Professional Documents
Culture Documents
Spherical Solutions - Field Eq Einstein - and - Brans-Dicke Gravity
Spherical Solutions - Field Eq Einstein - and - Brans-Dicke Gravity
© 2019
Master of Science
Nel mezzo del cammin di nostra vita
Mi ritrovai per una selva oscura,
Ché la diritta via era smarrita.
Ahi quanto, a dir qual era, è cosa dura,
Questa selva selvaggia ed aspera e forte,
Che nel pensier rinnova la paura!
The uniqueness of the Kottler/Schwarzschild-de Sitter solution (KSdS) of the vacuum Einstein
equations with positive cosmological constant is discussed and certain putative alternatives are shown
to either solve different equations or to be the KSdS solution in disguise. A simultaneous no-hair
and cosmic no-hair theorem for the KSdS geometry in the presence of an imperfect fluid is proved.
Also, we report the explicit form of the general static, spherically symmetric, and asymptotically flat
solution of vacuum Brans-Dicke gravity in the Jordan frame, assuming that the Brans-Dicke scalar
field has no singularities or zeros (except possibly for a central singularity). This general solution
is conformal to the Fisher-Wyman geometry of Einstein theory and its nature depends on a scalar
charge parameter. Apart from the Schwarzschild black hole, only wormhole throats and central naked
singularities are possible.
3
Acknowledgements
It is impossible to express with words the deep appreciation I have towards my supervisor Dr.
Valerio Faraoni. He was the first to give me the opportunity to do research work, and the trust he
has placed in me has no comparison in my life. Also, my deep thanks to Dr. Lorne Nelson and Dr.
Ariel Edery, who together with Dr. Faraoni, have given me the opportunity to advance directly to
the Master in Physics without having to finish the Honours. This support tells me that they have
trusted my capacity, which I always have doubts about. I wish to thank Dr. Sylvain Turcotte for his
humanity as a teacher. As well, without a doubt, an enormous gratitude to Dr. Anca Nedelcescu,
who made it possible for me to see experimental physics with other eyes. Besides, I am grateful to
Dr. Gilles Couture from the Department of Earth and Atmospheric Sciences at UQAM, who one
day, while I was in his office asking him about his Electromagnetism course, told me: "get out of
here and go study physics, the meteorology is not for you, but study it in English!". I was in this way
that I came to study physics at this University. Neither can I forgot two more teachers of my career
in physics from UNSL, Argentina, Dr Amílcar Fasulo and Engineer Héctor Gellón, who believed in
me before anyone else. To all of you, my extreme gratitude!!!
4
Dedications
I dedicate this thesis, very specially, to my husband Eduardo, for his unconditional support with
mathematics since the start of our careers at the University of Buenos Aires (UBA). I also dedicate
it to my son, Leonardo, for his collaboration with coloring the figures and supervising his little sister
Pamina when I was studying or attending my classes at university. I would like to dedicate it also to
the two wonderful persons: Samuel Finkielsztein from Argentina, who passed away at the incredible
age of 92, and was the main engine for my family and I to emigrate to Canada; and to Monique
Ross, retired teacher who for three years has helped me weekly with my English in an absolutely
disinterested way. May this thesis make them all as proud of me as I am thankful to them.
5
Contents
Introduction 1
2 Relativistic Cosmology 34
2.1 Friedmann-Lemaître-Robertson-Walker metric . . . . . . . . . . . . . . . . . . . . . . 36
2.1.1 Friedmann equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1.2 The energy content of the universe . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2 Cosmological solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.2.1 Einstein static universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.2.2 de Sitter space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
6
3.2.1 Abbassi-Meissner proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2.2 McVittie and generalized McVittie solutions . . . . . . . . . . . . . . . . . . . 51
3.2.3 Non-rotating Thakurta solution . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.2.4 Castelo Ferreira metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.3 Simultaneous Baldness and Cosmic Baldness . . . . . . . . . . . . . . . . . . . . . . . 55
4 Scalar-Tensor Theory 59
4.1 Conformal Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.2 Equations of Big Bang and Inflationary Cosmology . . . . . . . . . . . . . . . . . . . 61
4.3 Brans-Dicke Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.3.1 The Jordan Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.3.2 The Equations of Brans-Dicke Cosmology in the Jordan Frame . . . . . . . . 65
4.3.3 The Einstein Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.4 f (R) Theories of Gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Conclusions 76
Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Appendix 79
A 80
A.1 Abbassi Meissner geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
A.2 McVittie geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
A.3 Non-rotating Thakurta geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
A.4 Castelo-Ferreira geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
7
List of Figures
2.1 The distribution of galaxies is clumpy on small scales, but becomes more uniform on
large scales. Ref. [22]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
8
2.2 The spacetime of the universe can be foliated into flat, positively curved, or negatively
curved spatial hypersurfaces. Ref. [22]. . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.3 Cosmic microwave background (CMB). Credits of WMAP. https://map.gsfc.nasa.
gov/media/121238/index.html . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.4 The scale factor for universes dominated by dust with different values of κ: a universe
with critical density (κ = 0, blue curve) corresponds to the static space of Einstein-
De Sitter; the subcritical case (κ = −1, red curve) expands faster and approaches to
Minkowski space; the supercritical case (κ = 1, green curve) expands very fast initially,
but is slowed by the large amount of matter and will collapse in a finite time. Credits
of https://ned.ipac.caltech.edu/level5/March01/Carroll3/Carroll8.html . . 39
2.5 In the Einstein static universe, the spatial sections are 3-spheres S 3 and light turns
around the universe in a finite time. Ref. [18] . . . . . . . . . . . . . . . . . . . . . . 45
2.6 de Sitter space can be described in FLRW coordinates with both flat (κ = 0), spherical
(κ = 1) or hyperboloidal (κ = −1) spatial sections, depending on the coordinate
system chosen. Ref. [18]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
9
Introduction
J.A.Wheeler
Historically, the Principle of General Relativity is much older than the Theory of Relativity, even
older than Newton’s classical mechanics. It was formulated by Galileo Galilei (1564-1642) around
1600, as an argument in the discussion of heliocentrism versus geocentrism.
The advocates of geocentrism argued that if it were true that the Earth moved around the Sun
and around its axis, why did not we notice it? In response to this argument Galileo introduced a
new idea: the inertia. From his multiple experiments with balls rolling on inclined planes, Galileo
concluded that a mass in uniform rectilinear motion will eternally maintain this motion as long as
no external force acts on it.
Galileo himself realized that the inverse formulation is also valid. Imagine an observer moving
with the same velocity as a mass in uniform and rectilinear motion. For the observer, the mass is at
rest, given that, by the principle of inertia, without external influences you will both maintain your
motion, for the observer this mass will remain at rest until an external force acts on it.
Therefore, Galileo’s conclusion was that an observer is not able to determine if he is in a system
that is at rest or in uniform and rectilinear motion. More generally, if a person is locked in a box,
with all the imaginable mechanical experiments at his disposal, he has no way of determining his
state of motion without looking out the window. This principle is called the Principle of Relativity,
and it says so:
1
Introduction
the Einstein equation using said metric, to obtain the Friedmann equations, and some cosmological
solutions, such as the Einstein static universe and the de-Sitter space with a positive cosmological
constant Λ.
In Chapter 3, we proceed to make a detailed analysis about certain articles presented where
alternative solutions to the vacuum Einstein equations are proposed. The Jebsen-Birkhoff theorem
[23, 24] stating that the Schwarzschild geometry is the unique vacuum, spherically symmetric, and
asymptotically flat solution of the Einstein equations is standard textbook material [26].
Almost Birkhoff theorems studying small deviations from spherical symmetry or vacuum have
also been discussed [27, 28]. Relaxing the assumptions of the Jebsen-Birkhoff theorem to allow for
an infinite distribution of matter leads to a variety of inhomogeneous universes [29, 30], which shows
that there is no unique spherical solution with Friedmann-Lemaître-Robertson-Walker asymptotics.
However, it is straightforward to extend the proof of the Jebsen-Birkhoff theorem to vacuum with a
cosmological constant Λ to deduce that the unique spherical solution of the vacuum Einstein equations
in this case is the Kottler/Schwarzschild-de Sitter metric [31] (KSdS) if Λ > 0 and the asymptotics
are de Sitter, or the Schwarzschild-anti-de Sitter metric (SAdS) if Λ < 0 and the asymptotics are
anti-de-Sitter.
More mathematically sophisticated proofs of the uniqueness of the KSdS is contained in old and
recent references [40]. Similar to the situation of the Schwarzschild solution, uniqueness implies that
the KSdS solution is stable with respect to perturbations, the stability being established in Refs.[41]-
[44]. In spite of all this evidence, various works purport the existence of spherical solutions of the
vacuum Einstein equations with Λ > 0 which are alternatives to the KSdS one. This clearly cannot be
true, or else these solutions must reduce to KSdS in disguise. There are also more general solutions of
the Einstein equations representing central inhomogeneities embedded in FLRW spaces, which seem
to reduce to alternatives to the KSdS solution in the special case when the FLRW "background"
reduces to de Sitter. Again, this cannot be the case. Although these other authors presenting these
solutions do not claim that they are alternatives to KSdS, nevertheless a situation was created which
is unclear about the unique status of KSdS.
On the other hand, in Chapter 4, we present the prototypical alternative to general relativity,
Brans-Dicke gravity [70], which adds to the metric tensor gab a scalar degree of freedom φ, approxi-
mately corresponding to the inverse of the effective gravitational coupling strength, which becomes
dynamical [70]. Brans-Dicke theory was generalized by promoting the constant Brans-Dicke param-
eter w to a function of φ and/or by including a potential V (φ) [71].
Finally, in Chapter 5, again, we proceed to analyze other proposals of to the scalar-tensor theory
in black holes.
The Jebsen-Birkhoff theorem breaks down in theories of gravity alternative to general relativity,
which are motivated by the need to explain the present acceleration of the Universe without an ad
hoc dark energy, and by unavoidable corrections to general relativity arising from any attempt to
quantize gravity.
The Jebsen-Birkhoff theorem [23]-[24], generalized in Ref. [74] to higher dimensional general
relativity, is extremely important because it makes general relativity black holes simple; there is
no need to single out the "physical" black hole solutions of general relativity, and the end point of
gravitational collapse is completely determined.
Black holes in scalar-tensor gravity are not arbitrary: an important no-hair theorem due to
Hawking states that all vacuum, stationary, and asymptotically flat black holes of Brans-Dicke gravity
2
Introduction
must reduce to Kerr black holes1 [45]. Hawking’s no-hair theorem has been generalized to include
more general scalar-tensor theories and a potential V (φ), provided that the latter has a minimum
which allows for φ to sit in a state of equilibrium [72]-[48]. An essential feature in the proof of no-hair
theorems is that φ becomes constant outside the horizon, reducing the theory to general relativity
and any black hole to Kerr.
A few maverick solutions are known which evade the no-hair theorems, but only at the price of
physical pathologies such as the divergence of φ on the horizon (the solution of [73] is one example,
but it is linearly unstable [75]). No-hair theorems for Horndeski and Galileon theories, and ways to
evade them, are the subject of a large literature ([46] and references therein).
In Brans-Dicke theory, things become more tricky. In fact, a theorem due to Agnese and La
Camera [80] states that in Brans-Dicke gravity the possible solutions describe only wormholes or
naked singularities. Its proof is incorrect, as shown below. In any case, it is clear that this theorem
contradicts Hawking’s no-hair result and its generalizations [45], [72]-[48], [12] because it excludes
the Schwarzschild black hole which is indeed a solution, as is easy to verify.
The objective of this thesis is to verify the validity of the generalized Jebsen-Birkhoff theorems
and to solve the problem of putative alternative solutions to Kottler/Schwarzschild-de Sitter proposed
by various authors in the literature. Then, we find the general solution of the vacuum Brans-Dicke
field equations which is static, spherical symmetry, and asymptotically flat (the black hole solution
being already well-known).
1
A Kerr black hole is a non-isotropic region that is delimited by an event horizon and an ergosphere presenting notable differences with respect
to the black hole of Schwarzschild. The Kerr spacetime corresponds to the gravitational field produced by body of mass M and angular momentum
L.
3
Chapter 1
Albert Einstein
General relativity is Einstein’s theory of space, time, and gravitation. The mathematics that
is used is differential geometry [2]. Space and time come together to form an entity called space-
time. This spacetime is a pseudo-Riemannian differentiable manifold with four dimensions, which
corresponds to a space that may have may have a complicated topology, but locally looks like Rn .
4
A Introduction to General Relativity Chapter 1
It is useful to distinguish the different signatures of the metric: a manifold is called Riemannian
if the metric is positive definite, that is (+ + + .... +), and Lorentzian if the first eigenvalue of the
metric has negative signature, that is (− + + .... +), so that the Lorentzian manifolds are spacetimes,
whereas the Riemannian manifolds only have spatial directions. The notations and conventions used
in this thesis are based on Ref. [2].
While in special relativity we work in a 4-dimensional Minkowski space with zero curvature, the
line element in Cartesian coordinate is
• gab is taken to be non-degenerate, meaning that the determinant g = |gab | does not vanish,
so this allows one to define the inverse metric g ab as
(
1 if a = b,
g ac gcb = δba = (1.4)
0 if a 6= b.
• gab is an invariant.
5
A Introduction to General Relativity Chapter 1
The expression of the metric in curvilinear coordinates is very different from that in Cartesian
coordinates and the space described by gab is still Rn locally with the same geometrical properties
as in Cartesian coordinates. A change of coordinates is just a different way of labeling the points of
a space and can never change its geometric properties.
The metric plays the same role as the Euclidean metric in Cartesian coordinates. In particular,
the metric and its inverse raise and lower the index of covariant and contravariant vectors and tensors.
Then, we can define the operations of "raising" and "lowering" of indices
( (
Va ≡ gab V b , Tab = gaα gbβ T αβ ,
and (1.5)
Va ≡ g ab Vb , T ab = g aα g bβ Tαβ ,
where the first-order tensors Va and V a are called covariant tensor and contravariant tensor re-
spectively; and the second-order tensor T ab and Tab are called doubly covariant tensor and doubly
contravariant tensor also respectively.
We can define the scalar product between vectors from the metric
U · V = gab U a V b . (1.6)
From the metric we can define the Levi-Civita connection or Christoffel symbols, whose compo-
nents on a coordinate basis are
1
Γcab = g cd (∂a gbd + ∂b gda − ∂d gab ) , (1.7)
2
∂gab
where ∂d gab ≡ .
∂xd
The connection coefficients do not form a tensor, but they are very important because they give
"form" to the manifold. Eq. (1.7) tell us that, for a given Riemannian manifold, there is a unique
connection ∇, called the Levi-Civita connection of the given metric gab . It is easy to verify that the
Levi-Civita connection is zero in Cartesian coordinates, but they are not in spherical or cylindrical
coordinates. The Levi-Civita connection or Christoffel symbols satisfy two conditions:
1. The connection is symmetric in the lower indexes1 , that is Γcab = Γcba , this implies that the
torsion tensor Tab
c is zero and the commutator of two covariant derivatives is given only by the
Riemann tensor.
[∇a , ∇b ]V c = Rabe
c
V e. (1.8)
2. The covariant derivative of the metric is zero, that is ∇c gab = 0, which is called the compat-
ibility with the metric, this is the connections that are compatible with the metric have several
properties that simplify many the geometrical properties of the manifold.
1
Thus it has at most 40 independent components.
6
A Introduction to General Relativity Chapter 1
The second advantage of the compatibility with the metric is that it implies that the covariant
derivative commutes with raising, lowering, or contracting indices, that is
gab ∇e S ab = ∇e (gab S ab ) = ∇e Saa . (1.14)
Then, given these two conditions, the connection is completely determined by the metric. Now,
consider the three covariant derivatives of the metric
c c
0 = ∇a gbe = ∂a gbe − Γab gce − Γae gbc ,
0 = ∇b gea = ∂b gea − Γcbe gca − Γcba gec , (1.15)
0 = ∇e gab = ∂e gab − Γceb gcb − Γceb gac .
If we add the first two expressions and subtract the third, and using the fact that the metric and
the connection are symmetric, we obtain
∂a gbe + ∂b gea − ∂e gab − 2Γcab gce = 0 (1.16)
and multiplying by g de we obtain
1
Γdab = g dc (∂a gcb + ∂b gac − ∂c gab ) . (1.17)
2
We can therefore express the Levi-Civita connection or Christoffel symbols completely as a func-
tion of the components of the metric and its derivatives.
7
A Introduction to General Relativity Chapter 1
1.3 Curvature
The curvature at each point of the manifold is determined by the Riemann tensor, which is
antisymmetric in the first and the last two indexes
Redab = −Rdeab
=⇒ Redab = Rdeba . (1.18)
Redab = −Redba
These conditions reduce the number of degrees of freedom of the Riemann tensor. In N dimensions
we have 12
1
N 2 (N 2 − 1) degrees of freedom, then N = 4 corresponds to 20 independent components
at most.
In a Riemannian manifold, the curvature tensor will be defined from the Levi-Civita connection.
The metric tensor g may be used to raise or lower indexes of the curvature tensor. In particular, the
fully covariant version of the tensor is a tensor of type 04 given by
e
Rαdab = gαe Rdab (1.19)
whose components on a coordinate basis are
e
Rdab = ∂a Γebd − ∂b Γead + Γeac Γcbd − Γebc Γcad . (1.20)
The Levi-Civita connection makes the curvature tensors Rdab
e be more symmetrical and satisfy
simpler relationships. We already know that the Riemann tensor is antisymmetric in the last two
indices
1
e
Rd(ab) e
≡ (Rdab e
+ Rdba )=0 (1.21)
2!
and
1 e
e
R[dab] = e
Rdab − Radb e
+ Rabd e
− Rbad e
+ Rb(da) e
− Rdba = 0. (1.22)
3!
As noted above, these properties reduce the maximum amount of independent components of the
Riemann tensor from 256 to 20. Also the Bianchi identities are satisfied
e
∇[c Rda]b e
= ∇c Rdab e
+ ∇b Rdca e
+ ∇a Rdbc = 0. (1.23)
The Ricci tensor Rab , is the only independent contraction of Riemann
e
Rab ≡ Raeb . (1.24)
For the curvature tensor formed from an arbitrary connection (not necessarily Levi-Civita), there
are a number of independent contractions to take. Using the Levi-Civita connection, for which (1.24)
is the only independent contraction, all others either vanish, or are related to this one. The Ricci
tensor associated with the Levi-Civita connection is automatically symmetric
Rab = Rba (1.25)
as a consequence of the symmetries of the Riemann tensor. The trace of the Ricci tensor is the Ricci
scalar or curvature scalar
R ≡ Raa = g ab Rab . (1.26)
8
A Introduction to General Relativity Chapter 1
1.4 Geodesics
Another advantage of using the Levi-Civita connection is in the study of geodesics.
The motion of a massive particle in free fall in general relativity takes place along a timelike
geodesic and its dynamics is described by the equation
ua ∇a ub = 0, (1.27)
where ua = dxa /dτ is the tangent vector (4-velocity) to the geodesic and τ the proper time used to
parametrize the position of the particle along that is geodesic. From eq. (1.9) we obtain
∂ub
ua + Γbac ua uc = 0. (1.28)
∂xa
Note that
d ∂ dxb ∂
= = ub b (1.29)
dτ ∂xb dτ ∂x
then eq. (1.28) is written as
dub d2 xb a
b dx dx
c
+ Γbac ua uc = 0 or + Γ ac = 0, (1.30)
dτ dτ 2 dτ dτ
which is called geodesic equation.
Figure 1.2: Different geodesics on a torus. The torus is a surface of revolution obtained by revolving about the z -axis with
azimuth χ ∈ [0; 2π), and another little circle perpendicular to x -y plane and the latitude θ ∈ [0; 2π). We have the outer equator
when χ = 0 (in red), the inner when χ = ±π (in green), and the meridian when θ = 0 (in blue). Ref. [21]
9
A Introduction to General Relativity Chapter 1
∇a Gab = 0. (1.32)
Torsion vanishes due to the symmetry of the connection, while Rabe e = 0 due to the property
(1.18).
We know from the Equivalence Principle that gravity is a manifestation of the curvature of
spacetime. But also we know that the source of this curvature is the matter, of which we have a
tensorial description, the energy-momentum tensor. But we still need to know how matter interacts
with spacetime. This interaction is given by the Einstein equations.
The Principle of Covariance tells us that the equation must be valid in all reference systems, and
therefore must have a tensorial form. The Einstein equations have the form
1 8πGN
Gab = Rab − Rgab = Tab (1.33)
2 c4
where Gab is the tensor that describes the curvature of space, Tab the energy-momentum tensor,
GN is the universal gravitation constant of Newton, and c is the speed of light. Then, as gravity
manifests itself as curvature of spacetime, this curvature is related to the distribution of matter and
energy by the Einstein equations. They form a system of non-linear partial differential equations of
second order for the components of the metric.
It is usual in the framework of general relativity to use the system of "geometric" units in which
c = 1 and GN = 1. This system of units will be used for the rest of this thesis.
Now, the tensor Gab has different mathematical and physical restrictions, such as
1. Gab must be symmetric in both indices, since Tab is also symmetric.
2. Gab is a purely geometric object, therefore, it has to be a function only of the metric gab
and its derivatives.
10
A Introduction to General Relativity Chapter 1
All these conditions determine the tensor Gab univocally, the most general expression for a sym-
0
metric tensor 2 , constructed from the metric gab and its derivatives and linear in Rabec is, except a
Requiring that ∇a Gab = 0 implies that α = −1/2 and that Λ is a constant, while demanding that
Gab = 0 for flat space implies that Λ = 0. Therefore, the only tensor that satisfies all the necessary
conditions is the Einstein tensor as (1.31)
1
Gab = Rab − gab R. (1.35)
2
Comparing with eq. (1.33), we finally have
1
Rab − gab R = 8πTab . (1.36)
2
The Einstein equations form a system of 10 coupled non-linear partial differential equations of
second order. So, we have 10 components, but the condition ∇a Gab = 0 imposes 4 constraints, so
that there are only 6 independent equations, and these correspond to physical degrees of freedom.
The other four components are non-physical and they express the freedom of choice of the coordinate
system.
We can derive the equations of motion for such a particle in one dimension with coordinates q(t)
by using the "principle of least action", that is
Z
S = dtL(q, q̇), (1.37)
where L(q, q̇) is the Lagrangian. The critical points of the action, i.e. trajectories q(t) for which S
remains stationary under small variations, are those that satisfy the Euler-Lagrange equations
∂L d ∂L
− = 0. (1.38)
∂q dt ∂ q̇
In field theory, we need to replace the single coordinates q(t) by a set of spacetime-dependent
fields, Φi (xa ), and S becomes a functional of these fields.2 In field theory, the Lagrangian can be
2
A functional is a function of an infinite number of variables, such as the values of a field in some region of spacetime. Each Φi is a function on
spacetime in some coordinate system, and i is an index labeling our individual fields.
11
A Introduction to General Relativity Chapter 1
expressed as an integral over space of a Lagrangian density L , which is a function of the fields Φi
and their spacetime derivative ∂a Φi
Z
L = d3 xL (Φi , ∂a Φi ), (1.39)
so the action is
Z Z
S= dtL = d4 xL (Φi , ∂a Φi ). (1.40)
The action SH , called Hilbert action, is the integral over spacetime of a Lagrangian density
Z
SH = d4 xLH , (1.41)
√
where LH is a tensor density, which can be written as −g times a scalar, that is
√
LH = −gR. (1.42)
The equations of motion should come from varying the action with respect to the metric, this is
principle of least action δS. In fact let us consider variations with respect to the inverse metric g ab
and using R = g ab Rab , we will have
Z h√ √ √ i
δSH = d4 x ( −g)g ab δRab + ( −g)Rab δg ab + Rδ( −g) = (δS)1 + (δS)2 + (δS)3 . (1.43)
From eq. (1.20), we proceed to make the variation of the Levi-Civita connection with respect to the
metric, this is
The variation δΓρab is the difference of two connections, and therefore is itself a tensor. We can
take its covariant derivative
Therefore, the contribution of the first term (δS)1 in (1.43) can be written as
√
Z
(δS)1 = d4 x −gg ab [∇c (δΓcab ) − ∇b (δΓcac )] =
√
Z h i
= d4 x −g∇σ g aσ (δΓcac ) − g ab (δΓdab ) , (1.47)
3
We will say "Lagrangian" when we mean "Lagrangian density."
12
A Introduction to General Relativity Chapter 1
where we have used the metric compatibility and relabeled some dummy indices. But now we have
the integral with respect to the natural volume element of the covariant divergence of a vector; by
Stokes theorem, this is equal to a boundary contribution at infinity which we can set to zero by
making the variation vanish at infinity.
The second term (δS)2 is already in the form of some expression times δgab , this remains un-
changed.
Using the fact that δgab = −gaρ gbσ δg ρσ to convert from δgab to δg ab , then
√ 1 1 g 1√
δ −g = √ δg = √ gab δg ab = − −ggab δg ab . (1.51)
2 −g 2 −g 2
Putting all these results together in eq. (1.43) (remembering that (δS)1 does not contribute), we
obtain
4 √
Z
1
δSH = d x −g Rab − Rgab δg ab . (1.52)
2
This vanishes for arbitrary variations, so we are led to the Einstein equations in vacuum
1 δSH 1
√ = Rab − Rgab = 0. (1.53)
−g δg ab 2
1 δS(m)
1 δSH 1 1
√ ab
= Rab − Rgab + √ = 0, (1.55)
−g δg 8π 2 −g δg ab
13
A Introduction to General Relativity Chapter 1
1 δS(m)
Tab = − √ , (1.56)
−g δg ab
Going back to the eq. (1.40), where each Φi is a function on spacetime, we can write the variation
of the action δS as
Z Z
∂L ∂L δS
δS = d x 4
i
− ∂a i
δΦ = d4 x i δΦi .
i
(1.58)
∂Φ ∂ (∂a Φ ) δΦ
We can express the notion that S is at a critical point by saying that the functional derivative
vanishes. The final equations of motion four a field theory are
δS ∂L ∂L
= − ∂a = 0. (1.59)
δΦi ∂Φi ∂ (∂a Φi )
tells us all we need to know about the energy-like aspects of a system: energy density, pressure,
stress, and so forth. In this way, we have
1.) T 00 is "the flow of p0 (energy) across a spatial surface x0 = const. (time)", or simply the
rest-frame energy-density ρ.5
3.) T ij are the spatial component (i, j = 1, 2, 3) and they are the momentum flux, or the stress;
they represent the forces that infinitesimal elements of the fluid exert on other nearby elements.
We can distinguish:
4
Sometimes called the stress-energy tensor.
5
Note that the index 0 corresponds to t component.
14
A Introduction to General Relativity Chapter 1
• If i = j, we have T ii that are the diagonal terms and represent the component F i of
the force on the surface xi = const., that is, the pressure pi in the direction xi .
• If i 6= j, we have T ij that are the off-diagonal terms and represent shearing terms, such
as those due to viscosity.
The pressure has three components, given in this perfect fluid rest frame by
pi = T ii . (1.60)
Notice that there is no sum over i. Therefore, for a set of particles that do not interact, which is
called cold matter or non-relativistic matter or dust, its energy-momentum tensor is given by
Tdab = ρ0 ua ub , (1.61)
where ρ0 is the density of the particles in the reference frame in which they are at rest, and ua is the
4-velocity of the dust. We defined
dxa
ua (τ ) = . (1.62)
dτ
In the reference system of a comoving observer, which moves along with the fluid6 , the only
non-zero component is T(d)00 = ρ , that is, the density of energy. In other words, non-interacting
0
matter has no pressure. The law of conservation of energy (mass) says that in these coordinates
∂t ρ0 = 0, or in components of the energy-momentum tensor, ∂0 T(d)00 = 0. Therefore, for the law of
Vacuum
The vacuum solution is a Lorentzian manifold whose Einstein tensor vanishes identically. Ac-
cording to the Einstein field equation (1.33), this means that the stress–energy tensor also vanishes
identically, that is Tab = 0, so that no matter or non-gravitational fields are present.
1
Gab = Rab − Rgab = 0 (1.66)
2
They are a special case of the more general exact solutions in general relativity.
6
That is, the comoving observer has the same velocity vector ua as the fluid around it.
15
A Introduction to General Relativity Chapter 1
Cosmological constant Λ
The cosmological constant Λ is the energy density of space, or vacuum energy, and it is closely
associated to the concepts of dark energy. Then, the field equations of General Relativity (1.36) can
be modified with the introduction of this cosmological constant, with which said equations turn out
to be
1
Rab − gab R + gab Λ = 8πTab . (1.67)
2
Vacuum with Λ
In this case, the Einstein equations in vacuum with cosmological constant Λ are
1
Rab = gab R − gab Λ (1.68)
2
which brings us to
R = 4Λ (1.69)
and we obtain
Rab = gab Λ. (1.70)
This tells us that if the cosmological constant is not equal to zero then, even in the total absence
of any trace of mass-energy, spacetime will have a curvature. The corresponding effective momentum-
energy tensor would be
(Λ) Λ
Tab = − gab (1.71)
8π
Λ
that is, a perfect fluid with ρΛ = −PΛ = .
8π
Recent cosmological observations seem to suggest that in our universe the cosmological constant
has a small positive value and forms 68% of the energy content of the universe [19].
Electrovacuum
The electrovacuum solution is an exact solution of the Einstein field equation in which the only
nongravitational mass-energy present is the field energy of an electromagnetic field, which must
satisfy the Maxwell equations appropriate to the given geometry. For this reason, electrovacuums
are sometimes called Einstein-Maxwell solutions.
∇ × B − ∂t E = J,
∇ · E = ρ,
∇ × E + ∂t B = 0,
∇ · B = 0. (1.72)
16
A Introduction to General Relativity Chapter 1
where E and B are the electric and magnetic field 3-vectors, J is the current, ρ is the charge density;
and B = ∇ × A where A is a vector potential and E = −∇ · φ = −∂t A where φ is a scalar potential.
These equations are invariant under Lorentz transformations. Writing these equations in component
notation, we obtain
ijk ∂j Bk − ∂0 E i = J i ,
∂i E i = J 0 ,
ijk ∂j Ek − ∂0 B i = 0,
∂i B i = 0. (1.73)
where ijk is the three-dimensional Levi-Civita symbol7 , and we replaced the charge density ρ by J 0
because the density and current together form the current 4-vector J a = (ρ, J x , J y , J z ). In addition,
we know the electromagnetic field strength tensor Fab which is written as
0 −E1 −E2 −E3
E1 0 B3 −B2
(Fab ) =
E2 −B3
= −(Fba ). (1.74)
0 B1
E3 B2 −B1 0
From (1.73) and the definition (1.74) of the strength tensor Fab it is easy to get a completely
tensorial version of Maxwell’s equations. Then we can see that
F 0i = E i ,
F ij = ijk Bk . (1.75)
∂j F ij − ∂0 F 0i = J i ,
∂i F 0i = J 0 . (1.76)
Using the antisymmetry of F ab , we see that these may be combined in to the single tensor equation
∂a F ba = J b or 2·F=J (1.77)
called covariant formulation of Maxwell’s equations. A similar line of reasoning reveals that the third
and fourth equations in (1.73) can be written as
∂a F bc + ∂b F ca + ∂c F ab = 0 (1.78)
Looking at the eq. (1.59), we can see that the fields Φi are the four components of vector potential
A, that is
Φi = (A0 , A1 , A2 , A3 ), (1.79)
7
Note that ijk = ijk = 1 is normalized.
17
A Introduction to General Relativity Chapter 1
∂a F ba = J b . (1.84)
Finally, from (1.81) we can derive the tensor energy-momentum for electromagnetism
1 ab cd
ab
T(EM ac b
) = F Fc − g F Fcd . (1.85)
4
Perfect fluid
A particular case is that of the perfect fluid, with density ρ and pressure P . The energy-momentum
tensor is given by
T ab = (ρ + P )ua ub + P g ab . (1.86)
A comoving observer will see an isotropic fluid locally and, for this observer, the energy-momentum
tensor has the simple form
ρ 0 0 0
0 P 0 0
T ab =
0 0 P 0 .
(1.87)
0 0 0 P
By imposing ∂a T ab = 0 in (1.86) for the case of flat space, we find the laws of conservation of
energy and momentum in the relativistic version
∂a (ρ0 ua ) + P ∂a ua = 0,
(ρ0 + P )ua ∂a ub − ∂ b P + ∂ a P ua ub = 0. (1.88)
There is a relationship between the pressure p and density ρ called the equation of state. In the
case of the perfect fluids, they obey the equation
18
A Introduction to General Relativity Chapter 1
P = wρ (1.89)
where the proportionality constant w is the equation of state parameter and it is independent of time.
The energy-momentum tensor (1.86) and the equation of state (1.89) are of great importance in
cosmology, where different types of energy are represented by perfect fluids with different values of
w, this is
• w = 0 corresponds to dust
• w = 1/3 corresponds to radiation
• w = −1 corresponds to vacuum.
Imperfect fluid
Another possibility is to allow for an imperfect fluid with energy-momentum tensor [3].
Tab = (P + ρ)ua ub + P gab + qa ub + qb ua (1.90)
with P = wρ and the 4-velocity uµ is defined as
!
µ 1 p
u = p , 0, 0, 0 , uµ = (− |gtt |, 0, 0, 0).
|gtt |
The vector q c is purely spatial and describes a radial energy flow
√
q µ = (0, q, 0, 0), qµ = (0, q grr , 0, 0) (1.91)
while
ua ua = −1 q c uc = 0. (1.92)
Scalar field
The simplest example of a field is a real scalar field, that means φ(xµ ) :→ R.
If we consider the classical mechanics for that real scalar field, we have an energy density that is
a local function of spacetime as
1 2
kinetic energy: φ̇ ,
2
1
gradient energy: (∇φ)2 ,
2
potential energy: V (φ). (1.93)
Although the potential is a Lorentz-invariant function, the kinetic and gradient energies are not
themselves Lorentz-invariant, but we can combine them into a manifestly Lorentz-invariant form
1 1 1
− g ab ∇a φ∇b φ = φ̇2 − (∇φ)2 . (1.94)
2 2 2
19
A Introduction to General Relativity Chapter 1
Let a vector ξa (xb ) be defined for each point of the space xb , in an infinitesimal translation in the
direction ξa we have
∂ξ d ∂ξ d
∂gab d
2 a b
δ(ds ) = δ(gab dx dx ) = ξ + gdb a + gad b dxa dxb dλ = 0. (1.99)
∂xd ∂x ∂x
In the previous expression it is taken into account that the operators commute, i.e., dδ = δd and
conclude with the equations
∂ξ d ∂ξ d
∂gab d
ξ + gdb a + gad b = 0 (1.100)
∂xd ∂x ∂x
which is known as the Lie derivative £ of the metric tensor gab along the vector field ξ
£ξ gab = 0 = ∇a ξb + ∇b ξa (1.101)
If they exist, the vectors ξ a satisfying (1.100) or (1.101) are called Killing vectors and eq. (1.101)
the Killing equation which describes a symmetry in spacetime.
If eq. (1.100) has no solution, then the spacetime has no continuous symmetry.
8
Analogous to L = K − V in the point-particle case.
20
A Introduction to General Relativity Chapter 1
In the absence of energy and matter, the energy-momentum tensor Tab is zero and the Einstein
equations reduce to
Rab = 0. (1.102)
At first glance, it is strange that the vacuum equations admit non-trivial solutions, apart from
Minkowski space. But, due to non-linearity, there exists a large class of non-trivial vacuum solutions.
The metrics that satisfy the vacuum equation (1.102) are called Ricci-flat. The solutions that are
Ricci-flat, in general, are not flat, but they are vacuum solutions.9
9
This is in the absence of the cosmological constant Λ.
21
A Introduction to General Relativity Chapter 1
Now, if we observe the Schwarzschild metric (1.103) one can see that the metric tensor is inde-
pendent of the coordinates t and ϕ. Therefore, the vectors
ξ t = (1, 0, 0, 0) ⇒ ξ t = ∂t ,
ξ ϕ = (0, 0, 0, 1) ⇒ ξ ϕ = ∂ϕ , (1.105)
are Killing vector fields.
We can quickly compute the associated conserved quantities using the equation
dxa
ξa = ξa pa = const. (1.106)
dτ
dxa
so, knowing that pa = m is the 4-momentum associated with a massive particle, and using ξ t
dτ
from eq. (1.105) and gtt from eq. (1.103), we have
dt t dt 2M dt
t
ξt p = ξt m = mgtt ξ = −m 1 − = −E (1.107)
dτ dτ r dτ
where the quantity conserved corresponds to the energy E of theparticle. Proceeding in the same
∂ a
way, the conserved quantity associated with the vector ξϕ =
a from (1.105) is
∂ϕ
dϕ dϕ dϕ
ξϕ p ϕ = ξϕ m
= mgϕϕ ξ ϕ = mr2 sin2 θ = L, (1.108)
dτ dτ dτ
where the quantity conserved this time is the angular momentum L of the particle. The vector
associated in the other two coordinates r and θ are
ξ r = (0, 0, cos ϕ, − cot θ sin ϕ)
22
A Introduction to General Relativity Chapter 1
1.8.3 Horizons
The universe has a curious property, which is that it is homogeneous. Homogeneous at cosmologi-
cal scales, that is, if we look around us, it is not homogeneous, nor at astronomical scales, because we
have planets, stars and galaxies. However, if we increase the distance, the universe is incredibly ho-
mogeneous (and all inhomogeneities are irrelevant to those scales). But this homogeneity is seen more
clearly in the cosmic background radiation (Fig. 2.3) that has a temperature of 2.7 degrees Kelvin
and where the temperature variations never exceed one part per one hundred thousand. But that is
a problem, so that two bodies are at the same temperature they have had to exchange energy and
the bodies have had to spend time in contact (or exchanging radiation). But energy can not spread
faster than the speed of light according to relativity. However, looking at the cosmic background
radiation we see that everything has the same temperature, although the spatial separation is such
that nothing has given time to connect diametrically opposite points in the sky. This is undoubtedly
the most interesting and "problematic" problem of all concerning the model of standard cosmology.
How and why is everything at the same temperature? Did things move faster than light? Is there
another mechanism to explain this? Ref. [25]. This problem is called "horizon problem". A horizon
is "a frontier between things observable and things unobservable" [3].
In a stationary black hole, an event horizon is a region in spacetime beyond which events cannot
affect an outside observer, it is defined as the shell of "points of no return", i.e., the boundary at which
the gravitational pull of a massive object becomes so great as to make escape impossible. However,
this definition turns out to be essentially useless for practical purposes in dynamical spacetimes. This
major obstacle manifests itself because knowing the event horizon requires the knowledge of future
null infinity, which is physically impossible to achieve [3].
In addition, we can define an apparent horizon as the surface that appears in the collapse of a
body to form a black hole that separates the region from which the light can not escape from that
from which it can still escape. It turns out that this horizon does not coincide in general with the
event horizon [25].
Finally, a naked singularity as one that is not hidden behind an event horizon. Let us review the
history of a black hole [25]
1.) We have a system that in the distant future will form a black hole. Then it begins to form
an event horizon long before the process of black hole formation begins.
2.) When the process of black hole formation begins, the event horizon is already formed.
However, in the collapse process an apparent horizon is formed.
3.) The event horizon and the apparent one coincide when the system has formed the black hole
and becomes stationary.
1.8.4 Singularities
Many times the metric that emerges as a solution to Einstein equations is singular in a certain
region of spacetime, that is, some components and/or the determinant of gab tend to zero or to
infinity. In this case we have to distinguish between two types of singularities.
23
A Introduction to General Relativity Chapter 1
• Physical singularity: This is a location where the curvature of space is infinite. Near a physical
singularity, the gravitational field is very strong and the tidal effects are very large. British physicists
Stephen Hawking (1942-2018) and Roger Penrose (1931-) demonstrated in the 1960s that there must
necessarily be a physical singularity within a surface that traps light, so that the singularity in the
centre of a black hole is physical. There is a conjecture, called the Cosmic Censorship Conjecture,
that any physical singularity is always surrounded by an event horizon, that the singularity itself is
not visible from the outside and can not influence the rest of the universe and create problems of
causality.
In particular, we call coordinate singularities the artifacts of the coordinate system that we
use. The origin in spherical coordinates in R3 is an example of a coordinate singularity, since the
determinant |g| = r2 sin2 θ tends to zero for r → 0, although in Cartesian coordinates it is clear that
the origin is a completely regular point.
The general way to distinguish between coordinate singularities and physical singularities is by
calculating the curvature invariants, which are scalar functions formed from contractions of the
Riemann tensor, as R, Rab Rab , or Rabcd Rabcd . If a curvature invariant is singular, then the singularity
is a physical singularity. On the other hand if the invariants are regular, it is probable, but not sure
that the singularity is a coordinate singularity. One of the most useful ways to check is by checking
for the finiteness of the Kretschmann scalar which sometimes is also called Riemann tensor squared,
in other words:
If the curvature invariant K is singular, then the singularity is a physical singularity. On the
other hand if K is regular, it is probable, but it is not certain that the singularity is a coordinate
singular. The definitive proof that a singularity is coordinate is to find a change of coordinates that
makes the singularity disappear.
24
A Introduction to General Relativity Chapter 1
(1.103), both R and Rab Rab are of little use, since both are identically zero by construction, due
to the fact that the Schwarzschild solution is Ricci-flat. However, by calculating the Kretschmann
invariant, we obtain for the metric (1.103) that
48M 2
K = Rabcd Rabcd = , (1.111)
r6
that is, the curvature invariant diverges as r → 0, but it remains perfectly regular at r = 2M . The
point r = 0 therefore, represents a physical singularity, while r = 2M turns out to be a coordinate
singularity.10
In contrast to the singularity of coordinates in θ = 0 and θ = π, the coordinate singularity at
r = 2M does have a physical meaning. The radius rs = 2M is called the Schwarzschild radius and
plays an important role in the physics of the solution (1.103).
Now, the fact that the Schwarzschild coordinates (t, r, θ, ϕ) are singular at r = 2M means that
they are not valid at the Schwarzschild radius. We will see that they are the appropriate coordinates
to describe the region r > 2M, but for r < 2M they are unreliable and for r = 2M completely
useless.
One way to see that something physically non-trivial happens at the Schwarzschild radius is by
looking at the gravitational Doppler effect. We have that
s v
u 1 − 2M
u
gtt (rd ) rd
Td = Te = Te t 2M
(1.112)
gtt (re ) 1 − re
where Te , νe and Td , νd are the periods and frequencies of a light beam at the time of emission and
detection, respectively. Therefore, if gtt (re ) < gtt (re ), we have νd < νe . In other words, a light beam
will lose energy when trying to escape the gravitational potential well and will suffer a redshift.Then,
when a clock falling toward the centre of star or black hole emits pulses with period Te , a fixed
observer at r = rd will notice that the period Td measured by him is greater the closer the clock
approaches the Schwarzschild radius, to such a point that the Doppler effect is infinite when the
clock is at rs = 2M . That means that the Schwarzschild radius is a surface of infinite redshift.
If we look at the causal structure of the Schwarzschild solution, through radial null geodesics,
then the null geodesics that go radially from r = 0 to r = ∞ or backwards. In particular they have
dθ = dϕ = 0 and ds2 = 0, so their equation is given by
2M −1 2
2M
2
ds = 0 = − 1 − 2
dt + 1 − dr , (1.114)
r r
10
Strictly speaking, the regularity of K at r = 2M is not enough to conclude that it is a coordinate singularity, nor is the fact that numerous
other invariants are finite at this point. The real argument is that there is a change in coordinates such that r = 2M becomes regular.
25
A Introduction to General Relativity Chapter 1
2M −1
dt
=± 1− . (1.115)
dr r
t = ± (r + 2M ln |r − 2M | + c0 ) , (1.116)
where c0 is an integration constant and the signs indicate the direction of the geodesics: the positive
sign corresponds to outgoing geodesics and the negative sign to incoming geodesics. (Fig. 1.3)
Figure 1.3: The radial null geodesics in the Schwarzschild solution at the Schwarzschild coordinates (t, r, θ, ϕ). For r 2M
the light cones behave as in Minkowski, but closer to the radius of Schwarzschild the cones close and degenerate at rs = 2M .
Within the radius of Schwarzschild the cones have changed their orientation and point towards the singularity at r = 0. In these
coordinates, however, it seems that it is not possible to cross the Schwarzschild radius from the outside. Ref. [18]
Far from the centre, for r 2M , the second term is negligible compared to the first and the light
cones behave more or less as in Minkowski, that is, they make 45° angles with the axes t and r. This is
to be expected, since the Schwarzschild solution is asymptotically flat and approaches Minkowski for
r → ∞. However, closer to the Schwarzschild radius, the logarithmic term becomes more important,
which indicates that for light rays it is more difficult to get out of the potential well created by the
gravitational field. Specifically, the light cones close the closer we get to the Schwarzschild radius
and at rs = 2M they are completely degenerate (Fig. 1.3). Inside the Schwarzschild radius, both the
ingoing and the outgoing radial null geodesics point towards the singularity and the light cones have
changed orientation and point towards the physical singularity at r = 0. This is a consequence of
the fact that, for r < 2M , the gtt and grr components of the metric change sign and the coordinates
t and r exchange the roles of temporal and spatial coordinates.
In Fig. 1.3 it seems that no light signal is able to cross the Schwarzschild radius, neither from the
outside inwards, or vice-versa. Since a massive particle always moves within its light cone, it seems
that no massive particle, nor any observer, is able to enter the zone r < 2M . At the Schwarzschild
radius, a particle that falls towards the centre, seems to approach asymptotically the Schwarzschild
radius, but never crosses it.
26
A Introduction to General Relativity Chapter 1
But this turns out to be an erroneous conclusion, because the coordinates (t, r, θ, ϕ) are not
appropriate to describe what happens near the Schwarzschild radius. In 1939, the American physicist
Robert Oppenheimer (1904-1967) realized that the time of fall is very different, if we express it in
the proper time of the particle in free fall. For this it is necessary to make a study of the timelike
radial geodesics.
To calculate the radial timelike geodesic we are interested in the time component of eq. (1.27)
where an overdot denote differentiation with respect to the proper time τ . For the Schwarzschild
solution, we can simplify this expression, multiplying the equation by gtt and writing the Christoffel
symbol explicitly as a function of the metric (1.17), so that eq. (1.117) becomes
d d 2M
(1.118)
gtt ṫ = 0 ⇒ 1− ṫ = 0,
dτ dτ r
The integration constant has been chosen so that the equation corresponds to that of a particle
falling from infinity with zero initial velocity, since from eq. (1.119) we deduce that dτ ≈ dt for
r 2M .
Near infinity, the effects of relativistic time dilatation between the observer in free fall and the
observer measuring with time t are negligible. Therefore, we can conclude that the t coordinate
corresponds to the proper time of an asymptotic observer at infinity, far from rs = 2M. This is already
a first indication that the coordinate system (t, r, θ, ϕ) is not appropriate near the Schwarzschild
radius, since we know that time does not run up and down in a gravitational potential. Therefore,
the equations for timelike geodesic is
−1
2M 2M
1− 2
ṫ − 1 − ṙ2 = 1. (1.120)
r r
2M 2 2
2M 2M
2
ṙ = 1 − ṫ − 1 − = , (1.121)
r r r
27
A Introduction to General Relativity Chapter 1
with solution11
2 3/2
τ= √ r0 − r3/2 . (1.123)
3 2M
In other words, falling from a distance r0 , an observer O0 in free fall arrives at the Schwarzschild
radius rs = 2M and even at the singularity r = 0 in finite interval of proper time τ . Therefore,
although it does not look like in Fig.1.3, it is perfectly possible to reach the Schwarzschild radius,
cross it, and reach the singularity. The problem is that this figure is drawn in coordinates (t, r, θ, ϕ),
which are the coordinates of an observer O locally at r = ∞.
Fig. 1.3 therefore does not reflect what actually happens with O0 near the Schwarzschild radius,
but rather what happens with the O0 observer according to the point of view of the observer O.
The observer O only receives the light signals that O0 emits. But we have already seen that these
signals follow the radial null geodesics (1.116) and therefore arrive at O with increasing intervals,
as a consequence of suffering a redshift that is increasing. O sees O0 approach asymptotically to
Schwarzschild radius, but never see it cross it, since the signal emitted by O0 in rs = 2M reaches O
at time t = ∞.
Figure 1.4: An observer in free fall O0 reaches the singularity after a finite proper time. The light signals emitted by O0 (dotted
curves) follow radial null geodesics and take longer to reach the far observer O, who erroneously believes that O0 follows path
A and never crosses the Schwarzschild radius. Once past the Schwarzschild radius, O0 has no way to communicate with O to
convince him/her otherwise. Ref. [18]
In Fig. 1.4 it can be seen that the signals that O0 emits after crossing the Schwarzschild radius
do not go outward and never reach O, not even after an infinite amount of time t.
11
The negative sign is due to the fact that it describes a free fall towards the black hole.
28
A Introduction to General Relativity Chapter 1
t = ±r∗ + c0 (1.124)
r∗ ≡ r + 2M ln |r − 2M | + c0 . (1.125)
d(±r∗ + co ) dr∗ 2M −1
dt
= =± =± 1− (1.126)
dr dr dr r
2M −1
∗
⇒ dr = 1 − dr (1.127)
r
Z Z
dr rdr
⇒ r∗ = 2M
= (1.128)
1− r r − 2M
⇒ r∗ = r + 2M ln |r − 2M |, (1.129)
where the integration constant has been set to zero. In terms of the tortoise coordinate, the
Schwarzschild metric becomes
2M
2
ds = 1 − (−dt2 + dr∗2 ) + r2 dΩ2(2) , (1.130)
r
where r is thought of as a function of r∗ . We introduce null coordinates u and v defined as
(
v ≡ t + r∗ ,
(1.131)
u ≡ t − r∗ .
They map the semi-infinite interval r ∈ [rs ; ∞) onto the infinite interval r∗ ∈ (−∞; ∞). Hence,
the horizon at rs has been pushed to infinity. Solutions with v = const. penetrate the future horizon,
while solutions with u = const. penetrate the past horizon.
The ingoing Eddington–Finkelstein coordinates are obtained by replacing the coordinate t with
the new coordinate v = t + r∗ advanced time, then these coordinates in the Schwarzschild metric
becomes
2M
2
ds = − 1 − dv 2 + 2dvdr + dr2 + r2 dΩ2(2) . (1.132)
r
29
A Introduction to General Relativity Chapter 1
Figure 1.5: Light ray trajectories in ingoing Eddington-Finkelstein v vs r coordinates. Ref. [18]
In Fig. 1.5 can be seen that the behaviour of the light cones in the coordinates of Eddington-
Finkelstein is different than in the Schwarzschild coordinates. For r 2M , the light cones behave as
the Minkowski space but, close to the r = 2M , the light cones begin to tilt towards the singularity.
The ingoing null radial geodesics always maintain the same 45° angle with the r-axis, but the outgoing
null geodesics have an increasing angle, the closer they are to r = 2M , since the outgoing null
geodesics are more and more difficult to get out to infinite, due to the curvature of space. The
Schwarzschild radius is a special point, because the inclination of the light cones of the outgoing
null geodesics form an angle of 90° with the axis r. A light signal emitted radially outward, will
remain at a fixed distance at r = 2M from the center. For r < 2M , the inclination of light cones
is even greater and even the outgoing null geodesics are directed towards the center and end at the
singularity not being able to escape beyond the Schwarzschild radius which acts as an event horizon.
This is the reason why black holes are "black".
Likewise, the outgoing Eddington–Finkelstein coordinates are obtained by replacing t with the
null coordinate u = t − r∗ or retarded time, then the metric is given by
2M
2
ds = − 1 − dv 2 − 2dvdr + dr2 + r2 dΩ2(2) . (1.133)
r
From Fig. 1.6, we can see that the outer part of the horizon is identical to that in the other
coordinates, but inside the situation is totally different. The Schwarzschild radius corresponds to a
membrane that only passes causal influences from the inside to the outside, a kind of white hole.
This indicates that neither of the two coordinate systems completely covers the manifold but only
patches of it. The advanced coordinates cover an asymptotically flat region with a singularity covered
by a horizon in the future, while the retarded ones describe another asymptotically flat area with
a singularity covered by a horizon in the past. These two coordinate systems together, completely
describe the whole manifold that is composed of a white hole and a black hole.
30
A Introduction to General Relativity Chapter 1
Figure 1.6: Light ray trajectories in outgoing Eddington-Finkelstein u vs r coordinates. Ref. [18]
where the upper sign (−) refers to the exterior region, and the lower sign (+) refers to the interior
region. Replacing eq. (1.131) in (1.134), and working out the resulting equation, we obtain
∗
V ≡ e(t+r )/4M ,
∗ (1.135)
U ≡ ∓ e−(t−r )/4M ,
r−2M
r
U V = ∓ e(r+2M ln | 2M
|)/2M
= ∓ er/2M −1 (1.137)
2M
and one obtains the Schwarzschild line element in Kruskal-Szekeres coordinates
32M 3 −r/2M
ds2 = − e dU dV + r2 dΩ2(2) , (1.138)
r
31
A Introduction to General Relativity Chapter 1
which is clearly regular at rs = 2M . We can see in Fig. 1.7 that the asymptotically flat part
corresponds to the region I where V > 0 and U < 0, and the event horizon rs = 2M corresponds
to the straight lines at 45 degrees. Behind the horizon there are two regions, II and IV, mutually
symmetrical in time, each with its own singularity.
The advanced coordinates cover regions I and II describing a Schwarzschild black hole, while
retarded coordinates cover regions III and IV describing a Schwarzschild white hole.
Figure 1.7: The Kruskal-Szekeres plane: the horizon rs = 2M correspond to the U and V axes, the singularity r = 0
to the hyperbola U V = 1 or U V = −1 (in red), and region I is covered by the Schwarzschild coordinates (t, r). The
Schwarzschild black hole is described by regions I and II, while the white hole is described by regions III and IV. Credits
of http://teoria-de-la-relatividad.blogspot.com/2009/03/28b-los-agujeros-negros-ii_18.html
32
A Introduction to General Relativity Chapter 1
and
rs
dz 2 = dr2 . (1.144)
r − rs
Replacing (1.142) into (1.140) we get
rs rs + r − rs
dl = 2 2 2 2 2
dr + dr + r dϕ = dr2 + r2 dϕ2 (1.145)
r − rs r − rs
rs −1 2
⇒ dl2 = 1 − dr + r2 dϕ2 = dΣ2 (1.146)
r
and this is the Schwarzschild metric in cylindrical coordinates as is shown in eq. (1.139).
To visualize, consider the plane (z, r). In that plane we have a curve given by z(r) we have
Figure 1.8: Representation of the Flamm paraboloid, whose geometric curvature coincides with that of the ecliptic or equatorial
plane of a spherically symmetric star. This should not be confused with the concept of a gravity well. Credit of Wikipedia.
33
Chapter 2
Relativistic Cosmology
Stephen Hawking
Cosmology, the study of the entire universe, is almost as old as human civilization. Already the
pre-Socratic philosophers, like Thales of Mileto or Anaximander (VI century BC) thought they could
deduce certain statements about the form and constitution of the universe. However, cosmology
has only become part of physics when, thanks to general relativity, the universe was conceived as a
dynamic system, governed by the same physical laws that govern matter within the universe. General
relativity predicts a close relationship between the structure of spacetime and its matter and energy
content. It is, therefore, a logical step to try to use Einstein’s equations to study the dynamics of
the whole universe: its form, its content and its evolution.
Figure 2.1: The distribution of galaxies is clumpy on small scales, but becomes more uniform on large scales. Ref. [22].
34
Relativistic Cosmology Chapter 2
Obviously, it is impossible to find an exact solution of the Einstein equations that describes the
entire universe with all the stars, galaxies, gravitational waves, electric and magnetic field that it
contains in every detail. But this is not what purports cosmology. Then, we can say that
Cosmology describes the dynamics of the entire universe at very large scales, where the influence
of individual galaxies and even clusters of galaxies are merely perturbations.
Relativistic cosmology is able of making very realistic models thanks to two principles:
1.) Cosmological Principle, which claims to say something about the form of the universe. The
idea is that, whatever the evolution of the universe, at any moment its appearance is the same in all
points and in all directions. This aspect can vary throughout the evolution, but at least it varies in
the same way in all sites. The Cosmological Principle is formulated as:
At any time, the universe is homogeneous and isotropic at very large scales.
Homogeneous means that all points on the surface are equivalent, there is no privileged point;
and isotropic means that there is no privileged direction, that the surfaces have the same appearance
in all directions.
Therefore, the Cosmological Principle implies that the metric of the universe can be written
as a family of spacelike hypersurfaces (3-dimensional), each spatially homogeneous and isotropic,
representing the universe at a constant time t and together they describe the evolution of time.
A description of a space in terms of hypersurfaces (not necessarily spacelike), such that each
point of space is located at exactly one hypersurface, is called a foliation (Fig. 2.2). The Cosmolog-
ical Principle says that the spaces that describe cosmological solutions are foliations with spacelike
sections of constant 3-dimensional curvature.
Figure 2.2: The spacetime of the universe can be foliated into flat, positively curved, or negatively curved spatial hypersurfaces.
Ref. [22].
Even so, we wonder to what extent the Cosmological Principle is true. We know from observations
that stars are concentrated in galaxies, they are in galaxies clusters that in turn form superclusters
35
Relativistic Cosmology Chapter 2
with large gaps between them, so at first glance they do not seem to satisfy the conditions demanded
by the Cosmological Principle. However, observations with radio waves and cosmic X-rays indicate
that the universe is indeed quite homogeneous at 109 light years scales. But the best indication of
the veracity of the Cosmological Principle came in 1965, when Arno Penzias (1933-) and Robert
W. Wilson (1936-) discovered the cosmic background radiation of microwaves (Fig. 2.3). This
background cosmic radiation is the residue of the thermal radiation of a much hotter past of the
universe and was predicted by the Russian physicist George Gamov (1904-1968) in 1948, as a direct
consequence of the Big Bang model. The background cosmic radiation, which gives us information
about when the universe was still very young, turns out to be very isotropic, confirming in an
extraordinary way the Cosmological Principle. Ref. [18]
2.) Weyl’s Postulate, which attempts to model the matter content of the universe. Just as the
Cosmological Principle states that density fluctuations are very small on a cosmological scale, Weyl’s
Postulate says that the velocities proper to matter are small in comparison with the cosmological
motion. Weyl’s Postulate is formulated as:
Matter at cosmological scales behaves like a perfect fluid, whose components move along timelike
geodesics, which do not intersect, except (possibly) at a point in the past.
The Weyl Postulate supposes a class of privileged observers in the universe: those that are at rest
with respect to the perfect fluid and whose motion therefore is only determined by the evolution of the
universe. These observers are usually called comoving observers. We can also define a cosmological
time, being the temporal direction of a comoving observer. This cosmological time is useful to
describe the evolution of the universe and to calculate its age.
36
Relativistic Cosmology Chapter 2
where t is the time coordinate, the function a(t) is the scale factor, a cosmological time function that
measures the expansion or the contraction of the universe, and dσ 2 is the metric in Σ, that is
where (x1 , x2 , x3 ) are coordinates on Σ and g̃ij is a maximally symmetric 3-dimensional metric. The
Riemann tensor R̃ijkl of the metric g̃ij satisfies the condition
If the space is to be maximally symmetric, then it will be spherically symmetric about every
spatial point. Therefore the metric can be written as
where r̄ is the radial coordinate. The Ricci tensor of a static, spherically symmetric spacetime is
2
R̃11 = ∂1 β,
r̄
R̃22 = e−2β (r̄∂1 β − 1) + 1,
h i
R̃33 = e−2β (r̄∂1 β − 1) + 1 sin2 θ. (2.7)
dr̄2
dσ 2 = + r̄2 dΩ2(2) . (2.9)
1 − κr̄2
37
Relativistic Cosmology Chapter 2
Notice from (2.4) that the value of κ sets the curvature, and therefore the size, of the spatial
surface.
and absorb the physical size of the manifold into the scale factor a(t). We have three cases
The physical interpretation of these cases is made by introducing a new radial coordinate χ
defined by
dr̄
dχ = √ . (2.11)
1 − κr̄2
By integrating we obtain
r̄ = Sκ (χ), (2.12)
where
sin χ, κ = +1
Sκ (χ) = χ, κ=0 (2.13)
sinh χ, κ = −1,
so that
38
Relativistic Cosmology Chapter 2
Figure 2.4: The scale factor for universes dominated by dust with different values of κ: a universe with critical density (κ = 0, blue
curve) corresponds to the static space of Einstein-De Sitter; the subcritical case (κ = −1, red curve) expands faster and approaches
to Minkowski space; the supercritical case (κ = 1, green curve) expands very fast initially, but is slowed by the large amount of
matter and will collapse in a finite time. Credits of https://ned.ipac.caltech.edu/level5/March01/Carroll3/Carroll8.html
Finally, the metric on spacetime describes one of these maximally-symmetric hypersurfaces evolv-
ing in size, and can be written as
1
2 2 2
ds = −dt + a (t) 2 2 2
dr + r dΩ(2) . (2.18)
1 − κr2
The line element (2.18) is called the Friedmann-Lemaître-Robertson-Walker (FLRW) metric,
named after the Russian physicist-meteorologist Alexander Friedmann (1888-1925), the Belgian
mathematician Georges Lemaître (1884-1966), the American physicist Howard Robertson (1903-
1961) and the English mathematician Arthur Walker (1909-2001). Friedmann proposed in 1922 the
metric (2.18) for the universe, deduced Friedmann’s equations, and obtained one of the first realistic
solutions of described expanding universe. In 1935 and 1936, Robertson and Walker independently
demonstrated that the metric proposed by Friedmann is the most general one that describes a homo-
geneous and isotropic universe. The coordinates in which the FLRW metric is written in eq. (2.18)
are usually called comoving coordinates, since the time t is the proper time of an observer that moves
with the expansion of the universe.
39
Relativistic Cosmology Chapter 2
explicitly independent of the coordinates used in the spatial sections, since the only thing we need
to know is that
All Christoffel symbols with two time indices vanish, i.e., Γa00 = Γ00β = 0, and the only non-zero
components are
ȧ i 1
Γ0ij = aȧg̃ij , Γi0j = δ , Γkij = g̃ il (∂j g̃kl + ∂k g̃jl − ∂l g̃jk ), (2.21)
a j 2
where an overdot denotes derivative with respect to the coordinate t and Γ̃kij are the Christoffel
symbols of the g̃ij metric. Similarly, the only non-zero components of the Ricci tensor, the Ricci
scalar, and the Einstein tensor are given by
ä
R00 = −3 , Rij = [2κ + aä + 2ȧ2 ]g̃ij , (2.22)
a
" 2 #
κ ä ȧ
R=6 2 + +2 , (2.23)
a a a
" 2 #
κ ȧ
Gij = − κ + 2aȧ + ȧ2 g̃ij . (2.24)
G00 =3 2 + ,
a a
To determine a(t), we must solve the system of differential equations obtained by substituting
the FLRW metric (2.19) in the Einstein equations (1.33), where the energy-moment tensor contains
the contributions of all types of energy present moment in the universe (dust, radiation, cosmological
constant, ...).
We assume the content of the universe to be a perfect fluid. We can write, therefore,
where ρ = α ρα and P = α Pα are respectively the total density and the pressure of all the types
P P
of energy and matter present in the universe. We can also express it in a more friendly way as
where uµ = (1, 0, 0, 0), this is because gtt = −1 so g tt = 1/gtt = −1, then, applying eq. (2.26) in the
metric (2.18), we obtain
ρ 0 0 0
0 P a2
0 0
(2.27)
Tab = 1 + κr2 .
P a2 r2
0 0 0
0 0 0 P a2 r2 sin2 θ
40
Relativistic Cosmology Chapter 2
Sometimes it is useful to write Einstein equations in another way. Remembering that the Einstein
equation is
Gab = 8πTab , (2.30)
we have
R = −8πT, (2.31)
Substituting eq. (2.31) in the Einstein equation (1.33), we have
1
Rab = 8π Tab − T gab . (2.32)
2
Using this expression of the Einstein equation, we obtain for (a, b) = (0, 0)
ä
−3 = 4π(ρ + 3P ) (2.33)
a
and for (a, b) = (i, j) (where Latin indices run from 1 to 3),
2
ä ȧ κ
+2 + 2 2 = 4π(ρ − P ). (2.34)
a a a
There is only one equation from (a, b) = (i, j), due to spatial isotropy. We can use eq. (2.33) to
eliminate the second derivatives in eq. (2.34), obtaining
2
ȧ 8π κ
= ρ− 2 (2.35)
a 3 a
and
ä 4π
= − (ρ + 3P ). (2.36)
a 3
Together, eqs. (2.35) and (2.36) are known as the Friedmann equations.
The Friedmann equation (2.35) relates the speed of expansion of the universe with the energy
density and with the curvature of the spatial sections. Note that it is only a first-order differential
equation and therefore is not really an equation of motion, but rather a constraint for a, ȧ, , ρ and κ,
and the Friedmann equation (2.36) is called the acceleration equation.
41
Relativistic Cosmology Chapter 2
∇a T ab = 0. (2.38)
Although eq. (2.38) is a vector equation, only the temporal component gives us a relation between
ρ, P and a(t): substituting eq. (2.21) and eq. (2.25) in the law of conservation of energy, we find in
comoving coordinates
ρ̇ + 3H(ρ + P ) = 0. (2.39)
With the law of conservation of energy we can show that the two Friedmann equations are not
really independent: differentiation the first Friedmann equation (2.35) with respect to t and using eq.
(2.39), we obtain after a bit of calculation the acceleration equation (2.36). The Friedmann equation
and the conservation of energy, therefore, imply the acceleration equation
If we work with perfect fluids, which satisfy the conservation of energy, in practice we only have
to solve the Friedmann equation to determine the evolution of the system.
To solve the energy conservation law (2.39), we need to specify what kind of energy we are dealing
with. The type of energy or matter is specified by the dependence of the pressure Pα on the density
ρα , expressed in the equation of state
Pα = w(α) ρα , (2.40)
where w(α) is the parameter of the equation of state.1 In principle, w(α) does not have to be constant,
but the homogeneity and isotropy of the FLRW metric forces w(α) to be independent of the x
coordinates. In addition, w(α) is also usually taken to be independent of t: each perfect fluid is
characterized by a value of w(α) and, as we shall see shortly, dilutes differently with the expansion
of the universe. If at different times the universe is dominated by different types of energy, it is
preferable to characterize these by several perfect fluids, rather than by a single one with a variable
state parameter in time.
• w = 0 corresponds to a perfect fluid, without pressure and therefore describes cold matter,
without interactions, or dust.
42
Relativistic Cosmology Chapter 2
Substituting the equation of state (1.89) in the energy conservation equation (2.39), we find the
differential equation for ρα in terms of a
ȧ
ρ̇α + 3(w(α) + 1) ρα = 0. (2.41)
a
By solving, we obtain
ρα (t) = ρ0 a−3(w(α) +1) (t), (2.42)
where the constant ρ0 is the density at a given time t = t0 and we normalize the scale factor as
a(t0 ) ≡ 1. Each type of perfect fluid evolves differently in the expansion of the universe
• For w = 0, the density ρ evolves as a−3 , that is, the matter is diluted inversely proportional
to the volume.
• For w = 1/3, the radiation energy density evolves as a−4 . Apart from being inversely diluted
proportional to the volume, it loses energy in the redshift of the radiation, since as the universe
expands, the wavelength of the radiation scales in a linear way with a.
• For w = −1, a cosmological constant Λ provides a constant energy density per unit volume.
An interesting observation arises when replacing the equation of state (2.40) in the acceleration
equation (2.36)
ä 4π
= − (1 + 3w)ρ. (2.43)
a 3
An accelerated expansion of the universe is only possible when the universe is dominated by a
perfect fluid with equation of state parameter w < −1/3. A universe with cold matter or radiation
will suffer a deceleration, due to the gravitational attraction of the energy content.
Guided by the scientific prejudices of the time, which required the universe to be static, and bound
by Friedmann’s equations, which make it clear that a universe dominated by normal matter (cold
matter or radiation, ...) is necessarily expanding or contracting, Einstein was forced to argue "that the
gravity equations that I have defended so far need a little modification." To get a static cosmological
solution, he had to introduce the cosmological constant Λ in eq. (1.36), with dimensions of M L−3 ,
1
Rab − gab R + gab Λ = 8πTab (2.44)
2
43
Relativistic Cosmology Chapter 2
Λ represents an energy density of the vacuum2 , ρΛ = Λ/8π and generates a type of repulsive cosmic
force if Λ > 0 and attractive one if Λ < 0. Interestingly, adding a cosmological constant is the only
way the Einstein-Hilbert action can be generalized
Z
R
(2.45)
4
p
SEH = d x |g| −Λ .
16π
If we insist on a static universe, i.e., ȧ(t) = ä(t) = 0, the acceleration equation (2.43) becomes
4π X
0=− (1 + 3w(α) )ρα . (2.46)
3 α
Taking into account that for the cold matter wd = 0, we see that the acceleration equation is
satisfied if the density of matter and the energy density of the vacuum are related as
ρd = 2ρΛ . (2.47)
This not only implies that the cosmological constant must be positive (since ρd > 0), but also
suggests a first problem with the solution: for the solution to remain static, it is necessary that the
densities of the two fluids are exactly adjusted according to the relation (2.47)
Actually there is no a priori reason to suppose that the densities behave as the solution re-
quires, since there is no dynamic mechanism that leads to it. Therefore the relationship (2.47) seems
completely ad hoc. Even so, taking into account (2.47), Friedmann’s equation (2.35) takes the form
8π κ
ρd − 2 ,
0= (2.48)
3 a
and since ρd > 0, we necessarily have that κ = 1, so, this corresponds to eq. (2.16). So, from eq.
(2.48), we obtain
8π 1 3
ρd = 2 ⇒ a2 = . (2.49)
3 a 8πρd
Then, one can represent the static Einstein universe
3 h 2 i
ds2 = −dt2 + dχ + sin2 χdΩ2(2) (2.50)
8πρd
as a cylinder, where the axis of the cylinder is time and the sections orthogonal to the axis represent
the spatial sections S 3 (Fig. 2.5). Light goes around the universe in a finite time and each observer
can see the back of his head. Objects located at the antipodes of the universe seem very close, since
the light cones of an event first diverge, but then they converge again, to coincide all at the same
moment in the opposite point of the universe they have left.
In 1930 Eddington showed that Einstein’s static universe is unstable, because it needs a delicate
balance between the gravitational attraction of matter and the repulsive force of the cosmological
constant, and any disturbance that breaks this equilibrium will cause one of the two forces to domi-
nate. It will cause the expansion or collapse of the universe. But the real coup de grace for this model
was Hubble’s discovery in 1929 that the universe is not static, but that galaxies show a redshift due
to the expansion of the universe. Einstein removed the cosmological constant from his equations and
referred to it as "the biggest mistake of his life".
2
This is a perfect fluid with wΛ = −1.
44
Relativistic Cosmology Chapter 2
Figure 2.5: In the Einstein static universe, the spatial sections are 3-spheres S 3 and light turns around the universe in a finite
time. Ref. [18]
From the FLRW metric (2.18) we can obtain the Friedmann equations with a cosmological con-
stant Λ > 0,
2
ȧ 8π κ Λ
= ρ− 2 + (2.51)
a 3 a 3
and
ä 4π Λ
= − (ρ + 3P ) + . (2.52)
a 3 3
Λ
Specifically, the de Sitter space has ρd = ρrad = 0 and ρΛ = −PΛ = > 0 with κ = 0. With
8π
these parameters, the Friedmann equation (2.51) can be integrated directly, obtaining
2 r
2 ȧ Λ ȧ Λ
H ≡ = ⇒ =
a 3 a 3
and
45
Relativistic Cosmology Chapter 2
Then, the FLRW metric (2.18) has the expression in the de Sitter space
h i
ds2 = −dt2 + a2∗ e2Ht dr2 + r2 dΩ2(2) . (2.54)
dR Rȧ
dr = − 2 dt. (2.55)
a a
This change of radius reduces the line element to the Painleve-Gullstrand form 3
The cross-term proportional to dtdR in the line element (2.56) is eliminated by the use of the
new time coordinate T defined by
HR
dT = dt + dR (2.57)
1 − H 2 R2
and dT is a locally exact differential.
Making all the necessary substitutions and calculations, we finally get the de Sitter metric in
static (Schwarzschild-like) coordinates
dR2
ds2 = −(1 − H 2 R2 )dT 2 + + R2 dΩ2(2) . (2.58)
1 − H 2 R2
In this space, two observers who are initially arbitrarily close will move away more and more,
until they literally lose sight of each of each other. There will come a time when the expansion is so
great that it does not give time to light signal to get from one observer to another and the observers
will no longer have a way to communicate. However, at the same time it is a space where nothing
changes: the acceleration is constant over time and has no matter or radiation that can be diluted.
Like the static universe, de Sitter space has no beginning, no end, it has always existed and will
continue to exist eternally.
Figure 2.6: de Sitter space can be described in FLRW coordinates with both flat (κ = 0), spherical (κ = 1) or hyperboloidal
(κ = −1) spatial sections, depending on the coordinate system chosen. Ref. [18].
3
The coordinates (t, R, θ, ϕ) are called Paintevé-de Sitter coordinates.
46
Chapter 3
Albert Einstein
We analyze alternative proposals to the Schwarzschild-de Sitter geometry for a spherical and
asymptotically flat in solution of the Einstein equations. (See Appendix A for a more detailed
calculation.)
dt2
2M
2
ds = − 1 − − H r dt2 +
2 2
+ rdΩ2(2) . (3.1)
r 2M
1− − H 2 r2
r
Here M and H = Λ/3 are positive constants and dΩ2(2) = dθ2 + sin2 θdϕ2 is the line element on
p
the unit 2-sphere. The KSdS geometry plays the role of the prototypical black hole embedded in de
Sitter space. The latter is extremely important for early universe inflation [32, 33] and is the late-
time attractor of many dark energy and modified gravity models attempting to explain the current
acceleration of the cosmic expansion [34] discovered in 1998 with type Ia supernovae.
47
Kottler/Schwarzschild-de Sitter Metric Chapter 3
Here we provide a proof of the uniqueness of the KSdS spacetime in the gauge (3.2) which employs
the areal radius R as the radial coordinate. The vacuum Einstein equations
2Ḃ
= 0, (3.4)
BR
2B 0 1 1
3
− 2 2 + 2 = Λ, (3.5)
B R B R R
2A0 B2 1
− + = −ΛB 2 , (3.6)
AR R2 R2
00
A0 B 0 RB 2 B̈ RȦḂB 2 RA0 B 0 RA B
−B − + − + = −ΛRB 3 , (3.7)
A A2 A3 A A
where an overdot and a prime denote differentiation with respect to t and R, respectively.1 Using
the consequence of eq. (3.4) that B = B(R), we drop the terms containing Ḃ or B̈ from equation
(3.7). So, eq. (3.4) gives
0
R
= 1 − ΛR2 , (3.8)
B2
which is integrated to
1
B2 = , (3.9)
C R2
1+ −Λ
R 3
where C is an integration constant. By imposing that one recovers the Schwarzschild solution for a
mass m as Λ → 0, one obtains C = −2m and
1
B 2 (R) = . (3.10)
2m ΛR2
1− −
R 3
1
The Einstein equation (3.18) gives the same information as eq. (3.3).
48
Kottler/Schwarzschild-de Sitter Metric Chapter 3
2A0 1 ΛR2 − 1
+ + = 0, (3.11)
R 2m ΛR2
A
R 1− −
R 3
and integrates to
2m ΛR2
A2 (t, R) = eD(t) 1 − − , (3.13)
R 3
where D(t) is an integration function of time. At this stage, one is not entitled to assume that Ȧ = 0.
However, by rescaling the time coordinate according to
the spherically symmetric line element necessarily takes the static form
2m ΛR2 dR2
2
ds = − 1 − − dT 2 + + R2 dΩ2(2) . (3.15)
R 3 2m ΛR2
1− −
R 3
This is the KSdS solution of the Einstein equations if Λ > 0 (and then H = Λ/3), the SAdS
p
solution if Λ < 0, and it reduces to the Schwarzschild solution if Λ = 0. The analysis of the spherical
vacuum Einstein equations mirrors that, performed for Λ = 0, which leads to the Jebsen-Birkhoff
theorem in most relativity textbooks. If is therefore appropriate to speak of a generalized Jebsen-
Birkhoff theorem when Λ 6= 0 and the KSdS solution is the unique solution of the vacuum Einstein
equations with positive cosmological constant in spherical symmetry.
49
Kottler/Schwarzschild-de Sitter Metric Chapter 3
e2Ht
ds2 = −f (t, r)dt2 + dr2 + e2Ht r2 dΩ2(2) , (3.16)
f (t, r)
where
q
f (t, r) = h(t, r) + h2 (t, r) + H 2 r2 e2Ht (3.17)
and
1 2M −Ht
h(t, r) = 2 2 2Ht
1−H r e − e . (3.18)
2 r
M is a constant mass parameter and H is the Hubble constant of the de Sitter background given
by H 2 = Λ/3. The areal radius of this spherically symmetric geometry is R(t, r) = a(t)r = eHt r.
Making use of the relation between differentials dr = a−1 (dR − HRdt), one rewrites the line element
(3.16) in terms of the areal radius as
2HR dR2
ds2 = −2h(0, R)dt2 − dtdR + + R2 dΩ2(2) = (3.19)
f (0, R) f (0, R)
2m
=− 1− − H R dt2 =
2 2
R
4HR
=− q dtdR
1 − 2m 2m 2 R2 2 + 4H 2 R2
− H 2 R2 + 1 − − H
R R
2dR2
+ q 2 + R2 dΩ2(2) . (3.20)
2m 2m
1− R − H 2 R2 + 1− R − H 2 R2 + 4H 2 R2
dT = dt + β(R)dR (3.21)
50
Kottler/Schwarzschild-de Sitter Metric Chapter 3
By setting
2HR
β(R) = p , (3.24)
A0 A0 + A20 + 4H 2 R2
the cross-term in dT dR is eliminated and the line element assumes the diagonal and locally static
form
2 2
2H R 2
ds2 = −A0 (R)dT 2 + p
2
1 + dR2 + R2 dΩ2(2)
A0 (R) + A0 (R) + 4H R2 2
p
2
A0 A0 + A0 + 4H R 2 2
(3.25)
which is not of the KSdS form. A posteriori one can check that dT = dt+βdR is an exact differential
(i.e., the time coordinate T is well defined) by noting that it is closed,
∂(1) ∂β
=0= . (3.26)
∂R ∂t
Although claiming a new solution alternative to the KSdS one, Abbassi [52] mentions a coordinate
transformation that brings the line element (3.16) to the standard KSdS form, but this coordinate
change fails to do so. Moreover, this author ascribes different physical meanings to the same geometry
described in different coordinate systems. The geometry, however, must be coordinate independent.
In particular, the static character of the metric is shown by the existence of a timelike Killing vector
field. In spite of what stated in [52, 53] the diagonal metric (3.16) does not solve the vacuum Einstein
equations Rab = Λgab but it is generated by matter sources. For example, there is a radial mass flow
given by
1 d(ln B 2 )
T01 = =
8π dT
H ḢR2
= p p
2π A0 + A20 + 4H 2 R2 A20 + 4H 2 R2
p
A0 A0 A20 + 4H 2 R2 + A20 + 2H 2 R2
−1 + p h p i 6= 0. (3.27)
A20 + 4H 2 R2 A0 A0 + A20 + 4H 2 R2 + 2H 2 R2
51
Kottler/Schwarzschild-de Sitter Metric Chapter 3
where M is a positive constant related to the mass of the central object and H(t) is the Hubble
parameter of the FLRW space in which this object is embedded. When the FLRW “background”
reduces to de Sitter, H = const., the transformation to the coordinate T given by
HR
dT = dt + q dR (3.29)
2m 2m
1− R (1 − R − H 2 R2 )
reduces the metric to the KSdS form (3.1). Therefore, the McVittie metric with H = const. is not
an alternative to KSdS but it contains it as a special case.
In the literature there is also a class of "generalized McVittie solutions" in which, contrary to
the original McVittie one, there is a spacelike radial heat flow q µ = (0, q, 0, 0) [4, 7]. McVittie
spaces are also solutions of cuscuton theory (a special case of Hořava-Lifschitz gravity [57]) and
generalized McVittie spaces are also solutions of Horndeski gravity and shape dynamics [58]. They
are substantially more complicated than the McVittie one, but they also reduce to the KSdS geometry
when the background is de Sitter [4, 7], in which case the spacelike radial energy flow q a vanishes.
where a(η) is the scale factor of the FLRW background, η and t are its conformal and comoving
times, respectively, with dt = adη, and m is a constant mass parameter. The line element (3.30) is
manifestly conformal to the Schwarzschild one. By using the areal radius R(t, r) = a(t)r and the
a − HRdη, the line element is rewritten as
relation between2 differentials dr = dR
!
2M (t) H 2 R2 dR2 2HR
2
ds = − 1 − − dt2 + − dtdR + R2 dΩ2(2) , (3.31)
R 1 − 2MR(t) 1− 2M (t)
R 1− 2M (t)
R
where
2
Here H ≡ ȧ/a and an overdot denotes differentiation with respect to the comoving time t.
52
Kottler/Schwarzschild-de Sitter Metric Chapter 3
The cross-term in dtdR can be eliminated from this line element [8]. Then, we proceed to use
A(t, R) ≡ 1 − 2M/R = 1 − 2m/r and a new time coordinate T defined by
1 HR
dT = dt + 2 dR (3.33)
F A − H 2 R2
∂ 1 ∂ HR
= (3.34)
∂R F ∂t F (A − H 2 R2 )
2
to guarantee that dT is an exact differential. Straightforward manipulations bring the line element
to the diagonal gauge
!
2M H 2 R2 dR2
ds2 = − 1 − − F 2
dT 2
+ H 2 R2
+ R2 dΩ2(2) . (3.35)
R 1 − 2MR 1 − 2M
R − 2M
1− R
Using the form (3.30) of the metric, the Einstein equations give [60]
3H 2
G00 = 8πT00 = − , (3.36)
A
2mH
G01 = 8πT10 = − 2 2 , (3.37)
r A
1 2ä
1 1 2
G1 = 8πT1 = 8πT2 = − 2
H + . (3.38)
A a
Assume a de Sitter background with H = Λ/3 and a(t) = a0 eHt ; then the time-radius Einstein
p
equation (3.37) satisfied by the non-rotating Thakurta solution clearly cannot reduce to the corre-
sponding equation satisfied by the KSdS metric, which would instead give 8πT10 = −Λg10 = 0 (the
vanishing of T10 means that, because the cosmological constant is repulsive, it does not accrete onto
a black hole and there is no radial energy flow). The two equations only coincide in the trivial cases
when m = 0 (de Sitter space) or when a = const. (Minkowski background). These two equations
cannot coincide because, as stated clearly in [60, 61], the source of the non-rotating Thakurta ge-
ometry is not a perfect fluid, to which the cosmological constant can be reduced, but is instead an
imperfect one with a spacelike radial heat flow which has components qµ = (0, −2mȧaA−3/2 /r2 , 0, 0)
in coordinates (t, r, θ, ϕ) [60].
It has been shown in Refs. [3, 9] that the non-rotating Thakurta solution is the late time attractor
of generalized McVittie solutions, but these references3 did not recognize the geometry as a special
case of the less known Thakurta solution and called it "comoving mass solution" instead. The non-
rotating Thakurta solution is also the limit to general relativity of a class of solutions of Brans-Dicke
theory found in Ref. [63] as the Brans-Dicke parameter ω → ∞ [3, 9].
3
Ref [61] studied the same geometry for different purposes and did not identify it with the Thakurta solution.
53
Kottler/Schwarzschild-de Sitter Metric Chapter 3
where α and m are constants and H = H(t) is the Hubble parameter of the FLRW "background".
This geometry does not satisfy the vacuum Einstein equations Gab = −Λgab but is sourced by an
imperfect fluid which has different tangential and radial pressures if α 6= 0 [64]. The metric (3.39)
reduces to the McVittie metric in the form (3.28) when α = 0 (in which case the two pressures
coincide). In spite of superficial similarities, it does not reduce to the non-rotating Thakurta solution
(3.35) for α = −1. Similarities and differences may be misleading because they depend on the
coordinates adopted. Let us change the time coordinate t → T , where T is defined by
1
dT = (dt + βdR) (3.40)
F
where 1/F is an integrating factor and β(t, R) is a function to be determined. The line element
(3.39) becomes
2m α
2 2m 2 2
ds = − 1 − −H R 1− F 2 dT 2
R R
( α α−1 )
2m 2m 1 2m 2
+ − 1− − H 2 R2 1 − β2 + + 2HRβ 1 − dR2
R R 1 − 2m
R
R
( α α−1 )
2m 2m 2m 2
+ 2F 1− − H 2 R2 1 − β − HR 1 − dT dR + R2 dΩ2(2) . (3.41)
R R R
By setting
α−1
2m
HR 1 − 2
β(t, R) = 2m
R ,
2m α
(3.42)
1− R − H 2 R2 1 − R
the cross-term in dT dR is eliminated and one obtains the line element in the diagonal gauge
2m α dR2
2m
ds2 = − 1 − − H 2 R2 1 − F 2 dT 2 + 2m
+ R2 dΩ2(2) . (3.43)
2m α
R R 2
1− R −H R 1− R2
If the background is de Sitter then H = const., β = β(R), and F = 1, and the line element (3.43)
reduces to the non-rotating Thakurta solution (3.35) for α = −1 and to the KSdS form (3.1) (which
is a special case of McVittie) for α = 0. It is clear, however, that in the general case the geometry is
different from the KSdS one.
54
Kottler/Schwarzschild-de Sitter Metric Chapter 3
and assume spherical symmetry, in which case the line element is given by eq. (3.2). Assume that
the solution of the Einstein equations is asymptotically de Sitter, that is, that there is a de Sitter-like
cosmological horizon of areal radius RH and the solution of the Einstein equations (3.44) reduces to
(3.1) as4 R → R− H.
Ḃ
= 4πT01 , (3.45)
BR
2B 0
1 1
A 2
− + = ΛA2 + 8πT00 , (3.46)
B 3 R B 2 R2 R2
2A0 B2 1
− 2 + 2 = −ΛB 2 + 8πT11 , (3.47)
AR R R
A0 B RB 2 B̈ RȦḂB 2 RA0 B 0 RA00 B B3
− B0 − + − + = (−ΛR 2
+ 8πT22 ) . (3.48)
A A2 A3 A A R
Further assume that matter is described by an imperfect fluid with constant equation of state
and a purely spatial radial heat flow (of the kind considered in the previous section),
55
Kottler/Schwarzschild-de Sitter Metric Chapter 3
The fluid 4-velocity and the radial energy flow have components
T00 = A2 ρ, (3.54)
T01 = −|A|B q,2
(3.55)
2
T11 = B P, (3.56)
T22 = R2 P, (3.57)
2
T33 = R P sin θ. 2
(3.58)
Here, (
T01 > 0 and q < 0 correspond to radial inflow,
T01 < 0 and q > 0 correspond to radial outflow.
therefore, the metric component B 2 = g11 increases with t. Assuming the metric coefficients to be
continuous and differentiable, there are then two possibilities:
1 ) lim B 2 (t, R) → +∞ or lim B 2 (t, R) → +∞ ∀R
t→∞ t→tmax
Assuming the metric coefficients to be continuous and differentiable, there are then two possibil-
ities: either B 2 (t, R) → +∞ for any fixed R as t → +∞ (or as t → tmax if there is a singularity at a
finite future tmax ), or B 2 (t, R) has an horizontal asymptote as t → +∞. Let us consider each case:
∇c R∇c R = 0,
56
Kottler/Schwarzschild-de Sitter Metric Chapter 3
horizon (which has areal radius RAH = H −1 in a spatially flat FLRW cosmos [3]) shrinks around a
comoving observer because the expansion of the universe superaccelerates, that is [16]
Ḣ = −4π(P + ρ) > 0.
By contrast, in a de Sitter space the Hubble parameter H remains constant although the expansion
itself accelerates, ä > 0. In a universe dominated by non-phantom dark energy (other than the
cosmological constant), it is instead Ḣ < 0 while ä > 0. These phantom asymptotics contradict our
assumption of de Sitter asymptotics and, therefore, we discard this possibility.
2 ) In this case, Ḃ → 0 as t → +∞ (which also implies that the apparent horizons located by the
equation 1/B 2 = 0 become less and less dynamical). Then equation (3.59) implies that the radial
flow q → 0 as t → +∞. In conjunction with eq. (3.54), the differentiation of eq. (3.46) yields
2 B0 ·
·
1 1
8π ρ̇ = − → 0 as t → +∞. (3.60)
R B3 R2 B 2
The assumption that P = wρ with constant w (or with w = w(R)) then implies that also Ṗ → 0
as t → +∞. Equations (3.56) and (3.47) give
0 · · 0 ·
2 A 1 1 2 A
8π Ṗ = 2
+ 2 2
≈ 2
→ 0 as t → +∞. (3.61)
R AB R B RB A
Therefore, also A2 becomes time-independent, and the metric becomes static as t → +∞.
To make progress, consider the covariant conservation equation ∇b Tab = 0 for the imperfect fluid
stress-energy tensor (3.49), which yields
Projecting this equation onto the time direction ua of comoving observers and using the orthog-
onality of 4-velocity and 4-acceleration ua ∇b ua = 0, one obtains
At late times q c and ρ̇ disappear from this equation, as we have deduced above, leaving
In general ∇b ub is different from zero (indeed, since the geometry must be asymptotically de
Sitter at large radii, ∇b ub reduces to 3H > 0 there) and we are left with P + ρ → 0 as t → +∞.
Either the matter fluid reduces to a cosmological constant, in which case the vacuum uniqueness
theorem for KSdS holds, or else both ρ and P = wρ become subdominant and the cosmological
constant dominates the expansion at late times while ρ and P become unimportant. Also in this
case the solution reduces to KSdS.
57
Kottler/Schwarzschild-de Sitter Metric Chapter 3
and, since B 2 is bounded from below by zero and it decreases as t → +∞, it must have a horizontal
asymptote with B 2 (t, R) → B02 (R)+ for any fixed R as t → +∞. Then Ḃ → 0 and q → 0. The
reasoning made in the case with inflow is then repeated from this point, reaching the same conclusion.
Hence it is proved that, assuming spherical symmetry, Λ > 0 and spatial de Sitter asymptotics, and
an imperfect fluid with constant equation of state and purely spatial radial energy flow, the late time
solution of the Einstein equations must be the KSdS geometry.
As a special case, one can consider a perfect fluid by setting q a = 0. In this case T01 = 0 and
eq. (3.45) gives B = B(R). It is then straightforward to prove that it must be P = −ρ and that the
KSdS geometry can be the only solution (this simple proof for a perfect fluid was already given in
Ref. [4]).
58
Chapter 4
Scalar-Tensor Theory
Albert Einstein
Robert H. Dicke and Carl H. Brans, in 1961, formulated a theory of gravitation based primarily
on the Mach principle. The theory contains a scalar field which describes gravity together with the
metric tensor familiar from general relativity. But, what does the Mach principle say? It says that
the inertia of any system is the result of its interaction with the rest of the Universe. In other words,
each particle of the universe exerts an influence on all other particles. This theory was, and still is,
one of the alternatives to general relativity most discussed. The reasons are:
• Brans-Dicke theory can be obtained as a derivation of a theory of Kaluza and Klein in which
the scalar field is generated by the presence of compactified extra dimensions, an essential aspect of
all modern unified theories.
• There is interest in these theories for their cosmological applications; it is believed that the
convergence of Brans-Dicke theory to general relativity could occur during the era dominated by
matter, or even during the inflationary phase of the early universe.
The problem starts from the confrontation between Newton and Leibnitz. The laws of Newton
are always referred to reference systems called inertial. The question is how to determine such
systems. For Newton there is an absolute space and the inertial systems are those that are at rest or
in uniform motion with respect to said absolute space. But for Leibnitz it was that it should not be
necessary to define a space independence of material objects. It was Ernst Mach who proposed the
idea that a phenomenon of inertia is due to accelerations with respect to the mass distribution of the
universe. The inertial masses of the various elementary particles would not be fundamental constants,
but would represent the interaction of the particles with some kind of cosmic field. But, since the
59
Scalar-Tensor Theory Chapter 4
inertial mass of the particles can only be determined by measuring the gravitational acceleration, an
equivalent conclusion is that the universal gravitation constant GN should be related to the average
value of a scalar field φ coupled to the mass density of the Universe. Making use of this assumption,
Brans and Dicke formulated the theory that bears their names.
However, the most important motivation in scalar-tensor gravity is due to use of these theories in
inflationary scenarios of the universe. While the attractor solution of general relativity in phase space
(the de Sitter exponentially expanding universe) fails to provide an exit from an inflationary epoch,
its analogue in Brans-Dicke theory (a power-law attractor solution) does the job. More recently, the
discovery that the universe is undergoing an accelerated expansion today, and the approximately
70% of the energy of the universe is in a form with negative pressure (dark energy), has interest in
scalar-tensor theories which account for such features with a long-range gravitational scalar field.
where Weil or conformal transformation Ω is the conformal factor. Conformal transformations stretch
or shrink the distances between the two points described by the same coordinates xa in the spaces
M and M˜, but keeping the angles between vectors, in particular the light-type vectors that define
the light cones. If we take constant Ω we find the so-called scale transformations and we can see the
conformal transformations as localized scale transformations. A vector that is timelike, spacelike, or
null with respect to the metric gab has the same character with respect to g̃ab , and vice-versa. The
following transformation properties of geometric quantities are useful [17, 66],
where n is the dimension of the spacetime manifold M . Then, the Christoffel symbols or Levi-Civita
connection are transformed as
where ∇a is the covariant derivative associated with the Levi-Civita connection of the metric. In the
same way, ∀, n ≥ 2, the tensor and scale of Ricci are transformed as
d d
R̃abc = Rabc + 2δ d [a ∇] b∇c (ln Ω) − 2g de gc [a ∇] b∇e (ln Ω)
d
+ 2∇ [a (ln Ωδb] ∇c (ln Ω) − 2∇ [a (ln Ω)gb] c g de ∇e (ln Ω)
d ef
− 2gc [a δb] g ∇e (ln Ω)∇f (ln Ω) (4.4)
60
Scalar-Tensor Theory Chapter 4
(Ua Vb −Ub Va )
where U [a Vb] = 2 denotes anti-symmetrization,
g ab ∇a Ω∇b Ω
−2
ab
R̃ ≡ g̃ R̃ab = Ω R − 2(n − 1) 2 (ln Ω) − (n − 1)(n − 2) . (4.6)
Ω2
In the case of n = 4 spacetime dimensions, the transformation property of the Ricci scalar can
be written as
62Ω
R̃ = Ω−2 R −
Ω
" √ ab ∇ Ω∇ Ω
#
12 2 Ω 3g a b
= Ω−2 R − √ + . (4.7)
Ω Ω2
but the same tensor with index raised or lowered with respect to Cabc
b is not. This property justifies
dr2
2 2 2
ds = −dt + a (t) 2 2 2
+ r (dθ + sin θdϕ )2
(4.9)
1 − κr2
4 √
Z
R 1 cd
S = d x −g − g ∇c φ∇d φ − V (φ) (4.10)
16πGN 2
where no matter action is included other than φ, since the inflaton field φ dominates the dynamics
during inflation.
61
Scalar-Tensor Theory Chapter 4
(φ̇)2
ρ = Tab ua ub = + V (φ), (4.12)
2
Tii (φ̇)2
P = Tab hab = = − V (φ), (4.13)
gii 2
where hba is the projection operator on the three-dimensional spatial sections defined by
The scalar field potential V (φ) is taken to be positive because during inflation, in which φ̇ ' 0
and V (φ) ' ρ is dominated by the potential of the scalar. Then, the Friedmann equations assume
the form
" #
8πG ( φ̇)2 κ
N
H2 = + V (φ) − 2 , (4.15)
3 2 a
" #
8πG ( φ̇)2
N
Ḣ = −H 2 − − V (φ) , (4.16)
3 2
Only two of the three equations (4.15)-(4.17) are independent. In fact, if φ̇ 6= 0, the Klein-
Gordon equation (4.17) can be derived from the conservation equation (2.39). The equations of
inflation (4.15)-(4.17) and those satisfied by cosmological perturbations are solved in the slow-roll
approximation
(φ̇)2
V (φ), φ̈ H φ̇ (4.18)
2
in which eqs. (4.15) and (4.17) reduce to
8πGN V
H2 ' , (4.19)
3
dV
3H φ̇ ' − . (4.20)
dφ
62
Scalar-Tensor Theory Chapter 4
The slow-roll approximation assumes that the solution φ(t) of the field equations rolls slowly over
a shallow plateau of the scalar field potential V (φ). Then, the expansion of the universe is almost
de Sitter-like
where
4 √
Z
ω cd
Sφ = d x −g − g ∇d φ∇c φ − V (φ) . (4.24)
φ
If we assume now couplings between both fields, the generalized action in the co-called Jordan
frame can be written as
4 √
Z
1 ω cd
S BD
= d x −g φR − g ∇d φ∇c φ − V (φ) + S (m) , (4.25)
16π φ
where
√
Z
S (m)
= d4 x −gL (m) (4.26)
is the action describing ordinary matter and ω is the (dimensionless) Brans-Dicke parameter.
63
Scalar-Tensor Theory Chapter 4
It is important to point out that, although initially Jordan admitted a scalar field that was
included in the Lagrangian of matter, Brans and Dicke did not, since only in this way is it possible to
preserve the Weak Equivalence Principle (WEP), 1 , the equations of motion in a gravitational field
are not modified because they depend only on the metric g and not on the scalar φ.
The factor φ in the denominator of the second term in brackets in the action (4.25) is introduced
to make ω dimensionless. Matter is not directly coupled to φ, in the sense that the Lagrangian
density L (m) does not depend on φ ("minimal coupling" for matter) but φ is directly coupled to the
Ricci scalar.
The gravitational field is described by the metric tensor gab and by the Brans-Dicke scalar field
φ, which together with the matter variables describe the dynamics. The scalar field potential V (φ)
constitutes a natural generalization of the cosmological constant and may reduce to a constant, or
to a mass term.
By varying the action (4.25) with respect to gab , and using the well-known properties [94] we
have the analogy to Einstein’s evolution equations
√ 1√
dg = g ab dgab −→ δ( −g) = − −ggab δg ab , (4.27)
2
√ √ √
1
δ( −gR) = −g Rab − gab R δg ab ≡ −gGab δg ab , (4.28)
2
and Gab is the Einstein tensor. One then obtains the field equations
8π (m) ω 1 1 V
Gab = c
Tab + 2 ∇a φ∇b φ − gab ∇ φ∇c φ + (∇a ∇b φ − gab 2 φ) − gab , (4.30)
φ φ 2 φ 2φ
where
−2 δ √
T (m) ≡ √ −gL (m)
(4.31)
−g δg ab
is the stress-energy tensor of ordinary matter. Variation of the action with respect to φ yields
2ω ω dV
2 φ + R − 2 ∇c φ∇c φ − =0 (4.32)
φ φ dφ
8πT (m) ω 3 2 φ 2V
R=− + 2 ∇c φ∇c φ + + (4.33)
φ φ φ φ
1
The Weak Equivalence Principle (WEP) also known as the universality of free fall or the Galilean equivalence principle, assumes falling bodies
are bound by non-gravitational forces only.
64
Scalar-Tensor Theory Chapter 4
and using the result of equation (4.33) to eliminate R from equation (4.32), one obtains
1 dV
2φ = 8πT (m)
+φ − 2V . (4.34)
2ω + 3 dφ
In eq. (4.34), the scalar φ has non-conformal matter (matter with trace) as its source, but the
scalar is not coupled directly to T (m) or L (m) . The field φ acts back on ordinary matter only through
the metric tensor gab in the manner described by eq. (4.30). The term proportional to φdV /dφ − 2V
on the right hand side of eq. (4.34) vanishes if the potential V (φ) consists of a pure mass term.
The form of the action (4.25) or of eq. (4.30) suggests that the Brans-Dicke field φ plays the role
of the inverse of the gravitational coupling
1
Gef f (φ) = , (4.35)
φ
which becomes a function of the spacetime point.
the time-time component of the Brans-Dicke field equation (4.30) yields the constraint equation
!2
8π (m) ω φ̇ φ̇ κ V
H =2
ρ + −H − 2 + (4.39)
3φ 6 φ φ a 6φ
κ
which provides a first integral. The trace equation (4.33) yields, using R = 6 Ḣ + 2H 2 + 2 ,
a
!2
κ 4πT (m) ω φ̇ 1 2φ V
Ḣ + 2H 2 + 2 = − − + + . (4.40)
a 3φ 6 φ 2 φ 3φ
By using eqs. (4.34) and (4.39) and the expression of the trace
65
Scalar-Tensor Theory Chapter 4
one obtains
!2
8π h i ω φ̇ φ̇ κ 1 dV
Ḣ = − (ω + 2)ρ(m) + ωP (m) − + 2H + 2 + φ − 2V .
(2ω + 3)φ 2 φ φ a 2(2ω + 3)φ dφ
(4.42)
dV φ2
The combination 2V − φ in equations (4.42) and (4.43) vanishes when V (φ) = m2 and
dφ 2
in this case the scalar field mass does not directly affect the dynamics of φ. When V (φ) = 0 or
φ2
V (φ) = m2 and in the absence of ordinary matter, or in the presence of a radiative fluid with
2
equation of state P (m) = ρ(m) /3, eq. (4.43) can be integrated, yielding the solution φ = const. or
the equation
C
φ̇ = , (4.44)
a3
where C is an integration constant. Equations (4.39) and (4.43) can be regarded as the two indepen-
dent equations regulating the dynamics of spatially homogeneous and isotropic BD cosmology. When
(m)
ordinary matter is described by a perfect fluid stress-energy tensor Tab , the conservation equation
(m)
∇b Tab = 0 becomes
˙ + 3H(ρ(m) + P (m) ) = 0.
ρ(m) (4.45)
Equation (4.43) is of second order in φ(t), while eq. (4.39) and (4.45) are of first order in a, φ(t)
and ρ, and one must specify an equation of state to solve them. Then, the equation of state
with γ = const. is assigned, and the equations solved. Integrating (4.45) obtaining
C0
ρ(m) = (4.47)
a3γ
where C is an integration constant.
The contribution of the scalar field φ to the right hand side of eq. (4.30) is often described as
that of another cosmic fluid.
66
Scalar-Tensor Theory Chapter 4
(4.48)
p
Ω = GN φ.
(where φ 6= 0, ω > −3/2 and φ0 = G−1N ) brings the scalar field kinetic energy density into canonical
form. The resulting Einstein frame action is
( " # q )
πGN
√
Z
R̃ 1 −8 2ω+3 φ̃
S = d4 x −g ˜ a φ̃∇
− g̃ ab ∇ ˜ b φ̃ − U (φ̃) + e L (m) [g̃] (4.50)
16πGN 2
where ∇
˜ a is the covariant derivative operator of the rescaled metric g̃ab and
v
πGN
u
u
−8t φ̃
U (φ̃) = V [φ(φ̃)]e 2ω + 3 . (4.51)
Note that the Jordan frame scalar φ has dimensions of G−1 N , while the Einstein frame scalar φ̃ has
−1/2
dimensions of GN and can be measured in Planck masses. If φ becomes constant (in the limit to
general relativity), the Jordan and the Einstein frame coincide. By looking only at the action (4.50)
one could say that, in the Einstein frame gravity is described by general relativity, but there are two
important differences.
• First, the free scalar φ̃ acting as a source of gravity, i.e., one cannot consider solutions of the
vacuum field equations R̃ab = 0 in the Einstein frame as one can do in general relativity. The field φ̃
permeates spacetime and cannot be eliminated, a feature that is reminiscent of the cosmological origin
of φ in the original Brans-Dicke theory. This is in spite of the fact that formally the gravitational
field is only described by the metric tensor g̃ab in the Einstein frame. The conformal transformation
moves part of the Jordan frame gravitational variables (φ) into Einstein frame matter (φ̃).
• Second, the difference between Einstein frame Brans-Dicke theory and general relativity is the
matter part if the Lagrangian is now multiplied by the exponential factor in equation (4.50), which
described an "anomalous"2 coupling of the ordinary matter to the scalar φ̃.
67
Scalar-Tensor Theory Chapter 4
action (2.45). These theories come about by a simple generalization of the Lagrangian, to become a
general function of R:
√ √
Z Z
1 1
S= d4 x −gR −→ S = d4 x −gf (R). (4.52)
2κ 2κ
See Appendix B for a more detailed discussion.
68
Chapter 5
Leonardo da Vinci
To continue, we clarify the problem of the most general solution of vacuum Brans-Dicke theory
[79]. The result constitutes an analogue of the Jebsen-Birkhoff theorem of general relativity in
Brans-Dicke theory. We work under the following physically reasonable assumptions:
(1) the vacuum Brans-Dicke equations in the Jordan frame hold with ω 6= −3/2,
(2) the spacetime metric is spherically symmetric, static, and asymptotically flat (staticity re-
flects a state of equilibrium, while asymptotic flatness characterizes isolated objects),
(3) the Brans-Dicke scalar φ depends only on the radial coordinate r ; it does not have poles or
zeros (except possibly for a central singularity), and φ(r) becomes constant as r → +∞.
√
−g
Z
ω c
SBD = 4
d x φR − ∇ φ∇c φ , (5.1)
16π φ
where R is the Ricci scalar and ω is the constant Brans-Dicke coupling, gives rise to the field equations
1 ω 1 1
Rab − gab R = 2 c
∇a φ∇b φ − gab ∇ φ∇c φ + (∇a ∇b φ − gab 2 φ) , (5.2)
2 φ 2 φ
2 φ = 0. (5.3)
69
The Analogue of the Jebsen-Birkhoff Theorem in Brans-Dicke Gravity Chapter 5
−(A+B)
2η 2
φALC (r) = φ0 1 − , (5.5)
r
and
ω+1 (A + B)2
1− = (5.6)
ω+2 2(1 + AB)
where A, B, and η are real constants.
In [80], this is assumed to be a gauge choice valid for any solution satisfying (1)–(3), but at this
stage this is instead a choice of a special solution, the Campanelli-Lousto one. The general form of
the Campanelli-Lousto solution of eqs. (5.4) and (5.5) is [81]
2η b0 +1 2 2η −a0 −1 2 2η −a0 2 2
2
dsCL = − 1 − dt + 1 − dr + 1 − r dΩ(2) , (5.7)
r r r
a0 −b0
2η 2
φCL (r) = φ0 1 − , (5.8)
r
where a0 and b0 are two parameters satisfying
−2(a20 + b20 − a0 b0 + a0 + b0 )
ω= (5.9)
(a0 − b0 )2
where
(
a0 = −B − 1,
(5.10)
b0 = A − 1.
Therefore, the results of [80] are true only for this particular solution. In spite of being advertised
as black holes, the Campanelli-Lousto family contains only wormhole throats and naked singularities
[13]. The conflict with the no-hair theorems is then resolved [45]. But what are the solutions satisfying
(1)–(3) which are not Schwarzschild?
70
The Analogue of the Jebsen-Birkhoff Theorem in Brans-Dicke Gravity Chapter 5
Since the conformal factor is Ω = φ(r), the Einstein frame geometry is also spherical, static,
p
and asymptotically flat. Formally, the action (5.1) describes general relativity with a free, minimally
coupled scalar field and the most general spherical, static, asymptotically flat solution is known to
be the Fisher-Janis-Newman-Winicour-Buchdahl-Wyman (FJNWBW) solution of general relativity
[82, 83]
4 2
−α/r γ/r −α/r γ/r
2 α/r 2
ds̃ = −e dt + e 2
dr + e r2 dΩ2(2) (5.14)
sinh(γ/r) sinh(γ/r)
where α and γ are constants, and with scalar field [82, 83]
φ∗ −σ
φ̃ = , φ∗ = √ , (5.15)
r 4 π
where σ is a scalar charge and one can take γ ≥ 0 without loss of generality. These three constants
are related by [83]
4γ 2 = α2 + 2σ 2 . (5.16)
• If σ = 0, the Einstein frame scalar vanishes, the Jordan frame scalar reduces to a constant, the
theory reduces to general relativity, and the solution reduces to Minkowski in both conformal frames,
which then coincide. In fact, the constants α and γ both vanish whenever σ does, thus turning (5.14)
into the Minkowski metric. However, as the notation followed here is that of Ref. [83], the relation
(5.16) between the constants α, γ and σ does not allow one to see this fact as it only implies 4γ 2 = α2
when σ = 0. The vanishing of α and γ could be seen only when tracing back the steps that led to
expression (5.14) as presented in Ref. [83], for then one clearly sees that whenever σ vanishes, so
does the constant α, which, in turn, makes γ vanish as well.
• If σ 6= 0, mapping the FJNWBW solution back to the Jordan frame, one obtains the most
general solution of the Brans-Dicke equations under the assumptions (1)–(3).1 Equation (5.12) yields
1
A remark to this regard was made in passing in [84].
71
The Analogue of the Jebsen-Birkhoff Theorem in Brans-Dicke Gravity Chapter 5
This is the most general solution of Brans-Dicke theory under the assumptions (1)–(3). It is
related to a Campanelli-Lousto solution.
It should be noted here that in Ref. [79], the most general solution of the generalized Brans-Dicke
scalar-tensor theory has also been found by Bronnikov for the case of electrovacuum. For the case of
vacuum, explicit forms of the solution corresponding to imaginary γ, for which the sinh function in
the metric (5.18) is replaced by the sine function, were given there.
The latter possibility, corresponding to what has been called in Refs. [76] a "cold black hole",
arises for the anomalous case 2ω + 3 < 0. This case, which we avoided in this paper by taking care
of using the absolute value of 2ω + 3 in our field redefinition (5.12), is anomalous for it makes the
Einstein frame field φ̃ imaginary which, in turn, makes the kinetic term in the Einstein frame action
(5.13) acquire the wrong sign. This case gives the ghost counterpart of the solution (5.14) and (5.15)
due to Bergman and Leipnik [82]. Indeed, when an imaginary field φ̃ is allowed, the scalar charge σ
becomes imaginary and the Wyman relation between the various constants becomes [79]
− 4γ 2 = α2 − 2σ 2 . (5.19)
The negative signs can be absorbed by letting both σ and γ be imaginary. This then turns the
sinh function into a sine function in (5.18).2 In Ref. [85], the special cases α = β, α = (2ω + 3)β,
and α = −(ω + 1)β in (5.18) were found explicitly.
Much later, a more exhaustive investigation of the general solutions of the Bergmann-Wagoner
class of scalar-tensor theories, in which Brans-Dicke gravity is a special case, was made in Ref.
[77]. It was shown there that, among these solutions, black hole geometries arise for the anomalous
versions of these theories. The thermodynamics of such black holes, also dubbed cold black holes,
were investigated.
Let us now come back to our general solution. When γ 6= 0 in (5.18), by performing the two
consecutive coordinate transformations
B 2
1 + B/ρ
eγ/r = , r̄ = ρ 1 + , (5.20)
1 − B/ρ ρ
2
For completeness, we give here the general solution of Brans-Dicke theory in the anomalous case.
α+β β−α 2 2
γ/r γ/r
It reads: ds2 = −e r dt2 + e r
sin(γ/r) sin(γ/r)
dr 2 + r 2 dΩ2
(2) .
72
The Analogue of the Jebsen-Birkhoff Theorem in Brans-Dicke Gravity Chapter 5
and setting
√
η = 2B = m2 + σ 2 ,
m/η = −α/(2γ),
p
σ/m = β |2ω + 3|/(2γ),
and rescaling the time coordinate by a factor |γ/(2B)|, the solution (5.18), (5.17) becomes
1 √ σ 1 m+ √ σ
2η η m− |2ω+3| 2η − η
|2ω+3|
2 2
ds = − 1 − dt + 1 − dr̄2
r̄ r̄
1 √ σ
2η 1− η m+ |2ω+3| 2 2
+ 1− r̄ dΩ(2) , (5.21)
r̄
√ σ
2η η |2ω+3|
φ = φ0 1− , (5.22)
r̄
which is a Campanelli-Lousto solution (5.7), (5.8) with
!
1 σ
a0 = −1 +
m+ p ,
η |2ω + 3|
! (5.23)
1 σ
b0 = −1 + m− p .
η |2ω + 3|
It must be noted here that, although this form contains only the absolute value of the term 2ω +3,
the anomalous case 2ω + 3 < 0 discussed above does not apply here as the coordinate redefinition
(5.20) would not be real-valued anymore since γ is imaginary in this case. Therefore, the Campanelli-
Lousto metric (5.7), as well as its other version (5.4) used in Ref. [80], is only valid for the normal
case 2ω + 3 > 0.
73
The Analogue of the Jebsen-Birkhoff Theorem in Brans-Dicke Gravity Chapter 5
while it must be 2Φ = 2(φeΨ ) = 0 in order for Φ to still be a solution. Using this equation to
eliminate ∇r ∇r Ψ in equation (5.25), one finds
2∇r φ
∇r Ψ + ∇r Ψ = 0 (5.26)
φ
which (apart from the irrelevant possibility Ψ = const.) integrates to Ψ(r) = −2 ln φ + const. and
Φ = C0 /φ. However, replacing (gab ; φ) with (gab ; Φ) = (gab ; C0 /φ) amounts to changing the exponent
β in eq. (5.17) into −β, or to changing the sign of the scalar charge σ, a possibility already included
in the form of the general solution (5.18), (5.17).
It must be noted, however, that what makes the conformal transformation one to one and protects
(5.18) and (5.17) against such redefinitions as φ → φeΨ(r) , that could have prevented them from being
the most general solution, is the homogeneous wave equation 2φ = 0. The latter, in turn, is always
guaranteed to hold in vacuo and electrovacuo for which the matter energy-momentum tensor is
traceless. Therefore, we conclude that (5.18) and (5.17) constitute indeed the most general solution
with the assumptions (1)–(3) above. Moreover, this analysis also applies, and therefore reinforces,
Bronnikov’s general solutions for vacuum and electrovacuum scalar-tensor theories found in Ref. [79].
is the areal radius. Horizons correspond to the roots of that equation; a single root describes a black
hole horizon while a double root describes a wormhole throat. With (5.27), the equation becomes
2 2
dR α−β
g rr 2
= sinh (γ/r) + coth(γ/r) = 0. (5.28)
dr 2γ
It is clear that, if roots exist, they are always double roots corresponding to wormhole throats.
They exist if (β − α)/γ > 0 and, in this case, they are given by
2γ γ
rH = = . (5.29)
β−α+2γ 2γ
ln β−α−2γ tanh−1 β−α
74
The Analogue of the Jebsen-Birkhoff Theorem in Brans-Dicke Gravity Chapter 5
If (β − α)/γ < 0, instead, there is a naked singularity. In fact, the general solution (5.18) has a
spacetime singularity at R = 0, as is deduced from the Ricci scalar
2
ωβ (α−β)/r
e sinh4 (γ/r) if γ 6= 0,
γ 4
ω
R = 2 ∇c ∇c φ = (5.30)
φ 2
ωβ e(α−β)/r
if γ = 0.
r4
If γ 6= 0 then when r → 0 we have, depending on whether γ is positive or negative
ωβ 2 (α−β±4γ)/r
R= e , (5.31)
16γ 4
respectively. Therefore, the Ricci scalar diverges as r → 0 only for β − α < 4γ or for β − α > 4γ,
respectively.
In the special case γ = 0, the FJNWBW metric reduces to the Yilmaz geometry [88] and its
Jordan frame cousin (5.18), (5.17) is the Brans Class IV solution [89]
φ = φ0 e−BC/r , (5.33)
where B = −(α + β)/2 and C = −2β/(α + β). The equation locating the apparent horizons reduces
2
to 1 − β−α
2r = 0, which has a double root rH = (β − α)/2 corresponding to a wormhole throat if
β > α and to a central naked singularity otherwise.
These results about the nature of the solutions of Brans-Dicke theory have already been worked
out in detail in [14]. Therefore, this analysis satisfactorily shows that it is possible to use the most
general solution (5.18) of Brans-Dicke theory to recover in a compact way the results already found
in Ref. [14] by going through each of the Brans classes of solutions individually. The general solution
(5.18) thus allows for a unified investigation of the physics behind the four Brans classes of solutions.
It must be noted here that, just as it was done in Ref. [14], the investigation of the nature of the
solution conducted here is based on the simple detection of possible black hole horizons or wormhole
throats. In fact, in contrast to the analysis made in Ref. [78], no additional requirements, such
as asymptotic flatness or regularity of the spacetime away from the throat when the latter exists,
are imposed before calling such a solution a wormhole. The wormhole definition that is implicitly
adopted here, and which was also adopted in Ref. [14], is that of Ref. [90] which consists of a
quasilocal definition involving only the properties of the local geometry of spacetime. Of course,
extended wormholes can also be studied, and they are related to the cold black holes of [76]. We
refer the reader to [76] for these situations.
75
Conclusions
Eppur si muove.
Galileo Galilei
In summary, Jebsen-Birkhoff theorem states that the Schwarzschild solution is the unique spheri-
cally symmetric and asymptotically flat solution of the vacuum Einstein field equations: a spherically
symmetric gravitational field in empty space outside a star must be static.
Jebsen-Birkhoff theorem can be generalized to some matter fields: any spherically symmetric and
asymptotically flat solution of the Einstein-Maxwell field equations must be static in the exterior
domain with a Reissner-Nordström metric3 as the general solution.
Generalizing the Jebsen-Birkhoff theorem to situations with matter present allows one to under-
stand the validity, or lack thereof, of this theorem in scalar-tensor gravity because the scalar-tensor
field equations can be rewritten as effective Einstein equations with the Brans-Dicke like scalar act-
ing as a form of effective matter. Using the Jordan frame description of scalar-tensor gravity, this
effective matter distribution must be static in order for the Jebsen-Birkhoff theorem to be valid. [109]
Contrary to General Relativity, even static, asymptotically flat, spherically symmetric black holes
in scalar-tensor gravity are not forced to be Schwarzschild: the Jebsen-Birkhoff theorem is peculiar
to Einstein theory and breaks down in more general contexts, even in the simplest Brans-Dicke case
(4.25). In this theory, what can be rescued is only a very weak form of the theorem: if the Brans-
Dicke scalar field is required to be time-independent in electrovacuo, then the metric is static (but
not necessarily the Schwarzschild or Reissner-Nordström).
However, it does not hold for matter such as baryons, or a perfect or an imperfect fluid: their
spherically symmetric solutions have a dynamic behavior. But systems such as the Solar System are
not dominated by any fluid or any other continuous matter: they are basically empty space with
isolated bodies (planets) embedded in the vacuum. To a large extent the same is true of galaxies,
mainly made up of isolated stars, and clusters of galaxies mainly made up of isolated galaxies.
As seen in Sec. 3.1, the KSdS solution is the unique spherically symmetric solution of the vacuum
Einstein equations with positive cosmological constant. This result is a simple generalization of the
ordinary Jebsen-Birkhoff theorem [23],[24], which makes the same assumptions except that it assumes
Λ = 0, and goes hand in hand with the perturbative analyses which established the stability of the
KSdS solution [41]-[44]. Putative alternative solutions of the Einstein equations under the same
conditions either solve different equations (for example, including an imperfect fluid with spacelike
radial flow) or are just the KSdS solution in disguise in an unusual coordinate system.
3
The Reissner-Nordström spacetime describes the geometry of a static, spherically symmetric, asymptotically flat, electrically charged black
hole which solves the Einstein-Maxwell equations with a purely electric radial field.
76
Conclusion
Since no-hair theorems reinforce the uniqueness of the Schwarzschild geometry [47] and cosmic
no-hair theorems establish that the de Sitter space is the unique late-time attractor in cosmology
(with few exceptions [65]), it is reasonable to expect that similar theorems should hold for the KSdS
spacetime, which is usually interpreted as describing a Schwarzschild black hole embedded in de-Sitter
space. Such theorems would prove the uniqueness of the KSdS solution.
We have proved a result of this kind by assuming spherical symmetry and the presence of an
imperfect fluid with constant equation of state P = wρ and a purely spatial radial energy flow, which
is a rather common ingredient in the construction of solutions of the Einstein equations representing
spherical inhomogeneous universes (see, e.g., [4, 6, 7, 10, 60]). The theorem proved in Sec. 3.3 does
not contradict the previous statement of Sec. 3.2. that the nonrotating Thakurta solution is the
late-time attractor of generalized McVittie solutions [8] because, in this case, the asymptotics are
(time-dependent) FLRW and not de-Sitter, which was one of the assumptions in our theorem.
In Chapter 5, the key to solving the Brans-Dicke equations under the assumptions (1)–(3) is
to map the problem into the Einstein frame and use a known result of Einstein-massless Klein-
Gordon theory [82]. By contrast, little progress is made when analyzing directly the Jordan frame
field equations. The previous section shows that the Schwarzschild black hole is obtained when the
scalar charge σ vanishes and that there is no other black hole solution under the assumptions made.
This result matches the no-hair theorems of [45]-[48],[12] (which are, however, more general). The
remaining solutions, corresponding to σ 6= 0, are necessarily of the Campanelli-Lousto form (5.7),
(5.8) or conformal to it. They can only describe wormhole throats or naked singularities, according
to the values of the parameters σ/m and ω (or of α and β).
The most general solution of Jordan frame Brans-Dicke theory under the assumptions (1)–(3) is
given by eqs. (5.21) and (5.17) and is conformal to the FJNWBW solution in an asymptotically flat
spacetime. Our analysis of Sec.5.1.3 established the general character of this solution and pointed to
the homogeneous wave equation that the scalar field obeys as being the ingredient that renders the
solution (5.21) really general. We also pointed out that it is only thanks to this constraint that the
conformal transformation trick remains a one-to-one mapping and allows one to extract the general
Jordan frame solution from the most general Einstein frame one. The homogeneity of the wave
equation, being guaranteed by the tracelessness of the matter energy-momentum tensor, makes the
conformal trick work both in vacuum and electrovacuum. The conformal transformation trick is
indeed what has allowed Bronnikov in Ref. [79] to extract the general solution for the electrovacuum
case as well.
We have investigated the general solution of vacuum Brans-Dicke gravity, a result that constitutes
an analogue of the Jebsen-Birkhoff theorem of general relativity in Brans-Dicke theory, which is a
rarity in alternative gravities. In principle, this result can be circumvented in the same ways already
conceived to evade the no-hair theorems for scalar-tensor black holes: by including matter, by allowing
the scalar field to depend on time while keeping the geometry static, or by letting the scalar field
diverge or vanish on the horizons [46].
As already noted above, generalizations of the results presented here have already been obtained.
The generalization of the Jebsen-Birkhoff theorem to multidimensional general relativity was given
in Ref. [74], and the general spherically symmetric solution of the Bergmann-Wagoner class of
scalar-tensor theories, of which Brans-Dicke gravity is a special case, has been found in Ref. [77].
Generalizations to situations in which a cosmological constant or a non-gravitational scalar field are
present and, more important, to axial symmetry, may be possible and will be explored elsewhere.
77
Conclusion
Future work
In general relativity, different types of potential scalar fields V (φ) can be proposed, such as
2m φ , or λφ4 , or V0 e−λφ , or even complex scalar fields like eiα φ where it would have two degrees of
1 2 2
freedom rather than just one, which would be interpreted as a particle and a distinct antiparticle.
In Brans-Dicke gravity, the scalar field potential V (φ) constitutes a natural generalization of the
cosmological constant and may reduce to a constant, or to a mass term; it is sometimes included
when Brans-Dicke theory is used in theories of the early universe or in quintessential scenarios 4 of
the present universe.
It is also possible to add the Maxwell’s electromagnetic theory to Brans-Dicke gravity, or also
called Brans-Dicke Maxwell solutions which are a class of static and nonstatic solutions of the Brans-
Dicke gravity that are obtained in presence of an electromagnetic field. [108]
The early 21st century is an exciting period for cosmology as the discovery that the present
expansion of the universe is accelerated. An exciting possibility is that the universe may not be just
accelerating, but even superaccelerating: measurements of the w-parameter of Brans-Dicke gravity
will confirm or reject this idea and could discriminate between simple models based on Einstein
gravity and less conventional models.
4
In cosmology, quintessence is scalar field "dark energy", that is a form of energy distinct from any normal matter or radiation, or even "dark
matter".
78
Appendices
79
Appendix A
In all cases, the Christoffel symbols and the Einstein equations were calculated using the computer
algebra program Sage.
Moreover
1 2 2 2Ht 2m 1 2 2 2m
h(t, r) = 1−H r e − Ht = 1−H R − = h(0, R),
2 re 2 R
q p
f (t, r) = h(t, r) + h2 (t, r) + H 2 r2 e2Ht = h(0, R) + h2 (0, R) + H 2 R2 = f (0, R),
e2Ht
ds2 = −f (t, r)dt2 + dr2 + e2Ht r2 dΩ2(2)
f (t, r)
e2Ht −2Ht 2 2 2
= −f (0, R)dt2 + e (H R dt − 2HRdtdR + dR2 ) + R2 dΩ2(2)
f (0, R)
H 2 R2 2 2HR 1
= −f (0, R)dt2 + dt − dtdR + dR2 + R2 dΩ2(2)
f (0, R) f (0, R) f (0, R)
H 2 R2
2HR 1
= −f + dt2 − dtdR + dR2 + R2 dΩ2(2) ,
f f f
where f = f (0, R)
80
Appendix A
H 2 R2 −f 2 + H 2 R2
−f + =
f f
√
−h − 2h h2 + H 2 R2 − h2 − H 2 R2 + H 2 R2
2
=
f
2
√
−2h − 2h h + H R2
2 2
=
f
√ !
h + h2 + H 2 R2
= −2h
f
= −2h.
dT = dt + βdR,
dT 2 = dt2 + 2βdtdR + β 2 dR2 ,
dt2 = dT 2 − 2βdtdR − β 2 dR2 .
2HR 1
ds2 = −2hdt2 − dtdR + dR2 + R2 dΩ2(2)
f f
2HR 1
= −2h(dT 2 − 2βdtdR − β 2 dR2 ) − dtdR + dR2 + R2 dΩ2(2)
f f
2HR 1
= −2hdT 2 + 4hβdtdR + 2hβ 2 dR2 − dtdR + dR2 + R2 dΩ2(2)
f f
2HR 1
= −2hdT 2 + 4hβ − dtdR + 2hβ 2 + dR2 + R2 dΩ2(2) .
f f
where
2HR
0 = 4hβ − ,
f
HR
β = ,
2hf
α = 2hf β 2 + 1.
81
Appendix A
0 1 0 0
1 ∂h
1 0 0 0
ΓTµν = ,
2h ∂R 0 0 0 0
0 0 0 0
2 ∂h
f ∂R 0 0 0
1 ∂α ∂f
ΓR −α
0 f 0 0
= ,
µν ∂R ∂R
2αf
−2Rf 2
0 0 0
2 2
0 0 0 −2Rf sin (θ)
0 0 0 0
0 0 R−1 0
Γθµν
=
0 R−1
,
0 0
0 0 0 − cos(θ) sin(θ)
0 0 0 0
0 0 0 R−1
Γϕ
µν =
0
.
0 0 cot(θ)
0 R−1 cot(θ) 0
∂2f ∂α
∂f
4ΛRα2 f + 2Rαf ∂R 2 − Rf ∂R − 4αf ∂R
GT T = 6= 0,
4Rα2
∂2f ∂α ∂α
∂f
4ΛRα2 + 2Rα ∂R 2 − 4f ∂R − R ∂R − 4α ∂R
GRR = − 6= 0,
4Rαf
∂α ∂f
2(ΛR2 − 1)α2 − Rf ∂R + 2Rα ∂R + 2αf
Gθθ = − 2
6= 0,
2α
sin2 (θ)
∂α ∂f 2 2
Gϕϕ = Rf − 2α R + f − 2(ΛR − 1)α 6= 0.
2α2 ∂R ∂R
However, as one can see, it does not satisfy the vacuum Einstein equations.
82
Appendix A
where
r
2m
f = f (r) = 1− . (A.1)
r
Using the transformation proposed (3.29) and replacing for (A.1), one has
HR
dT = dt + dR,
f (f 2
− H 2 R2 )
H 2 R2 2HR
dT 2 = dt2 + 2 2 2 2 2
dR2 + dtdR,
f (f − H R ) f (f − H 2 R2 )
2
H 2 R2 2HR
dt2 = dT 2 − 2 2 dR2 − dtdR.
f (f − H 2 R2 )2 f (f 2 − H 2 R2 )
1
ds2 = − f 2 − H 2 R2 dT 2 + 2 dR2 + R2 dΩ2(2) .
f − H 2 R2
83
Appendix A
1 ∂α
2 α ∂R 0 0 0
1 ∂α
Γrµν = 0 − 0 0 ,
2α ∂R
0 0 −Rα 0
0 0 0 −Rα sin2 θ
0 0 0 0
0 0 R−1 0
Γθµν
=
0 R−1
,
0 0
0 0 0 − cos θ sin θ
0 0 0 0
0 0 0 R−1
Γϕ
µν =
0
.
0 0 cot(θ)
0 R−1 cot(θ) 0
From the vacuum Einstein equations Gµν = Rµν − Λgµν = 0, one obtains
∂ α 2
∂α
2ΛRα + Rα ∂R 2 + 2α ∂R
GT T = = 0,
2R
2
∂ α ∂α
2ΛR + R ∂R 2 + 2 ∂R
GRR = − = 0,
2Rα
∂α
Gθθ = −ΛR2 − R − α + 1 = 0,
∂R
∂α
Gϕϕ = − ΛR2 + R + α − 1 sin2 (θ) = 0.
∂R
84
Appendix A
2m
where f (r) = − .
r
The non-vanished Christoffel symbols are:
∂f
0 f 0 0
∂r
∂f ∂a
f
1 2a 0 0
Γtµν = 2 ∂r ∂t ,
2f 0 ∂a
0 2r2 f 0
∂t
∂a
0 0 0 2r2 f a sin2 (θ)
∂t
f ∂f 1 ∂a
0 0
2a2 ∂r a ∂t
1 ∂a 1 ∂f
Γrµν = a ∂t
− 0 0 ,
2f ∂r
0 0 −rf 0
2
0 0 0 −rf sin (θ)
1 ∂a
0 0
a ∂t
0
0 0 r−1 0
Γθµν = 1 ∂a ,
r−1 0 0
a ∂t
0 0 0 − cos(θ) sin(θ)
1 ∂a
0 0 0
a ∂t
0 0 0 r−1
Γϕ
µν = .
0 0 0 cot(θ)
1 ∂a
r−1 cot(θ) 0
a ∂t
85
Appendix A
Again, from vacuum Einstein equation, we should obtain Gµν = Rµν − Λgµν = 0
2 2
2Λra2 f − 6ra ∂∂t2a + rf ∂∂rf2 + 2f ∂f
∂r
Gtt = 6= 0,
2ra2
2 2 2
2Λra2 f − 4r ∂a∂t − 2ra ∂∂t2a + rf ∂∂rf2 + 2f ∂f
∂r
Grr = − 6= 0,
2rf 2
2 2
Λr2 a2 f − 2r2 ∂a
∂t − r2 a ∂∂t2a + rf ∂f 2
∂r + f − f
Gθθ = − 6= 0,
f
" 2 !#
sin2 (θ) ∂a ∂ 2a ∂f
Gϕϕ =− Λr2 a2 f − 2r2 + r2 a 2 − rf − f2 + f 6= 0,
f ∂t ∂t ∂r
∂a ∂f
∂t ∂r
Gtr = Grt = 6= 0.
af
That does not satisfy the vacuum Einstein equations.
1
ds2 = − f 2 − H 2 R2 f 2α dt2 − 2HRf α−1 dtdR + 2 dR2 + R2 dΩ2(2)
f
2m
where f 2 = f 2 (R) = 1 − .
R
Using the transformation proposed (3.40), one obtains
1
dT = (dt + βdR) ,
F
1
dT 2 = 2 dt2 + β 2 dR2 + 2βdtdR ,
F
dt2 = F 2 dT 2 − β 2 dR2 − 2βdtdR.
1
ds2 = − f 2 − H 2 R2 f 2α F 2 dT 2 − β 2 dR2 − 2βdtdR − 2HRf α−1 dtdR + 2 dR2 + R2 dΩ2(2)
f
= − f 2 − H 2 R2 f 2α F 2 dT 2 + f 2 − H 2 R2 f 2α β 2 dR2 + 2 f 2 − H 2 R2 f 2α β − 2HRf α−1 dtdR
1
+ 2 dR2 + R2 dΩ2(2) .
f
86
Appendix A
2 f 2 − H 2 R2 f 2α β − 2HRf α−1 = 0,
2 f 2 − H 2 R2 f 2α β = 2HRf α−1 ,
HRf α−1
β= ,
f 2 − H 2 R2 f 2α
and,
H 2 R2 f 2α−1 1
ds2 = − f 2 − H 2 R2 f 2α F 2 dT 2 + dR2 + 2 dR2 + R2 dΩ2(2)
2 2
(f − H R f ) 2 2α f
H R f + f − H R f 2α 2
2 2 2α 2 2 2
= − f 2 − H 2 R2 f 2α F 2 dT 2 + dR + R2 dΩ2(2)
f 2 (f 2 − H 2 R2 f 2α )
1
= − f 2 − H 2 R2 f 2α F 2 dT 2 + 2 dR2 + R2 dΩ2(2) .
(f − H 2 R2 f 2α )
1
ds2 = − F 2 ξdT 2 + dR2 + R2 dΩ2(2)
ξ
This expression is looks like eq. A.2, then the Einstein equations are Gµν = Rµν − Λgµν
2
∂ ξ
2F 2 ΛRξ + F 2 Rξ ∂R 2 ∂ξ
2 + 2F ξ ∂R
GT T = , (A.3)
2R
∂2ξ ∂ξ
2ΛR + R ∂R 2 + 2 ∂R
GRR =− , (A.4)
2Rξ
∂ξ
Gθθ = −ΛR2 − R − ξ + 1, (A.5)
∂R
∂ξ
Gϕϕ 2
= − ΛR + R + ξ − 1 sin2 (θ). (A.6)
∂R
ξ = 1 − 2mR−1 − H 2 R2 (1 − 2mR−1 )α ,
∂ξ
= 2mR−2 − 2H 2 R(1 − 2mR−1 )α − α2mH 2 (1 − 2mR−1 )α−1 ,
∂R
∂2ξ
= −4mR−3 − 2H 2 (1 − 2mR−1 )α − 2H 2 Rα(1 − 2mR−1 )α−1 2mR−2
∂R2
− α(α − 1)2mH 2 (1 − 2mR−1 )α−2 2mR−2
= −4mR−3 − 2H 2 f 2α − 4H 2 mαf 2(α−1) R−1 − α(α − 1)4m2 H 2 f 2(α−2) R−2 .
87
Appendix A
4m
GRR = 6H 2 R − − 2H 2 R(1 − 2mR−1 )α − 4H 2 mα(1 − 2mR−1 )α−1
R2
4α(α − 1)H 2 m2 4m
− (1 − 2mR−1 )α−2 + 2 − 4H 2 R(1 − 2mR−1 )α − 4mαH 2 (1 − 2mR−1 )α−1
R R
4mα(α − 1)H 2 2(α−2)
= 6H 2 R − 6H 2 Rf 2α − 8H 2 mαf 2(α−1) − f
R
∂ξ
Gθθ = −ΛR2 − R −ξ+1
∂R
= −ΛR2 − 2mR−1 + 2H 2 R2 (1 − 2mR−1 )α + 2mαH 2 R(1 − 2mR−1 )α−1
− 1 + 2mR−1 + H 2 R2 (1 − 2mR−1 )α + 1
= −ΛR2 + 3H 2 R2 f 2α + 2mαH 2 Rf 2(α−1) .
(
3R2 − 3R2 f 2α − 4mαRf 2(α−1) − 2mα(α − 1)f 2(α−2) = 0,
−3R + 3Rf 2α + 2mαf 2(α−1) = 0,
(
3R2 − 3R2 f 2α − 4mαRf 2(α−1) − 2mα(α − 1)f 2(α−2) = 0,
⇒
−3R2 + 3R2 f 2α + 2mαRf 2(α−1) = 0,
(
2αRf 2(α−1) + 2mα(α − 1)f 2(α−2) = 0,
⇒
−3R2 + 3R2 f 2α + 2mαRf 2(α−1) = 0.
88
Appendix A
If α 6= 0, one has
(
Rf 2(α−1) + m(α − 1)f 2(α−2) = 0,
(A.7)
−3R + 3Rf 2α + 2mαf 2(α−1) = 0.
µ − 2 = 3µ,
−2 = 2µ,
−1 = µ.
89
Appendix B
Beginning from the action (4.52) and adding a matter term SM , the total action for f (R) gravity
takes the form
√
Z
1
Smetric = d4 x −gf (R) + SM (gab , φ), (B.1)
2κ
where φ denotes the matter fields. Variation with respect to the metric gives
1
f 0 (R)Rab − f (R)gab − [∇a ∇b − gab 2]f 0 (R) = κTab , (B.2)
2
where
−2 δSM
Tab = √ . (B.3)
−g δg ab
The trace of eq. (B.2) is
where T = g ab Tab relates R with T differentially and not algebraically as in general relativity, where
R = −κT . This is an indication that the field equations of f (R) theories will admit a larger variety
of solution than Einstein’s theory.
One begins with the metric f (R) gravity (B.1). One can introduce a new field χ and write the
dynamically equivalent action
√
Z
1
Smetric = d4 x −g[f (χ) + f 0 (χ)(R − χ)] + SM (gab , ψ). (B.5)
2κ
Variation with respect to χ leads to
f 0 (χ)(R − χ) = 0. (B.6)
90
Appendix B
Therefore, if f 0 (χ) 6= 0 then R = χ, which reproduces the action (B.1).1 Redefining the field χ by
φ = f 0 (χ) and setting
R = V 0 (φ). (B.10)
By taking the trace of eq. (B.9) in order to replace R in equation (B.10), one gets
dV
32φ + 2V (φ) − φ = κT. (B.11)
dφ
This last equation determines the dynamics of φ for given matter sources.
Finally, as usual in Brans-Dicke and scalar-tensor theories, one can perform a conformal trans-
formation and rewrite the action (B.8) in the Einstein frame (as opposed to the Jordan frame),
a scalar-tensor theory is mapped into the Einstein frame in which the "new" scalar field φ̃ couples
minimally to the Ricci curvature and has canonical kinetic energy, as described by the gravitational
action
" #
4 √
Z
R̃ 1 a
(g)
S = d x −g − ∂ φ̃∂a φ̃ − U (φ̃) . (B.14)
2κ 2
Rf 0 (R) − f (R)
U (φ̃) = , (B.16)
2κ(f 0 (R))2
1
The action is sometime called "R regular" by mathematical physicists if f 0 (R) 6= 0 [107].
91
Appendix B
It is important to stress that the actions (B.1), (B.8), and (B.17) are nothing but different
representations of the same theory.
92
Bibliography
[1] X. Gu, Y. Wang, and S.-T. Yau, Geometric compression using Riemann surface structure, Comunications in Information
and Systems, Vol. 3, Nr 3, pp. 171-182 (2004).
[2] S. M. Carroll, Spacetime and Geometry, An Introduction to General Relativity, Addison Wesley (2004).
[3] V. Faraoni, Cosmological and Black Hole Apparent Horizons Springer, New York (2015).
[5] T.P. Sotiriou and V. Faraoni, Phys. Rev. Lett. 108, 081103 (2012).
[7] C. Gao, X. Chen, V. Faraoni, and Y.-G. Shen, Phys. Rev. D 78, 024008 (2008).
[9] V. Faraoni, C. Gao, X. Chen, and Y.-G. Shen, Phys. Lett. B 671, 7 (2009).
[11] V. Faraoni, Cosmology in Scalar-Tensor Gravity, Kluwer Academic Publishers, Boston/Dordrecht/London (2004).
[13] L. Vanzo, S. Zerbini, and V. Faraoni, Phys. Rev. D 86, 084031 (2012).
[14] V. Faraoni, F. Hammad, and S. D. Belknap-Keet, Phys. Rev. D 94, 104019 (2016).
[15] https://nms.kcl.ac.uk/eugene.lim/teach/GR/GR.pdf
[16] R.R. Caldwell, M. Kamionkowski, and N.N. Weinberg, Phys. Rev. Lett., 91, 071301 (2003);
S. Nesseris and L. Perivolaropoulos, Phys. Rev. D, 70, 123529 (2004).
93
Appendix B
[17] R.M. Wald. General Relativity The University of Chicago Press, Chicago (1984).
[19] https://science.nasa.gov/astrophysics/focus-areas/what-is-dark-energy
[20] D.L.W. Vleeshouwers, Black Hole Information and Thermodynamics, Ludwig Maximilians Universität, München (2018).
arXiv.org > gr-qc > arXiv:1809.01403
[23] J.T. Jebsen, Ark. Mat. Ast. Fys. (Stockholm), 15, nr 18 (1921), reprinted in Gen. Relat. Gravit, 37, 2253 (2005).
[24] G.D. Birkhoff, Relativity and Modern Physics, Harvard University Press, Cambridge, USA (1923).
[25] https://cuentos-cuanticos.com/tag/horizonte-aparente/
[26] P.T. Chrus̀ciel, J. Lopes Costa, and M. Heusler, Living Rev. Relativity 15, 7 (2012).
[27] R. Goswami and G.F.R. Ellis, Gen. Relat. Gravit. 43, 2157 (2011); Gen. Relat. Gravit. 44 2037 (2012);
G.F.R. Ellis and R. Goswami, Gen. Relat. Gravit. 45, 2123 (2013).
[28] A.M. Nzioki, R. Goswami, and P.K.S. Dunsby, Phys. Rev. D 89, 064050 (2014).
[29] A. Krasiński, Inhomogeneous Cosmological Models, Cambridge University Press, Cambridge, UK (1997).
[30] H. Stephani, D. Kramer, M. MacCallum, C. Hoenselaers, and E. Herlt, Exact Solutions of Einstein Field Equations,
Cambridge University Press, Cambridge, UK (2003).
[32] A.D. Linde, Particle Physics and Inflationary Cosmology, Harwood, Chur, Switzerland (1990).
[33] A.R. Liddle and D.H. Lyth, Cosmological Inflation and Large Scale Structure.
Cambridge: Cambridge University Press (2000).
[34] L. Amendola and S. Tsujikawa, Dark Energy, Theory and Observations, Cambridge University Press, Cambridge, UK
(2010).
[38] A.C. Fabian and A.N. Lasenby, General Relativity: The Most Beautiful of Theories, ed by C. Rovelli, De Gruyter Studes
in Mathematical Physics vol. 28, De Gruyter, Berlin (2015).
[39] J.L. Synge, Relativity: the General Theory, North Holland, Amsterdam (1960).
[40] M. Boucher, G.W. Gibbons, and G.T. Horowitz, Phys. Rev. D, 30, 2447 (1984);
H. Kodama, J. Korean Phys. Soc. 45, S68 (2004);
P.G. LeFloch and L. Rozoy, Compt. Rendu. Acad. Sci. Paris Ser I, 348, 1129 (2010);
W. Masood-ul-Alam and Yu, Comm. Analys. Geom., 23, 377 (2015).
94
Appendix B
[44] H. Otsuki and T. Futamase, Prog. Theor. Phys 85, 721 (1991).
[46] J. D. Bekenstein, in Second International A. D. Sahkarov Conference on Physics, Moscow, 1996, edited by I. M. Dremin
and A. M. Semikhatov, World Scientific, Singapore (1997);
T. P. Sotiriou and S.-Y. Zhou, Phys. Rev. Lett. 112, 251102 (2014);
C. A. R. Herdeiro and E. Radu, Int. J. Mod. Phys. D 24, 1542014 (2015);
T. P. Sotiriou, Classical Quantum Gravity 32, 214002 (2015);
E. Babichev and C. Charmousis, J. High Energy Phys. 08, 106 (2014).
[47] R. Ruffini and J.A. Wheeler, Phys. Today 24 (1), 30, (1971);
J. Chase, Comm. Math. Phys. 19, 276 (1970);
J.D. Bekenstein, Phys. Rev. D 5, 1239 (1972); Phys. Rev. D 5, 2403 (1972); Phys. Rev. Lett. 28, 452 (1972);
arXiv:gr-qc/9605059;
C. Teitelboim, Lett. Nuovo Cimento 3S2, 326 (1972);
T. Zannias, J. Math. Phys. 36, 6970 (1995);
J.D. Bekenstein, Phys. Rev. D 51, 6608 (1995);
A. Saa, J. Math. Phys. 37, 2346 (1996);
K.A. Bronnikov, Phys. Rev. D 64, 064013 (2001);
C.A.R. Herdeiro and E. Radu, Int. J. Mod. Phys. D 24, 1542014 (2015);
T.P. Sotiriou, Class. Quantum Grav., 32, 214002 (2015).
[48] S. Bhattacharya, K.F. Dialektopoulos, A.E. Romano, and T.N. Tomaras, Phys. Rev. Lett. 115, 181104 (2015).
[49] K. Henttunen, T. Multamäki, and I. Viljia, Phys. Rev. D 77, 024040 (2008);
K. Henttunen and I. Viljia, Phys. Lett. B 731, 110 (2014);
K. Henttunen and I. Viljia, J. Cosmol. Astropart. Phys. 1505, 001 (2015).
[54] G.C. McVittie, Mon. Not. Roy. Astron. Soc. 93, 325 (1933).
[55] R. Nandra, A.N. Lasenby, and M.P. Hobson, Mon. Not. Roy. Astron. Soc. 422, 2931 (2012); Mon. Not. Roy. Astron.
Soc. 422, 2945 (2012).
[57] N. Afshordi, D.J.H. Chung, M. Doran, G. Geshnizjani, Phys. Rev. D 75, 123509 (2007);
N. Afshordi, Phys. Rev. D 80, 081502 (2009).
[58] N. Afshordi, M. Fontanini, and D.C. Guariento, Phys.Rev. D 90, 084012 (2014);
E. Abdalla, N. Afshordi, M. Fontanini, D.C. Guariento, and E. Papantonopoulos, Phys. Rev. D 89, 104018 (2014).
[60] M.M.C. Mello, A. Maciel, and V. Zanchin, Phys. Rev. D 95, 084031 (2017).
[62] D.D. McNutt and D.N. Page, Phys. Rev. D, 95, 084044 (2017).
[63] T. Clifton, D.F. Mota, and J.D. Barrow, Mon. Not. Roy. Astron. Soc. 358, 601 (2005).
[64] P. Castelo Ferreira, Phys. Lett. B 684, 73 (2010); Adv. Sp. Res. 51, 1266 (2013); arXiv:0907.0847; arXiv:1203.1844.
95
Appendix B
[66] J.L. Synge, Relativity: The General Theory, North Holland, Amsterdam (1955).
[67] M.A. Lorentz, Collected Papers, vol. 5, p. 363, The Hague, Njhoff (1937).
[68] E.W. Kolb and M.S. Turner, The Early Universe, Reading, Massachusetts: Addison-Wesley (1994).
[70] C.H. Brans and R.H. Dicke, Phys. Rev 124, 925 (1961).
[71] P. G. Bergmann, Int. J. Theor. Phys. 1, 25 (1968); R. V.Wagoner, Phys. Rev. D 1, 3209 (1970); K. Nordvedt, Astrophys.
J. 161, 1059 (1970); K. Nordtvedt, Phys. Rev. 169, 1017 (1968).
[73] N. Bocharova, K. Bronnikov, and V. Melnikov, Vestn. Mosk. Univ., Ser. 3: Fiz., Astron. 6, 706 (1970).
[74] K.A. Bronnikov and V.N. Melnikov, Gen. Relativ. Gravit 27, 465 (1995).
[77] K. A. Bronnikov, C. P. Constantinidis, R. L. Evangelista, and J. C. Fabris, Int. J. Mod. Phys. D 08, 481 (1999).
[78] K. A. Bronnikov, M. V. Skvortsova, and A. A. Starobinsky, Gravitation Cosmol. 16, 216 (2010).
[81] M. Campanelli and C. Lousto, Int. J. Mod. Phys. D 02, 451 (1993).
[84] A. Bhadra and K. Sarkar, Gen. Relativ. Gravit. 37, 2189 (2005).
[85] N. Van Den Bergh, Gen. Relativ. Gravit. 12, 863 (1980).
[87] A. B. Nielsen and M. Visser, Class. Quantum Grav. 23, 4637 (2006).
[93] P. Jordan, Schwerkraff und Weltfall, Grundlagen der Theorestische Kosmologie, Braunschweig: Vieweg und Sohns (1922).
96
Appendix B
[94] L.D. Landau and E.M. Lifschitz, The Classical Theory of Fields. Oxford: Pergamon, Press (1989).
[99] A.V. Filippenko, and A.G. Riess, Phys Rep. Lett 79, 4740 (1998).
[100] G. Efstathiou, S.L. Bride, A.N. Lasenby, M.P. Hobson, and E.S. Ellis, astro-ph/9812226 (1998).
[107] G. Magnano and L.M. Sokolowski, Phys. Rev. D 50, 5039 (1994).
[108] V.B.Johri, G.K. Goswami, Electromagnetic solutions of Brans-Dicke theory of gravitation from Einstein theory, Univer-
sity of Gorakhpur, India, 273001, (1977).
[109] V. Faraoni, The Jebsen-Birkhoff theorem in alternative gravity, Bishop’s University, Canada (2010).
97