Physik-Institut der Universität Zürich

in conjunction with ETH Zürich

General Relativity
Autumn semester 2016

Prof. Philippe Jetzer

Original version by Arnaud Borde

Revision: Antoine Klein, Raymond Angélil, Cédric Huwyler, Simone Balmelli, Yannick Boetzel

Last revision of this version: January 4, 2017

Sources of inspiration for this course include

• S. Carroll, Spacetime and Geometry, Pearson, 2003

• S. Weinberg, Gravitation and Cosmology, Wiley, 1972

• N. Straumann, General Relativity with applications to Astrophysics, Springer Verlag, 2004

• C. Misner, K. Thorne and J. Wheeler, Gravitation, Freeman, 1973

• R. Wald, General Relativity, Chicago University Press, 1984

• T. Fliessbach, Allgemeine Relativitätstheorie, Spektrum Verlag, 1995

• B. Schutz, A first course in General Relativity, Cambridge, 1985

• R. Sachs and H. Wu, General Relativity for mathematicians, Springer Verlag, 1977

• J. Hartle, Gravity, An introduction to Einstein’s General Relativity, Addison Wesley, 2002

• H. Stephani, General Relativity, Cambridge University Press, 1990, and

• M. Maggiore, Gravitational Waves: Volume 1: Theory and Experiments, Oxford University

Press, 2007.

• A. Zee, Einstein Gravity in a Nutshell, Princeton University Press, 2013

As well as the lecture notes of

• Sean Carroll ([Link]

• Matthias Blau ([Link]

• Gian Michele Graf

2
 CONTENTS

Contents

I Introduction 6

1 Newton’s theory of gravitation 6

2 Goals of general relativity 7

II Special Relativity 9

3 Lorentz transformations 9
3.1 Galilean invariance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2 Lorentz transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.3 Proper time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

4 Relativistic mechanics 13
4.1 Equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2 Energy and momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.3 Equivalence between mass and energy . . . . . . . . . . . . . . . . . . . . . . . . . . 14

5 Tensors in Minkowski space 14

6 Electrodynamics 17

7 Accelerated reference systems in special relativity 18

III Towards General Relativity 20

8 The equivalence principle 20

8.1 About the masses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
8.2 About the forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
8.3 Riemann space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

9 Physics in a gravitational field 25

9.1 Equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
9.2 Christoffel symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
9.3 Newtonian limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

10 Time dilation 28
10.1 Proper time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
10.2 Redshift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
10.3 Photon in a gravitational field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3
 CONTENTS

11 Geometrical considerations 30
11.1 Curvature of space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

IV Differential Geometry 32

12 Differentiable manifolds 32
12.1 Tangent vectors and tangent spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
12.2 The tangent map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

13 Vector and tensor fields 37

13.1 Flows and generating vector fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

14 Lie derivative 40

15 Differential forms 42
15.1 Exterior derivative of a differential form . . . . . . . . . . . . . . . . . . . . . . . . . 43
15.2 Stokes theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
15.3 The inner product of a p-form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

16 Affine connections: Covariant derivative of a vector field 48

16.1 Parallel transport along a curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
16.2 Round trips by parallel transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
16.3 Covariant derivatives of tensor fields . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
16.4 Local coordinate expressions for covariant derivative . . . . . . . . . . . . . . . . . . 55

17 Curvature and torsion of an affine connection, Bianchi identities 57

17.1 Bianchi identities for the special case of zero torsion . . . . . . . . . . . . . . . . . . 59

18 Riemannian connections 60

V General Relativity 64

19 Physical laws with gravitation 64

19.1 Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
19.2 Electrodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
19.3 Energy-momentum tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

20 Einstein’s field equations 66

20.1 The cosmological constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

21 The Einstein-Hilbert action 68

4
 CONTENTS

22 Static isotropic metric 71

22.1 Form of the metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
22.2 Robertson expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
22.3 Christoffel symbols and Ricci tensor for the standard form . . . . . . . . . . . . . . 72
22.4 Schwarzschild metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

23 General equations of motion 74

23.1 Trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

VI Applications of General Relativity 80

24 Light deflection 80

25 Perihelion precession 83
25.1 Quadrupole moment of the Sun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

26 Lie derivative of the metric and Killing vectors 87

27 Maximally symmetric spaces 88

28 Friedmann equations 91

5
 1 NEWTON’S THEORY OF GRAVITATION

Part I

Introduction
1 Newton’s theory of gravitation
In his book Principia in 1687, Isaac Newton laid the foundations of classical mechanics and made a
first step in unifying the laws of physics.

The trajectories of N point masses, attracted to each other via gravity, are the solutions to the equation
of motion
N
d2~ri X mi mj (~ri − ~rj )
mi = −G i = 1 . . . N, (1.1)
dt2 j=1
|~ri − ~rj |3
j6=i

with ~ri (t) being the position of point mass mi at time t. Newton’s constant of gravitation is determined
experimentally to be
G = 6.6743 ± 0.0007 × 10−11 m3 kg−1 s−2 (1.2)

The scalar gravitational potential φ(~r) is given by

N
ρ(~r 0 )
Z
X mj
φ(~r) = −G = −G d3 r0 , (1.3)
j=1
|~r − ~rj | |~r − ~r 0 |

where it has been assumed that the mass is smeared out in a small volume d3 r. The mass is given by
dm = ρ(~r 0 )d3 r0 , ρ(~r 0 ) being the mass density. For point-like particles we have ρ(~r 0 ) ∼ mj δ (3) (~r 0 −~rj ).
The gradient of the gravitational potential can then be used to produce the equation of motion:
d2~r
m = −m∇φ(~r). (1.4)
dt2
According to (1.3), the field φ(~r) is determined through the mass of the other particles. The corre-
sponding field equation derived from (1.3) is given by1

∆φ(~r) = 4πGρ(~r) (1.5)

The so called Poisson equation (1.5) is a linear partial differential equation of 2nd order. The source of
the field is the mass density. Equations (1.4) and (1.5) show the same structure as the field equation
of electrostatics:
∆φe (~r) = −4πρe (~r), (1.6)

and the non-relativistic equation of motion for charged particles

d2~r
m = −q∇φe (~r). (1.7)
dt2
Here, ρe is the charge density, φe is the electrostatic potential and q represents the charge which acts
as coupling constant in (1.7). m and q are independent characteristics of the considered body. In
1
1∆
|~
r −~r 0|
= −4πδ (3) (~ r 0)
r−~

6
 2 GOALS OF GENERAL RELATIVITY

analogy one could consider the “gravitational mass” (right side) as a charge, not to be confused with
the “inertial mass” (left side). Experimentally, one finds to very high accuracy (∼ 10−13 ) that they
are the same. As a consequence, all bodies fall at a rate independent of their mass (Galileo Galilei).
This appears to be just a chance in Newton’s theory, whereas in GR it will be an important starting
point.
For many applications, (1.4) and (1.5) are good enough. It must however be clear that these
equations cannot be always valid. In particular (1.5) implies an instantaneous action at a distance,
what is in contradiction with the predictions of special relativity. We therefore have to suspect that
Newton’s theory of gravitation is only a special case of a more general theory.

2 Goals of general relativity

In order to get rid of instantaneous interactions, we can try to perform a relativistic generalization of
Newton’s theory (eqs. (1.4) and (1.5)), similar to the transition from electrostatics (eqs. (1.6) and
(1.7)) to electrodynamics.

The Laplace operator ∆ is completed such as to get the D’Alembert operator (wave equation)

1 ∂2
∆⇒= −∆ (2.1)
c2 ∂t2
Changes in ρe travel with the speed of light to another point in space. If we consider inertial coordinate
frames in relative motion to each other it is clear that the charge density has to be related to a
current density. In other words, charge density and current density transform into each other. In
electrodynamics we use the current density j α (α = 0, 1, 2, 3):

ρe → (ρe c, ρe v i ) = j α , (2.2)

where the v i are the cartesian components of the velocity ~v (i = 1, 2, 3). An analogous generalization
can be performed for the potential:
φe → (φe , Ai ) = Aα . (2.3)

The relativistic field equation is then

4π α
∆φe = −4πρe → Aα = j . (2.4)
c
In the static case, the 0-component reduces to the equation on the left.
Equation (2.4) is equivalent to Maxwell’s equations (in addition one has to choose a suitable gauge
condition). Since electrostatics and Newton’s theory have the same mathematical structure, one may
want to generalize it the same way. So in (1.5) one could introduce the change ∆ → . Similarly one
generalizes the mass density. But there are differences with electrodynamics. The first difference is
that the charge q of a particle is independent on how the particle moves; this is not the case for the
mass: m = pm0 2 .
1− vc2
As an example, consider a hydrogen atom with a proton (rest mass mp , charge +e) and an electron
(rest mass me , charge −e). Both have a finite velocity within the atom. The total charge of the atom

7
 2 GOALS OF GENERAL RELATIVITY

is q = qe + qp = 0, but for the total mass we get mH 6= mp + me (binding energy). Formally this
means that charge is a Lorentz scalar (does not depend on the frame in which the measurement is
performed). Therefore we can assign a charge to an elementary particle, and not only a “charge at
rest”, whereas for the mass we must specify the rest mass.
δq
Since charge is a Lorentz scalar, the charge density (ρe = δV ) transforms like the 0-component
1 √ 1
of a Lorentz vector ( δV gets a factor γ = 2 2
due to length contraction). The mass density
1−v /c
δm
(ρ = δV ) transforms instead like the 00-component of a Lorentz tensor, which we denote as the
energy-momentum tensor Tαβ . This follows from the fact that the energy (mass is energy E = mc2 )
is the 0-component of a 4-vector (energy-momentum vector pα ) and transforms as such. Thus, instead
of (2.2), we shall have !
ρc2 ρcv i
ρ⇒ ∼ T αβ i, j = 1, 2, 3 (2.5)
ρcv i ρv i v j
This implies that we have to generalize the gravitational potential φ to a quantity depending on 2
indices which we shall call the metric tensor g αβ . Hence we get

∆φ = 4πGρ ⇒ g αβ ∼ GT αβ . (2.6)

In GR one finds (2.6) for a weak gravitational field (linearized case), e.g. used for the description of
gravitational waves.
Due to the equivalence between mass and energy, the energy carried by the gravitational field is
also mass and thus also a source of the gravitational field itself. This leads to non-linearities. One can
note that photons do not have a charge and thus Maxwell’s equations can be linear.
To summarize:

1. GR is the relativistic generalization of Newton’s theory. Several similarities between GR and

electrodynamics exist.

2. GR requires tensorial equations (rather than vectorial as in electrodynamics).

3. There are non-linearities which will lead to non-linear field equations.

8
 3 LORENTZ TRANSFORMATIONS

Part II

Special Relativity
3 Lorentz transformations
A reference system with a well defined choice of coordinates is called a coordinate system. Inertial
reference systems (IS) are (from a “practical” point of view) systems which move with constant speed
with respect to distant (thus fixed) stars in the sky. Newton’s equations of motion are valid in IS. Non-
IS are reference systems which are accelerated with respect to an IS. In this chapter we will establish
how to transform coordinates between different inertial systems.

3.1 Galilean invariance

Galilei stated that “all IS are equivalent”, i.e. all physical laws are valid in any IS: the physical laws
are covariant under transformations from an IS to another IS’. Covariant means here form invariant.
The equations should have the same form in all IS.
With the coordinates xi (i = 1, 2, 3) and t, an event in an IS can be defined. In another IS’, the
same event has different coordinates x0i and t0 . A general Galilean transformation can then be written
as:

x0i = αi k xk + v i t + ai , (3.1)

t0 = t + τ, (3.2)

where

• xi , v i and ai are cartesian components of vectors

• ~v = v i~ei where ~ei is a unit vector

• we use the summation rule over repeated indices: αi k xk = α i k xk

P
k

• latin indices run on 1,2,3

• greek indices run on 0,1,2,3

• ~v is the relative velocity between IS and IS’

• ~a is a constant vector (translation)

• αi k is the relative rotation of coordinates systems, α = (αi k ) is defined by

αi n (αT )nk = δ i k or ααT = I i.e. α−1 = αT (3.3)

9
 3 LORENTZ TRANSFORMATIONS

The condition ααT = I ensures that the line element

ds2 = dx2 + dy 2 + dz 2 (3.4)

remains invariant. α can be defined by giving 3 Euler angles. Eqs. (3.1) and (3.2) define a 10
(a = 3, v = 3, τ = 1 and α = 3) parametric group of transformations, the so-called Galilean
group.

The laws of mechanics are left invariant under transformations (3.1) and (3.2). But Maxwell’s equations
are not invariant under Galilean transformations, since they contain the speed of light c. This led
Einstein to formulate a new relativity principle (special relativity, SR): All physical laws, including
Maxwell’s equations, are valid in any inertial system. This leads us to Lorentz transformations (instead
of Galilean), thus the law of mechanics have to be modified.

3.2 Lorentz transformations

We start by introducing 4-dimensional vectors, glueing time and space together to a spacetime. The
Minkowski coordinates are defined by

x0 = ct, x1 = x, x2 = y, x3 = z. (3.5)

xα is a vector in a 4-dimension space (or 4-vector). An event is given by xα in an IS and by x0α in an

IS’. Homogeneity of space and time imply that the transformation from xα to x0α has to be linear:

x0α = Λα β xβ + aα , (3.6)

where aα is a space and time translation. The relative rotations and boosts are described by the 4 × 4
matrix Λ. Linear means in this context that the coefficients Λα β and aα do not depend on xα . In
order to preserve the speed of light appearing in Maxwell’s equations as a constant, the Λα
β have to be
such that the square of the line element

ds2 = ηαβ dxα dxβ = c2 dt2 − d~r2 (3.7)

remains unchanged, with the Minkowski metric

 
1 0 0 0
 
0 −1 0 0
ηαβ =  . (3.8)
0
 0 −1 0

0 0 0 −1

Because of ds2 = ds02 ⇔ c2 dτ 2 = c2 dτ 02 , the proper time is an invariant under Lorentz transformations.
2
+dy 2 +dz 2
 x 2
Indeed for light dτ 2 = dt2 − dx
 
= 0. Thus, c2 =  d~  and c =  d~x . Applying a Lorentz
 0 c2 dt dt
 x
transformation results in c =  d~
dt 0 . This has the important consequence that the speed of light c is
the same in all coordinate systems (what we intended by the definition of (3.7)).

10
 3 LORENTZ TRANSFORMATIONS

A 4-dimensionial space together with this metric is called a Minkowski space. Inserting (3.6) into the
invariant condition ds2 = ds02 gives

ds02 = ηαβ dx0α dx0β

= ηαβ Λα γ dxγ Λβ δ dxδ

= ηγδ dxγ dxδ = ds2 . (3.9)

Then we get
Λα γ Λβ δ ηαβ = ηγδ or ΛT ηΛ = η. (3.10)
Rotations are special subcases incorporated in Λ: x0α = Λα β xβ with Λi k = αi k , and Λ0 0 = 1,
Λi 0 = Λ0 i = 0. The entire group of Lorentz transformations (LT) is the so called Poincaré group (and
has 10 parameters). The case aα 6= 0 corresponds to the Poincaré group or inhomogeneous Lorentz
group, while the subcase aα = 0 can be described by the homogeneous Lorentz group. Translations and
rotations are subgroups of Galilean and Lorentz groups.
Consider now a Lorentz ’boost’ in the direction of the x-axis: x02 = x2 , x03 = x3 . v denotes the
relative velocity difference between IS and the boosted IS’. Then
 
Λ0 0 Λ0 1 0 0
 1
Λ 0 Λ1 1 0 0

Λα β = 
 0
. (3.11)
 0 1 0 
0 0 0 1

Evaluating eq. (3.10):

(γ, δ) = (0, 0) (Λ0 0 )2 − (Λ1 0 )2 = 1 (3.12a)

= (1, 1) (Λ0 1 )2 − (Λ1 1 )2 = −1 (3.12b)

= (0, 1) or (1, 0) Λ0 0 Λ0 1 − Λ1 0 Λ1 1 = 0 (3.12c)

The solution to this system is

! !
Λ0 0 Λ0 1 cosh Ψ − sinh Ψ
= . (3.13)
Λ1 0 Λ1 1 − sinh Ψ cosh Ψ

For the origin of IS’ we have x01 = 0 = Λ1 0 ct + Λ1 1 vt. This way we find
Λ1 0 v
tanh Ψ = − = , (3.14)
Λ0 0 c
and as a function of velocity:
1
Λ0 0 = Λ1 1 = γ = q , (3.15a)
v2
1− c2

−v/c
Λ0 1 = Λ1 0 = q . (3.15b)
2
1 − vc2

11
 3 LORENTZ TRANSFORMATIONS

A Lorentz transformation (called a boost) along the x-axis can then be written explicitly as
x − vt
x0 = q , (3.16a)
2
1 − vc2

y 0 = y, (3.16b)

z 0 = z, (3.16c)

ct − x vc
ct0 = q , (3.16d)
2
1 − vc2

which is valid only for |v| < c. For |v|  c, (3.16) recovers the special (no rotation) Galilean transfor-
mation x0 = x − vt, y 0 = y, z 0 = z and t0 = t. The parameter
v
Ψ = arctanh (3.17)
c
is called the rapidity. From this we find for the addition of parallel velocities:

Ψ = Ψ1 + Ψ2

v1 + v2
⇒v= (3.18)
1 + v1c2v2

3.3 Proper time

The time coordinate t in IS is the time shown by clocks at rest in IS. We next determine the proper
time τ shown by a clock which moves with velocity ~v (t). Consider a given moment t0 an IS’, which
moves with respect to IS with a constant velocity ~v0 (t0 ). During an infinitesimal time interval dt0 the
clock can be considered at rest in IS’, thus:
r
0 v02
dτ = dt = 1− dt. (3.19)
c2
q
0 t(1−v02 /c2 ) v02
Indeed (3.16) with x = v0 t gives t = q =t 1− c2 and thus (3.19).
v2
1− c20
At the next time t0 + dt, we consider an IS” with velocity ~v0 = ~v (t0 + dt) and so on. Summing up
all infinitesimal proper times dτ gives the proper time interval:

Zt2 r
v 2 (t)
τ= dt 1 − (3.20)
c2
t1

This is the time interval measured by an observer moving at a speed v (t) between t1 and t2 (as given
by a clock at rest in IS). This effect is called time dilation.

12
 4 RELATIVISTIC MECHANICS

4 Relativistic mechanics
Let us now perform the relativistic generalization of Newton’s equation of motion for a point particle.

4.1 Equations of motion

The velocity ~v can be generalized to a 4-velocity vector uα :
dxi dxα
vi = → uα = (4.1)
dt dτ
Since dτ = ds
c , dτ is invariant. With dx0α = Λα β α α
β dx it follows that u transforms like dx :

u0α = Λα β uβ (4.2)

All quantities which transforms this way are Lorentz vectors or form-vectors. The generalized equation
of motion is then
duα
m = f α. (4.3)
dτ
α
du
Both dτ and f α are Lorentz vectors, therefore, (4.3) is a Lorentz vector equation: if we perform a
0α
0α
Lorentz transformation, we get m du dτ = f . Eq. (4.3) is covariant under Lorentz transformations
d~v


 v  c itreduces
and for  to Newton’s equations. (left hand side becomes m 0, dt and the right hand
0 ~ ~
side f , f = 0, K ). The Minkowski force f 0α is determined in any IS through a corresponding
 2
−1/2 1
LT: f 0α = Λα β f β . For example ~v = −v~e1 with γ = 1 − vc2 , leads to f 00 = γvK
c , f
01
= γK 1 ,
f 02 = K 2 and f 03 = K 3 . For a general direction of velocity (−~v ) we get:
~
~v · K ~
f 00 = γ , ~ + (γ − 1) ~v ~v · K .
f~0 = K (4.4)
c v 2

4.2 Energy and momentum

α
The 4-momentum pα = muα = m dx
dτ is a Lorentz vector. With (3.19) we get
 
i
 
mc mv  = E , p~ .
pα =  q ,q (4.5)
2
1 − vc2
2
1 − vc2 c

This yields the relativistic

mc2
energy : E=q = γmc2 (4.6a)
v2
1 − c2

m~v vc
momentum : p~ = q = γm~v −−−→ p~ = m~v . (4.6b)
2
1 − vc2

With (4.4), the 0-component of (4.3) becomes (in the case v  c)

dE ~
= ~v .K . (4.7)
dt |{z}
power given
to the particle

13
 5 TENSORS IN MINKOWSKI SPACE

This justifies to call the quantity E = γmc2 an energy. From ds2 = c2 dτ 2 = ηαβ dxα dxβ it follows
ηαβ pα pβ = m2 c2 and thus
E 2 = m2 c4 + c2 p~ 2 , (4.8)

the relativistic energy-momentum relation. The limit cases are


p mc2 + p2 v  c or p  mc2
2 4 2 2 2m
E = m c + c p~ ≈ (4.9)
cp v ∼ c or p  mc2

with p = |~
p|. For particles with no rest mass (photons): E = cp (exact relation).

4.3 Equivalence between mass and energy

One can divide the energy into the rest energy

E0 = mc2 (4.10)

and the kinetic energy Ekin = E − E0 = E − mc2 . The quantities defined in (4.6) are conserved when
more particles are involved. Due to the equivalence between energy and mass, the mass or the mass
density becomes a source of the gravitational field.

5 Tensors in Minkowski space

Let us discuss the transformation properties of physical quantities under a Lorentz transformation.
We have already seen how a 4-vector is transformed:

V α → V 0α = Λα β V β . (5.1)

This is a so-called contravariant 4-vector (indices are up). The coordinate system transforms according
to X α → X 0α = Λα β X β . A covariant 4-vector is defined through

Vα = ηαβ V β . (5.2)

Let us now define the matrix η αβ as the inverse matrix to ηαβ :

η αβ ηβγ = δγα . (5.3)

In our case  
1 0 0 0
 
0 −1 0 0
η αβ = ηαβ =  . (5.4)
0 0
 −1 0

0 0 0 −1
With (5.3) we can express (5.2) equally as

V α = η αβ Vβ . (5.5)

14
 5 TENSORS IN MINKOWSKI SPACE

The transformation of a covariant vector is then given by

Vα0 = ηαβ V 0β = ηαβ Λβ γ V γ = ηαβ Λβ γ η γδ Vδ = Λ̄δ α Vδ , (5.6)

with
Λ̄δ α = ηαβ Λβ γ η γδ (5.7)

(one can use Λα β instead of Λ̄βα but one should be very careful in writings since Λα β 6= Λβ α ). Thanks
to (3.10), we find
Λ̄γ α Λα β = ηαδ η γ Λδ  Λα β = η γ ηβ = δ γ β (5.8)

And similarly, we get Λβ α Λ̄αγ = δ β γ . In matrix notation, we have ΛΛ̄ = Λ̄Λ = I and thus Λ̄ = Λ−1 .
To summarize the transformations of 4-vectors:

• A contravariant vector transforms with Λ

• A covariant vector transforms with Λ−1 = Λ̄

The scalar product of two vectors V α and U β is defined by

V α Uα = Vα U α = η αβ Vα Uβ = ηαβ V α U β (5.9)

and is invariant under Lorentz transformations: V 0α Uα0 = Λα β Λ̄δ α V β Uδ = V β Uβ .

| {z }
δ
δβ
∂ ∂ ∂xβ ∂ ∂xβ ∂
The operator ∂xα transforms like a covariant vector: ∂x0α = ∂x0α ∂xβ . Since ∂x0α = Λ̄βα ⇒ ∂x0α =
∂ ∂
β
Λ̄ α ∂xβ . We will now use the notations ∂α ≡ ∂xα (covariant vector) and ∂ ≡ ∂x∂α
α
(contravariant
∂2
vector). The D’Alembert operator can be written as  = ∂ α ∂α = η αβ ∂α ∂β = c12 ∂t2 − ∆ and is a
Lorentz scalar.

A quantity is a rank r contravariant tensor if its components transform like the coordinates xα :

T 0α1 ...αr = Λα1 β1 . . . Λαr βr T β1 ...βr (5.10)

Tensors of rank 0 are scalars, tensors of rank 1 are vectors. For “mixed” tensors we have for example:

T 0α βγ = Λα δ Λ̄ β Λ̄µγ T δ µ

The following operations can be used to form new tensors:

1. Linear combinations of tensors with the same upper and lower indices: T α β = aRα β + bS α β

2. Direct products of tensors: T αβ V γ (works with mixed indices)

3. Contractions of tensors: T αβ β or T αβ Vβ (lowers a tensor by 2 in rank)

4. Differentiation of a tensor field: ∂α T βα (the derivative ∂α of any tensor is a tensor with one
additional lower index α: ∂α T βγ ≡ Rα βγ )

5. Going from a covariant to a contravariant component of a tensor is defined like in (5.2) and (5.5)
(lowering and raising indices with η αβ , ηαβ ).

15
 5 TENSORS IN MINKOWSKI SPACE

One must be aware that

• the order of the upper and lower indices is important,

• Λα β is not a tensor.

η can be considered as a tensor: η = η αβ = ηαβ is the Minkowski tensor.

0 (3.10) (5.8)
ηαβ = Λ̄γ α Λ̄δ β ηγδ = Λ̄γ α Λ̄δ β Λµ γ Λν δ ηµν = ηαβ

η appears in the line element (ds2 = ηαβ dxα dxβ ) and is thus the metric tensor in Minkowski space.
We also have η α β = η αγ ηγβ = δ α β = ηβ α , and thus the Kronecker symbol is also a tensor.

We define the totally antisymmetric tensor or (Levi-Civita tensor) as


+1 (α, β, γ, δ) is an even permutation of (0, 1, 2, 3)



αβγδ
 = −1 (α, β, γ, δ) is an odd permutation of (0, 1, 2, 3) (5.11)



0 otherwise

Without proof we have: (det (Λ) = 1)

0αβγδ = αβγδ ,
0 0
γ 0 δ0
αβγδ = ηαα0 ηββ 0 ηγγ 0 ηδδ0 α β = −αβγδ .

The functions S(x), V α (x), T αβ . . . with x = (x0 , x1 , x2 , x3 ) are a scalar field, a vector field, or a tensor
field . . . respectively if:

S 0 (x0 ) = S(x)

V 0α (x0 ) = Λα β
β V (x)

T 0αβ (x0 ) = Λα β δγ
δ Λγ T (x)

..
.

Also the argument has to be transformed, thus x0 has to be understood as x0α = Λα β xβ .

16
 6 ELECTRODYNAMICS

6 Electrodynamics
~ r, t), B(~
Maxwell’s equations relate the fields E(~ ~ r, t), the charge density ρe (~r, t) and the current density
~(~r, t):
 

 ~ = 4πρc
div E 
 ~ =0
div B
 
inhomogeneous homogeneous (6.1)
~ ~
~ = 4π ~ + 1 ∂ E ~ = − 1 ∂B
 
 rot B
  rot E

c c ∂t c ∂t
The continuity equation
div ~ + ρ̇c = 0 → ∂α j α = 0 (6.2)
α
Z j = (cρe , ~) follows from the conservation of charge, which for an isolated system implies
with
∂t j 0 d3 r = 0. ∂α j α is a Lorentz scalar. We can define the field strength tensor which is given
by the antisymmetric matrix
 
0 −Ex −Ey −Ez
 
Ex 0 −Bz By 
F αβ =  . (6.3)
E
 y Bz 0 −Bx 

Ez −By Bx 0

Using this tensor we can rewrite the inhomogeneous Maxwell equations

4π
∂α F αβ = jβ , (6.4)
| {z } c |{z}
4−vector 4−vector

and also the homogeneous ones:

αβγδ ∂β Fγδ = 0. (6.5)

Both equations are covariant under a Lorentz transformation. Eq. (6.5) allows to represent F αβ as a
“curl” of a 4-vector Aα :
F αβ = ∂ α Aβ − ∂ β Aα . (6.6)

We can then reformulate Maxwell’s equations for Aα = (φ, Ai ). From (6.6) it follows that the gauge
transformation
Aα → Aα + ∂ α Θ (6.7)

of the 4-vector Aα leaves F αβ unchanged, where Θ(x) is an arbitrary scalar field. The Lorenz gauge
∂α Aα = 0 leads to the decoupling of the inhomogeneous Maxwell’s equation (6.4) to
4π α
Aα = j . (6.8)
c
The generalized equation of motion for a particle with charge q is
duα q
m = F αβ uβ (6.9)
dτ c

17
 7 ACCELERATED REFERENCE SYSTEMS IN SPECIAL RELATIVITY

 
p
d~ ~ + ~v ∧ B
~ with p~ = γm~v .
The spatial components give the expression of the Lorentz force =q E
dt c
The energy-momentum tensor for the electromagnetic field is
 
αβ 1 1
Tem = F α γ F γβ + η αβ Fγδ F γδ (6.10)
4π 4
 
The 00-component represents the energy density of the field Tem 00
= uem = 8π 1 ~2 + B
E ~ 2 and the
 i
0i-components the Poynting vector S ~ i = cTem0i c
= 4π ~ ∧B
E ~ . In terms of these tensors, Maxwell’s
αβ 1
equations are ∂α Tem = − F βγ jγ . Tem αβ αβ
is symmetric and conserved: ∂α Tem = 0. Setting β = 0
c
αk
leads to energy conservation whereas ∂α Tem = 0 leads to conservation of the k th component of the
αβ
momentum. One should note that ∂α Tem = 0 is valid only if there is no external force, otherwise we
αβ
can write ∂α Tem = f β , where f β is the external force density. Such an external force can often be
included in the energy-momentum tensor.

7 Accelerated reference systems in special relativity

Non inertial systems can be considered in the context of special relativity. However, then the physical
laws no longer have their simple covariant form. In e.g. a rotating coordinate system, additional terms
will appear in the equations of motion (centrifugal terms, Coriolis force, etc.).
Let us look at a coordinate system KS’ (with coordinates x0µ ) which rotates with constant angular
speed with respect to an inertial system IS (xα ):



 x = x0 cos(ωt0 ) − y 0 sin(ωt0 ),



 y = x0 sin(ωt0 ) + y 0 cos(ωt0 ),



(7.1)
z = z0,








 t = t0 ,


and assume that ω 2 (x02 + y 02 )  c2 . Then we insert (7.1) into the line element ds (in the known IS
form):

ds2 = ηαβ dxα dxβ = c2 dt2 − dx2 − dy 2 − dz 2

= c2 − ω 2 (x02 + y 02 ) dt02 + 2ωy 0 dx0 dt0 − 2ωx0 dy 0 dt0 − dx02 − dy 02 − dz 02

 

= gµν dx0µ dx0ν . (7.2)

The resulting line element is more complicated. For arbitrary coordinates x0µ , ds2 is a quadratic form
of the coordinate differentials dx0µ . Consider a general coordinate transformation from xµ (in IS) to
x0µ (in KS’):
xα ≡ xα (x0 ) = xα (x00 , x01 , x02 , x03 ), (7.3)

18
 7 ACCELERATED REFERENCE SYSTEMS IN SPECIAL RELATIVITY

then we get for the line element

∂xα ∂xβ 0µ 0ν
ds2 = ηαβ dxα dxβ = ηαβ dx dx = gµν (x0 )dx0µ dx0ν , (7.4)
∂x0µ ∂x0ν
with
∂xα ∂xβ
gµν (x0 ) = ηαβ . (7.5)
∂x0µ ∂x0ν
The quantity gµν is the metric tensor of the KS’ system. It is symmetric (gµν = gνµ ) and depends on
x0 . It is called metric because it defines distances between points in coordinate systems.
In an accelerated reference system we get inertial forces. In the rotating frame we expect to
~ which can be written in terms of a centrifugal potential φ:
experience the centrifugal force Z,
ω 2 02 ~ = −m∇φ.
~
φ=− (x + y 02 ) and Z (7.6)
2
This enables us to see that g00 from (7.2) is
2φ
g00 = 1 + . (7.7)
c2
The centrifugal potential appears in the metric tensor. We will see later that the first derivatives of
the metric tensor are related to the forces in the relativistic equations of motion. To get the meaning
of t0 in KS’ we evaluate (7.2) at a point with dx0 = dy 0 = dz 0 = 0:
r r
dsclock √ 0 2φ 0 v2
dτ = = g00 dt = 1 + 2 dt = 1− 2 dt (7.8)
c c | {z c }
correspond to clocks
time computed in
an inertial system

τ represents the time of a clock at rest in KS’.

p
In an inertial system we have gµν = ηµν and the clock moves with speed v = ωρ (ρ = x02 + y 02 ).
With (7.6) we see that both expressions in (7.8) are the same.
The coefficients of the metric tensor gµν (x0 ) are functions of the coordinates. Such a dependence
will also arise when one uses curved coordinates. Consider for example cylindrical coordinates:

x00 = ct = x0 , x01 = ρ, x02 = θ, x03 = z.

This results in the line element

ds2 = c2 dt2 − dx2 − dy 2 − dz 2 = c2 dt2 − dρ2 − ρ2 dθ2 − dz 2 = gµν (x0 )dx0µ dx0ν . (7.9)

Here gµν is diagonal:

 
1
 
 −1 
gµν = . (7.10)

 −ρ2 

−1
The fact that the metric tensor depends on the coordinates can be either due to the fact that the
considered coordinate system is accelerated or that we are using non-cartesian coordinates.

19
 8 THE EQUIVALENCE PRINCIPLE

Part III

Towards General Relativity

8 The equivalence principle
The principle of equivalence of gravitation and inertia tells us how an arbitrary physical system re-
sponds to an external gravitational field (with the help of tensor analysis). The physical basis of
general relativity is the equivalence principle as formulated by Einstein:

1. Inertial and gravitational mass are equal

2. Gravitational forces are equivalent to inertial forces

3. In a local inertial frame, we experience the known laws of special relativity without gravitation

8.1 About the masses

The inertial mass mt is the quantity appearing in Newton’s law F~ = mt ~a which acts against accelera-
tion by external forces. In contrast, the gravitational mass ms is the proportionality constant relating
the gravitational force between mass points to each other. For a particle moving in a homogeneous
gravitational field, we have the equation mt z̈ = −ms g, whose solution is
1 ms 2
z(t) = − gt (+v0 t + z0 ). (8.1)
2 mt
ms
Galilei stated that “all bodies fall at the same rate in a gravitational field”, i.e. mt is the same for
all bodies. Another experiment is to consider the period T of a pendulum (in the small amplitude
T 2
=m s l


approximation): 2π mt g , where l is the length of the pendulum. Newton verified that this period
is independent on the material of the pendulum to a precision of about 10−3 . Eötvös (∼1890), using
torsion balance, got a precision of about 5 × 10−9 . Today’s precision is about 10−11 ∼ 10−12 , this is
way we can make the assumption ms = mt on safe grounds.
Due to the equivalence between energy and mass (E = mc2 ), all forms of energy contribute to
mass, and due to the first point of the equivalence principle, to the inertial and to the gravitational
masses.

8.2 About the forces

As long as gravitational and inertial masses are equal, then gravitational forces are equivalent to inertial
forces: going to a well-chosen accelerated reference frame, one can get rid of the gravitational field. As
an example take the equation of motion in the homogeneous gravitational field at Earth’s surface:

d2~r
mt = ms~g (8.2)
dt2

20
 8 THE EQUIVALENCE PRINCIPLE

This expression is valid for a reference system which is at rest on Earth’s surface (∼ to a good
approximation an IS). Then we perform the following transformation to an accelerated KS system:
1
~r = ~r 0 + gt02 , t = t0 , (8.3)
2
and we assume gt  c. The origin of KS ~r 0 = 0 moves in IS with ~r(t) = 21 gt2 . Then, inserting (8.3)
into (8.2) results in
d2
 
0 1 02
mt 02 ~r + gt = ms~g
dt 2

d2~r 0
⇒ mt = (ms − mt ) ~g . (8.4)
dt02
r0
d2 ~
If ms = mt , the resulting equation in KS is the one of a free moving particle dt02 = 0; the gravitational
force vanishes. As another example in a “free falling elevator” the “observer” does not feel any gravity.
Einstein generalized this finding postulating that (this is the Einstein equivalence principle) “in a
free falling accelerated reference system all physical processes run as if there is no gravitational field”.
Notice that the “mechanical” finding is now expanded to all types of physical processes (at all times
and places). Moreover also non-homogeneous gravitational fields are allowed. The equality of inertial
and gravitational mass is also called the weak equivalence principle (or universality of the free fall).
As an example of a freely falling system, consider a satellite in orbit around Earth (assuming that
the laboratory on the satellite is not rotating). Thus the equivalence principle states that in such a
system all physical processes run as if there would be no gravitational field. The processes run as in an
inertial system: the local IS. However, the local IS is not an inertial system, indeed the laboratory on
the satellite is accelerated compared to the reference system of the fixed distant stars. The equivalence
principle implies that in a local IS the rules of special relativity apply.

• For an observer on the satellite laboratory all physical processes follow special relativity and
there are neither gravitational nor inertial forces.

• For an observer on Earth, the laboratory moves in a gravitational field and moreover inertial
forces are present, since it is accelerated.

The motion of the satellite laboratory, i.e. its free falling trajectory, is such that the gravitational
forces and inertial forces just compensate each other (cf (8.4)). The compensation of the forces is
exactly valid only for the center of mass of the satellite laboratory. Thus the equivalence principle
applies only to a very small or local satellite laboratory (”how small” depends on the situation).

The equivalence principle can also be formulated as follows:

“At every space-time point in an arbitrary gravitational field, it is possible to choose a

locally inertial coordinate system such that, within a sufficiently small region around the
point in question, the laws of nature take the same form as in non-accelerated Cartesian
coordinate systems in the absence of gravitation.”2
2 Notice the analogy with the axiom Gauss took as a basis of non-Euclidean geometry: he assumed that at any point

on a curved surface we may erect a locally Cartesian coordinate system in which distances obey the law of Pythagoras.

21
 8 THE EQUIVALENCE PRINCIPLE

The equivalence principle allows us to set up the relativistic laws including gravitation; indeed one can
just perform a coordinate transformation to another KS:
special relativity laws
o nrelativistic laws
coordinate
without −−−−−−−−−−→ with
gravitation transf ormation gravitation

The coordinate transformation includes the relative acceleration between the laboratory system and
KS which corresponds to the gravitational field. Thus from the equivalence principle we can derive the
relativistic laws in a gravitational field. However, it does not fix the field equations for gµν (x) since
those equations have no analogue in special relativity.
From a geometrical point of view the coordinate dependence of the metric tensor gµν (x) means
that space is curved. In this sense the field equations describe the connection between curvature of
space and the sources of the gravitational field in a quantitative way.

8.3 Riemann space

We denote with ξ α the Minkowski coordinates in the local IS (e.g. the satellite laboratory). From the
equivalence principle, the special relativity laws apply. In particular, we have for the line element

ds2 = ηαβ dξ α dξ β . (8.5)

Going from the local IS to a KS with coordinates xµ is given by a coordinate transformation ξ α =

ξ α (x0 , x1 , x2 , x3 ). Inserting this into (8.5) results in

∂ξ α ∂ξ β µ ν
ds2 = ηαβ dx dx = gµν (x)dxµ dxν , (8.6)
∂xµ ∂xν
∂ξ α ∂ξ β
and thus gµν (x) = ηαβ . A space with such a path element of the form (8.6) is called a Riemann
∂xµ ∂xν
space.
The coordinate transformation (expressed via gµν ) also describes the relative acceleration between
KS and the local IS. Since at two different points of the local IS the accelerations are (in general)
different, there is no global transformation in the form (8.6) that can be brought to the Minkowski
form (8.5). We shall see that gµν are the relativistic gravitational potentials, whereas their derivatives
determine the gravitational forces.

22
 8 THE EQUIVALENCE PRINCIPLE

Figure 1: An experimenter and his two stones freely floating somewhere in outer space, i.e. in the
absence of forces.

Figure 2: Constant acceleration upwards mimics the effect of a gravitational field: experimenter and
stones drop to the bottom of the box.

23
 8 THE EQUIVALENCE PRINCIPLE

Figure 5: The experimenter and his stones in a

Figure 3: The effect of a constant gravitational non-uniform gravitational field: the stones will ap-
field: indistinguishable for our experimenter from proach each other slightly as they fall to the bot-
that of a constant acceleration as in figure 2. tom of the elevator.

Figure 6: The experimenter and stones freely

Figure 4: Free fall in a gravitational field has the falling in a non-uniform gravitational field: the ex-
same effect as no gravitational field (figure 1): ex- perimenter floats, so do the stones, but they move
perimenter and stones float. closer together, indicating the presence of some
external forces.

24
 9 PHYSICS IN A GRAVITATIONAL FIELD

9 Physics in a gravitational field

9.1 Equations of motion
According to the equivalence principle, in a local IS the laws of special relativity hold. For a mass
point on which no forces act we have
d2 ξ α
= 0, (9.1)
dτ 2
where the proper time τ is defined through ds2 = ηαβ dξ α dξ β = c2 dτ 2 . We can also define the 4-velocity
dξ α
as uα = dτ . Solutions of (9.1) are straight lines

ξ α = aα τ + bα . (9.2)

Light (or a photon) moves in the local IS on straight lines. However, for photons τ cannot be identified
with the proper time since on the light cone ds = cdτ = 0. Thus we denote by λ a parameter of the
trajectory of photons:
d2 ξ α
= 0. (9.3)
dλ2
Let us now consider a global coordinate system KS with xµ and metric gµν (x). At all points xµ , one can
locally bring ds2 into the form ds2 = ηαβ dξ α dξ β . Thus at all points P there exists a transformation
ξ α (x) = ξ α (x0 , x1 , x2 , x3 ) between ξ α and xµ . The transformation varies from point to point. Consider
a small region around point P . Inserting the coordinate transformation into the line element, we get
∂ξ α ∂ξ β
ds2 = ηαβ dξ α dξ β = ηαβ µ ν dxµ dxν . (9.4)
| ∂x{z ∂x }
≡gµν (x) metric tensor

We write (9.1) in the form

∂ξ α dxµ ∂ξ α d2 xµ ∂ 2 ξ α dxµ dxν
 
d
0= = + ,
dτ ∂xµ dτ ∂xµ dτ 2 ∂xµ ∂xν dτ dτ
∂xκ ∂ξ α ∂xκ d2 xµ
multiply it by ∂ξ α and make use of ∂xµ ∂ξ α = δ κ µ . This way we can solve for dτ 2 and get the following
equation of motion in a gravitational field
d2 xκ dxµ dxν
2
= −Γκµν , (9.5)
dτ dτ dτ
with
∂xκ ∂ 2 ξ α
Γκµν = . (9.6)
∂ξ α ∂xµ ∂xν
The Γκµν are called the Christoffel symbols and represent a pseudo force or fictive gravitational field
(like centrifugal or Coriolis forces) that arises whenever one describes inertial motion in non-inertial
coordinates. Eq. (9.5) is a second order differential equation for the functions xµ (τ ) which describe
duα α
the trajectory of a particle in KS with gµν (x). Eq. (9.5) can also be written as m = f α , uα = dx
dτ .
dτ
Comparing with (4.3) one infers that the right hand side of (9.5) describes the gravitational forces.
dxµ
Due to (9.4), the velocity dτ has to satisfy the condition
dxµ dxν
c2 = gµν (for m 6= 0) (9.7)
dτ dτ

25
 9 PHYSICS IN A GRAVITATIONAL FIELD

dxµ
(assume dτ 6= 0 and m 6= 0). Due to (9.7) only 3 of the 4 components of dτ are independent (this
is also the case for the 4-velocity in special relativity). For photons (m = 0) one finds instead, using
(9.3), a completely analogous equation for the trajectory:
d2 xκ µ
κ dx dx
ν
= −Γµν , (9.8)
dλ2 dλ dλ
and since dτ = ds = 0, one has instead of (9.7):
dxµ dxν
0 = gµν (for m = 0).
dλ dλ

9.2 Christoffel symbols

The Christoffel symbols can be expressed in terms of the first derivatives of gµν . Consider with (9.4):
 
∂gµν ∂gλν ∂gµλ  ∂ 2 ξ α ∂ξ β ∂ξ α ∂2
ξβ 

λ
+ µ
− ν
=ηαβ  µ λ ν
+ µ
 ν

λ
∂x ∂x ∂x  ∂x ∂x ∂x |∂x ∂x

{z ∂x }
1
 
 ∂ ξ 2 α β α β
2
∂ξ ∂ξ ∂ ξ 
+ ηαβ 
 ∂xλ ∂xµ ∂xν + ∂x 
λ ∂xν ∂xµ 
| {z }
2
 
 ∂ 2 ξ α ∂ξ
β

∂ξ α ∂2
ξβ 

− ηαβ  µ
 ∂x  ν λ
+ µ
 λ ν
.
| ∂x {z ∂x }  |∂x ∂x{z ∂x }


2 1

Using ηαβ = ηβα this becomes

∂ 2 ξ α ∂ξ β
= 2ηαβ . (9.9)
∂xµ ∂xλ ∂xν
On the other hand
gνσ Γσ
µλ
z }| {z }| {
∂ξ α ∂ξ β ∂xσ ∂ 2 ξ ρ
gνσ Γσµλ = ηαβ ν σ
∂x ∂x ∂ξ ρ ∂xµ ∂xλ
| {z }
δρβ

∂ξ α ∂ 2 ξ β
= ηαβ
∂xν ∂xµ ∂xλ
 
1 ∂gµν ∂gλν ∂gµλ
= + − . (9.10)
2 ∂xλ ∂xµ ∂xν
We introduce the inverse matrix g µν such that g µν gνσ = δ µ σ . Therefore we can solve with respect to
the Christoffel symbols:  
1 κν ∂gµν ∂gλν ∂gµλ
Γκµλ = g + − . (9.11)
2 ∂xλ ∂xµ ∂xν

26
 9 PHYSICS IN A GRAVITATIONAL FIELD

Note that the Γ’s are symmetric in the lower indices Γκµν = Γκνµ . The gravitational forces on the right
hand side of (9.6) are given by derivatives of gµν . Comparing with the equation of motion of a particle
in a electromagnetic field shows that the Γλµν correspond to the field F αβ , whereas the gµν correspond
to the potentials Aα .

9.3 Newtonian limit

Let us assume that v i  c and the fields are weak and static (i.e. not time dependent). Thus
dxi dx0
dτ  dτ . Inserting this into (9.5) leads to

small
velocity 2
2 κ µ ν
dx0

d x dx dx
= −Γκµν ≈ −Γκ00
z}|{
. (9.12)
dτ 2 dτ dτ dτ

For static fields we get from (9.11):

staticity
g κi ∂g00
Γκ00 = −
z}|{
(i = 1, 2, 3) (9.13)
2 ∂xi
(the other terms contain partial derivative with respect to x0 which are zero by staticity). We write
the metric tensor as gµν = ηµν + hµν . For weak fields we have |hµν | = |gµν − ηµν |  1. In this case the
coordinates (ct, xi ) are “almost” Minkowski coordinates. Inserting the expansion for gµν into (9.13)
(taking only linear terms in h) gives
 
1 ∂h00 k
Γκ00 = 0, δ . (9.14)
2 ∂xi i

Then, let us compute (9.12) for κ = 0, κ = j:

choice
d2 t dt
=0⇒
z}|{
2
= constant = 1, (9.15a)
dτ dτ
 2
d2 xj c2 ∂h00 dt
= − . (9.15b)
dτ 2 2 ∂xj dτ
| {z }
12

Taking (xj ) = ~r, we can write

d2~r c2
= − ∇h00 (~r), (9.16)
dt2 2
d2~r
which is to be compared with the Newtonian case = −∇φ(~r). Therefore:
dt2
2φ(~r)
g00 (~r) = 1 + h00 (~r) = 1 + . (9.17)
c2
Notice that the Newtonian limit (9.16) gives no clue on the other components of hµν . The quantity
2φ
c2 is a measure of the strength of the gravitational field. Consider a spherically symmetric mass

27
 10 TIME DILATION

distribution. Then
2φ(R)
≈ 1.4 × 10−9 at Earth surface,
c2

≈ 4 × 10−6 on the Sun (and similar stars),

≈ 3 × 10−4 on a white dwarf,

≈ 3 × 10−1 on a neutron star → GR required.

10 Time dilation
We study a clock in a static gravitational field and the phenomenon of gravitational redshift.

10.1 Proper time

The proper time τ of the clock is defined through the 4-dimensional line element as
q 
dsclock 1
dτ = = gµν (x)dxµ dxν , (10.1)
c c clock

x = (xµ ) are the coordinates of the clock. The time shown by the clock depends on both the gravita-
tional field and of its motion (the gravitational field being described by gµν ).

Special cases:

1. Moving clock in an IS without gravity :

r
v2
dτ = 1− dt
c2
(gµν = ηµν , dxi = v i dt, dx0 = cdt).

2. Clock at rest in a gravitational field (dxi = 0)

√
dτ = g00 dt.

For a weak static field, one has with (9.17):

r
2φ(r)
dτ = 1 + dt (|φ|  c2 ). (10.2)
c2
The fact that φ is negative implies that a clock in a gravitational field goes more slowly than a
clock outside the gravitational field.

10.2 Redshift
Let us now consider objects which emit or absorb light with a given frequency. Consider only a static
gravitational field (gµν does not depend on time). A source in ~rA (at rest) emits a monochromatic

28
 10 TIME DILATION

electromagnetic wave at a frequency νA . An observer at ~rB , also at rest, measures a frequency νB .

p
At source: dτA = g00 (~rA )dtA
p (10.3)
At observer: dτB = g00 (~rB )dtB

As a time interval we consider the time between two following peaks departing from A or arriving
at B. In this case dτA and dτB correspond to the period of the electromagnetic waves at A and B,
respectively, and therefore
1 1
dτA = , dτB = . (10.4)
νA νB
Going from A to B needs the same time ∆t for the first and the second peak of the electromagnetic
wave. Consequently, they will arrive with a time delay which is equal to the one with which they were
emitted, thus dtA = dtB . With (10.3) and (10.4) we get:
s
νA g00 (~rB ) νA λB
= , with z = −1= − 1. (10.5)
νB g00 (~rA ) νB λA

The quantity z is the gravitational redshift:

s
g00 (~rB )
z= − 1. (10.6)
g00 (~rA )

2φ
For weak fields with g00 = 1 + c2 we have

φ(~rB ) − φ(~rA )
z= (|φ|  c2 ), (10.7)
c2
such a redshift is observed by measuring spectral lines from stars. As an example take solar light with
(10.7)
φ(rB ) − φ(rA ) φ(rA ) GM
z= 2
≈− 2 = 2 ≈ 2 × 10−6 ,
c c c R
with M ≈ 2 × 1030 kg and R ≈ 7 × 108 m. For a white dwarf we find z ≈ 10−4 and for a neutron star
z ≈ 10−1 . In general there are 3 effects which can lead to a modification in the frequency of spectral
lines:

1. Doppler shift due to the motion of the source (or of the observer)

2. Gravitational redshift due to the gravitational field at the source (or at the observer)

3. Cosmological redshift due to the expansion of the Universe (metric tensor is time dependent)

10.3 Photon in a gravitational field

Consider a photon with energy Eγ = ~ω = 2π~ν, travelling upwards in the homogeneous gravity field
of the Earth, covering a distance of h = hB − hA (h small). The corresponding redshift is

νA φ(rB ) − φ(rA ) g(hB − hA ) gh

z= −1= = = 2, (10.8)
νB c2 c2 c

29
 11 GEOMETRICAL CONSIDERATIONS

resulting in a frequency change ∆ν = νB − νA (νA > νB , νB = ν) and thus

∆ν gh
=− 2. (10.9)
ν c
E E
The photon changes its energy by ∆Eγ = − c2γ gh (like a particle with mass c2γ = m). This effect has
∆νexp
been measured in 1965 (through the Mössbauer effect) as = 1.00 ± 0.01 (1% accuracy)3 .
∆νth

11 Geometrical considerations
In general, the coordinate dependence of gµν (x) means that spacetime, defined through the line element
ds2 , is curved. The trajectories in a gravitational field are the geodesic lines in the corresponding
spacetime.

11.1 Curvature of space

The line element in an N -dimensional Riemann space with coordinates x = (x1 , . . . , xN ) is given as

ds2 = gµν dxµ dxν (µ, ν = 1, . . . , N ).

Let us just consider a two dimensional space x = (x1 , x2 ) with

ds2 = g11 dx1 dx1 + 2g12 dx1 dx2 + g22 dx2 dx2 . (11.1)

Examples:

• Plane with Cartesian coordinates (x1 , x2 ) = (x, y):

ds2 = dx2 + dy 2 , (11.2)

or with polar coordinates (x1 , x2 ) = (ρ, φ):

ds2 = dρ2 + ρ2 dφ2 (11.3)

• Surface of a sphere with angular coordinates (x1 , x2 ) = (θ, φ):

ds2 = a2 dθ2 + sin2 θdφ2

(11.4)

The line element (11.2) can, via a coordinate transformation, be brought into the form (11.3). However,
there is no coordinate transformation which brings (11.4) into (11.2). Thus:

• The metric tensor determines the properties of the space, among which is also the curvature.

• The form of the metric tensor is not uniquely determined by the space, in other words it depends
on the choice of coordinates.
3 Pound, R. V. and Snider, J. L., Effect of Gravity on Gamma Radiation, Physical Review, 140

30
 11 GEOMETRICAL CONSIDERATIONS

The curvature of the space is determined via the metric tensor (and it does not depend on the coordinate
choice)4 . If gik = const then the space is not curved. In an Euclidian space, one can introduce Cartesian
coordinates gik = δik . For a curved space gik 6= const (does not always imply that space is curved).
For instance by measuring the angles of a triangle and checking if their sum amounts to 180 degrees
or differs, one can infer if the space is curved or not (for instance by being on the surface of a sphere).

4 Beside the curvature discussed here, there is also an exterior curvature. We only consider intrinsic curvatures here.

31
 12 DIFFERENTIABLE MANIFOLDS

Part IV

Differential Geometry
12 Differentiable manifolds
A manifold is a topological space that locally looks like the Euclidean Rn space with its usual topology.
A simple example of a curved space is the S 2 sphere: one can setup local coordinates (θ, ϕ) which map
S 2 onto a plane R2 (a chart). Collections of charts are called atlases. There is no one-to-one map of
S 2 onto R2 ; we need several charts to cover S 2 .

Definition: Given a (topological) space M, a chart on M is a one-to-one map φ from an open subset
U ⊂ M to an open subset φ(U) ⊂ Rn , i.e. a map φ : M → Rn . A chart is often called a coordinate
S
system. A set of charts with domain Uα is called an atlas of M, if Uα = M, {φα |α ∈ I}.
α

Definition: dim M = n

Definition: Two charts φ1 , φ2 are C ∞ -related if both the maps φ2 ◦ φ−1

1 : φ1 (U1 ∩ U2 ) → φ2 (U1 ∩ U2 )
and its inverse are C ∞ . φ2 ◦ φ−1
1 is the so-called transition function between the two coordinate charts.
A collection of C ∞ related charts such that every point of M lies in the domain of at least one chart
forms an atlas (C ∞ : derivatives of all orders exist and are continuous).

The collection of all such C ∞ -related charts forms a maximal atlas. If M is a space and A its maximal
atlas, the set (M, A) is a (C ∞ )-differentiable manifold. (If for each φ in the atlas the map φ : U → Rn
has the same n, then the manifold has dimension n.)

Important notions:

• A differentiable function f : M → R belongs to the algebra F = C ∞ (M), sum and product of

such functions are again in F = C ∞ (M).

• Fp is the algebra of C ∞ -functions defined in any neighbourhood of p ∈ M (f = g means f (q) =

g(q) in some neighbourhood of p).

• A differentiable curve is a differentiable map γ : R → M.

• Differentiable maps F : M → M0 are differentiable if φ2 ◦ F ◦ φ−1

1 is a differentiable map for all
suitable charts φ1 of M and φ2 of M0 .

The notions have to be understood by means of a chart, e.g. f : M → R is differentiable if x 7→

f (p(x)) ≡ f (x) is differentiable. This is independent of the chart representing a neighbourhood of p.
|{z}
∈M

32
 12 DIFFERENTIABLE MANIFOLDS

M
U1

U2

φ1 φ2

Rn Rn
x x̄

φ2 ◦ φ−1
1
X X̄

(Chart φ1 ) ⊂ Rn (Chart φ2 ) ⊂ Rn

Figure 7: Manifold, charts and transition function.

12.1 Tangent vectors and tangent spaces

At every point p of a differentiable manifold M one can introduce a linear space, called tangent space
Tp (M). A tensor field is a (smooth) map which assigns to each point p ∈ M a tensor of a given type
on Tp (M).

Definition: a C ∞ -curve in a manifold M is a map γ of the open interval I = (a, b) ⊂ R → M such

that for any chart φ, φ ◦ γ : I → Rn is a C ∞ map.

33
 12 DIFFERENTIABLE MANIFOLDS

Let f : M → R be a smooth function on M. Consider the map f ◦ γ : I → R, t 7→ f (γ(t)). This has

a well-defined derivative: the rate of change of f along the curve. Consider f ◦ φ−1 ◦ φ◦γ and
| {z } | {z }
n n
R →R I→R
xi →f (xi ) t→xi (γ(t))
=f (φ−1 (xi ))
use the chain rule:
n
dxi (γ(t))
 
d X ∂
(f ◦ γ) = f (xi ) . (12.1)
dt i=1
∂xi dt
 
d 
Thus, given a curve γ(t) and a function f , we can obtain a qualitatively new object (f ◦ γ)  ,
dt t=t0
the rate of change of f along the curve γ(t) at t = t0 .

Definition: The tangent vector γ̇p to a curve γ(t) at a point p is a map from the set of real functions
f defined in a neighbourhood of p to R defined by
 
d
γ̇p : f 7→ (f ◦ γ) = (f ◦ γ)• p = γ̇p (f ). (12.2)
dt p

Given a chart φ with coordinates xi , the components of γ̇p with respect to the chart are
 
i • d i
(x ◦ γ) p = x (γ(t)) . (12.3)
dt p

The set of tangent vectors at p is the tangent space Tp (M) at p.

Theorem: If the dimension of M is n, then Tp (M) is a vector space of dimension n (without proof).

We set γ(0) = p (t = 0), Xp = γ̇p , and Xp f = γ̇p (f ). Eq. (12.3) determines Xp (xi ), the components
of Xp with respect to a given basis:

Xp f = [f ◦ γ]• (0)
•
= f ◦ φ−1 ◦ φ ◦ γ (0)


n
X ∂ d (12.4)
= i
(f ◦ φ−1 ) (xi ◦ γ)(0)
i=1
∂x dt

X ∂ 
f (x , . . . , x ) Xp (xi ) .
1 n


= i
i
∂x

This way we see that  

X
i ∂
Xp = Xp (x ) , (12.5)
i
∂xi p
 
∂
and so the span Tp (M). From (12.5) we see that Xp (xi ) are the components of Xp with
∂xi p
respect to the given basis (Xp (xi ) = Xpi or X i ).

34
 12 DIFFERENTIABLE MANIFOLDS

Suppose that f, g are real functions on M and f g : M → R is defined as f g(p) = f (p)g(p). If

Xp ∈ Tp (M), then (Leibniz rule)

Xp (f g) = (Xp f )g(p) + f (p)(Xp g). (12.6)

Notation: (Xf )(p) = Xp f , p ∈ M.

Basis of Tp (M): Tp = TP (M) has dimension n. In any basis (e1 , . . . , en ) we have X = X i ei . Changes
of basis are given by
ēi = φi k ek , X̄ i = φi k X k . (12.7)

The transformations φi k and φi k are inverse transposed of each other. In particular, ei = ∂

∂xi is called
coordinate basis (with respect to a chart). Upon change of chart x 7→ x̄,

∂xk ∂ x̄i
φi k = , φi k = . (12.8)
∂ x̄i ∂xk

Definition: The cotangent space Tp∗ (or dual space Tp∗ of Tp ) consists of covectors ω ∈ Tp∗ , which are
linear one-forms ω : X 7→ ω(X) ≡< ω, X >∈ R (ω : Tp → R).
 
∂f
In particular for functions f , df : X 7→ Xf is an element of Tp∗ . The elements df = f,i dxi = ∂xi dxi
form a linear space of dimension n, therefore all of Tp∗ .
We can define a dual basis (e1 , . . . , en ) of Tp∗ : ω = ωi ei . In particular the dual basis of a basis
(e1 , . . . , en ) of Tp is given by < ei , X >= X i or < ei , X j ej >= X j < ei , ej > = X i . Thus ωi =< ω, ei >.
| {z }
δi j
Upon changing the basis, the ωi transform like the ei and the ei like the X i (see (12.7)). In particular
∂ ∂
we have for the coordinate basis ei = ∂xi , ei = dxi (< ei , ej >=< dxi , ∂x i
j >= δ j ). The change of

basis is:

∂ ∂xk ∂ ∂
i
= i k
= φi k k
∂ x̄ ∂ x̄ ∂x ∂x
∂ x̄i k
dx̄i = dx = φi k dxk
∂xk
(Similar to co- and contravariant vectors.)

1
Tensors on Tp are multilinear forms on Tp∗ and Tp , i.e. a tensor T of type


2 (for short T ∈ ⊗12 Tp ):
T (ω, X, Y ) is a trilinear form on Tp∗ × Tp × Tp . The tensor product is defined between tensors of any
type, i.e. T (ω, X, Y ) = R(ω, X)S(Y ) : T = R ⊗ S. In components:

T (ω, X, Y ) = T (ei , ej , ek ) ωi X j Y k , (12.9)

| {z } | {z }
≡T i jk ei (ω)ej (X)ek (Y )

hence T = T i jk ei ⊗ ej ⊗ ek . Any tensor of any type can therefore be obtained as a linear combination
of tensor products X ⊗ ω ⊗ ω 0 with X ∈ Tp , ω, ω 0 ∈ Tp∗ . A change of basis can be performed similarly

35
 12 DIFFERENTIABLE MANIFOLDS

to the ones for vectors and covectors:

i
T̄jk = T α βγ φi α φj β φk γ (12.10)

Trace: any bilinear form b ∈ Tp∗ ⊗ Tp determines a linear form l ∈ (Tp ⊗ Tp∗ )∗ such that l(X ⊗ ω) =
b(X, ω). In particular tr T is a linear form on tensors T of type 11 , defined by tr(X ⊗ ω) =< ω, X >.

In components with respect to a dual pair of bases we have: tr T = T α α . Similarly T i jk 7→ Sk = T i ik

defines for instance a map from tensors of type 12 to tensors of type 01 .

12.2 The tangent map

Definition: Let ϕ be a differentiable map: M → M̄ and let p ∈ M, p̄ = ϕ(p). Then ϕ induces a linear
map (“push-forward”):
ϕ∗ : Tp (M) → Tp̄ (M̄),

which we can describe in two ways:

(a) For any f¯ ∈ Fp (M̄) (F: space of all smooth functions on M (or M̄), that is C ∞ map f : M → R):

(ϕ∗ X)f¯ = X(f¯ ◦ ϕ)

(b) Let γ be a representative of X (X = γ̇p , see (12.2) and (12.3)). Then let γ̄ = ϕ ◦ γ be a
d ¯ d ¯
 
representative of ϕ∗ X. This agrees with (a) since dt f (γ̄(t))t=0 = dt (f ◦ ϕ)(γ(t))t=0 .

With respect to bases (e1 , . . . , en ) of Tp and (ē1 , . . . , ēn ) of Tp̄ (M̄), this reads X̄ = ϕ∗ X: X̄ i = (ϕ∗ )i k X k
∂ x̄i
with (ϕ∗ )i k =< ēi , ϕ∗ ek > or in case of coordinate bases: (ϕ∗ )i k = .
∂xk

Definition: The adjoint map (or “pull-back”) ϕ∗ of ϕ∗ is defined as ϕ∗ : Tp̄∗ → Tp∗ , ω̄ 7→ ϕ∗ ω̄ (= ω in

Tp∗ ) with < ϕ∗ ω̄, X >=< ω̄, ϕ∗ X >. The same result is obtained from the definition

ϕ∗ : df¯ 7→ d(f¯ ◦ ϕ), f¯ ∈ F(M̄). (12.10a)

In components, ω = ϕ∗ ω̄ reads ωk = ω̄i (ϕ∗ )i k .

−1
 i.e. maps ϕ such that ϕ exists in a neighbourhood of p̄. Note that
Consider (local) diffeomorphisms,

∂ x̄i
dim M = dim M̄ and det ∂xj 6= 0. Then ϕ∗ and ϕ∗ , as defined above, are invertible and may be
extended to tensors of arbitrary types.

1


Example: tensor of type 1

(ϕ∗ T )(ω̄, X̄) = T (ϕ∗ ω̄ , ϕ−1 X̄ ),

|{z} | ∗{z }
ω X

(ϕ∗ T̄ )(ω, X) = T̄ ((ϕ∗ )−1 ω , ϕ∗ X ).

| {z } | {z }
ω̄ X̄

36
 13 VECTOR AND TENSOR FIELDS

Here, ϕ∗ and ϕ∗ are the inverse of each other and we have

ϕ∗ (T ⊗ S) = (ϕ∗ T ) ⊗ (ϕ∗ S),

(12.11)
tr(ϕ∗ T ) = ϕ∗ (tr T ),

and similarly for ϕ∗ . In components T̄ = ϕ∗ T reads

∂ x̄i ∂xβ
T̄ki = T α β (in a coordinate basis). (12.12)
∂xα ∂ x̄k
This is formally the same as for transformation (12.10) when changing basis.

13 Vector and tensor fields

Definition: If to every point p of a differentiable manifold M a tangent vector Xp ∈ Tp (M) is assigned,
then we call the map X: p 7→ Xp a vector field on M.

∂


Given a coordinate system xi and associated basis ∂x i
p
for each Tp (M), Xp has components Xpi with
∂


Xp = Xpi ∂x i
p
and Xpi = Xp (xi ) (see (12.5)). Eq. (12.8) shows how Xpi transform under coordinate
transformations. The quantity Xf is called the derivative of f with respect to the vector field X. The
following rules apply:

X(f + g) = Xf + Xg,
(13.1)
X(f g) = (Xf )g + f (Xg) (Leibnitz rule).
The vector fields on M form a linear space on which the following operations are defined as well:

X 7→ f X (multiplication by f ∈ F),

X, Y 7→ [X, Y ] = XY − Y X (commutator).

[X, Y ], unlike XY , satisfies the Leibniz rule (13.1). The components of the commutator of two vector
fields X, Y relative to a local coordinate basis can be obtained by its action on xi . Thus using
∂ k ∂
X = X i ∂x i and Y = Y ∂xk
we get
j
[X, Y ] = (XY − Y X)xj

∂xj
Y xj = Y k = Y k δj k = Y j
∂xk
∂
XY j = X k (Y j ) = X k Y j ,k
∂xk |{z}
∂Y j
∂xk

⇒ XY j − Y X j = X k Y j ,k − Y k X j ,k

In a local coordinate basis, the bracket [∂k , ∂j ] clearly vanishes (X k = 1 and Y k = 1, and thus Y,jk = 0).
The Jacobi identity holds:
[X, [Y, Z]] + [Y, [Z, X]] + [Z, [X, Y ]] = 0. (13.2)

37
 13 VECTOR AND TENSOR FIELDS

Definition: Let Tp (M)rs be the set of all tensors of rank (r, s) defined on Tp (M) (contravariant of rank
r, covariant of rank s). If we assign to every p ∈ M a tensor tp ∈ Tp (M)rs , then the map t : p 7→ tp
defines a tensor field of type rs .

Algebraic operations on tensor fields are defined point-wise; for instance the sum of two tensor fields
is defined by (t + s̃)p = tp + s̃p where t, s̃ ∈ Tp (M)rs . Tensor products and contractions of tensor fields
are defined analogously. Multiplication by a function f ∈ F(M) is given by (f t)p = f (p)tp . In a
neighbourhood U of p, having coordinates (x1 , . . . , xn ) a tensor field can be expanded in the form
 
i1 ...ir ∂ ∂
⊗ dxj1 ⊗ . . . ⊗ dxjs .


t= t j1 ...js i
⊗ ... ⊗ i
(13.3)
| {z } ∂x 1 ∂x r

components of t relative
to the coordinate system
(x1 , . . . , xn )

If the coordinates are transformed to (x̄1 , . . . , x̄n ) the components of t transform according to

∂ x̄i1 ∂ x̄ir ∂xl1 ∂xls

t̄i1 ...ir j1 ...js ≡ tk1 ...kr l1 ...ls . . . . . . . (13.4)
∂xk1 ∂xkr ∂ x̄j1 ∂ x̄js
(We shall consider C ∞ tensor fields). Covariant tensors of order 1 are also called one-forms. The set
of tensor fields of type rs is denoted by Tsr (M).

Definition: A pseudo-Riemannian metric on a differentiable manifold M is a tensor field g ∈ T20 (M)

having the properties:

(i) g(X, Y ) = g(Y, X) for all X, Y

(ii) For every p ∈ M, gp is a non-degenerate (6= 0) bilinear form on Tp (M). This means that
gp (X, Y ) = 0 for all X ∈ Tp (M) if and only if Y = 0.

The tensor field g ∈ T20 (M) is a (proper) Riemannian metric if gp is positive definite at every point p.

Definition: A (pseudo-)Riemannian manifold is a differentiable manifold M, together with a (pseudo-)

Riemannian metric g.

13.1 Flows and generating vector fields

A flow is a 1-parametric group of diffeomorphisms: ϕt : M → M, s, t ∈ R with ϕt ◦ ϕs = ϕt+s . In
particular ϕ0 = id. Moreover, the orbits (or integral curves) of any point p ∈ M, t 7→ ϕt (p) ≡ γ(t)
shall be differentiable. A flow determines a vector field X by means of

d 
Xf = (f ◦ ϕt ) (13.5)
dt t=0

d

i.e. Xp = dt γ(t)t=0 = γ̇(0) (see (12.2) and (12.3)). γ̇(0) is the tangent vector to γ at the point
p = γ(0). At the point γ(t) we have then

d d 
γ̇(t) = ϕt (p) = (ϕs ◦ ϕt ) (p) = Xϕt (p)
dt ds s=0

38
 13 VECTOR AND TENSOR FIELDS

i.e. γ(t) solves the ordinary differential equation:

γ̇(t) = Xγ(t) , γ(0) = p. (13.6)

The generating vector field determines the flow uniquely. Not always does (13.6) admit global solutions
(i.e. for all t ∈ R), however for most purposes, “local flows” are good enough.

39
 14 LIE DERIVATIVE

14 Lie derivative
The derivative of a vector field V rests on the comparison of Vp and Vp0 at nearby points p, p0 . Since
Vp ∈ Tp and Vp0 ∈ Tp0 belong to different spaces their difference can be taken only after Vp0 has been
transported to Vp . This can be achieved by means of the tangent map ϕ∗ (Lie transport). The Lie
derivative LX R of a tensor field R in direction of a vector field X is defined by

d ∗ 
LX R = ϕt R , (14.1)
dt t=0

d ∗ 
or more explicitly (LX R)p = ϕt Rϕt (p)  . Here ϕt is the (local) flow generated by X, where
dt t=0
ϕ∗t Rϕt (p) is a tensor on Tp depending on t.

R ◦ ϕt (p)

Rp
ϕ−t∗ (R ◦ ϕt (p)) = ϕ∗t (R ◦ ϕt (p))

ϕt (p)
p


d ∗  1
LX R = ϕ R = lim (ϕ∗ R − R)
dt t t=0 t→0 t t
d
t → ϕt (p) = γ(t); Xp = dt γ(t)|t=0 = γ̇(0)
(ϕ∗ is the inverse of ϕ∗ )

Figure 8: Illustration of the Lie derivative

In order to express LX in components we write ϕt in a chart: ϕt : x 7→ x̄(t), and linearize it for small
2 i 2 i
∂ x̄ ∂ x
t: x̄i = xi + tX i (x) + O(t2 ), xi = x̄i − tX i (x̄) + O(t2 ), thus ∂x i
k ∂t = − ∂ x̄k ∂t = X ,k at t = 0.
i
∂ x̄β
As an example, let R be of type 11 . By (12.12) we have (ϕ∗t R)i j (x) = Rα β (x̄) ∂∂x


x̄α ∂xj . Taking
(according to (14.1)) a derivative with respect to t at t = 0 yields

(LX R)i j = Ri j,k X k − Rα j X i ,α + Ri β X β ,j (14.2)

∂ x̄k ∂xi ∂ x̄β 


∂ α
(first term: k
R β (x̄) α ∂xj 
= Ri j,k X k ).
∂
| x̄ {z ∂t
} |{z} ∂ x̄ t=0
Rα β,k (x̄) Xk

40
 14 LIE DERIVATIVE

Properties of LX :

(a) LX is a linear map from tensor fields to tensor fields of the same type.

(b) LX (tr T ) = tr(LX T )

(c) LX (T ⊗ S) = (LX T ) ⊗ S + T ⊗ (LX S)

(d) LX f = Xf (f ∈ F(M))

(e) LX Y = [X, Y ] (Y vector field)

(proof: (a) follows from (14.1), (b) and (c) from (12.11), (d) from (13.5), whereas (e) is more involved).

Further properties of LX : if X,Y are vector fields and λ ∈ R, then

(i) LX+Y = LX + LY , LλX = λLX

(ii) L[X,Y ] = [LX , LY ] = LX ◦ LY − LY ◦ LX

“Proof” of (ii): Apply it to f ∈ F(M),

[LX , LY ]f = LX ◦ LY f − LY ◦ LX f = LX (Y f ) − LY (Xf ) = XY f − Y Xf = [X, Y ]f = L[X,Y ] f.

Next apply it on a vector field Z:

[LX , LY ]Z = [X, [Y, Z]] − [Y, [X, Z]] = [[X, Y ], Z] = L[X,Y ] Z.

(e) Jacobi
identity

For higher rank tensors the derivation follows from the use of (c).
If [X, Y ] = 0 then LX LY = LY LX and for φ and ψ, which are the flows generated by X and Y , one
finds: φs ◦ ψt = ψt ◦ φs .

41
 15 DIFFERENTIAL FORMS

15 Differential forms
0


Definition: A p-form Ω is a totally antisymmetric tensor field of type p

Ω(Xπ(1) , . . . , Xπ(p) ) = (sign π)Ω(X1 , . . . , Xp )

for any permutation π of {1, . . . , p} (π ∈ Sp (group of permutations)) with sign π being its parity. For
p > dim M, Ω ≡ 0. Any tensor field of type p0 can be antisymmetrized by means of the operation A:

1 X
(AT )(X1 , . . . , Xp ) = (sign π)T (Xπ(1) , . . . , Xπ(p) ) (15.1)
p!
π∈Sp

with A2 = A. The exterior product of a p1 -form Ω1 with a p2 -form Ω2 is the (p1 + p2 )-form:

(p1 + p2 )!
Ω1 ∧ Ω2 = A(Ω1 ⊗ Ω2 ) (15.2)
p1 ! p2 !
Properties:

• Ω1 ∧ Ω2 = (−1)p1 p2 Ω2 ∧ Ω1
(p1 + p2 + p3 )!
• Ω1 ∧ (Ω2 ∧ Ω3 ) = (Ω1 ∧ Ω2 ) ∧ Ω3 = A(Ω1 ⊗ Ω2 ⊗ Ω3 )
p1 ! p2 ! p3 !
The components in a local basis (e1 , . . . , en ) of 1-forms are

Ω = Ωi1 ...ip ei1 ⊗ . . . ⊗ eip = AΩ

= Ωi1 ...ip A(ei1 ⊗ . . . ⊗ eip )

n n
X X 1 i1
= ··· Ωi1 ...ip e ∧ . . . ∧ e ip
i1 =1 ip =1
p!
X
= Ωi1 ...ip ei1 ∧ . . . ∧ eip (15.3)
1≤i1 <...<ip ≤n

A covariant tensor of rank p, which is antisymmetric under exchange of any pair of indices (i.e. is a
p-form), in n dimensions has np = (n−p)!p!
n!


independent components.

Examples:

• For 1-forms A, B (vector fields) we have

(A ∧ B)ik = Ai Bk − Ak Bi = (−1)(B ∧ A)ik .

• For a 2-form A and a 1-form B

(A ∧ B)ikl = Aik Bl + Akl Bi + Ali Bk , (15.4)

42
 15 DIFFERENTIAL FORMS

since
(1 + 2)!
A∧B = A(A ⊗ B)
1! 2!
3! 1
= (Aik Bl ) ei ∧ ek ∧ el
1! 2! 3!
1
= (Aik Bl )ei ∧ ek ∧ el
2
11
= (Aik Bl + cyclic permutations)ei ∧ ek ∧ el
23
1 i
= (Aik Bl + cyclic permutations) e ∧ ek ∧ el .
3!

Thus by comparing with (15.3) we get (15.4).

15.1 Exterior derivative of a differential form

The derivative df of a 0-form f ∈ F is the 1-form df (X) = Xf : the argument X (vector) acts as
∂f
a derivation. In a local coordinate basis: df = dxi . The exterior derivative is performed by an
∂xi
operator d applied to forms, converting p-forms to (p + 1)-forms. The derivative dΩ of a 1-form Ω is
given by
dΩ(X1 , X2 ) = X1 Ω(X2 ) − X2 Ω(X1 ) − Ω([X1 , X2 ]). (15.5)

This expression is verified as follows:

∂
X1 Ω(X2 ) = X1 hΩ, X2 i = X1i Ωk X2k = X1i Ωk,i X2k + X1i Ωk X2,i
k


i
,
| {z } ∂x
|{z}
1-form
,i

X2 Ω(X1 ) = X2k Ωi,k X1i + X2k Ωi X1,k

i
,

Ω([X1 , X2 ]) = hΩ, X1 X2 − X2 X1 i = Ωi (X1 X2 − X2 X1 )i = Ωi X1k X2,k

i
− X2k X1,k
i


,

then

dΩ(X1 , X2 ) = (Ωk,i − Ωi,k )X1i X2k .

1 (1+1)!
This is manifestly a 2-form (the coefficient also fits the expectations: 2! 1!1! = 1). One can easily
verify that
dΩ(f X1 , X2 ) = f dΩ(X1 , X2 ). (15.6)

For Ω ∧ f = f Ω (as f is a 0-form), the product rule

d(Ω ∧ f ) = dΩ ∧ f − Ω ∧ df

43
 15 DIFFERENTIAL FORMS

applies, as one can verify

d(Ω ∧ f )(X1 , X2 ) = X1 (f Ω)(X2 ) − X2 (f Ω)(X1 ) − (f Ω)([X1 , X2 ]),

(15.5)

and

∂ ∂ ∂f
X1 (f Ω)(X2 ) = X1i (f Ωk X2k ) = f X1i i Ωk X2k + X1i i Ωk X2k .


∂x i ∂x ∂x
| {z } | {z }
f X1 Ω(X2 ) df (X1 )Ω(X2 )

So

d(f ∧ Ω)(X1 , X2 ) = f dΩ(X1 , X2 ) + Ω(X2 )df (X1 ) − Ω(X1 )df (X2 ) . (15.7)
| {z } | {z }
dΩ∧f −Ω∧df
2
Moreover we have d f = 0, since

d2 f (X1 , X2 ) = X1 df (X2 ) − X2 df (X1 ) − df ([X1 , X2 ])

(15.5)

= X1 X2 f − X2 X1 f − [X1 , X2 ]f = 0.

(15.8)

The generalization of the definition to a p-form Ω gives

p+1
X
dΩ(X1 , .. , Xp+1 ) = (−1)i−1 Xi Ω(X1 , .. , X̂i , .. , Xp+1 )
i=1

p+1
X
+ (−1)i+j Ω([Xi , Xj ], X1 , .. , X̂i , .. , X̂j , .. , Xp+1 ), (15.9)
i<j

whereˆmeans omission, e.g. (X1 , X̂2 , X3 ) = (X1 , X3 ).

One can show that the following properties hold:

(a) d is a linear map from p-forms to p + 1-forms,

(b) d(Ω1 ∧ Ω2 ) = dΩ1 ∧ Ω2 + (−1)p1 Ω1 ∧ dΩ2 ,

(c) d2 = 0, i.e. d(dΩ) = 0,

(d) df (X) = Xf (f ∈ F),

By means of (a)-(d) we have an alternative definition of d. By eq. (15.3) we have with respect to a
coordinate basis
1
Ω= Ωi ...i dxi1 ∧ . . . ∧ dxip , and hence (15.10a)
p! 1 p
1
dΩ |{z}
= dΩi1 ...ip ∧ dxi1 ∧ . . . ∧ dxip (15.10b)
p!
ddxip =0

44
 15 DIFFERENTIAL FORMS

Components:

p! dΩ = Ωi1 ...ip ,i0 dxi0 ∧ dxi1 ∧ . . . ∧ dxip

= −Ωi0 i2 ...ip ,i1 dxi0 ∧ . . . ∧ dxip

= (−1)k Ωi0 ...îk ...ip ,ik dxi0 ∧ . . . ∧ dxip (k = 0, . . . , p)

p
1 1 X
⇒ dΩ = (−1)k Ωi0 ...îk ...ip ,ik dxi0 ∧ . . . ∧ dxip (15.11)
p! p + 1
| {z } k=0
| {z }
1
(p+1)! (dΩ)i0 ...ip

Examples:
• p = 1:
(dΩ)ik = Ωk,i − Ωi,k (15.12)

• p = 2:
(dΩ)ikl = Ωik,l + Ωkl,i + Ωli,k (15.13)

Consider a map ϕ : M → M̄ and ϕ∗ : Tp̄∗ (M̄) → Tp∗ (M); then

ϕ∗ ◦ d = d ◦ ϕ∗ . (15.14)

A “proof” is found by using (15.10), (12.11) and property (b). It suffices to verify (15.14) on 0-forms
and 1-forms. For 0-forms f¯, (15.14) is identical to (12.10a). For 1-forms which are differentials df¯,
due to (c) we have

(ϕ∗ ◦ d)(df¯) = 0 (d2 f¯ = 0),

(d ◦ ϕ∗ )(df¯) = d(ϕ∗ ◦ df¯) = d(d(f¯ ◦ ϕ)) = d2 (f¯ ◦ ϕ) = 0.

(12.10a)
(ϕ∗ ◦ df¯)
=d(f¯◦ϕ)

d ∗ 

Setting ϕ = ϕt (the flow generated by X) and forming (14.1) (LX R = dt ϕt R t=0 ), one obtains the
infinitesimal version of (15.14):
LX ◦ d = d ◦ LX . (15.15)

Definition: A p-form ω with

• ω = dη is exact

• dω = 0 is closed
An exact p-form is always closed (d2 η = 0), but the converse is not generally true (Poincaré lemma
gives conditions under which the converse is valid).5 6

5η is not unique since gauge transformations η 7→ η + dρ, with ρ any (p − 1)-form, leave dη unchanged.
6 This is a generalization of the results of three-dimensional vector analysis: rot grad f = 0 and div rot ~k = 0.

45
 15 DIFFERENTIAL FORMS

The integral of an n-form:

∂ x̄i
 
M is orientable within an atlas of “positively oriented” charts, if det > 0 for any change of
∂xj
coordinates. For an n-form ω (n = dim M):
1 i1
ω = ωi1 ...in dx ∧ . . . ∧ dxin = ω1...n dx1 ∧ . . . ∧ dxn (15.16)
n! | {z }
ω(x)

is determined by the single component ω(x); under a change of coordinates ω(x) transforms like

∂xi1 ∂xin
 i
∂x
ω̄(x̄) = ω̄1...n = ωi1 ...in ··· = ω(x) det . (15.17)
| {z } ∂ x̄1 ∂ x̄n ∂ x̄j
totally
antisymmetric

The integral of a n-form is defined as follows:

Z Z
ω = dx1 . . . dxn ω(x1 , . . . , xn ) (if the support of ω is contained in a chart U).
M U

ThisZ integral is independent

Z of the choice of coordinates,
 i  since in different coordinates
∂x 
dx1 . . . dxn ω(x) = dx̄1 . . . dx̄n ω(x) det

and (15.17) applies. 7 8
∂ x̄j 

15.2 Stokes theorem

Let D be a region in a n dimensional differentiable manifold M. The boundary ∂D consists of those
p ∈ D whose image x in some chart satisfies e.g. x1 = 0. One can show that ∂D is a closed (n − 1)
dimension submanifold of M. If M is orientable then ∂D is also orientable. D shall have a smooth
boundary and be such that D̄ is compact. Then for every (n − 1)-form ω we have
Z Z
dω = ω (15.18)
D ∂D

15.3 The inner product of a p-form

Definition: Let X be a vector field on M. For any p-form Ω we define the inner product as

(iX Ω)(X1 , . . . , Xp−1 ) ≡ Ω(X, X1 , . . . , Xp−1 ) (15.19)

(and zero if p = 0).

Properties:
7 Actually,it is often impossible to cover the whole manifold with a single set of coordinates. In the general case it is
necessary to introduce different sets of coordinates in different overlapping patches of the manifold, with the constraint
that in the overlap between the patch with coordinate xi and another patch with coordinate x̄i , the xi can be expressed
in a smooth one-to-one way as functions of x̄i and vice-versa (orientable manifold).
8 The integral over a p-form over the overlap between two patches (xi and x̄i ) can be evaluated using either coordinate
 
∂xi
system, provided det ∂ x̄j
> 0.

46
 15 DIFFERENTIAL FORMS

(a) iX is a linear map from p-forms to (p − 1)-forms,

(b) iX (Ω1 ∧ Ω2 ) = (iX Ω1 ) ∧ Ω2 + (−1)p1 Ω1 ∧ (iX Ω2 ),

(c) iX 2 = 0,

(d) iX df = Xf = hdf, Xi with f ∈ F(M),

(e) LX = iX ◦ d + d ◦ iX .

Proof of (e): for 0-forms f we have

LX f = Xf,

iX ◦ df + 
d ◦iX
f = iX df = Xf,
=0

and for 1-forms df

LX df = d(LX f ) = d(Xf ),
(15.15)
LX ◦d=d◦LX

iX ◦  + d ◦ iX df = d(Xf ).
ddf
=0

Application: Gauss theorem

Let X be a vector field. Then d(iX η) is an n-form with dim M = n. η is an n-form, and if ηp 6= 0 ∀p ∈
M, then η is a “volume form”. A function divη X ∈ F is defined through

(divη X)η = d(iX η) = LX η.9 (15.20)

We can apply Stokes theorem since d(iX η) is an n-form and thus iX η an (n − 1)-form:
Z Z Z
d(iX η) = (divη X)η = iX η. (15.21)
D D ∂D
p
The standard volume form η is given by η = |g| dx1 ∧ . . . ∧ dxn .

9 dη = 0, thus LX = iX ◦ d + d ◦ iX applied on η gives LX η = iX ◦ 

d η + d(iX η).


47
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

Expression for divη X in local coordinates:

∂
Let η = a(x) dx1 ∧ . . . ∧ dxn , X = X i ∂x i . Then since (divη X)η = LX η, we have (using property (c)

of the Lie derivative):

n
X
LX η = (Xa) dx1 ∧ . . . ∧ dxn + a dx1 ∧ . . . ∧ d(Xxi ) ∧ . . . ∧ dxn .
i=1

∂ i
Since d(Xxi ) = d(X k k
x ) = dX i (x) = X i ,j dxj , but dx1 ∧ . . . ∧ dxj ∧ . . . ∧ dxn 6= 0 only if j = i
∂x
| {z }
i
δk
(otherwise we have two identical dxi ) we find
n
X
Xa dx1 ∧ . . . ∧ dxn + a
LX η = |{z} X i ,i dx1 ∧ . . . ∧ dxn
Xi ∂a i=1
∂xi

1
= (X i a,i + aX i ,i ) η
a
1 1 p 
⇒ divη X = (aX i ),i = p |g|X i for the “standard” η. (15.22)
a |g| ,i

16 Affine connections: Covariant derivative of a vector field

Definition: An affine (linear) connection or covariant differentiation on a manifold M is a mapping
∇ which assigns to every pair X, Y of C ∞ vector fields on M another C ∞ vector field ∇X Y with the
following properties:

(i) ∇X Y is bilinear in X and Y ,

(ii) if f ∈ F(M), then

∇f X Y = f ∇X Y,

∇X (f Y ) = f ∇X Y + X(f )Y.

(16.1)

Lemma: Let X and Y be vector fields. If X vanishes at the point p on M, then ∇X Y also vanishes at
p.

∂
Proof: Let U be a coordinate neighbourhood of p. On U we have the representation X = ξ i ∂x i,

ξ i ∈ F(U) with ξ i (p) = 0. Then (∇X Y )p = ∇ξi ∂ Y = ξ i (p)[∇ ∂ i Y ]p = 0.

∂xi | {z } ∂x
=0
Since ∇X Y produces again a vector field, the result of the covariant differentiation can only be a linear
combination of again the basis in the current chart. This leads us to the following statement:

48
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

Definition: One sets, relative to a chart (X 1 , . . . , X n ) for U ⊂ M:

 
∂ ∂
∇ ∂ = Γkij (16.2)
∂xi ∂xj ∂xk

The n3 functions Γkij ∈ F(U) are called Christoffel symbols (or connection coefficients) of the connection
∇ in a given chart.10

The Christoffel symbols are not tensors:

∂xk ∂
 
∂ ∂
∇ ∂a b
= Γ̄cab c = Γ̄cab c k . (16.3)
∂ x̄ ∂ x̄ ∂ x̄ ∂ x̄ ∂x
If we use (16.1):

∂xj ∂ ∂xi ∂xj k ∂

      j 
∂ ∂ ∂x ∂
∇ ∂ =∇ ∂xi ∂

= Γ +
∂ x̄a ∂ x̄b ∂ x̄a ∂xi ∂ x̄b ∂xj ∂ x̄a ∂ x̄b ij ∂xk ∂xi ∂ x̄b ∂xj

∂xi ∂xj k ∂ ∂ 2 xj ∂
= a b
Γij k + .
∂ x̄ ∂ x̄ ∂x ∂ x̄a ∂ x̄b ∂xj

Comparison with 16.3:

∂xk c ∂xi ∂xj k ∂ 2 xk

c
Γ̄ab = a b
Γij +
∂ x̄ ∂ x̄ ∂ x̄ ∂ x̄a ∂ x̄b
∂xi ∂xj ∂ x̄c k ∂ x̄c ∂ 2 xk
⇒ Γ̄cab = Γij + (16.4)
∂ x̄a ∂ x̄b ∂xk ∂xk ∂ x̄a ∂ x̄b
The second term is not compatible with being a tensor.
If for every chart there exist n3 functions Γkij which transform according to (16.4) under a change
of coordinates, then one can show that there exists a unique affine connection ∇ on M which satisfies
(16.3).

Definition: for every vector field X we can introduce the tensor ∇X ∈ T11 (M) defined by

∇X(Y, ω) ≡ hω, ∇Y Xi , (16.5)

where ω is a one-form. ∇X is called the covariant derivative of X.

In a chart (x1 , . . . , xn ), let X = ξ i ∂i and ∇X = ξ i ;j dxj ⊗ ∂i (< dxi , ∂i >= δki ):

ξ i ;j = ∇X(∂j , dxi ) = dxi , ∇∂j X = dxi , ξ k ,j ∂k + ξ k Γδjk ∂δ = ξ i ,j + Γijk ξ k


 11
(16.6)

10 Fora pseudo-Riemannian manifold, the corresponding connection coefficients are given by (9.6) or (9.11).
11 semicolonshall denote the covariant derivative (”normal derivative” + additional terms, that vanish in (cartesian)
Euclidean or Minkowski space)

49
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

16.1 Parallel transport along a curve

Definition: let γ : I → M be a curve in M with velocity field γ̇(t), and let X be a vector field on some
open neighbourhood of γ(I). X is said to be autoparallel along γ if

∇γ̇ X = 0. (16.7)

DX ∇X
The vector ∇γ̇ X is sometimes denoted as dt or dt (covariant derivative along γ). In terms of
i
i dx
coordinates, we have X = ξ ∂i , γ̇ = dt ∂i (see (12.3)). With (16.1) and (16.2) we get

∇γ̇ X = ∇ dxi ∂ (ξ k ∂k )
dt i

dxi
= ∇∂i (ξ k ∂k )
dt
dxi h k j i
= ξ Γik ∂j + ∂i ξ k ∂k
dt
dxi  j k
ξ Γij ∂k + ∂i ξ k ∂k

=
dt
 k i

dξ k dx j
= + Γij ξ ∂k , (16.8)
dt dt

dxi ∂ξ k dξ k
where we used = . This shows that ∇γ̇ X only depends on the values of X along γ. In
dt ∂xi dt
terms of coordinates we get for (16.7)

dξ k dxi j
+ Γkij ξ = 0. (16.9)
dt dt

For a curve γ and any two point γ(s) and γ(t) consider the mapping

τt,s : Tγ(s) (M) → Tγ(t) (M),

which transforms a vector v(s) at γ(s) into the parallel transported vector v(t) at γ(t). The mapping
τt,s is the parallel transport along γ from γ(s) to γ(t). We have τs,s = 1 and τr,s ◦ τs,t = τr,t .
We can now give a geometrical interpretation of the covariant derivative that will be generalized
to tensors. Let X be a vector field along γ, then

d 
∇γ̇ X(γ(t)) = τt,s X(γ(s)), (16.10)
ds s=t

Proof: Let’s work in a given chart. By construction, v(t) = τt,s v(s) with v(s) ∈ Tγ(s) (M) and due to
(16.8) it satisifies: v̇ i + Γikj ẋk v j = 0. If we write (τt,s v(s))i = (τt,s )i j v j (s) = v i (t) (with τt,s = (τs,t )−1

50
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

τt,s v(t)

v(s) M

γ
γ(t)

γ(s)

Figure 9: Illustration of parallel transport.

and τs,s = 1), we get


i d 
v̇ (s) = v i (t)
dt t=s

d  h
i j
i
= (τ t,s ) j v (s)
dt t=s
  
d 
= (τt,s ) j v j (s)
i
dt t=s

= −Γikj ẋk v j (s).


d 
⇒ (τt,s )i j = −Γikj ẋk (16.11)
dt t=s
 
Since τt,s = (τs,t )−1 , ds
d 
(τ )i = − dt
s=t t,s j
d
(τ )i = Γikj ẋk . Then
t=s t,s j
   i 
d  i d  j d 
[τt,s X(γ(s))] = τt,s X + X i (γ(s))
ds s=t ds s=t j
ds 
s=t

dxj (γ(s)) 

= Γikj ẋk X j +X i
,j ,
ds 
s=t

dξ i
which is again (16.8) (X = ξ i ∂i and the second term gives dt ).

51
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

Definition: If ∇X Y = 0, then Y is said to be parallel transported with respect to X.

Geometrical interpretation of parallel transport: Consider the differential dAi = Ai ,j dxj = Ai (x +

dx) − Ai (x). In order that the difference of two vectors be a vector, we have to consider them at the
same position. The transport has to be chosen such that for cartesian coordinates there is no change
in transporting it. The covariant derivative exactly achieves this.

Definition: Let X be a vector field such that ∇X X = 0. Then the integral curves of X are called
geodesics.

d i
In local coordinates xi the curve is given by (using (12.3) and (13.6)) the requirement dt x (t) =
2 i i
i d x dX
X (x(t)). Inserting this into (16.8) and using dt2 = dt , we get

ẍk + Γkij ẋi ẋj = 0. (16.12)

For a vector parallel transported along a geodesic, its length and angle with the geodesic does not
change.

16.2 Round trips by parallel transport

Consider (16.8) and denote ξ i = v i , thus

v̇ i = −Γikj ẋk v j . (16.13)

Let γ : [0, 1] → M be a closed path, wih γ(0) = p = γ(1). Displace a vector v0 ∈ Tp (M) parallel along
γ and obtain the field v(t) = τt,0 v0 ∈ Tγ(t) (M). We assume that the closed path is sufficiently small
(such that we can work in the image of some chart), thus we can expand Γikj (x) around the point
x(0) = x0 on the curve:

∂ i
Γikj (x) Γikj (x0 ) ρ
xρ0 )

' + (x − Γ kj (x) + ··· (16.14)
∂xρ 
x=x0

Thus (16.13) is to first order in (xk − xk0 ):

Zt Zt Zt
i i
v̇ dt = v (t) − v0i =− Γikj j k
v ẋ dt ≈
|{z} −Γikj (x0 )v0j ẋk dt,
0 0 'v0j 0
| {z }
xk (t)−xk
0

taking only the first term in the expansion of Γ. And hence,

v i (t) = v0i − Γikj (x0 )(xk (t) − xk0 )v0j + · · · (16.15)

52
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

N 1

W
C E
α

4 3

Figure 10: Illustration of the path dependence of parallel transport on a curved space: vector 1 at N
can be parallel transported along the geodesic N-S to C, giving rise to vector 2. Alternatively, it can
be first transported along the geodesic N-S to E (vector 3) and then along E-W to C to give the vector
4. Clearly these two are different. The angle α between them reflects the curvature of the two-sphere.

By plugging (16.14) and (16.15) into (16.13), we obtain an equation valid to second order:
Z1 Z1
i
v̇ dt = − Γikj ẋk v j dt (16.16)
0 0

Z1  
∂ i
i
v (1) − v0i '− Γikj (x0 ) + (x − ρ
xρ0 ) Γ (x0 ) + · · · ×
∂xρ kj
0

 
v0j − Γjk̃j̃ (x0 )(xk̃ (t) − xk̃0 )v0j̃ + · · · ẋk dt.
53
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

Multiplying out and discarding terms of third order or higher in xk − xk0 , we get:

Z1   Z1
∂ i
i
v (1) ' v0i − Γikj (x0 )v0j k
ẋ dt − Γ (x0 ) − Γkj̃ (x0 )Γρj (x0 ) v0j (xρ − xρ0 )ẋk dt.
i j̃
∂xρ kj
0 0
| {z }
xk (1)−xk (0)
=0

R1
Since we are considering a closed path ( 0 ẋ dt = xk (1) − xk (0) = 0),

  Z1
∂ i
∆v i = v i (1) − v i (0) = − Γ (x 0 ) − Γi
(x
kl 0 )Γl
(x0 ) v0j xρ ẋk dt,
∂xρ kj ρj
0

with

I1 I1 I1 I1
ρ k d ρ k
x ẋ dt = (x x ) dt − ẋ x dt = − ẋρ xk dt,
ρ k
dt
0 0 0 0
| {z }
=0

antisymmetric in (ρ, k). Then

  Z1
i 1 ∂ i i l ∂ i j
∆v = − Γ − Γkl Γρj − Γ + Γρl Γkj (x0 ) v0 xρ ẋk dt,
i l
2 ∂xρ kj ∂xk ρj
| {z } 0
i
−Rjkρ

Z1
1 i
i
∆v = Rjkρ (x0 )v0j xρ ẋk dt. (16.17)
2
0
i
We shall see that Rjkρ is the curvature tensor.

i ∂ i ∂ i
Rjkρ = Γρj − Γ + Γlρj Γikl − Γlkj Γiρl (16.18)
∂xk ∂xρ kj
Thus an arbitrary vector v i will not change when parallel transported around an arbitrary small closed
i
curve at x0 if and only if Rjkρ vanishes at x0 .

16.3 Covariant derivatives of tensor fields

The parallel transport is extended to tensors by means of the requirements:

τs,t (T ⊗ S) = (τs,t T ) ⊗ (τs,t S),

τs,t tr(T ) = tr(τs,t T ),

τs,t c = c (c ∈ R).

54
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

1


For e.g. a covariant vector ω, hτs,t ω, τs,t Xiγ(s) = hω, Xiγ(t) and for a tensor of type 1 : τs,t T (τs,t ω, τs,t X) =
T (ω, X). In components:
i β
(τs,t T ) k = T α β (τs,t )i α (τs,t )k (16.19)

(τi k is inverse transpose of τ i k ). The covariant derivative ∇X (X vector field, T tensor field) associated
to τ is 
d 
(∇X T )p = τ0,t Tγ(t)  , (16.20)
dt t=0

with γ(t) any curve with γ(0) = p and γ̇(0) = Xp (generalization of (16.10)).

Properties of the covariant derivative:

r


(a) ∇X is a linear map from tensor fields to tensor fields of the same type r ,

(b) ∇X f = Xf ,

(c) ∇X (tr T ) = tr(∇X T ),

(d) ∇X (T ⊗ S) = (∇X T ) ⊗ S + T ⊗ (∇X S).

This follows from the properties of τs,t . For a 1-form ω we have:

(∇X ω)(Y ) = tr(∇X ω ⊗ Y )

= tr(∇X (ω ⊗ Y )) − tr(ω ⊗ ∇X Y )

= ∇X tr(ω ⊗ Y ) − ω(∇X Y )

= Xω(Y ) − ω(∇X Y ). (16.21)

1


General differentiation rule for a tensor field of type 1 :

(∇X T )(ω, Y ) = XT (ω, Y ) − T (∇X ω, Y ) − T (ω, ∇X Y ) (16.22)

Due to (a)-(d), the operation ∇X is completely determined by its action on vector fields Y , which are
the affine connections (see (16.1) and (16.2)).

16.4 Local coordinate expressions for covariant derivative

Let T ∈ Tpq (U) be a tensor of rank (p, q) with local coordinates (x1 , . . . , xn ) valid in a region U. We
have T i1 ...ip j1 ...jq ∂i1 ⊗ . . . ⊗ ∂ip ⊗ dxj1 ⊗ . . . ⊗ dxjq and X = X k ∂k . Let us use

XT i1 ...ip j1 ...jq = X k T i1 ...ip j1 ...jq ,k (16.23)

and write (16.2):

∇X (∂i ) = X k ∇∂k ∂i = X k Γlki ∂l . (16.24)

55
 16 AFFINE CONNECTIONS: COVARIANT DERIVATIVE OF A VECTOR FIELD

Moreover,
(16.21)
(∇X dxj )(∂i ) X dxj , ∂i − dxj , ∇X ∂i



=
| {z }
δj i
| {z }
0

= −X k Γjki ,

or ∇X dxj = −X k Γjki dxi . (16.25)

Using (16.23), (16.24) and (16.25) for ω j = dxj , Yi = ∂i we obtain the following expression for ∇X T :

i
T i1 ...ip j1 ...jq ;k = T i1 ...ip j1 ...jq ,k + Γikl1 T li2 ...ip j1 ...jq + . . . + Γklp T i1 ...ip−1 l j1 ...jq

− Γlkj1 T i1 ...ip lj2 ...jq − . . . − Γlkjq T i1 ...ip j1 ...jq−1 l . (16.26)

Examples:

• Contravariant and covariant vector fields:

ξ i ;k = ξ i ,k + Γikl ξ l ,

ηi;k = ηi,k − Γlki ηl ,

• Kronecker tensor:
δ i j;k = 0,

1


• Tensor 1 :
T i k;r = T i k,r + Γirl T l k − Γlrk T i l .

The covariant derivative of a tensor is again a tensor. Consider the covariant derivative of the metric
gµν :
∂gµν
gµν;λ = − Γρλµ gρν − Γρλν gρµ . (16.27)
∂xλ
Inserting into this the expressions of Γρλµ given by (9.11) leads us to

gµν;λ = 0. (16.28)

This is not surprising since gµν;λ vanishes in locally inertial coordinates and being a tensor it is then
zero in all systems.

Covariance principle: Write the appropriate special relativistic equations that hold in the absence of
gravitation, replace ηαβ by gαβ , and replace all derivatives with covariant derivatives (, → ;). The resulting
equations will be generally covariant and true in the presence of gravitational fields.

56
 17 CURVATURE AND TORSION OF AN AFFINE CONNECTION, BIANCHI IDENTITIES

17 Curvature and torsion of an affine connection, Bianchi iden-

tities
Let an affine connection be given on M, let X, Y , Z be vector fields.

Definition:

T (X, Y ) = ∇X Y − ∇Y X − [X, Y ] (17.1)

R(X, Y ) = ∇X ∇Y − ∇Y ∇X − ∇[X,Y ] (17.2)

1


T (X, Y ) is antisymmetric and f -linear in X, Y and then defines a tensor of type 2 through:
(ω, X, Y ) → hω, T (X, Y )i is thus a 12 tensor field called the torsion tensor.

f -linearity:
T (f X, gY ) = f gT (X, Y ) f, g, ∈ F(M).

In local coordinates, the components of the torsion tensor are given by:
* +
k

k  k
T ij = dx , T (∂i , ∂j ) = dx , ∇∂i ∂j −∇∂j ∂i − [∂i , ∂j ]
| {z } | {z }
=Γlij ∂l =0

= Γkij − Γkji (17.3)


(using that dxk , ∂l = δ k l ). In particular, we have T k ij = 0 ⇔ Γkij = Γkji .

R(X, Y ) = −R(Y, X) is antisymmetric in X,Y . The vector field R(X, Y )Z is f -linear in X, Y , Z:

(R(f X, gY ) hZ = f gh R(X, Y )Z; f, g, h ∈ F(M)). R determines a tensor of type 13 : the Riemann

tensor or curvature tensor.

(ω, Z, X, Y ) → hω, R(X, Y )Zi ≡ Ri jkl ωi Z j X k Y l

In components with respect to local coordinates:

Ri jkl = dxi , R(∂k , ∂l )∂j = dxi , (∇∂k ∇∂l − ∇∂l ∇∂k )∂j 12




= dxi , ∇∂k (Γslj ∂s ) − ∇∂l (Γskj ∂s )



= Γilj,k − Γikj,l + Γslj Γiks − Γskj Γils . (17.4)

12 Notice that ∇[∂ , ∂ ] ∂j = 0.

k l
| {z }
=0

57
 17 CURVATURE AND TORSION OF AN AFFINE CONNECTION, BIANCHI IDENTITIES

Eq. (17.4) is exactly the the same as defined in (16.18). It is antisymmetric in the last two indices:
Ri jkl = −Ri jlk .

Definition: The Ricci tensor is the following contraction of the curvature tensor:

Rjl ≡ Ri jil = Γilj,i − Γiij,l + Γslj Γiis − Γsij Γils (17.5)

The scalar curvature is the trace of the Ricci tensor:

R ≡ g lj Rjl = Rl l (17.6)

Example: For a pseudo-Riemannian manifold the connection coefficients are given by (9.11). Consider
a two-sphere (which is a pseudo Riemannian manifold) with the metric ds2 = a2 (dθ2 + sin2 θdφ2 ),
then: ! !
2 1 0 θφ 1 1 0
gθφ = a , g = 2 1
.
0 sin2 θ a 0 sin2 θ
The non-zero Γ are:

Γθφφ = − sin θ cos θ,

Γφθφ = Γφφθ = cot θ.

The Riemann tensor is given by

θ
Rφθφ = ∂θ Γθφφ − ∂φ Γθθφ + Γθθλ Γλφθ − Γθφλ Γλθφ

= (sin2 θ − cos2 θ) − 0 + 0 − (− sin θ cos θ) cot θ

= sin2 θ.

The Ricci tensor has the following components:

θ φ
Rφφ = Rφθφ + Rφφφ = sin2 θ,
| {z }
=0

Rθθ = 1,

Rθφ = Rφθ = 0.

The Ricci scalar is

R = g θθ Rθθ + g φφ Rφφ +g θφ Rθφ +g φθ Rφθ

|{z} |{z} |{z} |{z} |{z} |{z}
1 1 1
sin2 θ 0 0
a2 a2 sin2 θ

1 1
= 2
+ 2 2 sin2 θ
a a sin θ
2
= .
a2

58
 17 CURVATURE AND TORSION OF AN AFFINE CONNECTION, BIANCHI IDENTITIES

The Ricci scalar is constant over this two-sphere and positive, thus the sphere is “positively curved”.
13 14 15

17.1 Bianchi identities for the special case of zero torsion

X, Y and Z are vector fields, then

R(X, Y )Z + cyclic = 0 (1st Bianchi identity), (17.7)

(∇X R)(Y, Z) + cyclic = 0 (2nd Bianchi identity). (17.8)

Proof of the 1st identity: Torsion = 0 ⇒ ∇X Y − ∇Y X = [X, Y ]. Then

(∇X ∇Y − ∇Y ∇X )Z + (∇Z ∇X − ∇X ∇Z )Y

+ (∇Y ∇Z − ∇Z ∇Y )X − ∇[X,Y ] Z − ∇[Z,X] Y − ∇[Y,Z] X

= ∇X (∇Y Z − ∇Z Y ) − ∇[Y,Z] X + cyclic

= [X, [Y, Z]] + cyclic

=0 due to the Jacobi identity (13.2).

(See textbooks for proof of the 2nd Bianchi identity.)

13 For a position independent metric (e.g. Cartesian coordinates) the Riemann tensor (and thus the scalar curvature)

vanishes as the Γ vanish.  

14 For 1 0
a plane with polar coordinates we get a position dependent metric : ds2 = dr2 + r2 dθ2 and thus the
0 r2
Γ do not vanish. However the curvature vanishes.
15 The curvature does not depend on the choice of coordinates.

59
 18 RIEMANNIAN CONNECTIONS

18 Riemannian connections
Metric: Let M be equipped with a pseudo-Riemannian metric: a symmetric, non-degenerate tensor
field: g(X, Y ) = hX, Y i of type 02 .

• Non-degenerate means that, for any point p ∈ M, X, Y ∈ Tp , one has gp (X, Y ) = 0 ∀Y ∈ Tp ⇒

X = 0.

• In components, hX, Y i = gik X i Y k , gik = gki (symmetric) and det gik 6= 0.

• With the metric we can lower and raise indices:

X̃i = gik X k , ω̃ i = g ik ωk ,

where g ik denotes the inverse of gik . It also works for tensor fields of different types: T i k =
Tlk g il = T il glk .

• Given a basis (e1 , . . . , en ) of Tp , the covectors of the dual basis (e1 , . . . , en ) become themselves
vectors; indeed ei = gij ej .

Riemann connection: The metric tensor g at a point p in M is a symmetric 02 tensor. It assigns a

p
magnitude |g(X, X)| to each vector X on Tp (M), denoted by d(X) and defines the angle between
any two vectors X, Y (6= 0) on Tp (M) via
 
g(X, Y )
a(X, Y ) = arccos . (18.1)
d(X)d(Y )
π
If a(X, Y ) = 2 then X and Y are orthogonal. Further observations:

Rt2
• The length of a curve with tangent vector X between t1 and t2 is L(t1 , t2 ) = d(X) dt.
t1

• If (ea ) is a basis of Tp (M), the components of g with respect to this basis are gab = g(ea , eb ).

• Like in special relativity we classify vectors at a point as timelike (g(X, X) > 0), null (g(X, X) =
0) and space like (g(X, X) < 0).

Definition: let (M, g) be a pseudo-Riemannian manifold. An affine connection is a metric connection

if parallel transport along any smooth curve γ on M preserve the inner product: for autoparallel fields
X(t), Y (t) (see (16.7)), gγ(t) (X(t), Y (t)) is independent of t along γ.

Theorem: an affine connection ∇ is metric if and only if (no proof)

∇g = 0. (18.2)

Eq. (18.2) is equivalent for (g(Y, Z)) to

∇X g = 0 = Xg(Y, Z) − g(∇X Y, Z) − g(Y, ∇X Z)

60
 18 RIEMANNIAN CONNECTIONS

or

Xg(Y, Z) = g(∇X Y, Z) + g(Y, ∇X Z). (18.3)

Theorem: For every pseudo-Riemannian manifold (M, g), there exists a unique affine connection such
that
(a) ∇ has vanishing torsion (∇ is symmetric),

(b) ∇ is metric.

Proof: T = 0 (vanishing torsion) means ∇X Y = ∇Y X + [X, Y ]. Inserting this into (18.3) (and the
linearity of g) gives

Xg(Y, Z) = g(∇Y X, Z) + g([X, Y ], Z) + g(Y, ∇X Z). (18.4)

By cyclic permutations one obtains as well

Y g(Z, X) = g(∇Z Y, X) + g([Y, Z], X) + g(Z, ∇Y X), (18.5)

Zg(X, Y ) = g(∇X Z, Y ) + g([Z, X], Y ) + g(X, ∇Z Y ). (18.6)

Taking the linear combination (18.5) + (18.6) - (18.4), we get (Koszul formula):

2g(∇Z Y, X) = −Xg(Y, Z) + Y g(Z, X) + Zg(X, Y )

−g([Z, X], Y ) − g([Y, Z], X) + g([X, Y ], Z).

(18.7)

The right hand side is independent of ∇. Since g is non-degenerate, the uniqueness of ∇ follows from
(18.7).

Definition: the unique connection on (M, g) from the above theorem is called the Riemannian or
Levi-Civita connection.

We determine the Christoffel symbols for the Riemannian connection in a given chart (U, x1 , . . . , xn ).
For this purpose we take X = ∂k , Y = ∂j , Z = ∂i in (18.7) and we use [∂i , ∂j ] = 0 as well as
h∂i , ∂j i = gij . The result is

h∇∂i ∂j , ∂k i = Γlij h∂l , ∂k i,

| {z }
glk

2Γlij glk = −∂k h∂i , ∂j i +∂j h∂i , ∂k i +∂i h∂k , ∂j i,

| {z } | {z } | {z }
gij gik gkj

61
 18 RIEMANNIAN CONNECTIONS

or

1
glk Γlij = (gkj,i + gik,j − gji,k ). (18.8)
2

g ij denoting the inverse matrix of gij , we obtain

1 lk
Γlij = g (gkj,i + gik,j − gji,k ), (18.9)
2
which is exactly equation (9.11).

Properties of the Riemannian connection:

(i) The inner product of any two vectors remains constant upon parallel transporting them along
any curve γ (g(X, Y )γ(t) = g(X, Y )γ(0) ).
i
(ii) The covariant derivative commutes with raising or lowering indices, e.g. Tk;l = (gkm T im );l
= gkm T im ;l , because gkm;l = 0.

Riemann tensor: the curvature tensor of a Riemannian connection has the following additional sym-
metry properties (without proof):

hR(X, Y )Z, U i = − hR(X, Y )U, Zi , (18.10)

hR(X, Y )Z, U i = hR(Z, U )X, Y i . (18.11)

In coordinate expression the Riemann tensor satisfies the following symmetries:

Ri jkl = −Ri jlk is always the case, (18.12)

X
Ri jkl = 0 1st Bianchi identity, (18.13)
(jkl)

X
Ri jkl;m = 0 2nd Bianchi identity. (18.14)
(klm)
P
Eqs. (18.13) and (18.14) are valid for vanishing torsion. Here denotes the cyclic sum. Additionally,
(jkl)

Rijkl = −Rjikl , (18.15)

Rijkl = Rklij , (18.16)

for the Riemannian connection with Rijkl = gis Rs jkl .

Ricci and Einstein tensor

62
 18 RIEMANNIAN CONNECTIONS

Rik = Rj ijk Ricci tensor (18.17)

R = Ri i scalar curvature (18.18)

1
Gik = Rik − Rgik Einstein tensor (18.19)
2

By symmetry, Rik = Rki , Gik = Gki and

1
Rk i;k = R;i , (18.20)
2

Gk i;k = 0, (18.21)

which are the contracted 2nd Bianchi identity.

Proof: Rik = g jl Rlijk = g jl Rjkli , 2nd Bianchi identity gives:

Ri jkl;m + Ri jlm;k + Ri jmk;l = 0.

Then we take the (ik)-trace:

Rjl;m + Ri jlm;i −Rjm;l = 0,

| {z }
−g ik Rjklm;i

Rj l;m − g ik Rj klm;i − Rj m;l = 0,

(jm)-trace:

Rj l;j + g ik Rkl;i −R;l = 0.

| {z }
2Rj l;j

⇒ (18.20)

For (18.21):
1 1
Gk i = Rk i − g k i R = Rk i − δ k i R
2 2
1 1 1
Gk i;k = Rk i;k − (δik R);k = R;i − R;i = 0
2 |2{z } 2
(18.20)

n2 (n2 −1)
Without proof in n dimensions, the Riemann tensor has cn = 12 independent components
(c1 = 0, c2 = 1, c3 = 6, c4 = 20).

63
 19 PHYSICAL LAWS WITH GRAVITATION

Part V

General Relativity
19 Physical laws with gravitation
19.1 Mechanics
The physical laws are relations among tensors (scalars and vectors being tensors of rank 0 and 1
respectively). Thus the physical laws read the same in all coordinate systems (provided the physical
quantities are transformed suitably) and satisfy general covariance (same form). Practically, this means
that from the special relativity laws that hold in absence of gravitation, we have to replace ηαβ by gαβ
and replace derivation by covariant derivation.

In an inertial system, we have the equation of motion (see (4.3))

duα
m = f α. (19.1)
dτ
According to the equivalence principle, (19.1) holds in a local IS. f α does not contain gravitational
forces as they would vanish in a local IS. We transform it to general KS (coordinate system), then the
∂xµ α
Lorentz vector f α gets transformed to f µ = ∂ξ α f (ξ α is in local IS, xµ is in KS). Equation (19.1)
holds in a local IS reads then
Duµ
m = f µ, (19.2)
| dτ
{z }
covariant derivative
given in (16.8)

D dxi
(∇X → dτ ) with ξ µ → uµ (dt → dτ and dτ = ui ) and thus
Duµ duµ
= + Γµνλ uν uλ .
dτ dτ
Then equation (19.2) reads
duµ
= f µ − mΓµνλ uν uλ .
m (19.3)
dτ
We see that on the right hand side there are now gravitational forces appearing explicitly (via Γµνλ ).
Equation (19.3) (or (19.2)) is covariant (it has the same form in all coordinate systems) and reduces for
gµν → ηαβ (thus Γµνλ = 0) to equation (19.1) (in a local IS). The components of uµ are not independent
but satisfy the condition gµν uµ uν = c2 .

19.2 Electrodynamics
According to the equivalence principle, Maxwell’s equations (see (6.4) and (6.5))
4π β
∂α F αβ = j and αβγδ ∂β Fγδ = 0
c
are valid in a local IS. Applying the covariance principle, they become as follows in a general KS:
4π µ
F µν ;ν = j and µνλκ Fλκ;ν = 0, (19.4)
c

64
 19 PHYSICAL LAWS WITH GRAVITATION

provided that going from coordinates ξ α in a local IS to the KS coordinates xµ we have

∂xµ α ∂xµ ∂xν αβ
jα → jµ = j and F µν = F .
∂ξ α ∂ξ α ∂ξ β
Gravity enters via the Γµνλ in the covariant derivative. The continuity equation ∂α j α = 0 translates
to j µ ;µ = 0. It can be shown that in the homogeneous equation the terms with Γ vanish. Thus the
16
covariant derivative reduces to the ordinary derivative (,) .

19.3 Energy-momentum tensor

For an ideal fluid, given by (in a local IS)
 p
T µν = ρ + 2 uµ uν − η µν p, (19.8)
c
with

• uµ : four-velocity,

• ρ: proper energy density,

• p: pressure of the fluid.

In a KS this becomes  p
T µν = ρ + 2 uµ uν − g µν p. (19.9)
c
In the IS the conservation law implies T µν ,ν = 0 and in the KS T µν ;ν = 0 (explicitely, T µν ;ν =
√
∂ g
T µν ,ν + Γµνλ T νλ + Γννλ T µλ = 0). With (19.5), Γννλ = √1
g ∂xλ , we get instead
√
1 ∂ gT µν
T µν ;ν = √ + Γµνλ T νλ = 0. (19.10)
g ∂xν
This is no longer a conservation law, as we cannot form any constant of motion from (19.10). This
should also not be expected, since the system under consideration can exchange energy and momentum
with the gravitational field.
n
∂g
P ∂gki ∂gki ∂gkm m
16 g = det(gik ) = i1 ...in g1i1 . . . gnin . Consider ∂xl
= i1 ...in g1i1 . . . ∂xl
k
. . . gnin and use ∂xl
k
= ∂xl
δ ik =
k=1
∂gkm mr
∂xl
g grik . Due to the antisymmetry of , only the term r = k survives. Thus
∂g ∂gkm mk
= g g.
∂xl ∂xl
Plugging this into the definition of Γkkl (one contraction):
√ √
g km ∂gmk ∂gml ∂gkl g km ∂gmk ∂ ln g 1 ∂ g
 
Γkkl = + − = = = √ . (19.5)
2 ∂xl ∂xk ∂xm 2 ∂x l ∂x l g ∂xl
| {z }
vanish by interchanging
(m↔k)

With (19.5) one can show that the inhomogeneous Maxwell equation in KS can be written as
√
1 ∂( gF µν ) 4π
√ = 2 jµ, (19.6)
g ∂xν c
and the continuity equation: j µ ;µ = 0 becomes
√
∂( gj µ )
= 0. (19.7)
∂xµ

65
 20 EINSTEIN’S FIELD EQUATIONS

20 Einstein’s field equations

The field equations cannot be derived by using the covariance principle, since there is no equivalent
equation in a local IS. We have to make some requirements/assumptions.

Requirements:
• The Newtonian limit is well confirmed through all observations: ∆φ = 4πGρ.

• From the Newtonian limit of the equation of motion for a particle we derived (equation (9.17))
g00 ≈ 1 + 2 cφ2 .

• The non-relativistic limit should then be

8πG
∆g00 = T00 , (20.1)
c4
with T00 ≈ ρc2 (other Tij are small).
8πG
Thus a generalization should lead to something of type Gµν = c4 Tµν where Gµν has to satisfy the
following requirements:
(1) Gµν is a tensor (Tµν is tensor).

(2) Gµν has the “dimension” of a second derivative. If we assume that no new dimensional constant
enter in Gµν then it has to be a linear combination of terms which are either second derivatives of
the metric gµν or quadratic in the first derivative of gµν .

(3) Since Tµν is symmetric, Gµν also has to be symmetric and due to the fact that Tµν is covariantly
conserved, i.e. T µν ;ν = 0, it follows that Gµν must satisfy Gµν = Gνµ and Gµν ;ν = 0.

(4) For a weak stationary field we shall get (20.1), thus G00 ' ∆g00 .
Conditions (1)-(4) determine Gµν uniquely. (1) and (2) imply that Gµν has to be a linear combination

Gµν = aRµν + bRgµν (20.2)

of Rµν , the Ricci tensor, and R, the Ricci scalar17 . The symmetry of Gµν is automatically satisfied.
The contracted Bianchi identity (18.20), (18.21) suggests that Gµν ;ν = 0 on the Einstein tensor, what
implies b = − a2 . Thus we find
1 8πG
Gµν = a(Rµν − gµν R) = 4 Tµν . (20.3)
2 c
The constant a has to be determined by performing the Newtonian limit. Consider weak fields:
gµν = ηµν + hµν , |hµν |  1 (non relativistic velocities: v i  c), then |Tik |  |T00 | ⇒ |Gik |  |G00 |.
Compute the trace of Gµν :



 = a(R − 2R) = −aR from (20.3)
  
g µν Gµν ≈ G00 = a R00 − R g00 = a(R00 − R/2) . (20.4)
 2 |{z}

≈η00


=1
17 It can be shown that indeed the Ricci tensor is the only tensor made of the metric tensor and first and second
deivatives of it, and which is linear in the second derivative.

66
 20 EINSTEIN’S FIELD EQUATIONS

Compairing the two results gives R ≈ −2R00 , thus

 
R
G00 ' a R00 − ' 2aR00 . (20.5)
2

For weak fields all terms quadratic in hµν can be neglected in the Riemann tensor; we get to leading
order:
∂Γρµν ∂Γρρµ
Rµν = Rρ µρν ' − (|hµν |  1).
∂xρ ∂xν
For weak stationary fields we find:

∂Γi00 1 ∂g00
R00 = with Γi00 = .
∂xi 2 ∂xi
∂Γi00 ! 18
Thus G00 ≈ 2a ∂xi = a∆g00 = ∆g00 , therefore a = 1. Einstein’s field equations are (found 1915 by
Albert Einstein):
R 8πG
Rµν − gµν = 4 Tµν (20.6)
2 c
Together with the geodesic equation ((16.12) or (19.3)), these are the fundamental equations of general
relativity. By contraction of (20.6), we find also

R µ 8πG
Rµ µ − δ µ = −R = 4 T. (20.7)
2 |{z} c
=4

R can be expressed in (20.6) in terms of T , and we get:

 
8πG T
Rµν = 4 Tµν − gµν (20.8)
c 2

an equivalent version of the field equations. For the vacuum case where Tµν = 0 we have

Rµν = 0. (20.9)

Significance of the Bianchi identity

Einstein’s equation constitutes a set of non-linear coupled partial differential equations whose general
solution is not known. Usually one makes some assumptions, for instance spherical symmetry. Because
the Ricci tensor is symmetric, the Einstein equations constitute a set of 10 algebraically independent
second order differential equations for gµν .
The Einstein equations are generally covariant, so that they can at best determine the metric
up to coordinate transformation (→ 4 functions). Therefore we expect only 6 independent generally
covariant equations for the metric. Indeed the (contracted) Bianchi identities tell us that (equation
(18.21)) Gν µ;ν = 0 and hence there are 4 differential relations among the Einstein’s equations. Bianchi
identities can also be understood as a consequence of the general covariance of the Einstein equations.
18 Depending
on the convention used for the Riemann tensor, one could also encounter a minus in front of the energy-
momentum tensor, as for example in Weinberg.

67
 21 THE EINSTEIN-HILBERT ACTION

20.1 The cosmological constant

19
As a generalization, one can relax condition (2) and have a linear term in gµν . The field equations
become
R 8πG
gµν − Λgµν = 4 Tµν ,
Rµν − (20.10)
2 c
where Λ is a constant: the cosmological constant ([Λ] = L−2 ). For point (4) the Newtonian limit of
(20.10) leads to
∆φ = 4πρG − c2 Λ. (20.11)
The right-hand side can also be written as 4πG(ρ − ρvacuum ), with
c4
c2 ρvac = Λ. (20.12)
4πG
Λ corresponds to the (constant) energy density of empty space (vacuum). Λ−1/2 (distance) has to be
much larger than the dimension of the solar system.

21 The Einstein-Hilbert action

The field equations (20.6) can be obtained from a covariant variational principle. The action for the
metric g is Z
SD [g] = R(g) dv, (21.1)
D
where D ⊂ M is a compact region space-time, R is a scalar curvature and dv a volume element:
p
dv = |g|d4 x (21.2)

(g = det gik , d4 x in 4 dimensions). The Euler-Lagrange equations are the field equations in vacuum:

δSD [g] = 0.

We have
√ µν √ √
Z Z Z Z
µν 4
δ R(g) dv = δ(g Rµν −g) d x = (δRµν )g −g d x + Rµν δ(g µν −g) d4 x.
4
(21.3)
D D D D

Consider first δRµν :

Rµν = ∂α Γα α ρ α ρ α
µν − ∂ν Γµα + Γµν Γρα − Γνα Γρµ . (21.4)
Let us compute the variation of Rµν at any point p in normal coordinates, whose center is in p itself
(x(p) = 0, then Γα
βγ (0) = 0). Thus δRµν reduces (at any such p) to

δRµν = δΓα α


20
µν ,α − δΓµα ,ν . (21.5)

Without proof one finds that δΓα α

µν is indeed a tensor although Γµν is not a tensor. (21.5) is thus a
tensor equation, it holds in every coordinate system and we can also take the covariant derivative:

δRµν = δΓα α



µν ;α − δΓµα ;ν (21.6)
19 Note that gµν;σ = 0.
20 variation (with respect to g) δ and normal derivative commute

68
 21 THE EINSTEIN-HILBERT ACTION

(Palatini identity). Since gµν;σ = 0 we can write (21.6) as

g µν δRµν = g µν δΓα µν α



µν ;α − g δΓµα ;ν (21.7)

= ω α ;α

= ω α ,α + Γααµ ωµ
|{z}
√
∂ −g
√1 ∂xµ
−g

√
1 ∂( −g ω µ )
=√ . (21.8)
−g ∂xµ
Inserting this into the integral (21.3) and applying (15.21) (Gauss theorem), we get
Z Z
(divg ω)η = iω η,
D ∂D
α
where divg ω = ω ;α and thus
µν √ √
Z Z
(δRµν )g 4
−g d x = ω α −g d0α .
D ∂D

d0α is the coordinate normal to ∂D and

ω α = g µν δΓα
µν − g
µα
δΓνµν (21.9)

is a vector field. If the variations of δg µν vanish outside a region contained in D, then the boundary
term vanishes as well.
√
As for the second term in (21.3) ( Rµν δ(g µν −g) d4 x), we recall that for an n × n matrix A(λ)
R
D
we have (see linear algebra):
d
det A = det A tr A−1 dA


i) dλ dλ ,

−1
ii) d
dλ (A )A = −A−1 dA
dλ .

Thus (δg µν )gνσ = −g µν δgνσ comes from ii) and δg = gg µν δgνµ comes from i) with A−1 = g µν . Hence
we find the desired expressions
√ 1√ 1√
δ −g = −g g µν δgνµ = − −g gαβ δg αβ ,
2 2
√ √ 1√
δ(g µν −g) = −g δg µν − −g g µν gαβ δg αβ . (21.10)
2
And thus
 

√ √
Z Z  1 
0= Rµν δ(g µν −g) d4 x = −g d4 x Rµν δg µν − Rµν g µν gαβ δg αβ 
 
| {z }  2 | {z } | {z }
D D dv R α→µ
β→ν

Z  
1
= dv Rµν − Rgµν δg µν .
2
D | {z }
=Gµν =0

69
 21 THE EINSTEIN-HILBERT ACTION

R√ R√
Therefore δSD [gµν ] = 0 ⇒ Gµν = Rµν − 12 Rgµν = 0. Since δ −g d4 x = −g d4 x 12 g µν δgµν =
R√ D D
− 21 −g d4 x gµν δg µν , it follows that if we have a cosmological constant , the Einstein’s vacuum
D
equations are obtained from the action principle applied on
√
Z
SD [g] = (R − 2Λ) −g d4 x. (21.11)
D

The variational principle extends to matter described by any field ψ = (ψA ) (A = 1, . . . , N ), (we include
also the electromagnetic field among the ψA ) transforming as a tensor under change of coordinates.
Consider an action of the form
√
Z
SD [ψ] = L(ψ, ∇g ψ) −g d4 x, (21.12)
D

where ∇g is the Riemannian connection of the metric g. If we know L in flat space, the equivalence
principle prescribes to replace ηαβ by gαβ and replace ordinary derivatives by covariant ones.

Example: electromagnetic field

1 1
L=− Fµν F µν = − Fµν Fσρ g σµ g ρν ,
16π 16π
and the Euler-Lagrange equations in this case (Fµν = Aν;µ − Aµ;ν = Aν,µ − Aµ,ν ) for the basic
4-potential Aµ field read:

∂L ∂L
− ∇µ = 0, with ∇µ Aν = Aν;µ ;
∂Aν ∂∇µ Aν
∂L ∂L 1
in this case ∂Aν = 0, and ∂∇µ Aν = − 4π F µν . The Euler-Lagrange equations are then F µν ;ν = 0, which
4π µ 1
are the Maxwell equations for vanishing current j µ (F µν ;ν = c j and L = − 16π Fµν F µν − 1c j µ Aµ
with j µ Aµ = g µν jν Aµ ).
Variations in (21.12) with respect to the fields ψA lead to the Euler-Lagrange equations, whereas
variations with respect to the metric (which is also a function and is determined by solving Einstein’s
equations) gives (without proof)
√ √
Z Z
1
4
δg L(ψ, ∇g ψ) −g d x = − T µν δgµν −g d4 x. (21.13)
2
D D

This term has to be added to the one proportional to δgµν in Einstein’s action:

√ c4
Z  
1
−g d4 x Gµν − Tµν δg µν
16πG 2
D | {z }
=0

8πG 1
Fµσ Fνσ − 14 Fσρ F σρ gµν (or T αβ = −Fµα F µβ −


and thus Gµν = c4 Tµν . For electrodynamics: Tµν = 4π
Lg αβ ). And similarly for other “matter” fields.

70
 22 STATIC ISOTROPIC METRIC

22 Static isotropic metric

22.1 Form of the metric
For the gravity field of Earth and Sun we assume a spherically symmetric distribution of the matter
(rotation velocities v i  c). Thus we need a spherically symmetric and static solution for the metric
gµν (x). We first give the general form of such a metric (static and isotropic) which we then use as an
ansatz to solve the field equations. For r → ∞, the Newtonian gravitational potential ϕ = − GM
r goes
to zero. Thus, asymptotically, the metric should be Minkowskian: ds2 = c2 dt2 − dr2 − r2 (dθ2 +
r→∞
sin2 θdφ2 ), in spherical coordinates r, θ, φ and t. Thus,

ds2 = B(r)c2 dt2 − A(r)dr2 − C(r)r2 (dθ2 + sin2 θdφ2 ). (22.1)

Due to isotropy and time independence, A, B and C cannot depend on θ, φ and t (and no linear terms
in dθ and dφ). Freedom in the choice of coordinates allows to introduce a new radial coordinate in
(22.1): C(r)r2 → r2 , thus C(r) can be absorbed into r. We get the standard form:

ds2 = B(r)c2 dt2 − A(r)dr2 − r2 (dθ2 + sin2 θdφ2 ) (22.2)

(θ and φ have the same significance as in Minkowski coordinates). Due to our asymptotic requirements
(r → ∞) we can assume that B(r) → 1 and A(r) → 1.

22.2 Robertson expansion

Even without knowing the solution to the field equations, we can give an expansion of the metric
for weak fields outside the mass distribution. The metric can only depend on the total mass of the
considered object (Earth or Sun for instance), on the distance from it and on the constants G, c. Since
GM
A and B are dimensionless, they can only depend on a combination of the dimensionless quantity c2 r .
GM
For c2 r  1 we can then have the following expansion:
 2
GM GM
B(r) = 1 − 2 + 2(β − γ) + ...
c2 r c2 r
(22.3)
GM
A(r) = 1 + 2γ 2 + . . .
c r
which is the Robertson expansion. The linear term in B(r) has no free parameter since it is constrained
by the Newtonian limit: g00 ' 1 + 2 cφ2 , φ = − GM
r (Newtonian potential), therefore B → g00 . The
coefficient 2(β − γ) comes from historical reasons, β and γ are independent coefficients. In the solar
system, GM
c2 r ≤ GM
c 2 R ' 2 × 10−6 , then only linear terms in γ and β play a role. For general relativity:
γ = β = 1 (Newtonian gravity: γ = β = 0).

71
 22 STATIC ISOTROPIC METRIC

22.3 Christoffel symbols and Ricci tensor for the standard form
The metric tensor gµν is diagonal.

g00 = B(r) g11 = −A(r) g22 = −r2 g33 = −r2 sin2 θ (22.4)

1 1 1 1
g 00 = g 11 = − g 22 = − g 33 = − (22.5)
B(r) A(r) r2 r2 sin2 θ

g σν
 
∂gµν ∂gλν ∂gµλ
The non-vanishing components of Γσλµ = + − are
2 ∂xλ ∂xµ ∂xν

B0 B0 A0
Γ001 = Γ010 = Γ100 = Γ111 =
2B 2A 2A
1 r r sin2 θ
Γ212 = Γ221 = Γ122 = − Γ133 = − (22.6)
r A A
1
Γ313 = Γ331 = Γ323 = Γ332 = cot θ Γ233 = − sin θ cos θ
r
where 0 stands for ∂
∂r . With
− g = r4 AB sin2 θ (22.7)

we get √  
A0 B0
 
∂ ln −g 2
Γρµρ


= = 0, + + , cot θ, 0 . (22.8)
∂xµ r 2A 2B
The Ricci tensor can then be calculated as
∂Γρµν ∂Γρρµ
Rµν = − + Γσµν Γρρσ − Γσρµ Γρνσ , (22.9)
∂xρ ∂xν
and we get as a result

B 00 A0 B 0 B 02 B0 2 A0 B0
 
R00 = − − + + + ,
2A 2A2 2AB 2A r 2A 2B

B 00 B 0 A0 B0 B0
 
= − + + , (22.10)
2A 4A A B rA

B 00 B 0 A0 B0 A0
 
R11 =− + + + , (22.11)
2B 4B A B rA
 0
B0

r A 1
R22 =1+ − − , (22.12)
2A A B A

R33 = R22 sin2 θ, (22.13)

The non-diagonal components Rµν with µ 6= ν vanish.

72
 22 STATIC ISOTROPIC METRIC

22.4 Schwarzschild metric

We assume a static, spherically symmetric, mass distribution with finite extension:

6= 0 r ≤ r
0
ρ(r) (22.14)
= 0 r > r
0

Similarly, the pressure P (r) is thought to vanish for r > r0 . The four velocity vector within the mass
distribution in the static case is uµ = (u0 = constant, 0, 0, 0). This way, the energy-momentum tensor
(describing matter) does not depend on time. We then adopt the ansatz for the metric elaborated
in (22.2): gµν = diag(B(r), −A(r), −r2 , −r2 sin2 θ). Outside the mass distribution (r ≥ r0 ), the Ricci
tensor vanishes: Rµν = 0. We have already calculated the coefficients Rµν in equations (22.10) –
(22.13). For µ 6= ν, Rµν = 0 is trivially satisfied while the diagonal components should be set to zero:
R00 = R11 = R22 = R33 = 0 (r ≥ r0 ).
 0
A0

R00 R11 1 B d
Consider + =− + = 0 and thus (ln AB) = 0 (since rA 6= 0) or AB = constant
B A rA B A dr
1
(or ln AB = constant). For r → ∞ we require A = B = 1, therefore AB = 1 ⇒ A(r) = B(r) .
Introducing this into R22 (22.12) and R11 (22.11) leads to

R22 = 1 − rB 0 − B = 0, (22.15)

B 00 B0 rB 00 + 2B 0 1 dR22
R11 = − − =− = = 0. (22.16)
2B rB 2rB 2rB dr
With (22.15), (22.16) is automatically satisified (since R22 = 0 also its derivative vanishes). We write
(22.15) as
d
(rB) = 1. (22.17)
dr
| {z } = r − 2a. Then
We integrate it and get rB = r + constant
−2a

2a
B(r) = 1 − ,
r
(22.18)
1
A(r) = ,
1 − 2a
r

for r ≥ r0 . This solution for the vacuum Einstein’ equations was found in 1916 by Schwarzschild. The
Schwarzschild solution is
dr2
 
2 2a 2 2
ds = 1 − c dt − − r2 (dθ2 + sin2 θdϕ2 ) (22.19)
r 1 − 2a
r

The constant can be determined by considering the Newtonian limit:

φ GM 2a
g00 = B(r) −−−→ 1 + 2 2
=1−2 2 =1− .
r→∞ c c r r
Thus one introduces the so called Schwarzschild radius:
2GM
rS = 2a =
c2

73
 23 GENERAL EQUATIONS OF MOTION

2GM
The Schwarzschild radius of the Sun is rs, = c2 ' 3 km (M ' 2 × 1030 kg, R = 7 × 105 km) so
rS, 2GM
√
R = c 2 R ' 4 × 10−6 21 dt
. A clock at rest in r has the proper time dτ = B dt, thus dτ diverges
at r → rS . This implies that a photon emitted at r = rS will be infinitely redshifted (t is not a good
coordinate either for events taking place at r ≤ rS ). A star, whose radius rstar is smaller than rS , is
a black hole since photons emitted at its surface cannot reach regions with r > rS .
rS
Expanding the Schwarzschild metric in power of r and comparing it with the Robertson expansion
(22.3), one finds β = γ = 1 for general relativity.

23 General equations of motion

We now consider the motion of a freely falling material particle or photon in a static isotropic gravi-
tational field (e.g. motion of planets around the Sun). For the relativistic orbit xk (λ) of a particle in
a gravitational field we have:
d2 xk µ
k dx dx
ν
= −Γ µν (23.1)
dλ2 dλ dλ
and 
 2  2  2
µ
dx dx ν
ds dτ c m 6= 0, λ=τ
gµν = = c2 = . (23.2)
dλ dλ dλ dλ 0 m=0
For a massive particle we can take the proper time as a parameter for the trajectory or orbit (dλ = dτ ).
For massless particles one has to choose another parameter. For the spherically symmetric gravitational
field, we use the metric (r > r , radius of the star)

ds2 = B(r)c2 dt2 − dr2 A(r) − r2 (dθ2 + sin2 θdϕ2 ), (23.3)

with the coordinates (x0 , x1 , x2 , x3 ) = (ct, r, θ, φ). Equations (23.1) – (23.3) define the relativistic
Kepler problem. Using the Christoffel symbols given in (22.6), we get for (23.1):

d2 x0 B 0 dx0 dr
2
=− , (23.4)
dλ B dλ dλ
2  2  2  2
d2 r B 0 dx0 A0 dr r sin2 θ dφ

r dθ
=− − + + , (23.5)
dλ2 2A dλ 2A dλ A dλ A dλ
2
d2 θ

2 dθ dr dφ
2
=− + sin θ cos θ , (23.6)
dλ r dλ dλ dλ

d2 φ 2 dφ dr dθ dφ
2
=− − 2 cot θ . (23.7)
dλ r dλ dλ dλ dλ
Equation (23.6) can be solved by
π
θ=
= constant. (23.8)
2
π
Without loss of generality we can choose the coordinate system such that θ = 2, this way the trajectory
2
π d θ
lies on the plane with θ = 2 . dλ2 = 0 corresponds to angular momentum conservation. With (23.8)
21 Apparently it seems that the Schwarzschild metric is singular for r = r , but this is not the case. It is only an
s
artefact of the coordinate choice. To be discussed later.

74
 23 GENERAL EQUATIONS OF MOTION

we get for (23.7) :  

1 d 2 dφ
r = 0, (23.9)
r2 dλ dλ
which leads to
dφ
r2
= constant = l. (23.10)
dλ
l can be interpreted as the (orbital) angular momentum (per unit mass). Equations (23.8) and (23.10)
follow from angular momentum conservation, which is a consequence of spherical symmetry (rotation
invariance).
Equation (23.4) can be written as (B = B(r(λ))
  0 
d dx
ln + ln B = 0, (23.11)
dλ dλ
h 0  i
which can be integrated as ln dxdλ B = constant or

dx0
B = constant = F. (23.12)
dλ
In (23.5) we use (23.8), (23.10) and (23.12) and get:
2
d2 r F 2B0 A0 l2

dr
2
+ 2
+ − = 0. (23.13)
dλ 2AB 2A dλ Ar3
dr


We multiply it with 2A dλ and get
"   #
2
d dr l2 F2 22
A + 2− = 0. (23.14)
dλ dλ r B

Integration gives
2
l2 F2

dr
A + − = − = constant. (23.15)
dλ r2 B
Integrating it once more we get r = r(λ). Inserting then this result into (23.10) and (23.12), we obtain
with one more integration φ = φ(λ) and t = t(λ). Next we eliminate λ and get r = r(t) and φ = φ(t).
π
Together with θ = 2, this is then a complete solution (generally it has to be done numerically).
Equation (23.2) becomes
2 2 2 2
dxµ dxν dx0
   
dr 2 dθ 2 2 dφ
gµν =B −A −r − r sin θ = , (23.16)
dλ dλ dλ dλ dλ dλ

using (23.8), (23.10), (23.12) and (23.15). On the other hand


c2 (m 6= 0)
= .
0 (m = 0)

We are left with two integration constants, F and l.

dA dr
22 Notice: = A0
dλ dλ

75
 23 GENERAL EQUATIONS OF MOTION

23.1 Trajectory
From (23.15) we get s
F2 2
dr B − rl 2 − 
= , (23.17)
dλ A
and with (23.10), s
dφ dφ dλ l A
= = 2 F2 2 . (23.18)
dr dλ dr r B − rl 2 − 
Thus, s
Z
dr A(r)
φ(r) = F2
. (23.19)
r2 B(r)l2 − r12 − 
l2

With this we can find the trajectory φ = φ(r) in the orbital plane. (Massive particles: 2 integration
F2  F2
constants l2 and l2 , massless particles: only l2 ).

Trajectory in Schwarzschild metric:

rS
Insert Schwarzschild metric: B(r) = A−1 (r) = 1 − r =1− 2a
r and write:
dt dr dφ
ṫ = , ṙ = , φ̇ = .
dλ dλ dλ
Then with (23.8), (23.10), (23.12) and (23.15) we get
 
π 2a
θ = , cṫ 1 − = F, r2 φ̇ = l. (23.20)
2 r
Multiplying (23.15) with B and using AB = 1, we have
ṙ2 a l2 al2 F2 − 
− + 2− 3 = = constant. (23.21)
2 r 2r r 2
The radial component can be written as
ṙ2
+ Veff (r) = constant, (23.22)
2
2GM
with the effective potential (2a = c2 ,  = {c2 , 0})

− GM + l2 − GM l2
(m 6= 0)
r 2r 2 c2 r 3
Veff (r) = . (23.23)
 l2 − GM l2 (m = 0)
2r 2 c2 r 3

A formal solution r = r(λ) of (23.22) is given through the following integral :

Z
dr
λ=± p . (23.24)
2(constant − Vef f (r))
1
Due to the r3 term (relativistic), this is an elliptical integral which has to be solved numerically.
For small values of r, centrifugal potential term dominates (as long as l is not too small), then for
23 24
even smaller values of r the attractive relativistic term takes over:
23 l l2
∼r×v → r2
∼ v2
2
24 v ∼ 10−4 → vc2 ∼ 10−8
c

76
 23 GENERAL EQUATIONS OF MOTION

Veff collision

circular unstable

scattering

ellipse with precession

circular stable

Figure 11: Effective potential for massive particles in Schwarzschild metric

GM l2 GM v 2
− 2 2
'− (23.25)
r c r r c2
1 dr dr
Eq. (23.22) differs from the non-relativistic case by an additional r3 term and ṙ = dτ differs from dt
2
v
by terms of order c2 .

Observations:

• Where Veff has a minimum there are bounded solutions, however due to the relativistic effects
there will be small deviations from the elliptical orbits (precession of the perihelion). As a special
case, with ṙ = 0, the circular orbit is a possible solution (in which case the constant in (23.22) is
equal to the value of Veff at its minimum).

• The solution at the maximum of Veff is an unstable circular orbit.

• If the constant is positive one gets non-bounded trajectories (corresponding to hyperbolic solu-
tions in the non-relativistic case).

77
 23 GENERAL EQUATIONS OF MOTION

• If the constant is larger than the maximum value of the potential, the particle falls into the
center.
dVef f
• At minimum and maximum we have dr = 0. For m 6= 0 we get

c2 2 r
2
r − 2 + 3 = 0. (23.26)
l rS
3c2 1
In order to have two real solutions, we need l2 < 2 .
rS
That means
√
l ≥ lcrit = 3 rS c. (23.27)

For l → lcrit , the angular momentum barrier gets smaller and smaller until the maximum and
minimum fall together for l = lcrit . For l < lcrit , the potential decreases monotonically for r → 0.

Veff collision

circular unstable

scattering

Figure 12: Effective potential for massless particles in Schwarzschild metric

78
 23 GENERAL EQUATIONS OF MOTION

Here both terms are proportional to l2 , thus the shape of Veff does not depend on l. At rmax = 32 rS
the potential has a maximum. At rmax the photons can move on a circular orbit, which is unstable. If
the constant in (23.22) is smaller than Veff (rmax ) then the incoming photon will be scattered, whereas
25
if the constant is bigger the photon will be absorbed at the center.

25 for r ≤ rS the Schwarzschild solution is not applicable

79
 24 LIGHT DEFLECTION

Part VI

Applications of General Relativity

24 Light deflection
The trajectory r = r(φ) of a photon in the gravitational field is given by (23.19) ( = 0):
Zr s
dr̃ A(r̃)
φ(r) = φ(r0 ) + F2
. (24.1)
r̃2 −
B(r̃)l2
1
r̃ 2
r0

∆φ

r0
φ r

Sun

light ray

Figure 13: Deflection of light by the Sun

We will now show that light is deflected by a massive body, carrying through calculations for the Sun.
In fig. 13, the following quantities are defined: light is deflected by ∆φ and r0 is the minimal distance
(or impact parameter) from the Sun. For simplification we assume also r0  rS .
As starting point of the integration we choose the minimum distance r0 , where we set φ(r0 ) = 0.
Going from r0 till r∞ the angle changes by φ(∞). Along the drawn trajectory the radial vector turns
by 2φ(∞). If the trajectory would be a straight line, then 2φ(∞) = π.
Thus ∆φ = π − π = 0 for a straight line and in general (φ(r0 ) = 0):

∆φ = 2φ(∞) − π. (24.2)

At r0 , r(φ) is a minimum, thus  

dr 
= 0. (24.3)
dφ r0

80
 24 LIGHT DEFLECTION

light ray

π π
φ(r) → φ(∞) = 2 φ(r) → φ(∞) = 2
Sun

Figure 14: Non-deflected ray of light

From (24.3) we get with (23.17) and (23.18) the condition

F2 B(r0 )
= . (24.4)
l2 r02
This way we can eliminate the constants F and l in terms of r0 with (24.1):
Z∞ v
dr u A(r)
u
φ(∞) = t B(r ) 2 . (24.5)
r 0 r
2 − 1
r0 B(r) r0

Let us compute the integral by inserting the Robertson expansion A(r) = 1 + γ 2a 2a

r , B(r) = 1 − r (with
rs GM a
a= 2 = c2 ). We keep terms up to withr

B(r0 ) r2 r2
  
1 1
− 1 ' 2 1 + 2a − −1
B(r) r02 r0 r r0
 2  
r 2ar
= 2 −1 1− .
r0 r0 (r + r0 )

√
We get using 1 + x = 1 + x2 ,

Z∞  
dr r0 a ar
φ(∞) ' p 1+γ +
r2 − r02 r r r0 (r + r0 )
r0
" p #∞
r2 − r02
r
r 
0 a a r − r0
= arccos +γ +
r r0 r r0 r + r0
r0

π a a
= +γ + . (24.6)
2 r0 r0
With (24.2) we get    
4a 1+γ 2rS 1+γ
∆φ = = . (24.7)
r0 2 r0 2

81
 24 LIGHT DEFLECTION

2GM
For general relativity, γ = 1, rS = c2 and thus

2rS
∆φ =
r0

For a light ray which just grazes the surface of the Sun (r0 = R = 7×105 km) we get (π = 180×360000 ):
 
00 1+γ
∆φ = 1.75 . 26 (24.8)
2

On May 29, 1919, an eclipse allowed experimental confirm of this result.

Figure 15: Gravitational lensing in the Abel 2218 galaxy cluster

26 “cheating” with Newton’s theory gives half this result that is 0.8400

82
 25 PERIHELION PRECESSION

25 Perihelion precession
Consider the elliptical orbit of a planet around the Sun:
y

Mercury
r

p
φ

rmax rmin
Sun x

Figure 16: Non-relativistic elliptical orbit of Mercury around the Sun

We will use the following notations: minimum distance r− = rmin , maximum distance r+ = rmax ,
φ± = φ(r± ), A± = A(r± ), B± = B(r± ). The relativistic orbit follows from equation (23.19) for
r = r(φ) with  = c2 . The integral gives for the change in angle between r− and r+ :

Zr+ s Zr+ s
dr A(r) dr A(r)
φ+ − φ− = F2 c2
= . (25.1)
r2 1
B(r)l2 − r 2 − l2
r2 K(r)
r− r−

For a full orbit the angle is 2π, i.e. twice the integral (25.1). The shift of the perihelion (per complete
orbit) is given by
∆φ = 2(φ+ − φ− ) − 2π. (25.2)
dφ dr
p
The integrand in (25.1) is equal to dr . For r = r± due to dφ = 0, K(r) r2 has to vanish, thus
K(r± ) = 0:
F2 1 c2
2
= 2 + 2. (25.3)
B± l r± l

83
 25 PERIHELION PRECESSION

This way we can express F and l through r± :

1 1
F2 2
r+
− 2
r−
2
r− 2
− r+
= 1 1 =  , (25.4)
l2 − 2 r2 1 1
B+ B− r+ − B+ − B−

B+ B− r2 r2
2 − r2
−
B+ − B−
+
c2 r+ −
= − = −  . (25.5)
l2 B+ − B− 2 r2
r+ 1
− 1
− B+ B−

This leads us to an expression for K(r):

   
2 1 1 2 1 1
r− B(r) − B− − r+ B(r) − B+ 1
K(r) =   − . (25.6)
2 r2 1 1 r2
r+ − B+ − B−

For A and B insert the Robertson expansion

2a
A(r) = 1 + γ + ..., (25.7)
r
2a  a 2
B(r) = 1 − + 2(β − γ) + ..., (25.8)
r r
1 2a  a 2
27
=1+ + 2(2 − β + γ) + .... (25.9)
B(r) r r
dφ
With eq. (25.9), K(r) becomes a quadratic form in 1r . Since dr = ∞ for r = r± , K+ = K− = 0. This
determines K(r) up to a constant c̃:
  
1 1 1 1
K(r) = c̃ − − . (25.10)
r− r r r+

c̃ can be obtained by comparing with (25.6) for r → ∞. With (25.9) one gets
 
a a
c̃ = 1 − (2 − β + γ) + . (25.11)
r+ r−

We get thus the following integral:

Zr+   − 21
1 dr  a 1 1 1 1
φ+ − φ− = √ 1+γ − − . (25.12)
c̃ r2 | {z r } r− r r r+
r− √
from A ≈ 1+γ a
r

We perform the following substitution:

   
1 1 1 1 1 1 1
= + + − sin ψ; (25.13)
r 2 r+ r− 2 r+ r−
27 v 2≈ ar ; terms g00 u0 u0 ≈ Bc2 and g11 u1 u1 ≈ Av 2 ≈ Ac2 ar show up both and have thus to be expanded to the
c2
same order in ar . Therefore B has to be expanded one order in ar more than A.

84
 25 PERIHELION PRECESSION

π
r+ and r− correspond to ψ = 2 and ψ = − π2 , respectively. With
   
1 1 1 1 1
d = − 2 dr = − cos ψ dψ, (25.14)
r r 2 r+ r−
 
1 1 1 1 1
− = − (1 + sin ψ), (25.15)
r− r 2 r− r+
 
1 1 1 1 1
− = − (1 − sin ψ), (25.16)
r r+ 2 r− r+

we get for the integral

π
Z2      
1 a 1 1 a 1 1
φ+ − φ− = √ dψ 1 + γ + +γ − sin ψ . (25.17)
c̃ 2 r− r+ 2 r+ r−
−π
2

Now introduce the parameter p of the ellipse (see figure 16):

2 1 1
= + . (25.18)
p r+ r−

Integration of eq. (25.17) leads to

      
π a a a a
φ+ − φ− = √ 1 + γ = π 1 + (2 − β + γ) 1+γ = π 1 + (2 − β + 2γ) . (25.19)
c̃ p p p p

Precession per orbit for the perihelion is:

 
6πa 2 − β + 2γ
∆φ = 2(φ+ − φ− ) − 2π = . (25.20)
p 3
2−β+2γ
In general relativity γ = β = 1 and so 3 = 1. Thus,

6πa
∆φ =
p

Consider Mercury: p = 55 × 106 km, 2a ≈ 3 km, π = 180◦ × 360000 which give ∆φ = 6πa
p =
00
0.104 (per full orbit). In 100 years Mercury fulfills 415 orbits around the Sun, this way we get
∆φ = 43 (per century). For more distant planets (Venus, Earth,. . . ) ∆φ is at most ∼ 500 per century.
00

Already in 1882, Newcomb found a perihelion precession of 4300 per century for Mercury. Full perihelion
precession amounts to 57500 per century of which 53200 are due to the influence of other planets (this
within Newtonian theory). One finds

2 − β + 2γ
= 1.003 ± 0.005, (25.21)
3
in good agreement with general relativity. So far, the parameters of the Robertson expansion are
constrained to |γ − 1| < 3 × 10−4 and |β − 1| < 3 × 10−3 . More recently (radar echoes delay from
Cassini spacecraft): |γ − 1| = (2.1 ± 2.3) × 10−5 .28
28 B. Bertotti et al. Nature 425, 374 (2003)

85
 25 PERIHELION PRECESSION

Sun

Mercury

Figure 17: Illustration of the perihelion precession of Mercury (effect strongly exagerated)

25.1 Quadrupole moment of the Sun

A quadrupole moment of the Sun could also influence a perihelion precession of Mercury, that is why
one has to study it. The mass quadrupole moment of the Sun (due to its rotation) is

2 2 Rk − R⊥
Q = J2 M R  with J2 = , (25.22)
5 R
and

• Rk : orthogonal (to R⊥ ) radius,

• R⊥ : radius orthogonal to the plane containing the planet orbits and parallel to the rotation axis
of the Sun.

The induced gravitational potential in the planet’s orbital plane (which is also the equatorial plane of
the Sun) is
GM GQ
φ(r) = − − 3. (25.23)
r 2r
The additional term has the same r dependence than the additional relativistic term:
GM l2 GM l2
Veff = − + 2− 2 3 . (25.24)
r 2r c r

86
 26 LIE DERIVATIVE OF THE METRIC AND KILLING VECTORS

GM
With l ∼ pv and v 2 ∼ p we can compare the two terms (their relative strength)

2 2 2
GQ J2 R J2 R J2 R
2 ∼ 2 ∼ ∼ .
GM cl 2 p2 vc2 p GM
c2
pa

We see that the full expression for the perihelion precession is given by
2 
J2 R

6πa 2 − β + 2γ
∆φ = + . (25.25)
p 3 2ap
2
J2 R 
From observations one finds J2 ∼ (1 − 1.7) × 10−7 ; thus the additional term is 2ap ≈ 5 × 10−4 ,
accounting for at most 1/10 of the error given in (25.21) and is thus negligible.

26 Lie derivative of the metric and Killing vectors

Consider the Lie derivative of the metric tensor gµν in the direction of the vector K. According to
equation (14.2) we get:
LK gµν = gµν,κ K κ + gµκ K κ ,ν + gκν K κ ,µ . (26.1)

To rewrite this expression we observe the identities

Kσ = gσµ K µ ,

∂K κ ∂(K κ gµκ ) ∂gµκ ∂Kµ ∂gµκ

K κ ,ν gµκ = ν
gµκ = ν
− Kκ ν
= ν
− Kκ .
∂x ∂x ∂x ∂x ∂xν
Hence eq. (26.1) can also be written as
 
∂Kσ ∂Kρ µ ∂gρσ ∂gµσ ∂gρµ
LK gσρ = + + K − −
∂xρ ∂xσ ∂xµ ∂xρ ∂xσ

∂Kσ ∂Kρ
= + − 2Kµ Γµρσ (26.2)
∂xρ ∂xσ

= Kσ;ρ + Kρ;σ .

An infinitesimal coordinate transformation is a symmetry of the metric if LK gµν = 0, thus if

Kσ;ρ + Kρ;σ = 0 (26.3)

29
Any 4-vector Kσ (x) satisfying this equation will be said to form a Killing vector.

Example: Consider a stationary gravitational field, for which there exists coordinates {xµ } such that the
components of gµν do not depend on ct = x0 (for instance Schwarzschild metric). Let K µ = δ µ 0 with
the corresponding vector field δ µ 0 ∂µ (→ ∂0 ). Inserting K µ into (26.1) one gets LK gµν = gµν,0 + 0 + 0 =
∂
∂x0 gµν = 0 (since gµν does not depend on x0 ). K is a Killing vector or Killing field or an infinitesimal
isometry.
29 Named after 19th century mathematician Wilhelm Killing

87
 27 MAXIMALLY SYMMETRIC SPACES

Notice that, due to the properties of the Lie derivative, if K1 and K2 are Killing vectors, LK1 gµν = 0,
LK2 gµν = 0 then [K1 , K2 ] is also a Killing vector since

[LK1 , LK2 ]gµν = L[K1 ,K2 ] gµν = 0. (26.4)

We are used to the fact that symmetries lead to conserved quantities: in classical mechanics the angular
momentum of a particle moving in a rotationally symmetric field is conserved. In the present context,
the concept of “symmetries of a gravitational field” is replaced by “symmetries of the metric” and we
therefore expect conserved quantities to be associated with the presence of Killing vectors.
Let K µ be a Killing vector and xµ (τ ) be a geodesic. Then the quantity Kµ ẋµ is constant along the
geodesic. Indeed,
D
(Kµ ẋµ ) = (∇ν Kµ ẋν )ẋµ + Kµ (∇ν ẋµ ) ẋν
dτ | {z }
=0 geodesic

1
= (∇ν Kµ + ∇µ Kν ) ẋµ ẋν = 0. (26.5)
2| {z }
=0 (26.3)

T µν is the covariantly conserved symmetric energy-momentum tensor with ∇µ T µν = 0. Then J µ =

T µν Kν is a covariantly conserved current:
1
∇µ J µ = (∇µ T µν ) Kν + T µν ∇µ Kν = T µν (∇µ Kν + ∇ν Kµ ) = 0,
| {z } 2 | {z }
=0 =0 (26.3)

to which we can associate a conserved charge.

27 Maximally symmetric spaces

Maximally symmetric spaces are spaces that admit the maximal number of Killing vectors (which below
will turn out to be n(n+1)
2 for an n-dimensional space). In the context of the cosmological principle
such spaces, which are simultaneously homogeneous (“the same at every point”) and isotropic (“the
same in every direction”), provide a description of space in a cosmological space-time.
From equation (17.2) we had (from definition of Riemann tensor and covariant derivative)

([∇X , ∇Y ] − ∇[X,Y ] )V λ = Rλ σµν X µ Y ν V σ , (27.1)

along with X = X µ ∂µ , Y = Y ν ∂ν and ∇Y ν ∂ν V λ = Y ν ∇∂ν V λ = Y ν ∇ν V λ , ∇X ∇Y V λ = X µ ∇µ (Y ν ∇ν V λ ) =

X µ (∇µ Y ν )∇ν V λ + X µ Y ν ∇µ ∇ν V λ etc, we get

[∇µ , ∇ν ]V λ = Rλ σµν V σ . (27.2)

Taking into account the first Bianchi identity, it is possible to find that for a Killing vector Kµ , one
has (no proof here)
ρ
∇λ ∇µ Kν (x) = Rλµν Kρ (x) (27.3)

88
 27 MAXIMALLY SYMMETRIC SPACES

for x = x0 . Thus a Killing vector K µ (x) is completely determined everywhere by the values of K µ (x0 )
(i)
and ∇µ Kν (x0 ) at a single point x0 (think of Taylor expansion). A set of Killing vectors {Kµ (x)} is
said to be independent if any linear relation of the form
X
ci Kµ(i) (x) = 0, (27.4)
i

with constant coefficients ci ; implying ci = 0. Since in an n-dimensional space-time there can be

(i) n(n−1)
at most n linearly independent vectors Kµ (x0 ) at a point, and at most 2 independent anti-
symmetric matrices (∇µ Kν (x0 )), we reach the conclusion that an n-dimensional space-time can have
at most
n(n − 1) n(n + 1)
n+ = (27.5)
2 2
independent Killing vectors.

• Homogeneous space is meaning that the n-dimensional space(-time) admits n-translational Killing
vectors.

• Isotropic space: ∇µ Kν (x0 ) is an arbitrary anti-symmetric matrix (→ rotation). We can choose

n(n−1)
a set of 2 Killing vectors.

• We define a maximally symmetric space to be a space with a metric with a maximal number of
n(n+1)
2 Killing vectors.

The Riemann curvature tensor of a maximally symmetric space becomes simpler. One can show (no
proof) that it becomes
Rijkl = k(gik gjl − gil gjk ) (27.6)

for some constant k. The Ricci tensor then becomes

Rij (x) = (n − 1)kgij . (27.7)

The Ricci scalar can be obtained to be

R(x) = n(n − 1)k, (27.8)

and the Einstein tensor

1  n
Gik = Rik − Rgik = k (n − 1) 1 − gik .
2 2
The Bianchi identity implies that k is a constant in order for Gik ;k to vanish. We shall deal with
space-times in which the metric is spherically symmetric and homogenous on each “plane” of constant
time. In our case n = 4 and the maximally symmetric subspace has 3 dimensions. Consider first the
metric on the 3-dimensional subspace

dσ 2 = A(r)dr2 + r2 dΩ2
|{z}
dθ 2 +sin2 θdφ2

For the Christoffel symbols, we use the ones for the general form of a static isotropic metric (22.6) with
B(r) = 0. Since the Christoffel symbols are invariant under an overall sign change of the metric, also

89
 27 MAXIMALLY SYMMETRIC SPACES

the Ricci tensor is and thus one can apply (22.10)-(22.13) with B(r) = 0 for this three-dimensional
space, without caring for the sign in front of A. Hence we get for equation (27.7)

A0
Rrr = R11 = ,
rA
(27.9)
1 rA0
Rθθ = R22 =− +1+ .
A 2A2
From eq. (27.7), we have Rrr = 2kA, and Rθθ = 2kgθθ = 2kr2 . Thus from equating the two first
equation leads us to
A0
2kA = ⇒ A0 = 2krA2 , (27.10)
rA
while we get for the second one

1 rA0 1 2kr2 A2 1 1
2kr2 = − +1+ 2
=− +1+ = − + 1 + kr2 ⇒ kr2 = − + 1,
A 2A A 2A2 A A
which leads to
1
A= , (27.11)
1 − kr2
and solves also (27.10). Then the metric on the 3-dimensional subspace (maximally symmetric) is

dr2
dσ 2 = + r2 dΩ2 . (27.12)
1 − kr2
It can be shown that k can have the following values: 0, ±1.



 +1 sphere, positive curvature

k = −1 hyperbola, negative curvature



0 plane, zero curvature

The full metric (with time coordinate) has then the form:

dr2
 
ds2 = c2 dt2 − a2 (t) + r 2
(dθ 2
+ sin 2
θ dφ 2
) (27.13)
1 − kr2

where a(t) is the cosmic scale factor, which has to be determined by solving the Einstein’s equations via
the matter content of the universe. This metric (first discovered by Friedmann, Lemaître, Roberston
and Walker) is a reasonable ansatz for describing the universe. There is good evidence that the universe
(on large scales) is surprisingly homogeneous and isotropic (from redshift surveys of galaxies and cosmic
microwave background radiation).

90
 28 FRIEDMANN EQUATIONS

28 Friedmann equations
We write the metric (27.13) as follows:

ds2 = c2 dt2 − a2 (t)g̃ij dxi dxj , (28.1)

where tildes denote 3-dimensional quantities calculated with the metric g̃ij . The Christoffel symbols
are given by (notice Γµ00 = 0):
ȧ i
Γijk = Γ̃ijk , Γij0 = δ j, Γ0ij = ȧ a g̃ij , (28.2)
a
where dot denotes derivation with respect to t. The relevant components of the Riemann tensor are:
ä
Ri 0j0 = − δ i j , R0 i0j = a ä g̃ij , Rk ikj = R̃ij + 2ȧ2 g̃ij (28.3)
a
We can make use of R̃ij = 2kg̃ij (maximal symmetry of the 3-dimensional subspace) to compute Rµν .
The non-zero components are then
ȧ2
 
ä ä k
R00 = −3 , Rij = (aä + 2ȧ2 + 2k)g̃ij = − +2 2 +2 2 gij , (28.4)
a a a a
where gij = −a2 g̃ij . The Ricci scalar becomes R = − a62 (aä + ȧ2 + k) and the non-zero components of
the Einstein tensor are
ȧ2 ä ȧ2
   
k k
G00 = 3 2
+ 2 , G0i = 0, Gij = 2 + 2+ 2 gij . (28.5)
a a a a a
Next we have to specify the matter content. We treat here the universe as non-interacting particles or
a perfect fluid. A perfect fluid has energy-momentum tensor (19.8)
p 
T µν = 2 + ρ uµ uν − g µν p, (28.6)
c
where p is the pressure, ρ the energy density and uµ the velocity field of the fluid (uµ = (c, 0, 0, 0) in
a comoving coordinate system). The trace of the energy-momentum tensor is then

T µ µ = ρc2 − 3p. (28.7)

The equation of state is p = p(ρ) and in particular one assumes

p = wρ, (28.8)

where w is the equation of state parameter.

Examples:

• For non-interacting particles we have p = 0, w = 0. Such matter is referred to as dust. The

energy-momentum tensor is T µν = ρ uµ uν .

• For radiation the energy-momentum tensor is (like in Maxwell’s theory) traceless and hence
radiation has the equation of state
1
p= ρ, (28.9)
3
thus w = 31 .

91
 28 FRIEDMANN EQUATIONS

• As we will see, a cosmological constant Λ corresponds to a “matter” contribution with w = −1.

The conservation law T νµ ;µ = 0 implies T 0µ ;µ = 0 or ∂µ T µ0 + Γµµν T ν0 + Γ0µν T µν = 0. For a perfect

fluid: ∂t ρ(t) + Γµµ0 ρ + Γ000 ρ + Γ0ij T ij = 0 (with i, j = 1, 2, 3). Inserting the expressions for the Christoffel
symbols (28.2) we get:
ȧ
ρ̇ = −3(ρ + p) . (28.10)
a
For dust (p = 0):
ρ̇ ȧ
= −3 . (28.11)
ρ a
Integration gives ρa3 = constant or ρ ∝ a−3 . For a radiation dominated universe we get
ρ ρ̇ ȧ
p= ⇒ = −4 . (28.12)
3 ρ a

Integration gives ρa4 = constant or ρ ∝ a−4 . More generally for (28.8) one gets:

ρa(t)3(1+w) = constant. (28.13)

The Einstein equations with Λ (equation (20.10)) are

8πG
Gµν = Tµν + Λgµν .
c4
Using (19.8) and that uµ = (c, 0, 0, 0) in a comoving coordinate system, let us write down the 00-
component and the ij-component of this equation:
 2 
ȧ k
3 + 2 = 8πGρ + Λ, (28.14)
a2 a

ä ȧ2
 
k
2 + 2 + 2 gij = (−8πGp + Λ) gij . (28.15)
a a a

One has in addition equation (28.10) from the conservation law. Using the first equation to eliminate
ȧ2 k
a2 + a2 from the second one, one obtains the Friedmann equations:

ȧ2 k 8πG Λ
2
+ 2 = ρ+ , (28.16)
a a 3 3
ä
−3 = 4πG(ρ + 3p) − Λ, (28.17)
a
ȧ
ρ̇ = −3(ρ + p) . (28.18)
a

Notice that one could also use the form (20.8) of the Einstein equations to derive the above equations.
ȧ(t)
Introducing the Hubble parameter: H(t) = a(t) and the deceleration parameter: q(t) = − a(t)ä(t)
ȧ2 (t) , with
their present day values denoted by H0 = H(t0 ) and q0 = q(t0 ), where t0 is the age of the universe;

92
 28 FRIEDMANN EQUATIONS

we get instead
8πG k Λ
H2 = ρ− 2 +
3 a 3
1
q= (4πG(ρ + 3p) − Λ) (28.19)
3H 2
d
(ρa3 ) = −3Hpa3
dt
3H 2 ρ
In the case of Λ = 0, we define a critical density ρcrit = 8πG and a density parameter Ω = ρcrit . Then



 ρ < ρcrit ⇔ k = −1 open universe

ρ = ρcrit ⇔ k = 0 flat universe


crit ⇔ k = +1 close universe

ρ > ρ

Let us now assume that the density is a combination of dust (that we shall simply denote as
“matter”) and radiation: ρ = ρm + ρr . Moreover, we assume that ρm ∼ a−3 and ρr ∼ a−4 . This is
valid if radiation and matter are decoupled, or if one density is much bigger that the other one (notice
8πG 3 8πG 4
that in today’s universe ρm  ρr ). Let us introduce the constants Km = 3 ρm a and Kr = 3 ρr a .
Inserting them into equation (28.16) leads to

Kr Km 1
ȧ2 − − − Λa2 = −k. (28.20)
a2 a 3
This equation reads as
ȧ2 + V (a) = −k,

where
Kr Km 1
V (a) = − − − Λa2 (28.21)
a2 a 3
plays the role of an effective potential, see figure 18.
Kr Km
Consider the solution for a → 0: in that case the terms a2 and a dominate and the behavior
does not depend neither on k nor on Λ.
Kr √
ȧ2 ≈ → a(t) ∼ t (28.22)
a2
Km 2
ȧ2 ≈ → a(t) ∼ t 3 (28.23)
a
√
For a → 0, ȧ goes to ∞. If Kr 6= 0, then for a → 0 a ∼ t. From figure 18 we can discriminate
different types of solutions:

1. For Λ < 0, there is for all k-values a maximal amax . Thus there will be a periodic solution going
between a = 0 and amax .

2. Λ = 0 (Einstein-de Sitter universe): for k = 1 there is a bounded solution as in the case 1. For
k = 0 the expansion velocity goes towards 0, instead for k = −1 it goes towards a constant value.

3. For Λ > 0 there are several cases:

93
 28 FRIEDMANN EQUATIONS

a) Λ = Λcrit and k = 1. The value of Λcrit is obtained for the horizontal line −k (figure 18)
just going through the maximum of the potential. Assuming Kr ≈ 0 (as in today’s universe)
dV
we get with ȧ = 0, V = −1 → da = 0:

4 3Km
Λ = Λcrit = 2
and a = astat = . (28.24)
9Km 2
This corresponds to the static Einstein solution. Einstein introduced a cosmological constant
in order to get such a static solution, which is however unstable. Small perturbations lead
either to a contraction or to an exponential growth.
b) Λ < Λcrit and k = 1. The horizontal line −k intercepts the curve V (a) in two points a1 and
a2 . We obtain either a periodic solution between 0 and a1 or an unbounded solution with
a > a2 .
c) Λ = Λcrit (1 + ) and k = 1. For 0 <   1 the horizontal line −k lies just above the
maximum value of V (a). Thus the expansion velocity ȧ will be very low in this region
(Lemaître universe).
d) Λ > Λcrit and k = 1. The line −k does not intercept the V (a) curve. Around the maximum
of V (a) the expansion is lowered.
e) Λ > 0 and k = −1, 0 : as in the previous case, but there may be less deceleration in the
region of the maximum of V (a).

Λ<0

−k

a
Λ=0

Λ>0

Figure 18: Sketch of the “effective potential” V (a) for different values of Λ.

For Λ > 0, all solutions are unbounded. In the limit a → ∞, V (a) is dominated by the Λ-term:

Λ 2
ȧ2 ≈ a ,
3

94
 28 FRIEDMANN EQUATIONS

thus r !
Λ
a(t) ∼ exp t . (28.25)
3
The expansion is exponentially accelerated.
According to the currently most accepted model (so-called ΛCDM-model), the main contributions
to the density are

• ordinary matter (baryons) Ωbaryons ∼ 0.05

• dark matter ΩDM ∼ 0.27

ρΛ Λ
• cosmological constant ΩΛ = ρcrit ∼ 0.68 with ρΛ = 8πG (also called dark energy).

Moreover, the universe seems to be almost flat: k ≈ 0. These cosmological parameters would thus
correspond to the case 3e) of the previous discussion.
Finally, H0−1 is related to the age of the universe. H0 ∼ 67.80 ± 0.77 sec/Mpc
km
gives an age of ∼ 13.8
billion years.30

30 Planck 2013 results. XVI. Cosmological parameters - [Link]

95