You are on page 1of 24

Introduction

1 Quantum gravity, who needs it?!


1.1 Unification
We believe that we live in a world that it is fundamentally quantum mechanical.
A belief that is supported by the consistent mathematical formalism of the theory,
and by thousands of experiments that agree with the theory. Before the advent of
quantum mechanics, we know that much of the world could be seen through the eye
of what we call Classical physics. In the heart of classical physics lies the two great
achievements of the founding fathers; Isaac Newtons laws of mechanics and gravity,
and Maxwells theory of electromagnetism.

Using Newtons theory, we can describe many phenomena that we observe here on
Earth in our everyday lives. Even more, we are able to explain the mechanics of
the solar system and even the galaxies beyond just by using F = ma. On the other
hand, electromagnetism came to explain the mysterious force that makes some ob-
jects attract or rebel each other through electric charges or magnets. It concluded
its victories by describing light itself as an electromagnetic phenomenon, and uni-
fying the electric and magnetic phenomena in the astonishingly succinct 4 laws of
Maxwells.

Early in the twentieth century, we were faced with the incompleteness of classical
physics. a set of different experiments were in a striking disagreement with the clas-
sical theory; for example the black body radiation and the photoelectric effect. So,
the physicists of the last century had the task to reformulate the laws of classical
physics to explain these experimental results. That quest led to the invention of
quantum mechanics which succeeded very well to justify the experiments. However,
that success came with a price for the theory appeared to imply a picture of reality
that is so counterintuitive such as the uncertainty principle and the probabilistic
interpretation.

This great work of physicists like Dirac and Feynman led to the theory of quantum
electrodynamics (QED); a theory that replaces the old classical electromagnetism
with its quantum version. This theory has achieved a ridiculous agreement with
experiment by ten parts in a billion 108 . Quantum mechanics also came with its
forces, the strong and the weak nuclear forces which live within the range of a nu-
cleus radius, that is why we did not notice them before, describe why atoms are
stable and why some atoms decay respectively. The strong and the weak forces are
pretty much explained by another theory we call quantum chromodynamics (QCD).

1
In fact, the theories of quantum electrodynamics and quantum chromodynamics did
not only replace classical physics, but they also reconcile quantum mechanics with
another tenet of physics, the special theory of relativity.

I think, by now, you are beginning to see what we are coming into. There is some-
thing missing in the picture. So, where is gravity?! Gravity is like electromagnetism
a classical theory that does not respect the laws of quantum mechanics, however,
we managed to get only a quantum theory for electromagnetism. And, of course,
gravity does not arise naturally in quantum mechanics like the strong and the weak
forces. The search for a quantum theory of gravity is the final piece in a picture
that seems to strive for unification. Should we came up with a theory for quantum
gravity, then we would have within our hands the theory of everything.

1.2 Clash of the titans


The appeal for the unification of the laws of physics or the mathematical beauty
may sound not so stringent requirement, and that is what I think for we should
describe nature the way it is not the way we want it to be. However, it is not only
the aesthetics that drive physicists to search for quantum gravity; and that what I
am going to explain below. First, I should tell you that by gravity we do not mean
Newtons theory of gravity
Gm1 m2
F =
r2
but we speak about the more accurate description of gravity given by the Einsteins
theory of general relativity. Note that I will give a brief account for the general
theory of relativity for the readers who are unfamiliar with the theory.

The fundamental equation in general relativity is


1
R Rg = T g
2
Qualitatively, we can say that the left hand side describes the curvature of space-
time, we will define what we mean by the space-time curvature later. On the right
hand side, there is the energy-momentum tensor of matter or fields, and the term
with the in it describes the vacuum energy of space-time itself. The first problem
is now apparent, for the true nature of the energy and momentum of matter and
fields is described by quantum field theory, on the other hand, the left hand side
is obviously a classical quantity. Simply, the math does not add up, for the energy
momentum tensor is an operator in a quantum field theory and the curvature is a
number; this is a mathematical absurdity.

As a first way to fix this problem is to replace the right hand side with the expec-
tation value of the energy-momentum tensor
1
R Rg = h| T |i g
2
. Now, we have restored the mathematical sense, but that does not solve the problem
entirely as it remains an approximation for a more fundamental quantum theory of

2
gravity, such an approximation is called semi classical gravity. The more alarming
issue is that this equation seems to be in conflict with quantum mechanics. For it
spoils a fundamental feature of quantum mechanics which is linearity. This happens
because the Schrdinger equation for the wave function now depends on the metric
which in turn depends on the wave function through the modified field equations.
What do nonlinearities in quantum mechanics do anyway?! Well, it spoils one of
the most fundamental principle of quantum mechanics; the superposition principle.

It is, in fact, the superposition principle that may suggests the need for a quantum
theory of gravity. In the 1957 Chapel Hill Conference on gravity, Richard Feynman
gave the following thought experiment. He considers a SternGerlach experiment in
which two spin-1/2 particles are put into a superposition of spin up and spin down
and is guided to two counters. These counters are connected to a ball of macroscopic
dimensions. The superposition of the particles is now allegedly transferred to a
superposition of the ball being simultaneously at two positions.But this means that
the balls gravitational field is also in a superposition. I will let Feynman speaks for
himself, Now, how do we analyze this experiment according to quantum mechanics?
We have an amplitude that the ball is up, and an amplitude that the ball is down.
That is, we have an amplitude (from a wave function) that the spin of the electron
in the first part of the equipment is either up or down. And if we imagine that the
ball can be analyzed through the interconnections up to this dimension 1cm by
the quantum mechanics, then before we make an observation we still have to give
an amplitude that the ball is up and an amplitude that the ball is down. Now, since
the ball is big enough to produce a real gravitational field . . .we could use that
gravitational field to move another ball, and amplify that, and use the connections to
the second ball as the measuring equipment. We would then have to analyze through
the channel provided by the gravitational field itself via the quantum mechanical
amplitudes.

1.3 When a theory crashes


Physicists come up with models that they claim they can pretty well explain the
experimental observations related to that phenomenon. So, when do we realize that
these models are gone astray and not any more describe reality? Well, an obvious
way is when the models just do not give the correct answers. A more drastic way
for the crash of the theory is the existence of singularity, i.e the theory predicts
quantities with arbitrary large values like infinite density or pressure. The existence
of singularities in a theory implies the domain where the theory fails and calls for
another model that would get red of these singularities.

We face this scenario in general relativity. It has been proven by the British math-
ematician, Roger Penrose, that the general theory of relativity inevitably predicts
the existence of singularities. As an example for these singularities is the interior of
a black hole and the very beginning of the universe in the Big Bang theory. These
regions are characterized by infinite curvature and density which is not a physical

3
situation.

The inevitability of singularities in general relativity seems to motivate the need for
a more encompassing theory that shares the original successes of general relativity
and furthermore could get rid of these pesky unphysical singularities. We strongly
believe that this more encompassing theory would need quantum mechanics as the
interior of black holes is a region characterized by small dimensions which is the
domain where all the quantum games live. That is the third motivation for the
theory of quantum gravity.

4
Symmetry in general relativity

It has been known that the study of symmetry in physics is of a great value. One can
know much about the form of the solution without actually solving the equations
by using symmetry arguments. General relativity is no exception, in fact the study
of symmetry in GR is very important; as the field equations are highly nonlinear, so
we have little hope of finding exact solutions. Here symmetry comes to the rescue.

We need to be careful in defining symmetry on a curved manifold. As usual, we


would need to find a way to describe the symmetry in a covariant way that does
not depend on a particular choice of the coordinate system.

1.4 Killing Vectors


The general idea of symmetry is a transformation that operates on an object that
leaves the object invariant. In other words, you cannot tell, after the act of trans-
forming, whether it happened or not. Of course in GR we would be interested in
the symmetries of the metric, and we define them as the transformations that leave
the metric form-invariant.
0
A tensor T (x) is said to be form invariant under a given transformation x x if
0 0 0
the transformed tensor T (x ) is the same function of its argument x as the original
tensor. So, applied to the metric, we have for all y
0
g (y) = g (y) (1)

Since g is a second-rank tensor, we know that:

0 0 x x
g (x ) = g (x) (2)
x0 x0
Or we can write the inverse transform as:
0 0
x x 0 0
g (x) = g (x ) (3)
x x
0 0 0
If the metric is form-invariant, then we have g (x ) = g (x ). So the condition of
isometry is
0 0
x x 0
g (x) = g (x ) (4)
x x
These are a set of partial differential equations for the metric components, and it
is very complicated to work with. So, we are going to focus on the transformations
that can be generated infinitesimally, a wisdom that has been acquired through
physics encounter with symmetry.

5
We write the infinitesimal transformation for || << 1,
0
x = x +  (x)

By substituting back in (4), we have:


0
x
= +  (5)
x x
And,
0
x
= +  (6)
x x
Also, we expand the metric in a Taylor series as,
g
g (x + ) = g (x) +  (7)
x
By performing the product of (5), (6) and (7), and keeping only terms of first order
in , we get after some minor manipulations,

g (x)

g (x) +
g (x) + (x) =0 (8)
x x x
We can write this equation in terms of the covariant derivative of the covariant
vector = g by noting that we can write (8) as,

g g g
+ + ( )=0
x x x x x
Recall the form of the Christoffel symbol,
1 g g g
= g ( )
2 x x x
So, the above equation can be written as:


+ 2 = 0
x x
And in terms of the covariant derivatives this simplifies to:

+ = 0 (9)

This is known as the Killing equation, and we refer to as the Killing vectors.
Dont worry no vectors are killed in the process, they are just named after their
discoverer Wilhelm Killing. Now, we are able to describe all the infinitesimal sym-
metries of the metric through its Killing vectors. Note that any linear combination,
with constant coefficients, of the Killing vectors does also satisfy the Killing equa-
tion, and hence they generate a symmetry for the metric.

6
1.5 Consequences of Symmetries
We know from our experience in classical or quantum mechanics that the existence
of symmetries implies the existence of conserved quantities. For example, the in-
variance of the system under translations in time implies the conservation of energy.
We should expect a similar scenario in GR where symmetries lead to conservation.

Recall the geodesic equation of a massive particle parametrized by its proper time
:
d2 x
dx dx

+ =0 (10)
d 2 d d
We can show that, by direct substitution, that we can write the geodesic equation

in terms of the 4-momentum P = m dx d
as:

P P = 0 (11)

But, since g = 0, we can write (11) in an equivalent form

P P = 0

by expanding the covariant derivative of P , we get:

P P P P = 0 (12)

The first term P P can be written as:

dx P dP
P P = m = m
d x d
The second term, however, can be written as:
1
P P = ( g + g g )P g P
2
1
= ( g + g g )P P
2
The symmetry of the product P P under the exchange of the two indices and
leads to the cancellation of the first and third term in the bracket, leaving us with:
1
P P = ( g )P P
2
From (12), we see that the change of momentum along a geodesic is given by:

dP 1
= ( g )P P (13)
d 2m
Now, see what happens when the metric is independent of a specific coordinate x
i.e g = 0 in equation (13) this leads to dP
d

= 0. The momentum component in
the direction of that specific coordinate is constant along the geodesic.

7
The most obvious symmetry of the metric is the one when the metric does not
depend on the coordinate, for example, look at the flat space-time metric

ds2 = dt2 + dx2 + dy 2 + dz 2

It is independent of all the coordinates (t, x, y, z) and this leads directly to the
conservation of energy and momentum of the free particle as expected. In the
language of Killing equation, we should be able to find 4 independent Killing vectors
that generate the symmetry. These are simply

(n) = n

where n takes the values 0,1,2 and 3.

1.6 Number of independent killing vectors


A set of Killing vectors is said to be dependent if they satisfy a linear relation of
the form X
cn n = 0
n

If and only if all the constant cn coefficients vanish, then they are independent. You
might think that since we are in N dimensional space, then the maximum number of
independent vectors is N, hence, there are N independent Killing vectors, but this
is wrong! The reason is that although in any N dimensional space any more than N
vectors would satisfy a linear relation of the above form, the cn coefficients would
not be all constant, and they would not form a set of independent Killing vectors
as required.

There is a quick way to figure out the maximum number of independent Killing
vectors in any N dimensional manifold. It is, however, not the most rigorous, but
in physics we love the quickest routes to the problem not the most rigorous.

We know form our intuition that the good old flat Euclidean space Rn is as symmet-
ric as it gets. symmetries of the Euclidean space, from the perspective of a point P ,
can be grouped to translations and rotations. More specifically, translations move
the point P around, and we know that there are n independent directions in which
the point can move into. For the rotations around P that leave the geometry invari-
ant, we can think of rotating the axis going through P toward the other n 1 axes.
In fact, we have n axes that each could be rotated to the other n1 axis, so in total,
we have n(n 1) ways. However, rotating x to y is the same as rotating y toward
x, so we should divide the result by 2, and we have a total of n(n1)
2
independent
rotations. The total symmetries of the space is henceforth,
n(n 1) n(n + 1)
n+ =
2 2
Although our counting refers to flat Euclidean space, we can actually extend the
argument easily to curved manifolds. The trick is that all the above transforma-
tions is applied locally in the neighborhood of the point P ; nowhere in our argument

8
did we appeal to the global nature of the space under consideration. That means,
in general, the maximum number of independent Killing vectors of N dimensional
space is N (N2+1) .

1.7 Maximally Symmetric Spaces


Spaces which admits the maximum number of independent Killing vectors are said
to be maximally symmetric. An obvious example is the pre-mentioned flat Eu-
clidean space Rn , and the flat Minkowski space-time. For the curved manifolds, N
dimensional spheres Sn are the most familiar maximally symmetric curved spaces.

Maximally symmetric spaces enjoy a punch of interesting properties that make them
worth of study. The invariance of the geometry under translations and rotations
makes it natural to think that the curvature (specified by the curvature tensor R )
is the same everywhere, and it suffices to compute the curvature around some point
in order to find it everywhere.

How to quantify these statements?! The idea is to choose locally inertial coordinates
at a point P so that g = . We know that by doing a Lorentz transformations
at the point P , the metric components do not change. We want the curvature
tensor R to have this property in a maximally symmetric space. So, we look for
the tensors that their components do not change under a Lorentz transformation,
and we demand the curvature tensor R to be proportional to these tensors. We
look into our toolkit to find tensors that have this property, we find the metric, the
Levi-Ceveta tensor and the Kronecker delta. However, to capture the symmetries
of the Curvature tensor, we find a unique possibility:
R g g g g
This relation is valid in any coordinate system by the virtue of it is a tensorial
relation. It is also valid everywhere in the manifold by the virtue of the maximally
symmetric nature of the manifold.

It remains to fix the proportionality constant in the above relation. First contract
with g :
g R = C(g g g g g g )
Since, g R = R , and g g = n, and g g g = g = g , then we have:
R = C(n 1)g
Finally, contract with g .
R
C=
n(n 1)
Then, in a maximally symmetric space the curvature tensor is defined everywhere
by the relation:
R
R = (g g g g ) (14)
n(n 1)

9
The converse is also true, for if we manage to write the curvature tensor as it is in the
above relation, then the space is maximally symmetric. We can see for this relation
how every flat manifold such as: Euclidean space or Minkowsky space-time is , in
fact, maximally symmetric as it automatically satisfies the above relation with R =
0. We can see that the above relation specifies the different categories of maximally
symmetric spaces in terms of R. R could be zero, positive or negative, which
refers to flat, positively curved or negatively curved spaces respectively.Positively
curved maximally symmetric spaces are called De Sitter spaces, in recognition of
the mathematical physicist Willem de Sitter who studied these spaces. Negatively
curved maximally symmetric spaces are named unimaginably Anti de Sitter spaces.

10
Basic features of 3-d gravity

We live in a 4-dimensional world, 3 space dimensions and 1 time dimension. So,


why do we need to study a clearly nonphysical model of the world?! We certainly
do not live in flat-land. It turns, however, that physicists have been resorting to
lower dimensional models in different areas of physics. Lower dimensional models
serves as an arena where the physical concepts are clear and easy to grasp without
the mathematical complications that block the vision in the realistic case. For ex-
ample, in quantum mechanics, we easily can see the quantization of energy and the
tunneling process in one dimensional setting. Another example now in statistical
physics, is the Ising model, which is a two dimensional model that exhibits phase
transitions.

As we know, physicists have quite a bad time trying to quantize gravity in the
4-dimensional world. This lead some physicists to think why not study gravity in
lower dimensions? This certainly makes the math easier, for we now have less com-
ponents of the metric g or the curvature R tensors. It is not just the relative
mathematical simplicity that we gain by studying lower dimensional gravity, but we
also could find an insight to help us go our way through the original problem. And
that exactly what happened, for we could actually write candidates for a quantum
theory of gravity in 3-d.

We should, however, admit the difference between the 4-dimensional and 3-dimensional
gravity, for simplicity comes with a price. For example, in 4 dimensions, the effect
of gravity can propagate even in empty space by the newly experimentally discov-
ered gravitational waves, the theory of gravitational waves was established 100 years
ago! On the other hand, the 3-dimensional model does not allow for gravitational
waves. Even more, we can show that in 3-dimensional gravity every solution of the
vacuum Einsteins equation is actually flat. This means that we can never obtain
the most important solution of Einsteins equation in 4-d, the black hole solution.
Surprisingly, this problem could be solved by introducing the cosmological constant
in Einsteins equation, and in 1992 Baados, Teitelboim and Zanelli discovered the
BTZ black hole solution in 3-d space-time with a negative cosmological constant.

In what follows, I will show why there cannot be gravitational waves in 3-dimensional
gravity, and it will be evident that the 3-dimensional gravity is a local theory. After
that, I will return to the claim that in empty space the geometry is plain flat in
3-d space-time. It turns, however, that this is not true! The problem is that in flat
space-time, we should be able to find coordinates in which the metric g =
everywhere. It turns out that the transformation that ought to do this is not gen-
erally defined everywhere, so the region as the a whole is not entirely identical to
the flat Minkowski space-time. Although the 3-dimensional theory does not admit
propagative degrees of freedom (gravitational waves), nontrivial global effects can
arise within the theory.

11
As an example for the nontrivial global effects of the theory, I will solve the Ein-
steins equation for a point source in 3-dimensional gravity. It will be shown that
outside the source, the space-time is not that of flat Minkowski space time, but it
is a cone with the particle on its tip.

Returning now to the disappearance of gravitational waves in 3-dimensional gravity,


I will first need to introduce the tensor, known as the Weyl tensor C , that is
responsible for the existence of gravitational waves in 4-dimensional gravity. That
is being done by counting the independent components of the Riemannian tensor
R , and noticing that not all of the components are determined by the local mat-
ter and energy distribution given by the energy-momentum tensor T . So, roughly
speaking, the Weyl tensor accounts for the parts of the Riemannian tensor that
are not constrained by the local distribution of matter and energy. Then, we will
show that the Weyl tensor vanishes identically in 3-dimensional gravity proving the
nonexistence of gravitational waves in the theory.

2 Introducing the Weyl tensor


2.1 Counting the number of independent components of the
curvature tensor R
In general N dimensional manifold, a fourth rank tensor T should have a total
of N N N N = N 4 independent components. However, the Reimannian
tensor R enjoys a number of symmetries that reduce the generic N 4 components
drastically.

So, what are the symmetries of the Riemannian tensor? First it is defined by:

R = + (15)
Where the Christoffel abc symbol is defined by:
1
abc = g ad (b gdc + c gdb d gbc )
2

It is easy to see form the definition that R is antisymmetric on its last pair of
indices

R = R (16)
The antisymmetry in the last 2 indices means that there are only N (N21) independent
components they could have instead of N 2 , so the N 4 independent components of the
3
Riemannian tensor is reduced to N (N2 1) . However, this is not the end of the story
for there are still additional symmetries that we should dig for. These symmetries

are best shown by dealing with the contracted Riemanniann tensor R = g R .

It could be shown that the contracted Riemannian tensor satisfies the following
equations:
R = R (17)

12
R = R (18)
R = R (19)
Also with the cyclic identity

R + R + R = 0 (20)

So, lets start counting! First, the antisymmetry in the first and the last two indices,
equations (3) and (4) respectively, means that there are a total of N (N21) indepen-
dent components of each pair. What can we make of the symmetry in the first and
last pair of indices?!

Well, we can think of R as an m m matrix, where the pair () counts as a


first index and the pair () counts as a second index. And, now the question shifts
to asking about the number of independent components of an m m matrix. this
number is given by:
m2 m m(m + 1)
m+ =
2 2
By substituting by N (N21) as m in the above equation, we get the number of inde-
pendent components till now:
1 N (N 1) N (N 1)
( + 1) (21)
2 2 2
What is left is to deal with the remaining symmetry of the cyclic equation (6).
This equation reduces the number of independent components by N (N 1)(N4!2)(N 3) .
Finally, the total number of independent components of the Riemannian tensor is
given by:
1 N (N 1) N (N 1) N (N 1)(N 2)(N 3) 1
( + 1) = N 2 (N 2 1) (22)
2 2 2 4! 12
so, for N = 4, we have 20 independent components instead of 256, and for N=3, we
have 6 components.

2.2 The decomposition of the Riemannian tensor


Recall that we can get tensors of lower rank form the Riemannian tensorR by
contracting with the metric tensor g . The symmetries of the Riemannian tensor,
however, restrict our options. Indeed, there is a unique possibility which is the Ricci
tensor given by:
R = g R (23)
We say that the Ricci tensor is the trace of the Riemannian tensor. We can further
contract the Ricci tensor to get the Ricci scalar given by:

R = g R (24)

This leads us to ask is there a way to decompose the Riemannian tensor into two
parts, one that is trace-full, and another traceless part? so that when we

13
contract with the metric we get the Ricci tensor from the trace-full part. So,
what are the properties of the traceless part? Well, besides being a fourth rank
tensor that gives vanishing tensors under any possible contraction with the metric,
it should also be endowed with all the symmetries of the Riemannian tensor. Well,
quite a lot to accomplish!

So, lets dig for our tensor which we give the symbol C . naturally, we will try to
construct it with tensors known to us, namely, R , R and the metric g , and
possibly with the Ricci scalar R.

To get a fourth-rank tensor from R , we may think of the product g R . Now


to get the antisymmetry in the first pair (), remember we need to retrieve the
symmetric properties of the Riemannian tensor, we think of the combination:
g R g R (25)
. To get the antisymmetry in the last pair (), we interchange the positions of
and in (11) to get
g R g R (26)
Then we subtract (12) from (11)
g R g R g R + g R (27)
You can easily check that equation (13) is antisymmetric in ( and ), and in (
and ) as required. As a bonus, equation (13) is also symmetric in the exchange of
the pair () with () as you can easily verify.

We can also construct a combination using the metric tensor which gives that same
symmetry properties of the Riemannian tensor, and we can insert the Ricci scalar
here, too.
R(g g g g ) (28)
Now, we can write C as:
C = R + a(g R g R g R + g R ) + bR(g g g g ) (29)
Where, a and b are constants yet to be determined.

We can determine the above constants using the last requirement we impose on
C which is that:
g C = 0 (30)
From equation (16), we get:
R + a(N R R R + g R) + bR(N g g ) = 0 (31)
I have used g g = N and R = g R .

Collecting the terms in (17), reach


R (1 + aN 2a) + Rg (bN b + a) = 0 (32)

14
But, the metric tensor g and the Ricci tensor R are independent, and we demand
the vanishing of both the brackets in equation (18), which gives us:
1
a= (33)
N 2
1
b= (34)
(N 1)(N 2)
We can write now the complete formula of the C tensor

1 R
C = R (g R g R g R +g R )+ (g g g g )
N 2 (N 1)(N 2)
(35)

C is known as the Weyl tensor, it has the same symmetry properties of the
Riemannian tensor and it is traceless by construction. We can think of equation
(21) in another way; it gives a decomposition of the Riemannian tensor in terms of
a trace-full part given by the Ricci tensor and the Ricci scalar, and a traceless
part given by the Weyl tensor.

2.3 Einsteins field equation and the Weyl tensor


Recall the form of Einsteins field equation
1
R Rg = T (36)
2
We see that the Ricci tensor R is determined by the local matter and energy
distribution T , and in source free regions, the Ricci tensor R vanishes. A look
at equation (21) reveals that when R = 0, we have

R = C (37)

Although, we are in empty space (no matter or energy is around), the curvature
tensor does not vanish, and it is , in fact, determined completely by the Weyl tensor
there.

Here, our previous counting for the independent components of the Riemannian
tensor proves its value. As we have seen, in 4-dimensional space-time the number
of independent components of the Riemannian tensor is equal to 20. We know that
the Ricci tensor is a symmetric tensor, so it has a total of 4(4+1)
2
= 10 independent
components. So, 10 components of the curvature tensor is governed by the 10 Ricci
tensor components, and the other 10 components is given by the Weyl tensor. In
3-dimensional space-time, however, there are only 6 independent components of
the Riemannian tensor, and for the Ricci tensor there are 3(3+1) 2
= 6 independent
components, too! This gives us a hint that the Weyl tensor would vanish in a theory
of gravity in 3-dimensional space-time.

15
The connection between gravitational waves and the Weyl tensor is now evident.As
the Einstein equation does not let us reconstruct the complete curvature tensor. It
only gives us ten pieces of information, although we need 20 to specify the curvature
tensor completely. There is a similar story in electromagnetism. If we know the
charge and current distributions everywhere in space, you might think that that
would let us find out the electric and magnetic fields everywhere. But, this is not
true! There is extra information in the fields which the sources of the fields does not
know any thing about. After all, this is what all the electromagnetic waves thing
about. they do not need any source, a wave coming form infinity can still change
the fields.

I hope that it is clear by now the part played by the Weyl tensor in the story
of Einsteins gravity, we have reached to the conclusion that the Weyl tensor is
responsible for effects on the space-time curvature, and hence the movement of free
particles for regions that are free of sources. Gravitational waves are one type of
these effects that do not need sources to be brought up by. Should we claim the
absence of gravitational waves form the theory of gravity in 3-dimensional space
time, then we had better prove the vanishing of the Weyl tensor in such a theory.

2.4 The Weyl tensor in 3-d gravity


Recall the fact that the number of independent components of the Riemannian
tensor is equal to those of the Ricci tensor, both are equal to 6. This may suggest
that we can write the Riemannian tensor in terms of the Ricci tensor. Returning to
equation (21), we see that if we could write the Riemannian tensor R in terms
of the quantity
1 R
(g R g R g R + g R ) (g g g g )
N 2 (N 1)(N 2)

then, we could prove that the Weyl tensor C vanishes identically. Our task now
is to prove that we can write the curvature tensor in 3-d as:
R
R = g R g R g R + g R (g g g g ) (38)
2
First, lets choose a coordinate system in which g is diagonal, i.e g = 0 for 6=
at some point P . Note that we can always do this step as the metric is a symmetric
matrix, and we can always diagonalize it via a coordinate transformation. The 6
independent components of the Riemannian tensor are:

R1323 (39)

R1212 (40)
R1223 (41)
R1213 (42)
R2323 (43)

16
R3131 (44)
For the first component (25), we know that
1
R12 = g 33 R1323 = R1323 (45)
g33
which means that
R1323 = g33 R12 (46)
This, indeed, agrees with equation (24) For the second component (26) we know
that
R11 = g 22 R1212 + g 33 R1313 (47)
R22 = g 33 R2323 + g 11 R1212 (48)
g22 R11 + g11 R22 = 2R1212 + g 33 (g22 R1313 + g11 R2323 ) (49)
Or, we can write it as

g22 R11 + g11 R22 = R1212 + g11 g22 (g 11 g 22 R1212 + g 11 g 33 R1313 + g 22 g 33 R2323 ) (50)

Solving for R1212


R
R1212 = g22 R11 + g11 R22
g11 g22 (51)
2
Which again satisfies equation (24). The other 4 components of equations (27) to
(30) can be obtained as a permutations of the 1,2,3 indices form R1212 and R1323 ,
and they also satisfy equation (24).

Although, We have managed to prove equation (24) for a special coordinate system
in which the metric is diagonal, it is written in a covariant form, so it is true in
general. By proving that we can write the Riemannian tensor in terms of the Ricci
tensor as in equation (24) in 3-dimensional space-time, we prove that the Weyl ten-
sor vanishes identically in this theory.

We have seen that, in 3-dimensional space-time, the curvature tensor is fully speci-
fied by the Ricci tensor as in equation (24). This implies that in regions where there
are no sources around this happens

R = 0 = R = 0 (52)

We know from general relativity that if R = 0, then we can find a coordinate


transformation that makes the metric Minkowski flat everywhere

R = 0 = g = (53)

That makes gravity in 3 dimensions seems too trivial in contrast with its nature in
4 dimensions. As for the 4 dimensions case,

R = 0 6 R = 0 (54)

This makes it possible for gravity in 4 dimensions to admit the peculiar solutions of
the famous black holes.

17
So, in order to have any hope of obtaining nontrivial results from the 3 dimensional
theory, then we better work with the modified Einsteins field equation that contains
the cosmological constant
1
R + g Rg = T (55)
2
So, if we put T = 0 in equation (41), and then contracting by g , we get a relation
between the cosmological constant and the Ricci scalar R

R = 6 (56)

And, now we can write the Ricci tensor in terms of the metric tensor as
R
R = g (57)
3
Returning to equation (24), after substituting by equation (43) we get

R
R = (g g g g ) (58)
6
Or in terms of the cosmological constant

R = (g g g g ) (59)

So, even after introducing the cosmological constant, we found that the space is
having a constant curvature for regions that do not include any sources. That is
not exactly what happens in 4-dimensions for the black hole curvature is varying
through space, and its curvature is singular at the origin. Nevertheless, we were
able to cook up a solution for Einsteins equation in 3 dimensions which is closely
related to the 4-dimensional one except for some subtle differences including the
disappearance of the singularity.

18
Point particle in 3-d gravity

We have shown that Wyel tensor vanishes identically in 3-d gravity; which means
that the theory does not admit propagative degrees of freedom i.e, there are no
gravitational waves. However, that does not render gravity in 3-d trivial as we will
see that there are solutions with interesting dynamics. As a simple example, let
us investigate the space-time around a massive point particle that is stationary at
some point.

A note about dimensions: the combination Gm c2


has the dimension of a length. So,
to simplify things I am going to work with a system of units in which c = 1 and
the length is dimensionless. Now, the gravitational constant has the dimension of
an inverse mass.

Einsteins field equation with zero cosmological constant:


1 8G
R Rg = 4 T (60)
2 c
Since we are dealing with a particle at rest, all the energy momentum tensor com-
ponents vanish except for
T00 = mc2 2 (~r r~0 ) (61)
Furthermore, we can write the delta function in terms of the magnitudes of the
vectors ~r and r~0 First, in spherical coordinates we have:
1
2 (~r r~0 ) = (r r0 )( 0 ) (62)
r
Now, in our situation, we have azimuthal symmetry; so we could project out the
part of the delta function by just put a factor of 2 in the denominator. We end up
with
1
2 (~r r~0 ) = (r r0 ) (63)
2r

Lets now turn to our main goal which is to solve the field equations. As Schwarzschild
taught us, a lot of the solution should be known by appealing to symmetry. First,
the metric should not depend on time (static) as the particle is at rest; there is
no dynamics whatsoever. Secondly, the metric should be spherically symmetric as
every direction is on equal footing with any other direction in space. so, we can
write the Ansatz:
ds2 = A(r)dt2 + B(r)dr2 + r2 d2 (64)
We could also make a redefinition of the time coordinate, and the metric simplifies
to
ds2 = dt2 + B(r)dr2 + r2 d2 (65)

19
Now, we get back to Einsteins equations which is nontrivial only for the (00) com-
ponent.
1
R00 Rg00 = 8GT00 (66)
2
Inserting the energy momentum tensor;
1 (r r0 )
R00 Rg00 = 4Gm (67)
2 r
We just need to calculate the (00) component of the Einsteins tensor by using the
Mathematica package, and we end up with this differential equation

B 2 dB = 8Gm(r r0 )dr (68)

Integrating both sides:


Z r0 Z r0
2
B dB = 8Gm(r r0 )dr

We demand that away from the particle we should restore the metric of flat space-
time, so B() = 1
We end up with
1
+ 1 = 8Gm
B(r0 )
Or
B = (1 8Gm)1
Note that when m = 0 (there is no particle) we get B = 1 and we restore the flat
metric as expected. By substituting back in our Ansatz, we get:

ds2 = dt2 + (1 8Gm)1 dr2 + r2 d2 (69)

By making a redefinition of r so that


1
re = (1 8Gm) 2 r (70)

we get:
ds2 = dt2 + dr2 + (1 8Gm)r2 d2 (71)
This looks just like a flat space-time only for this factor 1 8Gm. We could fur-
thermore redefine the angle so that

e = 1 8Gm (72)

we can expand the square root to examine the neighborhood of the particle, so we
keep only the first order in Gm.

e = (1 4Gm)

20
Remember that Gm is a dimensionless parameter, so it makes sense to speak of its
order. Now, equation (12) becomes

ds2 = dt2 + dr2 + r2 d2 (73)

And, you might think that we have been wasting our time playing around with a flat
sapce-time metric. However, you are missing a seemingly minor detail that makes
all the difference in the world. For, our new angle does not now cover all of space!
It doesnt have the correct periodicity.
So, Watch closely, remember our new angle is defined by e = (1 4Gm). So,
as completes a cycle from 0 2 , e goes through 0 2 8Gm. We call
= 8Gm the deficit angle which refers to the failure to cover all of space by the
above metric. In fact, this geometry is a one of a cone with the particle on the tip
of it, hence the name conical singularity.

21
Black hole thermodynamics

3 Hawking Temperature of a Schwarzchild black


hole
In 1975, the renowned physicist Steven Hawking could manage to derive the exact
formula of the Black holes temperature and entropy. His method of attack was to
solve a quantum field theory in the vicinity of the black hole event horizon. This,
however, is a messy business to do in our level of study, and instead we are going
to do it in a different way. It was also Hawking who came up with this idea which
to use the path integral approach of quantum mechanics, and use its connection to
the thermal partition function.

We know that in quantum mechanics, the evolution of the wave function is dictated
by the unitary operator
U = eiHt/~ (74)
Where H is the Hamiltonian of the system. So, the amplitude of going from state
|ii to the state |if is given by

Z = hf | eiHt/~ |i i (75)
And we know that the partition function of the system is given by:
X X
Z= eEn = hn| e H |ni = T re H (76)
n n

Now, notice the intriguing similarity between the quantum mechanical probability
amplitude and the thermal partition function. In fact, we can make them identical
if we define an imaginary time in (2) by the relation:
t = i~ (77)
If we now demand every initial state |i i=|ni to evolve in time by returning to itself,
we then see that equation (2) becomes

Z = hn| e H |ni (78)


What we have done is, we have made time imaginary and periodic with a period
given by the inverse temperature .

This nifty trick would allow us to get the Hawking temperature of a Schwrzchild
black hole in a few lines! Remember the Schwarzchild solution is given by the metric:
rs rs
ds2 = (1 )dt2 + (1 )1 dr2 + r2 d2 (79)
r r
22
where d2 is the metric of a 2-d sphere:

d2 = d2 + sin2 d2 (80)

by putting equation t = i in (6), we have:


rs rs
ds2 = (1 )d 2 + (1 )1 dr2 + r2 d2 (81)
r r
Next, we make a change of variables that would make it easy to investigate the near
horizon region, which is our primary concern

r = rs (1 + 2 ) (82)

Then
dr = 2rs d (83)
And,
dr2 = 4rs2 2 d2 (84)
Equation (8) then becomes
1 1
ds2 = (1 )d 2 + 4r2 2 d2 + rs2 (1 + 2 )2 d2 (85)
1+ 2 1 (1 + 2 )1 s

Now, we want to investigate the limit when r rs or, equivalently, the limit when
0. We do this by noting that, when 0, we can say that

(1 + 2 )1 1 2 (86)

So, equation (12) becomes

ds2 = 2 d 2 + 4rs2 d2 + rs2 d2 (87)

Or
2 2 2 1
ds = 4rs2 ( 2
d + d2 + 2 ) (88)
4rs 4
The first two terms in equation (15) is very similar to the flat space metric in polar
coordinates
ds2 = dr2 + r2 d2 (89)
But, its angle is given by

= (90)
2rs
What have been done is that the (3+1)-dimensional space-time has been analytically
continued into a 4-dimensional Euclidean space consisting of a plane attached to its
points a sphere of radius rs .

In order for the flat metric to have the correct periodicity in its angle , we demand
that the period is given by
= 4rs (91)

23
But, referring to equation (4), where we identify the period with the inverse
temperature , then actually equation (18) is an expression for the temperature, so

~ = 4rs (92)

Or
~
= 4rs (93)
kT
Where k is the Boltzmann constant. Recall that the Shwarzchild radius rs is given
by
2GM
rs = (94)
c2
Then, solving for the temperature, we get the celebrated Hawking Temperature

~c3
T = (95)
8GM

24

You might also like