Emergent Spacetime

Udit Gupta

May 14, 2018

Abstract
There have been recent theoretical hints that spacetime should not be thought of as
a fundamental concept but rather as an emergent property of an underlying microscopic
theory. In this paper we give an overview of proposed microscopic models that suggest
that of spacetime should emerge as an effective description.

Contents
1 Introduction 2
1.1 Emergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Emergent Spacetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Quantum Gravity 3
2.1 Hints from String Theory: Ambiguity of Spacetime . . . . . . . . . . . . . . 4
2.2 Gravitons and Weinberg-Witten theorem . . . . . . . . . . . . . . . . . . . . 5
2.3 Gravity from Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 Emergence of Gravity in AdS/CFT 7

4 Comparison with Experiment 9

5 Conclusion 9

1
1 Introduction
1.1 Emergence
The reductionist philosophy seeks to find the ”true” laws of nature by breaking down its
constituents into its most basic forms. However, this is not the complete story because even
if one knows the fundamental description of nature the challenge still remains to construct
exact descriptions of large systems using these fundamental laws [1].

One can attempt to understanding fish behavior first by dissection to understand the
individual fish parts. The model that explains how the fish moves would be a complicated
mechanical mess of fins and muscles. Even after this respectable effort the model wouldn’t
adequately describe all phases of fish behavior. For example, if one wanted to explain the
schooling of fish, this model would be incomprehensibly complicated to use. Instead one
should view the school as a new being because on average it behaves differently than any
individual fish. This fish phase then has to be described by an emergent theory that is not
obviously determined in terms to the microscopic theory of fish dynamics. This is the beau-
tiful principle of emergence.

It happens in physics as well. For example, one can attempt to solve the many-body
problem of 1023 interacting electrons in a metal or one can understand that electrons are
fermions which leads to the idea of a Fermi surface and finally Fermi liquid theory. When
viewing a metal in this way, one easily finds other microscopic systems (not electrons in a
metal) whose emergent description is the same! This perspective is completely opposite that
of the reductionist approach where to understand a metal we would have to construct each
electron as an excitation in a quantum field. This would be an incomprehensibly intractable
problem both analytically and qualitatively. One understands a metal to have some effective
phenomena such as conduction with a finite resistivity which would not be evident by con-
structing the theory of a metal in terms of QED. Rather, the emergent property of a Fermi
liquid gives a much clearer picture of a metal while only having to know a few experimentally
measured parameters rather the individual behavior of all electrons and atoms in the lattice.
The idea of emergence in physics is pervasive especially in condensed matter and cold atom
physics. For example, superconductivity and superfluidity are both described by a conden-
sate, which is an emergent state of matter with finite fraction of particles in the ground state
and long range order.

1.2 Emergent Spacetime

There is an idea amongst theoretical physicists that spacetime is not fundamental but itself
an effective description. More importantly, they believe that the notion of particles and fields
living in a spacetime is an emergent property from the dynamics of some other underlying
microscopic theory. Most believe that these degrees of freedom of the underlying theory must
be also be quantum mechanical in nature. We know that a fundamental feature of quantum
mechanics is the notion of entanglement between systems. In the picture of emergent space-
time, a coarse-grained view of the entanglement structure between these quantum degrees of

2
freedom might be most economically described in terms of smooth spacetime. This means
that at some scale the notion of a smooth spacetime has to break down.

However, we know that fundamental quantum field theories such as those in the Standard
Model rely on a background spacetime to define a notion of locality and causality. These
notions tell us that quantum states in these theories can be well localized in the infinite past
and future. We do exactly this in particle accelerators; locality and causality are well-tested
features of quantum field theories describing particles in the Standard Model, which has
itself been tested rigorously to the TeV scale. Thus spacetime is a key concept in defining
consistent quantum field theories that describe the interactions of these particles.

This implies that the scale at which spacetime breaks down has to be much smaller than
the scales probed by particle accelerators. This scale might be anywhere between the TeV
scale and the Planck scale. Thus a theory of emergent spacetie could be related to a theory
of quantum gravity, which is the purported UV completion (which means the underlying
microscopic theory) of general relativity. Therefore it is fruitful to try to understand what
quantum gravity is before we can analyze whether a description of the universe in terms of
a smooth spacetime is emergent or fundamental.

2 Quantum Gravity
Quantum gravity is hard. Many attempts have been made to quantize ordinary gravitational
theories. Most of these ideas are plagued with the lack of a consistent conceptual framework
or a lack of analytic control. For example, if we simply take a well tested known theory of
gravity, Einstein gravity, and interpret it as an ordinary quantum field theory by attempting
to canonically quantize classical fields we get divergences that cannot be removed. Ordinarily
we choose to trust a quantum field theory even if divergences appear as long as the theory
has a property known as renormalizability. This means that once we know a finite number
of experimental parameters of a theory (particle masses, interaction couplings) at a certain
scale, everything else can be predicted.

This is not the case for Einstein gravity and hence we call it a non-renormalizable theory.
If one attempts to quantize this theory using the standard Feynman path integral prescrip-
tion, the quantum corrections to the classical theory diverge and cannot be removed by
renormalization can be done in the Standard Model. Renormalizable theories are sometimes
viewed as more fundamental than non-renormalizable ones because these divergences can be
systematically removed to sensical predictions (given that we ask the right questions). More
specifically this means there are a finite number of independent parameters that have to be
measured to fully specify the theory. In the case of a non-renormalizable theory one would
need to experimentally measure an infinite number of parameters which demolishes all hope
of making sensible predictions from such a theory. Therefore it seems to be the case that
attempting to interpret Einstein gravity as an ordinary quantum field theory isn’t consistent
with the typical framework of a quantum field theory. This means that merging together
quantum mechanics and gravity is more complicated than we once hoped.

3
 One thing we know for certain about quantum gravity is that its semiclassical approxi-
mation should include virtual effects on top of the classical theory of gravity. This leads to
the conceptually clean idea that if we linearize the equations of motion of a theory of gravity,
we should get a gauge theory with a massless particle that mediates the force of gravity.
This particle is is a massless spin-2 particle known as a graviton. Essentially we employ the
background field method of quantum field theory. Choose a particular background metric
and linearize fluctuations around it:

gµν = g̃µν + hµν (1)

where gµν is the full metric that splits up into the background g̃µν and the fluctuations hµν .
We can then linearize the action around this and find an equation of motion for the graviton.
Schematically, this looks like

√
Z
1
S= gR (2)
16πGN
Z
S ∂h∂h + h∂h∂h + O(∂ 2 h) (3)

The first term is just the standard kinetic term of a quantum fields. The following terms
are self-interactions, which follows from the fact that gravitons themselves gravitate. Thus
for any theory that claims to be a theory of emergent spacetime must have a graviton in its
spectrum.

2.1 Hints from String Theory: Ambiguity of Spacetime

String theory is a potential candidate for a UV completion of General Relativity, or a theory
of quantum gravity. We know from popular science that string theory says that all particles
and forces can be reproduced from vibrating strings. If one works out the modes of a quantum
string, it is ”easy” to see that a massless spin-2 particle always exists in the excitation spec-
trum of any string theory! If we want the emergence of spacetime to happen at the Planck
scale, we would expect that string theory has something to say about emergent spacetime.
In string theory, geometry is emergent from the collective behavior of the low-energy modes
of strings. In some sense, spacetime itself is a coherent state of gravitons, which we know
always exist as modes of the vibrating strings.

One of the reasons that string theory receives so much attention is because a description
of spacetime (or Einstein’s equations) emerges quite naturally from string theory. If strings
are the fundamental building blocks of nature, they should themselves be featureless and it
doesn’t make much sense to ask the question, ”What are strings made of?”. This is done
mathematically by imposing that the coordinates that parameterize strings should not be
”physical”, meaning that Lagrangian describing the dynamics of strings are parameterization
invariant. An analogous statement can be made for the parameterization of time along a
particle’s worldline. For a string this amounts to imposing conformal invariance (invariance
under local ”streching” of spacetime that preserve angles between vectors). If we want

4
conformal invariance to persist at the quantum level we need to look at the fixed points
of the theory, or when the beta functions go to zero. In a quantum field theory the beta
functions tell you how the couplings, or the interaction strengths, change with energy scale.
In fact, Einstein’s equations (along with additional matter fields) fall right out of imposing
conformal invariance at the quantum level [3]!
X
i 1 matter
βµν ≈ (Rµν − gµν R) − Tµν =0 (4)
i
2

Furthermore, even after this spacetime is set by the dynamics of the string it is not in any
sense unique. For example, in string theory there is an exact symmetry known as a T-duality.
What it tells us is that the spectrum of a string that is wrapped around a compactified circle
of a certain radius R is exactly the same as that of a string wrapped around a circle of radius
proportional to 1/R. This leads to a dual description of the string theory in these distinct
spacetimes. In this case, both have the same topology, but in one of them, if R happens to be
very small, meaning this compactified dimension is immeasurably small and is not accessible
in the low energy theory. In the other spacetime, the radius of this compact dimension is
1/R, which is a large extra dimension. And yet they are somehow described by exactly the
same solution of string theory. This suggests that there is an ambiguity in the description
of the world in terms of a unique spacetime. This suggest further evidence that spacetime
emerges from a more fundamental theory [2].

2.2 Gravitons and Weinberg-Witten theorem

Given that we know a theory of gravity has to contain a graviton, it is appealing to think that
if gravity should be emergent from a different microscopic theory then the graviton might
be a composite particle of an underlying quantum field theory living in flat space. This
is analogous to how cooper pairs are composite quasi-particles. This line of reasoning was
analyzed by Weinberg and Witten in 1980. They found that massless particles in a quantum
field theory with spin greater than 1 cannot carry a conserved stress-energy tensor [4]. If the
theory doesn’t have gravity to begin with (no diffeomorphism invariance), the stress energy
tensor is gauge-invariant and this theorem applies.

This is to say that if we have a theory with quantum fields on flat spacetime, an emer-
gent/composite graviton cannot exist. To get emergent spacetime there needs to be some
mechanism that allows the proposed theory to evade the Weinberg-Witten theorem. This
of course happens for General Relativity because we do have diffeomorphism invariance and
the stress energy tensor isn’t gauge invariant.

2.3 Gravity from Thermodynamics

Another interesting direction is the idea that Einsteins equations are actually a thermody-
namic equation of state. It is generally true that in thermodynamics, if one has information
about the dependence of the entropy on other thermodynamic variables it is possible to easily
find the equation of state. In gravitational theories we know about how to relate the entropy

5
Figure 1: δQ is matter flux that crosses the horizon of a codimension 2 space-like surface P.

to geometric data in certain situations from Bekensteins original argument that black holes
should have an entropy associated with them that is proportional to the area of the horizon.

A
S= /, . (5)
4GN

We make the assumption that this true of all horizons. We also need the first law of thermo-
dynamics in the gravitational context

δQ = T δS (6)
Z
= Tab χa dΣb , (7)
H

where Tab is the stress energy tensor, χa is vector field that generates the horizon and the
integration region is over the horizon.
Given that we have information about the matter flux through the horizon, the author
uses Raychudhuri’s equations (which are a statement about how the curvature of a spacetime
affects how light bends) to derive Einstein’s equations as a thermodynamic equation of state.
One starts with Raychudhuri’s equations:

dθ 1
= − θ2 − σ 2 − Rab k a k b (8)
dλ 2

6
where k a are the null vectors describing the light-like geodesic. θ is function of the affine pa-
rameter λ (which parameterizes the light-like curve) at which the light-like geodesic ”bends”
locally due to the curvature of spacetime. We can ignore the first two terms, and Rab is
the Ricci tensor describing the curvature of spacetime. Then relate this to the variation of
the entropy (which is related to the variation in the area) and one can extract Einstein’s
equations.

If Einstein’s equations are thermodynamic in nature, this suggests that trying to quantize
gravity by canonical quantization of the metric makes about as much sense as trying to canon-
ically quantize the equation of state of hydrodynamics of fluids (Navier-Stokes equations) by
quantizing the velocity field. Since the assumptions of this model involve pure thermodynam-
ical statements, this theory doesn’t necessarily rule out a connection to a deeper underlying
microscopic theory. If we make the further identification of the thermodynamic entropy
with boltzmann entropy (in equilibrium situations), as long as the microscopic theory can
reproduce the area law the thermodynamic argument of Jacobson is not ruled out.

3 Emergence of Gravity in AdS/CFT

Another example of emergent spacetime comes from the program of AdS/CFT. The state-
ment of the correspondence suggests that the degrees of freedom of gravity should have a
description in the quantum field theory that lives on the boundary of the spacetime where
the gravitational theory lives.

Figure 2: AdS/CFT in a picture

7
One should note that this is a strange form of emergence, because normally in condensed
matter systems the dynamics of electrons and atoms conspire in a certain way that can be
described by a spatially varying order parameter. This order parameter field usually lives on
top of the existing spacetime that the atoms live in.
In the case of AdS/CFT, the dynamics of the QFT degrees of freedom conspire to be
described by a theory in one more dimension and the degrees of freedom in this new theory
act like they live on a dynamical curved spacetime! This is very strange and we would like
to understand how exactly this happens. One idea is that the entanglement between these
QFT degrees of freedom somehow generates the gravitational degrees of freedom.

A simple example of this is the Ryu - Takayanagi conjecture, where the entanglement
entropy of the quantum fields in the boundary (a measure of entanglement in quantum field
theories) can be calculated in the bulk theory by computing the area of a minimal surface
whose boundary is the entanglement cut [6].

Figure 3: The bulk surface à has a boundary that coincides with the boundary of the sub-
region A in the CFT.

1
S(A) = Area(Ã) (9)
4GN
The left-hand side of this formula is the entanglement entropy of the quantum field theory
restricted to the subregion A. It measures the amount of entanglement between degrees of
freedom inside A and the complement Ā.

One could imagine varying the region A on the QFT by deforming the entanglement cut,
which would imply a slightly deformed minimal surface. If the Ryu-Takyanagi conjecture
holds, the change in the entanglement entropy of the QFT living on the boudary should
exactly reproduce the change in the area of the minimal surface. If this is the case, one could
map out all the bulk simply by changing the enclosed area on the boundary.

8
 The most important facet of this story is that the QFT has certain operators (or quantum
fields) that have dual descriptions in the bulk theory. We expect that for a theory of quantum
gravity, there should be a particle that mediates the gravitational force. This is known as
a graviton. This is an important concept because, in some sense spacetime itself can be
considered a condensate of gravitons. These gravitons have a dual operator in the CFT: the
stress energy tensor. This should always exist in any reasonable CFT.
If we can motivate a general mapping between the operators on the boundary and oper-
ators in the bulk this suggests that there is an isomorphism between the Hilbert spaces of
the two theories. For a particular CFT, there might exist a certain sector that can actually
be described by a gravitational theory. To check this, we use the mapping we have already
established to define a dual bulk operator to the stress-energy tensor:

CF T
Tµν ⇐⇒ gµν . (10)

4 Comparison with Experiment

There are no experiments that can definitely disprove any of these theories, but we can put
constraints at which scale an emergent spacetime should breakdown. However a breakdown
of spacetime implies that Lorentz-invariance breaks down at some scale. At the level of a
photon, Lorentz invariance implies that the dispersion relation of a photon should be pro-
portional to the frequency by a constant. If there is any energy dependence at all in this
constant, we cannot say that Lorentz invariance holds for all length scales. This test has
been done repeatedly and the most recent experiment suggests that the photon dispersion
relation holds up to energy scales above the Planck scale [7]!

5 Conclusion
It might be the case that gravitational degrees of freedom are actually lurking in quantum
field theories. Several directions of research suggest theoretical ideas of why this is the
case, but as of today no one can say for certain whether gravity and spacetime are actually
emergent in our universe.

References
[1] P. W. Anderson, Science 177, no. 4047, 393 (1972). doi:10.1126/science.177.4047.393
[2] N. Seiberg, hep-th/0601234.
[3] C. G. Callan, Jr., E. J. Martinec, M. J. Perry and D. Friedan, Nucl. Phys. B 262, 593
(1985). doi:10.1016/0550-3213(85)90506-1
[4] S. Weinberg and E. Witten, Phys. Lett. 96B, 59 (1980). doi:10.1016/0370-
2693(80)90212-9

9
[5] T. Jacobson, Phys. Rev. Lett. 75, 1260 (1995) doi:10.1103/PhysRevLett.75.1260 [gr-
qc/9504004].

[6] S. Ryu and T. Takayanagi, JHEP 0608, 045 (2006) doi:10.1088/1126-6708/2006/08/045

[hep-th/0605073].

[7] M. Ackermann et al. [Fermi GBM/LAT Collaboration], Nature 462, 331 (2009)
doi:10.1038/nature08574 [arXiv:0908.1832 [astro-ph.HE]].

10