Physics, General Relativity: Homework Due Wednesday, November

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Physics , General Relativity

Homework
Due Wednesday, th November
Jacob Lewis Bourjaily

Problem 11
We are asked to determine the ratio of frequencies observed at two fixed2 points in a spacetime with
a static metric gab ; we should use this to determine the redshift of light emitted from the surface of the
Sun which is observed on the surface of the Earth.

Imagine a clock at a fixed point x1 which ticks with a regular interval s. Because the
point is stationary, we may use the definition of the spacetime metric gab to see that
this interval is related to the coordinate time interval t by3
s2 = t21 g00 (x1 ). (1.1)
We have included a subscript on the coordinate time interval to make its position-
dependence manifest. The invariant interval s, however, must certainly be position-
independent for any reliable clock. Therefore, we naturally have that
s2 = t21 g00 (x1 ) = t22 g00 (x2 ), (1.2)
for any other point x2 . This implies that
t21 g00 (x2 )
= . (1.3)
t22 g00 (x1 )
It is important to note that this discussion is not limited to clocks ticking regularly: any
process with a well-defined, constant time interval observed at two distinct points
will obey equation (1.3). Indeed, consider an atomic transition which emits photons
with frequency 1 t1 1 at point x1 . Equation (1.3) implies that the frequency at
x1 will be related to the frequency 2 at x2 by
s
2 g00 (x2 )
= . (1.4)
1 g00 (x1 )

o o

To determine the redshift of light emitted from the Sun and observed on the Earth we
recall that in the Newtonian (weak-field) approximation,
g00 (x) = 1 2N (x), (1.5)
where N (x) is the Newtonian potential at x. The only subtlety is that we should
make sure to be careful about units when computing N (x). Notice that because
the 1 in 1 2N (x) is dimensionless, so should N (x) be. This will be the case if
we judiciously set c = 1. In these units, we find
N (R ) = 2.12 106 and N (R ) = 1.06 108 , (1.6)
which gives a redshift of 2.11 parts per million.

1Note added in revision: this solution is bad. The argument presented for equation (1.4) is not valid (even though the
right answer emerges). One should be very careful about the thought experiment under consideration (because the inverse
result is easy to obtain under a different situation).
2The equation which the problem set asks us to demonstrate is only valid for stationary sources and observersotherwise
there would be a doppler-shift term obfuscating the equation.
3In his textbook, Weinberg has an interesting discussion on why it is fundamentally not possible to disentangle s from
t at a particular point. However, it is possible to compare the metric at two distinct pointsby observing a gravitational
redshiftas described presently.
1
2 JACOB LEWIS BOURJAILY

Problem 2
We are to find the natural generally covariant generalization of the flat-space Klein-Gordon La-
grangian (which was shown to be Weyl invariant in the last problem set). We should use this to
determine the matter stress-energy tensor and show that it is traceless.
The striking similarity between
1
g ab a b R = 0, (2.1)
6
and the massive Klein-Gordon equation makes us guess that the action from which
this is derived is
Z
1 4 ab 1 2
S= d x gg a b + Rab . (2.2)
2 6
Our intuition is confirmed by calculating the equation of motion:

L L 1
0 = a = a g ab b R,
a 6
1
= g ab a b R. (2.3)
6
Therefore the action (2.2) does indeed give rise to the desired equation of motion for
as desired.
We now must compute the stress-energy tensor for this matter Lagrangian. Recall that
the stress-energy tensor T ab of a system with action S is defined according to
Z
1
S = d4 x g T ab gab , (2.4)
2
from the variation gab 7 gab + gab . To compute the metric variation for the action
given in (2.2) we first recall some useful identities:
1
g ab = g ac g bd gcd ; ( g) = gg ab gab ; (2.5)
2
and g ab Rab = a wa , where wa b (gab ) g cd a (gcd ) . (2.6)
This last identity, (2.6), follows from work done in lecture. Although brevity tempts
us to simply quote Walds textbook, it is sufficiently important to warrant a full
derivation. Therefore, to please the reader, a proof of this identity has been included
as an Appendix to this problem set.
We are now prepared to compute the metric variation of the action (2.2). As we proceed,
any total divergence will be assumed to integrate to zero.
Z
1 1 ab 1 1 1
S = d4 x g g gab g cd c d + Rcd 2 + g ab a b + Rab 2 + g ab Rab 2 ,
2 2 6 6 6
Z
1 4 1 ab cd 1 2 a b 1 ab 2 1 cd 2
= d x g gab g g c d + Rcd + R + g Rcd ,
2 2 6 6 6
Z
1 4 1 ab cd 1 2 a b 1 ab 2 1 c 2
= d x g gab g g c d + Rcd + R + ( wc ) ,
2 2 6 6 6
Z
1 4 1 ab cd 1 2 a b 1 ab 2 1 c 2
= d x g gab g g c d + Rcd + R wc .
2 2 6 6 6
(2.7)
The last term in the expression above is qualitatively different from the first two. Let
us try to recast it into a form which makes the gab -dependence manifest. Using the
definition of wc and making repeated use of integration by parts, we see
Z Z

d x g ( )wa = d4 x g a (2 ) b (gab ) g cd a (gcd ) ,
4 a 2

Z !

4 b a 2 b a 2 cd a 2 ab c 2
= d x g gab ( ) gab ( ) g a gcd ( ) + g gab c ( ) ,
Z

= d4 x g gab g ab g cd c d 2 a b 2 . (2.8)
PHYSICS : GENERAL RELATIVITY HOMEWORK 3

We are now ready to put everything together and find T ab . To make our result a bit
more transparent, let us agree to call g cd c d 2. Also, the identity g cd c d =
1 2
2 2 2 will allow us to tidy up our expressions substantially. Combining all
of this, we can continue our work on the total variation (2.7) using the result from
(2.8) to find
Z
1 4 1 ab cd 1 2 a b 1 ab 2 1 ab cd 2 a b 2
S = d x ggab g g c d + Rcd + R g g c d ,
2 2 6 6 6
Z
1 1 ab cd 1 1 1 1
= d4 x ggab g g c d c d 2 + Rcd 2 a b Rab 2 + a b 2 ,
2 2 3 6 6 6
Z
1 1 1 1 1 1
= d4 x ggab g ab 22 + R2 2 a b 2 + a b Rab 2 . (2.9)
2 2 6 6 3 6
This allows us to read-off

1 1 1 1 1
T ab = g ab 22 + R2 2 a b 2 + a b Rab 2 . (2.10)
2 6 6 3 6

o o

As anyone whos seen conformal field theory knows, the trace of the stress-energy tensor
must vanish. Lets see how this magically works out in the situation considered
presently.
1 1 1 1
gab T ab = 22 + R2 22 22 + 2 R2 ,
3 3 3 6
1 2
= R 2,
6
1
= 2 R ,
6
= 0.
Notice that the last line required using the equations of motionwhich wasnt entirely
anticipatedat least by us.
4 JACOB LEWIS BOURJAILY

Problem 3: Killing Vectors

a. If a (x) is a Killing field and pa () is the tangent vector to a geodesic curve (), then pa a (x)
is constant along .
proof: The derivative of pa a along is
pb b (pa a ) = pa pb b a + a pb b pa . (3.1)
a b
The first term vanishes because p p is symmetric while b a is antisymmetric (be-
cause it is Killing). The second term vanishes because pa is the tangent of a ge-
odesic, which practically by definition implies that it obeys the geodesic equation,
pb b pa = 0.
o 
b. We are to list the ten independent Killing fields of Minkowski spacetime.
The ten independent Killing fields correspond to the ten generators of the Poincare
algebra: four translations, three rotations, and three boosts. Given in terms of the
basis vectors ~ea , we the Killing vector fields are therefore
Translations : ~et , ~ex , ~ey , ~ez ;
Rotations : y~ex x~ey , z~ey y~ez , x~ez z~ex ;
Boosts : x~et + t~ex , y~et + t~ey , z~et + t~ez ;
Each of these ten vector fields manifestly satisfies Killings equation. That they are
linearly independent is also manifest4.

c. If a and a are Killing fields and , constants, then a + a is Killing.


proof: As should be obvious to all but the most casual observer,
b (a + z ) = b a + b a = a b a b = a (b + b ) , (3.2)
because, being constants, , commute with the gradient and a , a are Killing.
Therefore equation (3.2) implies that (a + a ) is Killing.
o 
d. We are to show that Lorentz transformations of the Killing vector fields listed in part (b) above
give rise to linear recombinations of the same fields with constant coefficients.
Because every Lorentz transformation can be built from infinitesimal ones, it is sufficient
to demonstrate the claim for infinitesimal Lorentz transformations. And this makes
our work exceptionally easy. Infinitesimal Lorentz transformations are simply the
identity plus a constant multiple of the generators of the Lorentz algebra; but (the last
six of) the Killing fields listed in part (b) are nothing but these Lorentz generators.
Therefore, any infinitesimal Lorentz transformation of the Killing fields listed in part (b)
is a linear combination of those same Killing fields with constant coefficients. And
by extension, the same is true for any finite Lorentz transformation.

4Although we should add that we were not requested to demonstrate thisso our lack of exposition here should be
forgiven.
PHYSICS : GENERAL RELATIVITY HOMEWORK 5

Appendix

In problem 2 we made use of an identity that didnt obviously follow from the work in lecture. We
remedy that deficiency presently5.

Lemma: Under the variation gab 7 gab + gab ,


g ab Rab = a wa , where wa b (gab ) g cd a (gcd ) . (A.1)
proof: We may begin with the related expression derived in lecture,

g ab Rab = a g bc abc g ac bcb . (A.2)
Rearranging this we find
n o
g ab Rab = a g bc gad dbc dad ; (A.3)
therefore, it suffices to show that the term in brackets is equal to wa . Expanding this
expression and using symmetry to collect and cancel terms, we find
1 bc 1
g bc gad dbc dad = g gad g de gbe,c + gce,b gbc,e + g bc ae gbe,c + gce,b gbc,e
2 2
1 be 1
g gae,b + gbe,a gab,e g be gae,b + gbe,a gab,e ,
2 2
1 bc ef d 1 1 1
= g g a gde gbf,c + gcf,b gbc,f + g bc gba,c + gca,b gbc,a + gde g dc g ef gcf,a g be gbe,a ,
2 2 2 2
1 1 1 1
= g bc g ef ad gde gbf,c g bc g ef ad gde gbc,f + g bc gba,c g bc gbc,a + gde g dc g ef gcf,a g bc gbc,a ,
2 2 2 2
1 1
=g bc gba,c g bc gbc,a g bc g ef gbf,c gae g bc g ef gbc,f gae + g dc g ef gcf,a gde .
2 2
Expanding the first two terms in the expression above in terms of covariant derivatives
and Christoffel symbols, we observe
g bc gab,c = b (gab ) + g bc geb eac + g bc gae ebc , (A.4)
and
g bc gbc,a = g bc a (gbc ) + g bc gbe eac + g bc gec eba . (A.5)
Noting that the terms with the covariant derivatives are what we are looking for
together, they give wa . Putting everything together,

1 1
g bc gad dbc dad = wa + g bc gae ebc g bc gec eba + g bc g ef gf c,a gbe + gbc,f gae gbf,c gae .
2 2
(A.6)
All that remains is for us to show that the terms in curly brackets above vanish. To do
this, we will expand our expressions one last timethis time using the definition of
the Christoffel symbols for a metric connection. Doing so, we find

1 1
g bc gad dbc dad wa = g bc g ef gf c,a gbe + gbc,f gae gbf,c gae
2 2
1 1 1
gbf,a gec gaf,b gec + gba,f gec
2 2 2
1 1 1
+ gbf,c gae + gcf,b gae gbc,f gae ,
2 2 2
= 0.
Here we have indicated the terms that cancel together in matching colours. With
this, we have shown that
n o n o
g ab Rab = a g bc gad dbc dad = a b (gab ) g cd a (gcd ) = a wa . (A.7)

o 

5We hope that there is an easier way to prove the following Lemma. But alas! too little time to be brief. Breviloquence
is a time-consuming luxury.

You might also like