This action might not be possible to undo. Are you sure you want to continue?
qxd 6/16/04 8:37 AM Page iv Quark03 Quark03:Desktop Folder:ChapterFM:
FFIRS.qxd 6/16/04 8:37 AM Page iv Quark03 Quark03:Desktop Folder:ChapterFM:
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
AN INTRODUCTION TO
BLACK HOLES, INFORMATION,
AND THE STRING THEORY
REVOLUTION
The Holographic Universe
Leonard Susskind
∗
James Lindesay
†
∗
Permanent address, Department of Physics, Stanford University, Stanford, CA 94305
4060
†
Permanent address, Department of Physics, Howard University, Washington, DC 20059
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
vi
British Library CataloguinginPublication Data
A catalogue record for this book is available from the British Library.
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
ISBN 9812560831
ISBN 9812561315 (pbk)
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.
Copyright © 2005 by World Scientific Publishing Co. Pte. Ltd.
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
Printed in Singapore.
AN INTRODUCTION TO BLACK HOLES, INFORMATION AND THE STRING
THEORY REVOLUTION
The Holographic Universe
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Preface
It is now almost a century since the year 1905, in which the principle
of relativity and the hypothesis of the quantum of radiation were intro
duced. It has taken most of that time to synthesize the two into the modern
quantum theory of ﬁelds and the standard model of particle phenomena.
Although there is undoubtably more to be learned both theoretically and
experimentally, it seems likely that we know most of the basic principles
which follow from combining the special theory of relativity with quantum
mechanics. It is unlikely that a major revolution will spring from this soil.
By contrast, in the 80 years that we have had the general theory of rel
ativity, nothing comparable has been learned about the quantum theory of
gravitation. The methods that were invented to quantize electrodynamics,
which were so successfully generalized to build the standard model, prove
wholly inadequate when applied to gravitation. The subject is riddled with
paradox and contradiction. One has the distinct impression that we are
thinking about the things in the wrong way. The paradigm of relativistic
quantum ﬁeld theory almost certainly has to be replaced.
How then are we to go about ﬁnding the right replacement? It seems
very unlikely that the usual incremental increase of knowledge from a com
bination of theory and experiment will ever get us where we want to go,
that is, to the Planck scale. Under this circumstance our best hope is an
examination of fundamental principles, paradoxes and contradictions, and
the study of gedanken experiments. Such strategy has worked in the past.
The earliest origins of quantum mechanics were not experimental atomic
physics, radioactivity, or spectral lines. The puzzle which started the whole
thing was a contradiction between the principles of statistical thermody
namics and the ﬁeld concept of Faraday and Maxwell. How was it possible,
Planck asked, for the inﬁnite collection of radiation oscillators to have a
ﬁnite speciﬁc heat?
vii
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
viii Black Holes, Information, and the String Theory Revolution
In the case of special relativity it was again a conceptual contradiction
and a gedanken experiment which opened the way. According to Einstein,
at the age of 15 he formulated the following paradox: suppose an observer
moved along with a light beam and observed it. The electromagnetic ﬁeld
would be seen as a static, spatially varying ﬁeld. But no such solution
to Maxwell’s equations exists. By this simple means a contradiction was
exposed between the symmetries of Newton’s and Galileo’s mechanics and
those of Maxwell’s electrodynamics.
The development of the general theory from the principle of equiva
lence and the manintheelevator gedanken experiment is also a matter of
historical fact. In each of these cases the consistency of readily observed
properties of nature which had been known for many years required revo
lutionary paradigm shifts.
What known properties of nature should we look to, and which paradox
is best suited to our present purposes? Certainly the most important facts
are the success of the general theory in describing gravity and of quantum
mechanics in describing the microscopic world. Furthermore, the two the
ories appear to lead to a serious clash that once again involves statistical
thermodynamics in an essential way. The paradox was discovered by Ja
cob Bekenstein and turned into a serious crisis by Stephen Hawking. By
an analysis of gedanken experiments, Bekenstein realized that if the sec
ond law of thermodynamics was not to be violated in the presence of a
black hole, the black hole must possess an intrinsic entropy. This in itself
is a source of paradox. How and why a classical solution of ﬁeld equations
should be endowed with thermodynamical attributes has remained obscure
since Bekenstein’s discovery in 1972.
Hawking added to the puzzle when he discovered that a black hole will
radiate away its energy in the form of Planckian black body radiation.
Eventually the black hole must completely evaporate. Hawking then raised
the question of what becomes of the quantum correlations between matter
outside the black hole and matter that disappears behind the horizon. As
long as the black hole is present, one can do the bookkeeping so that it is the
black hole itself which is correlated to the matter outside. But eventually
the black hole will evaporate. Hawking then made arguments that there is
no way, consistent with causality, for the correlations to be carried by the
outgoing evaporation products. Thus, according to Hawking, the existence
of black holes inevitably causes a loss of quantum coherence and breakdown
of one of the basic principles of quantum mechanics – the evolution of
pure states to pure states. For two decades this contradiction between
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Preface ix
the principles of general relativity and quantum mechanics has so puzzled
theorists that many now see it as a serious crisis.
Hawking and much of the traditional relativity community have been of
the opinion that the correct resolution of the paradox is simply that quan
tum coherence is lost during black hole evaporation. From an operational
viewpoint this would mean that the standard rules of quantum mechanics
would not apply to processes involving black holes. Hawking further ar
gued that once the loss of quantum coherence is permitted in black hole
evaporation, it becomes compulsory in all processes involving the Planck
scale. The world would behave as if it were in a noisy environment which
continuously leads to a loss of coherence. The trouble with this is that there
is no known way to destroy coherence without, at the same time violating
energy conservation by heating the world. The theory is out of control
as argued by Banks, Peskin and Susskind, and ’t Hooft. Throughout this
period, a few theorists, including ’t Hooft and Susskind, have felt that the
basic principles of quantum mechanics and statistical mechanics have to be
made to coexist with black hole evaporation.
’t Hooft has argued that by resolving the paradox and removing the
contradiction, the way to the new paradigm will be opened. The main
purpose of this book is to lay out this case.
A second purpose involves development of string theory as a uniﬁed de
scription of elementary particles, including their gravitational interactions.
Although still very incomplete, string theory appears to be a far more con
sistent mathematical framework for quantum gravity than ordinary ﬁeld
theory. It is therefore worth exploring the diﬀerences between string the
ory and ﬁeld theory in the context of black hole paradoxes. Quite apart
from the question of the ultimate correctness and consistency of string the
ory, there are important lessons to be drawn from the diﬀerences between
these two theories. As we shall see, although string theory is usually well
approximated by local quantum ﬁeld theory, in the neighborhood of a black
hole horizon the diﬀerences become extreme. The analysis of these diﬀer
ences suggests a resolution of the black hole dilemma and a completely new
view of the relations between space, time, matter, and information.
The quantum theory of black holes, with or without strings, is far from
being a textbook subject with well deﬁned rules. To borrow words from
Sidney Coleman, it is a “trackless swamp” with many false but seductive
paths and no maps. To navigate it without disaster we will need some
beacons in the form of trusted principles that we can turn to for direction.
In this book the absolute truth of the following four propositions will be
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
x Black Holes, Information, and the String Theory Revolution
assumed: 1) The formation and evaporation of a black hole is consistent
with the basic principles of quantum mechanics. In particular, this means
that observations performed by observers who remain outside the black
hole can be described by a unitary time evolution. The global process,
beginning with asymptotic infalling objects and ending with asymptotic
outgoing evaporation products is consistent with the existence of a uni
tary Smatrix. 2) The usual semiclassical description of quantum ﬁelds
in a slowly varying gravitational background is a good approximation to
certain coarse grained features of the black hole evolution. Those features
include the thermodynamic properties, luminosity, energy momentum ﬂux,
and approximate black body character of Hawking radiation. 3) Thirdly we
assume the usual connection between thermodynamics and quantum sta
tistical mechanics. Thermodynamics results from coarse graining a more
microscopic description so that states with similar macroscopic behavior
are lumped into a single thermodynamic state. The existence of a thermo
dynamics will be taken to mean that a microscopic set of degrees of free
dom exists whose coarse graining leads to the thermal description. More
speciﬁcally we assume that a thermodynamic entropy S implies that ap
proximately exp(S) quantum states have been lumped into one thermal
state.
These three propositions, taken by themselves, are in no way radical.
Proposition 1 and 3 apply to all known forms of matter. Proposition 2
may perhaps be less obvious, but it nevertheless rests on wellestablished
foundations. Once we admit that a black hole has energy, entropy, and
temperature, it must also have a luminosity. Furthermore the existence of a
thermal behavior in the vicinity of the horizon follows from the equivalence
principle as shown in the fundamental paper of Unruh. Why then should
any of these principles be considered controversial? The answer lies in a
fourth proposition which seems as inevitable as the ﬁrst three: 4) The
fourth principle involves observers who fall through the horizon of a large
massive black hole, carrying their laboratories with them. If the horizon
scale is large enough so that tidal forces can be ignored, then a freely
falling observer should detect nothing out of the ordinary when passing the
horizon. The usual laws of nature with no abrupt external perturbations
will be found valid until the inﬂuence of the singularity is encountered. In
considering the validity of this fourth proposition it is important to keep in
mind that the horizon is a global concept. The existence, location, size, and
shape of a horizon depend not only on past occurrences, but also on future
events. We ourselves could right now be at the horizon of a gigantic black
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Preface xi
hole caused by matter yet to collapse in the future. The horizon in classical
relativity is simply the mathematical surface which separates those points
from which any light ray must hit a singularity from those where light may
escape to inﬁnity. A mathematical surface of this sort should have no local
eﬀect on matter in its vicinity.
In Chapter 9 we will encounter powerful arguments against the mutual
consistency of propositions 1–4. The true path through the swamp at times
becomes so narrow it seems to be a dead end, while all around false paths
beckon. Beware the willo’thewisp and don’t lose your nerve.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
xii Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Contents
Preface vii
Part 1: Black Holes and Quantum Mechanics 1
1. The Schwarzschild Black Hole 3
1.1 Schwarzschild Coordinates . . . . . . . . . . . . . . . . . . . 3
1.2 Tortoise Coordinates . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Near Horizon Coordinates (Rindler space) . . . . . . . . . . 8
1.4 Kruskal–Szekeres Coordinates . . . . . . . . . . . . . . . . . 10
1.5 Penrose Diagrams . . . . . . . . . . . . . . . . . . . . . . . 14
1.6 Formation of a Black Hole . . . . . . . . . . . . . . . . . . . 15
1.7 Fidos and Frefos and the Equivalence Principle . . . . . . . 21
2. Scalar Wave Equation in a Schwarzschild Background 25
2.1 Near the Horizon . . . . . . . . . . . . . . . . . . . . . . . . 28
3. Quantum Fields in Rindler Space 31
3.1 Classical Fields . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 Entanglement . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3 Review of the Density Matrix . . . . . . . . . . . . . . . . . 34
3.4 The Unruh Density Matrix . . . . . . . . . . . . . . . . . . 36
3.5 Proper Temperature . . . . . . . . . . . . . . . . . . . . . . 39
4. Entropy of the Free Quantum Field in Rindler Space 43
4.1 Black Hole Evaporation . . . . . . . . . . . . . . . . . . . . 48
xiii
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
xiv Black Holes, Information, and the String Theory Revolution
5. Thermodynamics of Black Holes 51
6. Charged Black Holes 55
7. The Stretched Horizon 61
8. The Laws of Nature 69
8.1 Information Conservation . . . . . . . . . . . . . . . . . . 69
8.2 Entanglement Entropy . . . . . . . . . . . . . . . . . . . . 71
8.3 Equivalence Principle . . . . . . . . . . . . . . . . . . . . 77
8.4 Quantum Xerox Principle . . . . . . . . . . . . . . . . . . 79
9. The Puzzle of Information Conservation in Black Hole
Environments 81
9.1 A Brick Wall? . . . . . . . . . . . . . . . . . . . . . . . . . 84
9.2 Black Hole Complementarity . . . . . . . . . . . . . . . . 85
9.3 Baryon Number Violation . . . . . . . . . . . . . . . . . . 89
10. Horizons and the UV/IR Connection 95
Part 2: Entropy Bounds and Holography 99
11. Entropy Bounds 101
11.1 Maximum Entropy . . . . . . . . . . . . . . . . . . . . . 101
11.2 Entropy on Lightlike Surfaces . . . . . . . . . . . . . . . 105
11.3 Friedman–Robertson–Walker Geometry . . . . . . . . . . 110
11.4 Bousso’s Generalization . . . . . . . . . . . . . . . . . . . 114
11.5 de Sitter Cosmology . . . . . . . . . . . . . . . . . . . . . 119
11.6 Anti de Sitter Space . . . . . . . . . . . . . . . . . . . . . 123
12. The Holographic Principle and Anti de Sitter Space 127
12.1 The Holographic Principle . . . . . . . . . . . . . . . . . 127
12.2 AdS Space . . . . . . . . . . . . . . . . . . . . . . . . . . 128
12.3 Holography in AdS Space . . . . . . . . . . . . . . . . . . 130
12.4 The AdS/CFT Correspondence . . . . . . . . . . . . . . 133
12.5 The Infrared Ultraviolet Connection . . . . . . . . . . . . 135
12.6 Counting Degrees of Freedom . . . . . . . . . . . . . . . 138
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Contents xv
13. Black Holes in a Box 141
13.1 The Horizon . . . . . . . . . . . . . . . . . . . . . . . . . 144
13.2 Information and the AdS Black Hole . . . . . . . . . . . 144
Part 3: Black Holes and Strings 149
14. Strings 151
14.1 Light Cone Quantum Mechanics . . . . . . . . . . . . . . 153
14.2 Light Cone String Theory . . . . . . . . . . . . . . . . . 156
14.3 Interactions . . . . . . . . . . . . . . . . . . . . . . . . . 159
14.4 Longitudinal Motion . . . . . . . . . . . . . . . . . . . . 161
15. Entropy of Strings and Black Holes 165
Conclusions 175
Bibliography 179
Index 181
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
PART 1
Black Holes and Quantum Mechanics
1
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
2
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 1
The Schwarzschild Black Hole
Before beginning the study of the quantum theory of black holes, one
must ﬁrst become thoroughly familiar with the geometry of classical black
holes in a variety of diﬀerent coordinate systems. Each coordinate system
that we will study has its own particular utility, and no one of them is in
any sense the best or most correct description. For example, the Kruskal–
Szekeres coordinate system is valuable for obtaining a global overview of the
entire geometry. It can however be misleading when applied to observations
made by distant observers who remain outside the horizon during the entire
history of the black hole. For these purposes, Schwarzschild coordinates, or
the related tortoise coordinates, which cover only the exterior of the horizon
are in many ways more valuable.
We begin with the simplest spherically symmetric static uncharged
black holes described by Schwarzschild geometry.
1.1 Schwarzschild Coordinates
In Schwarzschild coordinates, the Schwarzschild geometry is manifestly
spherically symmetric and static. The metric is given by
dτ
2
= (1 −
2MG
r
)dt
2
−(1 −
2MG
r
)
−1
dr
2
−r
2
dΩ
2
= g
µν
dx
µ
dx
ν
.
(1.1.1)
where dΩ
2
≡ dθ
2
+ sin
2
θdφ
2
.
The coordinate t is called Schwarzschild time, and it represents the time
recorded by a standard clock at rest at spatial inﬁnity. The coordinate r
is called the Schwarzschild radial coordinate. It does not measure proper
3
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
4 Black Holes, Information, and the String Theory Revolution
spatial distance from the origin, but is deﬁned so that the area of the 2
sphere at r is 4πr
2
. The angles θ, φ are the usual polar and azimuthal
angles. In equation 1.1.1 we have chosen units such that the speed of light
is 1.
The horizon, which we will tentatively deﬁne as the place where g
00
vanishes, is given by the coordinate r = 2MG. At the horizon g
rr
becomes
singular. The question of whether the geometry is truly singular at the
horizon or if it is the choice of coordinates which are pathological is subtle.
In what follows we will see that no local invariant properties of the geometry
are singular at r = 2MG. Thus a small laboratory in free fall at r =
2MG would record nothing unusual. Nevertheless there is a very important
sense in which the horizon is globally special if not singular. To a distant
observer the horizon represents the boundary of the world, or at least that
part which can inﬂuence his detectors.
To determine whether the local geometry is singular at r = 2MG we can
send an explorer in from far away to chart it. For simplicity let’s consider
a radially freely falling observer who is dropped from rest from the point
r = R. The trajectory of the observer in parametric form is given by
r =
R
2
(1 + cosη) (1.1.2)
τ =
R
2
R
2MG
1/2
(η + sinη) (1.1.3)
t = (
R
2
+ 2MG)
R
2MG
−1
1/2
η +
R
2
R
2MG
−1
1/2
sinη
+2MGlog
(
R
2MG
−1
)
1/2
+tan
η
2
(
R
2MG
−1
)
1/2
−tan
η
2
[0 < η < π]
(1.1.4)
where τ is the proper time recorded by the observer’s clock. From these
overly complicated equations it is not too diﬃcult to see that the observer
arrives at the point r = 0 after a ﬁnite interval
τ =
π
2
R
R
2MG
1
2
(1.1.5)
Evidently the proper time when crossing the horizon is ﬁnite and smaller
than the expression in equation 1.1.5.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 5
What does the observer encounter at the horizon? An observer in free
fall is not sensitive to the components of the metric, but rather senses the
tidal forces or curvature components. Deﬁne an orthonormal frame such
that the observer is momentarily at rest. We can construct unit basis
vectors, ˆ τ, ˆ ρ,
ˆ
θ,
ˆ
φ with ˆ τ oriented along the observer’s instantaneous time
axis, and ˆ ρ pointing radially out. The nonvanishing curvature components
are given by
R
ˆ τ
ˆ
θˆ τ
ˆ
θ
= R
ˆ τ
ˆ
φˆ τ
ˆ
φ
= −R
ˆ ρ
ˆ
θˆ ρ
ˆ
θ
= −R
ˆ ρ
ˆ
φˆ ρ
ˆ
φ
=
MG
r
3
R
ˆ
θ
ˆ
φ
ˆ
θˆ τ
= −R
ˆ τ ˆ ρˆ τ ˆ ρ
=
2MG
r
3
(1.1.6)
Thus all the curvature components are ﬁnite and of order
R(Horizon) ∼
1
M
2
G
2
(1.1.7)
at the horizon. For a large mass black hole they are typically very small.
Thus the infalling observer passes smoothly and safely through the horizon.
On the other hand the tidal forces diverge as r → 0 where a true local
singularity occurs. At this point the curvature increases to the point where
the classical laws of nature must fail.
Let us now consider the history of the infalling observer from the view
point of a distant observer. We may suppose that the infalling observer
sends out signals which are received by the distant observer. The ﬁrst
surprising thing we learn from equations 1.1.2, 1.1.3, and 1.1.4 is that the
crossing of the horizon does not occur at any ﬁnite Schwarzschild time. It is
easily seen that as r tends to 2MG, t tends to inﬁnity. Furthermore a signal
originating at the horizon cannot reach any point r > 2MG until an inﬁ
nite Schwarzschild time has elapsed. This is shown in Figure 1.1. Assuming
that the infalling observer sends signals at a given frequency ν, the distant
observer sees those signals with a progressively decreasing frequency. Over
the entire span of Schwarzschild time the distant observer records only a
ﬁnite number of pulses from the infalling transmitter. Unless the infalling
observer increases the frequency of his/her signals to inﬁnity as the horizon
is approached, the distant observer will inevitably run out of signals and
lose track of the transmitter after a ﬁnite number of pulses. The limits
imposed on the information that can be transmitted from near the horizon
are not so severe in classical physics as they are in quantum theory. Accord
ing to classical physics the infalling observer can use an arbitrarily large
carrier frequency to send an arbitrarily large amount of information using
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
6 Black Holes, Information, and the String Theory Revolution
Black
Hole
c = Signal
originating
near horizon
d = Signal
from infalling
to distant
a = distant
observer
b = infalling
observer
c d
a
b
b
Fig. 1.1 Infalling observer sending signals to distant Schwarzschild observer
an arbitrarily small energy without signiﬁcantly disturbing the black hole
and its geometry. Therefore, in principle, the distant observer can obtain
information about the neighborhood of the horizon and the infalling sys
tem right up to the point of horizon crossing. However quantum mechanics
requires that to send even a single bit of information requires a quantum of
energy. As the observer approaches the horizon, this quantum must have
higher and higher frequency, implying that the observer must have had a
large energy available. This energy will back react on the geometry, dis
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 7
turbing the very quantity to be measured. Thereafter, as we shall see, no
information can be transmitted from behind the horizon.
1.2 Tortoise Coordinates
A change of radial coordinate maps the horizon to minus inﬁnity so that
the resulting coordinate system covers only the region r > 2MG. We deﬁne
the tortoise coordinate r
∗
by
1
1 −
2MG
r
dr
2
=
1 −
2MG
r
(dr
∗
)
2
(1.2.8)
so that
dτ
2
=
1 −
2MG
r
[dt
2
−(dr
∗
)
2
] −r
2
dΩ
2
(1.2.9)
The interesting point is that the radialtime part of the metric now has a
particularly simple form, called conformally ﬂat. A space is called confor
mally ﬂat if its metric can be brought to the form
dτ
2
= F(x) dx
µ
dx
ν
η
µν
(1.2.10)
with η
µν
being the usual Minkowski metric. Any twodimensional space
is conformally ﬂat, and a slice through Schwarzschild space at ﬁxed θ, φ
is no exception. In equation 1.2.9 the metric of such a slice is manifestly
conformally ﬂat. Furthermore it is also static.
The tortoise coordinate r
∗
is given explicitly by
r
∗
= r + 2MGlog
r −2MG
2MG
(1.2.11)
Note: r
∗
→ −∞ at the horizon.
We shall see that wave equations in the black hole background have a very
simple form in tortoise coordinates.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
8 Black Holes, Information, and the String Theory Revolution
1.3 Near Horizon Coordinates (Rindler space)
The region near the horizon can be explored by replacing r by a coor
dinate ρ which measures proper distance from the horizon:
ρ =
r
2MG
g
rr
(r
) dr
=
r
2MG
(1 −
2MG
r
)
−
1
2
dr
=
r (r −2MG) + 2MGsinh
−1
(
r
2MG
−1)
(1.3.12)
In terms of ρ and t the metric takes the form
dτ
2
=
1 −
2MG
r(ρ)
dt
2
− dρ
2
− r(ρ)
2
dΩ
2
(1.3.13)
Near the horizon equation 1.3.12 behaves like
ρ ≈ 2
2MG(r −2MG) (1.3.14)
giving
dτ
2
∼
= ρ
2
dt
4MG
2
− dρ
2
− r
2
(ρ) dΩ
2
(1.3.15)
Furthermore, if we are interested in a small angular region of the horizon
arbitrarily centered at θ = 0 we can replace the angular coordinates by
Cartesian coordinates
x = 2MGθ cosφ
y = 2MGθ sinφ
(1.3.16)
Finally, we can introduce a dimensionless time ω
ω =
t
4MG
(1.3.17)
and the metric then takes the form
dτ
2
= ρ
2
dω
2
− dρ
2
− dx
2
− dy
2
(1.3.18)
It is now evident that ρ and ω are radial and hyperbolic angle variables
for an ordinary Minkowski space. Minkowski coordinates T, Z can be
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 9
deﬁned by
T = ρ sinhω
Z = ρ coshω
(1.3.19)
to get the familiar Minkowski metric
dτ
2
= dT
2
− dZ
2
− dX
2
− dY
2
(1.3.20)
It should be kept in mind that equation 1.3.20 is only accurate near r =
2MG, and only for a small angular region. However it clearly demonstrates
that the horizon is locally nonsingular, and, for a large black hole, is locally
almost indistinguishable from ﬂat spacetime.
In Figure 1.2 the relation between Minkowski coordinates and the ρ, ω
coordinates is shown. The entire Minkowski space is divided into four
quadrants labeled I, II, III, and IV. Only one of those regions, namely
t=0
ρ=ρ
1
ω=ω
1
ω=ω
2
ρ=ρ
2
I
II
III
IV
Fig. 1.2 Relation between Minkowski and Rindler coordinates
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
10 Black Holes, Information, and the String Theory Revolution
Region I lies outside the black hole horizon. The horizon itself is the
origin T = Z = 0. Note that it is a twodimensional surface in the four
dimensional spacetime. This may appear surprising, since originally the
horizon was deﬁned by the single constraint r = 2MG, and therefore ap
pears to be a three dimensional surface. However, recall that at the horizon
g
00
vanishes. Therefore the horizon has no extension or metrical size in the
time direction.
The approximation of the nearhorizon region by Minkowski space is
called the Rindler approximation. In particular the portion of Minkowski
space approximating the exterior region of the black hole, i.e. Region I, is
called Rindler space. The timelike coordinate, ω, is called Rindler time.
Note that a translation of Rindler time ω →ω + constant is equivalent to
a Lorentz boost in Minkowski space.
1.4 Kruskal Szekeres Coordinates
Finally we can bring the black hole metric to the form
dτ
2
= F(R) [R
2
dω
2
− dR
2
] − r
2
dΩ
2
(1.4.21)
For small ρ equation 1.3.15 shows that ρ ≈ R. A more accurate comparison
with the original Schwarzschild metric gives the following requirements:
R
2
F(R) = 16M
2
G
2
1 −
2MG
r
(1.4.22)
F(R) dR
2
=
1
1 −
2MG
r
dr
2
(1.4.23)
from which it follows that
4MGlog
R
MG
= r + 2MGlog
r −2MG
2MG
= r
∗
(1.4.24)
or
R = MGexp
r
∗
4MG
(1.4.25)
R and ω can be thought of as radial and hyperbolic angular coordinates
of a space which is conformal to ﬂat 1+1 dimensional Minkowski space.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 11
Letting
Re
ω
= V
Re
−ω
= −U
(1.4.26)
be “radial lightlike” variables, the radialtime part of the metric takes the
form
dτ
2
= F(R) dU dV (1.4.27)
The coordinates U, V are shown in Figure 1.3. The surfaces of constant
r are the timelike hyperbolas in Figure 1.3. As r tends to 2MG the hy
perbolas become the broken straight lines H
+
and H
−
which we will call
the extended past and future horizons. Although the extended horizons
lie at ﬁnite values of the Kruskal–Szekeres coordinates, they are located
at Schwarzschild time ±∞. Thus we see that a particle trajectory which
crosses H
+
in a ﬁnite proper time, crosses r = 2MG only after an inﬁnite
Schwarzschild time.
+
H

H
II
I
V U
IV
III
r
=
c
o
n
s
t
a
n
t
Fig. 1.3 U, V Kruskal–Szekeris coordinates
The region of Schwarzschild space with r < 2MG can be taken
to be Region II. In this region the surfaces of constant r are the
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
12 Black Holes, Information, and the String Theory Revolution
spacelike hyperboloids
U V = positive constant (1.4.28)
The true singularity at r = 0 occurs at R
2
= −(MG)
2
, or
U V = (MG)
2
(1.4.29)
The entire maximal analytic extension of the Schwarzschild geometry is
easily described in Kruskal–Szekeres coordinates. It is shown in Figure 1.4.
+
H
S
in
g
u
la
rity
F
u
tu
re
Singularity Past
r
=
2
M
G
t
=
8
r
=
2
M
G
t
=
 8
III
H

U
External
I
H
o
r
i
z
o
n
r
=
c
o
n
s
t
a
n
t
H
o
r
i
z
o
n
IV
II
2
2
R
=
(M
G
)
=
U
V
V
V
V
V
V
V
V
V
V
V
V
V
V
VVVVV
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
VVVVV
V
V
V
V
V
V
V
V
V
V
Fig. 1.4 Maximal analytic extension of Schwarzschild in Kruskal–Szekeris coor
dinates
A useful property of Kruskal–Szekeres coordinates is the fact that light
rays and timelike trajectories always lie within a twodimensional light cone
bounded by 45
o
lines. A radial moving light ray travels on a trajectory
V = constant or U = constant. A nonradially directed light ray or time
like trajectory always lies inside the twodimensional light cone. With this
in mind, it is easy to understand the causal properties of the black hole
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 13
geometry. Consider a point P
1
in Region I. A radially outgoing light ray
from P
1
will escape falling into the singularity as shown in Figure 1.5. An
incoming light ray from P
1
will eventually cross H
+
and then hit the future
singularity. Thus an observer in Region I can send messages to inﬁnity as
well as into Region II.
Future Singularity
Past Singularity
r
=
2
M
G
t
=
8
I
n
f
a
l
l
i
n
g
r
a
d
i
a
l
l
i
g
h
t
r
a
y
O
u
t
g
o
i
n
g
r
a
d
i
a
l
l
i
g
h
t
r
a
y
III
H

U
External
I
H
o
r
i
z
o
n
IV
II
P
1
H
o
r
i
z
o
n
+
H
2 2
R = (MG) = UV
r
=
2
M
G
t
=

8
V
V
V
V
V
V
V
V
V
V
V
V
V
VVVVV
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
VVVVV
V
V
V
V
V
V
V
V
V
V
P
2
Fig. 1.5 Radial light rays from a point in Region I and Region II using Kruskal–
Szekeris coordinates
Consider next Region II. From any point P
2
any signal must eventually
hit the singularity. Furthermore, no signal can ever escape to Region I.
Thus no observer who stays outside r = 2MG can ever be inﬂuenced by
events in Region II. For this reason Region II is said to be behind the
horizon. Regions III and IV, as we will see, are not relevant to the classical
problem of black holes formed by collapsing matter. Nevertheless let us
consider them. From Region III no signal can ever get to Region I, and so
it is also behind the horizon. On the other hand, points in Region IV can
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
14 Black Holes, Information, and the String Theory Revolution
communicate with Region I. Region I however cannot communicate with
Region IV. All of this is usually described by saying that Regions II and
III are behind the future horizon while Regions III and IV are behind the
past horizon.
1.5 Penrose Diagrams
Penrose diagrams are a useful way to represent the causal structure
of spacetimes, especially if, like the Schwarzschild black hole, they have
spherical symmetry. They represent the geometry of a twodimensional
surface of ﬁxed angular coordinates. Furthermore they “compactify” the
geometry so that it can be drawn in total on the ﬁnite plane. As an example,
consider ordinary ﬂat Minkowski space. Ignoring angular coordinates,
dτ
2
= dt
2
− dr
2
− angular part = (dt + dr)(dt − dr) − angular part
(1.5.30)
Radial light rays propagate on the light cone dt ±dr = 0.
Any transformation that is of the form
Y
+
= F(t + r)
Y
−
= F(t −r)
(1.5.31)
will preserve the form of the light cone. We can use such a transformation
to map the entire inﬁnite space 0 ≤ r ≤ ∞, −∞ ≤ t ≤ +∞ to a ﬁnite
portion of the plane. For example
Y
+
= tanh(t + r)
Y
−
= tanh(t −r)
(1.5.32)
The entire spacetime is mapped to the ﬁnite triangle bounded by
Y
+
= 1
Y
−
= −1
Y
+
− Y
−
= 0
(1.5.33)
as shown in Figure 1.6. Also shown in Figure 1.6 are some representative
contours of constant r and t.
There are several inﬁnities on the Penrose diagram. Future and past
timelike inﬁnities (t = ±∞) are the beginnings and ends of timelike tra
jectories. Spacelike inﬁnity (r = ∞) is where all spacelike surfaces end.
In addition to these there are two other inﬁnities which are called I
±
.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 15
8
t=
8
t=+
8
r=
t
=
3
t
=
2
t=
1
t=0
r
=
3
r
=
2
r
=
1
r
=
0
+

Y =1

Y =+1
+
Y =Y
+

Fig. 1.6 Penrose diagram for Minkowski space
They are past and future lightlike inﬁnity, and they represent the origin
of incoming light rays and the end of outgoing light rays.
Similar deformations can be carried out for more interesting geometries,
such as the black hole geometry represent by Kruskal–Szekeres coordinates.
The resulting Penrose diagram is shown in Figure 1.7.
1.6 Formation of a Black Hole
The eternal black hole described by the static Schwarzschild geometry
is an idealization. In nature, black holes are formed from the collapse of
gravitating matter. The simpest model for black hole formation involves a
collapsing thin spherical shell of massless matter. For example, a shell of
photons, gravitons, or massless neutrinos with very small radial extension
and total energy M provides an example.
To construct the geometry, we begin with the empty space Penrose
diagram with the infalling shell represented by an incoming lightlike line
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
16 Black Holes, Information, and the String Theory Revolution
r = 8
t=+ 8
t= 8
Future Singularity
Past Singularity
III
IV
II
I
H
+
H

+

r= 8
t=+ 8
t= 8
Future Singularity
Past Singularity
III
IV
II
H
o
r
i
z
o
n
H
o
r
i
z
o
n
t=
2
t=
2
t=0
t=1
t=
1
+

r
=
r 1
r
=
r
2
Fig. 1.7 Penrose diagram for Schwarzschild black hole, showing regions (top) and
curves of ﬁxed radial position and constant time (bottom)
(see Figure 1.8). The particular value of Y
+
chosen for the trajectory is
arbitrary since any two such values are related by a time translation. The
infalling shell divides the Penrose diagram into two regions, A and B. The
Region A is interior to the shell and represents the initial ﬂat spacetime
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 17
Y

Y
+
B
A

Fig. 1.8 Minkowski space Penrose diagram for radially infalling spherical shell of
massless particles with energy M
before the shell passes. Region B is the region outside the shell and must
be modiﬁed in order to account for the gravitational ﬁeld due to the mass
M.
In Newtonian physics the gravitational ﬁeld exterior to a spherical mass
distribution is uniquely that of a point mass located at the center of the
distribution. Much the same is true in general relativity. In this case
Birkoﬀ’s theorem tells us that the geometry outside the shell must be the
Schwarzschild geometry. Accordingly, we consider the Penrose diagram for
a black hole of mass M divided into regions A’ and B’ by an infalling
massless shell as in Figure 1.9. Once again the particular value of Y
+
chosen for the trajectory is immaterial. Just as in Figure 1.8 where the
Region B is unphysical, in Figure 1.9 the Region A’ is to be discarded.
To form the full classical evolution the regions A of Figure 1.8 and B’ of
Figure 1.9 must be glued together. However this must be done so that
the “radius” of the local two sphere represented by the angular coordinates
(θ, φ) is continuous. In other words, the mathematical identiﬁcation of
the boundaries of A and B’ must respect the continuity of the variable r.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
18 Black Holes, Information, and the String Theory Revolution
Y
+
Y

A’
B’

Fig. 1.9 Penrose diagram of Schwarzschild black hole with radially infalling shell
of massless particle with energy M
Since in both cases r varies monotonically from r = ∞ at I
−
to r = 0,
the identiﬁcation is always possible. One of the two Penrose diagrams will
have to undergo a deformation along the Y
−
direction in order to make
the identiﬁcation smoothly, but this will not disturb the form of the light
cones. Thus in Figure 1.10 we show the resulting Penrose diagram for the
complete geometry. On Fig 1.10, a lightlike surface H is shown as a dotted
line. It is clear that any light ray or timelike trajectory that originates to
the upper left of H must end at the singularity and cannot escape to I
+
(or t = ∞). This identiﬁes H as the horizon. In Region B’ the horizon
is identical to the surface H
+
of Figure 1.7, that is it coincides with the
future horizon of the ﬁnal black hole geometry and is therefore found at
r = 2MG. On the other hand, the horizon also extends into the Region A
where the metric is just that of ﬂat spacetime. In this region the value of
r on the horizon grows from an initial value r = 0 to the value r = 2MG
at the shell.
It is evident from this discussion that the horizon is a global and not
a local concept. In the Region A no local quantity will distinguish the
presence of the horizon whose occurence is due entirely to the future collapse
of the shell.
Consider next a distant observer located on a trajectory with r >>
2MG. The observer originates at past timelike inﬁnity and eventually
ends at future timelike inﬁnity, as shown in Figure 1.11. The distant
observer collects information that arrives at any instant from his backward
light cone. Evidently such an observer never actually sees events on the
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 19
M
a
tc
h
a
re
a
s
o
f
2
s
p
h
e
re
A
B’
r=0
r
=
0
H
r
<
2
M
G
r
=
2
M
G
r=
8
t=
8
+

Fig. 1.10 Penrose diagram for collapsing shell of massless particles
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
20 Black Holes, Information, and the String Theory Revolution
D
i
s
t
a
n
t
o
b
s
e
r
v
e
r
Fig. 1.11 Distant observer to collapsing spherical shell
horizon. In this sense the horizon must be regarded as at the end of time.
Any particle or wave which falls through the horizon is seen by the distant
observer as asymptotically approaching the horizon as it is inﬁnitely red
shifted. At least that is the case classically.
This basic description of black hole formation is much more general than
might be guessed. It applies with very little modiﬁcation to the collapse of
all kinds of massive matter as well as to nonspherical distributions. In all
cases the horizon is a lightlike surface which separates the spacetime into
an inner and an outer region. Any light ray which originates in the inner
region can never reach future asymptotic inﬁnity, or for that matter ever
reach any point of the outer region. The events in the outer region can
send light rays to I
+
and timelike trajectories to t = ∞.
The horizon, as we have seen, is a global concept whose location depends
on all future events. It is composed of a family of light rays or null geodesics,
passing through each spacetime point on the horizon. This is shown in
Figure 1.12. Notice that null geodesics are vertical after the shell crosses
the horizon and essentially at 45
o
prior to that crossing. These light rays
are called the generators of the horizon.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 21
Shell of
radially
moving
lightrays
Horizon
Shell crosses
Singularity forms
Horizon
Horizon Forms
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Fig. 1.12 The horizon as family of null geodesics
1.7 Fidos and Frefos and the Equivalence Principle
In considering the description of events near the horizon of a static
black hole from the viewpoint of an external observer
[
1
]
, it is useful to
imagine space to be ﬁlled with static observers, each located at a ﬁxed
(r, θ, φ). Such observers are called ﬁducial observers, or by the whimsical
abbreviation, FIDOS. Each Fido carries a clock which may be adjusted to
record Schwarzschild time t. This means that Fidos at diﬀerent r values
see their own clocks running at diﬀerent proper rates. Alternatively, they
could carry standard clocks which always record proper time τ. At a given
r the relation between Schwarzschild time t and the Fidos proper time τ is
given by
dτ
dt
=
√
g
00
= [1 −
2MG
r
]
1
2
(1.7.34)
Thus, to the Fido near r = 2MG, the Schwarzschild clock appears to run
at a very rapid rate. Another possible choice of clocks would record the
dimensionless hyperbolic angle ω deﬁned by equation 1.3.17.
The spatial location of the Fidos can be labeled by the angular coordi
nates (θ, φ) and any one of the radial variables r, r
∗
, or ρ. Classically the
Fidos can be thought of as mathematical ﬁctions or real but arbitrarily light
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
22 Black Holes, Information, and the String Theory Revolution
systems suspended by arbitrarily light threads from some sort of suspension
system built around the black hole at a great distance. The acceleration
of a Fido at proper distance ρ is given by
1
ρ
for ρ << MG. Quantum me
chanically we have a dilemma if we try to imagine the Fidos as real. If they
are extrememly light their locations will necessarily suﬀer large quantum
ﬂuctuations, and they will not be useful as ﬁxed anchors labeling space
time points. If they are massive they will inﬂuence the gravitational ﬁeld
that we wish to describe. Quantum mechanically, physical Fidos must be
replaced by a more abstract concept called gauge ﬁxing. The concept of
gauge ﬁxing in gravitation theory implies a mathematical restriction on the
choice of coordinates. However all real observables are required to be gauge
invariant.
Now let us consider a classical particle falling radially into a black hole.
There are two viewpoints we can adopt toward the description of the par
ticle’s motion. The ﬁrst is the viewpoint of the Fidos who are permanently
stationed outside the black hole. It is a viewpoint which is also useful to
a distant observer, since any observation performed by a Fido can be com
municated to distant observers. According to this viewpoint, the particle
never crosses the horizon but asymptotically approaches it. The second
viewpoint involves freely falling observers (FREFOS) who follow the par
ticle as it falls. According to the Frefos, they and the particle cross the
horizon after a ﬁnite time. However, once the horizon is crossed, their
observations cannot be communicated to any Fido or to a distant observer.
Once the infalling particle is near the horizon its motion can be described
by the coordinates (T, Z, X, Y ) deﬁned in equations 1.3.16 and 1.3.19. Since
the particle is freely falling, in the Minkowski coordinates its motion is a
straight line
dZ
dτ
=
p
Z
m
= −
p
Z
m
dT
dτ
=
p
T
m
(1.7.35)
where p
Z
and p
T
are the Z and T components of momentum, and m is
the mass of the particle. As the particle freely falls past the horizon, the
components p
Z
and p
T
may be regarded as constant or slowly varying.
They are the components seen by Frefos.
The components of momentum seen by Fidos are the components p
ρ
and p
τ
which, using equation 1.3.19, are given by
p
ρ
= p
Z
coshω + p
T
sinhω
p
τ
= p
Z
sinhω + p
T
coshω
(1.7.36)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Schwarzschild Black Hole 23
For large times we ﬁnd
p
ρ
≈ p
τ
≈ 2p
Z
expω = 2p
Z
exp(
t
4MG
) (1.7.37)
Thus we ﬁnd the momentum of an infalling particle as seen by a Fido
grows exponentially with time! It is also easily seen that ρ, the proper
spatial distance of the particle from the horizon, exponentially decreases
with time
ρ(t) ≈ ρ(0)exp(−
t
4MG
) (1.7.38)
Locally the relation between the coordinates of the Frefos and Fidos is
a time dependent boost along the radial direction. The hyperbolic boost
angle is the dimensionless time ω. Eventually, during the lifetime of the
black hole this boost becomes so large that the momentum of an infalling
particle (as seen by a Fido) quickly exceeds the entire mass of the universe.
As a consequence of the boost, the Fidos see all matter undergoing
Lorentz contraction into a system of arbitrarily thin “pancakes” as it ap
proaches the horizon. According to classical physics, the infalling matter is
stored in “sedimentary” layers of diminishing thickness as it eternally sinks
toward the horizon (see Figure 1.13). Quantum mechanically we must ex
pect this picture to break down by the time the infalling particle has been
squeezed to within a Planck distance from the horizon. The Frefos of course
see the matter behaving in a totally unexceptional way.
H
o
r
i
z
o
n
Fig. 1.13 Sedimentary layers of infalling matter on horizon
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
24 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 2
Scalar Wave Equation in a
Schwarzschild Background
In Chapters 3 and 4 we will be concerned with the behavior of quantum
ﬁelds near horizons. In this lecture we will study the properties of a scalar
wave equation in the background of a black hole.
Let us consider a conventional massless free Klein–Gordon ﬁeld χ in the
Schwarzschild background. Here we will ﬁnd great advantage in utilizing
tortoise coordinates in which the metric has the form
dτ
2
= F(r
∗
) [dt
2
− (dr
∗
)
2
] − r
2
[dθ
2
+ sin
2
θdφ
2
] (2.0.1)
The action for χ is
I =
1
2
√
−g g
µν
∂
µ
χ∂
ν
χd
4
x
=
1
2
dt dr
∗
dθ dφ{
(∂
t
χ)
2
−(∂
r
∗ χ)
2
F
−
1
r
2
(
∂χ
∂θ
)
2
−
1
r
2
sin
2
θ
(
∂χ
∂φ
)
2
} F r
2
sinθ
(2.0.2)
Now deﬁne
ψ = r χ (2.0.3)
and the action takes the form
I =
1
2
[(∂
t
ψ)
2
− (
∂ψ
∂r
∗
−
∂(lnr)
∂r
∗
ψ)
2
−
F
r
2
(sinθ
∂ψ
∂θ
2
+
1
sinθ
∂ψ
∂φ
2
)] dt dr
∗
dθ dφ
(2.0.4)
25
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
26 Black Holes, Information, and the String Theory Revolution
which, after an integration by parts and the introduction of spherical har
monic decomposition becomes
I =
¸
m
1
2
[(
˙
ψ
m
)
2
−
∂ψ
m
∂r
∗
2
+
−{
∂lnr
∂r
∗
2
+
∂
∂r
∗
∂lnr
∂r
∗
} ψ
2
m
−
F
r
2
( + 1) ψ
2
m
]dt dr
∗
(2.0.5)
Using the relation between r and r
∗
r
∗
= r + 2MGln(r −2MG)
gives for each , m an action
I
m
=
1
2
dt dr
∗
¸
∂ψ
m
∂t
2
−
∂ψ
m
∂r
∗
2
− V
(r
∗
) ψ
2
m
¸
(2.0.6)
where the potential V
(r
∗
) is given by
V
(r
∗
) =
r −2MG
r
( + 1)
r
2
+
2MG
r
3
(2.0.7)
The equation of motion is
¨
ψ
m
=
∂
2
ψ
m
(∂r
∗
)
2
− V
(r
∗
) ψ
m
(2.0.8)
and for a mode of frequency ν
−
∂
2
ψ
m
(∂r
∗
)
2
+ V
(r
∗
) ψ
m
= ν
2
ψ
m
(2.0.9)
The potential V is shown in Figure 2.1 as a function of the Schwarzschild
coordinate r. For r >> 3MG the potential is repulsive. In fact it is
just the relativistic generalization of the usual repulsive centrifugal barrier.
However as the horizon is approached, gravitational attraction wins and the
potential becomes attractive, and pulls a wave packet toward the horizon.
The maximum of the potential, where the direction of the force changes,
depends weakly on the angular momentum . It is given by
r
max
= 3MG
1
2
1 +
1 +
14
2
+ 14 + 9
9
2
( + 1)
2
−
1
2( + 1)
(2.0.10)
For →∞ the maximum occurs at r
max
( →∞) = 3MG.
The same potential governs the motion of massless classical particles.
One can see that the points r
max
() represent unstable circular orbits, and
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Scalar Wave Equation in a Schwarzschild Background 27
Region of
Thermal
Atmosphere
2MG 3MG 4MG 5MG 6MG
=1
=0
=2
r
V (r)
Fig. 2.1 Eﬀective potential for free scalar ﬁeld vs Schwarzschild radial coordinate
the innermost such orbit is at r = 3MG. Any particle that starts with
vanishing radial velocity in the region r < 3MG will spiral into the horizon.
In the region of large negative r
∗
where we approach the horizon, the
potential is unimportant, and the ﬁeld behaves like a free massless Klein–
Gordon ﬁeld. The eigenmodes in this region have the form of plane waves
which propagate with unit velocity
dr
∗
dt
= ∓1
ψ →e
i k (r
∗
±t)
(2.0.11)
Let us consider a ﬁeld quantum of frequency ν and angular momentum
propagating from large negative r
∗
toward the barrier at r ≈ 3MG. Will it
pass over the barrier? To answer this we note that equation 2.0.9 has the
form of a Schrodinger equation for a particle of energy ν
2
in a potential V .
The particle has enough energy to overcome the barrier without tunneling
if ν
2
is larger than the maximum height of the barrier. For example, if
= 0 the height of the barrier is
V
max
=
1
2M
2
G
2
3
8
3
(2.0.12)
An swave quantum will therefore escape if
ν >
0.15
MG
(2.0.13)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
28 Black Holes, Information, and the String Theory Revolution
Similarly an swave quantum with ν >
0.15
MG
will be able to penetrate the
barrier from the outside and fall to the horizon. Less energetic particles
must tunnel through the barrier.
A particle of high angular momentum, whether on the inside or outside
of the barrier will have more diﬃculty penetrating through. For large
V
max
≈
1
27
2
M
2
G
2
(2.0.14)
Therefore the threshold energy for passing over the barrier is
ν ∼
1
√
27
MG
(2.0.15)
2.1 Near the Horizon
Near the horizon the exterior of the black hole may be described by the
Rindler metric
dτ = ρ
2
dω
2
− dρ
2
− dX
2
− dY
2
It is useful to replace ρ by a tortoiselike coordinate which again goes to
−∞ at the horizon. We deﬁne
u = logρ (2.1.16)
and the metric near the horizon becomes
dτ
2
= exp(2u)
dω
2
− du
2
− dX
2
− dY
2
(2.1.17)
The scalar ﬁeld action becomes
I =
1
2
dX dY du dω
∂χ
∂ω
2
−
∂χ
∂u
2
− e
2u
(∂
⊥
χ)
2
(2.1.18)
where ∂
⊥
χ = (∂
X
, ∂
Y
). Instead of using spherical waves, near the horizon
we can decompose χ into transverse plane waves with transverse wave vector
k
⊥
χ =
d
2
k
⊥
e
i k
⊥
x
⊥
χ(k
⊥
, u, ω) (2.1.19)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Scalar Wave Equation in a Schwarzschild Background 29
the action for a given wave number k is
I =
1
2
dω du
(∂
ω
χ)
2
− (∂
u
χ)
2
− k
2
e
2u
χ
2
(2.1.20)
Thus the potential is
V (k, u) = k
2
e
2u
(2.1.21)
The correspondence between the momentum vector k and the angular
momentum is given by the usual connection between momentum and
angular momentum. If the horizon has circumference 2π(2MG), then
a wave with wave vector k
⊥
will correspond to an angular momentum
 = k r = 2MGk. Thus the potential in equation 2.1.21 is seen to
be proportional to
2
. For very low angular momentum the approximation
is not accurate, but qualitatively is correct for > 0. In approximating a
sum over and m by an integral over k, the integral should be infrared cut
oﬀ at k ∼
1
MG
.
From the action in equation 2.1.20 we obtain the equation of motion
∂
2
χ
∂ω
2
−
∂
2
χ
∂u
2
+ k
2
exp(2u) χ = 0 (2.1.22)
A solution which behaves like exp(iνt) in Schwarzschild time has the
form
e
i ν [4MGω]
= e
i λω
(2.1.23)
The time independent form of the equation of motion is
−
∂
2
χ
∂u
2
+
k
2
exp(2u)
χ = λ
2
χ (2.1.24)
Once again we see that unless k = 0, there is a potential conﬁning quanta to
the region near the horizon. Qualitatively, the behavior of a quantum ﬁeld
in a black hole background diﬀers from the Rindler space approximation
in that for the black hole, the potential barrier is cut oﬀ when ρ = e
u
is greater than MG. By contrast, in the Rindler case V increases as e
u
without bound.
We have not thus far paid attention to the boundary conditions at the
horizon where u → −∞. Since in this region the ﬁeld χ(u) behaves like a
free massless ﬁeld, the boundary condition would be expected to be that
the ﬁeld is in the usual quantum ground state. In the next section we will
see that this is not so.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
30 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 3
Quantum Fields in Rindler Space
According to Einstein, the study of a phenomenon in a gravitational
ﬁeld is best preceeded by a study of the same phenomenon in an accelerated
coordinate system. In that way we can use the special relativistic laws of
nature to understand the eﬀect of a gravitational ﬁeld.
As we have seen, the relativistic analogue of a uniformly accelerated
frame is Rindler space. Because Rindler space covers only a portion of the
spacetime geometry (Region I) there are new and subtle features to the
description of quantum ﬁelds. These features are closely associated with
the existence of the horizon. The method we will use applies to any rela
tivistic quantum ﬁeld theory including those with nontrivial interactions.
For illustrative purposes we will consider a free scalar ﬁeld theory. It is
important to bare in mind that such a noninteracting description is of lim
ited validity. As we shall see, interactions become very important near the
horizon of a black hole. Ignoring them leads to an inconsistent description
of the Hawking evaporation process.
3.1 Classical Fields
First let us consider the evolution of a classical ﬁeld in Rindler space.
The ﬁeld in Region I of Figure 1.2 can be described in a self contained way.
Obviously inﬂuences from Regions II and III can never be felt in Region I
since no point in Regions II or III is in the causal past of any point in
Region I. Signals from Region IV can, of course, reach Region I, but to do
so they must pass through the surface ω = −∞. Therefore signals from
Region IV are regarded as initial data in the remote past by the Rindler
observer. Evidently the Rindler observer sees a world in which physical
phenomena can be described in a completely self contained way.
31
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
32 Black Holes, Information, and the String Theory Revolution
The evolution from one surface of constant ω to another is governed
by the Rindler Hamiltonian. Using conventional methods the generator of
ωtranslations is given by
H
R
=
∞
ρ=0
dρ dX dY ρ T
00
(ρ, X, Y ) (3.1.1)
where T
00
is the usual Hamiltonian density used by the Minkowski observer.
For example, for a massive scalar ﬁeld with potential V , T
00
is given by
T
00
=
Π
2
2
+
1
2
(∇χ)
2
+ V (χ) (3.1.2)
where Π is the canonical momentum conjugate to χ. The Rindler Hamil
tonian is
H
R
=
dρ dx
⊥
ρ
2
Π
2
+
∂χ
∂ρ
2
+
∂χ
∂x
⊥
2
+ 2 V (χ)
(3.1.3)
The origin of the factor ρ in the Rindler Hamiltonian density is straight
forward. In Figure 3.1 the relation between neighboring equal Rindlertime
surfaces is shown. The proper time separation between the surfaces is
δτ = ρ δω (3.1.4)
Thus, to push the ωsurface ahead requires a ρdependent time translation.
This is the reason that T
00
is weighted with the factor ρ. The Rindler
Hamiltonian is similar to the generator of Lorentz boosts from the viewpoint
of the Minkowski observer. However it only involves the degrees of freedom
in Region I.
3.2 Entanglement
Quantum ﬁelds can also be described in a selfcontained fashion in
Rindler space, but a new twist is encountered. Our goal is to describe
the usual physics of a quantum ﬁeld in Minkowski space, but from the
viewpoint of the Fidos in Region I, i.e. in Rindler space. To understand
the new feature, recall that in the usual vacuum state, the correlation be
tween ﬁelds at diﬀerent spatial points does not vanish. For example, in free
massless scalar theory the equal time correlator is given by
0 χ(X, Y, Z) χ(X
, Y
, Z
) 0 ∼
1
∆
2
(3.2.5)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Quantum Fields in Rindler Space 33
ρ=0
(Horizon)
Light
cone
T
p
=
1
p
=
2
Z
ω=0
ω=−3
ω=−4
ω=+4
ω=+3
ω=+2
ω=+1
ω=−2
ω=−1
r > 2MG
ω
=
8
,
t
=
8
r < 2MG
Fig. 3.1 Equal time and proper distance surfaces in Rindler space
where ∆ is the spacelike separation between the points (X, Y, Z) and
(X
, Y
, Z
)
∆
2
= (X −X
)
2
+ (Y −Y
)
2
+ (Z −Z
)
2
(3.2.6)
The two points might both lie within Region I, in which case the correla
tor in equation 3.2.5 represents the quantum correlation seen by Fido’s in
Region I. On the other hand, the two points might lie on opposite sides
of the horizon at Z = 0. In that case the correlation is unmeasurable to
the Fidos in Region I. Nevertheless it has signiﬁcance. When two sub
systems (ﬁelds in Regions I and III) become correlated, we say that they
are quantum entangled, so that neither can be described in terms of pure
states. The appropriate description of an entangled subsystem is in terms
of a density matrix.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
34 Black Holes, Information, and the String Theory Revolution
3.3 Review of the Density Matrix
Suppose a system consists of two subsystems, A and B, which have
previously been in contact but are no longer interacting. The combined
system has a wavefunction
Ψ = Ψ(α, β) (3.3.7)
where α and β are appropriate commuting variables for the subsystems A
and B.
Now suppose we are only interested in subsystem A. A complete descrip
tion of all measurements of A is provided by the density matrix ρ
A
(α, α
).
ρ
A
(α, α
) =
β
Ψ
∗
(α, β) Ψ(α
, β) (3.3.8)
Similarly, experiments performed on B are described by ρ
B
(β, β
).
ρ
B
(β, β
) =
α
Ψ
∗
(α, β) Ψ(α, β
) (3.3.9)
The rule for computing an expectation value of an operator a composed
of A degrees of freedom is
a = Tr aρ
A
(3.3.10)
Density matrices have the following properties:
1) Tr ρ = 1 (total probability=1)
2) ρ = ρ
†
(hermiticity)
3) ρ
j
≥ 0 (all eigenvalues are positive or zero)
In the representation in which ρ is diagonal
ρ =
ρ
1
0 0 ... ...
0 ρ
2
0 ... ...
0 0 ρ
3
... ...
... ... ... ... ...
(3.3.11)
The eigenvalues ρ
j
can be considered to be probabilities that the system
is in the j
th
state. However, unlike the case of a coherent superposition of
states, the relative phases between the states j are random.
There is one special case when the density matrix is indistinguishable
from a pure state. This is the case in which only one eigenvalue ρ
j
is
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Quantum Fields in Rindler Space 35
nonzero. This case can only result from an uncorrelated product wave
function of the form
Ψ(α, β) = ψ
A
(α) ψ
B
(β) (3.3.12)
A quantitative measure of the departure from a pure state is provided
by the Von Neumann entropy
S = −Tr ρ log ρ = −
j
ρ
j
log ρ
j
. (3.3.13)
S is zero if and only if all the eigenvalues but one are zero. The one non
vanishing eigenvalue is equal to 1 by virtue of the trace condition on ρ. The
entropy is also a measure of the degree of entanglement between A and B.
It is therefore called the entropy of entanglement.
The opposite extreme to a pure state is a completely incoherent den
sity matrix in which all the eigenvalues are equal to
1
N
, where N is the
dimensionality of the Hilbert space. In that case S takes its maximum
value
S
max
= −
j
1
N
log
1
N
= log N (3.3.14)
More generally, if ρ is a projection operator onto a subspace of dimension
n, we ﬁnd
S = log n (3.3.15)
Thus we see that the Von Neumann entropy is a measure of the number
of states which have an appreciable probability in the statistical ensemble.
We may think of e
S
as an eﬀective dimensionality of the subspace described
by ρ.
The Von Neumann (or entanglement) entropy should not be confused
with the thermal entropy of the second law of thermodynamics. This en
tropy has its origin in coarse graining. If a system with Hamiltonian H is
in thermal equilibrium at temperature T = 1/β then it is described by a
Maxwell–Boltzman density matrix
ρ
M.B.
=
e
−β H
Tre
−β H
. (3.3.16)
In this case the thermal entropy is given by
S
thermal
= −Tr ρ
M.B.
logρ
M.B.
(3.3.17)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
36 Black Holes, Information, and the String Theory Revolution
3.4 The Unruh Density Matrix
Now let us consider the space of states describing a Lorentz invariant
quantum ﬁeld theory in Minkowski space. In Figure 3.2 the surface T = 0
of Minkowski space is shown divided into two halves, one in Region I and
one in Region III. For the case of a scalar ﬁeld χ the ﬁelds at each point
R
χ
I
χ
L
III
Fig. 3.2 Fields on the spacelike surface T = 0 in Minkowski space
of space form a complete set of commuting operators. These ﬁelds may be
decomposed into two subsets associated with regions I and III. we call them
χ
R
and χ
L
respectively. Thus
χ(X, Y, Z) = χ
R
(X, Y, Z) Z > 0
χ(X, Y, Z) = χ
L
(X, Y, Z) Z < 0
(3.4.18)
The general wave functional of the system is a functional of χ
L
and χ
R
Ψ = Ψ(χ
L
, χ
R
) (3.4.19)
We wish to compute the density matrix used by the Fidos in Region I to
describe their Rindler world. In particular we would like to understand the
density matrix ρ
R
which represents the usual Minkowski vacuum to the
Fidos in Region I.
First let us see what we can learn from general principles. Obviously the
state Ψ is translationally invariant under the usual Minkowski space trans
lations. Thus the Fidos must see the vacuum as invariant under translations
along the X and Y axes. However, the translation invariance along the Z
axis is explicitly broken by the act of singling out the origin Z = 0 for
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Quantum Fields in Rindler Space 37
special consideration. From the X, Y translation invariance we conclude
that ρ
R
commutes with the components of momentum in these directions
[p
X
, ρ
R
] = [p
Y
, ρ
R
] = 0 (3.4.20)
A very important property of ρ
R
is that it is invariant under Rindler
time translations ω → ω + constant. This follows from the Lorentz boost
invariance of Ψ. Thus
[H
R
, ρ
R
] = 0 (3.4.21)
To proceed further we must use the fact that Ψ(χ
L
, χ
R
) is the ground
state of the Minkowski Hamiltonian. General path integral methods may
be brought to bare on the computation of the ground state wave functional.
Let us assume that the ﬁeld theory is described in terms of an action
I =
d
3
X dT L (3.4.22)
The socalled Euclidean ﬁeld theory is deﬁned by replacing the time coor
dinate T by i X
0
. For example, the Euclidean version of ordinary scalar
ﬁeld theory is obtained from the usual Minkowski action
I =
d
3
X dT
1
2
˙ χ
2
− (∇χ)
2
− 2 V (χ)
(3.4.23)
Letting T →i X
0
we obtain the Euclidean action
I
E
=
d
4
X
1
2
(∂
X
χ)
2
− 2 V (χ)
(3.4.24)
Now a standard method of computing the ground state by path integra
tion is to use the Feynman–Hellman theorem. Suppose we wish to compute
Ψ(χ
L
, χ
R
). Then we consider the path integral
Ψ(χ
L
, χ
R
) =
1
√
Z
dχ(x) e
−I
E
(3.4.25)
where the path integral is over all χ(x) with X
0
> 0 and Z is an appropriate
normalization factor . The ﬁeld χ(x) is constrained to equal (χ
L
, χ
R
) on
the surface X
0
= 0. Finally the action I
E
is evaluated as an integral over
the portion of Minkowski space with X
0
> 0.
The boost invariance of the original Minkowski action insures that the
Euclidean action has four dimensional rotation invariance. In particular,
the invariance under ωtranslations becomes invariance under rotations in
the Euclidean (Z, X
0
) plane. This suggests a new way to carry out the
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
38 Black Holes, Information, and the String Theory Revolution
path integral. Let us deﬁne the Euclidean angle in the (Z, X
0
) plane to be
θ. The angle θ is the Euclidean analogue of the Rindler time ω. Now let
us divide the region X
0
> 0 into inﬁnitesimal angular wedges as shown in
Figure 3.3.
T
Z
θ=0
Horizon
ρ=0
χ
L
χ
R
δθ
Fig. 3.3 Euclidean analogue of Rindler space for path integration
The strategy for computing the path integral is to integrate over the
ﬁelds in the ﬁrst wedge between θ = 0 and θ = δθ. The process can be
iterated until the entire region X
0
> 0 has been covered.
The integral over the ﬁrst wedge is deﬁned by constaining the ﬁelds at
θ = 0 and θ = δθ. This deﬁnes a transfer matrix G in the Hilbert space of
the ﬁeld conﬁguration χ
R
. The matrix is recognized to be
G = (1 − δθ H
R
) . (3.4.26)
To compute the full path integral we raise the matrix G to the power
π
δω
giving
Ψ(χ
L
, χ
R
) =
1
√
Z
χ
L
 e
−π H
R
χ
R
(3.4.27)
In other words, the path integral deﬁning Ψ is computed as a transition ma
trix element between initial state χ
R
and ﬁnal state χ
L
. The inﬁnitesimal
generator which pushes θ surfaces forward is just the Rindler Hamiltonian.
Now we are prepared to compute the density matrix ρ
R
. According to
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Quantum Fields in Rindler Space 39
the deﬁnition in equation 3.3.8, the density matrix ρ
R
is given by
ρ
R
(χ
R
, χ
R
) =
Ψ
∗
(χ
L
, χ
R
) Ψ(χ
L
, χ
R
) dχ
L
(3.4.28)
Now using equation 3.4.27 we get
ρ
R
(χ
R
, χ
R
) =
1
Z
χ
R
 e
−π H
R
χ
L
χ
L
 e
−π H
R
χ
R
dχ
L
=
1
Z
χ
R
 e
−2 π H
R
χ
R
(3.4.29)
In other words, the density matrix is given by the operator
ρ
R
=
1
Z
exp(−2 π H
R
) (3.4.30)
This remarkable result, discovered by William Unruh in 1976 , says that
the Fidos see the vacuum as a thermal ensemble with a density matrix of
the Maxwell–Boltzmann type. The temperature of the ensemble is
T
R
=
1
2π
=
1
β
R
(3.4.31)
The derivation of the thermal character of the density matrix and the
value of the Rindler temperature in equation 3.4.31 is entirely independent
of the particulars of the relativistic ﬁeld theory. It is equally correct for a
free scalar quantum ﬁeld, quantum electrodynamics, or quantum chromo
dynamics.
3.5 Proper Temperature
It is noteworthy that the temperature T
R
is dimensionless. Ordinarily,
temperature has units of energy, or equivalently, inverse length. The origin
of the dimensionless temperature lies in the dimensionless character of the
Rindler time variable ω. Nevertheless we should be able to assign to each
Fido a conventional temperature that would be recorded by a standard
thermometer held at rest at the location of that Fido. We can consider
a thermometer to be a localized object with a set of proper energy levels
i
. The levels
i
are the ordinary energy levels of the thermometer when it
is at rest. The thermometer is assumed to be very weakly coupled to the
quantum ﬁelds so that it eventually will come to thermal equilibrium with
them. Let us suppose that the thermometer is at rest with respect to the
Fido at position ρ so that it has proper acceleration
1
ρ
. The Rindler energy
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
40 Black Holes, Information, and the String Theory Revolution
in equation 3.1.1 evidently receives a contribution from the thermometer of
the form
H
R
(thermometer) =
i
ρ i i
i
(3.5.32)
In other words the Rindler energy level of the i
th
state of the thermometer
is ρ
i
.
When the quantum ﬁeld at Rindler temperature
1
2π
equilibrates with
the thermometer, the probability to ﬁnd the thermometer excited to the
i
th
level is given by the Boltzmann factor
P
i
=
e
−2π ρ
i
j
e
−2π ρ
j
(3.5.33)
Accordingly, the thermometer registers a proper temperature
T(ρ) =
1
2π ρ
=
1
ρ
T
R
(3.5.34)
Thus each Fido experiences a thermal environment characterized by
a temperature which increases as we move toward the horizon at ρ = 0.
The proper temperature T(ρ) can also be expressed in terms of the proper
acceleration of the Fido which is equal to
1
ρ
. Thus, calling the acceleration
a, we ﬁnd
T(ρ) =
a(ρ)
2π
(3.5.35)
The reader may wonder about the origin of the thermal ﬂuctuations felt
by the Fidos, since the system under investigation is the Minkowski space
vacuum. The thermal ﬂuctuations are nothing but the conventional virtual
vacuum ﬂuctuations, but now being experienced by accelerated appara
tuses. It is helpful in visualizing these eﬀects to describe virtual vacuum
ﬂuctuations as short lived particle pairs. In Figure 3.4 ordinary vacuum
ﬂuctuations are shown superimposed on a Rindler coordinate mesh. One
virtual loop (a) is contained entirely in Region I. That ﬂuctuation can be
thought of as a conventional ﬂuctuation described by the quantum Hamil
tonian H
R
. The ﬂuctuation (b) contained in Region III has no signiﬁcance
to the Fidos in Region I. Finally there are loops like (c) which are partly
in Region I but which also enter into Region III. These are the ﬂuctuations
which lead to nontrivial entanglements between the degrees of freedom χ
L
and χ
R
, and which cause the density matrix of Region I to be a mixed
state.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Quantum Fields in Rindler Space 41
(b)
(c)
(a)
ω=
8
8
ω=−
Fig. 3.4 Vacuum pair ﬂuctuations near the horizon
A virtual ﬂuctuation is usually considered to be short lived because it
“violates energy conservation”. If the virtual ﬂuctuation of energy needed
to produce the pair is E, then the lifetime of the ﬂuctuation ∼ E
−1
.
Now consider the portion of the loop (b) which is found in Region I.
From the viewpoint of the Fidos, a particle is injected into the system at
ω = −∞ and ρ = 0. The particle travels to some distance and then falls
back towards ρ = 0 and ω = +∞. Thus, according to the Fidos, the
ﬂuctuation lasts for an inﬁnite time and is therefore not virtual at all. Real
particles are seen being injected into the Rindler space from the horizon,
and eventually fall back to the horizon. To state it diﬀerently, the horizon
behaves like a hot membrane radiating and reabsorbing thermal energy.
A natural question to ask is whether the thermal eﬀects are “real”.
For example, we may ask whether any such thermal eﬀects are seen by
freely falling observers carrying their thermometers with them as they pass
through the horizon. Obviously the answer is no. A thermometer at rest
in an inertial frame in the Minkowski vacuum will record zero temperature.
It is tempting to declare that the thermal eﬀects seen by the Fidos are
ﬁctitious and that the reality is best described in the frame of the Frefos.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
42 Black Holes, Information, and the String Theory Revolution
However, by yielding to this temptation we risk prejudicing ourselves too
much toward the viewpoint of the Frefos. In particular, we are going to
encounter questions of the utmost subtlety concerning the proper relation
between events as seen by observers who fall through the horizon of a black
hole and those seen by observers who view the formation and evaporation
process from a distance. Thus for the moment, it is best to avoid the
metaphysical question of whose description is closer to reality. Instead we
simply observe that the phenomena are described diﬀerently in two diﬀer
ent coordinate systems and that diﬀerent physical eﬀects are experienced
by Frefos and Fidos. In particular a Fido equipped with a standard ther
mometer, particle detector, or other apparatus, will discover all the physical
phenomena associated with a local proper temperature T(ρ) =
1
2π ρ
. By
contrast, a Frefo carrying similar apparatuses will see only the zero tem
perature vacuum state. Later we will discuss the very interesting question
of how contradiction is avoided if a Frefo attempts to communicate to the
stationary Fidos the information that no thermal eﬀects are present.
We can now state the sense in which a self contained description of
Rindler space is possible in ordinary quantum ﬁeld theory. Since Rindler
space has a boundary at ρ = 0, a boundary condition of some sort must
be provided. We see that the correct condition must be that at some
small distance ρ
o
, an eﬀective “membrane” is kept at a ﬁxed temperature
T(ρ
o
) =
1
2π ρ
o
by an inﬁnite heat reservoir. It will prove useful later to
locate the membrane at a distance of order the Planck length
P
=
G/c
3
where quantum gravitational or string eﬀects become important. Such a
ﬁctitious membrane at Planckian distance from the horizon is called the
stretched horizon. We will see later that the stretched horizon has many
other physical properties besides temperature, although it is completely
unseen by observers who fall freely through it.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 4
Entropy of the Free Quantum Field in
Rindler Space
In the real world, a wide variety of diﬀerent phenomena take place at
diﬀerent temperature scales. At the lowest temperatures where only mass
less quanta are produced by thermal ﬂuctuations, one expects to ﬁnd a very
weakly interacting gas of gravitons, photons, and neutrinos. Increase the
temperature to the e
+
, e
−
threshold and electronpositron pairs are pro
duced. The free gas is replaced by a plasma. At higher temperatures, pions
are produced which eventually dissociate into quarks and gluons, and so it
goes, up the scale of energies. Finally, the Planck temperature is reached
where totally new phenomena of an as yet unimagined kind take place.
All of these phenomena have their place in the Fido’s description of the
region near a horizon. In this lecture we will consider an enormously over
simpliﬁed description of the world in which only a single free ﬁeld is present
in a ﬁxed spacetime background. There is serious danger in extrapolating
far reaching conclusions from so oversimpliﬁed a situation. In fact, the para
doxes and contradictions associated with black holes, quantum mechanics,
and statistical thermodynamics that these lectures are concerned with are
largely a consequence of such unjustiﬁed extrapolation. Nevertheless, the
study of a free quantum ﬁeld in Rindler space is a useful starting point.
We consider the ﬁeld theory deﬁned by equation 2.1.18. Fourier decom
posing the ﬁeld χ in equation 2.1.19 leads to the wave equation in equation
2.1.24.
−
∂
2
χ
k
∂u
2
+
k
2
e
2u
χ
k
= λ
2
χ
k
(4.0.1)
In order to quantize the ﬁeld χ it is necessary to provide a boundary condi
tion when u →−∞. The simplest method of dealing with this region is to
introduce a cutoﬀ at some point u
o
= log at which point the ﬁeld (or its
ﬁrst derivative) is made to vanish. The parameter represents the proper
43
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
44 Black Holes, Information, and the String Theory Revolution
distance of the cutoﬀ point to the horizon. Physically we are introducing
a perfectly reﬂecting mirror just outside the horizon at a distance . Later
we will remove the cutoﬀ by allowing u
o
→−∞.
It is by no means obvious that a reﬂecting boundary condition very near
the horizon is a physically reasonable way to regularize the theory. However
it will prove interesting to separate physical quantities into those which are
sensitive to and those which are not. Those things which depend on are
sensitive to the behavior of the physical theory at temperatures of order
1
2π
and greater.
Each transverse Fourier mode χ
k
can be thought of as a free 1+1 di
mensional quantum ﬁeld conﬁned to a box. One end of the box is at the
reﬂecting boundary at u = u
o
= log . The other wall of the box is provided
by the repulsive potential
V (u) = k
2
exp(2u)
which becomes large when u > −log k. Thus we may approximate the
potential by a second wall at u = u
1
= −log k. The total length of the box
depends on k and according to
L(k) = −log( k) (4.0.2)
For each value of k the ﬁeld χ
k
can be expanded in mode functions and
creation and annihilation operators according to
χ
k
(u) =
n
a
+
(n, k) f
n,k
(u) + a
−
(n, k) f
∗
n,k
(u)
(4.0.3)
where the mode (n, k) has frequency λ(n, k). The Rindler Hamiltonian is
given by
H
R
=
d
2
k
n
λ(n, k) a
†
(n, k) a(n, k)
=
n
λ(n, k) N(n, k)
(4.0.4)
where
N(n, k) = a
†
(n, k) a(n, k) (4.0.5)
Thus far the quantization rules are quite conventional. The new and
unusual feature of Rindler quantization, encountered in Chapter 3, is that
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy of the Free Quantum Field in Rindler Space 45
we do not identiﬁy the vacuum with the state annihilated by the a(n, k),
but rather with the thermal density matrix
ρ
R
=
n,k
ρ
R
(n, k) (4.0.6)
with
ρ
R
(n, k) ∼ exp
−2πλ(n, k) a
†
(n, k) a(n, k)
(4.0.7)
Thus the average occupation number of each mode is
N(n, k) =
1
exp[2πλ(n, k)] − 1
(4.0.8)
These particles constitute the thermal atmosphere.
The reader might wonder what goes wrong if we choose the state which
is annihilated by the a’s. Such a state is not at all invariant under trans
lations of the original Minkowski coordinates Z and T. In fact, a careful
computation of the expectation value of T
µν
in this state reveals a singular
behavior at the horizon. Certainly this is not a good candidate to represent
the original Minkowski vacuum.
A black hole, on the other hand, is not a translationally invariant system.
One might therefore suppose that the evolution of the horizon might lead
to the Fock space vacuum with no quanta rather than the thermal state.
This however would clearly violate the fourth guiding principle stated in
the introduction: To a freely falling observer, the horizon of a black hole
should in no way appear special. Moreover, the large back reaction on the
gravitational ﬁeld that would result from the divergent expectation value
of T
µν
makes it unlikely that this state can exist altogether.
Physical quantities in Rindler space can be divided into those which
are sensitive to the cutoﬀ at and those which are not. As an example of
insensitive quantities, the ﬁeld correlation functions such as
χ(X, Y, u) χ(X
, Y
, u
) = Tr ρ χ(X, Y, u) χ(X
, Y
, u
), (4.0.9)
are found to have smooth limits as → 0, as long as the points (X, Y, u)
and (X
, Y
, u
) are kept away from the horizon. Therefore such quantities
can be said to decouple from the degrees of freedom within a distance of
the horizon. A much more singular quantity which will be of great concern
in future lectures is the entropy of the vacuum state.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
46 Black Holes, Information, and the String Theory Revolution
Since the relevant density matrix has the Maxwell–Boltzmann form, we
can use equations 3.3.16 and 3.3.17 to obtain the entropy. Deﬁning
Tr e
−βH
= Z(β) (4.0.10)
and using the identity
ρ log ρ =
∂
∂N
ρ
N
N=1
(4.0.11)
we obtain
S = −Tr
∂
∂N
e
−NβH
Z(β)
N
N=1
= +Tr βH
e
−βH
Z
+ lnZ
= βH + lnZ
(4.0.12)
Deﬁning E = H and F = −
1
β
logZ we ﬁnd the usual thermodynamic
identity
S = β(E −F) (4.0.13)
Another identity follows from using E = −
∂logZ
∂β
where we ﬁnd
S = −β
2
∂(logZ/β)
∂β
(4.0.14)
The entropy S in equations 4.0.13 and 4.0.14 can be thought of as both
entanglement and thermal entropy in the special case of the Rindler space
density matrix. This is because the eﬀect of integrating over the ﬁelds χ
L
in equation 3.4.28 is to produce the thermal density matrix in equation
3.4.30. Thus the computation of the entropy of Rindler space is reduced to
ordinary thermodynamic methods. For the present case of free ﬁelds the
entropy is additive over the modes and can be estimated from the formula
for the thermodynamics of a free 1+1 dimensional scalar ﬁeld.
To compute the total entropy we begin by replacing the inﬁnite trans
verse X, Y plane by a ﬁnite torus with periodic boundary conditions. This
has the eﬀect of discretizing the values of k. Thus
k
X
=
2n
X
π
B
, k
Y
=
2n
Y
π
B
(4.0.15)
where B is the size of the torus.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy of the Free Quantum Field in Rindler Space 47
The entropy stored in the ﬁeld χ
k
can be estimated from the entropy
density of a 1+1 dimensional massless free boson at temperature T. A
standard calculation gives the entropy density
S
L
to be given by
S
L
=
π
3
T (4.0.16)
where T is the temperature. Substituting T =
1
2π
and equation 4.0.2 for
the length L gives the entropy of χ
k
S(k) =
1
6
log k  (4.0.17)
To sum over the values of k we use equation 4.0.15 and let B →∞
S
Total
=
B
2
24π
2
d
2
k log k  (4.0.18)
In evaluating equation 4.0.18, the integral must be cut oﬀ when k >
1
.
This is because when k =
1
the potential is already large at u = u
o
so that
the entire contribution of χ
k
is supressed. We ﬁnd that S is approximately
given by
S
Total
≈
1
96π
2
B
2
2
(4.0.19)
From equation 4.0.19 we see two important features of the entropy of
Rindler space. The ﬁrst is that it is proportional to the transverse area
of the horizon, B
2
. One might have expected it to diverge as the volume
of space, but this is not the case. The entropy is stored in the vicinity of
the stretched horizon and therefore grows only like the area. The second
feature which should alarm us is that the entropy per unit area diverges like
1
2
. As we shall see, the entropy density of the horizon is a physical quantity
whose exact value is known. Nevertheless the divergence in S indicates that
its value is sensitive to the ultraviolet physics at very small length scales.
Further insight into the form of the entropy can be gained by recalling
that the proper temperature T(ρ) is given by T(ρ) =
1
2πρ
. Furthermore
the entropy density of a 3+1 dimensional free scalar ﬁeld is given by
S(T) = V
2
π
2
ζ(4)k
B
k
B
T
c
3
= V
2π
2
45
T
3
(4.0.20)
Now consider the entropy stored in a layer of thickness δρ and area B
2
at
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
48 Black Holes, Information, and the String Theory Revolution
a distance ρ from the horizon
δS(ρ) =
2π
2
45
T
3
(ρ) δρ B
2
=
2π
2
45
1
(2πρ)
3
δρ B
2
(4.0.21)
To ﬁnd the full entropy we integrate with respect to ρ
S =
B
2
(2π)
3
2π
2
45
∞
dρ
ρ
3
=
B
2
(2π)
3
2π
2
45
2
2
(4.0.22)
Now we see that the entropy is mainly found near the horizon because that
is where the temperature gets large.
4.1 Black Hole Evaporation
The discovery of a temperature seen by an accelerated ﬁducial observer
adds a new dimension to the equivalence principle. We can expect that
identical thermal eﬀects will occur near the horizon of a very massive black
hole. However, in the case of a black hole a new phenomenon can take
place – evaporation. Unlike the Rindler case, the thermal atmosphere is
not absolutely conﬁned by the centrifugal potential in equation 2.0.7. The
particles of the thermal atmosphere will gradually leak through the barrier
and carry oﬀ energy in the form of thermal radiation. A good qualitative
understanding of the process can be obtained from the Rindler quantum
ﬁeld theory in equation 2.1.22 by observing two facts:
1) The Rindler time ω is related to the Schwarzschild time t by the equation
ω =
t
4MG
(4.1.23)
Thus a ﬁeld quantum with Rindler frequency ν
R
is seen by the distant
Schwarzschild observer to have a red shifted frequency ν
ν =
ν
R
4MG
(4.1.24)
The implication of this fact is that the temperature of the thermal at
mosphere is reckoned to be red shifted also. Thus the temperature as
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy of the Free Quantum Field in Rindler Space 49
seen by the distant observer is
T =
1
2π
×
1
4MG
=
1
8πMG
, (4.1.25)
a form ﬁrst calculated by Stephen Hawking.
2) The centrifugal barrier which is described in the Rindler theory by the
potential k
2
exp(2u) is modiﬁed at distances r ≈ 3MG as in Figure 2.1.
In particular the maximum value that V takes on for angular momentum
zero is
V
max
( = 0) =
27
1024
1
M
2
G
2
(4.1.26)
Any swave quanta with frequencies of order (V
max
)
1/2
=
3
√
3
32MG
or
greater will easily escape the barrier. Since the average energies of
massless particles in thermal equilibrium at temperature T is of course
of order T, equation 4.1.25 indicates that some of the swave particles
will easily escape to inﬁnity. Unless the black hole is kept in equilibrium
by incoming radiation it will lose energy to its surroundings.
Particles of angular momenta higher than swaves cannot easily escape
because the potential barrier is higher than the thermal scale. The black
hole is like a slightly leaky cavity containing thermal radiation. Most quanta
in the thermal atmosphere have high angular momenta and reﬂect oﬀ the
walls of the cavity. A small fraction of the particles carry very low angular
momenta. For these particles, the walls are semitransparent and the cavity
slowly radiates its energy. This is the process ﬁrst discovered by Hawking
and is referred to as Hawking radiation.
The above description of Hawking radiation does not depend in any
essential way on the free ﬁeld approximation. Indeed it only makes sense
if there are interactions of suﬃcient strength to keep the system in equi
librium during the course of the evaporation. In fact, most discussions of
Hawking radiation rely in an essential way on the free ﬁeld approximation,
and ultimately lead to absurd results. At the end of the next lecture, we
will discuss one such absurdity.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
50 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 5
Thermodynamics of Black Holes
We have seen that a large black hole appears to a distant observer as a
body with temperature
T =
1
8πMG
(5.0.1)
and energy M. It follows thermodynamically that it must also have an
entropy. To ﬁnd the entropy we use the ﬁrst law of thermodynamics in the
form
dE = T dS (5.0.2)
where E, the black hole energy, is replaced by M. Using equation 5.0.1
dM =
1
8πMG
dS
from which we deduce
S = 4πM
2
G (5.0.3)
The Schwarzschild radius of the black hole is 2MG and the area of the
horizon is 4π(4M
2
G
2
) so that
S
BH
=
Area
4G
(5.0.4)
This is the famous Bekenstein–Hawking entropy. It is gratifying that it is
proportional to the area of the horizon. This, as we have seen, is where all
the infalling matter accumulates according to external observers. We have
seen in Chapter 4 equation 4.0.19 that the matter ﬁelds in the vicinity of
the horizon give rise to an entropy. Presumably this entropy is part of the
entropy of the black hole, but unfortunately it is inﬁnite as →0. Evidently
something cuts oﬀ the modes which are very close to the horizon. To get an
51
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
52 Black Holes, Information, and the String Theory Revolution
idea of where the cut oﬀ must occur, we can require that the contribution
in equation 4.0.19 not exceed the entropy of the black hole S
BH
1
96π
2
2
<
∼
1
4G
(5.0.5)
or
>
∼
√
G
15
(5.0.6)
In other words, the cutoﬀ must not be much smaller than the Planck length,
where the Planck length is given in terms of Newton’s constant as
P
=
c
3
G. This is of course not surprising. It is widely believed that the nasty
divergences of quantum gravity will somehow be cut oﬀ by some mechanism
when the distance scales become smaller than
√
G.
What is the real meaning of the black hole entropy? According to the
principles stated in the introduction to these lectures, the entropy reﬂects
the number of microscopically distinct quantum states that are “coarse
grained” into the single macroscopic state that we recognize as a black
hole. The number of such states is of order exp S
B.H.
= exp
4πM
2
G
.
Another way to express this is through the level density of the black hole
dN
dM
∼ exp
4πM
2
G
(5.0.7)
where dN is the number of distinct quantum states with mass M in the
interval dM.
The entropy of a large black hole is an extensive quantity in the sense
that it is proportional to the horizon area. This suggests that we can
understand the entropy in terms of the local properties of a limiting black
hole of inﬁnite mass and area. The entropy diverges, but the entropy per
unit area is ﬁnite. The local geometry of a limiting black hole horizon is of
course Rindler space.
Let us consider the Rindler energy of the horizon. By deﬁnition it is
conjugate to the Rindler time ω. Accordingly we write
[E
R
(M), ω] = i (5.0.8)
Here E
R
is the Rindler energy which is of course the eigenvalue of the
Rindler Hamiltonian. We assume that for a large black hole the Rindler
energy is a function of the mass of the black hole.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Thermodynamics of Black Holes 53
The mass and Schwarzschild time are also conjugate
[M, t] = i (5.0.9)
Now use ω =
t
4MG
to obtain
E
R
(M),
t
4MG
= i
or
[E
R
(M), t] = 4MGi (5.0.10)
Finally, the conjugate character of M and t allows us to write equation
5.0.10 in the form
∂E
R
∂M
= 4MG (5.0.11)
and
E
R
= 2M
2
G (5.0.12)
The Rindler energy and the Schwarzschild mass are both just the energy
of the black hole. The Schwarzschild mass is the energy as reckoned by
observers at inﬁnity using tclocks, while the Rindler energy is the (dimen
sionless) energy as deﬁned by observers near the horizon using ωclocks. It
is of interest that the Rindler energy is also extensive. The area density of
Rindler energy is
E
R
A
=
1
8πG
(5.0.13)
The Rindler energy and entropy satisfy the ﬁrst law of thermodynamics
dE
R
=
1
2π
dS (5.0.14)
where
1
2π
is the Rindler temperature. Thus we see the remarkable fact
that horizons have universal local properties that behave as if a thermal
membrane or stretched horizon with real physical properties were present.
As we have seen, the stretched horizon also radiates like a black body.
The exact rate of evaporation of the black hole is sensitive to many
details, but it can easily be estimated. We ﬁrst recall that only the very low
angular momentum quanta can escape the barrier. For simplicity, suppose
that only the swave quanta get out. The swave quanta are described
in terms of a 1+1 dimensional quantum ﬁeld at Rindler temperature
1
2π
.
In the same units, the barrier height for the swave quanta is comparable
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
54 Black Holes, Information, and the String Theory Revolution
to the temperature. It follows that approximately one quantum per unit
Rindler time will excape. In terms of the Schwarzschild time, the ﬂux of
quanta is of order
1
MG
. Furthermore each quantum carries an energy at
inﬁnity of order the Schwarzschild temperature
1
8πMG
. The resulting rate
of energy loss is of order
1
M
2
G
2
. We call this L, the luminosity. Evidently
energy conservation requires the black hole to lose mass at just this rate
dM
dt
= −L = −
C
M
2
G
2
(5.0.15)
where C is a constant of order unity. The constant C depends on details
such as the number of species of particles that can be treated as light enough
to be thermally produced. It is therefore not really constant. When the
mass of the black hole is large and the temperature low, only a few species
of massless particles contribute and C is constant.
If we ignore the mass dependence of C, equation 5.0.15 can be integrated
to ﬁnd the time that the black hole survives before evaporating to zero mass.
This evaporation time is evidently of order
t
evaporation
∼ M
3
G
2
(5.0.16)
It is interesting that luminosity in equation 5.0.15 is essentially the
Stephan–Boltzmann law
L ∼ T
4
· Area (5.0.17)
Using T ∼
1
MG
and Area ∼ M
2
G
2
in equation 5.0.15 gives equation 5.0.17.
However the physics is very diﬀerent from that of a radiating star. In
that case the temperature and size of the system are related in an entirely
diﬀerent way. The typical wavelength of a photon radiated from the sun is
∼ 10
−5
cm, while the radius of the surface of the sun is ∼ 10
11
cm. The sun
is for all intents and purposes inﬁnite on the scale of the emitted photon
wavelengths. The black hole on the other hand emits quanta of wavelength
∼
1
T
∼ MG, which is about equal to the Schwarzschild radius. Observing
a black hole by means of its Hawking radiation will always produce a fuzzy
image, unlike the image of the sun.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 6
Charged Black Holes
There are a variety of ways to generalize the conventional Schwarzschild
black hole. By going to higher dimensions we can consider not only black
holes, but black strings, black membranes, and so forth. Typically black
strings and branes are studied as systems of inﬁnite extent, and therefore
have inﬁnite entropy. For this reason they can store inﬁnite amounts of
information. Higher dimensional Schwarzschild black holes are quite similar
to their fourdimensional counterparts.
Another way to generalize the ordinary black hole is to allow it to carry
gauge charge and/or angular momentum. In this lecture we will describe
the main facts about charged black holes. The most important fact about
them is that they cannot evaporate away completely. They have ground
states with very special and simplifying features.
Thus, let us consider electrically charged black holes. The metric for a
Reissner–Nordstrom black hole is
ds
2
= −
1 −
2MG
r
+
Q
2
G
r
2
dt
2
+
1 −
2MG
r
+
Q
2
G
r
2
−1
dr
2
+ r
2
dΩ
2
(6.0.1)
The electric ﬁeld is given by the familiar Coulomb law
E
r
=
Q
r
2
E
θ,φ
= 0
(6.0.2)
If the electric ﬁeld is too strong at the horizon, it will cause pair produc
tion of electrons, which will discharge the black hole in the same manner
as a nucleus with Z >> 137 is discharged. Generally the horizon occurs
at r ∼ MG, and the threshold ﬁeld for unsupressed pair production is
E ∼ m
2
e
, where m
e
is the electron mass. Pair production is exponentially
55
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
56 Black Holes, Information, and the String Theory Revolution
suppressed if
Q
M
2
G
2
<< m
2
e
(6.0.3)
or alternativley
M
2
Q
>>
1
m
2
e
G
2
.
For Q
2
> M
2
G the metric in equation 6.0.1 has a timelike singularity
with no horizon to cloak it. Such “naked singularities” indicate a break
down of classical relativity visible to a distant observer. The question is
not whether objects with Q
2
> M
2
G can exist. Clearly they can. The
electron is such an object. The question is whether they can be described
by classical general relativity. Clearly they cannot. Accordingly we restrict
our attention to the case M
2
>
Q
2
G
or
M
2
Q
>
Q
G
. A Reissner–Nordstrom
black hole that saturates this relationship M
2
=
Q
2
G
is called an extremal
black hole. Thus equation 6.0.3 is satisﬁed if
Q >>
1
m
2
e
G
∼ 10
44
Black holes with charge >> 10
44
can only discharge by exponentially sup
pressed tunneling processes. For practical purposes we regard them as
stable.
The Reissner–Nordstrom solution has two horizons, an outer one and
an inner one. They are deﬁned by
1 −
2MG
r
±
+
Q
2
G
r
2
±
= 0 (6.0.4)
where r
+
(r
−
) refers to the outer (inner) horizon:
r
±
= MG
1 ±
1 −
Q
2
M
2
G
(6.0.5)
The metric can be rewritten in the form
ds
2
= −
(r − r
+
)(r −r
−
)
r
2
dt
2
+
r
2
dr
2
(r −r
+
)(r −r
−
)
+ r
2
dΩ
2
(6.0.6)
Note that in the extremal limit M
2
=
Q
2
G
the inner and outer horizons
merge at r
±
= MG.
To examine the geometry near the outer horizon, let us begin by comput
ing the distance fromr
+
to an arbitrary point r > r
+
. Using equation 6.0.6
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Charged Black Holes 57
we compute the distance ρ to be
ρ =
r
(r −r
+
)(r −r
−
)
dr (6.0.7)
We deﬁne the following
r
+
+ r
−
≡ Σ
r
+
− r
−
≡ ∆
y ≡ r −
Σ
2
(6.0.8)
We ﬁnd
ρ =
y+
Σ
2
y
2
−
(
∆
2
)
2
dy
=
y
2
−
∆
2
4
+
Σ
2
cosh
−1
2
∆
y
(6.0.9)
The radialtime metric is given by
ds
2
= −
y
2
−
∆
2
4
y +
Σ
2
2
dt
2
+ dρ
2
(6.0.10)
Expanding equation 6.0.9 near the horizon r
+
one ﬁnds
ρ ≈
y −
∆
2
1/2
2r
+
∆
1/2
. (6.0.11)
Note that the proper distance becomes inﬁnite for extremal black holes.
For nonextremal black holes, equation 6.0.10 becomes
ds
2
∼
=
∆
2
4r
4
+
ρ
2
dt
2
− dρ
2
∼
= ρ
2
dω
2
− dρ
2
(6.0.12)
where
ω ≡
∆
2r
2
+
t (6.0.13)
Evidently the horizon geometry is again well approximated by Rindler
space. The charge density on the horizon is
Q
4πr
2
+
. Since r
+
∼ MG the
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
58 Black Holes, Information, and the String Theory Revolution
charge density is ∼
Q
4πM
2
G
2
. Thus for near extremal black holes, the charge
density is of the form
Q
4πr
2
+
∼
M
√
G
4πM
2
G
2
∼
=
1
4πMG
3/2
(6.0.14)
For very massive black holes the charge density becomes vanishingly small.
Therefore the local properties of the horizon cannot be distinguished from
those of a Schwarzschild black hole. In particular, the temperature at a
small distance ρ
o
from the horizon is
1
2πρ
o
. From equation 6.0.13 we can
compute the temperature as seen at inﬁnity.
T(∞) =
∆
2r
2
+
1
2π
(6.0.15)
Using
∆ = 2MG
1 −
Q
2
M
2
G
r
+
= MG
1 +
1 −
Q
2
M
2
G
(6.0.16)
We ﬁnd
T(∞) =
2MG
1 −
Q
2
M
2
G
4πM
2
G
2
1 +
1 −
Q
2
M
2
G
2
(6.0.17)
As the black hole tends to extremality, the horizon becomes progres
sively more removed from any ﬁducial observer. From equation 6.0.9 we
see that as ∆ →0
ρ → y +
Σ
2
log(2y) − log∆ (6.0.18)
Thus for a ﬁducial observer at a ﬁxed value of r the horizon recedes to
inﬁnite proper distance as ∆ →0.
In the limit ∆ →0 the geometry near the horizon simpliﬁes to the form
ds
2
=
−
r
+
sinh
ρ
r
+
2
dω
2
+ dρ
2
+ r
2
+
dΩ
2
(6.0.19)
which, although inﬁnitely far from any ﬁducial observer with r = r
+
, is
approximately Rindler.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Charged Black Holes 59
We note from equation 6.0.17 that in the extremal limit the temperature
at inﬁnity tends to zero. The entropy, however, does not tend to zero. This
can be seen in two ways, by focusing either on the region very near the
horizon or the region at inﬁnity. As we have seen, the local properties of
the horizon even in the extreme limit are identical to the Schwarzschild case
from which we deduce an entropy density
1
4G
. Accordingly,
S =
Area
4G
=
πr
2
+
G
= πM
2
G (6.0.20)
We can deduce this result by using the ﬁrst law together with equation
6.0.17
dM = T dS (Fixed Q)
to obtain S =
Area
4G
as a general rule.
The fact that the temperature goes to zero in the extreme limit indicates
that the evaporation process slows down and does not proceed past the
point Q = M
√
G. In other words, the extreme limit can be viewed as the
ground state of the charged black hole. However it is unusual in that the
entropy does not also tend to zero. This indicates that the ground state
is highly degenerate with a degeneracy ∼ e
S
. Whether this degeneracy
is exact or only approximate can not presently be answered in the general
case. However in certain supersymmetric cases the supersymmetry requires
exact degeneracy.
The metric in equation 6.0.19 for extremal black holes can be written
in a form analogous to equations 1.3.18 and 1.4.21 by introducing a radial
variable
R
r
+
=
e
ρ/r
+
− 1
e
ρ/r
+
+ 1
(6.0.21)
The metric then takes the form
dτ
2
=
2
1 −
R
2
r
2
+
R
2
dω
2
− dR
2
− r
2
+
dΩ
2
(6.0.22)
Obviously the physics near
R
r
+
→0 is identical to Rindler space, from which
it follows that the horizon will have the usual properties of temperature,
entropy, and a thermal atmosphere including particles of high angular mo
menta trapped near the horizon by a centrifugal barrier.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
60 Black Holes, Information, and the String Theory Revolution
Although the external geometry of an extreme or near extreme Reissner–
Nordstrom black hole is very smooth with no large curvature, one can nev
ertheless expect important quantum eﬀects in its structure. To understand
why, consider the fact that as ∆ →0 the horizon recedes to inﬁnity. Clas
sically, if we drop the smallest amount of energy into the extreme black
hole, the location of the horizon, as measured by its proper distance, jumps
an inﬁnite amount. In other words, the location of the horizon of an ex
tremal black hole is very unstable. Under these circumstances, quantum
ﬂuctuations can be expected to make the location very uncertain. Whether
this eﬀect leads to a lifting of the enormous degeneracy of ground states
or any other physical phenomena is not known at present except in the
supersymmetric case.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 7
The Stretched Horizon
Thus far our description of the nearhorizon region of black holes, or
Rindler space, has been in terms of quantum ﬁeld theory in a ﬁxed back
ground geometry. But we have already run into a contradiction in applying
quantum ﬁeld theory, although we didn’t spell it out. The problem arose
in Chapter 4 when we found that the entropy per unit area of the horizon
diverges as the cutoﬀ tends to zero (see equation 4.0.22). That in itself
is not a problem. What makes it a problem is that we later found that
black hole thermodynamics requires the entropy to be
A
4G
. Free quantum
ﬁeld theory is giving too much entropy in modes very close to the horizon,
where the local temperature diverges. The fact that the entropy is inﬁ
nite in quantum ﬁeld theory implies that any quantity that depends on the
ﬁniteness of the entropy will be miscalculated using quantum ﬁeld theory.
One possibility is that we have overestimated the entropy by assuming
free ﬁeld theory. Equation 4.0.20 could be modiﬁed by interactions. Indeed
that is so, but the eﬀect goes in the wrong direction. The correct entropy
density for a general ﬁeld theory can always be parameterized by
S(T) = γ(T) T
3
where γ(T) represents the number of “eﬀective” degrees of freedom at tem
perature T. It is widely accepted and in many cases proven that γ(T) is
a monotonically increasing function of T. Thus, conventional interactions
are only likely to make things worse. What we need is some new kind of
theory that has the eﬀective number of degrees of freedom going to zero
very close to the horizon. Let’s suppose that ordinary quantum ﬁeld theory
is adequate down to distance scale . In order that the entropy at distance
greater than not exceed the Bekenstein–Hawking value, we must have the
rough inequality
61
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
62 Black Holes, Information, and the String Theory Revolution
2
<
∼
G =
2
P
Evidently at distances less than
√
G from the horizon the degrees of free
dom must be very sparse, or even nonexistent. This leads to the idea that
the mathematical horizon should be replaced by an eﬀective membrane, or
“stretched” horizon at a distance of roughtly one Planck length from the
mathematical horizon.
Stretching the horizon has another beneﬁt. Instead of being lightlike, a
system at the stretched horizon is timelike. This means that real dynamics
and evolution can take place on the stretched horizon. As we will see, the
stretched horizon has dynamics of its own that includes such phenomena
as viscosity and electrical conductivity.
To see that the horizon of a black hole has electrical properties, it is
suﬃcient to study electrodynamics in Rindler space. First let us deﬁne the
stretched horizon. The metric is
dτ
2
= ρ
2
dω
2
− dρ
2
− dx
2
⊥
(7.0.1)
The stretched horizon is just the surface
ρ = ρ
o
(7.0.2)
where ρ
o
is a length of order the Planck length.
The action for the electromagnetic ﬁeld in Rindler space is
W =
¸√
−g
16π
g
µν
g
στ
F
µσ
F
ντ
+ j
µ
A
µ
dω dρ d
2
x
⊥
(7.0.3)
or, substituting the form of the metric
W =
1
8π
¸
¸
˙
A +
∇φ
2
ρ
− ρ
∇×
A
2
¸
+ j · A
¸
¸
¸
dω dρ d
2
x
⊥
(7.0.4)
where
˙
A means
∂
A
∂ω
and φ = −A
0
, and j is a conserved current in the usual
sense ∂
µ
j
µ
= 0. As usual
E = −
∇φ −
˙
A
B =
∇×
A
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Stretched Horizon 63
With these deﬁnitions, the action becomes
W =
1
8π
¸
¸
E
2
ρ
− ρ
B
2
¸
+ j · A
¸
¸
¸
dω dρ d
2
x
⊥
(7.0.5)
and the Maxwell equations are
1
ρ
˙
E −
∇×(ρB) = −4πj
˙
B +
∇×
E = 0
∇·
1
ρ
E
= 4πj
0
∇·
B = 0
(7.0.6)
We begin by considering electrostatics. By electrostatics we mean the
study of ﬁelds due to stationary or slowly moving charges placed outside
the horizon. Since the charges are slowly moving in Rindler coordinates,
it means that they are experiencing proper acceleration. We also assume
all length scales associated with the charges are much larger than ρ
o
. In
particular, the distance of the charges from the stretched horizon is macro
scopic.
The surface charge density on the stretched horizon is easily deﬁned.
It is just the component of the electric ﬁeld perpendicular to the stretched
horizon, or more precisely
σ =
1
4πρ
E
ρ
ρ=ρ
o
= −
1
4πρ
∂
ρ
φ
ρ=ρ
o
(7.0.7)
Working in the Coulomb gauge, the third expression in equation 7.0.6 be
comes
∇·
1
ρ
E
= −
∇·
1
ρ
∇φ
= 0 (7.0.8)
near the stretched horizon. Thus
∂
2
ρ
φ −
1
ρ
∂
ρ
φ = −∇
2
⊥
φ (7.0.9)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
64 Black Holes, Information, and the String Theory Revolution
We can solve this equation near the horizon by the ansatz φ ∼ ρ
α
. The
right hand side will be smaller than the left hand side by 2 powers of ρ and
can therefore be ignored. We easily ﬁnd that α = 0 or α = 2. Thus we
assume
φ = F(x
⊥
) + ρ
2
G(x
⊥
) + terms higher order inρ (7.0.10)
Plugging equation 7.0.10 into equation 7.0.9 and evaluating at ρ = ρ
o
gives
∇
2
⊥
F + ρ
2
o
∇
2
⊥
G = 0 (7.0.11)
If ρ
o
is much smaller than all other length scales, then equation 7.0.11 is
simpliﬁed to
∇
2
⊥
F = 0 (7.0.12)
A similar equation can also be derived for the ﬁnite mass black hole.
Since the black hole horizon is compact, equation 7.0.12 proves that
φ = constant on the horizon. This is an interesting result, which proves
that the horizon behaves like an electrical conductor.
We can easily identify the surface current density. Taking the time
derivative of equation 7.0.7 and using Maxwell’s equations 7.0.6 gives
4π ˙ σ =
1
ρ
o
˙
E
ρ
=
∇×ρ
B
ρ
(7.0.13)
Evidently this is a continuity equation if we deﬁne:
4π j
x
= −ρ B
y
4π j
y
= ρ B
x
(7.0.14)
Now let us consider an electromagnetic wave propagating toward the
stretched horizon along the ρ axis. From Maxwell’s equations we obtain
˙
B
x
= ∂
ρ
E
y
˙
B
y
= −∂
ρ
E
x
1
ρ
˙
E
x
= −∂
ρ
(ρ B
y
)
1
ρ
˙
E
y
= ∂
ρ
(ρ B
x
)
(7.0.15)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Stretched Horizon 65
To make the equation more familiar, we can redeﬁne the magnetic ﬁeld
ρ
B =
β (7.0.16)
and use tortoise coordinates
u = log ρ
Equation 7.0.15 then becomes
˙
β
x
= ∂
u
E
y
˙
β
y
= −∂
u
E
x
˙
E
x
= ∂
u
β
y
˙
E
y
= −∂
u
β
x
(7.0.17)
The mathematical equations allow solutions in which the wave propa
gates in either direction along the uaxis. However the physics only makes
sense for waves propagating toward the horizon from outside the black hole.
For such waves, the Maxwell equations 7.0.17 give
β
x
= E
y
β
y
= −E
x
(7.0.18)
or from equation 7.0.14
j
x
=
1
4π
E
x
j
y
=
1
4π
E
y
Evidently the horizon is an ohmic conductor with a resistivity of 4π.
That corresponds to a surface resistance of 377Ω/square.
∗
For example, if
a circuit is constructed as in Figure 7.1, a current will ﬂow precisely as if
the horizon were a conducting surface.
As a last example let us consider dropping a charged point particle
into the horizon. Since the horizon is just ordinary ﬂat space, one might
conclude that the point charge just asymptotically approaches the horizon
∗
The unit Ω/square is not a misprint. The resistance of a twodimensional resistor is
scale invariant and only depends on the shape.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
66 Black Holes, Information, and the String Theory Revolution
Horizon
A
A
Fig. 7.1 Battery, ammeter attached to horizon
with the transverse charge density remaining pointlike. However this is
not at all what happens on the stretched horizon. This process is shown
in Figure 7.2. Without loss of generality, we can take the charge to be
at rest at position z
o
in Minkowski coordinates. To compute the surface
charge density on the stretched horizon, we need to determine the ﬁeld
component E
ρ
. The calculation is easy because at any given time the
Rindler coordinates are related to the Minkowski coordinates by a boost
along the zaxis. Since the component of electric ﬁeld along the boost
direction is invariant, we can write the standard Coulomb ﬁeld
E
ρ
= E
z
=
e (z−z
o
)
[
(z−z
o
)
2
+x
2
⊥
]
3/2
=
e (ρ coshω −z
o
)
[
(ρ coshω −z
o
)
2
+x
2
⊥
]
3/2
(7.0.19)
Using 4π σ =
E
ρ
ρ
ρ
o
we ﬁnd
σ =
e
4πρ
o
ρ
o
coshω − z
o
(ρ
o
coshω − z
o
)
2
+ x
2
⊥
3/2
(7.0.20)
Now let’s consider the surface density for large Rindler time.
σ =
e
4πρ
o
ρ
o
e
ω
[ρ
2
o
e
2ω
+ x
2
⊥
]
3/2
(7.0.21)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Stretched Horizon 67
Stretched
horizon
T
Z
H
o
r
i
z
o
n
Z
0
Freely falling charge
Fig. 7.2 Charge falling past the stretched horizon
To better understand this expression, it is convenient to rescale x
⊥
using
x
⊥
= e
ω
y
⊥
to obtain
σ =
e
4π
e
−2ω
(ρ
2
o
+ y
2
⊥
)
3/2
(7.0.22)
It is evident that the charge spreads out at an exponential rate with Rindler
time. For a real black hole, it would spread over the horizon in a time
ω = log (R
s
− ρ
o
) = log (2MG − ρ
o
)
or in terms of the Schwarzschild time
t = 4MGω = 4MGlog (2MG − ρ
o
)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
68 Black Holes, Information, and the String Theory Revolution
The exponential spreading of the charge is characteristic of an Ohm’s
law conductor. To see this we use Ohm’s law j = conductivity E. Taking
the divergence gives
∇·
j ∼
∇·
E ∼ σ. Now use the continuity equation get
the relation ˙ σ ∼ −σ. Evidently the surface charge density will exponentially
decrease, and conservation of charge will cause it to spread exponentially.
Thus we see that the horizon has the properties of a more or less con
ventional hot conducting membrane. In addition to temperature, entropy,
and energy, it exhibits dissipative eﬀects such as electrical resistivity and
viscosity. The surprising and puzzling thing is that they are completely
unnoticed by a freely falling observer who falls through the horizon!
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 8
The Laws of Nature
In this chapter, we want to review three fundamental laws of nature
whose compatibility has been challenged. These laws are:
1) The principle of information conservation
2) The equivalence principle
3) The quantum xerox principle
8.1 Information Conservation
In both classical and quantum mechanics there is a very precise sense in
which information is never lost from a closed isolated system. In classical
physics the principle is embodied in Liouville’s theorem: the conservation
of phase space volume. If we begin following a system with some limited
knowledge of its exact state, we might represent this by specifying an initial
region Γ(0) in the system’s phase space. The region Γ(0) has a volume V
Γ
in the phase space.
Now we let the system evolve. The region Γ(0) = Γ evolves into the
region Γ(t). Liouville’s theorem tells us that the volume of Γ(t) is exactly
the same as that of Γ. In this sense the amount of information is conserved.
In a practical sense, information is lost because for most cases of interest
the region Γ becomes very complicated like a fractal, and if we coarse grain
the phase space, it will appear that Γ is growing. As a deﬁnition of coarse
graining, if one takes every point in the phase space and surrounds it by
solid spheres of ﬁxed volume, the union of those spheres is the “coarse
grained” volume of phase space, which indeed grows. This is the origin of
the second law of thermodynamics. This is illustrated in Figure 8.1.
In quantum mechanics, the conservation of information is expressed as
69
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
70 Black Holes, Information, and the String Theory Revolution
Fig. 8.1 Evolution of a ﬁxed volume in phase space
the unitarity of the Smatrix. If we again approach a system with limited
knowledge, we might express this by a projection operator onto a subspace,
P, instead of a deﬁnite state. The analog of the phase space volume is the
dimensionality of the subspace
N = Tr P
The unitarity of the time evolution insures that N is conserved with time.
A more reﬁned deﬁnition of information is provided by the concept of
entropy. Suppose that instead of specifying a region Γ in phase space, we
instead specify a probability density ρ(p, q) in phase space. A generalization
of the volume is given by the exponential of the entropy V
Γ
→exp S, where
S = −
dp dq ρ(p, q) log ρ(p, q) (8.1.1)
It is easy to check that if ρ =
1
V
Γ
inside Γ and zero outside, then S = log V
Γ
.
Similarly, for quantum mechanics the sharp projector P can be replaced
by a density matrix ρ. In this case the ﬁne grained or Von Neumann entropy
is
S = −Tr ρ log ρ (8.1.2)
For the case ρ =
P
Tr P
the entropy is log N. Thus the entropy is an esti
mate of the logarithm of the number of quantum states that make up the
initial ensemble. In both quantum mechanics and classical mechanics the
equations of motion insure the exact conservation of S for a closed, isolated
system.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Laws of Nature 71
8.2 Entanglement Entropy
In classical physics, the only reason for introducing a phase space prob
ability is a lack of detailed knowledge of the state. In quantum mechanics,
there is another reason, entanglement. Entanglement refers to quantum
correlations between the system under investigation and a second system.
More precisely, it involves separating a system into two or more subsystems.
Consider a composite system composed of 2 subsystems A and B. The
subsystem A(B) is described by some complete set of commuting observ
ables α(β). Let us assume that the composite system is in a pure state
with wave function Ψ(α, β). Consider now the subsystems separately. All
measurements performed on A(B) are describable in terms of a density
matrix ρ
A
(ρ
B
).
(ρ
A
)
αα
=
β
Ψ
∗
(α, β) Ψ(α
, β)
(ρ
B
)
ββ
=
α
Ψ
∗
(α, β) Ψ(α, β
)
(8.2.3)
The fact that a subsystem is described by a density matrix and not
a pure state may not be due to any lack of knowledge of the state of
the composite system. Even in the case of a pure state, the constituent
subsystems are generally not described by pure states. The result is an
“entanglement entropy” for the subsystems.
Let us consider some properties of the density matrix. For deﬁniteness,
consider ρ
A
, but we could equally well focus on ρ
B
.
1) The density matrix is Hermitian
(ρ
A
)
αα
= (ρ
A
)
∗
α
α
(8.2.4)
2) The density matrix is positive semideﬁnite. This means its eigenvalues
are all either positive or zero.
3) The density matrix is normalized to 1.
Tr ρ = 1 (8.2.5)
It follows that all the eigenvalues are between zero and one. If one of the
eigenvalues of ρ
A
is equal to 1, all the others must vanish. In this case
the subsystem A is in a pure state. This only happens if the composite
wave function factorizes
Ψ(α, β) = ψ
A
(α) ψ
B
(β) (8.2.6)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
72 Black Holes, Information, and the String Theory Revolution
In this case B is also in a pure state.
4) The nonzero eigenvalues of ρ
A
and ρ
B
are equal if the composite system
is in a pure state. To prove this, we start with the eigenvalue equation
for ρ
A
. Call φ the eigenvector of ρ
A
. Then the eigenvalue condition is
βα
Ψ
∗
(α, β) Ψ(α
, β) φ(α
) = λφ(α)
We assume λ = 0. Now we deﬁne a candidate eigenvector of ρ
B
by
χ(β
) ≡
α
Ψ
∗
(α
, β
) φ
∗
(α
).
Then
β
(ρ
B
)
ββ
χ(β
) =
αβ
Ψ
∗
(α, β) Ψ(α, β
) χ(β
)
=
αα
β
Ψ
∗
(α, β) Ψ(α, β
) Ψ
∗
(α
, β
) φ
∗
(α
)
= λ
α
Ψ
∗
(α, β) φ
∗
(α)
= λ χ(β)
Thus χ(β) is an eigenvector of ρ
B
with eigenvalue λ.
From the equality of the nonvanishing eigenvalues of ρ
A
and ρ
B
an
important property of entanglement entropy follows:
S
A
= −Tr ρ
A
log ρ
A
= S
B
(8.2.7)
Thus we can just refer to the entanglement entropy as S
E
.
The equality of S
A
and S
B
is only true if the combined state is pure.
In that case, the entropy of the composite system vanishes
S
A+B
= 0
Evidently entropy is not additive in general.
Next, let us consider a large system Σ that is composed of many similar
small subsystems σ
i
. Let us suppose the subsystems weakly interact, and
the entire system is in a pure state with total energy E. Each subsystem
on the average will have energy .
It is a general property of most complex interacting systems that the
density matrix of a small subsystem will be thermal
ρ
i
=
e
−β H
i
Z
i
(8.2.8)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Laws of Nature 73
where H
i
is the energy of the subsystem. The thermal density matrix max
imizes the entropy for a given average energy . In general large subsystems
or the entire system will not be thermal. In fact, we will assume that the
entire system Σ is in a pure state with vanishing entropy.
The coarse grained or thermal entropy of the composite system is deﬁned
to be the sum of the entropies of the small subsystems
S
Thermal
=
i
S
i
(8.2.9)
By deﬁnition it is additive. The coarse grained entropy is what we usually
think of in the context of thermodynamics. It is not conserved. To see
why, suppose we start with the subsystems in a product state with no
corelations. The entropy of each subsystem S
i
as well as the entropy of
the whole system Σ given by σ
Σ
, and the coarse grained entropy of Σ all
vanish.
Now the subsystems interact. The wave function develops correlations,
meaning that it now fails to factorize. In this case, the subsystem entropies
become nonzero
S
i
= 0
and the coarse grained entropy also becomes nonzero
S
Thermal
=
i
S
i
= 0
However, the “ﬁne grained” entropy of Σ is exactly conserved and therefore
remains zero.
Let us consider an arbitrary subsystem Σ
1
of Σ which may consist of
one, many, or all of the subsystems σ
i
. Typically the ﬁne grained entropy of
Σ
1
is deﬁned as the entanglement entropy S(Σ
1
) of Σ
1
with the remaining
subsystem Σ−Σ
1
. This will always be less than the coarse grained entropy
of Σ
1
S
Thermal
(Σ
1
) > S(Σ
1
) (8.2.10)
For example, as Σ
1
approaches Σ, the ﬁne grained entropy S(Σ
1
) will tend
to zero.
Another concept that we can now make precise is the information in a
subsystem. The information can be deﬁned by
I = S
Thermal
− S (8.2.11)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
74 Black Holes, Information, and the String Theory Revolution
Often the coarse grained entropy is the thermal entropy of the system,
so that the information is the diﬀerence between coarse grained and ﬁne
grained entropy.
Since typical small subsystems have a thermal density matrix, the in
formation in a small subsystem vanishes. At the opposite extreme the
information of the combined system Σ is just its total thermal entropy. It
can be thought of as the hidden subtle correlations between subsystems
that make the state of Σ pure.
How much information are in a moderately sized subsystem? One
might think that the information smoothly varies from zero (for the σ
i
)
to S
Coarse Grained
(for Σ). However, this is not so. What actually hap
pens is that for subsystems smaller than about 1/2 of the total system, the
information is negligible.
Entropy and information are naturally measured in “bits”. A bit is the
entropy of a two state system if nothing is known
[
2
]
. The numerical value
of a bit is log 2. Typically for subsystems less than half the size of Σ the
information is smaller than 1 bit. The subsystem
1
2
Σ has about 1 bit of
information. Thus for Σ
1
<
1
2
Σ
S(Σ
1
)
∼
= S
Thermal
(Σ
1
)
I(Σ
1
) ≈ 0
Next consider a subsystem with Σ
1
>
1
2
Σ. How much information does
it have? To compute it, we use two facts:
S(Σ −Σ
1
) = S(Σ
1
)
S(Σ −Σ
1
)
∼
= S
Thermal
(Σ −Σ
1
)
(8.2.12)
Thus
S(Σ
1
)
∼
= S
Thermal
(Σ −Σ
1
) (8.2.13)
The coarse grained entropy of Σ−Σ
1
will be of order (1 −f) S
Thermal
(Σ),
where f is the fraction of the total degrees of freedom contained in Σ
1
.
Thus, for Σ
1
<
1
2
Σ the information in Σ
1
is essentially zero. However for
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Laws of Nature 75
Σ
1
>
1
2
Σ we get the information to be
I(Σ
1
) = S
Thermal
(Σ
1
) −S(Σ
1
)
= f S
thermal
(Σ) − (1 −f) S
Thermal
(Σ)
= ( 2 f − 1 ) S
Thermal
(Σ)
(8.2.14)
We will eventually be interested in the information emitted by a black
hole when it evaporates. For now let’s consider a conventional system which
is described by known laws of physics. Consider a box with perfectly re
ﬂecting walls. Inside the box we have a bomb which can explode and ﬁll
the box with radiation. The box has a small hole that allows the thermal
radiation to slowly leak out. The entire system Σ consists of the subsys
tem B that includes everything in the box. The subsystem A consists of
everyting outside of the box, in this case, outgoing photons.
Initially the bomb is in its ground state, and B has vanishing entropy.
When the bomb explodes, it ﬁlls the box with thermal radiation. The
thermal, or coarse grained, entropy of the box increases, but its ﬁne grained
entropy does not. Furthermore, no photons have yet escaped, so S(A) = 0
at this time.
S
Thermal
(B) = 0
S(B) = 0
S(A) = 0
(8.2.15)
Next, photons slowly leak out. The result is that the interior and ex
terior of the box become entangled. The entanglement entropy, which is
equal for A and B, begins to increase. The thermal entropy in the box
decreases:
S
Entanglement
= 0
S
Thermal
(B) = 0
S
Thermal
(A) = 0
(8.2.16)
Eventually, all of the photons escape the box. The thermal or coarse
grained entropy as well as the ﬁne grained entropy in the box tends to zero.
The box is in a pure state; its ground state.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
76 Black Holes, Information, and the String Theory Revolution
1
(Σ ) S
1
f
Pure state
2

1
2

1
1
f
S (Σ )
1 Thermal
Additive coarsegrained
entropy
2

1
1
f
(Σ )
1 I
Fig. 8.2 Top: Von Neumann entropy of Σ
1
vs fraction f. Middle: coarse grained
entropy of Σ
1
vs fraction f. Bottom: information vs fraction of total degrees of
freedom in Σ
1
At this time, the thermal or coarse grained entropy of the exterior ra
diation has increased to its ﬁnal value. The second law of thermodynamics
insures that S
Thermal
(A) is larger than S
Thermal
(B) just after the explo
sion. But the ﬁne grained entropy of A must vanish, since the entanglement
has gone to zero.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Laws of Nature 77
The actual entanglement entropy must be less than the thermal entropy
of A or B. Thus a plot of the various entropies looks like Figure 8.3. Note
that the point at which S
Thermal
(A) = S
Thermal
(B) deﬁnes the time at
which the information in the outside radiation begins to grow. Before that
point, a good deal of energy has escaped, but no information. Roughly the
point where information appears outside of the box is the point where half
of the ﬁnal entropy of the photons has emerged.
It is useful to deﬁne this time at which information begins to emerge.
This time is called the information retention time. It is the amount of time
that it takes to retrieve a single bit of information about the initial state of
the box.
Thus we see how information conservation works for a conventional
quantum system. The consequence of this principle is the ﬁnal radiation
ﬁeld outside the box must be in a pure state. However, this does not mean
that localized regions containing a small fraction of the photons cannot be
extremely thermal. They typically carry negligible information.
The description of the evolution of the various kinds of entropy follow
from very general principles. Thus we regard the conservation of informa
tion in black hole evaporation as a fundamental law of nature. Note that
it applies to observations made from outside the black hole.
8.3 Equivalence Principle
A second law of nature concerns the nature of gravitation. In its sim
plest form the equivalence principle says that a gravitational ﬁeld is locally
equivalent to an accelerated frame. More exactly, it says that a freely falling
observer or system will not experience the eﬀects of gravity except through
the tidal forces, or equivalently, the curvature components. We have seen
that the magnitude of the curvature components at the horizon are small
and tend to zero as the mass and radius of the black hole increase. The
curvature typically satisﬁes
R ∼
1
(MG)
2
Any freely falling system of size much smaller than MGwill not be distorted
or otherwise disrupted by the presence of the horizon.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
78 Black Holes, Information, and the String Theory Revolution
Information
retention
time
t
Information
E
n
ta
n
g
le
m
e
n
t
e
n
tr
o
p
y
A (External)
B (Box)
t
Information
retention
time
Thermal
S
Fig. 8.3 Top: evolution of the thermal entropies of box and exterior. Bottom:
evolution of entanglement entropy and information
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Laws of Nature 79
The equivalence principle requires the horizon of a very large black hole
to have the same eﬀects on a freely falling observer as the horizon of Rindler
space has; namely, no eﬀect at all.
8.4 Quantum Xerox Principle
The third law of nature that plays an important role in the next lecture
is the impossibility of faithfully duplicating quantum information. What
it says is that a particular kind of apparatus cannot be built. We call it
a Quantum Xerox Machine.
∗
It is a machine into which any system can
be inserted, and which will copy that system, producing the original and a
duplicate. To see why such a system is not possible, imagine that we insert
a spin in the input port as in Figure 8.4. If the spin is in the up state with
In Out
Fig. 8.4 Schematic diagram of quantum Xerox machine
respect to the zaxis, it is duplicated
↑ → ↑ ↑ (8.4.17)
Similarly, if it is in the down state, it is duplicated
↓ → ↓ ↓ (8.4.18)
Now suppose that the spin is inserted with its polarization along the xaxis,
i.e.
1
√
2
(↑ + ↓) (8.4.19)
∗
The quantum Xerox principle is sometimes called the nocloning principle
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
80 Black Holes, Information, and the String Theory Revolution
The general principles of quantum mechanics require the state to evolve
linearly. Thus from equations 8.4.17 and 8.4.18
1
√
2
(↑ + ↓) →
1
√
2
(↑ ↑ + ↓ ↓) (8.4.20)
On the other hand, a true quantum Xerox machine is required to duplicate
the spin in equation 8.4.19
1
√
2
(↑ + ↓) →
1
√
2
(↑ + ↓)
1
√
2
(↑ + ↓)
=
1
2
↑ ↑ +
1
2
↑ ↓ +
1
2
↓ ↑ +
1
2
↓ ↓
(8.4.21)
The state in equation 8.4.20 is obviously not the same as that in equation
8.4.21. Thus we see that the principle of linearity forbids the existence of
quantum Xerox machines. If we could construct Xeroxed quantum states,
we would be able to violate the Heisenberg uncertainty principle by a set
of measurements on those states.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 9
The Puzzle of Information
Conservation in Black Hole
Environments
In 1976 Hawking raised the question of whether information is lost in
the process of formation and evaporation of black holes. By information
loss, Hawking did not mean the practical loss of information such as would
occur in the bombinthebox experiment in Chapter 8. He meant that in
a ﬁne grained sense, information would be lost. In other words, the ﬁrst
of the laws of nature described in the Introduction would be violated. The
argument was simple and persuasive. It was based on the only available tool
of that time, namely local quantum ﬁeld theory in the ﬁxed background of a
black hole. Although Hawking’s conclusion is undoubtedly wrong, it played
a central role in replacing the old ideas of locality with a new paradigm.
As we have already seen, quantum ﬁeld theory has a serious defect when
it comes to describing systems with horizons. It gives rise to an inﬁnite
entropy density on the horizon, instead of the correct Bekenstein–Hawking
value of
c
3
4G
. As we will see, quantum ﬁeld theory must be replaced with an
entirely new paradigm in which the concept of locality is radically altered.
To state the problem, let’s begin with a black hole that has been formed
during a collapse. The Penrose diagram is shown in Figure 9.1. Following
Hawking, we think of the geometry as a background on which we can for
mulate quantum ﬁeld theory. Let us concentrate on a theory of massless
particles.
According to the principles of quantum mechanics, the evolution of an
initial state ψ
in
is governed by a unitary Smatrix, so that the ﬁnal state
is given by
ψ
out
= S ψ
in
(9.0.1)
One way of stating the principle of information conservation is through the
unitarity of S. The point is that a unitary matrix has an inverse, so that
81
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
82 Black Holes, Information, and the String Theory Revolution
Collapsing
energy
Singularity
H
o
r
i
z
o
n
+

Fig. 9.1 Penrose diagram of black hole formed by collapse
in principle the initial state can be recovered from the ﬁnal state
ψ
in
= S
†
ψ
out
(9.0.2)
Now let us consider the Smatrix in a black hole background. The
Hilbert space of initial states h
in
is clearly associated with quanta coming
in from I
−
. These incoming particles will interact and scatter by means of
Feynman diagrams in the black hole background. It is evident from Figure
9.2 that some of the ﬁnal particles will escape to I
+
and some will be
lost behind the horizon. Evidently the ﬁnal Hilbert space, h
out
is a tensor
product of the states on I
+
and those at the singularity S. Thus
h
in
= h
I
−
h
out
= h
I
+ ⊗h
S
(9.0.3)
In arguing that the ﬁnal Hilbert space is a tensor product, Hawking
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Puzzle of Information Conservation in Black Hole Environments 83
S
+

Fig. 9.2 Feynman diagrams in black hole background
relied on an important property of quantum ﬁeld theory: locality. Since
every point on the singularity S is spacelike relative to every point on I
+
,
the operators on S all commute with those on I
+
. This was central to
Hawking’s analysis.
Now suppose that the ﬁnal state is given by an Smatrix which connects
h
in
and h
out
as in equation 9.0.1. Let us consider the description of the
ﬁnal state from the viewpoint of the observers at I
+
. Since they have no
access to S, all experiments on the outgoing particles are described by a
density matrix
ρ
out
= Tr
singularity
ψ
out
ψ
out
 (9.0.4)
where Tr
singularity
means a trace over the states on the singularity. Thus, in
general the observer external to the black hole will see a state characterized
by a density matrix.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
84 Black Holes, Information, and the String Theory Revolution
A simple example might involve a pair of correlated particles which are
thrown in from I
−
. By correlated, we mean that the two particles wave
function is entangled. If one particle ends up on S and the other ends up
on I
+
, then the ﬁnal state will be entangled. In this case the state on I
+
will not be pure.
Hawking went on to make arguments that the purity of the state would
not be restored if the black hole evaporated. In fact, the only possibilities
would seem to be that either information is lost during the entire process
of formation and evaporation, or the information is restored to the outside
world at the very end of the evaporation process, when the singularity is
“exposed” at Planckian temperature.
However, we have seen in Chapter 8 that the maximum amount of
information that can be hidden in a system is its entropy. Once the entropy
of the black hole has evaporated past
1
2
its original value, it must begin to
come out in the emitted radiation. This is a fundamental law of quantum
mechanics. By the time the black hole has small mass and entropy, the
entanglement entropy of the radiation cannot be larger than the black hole’s
remaining entropy. Thus, even if all information were emitted at the very
end of the evaporation process, a law of nature would be violated from
the viewpoint of the external observer. The situation is even worse if the
information is not emitted at all.
A ﬁnal possibility that was advocated by some authors is that black
holes never completely evaporate. Instead they end their lives as stable
Planckmass remnants that contain all the lost information. Obviously
such remnants would have to have an enormous, or even inﬁnite entropy.
Such remnants would be extremely pathological, and we will not pursue
that line further.
9.1 A Brick Wall?
There are two more possibilities worth pointing out. One is that the
horizon is not penetrable. In other words, from the viewpoints of an in
falling system, the horizon bounces everything out. A freely falling observer
would encounter a “brick wall” just above the horizon.
The reason that this was never seriously entertained, especially by rel
ativists, is that it badly violates the equivalence principle. Since the near
horizon region of a Schwarzschild black hole is essentially ﬂat spacetime,
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Puzzle of Information Conservation in Black Hole Environments 85
any violent disturbance to a freely falling system would violate the second
law of nature in the Introduction. Even more convincing is the fact that
the horizon of a black hole formed by a lightlike shell forms before the shell
gets to the center (see Figures 1.10 and 1.12).
Finally, the quantum Xerox principle closes out the last possibility. The
information conservation principle requires all information to be returned
to the outside in Hawking radiation. The equivalence principle, on the
other hand, requires information to freely pass through the horizon. The
quantum Xerox principle precludes both happening. In other words, the
horizon cannot duplicate information, and send one copy into the black hole
while sending a second copy out. Evidently we have come to an impasse. It
seems that some law of nature must break down, at least for some observer.
9.2 Black Hole Complementarity
In its simplest form, black hole complementarity just says that no ob
server ever witnesses a violation of a law of nature. Thus, for an external
observer it says:
A black hole is a complex system whose entropy is a measure of its
capacity to store information. It tells us that the entropy is the log of the
number of microstates of the degrees of freedom that make up the black
hole. It does not tell us what those microdegrees of freedom are, but it
allows us to estimate their number. That number is of order the area of
the horizon in Planck units.
Furthermore, it tells us that the microdegrees of freedom can absorb,
thermalize, and eventually reemit all information in the form of Hawking
radiation. At any given time, the ﬁne grained entropy of the radiation ﬁeld
(due to entanglement) cannot exceed the entropy of the black hole. At the
end of evaporation, all information is carried oﬀ in Hawking radiation.
For a freely falling observer, black hole complementarity tells us that
the equivalence principle is respected. This means that as long as the
black hole is much larger than the infalling system, the horizon is seen as
ﬂat featureless spacetime. No high temperatures or other anomolies are
encountered.
No obvious contradiction is posed for the external observer, since the
infalling observer cannot send reports from behind the horizon. But a
potential contradiction can occur for the infalling observer.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
86 Black Holes, Information, and the String Theory Revolution
Singularity
M
e
s
s
a
g
e
I
n
f
o
r
m
a
t
i
o
n
A
B
C
Fig. 9.3 Information exchange from External to infalling observer
Consider a black hole, shown in Figure 9.3 along with an infalling sys
tem A. System A is assumed to contain some information. According to
observations done by A, it passes through the horizon without incident.
Next, consider an observer B who hovers above the black hole monitor
ing the Hawking radiation. According to assumption, the photons recorded
by observer B encode the information carried in by system A. After col
lecting some information about A (from the Hawking radiation ), observer
B then jumps into the black hole. We don’t actually need observer B to
decode the information. All we really need is a mirror outside the black
hole horizon to reﬂect the Hawking radiation back into the black hole.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Puzzle of Information Conservation in Black Hole Environments 87
We now have two copies of the original information carried by A. We
can imagine A sending a signal to observer B so that observer B discovers
the duplicate information at point C in Figure 9.3. Now we have a contra
diction, since observer B has discovered a quantum Xeroxing of information
from observer A. If this experiment is possible, then black hole complemen
tarity is not selfconsistent. To see why the experiment fails, we need to
remember that no information will be emitted until about
1
2
the entropy of
the black hole has evaporated. From equation 5.0.16, this takes a time of
order
t
∗
≈ M
3
G
2
(9.2.5)
Let us also assume that the observer B hovers above the horizon at a
distance at least of the order of the Planck length
P
. In other words,
observer B hovers above the stretched horizon. This means that observer
B’s jump oﬀ point must correspond to Rindler coordinates satisfying
ω
∗
≥
t
∗
4MG
≈ M
2
G
ρ
∗
≥
P
(9.2.6)
In terms of the light cone coordinates x
±
= ρe
±ω
we have
x
+
∗
x
−
∗
>
2
P
x
+
∗
>
≈
P
exp (ω
∗
)
(9.2.7)
From Figure 9.4 we can see why there might be a problem with the
experiment. Observer B might be forced to the singularity before a message
can arrive. In fact, the singularity is given by
x
+
x
−
= (M G)
2
(9.2.8)
Thus, observer B will hit the singularity at a point with
x
−
<
∼
(M G) e
−ω
∗
(9.2.9)
The implication is that if A is to send a signal that B can receive, it must
all occur at x
−
< (M G) e
−ω
∗
. This means that A has a time of order
∆t ≈ (M G) e
−ω
∗
to send the message.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
88 Black Holes, Information, and the String Theory Revolution
Singularity
X X = (MG)
+  2
H
o
r
i
z
o
n
Information
from A
B gathers
information
and enters
horizon
M
e
s
s
a
g
e
f
r
o
m
A
A
B
X

X
+
V
V
V
V
V
V
V
V
V
V
V
V
VVVVV
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
VVVVV
V
V
V
V
V
V
V
V
V
V
P
Fig. 9.4 Resolution of Xerox paradox for observers within horizon
Now, in the classical theory, there are no limits on how much infor
mation can be sent in an arbitrarily small time with arbitrarily small en
ergy. However in quantum mechanics, to send a single bit requires at least
one quantum. Since that quantum must be emitted between x
−
= 0 and
x
−
= MGe
−ω
∗
, its energy must satisfy
E >
1
MG
e
ω
∗
(9.2.10)
From equation 9.2.6 we see that this energy is exponential in the square
of the black hole mass (in Planck units) E >
e
M
2
G
MG
. In other words, for
observer A to be able to signal observer B before observer B hits the singu
larity, the energy carried by observer A must be many orders of magnitude
larger than the black hole mass. It is obvious that A cannot ﬁt into the
horizon, and that the experiment cannot be done.
This example is one of many which show how the constraints of quan
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Puzzle of Information Conservation in Black Hole Environments 89
tum mechanics combine with those of relativity to forbid violations of the
complementarity principle.
9.3 Baryon Number Violation
The conservation of baryon number is the basis for the incredible stabil
ity of ordinary matter. Nevertheless, there are powerful reasons to believe
that baryon number, unlike electric charge, can at best be an approximate
conservation law. The obvious diﬀerence between baryon number and elec
tric charge is that baryon number is not the source of a long range gauge
ﬁeld. Thus it can disappear without some ﬂux having to suddenly change
at inﬁnity.
Consider a typical black hole of stellar mass. It is formed by the collapse
of roughly 10
57
nucleons. Its Schwarzschild radius is about 1 kilometer, and
its temperature is 10
−8
electron volts. It is far too cold to emit anything
but very low energy photons and gravitons. As it radiates, its temperature
increases, and at some point it is hot enough to emit massive neutrinos and
antineutrinos, then electrons, muons, and pions. None of these carry oﬀ
baryon number. It can only begin to radiate baryons when its temperature
has increased to about 1 GeV. Using the connection between mass and
temperature in equation 4.0.22, the mass of the black hole at this point is
about 10
10
kilograms. This is a tiny fraction of the original black hole mass,
and even if it were to decay into nothing but protons, it could produce
only about 10
37
of them. Baryon number must be violated by quantum
gravitational eﬀects.
In fact, most modern theories predict baryon violation by ordinary quan
tum ﬁeld theoretic processes. As a simpliﬁed example, let’s suppose there
is a heavy scalar particle X which can mediate a transition between an
elementary proton and a prositron, as well as between two positrons, as
in Figure 9.5. Since the Xboson is described by a real ﬁeld, it cannot
carry any quantum numbers, and the transition evidently violates baryon
conservation. The proton could then decay into a positron and an electron
positron pair. Let’s also assume that the coupling has the usual Yukawa
form
g
¯
ψ
p
ψ
e
+ X +
¯
ψ
e
+ ψ
p
X
where g is a dimensionless coupling.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
90 Black Holes, Information, and the String Theory Revolution
p
X
X
e
+
e
+
e
+
Fig. 9.5 Feynman diagrams of p→Xe
+
and e
+
→Xe
+
If the mass M
X
is suﬃciently large, baryon conservation will be a very
good symmetry at ordinary energies, just as it is in the real world. The
proton will have a lifetime in excess of 10
32
years.
Now let us consider what happens when a proton falls into a black
hole. The baryon number is lost, and will not be radiated back out in the
Hawking radiation. The question is: where does the baryon violation take
place? One possible answer is that it occurs when the freely falling proton
encounters very large curvature invariants as the singularity is approached.
From the proton’s viewpoint, there is nothing that would stimulate it to
decay before that.
On the other hand, from the vantage point of the external observer,
the proton encounters enormously high temperatures as it approaches the
horizon. Temperatures higher than M
X
can certainly excite the proton to
decay. So the external observer will conclude that baryon violation can
take place at the horizon. Who is right?
The answer that black hole complementarity implies is that they are
both right. On the face of it, that sounds absurd. Surely the event of
proton decay takes place in some deﬁnite place. For a very large black
hole, the time along the proton trajectory between horizon crossing and
the singularity can be enormous. Thus, it is diﬃcult to understand how
there can be an ambiguity.
The real proton propagating through spacetime is not the simple struc
tureless bare proton. The interactions cause it to make virtual transitions
from the bare proton to a state with an Xboson and a positron. The com
plicated history of the proton is described by Feynman diagrams such as
shown in Figure 9.6. The diagrams make evident that the real proton is
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Puzzle of Information Conservation in Black Hole Environments 91
p
p
X
e
+
p p
X
X
e
+
e
+
e
+
Fig. 9.6 Proton virtual ﬂuctuation into X e
+
pair
a superposition of states with diﬀerent baryon number; in the particular
processes shown in Figure 9.6, the intermediate state has vanishing baryon
number.
Nothing about virtual baryon nonconservation is especially surprising.
As long as the Xboson is suﬃciently massive, the rate for real proton decay
will be very small, and the proton will be stable for long times.
What is surprising is that the probability to ﬁnd the proton in a con
ﬁguration with vanishing baryon number is not small. This probability is
closely related to the wave function renormalization constant of the proton,
and is of the order
Probability ∼
g
2
4π
log
κ
M
X
where κ is the cutoﬀ in the ﬁeld theory. For example, for g ∼ 1, κ of the
order of the Planck mass, and M
X
of the order 10
16
GeV, the probability
that the proton has the “wrong” baryon number is of order unity. This
might seem paradoxical, since the proton is so stable.
The resolution of the paradox is that the proton is continuously making
extremely rapid transitions between baryon number states. The transitions
take place on a time scale of order δt ∼
1
M
X
. Ordinary observations of the
proton do not see these very rapid ﬂuctuations. The quantity that we
normally call baryon number is really the time averaged baryon number
normalized to unity for the proton.
Now let us consider a proton passing through a horizon, as shown in
Figure 9.7. Since the probability that the proton is actually an X, e
+
system is not small, it is not unlikely that when it passes the horizon,
its instantaneous baryon number is zero. But it is clear from Figure 9.7
that from the viewpoint of an external observer, this is not a shortlived
intermediate state. A ﬂuctuation that is much too rapid to be seen by a low
energy observer falling with the proton appears to be a real proton decay
lasting to eternity from outside the horizon.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
92 Black Holes, Information, and the String Theory Revolution
Time resolution
becomes very dense
∆
t
∆t
∆t
∆t
∆t
∆t
∆t
t=0
Fig. 9.7 Proton ﬂuctuations while falling through horizon
This, of course, is nothing but the usual time dilation near the horizon.
As a proton or any other system approaches the horizon, internal oscilla
tions or ﬂuctuations appear to indeﬁnitely slow down, so that a shortlived
virtual ﬂuctuation becomes stretched out into a real process.
The process of baryon violation near the horizon should not be totally
surprising to the external observer. To him, the proton is falling into a
region of increasing temperature. At a proper distance M
−1
X
from the
horizon, the temperature becomes of order M
X
. Baryon violating processes
are expected at that temperature.
An interesting question is whether an observer falling with the proton
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Puzzle of Information Conservation in Black Hole Environments 93
can observe the baryon number just before crossing the horizon, and send
a message to the outside world that the proton has not decayed.
The answer is interesting. In order to make an observation while the
proton is in a region of temperature M
X
the observer must do so very
quickly. In the proton’s frame, the time spent in the hot region before
crossing the horizon is
1
M
X
. Thus, to obtain a single bit of information
about the state of the proton, the observer has to probe it with at least one
quantum with energy of order M
X
. But such an interaction between the
proton and the probe quantum is at high enough energy that it can cause a
baryon violating interaction even from the perspective of the proton’s frame
of reference. Thus the observer cannot measure and report the absence of
baryon violation at the horizon without himself causing it.
It is evident from this example that the key to understanding the enor
mous discrepancies in the complementary description of events lies in un
derstanding the ﬂuctuations of matter at very high frequencies; frequencies
so high that the ordinary low energy observer has no chance of detecting
them.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
94 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 10
Horizons and the UV/IR Connection
The overriding theme of 20th century physics was the inverse relation
between size and momentum/energy. According to conventional relativistic
quantum mechanics, a size ∆x can be probed with a quantum of energy
E
∆x
≈
c
∆x
(10.0.1)
But we already know that this trend is destined to be reversed in the physics
of the 21st century. This can be seen in many ways. Let’s begin with a
traditional attempt to study interactions at length scales smaller than the
Planck scale. According to conventional thinking, what we need to do is to
collide, head on, particles with center of mass energy E
∆x
. We expect to
discover high energy collision products ﬂying out at all angles. By analyzing
the highest energy fragments, we hope to reconstruct very short distance
events.
The problem with this thinking is that at energies far above the Planck
mass, the collision will create a black hole of mass ∼ E
∆x
. The interesting
short distance eﬀects that we want to probe will be hidden behind a horizon
of radius
R
S
≈
2G
c
2
E
∆x
and are inaccessible. To make matters worse, the products of collision will
not be high energy particles, but rather low energy Hawking radiation.
The typical Hawking particle has energy ∼
c
R
S
which decreases with the
incident energy. Thus we see that a giant “Super Plancketron” Collider
(SPC) would fail to discover fundamental length scales smaller than the
Planck scale
P
, no matter how high the energy. In fact, as the energy
95
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
96 Black Holes, Information, and the String Theory Revolution
increases we would be probing ever larger scales
∆x ∼
2G
c
2
E
c
2
. (10.0.2)
We might express this in another form as a kind of spacetime uncertainty
principle:
∆x∆t ∼
2G
c
4
≈
2
P
(10.0.3)
This is the simplest example of the ultraviolet/infrared connection that will
control the relation between frequency and spatial size. Very high frequency
is related to large size scale.
The UV/IR connection is deeply connected to black hole complemen
tarity. As we saw in the previous lecture, the enormous diﬀerences in the
complementary descriptions of matter falling into a black hole are due to the
very diﬀerent time resolutions available to the complementary observers.
Let’s consider further: what happens to a proton falling into the black
hole? The proton carries some information with it; its charge, particle type,
momentum, location on the horizon through which it falls, etc. From the
viewpoint of the external observer, the proton is falling into an increasingly
hot region. The proton is like a tiny piece of ice thrown into a tub of very
hot water. The only reasonable expectation is that the constituents of the
proton “melt” and diﬀuse throughout the horizon. In fact, in Chapter 7, we
saw just such a phenomenon involving a charge falling onto the stretched
horizon. More generally, all of the information stored in the incident proton
should quickly be spread over the horizon. On the other hand, the observer
who follows the proton does not see it spread at all as it falls.
The need to reconcile the complementary descriptions gives us an im
portant clue to the behavior of matter at high frequencies. Following the
proton as it falls, the amount of proper time that it has before crossing the
horizon tends to zero as the Schwarzschild time tends to inﬁnity. Call the
proper time ∆τ. Then ∆τ varies with Schwarzschild time like
∆τ ∼
√
8MGδR e
−
∆t
4MG
. (10.0.4)
Thus in order to observer the proton before it crosses the horizon, we have
to do it in a time which is exponentially small as t → ∞. Thus, what we
need in order to be consistent with the thermal spreading of information is
now clear. As the proton is observed over shorter and shorter time intervals,
the region over which it is localized must grow.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Horizons and the UV/IR Connection 97
In fact, if we assume that it grows consistently with the uncertainty
relation in equation 10.0.3 (substituting ∆τ for ∆t) we obtain
∆x ∼
2
P
∆τ
∼
2
P
M G
e
t/4MG
Thus, if the proton size depends on the time resolution according to equa
tion 10.0.3, it will spread over the horizon exponentially fast.
The idea of information spreading as the resolving time goes to zero is
very unfamiliar, but it is at the heart of complementarity. It is implicit in
the modern idea of the UV/IR connection. Fortunately it is also built into
the mathematical framework of string theory.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
98 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
PART 2
Entropy Bounds and Holography
99
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
100
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 11
Entropy Bounds
11.1 Maximum Entropy
Quantum ﬁeld theory has too many degrees of freedom to consistently
describe a gravitational theory. The main indication that we have seen
of this overabundance of degrees of freedom is the fact that the horizon
entropydensity is inﬁnite in quantum ﬁeld theory. The divergence arises
from modes very close to the horizon. One might think that this is just an
indication that a more or less conventional ultraviolet regulator is needed
to render the theory consistent. But the divergent horizon entropy is not
an ordinary ultraviolet phenomenon. The modes which account for the
divergence are very close to the horizon and would appear to be ultra short
distance modes. But they also carry very small Rindler energy and therefore
correspond to very long times to the external observer. This is an example
of the weirdness of the Ultraviolet/Infrared connection.
A quantitative measure of the overabundance of degrees of freedom in
QFT is provided by the Holographic Principle. This principle says that
there are vastly fewer degrees of freedom in quantum gravity than in any
QFT even if the QFT is regulated as, for example, it would be in lattice
ﬁeld theories.
The Holographic Principle is about the counting of quantum states of a
system. We begin by considering a large region of space Γ. For simplicity
we take the region to be a sphere. Now consider the space of states that
describe arbitrary systems that can ﬁt into Γ such that the region outside
Γ is empty space. Our goal is to determine the dimensionality of that
statespace. Let us consider some preliminary examples.
Suppose we are dealing with a lattice of discrete spins. Let the lattice
spacing be a and the volume of Γ be V . The number of spins is then V/a
3
101
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
102 Black Holes, Information, and the String Theory Revolution
and the number of orthogonal states supported in Γ is given by
N
states
= 2
V/a
3
(11.1.1)
A second example is a continuum quantum ﬁeld theory. In this case the
number of quantum states will diverge for obvious reasons. We can limit
the states, for example by requiring the energy density to be no larger than
some bound ρ
max
. In this case the states can be counted using some con
cepts from thermodynamics. One begins by computing the thermodynamic
entropy density s as a function of the energy density ρ. The total entropy
is
S = s(ρ)V (11.1.2)
The total number of states is of order
N
states
∼ exp S = exp s(ρ
max
)V (11.1.3)
In each case the number of distinct states is exponential in the volume V .
This is a very general property of conventional local systems and represents
the fact that the number of independent degrees of freedom is additive in
the volume.
In counting the states of a system the entropy plays a central role. In
general entropy is not really a property of a given system but also involves
one’s state of knowledge of the system. To deﬁne entropy we begin with
some restrictions that express what we know, for example, the energy within
certain limits, the angular momentum and whatever else we may know. The
entropy is essentially the logarithm of the number of quantum states that
satisfy the given restrictions.
There is another concept that we will call the maximum entropy. This
is a property of the system. It is the logarithm of the total number of
states. In other words it is the entropy given that we know nothing about
the state of the system. For the spin system the maximum entropy is
S
max
=
V
a
3
log 2 (11.1.4)
This is typical of the maximum entropy. Whenever it exists it is propor
tional to the volume. More precisely it is proportional to the number of
simple degrees of freedom that it takes to describe the system.
Let us now consider a system that includes gravity. For deﬁniteness we
will take spacetime to be fourdimensional. Again we focus on a spherical
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 103
region of space Γ with a boundary ∂Γ. The area of the boundary is A. Sup
pose we have a thermodynamic system with entropy S that is completely
contained within Γ. The total mass of this system can not exceed the mass
of a black hole of area A or else it will be bigger than the region.
Now imagine collapsing a spherically symmetric lightlike shell of matter
with just the right amount of energy so that together with the original mass
it forms a black hole which just ﬁlls the region. In other words the area
of the horizon of the black hole is A. This is shown in Figure 11.1. The
R
Γ
E
M
E g
Fig. 11.1 Inmoving (zero entropy) spherical shell of photons
result of this process is a system of known entropy, S = A/4G. But now
we can use the second law of thermodynamics to tell us that the original
entropy inside Γ had to be less than or equal to A/4G. In other words the
maximum entropy of a region of space is proportional to its area measured
in Planck units. Thus we see a radical diﬀerence between the number of
states of any (regulated) quantum ﬁeld theory and a theory that includes
gravity.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
104 Black Holes, Information, and the String Theory Revolution
Spacetime depiction of horizon formation
Consider the collapsing spherically symmetric shell of lightlike energy de
picted in Figure 11.1. As the photonic shell approaches the center, the
horizon forms prior to the actual crossing of the shell. Inside of the shell,
the geometry is Schwarzschild with low curvature (no black hole) prior to
the shell crossing the Schwarzschild radius. However, just outside of the
shell the geometry becomes increasingly curved as the Schwarzschild radius
is approached (see Figure 11.2). The horizon grows until the collapsing
R
Low
curvature
Infalling
Shell E
Horizon
is a
lightlike
surface
Light beams get more
vertical as the energy
becomes more concentrated
γ
Fig. 11.2 Spacetime depiction of radially inmoving shell of photons
shell crosses, and a singularity forms at a later time. The energy of the
infalling photonic shell E
γ
has been tuned such that the collapsing shell
crosses the horizon exactly at the radius R in Figure 11.1. However, at
that time the system winds up with entropy
S =
A
4G
.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 105
Therefore, unless the second law of thermodynamics is untrue, the entropy
of any system is limited by
S
max
≤
A
4G
. (11.1.5)
The coarse grained volume in phase space cannot decrease, so this “holo
graphic” limit must be satisﬁed.
Aside: Scale of Entropy Limit
Example: To get some idea of how big a typical system must be in
order to saturate the maximum entropy, consider thermal radiation at
a temperature of 1000
◦
K, which corresponds to photons of wavelength
∼10
−5
cm. The number of photons N
γ
in a volume of radius R satis
ﬁes N
γ
∼
V
λ
3
∼
R(cm) ⊗10
5
3
. Entropy is proportional to the number of
photons, and thus one expects
S ∝
R(cm) ⊗10
5
3
.
Compare this with the maximum entropy calculated using the holographic
limit
S
max
≈
R
l
Planck
2
∼
=
R(cm) ⊗10
33
2
. (11.1.6)
Evidently the maximum entropy will only be saturated for the photon gas
when the radius is huge, R ∼ 10
51
cm, considerable larger than the ob
servable universe 10
28
cm.
11.2 Entropy on Lightlike Surfaces
So far we have considered the entropy that passes through spacelike
surfaces. We will see that it is most natural to deﬁne holographic entropy
bounds on lightlike surfaces
[
3
]
as opposed to spacelike surfaces. Under
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
106 Black Holes, Information, and the String Theory Revolution
certain circumstances the entropy bounds of lightlike surfaces can be trans
lated to spacelike surfaces, but not always. The case described above is
one of those cases where a spacelike bound is derivable.
Let us start with an example in asymptotically ﬂat spacetime. We as
sume that ﬂat Minkowski coordinates X
+
, X
−
, x
i
can be deﬁned at asymp
totic distances. In this chapter we will revert to the usual convention in
which X
+
is used as a light cone time variable. We will now deﬁne a “light
sheet”. Consider the set of all light rays which lie in the surface X
+
= X
+
0
in the limit X
−
→ +∞. In ordinary ﬂat space this congruence of rays
deﬁnes a ﬂat threedimensional lightlike surface. In general, they deﬁne
a lightlike surface called a light sheet. The light sheet will typically have
singular caustic lines, but can be deﬁned in a unique way
[
4
]
. When we vary
X
+
0
the light sheets ﬁll all spacetime except for those points that lie behind
black hole horizons.
Now consider a spacetime point p. We will assign it light cone coor
dinates as follows. If it lies on the light sheet X
+
0
we assign it the value
X
+
= X
+
0
. Also if it lies on the light ray which asymptotically has trans
verse coordinate x
i
0
we assign it x
i
= x
i
0
. The value of X
−
that we assign
will not matter. The twodimensional x
i
plane is called the Screen. Next
assume a black hole passes through the light sheet X
+
0
. The stretched
Incoming
Parallel
Light Rays
Far off screen
x
1
and x
2
plane
x
j
X

X
+
Fig. 11.3 Light propagating on lightlike surface X
+
= constant
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 107
X
+
Image of
Black hole
on screen
Asymptotic
screen
X

x
j
C
a
u
s
t
i
c
Black
Hole
Foliation
corresponds
to different
x
+
slices
Fig. 11.4 Family of light rays on ﬁxed X
+
surface in presence of black hole
horizon of the black hole describes a twodimensional surface in the three
dimensional light sheet as shown in Figure 11.4. Each point on the stretched
horizon has unique coordinates X
+
, x
i
, as seen in Figure 11.5. More gen
erally if there are several black holes passing through the light sheet we
can map each of their stretched horizons to the screen in a single valued
manner.
Since the entropy of the black hole is equal to 1/4G times the area of the
horizon we can deﬁne an entropy density of 1/4G on the stretched horizon.
The mapping to the screen then deﬁnes an entropy density in the x
i
plane,
σ(x). It is a remarkable fact that σ(x) is always less than or equal to 1/4G.
To prove that σ(x) ≤
1
4G
we make use of the focusing theorem of general
relativity. The focusing theorem depends on the positivity of energy and is
based on the tendency for light to bend around regions of nonzero energy.
Consider a bundle of light rays with cross sectional area α. The light rays
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
108 Black Holes, Information, and the String Theory Revolution
Black
hole
Stretched
horizon
Map of
stretched
horizon
Patch on
stretched
horizon
Map of patch
on stretched
horizon
Fig. 11.5 Image of “stretched horizon” on asymptotic screen
are parameterized by an aﬃne parameter λ. The focusing theorem says
that
d
2
α
dλ
2
≤ 0 (11.2.7)
Consider a bundle of light rays in the light sheet which begin on the
stretched horizon and go oﬀ to X
−
= ∞. Since the light rays deﬁning the
light sheet are parallel in the asymptotic region dα/dλ → 0. The focusing
theorem tells us that as we work back toward the horizon, the area of the
bundle decreases. It follows that the image of a patch of horizon on the
screen is larger than the patch itself. The holographic bound immediately
follows.
σ(x) ≤
1
4G
(11.2.8)
This is a surprising conclusion. No matter how we distribute the black
holes in threedimensional space, the image of the entropy on the screen
always satisﬁes the entropy bound equation 11.2.8. An example which
helps clarify how this happens involves two black holes. Suppose we try to
hide one of them behind the other along the X
−
axis, thus doubling the
entropy density in the x plane. The bending and focusing of light always
acts as in Figures 11.6 to prevent σ(x) from exceeding the bound. These
considerations lead us to the more general conjecture that for any system,
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 109
Motion of
second black
hole
CAUSTIC
Image of
second black
hole
Motion of
second black
hole
CAUSTIC
Image of
second black
hole
Fig. 11.6 Initial and later motions and images of second black hole
when it is mapped to the screen the entropy density obeys the bound in
equation 11.2.8.
Thus far we have assumed asymptotically ﬂat boundary conditions.
This allowed us to choose the screen so that the light rays forming the
light sheet intersect the screen at right angles. Equivalently da/dλ equals
zero at the screen. We note for future use that the conclusions concerning
the entropy bound would be unchanged if we allowed screens for which the
light rays were diverging as we move outward, i.e. da/dλ > 0. However,
if we attempt to use screens for which the light rays are converging then
the argument fails. This will play an important role in generalizing the
holographic bound to more general geometries.
Aside: Apparent motions
Consider a single point particle external to the black hole undergoing
motions near a caustic. Examine the projection of those motion upon the
screen, demonstrated in Figures 11.7. One sees that due to gravitational
lensing, the image of the particle can move at arbitrarily large speeds!
Small
motions near
caustic
CAUSTIC
produce large, rapid
motions on the screen
CAUSTIC
Fig. 11.7 Initial and later path and image during “slow” motion near caustic
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
110 Black Holes, Information, and the String Theory Revolution
11.3 Friedman Robertson Walker Geometry
The holographic bound can be generalized to ﬂat F.R.W. geometries,
where it is called the Fischler–Susskind (FS) bound
[
5
]
and to more general
geometries by Bousso
[
6
]
. First we will review the F.R.W. case. Consider
the general case of d + 1 dimensions. The metric has the form
dτ
2
= dt
2
−a(t)
2
dx
m
dx
m
(11.3.9)
where the index m runs over the d spatial directions. The function a(t) is
assumed to grow as a power of t.
a(t) = a
0
t
p
(11.3.10)
Let’s also make the usual simplifying cosmological assumptions of homo
geneity. In particular we assume that the spatial entropy density (per unit
d volume) is homogeneous. Later, we will relax these assumptions.
At time t we consider a spherical region Γ of volume V and area A.
The boundary (d − 1)sphere, ∂Γ, will play the same role as the screen in
the previous discussion. The light sheet is now deﬁned by the backward
light cone formed by light rays that propagate from ∂Γ into the past (See
Figure 11.8).
t
Screen
Past
light
cone
t>0
Projection from
screen onto
light cone
Fig. 11.8 Holographic surface for calculating entropy bound with a spherical sur
face as the screen
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 111
As in the previous case the holographic bound applies to the entropy
passing through the light sheet. This bound states that the total entropy
passing through the light sheet does not exceed A/4G. The key to a proof
is again the focusing theorem. We observe that at the screen the area of the
outgoing bundle of light rays is increasing as we go to later times. In other
words the light sheet has positive expansion into the future and negative
expansion into the past. The focusing theorem again tells us that if we map
the entropy of black holes passing through the light sheet to the screen, the
resulting density satisﬁes the holographic bound. It is believed that the
bound is very general.
It is now easy to see why we concentrate on light sheets instead of space
like surfaces. Obviously if the spatial entropy density is uniform and we
choose Γ big enough, the entropy will exceed the area. However if Γ is
larger than the particle horizon at time t the light sheet is not a cone, but
rather a truncated cone as in Figure 11.9, which is cut oﬀ by the big bang at
t = 0. Thus a portion of the entropy present at time t never passed through
the light sheet. If we only count that portion of the entropy which did pass
through the light sheet, it will scale like the area A. We will return to the
question of spacelike bounds after discussing Bousso’s generalization
[
6
]
of
the FS bound.
t
Sphere
Entropy not
bounded
Past
light
cone
Big
Bang
t=0
Entropy flux through
light cone bounded
Fig. 11.9 Region of space causally connected to particle horizon
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
112 Black Holes, Information, and the String Theory Revolution
Test: Is the observed horizon entropy bounded by its area?
We will check whether the observed particle horizon satisﬁes the entropy
bound:
S
Horizon
≤
A
Horizon
4 G
???
We can check by recognizing that the entropy is primarily given by the
number of black body cosmic background photons, N
γ
≈ 10
90
. The proper
size of the horizon is approximately given by 10
18
(light) seconds, and the
Planck time is approximately 10
−43
seconds. This gives a proper size for the
horizon of about 10
61
Planck units. We can use these numbers to compare
the cosmic entropy with the area of the horizon:
S
Horizon
≤
??? A
Horizon
4 G
10
90
≤
10
61
2
= 10
122
We ﬁnd this inequality to deﬁnitely be true today.
Next, using the F.R.W. geometry, we will determine if the entropy per
horizonarea contained within the particle horizon is increasing or decreas
ing. Let R
Horizon
represent the coordinate size of the particle horizon
(Figure 11.10), and d represent the number of spatial dimensions. Let σ be
the entropy volume density, so that
S ∼ σ R
d
Horizon
dτ
2
= dt
2
−a
2
(t)
d
j=1
dx
j
dx
j
This means that the proper size of the horizon is given by a(t)R
Horizon
. We
want to check whether
S ∼ σ R
d
Horizon
<
?
(a(t) R
Horizon
)
d−1
4 G
(11.3.11)
An outgoing light ray (null geodesic) which would generate the particle
horizon satisﬁes dt = a(t)dx, which gives the form for the time dependence
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 113
t
Past
light
cone
Entropy
flux
Big
Bang
t=0
R
Horizon
Fig. 11.10 Particle Horizon
of the size of the particle horizon:
R
Horizon
(t) =
t
0
dt
a(t
)
(11.3.12)
If we assume the form a(t) = a
o
t
p
then the particle horizon evolves accord
ing to the formula
R
Horizon
(t) =
t
1−p
a
o
. (11.3.13)
Therefore for the entropy bound to continue to be valid, the time de
pendence must satisfy t
(1−p)d
< t
d−1
. This bounds the expansion rate
coeﬃcient to satisfy
p >
1
d
. (11.3.14)
One sees that if the expansion rate is too slow, then the coordinate volume
will grow faster than the area, and the entropy bound will eventually be
contradicted.
Suppose that the matter in the F.R.W. cosmology satisﬁes the equation
of state
P = w u (11.3.15)
where P is the pressure and u is the energy density and w is a constant.
Given w and the scale factor for the expansion p, one can use the Einstein
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
114 Black Holes, Information, and the String Theory Revolution
ﬁeld equation to calculate a relationship between them:
p =
2
d (1 + w)
>
1
d
(11.3.16)
We see that the number of spatial dimensions cancels and that the entropy
bound is satisﬁed so long as
w ≤ 1. (11.3.17)
This is an interesting result. Recall that the speed of sound within a
medium is given by
v
2
s
=
∂P
∂u
= w
Therefore, in the future, the bound will always be satisﬁed, since the speed
of sound is always less than the speed of light. The relation satisﬁed by
the scale factor v
2
s
= w ≤ 1 is just the usual causality requirement. As one
moves forward in time, the entropy bound then becomes more satisﬁed,
not less.
Next, go back in time using the black body radiation background as
the dominant entropy. Using a decoupling time t
decoupling
∼ 10
5
years
(when the background radiation fell out of equilibrium with the matter)
and extrapolating back using the previously calculated entropy relative to
the bound, one gets
S
A/(4G)
= 10
−28
t
decoupling
t
1
2
(11.3.18)
The entropy bound S =
A
4G
is reached when
t
decoupling
t
1
2
= 10
28
⇒ t =
t
decoupling
10
56
∼ 10
−44
sec (11.3.19)
This time is comparable to the Planck time (by coincidence??). Therefore
the entropy bound is not exceeded after the Planck time.
11.4 Bousso’s Generalization
Consider an arbitrary cosmology. Take a spacelike region Γ bounded
by the spacelike boundary ∂Γ. At any point on the boundary we can con
struct four light rays that are perpendicular to the boundary
[
6
]
. We will
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 115
call these the four branches. Two branches go toward the future. One of
them is composed of outgoing rays and the other is ingoing. Similarly two
branches go to the past. On any of these branches a light ray, together with
its neighbors deﬁne a positive or negative expansion as we move away from
the boundary. In ordinary ﬂat spacetime, if ∂Γ is convex the outgoing
(ingoing) rays have positive (negative) expansion. However in nonstatic
universes other combinations are possible. For example in a rapidly con
tracting universe both future branches have negative expansion while the
past branches have positive expansion.
If we consider general boundaries the sign of the expansion of a given
branch may vary as we move over the surface. For simplicity we restrict
attention to those regions for which a given branch has a unique sign. We
can now state Bousso’s rule: From the boundary ∂Γ construct all light
sheets which have negative expansion as we move away. These light sheets
may terminate at the tip of a cone or a caustic or even a boundary of the
geometry. Bousso’s bound states that the entropy passing through these light
sheets is less than A/4G where A is the boundary of ∂Γ.
To help visualize how Bousso’s construction works we will consider
spherically symmetric geometries and use Penrose diagrams to describe
them. The Penrose diagram represents the radial and time directions.
Each point of such a diagram really stands for a 2sphere (more gener
ally a (d − 1)sphere). The four branches at a given point on the Penrose
diagram are represented by a pair of 45 degree lines passing through that
point. However we are only interested in the branches of negative expan
sion. For example in Figure 11.11 we illustrate ﬂat spacetime and the
negative expansion branches of a typical local 2sphere. In general as we
move around in the Penrose diagram the particular branches which have
negative expansion may change. For example if the cosmology initially ex
pands and then collapses, the outgoing future branch will go from positive
to negative expansion. Bousso introduced a notation to indicate this. The
Penrose diagram is divided into a number of regions depending on which
branches have negative expansion. In each region the negative expansion
branches are indicated by their directions at a typical point. Thus in Figure
11.12 we draw the Penrose–Bousso (PB) diagram for a positive curvature,
matter dominated universe that begins with a bang and ends with a crunch.
It consists of four distinct regions.
In Region I of Figure 11.12 the expansion of the universe causes both
past branches to have negative expansion. Thus we draw light surfaces
into the past. These light surfaces terminate on the initial boundary of the
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
116 Black Holes, Information, and the String Theory Revolution
Fig. 11.11 Negative expansion branches of 2sphere in ﬂat spacetime
geometry and are similar to the truncated cones that we discussed in the
ﬂat F.R.W. case. The holographic bound in this case says that the entropy
passing through either backward light surface is bounded by the area of the
2sphere at point p. Bousso’s rule tells us nothing in this case about the
entropy on spacelike surfaces bounded by p.
Now move on to Region II. The relevant light sheets in this region be
gin on the 2sphere q and both terminate at the spatial origin. These
are untruncated cones and the entropy on both of them is holographically
bounded. There is something new in this case. We ﬁnd that the entropy is
bounded on a future light sheet. Now consider a spacelike surface bounded
by q and extending to the spatial origin (shown in Figure 11.13). It is ev
ident that any matter which passes through the spacelike surface must
also pass through the future light sheet. By the second law of thermody
namics the entropy on the spacelike surface can not exceed the entropy on
the future light sheet. Thus in this case the entropy in a spacelike region
can be holographically bounded. Therefore, one condition for a spacelike
bound is that the entropy is bounded by a corresponding future light sheet.
With this in mind we return to Region I. For Region I there is no future
bound, and therefore the entropy is not bounded on spacelike regions with
boundary p.
In Region III the entropy bounds are both on future light sheets. Nev
ertheless there is no spacelike bound. The reason is that not all matter
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 117
TOWARDS
r = 0
TOWARDS
LARGE
RADIUS
t = 0
t = t
final
CONTRACTION
I
II
III
IV
EXPANSION
Point
corresponds
to finite area
of sphere
Backwards
lightcone
Forward
light cone
Decreasing
area
Backwards
light cone
Light cone
intercepts
singularity
Future
directed
p
q
Fig. 11.12 Penrose–Bousso diagram for matter dominated universe
which pass through spacelike surfaces are forced to pass through the future
light sheets.
Region IV is identical to Region II with the spatial origin being replaced
by the diametrically opposed antipode. Thus we see that there are lightlike
bounds in all four regions but only in II and IV are there holographic bounds
on spacelike regions. (See Figure 11.13.) Figure 11.14 demonstrates a
region that does not satisfy an entropy bound for this cosmology.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
118 Black Holes, Information, and the String Theory Revolution
Entropy on this space
q
like surface is also
bounded by flux
through lightlike
surface
Fig. 11.13 Bounding surfaces for inﬂowing entropy
t = 0
t
final
Area is INCREASING
in part of this lightlike
surface. The entropy
thru such a surface is
NOT bounded in a
Robertson–Walker
cosmology
Fig. 11.14 Lightlike surface which does not satisfy entropy bound
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 119
11.5 de Sitter Cosmology
de Sitter space holds special interest because of its close connection
with observational cosmology. The model of cosmology which is presently
gaining the status of a “standard model” involves de Sitter space in two
ways. First it is believed that the early universe underwent an epoch of
rapid inﬂation during which the spacetime geometry was very close to de
Sitter. The inﬂationary theory is more than two decades old and by now
has been well tested.
More recently de Sitter space has entered cosmology as the likely candi
date for the ﬁnal fate of the universe. The history of the universe seems to
be a transition from an early de Sitter epoch in which the vacuum energy or
cosmological constant was very large to a late de Sitter phase characterized
by a very small but not zero cosmological constant.
de Sitter space is the solution of Einstein’s ﬁeld equations with a positive
cosmological constant that exhibits maximal symmetry. Fourdimensional
de Sitter space may be deﬁned by embedding it in (4 + 1) dimensional ﬂat
Minkowski space. It is the hyperboloid given by
4
i=1
(x
i
)
2
−(x
0
)
2
= R
2
. (11.5.20)
The radius of curvature R is related to the cosmological constant, λ.
R
2
=
3
Gλ
(11.5.21)
de Sitter space can also be written in the form
dτ
2
= dt
2
−a(t)
2
dΩ
2
3
(11.5.22)
where dΩ
2
3
is the metric of a unit 3sphere, and the scale factor a is given
by
a(t) = Rcosh(t/R). (11.5.23)
For our purposes we want to put this in a form that will allow us to
read oﬀ the Penrose diagram. To that end deﬁne the conformal time T by
dT = dt/a(t). (11.5.24)
One easily ﬁnds that the geometry has the form
dτ
2
= (a(t))
2
(dT
2
−dΩ
2
3
). (11.5.25)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
120 Black Holes, Information, and the String Theory Revolution
π
0
0
−π
Fig. 11.15 Penrose diagram for de Sitter spacetime
S
Fig. 11.16 Extremal spacelike surface in de Sitter spacetime
Furthermore when t varies between ±∞, the conformal time (T =
tan
−1
(tanh(t/R)) −
π
2
) varies from −π to 0. Since the polar angle on
the sphere also varies from 0 to π the Penrose diagram is a square as in
Figure 11.15.
Once again the Bousso construction divides the Penrose diagram into
four quadrants. However, the contracting light sheets in the upper and
lower quadrants are oriented oppositely to the usual F.R.W. case. The
reason is that the geometry rapidly expands as we move toward the upper
and lower boundary of the de Sitter space.
Of particular interest is the bound on a spacelike surface beginning
at the center of the Penrose diagram and extending to the left edge as
in Figure 11.16. We leave it as an exercise to show that the entropy is
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 121
t
Period of inflation
Postinflation
particle horizon
Larger area
Fig. 11.17 Inﬂationary universe
bounded by the area of the 2sphere at the center of the diagram. In fact
the maximum entropy on any such surface is given by
S =
4πR
2
4G
. (11.5.26)
de Sitter space has a special importance because of its role during the
early evolution of the universe. According to the inﬂationary hypothesis,
the universe began with a large vacuum energy which mimicked the eﬀects
of a positive cosmological constant. During that period the geometry was de
Sitter space. But then at some time the vacuum energy began to decrease
and the universe made a transition to an F.R.W. universe. The transition
was accompanied by the production of a large amount of entropy, and is
called reheating.
Inﬂationary cosmology is illustrated in Figure 11.17. In order to con
struct the Penrose–Bousso diagram we begin by drawing Penrose diagrams
for both de Sitter space and either radiation or matter dominated F.R.W.
The Penrose–Bousso diagram for de Sitter space is shown in Figure 11.18.
In order to describe inﬂationary cosmology we must terminate the de Sitter
space at some late time and attach it to a conventional F.R.W. space as in
Figure 11.19. The dotted line where the two geometries are joined is the
reheating surface where the entropy of the universe is created.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
122 Black Holes, Information, and the String Theory Revolution
Point corresponds to
finite spherical area
Fig. 11.18 Buosso wedges for de Sitter geometry
EXPANDING
Join
de Sitter
Space
Reheating: Entropy is created
during reheating. There is no
entropy prior.
Robertson–Walker
Space
+
t
inflation
Fig. 11.19 Joining of inﬂationary and postinﬂationary geometries
Let us focus on the point p in Figure 11.20. It is easy to see that in
an ordinary inﬂationary cosmology p can be chosen so that the entropy on
the spacelike surface p −q is bigger than the area of p. However Bousso’s
rule applied to point p only bounds the entropy on the past light sheet.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 123
Robertson–
Walker
Inflation
Entropy flux
only after
reheating
No entropy
flux here
REHEATING
p
q
Backwards
light cone
Fig. 11.20 Entropy bound on spherical region causally connected to inﬂationary
period
In this case most of the newly formed entropy on the reheating surface is
not counted since it never passed through the past light sheet. Typical
inﬂationary cosmologies can be studied to see that the past light sheet
bound is not violated.
11.6 Anti de Sitter Space
de Sitter space is important because it may describe the early and late
time behavior of the real universe. Anti de Sitter space is important for
an entirely diﬀerent reason. It is the background in which the holographic
principle is best understood. In Chapter 12 the geometry of AdS will be
reviewed, but for our present purpose all we need are the properties of its
PB diagram.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
124 Black Holes, Information, and the String Theory Revolution
S
p
L
ANY spacelike surface S will have
its entropy bounded by the area
B
o
u
n
d
a
r
y
r=0
Fig. 11.21 Static surface of large area in AdS space
AdS space is the vacuum of theories with a negative cosmological con
stant. The spacetime in appropriate coordinates is static but unlike de
Sitter space the static coordinates cover the entire space. The vacuum of
AdS is a genuine zero temperature state.
The PB diagram consists of an inﬁnite strip bounded on the left by the
spatial origin and on the right by a boundary. The PB diagram consists of a
single region in which both negative expansion light sheets point toward the
origin. Let us consider a static surface of large area A far from the spatial
origin. The surface is denoted by the dotted vertical line L in Figure 11.21.
We will think of L as an infrared cutoﬀ. Consider an arbitrary point p on L.
Evidently Bousso’s rules bound the entropy on past and future light sheets
bounded by p. Therefore the entropy on any spacelike surface bounded by
p and including the origin is also holographically bounded. In other words
the entire region to the left of L can be foliated with spacelike surfaces
such that the maximum entropy on each surface is A/4G.
AdS space is an example of a special class of geometries which have
timelike Killing vectors and which can be foliated by spacelike surfaces
that satisfy the Holographic bound. These two properties imply a very far
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy Bounds 125
reaching conclusion. All physics taking place in such backgrounds (in the
interior of the infrared cutoﬀ L) must be described in terms of a Hamiltonian
that acts in a Hilbert space of dimensionality
N
states
= exp(A/4G) (11.6.27)
The holographic description of AdS space is the subject of the next chapter.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
126 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 12
The Holographic Principle and Anti
de Sitter Space
12.1 The Holographic Principle
As we have seen in Chapter 11, the number of possible quantum states
in a region of ﬂat space is bounded by the exponential of the area of the
region in Planck units. That fact together with the Ultraviolet/Infrared
connection and Black Hole Complementarity has led physics to an entirely
new paradigm about the nature of space, time and locality. One of the
elements of this paradigm is the Holographic Principle and its embodiment
in AdS space.
Let us consider a region of ﬂat space Γ. We have seen that the maximum
entropy of all physical systems that can ﬁt in Γ is proportional to the area
of the boundary ∂Γ, measured in Planck units. Typically, as in the case
of a lattice of spins, the maximum entropy is a measure of the number of
simple degrees of freedom
∗
that it takes to completely describe the region.
This is almost always proportional to the volume of Γ. The exception
is gravitational systems. The entropy bound tells us that the maximum
number of nonredundant degrees of freedom is proportional to the area.
For a large macroscopic region this is an enormous reduction in the required
degrees of freedom. In fact if the linear dimensions of the system is of
order L then the number of degrees of freedom per unit volume scales like
1/L in Planck units. By making L large enough we can make the degrees
of freedom arbitrarily sparse in space. Nevertheless we must be able to
describe microscopic processes taking place anywhere in the region. One
way to think of this is to imagine the degrees of freedom of Γ as living
on ∂Γ with an area density of no more than ∼ 1 degree of freedom per
∗
By a simple degree of freedom we mean something like a spin or the presence or absence
of a fermion. A simple degree of freedom represents a single bit of information.
127
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
128 Black Holes, Information, and the String Theory Revolution
Planck area. The analogy with a hologram is obvious; threedimensional
space described by a twodimensional hologram at its boundary! That this
is possible is called the Holographic Principle.
What we would ideally like to do is to have a solution of Einstein’s
equations that describes a ball of space with a spherical boundary and
then to count the number of degrees of freedom. Even better would be
to construct a description of the region in terms of a boundary theory
with a limited density of degrees of freedom. Ordinarily, it does not make
sense to consider a balllike region with a boundary in the general theory
of relativity. But there is one special situation which is naturally balllike.
It occurs when there is a negative cosmological constant; Anti de Sitter
space. Thus AdS is a natural framework in which to study the Holographic
Principle.
12.2 AdS Space
We saw in Chapter 11 that AdS space enjoys certain properties which
make it a natural candidate for a holographic Hamiltonian description. In
this lecture we will describe a very precise version of AdS “Holography”
which grew out of the mathematics of string theory. The remarkable pre
cision is due to the unusually high degree of symmetry of the theory which
includes a powerful version of supersymmetry. However we will downplay
the mathematical aspects of the theory and concentrate on those physical
principles which are likely to be general.
The particular space that we will be interested in is not simple 5
dimensional AdS but rather AdS(5)⊗S(5)
[
7
][
8
][
9
]
. This is a 10dimensional
product space consisting of two factors, the 5dimensional AdS and a
5sphere S(5). Why the S(5)? The reason involves the high degree of
supersymmetry enjoyed by superstring theory. Generally the kind of su
pergravity theories that emerge from string theory don’t have cosmological
constants. But by bending some of the directions of space into compact
manifolds it becomes possible to generate a cosmological constant for the
resulting lower dimensional Kaluza–Klein type theory. From a conceptual
point of view the extra internal 5sphere is not important. From a mathe
matical point of view it is essential if we want to be able to make precision
statements.
We will begin with a brief review of AdS geometry. For our purposes
5dimensional AdS space may be considered to be a solid 4dimensional
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Holographic Principle and Anti de Sitter Space 129
spatial ball times the inﬁnite time axis. The geometry can be described by
dimensionless coordinates t, r, Ω where t is time, r is the radial coordinate
(0 ≤ r < 1) and Ω parameterizes the unit 3sphere. The geometry has
uniform curvature R
−2
where R is the radius of curvature. The metric we
will use is
dτ
2
=
R
2
(1 −r
2
)
2
(1 + r
2
)
2
dt
2
−4dr
2
−4r
2
dΩ
2
(12.2.1)
There is another form of the metric which is in common use,
dτ
2
=
R
2
y
2
dt
2
−dx
i
dx
i
−dy
2
(12.2.2)
where i runs from 1 to 3.
The metric in equation 12.2.2 is related to 12.2.1 in two diﬀerent ways.
First of all it is an approximation to equation 12.2.1 in the vicinity of a
point on the boundary at r = 1. The 3sphere is replaced by the ﬂat
tangent plane parameterized by x
i
and the radial coordinate is replaced by
y, with y = (1 −r).
The second way that equations 12.2.1 and 12.2.2 are related is that
12.2.2 is the exact metric of an incomplete patch of AdS space. A timelike
geodesic can get to y = ∞ in a ﬁnite proper time so that the space in
equation 12.2.2 is not geodesically complete. It has a horizon at y = ∞.
When interpreted in this manner, time coordinates appearing in 12.2.1 and
12.2.2 are not the same.
The metric 12.2.2 may be expressed in terms of the coordinate z = 1/y.
dτ
2
= R
2
z
2
(dt
2
−dx
i
dx
i
) −
1
z
2
dz
2
(12.2.3)
In this form it is clear that there is a horizon at z = 0 since the time–time
component of the metric vanishes there. The boundary is at z = ∞.
To construct the space AdS(5) ⊗S(5) all we have to do is deﬁne 5 more
coordinates ω
5
describing the unit 5sphere and add a term to the metric
ds
2
5
= R
2
dω
2
5
(12.2.4)
Although the boundary of AdS is an inﬁnite proper distance from any
point in the interior of the ball, light can travel to the boundary and back
in a ﬁnite time. For example, it takes a total amount of (dimensionless)
time t = π for light to make a round trip from the origin at r = 0 to the
boundary at r = 1 and back. For all practical purposes AdS space behaves
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
130 Black Holes, Information, and the String Theory Revolution
like a ﬁnite cavity with reﬂecting walls. The size of the cavity is of order
R. In what follows we will think of the cavity size R as being much larger
than any microscopic scale such as the Planck or string scale.
Supplement on Properties of AdS metric
1) The point r = 0 is the center of the Anti de Sitter space and r varies from
0 to 1 in the space. A radial null geodesic satisﬁes
1 + r
2
2
dt
2
= 4dr
2
,
which means that a light beam will traverse the inﬁnite proper distance
in r from 0 to 1 back to 0 in a round trip time given by π, which makes
the space causally ﬁnite.
2) The metric is singular at r = 1 in all components. A unit coordinate
time interval corresponds to increasingly large proper time intervals.
3) Near r = 1, the metric is approximately conformal, which means that
light rays move at 45
◦
angles near the boundary.
ds
2
∼
=
4R
2
(1−r
2
)
2
−dt
2
+ dr
2
+ r
2
dΩ
2
D−2
Light rays move slower (by a factor of 2) near the center of the Anti de
Sitter space.
4) Generally, the spatial metric is that of a uniformly (negatively) curved
space, a hyperbolic plane (or the Poincar´e disk).
12.3 Holography in AdS Space
We will refer to AdS(5) ⊗S(5) as the bulk space and the 4dimensional
boundary of AdS at y = 0 as the boundary. According to the Holographic
Principle we should be able to describe everything in the bulk by a theory
whose degrees of freedom can be identiﬁed with the boundary at y = 0.
However the Holographic Principle requires more than that. It requires
that the boundary theory has no more than 1 degree of freedom per Planck
area. To see what this entails, let us compute the area of the boundary.
From equation 12.2.1 we see that the metric diverges near the boundary.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Holographic Principle and Anti de Sitter Space 131
Later we will regulate this divergence by moving in a little way from y = 0,
but for the time being we can assume that the number of degrees of freedom
per unit coordinate area is inﬁnite. That suggests that the boundary theory
might be a quantum ﬁeld theory, and that is in fact the case.
Another important fact involves the symmetry of AdS. Let us consider
the metric in the form 12.2.2. It is obvious that the geometry is invariant
under ordinary Poincar´e transformations of the 4dimensional Minkowski
coordinates t, x
i
. In addition there is a “dilatation” symmetry
t → λt
x
i
→ λx
i
y → λy (12.3.5)
On the other hand if we consider the representation of AdS in 12.2.1 we
can see additional symmetry. For example the rotations of the sphere Ω are
symmetries. The full symmetry group of AdS(5) is the group O(42). In
addition there is also the symmetry O(6) associated with rotations of the
internal 5sphere.
Since our goal is a holographic boundary description of the physics in
the bulk spacetime it is very relevant to ask how the symmetries act on
the boundary of AdS. Obviously, the 4dimensional Poincar´e symmetry
acts on the boundary straightforwardly. The dilatation symmetry also acts
as a simple dilatation of the coordinates t, x. All of the transformations
act on the boundary as conformal transformations which preserve lightlike
directions on the boundary. In fact the full AdS symmetry group, when
acting on the boundary at y = 0 is precisely the conformal group of 4
dimensional Minkowski space.
The implication of this symmetry of the boundary is that the holo
graphic boundary theory must be invariant under the conformal group.
This together with the fact that the boundary has an inﬁnite (coordinate)
density of degrees of freedom suggests that the holographic theory is a
Conformal Quantum Field Theory, and so it is.
As we mentioned, AdS(5) ⊗ S(5) is a solution of the 10dimensional
supergravity that describes low energy superstring theory. Indeed the space
has more symmetry than just the conformal group and the O(6) symmetry
of the internal 5sphere. The additional symmetry is the socalled N = 4
supersymmetry. This symmetry must also be realized by the holographic
theory. All of this leads us to the remarkable conclusion that quantum
gravity in AdS(5) ⊗ S(5) should be exactly described by an appropriate
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
132 Black Holes, Information, and the String Theory Revolution
superconformal Lorentz invariant quantum ﬁeld theory associated with the
AdS boundary.
In order to have a benchmark for the counting of degrees of freedom in
AdS(5) ⊗ S(5) imagine constructing a cutoﬀ ﬁeld theory in the bulk. A
conventional cutoﬀ would involve a microscopic length scale such as the 10
dimensional Planck length l
p
. One way to do this would be to introduce a
spatial lattice in 9dimensional space. It is not generally possible to make a
regular lattice, but a random lattice with an average spacing l
p
is possible.
We can then deﬁne a simple theory such as a Hamiltonian lattice theory
on the space. In order to count degrees of freedom we also need to regulate
the area of the boundary of AdS which is inﬁnite. To do so we introduce
a surface L at r = 1 − δ. The total 9dimensional spatial volume in the
interior of L is easily computed using the metric 12.2.2, and is seen to be
critically divergent.
V (δ) ∼
R
9
δ
3
. (12.3.6)
The number of bulk lattice sites and therefore the number of degrees of
freedom is
V
l
9
p
∼
1
δ
3
R
9
l
9
p
(12.3.7)
In such a theory we also will ﬁnd that the maximum entropy is of the same
order of magnitude. However the Holographic Principle suggests that this
entropy is overestimated.
The holographic bound discussed in Chapter 11 requires the maximum
entropy and the number of degrees of freedom to be of order
S
max
∼
A
l
8
p
(12.3.8)
where A is the 8dimensional area of the boundary L. This is also easily
computed. We ﬁnd
S
max
∼
1
δ
3
R
8
l
8
p
(12.3.9)
In other words when R/l
p
becomes large the holographic description re
quires a reduction in the number of independent degrees of freedom by a
factor l
p
/R. To say it slightly diﬀerently, the Holographic Principle im
plies a complete description of all physics in the bulk of a very large AdS
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Holographic Principle and Anti de Sitter Space 133
space in terms of only l
p
/R degrees of freedom per spatial Planck volume.
Nonetheless the theory must be able to describe microscopic events in the
bulk even when R becomes extremely large.
12.4 The AdS/CFT Correspondence
The search for a holographic description of AdS(5) ⊗ S(5) is consider
ably narrowed by the symmetries. In fact there is only one known class of
systems with the appropriate N = 4 supersymmetry; the SU(N) Super
symmetric Yang–Mills (SYM) theories.
The correspondence between gravity or its string theoretic general
ization in AdS(5) ⊗ S(5) and Super Yang–Mills (SYM) theory on the
boundary is the subject of a vast literature. We will only review some
of the salient features. The correspondence states that there is a complete
equivalence between superstring theory in the bulk of AdS(5) ⊗ S(5) and
N = 4, 3 + 1dimensional, SU(N), SYM theory on the boundary of the
AdS space
[
7
][
8
][
9
]
. In these lectures SYM theory will always refer to this
particular version of supersymmetric gauge theory, N represents the num
ber of supersymmetries, and N is the dimension of the Yang–Mills gauge
theory.
It is well known that SYM is conformally invariant and therefore has no
dimensional parameters. It will be convenient to deﬁne the theory to live
on the boundary parametrized by the dimensionless coordinates t, Ω or t, x.
The corresponding momenta are also dimensionless. In fact we will use the
convention that all SYM quantities are dimensionless. On the other hand
the bulk gravity theory quantities such as mass, length and temperature
carry their usual dimensions. To convert from SYM to bulk variables, the
conversion factor is R. Thus if E
SYM
and M represent the energy in the
SYM and bulk theories
E
SYM
= MR.
Similarly bulk time intervals are given by multiplying the t interval by R.
There is one question that may be puzzling to the reader. Since
AdS(5) ⊗ S(5) is a 10dimensional spacetime one might think that its
boundary is (8 + 1) dimensional. But there is an important sense in which
it is 3 + 1 dimensional. To see this let us Weyl rescale the metric by a
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
134 Black Holes, Information, and the String Theory Revolution
factor
R
2
(1−r
2
)
2
so that the rescaled metric at the boundary is ﬁnite. The
new metric is
dS
2
=
(1 + r
2
)
2
dt
2
−4dr
2
−4r
2
dΩ
2
+
(1 −r
2
)
2
dω
2
5
(12.4.10)
Notice that the size of the 5sphere shrinks to zero as the boundary at r = 1
is approached. The boundary of the geometry is therefore 3+1 dimensional.
Let us return to the correspondence between the bulk and boundary the
ories. The 10dimensional bulk theory has two dimensionless parameters.
These are:
1) The radius of curvature of the AdS space measured in string units R/
s
.
Alternately we could measure R in 10dimensional Planck units. The
relation between string and Planck lengths is given by
g
2
8
s
= l
8
p
2) The dimensionless string coupling constant g.
The string coupling constant and length scale are related to the 10
dimensional Planck length and Newton constant by
l
8
p
= g
2
8
s
= G (12.4.11)
On the other side of the correspondence, the gauge theory also has two
constants. They are
1) The rank of the gauge group N
2) The gauge coupling g
ym
Obviously the two bulk parameters R and g must be determined by N
and g
ym
. In these lectures we will assume the relation that was originally
derived in
[
7
]
.
R
s
= (Ng
2
ym
)
1
4
g = g
2
ym
(12.4.12)
We can also write the 10dimensional Newton constant in the form
G = R
8
/N
2
(12.4.13)
There are two distinct limits that are especially interesting, depend
ing on one’s motivation. The AdS/CFT correspondence has been widely
studied as a tool for learning about the behavior of gauge theories in the
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Holographic Principle and Anti de Sitter Space 135
strongly coupled ’t Hooft limit. From the gauge theory point of view the
’t Hooft limit is deﬁned by
g
ym
→ 0
N → ∞
g
2
ym
N = constant (12.4.14)
From the bulk string point of view the limit is
g → 0
R
s
= constant (12.4.15)
Thus the strongly coupled ’t Hooft limit is also the classical string theory
limit in a ﬁxed and large AdS space. This limit is dominated by classical
supergravity theory.
The interesting limit from the viewpoint of the holographic principle is
a diﬀerent one. We will be interested in the behavior of the theory as the
AdS radius increases but with the parameters that govern the microscopic
physics in the bulk kept ﬁxed. This means we want the limit
g = constant
R/
s
→ ∞ (12.4.16)
On the gauge theory side this is
g
ym
= constant
N → ∞ (12.4.17)
Our goal will be to show that the number of quantum degrees of freedom in
the gauge theory description satisﬁes the holographic behavior in equation
12.3.8.
12.5 The Infrared Ultraviolet Connection
In either of the metrics in equation 12.2.1 or 12.2.2 the proper area of any
ﬁnite coordinate patch tends to ∞ as the boundary of AdS is approached.
Thus we expect that the number of degrees of freedom associated with
such a patch should diverge. This is consistent with the fact that a con
tinuum quantum ﬁeld theory such as SYM has an inﬁnity of modes in any
ﬁnite threedimensional patch. In order to do a more reﬁned counting
[
9
]
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
136 Black Holes, Information, and the String Theory Revolution
we need to regulate both the area of the AdS boundary and the number
of ultraviolet degrees of freedom in the SYM. As we will see, these appar
ently diﬀerent regulators are really two sides of the same coin. We have
already discussed infrared (IR) regulating the area of AdS by introducing
a surrogate boundary L at r = 1 −δ or similarly at y = δ.
That the IR regulator of the bulk theory is equivalent to an ultraviolet
(UV) regulator in the SYM theory is called the IR/UV connection
[
9
]
. It
is in many ways similar to the behavior of strings as we study them at
progressively shorter time scales. In Chapter 14 we will ﬁnd the interesting
behavior that a string appears to grow as we average its properties over
smaller and smaller time scales. To understand the relation between this
phenomenon and the IR/UV connection in AdS we need to discuss the
relation between AdS(5) ⊗S(5) and Dbranes.
Dbranes are objects which occur in superstring theory. They are sta
ble “impurities” of various dimensionality that can appear in the vacuum.
A Dpbrane is a pdimensional object. We are especially interested in
D3branes. Such objects ﬁll 3 dimensions of space and also time. Their
properties are widely studied in string theory and we will only quote the
results that we need. The most important property of D3branes is that
they are embedded in a 10dimensional space. Let us assume that they ﬁll
time and the 3 spatial coordinates x
i
. Let the other 6 coordinates be called
z
m
and let z ≡
√
z
m
z
m
. We will place a “stack” of N D3branes at z = 0.
Now a single Dbrane has local degrees of freedom. For example the
location in z may ﬂuctuate. Thus we can think of the z location as a
scalar ﬁeld living on the Dbrane. In addition there are modes of the brane
which are described by vector ﬁelds with components in the t, x direction as
well as fermionic modes needed for supersymmetry. Our main concern will
be with the z(x, t) ﬂuctuations whose action is known from string theory
calculations to be a that of conventional 3+1 dimensional scalar ﬁeld theory.
Dbranes can also be juxtaposed to form stacks of Dbranes. A stack of
N Dbranes has a mass and Dbrane charge which grow with N. The mass
and charge are sources of bulk ﬁelds such as the gravitational ﬁeld. What
makes the Dbrane stack interesting to us is that the geometry sourced by
the stack is exactly that of AdS(5) ⊗S(5). In fact the geometry deﬁned by
12.2.3 and 12.2.4 is closely related to that of a Dbrane stack.
Speciﬁcally the geometry sourced by the Dbranes is a particular solu
tion of the supergravity equations of motion:
ds
2
= F(z)(dt
2
−dx
2
) −F(z)
−1
dzdz (12.5.18)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Holographic Principle and Anti de Sitter Space 137
where
F(z) =
1 +
cg
s
N
z
4
−1/2
(12.5.19)
and c is a numerical constant. If we consider the limit in which
cg
s
N
z
4
>> 1
then we can replace F(z) by the simpler expression
F(z)
∼
=
z
2
(cg
s
N)
1
2
. (12.5.20)
It is then a simple exercise to see that the Dbrane metric is of the form
12.2.3, 12.2.4.
Furthermore the theory of the ﬂuctuations of the stack is N = 4 SYM.
All of the ﬁelds in this theory form a single supermultiplet belonging to the
adjoint (N ×N matrix) representation of SU(N).
In this lecture we give an argument for the IR/UV connection based
on the quantum ﬂuctuations of the positions of the D3branes which are
nominally located at the origin of the coordinate z in equation 12.2.3. The
location of a point on a 3brane is deﬁned by six coordinates z, ω
5
. We may
also choose the six coordinates to be Cartesian coordinates (z
1
, ..., z
6
). The
original coordinate z is deﬁned by
z
2
= (z
1
)
2
+ ... + (z
6
)
2
(12.5.21)
As we indicated, the coordinates z
m
are represented in the SYM theory
by six scalar ﬁelds on the world volume of the branes. If the six scalar ﬁelds
φ
n
are canonically normalized, then the precise connection between the z
s
and φ
s is
z =
g
ym
2
s
R
2
φ (12.5.22)
Strictly speaking equation 12.5.22 does not make sense because the ﬁelds
φ in SU(N) are N × N matrices, where we identify the N eigenvalues of
the matrices in equation 12.5.18 to be the coordinates z
m
of the N D3
branes
[
10
]
. The geometry is noncommutative and only conﬁgurations in
which the six matrix valued ﬁelds commute have a classical interpretation.
However the radial coordinate z =
√
z
m
z
m
can be deﬁned by
z
2
=
g
ym
2
s
R
2
2
1
N
Trφ
2
(12.5.23)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
138 Black Holes, Information, and the String Theory Revolution
A question which is often asked is: where are the D3branes located in
the AdS space? The usual answer is that they are at the horizon z = 0.
However our experiences in Chapter 14 with similar questions will warn us
that the answer may be more subtle. What we will ﬁnd there is that the
way information is localized in space depends on what frequency range it
is probed with. High frequency or short time probes see the string widely
spread in space while low frequency probes see a well localized string.
To answer the corresponding question about D3branes we need to study
the quantum ﬂuctuations of their position. The ﬁelds φ are scalar quantum
ﬁelds whose scaling dimensions are known to be exactly (length)
−1
. From
this it follows that any of the N
2
components of φ satisﬁes
φ
2
ab
∼ δ
−2
(12.5.24)
where δ is the ultraviolet regulator of the ﬁeld theory. It follows from
equation 12.5.20 that the average value of z satisﬁes
< z >
2
∼
g
ym
2
s
R
2
2
N
δ
2
(12.5.25)
or, using equation 12.4.12
< z >
2
∼ δ
−2
(12.5.26)
In terms of the coordinate y which vanishes at the boundary of AdS
< y >
2
∼ δ
2
. (12.5.27)
Here it is seen that the location of the brane is given by the ultraviolet
cutoﬀ of the ﬁeld theory on the boundary. Evidently low frequency probes
see the branes at z = 0 but as the frequency of the probe increases the brane
appears to move toward the boundary at z = ∞. The precise connection
between the UV SYM cutoﬀ and the bulk theory IR cutoﬀ is given by
equation 12.5.23.
12.6 Counting Degrees of Freedom
Let us now turn to the problem of counting the number of degrees of
freedom needed to describe the region y > δ
[
9
]
. The UV/IR connection
implies that this region can be described in terms of an ultraviolet regulated
theory with a cutoﬀ length δ. Consider a patch of the boundary with
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
The Holographic Principle and Anti de Sitter Space 139
unit coordinate area. Within that patch there are 1/δ
3
cutoﬀ cells of size
δ. Within each such cell the ﬁelds are constant in a cutoﬀ theory. Thus
each cell has of order N
2
degrees of freedom corresponding to the N ⊗ N
components of the adjoint representation of U(N). Thus the number of
degrees of freedom on the unit area is
N
dof
≈
N
2
δ
3
(12.6.28)
On the other hand the 8dimensional area of the regulated patch is
A =
R
3
δ
3
×R
5
=
R
8
δ
3
(12.6.29)
and the number of degrees of freedom per unit area is
N
dof
A
∼
N
2
R
8
(12.6.30)
Finally we may use equation 12.4.13
N
dof
A
∼
1
G
(12.6.31)
This is very gratifying because it is exactly what is required by the Holo
graphic Principle.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
140 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 13
Black Holes in a Box
The apparently irreconcilable demands of black hole thermodynamics
and the principles of quantum mechanics have led us to a very strange view
of the world as a hologram. Now we will return, full circle, to see how the
holographic description of AdS(5) ⊗ S(5) provides a description of black
holes.
We have treated Schwarzschild black holes as if they were states of
thermal equilibrium, but of course they are not. They are longlived objects,
but eventually they evaporate. We can try to prevent their evaporation by
placing them in a thermal heat bath at their Hawking temperature but
that does not work. The reason is that their speciﬁc heat is negative;
their temperature decreases as their energy or mass increases. Any object
with this property is thermodynamically unstable. To see this, suppose a
ﬂuctuation occurs in which the black hole absorbs an extra bit of energy
from the surrounding heat bath. For an ordinary system with positive
speciﬁc heat this will raise its temperature which in turn will cause it to
radiate back into the environment. The ﬂuctuations are selfregulating.
But a system with negative speciﬁc heat will lower its temperature when
it absorbs energy and will become cooler than the bath. This in turn will
favor an additional ﬂow of energy from the bath to the black hole and a
runaway will occur. The black hole will grow indeﬁnitely. If on the other
hand the black hole gives up some energy to the environment it will become
hotter than the bath. Again a runaway will occur that leads the black hole
to disappear.
A well known way to stabilize the black hole is to put it in a box so that
the environmental heat bath is ﬁnite. When the black hole absorbs some
energy it cools but so does the ﬁnite heat bath. If the box is not too big
the heat bath will cool more than the black hole and the ﬂow of heat will
141
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
142 Black Holes, Information, and the String Theory Revolution
be back to the bath. In this lecture we will consider the properties of black
holes which are stabilized by the natural box provided by Anti de Sitter
space. More speciﬁcally we consider large black holes in AdS(5)⊗S(5) and
their holographic description in terms of the N = 4 Yang–Mills theory.
The black holes which are stable have Schwarzschild radii as large or
larger than the radius of curvature R. They homogeneously ﬁll the 5sphere
and are solutions of the dimensionally reduced 5dimensional Einstein equa
tions with a negative cosmological constant. The thermodynamics can be
derived from the black hole solutions by ﬁrst computing the area of the
horizon and then using the Bekenstein–Hawking formula.
Before writing the AdS–Schwarzschild metric, let us write the metric of
AdS in a form which is convenient for generalization.
dτ
2
=
1 +
r
2
R
2
dt
2
−
1 +
r
2
R
2
−1
dr
2
−r
2
dΩ
2
(13.0.1)
where in this formula r runs from 0 to the boundary at r = ∞. Note that
the coordinates r, t are not the same as in equation 12.2.1.
The AdS black hole is given by modifying the function
1 +
r
2
R
2
:
dτ
2
=
1 +
r
2
R
2
−
µG
R
5
r
2
dt
2
−
1 +
r
2
R
2
−
µG
R
5
r
2
−1
dr
2
−r
2
dΩ
2
(13.0.2)
where the parameter µ is proportional to the mass of the black hole and G
is the 10dimensional Newton constant. The horizon of the black hole is at
the largest root of
1 +
r
2
R
2
−
µG
R
5
r
2
= 0
The Penrose diagram of the AdS black hole is shown in Figure 13.1. One
ﬁnds that the entropy is related to the mass by
S = c
M
3
R
11
G
−1
1
4
(13.0.3)
where c is a numerical constant. Using the thermodynamic relation dM =
TdS we can compute the relation between mass and temperature:
M = c
R
11
T
4
G
(13.0.4)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Black Holes in a Box 143
Fig. 13.1 Penrose diagram of the AdS black hole
or in terms of dimensionless SYM quantities
E
sym
= c
R
8
G
T
4
sym
= cN
2
T
4
sym
(13.0.5)
Equation 13.0.5 has a surprisingly simple interpretation. Recall that
in 3 + 1 dimensions the Stephan–Boltzmann law for the energy density of
radiation is
E ∼ T
4
V (13.0.6)
where V is the volume. In the present case the relevant volume is the
dimensionless 3area of the unit boundary sphere. Furthermore there are
∼ N
2
quantum ﬁelds in the U(N) gauge theory so that apart from a nu
merical constant equation 13.0.5 is nothing but the Stephan–Boltzmann
law for black body radiation. Evidently the holographic description of the
AdS black holes is as simple as it could be; a black body thermal gas of N
2
species of quanta propagating on the boundary hologram.
The constant c in equation 13.0.3 can be computed in two ways. The
ﬁrst is from the black hole solution and the Bekenstein–Hawking formula.
The second way is to calculate it from the boundary quantum ﬁeld theory in
the free ﬁeld approximation. The calculations agree in order of magnitude,
but the free ﬁeld gives too big a coeﬃcient by a factor of 4/3. This is
not too surprising because the classical gravity approximation is only valid
when g
2
Y M
N is large.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
144 Black Holes, Information, and the String Theory Revolution
13.1 The Horizon
The high frequency quantum ﬂuctuation of the location of the D3branes
are invisible to a low frequency probe. Roughly speaking this is insured by
the renormalization group as applied to the SYM description of the branes.
The renormalization group is what insures that our bodies are not severely
damaged by constant exposure to high frequency vacuum ﬂuctuations. We
are not protected in the same way from classical ﬂuctuations. An exam
ple is the thermal ﬂuctuations of ﬁelds at high temperature. All probes
sense thermal ﬂuctuations of the brane locations. Let us return to equation
12.5.24 but now, instead of using equation 12.5.25 we use the thermal ﬁeld
ﬂuctuations of φ. For each of the N
2
components the thermal ﬂuctuations
have the form
< φ
2
>= T
2
sym
(13.1.7)
and we ﬁnd equations 12.5.26 and 12.5.27 replaced by
< z >
2
∼ T
2
sym
< y >
2
∼ T
−2
sym
(13.1.8)
It is clear that the thermal ﬂuctuations will be strongly felt out to a coor
dinate distance z = T
sym
. In terms of r the corresponding position is
1 − r ∼ 1/T
sym
(13.1.9)
In fact this coincides with the location of the horizon of the AdS black hole.
13.2 Information and the AdS Black Hole
The AdS black hole is an ideal laboratory for investigating how bulk
quantum ﬁeld theory fails when applied to the ﬁne details of Hawking ra
diation. Let us consider some ﬁeld that appears in the supergravity de
scription of the bulk. Such objects are 10dimensional ﬁelds and should
not be confused with the 4dimensional quantum ﬁelds associated with the
boundary theory. A simple example is the minimally coupled scalar dilaton
ﬁeld φ. We will only consider dilaton ﬁelds which are constant on the 5
sphere. In that case the action for φ is the minimally coupled scalar action
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Black Holes in a Box 145
in 5dimensional AdS.
I =
d
5
x
√
−gg
µν
∂
µ
φ∂
ν
φ. (13.2.10)
The appropriate boundary conditions are φ → 0 at the boundary of the
AdS, i.e. r →∞.
Plugging in the black hole metric given in equation 13.0.2, we ﬁnd a
number of things. First φ(r) ∼ r
−4
Φ as r → ∞. It is the value of Φ on
the boundary which is identiﬁed with a local ﬁeld in the boundary Super
Yang–Mills theory. This is true both in pure AdS as well as in the AdS–
Schwarzschild metrics. Secondly, in the pure AdS background φ is periodic
in time, but in the black hole metric φ goes to zero exponentially with time:
φ →exp−γt. (13.2.11)
Equation 13.2.11 has implications for quantum correlation functions in
the black hole background. Consider the correlator φ(t)φ(t
). Equation
13.2.11 requires it to behave like
φ(t)φ(t
) ∼ exp−γt −t
. (13.2.12)
for large t − t
 The parameter γ depends on the black hole mass or tem
perature and has the form
γ =
H(µGR
−7
)
R
(13.2.13)
where H is a dimensionless increasing function of its argument.
The meaning of this exponential decrease of the correlation function is
that the eﬀects of an initial perturbation at time t dissipate away and are
eventually lost. In other words the system does not preserve any memory
of the initial perturbation. This type of behavior is characteristic of large
thermal systems where γ would correspond to some dissipation coeﬃcient.
However, exponential decay is not what is really expected for systems of
ﬁnite entropy such as the AdS black hole that we are dealing with. Any
quantum system with ﬁnite entropy preserves some memory of a perturba
tion. Since AdS is exactly described by a conventional quantum system it
follows that the correlator should not go to zero. We shall now prove this
assertion.
The essential point is that any quantum system with ﬁnite thermal
entropy must have a discrete spectrum. This is because the entropy is
essentially the logarithm of the number of states per unit energy. Indeed
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
146 Black Holes, Information, and the String Theory Revolution
the spectrum of the boundary quantum ﬁeld theory is obviously discrete
since it is a theory deﬁned on a ﬁnite sphere.
Let us now consider a general ﬁnite closed system described by a thermal
density matrix and a thermal correlator of the form
F(t) = A(0)A(t) =
1
Z
Tre
−βH
A(0)e
iHt
A(0)e
−iHt
. (13.2.14)
By ﬁnite we simply mean that the spectrum is discrete and the entropy
ﬁnite. Inserting a complete set of (discrete) energy eigenstates gives
F(t) =
1
Z
ij
e
−βE
i
e
i(E
j
−E
i
)t
A
ij

2
. (13.2.15)
For simplicity we will assume that the operator A has no matrix elements
connecting states of equal energy. This means that the time average of F
vanishes.
Let us now consider the long time average of F(t)F
∗
(t).
L = lim
T→∞
1
2T
+T
−T
dtF(t)F
∗
(t) (13.2.16)
Using equation 13.2.15 it is easy to show that the long time average is
L =
1
Z
2
ijkl
e
−β(E
i
+E
k
)
A
ij

2
A
kl

2
δ
(E
j
−E
l
+E
k
−E
i
)
. (13.2.17)
where the delta function is deﬁned to be zero if the argument is nonzero and
1 if it is zero. The long time average L is obviously nonzero and positive.
Thus it is not possible for the correlator F(t) to tend to zero as the time
tends to inﬁnity and the limits required by the AdS/CFT correspondence
cannot exist. The value of the long time average for such ﬁnite systems can
be estimated, and it is typically of the order e
−S
where S is the entropy
of the system. This observation allows us to understand why it tends to
zero in the (bulk) QFT approximation. In studying QFT in the vicinity
of a horizon we have seen that the entropy is UV divergent. This is due
to the enormous number of short wave length modes near the horizon.
This leads us to a very important and general conclusion: any phenomenon
which crucially depends on the ﬁniteness of horizon entropy will be gotten
wrong by the approximation of QFT in a ﬁxed background. This includes
questions of information loss and of particular interest in this lecture, the
long time behavior of correlation functions.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Black Holes in a Box 147
How exactly do the correlations behave in the long time limit? The
answer is not that they uniformly approach constants given by the long
time averages. The expected behavior is that they ﬂuctuate chaotically. A
large ﬂuctuation which reduces the entropy by amount ∆S has probability
e
−∆S
. Thus we can expect large ﬂuctuations in the correlators at intervals
of order e
S
. These ﬂuctuations are analogous to the classical phenomenon
of Poincar´e recurrences. It is generally found that the large time behavior
of correlators is chaotic “noise” with the long time average given by
L ∼ e
−S
. (13.2.18)
This long time behavior, missed by bulk quantum ﬁeld theory, is a small
part of the encoding of information in the thermal atmosphere of the AdS
black hole.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
148 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
PART 3
Black Holes and Strings
149
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
150
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 14
Strings
We have learned a great deal about black holes by considering the be
havior of quantum ﬁelds near horizons. But ultimately local quantum ﬁeld
theory fails in a number of ways. In general the failures can be attributed to
a common cause – quantum ﬁeld theory has too many degrees of freedom.
The earliest evidence that QFT is too rich in degrees of freedom was
the uncontrollable short distance divergences in gravitational perturbation
theory. As a quantum ﬁeld theory, Einstein’s general relativity is very badly
behaved in the ultraviolet.
Even more relevant for our purposes is the divergence in the entropy
per unit area of horizons that was found in Chapter 4. Entropy is a direct
measure of the number of active degrees of freedom of a system. Evidently
there are far too many degrees of freedom very close to a horizon in QFT.
Later in Chapter 12 we quantiﬁed just how overrich QFT is.
The remaining portion of this book deals, in an elementary way, with
a theory that seems to have just the right number of degrees of freedom:
string theory. The problems posed by black holes for a fundamental theory
of quantum gravity are nonperturbative. Until relatively recently, string
theory was mostly deﬁned by a set of perturbation rules. Nevertheless, even
in perturbation theory, we will see certain trends that are more consistent
with black hole complementarity than the corresponding trends in QFT.
In Chapter 9 we explained that the key to understanding black hole
complementarity lies in the ultrahigh frequency oscillations of ﬂuctuations
of matter in its own rest frame. The extreme red shift between the freely
falling frame and the Schwarzschild frame may take phenomena which are
of too high frequency to be visible ordinarily and make them visible to the
outside observer. As an example, imagine a freely falling whistle that emits
a sound of such high frequency that it cannot be heard by the human ear.
151
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
152 Black Holes, Information, and the String Theory Revolution
As the whistle approaches the horizon, the observer outside the black hole
hears the frequency red shifted. Eventually it becomes audible, no matter
how high the frequency in the whistle’s rest frame.
On the other hand, the freely falling observer who accompanies the
whistle never gets the beneﬁt of the increasing red shift. She never hears
the whistle.
This suggests that the consistency of black hole complementarity is a
deep constraint on how matter behaves at very short times or high frequen
cies. Quantum ﬁeld theory gets it wrong, but string theory seems to do
better. The qualitative behavior of strings is the subject of this lecture.
In order to compare string theory and quantum ﬁeld theory near a
horizon, we will ﬁrst study the case of a free particle falling through a
Rindler horizon. As we will see, it is natural to use light cone coordinates
for this problem. The process and conventions are illustrated in Figure
14.1. The coordinates X
±
are deﬁned by
X
±
=
X
0
± X
√
2
= ∓
ρ
√
2
e
∓τ
(14.0.1)
and the metric is given by
dτ
2
= 2dX
+
dX
−
−
dX
i
2
(14.0.2)
where X
i
run over the coordinates in the plane of the horizon. We will
refer to X
i
as the transverse coordinates, because they are transverse to
the direction of motion of the point particle. The trajectory of the particle
is taken to be
X
i
= 0
X
−
− X
+
=
√
2L
(14.0.3)
As the particle falls closer and closer to the horizon, the constant τ
surfaces become more and more lightlike in the particle’s rest frame. In
other words, the particle and the Rindler observer are boosted relative to
one another by an everincreasing boost angle.
Near the particle trajectory X
+
and τ are related by
X
+
∼
= −2Le
−2τ
(14.0.4)
for large τ. This suggests that the description of mechanics in terms of the
Rindler (or Schwarzschild) time be replaced by a description in light cone
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Strings 153
L
X
+
X

2
L
Fig. 14.1 Free particle falling through a Rindler horizon
coordinates with X
+
playing the role of the independent time coordinate.
We will therefore brieﬂy review particle mechanics in the light cone frame.
14.1 Light Cone Quantum Mechanics
In order to write the action for a relativistic point particle we introduce
a parameter σ along the world line of the particle. Since the action only
depends on the world line and not the way we parameterize it, the action
should be invariant under a reparameterization. Toward this end we also
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
154 Black Holes, Information, and the String Theory Revolution
introduce an “einbein” e(σ) that transforms under σreparameterizations.
e
1
(σ
1
) dσ
1
= e(σ) dσ (14.1.5)
The action is given by
W =
Ldσ
L = −
1
2
1
e
dX
µ
dσ
dX
µ
dσ
− e m
2
(14.1.6)
where m is the mass of the particle. The action in equation 14.1.6 is
invariant under σreparameterizations.
Let us now write equation 14.1.6 in terms of light cone coordinates and,
at the same time use our gauge freedom to ﬁx σ = X
+
(which is then
treated as a time variable). Then L takes the form
L =
1
2
−
2
e
dX
−
dσ
+
1
e
dX
i
dσ
dX
i
dσ
− e m
2
(14.1.7)
The conserved canonical momenta are given by
P
−
=
∂L
∂
˙
X
−
= −
1
e
P
i
=
∂L
∂
˙
X
i
=
1
e
˙
X
i
(14.1.8)
where dot refers to σ derivative. Note that the conservation of P
−
insures
that e(σ) has a ﬁxed constant value.
The Hamiltonian is easily obtained by the standard procedure:
H =
eP
2
i
2
+
m
2
e
2
(14.1.9)
This form of H manifests a well known fact about light cone physics. If
we focus on the transverse degrees of freedom, the Hamiltonian has all the
properties of a nonrelativistic system with Galilean symmetry. The sec
ond term in H is just a constant, and can be interpreted as an internal
energy that has no eﬀect on the transverse motion. The ﬁrst term has
the usual nonrelativistic form with e
−1
playing the role of an eﬀective
transverse mass. This Hamiltonian and its associated quantum mechan
ics exactly describes the point particles of conventional free quantum ﬁeld
theory formulated in the light cone gauge.
Now let us consider the transverse location of the particle as it falls
toward the horizon. In particular, suppose the particle is probed by an
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Strings 155
experiment which takes place over a short time interval δ just before horizon
crossing. In other words, the particle is probed over the time interval
−δ < X
+
< 0 (14.1.10)
by a quantum of (Minkowski) energy ∼
1
δ
. This experiment is similar to the
one discussed in Chapter 9.3, except that the probe carries out information
about the transverse location of the particle instead of its baryon number.
Since the interaction is spread over the time interval in equation 14.1.10,
the instantaneous transverse position should be replaced by the time aver
aged coordinate X
i
δ
X
i
δ
=
1
δ
0
−δ
X
i
(σ) dσ (14.1.11)
To evaluate equation 14.1.11, we use the nonrelativistic equations of motion
X
i
(σ) = X
i
(0) + eP
i
σ (14.1.12)
to give
X
i
δ
= X
i
(0) +
eP
i
δ
2
. (14.1.13)
Finally, let us suppose that the particle wave function is initially a
smooth wave packet well localized in transverse position with uncertainty
∆X
i
. Let us also assume the very high momentum components of the wave
function are negligible. Under these conditions nothing singular happens
to the probability distribution for X
i
δ
as δ → 0. No matter how small δ
is, the eﬀective probability distribution for X
δ
is concentrated in a well
localized region of ﬁxed extent, δX. There is no tendency for information
to transversely spread over a stretched horizon.
All of this is exactly what is expected for an ordinary particle in free
quantum ﬁeld theory. For the more interesting case of an interacting quan
tum ﬁeld theory, we could study the transverse properties of an interacting
or composite particle such as a hydrogen atom. For example, a time aver
aged relative coordinate or charge density can be deﬁned, and it too shows
no sign of spreading as the sampling interval δ tends to zero.
Why is this a problem? The reason is that it conﬂicts with the com
plementarity principle. Complementarity requires the probe to report that
the particle fell into a very high temperature environment in which it re
peatedly suﬀered high energy collisions. In this kind of environment the
information stored in the infalling system would be thermalized and spread
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
156 Black Holes, Information, and the String Theory Revolution
over the horizon. The implication for the probing experiment is that the
particle should somehow spread or diﬀuse roughly the way the eﬀective
charge distribution did in Chapter 7.
14.2 Light Cone String Theory
Although naive pertubative string theory cannot capture this eﬀect com
pletely correctly, the tendency is already there in the theory of free strings.
A free string is a generalization of a free particle. There are a number of
excellent textbooks on string theory that the reader who is interested in
technical details can consult. For our purposes, only the most elementary
aspects of string theory will be needed.
A string is a onedimensional continuum whose points are parameterized
by a continuous parameter σ
1
. The transverse coordinates of the point at
σ
1
are labeled X
i
(σ), where σ
1
runs from 0 to 2π. It is also a function of a
timelike parameter σ
0
, which is identiﬁed with light cone time X
+
. Thus
X
i
(σ
0
, σ
1
) is a ﬁeld deﬁned on a 1+1 dimensional parameter space (σ
a
).
In addition to X
i
(σ), the canonical momentum density P
i
(σ) can also be
deﬁned. At equal times X and P satisfy
X
i
(σ), P
j
(σ
)
= i δ
i
j
δ(σ − σ
) (14.2.14)
The light cone Hamiltonian for the free string is a natural generalization of
that for a free particle;
H =
1
P
−
2π
0
dσ
2
P
i
(σ
)
2
+
∂X
i
∂σ
2
(14.2.15)
We have used units in which the string tension (energy per unit length in
the rest frame) is unity.
The equation of motion following from equations 14.2.14 and 14.2.15 is
a simple wave equation
∂
2
X
i
(∂σ
0
)
2
−
∂
2
X
i
(∂σ
1
)
2
= 0. (14.2.16)
Quantization of the string is straightforward. X
i
(σ) becomes a free scalar
ﬁeld in 1+1 dimensions satisfying equation 14.2.16 with periodic boundary
conditions in σ
1
, X(σ
0
, 2π) = X(σ
0
, 0).
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Strings 157
The string diﬀers in important ways from the free particle, especially in
its short time behavior. As we have repeatedly emphasized, it is the short
time behavior that is key to complementarity.
Let us consider the analog of the question that we addressed about the
time averaged location of the point particle. Now we consider the time
averaged location of a point on the string. Thus, deﬁne
X
δ
=
1
δ
δ
0
dσ
0
X(σ) (14.2.17)
Since all points σ
1
are equivalent, it doesn’t matter what value σ
1
takes
on the right hand side when we evaluate X
δ
. A useful measure of how
much the information in a string is spread as it falls towards the horizon is
provided by the ﬂuctuations in X
δ
, that is
(∆X)
2
=
X
2
δ
− X
δ
2
(14.2.18)
The state used for the expectation value in equation 14.2.18 is the ground
state string. This quantity is easily calculated and diverges logarithmically
as δ → 0. In other words, as the string approaches the horizon, any experi
ment (from the outside) to determine how its internal parts are distributed
will indicate a logarithmic increase in the area it occupies
(∆X)
2
∼  log δ . (14.2.19)
Another way to write equation 14.2.19 is to use the connection between
Rindler time and light cone time in equation 14.0.4
(∆X)
2
∼  log (2Le
−2τ
)  ∼ 2τ.
Finally, we can use the relation betwen Rindler time and Schwarzschild
time given by τ = t/4MG to obtain
(∆X)
2
∼
α
t
4MG
. (14.2.20)
In equation 14.2.20 we have restored the units by including the factor α
,
the inverse string tension.
Here we see the beginnings of an explanation of complementarity. The
observer outside the black hole will ﬁnd the string diﬀusing over an increas
ing area of the horizon as time progresses. But an observer falling with the
string and doing low energy experiments on it would conclude that the
string remains a ﬁxed ﬁnite size as it falls.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
158 Black Holes, Information, and the String Theory Revolution
The linear growth of the area in equation 14.2.20 is much slower than the
growth of a charged particle described in Chapter 7. In that case inspection
of 7.0.21 indicates that the growth is exponential. A completely consistent
theory would require these growth patterns to match. The true exponential
asymptotic growth is undoubtedly a nonperturbative phenomenon that
involves string interactions in an essential way.
To see how interactions inﬂuence the evolution, let’s determine the av
erage total length of string, projected onto the twodimensional transverse
plane
=
2π
0
dσ
1
∂X
i
∂σ
1
2
(14.2.21)
As a preliminary, let us consider the ground state average of
∂X
i
∂σ
1
2
. This
is another exercise in free scalar quantum ﬁeld theory, and the result is
quadratically divergent.
If however
∂X
i
∂σ
is averaged over the time interval δ, we ﬁnd that the
ground state average of
∂X
i
∂σ
1
2
is given by
∂X
i
∂σ
1
∂X
i
∂σ
1
∼
1
δ
2
(14.2.22)
Using the fact that the probability distribution for
∂X
i
∂σ
1
is Gaussian in free
ﬁeld theory, we can conclude that
∂X
∂σ
1
2
or scales as
≈
1
δ
(14.2.23)
or using equation 14.0.4
≈
1
2L
e
2τ
. (14.2.24)
In other words, as the string falls toward the horizon, it grows exponentially
in length.
Another quantity which exponentially grows is the ρ component of the
Rindler momentum. To see this, we use the transformation in equation
14.0.1 to derive
∂
∂ρ
=
1
√
2
−e
−τ ∂
∂X
+
−+e
τ ∂
∂X
−
,
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Strings 159
or in terms of momenta
P
ρ
=
1
√
2
e
τ
P
−
− e
−τ
P
+
(14.2.25)
In the Rindler approximation to a black hole horizon, P
±
are conserved,
and therefore as τ →∞ the radial momentum P
ρ
grows like e
τ
. Evidently
then the ratio of the string length to its total radial momentum is ﬁxed. As
the string falls toward the horizon, its radial momentum increases by the
mechanism of its physical length increasing.
14.3 Interactions
In Chapter 7 we saw that a charge falling toward the stretched horizon
spreads over an area which grows exponentially. The area occupied by a
free string only grows linearly. However, this not the end of the story. The
the total length of the string grows exponentially with τ. It is clear that
this behavior cannot continue indeﬁnitely. The exponential growth of string
length and linear growth of area imply that the transverse density of string
increases to the point where string interactions must become important
and seriously modify the free string picture. Roughly speaking, when a
piece of string gets within a distance of order
√
α
of another piece, they
can interact. The number of such string encounters will obviously increase
without bound as τ →∞.
String interactions are governed by a dimensionless coupling constant g
which determines the amplitude for strings to rearrange when they cross.
Obviously the importance of interactions is governed not only by g, but
also by the local density of string crossings. Let ρ be the number of such
crossings per unit horizon area. When g
2
ρ becomes large, interactions can
no longer be ignored.
Now, the form g
2
ρ is not dimensionless. There is only one dimensional
constant in string theory, the inverse string tension α
with units of area.
Sometimes α
is replaced by a length
s
=
√
α
. The dimensionally correct
statement is that string interactions become important when
g
2
ρ ≥
1
2
s
≈
1
g
2
α
(14.3.26)
This criterion has a profound signiﬁcance. The quantity g
2
α
in string
theory also governs the gravitational interaction between masses. It is
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
160 Black Holes, Information, and the String Theory Revolution
none other than the gravitational coupling constant (in units with c =
= 1). The implication is that interactions become important when the
area density of the string approaches
1
G
, the area density of horizon entropy.
Although we cannot follow the string past the point where interactions
become important, we can be sure that something new will happen. A
good guess is that the density of string saturates at order
1
G
. Since the
total length of string grows like ≈ e
τ
the area that it occupies must also
grow exponentially. This is reminiscent of the pattern of growth that we
encountered in Chapter 7.
Verifying that as δ → 0 the string grows as if the density saturates is
beyond the current technology of string theory. But the simple assumption
that splitting and joining interactions cause eﬀective short range repulsion,
and that the repulsion prevents the density from increasing indeﬁnitely,
provides a phenomenological description of how information spreads over
the horizon. Since the spreading is associated with a decreasing time of
averaging it is not seen by a freely falling observer. This is the essence of
complementarity.
In general, string theory is not a 4dimensional theory. It is important
to check if the same logic applies in higher dimensions. Let D be the space
time dimension. The general case goes as follows:
Since the number of transverse directions is D2, equation 14.3.26 is re
placed by
g
2
ρ ≥
1
D−2
s
=
1
(α
)
D−2
2
(14.3.27)
or
ρ ≈
1
g
2
D−2
s
(14.3.28)
In D dimensions, the gravitational and string couplings are related by
G ≈ g
2
D−2
s
(14.3.29)
so that the perturbative limit is again reached when the density is of order
ρ ≈
1
G
(14.3.30)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Strings 161
14.4 Longitudinal Motion
In discussing the properties of horizons, we have repeatedly run into
the idea of the stretched horizon, located a microscopic distance above
the mathematical horizon. It is natural to ask if the stretched horizon
has any reality or whether it is just a mathematical ﬁction? Of course,
to a Frefo observer, neither the stretched nor the mathematical horizon
appears real. But to a Fido the stretched horizon is the layer containing
the physical degrees of freedom that give rise to the entropy of the horizon.
Thus, consistency requires that in some appropriate sense, the degrees of
freedom of an infalling object should get deposited in this layer of ﬁnite
thickness. To study this question we must examine how strings move in
the X
−
direction.
It is a curious property of string theory that X
−
(σ) is not an indepen
dent degree of freedom. The only degrees of freedom carried by the string
are the transverse coordinates X
i
(σ) which lie in the plane of the horizon.
The longitudinal location is deﬁned by an equation whose origin is in the
gauge ﬁxing to the light cone gauge. The derivation is provided in the
supplement that follows this discussion.
∂X
−
∂σ
1
=
∂X
i
∂σ
1
∂X
i
∂σ
0
(14.4.31)
Recall that the quantum theory of X(σ) is a simple (1+1 dimensional)
quantum ﬁeld theory of (D2) free scalar ﬁelds deﬁned on a unit circle. In
such a theory, local operators can be characterized by a mass dimension.
For example, X
i
has dimension zero, while
∂X
i
∂σ
has dimension 1. The right
side of equation 14.4.31 has dimension 2. Thus it is apparent that X
−
(σ)
has dimension 1. It immediately follows that the ﬂuctuation in X
−
satisﬁes
∆X
−
∼
2
s
δ
(14.4.32)
The factor
2
s
is needed for dimensional reasons.
Another way to write equation 14.4.32 is to observe that δ is an aver
aging time in light cone coordinates. In other words δ = ∆X
+
. Equation
14.4.32 then takes the form of an uncertainty principle
∆X
−
∆X
+
≈
2
s
. (14.4.33)
Now the geometric meaning of equation 14.4.33 is very interesting. Let us
draw the motion of the ﬂuctuating string in the X
±
plane as it falls towards
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
162 Black Holes, Information, and the String Theory Revolution
x
+ }
∆τ
x

Region of
increasing
fluctuation
∆
Fig. 14.2 Near horizon Rindler time slices
the horizon. Evidently as X
+
tends to zero the ﬂuctuation in X
−
required
by equation 14.4.33 must increase. This is shown in Figure 14.2. What the
ﬁgure illustrates is that the stringy material tends to ﬁll a region out to a
ﬁxed proper distance from the mathematical horizon at X
+
= 0. In other
words, unlike a point particle, the stringy substance is seen by a probe to
hover at a distance ∼
s
above the horizon. Once again this surprising result
is a direct consequence of arbitrarily high frequency ﬂuctuations implicit
in the stringy structure of matter. Note that if the string coupling satisﬁes
g
s
<< 1, the string length
s
can be much greater than the Planck length,
P
=
s
g. In this case it is the string length and not the Planck length
that controls the distance of the stretched horizon from the mathematical
horizon.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Strings 163
Supplement: Light cone gauge ﬁxing of longitudinal string
motions
The coordinate X
−
(τ,σ) is not an independent degree of freedom, but is
given by equation 14.4.31. To see this multiply 14.4.31 by 1 =
∂X
+
∂τ
, which
is true for light cone coordinates. Thus we must check the validity of the
following equation
∂X
+
∂τ
∂X
−
∂σ
−
∂X
j
∂τ
∂X
j
∂σ
= 0. (14.4.34)
The content of this equation expresses the underlying invariance of string
theory to a reparameterization of the σ coordinate. Under the transforma
tion
σ →σ + δσ
the X
s transform as
X →X +
∂X
∂σ
δσ. (14.4.35)
The Noether charge for this invariance is exactly the quantity in equation
14.4.34. Setting this quantity to zero insures that the physical spectrum
consists only of states which are invariant under σ reparameterization. In
going to the light cone frame the constraint serves to deﬁne X
−
in terms of
the transverse coordinates. However even in the light cone gauge there is
a bit of residual gauge invariance, namely shifting σ by a constant. In the
light cone frame only the transverse X
s are dynamical and the generator
of these rigid shifts is
σ
=0
σ
’=0
vs
Fig. 14.3 String parameter translational invariance
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
164 Black Holes, Information, and the String Theory Revolution
dσ
∂X
j
∂σ
dX
j
∂τ
(14.4.36)
which by equation 14.4.31 is equal to
dσ
∂X
−
∂σ
. (14.4.37)
Setting this to zero simultaneously insures invariance under shifts of σ and
periodicity of X
−
(σ).
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Chapter 15
Entropy of Strings and Black Holes
The Bekenstein–Hawking entropy of black holes points to some kind of
microphysical degrees of freedom, but it doesn’t tell us what they are. A
real theory of quantum gravity should tell us and also allow us to com
pute the entropy by quantum statistical mechanics, that is, counting mi
crostates. In this lecture we will see to what extent string theory provides
the microstructure and to what extent it enables us to compute black hole
entropy microscopically.
String theory has many diﬀerent kinds of black holes, some in 3 + 1
dimensions, some in higher dimensions. The black holes can be neutral or
be charged with the various charges that string theory permits. We will see
that for the entire range of such black holes, the statistical mechanics of
strings allows us to compute the entropy up to numerical factors of order
unity. In every case the results nontrivially agree with the Bekenstein–
Hawking formula. What is more, in one or two cases in which the black
holes are invariant under a large amount of supersymmetry the calculations
can be reﬁned and give the exact numerical coeﬃcients. All of this is in
cases where quantum ﬁeld theory would give an inﬁnite result.
Because string theory is not necessarily a 4dimensional theory, it is
worth exploring the connections between strings and black holes in any
dimension. Let us begin with the formula for the entropy of a Schwarzschild
black hole in an arbitrary number of dimensions. Call the number of space
time dimensions D. The black hole metric found by solving Einstein’s
equation in D dimensions is given by
dτ
2
=
1 −
R
D−3
S
r
D−3
dt
2
−
1 −
R
D−3
S
r
D−3
−1
dr
2
− r
2
dω
D−2
(15.0.1)
165
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
166 Black Holes, Information, and the String Theory Revolution
The horizon is deﬁned by
R
S
=
16π(D−3)GM
Ω
D−2
(D −2)
1
D−3
(15.0.2)
and its D2 dimensional “area” is given by
A = R
D−2
S
dΩ
D−2
= R
D−2
S
Ω
D−2
. (15.0.3)
Finally, the entropy is given by
S =
A
4G
=
(2GM)
D−2
D−3
Ω
D−2
4G
. (15.0.4)
The entropy in equation 15.0.4 is what is required by black hole thermody
namics.
Supplement: Schwarzschild geometry in D = d +1
dimensions
To extend a static spherically symmetric geometry to D = d +1 dimen
sions, the metric can be assumed to be of the form
ds
2
= −e
2Φ
dt
2
+ e
2∆
dr
2
+r
2
dθ
2
1
+ sin
2
θ
1
dθ
2
2
+ ... + sin
2
θ
1
...sin
2
θ
d−2
dθ
2
d−1
.
(15.0.5)
Using orthonormal coordinates, the G
ˆ
t
ˆ
t
component of the Einstein tensor
can be directly calculated to be of the form
G
ˆ
t
ˆ
t
= −
¸
(D −2)∆
e
−2∆
r
+
(D −2)(D−3)
2r
2
(1 −e
−2∆
)
(15.0.6)
From Einstein’s equation for ideal pressureless matter, G
ˆ
t
ˆ
t
= κρ. This
means
G
ˆ
t
ˆ
t
= −
(D −2)
2r
D−2
d
dr
r
D−3
(1 −e
−2∆
)
= −κρ (15.0.7)
which can be solved to give
(1 −e
−2∆
)r
D−3
=
2κ
D−2
r
0
ρ(r
)r
D−2
dr
=
2κ
(D −2)
M
Ω
D−2
(15.0.8)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy of Strings and Black Holes 167
where the solid angle is given by Ω
D−2
=
2π
(D−1)/2
Γ((D−1)/2)
. The G
ˆ r
ˆ r
component
of the Einstein tensor satisﬁes
G
ˆ r
ˆ r
= −
−(D−2)Φ
e
−2∆
r
+
(D−2)(D−3)
2r
2
(1 −e
−2∆
)
= −
−(D−2)(Φ
+ ∆
)
e
−2∆
r
+κρ
(15.0.9)
For pressureless matter in the exterior region (ρ = 0 = P), we can im
mediately conclude that Φ = −∆. Deﬁning the Schwarzschild radius
R
D−3
S
=
2κM
(D−2)Ω
D−2
we obtain the form of the metric
e
−2∆
= 1 −
R
S
r
D−3
= e
2Φ
(15.0.10)
If we write F(r) ≡ e
2Φ
, a useful shortcut for calculating the solution to Ein
stein’s equation 15.0.7 is to note its equivalence to the Newtonian Poisson
equation in the exterior region
∇
2
F(r) = −κρ , F(r) = 1 + 2φ
Newton
. (15.0.11)
The Hawking temperature can be calculated by determining the dimen
sional factor between the Rindler time and Schwarzschild time. Near the
horizon, the proper distance to the horizon is given by
ρ =
2R
S
(D −3)
r
R
s
D−3
−1 (15.0.12)
which gives the relation between Rindler time/temperature units and
Schwarzschild time/temperature units
dω =
(D −3)
2R
S
dt (15.0.13)
Thus, the Hawking temperature of the black hole is given by
T
Hawking
=
1
2π
(D −3)
2R
S
. (15.0.14)
Using the ﬁrst law of thermodynamics, the entropy can be directly calcu
lated to be of the form
S =
2π(D −3)A
κ
(15.0.15)
Substituting the form κ = 8π(D−3)G for the gravitational coupling gives
the previous results in Ddimensions.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
168 Black Holes, Information, and the String Theory Revolution
All calculations of entropy in string theory make use of a well known
trick of quantum mechanics. The trick consists of identifying some kind
of control parameter that can be adiabtically varied. In the process of
adiabatic variation, energy levels are neither created nor destroyed. Thus
if we can follow the system to a value of the control parameter where the
system is tractable we can count the states easily even if the nature of the
object changes during the variation. Basically we are using the quantum
analog of the method of adiabatic invariants.
The trick in the string theory context is to vary the strength of the string
coupling adiabatically until we arrive at a point where the gravitational
forces are so weak that the black hole “morphs” into some more tractable
object. Thus we begin with a black hole of mass M
o
in a theory with string
coupling g
o
. Adiabatically varying a control parameter like g
o
will cause a
change in the black hole mass and other internal structural features. But
such a variation will not alter its entropy. Entropy is an adiabatic invariant.
Let us imagine decreasing the string coupling g. What happens to the
black hole as g tends to zero? The answer is obvious. It must turn into
a collection of free strings. String theory has all kinds of nonperturbative
objects, branes of various dimensionality such as membranes, Dbranes,
monopoles, and so on. But only the free strings have ﬁnite energy in the
limit g →0. Therefore a neutral black hole must evolve into a collection of
free strings. A very massive black hole might evolve into a large number of
low mass strings or, at the opposite extreme, a single very highly excited
string.
Very highly excited free strings have an enormously rich spectrum. They
can be thought of as a mass of tangled string that forms a timevarying
random walk in space. Such random walking strings have a large entropy
and can be studied statistically.
The entropy of a string of mass m can be calculated by returning to the
light cone quantization of the previous lecture. For any eigenstate of the
Hamiltonian with vanishing transverse momentum and unit P
−
the light
cone energy is m
2
/2.
On the other hand the quantization of the string deﬁnes a 1+1 dimen
sional quantum ﬁeld theory in which the (D2) transverse coordinates X
i
(σ)
play the role of free scalar ﬁelds. The spatial coordinate of this ﬁeld theory
is σ
1
, and it runs from 0 to 2π.
The counting of the states of a free string is best done in the light
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy of Strings and Black Holes 169
cone version of the theory that we discussed in the last lecture. In order to
describe the highly excited string spectrum, a formal light cone temperature
T can be deﬁned. Recall that the free string is described by means of a
1 + 1 dimensional quantum theory containing D −2 ﬁelds X
i
.
The entropy and energy of such a quantum ﬁeld theory can be calculated
by standard means. The leading contribution for large energy is (setting
s
= 1)
E = π T
2
(D −2)
S = 2πT (D −2)
(15.0.16)
Using E =
m
2
2
and eliminating the temperature yields S =
2(D−2)π m
or, restoring the units
S =
2(D−2)π m
s
. (15.0.17)
Subleading corrections can also be calculated to give
S =
2(D−2)π m
s
− c log (m
s
) (15.0.18)
where c is a positive constant. The entropy is the log of the density of
states. Therefore the number of states with mass m is
N
m
=
1
m
s
c
exp
2π(D−2) m
s
(15.0.19)
The formula 15.0.19 is correct for the simplest bosonic string, but similar
formulae exist for the various versions of superstring theory.
Now let us compare the entropy of the single string with that of n
strings, each carrying mass
m
n
. Call this entropy S
n
(m). Then
S
n
(m) = nS(m/n) (15.0.20)
or
S
n
(m) =
2(D−2)π m
s
− nc log
m
s
n
(15.0.21)
Obviously for large n the single string is favored. This is actually quite
general. For a given total mass, the statistically most likely state in free
string theory is a single excited string. Thus it is expected that when the
string coupling goes to zero, most of the black hole states will evolve into
a single excited string.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
170 Black Holes, Information, and the String Theory Revolution
These observations allow us to estimate the entropy of a black hole. The
assumptions are the following:
• A black hole evolves into a single string in the limit g →0
• Adiabatically sending g to zero is an isentropic process; the entropy of
the ﬁnal string is the same as that of the black hole
• The entropy of a highly excited string of mass m is of order
S ∼ m
s
(15.0.22)
• At some point as g →0 the black hole will make a transition to a string.
The point at which this happens is when the horizon radius is of the
order of the string scale.
To understand this last assumption begin with a massive black hole.
Gravity is clearly important and cannot be ignored. But no matter how
massive the black hole is, as we decrease g a point will come where the
gravitational constant is too weak to matter. That is the point where the
black hole makes a transition and begins to act like a string.
The string and Planck length scales are related by
g
2
D−2
s
=
D−2
p
. (15.0.23)
Evidently as g decreases the string length scale becomes increasingly big in
Planck units. Eventually, at some value of the coupling that depends on
the mass of the black hole, the string length will exceed the Schwarzschild
radius of the black hole. This is the point at which the transition from black
hole to string occurs. In what follows we will vary the g while keeping ﬁxed
the string length
s
. This implies that the Planck length varies.
Let us begin with a black hole of mass M
o
in a string theory with
coupling constant g
o
. The Schwarzschild radius is of order
R
S
∼ (M
o
G)
1
D−3
, (15.0.24)
and using
G ≈ g
2
D−2
s
(15.0.25)
we ﬁnd
R
S
s
≈
s
M
o
g
2
o
1
D−3
. (15.0.26)
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy of Strings and Black Holes 171
Thus for ﬁxed g
o
if the mass is large enough, the horizon radius will be
much bigger than
s
.
Now start to decrease g. In general the mass will vary during an adia
batic process. Let us call the gdependent mass M(g). Note
M(g
o
) = M
o
(15.0.27)
The entropy of a Schwarzschild black hole (in any dimension) is a func
tion of the dimensionless variable M
P
. Thus, as long as the system re
mains a black hole,
M(g)
P
= constant. (15.0.28)
Since
P
≈
s
g
2
D−2
we can write equation 15.0.28 as
M(g) = M
o
g
2
o
g
2
1
D−2
. (15.0.29)
Now as g → 0 the ratio of the gdependent horizon radius to the string
scale decreases. From equation 15.0.2 it becomes of order unity at
M(g)
D−2
P
≈
D−3
s
(15.0.30)
which can be written
M(g)
s
≈
1
g
2
. (15.0.31)
Combining equations 15.0.29 and 15.0.31 we ﬁnd
M(g)
s
≈ M
D−2
D−3
o
G
1
D−3
o
. (15.0.32)
As we continue to decrease the coupling, the weakly coupled string
mass will not change signiﬁcantly. Thus we see that a black hole of mass
M
o
will evolve into a free string satisfying equation 15.0.32. But now we
can compute the entropy of the free string. From equation 15.0.22 we ﬁnd
S ≈ M
D−2
D−3
o
G
1
D−3
o
. (15.0.33)
This is a very pleasing result in that it agrees with the Bekenstein–Hawking
entropy in equation 15.0.4. However, in this calculation the entropy is
calculated as the microscopic entropy of fundamental strings.
The evolution from black hole to string can be pictorially represented
by starting with a large black hole. The stretched horizon is composed of a
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
172 Black Holes, Information, and the String Theory Revolution
R
S
c
h
w
a
r
z
s
c
h
i
l
d
S
Fig. 15.1 Evolution from black hole to string. (a) A black hole with stringy
stretched horizon smaller than Schwarzschild radius, (b) with stretched horizon
and string scale comparable to radius scale, and (c) turned into a string
stringy mass to a depth of ρ =
s
as in the diagram Figure 15.1a. The area
density of string is saturated at ∼
1
G
. Another important property of the
stretched horizon is its proper temperature. Since the proper temperature
of a Rindler horizon is
1
2πρ
, the temperature of the stringy mass will be
T
Stretched
≈
1
s
This temperature is close to the socalled Hagedorn temperature, the max
imum temperature that a string can achieve.
As the Schwarzschild radius is decreased (in string units), the area of
the horizon decreases but the depth of the stretched horizon stays ﬁxed
as in Figure 15.1b. Finally the horizon radius is no larger than
s
(Figure
15.1c) and the black hole turns into a string.
By now a wide variety of black holes that occur in string theory have
been analyzed in this manner. The method is always the same. We adiabat
ically allow g to go to zero and identify the appropriate string conﬁguration
that the black hole evolves into.
A particularly interesting situation is that of charged extremal black
holes which may be supersymmetric conﬁgurations of a supersymmetric
theory. In this case the extremal black hole is absolutely stable and in
addition, its mass is completely determined by supersymmetry. When this
occurs there is no need to follow the mass of the black hole as g varies;
the mass is ﬁxed. Under these conditions the black hole can be compared
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Entropy of Strings and Black Holes 173
directly to the corresponding weakly coupled string conﬁguration and the
entropy read oﬀ from the degeneracy of the string theory spectrum. In the
cases where exact calculations are possible the charges carried by black holes
are associated not with fundamental strings but Dbranes. Nevertheless the
principles are that same as those that we used to study the Schwarzschild
black hole in D dimensions. The results in these more complicated examples
are in precise agreement with the Hawking–Bekenstein entropy.
Hagedorn Temperature Supplement
On general grounds, one can determine the density of states η for the
various string modes m:
η(m) ∼ exp(4πm
√
α
)
This allows the partition function to be written as
Z ∼
∞
0
exp(4πm
√
α
)exp
−
m
T
dm
which diverges if the temperature T is greater than the Hagedorn temper
ature deﬁned by
T
Hagedorn
≡
1
4π
√
α
The Hagedorn temperature scales with the inverse string length
T
Hagedorn
∼
1
l
s
.
To get a feel for the scale of the Hagedorn temperature, recall the be
havior of the entropy given by S∼log(density of states). Using dimensional
considerations, we have seen that the entropy of the string scales like
S
string
∼
d
f
M
s
l
s
,
where d
f
is the number of internal degrees of freedom available. Thus, the
density of states behaves like
e
S
string
∼ e
√
d
f
M
s
l
s
∼ e
1/T
Hagedorn
which gives the scale T
Hagedorn
∼ 1/l
s
. If one examines multistring ﬂuc
tuations as a function of temperature, the Hagedorn temperature is the
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
174 Black Holes, Information, and the String Theory Revolution
“percolation” temperature for multiple strings ﬂuctuations to coalesce into
ﬂuctuations of a single string as represented in Figure 15.2.
Increase
Temp
Increase
Temp
T<TH T=TH
T>TH
Fig. 15.2 String “percolation”
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Conclusions
The views of space and time that held sway during most of the 20th century
were based on locality and ﬁeld theory, ﬁrst classical ﬁeld theory and later
quantum ﬁeld theory. The most fundamental object was the spacetime
point or better yet, the event. Although quantum mechanics made the
event probabilistic and relativity made simultaneity nonabsolute, it was
assumed that all observers would agree on the usual invariant relationships
between events. This view persisted even in classical general relativity. But
the paradigm is gradually shifting. It was never adequate to deal with the
combination of quantum mechanics and general relativity.
The ﬁrst sign of this was the failure of standard quantum ﬁeld the
ory methods when applied to the Einstein action. For a long time it was
assumed that this just meant that the theory was incomplete at short dis
tances in the same way that the Fermi theory of weak interactions was
incomplete. But the dilemma of apparent information loss in black hole
physics that was uncovered by Hawking in 1976 said otherwise. In order
to reconcile the equivalence principle with the rules of quantum mechanics
the rules of locality have to be massively modiﬁed. The problem is not a
pure ultraviolet problem but an unprecedented mix of short distance and
long distance physics. Radical changes are called for.
The new paradigm that is gradually emerging is based on four closely
related concepts. The ﬁrst is Black Hole Complementarity. This principle is
a new kind of relativity in which the location of phenomena depends on the
resolution time available to the experimenter who probes the system. An
extreme example would be the fate of someone, call her Alice, falling into an
enormous black hole with Schwarzschild radius of a billion years. According
to the low frequency observer, namely Alice herself, or someone falling with
her, nothing special is felt at the horizon. The horizon is harmless and she
175
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
176 Black Holes, Information, and the String Theory Revolution
or her descendants can live for a billion years before being crushed at the
singularity.
In apparent complete contradiction, the high frequency observer who
stays outside the black hole ﬁnds that his description involves Alice falling
into a hellish region of extreme temperature, being thermalized, and even
tually reemitted as Hawking radiation. All of this takes place just outside
the mathematical horizon. Obviously this has to do with more than just a
modiﬁcation of the short distance physics. As we have seen, the key to black
hole complementarity is the extreme red shift of the quantum ﬂuctuations
as seen by the external observer.
The second new idea is the Infrared/Ultraviolet connection. Very closely
related to Black Hole Complementarity, the IR/UV connection reverses one
of the most fundamental trends of 20th century physics. Throughout that
century a close connection between energy and size prevailed. If one wished
to study progressively smaller and smaller objects one had to use higher and
higher energy probes. But once gravity is involved that trend is reversed.
At energies above the Planck scale any possible short distance physics that
we might look for is shrouded behind a black hole horizon. As we raise the
energy we wind up probing larger and larger distance scales. The ultimate
implications of this, especially for cosmology are undoubtedly profound but
still unknown.
Third is the Holographic Principle. In many ways this is the most
surprising ingredient. The nonredundant degrees of freedom that describe
a region of space are in some sense on its boundary, not its interior as they
would be in ﬁeld theory. At one per Planck area, there are vastly fewer
degrees of freedom than in a ﬁeld theory, cutoﬀ at the Planck volume. The
number of degrees of freedom per unit volume becomes arbitrarily small as
the volume gets large. Although the Holographic Principle was regarded
with skepticism at ﬁrst it is now part of the mainstream due to Maldacena’s
AdS/CFT duality. In this framework the Holographic Principle, Black
Hole Complementarity and the IR/UV connection are completely manifest.
What is less clear is the dictionary for decoding the CFT hologram.
Finally, the existence of black hole entropy indicates the existence of
microscopic degrees of freedom which are not present in the usual Ein
stein theory of gravity. It does not tell us what they are. String theory
does provide a microscopic framework for the use of statistical mechanics.
In all cases the entropy of the appropriate string system agrees with the
Bekenstein–Hawking entropy. This, if nothing else, provides an existence
proof for a consistent microscopic theory of black hole entropy.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Conclusions 177
The theory of black hole entropy is incomplete. In each case a trick,
speciﬁc to the particular kind of black object under study, is used to deter
mine the relation between entropy and mass for the speciﬁc stringtheoretic
object that is believed to represent a particular black hole. Then classi
cal general relativity is used to determine the area–mass relation and the
Bekenstein–Hawking entropy. In no case do we use string theory directly
to compare entropy and area. In this sense the complete universality of the
area–entropy relation is still not fully understood.
One very large hole in our understanding of black holes is how to think
about the observer who falls through the horizon. Is this important? It is
if you are that observer. And in some ways, an observer in a cosmological
setting is very much like one behind a horizon. At the time of the writing
of this book there are no good ideas about the quantum world behind the
horizon. Nor for that matter is there any good idea of how to connect the
new paradigm of quantum gravity to cosmology. Hopefully our next book
will have more to say about this.
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
178 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Bibliography
1. Kip S. Thorne, Richard H. Price and Douglas A. MacDonald.
Black Holes: The Membrane Paradigm (Yale University Press, 1986).
2. Don N. Page. Information in Black Hole Radiation, hepth/9306083, Phys.
Rev. Lett. 71 (1993) 3743–3746.
3. Leonard Susskind. The World as a Hologram, hepth/9409089, J. Math. Phys.
36 (1995) 6377–6396.
4. S. Corley and T. Jacobson. Focusing and the Holographic Hypothesis, gr
qc/9602043, Phys. Rev. D53 (1996) 6720–6724.
5. W. Fischler and L. Susskind. Holography and Cosmology, hepth/9806039.
6. Raphael Bousso. The Holographic Principle, hepth/0203101, Rev. Mod.
Phys. 74 (2002) 825–874.
7. Juan M. Maldacena. The Large N Limit of Superconformal Field Theories
and Supergravity, hepth/9711200, Adv. Theor. Math. Phys. 2 (1998) 231–
252; Int. J. Theor. Phys. 38 (1999) 1113–1133.
8. Edward Witten. Anti De Sitter Space and Holography, hepth/9802150, Adv.
Theor. Math. Phys. 2 (1998) 253–291.
9. L. Susskind and Edward Witten. The Holographic Bound in Antide Sitter
Space, hepth/9805114.
10. T. Banks, W. Fischler, S.H. Shenker and L. Susskind. M Theory as a Matrix
Model: A Conjecture, hepth/9610043, Phys. Rev. D55 (1997) 5112–5128.
179
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
180 Black Holes, Information, and the String Theory Revolution
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Index
AdS(5) ⊗S(5), 128, 141
’t Hooft limit, 135
adiabatic invariants, 168
adiabatic variation (string coupling),
170
AdS black hole, 144
AdS/CFT correspondence, 133
angular momentum, 27
anti de Sitter space, 123, 128
barrier penetration, 28
baryon number violation, 89
Birkoﬀ’s theorem, 17
black hole formation, 15
Boltzmann factor, 40
Bousso’s construction, 114, 120, 121
brick wall, 84
caustic lines, 106
charge, 55
charged black holes, 55
classical ﬁelds, 31
coarse graining, 69
collapsing lightlike shell, 103
complementarity, 85, 97, 157, 160
conductivity, 68
conformal ﬁeld theory, 131
conformally ﬂat, 7
cosmological constant, 119
curvature, 5
Dbranes, 136
de Sitter space, 119
degrees of freedom, 138
density matrix, 34, 45, 71
eﬀective potential, 26
electrical properties, 62
electromagnetic ﬁeld, 62
electrostatics, 63
energy (horizon), 52
entanglement, 32, 71, 85
entanglement entropy (equality), 72
entropy, 52, 61, 102, 165
entropy (Bekenstein–Hawking), 51,
165
entropy (bounds), 101
entropy (calculation), 36, 43, 168
entropy (coarse grained), 73, 76
entropy (ﬁne grained), 70, 73
entropy (inﬁnite), 81
entropy (maximum), 102
entropy (quantum ﬁeld theory), 81
entropy (strings), 165, 168
entropy (thermal), 35, 46, 73
entropy (vacuum), 45
entropy (Von Neumann), 35, 70, 76
entropy of entanglement, 35, 46, 71
equivalence principle, 21, 31, 69, 77
evaporation, 48
evaporation time, 54
expansion rate, 113
extended horizon, 11
181
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
182 Black Holes, Information, and the String Theory Revolution
extremal black hole, 56
Feynman–Hellman theorem, 37
Fidos, 21, 32, 39, 41, 48, 58, 161
ﬁducial observer (see also Fidos), 21
Fischler–Susskind bound, 110
ﬂuctuations, 41, 60, 93
focusing theorem, 107
free ﬁeld approximation, 49
freely falling observer (see also
Frefos), 22
Frefos, 21, 41, 85
Friedman–Robertson–Walker
geometry, 110
gauge ﬁxing, 22
geodesic completeness, 129
ground state (charged black hole), 59
Hagedorn temperature, 172
Hawking, 81
Hawking radiation, 49, 85
high frequency phenomena, 151
holographic principle, 101, 127, 130
holography, 101, 127
holography (AdS space), 130
horizon, 4, 8, 20, 25, 44, 52, 57, 144,
152
hyperbolic plane, 130
infalling observer, 5
inﬂation, 121
information, 74, 97, 144
information conservation, 69, 81
information retention time, 77
Kruskal–Szekeres coordinates, 10
lattice of discrete spins, 101
laws of nature, 69
level density, 52
light cone gauge ﬁxing, 163
light cone quantum mechanics, 153
light cone string theory, 156
Liouville’s theorem, 69
longitudinal string motions, 161
luminosity, 54
Minkowski space, 9, 106
mirror boundary condition, 44
momentum, 23, 32
near horizon coordinates, 8
near horizon wave equation, 28
Newton’s constant, 134
nocloning principle, 79
pair production, 55
particle horizon, 112
path integral, 37
Penrose diagram (AdS black hole),
142
Penrose diagram (de Sitter space),
120
Penrose diagram (F.R.W. space), 115
Penrose diagrams, 14
Penrose–Bousso diagram, 117
Penrose–Bousso diagram (AdS), 123
Planck distance, 23
Planck length, 52, 62, 134, 170
Planck units, 127
Poincar´e disk, 130
proper distance, 8, 167
proton decay, 89
quantization rules, 44
quantum ﬁeld theory, 43
quantum ﬁelds, 25, 61, 102, 151
quantum ﬁelds (Rindler space), 31
quantum Xerox principle, 69, 79
red shift, 48
reheating, 121
Reissner–Nordstrom black hole, 55
resistance, 65
Rindler energy, 53
Rindler Hamiltonian, 32, 44
Rindler space, 8, 31, 152, 167
Smatrix, 81
scalar wave equation, 25
Schwarzschild black hole, 3
October 25, 2004 15:0 WSPC/Book Trim Size for 9in x 6in blkhlphy
Index 183
Schwarzschild coordinates, 3
Schwarzschild geometry
(Ddimensions), 166
second law of thermodynamics, 103
solid angle, 167
standard thermometer, 39
static observers, 21
stretched horizon, 42, 61, 62, 161
string coupling constant, 134
string interactions, 159
string percolation, 174
strings, 151
Supersymmetric Yang–Mills (SYM),
133
supersymmetry, 131, 133
surface charge density, 63
temperature (Hawking), 48, 167
temperature (proper), 39
temperature (Rindler), 39
thermal atmosphere, 45, 48
thermal ensemble, 39
thermal ﬂuctuations, 144
thermodynamic instability, 141
thermodynamics, 51
tidal forces, 5
tortoise coordinates, 3, 7, 65
tortoiselike coordinates, 28
transfer matrix, 38
transverse spreading, 155
Unruh, 36, 39
UV/IR connection, 95, 101, 135
vacuum, 45
.
INFORMATION. Stanford University. DC 20059 . CA 943054060 † Permanent address. Department of Physics. Stanford. Washington. Department of Physics. Howard University. AND THE STRING THEORY REVOLUTION The Holographic Universe Leonard Susskind∗ James Lindesay† ∗ Permanent address.AN INTRODUCTION TO BLACK HOLES.
vi .
All rights reserved. Danvers. . Covent Garden. Ltd. Singapore 596224 USA office: 27 Warren Street. or parts thereof. In this case permission to photocopy is not required from the publisher.. MA 01923. 5 Toh Tuck Link.Published by World Scientific Publishing Co. without written permission from the Publisher. Pte. may not be reproduced in any form or by any means. NJ 07601 UK office: 57 Shelton Street. electronic or mechanical. Inc. INFORMATION AND THE STRING THEORY REVOLUTION The Holographic Universe Copyright © 2005 by World Scientific Publishing Co. Pte. Ltd. ISBN 9812560831 ISBN 9812561315 (pbk) Printed in Singapore. 222 Rosewood Drive. London WC2H 9HE British Library CataloguinginPublication Data A catalogue record for this book is available from the British Library. For photocopying of material in this volume. Hackensack. including photocopying. This book. USA. please pay a copying fee through the Copyright Clearance Center. Suite 401402. recording or any information storage and retrieval system now known or to be invented. AN INTRODUCTION TO BLACK HOLES.
One has the distinct impression that we are thinking about the things in the wrong way. The puzzle which started the whole thing was a contradiction between the principles of statistical thermodynamics and the ﬁeld concept of Faraday and Maxwell. Planck asked. The methods that were invented to quantize electrodynamics. The subject is riddled with paradox and contradiction. or spectral lines.Preface It is now almost a century since the year 1905. in the 80 years that we have had the general theory of relativity. By contrast. How was it possible. Such strategy has worked in the past. How then are we to go about ﬁnding the right replacement? It seems very unlikely that the usual incremental increase of knowledge from a combination of theory and experiment will ever get us where we want to go. radioactivity. It has taken most of that time to synthesize the two into the modern quantum theory of ﬁelds and the standard model of particle phenomena. Under this circumstance our best hope is an examination of fundamental principles. The paradigm of relativistic quantum ﬁeld theory almost certainly has to be replaced. to the Planck scale. The earliest origins of quantum mechanics were not experimental atomic physics. which were so successfully generalized to build the standard model. for the inﬁnite collection of radiation oscillators to have a ﬁnite speciﬁc heat? vii . It is unlikely that a major revolution will spring from this soil. it seems likely that we know most of the basic principles which follow from combining the special theory of relativity with quantum mechanics. in which the principle of relativity and the hypothesis of the quantum of radiation were introduced. nothing comparable has been learned about the quantum theory of gravitation. that is. prove wholly inadequate when applied to gravitation. Although there is undoubtably more to be learned both theoretically and experimentally. and the study of gedanken experiments. paradoxes and contradictions.
But no such solution to Maxwell’s equations exists. one can do the bookkeeping so that it is the black hole itself which is correlated to the matter outside. This in itself is a source of paradox. The development of the general theory from the principle of equivalence and the manintheelevator gedanken experiment is also a matter of historical fact.viii Black Holes. consistent with causality. For two decades this contradiction between . But eventually the black hole will evaporate. Bekenstein realized that if the second law of thermodynamics was not to be violated in the presence of a black hole. As long as the black hole is present. at the age of 15 he formulated the following paradox: suppose an observer moved along with a light beam and observed it. Eventually the black hole must completely evaporate. In each of these cases the consistency of readily observed properties of nature which had been known for many years required revolutionary paradigm shifts. Thus. What known properties of nature should we look to. According to Einstein. How and why a classical solution of ﬁeld equations should be endowed with thermodynamical attributes has remained obscure since Bekenstein’s discovery in 1972. the two theories appear to lead to a serious clash that once again involves statistical thermodynamics in an essential way. Information. The paradox was discovered by Jacob Bekenstein and turned into a serious crisis by Stephen Hawking. and the String Theory Revolution In the case of special relativity it was again a conceptual contradiction and a gedanken experiment which opened the way. and which paradox is best suited to our present purposes? Certainly the most important facts are the success of the general theory in describing gravity and of quantum mechanics in describing the microscopic world. the black hole must possess an intrinsic entropy. The electromagnetic ﬁeld would be seen as a static. By an analysis of gedanken experiments. according to Hawking. Furthermore. By this simple means a contradiction was exposed between the symmetries of Newton’s and Galileo’s mechanics and those of Maxwell’s electrodynamics. Hawking added to the puzzle when he discovered that a black hole will radiate away its energy in the form of Planckian black body radiation. spatially varying ﬁeld. for the correlations to be carried by the outgoing evaporation products. Hawking then made arguments that there is no way. Hawking then raised the question of what becomes of the quantum correlations between matter outside the black hole and matter that disappears behind the horizon. the existence of black holes inevitably causes a loss of quantum coherence and breakdown of one of the basic principles of quantum mechanics – the evolution of pure states to pure states.
including ’t Hooft and Susskind. Quite apart from the question of the ultimate correctness and consistency of string theory. Hawking further argued that once the loss of quantum coherence is permitted in black hole evaporation. The world would behave as if it were in a noisy environment which continuously leads to a loss of coherence. The main purpose of this book is to lay out this case. with or without strings. To borrow words from Sidney Coleman. the way to the new paradigm will be opened. It is therefore worth exploring the diﬀerences between string theory and ﬁeld theory in the context of black hole paradoxes. Although still very incomplete. have felt that the basic principles of quantum mechanics and statistical mechanics have to be made to coexist with black hole evaporation. To navigate it without disaster we will need some beacons in the form of trusted principles that we can turn to for direction. matter. at the same time violating energy conservation by heating the world. it becomes compulsory in all processes involving the Planck scale. The trouble with this is that there is no known way to destroy coherence without. Throughout this period. there are important lessons to be drawn from the diﬀerences between these two theories. Peskin and Susskind. in the neighborhood of a black hole horizon the diﬀerences become extreme. and information. Hawking and much of the traditional relativity community have been of the opinion that the correct resolution of the paradox is simply that quantum coherence is lost during black hole evaporation. it is a “trackless swamp” with many false but seductive paths and no maps. string theory appears to be a far more consistent mathematical framework for quantum gravity than ordinary ﬁeld theory. The analysis of these diﬀerences suggests a resolution of the black hole dilemma and a completely new view of the relations between space.Preface ix the principles of general relativity and quantum mechanics has so puzzled theorists that many now see it as a serious crisis. ’t Hooft has argued that by resolving the paradox and removing the contradiction. a few theorists. As we shall see. and ’t Hooft. time. The quantum theory of black holes. A second purpose involves development of string theory as a uniﬁed description of elementary particles. including their gravitational interactions. although string theory is usually well approximated by local quantum ﬁeld theory. The theory is out of control as argued by Banks. is far from being a textbook subject with well deﬁned rules. In this book the absolute truth of the following four propositions will be . From an operational viewpoint this would mean that the standard rules of quantum mechanics would not apply to processes involving black holes.
beginning with asymptotic infalling objects and ending with asymptotic outgoing evaporation products is consistent with the existence of a unitary Smatrix. Thermodynamics results from coarse graining a more microscopic description so that states with similar macroscopic behavior are lumped into a single thermodynamic state. More speciﬁcally we assume that a thermodynamic entropy S implies that approximately exp(S) quantum states have been lumped into one thermal state. but it nevertheless rests on wellestablished foundations. then a freely falling observer should detect nothing out of the ordinary when passing the horizon. it must also have a luminosity. In particular. luminosity. and approximate black body character of Hawking radiation.x Black Holes. Proposition 1 and 3 apply to all known forms of matter. Furthermore the existence of a thermal behavior in the vicinity of the horizon follows from the equivalence principle as shown in the fundamental paper of Unruh. If the horizon scale is large enough so that tidal forces can be ignored. and shape of a horizon depend not only on past occurrences. The existence of a thermodynamics will be taken to mean that a microscopic set of degrees of freedom exists whose coarse graining leads to the thermal description. taken by themselves. These three propositions. 2) The usual semiclassical description of quantum ﬁelds in a slowly varying gravitational background is a good approximation to certain coarse grained features of the black hole evolution. and temperature. The usual laws of nature with no abrupt external perturbations will be found valid until the inﬂuence of the singularity is encountered. We ourselves could right now be at the horizon of a gigantic black . The global process. and the String Theory Revolution assumed: 1) The formation and evaporation of a black hole is consistent with the basic principles of quantum mechanics. Information. 3) Thirdly we assume the usual connection between thermodynamics and quantum statistical mechanics. are in no way radical. carrying their laboratories with them. Those features include the thermodynamic properties. entropy. energy momentum ﬂux. The existence. Why then should any of these principles be considered controversial? The answer lies in a fourth proposition which seems as inevitable as the ﬁrst three: 4) The fourth principle involves observers who fall through the horizon of a large massive black hole. Proposition 2 may perhaps be less obvious. size. this means that observations performed by observers who remain outside the black hole can be described by a unitary time evolution. but also on future events. location. Once we admit that a black hole has energy. In considering the validity of this fourth proposition it is important to keep in mind that the horizon is a global concept.
Beware the willo’thewisp and don’t lose your nerve. In Chapter 9 we will encounter powerful arguments against the mutual consistency of propositions 1–4. while all around false paths beckon. The true path through the swamp at times becomes so narrow it seems to be a dead end.Preface xi hole caused by matter yet to collapse in the future. . The horizon in classical relativity is simply the mathematical surface which separates those points from which any light ray must hit a singularity from those where light may escape to inﬁnity. A mathematical surface of this sort should have no local eﬀect on matter in its vicinity.
xii Black Holes. and the String Theory Revolution . Information.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proper Temperature . . . . . . . . .3 1. .2 3. . . . . . . . . . . . . 1 3 3 7 8 10 14 15 21 25 28 31 2. xiii . . . . . . Tortoise Coordinates . . . . . . . . . Scalar Wave Equation in a Schwarzschild Background 2. . . . . . . . . . . . . . . . . . . . .1 1. . . .1 Black Hole Evaporation . . . . .6 1.1 3. . . The Schwarzschild Black Hole 1. . . . . . . . . .2 1.7 Schwarzschild Coordinates . . . 31 32 34 36 39 43 48 4. . . . . . . . . . . . . Fidos and Frefos and the Equivalence Principle . . . . . . . . . . . . . . . . . . .5 1. . . . . . . . . . . . . . . . 3. . . . . . . . . . . . . . . . .4 1. . . Entanglement . . . . . . . . . . . .4 3. . . . . . . .5 Classical Fields . . . . . . . . . .Contents Preface vii Part 1: Black Holes and Quantum Mechanics 1. . . . . . . . . . . . . Review of the Density Matrix The Unruh Density Matrix . . Kruskal–Szekeres Coordinates . . . . . . Quantum Fields in Rindler Space 3. Penrose Diagrams . . . . . . . . . . . . Near Horizon Coordinates (Rindler space) . . . . . . .1 Near the Horizon . . Entropy of the Free Quantum Field in Rindler Space 4. . . . . . . . . . . . . . .3 3. . Formation of a Black Hole . . . . .
1 11. . . . . . Baryon Number Violation . . . .3 12. . Thermodynamics of Black Holes 6. . . . . . . . . . de Sitter Cosmology . . . . . . . . . . . . . . . . . The Infrared Ultraviolet Connection Counting Degrees of Freedom . . . . . The Laws of Nature 8.3 11. . . . . . Horizons and the UV/IR Connection Part 2: Entropy Bounds and Holography 11. . . . . . .3 8. . . . . . . . . . Quantum Xerox Principle . . . . 81 84 85 89 95 10. . . . . . . . . . . . . . . . . . . . . . . . . The Holographic Principle and Anti de Sitter Space 12.4 12.2 12. . . . . . . . . .xiv Black Holes. . . . .1 9. . . . . . . Entropy on Lightlike Surfaces . 51 55 61 69 69 71 77 79 9. . .2 9. . . . . . 127 128 130 133 135 138 12. Equivalence Principle . .4 Information Conservation Entanglement Entropy . . . . . . . . . . . . . . Black Hole Complementarity . . . . . . . . . . . . . . . . . . . . .5 11. . . . . . . . . . . . . . . . . . AdS Space .1 8. . . . . . . .2 8. . . . . . . . . . . . . . . . . . . . . . 99 101 101 105 110 114 119 123 127 . . . . . . . . . . . . . . . . . . . . The Stretched Horizon 8. . . . . . . . . . . . . . . . . . . . . The AdS/CFT Correspondence . . . . . . .5 12. . . . . . . . . . . . . . and the String Theory Revolution 5. . . . . . . . . . Holography in AdS Space . . . . . .6 Maximum Entropy . . . . .4 11. The Puzzle of Information Conservation in Black Hole Environments 9. . . . . . . . . . . . . . . . Entropy Bounds 11. . .2 11. . . . Anti de Sitter Space . . . . . . . .3 A Brick Wall? . . . . . . . . . . . . . . . . . . .6 The Holographic Principle . . . . . . . . .1 12. Information. . . . . Friedman–Robertson–Walker Geometry Bousso’s Generalization . Charged Black Holes 7. . . . . .
. . . . . . 144 Part 3: Black Holes and Strings 14. . . . . . . .2 14. . . . . . . . . . . . . . . Entropy of Strings and Black Holes Conclusions Bibliography Index . . . . . . . . . . . . . Longitudinal Motion . . . . . . . . . . . . . . . .4 Light Cone Quantum Mechanics Light Cone String Theory . . Strings 14. . . . . . . . . . . . . . . .Contents xv 13. . . . .2 141 The Horizon . .1 13. . . . . . 149 151 153 156 159 161 165 175 179 181 15. . . .1 14. . Interactions . . . . . .3 14. . Black Holes in a Box 13. . . . . . . . . . . . 144 Information and the AdS Black Hole . . . . . . .
PART 1 Black Holes and Quantum Mechanics .
.
Each coordinate system that we will study has its own particular utility. or the related tortoise coordinates.1) = gµν dxµ dxν . and it represents the time recorded by a standard clock at rest at spatial inﬁnity. The coordinate r is called the Schwarzschild radial coordinate. For example.1 Schwarzschild Coordinates In Schwarzschild coordinates. and no one of them is in any sense the best or most correct description.1. The coordinate t is called Schwarzschild time. It does not measure proper 3 . 1. For these purposes. the Kruskal– Szekeres coordinate system is valuable for obtaining a global overview of the entire geometry. Schwarzschild coordinates.Chapter 1 The Schwarzschild Black Hole Before beginning the study of the quantum theory of black holes. It can however be misleading when applied to observations made by distant observers who remain outside the horizon during the entire history of the black hole. one must ﬁrst become thoroughly familiar with the geometry of classical black holes in a variety of diﬀerent coordinate systems. We begin with the simplest spherically symmetric static uncharged black holes described by Schwarzschild geometry. the Schwarzschild geometry is manifestly spherically symmetric and static. The metric is given by dτ 2 = (1 − 2MG 2 r )dt − (1 − 2MG −1 2 dr r ) − r2 dΩ2 (1. which cover only the exterior of the horizon are in many ways more valuable. where dΩ2 ≡ dθ2 + sin2 θdφ2 .
is given by the coordinate r = 2M G.4 Black Holes.1.1. The horizon.4) R ( 2M G −1) + tan η 2 R ( 2M G −1)1/2 −tan η 2 [0 < η < π] where τ is the proper time recorded by the observer’s clock. The angles θ.1. To a distant observer the horizon represents the boundary of the world.3) t = ( R + 2M G) 2 +2M G log R 2MG 1/2 η + R 2 R 2MG −1 1/2 sinη (1.1.2) τ= R 2 R 2M G −1 1/2 (η + sinη) (1. For simplicity let’s consider a radially freely falling observer who is dropped from rest from the point r = R.5) Evidently the proper time when crossing the horizon is ﬁnite and smaller than the expression in equation 1.1. . In equation 1. which we will tentatively deﬁne as the place where g00 vanishes. At the horizon grr becomes singular. but is deﬁned so that the area of the 2sphere at r is 4πr2 . The question of whether the geometry is truly singular at the horizon or if it is the choice of coordinates which are pathological is subtle. In what follows we will see that no local invariant properties of the geometry are singular at r = 2M G. φ are the usual polar and azimuthal angles.5. Nevertheless there is a very important sense in which the horizon is globally special if not singular. From these overly complicated equations it is not too diﬃcult to see that the observer arrives at the point r = 0 after a ﬁnite interval π τ = R 2 R 2M G 1 2 (1.1.1 we have chosen units such that the speed of light is 1. and the String Theory Revolution spatial distance from the origin. Information. The trajectory of the observer in parametric form is given by r= R (1 + cosη) 2 1/2 (1. To determine whether the local geometry is singular at r = 2M G we can send an explorer in from far away to chart it. or at least that part which can inﬂuence his detectors. Thus a small laboratory in free fall at r = 2M G would record nothing unusual.
τ . According to classical physics the infalling observer can use an arbitrarily large carrier frequency to send an arbitrarily large amount of information using .4 is that the crossing of the horizon does not occur at any ﬁnite Schwarzschild time. Furthermore a signal originating at the horizon cannot reach any point r > 2M G until an inﬁnite Schwarzschild time has elapsed. The ﬁrst surprising thing we learn from equations 1. Thus the infalling observer passes smoothly and safely through the horizon. This is shown in Figure 1. t tends to inﬁnity. The nonvanishing curvature components ˆ are given by Rτ θˆθ = Rτ φˆφ = −Rρθρθ = −Rρφˆφ = ˆ ˆτ ˆ ˆ ˆτ ˆ ˆˆˆˆ ˆ ˆρ ˆ Rθφθˆ = −Rτ ρˆρ = ˆ ˆ ˆτ ˆ ˆτ ˆ 2MG r3 MG r3 (1.6) Thus all the curvature components are ﬁnite and of order R(Horizon) ∼ 1 M 2 G2 (1. Unless the infalling observer increases the frequency of his/her signals to inﬁnity as the horizon is approached. Let us now consider the history of the infalling observer from the viewpoint of a distant observer. The limits imposed on the information that can be transmitted from near the horizon are not so severe in classical physics as they are in quantum theory.1. We may suppose that the infalling observer sends out signals which are received by the distant observer. the distant observer will inevitably run out of signals and lose track of the transmitter after a ﬁnite number of pulses. We can construct unit basis vectors. and 1. ρ.1.1.1. At this point the curvature increases to the point where the classical laws of nature must fail.1. For a large mass black hole they are typically very small. Assuming that the infalling observer sends signals at a given frequency ν. Deﬁne an orthonormal frame such that the observer is momentarily at rest. φ with τ oriented along the observer’s instantaneous time ˆ ˆ ˆ ˆ ˆ axis. On the other hand the tidal forces diverge as r → 0 where a true local singularity occurs. It is easily seen that as r tends to 2M G.The Schwarzschild Black Hole 5 What does the observer encounter at the horizon? An observer in free fall is not sensitive to the components of the metric. Over the entire span of Schwarzschild time the distant observer records only a ﬁnite number of pulses from the infalling transmitter. but rather senses the tidal forces or curvature components.1. and ρ pointing radially out. 1. θ.2. the distant observer sees those signals with a progressively decreasing frequency.7) at the horizon.3.
1.1 Infalling observer sending signals to distant Schwarzschild observer an arbitrarily small energy without signiﬁcantly disturbing the black hole and its geometry. Therefore. As the observer approaches the horizon. However quantum mechanics requires that to send even a single bit of information requires a quantum of energy. in principle. This energy will back react on the geometry. and the String Theory Revolution b c = Signal originating near horizon c d d = Signal from infalling to distant Black Hole a = distant observer a b = infalling observer b Fig. dis . this quantum must have higher and higher frequency. implying that the observer must have had a large energy available. Information.6 Black Holes. the distant observer can obtain information about the neighborhood of the horizon and the infalling system right up to the point of horizon crossing.
8) The interesting point is that the radialtime part of the metric now has a particularly simple form. Any twodimensional space is conformally ﬂat. 1.2. as we shall see. called conformally ﬂat. The tortoise coordinate r∗ is given explicitly by r∗ = r + 2M G log r − 2M G 2M G (1. We shall see that wave equations in the black hole background have a very simple form in tortoise coordinates. We deﬁne the tortoise coordinate r∗ by 1 1− so that dτ 2 = 1− 2M G r [dt2 − (dr∗ )2 ] − r2 dΩ2 (1.2.9) 2MG r dr2 = 1− 2M G r (dr∗ )2 (1. .The Schwarzschild Black Hole 7 turbing the very quantity to be measured. Thereafter.2.2.2 Tortoise Coordinates A change of radial coordinate maps the horizon to minus inﬁnity so that the resulting coordinate system covers only the region r > 2M G. φ is no exception. and a slice through Schwarzschild space at ﬁxed θ.11) Note: r∗ → −∞ at the horizon.9 the metric of such a slice is manifestly conformally ﬂat. In equation 1.2.10) with ηµν being the usual Minkowski metric. Furthermore it is also static. A space is called conformally ﬂat if its metric can be brought to the form dτ 2 = F (x) dxµ dxν ηµν (1. no information can be transmitted from behind the horizon.
3.18) t 4M G (1.3. if we are interested in a small angular region of the horizon arbitrarily centered at θ = 0 we can replace the angular coordinates by Cartesian coordinates x = 2M G θ cosφ (1.3. Z can be .3.12) − 1) r (r − 2M G) + 2M G sinh−1 ( In terms of ρ and t the metric takes the form dτ 2 = 1− 2M G r(ρ) dt2 − dρ2 − r(ρ)2 dΩ2 (1. Minkowski coordinates T .8 Black Holes.3 Near Horizon Coordinates (Rindler space) The region near the horizon can be explored by replacing r by a coordinate ρ which measures proper distance from the horizon: ρ = = = r 2MG grr (r ) dr 2MG − 1 2 r ) r (1 2MG − dr r 2MG (1.17) It is now evident that ρ and ω are radial and hyperbolic angle variables for an ordinary Minkowski space.12 behaves like ρ ≈ 2 2M G(r − 2M G) giving dτ 2 ∼ ρ2 = dt 4M G 2 (1.14) − dρ2 − r2 (ρ) dΩ2 (1.15) Furthermore.3.3.13) Near the horizon equation 1. Information. we can introduce a dimensionless time ω ω = and the metric then takes the form dτ 2 = ρ2 dω 2 − dρ2 − dx2 − dy 2 (1. and the String Theory Revolution 1.3.16) y = 2M G θ sinφ Finally.3.
20 is only accurate near r = 2M G.2 the relation between Minkowski coordinates and the ρ. for a large black hole. III. and.19) Z = ρ coshω to get the familiar Minkowski metric dτ 2 = dT 2 − dZ 2 − dX 2 − dY 2 (1. The entire Minkowski space is divided into four quadrants labeled I.20) It should be kept in mind that equation 1. 1. II. In Figure 1.3.2 Relation between Minkowski and Rindler coordinates .3. is locally almost indistinguishable from ﬂat spacetime. namely II ω=ω 2 ω=ω 1 III ρ=ρ1 t=0 I IV ρ=ρ2 Fig. However it clearly demonstrates that the horizon is locally nonsingular. Only one of those regions. ω coordinates is shown. and IV.The Schwarzschild Black Hole 9 deﬁned by T = ρ sinhω (1. and only for a small angular region.3.
21) For small ρ equation 1. . In particular the portion of Minkowski space approximating the exterior region of the black hole. Note that it is a twodimensional surface in the fourdimensional spacetime. and therefore appears to be a three dimensional surface.15 shows that ρ ≈ R. 1. since originally the horizon was deﬁned by the single constraint r = 2M G. is called Rindler time. and the String Theory Revolution Region I lies outside the black hole horizon.25) R and ω can be thought of as radial and hyperbolic angular coordinates of a space which is conformal to ﬂat 1+1 dimensional Minkowski space. The approximation of the nearhorizon region by Minkowski space is called the Rindler approximation. A more accurate comparison with the original Schwarzschild metric gives the following requirements: R2 F (R) = 16M 2 G2 1 − 1 1− 2MG r 2M G r dr2 (1.4.4. Note that a translation of Rindler time ω → ω + constant is equivalent to a Lorentz boost in Minkowski space.23) R = r + 2M G log MG r − 2M G 2M G r∗ 4M G = r∗ (1. recall that at the horizon g00 vanishes.22) F (R) dR2 = from which it follows that 4M G log or (1.4. i.3.4 Kruskal Szekeres Coordinates Finally we can bring the black hole metric to the form dτ 2 = F (R) [R2 dω 2 − dR2 ] − r2 dΩ2 (1.24) R = M G exp (1. This may appear surprising.4.10 Black Holes. is called Rindler space. However.4. ω. Region I. The timelike coordinate. Information. Therefore the horizon has no extension or metrical size in the time direction. The horizon itself is the origin T = Z = 0.e.
The Schwarzschild Black Hole 11 Letting R eω = V (1.3. In this region the surfaces of constant r are the + r = con nt sta H . Thus we see that a particle trajectory which crosses H + in a ﬁnite proper time. crosses r = 2M G only after an inﬁnite Schwarzschild time.3. Although the extended horizons lie at ﬁnite values of the Kruskal–Szekeres coordinates. The surfaces of constant r are the timelike hyperbolas in Figure 1. the radialtime part of the metric takes the form dτ 2 = F (R) dU dV (1.27) The coordinates U . V Kruskal–Szekeris coordinates The region of Schwarzschild space with r < 2M G can be taken to be Region II.3 U. As r tends to 2M G the hyperbolas become the broken straight lines H + and H − which we will call the extended past and future horizons. U V H II III I IV Fig. V are shown in Figure 1. 1.26) Re −ω = −U be “radial lightlike” variables.4. they are located at Schwarzschild time ±∞.4.
it is easy to understand the causal properties of the black hole 8 VVVVVVVVV VVVV VVV VV V VV VV VV VV UV II (MG 2 ) = eSing Futur y ularit r t= =2M G V R 2= 8 I External Ho riz on IV H G 2M r= t= . or U V = (M G) 2 2 (1.12 Black Holes. 1.4.4. A nonradially directed light ray or timelike trajectory always lies inside the twodimensional light cone.4. and the String Theory Revolution spacelike hyperboloids U V = positive constant The true singularity at r = 0 occurs at R2 = − (M G) .29) The entire maximal analytic extension of the Schwarzschild geometry is easily described in Kruskal–Szekeres coordinates. Information. It is shown in Figure 1.4 Maximal analytic extension of Schwarzschild in Kruskal–Szekeris coor dinates A useful property of Kruskal–Szekeres coordinates is the fact that light rays and timelike trajectories always lie within a twodimensional light cone bounded by 45o lines. With this in mind.28) (1. U r = con stan t H or i H +zon III VVVVVVVV VVVV V VV V VV VV Past Singularity VV VV VV V Fig. A radial moving light ray travels on a trajectory V = constant or U = constant.
For this reason Region II is said to be behind the horizon. and so it is also behind the horizon.5 Radial light rays from a point in Region I and Region II using Kruskal– Szekeris coordinates Consider next Region II. On the other hand. III H or i H . From any point P2 any signal must eventually hit the singularity. Thus an observer in Region I can send messages to inﬁnity as well as into Region II. An incoming light ray from P1 will eventually cross H + and then hit the future singularity. 1. From Region III no signal can ever get to Region I. Regions III and IV. no signal can ever escape to Region I.5. as we will see. points in Region IV can O ut lig rad goin ht ial g ra y U VVVVVVVV VV V VVVV VV VV V VV VV V VV R2 = (MG)2 = UV ng lli fa ial ay In rad t r h lig Future Singularity V r= 2 t= MG I External 8 . Furthermore.The Schwarzschild Black Hole 13 geometry. are not relevant to the classical problem of black holes formed by collapsing matter. Thus no observer who stays outside r = 2M G can ever be inﬂuenced by events in Region II. Consider a point P1 in Region I.G 8 Fig. Nevertheless let us consider them. A radially outgoing light ray from P1 will escape falling into the singularity as shown in Figure 1.zon H or H + izo n II P 2 P 1 IV VV VVVVVVVVVVVVV VVV VV VV VV Past Singularity VV V r= t= 2M .
As an example. consider ordinary ﬂat Minkowski space.32) as shown in Figure 1. 1. Information. Also shown in Figure 1.14 Black Holes. Furthermore they “compactify” the geometry so that it can be drawn in total on the ﬁnite plane. Spacelike inﬁnity (r = ∞) is where all spacelike surfaces end. Ignoring angular coordinates. In addition to these there are two other inﬁnities which are called I ± .33) (1. For example Y + = tanh(t + r) Y − = tanh(t − r) The entire spacetime is mapped to the ﬁnite triangle bounded by Y+ = 1 Y − = −1 + Y − Y− = 0 (1. . and the String Theory Revolution communicate with Region I. −∞ ≤ t ≤ +∞ to a ﬁnite portion of the plane. especially if. they have spherical symmetry. Any transformation that is of the form Y + = F (t + r) Y − = F (t − r) (1. Region I however cannot communicate with Region IV.30) Radial light rays propagate on the light cone dt ± dr = 0.5.6.5. There are several inﬁnities on the Penrose diagram.31) will preserve the form of the light cone.5. Future and past timelike inﬁnities (t = ±∞) are the beginnings and ends of timelike trajectories.5 Penrose Diagrams Penrose diagrams are a useful way to represent the causal structure of spacetimes. They represent the geometry of a twodimensional surface of ﬁxed angular coordinates. All of this is usually described by saying that Regions II and III are behind the future horizon while Regions III and IV are behind the past horizon.6 are some representative contours of constant r and t. dτ 2 = dt2 − dr2 − angular part = (dt + dr)(dt − dr) − angular part (1. We can use such a transformation to map the entire inﬁnite space 0 ≤ r ≤ ∞.5. like the Schwarzschild black hole.
black holes are formed from the collapse of gravitating matter.The Schwarzschild Black Hole 15 t=+ 8 Y +=+1 + t=3 t=2 t=1 t=0 r= 8 Y +=Y r=0 r=1 r=2 r= 3  Y . and they represent the origin of incoming light rays and the end of outgoing light rays. Similar deformations can be carried out for more interesting geometries. we begin with the empty space Penrose diagram with the infalling shell represented by an incoming lightlike line . To construct the geometry.7. or massless neutrinos with very small radial extension and total energy M provides an example.=1 t= Fig. such as the black hole geometry represent by Kruskal–Szekeres coordinates.6 Formation of a Black Hole The eternal black hole described by the static Schwarzschild geometry is an idealization. a shell of photons. 8 1. The resulting Penrose diagram is shown in Figure 1.6 Penrose diagram for Minkowski space They are past and future lightlike inﬁnity. 1. gravitons. The simpest model for black hole formation involves a collapsing thin spherical shell of massless matter. In nature. For example.
The infalling shell divides the Penrose diagram into two regions. A and B. 1. showing regions (top) and curves of ﬁxed radial position and constant time (bottom) (see Figure 1.7 Penrose diagram for Schwarzschild black hole.16 Black Holes. and the String Theory Revolution Future Singularity t=+ 8 II H+ + III  IV t=Past Singularity Future Singularity t=+ 8 r 2 8 II H or iz on r = r r1 = + t=2 t=1 t=0 III on iz or H t=1 =2 t  r= IV t=Past Singularity Fig. Information. The particular value of Y + chosen for the trajectory is arbitrary since any two such values are related by a time translation. The Region A is interior to the shell and represents the initial ﬂat spacetime 8 8 8 I r = H .8).
Accordingly. 1. the mathematical identiﬁcation of the boundaries of A and B’ must respect the continuity of the variable r. Region B is the region outside the shell and must be modiﬁed in order to account for the gravitational ﬁeld due to the mass M. Just as in Figure 1. To form the full classical evolution the regions A of Figure 1. in Figure 1. we consider the Penrose diagram for a black hole of mass M divided into regions A’ and B’ by an infalling massless shell as in Figure 1. In this case Birkoﬀ’s theorem tells us that the geometry outside the shell must be the Schwarzschild geometry.8 and B’ of Figure 1.The Schwarzschild Black Hole 17 YB Y+ A  Fig.9 the Region A’ is to be discarded.8 Minkowski space Penrose diagram for radially infalling spherical shell of massless particles with energy M before the shell passes.9 must be glued together. In Newtonian physics the gravitational ﬁeld exterior to a spherical mass distribution is uniquely that of a point mass located at the center of the distribution.9. In other words. However this must be done so that the “radius” of the local two sphere represented by the angular coordinates (θ. . φ) is continuous. Much the same is true in general relativity. Once again the particular value of Y + chosen for the trajectory is immaterial.8 where the Region B is unphysical.
Information. and the String Theory Revolution  Y B’ A’  Fig.18 Black Holes. The distant observer collects information that arrives at any instant from his backward light cone. In this region the value of r on the horizon grows from an initial value r = 0 to the value r = 2M G at the shell. On the other hand. It is clear that any light ray or timelike trajectory that originates to the upper left of H must end at the singularity and cannot escape to I + (or t = ∞). 1. the identiﬁcation is always possible. Evidently such an observer never actually sees events on the Y + . This identiﬁes H as the horizon. It is evident from this discussion that the horizon is a global and not a local concept. the horizon also extends into the Region A where the metric is just that of ﬂat spacetime. as shown in Figure 1. Thus in Figure 1.11. On Fig 1.10. Consider next a distant observer located on a trajectory with r >> 2M G.10 we show the resulting Penrose diagram for the complete geometry. In Region B’ the horizon is identical to the surface H + of Figure 1. One of the two Penrose diagrams will have to undergo a deformation along the Y − direction in order to make the identiﬁcation smoothly. The observer originates at past timelike inﬁnity and eventually ends at future timelike inﬁnity. but this will not disturb the form of the light cones.9 Penrose diagram of Schwarzschild black hole with radially infalling shell of massless particle with energy M Since in both cases r varies monotonically from r = ∞ at I − to r = 0. a lightlike surface H is shown as a dotted line.7. In the Region A no local quantity will distinguish the presence of the horizon whose occurence is due entirely to the future collapse of the shell. that is it coincides with the future horizon of the ﬁnal black hole geometry and is therefore found at r = 2M G.
1.The Schwarzschild Black Hole 19 areas Match here of 2sp B’ A r=0 H r < 2M G t= G M 8 =2 + r r= r=0 8  Fig.10 Penrose diagram for collapsing shell of massless particles .
20
Black Holes, Information, and the String Theory Revolution
Fig. 1.11
Distant observer to collapsing spherical shell
horizon. In this sense the horizon must be regarded as at the end of time. Any particle or wave which falls through the horizon is seen by the distant observer as asymptotically approaching the horizon as it is inﬁnitely red shifted. At least that is the case classically. This basic description of black hole formation is much more general than might be guessed. It applies with very little modiﬁcation to the collapse of all kinds of massive matter as well as to nonspherical distributions. In all cases the horizon is a lightlike surface which separates the spacetime into an inner and an outer region. Any light ray which originates in the inner region can never reach future asymptotic inﬁnity, or for that matter ever reach any point of the outer region. The events in the outer region can send light rays to I + and timelike trajectories to t = ∞. The horizon, as we have seen, is a global concept whose location depends on all future events. It is composed of a family of light rays or null geodesics, passing through each spacetime point on the horizon. This is shown in Figure 1.12. Notice that null geodesics are vertical after the shell crosses the horizon and essentially at 45o prior to that crossing. These light rays are called the generators of the horizon.
Di sta nt
ob
se rv er
The Schwarzschild Black Hole
21
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
Horizon
Singularity forms
Shell crosses Horizon
Shell of radially moving lightrays
Horizon Forms
Fig. 1.12
The horizon as family of null geodesics
1.7
Fidos and Frefos and the Equivalence Principle
In considering the description of events near the horizon of a static black hole from the viewpoint of an external observer[1], it is useful to imagine space to be ﬁlled with static observers, each located at a ﬁxed (r, θ, φ). Such observers are called ﬁducial observers, or by the whimsical abbreviation, FIDOS. Each Fido carries a clock which may be adjusted to record Schwarzschild time t. This means that Fidos at diﬀerent r values see their own clocks running at diﬀerent proper rates. Alternatively, they could carry standard clocks which always record proper time τ . At a given r the relation between Schwarzschild time t and the Fidos proper time τ is given by √ 2M G 1 dτ = g00 = [1 − ]2 dt r (1.7.34)
Thus, to the Fido near r = 2M G, the Schwarzschild clock appears to run at a very rapid rate. Another possible choice of clocks would record the dimensionless hyperbolic angle ω deﬁned by equation 1.3.17. The spatial location of the Fidos can be labeled by the angular coordinates (θ, φ) and any one of the radial variables r, r∗ , or ρ. Classically the Fidos can be thought of as mathematical ﬁctions or real but arbitrarily light
22
Black Holes, Information, and the String Theory Revolution
systems suspended by arbitrarily light threads from some sort of suspension system built around the black hole at a great distance. The acceleration 1 of a Fido at proper distance ρ is given by ρ for ρ << M G. Quantum mechanically we have a dilemma if we try to imagine the Fidos as real. If they are extrememly light their locations will necessarily suﬀer large quantum ﬂuctuations, and they will not be useful as ﬁxed anchors labeling spacetime points. If they are massive they will inﬂuence the gravitational ﬁeld that we wish to describe. Quantum mechanically, physical Fidos must be replaced by a more abstract concept called gauge ﬁxing. The concept of gauge ﬁxing in gravitation theory implies a mathematical restriction on the choice of coordinates. However all real observables are required to be gauge invariant. Now let us consider a classical particle falling radially into a black hole. There are two viewpoints we can adopt toward the description of the particle’s motion. The ﬁrst is the viewpoint of the Fidos who are permanently stationed outside the black hole. It is a viewpoint which is also useful to a distant observer, since any observation performed by a Fido can be communicated to distant observers. According to this viewpoint, the particle never crosses the horizon but asymptotically approaches it. The second viewpoint involves freely falling observers (FREFOS) who follow the particle as it falls. According to the Frefos, they and the particle cross the horizon after a ﬁnite time. However, once the horizon is crossed, their observations cannot be communicated to any Fido or to a distant observer. Once the infalling particle is near the horizon its motion can be described by the coordinates (T, Z, X, Y ) deﬁned in equations 1.3.16 and 1.3.19. Since the particle is freely falling, in the Minkowski coordinates its motion is a straight line
dZ dτ
=
dT dτ
pZ m
= − pZ m
pT m
(1.7.35)
=
where pZ and pT are the Z and T components of momentum, and m is the mass of the particle. As the particle freely falls past the horizon, the components pZ and pT may be regarded as constant or slowly varying. They are the components seen by Frefos. The components of momentum seen by Fidos are the components pρ and pτ which, using equation 1.3.19, are given by pρ = pZ coshω + pT sinhω pτ = pZ sinhω + pT coshω (1.7.36)
7. Quantum mechanically we must expect this picture to break down by the time the infalling particle has been squeezed to within a Planck distance from the horizon. Eventually. pρ ≈ pτ ≈ 2pZ expω = 2pZ exp( Fig.37) 4M G Thus we ﬁnd the momentum of an infalling particle as seen by a Fido grows exponentially with time! It is also easily seen that ρ. the Fidos see all matter undergoing Lorentz contraction into a system of arbitrarily thin “pancakes” as it approaches the horizon.The Schwarzschild Black Hole 23 For large times we ﬁnd t ) (1. the proper spatial distance of the particle from the horizon. The hyperbolic boost angle is the dimensionless time ω. the infalling matter is stored in “sedimentary” layers of diminishing thickness as it eternally sinks toward the horizon (see Figure 1.13). during the lifetime of the black hole this boost becomes so large that the momentum of an infalling particle (as seen by a Fido) quickly exceeds the entire mass of the universe.7.38) ρ(t) ≈ ρ(0)exp(− 4M G Locally the relation between the coordinates of the Frefos and Fidos is a time dependent boost along the radial direction. According to classical physics. The Frefos of course see the matter behaving in a totally unexceptional way. As a consequence of the boost. exponentially decreases with time t ) (1.13 Sedimentary layers of infalling matter on horizon Horizo n . 1.
Information. and the String Theory Revolution .24 Black Holes.
Here we will ﬁnd great advantage in utilizing tortoise coordinates in which the metric has the form dτ 2 = F (r∗ ) [dt2 − (dr∗ )2 ] − r2 [dθ2 + sin2 θdφ2 ] The action for χ is I = = 1 2 1 2 (2.0.4) )] dt dr dθ dφ ∗ (sinθ + 25 .2) − r12 ( ∂χ )2 − ∂θ Now deﬁne ( ∂χ )2 } F r2 sinθ ∂φ ψ = rχ and the action takes the form I = − F r2 1 2 ∂ψ [(∂t ψ)2 − ( ∂r∗ − ∂ψ ∂θ 2 1 sinθ ∂ψ ∂φ ∂(lnr) 2 ∂r ∗ ψ) 2 (2.0.0.0.1) √ −g g µν ∂µ χ ∂ν χ d4 x 2 dt dr∗ dθ dφ { (∂t χ) 1 r 2 sin2 θ − (∂r∗ χ)2 F (2.3) (2.Chapter 2 Scalar Wave Equation in a Schwarzschild Background In Chapters 3 and 4 we will be concerned with the behavior of quantum ﬁelds near horizons. In this lecture we will study the properties of a scalar wave equation in the background of a black hole. Let us consider a conventional massless free Klein–Gordon ﬁeld χ in the Schwarzschild background.
depends weakly on the angular momentum .8) = ν2 ψ m (2.0. m an action I m = 1 2 dt dr∗ ∂ψ m ∂t 2 − ∂ψ m ∂r∗ 2 − V (r∗ ) ψ 2m (2.1 as a function of the Schwarzschild coordinate r. One can see that the points rmax ( ) represent unstable circular orbits.10) → ∞ the maximum occurs at rmax ( → ∞) = 3M G. gravitational attraction wins and the potential becomes attractive.0.0. The same potential governs the motion of massless classical particles. It is given by rmax = 3M G For 1 2 1+ 1+ 14 2 + 14 + 9 9 2 ( + 1)2 − 1 2 ( + 1) (2.5) + } ψ 2m ( + 1) ψ 2m ]dt dr∗ Using the relation between r and r∗ r∗ = r + 2M G ln(r − 2M G) gives for each .0. after an integration by parts and the introduction of spherical harmonic decomposition becomes I = −{ ∂lnr 2 ∂r ∗ 1 m 2 ∂ ∂r ∗ ∂lnr ∂r ∗ ˙ [(ψ 2 m) − − F r2 ∂ψ m ∂r ∗ 2 + (2.6) where the potential V (r∗ ) is given by V (r∗ ) = The equation of motion is ∂2ψ m ¨ − V (r∗ ) ψ ψm = (∂r∗ )2 and for a mode of frequency ν − ∂ 2ψ m + V (r∗ ) ψ (∂r∗ )2 m m r − 2M G r ( + 1) 2M G + r2 r3 (2.9) The potential V is shown in Figure 2. where the direction of the force changes. and the String Theory Revolution which. However as the horizon is approached.7) (2.0. In fact it is just the relativistic generalization of the usual repulsive centrifugal barrier. For r >> 3M G the potential is repulsive. and .26 Black Holes. Information. and pulls a wave packet toward the horizon.0. The maximum of the potential.
and the ﬁeld behaves like a free massless Klein– Gordon ﬁeld. The particle has enough energy to overcome the barrier without tunneling if ν 2 is larger than the maximum height of the barrier. if = 0 the height of the barrier is Vmax = 1 2M 2 G2 3 8 3 (2.12) An swave quantum will therefore escape if ν > 0.9 has the form of a Schrodinger equation for a particle of energy ν 2 in a potential V .11) ±t) ψ → ei k (r Let us consider a ﬁeld quantum of frequency ν and angular momentum propagating from large negative r∗ toward the barrier at r ≈ 3M G.Scalar Wave Equation in a Schwarzschild Background 27 V (r) Region of Thermal Atmosphere =2 =1 =0 r 2MG 3MG 4MG 5MG 6MG Fig. Any particle that starts with vanishing radial velocity in the region r < 3M G will spiral into the horizon. Will it pass over the barrier? To answer this we note that equation 2.0. For example. The eigenmodes in this region have the form of plane waves which propagate with unit velocity dr ∗ dt = ∓1 ∗ (2.13) . In the region of large negative r∗ where we approach the horizon.0.0.1 Eﬀective potential for free scalar ﬁeld vs Schwarzschild radial coordinate the innermost such orbit is at r = 3M G. the potential is unimportant. 2.0.15 MG (2.
and the String Theory Revolution 0. ω) (2. whether on the inside or outside of the barrier will have more diﬃculty penetrating through. u. Less energetic particles must tunnel through the barrier. Instead of using spherical waves.16) (2.1. near the horizon we can decompose χ into transverse plane waves with transverse wave vector k⊥ χ = d2 k⊥ ei k⊥ x⊥ χ(k⊥ . Information.1.0. A particle of high angular momentum. ∂Y ).18) where ∂⊥ χ = (∂X .1.15 Similarly an swave quantum with ν > MG will be able to penetrate the barrier from the outside and fall to the horizon.17) − ∂χ ∂u 2 − e2u (∂⊥ χ)2 (2.15) 2. For large Vmax ≈ 2 1 27 M 2 G2 (2.14) Therefore the threshold energy for passing over the barrier is 1 ν ∼ √ 27 M G (2.1.19) .0. We deﬁne u = logρ and the metric near the horizon becomes dτ 2 = exp(2u) dω 2 − du2 − dX 2 − dY 2 The scalar ﬁeld action becomes I = 1 2 dX dY du dω ∂χ ∂ω 2 (2.28 Black Holes.1 Near the Horizon Near the horizon the exterior of the black hole may be described by the Rindler metric dτ = ρ2 dω 2 − dρ2 − dX 2 − dY 2 It is useful to replace ρ by a tortoiselike coordinate which again goes to −∞ at the horizon.
If the horizon has circumference 2π(2M G). Since in this region the ﬁeld χ(u) behaves like a free massless ﬁeld.21) The correspondence between the momentum vector k and the angular momentum is given by the usual connection between momentum and angular momentum. there is a potential conﬁning quanta to the region near the horizon.20) Thus the potential is V (k. the integral should be infrared cut 1 oﬀ at k ∼ MG .1. the potential barrier is cut oﬀ when ρ = eu is greater than M G.1.1.1. − (2. then a wave with wave vector k⊥ will correspond to an angular momentum   = k r = 2M Gk.23) . u) = k 2 e2u (2.1. but qualitatively is correct for > 0.Scalar Wave Equation in a Schwarzschild Background 29 the action for a given wave number k is I = 1 2 dω du (∂ω χ)2 − (∂u χ)2 − k 2 e2u χ2 (2. For very low angular momentum the approximation is not accurate. Qualitatively.24) ∂u2 Once again we see that unless k = 0. in the Rindler case V increases as eu without bound. In the next section we will see that this is not so. the behavior of a quantum ﬁeld in a black hole background diﬀers from the Rindler space approximation in that for the black hole.1. We have not thus far paid attention to the boundary conditions at the horizon where u → −∞. From the action in equation 2. Thus the potential in equation 2. By contrast. the boundary condition would be expected to be that the ﬁeld is in the usual quantum ground state.1. In approximating a sum over and m by an integral over k.22) A solution which behaves like exp(iνt) in Schwarzschild time has the form ei ν [4MGω] = ei λ ω The time independent form of the equation of motion is ∂ 2χ + k 2 exp(2u) χ = λ2 χ (2.20 we obtain the equation of motion ∂2χ ∂2χ − + k 2 exp(2u) χ = 0 ∂ω 2 ∂u2 (2.21 is seen to be proportional to 2 .
and the String Theory Revolution . Information.30 Black Holes.
interactions become very important near the horizon of a black hole. 31 . The method we will use applies to any relativistic quantum ﬁeld theory including those with nontrivial interactions. As we have seen. The ﬁeld in Region I of Figure 1. In that way we can use the special relativistic laws of nature to understand the eﬀect of a gravitational ﬁeld. For illustrative purposes we will consider a free scalar ﬁeld theory. Signals from Region IV can.1 Classical Fields First let us consider the evolution of a classical ﬁeld in Rindler space. Because Rindler space covers only a portion of the spacetime geometry (Region I) there are new and subtle features to the description of quantum ﬁelds. It is important to bare in mind that such a noninteracting description is of limited validity.Chapter 3 Quantum Fields in Rindler Space According to Einstein. but to do so they must pass through the surface ω = −∞. the study of a phenomenon in a gravitational ﬁeld is best preceeded by a study of the same phenomenon in an accelerated coordinate system.2 can be described in a self contained way. 3. reach Region I. These features are closely associated with the existence of the horizon. Ignoring them leads to an inconsistent description of the Hawking evaporation process. Evidently the Rindler observer sees a world in which physical phenomena can be described in a completely self contained way. Therefore signals from Region IV are regarded as initial data in the remote past by the Rindler observer. of course. Obviously inﬂuences from Regions II and III can never be felt in Region I since no point in Regions II or III is in the causal past of any point in Region I. As we shall see. the relativistic analogue of a uniformly accelerated frame is Rindler space.
1 the relation between neighboring equal Rindlertime surfaces is shown. the correlation between ﬁelds at diﬀerent spatial points does not vanish. X. However it only involves the degrees of freedom in Region I. T 00 is given by T 00 = 1 Π2 2 + (∇χ) + V (χ) 2 2 (3.3) The origin of the factor ρ in the Rindler Hamiltonian density is straightforward. Using conventional methods the generator of ωtranslations is given by ∞ HR = ρ=0 dρ dX dY ρ T 00 (ρ. To understand the new feature. for a massive scalar ﬁeld with potential V . 3. The proper time separation between the surfaces is δτ = ρ δω (3. The Rindler Hamiltonian is HR = dρ dx⊥ ρ 2 Π2 + ∂χ ∂ρ 2 + ∂χ ∂x⊥ 2 + 2 V (χ) (3. Y. The Rindler Hamiltonian is similar to the generator of Lorentz boosts from the viewpoint of the Minkowski observer. but a new twist is encountered. i. For example. in Rindler space.1. Y .1. and the String Theory Revolution The evolution from one surface of constant ω to another is governed by the Rindler Hamiltonian. in free massless scalar theory the equal time correlator is given by 0 χ(X. recall that in the usual vacuum state.1) where T 00 is the usual Hamiltonian density used by the Minkowski observer. Y ) (3. Our goal is to describe the usual physics of a quantum ﬁeld in Minkowski space. to push the ωsurface ahead requires a ρdependent time translation.2 Entanglement Quantum ﬁelds can also be described in a selfcontained fashion in Rindler space.5) .32 Black Holes. Z ) 0 ∼ 1 ∆2 (3. but from the viewpoint of the Fidos in Region I.2.1.2) where Π is the canonical momentum conjugate to χ.e. Z) χ(X .1. This is the reason that T 00 is weighted with the factor ρ.4) Thus. In Figure 3. For example. Information.
the two points might lie on opposite sides of the horizon at Z = 0. t= 8 8 r > 2MG 2 ω=+4 ω=+3 r < 2MG p= 1 p= ω=+2 ω=+1 ω=0 ρ=0 Z (Horizon) ω=−1 ω=−2 Light cone ω=−3 ω=−4 Fig.1 Equal time and proper distance surfaces in Rindler space where ∆ is the spacelike separation between the points (X.5 represents the quantum correlation seen by Fido’s in Region I. in which case the correlator in equation 3. we say that they are quantum entangled.2. Z) and (X . 3. When two subsystems (ﬁelds in Regions I and III) become correlated.2. On the other hand. . Y. The appropriate description of an entangled subsystem is in terms of a density matrix.6) The two points might both lie within Region I.Quantum Fields in Rindler Space 33 ω= T . Z ) ∆2 = (X − X )2 + (Y − Y )2 + (Z − Z )2 (3. Nevertheless it has signiﬁcance. so that neither can be described in terms of pure states. Y . In that case the correlation is unmeasurable to the Fidos in Region I.
.. ρA (α. . which have previously been in contact but are no longer interacting. the relative phases between the states j are random. . . However.3. α ) = β Ψ∗ (α. β ) (3. There is one special case when the density matrix is indistinguishable from a pure state.8) Similarly...3. ρ = 0 0 ρ3 . ρB (β.7) where α and β are appropriate commuting variables for the subsystems A and B.. Information. experiments performed on B are described by ρB (β. . (3. The combined system has a wavefunction Ψ = Ψ(α.34 Black Holes.3... β) (3. . Now suppose we are only interested in subsystem A.....9) The rule for computing an expectation value of an operator a composed of A degrees of freedom is a = T r a ρA Density matrices have the following properties: 1) T r ρ = 1 (total probability=1) 2) ρ = ρ† (hermiticity) 3) ρj ≥ 0 (all eigenvalues are positive or zero) In the representation in which ρ is diagonal ρ1 0 0 .10) (3. ...3.. . and the String Theory Revolution 3. 0 ρ2 0 .11) The eigenvalues ρj can be considered to be probabilities that the system is in the j th state... A and B... .. A complete description of all measurements of A is provided by the density matrix ρA (α. This is the case in which only one eigenvalue ρj is . β ) = α Ψ∗ (α. β ).3 Review of the Density Matrix Suppose a system consists of two subsystems. unlike the case of a coherent superposition of states. β) Ψ(α . β) (3. β) Ψ(α...3. α )..
The Von Neumann (or entanglement) entropy should not be confused with the thermal entropy of the second law of thermodynamics. It is therefore called the entropy of entanglement.B.12) A quantitative measure of the departure from a pure state is provided by the Von Neumann entropy S = −T r ρ log ρ = − j ρj log ρj .B. This case can only result from an uncorrelated product wave function of the form Ψ(α.3. T re−β H (3.16) In this case the thermal entropy is given by Sthermal = −T r ρM. = e−β H . β) = ψA (α) ψB (β) (3. If a system with Hamiltonian H is in thermal equilibrium at temperature T = 1/β then it is described by a Maxwell–Boltzman density matrix ρM. (3.3.17) .3.3. (3.14) More generally. This entropy has its origin in coarse graining.15) Thus we see that the Von Neumann entropy is a measure of the number of states which have an appreciable probability in the statistical ensemble. if ρ is a projection operator onto a subspace of dimension n.13) S is zero if and only if all the eigenvalues but one are zero. The entropy is also a measure of the degree of entanglement between A and B. The one nonvanishing eigenvalue is equal to 1 by virtue of the trace condition on ρ. we ﬁnd S = log n (3. We may think of eS as an eﬀective dimensionality of the subspace described by ρ.3. The opposite extreme to a pure state is a completely incoherent den1 sity matrix in which all the eigenvalues are equal to N .3. logρM.B. In that case S takes its maximum value Smax = − j 1 1 log = log N N N (3. where N is the dimensionality of the Hilbert space.Quantum Fields in Rindler Space 35 nonzero.
In particular we would like to understand the density matrix ρR which represents the usual Minkowski vacuum to the Fidos in Region I.2 Fields on the spacelike surface T = 0 in Minkowski space of space form a complete set of commuting operators. 3. Y. we call them χR and χL respectively. Z) Z < 0 The general wave functional of the system is a functional of χL and χR Ψ = Ψ(χL .4.2 the surface T = 0 of Minkowski space is shown divided into two halves. Z) = χR (X. For the case of a scalar ﬁeld χ the ﬁelds at each point III χL χR I Fig.18) χ(X. the translation invariance along the Z axis is explicitly broken by the act of singling out the origin Z = 0 for . Thus the Fidos must see the vacuum as invariant under translations along the X and Y axes. Z) Z > 0 (3. Thus χ(X. and the String Theory Revolution 3. In Figure 3. Z) = χL (X. Information. one in Region I and one in Region III.19) We wish to compute the density matrix used by the Fidos in Region I to describe their Rindler world. First let us see what we can learn from general principles. Y. Obviously the state Ψ is translationally invariant under the usual Minkowski space translations. These ﬁelds may be decomposed into two subsets associated with regions I and III. Y. Y.4. χR ) (3.36 Black Holes. However.4 The Unruh Density Matrix Now let us consider the space of states describing a Lorentz invariant quantum ﬁeld theory in Minkowski space.
the invariance under ωtranslations becomes invariance under rotations in the Euclidean (Z.23) Letting T → i X 0 we obtain the Euclidean action IE = d4 X 1 (∂X χ)2 − 2 V (χ) 2 (3. χR ) = √ Z dχ(x) e−IE (3. Finally the action IE is evaluated as an integral over the portion of Minkowski space with X 0 > 0. χR ) on the surface X 0 = 0.21) To proceed further we must use the fact that Ψ(χL .4. χR ) is the ground state of the Minkowski Hamiltonian.25) where the path integral is over all χ(x) with X 0 > 0 and Z is an appropriate normalization factor .22) The socalled Euclidean ﬁeld theory is deﬁned by replacing the time coordinate T by i X 0 . This follows from the Lorentz boost invariance of Ψ. The boost invariance of the original Minkowski action insures that the Euclidean action has four dimensional rotation invariance. Thus [HR . Suppose we wish to compute Ψ(χL .4. This suggests a new way to carry out the . General path integral methods may be brought to bare on the computation of the ground state wave functional. X 0 ) plane.4.Quantum Fields in Rindler Space 37 special consideration. For example. χR ).4. Y translation invariance we conclude that ρR commutes with the components of momentum in these directions [pX . In particular. ρR ] = [pY . ρR ] = 0 (3.20) A very important property of ρR is that it is invariant under Rindler time translations ω → ω + constant. Let us assume that the ﬁeld theory is described in terms of an action I = d3 X dT L (3. Then we consider the path integral 1 Ψ(χL . the Euclidean version of ordinary scalar ﬁeld theory is obtained from the usual Minkowski action I = d3 X dT 1 2 χ − (∇χ)2 − 2 V (χ) ˙ 2 (3. From the X.4. ρR ] = 0 (3. The ﬁeld χ(x) is constrained to equal (χL .4.24) Now a standard method of computing the ground state by path integration is to use the Feynman–Hellman theorem.
3 Euclidean analogue of Rindler space for path integration The strategy for computing the path integral is to integrate over the ﬁelds in the ﬁrst wedge between θ = 0 and θ = δθ.3. 3. Now let us divide the region X 0 > 0 into inﬁnitesimal angular wedges as shown in Figure 3. X 0 ) plane to be θ. According to . The inﬁnitesimal generator which pushes θ surfaces forward is just the Rindler Hamiltonian. χR ) = √ χL  e−π HR χR Z (3. The process can be iterated until the entire region X 0 > 0 has been covered.38 Black Holes.4. The integral over the ﬁrst wedge is deﬁned by constaining the ﬁelds at θ = 0 and θ = δθ.27) In other words. Information. Let us deﬁne the Euclidean angle in the (Z. T δθ θ=0 χL Horizon ρ=0 χR Z Fig. (3.26) π δω To compute the full path integral we raise the matrix G to the power giving 1 Ψ(χL . Now we are prepared to compute the density matrix ρR . The angle θ is the Euclidean analogue of the Rindler time ω. The matrix is recognized to be G = (1 − δθ HR ) . This deﬁnes a transfer matrix G in the Hilbert space of the ﬁeld conﬁguration χR . the path integral deﬁning Ψ is computed as a transition matrix element between initial state χR and ﬁnal state χL .4. and the String Theory Revolution path integral.
says that the Fidos see the vacuum as a thermal ensemble with a density matrix of the Maxwell–Boltzmann type. discovered by William Unruh in 1976 .4.27 we get ρR (χR .28) Now using equation 3. χR ) dχL (3.Quantum Fields in Rindler Space 39 the deﬁnition in equation 3. The origin of the dimensionless temperature lies in the dimensionless character of the Rindler time variable ω.4.31) The derivation of the thermal character of the density matrix and the value of the Rindler temperature in equation 3.4. χR ) = 1 Z χR  e−π HR χL χL  e−π HR χR dχL (3. Nevertheless we should be able to assign to each Fido a conventional temperature that would be recorded by a standard thermometer held at rest at the location of that Fido.8. χR ) Ψ(χL .5 Proper Temperature It is noteworthy that the temperature TR is dimensionless. Let us suppose that the thermometer is at rest with respect to the 1 Fido at position ρ so that it has proper acceleration ρ . Ordinarily.4. The thermometer is assumed to be very weakly coupled to the quantum ﬁelds so that it eventually will come to thermal equilibrium with them.3. 3.4. temperature has units of energy.4. or equivalently.29) = 1 Z χR  e −2 π HR χR In other words.31 is entirely independent of the particulars of the relativistic ﬁeld theory. inverse length. We can consider a thermometer to be a localized object with a set of proper energy levels i . quantum electrodynamics. The Rindler energy . or quantum chromodynamics. χR ) = Ψ∗ (χL . the density matrix ρR is given by ρR (χR . the density matrix is given by the operator ρR = 1 exp(−2 π HR ) Z (3. It is equally correct for a free scalar quantum ﬁeld. The levels i are the ordinary energy levels of the thermometer when it is at rest.30) This remarkable result. The temperature of the ensemble is TR = 1 1 = 2π βR (3.
5. 1 When the quantum ﬁeld at Rindler temperature 2π equilibrates with the thermometer.1 evidently receives a contribution from the thermometer of the form HR (thermometer) = i ρ i i i (3.33) Accordingly.35) The reader may wonder about the origin of the thermal ﬂuctuations felt by the Fidos. and the String Theory Revolution in equation 3.34) Thus each Fido experiences a thermal environment characterized by a temperature which increases as we move toward the horizon at ρ = 0. and which cause the density matrix of Region I to be a mixed state. The thermal ﬂuctuations are nothing but the conventional virtual vacuum ﬂuctuations. the thermometer registers a proper temperature T (ρ) = 1 1 = TR 2π ρ ρ (3. The ﬂuctuation (b) contained in Region III has no signiﬁcance to the Fidos in Region I. The proper temperature T (ρ) can also be expressed in terms of the proper 1 acceleration of the Fido which is equal to ρ . That ﬂuctuation can be thought of as a conventional ﬂuctuation described by the quantum Hamiltonian HR . . Finally there are loops like (c) which are partly in Region I but which also enter into Region III.5. we ﬁnd T (ρ) = a(ρ) 2π (3. In Figure 3.5. since the system under investigation is the Minkowski space vacuum. It is helpful in visualizing these eﬀects to describe virtual vacuum ﬂuctuations as short lived particle pairs.40 Black Holes.32) In other words the Rindler energy level of the ith state of the thermometer is ρ i . Information. but now being experienced by accelerated apparatuses.1. the probability to ﬁnd the thermometer excited to the ith level is given by the Boltzmann factor Pi = e−2π ρ i −2π ρ je j (3. calling the acceleration a.4 ordinary vacuum ﬂuctuations are shown superimposed on a Rindler coordinate mesh. One virtual loop (a) is contained entirely in Region I. These are the ﬂuctuations which lead to nontrivial entanglements between the degrees of freedom χL and χR . Thus.5.
From the viewpoint of the Fidos. Obviously the answer is no. Thus. Real particles are seen being injected into the Rindler space from the horizon. according to the Fidos. Now consider the portion of the loop (b) which is found in Region I. a particle is injected into the system at ω = −∞ and ρ = 0. the ﬂuctuation lasts for an inﬁnite time and is therefore not virtual at all. A thermometer at rest in an inertial frame in the Minkowski vacuum will record zero temperature. 3. For example. A natural question to ask is whether the thermal eﬀects are “real”. the horizon behaves like a hot membrane radiating and reabsorbing thermal energy. If the virtual ﬂuctuation of energy needed to produce the pair is E. It is tempting to declare that the thermal eﬀects seen by the Fidos are ﬁctitious and that the reality is best described in the frame of the Frefos. then the lifetime of the ﬂuctuation ∼ E −1 .Quantum Fields in Rindler Space 41 ω= 8 (a) (b) (c) ω=− Fig. and eventually fall back to the horizon. we may ask whether any such thermal eﬀects are seen by freely falling observers carrying their thermometers with them as they pass through the horizon. To state it diﬀerently.4 Vacuum pair ﬂuctuations near the horizon A virtual ﬂuctuation is usually considered to be short lived because it “violates energy conservation”. The particle travels to some distance and then falls back towards ρ = 0 and ω = +∞. 8 .
We will see later that the stretched horizon has many other physical properties besides temperature. Such a ﬁctitious membrane at Planckian distance from the horizon is called the stretched horizon. it is best to avoid the metaphysical question of whose description is closer to reality. In particular a Fido equipped with a standard thermometer.42 Black Holes. Instead we simply observe that the phenomena are described diﬀerently in two diﬀerent coordinate systems and that diﬀerent physical eﬀects are experienced by Frefos and Fidos. We can now state the sense in which a self contained description of Rindler space is possible in ordinary quantum ﬁeld theory. by yielding to this temptation we risk prejudicing ourselves too much toward the viewpoint of the Frefos. In particular. will discover all the physical 1 phenomena associated with a local proper temperature T (ρ) = 2π ρ . an eﬀective “membrane” is kept at a ﬁxed temperature T (ρo ) = 2π1ρo by an inﬁnite heat reservoir. we are going to encounter questions of the utmost subtlety concerning the proper relation between events as seen by observers who fall through the horizon of a black hole and those seen by observers who view the formation and evaporation process from a distance. or other apparatus. a Frefo carrying similar apparatuses will see only the zero temperature vacuum state. Thus for the moment. although it is completely unseen by observers who fall freely through it. Information. a boundary condition of some sort must be provided. We see that the correct condition must be that at some small distance ρo . . and the String Theory Revolution However. Later we will discuss the very interesting question of how contradiction is avoided if a Frefo attempts to communicate to the stationary Fidos the information that no thermal eﬀects are present. Since Rindler space has a boundary at ρ = 0. particle detector. It will prove useful later to G/c3 locate the membrane at a distance of order the Planck length P = where quantum gravitational or string eﬀects become important. By contrast.
the paradoxes and contradictions associated with black holes. up the scale of energies. At higher temperatures. pions are produced which eventually dissociate into quarks and gluons. photons. the study of a free quantum ﬁeld in Rindler space is a useful starting point.Chapter 4 Entropy of the Free Quantum Field in Rindler Space In the real world. Fourier decomposing the ﬁeld χ in equation 2. Finally. The parameter represents the proper 43 .1. All of these phenomena have their place in the Fido’s description of the region near a horizon. e− threshold and electronpositron pairs are produced. At the lowest temperatures where only massless quanta are produced by thermal ﬂuctuations. and statistical thermodynamics that these lectures are concerned with are largely a consequence of such unjustiﬁed extrapolation. In this lecture we will consider an enormously oversimpliﬁed description of the world in which only a single free ﬁeld is present in a ﬁxed spacetime background. The simplest method of dealing with this region is to introduce a cutoﬀ at some point uo = log at which point the ﬁeld (or its ﬁrst derivative) is made to vanish.18.1. We consider the ﬁeld theory deﬁned by equation 2.1) In order to quantize the ﬁeld χ it is necessary to provide a boundary condition when u → −∞. There is serious danger in extrapolating far reaching conclusions from so oversimpliﬁed a situation.1. one expects to ﬁnd a very weakly interacting gas of gravitons. and so it goes. quantum mechanics.24. − ∂ 2 χk + k 2 e2u χk = λ2 χk ∂u2 (4. In fact.19 leads to the wave equation in equation 2.0. Increase the temperature to the e+ . Nevertheless. the Planck temperature is reached where totally new phenomena of an as yet unimagined kind take place. and neutrinos. a wide variety of diﬀerent phenomena take place at diﬀerent temperature scales. The free gas is replaced by a plasma.
Physically we are introducing a perfectly reﬂecting mirror just outside the horizon at a distance .k (u) + a− (n. The other wall of the box is provided by the repulsive potential V (u) = k 2 exp(2u) which becomes large when u > −log k. is that . It is by no means obvious that a reﬂecting boundary condition very near the horizon is a physically reasonable way to regularize the theory.k (u) (4. encountered in Chapter 3. Thus we may approximate the potential by a second wall at u = u1 = −log k.0. k) N (n. k) a(n.0. k) Thus far the quantization rules are quite conventional. k) fn. Information. k) has frequency λ(n. Each transverse Fourier mode χk can be thought of as a free 1+1 dimensional quantum ﬁeld conﬁned to a box. k) a† (n.2) For each value of k the ﬁeld χk can be expanded in mode functions and creation and annihilation operators according to χk (u) = n ∗ a+ (n. k). and the String Theory Revolution distance of the cutoﬀ point to the horizon. k) = a† (n. However it will prove interesting to separate physical quantities into those which are sensitive to and those which are not.44 Black Holes.3) where the mode (n. k) (4. One end of the box is at the reﬂecting boundary at u = uo = log . k) (4.4) n λ(n. The Rindler Hamiltonian is given by HR = d2 k = where N (n. Those things which depend on are sensitive to the behavior of the physical theory at temperatures of order 1 2π and greater. Later we will remove the cutoﬀ by allowing uo → −∞. k) a(n. The new and unusual feature of Rindler quantization.0.0.5) n λ(n. The total length of the box depends on k and according to L(k) = −log( k) (4. k) fn.
u) and (X . Y. k) = 1 exp[2πλ(n. Moreover. Therefore such quantities can be said to decouple from the degrees of freedom within a distance of the horizon. k)] − 1 (4. Y.k ρR (n. A much more singular quantity which will be of great concern in future lectures is the entropy of the vacuum state. on the other hand. A black hole.0. Y . Y .Entropy of the Free Quantum Field in Rindler Space 45 we do not identiﬁy the vacuum with the state annihilated by the a(n.8) (4. Physical quantities in Rindler space can be divided into those which are sensitive to the cutoﬀ at and those which are not. the ﬁeld correlation functions such as χ(X. (4. k) ∼ exp −2πλ(n. In fact. Y .0.6) with ρR (n. u) χ(X . This however would clearly violate the fourth guiding principle stated in the introduction: To a freely falling observer. k) a(n. the horizon of a black hole should in no way appear special. u ) are kept away from the horizon. is not a translationally invariant system. k) Thus the average occupation number of each mode is N (n. Y. but rather with the thermal density matrix ρR = n. k) a† (n.9) are found to have smooth limits as → 0. Such a state is not at all invariant under translations of the original Minkowski coordinates Z and T . u ) = T r ρ χ(X. One might therefore suppose that the evolution of the horizon might lead to the Fock space vacuum with no quanta rather than the thermal state. k). u) χ(X .0. the large back reaction on the gravitational ﬁeld that would result from the divergent expectation value of T µν makes it unlikely that this state can exist altogether. k) (4.0. The reader might wonder what goes wrong if we choose the state which is annihilated by the a’s. Certainly this is not a good candidate to represent the original Minkowski vacuum. a careful computation of the expectation value of T µν in this state reveals a singular behavior at the horizon.7) These particles constitute the thermal atmosphere. as long as the points (X. As an example of insensitive quantities. u ). .
14) The entropy S in equations 4.13 and 4. Information. This is because the eﬀect of integrating over the ﬁelds χL in equation 3.0.0.16 and 3. Deﬁning T r e−βH = Z(β) and using the identity ρ log ρ = we obtain S = −T r ∂ e−N βH ∂N Z(β)N N =1 e−βH Z (4.46 Black Holes.13) (4.0.10) ∂ N ρ ∂N (4.14 can be thought of as both entanglement and thermal entropy in the special case of the Rindler space density matrix. Thus kX = 2nX π B .4.4.17 to obtain the entropy. For the present case of free ﬁelds the entropy is additive over the modes and can be estimated from the formula for the thermodynamics of a free 1+1 dimensional scalar ﬁeld. Y plane by a ﬁnite torus with periodic boundary conditions.28 is to produce the thermal density matrix in equation 3. kY = 2nY π B (4.11) N =1 = +T r βH + lnZ (4.0.0. Thus the computation of the entropy of Rindler space is reduced to ordinary thermodynamic methods. To compute the total entropy we begin by replacing the inﬁnite transverse X.30. and the String Theory Revolution Since the relevant density matrix has the Maxwell–Boltzmann form.3.0. This has the eﬀect of discretizing the values of k.0. we can use equations 3.12) = β H + lnZ 1 Deﬁning E = H and F = − β logZ we ﬁnd the usual thermodynamic identity S = β(E − F ) Another identity follows from using E = − ∂logZ where we ﬁnd ∂β S = −β 2 ∂(logZ/β) ∂β (4.0.15) where B is the size of the torus.3. .
18) In evaluating equation 4.19) From equation 4.20) Now consider the entropy stored in a layer of thickness δρ and area B 2 at . Substituting T = the length L gives the entropy of χk S(k) = 1 log k  6 1 2π (4.0. The ﬁrst is that it is proportional to the transverse area of the horizon. As we shall see. We ﬁnd that S is approximately given by ST otal ≈ 1 B2 96π 2 2 (4.Entropy of the Free Quantum Field in Rindler Space 47 The entropy stored in the ﬁeld χk can be estimated from the entropy density of a 1+1 dimensional massless free boson at temperature T .0.0. but this is not the case. The entropy is stored in the vicinity of the stretched horizon and therefore grows only like the area. Further insight into the form of the entropy can be gained by recalling 1 that the proper temperature T (ρ) is given by T (ρ) = 2πρ . Furthermore the entropy density of a 3+1 dimensional free scalar ﬁeld is given by S(T ) = V 2 ζ(4)kB π2 kB T c 3 = V 2π 2 3 T 45 (4.0.15 and let B → ∞ ST otal = B2 24π 2 d2 k log k  (4.0. B 2 . One might have expected it to diverge as the volume of space. Nevertheless the divergence in S indicates that its value is sensitive to the ultraviolet physics at very small length scales. A S standard calculation gives the entropy density L to be given by S π = T L 3 where T is the temperature.17) To sum over the values of k we use equation 4.0. This is because when k = 1 the potential is already large at u = uo so that the entire contribution of χk is supressed.16) and equation 4. the entropy density of the horizon is a physical quantity whose exact value is known.0.0. the integral must be cut oﬀ when k > 1 .19 we see two important features of the entropy of Rindler space.0. The second feature which should alarm us is that the entropy per unit area diverges like 1 2 .18.2 for (4.
A good qualitative understanding of the process can be obtained from the Rindler quantum ﬁeld theory in equation 2.22 by observing two facts: 1) The Rindler time ω is related to the Schwarzschild time t by the equation ω = t 4M G (4.1.0.24) The implication of this fact is that the temperature of the thermal atmosphere is reckoned to be red shifted also. Thus the temperature as . However.48 Black Holes.1. Unlike the Rindler case.23) Thus a ﬁeld quantum with Rindler frequency νR is seen by the distant Schwarzschild observer to have a red shifted frequency ν ν = νR 4M G (4.0.0. The particles of the thermal atmosphere will gradually leak through the barrier and carry oﬀ energy in the form of thermal radiation.22) = Now we see that the entropy is mainly found near the horizon because that is where the temperature gets large. 4. and the String Theory Revolution a distance ρ from the horizon δS(ρ) = = 2π 2 45 T 3 (ρ) δρ B 2 (4. the thermal atmosphere is not absolutely conﬁned by the centrifugal potential in equation 2. We can expect that identical thermal eﬀects will occur near the horizon of a very massive black hole. in the case of a black hole a new phenomenon can take place – evaporation.7.21) δρ B 2 2π 2 1 45 (2πρ)3 To ﬁnd the full entropy we integrate with respect to ρ S = B 2 2π 2 (2π)3 45 B2 45 (2π)3 2 2 2π 2 ∞ dρ ρ3 (4.1 Black Hole Evaporation The discovery of a temperature seen by an accelerated ﬁducial observer adds a new dimension to the equivalence principle. Information.1.
26) Vmax ( = 0) = 2 G2 1024 M T = 3 3 Any swave quanta with frequencies of order (Vmax ) = 32MG or greater will easily escape the barrier.1.25 indicates that some of the swave particles will easily escape to inﬁnity. 1/2 √ Particles of angular momenta higher than swaves cannot easily escape because the potential barrier is higher than the thermal scale. In fact. Since the average energies of massless particles in thermal equilibrium at temperature T is of course of order T . and ultimately lead to absurd results.1.1. equation 4. 2) The centrifugal barrier which is described in the Rindler theory by the potential k 2 exp(2u) is modiﬁed at distances r ≈ 3M G as in Figure 2.25) 2π 4M G 8πM G a form ﬁrst calculated by Stephen Hawking.1. In particular the maximum value that V takes on for angular momentum zero is 1 27 (4. we will discuss one such absurdity. the walls are semitransparent and the cavity slowly radiates its energy. Unless the black hole is kept in equilibrium by incoming radiation it will lose energy to its surroundings. For these particles. Indeed it only makes sense if there are interactions of suﬃcient strength to keep the system in equilibrium during the course of the evaporation. . most discussions of Hawking radiation rely in an essential way on the free ﬁeld approximation. A small fraction of the particles carry very low angular momenta. Most quanta in the thermal atmosphere have high angular momenta and reﬂect oﬀ the walls of the cavity. At the end of the next lecture. The black hole is like a slightly leaky cavity containing thermal radiation. The above description of Hawking radiation does not depend in any essential way on the free ﬁeld approximation. (4. This is the process ﬁrst discovered by Hawking and is referred to as Hawking radiation.Entropy of the Free Quantum Field in Rindler Space 49 seen by the distant observer is 1 1 1 × = .
and the String Theory Revolution .50 Black Holes. Information.
It follows thermodynamically that it must also have an entropy.3) 1 8πMG dS The Schwarzschild radius of the black hole is 2M G and the area of the horizon is 4π(4M 2 G2 ) so that SBH = Area 4G (5. but unfortunately it is inﬁnite as → 0.0. Using equation 5. To get an 51 .4) This is the famous Bekenstein–Hawking entropy. Presumably this entropy is part of the entropy of the black hole. Evidently something cuts oﬀ the modes which are very close to the horizon.Chapter 5 Thermodynamics of Black Holes We have seen that a large black hole appears to a distant observer as a body with temperature T = 1 8πM G (5.1) and energy M .2) where E.1 dM = from which we deduce S = 4πM 2 G (5.0.0. This.0. We have seen in Chapter 4 equation 4. It is gratifying that it is proportional to the area of the horizon. as we have seen.0. the black hole energy.0.19 that the matter ﬁelds in the vicinity of the horizon give rise to an entropy. is replaced by M . To ﬁnd the entropy we use the ﬁrst law of thermodynamics in the form dE = T dS (5. is where all the infalling matter accumulates according to external observers.
6) In other words. = exp 4πM 2 G . Let us consider the Rindler energy of the horizon. and the String Theory Revolution idea of where the cut oﬀ must occur.52 Black Holes. but the entropy per unit area is ﬁnite. Information. This suggests that we can understand the entropy in terms of the local properties of a limiting black hole of inﬁnite mass and area. This is of course not surprising.5) G 15 (5. the entropy reﬂects the number of microscopically distinct quantum states that are “coarse grained” into the single macroscopic state that we recognize as a black hole.0. ω] = i (5. It is widely believed that the nasty divergences of quantum gravity will somehow be cut oﬀ by some mechanism √ when the distance scales become smaller than G. the cutoﬀ must not be much smaller than the Planck length.7) where dN is the number of distinct quantum states with mass M in the interval dM . What is the real meaning of the black hole entropy? According to the principles stated in the introduction to these lectures. where the Planck length is given in terms of Newton’s constant as P = c3 G.0. Accordingly we write [ER (M ).0.0. The entropy diverges. By deﬁnition it is conjugate to the Rindler time ω. We assume that for a large black hole the Rindler energy is a function of the mass of the black hole. The local geometry of a limiting black hole horizon is of course Rindler space.H.19 not exceed the entropy of the black hole SBH 1 96π 2 or √ > ∼ < 2 ∼ 1 4G (5.0. The entropy of a large black hole is an extensive quantity in the sense that it is proportional to the horizon area. we can require that the contribution in equation 4. .8) Here ER is the Rindler energy which is of course the eigenvalue of the Rindler Hamiltonian. The number of such states is of order exp SB. Another way to express this is through the level density of the black hole dN ∼ exp 4πM 2 G dM (5.
0.10 in the form ∂ER = 4M G ∂M and ER = 2M 2 G (5. but it can easily be estimated.9) to obtain ER (M ).Thermodynamics of Black Holes 53 The mass and Schwarzschild time are also conjugate [M.0. The area density of Rindler energy is 1 ER = A 8πG 1 dS 2π (5. t] = i Now use ω = t 4MG (5. The swave quanta are described 1 in terms of a 1+1 dimensional quantum ﬁeld at Rindler temperature 2π . t 4MG = i or [ER (M ).13) The Rindler energy and entropy satisfy the ﬁrst law of thermodynamics dER = (5. In the same units. We ﬁrst recall that only the very low angular momentum quanta can escape the barrier.10) Finally.0.0. For simplicity.12) (5. It is of interest that the Rindler energy is also extensive.0. As we have seen. Thus we see the remarkable fact that horizons have universal local properties that behave as if a thermal membrane or stretched horizon with real physical properties were present. suppose that only the swave quanta get out. the barrier height for the swave quanta is comparable . the conjugate character of M and t allows us to write equation 5.11) The Rindler energy and the Schwarzschild mass are both just the energy of the black hole. t] = 4M G i (5. The Schwarzschild mass is the energy as reckoned by observers at inﬁnity using tclocks. the stretched horizon also radiates like a black body.0.14) 1 where 2π is the Rindler temperature. while the Rindler energy is the (dimensionless) energy as deﬁned by observers near the horizon using ωclocks.0. The exact rate of evaporation of the black hole is sensitive to many details.
Furthermore each quantum carries an energy at 1 inﬁnity of order the Schwarzschild temperature 8πMG . and the String Theory Revolution to the temperature.0.15 gives equation 5. When the mass of the black hole is large and the temperature low. unlike the image of the sun. Information.0. In terms of the Schwarzschild time. This evaporation time is evidently of order tevaporation ∼ M 3 G2 (5. The typical wavelength of a photon radiated from the sun is ∼ 10−5 cm.0. The resulting rate 1 of energy loss is of order M 2 G2 . only a few species of massless particles contribute and C is constant. The constant C depends on details such as the number of species of particles that can be treated as light enough to be thermally produced. Observing a black hole by means of its Hawking radiation will always produce a fuzzy image. while the radius of the surface of the sun is ∼ 1011 cm.54 Black Holes. It follows that approximately one quantum per unit Rindler time will excape. .0.0. The black hole on the other hand emits quanta of wavelength 1 ∼ T ∼ M G.17) 1 Using T ∼ MG and Area ∼ M 2 G2 in equation 5. the ﬂux of 1 quanta is of order MG . It is therefore not really constant. The sun is for all intents and purposes inﬁnite on the scale of the emitted photon wavelengths. Evidently energy conservation requires the black hole to lose mass at just this rate dM C = −L = − 2 2 (5.15 is essentially the Stephan–Boltzmann law L ∼ T 4 · Area (5. equation 5. If we ignore the mass dependence of C. In that case the temperature and size of the system are related in an entirely diﬀerent way. which is about equal to the Schwarzschild radius.17. We call this L.0.16) It is interesting that luminosity in equation 5. the luminosity. However the physics is very diﬀerent from that of a radiating star.15) dt M G where C is a constant of order unity.0.15 can be integrated to ﬁnd the time that the black hole survives before evaporating to zero mass.
but black strings. let us consider electrically charged black holes. black membranes. Higher dimensional Schwarzschild black holes are quite similar to their fourdimensional counterparts. and so forth. Another way to generalize the ordinary black hole is to allow it to carry gauge charge and/or angular momentum. and the threshold ﬁeld for unsupressed pair production is E ∼ m2 . The most important fact about them is that they cannot evaporate away completely. The metric for a Reissner–Nordstrom black hole is ds2 = − 1 − Q2 G 2M G + r r2 dt2 + 1 − Q2 G 2M G + r r2 −1 dr2 + r2 dΩ2 (6.0. which will discharge the black hole in the same manner as a nucleus with Z >> 137 is discharged.Chapter 6 Charged Black Holes There are a variety of ways to generalize the conventional Schwarzschild black hole.0.2) If the electric ﬁeld is too strong at the horizon. For this reason they can store inﬁnite amounts of information. Thus. They have ground states with very special and simplifying features. where me is the electron mass. and therefore have inﬁnite entropy. Pair production is exponentially e 55 . Typically black strings and branes are studied as systems of inﬁnite extent. it will cause pair production of electrons. By going to higher dimensions we can consider not only black holes.1) The electric ﬁeld is given by the familiar Coulomb law Q Er = r2 Eθ. Generally the horizon occurs at r ∼ M G.φ = 0 (6. In this lecture we will describe the main facts about charged black holes.
an outer one and an inner one. The Reissner–Nordstrom solution has two horizons.0.1 has a timelike singularity with no horizon to cloak it.0.0. They are deﬁned by 1− Q2 G 2M G + 2 r± r± = 0 (6.5) (6.6 .4) where r+ (r− ) refers to the outer (inner) horizon: r± = M G 1 ± The metric can be rewritten in the form ds2 = − (r − r+ )(r − r− ) 2 r2 dr2 + r2 dΩ2 dt + 2 r (r − r+ )(r − r− ) 2 1− Q2 M 2G (6.0.0.6) Note that in the extremal limit M 2 = Q the inner and outer horizons G merge at r± = M G. let us begin by computing the distance from r+ to an arbitrary point r > r+ .0. Q e For Q2 > M 2 G the metric in equation 6. Accordingly we restrict 2 2 our attention to the case M 2 > Q or M > Q . To examine the geometry near the outer horizon.3 is satisﬁed if Q >> 1 m2 G e Q2 G is called an extremal ∼ 1044 Black holes with charge >> 1044 can only discharge by exponentially suppressed tunneling processes.0. The question is whether they can be described by classical general relativity. Clearly they can.3) or alternativley M >> m21G2 . Clearly they cannot. A Reissner–Nordstrom G Q G black hole that saturates this relationship M 2 = black hole. Using equation 6. Information. and the String Theory Revolution suppressed if Q << m2 e M 2 G2 2 (6. The question is not whether objects with Q2 > M 2 G can exist. Such “naked singularities” indicate a breakdown of classical relativity visible to a distant observer.56 Black Holes. For practical purposes we regard them as stable. The electron is such an object. Thus equation 6.
0.7) (6.13) Evidently the horizon geometry is again well approximated by Rindler Q space. The charge density on the horizon is 4πr2 .0. For nonextremal black holes. (6.0.Charged Black Holes 57 we compute the distance ρ to be ρ = We deﬁne the following r+ + r− ≡ Σ r+ − r− ≡ ∆ y ≡ r − Σ 2 We ﬁnd ρ = y+ Σ 2 y 2 −( ∆ ) 2 ∆2 4 Σ 2 2 r (r − r+ )(r − r− ) dr (6.11) Note that the proper distance becomes inﬁnite for extremal black holes.0. Since r+ ∼ M G the + .0.0.0.10) Expanding equation 6.9 near the horizon r+ one ﬁnds ρ ≈ y − ∆ 2 1/2 2r+ ∆1/2 .9) −1 2 ∆y = y2 − + cosh The radialtime metric is given by ds = − 2 y2 − y+ ∆2 4 Σ 2 2 dt2 + dρ2 (6.8) dy (6.0.0.12) ∼ ρ2 dω 2 − dρ2 = where ω ≡ ∆ 2 t 2r+ (6. equation 6.10 becomes ds2 ∼ = ∆2 4 4r+ ρ2 dt2 − dρ2 (6.
In the limit ∆ → 0 the geometry near the horizon simpliﬁes to the form ds2 = − r+ sinh ρ r+ 2 dω 2 + dρ2 2 + r+ dΩ2 (6.58 Black Holes. Therefore the local properties of the horizon cannot be distinguished from those of a Schwarzschild black hole.0.0.18) Thus for a ﬁducial observer at a ﬁxed value of r the horizon recedes to inﬁnite proper distance as ∆ → 0.17) As the black hole tends to extremality.0.16) 1− Q2 M2G 1− Q2 M2G 2 (6. the temperature at a 1 small distance ρo from the horizon is 2πρo .19) which.0.13 we can compute the temperature as seen at inﬁnity. . although inﬁnitely far from any ﬁducial observer with r = r+ .0. is approximately Rindler. and the String Theory Revolution Q charge density is ∼ 4πM 2 G2 . the horizon becomes progressively more removed from any ﬁducial observer.0. From equation 6.15) (6. From equation 6.14) 2 ∼ 4πM 2 G2 = 4πr+ 4πM G3/2 For very massive black holes the charge density becomes vanishingly small. Thus for near extremal black holes. In particular. Information.0.0. the charge density is of the form √ M G ∼ 1 Q (6.9 we see that as ∆ → 0 ρ → y + Σ log(2y) − log∆ 2 (6. T (∞) = Using ∆ = 2M G 1 − r+ = M G 1 + We ﬁnd T (∞) = 2M G 1 − 4πM 2 G2 1+ Q2 M2G Q2 M2G ∆ 1 2 2r+ 2π (6.
Charged Black Holes
59
We note from equation 6.0.17 that in the extremal limit the temperature at inﬁnity tends to zero. The entropy, however, does not tend to zero. This can be seen in two ways, by focusing either on the region very near the horizon or the region at inﬁnity. As we have seen, the local properties of the horizon even in the extreme limit are identical to the Schwarzschild case 1 from which we deduce an entropy density 4G . Accordingly, S =
2 πr+ Area = = πM 2 G 4G G
(6.0.20)
We can deduce this result by using the ﬁrst law together with equation 6.0.17 dM = T dS (Fixed Q)
to obtain S = Area as a general rule. 4G The fact that the temperature goes to zero in the extreme limit indicates that the evaporation process slows down and does not proceed past the √ point Q = M G. In other words, the extreme limit can be viewed as the ground state of the charged black hole. However it is unusual in that the entropy does not also tend to zero. This indicates that the ground state is highly degenerate with a degeneracy ∼ eS . Whether this degeneracy is exact or only approximate can not presently be answered in the general case. However in certain supersymmetric cases the supersymmetry requires exact degeneracy. The metric in equation 6.0.19 for extremal black holes can be written in a form analogous to equations 1.3.18 and 1.4.21 by introducing a radial variable eρ/r+ − 1 R = ρ/r r+ e + +1 The metric then takes the form 2 2 dτ 2 = R2 dω 2 − dR2 − r+ dΩ2 2 1 − R2 r
+
(6.0.21)
(6.0.22)
R Obviously the physics near r+ → 0 is identical to Rindler space, from which it follows that the horizon will have the usual properties of temperature, entropy, and a thermal atmosphere including particles of high angular momenta trapped near the horizon by a centrifugal barrier.
60
Black Holes, Information, and the String Theory Revolution
Although the external geometry of an extreme or near extreme Reissner– Nordstrom black hole is very smooth with no large curvature, one can nevertheless expect important quantum eﬀects in its structure. To understand why, consider the fact that as ∆ → 0 the horizon recedes to inﬁnity. Classically, if we drop the smallest amount of energy into the extreme black hole, the location of the horizon, as measured by its proper distance, jumps an inﬁnite amount. In other words, the location of the horizon of an extremal black hole is very unstable. Under these circumstances, quantum ﬂuctuations can be expected to make the location very uncertain. Whether this eﬀect leads to a lifting of the enormous degeneracy of ground states or any other physical phenomena is not known at present except in the supersymmetric case.
Chapter 7
The Stretched Horizon
Thus far our description of the nearhorizon region of black holes, or Rindler space, has been in terms of quantum ﬁeld theory in a ﬁxed background geometry. But we have already run into a contradiction in applying quantum ﬁeld theory, although we didn’t spell it out. The problem arose in Chapter 4 when we found that the entropy per unit area of the horizon diverges as the cutoﬀ tends to zero (see equation 4.0.22). That in itself is not a problem. What makes it a problem is that we later found that A black hole thermodynamics requires the entropy to be 4G . Free quantum ﬁeld theory is giving too much entropy in modes very close to the horizon, where the local temperature diverges. The fact that the entropy is inﬁnite in quantum ﬁeld theory implies that any quantity that depends on the ﬁniteness of the entropy will be miscalculated using quantum ﬁeld theory. One possibility is that we have overestimated the entropy by assuming free ﬁeld theory. Equation 4.0.20 could be modiﬁed by interactions. Indeed that is so, but the eﬀect goes in the wrong direction. The correct entropy density for a general ﬁeld theory can always be parameterized by S(T ) = γ(T ) T 3 where γ(T ) represents the number of “eﬀective” degrees of freedom at temperature T . It is widely accepted and in many cases proven that γ(T ) is a monotonically increasing function of T . Thus, conventional interactions are only likely to make things worse. What we need is some new kind of theory that has the eﬀective number of degrees of freedom going to zero very close to the horizon. Let’s suppose that ordinary quantum ﬁeld theory is adequate down to distance scale . In order that the entropy at distance greater than not exceed the Bekenstein–Hawking value, we must have the rough inequality
61
4) ˙ ∂A where A means ∂ω and φ = −A0 . it is suﬃcient to study electrodynamics in Rindler space.3) − ρ ∇×A 2 W = + j · A dω dρ d2 x⊥ (7. As usual ˙ E = −∇φ − A B = ∇×A . and j is a conserved current in the usual sense ∂µ j µ = 0.2) (7. or “stretched” horizon at a distance of roughtly one Planck length from the mathematical horizon. This means that real dynamics and evolution can take place on the stretched horizon. and the String Theory Revolution 2 < ∼ G = 2 P √ Evidently at distances less than G from the horizon the degrees of freedom must be very sparse. Information.0. Instead of being lightlike.0. or even nonexistent. the stretched horizon has dynamics of its own that includes such phenomena as viscosity and electrical conductivity. Stretching the horizon has another beneﬁt. The metric is dτ 2 = ρ2 dω 2 − dρ2 − dx2 ⊥ The stretched horizon is just the surface ρ = ρo where ρo is a length of order the Planck length. To see that the horizon of a black hole has electrical properties.1) (7.0. The action for the electromagnetic ﬁeld in Rindler space is √ −g µν στ g g Fµσ Fντ + j µ Aµ dω dρ d2 x⊥ W = 16π or. First let us deﬁne the stretched horizon. a system at the stretched horizon is timelike. This leads to the idea that the mathematical horizon should be replaced by an eﬀective membrane.0.62 Black Holes. As we will see. substituting the form of the metric ˙ 1 A + ∇φ 8π ρ 2 (7.
the distance of the charges from the stretched horizon is macroscopic. or more precisely σ = =− 1 4πρ 1 4πρ Eρ ∂ρ φ ρ=ρo (7. Since the charges are slowly moving in Rindler coordinates. it means that they are experiencing proper acceleration.0. The surface charge density on the stretched horizon is easily deﬁned. the action becomes 2 E 2 1 W = − ρ B + j · A dω dρ d2 x⊥ 8π ρ and the Maxwell equations are 1 ρ (7. It is just the component of the electric ﬁeld perpendicular to the stretched horizon. Thus 2 ∂ρ φ − 1 ∂ρ φ = −∇2 φ ⊥ ρ (7.The Stretched Horizon 63 With these deﬁnitions. We also assume all length scales associated with the charges are much larger than ρo .0.0.6) ∇· 1 ρ E = 4πj 0 ∇·B = 0 We begin by considering electrostatics. the third expression in equation 7.7) ρ=ρo Working in the Coulomb gauge.6 becomes ∇· 1 E ρ = −∇ · 1 ∇φ ρ = 0 (7.5) ˙ E − ∇ × (ρB) = −4πj ˙ B + ∇×E = 0 (7. By electrostatics we mean the study of ﬁelds due to stationary or slowly moving charges placed outside the horizon.0.9) . In particular.8) near the stretched horizon.0.0.
0.7 and using Maxwell’s equations 7.64 Black Holes. Since the black hole horizon is compact.0. We can easily identify the surface current density.11 is simpliﬁed to ∇2 F = 0 ⊥ (7.9 and evaluating at ρ = ρo gives ∇2 F + ρ2 ∇2 G = 0 ⊥ o ⊥ (7.0.6 gives 4π σ = ˙ 1 ˙ Eρ = ρo ∇ × ρB (7.10) Plugging equation 7. From Maxwell’s equations we obtain ˙ Bx = ∂ρ Ey ˙ By = −∂ρ Ex 1 ˙ ρ Ex 1 ˙ ρ Ey (7. Taking the time derivative of equation 7.0. This is an interesting result. and the String Theory Revolution We can solve this equation near the horizon by the ansatz φ ∼ ρα . which proves that the horizon behaves like an electrical conductor.15) = −∂ρ (ρ By ) = ∂ρ (ρ Bx ) . then equation 7.0.0.0. We easily ﬁnd that α = 0 or α = 2.14) 4π jy = ρ Bx Now let us consider an electromagnetic wave propagating toward the stretched horizon along the ρ axis.0.10 into equation 7. equation 7. Thus we assume φ = F (x⊥ ) + ρ2 G(x⊥ ) + terms higher order in ρ (7. The right hand side will be smaller than the left hand side by 2 powers of ρ and can therefore be ignored.12) A similar equation can also be derived for the ﬁnite mass black hole.12 proves that φ = constant on the horizon.11) If ρo is much smaller than all other length scales.0.0.13) ρ Evidently this is a continuity equation if we deﬁne: 4π jx = −ρ By (7.0.0. Information.
For such waves. The resistance of a twodimensional resistor is scale invariant and only depends on the shape.1.0.The Stretched Horizon 65 To make the equation more familiar.0.0.15 then becomes ˙ βx = ∂u Ey ˙ βy = −∂u Ex (7.14 jx = jy = 1 4π 1 4π (7.18) βy = −Ex or from equation 7. That corresponds to a surface resistance of 377Ω/square. one might conclude that the point charge just asymptotically approaches the horizon ∗ The unit Ω/square is not a misprint. Since the horizon is just ordinary ﬂat space.0.0. .0. the Maxwell equations 7.17 give βx = Ey (7.16) Ex Ey Evidently the horizon is an ohmic conductor with a resistivity of 4π. However the physics only makes sense for waves propagating toward the horizon from outside the black hole. we can redeﬁne the magnetic ﬁeld ρB = β and use tortoise coordinates u = log ρ Equation 7. a current will ﬂow precisely as if the horizon were a conducting surface. if a circuit is constructed as in Figure 7.∗ For example.17) ˙ Ex = ∂u βy ˙ Ey = −∂u βx The mathematical equations allow solutions in which the wave propagates in either direction along the uaxis. As a last example let us consider dropping a charged point particle into the horizon.
21) .0. and the String Theory Revolution A A Horizon Fig. ammeter attached to horizon with the transverse charge density remaining pointlike. Without loss of generality. we can take the charge to be at rest at position zo in Minkowski coordinates. However this is not at all what happens on the stretched horizon. 7.2. The calculation is easy because at any given time the Rindler coordinates are related to the Minkowski coordinates by a boost along the zaxis.0.19) 3/2 [(ρ cosh ω − zo )2 + x2 ] ⊥ we ﬁnd ρo σ = e 4πρo ρo cosh ω − zo (ρo cosh ω − zo ) + x2 ⊥ 2 3/2 (7. To compute the surface charge density on the stretched horizon. Since the component of electric ﬁeld along the boost direction is invariant. Information.20) Now let’s consider the surface density for large Rindler time.0. σ = e ρo e ω 4πρo [ρ2 e2ω + x2 ]3/2 o ⊥ (7. we need to determine the ﬁeld component Eρ .1 Battery.66 Black Holes. This process is shown in Figure 7. we can write the standard Coulomb ﬁeld Eρ = Ez = = Using 4π σ = Eρ ρ [(z−zo )2 + x2 ]3/2 ⊥ e (ρ cosh ω − zo ) e (z−zo ) (7.
For a real black hole.0. it is convenient to rescale x⊥ using x⊥ = eω y⊥ to obtain σ = e−2ω e 4π (ρ2 + y 2 )3/2 o ⊥ (7.22) It is evident that the charge spreads out at an exponential rate with Rindler time.The Stretched Horizon 67 T H or iz o n Freely falling charge Z0 Z Stretched horizon Fig.2 Charge falling past the stretched horizon To better understand this expression. it would spread over the horizon in a time ω = log (Rs − ρo ) = log (2M G − ρo ) or in terms of the Schwarzschild time t = 4M Gω = 4M G log (2M G − ρo ) . 7.
and conservation of charge will cause it to spread exponentially. entropy. it exhibits dissipative eﬀects such as electrical resistivity and viscosity. Evidently the surface charge density will exponentially ˙ decrease.68 Black Holes. The surprising and puzzling thing is that they are completely unnoticed by a freely falling observer who falls through the horizon! . and the String Theory Revolution The exponential spreading of the charge is characteristic of an Ohm’s law conductor. In addition to temperature. To see this we use Ohm’s law j = conductivity E. and energy. Information. Now use the continuity equation get the relation σ ∼ −σ. Taking the divergence gives ∇ · j ∼ ∇ · E ∼ σ. Thus we see that the horizon has the properties of a more or less conventional hot conducting membrane.
information is lost because for most cases of interest the region Γ becomes very complicated like a fractal. Now we let the system evolve. As a deﬁnition of coarse graining. The region Γ(0) = Γ evolves into the region Γ(t).Chapter 8 The Laws of Nature In this chapter. it will appear that Γ is growing. the conservation of information is expressed as 69 . This is the origin of the second law of thermodynamics. In a practical sense. The region Γ(0) has a volume VΓ in the phase space. we might represent this by specifying an initial region Γ(0) in the system’s phase space.1. In this sense the amount of information is conserved. These laws are: 1) The principle of information conservation 2) The equivalence principle 3) The quantum xerox principle 8. This is illustrated in Figure 8. In quantum mechanics. and if we coarse grain the phase space. Liouville’s theorem tells us that the volume of Γ(t) is exactly the same as that of Γ. If we begin following a system with some limited knowledge of its exact state. which indeed grows. we want to review three fundamental laws of nature whose compatibility has been challenged.1 Information Conservation In both classical and quantum mechanics there is a very precise sense in which information is never lost from a closed isolated system. the union of those spheres is the “coarse grained” volume of phase space. if one takes every point in the phase space and surrounds it by solid spheres of ﬁxed volume. In classical physics the principle is embodied in Liouville’s theorem: the conservation of phase space volume.
and the String Theory Revolution Fig. The analog of the phase space volume is the dimensionality of the subspace N = TrP The unitarity of the time evolution insures that N is conserved with time. 8.1 Evolution of a ﬁxed volume in phase space the unitarity of the Smatrix. Similarly. A more reﬁned deﬁnition of information is provided by the concept of entropy. then S = log VΓ . q) log ρ(p. Thus the entropy is an estir mate of the logarithm of the number of quantum states that make up the initial ensemble.1.70 Black Holes. q) in phase space. where S = − dp dq ρ(p.2) For the case ρ = T P P the entropy is log N . instead of a deﬁnite state. isolated system. P . In this case the ﬁne grained or Von Neumann entropy is S = −T r ρ log ρ (8. q) (8. we instead specify a probability density ρ(p. we might express this by a projection operator onto a subspace. A generalization of the volume is given by the exponential of the entropy VΓ → exp S. If we again approach a system with limited knowledge.1. Information. Suppose that instead of specifying a region Γ in phase space. .1) 1 It is easy to check that if ρ = VΓ inside Γ and zero outside. In both quantum mechanics and classical mechanics the equations of motion insure the exact conservation of S for a closed. for quantum mechanics the sharp projector P can be replaced by a density matrix ρ.
the only reason for introducing a phase space probability is a lack of detailed knowledge of the state.6) . Trρ = 1 (8. Entanglement refers to quantum correlations between the system under investigation and a second system. Consider a composite system composed of 2 subsystems A and B. The subsystem A(B) is described by some complete set of commuting observables α (β). In this case the subsystem A is in a pure state.2. β) Ψ(α. Let us assume that the composite system is in a pure state with wave function Ψ(α.3) ∗ α Ψ (α.The Laws of Nature 71 8. all the others must vanish. the constituent subsystems are generally not described by pure states. More precisely. it involves separating a system into two or more subsystems. but we could equally well focus on ρB . This only happens if the composite wave function factorizes Ψ(α. β ) The fact that a subsystem is described by a density matrix and not a pure state may not be due to any lack of knowledge of the state of the composite system. If one of the eigenvalues of ρA is equal to 1.5) It follows that all the eigenvalues are between zero and one.2. β) Ψ(α . entanglement. For deﬁniteness.2 Entanglement Entropy In classical physics. β) = ψA (α) ψB (β) (8. β) (8. 3) The density matrix is normalized to 1. β). All measurements performed on A(B) are describable in terms of a density matrix ρA (ρB ). The result is an “entanglement entropy” for the subsystems. Consider now the subsystems separately. there is another reason. consider ρA . (ρA )αα = (ρB )ββ = β Ψ∗ (α. Even in the case of a pure state.2. Let us consider some properties of the density matrix. This means its eigenvalues are all either positive or zero. In quantum mechanics. 1) The density matrix is Hermitian (ρA )αα = (ρA )∗ α α (8.2.4) 2) The density matrix is positive semideﬁnite.
2. β) φ∗ (α) = λ χ(β) Thus χ(β) is an eigenvector of ρB with eigenvalue λ. From the equality of the nonvanishing eigenvalues of ρA and ρB an important property of entanglement entropy follows: SA = −T r ρA log ρA = SB (8. The equality of SA and SB is only true if the combined state is pure. It is a general property of most complex interacting systems that the density matrix of a small subsystem will be thermal ρi = e−β Hi Zi (8. In that case. and the String Theory Revolution In this case B is also in a pure state. β ) φ∗ (α ) = λ α Ψ∗ (α. Each subsystem on the average will have energy . β ) χ(β ) Ψ∗ (α. the entropy of the composite system vanishes SA+B = 0 Evidently entropy is not additive in general. 4) The nonzero eigenvalues of ρA and ρB are equal if the composite system is in a pure state. β) Ψ(α. Then the eigenvalue condition is βα Ψ∗ (α. To prove this. (ρB )ββ χ(β ) = = αα β αβ Ψ∗ (α. Next. β) Ψ(α . β ) Ψ∗ (α . let us consider a large system Σ that is composed of many similar small subsystems σi . and the entire system is in a pure state with total energy E. we start with the eigenvalue equation for ρA .2. β) Ψ(α.72 Black Holes. β ) φ∗ (α ). Information. Call φ the eigenvector of ρA .8) . β) φ(α ) = λ φ(α) We assume λ = 0.7) Thus we can just refer to the entanglement entropy as SE . Let us suppose the subsystems weakly interact. Now we deﬁne a candidate eigenvector of ρB by χ(β ) ≡ Then β α Ψ∗ (α .
we will assume that the entire system Σ is in a pure state with vanishing entropy. as Σ1 approaches Σ. The information can be deﬁned by I = ST hermal − S (8.11) . In general large subsystems or the entire system will not be thermal. many. Another concept that we can now make precise is the information in a subsystem. The entropy of each subsystem Si as well as the entropy of the whole system Σ given by σΣ . The thermal density matrix maximizes the entropy for a given average energy . and the coarse grained entropy of Σ all vanish. It is not conserved.10) For example. To see why. the subsystem entropies become nonzero Si = 0 and the coarse grained entropy also becomes nonzero ST hermal = i Si = 0 However. Typically the ﬁne grained entropy of Σ1 is deﬁned as the entanglement entropy S(Σ1 ) of Σ1 with the remaining subsystem Σ − Σ1 . or all of the subsystems σi . meaning that it now fails to factorize.2. This will always be less than the coarse grained entropy of Σ1 ST hermal (Σ1 ) > S(Σ1 ) (8.The Laws of Nature 73 where Hi is the energy of the subsystem.2. In this case. The wave function develops correlations.9) By deﬁnition it is additive. Now the subsystems interact. In fact. Let us consider an arbitrary subsystem Σ1 of Σ which may consist of one.2. the ﬁne grained entropy S(Σ1 ) will tend to zero. The coarse grained entropy is what we usually think of in the context of thermodynamics. suppose we start with the subsystems in a product state with no corelations. The coarse grained or thermal entropy of the composite system is deﬁned to be the sum of the entropies of the small subsystems ST hermal = i Si (8. the “ﬁne grained” entropy of Σ is exactly conserved and therefore remains zero.
How much information does 2 it have? To compute it. and the String Theory Revolution Often the coarse grained entropy is the thermal entropy of the system. It can be thought of as the hidden subtle correlations between subsystems that make the state of Σ pure. A bit is the entropy of a two state system if nothing is known[2]. Typically for subsystems less than half the size of Σ the information is smaller than 1 bit. Thus for Σ1 < 1 Σ 2 S(Σ1 ) ∼ ST hermal (Σ1 ) = I(Σ1 ) ≈ 0 Next consider a subsystem with Σ1 > 1 Σ.13) (8. The numerical value of a bit is log 2. the information is negligible. the information in a small subsystem vanishes. Since typical small subsystems have a thermal density matrix. so that the information is the diﬀerence between coarse grained and ﬁne grained entropy. What actually happens is that for subsystems smaller than about 1/2 of the total system. However for 2 . The subsystem 1 Σ has about 1 bit of 2 information. this is not so.2.12) The coarse grained entropy of Σ − Σ1 will be of order (1 − f ) ST hermal (Σ). for Σ1 < 1 Σ the information in Σ1 is essentially zero. Thus. How much information are in a moderately sized subsystem? One might think that the information smoothly varies from zero (for the σi ) to SCoarse Grained (for Σ). Entropy and information are naturally measured in “bits”.74 Black Holes. where f is the fraction of the total degrees of freedom contained in Σ1 . we use two facts: S(Σ − Σ1 ) = S(Σ1 ) S(Σ − Σ1 ) ∼ ST hermal (Σ − Σ1 ) = Thus S(Σ1 ) ∼ ST hermal (Σ − Σ1 ) = (8. Information.2. At the opposite extreme the information of the combined system Σ is just its total thermal entropy. However.
it ﬁlls the box with thermal radiation. its ground state.2. in this case.15) (8. outgoing photons. When the bomb explodes. The thermal or coarse grained entropy as well as the ﬁne grained entropy in the box tends to zero.2. but its ﬁne grained entropy does not. Consider a box with perfectly reﬂecting walls. begins to increase. no photons have yet escaped. which is equal for A and B. Initially the bomb is in its ground state. so S(A) = 0 at this time. The thermal entropy in the box decreases: SEntanglement = 0 ST hermal (B) = 0 ST hermal (A) = 0 Eventually. all of the photons escape the box. (8. The box has a small hole that allows the thermal radiation to slowly leak out. For now let’s consider a conventional system which is described by known laws of physics. and B has vanishing entropy. or coarse grained. Inside the box we have a bomb which can explode and ﬁll the box with radiation. The entanglement entropy. The thermal. Furthermore. entropy of the box increases.14) .16) (8. The subsystem A consists of everyting outside of the box. The box is in a pure state.The Laws of Nature 75 Σ1 > 1 Σ we get the information to be 2 I(Σ1 ) = ST hermal (Σ1 ) − S(Σ1 ) = f Sthermal (Σ) − (1 − f ) ST hermal (Σ) = ( 2 f − 1 ) ST hermal (Σ) We will eventually be interested in the information emitted by a black hole when it evaporates. The entire system Σ consists of the subsystem B that includes everything in the box. ST hermal (B) = 0 S(B) = 0 S(A) = 0 Next.2. The result is that the interior and exterior of the box become entangled. photons slowly leak out.
the thermal or coarse grained entropy of the exterior radiation has increased to its ﬁnal value.2 Top: Von Neumann entropy of Σ1 vs fraction f. 8. Middle: coarse grained entropy of Σ1 vs fraction f. But the ﬁne grained entropy of A must vanish. since the entanglement has gone to zero.76 Black Holes. . The second law of thermodynamics insures that ST hermal (A) is larger than ST hermal (B) just after the explosion. Bottom: information vs fraction of total degrees of freedom in Σ1 At this time. Information. and the String Theory Revolution S (Σ 1 ) Pure state 1 2  1 f SThermal (Σ 1 ) Additive coarsegrained entropy 1 2  1 f I (Σ 1 ) 1 2  1 f Fig.
The curvature typically satisﬁes R ∼ 1 (MG)2 Any freely falling system of size much smaller than M G will not be distorted or otherwise disrupted by the presence of the horizon. Roughly the point where information appears outside of the box is the point where half of the ﬁnal entropy of the photons has emerged.3. It is useful to deﬁne this time at which information begins to emerge.3 Equivalence Principle A second law of nature concerns the nature of gravitation. In its simplest form the equivalence principle says that a gravitational ﬁeld is locally equivalent to an accelerated frame. but no information. Thus we see how information conservation works for a conventional quantum system. The description of the evolution of the various kinds of entropy follow from very general principles. We have seen that the magnitude of the curvature components at the horizon are small and tend to zero as the mass and radius of the black hole increase. More exactly. However.The Laws of Nature 77 The actual entanglement entropy must be less than the thermal entropy of A or B. Thus we regard the conservation of information in black hole evaporation as a fundamental law of nature. 8. this does not mean that localized regions containing a small fraction of the photons cannot be extremely thermal. Note that it applies to observations made from outside the black hole. The consequence of this principle is the ﬁnal radiation ﬁeld outside the box must be in a pure state. Before that point. it says that a freely falling observer or system will not experience the eﬀects of gravity except through the tidal forces. It is the amount of time that it takes to retrieve a single bit of information about the initial state of the box. . or equivalently. This time is called the information retention time. Note that the point at which ST hermal (A) = ST hermal (B) deﬁnes the time at which the information in the outside radiation begins to grow. the curvature components. a good deal of energy has escaped. Thus a plot of the various entropies looks like Figure 8. They typically carry negligible information.
78 Black Holes. Bottom: evolution of entanglement entropy and information . 8.3 Top: evolution of the thermal entropies of box and exterior. and the String Theory Revolution S Thermal A (External) B Information retention time (Box) t Information trop t en emen ngl Enta y Information retention time t Fig. Information.
If the spin is in the up state with In Out Fig.17) Now suppose that the spin is inserted with its polarization along the xaxis. 8. 8.4 Quantum Xerox Principle The third law of nature that plays an important role in the next lecture is the impossibility of faithfully duplicating quantum information.4. What it says is that a particular kind of apparatus cannot be built.4 Schematic diagram of quantum Xerox machine respect to the zaxis. namely.18) (8. and which will copy that system. 1 √ (↑ + ↓ ) 2 ∗ The (8.∗ It is a machine into which any system can be inserted. We call it a Quantum Xerox Machine. imagine that we insert a spin in the input port as in Figure 8. i. To see why such a system is not possible.4.19) quantum Xerox principle is sometimes called the nocloning principle . it is duplicated ↑ → ↑ ↑ Similarly. if it is in the down state.4.4.The Laws of Nature 79 The equivalence principle requires the horizon of a very large black hole to have the same eﬀects on a freely falling observer as the horizon of Rindler space has. producing the original and a duplicate. no eﬀect at all. it is duplicated ↓ → ↓ ↓ (8.e.
4.4.4.21) 1 2 = ↑ ↑ + 1 2 ↑ ↓ + ↓ ↑ + ↓ ↓ The state in equation 8. Thus from equations 8.17 and 8. Thus we see that the principle of linearity forbids the existence of quantum Xerox machines. .20) On the other hand.19 1 √ 2 (↑ + ↓ ) → 1 2 1 √ 2 (↑ + ↓ ) 1 2 1 √ 2 (↑ + ↓ ) (8.4.4.20 is obviously not the same as that in equation 8. we would be able to violate the Heisenberg uncertainty principle by a set of measurements on those states.18 1 1 √ (↑ + ↓ ) → √ (↑ ↑ + ↓ ↓ ) 2 2 (8.4.21. If we could construct Xeroxed quantum states.4. Information.80 Black Holes. a true quantum Xerox machine is required to duplicate the spin in equation 8. and the String Theory Revolution The general principles of quantum mechanics require the state to evolve linearly.
the evolution of an initial state ψin is governed by a unitary Smatrix.1. It gives rise to an inﬁnite entropy density on the horizon.1) One way of stating the principle of information conservation is through the unitarity of S. Following Hawking. Let us concentrate on a theory of massless particles. The point is that a unitary matrix has an inverse. Hawking did not mean the practical loss of information such as would occur in the bombinthebox experiment in Chapter 8. the ﬁrst of the laws of nature described in the Introduction would be violated. namely local quantum ﬁeld theory in the ﬁxed background of a black hole. let’s begin with a black hole that has been formed during a collapse. instead of the correct Bekenstein–Hawking c3 value of 4G . so that the ﬁnal state is given by ψout = S ψin (9. it played a central role in replacing the old ideas of locality with a new paradigm. By information loss. information would be lost. To state the problem. Although Hawking’s conclusion is undoubtedly wrong. As we have already seen. quantum ﬁeld theory must be replaced with an entirely new paradigm in which the concept of locality is radically altered. so that 81 . As we will see. we think of the geometry as a background on which we can formulate quantum ﬁeld theory. The Penrose diagram is shown in Figure 9.0. In other words. According to the principles of quantum mechanics. He meant that in a ﬁne grained sense. It was based on the only available tool of that time. The argument was simple and persuasive.Chapter 9 The Puzzle of Information Conservation in Black Hole Environments In 1976 Hawking raised the question of whether information is lost in the process of formation and evaporation of black holes. quantum ﬁeld theory has a serious defect when it comes to describing systems with horizons.
9. and the String Theory Revolution Singularity H or iz on + Collapsing energy  Fig. It is evident from Figure 9.82 Black Holes.3) In arguing that the ﬁnal Hilbert space is a tensor product.0. Thus hin = hI− hout = hI+ ⊗ hS (9.2 that some of the ﬁnal particles will escape to I + and some will be lost behind the horizon.2) Now let us consider the Smatrix in a black hole background. Information. These incoming particles will interact and scatter by means of Feynman diagrams in the black hole background.1 Penrose diagram of black hole formed by collapse in principle the initial state can be recovered from the ﬁnal state ψin = S † ψout (9. Hawking . hout is a tensor product of the states on I+ and those at the singularity S. The Hilbert space of initial states hin is clearly associated with quanta coming in from I − . Evidently the ﬁnal Hilbert space.0.
2 Feynman diagrams in black hole background relied on an important property of quantum ﬁeld theory: locality. Let us consider the description of the ﬁnal state from the viewpoint of the observers at I + .The Puzzle of Information Conservation in Black Hole Environments 83 S +  Fig.0. Since every point on the singularity S is spacelike relative to every point on I + . Thus.1. in general the observer external to the black hole will see a state characterized by a density matrix. Now suppose that the ﬁnal state is given by an Smatrix which connects hin and hout as in equation 9. Since they have no access to S. . 9.4) where T rsingularity means a trace over the states on the singularity.0. This was central to Hawking’s analysis. all experiments on the outgoing particles are described by a density matrix ρout = T rsingularity ψout ψout  (9. the operators on S all commute with those on I + .
Thus. especially by relativists. when the singularity is “exposed” at Planckian temperature. a law of nature would be violated from the viewpoint of the external observer.1 A Brick Wall? There are two more possibilities worth pointing out. the horizon bounces everything out. The reason that this was never seriously entertained. In other words. or even inﬁnite entropy. it must begin to 2 come out in the emitted radiation. If one particle ends up on S and the other ends up on I + . then the ﬁnal state will be entangled. 9. By correlated. Instead they end their lives as stable Planckmass remnants that contain all the lost information. A freely falling observer would encounter a “brick wall” just above the horizon. from the viewpoints of an infalling system. . the entanglement entropy of the radiation cannot be larger than the black hole’s remaining entropy. This is a fundamental law of quantum mechanics. is that it badly violates the equivalence principle. The situation is even worse if the information is not emitted at all. we mean that the two particles wave function is entangled. Obviously such remnants would have to have an enormous. Information. In this case the state on I + will not be pure. we have seen in Chapter 8 that the maximum amount of information that can be hidden in a system is its entropy. Once the entropy of the black hole has evaporated past 1 its original value. Such remnants would be extremely pathological. One is that the horizon is not penetrable. or the information is restored to the outside world at the very end of the evaporation process. A ﬁnal possibility that was advocated by some authors is that black holes never completely evaporate. In fact. However. Hawking went on to make arguments that the purity of the state would not be restored if the black hole evaporated. and the String Theory Revolution A simple example might involve a pair of correlated particles which are thrown in from I − . and we will not pursue that line further. even if all information were emitted at the very end of the evaporation process. By the time the black hole has small mass and entropy. the only possibilities would seem to be that either information is lost during the entire process of formation and evaporation. Since the near horizon region of a Schwarzschild black hole is essentially ﬂat spacetime.84 Black Holes.
and eventually reemit all information in the form of Hawking radiation. Thus. In other words. Furthermore. the horizon cannot duplicate information. Even more convincing is the fact that the horizon of a black hole formed by a lightlike shell forms before the shell gets to the center (see Figures 1. But a potential contradiction can occur for the infalling observer. on the other hand. Evidently we have come to an impasse. No high temperatures or other anomolies are encountered. at least for some observer. the horizon is seen as ﬂat featureless spacetime. black hole complementarity just says that no observer ever witnesses a violation of a law of nature.2 Black Hole Complementarity In its simplest form. The information conservation principle requires all information to be returned to the outside in Hawking radiation.10 and 1. the quantum Xerox principle closes out the last possibility. . At any given time. and send one copy into the black hole while sending a second copy out. black hole complementarity tells us that the equivalence principle is respected. but it allows us to estimate their number. the ﬁne grained entropy of the radiation ﬁeld (due to entanglement) cannot exceed the entropy of the black hole.The Puzzle of Information Conservation in Black Hole Environments 85 any violent disturbance to a freely falling system would violate the second law of nature in the Introduction. This means that as long as the black hole is much larger than the infalling system. thermalize. all information is carried oﬀ in Hawking radiation. for an external observer it says: A black hole is a complex system whose entropy is a measure of its capacity to store information. It tells us that the entropy is the log of the number of microstates of the degrees of freedom that make up the black hole. It seems that some law of nature must break down. That number is of order the area of the horizon in Planck units. The equivalence principle. For a freely falling observer. requires information to freely pass through the horizon. it tells us that the microdegrees of freedom can absorb. 9. since the infalling observer cannot send reports from behind the horizon.12). The quantum Xerox principle precludes both happening. Finally. At the end of evaporation. It does not tell us what those microdegrees of freedom are. No obvious contradiction is posed for the external observer.
We don’t actually need observer B to decode the information. All we really need is a mirror outside the black hole horizon to reﬂect the Hawking radiation back into the black hole. and the String Theory Revolution Singularity C M es sa g In fo rm at io n e B A Fig.86 Black Holes. Next. it passes through the horizon without incident. shown in Figure 9. consider an observer B who hovers above the black hole monitoring the Hawking radiation. . After collecting some information about A (from the Hawking radiation ). observer B then jumps into the black hole. 9. Information. System A is assumed to contain some information. According to observations done by A. the photons recorded by observer B encode the information carried in by system A.3 along with an infalling system A. According to assumption.3 Information exchange from External to infalling observer Consider a black hole.
it must ∗ all occur at x− < (M G) e−ω . If this experiment is possible.The Puzzle of Information Conservation in Black Hole Environments 87 We now have two copies of the original information carried by A. This means that observer B’s jump oﬀ point must correspond to Rindler coordinates satisfying ω∗ ≥ ρ∗ ≥ t∗ 4MG P ≈ M2 G (9.8) Thus.9) The implication is that if A is to send a signal that B can receive. In other words. observer B hovers above the stretched horizon.2.2.6) In terms of the light cone coordinates x± = ρe±ω we have x+ x− > ∗ ∗ x+ ∗ > ≈ P 2 P exp (ω ∗ ) (9. In fact. Now we have a contradiction. We can imagine A sending a signal to observer B so that observer B discovers the duplicate information at point C in Figure 9.3.16. we need to remember that no information will be emitted until about 1 the entropy of 2 the black hole has evaporated.2. To see why the experiment fails. the singularity is given by x+ x− = (M G) 2 (9. this takes a time of order t∗ ≈ M 3 G2 (9. .2.0. then black hole complementarity is not selfconsistent.5) Let us also assume that the observer B hovers above the horizon at a distance at least of the order of the Planck length P .4 we can see why there might be a problem with the experiment. since observer B has discovered a quantum Xeroxing of information from observer A. observer B will hit the singularity at a point with x− < ∼ (M G) e−ω ∗ (9. This means that A has a time of order ∗ ∆t ≈ (M G) e−ω to send the message.2.7) From Figure 9. From equation 5. Observer B might be forced to the singularity before a message can arrive.
to send a single bit requires at least one quantum.4 Now.= (MG) + X B gathers information and enters horizon Information from A A P B VV VVVVVVVVVVVVV VVV VV VV VV VV V X Resolution of Xerox paradox for observers within horizon (9. 9.88 Black Holes. Information. This example is one of many which show how the constraints of quanM2 G VVVVVVVVV VVVV VV VV VV V VV VV V VV X + Singularity 2 X . the energy carried by observer A must be many orders of magnitude larger than the black hole mass. In other words. Since that quantum must be emitted between x− = 0 and ∗ x− = M G e−ω .2. its energy must satisfy E > ∗ 1 eω MG From equation 9. in the classical theory.6 we see that this energy is exponential in the square of the black hole mass (in Planck units) E > eM G . and the String Theory Revolution M es sa ge H or fro iz m on A Fig. However in quantum mechanics. It is obvious that A cannot ﬁt into the horizon.2. there are no limits on how much information can be sent in an arbitrarily small time with arbitrarily small energy. and that the experiment cannot be done. for observer A to be able to signal observer B before observer B hits the singularity.10) .
and its temperature is 10−8 electron volts. Baryon number must be violated by quantum gravitational eﬀects. the mass of the black hole at this point is about 1010 kilograms. Thus it can disappear without some ﬂux having to suddenly change at inﬁnity. as in Figure 9. It is far too cold to emit anything but very low energy photons and gravitons. The obvious diﬀerence between baryon number and electric charge is that baryon number is not the source of a long range gauge ﬁeld. Since the Xboson is described by a real ﬁeld. Using the connection between mass and temperature in equation 4. unlike electric charge. and even if it were to decay into nothing but protons. It can only begin to radiate baryons when its temperature has increased to about 1 GeV. and at some point it is hot enough to emit massive neutrinos and antineutrinos. and pions.0. most modern theories predict baryon violation by ordinary quantum ﬁeld theoretic processes. Nevertheless. As a simpliﬁed example. and the transition evidently violates baryon conservation. its temperature increases. It is formed by the collapse of roughly 1057 nucleons.5. it cannot carry any quantum numbers. it could produce only about 1037 of them. . Consider a typical black hole of stellar mass. As it radiates. The proton could then decay into a positron and an electronpositron pair. then electrons. This is a tiny fraction of the original black hole mass. as well as between two positrons. None of these carry oﬀ baryon number. muons. Its Schwarzschild radius is about 1 kilometer.The Puzzle of Information Conservation in Black Hole Environments 89 tum mechanics combine with those of relativity to forbid violations of the complementarity principle. can at best be an approximate conservation law. there are powerful reasons to believe that baryon number.22. let’s suppose there is a heavy scalar particle X which can mediate a transition between an elementary proton and a prositron.3 Baryon Number Violation The conservation of baryon number is the basis for the incredible stability of ordinary matter. 9. Let’s also assume that the coupling has the usual Yukawa form ¯ ¯ g ψp ψe+ X + ψe+ ψp X where g is a dimensionless coupling. In fact.
The diagrams make evident that the real proton is . and the String Theory Revolution e+ e+ X X p e+ Fig. 9. Surely the event of proton decay takes place in some deﬁnite place.5 Feynman diagrams of p→ Xe+ and e+ → Xe+ If the mass MX is suﬃciently large. Temperatures higher than MX can certainly excite the proton to decay. Now let us consider what happens when a proton falls into a black hole. From the proton’s viewpoint.90 Black Holes. The complicated history of the proton is described by Feynman diagrams such as shown in Figure 9. On the other hand. The interactions cause it to make virtual transitions from the bare proton to a state with an Xboson and a positron. Thus. On the face of it. The baryon number is lost. just as it is in the real world. the proton encounters enormously high temperatures as it approaches the horizon. baryon conservation will be a very good symmetry at ordinary energies. and will not be radiated back out in the Hawking radiation. The real proton propagating through spacetime is not the simple structureless bare proton. Information. The proton will have a lifetime in excess of 1032 years. that sounds absurd. The question is: where does the baryon violation take place? One possible answer is that it occurs when the freely falling proton encounters very large curvature invariants as the singularity is approached. Who is right? The answer that black hole complementarity implies is that they are both right. there is nothing that would stimulate it to decay before that. from the vantage point of the external observer. it is diﬃcult to understand how there can be an ambiguity. So the external observer will conclude that baryon violation can take place at the horizon. the time along the proton trajectory between horizon crossing and the singularity can be enormous. For a very large black hole.6.
The quantity that we normally call baryon number is really the time averaged baryon number normalized to unity for the proton. for g ∼ 1. 9. Since the probability that the proton is actually an X. Nothing about virtual baryon nonconservation is especially surprising. This might seem paradoxical. As long as the Xboson is suﬃciently massive. and is of the order P robability ∼ g2 4π κ log MX where κ is the cutoﬀ in the ﬁeld theory. The transitions 1 take place on a time scale of order δt ∼ MX . For example. it is not unlikely that when it passes the horizon.7. Now let us consider a proton passing through a horizon. This probability is closely related to the wave function renormalization constant of the proton. its instantaneous baryon number is zero. But it is clear from Figure 9. . e+ system is not small. κ of the order of the Planck mass. the rate for real proton decay will be very small. and the proton will be stable for long times. A ﬂuctuation that is much too rapid to be seen by a low energy observer falling with the proton appears to be a real proton decay lasting to eternity from outside the horizon. Ordinary observations of the proton do not see these very rapid ﬂuctuations. What is surprising is that the probability to ﬁnd the proton in a conﬁguration with vanishing baryon number is not small. the intermediate state has vanishing baryon number. in the particular processes shown in Figure 9.6. as shown in Figure 9. since the proton is so stable. The resolution of the paradox is that the proton is continuously making extremely rapid transitions between baryon number states.7 that from the viewpoint of an external observer.6 Proton virtual ﬂuctuation into X e+ pair a superposition of states with diﬀerent baryon number.The Puzzle of Information Conservation in Black Hole Environments e+ X p p p e+ X p 91 X e+ e+ Fig. and MX of the order 1016 GeV. the probability that the proton has the “wrong” baryon number is of order unity. this is not a shortlived intermediate state.
As a proton or any other system approaches the horizon. The process of baryon violation near the horizon should not be totally surprising to the external observer. An interesting question is whether an observer falling with the proton . To him. is nothing but the usual time dilation near the horizon. At a proper distance MX from the horizon. the temperature becomes of order MX . of course. 9. Baryon violating processes are expected at that temperature.7 Proton ﬂuctuations while falling through horizon This. internal oscillations or ﬂuctuations appear to indeﬁnitely slow down.92 Black Holes. the proton is falling into a −1 region of increasing temperature. and the String Theory Revolution Time resolution becomes very dense ∆t ∆t ∆t ∆t ∆t ∆t ∆t t=0 Fig. Information. so that a shortlived virtual ﬂuctuation becomes stretched out into a real process.
Thus. But such an interaction between the proton and the probe quantum is at high enough energy that it can cause a baryon violating interaction even from the perspective of the proton’s frame of reference. Thus the observer cannot measure and report the absence of baryon violation at the horizon without himself causing it. frequencies so high that the ordinary low energy observer has no chance of detecting them. and send a message to the outside world that the proton has not decayed. . the observer has to probe it with at least one quantum with energy of order MX . to obtain a single bit of information about the state of the proton. The answer is interesting. It is evident from this example that the key to understanding the enormous discrepancies in the complementary description of events lies in understanding the ﬂuctuations of matter at very high frequencies. In order to make an observation while the proton is in a region of temperature MX the observer must do so very quickly. In the proton’s frame.The Puzzle of Information Conservation in Black Hole Environments 93 can observe the baryon number just before crossing the horizon. the time spent in the hot region before 1 crossing the horizon is MX .
94 Black Holes. Information. and the String Theory Revolution .
Let’s begin with a traditional attempt to study interactions at length scales smaller than the Planck scale. no matter how high the energy. To make matters worse.0. According to conventional relativistic quantum mechanics. Thus we see that a giant “Super Plancketron” Collider (SPC) would fail to discover fundamental length scales smaller than the Planck scale P . The typical Hawking particle has energy ∼ Rc which decreases with the S incident energy. The interesting short distance eﬀects that we want to probe will be hidden behind a horizon of radius RS ≈ 2G c2 E∆x and are inaccessible. but rather low energy Hawking radiation. the products of collision will not be high energy particles. According to conventional thinking.1) But we already know that this trend is destined to be reversed in the physics of the 21st century. head on. a size ∆x can be probed with a quantum of energy E∆x ≈ c ∆x (10. In fact. By analyzing the highest energy fragments. The problem with this thinking is that at energies far above the Planck mass. We expect to discover high energy collision products ﬂying out at all angles. particles with center of mass energy E∆x . we hope to reconstruct very short distance events.Chapter 10 Horizons and the UV/IR Connection The overriding theme of 20th century physics was the inverse relation between size and momentum/energy. as the energy 95 . what we need to do is to collide. This can be seen in many ways. the collision will create a black hole of mass ∼ E∆x .
96 Black Holes. the amount of proper time that it has before crossing the horizon tends to zero as the Schwarzschild time tends to inﬁnity. particle type.4) ∆τ ∼ 8M G δR e− 4M G . Thus in order to observer the proton before it crosses the horizon. the region over which it is localized must grow. The UV/IR connection is deeply connected to black hole complementarity.0. etc. and the String Theory Revolution increases we would be probing ever larger scales ∆x ∼ 2G E . we saw just such a phenomenon involving a charge falling onto the stretched horizon. More generally. location on the horizon through which it falls. Very high frequency is related to large size scale. Then ∆τ varies with Schwarzschild time like √ ∆t (10. The need to reconcile the complementary descriptions gives us an important clue to the behavior of matter at high frequencies. the observer who follows the proton does not see it spread at all as it falls. we have to do it in a time which is exponentially small as t → ∞. Let’s consider further: what happens to a proton falling into the black hole? The proton carries some information with it. On the other hand. In fact. the proton is falling into an increasingly hot region. The only reasonable expectation is that the constituents of the proton “melt” and diﬀuse throughout the horizon. From the viewpoint of the external observer.0. As we saw in the previous lecture.3) This is the simplest example of the ultraviolet/infrared connection that will control the relation between frequency and spatial size.2) We might express this in another form as a kind of spacetime uncertainty principle: ∆x ∆t ∼ 2G c4 ≈ 2 P (10. The proton is like a tiny piece of ice thrown into a tub of very hot water. Call the proper time ∆τ .0. its charge. . Following the proton as it falls. what we need in order to be consistent with the thermal spreading of information is now clear. in Chapter 7. momentum. Thus. Information. As the proton is observed over shorter and shorter time intervals. the enormous diﬀerences in the complementary descriptions of matter falling into a black hole are due to the very diﬀerent time resolutions available to the complementary observers. all of the information stored in the incident proton should quickly be spread over the horizon. c2 c2 (10.
but it is at the heart of complementarity. if we assume that it grows consistently with the uncertainty relation in equation 10. Fortunately it is also built into the mathematical framework of string theory. . The idea of information spreading as the resolving time goes to zero is very unfamiliar.3.Horizons and the UV/IR Connection 97 In fact.0. if the proton size depends on the time resolution according to equation 10.0. it will spread over the horizon exponentially fast. It is implicit in the modern idea of the UV/IR connection.3 (substituting ∆τ for ∆t) we obtain ∆x ∼ ∆τ 2 P ∼ 2 P MG et/4MG Thus.
98 Black Holes. Information. and the String Theory Revolution .
PART 2 Entropy Bounds and Holography .
.
it would be in lattice ﬁeld theories. Suppose we are dealing with a lattice of discrete spins. The divergence arises from modes very close to the horizon. The main indication that we have seen of this overabundance of degrees of freedom is the fact that the horizon entropydensity is inﬁnite in quantum ﬁeld theory. We begin by considering a large region of space Γ. Let the lattice spacing be a and the volume of Γ be V .Chapter 11 Entropy Bounds 11. This principle says that there are vastly fewer degrees of freedom in quantum gravity than in any QFT even if the QFT is regulated as. Let us consider some preliminary examples. For simplicity we take the region to be a sphere. This is an example of the weirdness of the Ultraviolet/Infrared connection. Now consider the space of states that describe arbitrary systems that can ﬁt into Γ such that the region outside Γ is empty space. One might think that this is just an indication that a more or less conventional ultraviolet regulator is needed to render the theory consistent. Our goal is to determine the dimensionality of that statespace. A quantitative measure of the overabundance of degrees of freedom in QFT is provided by the Holographic Principle. The number of spins is then V /a3 101 .1 Maximum Entropy Quantum ﬁeld theory has too many degrees of freedom to consistently describe a gravitational theory. The Holographic Principle is about the counting of quantum states of a system. for example. The modes which account for the divergence are very close to the horizon and would appear to be ultra short distance modes. But they also carry very small Rindler energy and therefore correspond to very long times to the external observer. But the divergent horizon entropy is not an ordinary ultraviolet phenomenon.
In other words it is the entropy given that we know nothing about the state of the system. This is a property of the system. In general entropy is not really a property of a given system but also involves one’s state of knowledge of the system. This is a very general property of conventional local systems and represents the fact that the number of independent degrees of freedom is additive in the volume. More precisely it is proportional to the number of simple degrees of freedom that it takes to describe the system. the energy within certain limits.1) A second example is a continuum quantum ﬁeld theory. We can limit the states. and the String Theory Revolution and the number of orthogonal states supported in Γ is given by Nstates = 2V /a 3 (11.2) In each case the number of distinct states is exponential in the volume V . There is another concept that we will call the maximum entropy.1. Let us now consider a system that includes gravity. It is the logarithm of the total number of states. for example. To deﬁne entropy we begin with some restrictions that express what we know. The total entropy is S = s(ρ)V The total number of states is of order Nstates ∼ exp S = exp s(ρmax )V (11. the angular momentum and whatever else we may know.3) (11.1.4) This is typical of the maximum entropy. For the spin system the maximum entropy is Smax = V log 2 a3 (11. For deﬁniteness we will take spacetime to be fourdimensional. Information. In counting the states of a system the entropy plays a central role.1.1. Again we focus on a spherical . One begins by computing the thermodynamic entropy density s as a function of the energy density ρ. for example by requiring the energy density to be no larger than some bound ρmax .102 Black Holes. In this case the states can be counted using some concepts from thermodynamics. The entropy is essentially the logarithm of the number of quantum states that satisfy the given restrictions. In this case the number of quantum states will diverge for obvious reasons. Whenever it exists it is proportional to the volume.
11. The area of the boundary is A. This is shown in Figure 11. Suppose we have a thermodynamic system with entropy S that is completely contained within Γ.1. S = A/4G.1 Inmoving (zero entropy) spherical shell of photons result of this process is a system of known entropy.Entropy Bounds 103 region of space Γ with a boundary ∂Γ. Thus we see a radical diﬀerence between the number of states of any (regulated) quantum ﬁeld theory and a theory that includes gravity. The Eg EM Γ R Fig. Now imagine collapsing a spherically symmetric lightlike shell of matter with just the right amount of energy so that together with the original mass it forms a black hole which just ﬁlls the region. But now we can use the second law of thermodynamics to tell us that the original entropy inside Γ had to be less than or equal to A/4G. In other words the area of the horizon of the black hole is A. In other words the maximum entropy of a region of space is proportional to its area measured in Planck units. . The total mass of this system can not exceed the mass of a black hole of area A or else it will be bigger than the region.
. at that time the system winds up with entropy S= A 4G . the horizon forms prior to the actual crossing of the shell.2 Spacetime depiction of radially inmoving shell of photons shell crosses. the geometry is Schwarzschild with low curvature (no black hole) prior to the shell crossing the Schwarzschild radius. and a singularity forms at a later time.104 Black Holes. Inside of the shell. However. The energy of the infalling photonic shell Eγ has been tuned such that the collapsing shell crosses the horizon exactly at the radius R in Figure 11. The horizon grows until the collapsing Light beams get more vertical as the energy becomes more concentrated Horizon is a lightlike surface R Low curvature Infalling Shell E γ Fig. Information.1. and the String Theory Revolution Spacetime depiction of horizon formation Consider the collapsing spherically symmetric shell of lightlike energy depicted in Figure 11. just outside of the shell the geometry becomes increasingly curved as the Schwarzschild radius is approached (see Figure 11. As the photonic shell approaches the center. However.1. 11.2).
We will see that it is most natural to deﬁne holographic entropy bounds on lightlike surfaces[3] as opposed to spacelike surfaces. R ∼ 1051 cm. considerable larger than the observable universe 1028 cm.2 Entropy on Lightlike Surfaces So far we have considered the entropy that passes through spacelike surfaces. consider thermal radiation at a temperature of 1000◦K. 4G (11. Compare this with the maximum entropy calculated using the holographic limit Smax ≈ R lP lanck 2 3 ∼ R(cm) ⊗ 1033 = 2 . the entropy of any system is limited by Smax ≤ A . Under . Entropy is proportional to the number of photons. which corresponds to photons of wavelength ∼10−5 cm.1.6) Evidently the maximum entropy will only be saturated for the photon gas when the radius is huge. unless the second law of thermodynamics is untrue. so this “holographic” limit must be satisﬁed.Entropy Bounds 105 Therefore. (11. Aside: Scale of Entropy Limit Example: To get some idea of how big a typical system must be in order to saturate the maximum entropy. 11. The number of photons Nγ in a volume of radius R satis3 V ﬁes Nγ ∼ λ3 ∼ R(cm) ⊗ 105 .1.5) The coarse grained volume in phase space cannot decrease. and thus one expects S ∝ R(cm) ⊗ 105 .
106
Black Holes, Information, and the String Theory Revolution
certain circumstances the entropy bounds of lightlike surfaces can be translated to spacelike surfaces, but not always. The case described above is one of those cases where a spacelike bound is derivable. Let us start with an example in asymptotically ﬂat spacetime. We assume that ﬂat Minkowski coordinates X + , X − , xi can be deﬁned at asymptotic distances. In this chapter we will revert to the usual convention in which X + is used as a light cone time variable. We will now deﬁne a “light+ sheet”. Consider the set of all light rays which lie in the surface X + = X0 in the limit X − → +∞. In ordinary ﬂat space this congruence of rays deﬁnes a ﬂat threedimensional lightlike surface. In general, they deﬁne a lightlike surface called a light sheet. The light sheet will typically have singular caustic lines, but can be deﬁned in a unique way[4]. When we vary + X0 the light sheets ﬁll all spacetime except for those points that lie behind black hole horizons. Now consider a spacetime point p. We will assign it light cone coor+ dinates as follows. If it lies on the light sheet X0 we assign it the value + + X = X0 . Also if it lies on the light ray which asymptotically has transverse coordinate xi we assign it xi = xi . The value of X − that we assign 0 0 will not matter. The twodimensional xi plane is called the Screen. Next + assume a black hole passes through the light sheet X0 . The stretched
X+
Incoming Parallel Light Rays
XFar off screen x1 and x 2 plane
xj
Fig. 11.3
Light propagating on lightlike surface X + = constant
Entropy Bounds
107
Asymptotic screen
X
+
Image of Black hole on screen

X
stic Cau
xj
Foliation corresponds to different + x slices Black Hole
Fig. 11.4
Family of light rays on ﬁxed X + surface in presence of black hole
horizon of the black hole describes a twodimensional surface in the threedimensional light sheet as shown in Figure 11.4. Each point on the stretched horizon has unique coordinates X + , xi , as seen in Figure 11.5. More generally if there are several black holes passing through the light sheet we can map each of their stretched horizons to the screen in a single valued manner. Since the entropy of the black hole is equal to 1/4G times the area of the horizon we can deﬁne an entropy density of 1/4G on the stretched horizon. The mapping to the screen then deﬁnes an entropy density in the xi plane, σ(x). It is a remarkable fact that σ(x) is always less than or equal to 1/4G. 1 To prove that σ(x) ≤ 4G we make use of the focusing theorem of general relativity. The focusing theorem depends on the positivity of energy and is based on the tendency for light to bend around regions of nonzero energy. Consider a bundle of light rays with cross sectional area α. The light rays
108
Black Holes, Information, and the String Theory Revolution
Patch on stretched horizon
Map of patch on stretched horizon
Black hole
Stretched horizon
Map of stretched horizon
Fig. 11.5
Image of “stretched horizon” on asymptotic screen
are parameterized by an aﬃne parameter λ. The focusing theorem says that d2 α ≤0 dλ2 (11.2.7)
Consider a bundle of light rays in the light sheet which begin on the stretched horizon and go oﬀ to X − = ∞. Since the light rays deﬁning the light sheet are parallel in the asymptotic region dα/dλ → 0. The focusing theorem tells us that as we work back toward the horizon, the area of the bundle decreases. It follows that the image of a patch of horizon on the screen is larger than the patch itself. The holographic bound immediately follows. σ(x) ≤ 1 4G (11.2.8)
This is a surprising conclusion. No matter how we distribute the black holes in threedimensional space, the image of the entropy on the screen always satisﬁes the entropy bound equation 11.2.8. An example which helps clarify how this happens involves two black holes. Suppose we try to hide one of them behind the other along the X − axis, thus doubling the entropy density in the x plane. The bending and focusing of light always acts as in Figures 11.6 to prevent σ(x) from exceeding the bound. These considerations lead us to the more general conjecture that for any system,
We note for future use that the conclusions concerning the entropy bound would be unchanged if we allowed screens for which the light rays were diverging as we move outward. da/dλ > 0. i. rapid motions on the screen Small motions near caustic CAUSTIC CAUSTIC Fig.e.8.Entropy Bounds 109 Motion of second black hole CAUSTIC CAUSTIC Image of second black hole Motion of second black hole Image of second black hole Fig. if we attempt to use screens for which the light rays are converging then the argument fails. demonstrated in Figures 11.7.2.7 Initial and later path and image during “slow” motion near caustic . Thus far we have assumed asymptotically ﬂat boundary conditions. However. One sees that due to gravitational lensing. Examine the projection of those motion upon the screen.6 Initial and later motions and images of second black hole when it is mapped to the screen the entropy density obeys the bound in equation 11. Equivalently da/dλ equals zero at the screen. 11. 11. Aside: Apparent motions Consider a single point particle external to the black hole undergoing motions near a caustic. This will play an important role in generalizing the holographic bound to more general geometries. This allowed us to choose the screen so that the light rays forming the light sheet intersect the screen at right angles. the image of the particle can move at arbitrarily large speeds! produce large.
will play the same role as the screen in the previous discussion. The metric has the form dτ 2 = dt2 − a(t)2 dxm dxm (11. Information. Later. geometries. The function a(t) is assumed to grow as a power of t. The light sheet is now deﬁned by the backward light cone formed by light rays that propagate from ∂Γ into the past (See Figure 11. At time t we consider a spherical region Γ of volume V and area A.W. The boundary (d − 1)sphere. ∂Γ. and the String Theory Revolution 11. we will relax these assumptions.3.10) Let’s also make the usual simplifying cosmological assumptions of homogeneity.8 Holographic surface for calculating entropy bound with a spherical surface as the screen . a(t) = a0 tp (11. In particular we assume that the spatial entropy density (per unit d volume) is homogeneous.3 Friedman Robertson Walker Geometry The holographic bound can be generalized to ﬂat F.110 Black Holes.R.3.W.R. Consider the general case of d + 1 dimensions.8). First we will review the F. where it is called the Fischler–Susskind (FS) bound[5] and to more general geometries by Bousso[6]. t Screen Past light cone Projection from screen onto light cone t>0 Fig. case. 11.9) where the index m runs over the d spatial directions.
This bound states that the total entropy passing through the light sheet does not exceed A/4G. However if Γ is larger than the particle horizon at time t the light sheet is not a cone.9. We observe that at the screen the area of the outgoing bundle of light rays is increasing as we go to later times. it will scale like the area A. 11. We will return to the question of spacelike bounds after discussing Bousso’s generalization[6] of the FS bound. but rather a truncated cone as in Figure 11.Entropy Bounds 111 As in the previous case the holographic bound applies to the entropy passing through the light sheet. t Sphere Past light cone Entropy flux through light cone bounded Big Bang t=0 Entropy not bounded Fig. It is believed that the bound is very general. the entropy will exceed the area. Thus a portion of the entropy present at time t never passed through the light sheet. It is now easy to see why we concentrate on light sheets instead of spacelike surfaces. the resulting density satisﬁes the holographic bound. Obviously if the spatial entropy density is uniform and we choose Γ big enough. The key to a proof is again the focusing theorem.9 Region of space causally connected to particle horizon . which is cut oﬀ by the big bang at t = 0. The focusing theorem again tells us that if we map the entropy of black holes passing through the light sheet to the screen. In other words the light sheet has positive expansion into the future and negative expansion into the past. If we only count that portion of the entropy which did pass through the light sheet.
Nγ ≈ 1090 .R.W. and d represent the number of spatial dimensions. which gives the form for the time dependence .10).112 Black Holes.11) An outgoing light ray (null geodesic) which would generate the particle horizon satisﬁes dt = a(t)dx. and the String Theory Revolution Test: Is the observed horizon entropy bounded by its area? We will check whether the observed particle horizon satisﬁes the entropy bound: SHorizon ≤ AHorizon 4G ??? We can check by recognizing that the entropy is primarily given by the number of black body cosmic background photons. we will determine if the entropy per horizonarea contained within the particle horizon is increasing or decreasing. Information. We want to check whether d S ∼ σ RHorizon <? (a(t) RHorizon ) 4G d−1 (11. The proper size of the horizon is approximately given by 1018 (light) seconds. using the F.3. so that d S ∼ σ RHorizon d j=1 dτ 2 = dt2 − a2 (t) dxj dxj This means that the proper size of the horizon is given by a(t)R Horizon . Let R Horizon represent the coordinate size of the particle horizon (Figure 11. This gives a proper size for the horizon of about 1061 Planck units. and the Planck time is approximately 10−43 seconds. We can use these numbers to compare the cosmic entropy with the area of the horizon: SHorizon ≤??? 1090 ≤ 1061 2 AHorizon 4G = 10122 We ﬁnd this inequality to deﬁnitely be true today. Next. Let σ be the entropy volume density. geometry.
one can use the Einstein . Given w and the scale factor for the expansion p.3. d (11. ao (11.14) One sees that if the expansion rate is too slow.13) Therefore for the entropy bound to continue to be valid. cosmology satisﬁes the equation of state P = wu (11.12) If we assume the form a(t) = ao tp then the particle horizon evolves according to the formula RHorizon (t) = t1−p .3.Entropy Bounds 113 t Past light cone RHorizon Entropy flux Big Bang t=0 Fig.15) where P is the pressure and u is the energy density and w is a constant.10 Particle Horizon of the size of the particle horizon: t RHorizon (t) = 0 dt a(t ) (11.W.R. the time dependence must satisfy t(1−p)d < td−1 .3. 11. Suppose that the matter in the F. and the entropy bound will eventually be contradicted.3. This bounds the expansion rate coeﬃcient to satisfy p > 1 . then the coordinate volume will grow faster than the area.
3. the bound will always be satisﬁed. Using a decoupling time tdecoupling ∼ 105 years (when the background radiation fell out of equilibrium with the matter) and extrapolating back using the previously calculated entropy relative to the bound. 11. Information.3.18) A 4G is reached when tdecoupling ∼ 10−44 sec 1056 (11.3.16) We see that the number of spatial dimensions cancels and that the entropy bound is satisﬁed so long as w ≤ 1.114 Black Holes. At any point on the boundary we can construct four light rays that are perpendicular to the boundary[6]. since the speed of sound is always less than the speed of light. Therefore the entropy bound is not exceeded after the Planck time. (11. The relation satisﬁed by 2 the scale factor vs = w ≤ 1 is just the usual causality requirement. Recall that the speed of sound within a medium is given by 2 vs = ∂P ∂u =w Therefore. in the future. Next.3. As one moves forward in time.19) = 1028 ⇒ t = This time is comparable to the Planck time (by coincidence??). not less. one gets S = 10−28 A/(4G) The entropy bound S = tdecoupling t 1 2 tdecoupling t 1 2 (11.17) This is an interesting result. the entropy bound then becomes more satisﬁed. go back in time using the black body radiation background as the dominant entropy. We will . Take a spacelike region Γ bounded by the spacelike boundary ∂Γ.4 Bousso’s Generalization Consider an arbitrary cosmology. and the String Theory Revolution ﬁeld equation to calculate a relationship between them: p= 1 2 > d (1 + w) d (11.
For example in a rapidly contracting universe both future branches have negative expansion while the past branches have positive expansion. On any of these branches a light ray. Similarly two branches go to the past. Thus we draw light surfaces into the past. In general as we move around in the Penrose diagram the particular branches which have negative expansion may change. Thus in Figure 11.12 we draw the Penrose–Bousso (PB) diagram for a positive curvature. It consists of four distinct regions. One of them is composed of outgoing rays and the other is ingoing.12 the expansion of the universe causes both past branches to have negative expansion. These light sheets may terminate at the tip of a cone or a caustic or even a boundary of the geometry. For example in Figure 11. Bousso introduced a notation to indicate this.Entropy Bounds 115 call these the four branches. The Penrose diagram is divided into a number of regions depending on which branches have negative expansion.11 we illustrate ﬂat spacetime and the negative expansion branches of a typical local 2sphere. The four branches at a given point on the Penrose diagram are represented by a pair of 45 degree lines passing through that point. For simplicity we restrict attention to those regions for which a given branch has a unique sign. For example if the cosmology initially expands and then collapses. If we consider general boundaries the sign of the expansion of a given branch may vary as we move over the surface. matter dominated universe that begins with a bang and ends with a crunch. In each region the negative expansion branches are indicated by their directions at a typical point. Each point of such a diagram really stands for a 2sphere (more generally a (d − 1)sphere). the outgoing future branch will go from positive to negative expansion. However in nonstatic universes other combinations are possible. Two branches go toward the future. To help visualize how Bousso’s construction works we will consider spherically symmetric geometries and use Penrose diagrams to describe them. if ∂Γ is convex the outgoing (ingoing) rays have positive (negative) expansion. However we are only interested in the branches of negative expansion. In ordinary ﬂat spacetime. together with its neighbors deﬁne a positive or negative expansion as we move away from the boundary. In Region I of Figure 11. Bousso’s bound states that the entropy passing through these light sheets is less than A/4G where A is the boundary of ∂Γ. We can now state Bousso’s rule: From the boundary ∂Γ construct all light sheets which have negative expansion as we move away. These light surfaces terminate on the initial boundary of the . The Penrose diagram represents the radial and time directions.
one condition for a spacelike bound is that the entropy is bounded by a corresponding future light sheet. By the second law of thermodynamics the entropy on the spacelike surface can not exceed the entropy on the future light sheet. For Region I there is no future bound. Bousso’s rule tells us nothing in this case about the entropy on spacelike surfaces bounded by p. The reason is that not all matter . case.W.116 Black Holes. It is evident that any matter which passes through the spacelike surface must also pass through the future light sheet. and the String Theory Revolution Fig. Nevertheless there is no spacelike bound.R. Now consider a spacelike surface bounded by q and extending to the spatial origin (shown in Figure 11. Now move on to Region II. There is something new in this case. Thus in this case the entropy in a spacelike region can be holographically bounded. 11. With this in mind we return to Region I. These are untruncated cones and the entropy on both of them is holographically bounded. Therefore. Information. In Region III the entropy bounds are both on future light sheets. The relevant light sheets in this region begin on the 2sphere q and both terminate at the spatial origin. We ﬁnd that the entropy is bounded on a future light sheet.11 Negative expansion branches of 2sphere in ﬂat spacetime geometry and are similar to the truncated cones that we discussed in the ﬂat F. The holographic bound in this case says that the entropy passing through either backward light surface is bounded by the area of the 2sphere at point p. and therefore the entropy is not bounded on spacelike regions with boundary p.13).
12 Penrose–Bousso diagram for matter dominated universe which pass through spacelike surfaces are forced to pass through the future light sheets. 11. Region IV is identical to Region II with the spatial origin being replaced by the diametrically opposed antipode.14 demonstrates a region that does not satisfy an entropy bound for this cosmology.) Figure 11.13. . Thus we see that there are lightlike bounds in all four regions but only in II and IV are there holographic bounds on spacelike regions. (See Figure 11.Entropy Bounds 117 t = tfinal Decreasing area Future directed CONTRACTION III Forward light cone TOWARDS r = 0 Backwards lightcone q Point corresponds to finite area of sphere TOWARDS LARGE RADIUS IV Backwards light cone II I p t = 0 EXPANSION Light cone intercepts singularity Fig.
13 Bounding surfaces for inﬂowing entropy t final Area is INCREASING in part of this lightlike surface. 11. 11.14 Lightlike surface which does not satisfy entropy bound .118 Black Holes. and the String Theory Revolution Entropy on this space like surface is also bounded by flux through lightlike surface q Fig. Information. The entropy thru such a surface is NOT bounded in a Robertson–Walker cosmology t = 0 Fig.
5.23) For our purposes we want to put this in a form that will allow us to read oﬀ the Penrose diagram. More recently de Sitter space has entered cosmology as the likely candidate for the ﬁnal fate of the universe. The model of cosmology which is presently gaining the status of a “standard model” involves de Sitter space in two ways. The inﬂationary theory is more than two decades old and by now has been well tested. It is the hyperboloid given by 4 i=1 (xi )2 − (x0 )2 = R2 . (11.25) . R2 = 3 Gλ (11. The history of the universe seems to be a transition from an early de Sitter epoch in which the vacuum energy or cosmological constant was very large to a late de Sitter phase characterized by a very small but not zero cosmological constant.21) de Sitter space can also be written in the form dτ 2 = dt2 − a(t)2 dΩ2 3 (11. and the scale factor a is given 3 by a(t) = R cosh(t/R). Fourdimensional de Sitter space may be deﬁned by embedding it in (4 + 1) dimensional ﬂat Minkowski space.5. λ.20) The radius of curvature R is related to the cosmological constant. (11.5. To that end deﬁne the conformal time T by dT = dt/a(t).5.24) (11.5 de Sitter Cosmology de Sitter space holds special interest because of its close connection with observational cosmology.22) where dΩ2 is the metric of a unit 3sphere. First it is believed that the early universe underwent an epoch of rapid inﬂation during which the spacetime geometry was very close to de Sitter.5. de Sitter space is the solution of Einstein’s ﬁeld equations with a positive cosmological constant that exhibits maximal symmetry. 3 2 (11. One easily ﬁnds that the geometry has the form dτ 2 = (a(t)) (dT 2 − dΩ2 ).Entropy Bounds 119 11.5.
120 Black Holes. Information. The reason is that the geometry rapidly expands as we move toward the upper and lower boundary of the de Sitter space. case. and the String Theory Revolution 0 −π Fig.15. 11. the conformal time (T = tan−1 (tanh(t/R)) − π ) varies from −π to 0.15 0 π Penrose diagram for de Sitter spacetime S Fig. We leave it as an exercise to show that the entropy is .R.W. However. Since the polar angle on 2 the sphere also varies from 0 to π the Penrose diagram is a square as in Figure 11. 11.16 Extremal spacelike surface in de Sitter spacetime Furthermore when t varies between ±∞. Of particular interest is the bound on a spacelike surface beginning at the center of the Penrose diagram and extending to the left edge as in Figure 11.16. Once again the Bousso construction divides the Penrose diagram into four quadrants. the contracting light sheets in the upper and lower quadrants are oriented oppositely to the usual F.
The Penrose–Bousso diagram for de Sitter space is shown in Figure 11. Inﬂationary cosmology is illustrated in Figure 11. 4G (11. The dotted line where the two geometries are joined is the reheating surface where the entropy of the universe is created. In fact the maximum entropy on any such surface is given by S= 4πR2 .17. But then at some time the vacuum energy began to decrease and the universe made a transition to an F. According to the inﬂationary hypothesis. In order to describe inﬂationary cosmology we must terminate the de Sitter space at some late time and attach it to a conventional F.5.Entropy Bounds 121 t Postinflation particle horizon Larger area Period of inflation Fig.W. 11. In order to construct the Penrose–Bousso diagram we begin by drawing Penrose diagrams for both de Sitter space and either radiation or matter dominated F. universe.R. The transition was accompanied by the production of a large amount of entropy. space as in Figure 11.17 Inﬂationary universe bounded by the area of the 2sphere at the center of the diagram.W.19. the universe began with a large vacuum energy which mimicked the eﬀects of a positive cosmological constant.26) de Sitter space has a special importance because of its role during the early evolution of the universe. and is called reheating.R.W.18. During that period the geometry was de Sitter space. .R.
19 Joining of inﬂationary and postinﬂationary geometries Let us focus on the point p in Figure 11. 11. 11. .18 Buosso wedges for de Sitter geometry de Sitter Space Robertson–Walker Space + Join t inflation EXPANDING Reheating: Entropy is created during reheating.20. Fig. However Bousso’s rule applied to point p only bounds the entropy on the past light sheet. It is easy to see that in an ordinary inﬂationary cosmology p can be chosen so that the entropy on the spacelike surface p − q is bigger than the area of p. and the String Theory Revolution Point corresponds to finite spherical area Fig.122 Black Holes. There is no entropy prior. Information.
but for our present purpose all we need are the properties of its PB diagram. . 11.Entropy Bounds 123 Entropy flux only after reheating p q Robertson– Walker No entropy flux here REHEATING Inflation Backwards light cone Fig.20 Entropy bound on spherical region causally connected to inﬂationary period In this case most of the newly formed entropy on the reheating surface is not counted since it never passed through the past light sheet. 11.6 Anti de Sitter Space de Sitter space is important because it may describe the early and late time behavior of the real universe. It is the background in which the holographic principle is best understood. Anti de Sitter space is important for an entirely diﬀerent reason. Typical inﬂationary cosmologies can be studied to see that the past light sheet bound is not violated. In Chapter 12 the geometry of AdS will be reviewed.
Let us consider a static surface of large area A far from the spatial origin. The surface is denoted by the dotted vertical line L in Figure 11. Consider an arbitrary point p on L. The PB diagram consists of an inﬁnite strip bounded on the left by the spatial origin and on the right by a boundary. The spacetime in appropriate coordinates is static but unlike de Sitter space the static coordinates cover the entire space. Therefore the entropy on any spacelike surface bounded by p and including the origin is also holographically bounded. The PB diagram consists of a single region in which both negative expansion light sheets point toward the origin.21. These two properties imply a very far .124 Black Holes. and the String Theory Revolution L Boundary r=0 S p ANY spacelike surface S will have its entropy bounded by the area Fig. Evidently Bousso’s rules bound the entropy on past and future light sheets bounded by p. Information. In other words the entire region to the left of L can be foliated with spacelike surfaces such that the maximum entropy on each surface is A/4G. AdS space is an example of a special class of geometries which have timelike Killing vectors and which can be foliated by spacelike surfaces that satisfy the Holographic bound. We will think of L as an infrared cutoﬀ.21 Static surface of large area in AdS space AdS space is the vacuum of theories with a negative cosmological constant. The vacuum of AdS is a genuine zero temperature state. 11.
.27) The holographic description of AdS space is the subject of the next chapter.6. All physics taking place in such backgrounds (in the interior of the infrared cutoﬀ L) must be described in terms of a Hamiltonian that acts in a Hilbert space of dimensionality Nstates = exp(A/4G) (11.Entropy Bounds 125 reaching conclusion.
and the String Theory Revolution .126 Black Holes. Information.
Let us consider a region of ﬂat space Γ.Chapter 12 The Holographic Principle and Anti de Sitter Space 12. the number of possible quantum states in a region of ﬂat space is bounded by the exponential of the area of the region in Planck units. 127 . Nevertheless we must be able to describe microscopic processes taking place anywhere in the region. the maximum entropy is a measure of the number of simple degrees of freedom∗ that it takes to completely describe the region. For a large macroscopic region this is an enormous reduction in the required degrees of freedom. In fact if the linear dimensions of the system is of order L then the number of degrees of freedom per unit volume scales like 1/L in Planck units. A simple degree of freedom represents a single bit of information. One of the elements of this paradigm is the Holographic Principle and its embodiment in AdS space. The entropy bound tells us that the maximum number of nonredundant degrees of freedom is proportional to the area. One way to think of this is to imagine the degrees of freedom of Γ as living on ∂Γ with an area density of no more than ∼ 1 degree of freedom per ∗ By a simple degree of freedom we mean something like a spin or the presence or absence of a fermion. This is almost always proportional to the volume of Γ. measured in Planck units.1 The Holographic Principle As we have seen in Chapter 11. The exception is gravitational systems. Typically. By making L large enough we can make the degrees of freedom arbitrarily sparse in space. We have seen that the maximum entropy of all physical systems that can ﬁt in Γ is proportional to the area of the boundary ∂Γ. time and locality. as in the case of a lattice of spins. That fact together with the Ultraviolet/Infrared connection and Black Hole Complementarity has led physics to an entirely new paradigm about the nature of space.
128 Black Holes. We will begin with a brief review of AdS geometry. threedimensional space described by a twodimensional hologram at its boundary! That this is possible is called the Holographic Principle. it does not make sense to consider a balllike region with a boundary in the general theory of relativity. Even better would be to construct a description of the region in terms of a boundary theory with a limited density of degrees of freedom. Why the S(5)? The reason involves the high degree of supersymmetry enjoyed by superstring theory. the 5dimensional AdS and a 5sphere S(5). Information. The particular space that we will be interested in is not simple 5dimensional AdS but rather AdS(5)⊗S(5)[7][8][9]. But by bending some of the directions of space into compact manifolds it becomes possible to generate a cosmological constant for the resulting lower dimensional Kaluza–Klein type theory. Ordinarily. What we would ideally like to do is to have a solution of Einstein’s equations that describes a ball of space with a spherical boundary and then to count the number of degrees of freedom. The remarkable precision is due to the unusually high degree of symmetry of the theory which includes a powerful version of supersymmetry. In this lecture we will describe a very precise version of AdS “Holography” which grew out of the mathematics of string theory. From a conceptual point of view the extra internal 5sphere is not important. Generally the kind of supergravity theories that emerge from string theory don’t have cosmological constants. However we will downplay the mathematical aspects of the theory and concentrate on those physical principles which are likely to be general. But there is one special situation which is naturally balllike. Anti de Sitter space.2 AdS Space We saw in Chapter 11 that AdS space enjoys certain properties which make it a natural candidate for a holographic Hamiltonian description. It occurs when there is a negative cosmological constant. For our purposes 5dimensional AdS space may be considered to be a solid 4dimensional . Thus AdS is a natural framework in which to study the Holographic Principle. From a mathematical point of view it is essential if we want to be able to make precision statements. The analogy with a hologram is obvious. and the String Theory Revolution Planck area. This is a 10dimensional product space consisting of two factors. 12.
It has a horizon at y = ∞. For all practical purposes AdS space behaves . dτ 2 = (12. The metric we will use is dτ 2 = R2 (1 + r2 )2 dt2 − 4dr2 − 4r2 dΩ2 (1 − r2 )2 R2 dt2 − dxi dxi − dy 2 y2 (12.2 may be expressed in terms of the coordinate z = 1/y. Ω where t is time.2.2.1 and 12.The Holographic Principle and Anti de Sitter Space 129 spatial ball times the inﬁnite time axis.2.1) There is another form of the metric which is in common use. To construct the space AdS(5) ⊗ S(5) all we have to do is deﬁne 5 more coordinates ω5 describing the unit 5sphere and add a term to the metric 2 ds2 = R2 dω5 5 (12.2 is not geodesically complete. A timelike geodesic can get to y = ∞ in a ﬁnite proper time so that the space in equation 12. The second way that equations 12.2 are not the same. The geometry has uniform curvature R−2 where R is the radius of curvature. When interpreted in this manner.2) where i runs from 1 to 3. dτ 2 = R2 z 2 (dt2 − dxi dxi ) − 1 2 dz z2 (12.3) In this form it is clear that there is a horizon at z = 0 since the time–time component of the metric vanishes there.1 in the vicinity of a point on the boundary at r = 1. it takes a total amount of (dimensionless) time t = π for light to make a round trip from the origin at r = 0 to the boundary at r = 1 and back.2 is the exact metric of an incomplete patch of AdS space. The boundary is at z = ∞.2.2.2. The metric in equation 12. time coordinates appearing in 12.2. The 3sphere is replaced by the ﬂat tangent plane parameterized by xi and the radial coordinate is replaced by y. light can travel to the boundary and back in a ﬁnite time. For example. First of all it is an approximation to equation 12. The geometry can be described by dimensionless coordinates t.2. The metric 12. r is the radial coordinate (0 ≤ r < 1) and Ω parameterizes the unit 3sphere.2.2. with y = (1 − r).1 and 12.2 are related is that 12.1 in two diﬀerent ways.2.2 is related to 12. r.2.4) Although the boundary of AdS is an inﬁnite proper distance from any point in the interior of the ball.2.2.
The size of the cavity is of order R. Supplement on Properties of AdS metric 1) The point r = 0 is the center of the Anti de Sitter space and r varies from 2 0 to 1 in the space. From equation 12. let us compute the area of the boundary. which means that light rays move at 45◦ angles near the boundary. e 12. 3) Near r = 1. the spatial metric is that of a uniformly (negatively) curved space. a hyperbolic plane (or the Poincar´ disk ).130 Black Holes.2. A unit coordinate time interval corresponds to increasingly large proper time intervals. According to the Holographic Principle we should be able to describe everything in the bulk by a theory whose degrees of freedom can be identiﬁed with the boundary at y = 0. However the Holographic Principle requires more than that. To see what this entails. . It requires that the boundary theory has no more than 1 degree of freedom per Planck area. Information. In what follows we will think of the cavity size R as being much larger than any microscopic scale such as the Planck or string scale. the metric is approximately conformal. A radial null geodesic satisﬁes 1 + r2 dt2 = 4dr2 . and the String Theory Revolution like a ﬁnite cavity with reﬂecting walls.1 we see that the metric diverges near the boundary. 4) Generally. which makes the space causally ﬁnite.3 Holography in AdS Space We will refer to AdS(5) ⊗ S(5) as the bulk space and the 4dimensional boundary of AdS at y = 0 as the boundary. ds2 ∼ = 4R2 (1−r 2 )2 −dt2 + dr2 + r2 dΩ2 D−2 Light rays move slower (by a factor of 2) near the center of the Anti de Sitter space. 2) The metric is singular at r = 1 in all components. which means that a light beam will traverse the inﬁnite proper distance in r from 0 to 1 back to 0 in a round trip time given by π.
2. For example the rotations of the sphere Ω are symmetries.The Holographic Principle and Anti de Sitter Space 131 Later we will regulate this divergence by moving in a little way from y = 0. This symmetry must also be realized by the holographic theory.2. In addition there is a “dilatation” symmetry t → λt xi → λxi y → λy (12. AdS(5) ⊗ S(5) is a solution of the 10dimensional supergravity that describes low energy superstring theory. xi . The full symmetry group of AdS(5) is the group O(42). In fact the full AdS symmetry group. Let us consider the metric in the form 12. and that is in fact the case. All of this leads us to the remarkable conclusion that quantum gravity in AdS(5) ⊗ S(5) should be exactly described by an appropriate . The dilatation symmetry also acts as a simple dilatation of the coordinates t. Another important fact involves the symmetry of AdS.1 we can see additional symmetry. when acting on the boundary at y = 0 is precisely the conformal group of 4dimensional Minkowski space. In addition there is also the symmetry O(6) associated with rotations of the internal 5sphere. and so it is.3. the 4dimensional Poincar´ symmetry e acts on the boundary straightforwardly. That suggests that the boundary theory might be a quantum ﬁeld theory. All of the transformations act on the boundary as conformal transformations which preserve lightlike directions on the boundary. The additional symmetry is the socalled N = 4 supersymmetry. As we mentioned. x. Since our goal is a holographic boundary description of the physics in the bulk spacetime it is very relevant to ask how the symmetries act on the boundary of AdS. Obviously. The implication of this symmetry of the boundary is that the holographic boundary theory must be invariant under the conformal group. but for the time being we can assume that the number of degrees of freedom per unit coordinate area is inﬁnite.2. It is obvious that the geometry is invariant under ordinary Poincar´ transformations of the 4dimensional Minkowski e coordinates t. Indeed the space has more symmetry than just the conformal group and the O(6) symmetry of the internal 5sphere. This together with the fact that the boundary has an inﬁnite (coordinate) density of degrees of freedom suggests that the holographic theory is a Conformal Quantum Field Theory.5) On the other hand if we consider the representation of AdS in 12.
A conventional cutoﬀ would involve a microscopic length scale such as the 10dimensional Planck length lp . To do so we introduce a surface L at r = 1 − δ. One way to do this would be to introduce a spatial lattice in 9dimensional space. In order to count degrees of freedom we also need to regulate the area of the boundary of AdS which is inﬁnite.8) where A is the 8dimensional area of the boundary L. and is seen to be critically divergent.2. It is not generally possible to make a regular lattice. To say it slightly diﬀerently. δ3 (12. We can then deﬁne a simple theory such as a Hamiltonian lattice theory on the space.2.6) The number of bulk lattice sites and therefore the number of degrees of freedom is 1 R9 V ∼ 3 9 9 lp δ lp (12. In order to have a benchmark for the counting of degrees of freedom in AdS(5) ⊗ S(5) imagine constructing a cutoﬀ ﬁeld theory in the bulk. However the Holographic Principle suggests that this entropy is overestimated.132 Black Holes.7) In such a theory we also will ﬁnd that the maximum entropy is of the same order of magnitude.9) In other words when R/lp becomes large the holographic description requires a reduction in the number of independent degrees of freedom by a factor lp /R.3.3. The holographic bound discussed in Chapter 11 requires the maximum entropy and the number of degrees of freedom to be of order Smax ∼ A 8 lp (12. Information. V (δ) ∼ R9 .3. This is also easily computed. The total 9dimensional spatial volume in the interior of L is easily computed using the metric 12. We ﬁnd Smax ∼ 1 R8 8 δ 3 lp (12.3. and the String Theory Revolution superconformal Lorentz invariant quantum ﬁeld theory associated with the AdS boundary. the Holographic Principle implies a complete description of all physics in the bulk of a very large AdS . but a random lattice with an average spacing lp is possible.
SYM theory on the boundary of the AdS space[7][8][9]. To see this let us Weyl rescale the metric by a . length and temperature carry their usual dimensions. and N is the dimension of the Yang–Mills gauge theory. To convert from SYM to bulk variables. Ω or t. In fact we will use the convention that all SYM quantities are dimensionless. On the other hand the bulk gravity theory quantities such as mass. Thus if ESYM and M represent the energy in the SYM and bulk theories ESYM = M R. The correspondence states that there is a complete equivalence between superstring theory in the bulk of AdS(5) ⊗ S(5) and N = 4. In these lectures SYM theory will always refer to this particular version of supersymmetric gauge theory. But there is an important sense in which it is 3 + 1 dimensional. 3 + 1dimensional. 12. It is well known that SYM is conformally invariant and therefore has no dimensional parameters. In fact there is only one known class of systems with the appropriate N = 4 supersymmetry. N represents the number of supersymmetries. Nonetheless the theory must be able to describe microscopic events in the bulk even when R becomes extremely large.4 The AdS/CFT Correspondence The search for a holographic description of AdS(5) ⊗ S(5) is considerably narrowed by the symmetries. x. The corresponding momenta are also dimensionless. We will only review some of the salient features. The correspondence between gravity or its string theoretic generalization in AdS(5) ⊗ S(5) and Super Yang–Mills (SYM) theory on the boundary is the subject of a vast literature.The Holographic Principle and Anti de Sitter Space 133 space in terms of only lp /R degrees of freedom per spatial Planck volume. SU (N ). There is one question that may be puzzling to the reader. the conversion factor is R. It will be convenient to deﬁne the theory to live on the boundary parametrized by the dimensionless coordinates t. Similarly bulk time intervals are given by multiplying the t interval by R. Since AdS(5) ⊗ S(5) is a 10dimensional spacetime one might think that its boundary is (8 + 1) dimensional. the SU (N ) Supersymmetric Yang–Mills (SYM) theories.
Let us return to the correspondence between the bulk and boundary theories. the gauge theory also has two constants.12) We can also write the 10dimensional Newton constant in the form G = R8 /N 2 (12.4.4. The 10dimensional bulk theory has two dimensionless parameters. R s 2 = (N gym ) 4 1 2 g = gym (12.11) On the other side of the correspondence.134 2 Black Holes.13) There are two distinct limits that are especially interesting. The string coupling constant and length scale are related to the 10dimensional Planck length and Newton constant by 8 lp = g 2 8 s =G (12. The AdS/CFT correspondence has been widely studied as a tool for learning about the behavior of gauge theories in the . depending on one’s motivation. Information. and the String Theory Revolution R factor (1−r2 )2 so that the rescaled metric at the boundary is ﬁnite.4. The new metric is 2 dS 2 = (1 + r2 )2 dt2 − 4dr2 − 4r2 dΩ2 + (1 − r2 )2 dω5 (12.4. These are: 1) The radius of curvature of the AdS space measured in string units R/ s . They are 1) The rank of the gauge group N 2) The gauge coupling gym Obviously the two bulk parameters R and g must be determined by N and gym . Alternately we could measure R in 10dimensional Planck units. In these lectures we will assume the relation that was originally derived in [7].10) Notice that the size of the 5sphere shrinks to zero as the boundary at r = 1 is approached. The relation between string and Planck lengths is given by g2 8 s 8 = lp 2) The dimensionless string coupling constant g. The boundary of the geometry is therefore 3+1 dimensional.
4. This is consistent with the fact that a continuum quantum ﬁeld theory such as SYM has an inﬁnity of modes in any ﬁnite threedimensional patch. This limit is dominated by classical supergravity theory.14) (12.3.16) gym = constant N →∞ (12. In order to do a more reﬁned counting[9] . We will be interested in the behavior of the theory as the AdS radius increases but with the parameters that govern the microscopic physics in the bulk kept ﬁxed. Thus we expect that the number of degrees of freedom associated with such a patch should diverge.5 The Infrared Ultraviolet Connection In either of the metrics in equation 12. This means we want the limit g = constant s → ∞ R/ On the gauge theory side this is (12.The Holographic Principle and Anti de Sitter Space 135 strongly coupled ’t Hooft limit.4.8. The interesting limit from the viewpoint of the holographic principle is a diﬀerent one.2.4. 12.15) Thus the strongly coupled ’t Hooft limit is also the classical string theory limit in a ﬁxed and large AdS space.4.17) Our goal will be to show that the number of quantum degrees of freedom in the gauge theory description satisﬁes the holographic behavior in equation 12. From the gauge theory point of view the ’t Hooft limit is deﬁned by gym → 0 N →∞ 2 gym N = constant From the bulk string point of view the limit is g→0 R = constant s (12.2 the proper area of any ﬁnite coordinate patch tends to ∞ as the boundary of AdS is approached.1 or 12.2.
Thus we can think of the z location as a scalar ﬁeld living on the Dbrane. A stack of N Dbranes has a mass and Dbrane charge which grow with N . Let us assume that they ﬁll time and the 3 spatial coordinates xi .5.2. t) ﬂuctuations whose action is known from string theory calculations to be a that of conventional 3+1 dimensional scalar ﬁeld theory. x direction as well as fermionic modes needed for supersymmetry. Now a single Dbrane has local degrees of freedom. The most important property of D3branes is that they are embedded in a 10dimensional space. In Chapter 14 we will ﬁnd the interesting behavior that a string appears to grow as we average its properties over smaller and smaller time scales. We are especially interested in D3branes. In fact the geometry deﬁned by 12. For example the location in z may ﬂuctuate. They are stable “impurities” of various dimensionality that can appear in the vacuum.2. Let the other 6 coordinates be called √ z m and let z ≡ z m z m . Information. Dbranes can also be juxtaposed to form stacks of Dbranes. We have already discussed infrared (IR) regulating the area of AdS by introducing a surrogate boundary L at r = 1 − δ or similarly at y = δ. It is in many ways similar to the behavior of strings as we study them at progressively shorter time scales. these apparently diﬀerent regulators are really two sides of the same coin. What makes the Dbrane stack interesting to us is that the geometry sourced by the stack is exactly that of AdS(5) ⊗ S(5). The mass and charge are sources of bulk ﬁelds such as the gravitational ﬁeld. Their properties are widely studied in string theory and we will only quote the results that we need. As we will see. That the IR regulator of the bulk theory is equivalent to an ultraviolet (UV) regulator in the SYM theory is called the IR/UV connection[9]. Dbranes are objects which occur in superstring theory. We will place a “stack” of N D3branes at z = 0. Our main concern will be with the z(x. To understand the relation between this phenomenon and the IR/UV connection in AdS we need to discuss the relation between AdS(5) ⊗ S(5) and Dbranes. Such objects ﬁll 3 dimensions of space and also time. In addition there are modes of the brane which are described by vector ﬁelds with components in the t.3 and 12. and the String Theory Revolution we need to regulate both the area of the AdS boundary and the number of ultraviolet degrees of freedom in the SYM. A Dpbrane is a pdimensional object. Speciﬁcally the geometry sourced by the Dbranes is a particular solution of the supergravity equations of motion: ds2 = F (z)(dt2 − dx2 ) − F (z)−1 dzdz (12.136 Black Holes.4 is closely related to that of a Dbrane stack.18) .
In this lecture we give an argument for the IR/UV connection based on the quantum ﬂuctuations of the positions of the D3branes which are nominally located at the origin of the coordinate z in equation 12..The Holographic Principle and Anti de Sitter Space 137 where F (z) = 1+ cgs N z4 −1/2 (12.4.2. The original coordinate z is deﬁned by z 2 = (z 1 )2 + . All of the ﬁelds in this theory form a single supermultiplet belonging to the adjoint (N × N matrix) representation of SU (N ).5.5.5. 12..2.. We may also choose the six coordinates to be Cartesian coordinates (z 1 .. ω5 . If the six scalar ﬁelds φn are canonically normalized. Furthermore the theory of the ﬂuctuations of the stack is N = 4 SYM. + (z 6 )2 (12..5.23) . The geometry is noncommutative and only conﬁgurations in which the six matrix valued ﬁelds commute have a classical interpretation. z 6 ).2.19) cgs N z4 and c is a numerical constant.22 does not make sense because the ﬁelds φ in SU (N ) are N × N matrices. (12.18 to be the coordinates z m of the N D3branes[10].3. . √ However the radial coordinate z = z m z m can be deﬁned by z2 = gym 2 s R2 2 1 T rφ2 N (12.5.3. where we identify the N eigenvalues of the matrices in equation 12. the coordinates z m are represented in the SYM theory by six scalar ﬁelds on the world volume of the branes.21) As we indicated.5.5.20) It is then a simple exercise to see that the Dbrane metric is of the form 12. then the precise connection between the z s and φ s is z= gym 2 s φ R2 (12. The location of a point on a 3brane is deﬁned by six coordinates z. If we consider the limit in which then we can replace F (z) by the simpler expression F (z) ∼ = z2 (cgs N ) 2 1 >> 1 .22) Strictly speaking equation 12.
Evidently low frequency probes see the branes at z = 0 but as the frequency of the probe increases the brane appears to move toward the boundary at z = ∞.24) where δ is the ultraviolet regulator of the ﬁeld theory.5. What we will ﬁnd there is that the way information is localized in space depends on what frequency range it is probed with. It follows from equation 12.4.6 Counting Degrees of Freedom Let us now turn to the problem of counting the number of degrees of freedom needed to describe the region y > δ [9].20 that the average value of z satisﬁes < z >2 ∼ or. and the String Theory Revolution A question which is often asked is: where are the D3branes located in the AdS space? The usual answer is that they are at the horizon z = 0. The precise connection between the UV SYM cutoﬀ and the bulk theory IR cutoﬀ is given by equation 12. (12.23.25) In terms of the coordinate y which vanishes at the boundary of AdS < y >2 ∼ δ 2 .27) Here it is seen that the location of the brane is given by the ultraviolet cutoﬀ of the ﬁeld theory on the boundary. However our experiences in Chapter 14 with similar questions will warn us that the answer may be more subtle. High frequency or short time probes see the string widely spread in space while low frequency probes see a well localized string. 12. The ﬁelds φ are scalar quantum ﬁelds whose scaling dimensions are known to be exactly (length)−1 . The UV/IR connection implies that this region can be described in terms of an ultraviolet regulated theory with a cutoﬀ length δ.138 Black Holes.5.12 < z >2 ∼ δ −2 (12. To answer the corresponding question about D3branes we need to study the quantum ﬂuctuations of their position. Consider a patch of the boundary with . Information.5.26) gym 2 s R2 2 N δ2 (12. using equation 12.5.5.5. From this it follows that any of the N 2 components of φ satisﬁes φ2 ∼ δ −2 ab (12.
The Holographic Principle and Anti de Sitter Space 139 unit coordinate area.6.13 (12.28) δ3 On the other hand the 8dimensional area of the regulated patch is Ndof ≈ R3 R8 × R5 = 3 δ3 δ and the number of degrees of freedom per unit area is A= Ndof N2 ∼ 8 A R Finally we may use equation 12.6. Thus the number of degrees of freedom on the unit area is N2 (12.30) Ndof 1 ∼ (12. Thus each cell has of order N 2 degrees of freedom corresponding to the N ⊗ N components of the adjoint representation of U (N ).6.31) A G This is very gratifying because it is exactly what is required by the Holographic Principle.6. Within each such cell the ﬁelds are constant in a cutoﬀ theory. Within that patch there are 1/δ 3 cutoﬀ cells of size δ. .29) (12.4.
and the String Theory Revolution .140 Black Holes. Information.
When the black hole absorbs some energy it cools but so does the ﬁnite heat bath. This in turn will favor an additional ﬂow of energy from the bath to the black hole and a runaway will occur. The reason is that their speciﬁc heat is negative. We can try to prevent their evaporation by placing them in a thermal heat bath at their Hawking temperature but that does not work. To see this. But a system with negative speciﬁc heat will lower its temperature when it absorbs energy and will become cooler than the bath. If the box is not too big the heat bath will cool more than the black hole and the ﬂow of heat will 141 . They are longlived objects. but eventually they evaporate. their temperature decreases as their energy or mass increases. suppose a ﬂuctuation occurs in which the black hole absorbs an extra bit of energy from the surrounding heat bath. but of course they are not. Now we will return. full circle. For an ordinary system with positive speciﬁc heat this will raise its temperature which in turn will cause it to radiate back into the environment. We have treated Schwarzschild black holes as if they were states of thermal equilibrium. If on the other hand the black hole gives up some energy to the environment it will become hotter than the bath. to see how the holographic description of AdS(5) ⊗ S(5) provides a description of black holes. Again a runaway will occur that leads the black hole to disappear. The black hole will grow indeﬁnitely. A well known way to stabilize the black hole is to put it in a box so that the environmental heat bath is ﬁnite.Chapter 13 Black Holes in a Box The apparently irreconcilable demands of black hole thermodynamics and the principles of quantum mechanics have led us to a very strange view of the world as a hologram. The ﬂuctuations are selfregulating. Any object with this property is thermodynamically unstable.
The black holes which are stable have Schwarzschild radii as large or larger than the radius of curvature R. The thermodynamics can be derived from the black hole solutions by ﬁrst computing the area of the horizon and then using the Bekenstein–Hawking formula.1) where in this formula r runs from 0 to the boundary at r = ∞.0.0.2.0. dτ 2 = 1+ r2 R2 dt2 − 1 + r2 R2 −1 dr2 − r2 dΩ2 (13.1. They homogeneously ﬁll the 5sphere and are solutions of the dimensionally reduced 5dimensional Einstein equations with a negative cosmological constant. Using the thermodynamic relation dM = T dS we can compute the relation between mass and temperature: M =c R11 T 4 G (13. More speciﬁcally we consider large black holes in AdS(5)⊗S(5) and their holographic description in terms of the N = 4 Yang–Mills theory.4) . One ﬁnds that the entropy is related to the mass by S = c M 3 R11 G−1 1 4 (13.142 Black Holes.0. The horizon of the black hole is at the largest root of 1+ µG r2 − 5 2 2 R R r =0 The Penrose diagram of the AdS black hole is shown in Figure 13. and the String Theory Revolution be back to the bath. Information. r2 The AdS black hole is given by modifying the function 1 + R2 : dτ 2 = 1+ r2 µG − 5 2 R2 R r dt2 − 1 + r2 µG − 5 2 R2 R r −1 dr2 −r2 dΩ2 (13.1. let us write the metric of AdS in a form which is convenient for generalization.3) where c is a numerical constant. Before writing the AdS–Schwarzschild metric.2) where the parameter µ is proportional to the mass of the black hole and G is the 10dimensional Newton constant. In this lecture we will consider the properties of black holes which are stabilized by the natural box provided by Anti de Sitter space. Note that the coordinates r. t are not the same as in equation 12.
This is not too surprising because the classical gravity approximation is only valid 2 when gY M N is large. In the present case the relevant volume is the dimensionless 3area of the unit boundary sphere. . 13.1 Penrose diagram of the AdS black hole or in terms of dimensionless SYM quantities Esym = c R8 4 T G sym (13.0. The second way is to calculate it from the boundary quantum ﬁeld theory in the free ﬁeld approximation.3 can be computed in two ways. The constant c in equation 13.5 is nothing but the Stephan–Boltzmann law for black body radiation.0. The ﬁrst is from the black hole solution and the Bekenstein–Hawking formula.0. The calculations agree in order of magnitude. Evidently the holographic description of the AdS black holes is as simple as it could be. Furthermore there are ∼ N 2 quantum ﬁelds in the U (N ) gauge theory so that apart from a numerical constant equation 13. a black body thermal gas of N 2 species of quanta propagating on the boundary hologram.0.5 has a surprisingly simple interpretation.Black Holes in a Box 143 Fig.5) 4 = cN 2 Tsym Equation 13.6) where V is the volume. Recall that in 3 + 1 dimensions the Stephan–Boltzmann law for the energy density of radiation is E ∼ T 4V (13.0. but the free ﬁeld gives too big a coeﬃcient by a factor of 4/3.
A simple example is the minimally coupled scalar dilaton ﬁeld φ. 13.144 Black Holes. An example is the thermal ﬂuctuations of ﬁelds at high temperature. Let us return to equation 12. Information. In that case the action for φ is the minimally coupled scalar action .5. We will only consider dilaton ﬁelds which are constant on the 5sphere.1 The Horizon The high frequency quantum ﬂuctuation of the location of the D3branes are invisible to a low frequency probe.5. We are not protected in the same way from classical ﬂuctuations. All probes sense thermal ﬂuctuations of the brane locations.8) It is clear that the thermal ﬂuctuations will be strongly felt out to a coordinate distance z = Tsym .5. For each of the N 2 components the thermal ﬂuctuations have the form 2 < φ2 >= Tsym (13.5. Roughly speaking this is insured by the renormalization group as applied to the SYM description of the branes.7) and we ﬁnd equations 12.27 replaced by 2 < z >2 ∼ Tsym 2 −2 < y > ∼ Tsym (13. and the String Theory Revolution 13.24 but now.9) In fact this coincides with the location of the horizon of the AdS black hole.2 Information and the AdS Black Hole The AdS black hole is an ideal laboratory for investigating how bulk quantum ﬁeld theory fails when applied to the ﬁne details of Hawking radiation.25 we use the thermal ﬁeld ﬂuctuations of φ.1.1.1. The renormalization group is what insures that our bodies are not severely damaged by constant exposure to high frequency vacuum ﬂuctuations.26 and 12. Such objects are 10dimensional ﬁelds and should not be confused with the 4dimensional quantum ﬁelds associated with the boundary theory. In terms of r the corresponding position is 1 − r ∼ 1/Tsym (13. instead of using equation 12. Let us consider some ﬁeld that appears in the supergravity description of the bulk.
Plugging in the black hole metric given in equation 13. (13. Secondly.e.2. in the pure AdS background φ is periodic in time. This is true both in pure AdS as well as in the AdS– Schwarzschild metrics.12) for large t − t  The parameter γ depends on the black hole mass or temperature and has the form γ= H(µGR−7 ) R (13.2. r → ∞.11 has implications for quantum correlation functions in the black hole background.2. i. (13. (13.2.13) where H is a dimensionless increasing function of its argument.2. The meaning of this exponential decrease of the correlation function is that the eﬀects of an initial perturbation at time t dissipate away and are eventually lost. It is the value of Φ on the boundary which is identiﬁed with a local ﬁeld in the boundary Super Yang–Mills theory.0. The essential point is that any quantum system with ﬁnite thermal entropy must have a discrete spectrum. This type of behavior is characteristic of large thermal systems where γ would correspond to some dissipation coeﬃcient.10) The appropriate boundary conditions are φ → 0 at the boundary of the AdS.Black Holes in a Box 145 in 5dimensional AdS. This is because the entropy is essentially the logarithm of the number of states per unit energy. exponential decay is not what is really expected for systems of ﬁnite entropy such as the AdS black hole that we are dealing with.2.11) Equation 13.11 requires it to behave like φ(t)φ(t ) ∼ exp−γt − t . but in the black hole metric φ goes to zero exponentially with time: φ → exp−γt. In other words the system does not preserve any memory of the initial perturbation. we ﬁnd a number of things.2. I= √ d5 x −gg µν ∂µ φ∂ν φ. We shall now prove this assertion. Consider the correlator φ(t)φ(t ) . Indeed . First φ(r) ∼ r−4 Φ as r → ∞. Since AdS is exactly described by a conventional quantum system it follows that the correlator should not go to zero. Equation 13. Any quantum system with ﬁnite entropy preserves some memory of a perturbation. However.
Let us now consider a general ﬁnite closed system described by a thermal density matrix and a thermal correlator of the form F (t) = A(0)A(t) = 1 T re−βH A(0)eiHt A(0)e−iHt .2. The long time average L is obviously nonzero and positive. This includes questions of information loss and of particular interest in this lecture.146 Black Holes. This leads us to a very important and general conclusion: any phenomenon which crucially depends on the ﬁniteness of horizon entropy will be gotten wrong by the approximation of QFT in a ﬁxed background. ijkl (13. Let us now consider the long time average of F (t)F ∗ (t). the long time behavior of correlation functions. This observation allows us to understand why it tends to zero in the (bulk) QFT approximation. Z (13. . This is due to the enormous number of short wave length modes near the horizon.2. ij (13. Inserting a complete set of (discrete) energy eigenstates gives F (t) = 1 Z e−βEi ei(Ej −Ei )t Aij 2 .15 it is easy to show that the long time average is L= 1 Z2 e−β(Ei +Ek ) Aij 2 Akl 2 δ(Ej −El +Ek −Ei ) .14) By ﬁnite we simply mean that the spectrum is discrete and the entropy ﬁnite. and it is typically of the order e−S where S is the entropy of the system.16) Using equation 13. and the String Theory Revolution the spectrum of the boundary quantum ﬁeld theory is obviously discrete since it is a theory deﬁned on a ﬁnite sphere.17) where the delta function is deﬁned to be zero if the argument is nonzero and 1 if it is zero. Information.2. Thus it is not possible for the correlator F (t) to tend to zero as the time tends to inﬁnity and the limits required by the AdS/CFT correspondence cannot exist. The value of the long time average for such ﬁnite systems can be estimated.2.2. L = lim 1 T →∞ 2T +T −T dtF (t)F ∗ (t) (13. This means that the time average of F vanishes. In studying QFT in the vicinity of a horizon we have seen that the entropy is UV divergent.15) For simplicity we will assume that the operator A has no matrix elements connecting states of equal energy.
2. The expected behavior is that they ﬂuctuate chaotically. missed by bulk quantum ﬁeld theory. Thus we can expect large ﬂuctuations in the correlators at intervals of order eS . A large ﬂuctuation which reduces the entropy by amount ∆S has probability e−∆S .18) This long time behavior.Black Holes in a Box 147 How exactly do the correlations behave in the long time limit? The answer is not that they uniformly approach constants given by the long time averages. . These ﬂuctuations are analogous to the classical phenomenon of Poincar´ recurrences. (13. It is generally found that the large time behavior e of correlators is chaotic “noise” with the long time average given by L ∼ e−S . is a small part of the encoding of information in the thermal atmosphere of the AdS black hole.
148 Black Holes. and the String Theory Revolution . Information.
PART 3 Black Holes and Strings .
.
In general the failures can be attributed to a common cause – quantum ﬁeld theory has too many degrees of freedom. with a theory that seems to have just the right number of degrees of freedom: string theory. Nevertheless. Later in Chapter 12 we quantiﬁed just how overrich QFT is. Even more relevant for our purposes is the divergence in the entropy per unit area of horizons that was found in Chapter 4. In Chapter 9 we explained that the key to understanding black hole complementarity lies in the ultrahigh frequency oscillations of ﬂuctuations of matter in its own rest frame. we will see certain trends that are more consistent with black hole complementarity than the corresponding trends in QFT. string theory was mostly deﬁned by a set of perturbation rules. The problems posed by black holes for a fundamental theory of quantum gravity are nonperturbative. Until relatively recently. But ultimately local quantum ﬁeld theory fails in a number of ways. Entropy is a direct measure of the number of active degrees of freedom of a system. As a quantum ﬁeld theory. imagine a freely falling whistle that emits a sound of such high frequency that it cannot be heard by the human ear.Chapter 14 Strings We have learned a great deal about black holes by considering the behavior of quantum ﬁelds near horizons. 151 . As an example. The earliest evidence that QFT is too rich in degrees of freedom was the uncontrollable short distance divergences in gravitational perturbation theory. even in perturbation theory. Evidently there are far too many degrees of freedom very close to a horizon in QFT. in an elementary way. The remaining portion of this book deals. Einstein’s general relativity is very badly behaved in the ultraviolet. The extreme red shift between the freely falling frame and the Schwarzschild frame may take phenomena which are of too high frequency to be visible ordinarily and make them visible to the outside observer.
152
Black Holes, Information, and the String Theory Revolution
As the whistle approaches the horizon, the observer outside the black hole hears the frequency red shifted. Eventually it becomes audible, no matter how high the frequency in the whistle’s rest frame. On the other hand, the freely falling observer who accompanies the whistle never gets the beneﬁt of the increasing red shift. She never hears the whistle. This suggests that the consistency of black hole complementarity is a deep constraint on how matter behaves at very short times or high frequencies. Quantum ﬁeld theory gets it wrong, but string theory seems to do better. The qualitative behavior of strings is the subject of this lecture. In order to compare string theory and quantum ﬁeld theory near a horizon, we will ﬁrst study the case of a free particle falling through a Rindler horizon. As we will see, it is natural to use light cone coordinates for this problem. The process and conventions are illustrated in Figure 14.1. The coordinates X ± are deﬁned by X± = and the metric is given by dτ 2 = 2dX + dX − − dX i
2
ρ X0 ± X √ = ∓ √ e∓τ 2 2
(14.0.1)
(14.0.2)
where X i run over the coordinates in the plane of the horizon. We will refer to X i as the transverse coordinates, because they are transverse to the direction of motion of the point particle. The trajectory of the particle is taken to be Xi = 0 X− − X+ = √ 2L (14.0.3)
As the particle falls closer and closer to the horizon, the constant τ surfaces become more and more lightlike in the particle’s rest frame. In other words, the particle and the Rindler observer are boosted relative to one another by an everincreasing boost angle. Near the particle trajectory X + and τ are related by X + ∼ −2L e−2τ = (14.0.4)
for large τ . This suggests that the description of mechanics in terms of the Rindler (or Schwarzschild) time be replaced by a description in light cone
Strings
153
X

X
2 L
+
L
Fig. 14.1
Free particle falling through a Rindler horizon
coordinates with X + playing the role of the independent time coordinate. We will therefore brieﬂy review particle mechanics in the light cone frame.
14.1
Light Cone Quantum Mechanics
In order to write the action for a relativistic point particle we introduce a parameter σ along the world line of the particle. Since the action only depends on the world line and not the way we parameterize it, the action should be invariant under a reparameterization. Toward this end we also
154
Black Holes, Information, and the String Theory Revolution
introduce an “einbein” e(σ) that transforms under σreparameterizations. e1 (σ 1 ) dσ 1 = e(σ) dσ The action is given by W = L = L dσ (14.1.6) −1 2
1 dX µ dXµ e dσ dσ
(14.1.5)
− em
2
where m is the mass of the particle. The action in equation 14.1.6 is invariant under σreparameterizations. Let us now write equation 14.1.6 in terms of light cone coordinates and, at the same time use our gauge freedom to ﬁx σ = X + (which is then treated as a time variable). Then L takes the form L = 1 dX i dX i 1 2 dX − + − e m2 − 2 e dσ e dσ dσ (14.1.7)
The conserved canonical momenta are given by P− = Pi =
∂L ˙ ∂X− ∂L ˙ ∂Xi
= −1 e (14.1.8) =
1 e
˙ Xi
where dot refers to σ derivative. Note that the conservation of P− insures that e(σ) has a ﬁxed constant value. The Hamiltonian is easily obtained by the standard procedure: H = m2 e ePi2 + 2 2 (14.1.9)
This form of H manifests a well known fact about light cone physics. If we focus on the transverse degrees of freedom, the Hamiltonian has all the properties of a nonrelativistic system with Galilean symmetry. The second term in H is just a constant, and can be interpreted as an internal energy that has no eﬀect on the transverse motion. The ﬁrst term has the usual nonrelativistic form with e−1 playing the role of an eﬀective transverse mass. This Hamiltonian and its associated quantum mechanics exactly describes the point particles of conventional free quantum ﬁeld theory formulated in the light cone gauge. Now let us consider the transverse location of the particle as it falls toward the horizon. In particular, suppose the particle is probed by an
Complementarity requires the probe to report that the particle fell into a very high temperature environment in which it repeatedly suﬀered high energy collisions. the instantaneous transverse position should be replaced by the time averi aged coordinate Xδ i Xδ = 1 δ 0 −δ X i (σ) dσ (14. No matter how small δ is.11. There is no tendency for information to transversely spread over a stretched horizon. Let us also assume the very high momentum components of the wave function are negligible. δX.12) eP i δ . the particle is probed over the time interval −δ < X + < 0 (14.1. a time averaged relative coordinate or charge density can be deﬁned. 2 (14. All of this is exactly what is expected for an ordinary particle in free quantum ﬁeld theory. and it too shows no sign of spreading as the sampling interval δ tends to zero.10) by a quantum of (Minkowski) energy ∼ 1 . Why is this a problem? The reason is that it conﬂicts with the complementarity principle.3. This experiment is similar to the δ one discussed in Chapter 9.1. except that the probe carries out information about the transverse location of the particle instead of its baryon number.Strings 155 experiment which takes place over a short time interval δ just before horizon crossing. In this kind of environment the information stored in the infalling system would be thermalized and spread .11) To evaluate equation 14. let us suppose that the particle wave function is initially a smooth wave packet well localized in transverse position with uncertainty ∆X i .13) Finally.1.1. For the more interesting case of an interacting quantum ﬁeld theory. In other words. Under these conditions nothing singular happens i to the probability distribution for Xδ as δ → 0.1. the eﬀective probability distribution for Xδ is concentrated in a well localized region of ﬁxed extent. we use the nonrelativistic equations of motion X i (σ) = X i (0) + eP i σ to give i Xδ = X i (0) + (14. For example.10.1. we could study the transverse properties of an interacting or composite particle such as a hydrogen atom. Since the interaction is spread over the time interval in equation 14.
14 and 14. The transverse coordinates of the point at σ 1 are labeled X i (σ). 0).2. The equation of motion following from equations 14. only the most elementary aspects of string theory will be needed. (14. At equal times X and P satisfy i X i (σ). which is identiﬁed with light cone time X + .2.2.2.15) We have used units in which the string tension (energy per unit length in the rest frame) is unity.16) Quantization of the string is straightforward.16 with periodic boundary conditions in σ 1 . H = 1 P− 2π 0 dσ 2 Pi (σ ) + 2 ∂X i ∂σ 2 (14. There are a number of excellent textbooks on string theory that the reader who is interested in technical details can consult.2. For our purposes.2 Light Cone String Theory Although naive pertubative string theory cannot capture this eﬀect completely correctly. σ 1 ) is a ﬁeld deﬁned on a 1+1 dimensional parameter space (σ a ). the canonical momentum density Pi (σ) can also be deﬁned.14) The light cone Hamiltonian for the free string is a natural generalization of that for a free particle.15 is a simple wave equation ∂2X i (∂σ 0 ) 2 − ∂2X i (∂σ 1 ) 2 = 0. Information. 14. the tendency is already there in the theory of free strings. The implication for the probing experiment is that the particle should somehow spread or diﬀuse roughly the way the eﬀective charge distribution did in Chapter 7. 2π) = X(σ 0 . X i (σ) becomes a free scalar ﬁeld in 1+1 dimensions satisfying equation 14. Pj (σ ) = i δj δ(σ − σ ) (14. and the String Theory Revolution over the horizon. A string is a onedimensional continuum whose points are parameterized by a continuous parameter σ 1 . Thus X i (σ 0 . In addition to X i (σ). It is also a function of a timelike parameter σ 0 . X(σ 0 . where σ 1 runs from 0 to 2π.156 Black Holes. A free string is a generalization of a free particle. .2.
2.2. deﬁne Xδ = 1 δ δ 0 dσ 0 X(σ) (14.0.2. Let us consider the analog of the question that we addressed about the time averaged location of the point particle. we can use the relation betwen Rindler time and Schwarzschild time given by τ = t/4M G to obtain (∆X)2 ∼ αt .Strings 157 The string diﬀers in important ways from the free particle. Now we consider the time averaged location of a point on the string. (14. it doesn’t matter what value σ 1 takes on the right hand side when we evaluate Xδ . But an observer falling with the string and doing low energy experiments on it would conclude that the string remains a ﬁxed ﬁnite size as it falls. Finally.2. the inverse string tension. especially in its short time behavior. The observer outside the black hole will ﬁnd the string diﬀusing over an increasing area of the horizon as time progresses. This quantity is easily calculated and diverges logarithmically as δ → 0. as the string approaches the horizon.20 we have restored the units by including the factor α . 4M G (14.2. As we have repeatedly emphasized.19 is to use the connection between Rindler time and light cone time in equation 14.18) The state used for the expectation value in equation 14.2.19) Another way to write equation 14. A useful measure of how much the information in a string is spread as it falls towards the horizon is provided by the ﬂuctuations in Xδ . that is 2 (∆X)2 = Xδ − Xδ 2 (14. any experiment (from the outside) to determine how its internal parts are distributed will indicate a logarithmic increase in the area it occupies (∆X)2 ∼  log δ . Thus. Here we see the beginnings of an explanation of complementarity.4 (∆X)2 ∼  log (2Le−2τ )  ∼ 2τ.2. . it is the short time behavior that is key to complementarity.17) Since all points σ 1 are equivalent.20) In equation 14. In other words.18 is the ground state string.
23) 1 δ (14. let us consider the ground state average of ∂X1 . 2L ∂X 2 ∂σ1 or scales as (14. it grows exponentially in length. The true exponential asymptotic growth is undoubtedly a nonperturbative phenomenon that involves string interactions in an essential way.0.4 ≈ 1 2τ e . we can conclude that ≈ or using equation 14. Another quantity which exponentially grows is the ρ component of the Rindler momentum. and the String Theory Revolution The linear growth of the area in equation 14. A completely consistent theory would require these growth patterns to match. In that case inspection of 7. we use the transformation in equation 14.2.1 to derive ∂ ∂ρ = 1 √ 2 −e−τ ∂ ∂X + − + eτ ∂ ∂X − . Information. we ﬁnd that the ∂σ ground state average of ∂X i ∂σ1 2 2 is given by ∼ 1 δ2 ∂X i ∂σ1 ∂X i ∂X i ∂σ 1 ∂σ 1 (14.22) is Gaussian in free Using the fact that the probability distribution for ﬁeld theory.2.2.158 Black Holes. as the string falls toward the horizon. . This ∂σ is another exercise in free scalar quantum ﬁeld theory.24) In other words. i If however ∂X is averaged over the time interval δ.2. projected onto the twodimensional transverse plane 2π = 0 dσ 1 ∂X i ∂σ 1 2 (14.2.21 indicates that the growth is exponential. To see how interactions inﬂuence the evolution.20 is much slower than the growth of a charged particle described in Chapter 7.0. To see this.21) i As a preliminary. let’s determine the average total length of string.0. and the result is quadratically divergent.
3. Now. Evidently then the ratio of the string length to its total radial momentum is ﬁxed.2. Obviously the importance of interactions is governed not only by g. It is clear that this behavior cannot continue indeﬁnitely. As the string falls toward the horizon. String interactions are governed by a dimensionless coupling constant g which determines the amplitude for strings to rearrange when they cross. when a piece of string gets within a distance of order α of another piece. The exponential growth of string length and linear growth of area imply that the transverse density of string increases to the point where string interactions must become important and seriously modify the free string picture. the inverse string tension α with units of area. they can interact.3 Interactions In Chapter 7 we saw that a charge falling toward the stretched horizon spreads over an area which grows exponentially. The number of such string encounters will obviously increase without bound as τ → ∞. The the total length of the string grows exponentially with τ . √ Sometimes α is replaced by a length s = α . and therefore as τ → ∞ the radial momentum Pρ grows like eτ . √ Roughly speaking. its radial momentum increases by the mechanism of its physical length increasing. The area occupied by a free string only grows linearly. When g 2 ρ becomes large. It is . The dimensionally correct statement is that string interactions become important when g2ρ ≥ 1 2 s ≈ 1 g2 α (14. 14. but also by the local density of string crossings. the form g 2 ρ is not dimensionless. There is only one dimensional constant in string theory. this not the end of the story. The quantity g 2 α in string theory also governs the gravitational interaction between masses. Let ρ be the number of such crossings per unit horizon area. interactions can no longer be ignored. However.26) This criterion has a profound signiﬁcance. P± are conserved.Strings 159 or in terms of momenta 1 Pρ = √ eτ P− − e−τ P+ 2 (14.25) In the Rindler approximation to a black hole horizon.
The general case goes as follows: Since the number of transverse directions is D2. This is reminiscent of the pattern of growth that we encountered in Chapter 7. provides a phenomenological description of how information spreads over the horizon. the gravitational and string couplings are related by G ≈ g2 D−2 s (14. But the simple assumption that splitting and joining interactions cause eﬀective short range repulsion. and the String Theory Revolution none other than the gravitational coupling constant (in units with c = = 1).29) so that the perturbative limit is again reached when the density is of order ρ ≈ 1 G (14. Although we cannot follow the string past the point where interactions become important. Let D be the spacetime dimension.3. string theory is not a 4dimensional theory. This is the essence of complementarity.27) (14. Information.30) .3. equation 14.28) In D dimensions.26 is replaced by g2 ρ ≥ or ρ ≈ 1 D−2 g2 s 1 D−2 s = 1 (α ) D−2 2 (14. Verifying that as δ → 0 the string grows as if the density saturates is beyond the current technology of string theory. The implication is that interactions become important when the 1 area density of the string approaches G . we can be sure that something new will happen.3. A 1 good guess is that the density of string saturates at order G . Since the spreading is associated with a decreasing time of averaging it is not seen by a freely falling observer. It is important to check if the same logic applies in higher dimensions.160 Black Holes. the area density of horizon entropy.3. Since the τ total length of string grows like ≈ e the area that it occupies must also grow exponentially. In general.3. and that the repulsion prevents the density from increasing indeﬁnitely.
4. Equation 14.4 Longitudinal Motion In discussing the properties of horizons. The only degrees of freedom carried by the string are the transverse coordinates X i (σ) which lie in the plane of the horizon. Thus. It is natural to ask if the stretched horizon has any reality or whether it is just a mathematical ﬁction? Of course.31 has dimension 2.32 then takes the form of an uncertainty principle ∆X − ∆X + ≈ 2 s.Strings 161 14. we have repeatedly run into the idea of the stretched horizon. X i has dimension zero. consistency requires that in some appropriate sense. ∂X i ∂X i ∂X − = ∂σ1 ∂σ1 ∂σ0 (14.33) Now the geometric meaning of equation 14. Let us draw the motion of the ﬂuctuating string in the X ± plane as it falls towards .4. But to a Fido the stretched horizon is the layer containing the physical degrees of freedom that give rise to the entropy of the horizon. (14.4. local operators can be characterized by a mass dimension.4. The right ∂σ side of equation 14. s Another way to write equation 14.4. It is a curious property of string theory that X − (σ) is not an independent degree of freedom. To study this question we must examine how strings move in the X − direction. located a microscopic distance above the mathematical horizon. It immediately follows that the ﬂuctuation in X − satisﬁes ∆X − ∼ 2 s δ (14. while ∂X has dimension 1.33 is very interesting. Thus it is apparent that X − (σ) has dimension 1.4.32 is to observe that δ is an averaging time in light cone coordinates. In other words δ = ∆X + . i For example. In such a theory.4.31) Recall that the quantum theory of X(σ) is a simple (1+1 dimensional) quantum ﬁeld theory of (D2) free scalar ﬁelds deﬁned on a unit circle.32) The factor 2 is needed for dimensional reasons. the degrees of freedom of an infalling object should get deposited in this layer of ﬁnite thickness. The derivation is provided in the supplement that follows this discussion. The longitudinal location is deﬁned by an equation whose origin is in the gauge ﬁxing to the light cone gauge. neither the stretched nor the mathematical horizon appears real. to a Frefo observer.
2 Near horizon Rindler time slices the horizon. Once again this surprising result is a direct consequence of arbitrarily high frequency ﬂuctuations implicit in the stringy structure of matter. In other words. What the ﬁgure illustrates is that the stringy material tends to ﬁll a region out to a ﬁxed proper distance from the mathematical horizon at X + = 0. P = s g. 14. unlike a point particle. the string length s can be much greater than the Planck length. and the String Theory Revolution ∆x  ∆τ} x+ Region of increasing fluctuation Fig. Note that if the string coupling satisﬁes gs << 1.2. the stringy substance is seen by a probe to hover at a distance ∼ s above the horizon. Evidently as X + tends to zero the ﬂuctuation in X − required by equation 14.162 Black Holes. . In this case it is the string length and not the Planck length that controls the distance of the stretched horizon from the mathematical horizon.4. This is shown in Figure 14.33 must increase. Information.
4.4.4.4.35) The Noether charge for this invariance is exactly the quantity in equation 14. In the light cone frame only the transverse X s are dynamical and the generator of these rigid shifts is σ’=0 vs σ=0 Fig. namely shifting σ by a constant.31. ∂σ (14. Thus we must check the validity of the following equation ∂X j ∂X j ∂X + ∂X − − = 0.34.4. but is + given by equation 14. 14. Under the transformation σ → σ + δσ the X s transform as X→X+ ∂X δσ.Strings 163 Supplement: Light cone gauge ﬁxing of longitudinal string motions The coordinate X− (τ . In going to the light cone frame the constraint serves to deﬁne X − in terms of the transverse coordinates. Setting this quantity to zero insures that the physical spectrum consists only of states which are invariant under σ reparameterization.3 String parameter translational invariance .σ) is not an independent degree of freedom. which ∂τ is true for light cone coordinates. To see this multiply 14.31 by 1 = ∂X . ∂τ ∂σ ∂τ ∂σ (14.34) The content of this equation expresses the underlying invariance of string theory to a reparameterization of the σ coordinate. However even in the light cone gauge there is a bit of residual gauge invariance.
∂σ (14.36) which by equation 14. .4.31 is equal to dσ (14.4.4. and the String Theory Revolution dσ ∂X j dX j ∂σ ∂τ ∂X − .37) Setting this to zero simultaneously insures invariance under shifts of σ and periodicity of X − (σ).164 Black Holes. Information.
in one or two cases in which the black holes are invariant under a large amount of supersymmetry the calculations can be reﬁned and give the exact numerical coeﬃcients. but it doesn’t tell us what they are.1) 165 . In this lecture we will see to what extent string theory provides the microstructure and to what extent it enables us to compute black hole entropy microscopically. String theory has many diﬀerent kinds of black holes. Let us begin with the formula for the entropy of a Schwarzschild black hole in an arbitrary number of dimensions. What is more. All of this is in cases where quantum ﬁeld theory would give an inﬁnite result. The black hole metric found by solving Einstein’s equation in D dimensions is given by dτ 2 = D−3 RS 1 − D−3 r dt − 2 D−3 RS 1 − D−3 r −1 dr2 − r2 dωD−2 (15.0. the statistical mechanics of strings allows us to compute the entropy up to numerical factors of order unity. A real theory of quantum gravity should tell us and also allow us to compute the entropy by quantum statistical mechanics. some in higher dimensions. some in 3 + 1 dimensions. Because string theory is not necessarily a 4dimensional theory. it is worth exploring the connections between strings and black holes in any dimension.Chapter 15 Entropy of Strings and Black Holes The Bekenstein–Hawking entropy of black holes points to some kind of microphysical degrees of freedom. In every case the results nontrivially agree with the Bekenstein– Hawking formula. counting microstates. The black holes can be neutral or be charged with the various charges that string theory permits. that is. Call the number of spacetime dimensions D. We will see that for the entire range of such black holes.
0. Information.4 is what is required by black hole thermodynamics.0.0.6) From Einstein’s equation for ideal pressureless matter. (15. Supplement: Schwarzschild geometry in D = d + 1 dimensions To extend a static spherically symmetric geometry to D = d + 1 dimensions.2) and its D2 dimensional “area” is given by D−2 A = RS D−2 dΩD−2 = RS ΩD−2 . + sin2 θ1 .166 Black Holes.sin2 θd−2 dθd−1 ..4) The entropy in equation 15. the metric can be assumed to be of the form ds2 = −e2Φ dt2 + e2∆ dr2 (15.0. and the String Theory Revolution The horizon is deﬁned by RS = 16π(D − 3)GM ΩD−2 (D − 2) 1 D−3 (15.0.8) .0. the Gt component of the Einstein tensor ˆ t can be directly calculated to be of the form Gt = − (D − 2)∆ ˆ t ˆ (D − 2)(D − 3) e−2∆ + (1 − e−2∆ ) r 2r2 (15.0. the entropy is given by S = (2GM ) D−3 ΩD−2 A = .3) Finally.. 4G 4G D−2 (15.7) which can be solved to give (1 − e−2∆ )rD−3 = 2κ D−2 r ρ(r )r 0 D−2 dr = M 2κ (D − 2) ΩD−2 (15. Gtt = κρ.0.. ˆ Using orthonormal coordinates.5) 2 2 2 + r2 dθ1 + sin2 θ1 dθ2 + . This ˆˆ means Gt = − ˆ t ˆ (D − 2) d D−3 r (1 − e−2∆ ) = −κρ 2rD−2 dr (15..
0. 167 where the solid angle is given by ΩD−2 = of the Einstein tensor satisﬁes ˆ Gr = − −(D − 2)Φ r ˆ e−2∆ r ˆ The Gr component r ˆ + (D−2)(D−3) (1 2r 2 −2∆ − e−2∆ ) = − −(D − 2)(Φ + ∆ ) e (15.0. 2π 2RS (15.Entropy of Strings and Black Holes 2π (D−1)/2 Γ((D−1)/2) .0.14) Using the ﬁrst law of thermodynamics.12) which gives the relation between Rindler time/temperature units and Schwarzschild time/temperature units dω = (D − 3) dt 2RS 1 (D − 3) . we can immediately conclude that Φ = −∆.7 is to note its equivalence to the Newtonian Poisson equation in the exterior region ∇2 F (r) = −κρ .0. the Hawking temperature of the black hole is given by THawking = (15.13) Thus.0. a useful shortcut for calculating the solution to Einstein’s equation 15.11) The Hawking temperature can be calculated by determining the dimensional factor between the Rindler time and Schwarzschild time. (15.15) Substituting the form κ = 8π(D − 3)G for the gravitational coupling gives the previous results in Ddimensions. Near the horizon. the entropy can be directly calculated to be of the form S = 2π(D − 3)A κ (15.10) If we write F (r) ≡ e2Φ .0.0. Deﬁning the Schwarzschild radius D−3 2κM = (D−2)ΩD−2 we obtain the form of the metric RS e−2∆ = 1 − RS r D−3 = e2Φ (15.0.9) r + κρ For pressureless matter in the exterior region (ρ = 0 = P ). F (r) = 1 + 2φN ewton . . the proper distance to the horizon is given by ρ = 2RS (D − 3) r Rs D−3 −1 (15.
A very massive black hole might evolve into a large number of low mass strings or. at the opposite extreme. Basically we are using the quantum analog of the method of adiabatic invariants. Very highly excited free strings have an enormously rich spectrum. and the String Theory Revolution All calculations of entropy in string theory make use of a well known trick of quantum mechanics. and it runs from 0 to 2π. On the other hand the quantization of the string deﬁnes a 1+1 dimensional quantum ﬁeld theory in which the (D2) transverse coordinates X i (σ) play the role of free scalar ﬁelds. For any eigenstate of the Hamiltonian with vanishing transverse momentum and unit P− the light cone energy is m2 /2. What happens to the black hole as g tends to zero? The answer is obvious. and so on. monopoles. They can be thought of as a mass of tangled string that forms a timevarying random walk in space. In the process of adiabatic variation. Therefore a neutral black hole must evolve into a collection of free strings. a single very highly excited string. Information. But such a variation will not alter its entropy. Such random walking strings have a large entropy and can be studied statistically. Thus we begin with a black hole of mass Mo in a theory with string coupling go . branes of various dimensionality such as membranes. It must turn into a collection of free strings.168 Black Holes. The spatial coordinate of this ﬁeld theory is σ1 . Let us imagine decreasing the string coupling g. Thus if we can follow the system to a value of the control parameter where the system is tractable we can count the states easily even if the nature of the object changes during the variation. The trick in the string theory context is to vary the strength of the string coupling adiabatically until we arrive at a point where the gravitational forces are so weak that the black hole “morphs” into some more tractable object. Adiabatically varying a control parameter like go will cause a change in the black hole mass and other internal structural features. But only the free strings have ﬁnite energy in the limit g → 0. energy levels are neither created nor destroyed. Entropy is an adiabatic invariant. The entropy of a string of mass m can be calculated by returning to the light cone quantization of the previous lecture. Dbranes. The trick consists of identifying some kind of control parameter that can be adiabtically varied. String theory has all kinds of nonperturbative objects. The counting of the states of a free string is best done in the light .
but similar formulae exist for the various versions of superstring theory. restoring the units S= 2(D − 2)π m s.0.21) Obviously for large n the single string is favored. The entropy and energy of such a quantum ﬁeld theory can be calculated by standard means. 2 2(D − 2)π m (15. each carrying mass m . This is actually quite general.18) where c is a positive constant. For a given total mass. the statistically most likely state in free string theory is a single excited string.0. Therefore the number of states with mass m is Nm = 1 m s c exp 2π(D − 2) m s (15.Entropy of Strings and Black Holes 169 cone version of the theory that we discussed in the last lecture. The leading contribution for large energy is (setting s = 1) E = π T 2 (D − 2) (15.0.17) Subleading corrections can also be calculated to give S = 2(D − 2)π m s − c log (m s) (15.20) − n c log m s n (15. The entropy is the log of the density of states.19 is correct for the simplest bosonic string. Then n Sn (m) = n S(m/n) or Sn (m) = 2(D − 2)π m s (15. Call this entropy Sn (m). a formal light cone temperature T can be deﬁned. In order to describe the highly excited string spectrum.19) The formula 15.16) S = 2πT (D − 2) Using E = m and eliminating the temperature yields S = 2 or. Thus it is expected that when the string coupling goes to zero. Now let us compare the entropy of the single string with that of n strings. .0.0. Recall that the free string is described by means of a 1 + 1 dimensional quantum theory containing D − 2 ﬁelds X i . most of the black hole states will evolve into a single excited string.0.0.
as we decrease g a point will come where the gravitational constant is too weak to matter.25) ≈ s 2 Mo go 1 D−3 . The string and Planck length scales are related by g2 D−2 s = D−2 .0.23) Evidently as g decreases the string length scale becomes increasingly big in Planck units. The point at which this happens is when the horizon radius is of the order of the string scale. p (15.0. That is the point where the black hole makes a transition and begins to act like a string. Let us begin with a black hole of mass Mo in a string theory with coupling constant go . and the String Theory Revolution These observations allow us to estimate the entropy of a black hole. and using G ≈ g2 we ﬁnd RS s D−2 s 1 (15. In what follows we will vary the g while keeping ﬁxed the string length s .22) • At some point as g → 0 the black hole will make a transition to a string.170 Black Holes.0. The assumptions are the following: • A black hole evolves into a single string in the limit g → 0 • Adiabatically sending g to zero is an isentropic process.0. Information. The Schwarzschild radius is of order RS ∼ (Mo G) D−3 . To understand this last assumption begin with a massive black hole. Gravity is clearly important and cannot be ignored.24) (15. This is the point at which the transition from black hole to string occurs. the string length will exceed the Schwarzschild radius of the black hole. (15. the entropy of the ﬁnal string is the same as that of the black hole • The entropy of a highly excited string of mass m is of order S ∼ m s (15. This implies that the Planck length varies. at some value of the coupling that depends on the mass of the black hole.0.26) . But no matter how massive the black hole is. Eventually.
as long as the system remains a black hole. However.0.28 as M (g) = Mo 2 go g2 1 D−2 . the weakly coupled string mass will not change signiﬁcantly. In general the mass will vary during an adiabatic process. But now we can compute the entropy of the free string.29) Now as g → 0 the ratio of the gdependent horizon radius to the string scale decreases. D−2 1 (15. M (g) Since P P = constant. the horizon radius will be much bigger than s .30) ≈ 1 .0.4. Note M (go ) = Mo (15. g2 (15.0.0.31 we ﬁnd M (g) s D−3 ≈ MoD−3 Go .32) As we continue to decrease the coupling. Now start to decrease g.0.0.32. D−2 1 (15.0.0. in this calculation the entropy is calculated as the microscopic entropy of fundamental strings.29 and 15. From equation 15. Let us call the gdependent mass M (g). Thus we see that a black hole of mass Mo will evolve into a free string satisfying equation 15.Entropy of Strings and Black Holes 171 Thus for ﬁxed go if the mass is large enough.0.0.22 we ﬁnd D−3 S ≈ MoD−3 Go .27) The entropy of a Schwarzschild black hole (in any dimension) is a function of the dimensionless variable M P . The stretched horizon is composed of a .2 it becomes of order unity at M (g) which can be written M (g) s D−2 P ≈ D−3 s (15.0.33) This is a very pleasing result in that it agrees with the Bekenstein–Hawking entropy in equation 15. (15.0.0. Thus.0. (15.28) ≈ s g 2 D−2 we can write equation 15. From equation 15. The evolution from black hole to string can be pictorially represented by starting with a large black hole.31) Combining equations 15.
1a. We adiabatically allow g to go to zero and identify the appropriate string conﬁguration that the black hole evolves into. the maximum temperature that a string can achieve. When this occurs there is no need to follow the mass of the black hole as g varies. The method is always the same. Another important property of the stretched horizon is its proper temperature. Information. The area 1 density of string is saturated at ∼ G . 15. the mass is ﬁxed. the temperature of the stringy mass will be TStretched ≈ 1 s This temperature is close to the socalled Hagedorn temperature. Under these conditions the black hole can be compared . In this case the extremal black hole is absolutely stable and in addition.1 Evolution from black hole to string. Finally the horizon radius is no larger than s (Figure 15. and (c) turned into a string stringy mass to a depth of ρ = s as in the diagram Figure 15. (b) with stretched horizon and string scale comparable to radius scale. (a) A black hole with stringy stretched horizon smaller than Schwarzschild radius. the area of the horizon decreases but the depth of the stretched horizon stays ﬁxed as in Figure 15. A particularly interesting situation is that of charged extremal black holes which may be supersymmetric conﬁgurations of a supersymmetric theory.172 Black Holes. Since the proper temperature 1 of a Rindler horizon is 2πρ . As the Schwarzschild radius is decreased (in string units). By now a wide variety of black holes that occur in string theory have been analyzed in this manner.1b. its mass is completely determined by supersymmetry.1c) and the black hole turns into a string. and the String Theory Revolution il d sch arz hw Sc R S Fig.
If one examines multistring ﬂuctuations as a function of temperature. Thus. we have seen that the entropy of the string scales like Sstring ∼ df Ms ls . recall the behavior of the entropy given by S∼log(density of states). To get a feel for the scale of the Hagedorn temperature. one can determine the density of states η for the various string modes m: √ η(m) ∼ exp(4πm α ) This allows the partition function to be written as Z∼ ∞ 0 √ exp(4πm α )exp − m dm T which diverges if the temperature T is greater than the Hagedorn temperature deﬁned by THagedorn ≡ 1 √ 4π α The Hagedorn temperature scales with the inverse string length THagedorn ∼ 1 ls . Using dimensional considerations. where d f is the number of internal degrees of freedom available. The results in these more complicated examples are in precise agreement with the Hawking–Bekenstein entropy. the Hagedorn temperature is the . Hagedorn Temperature Supplement On general grounds. Nevertheless the principles are that same as those that we used to study the Schwarzschild black hole in D dimensions. the density of states behaves like √ eSstring ∼ e df Ms ls ∼ e1/THagedorn which gives the scale T Hagedorn ∼ 1/l s .Entropy of Strings and Black Holes 173 directly to the corresponding weakly coupled string conﬁguration and the entropy read oﬀ from the degeneracy of the string theory spectrum. In the cases where exact calculations are possible the charges carried by black holes are associated not with fundamental strings but Dbranes.
174 Black Holes.2 String “percolation” . Increase Temp Increase Temp T<TH T=TH T>TH Fig. Information.2. 15. and the String Theory Revolution “percolation” temperature for multiple strings ﬂuctuations to coalesce into ﬂuctuations of a single string as represented in Figure 15.
ﬁrst classical ﬁeld theory and later quantum ﬁeld theory. Radical changes are called for. nothing special is felt at the horizon. An extreme example would be the fate of someone. The problem is not a pure ultraviolet problem but an unprecedented mix of short distance and long distance physics. This principle is a new kind of relativity in which the location of phenomena depends on the resolution time available to the experimenter who probes the system. The ﬁrst is Black Hole Complementarity. It was never adequate to deal with the combination of quantum mechanics and general relativity. For a long time it was assumed that this just meant that the theory was incomplete at short distances in the same way that the Fermi theory of weak interactions was incomplete. But the dilemma of apparent information loss in black hole physics that was uncovered by Hawking in 1976 said otherwise. it was assumed that all observers would agree on the usual invariant relationships between events. The new paradigm that is gradually emerging is based on four closely related concepts. This view persisted even in classical general relativity. In order to reconcile the equivalence principle with the rules of quantum mechanics the rules of locality have to be massively modiﬁed. Although quantum mechanics made the event probabilistic and relativity made simultaneity nonabsolute. call her Alice.Conclusions The views of space and time that held sway during most of the 20th century were based on locality and ﬁeld theory. According to the low frequency observer. namely Alice herself. The ﬁrst sign of this was the failure of standard quantum ﬁeld theory methods when applied to the Einstein action. the event. falling into an enormous black hole with Schwarzschild radius of a billion years. The most fundamental object was the spacetime point or better yet. or someone falling with her. But the paradigm is gradually shifting. The horizon is harmless and she 175 .
In this framework the Holographic Principle. Information. . Although the Holographic Principle was regarded with skepticism at ﬁrst it is now part of the mainstream due to Maldacena’s AdS/CFT duality. The number of degrees of freedom per unit volume becomes arbitrarily small as the volume gets large. In apparent complete contradiction. being thermalized. and eventually reemitted as Hawking radiation. The ultimate implications of this. cutoﬀ at the Planck volume. if nothing else. If one wished to study progressively smaller and smaller objects one had to use higher and higher energy probes. the IR/UV connection reverses one of the most fundamental trends of 20th century physics. But once gravity is involved that trend is reversed. In all cases the entropy of the appropriate string system agrees with the Bekenstein–Hawking entropy. All of this takes place just outside the mathematical horizon. Black Hole Complementarity and the IR/UV connection are completely manifest. In many ways this is the most surprising ingredient. The second new idea is the Infrared/Ultraviolet connection. The nonredundant degrees of freedom that describe a region of space are in some sense on its boundary. especially for cosmology are undoubtedly profound but still unknown. and the String Theory Revolution or her descendants can live for a billion years before being crushed at the singularity. As we raise the energy we wind up probing larger and larger distance scales. the key to black hole complementarity is the extreme red shift of the quantum ﬂuctuations as seen by the external observer. the existence of black hole entropy indicates the existence of microscopic degrees of freedom which are not present in the usual Einstein theory of gravity. String theory does provide a microscopic framework for the use of statistical mechanics. there are vastly fewer degrees of freedom than in a ﬁeld theory. At energies above the Planck scale any possible short distance physics that we might look for is shrouded behind a black hole horizon. Finally. At one per Planck area. Third is the Holographic Principle. the high frequency observer who stays outside the black hole ﬁnds that his description involves Alice falling into a hellish region of extreme temperature. not its interior as they would be in ﬁeld theory. What is less clear is the dictionary for decoding the CFT hologram. It does not tell us what they are.176 Black Holes. Very closely related to Black Hole Complementarity. Obviously this has to do with more than just a modiﬁcation of the short distance physics. This. Throughout that century a close connection between energy and size prevailed. As we have seen. provides an existence proof for a consistent microscopic theory of black hole entropy.
Conclusions 177 The theory of black hole entropy is incomplete. And in some ways. In this sense the complete universality of the area–entropy relation is still not fully understood. At the time of the writing of this book there are no good ideas about the quantum world behind the horizon. In no case do we use string theory directly to compare entropy and area. Hopefully our next book will have more to say about this. Then classical general relativity is used to determine the area–mass relation and the Bekenstein–Hawking entropy. In each case a trick. Nor for that matter is there any good idea of how to connect the new paradigm of quantum gravity to cosmology. One very large hole in our understanding of black holes is how to think about the observer who falls through the horizon. . speciﬁc to the particular kind of black object under study. Is this important? It is if you are that observer. is used to determine the relation between entropy and mass for the speciﬁc stringtheoretic object that is believed to represent a particular black hole. an observer in a cosmological setting is very much like one behind a horizon.
178 Black Holes. and the String Theory Revolution . Information.
Theor. Kip S. Focusing and the Holographic Hypothesis. Shenker and L. Price and Douglas A. Phys. Phys. 38 (1999) 1113–1133. hepth/9802150. Phys. 71 (1993) 3743–3746. hepth/9806039. Susskind and Edward Witten. Int.H. Phys. Rev. hepth/0203101. J. Adv. 7. hepth/9711200. Corley and T. W. Phys. Theor. Rev. Adv. Edward Witten. L. hepth/9306083. Raphael Bousso. Susskind. 9. 74 (2002) 825–874. The Holographic Bound in Antide Sitter Space. Rev. Lett. The Holographic Principle. 4. 2 (1998) 253–291. hepth/9805114. hepth/9610043.Bibliography 1. Page. D55 (1997) 5112–5128. Holography and Cosmology. Rev. S. 36 (1995) 6377–6396. D53 (1996) 6720–6724. Math. Don N. Phys. Maldacena. Information in Black Hole Radiation. W. 3. Theor. Leonard Susskind. Fischler. Anti De Sitter Space and Holography. Thorne. Fischler and L. Banks. The Large N Limit of Superconformal Field Theories and Supergravity. Black Holes: The Membrane Paradigm (Yale University Press. 8. grqc/9602043. M Theory as a Matrix Model: A Conjecture. Phys. 2 (1998) 231– 252. Math. J. 5. hepth/9409089. Susskind. 2. Juan M. Jacobson. 10. Math. 179 . The World as a Hologram. Richard H. S. MacDonald. 6. 1986). Phys. Mod. T.
Information.180 Black Holes. and the String Theory Revolution .
5 181 Dbranes. 26 electrical properties. 62 electrostatics. 71 eﬀective potential. 72 entropy. 114. 61. 168 entropy (thermal). 71 equivalence principle. 46. 106 charge. 119 curvature. 168 entropy (coarse grained). 55 charged black holes. 81 entropy (maximum). 54 expansion rate. 123. 102. 85 entanglement entropy (equality). 165. 45 entropy (Von Neumann). 128 barrier penetration. 120. 136 de Sitter space. 17 black hole formation. 35. 31. 48 evaporation time. 97. 62 electromagnetic ﬁeld. 71. 32. 103 complementarity. 70. 113 extended horizon. 131 conformally ﬂat. 21. 135 adiabatic invariants. 170 AdS black hole. 84 caustic lines. 138 density matrix. 76 entropy (ﬁne grained). 15 Boltzmann factor. 101 entropy (calculation). 133 angular momentum. 128. 40 Bousso’s construction. 69 collapsing lightlike shell.Index AdS(5) ⊗ S(5). 73. 45. 7 cosmological constant. 157. 35. 77 evaporation. 63 energy (horizon). 89 Birkoﬀ’s theorem. 69. 102 entropy (quantum ﬁeld theory). 52 entanglement. 34. 51. 121 brick wall. 68 conformal ﬁeld theory. 144 AdS/CFT correspondence. 76 entropy of entanglement. 85. 52. 73 entropy (inﬁnite). 141 ’t Hooft limit. 165 entropy (bounds). 46. 31 coarse graining. 73 entropy (vacuum). 35. 168 adiabatic variation (string coupling). 36. 160 conductivity. 70. 55 classical ﬁelds. 43. 27 anti de Sitter space. 28 baryon number violation. 119 degrees of freedom. 165 entropy (Bekenstein–Hawking). 81 entropy (strings). 11 .
161 ﬁducial observer (see also Fidos). 120 Penrose diagram (F. 49 freely falling observer (see also Frefos).W. 112 path integral. 152. 60. 106 mirror boundary condition. 49. 41. 101 laws of nature. 144 information conservation. 9. 44. 117 Penrose–Bousso diagram (AdS). 156 Liouville’s theorem. 81 information retention time. 130 holography. 151 quantum ﬁelds (Rindler space). 14 Penrose–Bousso diagram. 115 Penrose diagrams. 44 quantum ﬁeld theory. 69. 161 luminosity. 151 holographic principle. 69 level density. 121 Reissner–Nordstrom black hole. 167 Smatrix. 43 quantum ﬁelds. 22 geodesic completeness. 130 e proper distance. 21. 107 free ﬁeld approximation. 152 hyperbolic plane. space). 41. 25. 53 Rindler Hamiltonian. 79 red shift. 134 nocloning principle. 48. 54 Minkowski space. 52. 3 . 85 high frequency phenomena. 97. Information. 8. 77 Kruskal–Szekeres coordinates. 25 Schwarzschild black hole. 32. 81 Hawking radiation. 110 gauge ﬁxing. 79 pair production. 74.182 Black Holes. 25. 8. 8 near horizon wave equation. 130 infalling observer. 58. 110 ﬂuctuations. 21. 127 Poincar´ disk. 5 inﬂation. 142 Penrose diagram (de Sitter space). 52. 32 near horizon coordinates. 102. 167 proton decay. 20. 23 Planck length. 32. 52 light cone gauge ﬁxing. 41. 62. 21 Fischler–Susskind bound. 23. 56 Feynman–Hellman theorem. 101. 28 Newton’s constant. 144. 121 information. 39. 129 ground state (charged black hole). 101. 31. 81 scalar wave equation. 93 focusing theorem. 153 light cone string theory. 10 lattice of discrete spins. 44 momentum. 123 Planck distance. 22 Frefos. 59 Hagedorn temperature. 31 quantum Xerox principle. 163 light cone quantum mechanics. 127. 44 Rindler space. 134. 55 resistance. 127 holography (AdS space). 69 longitudinal string motions. 89 quantization rules. 172 Hawking. 65 Rindler energy. 37 Penrose diagram (AdS black hole). 4.R. 57. 61. 37 Fidos. and the String Theory Revolution extremal black hole. 8. 130 horizon. 170 Planck units. 55 particle horizon. 69. 85 Friedman–Robertson–Walker geometry. 48 reheating.
39 UV/IR connection. 144 thermodynamic instability. 38 transverse spreading. 5 tortoise coordinates. 42. 45. 36. 28 transfer matrix. 161 string coupling constant. 61. 39 temperature (Rindler). 51 tidal forces. 63 temperature (Hawking). 39 thermal atmosphere. 45 . 134 string interactions. 21 stretched horizon. 48. 133 supersymmetry. 3 Schwarzschild geometry (Ddimensions). 39 static observers. 167 standard thermometer. 141 thermodynamics. 101. 135 vacuum. 133 surface charge density. 131. 48 thermal ensemble. 39 thermal ﬂuctuations. 95. 151 Supersymmetric Yang–Mills (SYM). 62. 65 tortoiselike coordinates. 167 temperature (proper). 159 string percolation. 166 second law of thermodynamics. 3. 174 strings. 155 Unruh. 103 solid angle. 7.Index 183 Schwarzschild coordinates.