General Relativity

MA4448
Transcribed by Debbie Ip
Taught by Dr Peter Taylor
Hilary Term 2012

Contents
1 Introduction
1.1 Newtonian Theory of Gravity . . . . . . . . . . . . . . . . . .
1.2 Special Relativity . . . . . . . . . . . . . . . . . . . . . . . . .
1.3 General Relativitiy (Guiding Principles) . . . . . . . . . . . .

1
1
2
3

2 Einstein Equations from an Action Principle

2.1 Principle of Least Action . . . . . . . . . . . . . . . . .
2.2 The Stress-Energy-Momentum Tensor . . . . . . . . . .
2.3 Varying the Metric Inverse and the Metric Determinant
2.4 The Einstein Hilbert Action . . . . . . . . . . . . . . .
2.5 Einsteins Field Equations . . . . . . . . . . . . . . . .
2.6 Further Remarks on the Field Equations . . . . . . . .

4
4
5
5
7
8
9

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

3 The Stress-Energy-Momentum Tensor

10
3.1 Decomposition of the Stress-Energy-Momentum Tensor in an
Orthonormal Tetrad . . . . . . . . . . . . . . . . . . . . . . . 10
3.2 Stress-Energy-Momentum Tensor for a Perfect Fluid and for
Dust . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4 The
4.1
4.2
4.3

4.4

Schwarzschild Solution
Canonical Form of a Spherically Symmetric Line-Element
The Schwarzschild Solution . . . . . . . . . . . . . . . .
Properties of the Schwarzschild Solution . . . . . . . . .
4.3.1 Limiting Cases M 0, r . . . . . . . . . .
4.3.2 The Coordinate Singularity at r = 2M . . . . . .
Birkhoffs Theorem . . . . . . . . . . . . . . . . . . . . .

5 Solar System Tests of GR

5.1 The Gravitational Red-Shift . . . . . . . . .
5.2 Planetary Motion and Perihelian Advance of
5.2.1 Geodesic Equations . . . . . . . . . .
5.2.2 Propagation Equation for ( ) . . . .
5.2.3 Newtonian Result . . . . . . . . . . .
5.2.4 Shape of General Relativistic Orbit .
5.3 Light Reflection . . . . . . . . . . . . . . . .

. . . . .
Mercury
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

14
14
15
16
16
17
17

.
.
.
.
.
.
.

18
18
20
20
21
22
23
24

6 Black Holes
27
6.1 Radial In-falling Photons . . . . . . . . . . . . . . . . . . . . . 27
6.2 Radially In-falling Particles . . . . . . . . . . . . . . . . . . . 28
6.3 The Kruskal Extension of the Schwarzschild Manifold . . . . . 29
2

7 Cosmology
7.1 The Cosmological Principle: . . . . . . . . . . . .
7.2 Kinematics of the Continuum . . . . . . . . . . .
7.2.1 Connecting Vectors . . . . . . . . . . . . .
7.2.2 Shear, Stress and Expansion . . . . . . . .
7.3 The Friedman-Robertson-Walker (FRW) Metric .
7.3.1 Isotropy . . . . . . . . . . . . . . . . . . .
7.3.2 Comoving Coordinates . . . . . . . . . . .
7.3.3 The Spacetime Metric . . . . . . . . . . .
7.4 Cosmological Red-Shift and Hubbles Law . . . .
7.5 Einsteins Equations . . . . . . . . . . . . . . . .
7.5.1 Matter Content of the Universe . . . . . .
7.5.2 The Friedman and Raychauduri Equations
7.6 Cosmological Models with Vanishing . . . . . .
7.6.1 Geometry of the Universe . . . . . . . . .
7.6.2 Big Bang Singularities . . . . . . . . . . .
7.6.3 The Fate of the Universe (Eschatology) . .
8 The
8.1
8.2
8.3
8.4

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

Linear Approximation
The Einstein Equations in the Linear Approximation
Gauge Transformations . . . . . . . . . . . . . . . . .
The Newtonian Limit . . . . . . . . . . . . . . . . . .
Gravitational Waves . . . . . . . . . . . . . . . . . .
8.4.1 The Linearised Vacuum Solutions . . . . . . .
8.4.2 Energy Transfer . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

31
31
31
31
34
36
36
38
39
42
44
44
45
46
46
47
48

.
.
.
.
.
.

50
50
52
53
53
53
53

9 Einstein-Maxwell Theory
53
9.1 The Field Equations . . . . . . . . . . . . . . . . . . . . . . . 53
9.2 The Reissner-Nordstrom Solution . . . . . . . . . . . . . . . . 55

Suggested Reading
R. DInverno, Introducing Einsteins Relativity (+ - - - )
B. Shutz, A First Course in General Realtivity (- + + +)
S.M.Carroll, Lecture Notes on GR, online (- + + +)
H. Stephani, General Relativity (+ - - -)
R.M. Wald, General Relativity (- + + +)
C.W.Misner, K.S. Throne, J.A.Wheeler, Gravitation (- + + +)

Introduction

1.1

Newtonian Theory of Gravity

For a distribution of matter with density (t, x, y, z), we have a gravitational

potential .
Field Equation:
2 = 4G
where

2
2
2
+
+
x2 y 2 z 2
is given at points inside the matter distribution by solving Poissons
equation with given.
Outside the matter distribution (i.e. in a vacuum) is given by solving
2 =

2 = 0

Equations of Motion
The equations of motion of a test particle are
xi = xi (t), i = 1, 2, 3

i.e.

d 2 xi
= ,i
dt2

d2 x

=
2
dt
x
2
dy

=
2
dt
y

d2 z

=
2
dt
z
Newtonian theory can be written as a metric theory of gravity, though
the spacetime is complicated to describe geometrically. (Requires additional
structures such as absolute time and simultaneous points forming a Euclidean
3-manifold)
E.g. We can rewrite the equations of motion in the geodesic equation
form.

Define
x = (t, x, y, z)
x = (1, x i )
x = (0, xi ) = (0, ,i )
i.e.
t = 0 and xi + ,i = 0
xi + ,i x 0 x 0 = 0
Compare with the geodesic equation

d 2 x
dx dx
=0
+

dt2
dt dt
We can read off the Newtonian connection

i N 00 = ,i and N = 0 otherwise
RN i0j0 = ,i and RN = 0 otherwise
RN 00 = 2 and RN = 0 otherwise
i.e. Newtonian spacetime is curved.

1.2

Special Relativity

Describes non gravitational physics such as electrodynamics, standard model

etc.
Discard the notion of absolute time, we introduce a 4D space continuum.
To each event in spacetime, we assign the coordinates (t, x, y, z) and the
infinitesimal interval ds between the infinitesimally separated events satisfies
the Minkowski line element.
ds2 = dt2 + dx2 + dy 2 + dz 2
= dx dx
where
= diag (1, 1, 1, 1)
The Minkowski line element is invariant under Lorentz transformations
0

x x
2

where

0 0 =
and under arbitrary translations
x x + d
where d are constants. Together, these define the Poincare group.
Null cones of describe light rays in a vacuum.
Time-like geodesics of describe force free motion of massive particles and
Z
Z
Z
1

12
= d = ( dx dx ) = (1 v 2 ) 2 dt
is the proper time measured by a standard clock associated with the particle
where the integral is taken along a time like path representing the particles
trajectory.
Note Used relativistic units c = 1. In non-relativistic units
1
Z 
v2 2
=
1 2
dt
c

1.3

General Relativitiy (Guiding Principles)

All observers are equivalent.

physics should be coordinate independent. (Principle of General
Covariance)
tensor equations are the most natural mathematical framework.
Should agree locally with SR
Admit a class of preferred relatively accelerated world lines representing
free fall.
Should admit a tensor related to the source of the gravitational field.
Should explain observed solar system phenonema such as light deflection, perihelian advance of Mercury, time-delay etc.
General relativity assumes spacetime is a pseudo-riemannian manifold
with signature(- + + +).

Null geodesics represent light rays.

Timelike geodesics represent paths of freely falling particles.
Locally, we can always choose a chart s.t. g = , i.e. SR valid locally.
The field equations are
G = T
where is a constant fixed by the Newtonian limit, G is the Einstein
Tensor, and T describes the source of the gravitational field.

Einstein Equations from an Action Principle

We first recall two important results:

(i) Fundamental Lemma of Calculus of Variations: If
Z x2
(x)(x)dx = 0
x1

where (x) is continuous, ad (x) twice differentiable and vanishes on boundary (x1 ) = (x2 ) = 0, then (x) = 0 on [x1 , x2 ]
(ii) Gauss Divergence Theorem:
Z
Z

d
d =
V

where is a vector density of weight 1. An immediate corollary is

Z
Z

g X d =
g X d
V

where X is a vector field.

2.1

Principle of Least Action

We start with an action

Z
Ld

S=
all space

where L is a Langrangian density of weight 1. We consider small variations in the metric tensor g g + Sg which inducs a variation in the
action functional S S + S. We also assume the metric variations and its
derivatives vanish at infinity.
4

The action principle implies

Z

L g d = 0

S =
all space

where L

2.2

L
g

is a

( 20 )

tensor density of weight 1.

The Stress-Energy-Momentum Tensor

In General Relativity, we must allow for the definition of a tensor related to

the source of the gravitational field, i.e. the action has contributions coming
from the matter fields and the gravitational fields
Z
S = S + Sg =
(L + Lg )d
all space

We define
Z
S =
all space

where we have defined

L
1
g d =
g
2

g T g d

all space

2 L
T =
g g

which is the stress-energy-momentum tensor.

2.3

Varying the Metric Inverse and the Metric Determinant

In what follows, we shall require g in terms of g . We note that

g g =
g g + g g = 0
Multiply by g
g = g g g
g = g g g

(2.1)

We also require g. We note that for any non-singular matrix a , with

inverse a and determinant a. Each element has a cofactor given by
A = a a
5

Also, the determinant is obtained by expanding across any row.

a = A() a()

(no sum over )

which imples that

a
= A = a a
a
Therefore
a =

a = g, this gives

a
a = a aa
a

1
1
( g) = (g) 2 g
2
1
1
= (g) 2 gg g
2
1
gg g
2

1
( g) =
g g g
2
An immediate consequence of equation (2.2) is

1
( g), =
g g g,
2

(2.2)

(2.3)

Example 2.3.1
Show that equation (2.3) leads to

( g) = 0

g is a scalar density of weight 1. The covariant derivative of a scalar

density of weight a is
= ,
We wish to show that

( g) = ( g), g = 0
We know that
0 = g = g, g g
g, = g + g
6

Equation (2.3) implies:

1
( g), =
gg g,
2
=

1
g g (g + g )
2
=

1
g( + )
2
=

1
g( + )
2

= g

( g) = 0

2.4

The Einstein Hilbert Action

We now consider the contribution to the action coming from the gravitational
field:
Z
Sg =
Lg d
space

The only scalar density of weight 1 involving the metric and its derivatives

up to second order is gR. i.e. we take

Lg = 1 gR = 1 g g R
Z

1
Sg =
[( g g )R + g g R ]
We require our expression for R schematically, we have
R = +
Thus
R = () () + +

is a well defined tensor (even though is not, since this involves the
difference of two connections) we can therefore convert partial derivatives to
covariant derivatives:

R
= ( ); ( );

The second term of the gravitational action is therefore

Z


1

g (g ); (g ); d
Now g and g are vectors, so we may apply the corollary to the
divergence theorem to convert to a surface integral
Z

g(g g )d = 0)
The gravitational action reduces to
Z

1
Sg =
( g g )R d
1

=
1

2.5

[( g)g R + gg R ]d

g g g g R g g g R d
2


Z

1
1

=
g RR
g
g d
2
Z

1
=
g G g d

Z 

Einsteins Field Equations

Putting the results together, we have

S = S + Sg
Z
Z

1
=
g T g d
g G g d
2


Z

1
1
=
g
T G
g d = 0
2
Since we assume metric variations vanish at the boundary, we must have
1
T 1 G = 0
2

= G = T
2
8

In the weak field slow moving approximations, they reproduce Poissons

equation only when
16G
=
(DInverno)
c4
or, in natural units c = g = 1, = 16.
G = T Recall the twice contracted Bianchi identities
G
; = 0
= T; = 0

2.6

Further Remarks on the Field Equations

The history of an isolated body in spacetime is a timelike world tube filled

with the world lines of the constituent particles. Inside the world tube, we
have T 6= 0 , and we solve the non vacuum Einstein Field Equations.
G = T
Outside the world tube, T = 0 and we solve the vacuum field equations
G = 0 R = 0
Agreement with Newtonian limit requires = 16.
G = 8T
The world line of a particle x (s) with non zero mass is timelike. Taking s
to be arc-length along the curve, we have
dx dx
= 1
ds ds
If the particle is a test particle (doesnt perturb the geometry of spacetime),
then the world line is a timelike geodesic satisfying
g

d 2 x
dx dx
+

=0

ds2
ds ds
where s is now the proper time along the curve.

We take the world line of massless particles to be null geodesics

d 2 x
dx dx
+

=0

dr2
dr dr

and
g

dx dx
=0
dr dr
9

3
3.1

The Stress-Energy-Momentum Tensor

Decomposition of the Stress-Energy-Momentum Tensor in an Orthonormal Tetrad

The stress-energy-momentum tensor satisfies

1
G
and T; = 0
8
T is a symmetric 4x4 matrix and in general will have 4 mutually orthogonal
eigenvectors; one timelike and three spacelike.
Let u be the unit timelike eigenvector of T with eigenvalue , i.e.
T =

T u = u , u u = 1
= G u
We take timelike worldlines tangent to u (i.e. the integral curves of u )
to be the worldlines of the constituent particles of the matter distribution.
We take to be the proper density of the matter (density observed in the
rest frame of the constituent particle). u is the 4-velocity of a constituent
particle, and it describes the interval motion of the body.
We further define
S = u u T = S
S u = u (u u ) T u
= u + u
=0
u is a unit timelike eigenvector of S with eigenvalue zero.
S has 6 independent components, and is called the stress tensor of the
matter distribution.
We now let {e(1) , e(2) , e(3) } = {e(i) }3i1 be the unit spacelike eigenvectors
of T with eigenvalues {p(i) }3i=1 , respectively.
T e(i) = p(i) e(i)

i = 1, 2, 3 no sum over i

Mutual orthogonality implies

u e(i) = u e(i) = 0
Therefore
S e(i) = u u e(i) T e(i)
= (i) e(i)
10

i = 1, 2, 3

Hence {e(i) }3i=1 are the unit spacelike eigenvectors of S with e-value {p(i) }3i=1 .
These are called the 3 principle stresses in the matter distribution.
For pressures p(i) > 0.
For tensions p(i) < 0.
So we have 4 mutually orthogonal eigenvectors satisfying
u u = 1
e(i) e(j) = (i)(j)
e(i) u = 0
We set u = e(0) , then we have
e(0) e(0) = 1
e(1) e(1) = (i)(j)
e(0) e(1) = 0
e(a) e(b) = (a)(b)

(a,b=0,1,2,3)

(parenthesis around indices to distinguish tetrad indices from spacetime indices)

e(a) e(b) = (a)(B)
= diag(1, 1, 1, 1)
g e(0) e(b) = (a)(b)
{e(a) }3a=0 is an orthonormal tetrad. (a)(b) are the components of the metric
tensor on this orthonormal tetrad.
Any vector or tensor may be projected onto the tetrad from, for example,
the components of the curvature tensor in the orthonormal tetrad are
R(a)(b)(c)(d) = Re(a) e(b) e(c) e(d)
We can also write the metric components g in terms of {e(a) } following
from the orthonormality conditions:
g = (a)(b) e(a) e(b)
Therefore, we can pass freely from tensor components to tetrad components
and vice-versa.
11

Note:
T(0)(0) = T e(0) e(0) = T u u = u u =
T(0)(i) = T e(0) e(i) = T u e(i) = 0

(i=1,2,3)

T(i)(j) = T e(i) e(j) = p(i) e(i) e(j) = p(i) (i)(j)

T(a)(b) = diag(, p(1) , p(2) , p(3) )

3.2

g e(a) e(b) = (a)(b)

(a)

g (a)(b) e(a) e(b)

(b)

Stress-Energy-Momentum Tensor for a Perfect Fluid

and for Dust

Writing (b) out explicitly:

g = (0)(b) e(0) e(b) + (1)(b) e(1) e(b) + (2)(b) e(2) e(b) + (3)(b) e(3) e(b)
= (0)(0) e(0) e(0) + (1)(1) e(1) e(1) + (2)(2) e(2) e(2) + (3)(3) e(3) e(3)

=u u +

3
X

e(i) e(i)

i=1

3
X

e(i) e(i) = g + u u

i=1

Recall
S e(i) = p(i) e(i)

3
X

S e(i) e(i)

3
X

i=1

LHS = S

p(i) e(i) e(i)

i=1
3
X

e(i) e(i) = S (g + u u ) = S + 0

i=1

3
X

p(i) e(i) e(i)

i=1

For a perfect fluid, the stress is an isotropic pressure (no preferred direction)
p(1) = p(2) = p(3) = p
12

 S = p

3
X

e(i) e(i)

i=1

S = p(g + u u )

(stress tensor for a perfect fluid)

where 4 is the 4-velocity.

By definition, we have
T = u u S
= u u + p(g + u u )
T = ( + p)u u + pg
(stress-energy-momentum tensor for a perfect fluid)

Example 3.2.1
Show that for incoherent matter with proper density , that changes along
integral curves of u according to
, u = u; = 0
Further show that the world lines of the dust particles are timelike geodesics.
We have T = u u . The conservation equations are
T; = 0
0 = (u u )
= , u u + ( u )u + u u
= u (, u + u; ) + u; u

u; )

0 = (, u +
+
1
But u u; = (u u ); = 0
2

(1)

u u;
(as required)

Sub this result back into (1)

u; u = 0
Du u = 0
i.e. the integral curve of the dust particle parallel transports its own tangent
vector geodesics.

13

The Schwarzschild Solution

4.1

Canonical Form of a Spherically Symmetric LineElement

We shall consider spherically symmetric solutions to Einsteins vacuum field

equations.
Spherical symmetry implies that there exists a coordinate system (t, r, , )
say, in which the line-element is invariant under the reflections
0 =
0 =
i.e. no cross terms of the form drd, drd, dd, dd, dtd, dtd and that
each 2D submanifold defined by t =const, r =const, are the 2-spheres.
dl2 = a2 (d2 + sin2 d2 )
Therefore, the spherically symmetric line-element has the form
ds2 = A(r, t)dt2 + 2B(r, t)dtdr + C(t, r)dr2 + D(t, r)(d2 + sin2 d2 )

Changing the radial coordinate r r = D

r)dt2 + 2B(t,
r)dtd
r)d
ds2 = A(t,
r + C(t,
r2 + r2 (d2 + sin2 d2 )
Introduce a new time coordinate by
r)dt + B(t,
r)d
dt = I(t, r)[A(t,
r]
r+B
2 d
dt2 = I(t, r)[A2 dt2 2ABdtd
r2 ]
2
dt2
B
2

Adt + 2Bdtd
r=
+
d
r2
2

I A
A
The line-element now reads
 2

dt2
B
2
ds = 2 +
+ C dr2 + r2 (d2 + sin2 d2 ) (dropped the tildes)
I A
A
Defining 2 new functions p = p(t, r); q = q(t, r) by
1
= ep ;
2
I A

B2
+ C = eq
A

Our canonical form of a spherically symmetric line-element reads

ds2 = ep dt2 + eq dr2 + r2 (d2 + sin2 d2 )
14

4.2

The Schwarzschild Solution

To determine the functions p(t, r), q(t, r) we must solve the vacuum field
equations G = 0. The non vanishing components of the Einstein tensor are


1 q
1
1
t
q
Gt = e
2 2
(i)
r r r
r
eq q
2
(ii)
Gt =
r 
t

1 p
1
1
Grr = eq
+ 2 2
(iii)
r r r
r

1
G = G = eq
2
1
ep
2

!
 2
1 p q 1 q 1 p 1 p
2p
+

2
2 r r r r r r 2 r
r
!
 2
2 q 1 q
1 1 p
+

2
t
2 t
2 t t

We see that the Einstein equations give us 4 non trivial equations. However,
they are not all independent. The twice contracted Bianchi identities G
; = 0
imply that vanishing of (i) (iii) implies vanishing of (iv). So we have 3
independent equations


1
1 q
1
q
e
2 + 2 =0
(a)
r r r
r
q
=0
(b)
 t

1
1
1 p
+ 2 2 =0
(c)
eq
r r r
r
It is immediately obvioius from (b) that q is a function of r only. i.e q = q(r)
and therefore, (a) becomes a simple ODE:
eq eq r

dq
=1
dr

d
(req ) = 1
dr
req = r = const

Taking our constant of integration to be 2M (which we will interpret later)

yields:

1
2M
q
e = 1
r
15

To optain p we note that adding (a) and (c) gives

p q
+
=0
r r
i.e. p + q = f (t)
ep = eq ef (t)


2M
= 1
ef (t)
r
Our line element reads



1


2M
2M
2
f (t) 2
ds = 1
e dt + 1
dr2 + r2 d2 + sin2 d2
r
r
Finally, we may eliminate f (t) by redefining our time coordinate by
1

e 2 f (t) dt = dt0
Z t
1
0
e 2 f (u) du
t =
c

which gives (after dropping primes)




1
2M
2M
2
2
ds = 1
dt + 1
dr2 + r2 (d2 + sin2 d2 )
r
r
(Schwarzschild Solution)

4.3
4.3.1

Properties of the Schwarzschild Solution

Limiting Cases M 0, r

It is clear that by setting M = 0 we retrieve the Minkowski metric in spherical

polar coordinates. The parameter M represents the mass/energy and one
may interpret the Schwarzschild solution as the geometry due to a point
mass M at the origin.
We further note that as r , we again retrieve the Minkowski metric. We did not impose asymptotic flatness! Spherically symmetric vacuum
solutions of Einsteins equations are necessarily asymptotically flat.

16

4.3.2

The Coordinate Singularity at r = 2M

The metric components of G are singular at r = 0, and r = 2M (r = 2GM

in
C2
non natural units). The r = 0 singularity is known as a curvature singularity
and is irremovable. The r = 2M singularity is a coordinate singularity
and may be rmoved by an appropriate coordinate transformation (though
r = 2M still has important physical implications). To see this, we make the
coordinate transformation (t, r, , ) (u, r, , ) where
u = t r 2M log(r 2M )
1

2M
dr
du = dt 1
r
In these coordinates, the metric reads

1
2M
2
ds = 1
du2 2dudr + r2 (d2 + sin2 d2 )
r
In coordinates (u, r, , ) the components of G are non singular at r = 2M

)
1
0
0
(1 2M
r

1
0 0
0

g =
0
0 r2
0
0
0 0 r2 sin2
We also note that in the standard form of the Schwarzschild metric


2M
gtt = 1
< 0, r > 2M
r
> 0, r > 2M
so that the signature of the metric is (+ ++) for r < 2M . In this region,
r takes on the character of a time coordinate and t a spatial coordinate. We
call the region r > 2M the exterior Schwarzschild geometry, and the region
0 < r < 2M the interior Schwarzschild geometry.

4.4

Birkhoff s Theorem

Definition: Static space time: A space time is said to be static if there exists
a coordinate system in which the metric components are time- independent
and the metric is time reversal invariant, i.e. there exists a coordinate system
such that g,t = 0, and there are no cross terms dtdxi (i = 1, 2, 3)

17

Note: The chart independent definition relies on the existence of a timelike killing vector that is hypersurface orthogonal.
We note that the Schwarzschild solution is static, but we did not impose
this!
Birkhoff s Theorem: A spherically symmetric vacuum solution in the
exterior region is necessarily static.
Corollary: For a spherically symmetric source in the region r a, where
a > 2M , the exterior Schwarzschild solution is the unique solution.

Solar System Tests of GR

In order for GR to be considered a viable theory of gravitation, it ought

to be able to explain various phenomena in our solar system such as light
deflection
We model the gravitational field by the Schwarzschild solution with M =
M , the mass of the sun. We model the planets as text particles which move
on timelike geodesics of the Schwarzschild spcacetime. There are 3 classical
tests we shall consider:

5.1

The Gravitational Red-Shift

Let Co and C1 be the timelike world lines of an emitter and receiver of light,
respectively. Let be the proper time along them. Let P0 and P1 be the null
worldline of a photon emitted at the even P0 on C0 and received at the event
P1 on C1 . Suppose in a short interval d0 of proper time on C0 , n photons
are emitted and these are received in an interval d1 of proper time on C1 .
Then,
0 = frequency of emission
= no. of photons per unit time
n
=
d0

18

Similarly,
1 = frequency of reception
n
=
d1
0 d0 = 1 d1
0
d1

=
1
d0
If 0 , 1 are the emitted and received wavelengths respectively, then
1
1
1 =
(c=1)
0 =
0
1
1
d1

=
0
d0
A signal is red shifted (loses energy) if 1 > 0 or if d1 > d0 .
Suppose the emitter is at rest on the surface of the sun. Then the world
line C0 would be given by
r = a = solar radius
= 0
= 0
On C0 :

2M
ds = 1
dt2
a
r

2M
d = ds2 = 1
dt
a
2

Similarly, on C1 :
r
d1 =

For

M
a

2M
dt
r
q

1 2M
d1
1
r
=
=q

0
d0
1 2M
a


small Mr small
q
 2 !
 2 !
1 2M
M
M
M
M
r
q
= 1
+O
1+
+O
r
r
a
a
1 2M
a

1+

M
M

a
r
19

M
M
1
1+

0
a
r

Since Ma > Mr , we have 1 > 0 . i.e. signals are red-shifted as they pass
through the gravitational field


0 1
1
M
M

=
=1
=

z=
0
0
0
a
r


1
1r in standard units)
(or GM
c2
a
Note: This is not a Doppler shift since there is no relative motion between
observers.

5.2
5.2.1

Planetary Motion and Perihelian Advance of Mercury

Geodesic Equations

We treat planets as test particles moving among timelike geodesics of Schwarzschild

spacetime. Line element



1
2M
2M
2
2
ds = 1
dt + 1
dr2 + r2 (d2 + sin2 d2 )
r
r



1
2M 2
2M
L = g x x = 1
t + 1
r 2 + r2 (2 + sin2 2 ) = 1
r
r
(5.1)
dt
t =
d(proper time)


d
L
L
EL=
=0

d x
x

20


1 #
2M
2 1
r
r


2M
1
t=E
r
"
#
1 # "
2

2M
2M 2
d
2M 2
2M
1
= r :0 =
r 2 t 1
r + 2r(2 + sin2 2 )
2
d
r
r
r
r

1


M
M
2M
2M 2
2
2
2
2

r = 2 1
r (2 2M )( + sin ) + 2 1
t =0
r
r
r
r
d
2r2 sin cos 2
= :0 =
(2r2 )
d
2 2
+ r
sin cos 2 (iv)
r
d
= :0 =
(2r2 sin2 )

d
r2 sin2 = h
d
=t:
d

"

5.2.2

Propagation Equation for ( )

(iv)

= 0.
is a solution. Assume (0) = 2 , (0)
= 0 (0)

Differentiating (iv) gives (0)

= 0.
all derivatives of vanish.
Consider = 1 > 0 close to = 0, then
1 ...
1
2
3

(0 ) = (0) + (0)
(0)1 + ...
1 + (0)1 +
2
3!
1 ...
2
1 ) = (0)
+ (0)

(
(0)1 + ...
1+
2

(1 ) = , (
1) = 0
2

Therefore we have shown that assuming (0) = 2 , (0)

= 0, then it remains
true for some nearby point. By induction, it is true for all values of ,
) = 0.
( ) = 2 , (
only consider equatorial plane. We now have


2M
t=E
1
r
21

r 2 +
Substitution: h =

h2
r2

r2 = h



2M
2M
2
1
E + 1
=0
r
r

r0
r

dr du
dr
=
d
du d
dr du d
=
du d d
 2
r
du

=
d
r0
h du
=
r0 d

r =


2

du
2M
h2 2
2M u
+ 2u 1
E2 + 1
=0
d
r0
r0
r0
 2
du
2M ur0 2M u3
r2

+
+ u2 = 0 (E 2 1) +
d
h
h2
r0

u2
r02

(omitting

2M u3
,
r0

we retrieve the Newtonian result)

Differentiation gives the more familiar form

M r0 3M u2
d2 u
+
u
=
+
d2
h2
r0
5.2.3

(Relativistic Binet Equation)

Newtonian Result

Ignoring

2M u3
r0

and writing 2M
=1
r0
2

uN r2
duN
r2

+ u2N = 0 (E 2 1) + 2 0
d
h
h

This can be solved exactly by writing the solution as uN + u0 + v., where u0

is a constant chosen to eliminate the term linear in v.
 2
dv
r2
r2

+ u20 + 2u0 v + v 2 = 02 (E 2 1) + 20 (u0 + v)

d
h
h
u0 is chosen such that
2u0 =

r02
1 r02

u
=
0
h2
2 h2
22

dv
d

2

+ v2 =

r02 2
r02 u0
2
(E

1)

u
+
= k2
0
h2
h2

v() = k sin( 0 )
uN = u0 (1 + e sin( 0 ))
with e = uk (defines ellipse for 0 < e < 1)
 2
3
r2 (c2 1)
du
0
+ u2 = 0 h2 + 2 Mhur
+ 2 Mr0u
2
d
Newtonian result obtained by ignoring u3 term. Solved with Ansatz

uN = u0 + v
= u0 (1 + esin( 0 ))
ellipse with period 2
5.2.4

Shape of General Relativistic Orbit

Again we take u = u0 + v where u0 is a constant chosen to eliminate the term

linear in v.
This leads to requiring that u0 satisfies the quadratic
3u20 2u0 + 

r02
=0
h2

where  = 2M
 1 and we choose the solution that is closest to the Newtonian
r0
result. Then v satisfies
 2
dv
(c2 1)
M r0 u0
+ u20 + v 2 = r02
+2
+ u30 + 3u0 v 2 + v 3
2
2
d
h
h
Ignoring the v 3 term and collecting constants
 2
dv
+ v 2 (1 3u0 ) = k 2
d
which is easily solved, yielding
v=

k
sin( 0 )

23

where 2 = 1 3u0 . i.e. the shape of the orbit as predicted by relativity is

an ellipse with a periodicity
2
3
2(1 + u0 )

2
the periolian advance is given by (in standard units)

= 3u0
 
GM u0
= 6 2
c
r0
To approximate ur00 we use the fact that each orbit is approximately Newtonian and we know for an ellipse
rmax = a(1 + e) rmin = a(1 e)
where a is the semi-major axis
r0
r0
=
rmin
a(1 e)
r0
r0
=
(uN )min = u0 (1 e) =
rmax
a(1 + e)
1
2
2u0
1
+
=

=
r0
a(1 e2 ) a(1 e2 )
a(1 e2 )

(uN )max = u0 (1 + e) =

6GM
e2 )

c2 a(1

For Mercury, this predicts a shift of 43 per century while the observed value
is 43.1 0.5

5.3

Light Reflection

We consider photon paths in the Schwarzschild gravitational field. We describe the photons by null geodesics
x + x x = 0
and
g x x
24

where x = dx
is an affine parameter.
ds
Again, without loss of generality, we take the photon path to be in the
equatorial plane (s) = 2 for all s. Our geodesic equations are


2M
1
t=E
r
r2 = h
M2
r 2
r


1


2M
2M 2
M
2
2
1
1
t =0
r (r 2M ) +
r
r
r

The 1st integral of the motion

0 = g x x



1
2M 2
2M
2 2
= 1
t + r + 1
r 2
r
r
Using the fact that
dr
h dr
= 2
d
r d
and the conservation equations to simplify

1 2  2

1
2M
h
dr
h2
2M
2
1
+ 2 E 1
=0
r
r4 d
r
r
r =

Again, we take u =

r0
r

du
d

2

du
d

2

+ u2

E 2 r2
2M u3
= 20
r0
h

or
+ u2 u3 =

E 2 r02
h2

where  = 2M
1
r0
Take u = u0 + u, and subbing into our equation and equating equal orders
of  gives
2

r2
h
du0
+ u20 = 02 where d =
(A)
d
d
E
and


2

du0
d



du1
d

+ 2u0 u1 u30 = 0

25

(B)

Equation (A) is easily solved

r0
sin
d

u0 =

taking

0 = 0

Then subbing this into equation (B)



du1
1 r2
cos
+ sin u1 02 sin3 = 0
d
2d
Try a solution of the form
u1 = A + B sin + C cos2
B + sin (A C) + sin3 (C
B + 0, A = C, C =

1 r02
)=0
2 d2

1 r02
2 d2

1 r02
(1 + cos2 )
2 d2
r0

u = sin + (a + cos2 )
d
2
We require the total deflection in the asymptotic regions r (u 0).
r , as 1
r , as + 2
subbing these into our equation gives
u1 =

1 r2
r0
(1 ) +  02 (1 + 1 + O(2 ))
d
2d
r0
1 = 
d
2
1r
r0
0 = 2 +  02 (1 + 1 + O(2 ))
d
rd
r0
2 = 
d
2r0 t
4M
= 1 2 =
=
d
d

0=

4GM
c2 d

(total deflection angle (in standard units))

Take M = M , d = R , gives = 1.75. Observed in 1919 by Sir Arthur

Eddington during a solar eclipse.
26

6
6.1

Black Holes
Radial In-falling Photons

Consider an observer at rest relative to the source of the Schwarzschild gravitational field. The observers world line is r = constant, = constant, =
constant and


2M
2
d = 1
dt2
r
where is proper time

1
d
2M 2

= 1
dt
r
For r  2M , then along the observer is world line
d
= 1 t = (choosing (0) = 0)
dt
Therefore, t corresponds to proper time measured by an observer at rest
at infinity. How does such an observer see a radially in-falling photon as
r 2M ?
The world line of a radially in-falling photon satisfies
1



2M
2M
2
dr = 1
dt2
1
r
r

1
dt
2M

= 1
dr
r
where + represents an outgoing photon and - represents an ingoing photon.
Solving gives
t = (r + 2M log(r 2M ) + C)
u = t (r + 2M log(r 2M ))
= constant along radially null geodisics.
Clearly, as r 2M, t . i.e. an observer at infinity will never see
the photon cross the horizon (r = 2M ), according to this observer, it takes
an infinite amount of time to reach r = 2M .
dt
Note: As r , we have dr
= 1 t = r + c.
i.e. as r , ingoing and outgoing null rays are straight lines with angle
45 .

27

6.2

Radially In-falling Particles

A radially in-falling particle will move on a timelike geodesic given by



2M
1
t=E
r


1

2M 2
2M
1
t = 1
r 2 = 1
r
r
If we consider a particle initially at rest at infinity
E=1
Then the geodesic equations give

1 
1
2M
2M
1
+ 1
r 2 = 1
r
r


d
dr

2

r
2M
 r  21
=

d
=
dr
2M
(minus sign reflects the fact that the particle is ingoing.
Integrating, we obtain
3
3
2
2
2
(r
=
1
0 r )
3(2M ) 2

where the particle is at r0 at = 0 . Now the proper time to reach the

singularity r = 0 is
3
2
2
= 0 +
1 r0
3(2M ) 2
which is finite.
According to his clock, he passes through the coordinate singularity r =
2M continuously, and reaches the curvature singularity r = 0 in a finite
proper time.
If we now describe the motion in terms of coordinate time t (time measured by an observer at rest at infinity), then
r


t
r
2M
dt
= =
1
(E=1)
dr
r
2M
r
Integrating, we obtain
28

t t0 =

2 r 2 + 6M r 2 6M r 2 )
1 (r
0
0

3(2M ) 2

+ 2M log

(r 2 + (2M ) 2 )(r02 (2M ) 2 )

(r02 + (2M ) 2 )(r 2 (2M ) 2 )

t , as t 2M . So again, an observer at infinity never sees the
particle cross the horizon despite the fact that according to a clock attached
to the particle, it reaches the singularity in an infinite time. According to
his clock, he passes through the coordinate singularity r = 2M continuously,
and reaches the curvature singularity r = 0 in a finite proper time.
The path of a photon is always tangent to the local radial null cone and
here two photon paths passing through P define the local null. The paths of
massive particles are always inside the null cone. In there (t, r) coordinates,
the local radial null cones are closing as r 2M . For r < 2M , the null cones
tip over and we can see that an observer cannot remain at rest but is forced
to move towards the singularity.

6.3

The Kruskal Extension of the Schwarzschild Manifold

It is clear that the coordinates (t, r) are a bad choice for wollowing an infalling particle.
We look for new coordinates (u, v) in terms of which the local radial null
cones do not close as r 2M . We employ Kruskal coordinates:


 r
 12 r
t
1 e 4M cosh
u=
2M
4M


 12 r
 r
t
v=
1 e 4M sinh
2M
4M




1
 2 r
 21 r
r  r
t
1  r
t
4M
4M
1
1 e sinh
du =
e cosh
dr+
dt
8M 2 2M
4M
4M 2M
4M




 12 r
 12 r
r  r
t
1  r
t
dv =
1
e 4M sinh
dr+
1 e 4M cosh
dt
8M 2 2M
4M
4M 2M
4M
" 


1

r
r
2M
2m
2
2
2
2
2M
du dv =
e
1
dt + 1
dr
32M 3
r
r

29

32M 3 r
e 2M (du2 dv 2 ) + r2 (d2 + sin2 d2 )
r
where r = r(u, v) is defined implicitly by
 r
 r
u2 v 2 =
1 e 2M
2M
ds2 =

Remarks:
1. The line element is singular only at r = 0.
2. r = 0 u2 v 2 = 1 hyperbola with u, v as Cartesian coordinates.
r = constant > 2M u2 v 2 = positive constant
r = constant < 2M u2 v 2 = negative constant
dv
3. Null radial geodesics are now defined by du
= 1. i.e. in R(u, v) are
Cartesian coordinates with local null cones as straight lines at 45 . This
is because (u, v) was chosen to satisfy

2M
1
r

2M
dt + 1
r
2

1

dr2 = F (u, v)(du2 dv 2 )

4.

tanh

t
4M


=

v
.
u

t = constant v = constant.u
5. r = 2M u = v.
Conclusions from Space-Time Diagram in Kruskal Coordinates
r = 2M is a null-cone (null hypersurface) separating I and III from II
and IV respectively.
Massive particles and photons can cross the radius r = 2M from I to II
but not from II to I. Hence, r = 2M is a one-way membrane known as
the event horizon; it is the boundary of the Schwarzschild black hole.
Region I corresponds to the exterior Schwarzschild solution (r > 2M ),
describing the gravitational field outside a spherically symmetric object
of mass M .

30

 Region II represents a black hole solution. Observers inside this region

cannot send signals to an observer in region I and all observers in this
region are destined for the future singularity r = 0.
Region III is a universe whose geometry is identical to that of Region
I, though the two regions are causally disconnected, i.e. no communication either way is possible.
Region IV is the Schwarzschild white hole, the time reversal of a black
hole. Generally not thought to be physical since nature abhors naked
singularities.

7
7.1

Cosmology
The Cosmological Principle:

Our position in the univers, with respect to the largest scales, is in no sense
preferred. Extends the familiar Copernician Principle which states that our
position in the solar system is in no sense preferred.
The cosmological principle is modelled by asserting that the universe is
globally spatially isotropic. (globally isotropy implies homogeneity.)
Definition Homogeneous: For each space-like hypersurface, there are no
privileged points, i.e. each point is a centre of spherical symmetry.
Definition Isotropy: For each space-like hyperspace, there are no privileged
directions about any point.

7.2
7.2.1

Kinematics of the Continuum

Connecting Vectors

We consider the spacetime of the universe to be filled with the trajectories of

massive particles (the galaxies) forming a congruence of timelike world lines
with one world line passing through each point of the space time.
Let be a space like hypersurface, then the normal to is timelike and
vectors tangent to are spacelike.
Let { i }3i=3 be an intrinsic coordinate system on . Since there is a line
of the congruence passing through every point of , we can use the points of
to label the lines of the congruence.

31

The congruence is timelike, so we use proper time as parameter along

each worldline. Then for {xi }3=0 a chart on the manifold, the parametric
equations of the congruence are
x = x (, i );

= 0, ..., 3;

i = 1, ..., 3.

A particular line of the congruence is given by

i = constants;

x = x (, i ),

the unit tangent to this line is

u =

x
;

u u = 1

the 4-velocity of particle with worldline i .

Consider now two neighbouring lines of the congruence i and i + i .

is an infinitesimal connecting vector defined along i .

connects points of equal parameter value on i and i + i
= x (, i + i ) x (, i )
Taylor expanding for small j .
=

x (, i ) j

So see how varies along the line of congruence i , we differentiate with

respect to


x (, i )

=
j

t
j



x (, j )
= j
j

u j
=

32

But
u
u x
=
j
x j

x

= u, j j

= u ,
x
= u,

x
, u = u,
[, u] = 0
We also know that
X , Y Y , X = X ; Y Y ; X
The propagation equation may be rewritten
; u = u;
= u;
where

D
=
= ; u
d
We may also define the orthogonal connecting vector
= h
where h = + u u is the projection tensor which projects vectors orthogonal to u .
h = + u u
h = g + u u
h = g + u u

It is straight forward to verify that h satisfies

i. h u = 0
ii. h h = h
33

iii. h = 3
Arbitrary tensors may be projected onto the plane orthogonal to u , e.g. for
a ( 20 ) tensor Q , its projection orthogonal to u is
= h h Q
Q

It is trivial to see that
= 0 = u Q

u Q
7.2.2

Shear, Stress and Expansion

The 3-velocity of the line of congruence i + i relative to i is defined as

v = h
(3-velocity since v u = 0 implies only 3 independent components)
Lemma 7.2(a)
v = A where A = u; h

34

Proof
v = h
= h
= ( + u u )
= + u u
= + u u + u (u ).
= u ; + u u + u (u ).
v = h
= h u ; + h u u + h u (u ).
= ( + u u )u; + ( + u u )u u
= u; + u (u u ; ) + u; u u + u u u ; u u
= u; + u; u u
= u; + u; u u
= u; ( + u u )
= u; h
= u;
= u; h nu
= A
where A = u; h
A = u; h 
Lemma 7.2(b)
A may be written as
1
A = u; h = + + h
3
where

1
= u(;) + u ( u) u; h
3
is a symmetric, trace free ( 02 ) tensor known as the shear tensor.
= u[;] + u [ u]

is an antisymmetric ( 02 ) tensor known as the vorticity tensor. = u; is the

expansion or contraction of the congruence.
35

Proof
1
1
RHS = u(;) + u ( u) u; h + u[;] + u [ u] + u; h
3
3
= u; + u u
= u; + u u; u
= u; ( + u u )
= u; h = A

Conclusion
In going from to + along i , the 3-space of orthogonal connecting
vectors undergoes a linear transformation or
1. a shear, or distortion
2. a twist, or rotation
3. an expansion, or contraction.

7.3
7.3.1

The Friedman-Robertson-Walker (FRW) Metric

Isotropy

The model universe is a spacetime manifold (M, g) filled with a congruence

K of timelike worldlines k K. We assume spatial isotropy.
Definition: If (M, g) is isotropic with respect to a k K, then all
directions orthogonal to k at each point of k are equivalent, i.e. if is a
hypersurface orthogonal to the tangent to k, then there are no preferred
directions in .
Consequence of isotropy of k
1. u = 0 along k, i.e. k is a geodesic.
Proof
u u = 1
u u = 0

36

 at each point of k, u is a special vector orthogonal to u .

Isotropy no such vector exists u = 0.
2. = 0 along k.
Proof
is a 4x4 symmetric trace-free matrix which is orthogonal to u in
both of its indices.
u = u = 0
has 4 mutually orthogonal eigenvectors, one timelike and 3 spacelike, where u is the unit timelike eigenvector with eigenvalue 0. Hence,
the 3 unit spacelike eigenvectors constitute 3 unique directions orthogonal to u (i.e. lying in the hypersurface ) at each point of k. By
isotropy, no such directions = 0.
3. = 0 along k.
Proof

Let be the Levi-Civita symbol and take = g as the

Levi-Civita tensor. We define the vorticity vector by
= u
u = 0. There is a unique vector orthogonal to u .
Isotropy = 0 = 0.
4. h , = 0 along k.
Proof
u (h , ) = 0, hence h , is a unique vector orthogonal to u .
Isotropy h , = 0.
Definition: Isotropic universe:(M, g) is spatially isotropic if it is isotropic
with respect to all k K. Since a worldline k passes through every
point of M , this implies:
u


h ,
37

=0
=0
=0
=0

7.3.2

Comoving Coordinates

Isotropy implies
= u[;] + u [ u] = 0
and
u = 0
u[;] = 0
u, = u,
there exists a scalar function t(x ) such that u = t,
u dx = t, dx = dt
the worldlines k K intersect the t =constant hypersurfaces orthogonally.
Since
x
u =

and
dt
t x
=
d
x
= t, u
= u u
=1
=t

(taking the integration constant to be zero)

Hence t is the proper time along each k K, this is known as cosmic

time.
Taking {x } = (t, xi ) to be the coordinates on M , then xi =constant
label each k K. In these coordinates
x
x
=
= t

t
u = (1, 0, 0, 0).

u =

Since the 4-velocity in these coordinates has no spatial components, we say

the coordinates are co-moving with the matter.
Galaxies move on integral curves of the vector field

38

.
t

7.3.3

The Spacetime Metric

We have shown that isotropy yields

u g u = t,
which, in co-moving coordinates, implies
g t = t, = t
gt = t
gtt = tt = 1
gti = 0 i = 1, 2, 3.
The metric reads
ds2 = dt2 + gij (t, xi )dxi dxj
Next we consider
h , = 0
where
h = + u u
wich in comoving coordinates gives
( t t ), = 0
, = t ,t
,i = 0
= (t)
Finally, we have that
= 0
The orthogonal connecting vector of two neighbouring galaxies x and x +
x in comoving coordinates is
= (0, xi ).
Let l be the invariant length of this vector
l2 = g = gij xi xj
2ll =

gij i j
x x
t

39

(i, j = 1, 2, 3)
(A)

We may also write = ln where n is a unit spacelike vector. Recall that

h = A
where

1
A = + + h
3

In our case = 0.
1
h = + h
3
+ h ln = ln + 1 h ln
h ln

Multiplying by n and using n n = 1 and n u = 0

1
l + ( + u u )ln n = ln n + ( + u u )ln n
3
+ u u = 0
ln n = 0

(since n n = 0)

(since anti-symmetric, n n symmetric)

1
l = l
3

(B)

where = (t) in comoving coordinates.

Subbing (B) into (A)
2
gij i j
(t)l2 =
x x
3
t
2
gij i j
(t)gij xi xj =
x x
3
t
2
gij
(t)gij =
3
t
A separable solution of the form gij = hij (xi )l2 (t) satisfies this equation, so
the metric now reads
ds2 = dt2 + l2 (t)hij (xi )dxi dxj
where l(t) satisfies
1
l = (t)l
3
and hij is a positive definite metric on a Riemannian 3-manifold which is
isotropic at each of its points. This is consistent with asserting that hij is
a positive definite 3-metric of constant curvature. There are only 3 distinct
possibilities: R3 , S3 , H3 .
40

1. Flat Space R3 : This is the familiar infinite Euclidean geometry

(3)

dx2 = hij dxi dxj

= dx2 + dy 2 + dz 2
= dr2 + r2 (d2 + sin2 d2 )

2. Three-Sphere S3 : The compact space of constant curvature. To visualise this, we embed S3 in R4 with points on S3 satisfying
x 2 + y 2 + z 2 + w 2 = a2
(3) ds2 = dx2 + dy 2 + dz 2 + dw2
(xdx + ydy + zdz)2
= dx2 + dy 2 + dz 2 + 2
a x2 y 2 z 2
which, in spherical polar coordinates, yields
(3)

ds2 = dr2 + r2 (d2 + sin2 d2 ) +

=

r2 dr2
a2 r 2

dr2
+ r2 (d2 + sin2 d2 )
r2
1 a2
0ra
0
0 2

3. Hyperbolic Space H3 : The infinite space of constant negative curvature.

To visualise this, we embed H3 in a 4 dimensional Lorentzian space
x2 + y 2 + z 2 w2 = a2
which yields
(3)

ds2 =

dr2
2
2
2
2
2 + r (d + sin d )
1 + ar 2

Scaling the radial variable

r ra
then we may rewrite the general form of the 3pmetric of constant curvature as
h dr2
i
2
2
2
2
(3)
2
2
ds = a
+ r (d + sin d )
1 kr2
41

where
k = +1 S3 (closed)
0 R3 (flat,open)
1 H3 (open)
The 4D metric in comoving coordinates is therefore
ds2 = dt2 + l2 (t)a2

h dr2
i
2
2
2
2
+
r
(d
+
sin
d
)
1 kr2

Finally, taking a(t) = l(t)a

h dr2
i
2
2
2
2
ds2 = dt2 + a2 (t)
+
r
(d
+
sin
d
)
1 kr2
The Friedmann-Robertson-Walker metric, where a(t) satisfies
a
1
= (t)
a
3
and is known as the scale factor.
Another particularly useful form of the FRW metric is obtained by the
transformation
dr2
d2 =
1 kr2
ds2 = dt2 + a2 (t)[d2 + f 2 ()(d2 sin2 d2 )]
where
f () = sin 0 < < S3
0
R3
sinh 0 < H3
The scale factor a(t) is determined by the Einstein field equations.

7.4

Cosmological Red-Shift and Hubbles Law

Consider light emitted by a galaxy with world line

r + r0 , = 0 , = 0

42

Suppose this light is received by our galaxy whose world line is

r + r1 , = 1 , = 1
Along photon worldline P0 P1 and Q0 Q1
ds = 0, = 0 , = 0
dt2
dr2
=
a2 (t)
1 kr2

Assuming, without loss of generality, that r1 > r0 and a(t) > 0

dr
dt
=
a(t)
1 kr2

If n photons are emitted by r = r0 in proper time dt0 , and received by r = r1

in proper time dt, then the emitted frequency is 0 = dtn0 while the received
frequency is 1 = dtn1
0
1
dt1

=
=
1
0
dt0
Integrating along P0 P1 and Q0 Q1 gives
Z t1
Z r1
dt
dr

=
1 kr2
t0 a(t)
r0
and
Z

t1 +dt1

dt
=
a(t)

t0 +dt0

t1

t0

r1

dt
=
a(t)

r0

dr
1 kr2

t1 +dt1

t0 +dt0

dt
a(t)

writing
df (t)
dt
f (t1 ) f (t0 ) f (t1 + dt1 ) + f (t0 + dt0 ) = 0
a1 (t) =

Taylor expanding, keeping only first order terms

 
 
df
df
dt1 +
dt0 = 0
dt t=t1
dt t=t0

dt1
a(t1 )
1
=
=
dt0
a(t0 )
0
43

Observations of distant galaxies a red-shift

1 > 0
a(t1 ) > a(t0 )
a > 0
(in the present epoch)
3a
=
>0
a
the universe is expanding.
Galaxies are receeding away from one another at a rate proportional to
the distance between them-Hubbles Law (1929).
Hubble Parameter
a(t)

H(t) =
a(t)

7.5

Einsteins Equations

The scale factor a(t) is determined by considering Einsteins field equations

with a cosmological constant term
1
R Rg + g = 8T
2
7.5.1

Matter Content of the Universe

Consistent with spatial homogeneity and isotropy in a comoving frame with

n = (1, 0, 0, 0) the e.m.s. tensor takes the form of a perfect fluid
T = ( + p)u u + pg
Also, isotropy = (t), p = p(t).
Conservation of e.m. T , = 0 gives constraint
+

3a
( + p) = 0
a

where
= proper energy density
p = isotropic pressure
To solve this, we further require an equation of state, we assume a barotropic
fluid of the form
P = ( 1)
(= a constant)
Two common cases are matter/radiation domination:
44

1. Dust: = 1, p = 0, i.e. pressureless not interacting matter

3a
=

a
3
a
0 a3
= 30
a

2. Radiation: = 43 , p =
4a

a
4
a
0 a4
= 40
a

7.5.2

(dilution of energy to expansion and redshift)

The Friedman and Raychauduri Equations

Non-zero components of the FRW-metric:

Rtt = 3

 2
a
ki
+2
+ 2 2 gij
Rij =
a
a
a
ha

R = g tt Rtt + g ij Rij
 2
ha
a

ki
=6 +2
+ 2
a
a
a
The tt-component of Einsteins equations
1
RH Rgtt + gtt = 8Ttt
2
   2
h
a

a
ki
3 + 3
+
+ 2 = 8
a
a
a
a
 2
a
k

8
+ 2 =
(Friedman Equation)

a
a
3
3

45

 ij-component
a

2 +
a

 2
a
k
+ 2 = 8p
a
a

Subtracting Friedmans equations

4
a

=
( + 3p)
a
3
3

(Raychauderi Equation)

Standard cosmological models are obtained by solving these for some equation of state.

7.6

Cosmological Models with Vanishing

 2
a
k
8
+ 2 =
a
a
3
 
a

4
=
( + 3p)
a
3

7.6.1

Geometry of the Universe

Critical density c : energy density of flat universe

 2
dH 2
3 a
(in theory, H0 is measurable)
c =
=
8 a
8
If we now define the density parameter of the universe to be
=
Friedman

k
a2 H

=1

k = +1 > 1 (closed)
0 = 1 (flat,open)
1 < 1 (open)
The actual geometry is still hotly debated.

46

7.6.2

Big Bang Singularities

Assuming the matter content of the universe obeys the strong energy condition (SEC)
1
T u u T
2

for any time-like vector u and T = T .

In our FRW metric with perfect fluid matter distribution, this implies
+ 3P 0
This in turn implies (from the Raychaudhuri equation)
a

0
a
H + H 2 0
Z
Z
dH

dt
H2
1
1
+

(t t0 )
H
H0
1
H 1
H0 + (t t0 )
Using the fact that H = aa and integrating again, we get
Z
Z
da
1

dt
1
a
H0 + (t t0 )

a(t) a(t0 )

H01 + (t t0 )
H01

i.e. a(t) is bounded above by a linear function of t, which has t-intercept at

t = t0 h1
0 .
For some finite time t > t0 H01 , we must have a(t) = 0. But as a(t)
0, , we have an infinite energy state, and all known laws of physics
breaks down. (Singularity Theorems-Hawking,Penrose,Geroch)
All FRW moels with = 0 with a matter distribution satisfying the SEC
predict a Big Bang singularity at some finite time in the past.

47

7.6.3

The Fate of the Universe (Eschatology)

(Matter Domination P = 0).

We define a conformal time by
dt
a

d =

and we denote derivatives with respect to conformal time by , i.e.

a0 =
Defining H =

a0
,
a

da
d

then

a
a0 d
H
=
=
a
a dt
a
The Friedmann equation in conformal time becomes
H=

H2 + k =

8a2
3

and the Raychaudhuri equation becomes

4
H0
=
2
a
3

(for dust P = 0)

Combining the two to eliminate , we obtain

2H0 + H2 + k = 0
For an open universe, k = 1: We have
Z
Z
1
dH
=
d
1 H2
2
This may be solved with the substitution
u
H = coth
 2 
H = coth
2
1  
1
0
H =
=
2
1 cosh
2 sinh 2
From the Raychaudhuri equation
H0
4
40 a30
=
=
a
3
3a3
48

 a( ) =

4
0 a30 (1 cosh )
3

We have
0 =

80
3H02

1
a( ) = 0 H02 a30 (1 cosh )
2
We also have
a20 H02 =

k
0 1
3

a30 = H03 (1 ) 2 )
3
1
a( ) = 0 H01 (1 0 ) 2 (cosh 1)
2

(A)

We also have
dt = ad
3
1
t = 0 H01 (1 0 ) 2
2

Z
(cosh 1)d

3
1
t = 0 H01 (1 0 ) 2 (sinh )
2

(B)

(A) and (B) together form a parametric solution for an open k = 1 universe.
As t , the k = 1 universe expands forever. As a(t) gets larger and
larger, in the Friedmann equation
 2
80 a30
a
1
2 =
a
a
3a3
the curvature term

1
a3

dominates the matter term

a3 

|k|
a2

(for a )

a
1
2
a
a
a(t) t

The universe undergoes a period of free expansion as the density drops off
rapidly.
t3

49

The Linear Approximation

8.1

The Einstein Equations in the Linear Approximation

We begin with the assumption that there exists coordinates in which the
metric of a weak gravitational field can be written as
g = + h
where
|h |  1
Introduce a fictitious book-keeping parameter
g = + h
We neglect all O(2 ) and higher terms. We further assume the boundary
condition
lim h = 0
r

i.e. the spacetime is asymptotically flat.

If we think of h as a tensor on , then we can raise indices of h by
contracting with , e.g.
h = h
It is easy to verify that the metric inverse (to first order) is
g = h
To see this, we check that
g g = + O(2 )
. The Christoffell symbols in the linear approximation are
1
= g (g, + g, g, )
2
1
= ( h )(h, + h, h, )
2
1
= (h , + h, h, ) + O(2 )
2

50

The Riemann curvature tensor is given by

R = g R
= ( + h )(, , + )
h1
i
1
= (h, + h, h, ) (h, + h, h, )
2
2
1
= (h, + h, h, h, ) + O(2 )
2
The Ricci tensor components in the linear approximation are
R = g R
1
= (h, + h, h, h, ) + O(2 )
2
1
= (h , + h, h, h, ) + O(2 )
2
where h = h and
 = = =

2
2
2
2
+
+
+
t2
x2 y 2 z 2

The Ricci scalar in the linear approximation is

1
R = ( h ) (h, + h, h, h ) + O(2 )
2

= (h , h)
Finally, the Einstein tensor in the linear approximation is
1
G = R Rg
2
1
1
= (h , + h, h, h ) (h, h) + O(2 )
2
2
1
G = (h , + h, h, h h, + h)
2
The linearised vacuum field equations would involve setting this to be zero
and solving for h . It is convenient to write the Einstein tensor in terms of
a new dependent tensor. the star conjugate of h
1
h = h h
2
h = h
51

(1)

Note that

1
h = h = h (4)h = h
2
Using this to invert (1)
1
h = h h
2
Note also, we can perform a double star conjugate
h
= h
In terms of h , the Einstein tensor reads
G

1 h
1
1
=  (h h ), + (h h ), + h,
2
2
2
i
1

1
(h h ) (h h ), h )
2
2

8.2

i
1 h

=  h + h , + h , h ,
2

Gauge Transformations

Let us consider coordinate transformations of the form

x x = x + 
(Sacrificed general covariance, considering only coordinate transformations
close to the identity).
Under this transformation
g =

x x
g
x x

But
x = x (x )
= x (
x (x ))
= x +  (
x) + O(2 )
x
=  , + O(2 )

x
g = g , , + O(2 )

52

and since

g = + h
= + (h , , )
i.e.
= h 2(,)
h

(Gauge Transformation of h )

We can check that

= R
R
i.e. the Riemann curvature tensor (and hence the Einstein tensor) are gauge
invariant (to first order).

8.3

The Newtonian Limit

8.4

Gravitational Waves

8.4.1

The Linearised Vacuum Solutions

8.4.2

Energy Transfer

Einstein-Maxwell Theory

9.1

The Field Equations

The variables of Einstein-Maxwell theory are:

The metric g
Maxwell tensor F
Stress-energy tensor T
Current 4-vector J
If is the proper electric density and u the 4-velocity of the charge, then
J = u
At each point p M , we construct an orthonormal tetrad
{(a) }3a=0

53

where
(0) (0) = 1
(0) (i) = 0

(i = 1, 2, 3)

(i) (j) = (i)(j)

(i, j = 1, 2, 3)

Then (a) (b) = (a)(b) = diag(1, 1, 1, 1). The orthonormal tetrad of the
Maxwell tensor are
F(a)(b) = F (a) (b) = F(b)(a)
These components define the electric and

0 E1
E 1
0
F(a)(b) =
E2 B3
E3 B2

magnetic 3-vectors

E2 E3
B3 B2

0
B1
B1
0

~ = (E1 , E2 , E3 )
E
~ = (B1 , B2 , B3 )
B
Maxwells equations in flat space are
~ B
~ t E
~ = J~

~ E
~ =

~ E
~ t B
~ =0

~ =0
~ B

These may be written in tensor notation as

F , = J

(i)

F, + F, + F, = 0

(ii)

~
where J = (, J)
F 0i = E i
Fij = ijk Bk

(i = 1, 2, 3)
(i, j, k = 1, 2, 3)

The covariant generalisations of these equations are obtained by the comma

goes to semi-colon rule:

54

F ; = J
F, + F, + F, = 0

(Maxwells equations in curved spacetime)

It is easy to prove that the second equation above is equivalent to (ii) and
hence there exists a 4-potential A such that
F = A, A, = A; A;
For source-free regions (J = 0) in a vacuum, we still have a contribution to
the energy-momentum tensor coming from the Maxwell tensor. The electromagnetic Lagrangian density is defined by

9.2

The Reissner-Nordstrom Solution

55