UNIVERSITY OF CINCINNATI
Consequences of quantum mechanics in general relativity
by Souvik Sarkar
A thesis submitted to the graduate school in partial fulfillment for the degree of Doctor of Philosophy
in the Department of Physics Cenalo Vaz, L.C.R. Wijewardhana
August 2018 UNIVERSITY OF CINCINNATI
Abstract
Cenalo Vaz, L.C.R. Wijewardhana
Department of Physics
Doctor of Philosophy
by Souvik Sarkar
We have presented a gauge theoretic approach for gravity in 2 + 1-D, starting from the
Chern-Simons action with the SO(2, 2) and ISO(2, 1) Lie group structures. With these groups, we are able to construct some known results from general relativity namely the
Banados, Teteilboim and Zanelli (BTZ) solution and the spinning point particle solution respectively. We obtain the canonical structure in each case. To get a more physi- cal picture we probe the canonical structure by embedding the Arnowitt-Deser-Misner
(ADM) metric into the corresponding spacetime. After several canonical transformations we found in each case that the spacetime is described by a two dimensional phase space with two degrees of freedom, the mass and the angular momentum. These are stationary solutions. In the second part of this thesis, we have extended this approach to non- rotating dynamical collapse. During a spherical collapse the end state does not always lead to a black hole. Instead, due to quantum effects, collapsing shells in the exterior of the apparent horizon are accompanied by outgoing Unruh radiation in its interior. Both collapsing shells and the outgoing Unruh radiation appear to stop at the apparent hori- zon. This solution is obtained by solving the Wheeler-DeWitt equation (WDW) which is the Hamiltonian constraint, elevated to operator form by applying Dirac’s quantization procedure. As a consequence, we can expect that the collapse process would end in a quasi-stable, static compact object without forming a black hole. We have shown that, using Einstein equations, such a stable configuration is possible where the BTZ horizon radius is 87% of the boundary radius, so the BTZ horizon lies inside the boundary. In the last part, we consider a scalar field model of dark matter(DM) which forms a Bose
Einstein condensate. We coupled a non-relativistic scalar field with φ4 interactions to linearized gravity in the Gravitoelectromagnetic formulation. This model predicts that halo sizes vary significantly depending on the mass of the scalar field particles and the nature of the self-interaction. Dark matter halos ranging from asteroid size to galaxy size are possible within this model. This was one of the major challenges in the Cold Dark
Matter (CDM) paradigm. We examined the stability of the halos by studying small os- cillations about the equilibrium using a collective coordinates approach with both types of self-interaction. We found oscillations about the minima of the energy with a time period of 10 billion years for attractive interactions and 1.3 billion years for repulsive interactions.
ii iii Acknowledgements
Firstly, I would like to express my sincere gratitude to my advisors Dr. Cenalo Vaz and
Dr. L. C. R. Wijewardhana for the continuous support during my PhD study and research and for their patience, motivation, enthusiasm and knowledge. Their persistent guidance helped me in every way in my reasearch and writing of this thesis. It has been an honour to be PhD student under them. Without their help, I could not have imagined this day.
I would like to use this opportunity to thank my thesis committee members Dr. Rostislav
A. Serota, Dr. Philip Argyres, Dr. Kay Kinoshita for their time, interest and helpful comments in my research.
I would like to convey my heartiest gratitude to Dr. Paul Esposito, Dr. Peter Suranyi,
Dr. Michael Ma for their Constructive discussions and sharing their thoughts. It has considerably improved the environment to carry on the research during my PhD.
All the group sessions with my other PhD colleagues have added another polished layer in my research journey through my PhD career. I thank all of them from the core for their interest and helping me in every way possible.
Last but not the least, I would mention my parents , Sreetama and my friends for their constant support, love and encouragement to reach my goal.
It has been a wonderful journey through my research career and many more people are responsible for making it memorable. I would like to thank all of them. Their contributions in my life is unforgettable and will be cherished through the rest of my life.
Thank you...
iv Contents
Abstract i
Acknowledgements iv
List of Figures vii
Physical Constants viii
1 Introduction1
2 Chern-Simons theory and quantization of BTZ black hole8 2.1 Introduction...... 8 2.2 The Connection...... 10 2.3 Chern-Simons gravity...... 14 2.4 Equations of motion...... 19 2.4.1 Static case...... 20 2.5 Fall-off conditions and boundary action...... 23 2.6 Embedding...... 25 2.6.1 Spinning particle...... 25 2.6.2 BTZ black hole...... 27 2.7 New canonical variables...... 28 2.8 The boundary action...... 30 2.9 Summary...... 34
3 Classical dust collapse model 36 3.1 Collapse wave functionals...... 39 3.2 A quasi-classical static configuration...... 46 3.3 Energy extraction...... 48 3.4 Summary...... 49
v Contents
4 Dark matter as Bose-Einstein condensate 51 4.1 Introduction...... 51 4.2 Dark matter and BEC...... 54 4.3 The action...... 56 4.3.1 The GPP Action...... 56 4.3.2 The Gravitational Action...... 59 4.3.3 Equations of Motion...... 61 4.3.4 Stable Configuration...... 64 4.3.4.1 Total energy...... 64 4.3.4.2 Minimum energy configuration...... 67 4.3.4.3 Single vortex ansatz...... 70 4.3.4.4 Size of condensates(halos)...... 72 4.3.5 Vortex Oscillations...... 74 4.4 Summary...... 78
5 Conclusion 81
A Simplification of the Hamiltonian constraint 92
B Derivation of unique solution of WDW equation 94
C Calculations of Israel’s junction conditions 97
D The Gross-Pitaevskii equation 99
vi List of Figures
4.1 Gravitational acceleration on the equatorial plane of the BEC for our ansatz 74 4.2 Gravitational acceleration on the equatorial plane of the BEC for our ansatz 77 4.3 Gravitational acceleration on equatorial plane of the BEC for our ansatz 78
vii Physical Constants
Speed of light c = 2.997 924 58 × 108 ms−1
Reduced Planck constant ~ = 1.054 571 800 × 10−34 J.s Gravitational constant G = 6.674 08 × 10−11 Nm2kg−2
Boltzmann constant k = 1.380 650 5 × 10−23 JK−1
−35 Planck length lP = 1.616 229 × 10 m
−8 Planck mass mP = 2.176 470 × 10 kg
−44 Planck time tP = 5.391 16 × 10 s
viii To my parents. . .
ix Chapter 1
Introduction
There are four fundamental forces in nature based on its properties e.g. strong, weak, electromagnetic, and gravitational force. While the description of the first three forces is well-built on the principles of quantum mechanics on the microscopic level with a very high precision, general relativity takes into account the gravitational force on the macroscopic (galactic) scales. So far, quantum mechanics and general relativity are proved to be two of the most successful theories in modern physics to account for these fundamental forces. More specifically, general relativity discards the concept of force and introduces a new way to look at gravity as the curvature of spacetime itself, which in turn, dictates the objects to move in the shortest path in the presence of matter. On the other hand, according to quantum mechanics, the microscopic forces are mediated by quantum particles which are the quanta of a force field which pervades all over the universe. This generation of quanta out of the field is known as ‘quantization’ of force. Since gravity is simply the curvature of spacetime, quantization of gravity means quantizing spacetime itself which is difficult to realize in our usual understanding.
The task of quantizing gravity is one of the most outstanding problems of modern theo- retical physics. Many attempts have been made to reconcile these two theories starting in 1930s. Decades of hard work has enabled us to develop insights into quantum field theory and quantization of constrained systems. Despite the progress in these areas, there is still 1 Quantization of gravity no self-consistent quantum theory of gravity. There are some unavoidable reasons that make quantum gravity very challenging to formulate. They are the following:
• General relativity is a highly non-linear theory and standard power counting argu- ments indicate that the theory is nonrenormalizable. This was finally confirmed in 1986 by explicit computations.
• Observables in quantum field theory are local, wherein, quantum gravity physical observables are nonlocal.
• Quantum field theory respects causality but for quantum gravity, light cones and causal structure are themselves subject to quantum fluctuations.
• It turns out, that quantum gravity is in itself a constrained system. Hamiltonian is identically zero when acting on physical states. Therefore time evolution in quantum gravity does not have any obvious meaning. It has to be handled carefully.
Our aim is to understand the quantum theory of gravity which is far more complicated in 3 + 1-D. In consideration of these problems, it is reasonable to look at a simpler theory namely 2 + 1-D gravity and develop our insight based on it. We might hope to generalize that understanding to 3 + 1 case. We could have looked at 1 + 1 case, but in one spatial dimension, the concept of Bekenstein-Hawking entropy of black holes does not make sense. This is because the entropy is proportional to the area or the circumference of the black hole. In 1 + 1-D, the boundary of a black hole consists of only two points on which the quantum entropy law cannot be applied. Keeping these shortcomings in mind, we expect that 2 + 1-D gravity may serve well as our prototype. In 1992, Banados, Teteilboim, and Zanelli (BTZ) showed in their paper [1] that in 2 + 1-D, it is possible to have a vacuum solution with a negative cosmological constant, which has many of the same properties as the more physical 3 + 1-D black holes. Due to these similarities, their solution has become the primary model to understand “quantum gravity”.
Gravity in 2 + 1-D has two convenient properties which makes it even more interesting. They are : 2 Quantization of gravity
• 2 + 1-D gravity is trivial, meaning it has no propagating modes or gravitational waves. This is because the Riemann curvature is fully determined by the Ricci tensor. The number of degrees of freedom in this system is exactly equal to the number of constraints given by Einstein’s equations. This makes 2 + 1-D grav- ity a “topological field theory” which is much easier to quantize than its 3 + 1-D counterpart.
• Unlike Schwarzschild and Kerr solutions, the BTZ spacetime is asymptotically anti- de Sitter which is a maximally symmetric vacuum solution with negative curvature. This manifold is exactly equal to the “group manifold” SL(2,R). This enables us to translate many difficult analytical questions into simple algebraic computations.
Quantum mechanics is believed to be the underlying theory of all physical processes known in nature. Nevertheless, it cannot be used to predict the outcome precisely but gives the probability for a particular process to happen. Quantum mechanics is based on one main principle, namely unitarity. All the information about a particle state is encoded in the wave function describing it until it collapses. Unitarity means that given a wave function at present time, its time evolution in later time as well as in the past time is uniquely determined by evolution operator(unitary). One consequence of this is that in quantum theory, information is never truly lost, nor is it truly copied, at least in principle. But a paradox arose in 1975 after Stephen Hawking [2] and Bekenstein [3–5] made a remarkable connection between thermodynamics, quantum mechanics, and black holes. They have shown within the field theoretic framework that black holes surrounded by quantum fields emit radiation (Hawking radiation) and slowly shrink in size, eventually evaporating completely leaving nothing behind. According to the quantum field theory in curved spacetime, Hawking radiation involves a pure state of two mutually entangled particles where an outgoing particle escapes as Hawking radiation and the infalling one is swallowed by the black hole. Therefore the exterior is entangled with the interior. For an observer with access only to the exterior, the outgoing particle is in a mixed state and since
3 Quantization of gravity
the quantum numbers of the particle inside the black hole can never escape, there will only be an exterior mixed state if the black hole evaporates completely. This mixed state is quantum mechanically described by a density matrix rather than a wave function. In transforming from a pure state to a mixed state one must lose information. For instance, in our example, we took a state described by a set of eigenvalues and coefficients, a large set of numbers, and transformed it into a state described by temperature, which is a single quantitative number. All the other structures of the state were lost in the transformation. In other words, during the process of falling into the black hole, a non- unitary transformation has been performed on the state of the system. As we recall, non- unitary evolution is not allowed to occur naturally in a quantum theory because it fails to preserve probability; that is, after non-unitary evolution, the sum of the probabilities of all possible outcomes of an experiment may be greater or less than 1. This is the famous ‘information paradox’. The fact that the information is lost is reflected in the thermal nature of the emitted radiation. But any thermal system can be assigned to an entropy via the Gibbs law dE = SdT . Since the quantum radiation is thermal, we can calculate the black hole entropy by just calculating the black hole temperature.
While it is true that Gibbs law gives the correct Bekenstein-Hawking entropy from the calculated temperature, no one has been able to explain the entropy directly from quan- tum mechanical or statistical grounds. In fact, it is proven that semi-classical gravity is insufficient to account for this entropy. This is a profound result since the thermodynamic entropy is obtained at the semi-classical level, thus, leaving us with two possible choices :
• Thermodynamic entropy does not always have a statistical mechanical basis.
• Gravity is not a fundamental interaction, but rather a composite effect of some more fundamental underlying theory.
The second option is from the point of view of String theory. It turns out that by using some known results regarding monopoles in certain types of field theories, string theory
4 Quantization of gravity is able to count for states that would contribute to a certain class (unphysical) of black hole of a given mass.
Several theories have been put forward to resolve the information paradox. The most straightforward solution is to assume that the evaporation process leaves behind a rem- nant by some unknown mechanism. But this is difficult to imagine since it would possess large degeneracy while remaining stable. To contain information from an evaporated black hole, the remnant would have to have an infinite number of internal states which does not seem to be a viable option.
In 1993, Leonard Susskind, working with Larus Thorlacius and John Uglum [6], build- ing on ideas of Gerard‘t Hooft and John Preskill, tried to give a revolutionary idea. They emphasized on the non-local behavior that would be needed in order to resolve the information paradox. Their principles of black hole complementarity [7–9] is based on following postulates :
• Unitarity: A distant observer sees a unitary S-matrix which describes the black hole evolution from infalling matter to outgoing Hawking radiation within the standard quantum theory.
• EFT : Outside the stretched horizon, physics can be described by an effective field theory of Einstein gravity and matter.
• No Drama : Equivalence principle holds at the horizon. A freely falling observer experiences nothing out of the ordinary while crossing the horizon.
According to this theory, different observers see the same bit of information at different places. For an external observer, the infinite time dilation at the horizon itself makes it appear as if it takes an infinite amount of time to reach the horizon. However, accord- ing to the infalling observer, nothing special happens at the horizon itself, and both the observer and the information will hit the singularity eventually. Though it sounds like cloning of information, there is only one bit of information in the Hilbert space, which is
5 Quantization of gravity not localized. Both the observers are right in their own point of view but since nothing can escape the black hole, there is no scope for them to compare both their stories simul- taneously. In a way, the notion of locality is lost, but not the quantum theory. Another aspect of this complementarity theory is the existence of the stretched horizon, which is a hot and physical membrane hovering about a Plank length outside the event horizon. According to an external observer, the infalling information stays on the membrane heat- ing it up which then radiates it as Hawking radiation, with the entire evolution being unitary.
However, in 2012, Almheiri, Marolf, Polchinski and Sully (AMPS) [10] showed (under certain assumptions) that the equivalence principle and unitarity contradict each other during evaporation process of the black hole. Considering that the effective field theory is valid outside the stretched horizon, a Bogoliubov transformation is possible from the frame of a distant observer (for whom the quantum field is in a pure state) to that of freely falling observer such that the freely falling observer will see a violation of equivalence principle as he crosses the horizon. AMPS proposed that the most conservative solution to this contradiction is that there would be drama at the horizon of the black hole in the form of a firewall that would destroy the infalling object. It seems surprising because the curvature is relatively small at the event horizon of a sufficiently large black hole where general relativity should hold. This conflict arises between physics at infinity and physics near the horizon, but far from the singularity.
Hawking recently proposed a resolution [11] to this paradox by suggesting that it may incidentally happen that collapse does not always lead to an event horizon. Matter stops collapsing once each shell reaches its apparent horizon due to some quantum mechanical mechanism. In other words, quantum mechanics is believed to play a major role in the collapsing phase in such a way that the collapse does not end up in a black hole. Therefore, no singularity is formed. And in the absence of event horizon, the paradox of information loss becomes irrelevant.
6 Quantization of gravity
With this in mind, we will attempt to quantize gravity with a view to constructing a quantum description of dust collapse model in 2 + 1 dimensions. Then we will consider an example of a static configuration and demonstrate that such conditions are possible under which the physical boundary of the system does not cross the BTZ horizon radius forming no black hole, hence proving Hawking’s proposal.
Like the quantization of other three forces, quantizing gravity naturally directs us to think of the Lagrangian (Hamiltonian) formalism of some kind of field variables. As Witten shows in his paper [12] that 2 + 1-D gravity has a simple gauge theoretic interpretation, it seems promising to explore more into this description. It is very interesting to see that if we take a matrix-valued gauge field which depends on local frame field, then the 3-D action for gravity can be derived from the boundary term of a 4-D action. This is the Chern-Simons action. General relativity is described by this action with an additional constraint that the gauge fields are invertible. The advantage of this approach is that it admits many different classes of solutions that may have metric interpretations. It also comes with solutions where metric of the spacetime cannot be interpreted in a usual way. However, it is interesting to explore this description in its own right. In the next chapter, we will elaborately describe this action principle and present that it can lead to some well-known solutions to Einstein’s equations.
7 Chapter 2
Chern-Simons theory and quantization of BTZ black hole
2.1 Introduction
In this chapter, we will explain the gauge-theoretic approach to 2 + 1-D gravity. 2 + 1-D pure gravity has no local, propagating degrees of freedom. In the absence of matter, gen- eral relativity describes a spacetime of constant curvature. The absence of local dynamics means that the gravity in three dimensions is totally determined by its global effect, hence it is a topological theory. Therefore, we can expect the problems with quantum gravity to be handled easily in the 2 + 1 case. Although this fact does not make anything easier. Nevertheless, it is far from being trivial [13]. The vacuum solutions of pure gravity are multi-conical spacetimes, obtained by identification of points in flat space [14–16]. In the presence of a cosmological constant, the solutions obtained are maximally symmetric, i.e. the Anti-de Sitter(AdS) and de Sitter(dS) spacetimes. The Ba˜nados, Teitelboim, and Zanelli (BTZ) [1] black hole solution is locally AdS but globally it is characterized by conserved charges at the boundary of the AdS spacetime [17]. Due to similarity with the 3 + 1 case, it provides much simpler ground to study quantum effects of gravity.
8 Quantization of gravity
Classical 2 + 1 dimensional gravity and super-gravity can be viewed as Chern-Simons gauge theories of the Poincare, Anti-de Sitter and de Sitter group and their supersym- metric generalizations [12, 18–20]. One can also show that the spinning black hole solution of Banados, Teitelboim, and Zanelli (BTZ) [1] can be recovered by identification of points by a discrete subgroup of SO(2, 2) [21]. We can construct the Chern-Simons action as
1 Z 1 I = γ Aa ∧ dAb + f b Ac ∧ Ad (2.1) C.S. 2 ab 3 cd
a by expanding the gauge super field in the bases of generators Tba. Here fbc are the structure constants of Lie algebra and γab = Tr(TbaTbb) plays the role of the metric of the Lie algebra. For the action to contain the kinetic terms of all gauge fields the metric has to be non- degenerate. By construction, the action is invariant under the gauge transformation
a δgAµ = −DµΛ, where Λ = Λ Ta and Dµ = ∂µ + [Aµ, ]. If the action (2.1) describes the correct gauge theory of (super)gravity, then gauge transformation is equivalent to diffeomorphism. This is true for small diffeomorphism on-shell. In general Chern-Simons action admits many solutions that have no metric interpretation. To have a metric interpretation it is necessary to impose the additional constraint that the gauge field is invertible. With this constraint, solutions to Einstein’s theory in 2+1 dimensions emerge from the Chern-Simons action. In my work, we will limit ourselves to those solutions with metric interpretations. In this case the gauge fields are the dreibein and spin connection, and the field strength torsion vanishes from the classical equations of motion and the curvature is constant. Following Kuchaˇr[22], we will describe the canonical structure of Chern-Simons action for metric compatible solutions. There are other approaches that have been pursued in [23–26]. They focus on solving constraints and using them to derive simplified Hamiltonian in space of finite numbers of degrees of freedom. In case of spacetime having symmetry, one can start with reduced action for the system.
In our approach [27], we will simplify the constraints via a series of canonical transforma- tions. The advantage of this approach is that these transformations can be modified and thus be useful in a variety of systems i.e. Einstein-Maxwell system [28], Lovelock gravity
9 Quantization of gravity
[29] and the cases when the matter is included [30, 31]. In section (2.2), we will provide basics of first order formalism. In section (2.3), we will discuss the Chern-Simons action with SO(2, 2) group structure. In section (2.4), we will calculate the equations of motion and showed that static cases correspond to two different solutions. In section (2.5), the fall-off conditions will be described which are important to find the boundary variations from the bulk action. In section (2.6), the canonical structure of the two spacetimes will be considered by embedding the ADM metric into the static solutions. In section (2.7) and (2.8), we will perform the quantization of the rotating BTZ black hole. In the end, we will show that the wave function that describes the system stays the same on each spacelike hypersurface and are fully characterized by two global conserved charges.
2.2 The Connection
Three dimensional gravity is topological by nature which can be seen from Einstein- Hilbert action on 3 − D manifold
Z 1 3 √ IEH = d x −g(R + 2Λ) (2.2) 16πG M
Gibbons, Hawking and York proposed an additional boundary term that depends on the boundary geometry [32, 33]
Z √ 1 2 IGHY = d x −hK (2.3) 8πG M where h is the determinant of the induced metric hµν defined by
hµν = gµν + nµnν (2.4)
on the boundary, nµ is the normal vector to the hypersurface at the boundary and K is the trace of the extrinsic curvature of the boundary. This term cancels the terms involving
δ(∂σgµν). Therefore, making δgµν = 0 is sufficient to make the action stationary. We can 10 Quantization of gravity map the diffeomorphism as a gauge transformation that leaves the total action invariant if we consider Einstein’s gravity as a gauge theory. This is called first order formalism
a where vierbeins e µ are the fundamental field variables. This is referred as frame field and a all the frame fields at a particular point makes the frame bundle. e µ can be thought of as a transformation matrix between the tangent space and coordinate frame at a particular point in spacetime. If we take the basis vectors of local tangent space to be orthonormal with respect to the Minkowski metric, then they satisfy the following relations
µν a b ab g e µe ν = η a b ηabe µe ν = gµν (2.5)
Here the Greek indices are spacetime indices and the Latin indices are vierbein indices. Given a vector field whose tangent space components are V a, the covariant derivative can
a be defined using the spin connection ω µ
a a a b DµV = ∂µV + ωµbV (2.6)
If V µ are the components of the same vector in the coordinate basis, then the covariant derivative of the vector is defined by
ν ν ν λ DµV = ∂µV + ΓλµV (2.7)
The above two equations are compatible if and only if we demand that the net parallel transport of vierbeins gives a vanishing covariant derivative
a a ρ a abc Dµe ν = ∂µe ν − Γµνe ρ + ωµbeνc = 0 (2.8)
ρ ρ a ρ a b From the last equation (2.8) it follows that Γµν = e a∂µeν +e aωµbeν . The torsion tensor ρ ρ is given by Tµν = 2Γ[µν] and can thus be written as
a a a b Tµν = 2(∂[µeν] + ω[µb eν]) (2.9)
11 Quantization of gravity
. If the connection is torsion free, then the above equation can be written in terms of exterior calculus notation as
a a a b Dωe = de + ω b ∧ e (2.10)
a a where e and ω b are the frame and spin connection one-form. This is Cartan’s first structure equation. In that case, the curvature tensor can be defined using usual expression
a a b for gauge field strength as [Dµ,Dν]V = RµνbV . Using the above definition of covariant derivative Riemann tensor takes the form
a µ ν a a c Rµνbdx ∧ dx = dω b + ω c ∧ ω b (2.11)
which is analogous to familiar gauge theory field strength F = dA + A ∧ A. To construct the Chern-Simons action in 3-D we will take a four dimensional manifold with the Chern- Pontryagin action of the form
Z Z 1 4 µνλρ a b IP = Tr.(F ∧ F ) = d x F µνF λργab (2.12) 4 M4
We will assume that γab is non-degenerate to make sure that the action contains kinetic terms of all gauge fields. The integrand can be written in the form
1 1 γ dAa + f a AcAd dAb + f b AeAf ab 2 cd 2 ef 1 ⇒ γ dAadAb + f a AeAf dAb (2.13) ab 2 ef
a where we have used the Jacobi identity of structure constants fbc to eliminate the last term containing four A’s. The Lagrangian now can be readily written as total divergence of a boundary term which is the Chern-Simons action
Z 1 I = dL ,L = γ AadAb + f a AcAeAb (2.14) P CS CS ab 3 ce
12 Quantization of gravity
In 2 + 1-D these fields can be written as one-forms
1 ea = ea dxµ, ωa = abcω dxµ (2.15) µ 2 µbc where abc is three dimensional Levi-Civita tensor and invariant under SO(3, 1). Einstein Hilbert action in 3 dimensions with cosmological constant can be written as
1 Z 1 Λ I = ea ∧ (dω + ωbωc) + ea ∧ eb ∧ ec (2.16) 8πG a 2 abc 6 abc
a Varying this action with respect to ωa and e will yield the equation of motion
b c Ta = dea + abcω ∧ e = 0 1 Λ R = dω + ωb ∧ ωc = − eb ∧ ec (2.17) a a 2 abc 2 abc
a respectively. The triads e µ are invertible to have any meaningful solution in terms of General relativity.The second equation is the Einstein’s equation in first order formalism. It can be shown that the action is invariant under two sets of gauge symmetries
• Local Lorentz Transformation(LLT ): given by
a abc δle = ebτc a a abc δlω = dτ + ωbτc (2.18)
• Local Translations (LT ):
a a abc δte = dρ + ωbρc a abc δtω = −Λ ebρc (2.19)
Here τ a and ρa are local functions and its components are equal to the number of Lorentz transformations and translations respectively.
13 Quantization of gravity
Gravity in 2 + 1-D behaves like a gauge theory in many ways because of the action (2.16) which is of the first order form. Starting from the action (2.1) Euler-Lagrange equation gives
F [A] = dA + A ∧ A = 0 (2.20)
Therefore A is a flat connection and its field strength vanishes.
In the following section, we show that under certain group structure Chern-Simons action leads to well-known solutions to Einstein equations.
2.3 Chern-Simons gravity
Vacuum three dimensional gravity can be treated as a gauge theory if we consider follow- ing three groups
• ISO(2, 1) for pure gravity(Λ = 0)
• SO(2, 2) for anti-de Sitter space(Λ > 0)
• SO(3, 1) for de Sitter space(Λ < 0) on a Chern-Simons action of the form
1 Z 2 ICS = Tr A ∧ (dA + A ∧ A) 2 M 3 Z 1 µνλ 2 = Tr Aµ ∂νAλ + AνAλ 2 M 3 Z 1 a b 1 b c d = γab A ∧ (dA + f cdA ∧ A ) (2.21) 2 M 3
µνλ These A’s are the gauge connections (one-forms), is completely antisymmetric tensor,γab
is the metric of the Lie algebra i.e., γab = Tr(TbaTbb) with Tba’s are the generators of the 14 Quantization of gravity
a Lie algebra and f bc are the structure constants of the group. For ISO(2, 1) group Aµ’s can be expanded in the basis of generators of the corresponding group as
a a Aµ = e µPca + ω µJba (2.22)
a a e µ and ω µ are the dreibeins and the spin connections respectively.Pca and Jba are the translation and Lorentz generators of ISO(2, 1) group satisfying the following commuta- tion rule
h i h i h i c c Pca, Pbb = 0, Pca, Jbb = abcPc, Jba, Jbb = abcJb (2.23)
It is worth mentioning that the above action is invariant under infinitesimal gauge trans-
a formation δAµ = −Dµα with α = α Tba, if we define the covariant derivative Dµ =
∂µ + [Aµ, ]. The quadratic invariant form (Casimirs), namely P.b Jb+ J.b Pb determines the
γab as
Tr(PcaJbb) = Tr(JbaPbb) = δab (2.24)
Substituting all these in (2.21) we can write the action in the form
Z 1 3 µνλ a b b c d a b ICS = δab d x e µ ∂νω λ + cdω νω λ + ω µ∂νe λ (2.25) 2 M
Chern-Simons action can also be seen as the surface term of the four dimensional gauge theory action of the form R Tr(F ∧ F ) as
Z Z µνρσ 2 Tr(F ∧ F ) = (Bulk term) + ∂σ Tr Aµ∂νAρ + AµAνAρ (2.26) S ∂S 3 where Fµν = [Dµ,Dν] = ∂µAν − ∂νAµ + [Aµ,Aν] is the non-abelian field strength tensor. We can generalize it by including Cosmological constant (Λ < or > 0). Spacetime is not flat anymore but locally homogeneous with a constant curvature determined by Λ. It turns out that for Λ > 0, the spacetime is going to be what we call anti-de Sitter spacetime. It is a vacuum solution of Einstein’s equations. This spacetime has a constant
15 Quantization of gravity
negative curvature proportional to −Λ. Anti-de Sitter spacetime has its own symmetry SO(2, 2). It is easy to see that gauge group for this spacetime will be SO(2, 2). On the other hand, for Λ < 0 the spacetime will have a constant positive curvature and it is de Sitter spacetime. It has a SO(3, 1) symmetry.
Moving on to anti-de Sitter spacetime, inclusion of Cosmological constant Λ corresponds to changing the commutation relations as
h i h i h i c c c Pca, Pbb = ΛabcJb , Pca, Jbb = abcPc, Jba, Jbb = abcJb (2.27)
Starting from (2.21) with this commutation relations we will have now one extra term in the action
Z 1 3 µνλ a b b c d Λ c d ICS = δab d x e µ ∂νω λ + cd ω νω λ + e νe λ 2 M 3 a b +ω µ∂νe λ (2.28)
For further calculation it is convenient to separate the time components of the action
Z 1 3 ij a b b c d c d ICS = δab d x e t 2∂iω j + cd ω iω j + Λe ie j 2 M a b b c d a b a b +ω t 2∂ie j + 2 cde iω j − e i∂tω j − ω i∂te j (2.29)
We can immediately see that dreibeins and spin connections are canonically conjugate
a i ij b to each other. In other words, the momentum conjugate to e i is Πa = δab ω j. In a a this first order form, if we treat {e i, ω i} on equal footing, then canonical momenta do not depend on time derivatives of field variables. This tells us that it is a second class primary constraint. This theory does not have any degrees of freedom since it has twelve
a second class and six first class constraints. We choose e i to be the configuration space variables, the Hamiltonian density will take the form
a b a b H = −δab e tF [ω] + ω tF [e] (2.30)
16 Quantization of gravity
where
dm c i Fa[e] ≡ acdδ e iΠm ≈ 0
1 F [ω] ≡ ∂ Π i + δcmδdnΠ kΠ l + Λijec ed ≈ 0 (2.31) a i a 2 acd kl m n i j
These are the six constraints of the theory. First three constraints represent vanishing of torsion tensor and last three enforce constancy of the spacetime curvature. In other words, Einstein’s gravity is torsion free gravity where the affine connections are symmetric in their lower two indices.
To retrieve the canonical structure of the spacetime, we need to write down the constraints in terms of the metric functions which we can do by considering the standard ADM metric
2 ds2 = N dt2 − A2(dr + N rdt)2 − B2(dφ + N φdt)2 − C2(dr + N rdt)(dφ + N φdt) (2.32)
We are interested in axisymmetric solutions. This ADM form implies foliation of three dimensional spacetime labeled by time parameter t such that each leaf (spatial hypersur- face) will have the following axisymmetric metric in circular coordinates (r, φ)
ds2 = A2(r)dr2 + B2(r)dφ2 + C2(r)drdφ (2.33)
Now (2.32) can be written as
Q 2 ds2 = N 2dt2 − L2(dr + N rdt)2 − R2 dφ + N φdt + dr (2.34) R
with the following identification
r c4 C2 L = A2 − ,R = BQ = 4B2 2B
C2 N = NN r = N r N φ = N φ + N r (2.35) 2B2
17 Quantization of gravity
Now we can apply the ADM formalism on this metric. We can define this particular kinds of combination of functions A, B, C as our canonical variable L, R, Q and N,N r,N φ are lapse and shift functions respectively which are linear combinations of old lapse and shift functions. We will see that these lapse and shift are nothing but Lagrange multipliers. We can find a dreibein which leads to the metric given in (2.34). It is given in lower triangular matrix form
N 0 0 a r e µ = N LL 0 (2.36) N φRQR
In terms of these variables the six constraints (2.31) will be
r φ r F0[e] ≡ LΠ2 − RΠ1 + QΠ1 ≈ 0 φ r F1[e] ≡ RΠ0 + QΠ0 ≈ 0 r 0 F2[e] ≡ LΠ0 + R ≈ 0 r r φ r φ F0[ω] ≡ ∂rΠ0 + Π1 Π2 − Π2 Π1 + ΛLR ≈ 0 r r φ r φ F1[ω] ≡ ∂rΠ1 + Π0 Π2 − Π2 Π0 ≈ 0 r r φ r φ F2[ω] ≡ ∂Π2 − Π0 Π1 + Π1 Π0 ≈ 0 (2.37)
r 0 φ 0 From the third and the second constraints we have Π0 = −R /L and Π0 = QR /LR. φ Substituting, these values and Π1 from the first equation, into the last three constraints will yield
r φ r φ 0 0 F0[ω] ≡ Π1 Π2 − Π2 Π1 + ΛLR − (R /L) ≈ 0 r 0 φ 0 r F1[ω] ≡ ∂rΠ1 − (R /L)Π2 − (QR /LR)Π2 ≈ 0 r 0 r F2[ω] ≡ ∂rΠ2 + (R /R)Π2 (2.38)
r r φ Defining PL = Π1 ,PQ = Π2 and PR = Π2 , we have the simplified Hamiltonian as
g r φ H = −NH − N Hr − N Hφ (2.39)
18 Quantization of gravity
g with the Hamiltonian constraint H and momentum constraints Hr, Hφ as
L Q R0 0 Hg = P P + ΛLR − P 2 + P P − ≈ 0 L R R Q R Q L L QR0 H = LP 0 − R0P − P ≈ 0 r L R R Q 0 Hφ = (RPQ) ≈ 0 (2.40)
The last constraint tells us that RPQ = α(t). Using the third constraint second constraint can be written as
0 0 0 LPL − R PR + QPQ ≈ 0 (2.41)
Therefore, we construct a six dimensional phase space spanned by L, R, Q and their conjugate momenta. These three constraints will dictate the dynamics of a point in this phase space, since the Hamiltonian is itself combination of these three constraints. As an example, axisymmetric solutions are obtained by taking Q = 0 and circularly symmetric solutions by Q = N φ = 0.
2.4 Equations of motion
The equations of motion are given by Poisson brackets with the Hamiltonian H. These are
˙ r 0 R = −NPL + N R NQ L˙ = −NP − P + (N rL)0 R R Q 0 2L Q QR 0 Q˙ = N P − P + N r + N φ R R Q R L R 0 0 2 N R NP 0 P˙ = NΛR − − Q + N φ R L R2 R 00 0 0 2 N N L NLP Q 0 P˙ = NΛL − + + Q − P P + (N rP )0 − N φ P R L L2 R2 R2 Q L R Q N R0 P˙ = P P + N rP 0 = − P (2.42) Q R Q L Q R Q 19 Quantization of gravity
where we have used the last constraint to get the result in the last equation of motion. This tells us that α(t) must be constant. So far we have made no additional assumptions except isotropy. Therefore, (2.40) and (2.42) must be satisfied for any isotropic classical solutions.
2.4.1 Static case
In this section we will discuss the static case and will work out two known examples starting from equations of motion. In static case, the time derivative of all canonical variables must vanish. Using RPQ = α with the first two constraints equations of motion will be of the form
(N rL)0 αQ P = − R N R2 N rR0 P = L N 0 2αL Q QR 0 N − P + N r + N φ R = 0 R2 R L R N 0R0 α2N NΛR − − + N rP 0 = 0 L2 R3 L 00 0 0 2 N N L α NL αQ α 0 NΛL − + + − + (N rP )0 − N φ = 0 L L2 R4 R3 R R α2L αQ R0 0 P P + ΛLR − + P − = 0 L R R3 R2 L L R0 αQR0 P 0 − P − = 0 (2.43) L L R LR2
We have eight unknown functions but seven equations. The system is over-determined. So we have freedom to choose one of the unknown functions and we have set N r = 0.
2 This substitution gives PL = 0, PR = −αQ/R . We can readily check the last equation is satisfied identically. Now we are left with four remaining equations with five unknowns, namely
2αNL 0 + N φ R = 0 R2 N 0R0 α2N NΛR − − = 0 L2 R3 20 Quantization of gravity
00 0 0 2 N N L α NL αQ α 0 NΛL − + + − − N φ = 0 L L2 R4 R3 R α2L R0 0 ΛLR − − = 0 (2.44) R3 L
We still have freedom to choose another function and we take Q = 0. We can solve the first equation, N φ0 = −2αNL/R3. The remaining equations are
N 0R0 α2N NΛR − − = 0 L2 R3 N 00 N 0L0 3α2NL NΛL − + + = 0 L L2 R4 α2L R0 0 ΛLR − − = 0 (2.45) R3 L
A close inspection reveals that the third equation can be obtained from the other two equations. So, there are actually two independent equations
N 0R0 α2N NΛR − − = 0 L2 R3 α2L R0 0 ΛLR − − = 0 (2.46) R3 L
These are two independent equations with three unknown functions. Therefore, we choose R(r) = r as expected in the static case. At this point we can consider two different cases, i)Λ = 0 and ii)Λ 6= 0.
For Λ = 0 case, the corresponding gauge group SO(2, 2) turns into the Poincar´egroup by a Wigner-Inonu contraction. With R(r) = r the equations in (2.46) will yield the following solutions
1/µ L(r) = p1 + α2/µ2r2 p 2 2 2 N(r) = N+ 1 + α /µ r φ φ 2 N (r) = N+ + N+α/µr (2.47)
21 Quantization of gravity
φ where µ,N+ and N+ are constants of integration. To see what this means, we set N+ = 1 φ and N+ = 0, the corresponding line element is given by
N −2 j 2 ds2 = N 2dt2 − dr2 − r2 dφ − dt (2.48) µ2 µr2
If we compare this with spinning point particle metric, µ can be identified as the mass of the particle and j as the angular momentum.
With Λ 6= 0, solving second equation of (2.46) gives the solution
α2 −1/2 L(r) = Λr2 − M + (2.49) r2
where M is the constant of integration. Substituting this in the first equation of (2.46), we have
α2 1/2 N(r) = N Λr2 − M + (2.50) + r2
φ φ 2 φ together with N = N+ + N+α/r . With N+ = 1 and N+ = 0 we recover the BTZ black hole metric
J 2 ds2 = N 2dt2 − N −2dr2 − r2 dφ − dt (2.51) r2
where we have identified M as the mass of the BTZ black hole and −α as the angular momentum.
22 Quantization of gravity
2.5 Fall-off conditions and boundary action
The total action on the hypersurface can be written as the sum of a bulk and boundary action
Z Z ˙ ˙ ˙ SΣ = dt dr PLL + PRR + PQQ − H + S∂Σ (2.52)
The boundary terms play an important role by canceling unwanted boundary variations in the action. Its values depend on the boundary conditions that are imposed. And it is reasonable to impose the boundary conditions such that the solutions asymptotically approach stationary spacetime as r → ∞. For the maximally extended point particle and the BTZ black hole, the boundary conditions are described below. The basic idea is to consider the asymptotic expansions of the canonical variables integer power of 1/r.
With this spirit, for spinning point particle, we adopt the boundary conditions such that it matches with (2.48) as r → ∞
R −→ r + O∞(r−2) 2 1 j± −2 ∞ −3 L −→ − 3 r + O (r ) µ± 2µ± Q −→ O∞(r−2)
∞ −1 PR −→ PR0 + O (r ) ∞ −1 PL −→ O (r ) −1 ∞ −2 PQ −→ j±r + O (r ) j± −2 ∞ −3 N −→ 1 + 2 r N± + O (r ) 2µ± N r −→ O∞(r−1)
φ φ j± −2 ∞ −3 N −→ N± + r + O (r ) (2.53) µ±
The higher order terms O∞(r−n) represents the asymptotic behavior that fall as r−n. It is implied that some function of t is multiplied with these terms. The ± sign refers to the quantities at right and left infinities respectively. These fall-off conditions are consistent with the constraints and the equations of motion (2.43). We will consider all the terms of 23 Quantization of gravity
the Hamiltonian density H whose variation actually contribute to the boundary terms. As it turns out that the variations with respect to L and PQ will contribute to the boundary terms
Z Z 0 φ φ φ dt NR δ (1/L) − N RδPQ |∂M = − dt[N+δµ+ + N−δµ− − N +δj+ − N −δj−] (2.54)
This unwanted boundary variations must be canceled in the boundary action. Therefore, we take the boundary action to be
Z φ φ S∂Σ = dt[N+µ+ + N−µ− − N +j+ − N −j−] (2.55)
This boundary action affirms the role of µ and j as the mass and angular momentum of the spinning particle.
For the BTZ black hole case, we adopt the following boundary conditions
R −→ r + O∞(r−2) r−1 M r−3 L −→ + ± + O∞(r−4) Λ1/2 2Λ3/2 Q −→ O∞(r−6)
∞ −2 PL −→ O (r ) ∞ −4 PR −→ O (r ) −1 ∞ −2 PQ −→ −J±r + O (r ) M N −→ Λ1/2r − ± r−1 N + O∞(r−2) 2Λ1/2 + N r −→ O∞(r−2)
φ φ −2 ∞ −4 N −→ N ± + J±r + O (r ) (2.56)
By explicitly calculating the variations of the contributing terms, we find that variations
with respect to L and PQ contribute to the variations as before. So the boundary action
24 Quantization of gravity
will be
Z 1 S = − dt (N M + N M ) + N φ J + N φ J (2.57) ∂Σ 2 + + − − + + − −
Though inclusion of these boundary terms are necessary as it would unfreeze the evolution at the two infinities, it would lead to the problem that the lapse and shift functions may also be varied at the boundary which would give µ± = j± = M± = J± = 0. As Kuchaˇr φ proposed, the solution to this problem is to assume N± and N± as a prescribed functions of t. This procedure is called parameterization at infinity.
2.6 Embedding
The canonical structure of a spacetime can be understood with the help of the embedding procedure. We know the canonical structure of the ADM metric. We will imagine the two metrics describing the spinning particle and the BTZ black hole as leaves of a particular foliation of spacetime and we will embed the hypersurfaces of the ADM metric into these metrics.
2.6.1 Spinning particle
The metric in this case can be cast in terms of the Killing time and area radius as
2 j 2 ds2 = F dT 2 − dR2 − R2 dφ − dT (2.58) µ2F µR2
j2 where F = 1 + µ2R2 , µ and j are the mass and angular momentum respectively. With the rescaling of the Killing time as T = T/µ, the above metric can be conveniently rewritten as
2 2 2 2 2 2 j ds = F1dT − dR − R dφ − 2 dT (2.59) F1 R 25 Quantization of gravity
2 j2 where F1 = µ + R2 . The functions T and R are assumed to be functions of the ADM variables t and r. Comparing corresponding terms of this metric with that of the ADM metric in (2.34), we find the following
0 R0T˙ − T R˙ N = L 0 F −1RR˙ 0 − F T˙ T N r = 1 1 L2 jT˙ N φ = − R2 2 −1 02 02 L = F1 R − F1T 0 jT Q = − (2.60) R
Inserting the lapse and shift functions into the first equation of (2.42) we get
0 1 ˙ r 0 F1T 0 LPL PL = (−R + N R ) = − ⇒ T = − (2.61) N L F1 which upon inserting into the expression for L2, gives
j2 R02 F = µ2 + = − P 2 (2.62) 1 R2 L2 L
From the last equation of (2.60) we also find
02 QR R 2 j = 2 − PL (2.63) LPL L
Thus we are able to write the mass and angular momentum in terms of canonical data.
Furthermore, differentiating F1 with respect to r, we find
R02 0 R0 R0 0 F 0 = − P 2 = −2P P 0 + 2 1 L2 L L L L L 2R0 2P 2R0 = − H − L H − P 2 (2.64) L g L r R Q
26 Quantization of gravity
where Hg and Hr are the Hamiltonian and momentum constraints respectively. Therefore,
j2 0 2R0 2P 2P µ2 + − P 2 = − H − L H − Q H ≈ 0 (2.65) R2 Q L g L r R φ is linear combination of constraints. But µ0 and j0 do not require to be zero separately.
2.6.2 BTZ black hole
For the BTZ black hole the metric is expressed as
1 J 2 ds2 = F (R)dT 2 − dR2 − R2 dφ − dT (2.66) F (R) R2 where F (R) = ΛR2 − M + J 2/R2, M and J being the mass and angular momentum respectively. Embedding (2.34) into the BTZ metric will yield the same result as (2.60), just F1 replaced by F . Inserting the lapse and the shift into first equation of (2.42) we obtain T 0 and substituting its value into the expressions for L2 and Q, we have
J 2 R02 F = ΛR2 − M + = − P 2 R2 L2 L 02 QR R 2 J = 2 − PL (2.67) LPL L
Here we again write the mass and angular momentum in terms of canonical data. We also have seen
R02 2 R0 R0 0 (F − ΛR2)0 = − P 2 − ΛR2 = −2P P 0 − 2ΛRR0 + 2 L2 L L L L L 2R0 2P 2R0 = − H − L H − P 2 (2.68) L g L r R Q
Thus we may write
J 2 0 2R0 2P 2P −M + − P 2 = − H − L H − Q H ≈ 0 (2.69) R2 Q L g L r R φ
27 Quantization of gravity
2.7 New canonical variables
The goal for seeking new variables is to cast the constraints in simplified form. From the expressions for µ(M) and j(J), it is clear that the mass and the angular momentum cannot be part of the same canonical chart since their Poisson brackets do not vanish. In the quantum theory, they are not simultaneously observable. We will work with non- zero cosmological constant as the spinning particle is the limit Λ → 0 taken together with M → −µ2. From the expression for F in (2.67) it is straightforward to see that if we define two canonical variables Z and PZ in the following way, they are canonically conjugate to each other i.e. {Z,PZ }P.B. = 1, where
R02 LP Z = − P 2 − ΛR2 − P 2 ,P = − L (2.70) L2 L Q Z 2F
The Poisson brackets of Z and PZ with other canonical variables are given below
2R0 0 {Z,R} = 0 {Z,P } = − − 2ΛR P.B. R P.B. L2 {Z,Q}P.B. = 2PQ {Z,PQ}P.B. = 0 R0P 0 {P ,R} = 0 {P ,P } = − L Z P.B. Z R P.B. F 2L {PZ ,Q}P.B. = 0 {PZ ,PQ}P.B. = 0 (2.71)
From now on we will use Z and PZ as our new canonical variables in stead of L and PL.
Looking at the Poisson brackets, the necessity of new canonical variables P R and Q is obvious. However, simple evaluation reveals that the Poisson brackets of
LP P Q = Q + L Q (2.72) F
with Z,PZ and R is zero and is also conjugate to PQ. The remaining problem is to find
P R and we will use generating functional procedure to achieve this. The canonical trans-
formation from original chart {L, R, Q, PL,PR,PQ} to new chart {Z,R, Q, PZ , P R,PQ}
28 Quantization of gravity
can be found to be generated by
Z 2 0 PQ 0 −1 R G[L, R, PL,PQ] = dr LPL 1 − − R tanh (2.73) F LPL
and from this P R is determined to be
0 0 ΛRLPL (R /LPL) P R = PR − − 0 2 (2.74) F 1 − (R /LPL)
The fall-off conditions for the new variables can be easily determined from fall-off condi- tions of old canonical variables (2.56). Once we have the new canonical variables, we can write the Hamiltonian constraint in terms of new variables
2FP R0 Hg = Z [QP + RP ] − [2P (RP )0 + RZ0] (2.75) RL Q R 2FRL Q Q
and the momentum constraint now will be
Q − 2P P H = Z0P − R0P + QP 0 − Z Q H (2.76) r Z R Q R φ
Using the expression in (2.65), the Hamiltonian constraint can be greatly simplified
0 g F R H = [QPQ + RP R] − 2 Hr (2.77) RPL L PL
Now the full Hamiltonian can be written as sum of constraints
g Hf = RP R + QPQ 0 0 0 Hfr = Z PZ − R P R + QPQ 0 Hfφ = (RPQ) (2.78)
g r φ and the Hamiltonian is H = −NeHf − Ne Hfr − Ne Hfφ where the new multipliers are
NF Ne = RPL
29 Quantization of gravity
0 r r NR Ne = N + 2 L PL 0 φ φ r NR Q Ne = N − N + 2 (2.79) L PL R
We also notice that
0 0 R g Q Z PZ = Hf + Hfr − Hfφ (2.80) R R
So we could just as well consider the constrained system
g Hf = RP R + QPQ 0 HfZ = Z PZ 0 Hfφ = (RPQ) (2.81)
In the next section we will absorb the boundary action into the bulk action. This enables us to write the constrained system in more simplified form.
2.8 The boundary action
In terms of new variables the action takes the form
Z h i ˙ ˙ ˙ g r φ S = dtdr PZ Z + P RR + PQQ − NeHf − Ne Hfr − Ne Hfφ + S∂Σ (2.82) where the boundary action is given by (2.57). If the lapse and shift functions at the boundary are allowed to be varied freely, it would imply that the mass and the angular momentum both vanish at infinity. To avoid this problem and allow for non-vanishing mass and angular momentum, those functions must be treated as prescribed functions of the ADM time t i.e., lapse and shift must have fixed ends. To determine what these functions are, we compare the asymptotic ADM metric in (2.34) at fixed r
2 2 2 φ2 2 2 φ 2 2 ds = (N± − R N± )dt − 2R N±dtdφ − R dφ (2.83) 30 Quantization of gravity
with the asymptotic metric in the co-moving frame
2 2 2 2 2 ds = t± + 2Ω±R dtdφ − R dφ (2.84)
where t and Ω denote proper time and angular velocity. To match, we must take
q 2 2 ˙ def N± = ± 1 + r Ω±t± = ±τ˙± φ ˙ def N± = ∓Ω±t± = ±ω˙ ± (2.85)
where t± and ω± represent the proper time and the angular velocity as measured along constant r worldlines at the infinities. The surface action now looks
Z 1 S = − dt (M τ˙ − M τ˙ ) + J ω˙ − J ω˙ (2.86) ∂Σ 2 + + − − + + − −
We will consider the Liouville form
Z ∞ 1 Θ1 := drPZ δZ − (M+δτ+ − M−δτ−) (2.87) ∞ 2
According to the fall-off conditions, limx→±∞ Z(r) = −M±. Therefore there exists a density function Γ(r) such that
Z r 0 0 0 Z(r) = −M− − dr Γ(r ),Z (r) = −Γ(r) (2.88) −∞
We rewrite the Liouville form
Z ∞ Z r 0 0 1 Θ1 := drPZ (r) −δM − dr δΓ(r ) − δ(τ+M+ − τ−M−) −∞ −∞ 2 1 1 + τ δM − τ δM 2 + + 2 − − Z ∞ Z ∞ Z r 1 0 0 = δM− − τ− − drPZ (r) − drPZ (r) dr δΓ(r ) 2 −∞ −∞ −∞ Z ∞ 1 0 0 1 + τ+ δM− + dr δΓ(r ) − δ(τ+M+ − τ−M−) 2 −∞ 2 Z ∞ Z ∞ Z r 1 0 0 = δM− (τ+ − τ−) − drPZ (r) − drPZ (r) dr δΓ(r ) 2 −∞ −∞ −∞ 31 Quantization of gravity
Z ∞ 1 0 0 1 + τ+ dr δΓ(r ) − δ(τ+M+ − τ−M−) (2.89) 2 −∞ 2
From the result above we are able to identify the conjugate variables as
1 Z ∞ m = M−, pm = (τ+ − τ−) − drPZ (r) (2.90) 2 −∞
In terms of these new variables the Liouville form is
Z ∞ Z r 1 0 0 Θ1 := pmδm + dr τ+δΓ(r) − PZ (r) dr δΓ(r ) (2.91) −∞ 2 −∞
Using the identity
Z ∞ Z r Z ∞ Z r 0 0 0 0 drPZ (r) dr δΓ(r ) = − drδΓ(r) dr PZ (r ) (2.92) −∞ −∞ −∞ ∞
we find the Liouville form
Z ∞ Z r 1 0 0 Θ1 := pmδm + dr τ+ + dr PZ (r ) δΓ(r) (2.93) −∞ 2 ∞
This form now enables us to identify two other new canonical variables
Z r 0 1 0 0 Γ(r) = −Z (r),PΓ(r) = τ+ + dr PZ (r ) (2.94) 2 ∞
0 0 We notice that PΓ = PZ = T /2. Therefore, the Killing time is identified with the
momentum PΓ up to a constant. We can choose the constant in such a way that T
matches with τ+ at infinity. Then we can write
Z r 0 0 T = 2PΓ = τ+ + 2 dr PZ (r ) (2.95) ∞
We can check that the momentum conjugate to the Killing time is PT = −Γ/2.
32 Quantization of gravity
Now we will look at the other Liouville form
Z ∞ Θ2 := dr P RδR + PQδQ − (J+δω+ − J−δω−) (2.96) −∞
−1 Again under the fall-off conditions, limx→±∞ PQ = −J±r . With the same logic, there exists a density function Σ(r) such that
Z r 0 0 R(r)PQ(r) = −J− + dr Σ(r ) (2.97) −∞
But according to the third constraint RPQ = constant. This means Σ(r) has to vanish.
We conclude, then, J+ = J− = J and
Z ∞ J Θ2 := dr P RδR − δQ − Jδ (ω+ − ω−) −∞ R Z ∞ JQ Z ∞ Q = dr P R + 2 δR − Jδ (ω+ − ω−) + dr −∞ R −∞ R Z ∞ JQ Z ∞ Q = dr P R + 2 δR + (ω+ − ω−) + dr δJ (2.98) −∞ R −∞ R
Thus this Liouville form tells us that we can choose
Z ∞ Q pJ = (ω+ − ω−) + dr (2.99) −∞ R as the momentum conjugate to J and on the other hand,
JQ P = P + (2.100) R R R2 conjugate to R. Now the new constraints will look like
HR = RPR 0 HT = T PT (2.101)
33 Quantization of gravity
and may be adjoined to the canonical action by means of new Lagrange multipliers. In reduced form the action will be of the form
Z ∞ h i ˙ ˙ ˙ T R S = pmm˙ + pJ J + dr PT T + PRR − (N PT + N PR) (2.102) −∞
The configuration space of vacuum 2 + 1 − D gravity is spanned by coordinates T , R and two degrees of freedom,m and J. Now we are in a position to quantize the system. The momenta are raised to operator status and constraints act as an operator on a state
wave functional Ψ(m, J, t; T,R). The two constraints PT = 0 and PR = 0 tell us that the wave functionals are independent of T and R. Therefore the spacetime is described by the wave functional Ψ(m, J, t). We can even further say that the wave functional is time independent. Since the Hamiltonian is a combination of constraints , Hamiltonian vanishes,
iΨ(˙ m, J, t) = 0 ⇒ Ψ(m, J, t) = Ψ(m, J) (2.103)
The wave functional is characterized by two global parameters m and J. The time independence arises from the fact that the spacetime is stationary and it has time-like killing vector field. Being time independent, the wave functional once prepared stays the same on all space-like hypersurfaces which is expected from the BTZ spacetime.
2.9 Summary
In this chapter, we have used the canonical reduction approach developed by Kuchaˇrto construct the canonical description of 2 + 1-D case of the BTZ black hole. The advantage of this approach is that it can be naturally extended to describe the quantum effects of dynamical canonical collapse. Here, we have presented that starting from the Chern- Simons action, some well-known solutions can be derived if we associate certain Lie group structure with the corresponding gauge field. To achieve this goal, we derived the
34 Quantization of gravity
Hamiltonian of the system which turns out to be identically zero since it is derived from a generally covariant system. The whole phase space is six-dimensional and therefore the constraints depend on six phase variables. Once we have the Hamiltonian, we readily found the equations of motion for the phase variables. The static case is given by the fact that the time derivative of the variables vanishes. Residual diffeomorphism constraints allow us to choose one of the variables to remove the under-determinacy of the solutions. These choices lead to two solutions, namely i) the spinning point particle (Λ = 0), ii) the BTZ solution (Λ 6= 0). To understand the underlying canonical structure we embed general axisymmetric metric in the ADM form into one of the solutions. Fall-off conditions are important since the metric has to approach the metric given by (2.48) and (2.51). With these fall-off conditions, it turns out that variations of some terms in the action do not die off at the boundary. To cancel these unwanted variations the boundary action plays a key role. To probe the underlying canonical structure, we embedded the two solutions into the ADM metric whose canonical structure is already known to us. This embedding procedure enables us to write the global parameters m and J in terms of canonical data. Then we sought for new canonical variables that will reduce the constraints into a simpler form. This has been achieved by a series of canonical transformations. The important point is that the information about the global parameters is preserved throughout the calculation. In terms of these new variables, the configuration space turns out to be two dimensional with two degrees of freedom. The constraints removed the dependence of phase space variables and the fact that the Hamiltonian is identically zero makes the wave functional time independent. Therefore the spacetime will be described by a wave functional with only two degrees of freedom, m, and J.
35 Chapter 3
Classical dust collapse model
The canonical formalism mentioned in the previous chapter becomes much simpler for non-rotating black holes and maybe generalized to describe the graviational collapse of matter. This allows us to consider the non-rotating gravitational collapse in 2 + 1-D and determine the collapse wave functional for the system. The earliest study of three- dimensional gravitational collapse without cosmological constant was carried out in [34]. In the case of circularly symmetric and homogeneous dust collapse, it has been shown that the collapse to a black hole sensitively depends on initial data. In the absence of the cosmological constant, the collapse may or may not occur depending on the initial velocity of the collapsing shells. If it collapses then it forms a naked, conical point source singularity [35]. Whereas in AdS spacetime the end state of collapse leads to BTZ black hole provided that the initial density is sufficiently large. If not, then the end state is again naked conical singularity. These results provide the ground for carrying out nu- merical studies of critical phenomena associated with the collapse [36]. It has been also confirmed for inhomogeneous dust collapse in [37, 38]. In [39–42] quantization procedure of dust collapse have been carried out that leads to interesting results. In this chapter, we are going to describe the quantum gravitational collapse in 2 + 1-D. Rotating collapse has been already described classically in [43].We will employ canonical quantization on Lemaˆıtre-Tolman-Bondi (LTB) family of solutions [37, 44–46] that describes the collapse
36 Quantization of gravity or expansion of spherically symmetric inhomogeneous mass distribution. Quantizing the LTB metric will enable us to understand the dynamics of the collapsing system under the influence of the quantum effects which will be described by functional solutions of the Wheeler-DeWitt equation(WDW) equation [47, 48]. We will see that the equation admits two sets of solutions. In the first solution, the collapse will form the apparent horizon and matter on both sides will coalesce at the horizon. The exterior infalling waves are collaps- ing dust shells which are accompanied by interior outgoing Unruh radiation. The second solution represents the case when matter moves away from the apparent horizon. Having found these solutions, now we may attempt to realize Stephen Hawking’s proposal in a quantized model of dust collapse in 2 + 1-D with negative cosmological constant. Hawk- ing’s proposal is in agreement with the first solution only. However, the superposition of these two solutions leads to the picture that both the exterior infalling and interior outgoing wave modes are transmitted through the apparent horizon. The outgoing modes are suppressed by the Boltzmann factor. This corresponds to the traditional picture of black hole evaporation.
For the moment if we take the first solution where continued collapse does not occur, we can expect the collapse process to end up in a spherically symmetric quasi-stable, static configuration of finite size. We will argue that the interior outgoing Unruh radiation dur- ing collapse will generate enough conditions to sustain such a quasi-stable configuration. The process of outgoing Unruh radiation will lead to a negative mass singularity which will weaken gravity slowly causing an expansion eventually at a later time[49].
We will consider a dust ball collapsing in 2 + 1-D spacetime with negative cosmological constant. In general, the collapsing solution is described by LTB family of solutions. In a comoving and synchronous coordinates (τ, ρ, φ), the metric takes the following form
Re2 ds2 = dτ 2 − dρ2 − R2dφ2 (3.1) 2(E − F )
37 Quantization of gravity
The physical radius R(τ, ρ)) and energy density (τ, ρ) obey the following
(R∗)2 = 2E − ΛR2 Fe 2πG(τ, ρ) = (3.2) RRe
∗ where Λ is the cosmological constant, G is Newton’s constant in 2 + 1-D, and e are derivatives with respect to comoving time (τ) and shell label (ρ) respectively. We note that without loss of generality we can rescale the shell labels in such a way that at some initial time the physical radius coincides with the shell label, i.e. R(0, ρ) = ρ. Exploiting this freedom gives an idea about the quantities E and F which may be given as
Z ρ F (ρ) = 2πG ρ0(0, ρ0)dρ0 0 2 2 E(ρ) = [∂τ R(0, ρ)] + Λρ (3.3)
The quantity 2F (ρ) represents total gravitational mass content inside the shell labeled by ρ. E(ρ)/2 is the total energy per unit mass in the shell labeled by ρ. They are called mass function and energy function. From the definition of F (ρ), we would expect that it is a positive and monotonically increasing function of ρ. We will exculde the possibility of shell crossing singularity by assuming Re > 0. In the case, when E is constant, the solution to (3.1) for collapsing dust is given by
r r ! 2E √ Λρ2 R(τ, ρ) = sin − Λτ − sin−1 (3.4) Λ 2E
From this solution, it is evident that shells are inevitably doomed to hit the singularity that is, the phyical radius will shrink to zero at finite proper time
r 2 1 −1 Λρ τ0(ρ) = √ sin (3.5) Λ 2E
Different shells will collapse to the singularity at different proper times which is indicated by the dependence of ρ. It can be shown that the apparent horizon is defined by ΛR2 −
38 Quantization of gravity
2F = 0 where the null divergence vanishes. A shell labeled by ρ becomes trapped once it crosses this apparent horizon i.e. R < p2F/Λ. Only those shells that satisfy F > 0 will be trapped at proper time
r 2 r ! 1 −1 Λρ −1 F τah(ρ) = √ sin − sin (3.6) Λ 2E E
As τ0(ρ) > τah(ρ), collapsing shells will be trapped first before they hit the singularity. In other words, singularity formation is not a necessary condition for the formation of trapped surface.
3.1 Collapse wave functionals
To describe the collapse wave functionals we need to consider the canonical structure of the spacetime which is well described by the Arnowitt-Deser-Misner(ADM) metric. To that goal, we embed cicularly symmetric ADM metric
ds2 = N 2dt2 − L2(dr + N rdt)2 − R2dφ2 (3.7) into LTB family of solution given by (3.1). N and N r are lapse and shift functions. The total action can be written in terms of the metric functions as
Z h i ˙ ˙ g r SEH = dtdr PLL + PRR − NH − N Hr + S∂Σ (3.8)
where the angular dependence has been integrated out, PL and PR are conjugate momenta of L(t, r) and R(t, r) respectively given by (G = 1)
1 h i P = − R˙ − N rR0 L N 1 h i P = − L˙ − (N rL)0 (3.9) R N
39 Quantization of gravity
S∂Σ is the surface term of the action. Since the lapse and shift appear as Lagrange multipliers in the action, the corresponding Hamiltonian and momentum constraints are given by
R0 0 Hg = −P P − ΛLR + ≈ 0 L R L 0 0 Hr = R PR − LPL ≈ 0 (3.10)
The matter content will also contribute to the collpase. Following the paper [50, 51],
k the dust is described by eight spacetime scalars (, τ, Z ,Wk) with decomposition of four velocity in the cobasis as
k Uµ = −τ,µ + WkZ ,µ (3.11) where is the dust proper energy density, τ is the proper time along particle flow lines, Zk are the comoving coordinates of the dust and W k ae the spatial components of four velocity in the dust frame. The dust action is given by
Z 3 p α SD = d x |g| (U Uα + 1) (3.12)
In the case when the particles do not move spatially in the dust frame i.e. W k = 0 the dust action can be simplified in the form
Z d r d SD = dtdr Pτ τ˙ − NH − N Hr (3.13) where
LR P = − (τ ˙ − N rτ 0) (3.14) τ N and the constraints look
r τ 02 Hd = P 1 + τ L2
40 Quantization of gravity
d 0 Hr = τ Pτ (3.15)
To derive the constraints we have used the fact that is a Lagrange multiplier since the action does not involve any time derivatives of dust proper energy density . The total Hamiltonian and momentum constraints are sum of these two constraints. It is obvious from the action that the Hamiltonian is identically zero. The equations of motion of all variables can readily be found out using Poisson brackets.
Embedding of the ADM metric (3.7) in the metric given by (3.1) is useful in order to find a suitable canonical variable. This procedure enables us to write the mass function F in terms of the canonical variables as
1 R02 1 F = P 2 − + ΛR2 = ΛR2 − F (3.16) 2 L L2 2
It has already been shown in [22, 52] that the mass function can be elevated to a canonical variable by making proper canonical transformation. If we take momentum conjugate to
the mass function as PF = LPL/F, by directly taking Poisson bracket with F one can show that they satisfy the commutation relation
{F,PF }P.B. = 1 (3.17)
It is to be noted that F = 0 defines the apparent horizon and plays an important role
in the collapse. F does not commute with PR. Therefore they do not form a canonical
chart. We need to make a canonical transformation from {R, L, PR,PL} to new canoni-
cal variables {R,F, PR,PF }. This transformation can be shown to be generated by the generating function G[R, L, PL]
Z 0 −1 0 G[R, L, PL] = dr LPL − R tanh (LPL/R ) (3.18)
41 Quantization of gravity
And PR is given by
0 0 ΛRLPL (LPL/R ) PR = PR + − 0 (3.19) F 1 − (LPL/R )
In terms of these new variables, the constraints can be simplified to the form
1 Hg = − FP P + F −1R0F 0 L F R 0 0 Hr = R PR + F PF (3.20)
The fall-off conditions of the variables are important as they are needed to asymptotically match the flat metric at infinity. The fall-off conditions are given in [40]. These fall-off conditions lead to
Z Z S∂Σ = − dtM+(t)τ˙+ + dtM0(t)τ˙0 (3.21)
This is called parameterization at infinity. This removes the necessity of fixing the lapse function at the boundary. We are interested in finding new variables that absorb the boundary terms of the action. To this end, we will aim to cast the homogeneous part of the action into Liouville form. This is done by introducing a a density function Γ(r) as
Z r 0 0 0 F (r) = M0 + dr Γ(r ), Γ(r) = F (r) (3.22) 0
Consider the Liouville form
Z ∞ Θ = PF δF + τ+δM+ − τ0δM0 (3.23) 0
Using the above definition of F we can write