Tutorial Exercises: Covariance of the Lagrangian Formalism

1. (a) Consider again the four Lagrangians L(t, x, y, x, y ) from a previous tutorial set. i. ii. iii. iv. x 2 + y 2 k2 (x + y )2 x 2 + y 2 k2 (x y )2 x 2 + y 2 + k2 (x + y )2 2 x y 2 + k2 (x + y )2

Consider the point transformation dened by x = u + v and y = u v . Rewrite each of the four Lagrangians in terms of the new variables. (b) Notice that none of the original coordinates were ignorable. Do any of the new Lagrangians have ignorable coordinates? (c) Solve the equations of motion taking full advantage of any ignorable coordinates. Look back at your earlier tutorial solution of the same problems and compare the amount of work required. Solution Taking a time derivative gives x =u +v and y =u v . The four Lagrangians become (b) 2u 2 + 2v 2 4k2 v 2 (d) 4u v + 4k2 u2 We look at them all at the same time to compare what happens. In the rst and third examples v is ignorable, and in both these cases the rst integral or constant of the motion is 4v = C1 , 1 which can be solved straight away to give v = 4 C1 t + C2 . In the second example u is ignorable, and the rst integral or constant of the motion is 4u = C1 , which can be solved straight away 1 to give u = 4 C1 t + C2 . In the fourth example v is ignorable, but the rst integral is actually 4u = C1 , which can be solved straight away to give u = 1 4 C1 t + C2 . The full Euler or Lagrange equation has to be derived for the non-ignorable co-ordinate. These are in the original order above (a) 4 u + 8k2 u = 0 (b) 4 v + 8k2 v = 0 (d) 4 v + 8k2 u = 0 (c) 4 u 8k2 u = 0 (a) 2u 2 + 2v 2 4k2 u2 (c) 2u 2 + 2v 2 + 4k2 u2

The rst and second equations have the general solutionC3 sin 2t + C4 cos 2t. The third equation has the general solution C3 exp( 2t) + C4 exp( 2t). The last equation simplies when the relevant expression for u is inserted and becomes v + 1 1 1 2 3 2 2 2 2k ( 4 C1 t+C2 ) = 0. Rearranging and integrating twice gives v = 12 C1 k t 4 C2 k t +C3 t+C4 . Its not hard to show all these solutions are equivalent to the ones on the earlier solution sheet. Finding them here was a lot easier!

2. Consider the transformation x = r cos , y = r sin , z = r2.

This transformation enforces a constraint between x, y and z . Eliminate r and to nd the constraint (i.e. and equation for the surface). What surface is described by this constraint? 60

Apply the transformation to 1 2 + y 2 + z 2 ) mgz L = m(x 2 Solution To nd the surface in cartesian coordinates, eliminate by taking x2 + y 2 and then eliminate r . Thus, x2 + y 2 = z is the surface, which is a paraboloid. First calculate the derivatives x =r cos r sin , Then substitute into L and simplify 1 2 + 4r 2 r L = m(r 2 + r2 2 ) mgr 2 2 3. Starting from the Kepler problem in cylindrical polar coordinates, namely, 1 2 + z 2 + 2 2) + L = m( 2 transform to parabolic coordinates dened by = Solution From the transformation equations 2 + z 2 , y =r sin + r cos , z = 2r r.

, = and z = 1 2 ( ).

1 1 2 + z 2 = + ( )2 = ( + )2 4 4 which means 2 + z 2 = Also z = = Substituting into the Lagrangian gives L = = 1 1 )2 + 2 2 + 1 ( m{ ( + )2 }) V, 2 4 4 + 1 2 ) + 2 } + 2 . m{ ( 2 + 2 4 + 1 ( ) 2 1 ) ( + 2 1 ( + ) 2

4. Covert intelligence operative Sydney Bristow discovers an old document written by the inventive genius Milo Rimbaldi that contains the following Lagrangian L(t, e, s, e, s ) = s 2 4e 2 + s3 s(1 + e2 )2 8ge s(1 + e2 )

where e is called elevation gradient, and s is called signal strength. The only other clue as to what physical system it describes is that g stands for gravity. 61

(a) If you feel algebraically macho, determine the Lagrange equations directly from this Lagrangian and solve them... OR ...nd a change of variables that makes the Lagrangian simpler. Hint: If you saw 1 + e2 in the denominator of an integral what substitution would you try? Remember e is called a gradient! Also try to get rid of the over s. (b) Keep applying more transformations if you can continue to simplify the Lagrangian. (c) Interpret your simplied Lagrangian and give the solution to the Lagrange equations. (d) Now go back and justify the naming of the two original co-ordinates. Solution The rst transformation to try is e = tan . Then = (1 + e2 ) e = sec2 Substituting gives 2 2 8g tan 8g 4 4 s sin L(t, , s, , ) = s 3 s 2 + = s 3 s 2 + s s s sec s Then try s = u2 to get rid of the square-root. Thus s = 2uu and 2 8g 2 8g 4 4 u L(t, , u, , ) = u6 4u2 u 2 + 2 sin = 4u4 u 2 + 2 sin u u u u Then try u = 1/r to simplify the terms, so u = r/r 2 and r 2 8gr sin = 4(r 2 ) 8gr sin L(t, , r, , ) = 4r 4 r 4 r 2 + 4r 2 2 + r2 Then L looks like something in polar coordinates, so try x = r cos , and y = r sin to make it look like cartesian coordinates. L(t, x, y, x, y ) = 4(x 2 + y 2 ) 8gy It is now obvious that this Lagrangian just describes a particle of mass m = 8 falling under gravity. The parameter e is the slope or gradient of the line-of-sight from the origin. The parameter s goes like inverse distance squared, so if we were tracking the particle with some radar or wave-based method, it would be the signal strength. This question is meant to drive home the realisation that good choices of coordinates makes an enormous dierence in both our ability to analyse the system but also to interpret the results. 5. Let F (x, y, z ) = xa y b z c . What is the degree of F . Show that x F F F +y +z = (a + b + c)F x y z

Note that the formula works even if a + b + c is not an integer. Check Eulers Homogeneous function theorem for the function F (x, y, z ) = 62 x3 + yz 2 + xyz

Solution The degree of F is a + b + c. Now, F = axa1 y b z c , x The result follows easily from there. The quantity inside the square-root has degree 3, so the second function has degree 3/2. Now, 3x2 + yz F = , x 2 x3 + yz 2 + xyz so x F = y 2 z 2 + xz x3 + yz 2 + xyz , 2yz + xy F = , 3 z 2 x + yz 2 + xyz F = bxa y b1 z c , y F = cxa y b z c1 z

F F F 3x3 + xyz + yz 2 + xyz + 2yz 2 + xyz 3 +y +z = = F x y z 2 2 x3 + yz 2 + xyz

6. The Lagrangian for a particle in an electromagnetic eld is 1 r eV (r, t) + er A(r, t). L = mr 2 Decide whether the Lagrangian can be split into homogeneous parts and the degree of each. Thus, show that the energy of a particle in an electromagnetic eld does not depend on the magnetic part of the eld.
1 mr r is homogeneous of degree 2 in the velocities, so it should appear Solution The term 2 unchanged in the energy.

The term eV (r, t) is homogeneous of degree 0 in the velocities, so it should appear with the opposite sign in the energy. The term er A(r, t) is homogeneous of degree 1 in the velocities, so it should not appear at all in the energy. Thus E= 7. Starting with q i =
k =1

1 mr r + eV (r, t). 2
m

qi qi Qk + Qk t

j . Note that the old coordinates qi do not depend on any of the new velocities calculate q i / Q j. Q are treated as independent quantities. What In the new coordinate system, all the velocities Q k / Q j when k = j and when k = j ? does this imply about Q Hence show that q i qi . = Qj Qj

Solution j , the required partial derivative of q Since the coordinates qi do not depend on the velocities Q i can be written as m k qi Q q i = . j j Qk Q Q k =1

63

The independence of the velocities from each other in the new coordinate system means that k / Q j = ij Q Thus, qi qi q i kj = . = Q Q Qj j k k =1 8. Starting with q i =
k =1 m m

qi qi Qk + Qk t

are treated as indecalculate q i /Qj . Note, in the new coordinate system, the Q and the Q pendent quantities. Now using the chain rule d = + dt t and the denition of Jij calculate
d dt Jij . m k =1

k Q Qk

Show that the two results are the same. [What technical condition do you need about mixed second order partial derivatives.] Solution First q i = Qj Also d d Jij = dt dt qi Qj =
k =1 m k =1

2 qi 2 qi Qk + Qj Qk Qj t
m

Qk

qi Qj

k + Q t

qi Qj

the two expressions are equal if we assume that mixed second-order partial derivatives commute.

64