Space Time and Gravity: Revision Notes

This set of notes is a summary of the main formulae and key

concepts introduced in the course “Space Time and Gravity”. They
have been thought as a guide for revising the course, but they are not
a complete summary of the course material.
Students should learn and understand the physical meaning and
the derivation of the concepts and formulae included in this file as
they had been presented during the lectures.
Einstein’s sum convention applies throughout the paper.

Basic concepts in General Relativity

• Before General Relativity NEWTONIAN GRAVITY was providing a co-
herent description of gravity by describing:
– the acceleration of a body in a gravitational potential φ(x):

~
~a = −∇φ(x)

which tells how gravitation influences matter’s behaviour;

– the gravitational potential generated by a mass distribution ρ(x):

∇2 φ(x) = 4πGρ(x)

which tells how a matter distribution determines gravitation.

• 1905 - SPECIAL RELATIVITY: but it is valid only for inertial frames.
How does special relativity generalise when we deal with accelerated frames?
• Is there a connection between gravity and accelerated frames? Encoded by
the Weak Equivalence Principle: “The inertial mass and the gravitational
mass of a body do coincide”.
• Experiments show that the acceleration felt by a body moving freely in a
gravitational field is universal and independent on the particle’s properties.
Thus gravity is an intrinsic property of spacetime rather than being the
effect of a propagating field.

1
 • Gravity has only a relative existence: a person in a small box which is
freely falling does not experience any gravitational field. Furthermore we
can “fake” the effects of a gravitational field by accelerating a box in empty
space.
⇒ EINSTEIN’S EQUIVALENCE PRINCIPLE : “In small enough re-
gions of spacetime the laws of physics reduce to those of special rela-
tivity; it is impossible to detect the existence of a gravitational field
by means of local experiments”.
• It exists a preferred set of trajectories, named inertial or freely falling,
which are those followed by accelerated particles (i.e., particles subjected
only to the action of gravity), and we should attribute the effects of gravity
to the curvature of spacetime, namely to the curvature of the trajectories
which freely falling particles follow.
⇒ The notion of inertial frame is replaced by that of Freely Falling
Frame or Locally Inertial Frame: a frame which is subjected only
to the local effects of gravity, namely which is moving freely in the
proximity of an object deforming spacetime and sourcing gravity, and
that is small enough to make the inhomogeneities of the gravitational
field undetectable.
⇒ We can rephrase the Einstein’s Equivalence Principle as: “At any
point in spacetime, in any arbitrary gravitational field, it is possible
to choose a Locally Inertial Frame such that in an arbitrary neigh-
bourhood of that frame the laws of Physics assume the form as in an
unaccelerated coordinate system in the absence of gravity”.
• We can get a measure of spacetime curvature already in the Newtonian
theory, by observing the non-local effect of gravity
– Observe the relative motion of two test particles in the proximity of
the Earth. The gravitational potential is φ(~x). One particle sits in ~x
y , and they are ξ~ apart: ~y = ~x + ξ.
and the other in ~ ~
– the variation of the relative velocity of the two particles is given by
the formula:
ξ¨i ≈ Kij ξ j , , i, j = 1, 2, 3
∂ 2 φ(~
x)
where a sum over repeated indexes is understood and where Kij = ∂xi ∂xj .
– Notice that: i Kii ≡ Tr K = ∇2 φ = 4πρ
P  

SPECIAL RELATIVITY
Postulates of Special Relativity:
• The fundamental laws of physics have the same form in any inertial ref-
erence frame.

2
 • The speed of light in the vacuum is a universal constant.

Lorentz Transformations
Two inertial reference frames K, described by the set of coordinates (t, x, y, z),
and K ′ , described by the set of coordinates (t′ , x′ , y ′ , z ′ ), moving with relative
speed ~v along the x-axes are related by the following set of Lorentz transforma-
tions (c = 1):

′ √t−vx
′
t = 1−v2 t = cosh ζt − sinh ζx

 
x′ = √x−vt
 
x′ = − sinh ζt − cosh ζx
1−v 2 −→
y ′ = y
 y ′ = y

 
 ′
z =z

z ′ = z

where we have introduced the rapidity ζ defined by v = tanh ζ.

• The Lorentz transformation leave invariant the spacetime interval:

(∆s)2 = −(∆t)2 + (∆x)2 + (∆y)2 + (∆z)2 .

• (∆s)2 is the distance between two points (events) in Minkowsky spacetime.

• If we consider two points in Minkowsky spacetime P in (tP , xP , yP , zP )
and Q in (tP + dt, xP + dx, yP + dy, zP + dz), the (infinitesimal) invariant
distance between P and Q is:

ds2 = −dt2 + dx2 + dy 2 + dz 2 .

• The (finite or infinitesimal) distance between two events in spacetime can

be greater, less, or equal to zero. According to this, the two events are:
– if (∆s)2 < 0 ⇒ timelike separated ⇒ this for example is the case
of events happening at the same place but at different time; can be
connected by a particle travelling at speed lower than the speed of
light;
– if (∆s)2 = 0 ⇒ lightlike (or null ) separated ⇒ can be connected by
light rays;
– if (∆s)2 > 0 ⇒ spacelike separated ⇒ this for example is the case of
events happening at the same place but at different time; cannot be
connected by a particle moving at speed lower or equal to the speed
of light.
• The light cone of a point P is the locus of points which are null separated
from P . Each point P in spacetime has its own light cone.
– The future light cone is generated by the light rays moving outward
from P .

3
 – The past light cone is generate by the light rays converging in P .
– The points that are timelike separated from P lay inside the light
cone. The points that are spacelike separated from P lay outside the
light cone.
– Particles with non zero rest mass move along timelike paths which
are always inside the light cones at each point along their trajectories.
• The proper time τ is defined by:

ds2
dτ 2 = −ds2 (dτ 2 = − if c 6= 1)
c2
and can be defined as the time that would be measured by a clock carried
by the observer himself.
• Let’s introduce the four vector xµ = (t, x, y, z), where µ is an index taking
values from 0 to 3 (namely x0 = t, x1 = x, etc.) . Then we can define the
scalar product (Einstein sum convention applies):

x · x = xµ ηµν xν = −t2 + x2 + y 2 + z 2

where
 
−1 0 0 0
0 1 0 0 is the
ηµν =
0
 ←−
0 1 0 Minkowsky metric
0 0 0 1

• The infinitesimal Minkowsky spacetime separation is rewritten as

ds2 = ηµν dxµ dxν

CURVED GEOMETRIES
• Can be described through their embedding in higher dimensional spaces.
Example: a two-dimensional spherical surface can be seen as the set of
points in R3 whose coordinates satisfy the equation x2 + y 2 + z 2 = r2 .
• Can otherwise be described through a set of intrinsic properties, which do
not depend on anything outside the surface itself. These are:
– The set of geodesics of a surface, namely the paths of minimal dis-
tance between two points in the curved space under consideration.
They are the equivalent of straight lines of Euclidean geometry, be-
cause they are the “straightest” possible path one can draw on the
surface.
Example: On the 2-dimensional sphere, the geodesics are half great
circles.

4
– The lenght of segments (portions of geodesics) on a given spacetime,
and their relations provide a way of deriving its intrinsic curvature.
Example: We know from Euclidean geometry that the ration be-
tween the circumference of a circle drawn on a flat surface and its
radius is Cr = 2π. On the other hand, if we draw a circle of ra-
dius r on the surface of a 2-dimensional sphere (see figure), we get:

C
C sin r
a
= 2π r a
r a

Expanding the relation above for small values of r we get:

C 1
≈ 2π(1 − Kr2 )
r 6
where K = a12 is the (intrinsic) Gaussian curvature of a 2 dimen-
sional surface in the neighbourhood of a point P.
– A general local definition for the intrinsic curvature of a 2-dimensional
surface is then:
3 2πr − C
K = lim
π r→0 r3
– Besides the circumference C, other geometrical properties of a space-
time which depend on its intrinsic curvature are:
∗ the area A of a circle on a 2-dimensional surface:
r2
A ≈ πr2 (1 − K + . . .)
12
∗ The second variation of the spread between two geodesics inter-
secting at a point P :
η̈ = −Kη
These properties depend on the sign of K:
∗ if K > 0, C and A are smaller than the flat space case, and two
geodesic spread with decreasing speed;
∗ if K < 0, C and A are bigger than the flat space case, and two
geodesic spread with increasing speed;
∗ if K = 0, we retrieve the flat space results.

5
Metric
The metric provides a description of a given spacetime geometry which is not
restrained to the neighbourhood of a point.
In an arbitrary 4-dimensional curved geometry the line element will be a
generalisation of the Minkowsky metric. In general, we define:
(n-dimensional) Riemannian Space: is an n-dimensional space parametrised
by a set of n coordinates xµ (µ = 0, . . . , n − 1), whose intrinsic geometry is de-
fined by a metric gµν (x) such that:
ds2 = gµν (xρ )dxµ dxν and det g 6= 0
Important remarks:
• The inverse metric g µν corresponds to the inverse matrix: gµρ g ρν = δµρ .
Example: A 2-dimensional Riemannian space is described by the line el-
2
ement ds2 = x2 dx2 + xydy 2 . The metric is gµν = x0 xy


0 . The inverse
−2
metric is gµν = x 0 (xy)0 −1 .

• The set of coordinates xµ is not necessary uniquely defined. One can have
different coordinate systems related by a non-degenerate change of coor-
dinates, and the form of the line element will not in general be preserved
when switching from one set of coordinates to another.
Example: Let’s consider the Euclidean plane R2 :
– in Cartesian coordinate (x, y):
 
2 2 2 1 0
ds = dx + dy ⇒ gij =
0 1

– in polar coordinates (r, θ)

 
1 0
ds2 = dr2 + r2 dθ2 ⇒ gij =
0 r2

• It may happen that a single coordinate set is not sufficient to describe the
full space. We can encounter a coordinate singularity. This is a point P ,
of coordinates xP such that one or more elements of the metric diverges
for x = xP , but such that the curvature scalars behave well in that point.
Example: The metric ds2 = ρ14 (dρ2 + ρ2 dθ2 ) has a singularity for ρ = 0.
Is this a true curvature singularity? Does the geometry behaves badly
for ρ = 0? If we perform the change of coordinates ρ = 1r the metric
becomes ds2 = dr2 + r2 dθ2 . This is just the metric on the Euclidean plane
in polar coordinates. The Euclidean plane is flat everywhere, the point
ρ = 0 corresponds to all the set of points at infinity which are taken to a
single point by the change of coordinates r → ρ. This simply means that
the set of coordinates (ρ, θ) are not suitable to describe this point.
Another example of coordinate singularity is the Schwarzschild radius (see
later).

6
• If a geometry can be described through its embedding into a higher di-
mensional space, its metric can be derived restraining the line element of
this space on the lower dimensional one.
Example: S 2 can be embedded in R3

– R3 line element: ds2 = dx2 + dy 2 + dz 2 ;

– each point on S 2 can be parametrised by two angular coordinates
(θ, φ) such that x = R sin θ cos φ, y = R sin θ sin φ, and z = R cos θ.
Implementing this in the line element above yields to the metric on
S 2 : ds2 |S2 = R2 (dθ2 + sin2 θdφ2 ).
• Spacetime is a 4-dimensional Riemannian space.
– The equivalence principle suggests that locally the properties of space-
time should be indistinguishable from those of the flat spacetime of
special relativity. This means that given a metric gµν (x) at each point
P of spacetime it is possible to introduce a new set of coordinates
x′µ (x) such that
′
gµν (x′α
P ) = ηµν

where ηµν is the flat Minkowsky metric and x′α

P are the coordinates
of the point P .
– Furthermore one can find a change of coordinates such that locally
in a point P :
′
′
∂gµν
gµν (x′α
P ) = ηµν and =0
∂x′ρ P

– This is the mathematical description of a freely falling frame or locally

inertial frame.
– If spacetime is curved, it is not possible to find a change of coordinates
such that the metric becomes flat everywhere. This can happen only
if the whole spacetime is flat, and there is no curvature (there is no
gravity → gravity is spacetime curvature).
• The definition of timelike, spacelike and lightlike separations hold in curved
space as well. Massive particles move along timelike trajectories, and the
set of lightlike curves connected with a point P form the light cone in P .
• Given a metric ds2 = gtt dt2 + gxx dx2 + gyy dy 2 + gzz dz 2
– the infinitesimal proper lenght along the directions t, x, y, and z are
√ √ √ √
respectively gtt dt, gxx dx, gyy dy, and gzz dz.
√
– the area element along the directions x and y is dA = gxx gyy dxdy;
similar formulae hold for infinitesimal area elements along different
directions;
√
– the infinitesimal 3-volume element is dV = gxx gyy gzz dxdydz;
√
– the infinitesimal 4-volume element is dW = −gtt gxx gyy gzz dtdxdydz;

7
Tensor calculus
Principle of General Covariance
The laws of Physics can be stated in a form which is independent on the specific
choice of spacetime coordinates.
The laws of Physics must be valid in any system of coordinates. Thus, they
must be expressible as tensor equation. In this way, they are covariant : they
retain the same form upon a coordinate transformation.

Coordinate transformations
Let’s focus on a d-dimensional spacetime geometry and let’s consider an invert-
ible coordinate transformation:

xµ → x′µ , µ = 0, . . . , d − 1

.
Under such a coordinate transformation:
• a scalar field is invariant:
φ(x) = φ′ (x′ )

• a contravariant vector transforms as:

Aµ (x) −→ A′µ (x′ ) = Λµν Aν (x)

∂x′µ
where Λµν = ∂xν ;

• a covariant vector transforms as:

Bµ (x) −→ Bµ′ (x′ ) = Λµν Bν (x)

∂xν
where Λµν = ∂x′µ .

The contraction of a contravariant and a covariant vector field is a scalar

under a coordinate transformation, hence: Λµν = (Λµν )−1 .

Tensors
µ µ ...µ
A tensor T 1 2 pν1 ν2 ...νq is a multi-index mathematical object which trans-
forms linearly and homogeneously under a spacetime coordinate transformation:
• every contravariant (up) index transforms with Λµν
• every covariant (down) index transforms with Λµν
Remarks.
• The total number of indexes denotes the rank of the tensor. Hence:

8
 – T µ1 µ2 ...µp is a contravariant tensor of rank p;
– Tµ1 µ2 ...µq is a covariant tensor of rank q;
µ1 µ2 ...µp
– T ν1 ν2 ...νq is a mixed tensor of rank p + q;
• The null tensor (every component equal to zero) is zero in any reference
frame.
• The sum of two tensor of one kind is still a tensor of the same kind:

Aµ (x) + Bµ (x) = Cµ (x)

• The product of two tensors is a tensor with rank equal to the sum of the
ranks of the two factors:

Aµ1 µ2 ...µp · Bν1 ν2 ...νq = C µ1 µ2 ...µpν1 ν2 ...νq

• The metric is a covariant tensor of rank 2. Under a coordinate transfor-

mation it transforms as:
′
gµν (x) −→ gµν (x′ ) = Λµρ Λν σ gρσ (x)

• Contraction of indexes:
– contracting a covariant and a contravariant tensor we obtain a scalar
(e.g Aµ B µ = A0 B 0 + A1 B 1 + A2 B 2 + A3 B 3 in 4 dimensions);
– we can also contract covariant and contravariant indexes in the same
tensor: T µµ ≡ T = T 00 + T 11 + T 22 + T 33 ;
– this generalises to the contraction of any pair of tensor indexes:
T ρµµ ≡ T ρ is a contravariant tensor of rank 1, Tµνσ µν ≡ Tσ is a
covariant tensor of rank 1, etc.
• we can use the metric gµν and the inverse metric g µν to raise and lower
the indexes:
– Tµ = gµν T ν
– T µ = g µν Tν
– T µµ = T µν gµν = Tµν g µν

The Covariant Derivative

The ordinary derivative of a scalar is a tensor, but the derivative of a covariant
(or contravariant) vector is not. In fact, it does not transforms covariantly as a
tensor under a coordinate transformation xµ → x′µ :
x→x′
∂µ φ(x) −−−→ Λµν ∂ν φ(x) X
2 ρ
x→x′ ∂ x
∂µ Vν (x) −−−→ Λµρ Λν σ ∂ρ Vσ (x) + Vρ (x) ×
∂x′µ ∂x′ν

9
 We overcome this problem by introducing the covariant derivative:

Dµ Vν (x) = ∂µ Vν (x) − Γα
µν Vα (x)

where the set of coefficient Γα

µν form the connection. The connection is not a
tensor: under a coordinate transformation, it transforms as:

x→x′ ν ∂ 2 x′ρ
Γα ′α α ρ µ
βγ −−−→ Γβγ = Λ µ Λβ Λγ Γνρ + Λρ
α
∂x′β ∂x′γ
The covariant derivative of a covariant vector Dµ Vν (x) transforms as a co-
variant rank-2 tensor.
We define similarly the covariant derivative of a contravariant vector:

Dµ B ν (x) = ∂µ B ν (x) + Γνµρ B ρ (x)

Under a coordinate transformation Dµ B ν (x) transforms as a mixed tensor of

x→x′
rank 2: Dµ B ν (x) −−−→ Dµ′ B ′ν (x) = Λµρ Λν σ Dρ B σ (x)
These definitions are easily generalised to tensors of arbitrary rank.
Remarks.
• The covariant derivative is distributive: Dµ (T S) = (Dµ T )S + T (Dµ S) (T
and S are tensors of arbitrary rank).
• The covariant derivative of scalar is equivalent to the ordinary derivative:
Dµ φ(x) = ∂µ φ(x)
• Demanding that the covariant derivative commute with the operation of
raising and lowering indexes, Dµ (Aρ g ρν ) ≡ (Dµ Aρ )g ρν , we derive the
further condition:
The covariant derivative of the
Dµ g ρν = 0
metric is identically zero

This last condition can be expanded and solved for the connection coeffi-
cient, yielding to the definition of the Christoffel Symbols:
1 αµ
Γα
βγ = g (∂β gµγ + ∂γ gµβ − ∂µ gβγ )
2

• We define the directional covariant derivative along a path xµ (s) as:

D dxµ
= Dµ .
Ds ds
The directional covariant derivative does not change the rank of a tensor,
it just moves it along the path!
µ ...µ
A tensor T 1 pν1 ...νq is parallely transported along the path xµ (s) if its di-
ρ µ1 ...µp
rectional covariant derivative along that path is zero: dx
ds Dρ T ν1 ...νq = 0

10
Geodesics Equations
Geodesics are the equivalent in curved space of what straight lines are in flat
space. They can then be defined as the curves whose tangent vector does not
vary when moved along: the Geodesics of a curved space are the set of curves
along which the tangent vector is parallely transported.
These are the set of curves satisfying the equations:

d2 xµ ν
µ dx dx
ρ
+ Γ νρ =0
ds2 ds ds
In a given geometry, let’s consider a curve parametrised by xµ (s). The vector
µ µ
tangent to the curve is uµ = dxds(s) , and it is constant along the curve if du
ds = 0.
But :
duµ dxρ COVARIANTISE dxρ
0= = ∂ρ uµ −−−−−−−−−−−−−−→ Dρ u µ = 0
ds ds ds
We obtain the geodesics equations by expanding the expression for the covariant
derivative in the last formula.
Remarks.
• Geodesics are the trajectories followed by unaccelerated particles, sub-
jected only to the action of gravity.
• In flat space and in Cartesian rectilinear coordinates Γµνρ = 0, and the
d2 xµ
geodesics equations reduce to ds2 = 0, i.e. they describe the motion of
a particle on a straight line.
• For timelike trajectories the parameter s can be any affine parameter,
namely any parameter related to the proper time τ by a linear transfor-
mation: s = aτ + b, a and b constants.
• It is possible to derive the geodesic equations as the equations of the
trajectories extremising the proper time between the starting point A and
end point B of the particle trajectory:
Z Br
dxµ dxν
τAB = −gµν ds
A ds ds

• Along the geodesics the following normalisation condition holds:

dxµ dxν
gµν = constant
ds ds
If the affine parameter s coincides with the proper time τ , then the con-
stant is equal to −1.

11
Locally Inertial Frames
Let’s consider a Riemannian spacetime whose geometry is described by the
metric gµν (x), in the set of coordinates xµ . Then, at any point P it is possible
to introduce a new set of coordinates, namely a new reference frame, which we
denote x′µ , such that:
′
′
∂gµν
gµν (x′P ) = ηµν and =0
∂x′ρ P

where ηµν is the Minkowsky metric and x′P are the coordinates of P in the new
set.
This is the formal definition of a Locally Inertial Frame.
Remarks:
∂g′
• The condition ∂xµν ′ρ |P = 0 is automatically satisfied if, in the new set of

coordinates, Γ′µ
νρ |P = 0.

• These relations hold locally in a small neighbourhood of a point P : one

cannot find a unique coordinate transformation under which they hold
everywhere, unless the geometry under consideration is flat.
• The second derivatives of the metric are non-zero and they enter in the
definition of the spacetime intrinsic curvature in the neighbourhood of the
point P .
• The spacetime distance between the point P and its neighbouring points
can be either computed in the old set of coordinates or in the LIF coordi-
nates, and they can be either negative, positive, or null. As usual light rays
move along null directions, while massive particles along timelike world-
lines. Locally, light-cone structure is the same as in special relativity, it is
the distribution of light cones in spacetime which varies.

The Riemann Curvature Tensor

Parallel transport between two points along two different paths yields to two
different results. This is related to the fact that we cannot exchange the action
of two subsequent covariant derivatives. In a curved geometry the commutator
of two covariant derivatives acting on a four-vector Aµ is non zero, and quantifies
the spacetime curvature:

Dσ Dν Aµ − Dσ Dν Aµ = Aρ Rρµνσ

The tensor Rρµνσ is the Riemann Curvature Tensor :

Rρµνσ = ∂σ Γρµν − ∂ν Γρµσ + Γα ρ α ρ

µσ Γαν − Γµν Γασ .

Remarks.

12
 • The components of the Riemann tensor depend on the second derivatives
of the metric: in a curved geometry they cannot be all put to zero by
changing to Riemann Normal Coordinates;
• If, in a given spacetime, it exists a coordinate choice for which Rρµνσ = 0
for every value of the indexes, then that spacetime is flat.
• Symmetries of the Riemann Tensor:
Rµνρσ = Rρσµν = −Rνµρσ = −Rµνσρ
where Rµνρσ = gµα Rανρσ

Ricci Tensor
Obtained by contracting the first and third indexes of the Riemann Tensor:
Rµν = g αβ Rαµβν

Ricci Scalar
Obtained by contracting the Ricci tensor with the metric:
R = g µν Rµν

Einstein Tensor
It is the combination:
1
Gµν = Rµν − Rgµν
2

Bianchi Identities
Dλ Rρσµν + Dρ Rσλµν + Dσ Rλρµν = 0
• The Bianchi Identities can be verified by rewriting Dλ Rρσµν in Riemann
Normal Coordinates (such that Γ = 0 but ∂Γ 6= 0), and then summing
over the cyclical permutations of λ, ρ and σ.
• If we contract the Bianchi identities written above with g ρσ g µν we get:
 1 
Dµ Gµλ = Dµ Rµλ − Rgµλ = 0
2
The Einstein tensor is conserved.

Equation of Geodesic Deviation

It describes the relative acceleration between two geodesics xµ (s) and x∗ µ (s) =
xµ (s) + ξ µ (s):
D2 ξ µ µ
ν
ρ dx dx
σ
= −R νρσ ξ
Ds2 ds ds
The parameter s is a generic affine parameter along the geodesics.

13
Minimal Coupling Between Matter and Gravity
Gravity is universal: it can be described as a feature of the background where
particles move rather than as a propagating field:
• free particles are those subjected only to the effects of gravity;
• free particles move along geodesics defined by the geodesic equation.
The minimal coupling between matter and gravity let us combine these find-
ings with the laws of physics that we gather from special relativity,
To minimally couple matter and gravity we:
1. take the laws of physics valid in an inertial frame in Minkowsky spacetime;
2. write them in a covariant form;
3. assert that the results which they describe remain valid in a curved ge-
ometry characterised by the metric gµν , then we get their curved space
counterpart by replacing:
• ηµν −→ gµν
• ∂µ −→ Dµ
Examples.
• GEODESIC EQUATIONS The minimal coupling is exactly what we did
when we derived the equations for the geodesics. We:
– wrote in a covariant form the equation demanding that the tangent
ρ
vector is constant along the geodesics: dx µ
ds ∂ρ u ;
– replaced the ordinary derivative ∂ρ with the covariant derivative Dρ .
• MAXWELL EQUATIONS In covariant form in Minkowsky spacetime they
read ∂µ F µν = −j ν . Minimally coupled with gravity, they become:

Dµ F µν = −j ν

where all the contractions are performed with the metric gµν .
• CONSERVATION LAWS. In a covariant form and in flat spacetime the
conservations of energy and momentum is encoded in the conservation of
the stress energy tensor T µν . Then:
[Link]
∂µ T µν = 0 −−−−−−−−−→ Dµ T µν = 0

14
Newtonian limit of GR
We must recover the results of Newtonian gravity under the following conditions:
1. The particles move slowly with respect to the speed of light.
2. Gravity is weak, and can be treated as a perturbation of flat spacetime.
3. Fields are static: no dependence on the coordinate time t.

Newtonian limit of the geodesics equations

Let’s focus on the trajectories of timelike particles, which we parametrise with
their proper time τ . At the level of the geodesic equation, the three conditions
above are implemented by requiring:
dxi dt
1. dτ ≪ dτ ∀ i = 1, 2, 3
2. gµν = ηµν + hµν with |hµν | << 1
∂
3. ∂t gµν =0 ∀ µ, ν = 0, 1, 2, 3.
Implementing this condition in the geodesic equation we get:
d2 xµ dxν dxρ dt
2
+ Γµνρ =0 if µ = 0 =⇒ = const.
dτ dτ dτ dτ
d2 xi 1
if µ = i 6= 0 =⇒ = ∂ i h00
dt2 2
• The first relation tells us that the coordinate time can be used as parameter
along the geodesics, while the second relation shows that, upon identifying
h00 ≡ −2φ(~x), where φ(~x) is the Newtonian potential, we get exactly the
Newton equation for gravity:
d2 ~x ~ ~(x))
= −∇φ(
dt2
• The identification between h00 and φ holds if we can write g00 = −(1 +
2φ(x)), namely if this component of the metric is a solution of the Einstein
equations (the equations that determine the explicit form of the metric).

Einstein Equations
• The Einstein equations are the relativistic generalisation of the Poisson
equation for the Newton potential ∇2 φ = 4πGρ(x), where ρ(x) is the
mass density.
• They have the form:
TENSOR DESCRIBING THE
TENSOR DESCRIBING THE
= SPACETIME ENERGY CON-
SPACETIME GEOMETRY
TENT

15
The Stress-Energy tensor
This is the tensor describing the matter and energy content of spacetime.

• Physical definition:

Flux of four-momentum pµ
T µν =⇒
across a surface of constant xµ
• Separating the “0” coordinate from ”i” = 1, 2, 3, we identify:
 00
T 0j

T
T i0 T ij

where
– T 00 : mass-energy density;
– T i0 : flux of energy;
– T 0j : momentum density;
– T ij : momentum flux: stress
• For a perfect fluid

T µν = (ρ + p)U µ U ν + pg µν
fluid
where
– ρ is the mass-energy density of the fluid;
– p is the pressure (same in every direction);
– U µ is the average macroscopic 4-velocity.
• The stress-energy tensor is conserved: Dµ T µν = 0. If we integrate this
relation over a volume V we recover the equations for the conservation of
energy (integrating the relation for ν = 0) and for the conservation of the
three components of the linear momentum (integrating the three relations
for ν = i 6= 0).

Complete form of the Einstein Equations

1
Rµν − Rgµν + Λgµν = 8πGTµν
2
where:
→ gµν is the metric describing the geometry of a given spacetime.
→ Rµν = Rαµβν g αβ is the Ricci tensor.
→ R = Rµν g µν is the Ricci scalar.

16
 → Tµν is the stress-energy tensor, which describes the matter and energy
content of a given spacetime.
→ Λ is the cosmological constant, which describes the vacuum energy density,
namely the energy density of empty space.

Einstein equations in the vacuum

In the vacuum Tµν = 0, then the Einstein equations reduce to:

Rµν = 0

Remarks.
• These equations can be derived comparing the equation for the geodesic
deviation with the equation for the tidal forces in the case when the mass
density of the gravitational source is zero.
• In the static and weak field limit they reduce to the Poisson equation in
empty space:
Rµν ∼ −δµν ∇2 φ(x)
then R00 = 0 ⇒ ∇2 φ(x) = 0.

The Schwarzschild Metric

It is the unique static and spherically symmetric solution of the Einstein equa-
tions in the vacuum Rµν = 0. In spherical coordinates (t, r, θ, φ):
 2GM  2  2GM −1 2
ds2 = − 1 − dt + 1 − dr + r2 (dθ2 + sin2 θdφ2 )
r r
where M is the mass sourcing the gravitational field and G is the universal
gravitational constant.

• The coefficients of the metric becomes singular for r = 0 and r = 2GM .

The quantity rS ≡ 2GM is called Schwarzschild radius and it is a coordi-
nate singularity of the Schwarzschild metric (if c 6= 1 rS ≡ 2GM
c2 ). When
r = 2GM the coefficient grr of the metric becomes singular, however this
is due just to the particular coordinate set we are using, and the geom-
etry behaves well. To prove it one must consider scalars built out of the
curvature Riemann tensor, and check if they are finite or if they diverge
in the singular points of the metric. For example one can consider

48G2 M 2
Rαβγδ Rαβγδ =
r6
r→0
We see that Rαβγδ Rαβγδ −−−→ ∞, so r = 0 is a true curvature singularity,
while Rαβγδ Rαβγδ |R=2GM = 4G43M 4 is finite at the Schwarzschild radius.

17
 • The Schwarzschild radius has a physical meaning only for ultra-compact
objects, such that this distance is still in the vacuum, and the neigh-
bouring geometry is still described by the Schwarzschild metric (e.g., the
2GM⊙
Schwarzschild radius of the Sun c2 is about 2.9km, well inside the
surface of the star, where the geometry is no longer described by the
Schwarzschild solution).
• The Schwarzschild radius is the distance of the event horizon of a black
hole.
• The weak field limit of the Schwarzschild metric holds far from the mass
sourcing the geometry, namely for r ≫ 2GM . At such distances the metric
can be seen as a perturbation of the Minkowsky metric: gµν = ηµν + hµν .
But we can rewrite g00 = −1 + 2GM r = η00 − 2φ(r), where φ(r) is the
Newtonian potential. Similarly, if we take the weak field limit of grr we
get:
 2GM −1 r≫2GM 2GM
grr = 1 − −−−−−→ 1 + = η11 − 2φ(r)
r r
and the metric becomes:
r≫2GM
ds2 −−−−−→ −(1 + 2φ(r))dt2 + (1 − 2φ(r))dr2 + r2 (dθ2 + sin2 θdφ2 )

To write this as a perturbation of Minkowsky, we can choose hµν =

−2φδµν .
• The metric tends to Minkowsky for r → ∞. This phenomenon is called
asymptotic flatness, and it is a consequence of the metric, not an ansatz.
• The metric tends to Minkowsky for M → 0 (no mass sourcing the geom-
etry).

Tests of General Relativity

Three phenomena predicted by General Relativity which can be considered as
experimental tests of the theory are:
• The precession of planet’s perihelia
• The bending of light rays in proximity of a gravitational source
• The gravitational redshift
To quantify the effects of the first two phenomena we need to know the orbits
of test particles in the Schwarzschild geometry. Spherical symmetry constraints
the orbits to lay on planes of constant θ. Let’s fix θ = π2 .

18
Orbits of massive particles
• Timelike orbits can be computed by exploiting the following conditions:
– Conservation of energy (because of no time dependence):
 2GM  dt
1− = constant = k
r dτ

– Conservation of angular momentum (because of spherical symmetry):

dφ
r2 = constant = l
dτ
µ
dxν
– normalisation condition along timelike geodesics gµν dx
dτ dτ = −1:
 2GM  d2 t  2GM −1 d2 r 2
2d φ
1− − 1 − − r =1
r dτ 2 r dτ 2 dτ 2
• Combining these three conditions we find that particles move along the
orbits of the effective potential

MG l2 l2 M G
Vef f (r) = − + 2−
r 2r r3

• We have bigger discrepancy with respect to the Newtonian effective po-

l2
tential VN (r) = − MGr + 2r 2 for smaller values of r: search for measurable
effects in the orbits of innermost planets of the Solar System.

Precession of planets’ perihelia

• Introducing u = r1 , the variation of u with r is encoded by the equation:

d2 u MG
+ u = 2 + 3M Gu2
dφ2 l
d2 u MG
while the Newtonian effective potential gives: dφ2 +u= l2 .

• The extra term is small for solar system planets, and can be treated as a
perturbation, leading to the approximate solution

GM h  3G2 M 2
i
uGR = 1 + e cos φ 1 −
l2 l2
2π
• uGR is a periodic function of period 2 2 > 2π.
1− 3Gl2M

2π
• For Mercury 2 2 − 2π ∼ 42.9” per century.
1− 3Gl2M

19
Photons’ orbits and deflection of light rays
• Lightlike trajectories can be worked out exactly in the same way as for
timelike, but now considering that for lightlike trajectories ds2 = 0, then
the normalisation condition along the geodesic reads:
 2GM  d2 t  2GM −1 d2 r 2
2d φ
1− − 1 − − r =0
r dτ 2 r dτ 2 dτ 2

• The equation for u(φ) now reads:

d2 u
+ u = 3M Gu2
dφ2

• If M = 0 the equation this is the equation for a straight line

• When M 6= 0 it is solved by:
1 sin φ M G
= + (3 + cos 2φ)
r R 2R
where R is an integration constant.
• in Cartesian coordinates the distance from the gravitational source along
the y axes varies as:

GM y 2 + 2x2
y =R− p ,
R x2 + y 2
and the total deflection from x → −∞ to x → +∞ is equal to
y 4GM
Θ ∼ −2 ∼
x x→+∞ R
4GM
(if c 6= 1 ⇒ Rc2 )

• The deflection of starlight by the Sun is about 1.75”.

Gravitational redshift
Light climbing a gravitational field gets redshifted.
Let’s consider an atom in a gravitational field sitting at a distance re , which
emits radiation with frequency νe . The emitted pulse travels radially towards
an observer sitting at a distance r0 . If the metric is static (i.e., it does not
depend on time), the observed radiation will have frequency νo such that:
s
ν0 g00 (re )
=
νe g00 (ro )

where g00 is the coefficient of dt2 in the metric.

20
 In Schwarzschild geometry:
v
u
u1 − 2GM
ν0 re
=t 2GM
νe 1− ro

Remarks.
• To derive the formula for the gravitational redshift one must explicitly
use the fact that the metric does not depend on the coordinate time t,
and then in coordinate time the period of the emitted and of the observed
wave is the same.
• If we are far from the source (r ≫ 2GM ), and ννoe ≈ 1 − GM GM
re + ro =
1 + φ(re )− φ(ro ), where φ(r) is the Newton potential. Since φ(re ) > φ(r0 ),
νo < νe , i.e., the radiation gets redshifted.

Effective speed of light

The curvature induced around a massive body increases the travel time of light
rays relative to what it would be in flat space. For timelike geodesic ds2 = 0,
and if we focus on radial trajectories dφ = dθ = 0, then:
 2GM  2  2GM −1 2 dr  2GM 
1− dt − 1 − dr = 0 ⇒ = 1− <1
r r dt r

Time delay
Time runs slower in a gravitational field is compared to flat spacetime. For a
stationary observer (dr = dθ = dφ = 0), the proper time is
q
• in Schwarzschild: dτ = 1 − 2GM r dt

• in flat spacetime: dτ = dt
where t is the coordinate time measured by an observer at infinity.

Black Holes
The Schwarzschild solution describes static black holes with no electric charge.
• The Schwarzschild radius Rs = 2GM is a coordinate singularity of the
geometry: spacetime curvature is finite for r = RS , and this feature is an
artifact of the coordinate system we are using, which are not suitable to
describe this geometry for r ≤ 2GM .
• What does it happen to the metric when r < 2GM ?
– if r > 2GM : gtt < 0 and grr > 0

21
 – if r < 2GM : gtt > 0 and grr < 0
Thus considering a worldline along the t-axes (describing two events hap-
pening in the same position at different time):
 2GM  2 ր < 0 for r > 2GM ⇒ TIMELIKE
ds2 = − 1 − dt
r ց > 0 for r < 2GM ⇒ SPACELIKE

thus passing from r > 2GM to r < 2GM the time and space character of
the coordinates do reverse.
 
r→2GM
• gtt = − 1 − 2GMr −−−−−→ 0 ⇒ surface of infinite redshift.

Behaviour of light cones along radial paths

Along radial lines dθ = dφ = 0, and along the path of light rays ds2 = 0, then
the slope of the light cone is:
dt  2GM −1
=± 1−
dr r
(− and + refer to outgoing and ingoing light rays)
dt
• dr → ±1 for r → ∞ ⇒ describes light cones in ordinary Minkowsky
spacetime
dt
• Focusing only on the outgoing curves (sign +), dr → +∞ for r → 2GM
⇒ the light cones become more and more narrow while getting closer to
the event horizon.
• Passing from r > 2GM to r < 2GM flips the inner and the outer part of
the light cones: the role of r and t is exchanged, and for outgoing particles
r decreases as the time flows: the particles are bound to move towards
r = 0.

Path of a radially infalling particle

We want to describe what happens to an observer falling freely and radially into
a spherical black hole from a fixed distance r0 > 2GM . We want to describe
his motion:
• from his point of view;
• from the point of view of another observer, sitting still at the same distance
r0 .
Let’s suppose also that the free-falling observer carries a lamp which emits a
sequence of light pulses at constant proper time separation ∆τ . We see that:

22
 • The proper time that the observer takes to reach a given distance r < r0
is:
1 3 3 3

τ = −√ r 2 − r02
2GM 2
For the infalling observer, nothing special happens at r = 2GM . Neglect-
ing the action of tidal forces, he will just cross the horizon and keep on
falling until when, after a finite amount of proper time, he will reach the
central singularity at r = 0.
• The other observer sitting still at r0 measures the time with the coordinate
time t. He sees the freely-falling observer reaching the distance r such that
2GM < r < r0 after a time:
 √ √  √ √ 
2 h 3 3 √ √
i r + 2GM r 0 − 2GM
t=− √ r 2 − r02 + 6GM r − r0 +2GM ln  √ √  √ √ 
3 2GM r0 + 2GM r + 2GM

This means that:

– if both r and r0 are much bigger than the Schwarzschild radius 2GM ,
than t ≈ τ , and the light pulses will reach the still observer at sepa-
ration ∆t ≈ ∆τ ;
– t → ∞ for r → 2GM . So, the still observer will measure an in-
finite Schwarzschild coordinate time for the free-falling observer to
reach the horizon, and only the light pulses emitted before the falling
observer crosses the horizon will reach him. The still observer will
receive the light pulses at intervals ∆t < ∆t′ < ∆t” . . . which grow
larger and larger, and the pulses of light will be more and more red-
shifted.

Eddington-Finkelstein coordinates
Rewritten in this set of coordinates the metric is no longer singular at r = 2GM .
They are defined as the set of coordinates (t̄, r, θ, φ) such that the new time
coordinate t̄ is:
t̄ = t + 2GM ln |r − 2GM |
The metric becomes:
 2GM  2  2GM  2 4GM
ds2 = − 1 − dt̄ + 1 + dr + drdt̄ + r2 (dθ2 + sin2 θdφ2 )
r r r
and it is regular for r = 2GM . Studying the behaviour of radial light cones we
find that:
dt̄
= −1 (1a)
dr
dt̄ r + 2GM
= (1b)
dr r − 2GM

23
 • (1a) is always oriented at −45 deg with respect to the vertical axis
• (1b) tends to +1 (flat space) for r → ∞ and it tends to infinity for
r → 2GM , and when 0 < r < 2GM it decreases until the limit value −1.
• the future light cone is always directed towards r → 0
Remarks:
• The surface r = 2GM is a coordinate singularity which can be removed
via a change of coordinates.
• r = 2GM defines a membrane called event horizon of the black hole. It
is possible for future directed trajectories to cross it from r > 2GM to
r < 2GM , but the reverse is not possible.
• Moving towards smaller r, the light cones begin to tip over. At r = 2GM
lightlike directions are stationary.
• For r < 2GM all future directed lightlike and timelike trajectories are
directed towards r = 0.

Cosmology
Cosmology is the study of the universe as a whole. It is based upon the:
Cosmological Principle: The universe is homogeneous and isotropic.

− Homogeneous: the metric is the same throughout space.

− Isotropic: from any point space look the same in any direction.

Remarks.
• The cosmological principle is supported by observation
• Observation shows that distant galaxies are recessing from us: the universe
is not static, hence it is homogeneous and isotropic only in space, and not
in time. The metric can be put in the form:

ds2 = −dt2 + a2 (t)γij (x)dxi dxj

where
– t is the cosmological time, xi , i = 1, 2, 3 are the three spatial coor-
dinates, orthogonal to the time direction for an observer sitting at
constant xi (comoving).
– a(t) is the scale factor : it says how much big the “space part” of
spacetime is at any given coordinate time t.
– γij (x) is the maximally symmetric metric on the three-dimensional
space.

24
 • Implementing the homogeneity and isotropy conditions one finds:
dr2
 
2 2 2 2 2 2 2
ds = −dt + a (t) + r (dθ + sin θdφ )
1 − Kr2
This metric is called Robertson-Walker metric. The parameter K can
assume the following values:
– K = 1 ⇒ the spatial part has constant positive curvature, it is closed;
– K = 0 ⇒ the spatial part has zero curvature, it is flat;
– K = −1 ⇒ the spatial part has constant negative curvature, it is
open.
• The Einstein equations become a set of differential equations for the scale
factor a(t).

Matter and energy content of the universe

The universe is modelled as a perfect fluid at rest in comoving coordinates:

Tµν = (ρ + p)Uµ Uν + pgµν

with U µ = (1, 0, 0, 0) ⇒ Uµ = (−1, 0, 0, 0). It is convenient to consider T µν =

g µα Tαν  
−ρ 0 0 0
µ
0 p 0 0
T ν =
0 0

p 0
0 0 0 p
• The trace of the stress-energy tensor is:

T µµ = −ρ + 3p

• The stress-energy tensor is conserved: Dµ T µν = 0. The equation for the

conservation of energy is obtained selecting ν = 0. One gets:
ȧ
ρ̇ + 3 (ρ + p) = 0
a
where the dot indicates a derivative by the cosmological time t.
• An equation of state relates the pressure p and the energy density ρ:

p = ωρ where ω = constant

• The conservation of energy leads to:

ρ̇ ȧ
= −3 (ω + 1) ⇒ ρ ∝ a−3(ω+1)
ρ a
Examples:

25
 – Dust: collisionless and non-relativistic matter ⇒ no pressure: p = 0.
Then ω = 0 and:
ρM ∝ a−3
The number density of particles decreases as the universe expands.
– Radiation: either proper EM radiation or particles moving with ve-
locity close to the speed of light. One has pR = 13 ρR and:

ρM ∝ a−4

The energy density of radiation falls faster than the energy density
of dust.
(vac)
– Vacuum: represented by a vacuum stress-energy tensor Tµν =
Λ
− 8πG gµν . Is like a perfect fluid with ω = −1, from which we get:

ρ(vac) = constant

the vacuum energy density is independent on size and shape of the

universe.
• If ρM prevails over the others: universe is matter dominated. If ρR pre-
vails, universe is radiation dominated. If ρ(vac) prevails: universe is vac-
uum dominated.

Einstein equations
Neglecting the cosmological constant term, the Einstein equations become a
pair of coupled equations determining the evolution of the scale factor:
ä 4πG
 ȧ 2 8πG K
=− ρ + 3p = ρ− 2
a 3 a 3 a
These equations are called Friedmann equations. The Robertson-Walker metric
whose scale factor is a solution of these equations is called Friedmann-Robertson-
Walker metric.

Cosmological parameters
• Hubble parameter: defines the expansion rate of the universe:

ȧ(t)
H(t) =
a(t)

• Hubble constant H0 is the Hubble parameter at present time: 70 km/sec/Mpc

< H0 < 100 km/sec/Mpc.
• Hubble lenght: dH = H0 c−1 ≈ 3 × 103 h−1 M pc;

26
 • Hubble time: tH = H0−1 = 9.78 × 109 h−1 yr.
• Deceleration parameter: measures the rate of change of the expansion
velocity:
aä
q=− 2
ȧ
• Density parameter:
8πG ρ
Ω= 2
ρ=
3H ρc
3H 2
where the critical density ρc = 8πG determines the sign of K, i.e., uni-
verse’s geometry:
– Ω < 1 (⇔ ρ < ρc ) ⇔ K < 0 (open universe);
– Ω = 1 (⇔ ρ = ρc ) ⇔ K = 0 (flat universe);
– Ω > 1 (⇔ ρ > ρc ) ⇔ K > 0 (closed universe);

Hubble Law
• Obtained experimentally, relates the recessing velocity of distant galaxies
to their distance:
v = H0 d
• Can be obtained directly from the Friedmann-Robertson-Walker metric
by computing the radial distance between distant galaxies.

Scale Factor evolution

Obtained by solving the Friedmann-Robertson-Walker equations once they are
given the spatial curvature K and the energy densities ρR , ρM , and ρ(vac) .

• If ρ > 0 AND p ≥ 0, and Λ = 0 (energy or matter dominated universe),

we get:
ä < 0
The universe is decelerating, and we know from observations that the
universe is expanding. It was expanding faster in the past, and we can
trace this expansion back to the singular state at t = 0: big bang. In
particular
– if K ≤ 0 (flat and open cases):
8πG 2
ȧ2 = ρa + |K|
3
The right hand side is greater than zero. ȧ has never passed through
the zero, and it must have been positive all the time: open and flat
universes expand forever. But
a(t)→∞
ȧ2 −−−−−→ |K|

27
 thus
∗ K = −1: evolution approaches the limiting value ȧ = 1;
∗ K = 0: universe keeps expanding but more and more slowly.
– K = +1 (closed universe). We get:

8πG 2
ȧ2 = ρa − 1
3
For ȧ2 not to become negative, the universe cannot expand forever,
but a(t) must have an upper bound avac . As a approaches avac , ä is
always negative: the universe keeps on contracting. Closed universes
are closed in time too.
• The precise form of a(t) changes accordingly if the universe is matter or
radiation dominated:
√
– Radiation dominated: a(t) ∝ t.
2
– Matter dominated: a(t) ∝ t 3 .
• Empty universes (only vacuum energy): have either p < 0 or ρ < 0, and
previous considerations don’t hold anymore. We can distinguish:

– Λ < 0: it exists only an open (K = −1) solution with a(t) ∝ eHt .

This is called anti de Sitter universe.
– Λ > 0: they exist solutions for K = −1, K = 0, and K = +1, all
called de Sitter universes.

28