Analysis of Some Classical and Quantum Aspects of Black Hole Physics

arXiv:2004.
07104v1 [gr-qc] 15 Apr 2020

Analysis of some classical and quantum
aspects of black hole physics
L UCIANO G ABBANELLI

c 2019
All the content of this work is licensed under the Creative Commons
iii
Analysis of some classical and quantum aspects

of black holes
by
Luciano Gabbanelli
Thesis submitted in fulfillment of the requirements for the degree of Philosophiæ

Doctor in the Faculty of Physics of the University of Barcelona.
PhD programme in Physics. Line of research: Particle Physics and Gravitation.
This thesis was supervised and tutorized by
Domènec Espriu Climent
Department of Quantum Physics and Astrophysics & Institute of Cosmos Sciences

of the University of Barcelona.
Barcelona, September 2019

To my biological family, for being who they are.
To my chosen family, for being here.

vii
Abstract
For ninety years we have known that our universe is in expansion. Cosmological
data favor an unknown form of intrinsic and fundamental uniform energy contribut-
ing approximately 68% of the total energy budget in the current epoch. The simplest
proposal in accordance with observations is the standard cosmological model con-
sisting of a small but positive cosmological constant producing a gravitational re-
pulsive effect driving the accelerated expansion. In standard cosmology general rel-
ativity is assumed as the theory for gravity, which in turn predicts that a sufficiently
compact mass can deform spacetime and form a black hole. At a mathematical level,
these objects are considered vacuum solutions described by very few parameters.
For instance, a stationary black hole solution is completely described by its mass,
angular momentum, and electric charge; and two black holes that share the same
values for these parameters, are indistinguishable from one another.
On the basis of the usual metrics describing black holes, it is generally believed
that all contained matter is localized in the center or, if rotating, on an infinitely thin
ring. Recent approaches challenge this unintuitive assumption and consider matter
just spread throughout the interior. Clearly, this begs for a quantum description in
curved space. In past years, a novel approach established a new bridge between
quantum information and the physics of black holes when an intriguing proposal
was made: black holes could possibly be understood as Bose–Einstein condensates
of soft interacting but densely packed gravitons. The aim of this thesis is to discuss
how to construct a graviton condensate structure on top of a classical gravitational
field describing black holes. A necessary parameter to be introduced for this anal-
ogy is a chemical potential which we discuss how to incorporate within general rel-
ativity. Next we search for solutions and, employing some very plausible assump-
tions, we find out that the condensate vanishes outside the horizon but is non-zero
in its interior. These results can be extended easily to a Reissner–Nordström black
hole. In fact, we find that the phenomenon seems to be rather generic and is associ-
ated with the presence of a horizon, acting as a confining potential. In order to see
whether a Bose–Einstein condensate is preferred, we use the Brown–York quasilocal
energy, finding that a condensate is energetically favourable in all cases in the classi-
cally forbidden region. The Brown–York quasilocal energy also allows us to derive a
quasilocal potential, whose consequences allow us to suggest a possible mechanism
viii
to generate a graviton condensate in black holes. On the contrary, this is not the case
for any kind of horizons; for instance, this mechanism appears not to be feasible in
order to generate a quantum condensate behind the cosmological de Sitter horizon.
Furthermore, when a pair of black holes merge, an immense amount of energy
should be given off as gravitational waves. Their wave forms have been recently
confirmed to be perfectly described by general relativity. We discuss why for low
frequency gravitational waves aimed to be detected by astrophysical PTA observa-
tions the fact that propagation should take place over an expanding (approximately
globally de Sitter) spacetime should be taken into account. In this manner, harmonic
waves produced in such mergers would become anharmonic when measured by
cosmological observers. This effect is tiny but appears to be observable for gravita-
tional waves to which PTA are sensitive. Therefore we have characterized modifica-
tions to the expected signal, and how it is related to the source and pulsar character-
istics that are employed by the IPTA collaboration. If the cosmological constant were
an intrinsic property, this experiment would be capable of confirming the relevance
of Λ at redshift z < 1.
ix
Contents
Abstract vii
1 Introduction 1
1.1 From general relativity to black holes . . . . . . . . . . . . . . . . . . . . 1
1.2 Self-completion via classicalization . . . . . . . . . . . . . . . . . . . . . 4
1.3 Black holes as self sustained quantum systems . . . . . . . . . . . . . . 7
1.4 Back to the geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5 The accelerating universe . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6 Gravitational waves propagating through an expanding universe . . . 14
1.7 Gravitational wave detection . . . . . . . . . . . . . . . . . . . . . . . . 15
2 Black holes as Bose–Einstein Condensates 19

2.1 Summary of our current understanding . . . . . . . . . . . . . . . . . . 19
2.2 Building up a condensate over a Schwarzschild black hole background 21
2.2.1 Einstein’s equations as Gross–Pitaevskii equations . . . . . . . . 22
2.3 Outside the horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.1 Asymptotic analysis . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.2 Numerical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.4 Inside the horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.1 Linearization near the event horizon . . . . . . . . . . . . . . . . 31
2.4.2 Numerical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.4.3 Exact solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.5 Consequences of the results and connection with previous proposals. . 38
3 Quasilocal energy for condensate structures 43

3.1 Quasilocal energy definitions in general relativity . . . . . . . . . . . . 43
3.2 Hamilton–Jacobi Formalism . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.2.1 Nonrelativistic mechanics review . . . . . . . . . . . . . . . . . . 47
3.2.2 Covariant Hamiltonian formalism . . . . . . . . . . . . . . . . . 48
3.3 Brown–York quasilocal energy . . . . . . . . . . . . . . . . . . . . . . . 50
3.4 Quasilocal energy for a graviton condensate black hole . . . . . . . . . 52
3.5 Gravitational binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4 Condensates beyond horizons 59

4.1 Bose–Einstein condensate extensions . . . . . . . . . . . . . . . . . . . . 59
4.2 Bose–Einstein condensate in Reissner–Nordström . . . . . . . . . . . . 59
x
4.2.1 Quasilocal energy of a Reissner–Nordström black hole . . . . . 62

4.3 Bose–Einstein condensate in de Sitter . . . . . . . . . . . . . . . . . . . . 63
4.3.1 Quasilocal energy analysis over a de Sitter horizon . . . . . . . 64
4.3.2 The case of the de Sitter horizon . . . . . . . . . . . . . . . . . . 66
5 Gravitational waves propagating over nonempty backgrounds 69

5.1 The standard cosmological Model . . . . . . . . . . . . . . . . . . . . . 69
5.2 The importance on the choice of coordinate systems . . . . . . . . . . . 74
5.3 Linearized gravity framework . . . . . . . . . . . . . . . . . . . . . . . . 76
5.4 Gravitational wave propagation in de Sitter . . . . . . . . . . . . . . . . 78
5.5 Non-relativistic matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.6 Gravitational wave propagation in an analytic ΛCDM- background . . 84
5.6.1 Non-interacting background components . . . . . . . . . . . . . 85
5.6.2 Non-relativistic matter over an expanding background . . . . . 86
5.7 Including relativistic matter . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.8 Gravitational wave propagation in a general background . . . . . . . . 90
6 Local measurements of Λ with pulsar timing array 93

6.1 Pulsars as gravitational wave detectors . . . . . . . . . . . . . . . . . . 93
6.2 Linearized perturbations in FLRW . . . . . . . . . . . . . . . . . . . . . 94
6.3 Relevance for Pulsar Timing Arrays . . . . . . . . . . . . . . . . . . . . 97
6.4 Characterization of the signal . . . . . . . . . . . . . . . . . . . . . . . . 103
6.4.1 Dependence with the source parameters . . . . . . . . . . . . . 103
6.4.2 Dependence with the position of the pulsars . . . . . . . . . . . 108
6.5 The International Pulsar Timing Array project . . . . . . . . . . . . . . 112
7 Conclusions 115
List of publications 119
Abstract in Spanish 121
Resume in Spanish 123
Acknowledgements 127
Appendices 129
A Perturbative aspects of the classical theory 131
B Action variation 135
C Extrinsic curvature of 3 B 137
D Cosmological horizon for a dust universe 139

xi
E Energy momentum tensor for dust in static coordinates 141
References 143
1
Chapter 1
Introduction
1.1 From general relativity to black holes

Even though gravity is ubiquitous, it is literally forty orders of magnitude weaker
than all the other known forces. Gravity is the dominant force on the largest scales
and the force we are most aware of in daily life and for longest; it literally grounds
us. Macroscopically, matter is neutral with respect to the other three forces, whereas
everything that has mass or energy experiences or generates gravity.
In 1915 Albert Einstein completed the general theory of relativity that changed
the way of understanding the gravitational interaction [1]. From this time on, space-
time would never be again a static place where events happen, but a dynamical struc-
ture in deep connection with those events. For instance, the local energy and momen-
tum within the spacetime curve the very spacetime. Paraphrasing John Wheeler in
[2]: "spacetime tells matter how to move; matter tells spacetime how to curve." This
interplay is seen at the level of the field equations where the local spacetime curva-
ture, expressed by the Einstein tensor Gµν on the left hand side, is altered because
of the local energy and momentum within that spacetime given by the stress-energy
tensor Tµν ; this is 1
Rµν − 12 R gµν = κ Tµν . (1.1)
A crucial guiding principle leading Einstein along the formulation of his theory was
the equivalence between the gravitational and the inertial masses; to the point that
he suggested that it should be ascended to the status of general principle. The equiv-
alence principle states that "the movement of a body subjected to a gravitational
field is physically indistinguishable to that of an accelerated body in flat spacetime".
This means that locally, the spacetime is Minkowskian and the laws of physics ex-
hibit local Lorentz invariance. This universal characteristic is what urged Einstein
to transfer the properties of the gravitational field to the structure of spacetime. The
spacetime curvature and its fluctuations are a measure of the gravitational field. As
an immediate consequence, gravity could not longer be viewed as an instantaneous
1 Theconvention used for obtaining the Ricci tensor is by contracting the first and third indices of
the Riemann tensor with the inverse metric as Rµν = gδλ Rδµλν . The scalar curvature is the trace of the
Ricci tensor R = gµν Rµν and κ = 8πGN /c4 is the Einstein constant. GN is the Newtonian constant of
gravitation.
2 Chapter 1. Introduction
force. In the absence of external forces, any body immersed in a given spacetime
must follow geodesics in accordance to the curvature of the dynamical spacetime it
lives in (and it curves as well).
Since its formulation, general relativity has not only been exhaustively tested
but also has predicted inconceivable phenomena. Having been born from a small
contribution to the Mercury’s perihelion, an explanation of a small precession with-
out the use of any arbitrary parameter, quickly changed the paradigm of how we
understand the oldest known natural force. Einstein jump to the fame soon after,
during the total Sun eclipse of 1919, when Eddington confirmed the prediction for
the deflection of starlight passing near the surface of the Sun. However it was not
until 1959 that the first experiment (not observation) was performed to test this the-
ory. Pound and Rebka measured the redshift of light when moving in the Earth’s
gravitational field. This research ushered in an era of precision tests of general rel-
ativity where more and more accurate tests began to be made. Some important ex-
amples are the Shapiro time delay of a signal when passing through a gravitational
field (1966); or in a stronger-field regime the orbital decay due to gravitational wave
emission by a Hulse–Taylor binary (1974), among many others. Nowadays, the first
direct measurement of gravitational waves by the LIGO (2015), tests the theory in
the very strong gravitational field limit, observing to date no deviations from the
theory. It is a common agreement in the relativistic community that we stand on the
threshold of the gravitational-wave astronomy era.
Although Einstein’s equations look elegantly simple, they are quite difficult to
solve. As the curvature has an associated energy and the energy curves the space-
time, the equations have a strongly non-lineal character. In fact, Einstein itself used
approximation methods in working out his initial predictions. In was in 1916 when
Schwarzschild found the first exact vacuum solution, other than the trivial flat solu-
tion [3]. It is a useful approximation describing slowly moving astronomical objects
such as the Earth or the Sun. Known as the Schwarzschild metric2 ,
−1
2 rS 2 2 rS
ds = gµν dx dx = − 1 −
µ ν
c dt + 1 − dr2 + r2 dΩ2 , (1.2)
r r
it is understood as the limit case of a static and spherically symmetric solution of

Einstein’s field equations. The variables range between 0 ≤ θ ≤ π, 0 ≤ φ < 2π and
the metric of the two-sphere is given by dΩ2 = dθ 2 + sin2 θ dϕ2 . The Schwarzschild
radius depends only on the object mass, rS = 2GN M/c2 .
The consequences of this solution were quite disconcerting and until today it is
said that they hid in plain sight the breakdown of the classical description and the
need of new physics to be understood. In the following year, Hilbert presented an
extensive study of the singularity structure and showed that the metric was non-
regular both at r = 0 and at r = rS . For the former, there was general consensus
2 The (− + ++) signature has been used in chapters 2, 3 and 4. We have changed to the (+ − −−)
signature in chapters 5 and 6.

1.1. From general relativity to black holes 3
about the genuine physical singularity behaviour of the curvature, which implies
that some invariant (scalar) built upon the Riemann’s tensor, such as curvature scalar
or the Kretschmann invariant (the sum of Riemann’s tensor components squared),
diverges3 . This means that the gravitational field becomes infinitely large at a grav-
itational singularity. On the contrary, the nature for the latter remained unclear for
many years.
It was Lemaître who showed in 1932 that this singularity was not physical (he
used Lemaître coordinates and stated that it was an artifice introduced because of
the coordinates that Schwarzschild had used); an illusion. Besides, seven years later,
Robertson showed that a free falling observer descending in the Schwarzschild metric
would cross the r = rS singularity in a finite amount of proper time ∆τ even though
this would take an infinite amount of time in terms of coordinate time t employed
in (1.2). In any case, it was not until 1960 when new tools of differential geometry
provided more precise definitions of what means for a Lorentzian manifold to be
singular. This led to the definitive identification of the r = rS coordinate singularity
as an event horizon: "a perfect unidirectional membrane: causal influences can cross
it in only one direction" [4]. Stated in other words, a null hypersurface from which
points in the spacetime are causally disconnected from outside. When we choose a
coordinate system where the metric does not diverge at the Schwarzschild radius,
the following fact is observed [5]: in the inner region, r < rS , everything, even pho-
tons directing radially outward, actually end up moving to decreasing values of r
and falling into the singularity. This indicates that even light remains trapped inside
the surface given by r = rS and there is no longer a causal relationship with the
exterior. This is what allows us to interpret Schwarzschild’s solution as a black hole.
The early 1970s brought a blossom in black hole physics when a close resem-
blance between the laws of black hole mechanics and the laws of thermodynamics
were discovered. The second law of thermodynamics requires an entropy for black
holes. In [6] Bekenstein conjectured that an entropy could be assigned to black holes
and that it should be proportional to the area of its event horizon, A H = 4πrS 2 ,
divided by the Planck length, ` P 2 = h̄GN /c3 ; this is
kB AH
S= . (1.3)
4 `P 2
k B is Boltzmann’s constant. Bekenstein also noticed that there was a positive quan-
tity, called the surface gravity κ H , which should be an analogue of the temperature
when comparing the laws of thermodynamics with that of black holes
h̄ c κ H
T= . (1.4)
2 π kB
In fact, it can be shown that having relations between the mass and the energy, the
3 In the original paper, the origin of coordinates has been placed over the Schwarzschild radius
(where nowadays the event horizon of a black hole is situated).

area and the entropy, the surface gravity and the temperature, black holes are con-
sistent with the three laws of thermodynamics (see [7] for a nice review).
It is reasonable to think that the emergence of the singularity structure of black
holes might be connected with the ignorance of the quantum mechanical effects. It
has been known for a long time that a quantum description of gravity is a prob-
lem and quantization of Einstein’s equations gives rise to a non-renormalizable the-
ory, uncontrollable at high energies. It was Hawking who first attempted to put
together quantum mechanics and black hole physics by semiclassical techniques
without tackling the problem of curvature singularities. In certain regimes, quan-
tum field theory (QFT) calculations can be performed in curved spacetime show-
ing that black holes are indeed thermodynamic bodies with a temperature inversely
proportional to its mass. Both their temperature and the entropy depend only on
asymptotic charges [8]. The main prediction of his calculations points towards a
slow evaporation of these objects due to the emission of thermal radiation with the
same spectrum of a black body. This led him to speculate that quantum information
may be destroyed, and that it is not possible to trace back from the emitted Hawking
quanta, the initial state that first formed a black hole [9].
This fundamental paradox is based on the fact that general relativity predicts
the existence of surfaces at which time, as measured as a clock at infinity, stands
still (event horizons). Gravity is well behaved at those surfaces (as we have already
discussed free falling observers would pass through in a finite proper time), but this
causes paradoxes in quantum mechanics. Other proposals apart from the Hawking
breakdown of predictability are in favour of revising the laws of quantum mechanics
[10]. This has turned into a fundamental problem in theoretical physics for more
than 40 years. If the breakdown of general relativity is encoded in the singularity
structure, the information paradox explicitly expresses an existing tension between
general relativity and QFT. This evidences the need for a theory of quantum gravity.
1.2 Self-completion via classicalization

Alternative ideas have been proposed to provide a bottom-up picture reproducing
the experimentally observed gravitational effects, but from a quantum field theoretic
approach and without referring to general relativity at all. The Newtonian potential
models quite comprehensively most of the features of the solar system. Hence, given
two masses, it is mandatory for the new theory to recover
m
VN (r ) = − G (1.5)
r
in the non-relativistic limit. The similarities with the Coulomb potential are obvi-
ous, VC = Q/4πr, and the difference as well, the gravitational potential is always
attractive, whereas the photon also permits a repulsive behaviour. This ensures that
a spin 1 field would not be suitable for describing the gravitational interaction; nor
1.2. Self-completion via classicalization 5
would a spin 0. In a scalar field theory, the only allowed coupling between the hy-
pothetical spin 0 gravitational mediator and the electromagnetic field is given by the
µ
Lagrangian density Lint ∼ ψTµ . If we recall the electromagnetic stress-tensor (4.3),
µ
one immediately sees that Tµ = 0; and how it was stated by Eddington during the
solar eclipse of 1919: "the light is drawn towards the sun".
Since fields higher than spin 2 are quite difficult to include consistently in a QFT,
efforts were directed to the quantization of hµν emerging from the weak-field limit
of the Einstein–Hilbert action
1
Z
d4 x
p
S HE = −g R . (1.6)
2κ
From a Wilsonian perspective, low energy effective field theories (EFT) are perfectly
well defined if we keep away from its cut-off scale. Therefore, in their domain of ap-
plicability an EFT singles out the suitable (quantum) degrees of freedom that would
serve as an adequate reference to describe a system at a given energy scale. The con-
sideration of "suitable" refers to whether the system can be consistently described
by means of a perturbative picture. In the case of gravity, the cut-off scale is given
by the Planck mass MP , and in fact, it has been possible to compute explicit quan-
tum corrections to classical spacetimes in the low-energy limit, E MP [11]. The
non-renormalizability of gravity does not represent a big deal for the theory per se at
their applicability domain; however at energies beyond its domain the theory breaks
down and gives rise to a non-renormalizable theory uncontrollable at high energies.
Let us now look at some orders of magnitude: neutron stars have masses lower than
2.16 M inside a radius of the order of 10 km; hence an average density about the
nuclear density. Some black holes have about the same mass, but they are about 10
times smaller and thus 1000 times more compact. Clearly, this begs for a quantum
description in curved space.
Both approaches, a QFT description in terms of a massless spin 2 tensor field liv-
ing in a Minkowskian space and Hawking’s theory of quantizing fields on curved
classical spacetimes, provides very interesting quantum mechanical effects. How-
ever, they also predict serious conceptual issues. As discussed above, the non-
renormalizability and the information loss paradox point towards an unavoidable
need of a more general theory. In what follows we will discuss an alternative per-
spective called self-completion via classicalization. In order to motivate the discussion,
let us briefly resume how the UV-completion of the Fermi’s interaction is performed
within the standard Wilsonian approach [12]. The UV-completion of the theory is
achieved by integrating new degrees of freedom that allow the recovery of the weak
coupling regime. This led to the replacement of the Fermi’s contact interaction by a
more complete theory, the electroweak theory. On the contrary, classicalization pro-
posed that gravity would require a non-Wilsonian treatment.
The Fermi theory was the precursor of the weak interaction. It consists of a Feyn-
man diagram of four fermions directly interacting with one another at one vertex
with the purpose of explaining the β−decay [13]. Roughly speaking, this theory is
given by a Dirac Lagrangian together with a four-fermion interaction term
Lint. ∼ GF (ΨΨ)2 , (1.7)
where GF is the Fermi coupling constant, which denotes the strength of the interac-
tion. From dimensional analysis one obtains the following: the field’s kinetic term
implies [Ψ] =mass1/2 · length−1 , then we have [ GF ] =length · mass−1 for the cou-
pling. In Planck units4 the Fermi coupling constant dimensions are [ GF ] =mass−2 .
It is common knowledge in QFT that interactions with coupling constants scaling as
the inverse square of the mass are non-renormalizable and this is what happens in
Fermi’s four-fermion interaction. While this theory describes the weak interaction
remarkably well, this happens only in its domain of applicability. The cross-section
grows as the square of the energy ∼ GF 2 E2 , indeed once the center of mass energy
of a certain process approaches the cut-off, all orders in the perturbative expansion
become relevant, the cross-section grows without bound and unitary is violated. As
a matter of fact, the theory is not valid at energies much higher than about 100 GeV.
The fact that all terms become relevant is a consequence of the non-weak interac-
tion between the chosen degrees of freedom; in fact they have rather entered in the
strong coupling regime. This sick behaviour of Fermi’s theory claims the need of a
UV–completion of the theory.
The electroweak theory represents the UV–completion of the Fermi weak inter-
actions. The cure for the later is managed by introducing three mediators for the
weak interaction; the vector bosons W ± and Z0 . Consequently, the Fermi coupling
constant [14] is usually expressed as
√ 2
2 gW
GF = (h̄c)3 . (1.8)
8 MW c2
Notice that the coupling constant of the weak interaction gW and the mass of the W
boson which mediates the decay in question, MW , do not appear separately; only
their ratio occurs. At high energies it is better to describe the β−decay as a non-
contact force field having a finite range, albeit very short. Now it is clearly seen
that one can understand Fermi’s theory as the low energy limit of the more general
electroweak theory.
The findings in [15, 16] lead us to the conclusion that Einsteinan gravity is self-
complete in a sense that is very different from the standard notion of the Wilsonian
completeness. The concepts of ` P being a minimal length and the absence of trans-
Planckian propagating quantum degrees of freedom [17], as well as UV–IR connec-
tion through the black holes [18] have been around for some time. The authors argue
that when the relevant degrees of freedom of our effective theory become strongly
coupled, any attempt of probing the physics beyond, automatically bounces us back
4 Planck units are given by c = GN = k B = h̄ = 1.
1.3. Black holes as self sustained quantum systems 7
to the macroscopic distances due to black hole formation. The theory completes itself
by producing some high-multiplicity states of the same particles that were already
in the theory. Each of the constituents would be in an extremely soft and weakly
interacting regime with one another. The occupation number of the newly produced
soft quanta is extremely high. The resulting state would approximately behave clas-
sically. Hence, they postulate black holes as the real UV states of the theory. As a
natural consequence of this barrier, high energy scattering amplitudes are unitarized
using black hole production.
1.3 Black holes as self sustained quantum systems

According to the usual metrics describing black holes, it is generally believed that
all contained matter is localized in the center, or if rotating, on an infinitely thin ring.
Radically different approaches challenge this unintuitive assumption of localized
matter and consider it spread throughout the interior [19]. The key point addressed
is how matter can withstand the enormous pressures associated with such high den-
sities. In [20] for the very first time it is assumed that matter near the horizon could
be in a Bose–Einstein condensate phase. Soon after, the Bose–Einstein condensate
(BEC) idea is extended to the whole interior and the "gravo-star" solution is dis-
cussed [21].
The idea that many-particle states might represent the right path toward a quan-
tum mechanical description of self-gravitating systems finds a natural explanation
from the classicalization approach, encoded in the 2 → N scattering process [22]. Ac-
cording to the Thorne’s Hoop conjecture (see e.g. [23] for the classical formulation
and [24] for its quantum mechanical extension) black holes can form in a collision if
the two colliding objects fall within their black disk. This acceptance is based on the
√
fact that if a source with center of mass energy s is confined within its gravitational
√
(Schwarzschild) radius R g (s) = s ` P 2 , must form a black hole. Then it follows that
√
when two high energy quanta are scattered with a center of mass energy s MP ,
the theory self-completes by producing collective states composed of a large number
s
N' (1.9)
MP 2
of soft virtual gravitons. Each of them is characterized by a typical quantum size

√
λ' N `P (1.10)
and in the mean-field approximation (N 1), the semi-classical behavior of macro-

scopic black holes is recovered 5 . Stated in other words, in the classicalization pic-
ture the role of the hard quanta is taken up by the collective effects of a self-bound
5 Note that the weakly interacting gravitons are characterized by an energy
h̄ h̄
ε= = . (1.11)
λ rS
state, composed of a large multiplicity of soft quanta produced in the process. The
weakly-interacting constituents result in a quasi-classical final state that in a mean-
field approximation (N 1) acquires some properties of classical black holes. This
is the reason why it is said that such a theory self-completes in the UV by classical-
ization.
Starting from the previous discussion, it is possible to draw an interesting quan-
tum picture for black holes without referring to any geometric arguments at all
[25, 26]. The microscopic corpuscular picture of black holes as N −graviton self-
bound states at a quantum critical point [27] enables us to translate the N −graviton
production processes into the black hole formation, both in field and string theory
scatterings.
Let us assume a 2 → N scattering process with a momentum exchange p. Then
the final state of N gravitons each of them with a dimensionless effective coupling α g .
According to the de Broglie relation λ = h̄/|p|, using (1.9) or (1.10), one can alterna-
tively write
G N | p |2 ` 2 M 2 1
αg = = P2 ' P ' . (1.14)
h̄ λ s N
It is explicitly seen that although each of the constituents interact very weakly among
the other individual corpuscles (α g 1), the system is globally in a strong coupling
regime with a collective dimensionless coupling g = α g N ∼ 1. In a Bose–Einstein con-
densate language, non-perturbative effects due to the collective interaction among
the N gravitons resulting from the scattering process become extremely relevant and
lead to the global (approximately) classical behaviour of the system.
Let us face this picture from another perspective and try to form a Bose–Einstein
condensate of N weakly interacting gravitons. Let us assume that each of them is
characterized by a typical wavelength λ ` P . It is obvious that each graviton would
feel the collective binding potential by the other N − 1 constituents. At first order of
approximation, it can be estimated to be U ' −h̄ α g N/λ. Furthermore, the kinetic
energy is K ' |p| ' h̄/λ. The resulting self-sustained state is expected to be in a
marginally bound condition; then the energy balance yields K + U ' 0. The direct
implication is the quantum criticality condition
αg N ' 1 (1.15)
that coincides with the scaling found in the classicalization approach. This relation
allows us to identify a black hole of mass M with a graviton Bose–Einstein conden-
sate on the verge of a quantum phase transition. The critical point separates two
Energy conservation arguments implies that
√
s=Nε, (1.12)
from where the expression (1.9) it is derived. This clearly implies that
N 1. (1.13)
1.3. Black holes as self sustained quantum systems 9
phases: for αN 1, the system is in the phase in which collective graviton-graviton

attraction is not enough to form a self-bound state and the graviton Bose-gas is es-
sentially free. For αN = 1 the bound-state is formed, then corresponds to the point
of black hole formation. The typical wavelength of the gravitons constituting the
condensate coincides with the Schwarzschild radius λ ' rS . Since (1.14) holds, the
scaling laws of corpuscular black holes are given by the equations (1.10), (1.14) together
with
√
M= N MP . (1.16)
Furthermore, at the critical point the system is characterized by collective Bo-

goliubov modes that become nearly gapless (up to 1/N resolution), and there is a
maximal degeneracy of states. These degeneracy of states, at the same time rep-
resents the black hole entropy factor. As one would expect for an interacting ho-
mogeneous Bose–Einstein condensate [28], the graviton wave-packet is "stuck" at
the critical point. It cannot really hold on all its constituents, then it is slowly loos-
ing gravitons through a quantum depletion process, reducing N, but maintaining the
quantum criticality condition for each N. This leakage is a fully-quantum progen-
itor of what in a semi-classical treatment is called Hawking’s radiation. Indeed, in
[25, 26] the depletion law

dN 1 1
' −√ +O (1.17)
dt N `P ` P N 3/2
is obtained. Figure 1.1 shows how depletion works, in the 2 → 2 scattering of two
constituent gravitons one gains enough energy to leave the ground state and hence
leaves the condensate. Considering the "scaling laws" previously discussed, the lat-
ter relation can be written in terms of the Arnowitt–Deser–Misner (ADM) mass M
as ! !
dM M 3 MP 5 T2
T4
' − P 2 +O ' − +O , (1.18)
dt `P M ` P M3 h̄ MP 2
h̄
√
where the temperature is given by T = h̄/GN M = h̄/ N ` P . The result (1.18)
coincides with Hawking’s arguments in the semiclassical limit given by N → ∞,
√
` P → 0, but keeping N ` P and h̄ fixed.
F IGURE 1.1: Leading order process responsible for quantum depletion of

graviton condensate. Figure taken from [26].
1.4 Back to the geometry

The corpuscular theory of gravity offers a way to unify most of the experimental
observations in a single framework (from astrophysical to galactic and cosmological
scales), solely based on gravity and matter. This enables us to argue for the first
time that the collective effects would be the correct explanation for the puzzling
behavior of black holes; and by direct implication the apparent incompatibility of
quantum mechanics and general relativity. Unlike black holes, the new solution is
thermodynamically stable and apparently bypasses the information loss paradox.
As an attempt to put together Dvali’s proposal with general relativity solutions, the
idea of black holes as graviton condensates is recovered [29].
The possibility of defining a quasilocal energy for different type of classical space-
times presenting horizons is well established; therefore certain volume containing
gravitons, each of them with an energy ε given by (1.11), would have an associated
energy density and a graviton number density. The depletion of the condensate
must be associated to variations on both magnitudes in that volume. In this manner,
it is thermodynamically consistent to associate a chemical potential µ to take into
account the change of the particle number N (according to Dvali’s picture, in the
phase transition).
In Chapter 2 we present a simple way to introduce the chemical potential at the
level of the Einstein’s equations. In this geometrical approach, this potential would
regulate the components of a 2−rank tensor hµν , associated to perturbations of the
causal structure of a black hole; for instance (1.2). The derived field equations resem-
ble the non-linear Gross–Pitaevskii equation usually employed for modeling Bose–
Einstein condensate. And the corresponding solution seems to correctly describe a
self-sustained graviton condensate over a confining potential given by the metric it-
self. Besides, the condensate is intimately related to the classical field that sustains
it and determines its characteristics. This suggests the possibility of interpreting the
solution as a collective wave function in the framework of mean field approxima-
tion (usually a fairly good description for Bose–Einstein condensates). Remarkably
enough, the characteristics of the resulting condensate are uniquely described by the
Schwarzschild radius.
In [30] analogous solutions for Reissner–Nordström and de Sitter metrics have
been derived. In Chapter 3 quasilocal-energy considerations, well established in
black hole physics, have been extended to the obtained condensate-like solutions.
In all cases the BEC structure is energetically favourable in the classically forbidden
region. As mandatory, all solutions recover the classical characteristics in the visible
side of the horizon, where the chemical potential vanishes. Besides, the quasilo-
cal energy can be used to define a quasilocal potential that may shed light upon
the mechanism regulating the absorption and depletion of gravitons. The deep well
structure around the horizon mediates the binding of matter falling into the black
hole. Gravitational radiation produced by the infalling matter seems to feed the
1.5. The accelerating universe 11
graviton condensate, then it is natural to interpret the chemical potential as the con-
jugate variable to the number of gravitons N.
On the contrary, in Chapter 4 it is shown that it is not possible to define an anal-
ogous mechanism behind a cosmological de Sitter horizon. Then one is not able to
generate a Newtonian potential as the result of a coherent state.
1.5 The accelerating universe

It is usual for imagination to go faster than technology and Einstein’s equations (1.1)
are not the exception. Only a year after Schwarzschild derived his first exact solu-
tion (1.2), theoretical consequences rapidly appear also in the large distances limit.
Einstein himself applied his theory to the universe as a whole.
At the time, the universe was thought to be static and the low relative velocities
of the observed stars support this belief. The "greatest blunder" story is well know;
Einstein knew that there was a missing piece for the universe’s existence to take
place. The dynamical nature of the field equations told him that the gravitational
attraction would cause everything to collapse, whatever the initial equilibrium con-
dition chosen. Hence, following the current observational presumptions, he added a
universal constant λ to counterbalance the effects of gravity and achieve the staticity;
an attractive gravitational force being balanced by a repulsive cosmological constant
force. Using the weak limit for the gravitational field and low velocities as a compass
that guided him to the well establish Newtonian Poisson’s equation, he proposed the
following modification
∇2 V (r ) − λ V (r ) = 4π GN ρm , (1.19)
where the subscript m stands always for matter, whatever its type; in this case the
visible matter energy density. In such a manner, a constant gravitational potential
V ≡ −4πGN ρm /λ would produce the desired static universe [31]. From now on, we
change to the terminology of Λ for the cosmological constant. In accordance to his
proposal of modifying the Poisson’s equation, Einstein modified his field equations
(1.1) introducing this new constant of nature6
Gαβ − Λ gαβ = κ Tαβ . (1.20)
initiating the field of relativistic cosmology; another natural playground for Ein-
stein’s theory of gravity. He obtained the value Λ = 4πGN ρm working in the non-
relativistic limit; and this would allow a static, homogeneous and spherically closed
universe with radius of curvature given by rcurv. ' Λ−1/2 .
6 The sign of the cosmological term depends on the signature of the metric. Equation (1.20) corre-
sponds to a (+ − −−) signature. On the contrary, for a signature (− + ++) the sign is a plus. See for
instance equation (4.16).
This exact balance was completely ad hoc, and in fact it was a thought that the
cosmological constant spoiled the beauty and simplicity of general relativity. How-
ever, it was not until more than a decade after, in 1929, that Hubble’s observations
seemed to indicate that the universe was expanding [32]. He confirmed experimen-
tally the very famous Hubble law, the linear relationship between the distance of
extragalactic sources and their recessional velocity v = H0 D. The Hubble constant
is the Hubble’s parameter at the time of observation
H ( T0 = 0) = H0 . (1.21)
A significant tension between local and non-local measurements of the Hubble con-
stant has been inferred from recent data. On the one side, the cosmic microwave
background (CMB) measurements [33] have reported a value H0 = (67.4 ± 0.5) km
s−1 Mpc−1 (Planck, 2018); contrary, direct measurements in the local universe using
supernovae type Ia as standard candles [34] have given H0 = (73.24 ± 1.74) km s−1
Mpc−1 (SNe Ia, 2018b). Latest measurements imply a discrepancy on the value for
H0 of more than 4σ and less than 6σ [35]. This shows that the discrepancy does not
seem to be dependent on the method, team, or source. Despite the controversy be-
tween late and early universe’s techniques, the dynamical nature of our universe is
commonly accepted. However, intriguing questions regarding the local behaviour
of the Hubble constant triggered an intense debate that is still ongoing.
Coming back to the cosmological constant story [36]. Einstein failed to see that
with or without cosmological constant, general relativity equations predict that the
universe cannot be static7 . The person who told everybody to change the idea of
space and time did not dare to predict what his theory was telling: the cosmic con-
traction or expansion with or without a cosmological constant, or even with the fine
tuned value he had chosen despite of the instability problems he overlooked. What’s
more, as pointed in [37], it may not be even true that the introduction of Λ was be-
cause of a matter of cosmology. Einstein knew about this term much earlier than his
cosmological work. In [38] Einstein derived the gravitational field equations from a
list of physical requirements. In a footnote he notice that their equations were not
the most general possible ones; between other possible terms he mentioned the cos-
mological term. Hence Λ was not an appendage added by Einstein to his theory to
account for observations. Einstein seemed to know about Λ being an integral and nat-
ural part of his theory. It was not a matter of "beauty" or "mystery", the cosmological
constant seems to be one more constant in our fundamental theories. It seems that
the "blunder" was not a physically wrong prediction (the existence of a cosmological
constant), but his silly prejudices that blinded him and made him loose the chance of
doing an easy and spectacular prediction. Einstein missed an instability8 that could
7 For a homogeneous and isotropic universe, Einstein’s equations (1.20) take the structure written in
(5.4). Taking the time derivative of the fist equation and considering ordinary matter with an energy
density obeying ρ ∼ a−3 , one finds that for small perturbations a = a0 + δa the equilibrium is always
unstable.
8 Pointed out later on by Eddington [39].
1.5. The accelerating universe 13
have led him to predict the cosmic expansion in 1917.

In the subsequent decades, much progress was made and cosmological data
strongly spoke in favor of a positive cosmological constant at very large scales. The
end of the 20th century brought the extraordinary discovery that our universe is
expanding at accelerated rates at late eras. Two independent groups carried out ac-
curate measurements of SNe Ia luminosity distances as a function of the redshift,
measuring Λ > 0 with a confidence level of more than 3σ [40]. These results have
been reinforced by different observations as CMB and the Baryon Acoustic Oscilla-
tions (BAO) among others. Today is widely accepted that the easiest explanation for
this accelerated expansion is the dominance of Λ over matter at late eras. Best fitting
data says that over the 68% of the energy density budget is in form of dark energy.
These results are compatible with the existence of a cosmological constant in the
current-epoch (see [33, 41] for recent articles). At the level of the equations this term
counteracts the gravitational pull of matter with an anti-gravity effect, maintaining
the general covariance of the theory; if small enough it passes the Solar System tests.
However, the elusive nature of the cosmological constant still remains a mystery,
and its very small value [42],
Λ ∼ 1.11 × 10−52 m−2 , (1.22)
does not make things easy. Until now the unobservable character remains at not such
large distances. The cosmological constant problem is still an outstanding theoretical
challenge: it is not so clear whether Λ is an effective property valid only at very large
scales or, on the contrary, a fundamental property of the spacetime. If the later is
the case, it may be observable, as a matter of principle, at any scale. As can be seen,
many questions concerning the ultimate nature of the cosmological constant are still
unanswered.
Once Λ is included in general relativity, its implications immediately surpass the
cosmological arena. One of its specific characteristic is that it does not suffer any
clustering effects (contrary for instance, to the scalar field dark energy). Interesting
efforts turn to local observable effects, where local stands for sub-cosmological scales.
For instance, whether it should affect the orbits of the Solar System and/or of dou-
ble pulsars. The resulting effects seem to be too small for being detected [43]. At
higher scales, effects are expected to be more relevant; corrections to the Virial re-
lation are considered for extended galaxy clusters [44]. Other interesting possibility
is assessing the presence of the cosmological constant for the case of light bending
from distant objects9 . Different studies derive different conclusions in which the or-
der of magnitude of the effects of Λ ranges from zero [45] to appreciable values [46],
moving through unobservably small ones [47]. All in all, these issues remain as an
open debate.
9 Even involving cosmological distances.
1.6 Gravitational waves propagating through an expanding

universe
It is well known that general relativity predicts the existence of gravitational waves.
For describing the inspiraling, merger and final relaxation (ringdown) of black hole
binaries one needs nothing other than vacuum spacetime dynamics. The field equa-
tions simply become
Rαβ = 0 (1.23)
and describe purely curved spacetime. When a pair of black holes merge, an im-
mense amount of energy is given off as ripples in the curvature of spacetime propa-
gating at the speed of light c.
It is well established that the inclusion of a spread energy density immediately
produces a non-trivial curvature of the spacetime. Therefore the logic would lead
us to expect that the propagation of gravitational waves in our universe differs from
that over a flat Minkowskian universe (1.23). The usual treatment of gravitational
waves is to perform small perturbations around a background, but in this case it
should be a non flat background. These issues and more have been reported in
[48], where linearized Einstein’s equations in the presence of Λ have been solved.
For obtaining the solution, a coordinate choice is mandatory. The presence of a
cosmological term adds new terms in the linearized equations. The relevance of
these terms together with a careful discussion on the coordinate choice has been dis-
cussed in [49]. In fact, the easiest choice to solve the set of equations is the Lorenz
gauge (or Λ gauge) where the linearized equations are closely related to those of
flat spacetime. These coordinates are nothing but different parametrizations of the
Schwarzschild-de Sitter (SdS) spacetime. However, the only coordinate system we
can make sense of is one corresponding with the cosmological coordinates (FLRW,
because the acronyms of Friedmann, Lemaître, Robertson and Walker); i.e. the
one where the universe appears to be homogeneous and isotropic. For a de Sitter
universe, the change of coordinates between both coordinate systems, from SdS to
FLRW, is intricate but it can be worked out. Nonetheless, for the purpose of the topic
under discussion we will see that it is enough only with its linearized version.
Once we know how to move between different coordinate systems, it is direct
to transform the gravitational wave solution (easily found in SdS) to coordinates
where we can make observable predictions. The conclusions of [49] are prominent:
the transformed wave functions get modifications in their amplitude and in their
√
phase of order O( Λ), because the propagation takes place over a dynamical back-
ground. We will see that the latter modifications would imply a possible way to
measure locally the effect of the cosmological constant. When waves travel they are
redshifted but in a different way from the usual gravitational redshift usually used
for electromagnetic radiation. The dispersion relation is not linear anymore; it re-
lates an effective frequency we f f with an effective wave number k e f f . Of course the
usual gravitational redshift for the frequency is recovered, but as it will be explained
1.7. Gravitational wave detection 15
in Chapter 5, there is much more to say about the consequences of we f f 6= k e f f .

This chapter presents the results obtained in [50]. Towards a realistic cosmology, we
derived how to incorporate dark matter as a different background to the existent so-
lution in de Sitter. This new type of matter has been modeled as a pressureless and
non-interacting cold fluid. Although mathematics get obscured because things get
rapidly difficult when combining different types of matter, according to the previ-
ous discussion we can take advantage of the linearized arguments. Then a linearized
version of the change of variables for these two components can be easily derived.
One step further has been given in [51], where it is argued that this change of vari-
ables can be generalized to an arbitrary background, showing also the inclusion of
radiation; however the validity of this goes only up to linear order. Certainly, the
effect of the background over the gravitational wave propagation is ruled by the
Hubble parameter today ( H0 ), and it is not possible to disentangle the individual
effect corresponding to each component; at least not at a linearized level.
1.7 Gravitational wave detection

Detection of gravitational waves is expected to be performed by pulsar timing array
(PTA) observations sometime in the future. The experimental setup consists of an
array of radiotelescopes for the accurate timing of a galaxy-sized net of rapidly-
spinning neutron stars across the Milky Way. These rapidly rotating and highly
magnetized stars called pulsars emit beams of electromagnetic radiation. Among
the thousands of existing pulsars, the millisecond pulsars (MSPs) are of particular
interest given their very stable periods and the absence of glitches10 .
Three PTAs consortia carrying out such kind of observations have joined efforts
in an international collaboration called the International Pulsar Timing Array (IPTA)
[53]. The aim of the collaboration is to observe variations in the time of arrival (ToA)
of the electromagnetic signals coming from the pulsar array. The differences between
the predicted and measured ToAs are known as "timing residuals". Pulsar-timing
observations, unlike LIGO [54] or Virgo [55], are sensitive to supermassive black
hole (SMBH) binary mergers [56]. These binary systems composed by black holes
with masses in the range ∼ (107 − 1010 ) M , are amongst the primary candidates
producing the kind of low-frequency gravitational waves (between the nanohertz
to microhertz) aimed to be measured by IPTA. In [57] it is argued that the square
kilometer array project (SKA) constitutes a significant improvement with respect to
current PTAs projects. SKA will detect many more MSPs than what is currently
possible and will allow to time them with very high precision, making them very
sensitive to the small spacetime perturbations of gravitational waves.
Gravitational waves are emitted by very distant sources, then it is natural to ex-
pect a gravitational wave signature on the timing residuals modeled as a plain wave
propagating in a approximately flat spacetime. Then the expansion of the universe is
10 Except for some very small glitches detected in some particular MSPs; see for instance [52].
included as an ad hoc redshift in the frequency. Nonetheless, the propagation is over

a nearly de Sitter universe, not a flat one, and we already know that there is more be-
sides the known redshift. In [51] the propagation over non-flat spacetimes has been
derived but from another point of view. Chapter 6 explicitly shows the wave equa-
tion in comoving coordinates. It is obtained from the linearized field equations for
small perturbations around a FLRW expanding background. As expected, the corre-
sponding wave-like solution coincides with the solution obtained in Chapter 5 in the
Lorenz gauge, after performing the change of variables to cosmological coordinates.
Analogously to the previous discussion, the aforementioned wave front suffers an
amplitude and a phase modification. In Chapter 6 we also show that the phase ef-
fect is reflected in an anomalous enhancement of the timing residual registered by
pulsars, but only for the ones located at a particular angle subtended between the
line observer-pulsar and the line observer-source.
In summary, if waves are modeled bearing in mind that the propagation takes
place in a dynamical universe, the local enhancement of the timing residual for cer-
tain pulsars could have an impact in PTA observations [58]. The signal develops a
peak-like pattern centered at a polar angle α with respect to the source direction as
seen from Earth. The location of the peak strongly depends on the distance to the
gravitational wave source (and on the value of Λ, or correspondingly to a realistic
cosmology, the value of H0 ). According to the validity of the linearized analysis, the
timing residual enhancement can be found at polar angles between α ∼ 10◦ − 40◦
with respect to the source direction. Even more, the theory states that the effect
has a symmetry of revolution around the azimuthal angle β perpendicular to the
source direction. This means that given a fixed source emitting gravitational waves,
at a characteristic distance where pulsars of our galaxy are placed, one can find en-
hancement rings where the timing residual registers the discussed anomalous en-
hancement. As mentioned in [59], where the differential timing residual along the
electromagnetic geodesic is examined, the peak corresponds to a valley where the
phase is (nearly) stationary.
The height of the peak depends considerably on the location of the pulsar emit-
ting the periodic electromagnetic beams; further away pulsars present significantly
bigger signals which in some cases are even an order of magnitude bigger than if the
computation is performed in a Minkowskian spacetime. We took a modest thresh-
old as an example stating that we can discern pulsars with an enhancement mag-
nitude of half the peak. Then any pulsar located inside the region bounded by the
corresponding angles for this half peak (i.e. a region α ± e) would experience an
anomalous timing residual in their ToAs. This interval corresponds to the width of
the previously defined enhancement ring, which may be up to tens of kpc.
Supporting the interesting detailed statistical analysis carried out in [58], where
a real array of pulsars taken from the ATNF catalog [60] has been used to compute
the statistical significance of the intensified timing residual, the last chapter is also
focused on the consequences of the characterization on the dimensions of peak-like
1.7. Gravitational wave detection 17
part of the signal. The linearized treatment has a range of validity. From our analysis
it follows that the effect is presented for sources located between hundred of Mpc to
a few Gpc away (which as mentioned before determines the α location). These dis-
tances might correspond to single sources SMBH merger observations for PTA and
because of the strength of the derived effect, the increased value can maybe facilitate
a first detection of gravitational waves by the IPTA collaboration. If this observable
effect is in fact present, we have an interesting method for choosing new pulsars
apart from the nearly half a hundred already monitored by IPTA. Also an hypo-
thetical detection of gravitational waves taking into account these enhanced signals
would imply experimental evidence in favour of local effects of the cosmological
constant.
19
Chapter 2
Black holes as Bose–Einstein

Condensates
2.1 Summary of our current understanding

In an interesting saga of papers, Dvali, Gómez and coworkers [25, 26, 27, 61] have
put forward an intriguing suggestion: black holes could perhaps be understood as
Bose–Einstein condensates of gravitons. This proposal might be scientifically useful
despite is based on a brutal approximation. Of course it might fail to capture aspects,
but it might work to some extent, especially in the long wavelength regime and the
implications can be conceptually staggering.
The condensate model of course cannot be a complete description of what hap-
pens in the black hole, but revealed a quite fruitful relation: it establishes a bridge
between quantum information and black hole physics never discussed before. This
would reveal new approaches on the deep quantum nature of such fascinating ob-
jects, and could lead to an alternative understanding of some of the possibly most
striking features of black holes. For instance, Hawking radiation [8, 62] could be un-
derstood as being due to leakage from the condensate. Besides, this picture brings
new ideas about the Bekenstein entropy [6], the absence of hair [63], as well as the
quantum nature of information storage and the possible information loss in black
holes [64].
The main point of these works is that the physics of black holes can be un-
derstood in terms of a single number N, the number of (off-shell) gravitons con-
tained in the Bose–Einstein condensate. These condensed gravitons have a wave-
√
length λ ∼ N ` P , being ` P the Planck length. They have a characteristic interaction
strength α g ∼ 1/N and the leakage of the condensate leads to a Hawking temper-
√
ature of order TH ∼ 1/( N ` P ), equal to the inverse of λ. The mass of the black
√ √
hole is M ∼ N MP and its Schwarzschild radius therefore is given by rS ∼ N ` P ,
thus agreeing with the Compton wavelength of the quantum gravitons, in accor-
dance with the uncertainty principle that dictates λ ' rS in the ground state of the
quantum system. Therefore, up to various factors of the Planck mass everything is
governed by N, the number of intervening gravitons. Modulo some assumptions, all
20 Chapter 2. Black holes as Bose–Einstein Condensates
these results stem from the basic relation postulated in [25]
N
rg = (2.1)
M
that relates the number of gravitons N, the mass of the gravitating object M and its
gravitational radius r g . For a Schwarzschild black hole r g = rS .
We found these results intriguing and we tried to understand them in a different
language, with the aim of being more quantitative. In [29] the previous relations
have been rederived from rather simple assumptions.
It is well known that a Bose–Einstein condensation of a spatially homogeneous
gas with attractive interactions is precluded by a conventional phase transition into
either a liquid or solid. Even when such a condensate can form, its occupation num-
ber is low. Repulsive forces act to stabilize the condensate against collapse, but gravi-
tons do not have repulsive interactions, at least naively, and therefore one would
conclude that a Bose–Einstein condensate is impossible to sustain, particularly as we
expect N to be very large. However, it is up to the equations themselves to establish
whether such a condensate is possible or not. In addition, in theories of emergent
gravity, the ultimate nature of gravitons may involve some kind of fermionic degrees
of freedom (such as e.g. in the model suggested in [65]). Then, repulsion is assured
at some scale and fundamental collapse prevented.
The results in [29] suggest that condensation is possible. However in our analysis
we have separated slightly from original line of thought of Dvali et al. This approach
is not geometric at all. There is no mention of horizon or metric. On the contrary,
we adopt a more conservative approach; we assumed the preexistence of a classical
gravitational field created by an unspecified source that generates the Schwarzschild
metric. As it is known, nothing can classically escape from a black hole, so if we
wish to interpret this in potential terms (which of course is not correct but serves us
for the purpose of creating a picture of the phenomenon) it would correspond to a
confining potential. On and above this classical potential one can envisage a number
of quantum fields being trapped. For sure there is a gravity quantum field present;
hence gravitons (longitudinal gravitons that is). It is expected that other quanta may
get trapped by the black hole potential as well, along the following sections we will
discuss such issue.
Continuing with our semiclassical analogy, these gravitons, and maybe other
quanta, have had a long time to thermalize in a static (and "eternal") Schwarzschild
black hole. It is therefore natural to expected that after cooling, a Bose–Einstein con-
densate of gravitons could eventually form. Of course these gravitons are in no way
freely propagating transverse gravitons. They are necessary off-shell (q2 6= 0) and
thus have some sort of effective mass1 .
In [29] we give a set of equations that could describe a Bose–Einstein condensate
1 It may help to get a picture of the phenomenon to think of them as quasiparticles.
2.2. Building up a condensate over a Schwarzschild black hole background 21
constructed on top of a classical field created by a black hole. This set of equa-
tions would be identified as the counterpart of the Gross–Pitaevskii equation [66].
Remarkably enough, the characteristics of the resulting condensate are uniquely de-
scribed in terms of the Schwarzschild radius of the black hole and the value of a
dimensionless parameter, interpreted as a chemical potential. A condensate appears
not only to be possible but actually intimately related to the classical field that sus-
tains it and determines its characteristics. It would therefore be tempting to go one
step beyond and reverse the order of the logical implication; and eventually attempt
to derive the classical field as a sort of mean field potential à la Hartree–Fock. How-
ever this will be left for future research.
2.2 Building up a condensate over a Schwarzschild black hole

background
Let us start with the Einstein field equations (1.1) describing the spacetime curvature.
These equations are constructed with the metric tensor gαβ which in this context pro-
vides the geometric and causal structure of the built condensate and the spacetime
where it is immersed. We will denote by g̃αβ to the background metric that in this
section will invariably be the Schwarzschild metric. Perturbations above this back-
ground metric will be denoted by hαβ , so as
gαβ = g̃αβ + hαβ . (2.2)
To initiate our program we should, first of all, identify an equation (or set of
equations) that could provide a suitable description of a Bose–Einstein condensate
in the present context. In other words, we have to find the appropriate generaliza-
tion of the Gross–Pitaevskii equation. The graviton condensate has necessarily to be
described by a second-degree, symmetric tensor field hαβ that within our philoso-
phy has to be connected necessarily with a perturbation of the classical metric. In
order to keep things as simple as possible we will attempt to describe only conden-
sates with quantum states having l = 0. This is translated to spherically symmetric
excitations of the metric of the form
r S −1

2 rS 2
ds = − 1 − + htt dt + 1 − + hrr dr2 + r2 dΩ2 . (2.3)
r r
The Einstein tensor derived from the previous metric will be expanded up to sec-
ond order in hαβ to retain the leading non-linearities (self-interactions of the desired
condensate).
It is well known that Birkhoff’s theorem [67] guarantees the uniqueness of the
solution of Einstein’s equations in vacuum with the properties of having spherical
symmetry and being static. As every dimensionful quantity can be expressed as a
function of the Schwarzschild radius rS = 2GN M, the difference between a given
Schwarzschild metric and any perturbed solution built upon it must necessarily cor-
respond to a change in the mass M → M + δM. Even though this result is well
known [68], it is interesting to give a short review in a perturbative formalism. The
relevant equations of the classical theory give the geometrical structure of the con-
densate, hence the corresponding part of the field equations. For the sake of simplic-
ity these equations are included in Appendix A.
The fact that spherical symmetric perturbations are related to shifts in mass, in-
dicates a correspondence between any perturbation hαβ (i.e. each graviton) and a
certain amount of energy that is reflected in a change of the black hole mass.
2.2.1 Einstein’s equations as Gross–Pitaevskii equations
The familiar Gross–Pitaevskii equation [66] employed to describe Bose–Einstein con-

densates is a non-linear Schrödinger equation; i.e. an equation of motion that con-
tains self-interactions (hence the non-linearity), a confining potential for the atoms
or particles constituting the condensate, and a chemical potential, the conjugate ther-
modynamic variable of the number of particles or atoms contained in the conden-
sate.
Among all these ingredients, perturbed Einstein’s equations already contain most
of them. They are already non-linear and while there is no an explicit confining po-
tential (as befits a relativistic theory) they do confine particles, at least classically,
because if the selected background corresponds to a Schwarzschild black hole, the
strong gravitational field classically traps particles inside the horizon. However,
there is one ingredient still missing, namely the equivalent of the chemical poten-
tial. Therefore we have to extend the formulation of perturbations around a classical
black holes solution to the grand-canonical ensemble by adding to the appropriate
action a chemical potential term.
As it is well known since the early days of quantum field theory [69] there are no
conserved currents or continuity equations for fundamental fields that are chargeless
(such as a real Klein–Gordon field). Therefore there is no way of defining a number
operator for freely propagating gravitons or photons.
However in the picture of [25, 26, 27, 61] the situation is different. If a Bose–
Einstein condensate is present with gravitons acquiring all the same momenta ∼ 1/rS
and being weakly interacting, for macroscopic black holes (recall α g ∼ 1/N) the
total energy stored in the condensate should be ∼ N/rS and this quantity would be
a conserved one. Therefore, lest rS change, N would be conserved too. The previous
reasoning shows very clearly that the gravitons contemplated in the present scenario
(of "maximum packing"), if realized, have nothing to do with freely propagating
gravitons as the number of massless on-shell gravitons is not a conserved quantity.
The energy contained in a given volume occupied by a non-interacting scalar
field is
1
Z
E=− dV φ ∇2 φ ; (2.4)
2
2.2. Building up a condensate over a Schwarzschild black hole background 23
by analogy, in the present case
1
Z Z
2 αβ
E= dV ε ĥαβ ĥ = dV ε ρĥ , (2.5)
2
where we assume that the energy per graviton ( ε ) is constant and approximately
given by ε = 1/λ with λ = rS , and
1 1
ρĥ ≡ ĥαβ ĥαβ . (2.6)
2 λ
While there is no formally conserved current, the above quantity can be interpreted
as a "graviton number density", which in principle is related to the perturbation
tensor as ĥαβ = MP hαβ to keep the correct dimensions. The Planck mass is used
is favour of the universal gravitational constant, MP 2 = h̄/GN . The integral of the
graviton density (2.6) in the interior of the black hole has to be interpreted as the
number of constituents of the condensate.
The above considerations can now be phrased in a Lagrangian language. The
chemical potential term in the action should be related to the graviton density of the
condensate inside a differential volume element dV. In order to respect the basic
symmetry of the general theory of relativity, the simplest form of introducing such a
term is by means of
1
Z
∆Schem.pot. d4 x
p
=− − g̃ µ ĥαβ ĥαβ . (2.7)
2
g̃ stands for the determinant of the background metric. This term does of course
resemble a mass term for the spin-2 excitation and indeed it is some sort of effec-
tive mass in practice as the gravitons in the condensate are quasi-non-interacting.
However we will see that general relativity eventually requires for the quantity µ to
transform as a scalar and to be position dependent, i.e. µ = µ(r ), from requirements
of self-consistency of the proposal.
The additional term it is invariant under the gauge group of diffeomorphism.
Part of this statement is shown in [70]: under an infinitesimal displacement in the
coordinates of the form δD x α = ξ α the full metric changes as the lie derivative
along the direction of the displacement, δgαβ = LD gαβ = ξ α;β + ξ β;α . The same
gauge transformation rules the background metric. As the perturbation is defined
by hαβ = gαβ − g̃αβ , under the same perturbation of the coordinate system, the same
rule of covariant transformation is obtained. Together with the fact that the chemi-
cal potential µ behaves as a scalar under a general coordinate transformation, ensure
automatically the general covariance of the theory. However, while the addition is
diff invariant, it is not background independent. The separation gαβ = g̃αβ + hαβ
leads to an action that depends on the choice of the background metric, which is the
one of the chosen black hole. Likewise it is expected for µ to also depends on the
background metric.
Therefore an appropriate action for the field hαβ is

Z Z
2 4 2
1
d4 x
p p
S( h) = MP d x −g R − 2 MP − g̃ µ hαβ hαβ . (2.8)
Indices are raised and lowered using the full metric gαβ = g̃αβ + hαβ in order to pre-
serve diffeomorphism invariance, that is an exact invariance of the above action. A
completely covariant expansion in powers of hαβ can be performed up to the desired
order of accuracy. By construction these equations will be non-linear, but we will
keep only the leading non-linearities to maintain the formalism simple and analyt-
ically tractable. The left hand side is just the Schrödinger part, the additional piece
is proportional to the chemical potential, both giving a Gross–Pitaevskii like equa-
tion(s). It is worth noting again that the action (2.8) is as shown reparametrization
invariant, but it is not background independent and it should not be. The fact that
fluctuations take place above a black hole background (in whatever coordinates one
chooses to describe it) does matter.
The action principle yields the two equations of motion for the perturbation field
hαβ = diag(htt , hrr , 0 , 0); namely
q
g̃
Gαβ ( g̃ + h) = µ g hαβ − hασ hσ β . (2.9)
that can be interpreted as the Gross–Pitaevskii equations describing the properties

of a graviton condensate "sitting" in the black hole interior. The explicit derivation
for these equations is included in Appendix B. It is important to keep in mind the
following: we are working here in the grand canonical ensemble; this implies that
the magnitude µ is an external field and does not vary in the action. In particular,
for these equations of motion it is independent of hαβ , so δµ/δhαβ = 0. This should
be the way of introducing the chemical potential. Otherwise, were µ not an external
field, it would be necessary to take it into account when performing variations to
derive the equations of motion. An equation of motion would be obtained for µ and
this equation would nullify automatically the perturbation as well as the chemical
potential itself, i.e. hαβ = 0 and µ = 0. Therefore, the external field µ is introduced
in the theory as some kind of Lagrange multiplier. As Lagrange multipliers have
constraint equations, the chemical potential of the theory may not be arbitrary at
all. It must satisfy binding conditions with hαβ . The main difference is that these
constrains are implicit in a general theory of relativity. The restriction of µ is through
the general covariance conditions for the action. In other words, the diffeomorphism
covariance implies the covariant conservation of the Einstein tensor, and this in itself
entails the same for the chemical potential term
h q i
g̃ ;β
∇ β Gαβ = 0 =⇒ µ g hαβ − hασ hσ β =0. (2.10)
The covariant derivative is defined using the full metric gαβ = g̃αβ + hαβ . This dif-
ferential equation of motion is valid up to every order in perturbation theory.
2.3. Outside the horizon 25
Subsequently, we will proceed to find acceptable solutions of these equations and

interpret them. We will separate the problem into two regions: outside and inside
of the black hole horizon in sections 2.3 and 2.4 respectively. We certainly expect
that the graviton condensate will disappear quickly in the outside region; imposing
this as a boundary condition for r rS we will see that in fact the condensate is
identically zero on this side of the horizon. On the contrary, a unique non-trivial
solution will be found in the interior of the black hole.
2.3 Outside the horizon

In this section it will be shown that even after the inclusion of µ there is not other
normalizable solution outside the black hole horizon than the trivial solution for the
perturbation.
With the aim of motivating this solution it is practical to start with a perturbative
analysis, retaining only the leading order in hαβ on the right hand side of (2.9) and
neglecting possible terms of the form µh2 on the grounds of being considered of
O(3). This treatment seems natural since the chemical potential (if present –despite
of being null in this region–), far away from the condensate, should behave at most
as a perturbative small parameter; and small changes in the chemical potential would
be responsible of producing a small perturbation over the background metric. As a
plus, the O(2) system of equation takes a simple form, very similar to the familiar
Gross–Pitaevskii equation,
(1)
Gαβ (2)
(h) + Gαβ (h2 ) = µ hαβ . (2.11)
The zero order corresponds to the vacuum solution and vanish. The left hand side
has been already introduced in (A.1).
2.3.1 Asymptotic analysis
If we place ourselves far away from the black hole, in the r → ∞ limit, one is able
to keep only dominant terms in the Einstein’s equations. As we want a vanishing
perturbation at infinity, the following ansatz is imposed in the faraway region:



 htt = A r −a




hrr = B r −b (2.12)




µ = C r −c



with a, b, c > 0.
Before proceeding, an additional consideration is needed: since for us h repre-
sents a localized Bose–Einstein condensate, in analogy with the wave function of a
confined particle, the perturbation must be a square-integrable function. This is

Z ∞ Z ∞
d3 x − g̃ h2 ≡ d3 x − g̃ htt 2 gtt 2 + hrr 2 grr 2 < ∞ .
p p
(2.13)
rS rS
There are at least main two reasons that lead us to the previous requirement. Let
us review first why one has to request square-integrability. Note that the solutions
of (2.9) should not be understood as a quantum field, but rather as solutions of the
Gross–Pitaevskii equation. Here we adopt Bogoliubov theory [28] and its interpre-
tation of the Gross–Pitaevskii wave function. This is the commonly accepted inter-
pretation and essentially it boils down to the fact that in the large occupation limit
(N → ∞) the creation and annihilation operators of the ground state can be ap-
proximately treated as commuting c-numbers, and hence the many body quantum
problem is described by a classical function called the "macroscopic wave function"
or simply the "order parameter". Because the modulus square of this quantity is pro-
portional to a† a (we introduce the well known annihilation and creation operators a
and a† to define the number operator), it is therefore proportional to N. Hence the
√
order parameter itself is proportional to ∝ N/V. If the potential is not uniform,
N may of course depend on the coordinates. From this the need to require square-
integrability follows. This interpretation is in the present case also supported by the
dimensionality of ρĥ in (2.6).
Why then this precise condition? The leading role of the geometry has been to
provide, in an indirect manner, a "confining potential" that traps gravitons inside the
horizon. Therefore, from this point of view, when giving a physical interpretation
to the chemical potential (see section 2.5) one implicitly assumes that the geome-
try of the spacetime is flat and that geometry acts via the external potential and the
√
chemical one. In the present situation, it turns out that − g̃ ∝ r2 sin θ coincides
with the one corresponding to a 3 dimensional spatial flat metric in spherical co-
√
ordinates; therefore d3 x − g̃ is reparametrization invariant in 3 dimensions. This
interpretation is equivalent to defining a wave function normalized by the tempo-
√
ral component of the metric tensor, ψ = 4 − g̃tt h. Then, any magnitude computed
√
by means of an integration over a 3-spatial volume d3 V = d3 x γ (where γ corre-
sponds to the determinant of the 3-spatial metric γij ) is effectively written in a fully
2 √ √ √
covariant way as d3 V ψ = d3 x γ − g̃tt h2 = d3 x − g̃ h2 , since the determinant
of the 4-dimensional metric can be decomposed as g̃ = g̃tt γ, with i, j = r, θ, φ if the
metric is diagonal. Therefore, the volume element is coordinate dependent as well
as the square modulus of the macroscopic wave function ψ, but the combination of
both is not; and we end up finally with the corresponding volume element to the
previously discussed flat spacetime.
Whatever the case, in this perturbative limit the components of the Schwarzschild
metric accomplish g̃tt 2 , g̃rr 2 −→ 1 in the limit of r → ∞; then the integral in equation
(2.13) should fulfill

Z ∞ Z ∞
h
2 2
i dr logarithmic
4π dr r 2
htt + hrr + O(h ) < 3
=∞ . (2.14)
rS rS r divergence
For this to happen, if a power law ansatz at infinity is imposed, the exponents must
obey a, b > 3/2.
After the inclusion of the chemical potential term, two of the there linearly in-
dependent equations coming from (2.11) becomes nonzero, the temporal and radial
due to the new term, while the angular remains unchanged,
B B2 AC
(1 − b ) − (1 − 2 b ) =
r b +2 r2b+2 r a+c
A2 B2

rS A B rS rS AB BC
+a − + +a − +a + =
r r a +2 r b +2 r r2a+2 r r a + b +2 r2b+2 r b+c
(2.15)
A2

rS 2A rS 2B 2rS
+ a2 + − b b +2 + +3a 2
r r a +2 r r r r2a+2
4 B2

rS rS 2 AB
− − b 2b+2 − (2 + b) + 2 a + a b a+b+2 = 0 .
r r r r
The requirement of the solution being square-integrable exclude that a and b could
be zero. This implies automatically the neglection of the terms proportional to the
Schwarzschild radius. Even more, with the previous considerations we can also ne-
glect all terms coming from second order in perturbations, as they vanish faster than
the linear ones. It is natural for the Einstein tensor to become Minkowskian after
these requirements. After all, (2.15) reduce to the following asymptotic identities
B AC
(1 − b ) =
r b +2 r a+ρ
Aa B BC
a + 2
− b +2 = (2.16)
r r r b+c
2 A a2 2 B b
− b +2 = 0 .
r a +2 r
The third equation, the angular component Gθ θ , make explicit how the components
of the perturbation behaves between each other; it is mandatory for both terms to
contribute at infinity. If this is not the case, this would nullify A (or B), and then, via
the two fist equations in (2.16), fix B = 0 (or A = 0) and C = 0; i.e. hαβ = µ = 0. The
competition of both terms is possible if and only if a = b. This condition modifies
the angular equation in (2.16) as
2a
( A a − B) =0 =⇒ Aa = B. (2.17)
r a +2
Then, from the second equation in (2.16) the following relationship can be read
1 BC
( A a − B) = 0 = . (2.18)
r a +2 r a+c
This automatically leads to a null chemical potential as the only possible solution for
this region with the proposed ansatz.
Faster decays
We have also explored the possibility of exponentially vanishing chemical potential

when r → ∞ with analogous conclusions. If we change the ansatz and impose an
exponential decreasing solution for the perturbations at infinity,



 htt = A e a r




hrr = B eb r (2.19)





= C ec r

 µ
with a, b, c < 0, the linear order dominates in front of the higher ones. As decreasing
solutions are expected, the parameters a, b and c must be not null. If one use this
ansatz in the Einstein’s equations partially evaluated at infinity, one finds that in the
angular equation there is no way for both leading terms to compete between each
other
eb r
Gθ θ : −2 A a2 e a r − 2 B b =0. (2.20)
r
In conclusion, depending on the fact whether if a is bigger or not than β, A = 0 or
B = 0. For the first case, if A = 0, the leading terms of the temporal equation
eb r
Gt t : Bb = A C e( a+c) r (2.21)
r
fixes B = 0. Being both, A and B null, the chemical potential disappears from the
theory in this region. For the second case, if B = 0, the same equation Gt t nullifies
A (as no null solutions for C are expected). No perturbation makes the chemical
potential senseless. To sum up, this ansatz implies as well that the only solution
at infinity is a null perturbation, hαβ = 0, and the disappearance of the chemical
potential from the theory, µ = 0.
Likewise it can be seen that a much faster Gaussian-like decay

 2


 htt = A e a r




2
hrr = B eb r (2.22)




2

= C ec r

 µ
is also excluded with analogous calculations. Null perturbation and null chemical
potential is the only solution in this zone.
2.3.2 Numerical analysis
Before concluding this section we will give additionally a numerical argument that
confirms the vanishing of the condensate and the chemical potential in the outer
region when the perturbations are null at infinity.
The following change of variables allows rewriting the relevant components of
the Einstein tensor (A.1) in terms of dimensionless quantities
rS


 z =
 r
(2.23)

X (z) = µ (r ) r 2 .


This quantities are universal as rS and µ are the only physical parameters. This
redefinition maps the exterior part into a compact interval: r = ∞ is transformed
into z = 0 and r = rS into z = 1. It is well known that Schwarzschild spacetime is
asymptotically flat; therefore, at infinity (z = 0) it is expected for the perturbation to
vanish (hαβ = 0). Nevertheless equations are difficult to deal with because they are
singular at z = 0, i.e. it is not possible to isolate the higher order derivatives.
Let us see how we can proceed. As mentioned before, we have to settle for a
point z ∼ 0 (but z 6= 0) to set up boundary conditions for the numerical integra-
tion procedure to start. Likewise setting hrr = 0 at any arbitrary value of z outside
the event horizon, immediately triggers divergences at the first step of the routine.
Therefore we take a "small" initial value for the perturbation in a point near infinity
(z ∼ 0), and then decrease this initial value e; this is
hαβ (r → ∞) −→ e. (2.24)
for a decreasing e 1.
With the set of solutions for the components of hαβ obtained for each of the initial
condition imposed, it is possible to compute the quantity dV h2 defined in equa-
R
tion (2.13). The integral extends over the exterior of the black hole; this is z ⊂ (0 ; 1).
In Figure 2.1 the behaviour of the integral of h2 outside the black hole is presented
when the initial condition for the integration approaches zero at a fixed point far
away from the condensate (near infinity or z ∼ 0). It is observed that this integral is
invariably small. The numerical analysis results very stable and the initial condition
can be reduced as many orders of magnitude as desired. The graph is presented in
a log–log scale, hence the linear behaviour. The points are fitted by a linear function
with a slope approaching 2 as more points with lower initial condition are added.
This implies that the integral converges quadratically to zero as the initial condition
for hαβ nullifies. This seems to confirm —also numerically— that there is no con-
densate, i.e. hαβ = 0, and no dimensionless chemical potential, X = 0, in the outer
region of the black hole.
-4
10
-7
10
-10
10
-13
10
2
-g h
-16
10
dz
-19
10
-22
10
-25
10
-2 -3 -4 -5 -6 -7 -8 -9 -10 -11 -12 -13 -14

10 10 10 10 10 10 10 10 10 10 10 10 10
Decreasing boundary condition
F IGURE 2.1: Each point represents a magnitude proportional to the integral of

h2 outside the event horizon, computed for a particular initial condition for the
perturbation placed near infinity, hαβ (∞). The graph is presented in a log–log
scale. A linear fit gives a slope that asymptotically approach 2 as the boundary
condition is nullified; this implies that the integral vanishes quadratically. The
closer the limit to an asymptotically flat spacetime is (i.e. decreasing the initial
condition near infinity), the smaller this integral becomes.
2.4 Inside the horizon

In order to study the behaviour of our equations when the horizon is crossed, it is
convenient to keep working in the z = rS /r and X = µ r2 variables introduced in
(2.23) from the previous section and redefine the perturbations as



 htt (r ) = (1 − z) γtt (z)
(2.25)


 hrr (r ) = (1 − z) γrr (z) .
2.4. Inside the horizon 31
With the new perturbations the Einstein’s equations (A.1) become more compact
and can be dimensionalized. The three relevant components of the nondimensional
tensor r2 Gα β read as
−(2z + 1)(z − 1)2 γrr − z(z − 1)3 γrr 0 + (4z + 1)(z − 1)4 γrr 2
+2z(z − 1)5 γrr γrr 0
−(z − 1)2 γrr − z(z − 1)γtt 0 − z(z − 1)γtt γtt 0 + z(z − 1)3 γrr γtt 0
+(z − 1)4 γrr 2
(2.26)
h
− 2z(z − 2)(z − 1)γrr + z(z − 2)(z − 1)2 γrr 0 − z(5z − 2)γtt 0
2
−2z2 (z − 1)γtt 00 − 4z(z − 2)(z − 1)3 γrr 2 − z2 (z − 1)γtt 0
+z(7z − 2)(z − 1)2 γrr γtt 0 + z2 (z − 1)3 γrr 0 γtt 0
−2z(z − 2)(z − 1)4 γrr γrr 0 − z(5z − 2)γtt γtt 0
i
−2z2 (z − 1)γtt γtt 00 + 2z2 (z − 1)3 γrr γtt 00 /4 .
Of course with the change of variables, now the prime stand for a derivative with
respect to the dimensionless inverse radial coordinate z.
2.4.1 Linearization near the event horizon
In order to get a feeling for possible solutions to these equations we consider their
linearized approximation. Only terms linear in γtt , γrr and their derivatives are
kept in the region z → 1. Beforehand no restriction for the dimensionless chemical
potential X is imposed, therefore Xγαβ is a priori considered as a linear contribution
in the perturbation. That is
3(z − 1)2 γrr + (z − 1)3 γrr 0 = X γtt
(z − 1)2 γrr + (z − 1)γtt 0 = − X (z − 1)2 γrr (2.27)
2(z − 1)γrr + (z − 1)2 γrr 0 + 3γtt 0 + 2(z − 1)γtt 00 = 0 .
Let us now make the following ansatz for the new perturbations



 γtt ∼ A ( z − 1) a
(2.28)
 b

 γrr ∼ B ( z − 1)
while we leave the chemical potential X = X (z) as a free function. The three equa-
tions take respectively the following form
B(3 + b)(z − 1)b+2 = X A(z − 1) a
B(z − 1)b+2 + Aa(z − 1) a = − X B(z − 1)b+2 (2.29)
B(2 + b)(z − 1)b+1 + Aa(1 + 2a)(z − 1) a−1 = 0 .
It is worth noting that if a = b + 2, all terms contribute as we get closer to the event
horizon and X behaves as a constant. In this situation, we obtain three equations for
the coefficients
B (1 + a ) = X A
B + aA = − X B (2.30)
a[ B + A(1 + 2a)] = 0 .
The system of equations is algebraic for the variable X (that as we have seen should
behave as a constant as z → 1); therefore, it is possible to eliminate X by combining
the temporal and radial equation. In the present ansatz this leads to
B2 (1 + a) + AB + A2 a = 0 . (2.31)
Together with the angular equation, this determines the solutions up to a single
constant. There are two possible solutions for this system

a = 0 

 γtt = A
(i) A = −B =⇒ A (2.32)
b =−2

 γrr = −
( z − 1)2


 A
a =−1

 γtt =
( z − 1)


(ii) A = B =⇒ (2.33)
 A
b =−3 
 γrr = .
( z − 1)3


In any case, at least one of the perturbations is divergent over the event horizon.
Nonetheless, the first solution appears to be integrable, while this is not the case for
the second one. In the case of (i), we get X = −1; while if (ii) is taken as solution,
X = 0 (i.e. no chemical potential at all)2 .
2 Solution ii) represents however a volume-preserving fluctuation at the linear order.
However, because these solutions do not vanish when z → 1, justified doubts

can be casted on the relevance of the linearized equations. Let us examine this point
taking into consideration only the solution (i), which is integrable according to the
considerations of the previous section.
In order to see if this solution is modified when non-linearities are switched on,
we substitute it back in the O(2) equations system where self-interactions matters,
and see if the solution survives or how would it get modified. For a compete O(2)
picture, we need to expand the right hand side of the field equations (2.9). Therefore,
the Einstein’s equations corresponding to the non trivial components of the chemical
potential tensor
q h i
r2 Gt t ( g̃ + h) = X g̃tt − g̃tt g̃rr 1 − 5
2 g̃tt htt − 1
2 g̃rr hrr htt
(2.34)
q h i
r2 Gr r ( g̃ + h) = X g̃rr − g̃rr g̃tt 1 − 1
2 g̃tt htt − 5
2 g̃rr hrr hrr .
In view of the foregoing, near z ∼ 1, the left hand side, given by (2.26), together with
the right hand side, changed to the γαβ perturbation variable defined on (2.25), give
the temporal, radial and angular equations
−3(z − 1)2 γrr − (z − 1)3 γrr 0 + 5(z − 1)4 γrr 2 + 2(z − 1)5 γrr γrr 0
h i
= − Xγtt 1 + 52 γtt − 12 (z − 1)2 γrr
−(z − 1)2 γrr − (z − 1)γtt 0 − (z − 1)γtt γtt 0 + (z − 1)3 γrr γtt 0

h i
+(z − 1)4 γrr 2 = X (z − 1)2 γrr 1 + 21 γtt − 52 (z − 1)2 γrr (2.35)
h
2(z − 1)γrr + (z − 1)2 γrr 0 + 3γtt 0 + 2(z − 1)γtt 00 − 4(z − 1)3 γrr 2
−2 (z − 1)4 γrr γrr 0 + 3γtt γtt 0 − 5(z − 1)2 γrr γtt 0 − (z − 1)3 γrr 0 γtt 0
i
2
+(z − 1)γtt 0 + 2(z − 1)γtt γtt 00 − 2(z − 1)3 γrr γtt 00 /4 = 0 ,
respectively. Replacing the possible solution, γtt = A –this eliminates any derivative
of γtt – and γrr = − A/(z − 1)2 into these equations, we obtain the following relations
3 A − 2 A + 5 A2 − 4 A2 = (1 + A ) A = − X A (1 + 3 A )
A + A2 = (1 + A) A = − X A (1 + 3A) (2.36)
2A 2A 4 A2 4 A2
− + − + =0.
( z − 1) ( z − 1) ( z − 1) ( z − 1)
Therefore, quite surprisingly, the linear solution is still an exact solution of the non-
linear quadratic differential equations. Thus, we conclude that (2.32) is solution of
the second order system of equations.
In addition, this exercise gives an interesting result:
1+A
X'− ∼ −1 + 2A + . . . , (2.37)
1 + 3A
where A is so far arbitrary, also in sign. Note that at linear level we got X = −1 and
the fact that the quadratic equation gives an O( A) correction to this result is consis-
tent as we are implicitly assuming that |hαβ | | g̃αβ |, i.e. | A| 1. The constant A
itself is arbitrary and is not determined by the structure of the equations. Changes
in A appear as an overall factor in the solution.
2.4.2 Numerical analysis
So far we have seen that the solution i) given in (2.32) satisfies not only the linearized
approximation but also the full quadratic equations. Could it happen that the pre-
vious analysis misses some solution? To try to answer this question, we have per-
formed a numerical integration of the basic equations retaining the leading order on
the right hand side of the equations, i.e. the set of equations (2.11) which are much
more easier to approach and would give a hint of the collective modes of the trapped
graviton condensate. Although the structure of these field equations resembles the
Gross–Pitaevskii equation, where the chemical potential term enters together with
a linear factor in the perturbation, we are really interested in solving the equations
given by (2.9); as they can be derived from an action principle.
Of course no matter at which order the truncation is taken, the dimensionless
character of the main equations remains; it is useful to introduce a nondimensional
radial variable r/rS . The integration is performed in a compact spacetime domain
[e ; 1 − e] ⊂ (0 ; 1). 3 It is consistent to use boundary conditions according to the
asymptotic solution derived in (2.32). Parametrized as
γtt (1 − e) = A
(1 − e )2 (2.38)
γrr (1 − e) = − A
e2
γtt 0 (1 − e) = 0
it can be seen that the constant A itself is arbitrary and is not determined by the
equations. Besides, changes in A appear as an overall factor in the solution.
3 It
is impossible to start integrating from either 0 (the black hole center) or 1 (its event horizon)
exactly because as can be seen from (A.1) the system of equations is singular at r = 0 and r = rS ; i.e.
one cannot determine in either point the value of the higher order derivatives because precisely at the
origin and at the horizon they are multiplied by factors that vanish or diverge.
For certain fixed values of A, varying the small value of e, numerical integrations
give curves with common structures as the one presented in Figure 2.2. The conclu-
sion from the analytical study is reinforced after performing this numerical analysis
of the basic O(2) equations. We found only one solution with the same general fea-
tures; in the first plot of Figure 2.2 it is seen that γtt as well as the function related
to the chemical potential X, turns out to be constant in the interior of the black hole.
No other solutions are found.
Along this section, we have seen how the equations simplifies in such a way that
they look simpler if expressed in terms of the functions γαβ . However this is not the
case for the solution; to understand what the solution (2.32) means is better to undo
the redefinition of the components of the wave function (i.e. γαβ → hαβ ) by means of
(2.25). It is of interest to raise one index of the components of the perturbation with
the inverse full metric. In the second plot of Figure 2.2 the results of the numerical
integration for ht t and hr r are shown; as can be seen, the obtained functions are
4 1.0
3 0.8
γ
2
0.6
γ
1
0.4
0
0.2
-1
0.0
0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8
0
5
-1
μ
-2
-5
μ
-3
-10
-4
0.2 0.4 0.6 0.8 0.0 0.5 1.0 1.5 2.0
F IGURE 2.2: In this plot A = −0.8 has been taken; this implies γtt = −0.8 and
X = −0.2 inside the black hole. On the first plot, γtt (dotted) and γrr (dashed)
are represented together with the dimensionless chemical potential X (solid).
On the second plot, the numerical integration shows that the solution of the
O(2) equations has constant and equal components ht t and hr r (same markers
are used for each component). On the third graphic, the components of the
perturbation htt (dotted), hrr (dashed) and the dimensionful chemical potential
µ (long dashed) are plotted. Finally, the fourth graphic gives the behaviour
of the chemical potential, both µ and X. Across the black hole horizon, this
variable experience a jump that in this case is ∆X = ∆µ ' 0.2. As already
emphasized, a whole family of solutions are obtained, parametrized by a real
arbitrary constant A. One can simply trade this constant for the value of ϕ.
revealed to be constant and equal throughout the interior of the black hole
ht t = hr r = constant . (2.39)
The solution for the initially defined variables, the perturbation hαβ , and the
chemical potential µ are presented in the third plot. The components of the per-
turbation turns out to be proportional to the corresponding metric element where
each of the two belongs; i.e. htt = ht t gtt and hrr = hr r grr with constant factors (2.39).
The angular degrees of freedom remain unchanged, hθ θ = hφ φ = 0. The chemical
potential is negative and presents a jump at the horizon; this is shown in the fourth
plot of Figure 2.2. Before drawing conclusions from the solution let us move to the
complete equations; we will see that it is possible to derive an analytic solution from
the system of equations that coincides with our previous asymptotic and numerical
perturbative analysis.
2.4.3 Exact solution
In turn to study the full equation of motion we are interested in, equation (2.9), let
us recover the exact chemical potential term for the right hand side of Einstein’s
equations q
g̃
Xαβ = µ g hαβ − hασ hσ β . (2.40)
We will abandon the perturbative analysis and show an analytic solution for the
field equations. Inspired by the previous analysis, both perturbative and numerical
of the main equations derived at O(2), we reformulate our picture starting from a
constant metric fluctuation with mixed indices (one covariant, one contravariant).
Whereas the numerical integration have proved to be extremely precise in giving
the relation (2.39) between components, the solution presents instabilities at some
very close scale. Consequently let us write first
ht t = ϕt ; hr r = ϕr ; hθ θ = hφ φ = 0 (2.41)
where ϕt and ϕr are r − independent. With such condition, the perturbation takes
the following form
!
ϕt ϕr
hαβ = diag g̃tt ; g̃rr ; 0 ; 0 . (2.42)
1 − ϕt 1 − ϕr
This implies for the total metric to become

1 1
gµν = diag g̃tt , g̃rr , g̃θθ , g̃φφ (2.43)
1 − ϕt 1 − ϕr
and seems to impose an upper limit for the constant values of ϕt and ϕr by the
requirement of a regular metric along the whole interior4 .
The exact Einstein’s equations of motion for the theory simplifies enormously
with respect to the perturbative cases, and reduces to the following two
ϕr 3/2
− 2
= µ ϕt 1 − ϕt (1 − ϕr )1/2
r
(2.44)
ϕr 1/2
− 2 = µ ϕr 1 − ϕ t (1 − ϕr )3/2 .
r
Of course if we expand these two equations up to O(2) in the components of the per-
turbations we recover the equations (2.34) when changing to the picture with mixed
indices, i.e. for ht t = ϕt and hr r = ϕr . These equations are valid also for the external
solution; however, in the case of the external region, we know that Minkowskian
metric must be recovered far away from the sources. This condition implies that
ϕt = ϕr = 0 and the same for the chemical potential µ = 0, as explained in section
2.3.
Among the two possible solutions of the latter algebraical system, only one is
compatible with an acceptable limit for small perturbations, namely ϕt = ϕr = ϕ
(we expect 0 < ϕ < 1). All things considered, the two equations in (2.44) are equal,
therefore the resulting equation of motion for a constant ϕ becomes
ϕ 2
− 2
= µ 1−ϕ ϕ. (2.45)
r
Within our philosophy (2.9) is understood as a Gross–Pitaevskii equation for a con-

densate wave function ϕ. Because the solution is constant within the chosen formu-
lation, the kinetic term drops and one is left with a purely algebraic, mean-field-like
equation5 . Added to this, it can be derived from an action principle. This unique
non-linear equation produces a unique branch of solutions that relate the value of
the condensate with the chemical potential directly from (2.45), and is described by
µ0 1
µ= , with µ0 = − . (2.46)
r2 (1 − ϕ )2
For completeness, let us mention that the picture of a Gross–Pitaevskii like equation
(2.45) providing a dimensionless constant µ0 given by (2.46) should be consistent
with the analytical and perturbative analysis performed to the O(2) system. And
of course it is so; we are retrieving the O(2) solution previously found, with ϕ =
4 The upper bound for the wave function will become clear when computing the number N of
constituents of the condensate; such value for the wave function corresponds to the limit N → ∞.
5 Gross–Pitaevskii equations containing no derivative terms are widely known. For instance when
describing a uniform gas of interacting atoms, the corresponding equation is
g ϕ2 = µ ,
where g is the (repulsive) coupling constant, ϕ the order parameter for the Bose–Einstein condensate,
and µ, needless to say, is the chemical potential.
− A(1 + A) playing the role of the free constant discussed along this section. This
relation between ϕ and A allows to write µ0 ∼ −1 + 2A + . . . which coincides with
(2.37).
Before moving on, it is mandatory to make a comment on the covariant con-
servation of our equations. At the end of section 2.2 we have pointed out that the
diffeomorphism invariance entails a differential equation for the chemical potential,
namely (2.10). Bianchi identities entails the covariant conservation of the Einstein
tensor, and this implies automatically the same for the left hand side of our equa-
tions of motion
√ √
− g̃ − g̃
µ √− g , ν hµν − hµσ hσ ν + µ √− g hµν − hµσ hσ ν ; ν = 0 . (2.47)
The latter equation is a set of four equations for α = t, r, θ, φ, but only one is non
trivial; this equation is the radial one, α = r. When the components of the pertur-
bation are equal and constant, i.e. ht t = hr r = ϕ, the general covariance condition
yields the following differential equation for the chemical potential
ϕ2

2µ
ϕ− ∂r µ + =0. (2.48)
2 r
The integration is direct, µ = µ0 r −2 , and the only degree of freedom is a boundary

condition when integrating this differential equation governing the chemical poten-
tial. This value should coincide with the value of our constant dimensionless chem-
ical potential in (2.46); this is µ0 = −1/(1 − ϕ)2 .
The qualitative result is the existence of a normalizable solution that can be inter-
preted as the collective wave function of a graviton condensate. A unique relation
is obtained between this (constant) wave function and a (also constant) dimension-
less chemical potential. It is to be expected that other quanta may form condensates
too, being described by Gross–Pitaevskii like equations similar in spirit to the one
considered for gravitons. This would lead to the introduction of chemical potentials
conjugate to their respective number of particles. However it is not clear a priori
which equations they obey; in particular which potential energy one should use in
these cases. We will advocate here to use a potential derived from the quasilocal en-
ergy and investigate some possible consequences of this in next chapter. However
let us first discuss in which manner our research connects with other proposals on
corpuscular black holes.
2.5 Consequences of the results and connection with previ-

ous proposals.
In order to present a physical interpretation for the obtained results, let us compute
the number of particles (bosons) present in the condensate. It is immediate to see that
the solution found for the wave function is of finite norm. Taking into account that
2.5. Consequences of the results and connection with previous proposals. 39
the perturbation hαβ is null from the event’s horizon onwards (we have seen that the
condensate is confined behind the event horizon), the endpoint on the integration
limit can be fixed at the Schwarzschild radius. This way, the number of particles is
related to the following integral
Z ∞ Z rS 8π 3 2
3
dr r2 ht t 2 + hr r 2 =
p α
d x − g̃ hα h β = 4π
β
rS ϕ (2.49)
0 0 3
which states that the integral of the square modulus of the wave function has a con-
stant value. Because of the definition of the chemical potential term, the volume
√
element for the background metric d3 x − g̃ has been used.
The fact that this magnitude is constant automatically ensures a constant be-
haviour for the probability density of the wave function defined in (2.6), as hα β h β α =
hαβ hαβ . Then, we are able to compute the total number of gravitons of the condensate
as Z
d3 x
p
N= g̃ ρĥ (2.50)
where the density has been introduced in equation (2.6). Assuming the "maximum
packaging" condition λ = rS explained in [25] and retrieving the missing constants,
we can relate the quantity in (2.49) to the integral of the density ρĥ . Then, with the
arguments of section 2.2, the number of particles (2.50) is expressed as
s
4π 3 √
N= M P 2 ϕ2 r S 2 =⇒ rS = N `P . (2.51)
3 4π ϕ2
These relations agrees nicely with the proposal of Dvali and Gómez. The rest of
relations of their work can basically be derived from this.
Possibly our more striking results are that the dimensionless chemical potential
µ0 = µ(r )r2 stays constant and non-zero throughout the interior of the black hole,
and that so does the quantity hα α = ϕ previously defined and entirely determined
by the value of µ0 , and viceversa. Therefore, it is totally natural to interpret µ0 as the
variable conjugate to N, the number of gravitons.
Of course the solution ϕ = 0 is also possible. If µ0 = 0 the only solution is ϕ =
0, N = 0 reproducing the familiar Schwarzschild metric in the black hole interior.
However, the interesting point is that solutions with ϕ 6= 0 exist and the value of
µ0 is uniquely given by (2.46). Assuming continuity of the solution the limit ϕ → 0
would correspond to µ0 = −1. This dimensionless number would then be a unique
property of a Schwarzschild black hole.
As seen above the dimensionless chemical potential has a rather peculiar be-
haviour. As µ0 = µ r2 is a constant function, then µ ∝ 1/r2 it is not null inside
and over the event horizon. Outside it appears to be exactly zero. Let us now for a
moment forget about the geometrical interpretation of black hole physics and let us
treat the problem as a collective many body phenomenon. It is clear why gravitons
are trapped behind the horizon: the jump of the chemical potential at r = rS would
prevent the particles inside from reaching infinity. From this point of view it is quite
natural to have a lower chemical potential inside the horizon than outside (where is
obviously zero) as otherwise the configuration would be thermodynamically unsta-
ble. In the present solution particles (gravitons in our case) cannot escape.
However this is not completely true as the picture itself suggests that one of
the modes can escape at a time without paying any energy penalty if the maximum
packaging condition is verified. Let us do a semiclassical calculation inspired by this
√
picture; using M ∼ MP N:
dM M dN 1 dN
' √P = . (2.52)
dt 2 N dt 2rS dt
To estimate dN/dt (which is negative) we can use geometrical arguments to deter-

mine the flux. If we assume that for a given value of rS only one mode can get
out (as hypothesized above) and that propagation takes place at the speed of light,
elementary considerations6 lead to
dM 3 1
'− . (2.53)
dt 2 rS 2
This agrees with the results of [25] –for instance Eq. (35)– and yields T ' 1/rS .
Within this picture, several questions concerning the long-standing issue of loss of
information may arise as the outcoming state looks thermal [9] but apparently is not;
or at least not totally so. However we shall refrain of dwelling on this any further at
this point.
The main result from the previous rather detailed analysis is that the black hole
is able to sustain a graviton Bose–Einstein condensate (and surely similar conden-
sate’s made of other quanta). But is that condensate really present? Our results
do not answer that question, but if we reflect on the case where the limit N → 0
is taken, without disturbing the black hole geometry (i.e. keeping rS constant) this
√
requires taking ϕ → 0 in a way that the ratio N/ϕ is fixed. Then one gets a di-
mensionless µ0 = X = −1. This value appears to be universal and independent of
any hypothesis. The metric is 100% Schwarzschild everywhere. We conclude that a
black hole produces necessarily a (trapping) non-zero chemical potential when the
physical system is expressed in terms of the grand-canonical ensemble. Outside the
black hole, µ = 0 (X = 0).
Another way of reaching this conclusion is by taking a closer look to our exact
equations in the previous section. If one sets µ = 0 then necessarily ϕ = 0 and one
gets the classical Schwarzschild solution everywhere, but the reverse is not true. One
6 To determine the rate of variation of N we have to multiply the flux times the surface 4πrS2 ,
dN 3
= ΦG A = − .
dt rS
The flux is negative and can be computed as the density of the mode times the velocity, where we have
assumed that c = 1 in our units. Since the density of the mode is constant in the interior, it is just
3/4πrS3 .
2.5. Consequences of the results and connection with previous proposals. 41
can have ϕ = 0 but this does not imply µ = 0. Let us emphasize that these results
go beyond the second order perturbative expansion used in parts of this article.
As gravitons cross the horizon and are trapped by the black hole classical gravita-
tional field they eventually thermalize and form a BEC. The eventual energy surplus
generated in this process is used to increase the mass and therefore the Schwarzschild
√
radius. ϕ is now non-zero; it is directly proportional to N, the number of gravitons,
and the dimensionless chemical potential departs from the value µ0 = X = −1, pre-
sumably increasing it in modulo. As soon as ϕ 6= 0 the metric inside the black hole is
not anymore Schwarzschild (but continues to be Schwarzschild outside). Note that
the metric is destabilized and becomes singular for ϕ = 1, so surely this is an upper
limit where N → ∞.
Yet another possible interpretation, that we disfavour, could be the following.
Each black hole has associated a given constant value of µ0 = X, hence of ϕ. Then
equations (2.51) would imply that after the emission of each graviton, the value of rS
is readjusted. The problem with this interpretation is that it would require a new di-
mensionless magnitude (µ0 or X) to characterize a Schwarzschild black hole; some-
thing that most black holes practitioners would probably find hard to accept. In
either case the metric in the black hole interior is different from the Schwarzschild
one.
43
Chapter 3
Quasilocal energy for condensate

structures
3.1 Quasilocal energy definitions in general relativity

It is a fundamental tenet of general relativity that there is no such concept as lo-
cal energy of the gravitational field. This can be understood as a consequence of
the equivalence principle, implying that there is no proper definition for the energy
density in gravitating systems. One of the greatest achievements has certainly been
the possibility of defining a global gravitational energy and momentum, both at spa-
tial and null infinity (Arnowitt–Deser–Misner, in short ADM, or Bondi masses [71]
between other formulations).
More ambitious efforts to associate an energy to extended, but finite, spacetime
domains have resulted in quasilocal definitions. Initial attempts in this direction in-
volved pseudotensor methods, where part of the "gravitational energy" stands as a
source term in the form of a stress energy momentum. However this separation is
not true for all observers, and there is no general definition for obtaining it. Ein-
stein was the first one to derive a stress tensor from a variational principle. From a
perspective of decades of research in this field, his definition, while not being dif-
feomorphism invariant, was quite successful. In the 60s, Møller discovered a bulk
action, instead of being quadratic on the Christoffel symbols, Γσµν , as it was in the
ΓΓ−action proposed by Einstein [38], it was quadratic on the tetrad connection co-
αβ
efficients, ωµ ; the ωω −action [72]. This proposal gained the property of being dif-
feomorphism invariant; however it was not fully invariant under "internal" transfor-
mations of the tetrad and it was not constructed purely with the metric as Einstein’s
formulation. A major step forward was made when it was possible to generalize
the quasilocal treatment into the language of two-spinors (see for instance the works
of Szabados [73]). From this powerful formalism numerous interesting results have
been obtained. Although several approaches have led to some interesting results,
no definitive consensus for the expression of the quasilocal energy seems to have
emerged (the most important proposals are reviewed in [74] and references therein).
In the present study we will make use of the canonical quasilocal formalism
based in the so-called Trace K −action [75]. Quasilocal properties can be derived
44 Chapter 3. Quasilocal energy for condensate structures
by means of a field–theoretic generalization of the Hamilton–Jacobi theory, without

inquiring into the detailed structure of the quasilocal (Hamiltonian) phase space.
In fact, it provides a formal basis for deriving from the action a stress energy mo-
mentum tensor with a purely metric formalism (as the ΓΓ−action), differing from
the standard Hilbert action by metric dependent boundary terms. It is worth noting
that (contrary to the ΓΓ−action) the TraceK −action is manifestly invariant under dif-
feomorphism and independent on any background structures1 . Besides, although it
is not clear how to apply the Noether theorem as this action is not constructed from
a bulk Lagrangian, it is possible to avoid the use of this theorem by means of the
Hamilton–Jacobi theory.
In what follows we will give the basic ingredients of this formalism. Then, in sec-
tion 3.3, we will review how to define a consistent quasilocal energy for spacetime
exhibiting horizons. In the remaining sections, following [30], first we will analyze
the consequences of applying this formalism to the zone beyond the horizon. Finally,
it will be shown that with the quasilocal energy, one can associate also a quasilocal
binding-like potential. In this manner, a consistent picture can be obtained to con-
strain the behaviour of a graviton condensate structure gravitationally.
3.2 Hamilton–Jacobi Formalism

There is no specific construction for the Hamilton–Jacobi equations. The starting
point is to define a Hamiltonian making use of the action, which is well known in
gravity. The mathematical formulation of this idea is to take a compact region of
the spacetime M. Time is globally defined inside this region and provides a smooth
foliation into a family of spacelike slices Σ. This foliation admits for the metric to
introduce the usual ADM decomposition [76],
gµν dx µ dx ν = − N 2 dt2 + γij (dxi + V i dt)(dx j + V j dt) . (3.1)
N is the lapse function and V i the shift vector. In the static and spherically symmetric
cases under consideration in section 2, this metric gets the simpler structure; for
convenience it will be written as
ds2 = −eN 2 dt2 + e f −2 dr2 + r2 dΩ2 ; (3.2)
with such definition N (r ) and f (r ) are always positive functions of the coordinate r
and e is either +1 or −1. If we are placed outside the horizon, t is a timelike coor-
dinate, then the temporal metric potential gtt is negative and in this region e = +1.
At the horizon, gtt and grr exchange sign, hence in the interior e = −1. Even more,
when a condensate is present the inverse relation between gtt and grr is broken.
1 Any
action admits subtraction terms, function of the fixed boundary data. In some cases, as the
one we will deal with, it is convenient to fix this ambiguity by introducing a background spacetime;
however, this is not a required feature of the canonical quasilocal formalism.
3.2. Hamilton–Jacobi Formalism 45
In this manner the decomposition M = [t0 ; t00 ] × Σ can be thought of as if it were

a family of hypersurfaces defined by a timelike and future pointing unit normal
uµ = −eN δµt . The metric tensor induced on each leaf Σt is
γµν = gµν + uµ uν . (3.3)
Hence γµν works as a projector tensor onto Σ. For space coordinates xi adapted to
the foliation Σ : i = (r, θ, φ), we can define a spacial metric γij , such that contracted
j j
with its inverse gives the Kronecker delta γi = δi .
Also curvature tensors for the intrinsic curvature Rijkl and the extrinsic curvature
Θij can be defined. It is interesting for us the latter one, which is defined as usual
1
Θij = − Lu γij . (3.4)
2
Each generic leaf Σ has a spatially closed two-boundary ∂Σ = B. The time

history of these two-surface boundaries B is the timelike three-surface boundary
3B = [t0 ; t00 ] × B, characterized by a spacelike normal nµ pointing outwards as it is
depicted in Figure 3.1. An overlined notation will be used for tensors living in this
submanifold, let us note that the boundary B and its history 3 B need not be physical
barriers. The induced metric is
γµν = gµν − nµ nν , (3.5)
γνµ projects onto 3 B. Analogously, intrinsic coordinates i = (t, θ, φ) defines tensors

j j
on 3 B: an intrinsic metric γij , the projector becomes the delta γi = δi , an extrinsic
curvature
1
Θij = − Ln γij , (3.6)
2
and so on and so forth.
As we will see, when Σ is taken to intersect 3 B orthogonally, the Brown–York
quasilocal energy arises from the boundary term in the action. This orthogonality
F IGURE 3.1: A generic spacelike slice Σ has two-boundary B. The spacetime

M has as boundaries the time history of B, a timelike three-surface 3 B, and the
initial and final spacelike hypersurfaces defined at t0 and at t00 .
implies for the three-boundary normal nµ to satisfy (u · n)|3 B = 0. Therefore, the

unit normal nµ in the spacetime region to the three-boundary 3 B ’coincides’ with the
outward unit normal in Σ to the two-boundary B. The normal vector to constant r
surfaces is
ni = e f δri ; (3.7)
with unit length in the hypersurface three-metric, i.e. n j γij n j |3 B = 1. This way of
defining objects, factorizing the signs via e, makes the extension of the definition
of the quasilocal energy to the interior zone, beyond the event horizon, straightfor-
ward. Details can be found in [77], where it is argued on physical grounds why
normal vectors change sign at r = rS . Basically they must keep pointing outwards
and to the future.
Let now N and N a be the lapse and the shift, this orthogonal restriction allows
the metric on 3 B to be decomposed as
γij dxi dx j = − N 2 dt2 + σab (dx a + V a dt)(dx b + V b dt) . (3.8)
Now a, b = θ, φ are the coordinates on B and σab is the two-metric. These two-
boundaries B can be thought as the intersection of the family of leaves Σ with the
three-boundary 3 B as shown in Figure 3.1. The metric can alternatively be written as
σµν = γµν − nµ nν = γµν + uµ uν = gµν − nµ nν + uµ uν . (3.9)
In the case of spherically symmetric and static spacetimes from the metric decom-
position (3.8) it is seen that the induced two-metric of B has clearly the structure for
a sphere; i.e. ds2 = r2 dΩ2 . If the connection is metric compatible as in the cases of
interest, the usual definition for the extrinsic curvature of being proportional to the
Lie derivative L along the corresponding unit normal (3.7), an analogous definitions
as the ones written in (3.4) or (3.6), in this case for the two-boundary B as embedded
in Σ, can be equivalently written as the covariant derivative in the spacelike slice of
the normalized tangent vector to this hypersurface as defined on Σ
Kij = −σik ∇k n j (3.10)
j
along a direction of the projector operator σi constructed with (3.9).
The purpose of this decomposition is to establish an analogy with this method
deduced in the context of nonrelativistic mechanics. To make the generalization
more tractable, let us summarized the notation in Table 3.1. Also some technical
expressions will be included in the Appendix C.
Unit Extrinsic Conjugate

Metric
Normal curvature momentum
Spacetime M gµν
Hypersurfaces Σ γij uµ Θij Pij
Three-boundary 3 B γij nµ Θij π ij
Two-boundary B σab ni Kab
TABLE 3.1: Summary of notation for the spacetime decomposition.
3.2.1 Nonrelativistic mechanics review
Given a classical mechanical system of n degrees of freedom with configuration

manifold Q and a Lagrangian L : TQ × R → R that is assumed to be first order
and may depend on time explicitly. In this mathematical formalism, given two con-
figurations (q0a , t0 ) and (q00a , t00 ), the corresponding action functional is
Z t00 Z t00
. .
S (nr )
[q(t)] =
t0
L [q a (t); q a (t); t] dt =
t0
∑ q a p a − H [q a , p a , t] dt ; (3.11)
a
where q a (t) are the generalized coordinates of the state space (phase space + time)2 ,
.
from q0a (t0 ) to q00a (t00 ), with tangent q a (t) at the parametrization parameter t. A Leg-
endre transformation of the Lagrangian provides the action in canonical form and
gives the usual definition for the Hamiltonian of the system in terms of the general-
ized coordinates q a and the generalized canonical momenta p a . It is expected that a
generalization of this theory provides a formal basis for identifying from the action
a stress-energy-momentum tensor for general relativity.
Varying this action with the two fixed points, one gets
t00 t00
equations of motion
(nr )
= + ∑ p a δq a 0 − H δt 0 . (3.12)

δS
terms a t t
Variations among classical solutions extremize S(nr) ; the terms giving the equations
of motion vanish. The requirement that the action has null variation leads to the
Hamilton–Jacobi equations
!
∂S ∂L ∂S
pa = = . ; H=− . (3.13)
∂q a ∂qa ∂t
Clearly, the action is not unique. Subtraction terms S(nr) [q(t)] − S0 [q(t)] of the form
R t00
S0 [q(t)] = t0 (dh/dt) dt, with any arbitrary smooth function h = h(q a (t), t) is an
equally good action for obtaining the same dynamics. Obviously, the subtraction
term S0 [q(t)] modifies both magnitudes, the canonical momenta and the energy ac-
cording to p a → p a − (∂h/∂q a ) and E → E + (∂h/∂t), respectively.
2 The pairs (q a , t) may be called a history or world line in the ‘spacetime’ Q × R for proceeding to a
relativistic formulation.
There is therefore an intimate connection between the classical Hamiltonian (the

energy), the change in time of the classical action and the fact that choice of the action
is not unique.
3.2.2 Covariant Hamiltonian formalism
The guiding principle for obtaining the quasilocal energy is the analogy between
the action variation in nonrelativistic mechanics and variation in general relativity.
Now that we have defined the tensors needed at the beginning of section 3.2 and in
the Appendix C, let us move to generalize the Hamilton–Jacobi method to general
relativity.
In order for the action to provide a well-defined variational principle for Einstein
theory, one fixes the intrinsic three-metrics over the boundaries (rather than all the
gµν ) [75]. Then we have a bulk Hilbert action over M, with boundary three-metric γij
fixed on 3 B and the hypersurface metric γij fixed on the end points of the world-lines
t0 and t00 , on where to apply variational principles
1
Z
d4 x
p
S= − g R + Sm
2κ M
t00 1 Z (3.14)
1 √ 1
Z Z
d3 x γ Θ 0 − d3 x − γ Θ + d3 x − γ Θ .
p p
+

κ Σ t κ 3B κ 3B
Sm may include the matter action as well as the chemical potential or the cosmolog-
ical constant terms when corresponds. The last term is the Gibbons–Hawking nor-
malizing factor [78]; Θ corresponds to the extrinsic curvature of the three-boundary
but as embedded in a flat background four-geometry (the spacetime in which we
want to obtain zero quasilocal quantities). This term is usually interpreted as our
freedom to shift the zero point of the energy.
Varying the action with respect to arbitrary variations in the metric and matter
fields, we get
t f Z
equations of motion
Z
3 ij
d3 x (π − π0 )ij δγij . (3.15)

δS = + d x P δγij +
terms Σ ti
3B
The Hilbert action minimally coupled to matter is stationary (δS = 0) if Einstein’s

equations are satisfied. When restricted to classical solutions, the equations of mo-
tion vanish and we are left with the terms given by the two momenta: when matter
fields are minimally coupled to gravity, the gravitational momentum canonically
conjugate to γij for the hypersurfaces Σ is defined as
1 √
Pij = γ Θ γij − Θij ; (3.16)
2κ
and the boundary momentum π ij conjugate to γij , as
1 p ij

π ij = − −γ Θ γ ij − Θ . (3.17)
2κ
ij
The π0 term is conjugate to the flat metric and is simply a shift due to the Gibbons–
Hawking normalizing factor; fixing this term means to choose a ‘reference config-
uration’. The analogues of the Hamilton–Jacobi equations (3.13) are the following
two relations (the equations of the nonrelativistic case are recalled on the right for
making the correspondence straightforward)
 
δS
Pij = ∂S

pa =

 



 δγ ij


 ∂q a
⇐⇒ (3.18)

 ij δS 
 ∂S
 (π − π0 ) = δγ H = −
 

  .
ij ∂t
The generalization is quite direct. The three-metric γij provides the metrical dis-
tance between all spacetime intervals in the boundary manifold 3 B (including time
between spacelike surfaces); therefore the notion of energy (the one equation for the
Hamiltonian) in nonrelativistic mechanics is generalized to a stress energy momen-
tum defined on 3 B that characterizes the entire system (gravitational field, chemical
potential fields and any matter fields and/or cosmological constant). In accordance
with the standard definition for the matter stress tensor
2 δSm
T µν = √ , (3.19)
− g δgµν
the following surface stress tensor is conveniently defined
2 δS 2 ij ij

τ ij ≡ √ =√ π − π0 . (3.20)
−γ δγij −γ
An important feature of this stress tensor is apparent when considering two con-
centric spherical surfaces B1 and B2 . In the limit where B1 and B2 approach each
other, the total surface stress energy momentum result in
2 ij ij

τ ij = √ π2 − π1 (3.21)
−γ
since τ1 ij is computed using the outward normal to B2 and reference terms π0 ij cancel
between each other. This tensor embodies the well-known result of Lanczos and
Israel in general relativity [79], which relates the jump in the momentum π ij to the
matter stress tensor of the surface layer. In the infinitesimally thin layer limit, the
three-geometries of each ‘side’ of the layer coincide and there is no gravitational
contribution to τ ij . The direct physical implication of this result is the widely known
absence of a local gravitational energy momentum [80].
3.3 Brown–York quasilocal energy

Let us give one step further and rederive the quasilocal energy elaborated by Brown
and York [81] in order to analyze its behaviour for several spacetime configurations
where horizons are present, with or without the possible graviton condensates just
discussed in Chapter 2. This magnitude is one of the promising candidates for ob-
taining the quasilocal energy in general relativity. According to this proposal, a
suitable fixing of the metric over the boundaries [75] provides a well-defined ac-
tion principle for gravity and matter, and dictates a natural choice for the defini-
tion of the quasilocal energy contained in the region of interest without the need
for any other geometric structures. Besides, its properties can be obtained through
a Hamilton–Jacobi type of analysis involving the canonical action as it has been re-
cently describes in the previous section. Among these properties we could mention
that the proposal of [81] agrees with the Newtonian limit for spherical stars, it is
applicable to thermodynamic problems [82, 83] and the asymptotic limit at infinity
is the Arnowitt–Deser–Misner expression [76] for the energy in asymptotically flat
spacetimes [84].
Physically, the Brown and York proposal [85] is meant to represent the total en-
ergy enclosed by a space-like surface. This information is encoded in the surface ten-
sor τ ij defined on (3.20). As it has been stated in (3.1), the 3+1 decomposition of the
spacetime, together with the orthogonal condition for the normal vectors uµ nµ = 0,
yields the 2+1 decomposition (3.8) of the metric γij on 3 B as well. Hence the nor-
mal and tangential projections of τ ij on the two-surface B defines the proper energy,
momentum, and spacial-stress surface densities, given by
ε = ui u j τ ij
ja = −σai u j τ ij (3.22)
s ab = σia σjb τ ij
respectively.
We are interested in the first magnitude; τ ij is proportional to the boundary mo-
mentum (3.17). Written as in (C.8), it is immediate to see that
√
2 ij 2 N σ K
√ ui u j π = √ K= . (3.23)
−γ N σ 2κ κ
At this stage, as the action is not unique and arbitrary functions of the fixed bound-
ary data can be included without modifying the equations of motion, the reference
ij
tensor π0 contributes with a trace curvature term K0 as embedded in Minkowski.
Integrating the energy density along a slice defining a two-boundary B, leads us
3.3. Brown–York quasilocal energy 51
directly to the definition of the quasilocal energy

Z
2
√ 1
Z √
E= d x σε= d2 x σ ( K − K0 ) . (3.24)
B κ B
The energy result in the subtraction of the total mean curvature of B es embedded
in Σ with the total mean curvature of B es embedded in a Minkowskian reference
frame, times the inverse of the Einstein’s gravitational constant.
Notice that the separation of the four dimensional space in a product M = Σ × t
implies that the quasilocal energy is invariant under general coordinate transfor-
mations on Σ but not under all four dimensional general coordinate transforma-
tions, which also involve time. Therefore quasilocal energy will be dependent on
the proper time of the observer [86].
Let us now focus this definition over arbitrary static and spherically symmetric
spacetimes generically written as in (3.2). For the calculation it is only needed the
following connection coefficients
f0 1
Γrrr = − Γrθ = Γrφ =
θ φ
; (3.25)
f r
the prime denotes r derivatives. The trace of the extrinsic curvature needed for com-
puting the quasilocal energy is
f (r )
K = −2 e (3.26)
r
and the subtraction term is obtain embedding the sphere B in a Minkowskian space-
time by setting f (r ) = e = 1, this is
2
K0 = − . (3.27)
r
The next step is straightforward, putting everything together in (3.24) with the de-
√
terminant of the induced two-metric given by σ = r2 sin2 θ, the quasilocal energy
is
r
E= 1 − e f (r ) . (3.28)
GN
Apart from the intrinsic interest of defining an energy, using a least-energy prin-
ciple could allow us to determine whether a graviton condensate is in some sense
preferred to the standard no-condensate situation. One might after all expect that
matter falling into a black hole will eventually thermalize and the formation of a
BEC may not seem such an exotic possibility then, but in the case of gravitons the
formation of the condensate is a lot less intuitive. Minimization of the quasilocal
energy may be a convenient tool to answer whether a BEC for gravitons exists or
not.
3.4 Quasilocal energy for a graviton condensate black hole

Let us now consider the change in the quasilocal energy, as defined above, when a
condensate with spherical symmetry, such as the one previously described, exists
in a static and spherically symmetric black hole. The significant metric elements in
(3.2), both within and beyond the event horizon, are obtained particularizing (2.43)
to the Schwarzschild solution
s s
e rS rS
N (r ) = 1− ; f (r ) = e 1−ϕ 1− . (3.29)
1−ϕ r r
In the outer region we get the known solution by setting e = 1 and ϕ = 0. In the
interior r is a timelike coordinate, e = −1 and ϕ may be different from zero. The
following expression describes the quasilocal energy (3.28) in both cases
 s 

r  rS 
E= 1−e e 1−ϕ 1− . (3.30)
GN r
We have written the expressions above in such a way that taking ϕ = 0 and
e = 1 for r > rS we get the quasilocal energy of the black hole as a whole as seen
from outside3 . The quasilocal energy inside an r = constant surface placed outside
the event horizon is given by
r !
r rS
E= 1− 1− . (3.31)
GN r
Assuming the existence of a BEC with ϕ 6= 0 and that inside the horizon the r coor-
dinate becomes timelike with e = −1
r !
r p rS
E= 1+ 1−ϕ −1 . (3.32)
GN r
These expressions were derived in [85] in the case ϕ = 0.

The quasilocal energy of the entire Schwarzschild metric is plotted in Figure 3.2.
This shows that a condensate structure lowers the energy with respect the familiar
Schwarzschild metric. The higher the value for ϕ < 1 chosen, the lower the quasilo-
cal energy will be. Figure 3.2 tells us that the total energy inside the event horizon
(including gravitational energy as well as the energy of the r = 0 singularity) is twice
the black hole mass
E(r = rS ) = 2M (3.33)
and twice the energy of the total spacetime as can be seen from the asymptotic be-
haviour for the quasilocal energy at infinity. An expansion of (3.31) around infinity
3 That is from the reference system of an observer at rest at r = ∞.
3.4. Quasilocal energy for a graviton condensate black hole 53
2.5
E(r)
Quasilocal Energy 2.0
1.5
1.0
0.5
0.0
0 1 2 3 4 5 6
r
F IGURE 3.2: Quasilocal energy for a Schwarzschild black hole computed in-
side as well as outside the event horizon. Both axes are in units of the mass
M, GN = 1. The event horizon is located at 2M. In the inner region we com-
pare a black hole in classical theory of general relativity where ϕ = 0 (solid
line), and various values for the condensate parameter where we understand
that a BEC structure is preferred minimizing the resulting energy of the object.
We use as a comparison two values for the condensate: ϕ = 0.4 dashed line
and ϕ = 0.7 dotted line.
yields  
" #
r  rS 1 r
lim E(r ) ' 1− 1− +O 2 = S ≡M, (3.34)
r →∞ GN  2r r  2GN
which is certainly an expected result.

The reinterpretation of a black hole as describing a graviton BEC does not modify
three remarkable features of the quasilocal energy presented in [77] that are perhaps
not widely acknowledged. The first one is that the quasilocal energy is zero at the
singularity. The infinity of the energy of the gravitational field at r = 0 predicted
by Newtonian gravity is removed. The second striking fact comes from the energy
distribution inside the black hole. Inside the horizon, the quasilocal energy has its
maximum value at !
1
rmax = GN M 1+ p . (3.35)
2−ϕ
In the limit of the classical theory, this maximum value is located at a radius equals to
√
(1 + 1/ 2) GN M. As the condensate parameter ϕ increases, the maximum moves
towards the event horizon and for the limiting value ϕ = 1, it lies exactly at the
horizon. All this means that the black hole looks like an extended object where most
of the energy seems to be ’stored’ not far from the horizon. The third feature is that
the derivative of the quasilocal energy matches across the horizon, but is infinite
there regardless of the value of ϕ.
Note that expanding the quasilocal energy and subtracting the black hole mass
M one gets for rS r
GN M2
E (r ) − M = +... (3.36)
2r
The leading term is the classical gravitational self-energy of a shell of matter with
total mass M and radius r.
3.5 Gravitational binding

For r rS , if we denote by E(r )| M the quasilocal energy of a black hole of mass M,
Mm 3 M2 m
E(r )| M+m − E(r )| M ' m + GN + GN 2 2 + . . .
r 2 r
(3.37)
∂
= m E(r ) + O(m2 ) .
∂M
The first term corresponds to the gravitational potential energy of masses m and M
separated by a distance r. It is also the energy (for large values of r) that it takes to
separate the small mass m from M and take it to infinity. This leads us to interpret
the quantity
r !
∂ r 2GN M 1
V (r ) = 1 − 1− = 1− q (3.38)
∂M GN r 1− 2GN M
r
as the binding potential energy associated to the quasilocal energy outside the hori-
zon. Now, if we take one step forward and extend these ideas to the black hole
interior, we will get
r ! p
∂ r p 2GN M 1−ϕ
V (r ) = 1 − 1−ϕ −1 = 1− q (3.39)
∂M GN r 2GN M
−1
r
as the analogous potential energy inside the horizon. As occurs with the quasilocal
energy, there are different ways to derive a quasilocal potential as well (both inside
and outside black holes). For instance, in [87] the authors define a quasilocal poten-
tial coming from the Brown–York quasilocal energy that does not coincide with our
definition. They compare their definition with the Misner–Sharp potential, derived
from its respective quasilocal energy too [88].
We note that V (r ) is singular at r = rS , but we also note that this singularity is
integrable. The following limits are readily found:
lim V (r ) = 1 , lim V (r ) = 0 . (3.40)

r →0 r →∞
A plot of the profile is presented in Figure 3.3. There we can see the potential that
seems to emerge from the Brown–York prescription for the quasilocal energy in the
3.5. Gravitational binding 55
0
Binding Potential V(r)
-1
-2
-3
-4
0 1 2 3 4 5 6
r
F IGURE 3.3: Potential profile derived from Brown–York energy for

Schwarzschild spacetime, with M = 1 and various values for the conden-
sate parameter ϕ. The solid line corresponds to ϕ = 0. As ϕ approaches the
limiting value ϕ = 1 the inner part of the potential tends to a square potential.
case where the background metric is Schwarzschild. As we observe, the binding

takes place in a region that is relatively close to the horizon, which is not at all unex-
pected after having seen the shape of the black hole quasilocal energy profile.
There are several considerations that may follow from this potential profile if we
accept the previous interpretation. Notice that a massive particle of mass m initially
at rest at infinity follows the trajectory described by (see e.g. [89])
2
.2 rS rS
r = 1− , (3.41)
r r
where the dot means derivative with respect the coordinate time t (i.e. not proper
.
time). The radial momentum will be pr = mγr. As the particle approaches r = rS
from the outside, pr tends to zero as it is well known. If we take Figure 3.3 at face
value the particle falls in the potential valley while emitting gravitational radiation.
However, quantum mechanics should stabilize it at some point. Let us use a simple
argument to make some energy considerations (a similar argument works fine in the
case of the hydrogen atom).
Indeed, assuming that ∆pr ∼ pr , according to the uncertainty principle ∆r >
1/pr . This leads to the following order of magnitude estimate:
r
rS
∆r ' . (3.42)
m
This gives us the characteristic width of the potential and the approximate energy
level. Introducing the relevant units, for a black hole of about 30 solar masses and
considering the fall of an electron, this gives ∆r ' 1mm. This result is surprising as it
would imply that falling matter accumulates in a very thin shell on both sides of the
√
horizon. The associated potential level would approximately be ε/m ' 1 − rS m.
Figure 3.4 may help us to understand the situation. In this figure we plot the ve-
locity of a particle falling in a black hole of mass M = 1 following a radial geodesic,
as seen from an observer at infinity. We also plot the Newtonian potential that such
an observer would deduce from measuring the motion of the falling particle along its
geodesic motion. This (fictitious) potential agrees for the most part of the trajectory
with the one derived from the quasilocal energy, which however is much deeper
close to r = rS . The difference between the two potentials should be observed by
the distant reference point as gravitational wave emission. The total energy emit-
√
ted should be approximately m rS m m. Obviously, only part of this energy is
emitted outside the black hole and eventually reaches a remote observer. It is to be
expected that a good part of it remains inside the black hole horizon. This is a natu-
ral explanation for the graviton condensate, present when ϕ 6= 0, whose properties
have been discussed in the previous sections.
Assuming for the sake of the discussion that (a) gravitons in the condensate are in
a state close to the maximum packing condition, i.e. have an energy ∼ 1/rS and (b)
that all the gravitational energy emitted by the falling particle is eventually stored
in the black hole interior, the capture by the black hole of a particle of mass m would
0.4
0.2
r (r)
•
V (r) ,
N
0.0
V(r) ,
-0.2
-0.4
0 5 10 15 20 25 30
r
F IGURE 3.4: Evolution of the velocity of a falling

√ object as seen from a distant
observer (blue curve). The maximum is 2c/(3 3) and is attained at r = 3 rS .
We also plot the (fictitious) Newtonian potential that such an observer would
‘measure’ from the fall of the object (red curve). In the same plot we show a
zoom of the quasilocal potential profile derived from Brown–York energy for
Schwarzschild spacetime, with M = 1 presented in Figure 3.3. In the interior
we plot the evolution of this magnitude as ϕ gets close to its limiting value
ϕ = 1 and the potential approaches a step function.
3.5. Gravitational binding 57
3
create n = (m rS ) 2 gravitons. This is a fairly large number so it is not unexpected
that the continuous capture of matter by a black hole brings the value of the graviton
condensate close to 1, its limiting value, N ∼ ( M/MP )2 . At that point, most of
the mass of the black hole is actually due to the graviton condensate: gravitational
(quasi)local energy has turned into black hole mass. This simple argument gives
strong plausibility to the thesis sustained by Dvali and Gómez.
Certainly, along the process just described, M and consequently the shape of the
potential does not stay constant, so a detailed dynamical analysis would be required
to understand this process more accurately. Of course all the previous considerations
hinge on the validity of the hypothesis behind the quasilocal energy principle.
It is actually interesting to see what happens when the value of the graviton BEC
approaches the limiting value ϕ = 1 beyond which the metric becomes intrinsically
singular. The quasilocal potential becomes a constant (the quasilocal energy is a
linearly rising function of r in the black hole interior in this limit). The result would
apparently be a barrier right at the horizon location for falling particles. In this limit
it would seem that gravitationally trapped objects would adhere to the black hole
external surface.
59
Chapter 4
Condensates beyond horizons
4.1 Bose–Einstein condensate extensions

In Chapter 2, we have derived a collective solution representing a bosonic conden-
sate constructed with gravitons on top of a confining classical Schwarzschild poten-
tial. The ground state of the system is dictated by a gravitational Gross–Pitaevskii-
like equation and is strongly background dependent. The characteristics of the solu-
tion present some interesting similarities with the usually employed approximated
schemes such as Hartree–Fock. A main characteristic is that the total wave-function
of the system of N bosons is taken as a product of single-particle functions in the
same quantum state. When this is so, and the single-particle wave function satisfies
the Gross–Pitaevskii equation, the total wave-function minimizes the expectation
value of the Hamiltonian under a normalization condition giving the total number
of constituents N. In the gravitational counterpart this aspect is translated into the
maximum packaging condition for the gravitons. As shown in Chapter 3 the energy
is minimized for the condensate structure and is preferred over vacuum solutions.
In the present chapter we will give the first two simplest generalizations to other
self-bounded classical field configurations: Reissner–Nordström and de Sitter back-
ground metrics. We have found that condensate-like solutions may be present in
all these cases too, and its structure is strongly dependent on the trapping geometry.
Here again, its constituents are regulated by some sort of chemical potential. We then
proceed to study the energy associated to these solutions and we found that they are
in fact energetically favourable. In the nonrelativistic case the mean-field theories
provide a correct qualitative picture of the phase transition. Analogous implications
are expected here and it is plausible to interpret the different static gravitational po-
tential with our effective quantum description.
4.2 Bose–Einstein condensate in Reissner–Nordström

In the present section, we shall extend the Bose–Einstein condensate formalism to
a Reissner–Nordström geometry [90]. When a black hole is coupled both to the
60 Chapter 4. Condensates beyond horizons
gravitational and to the electromagnetic fields, the geometry is given by

! ! −1
rS rQ 2 2 rS rQ 2
g̃αβ dx dx = − 1 −
α β
+ 2 dt + 1− + 2 dr2 + r2 dΩ2 . (4.1)
r r r r
This is a static solution to the Einstein–Maxwell field equations that describes an

spherically symmetric electrically charged body of mass M and charge Q 1 . In order
to introduce the electromagnetic interaction into the condensate formalism at the
level of the action, (2.8) should be modified by adding the following Lagrangian
density Z
∆Sem = d4 x
p
− g Lem ; (4.2)
where Lem = − Fαβ F αβ /4. The corresponding electromagnetic stress–energy tensor

is expressed in terms of the field strength Fαβ = ∂α A β − ∂ β Aα as usual
1
Tαβ = Fασ Fβ σ − Fρσ F ρσ gαβ . (4.3)
4
In the above metric rQ 2 = GN Q2 is a characteristic length. The tensor (4.3) describes

the electromagnetic field in the outer region and is constructed by means of the elec-
tric potential Aα = ( Q/r, 0, 0, 0).
Equivalent arguments as the one discussed in (2.3) leads us to the classical the-
ory in this zone. No perturbations nor chemical potential are present in our visible
region; therefore, the Einstein–Maxwell equations are the usual for describing the
electromagnetic field in curved spacetime all the way from r = ∞ up to the (outer)
event horizon located at 2

1
q
r+ = rS + 2
rS − 4 r Q 2 , (4.4)
2
where the metric becomes singular in these coordinates. There is another point
where the metric presents a singularity; this is

1
q
r− = rS − rS 2 − 4 r Q 2 . (4.5)
2
This is called the inner horizon. In the region between horizons r− < r < r+ the
metric elements g̃tt and g̃rr exchange signs.
Let us entertain the possible existence of a condensate beyond the outer event
horizon. In this region the chemical potential term should be considered, hence the
the weak gravitational field limit we have − gtt ' 1 + 2Φ N , with Φ N the Newtonian potential.
1 In
Then it is see that the electromagnetic field produces a gravitational repulsion by means of the term
+rQ 2 /r2 . This is not an effect of the electromagnetic interaction, since it forms part of the metric, and
hence, also acts on neutral particles.
2 As long as 2r
Q ≤ rS ; otherwise there can be no physical event horizon. Objects with a charge
greater than their mass can exist in nature, but they cannot collapse down to a black hole unless they
display a naked singularity.
4.2. Bose–Einstein condensate in Reissner–Nordström 61
set of equations to solve are
Gµν ( g̃ + h) = κTµν + Xµν , (4.6)
where Xµν is the chemical potential term defined in (2.40). We shall explore the ex-
istence of solutions of the type (2.43). For doing so, we redefine the electromagnetic
tensor of the charged black hole in (4.3) with a new unknown factor
Fµν → C Fµν (4.7)
(this can be understood as a renormalization of the charge Q).

It can be seen that a family of solutions describing a condensate exists. Indeed
from the angular equations of motion we get
Q2 2 tt rr Q
2
(1 − ϕ ) = − C g g , (4.8)
r4 r4
where the effective metric (g = g̃ + h) is used to raise and lower indices. Because
g̃ tt g̃ rr = −1 this equation defines the relation of the new constant C to the value of
the condensate
1
C2 = . (4.9)
1−ϕ
On the other hand, we have the sector where the chemical potential acts, given by the
(t, r )− sector. Both equations entail the same information and the resulting equation
is
Q2 ϕ 2 Q2
− 1 − ϕ 4 − 2 = µ 1 − ϕ ϕ + C2 gtt grr 4 . (4.10)
r r r
The value for C in (4.9) makes the terms with Q to cancel and this equation exactly
reproduces the one obtained in the Schwarzschild case in equation (2.45); hence the
chemical potential yields
1
µr2 = − . (4.11)
(1 − ϕ )2
As we have seen in previous cases, the value of the BEC is in principle arbitrary
0 < ϕ < 1 and its relation to the chemical potential remains unchanged with respect
to the Schwarzschild case.
All obvious limiting cases behave as expected. If we turn off the charge, this
is Q = 0, the field equations and their corresponding solutions reduces to the un-
charged Schwarzschild case. Also, if we make the chemical potential to vanish, the
condensate disappear and we are left with the classical Reissner–Nordström vacuum
solution. Maxwell equations in a curved vacuum are still satisfied in this solution
for any arbitrary Q
D ν Fµν = 0 ,
(4.12)
Fνρ,µ + Fρµ,ν + Fµν,ρ = 0 .
The electromagnetic energy ( E2 + B2 )/2 is scaled by the factor C2 = 1/(1 − ϕ) if

ϕ 6= 0. This is indeed equivalent to scaling the charge. Notice that the redefinition of
the charge does not change the background metric, nor, consequently, the location
of the two horizons.
Reissner–Nordström is not a particularly physically relevant solution as the exis-
tence of a macroscopic electrically charged black hole is very unlikely, so its interest
is mostly theoretical as it would imply an accumulation of charged matter. Taking
this into account one can possibly understand the modification of the charge im-
plied by the redefinition Q → CQ as the consequence of the fact that the condensate
density increases too. This redefinition makes Q to diverge when ϕ → 1.
4.2.1 Quasilocal energy of a Reissner–Nordström black hole
We shall use again the results of section 3. In this case, the radial metric element is
v !
u
r r 2
S Q
p u
f (r ) = 1 − ϕ t e 1 − + 2 . (4.13)
r r
In the outer horizon the same change of sign of e as in the Schwarzschild case takes
place. As we move inwards and cross the inner horizon the signature of the t and r
coordinates are exchanged again, and t recovers its timelike character.
Including the electromagnetic field into the theory will not change the definition
of the quasilocal energy. The quasilocal energy is meant to measure the gravitational
energy associated to a specific geometry, and so only the gravitational action is im-
portant. Of course, the addition of a new field changes the metric and this is the way
that adding the electromagnetism influences the quasilocal energy.
The quasilocal energy in this case is given by
 v 
u !
r  p u 2m Q  2
E (r ) = 1 − e 1 − ϕ t e 1 − + 2  , (4.14)
GN r r
with e = 1 when t is timelike and e = −1 when the temporal coordinate is spacelike.

A striking feature is that the energy becomes negative for r < Q2 /2M. This radius is
always inside the inner horizon. The value of the quasilocal energy at the singularity
is E(r = 0) = −| Q|. The singularity at r = 0 has the electric field of a point charge,
and so, using just classical electromagnetism, its self energy should diverge.
In Figure 4.1 the quasilocal energy for these type of black holes is presented.
As we see, in the regions r > r+ and r < r− the usual (no condensate) solution is
energetically preferred, while in the region between horizons the opposite behaviour
is taken place and a condensate is viable. In particular, inside the inner horizon there
should not be a condensate. We conclude that for this case the condensate has the
structure of a spherical shell. According to the proponents of the quasilocal energy,
this energy includes the contribution of all fields.
4.3. Bose–Einstein condensate in de Sitter 63
The conclusions that can be drawn are in line with those discussed in the previ-
ous section for a Schwarzschild black hole.
1.5
1.0
E(r)
Quasilocal Energy
0.5
0.0
-0.5
-1.0
0.0 0.5 1.0 1.5 2.0 2.5
r
F IGURE 4.1: Quasilocal energy of a Reissner–Nordström black hole, with a

charge given by Q2 = 0.8M2 . Both horizons are clearly identified by the di-
vergence of the derivative of the energy. Inside the external event horizon we
compare the classical theory where the wave function is null, ϕ = 0, with a
condensate structure with the same values for ϕ used in the Schwarzschild
case. As we can see from the plot, inside the Cauchy horizon (where the t and
r coordinates exchange signs again), the condensate structure is not favoured
any more; hence the theory dictates a ring condensate between both horizons.
Both axes are in units of the mass M of the black hole.
4.3 Bose–Einstein condensate in de Sitter

It is quite peculiar for reasons that we will discuss below that also in this case the
presence of an event horizon seems associated to a possible Bose–Einstein conden-
sate. Let us repeat the previous calculations. In a de Sitter universe a cosmological
term must be included to the Hilbert–Einstein action (1.6)
Z
2
d4 x
p
SdS = MP − g ( R − 2Λ) . (4.15)
Including also the chemical potential action (2.7), variational principles lead us to
the Einstein field equations with a cosmological constant and a chemical potential
term which are
Gµν = −Λgµν + Xµν . (4.16)
The following maximally symmetric solution

−1
Λ Λ

ds = − 1 − r2
2
dt + 1 − r2
2
dr2 + r2 dΩ2 (4.17)
3 3
is obtained in our part of the universe. The reasoning goes as in section 2.3, but
using r = 0 as the Minkowskian location where it is mandatory for the condensate
to vanish. As it vanishes at r = 0 it remains null up to the cosmological horizon,
which is located at r
3
rΛ = . (4.18)
Λ
However, when we go thorough this horizon, it is expected a condensate to be
present. Searching for solutions with a nonzero chemical potential, we redefine the
cosmological constant term with the help of a new constant C: Λ → C Λ, while the
metric is modified to
!
1 1
gµν = diag g̃tt ; g̃rr ; g̃θθ ; g̃φφ , (4.19)
1−ϕ 1−ϕ
as in the Schwarzschild and Reissner–Nordström cases.

Any of the two angular components of the Einstein’s equations fixes this constant
with respect to the condensate as
C = 1−ϕ. (4.20)
This election for C makes the remaining field equations to take the same form as
in (2.45); the cosmological contribution inside the Einstein tensor cancel out with
the modified cosmological term. As the equation is the same, the structure for the
chemical potential remain unchanged with respect to the Schwarzschild case (2.46).
That is
µ0
µ (r ) = . (4.21)
r2
In the visible part of our universe there is no condensate. This case is fairly obvious
as at r = 0 the metric is Minkowskian. At r = rΛ there is a discontinuity in the chem-
ical potential. There is a peculiar situation here: contrary to the Reissner–Nordström
case, where the electromagnetic field is enhanced by the condensate structure, the
cosmological constant term is screened by the condensate structure. The presence
of the graviton condensate decreases the value of the cosmological constant which
becomes zero for the limiting value ϕ = 1.
4.3.1 Quasilocal energy analysis over a de Sitter horizon
The fundamentals of the previous derivations of the quasilocal energy do not change
if a cosmological constant is considered. Now f (r ) in (3.2) is given by
s
Λ 2

p
f (r ) = 1−ϕ e 1− r , (4.22)
3
2.5
2.0
E(r)
Quasilocal Energy 1.5
1.0
0.5
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
r
F IGURE 4.2: Quasilocal energy for de Sitter spacetime. The axes are in units
of the cosmological horizon, i.e. rΛ = 1. Beyond the horizon we present the
curves for the classical solution in solid line and two different condensate val-
ues, ϕ = 0.2 and ϕ = 0.4, with dashed line and dotted respectively.
where ϕ is zero in our visible universe. When e = 1, gtt is negative and t is a timelike
coordinate. At a horizon, gtt and grr exchange signs and so e = −1. Then
 s 
Λ

r  p
E (r ) = 1−e 1 − ϕ e 1 − r2  . (4.23)
GN 3
The quasilocal energy for a de Sitter metric is shown in Figure 4.2.

As expected, the energy continually grows with increasing r. The horizon forms
when the energy at the surface is larger than E(rc ) = rΛ /GN . As in the Schwarzschild
case, the formation of a condensate diminishes the energy with respect to the vac-
uum case, so it seems that the condensate structure is favorable against the vacuum
itself.
Just for completeness the solution combining the Schwarzschild and de Sitter
spacetimes can be easily computed too. The quasilocal energy is plotted in Figure
4.3. Both horizons, the black hole horizon and the cosmological horizon, can be
directly seen in the curve. In both points even though the derivative of the energy
matches across the horizon, it is divergent. Also in both points, the horizon forms
when the quasilocal energy crossed the value r/GN .
The above picture provides a criterion to discern whether a condensate can be
formed or not. Any time the radial coordinate transform its character from spacelike
to timelike, when crossing a horizon, a condensate structure is energetically favored
with respect to the vacuum solution.
1.4
1.2
E(r)
1.0
Quasilocal Energy
0.8
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
r
F IGURE 4.3: Quasilocal energy for a Schwarzschild black hole embedded in a

de Sitter space. The units are such that rΛ = 1. The black hole mass is unre-
alistically large, rS = 0.1 rc to make details notorious. Beyond both horizons
we plot a condensate structure for comparison, with ϕ = 0.2 and ϕ = 0.4 in
dashed and dotted lines respectively.
4.3.2 The case of the de Sitter horizon
Before entering in a more detailed discussion about possible physical interpretations

in the case of the cosmological horizon in de Sitter, let us analyze this spacetime but
using an inverse radial coordinate

 1

 r=
z


(4.24)

 1
dr = − 2 dz .


z

With this change, the metric reads as

−1
Λ 1 Λ 1

2 2 1 1
ds = − 1 − dt + 1 − dz2 + 2 dΩ2 . (4.25)
3 z2 3 z2 z 4 z
This change of variables maps the cosmological horizon to a concentric sphere cen-
tered at z = 0. The Einstein tensor is constant in r coordinates, then it is not changed
by this change of variables and the equations of motion remains the same as well.
Let us now consider the geodesics in this spacetime for straight trajectories char-
acterized by constant angles θ and φ. Some Christoffel symbols used to compute
geodesics are
3 Λ z2 − Λ2
Γztt =
9z
Λ
Γttz = (4.26)
3 z3 −Λz
6 z2 − Λ
Γzzz = − .
3 z3 − Λ z
The geodesic equation is

2
d2 z 3z2 − Λ 6z2 + Λ

dz
= − Λ + . (4.27)
dt2 9z 3z3 − Λz dt
This equation is virtually identical in structure to the one corresponding to a particle

moving towards the horizon in a Schwarzschild metric with vanishing angular mo-
mentum and, exactly like in that case, the observer sees the particle to slow down as
it approaches the horizon that —at least classically— never crosses (according to the
observer at infinity in the case of Schwarzschild, at r = 0 in the case of de Sitter, i.e.
at z = ∞).
Continuing with this analogy we derive the quasilocal energy using the z coordi-
nate where the space beyond the de Sitter horizon is mapped onto a sphere of radius
zΛ = 1/rΛ . It is given by
r !
1 Λ 1
E(z) = 1− 1− (4.28)
z 3 z2
for z > zΛ and r !

1 p Λ 1
E(z) = 1+ 1−ϕ −1 (4.29)
z 3 z2
for z < zΛ . The derivation for this expression was made directly using the inverse
z coordinate. If we go back to the radial coordinate z → r we obtain the expression
(4.23) but with a difference of a global minus sign.
Radial geodesics (in the usual coordinate r) can be obtained easily from the ex-
pression
!2
2 Λr2 Λr2
ṙ = 1− , (4.30)
3 3
where Λr2 /6 would be the analogous of the Newtonian potential. Note that the
usual interpretation that one element of vacuum repeals each other, making the ex-
pansion of the universe obvious, is very intuitive in FLRW coordinates, but is not so
obvious in the static coordinates used in this section.
In spite of the formal similarity between the expressions for the quasilocal energy
(4.28) and (3.31), the situation is different because it does not seem possible to derive
a potential such as the one in (3.38). The reason is that (4.28) is mass-independent.
As a consequence, the argument that was presented in the case of a Schwarzschild
black hole, where it was argued that the presence of the potential well implied a sub-
stantial amount of gravitational radiation that could be stored in the inner part of the
black hole in form of a graviton condensate does not hold. In the present case the
potential well appears to be absent and consequently there is no substantial energy
loss in form of gravitational radiation.
Intuitively, the de Sitter horizon appears to be somewhat different from the one
associated to a Schwarzschild black hole (or to a Reissner–Nordström for that mat-
ter). To the external observer at infinity, the horizon at r = rS has a very clear mean-
ing and objective existence. On the contrary, in the de Sitter spacetime the horizon
is dependent on the observer situation (i.e. where the r = 0 point is assumed to be).
However as discussed in some detail in the seminal paper of Gibbons and Hawking
[91] this might not pose any conceptual problem in principle. It is only measurable
effects by an observer what matters.
As it is well known, there is a temperature associated to the de Sitter horizon,
given by [91] r
1 Λ
T= . (4.31)
2π 3
This raises a puzzling possibility: matter falling into the de Sitter horizon in the
inverse coordinate z; actually accumulating close to the horizon in the static coor-
dinates (r, t). In the Schwarzschild case we concluded that this would most likely
imply a growth in the value of the graviton condensate ϕ. If this were the case, we
immediately notice that because of the change in the cosmological constant beyond
the horizon Λ → CΛ with C = 1 − ϕ the vacuum energy would decrease with time.
However, this does not affect neither the position of the horizon, as the change in
the metric is merely a multiplicative factor in the time and radial components, nor
the value of the cosmological constant in the visible part of the universe.
However, the fact that there is no gravitational binding potential of the form
(3.38), obtained in the Schwarzschild case, would imply that there is no way to create
of a gravitational BEC and therefore in spite of being energetically favourable, the
value of the condensate should be zero and consequently there would be no change
in the cosmological constant neither inside nor outside the de Sitter horizon.
69
Chapter 5
Gravitational waves propagating

over nonempty backgrounds
5.1 The standard cosmological Model

The ΛCDM (Lambda cold dark matter) is the simplest parametrization of the Big
Bang cosmological model. The three main contributions are the cosmological con-
stant, which is associated with dark energy, the cold dark matter (CDM) and ordi-
nary matter (SM). This model is "almost universally accepted by cosmologists as the best
description of the present data" [92] consistent with the history of our universe, since a
very small fraction of a second after the Big Bang until the undergoing accelerated
phase, providing a reasonably good description of the various processes occurred
along its 13.8 billion years of existence. This concordance model describes the cos-
mic dynamics using the theory of general relativity with its field equations, together
with very different branches of physics, from astronomy and thermodynamics to nu-
clear and particle physics. The success of the model lies in the capability of properly
fitting a large variety of observational data.
One of the fundamental pillars upon which the concordance model is built upon
is the cosmological principle, which postulates that our location in the universe is not
special; the large scale isotropy (it looks the same in all directions) and homogene-
ity (from every location) are the two key features [93]. These two constraints can of
course only be checked in the region where we have observational access and only in
our world line. However no world line is special: neither ours nor any other for any
galaxy. Therefore, isotropy about all galaxies implies homogeneity, and the cosmo-
logical principle can be deduced from the isotropic hypothesis, which is exceedingly
sustained by radiation backgrounds1 . Theoretical predictions of the ΛCDM model
state that the matter distribution in our universe becomes homogeneous at scales
larger than R H = 62.0 h−1 Mpc (∼ 100 Mpc), in close agreement with data. For
instance, in [94] a transition value to cosmic homogeneity of R H = (63.3 ± 0.7) h−1
Mpc is inferred from large scale structures. The values are given in terms of the re-
duced Planck constant h = H0 /[100 km s−1 Mpc−1 ]; recall the different values for
the Hubble constant discuss below equation (1.21).
1 For instance, CMB observations shows very faint anisotropies of the order of ∆T/T ∼ 10−4 − 10−5 .
70 Chapter 5. Gravitational waves propagating over nonempty backgrounds
Different collaborations studying various cosmological phenomena [42], from

temperature fluctuations measurements to high-redshift supernovae observations,
CMB and BAO measurements, weak lensing techniques (such as cosmic shear and
redshift space distortion), etc. agrees on a structure for the metric tensor accomplish-
ing the cosmological principle. The model uses the FLRW metric2 , which is written
as  
dR2
ds2 = dT 2 − a2 ( T )  + R2 dΩ2  . (5.1)
R2
1−k R0 2
The coordinate T refers to the cosmic time and with the radial coordinate ranging
from 0 ≤ R < R0 , we get the well known comoving coordinates; we will refer to
them with capital letters: ( T, R, θ, φ). a( T ) is the scale factor, a dimensionless quan-
tity which gives a mapping between the theoretical predictions and the observations;
i.e. between the (constant) comoving distance and the (physical) proper distance at
a given cosmic time. The usual election of a( T0 ) = a0 = 1 at the present age of the
universe has been taken as the reference time. The curvature parameter k can take
three different values, +1, −1 or 0, depending on the geometry of the spacetime,
whether it is closed, open or flat respectively.
Suppose that the history of the universe is reasonably well described by the Ein-
stein’s equations with a cosmological constant (1.20). At a background level we
can treat the components of the universe as some kind of fluids uniformly distributed
along the whole space. In thermodynamic equilibrium, the energy momentum ten-
sor for each component can be expressed as

T αβ = ρ + P U α U β − P gαβ , (5.2)
in the time-positive metric signature tensor notation. U α is the 4-velocity of the fluid
with respect to the observer; the energy density ρ and the isotropic pressure P are
measured in the rest frame of the fluid. The recently introduced FLRW comoving
coordinates ( T, R, θ, φ), belong to the so-called rest frame and implies U T = UT = 1,
whereas UR = Uθ = Uφ = 0. The normalization condition gαβ U α U β = 1 is fulfilled.
In order to determine the time evolution of the scale factor, Einstein field equations
(1.20) are required together with some equations of state that relate the pressure and
the density for every single component present; these are
Pi = ωi ρi . (5.3)
Using the recently introduced geometric properties of homogeneity and isotropy

with the metric (5.1) and the aforementioned energy-momentum tensors (5.2), the
2 At this point we change to the (+ − −−) signature to follow the usual standards in cosmology.
5.1. The standard cosmological Model 71
field equations are

. 2
a k Λ κ
+ 2 − = ρm
a a 3 3
(5.4)
..
a Λ κ
− = − (ρm + 3 Pm ) .
a 3 6
They are the so-called Friedmann equations. The dot stands for derivatives with
respect to the cosmic time T. It is usually useful to work with the Hubble function
.
defined as H = a/a. At this point when we talk about matter, we are not distinguish-
ing between nonrelativistic matter (baryonic or dark matter) and relativistic matter
(radiation). Further on we will study each component separately.
It is customary to divide the first equation in (5.4) by the square of the Hubble
parameter today, H ( a0 ) = H0 , and define the following density parameters
κ ρm
Ωm = (5.5)
3 H0 2
Λ
ΩΛ = (5.6)
3 H0 2
k
Ωk = − . (5.7)
a2 H0 2
The sum of the three equals unity. Each density parameter corresponds to the ratio
between the actual (or observed) energy density ρi and the critical energy density
ρc = 3H0 2 /κ; this is Ωi = ρi /ρc . From Figure 5.1 it is seen that observations favour
a flat universe within a small amount Ωk ∼ 0. These analysis assumed ωΛ = 1 for
the equation of state parameter for the cosmological constant [95], along this lines
the same assumption has been made. While current observations are consistent with
k = 0, they are very far from selecting out only this value.
For the sake of simplicity and in accordance with future approximations, we as-
sume k = 0 and consider the remaining two parameters (5.5) and (5.6). At this point,
we can distinguish different types of matter characterized by different cosmological
parameters wi in (5.3). Each contribution has a weight in the effective matter density
parameter
Ωm = Ωnonrelativistic matter + Ωrelativistic matter . (5.8)
Several independent techniques like cluster mass-to-light ratios [97], baryon densi-
ties in clusters [98], weak lensing of clusters [99], the existence of massive clusters
at high redshift [100] constrain the model such that baryons account for the ∼ 5%
energy content of the universe, whereas dark matter for the ∼ 25%. Relativistic
matter encompasses the measure of the present mass in the relativistic 3 K thermal
cosmic microwave background radiation that almost homogeneously fills the space,
2.0
F IGURE 5.1: ΩΛ − Ωm plane from
No Big Bang the CMB, BAO and the Union SNe
Ia set, together with their combina-
tion [96]. The confidence level con-
tours of 68.3%, 95.4% and 99.7% are
shown in color gradient. The val-
1.5
ues above the "Flat" line indicate
a closed universe. These observa-
tions appear to point towards ΩΛ +
Ωm ∼ 1 within a small percentage,
meaning that the universe seems to
be approximately flat on the scale of
1.0 our horizon. More precisely, the spa-
SNe tial curvature density today (5.7) in-
dicates that the radius of the spacial
universe is at least about one order
of magnitude larger than the radius
of the visible universe. Clearly, we
are surely influenced by our expe-
0.5
rience in Earth, which is flat within
our visible horizon. The impossibil-
CM ity of conceiving a flat infinite Earth
B
Fl is surely a well-founded prejudice
at
BAO on our unnatural perspective for a
perfectly flat universe.
0.0
0.0 0.5 1.0
together with the accompanying low mass neutrinos, accounting for an almost neg-
ligible contribution of
Ωrel ∼ 10−4 − 10−5 . (5.9)
The remaining component corresponds to dark energy, which is roughly 30 times

bigger than the observed matter energy [101]. We live in the precision Cosmology
era and the ΛCDM model is still resisting the new generation of galaxy surveys and
CMB probes.
The equation of state (5.3) for the main components of the universe provides
enormous information. The accepted values for the state parameters of the three
main contributions of the model are
ωΛ = −1
ωnr = 0 (presureless) (5.10)
1
ωrel = .
3
5.1. The standard cosmological Model 73
The possibility of relating the pressure and the density separately for each compo-
nents states that the second equation in (5.4) can be rewritten as
..
a κ κ
a
=−
6 ∑ (ρi + 3 Pi ) = 6 (2ρ Λ − 2ρr − ρnr ) . (5.11)
i =1
We favour the use of the density of the cosmological constant, rather than the cos-
mological constant itself. We also distinguish the relativistic from the nonrelativistic
matter (either baryonic or non-baryonic). Interestingly, this equation that describes
the acceleration of the universe clearly brings that ρΛ speeds up the expansion, whilst
the matter content slows it down. Generally, any background component satisfying
ρi + 3 Pi < 0 contributes positively to the acceleration.
With the purpose of discussing a novel effect coming from the fact that gravita-
tional waves propagate over an expanding background. Next section reviews pre-
vious results about this effect obtained in a theoretical de Sitter universe. This line
of research argues that this effect seems to have a window to be measured in Pulsar
Timing Array (PTA) observations; in fact a detailed characterization of the expected
signal in these observations is presented in Chapter 6. In section 5.3 we employ some
basic ingredients of the ΛCDM cosmology recently reviewed to study the lineariza-
tion of Einstein’s equations and argue why this is the proper conceptual framework.
Next, in section 5.4, we give the main issues related to the choice of the coordinate
system for the de Sitter case. This discussion actually contains the essential ingredi-
ent of the present study. Later we consider the case where there is no cosmological
constant but only non-relativistic matter; and finally in section 5.6, we extend our
considerations to the combined ΛCDM case of two main components (with known
analytic solution): dark matter and dark energy. We have derived the coordinate
transformation between spherically symmetric coordinates centered at a coalescence
of two black holes where gravitational waves are produced, and FLRW expanding
coordinates, where gravitational waves propagate and are measured. One step fur-
ther is given in section 5.7 where the same procedure is performed for relativistic
matter. This allows one to infer a linearized picture in terms of the Hubble parameter
H0 which would take into account all constituents of the universe as a background
entity in the last section.
In this manner, we have found a direct link to see whether this effect has a real-
istic correlation with current measurements and describe our universe in a realistic
way. In this chapter we focus on the more fundamental aspects concerning the def-
inition of and relation between the different coordinate systems, while we leave the
discussion of their impact in PTA data for next chapter.
5.2 The importance on the choice of coordinate systems

In [58, 59] the influence of a non-vanishing cosmological constant on the detection of
gravitational waves in PTA was discussed3 . It was found that the value of the timing
residuals observed in PTA for a given range of angles subtended by the source and
the pulsars were dependent on the value of Λ. Given that the gravitational waves
potentially observed in PTA would correspond to the final phase of the fusion of
two very massive black holes lurking at the center of two colliding galaxies at rel-
atively low redshift, this result opens the possibility of ‘local’ measurements of the
cosmological constant; namely at subcosmological distances.
To simplify things, imagine a large mass M and a much smaller mass m orbiting
around it. For the time being let us assume that there is no cosmological constant
and no matter density. In the situation just described, the geometry is very approxi-
mately described by the well known Schwarzschild metric corresponding to a mass
M [3]. This metric is expressed in terms of the radial coordinate r and a time coordi-
nate t. This time coordinate t is just the time ticked by a clock located at r → ∞, i.e.
an observer at rest at infinity.
Next let us assume that a cosmological constant Λ 6= 0 is present. In this case the
relevant metric is also well known [102]; the Schwarzschild–de Sitter (SdS) metric:
−1
2GN M Λ 2 2GN M Λ 2

2 2
ds = 1− − r dt − 1 − − r dr2 − r2 dΩ2 . (5.12)
r 3 r 3
This metric is unique once the requirements of spherical symmetry and time inde-
pendence of the metric elements are imposed [67]. It exhibits two horizons at
q
3
rS = 2GN M and rΛ = Λ . (5.13)
Here r has the same physical realization as in the case of the Schwarzschild metric
mentioned above, but the metric is not anymore Minkowskian when r → ∞ so one
must understand carefully what the meaning of the time coordinate t is. In this work
we will consistently denote with lower case letters, i.e. (t, r, θ, φ), the coordinates
relevant in the location where gravitational waves are emitted.
In the situation just described, if gravitational waves are produced it is obvious
that they will be periodic in the time coordinate t (see also the discussion in section
5.3), and at a distance sufficiently large from the source but very small compared to
cosmological distances they will be described by (a superposition of) approximately
harmonic functions of the form
eµν
hµν (w; t, r ) = cos w (t − r ) , (5.14)
r
with eµν being the polarization tensor. This is so because the metric is dominated by
3 That is, beyond the frequency redshift due to the expansion of the universe.
5.2. The importance on the choice of coordinate systems 75
the strong gravitational field created by the mass M, and in its vicinity the contribu-
tion of the Λr2 term (that produces very small corrections in the orbit parameters)
is totally negligible. On the other hand, determining the precise form of the front
wave (namely the actual superposition of harmonic functions) is a non-trivial task
that in general requires a detailed analysis. Here we assume that the gravitational
wave emission is strongly dominated by a single harmonic with a frequency that is
governed by the period of rotation. There would be no problem to include in our
considerations a dependence of the amplitude on t, particularly relevant in the last
stages of the spiraling down process, but we do not consider it here for simplicity.
If at long distances from the source the universe is approximately Minkowskian,
as it would be the case if the metric is globally Schwarzschild, this wave front would
be seen by a remote observer with exactly the same functional form. This is not
the case when a cosmological constant or dust are present. The universe cannot
be considered to be globally Minkowskian although at very short distances this is a
good approximation, except in the case of strong gravitational sources being present.
In the vicinity of the very massive object of mass M, Einstein’s equations together
with symmetry considerations dictate the form of the solution.
However, as discussed in section 1.5 our spacetime is not Minkowskian far away
from the gravitational wave source source and we know that the distribution of mat-
ter and energy governs the way clocks tick. Exactly as clocks in the vicinity of a
strong gravitational field created by a very massive object signal a time t to a remote
observer, at cosmological scales the universe being homogeneous and isotropic dic-
tates at what rate comoving clocks, far from strong gravitational sources, tick. We
know that the latter physical situation is described by a FLRW metric described by
coordinates ( T, R, θ, φ)

ds2 = dT 2 − a2 ( T ) dR2 + R2 dΩ2 . (5.15)
Then a cosmological observer located very far away from the black hole merger will
not see the same functional dependence in the wave front simply because T 6= t and
R 6= r.
It is particularly interesting to consider the case where only a cosmological con-
stant is present because then the coordinate transformation between the two coordi-
nate systems is precisely known when one is placed far away from the source of the
gravitational field [49], as is obviously the case if r 2MG. The universe is globally
de Sitter; SdS and FLRW coordinates correspond to slicing the same spacetime in
two different ways. Both are solutions very far from the source. Clocks tick with
time t near the large mass M and with time T at the cosmological distances where
we observe the phenomenon. In both cases the distribution of matter and energy
dictates what time is.
The cosmological time and comoving space coordinates will in fact be non-trivial
functions of the SdS coordinates i.e. T (t, r ), R(t, r ). These transformations are well
known in the case where the universe is globally de Sitter, and will be reviewed in
the next sections. Adding matter further modify these transformations. As a con-
sequence of the different coordinate systems, some anharmonicities appear when
(5.14) is written in terms of the observer coordinates T, R. Given that the cosmo-
logical constant is small, these anharmonicities are numerically small too, but the
results of [58] suggest that nevertheless they could be measurable in a realistic PTA
observation.
Conventionally, the expansion of the universe is taken into account by replacing
w by its redshifted counterpart in (5.14). If z is small
w 0 ' w (1 − z ) , (5.16)
with z being the redshift factor. We will see that this effect on the frequency is indeed
reproduced, but there is more than this.
Let us justify why in PTA observations the coordinates that an astronomer uses
to study the phenomenon are (up to some modifications described in Chapter 6) T
and R to a high degree of accuracy. Galaxies are following the cosmic expansion
with respect the source of the gravitational waves and therefore the relevant change
of coordinates for an observer sitting at, say, the center of mass of a galaxy would
be the one relating (t, r ) with ( T, R). Needless to say, gravitational fields inside the
galaxy locally modify this relation. These modifications are truly negligible, except
for those that will be considered in the next chapter. Therefore in this work we will
consistently denote the coordinates used by an Earth-bound observer by ( T, R). Nat-
urally, the Earth and pulsars are gravitationally bound to the Galaxy and the distance
of a given pulsar to an observer on the Earth is a constant distance L in coordinate
R. On the other hand, the gravitational wave source is not gravitationally bound,
so it feels the expansion of the universe. In short, the effect that will be discussed
and analyzed in this work has nothing to do with an hypothetical expansion of the
distance between the pulsar and the observer (which is not present) but rather with
the non-trivial transformation relating coordinates (t, r ) and ( T, R).
The results obtained in [58, 59] were not yet directly applicable to a realistic mea-
surements of PTA timing residuals because at the very least non-relativistic dust
needs to be included to get a realistic description of the cosmology. This is in fact the
main purpose of the following part of this thesis.
5.3 Linearized gravity framework

The linearized theory of gravity is nothing else than perturbation theory around
Minkowski spacetime. This is, starting with a small deviation |hµν | 1 from a flat
metric tensor,
gµν = ηµν + hµν . (5.17)
5.3. Linearized gravity framework 77
The usual procedure for obtaining the linearized Einstein tensor can be found in any
textbook on general relativity (see e.g. [103]). However, a gauge choice is mandatory
to obtain a solution to these equations in order to avoid redundancy under coordi-
nate transformations x µ → x µ + ξ µ ( x ). Although the following discussion is valid
in any gauge, it is simplest in the familiar Lorenz gauge
1
∂µ hµ ν − ∂ν h = 0 , (5.18)
2
or defining the trace-reversed perturbation variable
1
h̃µν = hµν − h ηµν ⇒ ∂µ h̃µ ν = 0 . (5.19)
2
In this gauge the linearized Einstein’s equations, including the cosmological con-
stant and the energy momentum tensor read
h̃µν = −2 Ληµν − 2 κ Tµν . (5.20)
As discussed in [49], up to first order in Λ the perturbation hµν in equation. (5.17)

can be decomposed into a gravitational wave perturbation h(µν
GW)
and a background
Λ)
modification due to the cosmological constant h(µν , being both small in magnitude. It
is straightforward to propose a new contribution due to dust h(µν
dust)
. This component
would contribute as an independent new background that will not interact with the
other one, the one corresponding to Λ, and that will be also small enough so that
the linearized approximation to be meaningful. Therefore, the total metric would be
written as
Λ)
gµν = ηµν + h(µν
GW)
+ h(µν + h(µν
dust)
. (5.21)
The first two contributions have been already discussed in [48]. The two indepen-
dent background would satisfy each a simpler equation of motion
Λ)
h̃(µν = −2 Ληµν (5.22)
for the cosmological constant and
h̃(µν
dust)
= −2 κTµν (5.23)
for a dark matter background uniformly distributed along the spacetime. Finally the
perturbation due to the gravitational wave itself satisfies the usual homogeneous
wave equation
h̃(µν
GW )
= 0. (5.24)
In the previous discussion we have not included the modification due to the grav-
itational field created by the very massive object of mass M as this is negligible at
large distances, but the perturbation is also additive as is easily seen by expanding
Schwarzschild in powers of GN . Thus the discussion is easiest in linearized grav-

ity. The contribution of the different terms simply adds to construct the total metric.
And because the deviations with respect to Minkowski spacetime are small away
from the gravitational wave source, linearization is a priori well justified.
If we stress the importance of the linearization process it is because it is not al-
ways possible to accomplish in all coordinate systems. For instance, it is impossible
in cosmological coordinates. This will be discussed in more detail in section 5.4. As
a consequence there are no harmonic gravitational waves waves in FLRW coordinates4 .
In SdS coordinates, however, linearization is possible [49, 58], and linearization re-
mains valid after including dust, which is one of the main results of this chapter.
The solution of equation (5.22) as explained in [49] should correspond to the lin-
earization (i.e. expanding in Λ at first order) of equation (4.17); i.e. the metric (5.28).
In fact this statement is not totally correct because in (5.20) the Lorenz gauge condi-
tion is assumed, while the SdS metric does not fulfill this gauge condition, as it can
be easily checked. In fact, as explained in [49] a trivial coordinate transformation
turns the SdS metric into one that complies with the Lorenz gauge. This coordinate
transformation is (a) time-independent and (b) is of order Λ. The discussion on grav-
itational waves and the separation in background and waves is however simplest in
the Lorenz gauge and it is the one just presented. In fact, the relevant effects to be
√
discussed later are all of order Λ and order Λ corrections are safe to be neglected.
5.4 Gravitational wave propagation in de Sitter

De Sitter geometry models a spatially flat universe and neglects matter, so the dy-
namic of the universe is dominated by the cosmological constant Λ. Obviously there
exist a number of useful coordinate choices for this spacetime. These consist in pick-
ing a convenient time choice and thus defining a family of spacelike surfaces. Some
slices make explicit a cosmologically expanding space with flat curved spatial part,
whereas some other do not.
There is as well a unique choice of coordinates in which the metric does not
depend on time at all; as is described in equation (5.12) if the black hole contribution
were omitted
−1
Λ Λ

2
ds = 1 − r2 dt − 1 − r2
2
dr2 − r2 dΩ2 . (5.25)
3 3
The mere existence of such a choice is enough to tell us that there is no fundamental
sense in which this is an expanding cosmological spacetime. The notion of expansion
is a concept that is linked to a coordinate choice. Yet, the choice of coordinates is not
totally arbitrary. Observers’ clocks tick at a given rate, and they can measure the
local space geometry at a fixed time.
4 This statement remains true even after taking into account the R−dependent redshift.
5.4. Gravitational wave propagation in de Sitter 79
Cosmological observers detect that the universe as we see it is spatially flat and
that gravitationally unbound objects separate from each other at a rate that is gov-
erned by the cosmological scale factor. a( T ) defines the metric (5.15) in cosmological
coordinates and the first –flat, i.e. with k = 0 in (5.7)– equation of motion from (5.4)
with ρm = 0, is a sufficient condition to get
. r √Λ
a( T ) Λ ∆T
≡H= =⇒ a ( T ) = a0 e 3 . (5.26)
a( T ) 3
The cosmological constant sets the expansion rate, given by the Hubble parameter
H. ∆T = T − T0 is the time interval and T0 is some initial time which is arbitrary
and associated with the initial choice of the scale factor such that a( T0 ) = a0 .
Then the physical coordinate system for discussing cosmology is given by the
FLRW metric with coordinates ( T, R, θ, φ), as this coordinate system encodes the
observed homogeneity and isotropy of the matter sources summarized in the cos-
mological principle [93]. Of course, space redefinitions of the metric are innocuous.
It is of no consequence to choose to describe the world around us using Cartesian
or polar coordinates. As long as coordinate transformations do not involve time in
any essential way, any coordinate choice will be equally good to describe a universe
conforming to the cosmological principle.
Time and space will however look very different to an observer in the vicinity
of a black hole or, for that matter, in the vicinity of any strong gravitational source
that is spherically symmetric, centered at one point that we conventionally denote
by r = 0. The spacetime there is isotropic, but not homogeneous. If the black hole
is static, we know that the solution is unique and it is given by the Schwarzschild
metric. If in addition, there is a cosmological constant, the corresponding metric will
be the one given in equation (5.12).
The SdS metric approximates a Schwarzschild space for small r; and for large r
the space approximates a de Sitter space. We are focusing right now on effects on
gravitational wave propagation, hence we are in a large r regime. Far away from a
SMBH with a typical Schwarzschild radius of ∼ 1011 m, the term rS /r can be safely
neglected from our considerations and only the Λ dependence needs to be taken
into account; i.e. under this regime the metric is given by (5.25). In this case the
transformation relating the SdS coordinates to the FLRW ones are well known (see
e.g. [49])  r r
 3 Λ
log 1 − a2 ( T ) R2

 t( T, R) = T −

Λ

3
(5.27)
 √Λ
 r ( T, R) = a( T ) R = a0 e 3 ∆T R ;



the θ and φ coordinates do not transform. Note that even though for r rS the influ-
ence of the black hole is negligible, its presence sets up a global coordinate system,
and r and t have a well defined physical meaning.
The SdS metric in the far away from the source limit can be linearized expanding
(5.25) in powers of Λ. For r → ∞, but Λr2 1 (i.e. well before the cosmological
horizon)
Λ 2 Λ

ds2 = 1− r dt2 − 1 + r2 dr2 − r2 dΩ2 . (5.28)
3 3
This metric satisfies a linearized version of Einstein field equations (1.20). On the
contrary, the FLRW metric does not fulfill any linearized version of Einstein’s equa-
tions5 ; in fact, no metric that depends only on time can be solution of the linearized
Einstein’s equations with a cosmological constant, whatever the gauge choice se-
lected.
If linearization were possible in FLRW coordinates, gravitational waves could be
harmonic in those coordinates; i.e. they could obey a (flat) wave equation. But this
as we have seen is impossible. As a consequence, gravitational waves are necessarily
anharmonic in FLRW coordinates. Therefore the proper procedure is to find a co-
ordinate system where linearization is possible and the wave perturbation and the
background can be treated additively. Luckily this is accomplished in SdS coordi-
nates because they are analytic in Λ and have a smooth Minkowski limit. Lineariza-
tion and, consequently, harmonic waves remain valid when dust is included. The
change from SdS to FLRW is necessarily not a "soft" one –it must have a singularity
somewhere, and it does.
If one attempts to solve directly the wave equation with a non-zero cosmological
constant in FLRW using perturbation theory in Λ, one generates secular terms that
grow with ( T, R) and invalidate perturbation theory.
The coordinate transformations relating SdS and FLRW can be expanded for
small values of ΛT 2 or ΛR2 (this will always be the relevant situation in the ensuing
discussions)  r !
 R2 Λ Λ
t( T, R) ≈ T + a20 + ∆T R2


 +...
2 3 3




" # (5.29)
r
Λ Λ


r ( T, R) ≈ a0 1 + ∆T + ∆T 2

R+...




 3 6
The change is of course still non–analytical in Λ. Applying this approximate trans-

formation to the SdS metric, one ends up with an approximate version of the FLRW
metric r !
Λ Λ
2
ds ≈ dT − 2
a20 1+2 ∆T + 2 ∆T 2 2 2
dR + R dΩ 2
(5.30)
3 3
which is the expansion of (5.1) for small values of the cosmological constant. In prac-
√
tice, the leading Λ term will be the relevant one for our purposes. The extension of
these considerations to the case where a cosmological constant and non-relativistic
matter coexist comes next.
5 This is easy to understand by realizing that in the FLRW metric Λ appears in a non-analytic form.
5.5. Non-relativistic matter 81
5.5 Non-relativistic matter

First we would like to repeat the discussion previously done in section 5.4 in the case
where there is no cosmological constant, just dust. Namely we ask ourselves what
would be the equivalent of equation (5.27) in such a case.
According to the discussion around (5.8), we will model the non-relativistic mat-
ter background as a pressureless dust-like perfect fluid, with a state parameter given
in (5.10). The gravitational field is produced entirely by matter characterized by a
positive mass density, given by the scalar function ρdust , but with the absence of pres-
sure. The stress-energy tensor is
T µν = ρdust U µ U ν . (5.31)
As said above, in FLRW comoving coordinates the four–velocity is given by

Uµ = (1, 0, 0, 0); so the stress-energy tensor reduces to T µν = diag(ρdust , 0, 0, 0). The
Einstein’s equations derived for this metric structure together with the equation of
state corresponding to ωdust = 0, give the solution [70]
2/3
T
a( T ) = a0 ,
T0
(5.32)
4
ρdust ( T ) = .
3 κ T2
Therefore, we are able to write the scale factor and the metric with a density depen-
dence exclusively
" #2/3
2 2 ρ d0
ds = dT − a20 dR2 + R2 dΩ2 . (5.33)
ρdust ( T )
Here, ρd0 = 4/(3 κ T0 2 ) is the density measured at a particular cosmological time T0 ;

the latter usually set according to a scale factor such that a0 = 1.
With the aim of deriving a full picture on how dust influences the observation of
gravitational waves, we need to find a coordinate system where dust is described in
‘static’ spherically symmetric coordinates with an origin coincident with the object
emitting the gravitational waves. The first step to find this (spherically symmet-
ric) set of coordinates is to impose on the angular element that in FLRW reads as
a( T )2 R2 dΩ2 , to become simply r2 dΩ2 . The coordinate transformation will be given
by 


 T (t, r ) = unknown function

(5.34)
r
 R(t, r ) = a( T ) .



Next we take into account the transformation properties of a rank 2 tensor
∂X µ ∂X ν
gµ 0 ν 0 = gµν . (5.35)
∂x µ 0 ∂x ν 0
Spherical symmetry is a requirement for both metrics, then the transformation for
the two angular variables is characterized by the identity. This means that in the
angular diagonal components of the metric the Jacobian corresponds to the identity
and the transformation for these components do not bring any new information.
Analogous is the case for the angular and off-diagonal elements of the metric. The
transformation is trivial and in both pictures the components of the metric are null.
However, there is one off-diagonal element that is not null per se; the gtr = grt ele-
ment. If a diagonal metric is desired we must impose the vanishing of this element.
This translates into the following differential equation for the cosmological time T
6rT κ ρ2dust r
∂r T = − ⇐⇒ ∂r ρdust = . (5.36)
9T − 4 r2
2 κρ
1 − 3dust r2
Of course we already know the structure for the bijection between the cosmologi-
√
cal time and the content of the universe, i.e. | T | ∝ 1/ ρdust coming from equation
(5.33). This allows to write the diagonal metric condition (5.36) in terms of the non-
relativistic matter density only. We have two remaining components of the metric
to be transformed: the TT and RR–components. If we use only the requirement for
the radial transformation (5.34) and impose a diagonal structure by (5.36), the metric
reads
2
2 ∂t ρdust κ ρdust 2 1
ds = 3
1− r dt2 − 2 2 2
κ ρdust 2 dr − r dΩ . (5.37)
3 κ ρdust 3 1− 3 r
Note that ρdust is a (scalar) known function of T and therefore it is expected to depend
on the new temporal variable t; but this dependence is unknown a priori (because t
is unknown so far). In any case we see that the new metric necessarily contains a t
dependence via ρdust (t).
Before continuing, let us reduce the number of dimensionful parameters of the
theory. The limit of weak coupling (κ → 0) and vanishing matter density (ρdust → 0)
are essentially the same because both parameters appear together in all the equations
–except in the energy momentum conservation which is trivially satisfied. Hence we
define: ρ̃dust = κρdust . Then the solution of (5.36) is
6 + r2 ρ̃dust
= C (t) ; (5.38)
ρ̃dust 1/3
where C (t) is a constant with respect to r, meaning that it can only depends on t.
Dimensional analysis do the rest of the work. In natural units, [ρ̃dust ] = L−2 , then
the units for the ‘constant’ should be [C (t)] = [ρ̃dust ]−1/3 = L2/3 . In this picture, κ
does not belong to the theory, hence there is only one dimensionful parameter to
5.5. Non-relativistic matter 83
give units to C (t); namely t. As [ t ] = L, then C (t) = A t2/3 with A a positive

dimensionless parameter to be determined.
It is expected for later times the dust density to be homogeneously diluted. This
requirement automatically fixes the remaining free parameter of the theory, A. This
is translated into the metric (5.37) to be Minkowskian–like when t → ∞. This is auto-
matically fulfilled except for the factor in front of the brackets in the gtt element. If a
Minkowskian limit is expected, the following enforcement of the prefactor is needed
(∂t ρdust )2 /(3κρdust 3 ) → 1. We have a relation for obtaining the temporal variation of
the dust density, namely (5.38) with the corresponding C (t) function. Therefore,
reintroducing κ and deriving both sides of the equality, one arrives to
(κρdust )4/3 A
∂t (κρdust ) = − κρ . (5.39)
1 − 3dust r2 3 t1/3
Squaring the last result, dividing by 3 (κρdust )3 , and finally using equation (5.38)
again for replacing 1/t2/3 , the prefactor for the gtt component in the limit of ρdust → 0
becomes
A2 A3
−→ . (5.40)
33 (κρdust )1/3 t2/3 33 6
After all, imposing the flat asymptotic limit to this factor, the constant is fixed to
√
A = 3 3 6. Once A is known, ∂t ρdust can be expressed as a function of ρdust and r
uniquely. Following the same steps as before; replacing 1/t2/3 from the prefactor by
A/C (t) from equation (5.38), we get simply
2
∂t ρdust 1
= 2
3 κ ρdust 3

κ ρdust κρdust
1+ 6 r2 1− 3 r2
where the asymptotic limit is shown explicitly. Then the sought for metric turns out
ot be
1 1
ds2 = dt2 − κ ρdust dr2 − r2 dΩ2 . (5.41)
1+
κ ρdust
r2 1−
κ ρdust
r2 1− 3 r2
6 3
This metric has a good Minkowskian limit for ρdust → 0. It has similar character-
istics to the ones of the SdS metric and it is clearly the relevant metric to describe
the Keplerian problem we alluded to in the introduction, in the presence of dust but
without cosmological constant, when properly extended with the black hole gravi-
tational potential (see below).
Further discussions concerning dust in these coordinates are included in the ap-
pendix. With the full structure for C (t), we present a detailed discussion on the
physical solutions to equation (5.38). In Appendix D we show that the mere exis-
tence of a solution for (5.38) entails the presence of a horizon in such coordinates. In
Appendix E, we include the full computation of the non-diagonal stress-energy tensor
also in these coordinates.
The corresponding change of variables to and from FLRW is given by


h i3/2
6 + (κρd0 )2/3 (κρdust )1/3 R2




t=


 p
 9 2 κρdust
(5.42)

 r

 ρ d0
r= 3 R



 ρdust
with ρdust given by equation (5.32). The metric (5.41) can be linearized for ρdust → 0 to

2 κ ρdust 2 2 κ ρdust 2
ds = 1+ r dt − 1 + r dr2 − r2 dΩ2 . (5.43)
6 3
Within this linearized approximation it is of course trivial to include the black hole
gravitational field. This would give rise to

rS κ ρdust 2 rS κ ρdust 2
ds2 = 1− + r dt2 − 1 + + r dr2 − r2 dΩ2 . (5.44)
r 6 r 3
This expression is valid in the region r rS and κρdust r2 1. Note that it is not
time-independent, unlike its SdS counterpart, but the time dependence enters only
via the matter density.
In (5.42) the coordinate transformation is expressed in terms of ρdust but the de-
pendence of ρdust on T is known; see equation (5.32). We can therefore expand in
powers of ∆T = T − T0 , where T0 is the time where the density of dust is ρd0 . Lin-
earization of the coordinate transformation leads to

R2 κρd0
r
κρ
+ ∆T R2 d0 + . . .

t=T+





 2 3 12
r ! (5.45)

κρd0 κρ
− ∆T 2 d0

r= 1 + ∆T R+...




 3 12
where the scale factor a0 is taken equal to 1. Note that the initial condition T = T0
for the cosmological time does not correspond to a unique value for t as the relation
does depend on R.
5.6 Gravitational wave propagation in an analytic ΛCDM-

background
Observations show that radiation density is very small in the current epoch; recall
the radiation density parameter (5.9). If radiation is neglected, the system of Fried-
mann equations has an analytic solution for the remaining two components: nonrel-
ativistic matter and the cosmological constant.
5.6. Gravitational wave propagation in an analytic ΛCDM- background 85
5.6.1 Non-interacting background components
Another key relation can be found thanks to the Bianchi identities. This is the co-
variant conservation of the stress-energy tensor
∇µ Tµν = 0 . (5.46)
One of these four equations, the temporal one (ν = 0), prides a very useful relation
for the evolution of all background energy densities:
.
. a
∑

ρi + 3 ρi + Pi =0, (5.47)
i =1
a
where i label each component of the universe. This equation does not provide any
new information; indeed is a linear combination of Friedmann equations: deriving
..
the first Friedmann equation and replacing the a term by the second equation in
(5.4), (5.47) is recovered.
If we consider a universe with non-coupled components only, there is no ex-
change of energy-momentum among them; to wit, each of the addend in the sum
(5.46) nullifies independently
µν
∇µ T( i ) = 0 . (5.48)
This means that their interaction is purely gravitational and (5.47) holds separately
for each individual component i. These equations can be rewritten as
dρi ρ
+ 3 i (1 + wi ) = 0 (5.49)
da a
for each component i. This simplification of non-interacting components is very

useful since provides one equation for each component apart from the initial two
Friedmann equations. These new relations depends only on the state parameter of
each component, which obeys
ρ i = ρ i 0 a −3(1+ ωi ) . (5.50)
The initial scale factor at T0 has been fixed to the unity, a0 = 1.

If radiation is neglected, the two relevant component for the following discussion
are dark matter and the cosmological constant, each of them with the corresponding
cosmological parameter already written in (5.10). For the latter, where ωΛ = −1,
equation (5.50) becomes
ρΛ ≡ ρΛ0 . (5.51)
This component has been already discussed. Here energy density of the cosmolog-
ical constant can be obtained from (5.6). The second component is to very good
approximation a dust-like component [104], characterized by ωdust = 0. Therefore,
ρdust ( a) = ρd0 a−3 . (5.52)

Finally, let us consider the combined situation where dust and cosmological con-
stant are both present, and let us derive the evolution of the corresponding cos-
mological scale factor in FLRW coordinates The total density is the addition of the
density for each component; in this way the first Friedmann equation (5.4) can be
rewritten as
. 2
a0 3

a
3 = κρE f f ( T ) = κρdust ( T ) + κρΛ = κρd0 + ρΛ . (5.53)
a a( T )
ρd0 is the initial density of "dust" and ρΛ is the (constant) density of "dark energy"
when a( T0 = 0) = a0 = 1. The solution for the latter equation is analytic and gives
s 2/3
Ωm

p T p T 
a( T ) =  1+ sinh 3H0 ΩΛ + cosh 3H0 ΩΛ . (5.54)
ΩΛ 2 2
In the limit when Ωd0 → 0 we recover the one component scale factor (5.26); or if
ΩΛ → 0, we get (5.32) as it should be.
Figure 5.2 shows the corresponding cosmological scale factor together with the
effective energy density. This approximation of gravitationally interacting compo-
nents is fairly accurate for a > 0.01 or equivalently for T > 10 million years with
respect to the Big Bang. On the contrary, when radiation is added, this analytic be-
haviour would not be the case anymore, and there is no analytic solution for the
scale factor.
4
ρeff. (T)
1.5
a(T)
1.0
Effective density
Scale Factor
0.5
1
0.0
0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4
Cosmological Time T Cosmological Time T
F IGURE 5.2: Scale factor a( T ) and the effective energy density ρE f f of the 2
main components of the ΛCDM model. The vertical line corresponds to the
value T0 = 0, where a( T0 ) = a0 = 1.
5.6.2 Non-relativistic matter over an expanding background
Now that we have combined both components, dust and cosmological constant, let
us discuss the same picture but from the point of view where gravitational waves are
produced. These components are considered as non-interacting fluids, each with
such an equation of state (5.3), composing a background over where gravitational
waves propagates. As the interaction is purely gravitational, there is no exchange
5.7. Including relativistic matter 87
energy-momentum directly among them, i.e. the stress tensor for each component
is covariantly conserved separately.
The same picture in cosmological coordinates centered at the gravitational waves
source is exceedingly simple in linearized gravity, as previously discussed. The met-
ric will be
κρdust 2 Λ 2 κρdust 2 Λ 2

2 2
ds = 1 + r − r dt − 1 + r + r dr2 − r2 dΩ2 ; (5.55)
6 3 3 3
where the black hole contribution ∼ rS /r has been omitted because is not relevant
at large distances. The additive picture permits to recover the individual linearized
theory for each component (taking the corresponding limit of the other going to
zero). One would expect the same additive behaviour in the perturbations for the
change of variables, but if one analyze how the full scale factor (5.54) expands in
Taylor series,
r
Λ + κρd0
a( T ) ' 1 + ∆T + . . . , (5.56)
3
it is clear that the additivity takes place inside the square root. Indeed it is easy to
see that the following coordinate transformation does the job of moving from FLRW
coordinates to the ones in which the metric (5.55) is expressed
 r
R2 Λ + κρd0 Λ κρd0


t∼T+ + ∆T R2 + +...





 2 3 3 12
r ! (5.57)
Λ + κρd0


r = a( T ) R ∼ 1 + ∆T


 R+...

 3
Of course in both limits, ρd0 → 0 and Λ → 0, we recover each 1−component uni-

verse coordinate transformation.
5.7 Including relativistic matter

The contribution of a relativistic background when gravitational waves are propa-
gating through the universe is the weakest among the components of the ΛCDM
model in the current epoch. As discussed along the previous lines, this contribution
is almost negligible; the interest of including it is that it provides a generalization to
take into account all background energy densities existent as a whole. In this man-
ner, we will find that the results will depend only on the Hubble parameter today,
which is a direct measurement of observations, and we would be able to apply our
results to a realistic cosmology.
The procedure presented in section 5.5 can be applied in an analogous manner
to a relativistic matter universe. For this single component, characterized by a state
parameter ωrad = 1/3, the equation of state (5.3) is written as
ρrad. = 3 Prad. (5.58)
and closes the Friedmann system of equations (5.4). In comoving coordinates the
scale factor can be written in terms of different parameters
s
T
r
ρr0
a ( T ) = a0 = a0 4 (5.59)
T0 ρr ( T )
as there is a direct link between the energy density and the cosmic time. Radiation
gets diluted as
3
ρr ( T ) = . (5.60)
4 κ T2
The integration constant ρr0 = 3/(4κT0 2 ) is the energy density measured today,
when the scale factor is a0 = 1. Let us repeat the steps for non-relativistic but for
a universe with this single component. Again it is needed a set of static coordinates
with spherical symmetry such that the FLRW 2-sphere S2 = a2 R2 dΩ2 becomes a
static 2-sphere S2 = r2 dΩ2 . This requirement fixes the radial change of variables
while leave the temporal transformation unknown. The structure is the same as the
one employed in (5.34).
The next step is to change the metric representation from the cosmological coor-
dinates X µ = ( T, R, θ, φ) to the new set x µ = (t, r, θ, φ) by means of the 2-rank tensor
transformation (5.35). Doing the math, it is seen that the diagonal angular compo-
nents of the metric transform trivially and that any non-diagonal component with
at least one angular index remains null. A diagonal metric in the new coordinates is
obtained if the following identity is satisfied
4 κρr 2 r 6 + 2 κρr r2
∂r ρr = =⇒ √ = C (t) = A t . (5.61)
3 − κρr r2 κρr
The energy density ρr is a function of the new variables t and r. According to the
first integral of the equation on the left, the integration constant must depend only
on the temporal coordinate. Dimensional analysis provides its structure: the units
goes as [C ] = [κρr ]−1/2 = L; therefore, using equivalent reasons as the ones dis-
cusses around equation (5.38), any integration constant for r can acquire units only
by means of t, hence C (t) = At.
The diagonal condition provides a metric which is almost unequivocally defined,
except for a global coefficient in the temporal component of the metric involving the
temporal derivative of the radiation density
2 ! ! −1
2 3 ∂ t ρr κρr r2 2 κρr r2
ds = 1− dt − 1− dr2 − r2 dΩ2 . (5.62)
16 κρr 3 3 3
Of course this metric resembles the metric of the non-relativistic case obtained in
5.7. Including relativistic matter 89
(5.37) but with different numerical factors. The temporal derivative can be obtained
deriving the right equation in (5.61)
√
κρr ρr
∂ t ρr = − A . (5.63)
3 − κρr r2
One more time, The homogeneously diluted limit for later times, t → ∞, pro-
vides the missing constant A. The requirement for the metric to be Minkowskian in
this asymptotic regime, gαβ → ηαβ while ρrad. → 0, is fulfilled for all the components
of the metric except for the temporal one. We need to impose (∂t ρr )2 /(κρr 3 ) → 16/3
√
in gtt from equation (5.62). Accordingly, A = 4 3. After all, we have fixed the
density relation, this is
3 + κρr r2 = 2
p
3 κρr t . (5.64)
Inserting (5.63) in (5.62) with the value of A, we obtain the metric representation
in some static and spherically symmetric coordinates
! −1 ! −1
2 κρr r2 2 κρr r2
ds = 1− dt − 1− dr2 − r2 dΩ2 . (5.65)
3 3
The change of variables back to the cosmological coordinate system is given by


κρr0 R2
r
2

t = T + a0


3 2



r (5.66)

 ρr0
r = a0 4 R,


ρr ( T )


where the dependence of the density on the cosmological time has been written in
(5.60). Let us mention that time dependence enters only via the energy density of
the relativistic matter.
It is enough for our purposes a linearized version of (5.66). The linearized metric
for ρrad. → 0 from (5.65) is
! !
κρr r2 κρr r2
ds2 = 1+ dt2 − 1+ dr2 − r2 dΩ2 (5.67)
3 3
and the linear change of variables, expanded in powers of ∆T = T − T0 , results


κρr0 R2
r
2

t ∼ T + a0 +...





 3 2
r ! (5.68)

 κρr0
r ∼ a0 1+ ∆T R+...




 3
p
The magnitudes measured today are related by 1/T0 = 2 κρr0 /3 .
5.8 Gravitational wave propagation in a general background

Although the contribution of radiation as a background component is almost negli-
gible and worse, it cannot be included in an analytical expression as found in (5.54),
it is interesting to include its effect anyway for reasons that will be seen below.
At this point where the linearized framework is a fair approximation, the inclu-
sion of radiation in (5.55) becomes obvious,

2 κ
2ρΛ − 2ρr − ρnr r + ... dt2
2
ds = 1 −
6
(5.69)
κ
ρΛ + ρnr + ρr r + ... dr2 − r2 dΩ2 .
2
− 1+
3
In the later equation we favour the density of dark matter over the cosmological
constant. Observations measure the Hubble parameter
p
H ( a) ≡ H0 Ωdm a−3 + Ωrad a−4 + ΩΛ (5.70)
rather than the separate contributions from the cosmological constant and the var-
ious matter densities. Its value at a time T0 when a0 = 1 is a measure of the total
energy density of the background on which gravitational waves propagate6 ; known
as the Hubble constant is
κ
H0 2 = ( ρ Λ + ρ d0 + ρ r 0 ) . (5.71)
3
–recall the discussion below equation (1.21).

Well away from the sources, in the limit where rS /r ρi , the black hole con-
tribution has been be safely neglected in equation (5.69), and the linearized picture
holds. In this approach, the gravitational wave would propagate over an expanding
spacetime, characterized by a scale factor approximately given by
a( T ) ' 1 + H0 ∆T + . . . (5.72)
The propagation should take place in coordinates where the expansion is explicit.
To change to those cosmological coordinates, is enough to use the linearized change
of variables given by 
 R2
t∼T+ H0 + . . .


2

(5.73)

r ∼ (1 + ∆T H0 ) R + . . . .



In conclusion, the harmonic wave front (5.14) can be transformed up to linear

order in the Hubble parameter, to expanding FLRW coordinates by means of (5.73).
Therefore, choosing T0 = 0 at the time when gravitational waves are emitted, the
6 In cosmology, its value is conventionally chosen as T0 = 0 in the current epoch.
5.8. Gravitational wave propagation in a general background 91
wave perturbation gets modified to

2
eµν

R
hµν ∼ (1 + H0 T ) cos w ( T − R) + w H0 − TR , (5.74)
R 2
√
keeping only changes of order Λ. eµν is the transformed polarization tensor that
will not be very relevant for the following discussion, although its precise form is
essential for precise comparison with observations. From this last expression we can
identify the effective frequency and the effective wave number
we f f = w (1 − H0 R)
(5.75)
R
ke f f = w 1 − H0 ;
2
which do not coincide.

In astrophysical observations, it is common to use the cosmological redshift z
for characterizing the relative changes between the observed we f f and emitted w
frequencies (or wavelengths) of distant objects. This magnitude is defined by
w
we f f = (5.76)
1+z
and at these scales, the redshift is very well approximated by the Hubble relation
z ' H0 R. Hence the observed frequency we f f is very well approximated by equa-
tion (5.16). We notice that the obtained frequency in (5.75) agrees with the usual
frequency redshift (5.16), or (5.76); so this well known result is well reproduced.
However the wave number is different. This discrepancy lies at the root of a possi-
ble local measurement of Λ in PTA, as we will discuss in next chapter.
Before moving on, let us draw some conclusions about the linearized results ob-
tained. Along this chapter, we have discussed extensively how gravitational waves
are modified when they propagate over a non-flat background, but over an expand-
ing one. If comoving coordinates are employed during the propagation, we have
shown that modifications to the gravitational waves are induced. Simultaneously,
the wavelength and the amplitude are increased with the radial coordinate R that
measures the distance to the source. There is also a discrepancy between the effec-
tive frequency we f f and the effective wave number k e f f ; see the equations in (5.75).
The study of the relevant coordinate transformations make us arrive to (5.73), where
linear modifications due to the effective background over where waves propagate
have been taken into account. The magnitude describing the energy density compo-
sition of the background is given by the Hubble parameter today (5.71) whose value
is dominated by the cosmological constant. This parameter provides the linear cor-
rections to the theory. Measuring these corrections at moderate values of the redshift
z, would immediate imply that we are measuring locally the cosmological constant.
The above calculations have been performed at a linear level, retaining O( H0 R)
terms7 , and considering the corrections of order O( H0 2 R2 ) and beyond as sublead-

ing. This truncation can be justified based on the smallness of H0 (and of all its
ingredients, of course) since we will be interested in sources at distance R H0 −1 .
This approximated picture has a range of validity. Given the current observed value
of the cosmological constant (5.4), for events happening at most at a distance of a
few Gpc away (R . 1025 m) modifications of O( H0 R) . 10−1 appears to be non
negligible. In any case, the expressions could be improved to order H0 2 accuracy
easily.
7 We
emphasize once more that at the next, non-linear, order the corrections cannot be all grouped
in terms of the Hubble constant, so when we speak of O( H0 2 R2 ) corrections this has to be understood
in the above sense.
93
Chapter 6
Local measurements of Λ with

pulsar timing array
6.1 Pulsars as gravitational wave detectors

A pulsar is a highly magnetized rotating neutron star that emits beams of electro-
magnetic radiation. They are born in supernova explosions, usually with periods
ranging between 10 − 20 ms and later on slow down gradually to periods of order
∼ 1 s over millions of years [105]. When the beam of emission points towards us,
we see an effect much like a lighthouse. Since its first discovery in 1968 [106] radio
telescopes have been monitoring repeatedly the corresponding pulsed appearance
of emission.
Among the pulsar population, the so-called millisecond pulsars (MSPs) are of
special interest to us. They have periods less than 100 ms and a period derivative
less than 10−17 . This means that they are very stable and regularly rotating stars.
The leading theory for such objects dates from 1982 [107] and states that the MSPs
result from the "recycling" of an old, slowly rotating and probably dead neutron
star through accretion from a low-mass companion. This accretion results in a mass
transferred from the companion that also carries angular momentum from the orbit
to the neutron star, spinning it up and reactivating the pulsar emission process. In
some cases the complete ablation of the companion star occurs. This is a viable
mechanism for the formation of solitary MSPs. However, globular clusters are much
more efficient factories for the production of MSPs.
Owing to their extraordinarily rapid and stable rotation, MSPs are usually used
by astronomers as galactic clocks rivaling the stability of the best atomic clocks on
Earth. Any factor affecting the arrival time of pulses by more than a few hundred
nanoseconds can be easily detected and used to make precise measurements. The
process of pulsar timing is fundamentally dependent on how accurate the descrip-
tion of everything that affects the times of arrival (ToAs) of the pulsed radiation at
the telescope is. Then, there are "timing models" that take into account all orbital,
location and motion parameters of the pulsars and of the interstellar medium along
where radio waves propagate, to predict the phase of the pulsar’s periodic signal at
any point in time. When a gravitational wave passes it disturbs everything around
94 Chapter 6. Local measurements of Λ with pulsar timing array
it, and it does it over the observed periods of pulsars. The gravitational waves de-
tection with pulsars depends on the great stability of the MSPs.
Even the more stable pulsars present intrinsic irregularities in their period at
some point. Therefore, observations of an ensemble of MSPs (this is a Pulsar Tim-
ing Array) are needed for gravitational wave detection. The observational method
is based on searching for correlated signals among pulsars in a PTA. A disturbance
from a passing gravitational wave across the ensemble of pulsars will have the par-
ticular quadrupolar spatial signature that corresponds to gravitational waves [108].
If it passes, it will thus be detected. In practice, the analysis consists in computing the
difference between the expected and actual ToA of pulses. This difference is called
"timing residuals" and contains information of all unmodeled effects that have not
been included in the fit, including gravitational waves [109].
Nevertheless, PTA observations have not yet found convincing evidence of grav-
itational waves. In the next section we will rederive in an alternative way the main
results obtained in the previous chapter. This time without moving from different
coordinate systems but just computing the wave equation directly in FLRW expand-
ing coordinates, and showing that the wave solution is in fact nothing less than the
same solution found before. As it was pointed out at the end of section 5.8, the so-
lution develops some anharmonic effects because propagation takes place through
an expanding background. Although globally very small, these anharmonic effect
result in a dramatic enhancement of the signals expected in PTA observations. How-
ever, the effect only takes place at some particular values of the angle subtended by
the source and the observed pulsar. In section 6.3 we will compute the timing resid-
ual numerically and show its main peak-like characteristics. As can be expected, if cal-
culations are performed in a Minkowskian background the well-known oscillating
signal with no enhancement is obtained. However when a dynamical background
is considered the anharmonicities appear. After this, in section 6.4 we will charac-
terize the signal; particularly which of the cosmological parameters or characteristic
distances modifies the position or width of the peak. To conclude, we will expose
in what manner these results can be tested by the IPTA collaboration. We will pro-
pose a way to hopefully detect this very relevant effect; which if present would also
provide a way to measure the Hubble constant H0 but at subcosmological distances.
6.2 Linearized perturbations in FLRW

As we have discussed in Chapter 5 the gravitational wave propagation takes place
not over a flat spacetime, but over an approximately globally de Sitter universe. In
[51] linearized gravity in curved backgrounds has been discussed but from the point
of view of a cosmological observer. Applying the same strategy of linearization on
a curved background, we consider small perturbations around a FLRW background
metric (5.15) of the form
( FLRW )
gµν = g̃µν + hµν . (6.1)
6.2. Linearized perturbations in FLRW 95
These perturbations are interpreted as gravitational waves propagating on this curved

background in analogy to the treatment on a flat spacetime; recall section 5.3.
The linearized equations of motions are provided by the Einstein’s equations and
are given by the following expansion

δGµν
(0) (1)

Gµν ( g̃ + h) = Gµν ( g̃) + hαβ + . . . = κ Tµν + Tµν + ... . (6.2)

δgαβ
g̃
Clearly, these equations are satisfied for the unperturbed FLRW background metric.
(0)
At zero order we have Gµν ( g̃) = κTµν , which in fact gives the well-known Fried-
mann equations by treating the energy content of the universe as perfect fluids. The
energy-momentum tensor includes as a matter of fact all cosmological fluids com-
posing the background, each one described by some equation of state (5.3). Then,
at first order, we end up with the following expression as a motion equation of the
perturbation tensor
δGµν
(1)
hαβ = κ Tµν . (6.3)

δgαβ
g̃
Let us make some reasonable assumptions to avoid redundancies under coor-

dinate transformations and go straightforward to the gravitational wave profile in
FLRW coordinates. In this new set of coordinates the spatial character of the grav-
itational wave is not expected to change. Besides, it is reasonable to expect that in
spherical coordinates gravitational waves propagate along the radial direction away
from the source, along a vector given by kµ = (w, k R , 0, 0). Then transversality to the
R−coordinate implies that the only non-vanishing h−components are the purely
angular ones, i.e.
h T α = h Rα = 0 (6.4)
for α = ( T, R, θ, φ). Moreover, the traceless nature of gravitational waves should not
change; hence the condition h = g̃µν hµν = 0 relates the diagonal components of the
metric perturbation by
hφφ = − sin2 θ hθθ . (6.5)
Finally, symmetry conditions imply
hθφ = hφθ . (6.6)
All in all, there are two nonzero independent components in agreement with the two
degrees of freedom corresponding to the two polarization directions of gravitational
waves.
After imposing these gauge conditions, the two equations for hφφ and hθφ coming
from (6.3) are rewritten up to first order in the perturbations as
" 2 #
R2 d

a d 1 d 1 d 1 ȧ 3 ä (1)
− 2 − 2 2− − hµν = κ Tµν ; (6.7)
2 dT a dT 2a dR R2 dR R a a a
where µ, ν = θ, φ. We notice that the equations for the two independent metric
components have exactly the same functional form. The previous discussion is valid
far away from the source, where gravitational waves propagates freely along the
expanding universe. Although equations (6.7) are valid up to order O(Λ), we will be
√
mostly interested in terms of order O( Λ) due to the smallness of the cosmological
constant value, and also to be able to compare with the previous chapter. Therefore,
(1)
in the right hand side of the latter equation, we can get rid of the term κ Tµν because
is linear with the densities and therefore of order O(Λ). Of course there is nothing
wrong in keeping this term (we will return to this later).
Gravitational waves are produced by very distant sources. In this manner, in the
far away field regime, when waves arrive to our local system, they can be described
very approximately by a plane wave front. A more useful coordinate system for
doing this analysis is the Cartesian set of coordinates;

ds2 = dT 2 − a( T )2 dX 2 + dY 2 + dZ2 . (6.8)
Let us define the radial direction joining the SMBH merger with the Earth as the
propagation direction for these waves; let us call it the Z −axis of the Cartesian coor-
dinates. For local observers, the usual treatment on linearized gravity applies very
well for the current discussion. In the transverse (hµα kα = 0) and traceless (hα α = 0)
gauge (TT), for a wave propagating with wave vector kµ = (ω, 0, 0, k Z ), the nonzero
components are hXX , hYY , hXY and hYX , however only two of the four are indepen-
dent. It is a customary to express the solution in terms of the "plus" polarization
h+ = (hXX + hYY )/2 and the "cross" polarization h× = hXY = hYX .
In this manner, gravitational waves coming from a very distant source are equiv-
alently described by Cartesian coordinates or correspondingly to spherically sym-
metric coordinates but in the very small polar angle case, θ ≈ 0. The relation be-
tween components is given by
1 1 cos θ
hXX = + cos(2φ) hθθ − 2 sin(2φ) hθφ
R2 R sin θ
1 1 cos θ
hYY = − 2
cos(2φ) hθθ + 2 sin(2φ) hθφ (6.9)
R R sin θ
1 1 cos θ
hXY = + sin(2φ) hθθ + 2 cos(2φ) hθφ .
R2 R sin θ
Note that the apparent singularity in the prefactor of hθφ has to be compensated
by the appropriate vanishing of this component. The characteristic signature of the
gravitational waves on the ToAs of radio pulses is a linear combination of these two
independent polarizations; see for instance [110]. Nevertheless
By direct substitution of hφφ (∝ hθθ ) and hθφ (= hφθ ) in the small angle limit θ ≈ 0,
6.3. Relevance for Pulsar Timing Arrays 97
we can easily reach the corresponding equations (6.7) for the Cartesian metric pertur-
bations but written in spherical coordinates, since the above differentials equations
do not involve angular derivatives and only the radial derivative plays a role in this
coordinate transformation
" #
ȧ 1 2
∂2T − ∂T − 2 ∂2R + ∂R hij = 0 ; (6.10)
a a R
where now ij = +, ×. In the derivation of these equations we have neglected O(Λ)

terms, as we mentioned before. We can understand this differential equation as the
wave equation for a universe with nonnull matter content; this is
g̃ hµν = 0 , (6.11)
which yields over the two polarization directions. In the limit of an empty universe,
with no cosmological constant or dust, the scale factor a( T ) can be taken to be a0 = 1,
so equations (6.10) reduce to a homogeneous wave equation
η hµν = 0 , (6.12)
as expected for the flat spacetime limit.

As expected, it is not (5.14) the solution of (6.10), because it solves the flat back-
ground wave equation (6.12). The solution for (6.10) or equivalently to (6.11) is
eµν h i
hµν = (1 + H0 T ) cos we f f T − k e f f R ; (6.13)
R
where the effective frequency and effective wave number have been introduced in
(5.75). This solution is nothing more that the previous obtained solution after chang-
ing coordinates with (5.73), from the set where waves are produced to the one where
they propagate and are measured. This solution has been presented in equation
(5.74). As we have seen in the previous chapter, if the universe were static, the co-
ordinate transformation (5.73) would simply become t = T and r = R, and the
corresponding solution is a superposition of harmonic waves of the form written in
equation (5.14). However now we have derived the anharmonic solution directly in
expanding coordinates. The phase velocity when gravitational waves propagate in
expanding coordinates is v p ∼ 1 − H0 ∆T + O( H0 2 ) [49]. On the other hand, with
respect to the ruler distance traveled (computed with gij ) the velocity is still 1.
6.3 Relevance for Pulsar Timing Arrays

From the previous discussion it is clear that one has to understand what are the
consequences of having k e f f 6= we f f . As discussed before, PTA observations involve
some non-local contributions as the signal is integrated along the line of sight joining
the pulsar and the observer. Therefore, the fact that H0 in (6.13) is nonzero may
render these contributions observable at some point.
Let us set the experimental framework that might measure this effect. Consider
the situation shown in Figure 6.1 describing the relative situation of a gravitational
wave source (possibly a very massive black hole binary), the Earth and a nearby
pulsar. The pulsar emits pulses with an electromagnetic field phase φ0 . If this mag-
nitude is measured from the Earth, it will be
h i
L
φ( T ) = φ0 T − c − τ0 ( T ) − τGW ( T ) . (6.14)
The corrections due to the spatial motion of the Earth within the Solar system and
the solar system with respect to the pulsar, together with the corrections when the
electromagnetic wave propagates through the interstellar medium are taken into
account in τ0 . The term τGW involves the corrections owing exclusively to the gravi-
tational wave strength hij(FLRW ) computed in comoving coordinates.
In the following we will focus in the phase corrections due to the passage of a
gravitational wave; this shift is given by [111]
1
τGW = − n̂i n̂ j Hij ( T ) . (6.15)
2
The unit vector giving the direction can be written using spherical coordinates as
F IGURE 6.1: Relative coordinates of the gravitational wave source (R = 0),

the Earth (at R = ZE with respect to the source) and the pulsar (located at
coordinates ~P = ( PX , PY , PZ ) referred to the source). In FLRW coordinates
centered on Earth, the angles α and β are the polar and azimuthal angles of the
pulsar with respect to the axis defined by the Earth-source direction.
n̂ = (− sin α cos β; − sin α sin β; cos α), where α is the angle subtended by the grav-
itational wave source and the pulsar as seen from the Earth and β is the azimuthal
angle around the source-Earth axis. Hij is the integral of the transverse-traceless
metric perturbation along the null geodesic from the pulsar to the Earth. Let us
chose the following parametrization:
~R( x ) = ~P + L(1 + x )n̂ , (6.16)
with x ∈ [−1; 0], L the distance to the pulsar and ~

P the pulsar location with respect
to the source. Making the dependencies explicit, this magnitude is given by
Z 0
L h i
Hij = dx hij TE + L
c x ; ~R( x ) . (6.17)
c −1
TE is the time of arrival of the wave that travels a distance ZE to the local system.
The speed of light has been restored in this formula. In the TT–gauge, a gravitational
wave propagating through the Ẑ axis has a polarization tensor with only nonzero
values in the X, Y components; or equivalently in the + and × components. Then
we can simply write
h i
n̂i n̂ j hij = sin2 α (cos2 β − sin2 β) h+ + 2 cos β sin β h× , (6.18)
where h+ and h× are the two nonnull polarizations of the gravitational wave ob-
tained in (6.13). For simplicity, we can also assume ε ∼ |eij | for i, j = +, ×. This
approximation does not affect in any significant way the results below, and in fact it
can be easily removed.
We have assumed that from the pulsar to the Earth the electromagnetic signal
follows the trajectory given by the line of sight and is the distortion created by
the intervening gravitational wave what give rise to the timing residual. The real
question is whether the observationally preferred small value of the cosmological
constant affects this timing residuals from a pulsar at all. This question was an-
swered in the affirmative in [58]. In order to see the magnitude of the effect we
take average values for the parameters involved: ε = 1.2 × 109 m for a gravita-
tional wave generated by some source placed at ZE ∼ 0.1 − 1 Gpc, giving a strength
|h| ∼ ε/R ∼ (10−16 − 10−15 ) which is within the expected accuracy of PTA [112].
Monitored MSPs are usually found at a distance of the order of L ∼ 0.1 − 10 kpc.
For the distance to the source we take values in the range Z = 100 Mpc − 1 Gpc
and we assume the value w = 10−8 Hz for the frequency. These, together with the
preferred value of H0 are all the parameters we need to know.
In formula (6.13) enters the modulus of the parametrized trajectory ~R( x ). It is
reasonable to assume that L ZE , then
R( x ) = |~R( x )| ∼ ZE + x L cos α + . . . (6.19)

The parametrization for the temporal variable has already been presented in terms
of the time of arrival of the wave to the local system TE , this is
L
T = TE + c x. (6.20)
On geometric grounds, we are always able to choose a plane containing the Earth,
the pulsar and the source and set β = 0. This requirement simplifies considerably
the contraction (6.18). After all these considerations, in order to obtain the timing
residual we need to perform the following integral
1 + H0 ( TE + x Lc )
Z 0
(ΛCDM) Lε
τGW =− sin2 α dx cos(w Θ) ; (6.21)
2c −1 ZE + x L cos α + . . .
where the argument of the trigonometrical function (less than a global phase) is
given by
h i
TE
Θ≡x L
c (1 − cos α) − H0 ( TE + x L
c cos α ) 2 +x L
c 1− cos α
2 . (6.22)
Snapshots of the resulting timing residuals as a function of the angle α can be ob-
tained for different values of the Hubble constant (5.71). In Figure 6.2 we compare
different universes characterized by various values of H0 (when different densities
are taken into account). The distance to the source and the gravitational waves fre-
quency are always chosen to be ZE = 100 Mpc and w = 10−8 Hz respectively, except
when specifically indicated. The figure speaks by itself and it strongly suggests that
the angular dependence of the timing residual is somehow influenced by the value
of H0 , whether this value is due to a cosmological constant, radiation and/or dust.
The feature that catches the eye immediately is an enhancement of the signal for a
specific small angle α, corresponding generally to a source of low galactic latitude or
a pulsar nearly aligned with the source (but not quite as otherwise eij0 n̂i n̂ j = 0). In
fact, as it is observed, the curves are extremely sensitive to small changes in the H0
parameters.
Surprising as it is, this enhancement is relatively easy to understand after a care-
ful analysis of the integral in equation (6.21) (analogous results as in [49] have been
obtained but for the Hubble parameter). In Figure 6.3 we plot the differential tim-
ing residual in the vertical axis, versus the subtended angle α and versus the line of
sight to the pulsar x. We show as a comparison the signal propagating through a flat
background on the left and through an expanding universe due to a cosmological
constant on the right. The change occurs around the angle where the enhancement
is located; dτGW registers a constant value along all the line of sight. This corresponds
to a valley along where the phase is nearly constant. The location of this stationary
phase path seems to be the responsible of the enhancement at that particular angle
α. As we will see further on, in the discussion around Figure 6.4, the position of this
valley strongly depends on the value of the Hubble constant.
The effect under discussion is largely degenerated with respect to the details of
the source of gravitational waves. Not much depends on the polarization or fre-
quency of the wave front (see e.g. [58]). This is so because the enhancement is
entirely a consequence of the characteristics of the FLRW metric and it cannot be
mimicked by changing the frequency or the amplitude of the gravitational wave,
or its polarization. All these changes would enhance or suppress the signal for all
angles, not a restricted range of them. The phenomenon described is thus indeed a
telltale signal. However, the curves shown in Figure 6.2 are only of theoretical inter-
est, to prove that the effect exists. What one needs to do is to define an observational
protocol.
The observable defined in (6.15) is in fact not necessarily the preferred one in PTA
collaborations. Typically, PTA use two-pulsars correlator and use the assumption
that the gravitational wave signal in the location in the two pulsars are uncorrelated
(see Hellings and Downs [108] and Anholm et al. [113]). This is so because it is
assumed that stochastic gravitational wave prevails and that a single event leading
to the direct observable presented here would normally be too small to be observed
Minkowskian universe radiation universe

1 1
timing residual ( τ x 10-7 )
0 0
-1 -1
α
α
-2 -2
-3 -3
0 π/2 π 0 π/2 π
angle ( α ) angle ( )
dust universe de Sitter universe

1 1
)
-7
timing residual ( τ x 10
0 0
-1 -1
α
α
-2 -2
-3 -3
0 π/2 π 0 π/2 π
angle ( α ) angle ( α )
F IGURE 6.2: Raw timing residual as a function of the angle α subtended by the
source and the measured pulsar as seen from the observer. The figures repre-
sent a snapshot of the timing residual at a given time TE in flat space time, in
a universe with only relativistic matter, with only non-relativistic matter, and
in a de Sitter universe. Clearly, at least in a superficial analysis, it seems plau-
sible to be able to disentangle both contributions. The figures depict the signal
at the time TE = ZE /c. Time evolution or small changes in the parameters
will change the phase of the oscillatory signal but not the enhancement clearly
visible at relatively low values of the angle α. The figures are symmetrical for
π ≤ α ≤ 2π.
directly. Unfortunately, the averaged pulsar-pulsar correlator when supplemented

with the uncorrelation hypothesis normally used, likely turns into a local observable.
Local observables are not affected by modifications in the wave number. One needs
to be able to measure the gravitational waves in two widely separated points for
the wave number to be relevant. Otherwise only the frequency of the gravitational
waves is relevant and as we have seen this is not at all varied with respect to the
usual treatment. It is also for this reason that this effect is totally invisible in an
interferometric experiment such as LIGO.
It is thus essential to validate the effect observationally to consider single events
observations, also in PTA, rather than looking for stochastic gravitational wave back-
ground. It is true that the latter may dominate, but the effect reported here brings
such a relevant enhancement that the search technique needs to be changed. In the
following sections we will provide a detailed characterization of the signal as well
as its dependence on various parameters.
F IGURE 6.3: In the fist two plots we compare differential timing residual com-
puted in Minkowski and in an expanding background given by H0 in (5.71).
We can observe the evolution of the phase along the line of sight from the pul-
sar (x = −1) to the Earth (x = 0), for different subtended angles α. The code
colours goes from the bluish for the more negative values of dτGW , to reddish
for the higher ones. Around the angle of enhancement αmax ∼ 0.2 it can be
seen that the signal takes a constant value along the line of sight (in green).
The third plot is the same as the second one, but rotated in such a way that the
valley of stationary phase is obvious.
6.4. Characterization of the signal 103
6.4 Characterization of the signal

Let us begin by taking into account the influence of all the components in the ΛCDM
model: relativistic and nonrelativistic matter together with the cosmological con-
stant. From its definition in equation (5.71), the Hubble constant H0 represents the
current rate of expansion of the universe and contains information of its compo-
sition. Technological progress produced a qualitative change in the observational
paradigm in recent decades. The uncertainty decreased significantly, making H0
one of the better measured cosmological parameters. However, there is still a great
controversy about its actual value. As we know, there is an apparent incompatibility
between ‘local’ measurements using standard candles [34] and ‘cosmological’ ones
performed with the CMB spectrum [33]. This fact has generated intriguing ques-
tions regarding the local behaviour of the Hubble constant and an intense debate
is still ongoing. The discrepancy on H0 is more than 4σ but less than 6σ; strongly
suggesting that it is not dependent on the method, team, or source. In fact, if lucky,
our analysis could provide an alternative method of measuring H0 locally, on top of
making PTA observations a lot easier.
6.4.1 Dependence with the source parameters
Now we would like to make a more detailed characterization of the possible signal.
We analyze how the signal changes when the main characteristics of the gravita-
tional wave front are modified.
Dependence with the distance to the source
In Figure 6.4 we plot the modulus of the timing residual (6.21) for three different
reported values of the Hubble constant. The angular position for the maximum,
αmax , is significantly dependent on the Hubble constant. In fact, it is possible to
derive an approximate formula relating the angle α (the effect is independent of the
angle β) to the value of H0 :
2c α
max

H0 ' sin2 . (6.23)
ZE 2
An analogous relationship has been discussed in [58], but when gravitational waves
propagates in a purely de Sitter universe. Taking a fixed distance to the source
ZE ∼ 100 Mpc, the authors verify in this way the validity of an analogous approx-
imate analytical expression to equation (6.23), but relating the angular position of
the maximum and the (local) value of the cosmological constant, i.e. Λ(α). This re-
lation is extremely interesting in two ways: on the one side, this prediction could
be eventually tested experimentally and may facilitate enormously the detection of
gravitational waves produced in SMBH mergers. On the other side, and the most
interesting for us, this relation may provide an indirect way to measure the Hubble
constant and so achieve a local manner to detect the cosmological constant at sub-
cosmological scales (once the distance to a SMBH binary ZE and the angular location
of the enhancement αmax are known). It is particularly interesting to measure H0 as
locally as possible as its main component Λ/3 is assumed to be an intrinsic property
of the spacetime, present to all scales, something that needs to be proven.
Obviously, for a fixed value of H0 (assumed to be the one reported in [34] from
now on), we can test the dependence of the effect on the distance to the source, and
also compare the results to equation (6.23). In Table 6.1 we have tested this relation
for a range of validity of the source location in the corresponding approximation. We
have selected the pulsar J2033+1734 located at L ∼ 2 kpc (all distances to the pulsars
used along our study are presented in Table 6.2). Leaving the pulsar fixed, we vary
the distance to the source for three different equally spaced values. We compare the
angle of the peak obtained from analogous plots as the one presented in Figure 6.4,
called α graphical , and the same angle computed by the approximate formula (6.23),
named αtheoretical . There is a nice agreement between both ways of finding the an-
gular location of the maximum. These results have been checked independently in
[114].
ΛCDM universe
3
2
α
0 π/ 2 π
0
π / 30 π / 10 π/6
angle ( )
F IGURE 6.4: Absolute timing residual |τGW | for three different values of the
Hubble constant: H0 = 73.24 ± 1.74 km s−1 Mpc−1 reported in [34] (green
dashed curve), H0 = 67.4 ± 0.5 km s−1 Mpc−1 from [33] (red dotted curve) and
H0 = 60 ± 6 km s−1 Mpc−1 [115] (blue curve). We also include the propaga-
tion in Minkowski in thick black line. The remaining cosmological parameters
employed for the integration do not vary from the ones used in the computa-
tion performed in Figure 6.2. The small plot contain the four signals shown in
their whole angular range for a comparison.
Source distance Strength α graphical αtheoretical Deg

100 Mpc 1.71 × 10−6 0.2125 / 0.2130 0.2101 ∼ 12◦
500 Mpc 1.60 × 10−6 0.4716 / 0.4720 0.4735 ∼ 27◦
1 Gpc 1.67 × 10−6 0.6748 / 0.6751 0.6761 ∼ 39◦
TABLE 6.1: Comparison of the angular location of the maximum where the
pulsar J2033+1734 would experience the enhancement on its timing residual.
We compare the angle αmax computed in two different ways, graphically by
means of numerical integrations as the one performed in Figure 6.4 and with
the formula (6.23). As can be seen the accuracy is quite remarkable.
Dependency on the wave frequency
From the previous analysis we have seen that the maximum turns out to be mod-
erately stable under changes of the parameters involved in the gravitational wave
production. Let us now focus on the dependency of the signal when the frequency
of the gravitational waves change in the range of the characteristic values where the
PTA experiment is expected to be performed. In Figure 6.5 the absolute value of
S
at 100 M S
at 100 M
2 w = 1 nHz 2 w = 10 nHz
)
6
-
w = 1.3 nHz 6
-
( τ x 10
( τ x 10
1.5 1
w = 5 nHz

α
α
1 1

r r
timing
timing
0 0
0 0
0 π/2 π 0 π/2 π
angle ( α ) angle ( )
S
at 100 M S
at 100 M
2 w = 100 nHz 2 w = 10 nHz

)
6
-
6
-
w = 50 nHz
( τ x 10
( τ x 10
1 1
w = 100 nHz

α
1 1

r r
timing
timing
0 0
0 0
0 π/2 π 0 20 0 21 0 22 0 2 0 2
F IGURE 6.5: All curves show monochromatic gravitational waves with the
frequency mentioned in the plot. The first graph shows three nearby frequen-
cies. We can see that the signal decreases quite fast its main value when the
frequency is increased in some decimals. On the contrary, the peak remains
nearly at the same angular position. In the second and third graphs we in-
crease the frequency in 2 orders of magnitude. Accordingly to the first plot, the
signal decreases even more while the peak does not. In the last plot we show
a zoom of the two previous curves with w = 10 nHz and w = 100 nHz, where
it is clearly seen that the signal stretches when the frequency is increased. The
same compression suffers the peak.

the timing residuals for the angular coordinate τGW (α) is plotted for different val-
ues of the frequency. This magnitude would be registered by the pulsar J2033+1734
located at 2 kpc.
In the fist graph we superpose three different harmonic functions with slightly
different values for the frequency: w = 1 nHz, w = 1.3 nHz and w = 5 nHz. For
values near ∼ 1 nHz the peak already doubles the rest of the signal. If we slightly
increase the frequency by a few decimals, the signal falls while the peak remains
stable with nearly the same height and location. We chose these values to show how
fast the decay results: increasing only 3 decimals in the frequency the signal falls by
half. This pattern remains valid for all frequencies tested. At w = 5 nHz the peak is
already one order of magnitude higher than the rest of the signal. It is also noticeable
that peak becomes narrower as the frequency increases. The fourth plot shows a
zoom on the enhancement region for the latter two figures. Here again it is clear
how the signal, peak and oscillations, stretch and fall while the peak remains at the
same place with the same height. Sometimes two peaks next to each other appear;
however, as can be seen in the signal computed with w = 50 nHz (green curve),
the value of the maximum is still approximately one order of magnitude above the
signal. In conclusion, the height of the signal turns out to be nearly constant as the
frequency of the gravitational wave varies.
It is interesting to explore signals consisting in the superposition of several fre-
quencies. In Figure 6.6 the absolute value of the timing residuals are computed em-
ploying bichromatic waves. The main characteristics of the signal remain the same if
we plot for the full angular range α ∈ [0, π ]. We can see an oscillating signal around
zero and a remarkable peak always at the same location. It turns out that the peak
seems to depend only on the value of the Hubble constant and on the distance to the
source. A closer look reveals that the peak suffers several deformations when two
or more frequencies are taken into account, but it is always present. In the figure
Source at 100 Mpc Source at 100 Mpc
10 nHz + 20 nHz 2 50 nHz + 100 nHz

3
-6 -6
)
10 nHz + 30 nHz 5 nHz + 50 nHz

10
10
x x
2
(τ
(τ

α
1
r r
1

t t
0 0
0.16 0.18 0.20 0.22 0.24 0.26 0.18 0.20 0.22 0.24 0.26
angle ( ) angle ( )
F IGURE 6.6: Absolute timing residual |τGW (α)| for non-monochromatic gravi-
tational waves. In the first plot be combine two frequency with nearer values.
It is seen that the peak suffers some modifications if its height. In the sec-
ond plot, the frequencies differ in an order of magnitude and the enhancement
zone becomes wider. The principal peak remains normally in the same posi-
tion.
we show two combinations of two frequencies each; one combines relatively similar
frequencies, while the other makes a superposition of rather different ones. In sum-
mary, the main peak of the signal remains at the same location when frequencies are
varied or combined. However, the combination of frequencies has a desirable effect
as it enlarges the range of angles where the enhancement is visible, and this is quite
welcome when it comes to a possible detection of the effect.
Time dependent amplitudes for the gravitational waves
Let us assume a varying global factor for the wavefront such that
dτGW → A( T ) dτGW . (6.24)
dτGW is the differential timing residual already introduced in the discussion regard-
ing Figure 6.3. The next step is to give a structure to the form in which the amplitude
varies. Let us impose two different behaviours: a signal modified by a very slowly
varying exponential and by means of a linear varying amplitude. For the former, we
would have
 " 
c #
L
A( T ) = exp T − TE − = exp (1 + x ) . (6.25)
L c 
For the second equality we have used the parametrization of the Earth-pulsar path.
Then we perform the numerical integration as in (6.21), with the argument of the
trigonometric function given by (6.22), but taking into account the new varying am-
plitude. In such manner, when the gravitational wave arrives to the pulsar location,
the amplitude equals 1 in (6.25) (the integration starts at x = −1), then it grows
while the integration is performed. At the endpoint, the amplitude reaches a value
equals to A( TE ) = e. While this may not be a realistic change of the wave front, it
serves us to test the dependence on A( T ).
The same steps have been done for a linear varying amplitude. We start with a
linear structure A( T ) = A0 ( T − T0 + b) and impose the same boundary conditions.
At the start of integration, we want that the amplitude takes a unit value; for this to
happen we must impose b = 1/A0 . We have used again the same parametrization
(6.20) and we take A0 = c/L. With these considerations, we use
A ( T ) = 1 + (1 + x ) (6.26)
to compute the timing residual (6.21).

In Figure 6.7 we show how the timing residual changes when the above time-
dependent amplitudes are introduced. The remaining parameters remain fixed with
the values taken in the curve in Figure 6.4 with the Hubble constant value reported
by the Planck collaboration. One more time we take the signals coming from the
pulsar J2033+1734. The timing dependent amplitudes modifies the overall charac-
teristics in the expected way, increasing the whole signal according to the function
taken.
S
at 100 Mpc S
at 100 Mpc
3 Exponential amplitude Exponential amplitude

3
)
)
- Linear amplitude - Linear amplitude
6
6
( τ x 10
( τ x 10
2 Constant amplitude Constant amplitude
2

α

r r
1 1
t t
0 0
π / 16 π/8 3 π /16 0.20 0.22 0.24 0.26
F IGURE 6.7: Absolute timing residual |τGW (α)| for gravitational waves charac-
terized by the two different time varying amplitudes mentioned. As a compar-
ison we also include the timing residual corresponding to a unitary amplitude
in blue.
6.4.2 Dependence with the position of the pulsars
Let us focus on the dependence of the peak on the distance L to the pulsar. Let us
fix the gravitational waves source at 100 Mpc and use the same values for the pa-
rameters as in previous analysis. We plot the absolute value for the timing residual
versus the angle subtended by the source and the measured pulsar as seen from the
observer, |τGW (α)|, for different MSPs from Table 6.2.
We choose five pulsars with locations halfway apart with respect to the Earth;
they are J1939+2134, J1802−2124, J2033+1734, J1603−7202 and J0751+1807 in a de-
creasing order of distance from us. In Figure 6.8 we show a zoom of the enhancement
region. As expected, as long as the Hubble parameter as well as the distance to the
source are fixed parameters, the angular position where the enhancement peak is lo-
cated do not vary for the different curves. On the contrary, the strength of the peak
grows to the extent that farther away pulsars are used in the calculation. There is a
Pulsar distance (kpc) C Pulsar distance (kpc) C

J0030+0451 0.28+ 0.10
−0.06 kpc N,E J1804−2717 0.8 ± 0.2 kpc E
J0610−2100 3.5±0.7 kpc E J2033+1734 2 ± 0.4 kpc E
J0751+1807 0.4+ 0.2
−0.1 kpc E J1939+2134 5+ 2
−1 kpc E,P
J1603−7202 1.2 ± 0.2 kpc P J1955+2908 4.6 ± 0.9 kpc N,E
J1802−2124 2.9 ± 0.6 kpc E
TABLE 6.2: Pulsars employed in the analysis [60]. The distances to the pulsars
have been taken from the reported values in [116]. The column C referees to
the collaboration monitoring the pulsar: N stands for the NanoGRAV collabo-
ration, E for the EPTA and P for PPTA.
logic behind this, as the accumulative effect of the enhancement becomes larger as it
does the optical path.
We emphasize that in all cases the enhancement peak is several orders of magni-
tude bigger that the rest of the signal. Even for the black curve in Figure 6.8 where
at fist sight and because of the scale seems that there is no peak, this reappears when
one plots for a wide range of values for α. Note that this pulsar, J0751+1907, is lo-
cated at a characteristic distance very similar to the one used in Figure 6.4 where
L = 0.32 kpc. In fact, its corresponding plot is very similar to those curves too where
it is clear that the peak is still significantly bigger than the whole signal.
What we also see from these figures is that while the peak becomes higher when
the distance to the pulsar grows, it becomes also narrower. So peaks resulting from
closer pulsars are broader. Figure 6.8 indicates that pulsars located in a region char-
acterized by α ∈ [α1 ; α2 ] register an anomalous enhancement in their timing residual
because of the passage of a gravitational wave produced at a distant source. This in-
terval is related with the width of the peak; in fact the global maximum is the center
of this interval: αmax = (α1 + α2 )/2. All in all, pulsars placed more near to us will
have a bigger range of values α where they would register a significant enhancement
in τGW .
As a more realistic example we will make use of all the real pulsars listed in Table
Source at 100 Mpc
5.0 kpc
4
2.9 kpc
)
-
6
timing residual ( τ x 10
3
2.0 kpc
α
2
1.2 kpc
0.4 kpc
1
0
0.20 0.21 0.22 0.23 0.24 0.25
angle ( )
F IGURE 6.8: Zoom of |τGW | on the enhancement region. Each curve represent each
of the 5 pulsars chosen with different distances to the Earth. It is worth to note
that all peaks are way higher than the rest of the signal. Even the lowest in height,
J0751+1807 (in black), is significantly bigger (a factor 5) than the rest of the re-
maining signal. For instance, take as an example the blue curve corresponding
to the pulsar J1603−7202; this pulsar has nearly the distance employed in Figure
6.4 (where for the red curve L was fixed to 1 kpc). The remaining pulsars are:
J2033+1734 in red, J1802−2124 in violet and J1939+2134 in green.
3.0
● ● for a source at 100 Mpc
Full width at half maximum (deg)

2.5
● for a source at 500 Mpc
● for a source at 1 Gpc

●
2.0
1.5
●
●
●
1.0 ●
●
●
0.5 ●
● ●
● ●
● ●
● ● ●
● ●
● ●
● ● ●
● ●
1 2 3 4 5
L: distance to the pulsar (kpc)
F IGURE 6.9: Each point represents the angular value using the FWHM criterion for
obtaining the main peak width of the signal. We compute this magnitude for each
of the 9 pulsars in Table 6.2, order in the x-axis by the distance from the pulsar to
the Earth. Each curve corresponds as indicated to a different source location.
6.2. Using three different source locations, we can draw the α profile of the modu-
lus of the timing residuals corresponding to each of the pulsars. Added to this, we
need a criterion to link the width of the peak with the interval α where the enhance-
ment happens. A conservative election is the full width at half maximum (FWHM)
parameter. It is defined as the distance between points on the curve at which the
function reaches half of its maximum value1 . In Figure 6.9 we show the result of
computing the FWHM for the 9 pulsars shown in Table 6.2 when the gravitational
wave source is fixed, arbitrarily, at 3 different distances (100 Mpc, 500 Mpc and 1
Gpc). Accordingly, we can remark the convenience of employing nearest pulsars
and nearer sources as much as possible. As reported in the first data release of the
IPTA collaboration in [116] almost the 25 % of the pulsars considered are at distances
L ≤ 0.5 kpc from us.
We can convert the angular interval where the enhancement takes place to as-
tronomical distances; i.e. α → δ. Taking into account at which distance the pulsar
would be located, we can use trigonometry to calculate the size in parsecs that would
correspond to a determined angular interval subtended at that distance L. For the 9
pulsars referred previously we obtain the plot in Figure 6.10. A source located at the
referred distance would "light up" a zone characterized by an angular width α. This
interval would correspond to a distance δ as it is shown in the plot on the left. Any
1 Pulsars further away from Earth would register a higher but narrower peak. The FWHM measure
can be flexibilized; as the half peak is quite a big threshold that can be reduced significantly in pursuit
of increasing the width where the enhancement occurs. This would significantly increase the areas
where this effect can be detected in the sky.
pulsar falling in this interval would experience a significant enhancement in their

timing residual. The three different curves show that the size of this zone is of a few
tens of parsecs and its width is nearly constant with respect to the distance L (the
fluctuations of each curve when varying the location L of the pulsar represent more
or less 10% of its value). We can conclude that the width of the peak seems to be
highly dependent on the location of the gravitational wave source ZE , as it was for
its position αmax derived in equation (6.23).
There is another important aspect to take into account. According to the defini-
tion of the timing residual, the effect is β independent and nearly L independent.
The β independency of the theory is translated into a symmetry of revolution for the
enhancement region δ around the Z-axis (the axis coinciding with the direction to the
source). This means that a SMBH merger would generate a ring of enhancement at
each constant distance L from the Earth characterized by the azimuthal angle rang-
ing between β ∈ [0; 2π ). The angular radius of the ring is given by the polar angle
αmax (recall the results on Table 6.1 where different angular locations of the enhance-
ment have been obtained by using different distances to the source) and its width δ is
given in parsecs on the plot next to the scheme of Figure 6.10, according to the corre-
sponding distance involved. This region where a significant deviation in the timing
residual would be registered is shown in the scheme as the blue ring and will be
called the enhancement region. The invariance with respect to L makes this region a
truncated cone-like surface more that a thick ring; however for practical issues that
will be discussed in the next section, let us retain the region as a ring-like image such
that pulsars placed there would be seen by us as experiencing a notorious deviation
on the time of arrival of their pulses. The width of this ring-like enhancement region
16 ● ● ● ●
● ●
●
●
Ring width δ (pc)
14 ●

12
10
● for a source at 1 Gpc
8
● ● ● ● ● ● ●
● ●
6 ● ● ● ● ● ● ● ● ●
1 2 3 4 5
L: distance to the pulsar (kpc)
F IGURE 6.10: Spatial region (in blue) where any falling pulsar would experi-
ence an anomalous enhancement in timing residual τGW . This region is shown
for only one value of the distance L, but appears for all values of L with nearly
the same characteristics (the complete region is a truncated cone-like surface).
On the left we show the width δ of the ring that seems to depend only on
changes of the source location ZE .
will be in the 1 to 20 pc range, depending on the distance to the source.
6.5 The International Pulsar Timing Array project

In order to detect correlated signals in observed pulsar timing data, an array of MSPs
is monitored with large and sensitive telescopes. Such array called PTA2 exploits the
great period stability of the MSPs where correlated timing variations among the ToA
pulses in the array are searched for [117].
Three different collaborations has been set up in three different places and have
been taking data for more than a decade. Specifically, the Australian Parkes PTA
(PPTA) [118], the North-American Nanohertz Observatory for gravitational waves
(NANOGrav) [119] and the European PTA (EPTA) [120]. Combined, they form the
International PTA (IPTA) [53]. This collaboration monitors around 50 pulsars in the
quest for finding footprints of the passage of gravitational waves in the aforemen-
tioned correlated signals.
The quest for detection of gravitational waves in PTA follows two main avenues.
One proposal focuses on stochastic (isotropic and incoherent) gravitational wave
background [109, 108, 121] produced by incoherent superposition of radiation from
the whole cosmic population. In [116] a brief review goes through the main sources
for this background and define the resulting signal that would (or not) be detected
by PTA.
More interesting for our purposes is the search for single or individual sources,
if they are sufficiently bright. Clearly, a nearby binary merger would be detectable
if and only if the gravitational signal stand out above the root-mean-square value
of the aforementioned background [122]. There is controversy on whether a single
merger event is detectable by PTA [123]. This debate and its somewhat pessimistic
conclusion assumes of course the standard analysis of the signal-to-noise ratio.
As of today there is no conclusive evidence of an indirect measurement of any
gravitational radiation in this regime of frequencies wherever they are produced.
Single-source limits show that all proposed binary mergers are still below current
sensitivity levels: recent limits have been derived in [124] from the EPTA, in [125]
from the NANOGrav data and in [126] from the PPTA collaboration. In addition,
similar conclusions have been reached for gravitational waves bursts [127]. Our
analysis reported here may suggest to reconsider the expected signal and how to
look for it.
When gravitational waves propagate over flat spacetime the phase effect dis-
cussed before is lost as we have shown in the first plot of Figure 6.2 (Minkowskian
propagation). Including the familiar redshift in frequency alone as the only way of
including the effect of the FLRW metric, without taking in consideration the effect
2 The acronym PTA stands for the set of pulsars as well as for the different collaborations monitoring
these pulsars.
6.5. The International Pulsar Timing Array project 113
discussed in this work, for each single-source event a ring-like region of enhance-
ment in the sky is possibly missed. The enhancement should be affecting any pulsar
in it, irrespective of the distance to the observer L. We recall that the enhancement
may represent an increase in the signal of several orders of magnitude. From the
study presented in the previous sections and the typical values of L we concluded
that the width of this ring is of the order of 1 to 20 pc.
The image presented in Figure 6.10, we reminds us of the coordinates of the pul-
sars monitored by the IPTA collaboration, irrespective of the value of L, the distance
between the pulsar and the Earth. In the graphical representation on Figure 6.11, for
each sphere of the sky at a distance L, a SMBH merger would induce a circular ring-
like region with a diameter given by twice the angle subtended between the source
and the pulsar with respect to the source. In Figure 6.11 we show in a Mollwide pro-
jection the location of these pulsars together with the enhancement ring for a source
located at 100 Mpc and at 1 Gpc.
Assuming that each collaboration measures the timing residual over an adequate
integration period a single-event should represent a simultaneous firing of at least
three pulsars. The collaboration should therefore look for such triple signals by mak-
ing all possible combinations. Each of such triple signals would define a vector
pointing towards the source. Once this direction is tentatively determined, a more
detailed analysis of the signal effect on other pulsars would be possible.
Note that along this chapter all signals referred to snapshots at a given specified
75°
60°
45°
30°
15°
-150° -120° -90° -60° -30° 0° 30° 60° 90° 120° 150°
Dec (°)
0°
-15°
-30°
-45°
-60°
-75°
RA (°)
F IGURE 6.11: Galactic distribution of all MSPs observed by the IPTA are indi-
cated by blue stars. Enhancement rings correspond to three different sources
placed at 100 Mpc in blue and at 1 Gpc in green. The red ellipse would be the
enhancement ring if the three further south pulsars are supposed to perceive
the enhancement. When we see that at least three pulsars register an anoma-
lous timing residual, we have enough information to know where the effect
has its origin. This three points determine an enhancement ring of αmax ' 25◦
with respect to a center where there should be placed an hypothetical gravita-
tional wave source. According to the relation (6.23), the angle of enhancement
constrain the spatial location of the source to ZE ' 443 Mpc. The angular
radius subtended between the source and the pulsar are for the other two en-
hancement rings: ∼ 39◦ for the green ellipse and ∼ 12◦ for the blue one.
time. Of course, there is a large background that obscures the signal (even taking
into account the enhancement) and therefore the signal has to be integrated over an
adequate period of time. This issue has not been considered here, but we refer the
reader to the work [58] where it was preliminarily estimated that observation of the
given set of pulsars over a period of three years with an eleven-day period would
provide a sufficiently good signal. Needless to say that this is a very crude value
that needs careful consideration.
115
Chapter 7
Conclusions
The purpose of this work is to provide some new insights into different aspects of
black hole physics, both at a quantum level and at a classical or observational point
of view. In reference to the former, we discuss its possible internal structure em-
ploying some common approximations in the context of quantum mean field theo-
ries and give different internal structures based on a graviton condensate structure.
With respect to the observable aspects, we discuss how in which manner the fact
that we live in an expanding universe would produce any measurable effect over
the gravitational waves produced in a black hole merger.
We have started analyzing a novel approach that reformulates the quantum the-
ory of black holes in a language of condensed matter physics. The key point of the
theory is to identify the black hole with a Bose–Einstein condensate of gravitons.
In Chapter 2 we have conjectured the set of equations that play the role of the
Gross–Pitaevskii non-linear equation that describes the ground state of a quantum
system of identical bosons using the Hartree–Fock approximation as an interacting
model; they are the field equations derived from the Einstein–Hilbert Lagrangian af-
ter adding a chemical potential-like term. We have used a number of different tech-
niques to analyze these equations when the perturbation (i.e. the tentative conden-
sate) has spherical symmetry. The equations appear at first sight rather intractable,
but by doing a perturbative analysis around the black hole Schwarzschild metric at
quadratic order (i.e. including the leading non-linearities) we found that the chem-
ical potential necessarily vanishes in the exterior of the black hole. On the contrary,
in the interior we have found two sets of solutions, one of them has to be discarded
as producing a non-normalizable result. The other one leads to a nonzero chemi-
cal potential in the interior of the black hole that behaves as 1/r2 . Therefore, there
is a finite jump on the chemical potential at the black hole horizon. Surprisingly –
or maybe not so– this solution modifies the coefficient of the tt and rr terms in the
Schwarzschild solution, but not its functional form. Of course if there is no chemical
potential at all, the modification vanishes, in accordance with well known theorems.
However, if the former is nonzero, the modification affecting the metric is also nec-
essarily nonzero.
The perturbative analysis triggers a unique consistent physical solution for the
non-perturbative (exact) theory. This solution is characterized by a constant density
116 Chapter 7. Conclusions
of the wave function for the condensate. From the existence and knowledge of this
solution, an unambiguous relation between the number of gravitons and the geo-
metric properties of the black hole is obtained. Hence, we find an expression for the
Schwarzschild radius that involves an a priori independent and tuneable parameter,
the dimensionless chemical potential X (related to the mean-field wave function of
the condensate ϕ). We find this somewhat strange as this would be a new black hole
parameter. Therefore we favour the universal value X = −1 as discussed in the text.
As should be obvious to the reader who has followed our discussion, the ap-
proach is somewhat different from the one developed in the initial papers by Dvali,
Gómez and coworkers. We assume from the start the existence of a classical geome-
try background that acts as confining potential for the condensate. The fact that the
functional form of the metric perturbation induced by the condensate is exactly the
same as the original background, of course gives a lot of credence to the possibility
of deriving the latter from the former in a sort of self-consistent derivation. We have
not explored this possibility in detail yet.
The picture presented in the context of Schwarzschild black holes has been ex-
tended to charged black holes (with Reissner–Nordström metric) and to the de Sitter
cosmological horizon. We have found that in all these situations a graviton conden-
sate is possible in the classically inaccessible zone when a chemical potential term
is appropriately introduced in Einstein’s equations. This seems to suggest that the
existence of an horizon is intimately linked to the existence of such condensates.
In order to discern whether this possibility describes in any manner the nature of
black holes, in Chapter 3 we analyze how the notion of quasilocal energy in the con-
text of the classical theory of general relativity can be extended so as to encompass
the condensate description. Even though there is no total consensus on the defini-
tion of a quasilocal energy, we consider the Brown–York prescription [81] as being
adequate for our purposes and physically well-motivated in the present context. If
this definition for the energy it chosen, it is found that a graviton Bose–Einstein con-
densate is energetically favourable for all the studied cases in the regions beyond the
horizon.
The possibility of defining a quasilocal energy entails the possibility of deriving
a quasilocal potential describing the binding of falling matter to a black hole. Its
well-like structure, makes the binding to take place around a fairly thin shell located
at the event horizon. This seems to indicate that matter falling into the black hole
accumulates at both sides of the horizon, but always in its vicinity. This confers to
the black hole a fuzzy boundary. When matter falls and get trapped by this po-
tential, gravitational wave emission is necessarily produced on energy conservation
grounds. Part of this energy is emitted towards infinity, however it is expected that
a large fraction is trapped in the inner region. This seems to be a very plausible ex-
planation on how the condensate of gravitons behaves and the value of the graviton
condensate order parameter ϕ increases. As explained in section 3.5, order of mag-
nitude considerations indicate that practically most of the black hole mass may be
Chapter 7. Conclusions 117
stored in the form of a graviton condensate state. Besides, the unceasing capture of
gravitons may indicate that the condensate value should be close to its limit value,
ϕ → 1. This statements gives strong plausibility to the original proposal of Dvali
and Gómez; namely that a black hole consist mostly of a BEC of gravitons.
In Chapter 4 we show that the extension of the energy considerations to the
Reissner–Nordström case is straightforward. However, is not always possible to
define a significant quasilocal potential for every spacetime. For instance, in de Sit-
ter we have been unable to do so, this concluding that while it is energetically viable,
a graviton condensate is actually not present.
The validity of the previous picture brings about many interesting points. One
of them is that the horizon, while being from the metric point of view, quite well
defined, becomes necessarily fuzzy due to quantum effects. A second consequence
is that matter falling in the black hole does not get ‘crunched’ by the singularity at
r = 0. When it comes to the quasilocal energy, it is perfectly regular at the origin,
and so is the associated potential (it tends to a constant). In the classical theory, the
location of the horizon is defined as a limit for timelike world lines to exist. Points
where 4-velocity turns null form a ‘one way’ spatial surface. Here, following our in-
terpretation, we see that matter is captured by the quasilocal potential and certainly
cannot escape, except for the occasional thermal fluctuation. However, no loss of
information would occurs here, except for the usual thermodynamic irreversibility.
If the mass of the black hole decreases for some reason and rS → 0, the potential
becomes progressively shallower and matter stored in its potential well can escape.
Needless to say that one mechanism whereby the mass of the black hole could de-
crease is through evaporation. This is an intrinsically quantum mechanism that is
accurately described in the usual way. However, if a graviton condensate is present,
and we have argued that this is very likely with a value close to the limiting value,
one should take into account that Bose–Einstein condensates are not localized ob-
jects and therefore some amount of leaking should be present. In their original pro-
posal Dvali and Gómez sustained the point of view that Hawking radiation could
be understood in this way. From the point of view that most of the black hole mass
is stored in the form of the condensate, this is a likely possibility, but we have not
considered this issue in the present study.
Next, in Chapter 5, we review shortly the main aspects of the ΛCDM model
that describes in a very good agreement the universe over where observational phe-
nomena are aimed to be measured. For instance it is well known that accelerated
masses generate gravitational radiation in the form of gravitational waves. We have
reviewed how a nonzero cosmological constant modifies the propagation of a grav-
itational wave, with an additional contribution to the one that is usually taken into
account –the redshift in frequency. The effect is entirely due to the change of coordi-
nate systems between the reference frame where the wave originates (e.g. the merger
of two gigantic black holes) and the reference frame where waves are measured in
PTA, namely cosmological FLRW coordinates. We have proceeded to extend these
118 Chapter 7. Conclusions
results to the case where non-relativistic dust is present, and later on to the com-
bined and analytical situation where non-relativistic matter and vacuum energy are
both present. Finally the inclusion of radiation motivates a picture dependent on the
Hubble constant, including all the components of the background in which gravita-
tional waves propagate. All the results have been derived in a linearized approxima-
tion where only the fist deviations are considered. This approximation is however
enough for the case of study, as subsequent corrections are seen to be extremely
small when the measured values of matter density and cosmological constant (the
two main components) are used, combined with the distances involved in the prob-
lem.
In the last chapter, we have confirmed the validity of the approximate solution
shown in equation (6.13) by solving directly, up to O( H0 ), the wave equation for
linearized perturbations in a FLRW geometry. The solution has an effective wave
number k e f f 6= we f f , where we f f is the familiar redshifted frequency. This does
not mean that gravitational waves propagate subluminically; they do so when the
proper ‘ruler’ distance is considered. This approximate solution is valid to describe
gravitational waves originating from sources up to ∼ 1 Gpc, or, equivalently, up to
redshift z = 0.2 approximately. It is important to emphasize that, at this order, all
cosmological parameters enter in the form of the Hubble constant H0 , but it is not so
when higher orders are included.
It came out as a surprise that the gravitational wave solution has a definite im-
pact in observations of gravitational waves involving non-local effects; there is an
integration over a path in comoving distances. This is the case of observations made
in the framework of the IPTA project (and, incidentally, also in LISA, even though the
analysis in this case is yet to be performed). Also in Chapter 6 we have discussed
in great detail the dependence of the effect on various parameters influencing the
observation: superposition of frequencies, time-varying amplitudes, distance to the
source, distance to the intervening pulsars. In all cases the signal should be perfectly
visible, well above the usual analysis assumed. The analysis here presented may
help to establish a well designed search strategy. We note that one aspect we have
not discussed at all are the integration times. In [58, 59] an observational strategy
was mentioned, but we consider that the concourse of the observational teams is
essential in this point.
So far no clear positive measurement of gravitational waves has come out from
the existing collaborations. The effect reported here appears to be firmly established,
and in conclusion it may provide an opportunity for the IPTA collaboration to de-
tect correlated signals in triplets of pulsars and from there on determine location
and other properties of the source of gravitational waves. Needless to say that an
independent local determination of H0 (and eventually Λ) would be of enormous
interest.
119
List of publications
The research carried out during this PhD thesis has given rise to the following list of
papers:
• Bose−Einstein graviton condensate in a Schwarzschild black hole.

J. Alfaro, D. Espriu and L. Gabbanelli
Class. Quant. Grav. 35, 015001 (2018); doi: 10.1088/1361-6382/aa9771
arXiv:1609.01639
• On the propagation of gravitational waves in a ΛCDM universe.

Class. Quant. Grav. 36, 025006 (2019); doi: 10.1088/1361-6382/aaf675
arXiv:1711.08315
• Condensates beyond the horizons.

arXiv:1905.01080
• Measuring H0 in PTA with gravitational waves.

D. Espriu, L. Gabbanelli and M. Rodoreda
arXiv:1908.08472
121
Abstract in Spanish
El objetivo de la presente tesis es profundizar en diversos aspectos de la física de

los agujeros negros. Tanto en lo que respecta a sus características constitutivas fun-
damentales, su "estructura" interna, como a la posibilidad de observar o detectar
mediante observaciones astrofísicas ciertos efectos producto de su dinámica.
Por un lado, hemos seguido las ideas de Dvali, Gómez et al. quienes han sug-
erido la posibilidad de que un agujero negro sea un condensado de Bose–Einstein de
gravitones débilmente interactuantes. En nuestro caso hemos estudiado la existencia
de este tipo de soluciones sobre diferentes métricas de agujero negro (Schwarzschild
y Reissner–Nordström) que actuarían como potencial confinante para dichos con-
densados. Un parámetro necesario para ello, es el equivalente a un potencial químico
que debe ser incorporado a la relatividad general. Cabe destacar que la solución en-
contrada puede ser interpretada como la función de campo medio del condensado.
Además resulta fuertemente ligada a la estructura clásica de la métrica que la sus-
tenta.
Por otro lado, es bien sabido que la aceleración de cuerpos muy masivos pro-
ducen perturbaciones de tipo onda en el espaciotiempo. Son de nuestro interés las
ondas gravitatorias de baja frecuencia, provenientes de la colisión de agujeros negros
supermasivos y que deberían poder ser detectadas mediante sistemas de púlsares
(Pulsar Timing Arrays). De acuerdo a una línea de investigación desarrollada por
Espriu et al. la presencian de una constante cosmológica podría tener un efecto en la
propagación y por lo tanto en la detección por parte de la colaboración IPTA de es-
tas ondas. En la presente tesis hemos generalizado el método para incluir diferentes
tipos de materia (relativista y no relativista) además de la constante cosmológica. Del
análisis se deriva que el efecto depende sensiblemente del valor de la constante de
Hubble (que engloba todos los tipos de materia presentes). Continuando dicha línea,
hemos caracterizado detalladamente el efecto en su dependencia con los parámetros
cosmológicas y las distancias involucradas, y cómo podría ser hallado. Esperamos
que nuestros resultados puedan contribuir a una definitiva detección por IPTA.
123
Resume in Spanish
Es un hecho que nuestro universo está en expansión. Las mediciones a escalas cos-
mológicas son consistentes con la presencia de una forma de energía intrínseca uni-
formemente distribuida. Si bien esta energía es hasta el momento completamente
desconocida, se estima que contribuye con aproximadamente un 68 % del contenido
energético del Universo. El modelo teórico más simple en acuerdo con las observa-
ciones es el modelo cosmológico estándar. En este, una constante cosmológica (Λ)
pequeña pero positiva produce un efecto gravitatorio repulsivo, resultando la ex-
pansión acelerada. Este modelo tiene sus comienzos en el desarrollo de la teoría
general de la relatividad de Albert Einstein y la mejora en las observaciones as-
tronómicas de objetos extremadamente distantes. Una de las predicciones más con-
troversiales de la teoría de la relatividad es la existencia de soluciones de tipo agujero
negro. Estas soluciones indican que una concentración de masa lo suficientemente
densa en una región acotada del espacio genera un campo gravitatorio tan extremo,
que ninguna partícula material, ni siquiera la luz, puede escapar de dicha región.
A un nivel matemático, estos objetos son soluciones de vacío y su estructura
puede ser descripta por muy pocos parámetros. Por ejemplo, existe un teorema
que afirma que un agujero negro en estado estacionario puede ser descripto com-
pletamente por tres parámetros: su masa, su carga eléctrica y su momento angular.
Como consecuencia, dos agujeros negros que comparten los mismos valores para
estos parámetros, son indistinguibles entre sí. De acuerdo a las métricas habituales
que describen este tipo de objetos estelares, la curvatura del espaciotiempo se vuelve
singular. En este lugar, la solución predice que se encuentra localizada toda la masa
(en el centro mismo del agujero negro, o si gira, en un anillo infinitamente delgado).
Estas singularidades se encuentran envueltas por superficies cerradas denominadas
horizontes de sucesos.
Enfoques recientes desafían esta estructura poco intuitiva y se enfocan en formas
de lograr que la materia se extienda en todo el interior en forma consistente con las
condiciones de altas presiones debidas a las altas densidades. Claramente, estas
condiciones extremas requieren de una teoría cuántica para la gravedad. En los
últimos años una novedosa propuesta estableció una conexión entre la información
cuántica y la física de los agujeros negros: los agujeros negros podrían entenderse
como condensados de Bose–Einstein de gravitones débilmente interactuantes pero
densamente empaquetados. De este modo el comportamiento "misterioso" de los
agujeros negros no sería más que el producto de efectos colectivos de un sistema
cuántico de bosones idénticos.
124
El objetivo de la presente tesis es llevar esta propuesta al terreno de las soluciones

conocidas de la teoría de la relatividad general. Uno de los problemas centrales de
esta teoría es que no se tiene una noción local y covariante de la densidad de energía
del campo gravitatorio. Por supuesto, esto se encuentra estrechamente vinculado
con los problemas históricos para cuantizar este campo. Por lo tanto, el problema
de su localidad es objeto de intenso debate. De cualquier manera, la posibilidad de
definir este tipo de energía en regiones compactas (que incluso pueden poseer hori-
zontes y singularidades) ha llevado a postular numerosas representaciones cuasilo-
cales de la energía. Esto permite asociar a la densidad de energía una densidad
de numero de gravitones en estado fundamental conformando el condensado. De
acuerdo a la correspondencia propuesta, la radiación semiclásica que Hawking ob-
tuvo al aplicar la teoría cuántica de campos a espacios curvos, no sería otra cosa que
la depleción cuántica usual que sufren los condensados de Bose–Einstein. Esta es
una propiedad puramente cuántica producto por la cual el condensado se encuentra
constantemente chorreando bosones debido a efectos de scattering donde un gravitón
es excitado fuera de su estado fundamental; esto le permite vencer el potencial de
los N − 1 restantes y abandonarlos.
De acuerdo a estas consideraciones, la depleción es responsable del la variación
de los parámetros previamente mencionados, en particular de la densidad de partícu-
las (en un sentido probabilístico). Por lo tanto, es termodinámicamente consistente
buscar alguna especie de potencial químico que de cuenta de la variación del número
de partículas N del condensado. En el capítulo 2 discutimos cómo introducir este
parámetro dentro de la relatividad general.
En el marco de teorías de materia condensada, bajo la aproximación de Hartree–
Fock la función de onda del estado fundamental de un sistema de N bosones es
descripta por la ecuación de Gross–Pitaevskii. Desde un aspecto geométrico, el po-
tencial químico regularía las componentes de un tensor de orden 2 (hµν ) asociado a
perturbaciones de una estructura causal clásica de tipo agujero negro. Esta métrica
background haría las veces del potencial confinante y las ecuaciones de campo para
hµν pueden ser interpretadas como las ecuaciones efectivas de tipo Gross–Pitaevskii
que rigen al condensado de gravitones. La solución para hµν posee característi-
cas fuertemente dependientes de la métrica clásica que sustenta el condensado y
por lo tanto puede ser interpretada como la función de campo medio de este con-
densado. Es interesante remarcar que las propiedades de los agujeros negros que
pueden derivarse de esta propuesta resultan siempre parametrizadas únicamente
por el radio de Schwarzschild.
Una de las propuestas mejor fundamentadas para definir la energía cuasilocal
tiene sus bases en una formulación covariante del formalismo de Hamilton–Jacobi.
En el capítulo 3 rediscutimos ciertas consideraciones energéticas que se derivan de
esta energía cuasilocal aplicada a los agujeros negros, pero desde el punto de vista
de considerarlos como condensados de gravitones. Sorprendentemente la estructura
condensada es energéticamente favorable en comparación a las soluciones de vacío.
125
Además, es posible definir un potencial cuasilocal efectivo que permitiría proveer un

posible mecanismo para dar cuenta de la absorción y depleción de los constituyentes
del condensado. El perfil de este parámetro presenta un pozo de potencial sobre
el horizonte de eventos donde la materia podría almacenarse en estado cuántico
confiriendo a estos objetos un "borde" difuso, como es bien sabido que sucede para
los condensados.
En el capítulo 4 mostramos que, de manera contraria a las métricas que presentan
horizontes de eventos (por ejemplo de tipo Schwarzschild o Reissner–Nordström),
nuestra formulación no es consistente para métricas con horizontes cosmológicos y
por lo tanto no pareciera ser posible justificar una estructura condensada sobre una
métrica de tipo de Sitter.
Pasemos ahora al análisis de la dinámica de un tipo particular de agujero negro:
los supermasivos, con masas entre 107 y 1010 masas solares (M ). Un sistema bina-
rio de este tipo de objetos es la mayor fuente de ondas gravitacionales del Universo;
y el perfil del frente de onda es otra de las sólidas predicciones de la teoría de Ein-
stein. En teoría este tipo particular de onda es de baja frecuencia (entre los nHz y
los mHz) y es sensible a ser detectada por observaciones astrofísicas de arreglos de
pulsares (PTA). En el capítulo 5 discutimos por qué en este tipo de observaciones
debe tenerse en cuenta el hecho de que la propagación se produce en un universo
en expansión (globalmente de tipo de Sitter) y no en un universo plano y estático.
Esta expansión haría que las ondas emitidas durante la fase espiral, que pueden ser
descompuestas en una suma de ondas armónicas, se vuelvan anarmónicas al llegar
al sistema local de coordenadas donde deberían ser detectadas. Esta anarmonicidad
es pequeña; sin embargo en el capítulo 6 damos evidencia de que puede ser sensible
a las observaciones de PTA. Caracterizamos las posibles modificaciones a la señal
que las colaboraciones que monitorean estas estrellas esperan obtener. Analizamos
en detalle la estructura de la nueva señal esperada, cómo se relaciona con las car-
acterísticas de la fuente de ondas gravitatorias y con las posiciones de los púlsares
empleados por la colaboración IPTA para la detección.
Vale la pena destacar que la locación del efecto encontrado es relativamente sen-
sible al valor de la constante de Hubble, por lo tanto la posibilidad de hallar ondas
gravitatorias por este medio podría traer aparejada la medición local de esta con-
stante H0 . Es claro que si la constante cosmológica fuera una propiedad intrínseca
del Universo, este efecto sería capaz de confirmar la existencia de Λ para redshifts
tales que z < 1.
127
Acknowledgements
Dicen algo así como que la suerte te engancha justo donde el mucho esfuerzo te dejó;
sin embargo a esta frase le falta un ingrediente clave: las personas que lo recorrieron
a tu lado y lo hicieron posible.
Quiero agradecer en primer lugar a Domènec Espriu con quien compartí a lo
largo de todos estos años dos grandes pasiones: hacer física y explorar el mundo.
Realmente me encantó haberlo compartido y haber aprendido a su lado. También
quiero darle las gracias a Jorge Alfaro por su hospitalidad y su recibimiento en San-
tiago de Chile. Voy a guardar en mi memoria los encuentros con ambos. A ambos,
gracias.
A Federico Mescia con quien disfruté muchísimo la docencia, otra de las cosas
que amo. Gracias a los jurados por aceptar ser parte: Oriol Pujolas, Jorge Russo
y Carlos Sopuerta, espero que disfruten leyendo esta tesis tanto como yo disfruté
escribiéndola.
Casi al comienzo de mi doctorado, en alguna conferencia, alguna persona dijo
algo que cambió mi percepción de la física. Las palabras fueron algo así como
"physics is about friendship". Esas palabras me dejaron una huella imborrable. Es
así que esta tesis va dedicada a mis colegas alrededor del mundo. A Jorge Ovalle y
Adrián Sotomayor, ojalá las casualidades nos sigan reuniendo en los rincones más
recónditos. Mientras, estimulemos las causalidades. A Ángel Rincón y Carlitos Ru-
bio Flores, porque hacer cuentas con ellos siempre fue un disfrute y ojalá lo siga
siendo.
También quiero dedicársela a les pilares de mi vida. A mi compañera, Guada,
quien me enseñó que lo único permanente es el cambio. A mi hermano, con quien
comparto una vida de mucho más que amistad. Siempre son todo lo que está bien y
les admiro profundamente. Gracias por estar.
A mi vieja, por su fuerza arrasadora, por sus ganas de vivir cada segundo y por
seguirme en los viajes más inesperados. A mi viejo, por su amor incondicional. A
Ivo, Agus, Fabi y Gaby, por la perseverancia y por elegir ser libres. A Marisol y Luís,
por estar siempre. A la abuela Ilda, a ver si pronto le festejamos el medio milenio :).
A mis amigos de la vida: Javito, Alan, el Pela, el Bishu, Dante, Emito y el Ne-
grus. Me encanta levantarme cada mañana con tres o cuatro ordenes de magnitud
de mensajes. Y sí, todavía me parece increíble que no hayan dudado un segundo en
venir en banda para acá. Una dedicatoria especial va para SanSan. A él va dedicado
todo el capítulo 3 porque escribirlo me llevó devuelta a la mesa redonda de Salguero
y los baldes de café.
128
A con quienes compartimos kilómetros, con el podio indiscutido para la Colo,

Rami, Haize, Ikercito, Jonny y Dieguis. A la inmensa Vane. A la familia extendida:
Juli, Gaby, Osqui, y Poli. Al Maicol, Emi y Ro porque cada vez que nos reencon-
tramos es como si el tiempo no pasara. A quienes cruzaron el charco y a quienes no,
y a les que siempre están yendo y viniendo: Virgi, Marian, Tomi, Yaguis, Robi y Tati.
A Ceci Villegas, la responsable de transformar los desayunos y meriendas.
A las geniales apariciones; a los conejos de todas las galeras, porque sin querer
saber el truco, me quedo con la magia. A Juan, porque descubrí un gran amigo de
sopetón. A Eze, slackeros y la gente de Can Batlló con quienes aprendí otra nueva
forma de autogestión. A toda la gente colgada, en especial a Kat, Gemma, Joan-
ito, Anna y Clara, por enseñarme a volar. A Marcelo, Darío y Alfonso, agradezco
haberme cruzado con ellos para darle vueltas a nuestro increíble instrumento. A
la orquesta Latitud y a la fila de bandoneones, es muy emocionante todo lo hecho
y seguro todo lo que vendrá. Para todas las personas que construyen mi vida en
Barcelona; ustedes son parte fundamental del todo. Por último y muy especial-
mente, va dedicada a mis cuadrúpedas amigas con quienes compartimos infinitas
horas silla de trabajo y estudio. Y a todes con quienes compartimos sonrisas. Gra-
cias por compartir.
129
Appendices
131
Appendix A
Perturbative aspects of the classical

theory
In the present appendix we will rederive a well known result from [68], but in conve-
nient perturbative formalism, useful for constructing a condensate. In this manner,
we will study the spacetime curvature part of the field equations, coming from the
Einstein tensor.
As we will see, every dimensionful quantity can be expressed as a function of
the Schwarzschild radius rS . Let us consider a perturbation on the Schwarzschild
metric with the form of (2.3). We will now expand the Einstein tensor up to second
order in hαβ . This will allow us to retain the leading non-linearities. Four of the com-
ponents of the Einstein tensor are nonnull but only three are linearly independent;
both angular components are the same. These are Gt t , Gr r and Gθ θ
r − rS r + rS (r + 2rS )(r − rS ) 2 (r − r S )2

0 0
− 3 hrr + (r − rS )hrr − hrr − 2 hrr hrr
r r r2 r

1
rS r2 (r − rS ) htt − r (r − rS )3 hrr − r3 (r − rS )2 htt 0 + (r − rS )4 hrr 2
r 4 (r − r S )2

3 2 2 4 0 2 3 0
+rS r htt − rS r (r − rS ) hrr htt − r (r − rS ) htt htt + r (r − rS ) hrr htt
1
(A.1)
− rS r3 (r − rS )(2r − rS )htt + rS r (2r − rS )(r − rS )3 hrr
4r5 (r − rS )3
+r4 (2r − 3rS )(r − rS )2 htt 0 + r2 (2r − rS )(r − rS )4 hrr 0 + 2r5 (r − rS )3 htt 00
−2rS (2r − rS )(r − rS )4 hrr 2 + 2rS r5 htt 2 − 2rS r2 (r − rS )3 hrr htt
+r5 (2r − 5rS )(r − rS )htt htt 0 + r6 (r − rS )2 htt 02 − 2r3 (r − rS )4 hrr htt 0
+rS r3 (r − rS )3 htt hrr 0 − r4 (r − rS )4 htt 0 hrr 0 − 2r (2r − rS )(r − rS )5 hrr hrr 0

6 2 00 4 4 00
+2r (r − rS ) htt htt − 2r (r − rS ) hrr htt .
The prime stand for derivative with respect to the radial coordinate r. Of course, the
(0)
zero order vanishes as Gαβ ( g̃αβ ) = 0. With this three components we get the three
132 Appendix A. Perturbative aspects of the classical theory
O(2) vacuum Einstein’s equations
Gα β ( g̃αβ + hαβ ) = 0 (A.2)
to solve.
It is well known that Birkhoff’s theorem [67] guarantees that any spherically
symmetric solution of the vacuum field equations must be static and asymptoti-
cally flat. So this means that the solution of the equations Gαβ ( g̃ + h) = 0 with
g̃αβ being the Schwarzschild’s metric, must necessarily be of the Schwarzschild type
again. In conclusion, a perturbation of a Schwarzschild metric in vacuum must end
in another Schwarzschild metric but taking into account a change in the parameters
responsible of inducing the perturbation. Stating this in other words, a spherically
symmetric gravitational field should be produced by some massive object at the ori-
gin characterized by the Schwarzschild radius, hence a perturbation in itself it can
only be produced by an alteration of this radius and therefore a change in its mass.
Indeed, imposing the following ansatz for the components of the perturbation
∞ ∞
B(n) A(n)
htt = ∑ rn
; hrr = ∑ rn
(A.3)
n =1 n =1
and substituting hrr in the corresponding field equation constructed with the first
component (temporal) of the Einstein tensor written in (A.1), is possible to obtain
the coefficients for each term of the radial expansion. We obtain
A A2 + 2rS A 2A3 + 3rS A2 + 3rS 2 A

hrr = + + +... (A.4)
r r2 r3
where A = A(0) . It is obvious that A = 0 is a solution as implies no perturbation of

the metric at all. But if A 6= 0 the rest of the coefficients of the expansion are fixed
once A is given. The same happens for htt using the radial equation from (A.1)
A A3 5A4 + 8rS A3
htt = +0+ 3 + +... (A.5)
r r 12r4
Taking into account the accuracy of the expansion in hαβ the components of the met-
ric become
!
rS A 2GN M
g̃tt + htt ' − 1 − + = − 1−
r r r
! (A.6)
rS r 2 A A2 + 2ArS 2GN M (2GN M )2
g̃rr + hrr ' 1+ + S2 + + 2
= 1+ + ;
r r r r r r2
M being the new mass associated to the black hole and r̄S = 2GN M the perturbed
Schwarzschild radius. Now we will use similar techniques in the ‘quantum’ case
and as we will see the results are dramatically different. However it is interesting to
Appendix A. Perturbative aspects of the classical theory 133
note that any perturbation hαβ (i.e. each graviton) has associated a certain amount of
energy that is reflected in a change of the black hole mass.
135
Appendix B
Action variation
Here we will shortly review the variation of the action (2.8) with respect to the per-
turbation (it is equivalent to compute the variation with respect to hαβ )
" √ √ #
R δ −g − g̃ µ δ(hρσ hρσ )
Z
2 4
p δR
0 = δS = MP d x −g √ + αβ − √ δhαβ . (B.1)
− g δhαβ δh −g 2 δhαβ
The first two terms corresponding to the Einstein tensor are widely known, let us
just clarified a few steps: making use of the chain rule, δ/δhαβ = (δgρσ /δhαβ )δ/δgρσ ,
we get the Ricci tensor from the second term, δR/δgρσ = Rρσ , and, with the follow-
√ √
ing identity δ − g/δgρσ = − − g gρσ /2, with the first term we get both terms of
the Einstein tensor. In the third term, the chemical potential term, is useful for the
variation to write it as hρσ hρσ = gρλ gσγ hλγ hρσ .
Since the action principle should hold for any variation δhαβ , yields
√ " #
µ − g̃ ( g g )

R δgρσ δh ρσ δ ρλ σγ
Rρσ − gρσ − √ 2gρλ gσγ hλγ αβ + hλγ hρσ = 0 . (B.2)
2 δhαβ 2 −g δh δhαβ
The metric has been defined as a perturbation of the background metric (2.2); the
background metric is independent with respect to the perturbation, hence gives just
ρ
Kronecker deltas δ( g̃ρσ + hρσ )/δhαβ = δα δβσ because of the symmetric property of
this tensor.
There is only one remaining term for obtaining the field equations. The final in-
gredient is the variation of the metric with respect to the inverse metric. For instance,
using that δgσγ /δgαβ = −( gσα gγβ + gσβ gγα )/2 and doing the the corresponding in-
dices contractions, we get the Einstein field equations (2.9) with a chemical potential
term √
µ − g̃
Gαβ − √ 2 hαβ − 2 hασ hσ β = 0 . (B.3)
2 −g
137
Appendix C
Extrinsic curvature of 3 B
In the present appendix some manipulation of the expressions will be presented to

make the generalization to general relativity and the derivation of the quasilocal
energy for static and spherically symmetric spacetime more tractable.
Let us make use of all the induced coordinates to write the extrinsic curvature
(3.6) of the three-boundary 3 B in terms of the extrinsic and intrinsic geometry of the
spacetime foliated into the spacelike hypersurfaces Σ. For doing so, let us write the
extrinsic curvature introducing two identity tensors written in terms of the metric
tensor defined on Σ, δµα = γµα − uµ uα . In fact, if we expand Θµν = δµα δν Θαβ we obtain
β
Θµν = −γµα γν γα ∇ρ n β − uµ uν uα u β γα ∇ρ n β + 2γ(αµ uν) u β γα ∇ρ n β ;

β ρ ρ ρ
(C.1)
where we have used that the extrinsic curvature is a symmetric tensor, because the
unit normal nµ is surface forming with vanishing vorticity
ρ ρ
γα ∇ρ n β − γ β ∇ρ nα = 0 . (C.2)
Let us manipulate each of the terms in (C.1). The orthogonal condition of the
unit normals on 3 B, i.e. uµ nµ = 0, makes the projectors on 3 B and Σ commute; i.e.
γµα γα = γαµ γα . Now the covariant derivative is projected onto Σ because of the two
ρ ρ
β ρ
projector tensors γν γα ∇ρ n β = Dα nν . Therefore the first term is simply the extrinsic
curvature of B as embedded in Σ defined in (3.10)
− γαµ Dα nν = −σµα Dα nν = Kµν . (C.3)
Both spacetime tensors Kµν and Dα nν are tangent to Σ. The second term can be
manipulated according to the differentiation of the orthogonal condition between
normal vectors u β ∇ρ n β + n β ∇ρ u β = 0 to be proportional to the acceleration a β =
uρ ∇ρ u β of the timelike hypersurface normal; this is
ρ
uµ uν uα γα n β ∇ρ u β = uµ uν n β a β . (C.4)
Finally, the last term in eq. (C.1) is related to the extrinsic curvature of Σ. In fact,
ρ ρ ρ ρ
using these relationships, γα u β ∇ρ n β = −γα n β ∇ρ u β and γµα γα = σµα γα on 3 B we
obtain for the second term in (C.1) an expression related to the extrinsic curvature
138 Appendix C. Extrinsic curvature of 3 B
defined in (3.4)
− 2γ(αµ uν) γα n β ∇ρ u β = −2σ(αµ uν) n β γα ∇ρ u β = 2σ(αµ uν) n β Θαβ .

ρ ρ
(C.5)
Plugging everything together, the extrinsic curvature can be written as
Θµν = Kµν + uµ uν n β a β + 2σ(αµ uν) n β Θαβ . (C.6)
The first term gives the projection onto B; the second term, the projection along the
normal uµ ; and the two symmetrized last terms give the off-diagonal projection of
Θµν . This components can be written as the off-diagonal projection of the hypersur-
face extrinsic curvature Θµν making use of the relationship σµα n β Θαβ = −σµα u β Θαβ .
From these decomposition the trace is directly obtained,
Θ = γµν Θµν = K − n β a β . (C.7)
It is also useful to employ the decomposition for the extrinsic curvature of 3 B in

(C.6) to rewrite the boundary momentum π ij in (3.17). Replacing one identifies the
gravitational momentum Pij conjugate to γij and finally write
√ i 2N √σ
ijN σ h ij (i
π = K + (n · a)σ − K γ − √ σα u j) Pαβ n β .
ij ij
(C.8)
2κ γ
Let us remark the difference in the tensor indices; in the latter expression i, j refer to
coordinates on 3 B while α, β are related to coordinates in the slices Σ.
139
Appendix D
Cosmological horizon for a dust

universe
Let us make a short comment on the equation that relates the density of dust and
the variables t and r in (5.38). Once the temporal function C (t) = A t3/2 is fixed, the
following change of variables
r
3 2 3 t2
u = r κρ , x=A (D.1)
r2
makes the relation (5.38) to correspond with the roots of an extended function
f (u) = u3 − xu + 6 . (D.2)
The procedure goes as follows: at any particular time t, the density profile is deter-
mined at each point r. The density takes a value such that the function f vanishes,
p
f (u = 3 r2 κρ) = 0. Or in other words, at a certain point of the spacetime, the
relation between the magnitudes u and x is given by the roots of the function f (u).
The new function accomplish f (−∞) = −∞ and f (0) = 6, hence at least one
of the roots of f (u) has a real and negative value, u1 < 0. This root of (D.2) has no
physical interpretation because as it has been defined in (D.1), u must be positive.
Therefore, we know that a positive solution must exist; as f (∞) = ∞ there must
be at least one double root (or two more roots). To find it (them), let’s analyze the
function f (u). The critical points are obtained by the roots of the first derivative of
f (u) r
0 x
f (uc ) = 3 u2c −x =0 =⇒ u c± = ± . (D.3)
3
The second derivative classifies whether the critical points are a maximum, a mini-
mum or a saddle point
f 00 (uc ) = 6uc . (D.4)
√ √
Therefore, being also x > 0, uc+ = x/3 is a local minimum and uc− = − x/3 is a
local maximum. We are interested in u > 0. The next step is to locate the image of
the positive critical point uc+ (the minimum), because this value will determine the
140 Appendix D. Cosmological horizon for a dust universe
existence of the wanted remaining root(s)
f ( uc+ ) = u3c+ − 3 u3c+ + 6 ≤ 0 . (D.5)
We are searching for positive real roots of (D.2), hence the value of the minimum
must be zero or negative (one double root or two simple roots). It is interesting
to note that the latter bound can be written as 35/3 ≤ x. Bringing the value of x
back, we will see that the previous constraint reflects the presence of a horizon. The
mere existence of a positive solution for f (u) = 0 with u ∝ ρ entails the presence
√
of a horizon in such coordinates. With the value of the constant A = 3 3 6, the
cosmological horizon equation is written as
r
r 2
≤ . (D.6)
t 3
141
Appendix E
Energy momentum tensor for dust

in static coordinates
In this appendix we will give the components of the stress-energy tensor explicitly
for a dust universe in the new set of coordinates. Dust is a pressureless perfect
fluid with an energy-momentum tensor given by (5.31); this definition is coordinate
independent. Once the coordinates are fixed one can find the explicit stress-energy
tensor by means of Einstein’s equations (1.1). The Einstein tensor is straightforward
once the metric is chosen; from (5.37) for instance the computations are particularly
simple –it would be equivalent to use the metric (5.41), at this point ∂t ρdust is already
know.
The product of the two nonnull off-diagonal components of the Einstein tensor
lead us to
(κρdust )3 r2
Gt r Gr t = − 2 . (E.1)
κρ
3 1 − 3dust r2
Notice that Gt r 6= Gr t . The right hand side of Einstein’s equations equals the previ-
ous relation to
κ 2 Tt r Tr t = (κρdust )2 Ut U t Ur U r . (E.2)
We aim at solving the 4–velocity and we have another equation to do so: the nor-
malization condition over the 4–velocity that yields
gµν U µ U ν = Ut U t + Ur U r = 1 . (E.3)
The system of two equations: (E.1) = (E.2) and (E.3) has the following solution
1
Ut U t = κρdust ; (E.4)
1− 3 r2
κρdust r2
Ur U r = − κρdust 2 . (E.5)
3 1− 3 r
We have chosen the solution noticing that Ut U t > 0 and Ur U r < 0. The next step
is to isolate each component of the velocity; using the prescription Uµ 2 = Uµ U µ gµν ,
142 Appendix E. Energy momentum tensor for dust in static coordinates
for µ = t, r we obtain

∂t ρdust
Ut = ; (E.6)
(3 κρ3dust )1/2
r
κρdust r
Ur = κρdust 2 . (E.7)
3 1− 3 r
A relevant comment is that if the temporal Einstein’s equation
r ∂r ρdust + 3 ρ
Gt t = κ = κ ρ Ut U t (E.8)
3
is combined with the corresponding velocity in equation (E.4), and then one solves
for the radial derivative one gets (5.36).
Finally,

t 3 ρdust r
∂t ρdust
Tt = Tt = − r
3 − κρdust r2 3
(E.9)
9 κρ3 κρ2
Tr t = dust
r Tr r = − dust
r2 .
∂t ρdust 3 − κρdust r2 2 3 − κρdust r2

143
Bibliography
[1] A. Einstein, "Die Feldgleichungen der Gravitation", Sitzungsberichte der

Preussischen Akademie der Wissenschaften zu Berlin, 844–847. Retrieved 2017-
08-21. (November 25, 1915).
[2] C.W. Misner, K.S. Thorne and J.A. Wheeler, Gravitation (Freeman, San Fran-
cisco, 1973).
[3] K. Schwarzschild, Über das Gravitationsfeld eines Massenpunktes nach der Einstein-
schen Theorie, Sitzungsber. Preuss. Akad. Wiss. 7, 189 (1916).
For a translation see S. Antoci and A. Loinger, On the gravitational field of a mass
point according to Einstein’s theory (1999); arXiv:physics/9905030.
[4] D. Finkelstein, Past-Future Asymmetry of the Gravitational Field of a Point Particle,

Phys. Rev. 110, 965 (1958).
[5] P.K. Townsend, Black Holes. Lecture notes for a ’Part III’ course ’Black Holes’
given in DAMTP, Cambridge. The course covers some of the developments in
Black Hole physics of the 1960s and 1970s, (1997).
[6] J.D. Bekenstein, Black Holes and Entropy, Phys. Rev. D 7, 2333 (1973).
[7] S.F. Ross, Black hole thermodynamics; arXiv:hep-th/0502195.
[8] S.W. Hawking, Particle Creation by Black Holes, Commun. Math. Phys. 43, 199
(1975).
[9] S.W. Hawking, Breakdown of predictability in gravitational collapse, Phys. Rev. D

14, 2460 (1976).
[10] G. ’tHooft, Class. Quant. Grav. 16, 3263 (1999).
[11] J.F. Donoghue, General relativity as an effective field theory: The leading quantum
corrections, Phys. Rev. D 50, 3874 (1994); gr-qc/9405057.
J.F. Donoghue, Leading quantum correction to the Newtonian potential, Phys. Rev.
Lett. 72, (1994) 2996; gr-qc/9310024.
N.E.J. Bjerrum-Bohr, J.F. Donoghue and B.R. Holstein, Quantum corrections to the
Schwarzschild and Kerr metrics, Phys. Rev. D 68, 084005 (2003); hep-th/0211071.
Erratum: Phys. Rev.D 71, 069904 (2005).
144 BIBLIOGRAPHY
[12] K.G. Wilson, The Renormalization Group and Strong Interactions, Phys. Rev. D 3,
1818 (1971).
K.G. Wilson and J.B. Kogut, The Renormalization group and the epsilon expansion,
Phys. Rept. 12, 75 (1974).
[13] E. Fermi, An attempt of a theory of beta radiation. I, Z. Phys. 88, 161 (1934).
[14] D.J. Griffiths, Introduction to elementary particles, 2nd rev. ed. (Wiley-VCH, Wein-
heim, 2008).
[15] G. Dvali and C. Gomez, Self-Completeness of Einstein Gravity; arXiv:1005.3497.
[16] G. Dvali, G.F. Giudice, C. Gomez and A. Kehagias, UV-Completion by Classical-

ization, JHEP 08, 108 (2011); arXiv:1010.1415.
G. Dvali, Strong Coupling and Classicalization, Subnucl. Ser. 53, 189 (2017);
arXiv:1607.07422.
[17] G. Veneziano, Europhys. Lett. 2, 199 (1986).

D.J. Gross and P.F. Mende, Nucl. Phys. B 303, 407 (1988).
D. Amati, M. Ciafolini and G. Veneziano. Phys. Lett. B 216, 41 (1989).
K. Konishi, G. Paffuti and P. Provero, Phys. Lett. B 234, 276 (1990).
E. Witten, Phys. Today, Apr. 24 (1996).
[18] T. Banks, A critique of pure string theory: Heterodox opinions of diverse dimensions;
hep-th/0306074.
T. Banks. hep-th/0305206.
L. Susskind. Twenty years of debate with Stephen; hep-th 0204027.
[19] A.D. Sacharov, Sov. Phys. JETP 22, 241 (1966).

I. Dymnikova, Int. J. Mod. Phys. D 12, 1015 (2003).
[20] G. Chapline, E. Hohlfield, R.B. Laughlin and D.I. Santiago, Philos. Mag. B 81,
235 (2001).
[21] P.O. Mazur and E. Mottola, Proc. Nat. Acad. Sci. 101, 9545 (2004).
[22] G. Dvali, C. Gomez, R.S. Isermann, D. Lüst and S. Stieberger, Black hole formation
and classicalization in ultra-Planckian 2 → N scattering, Nucl. Phys. B 893, 187
(2015); arXiv:1409.7405.
G. Dvali, C. Gomez and A. Kehagias, Classicalization of Gravitons and Goldstones,
JHEP 11, 070 (2011); arXiv:1103.5963.
[23] K.S. Thorne, Nonspherical Gravitational Collapse: A Short Review, in: Magic With-
out Magic: John Archibald Wheeler, A Collection of Essays in Honor of his
Sixtieth Birthday. Edited by John R. Klauder. San Francisco: W.H. Freeman.
BIBLIOGRAPHY 145
E. Flanagan, sl Hoop conjecture for black-hole horizon formation, Phys. Rev. D

44, 2409 (1991).
[24] R. Casadio, O. Micu and F. Scardigli, Quantum hoop conjecture: Black hole forma-
tion by particle collisions, Phys. Lett. B 732, 105 (2014); arXiv:1311.5698 [hep-th].
[25] G. Dvali and C. Gomez, Black Hole’s Quantum N–Portrait, Fortsch. Phys. 61, 742
(2013); arXiv:1112.3359.
[26] G. Dvali and C. Gomez, Black Hole’s 1/N Hair, Phys. Lett. 719, 419 (2013);
arXiv:1203.6575.
[27] G. Dvali and C. Gomez, Black Holes as Critical Point of Quantum Phase Transition,
Eur. Phys. J. C 74, 2752 (2014); arXiv:1207.4059.
D. Flassig, A. Pritzel and N. Wintergerst, Black Holes and Quantumness on Macro-
scopic Scales, Phys. Rev. D 87, 084007 (2013); arXiv:1212.3344.
G. Dvali, D. Flassig, C. Gomez, A. Pritzel and N. Wintergerst, Scrambling in the
Black Hole Portrait, Phys. Rev. D 88, 124041 (2013); arXiv:1307.3458.
G. Dvali, A. Franca, C. Gomez and N. Wintergerst, Nambu-Goldstone Effective
Theory of Information at Quantum Criticality, Phys. Rev. D 92, 125002 (2015);
arXiv:1507.02948.
[28] N.N. Bogoliubov, On the theory of superfluidity, J. Phys. (USSR) 11, 23 (1947). Izv.
Akad. Nauk Ser. Fiz. 11, 77(1947).
[29] J. Alfaro, D. Espriu and L. Gabbanelli, Bose−Einstein graviton condensate in a

Schwarzschild black hole, Class. Quant. Grav. 35, 015001 (2018); arXiv:1609.01639.
[30] J. Alfaro, D. Espriu and L. Gabbanelli, Condensates beyond the horizons;

arXiv:1905.01080.
[31] A. Einstein, Kosmologische Betrachtungen zur allgemeinen Relativitätstheorie,

Sitzungsber. Königl. Preuss. Akad. Wiss., 142 (1917).
[32] E. Hubble, A relation between distance and radial velocity among extra-galactic neb-
ulae, Proc. Nat. Acad. Sci. 15, 168 (1929).
[33] N. Aghanim et al. (Planck Collaboration), Planck 2018 results. VI. Cosmological
parameters, (2018); arXiv:1807.06209.
[34] A.G. Riess et al., Milky Way Cepheid Standards for Measuring Cosmic Distances and
Application to Gaia DR2: Implications for the Hubble Constant, ApJ 861, 126 (2018);
arXiv:1804.10655.
[35] K.C. Wong et al., H0LiCOW XIII. A 2.4% measurement of H0 from lensed quasars:
5.3σ tension between early and late-Universe probes (2019); arXiv:1907.04869.
L. Verde, T. Treu and A.G. Riess, Tensions between the Early and the Late Universe
(2019); arXiv:1907.10625.
146 BIBLIOGRAPHY
[36] G. Gamow, My World Line, (Viking Adult, New York 1970).

"Much later, when I was discussing cosmological problems with Einstein, he re-
marked that the introduction of the cosmological term was the biggest blunder
he ever made in his life."
[37] E. Bianchi, C. Rovelli and R. Kolb, Is dark energy really a mystery?, Nature 466,
321 (2010).
E. Bianchi and C. Rovelli, Why all these prejudices against a constant?;
arXiv:1002.3966.
[38] A. Einstein, Die Grundlage der allgemeinen Relativitätsththeorie Ann. Phys. br 49,
50 (1916). English version: The foundations of the general theory of relativity, in The
collected papers of Albert Einstein, Vol. 6, A.J. Kox, M.J. Klein, R. Schulmann
editors (Princeton University Press 1997).
[39] A.S. Eddington, On the Instability of Einstein’s Spherical World, MNRAS 90, 668
(1930).
[40] A.G. Riess et al. (Supernova Search Team Collaboration), Observational Evidence
from Supernovae for an Accelerating Universe and a Cosmological Constant, Astron.
J. 116, 1009 (1998); arXiv:astroph/9805201.
S. Perlmutter et al. (Supernova Cosmology Project Collaboration), Measurements
of Ω and Λ from 42 High-Redshift Supernovae, ApJ 517, 565 (1999); arXiv:astro-
ph/9812133.
E. Linder and S. Perlmutter, Dark energy: the decade ahead, Phys. World 20, 24
(2007).
[41] T.M.C. Abbott et al. (DES Collaboration), First Cosmology Results using Type Ia
Supernovae from the Dark Energy Survey: Constraints on Cosmological Parameters,
The Astrophysical Journal Letters 872, L30 (2019); arXiv:1811.02374.
[42] A.G. Riess, The Case for an Accelerating Universe from Supernovae, Publ. Astron.
Soc. Pac. 112, 1284 (2000); astro-ph/0005229.
M. Kowalski et al. (Supernova Cosmology Project Collaboration), Improved Cos-
mological Constraints from New, Old and Combined Supernova Datasets, ApJ 686,
749 (2008); arXiv:0804.4142.
W.M. Wood–Vasey et al. (ESSENCE Collaboration), Observational Constraints on
the Nature of the Dark Energy: First Cosmological Results from the ESSENCE Super-
nova Survey, ApJ 666, 694 (2007); astro-ph/0701041.
A.G. Riess and M. Livio, The First Type Ia Supernovae: An Empirical Approach to
Taming Evolutionary Effects in Dark Energy Surveys from SNe Ia at z > 2, ApJ 648,
884 (2006); astro-ph/0601319.
BIBLIOGRAPHY 147
A.G. Riess et al. (Supernova Search Team Collaboration), Type Ia Supernova Dis-
coveries at z > 1 from the Hubble Space Telescope: Evidence for Past Deceleration and
Constraints on Dark Energy Evolution, ApJ 607, 665 (2004); astro-ph/0402512.
D.N. Spergel et al. (WMAP Collaboration), Wilkinson Microwave Anisotropy Probe
(WMAP) Three Year Results: Implications for Cosmology, Astrophys. J. Suppl. 170,
377 (2007); astro-ph/0603449.
D.J. Eisenstein et al. (SDSS Collaboration), Detection of the Baryon Acoustic Peak
in the Large-Scale Correlation Function of SDSS Luminous Red Galaxies, ApJ 633,
560 (2005); astro-ph/0501171.
H.J. Seo and D.J. Eisenstein, Probing Dark Energy with Baryonic Acoustic Os-
cillations from Future Large Galaxy Redshift Surveys, ApJ 598, 720 (2003); astro-
ph/0307460.
L. Fu et al., Very weak lensing in the CFHTLS wide: cosmology from cosmic shear in
the linear regime, Astronomy & Astrophysics 479, 9 (2008).
L. Guzzo et al., A test of the nature of cosmic acceleration using galaxy redshift distor-
tions, Nature 451, 541 (2008); arXiv:0802.1944.
[43] M. Sereno and P. Jetzer, Phys. Rev. D 73, 063004 (2006); astro-ph/0602438.
A. Balaguera-Antolinez, C.G. Boehmer, and M. Nowakowski, Class. Quant.
Grav. 23, 485 (2006); gr-qc/0511057.
L. Iorio, Adv. Astron. Astrophys., doi: 10.1155/2008/268647 (2008); New As-
tron. Rev. 14, 196 (2009); arXiv:0808.0256.
[44] Y. Suto, Prog. Theor. Phys. 90, 1173 (1993); also in the Proceedings of Asia Pacific
Center for Theoretical Physics Inaugration Conference, RESCEU-22/96, UTAP-
235/96 (1996); astro-ph/9609014.
P.J. Peebles, Astrophys. Sp. Sci. 45, 3 (1976); The Large Scale Structure of The Uni-
verse (Princeton University Press, New Jersey, 1980).
G.S. Adkins, J. McDonnell, and R.N. Fell, Phys. Rev. D 75, 064011 (2007); gr-
qc/0612146.
[45] I.B. Khriplovich and A.A. Pomeransky, Int. J. Mod. Phys. D 17, 2255 (2008).
M. Park, Phys. Rev. D 78, 023014 (2008).
F. Simpson, J.A. Peacock, and A.F. Heavens, On lensing by a cosmological constant,
MNRAS 402, 2010; arXiv:0809.1819.
[46] W. Rindler and M. Ishak, Phys. Rev. D 76, 043006 (2007).

M. Ishak, W. Rindler, and J. Dossett, More on Lensing by a Cosmological Constant,
MNRAS 403, 2152 (2010); arXiv:0810.4956.
G.W. Gibbons, C.M. Warnick and M.C. Werner, Class. Quant. Grav. 25, 245009
(2008).
148 BIBLIOGRAPHY
T. Schücker, Gen. Rel. Grav. 42, 1991 (2010); arXiv:0903.2940.

M. Ishak, W. Rindler, J. Dossett, J. Moldenhauer, and C. Allison, MNRAS 388,
1279 (2008).
[47] M. Sereno, Phys. Rev. Lett. 102, 021301 (2009); Phys. Phys. Rev. D 81, 084002
(2010); Rev. D 77, 043004 (2008).
R. Kantowski, B. Chen and X. Dai, Gravitational Lensing Corrections in Flat Lamb-
daCDM Cosmology, ApJ 718, 913 (2010); arXiv:0909.3308.
[48] J. Bernabeu, C. Espinoza and N.E. Mavromatos, Cosmological Constant and Local
Gravity, Phys. Rev. D 81, 084002 (2010); arXiv:0910.3637.
[49] J. Bernabeu, D. Espriu, D. Puigdomènech, Gravitational waves in the presence of a

cosmological constant, Phys. Rev. D 84, 063523 (2011); arXiv:1106.4511.
[50] J. Alfaro, D. Espriu and L. Gabbanelli, On the propagation of gravitational waves

in a ΛCDM universe, Class. Quant. Grav. 36, 025006 (2019); arXiv:1711.08315.
[51] D. Espriu, L. Gabbanelli and M. Rodoreda, Measuring H0 in PTA with gravita-

tional waves; arXiv:1908.08472.
[52] I. Cognard and D.C. Backer, A Micro-glitch in the Millisecond Pulsar B1821−24 in
M28, ApJ 612, L125 (2004); astro-ph/0407546.
J.W. Mckee, G.H. Janssen, B.W. Stappers, A.G. Lyne, R.N. Caballero et al., A
glitch in the millisecond pulsar J0613˘0200, MNRAS 461, 2809 (2016).
[53] R.N. Manchester (for the IPTA), Class. Quant. Grav. 30, 224010 (2013).
G. Hobbs, A. Archibald, Z. Arzoumanian, D. Backer, M. Bailes, N.D.R. Bhat,
M. Burgay, S. Burke-Spolaor, D. Champion, I. Cognard and W. Coles, The Inter-
national Pulsar Timing Array project: using pulsars as a gravitational wave detector,
Class. Quant. Grav. 27, 084013 (2010); arXiv:0911.5206.
[54] The LIGO Scientific Collaboration et al., Class. Quant. Grav. 32, 074001 (2015);
arXiv:1411.4547.
[55] F. Acernese et al., Class. Quant. Grav. 32, 024001 (2015).
[56] R.M. Shannon, V. Ravi, L.T. Lentati et al., Science 349, 1522 (2015).
[57] G.H. Janssen et al., Gravitational wave astronomy with the SKA, PoS AASKA 14,
037 (2015); arXiv:1501.00127.
J.A. Font, A.M. Sintes and C.F. Sopuerta, Gravitational waves with the SKA, in
"The Spanish Square Kilometre Array White Book", published by "Sociedad
Española de Astronomía" (2015); arXiv:1506.03474.
[58] D. Espriu and D. Puigdomènech, Local measurement of Λ using pulsar timing ar-
rays, ApJ 764, 163 (2013); arXiv:1209.3724.
BIBLIOGRAPHY 149
[59] D. Espriu, Pulsar Timing Arrays and the cosmological constant, Proceedings.
Invited talk at the 2n d Russian–Spanish meeting on particle physics at all
scales, Saint Petersburg, Russia, October 2013. AIP Conf. Proc. 1606, 86 (2014);
arXiv:1401.7925.
[60] R.N. Manchester, G.B. Hobbs, A. Teoh and M. Hobbs, ApJ 129, 1993 (2005).
ATNF pulsar catalogue, www.atnf.csiro.au/research/pulsar/psrcat.
[61] G. Dvali, C. Gomez and S. Zell, Quantum Break-Time of de Sitter, J. Cosmol. As-
tropart. Phys. 06, 028 (2017); arXiv:1701.08776.
G. Dvali and C. Gomez, Landau-Ginzburg Limit of Black Hole’s Quantum Por-
trait: Self Similarity and Critical Exponent, Phys. Lett. B 716, 240 (2012);
arXiv:1203.3372.
[62] M.K. Parikh and Frank Wilczek, Hawking Radiation as Tunneling, Phys. Rev. Lett.
85, 5042 (2000); hep-th/9907001.
J.D. Bekenstein, Generalized second law of thermodynamics in black-hole physics,
Phys. Rev. D 9, 3292 (1974).
S.W. Hawking, Black holes and thermodynamics, Phys. Rev. D 13, 191 (1976).
[63] J.D. Bekenstein, Phys. Rev. Lett. 28, 452 (1972); Phys. Rev. D 5, 1239 (1972); 5,
2403 (1972);
C. Teitelboim, Phys. Rev. D 5 (1972) 294;
J. Hartle, Phys. Rev. D 3, 2938 (1971).
[64] S.W. Hawking, Information Loss in Black Holes, Phys. Rev. D 72, 084013 (2005);
hep-th/0507171.
S.W. Hawking, The Unpredictability of Quantum Gravity, Commun. Math. Phys.
87, 395 (1982).
For a review on information loss paradox, see, J. Preskill, Do Black Holes Destroy
Information?, International Symposium on Black Holes, Membranes, Worm-
holes, and Superstrings, Houston Advanced Research Center (1992); arXiv:hep-
th/9209058.
[65] J. Alfaro, D. Espriu, D. Puigdomènech, Spontaneous generation of geometry in four

dimensions, Phys. Rev. D 86 025015 (2012); arXiv:1201.4697.
J. Alfaro, D. Espriu, D. Puigdomènech, The emergence of geometry: a two-
dimensional toy model, Phys. Rev. D 82, 045018 (2010); arXiv:1004.3664.
[66] F. Dalfovo, S. Giorgini, L.P. Pitaevskii and S. Stringari, Theory of Bose–Einstein

condensation in trapped gases, Rev. Mod. Phys. 71, 463 (1999).
[67] G.D. Birkhoff and R. Langer, Relativity and Modern Physics (Harvard Univ. Press,
1923).
150 BIBLIOGRAPHY
[68] T. Regge and J.A. Wheeler, Stability of a Schwarzschild Singularity. Phys. Rew.
108, 1063 (1957);
[69] W. Pauli and V. Weisskopf, Helv. Phys. Acta 7, 709 (1934).
[70] S. Weinberg, Cosmology, Oxford University Press, New York, First Edition
(2008).
[71] R. Arnowitt, S. Deser and C.W. Misner, Dynamical Structure and Definition of
Energy in General Relativity,Phys Rev. 116, 1322 (1959); Coordinate invariance and
energy expressions in general relativity, Phys. Rev. 122, 997 (1961).
H. Bondi, M. G. Van der Burg and A. W. K. Metzner, Proc. Roy. Soc. Lond. A
269, 21 (1962).
[72] C. Møller, Ann. Phys. 12, 118 (1961).
[73] L.B. Szabados, Class. Quant. Grav. 11, 1833 (1994); 1847 (1994); 13, 1661 (1996);
16, 2889 (1999).
[74] L.B. Szabados, Quasi-Local Energy-Momentum and Angular Momentum in General

Relativity, Living Rev. Relativ. 12, 4 (2009).
[75] J.W. York, Boundary terms in the action principles of general relativity, Found. Phys.
16, 249 (1986).
J.W. York, Role of Conformal Three-Geometry in the Dynamics of Gravitation, Phys.
Rev. Lett. 28, 1082 (1972).
[76] R. Arnowitt, S. Deser and C.W. Misner, The Dynamics of General Relativity, in
Gravitation: An Introduction to Current Research, edited by L. Witten (Wiley,
New York, 1962), pp. 227–264.
Republication of: The dynamics of general relativity, General Relativity and Gravi-
tation 40, 1997 (2008); arXiv:gr-qc/0405109.
[77] A.P. Lundgren, B.S. Schmekel and J.W. York, Self-renormalization of the classical
quasilocal energy, Phys. Rev. D 75, 084026 (2007).
[78] G.W. Gibbons and S.W. Hawking, Phys. Rev. D 15, 2752 (1977).
[79] C. Lanczos, Phys. Z. 23, 539 (1922); Ann. Phys. (Leipzig) 74, 518 (1924).
W. Israel, Nuovo Cim. B 44, (1966) 1; Erratum-ibid B 48, (1967) 463.
[80] C.W. Misner, K.S. Thorne and J.A. Wheeler, Gravitation (Freeman, San Fran-
cisco, 1973).
[81] J.D. Brown and J.W. York, Quasilocal energy in general relativity, in Mathemati-
cal Aspects of Classical Field Theory, edited by M.J. Gotay, J.E. Marsden, and
V.E. Moncrief (American Mathematical Society, Providence, RI, USA, 1992), pp.
129–142.
BIBLIOGRAPHY 151
[82] H.W. Braden, J.D. Brown, B.F. Whiting and J.W. York, Phys. Rev. D 42, 3376
(1990).
[83] J.W. York, Phys. Rev. D 33, 2092 (1986).
[84] T. Regge and C. Teitelboim, Role of Surface integrals in the Hamiltonian Formula-
tion of General Relativity, Annals of Physics 88, 286 (1974).
[85] J.D. Brown and J.W. York, Quasilocal Energy and Conserved Charges Derived from
the Gravitational Action, Phys. Rev. D 47, 1407 (1993); arXiv:gr-qc/9209012.
[86] M.-F. Wu, C.-M. Chen, J.-L. Liu and J.M. Nester, Optimal choices of reference for a
quasi-local energy, Phys. Lett. A 374, 3599 (2010).
[87] M. Blau and B. Rollier, Brown–York Energy and Radial Geodesics,Class. Quant.
Grav. 25, 105004 (2008); arXiv:0708.0321.
[88] C.W. Misner and D.H. Sharp, Relativistic Equations for Adiabatic, Spherically Sym-
metric Gravitational Collapse, Phys. Rev. 136, B571 (1964).
W.C. Hernandez and C.W. Misner, Observer time as a coordinate in relativistic
spherical hydrodynamics, Astrophys. J. 143, 452 (1966).
M.E. Cahill and G.C. McVittie, Spherical symmetry and mass-energy in general rel-
ativity I. General theory, J. Math. Phys. 11, 1382 (1970).
[89] W. Rindler, Relativity: Special, General, and Cosmological, Second Edition 2006
(Oxford: Oxford University Press).
[90] H. Reissner, Über die Eigengravitation des elektrischen Feldes nach der Einsteinschen
Theorie, Ann. Phys. 355, (1916) 106; Ann. Phys.
G. Nordström, On the Energy of the Gravitational Field in Einstein’s Theory," Proc.
Kon. Ned. Akad. Wet. 20, 1238 (1918).
[91] G.W. Gibbons and S.W. Hawking, Cosmological Event Horizons, Thermodynamics,
and Particle Creation, Phys. Rev. D 15, 2738 (1977).
[92] O. Lahav and A.R. Liddle, The Cosmological Parameters 2010, review arti-
cle for The Review of Particle Physics 2010 (aka the Particle Data Book);
arXiv:1002.3488.
[93] S. Weinberg, Gravitation and cosmology: Principles and applications of the

general theory of relativity, (Wiley, New York, 1972).
C.W. Misner, K.S. Thorne, and J.A. Wheeler, Gravitation, (Freeman, San Fran-
cisco, 1973).
P.J.E. Peebles, Principles of physical cosmology, (Princeton, 1993).
[94] P. Ntelis et al., Exploring cosmic homogeneity with the BOSS DR12 galaxy sample,
JCAP 06, 019 (2017); arXiv:1702.02159.
152 BIBLIOGRAPHY
[95] P.J.E. Peebles and B. Ratra, The Cosmological Constant and Dark Energy, Rev. Mod.
Phys. 75, 559 (2003); arXiv:astro-ph/0207347.
[96] M. Kowalski et al. (The Supernova Cosmology Project), Improved Cosmological

Constraints from New, Old and Combined Supernova Datasets, ApJ 686, 749 (2008);
arXiv:0804.4142.
[97] R.G. Carlberg, H.K.C. Yee and E. Ellingson, The Average Mass-to-Light Profile of
Galaxy Clusters, ApJ 478, 462 (1997); arXiv:astro-ph/9512087.
[98] J.J. Mohr, B. Mathiesen and A.E. Evrard, Properties of the Intracluster Medium
in an Ensemble of Nearby Galaxy Clusters, ApJ 517, 627 (1999); arXiv:astro-
ph/9901281.
L. Grego, J.E. Carlstrom, E.D. Reese, G.P. Holder, W.L. Holzapfel, M.K. Joy, J.J.
Mohr and S. Patel, Galaxy Cluster Gas Mass Fractions from Sunyaev-Zel’dovich Ef-
fect Measurements: Constraints on Ω M , ApJ 552, 2 (2001); arXiv:astro-ph/0012067.
[99] Y. Mellier, Probing the Universe with Weak Lensing, Ann. Rev. Astron. Astrophys.
37, 127 (1999); arXiv:astro-ph/9812172.
G. Wilson, N. Kaiser and G.A. Luppino, Mass and Light in the Universe, ApJ 556,
601 (2001); astro-ph/0102396.
[100] Z.L. Wen and J.L. Han, Galaxy clusters at high redshift and evolution of brightest
cluster galaxies, ApJ 734, 68 (2011); arXiv:1104.1667.
[101] G. Hinshaw et al., (WMAP Collaboration), Five-Year Wilkinson Microwave

Anisotropy Probe Observations: Data Processing, Sky Maps, and Basic Results, As-
trophys. J. Suppl. 180, 225 (2009); arXiv:0803.0732.
[102] The spherically-symmetric Schwarzschild–de Sitter solution was first dis-

cussed in F. Kottler, Ann. Phys. 361, 401 (1918).
[103] T.P. Cheng. Relativity, Gravitation, and Cosmology – A Basic Introduction.

Oxford University Press, New York, Second Edition (2010).
[104] J.A. Frieman, M.S. Turner and D. Huterer, Dark Energy and the Accelerating Uni-
verse, Ann. Rev. Astron. Astrophys. 46, 385 (2008).
[105] T. Gold, Nature 218, 731 (1968).

D.W. Richards and J.M. Comella, Nature 222, 551 (1969).
V. Radhakrishnan, D.J. Cooke, M.M. Komesaroff and D. Morris, Nature 221, 443
(1969).
[106] A. Hewish, S.J. Bell, J.D.H. Pilkington, P.F. Scott and R.A. Collins, Nature 217,
709 (1968).
J.D.H. Pilkington, A. Hewish, S.J. Bell and T.W. Cole, Nature 218, 128 (1968).
BIBLIOGRAPHY 153
[107] V.Radhakrishnan and G. Srinivasan, Curr. Sci. 51, 1096 (1982).

M.A. Alpar, A.F. Cheng, M.A. Ruderman, J. Shaham, Nature 300, 728 (1982).
[108] R.W. Hellings and G.S. Downs, Upper limits on the isotropic gravitational radia-
tion background from pulsar timing analysis, ApJ 265, L39 (1983).
[109] F.A. Jenet, G.B. Hobbs, K.J. Lee and R.N. Manchester, Detecting the stochastic
gravitational wave background using pulsar timing, ApJ 625, L123 (2005); astro-
ph/0504458.
[110] M.V. Sazhin, Soviet Astron. 22, 36 (1978).

S. Detweiler, ApJ 234, 1100 (1979).
B. Bertotti, B.J. Carr and M.J. Rees, MNRAS 203, 945 (1983).
[111] X. Deng and L.S. Finn, Pulsar timing array observations of gravitational wave
source timing parallax, MNRAS 414, 50 (2011); arXiv:1008.0320.
[112] F.A. Jenet et al., Upper bounds on the low-frequency stochastic gravitational wave
background from pulsar timing observations: current limits and future prospects, ApJ
653, 1571 (2006); astro-ph/0609013.
[113] M. Anholm et al., Optimal strategies for gravitational wave stochastic background
searches in pulsar timing data, Phys. Rev. D 79, 084030 (2009).
[114] J. Alfaro and M. Gamonal, An alternative way to locally measure the Hubble con-
stant using Gravitational Waves and PTA; arXiv:1902.04550.
[115] G.A. Tammann, B. Reindl, F. Thim, A. Saha and A. Sandage, in A New Era in
Cosmology, edited by T. Shanks and N. Metcalf, Astron. Soc. Pac. Conf. Proc. (in
press) 2001.
[116] J.P.W. Verbiest et al., The International Pulsar Timing Array: First Data Release,
MNRAS 458, 1267 (2016); arXiv:1602.03640.
[117] R.W. Romani, in H. Ögelman, E.P.J. van den Heuvel, eds, Timing Neutron Stars
Timing a millisecond pulsar array, Kluwer, Dordrecht, 113 (1989).
R.S. Foster and D.C. Backer, ApJ 361, 300 (1990).
[118] R.N. Manchester, G. Hobbs, M. Bailes, W.A. Coles, W. van Straten and M.J.
Keith, PASA 30, 17 (2013).
[119] Z. Arzoumanian, A. Brazier, S. Burke-Spolaor, S. Chamberlin, S. Chatterjee,

B. Christy, J. Cordes, N. Cornish, P. Demorest, X. Deng, T. Dolch, J. Ellis, R.
Ferdman, E. Fonseca, N. Garver-Daniels et al., ApJ 821, 23 (2016).
[120] G. Desvignes, R.N. Caballero, L. Lentati, J.P.W. Verbiest, D.J. Champion, B.W.
Stappers, G.H. Janssen, P. Lazarus et al., MNRAS 458, 3341 (2016).
154 BIBLIOGRAPHY
[121] R.S. Foster and D.C. Backer, ApJ 361, 300 (1990).
V.M. Kaspi, J.H. Taylor and M. Ryba, ApJ 428, 713 (1994).
[122] A. Sesana, A. Vecchio and M. Volonteri, MNRAS 394, 2255 (2009).
[123] N. Seto, MNRAS 400, L38 (2009).

R. van Haasteren and Y. Levin, MNRAS 401, 2372 (2010).
M.S. Pshirkov, D. Baskaran and K.A. Postnov, MNRAS 402, 417 (2010).
M. Favata, ApJ 696, L159 (2009).
[124] S. Babak, A. Petiteau, A. Sesana, P. Brem, P.A. Rosado, S.R. Taylor, A. Lassus,
J.W.T. Hessels, C.G. Bassa et al., MNRAS 455, 1665 (2015).
[125] Z. Arzoumanian, A. Brazier, S. Burke-Spolaor, S.J. Chamberlin, S. Chatterjee,

J.M. Cordes, P.B. Demorest, X. Deng, ApJ 794, 141 (2014).
[126] X.-J. Zhu, G. Hobbs, L. Wen, W.A. Coles, J.-B. Wang, R.M. Shannon, R. N.
Manchester, M. Bailes, N.D.R. Bhat, S. Burke-Spolaor, S. Dai, M.J. Keith, M.
Kerr, Y. Levin, D.R. Madison, S. Os lowski, V. Ravi, L. Toomey, W. van Straten,
MNRAS 444, 3709 (2014).
[127] J.B. Wang, G. Hobbs, W. Coles, R.M. Shannon, X.J. Zhu, D.R. Madison, M.
Kerr, V. Ravi, M.J. Keith, R.N. Manchester, Levin Y., Bailes M., N.D.R. Bhat, S.
Burke-Spolaor, S. Dai, S. Os lowski, W. van Straten, L. Toomey, N. Wang, L.
Wen, MNRAS 446, 1657 (2015).

Analysis of Some Classical and Quantum Aspects of Black Hole Physics

Uploaded by

Copyright:

Available Formats

Analysis of Some Classical and Quantum Aspects of Black Hole Physics

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Analysis of Some Classical and Quantum Aspects of Black Hole Physics

Uploaded by

Copyright:

Available Formats

arXiv:2004.

07104v1 [gr-qc] 15 Apr 2020

Analysis of some classical and quantum aspects

Thesis submitted in fulfillment of the requirements for the degree of Philosophiæ

PhD programme in Physics. Line of research: Particle Physics and Gravitation.

This thesis was supervised and tutorized by

Domènec Espriu Climent

Department of Quantum Physics and Astrophysics & Institute of Cosmos Sciences

Barcelona, September 2019

To my chosen family, for being here.

2 Black holes as Bose–Einstein Condensates 19

3 Quasilocal energy for condensate structures 43

4 Condensates beyond horizons 59

4.2.1 Quasilocal energy of a Reissner–Nordström black hole . . . . . 62

5 Gravitational waves propagating over nonempty backgrounds 69

6 Local measurements of Λ with pulsar timing array 93

List of publications 119

Abstract in Spanish 121

Resume in Spanish 123

A Perturbative aspects of the classical theory 131

B Action variation 135

C Extrinsic curvature of 3 B 137

D Cosmological horizon for a dust universe 139

E Energy momentum tensor for dust in static coordinates 141

1.1 From general relativity to black holes

it is understood as the limit case of a static and spherically symmetric solution of

signature in chapters 5 and 6.

(where nowadays the event horizon of a black hole is situated).

1.2 Self-completion via classicalization

Lint. ∼ GF (ΨΨ)2 , (1.7)

1.3 Black holes as self sustained quantum systems

of soft virtual gravitons. Each of them is characterized by a typical quantum size

and in the mean-field approximation (N  1), the semi-classical behavior of macro-

phases: for αN  1, the system is in the phase in which collective graviton-graviton

Furthermore, at the critical point the system is characterized by collective Bo-

F IGURE 1.1: Leading order process responsible for quantum depletion of

1.4 Back to the geometry

1.5 The accelerating universe

Gαβ − Λ gαβ = κ Tαβ . (1.20)

have led him to predict the cosmic expansion in 1917.

Λ ∼ 1.11 × 10−52 m−2 , (1.22)

1.6 Gravitational waves propagating through an expanding

in Chapter 5, there is much more to say about the consequences of we f f 6= k e f f .

1.7 Gravitational wave detection

included as an ad hoc redshift in the frequency. Nonetheless, the propagation is over

Black holes as Bose–Einstein

2.1 Summary of our current understanding

these results stem from the basic relation postulated in [25]

2.2 Building up a condensate over a Schwarzschild black hole

gαβ = g̃αβ + hαβ . (2.2)

2.2.1 Einstein’s equations as Gross–Pitaevskii equations

The familiar Gross–Pitaevskii equation [66] employed to describe Bose–Einstein con-

by analogy, in the present case

Therefore an appropriate action for the field hαβ is

that can be interpreted as the Gross–Pitaevskii equations describing the properties

Subsequently, we will proceed to find acceptable solutions of these equations and

2.3 Outside the horizon

2.3.1 Asymptotic analysis

and in the mean-field approximation (N 1), the semi-classical behavior of macro-

phases: for αN 1, the system is in the phase in which collective graviton-graviton