-linear element can be substituted by a battery
ance is 2 R5 = 0.4R. So, we can conclude that 0.4R ≤ r ≤ 0.5R. dI ≡ r of
the internal resistance of which is equal to the slope dV
Since the wires we cut off did have current, and the nodes which the corresponding straight line segment, and the electromotive
we connected with wires did have a voltage difference, the new force is equal to the V -intercept of that line. Instead of a bat-
current- and voltage distributions are sub-optimal and we can tery, sometimes it is better to use a current source connected
exclude equality signs: 0.4R < R < 0.5R in parallel to the internal resistance r and supplying a current
equal to the I-intercept of the graph segment; use this method
in particular when r = ∞.

pr 29. [EstOPhC-2012] Find the power dissipation on each

of the diodes in the figure below. These diodes open at the
forward voltage V0 = 1.0 V. It can be assumed that the diode
voltage remains equal to V0 for any forward current, and that
for voltages less than V0 , there is no current through the diode.
pr 27. Improve the upper bound r < 0.5R for the previ- The values of the resistances and of the electromotive force are
given in the figure.
ous problem (do not “cut off” as many wires as we did before),
as well as the lower bound (short-circuit a lesser number of
nodes).
Finally, let us consider circuits including ideal diodes.
idea 28: If there are ideal diodes included into the circuit
(which have zero resistance for forward current, and infinite
19 Proof is provided in the appendix 3, page 15.
20 Proof is also provided in the appendix 3.
21 This is similar to the problems with dry friction between solid bodies when you consider separately the cases when (a) the bodies splip and there

is a friction force defined by the kinetic coefficient of friction, and (b) the bodies don’t slip.

— page 12 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
Problems involving ideas 1–28 pr 37.
The drawing below depicts octahedron made from
pr 30. Determine the maximal power which can be dissip- wire; the number near to each edge shows the resistance of the
ated on a load connected to the leads of the circuit. corresponding wire in ohms. The resistance of the wires con-
necting the ammeters are negligibly small. Find the readings
of the ammeters.

pr 31. Find the current through the diode in the circuit

given below; for the diode, use the I(V ) dependence from the
problem 24.
pr 38. In the figure, all three voltmeters are identical,
and all three resistors are identical. The first voltmeter shows
V1 = 10 V, the third — V3 = 8 V. What does show the second?

pr 32. For an overcurrent protection, there are two fuses

connected in parallel: fuse A has resistance RA = 1 Ω and
maximal current (by which it melts) IAmax = 1 A; fuse B
has resistance RB = 2 Ω and maximal current (by which it
pr 39. Determine the potential of the lead A. (Note that
melts) IBmax = 1.2 A. What is the maximal total current for
the ground potential is always assumed to be 0.)
such a system of fuses? What is the total current when the
fuse B is substituted with a fuse C which has RC = 2 Ω and
ICmax = 1.7 A?

pr 33. The two voltmeters in the circuit below are identical;

their readings are V1 = 30 V and V2 = 20 V. The reading of
the ammeter is I = 750 µA. All the five resistors have equal
resistance R; find the numerical value of R. pr 40. In the circuit below, the “device” takes the read-
ing of the ammeter and adjusts the resistance of the rheo-
stat so that the ammeter reading becomes zero. Find the
voltage on the resistance R3 . It is known that V = 5 V,
R1 = 10 Ω,R2 = 1 kΩ, R3 = 100 kΩ, R4 = 4.99 kΩ.

pr 34. Assuming that the resistance of a light bulb’s wire is

proportional to its temperature T and its heat radiation power
is proportional to T 4 , find the power law exponent of its V –I
dependence. Neglect the heat conductivity and assume that T
is much higher than the room temperature.

pr 35. [EstPhO-1999] All the resistors have equal resist- pr 41. Eight identical lamps of nominal voltage V = 4 V
ance R = 1 Ω Ammeters and the battery are ideal, E = 1 V.
and nominal current I = 0.25 A are connected to a battery via
Determine the readings of all the ammeters.
a resistor as shown in figure. The resistor is such that the lamps
will operate at the nominal regime (with nominal voltage and
current). One of the lamp burns out (the lamp is essentially re-
moved). How many times does change the overall power which
is dissipated by the lamps? (The power dissipation on the res-
istor is NOT included.) Neglect the dependence of the lamp
resistances on the temperature.
pr 36. Find the reading of the ammeter in the circuit below.
1Ω 2Ω 3Ω
2Ω

2Ω

A
pr 42. The figure below depicts a cube, the edges of which
3Ω 2Ω 1Ω (blue lines) are made of a resistive wire, so that the resistance
8V of each edge is R = 1 kΩ. The ammeters are connected with
— page 13 —
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
copper wires of negligible resistance to the vertices of the cube. are resistors of equal resistance R. What is the resistance
The battery voltage is E = 9 V; the wires make electrical con- between two neighbouring nodes of the n-gon?
tact only at the vertices of the cube. Find the readings of the
ammeters.
pr 43. Four ammeters with identical internal resistances r
and a resistor of resistance R are connected to a current source
as shown in figure. It is known that the reading of the ammeter
A1 is I1 = 3 A and the reading of the ammeter A2 is I2 = 5 A.
Determine the ratio of the resistances R/r.
pr 44. How many times does change the current through
the battery if the polarity of the battery is reversed? All the
resistors are identical, diodes are ideal and internal resistance
of the battery is negligible.
pr 45. Determine the resistance between two neighbouring
nodes A and B of an infinite cubic lattice assuming that the
edges of the lattice are made of wire, and the resistance of each pr 53.
edge is R.
pr 46. There is an infinite honeycomb lattice; the edges of
the lattice are made of wire, and the resistance of each edge is
R. Let us denote two neighbouring vertices of a vertex B by A
and C. Determine the resistance between A and C.
pr 47. There is an infinite triangular lattice; the edges of
the lattice are made of wire, and the resistance of each edge is
R. Let us denote the corners of a triangular lattice face by A,
B, and C. The wire connecting B and C is cut off. Determine
the resistance between A and B.
pr 48. There is a n-gon all sides and all diagonals of which
— page 14 —
pr 49. There is a decagon all sides and all diagonals of
which are resistors of equal resistance R; let A and C denote
the two neighbours of a vertex B, and let D be a vertex which
is not neighbouring any of the three mentioned vertices. The
wires corresponding to the sides AB and BC are cut off. De-
termine the resistance between A and D.
pr 50. There is an octagon all diagonals of which are res-
istors of equal resistance R; the sides of the octagon are made
of an insulating material. Find lower and upper bounds for
the resistance between two opposing nodes of such an octagon
without exactly calculating its value. Verify the result by cal-
culating this resistance also exactly.
pr 51. Find the resistance between the terminals A and B
for the infinite chain shown below. The resistances are as shown
and increase by a factor of two for each consecutive link.
pr 52. Find the voltage between the terminals A and B for
the infinite chain shown below.
R E R E R E
r r r
Which unequalities must be valid for the resist-
ance between two neighbouring vertices A and B of an infinite
square lattice, if the edges of the lattice were made of wire so
that the resistance of each edge was R, but some parts of the
lattice have been damaged: some wires have been broken and
some of the broken wires have been replaced by copper wires of
negligible resistance. However, within the distance of two edge
lengths from the wire AB, the lattice is completely intact (this
includes 13 wires parallel to AB, and 12 wires perpendicular
to it).
pr 54. A wheel circuit is a circuit which can be drawn as
a regular n-gon so that the rim of the “wheel” is formed by
n resitors of resistance R connecting neighboring vertices of
the n-gon, and the “spokes” of the “wheel” are formed by n
resistors of resistance r connecting the centre of the “wheel”
with each of the vertices. Let R1 be the resistance between
two neighbouring vertices of such “wheel”, and R2 — the res-
istance between one of the vertices and the centre. Express R1
in terms of R2 , R and r (without using n).
appendix 1: Proof of the Y − ∆ circuit theorem
Two circuits, one with ports A, B, C , and the other with ports
a, b, c are equivalent if their response to external forcing is
identical. This means that if we drive a current I into the
 1. CIRCUITS WITH RESISTORS, BATTERIES, AMMETERS AND VOLTMETERS
lead A (or into a) and drive it out from B (or from b), Let us denote the potential of the j -th node of the circuit in
the lead voltages must be pair-wise equal: VAB = Vab , and the rst case by φj (j = 1, . . . n), and in the second case by
VAC = Vac . Due to the linear nature of the Kircho 's and ψj ; the rst four nodes (j = 1, 2, 3, 4) are the four input leads
Ohm's laws, we know that all these voltages are proportional A,B ,C , and D. Due to Ohm's law, for any pair of nodes (i, j )
to the current: VAB = IRAB , Vab = IRab , VAC = IRACB , and directly connected by a wire (over a resistor), there is equality
Vac = IRacb , where RAB is the resistance between A and B , (φj − φi )/Iji = (ψj − ψi )/Jji ,
RAC is the A − C resistance. Proportionality coecients Rijk where Iji and Jji are the wire's currents in the rst and the
(i, j, k ∈ {A, B, C, a, b, c}) relate the i − k -voltage to the i − j
second case, respectively. This can be rewritten as
current. Equivalence of the circuits means that
(φj − φi )Jji = (ψj − ψi )Iji ;
RAB = Rab , RAC = Rac , RBC = Rbc , (3) Summing this equality over all the node pairs we obtain
∑ ∑ ∑ ∑ ∑ ∑ ∑ ∑
RACB = Rabc , RABC = Rabc , RBCA = Rbca , etc. (4) φj Jji − φi Jji = ψj Iji − ψi Iji .
These 9 equations represent necessary and sucient condi- j i i j j i i j
Note that due to the Kircho 's current law, for any j ̸= 3, 4,
tions for the equivalence between an arbitrary 3-lead circuit ∑ ∑
J = 0 ; similarly, for j ̸
= 1, 2, I = 0 ; for i ̸= 3, 4,
(with leads A, B, C ) and a Y -connection (or a ∆-connection) ∑i ji ∑ i ji ∑
J = 0 ; for i ̸
= 1, 2 I = 0 . Further, J =
of three resistors. Nine equations seems to be too many for ∑
j ji ∑ ∑ j ji ∑ ∑ i 4i
− i J3i = I = − i I1i = − j Jj4 = j Jj4 =
determining the values of the three resistances. Luckily, it ∑ ∑ i 2i
appears that if the rst three equations (3) are satised, all
− j Ij2 = j I1i = I . Therefore, the above equality sim-
plies into
the rest are satised automatically. It is (relatively) easy to
verify via direct arithmetical calculations that for any triplet
2(φ4 − φ3 )I = 2(ψ2 − ψ1 )I.
of resistances RAB , RAC , and RBC , these three equations
Finally, as φ4 − φ3 = V and ψ2 − ψ1 = U , we arrive at V = U,
QED.
can always be solved with respect to the three resistances
of the Y- ∆-connection, and as long as the triangle in-
or There is one quite difficult problem which can be solved in
equalities of the form RAB ≤ RAC + RBC are satised, the a somewhat similar manner to how we proved the reciprocity
three resistances (of Y or ∆-connection) are non-negative. In- theorem. We might also try to formulate a respective hint.
deed, for Y -connection, Rab = ra + rb , Rac = ra + rc , and
idea 30: Sometimes it is possible to combine the equations
Rbc = rb + rc ; if we put these expressions into Eqns. (3), we
of a long system of equations so that almost everything cancels
obtain ra = (RAB + RAC − RBC ), and analogous expressions
1
2 out, leaving only few non-zero terms.
for rb and rc . The calculations for ∆-connection are analogous
[alternatively, Eq. (2) can be used to nd the ∆-connection-
pr 55. m identical resistors of resitance R are connected
resistances from ra , rb and rc ]. It appears that the triangle
in an arbitrary way; though, none of the resistors is short-
inequalities are, indeed, satised for any three-lead circuit, see
circuited (there is no direct wire connection between the two
appendix 4 below.
leads of a resistor), and all the resistors are connected together
What is left to do is to show that the equations (4) dealing (the resistance between any pair of nodes is finite). Overall,
with the three-lead-resistances follow from the equations (3). this resistor network has n nodes For each resistor, the resist-
First, from the Kircho 's voltage law we can conclude that ance between the adjacent nodes (to which it is connected) is
RACB + RCAB = RAC (and similar expressions for RCBA and determined, and the results are added up. Show that this sum
RBAC ). Indeed, if a current I is driven into A and taken of m resistances equals always to (n − 1)R.23
out from C , we can express the voltages as VAC = IRAC ,
VAB = IRACB , and VBC = IRCAB ; due to the Kircho 's appendix 3: Proof of the dissipation minimum the-
voltage law, VAC = VAB + VBC , hence RAC = RACB + RCAB .
Second, the equality RACB = RABC (and similar expressions
orem
In order to prove the rst part (when the Kircho 's voltage
for RCBA and RBAC ) follows directly from the reciprocity the-
law remains satised), consider the power dissipation
orem
22
; however, this theorem is more tricky to prove, see ap- ∑
pendix 2.
P = (φi − φj )2 /Rij ,
ij
where φi is the potential of the i-th node (for a ctitious po-
appendix 2: Proof of the reciprocity theorem tential distribution), and the sum is taken over all such pairs
The theorem states that if we have a four-lead system of resist- of nodes (i, j) which are directly connected via a resistor Rij .
ors, the leads being denoted by A, B ,C ,and D, and we consider If the potential of the i-th node is changed by a small incre-
two cases, (i) current I is driven into A and out from B , and (ii) ment ∆φi (while keeping the other potentials intact), the total
the same current I is driven into C and out from D , then the power dissipation is changed by
∑
voltage VCD induced between C and D in the rst case equals ∆P = [2∆φi (φi − φj ) + ∆φ2i ]/Rij .
to the voltage UAB between A and B in the second case (the j
equality required for the proof of the Y − ∆ circuit theorem The last term here can be neglected for very small potential
corresponds to the particular case when D coincides with A). increments, and we can denote (φi − φj )/Rij ≡ Iij : this is
22 H.A. Lorentz, 1896
23 The theorem can be generalized: the resistors are different, each term in the sum is divided by the resistance of the respective resistor, the sum
equals to n − 1.

— page 15 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
the current owing from i-thj -th node. So, at the limit of Indeed, there is no change in the region φ ≤ V1 as compared
to
innitesimally small increments (∆φi → dφi ), we obtain to the case (i), which means that the Kircho 's current law is
∑
dP = 2dφi Iij . satised there; in the region which had originally φ > V1 , the

j potential is now constant, hence there is no current, hence the

∑
Kircho 's current law is also satised. All the threshold points
If j Iij > 0 then the power dissipation can be decreased by
increasing the potential φi (dφi < 0 leads to dP > 0); simil- φ = V1 are connected by a wire which directs all the total
∑
arly, if j Iij < 0, we can take dφ∑i > 0. So, the dissipation current I0 into the lead B so that the Kircho 's current law
minimum can take place only for j Iij = 0 , i.e. when the remains still satised. For such a modied circuit, the power
∑
dissipation is V1 I0 ; due to the power dissipation theorem, this
Kircho 's current law is satised. Further, if
∑ j Iij = 0 then
2
∆P = j ∆φi /Rij > 0, i.e. we have, indeed, a minimum. is larger or equal to the actual dissipation V1 I1 , hence I1 ≤ I0 .

The second half of the theorem is proved in the same way as Finally, let us introduce case (iii): we disconnect A from the

the rst half: we assume that there is a ctitious current distri- ground, and connect B and C to the voltage sources which

bution which satises the Kircho 's current law and hence, can provide potentials V1 and V2 , respectively. Analogously to
∑
be represented as a sum of loop currents ν what we did before, we can show that the emerging current
Iµ = ν Iνµ , where
enumerates the loops and Iνµ µ-th I2 ≤ I0 (through B and C ). Now we can write inequalities for
represents the current in a
wire contributed by the ν -th loop. Note that Iνµ = 0 if the µ-th the resistances:
V1 V1 V0 − V1 V0 − V1
wire does not belong to the ν -th loop; otherwise Iνµ = ±iν  RAB = ≥ , RBC = ≥ ;
I1 I0 I1 I0
the loop current has the same magnitude everywhere, and the
if we sum up these inequalities we obtain
sign of the contribution depends on which current direction is V1 V0 − V1 V0
assumed to be positive for the given wire. Let us assume that
RAB + RBC ≥ + = = RAC ,
I0 I0 I1
all the contributions of the ν -th loop current are positive (if not, QED.
we can re-dene the positive directions of the relevant wires).
Note that owing to the triangle inequalities, the resistance can
Then
( )2 be used to dene the distance between two circuit points (or
∑ ∑
between two points of a continuous conducting medium); then,
P = Rµ Iνµ ,
µ ν instead of meters, the distance will be characterized in ohms.
and for an increment diν ν -th loop current,
of the
∑
dP = diν · 2 Iµ Rµ ;
µ
2 Circuits including capacitors and inductances
∑
for a minimum, we need to have µ Iµ Rµ In order to be able to solve circuits involving capacitors and
= 0, which is the
Kircho 's voltage law for the ν -th loop. inductances, the knowledge of several facts is needed. Some
facts will be provided here without proof; more insight will be
appendix 4: Proof of the triangle inequality given in the section “Electromagnetism’.
For a three-lead circuit, let us ground the lead A. (i) First, let
Let us begin with capacitors. A capacitor can be thought
us connect the lead C to a voltage source providing a potential
of as consisting of two parallel conducting sheets (plates) which
V0 > 0 while keeping the lead B disconnected externally; this are very close to each other, and separated by a thin dielectric
gives rise to a certain current I0 which ows from C through
(insulating) layer 24 . We mentioned in the introduction of Sec-
the circuit and through the lead A into the grounding wire; this
tion 1 that typically, we can neglect charges on the wires; this
will also induce a certain potential V1 on the lead B ; apparently
is because any non-negligible charge on wires would give rise
0 ≤ V1 ≤ V0 . to a huge electric field, and hence, to a huge voltage. How-
ever, situation is different if we have two parallel conducting
(ii) Second, let us disconnect C from the voltage source and
plates: if these two plates have equal and opposite charges,
connect B to a voltage source providing the same potential V1
so that the system as a whole is electrically neutral, the huge
what it had previously; this gives rise to a current I1 via A
electric field is constrained into the narrow layer between the
and B , and an induced potential V2 on the lead C . Now, let
plates, hence the voltage (the product of the layer thickness
us apply the minimal power dissipation theorem. For case (ii),
and field strength) can remain moderate. Typically 25 , the
we construct a ctitious potential distribution based on the po-
voltage between the plates is proportional to the charge sitting
tential distribution of case (i): all these internal circuit points
on one of the plates. Since a capacitor is electrically neutral as
which have potential φ less than V1 will have the same poten-
a whole, the Kirchoff’s current law remains valid for capacitors,
tial which they had previously, and all those internal circuit
as well: current flowing along a wire to one plate (increasing
points which have φ ≥ V1 will have potential V1 (if a certain
the charge there) equals to the current flowing from the other
resistor extends over the threshold potential V1 , we imagine
plate (decreasing the charge there) along another wire.
the resistor as if being made of a resistive wire and cut this
wire into two segments at the point where φ = V1 ). Such a fact 10: Capacitance is defined as
ctitious potential distribution would be the real potential dis- C = q/V,
tribution of a moded circuit for which all the circuit points where q is the charge on the plates of the capacitor (one plate
with potential φ = V1 are connected via a wire to the lead B. has +q, the other one −q) and V is the potential difference
24 There are different types of capacitors with different shapes, but such details are not important for the time being.
25 if the inter-plate distance and the dielectric permeability of the insulator remain constant

— page 16 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
between the plates of the capacitor. Unless otherwise men- where I is the current flowing through the inductor, and Φ is
tioned, C is independent of the applied voltage V . the magnetic flux created by that current passing through the
inductor itself 26 . Unless otherwise noted, the inductance may
fact 11: The energy of a charged capacitor is be assumed to be independent of current 27 .
2
W = CV /2.
fact 15: Electromotive force created in a circuit due to
Indeed, consider a charging of a capacitor. If a charge dq crosses changing magnetic field
a potential difference V , electrical work dA = V I · dt = V · dq
∫V E = −dΦ/dt,
needs to be done. So, the total work done A = 0 V · dq =
∫V ∫V where Φ is the magnetic flux through the circuit. If Φ is created
0
V · d(CV ) = C 0 V dV = CV 2 /2.
by the self-inductance effect in an inductor, we obtain
fact 12: The voltage on a capacitor cannot change moment- E = −LdI/dt.
arily, because a momentary change of the charge would require The minus sign refers to the fact that this electromotive force
an infinite current; the characteristic time of the voltage change tries to oppose the current change.
(with which the voltage will relax towards its equilibrium value)
is fact 16: The energy stored in an inductor
τ = CR, W = LI 2 /2.
where R is the net resistance of the circuit connected to the
Indeed, consider the electrical work needed to create a current
capacitor’s leads. ∫ ∫ ∫
in an inductance: A = E · dq = E · Idt = L dI dt · Idt =
∫
Indeed, consider a capacitor with voltage V , the leads of which L IdI = LI 2 /2.
are attached to a resistance R. According to Kirchoff’s laws,
R dq q fact 17: The current through an inductance cannot change
dt + C = 0, hence
dq dt t momentarily, because this would cause an infinite electromot-
=− ⇒ ln q − ln q0 = − ⇒ q = q0 e−t/RC . ive force; the characteristic time of the current change (with
q CR CR
Here, − ln q0 serves as an integration constant. which the current will relax towards its equilibrium value) is
fact 13: In a simple R − C-circuit, charge (and voltage) τ = L/R,
on the capacitor, as well as the current decay exponentially, where R is the net resistance of the circuit connected to the
∝ e−t/τ . inductance leads.

Now, let us consider inductors. In the section “Electro- Indeed, consider an inductance with current I, the leads of
magnetism’ we’ll learn that similarly to how electrical charges which are attached to a resistance R. According to Kirchoff’s
give rise to an electric field, currents (moving charges) give rise laws, RI + L dI
dt = 0, hence
to a magnetic field, which is characterized by magnetic induc- dI Rdt Rt
=− ⇒ ln I − ln I0 = − ⇒ I = I0 e−Rt/L .
tion B (also referred to as the magnetic B-field). We’ll need I L L
also the concept of magnetic flux Φ, which can be interpreted Here, − ln I0 serves as an integration constant.
intuitively (and loosely) as the number of magnetic field lines fact 18: In a simple L − R-circuit, inductor current (and
passing through a closed (possibly fictitious) loop; in the case voltage) decays exponentially, ∝ e−t/τ .
of an homogeneous magnetic field perpendicular to the loop, With this result, we are finally ready to return to the prob-
Φ = BS, where S is the surface area of the loop. The import- lem of stability of circuits with nonlinear elements obeying neg-
ance of the concept of magnetic flux lies in the fact that if it ative differential resistance, cf. idea 24 and fact 9.
changes in time, an emf. is created in the loop (circuit), see
below. idea 31: When you are asked to perform a stability analysis,
keep in mind that
So, any current in a circuit gives rise to a magnetic field,
(a) the most standard way of doing it is by assuming that the
which, in its turn, will cause a magnetic flux passing through
departure from a stationary state is very small, hence the idea
that electric circuit. Typically, however, that flux is relatively
26 can be applied, i.e. all the nonlinear dependences can be
small so that the emf. caused by it can be neglected. In order
linearized28 ;
to create a larger flux, coils (inductors) are used. Increasing the
(b) the system needs to involve inertia which, in the case of cir-
number N of overlapping wire loops has two-fold effect: first,
cuits, is most typically provided by inductance: every wire has
the current in the circuit will pass N times parallel to itself,
a non-zero inductance (a very rough rule is that one millimetre
giving rise to a N -fold increase of the magnetic field; second,
of wire length contributes 0.5 nH to the overall inductance). In-
the magnetic field lines pass now the circuit N times, giving
rise to another factor N for the magnetic flux. troduction of inductance may not be needed for systems with
capacitors which already obey inertia of capacitor charge (and
fact 14: Self-inductance of an inductor (often called just hence, of capacitor voltage).
“inductance”) is defined as (c) the linearized differential equation may have more than one
L = Φ/I, solution; the system as a whole is stable if none of the solutions
26 One can also speak about the inductance of simple circuit wires: although the inductance of simple wires is small, there are applications where it
cannot be neglected
27 However, in the case of inductors with ferromagnetic coils, there is an essential non-linearity: the inductance will decrease with increasing current.
28 Alternatively, conservation laws can be studied, cf. http://www.ipho2012.ee/physicscup/physics-solvers-mosaic/1-minimum-or-maximum/

— page 17 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
is unstable (if there is even one unstable solution, its exponen-
tial growth would lead to an eventual departure of the system
from the equilibrium state).

pr 56. Consider a tunnel diode which is connected in series

with a resistance R to a battery (for a typical V − I-curve of
a tunnel diode, see problem 25). Let the parameters of the
system be such that there is a stationary state at such diode
voltage that diode’s differential resistance Rdiff ≡ dV
dI is negat-
ive. Under which condition will this state be stable?
pr 59. A capacitor is charged by connecting it to a series
The answer of this problem has an interesting implication. connection of a battery of electromotive force E, inductor of in-
Indeed, suppose we try to measure the V-I-curve of a tunnel ductance L, and a diode. For the V –I dependence of the diode
diode; we would use the same circuit as in the case of this prob- use the graph of the previous problem; internal resistance of
lem (with the addition of an ammeter), but according to the the battery is negligible. To which voltage the capacitor will
results this problem, the state with Rdiff < 0 would be either be charged, assuming that E > Vd ?
unstable or cannot be reached due to fact 9. If that is so then
how can we measure the full V − I-curve? It appears that for idea 34: If a circuit includes a set of the plates of capacit-
this purpose, the tunnel diode can be stabilized if a suitably ors which is isolated electrically from the rest of circuit by the
selected capacitor and resistor are connected in parallel to the dielectric insulating layers of the capacitors, the net charge on
diode. In that case, the stability analysis becomes more com- these plates is conserved.
plicated; then, the most efficient approach makes use few more For instance, consider a series connection of two capacitors
ideas and therefore the corresponding problem is considered at which were initially charge-free. Then the set of two plates
the very end of the booklet (see problem 82). (shown in figure) forms an electrically insulated system, hence
the net charge there will remain always zero, i.e. the two capa-
idea 32: Energy conservation law can be used to calculate
citors will bear always equal by modulus charge.
heat dissipation. In addition to capacitors’ and inductances’
energies (c.f. facts 11 and 16), the work done by electromotive
∫ ∫
force needs to be taken into account: A = EI · dt = Edq; if
E is constant, this simplifies into
A = E · δq,
pr 60. Show that the series connection of capacitors of ca-
where δq is the charge passing through the electromotive force. pacitance C , C ,. . . C has net capacitance C = (C −1 + C −1 +
1 2 n 1 2
. . . Cn−1 )−1 .
idea 33: If a battery is connected in series to a capacitor,
the charge passing through the battery can be found as the pr 61. Three identical charge-less capacitors of capacitance
change of charge on a plate of the capacitor: C are connected in series. The capacitors are charged by con-
δq = C · δV, necting a battery of electromotive force E to the terminal leads
of this circuit. Next, the battery is disconnected, and two res-
where δV is the change of the capacitor’s voltage.
istors of resistance R are connected simultaneously as shown
in figure below. Find the net heat which will be dissipated on
pr 57. A capacitor of capacitance C is charged using a bat- each of the resistances.
tery of electromotive force E. Find the heat dissipated during
the charging process (either via a spark or in the wires or in
the battery due to (internal) resistance.

This is a very simple problem which we solve here to show

the procedure. During the charging process, a plate of the
capacitor obtains charge q; this charge necessarily needs to
come through the battery, hence the work done by the battery
A = qE = CE 2 . Part of this work is accumulated as the poten-
tial energy of the capacitor, W = CE 2 /2; the rest is dissipated idea 35: Extremal currents and voltages can be often found
as a heat, Q = A − W = CE 2 /2. from the energy conservation law by noting that (a) at the mo-
ment of an inductor’s current extremum, dI dt = 0, hence the
pr 58. A capacitor of capacitance C is charged so that its voltage on the inductor V = L dI dt = 0; (b) at the moment of
dV −1 dq
voltage is V0 . The capacitor is discharged on a series connec- a capacitor’s voltage extremum, dt = C dt = 0, hence the
dq
tion of a diode and resistor R. Assume that the following graph current through capacitor’s leads I = dt = 0.
provides a good approximation for the V –I dependence of the
diode and that the capacitor is discharged down to the voltage pr 62. Consider the electrical circuit given below: initially
Vd . Find the amount of heat which is dissipated on the resistor. chargeless capacitors C1 and C2 were connected to a battery,
— page 18 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
and at certain moment, the key K will be closed. After that nonhomogeneous part of the equations are constant terms then
moment, current and voltage will start oscillating. For these one solution can be easily found as a stationary solution which
oscillations, you need to find (a) the maximal current Imax is constant in time and for which all the time derivatives are
through the inductor; (b) the maximal voltage Vmax on the equal to zero. In the case of our circuits, the nonhomogen-
capacitor C1 . eous terms are indeed constant, and usually the stationary solu-
tions (currents in all wires and voltages on all elements) can be
figured out without writing down the differential equations.
In order to find the generic solution of the homogeneous
equations, we need to notice first that the homogeneous part of
the equations corresponds to the case when all the electromot-
ive forces and currents supplied by current sources are set equal
One way of solving this problem is using the idea 35, to- to zero; this corresponds to short-circuiting all the electromot-
gether with the energy conservation law. The second way is to ive forces and removal of all the current sources, as suggested
study the voltage (and current) oscillations in the circuit. LC- by idea 36. Now, the task of finding the generic solution may
circuit oscillations will be studied later in more details; here it become already easy enough: if parallel and series connections
is enough to formulate one more “fact”. of resistors can be combined into one single resistor, and the
same applies to capacitors, as well as to inductors then we’ll be
fact 19: In a closed circuit consisting of a capacitor C and having either a R−C circuit, L−R-circuit, or L−C-circuit. In
an inductor L, current through the inductor and voltage on each of these cases, we already know the solution as provided by
the capacitor
√ will oscillate sinusoidally with circular frequency the facts 13, 18, and 19, which is either an exponential decay
ω0 = 1/ LC, e.g. V = V0 sin(ω0 t + φ).29 I = I0 e−t/τ or a sinusoidal oscillation. In more complicated
Indeed, for such a circuit, Kirchoff’s voltage law states that cases, it is still possible to find the generic solution without
dq
q/C + L dIdt = 0; here, q is the capacitor’s charge, and I = dt ,
writing down the system of equations, by using AC-resonance
2
hence q + LC 2 = 0. This is a second order linear differential as explained by idea 48.
d q
dt
equation, the solution of which is given by q = q0 sin(ω0 t + φ),
pr 63. Under the assumptions of the previous problem,
where the constants q0 and φ can be found using the initial
sketch the voltage on the capacitor C1 as a function of time.
conditions (e.g. the current and voltage values at t = 0), c.f.
Formula sheet I-3. idea 37: If a constant voltage V is applied to the leads
idea 36: If the task is to find a temporal dependence of a of an inductor, its current will start changing linearly in time:
voltage or current, and the circuit contains one or more batter- dt = V ⇒ I = I0 + V t/L.
L dI
ies or constant current sources, the solution can be found as
a superposition of a stationary solution (when all the voltages pr 64. The circuit below makes it possible to charge a re-
and currents are constant), and a solution obtained for a sim- chargeable battery of voltage E = 12 V with a direct voltage
plified circuit, where all the ideal batteries are substituted with source of a voltage lower than E, V0 = 5 V. To that end, the
wires, and the current sources are “cut off”. key K is periodically switched on and off — the open and
closed periods have equal length of τ = 10 ms. Find the aver-
This idea is based on the fact that if there are neither batteries
age charging current assuming that L = 1 H. The diode can
nor current sources present, circuits containing linear resistors,
be considered to be ideal; neglect the ohmic resistance of the
capacitors, and/or inductors are described by Kirchoff’s laws
inductor.
which represent a set of homogeneous linear differential equa-
tions. The word homogeneous here means that each term in
these equations contains exactly one unknown function (or its
time derivative), Ik (t) or Vl (t) (the current in the k-th wire seg-
ment and the voltage on the l-th circuit element). On the other
hand, if there are also batteries or current sources in the circuit,
there would be also terms without any of the unknown functions
being involved — these are the terms involving the correspond- idea 38: For circuits containing L and R or C and R, at
ing electromotive forces and/or constant current values; such time-scales much shorter then the characteristic times
systems of equations are called linear nonhomogeneous differ- τ = RC or τ = L/R,
ential equations, and if we remove from these equations the the capacitor’s charge and inductor’s current remain almost
terms without unknown functions, we obtain what is called the constant. In particular, if a capacitor was chargeless, its voltage
homogeneous part of the equations. remains almost zero, i.e. it is essentially short-circuited; if there
General theory of linear differential equations tells us that was no current in an inductor, its current remains zero, i.e. the
the generic solution of the nonhomogeneous differential equa- wire leading to the inductor can be considered as broken. If
tions is obtained as the sum of (a) the generic solution of the a capacitor had a charge Q corresponding to a voltage V0 , its
homogeneous part of these differential equations and (b) one voltage remains essentially constant, i.e. it acts as (and can
single (any) solution of the nonhomogeneous equations. If the be substituted by) a battery of emf. E = V0 . Similarly, if an
√
29 If there is also a small resistance R connected in series then V = V0 e−γt sin(t ω02 − γ 2 + φ) with γ = R/2L; this will be derived after idea 48

— page 19 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
inductor had a current I0 , it can be substituted by a respect- The last part of this problem requires one more idea.
ive constant current source. If we try to forcefully break the
current through an inductor by switching it off, a rapid fall of idea 40: Suppose that a periodic signal is applied to a circuit
current I creates a huge voltage L dI containing two or more of the following elements: resistances
dt which usually leads to a
spark at the switch 30 (R), capacitors (C), inductances L, nonlinear elements such as
diodes. If the system has evolved long enough (much longer
At time-scales which are much longer than the characteristic than RC and L/R, so that the system response has also be-
times, the situation is reversed: inductor can be considered as a come periodic), the average voltage on the leads of an inductor
short-circuiting wire, and capacitor as an insulator. This is be- is zero, and the average current through each capacitor is zero.
cause all the currents and voltages tend exponentially towards
the equilibrium state so that the difference from the equilib- Indeed, the voltage on an inductor can be expressed via its cur-
dI
rium value ∆ ∝ e−t/τ : the capacitor charge is almost constant, rent, V = L dt , and an average non-zero voltage would imply
d⟨I⟩
hence there is no current, and the inductor current is almost a non-constant average current, dt = ⟨V ⟩ ̸= 0, which violates
constant, hence no electromotive force. the assumption that the system response has become periodic.
Similarly, a non-zero average current through the wires leading
pr 65. The key of the circuit given below has been kept to a capacitor would imply a non-constant average charge on
the plates of it.
open; at certain moment, it is closed. (a) What is the ammeter
reading immediately after the key is closed? (b) The key is This idea is illustrated with one more problem.
kept closed until an equilibrium state is achieved; what is the
pr 67. Alternating voltage V = V0 cos(2πνt) is applied to
ammeter reading now? (c) Now, the key is opened, again; what
the leads of the circuit shown below. Sketch the graphs of the
is the ammeter reading immediately after the key is opened?
resistor’s and diode’s currents as a function of time.

The short-time-approximation of the idea 38 can be further

improved with the help of the following idea.
idea 39: If the considered time interval is much less than
RC or L/R, the time dependence of the capacitors’ charges
and inductors’ currents can be linearized: q = q0 + Ic t, where
Ic is an almost constant current feeding the capacitor, and
IL = I0 + VL t/L, where VL is an almost constant voltage ap-
plied to the leads of the inductor.
pr 66. Capacitor of capacitance C and resistor of resistance
R are connected in parallel, and rectangular current pulses (see
figure) are applied to the leads of the system. Assuming that
I2 = −I1 and that at the moment t = 0, the capacitor had
no charge, sketch the voltage on the capacitor as a function of
time (a) if T ≫ RC, and (b) if T ≪ RC.
Now assume that the periodic input current has been applied
for a very long time (for much longer than RC), and let us
no longer assume I2 = −I1 . Find the average voltage and
the amplitude of the voltage oscillations on the capacitor if (c)
T ≫ RC, and (d) if T ≪ RC.
Finally, there is one more idea which can be used when it
is needed to find a charge passing through a resistor.
idea 41: If a circuit contains a current loop (as defined for
idea 13) which contains a resistor R, an inductor L, and/or
embraces an externally applied magnetic flux Φe , the charge
passing through the resistor can be expressed in terms of the
change of the magnetic flux (both external and self-induced):
q = (δΦe + LδI)/R.
Indeed, this follows immediately from the Kirchoff’s voltage
dt + L dt = −RI = −R dt (where Φi = LI), which can
dq
law dΦ e dΦi
be written as dΦ + L · dI = −R · dq and easily integrated.
pr 68. In order to measure magnetic induction, the follow-
ing device can be used. A small coil with N loops, surface area
S and inductivity L is connected to a ballistic galvanometer
which is graduated to show the total charge of a current pulse
31
. The coil is placed into a magnetic field so that the axis of
the coil is parallel to the magnetic field. With a fast motion,
the coil is flipped around by 180◦ (so that axis is again parallel
to the magnetic field); find the total charge of the current pulse
passing through the galvanometer if the total ohmic resistance
of the coil and wires is R

The next idea can be considered to be a limit case of the

idea 38, but it can be formulated as a more generic conservation
law.
30 This effect can be used intentionally to create a short pulse of high voltage.
31 The needle of the ballistic galvanometer has a large inertia, it will take some time before it will reach the equilibrium position; because of that, if
a short current pulse is let through such a galvanometer (shorter than the response time of the galvanometer), the maximal declination of the needle
will be proportional to the total charge of the pulse.

— page 20 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
idea 42: If a circuit includes a current loop which is entirely I I
(b-i)
in a superconducting state (i.e. with strictly zero resistance), I/2
the magnetic flux through it is conserved, Φ = Const. This t
t t t t
follows directly from the Kirchoff’s voltage law for the super- rn r (b-ii)
conducting circuit, dΦ
dt = 0. If the flux is only due to the
t
self-inductance, and there is only one inductor of inductance L t t t t
in the circuit then LI = Const; if L is constant then also I is I I
(b-iii)
constant. I/2
t
In the Section “Electromagnetism’, there will be more ex- t t t t
I I
amples for the application of this idea (involving external fields (b-iv)
and mutual induction); here just one problem is provided. I/2
t
pr 69. [IPhO-1994] Superconducting magnets are widely t t t t
used in laboratories. The most common form of superconduct- (b) Suppose the power switch K is turned on at time t = 0
ing magnets is a solenoid made of superconducting wire. The when r = 0, I1 = 0 and R = 7.5 Ω, and the total current I is
wonderful thing about a superconducting magnet is that it pro- 0.5 A. With K kept closed, the resistance r of the supercon-
duces high magnetic fields without any energy dissipation due ducting switch is varied in he way shown in Fig. (c)-ii. Plot the
to Joule heating, since the electrical resistance of the supercon- corresponding time dependences of I, I1 and I2 in Figs. (c)-i,
ducting wire becomes zero when the magnet is immersed in (c)-iii and (c)-iv respectively.
liquid helium at a temperature of 4.2 K. Usually, the magnet
is provided with a specially designed superconducting switch,
as shown in Fig. 1. The resistance r of the switch can be
controlled: either r = 0 in the superconducting state, or in
the normal state. When the persistent mode, with a current 0,5A I
circulating through the magnet and superconducting switch in- t
definitely. The persistent mode allows a steady magnetic field rn r 1 2 3 min
t
to be maintained for long periods with the external source cut
1 2 3 min
off. 0,5A I
t (c)
The details of the superconducting switch are not given in 0,5A I 1 2 3 min
Fig. (a). It is usually a small length of superconducting wire t
wrapped with a heater wire and suitably thermally insulated 1 2 3 min
from the liquid helium bath. On being heated, the temperat- (c) Only small currents, less than 0.5 A, are allowed to flow
ure of the superconducting wire increases and it reverts to the through the superconducting switch when it is in the normal
resistive normal state. The typical value of is a few ohms. Here state, with larger currents the switch will be burnt out. Sup-
we assume it to be 5 Ω. The inductance of a superconducting pose the superconducting magnet is operated in a persistent
magnet depends on its size; assume it be 10 H for the magnet mode, i. e. I = 0, and I1 = i1 (e.g. 20 A), I2 = −i1 , as shown
in Fig. (a). The total current I can be changed by adjusting in Fig. (d), from t = 0 to t = 3 min. If the experiment is to be
the resistance R. stopped by reducing the current through the magnet to zero,
how would you do it? This has to be done in several operation
steps. Plot the corresponding changes of I, r, I1 and I2 in
Fig. (d)

20A I
t
rn r 3 6 9 12 min
t
3 6 9 12 min
20A I (d)
The arrows denote the positive direction of I, I1 and I2 . t
3 6 9 12 min
(a) If the total current I and the resistance r of the super-
20A I
t
conducting switch are controlled to vary with time in the way r 3 6 9 12 min
shown in Figs. (b)-i and (b)-ii respectively, and assuming the -20A
currents I1 and I2 flowing through the magnet and the switch (d) Suppose the magnet is operated in a persistent mode with
respectively are equal at the beginning (Fig. (b)-iii and Fig. (b)- a persistent current of 20 A [t = 0 to t = 3 min. See Fig. (e)].
iv), how do they vary with time from t1 to t4 ? Plot your answer How would you change it to a persistent mode with a current
in Fig. (b)-iii and Fig. (b)-iv. of 30 A? plot your answer in Fig. (e).
— page 21 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
30A I
20A
t adjustable current source. We start with increasing the current
rn r 3 6 9 12 15 min
t
30A I 3 6 9 12 15 min
20A (e)
t If the final value of the current in the first loop is I1 then the
30A I 3 6 9 12 15 min
20A
t 16). Next we start increasing the current in the second loop
3 6 9 12 15 min
-20A
-30A
Previously we introduced inductance with facts 14–16. In
particular, we learned that electric current I in a circuit will
cause a flux of magnetic field Φ = LI, where L is the self-
inductance of the loop. Consider now two loops which are
positioned in each other’s neighbourhood. Then, similarly to
what we observed for a single loop, a current I1 in the first
loop will cause a magnetic field in the position of the other
loop, and hence, gives rise to a flux Φ21 through the second
loop. Due to the linearity of the Maxwell’s equations32 , the
flux is proportional to the current, so Φ21 ∝ I1 ; the coefficient
of proportionality L21 is called the mutual inductance. Now we
can express the total flux in the second loop as
Φ2 = L21 I1 + L2 I2 ,
where L2 is the self-inductance of the second loop. Similarly,
the flux through the first loop is expressed as
Φ1 = L12 I2 + L1 I1 ,
where L1 is the self-inductance of the first loop.
fact 20: If there are two loops (or two inductors) then the mu-
tual inductance L21 is defined as the coefficient of proportional-
ity between the flux in the second loop, caused by the current
I1 in the first loop, and the current I1 . L12 is defined analog-
ously; it appears that always, L12 = L21 . If there are only two
magnetically coupled circuits, both are usually denoted as M ,
so that total flux through the first loop is expressed as
Φ1 = M I1 + L1 I1 .
The total energy of the( system is ) Note that x can take any values, including negative ones, but
W = 21 L1 I12 + 2M I1 I2 + L2 I22 .
(Both equalities will be proved below.) As long as the currents
I1 and I2 flow in independent circuits, the sign of M is not
important as we can just change the sign of one of the currents.
However, if I1 and I2 are connected to each other via Kirchoff’s
current law33 , we need to be careful to select the correct sign
of M .
The equality L12 = L21 is very useful, it is analogous to
the reciprocity theorem (cf. appendix 2) and to Newton’s 3rd
law. Sometimes it is difficult to calculate a force exerted by a
body A to a body B, but it is easy to calculate the force exer-
ted by the body B to the body A (for instance, when A is a
point charge and B — an homogeneously charged plate); simil-
arly, it is sometimes difficult to calculate L12 , but it is easy to
calculate L21 (naturally we determine M then by calculating
L21 ).
In order to prove this fact, let us first derive an expression
for the total energy for the system consisting of the two current-
carrying circuits similarly to how we proved the fact 16. So we
32 To be discussed in more details in the Electromagnetism booklet
33 For instance, by problem 85
— page 22 —
have two simple circuits, each consisting of an inductor and an
in the first loop while keeping the current in the second loop
zero. The work done by the current source in the second loop
is zero because no charge will pass through the current source.
work done by the first current source is 12 L1 I12 (due to the fact
while keeping the current in the first loop constant. Let us con-
sider a current increment dI2 ; this will give rise to a voltage in
the first loop which needs to be compensated by an electromot-
ive force of the current source E1 = L12 dI dt ; similarly for the
2
current source of the second loop E2 = L2 dt . The work done
dI2
by the first current source
∫ ∫ ∫ ∫
dI2 dq1
A1 == E1 dq1 = L12 ·dq1 = L12 dI2 = L12 I1 dI2 .
dt dt
Now let us recall that in our process, I1 is constant, so the
integral is easily taken, resulting in A1 = L12 I1 I2 . The work
done by the second current source is calculated similarly, res-
ulting in A2 = E2 dq2 = 12 L2 I22 . Therefore, the total amount of
work which has been done is
( )
W = 12 L1 I12 + A1 + A2 = 12 L1 I12 + 2L12 I1 I2 + L2 I22 .
Now it is clear that if we increase the currents in the reverse
order, we shall obtain a result where the indices 1 and 2 are
( )
swapped: W = 12 L1 I12 + 2L21 I1 I2 + L2 I22 . The work of the
current sources was transferred to the energy of the magnetic
field and its value can depend only on the final state of the sys-
tem, hence these two expressions must provide the same result
and therefore, L12 = L21 .
Magnetic field energy density is proportional to the squared
magnetic field strength and hence, is always positive; because
of that, any system of magnetically coupled current-carrying
inductors must also have a positive total energy. From this
condition, we can derive a useful inequality for the mutual
inductance M . Indeed, for two loops, the total energy can
( )
expressed as W = 12 I12 L1 + 2M x + L2 x2 , where x = I2 /I1 .
W must remain non-negative. Therefore, the roots x1 and x2
of the quadratic equation L1 + 2M x + L2 x2 = 0 cannot take
real values (unless the two roots are equal): otherwise, for any
value of x between x1 and x2 , W would be negative. Thus, the
discriminant must be nonnegative, M 2 ≤ L1 L2 .
Equality M 2 = L1 L2 would mean that the total energy of
the system can be zero even if there are non-zero currents in
the loops. Zero energy means zero magnetic field: the magnetic
field created by a current in one loop must negate everywhere
the magnetic field created by the current in the other loop.
This is possible only if the shape of the field lines of the both
loops are identical. To achieve this, there are two possibilities:
(a) they must have identical and overlapping in space geomet-
rical shape (two solenoids of identical length and cross-sectional
area but possibly with different winding densities, one of which
is inserted tightly into the other), or
(b) the windings of the both inductors must be made around
the same closed ferromagnetic core as shown in figure — such
devices are called transformers. As we shall learn in electro-
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
magnetism, magnetic field lines keep, if possible, inside fer- (a) Find the current in the both loops immediately after the
romagnetic materials; if the ferromagnetic core is closed (e.g. switch is closed.
having a toroidal shape), there is no need for the field lines (b) Find the currents as a function of time.
caused by the current in the winding to exit the core. In that
case, the shape of the field lines inside the core is defined by
the shape of the core, and not by where and how the winding
is made. Note that electrical transformers are usually made
using a closed ferromagnetic core as shown in the figure below;
for the electrical symbol of such transformer, see problem 70.

Problems involving ideas 32–43

pr 71. There are three identical lamps which are connected
to a battery as shown in figure; the current through each lamp
is I. Find the currents immediately after the key is opened.
fact 21: Mutual inductance cannot be larger than the
√
geo-
metric average of the self-inductances; equality M = L1 L2
is achieved for transformers when all the magnetic field lines
created by the both coils have identical shapes.

idea 43: If there is no leakage√

of magnetic field lines from
a transformer, i.e. if M = L1 L2 then the total inductive pr 72. In the circuit shown below, the key K has been kept
electromotive force in both inductors is defined by the same
√ √ in the position 1; after an equilibrium state has been reached,
linear combination of currents J ≡ I1 L1 + I2 L2 (indeed,
√ √ the key is thrown over to the position 2. This is done much
Φ1 = J L1 and Φ2 = J L2 ). This has two important con- faster than the characteristic time (L1 + L2 )/R. After that,
sequences. we wait for a very long time until a new equilibrium state is
(i) The idea 38 states that if there are inductors, the inductor reached. Find the amount of heat Q which was dissipated in
currents need to be continuous; in the case of transformers with
√ the resistor R after the key was switched to the position 2.
M = L1 L2 , this statement has to be modified: what need to Also, find the total charge q which flowed through the resistor
be continuous are the fluxes, and this is achieved as long as J R during the same period of time. Neglect the internal resist-
is a contiuous function√of time, i.e. current jumps satsify the ance of the inductors. Note that the resistance of all the wires
condition ∆I1 = −∆I2 I2 /I1 . is also negligible, but there is no superconductivity.
(ii) Assuming that there are no other inductors affected by the
currents I1 and I2 , if we write down all the Kirchoff’s voltage
laws, we can reduce the order of the system of differential equa-
√
tions by one if we use J (or J/ L1 which has the dimensionality
of an electric current) as one unknown function, and any other
linear combination of I1 and I2 as the other unknown.
Let us elaborate on the statement (i) and consider the simplest
case of only two unknown functions, I1 and I2 . Then we would
be able to get two equations given by the Kirchoff’s voltage
law. Inductors yield us terms containing derivatives of I1 and
√ pr 73. Capacitor of capacitance C and resistor of resistance
I2 so that with M < L1 L2 we would have a second order
system of linear differential equations (if we take derivative of R are connected in series, and rectangular voltage pulses (see
both equations, and eliminate from the four equations I2 and figure) are applied to the leads of the system. Find the aver-
dI2 age power which dissipates on the resistors assuming (a) that
dt , we obtain
√
a second order differential equation for I1 ). Now
with M = L1 L2 , however, we can use J as one of the vari- T ≫ RC; (b) that T ≪ RC.
ables, and eliminate dJ dt from the set of equations; we result
in an expression relating I1 and I2 to each other, i.e. we can
express I2 in terms of I1 . If we substitute I2 in one of the
original differential equations using this expression, we obtain
a first order differential equation for I1 .

pr 70. An electrical transformer is connected as shown in

the circuit below. Both windings of the transformer have the
same number of loops and the self-inductance of the both coils pr 74. Determine the time constant for the circuit shown
is equal to L; there is no leakage of the magnetic field lines in figure (i.e. for the process of charging the capacitor, time
from the core so that the mutual inductance is also equal to L. interval during which the charging rate drops e times).
— page 23 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
(a) Find the average power dissipation at the diode at the work-
ing regime of such a circuit.
(b) Determine the amplitude of the AC voltage (with frequency
ν = 50 Hz), which has to be applied at the input of the circuit.
(c) Find the required capacitance C.
(d) Find the average power dissipation at the diode during the
first period (of AC input voltage) immediately following the
application of AC voltage to the input of the circuit.
pr 75. A boy wants to build decorative lights using 50 light
emitting diodes, to be fed by AC-voltage V = V0 cos(2πνt), pr 78. [Est-Fin-2010]
with V0 = 311 V and ν = 50 Hz. The circuit he plans to
(a) Consider the circuit given in Fig (a), where the diode can
use is given below. The voltage of his light emitting diodes
be assumed to be ideal (i.e. having zero resistance for forward
can be taken equal to 3 V (it remains constant for a wide
current and infinite resistance for reverse current. The key is
range of forward currents); the nominal current is 20 mA.
switched on for a time τc and then switched off, again. The
Find the optimal value of the resistor R (ensuring a nom-
input and output voltages are during the whole process con-
inal operation of the diodes), and minimal value of the ca-
stant and equal to Ui and Uo , respectively (2Ui < Uo ). Plot
pacitance C, if the current variations need to be less than
the graphs of input and output currents as functions of time.
5%. The rectifying diode D can be considered to be ideal.

(b) Now, the key is switched on and off periodically; each time,

the key is kept closed for time interval τc and open — also for
pr 76. [EstFin-2012] For the circuit shown in Figure, R1 = τc . Find the average output current.
3R, R2 = R, C1 = C2 = C, and L1 = L2 = L. The electromot-
ive force of the battery is E. Initially the switch is closed and
the system is operating in a stationary regime.
(a) Find the reading of the voltmeter in the stationary regime.
(b) Now, the switch is opened. Find the reading of the volt-
meter immediately after the opening. (c) Now, circuit (a) is substituted by circuit (b); the switch
(c) Find the total amount of heat which will be dissipated on is switched on and off as in part ii. What will be the voltage
each of the resistors after opening the switch, and until a new on the load R, when a stationary working regime has been
equilibrium state is achieved. reached? You may assume that τc ≪ RC, i.e. the voltage vari-
V ation on the load (and capacitor) is negligible during the whole
R1 R2 period (i.e. the charge on the capacitor has no time to change
significantly).

C1 L2 pr 79. [IPhO-2001] A sawtooth voltage waveform V0 can

L1 C2
ε ing voltage Vf , the air between the electrodes breaks down,
pr 77. [EstFin-2008] A voltage rectifier is made according
to the circuit depicted in Figure. The load R = 10 kΩ is fed
with DC, equal to I = 2 mA. In what follows we approximate
the U-I characteristic of the diode with the curve depicted in
Figure. The relative variation of the current at the load has to
satisfy the condition ∆I/I < 1%.
— page 24 —
be obtained across the capacitor C in Fig. (A); R is a variable
resistor, Vi is an ideal battery, and SG is a spark gap con-
sisting of two electrodes with an adjustable distance between
them. When the voltage across the electrodes exceeds the fir-
hence the gap becomes a short circuit and remains so until the
voltage across the gap becomes very small.
(a) Draw the voltage waveform V0 versus time t, after the switch
is closed.
(b) What condition must be satisfied in order to have an almost
linearly varying sawtooth voltage waveform V0 ?
(c) Provided that this condition is satisfied, derive a simplified
expression for the period T of the waveform.
(d) What should you vary( R and/or SG ) to change the period
only?
(e) What should you vary (R and/or SG ) to change the amp-
litude only?
(f) You are given an additional, adjustable DC voltage supply.
Design and draw a new circuit indicating the terminals where
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
you would obtain the voltage waveform described in Fig. (B).

pr 80. [Est-Fin-2013] An inductance L and a capacitor C

are connected in series with a switch. Initially the switch is (a) Find the current through the resistors R1 and R2 at the
open and the capacitor is given a charge q0 . Now the switch is moment of time t1 = 5 ms.
closed. (b) Find the current through the resistors R1 and R2 at the
(a) What are the charge q on the capacitor and the current I moment of time t2 = 15 ms.
in the circuit as functions of time? Draw the phase diagram (c) What is the net charge passing through the resistor R2 ?
of the system — the evolution of the system on a I − q graph
— and note the curve’s parameters. Note the direction of the pr 82. As we have learned with problem 56, if we want to
system’s evolution with arrow(s). measure the full V − I-curve of a tunnel diode, it needs to be
specifically stabilized34 . Let us study such a stabilization in
more details. Let a tunnel diode be connected in series with a
A Zener diode is a non-linear circuit element that acts as a resistance R to a battery, and let the parameters of the system
bi-directional diode: it allows the current to flow in the pos- be such that at the stationary state, the diode’s voltage is such
itive direction when a forward voltage on it exceeds a certain that the diode’s differential resistance Rdiff < −R.
threshold value, but it also allows a current to flow in the oppos- (a) According to the results of problem 56, this stationary state
ite direction when exposed to sufficiently large negative voltage. is unstable; in order to stabilize it, a series connection of a ca-
Normally the two voltage scales are quite different, but for our pacitance C and a resistance r are connected in parallel to the
purposes we will take a Zener diode with the following volt- diode. The wires connecting the capacitor and resistor r to the
ampere characteristics: for forward currents, the voltage on diode are so short that the corresponding inductance Ld can
the diode is Vd , for reverse currents, the voltage on the diode is be assumed to be negligibly small; the inductance of the wires
−Vd , for zero current the voltage on the diode is −Vd < V < Vd . connecting the diode, battery, and resistor R to each other is
L, (see figure; neglect the capacitance Cd ). Which condition(s)
need to be satisfied for C, R, Rdiff , r and L for stabilization?
Now we connect the inductance L, the capacitor C all in series
with a switch and a Zener diode. The switch is initially open.
The capacitor is again given the charge q0 > CVd and the
switch is then closed.
(b) Make a drawing of the phase diagram for the system. Note
the direction of the system’s evolution with arrow(s).
(c) Does the evolution of the system only necessarily stop for (b) In task (a) we addressed a relatively slow instability; in
q = 0? Find the range of values of q on the capacitor for which that case, the characteristic time of current variation cannot
the evolution of the system will necessarily come to a halt. be much less than ( )
L
(d) Find the decrease ∆q in the maximum positive value of the τs ≡ min , C|r + R diff | .
|R + Rdiff |
capacitor’s charge q after one full oscillation. How long does it However, with the new addition to the circuit, there is now a
take before oscillation halts? current loop consisting of the diode, resistance r, capacitance
(e) Suggest a mechanical system which is analogous to this C, and inductance Ld in which the characteristic time scale τf
circuit. can be much smaller than τs , of the order of Ld /|r + Rdiff |. For
such fast current fluctuations, we can no longer neglect the in-
ductance Ld (its impedance becomes comparable with |Rdiff |).
The negative differential resistance in that loop may give rise to
pr 81. [EstFin-2009] Consider an electric circuit consisting emergence of instabilities which develop within the time-scale
of a coil of negligibly small inductance, consisting of N = 10 τf ≪ τs . For the analysis of instabilities within that loop, it is
turns and with the surface area of a single loop S = 10 cm2 , also important to notice that the tunnel diode has electrodes
resistors R1 = R2 = 3 Ω, capacitor C = 0.2 F, and an induct- which perform as a capacitor of capacitance Cd ≪ C, connected
ance L = 1H, connected as shown in Fig. At the moment of in parallel to the differential resistance Rdiff (for very fast cur-
time t = 0, a magnetic field, parallel to the axis of the coil is rent fluctuations, the impedance of Cd may become comparible
switched on. The induction of the magnetic field starts growing with |Rdiff |). Find the condition for the lack of fast instabilit-
linearly, starting from B = 0 until the maximal value B = 1 T ies, i.e. instabilities which would develop within a time-scale
is achieved at t = 10 ms. Further, the induction of the magnetic τf ≪ τs ; simplify your calculations by using appropriate ap-
field remains constant (and equal to 1 T). proximations.
34 Such a stabilization has been done for the black-box-experiment at IPhO-2012

— page 25 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES

pr 83. [IPhO-2016] The I − V characteristics of a thyristor (e) Find expressions for the times t1 and t2 that the system
can be approximated by a piece-wise linear graph as shown spends on each branch of the I −V graph during the oscillation
in the figure below. Henceforth we consider such an idealized cycle. Determine their numerical values. Find the numerical
thyristor, the I − V curve of which is given by figure, and value of the oscillation period T assuming that the time needed
refer to it as the “nonlinear element X”. In the voltage range for jumps between the branches of the I − V graph is negligible.
between Uh = 4.00 V (the holding voltage) and Uth = 10.0 V (f) Estimate the average power P dissipated by the non-linear
(the threshold voltage) this I − V curve is multivalued. element over the course of one oscillation. An order of mag-
nitude is sufficient.
(g) A neuron in a human brain has the following property:
when excited by an external signal, it makes one single oscilla-
tion and then returns to its initial state. This feature is called
excitability. Due to this property, pulses can propagate in the
network of coupled neurons constituting the nerve systems. A
semiconductor chip designed to mimic excitability and pulse
propagation is called a neuristor (from neuron and transistor).
We attempt to model a simple neuristor using a circuit that
includes the non-linear element X that we investigated previ-
ously. To this end, the voltage E in the circuit above is de-
creased to the value E ′ = 12.0 V. The oscillations stop, and
the system reaches its stationary state. Then, the voltage is
rapidly increased back to the value E = 15.0 V, and after a
period of time τ (with τ < T ) is set again to the value E ′ (see
(a) Using the graph, determine the resistance Ron of the ele- figure below). It turns out that there is a certain critical value
ment X on the upper branch of the I − V characteristics, and τ , and the system shows qualitatively different behavior for
crit.
Roff on the lower branch, respectively. The middle branch is τ < τ
crit and for τ > τcrit .
described by the equation I = I0 − RUint . Find the values of the
16
parameters I0 and Rint .
(b) The element X is connected in series with a resistor R, an 15

inductor L and an ideal voltage source E (see figure below). 14

ℰ [V]
The circuit is said to be in a stationary state if the current is
constant in time, I(t) = const. 13

R L
12
t0 t0 + τ
11
ℰ t
Sketch the graphs of the time dependence of the current IX (t)
How many different stationary states this circuit may have for on the non-linear element X for τ < τcrit and for τ > τcrit .
a fixed value of E and for R = 3.00 Ω (consider different values
(h) Find the expression and the numerical value of the critical
of E)? How does the answer change for R = 1.00 Ω?
time τcrit for which the scenario switches.
(c) Let R = 3.00 Ω, L = 1.00 µH and E = 15.0 V in the cir-
cuit shown above. Determine the values of the current Ist and pr 84. [Est-Fin-2014]
the voltage Vst on the non-linear element X in the stationary In order to obtain high voltage supply using a battery, the
state. Is this state stable or unstable (study the effect of a following circuit is used.
small departure of the current strength from Ist )?
(d) We now investigate a new circuit configuration, see figure
below. This time, the non-linear element X is connected in par-
allel to a capacitor of capacitance C = 1.00 µF. This block is
then connected in series to a resistor of resistance R = 3.00 Ω
and an ideal constant voltage source of voltage E = 15.0 V.
It turns out that this circuit undergoes oscillations with the
non-linear element X jumping from one branch of the I − V
characteristics to another over the course of one cycle. An electromagnetic switch K1 connects a battery of electromot-
R ive force E to an inductor of inductance L: it is closed if there
is no current in the inductor (a spring keeps it closed), but if
C ℰ the inductor current reaches a critical value I0 , magnetic field
created by the inductor pulls it open. Due to inertia, once the
Draw the oscillation cycle on the I − V graph, including its key is open, it takes a certain time τK to close again even if
direction (clockwise or anticlockwise). the current falls to zero.
— page 26 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
For the diode D you may assume that its current is zero for any numbers:
reverse voltage (VD < 0), and also for any forward voltage smal- ei(ωt+φ) = cos(ωt + φ) + i sin(ωt + φ).
ler than the opening voltage V0 (i.e. for 0 < VD < V0 ). For any So, instead of using a sine or cosine, we write I = I0 ei(ωt+φ) .
non-zero forward current, the diode voltage VD remains equal The exponential function is much easier to deal with, because
to V0 . if we add different voltages or currents with the same fre-
quency, the term eiωt can be factorised, owing to the property
You may express your answers in terms of L, E, I0 , V0 , R, and a+b
e = ea · eb (see appendix 6). There is no need to worry that
the capacitance C (see figure).
physical quantities are typically measured in real numbers, and
a) At first, let the key K2 be open. If the initial inductor now we have suddenly a complex current (and voltage): cur-
current is zero, how long time τL will it take to open the key rent remains to be a real-valued quantity; when we write it in a
K1 ? complex form, we just keep in mind that what we actually have
b) Assuming (here and in what follows) that L/R ≪ τK ≪ τL , (in physical reality) is the real part of that complex number. So,
plot the inductor current as a function of time t (for 0 ≤ t < if we write I = I0 ei(ωt+φ) , we assume that the physically meas-
3τL ). urable current is Ir = ℜI0 ei(ωt+φ) = I0 cos(ωt + φ) (ℜz stands
for “real part of z”).
c) What is the maximal voltage Vmax on the resistor R?
Now, if we accept the complex form I = I0 ei(ωt+φ) =
d) Assuming that Vmax ≫ V0 , what is the average power dis- I0 e · eiωt , it is convenient to combine I0 and eiφ into what
iφ

sipation on the diode? we call the complex amplitude of the current,

Ic = I0 eiφ .
e) Now, let the key K2 be closed, and let us assume √ simplify-
Then, all the currents and voltages are products of eiωt with the
ingly that V0 = 0; also, RC ≫ τL and τK > π LC. Suppose
complex amplitude, which means that for any linear combin-
that the circuit has been operated for a very long time. Find
ation of currents and voltages, the time-dependent factor eiωt
the average voltage on the resistor.
can be brought before the braces. If so, there is no need to write
f) Find the amplitude of voltage variations on the resistor. always that term: typically, all the calculations are done just
with the complex amplitudes, the modulus of which gives us
pr 85. An electrical transformer is connected as shown in the amplitude, |Ic | = |I0 eiφ| = |I0 ||eiφ | = I0 , and the argument
the circuit below. Both windings of the transformer have the of which gives us the phase shift, φ = arg I0 = arctan ℑIc /ℜIc
same number of loops and the self-inductance of the both coils (for more details about those properties of complex numbers
is equal to L; there is no leakage of the magnetic field lines which have been used here, see appendix 8).
from the core so that the mutual inductance is also equal to L. From this brief theory we can draw the following conclu-
sions. Operating with complex amplitudes works well as long
as we have a single sinusoidal signal, and only linear circuit
elements are included. Inversely, complex amplitudes cannot
be used if (a) the signal is not sinusoidal, e.g. rectangular;
(b) if there are nonlinear elements, e.g. diodes, capacitors for
(a) Consider the case when the coil windings are oriented so which capacitance depends on the charge, etc. If we have a su-
that if the both coils have current flowing from left to right perposition of different frequencies and these assumptions are
then the magnetic fields in the transformer core add up de- satisfied, the different frequency signals need to be studied sep-
structively Find the currents in the resistors immediately after arately (superposition principle can be applied), and for each
the switch is closed. component-signal, the complex amplitudes can be used. An
(b) Under the assumption of the task (a), find the current in important case is the power dissipated in the circuit: this is a
the left resistor as a function of time. nonlinear function of the voltage and current, and so we need
(c) Now, let one of the coils have a reversed direction of wind- to be careful. Let I and V be the complex amplitudes of the
ing; find the current in the right resistor as a function of time. current and voltage. Then⟨ ⟩
⟨ ⟩ Ieiωt + Ie−iωt V eiωt + V e−iωt
P = ℜIe · ℜV e
iωt iωt
= · ,
2 2
where ⟨. . .⟩ denotes averaging over time, and bar over a sym-
Alternating current
bol denotes a complex conjugate (a + bi ≡ a − bi; eiω = e−iω ).
⟨ ⟩
Alternating current (AC) and voltage are assumed to be sinus- Upon opening the braces and using the fact that ei2ωt =
⟨ ⟩
oidal, e.g. I = I0 sin(ωt + φ). Kirchoff’s laws are linear — they e−i2ωt = ⟨cos 2ωt⟩ + i ⟨sin 2ωt⟩ = 0, we obtain
involve only adding first powers of voltages and currents; hence, IV + V I eiφ1 e−iφ2 + e−iφ1 eiφ2
as long as the circuit elements are linear (i.e. their properties do P = = |I||V | ;
4 4
1 ix −ix
not depend on the amplitude of the current or voltage), dealing using the formula cos x = 2 (e + e ), we end up with
with Kirchoff’s laws means dealing with linear combinations of ei(φ1 −φ2 ) + e−i(φ1 −φ2 ) 1
P = |I||V | = |I||V | cos(φ1 − φ2 ).
voltages and currents. However, sine and cosine are not very 4 2
convenient functions for adding, in particular if different terms Note that this can be rewritten as P = 12 ℜV I, because
iφ1 −iφ2
have different phase shift φ. Luckily, using the Euler’s formula ℜV I = |V ||I|ℜe e 2 = |V2 ||I| cos(φ1 − φ2 ). Also, since
(see appendix 5), sine and cosine can be substituted with ex- ℜV I = ℜ(ZI)I = ℜZ|I| = |I| 1 2
ℜZ, we can write
ponential function, if we switch from real numbers to complex P = |I| ℜZ.
2
— page 27 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
In order to get rid of the factor 12 , amplitudes are often sub- would be a good idea to equip some appliances with capacit-
√
stituted by root mean square (rms) amplitudes: Ũ = U/ 2, ors, which make the phase shift opposite without introducing
√
I˜ = I/ 2 (always make clear with which amplitude you are any additional power dissipation: when currents of opposite
dealing with). (or nearly opposite) phase shift are added in the power lines,
And so, in the case of AC currents, it is convenient to deal the large and opposite imaginary parts of the complex current
with complex amplitudes. Most often, the temporal depend- amplitudes cancel out, giving rise to a significant reduction of
ence in the form of eiωt is never written, and calculations in- the net current.
volve only the complex amplitudes.
fact 22: It should be also mentioned that sometimes, the
Let us recall that at the leads of an inductor, U = L dI dt . concepts of reactive and apparent powers, P and P are used,
r a
Once we substitute here I = I0 eiωt we obtain immediately
defined as
U = iωLI0 eiωt . The prefactor of the exponent here is the com-
Pa = |V I| and Pr = ℑV I,
plex amplitude U0 = iωLI0 of the capacitor’s voltage; upon
where ℑz stands for “imaginary part of z”.
denoting
ZL = iωL However, these concepts are not very useful, and serve mainly
we can rewrite the last equality as U0 = ZL I0 ; here, ZL is as tools to emphasize the importance of having large cos φ —
called impedance. So, if dealing with complex amplitudes, small reactive power.
an inductor’s voltage and current satisfy the Ohm’s law in the
same way as in the case of resistors with a direct current (DC)
pr 86. Consider a soldering gun of nominal power P = 30 W
— just instead of a resistance, its impedance is to be used. and nominal voltage V = 220 V (AC voltage with frequency
∫
Similarly, for a capacitor we have U = q/C = I · dt/C = ν = 50 Hz). Which capacitance needs to be connected in series
∫ to the iron in order to reduce the power down to P1 = 20 W?
I0 eiωt · dt/C = I0 /iωC, i.e. U0 = I0 ZC with
1
ZC = . pr 87. [IPhO-1982] An alternating voltage of 50 Hz fre-
iωC
Finally, for a resistor we have still the Ohm’s law U = IR = quency is applied to the fluorescent lamp as shown in the ac-
iωt
I0 e , hence U0 = ZR I0 with companying circuit diagram. The following quantities are meas-
ZR = R. ured: overall voltage (main voltage) V = 228.5 V, electric cur-
rent I = 0.6 A, voltage across the fluorescent lamp U ′ = 84 V,
Sometimes this is called the “active resistance”, or “ohmic res-
ohmic resistance of the inductor Rd = 26.3 Ω. The fluorescent
istance”, emphasizing the difference from the “reactive” and
lamp itself may be considered as an ohmic resistor in the cal-
non-ohmic impedances ZL and ZC .
culations.
As a conclusion:
(a) What is the inductance L of the series reactor?
idea 44: For AC circuits, all the techniques learnt for DC cur- (b) What is the phase shift φ between voltage and current?
rents can be used (Kirchoff’s laws, method of potentials etc.), (c) What is the active power Pw transformed by the apparatus?
if calculations are made with the complex amplitudes, and im- (d) Apart from limiting the current the series inductor has an-
pedances are used as resistors: for the complex amplitudes of other important function. Name and explain this function!
the voltage and current, V = IZ, where Z is the circuit’s full Hint: The starter (denoted by a circle with “S”) includes a con-
impedance; the phase shift between the voltage and current is tact which closes shortly after switching on the lamp, opens up
given by φ = arg Z. again and stays open.
The only difference is in the way how the power dissipation is (e) In a diagram with a quantitative time scale sketch the time
to be calculated (see above). sequence of the luminous flux emitted by the lamp.
(f) Why has the lamp to be ignited only once although the ap-
idea 45: For AC circuits, the dissipated power plied alternating voltage goes through zero in regular intervals?
P = |I||V | cos φ = ℜV I = |I|2 ℜZ. (g) According to the statement of the manufacturer, for a
NB! Here V and I are assumed to be the rms amplitudes; if we fluorescent lamp of the described type a capacitor of about
deal with the real amplitudes, the factor 12 is to be added. Al- C = 4.7 µF can be switched in series with the series reactor.
ternatively, since there is no power dissipation on the inductors How does this affect the operation of the lamp and to what
and capacitors (for which φ = π2 so that cos φ = 0), the power intent is this possibility provided for?
can be calculated as the power dissipated in all the resistors,
2
for each of which P = RIR (IR being the resistor’s current).
Note that if we deal with AC appliances and cos φ is small,
for a given required power dissipation, the current needs to
be larger than what would be in the case of larger values of
cos φ. Unnecessarily large current means unnecessarily large
dissipation losses in the power lines. The appliances based on idea 46: With alternating currents, voltages caused by mu-
inductors (in particular those including electromagnetic mo- tual inductances can be calculated in the same way as in the
tors) have intrinsically small cos φ. Therefore, if several ap- case of inductors: if there is a current of complex amplitude I1
pliances of small cos φ are plugged simultaneously into a AC in a coil of inductance L1 which is magnetically coupled (with
outlet, in order to reduce the net current in the power lines, it mutual inductance M ) to a second coil with current I2 then
— page 28 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
the total voltage on the first inductor is iωL1 I1 + iωM I2 . NB! If the circuit contains resistors, the natural frequencies are
Be careful with the sign of the mutual inductance, cf. fact 20. typically complex numbers; then, imaginary part of the com-
plex number gives the exponential decay of the corresponding
−iω t
pr 88. Around a toroidal ferromagnetic core of a very large mode: I = e i sin(ωr t + φ), where ωr and ωi stand for the
magnetic permeability, a coil is wound; this coil has a large real and imaginary parts of the natural frequency, respectively.
number of loops and its total inductivity is L. A capacitor of If the impedance of a circuit is not dominated by inductors and
capacitance C is connected to the middle point of the coil’s capacitors, there are frequencies for which ωi = 0. For instance,
wire as shown in figure. AC voltage V0 of circular frequency ω one can say that the natural frequency of a simple RC-circuit
iωt −t/RC
is applied to the input leads of the circuit; what is the reading is ω = i/RC; indeed, with I = I0 e = I0 e we recover
of the ammeter (which can be considered to be ideal)? the fact 12.

fact 24: Suppose at a certain circular frequency ω, imped-

fact 23: In AC circuits the impedance of which is domin-
ated by capacitors and inductors, free oscillations of current
and voltage can take place; the decay rate of oscillations is
defined by the ohmic resistance of the circuit. The frequency
of such an oscillation is called the natural frequency, or eigenfre-
quency; the corresponding current- and voltage oscillations are
referred to as the natural modes. If the circuit contains more
than one current loop, there may be more than one natural
frequencies. Then, if the circuit is left isolated from external
inputs, any current- and voltage dynamics in that circuit can
be represented as a superposition of the natural modes.
That superposition thing means mathematically that if we have
n nodes characterized by the node potentials ϕj , j = 1, . . . , n,
and m natural frequencies ωµ , µ = 1, . . . m, and in the case of
the µ-th natural frequency, the node voltages oscillate accord-
ing to the law
ϕj = Vµj eiωµ t ,
where Vµj is the complex amplitude of the potential of the j-
th node in the case of the µ-th natural mode, then arbitrary
motion of the system can be represented as
∑m
ϕj = Aµ Vµj eiωµ t ,
µ=1
where Aµ is a constant — the amplitude of the µ-th natural
motion. Such a decomposition into natural modes will be re-
visited in the section “Oscillations and waves”.
idea 47: If the impedance of a circuit is dominated by in-
ductors and capacitors35 the number of its natural oscillation
modes can be found as follows. Find the number of linearly
independent loops Nl for the given circuit (cf. idea 13); find
the number of such linearly independent loops N= which con-
tain only one type of elements (all resistors, all inductors, or
all capacitors); the number of natural modes is then given by
Nm = Ni − N= . In addition to that, each linearly independent
loop containing only inductors contributes one zero-frequency
mode: a constant current can circulate in each such a loop.
the corresponding characteristic equation is Lλ2 + Rλ + C −1 =
The proof of this idea (together with its generalization to cir- 0. Meanwhile, if we “cut” one of the wires then the impedance
cuits for which resistances play an important role) is given in will be Z = iωL + R + iωC
appendix 9. as Z(ω) ≡ −ω L + iωR + C −1 = 0, which is identical to the
1
ance is very large, Z(ω) ≈ 0. Then, a very small current driven
to the leads will give rise to a very large voltage V = IZ; this
phenomenon is called the voltage resonance. Similarly, if the
impedance is very small, Z(ω) ≈ 0, we have a current reson-
ance: small input voltage will lead to a large current.
idea 48: The natural frequencies can be found as the reson-
ance frequencies; there are two options. First, you can select
two points A and B at the circuit, henceforth referred to as
the fictitious terminals, and equate the impedance of the cir-
cuit between A and B to infinity and solve the equation with
respect to the frequency: although there is no input current,
there can be voltage oscillations at a resonant frequency, be-
cause with V = IZ, I = 0, and Z = ∞, V can take any value.
Second option: select a point A the circuit and cut the cir-
cuit fictitiously at that point. Thus, one “half” of the point A
becomes the first terminal A1 of the new circuit, and another
“half” becomes the second terminal A2 . Since in the original
circuit, A1 and A2 coincide, they must have the same voltage:
voltage between A1 and A2 is zero. Finally, equate the imped-
ance between A1 and A2 to zero and solve it: although the
voltage is zero, there can be a non-zero current I = U/R.
The technique described by this idea is a shortcut substitut-
ing the standard method for finding natural frequencies. The
standard method involves two steps: (a) writing down the
full set of linear homogeneous differential equations using the
Kirchoff’s laws for the circuit; (b) writing down the charac-
teristic equations where derivatives of an unknown function
are substituted by powers of an unknown variable (cf. Formula
sheet pt. I-3). When we use the idea 48, we basically bypass
the first step and obtain directly the characteristic equation by
equating the impedance (or its reciprocal) to zero.
As an illustration, let us consider a simple circuit where a
resistor of resistance R, inductor of inductance L, and a capa-
citor of capacitance C are connected in series. If we denote
the charge of the capacitor as q, the current in the circuit is
expressed as I = dq dt so that the Kirchoff’s voltage law gives us
a differential equation
d2 q dq q
+R + L= 0;
dt2 dt C
1
= 0; this equation can be rewritten
2

35 Resistances connected in series to inductors are small, and resistances connected in parallel to capacitors are large.

— page 29 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
characteristic equation with λ = iω. This equation yields want to recover), the voltage between the fictitious terminals
√
iR R 2 is constantly zero, hence we can short-circuit these terminals
ω= ± (LC)−2 − 2
. and thereby simplify the circuit. For those “lost” modes, the
2L 4L
When using impedances we assume that the temporal depend- new circuit is no different from the old circuit, so among the
ence of all the quantities is proportional to eiωt . Therefore, the solutions of the new circuit, there must be the “lost” frequency.
impedance is zero and free current oscillations are possible with Analogously, if the solutions were lost when the voltage reson-
I = I0 eiωt = I0 e− 2L t e±iω0 t ,
R
ance condition was used, the lost solutions are among the nat-
√
ural frequencies of the cut circuit (terminals A1 and A2 remain
where ω0 ≡ (LC)−2 − 4L2 . Let us recall now that with
R 2

disconnected).
complex-number-representation, we assume that the real cur-
rent value is given by the real part of a complex number; with
However, one must keep in mind that such a simplified cir-
a complex amplitude I0 = |I0 |eiφ , this leads us to a real-valued
cuit may have other natural frequencies (additional to the pre-
solution I = |I0 |e− 2L t cos(ω0 t ± φ).
R

viously “lost” solutions), possibly different from the natural

It is useful to keep in mind that the idea 48 can be applied
frequencies of the original circuit. So, the method is safe, if
not only in the case of LC-circuits, but also in the case of L−R
the simplified circuit has as many natural frequency as is the
circuits and R − C circuits in which case the solution of the
number of the “lost” solutions (otherwise we would need to
equation Z(ω) = 0 will be purely imaginary, corresponding to
study, which of the solutions of the simplified circuit are the
an oscillation-less exponential decay. In such cases, it is more
“lost” solution of the original circuit).
convenient to subsitute iω = −γ which corresponds to expo-
nential dependence of currents and voltages, proportional to Let us sum up what has been said above as
e−γt . With such substitution, the impedance of a capacitance
C becomes − γC 1
, and that of an inductance L — −γL; the idea 49: When finding the natural frequencies of a circuit
current resonant condition remains still the same, Z(γ) = 0. obeying certain symmetries while using the idea 48, it is useful
Finally, it should emphasized that with a low probability, to exploit the symmetry: select symmetric positions for the fic-
for symmetric circuits, one or more natural frequencies may be titious terminals (for which Z = ∞), or select symmetric point
lost with this method. Therefore, it is a good idea to compare for “cutting” (Z = 0). This will lead to the loss of one or more
the number of obtained frequencies ω(> 0) with the expected solutions which can be found as the natural frequencies of a
number (deduced using the idea 47). Losing one or more solu- simplified circuit — we either short-circuit the fictitious ter-
tions in such a way is actually not a bad thing, because losing minals (if Z = ∞ was used), or we leave the “cut” wire broken
a solution means obtaining a lower-degree equation for x = ω 2 (if Z = 0 was used). NB! count carefully the number of solu-
which is easier to solve. The “lost” solutions can be recovered; tions36 : if the number of degrees of freedom of the simplified
one option is to “cut” the circuit at a different place (or to circuit exceeds the number of “lost” solutions then some of the
add fictitious nodes to different places if the voltage resonance natural frequencies of the simplified circuit may differ from the
approach is used). If the resulting circuit is non-symmetric, frequencies of the original circuit.
the obtained equation for x = ω 2 will have its degree (and the
number of different solutions) equal to the number of degrees
To show how to count the number of modes and to make use
of freedom of the original circuit. Among the solutions, there
of the symmetry of the problem, let us consider the following
are also the ones we already know based on the symmetric cir-
problem.
cuit. Therefore, it would be possible to reduce the degree by
long dividing the corresponding polynomial with x − x1 , where
x1 is a known solution.
However, in many cases, there is actually no need to long pr 89. Find the natural frequencies of the circuit shown in
divide the polynomial, because an easier method exists. Be- the figure.
fore we proceed to this method, let us analyse first why one or
more solutions were lost. When the current resonance method
is used, some natural frequencies will be lost if the circuit is
“cut” at such a point A where one of the natural modes has (due
to symmetry) always zero current: for such an oscillation mode,
the circuit terminals A1 and A2 will have, in addition to a zero
voltage, also a zero current; hence the impedance does not need
to be vanishing. Similarly, when the voltage resonance method
is used, a natural frequency is lost if the corresponding mode
has always zero voltage between the fictitious terminals A and
B.
Keeping this is mind, it becomes evident how to recover the Using the fact 8, we can easily conclude that the maximal
lost solutions in the most efficient way. In the case of current number of linearly independent loops is four (red arrows in fig-
resonance, we use the fact that for the lost modes (which we ure).
36 When counting the number of frequencies, keep in mind that an equation for finding frequencies may have repeated roots; double root needs to
be counted twice, triple root trice, etc.

— page 30 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
already two solutions. However, that would have been long
way. Meanwhile, we know that the missing solution is, in fact,
very symmetric; would it be possible to use this fact? It ap-
pears, yes! We can make use of the voltage equalities VB = VE
and VD = VE to conclude that in the case of the missing os-
cillation mode, the loop currents 1–4 (as shown in the figure)
satisfy equalities I2 = I3 = 2I1 = 2I4 . With this knowledge,
we can write down Kirchoff voltage law, for instance, for the
first loop: (I1 + I4 )/(iωC) + (I1 + I2 )/(2iωC) + 2iI1 ωL from
Since the circuit consists of only inductors and resistors, we where we obtain easily the same result as before.
could expect the number of natural frequencies to be also
pr 90. [IPhO-1983] Let us consider the electric circuit in
four. However, we need to pay attention that one can find
the figure, for which L1 = 10 mH, L2 = 20 mH, C1 = 10 nF,
one current loop consisting only of inductors (blue arrow).
C2 = 5 nF, R = 100 kΩ. The switch K being closed, the circuit
Therefore, we have 4-1=3 non-zero frequencies, and 1 zero fre-
is coupled with a source of alternating current. The current fur-
quency. The circuit has mirror symmetry which we are go-
nished by the source has constant intensity while the frequency
ing to use: we use nodes A and D (see figure) for voltage
of the current may be varied.
resonance, and require ZAD = ∞. When calculating ZAD ,
L1
the nodes A and D serve as the terminals; then, due to sym- i 01
metry, potentials of the nodes B, O, and E are equal (because
C1 A
zAB /zBD = zAO /zOD = zAE /zED , where zXY denotes the
impedance of the component connected directly between the B
C2
nodes X and Y ), hence there is no current in the wires BO i 02
and OE which can be removed without changing the imped- L2
ance ZAD . Then there are three parallel paths connecting A K
and D, out of which the paths ABD and AED can be re- R
placed with a single equivalent inductor of inductance 23 L; the
two capacitors in the path AOD can be replaced with a single (a) Find the ratio of frequency νM /∆ν, where νm is the fre-
capacitor of capacitance 23 C. So, we resulted in a circuit with quency for which the active power in circuit has the maximum
one capacitor and one inductor,√ and which has clearly only one value Pm , and the frequency difference ∆ν = ν+ − ν− , where
resonance frequency ω1 = 1/ LC. Next, we’ll proceed in the ν+ and ν− are the frequencies for which the active power in
same way, but now we select the nodes B and E as the ter- the circuit is half of the maximum power P = 21 Pm .
minals for voltage resonance. In a very similar way, we end up, The switch is opened in the moment t when there is no current
0
again, with an equivalent circuit consisting of one inductor of through the resistor. Immediately after the switch is open, the
inductance 34 L and one capacitor of capacitance C; the corres- intensities of the currents in the coils L and L are respect-
√ 1 2
ponding natural frequency ω2 = 1/ 43 LC. ively i01 = 0.1 A and i02 = 0.2 A. (the currents flow as in the
We have found two frequencies, but there are three in total. figure); at the same moment, the potential difference on the
In the first step, we missed those natural modes which had zero capacitor with capacity C1 is U0 = 40 V.
voltage between A and D; in the second step, we missed those (b) Calculate the frequency of electromagnetic oscillation in
natural modes which had zero voltage between B and E. So, L1 C1 C2 L2 circuit;
now we can be missing only those modes for which both these (c) Determine the intensity of the electric current in the AB
voltages are zero, i.e. we can connect the nodes B and E with conductor;
a wire, and also connect the nodes A and D with a wire; the (d) Calculate the amplitude of the oscillation of the intensity
resulting circuit must have the missing frequency as one of its of electric current in the coil L1 .
natural frequencies. Once we merge nodes B with E and D The idea 38 is useful in the case of AC, as well; let us
with A, we have only three nodes: B (with E) is connected to formulate this as another idea, which can be used to find qual-
O via an effective capacitance 4C; A (with D) is connected to itatively or asymptotically37 the dependence of something on
O via an effective capacitance 3C; A (with D) is connected to the frequency of the input signal, or to simplify the analysis ac-
B (with E) via an effective inductance 13 L. This circuit has cording to the idea 90 in those cases when the circuit includes
only one degree of freedom and hence, its natural frequency
√ is both large and small inductances and/or capacitors.
4
the “lost” frequency of the original circuit, ω3 = 1/ 7 LC. idea 50: At the limit of low frequencies, capacitors can be
So, we were lucky in that the number of missing natural “cut off”, and inductors — “short-circuited”; similarly, at the
frequencies was equal to the number of degrees of freedom of limit of high frequencies, inductors can be “cut off”, and capa-
the simplified circuit. What should have been done if that were citors — “short-circuited”. Systematic analysis assumes that all
not the case? Of course, we could have written down the im- the appropriate limit cases are considered, e.g. for ω ≪ 1/RC,
pedance for a pair of non-symmetric terminals; in that case we |ZC | ≪ |ZR | and hence, if connected in parallel, the resistor
would have ended up with a cubic equation for which we know can be “cut off”, and if connected in series, the capacitor can
37 at the limit of high- or low frequencies

— page 31 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
be short-circuited. ℜz1 z2 equals to the scalar product of the respective vectors
Keep also in mind that at a voltage resonance, a parallel L − C z⃗1 and z⃗2 , and ℑz1 z2 equals to the z-component of the vector
connection has an infinite impedance, and can be cut off; sim- product z⃗2 × z⃗1 (assuming that the real axis corresponds to
ilarly, at the current resonance, a series L − C connection has the x-axis, and the imaginary axis — to the y axis). However,
a zero impedance, and can be short-circuited. regardless of what have been said, there are cases when it is
more convenient to deal with vector diagrams of voltages and
pr 91. In a black box with two ports, there are three com- currents, rather than with the complex amplitudes.
ponents connected in series: a capacitor, an inductance, and a idea 52: If an AC-circuit problem turns out to be essentially
resistor. Devise a method to determine the values of all three a geometrical problem, it is better to use vector diagrams in-
components, if you have a sinusoidal voltage generator with stead of complex amplitudes, i.e. to draw the vectors corres-
adjustable output frequency ν, an AC-voltmeter and an AC- ponding to the complex amplitudes, and to study the problem
ammeter. geometrically; keep in mind that using scalar product and rms.
amplitudes, P = U ⃗ · I.
⃗ Geometrical knowledge which can be
idea 51: Mathematically, electrical oscillations are ana- useful: Thales theorem, inscribed angle theorem, laws of sines
logous to the mechanical ones, which are usually studied by and cosines
writing down differential equations. Method of impedances al-
lows us to bypass this step. Based on the resonance condition pr 93. The circuit consists of a capacitor, inductance, and
Z(ω) = 0, it is also possible to make a “reverse-engineering” two resistors, see figure. The voltage on both resistors is 10 V,
and deduce the corresponding differential equation using a and the voltage between the leads A and B is also 10 V. Find
simple rule: a factor iω corresponds to a time derivative (hence, the applied voltage U0 .
a factor −ω 2 corresponds to a second derivative). The safest
method for verifying the equivalence of a mechanical system
with an electrical one is to write down the differential equations
(or systems of differential equations, if appropriate) describing
the both systems and verify that these two are mathematically
equivalent. The matching scheme is usually as follows: a ca-
pacitor’s charge or loop current’s time integral corresponds to
a coordinate of a point mass; an inductance — to a mass; a
Problems involving ideas 44–52
capacitance — to a spring’s stiffness. pr 94. Consider a so-called Maxwell’s bridge shown in fig-
ure below, which is used for measuring the inductance L and
pr 92. [IPhO-1987] When sine waves propagate in an infin- the ohmic resistance R of an inductor. To that end, the other
ite LC-grid (see the figure below) the phase of the AC voltage parameters are adjusted so that the voltage reading will be
across two successive capacitors differs by φ. zero. Assuming that such a state has been achieved, express L
(a) Determine how φ depends on ω, L and C (ω is the angular and R in terms of R1 , R2 , RC and C.
frequency of the sine wave).
(b) Determine the velocity of propagation of the waves if the
length of each unit is l.
(c) State under what conditions the propagation velocity of the
waves is almost independent of ω. Determine the velocity in
this case.
(d) Suggest a simple mechanical model which is an analogue pr 95. Below a circuit is given which makes it possible to
to the above circuit and derive equations which establish the adjust the phase of a voltage signal. Show that if the output
validity of your model. current is negligibly small, its voltage amplitude will be the
L L same as at the input leads, but with a different phase. Find
C C C
the phase shift.
l l

In general, when dealing with two-dimensional problems,

complex number analysis is a more powerful tool than a vector
analysis, because everything what can be done with vectors,
can be also done with complex numbers: addition and subtrac-
tion, multiplication with a constant, and even the scalar and
vector products (although this is a little bit more tricky, see be-
low), but there are many more things what can be done with pr 96. A remote summer house receives electricity from a
the complex numbers (addition, division, taking powers and power station over a rather long cable. To check the status
exponents, etc). A hint for the way of obtaining scalar- and of the cables, the power meter can also measure the voltage
vector products can be found in the idea 45: if we take two com- supplied to the household. People left the summer house, and
plex numbers z1 and z2 , and consider the product z1 z2 , then switched all the other electricity devices off, but forgot the
— page 32 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
transformer under the voltage (the transformer was used for
feeding low-voltage lamps). The transformer can be considered
as a series connection of an inductance L and ohmic resist-
ance r. Readings at the power meter of the house: when the
transformer was switched on, voltage U1 = 234.0 V and power
consumption P1 = 5 W;
when everything was switched off, voltage U0 = 236.0 V;
when the transformer is off, but an electric oven is switched on:
electrical power P2 = 1200 W, and voltage U2 = 219.6 V (oven
is a purely ohmic resistance R).

You may assume that the voltage at the power station (to which pr 101. Determine all the natural frequencies of the circuit
the cables are connected) is always constant. Determine the shown in Figure. You may assume that all the capacitors and
power of electrical energy which was dissipated in the power inductances are ideal, and that the following strong inequalities
cables (connecting the house with the power station). are satisfied: C1 ≪ C2 , and L1 ≪ L2 . Note that your answers
need to be simplified according to these strong inequalities.

pr 97. A circuit consists of two identical inductances, two

identical resistors, and two identical capacitors, see figure. The
applied voltage U0 = 10 V; the voltage on the lower inductance
is 10 V, and the voltage between the leads D and E is also 10 V.
Determine the voltage between the leads B and D.

pr 102. [Adapted from IPhO-1984] An electronic frequency

pr 98. A circuit consists of two identical inductances, two
identical capacitors, and one resistor, see figure. The applied
voltage is U0 = 10 V, and the total current at the input leads
is I0 = 1 A; the voltage measured at the left capacitor is 10 V,
and 10 V is also measured at the left inductance. What is the
active power dissipated in this circuit and what is the resistance
of the resistor?
filter consists of four components as shown in figure: there are
two capacitors of capacitance C, an inductor L, and a resistor
R. An input voltage Vin is applied to the input leads, and
the output voltage Vout is measured with an ideal voltmeter
at the output leads, see figure. The frequency ν of the input
voltage can be freely adjusted. Find the ratio of Vout /Vin and
the phase shift between the input- and output voltages for the
following cases: (a) at the limit of very high frequencies; (b)
at the limit of very low frequencies; (c) in the case of such a
frequency ν0 for which there is no voltage on the resistor; (d) in
the case of such a frequency ν1 for which the power dissipation
in the circuit is maximal (assuming that the amplitude of the
input voltage is kept constant). Find also the frequencies ν0
and ν1 .

pr 99. Find the natural frequencies of the circuit given be-

low.

pr 103. [IPhO-2014] Initially: switch S in the circuit below

is open; the capacitor of capacitance 2C carries electric charge
q0 ; the capacitor of capacitance C is uncharged; and there are
no electric currents in either the coil of inductance L or the
coil of inductance 2L. The capacitor starts to discharge and at
the moment when the current in the coils reaches its maximum
value, the switch S is instantly closed. Find the maximum
pr 100. Edges of a dodecahedron are made of wire of neg-
current Imax through the switch S thereafter.
ligible electrical resistance; each wire includes a capacitor of
capacitance C, see figure. Let us mark a vertex A and its three
neighbours B, D and E. The wire segments AB and AD are
removed. What is the capacitance between the vertices B and
E?
— page 33 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
√
pr 104. [Est-Fin-2016] A dimmer for controlling the bright- pr 105. In the circuit below, R = L/C. A rectangular
ness of lighting consists of a rheostat, a capacitor, a diac and voltage waveform of period τ and amplitude V0 is applied to the
a triac, connected as in the circuit. input ports (this means that during half of the period, the in-
put voltage is V0 , and during the other half-period, the voltage
I
: is −V0 ). Find the shape and amplitude of the current flowing
through the input ports.
R
∼ U, f −Ub −Ud I b U
Ud Ub
C

A diac is a component whose behaviour is determined by pr 106. Find the natural frequencies for the circuit below.
the voltage-current diagram shown above. A triac , on the
other hand, can be thought of as a switch controlled by current
— look at the following equivalent schematics.
anode 2
anode 2
≈ Kt
gate gate Rt
anode 1 pr 107. For the circuit shown below, the frequency of the
sinusoidal input voltage is unknown; given the capacitance C,
anode 1
inductances L1 , L2 , resistances R1 , R2 , amplitude of the input
The switch Kt is open as long as the current through the voltage V0 , and the phase shift φ between the currents through
triac’s gate stays under the threshold current It ; closes when the nodes A and B, determine: (a) the amplitude of the voltage
the threshold current is applied (in either direction) and stays VAD between the nodes A and D; (b) the phase shift between
closed while a current is flowing through the switch Kt (the the voltages VAD and VDB .
gate current is irrelevant until the switch opens again).
(a) Assume that the resistance Rt is large enough that the
charge moving through the diac can be neglected. Let the si-
nusoidal supply voltage have a maximum value of U and a
frequency of f ; the rheostat be set to the resistance R and the
capacitor’s capacitance be C. Find the maximum value of the
voltage UC on the capacitor, and its phase shift φ with respect
pr 108. For the circuit shown below, the frequency of the si-
to the supply voltage.
nusoidal input voltage is unknown; given the capacitance C, in-
(b) What inequality should be satisfied by the diac’s charac- ductance L, resistances R1 , R2 , amplitude of the input voltage
teristic voltages Ub and Ud , triac’s threshold current It and V0 , and the phase shift φ between the inductor current and
gate resistance Rt to ensure that when the diac starts to con- input voltage, what is the phase shift between the capacitor
duct (while the voltage on the capacitor rises), then the triac voltage and output voltage?
would also immediately start to conduct? You may assume
that Ib < It and that the diac’s voltage at current It is Ud .

pr 109. In the circuit shown in the figure, the sinusoidal

input voltage has a fixed amplitude V0 and frequency f . What
is the maximal amplitude of the output voltage, and for which
values of the variable resistances R1 , R2 , and R3 is the maximal
(c) The voltage Ul on the lamp follows the plot above. Let’s
amplitude achieved?
assume that the assumption of part i) and the inequality of
part ii) hold. Find the time t0 during which the voltage on the
lamp is zero.
(d) Express through t0 and f , how many times the average
power of the lamp is lower than the one of a lamp without a
dimmer, assuming that the resistance of the lamp is unchanged.

— page 34 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES

pr 110. Find such frequencies of the input voltage ω for For real-valued arguments, the property ea+b = ea ·eb is an easy

which the circuit shown below has zero impedance. generalization from the same rule for integer arguments. This
is a very useful property, and actually the only reason why the
exponential function is easier to deal with than sine or cosine,
but it is not obvious why it should held for complex-valued ar-
x
guments. Since we generalized e to complex-valued arguments
via the Taylor expansion, this series is the only thing we can
use to prove the validity of this property. So, we start with
∞
∑
a+b (a + b)n
e = ,
pr 111. Find such frequencies of the input voltage ω for n=0
n!
which the circuit shown below has zero impedance. where a and b are complex numbers. Here we need the binomial
theorem (see appendix 7),
∑n ( ) ∑
n
n n k n−k n!
(a + b) = a b = ak bn−k ,
k k!(n − k)!
k=0 k=0
which leads us to
∞
∑∑ n ∑∞ ∑ n
a+b n! ak bn−k ak bn−k
e = == .
n=0 k=0
k!(n − k)! n! n=0 k=0
k! (n − k)!
Now, let us substitute m = n − k ; in this double sum, as n goes
from 0 to ∞, and k goes from 0 to n, the pair of numbers k, m
appendix 5: Euler’s formula takes all the possible integer-valued combinations, with both
The standard way of generalizing a function F (x) from its real
m and k varying from 0 to ∞:
argument x to complex argument z is by using a Taylor expan- ∑∞ ∑ n k n−k ∞ ∑
∑ ∞
a b ak bm
sion: = .
∑ k! (n − k)! k! m!
F (x) = F (x0 ) + F (n) (x0 )(x − x0 )n /n! : n=0 k=0 k=0 m=0
This double sum can be factorised, because one factor doesn't
in such a power series, we can just substitute x with z . Here,
depend on m, and the other one  on k, and constant terms
F (n) (x0 ) stands for the n-th derivative of F (x), calculated at
(independent of the summation index) can be brought before
x = x0 . The structure of this power series is quite easy to un-
the summation sign (before the braces):
derstand: if we truncate it by keeping only the rst N terms, ∞ ∑ ∞ ∞
( ∞ )
∑ k m
a b ∑ ∑ k a bm
it approximates the function F (x) with such a polynomial of = =
k! m! k! m!
N -th order for which the rst N derivatives at x = x0 are equal
∞
[ ( ∞
k=0 m=0 )] ( ∞ k=0 )( ∞
m=0 )
to those of the function F (x). Furthermore, the thrown-away
∑ ak ∑ bm ∑ ak ∑ bm
= = ea · eb .
terms with n > N are small if |x−x0 | is not very large, because k! m=0 m! k! m!
k=0 m=0 k=0
the denominator grows rapidly with n. By keeping more and
more terms, the approximation becomes increasingly accurate, appendix 7: Newton binomial formula
so that at the limit N → ∞, the series becomes equal to the If we open the n braces in the expression (a + b)n , we'll have a
function. sum of terms where each term is a product of n factors, each
k n−k
of which is either a or b. In that sum, a b will arise as
With x0 = 0, the Taylor series for the exponent, sine, and
many times as many dierent possibilities there is for select-
cosine functions are written as
∑∞ ing exactly k braces out of the total n braces: from the se-
x xn
e = , lected braces we pick a as the factor entering a term in the
n=0
n!
de-factorised sum, and from the non-selected braces we pick
∞
∑ (−1)k x2k+1 b). This so-called number of k -combinations from a set of n
sin(x) =
(2k + 1)!
, ( n)
elements is denoted by
k=0 k .
∑∞
(−1)k x2k In order to nd the number of possibilities for selecting k ob-
cos(x) = .
(2k)! jects out of n objects, let us enumerate all the objects with
k=0
Now, if we substitute in the expression of ex the argument x numbers from 1 till n. The number of permutations (dierent
with ix, we obtain ways for ordering these enumerated objects) is n! (there is n
∑∞ ∞ [
∑ ]
ix (ix)n (ix)2k(ix)2k+1 dierent ways for picking the rst object, n−1 for picking the
e = = + =
n=0
n! (2k)! (2k + 1)! second, etc). In the case of each ordering of the objects, we
k=0
∑∞ [ 2k 2k ] select the rst k ones. If we go through all the dierent or-
i x i · i2k · x2k+1
= + . derings, we denitely obtain all the dierent ways of selecting
(2k)! (2k + 1)!
k=0 k objects, but each selection will be obtained many times: as
2k
Let us notice that i = (i2 )k = (−1)k ; if we compare now the
many times as we can re-order objects within a given selection.
series expansion for eix , and those of sin x and cos x, we see
The selected object can be re-ordered in k! dierent ways, and
that
the non-selected objects  in (n − k)! dierent ways. In order
eix = cos x + i sin x.
to obtain the number of dierent ways of selecting k objects,
we divide the overall number of permutations by the number
( n)
appendix 6: Exponent of a sum of two complex n!
of dierent ways of re-ordering, which results in
k = k!(n−k)! .
numbers
— page 35 —
 2.
appendix 8: Basic properties of complex numbers
Complex numbers can be thought of as two-dimensional vec-
tors: the real part of a complex number z = x + iy
the x-coordinate of a vector, and the imaginary part  the
y -coordinate. What diers complex numbers from vectors is
that two complex numbers can be multiplied so that the result
is still a complex number (there is a vector product of two vec-
tors, but if we have two-dimensional vectors lying in
the resulting vector will no longer lie in that plane). Because
of that, you can also divide two non-complex numbers with a
uniquely-dened result  as long as the divisor is not zero (you
cannot divide two non-parallel vectors!).
CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
so this set of linear equations is a homogeneous system which
has non-zero solutions (which are the natural oscillation modes)
denes only if its determinant is zero. Equating the determinant of the
system to zero gives us an algebraic equation for nding the os-
cillation frequencies ω. The degree of the equation is found
as the sum of the degrees of the individual equations. An in-
dividual equation describing a certain current loop has degree
x−y -plane, 2 if it includes at least one capacitor and one inductor; it has
degree 1 if it includes at least one resistor together with at least
one capacitor or at least one inductor; nally, it has degree 0
if all its elements are of the same type (resistors, inductors, or
capacitors). If we denote the number of such loop types as

The modulus of a complex number is dened as the length

NLC , N̸= , and N= respectively, the degree of the equation is
√
expressed as
of the corresponding vector, |z| = x2 + y 2 . Bearing in mind
the geometrical (vectorial) representation and using Euler's for-
Nd = 2NLC + N̸= .
According to the fundamental theorem of algebra, the num-
mula, we can write

z = |z|(cos α + i sin α) = |z|eiα , ber of complex-valued roots of such equation is also Nd . How-
ever, not all roots will correspond necessarily to dierent os-
where α is the angle between the vector and the x-axis; this
cillation modes. Indeed, if there is a pair of solutions in the
is called the exponential form of a complex number, and α is
form ω = ±ω0 − γ , the corresponding current oscillations can
called the argument (arg) of the complex number. Apparently,
be combined into one mode
α = arctan y/x = arctan ℑz/ℜz.
I1 eiω0 t−γt + I2 e−iω0 t−γt = I0 e−γt cos(ω0 t + φ),
Now, if we consider the product of two complex numbers,
where φ is the oscillations phase. Conversely, if there is an os-
z1 · z2 = |z1 |eiα1 |z2 |eiα2 = |z1 ||z2 |ei(α1 +α2 ) . cillating mode (possibly decaying-in-time) I0 e
−γt
cos(ω0 t + φ),
Here, the right-hand-side of the equality is an exponential rep-
there must be two corresponding solutions of the algebraic equa-
resentation of the complex number z1 z2 ,which means that
tion for ω (with equal imaginary parts and opposite real parts).
|z1 z2 | = |z1 ||z2 |, To sum up, if we denote with No the number of oscillatory
and modes, and with Ne the number of such modes which decay
arg z1 z2 = arg z1 + arg z2 . without oscillations, the following equality holds:
Similarly, of course, |z1 /z2 | = |z1 /z2 | and arg z1 /z2 = arg z1 −
2No + Ne = 2NLC + N̸= .
arg z2 .
Finally, the number of zero-frequency modes equals to the num-
Here is a list of simple but sometimes useful formulae: ber of such linearly independent loops which have only induct-
1
ℜz = (z + z), ors in it, because in that case, with ω = 0, the corresponding
2 loop current is eliminated from all the equations so that it can
where z = x − iy is called the complex conjugate of z;
take arbitrary value.
|z| = zz.
2

Note that z is a vector symmetric to z with respect to the

x-axis, and therefore
Hints
eiα = e−iα ;
in particular, applying these two formulae for z = eiα results
1. Determine the surface area under the graph(count the cells
in
or approximate the shape with a set of trapezoids; pay at-
eiα + e−iα eiα + e−iα tention to the units of your surface area (mm·m).
cos α = , sin α = .
2 2i 2. The sum of the voltages on R1 and R2 is constant,
If you need to get rid of a complex number in a denominator
of a fraction, you can use equality
I1 R1 + I2 R2 = Const, hence one can find the change of
z1 z 1 z2 current through R2 . Initially, I1 = I2 ; later, the difference
= .
z2 |z2 |2 of these two currents goes to the lamp.
3. Resistors 4 and 5 are connected in parallel between B and
appendix 9: Determining the number of natural C.
modes of electrical circuits 4. Due to symmetry, there is no current through the bridge
If we need to nd the natural frequencies of a certain circuit, it
resistor, hence it can be removed (the both leads of it
can be done by using the method of loop currents: we need to
divide the overall voltage between the input leads of the
select a full set of linearly independent current loops (cf. idea
circuit in 2:3-proportion).
13), consider the corresponding loop currents (which are now
AC currents dened by complex amplitudes), and write down
5. Find the currents in the upper resistors (2 Ω and 3 Ω) by
for each current loop the Kircho 's voltage law. This will give
short-circuit the ammeter; the difference of these two cur-
us a set of N linear equations for nding N unknown currents,
rents goes to the ammeter.
where N denotes the total number of current loops. For nat- 6. For a voltmeter, the reading is proportional to the current
ural oscillation modes, there are no AC sources in the circuit, through it. Hence, you can find the current through V2 ;
— page 36 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
use the Kirchoff’s current law for finding the ammeter’s vertex of the vertex P , and I/19 is driven into all the other
current. 19 vertices.
7. The sum of the voltmeters’ readings is related to the sum 22. Reduce this problem to the previous one by representing
∑15 ∑15 ∑15
of their currents: i=2 Vi = i=2 rIi = r i=2 Ii , which the missing wire as a parallel connection of R and −R: the
equals to A2 , as it follows from the idea 8 for the region resistance of all the other resistors, except for the −R, is
marked with a red line in the figure below; here, r is the given by the answer of the previous problem.
voltmeters’ internal resistance. 23. For the first half of the process, the transitions of the states
of the resistors will take place at the overall voltage values
V1 = 1.5 V and V2 = 5 V; for the second half, the respect-
ive transition voltages are V2′ = 3 V and V1′ = 1.25 V.

8. The left three resistors form a ∆-connection which can be 24. The diode current can be expressed as E − IR; draw this
substituted by a Y -connection consisting of 1 Ω-resistors. straight line onto the graph provided, and find the inter-
section point.
10. Substitute the entire circuit in figure with an equivalent
battery with Eeff = E r+RR12+R2 . 25. Find the tunnel diode current exactly in the same way as
for problem 24; let the intersection point voltage be V0 .
11. When solving using the idea 10: first find the internal
Find the cotangent Rt = ∆V /∆I of the I(V )-curve at
resistance of an equivalent battery; the equivalent electro-
that voltage (note that Rt < 0). Substitute the tunnel di-
motive force will be found by comparing the currents of
ode with a series connection of a battery of electromotive
the two systems when the ports are short-circuited. When
force V0 and resistance Rt . If the input voltage changes
using the idea 11: substitute all the batteries with current
by ∆V , the current will change by ∆I = ∆V /(R + Rt ),
sources; for a parallel connection of current sources, the
and the output voltage will change by R∆I, hence the
currents are just added, so it is easy to substitute a set
amplification factor n = R∆I/∆V .
of current sources with a single equivalent current source
(and at the final step, the current source with a battery). 27. Keep 4 more wires(four red ones in the left figure below);
short-circuit only four nodes as shown in figure; then, due
12. Assume that E1 is short-circuited and calculate the cur-
to symmetry, the middle point C of the green wire can be
rents through the batteries I1n , n = 1 . . . 4; then as-
also merged with the short-circuited nodes (because both
sume that E2 is short-circuited and calculate the currents
will have the same potential if a voltage is applied between
through the batteries I2n n = 1 . . . 4. The final answers
the leads A and B).
will be I1n +I2n , n = 1 . . . 4. Alternatively, show that after
mirroring the circuit with respect to the vertical axis, it re-
mains identical to itself, but the currents through R2 and
R4 will be reversed, hence these currents need to be zero,
hence these resistors can be “cut” off.
15. As mentioned, the numerical values are such that in SI-
units, each “resistance” of the dual circuit is numerically 28. For the original polarity, the two leftmost diodes are open,
exactly 4 times smaller than the corresponding resistance i.e. these can be short-circuited (resulting in a parallel
of the original circuit. Therefore, the total “resistance” of connection of the three leftmost resistors), and the third
the dual circuit must be also 4 times smaller, 4R⋆ · Ω2 = R. diode is closed, i.e. can be “cut” off. For the reversed po-
On the other hand, the “resistance” of the dual circuit larity, the states of the diodes are reversed: the rightmost
equals to the resistance of the original circuit, R⋆ = 1/R; diode is open and short-circuits the four rightmost resist-
from here, we can immediately obtain the answer. ors (which can thus be removed), and the other diodes are
17. Assume that the circuit is equivalent to a battery of elec- closed so that the three leftmost resistors are connected in
tromotive force E and internal resistance r; then write series.
equations for E and r analogously to how it was done for 29. The voltage on the first diode is 1 V, hence the other di-
problem 16. odes have a lesser voltage applied (resistors take also some
18. If a current I is let into one of the leads, it is distributed voltage), and are closed. As a result, the first diode can
equally between the three branches: I/3 flows in each. At be replaced by a battery of 1 V, and the other diodes —
the next junction, each of these currents is divided equally, “cut off”. The power dissipation on the first diode is found
again, so that the next wires have current equal to I/6. as its voltage (1 V) times the current through it.
19. For node-merging: merge B with F and C with E so that 30. Calculate the electromotive force and internal resistance of
A will be connected with the merged BF node (and BF the battery which is equivalent to the set of four resistors
with CE) via a R/2-resistance. For edge-splitting: split and the battery; internal resistance calculation can be sim-
OD into resistors OD′ and OD′′ , each of resistance 2R. plified, if the three ∆-connected resistors are substituted
21. Consider symmetric current distributions: (A) I is driven with a Y -connection.
into one vertex (P ), and I/19 is driven out from all the 31. Substitute the three resistors and the battery with an equi-
other 19 vertices; (B) I is driven out from a neighbouring valent battery with an internal resistance r; thus the prob-
— page 37 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
lem is reduced to Pr. 24.
32. Note that IA /IB = RB /RA = 2, hence the fuse A will
melt first. Pay attention that in the second case, the fuse 45. Apply the solving technique of problem 20.
B tolerates more current than the overall current by which
the fuse A melts.
33. The circuit breaks down into a combination of series- and
parallel connections. Using the given voltage values, one
Rr
can conclude that R+r = 23 2R+r
2Rr
, where r is the volt-
meter’s resistance.
34. The radiated heat AT 4 equals to the electrical power V I;
also, R = V /I = BT (where A and B are constants).
35. First we apply the idea 6 and substitute the ammeters
with wires; then we apply the idea 5: the two vertically
positioned resistors connect directly the battery leads; the
other four resistors form two pairs of parallel-connected
resistors (these pairs are connected in series).
36. Apply the idea 6: the circuit becomes a combination of
parallel and series connections; find ammeter current us-
ing Kirchoff’s current law.
37. Apply the idea 6; find ammeter currents using Kirchoff’s
current law (keep in mind that due to symmetry, all the
ammeters have the same current by modulus).
38. The voltage distribution between the voltmeters in this
circuit is defined by the ratio of the resistor’s resistance
R and the voltmeter’s internal resistance r. If we express
the ratio V3 /V1 = 0.8 in terms of r/R (the battery voltage
cancels out from this equation), we obtain a square equa-
tion for x = r/R. Once we know r/R, we can also find
V2 /V1 .
39. Apply the result of the problem 11 for the particular case
of three batteries (one of which has E = 0).
40. Since there is no current through the ammeter, it can be
“cut” off. Additionally, since there was no current in the
ammeter, the voltages on R1 and R2 are equal; owing to
this, we can find the voltage V2 on R2 . In order to find
the voltage on R3 , we need its current, which goes also
through R2 and can be found as V2 /R2 .
41. Apply the idea 1 for the two cases (7 lamps and 8 lamps).
54. Notice that a wheel graph is self-dual, so all we need to
42. Solution is completely analogous to 35.
43. Apply the idea 9: it is not good to combine A1 , A2 and
A3 into a Y , because we loose information supplied by the
problem conditions (currents I1 and I2 would be merged
into a single wire of the equivalent Y -connection). There-
fore, we substitute A2 , A3 and A4 with a ∆-connection
(with each resistor having a resistance of 3r). We know the
sum of currents I2 in the two wires of the ∆-connection,
and we can find the current through that 3r-resistor which
is parallel to A1 as I1 /3 (see idea 1). Hence we can de- 55. First, we need to write down expression for the resistance
termine all the currents (knowing the currents, it is easy
to find also R/r). Alternatively, the problem can be solved
by making use of the idea 13.
44. For the first polarity, the leftmost diode short-circuits all
the other resistors, except for the two at the left upper
corner of the circuit. For the reversed polarity, the left-
most diode is closed (can be “cut” off), and the next diode
— page 38 —
short-circuits five resistors which remain rightwards from
it.
46. Apply the solving technique of problem 20.
47. Apply ideas 22 and 23 (use a negative resistor between
B and C); calculate all the pair-wise resistances (A − B,
A − C,and B − C) for the symmetric lattice; apply idea 9
to “cut off” the wire between B and C.
48. Apply the solving technique of problem 21.
49. Consider this decagon (with few missing sides) as the deca-
gon of the problem 48, which has additionally negative
resistances −R connected parallel between the sides AB
and BC (idea 23). Further, apply the generalized idea 9:
represent the decagon of the problem 48 as a Y -connection
of 10 identical resistors. Finally, calculate the resistance
using the idea 1.
50. (I) leave the wires as shown below by red, blue and green
lines; (II) short-circuit the nodes as shown by black lines
below.
For a precise calculation, consider this side-less octagon as
the octagon of the problem 48, which has additionally neg-
ative resistances −R connected parallel to the sides (idea
23). Further, apply the generalized idea 9: represent the
octagon of the problem 48 as a Y -connection of 8 identical
resistors. Finally, calculate the resistance using the idea
1.
51. Note that removing the first cell of this infinite chain will
increase its resistance exactly two times; use this fact to
apply the idea 18.
52. Use the same approach as for problem 17.
53. Apply the idea 27: short-circuit (and later, cut off) all
these wires which are not known to be intact; the ideas 20,
21, and 5 will be also useful.
do is to use the idea 17. More specifically, assume that
in the case of one of the spokes, there is additionally a
battery connected in series to the resistor r. In the dual
circuit, spokes become rim segments (of “resistance” 1/r)
and vice versa. So, the circuit is almost self-dual (with
resistance-to-“resistance” ratio of Rr, just the “battery”
is now on a rim segment (and not on a spoke as the ori-
ginal battery). There is also an alternative solution which
exploits the result of the generalized problem 55.
between i-th node and j-th node if there is a direct con-
nection between these two, and this can be done easily:
Rij = (φ
(ij) (ij)
− φ )/I, where φ
(ij)
denotes the potential
i j k
of the k-th node when a current I is driven into the i-th
node, and driven out from the j-th node. Our plan is to
add up all the equations, and because of that, we don’t
want to have in our expression “if”-conditions, because it
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
is difficult to expect that such expressions will cancel out 64. During the first half-period when K is closed, the diode
when a sum is taken. The solution here is to multiply receives the reverse voltage of the battery and, hence, is
this expression by the conductance σij of a direct connec- closed; according to the idea 37, the current through the
tion between the two nodes: if there is a direct connection, diode grows linearly according to the applied voltage 5 V.
σij = 1/R, and the result will be Rij /R. If there is no At the beginning of the second half-period, when the key
direct connection, the result will be zero, and hence, there is opened, this current will be re-directed through the di-
will be no contribution to the sum. The second issue is ode; the diode will open since this is a forward current.
(ij)
that if we keep using the potentials φk , nothing will can- Now, the inductor will receive voltage −7 V, which corres-
cel out upon taking a sum: for each resistance Rij , we ponds to a linear decrease of the current. Once the current
introduce a new distribution of potentials (as indicated by reaches 0 A, the diode will close again (a reverse voltage of
the upper indices). We clearly need to reduce the num- −7 V will be applied to it). The average charging current is
ber of potential distributions. Luckily, this can be done found as the charge through the battery (the surface area
by using the superposition principle (similarly to the idea under the I(t)-graph for the second half-period), divided
22): we introduce a reference node, let it have index i = 1; by the period.
(k)
we consider n − 1 potential distributions φj (2 ≤ k ≤ n)
65. (a) Initially, there was no charge nor current in the system;
when a current I is driven into the i-th node, and out of
(ij) (i) (j) hence, immediately after the key is closed, there is still no
the 1st node. Then, φk = φk − φk . What is left to do,
∑ current in L, according to the idea 38, we “cut” it off, and
is to take a sum i,j Rij ; note that the summation order
∑ ∑ ∑ ∑ there is still no charge nor voltage on the capacitor, hence
can be chosen as needed, either i j Rij or j j Rij ;
we short-circuit it; for such a simplified circuit, we can
don’t forget that with this sum, each node pair is counted
easily calculate the ammeter current.
twice.
(b) Once a new equilibrium is achieved, according to the
56. According to the idea 31, perturbation current I˜ is de- idea 38, we “cut off” the capacitor, and short-circuit the
scribed by a circuit which we obtain if we substitute the inductance.
diode with a resistance Rdiff and remove the electromotive (c) For the final part, the inductor will retain the current
force. For this equivalent circuit, use the fact 17. of part (b), hence we substitute with the respective ideal
58. Repeat the solution of pr 57 while using the ideas 32, 33, current source; the capacitor will retain the voltage of the
and 32 (which means that the energy loss at the diode part (b), hence we substitute it with the respective ideal
equals to Vd ∆q, where ∆q is the charge change). battery. The circuit is further simplified by using the idea
59. Use idea 32 to obtain an equation for the final voltage; 6.
apply also ideas 33, and 32, and notice that at the initial 66. (a) During each half-period, the capacitor will reach very
and final states, there is no current, hence no energy of fast (as compared with the period length) a stationary
the inductor. state: constant charge on the capacitor means no current,
61. Initially, all the capacitors are charged to the voltage 13 E, so that all the current flows through the resistor.
i.e. q = 13 CE. The total charge of the system “right plate (b) According to the idea 38, capacitor remains essentially
of the 1st + left plate of the 2nd capacitor + right plate of short-circuited, so that almost all the current goes to the
the 3rd capacitor” (the system A) is conserved; at the final capacitor plates (and nothing goes to the resistor). Hence,
state, when there is no current through the resistor, the the charge q on the capacitor can be found using the idea
system A will be equipotential, and the applies to the sys- 39; V = q/C.
tem B (consisting of all the other capacitor plates), hence (c) Use the graph from part (a); keep in mind that the
all the plates of the system A will have the same charge. amplitude of oscillations is half of the difference between
62. Part (a): apply ideas 32 and 35: for dI the minimal and maximal values.
dt = 0, there is no
voltage of L, and hence no voltage on C2 , and hence a (d) There are still the same saw-tooth oscillations as in
voltage E on C1 . part (b), but the value V0 around which the voltage oscil-
Part (b): use the same approach by noting that when lates changes slowly, until a completely periodic behaviour
there is an extremal voltage on C1 , there is also an ex- is reached; the stationary value of V0 can be found by ap-
tremal voltage on C2 (as the sum of these voltages gives a plying the idea 40. the mean current
constant E). Hence, there is no current through the wires 67. Characteristic time L/R is much larger than the period:
leading to C1 , and there is no current through the wires the current in the inductor will remain almost constant
leading to C2 ; from the Kirchoff’s current law, there is also during a period. Suppose the AC input voltage is switched
no current through L. on; first, there is no current in inductor, it is as if “cut off”.
63. Apply idea 36: find the voltage on C1 for the stationary The diode is opened during a half-period: then, the diode
state of the system (constant I implies no voltage on L, works as a resistance-less wire, and the inductor’s voltage
hence full E on C1 ). Next, short-circuit E (C1 and C2 be- equals to the input voltage; during the other half-period,
come parallel), and find the free voltage oscillations on C1 the diode is closed, “cutting off” the right-hand-part of the
in the form V = V0 cos(ωt+φ) (what is ω?) and determine circuit: since there is no current in the inductor, there is
V0 and φ from the initial condition (i.e. initial values of V also no current in the resistor, and hence, no voltage on the
dV
and dt ). inductor. This means that when averaged over the entire
— page 39 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
period, there is a positive voltage on the inductor: the in- 73. (a) Apply the idea 38: during each half-period, a new equi-
ductor current starts slowly increasing. Constant inductor librium will be reached, and the dissipated energy can be
current means that the diode will remain open for a longer found using the idea 32.
time than a half-period. Now, in order to apply the idea (b) Apply the idea 39: the capacitor’s voltage VC remains
40, sketch a graph for the inductor voltage as a function essentially constant, hence the resistor’s voltage will be
of time, and try to figure out, under which condition the V2 − VC , or V1 − VC , depending on the half-period. Upon
average inductor voltage will become equal to zero. long-term evolution, the average current through the ca-
68. Notice that the external magnetic field changes from pacitor will be equal to 0, hence VC = 21 (V1 + V2 ).
−N BS to +N BS (or vice versa), and that the self-induced 74. According to the idea 36, short-circuit the battery, upon
flux is 0 both at the beginning and at the end (although which the parallel/series connection of the three resistors
takes non-zero values in between); apply the idea 41. can be substituted by a single equivalent resistor.
69. (a) Notice that if r = 0, there is a superconducting loop 75. To begin with, notice that RC needs to be large, so that
containing L and r, which means that I2 = Const); use the capacitor will keep almost a constant voltage during
the Kirchoff’s current law to obtain I1 for t < t3 ; notice each period (otherwise there would be large current fluctu-
that for t = t3 , I1 = 0, so that even for t > t3 , there is no ations). Next, the capacitor’s voltage will be equal to the
voltage on r, hence I2 = Const. maximal voltage of the sinusoidal input voltage (the diode
(b) At the moment t = 1 min, the characteristic time of is opened once during each period, when the input voltage
the circuit containing L will drop from infinity to a new is maximal). Now, the resistance R can be found by com-
value (use the idea 36 to find it); analyse now the prob- bining the Kirchoff’s voltage law (to obtain the resistor’s
lem using the idea 38 (the system will relax to the new voltage), and the Ohm’s law. During each period when the
equilibrium state with the above mentioned characteristic diode is closed, the capacitor’s voltage will decrease by the
time). amount ∆V corresponding to the charge ∆Q which flew
(c) There are three stages: first, reduce the current in r through the light emitting diodes. On the other hand, the
(by increasing the total current) so that it falls below 0.5 A; allowed voltage variations can be expressed in terms of the
second, switch it to a normal conduction state and while allowed current variations by using the Ohm’s law.
keeping I1 small, reduce I2 by reducing the total current 76. (a) Apply the ideas 38 (cut off the capacitors) and 6 (cut
1 −1
I2 (in order to keep I1 small, I1 = L dI dt rn needs to be off the voltmeter).
small, i.e. the process needs to be slow); third, switch r (b) Use the same approach as in part (a): substitute capa-
back to superconducting state. citors with batteries and inductors — with current sources.
(d) First step is the same as for (c), second step is to in- (c) Apply energy conservation law (notice that the circuit
crease the total current further (from 20 Ato30 A) while breaks down into two independent circuits, so that the
r = rn , third step is to make r = 0 and to decrease the power dissipation can be calculated separately for each of
total current back to zero. the circuits).
70. (a) Use the idea 43-i: the magnetic flux L(I1 + I2 ) in the 77. (a) Use the fact 5: P (t) = V · I(t) , where V is the diode
ferromagnetic core (cf. fact 20) does not change instant- voltage, which is constant and can be brought before the
aneously, and immediately before switching, the flux was braces (i.e. the averaging sign), ⟨P ⟩ = V ⟨I⟩; apply the
zero (this gives us our first equation). Following the idea idea 40 (together with the Kirchoff’s current law) to con-
43-ii, the inductive electromotive forces can be eliminated clude that the average current through the diode equals to
from the Kirchoff’s voltage laws (written for the two loops), the current through the load.
giving us a second equation. (b) Proceed in the same way as in Problem 75; do not
(b) We just need to apply the idea 43-ii. More specifically, forget the diode opening voltage 1 V, which needs to be
the Kirchoff’s voltage laws for the two loops serve us as subtracted from the AC voltage amplitude to obtain the
a system of differential equations for the two currents I1 maximal voltage of the capacitor.
and I2 ; if we multiply one equation by r, another one by R, (c) Proceed in the same way as in Problem 75.
and add the two equations, we end up in a single equation (d) Apply the idea 29 to the first period when the capacitor
for I = I1 + I2 ; the solution of that equation can be found is charged from zero volts up to the full working voltage.
using the fact 17 and idea 36.
78. (a) This problem is very similar to Pr 64: the inductor re-
71. Apply the idea 38: immediately after the key is closed, ceives a constant voltage Ui when the key is closed, U0 −Ui
all the inductor currents are the same as before, which when the key is opened and there is a forward current
you can find from the Kirchoff’s current law; knowing the through the diode, and 0 V when the key is opened and
inductor currents, you can also find the currents in the there is no current. This corresponds respectively to a lin-
lamps. early increasing, to linearly decreasing, or to a constantly
72. Notice that L1 is short-circuited, hence its voltage is zero zero input current. Output current can be found from the
and its current is constant; keep this in mind while using Kirchoff’s current law.
the idea 32 to find heat; for finding the charge, consider (b) See the hints of Pr 64.
the Kirhoff’s voltage law for the loop involving the upper (c) Apply the ideas 38, 40, and Kirchoff’s current law to
wire and inductance L2 , use the idea 41. conclude the resistor current equals to the average output
— page 40 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
current from part (b); then apply the Ohm’s law. current can enter the inductor L and resistor R: this part
79. (a) Consider separately two cases: SG is open, we have a of the circuit can be “cut off”. Similarly, for such fast
simple RC-circuit with a battery; SG is closed, essentially processes, the impedance of the capacitor C is negligibly
short-circuiting the capacitor and almost immediately dis- small, so it can be short-circuited in the equivalent circuit.
charging it. Otherwise, the analysis repeats the steps of task (a).
(b) During the charging cycle, the capacitor current needs 83. (a) resistances are found straightforwardly as the cotan-
to remain almost constant; this current is defined by the gents of the line segment slopes;
voltage falling onto the resistor throughout the cycle. (b) use the idea 24 and determine the number of solutions
(c) Charging current times T gives the break-down charge as the number of intersection points between the I − U -
of the capacitor; equate this to the value defined by the curve and the straight line describing the Kirchoff’s voltage
break-down voltage and C. law;
(d) Look at your expression for T . (c) apply the ideas 31 and 26 (cf. problem 56);
(e) Notice that amplitude equals to V0 /2. (f) Notice that (d) apply the idea 29 together with the fact 9: notice that
the required waveform can be obtained from the waveform as long as the thyristor operates at the lower branch of
Vb (t) from the question (b) as V0 − Vb (t); construct circuit the V-I-curve, there will be always a positive capacitor
which gives such an output. current charging it, and as long as the thyristor operates
80. (a) Apply fact 19; if V = V0 cos(ωt), current can be ex- at the upper branch, there is a is negative capacitor cur-
dq
pressed as I = dt , where q = V C. rent discharging it;
(b) Apply the ideas 29 and 36: consider separately the (e) apply the idea 29 together with the facts 12 and 13;
cases of forward and reverse currents; in both cases, there (f) since were are asked about an estimate only, there are
is a sinusoidal signal with a shifted symmetry axis. many methods what can be applied, e.g. calculating the
(c) The system will stop if there is no current, i.e. when minimal and maximal instantaneous dissipation power val-
−Vd < V < Vd . ues based on the cycle drawn for task (d) and approxim-
(d) Use your graph for question (b) to find the corres- ating the average power with the average of the minimal
ponding change of the capacitor’s voltage ∆V (of course, and maximal values;
∆q = C∆V ). (g) with battery voltage E ′ , there is a stationary state for
(e) If q corresponds to the coordinate x of a spring-mass the system at the lower branch of the V − I-curve, but not
oscillator then to which physical quantity X will corres- at the upper branch; if the battery voltage is increased, the
pond a voltage applied to an inductor? If the current is stationary state disappears and the system starts evolving
positive then what can be said about the mechanical sys- as described for task (f); the behavour depends how fast
tem? Zener diode provides a constant voltage if current is we switch the voltage back: was there enough time for the
positive; what type of X would have an equivalent prop- system to jump over to the upper branch or not [for more
erty? detailed analysis, use the same methods as in the case of
the task (d)];
81. (a) During the linear growth of B (from 0 ms to 10 ms),
(h) write down the qualitative criterion found for task (g)
the coil serves as an ideal battery of emf. N S dB dt . Estim- using the facts 12 and 13.
ate the characteristic times for two current loops: first
involving the coil and C, and second, involving the coil 84. (a) Write down the Kirchoff’s current law for the loop con-
and L; compare this with the time-scale of 10 ms. Apply taining L and E and keep in minde idea 37;
the idea 38. (b) when K1 opens, the current flows in loop L-R: study
(b) Now, the coil operates essentially as a wire (as its in- the behaviour of current in time, keep in mind the idea 17
ductance is negligible); the current in R1 is defined by the and pay attention to the fact that L/R ≪ τK ≪ τL ;
voltage obtained by the capacitor during the first 10 ms, (c) recall the intermediate result of the previous question
and the current in R2 is defined by the current induced in — how behaves I(t) on the inductor;
the inductor during the first 10 ms. (d) use the idea 29; charge can be found by either using
(c) apply the idea 41 for the current loop consisting of the the idea 41, or direct integration of the I(t)-dependance
coil, L, and R2 . (cf. fact 18);
(e) apply the idea 32: compare the energy released by the
82. (a) According to the idea 31, perturbation current I˜ is de-
inductor and the heat dissipation by keeping in mind that
scribed by a circuit which we obtain if we substitute the
in average, the capacitor’s energy remains constant;
diode with a resistance Rdiff and remove the electromot-
(f) study, how large charge is lost on capacitor when the
ive force. Further we apply idea 48; for instance, we can
diode is closed: keep in mind the idea 39.
“cut” the circuit near the battery and equate the imped-
ance of the resulting circuit segment to zero, Z(γ) = 0 85. (a) Apply the idea 43-i: the fluxes of parallel currents add
with γ = −iω. According to the idea 31, the system is up destructively, hence, at the first moment, the fluxes in
stable if all the solutions are stable, so all the roots for γ the two coils must be equal and parallel; keep in mind that
must have a non-negative real part. the total flux in both coils are equal, hence their voltages
(b) For very fast perturbations, the impedance of the in- are also equal.
ductance L can be considered infinitely large, hence no (b) Follow the idea 43-ii: write down Kirchoff’s voltage
— page 41 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
laws, eliminate current derivatives, express one current sides of the circuit to infinity).
through the other and substitute this expression back to (c,d) Note that there is a zero-frequency-mode: a constant
one of the Kirchoff’s equations, solve this differential equa- current can circulate in the loop formed by the two induct-
tion for the initial conditions found in task (a), cf. fact ors. The total current is the sum of such a constant cur-
17. rent, and a sinusoidally oscillating current; use the initial
(c) Note that a current flowing along the loop defined by conditions (the values of i01 , i02 , and U0 ), together with
the two coils does not cause any flux, hence it can be the Kirchoff’s current law, and the fact that the ratio of
switched on instantaneously without causing any voltage capacitor currents equals to the ratio of the capacitances,
on the inductors. to find the respective amplitudes.
86. Apply ideas 44 and 45: calculate the ohmic resistance 91. Study the behaviour of the impedance of the black box
R of the gun from the nominal values, express the new as a function of frequency, and pay attention to the low-
power dissipation as P1 = |I1 |2 R, where I1 = V /Z, with frequency asymptotics, to the high-frequency asymptotics,
Z = R + 1/iωC. and to the minimum of the modulus of the impedance.
87. (a) calculate the ohmic resistance of the lamp as U ′ /I, and 92. (a) Apply the idea 44, together with either the method
apply the idea 44 (express the total impedance containing of loop currents, or node potentials; keep in mind that
L as an unknown, and take the modulus from the Ohm’s each next node has potentials and currents phase-shifted
law, written for the entire circuit, to obtain equation for by φ, for instance, ϕj+1 = ϕj eiφ , where ϕj is the poten-
finding L). tial of the j-th node (we assume that the lower wire has
(b) Apply idea 44 (the formula φ = arg Z). a zero-potential); once you eliminate potentials (or cur-
(c) Apply the idea 45 (keep in mind that we are dealing rents) from your system of equations, you should obtain
with the rms amplitudes here). an equation relating ω to φ .
(d) Apply the idea 38 (for a brief instance, the inductor (b) Keep in mind that the phase speed of a wave is
will act as a constant current source). vp = ω/k, and the phase shift is related to the wave vector
(e) Express the instantaneous power as a function of time via equality φ = kl.
by substituting I = I0 cos(ωt) into the Joule’s law. (c) Study the low-frequency limit of your result for part
(f) Ignition is needed when the gas is insulating, i.e. when (b).
there are almost no ions in the gas. (d) Use the idea 51 to compare two systems: an infinite
(g) Calculate the new power dissipation; does it change? chain of springs and masses, and the given circuit; now it is
Compare the magnitudes of the active and reactive powers more convenient to work with the loop currents, and write
of this device (see fact 22). differential equations for charges passing through induct-
88. Notice that each of the halves of the coil have inductance ors. This is because for the mechanical system, the second
L L derivative of each coordinate enters only into one of the dif-
4 , and due to the fact 21, M = 4 . (This can be under-
stood if we cut off the capacitor and consider a current ferential equations (the Newton’s II law); meanwhile, when
I through the inductor: let the inductance of one half be using the node potential method and capacitors’ charges,
′ ′
L so that also M = L ; then each of the halves will have the inductors’ currents (terms giving rise to second deriv-
voltage L′ dI dI ′ dI atives) are expressed from the Kirchoff’s current law and
dt + M dt = 2L dt , i.e. the total voltage on
the full inductor would be 4L′ dI .) Let us use clock-wise would involve the charges of all the capacitors).
dt
current loops I1 and I2 in the both halves of the circuit; 93. Apply the idea 52. More specifically, notice that the four
then the voltage on a half of the inductor will depend vectors form a quadrilateral, opposing angles A and B of
on I1 + I2 (use idea 46!) and voltage on the capacitor which are right angles, hence this is an inscribed quadri-
— on I1 − I2 ; write down the two equations given by the lateral, and the other diagonal (other than AB) is the dia-
Kirchoff’s voltage law, first express both I1 −I2 and I1 +I2 meter. Pay attention to the fact that the quadrilateral is
in terms of V0 , and from these expressions find I2 . not convex, because the direction of the voltage vector on
90. (a) Apply the ideas 44 and 45. Note that for the frequen- L is obtained from that of R1 by a 90◦ -counter-clockwise
cies ν+ and ν− , you’ll have a fourth order equation with rotation (multiplication by iLω/R1 ), and the direction of
breaks easily down into two quadratic equations, one of the voltage vector on R2 is obtained from that of C by
which has roots ν+ and −ν− , and the other one — −ν+ the same rotation (multiplication by iCωR2 . The prob-
and ν− . Indeed, both negative and positive frequencies lem simplifies further owing to the fact that two sides and
must be the solutions, because they provide physically the one diagonal of the inscribed quadrilateral are all equal to
same signal, cos(ωt) = cos(−ωt), and each of these quad- each other.
ratic equations have roots which are clearly different by 94. Apply the idea 44, together with the idea 6: the circuit
modulus. Because of that, the difference ν+ − ν− is ac- breaks down into parallel and series connections. Express
tually the sum of the two roots of a quadratic equation, the voltage on the voltmeter and equate it to zero; pay at-
which can be found using the Vieta’s formula. tention to the fact that a complex number is zero if both
(b) Apply the idea 47 to conclude that there is only one real and imaginary parts are zeros, i.e. one equation for
non-zero natural frequency, and the idea 48 to find it complex numbers gives actually two equations for the real-
(equate the impedance between the left- and right-hand- valued quantities.
— page 42 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
95. You can solve it either by straightforward calculations by 99. Proceed according to the idea 48: “cut” the circuit near
applying idea 44, or geometrically (idea 52). one of the inductors and equate the impedance of the res-
ulting circuit to zero.
96. Knowing P2 , U2 , and U0 , one can easily find the resistance
of the power lines Rl . Further there are two options. First, 100. Apply the idea 44 to reduce the problem to a problem
one can proceed via a brute-force approach and using the of resistances, very similar to the problem 22. Similarly
idea 44: with the known voltage at the power station U0 , to that problem, you need to apply the ideas 23 and 22.
the values U1 and P1 yield two equations for finding two Notice that the segment AE is also essentially broken, and
unknowns, r and L. once AB and AE are broken, we can keep DA because if
Another and mathematically easier way is to use the idea the output leads are B and E, there is no current in the
52: apply the cosine theorem to express the known voltage segment DA (due to symmetry); breaking BA and AE
at the power station via the voltage on the power lines is the same as connecting respective negative resistances;
Ul = IRl , and the voltage on the transformer. This equa- with these negative resistances, we can perform a node
2
tion can be solved directly with respect to Pl = I Rl once splitting at A, so that a negative resistance is connected
you notice that the term with cosine can be written as only between B and E.
2P Rl and hence, is already known. 101. First, count the number of degrees of freedom (i.e. the
number of natural frequencies). Further, notice that there
97. Proceed similarly to the problem 93 (though, the quadrilat-
is one loop current, which involves only inductors through
eral is not inscribed): show that the diagonal AE divides
which a permanent current can circulate (this yields one
the quadrilateral of voltages into two equilateral triangles.
frequency). Next, apply the idea 50: two limit cases are
More specifically, notice that lower and upper branches of
obtained: one circuit contains only L1 and C1 (use the
the circuit have identical impedances and hence, there is
result of problem 99!), and the other — L2 and C2 .
no phase shift between the currents in them; because of
that, the voltage vector on AB is equal (hence also paral- 102. Use the idea 38: for (a), inductors can be “cut off”, and
lel) to the one on DE; the same applies to the voltages on the capacitors short-circuited; for (b) it is vice versa. For
BE and AD. Therefore, the triangle formed by voltage (c), the impedance of the connection of inductors and ca-
vectors on AB, BE, and AE is equilateral; the same ap- pacitors needs to be inifinite (there is a voltage resonance),
plies to the remaining triplet of voltages. and since there is no voltage on the resistor, it can be short-
circuited. For (d), the voltage on R needs to be maximal,
98. Apply the idea 52. Show that similarly to the problem 97,
hence the modulus of the overall impedance — minimal;
the voltage vectors form two equilateral triangles. More
this means that the impedance of the connection of induct-
specifically, use the symmetry to show that the voltage
ors and capacitors is zero (current resonance).
vectors on the two capacitors are equal to each other, and
the voltage vectors on the two inductors are equal to each 103. Keep in mind the idea 34. Find the maximal current
other. Details of exploiting the symmetry are as follows. Ix when the switch is opened — either using the idea
◦
Rotate the circuit by 180 , upon which the capacitors (and 35, or using the idea 36 with fact 19. When the switch
inductors) are swapped, and the applied voltage becomes is closed, we have two independent LC-circuits with the
negative; further, rotate the input voltage vector by 180◦ : same frequency, so the current in the switch is found as the
the new and old circuits become identical, hence, all the difference between the two currents in these LC-circuits.
corresponding voltages are equal. In particular, if origin- Amplitude (and hence, the maximal value) can be found
ally, the left capacitor had voltage U⃗C1 (from the input using phasor diagram, phase shift is to be found from the
lead towards the resistor), the originally right capacitor initial charges on capacitors and initial currents in induct-
⃗
has now also voltage UC1 , i.e. before the input voltage re- ors when the switch was closed.
versal, it had voltage −U⃗C1 (towards the resistor), which 104. (a) diac current can be neglected, hence R and C are con-
means that originally, it had voltage U⃗C1 , as measured nected in series directly to the AC voltage source so that
from the resistor to the input lead. Next, study the quad- we can apply the idea 44;
rilateral of the voltage vectors: its one diagonal gives the (b) we need to study, how will change the diac’s current
input voltage, and the other one — the voltage on the res- while the capacitor voltage grows, to that end we use the
istor. While the current vector of the resistor is parallel idea 24 and fact 9: if we start with a small voltage, there is
to the voltage vector, in the case of a capacitor it is ro- only one solution for the current, but with a small enough
tated by 90◦ clockwise, and in the case of an inductor — Rt , at a certain voltage two more solutions appear, and at
by 90◦ counter-clockwise; use this observation when writ- an even larger voltage (around Ub ), two smaller solutions
ing down the Kirchoff’s current law (for the node where disappear; at that moment, the diac’s current is forced to
R, L, and C meet each other) to conclude that the cur- jump to the only remaining solution (now it becomes also
rents in C, L, and R are all equal by modulus. Keep in clear what does mean “diac starts conducting”); the con-
mind that while the difference of the current vectors of L dition “triac starts immediately conducting” means that
and C gives the resistor’s current, the sum of those gives the new diac’s current must be larger than It ;
the input current. Finally, find the power dissipation and (c) according to (a), the capacitor voltage lags behind the
the resistor’s resistance by using the resistor’s voltage and overall voltage and hence, behind the lamp’s current; when
current values. the lamp’s current goes through zero (at t = 0 on the
— page 43 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
graph) and triac closes, the capacitor’s and diac’s voltage a symmetric circuit which has three linearly independent
is still negative; the triac will open again when the capa- loops, but there is one loop which consists of only capacit-
citor’s voltage becomes positive and equal to Ub at t = t1 ; ors, so there are two natural oscillation modes. We apply
(d) the energy disspated on the lamp is calculated as ideas 48 and 49 by splitting the circuit at the position of
∫ −1
R [Ul (t)]2 dt. one of the inductors and requiring Z = 0. Due to the sym-
105. Notice that the circuit is self-dual. Calculate the imped- metry of the obtained circuit we can remove the bridge
ance for sinusoidal voltage input using the idea 15. Since connection without affecting its impedance; the condition
the impedance is independent of the sinusoidal circular fre- Z = 0 yields us one frequency. The missing frequency is
quency ω, it is equivalent to an active resistor. NB! The obtained as the natural frequency of the simplified circuit:
analysis would have been much more difficult if the overall the circuit where the “cut” point is left disconnected.
impedance were to depend on ω as the impedances cannot
be applied directly in the case of non-sinusoidal signals. Answers
106. The number of degrees of freedom is two, so is the num- 1. R ≈ 14 Ω
ber of natural oscillation modes. According to the ideas 2. I = 0.5 A
48 and 49, consider voltage resonance (ZAB = ∞) for fic-
3. R = 0.5 Ω
titious terminals at the symmetry axis (A — middle point
of the inductor L2 dividing it into two inductances, L2 /2 4. R = 2.5 Ω
3
each, and B — the middle point of the lower wire); the im- 5. I= 22 A
pedance ZAB simplifies due to symmetry to such a degree 6. I = 196 µA
that the condition ZAB = ∞ gives us only one frequency.
7. VΣ = 78 V
According to the idea 49, the missing frequency is found by 21
short-circuiting nodes A and B, which makes C connected 8. I= 19 A

to a parallel connection of L1 and L2 /2. 9. I4 = 3 A, I3 = 2 A

4 (r+R1 +R2 )(R1 +r) E
1 R2 2
107. According to the idea 52, consider phasor diagram; ac- 10. P =
∑n ∑n
cording to the Thales’ theorem, the voltages (with respect 11. r = ( i=1 ri−1 )−1 , E = r i=1 Ei ri−1
to the one of the input terminals) of the nodes A and B lie
12. I2 = I4 = 0, I1 = I3 = E/R
on circle drawn around the input voltage vector as its diag-
15. R = 2Ω
onal; the voltage of the node D is the circle’s centre. Keep √ ( )
R1
in mind that due to the Thales’ theorem, the median of a 16. R = 1 + 1 + 4R2 /R1
2
right triangle equals to half of its hypotenuse; don’t forget ( √ )
17. r′ = R2 1 + 1 + 4R/r , E ′ = E
that central angle is twice as large as the corresponding
inscribed angle. 18. R = 56 Ω
9
108. Similarly to the problem 107, the voltages (with respect 19. RAO = 20 R.
to one of the input terminals) of the output nodes lie on 21. r = 19 R
30
circle drawn around the input voltage vector as its diag-
22. r = 19
11 R
onal; apply the inscribed angle theorem.
23. Straight lines connecting the following points: (0 s, 0 A);
109. Proceed similarly to the problem 108; the only difference (1.5 s, 1 A); (1.5 s, 0.6 A); (5 s, 2 A); (5 s, 53 A); (10 s, 10
3 A);
is that the relevant voltages do not form right angles, but (17 s, 1 A); (17 s, 1.2 A); (18.75 s, 0.5 A); (18.75 s, 56 A);
angles ≥ π/2; correspondingly, the voltages of the out- (20 s, 0 A)
put terminals lie inside the circle. Note that the segment
24. I ≈ 8 mA
which lies completely inside a circle cannot be longer than
the diameter, and can be equal in length to the diameter 25. approximately −1.4 times
if its endpoints lie on the circle. 27. 2
29 R ≈ 0.414R < r < 94 R ≈ 0.444R.
16
110. Reversing the idea 48, we find these frequencies as the 28. increases 9 times
natural frequencies of the circuit for which the termin- 29. 0.75 mW, 0 W, 0 W.
als are short-circuited. We result in a symmetric circuit E2R
30. r′ = R 3R+2r ′
5R+3r , E = E 5R+3r , Pmax =
R
4(5R+3r)(3R+2r)
which has two parallel connections of L and C, connec-
ted in series with 3C. We represent the capacitor 3C 31. I ≈ 3 mA
as a series connection of two capacitances 6C, and use 32. I1 = 1.5 A; I2 = 1.7 A
the idea 49 together with the voltage reonance condition 33. R = 40 kΩ
(ZAB = ∞, where A and B are the symmetrical nodes of 34. I ∝ V 0.6 or equivalently V ∝ I 5/3
the circuit) to deduce the first natural frequency. Finally,
35. I1 = 0, I2 = 3E/R, I3 = I4 = 1.5E/R
we short-circuit nodes A and B to find the missing natural
frequency. 36. 4 A
111. We start solving the same way as in the case of problem 37. all ammeters: 0.75 A.
110. Once we short-circuit the input terminals, we obtain 38. V2 ≈ 8.65 V
— page 44 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
39. V = (E1 R2 − E2 R1 )/(R1 + R2 + R1 R2
R3 ) 77. P = 2 mW; U0 = 21 V; C ≥ 200 µF; P1 = 200 mW
40. V3 = 1 V 78.
41. increases ≈ 1.14 times
42. 3 mA, 6 mA, 7 mA, and 14 mA.
43. R/r = 9
44. I1 = 0.7I0
45. RAB = R/3
[ ( √ )]
46. r = R J= τc Ui2
Ui τc R
4L U0 −Ui ; U0 = max 2Ui , 2 1+ 1+ L
47. r = 38 R
79. sawtooth profile consisting of curve segments V0 = Vi (1 −
48. r = 2R/n
e−t/RC ) (from V0 = 0 till V0 = Vf ) and vertical line seg-
67
49. r = 315 R ments (from V0 = Vf till V0 = 0); Vi ≫ Vf ; T = Vf RC/Vi ;
R 5R6 R; both; for instance: use tha same circuit, but connect
50. 3 <r< 11 ;
r = 17 R
√ another battery of emf. Vi − Vf and opposed polarity to
51. r = R 14 (3 + 17)
( ) ( √ ) the node between the battery and the resistance, and take
52. E ′ = E 1 + rr′ with r′ = R
2 1+ 1 + 4 Rr output signal between the ‘−’-lead of the new battery, and
the top lead of the SG (many other solutions are possible).
53. 40
87 R ≤r≤ 47
87 R
80.
54. R(1 − R2
r )
56. R + Rdiff > 0
58. 1
2 (V0 − V d )2 C
59. 2(E − Vd )
2 2
61. 27 CE
√ ( )
62. C1 E/ L(C1 + C2 ), E 1 + C1
C1 +C2

63. Sinusoid with minima at V = E C1C+C

1
and maxima at
√ 2

V = Vmax ; ω = 1/ (C1 + C2 )L √
−CV
⌊ d < ⌋q < CVd ; ∆q = −4CVd , t = N π LC, N =
64. 8.9 mA |q0 |−CVd
2CVd ; a mass-spring system where the mass is sub-
65. 0 mA, 0 mA, and 0 mA ject to a dry friction force.
66. V = IR for the first half-period, and V = −IR for the 81. I1 ≈ 0.33 A, I2 = 5 mA; I1′ ≈ 5.6 mA, I2′ = U τ /L =
second half-period [more precisely, for each half-period, 10 mA; ∆q = LI2′ /R2 = 3.3 mC.
the asymptotic values are reached exponentially, V =
82. (a) r + Rdiff < 0 and C L
< R|Rdiff | + r|R + Rdiff |;
±I1 R(1 − 2e−∆t/RC ), where ∆t is the time elapsed since
(b) additionally, Ld < r|Rdiff |Cd
the beginning of the half-period]; saw-tooth profile which
grows linearly from 0 to I1 T /C, and decreases linearly 83. (a) Roff = 10 Ω, Ron = 1 Ω, Rint = 2 Ω, I0 = 6 A; (b) 3 Ω:
down to 0 during the second half-period; (I2 + I1 )R/2, always one state; 1 Ω: 1, 2 or 3 states; (c) 3 A, 6 V, stable;
(I2 − I1 )R/2; (I2 + I1 )R/2, (I2 − I1 )T /8C (d) from 4 V to 10 V moves along the lower branch, jumps
67. IR = V0 cos(2πνt)/R, ID = V0 [cos(2πνt) + 1]/R from 1 A to 10 A, returns down to 4 V along the upper
we don’t use approximation Lω ≫ R then ID =
(if √ branch, and completes the cycle by jumping down to 0.4 A;
V0 R−2 + (2πνL)−2 [cos(2πνt + φ) + 1]). (e) 2.41 µs, 3.71 µs, 6.12 µs; (f) P ∼ 20 W; (g) τ < τcrit :
relaxation towards a higher stationary current, followed
68. 2BN S/R by a relaxation towards the original stationary current;
69. See at the website of IPhO τ > τcrit : relaxation towards a higher stationary current
70. (a) Both by modulus E/(R + r); along the lower branch, followed by a jump to the upper
E E
(b) R+r e−t/τ and R (1 − R+r
r
)e−t/τ . branch, current decrease along the upper branch, jump
to the lower branch, and relaxation towards the original
71. 2I, I, and I
stationary state; (h) 0.936 µs.
2E
72. 12 L2 Er2 ; LrR
2

0E
84. a) τL = LI0 /E; b) Vmax = RI0 ; c) P = VR ; d) Vav =
73. P = (U2 − U1 ) /4R, and P = C(U2 − U1 ) /T
2 2 √ √
EI0 R I0 L I0
( ) 2 ; e) U0 = 2C 2RE .
R2 R3
74. τ = R1 + C. E E E −t/τ
R2 + R3 85. (a) 25 R and 15 R ; (b) 25 R e , τ = 5L
R ; (c) constant, equal
75. 8.06 kΩ, > 50 µF to E/R
76. E; −2E; LE 2 /(2R2 ), CE 2 /2 + LE 2 /(2R2 ). 86. 2.8 µF
— page 45 —
 2. CIRCUITS INCLUDING CAPACITORS AND INDUCTANCES
87. 1.09 H; 64.1◦ 59.9 W; to create huge voltage to ionize the
gas, graph ∝ 1+cos(2πνt) [or slightly raised: a+cos(2πνt)
with a > 1]; recombination time is large enough to keep
vapors in the plasma state; the current is almost the same
as before, the phase −63.6◦ , this is to reduce the reactive
power if two lamps are in parallel
1
88. V0 ( ωL + ωC
4 .
√ √ L1 L2
90. R C/L; 1/ LC with L = L1 +L2 and C = C1 +C2 ; 0.1 A;
0.2 A
91. C = 1/ω limω→0 |Z(ω)|, L = limω→∞ |Z(ω)|/ω, R =
minω |Z(ω)|
( √ ) √
92. φ = 2 arcsin 12 ω LC ; ωl/φ; φ ≪ 1 when v0 = l/ LC;
infinite chain of masses connected by springs
93. 20 V
94. L = R1 R2 C, R = R1 R2 /RC
95. 2 arctan(ωRC)
96. ≈ 300 W
√
97. 10 3V
98. 10 W, 30 Ω
√
99. ω = √5±1
2 LC
11
100. 18 C
√
101. 0; √ 2 ; √ 5±1
L2 C2 2 L1 C1
√ √
102. 1, 0; 1, 0; 1, π, ν0 = 4π/ 2LC; 1 + RL2 C , arctan RL2 C ,
√
ν1 = 2π/ LC
√
103. q0 / 2LC.
√
104. (a) UC = U |k| = U/ 1 + (2πf RC)2 and φ = arg k =
− arctan(2πf RC); (b) Rt It < Ub − Ud ; (c) t0 =
[ ]−1
[arcsin(Ub /UC ) − φ]/(2πf ); (d) 1 − 2f t0 + sin(4πf
2π
t0 )
.
√
105. rectangular waveform of amplitude V0 C L.
√ √
106. ω1 = 1/ L1 C; ω2 = 2L 1 +L2
L1 L2 C .

107. V0 /2; 2φ.

108. φ
109. V0 ; R1 = R2 = 0, R3 = L/Cr.
√ √
110. 1/ LC; 1/ 7LC.
√ √
111. 2/3LC; 3/4LC.

— page 46 —
 1. BASICS. DOUBLE SLIT DIFFRACTION.
WAVE OPTICS where we have substituted ω = kv. Here, E0 and B0 can be
Jaan Kalda Version: 24th May 2016 complex numbers, so that E0 = |E0 |eiϕ ; then, E0 is referred to
as the complex amplitude, and ϕ = arg E0 is the wave’s phase.
1 Basics. Double slit diffraction. Since the electric field of an electromagnetic wave defines im-
Visible light is an electromagnetic wave; in vacuum, its speed is mediately also the magnetic field, in what follows we consider
constant and equal to c = 3 × 108 m/s; in a dielectric medium, only its electric field.
√ In real life, the wave fronts are not necessarily plane. In
the speed is reduced by a factor n = ε, where n = n(ω) is
the refraction coefficient, and ε is the relative dielectric per- particular, a point source emits spherical waves, and a line
meability; both depend on the angular frequency of the electric source emits cylindrical waves. However, if the distance to the
field (here we assume that the magnetic permeability µ ≈ 1 for wave source of an arbitrary shape is much larger than the wave
dielectric materials). length, within a small neighbourhood of an observation point
Maxwell equations admit several solutions; for instance, (of a radius of few wavelengths), the wavefront curvature is neg-
time-independent (stationary) solutions are possible. In partic- ligibly small2 . Because of that, the interference of non-planar
~ = 1 q2 ~er ,
ular, a point charge q creates an electrostatic field E waves can be studied as the interference of locally plane waves.
4πε0 r
where ~er is a unit vector pointing from the charge towards the Still, an important aspect should be kept in mind: while for
observation point. Note that stationary electric fields are cre- plane waves, the oscillation amplitude is constant throughout
ated by electric charges, and stationary magnetic fields — by the space, for non-planar waves, the amplitude is a function of
electric currents. However, Maxwell equations include also coordinates.
terms with time derivatives (e.g. the time derivative of the In particular, for spherical waves, the amplitude is inversely
magnetic flux in the Faraday’s law); owing to these terms, wave- proportional to the distance from the point source, as it follows
like solution are also possible. In particular, one can have a from the energy flux continuity. Indeed, the intensity of the
sinusoidal plane wave, for which the wave fronts1 form a set wave (the energy flux density) is proportional to the squared
of parallel planes: wave amplitude, I ∝ E02 ; the energy flux (i.e. the total radi-
~ = ~ex E0 cos(kz − ωt), B
E ~ = ~ez B0 cos(kz − ωt), ation power transmitted through a fictitious surface) equals to
where z is the propagation direction axis, the product of the intensity, the surface area, and the cosine
k = 2π/λ, (1) of the angle between the wave vector and the surface normal,
is the wave vector, related to the circular frequency and wave Φ = IA cos φ. With the origin being at the point source, let
speed via us consider the energy balance for the volume between two
ω c c
=v= = √ , (2) concentric spherical surfaces (of radii r1 and r2 ), within a solid
k n ε angle Ω: the incoming energy flux equals to Ωr12 I1 , and the
and the field amplitude ratio E0 = B0 v. Let us notice that at
any point in space, the electric field, the magnetic field, and outgoing one — to Ωr22 I2 . In a stationary state and assuming
the propagation direction are all perpendicular to each other that there is no energy loss due to dissipation, these√ two fluxes
(propagation direction corresponds to the motion of a screw must be equal, i.e. I1 r12 = I2 r22 and hence, E ∝ I ∝ 1/r.
√
when rotated from E to B). In the figure (a) below, the vectors Similarly, for cylindrical wave, E ∝ 1/ r.
of the electric- and magnetic field are depicted for a certain As long as the propagation speed is constant, e.g. in a
time moment t = τ , for a series of points lying on the z-axis; vacuum with v = c, any electromagnetic pulse will propagate
the endpoints of these vectors lay on sinusoids, which are drawn with a constant shape and speed3 :
~ = ~ex E(z − vt), B
E ~ = ~ez E(z − vt)/v
for t = τ and t = 0 (the dotted curves). In figure (b), electric
field lines are depicted for the same wave. However, if n in Eq. (2) depends on the angular frequency ω,
x the pulse shape will change in time; furthermore, the speed of
(a)
k E the pulse will be
dω d ck
E z B vg = = , (5)
dk dk n(k(ω))
B t= t= k which is referred to as the group speed (then, ω/k yields the
v 0
y speed of a fixed phase, e.g. of a wave-crest, and is called the
x phase speed).
(b) electric
field lines To understand why, let us consider the superposition of two
z beating waves of wave vectors k − ∆k and k + ∆k, respectively:
ei[(k−∆k)z−(ω−∆ω)t]
h + ei[(k+∆k)z−(ω+∆ω)t]
i =
i(kz−ω)t −i(∆kz−∆ω)t i(∆kz−∆ω)t
In the complex number form, such a wave can be expressed e e +e =
as 2ei(kz−ω)t cos(∆kz − ∆ωt).
E~ = ~ex E0 ei(~k~r−ωt) , B
~ = ~ez B0 ei(~k~r−ωt) , (3) Here, the first factor ei(kz−ω)t corresponds to the wave itself,
where we have used the dot product of the radius vector and the second factor cos(∆kz − ∆ωt) describes its envelope;
~r ≡ (x, y, z) with the wave vector ~k (which is parallel to the the speed of the envelope vg = ∆ω/∆k.
propagation direction of the wave). Alternatively, we can write In what follows, we consider only monochromatic waves,
E~ = ~ex E0 eik(z−vt) , B ~ = ~ez B0 eik(z−vt) , (4) i.e. sinusoidal waves of a fixed frequency ω. This is because
1A wave front is defined as the set of points of constant oscillation phase (for instance, wave crests).
2 Unlessa concave source shape leads to a wave focusing near the observation point.
3 The proof is provided in Appendix 1 on pg. 12.

— page 1 —
 1. BASICS. DOUBLE SLIT DIFFRACTION.
we shall study the interference of light waves, and typically, an observed at a far-away screen where two parallel rays (drawn
interference pattern can be observed only for light beams origin- in figure) meet.
ating from a single source (this will be discussed in more details
below). Now, if all the waves originate from the same source, e
4 s ou
rc
l1
they must have also the same frequency . Note that if the wave s
g en
enters a refractive transparent medium, the wavelength may uy
A H
change, but the frequency remains constant.
l2
fact 1: The frequency of a wave remains constant along its a
entire path if the wave speed v depends only on coordinates

wavefronts
C
and not on time (for light: if n = n(x, y, z) does not depend on B
time). Huyg
ens s
ource
Indeed, the time required for a wave crest to travel from the
source to a given destination point is defined by the integral To begin with, it is quite easy to figure out, where are the
R dl intensity maxima and minima. Indeed, as it can be seen from
v(x,y,z) , taken over the wave trajectory, which remains con-
stant in time; hence, the time delay between neighbouring wave the figure above, the optical path difference between the two
crests at the destination remains equal to what it was at the rays is ∆l = a sin ϕ. The two rays add up constructively (giving
source. rise to an intensity maximum) if the two waves arrive to the
screen at the same phase, i.e. an integer number of wavelengths
We also assume that the coherence length of the waves is
fits into the interval: ∆l = nλ. Similarly, there is a minimum if
larger than the system size. Coherence length is a distance upon
the waves arrive in an opposite phase:
which the wave “forgets” its phase. One can imagine this as
1
having a sinusoid with slightly varying wavelength; upon certain sin ϕmax = nλ/a, sin ϕmin = (n + )λ/a. (6)
2
distance, the variations accumulate into such an error that the Now, let us proceed with the calculation of the intensity
phase difference between this wave and an ideal sinusoid will distribution.
be of the order of π (which corresponds to an opposite phase).
Quantitative calculation of the interference pat-
method 1:
For the light sources other than lasers, the coherence length is
terns is most conveniently done by adding up the complex
really short; for lasers, it can reach the values around tens of
amplitudes of interfering waves (similarly to alternating cur-
meters.
rents and voltages). Mathematically, if the wave amplitudes of
fact 2: (The Huygens principle.) Consider an arbitrary wave M interfering are a , m = 1, 2, . . . M , and the corresponding
propagation, for which a certain wave front is known. The wave optical paths are denoted by l the resulting wave's complex
m

propagates beyond that wave front in the same way it would amplitude is
m

propagate if a densely populated array of small wave sources X

iklm

were placed along the wave front.

a= a e . m

Note that complex numbers can be considered as two-

m

This fact is the main tool for calculating diffraction patterns dimensional vectors (the x- and y -coordinates of which are their
in majority of cases. Let us analyse this using the example of real- and imaginary parts, respectively); because of that, altern-
double slit diffraction (for some problems, including the first atively, vector diagrams can be used (each wave is represented
one, solutions are provided after the problem text). by a vector, the length of which reects the wave amplitude,
and the direction  the wave's phase). Here, the amplitudes
pr 1. Consider a non-transparent wall with two parallel nar- a are proportional to the sizes of the Huygens sources. In
row slits (much narrower than the wavelength) at distance a the case of 3-dimensional geometry, they are and inversely pro-
m

from each other. Parallel light beam falls perpendicularly on portional to the distance l , and in the case of 2-dimensional
that wall. Find the diffraction pattern behind the wall: the geometry  inversely proportional to √l .
m
m
propagation angles for which there are light intensity maxima
To understand the origin of the source-size-proportionality, one
ϕmax and minima ϕmin , as well as the intensity distribution as
can consider two identical near-by sources: due to the negligible
a function of the angle ϕ.
distance, the corresponding waves have the same phase and
The wall blocks almost all the wave front of the original therefore add up into a wave of double amplitude.
wave, leaving only two points in a cross-section perpendicular It should be noted that the formulation of this method
to the slits (see figure below). To be precise, these are actually ignores the dependence of the wave amplitude of the contribu-
segments, but their size is much smaller than the wavelength; so, tions of the Huygens sources on the propagation direction. This
from the point of view of wave propagation, the segments can approximation is valid as long as the angle between the surface
be considered as points. According to the Huygens principle, normal of a Huygens source and the vector pointing to the obser-
two point sources of electromagnetic waves of wavelength λ vation point is small (its cosine is approximately equal to one).
will be positioned into these two points (A and B). The point If this angle is not small so that strictly speaking, dropping the
sources radiate waves in all the directions, and we need to study angle-dependent factor would not be correct, doing so retains
the interference of this radiation. Let us study, what will be still the qualitative properties of the diffraction pattern if all
4 The source can emit different frequencies, but such a radiation can be decomposed into a superposition of sinusoidal waves, as taught by the
Fourier analysis.

— page 2 —
 1. BASICS. DOUBLE SLIT DIFFRACTION.
the contributing beams are characterized by the same angle of the modulus can be calculated via Pythogorean theorem as
(like in the case of Pr. 1), because then, the angle-dependent the sum of the squares of the real- and imaginary parts:
factor has the same value for all the beams and can be brought {1 + cos[k(l2 − l1 )]}2 + sin2 [k(l2 − l1 )] = 2{1 + cos[k(l2 − l1 )]}.
before the braces.
Finally, if we denote l2 − l1 = a sin ϕ, we end up with
In the case of light waves travelling in the z-direction, the
amplitudes am and a are to be interpreted as x- or y-components I = 2I0 [1 + cos(ka sin ϕ)]. (7)
of the E- or B-field. It is not important, which quantity is con- Now we can also recover the earlier result (6) regarding the
sidered, because as long as there is no double refraction, for positions of the intensity minima (cosine gives −1, I = 0) and
any contributing wave, there is no phase shift between Ex , Ey , maxima (cosine gives +1, intensity becomes quadruple): for
Bx , and By . Meanwhile, for polarized light in double-refracting minima, ka sin ϕ = (2n + 1)π, and for maxima, ka sin ϕ = 2nπ.
materials, x- and y-components need to be studied separately: Note that if there were non-coherent light sources at A
then, the phase shift will depend on the oscillation axis; this will and B, there would have been an additional time-dependent
be discussed later. In order to avoid emphasizing which field is phase shift ψ(t) which should have been added to the phase
considered, in what follows the amplitudes will be denoted by a. shift ka sin ϕ due to the optical path difference. In that
Let us recall that the modulus of the complex amplitude gives case, I = 2I0 {1 + cos[ka sin ϕ + ψ(t)]}; owing to the fluc-
the real amplitude of the sinusoidal signal, and the argument of tuating phase ψ(t), the diffraction maxima (and minima)
it gives the phase shift. Thus, the real field a(x, t), at the given would move so fast that human eyes would register only the
point as a function of time, is given by mean value of the intensity, h2I0 {1 + cos[ka sin ϕ + ψ(t)]}i =
Re a(l)eikl · e−iωt = |a(l)| cos[ωt + arg a(l)]. 2I0 {1+hcos[ka sin ϕ + ψ(t)]i} = 2I0 ; here angular braces denote
averaging, and averaged cosine gives zero.
Typically, momentary field values of electromagnetic waves are
Regarding the fluctuations of the interference patterns from
never needed: the oscillations are so fast that what is measured
non-coherent light sources, we can make a simple estimation:
is the root-mean-square average. Therefore, the only things
let us have two point sources of fairly monochromatic light, for
of interest are the modulus of the field, and its phase shift.
instance from two identical good lasers with a coherence length
Because of that, it is enough to work with the complex
of l = 10 m and wavelength λ1 = 658 nm. Then the coherence
amplitudes; there is no need to write down the full spatio-
i(kl−ωt) time τ = l/c ≈ 3 × 10−8 s gives us the fluctuation time of the
temporal dependence of the complex signal a(l)e , and
random phase ψ(t), and also the characteristic time-scale during
there is no need to add Re to separate its real part which
which the diffraction pattern fluctuates. This is well beyond
corresponds to a real physical quantity.
anything what an human eye can resolve: we’ll see an averaged
So, at our infinitely-remote-screen, we have two waves
picture without any interference stripes.
a1 (l1 )eikl1 and a2 (l2 )eikl2 adding up. The relative difference
So, as long as we are not studying phenomena at ultra-short
between l1 and l2 is small; hence, the dependence of the wave
time-scale (ato- and picosecond-scale-physics), in order to be
amplitude on distance affects the both waves in the same
able to see an interference pattern, the light needs to come
way, i.e. |a1 (l1 )| = |a2 (l2 )|. The two Huygens sources are
from the same light source, even in the case of lasers. Addi-
at the same wavefront, which means that at the respective
tionally, the optical path differences of the interfering rays must
sources, there is no phase shift between the emitted waves,
not exceed the coherence length of the given light source. In
hence arg a1 (l1 ) = arg a2 (l2 ); combining the last two equalities
the case of the double slit interference, it is sufficient if the light
yields a1 (l1 ) = a1 (l2 ). Since we are interested in the relative
falls onto the slits from the same point source, not necessarily a
intensity of the light at the screen, and not in how √ it decreases laser. On the other hand, if it is not a point source, but instead
with l, we can drop the dependence a(l) ∝ 1/ l and denote
a light bulb with a considerable size, the coherence length may
a1 (l1 ) = a1 (l2 ) ≡ a (the sign “∝” means “is proportional to”).
become too short to be able to observe a two-slit-interference;
Finally, we can combine the term eikl1 into the complex amp-
a practical guideline here is that the angular size of the light
litude (this only rotates the complex ampitude as |eikl1 | = 1)
ikl1 source needs to be smaller than the angular distance between
by denoting ae = ã, in which case
the diffraction maxima (otherwise two non-coherent halves of
aeikl2 = ãe−ikl1 · eikl2 = ãeik(l2 −l1 )
the source would give rise to two shifted diffraction patterns
We can also say that the amplitudes are normalized to the light which become smoothed due to overlapping).
wave amplitude from the first slit, and put ã = 15 .
So, the superposition amplitude is given by pr 2. Consider the same situation as in the case of pr 1,
E =1+e ik(l2 −l1 )
. however with three slits at equal distance a from each other.
Find the positions of diffraction minima and maxima.
The intensity is proportional to the square of the modulus,
which is given by the product of E with its complex conjugate
Ē = 1 + e−ik(l2 −l1 ) : pr 3. Show that for three parallel slits of equal size and
with neighbour-to-neighbour distances being equal to a and b,
I/I0 = (1 + eik(l2 −l1 ) )(1 + e−ik(l2 −l1 ) ) =
respectively, the intensity at the diffraction minima is non-zero
= 2 + eik(l2 −l1 ) + e−ik(l2 −l1 ) = 2{1 + cos[k(l2 − l1 )]}; unless ab = m n
, where n and m are integers and n − m is a
here I0 is such an intensity which would be recorded at the multiple of three.
screan when one of the slits is closed. Alternatively, the square
5 When solving wave optics problems, this paragraph can be summarized as on sentence: “at large distances, the contributing waves have equal
amplitudes which will be normalized to the amplitude of a single wave

— page 3 —
 2. SINGLE SLIT DIFFRACTION; DIFFRACTION GRATING
2 Single slit diffraction; diffraction grating localized into the main maximum (between a| sin ϕ| < λ).
pr 4. Consider a non-transparent wall with a slit of width a. I
Parallel light beam falls perpendicularly onto that wall. Find
the diffraction pattern behind the wall: the propagation angles
for which there are light intensity maxima ϕmax and minima
ϕmin , as well as the intensity distribution as a function of the
angle ϕ.

A
From the practical point of view, an important case is the
wavefronts

A' diffraction behind a circular opening of diameter d — this will

 C
B
C'
B'
 lenging task, and involves Bessel functions; it appears that for
To begin with, let us find the positions of the diffraction
minima — where the intensity is zero. To this end, let us divide
the slit (fictitiously) into two halves; the Huygens sources from
the upper half are marked with blue, and those from the lower
half — with red, see figure. If the segment lenth BC equals to
a half-integer-multiple of the wavelength then the contributions
of the red and blue sources (from A and B) will cancel out;
the same applies to any another matching pair, e.g. for points
A0 and B 0 . Indeed, the corresponding optical path difference
|B 0 C 0 | = |BC|. So, the contributions from all the red and blue
sources will cancel pair-wise out: the intensity is zero if
a 1
sin ϕmin = λ(n + ) ⇒ a sin ϕmin = λ(2n + 1).
2 2
Next, we divide the slit into four segments; then into eight
segments, etc. in general, into 2m segments; as a result, we
find that a zero intensity is observed for
a sin ϕmin = λ2m−1 (2n + 1).
One can see that here the factor of λ can take all the integer
values except for zero6 ; so we can write
a sin ϕmin = nλ n 6= 0.
This result means that the main intensity maximum (at ϕ = 0)
has a double width.
In order to find intensity distribution behind the slit, we
need to integrate over the Huygens sources. Let us take the
happen in the case of optical devices such as lenses, telescopes
and microscopes. Finding the diffraction behind a circular
opening of diameter d is mathematically significantly more chal-
λ  d, the first intensity minimum is observed for the angle
ϕ ≈ 1.22λ/d.
Let us assume that the front lens of a telescope (the object-
ive) has a diameter d and creates an image of a twin star, the
angular distance between the stars being α. It may be confusing
that we have two effects: the diffraction on a circular opening,
and ray convergence due to the lens. These two effects can be
fortunately decoupled: suppose we remove the lens; then, the
circular opening will create a diffraction pattern at an infinitely
remote “screen”. Now, we “put back” the lens, which creates
the image of that infinitely remote pattern at its focal plane,
according to the rules of geometrical optics.
It is said that the images of the two stars are resolved if the
centre of the image of one star lies beyond the first intensity
minimum of the diffraction pattern of the other star. According
to angular position of the first diffraction minimum behind a
circular hole, this means that the telescope resolves angular
distances larger than 1.22λ/d.
Next, let us consider a diffraction grating, which has N slits,
neighbouring slits being at a distance a from each other.
pr 5. Calculate the diffraction pattern behind a diffraction
grating assuming that the slit width is half of the grid pitch a.
Na
D' C' B' A' B A D C
n
(m+ ition for

a
n
cond um

x-axis as the line AB, with the origin at B (i.e. at the centre
minim

cond
main
½

of the slit). Then, each Huygens source contribution to the net



tion fi

wave amplitude E is proportional to its length dx; the optical

maxi

path difference of this wave with respect to the wave arriving

or n-
m um

from B is given by ∆ = x sin ϕ, which corresponds to the phase

th

shift x sin ϕ. Hence, the sum of all the waves can be written as
Z a2 ak ak Main maxima behind a grating can be found from the condi-
ei 2 sin ϕ − e−i 2 sin ϕ 2 sin( ak
2 sin ϕ)
E∝ eikx sin ϕ dx = = .
tion that the contributions from the neighbouring slits arrive at
−a ik sin ϕ k sin ϕ
2 the same phase: the length of the thick blue lines in the figure
Intensity is proportional to the squared amplitude, so that
" #2 needs to be an integer multiple of the wavelength, i.e.
sin( ak
2 sin ϕ) a sin ϕmax = nλ.
I∝
sin ϕ Apart from the main maxima, there are numerous side max-
(we have dropped here constant factors 4 and k −2 ). This ima; similarly to the case of a single-slit-diffraction, instead
dependence is shown in the figure below. Pay attention of finding the positions of these maxima, it is easier to find
to the fact that dominating majority of the light energy is the positions of the minima; the maxima are just between the
6m = 1 gives all the odd numbers; m = 2 gives all such even numbers which give reminder 2 if divided by 4, etc

— page 4 —
 2. SINGLE SLIT DIFFRACTION; DIFFRACTION GRATING
minima. The approach is also the same: we divide the whole Pay attention that at the right-hand-side of this expression,
diffraction grating into two halves, and consider the interfer- nN equals to the number of wavelengths which fit into the op-
ence of the contributions of the both halves: the pair-wise tical path difference between the shortest path and the longest
cancellation of the light rays will take place if the length of path through our spectral device (for the direction at which
the red thick line is an half-integer-multiple of the wavelength, the n-th main diffraction is observed). This is a very generic
i.e. 12 aN sin ϕ = (m + 12 )λ. Further we divide the grating into result, applicable to any spectral device, e.g. to a Fabri-Perot’
four, eight etc pieces, to conclude that the minima (with zero or Mach-Zehnder interferometer or to an echelle grating.
intensity) are observed for aN sin ϕmin = mλ, where the integer The last result implies that larger physical size of a grating
m 6= nN (m = nN corresponds to the n-th main maximum). provides a better resolving power. However, in practice, this
In order to calculate intensity distribution behind such a is not always the case. The reason is that the above derived
grating, we can sum over the contributions of single slits. For formula assumes that the grating is ideal, with a strictly con-
the observation angle ϕ, we can use the expression for the elec- stant pitch. In practice, however, the pitch can fluctuate, and
tric field created by a single slit which was calculated earlier in that case the limiting factor will be the coherence length
(we need to substitute a with a/2): of the stripes — the length l which can be covered either by
sin( ak
4 sin ϕ) N or N + 1 stripes, the uncertainty being due to the pitch
E0 = . fluctuations.
sin ϕ
Neighbouring slits have additional optical path difference a sin ϕ, pr 7. Estimate the resolving power of a Fabri-Perot interfer-
which corresponds to the phase shift ka sin ϕ, and can be reflec- ometer as a spectral filter, and find its spectral transmittance
ted by an additional term eika sin ϕ for the complex amplitude as a function of wavelength. This interferometer consists of two
of the electric field. Thus, semi-transparent and semi-reflecting parallel surfaces with a
N/2
X sin( ak
4 sin ϕ) ikan sin ϕ very high reflectivity R (this gives the fraction of light energy
E= e =
sin ϕ which is reflected — as compared with the incident beam’s
n=−N/2
energy), which are separated by a distance a.
N/2
sin( ak
4 sin ϕ)
X
= eikan sin ϕ .
sin ϕ
n=−N/2
This is a geometric progression, and the sum can be easily
taken:
sin( ak ika( N N
2 +1) sin ϕ − e−ika 2 sin ϕ
4 sin ϕ) e
E= =
sin ϕ e−ika sin ϕ − 1
sin( ak sin ka N2+1 sin ϕ


4 sin ϕ)
= ·
=
sin ϕ sin ka2 sin ϕ
sin ka N2+1 sin ϕ


=
.
2 sin ϕ cos ka 4 sin ϕ
Diffraction gratings are often used as spectral devices — to
measure the spectrum of a light. In that case, it is important The resolving power can be estimated easily using the above-
to have a good resolving power. mentioned generic rule. The shortest optical path is the one
pr 6. Find the resolving power of a diffraction described by which goes through the interferometer directly without being
reflected; the longest one performs many reflections. Strictly
the previous problem, i.e. determine minimal value of ∆λ, such
speaking, such a multiply-reflected beam has no upper length
that two spectral lines λ and λ + ∆λ can be resolved with such
limit, but too many reflections lead to an almost vanishing
a grating.
intensity of the beam. A beam will participate efficiently in
In the case of a telescope, two points were assumed to be diffraction, if its intensity is not much smaller than that of
resolved, if the centre of one image lays beyond the first intens- the other beams. The number of reflections can be estimated
ity minimum of the second image. In the case of a grating, we as N ≈ 1  1. Indeed, upon N reflections, the remaining
1−R
proceed in the same way: two spectral lines are resolved, if the intensity of the light is reduced by a factor of RN ; let us take
centre of one line is beyond the nearest diffraction minimum of the bordeline value for this factor to be e−1 . Then RN ≈ e−1
the other line. At the borderline case, these two things coincide; from where N ≈ − 1 ≈ 1 . So, we obtain
ln R 1−R
let the centre of one line’s n-th main maximum be at ϕ; then λ 1
≈ .
a sin ϕ = nλ; ∆λ 1−R
The spectral transmittance can be found in two ways. The
if this coincides with the nearest minimum of the second spectral
first way is to sum the contributions of several reflections using
line then
the formula for a geometric progression. The second way is to
aN sin ϕ = (nN − 1) (λ + ∆λ).
combine all the upwards propagating waves into a single wave,
We can eliminate ϕ from these two equations to obtain
and all the downwards propagating waves into another wave.
N nλ = (nN − 1) (λ + ∆λ); Then, the amplitudes of these effective waves can be tailored to
upon opening the braces and neglecting the term ∆λ for N  1, each other via the reflection condition (see below), and eventu-
we obtain ally expressed in terms of the incident wave amplitude. This
λ
= N n. approach is valid because of the following idea.
∆λ
— page 5 —
 3.
The sum of several sinusoidal waves of equal
BRAGG REFLECTION
idea 1: Let us pay attention to the fact that at the transmission max-
wavelength propagating in the same direction is also a sinus-  2

oidal wave propagating in the same direction. imum, t = 1, and at the transmission minimum, t = 1−R 1+R
The reflected wave (the purple up-moving one) can be found
.

Indeed, let us have N sinusoidal waves, and let the n-th from the energy conservation law: the effective reflectivity
wave be represented in a complex form as r = 1 − t. Alternatively, it can be found as the superposition of
the reflected part of Ei and the transmitted part of Eu . Here,
an (z) = An ei(kz−ωt) ,
however, we need to take into account additional phase shifts
where An its complex amplitude. Then, the sum of the waves during reflections.
If electromagnetic wave is reected from the interface
is given by
fact 3:
X
An ei(kz−ωt) .
of two dielectric media, it will partially reected and partially
a(x) =

refracted (as long as it is not a total internal reection); in

n
The exponential term is the same for all the waves and hence,
can be brought before the braces (i.e. the summation sign): optically sparser medium, the reected wave obtains an addi-
a(x) = ei(kz−ωt)
X
An . tional phase shift of π at the interface. There is no phase shift
for other waves (refracted waves and reected beam in the op-
tically denser medium).
n
P
Now, n An is a complex number, let us denote it by A. Then,
a(z) = Aei(kz−ωt) , i.e. we have a wave with the same wavelength
Partial reflection can take place on various interfaces, for
and the same direction of propagation as the component-waves.
instance on very thin metal films. It can be proved using the
Thus, we can combine all the purple up-moving-waves (see
energy conservation law that regardless of what kind of interface
the figure above) into a single wave of amplitude Er , and all
there is, the sum of phase shifts between the transmitted and
the green down-moving-waves into a single wave Et (indices t
reflected waves (for the both directions of the incident wave)
and r standing for “transmitted” and “reflected”). Similarly we
equals to π. In the case of thin metal film, there is a mirror
can combine all the blue up-moving waves into Eu and all the
symmetry, so the phase shifts cannot depend on the direction
blue down-moving-waves into Ed . Finally, let the amplitude of
of incidence and hence, the phase shift between the transmitted
the incident wave (the red one) be denoted by Ei . Then we
and reflected waves is π/2.
can say that Et is the transmitted part of Ed , and Eu is the
reflected part of Ed : pr 8. Prove that for an arbitrary semi-reflecting dissipation-
√ √
Et = 1 − REd , Eu = REd . less interface, the sum of phase shifts between the transmitted
Note that we need a square root because reflectivity R and and reflected waves for the both directions of the incident wave
transmittance 1 − R are related to the intensity (we deal with equals to π.
the amplitudes, which are proportional to the square root of the Returning to the case of Fabri-Perot interferometer, one can
intensity). Further, Ed is made up from the transmitted part of assume that the mirrors are dielectric and between the mirrors,
Ei and reflected part of Eu . Now we need to pay attention to there is optically denser medium. Then, there is a phase shift
the phase shift between these three complex amplitudes. The π between the red and purple waves (see figure above), so that
√ √
phase shift of a wave will be changed by kl, if we shift the origin Er = 1 − REu e2ika − REi .
by distance l; since the “blue down wave” and the “orange
incident wave” propagate in different (non-overlaping) regions
3 Bragg reflection
, we can use different origins for them and make the phase
shift of both waves equal to zero. Thus, we can assume that The interaction of X-rays with ordinary matter is typically
the transmitted component of Ei contributes to the complex very weak. This is because the frequency of X-rays is much
amplitude of the “blue down wave” without phase shift7 , with higher than the natural frequencies of electrons around the
√
1 − REi . Then, upon travelling down and up, the optical atoms and molecules. What happens is completely analogous
path length accumulated by the blue wave equals to 2a, and to the mechanical resonance which is described by its equation
hence, the reflected component of Eu comes with a phase factor of motion
e2ika : mẍ = −kx + f0 cos(ωt),
√ √ 2ika
Ed = 1 − REi + REu e . where m is the mass of a particle attatched to a spring of
iωt
Using these three equations, we can express Et in terms of Ei . stiffness k, x is the displacement of the particle, and f0 e —
√ √
Indeed, Ed = Et / 1 − R and Eu = Et √1−R R
; thus, the external forcing. If the forcing frequency is close to the
√ natural frequency of an oscillator, the oscillation amplitude will
E R
√ t = 1 − REi + √ Et e2ika , become very large; for low frequencies, the oscillator will take a
1−R 1−R
quasi-equilibrium position: the displacement of the system is
from where
1−R given by the momentary value of the forcing, x = f0 cos(ωt)/k.
Et 1 − Re2ika = (1 − R)Ei ⇒ Et =


Ei . For very high frequencies, the strain force of the spring will
1 − Re2ika
By definition, effective transmittance t = |Et | /|Ei | = play a negliglible role and the oscillator will behave as almost a
2 2

Et Ēt /Ei Ēi ; therefore, free particle, mẍ = f0 cos(ωt), which can be integrated twice to
yield x = −f0 cos(ωt)/ω 2 . This result means that the oscillation
1−R 1−R (1 − R)2
t= · = . amplitude will decrease inversely proportionally to the squared
1 − Re2ika 1 − Re−2ika 1 + R2 − 2R cos(2ka)
7 The semitransparent mirror can cause a phase shift, but in that case, we use an appropriately shifted coordinate systems for the interior region.

— page 6 —
 3. BRAGG REFLECTION
frequency, and the displacement vector of the oscillator will be form as weakly reflecting (mostly transparent) surfaces. Crystal
in the opposite phase with the forcing. planes are fictitious planes at which a large number of ions
In the case of low-frequency electromagnetic waves (such is situated (see figure with a cubic lattice: cross-sections of
as radio waves), the forcing frequency is much smaller than crystal planes are depicted by black solid, blue dashed and
the natural frequencies of the electrons, the molecules will be orange dotted lines); the planes with higher ion surface density
deformed and polarized exactly in the same way as when being reflect X-rays more efficiently (in the figure, surfaces marked
put into an electrostatic fields. Hence, the frequency-dependent with black solid lines have the highest density).
dielectric permeability takes its stationary value. Inpparticular,
the refractive index of water for such waves is n = ε(0) ≈ 9.
Near natural frequencies (close to a resonance), the wave
energy is pumped into oscillations of electrons, which leads
to dissipation: matter becomes opaque. For frequencies much
higher than the natural frequencies of the orbital electrons
(e.g. for X-rays), the electrons respond to the forcing in the
same way as free electrons (not bound to molecules). Indeed,
the natural frequencies are of the same order of magnitude as
the orbital rotation frequency, which means that during one
forcing period, the orbital displacement of electrons remains
Although each surface reflects only a small amount of light,
much smaller than the orbital radius. We have seen above that
if the reflections from many surfaces add up in the same phase,
for a free particle in a sinusoidal force field, the displacement
the total reflected light can be significant; in fact, almost all
is in the opposite phase with the forcing. In the case of low
the light can be reflected. So, the condition of Bragg reflection
frequencies, the displacement is in the same phase with the
is that the contributions from neighbouring crystal surfaces are
forcing, which leads to a decrease of the overall electric field
in the same phase, i.e.
due to the polarization of the molecules; this effect is described
by the relative dielectric permeability ε > 1. In the case of ∆l = |AB2 | + |B2 C2 | − |AB1 | + |B1 C1 | = nλ.
high frequencies,
p the effect is opposite, hence ε(ω) < 1, and Calculating ∆l in such a way is not easy enough — it can be
also n(ω) = (ε) < 1. Such values imply that the phase speed done in an easier way. Indeed, instead of studying the reflec-
of light v = c/n > c, which may seem to be in a contradiction tions of the same ray AB1 , let us consider different rays of the
with the theory of relativity; however, relativistic constraint incident beam of X-rays, AB1 and KH.
applies to the energy and information transfer rate only, which
idea 2: Dierent rays of the incident beam have the same
is given by the group speed of electromagnetic waves.
phase at those points which lay on the same wave front. Wave
Unlike in the case of gamma rays, the frequency of X-rays is
lower than the natural frequencies of nuclear oscillation modes,
front is a surface perpendicular to the rays.
hence they interact only with the orbital electrons (to be correct,
they do interact with the whole nuclei as with charged particles,
but the mass of a nucleus is much larger than the electron mass,
so the interaction is much more efficient in the case of electrons).
As we have argued, the electrons behave as free particles, hence
the interaction strength is defined only by the volume density
of electrons: higher density of electrons implies larger value of
n − 1. In particular, this is why iridium is used for the mirrors
of X-ray telescopes (even though n − 1 remains still small, for
small grazing angles, total internal reflection can be achieved).
A weak refraction is not the only way in which the X-rays
can interact with matter. There is also a possibility that they
are absorbed or scattered from the electrons; in that case, an
X-ray beam behaves as a beam of particles — photons, which Therefore, the incident wave has the same phase at the
collide with the electrons (in the case of absorption, an electron points A and D; the same applies to the pair of points B1 and
at a lower orbital receives the photon’s energy, and “jumps” to F . Similarly, for the reflected wave, C1 and E have the same
a free orbital). These are probabilistic effects; the scattering (or phase, and B1 and G have the same phase. So, Bragg reflection
absorption) probabilities are to be calculated using equations of condition can be written as condition that
quantum mechanics. Scattering of photons on electrons (when ∆l = |F H| + |HG| = 2a sin θ = nλ,
electrons and photons are considered as elastic balls) is studied
where a is the distance between neighbouring crystal planes.
in the section “Quantum mechanics”; calculation of scattering
This is the main formula for Bragg condition: the X-rays are
probabilities is beyond the scope of the IPhO Syllabus.
reflected by a crystal if the grazing angle θ between the X-rays
Finally, X-rays can be reflected by a regularly arranged array
and the crystal plane satisfies the condition
of ions (present in crystals); this phenomenon called the Bragg
nλ
reflection. According to the Bragg model, crystal planes per- sin θ = .
2a
— page 7 —
 4. POLARIZATION. DOUBLE REFRACTION.
√
Pay attention to the fact that for the same crystal, a can take of molecules are around N . So, the relative √ fluctuations in
different values and can be arbitrarily small (e.g. in the figure the density of air are of the order of 1/ N : they grow with
above, purple planes are significantly closer than the red ones). decreasing N . The departure of the air’s coefficient of refraction
However, there is always a maximal value for a (corresponds to n from unity is proportional to the density of air. Hence, small
the red planes above). So, a crystal can reflect X-rays under fictitious volumes of air behave as media of different coeffi-
many angles, due to the different values of a and n; the smallest cient of refraction: there is partial reflection from these volume
value of θ corresponds to n = 1 and to the largest value of a. boundaries. The amount of reflected light is still small, because
Finally, don’t forget that the angle by which the X-rays are the difference in n is small. The effect is stronger for smaller
deflected equals to 2θ. fictitious volumes, but a wave cannot “discern” anything smaller
Note that the equality |AB2 | + |B2 C2 | − |AB1 | − |B1 C1 | = than ca a quarter of the wavelength. This is the reason why
|F H| + |HG| can be also verified geometrically. Indeed, the sky is blue: blue light has shorter wavelength and hence,
|KH| + |HL| = |AB2 | + |B2 C2 |; due to the congruence of can “see” smaller volumes with higher fluctuations in n than
grey triangles, |KH| + |HL| = |DH| + |HE|. Thus, |AB2 | + the other components of the sunlight. Now, what we see as a
|B2 C2 | = |DH| + |HE|, i.e. |AB2 | + |B2 C2 | − |AB1 | − |B1 C1 | = blue sky is a light reflected by a dielectric “interface”, which
|DH| + |HE| − |AB1 | − |B1 C1 | = |DH| + |HE| − |DF | − |GE| = is partially polarized. The polarization is the strongest for
|F H| + |HG|. Brewster angle, when the reflected and refracted beams are
perpendicular. Since n is very close to one, the refracted beam
goes almost along a straight line, parallel to a vector pointing
4 Polarization. Double refraction. to Sun. So, in the case of the Brewster angle, the reflected
Up till now we have implicitly assumed the light to be linearly (scattered) light is perpendicular to the direction of Sun: if you
polarized — by assuming a fixed axis (x) for the direction of elec- look into sky perpendicularly to the Sun, you see a strongly
tric field. Natural light, however, is in most cases non-polarized. polarized blue light.
This means that the direction of the electric field fluctuates in There are materials which have the so called double refrac-
time. This is effectively another aspect of non-coherence: after tion property; for a linearly polarized light, the coefficient of
a certain time period (it may be many wave periods, but in refraction depends on the polarization plane. Furthermore,
seconds, still a really tiny amount). The electric field of the some materials are transparent for one polarization plane, and
electromagnetic wave “forgets” its previous direction and takes opaque for the perpendicular one. These materials are used to
a new arbitrary direction (perpendicular to the direction of make linear polarizers — filters which let through only a light
propagation, see Section 1). This happens so fast that neither which is polarized in a specific plane. When a non-polarized
human eye nor common measuring devices are able to discern light goes through such a filter, half of the light energy is dissip-
the momentary directions of the electric field. From the point ated (the light which was polarized in a wrong direction), and
of view of diffraction studied above, this is not really important at the output, we have a completely polarized light. Such filters
since two coherent light beams (from the same source!) have are used in photography to reduce reflections from dielectric
the same momentary direction of the electric field (as long as surfaces (such as water or glass); in the case of Brewster angle,
the optical path difference of the beams does not exceed the the reflections can be removed entirely. Also, these filters can
coherence length). So, non-polarized light is an electromagnetic make sky darker and remove blue haze obscuring distant objects
wave which has randomly fluctuating direction of the electric (e.g. mountains); don’t forget that such a haze-removal works
field. best if the Sun is perpendicular to the direction of observation
However, it is possible to have light, the electric field of (see above).
which is always parallel to a fixed axis; the plane defined by The Brewster reflection can be used for precise measure-
this axis and propagation direction is called the polarization ments of the coefficient of refraction. For instance, when a
plane. In particular, when light is reflected by a dielectric inter- totally polarized laser light falls onto a dielectric surface, the
face under Brewster angle αB = arctan n, the reflected beam is reflected beam disappears for Brewster angle αB assuming that
completely polarized (electric field is parallel to the interface). the polarization plane of the laser light is perpendicular to
In this case, the reflected an refracted beams are perpendicular the surface. Then, αB can be measured, and n is found as
to each other. The refracted beam is also polarized, but only n = tan αB .
partially (most of the light is polarized perpendicularly to the Now, consider a case when a polarized light falls onto a
interface). Partially polarized light can be thought as a super- polarizer so that the polarization planes form an angle α. Let
position of two non-coherent polarized waves with perpendicular the polarizer’s polarization plane define the x-axis; suppose that
planes of polarization. before the polarizer, the electric field vector at its maximum
Light reflected by a dielectric surface is always somewhat is E~ 0 . This vector can be represented as a sum of two vectors
polarized, and the non-polarized component decreases as the ~
E0 = ~ex E0 cos α + ~ey E0 sin α, which represents the decomposi-
incidence angle approaches the Brewster angle. Similarly, the tion of the initial wave into two perpendicularly polarized com-
blue light from the sky is also partially polarized. This is be- ponents. The polarizer dissipates completely the y-component,
cause we see blue light from the sky due to Rayleigh scattering. and at the output we have electric field amplitude vector equal
If we consider a small fictitious volume of air, the number of to ~ex E0 cos α (~ex and ~ey are the unit vectors along x and y axis,
molecules in it fluctuates somewhat; if the average (expected) respectively). Let us recall that the intensity is proportional
number of molecules is N , typical fluctuations in the number to the squared amplitude; therefore, the transmitted light’s
— page 8 —
 4. POLARIZATION. DOUBLE REFRACTION.
intensity Transmittency
I = I0 cos2 α, 70 (%)
which is referred to as the Malus’ law (I0 is the intensity of the 60
incoming polarized light).
50
pr 9. Let us have two polarizers with perpendicular planes
40
of polarization. A non-polarized light beam of intensity I0 falls
onto such a system, and of course, no light can be detected at 30
the output. Now, a third polarizer is inserted between the two
polarizers, so that it forms an angle α with the polarization 20
plane of the first polarizer. What is the intensity of light at the
10 f (THz)
output?
500 520 540 560 580 600 620

Apart from the non-polarized light and linearly polarized

light, there is also a circularly polarized light and elliptically pr 11. Anemometer is a device measuring flow rate of a
polarized light. These can be obtained from the linearly po- gas or a fluid. Let us look the construction of a simple laser-
larized light by using double refracting plates. As mentioned anemometer. In a rectangular pipe with thin glass walls flows
above, the refraction coefficient of double refracting materials a fluid (refractive index n = 1,3), which contains light dissip-
depends on the polarization plane. In the case of a so-called ating particles. Two coherent plane waves with wavelength
quarter-wavelength plate, this effect leads to the optical path λ = 515 nm and angle α = 4◦ between their wave vectors,
difference between two components equal to λ4 . Let us choose are incident on a plate so that (a) angle bisector of the angle
x and y axis at the plate’s plane so that the y-polarized light between wave vectors is normal to one wall of the pipe and (b)
is retarded with respect to the x-polarized one by λ4 . Let us pipe is parallel to the plane defined by wave vectors. Behind
have linearly polarized light falling onto such a plate so that the pipe is a photodetector, that measures the frequency of
the polarization plane forms an angle α with the x-axis. Then, changes in dissipated light intensity.
at a certain point before the plate, the time-dependence of the
(i) How long is the (spatial) period ∆ of the interference pattern
electric field components are given by
created along x-axis (see Figure)?
Ex = E0 cos α sin(ωt), Ey = E0 sin α sin(ωt). (ii) Let the oscillation frequency of the photometer signal be
ν = 50 kHz. How large is the fluid’s speed v? What can be said
Thus, at any moment of time, Ey /Ex = tan α, i.e. the electric
about the direction of the fluid flow?
field vector oscillates along a line Ey = Ex tan α. After the
plate (iii) Let us consider a situation, when the wavelengths of the
π plane waves differ by δλ = 4,4 fm (1 fm= 10−15 m). What is the
Ex = E0 cos α sin(ϕ0 + ωt + ), Ey = E0 sin α sin(ϕ0 + ωt), frequency of signal oscillations now (fluid’s speed is the same
2
as in previous section)? Is it possible to determine the flow
where ϕ0 is the phase depending on the point of observation. direction with such a device?
From this system of equations, we can easily obtain
 2  2
Ex Ey
+ = E02 ,
cos α sin α
which is the equation of an ellipse: the endpoint of the electric
field vector draws such an ellipse. Therefore, such a light is said
to be elliptically polarized; in the particular case of α = π4 , it is
circularly polarized.
Sometimes it is important to use a linear polarizer, but to
avoid linearly polarized light at the output (e.g. in modern
cameras semi-transparent mirrors are used to split the light,
and if the light is linearly polarized, the balance between the
intensities of split beams becomes non-predictable). Then, a pr 12.
quarter-wavelength plate is attached to a polarizer so that As it is well known, a telescope makes it possible to see the
the output light becomes circularly polarized; these are called stars in daylight. Let us study the problem in more details.
circular polarizers. Consider a simplified model of the eye: a single lens with focal
length f = 4 cm and diameter d = 3 mm creating an image
pr 10. A thick glass plate is coated by a thin transparent on screen (retina). The model of a telescope is similar: a lens
film. The transmission spectrum of the system is depicted in of focal length F = 2 m and diameter D = 20 cm creating an
graph (light falls normal to the plate). The refractive index of image in focal plane (where eg. a film can be put). In your
the film n ≈ 1.3. What is the thickness of the film d? calculations, the following quantities can be used: the density of
— page 9 —
 4. POLARIZATION. DOUBLE REFRACTION.
the light energy radiated from a unit Solar surface in unit time diffraction of a plane, monochromatic wave that falls perpen-
w0 (the light power surface density); the ratio of the star and dicularly on a 2-dimensional grid that consists of N1 × N2 slits
Sun distances q = 4 · 105 (we assume that the star is identical with separations d1 and d2 . The diffraction pattern is observed
to the Sun); Solar angular diameter φ ≈ 9 mrad. Remark: If on a screen at a distance L from the grid. The screen is parallel
the answer contains w0 then numerical answer is not required. to the grid and L is much larger than d1 and d2 .
(i) Consider a sheet of paper, the normal of which is directed
towards the Sun. What is the surface density of the light power
w1 arriving to the sheet from the Sun?
(ii) Find the net power P2 of the light, which is focused by the
telescope into the image of the star.
(iii) Assume that blue sky is as bright as a sheet of grey paper
illuminated by Sun. You may assume that in the direction, per-
pendicular to the sheet, the ratio of the light power scattered
by the paper into a 1-steradian space angle, to the net light
power arriving to the sheet, is α ≈ 0,1 (this corresponds to the
dissipation of ca 70 % light energy in the grey paper). What is
the surface density of the light power in the focal plane of the
telescope w3 , due to the blue sky?
(iv) While studying the star image, let us ignore all the effects
other than diffraction. Estimate the surface density of the light
power in the centre of the star image w2 (in the focal plane of (i) Determine the positions and widths of the principal max-
the telescope), due to the light arriving from the star. imum on the screen. The width is defined as the distance
(v) Provide an expression for the ratio of the surface densities between the minima on either side of the maxima.
of the light powers k in the middle of the star image, and in a (ii) We consider now a cubic crystal, with lattice spacing a and
point farther away from it. size N0 a × N0 a × N1 a, where N1  N0 . The crystal is placed
(vi) Is it possible to see a star in daylight using a telescope? in a parallel X-ray beam along the z-axis at an angle θ (see
Plain eye? Motivate yourself. Fig. b). The diffraction pattern is again observed on a screen
at a great distance L  N0 a from the crystal. Calculate the
pr 13. [IPhO-1981] A detector of radiowaves in a radioastro- position and width of the maxima as a function of the angle θ
nomical observatory is placed on the sea beach at height h = 2 m for θ  1. What in particular are the consequences of the fact
above the sea level. After the rise of a star, radiating electro- that N1  N0 ?
magnetic waves of wavelength λ = 21 cm, above the horizon (iii) The diffraction pattern can also be derived by means of
the detector registers series of alternating maxima and minima.
Bragg’s theory, in which it is assumed that the X-rays are reflec-
The registered signal is proportional to the intensity of the de-
ted from atomic planes in the lattice. The diffraction pattern
tected waves. The detector registers waves with electric vector,
then arises from interference of these reflected rays with each
vibrating in a direction parallel to the sea surface. other. Show that this so-called Bragg reflection yields the same
(i) Determine the angle between the star and the horizon in conditions for the maxima as those that you found in (ii).
the moment when the detector registers maxima and minima (iv) In some measurements the so-called powder method is em-
(in general form). ployed. A beam of X-rays is scattered by a powder of very many,
(ii) Does the signal decrease or increase just after the rise of small crystals (Of course the sizes of the crystals are much larger
the star? than the lattice spacing, a.) Scattering of X-rays of wavelength
λ = 0.15 nm by Potassium Chloride [KCl] (which has a cubic
(iii) Determine the signal ratio of the first maximum to the
lattice, see Fig a) results in the production of concentric dark
next minimum. At reflection of the electromagnetic wave on
circles on a photographic plate. The distance between the crys-
the water surface, the ratio of the intensities of the electric field
tals and the plate is L = 0.10 m and the radius of the smallest
of the reflected (Er ) and incident (Ei ) wave follows the low:
circle is R = 0.053 m. (see Fig c). K+ and Cl− ions have almost
Er /Ei = (n − cos ϕ)/(n + cos ϕ),
the same size and they may be treated as identical scattering
where n is the refraction index and ϕ is the incident angle of the centres. Calculate the distance between two neighbouring K+
wave. For the surface “air-water” for λ = 21 cm, the refraction ions in the crystal.
index n = 9.
(iv) Does the ratio of the intensities of consecutive maxima and pr 15. [Est-PhO-2009] A hall of a contemporary art instal-
minima increase or decrease with rising of the star? Assume ment has white walls and white ceiling; the walls and the
that the sea surface is flat. ceiling are lit with a monochromatic green light of wavelength
λ = 550 nm. The floor of the hall is made of flat transparent
pr 14. [IPhO-1990] We wish to study X-ray diffraction by a glass plates. The lower surfaces of the glass plates are matte
cubic crystal lattice (see figure a). To do this we start with the and painted black; the upper surfaces are polished and covered
— page 10 —
 4. POLARIZATION. DOUBLE REFRACTION.
with thin transparent film. A visitor standing somewhere in pr 17. [Est-PhO-2004] Screen, two mirrors, and a source of
the room will see circular concentric bright and dark stripes on monochromatic light are positioned as shown in figure. Due to a
the floor, centred around himself. A curious visitor investigates shade, only reflected light from the source can reach the screen.
the phenomenon and concludes the following: in order to see There will be a striped interference pattern on the screen; the
the largest bright stripes, he needs to lower his viewpoint; the distance between the stripes is d. Express the wavelength of the
maximal number of observable stripes is N = 20. Determine light λ in terms of d and the distance a (see figure). Assume
the thickness of the film if the film’s coefficient of refraction is that a  d.
known to be n0 = 1.4, and that of the glass plates — n1 = 1.6.

screen
ror
mir
a
pr 16. [Est-PhO-2002] Circular resonator is a device used in a
fiber optics; it consists of circular loop made of an optical fibre,

mir
ror
coupled to two straight fibres as shown in the figure. Fibre
coupling is achieved by bringing the light-conducting cores
so close that electromagnetic waves can “tunnel” through the
inter-fibre-gap, from one fibre into the other one. In the case pr 18. [Est-PhO-2001] In fibre optics, devices called equal ra-
of circular resonators, the coupling between the fibres is very tio splitters are often used: these are devices where two optical
weak: if a light pulse propagates along the fibre A from left to fibres are brought into such a contact that if an electromagnetic
right, most of the light energy will pass the coupling point and wave is propagating in one fibre, at the contact point it splits
continue propagation towards A2 , and only a small fraction α into two equal amplitude waves, travelling in each of the fibres,
of the incident energy “jumps” over to the circular fibre B; let see figure.
incident wave outgoing waves
α = 0.01. Let us assume the following: (i) all three fibres have fibre “2”
identical properties; (ii) these are so called single-mode fibres,
i.e. light can travel only parallel to the fibre’s axis, without
fibre “1”
“bouncing” between the walls; (iii) the coupling between the
1. Show that if an equal ratio splitter splits an electro-
fibres B and C is identical to that of between A and C; (iv)
magnetic wave into two, after the contact point, there is a
monochromatic infrared light of intensity I0 is being led to the
phase shift of π2 between the two waves. Hint: use the en-
inlet A1 of the fibre A.
ergy conservation law; depending on your solution, equality
sin α + sin β = 2 sin α+β α+β
2 cos 2 can be useful.
ii. Consider now two sequentially positioned equal ratio split-
Graph below shows the dependence of the light intensity at the ters, as shown in the figure below (a device called the Mach-
outlet A2 on its wavelength λ. Zehnder interferometer). The optical path difference between
i. Sketch the intensity of light at the outlets C1 and C2 as a the inter-splitter segments of the two fibres is ∆ = 30 µm. As-
function of λ. suming that the wavelength of the incoming monochromatic
ii. What is the intensity of light in the fibre B for λ = 1600 nm? light varies from λ1 = 610 nm to λ2 = 660 nm, for which
iii. How long is the fibre B? The fibre’s coefficient of refraction wavelengths all the light energy is directed into the fibre “2”?
for the inrared light is n = 1.66. incident wave fibre “2” outgoing
C1 C2 wave “2”
C
outgoing
fibre “1” wave “1”

B
pr 19. [PhysCup-2012] In the figure below, there are five
0.01I0 images which were obtained as follows. Laser beam has been
I0
directed to a sheet of white paper (in one case, to a white wall
0.99I0 A
A1 A2 with a rough surface). The emerging bright spot of the laser
beam has been photographed with a digital camera; the lens
I0
axis was kept approximately perpendicular to the sheet of pa-
per. The camera has been focused to infinity (not to the bright
spot!); so, the sensor surface coincided with the focal plane of
I the camera lens. The images were taken with a lens of focal
length F = 300 mm; the diameter of the lens was D = 75 mm
(for one image, the effective diameter of the lens was reduced
by a diaphragm down to 38mm). Each image is a square crop
1650 1660 (nm) 1670 1680 from the image recorded by the sensor; the scaling factor of
these images can be calculated from the fact the pixel length
on the sensor was 9.6 µm; the respective size of each image is
— page 11 —
 4. POLARIZATION. DOUBLE REFRACTION.
also indicated in millimetres. 3. Apply method 1; note that the sum of three vectors of
670nm
(a) (c) 532nm equal length can be zero only if they form an equilateral
triangle.
9. Decompose the light after the first polarizer into two com-
ponents according to the axis defined by the middle polar-
izer; do the same for the light before the final polarizer.
670nm (b)
Answers
λ
± 31 ) ; ϕmax = arcsin n λa


2. ϕmin = arcsin a (n
7.5mm 7.5mm
404nm 404nm
9. I = 21 I0 cos2 α sin2 α = 1
8 sin2 2α.
10. The short-wavelength oscillations on the graph are due to
(f)
4.7mm the diffraction on the film, therefore the local maximum
condition is 2dn = λN = cN/ν. So, 2dnν = cN and
(f) 2dn(ν + δ)ν = c(N + 1), hence 2dnδν = c and d = c/2nδν.
In order to measure the distance between two maxima
more precisely, we take a longer frequency interval , e.g.
∆ν = 80 THz and count the number of maxima between
them, m ≈ 34. Consequently, δν = ∆ν/m ≈ 2.35 THz,
7.4mm (d) (e) 6.7mm and d ≈ 50 µm

Figure: images recorded by the sensor of a digital camera when a bright spot
was created on a sheet of white laser printer paper [except for image (e), which
was created on a white wall with a rough surface; the surface height fluctu-
ations were around 0.2 mm]. For each image, the number with nanometers
indicates the laser wavelength, and the number with millimetres indicates the
size (the side length of the square) of the image crop on the sensor. Image (f)
was uploaded two weeks after the publication of the problem as a hint: you
can see two brighter spots, because half of the laser beam hits a white paper,
and the other half hits the white wall; what was supposed to draw attention
(asking for an explanation) is that the spot on the paper is much brighter and
has a different tint (whiter) than the one on the wall.
Part A. Using these images, estimate the size of the bright
spot created by the red laser seen on Figures 1(a) and 1(b).
(So, inaccuracies within a factor of 2 are acceptable.) Note that
the bright spot was slightly elliptical and approximately of the
same size for all the three lasers used.
Part B. Explain why the image of the bright spot created
by the violet laser on a sheet of white paper [Fig. 1(d)] is
qualitatively different from all the other images.
11. (i) First we need to find the angle after the refraction
β: For small incidence angles we find approximately
β = α/n. In the liquid, the wavelength is decreased
n times: λ0 = λ/n. The requested wavelength can be
found as the distance between the lines connecting the
intersection points of the equal phase lines of the two
beams. Alternatively (and in a simpler way), it is found
as the difference of the two wavevectors: k 0 = kβ, where
k = 2π/λ0 = 2πn/λ is the wavevector of the incident beams.
So, ∆ = 2π/k 0 = λ/α ≈ 7,4 µm.
(ii) The scattered light fluctuates due to the motion of
the scattering particles; the frequency is ν = v/∆ = vα/λ.
There is no way to determine the direction of the flow, but
the modulus is obtained easily: v = νλ/α ≈ 0.37 m/s.
(iii) The spatial structure of the interference pattern re-
mains essentially unchanged (the wavelength difference is
negligible). However, the pattern obtains temporal fre-
quency δω = δ(c/λ) ≈ cδλ/λ2 . The velocity of the in-
terference pattern u = ∆δω = αc δλ λ . If the fluid speed is

appendix 1: Mathematically, we can derive this from the v ≈ 0.37 m/s, then the relative speed of the pattern and
superposition principle (Maxwell Eqns are linear, hence super- the fluid is ν 0 = αc δλ
λ ± v, depending on the direction of

position principle holds: any linear combination of solutions is the flow (in both cases, ν 0 ≈ 740 kHz). So, the output
also a solution), and from a branch of mathematics called the frequency allows us to determine the flow direction as long
Fourier analysis. The latter states that any function Rof x can be as we can be sure that the interference pattern velocity is
represented as a sum of sinusoidal functions: f (x) = R f e dk. ikx larger than the flow velocity.
Assume that for t = 0, E(z, E e dk .
k
~ t = 0) = ~e E (z) = ~e ikz 12. (i) The light flux density decreases inversely proportionally
Each of the sinusoidal components will evolve in time according
x ? x k
to the square of the distance, therefore w1 = w0 Rp2 /L2p ,
to Eq. (3), so that with ζ = z − vt, Z where Rp is the solar radius, and Lp — the solar distance.
Due to φ = 2Rp /Lp , we obtain w1 = w0 φ2 /4.
Z
~ t) = ~ex
E(z, Ek eik(z−vt) dk = ~ex Ek eikζ dk = ~ex E? (ζ),
(ii) The previous result can be applied to the star flux
i.e E(z,
~ t) = ~e E (z − vt).
x ? density, which is q −2 w1 ; hence P2 = 14 πD2 w1 q −2 =
w0 π(φD/4q)2 .
(iii) The paper surface area S radiates towards the lens of
Hints the telescope the power P3 = w1 αS( π4 D2 /L2 ), where L is
2. Apply method 1. the telescope distance. The image of this piece of paper
— page 12 —
 4. POLARIZATION. DOUBLE REFRACTION.
has size s = SF 2 /L2 ; thus, w3 = P3 /s = w1 α( π4 D2 /F 2 ) = then
w0 απ(φD/4F )2 .  
n − cos ϕ
(iv) The angular distance of the first diffraction min- Emax = Ei 1 + ,
n + cos ϕ
imum (using the single slit approximation — circle is
actually not a slit) is λ/D. Hence, the bright circle π
From the figure it is seen that ϕmax = 2 − αmax , we obtain
radius can be estimated as δ = F λ/D. Consequently, ø n - sin a max ˆ 2n
Emax = Ei ??1 + ˜˜ = Ei . (4)
w2 = P2 /πδ 2 = w0 (φD2 /4qF λ)2 . ? n + sin a max ∏ n + sin( 2a max )
At the interference minimum, the resulting intensity is:
(v) k = (w2 + w3 )/w3 = 1 + (απ)−1 (D/λq)2 ≈ 4 (assuming 2 sin a min
Emin = Ei - Er = Ei . (5)
λ ≈ 500 nm). n + sin a min
The intensity I of the signal is proportional to the square of the intensity of the electric
(vi) k − 1 ∼ 1 (or k − 1 > 1) means that the star can be eas- field E, therefore the ratio of the intensities of the consecutive maxima and minima is:
2
I max øEmax ˆ n2 (n + sin a min ) 2
ily seen (as is the case for the telescope); k − 1  1 means =? ˜ = . (6)
I min ?? Emin ˜∏ sin 2 a min (n + sin a max ) 2
that the star cannot be seen (for the eye, k − 1 ≈ 1 · 10−4 ). Using the eqs. (2) and (3), the eq. (6) can be transformed into the following form:
2
È l ?
14. (i) The signal, registered by the detector A, is result of the I max 4n 2 h 2? n + k 2h ?
= 2 2 ? ? .
interference of two rays: the ray 1, incident directly from I min k l ? n + (2k - 1) l ?
? 4h ?
the star and the ray 2, reflected from the sea surface (see Using this general formula, we can determine the ratio for the first maximum (k =1) and the
the figure) next minimum: 2
ø l ˆ
n+
I max 4n 2 h 2 ? 2 h
˜
= ? ˜ = 3.104
1 I min l2 ? n + l ˜
? ˜
? 4h ∏

(iv) Using that n  λ, from the Eq. two lines above it

B
follows :
A
2 2 Imax 4n2 h2
= 2 2 .
Imin k λ
h
So, with the rising of the star the ratio of the intensities of
C the consecutive maxima and minima decreases.

The phase of the second ray is shifted by π due to the re-

flection by a medium of larger refractive index. Therefore,
the phase difference between the two rays is: 14.
l h l ø h ˆ
D = AC + - AB = + -? ˜ cos(2a ) =
2 sin a 2 ? sin a ∏
l h l
= + [1 - cos(2a )] = + 2h sin a
2 sin a 2
The condition for an interference maximum is:
l
+ 2h sin a max = kl or
2
1 l l
sin a max = (k - ) = (2k - 1)
2 2h 4h

where k = 1, 2, 3, . . . , 19. (the difference of the optical

paths cannot exceed 2h, therefore k cannot exceed 19).
The condition for an interference minimum is:

λ λ
+ 2h sin αmax = (2k + 1) ⇒
2 2
kλ
sin αmax =
2h
where k = 1, 2, 3, . . . , 19.
(ii) Just after the rise of the star the angular height α is
zero, therefore the condition for an interference minimum
is satisfied. By this reason just after the rise of the star,
the signal will increase.
(iii) If the condition for an interference maximum is sat-
isfied, the intensity of the electric field is a sum of the
intensities of the direct ray Ei and the reflected ray Er ,
respectively: Emax = Ei + Er .
Because
n − cos ϕ
Er = Ei ,
n + cos ϕ
— page 13 —
 4. POLARIZATION. DOUBLE REFRACTION.

N√
λ
15. ≈ 13 µm.
2(n1 − n21 −1

— page 14 —
A Glimpse into the Special Theory of Relativity
Siim Ainsaar
1 Why relativity?..............................1
2 Postulates of special relativity.......2
3 Basic thought experiments............2
3.1 Time dilation................................2
3.2 Length contraction.......................3
3.3 Proper time..................................3
4 Lorentz transformations................3
4.1 Spacetime interval.......................3
4.2 Minkowski spacetime,
Poincaré transformations.............4
4.3 Rapidity.......................................4
4.4 Hyperbolic trigonometry..............5
4.5 Length contraction, time dilation
and velocity addition....................5
1 Why relativity?
Relativity is often seen as an intric­
ate theory that is necessary only
when dealing with really high
speeds or ultra­precise measure­
ments. However, there are some
quite often­encountered topics that
are paradoxical if treated non­
relativistically. These are also some
of the main sources of Olympiad
problems on relativity.
Think for a moment about two
charged initially stationary
particles. They “feel” only the elec­
trostatic force from each another.
But in another, moving reference
frame there is also the magnetic
force, in general, in a different dir­
ection! How could force depend on
the choice of inertial reference
frame? What principles forbid the
particles from colliding in one
1
January 2013
4.6 Light­cones, simultaneity and
causality......................................6
4.7 Lorentz transformations
algebraically.................................7
5 Dynamics.......................................7
5.1 Four­velocity and
four­acceleration..........................7
5.2 Mass, momentum and energy......8
5.3 Force............................................9
6 Optical effects................................9
7 Electromagnetism..........................9
8 Basic problems...............................9
9 Olympiad problems........................9
10 Further reading............................9
frame and departing in another?
Electromagnetism needs relativity
for an explanation.
Photons (thus, much of optics) are
always relativistic and other
particles often are. Anything where
the speed of light matters – for ex­
ample, the GPS measuring the time
for a radio signal to travel from
satellites – uses relativity.
Particle physics needs relativity in
several aspects. Particles cannot be
controlled in a modern accelerator
without taking into account their
relativistic dynamics. The only suc­
cessful quantum theory predicting
the outcomes of particle collisions,
quantum field theory, is relativistic.
Muons in cosmic rays would decay
long before reaching the ground,
but we still detect them thanks to
relativistic time dilation.
Relativistic theory of gravity – gen­
eral relativity – allows to formulate
the physics independently of
whether the reference frame is iner­
tial or not, thus unifying time and
space even more tightly. It is neces­
sary for astrophysics (precession of
planets' orbits, gravitational lens­
ing, black holes), and cosmology
(history and future of large­scale
structures).
* * *
In the following, we shall derive the
most important results of the spe­
cial theory of relativity, starting
from the fundamental postulates.
Most steps of the derivation are giv­
en as problems, which are also
good examples of what one can ask
in relativity and exercise the
reader's ability to use the theory.
The most important general tech­
nique for problem­solving is rota­
tion of Minkowski spacetime in
complex coordinates, this is de­
scribed in section 4. Section 3 shows
the way from postulates to the use­
ful techniques, its problems may be
skipped if concentrating purely on
Olympiad preparation. Section 4 is
mostly on kinematics, the following
ones develop dynamics, optics and
(briefly) electromagnetism from it.
Finally, some problems for prac­
tising are given.
2
2 Postulates of special 
relativity
1. The laws of all physics are
the same in every inertial ref­
erence frame.
A reference frame is inertial if and
only if objects onto which no force
acts move in a straight line with
constant velocity.
2. The speed of light in vacu­
um (c) is the same in every
inertial reference frame.
In SI, after defining the second,* the
metre is defined through fixing (ex­
actly!) c = 299,792,458 m/s.
3 Basic thought 
experiments
3.1 Time dilation
Problem 1. Consider a “light
clock” that works as follows. A
photon is emitted towards a
mirror at a known distance l
and reflected back. It is detec­
ted (almost) at the emitter
again. The time from the emis­
sion to the detection (a “tick”)
is measured to be t. Now we
* One second is the duration of
9,192,631,770 periods of the
radiation corresponding to the
transition between the two
hyperfine levels of the ground state
of the caesium-133 atom at rest at a
temperature of 0 K.
 look at the clock from a refer­
ence frame where the whole
apparatus is moving with velo­
city v perpendicularly to the light
beam. Assume that the lengths
perpendicular to the motion do
not change. How long is the
tick for us? (Hint: the light
beam follows a zig­zag path.)
The answer is given by the follow­
ing fact.
Fact 1. If the time interval
between to events happening
at a stationary point is t, then
in a reference frame where the
speed of the point is v the
time interval is γ t , where
the Lorentz factor
1 time τ – time measured by a
γ= . passenger on the spaceship –
√ v2
1− 2
c
Another useful quantity in relativ­
istic calculations is β=v /c . As
γ>1 , we see everything in a mov­
ing vehicle take longer than in a sta­
tionary one – time is dilated
(stretched) in a moving reference
frame.
3.2 Length contraction
Problem 2. Now consider the
same “light clock” as in Prob­
lem 1., but moving in parallel to
the light beam, with velocity v.
What is the distance to the mir­
3
ror in our reference frame, if
the distance in the stationary
frame is l?
The answer is in the following fact.
Fact 2. If the length of a station­
ary rod is l, then its length in a
reference frame moving in
parallel to the rod with speed
v is l / γ .
Lengths are contracted (compressed)
in the direction of motion.
3.3 Proper time
Problem 3. A spaceship flies
freely from (t 1 ,x 1 , y 1 , z 1 )
(event 1) to (t 2 , x2 , y 2 , z 2 )
(event 2). What is the proper
between these events? [Answer:
2 2 2 2
c τ =c (t 2 −t 1 ) −(x 2 −x 1 )
2
−( y 2 − y 1 ) −( z 2 −z 1 )
2
2
]
4 Lorentz transformations
4.1 Spacetime interval
In ordinary, Galilean relativity,
lengths and time intervals are abso­
lute. As we have now seen, the pos­
tulates of Einsteinian relativity im­
ply that neither is so, once speeds
become comparable to c. However,
proper time – time in a comoving
frame – must clearly be independ­
ent of our reference frame. There­
fore we can define a new invariant
quantity with the dimension of
length.
Fact 3. The spacetime interval
s= √ c 2 (Δ t)2 −(Δ x)2 −(Δ y)2 −(Δ z )2
is independent of the choice
of reference frame.
If s is a real number, the interval is
called time­like; if s is imaginary, the
interval is space­like. If s is zero, the
interval is light­like.
Fact 4. The interval between two
events on the same light­ray
(in vacuum) is zero – thus,
light­like.
4.2 Minkowski spacetime,
Poincaré transformations
We can say that spacetime points
are represented by position four­vec­
μ and, if we only rotate and do not
tors* x =(ct , x , y , z) and the in­
terval calculates the length of the
displacement four­vector Δ xμ .
However, this law of calculating the
length has important minus signs in
it, so these four­vectors form a
Minkowski spacetime, not the usual
Euclidean space, where lengths
would be calculated using the usual
Pythagoras' law.
We can reuse our familiar laws of
geometry if we introduce complex
* Four-vectors are customarily
labelled by Greek indices written
as superscripts; subscripts have a
meaning in more advanced theory.
4
numbers. Namely, the invariant
quantity
is= √(i c Δ t)2 +(Δ x)2 +(Δ y )2 +(Δ z)2
is now expressed just like Py­
thagorean theorem. So, the Euc­
lidean distance between two events
in the spacetime of (ict, x, y, z) is in­
dependent of reference frame.
What transformations of Euclidean
space leave lengths invariant? Rota­
tions and translations and combina­
tions thereof!
Fact 5. Changes of inertial refer­
ence frames correspond to ro­
tations and shifts in the space­
time coordinates ict, x, y and z.
In general, such transformations are
called the Poincaré transformations*
shift the coordinates, the Lorentz
transformations.
4.3 Rapidity
By what angle should we rotate the
axes? Clearly, as one axis has ima­
ginary numbers on it, the angle
must also be complex.
Luckily this poses no problems in
drawing the angle, as long as we
consider only one­dimensional mo­
tion: it turns out to be a purely ima­
ginary angle, so its cosine
* The Poincaré transformations are
also known as the inhomogeneous
Lorentz transformations.
 iα −i α
e +e
cos α= (a projection of the
2 sin α .
unit direction vector) is real (can be
drawn on the real x­axis) and its
iα
e −e
−i α
sine sin α= is purely ima­
2i
ginary (can be drawn on the ima­
ginary ict­axis).
Problem 4. Take two coordinate
systems, O and O', with the
spatial axes parallel and the
(spatial) origin* of O' moving in
the x­direction with velocity v.
Calculate the angle α between
the x­ and x'­axis. (Hint: make
a diagram with ict on one axis
and x on another. Add the ict'­
and x'­axes. Calculate the x­
and ict­coordinate of one arbit­
rary point the spatial origin of
O' passes through. The ratio of
these coordinates is tan α .)
Such a Lorentz transformation in­
volving only time and one spatial
coordinate is called the Lorentz boost
in the x­direction. The answer to the
problem is the following useful fact.
Fact 6. A Lorentz boost in the x­
direction from standstill to ve­
locity v corresponds to rota­
tion of x­ and ict­axis by an
angle of
v β
α=arctan =arctan .
ic i
* In the spatial origin, x = y = z = 0,
but ict changes.
5
Problem 5. Calculate cos α and
Fact 7. cos α=γ ,  sin α=β γ /i .
The quantity ϕ=α /i is a real di­
mensionless number and is called
the rapidity.
4.4 Hyperbolic trigonometry
Some imaginary units i and some
minuses can be eliminated by using
hyperbolic trigonometry. Employ­
α −α
e −e
ing the formulae sinh α= ,
2
α −α
e +e sinh α
cosh α= , tanh α=
2 cosh α
2 2
and cosh α−sinh α=1 you can
prove the following.
Problem 6. Prove for the rapidity
ϕ that tanh ϕ=β , cosh ϕ=γ
and sinh ϕ=β γ .
Consequently, using the inverse
function of hyperbolic tangent,
α=i ϕ=i artanh β .
4.5 Length contraction, time
dilation and velocity addition
Problem 7. Prove again the length
contraction formula of Fact 2.
Here use rotation of Minkowski
spacetime.
Problem 8. Prove similarly the
time dilation formula of Fact 1.
Fact 8. If an object moves with re­
spect to reference frame O'
with velocity u and O' moves
 with respect to frame O with
velocity v in the same direc­
tion, then the velocity of the
object in O is
w= u+v .
uv
1+ 2
c
Problem 9. Prove the velocity ad­
dition formula in the last fact.
tan α+tan β
(Hint: tan(α+β)=
1−tan α tan β
and
tanh α +tanh β
tanh(α+β)= .)
1 +tanh α tanh β
Problem 10. Show that the velocity
addition formula implies the
postulate that the speed of light
is universal. (Hint:  u =±c .)
Problem 11. Prove that if u and v in
the velocity addition formula
are both between −c and c ,
then so is w. (Hint: show that
dw
>0 – hence w is monoton­
du
ous – and use the result of the
last problem that u=±c cor­
responds to w=±c .)
Fact 9. If there exists a reference
frame where an object moves
slower than light, then it does
so in every reference frame.
4.6 Light­cones, simultaneity and
causality
The trajectory of a particle in the
space­time is called its world­line.
6
The world­line of a photon cuts a
very special wedge from the dia­
gram: the inside of the wedge can
be influenced event at the tip of the
cone; the outside cannot. The region
where an event can have influence
in is called the light­cone of the
event.
Fact 10. If the spacetime diagram
is scaled so that i metres (on
the ict­axis) is at the same dis­
tance from the origin as 1
metre (on the x­axis), then the
world­line of a photon is at
45º from either axis.
Fact 11. Simultaneity is relative.
Problem 12. In reference frame O,
two events take place at the
same time t = 0, but with spa­
tial separation Δ x . What is the
time Δt ' between them in ref­
erence frame O', which is mov­
ing in the x­direction with velo­
city v? [Answer:
2
Δt '=−γ v Δ x/ c ]
Fact 12. The order of two events
with time­like or light­like
separation is absolute. For
space­like separation, the or­
der depends on the reference
frame.
This means that only time­like or
light­like separation allows one
event to be the cause of another. De­
manding that the causality should
hold and, thus, no information may
be sent to the past, we get the fol­
lowing fact.
Fact 13. Information cannot
propagate faster than light in
vacuum.
This means, among many other im­
plications, that everything must be
somewhat deformable: if we push
one end of a long rod, then the push
will propagate to the other end
slower than c (probably much
slower).
4.7 Lorentz transformations
algebraically
Fact 14. When going to a reference
frame moving in the x­direc­
tion with velocity v, the time
and space coordinates of an
event transform under Lorentz
transformations as follows.
vx
t '= γ(t− 2 )
c boosts that act as rotations in the
x ' =γ (x−vt )
y '= y
z ' =z
Problem 13. Prove the last fact.
Problem 14. Show algebraically
that if boosting in both x­ and
y­directions, the order of boosts
matters.
Intuitively, as boosts are rotations,
their order should matter just like
the order of ordinary spatial rota­
tions matters: try turning a book
7
over around two different axes, re­
member the result and then repeat,
switching the axes. The result of
two successive boosts in different
directions is actually not just a boost
in a third direction, but adds some
rotation that depends on the order
of the boosts.
5 Dynamics
5.1 Four­velocity and
four­acceleration
Generalising from the position four­
vector x μ=(ct , x , y ,z ) intro­
duced in section 4.2, we now in gen­
erality define a four­vector as a col­
lection of four numbers
qμ=(qt ,q x ,q y , q z )
that trans­
forms under Lorentz transforma­
tions. The spatial components
(q x , q y , q z )≡⃗
q rotate just like a
usual vector. The time­ and space­
components are mixed by Lorentz
four­space of (iq , q
t
⃗ ) . A boost in
the x­direction is given just as in
Fact 14.
q t '= γ(qt −βq x )
q x '= γ(q x−β qt )
The Lorentz­invariant length of the
four­vector is
∣q ∣= √(q ) −(q ) −(q ) −(q ) .
μ t 2 x 2 y 2 z 2
We already know that periods of
proper time d τ are Lorentz­
invariant. Thus, the following deriv­
atives can be formed.
d xμ
Fact 15. Four­velocity v μ=
dτ
and four­acceleration
d vμ
aμ= of a particle are
dτ
four­vectors.
Problem 15. Show that the four­
velocity of a particle moving
with speed v in the x­direction
is (γ c , γ v , 0,0) .
Problem 16. What Lorentz­invari­
ant quantity is the length of the
four­velocity from the last
problem? [Answer: c]
As the x­direction was arbitrary, we
can generalize the answer as fol­
lows.
Fact 16. The length of any four­
velocity is c.
Problem 17. Show that the four­
acceleration of a particle mov­
ing and accelerating in the x­
direction with a three­accelera­
tion of magnitude a=dv/ dt is
(β γ4 a , γ 4 a , 0, 0) with in­
variant length a.
5.2 Mass, momentum and energy
Some texts about relativity distin­
guish the rest mass or invariant mass
m from the relativistic mass γ m ,
but this would be misleading for
8
discussing motion in several dimen­
sions. Therefore, in this studying
material, we refer to m as just the
mass. This mass is an intrinsic prop­
erty of any object and does not de­
pend on the reference frame.
Fact 17. The four­momentum of a
particle with mass m is the
μ μ
four­vector  p =m v .
Fact 18. pμ=( E/c ,⃗p ) where the
2
total energy E=γ m c and
the relativistic momentum
⃗p =γ m ⃗
v.
Note that here ⃗ v is the usual three­
velocity and not the spatial part of
the four­velocity that has an addi­
tional γ  in it.
Fact 19. The length of the four­
momentum is mc, whatever
the velocity is. Therefore,
E2 =( p c)2 +(m c 2)2 .
For massless particles (such as
photons),  E= pc .
Fact 20. In interactions, four­
momentum is conserved.
This encompasses both the conser­
vation of energy and the conserva­
tion of momentum.
Fact 21. The total energy can be
separated into the rest energy
E rest=m c 2 and the kinetic
energy E k=(γ−1)m c 2 .
Problem 18. Show that for low 6 Optical effects
2
mv
speeds, E k≈ . Problem 20. What is the apparent
2
length of a rod with rest length
Note that if an object has any intern­ l moving with velocity v in par­
al structure and, thus, internal en­
allel to the rod, if you take into
ergy, then it must be taken into ac­ account the finite travel times
count in its rest energy and, thus, its
of photons from its ends to our
(rest) mass. eyes?
On the other hand, for any ul­
Fact 25. Doppler shift of the fre­
trarelativistic object moving almost
with a speed of c, the rest energy
and the rest mass can be neglected;
quency of light:  ν' =ν 0
√
c−v
c+v
.

thus, E≈ pc . Problem 21. Prove the formula,

Since the speed of light, c, corres­ considering the world­lines of
ponds to γ=∞ , we can deduce the two wave­crests.
following.
Problem 22. Reprove the for­
mula using E=h ν .
Fact 22. It takes infinite energy to
accelerate a massive object At least two important relativistic
to c. Massless particles move optical effects have been left out of
only with a speed of c. this studying material, but are still
worthwhile to think about:
5.3 Force
• Measuring the Astronomic­
⃗ d ⃗p = d (γ m v ) .
Fact 23. F=
al Unit through aberration
dt dt • Compton scattering
Fact 24. Four­force
d pμ 7 Electromagnetism
Fμ =m aμ= .
dτ The Lorentz force acting on a particle
Problem 19. Show that if all the with charge q moving in an electro­
motion is in x­direction, then magnetic field is ⃗ F =q ⃗
E + q⃗v × ⃗
B . If
we separate the fields into compon­
Fμ =(β γ F ,γ F , 0, 0) .
ents parallel and perpendicular to
In general,Fμ =(γ ⃗v⋅⃗ F /c , γ ⃗
F) ⃗v , it can be shown that the electric
v⋅⃗
where  ⃗ F=dE/ dt  is the power. and magnetic fields transform into

9
each other upon Lorentz transform­
ations:
Fact 26. ⃗ ⃗ ∥ ,  ⃗
E∥ '= E B ∥' = B⃗∥ ,
⃗ ⃗ ⊥ +⃗v × ⃗
E ⊥ '= γ( E B⊥ ) ,
⃗
B ⊥ '=γ ( ⃗
B⊥ −⃗ v ×E ⃗ ⊥ /c 2 ) .
8 Additional problems
The following problems have been
translated from an Estonian book.*
Problem 23. A rod with rest length
l 0 is moving translationally
with speed v in such a way that
the line connecting its end­
points at an instant forms an
angle ϕ with the direction of
motion. Find its length. [An­
l0
swer: ] (PK200)
γ √ 1−β sin ϕ
2 2
Problem 24. A body is moving uni­
formly in a circle, an orbit takes
t = 3 h. A clock inside the body
sees it to take τ=30min . Find
the radius R of the orbit. [An­
swer: c √ t −τ /(2 π) ] (PK201)
2 2
* Paul Kard, “Elektrodünaamika ja
spetsiaalse relatiivsusteooria
ülesannete kogu” (“A collection of
problems on electrodynamics and
special relativity”),
Tartu State University 1961. Here
we cite it as “PK”, followed by the
problem number.
10
Problem 25. The characteristic life­
time of a muon at rest is
−6
τ=2.2⋅10 s. How long a path
s can it travel since its creation,
if its speed is v = 0.999c? [An­
swer: γ v τ=14.7 km] (PK202)
Problem 26. A pion at rest decays
into a muon and a neutrino.
Find the total energy E and the
kinetic energy T of the muon, if
the rest masses of the pion and
the muon are, respectively, m π
and m μ ; the rest mass of the
neutrino is zero. [Answers:
2 2
(m π+mμ )c
2
E= ,
2 mπ
2 2
(m π −m μ ) c (PK234)
T= ]
2 mπ
Problem 27. A muon at rest decays
into an electron and two neutri­
nos. The rest mass of the muon
is μ , the mass of the electron
is m, the mass of the neutrino is
zero. Find the maximum pos­
sible energy E max of the elec­
1+(μ / m)
2
tron. [Answer: mc 2 ]
2 (μ /m)
(PK235)
Problem 28. At least how big must
be the energy E of a pion, if its
collision with a nucleon at rest
produces a nucleon­antinucle­
on pair and the pion is ab­
sorbed? The rest masses of the
nucleon and the pion are, re­
 2 2 −1
spectively, M and m. [Answer: c(γ +3 γ /4)(2+3 γ+5 γ / 4) ]
2 2 2 (Part of PK249)
(8 M −m )c /(2M ) ] (PK244)

Problem 29. At least how big

should be the energy E of a nuc­
leon, if its collision with a nucle­
on at rest produces a nuc­
leon­antinucleon pair and the
original nucleons are both
intact? The rest mass of the nuc­
2
leon is M. [Answer:  7 M c ] (PK245)
Problem 30. An atom with
rest mass m, at rest, radiates a
photon with frequency ν . What
is the rest mass m 0 of the atom
after the process? [Answer:
m √ 1−2 hν/(mc2 ) ] (PK248)
Problem 31. The difference
between an excited energy level
and the ground level of an atom
is Δ E . What should the speed v
of the excited atom be, if we
want a photon, that is radiated
in the direction of motion, have a
frequency of Δ E /h ? The rest
mass of the atom in its ground
state is m. [Answer:
9 Olympiad problems
See the following pages for the ori­
ginal texts of the problems.
• Cuba 1991 (Relativistic
Square)
• Iceland 1998 (Faster than
Light?)
• Taiwan 2003 (Neutrino De­
cay)
• China 1994 (Relativistic
Particle)
• Australia 1995 (Gravitation­
al Red­shift)
• Physics Cup 2012 (Electron­
Positron annihilation)
10 Further reading
• Ta­Pei Cheng, “Relativity,
gravitation, and cosmology:
a basic introduction”, Ox­
ford University Press, 2005,
2006

11
 T H EO RE TI CAL PROBLEMS

Pr oblem 1
The figure 1.1 shows a solid, homogeneous ball radius R. Before falling to the floor its center of mass is
at rest, but the ball is spinning with angular velocity ω0 about a horizontal axis through its center. The
lowest point of the ball is at a height h above the floor.

When released, the ball falls under gravity, and rebounds to a new height such that its lowest point is now
ah above the floor. The deformation of the ball and the floor on impact may be considered negligible.
Ignore the presence of the air. The impact time, although, is finite.

The mass of the ball is m, the acceleration due the gravity is g, the dynamic coefficient of friction between
the ball and the floor is µk, and the moment of inertia of the ball about the given axis is:
2mR 2
I=
5
You are required to consider two situations, in the first, the ball slips during the entire impact time, and in
the second the slipping stops before the end of the impact time.

Situation I: slipping throughout the impact.

Find:
a) tan θ , where θ is the rebound angle indicated in the diagram;
b)the horizontal distance traveled in flight between the first and second impacts;
c) the minimum value of ω0 for this situations.

Situation II: slipping for part of the impacts.

Find, again:
a) tan θ;
b)the horizontal distance traveled in flight between the first and second impacts.
Taking both of the above situations into account, sketch the variation of tan θ with ω0.

Pr oblem 2
In a square loop with a side length L, a large number of balls of negligible radius and each with a charge q
are moving at a speed u with a constant separation a between them, as seen from a frame of reference that
is fixed with respect to the loop. The balls are arranged on the loop like the beads on a necklace, L being
much greater than a, as indicated in the figure 2.1. The no conducting wire forming
the loop has a homogeneous charge density per unit length in the in the frame of the loop. Its total charge
is equal and opposite to the total charge of the balls in that frame.
Consider the situation in which the loop moves with velocity v parallel to its side AB (fig. 2.1) through a
homogeneous electric field of strength E which is perpendicular to the loop velocity and makes an angle θ
with the plane of the loop.

Taking into account relativistic effects, calculate the following magnitudes in the frame of reference of an
observer who sees the loop moving with velocity v:
a) The spacing between the balls on each of the side of the loop, aAB , aBC , aCD , y aDA.
b)The value of the net charge of the loop plus balls on each of the side of the loop: QAB , QBC , QCD y, QDA
c) The modulus M of the electrically produced torque tending to rotate the system of the loop and the
balls.
d)The energy W due to the interaction of the system, consisting of the loop and the balls with the electric
field.
All the answers should be given in terms of quantities specified in the problem.
Note. The electric charge of an isolated object is independent of the frame of reference in which the
measurements takes place. Any electromagnetic radiation effects should be ignored.

Some formulae of special relativity

Consider a reference frame S’ moving with velocity V with reference to another reference frame S. The
axes of the frames are parallel, and their origins coincide a t = 0. V is directed along the positive direction
of the x axis.

Relativistic sum of velocities

If a particle is moving with velocity u’ in the x’ direction , as measured in S’, the velocity of the particle
measured in S is given by:
u′ + V
u=
u′V
1+ 2
c
Relativistic Contraction

If an object at rest in frame S has length L0 in the x-direction, an observer in frame S’ (moving at velocity
V in the x-direction} will measure its length to be:

v2
L = L0 1 −
c2
� ������ ���� ������
��� ������� ����
ÁÒ Ø × ÔÖÓ Ð Ñ Û Ò ÐÝÞ Ò ÒØ ÖÔÖ Ø Ñ ×ÙÖ Ñ ÒØ× Ñ Ò ½ ÓÒ Ö Ó Û Ú
Ñ ×× ÓÒ ÖÓÑ ÓÑÔÓÙÒ ×ÓÙÖ Û Ø Ò ÓÙÖ Ð Üݺ
Ì Ö Ú Ö Û × ØÙÒ ØÓ ÖÓ Ò Ó Ö Ó Û Ú × Ó Û Ú Ð Ò Ø × Ó × Ú Ö Ð
ÒØ Ñ Ø Ö׺ ÙÖ ¿º½ × ÓÛ× × Ö × Ó Ñ × Ö ÓÖ Ø Ö ÒØ Ø Ñ ×º Ì ÓÒØÓÙÖ×
Ò Ø ÓÒ×Ø ÒØ Ö Ø ÓÒ ×ØÖ Ò Ø Ò ÑÙ Ø × Ñ Û Ý × ÐØ ØÙ ÓÒØÓÙÖ× ÓÒ
Ó Ö Ô Ð Ñ Ôº ÁÒ Ø ÙÖ Ø ØÛÓ Ñ Ü Ñ Ö ÒØ ÖÔÖ Ø × × ÓÛ Ò ØÛÓ Ó Ø×
ÑÓÚ Ò Û Ý ÖÓÑ ÓÑÑÓÒ ÒØ Ö × ÓÛÒ Ý ÖÓ×× × Ò Ø Ñ ×º ´Ì ÒØ Ö¸ Û
× ××ÙÑ ØÓ Ü Ò ×Ô ¸ × Ð×Ó ×ØÖÓÒ Ö Ø ÓÒ Ñ ØØ Ö ÙØ Ñ ÒÐÝ Ø ÓØ Ö
Û Ú Ð Ò Ø ×µº Ì Ñ ×ÙÖ Ñ ÒØ× ÓÒ Ù Ø ÓÒ Ø Ú Ö ÓÙ× Ø × Û Ö Ñ ØØ × Ñ
ØÑ Ó Ýº
Ì × Ð Ó Ø ÙÖ × Ú Ò Ý Ð Ò × Ñ ÒØ × ÓÛ Ò ÓÒ Ö × ÓÒ ´ ×µº ´½ ×
½ ¿ ¼¼ Ó Ö µº Ì ×Ø Ò ØÓ Ø Ð ×Ø Ð Ó Ý Ø Ø ÒØ Ö Ó Ø ÙÖ ¸ Ò Ø
Ý ÖÓ×× ×¸ × ×Ø Ñ Ø ØÓ Ê ½¾º Ô º ÐÓÔ Ö× ´ Ô µ ÕÙ Ð× ¿º¼ �½¼�� Ѻ Ì
×Ô Ó Ð Ø × ¿º¼¼ �½¼� ѻ׺ ÖÖÓÖ Ð ÙÐ Ø ÓÒ× Ö ÒÓØ Ö ÕÙ Ö Ò Ø ×ÓÐÙØ ÓÒº

µ ´¾ ÔÓ ÒØ×µ Ï ÒÓØ Ø Ò ÙÐ Ö ÔÓ× Ø ÓÒ× Ó Ø ØÛÓ Ø Ö Ó Ñ ØØ Ö׸ Ö Ð Ø Ú

ØÓ Ø ÓÑÑÓÒ ÒØ Ö¸ Ý � ´Øµ Ò � ´Øµ¸ Û Ö Ø ×Ù × Ö ÔØ× ½ Ò ¾ Ö Ö ØÓ Ø Ð Ø
Ò Ö Ø Ò ÓÒ ×¸ Ö ×Ô Ø Ú Ðݸ Ò Ø × Ø Ø Ñ Ó Ó × ÖÚ Ø ÓÒº Ì Ò ÙÐ Ö ×Ô ×¸ ×
× Ò ÖÓÑ Ø ÖØ ¸ Ö � Ò � º Ì ÓÖÖ ×ÔÓÒ Ò ÔÔ Ö ÒØ ØÖ Ò×Ú Ö× Ð Ò Ö ×Ô ×
Ó Ø ØÛÓ ×ÓÙÖ × Ö ÒÓØ Ý Ú�� Ò Ú�� º
�
�
�
�

Í× Ò ÙÖ ¿º½¸ Ñ Ö Ô ØÓ Ò Ø ÒÙÑ Ö Ð Ú ÐÙ × Ó � Ò � Ò Ñ ÐÐ ¹ Ö ¹
× ÓÒ × Ô Ö Ý ´Ñ ×» µº Ð×Ó Ø ÖÑ Ò Ø ÒÙÑ Ö Ð Ú ÐÙ × Ó Ú�� Ò Ú�� ¸ Ò ÛÖ Ø �
�
�
�

ÐÐ Ò×Û Ö× ÓÒ Ø Ò×Û Ö × Øº ´ ÓÙ Ñ Ý ÔÙÞÞÐ Ý ×ÓÑ Ó Ø Ö ×ÙÐØ×µº

µ ´¿ ÔÓ ÒØ×µ ÁÒ ÓÖ Ö ØÓ Ö ×ÓÐÚ Ø ÔÙÞÞÐ Ö×Ò Ò Ô ÖØ ´ µ¸ ÓÒ× Ö Ð Ø¹×ÓÙÖ

ÑÓÚ Ò Û Ø Ú ÐÓ ØÝ Ú Ø Ò Ò Ð ´¼ � � µ ØÓ Ø Ö Ø ÓÒ ØÓÛ Ö × ×Ø ÒØ
Ó × ÖÚ Ö Ç ´ ÙÖ ¿º¾µº Ì ×Ô Ñ Ý ÛÖ ØØ Ò × Ú ¬ ¸ Û Ö × Ø ×Ô Ó
Рغ Ì ×Ø Ò ØÓ Ø ×ÓÙÖ ¸ × Ñ ×ÙÖ Ý Ø Ó × ÖÚ Ö¸ × Êº Ì Ò ÙÐ Ö ×Ô
Ó Ø ×ÓÙÖ ¸ × × Ò ÖÓÑ Ø Ó × ÖÚ Ö¸ × ¸ Ò Ø ÔÔ Ö ÒØ Ð Ò Ö ×Ô Ô ÖÔ Ò ÙÐ Ö
ØÓ Ø Ð Ò Ó × Ø × Ú º �
�

Ò Ò Ú Ò Ø ÖÑ× Ó ¬¸ Ê Ò
�
�
Ò ÛÖ Ø ÝÓÙÖ Ò×Û Ö ÓÒ Ø Ò×Û Ö × Øº

µ ´½ ÔÓ Òص Ï ××ÙÑ Ø Ø Ø ØÛÓ Ø Ó Ø׸ × Ö ÒØ ÒØÖÓ Ù Ø ÓÒ Ò Ò

Ô ÖØ ´ µ¸ Ö ÑÓÚ Ò Ò ÓÔÔÓ× Ø Ö Ø ÓÒ× Û Ø ÕÙ Ð ×Ô × Ú ¬ º Ì Ò Ø Ö ×ÙÐØ× Ó
Ô ÖØ ´ µ Ñ Ø ÔÓ×× Ð ØÓ Ð ÙÐ Ø ¬ Ò ÖÓÑ Ø Ò ÙÐ Ö ×Ô × � Ò � Ò Ø
×Ø Ò Êº À Ö ×Ø Ò Ð Ò Ò Ô ÖØ ´ µ¸ ÓÖ Ø Ð Ø Ò Ó Ø¸ ÓÖÖ ×ÔÓÒ Ò
ØÓ ×Ù × Ö ÔØ ½ Ò Ô ÖØ ´ µº
Ö Ú ÓÖÑÙÐ × ÓÖ ¬ Ò Ò Ø ÖÑ× Ó ÒÓÛÒ ÕÙ ÒØ Ø × Ò Ø ÖÑ Ò Ø Ö ÒÙÑ Ö Ð
Ú ÐÙ × ÖÓÑ Ø Ø Ò Ô ÖØ ´ µº ÏÖ Ø ÝÓÙÖ Ò×Û Ö× Ò Ø × Ò Ø Ð × ÓÒ Ø Ò×Û Ö
× Øº

µ ´¾ ÔÓ ÒØ×µ ÁÒ Ø ÓÒ ¹ Ó Ý × ØÙ Ø ÓÒ Ó Ô ÖØ ´ µ¸ Ò Ø ÓÒ Ø ÓÒ ÓÖ Ø ÔÔ Ö ÒØ
Ô ÖÔ Ò ÙÐ Ö ×Ô Ú ØÓ
�
�
Ð Ö Ö Ø Ò Ø ×Ô Ó Ð Ø º
� �������� ����� ������������ ������ ������� ��� ���������� ������������

½
ÙÖ ¿º½ Ê Ó Ñ ×× ÓÒ ÖÓÑ ×ÓÙÖ Ò ÓÙÖ Ð Üݺ

½
 Ú

Ç Ê

ÙÖ ¿º¾ Ì Ó × ÖÚ Ö × Ø Ç Ò Ø ÓÖ Ò Ð ÔÓ× Ø ÓÒ Ó Ø Ð Ø ×ÓÙÖ × Ø º Ì

Ú ÐÓ ØÝ Ú ØÓÖ × Úº

ÏÖ Ø Ø ÓÒ Ø ÓÒ Ò Ø ÓÖÑ ¬ ´ µ Ò ÔÖÓÚ Ò Ò ÐÝØ ÜÔÖ ×× ÓÒ ÓÖ Ø

ÙÒ Ø ÓÒ ÓÒ Ø Ò×Û Ö × Øº
Ö Û ÓÒ Ø Ö Ô Ò×Û Ö × Ø Ø Ô Ý× ÐÐÝ Ö Ð Ú ÒØ Ö ÓÒ Ó Ø ´¬ µ¹ÔÐ Ò º
Ë ÓÛ Ý × Ò Ò Û Ô ÖØ Ó Ø × Ö ÓÒ Ø ÓÒ Ø ÓÒ Ú ÓР׺�
�

µ ´½ ÔÓ Òص ËØ ÐÐ Ò Ø ÓÒ ¹ Ó Ý × ØÙ Ø ÓÒ Ó Ô ÖØ ´ µ¸ Ò Ò ÜÔÖ ×× ÓÒ ÓÖ Ø Ñ Ü ÑÙÑ

Ú ÐÙ ´Ú µ��� Ó Ø
�
�
ÔÔ Ö ÒØ Ô ÖÔ Ò ÙÐ Ö ×Ô Ú ÓÖ Ú Ò ¬ Ò ÛÖ Ø Ø Ò Ø
�
�

× Ò Ø Ð ÓÒ Ø Ò×Û Ö × Øº ÆÓØ Ø Ø Ø × ×Ô Ò Ö × × Û Ø ÓÙØ Ð Ñ Ø Û Ò

¬ � ½º

µ ´½ ÔÓ Òص Ì ×Ø Ñ Ø ÓÖ Ê Ú Ò Ò Ø ÒØÖÓ Ù Ø ÓÒ × ÒÓØ Ú ÖÝ Ö Ð Ð º Ë ÒØ ×Ø×

Ú Ø Ö ÓÖ ×Ø ÖØ ×Ô ÙÐ Ø Ò ÓÒ ØØ Ö Ò ÑÓÖ Ö Ø Ñ Ø Ó ÓÖ Ø ÖÑ Ò Ò Êº
ÇÒ ÓÖ Ø × Ó × × ÓÐÐÓÛ׺ ××ÙÑ Ø Ø Û Ò ÒØ Ý Ò Ñ ×ÙÖ Ø ÓÔÔÐ Ö
× Ø Û Ú Ð Ò Ø × � Ò �Ó Ö Ø ÓÒ ÖÓÑ Ø ØÛÓ Ø Ó Ø׸ ÓÖÖ ×ÔÓÒ Ò ØÓ
Ø × Ñ ÒÓÛÒ ÓÖ Ò Ð Û Ú Ð Ò Ø � Ò Ø Ö ×Ø Ö Ñ × Ó Ø Ó Ø׺
ËØ ÖØ Ò ÖÓÑ Ø ÕÙ Ø ÓÒ× ÓÖ Ø Ö Ð Ø Ú ×Ø ÓÔÔÐ Ö × Ø¸
� ���
� ´½ � ¬ Ó× µ´½ � ¬ µ ¸ Ò ××ÙÑ Ò ¸ × ÓÖ ¸ Ø Ø ÓØ Ó Ø× Ú Ø × Ñ
�

×Ô ¸ Ú¸ × ÓÛ Ø Ø Ø ÙÒ ÒÓÛÒ ¬ Ú Ò ÜÔÖ ×× Ò Ø ÖÑ× Ó � ¸ � ¸ Ò � ×

� �
« �
¬ ½� �
´¿º½µ
´ �· �µ

ÏÖ Ø Ø ÒÙÑ Ö Ð Ú ÐÙ Ó Ø Ó ÒØ « Ò Ø × Ò Ø Ð ÓÒ Ø Ò×Û Ö × Øº
ÓÙ Ñ Ý ÒÓØ Ø Ø Ø × Ñ Ò× Ø Ø Ø ×Ù ×Ø Û Ú Ð Ò Ø Ñ ×ÙÖ Ñ ÒØ× Û ÐÐ Ò
ÔÖ Ø ÔÖÓÚ Ò Û ×Ø Ñ Ø Ó Ø ×Ø Ò º

��� ��������

µ ÇÒ ÙÖ ¿º½ Û Ñ Ö Ø ÒØ Ö× Ó Ø ×ÓÙÖ × × Ò ØÐÝ × Û Òº Ä Ø � ´Øµ

Ø Ò ÙÐ Ö ×Ø Ò Ó Ø Ð Ø ÒØ Ö ÖÓÑ Ø ÖÓ×× × ÙÒ Ø ÓÒ Ó Ø Ñ Ò � ´Øµ Ø
Ò ÙÐ Ö ×Ø Ò Ó Ø Ö Ø ÒØ Öº Ï Ñ ×ÙÖ Ø × ÕÙ ÒØ Ø × ÓÒ Ø ÙÖ Ø Ø
Ú Ò Ø Ñ × Ý ÖÙÐ Ö Ò ÓÒÚ ÖØ ØÓ Ö × ÓÒ × ÓÖ Ò ØÓ Ø Ú Ò × Ð º Ì × Ö ×ÙÐØ×
ÒØ ÓÐÐÓÛ Ò ÒÙÑ Ö Ð Ø

½
 Theoretical Question 3
Part A
Neutrino Mass and Neutron Decay
A free neutron of mass mn decays at rest in the laboratory frame of reference into
three non-interacting particles: a proton, an electron, and an anti-neutrino. The rest
mass of the proton is mp, while the rest mass of the anti-neutrino mv is assumed to be
nonzero and much smaller than the rest mass of the electron me. Denote the speed of
light in vacuum by c. The measured values of mass are as follows:
mn＝939.56563 MeV/c2, mp＝ 938.27231 MeV/c2, me＝0.5109907 MeV/c2
In the following, all energies and velocities are referred to the laboratory frame. Let E
be the total energy of the electron coming out of the decay.
(a) Find the maximum possible value Emax of E and the speed vm of the anti-neutrino
when E = Emax. Both answers must be expressed in terms of the rest masses of the
particles and the speed of light. Given that mv < 7.3 eV/c2, compute Emax and the
ratio vm /c to 3 significant digits. [4.0 points]
Theoretical Problem 1
RELATIVISTIC PARTICLE
In the theory of special relativity the relation between energy E and momentum P
or a free particle with rest mass m0 is

E  p 2 c 2  m02 c 4 mc 2

When such a particle is subject to a conservative force, the total energy of the

particle, which is the sum of p 2 c 2  m02 c 4 and the potential energy, is conserved. If

the energy of the particle is very high, the rest energy of the particle can be ignored
(such a particle is called an ultra relativistic particle).
1) consider the one dimensional motion of a very high energy particle (in which
rest energy can be neglected) subject to an attractive central force of constant
magnitude f. Suppose the particle is located at the centre of force with initial
momentum p0 at time t=0. Describe the motion of the particle by separately
plotting, for at least one period of the motion: x against time t, and momentum
p against space coordinate x. Specify the coordinates of the “turning points” in
terms of given parameters p0 and f. Indicate, with arrows, the direction of the
progress of the mothon in the (p, x) diagram. There may be short intervals of
time during which the particle is not ultrarelativistic. However, these should be
neglected.
Use Answer Sheet 1.
2) A meson is a particle made up of two quarks. The rest mass M of the meson is
equal to the total energy of the two-quark system divided by c2.
Consider a one--dimensional model for a meson at rest, in which the two
quarks are assumed to move along the x-axis and attract each other with a force
of constant magnitude f It is assumed they can pass through each other freely.
For analysis of the high energy motion of the quarks the rest mass of the quarks
can be neglected. At time t=0 the two quarks are both at x=0. Show separately
the motion of the two quarks graphically by a (x, t) diagram and a (p, x)
diagram, specify the coordinates of the “turning points” in terms of M and f,
indicate the direction of the process in your (p, x) diagram, and determine the
maximum distance between the two quarks.
Use Answer Sheet 2.
3) The reference frame used in part 2 will be referred to as frame S, the Lab frame,
referred to as S, moves in the negative x-direction with a constant velocity
v=0.6c. the coordinates in the two reference frames are so chosen that the point

2
 x=0 in S coincides with the point x  0 in S  at time t t  0 . Plot the
motion of the two quarks graphically in a ( x , t  ) diagram. Specify the
coordinates of the turning points in terms of M, f and c, and determine the
maximum distance between the two quarks observed in Lab frame S  .
Use Answer Sheet 3.
The coordinates of particle observed in reference frames S and S  are related
by the Lorentz transformation

 x ( x  ct )

 x
 t  (t   c )

where  v / c ,  1 / 1  2 and v is the velocity of frame S moving relative

to the frame S  .
4) For a meson with rest energy Mc2=140 MeV and velocity 0.60c relative to the
Lab frame S  , determine its energy E in the Lab Frame S  .

ANSWER SHEET 1 ANSWER SHEET 2

1) 2)
x x1, x2

t t
O

p p1 p2

x x1 x2
O O O

Quark1 Quark2
The maximum distance between
the two quarks is d=

3
Theoretical Question 1
Gravitational Red Shift and the Measurement of Stellar Mass
(a) (3 marks)
A photon of frequency f possesses an effective inertial mass m determined by its energy. Assume
that it has a gravitational mass equal to this inertial mass. Accordingly, a photon emitted at the
surface of a star will lose energy when it escapes from the star’s gravitational field. Show that the
frequency shift Δf of the photon when it escapes from the surface of the star to infinity is given by
Δf GM
�− 2
f Rc
for Δf � f where:
• G = gravitational constant
• R = radius of the star
• c = velocity of light
• M = mass of the star.
Thus, the red-shift of a known spectral line measured a long way from the star can be used to
measure the ratio M/R. Knowledge of R will allow the mass of the star to be determined.
(b) (12 marks)
An unmanned spacecraft is launched in an experiment to measure both the mass M and radius
R of a star in our galaxy. Photons are emitted from He+ ions on the surface of the star. These
photons can be monitored through resonant absorption by He+ ions contained in a test chamber
in the spacecraft. Resonant absorption accors only if the He+ ions are given a velocity towards the
star to allow exactly for the red shifts.
As the spacecraft approaches the star radially, the velocity relative to the star (v = βc) of the He +
ions in the test chamber at absorption resonance is measured as a function of the distance d from
the (nearest) surface of the star. The experimental data are displayed in the accompanying table.
Fully utilize the data to determine graphically the mass M and radius R of the star. There is no
need to estimate the uncertainties in your answer.
Data for Resonance Condition
Velocity parameter β = v/c (×10−5 ) 3.352 3.279 3.195 3.077 2.955
Distance from surface of star d (×108 m) 38.90 19.98 13.32 8.99 6.67

(c) (5 marks)
In order to determine R and M in such an experiment, it is usual to consider the frequency
correction due to the recoil of the emitting atom. [Thermal motion causes emission lines to be
broadened without displacing emission maxima, and we may therefore assume that all thermal
effects have been taken into account.]
(i) (4 marks)
Assume that the atom decays at rest, producing a photon and a recoiling atom. Obtain the
relativistic expression for the energy hf of a photon emitted in terms of ΔE (the difference in
rest energy between the two atomic levels) and the initial rest mass m0 of the atom.
(ii) (1 mark) � �
Δf
Hence make a numerical estimate of the relativistic frequency shift for the case of
f recoil
He+ ions.
Your answer should turn out to be much smaller than the gravitational red shift obtained in
part (b).

Data:
Velocity of light c = 3.0 × 108 ms−1
Rest energy of He m 0 c2 = 4 × 938(MeV)
13.6Z 2
Bohr energy En = − (eV)
n2
Gravitational constant G = 6.7 × 10−11 Nm2 kg−2
Problem No 9 | IPhO Estonia 2012 http://www.ipho2012.ee/physicscup/problem-no-9/

Home
First Circular
Second Circular
Third Circular
Registration
Problem No 9 Program
Leaders and observers
Electron, initially at rest, is accelerated with a voltage , where is the electron's rest mass, Students
– the elementary charge, – the speed of light, and – a dimensionless number. The electron hits a
Opening Ceremony
motionless positron and annihilates creating two photons. The direction of one emitted photon defines the
direction of the other one. Find the smallest possible value of the angle between the directions of the Lecture: Sir Harold Kroto
two emitted photons (express it in terms of and provide a numrical value for ). Closing Ceremony
Problems
Solutions
Experimental apparatus

Results
Gold medalists
Silver medalists
Bronze medalists
Honorable Mentioned
Special prizes
Statistics

Steering Committee
Academic Committee
Organizing Committee

IPhO Homepage
History
Statutes
Syllabus

Newsletter
Press Releases
Invitation movie
Short version
Long version

Sponsors
Travel and accommodation
Estonia
Tallinn
Tartu

IPhO 2012 Eestis

Feedback
Competition “Physics Cup –
IPhO2012”
Formula sheet
Frequently asked questions
Physics solver’s mosaic
1. Minimum or maximum?
2. Fast or slow?
3. Force diagrams or
generalized coordinates?
4. Are Trojans stable?
5. Images or roulette?
Problem 0
Solution
Problem No 1
Results after Problem 1
Solution
Problem No 10
Intermediate conclusion
Solution
Problem No 2
Results after Problem 2
Solution
Problem No 3
Results after Problem 3
Solution

1 of 2 01/24/2013 10:53 PM
Quantum Mechanics There are at least two good reasons for this. Firstly, any state of a
quantum-mechanical system is representable as a superposition of station-
Original Estonian text by Jaan Kalda, translated and edited by
ary states (exactly in the same way as the random motion of connected
Stanislav Zavjalov and dated by January 7, 2013. Please send
oscillators is representable as a superposition of normal modes). Secondly,
suggestions and comments to stanleon@gmail.com.
generally a quantum system that has been brought of the state of equilib-
Introduction rium will quickly find itself in a state with the lowest possible energy such

When talking about quantum mechanics, it is often said that that no conservation laws are violated. After all, the lowest energy is also

matter has both wave-like and particle-like properties. Above a well-defined, unique energy.

all, that approach is an attempt to explain the non-intuitive
aspect of the subject with familiar notions. In fact, a quantum-
mechanical particle is an object that is a curious combination
of these two aspects — an object the exact likes of which we
do not see in the familiar macroscopic environment and which
is therefore inherently non-intuitive. Still, quantum mechanical
particles are most successfully treated like waves, and it turns
out that some seemingly particle-like properties could still be
explained using wave considerations. Let us dive in! The text in
small print could be omitted on the first reading, the majority
of IPhO problems could be solved without knowing these facts.
Wavefunction
In classical mechanics, the state of a particle without internal
degrees of freedom1 can be completely described by its momen-
tum and co-ordinate; in quantum mechanics, the complete
state of a particle is described by its so-called wavefunction
Ψ, which is generally a complex number and which could be
represented, for example, as a function of co-ordinates and time,
like so: Ψ = Ψ(x, y, z, t). The probability to find the particle at
time t at position with co-ordinates x, y and z is then propor-
tional 2 to the modulus-squared of the complex wavefunction,
|Ψ|2 = Ψ(x, y, z, t)Ψ∗ (x, y, z, t), where Ψ∗ is the complex con-
jugate of Ψ. What is actually meant by ”finding” the particle
at a given point of space is actually a tricky question, we will
return to it later. In short, this is a short-hand for saying that
a measurement has been conducted to find the position of the
particle, and the result of that measurement was (x, y, z).
The example of the photon could be helpful to understand
the idea of the wavefunction: the real part of the photon’s
wavefunction is the electric field vector and the imaginary part
is the magnetic field vector (this statement is true in the Gauss
system of units; in SI, the wavefunction’s imaginary part is the
product of the magnetic induction and the speed of light). The
probability to find a photon at any point is proportional to the
electromagnetic energy density (light intensity) at that point.
Comparing the electron and the photon we should note that for pho-
tons one uses a vectorial wavefunction, for electrons — a scalar one. The
difference arises because the angular moment of a photon about an axis
through its ”centre of mass” can have three different values (which cor-
respond to linear and two circular — clockwise and counter-clockwise —
polarisations). We will not discuss this nuance further.
Energy and momentum in Quantum Mechanics
In quantum mechanics, the so-called stationary states or eigen-
states of energy play an important role. In such a state, the
particle has a well-defined single value of energy E (the so-called
eigenvalue).
1 Suchan internal degree of freedom could be, for example, rotation about its own axis.
2 Innon-relativistic quantum mechanics the particles are never created nor destroyed, and the probability to find the particle in all space is therefore
unity. Hence the integral of |Ψ|2 over all space must also be equal to unity. Performing the integral fixes the constant in front of the wavefunction —
we are normally referring to just that constant when we say, in the following, that the wavefunction is proportional to something.
It turns out (the reason being the Schroedinger equation,
which could be considered as a postulate of quantum mechanics),
that in a state with a well-defined energy E the wavefunction
evolves as follows: Ψ(x, y, z, t) = ψ(x, y, z)e−iEt/~ , so that the
wavefunction oscillates with angular frequency ω, which is re-
lated to the energy by the important formula
E = ~ω = hν. (1)
The linear frequency ν = ω/2π; the quantity h is known as
Planck’s constant, while ~ = h/2π is better known as the
reduced Planck’s constant.
Note that the energy in (1) is the total (kinetic plus poten-
tial) energy of a quantum-mechanical system (e.g. of a particle),
while the zero-level of potential energy could be freely chosen.
Note how a change of the zero level of the potential energy by
a value U means that the wavefunction gets multiplied by an
extra factor eiU t/~ (this statement also holds for non-stationary
states), which does not at all alter the physically tangible quan-
tity — the probability of finding the particle — because the
modulus of this factor is still just unity. There are some impor-
tant practical lessons to be learned from this:
(a) For motion and collisions of non-relativistic electrons and
neutrons, one can use the expression for non-relativistic kinetic
energy, p2 /2m, as the particles don’t vanish and it makes sense
to measure the kinetic energy with respect to the motionless
state (in other words, the rest mass energy can be ignored);
(b) For photon absorption and emission, it is natural to use the
relativistic energy expression E = mc2 ;
(c) The zero level of potential energy can be chosen arbitrarily.
In quantum mechanics, energy conservation law could be
regarded as the condition for resonance. Suppose there is a
system with two energy levels E1 < E2 . Suppose the system
transfers from the low-lying energy state to the other state by
absorbing a particle (a photon for concreteness) with energy E3 .
Energy conservation law then reads E2 = E1 + E3 , which could
be re-written using angular frequencies in the form ω2 −ω1 = ω3 .
On the left hand side of the equation is the angular frequency of
the system’s wavefunction, if we chose the zero level of energy
to be at E = E1 , while on the other side of the equation is the
frequency of the photon. Therefore the interaction (absorption
of the photon) can only happen if the condition for resonance is
fulfilled: the electromagnetic frequency is equal to the difference
in frequencies of the two stationary states.
Consider a quantum-mechanical simple harmonic oscillator:
a particle that moves in parabolic potential (potential energy
is proportional to the square of the displacement from equilib-
rium). Suppose the classical angular frequency of this particle is
ω0 . This oscillator can go from one quantum-mechanical energy

— page 1 —
level (Ei ) to another (Ej , corresponding to higher amplitude os-
cillations) by absorbing (n = 1, 2, 3 . . .) photons. These photons
are electromagnetic waves, which have to be in resonance with
the oscillator, so that the electromagnetic angular frequency
has to be equal to the eigenfrequency of the oscillator ω0 . From
the energy conservation law Ej − Ei = ~nω0 we deduce that
the energy levels of the simple harmonic oscillator have to be
of the form
En = ~nω0 + C.
A rigorous solution of Schroedinger’s equation shows that for the zero
level of potential energy at the minimum of the potential, the constant
C = ~ω0 /2.
This result can easily be generalised unto oscillators with m degrees of
freedom. From classical mechanics we know that in this case, the system
has m eigenfrequencies ωj , j = 1, 2, . . . m. j-th eigenfrequency can be
excited if the resonance condition is satisfied, so that the electromagnetic
frequency is ωj . Therefore one has to employ m different integers nj to
describe the stationary energy levels of such an oscillator:
E= m
P
j=1 ~ωj nj + C.
This energy expression looks just as if we had different particles — of
energies ωj , j = 1, 2, . . . m — and the integers nj describe the numbers
of these particles. In case of elastic oscillations (standing waves) of the
crystal lattice these so-called quasi-particles (so not truly real, physical
particles) are called phonons.
We have seen that the energy of the oscillator is quantized,
it can only assume certain discrete values.
As we have seen, in a stationary state the time-dependence
of the wavefunction is known, and therefore the interesting
part is its spatial dependence ψ(x, y, z). It turns out (from
Schroedinger’s Equation) that in a state with uniquely defined
momentum p~, the wavefunction is a sinusoidal plane wave,
~
ψ = eik~r , where the wave vector ~k is related to momentum
through the formula
p~ = ~~k (2)
Wave vector points along the direction of the wave’s propagation,
and its modulus is k = 2π/λ, where λ is the wavelength. The
function ψ = ei~p~r/~ is called an eigenfunction of momentum.
Note that for a particle, the group velocity of the wavefunction consid-
ered as a wave is equal to the particle’s classical velocity: ~ω = p2 /2m =
(~k)2 /2m+U . Upon differentiating, this gives vg = dω
= ~k/m = p/m = v.
dk
dω
This also holds for photons: ~ω = mc2 = pc = ~kc, and vg = dk
= c.
Wavefunction as a function of momentum
As seen before, the wavefunction of a particle with well-defined
momentum is a sinusoidal plane wave that fills the whole space,
so that the position of the particle is entirely unknown.
The square of the modulus of the wavefunction |ψ(~r)|2 ≡
|ψ(x, y, z)|2 gives the probability to find the particle at a given
point. This statement should be considered as a postulate given
on empirical (experimental) basis. The question of what is
actually meant by “finding the particle” will be addressed later.
For the moment, it is important to note that the wavefunction
can not only be given as a function of the co-ordinate ~r, but also
as that of the momentum p~ ≡ (px , py , pz ), or ψ = ψ(~ p) ≡ ψp~ .
Here, we are dealing with a mathematical trick (called “Fourier
analysis”), where an arbitrary wavefunction can be represented
as the sum of various states of well-defined momentum, ψp~ ei~p~r/~ .
In this representation, the square of the absolute value of each
component’s amplitude |ψp~ |2 gives the probability to find the
particle with the corresponding momentum p~ (generally, the
— page 2 —
amplitude ψp~ is complex number whose argument gives the
phase-shift of the corresponding component wave).
Let us approach the issue mathematically. Fourier analysis tells
us, that any function f (x) couldR ∞ be ikxrepresented as the sum of sinu-
soidal functions, f (x) = √1 −∞ fk e dk, where the quantity fk =
2π
√1
R∞
f (x)e−ikx dx, which depends on the wave vector k, is called a
2π −∞
Fourier component of the function f (x). Changing the variables k → p/~,
we can write the given integral for the wavefunction in the form
Z ∞
ψ(x) = √
1
ψp eipx/~ dp, (3)
2π~ −∞
where the factor Z ∞
1
ψp = √ ψ(x)e−ipx/~ dx. (4)
2π~ −∞
Formulae (3,4) are not for learning by heart at this point, but to supply
evidence to this statement: each wavefunction can be represented as a su-
perposition of states (wavefunctions ψ = eipx/~ ) such that the momentum
has a well-defined unique value p; the amplitudes ψp of those states can
be regarded as providing the dependence of the given wavefunction on
momentum.
To make a further comparison, consider the state where the particle has
a well-defined unique co-ordinate x0 ; then its wavefunction is ψ = δ(x−x0 ).
The function δ(x) is called the Dirac delta-function (the area under the
graph of this function is 1 and it only has a non-zero value at point x = 0).
Now we can express an arbitrary wavefunction ψ(x) as a superposition
where every component has a unique co-ordinate:
Z
ψ(x) = ψ(x0 )δ(x − x0 )dx0 .
Note that on the right hand side of the equality sign, the prefactor in
front of the state of well-defined position δ(x − x0 ) is the initial function
ψ(x) at position x = x0 , whose modulus-squared is the probability to find
the particle at x0 . Therefore it should not be surprising that in (3), the
modulus-squared of the prefactor in front of eipx/~ (which is |ψp |2 ) gives
the probability to find the particle with momentum p.
The transfer from the function ψ(x) to another ψp could be regarded
as a spatial rotation in the functional (Hilbert) space (where the function
ψ is a vector), because just as in the ordinary spatial rotation formulae,
the new co-ordinates of the vector ψp are represented through the linear
combination of the old co-ordinates ψ(x). The only difference is that as the
number of co-ordinates in ordinary spaces is finite, the linear combination
is written with a sum; in the Hilbert space, however, the set of co-ordinates
is uncountable (is a continuum), and therefore the sum is replaced by the
integral. While in the ordinary space a co-ordinate-independent vector is
signified with a small arrow upon the symbol, in the Hilbert space the
notation |ψi is normally used. Therefore |ψi represents the wavefunction
for which it is undecided or unimportant whether we should investigate
its dependence based on the co-ordinate x, momentum p or a different
physical quantity altogether.
The formulae above assume a one-dimensional motion, which can be
described with one co-ordinate x and momentum p. In three dimensions,
a single integral is replaced by triple integration (over x, y and z or over
px , py and pz ), the product px by the scalar product p ~·~r and the factor
√1 by a factor (2π~)−3/2 .
2π~
The uncertainty principle
The uncertainty principle is a purely mathematical result, which
comes from the connection between the two representations of
the wavefunction: in the co-ordinate and momentum space
(formulae 3,4). It is possible to prove that if one defines the
co-ordinate and momentum uncertainties δx and δp as root-
mean-square deviations from the corresponding average values
x̄ and p̄, then the following inequality holds:
δx · δp ≥ ~/2. (5)
Mathematically, one can write these definitions as
R∞
x̄ ≡ −∞
R∞
|ψ(x)|2 xdx, p̄ ≡ −∞ |ψp |2 pdp,
qR
∞
δx ≡ 2 2
−∞ |ψ(x)| (x − x̄) dx,
qR
∞
δp ≡ 2 2
−∞ |ψp | (p − p̄) dp.
In inequality (5) equality is observed if the wavefunction is a
2 2
Gaussian, ψ(x) = π 1/4 (δx)−1/2 e−x /4(δx) ; in all other cases, it
is a strict inequality.
 The uncertainty principle is often used in quantum- The second figure represents the probability distribution
mechanical estimations, where one replaces the inequality sign |ψ(x)|2 = (1 − x2 )2 for |x| ≤ 1 and |ψ(x)|2 = 0 for |x| ≥ 1
with an approximate equality and the root-mean-square devi- of finding the particle. In this case the wavefunction ψ(x) is
ations with characteristic widths ∆x and ∆p. In which case continuous and therefore the probability distribution in momen-
would we get a more precise estimate, is it with the expression tum space is localised more strongly than in the case of the slit,
∆x · ∆p ∼ h, (6)and the root-mean-square deviation of momentum is, in fact,
∆x · ∆p ∼ ~ or ∆x · ∆p ∼ ~/2? The answers could differ by up finite: δp ≈ 1,34.
√ By computing the integral it is not hard to
to 4π ≈ 13 times and an error this big should be avoided even find δx = 1/ 7, so that δx · δp ≈ 0,507~ (and the solution is
in estimates (even though when estimating, one cannot discard close to the absolute minimum ~/2). Calculations show that
as wrong even those answers that differ from the true answer the widths of the two peaks at half-maximum are ∆p ≈ 3,6~
by a factor of, say, (2π)3 ≈ 248, by carefully thinking through and ∆x ≈ 1,1. Thus ∆x · ∆p ≈ 4~ ≈ 0,63h.
one’s actions it is normally possible to achieve the difference by The two given examples show that the estimate (6) is fit-
less than a factor of two). Let us investigate this question on ting when one takes the characteristic widths of the interval
two concrete example wavefunctions, see the figures. as the uncertainties ∆p and ∆x; the estimate (5) is valid for
|ψ| 2 |ψp|2 the root-mean-square deviations (unless the localisation is weak
and the root-mean-square deviation is actually infinite).
δx The uncertainty principle for the pair px and x holds because
the wavefunction for a state of definite px is a sinusoidal func-
-1 1 x tion of x (and any state can be represented as a superposition
of sinusoids for different px ). Obviously the same thing holds
for py and y as well as for pz and z, but if we remember formula
(1), then it also holds for E and t (and, in fact, also for angular
momentum and the rotation angle) — the variables forming
such pairs are called conjugate variables. The connection be-
tween E and t is typically useful for describing for the excited
state of an atom or a molecule: if theoretically the value of
the energy in a stationary state is En , then due to instabilities,
-2π -π 0 π 2π p/h collisions etc. the atom or molecule cannot reside in that state
The first figure describes an equal probability for the particle for longer than some characteristic time τ . Thus, its energy
to be found along the x axis if its co-ordinate is in the range in the excited state will not be exactly En , but can be in the
−1 ≤ x ≤ 1, while outside of this range the particle cannot range of width Γ around the central value En , where
possibly be found. This is exactly the type of distribution that Γ · τ ∼ h. (7)
arises when an electron, photon etc. passes through a single The quantity Γ is called the width of the energy level and τ is
slit; in that case the probability distribution as a function of known as the excited level’s lifetime.
momentum, |ψp |2 , is exactly the same as the intensity distribu-
tion in the diffraction pattern (on the screen). We will normally Quasi-classical approximation
consider the typical width of the diffraction maximum to be If one wishes to obtain the precise forms of the stationary state
of the order of ∆p = π~ (it is approximately the width of the wavefunctions and corresponding energy levels, one will gener-
peak at half-maximum). The probability |ψp |2 oscillates and ally have to solve Schroedinger’s equation. However, one can
the oscillation amplitude is inversely proportional to the square advance quite far (and sometimes even get the correct result)
of momentum (see fig.); therefore the root-mean-square devia- by employing the so-called quasi-classical approximation (also
tion is not finite, δp = ∞. The uncertainty√ in the position is, known as the WKB(J) approximation in the English-language
according to the integral above, δx = 1/ 3, which is about 3,5 literature). In this approach, one first considers a particle mov-
times smaller than the width of the slit ∆ = 2. Thus, we find ing in a known potential as a classical particle and finds the
δp · δx = ∞ and ∆p · ∆x = 2π~ ≡ h. dependence of its momentum p on co-ordinate x. Then one
|ψp|2 |ψ| 2 says that the particle’s wavefunction is almost a sinusoid with
a variable wave vector: ψ(x) ≈ exp[ ~i p(x)dx]. This is, of
R

course, an approximate relationship — the equality only holds

if the particle is free (the momentum is constant). Still, the
equality almost holds if the relative change of momentum in one
∆x wavelength is small [so that |p(x) − p(x + ~/p)|  p(x)]. Sup-
δx pose this variable-wavelength wave propagates along the x-axis
x or along a circular trajectory (e.g. in the case of an electron
∆p -1 0 1 orbiting around an atom). If the particle moves in a potential
well so deep that its walls keep reflecting it back, then the wave
δp is also reflected and starts bouncing (oscillating) back and forth.
The wave pattern is stable if there is an integer number of
p/h wavelengths in one oscillation period (then a stationary wave is
-5 0 5 10 formed). For example, if we are dealing with a potential well
— page 3 —
with vertical walls and flat horisontal bottom of length L, then
the momentum inside the well is constant and the stationary
wave condition gives 2Lp = nh, which means that p = nh/2L
and the n-th energy level is En = p2 /2m = n2 h2 /8L2 m.
In a hydrogen atom, an electron of mass m orbits around
the nucleus in a symmetric potential well U = −kZe2 /r (where
1 important lesson to be learnt is that every measurement changes
k = 4πε 0
). The kinetic energy of an electron moving in a circu-
lar orbit is p2 /2m = −U/2, therefore p2 r = kmZe2 . We write
the stationary wave condition as 2πrp = 2π~n (where n is the
number of stationary waves), and then rp = n~.
This last relation shows that the angular momentum rp is quantized.
It turns out that this conclusion is more general and is not only charac-
teristic of circular orbits: the angular momentum with respect to a fixed
axis can only be an integer number of ~ (an exception to this rule are
the so-called Fermi particles or fermions (for example, the electron): the
internal angular momentum — so-called spin — of fermions can only be a
half-integer number of ~. In the case of electrons, it is always ±~/2).
Combining these two last results, we find p = kmZe2 /n~
and therefore the total energy
m probability of two independent events is the product of two individual prob-
En = U/2 = −p2 /2m = − (kZe2 /n~)2 .
2 possible: one has to solve the seven-dimensional Schroedinger equation.
Quantum tunnelling
Quantum tunnelling is a striking non-classical effect, where
a particle of energy E is capable of passing through a po-
tential energy barrier U > E in a way, where beneath the
barrier it acquires a negative kinetic energy p2 /2m = E − U .
A negative kinetic energy p means an imaginary momentum
and wave vector k = ~i 2m(U − E), and the wavefunction
is no longer√a sinusoid, but a decaying exponential function:
ψ(x) = e−xp 2m(U −E)/~ . Thus, if the width of the barrier L
is large [L 2m(U − E)  ~], then the particle’s wavefunc-
tion drops exponentially below the barrier, and the particle
cannot
p pass through (is reflected). On the other hand, if
L 2m(U − E) ∼ ~, then the wavefunction still decays (the
particle is reflected back with a certain probability), but not
extremely so: there is a non-negligible probability of tunnelling.
A neat approach relates quantum tunnelling to the uncertainty prin-
ciple for energy: the condition for tunnelling could be rewritten in the
form 2(U − E) · √ L
∼ ~ and could be interpreted as follows: the
2(U −E)/m
p
particle “borrows”, for a short time τ = L/v [where v = 2(U − E)/m]
the energy 2(U − E), out of which U − E is spent to overcome the barrier
and a further U − E is left as kinetic energy to pass the barrier; the energy
could not be borrowed for any longer than the uncertainty principle allows.
Measurement in quantum mechanics
The classical or so-called Copenhagen interpretation of quantum
mechanics states that as a result of a quantum-mechanical mea-
surement, the wavefunction collapses. Suppose an experiment
to measure the particle’s momentum is performed. Before the
measurement, the wavefunction of the particle was R a superpo-
sition of states of well-defined momentum, ψ = ψp eipx/~ dp.
Upon the measurement, a certain result pm is obtained: thus,
from the state ψp eipx/~ dp a state ψpm eipm x/~ has been ob-
R
tained. A repeated measurement of the particle’s momentum
from now on will always find the same value pm of the particle’s
momentum, unless the particle is affected by something else.
This change of state from a superposition of eigenfunctions
of the measured quantity — which also included the eventual
result among many other contributions — into the state cor-
responding to the eigenfunction of only the measured result
— page 4 —
is exactly what is known as the collapse of the wavefunction.
Recall that an eigenfunction of a physical quantity is defined as
a wavefunction that the particle must have in order to have this
physical quantity well-defined. The probability that any of these
possible results is found is proportional to the modulus-squared
of the corresponding amplitude in the initial superposition. An
the state of the system.
This interpretation works perfectly well in practice, yet can be un-
satisfying from the philosophical point of view. The issue is that the
interaction of a particle with a macroscopic measurement device is treated
differently from inter-particle interactions. In this last case, no collapse
happens. For concreteness, let us look at the interaction of two particles,
e.g. electrons. If one particle is described by the wavefunction that is (in
the case of spatial representation) a function of three spatial co-ordinates
and time, ψ(~ r, t), then the two-particle wavefunction is already a func-
tion of six co-ordinates: Ψ = Ψ(~ r1 , ~
r2 , t), where ~
r1 and ~
r2 are the two
particle’s respective position vectors. If the particles do not interact, this
seven-variable function becomes a product of two four-variable functions,
Ψ(~r1 , ~
r2 , t) = ψ1 (~r1 , t)ψ2 (~
r2 , t) (much in the same way as the combined
abilities); in case of interactions such separation of variables is no longer
Interactions of more than two particles are treated analogously, and for n
particles one deals with a wavefunction that depends on 3n + 1 variables.
And there is not even a mention of the collapse of the wavefunction! So
how does this collapse come into play? Classical interpretation leaves
this question unanswered. The tumultuous interplay between the macro-
scopic and microscopic worlds gives birth to various paradoxes, such as
the Schroedinger’s cat and the teleportation (Einstein-Podolsky-Rosen)
paradox.
An alternative interpretation of quantum mechanics, the so-called
many-worlds interpretation due to Hugh Everett and others, is capable
of justifying both the collapse of the wavefunction and the mentioned
paradoxes. The practical implications of this interpretation are mostly the
same as in the classical interpretation. An additional advantage, however,
is that the modulus-squared of the wavefunction is identified with the
probability not by the means of a postulate, but via a mathematical proof.
In this interpretation, the physical reality is postulated to correspond to
the Universe’s wavefunction Ψ, which is a function of co-ordinates of all the
particles in the Universe (including those comprising the living beings); it
evolves all the time according to the Schroedinger equation and experiences
no collapses whatsoever. Thus, Ψ includes all the possible measurement
results for any conceivable measurement. For simplicity, let us consider the
wavefunction of the combined system “observer (experimenter)+particle”
and let us suppose that there are only two possible outcomes to this mea-
surement, ‘1’ and ‘2’. Before and after the measurement, the observer and
the particle do not interact, and so one can separate the variables and
represent the system’s initial state as the product |M0 i |O0 i, where |M0 i
is the initial state of the observer, while the initial state of the particle
is the superposition of two possible states, |O0 i = α |O1 i + β |O2 i; here
|O1 i and |O2 i are the particles’ states in which the measured quantity
is well-defined. After the measurement, the state of system assumes the
form α |M1 i |O1 i + β |M2 i |O2 i, where |Mj i is the state of the observer
according to whom the result of the measurement is ‘j’. But to ensure
that a concrete observer have one and only one opinion of the result of
the measurement, we say that the worlds have split, or that the Universe
has branched into two: in one branch, there is an observer with the state
|M1 i, in the other — with the state |M2 i. In which Universe each concrete
individual (including yourself) will end up is purely arbitrary; in either
Universe it seems that a wavefunction collapse has occured (so that |M1 i
sees the particle in the pure state |O1 i).
The branching of the Universe does not only happen during a mea-
surement, but rather at any time when an energy dissipation (irreversible
process) occurs. Due to irreversibility, the disconnected worlds cannot
reconnect again. One can say that all the quantum mechanical probabil-
ities come to life in different worlds, and us here are only witnesses to
one possible realisation. There may be a world where the dinosaurs never
perished.