University of California at San Diego Department of Physics Prof.
John McGreevy
General Relativity (225A) Fall 2013
Assignment 2 Solutions
Posted October 3, 2013 Due Monday, October 15, 2013
1. Special relativity reminders
This problem is optional. Do it if you have to think about it or if you find it a useful
reminder.
dxi
(a) Derive the special relativity addition law for coordinate velocities dt
using the
Lorentz transformation x = x .
The infinitesimal coordinate lengths are related by
dx = dx
where is a boost; wlog take the relative velocity between the frames in the x
direction. And for simplicity lets take v = dx/dt in this direction, too.
dx
dx dx + dt + v+
v = = = dt dx = .
dt dt + dx 1 + dt 1 + v
In this way of doing it we didnt need to worry about the fact that along the
trajectory there is a relation between dx and dt; if we wanted to do it by
d x d
v = dx = 1 0 x dx
dt dx
we would have to worry about this.
(b) Beginning with the assumption that the speed of light is constant, derive the time
dilation effect as follows. Build a clock by bouncing a light ray between mirrors.
Watch the clock tick in a frame which is boosted transverse to the motion of the
light.
Compare Bobs experience in the rest frame of the clock (call the distance between
the mirrors ; the time for light to go from one mirror to the other is t0 = /c):
1
with that of Alice who sees the clock move to the right at speed u:
2. Electric-magnetic (Poincare) duality
The vacuum Maxwell equations
0 = F , 0 = F
~ B)
are invariant under the exchange of the electric and magnetic fields (E, ~ 7 (B,
~ E).
~
1
This is called electric-magnetic duality.
Show that the duality transformation can be written in a covariant manner as
1
F 7 (?F ) F .
2
If I identify Fi0 = Ei and Fij = ijk B k (so B i = 12 ijk Fjk ) as I did in lecture, and raise
spatial indices with ij , so E i = ij Ej , then I should have an extra minus sign (relative
to the original problem statement):
1 1
(?F )ij = ij F = ijk0 F k0 F 0k = ijk F k0 = +ijk (Fk0 ) = +ijk Ek
2 2
(so Bk ?Bk = +Ek ) and
1 1 1 1 1 1
(?F )i0 = i0 F = i0jk F jk = ijk F jk = ijk jkl Bl = ijk ljk Bl = 2Bi = Bi
2 2 2 2 2 2
(so Ek ?Ek = +Bk ). Notice that (E, B) (E, B) is also a symmetry of
Maxs equations (namely charge conjugation) so the overall sign of the transforma-
tion shouldnt trouble us too much both are symmetries.
Verify that the (vacuum) Maxwell equations are invariant under this replacement.
The Bianchi identity becomes
1
0 = (?F ) = F
2
1
In the presence of electric charges, the duality transformation is only a symmetry if we allow for magnetic
monopole charge; the duality transformation then must exchange electric charges and magnetic charges.
2
1
= F
2| {z}
=2( ())
1
= 2 F F = 2 F
(1)
2
which is the vacuum Maxwell equation. The vacuum Maxwell equation becomes
1
0 = (?F ) = F
2
which is the Bianchi identity.
Check that doing the duality map twice gives 1: ? (?F ) = F .
1 1
? (?F ) = (?F ) = F
2 4
But now we can use the identity
= a
where a a constant that I determine by tracing both sides with :
= a(4 1)
| {z }
=3!
so a = 2. Therefore:
1
? (?F ) = 2 (F F ) = F .
4
3. Show that E~2 B~ 2 and E~ B
~ are respectively scalar and pseudo-scalar under the
Lorentz group by expressing them in terms of F .
1
~2 B
~2
F F = E
2
is a scalar since all tensor indices are contracted with the Minkowski metric.
1 1 ~ B
~
F F = (?F ) F = E
8 4
is a pseudo scalar since some of the indices are contracted with the Levi-Civita symbol.
4. Lorentz transformations of the EM fields
This problem is to convince yourself that the formula we gave for the Lorentz trans-
formation of the Maxwell field strength:
F (x) 7 F (x) , F (x) = F (x) (2)
is correct, i.e. it is a Lorentz tensor.
3
(a) Consider the case where is a Lorentz boost in some direction. What trans-
formation laws does this imply for the components of the electric field along the
boost Ek ? And the component transverse to the boost E ?
This is a problem in unpacking, but an important one. Take the boost along x,
so that it takes the form of eqn (15) of the lecture notes. Then y and z indices
dont acquire matrices. For example, Ey = Fy0 becomes
Fy0 = y 0 F = yy 00 Fy0 + 0x Fyx = Ey vBz
|{z}
=1
which is the same (restoring a factor of c) as
~ 7 E ~
v
~ + B ~ .
E
c
The longitudinal bit of E is Ex = Fx0 which becomes
v2
x 0 0 x 2
Fx0 = x 0 F = x 0 Fx0 + x 0 F0x = 1 2 Fx0 = Fx0
c
~ are of the same form with E
it doesnt change. The results for B ~ B,
~ B~ E.
~
~ ~ ~v ~ ~k = B
~ k.
B = B E , B
c
In fact, we can deduce this immediately from the duality transformation.
(b) Describe a charge configuration which creates a constant electric field in some
region of space.
Parallel plates with opposite uniform charge!
(c) For the electric field case, boost the charge configuration along the direction of the
field and transverse to the direction of the field, and find the new fields that they
create. You may assume the experimental fact that electric charge is a scalar
quantity. Convince yourself that they agree with the answers from the fancy
formula (2), in the case where there is no magnetic field.
This is really two problems. If we boost a capacitor along the plates, the charge
density gets contracted along the boost, so it goes up by a factor of .
E = 4 = E 7 E .
If we boost transverse to the plates, the plates are closer together in the new
frame, but the charge is the same, so:
Ek 7 Ek .
4
(d) Admit to yourself that magnetic charge could exist. Use this fact and the idea in
problem 2 to figure out how Bk and B transform in the absence of an electric
field.
Just imagine a magnetic capacitor rather, just parallel sheets of uniform mag-
netic charge. This produces a uniform magnetic field in between. Never mind
what the plates are made of. You can get the Bk answer pretty easily by boosting
a solenoid, but B is harder to get that way.
(e) Consider current in a neutral wire made from negative charge per unit length
(electrons) moving at velocity vdrift superposed with a static positive charge
density. What field does this create? By boosting to the rest frame of the negative
charges, find the transformation rule for E , B when both are nonzero.
In the rest from of the electrons, the linear charge density of the negative charges
is dilated to = / ( = 12 2
) while that of the positive charges is
1vdrift /c
contracted to = +, since they now move with velocity vdrift . The magnetic
field There is therefore a net charge density, which is the sum of the two
v2
t = + + = 1 = 2 > 0.
c
This produces an electric field pointing away from the wire
0
~ = r 2t .
E
r
This is related to B~ = 2 by E~ = ~v B.
~ The magnetic field changes too: the
c2 r c
current density is 0 + + (vdrift ) = +vdrift , so the new magnetic field is B .
Combining this with the EM dual case of a magnetic current, we learn that
~ ~ ~v ~
~ ~ ~v ~
E = E + B B = B E
c c
as we found above.
5. Easy. Consider the worldline of a particle which sits at the origin for all time. In lecture
we showed that a tangent vector to this curve has negative proper length-squared,
dx dx
<0,
d d
where is an arbitrary parameter along the worldline. That is, the tangent vector is
timelike. How do I know that a Lorentz boost cannot take this vector to one which is
spacelike?
The proper length is Lorentz invariant.
6. Light-cone accounting
5
(a) Show that any timelike vector u for which u0 > 0 and u u = 1 is the four-
velocity of some worldline.
We may consider
1 = u u = (u0 )2 + ~u ~u
as a differential equation for a path x (s) whose instantaneous velocity is dx
ds
.
More explicitly, we can choose s = t so the proper length along a segment of the
path is
p d~x
d = dx dx = 1 ~v 2 dt, ~v = .
dt
Then the four-velocity of this worldline is
dx 1
u = = (1, ~v )
d 1v 2
which indeed has
1
u u = 2
1 + ~
v = 1
1 ~v 2
and u0 = 1 > 0. So the relevant path has
1v 2
d~x ~u
= ~v = 0
dt u
which we could regard as a differential equation for ~x(t), treating u as a constant.
OK hold on that sounded really complicated. What I mean is: you can choose
the path to be a straight line with constant 4-velocity u .
(b) Use the previous result to show that for any timelike vector v there is a Lorentz
frame in which v has zero spatial components.
Our goal is to find a Lorentz transformation which sets to zero the spatial compo-
nents of v. WLOG assume v 0 > 0, since we can reverse its sign by an improper
Lorentz transformation. Let u = vv v . This u satisfies the assumptions of the
previous result. Our statement now is just that every particle has a rest frame.
This answer is not as satisfying as I thought it would be. Here is a more direct
argument. First, WLOG, we can rotate the spatial components of v so that in
only points in the x direction (recall that a rotation is a special case of a Lorentz
transformation). So now the problem we are trying to solve is to find a D = 1 + 1
boost by velocity u such that
0 0
u v w
v = = (3)
u vx 0
given that
0 > v 2 = (v 0 )2 + ~v 2 = (v 0 )2 + vx2 = (w0 )2 . (4)
Lets just solve the bottom equation of (3) for u: it says
v 0 + uvx = 0
6
which says
v0
u=
vx
which is smaller in magnitude than the speed of light by the assumption that v
is timelike (4).
(c) Show that the sum of any two orthogonal2 spacelike vectors is spacelike.
V, W spacelike means their norms are positive: V 2 V V > 0, W 2 > 0.
V, W ortthogonal means V W V W = 0. The sum V + W has norm
(V + W )2 = V V +2 V W + W W > 0.
| {z } | {z } | {z }
>0 =0 >0
(d) Show that a (nonzero) timelike vector and a (nonzero) null vector cannot be
orthogonal.
A timelike vector T has T 2 < 0, and a null vector has N 2 = 0. Their inner
product T N is Lorentz invariant. By part (b) theres a Lorentz frame where
T has zero spatial components; N is still a nonzero null vector in that frame. In
that frame
T N = T 0 N 0 .
If T and N orthogonal, then, we must have either T 0 = 0 (in which case T is zero
which its not) or N 0 = 0. But if N 0 = 0 then clearly N cant be null, since
N is Null ~2 N
~2
0 = N 2 = N N = (N 0 )2 + N
2
Note that the assumption of orthogonal is necessary: otherwise V = (v 0 , ~v ) and V (v 0 , ~v ) is a
counterexample.
