Вы находитесь на странице: 1из 143

A First Course in String Theory.

Solutions for problems in Part I.

The following pages contain the solutions for all the problems to be found in Part I of the
textbook A First Course in String Theory. The handwritten solutions are all due to Jeffrey
Goldstone. They are clear and elegant, and I hope that he will publish them someday. If you
distribute some solutions to students please let them know that they are due to Goldstone.
The rest of the solutions have been typeset. They were mostly written by me, with some
help by Stefanos Marnerides and Ian Ellwood.

Barton Zwiebach
MIT
Cambridge, MA
June 2004
Quick Calculation 2.3.
Since we are leaving the coordinates x2 and x3 invariant, we only consider boosts along
the x1 axis, for which
x0 = γ(x0 − βx1 ) .
To get
1
x0 = x+ = √ (x0 + x1 ),
2
we would need
1
γ = √ , and β = −1 ,
2

which is not possible given that γ = 1/ 1 − β 2 .
More conceptually, under boosts −(x0 )2 + (x1 )2 = −(x0 )2 + (x1 )2 = −(x+ )2 + (x− )2 .
The last equality would be needed if there was a boost that set (x0 , x1 ) = (x+ , x− ).

1
2.1. Exercises with units
(a) By definition, two charges of one esu each placed a distance 1cm from each other, repel
with a force of magnitude 1 dyne. Let q denote the charge in Coulombs that is equivalent
to one esu. We use the equations

1 |q1 q2 | 1 N m2
|F | = with = 8.99 × 109 2
4π0 r2 4π0 C
and since 1 dyne is equivalent to 10−5 Newtons,

N · m2 q2
10−5 N = 8.99 × 109 2 → q = 3.334 × 10−10 C.
C (10−2 m)2

From this we conclude that 1 esu is equivalent to 3.34×10−10 Coulombs, or alternatively, 1


Coulomb is equivalent to 2.998×109 esu.

(b) The fundamental temperature τ is defined from τ1 = ∂S/∂E where S is entropy (unitless)
and E is energy. So, the fundamental temperature has units of energy. The Kelvin temper-
ature T arises from τ = kB T , so [kB T ] = [E]. The constant kB can be measured once we fix
the Kelvin scale. This is done by specifying arbitrarily the value for the temperature Ttp of
the triple point of water: Ttp ≡ 273.16K. Then kB can be measured, say from P V = N kB T ,
and from measurements on a dilute gas kB  1.38 × 10−16 erg/K. Just as Coulombs always
appear with 0 , so that Coulombs always cancel in formulae, Kelvin temperature T always
appears with kB , and K cancels.

(c) Recall that [e] = M 1/2 L3/2 T −1 , [~] = M L2 T −1 , and [c] = LT −1 . Therefore [~c] =
M L3 T −2 = [e2 ] and e2 /~c is dimensionless. With e = 4.802×10−10 esu, ~ = 1.504×10−27 erg·s,
and c = 2.997 × 1010 cm/s, we find

e2 1
= 7.297 × 10−3 ≈ .
~c 137
This is the fine structure constant.

1
2.2. Lorentz Transformation for light cone coordinates.
(a) The nontrivial Lorentz tranformations for this boost are:

x0 = γ(x0 − βx1 )


x1 = γ(x1 − βx0 ) .

We then find
x0 ± x1 γ
x± = √ = √ (x0 − βx1 ± x1 ∓ βx0 )
2 2
γ(1 ∓ β) 0
= √ (x ± x1 ) = γ(1 ∓ β)x± ,
2
Therefore,
 
1−β + 1+β −
x+ = x , x− = x , x2 = x2 , x3 = x3 . (1)
1+β 1−β

The light-cone coordinates x+ and x− do not mix under boosts along x1 !

(b)

x2
/
x2

/
x1

e
x1

Figure 1: A rotation of coordinates.

The new, rotated coordinates are given by

x0 = x0
x1 = cos θ x1 + sin θ x2
x2 = − sin θ x1 + cos θ x2
x3 = x3 .

1
After some algebra, we find
1 1 sin θ 2
x+ = (1 + cos θ) x+ + (1 − cos θ)x− + √ x ,
2 2 2
1 1 sin θ 2
x− = (1 − cos θ)x+ + (1 + cos θ) x− − √ x ,
2 2 2
sin θ sin θ
x2 = − √ x+ + √ x− + cos θ x2 ,
2 2
x3 = x3 .

(c) For a boost with velocity parameter β along x3 , the Lorentz transformations are:

x0 = γ(x0 − βx3 )


x1 = x1
x2 = x2
x3 = γ(x3 − βx0 ) .

Therefore
x0 + x1 γ(x0 − βx3 ) + x1
x+ = √ = √
2 2
γ + γβx3 1
= (x + x− ) − √ + (x+ − x− ) .
2 2 2
Working similarly for x− , x2 , and x3 we obtain
1 1 γβ
x+ = (γ + 1)x+ + (γ − 1) x− − √ x3 ,
2 2 2
1 1 γβ
x− = (γ − 1)x+ + (γ + 1)x− − √ x3 ,
2 2 2
x2 = x2 ,
γβ
x3 = γx3 − √ (x+ + x− ) .
2

2
2.5. Spacetime diagrams and Lorentz transformations.
Consider the boost
x0 = γ(x0 − βx1 )
x1 = γ(x1 − βx0 ) .
The x0 axis is the set of points with x1 = 0, so it corresponds to the line x1 = βx0 . In other
words, it is a line that goes through the origin and the point x0 = 1, x1 = β. This point
gives x0 = γ(1 − β 2 ) > 0.
The x1 axis is the set of points with x0 = 0, so it corresponds to the line x0 = βx1 . In other
words, it is a line that goes through the origin and the point x0 = β, x1 = 1. This point
gives x1 = γ(1 − β 2 ) > 0.
In the figure we show the old and new axes, for β > 0 (left-side) and for β < 0 (right-
side). In both cases the angle φ between the axes (x0 and x0 , or x1 and x1 ) has magnitude
tan |φ| = |β|. Note that |φ| ≤ π/4. The direction of the arrows indicates increasing value for
the coordinates.

Figure 1: Oblique axes for boosts.

1
2.6. Light-like compactification
     
x x R
∼ + 2π (1)
ct ct −R

(a) We use the above to compute


ct + x 1
x+ = √ ∼ √ (ct − 2πR + x + 2πR) = x+
2 2
ct − x 1 √
x− = √ ∼ √ (ct − 2πR − x − 2πR) = x− − 2π( 2R) .
2 2
We therefore have the light-cone identifications

x+ ∼ x+ , x− ∼ x− − 2π( 2R) .

The compactification does not involve the light-cone time x+ .

(b) With the Lorentz transformations x = γ(x − βct) and ct = γ(ct − βx), we find

x ∼ γ(x + 2πR − β[ct − 2πR]) = x + 2πRγ(1 + β)


ct ∼ γ(ct − 2πR − β[x + 2πR]) = ct − 2πRγ(1 + β)

thus
      
x x 1+β R
∼ + 2π . (2)
ct ct 1 − β −R
Comparing with (1) we learn that in the boosted frame the effective radius of compactification
is increased.
Now consider the compactification
     2 
x x R + Rs2
∼ + 2π .
ct ct −R

(c) We search for a frame (x , ct )

x = γ(x − βct)
ct = γ(ct − βx)

in which this compactification is standard. For this we need ct ∼ ct . So,

ct ∼ γ(ct − 2πR − β [ x + 2π R2 + Rs2 ] )

= ct − γ(2π)(R + β R2 + Rs2 )

1

Thus, we need R + β R2 + Rs2 = 0, so
R
β=− . (3)
R + Rs2
2

Now examine the x identification to find the radius:



x ∼ γ(x + 2π R2 + Rs2 − β(ct − 2πR))

x ∼ x + 2πγ( R2 + Rs2 + βR) .
Simplify,
 1  R2
γ( R2 + Rs2 + βR) =  ( R2 + Rs2 −  )
1 − β2 R2 + Rs2

R2 + Rs2 R2 + Rs2 − R2
= (  ) = Rs .
Rs R2 + Rs2
 
x ∼ x + 2πRs . The radius of compactification is Rs . It is interesting to note that
Thus,
γ = R2 + Rs2 /Rs . In the limit as Rs is small, this is a very large Lorentz factor γ  R/Rs .
(d) The required spacetime diagram is shown in Figure 1.

/
ct
ct

2 2
2/ R+Rs x

- 2/R /
x

Figure 1: The heavy dots are identified so that the x axis must go through them. The world line
of the origin of the primed system is the ct line. This origin is moving in the negative x-direction,
as the sign of β indicates.

(e) Light like compactification with radius R arises by boosting a standard compactification
with radius Rs with Lorentz factor γ  R/Rs , in the limit as Rs → 0.

2
2.7. A spacetime orbifold in two-dimensions.

(a) From Problem 2.2 we know that under a boost with velocity parameter β along the x1
direction, x+ and x− transform as
 
1 − β 1+β −
x+ → x+ and x− → x .
1+β 1−β

Since these positive scale factors are inverses of each other we may define

1+β
eλ ≡ , −∞ ≤ λ ≤ ∞ .
1−β

λ is called the rapidity. Given two boosts with rapidities λ1 and λ2 , the rapidity of the
combined transformation is clearly λ1 + λ1 . The range −∞ ≤ λ ≤ ∞ arises because β ∈
(−1, 1), so eλ ∈ (0, ∞). We can write the identification as

(x+ , x− ) ∼ (e−λ x+ , eλ x− ) .

For arbitrary λ the orbifold fixed point is the origin x+ = x− = 0.

(b) Below we plot curves of constant a2 for two values of a2 . Note that there are upper and
lower branches.
larger a 2

x
ï x0 +
x

line of constant a 2

1
x
larger a 2

Note that x+ x− is invariant under boosts since x+ and x− get multiplied by e−λ and eλ ,
respectively. Since the signs of x+ and x− are not changed under boosts, the identification
equates points on the same x+ x− = a2 curve.

1
(c) We have x+ = a2 /x− so that dx+ = −( xa− )2 dx+ . This yields
 2
a
−ds = −2dx dx = 2 − (dx− )2 > 0 .
2 + −
x

Thus ds2 < 0 and the curve is spacelike.



(d) From part (c) we have that −ds2 = 2a2 ( dx
x−
)2 so the appropriate root is
√ dx−
2a .
x−
To find the invariant length of the circle we must integrate the above differential from a
given x− λ −
∗ to its immediate image e x∗ . This gives an invariant length

 eλ x−
eλ x−∗
√ ∗
dx− √ 
−
√ √
2a −
= 2 a log x  = 2 a log(eλ ) = 2 aλ .
x−

x x−

2
2.8. Extra dimensions and statistical mechanics.
(a) The states of the system are labeled by k and l and the partition function is given by
   2   2  2

~ kπ l
Z(a, R) = exp −β + .
k,l
2m a R

Here k = 1, . . . , ∞, while l = 0, 1, . . . , ∞. Due to the degeneracy of the energy levels we


must include an extra factor of 2 for each term in the l sum for l > 0. Defining
 ∞   2  2
~ kπ
Z(r) ≡ exp −β ,
k=1
2m r

we have that 

Z(a, R) = Z(a)
1 + 2Z(πR) . (1)

We need to evaluate Z(r) for small β. In this case, the terms in the sum vary very slowly
with k and many of them are nearly equal to one making the total sum very large. It is thus
possible to replace the sum with an integral that runs from zero to infinity:
 ∞   2 
β ~2
kπ mr2
Z(r)  dk exp − = 1, β → 0. (2)
0 2m r 2πβ ~2

For sufficiently small β, (2) applies both for r = a and r = 2πR. So (1) gives:
   
ma 2 m(πR) 2 ma 2 m(2πR)2

Z(a, R)  Z(a)
· 2Z(πR)  ×2 = .
2πβ ~ 2 2πβ ~2 2πβ ~ 2 2πβ ~2
This partition function is just the product of two familiar factors, one using the length a and
the other using 2πR. For high temperature the partition function is approximately that of
a particle in a two-dimensional box with sides a and 2πR.

(b) The stated regime implies that the thermal energy kT satisfies the inequalities
~2 ~2
kT .
mR2 ma2

We can use the high-temperature result (2) for Z(a),
but not for Z(πR):

  

ma2 ma2 β ~2
Z(a, R) = 1 + 2Z(πR) = 1 + 2 exp − + ...
2πβ ~2 2πβ ~2 2mR2


The leading correction arises from the first term in the sum that defines Z(πR). This is a
“low-temperature” small correction; the argument of the exponential is large and negative.

1
3.1. Lorentz covariance for motion in electromagnetic fields
We rearrange the indices by raising the µ in equation (1) of the problem statement:

dpµ q dxν
= F µν .
ds c ds
Multiplying both sides of the equation by ds/dt we find

dpµ q dxν
= F µν .
dt c dt
dxν
We test this equation using F µν , as given in the text, and dt
= (−c, vx , vy , vz ):

dp1 q q
= (F 10 (−c) + F 12 vy + F 13 vz ) = qEx + (vy Bz − vz By )
dt c  c
1
 + v × B
=q E good!
c x
dp2 q q
= (F 20 (−c) + F 21 vx + F 23 vz ) = qEy + (vz Bx − vx Bz )
dt c  c
1
= q E + v × B
  good!
c y
dp3 q q
= (F 30 (−c) + F 31 vx + F 32 vy ) = qEz + (vx By − vy Bx )
dt c  c
1
 + v × B 
=q E good!
c z

The last equation is


dp0 q dxi q
= F 0i = E · v .
dt c dt c
With p0 = E/c, it becomes

dE  · v = force × velocity
= (q E)
dt
The rate of change of the particle energy equals the rate at which the fields do work on the
particle. The magnetic force is perpendicular to the velocity and does not do work.

1
3.2. Maxwell equations in four dimensions
(a) T is totally antisymmetric, so T vanishes unless all the indices are different. This yields
four equations: T012 = 0, T013 = 0, T023 = 0, and T123 = 0. The first three of them give

T012 = 0 → ∂0 F12 + ∂1 F20 + ∂2 F01 = 0


1∂ ∂ ∂
Bz + Ey − Ex = 0
c ∂t ∂x ∂y
∂Ey ∂Ex 1 ∂Bz
→ − =−
∂x ∂y c ∂t
T013 = 0 → ∂0 F13 + ∂1 F30 + ∂3 F01 = 0
1∂ ∂ ∂
(−By ) + Ez − Ex = 0
c ∂t ∂x ∂z
∂Ex ∂Ez 1 ∂By
→ − =−
∂z ∂x c ∂t
T023 = 0 → ∂0 F23 + ∂2 F30 + ∂3 F02 = 0
1∂ ∂ ∂
(Bx ) + Ez − Ey = 0
c ∂t ∂y ∂z
∂Ez ∂Ey 1 ∂Bx
→ − =− .
∂y ∂z c ∂t

 = − 1 ∂ B . Finally T123 = ∂1 F23 + ∂2 F31 + ∂3 F12 = 0


They are the three components of ∇ × E c ∂t
gives
∂Bx ∂By ∂Bz
+ + =0 → ∇·B  = 0.
∂x ∂y ∂z

(b) Test values for µ in the equation ∂F µν /∂xν = j µ /c. For µ = 0:

∂F 0i j0
= = ρ which is ∇ · E
 = ρ.
∂xi c
µ = i: To do the 3 indices at once note that F ij = ijk B k where  is totally antisymmetric
and 123 = 1. For example F 12 = 12k B k = 123 B 3 = B 3 . Using this, we have

1 ∂F i0 ∂F ij ji 1 ∂E i ∂ ji
+ = → − + j (ijk B k ) =
c ∂t ∂xj c c ∂t ∂x c
A little rearrangement gives
  
∂ k j i 1 ∂E i j 1 ∂ E
 ijk
B = + , which is (∇ × B)
 i= + .
∂xj c c ∂t c c ∂t i

1
3.3 Electromagnetism in three dimensions
(a) Using equation (3.9) and noting that there are no currents nor velocity in the z direction,
we have
Bx = By = 0, Ez = 0, jz = 0, vz = 0 .
Maxwell’s equations then become


⎪ ∂Ex ∂Ey

⎪ + =ρ from ∇ · E
 =ρ

⎪ ∂x ∂y



⎪ ∂Ey − ∂Ex = − 1 ∂Bz from ∇ × E
Maxwell’s ⎪
⎪  = − 1 ∂B
⎨ ∂x ∂y c ∂t c ∂t
equations ⎫

⎪ ∂Bz jx 1 ∂Ex ⎪ ⎪

down to 3D ⎪⎪
⎪ = + ⎬ 

⎪ ∂y c c ∂t
∇ ×  = j + 1 ∂E

⎪ from B

⎪ ∂Bz jy 1 ∂Ey ⎪ ⎪ c c ∂t
⎩ − ∂x = c + c ∂t ⎪
⎪ ⎭

while we get nothing from ∇ · B


 = 0. The force law gives nontrivial equations only for the
x and y components:
dpx vy dpy vx
= q(Ex + Bz ) , = q(Ey − Bz ) .
dt c dt c

(b) In three dimensions we have Aµ = (Φ, A1 , A2 ), Aµ = (−Φ, A1 , A2 ), and j µ = (cρ, j 1 , j 2 ).


Moreover, Fµν = ∂µ Aν − ∂ν Aµ , so
1 ∂Ai ∂Φ ∂Ay ∂Ax
F0i = + i ≡ −Ei , F12 = − ≡ Bz just a name!
c ∂t ∂x ∂x ∂y
Thus the field strength looks like
⎛ ⎞ ⎛ ⎞
0 −Ex −Ey 0 Ex Ey
⎜ ⎟ ⎜ ⎟
Fµν = ⎝Ex 0 Bz ⎠ F µν = ⎝−Ex 0 Bz ⎠ .
Ey −Bz 0 −Ey −Bz 0

The above 3-dimensional F can be viewed as the 4-dimensional one with Ez = Bx = By = 0.


The 3-dimensional j can be viewed as the 4-dimensional j with jz = 0. The Maxwell
equations that arise in 3 dimensions are therefore the truncation to Ez = Bx = By = 0 and
jz = 0 of the original four-dimensional ones. Those are simply the ones given in part (a), so
there is no need to check anything further.
The force law
dpµ q dxν
= F µν
dt c dt
1
gives
 
dpx F 10 F 12 vy
=q (−c) + q vy = q Ex + Bz ,
dt c c c
 
dpy F 20 F 21 vx
=q (−c) + q vx = q Ey − Bz ,
dt c c c

so everything is as expected. Also,

dp0 qF 0i q dE
= vi = E · v ⇒  · v
= qE familiar result!
dt c c dt

2
3.4 Electric fields and potentials of point charge
(a) T0ij = ∂0 Fij + ∂i Fj0 + ∂j F0i . For time-independent fields ∂0 Fij = 0, so T0ij = 0 implies

∂i F0j − ∂j F0i = 0 → ∂i Ej − ∂j Ei = 0 . (1)

 = −∇Φ says Ei = −∂i Φ, so condition (1) is satisfied since ∂i ∂j Φ = ∂j ∂i Φ.


E
(b) With d spatial dimensions and a point charge q at x = 0,

Γ( d ) q

E(r) = 2d d−1 (equation (3.73)) .
2π 2 r
Since E(r) = − dΦ
dr
(r),
Γ( d2 ) q 1
Φ(r) = d d−2 d − 2
,
2π 2 r
setting Φ = 0 at r = ∞ for d > 2. Since Γ(x + 1) = xΓ(x), Γ( d2 ) = ( d2 − 1)Γ( d2 − 1), so we get

Γ( d2 − 1) q
Φ(r) = d , d > 2.
4π 2 rd−2

1
3.5 Calculating the divergence in higher dimensions.
The thin shell has volume vol(S d−1 (r)) dr, so applying the divergence theorem we find

(∇ · f) vol(S d−1 (r)) dr = f (r + dr) vol(S d−1 (r + dr)) − f (r) vol(S d−1 (r)) .

The left-hand side is the volume integral of the divergence of f over the shell. The right-hand
side is the net flux of f out of the shell. Dividing both sides of this equation by vol(S d−1 (r) dr
and taking the limit dr → 0, we find
1 d 
∇ · f = f (r) vol (S d−1
(r)) .
vol (S d−1 (r)) dr

Since vol (S d−1 (r)) ∼ rd−1 , with a constant of proportionality that cancels out between the
numerator and the denominator, we have
1 d d−1 
∇ · f = r f (r) .
rd−1 dr
1 d
This is the final formula. For d = 3 it gives ∇ · f = r 2 dr
(r2 f (r)) = df
dr
+ 2r f (r) .

1
3.6 Analytic continuation for gamma functions.
We begin by breaking up the integral into two pieces
 1  ∞
−t z−1
Γ(z) = dt e t + dt e−t tz−1 . (1)
0 1

The second term converges for any z since the problematic region of integration near t = 0
has been removed; whatever the value of z, we have e−t tz−1 → 0 for t → ∞. Using the
expansion for e−t , the first term in (1) can be written as
 1  1 

(−t)n
dt e−t tz−1 = dt tz−1 . (2)
0 0 n=0
n!

Note that in the sum the nth term converges for (z) > −n. If we remove the first N terms,
the remaining sum from N + 1 to ∞ will converge for (z) > −(N + 1). We thus rewrite
(2) as
 1 ∞  1 N
z−1 (−t)n z−1 (−t)n
dt t + dt t
0 n=N +1
n! 0 n=0
n!

The first term can be rewritten as


 1 ∞  1  N 
(−t)n −t (−t)n
z−1
dt t = dt tz−1
e − .
0 n=N +1
n! 0 n=0
n!

In the second term we can perform the integration assuming that (z) > 0. This gives
 1 
N 
N  1 
N
z−1 (−t)n (−1)n n+z−1 (−1)n 1
dt t = dt t = .
0 n=0
n! n=0
n! 0 n=0
n! z+n

Putting all the terms together, we have


 1  
N  
N  ∞
−t (−t)n (−1)n 1
Γ(z) = dt t z−1
e − + + dt e−t tz−1 .
0 n=0
n! n=0
n! z+n 1

While this was derived for (z) > 0, the integrals converge for (z) > −N − 1, so the
right-hand side provides the analytic continuation. The result is a meromorphic function,
namely, a function with poles. These arise from the term


N
(−1)n 1
. (3)
n=0
n! z+n

1
which is just a sum of simple poles at the integers n = 0, −1, −2, . . . , N . Since we can take
N to be arbitrarily large we see that Γ(z) has a simple pole at every integer, n ≤ 0. Finally
we may read off the residue of the poles from (3)

(−1)n
Res (Γ(z)) = .
z=−n n!

2
3.7 Simple quantum gravity effects are small.
(a) The standard Bohr radius

~2 e2
a0 = = 5.29 × 10−9 cm, arises from the potential V = − .
me e2 r
In the gravitational case, the potential is V = −Gme mp /r2 , so, the “gravitational” Bohr
radius ag0 is obtained by replacing e2 with Gme mp :
 
g ~2 ~2 e2
a0 = = .
me (Gme mp ) me e2 Gme mp
The ratio between the “gravitational” and the conventional Bohr radius is therefore
ag0 e2 (4.8 × 10−10 )2
= =  2.27 × 1039 .
a0 Gme mp (6.67 × 10−8 )(0.911 × 10−27 )(1.672 × 10−24 )
This gives
ag0 = 1.20 × 1031 cm.
Since 1 light year ≈ 9.5 × 1017 cm, ag0 = 1.3 × 1013 light years !! The universe is about 1010
years old, so this distance is fantastically large.
(b) We introduce factors of G, c, and ~ into the formula, with exponents α, β, and γ to be
determined:
1 Gα cβ ~γ
kT = → kT = .
8πM 8πM
Since [kT ] = [E] = M L2 /T 2 , using [G] = L3 M −1 T −2 , [c] = LT −1 , and [~] = M L2 T −1 , we
find
M L2
= L3α+β+2γ M −α+γ−1 T −2α−β−γ
T2
This gives three equations for three unknowns:

3α + β + 2γ = 2 , − α + γ = 2, − 2α − β − γ = −2 .

The solution is α = −1, β = 3, and γ = 1. As a result,


~c3 ~c c2 m 2 c2
kT = =( )= P ,
8πGM G 8πM 8πM
where mP = 2.17 × 10−5 gr is the Planck mass. For a mass with million solar masses M =
106 Msun ≈ 2 × 1039 gr,
(2.17 × 10−5 )2 (3 × 1010 )2
kT = = 8.43 × 10−30 eV =⇒ T ≈ 6.1 × 10−14 K .
8π 2 × 10 39

1
This is much colder than the nanokelvins attained by Ketterle et al. in Bose-Einstein con-
densation experiments. For a room temperature black hole, T  300 K, and we find

6.1 × 10−14 K
M= 2 × 1039 gr ≈ 4.1 × 1020 kg ,
300 K
which is approximately 6.8 × 10−5 Mearth or 5.6 × 10−3 Mmoon .

2
3.8 Planetary motion in four and higher dimensions.

ee
er P
e

Figure 1: Motion in the plane described with polar coordinates.

We begin by deriving the usual results concerning effective potentials for the motion in a
centrar force field. Using time-dependent unit vectors er (t) and eθ (t) in the radial and in
the tangential direction we have:
x(t) = r(t)er (t)
x˙ (t) = ṙer (t) + rθ̇eθ (t)
x¨(t) = [r̈ − rθ̇2 ]er + [2ṙθ̇ + rθ̈]eθ
If the force is central (along er ) we have
d 2
2ṙθ̇ + rθ̈ = 0 → (r θ̇) = 0 .
dt
The constancy of r2 θ̇ implies the constancy of the angular momentum L = mr2 θ̇. We thus
have
L
θ̇ = , (1)
mr2
with L a constant of motion. Let F be the gravitational (central) force associated to the
potential energy Vgr :
∂Vgr
F = − er = m[r̈ − rθ̇2 ]er .
∂r
Together with (1) this equation gives
 
L2 1 ∂Vgr ∂ Vgr L2
r̈ − 2 3 = − → r̈ = − + .
mr m ∂r ∂r m 2m2 r2
  
effective potential

1
This completes our derivation of the effective potential. Noting that the gravitational po-
tential energy Vgr divided by the mass is simply the gravitational potential Vg , we have the
familiar result
L2
Veff = Vg + . (2)
2m2 r2

(4)
Veff(r)

r0 r

Figure 2: Effective potential for four-dimensional planetary motion.

In four dimensions
(4) GM L2
Veff = − + . (0.1)
r 2m2 r2
As seen by the shape of the potential in Figure 2, the radial motion is stable around the
equilibrium position r0 . Indeed

∂Veff  GM L2 L2
= + − = 0 ⇒ r = orbit radius .
∂r r0
0
r02 m2 r03 GM m2

Computing the second derivative of the potential at the critical point we confirm it is a
minimum:

∂ 2 Veff  2GM 3L2 2GM 3GM GM
2  = − 3
+ 2 4
=− 3 + 3
= 3 > 0 −→ stable!
∂r r0 r0 m r0 r0 r0 r0

Our familiar world has the additional property that non-circular closed orbits exist, but we
can’t see that from the effective potential.
In five dimensions
(5) αG(5) M L2
Veff = − + ,
2r2 2m2 r2
2
(5)
Veff(r)

Figure 3: Five spacetime dimensions and L > L0 . Planet will go out to r → ∞

(5)
Veff(r)

Figure 4: Five spacetime dimensions and L < L0 . Planet will fall into the star at r = 0.

with α > 0, a dimensionless constant that can be determined using Gauss’ law. Note that
(5)
both terms have the same r dependence! There is a critical L0 which makes Veff vanish:

L20 = αG(5) M m2 .

In terms of L0 the potential is

(5) 1
Veff = (L2 − L20 ) .
2m2 r2
For L > L0 , see Figure 3. For L < L0 , see Figure 4. The orbits are unstable. They can
exist for any radius if L = L0 exactly, but any small perturbation that changes the angular
momentum will either make the planet spiral in or out.

3
For spacetime dimensions D > 5

(D) G(D) M L2
Veff = −α(D) + .
rD−3 2m2 r2
(D)
Since D − 3 > 2, Veff is as in Figure 5. An unstable circular orbit is possible with an
L-dependent radius r0 .

Figure 5: Effective potential for D ≥ 6 spacetime dimensions.

4
3.9 Gravitational field of a point mass in compactified five-dimensional
world.
(a) We examine the gravitational equation

∇2 Vg(5) = 4πG(5) ρm ,

for a point particle of mass M fixed at the origin r = x2 + y 2 + z 2 + w2 = 0 of a five
dimensional spacetime. Consider a four-dimensional ball B 4 (r) that contains the mass. The
boundary of this ball is the 3-sphere S 3 (r). Integrating the above equation over B 4 (r):

d(Vol)∇ · ∇Vg(5) = 4πG(5) M .
B 4 (r)

Using the divergence theorem the left-hand side can be written as


  (5) (5)
∂Vg ∂Vg
d(Vol) Flux of ∇Vg(5) = d(Vol) = Vol(S 3 (r)) · (r)
S 3 (r) S 3 (r) ∂r ∂r
(5)
because ∇Vg is radial and of constant magnitude over the three sphere. Since Vol(S3 (r)) =
2π 2 r3 we get
(5) (5)
2 3 ∂Vg (5) ∂Vg 2 G(5) M
2π r = 4πG M → = .
∂r ∂r π r3
Integrating, with zero potential at infinity, we find

1 G(5) M
Vg(5) (r) = − . (1)
π r2

(b) We consider the compactification w ∼ w + 2πa of the w direction into a circle of radius
a (see Figure 1). Note that the potential at the point P with coordinates (x, y, z, 0) cannot
just be
1 G(5) M
Vg(5) (x, y, z, 0) = − not true
π x2 + y 2 + z 2
as if we could use (1) when the space is compactified. If (1) could be used directly, for a
point P  displaced vertically from P by 2πa we would also write

1 G(5) M
Vg(5) (x, y, z, 2πa) = − not true .
π x2 + y 2 + z 2 + (2πa)2

But then, the potentials at P and P  would be different, in contradiction with the compact-
ification statement that P and P  are the same point.

1
w

M
3/ a

M 2
/a
1 x,y,z
M 3 P
</a

Figure 1:

To implement the w periodicity implied by the compactification we must introduce a periodic


array of masses; a mass at each w = 2πna, for every integer n ∈ Z. This is an infinite number
of masses! Note that with periodic masses the potential will also be periodic. As shown in
Figure 2, the observer actually “sees” the additional masses because the light rays wrap
around the cylinder. Adding the contributions of each mass using (1), we find the potential:

1 (5) ∞
1 1 (5) ∞
1
Vg(5) (x, y, z, 0) =− G M 2 2 2 2
=− G M
π n=−∞
x + y + z + (2πna) π n=−∞
R + (2πna)2
2

where R2 ≡ x2 + y 2 + z 2 . We thus have

1 G(5) M 

1
Vg(5) (x, y, z, 0) = − (2)
π R n=−∞ 1 + (2π Ra n)2
2

(5)
(c) Let us evaluate the above Vg when a
R
 1. In this case the terms in the sum are slowly

2
2
M 1 P

Figure 2:

varying and

∞  ∞
1 ∼ dn
2
= .
n=−∞
1 + (2π a
R
n) −∞ 1 + (2π Ra n)2
R
Letting n = 2πa
x,

∞  ∞
1 ∼ R dx R
a 2
= 2
=
n=−∞
1 + (2π R n) 2πa −∞ 1+x 2a

Back in (2) we get


 (5) 
∼ G(5) M 1 R G M
V (x, y, z, 0) = − 2
=− ,
π R 2a 2πa R

that is, V ∼
= − GM
R
with G = G(5)
2πa
, which is exactly what we wanted to show!

3
3.10 Exact answer for gravitational potential.
(a) Using the cited identity with x = 2a/R, the potential in equation (2) of Problem 3.9
becomes:

 (5) 

(5) G(5) M 1 R R G M R
Vg (x, y, z, 0) = − 2
coth =− coth (exact) .
π R 2a 2a 2πa R 2a
!
(b) We can do an asymptotic expansion of coth R
2a
for R a:
!

cosh R
e 2a + e− 2a
R R
R 2a
coth = ! =
e 2a − e− 2a
R R
2a sinh R
2a

1 + e− a 2e− a
R R

= =1+
−a
1 − e− a 
R R
1 − e
2R
= 1 + 2 e− a + e−
R
a + ... .

So the potential with its first leading correction is


 (5)  
(5) G M 2R
1 + 2e− a + O(e− a ) ,
R
Vg = − a  R.
2πa R
1
Correction is 1% when 2e− a =
R
100
. This gives R
a
= ln 200  5.3.
(c) When R  a, we are near the mass. For small x, coth(x)  x1 + x3 + O(x3 ). So we have

 
R ∼ 2a R 2a R2
coth = + + ... = 1+ + ... .
2a R 6a R 12a2

We then find
   
∼ G(5) M 2a R2 G(5) M R2
Vg(5) =− 1+ + ... = − 1+ + ... , R  a.
2πa R R 12a2 πR2 12a2

The leading term gives the 5-dimensional answer (see equation (1) of Problem 3.9). This
makes sense; very near the mass the effects of compactification must be negligible.

1
4.2 Longitudinal waves on strings.
Let η(x, t) denote the longitudinal displacement at x, for any time t. One may then ask,
what is the tension T (x, t)? Given the definition of the coefficient of tension, we can write

∆L
T (x, t) = T0 + τ0 .
L
To evaluate the second term on the right-hand side, we look at a small piece of parameter
space from x −  to x + . The corresponding piece of static string has length L = 2, and
it is stretched ∆L = η(x + , t) − η(x − , t). So

∆L 1  ∂η
= η(x + , t) − η(x − , t)  , when  → 0.
L 2 ∂x
As a result, we find
∂η
T (x, t) = T0 + τ0 . (1)
∂x
This gives the tension at any point on the string. Now consider the piece of string associated
with the parameter space interval from x to x + dx. The mass of this piece of string is
2
µ0 dx, its average acceleration is ∂∂t2η , and the net force acting on it is T (x + dx) − T (x). So,
Newton’s law gives
∂2η ∂T
(µ0 dx) 2 = T (x + dx) − T (x) = dx .
∂t ∂x
Cancelling the common factor of dx and using (1), we find

∂2η ∂2η
µ0 = τ 0 .
∂t2 ∂x2
This takes the form of the familiar wave equation for transverse oscillations, with the tension
T0 replaced
 by the coefficient of tension τ0 . The velocity v of the longitudinal waves is
v = τ0 /µ0 . You can also prove that the tension T (x, t) satisfies exactly the same wave
equation.

1
4.4 A configuration with two joined strings.
The equation for the oscillation frequencies breaks into two equations

d2 y1 µ1 2
+ ω y1 (x) = 0 , for 0 ≤ x ≤ a ,
dx2 T0
d2 y2 µ2 2
+ ω y2 (x) = 0 , for a ≤ x ≤ 2a .
dx2 T0
Here y1 (x) = y(x) for x ∈ [0, a] and y2 (x) = y(x) for x ∈ [a, 2a]. We let
µ1 2 µ2 2
λ21 = ω and λ22 = ω .
T0 T0
With this notation, the two equations take similar form
"
d2 yi 2 µi
2
+ λi y(x) = 0 , λi = ω, i = 1, 2.
dx T0

Since y(x = 0) = y(x = 2a) = 0 we write

y1 (x) = A sin(λ1 x) , and y2 (x) = B sin(λ2 (2a − x)) ,

which satisfy the differential equation and the boundary conditions at x = 0 and x = 2a.
(a) Since the string does not break at x = a, we must have y1 (a) = y2 (a). Since the there is
no mass at the joining point the slopes must agree too: y1 (a) = y2 (a).
(b) Applying the boundary conditions in (a) we find

A sin λ1 a = B sin λ2 a ,
Aλ1 cos λ1 a = −Bλ2 cos λ2 a .

These two equations can be written as


# $# $
sin λ1 a − sin λ2 a A
= 0.
λ1 cos λ1 a λ2 cos λ2 a B

Since we need solutions where both A and B are not zero, the determinant of the above
matrix must vanish:
λ2 sin λ1 a cos λ2 a + λ1 cos λ1 a sin λ2 a = 0 . (1)
This equation can be used to find the oscillation frequencies. As a check, assume µ1 = µ2 =
µ0 , in which case λ1 = λ2 ≡ λ. The condition becomes sin(2λa) = 0, which gives λ(2a) = nπ.
This gives the familiar frequencies.

1
(c) With µ1 = µ0 and µ2 = 2µ0 we write
" "
µ0 2µ0 √
λ1 = ω ≡ λ and λ2 = ω = 2λ .
T0 T0

Equation (1) then gives


√ √ √
2 sin λa cos 2λa + cos λa sin 2λa = 0 .

Solving this equation numerically we find that the lowest root is λa  1.26567. This gives
the lowest frequency  
T0 1.26567 T0 2.53134
ωL = = .
µ0 a µ0 (2a)
Note that ωL is bounded as
 √ 
T0 π/ 2 T0 π
< ωL < .
µ0 (2a) µ0 (2a)

The lower and upper bounds√are the lowest frequencies of strings of constant mass density
2µ0 and µ0 , respectively (π/ 2  2.22144).

2
4.6. Deriving Euler-Lagrange equations.
dδq
(a) Under a variation δq(t) of the coordinate, the variation δ q̇(t) of the velocity is dt
. Since
the Lagrangian L depends on q and q̇, the full variation is
 
∂L ∂L dδq
δS = dt δq + .
∂q ∂ q̇ dt
The second term in the brackets is rewritten in terms of a total time derivative and a term
proportional to δq:
     
∂L d ∂L d ∂L
δS = dt δq + δq − δq
∂q dt ∂ q̇ dt ∂ q̇
The total time derivative does not contribute when we set the variations to vanish at the
initial and final times (we always assume this). The variation δS is therefore
   
∂L d ∂L
δS = dt δq(t) − .
∂q dt ∂ q̇
If this variation is to vanish for all δq(t) we must have
 
d ∂L ∂L
− = 0. (1)
dt ∂ q̇ ∂q
This is the Euler-Lagrange equation.
(b) Under a variation δφ(x) of the field, the variation δ(∂µ φ) of the field derivative is equal
to ∂µ (δφ). Since the Lagrangian density L depends on φ and ∂µ φ, the full variation is
 
D ∂L ∂L
δS = d x δφ + ∂µ (δφ)
∂φ ∂(∂µ φ)
     
∂L ∂L ∂L
= d x D
δφ + ∂µ δφ − ∂µ δφ .
∂φ ∂(∂µ φ) ∂(∂µ φ)
The total derivatives (middle term) do not contribute if the variations vanish at infinity or,
more precisely, if δφ(x0 , x1 , x2 , . . . , xd ) vanishes when any of its arguments is equal to plus
or minus infinity. Consider, for example, the term with µ = 1:
   
x1 =∞
0 1 2 d ∂ ∂L 0 2 d ∂L
dx dx dx . . . dx δφ = dx dx . . . dx δφ = 0,
∂x1 ∂(∂1 φ) ∂(∂1 φ) x1 =−∞

if δφ(x0 , x1 = ±∞, x2 , . . . , xd ) = 0. Back to the variation, we therefore have


     
∂L ∂L ∂L ∂L
D
δS = d x δφ(x) − ∂µ = 0 → ∂µ − = 0,
∂φ ∂(∂µ φ) ∂(∂µ φ) ∂φ
which is the Euler-Lagrange equation for the field.

1
5.2 Particle equation of motion with arbitrary parameterization.
We are told to vary xµ (τ ) → xµ (τ ) + δxµ (τ ) in the action
 τf "
dxµ dxν
S = −mc −ηµν dτ .
τi dτ dτ
The variation gives
 µ ν  dδxµ dxµ
τf
1 − 2ηµν dδx dx τf
δS = −mc % dτ dτ
dτ = mc % dτ dτ dτ .
τi 2 −η dxα dxβ τi −ηαβ dx
α dxβ
αβ dτ dτ dτ dτ

Integrating by parts and dropping total derivatives (δxµ (τi ) = δxµ (τf ) = 0) we find
 τf

d mc dx µ

δS = − µ
dτ δx (τ ) % dτ
.
τi dτ −ηαβ dτ dτ
dxα dxβ

The action is stationary if


d mc dx µ

% dτ
=0 . (1)
dτ −ηαβ dx
α dxβ
dτ dτ

This is the equation of motion for the particle in “manifestly” reparameterization invariant
form. Indeed, the object between the brackets is clearly reparameterization invariant:

mc dx µ
mc dxµ
dτ 
% dτ
=% .
−ηαβ dx −ηαβ dx
α dxβ α dxβ
dτ dτ dτ  dτ 


(This follows immediately using the chain rule dx

= dτ dx
dτ dτ 
.) The derivative in front of the
brackets (see (1)) does not spoil the reparameterization invariance since:
d dτ  d d
[· · · ] = [· · · ] = 0 → [· · · ] = 0 .
dτ dτ dτ  dτ 

When we choose τ equal to s


dxµ dxν dxµ dxν (ds)2
−ηµν = −ηµν = = 1.
dτ dτ ds ds (ds)2
Equation (1) then becomes the familiar
d dxµ ! dpµ
mc = = 0.
ds ds ds

1
5.5 Equations of motion for a charged particle.
We are interested in the variation of the action
 
q dxµ
S = −mc ds + I , I = dτ Aµ (x(τ )) (τ ) , (1)
P c P dτ
when we let xµ (τ ) → xµ (τ ) + δxµ (τ ). We note that
  ∂A
µ
δAµ (x(τ )) ≡ Aµ x(τ ) + δx(τ ) − Aµ x(τ ) = δxν (τ ) ,
∂xν
∂Aµ
where ∂xν
is calculated at x = x(τ ). We therefore have
 
∂Aµ ν dxµ dδxµ
δI = dτ δx + dτ A µ .
P ∂xν dτ P dτ
We exchange µ ↔ ν in the first term and rewrite the second using a total derivative:
 ν 

µ ∂Aν dx d µ dAµ
δI = dτ δx + dτ (Aµ δx ) − δx
µ
.
P ∂xµ dτ P dτ dτ

We assume that δx vanishes at the ends of P, so the total derivative (first term in brackets)
vanishes. Using the chain rule for the second term in brackets we find
 ν  ν
µ ∂Aν dx µ ∂Aµ dx
δI = dτ δx − dτ δx .
P ∂xµ dτ P ∂xν dτ
The two terms can now be combined to write
   
∂Aν ∂Aµ dxν dxν
δI = dτ δxµ
− = dτ δxµ
Fµν .
P ∂xµ ∂xν dτ P dτ

This concludes our computation of the variation of I. The variation of the first term in (1)
was calculated in the textbook (equation (5.29)). The total variation is therefore
 
µ dpµ q dxν
δS = − dτ δx + dτ δxµ Fµν
dτ c P dτ
 P  ν

dpµ dx
= dτ δxµ − + Fµν .
P dτ dτ
Setting the variation equal to zero, we find the expected equation of motion
dpµ q dxν
= Fµν .
dτ c dτ

1
5.6 Electromagnetic field dynamics with a charged point particle.
Write 
q dxµ
Sint = dτ Aµ (x(τ )) (τ ) .
c P dτ
Since 
Aµ (x(τ )) = dD xδ D (x − x(τ )) Aµ (x) ,

we have
 

1 dxµ 1
Sint = 2 d x Aµ (x) qc dτ δ (x − x(τ ))
D D
= 2 dD x Aµ (x) j µ (x) ,
c P dτ c

where j µ (x) is the current of Problem 5.3. Under a variation δAµ (x) of the gauge field

1 j µ (x)
δSint = dD x δAµ (x) .
c c

Let now 
1
Sem =− dD xFµν F µν .
4c
We then have
  
1 1
δSem =− d x 2 · F δFµν = −
D µν
dD x · F µν ∂µ δAν − ∂ν δAµ
4c 2c

1 ∂δA µ
=+ dD x F µν (since F µν = −F νµ )
c ∂xν

1 ∂F µν
=− dD x δAµ (throwing away boundary terms) .
c ∂xν
Collecting variations we have
  µν 
1 ∂F j µ (x)
δSint + δSem =− D
d x δAµ (x) − .
c ∂xν c

Setting the total variation to vanish we find the field equations


∂F µν jµ
= .
∂xν c

1
5.7 Point particle action in curved space.
Our starting point is the variation

δS = −mc δ(ds) . (1)

With the path parameterized by an arbitrary τ we have


dxµ dxν
(ds)2 = −gµν (x(τ )) (dτ )2 . (2)
dτ dτ
Under a variation x(τ ) → x(τ ) + δx(τ ) the variation of the metric is
∂gµν α
δgµν (x(τ )) = gµν (x + δx) − gµν (x) = δx (τ ) ,
∂xα
so the variation of (2) gives
∂gµν α dxµ dxν 2 dxµ dδxν
2 ds δ(ds) = − δx (dτ ) − 2gµν (dτ )2 .
∂xα dτ dτ dτ dτ
Dividing by 2ds and cancelling one factor of dτ , we find
1 ∂gµν α dxµ dxν dxµ dδxα
δ(ds) = − δx dτ − gµα dτ .
2 ∂xα ds dτ ds dτ
Back into (1), we have
  
1 ∂gµν α dxµ dxν dxµ dδxα
δS = mc dτ δx + gµα .
2 ∂xα ds dτ ds dτ
Integrating by parts and dropping total derivatives that vanish since δx = 0 at the endpoints
of the world-line
  µ ν


α 1 ∂gµν dx dx d dxµ
δS = mc dτ δx − gµα .
2 ∂xα ds dτ dτ ds
It is now convenient to choose the arbitrary parameter τ equal to s, so that
  µ ν


α 1 ∂gµν dx dx d dxµ
δS = mc ds δx − gµα .
2 ∂xα ds ds ds ds
[Note: if we had used s as a parameter from the start, we might worry about the change in
the range of s when the path changes.] The equation of motion simply requires the vanishing
of the object inside parentheses. Expanding it out
d2 xµ ∂gµα dxν dxµ 1 ∂gµν dxµ dxν
gµα + − = 0.
ds2 ∂xν ds ds 2 ∂xα ds ds
1
The last two terms in the left-hand side can be combined:
 
d2 xµ 1 ∂gµα ∂gµν dxµ dxν
gµα + 2 ν − = 0.
ds2 2 ∂x ∂xα ds ds
µ ν
Since the expression in parenthesis multiplies dx dx
ds ds
, which is symmetric in µ, ν, we rewrite
it in a form which is also symmetric in µ, ν, to get
 
d2 xµ 1 ∂gµα ∂gνα ∂gµν dxµ dxν
gµα + + − = 0.
ds2 2 ∂xν ∂xµ ∂xα ds ds

Multiplying the equation by g λα and recalling that g λα gµα = δµλ , we have


 
d2 xλ 1 λα ∂gαµ ∂gαν ∂gµν dxµ dxν
+ g + − = 0.
ds2 2 ∂x ν ∂x µ ∂xα ds ds

The equation has taken the form

d2 xλ µ
λ dx dx
ν
+ Γ = 0,
ds2 µν
ds ds
as we wanted to show. [For the quoted form let λ → µ, µ → α, and ν → β.]

2
6.1 Stretched string and a nonrelativistic limit.
We are told to consider the string action in the form
  "
tf 2
v⊥
S = −T0 dt ds 1− . (1)
ti c2

The second integral here is simply an integral along the string; it is not appropriate to write
limits of integration because the integration variable has not been chosen yet.
We describe the motion using the function y (t, x), where x is a (horizontal) coordinate
along the equilibrium direction of the string and y represents transverse displacement. To
perform the integral along the string in (1), we examine the contribution from the piece of
string that stretches between x and x + dx. This piece is represented by the spatial vector
(dx , dX 2 , . . . , dX d ) = (dx, dy ). The length ds of this small piece of string is given by the
Pythagorean theorem:
ds2 = dx2 + dy · dy .
This implies that " ∂y 2

1 ∂y 2
ds = dx 1+  dx 1 + , (2)
∂x 2 ∂x
where we used the small-oscillation approximation | ∂x∂
y
|  1. Additionally, since the string is
almost horizontal everywhere and at all times (| ∂x |  1, again) the perpendicular velocity
∂
y

is approximately equal to the vertical velocity ∂


y
∂t
, so we write

∂y
v⊥  .
∂t
Since |v⊥ |  c (problem statement) we can approximate
"
2
v⊥ 2
1 v⊥ 1 ∂y 2
1− 2  1−  1− 2 . (3)
c 2 c2 2c ∂t
Using (2) and (3) in (1), we find
 tf 
 
a
1 ∂y 2 1 ∂y 2
S  −T0 dt dx 1 + 1− 2 .
ti 0 2 ∂x 2c ∂t

Since the terms in brackets and parentheses are of the form 1 + , with  a small correction,
multiplying out we only keep terms linear in the corrections
 tf  a

1 ∂y 2 1 ∂y 2
S  −T0 dt dx 1 + − 2 .
ti 0 2 ∂x 2c ∂t

1
The action thus becomes
 tf  

tf a
1 T0 ∂y 2 1 ∂y 2
S dt(−T0 a) + dt dx − T0 .
ti ti 0 2 c2 ∂t 2 ∂x

This can be compared with the action in equation (4.36). We have indeed obtained the
action of a nonrelativistic string with tension T0 and mass density µ0 = T0 /c2 . The additional
constant corresponds to a potential energy T0 a, namely tension times the length of the string.

2
6.2 Alternative derivation of the non-relativistic limit.
In this problem we work in the static gauge X 0 ≡ ct = cτ supplemented by the additional
gauge condition X 1 = x = aσ/σ1 . The other embedding coordinates X i (τ, σ), with i =
2, 3, . . . d can then be viewed as functions X i (t, x) of parameters t and x. They will be
collectively denoted as y (t, x). The nonrelativistic approximation is based on
1 
 
|∂x y |  1 and  ∂t y  1 . (1)
c

We begin with the action


  %&
T0 σ1 '2 & '2 & '2
S=− dτ dσ ∂σ X · ∂τ X − ∂τ X ∂σ X .
c 0

Because of reparameterization invariance we can replace τ by t and σ by x to get:


 2
  a
1 1 & '2 & '2
S = −T0 dt dx ∂x X · ∂t X − 2 ∂t X ∂x X . (2)
0 c c

Noting that ∂x X 0 = 0 and ∂x X 1 = 1, we have


1 & '2 1
2
∂t X = −1 + 2 (∂t y )2 ,
c c
& '2 & '2
∂x X = 1 + ∂x y ,
 2  2
1 1
∂x X · ∂t X = ∂x y · ∂t y .
c c

The term in the last line does not contribute since, on account of (1), it is a fourth power of
a small number. The terms in the first two lines will contribute; each contains corrections
quadratic in the small number. We can approximate S in (2) by action as
  a  
1 & '2 
S  −T0 dt dx 1 − 2 (∂t y )2 1 + ∂x y .
0 c

Multiplying out, keeping only quadratic terms, and using 1 +   1 + 12  we find the
nonrelativistic action we derived previously in Problem 6.1:
  a  
1 T0 2 1 2
S  dt dx −T0 + (∂t y ) − T0 (∂x y ) .
0 2 c2 2

1
6.3 Planar motion for open string with attached endpoints.
2
To solve this problem we must find an expression for v⊥ in terms of y. Since the motion
occurs in two spatial dimensions, the location of the string is described by
 = (x , y(x, t) ) .
X (1)

 given by
Associated with an infinitesimal dx we have the small string segment dX
 = (dx , y  dx ) = (1 , y  ) dx ,
dX

where prime denotes x derivative. The length ds of this piece of string is



ds = |dX|
 = 1 + y 2 dx . (2)

We now find

∂X 
dx ∂ X 1
= = ( 1 , y ) , (3)
∂s ds ∂x 1+y 2

and we note that


∂X 
= ( 0 , ẏ ) . (4)
∂t
2
Using the formula for v⊥ from the notes
  2    2
2 ∂X ∂X ∂X
v⊥ = − · ,
∂t ∂t ∂s

and evaluating the dot products using (3) and (4) we get
 2
2 2 ẏy  ẏ 2 y 2 ẏ 2
v⊥ = ẏ −  = ẏ 2 − = . (5)
1 + y 2 1 + y 2 1 + y 2

Finally, using (2) and (5), we find that the string Lagrangian becomes
 "  a  
2
v⊥ 1 ẏ 2
L = −T0 ds 1 − 2 = −T0 dx 1 + y 2 1 − 2
c 0 c 1 + y 2
 a "
ẏ 2
= −T0 dx 1 + y 2 − 2 .
0 c

1
7.5 Planar motion for an open string with fixed endpoints.

(a) We check first that |F  (u)|2 = 1. We readily confirm that it holds:
πu  πu 
|F  (u)|2 = cos2 γ cos + sin2 γ cos = 1. (1)
σ1 σ1
The other condition is
F (u + 2σ1 ) − F (u) = (2a, 0) . (2)
Differentiating both sides with respect to u gives the periodicity condition

F  (u + 2σ) = F  (u) ,

which holds since F  (u) is a function of cos πu


σ1
only. In full detail, condition (2) requires
 2σ1
du F  (u) = (2a, 0), (3)
0

which implies a relation among the constants a, γ, and σ1 , to be discussed in part (d).
(b) We have  
  (0, σ) = dx dy 1  
X , = F (σ) + F  (−σ) = F  (σ) ,
dσ dσ 2
where we used F  (u) = F  (−u). This gives

dy πσ 
= sin γ cos .
dσ σ1
Below we plot dy/dσ for γ = π/10 and σ1 = π.
dy
dm

0.3

0.2

0.1

m
0.5 1 1.5 2 2.5 3

-0.1

-0.2

-0.3

Integrating dy/dσ gives y(σ), as shown in the following figure:

1
y

0.3

0.25

0.2

0.15

0.1

0.05

m
0.5 1 1.5 2 2.5 3

(c) We have
#

$

  (t, 0) = 1 F  (ct + 0) + F  (ct − 0) = F  (ct) =
X cos γ cos
πct
, sin γ cos
πct
.
2 σ1 σ1

Note that X   (t, 0) is just the unit tangent vector along the path of the string at its left
endpoint. The angle θ between the tangent vector and horizontal is given by
πct
θ(t) = γ cos ,
σ1
so that θ(t = 0) = γ. Thus γ is the initial and maximum angle of the end of the string.
(d) The first component of the condition in (3) gives
 2σ1
du Fx (u) = 2a.
0

where F  = (Fx , Fy ). The second component of equation (3) is automatically satisfied (check
this!). More explicitly, the above equation gives
 2σ1

πu
du cos γ cos = 2a .
0 σ1
Letting πu/σ1 = x, we rewrite the above equation as
 2π
a 1
= dx cos(γ cos x) . (4)
σ1 2π 0
For γ  1, we expand the integrand in powers of γ:
 2π 
a 1 γ2 γ2
= dx 1 − cos2 x + O(γ 4 ) = 1 − + O(γ 4 ) .
σ1 2π 0 2 4
This is the approximate relation between a, σ1 and γ, for small γ.
1
( 2π
(e) With the integral representation J0 (x) = 2π 0
dx cos(γ cos x) for the Bessel function,
we immediately recognize that (4) is just equivalent to a/σ1 = J0 (γ).

2
7.6 Planar motion (continued) and the formation of cusps.
(a) We have that
  
  ∂X 1  
X (t, σ) = = F (ct + σ) + F (ct − σ) ,
 (1)
∂σ 2
and 


πu πu
F  (u) = cos γ cos , sin γ cos .
σ1 σ1
Consider the x component of equation (1):

 x = 1 cos γ cos π(ct + σ) + 1 cos γ cos π(ct − σ)


X
2 σ1 2 σ1
It is convenient to define
π(ct ± σ) πct πσ πct πσ
γ cos ≡ β ∓ α , so β = γ cos cos , α = γ sin sin . (2)
σ1 σ1 σ1 σ1 σ1
With this notation,
& '
 x = 1 cos(β − α) + cos(β + α) = cos α cos β .
X
2
Similarly,
& '
X y = 1 sin(β − α) + sin(β + α) = cos α sin β .
2
Assembling together these results,
   

 (t, σ) = cos α cos β, sin β = cos γ sin πct πσ
X sin cos β, sin β . (3)
σ1 σ1
Note that when ct = σ1 /2, we obtain β = 0 so that
   
  π πσ πσ
X = cos γ sin sin (1, 0) = cos γ sin (1, 0).
2 σ1 σ1
  points along the x axis for all σ, the string is horizontal.
Since X

(b) We have

1 ∂X 1   
= F (ct + σ) − F  (ct − σ) .
c ∂t 2
  . Thus we get
There is only one sign difference between this expression and that for X
immediately that
x
1 ∂X 1 
= cos(β − α) − cos(β + α) = sin α sin β ,
c ∂t 2
1
and similarly,
1 ∂Xy 1 
= sin(β − α) − sin(β + α) = − cos β sin α .
c ∂t 2
Together, the above equations give

1 ∂X 
= sin α sin β, − cos β .
c ∂t
It follows that     
 1 ∂X
  
  = | sin α| =  sin γ sin πct sin πσ  ,
 c ∂t   σ1 σ1 
which is what we wanted to show.
At t = 0 the velocity is clearly zero. When γ < π/2, we have
π πct πσ π
− < γ sin sin =α< ,
2 σ1 σ1 2
so | sin α| < 1 and the velocity of any point on the string at any time is strictly smaller
than c. For γ = π/2 and ct = σ1 /2 we have
  
   π
 1 ∂X πσ


   
 c ∂t  = sin 2 sin σ1  .

At the midpoint of the string, σ = σ1 /2 this gives | 1c ∂∂tX | = | sin π2 | = 1.





(c) Setting γ = 2(π/2) and ct = σ1 /4 we have
    
   √ π
 1 ∂X π πσ  

π πσ


    =  sin .
 c ∂t  =  sin 2 sin sin
2 4 σ1   2
sin
σ1 
This is the same formula that we found for the case γ = π/2 and ct = σ1 /2, so we again find
that only one point, σ = σ1 /2 has reached the speed of light.

Setting γ = 2(π/2) and ct = σ1 /3 we have
  
   √ π   " 
 1 ∂X π πσ   3 π πσ 
    = sin .
 c ∂t  = sin 2 sin sin
2 3 σ1   2 2
sin
σ1 
This equals one when "
πσ 2
sin = ,
σ1 3
which occurs at two values σl , σr of σ:
" "
πσl −1 2 πσr 2
= sin and = π − sin−1 ,
σ1 3 σ1 3

2
To see that these points are cusps, we examine X  near σ = σl using equation (3):
# " $
σ  π 3 πσ 
 1
 t = , σ = cos
X sin cos β, sin β . (4)
3c 2 2 σ1

Here
πct πσ π 1 πσ
β = γ cos cos = √ cos .
σ1 σ1 2 2 σ1
We note that at σ = σl the value βl of β is completely regular:
π 1
βl = √ .
2 6
We can therefore set β = βl in equation (4). The prefactor on the right-hand side of (4) is
more delicate. Putting σ = σl + δσ we have to lowest order in δσ that
"
π(σl + δσ) 2 1 π
sin = +√ δσ + O(δσ 2 ) .
σ1 3 σ
3 1
Therefore, "
π 3 π(σl + δσ) π 1 π
sin = +√ δσ + O(δσ 2 ) ,
2 2 σ1 2 2 σ1
which implies that
 " 
π 3 π(σl + δσ) 1 π
cos sin = −√ δσ + O(δσ 2 ).
2 2 σ1 2 σ1
Finally, back in (4), we have
  
X  t = σ1 , σl + δσ = δσ − √1 π cos βl , sin βl = δσ (−Vl ) .
3c 2 σ1
Near σl the tangent vector X   to the string is equal to a constant vector Vl times δσ. For σ
slightly smaller than σ1 and growing we are approaching the cusp, δσ < 0, and the tangent
points along +Vl . For σ slightly larger than σ1 we are moving away from the cusp, δσ > 0
and the tangent points along (−Vl ). The tangent to the string reverses direction at σl , and
it does so discontinuously. This is a cusp. Note that βl is simply the angle that the cusp
makes with respect to the horizontal.

(d) Setting a = 1, requires


1
σ1 = √   10.154868.
π 2
J0 2

For these values of σ1 and γ we plot the strings that occur for ct = 0, σ1 /4, and σ1 /3,

3
4

-1 -0.5 0.5 1 1.5 2

4
8.2 A generalization for charges and a special case for currents.
(a) With several coordinates q a and symmetry parameters i , the variations are written as

δq a = i hai (q; t) . (1)

We now claim that the charges are


∂L a
i Qi = δq . (2)
∂ q̇ a
This time the Euler-Lagrange and invariance equations are
 
d ∂L ∂L ∂L a ∂L d
− a = 0, δq + a (δq a ) = 0 . (3)
dt ∂ q̇ a ∂q ∂q a ∂ q̇ dt

We verify conservation as in the book: we compute the time derivative of the charges in (2)
and then use equations (3) to find that
 
i dQi d ∂L ∂L d ∂L ∂L d
 = δq a + a (δq a ) = a δq a + a (δq a ) = 0. (4)
dt dt ∂ q̇ a ∂ q̇ dt ∂q ∂ q̇ dt

Since the i are independent parameters, we conclude that


dQi
= 0. (5)
dt

(b) The index α can only take the value zero – there is just ξ 0 , which plays the role of time
in this world. The field φa is only a function of time. The action becomes

S = dξ 0 L(φa , ∂0 φa ) . (6)

The quoted equations become


∂L
i ji0 ≡ δφa , ∂0 ji0 = 0 , Qi = ji0 , (7)
∂(∂0 φa )

where the last equation is “obtained” because there are no spatial coordinates to integrate
over; since there is no space, the charge density ji0 is really just a charge. The second equation
in (7) is charge conservation, in correspondence with (5). The first equation in (7), takes
the form of (2), once we note that ∂0 is really a dot derivative, and that q a and φa are the
labels of the corresponding dynamical variables, both of which depend only on time.

1
8.4 Simple estimates regarding α , T0 , and s .
We know that
1 √
T0 = and s = ~c α .
2πα ~c
Moreover, for quick estimates we can use

~c = 198 MeV × 10−15 m = 1.98 × 10−14 GeV · cm ,

together with
GeV
= 1.6 × 10−8 N  1.63 × 10−12 ton .
cm

(a) For α  0.95 GeV−2 , we get

1 GeV GeV
T0 =  8.47 × 1012  13.8 ton .
2 · 3.14 · 0.95 · 1.98 × 10−14 cm cm
The string length is then

s = 1.98 × 10−14 × 0.95 cm = 1.93 × 10−14 cm .

(b) With s ∼ 10−30 cm, we have


 2
s 10−60

α = = GeV−2 = 2.55 × 10−33 GeV−2 .
~c 3.92 × 10−28

Since the string tension is inversely proportional to α we have


 
0.95 GeV−2
T0 = 13.8 ton × = 5.14 × 1033 ton .
2.55 × 10−33 GeV−2
This is a fantastically large force. The gravitational force which keeps the Earth in orbit
around the sun is about 3.6 × 1018 ton. The above string tension is about 2.6 million times
larger than the “weight” of the sun (the mass of the sun is  2 × 1030 kg).

1
8.5 Angular momentum of Kasey’s jumping rope.
From the solution to the jumping rope problem we have
L0 πσ
(X1 , X2 ) = tan γ sin (cos ωt , sin ωt ) , (1)
π σ1
and, moreover,
πc E L0
ω= cos γ , σ1 = = . (2)
L0 T0 cos γ
The momentum density is computed using (1) and (2)


 τ = T0 ∂ X
P → (P1τ , P2τ ) =
T0
sin γ sin
πσ
(− sin ωt , cos ωt ) (3)
c2 ∂t c σ1
We compute the z-component Jz of the angular momentum of the string, measured with
respect to the origin:  σ1
Jz = M12 = dσ (X1 P2τ − X2 P1τ ) . (4)
0

Using (1) and (3) we find


  2
L0 T0 σ1
2πσ L0 T0 σ1 sin2 γ T0 L0
Jz = tan γ sin γ dσ sin = = sin2 γ . (5)
π c 0 σ1 2πc cos γ 2πc cos γ

We express this result in terms of energy using (2):

T0 E 2
Jz = sin2 γ .
2πc T02

Equivalently,
Jz 1
= E 2 sin2 γ = sin2 γ α E 2 . (6)
~ 2π T0 ~c
The above Jz /~ is smaller than α E 2 , the value of the angular momentum of a rigidly rotating
straight open string of the same energy.

1
8.7 Generalizing the construction of conserved currents.
To show that the current is conserved we take its divergence
 
∂L ∂L & '
i α
 ∂α ji = ∂α a
δφa
+ a
∂α δφa
− i ∂α Λαi (1)
∂(∂α φ ) ∂(∂α φ )
We now use the equations of motion that follow from the Lagrangian
∂L ∂L
∂α a
=
∂(∂α φ ) ∂φa
to simplify the first term of (1). We then find
∂L a ∂L & '
i ∂α jiα = a
δφ + a
∂α δφa − i ∂α Λαi .
∂φ ∂(∂α φ )
The first two terms on the right-hand side are simply the variation δL of the Lagrangian
under the symmetry transformation. By equation (1) of the problem statement we have
i ∂α jiα = δL − i ∂α Λαi = 0 .
Since the parameters i are independent constants, the currents ji are conserved.
We now consider a L with no dependence on the coordinates ξ β . Under a change φa →
φa + β ∂β φa , the variation of the Lagrangian is given by
 
∂L β a ∂L β a β ∂L a ∂L a
δL =  ∂β φ +  ∂α ∂β φ =  ∂β φ + ∂β (∂α φ ) .
∂φa ∂(∂α φa ) ∂φa ∂(∂α φa )
We therefore recognize that
∂L ∂
= β (β L),
δL = β
β
∂ξ ∂ξ
β
where we also used the constancy of the  . Relabelling indices
∂ α ∂ ∂
δL = ( L) = α (β δβα L) = α (β Λαβ ) with Λαβ = δβα L .
∂ξ α ∂ξ ∂ξ
We then have, using equation (2) of the problem statement
∂L ∂L
β jβα = β ∂β φa − β δβα L → jβα = ∂β φa − δβα L ,
∂(∂α φa ) ∂(∂α φa )
as desired. For α = β = 0
∂L
j00 = ∂0 φa − L = H .
∂(∂0 φa )
where H is the Hamiltonian density of the system. This is clear because the first term on the
right-hand side is the product of the canonical momentum associated with the coordinate
φa times the velocity of the coordinate φa .

1
Problem 10.3. Light-cone components of Lorentz tensors.

(a) Since every component of A equals the corresponding component of B, we have that
1 1
A± = √ (A0 ± A1 ) = √ (B 0 ± B 1 ) = B ±
2 2
The fact that AI = B I follows from Aµ = B µ setting µ = I.

(b) The fact that our definition works for tensors of the form Rµν = Aµ B ν implies that
1 1
R++ = A+ B + = √ (A0 + A1 ) √ (B 0 + B 1 )
2 2
1 0 0
= (A B + A0 B 1 + A1 B 0 + A1 B 1 )
2
1 00
= (R + R01 + R10 + R11 ) (1)
2
Similarly,
1 1
R+− = A+ B − = √ (A0 + A1 ) √ (B 0 − B 1 )
2 2
1 0 0
= (A B − A0 B 1 + A1 B 0 − A1 B 1 )
2
1 00
= (R − R01 + R10 − R11 ) (2)
2

1 1
R−+ = A− B + = √ (A0 − A1 ) √ (B 0 + B 1 )
2 2
1 0 0
= (A B + A0 B 1 − A1 B 0 − A1 B 1 )
2
1 00
= (R + R01 − R10 − R11 ) (3)
2

1 1
R−− = A− B − = √ (A0 − A1 ) √ (B 0 − B 1 )
2 2
1
= (A0 B 0 − A0 B 1 − A1 B 0 + A1 B 1 )
2
1 00
= (R − R01 − R10 + R11 ) (4)
2

1
In general we may write the change of coordinates of a vector field as Aµ̂ = Mνµ̂ Aν where
ˆ ˆ
the hatted index is a light-cone index and M0± = √12 and M1± = ± √12 and MJI = δJI . Since
each index of a tensor transforms like a vector, we can write

Rµ̂ν̂ = Mµµ̂ Mνν̂ Rµν

This immediately implies that if Rµν = S µν that

Rµ̂ν̂ = Mµµ̂ Mνν̂ Rµν = Mµµ̂ Mνν̂ S µν = S µ̂ν̂

(c) We have
1 1
η ++ = (η 00 + η 01 + η 10 + η 11 ) = (−1 + 0 + 0 + 1) = 0
2 2
1 1
η +− = (η 00 − η 01 + η 10 − η 11 ) = (−1 − 0 + 0 − 1) = −1
2 2
1 00 1
η = (η + η 01 − η 10 − η 11 ) =
−+
(−1 + 0 − 0 − 1) = −1
2 2
1 00 1
η = (η − η 01 − η 10 + η 11 ) =
−−
(−1 − 0 − 0 + 1) = 0
2 2

(d) We find
1
F +− = (F 00 − F 01 + F 10 − F 11 ) = F 10 = −E x .
2
Similarly
1 1
F +I = √ (F 0I + F 1I ) = √ (E I + 1IJ BJ )
2 2
and
1 1
F −I = √ (F 0I − F 1I ) = √ (E I − 1IJ BJ )
2 2
Finally, there is only one non-zero component of F IJ given by F 23 = −F 32 = Bx .

2
Problem 10.7. Massive vector field.

We have been given the action


 1 
1 1 1
S = dD x − ∂µ Aν ∂ µ Aν + ∂ν Aµ ∂ µ Aν − m2 Aµ Aµ − ∂µ φ ∂ µ φ − (∂ · A)mφ . (1)
2 2 2 2

The first two terms can be written as − 12 (∂µ Aν − ∂ν Aµ ) ∂ µ Aν = − 14 Fµν F µν , so the action is
 1 
1 1
S = dD x − Fµν F µν − m2 Aµ Aµ − ∂µ φ ∂ µ φ − (∂ · A)mφ . (2)
4 2 2

(a) We consider the candidate gauge transformations

δAµ = ∂µ  , δφ = b m  .

The variation of the action is computed using (2), in which the first term inside the paren-
theses is gauge invariant. We find
 
δS = dD x −m2 ∂µ  Aµ − b m(∂µ ) ∂ µ φ − (∂ 2 )mφ − b(∂ · A)m2  .

We integrate by parts the derivatives that appear in the underlined terms to find:
 
δS = dD x m2  ∂ · A + b m(∂ 2 ) φ − (∂ 2 )mφ − b(∂ · A)m2  .

To get δS = 0 we need b = 1. So the gauge transformations are

δAµ = ∂µ  , δφ = m  . (3)

(b) The variation of the action (2) gives:


 
δS = dD x ∂ν (δAµ ) F µν − m2 (δAµ )Aµ − ∂µ (δφ) ∂ µ φ − (∂ µ δAµ )mφ − (∂ · A)mδφ .

We integrate by parts all derivatives that act on variations:


 
δS = dD x −δAµ ∂ν F µν − δAµ m2 Aµ + δAµ m∂ µ φ + δφ ∂ 2 φ − δφ m ∂ · A .

The equations of motion are therefore

−∂ν F µν − m2 Aµ + m∂ µ φ = 0 , ∂ 2 φ − m∂ · A = 0 . (4)

1
One readily verifies that the equations of motion are gauge invariant. This is good, but not
strictly necessary; it suffices for consistency if the variations vanish when the equations of
motion hold.
(c) The gauge transformation δφ = m allows us to gauge away φ: setting  = −φ/m, one
finds φ → φ + m = 0. With φ = 0 the equations of motion (4) become:

−∂ν F µν − m2 Aµ = 0 , ∂ · A = 0.

More explicitly:
−∂ µ (∂ · A) + ∂ 2 Aµ − m2 Aµ = 0 , ∂ · A = 0.
This is equivalent to
∂ 2 Aµ − m2 Aµ = 0 , ∂ · A = 0. (5)
The above are the equations of motion in the gauge where the scalar field has disappeared.
(d) In momentum space equations (5) become

(p2 + m2 )Aµ = 0 , p · A = 0. (6)

When p2 = −m2 the first equation gives Aµ (p) = 0. Consider now p2 = −m2 , in which case
the first equation is satisfied for arbitrary Aµ (p). Using a suitable Lorentz transformation
any timelike vector pµ can be rotated into a vector with only a time component. We can
thus assume that pµ = (m, 0) since this ansatz satisfies p2 = −m2 . The second equation
gives p · A = p0 A0 + p · A
 = −mA0 = 0. So we conclude that A0 = 0. The remaining D − 1
components A with i = 1, . . . D − 1 remain unconstrained. We have shown that the massive
i

vector field has D − 1 degrees of freedom.

2
11.3 Classical dynamics in Hamiltonian language.

As stated, to derive the classical equations of motion in Hamiltonian language we vary


 tf   tf dq dδq ∂H ∂H 
δ dt pq̇ − H(p, q; t) = dt δp +p − δp − δq ,
ti ti dt dt ∂p ∂q

with independent variations δq(t) and δp(t) that vanish at ti and tf . Replacing

dδq d(δq p) dp
p = − δq ,
dt dt dt
and integrating the total derivative, the variation becomes
!tf  tf dq ∂H ! dp ∂H !
δq(t)p(t) + dt δp − − δq + .
ti ti dt ∂p dt ∂q

The terms at initial and final times vanish because of the boundary conditions on the vari-
ations, and the condition that the variations vanish for otherwise arbitrary δq(t) and δp(t)
give
dq ∂H dp ∂H
= , and =− .
dt ∂p dt ∂q
Using these equations we readily find that
dv ∂v ∂v dq ∂v dp ∂v ∂v ∂H ∂v ∂H ∂v
= + + = + − = + {v , H} .
dt ∂t ∂q dt ∂p dt ∂t ∂q ∂p ∂p ∂q ∂t
This is what we wanted to show. The Hamiltonian generates time evolution in classical
mechanics via the Poisson brackets.

1
12.4 Analytic continuation of the zeta function.
Replacing the integration variable t with nt in the definition of Γ(s) gives
 ∞  ∞
−nt s−1 s−1
Γ(s) = d(nt)e n t =n s
dte−nt ts−1 .
0 0


∞ ∞ 
 ∞  ∞ 

−s −nt s−1
→ Γ(s)ζ(s) = Γ(s)n = dte t = s−1
dt t e−nt .
n=1 n=1 0 0 n=1
)∞ e−t 1
The geometric series gives n=1 e−nt = 1−e−t
= et −1
, so we get
 ∞
ts−1
Γ(s)ζ(s) = dt . (1)
0 et − 1

We now consider the expansion of 1/(et − 1) around t = 0. Expanding the exponential,

1 1 1 1 1 1 t
= = · = − + + O(t2 ).
et −1 t+ t2
2
+ t3
3!
+ O(t4 ) t t2
t 1 + 2 + 6 + O(t3 ) t 2 12

1
where we used 1+x
= 1 − x + x2 − . . .. We now write Γ(s)ζ(s) in (1) as
 1  ∞
ts−1 ts−1
Γ(s)ζ(s) = dt t + dt . (2)
0 e −1 1 et − 1

Only the first integral on the right hand side may diverge near t = 0. We rewrite it as
 1  1    1  
ts−1 1 1 1 t 1 1 t
dt t = dt ts−1
− + − + dt ts−1
− +
0 e −1 0 et − 1 t 2 12 0 t 2 12
 1  
1 1 1 t 1 1 1
= dt ts−1 t − + − + − + .
0 e − 1 t 2 12 s − 1 2s 12(s + 1)

Hence, as quoted in the problem statement, (2) becomes


 1    ∞
1 1 1 t 1 1 1 ts−1
Γ(s)ζ(s) = s−1
dt t − + − + − + + dt . (3)
0 et − 1 t 2 12 s − 1 2s 12(s + 1) 1 et − 1

The last integral is convergent for all s. Since (the terms enclosed by the large parentheses
1
are of order t2 , the first integral is of the form 0 dtts+1 and converges for (s) > −2.

We recall that Γ(s) has a simple pole at s = 0 with residue one and a simple pole at
s = −1 with residue −1 [ the general result is that the residue at s = −n < 0, with n integer

1
is (−1)n /n!]. Equation (3) shows that Γ(s)ζ(s) also has simple poles at s = 0 and s = −1.
We therefore deduce that ζ(s) is regular for s = 0, −1, and

Res s=−n (Γ(s)ζ(s)) = (Res s=−n Γ(s)) ζ(−n) , n = 0, 1 .

Reading off the residue at s = −1 from (3) we get the celebrated result:


1 1
= (−1) ζ(−1) → ζ(−1) = − .
12 12
Similarly, reading off the residue at s = 0 we find that
1 1
− = (1) ζ(0) → ζ(0) = − .
2 2
It is interesting to note in (3) that Γ(s)ζ(s) has a simple pole at s = 1. Since Γ(s) has no
pole at s = 1, ζ(s) does. The pole at s = 1, with residue one, is in fact the only singularity
of the zeta function ζ(s).

2
13.4 L⊥ ⊥
0 − L̄0 as world-sheet momentum. (Continued)

(b) With δX I =  ∂σ X I we find:


1 I 2 I
 1
I 2 I 2

δL =  ∂τ X I
∂τ ∂σ X I
− ∂σ X ∂ X =  ∂ σ (∂τ X ) − (∂σ X ) .
2πα σ
4πα
With a little rearrangement we write
1 I 2 I 2
!
δL = ∂σ  (∂ τ X ) − (∂σ X ) .
4πα
Since the variation of L is a total derivative, the σ-translation δX I =  ∂σ X I is a symmetry.
In the notation of equation (1) of Problem 8.7, and letting (ξ 0 , ξ 1 ) = (τ, σ) we write

∂ 0 1 1 I 2 I 2

δL = α ( Λα ) with Λ = 0 and Λ =  (∂τ X ) − (∂σ X ) . (1)
∂ξ 4πα

The conserved current is (equation (2) in Problem 8.7):

∂L
 jα = δX I −  Λα ,
∂(∂α X I )

so the charge density j 0 is


∂L 1 
j0 = I
∂σ X I − Λ0 = 
Ẋ I X I .
∂(∂0 X ) 2πα

Note that Λα , computed to prove that we had a symmetry, does not enter into the charge
density since Λ0 = 0. The conserved charge Q is
 2π  2π
0 1 
Q= dσ j = 
dσ Ẋ I X I = −(L⊥ ⊥
0 − L̄0 ) ,
0 2πα 0

comparing with the result stated in part (a) of the problem.

Вам также может понравиться