Physics , General Relativity

Homework
Due Friday, th November
Jacob Lewis Bourjaily

Problem 1
Let us consider a manifold with a torsion free connection R(X, Y ) which is not necessarily metric
compatible. We are to prove that
R(X, Y )Z + R(Y, Z)X + R(Z, X)Y = 0, (1.1)
and the Bianchi identity

X R(X, Y ) V + Y R(Z, X) V + Z R(X, Y ) V = 0. (1.2)

The first identity is relatively simple to proveit follows naturally from the Jacobi
identity for the Lie derivative. Let us first prove the Jacobi identity:
[X, [Y, Z]] + [Y, [Z, X]] + [Z, [X, Y ]] = 0. (1.3)
Using the antisymmetry of the Lie bracket and our result from last homework problem
3, we have
[X, [Y, Z]] = X[Y, Z] = [Y,Z]X = ZY X Y ZX = [Z, [X, Y ]] [Y, [Z, X]].

o 
The condition of a connection being torsion free is that
XY = X Y Y X. (1.4)
Expanding the Lie brackets encountered in the statement of the Jacobi identity,
0 =XY Z + Y ZX + ZXY,
=X (Y Z Z Y ) + Y (Y X X Y ) + Z (X Y Y X) ,
=X Y Z X Z Y [Y,Z] X + Y Z X Y X Z [Z,X] Y + Z X Y Z Y X [X,Y ] Z,

= X Y Y X [X,Y ] Z + Y Z Z Y [Y,Z] X + Z X X Z [Z,X] Y ;
R(X, Y )Z + R(Y, Z)X + R(Z, X)Y = 0. (1.5)

o 

To prove the Bianchi identity, we will dirty our expressions with explicit indices in
hope of a quick solution. It is rather obvious to see that (1.2) is equivalent to the
component expression
Rabcd;e + Rabde;c + Rabec;d = 0. (1.6)
Worse than introducing components, let us use our (gauge) freedom to consider the
Bianchi identity evaluated at a point p in spacetime in Riemann normal coordinates1.
If we show that the Bianchi identity (1.6) holds in any particular coordinates at a
point p, it necessarily must hold in any other coordinate systemand if p is arbitrary,
then it follows that the Bianchi identity holds throughout spacetime.
Recall from lecture or elsewhere that Riemann normal coordinates at p are such that
abc (p) = 0. This implies that the covariant derivative of the Riemann tensor is simply
a normal derivative at p. Using the definition of Rabcd in terms of the Christoffel
symbols, we see at once that
Rabcd;e (p) + Rabde;c (p) + Rabec;d (p) = abd,ce (p) abc,de (p) + abe,dc (p) abd,ec (p) + abc,ed (p) abe,cd (p);
Rabcd;e (p) + Rabde;c (p) + Rabec;d (p) = 0. (1.7)

o 

1Riemann normal coordinates are constructed geometrically as follows: in a sufficiently small neighbourhood about p,
every point can be reached by traversing a certain geodesic through p a certain distance. If we choose to define all families
of geodesics through p using the same affine parameter then if we fix , there is a (smooth) bijection between tangent
vectors in Tp M to points in the neighbourhood about p: the direction of v Tp M tells the direction to the nearby points
and its magnitude (for fixed ) tells the distance to travel along the geodesic. Needless to say this construction does not
require a metric.
1
 2 JACOB LEWIS BOURJAILY

Problem 2
We are to compute the Riemann tensor, the Ricci tensor, the Weyl tensor and the scalar curvature
of a conformally-flat metric,
gab (x) = e2(x) ab . (2.1)

Using the definition of the Christoffel symbol with our metric above, we find
1
abc = g am {gam,b + gbm,a gab,m } ,
2
1
= e2 am bm e2 c + cm e2 b e2 bc m ,
2
abc = ba c + ca b bc am m . (2.2)
Using this together with the (definition of the) Riemann tensors components
Rabcd = abd,c abc,d + m a m a
bd cm bc dm , (2.3)
2
we may compute directly ,
Rabcd = da c b bd am c m ca b d + bc am d m da (b ) (c ) + bd am (c )(m )
da (c )(b ) ca (b )(b ) + bc da mn (m )(n ) + ca (b )(d ) bc am (d )(m )
+ ca (d )(b ) + da (b )(c ) bd dac mn (m )(n ) bd am (c )(m ) + bd am (c )(m )
n o
= bm (ca dn da cn ) + bd ( an cm mn ca ) + bc ( mn da an dm ) (m ) (n )
+ (da c ca d ) b + am (bc d m bd c m ) .

o o

It will be helpful to recast this into the form where all the indices are lowered. We can
do this by acting with the metric tensor. Doing so we find,
n o
e2 Rabcd = bm (ac dn ad cn ) + bd (an cm ac mn ) + bc (ad mn am dn ) (m )(n )
+ ad c b ac d b + bc d a bd c a ,
n o
= ad bm cn ac bm dn + bc am dn bd am cn m n (m )(n ) + ad bc ac bd mn (m )(n ).
(2.4)
Although we will not have any use for such frivolities, we can further compress this
expression to

e2 Rabcd = 4[a
r n s m
b] [d c] rs m n (m )(n ) + ad bc ac bd mn (m )(n ). (2.5)
Now, we can then find the Ricci tensor by acting on equation (2.4) with g ac . Letting D
be the dimensionality of our manifold, we find
n o
Rbd = dm bn Ddm bn + dm bn bd mn m n (m )(n ) + mn bd Dbd (m )(n ),

=(2 D) b d (b )(d ) + (2 D)bd mn (m )(n ) bd mn m n . (2.6)

o o

Lastly, contracting this, we find the scalar curvature,

e2 R =(2 D) mn m n (m )(n ) + D(2 D) mn (m )(n ) D mn m n ,
=2(1 D) mn m n (2 D)(1 D) mn (m )(n ). (2.7)

o o

2To be absolutely precise, there are two terms which manifestly cancel that appear when expanding this expression,
which we have left out for typographical and aesthetic considerations.
 PHYSICS : GENERAL RELATIVITY HOMEWORK 3

All that remains for us to compute is the Weyl tensor. Any exposure to conformal
geometry immediately tells us that the Weyl tensor vanishes. That is, that
1 1
Rabcd = gac Rdb + gdb Rac gad Rbc gbc Rad R gac gdb gad gbc . (2.8)
(D 2) (D 1)(D 2)
We will try as hard as possible to avoid actually computing the right hand side by
expanding our expressions above. To show that the Weyl tensor vanishes, we must
build Rabcd out of Rbc , R and the metric gab . This statement alone essentially gives
us the expression at first glance.
The first important thing to notice is that Rabcd has no term proportional to mn m n
while both Rab and R do. This means that if Rabcd can only be composed of linear
combinations of Rab and R which do not contain mn m n . Looking at expressions
(2.4) and (2.6), we see that they can only appear in the combination
e2 bd gbd
Rbd + R = Rbd + R. (2.9)
2(1 D) 2(1 D)
Any multiple of this combination will automatically have no mn m n contribution.
Staring a bit more at equations (2.4) and (2.6), we notice that the first set of terms in
(2.4) are all of the form gac Rbd . Indeed, we see that
1 n o n o
ad Rbc ac Rbd +bc Rad bd Rac = ad bm cn ac bm dn +bc am dn bd am cn m n (m )(n ) +. . . .
2D
(2.10)
Notice that multiplying both sides of the above equation by e2 will convert all of
the ab s into gab s3 . This is all we need to construct the Riemann tensor from the
Ricci tensor and scalar curvature: knowing the combination of Ricci tensors which
gives part of the Riemann tensor, we can use (2.9) to determine the rest. Indeed, we
see that
1 n o R
Cabcd + Rabcd = gab Rbc gad Rbd + gbc Rad gbd Rac + gad gbc gac gbd + gbc gad gbd gac ,
2D 2(1 D)(2 D)
1 n o R
= gac Rbd gad Rbc gbc Rad + gbd Rac gac gbd gad gbc ,
D2 (D 1)(D 2)
n o
= gad bm cn gac bm dn + gbc am dn gbd am cn m n (m )(n ) + gad gbc gac gbd g mn (m )(n ),
=Rabcd ;
Cabcd = 0. (2.11)

o o

3The conversion from

ab gab is completely natural. The only possibly non-trivial step comes from the last term
in the expression (2.4) for the Riemann tensor: bringing e2 to the right hand side of (2.4), we have a term which has
two lowered ab s and one upper ab ; now, e2 mn = e4 g mn and how these two factors of e2 can be absorbed into the
lowered s as desired.
4 JACOB LEWIS BOURJAILY

Problem 3
We are to show that if (x) satisfies the flat-space, massless Klein-Gordon equation, then if gab =
e2(x) ab , the transformed field e(x) (x) 0 (x) satisfies the equation
g ab 0;ab R0 = 0, (3.1)
for appropriate values of and dependant on the spacetime dimension but independent of (x).

Let us agree to call ab a b . Then the flat-space Klein-Gordon equation is sim-

ply (x) = 0. Recall the expression for the scalar curvature R in D spacetime
dimensions for a metric which is conformally-related to the Minkowski metric (2.7):
R = 2(1 D)e2 (2 D)(1 D)e2 mn (m )(n ). (3.2)
We would like to explicitly state g ab b a in terms of and . This can be done quite
explicitly, recalling the Christoffel symbols for a conformally-flat spacetime(2.2),
g ab b a = g ab a b g ab cab c ,
n o
= e2 ab ac (b )c + bc (a )c ab cm (m )c ,
n o
= e2 cb (b n)c ac (a )c + D cm (m )c ,
n o
= e2 (D 2) ab (a )b .

Acting with g ab b a on 0 we find,

n o
g ab b a 0 = e2 e + (D 2) ab (a ) b e ,
n o
= e2 0 () + ( + D 2)0 ab (a )(b ) + 2e ab (a )(b ) + (D 2)e ab (a )(b ) .

Although only one equation, if (3.1) is to hold for arbitrary (x), there are actually
three constraints implied by (3.1)one for each functionally distinct contribution.
Actually, well find that there are only two independent conditionsjust enough to
uniquely determine and .
First, notice that R does not contain any derivatives of (x). Therefore equation (3.1)
implies that
2e ab (a )(b ) + (D 2)e ab (a )(b ) = 0, (3.3)
ab 0
arising from the g b a term in (3.1). This obviously implies that
D2
= . (3.4)
2
The next condition(s) come form matching the remaining two functionally distinct terms
in (3.1), namely4
g ab b a 0 R0 + ( + D 2) ab (a )(b ) 2(1 D) + (D 2)(D 1) ab (a )(b ).
(3.5)
Matching the corresponding terms, we see that
( + D 2)
= and = . (3.6)
2(1 D) (D 2)(D 1)
We see that = 21 (D 2) is consistent with both of thesemore concretely, any two
of these three constraints is sufficient to imply the third. Therefore, we have shown
that 0 = e will satisfy the modified Klein-Gordon equation (3.1) for any (x) if

2D 1D2
= and = . (3.7)
2 4D1

o o

4We are not including those pieces eliminated by the choice (3.4).