FFIRS.

qxd 6/16/04 8:37 AM Page iv Quark03 Quark03:Desktop Folder:Chapter-FM:
FFIRS.qxd 6/16/04 8:37 AM Page iv Quark03 Quark03:Desktop Folder:Chapter-FM:
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 Cataloguing-in-Publication 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 981-256-083-1
ISBN 981-256-131-5 (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 401-402, 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 man-in-the-elevator 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 co-exist 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 S-matrix. 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 well-established
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 will-o’-the-wisp 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 Light-like 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

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 non-vanishing 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

2
=

1 −
2MG
r

[dt
2
−(dr

)
2
] −r
2
dΩ
2
(1.2.9)
The interesting point is that the radial-time 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

2
= F(x) dx
µ
dx
ν
η
µν
(1.2.10)
with η
µν
being the usual Minkowski metric. Any two-dimensional 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

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

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

2
= ρ
2

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

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 space-time.
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 two-dimensional surface in the four-
dimensional space-time. 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 near-horizon 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 time-like 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

2
= F(R) [R
2

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 light-like” variables, the radial-time part of the metric takes the
form

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 two-dimensional 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 two-dimensional 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 two-dimensional
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,

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 space-time 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
time-like infinities (t = ±∞) are the beginnings and ends of time-like tra-
jectories. Space-like infinity (r = ∞) is where all space-like 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 light-like 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 light-like 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 space-time
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 light-like 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 space-time. 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 time-like infinity and eventually
ends at future time-like 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 non-spherical distributions. In all
cases the horizon is a lightlike surface which separates the space-time 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 time-like 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 space-time 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
light-rays
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

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

=
p
Z
m
= −
p
Z
m
dT

=
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

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 s-wave 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 s-wave 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

2
− dρ
2
− dX
2
− dY
2
It is useful to replace ρ by a tortoise-like coordinate which again goes to
−∞ at the horizon. We define
u = logρ (2.1.16)
and the metric near the horizon becomes

2
= exp(2u)


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 = 2MG|k|. 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
space-time 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 non-interacting 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
ω-translations 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 Rindler-time
surfaces is shown. The proper time separation between the surfaces is
δτ = ρ δω (3.1.4)
Thus, to push the ω-surface ahead requires a ρ-dependent 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 self-contained 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 space-like 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 so-called 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 ω-translations 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

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

R
, χ

R
) =

Ψ


L
, χ
R
) Ψ(χ
L
, χ

R
) dχ
L
(3.4.28)
Now using equation 3.4.27 we get
ρ
R

R
, χ

R
) =
1
Z

χ
R
| e
−π H
R

L
χ
L
| e
−π H
R

R

L
=
1
Z
χ
R
| e
−2 π H
R

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

=
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

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)

(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 electron-positron 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 space-time 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

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

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

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) =

2
45
T
3
(ρ) δρ B
2
=

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

2
45


ρ
3
=
B
2
(2π)
3

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

×
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 s-wave 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 s-wave 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 s-waves 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 semi-transparent 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 t-clocks, while the Rindler energy is the (dimen-
sionless) energy as defined by observers near the horizon using ω-clocks. 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

dS (5.0.14)
where
1

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 s-wave quanta get out. The s-wave quanta are described
in terms of a 1+1 dimensional quantum field at Rindler temperature
1

.
In the same units, the barrier height for the s-wave 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 four-dimensional 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 time-like 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 radial-time 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 non-extremal black holes, equation 6.0.10 becomes
ds
2

=

2
4r
4
+
ρ
2
dt
2
− dρ
2

= ρ
2

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

(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

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

2
=




2
1 −
R
2
r
2
+

R
2

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 near-horizon 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 light-like, a
system at the stretched horizon is time-like. 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

2
= ρ
2

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

_
_
_
_
˙

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

_
_
_
¸
¸
¸

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 u-axis. 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

E
x
j
y
=
1

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 two-dimensional 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 point-like. 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 z-axis. 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


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

2
o
e

+ 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

e
−2ω

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 S-matrix. 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

B
).

A
)
αα

=

β
Ψ

(α, β) Ψ(α

, β)

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

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

, β

) φ

).
Then

β


B
)
ββ

χ(β

) =

αβ

Ψ

(α, β) Ψ(α, β

) χ(β

)
=

αα

β

Ψ

(α, β) Ψ(α, β

) Ψ

, β

) φ

)
= 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 non-vanishing 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

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

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

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 coarse-grained
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 z-axis, 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 x-axis,
i.e.
1

2
(|↑ + |↓) (8.4.19)

The quantum Xerox principle is sometimes called the no-cloning 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 bomb-in-the-box 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 S-matrix, so that the final state
is given by

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

in
= S


out
(9.0.2)
Now let us consider the S-matrix 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 space-like 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 S-matrix 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

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
Planck-mass 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 space-time,
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 light-like 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 micro-degrees 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 micro-degrees of freedom can absorb,
thermalize, and eventually re-emit 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 space-time. 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 self-consistent. 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
anti-neutrinos, 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 X-boson 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 space-time 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 X-boson 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 non-conservation is especially surprising.
As long as the X-boson 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

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 short-lived
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 short-lived
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 space-time 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
entropy-density 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
state-space. 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 four-dimensional. 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 light-like 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 In-moving (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
Space-time depiction of horizon formation
Consider the collapsing spherically symmetric shell of light-like 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
light-like
surface
Light beams get more
vertical as the energy
becomes more concentrated
γ
Fig. 11.2 Space-time depiction of radially in-moving 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 Light-like Surfaces
So far we have considered the entropy that passes through space-like
surfaces. We will see that it is most natural to define holographic entropy
bounds on light-like surfaces
[
3
]
as opposed to space-like 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 light-like surfaces can be trans-
lated to space-like surfaces, but not always. The case described above is
one of those cases where a space-like bound is derivable.
Let us start with an example in asymptotically flat space-time. 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 three-dimensional light-like surface. In general, they define
a light-like 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 space-time except for those points that lie behind
black hole horizons.
Now consider a space-time 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 two-dimensional 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 light-like 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 two-dimensional 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

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 three-dimensional 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

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 space-like 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
horizon-area 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

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 space-like region Γ bounded
by the space-like 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 space-time, if ∂Γ is convex the outgoing
(ingoing) rays have positive (negative) expansion. However in non-static
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 2-sphere (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 space-time and the
negative expansion branches of a typical local 2-sphere. 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 2-sphere in flat space-time
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
2-sphere at point p. Bousso’s rule tells us nothing in this case about the
entropy on space-like surfaces bounded by p.
Now move on to Region II. The relevant light sheets in this region be-
gin on the 2-sphere 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 space-like 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 space-like surface must
also pass through the future light sheet. By the second law of thermody-
namics the entropy on the space-like surface can not exceed the entropy on
the future light sheet. Thus in this case the entropy in a space-like region
can be holographically bounded. Therefore, one condition for a space-like
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 space-like regions with
boundary p.
In Region III the entropy bounds are both on future light sheets. Nev-
ertheless there is no space-like 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 space-like 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 light-like
bounds in all four regions but only in II and IV are there holographic bounds
on space-like 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 light-like
surface
Fig. 11.13 Bounding surfaces for inflowing entropy
t = 0
t
final
Area is INCREASING
in part of this light-like
surface. The entropy
thru such a surface is
NOT bounded in a
Robertson–Walker
cosmology
Fig. 11.14 Light-like 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 space-time 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. Four-dimensional
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

(11.5.21)
de Sitter space can also be written in the form

2
= dt
2
−a(t)
2
dΩ
2
3
(11.5.22)
where dΩ
2
3
is the metric of a unit 3-sphere, 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

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 space-time
S
Fig. 11.16 Extremal space-like surface in de Sitter space-time
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 space-like 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
Post-inflation
particle horizon
Larger area
Fig. 11.17 Inflationary universe
bounded by the area of the 2-sphere 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 post-inflationary 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 space-like 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 space-time 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 space-like 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 space-like 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
time-like Killing vectors and which can be foliated by space-like 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 non-redundant 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; three-dimensional
space described by a two-dimensional 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 ball-like region with a boundary in the general theory
of relativity. But there is one special situation which is naturally ball-like.
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 10-dimensional
product space consisting of two factors, the 5-dimensional AdS and a
5-sphere 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 5-sphere 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
5-dimensional AdS space may be considered to be a solid 4-dimensional
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 3-sphere. The geometry has
uniform curvature R
−2
where R is the radius of curvature. The metric we
will use is

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,

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 3-sphere 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 time-like
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.

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 5-sphere and add a term to the metric
ds
2
5
= R
2

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 4-dimensional
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 4-dimensional 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(4|2). In
addition there is also the symmetry O(6) associated with rotations of the
internal 5-sphere.
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 4-dimensional 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 light-like
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 10-dimensional
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 5-sphere. The additional symmetry is the so-called 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 9-dimensional 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 9-dimensional 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 8-dimensional 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 + 1-dimensional, 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 10-dimensional 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

2
5

(12.4.10)
Notice that the size of the 5-sphere 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 10-dimensional 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 10-dimensional 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 10-dimensional 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 three-dimensional 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 D-branes.
D-branes are objects which occur in superstring theory. They are sta-
ble “impurities” of various dimensionality that can appear in the vacuum.
A Dp-brane is a p-dimensional object. We are especially interested in
D3-branes. 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 D3-branes is that
they are embedded in a 10-dimensional 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 D3-branes at z = 0.
Now a single D-brane 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 D-brane. 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.
D-branes can also be juxtaposed to form stacks of D-branes. A stack of
N D-branes has a mass and D-brane charge which grow with N. The mass
and charge are sources of bulk fields such as the gravitational field. What
makes the D-brane 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 D-brane stack.
Specifically the geometry sourced by the D-branes 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 D-brane 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 D3-branes which are
nominally located at the origin of the coordinate z in equation 12.2.3. The
location of a point on a 3-brane 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 D3-branes 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 D3-branes 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 8-dimensional 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 long-lived 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 self-regulating.
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 5-sphere
and are solutions of the dimensionally reduced 5-dimensional 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.

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

:

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 10-dimensional 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 3-area 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 D3-branes
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 10-dimensional fields and should
not be confused with the 4-dimensional 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 5-dimensional 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−γ|t −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 over-rich 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 non-perturbative. 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

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 light-like in the particle’s rest frame. In
other words, the particle and the Rindler observer are boosted relative to
one another by an ever-increasing 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 σ-reparameterizations.
e
1

1
) dσ
1
= e(σ) dσ (14.1.5)
The action is given by
W =

Ldσ
L = −
1
2

1
e
dX
µ

dX
µ

− e m
2

(14.1.6)
where m is the mass of the particle. The action in equation 14.1.6 is
invariant under σ-reparameterizations.
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


+
1
e
dX
i

dX
i

− 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 non-relativistic 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 non-relativistic 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 non-relativistic 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 one-dimensional 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
time-like parameter σ
0
, which is identified with light cone time X
+
. Thus
X
i

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

)

= 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


0

2

|P
i

)|
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

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 non-perturbative 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 two-dimensional transverse
plane
=


0

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

. (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 4-dimensional 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 D-2, equation 14.3.26 is re-
placed by
g
2
ρ ≥
1

D−2
s
=
1

)
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 (D-2) 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


∂X
j
∂σ
dX
j
∂τ
(14.4.36)
which by equation 14.4.31 is equal to


∂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 4-dimensional 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

2
=

1 −
R
D−3
S
r
D−3

dt
2

1 −
R
D−3
S
r
D−3

−1
dr
2
− r
2

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

D−2
(D −2)
_ 1
D−3
(15.0.2)
and its D-2 dimensional “area” is given by
A = R
D−2
S
_
dΩ
D−2
= R
D−2
S

D−2
. (15.0.3)
Finally, the entropy is given by
S =
A
4G
=
(2GM)
D−2
D−3

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

dt
2
+ e
2∆
dr
2
+r
2
_

2
1
+ sin
2
θ
1

2
2
+ ... + sin
2
θ
1
...sin
2
θ
d−2

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
=

D−2
_
r
0
ρ(r

)r

D−2
dr

=

(D −2)
M

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
=

(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

(15.0.10)
If we write F(r) ≡ e

, 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

(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 D-dimensions.
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 non-perturbative
objects, branes of various dimensionality such as membranes, D-branes,
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 time-varying
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 (D-2) 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 g-dependent 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 g-dependent 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 so-called 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 D-branes. 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


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 multi-string 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 space-time
point or better yet, the event. Although quantum mechanics made the
event probabilistic and relativity made simultaneity non-absolute, 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 re-emitted 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 non-redundant 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 string-theoretic
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, hep-th/9306083, Phys.
Rev. Lett. 71 (1993) 3743–3746.
3. Leonard Susskind. The World as a Hologram, hep-th/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, hep-th/9806039.
6. Raphael Bousso. The Holographic Principle, hep-th/0203101, Rev. Mod.
Phys. 74 (2002) 825–874.
7. Juan M. Maldacena. The Large N Limit of Superconformal Field Theories
and Supergravity, hep-th/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, hep-th/9802150, Adv.
Theor. Math. Phys. 2 (1998) 253–291.
9. L. Susskind and Edward Witten. The Holographic Bound in Anti-de Sitter
Space, hep-th/9805114.
10. T. Banks, W. Fischler, S.H. Shenker and L. Susskind. M Theory as a Matrix
Model: A Conjecture, hep-th/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 light-like shell, 103
complementarity, 85, 97, 157, 160
conductivity, 68
conformal field theory, 131
conformally flat, 7
cosmological constant, 119
curvature, 5
D-branes, 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
no-cloning 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
S-matrix, 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
(D-dimensions), 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
tortoise-like coordinates, 28
transfer matrix, 38
transverse spreading, 155
Unruh, 36, 39
UV/IR connection, 95, 101, 135
vacuum, 45

.

Howard University. CA 943054060 † Permanent address. Stanford University. Department of Physics. DC 20059 . AND THE STRING THEORY REVOLUTION The Holographic Universe Leonard Susskind∗ James Lindesay† ∗ Permanent address. INFORMATION. Stanford. Department of Physics. Washington.AN INTRODUCTION TO BLACK HOLES.

vi .

For photocopying of material in this volume. or parts thereof. MA 01923. . INFORMATION AND THE STRING THEORY REVOLUTION The Holographic Universe Copyright © 2005 by World Scientific Publishing Co. Covent Garden. without written permission from the Publisher. please pay a copying fee through the Copyright Clearance Center. Ltd. All rights reserved. ISBN 981-256-083-1 ISBN 981-256-131-5 (pbk) Printed in Singapore. Pte. Danvers. 222 Rosewood Drive. AN INTRODUCTION TO BLACK HOLES. including photocopying. 5 Toh Tuck Link. London WC2H 9HE British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.Published by World Scientific Publishing Co. recording or any information storage and retrieval system now known or to be invented. may not be reproduced in any form or by any means. electronic or mechanical. This book. In this case permission to photocopy is not required from the publisher. Ltd. Hackensack.. USA. Inc. Singapore 596224 USA office: 27 Warren Street. Suite 401-402. NJ 07601 UK office: 57 Shelton Street. Pte.

The paradigm of relativistic quantum field theory almost certainly has to be replaced. The earliest origins of quantum mechanics were not experimental atomic physics. to the Planck scale. paradoxes and contradictions. The subject is riddled with paradox and contradiction. Planck asked. nothing comparable has been learned about the quantum theory of gravitation. 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. The methods that were invented to quantize electrodynamics. 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. and the study of gedanken experiments. Under this circumstance our best hope is an examination of fundamental principles. radioactivity. that is. or spectral lines. Although there is undoubtably more to be learned both theoretically and experimentally. in the 80 years that we have had the general theory of relativity. for the infinite collection of radiation oscillators to have a finite specific heat? vii . Such strategy has worked in the past. How was it possible.Preface It is now almost a century since the year 1905. The puzzle which started the whole thing was a contradiction between the principles of statistical thermodynamics and the field concept of Faraday and Maxwell. prove wholly inadequate when applied to gravitation. By contrast. One has the distinct impression that we are thinking about the things in the wrong way. which were so successfully generalized to build the standard model. 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. in which the principle of relativity and the hypothesis of the quantum of radiation were introduced.

By this simple means a contradiction was exposed between the symmetries of Newton’s and Galileo’s mechanics and those of Maxwell’s electrodynamics. As long as the black hole is present. By an analysis of gedanken experiments. one can do the bookkeeping so that it is the black hole itself which is correlated to the matter outside. 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. Information. 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. Thus. Furthermore. The paradox was discovered by Jacob Bekenstein and turned into a serious crisis by Stephen Hawking. Bekenstein realized that if the second law of thermodynamics was not to be violated in the presence of a black hole. the two theories appear to lead to a serious clash that once again involves statistical thermodynamics in an essential way. 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 electromagnetic field would be seen as a static. consistent with causality. Eventually the black hole must completely evaporate. 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. What known properties of nature should we look to. But eventually the black hole will evaporate. According to Einstein. The development of the general theory from the principle of equivalence and the man-in-the-elevator gedanken experiment is also a matter of historical fact. This in itself is a source of paradox. the black hole must possess an intrinsic entropy. For two decades this contradiction between . 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.viii Black Holes. at the age of 15 he formulated the following paradox: suppose an observer moved along with a light beam and observed it. Hawking then made arguments that there is no way. for the correlations to be carried by the outgoing evaporation products. according to Hawking. 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. But no such solution to Maxwell’s equations exists. spatially varying field. In each of these cases the consistency of readily observed properties of nature which had been known for many years required revolutionary paradigm shifts.

Quite apart from the question of the ultimate correctness and consistency of string theory. with or without strings. ’t Hooft has argued that by resolving the paradox and removing the contradiction. it becomes compulsory in all processes involving the Planck scale. string theory appears to be a far more consistent mathematical framework for quantum gravity than ordinary field theory. The trouble with this is that there is no known way to destroy coherence without. including ’t Hooft and Susskind. The theory is out of control as argued by Banks. A second purpose involves development of string theory as a unified description of elementary particles. As we shall see. matter. The analysis of these differences suggests a resolution of the black hole dilemma and a completely new view of the relations between space. Throughout this period. The world would behave as if it were in a noisy environment which continuously leads to a loss of coherence. To navigate it without disaster we will need some beacons in the form of trusted principles that we can turn to for direction. including their gravitational interactions. in the neighborhood of a black hole horizon the differences become extreme. 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. The quantum theory of black holes. The main purpose of this book is to lay out this case. time. and information. In this book the absolute truth of the following four propositions will be . Although still very incomplete. there are important lessons to be drawn from the differences between these two theories. it is a “trackless swamp” with many false but seductive paths and no maps. It is therefore worth exploring the differences between string theory and field theory in the context of black hole paradoxes. at the same time violating energy conservation by heating the world. have felt that the basic principles of quantum mechanics and statistical mechanics have to be made to co-exist with black hole evaporation. and ’t Hooft. From an operational viewpoint this would mean that the standard rules of quantum mechanics would not apply to processes involving black holes. the way to the new paradigm will be opened.Preface ix the principles of general relativity and quantum mechanics has so puzzled theorists that many now see it as a serious crisis. although string theory is usually well approximated by local quantum field theory. a few theorists. To borrow words from Sidney Coleman. Hawking further argued that once the loss of quantum coherence is permitted in black hole evaporation. is far from being a textbook subject with well defined rules. Peskin and Susskind.

The existence. 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. but it nevertheless rests on well-established foundations. 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. but also on future events. Thermodynamics results from coarse graining a more microscopic description so that states with similar macroscopic behavior are lumped into a single thermodynamic state. and shape of a horizon depend not only on past occurrences. These three propositions. Proposition 2 may perhaps be less obvious. entropy. size. The global process. taken by themselves. luminosity. The usual laws of nature with no abrupt external perturbations will be found valid until the influence of the singularity is encountered. this means that observations performed by observers who remain outside the black hole can be described by a unitary time evolution. energy momentum flux. Information. 3) Thirdly we assume the usual connection between thermodynamics and quantum statistical mechanics. and temperature. then a freely falling observer should detect nothing out of the ordinary when passing the horizon. Those features include the thermodynamic properties. location. If the horizon scale is large enough so that tidal forces can be ignored. More specifically we assume that a thermodynamic entropy S implies that approximately exp(S) quantum states have been lumped into one thermal state. Proposition 1 and 3 apply to all known forms of matter. are in no way radical.x Black Holes. and the String Theory Revolution assumed: 1) The formation and evaporation of a black hole is consistent with the basic principles of quantum mechanics. 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. We ourselves could right now be at the horizon of a gigantic black . it must also have a luminosity. 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. 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 approximate black body character of Hawking radiation. beginning with asymptotic infalling objects and ending with asymptotic outgoing evaporation products is consistent with the existence of a unitary S-matrix. In particular. carrying their laboratories with them.

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. Beware the will-o’-the-wisp and don’t lose your nerve. while all around false paths beckon. A mathematical surface of this sort should have no local effect on matter in its vicinity. .Preface xi hole caused by matter yet to collapse in the future. 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.

Information.xii Black Holes. and the String Theory Revolution .

. . . . . . . . . Review of the Density Matrix The Unruh Density Matrix . Kruskal–Szekeres Coordinates . .2 3. . . . 31 32 34 36 39 43 48 4. . . . . . . . . . 1 3 3 7 8 10 14 15 21 25 28 31 2. . Penrose Diagrams . . . . . Quantum Fields in Rindler Space 3. . . . . . . . . . . . . . . . . Tortoise Coordinates . . .5 Classical Fields . . . . . . . . . . . . . . Fidos and Frefos and the Equivalence Principle .2 1. . . . . . . . . . . . . .1 Black Hole Evaporation . Entropy of the Free Quantum Field in Rindler Space 4. . . . . . . . .4 1. . . . . . . . . . . . . . . . .5 1. . . . . Entanglement . . . . .1 1. . . . . . . . . . . . xiii . . . . . . . . . .4 3. .7 Schwarzschild Coordinates . . .1 3. . . .Contents Preface vii Part 1: Black Holes and Quantum Mechanics 1. . . . .1 Near the Horizon . .6 1. . . . . . . . . . . . . . . . Scalar Wave Equation in a Schwarzschild Background 2. . . .3 3. . . . . . Formation of a Black Hole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Near Horizon Coordinates (Rindler space) .3 1. . . . . . . . . . The Schwarzschild Black Hole 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. . . . . Proper Temperature .

. . . . .2 11. . . . . and the String Theory Revolution 5. . . . . . . . . .xiv Black Holes. . . . . . . . . . . . Holography in AdS Space . . . . . . . . . . . Horizons and the UV/IR Connection Part 2: Entropy Bounds and Holography 11. . . . . . . . . . . . . . .3 A Brick Wall? . . . . . . . . . . . .4 Information Conservation Entanglement Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 9. . . .1 12. . . . . .3 12. . . . . . . . . . . . . Anti de Sitter Space . . . . . . . Thermodynamics of Black Holes 6. . The Puzzle of Information Conservation in Black Hole Environments 9.2 12.4 12.2 9. . . . . . . . . . .5 11. . . . . Entropy Bounds 11. . . . . . de Sitter Cosmology . . . . . . . . . 51 55 61 69 69 71 77 79 9. . . AdS Space . . . . . . . . . . . . . . . .5 12. . . . . . . . Entropy on Light-like Surfaces . . . . . . . . . . . . . . . . . . The Laws of Nature 8. . . . Baryon Number Violation . . . . . . .4 11. . . . . . . Black Hole Complementarity . . . . . . . . . . . . The Holographic Principle and Anti de Sitter Space 12. The Infrared Ultraviolet Connection Counting Degrees of Freedom . . . . . . . . . . . . . . . . . . . . . . . . . . Charged Black Holes 7. . . . .3 11. . . . . . . . . . . . . . . . . . . 127 128 130 133 135 138 12. .6 The Holographic Principle .1 11. . . . . . . . . . . . . . . . Friedman–Robertson–Walker Geometry Bousso’s Generalization . . . . . The Stretched Horizon 8. . .6 Maximum Entropy . . . . . .2 8. 81 84 85 89 95 10. . . . The AdS/CFT Correspondence . . . . . . . .1 8. Information. . . . Equivalence Principle . . . . . . . . . . . . 99 101 101 105 110 114 119 123 127 . . . . . . . . . . . .3 8. . . . . . Quantum Xerox Principle .

. . . . . Interactions . . . . 144 Part 3: Black Holes and Strings 14. . . .1 14. Black Holes in a Box 13. . . . . . . .2 14. . . . . . . . . . . . . . .4 Light Cone Quantum Mechanics Light Cone String Theory . . . . 149 151 153 156 159 161 165 175 179 181 15. . Entropy of Strings and Black Holes Conclusions Bibliography Index . .1 13. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 141 The Horizon . . . . . . . . . . . . .Contents xv 13. . . . . . Strings 14. . . . . 144 Information and the AdS Black Hole . . . . . .3 14. Longitudinal Motion . . . . . . . . .

PART 1 Black Holes and Quantum Mechanics .

.

and it represents the time recorded by a standard clock at rest at spatial infinity.Chapter 1 The Schwarzschild Black Hole Before beginning the study of the quantum theory of black holes. which cover only the exterior of the horizon are in many ways more valuable. 1. the Schwarzschild geometry is manifestly spherically symmetric and static. It does not measure proper 3 . We begin with the simplest spherically symmetric static uncharged black holes described by Schwarzschild geometry.1. The coordinate r is called the Schwarzschild radial coordinate. The coordinate t is called Schwarzschild time. The metric is given by dτ 2 = (1 − 2MG 2 r )dt − (1 − 2MG −1 2 dr r ) − r2 dΩ2 (1. For example. or the related tortoise coordinates. and no one of them is in any sense the best or most correct description. For these purposes. 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. Schwarzschild coordinates. 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.1 Schwarzschild Coordinates In Schwarzschild coordinates. where dΩ2 ≡ dθ2 + sin2 θdφ2 . the Kruskal– Szekeres coordinate system is valuable for obtaining a global overview of the entire geometry.1) = gµν dxµ dxν .

or at least that part which can influence his detectors.1. is given by the coordinate r = 2M G. To a distant observer the horizon represents the boundary of the world.1.3) t = ( R + 2M G) 2 +2M G log R 2MG 1/2 η + R 2 R 2MG −1 1/2 sinη (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.5. Information. The horizon.5) Evidently the proper time when crossing the horizon is finite and smaller than the expression in equation 1. Nevertheless there is a very important sense in which the horizon is globally special if not singular. For simplicity let’s consider a radially freely falling observer who is dropped from rest from the point r = R. but is defined so that the area of the 2sphere at r is 4πr2 . At the horizon grr becomes 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. which we will tentatively define as the place where g00 vanishes.4 Black Holes. . The trajectory of the observer in parametric form is given by r= R (1 + cosη) 2 1/2 (1. In equation 1.1 we have chosen units such that the speed of light is 1. In what follows we will see that no local invariant properties of the geometry are singular at r = 2M G.2) τ= R 2 R 2M G −1 1/2 (η + sinη) (1.4) 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.1. φ are the usual polar and azimuthal angles. The angles θ.1. Thus a small laboratory in free fall at r = 2M G would record nothing unusual. 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. and the String Theory Revolution spatial distance from the origin.1.1.

Assuming that the infalling observer sends signals at a given frequency ν. The non-vanishing 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. φ with τ oriented along the observer’s instantaneous time ˆ ˆ ˆ ˆ ˆ axis. θ. τ .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. For a large mass black hole they are typically very small.1.7) at the horizon. ρ.1. Let us now consider the history of the infalling observer from the viewpoint of a distant observer. At this point the curvature increases to the point where the classical laws of nature must fail.4 is that the crossing of the horizon does not occur at any finite Schwarzschild time.6) Thus all the curvature components are finite and of order R(Horizon) ∼ 1 M 2 G2 (1. It is easily seen that as r tends to 2M G. the distant observer sees those signals with a progressively decreasing frequency. This is shown in Figure 1. Over the entire span of Schwarzschild time the distant observer records only a finite number of pulses from the infalling transmitter. Define an orthonormal frame such that the observer is momentarily at rest. On the other hand the tidal forces diverge as r → 0 where a true local singularity occurs.1. Unless the infalling observer increases the frequency of his/her signals to infinity as the horizon is approached. Thus the infalling observer passes smoothly and safely through the horizon. but rather senses the tidal forces or curvature components. According to classical physics the infalling observer can use an arbitrarily large carrier frequency to send an arbitrarily large amount of information using . 1. The first surprising thing we learn from equations 1.1.1. t tends to infinity. and ρ pointing radially out. 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.3.2. Furthermore a signal originating at the horizon cannot reach any point r > 2M G until an infinite Schwarzschild time has elapsed.1. and 1. We may suppose that the infalling observer sends out signals which are received by the distant observer. We can construct unit basis vectors. the distant observer will inevitably run out of signals and lose track of the transmitter after a finite number of pulses.

6 Black Holes. Therefore. Information. 1.1 Infalling observer sending signals to distant Schwarzschild observer an arbitrarily small energy without significantly disturbing the black hole and its geometry. in principle. As the observer approaches the horizon. dis- . implying that the observer must have had a large energy available. However quantum mechanics requires that to send even a single bit of information requires a quantum of energy. the distant observer can obtain information about the neighborhood of the horizon and the infalling system right up to the point of horizon crossing. 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. this quantum must have higher and higher frequency.

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. and a slice through Schwarzschild space at fixed θ.11) Note: r∗ → −∞ at the horizon. Any two-dimensional space is conformally flat. The tortoise coordinate r∗ is given explicitly by r∗ = r + 2M G log r − 2M G 2M G (1. 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. Thereafter. φ is no exception. We define the tortoise coordinate r∗ by 1 1− so that dτ 2 = 1− 2M G r [dt2 − (dr∗ )2 ] − r2 dΩ2 (1.2.2.10) with ηµν being the usual Minkowski metric. In equation 1. as we shall see.8) The interesting point is that the radial-time part of the metric now has a particularly simple form.2. called conformally flat.2.2. We shall see that wave equations in the black hole background have a very simple form in tortoise coordinates.9) 2MG r dr2 = 1− 2M G r (dr∗ )2 (1.The Schwarzschild Black Hole 7 turbing the very quantity to be measured.9 the metric of such a slice is manifestly conformally flat. . Furthermore it is also static. 1.

15) Furthermore.14) − dρ2 − r2 (ρ) dΩ2 (1. Minkowski coordinates T .3. and the String Theory Revolution 1.18) t 4M G (1. 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 (1.3 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 (1.3.13) Near the horizon equation 1.3. we can introduce a dimensionless time ω ω = and the metric then takes the form dτ 2 = ρ2 dω 2 − dρ2 − dx2 − dy 2 (1.16) y = 2M G θ sinφ Finally.8 Black Holes.12 behaves like ρ ≈ 2 2M G(r − 2M G) giving dτ 2 ∼ ρ2 = dt 4M G 2 (1.3.3. Information. 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 (1.17) It is now evident that ρ and ω are radial and hyperbolic angle variables for an ordinary Minkowski space.3.3.

3.20) It should be kept in mind that equation 1.19) Z = ρ coshω to get the familiar Minkowski metric dτ 2 = dT 2 − dZ 2 − dX 2 − dY 2 (1. namely II ω=ω 2 ω=ω 1 III ρ=ρ1 t=0 I IV ρ=ρ2 Fig.The Schwarzschild Black Hole 9 defined by T = ρ sinhω (1. In Figure 1. II. III.2 the relation between Minkowski coordinates and the ρ. Only one of those regions. The entire Minkowski space is divided into four quadrants labeled I. 1.2 Relation between Minkowski and Rindler coordinates . ω coordinates is shown. and only for a small angular region. for a large black hole. is locally almost indistinguishable from flat space-time.20 is only accurate near r = 2M G.3.3. and. and IV. However it clearly demonstrates that the horizon is locally nonsingular.

4. The time-like coordinate.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. ω. The approximation of the near-horizon region by Minkowski space is called the Rindler approximation.21) For small ρ equation 1. is called Rindler time. However. The horizon itself is the origin T = Z = 0. Note that it is a two-dimensional surface in the fourdimensional space-time.22) F (R) dR2 = from which it follows that 4M G log or (1.4.23) R = r + 2M G log MG r − 2M G 2M G r∗ 4M G = r∗ (1.4. i. since originally the horizon was defined by the single constraint r = 2M G.24) R = M G exp (1. 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. Region I. 1.4. This may appear surprising. In particular the portion of Minkowski space approximating the exterior region of the black hole. Therefore the horizon has no extension or metrical size in the time direction.10 Black Holes.e.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. recall that at the horizon g00 vanishes. Information. and the String Theory Revolution Region I lies outside the black hole horizon. and therefore appears to be a three dimensional surface.15 shows that ρ ≈ R.4. is called Rindler space. Note that a translation of Rindler time ω → ω + constant is equivalent to a Lorentz boost in Minkowski space. .

27) The coordinates U . crosses r = 2M G only after an infinite Schwarzschild time.3.The Schwarzschild Black Hole 11 Letting R eω = V (1.4. Although the extended horizons lie at finite values of the Kruskal–Szekeres coordinates. In this region the surfaces of constant r are the + r = con nt sta H . they are located at Schwarzschild time ±∞. The surfaces of constant r are the timelike hyperbolas in Figure 1. the radial-time part of the metric takes the form dτ 2 = F (R) dU dV (1. V are shown in Figure 1. 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.4.3 U. V Kruskal–Szekeris coordinates The region of Schwarzschild space with r < 2M G can be taken to be Region II.26) Re −ω = −U be “radial light-like” variables.3. 1. U V H II III I IV Fig. Thus we see that a particle trajectory which crosses H + in a finite proper time.

and the String Theory Revolution spacelike hyperboloids U V = positive constant The true singularity at r = 0 occurs at R2 = − (M G) . 1.4. A nonradially directed light ray or timelike trajectory always lies inside the two-dimensional light cone. 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= .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 two-dimensional light cone bounded by 45o lines.28) (1. With this in mind.4. A radial moving light ray travels on a trajectory V = constant or U = constant.12 Black Holes. It is shown in Figure 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. Information.29) The entire maximal analytic extension of the Schwarzschild geometry is easily described in Kruskal–Szekeres coordinates.4. or U V = (M G) 2 2 (1.

A radially outgoing light ray from P1 will escape falling into the singularity as shown in Figure 1. From Region III no signal can ever get to Region I. 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. III H or i H .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 we will see. Consider a point P1 in Region I. Furthermore.The Schwarzschild Black Hole 13 geometry.5. 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. 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 . Nevertheless let us consider them. and so it is also behind the horizon. An incoming light ray from P1 will eventually cross H + and then hit the future singularity. For this reason Region II is said to be behind the horizon.G 8 Fig. no signal can ever escape to Region I.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. 1. Regions III and IV.

dτ 2 = dt2 − dr2 − angular part = (dt + dr)(dt − dr) − angular part (1. Also shown in Figure 1.32) as shown in Figure 1.5. Space-like infinity (r = ∞) is where all space-like surfaces end. 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.5.31) will preserve the form of the light cone.14 Black Holes. like the Schwarzschild black hole. especially if.5 Penrose Diagrams Penrose diagrams are a useful way to represent the causal structure of spacetimes.5. and the String Theory Revolution communicate with Region I. Information. Furthermore they “compactify” the geometry so that it can be drawn in total on the finite plane.6 are some representative contours of constant r and t. They represent the geometry of a two-dimensional surface of fixed angular coordinates. they have spherical symmetry. Ignoring angular coordinates. We can use such a transformation to map the entire infinite space 0 ≤ r ≤ ∞. Future and past time-like infinities (t = ±∞) are the beginnings and ends of time-like trajectories.6. In addition to these there are two other infinities which are called I ± . 1. . consider ordinary flat Minkowski space. For example Y + = tanh(t + r) Y − = tanh(t − r) The entire space-time is mapped to the finite triangle bounded by Y+ = 1 Y − = −1 + Y − Y− = 0 (1.5. As an example.30) Radial light rays propagate on the light cone dt ± dr = 0. There are several infinities on the Penrose diagram.33) (1. Region I however cannot communicate with Region IV. −∞ ≤ t ≤ +∞ to a finite portion of the plane. Any transformation that is of the form Y + = F (t + r) Y − = F (t − r) (1.

6 Penrose diagram for Minkowski space They are past and future light-like infinity. gravitons. For example.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 . a shell of photons. and they represent the origin of incoming light rays and the end of outgoing light rays. black holes are formed from the collapse of gravitating matter.6 Formation of a Black Hole The eternal black hole described by the static Schwarzschild geometry is an idealization. or massless neutrinos with very small radial extension and total energy M provides an example. In nature.=-1 t=- Fig. Similar deformations can be carried out for more interesting geometries. The resulting Penrose diagram is shown in Figure 1. To construct the geometry. such as the black hole geometry represent by Kruskal–Szekeres coordinates.7. 1. we begin with the empty space Penrose diagram with the infalling shell represented by an incoming light-like line . The simpest model for black hole formation involves a collapsing thin spherical shell of massless matter. 8 1.

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. 1. The infalling shell divides the Penrose diagram into two regions.7 Penrose diagram for Schwarzschild black hole. The particular value of Y + chosen for the trajectory is arbitrary since any two such values are related by a time translation.16 Black Holes. showing regions (top) and curves of fixed radial position and constant time (bottom) (see Figure 1. A and B.8). The Region A is interior to the shell and represents the initial flat space-time 8 8 8 I r = H . Information.

9. To form the full classical evolution the regions A of Figure 1. 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. φ) is continuous.The Schwarzschild Black Hole 17 YB Y+ A - Fig. in Figure 1.8 and B’ of Figure 1. Once again the particular value of Y + chosen for the trajectory is immaterial. However this must be done so that the “radius” of the local two sphere represented by the angular coordinates (θ. In this case Birkoff’s theorem tells us that the geometry outside the shell must be the Schwarzschild geometry.8 Minkowski space Penrose diagram for radially infalling spherical shell of massless particles with energy M before the shell passes. the mathematical identification of the boundaries of A and B’ must respect the continuity of the variable r. Just as 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 other words.9 the Region A’ is to be discarded. Accordingly. Much the same is true in general relativity. 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. 1.9 must be glued together.8 where the Region B is unphysical.

a light-like surface H is shown as a dotted line. 1. 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. the horizon also extends into the Region A where the metric is just that of flat space-time. The distant observer collects information that arrives at any instant from his backward light cone.7. Evidently such an observer never actually sees events on the Y + . On the other hand.10 we show the resulting Penrose diagram for the complete geometry. On Fig 1.10. The observer originates at past time-like infinity and eventually ends at future time-like infinity. as shown in Figure 1. Consider next a distant observer located on a trajectory with r >> 2M G. Information. but this will not disturb the form of the light cones. 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 = ∞).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. and the String Theory Revolution - Y B’ A’ - Fig. that is it coincides with the future horizon of the final black hole geometry and is therefore found at r = 2M G.11. Thus in Figure 1. the identification is always possible. 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. One of the two Penrose diagrams will have to undergo a deformation along the Y − direction in order to make the identification smoothly. In Region B’ the horizon is identical to the surface H + of Figure 1.18 Black Holes. This identifies H as the horizon.

1.10 Penrose diagram for collapsing shell of massless particles .The Schwarzschild Black Hole 19 areas Match here of 2-sp B’ A r=0 H r < 2M G t= G M 8 =2 + r r= r=0 8 - Fig.

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 non-spherical distributions. In all cases the horizon is a lightlike surface which separates the space-time 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 time-like 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 space-time 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 light-rays

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)

The Schwarzschild Black Hole 23 For large times we find t ) (1.37) 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.13 Sedimentary layers of infalling matter on horizon Horizo n . As a consequence of the boost.7. exponentially decreases with time t ) (1. pρ ≈ pτ ≈ 2pZ expω = 2pZ exp( Fig.7. 1. The Frefos of course see the matter behaving in a totally unexceptional way.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.13). 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. 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. the Fidos see all matter undergoing Lorentz contraction into a system of arbitrarily thin “pancakes” as it approaches the horizon. Eventually. 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. The hyperbolic boost angle is the dimensionless time ω. the proper spatial distance of the particle from the horizon.

and the String Theory Revolution . Information.24 Black Holes.

4) )] dt dr dθ dφ ∗ (sinθ + 25 .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 (2.0. Let us consider a conventional massless free Klein–Gordon field χ in the Schwarzschild background.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.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. 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. In this lecture we will study the properties of a scalar wave equation in the background of a black hole.3) (2.0.

and pulls a wave packet toward the horizon. In fact it is just the relativistic generalization of the usual repulsive centrifugal barrier. where the direction of the force changes.1 as a function of the Schwarzschild coordinate r.10) → ∞ the maximum occurs at rmax ( → ∞) = 3M G.0.5) + } 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 .26 Black Holes. However as the horizon is approached. The maximum of the potential.0. depends weakly on the angular momentum .0. One can see that the points rmax ( ) represent unstable circular orbits.0. For r >> 3M G the potential is repulsive. and . It is given by rmax = 3M G For 1 2 1+ 1+ 14 2 + 14 + 9 9 2 ( + 1)2 − 1 2 ( + 1) (2. The same potential governs the motion of massless classical particles.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. and the String Theory Revolution which. Information. m an action I m = 1 2 dt dr∗ ∂ψ m ∂t 2 − ∂ψ m ∂r∗ 2 − V (r∗ ) ψ 2m (2.0.7) (2.9) The potential V is shown in Figure 2.8) = ν2 ψ m (2.0. 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. gravitational attraction wins and the potential becomes attractive.

In the region of large negative r∗ where we approach the horizon. Any particle that starts with vanishing radial velocity in the region r < 3M G will spiral into the horizon. and the field behaves like a free massless Klein– Gordon field.1 Effective potential for free scalar field vs Schwarzschild radial coordinate the innermost such orbit is at r = 3M G. 2. The eigenmodes in this region have the form of plane waves which propagate with unit velocity dr ∗ dt = ∓1 ∗ (2.0.13) . if = 0 the height of the barrier is Vmax = 1 2M 2 G2 3 8 3 (2.0.11) ±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.0. The particle has enough energy to overcome the barrier without tunneling if ν 2 is larger than the maximum height of the barrier.0. Will it pass over the barrier? To answer this we note that equation 2.15 MG (2. For example.9 has the form of a Schrodinger equation for a particle of energy ν 2 in a potential V .12) An s-wave quantum will therefore escape if ν > 0. the potential is unimportant.

17) − ∂χ ∂u 2 − e2u (∂⊥ χ)2 (2. Less energetic particles must tunnel through the barrier.1.1 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 tortoise-like coordinate which again goes to −∞ at the horizon.14) Therefore the threshold energy for passing over the barrier is 1 ν ∼ √ 27 M G (2.0. ω) (2. For large Vmax ≈ 2 1 27 M 2 G2 (2. whether on the inside or outside of the barrier will have more difficulty penetrating through.15) 2.1. Information.0. A particle of high angular momentum.1. 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 (2.16) (2.18) where ∂⊥ χ = (∂X .15 Similarly an s-wave quantum with ν > MG will be able to penetrate the barrier from the outside and fall to the horizon. Instead of using spherical waves. u. and the String Theory Revolution 0.1. ∂Y ).28 Black Holes.19) . near the horizon we can decompose χ into transverse plane waves with transverse wave vector k⊥ χ = d2 k⊥ ei k⊥ x⊥ χ(k⊥ .

there is a potential confining quanta to the region near the horizon. By contrast.21 is seen to be proportional to 2 .23) . in the Rindler case V increases as eu without bound. Qualitatively.24) ∂u2 Once again we see that unless k = 0. the behavior of a quantum field in a black hole background differs from the Rindler space approximation in that for the black hole. the boundary condition would be expected to be that the field is in the usual quantum ground state.1. − (2.1.1. In the next section we will see that this is not so. For very low angular momentum the approximation is not accurate. but qualitatively is correct for > 0. From the action in equation 2. the potential barrier is cut off when ρ = eu is greater than M G.1.1. then a wave with wave vector k⊥ will correspond to an angular momentum | | = |k| r = 2M G|k|. Thus the potential in equation 2.21) The correspondence between the momentum vector k and the angular momentum is given by the usual connection between momentum and angular momentum. u) = k 2 e2u (2.22) 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 (2.1.20 we obtain the equation of motion ∂2χ ∂2χ − + k 2 exp(2u) χ = 0 ∂ω 2 ∂u2 (2. If the horizon has circumference 2π(2M G).20) Thus the potential is V (k.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. We have not thus far paid attention to the boundary conditions at the horizon where u → −∞. the integral should be infrared cut 1 off at |k| ∼ MG .1. In approximating a sum over and m by an integral over k. Since in this region the field χ(u) behaves like a free massless field.

and the String Theory Revolution . Information.30 Black Holes.

In that way we can use the special relativistic laws of nature to understand the effect of a gravitational field. the study of a phenomenon in a gravitational field is best preceeded by a study of the same phenomenon in an accelerated coordinate system. These features are closely associated with the existence of the horizon. the relativistic analogue of a uniformly accelerated frame is Rindler space. The field in Region I of Figure 1.2 can be described in a self contained way. Evidently the Rindler observer sees a world in which physical phenomena can be described in a completely self contained way.Chapter 3 Quantum Fields in Rindler Space According to Einstein. interactions become very important near the horizon of a black hole.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 space-time geometry (Region I) there are new and subtle features to the description of quantum fields. As we shall see. Signals from Region IV can. The method we will use applies to any relativistic quantum field theory including those with nontrivial interactions. 3. Ignoring them leads to an inconsistent description of the Hawking evaporation process. It is important to bare in mind that such a non-interacting description is of limited validity. As we have seen. 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. 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. 31 . For illustrative purposes we will consider a free scalar field theory. reach Region I. of course.

e. i. The Rindler Hamiltonian is HR = dρ dx⊥ ρ 2 Π2 + ∂χ ∂ρ 2 + ∂χ ∂x⊥ 2 + 2 V (χ) (3. In Figure 3.3) The origin of the factor ρ in the Rindler Hamiltonian density is straightforward. Z ) |0 ∼ 1 ∆2 (3.32 Black Holes. 3.2 Entanglement Quantum fields can also be described in a self-contained fashion in Rindler space.2. Our goal is to describe the usual physics of a quantum field in Minkowski space.2) where Π is the canonical momentum conjugate to χ. The Rindler Hamiltonian is similar to the generator of Lorentz boosts from the viewpoint of the Minkowski observer.1. Information. The proper time separation between the surfaces is δτ = ρ δω (3.1. in Rindler space. Using conventional methods the generator of ω-translations is given by ∞ HR = ρ=0 dρ dX dY ρ T 00 (ρ. the correlation between fields at different spatial points does not vanish. but from the viewpoint of the Fidos in Region I. and the String Theory Revolution The evolution from one surface of constant ω to another is governed by the Rindler Hamiltonian.1 the relation between neighboring equal Rindler-time surfaces is shown. for a massive scalar field with potential V . Z) χ(X .1) where T 00 is the usual Hamiltonian density used by the Minkowski observer.5) . in free massless scalar theory the equal time correlator is given by 0| χ(X. Y .1. However it only involves the degrees of freedom in Region I. For example. to push the ω-surface ahead requires a ρ-dependent time translation. To understand the new feature. T 00 is given by T 00 = 1 Π2 2 + (∇χ) + V (χ) 2 2 (3. Y ) (3. This is the reason that T 00 is weighted with the factor ρ. but a new twist is encountered. X. For example.1.4) Thus. recall that in the usual vacuum state. Y.

2. When two subsystems (fields in Regions I and III) become correlated. Y. we say that they are quantum entangled.6) The two points might both lie within Region I.5 represents the quantum correlation seen by Fido’s in Region I. . the two points might lie on opposite sides of the horizon at Z = 0. The appropriate description of an entangled subsystem is in terms of a density matrix. 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. In that case the correlation is unmeasurable to the Fidos in Region I. in which case the correlator in equation 3.1 Equal time and proper distance surfaces in Rindler space where ∆ is the space-like separation between the points (X. 3. Nevertheless it has significance. Z ) ∆2 = (X − X )2 + (Y − Y )2 + (Z − Z )2 (3.Quantum Fields in Rindler Space 33 ω= T . Y . On the other hand.2. Z) and (X . so that neither can be described in terms of pure states.

. (3.11) The eigenvalues ρj can be considered to be probabilities that the system is in the j th state. .. This is the case in which only one eigenvalue ρj is .3.3.. The combined system has a wavefunction Ψ = Ψ(α.3. β) (3. β ).. ....   ρ =   0 0 ρ3 .. α ) = β Ψ∗ (α. There is one special case when the density matrix is indistinguishable from a pure state. which have previously been in contact but are no longer interacting.. A and B. .7) where α and β are appropriate commuting variables for the subsystems A and B...  0 ρ2 0 . . unlike the case of a coherent superposition of states.. experiments performed on B are described by ρB (β.. β ) (3. . Now suppose we are only interested in subsystem A. ρA (α.3 Review of the Density Matrix Suppose a system consists of two subsystems. ρB (β.3. However..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 . and the String Theory Revolution 3..10) (3.  . α ). β ) = α Ψ∗ (α.34 Black Holes. A complete description of all measurements of A is provided by the density matrix ρA (α..8) Similarly... Information... β) (3.. β) Ψ(α. β) Ψ(α . . the relative phases between the states |j are random. ..3.

3. T re−β H (3. We may think of eS as an effective dimensionality of the subspace described by ρ. logρM. This case can only result from an uncorrelated product wave function of the form Ψ(α. we find S = log n (3.B. if ρ is a projection operator onto a subspace of dimension n. The Von Neumann (or entanglement) entropy should not be confused with the thermal entropy of the second law of thermodynamics.3. In that case S takes its maximum value Smax = − j 1 1 log = log N N N (3.B. The one nonvanishing eigenvalue is equal to 1 by virtue of the trace condition on ρ. It is therefore called the entropy of entanglement.Quantum Fields in Rindler Space 35 nonzero.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.B. This entropy has its origin in coarse graining.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 .3. The opposite extreme to a pure state is a completely incoherent den1 sity matrix in which all the eigenvalues are equal to N . β) = ψA (α) ψB (β) (3.13) S is zero if and only if all the eigenvalues but one are zero.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. (3. (3.17) .14) More generally.3.16) In this case the thermal entropy is given by Sthermal = −T r ρM. where N is the dimensionality of the Hilbert space. The entropy is also a measure of the degree of entanglement between A and B.3. = e−β H .

4. In particular we would like to understand the density matrix ρR which represents the usual Minkowski vacuum to the Fidos in Region I. However. Y.36 Black Holes. Information. Z) = χR (X. the translation invariance along the Z axis is explicitly broken by the act of singling out the origin Z = 0 for .4. Thus the Fidos must see the vacuum as invariant under translations along the X and Y axes. These fields may be decomposed into two subsets associated with regions I and III. Z) = χL (X. Z) Z < 0 The general wave functional of the system is a functional of χL and χR Ψ = Ψ(χL . Thus χ(X. 3. χR ) (3. For the case of a scalar field χ the fields at each point III χL χR I Fig. Y. First let us see what we can learn from general principles. Obviously the state Ψ is translationally invariant under the usual Minkowski space translations. one in Region I and one in Region III. Y. and the String Theory Revolution 3. In Figure 3. Z) Z > 0 (3.19) We wish to compute the density matrix used by the Fidos in Region I to describe their Rindler world.2 the surface T = 0 of Minkowski space is shown divided into two halves.18) χ(X.4 The Unruh Density Matrix Now let us consider the space of states describing a Lorentz invariant quantum field theory in Minkowski space. Y. we call them χR and χL respectively.2 Fields on the spacelike surface T = 0 in Minkowski space of space form a complete set of commuting operators.

χR ).21) To proceed further we must use the fact that Ψ(χL . The field χ(x) is constrained to equal (χL . 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. χR ) = √ Z dχ(x) e−IE (3. ρR ] = 0 (3. From the X. χR ) on the surface X 0 = 0.4. 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. The boost invariance of the original Minkowski action insures that the Euclidean action has four dimensional rotation invariance.23) Letting T → i X 0 we obtain the Euclidean action IE = d4 X 1 (∂X χ)2 − 2 V (χ) 2 (3.25) where the path integral is over all χ(x) with X 0 > 0 and Z is an appropriate normalization factor . the invariance under ω-translations becomes invariance under rotations in the Euclidean (Z.Quantum Fields in Rindler Space 37 special consideration. Let us assume that the field theory is described in terms of an action I = d3 X dT L (3. In particular.4. For example. ρR ] = 0 (3. ρR ] = [pY . Y translation invariance we conclude that ρR commutes with the components of momentum in these directions [pX . Finally the action IE is evaluated as an integral over the portion of Minkowski space with X 0 > 0.24) Now a standard method of computing the ground state by path integration is to use the Feynman–Hellman theorem.4. Suppose we wish to compute Ψ(χL .4. Thus [HR . X 0 ) plane.22) The so-called Euclidean field theory is defined by replacing the time coordinate T by i X 0 .4.4. This follows from the Lorentz boost invariance of Ψ. χR ) is the ground state of the Minkowski Hamiltonian. Then we consider the path integral 1 Ψ(χL .20) A very important property of ρR is that it is invariant under Rindler time translations ω → ω + constant.

3. (3. The process can be iterated until the entire region X 0 > 0 has been covered. T δθ θ=0 χL Horizon ρ=0 χR Z Fig. According to .26) π δω To compute the full path integral we raise the matrix G to the power giving 1 Ψ(χL . Let us define the Euclidean angle in the (Z. Now let us divide the region X 0 > 0 into infinitesimal angular wedges as shown in Figure 3. the path integral defining Ψ is computed as a transition matrix element between initial state χR and final state χL . Information. The angle θ is the Euclidean analogue of the Rindler time ω.38 Black Holes. The infinitesimal generator which pushes θ surfaces forward is just the Rindler Hamiltonian.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 integral over the first wedge is defined by constaining the fields at θ = 0 and θ = δθ. Now we are prepared to compute the density matrix ρR . 3. X 0 ) plane to be θ.4.4. and the String Theory Revolution path integral. This defines a transfer matrix G in the Hilbert space of the field configuration χR .27) In other words. χR ) = √ χL | e−π HR |χR Z (3. The matrix is recognized to be G = (1 − δθ HR ) .

We can consider a thermometer to be a localized object with a set of proper energy levels i . quantum electrodynamics. The origin of the dimensionless temperature lies in the dimensionless character of the Rindler time variable ω. χR ) Ψ(χL . 3.3.Quantum Fields in Rindler Space 39 the definition in equation 3.4. inverse length.31) The derivation of the thermal character of the density matrix and the value of the Rindler temperature in equation 3. It is equally correct for a free scalar quantum field. The Rindler energy .4.4. The levels i are the ordinary energy levels of the thermometer when it is at rest.28) Now using equation 3. The thermometer is assumed to be very weakly coupled to the quantum fields so that it eventually will come to thermal equilibrium with them. χR ) dχL (3.5 Proper Temperature It is noteworthy that the temperature TR is dimensionless. the density matrix ρR is given by ρR (χR . χR ) = 1 Z χR | e−π HR |χL χL | e−π HR |χR dχL (3.8. the density matrix is given by the operator ρR = 1 exp(−2 π HR ) Z (3.30) This remarkable result. 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.4.27 we get ρR (χR . says that the Fidos see the vacuum as a thermal ensemble with a density matrix of the Maxwell–Boltzmann type. or equivalently. or quantum chromodynamics. The temperature of the ensemble is TR = 1 1 = 2π βR (3.4.31 is entirely independent of the particulars of the relativistic field theory. discovered by William Unruh in 1976 . χR ) = Ψ∗ (χL . Ordinarily. temperature has units of energy.4. 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 .29) = 1 Z χR | e −2 π HR |χR In other words.

5.1. 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. . 1 When the quantum field at Rindler temperature 2π equilibrates with the thermometer. Information. One virtual loop (a) is contained entirely in Region I.5.32) In other words the Rindler energy level of the ith state of the thermometer is ρ i .33) Accordingly. It is helpful in visualizing these effects to describe virtual vacuum fluctuations as short lived particle pairs. calling the acceleration a.40 Black Holes. 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.4 ordinary vacuum fluctuations are shown superimposed on a Rindler coordinate mesh. the thermometer registers a proper temperature T (ρ) = 1 1 = TR 2π ρ ρ (3.35) 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 proper temperature T (ρ) can also be expressed in terms of the proper 1 acceleration of the Fido which is equal to ρ .5. and the String Theory Revolution in equation 3. Thus.1 evidently receives a contribution from the thermometer of the form HR (thermometer) = i ρ |i i| i (3.5. we find T (ρ) = a(ρ) 2π (3. 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. These are the fluctuations which lead to nontrivial entanglements between the degrees of freedom χL and χR . but now being experienced by accelerated apparatuses. In Figure 3.34) Thus each Fido experiences a thermal environment characterized by a temperature which increases as we move toward the horizon at ρ = 0. That fluctuation can be thought of as a conventional fluctuation described by the quantum Hamiltonian HR .

Thus. 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. If the virtual fluctuation of energy needed to produce the pair is E. then the lifetime of the fluctuation ∼ E −1 . the fluctuation lasts for an infinite time and is therefore not virtual at all. according to the Fidos. 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. Now consider the portion of the loop (b) which is found in Region I. To state it differently. Real particles are seen being injected into the Rindler space from the horizon. and eventually fall back to the horizon.4 Vacuum pair fluctuations near the horizon A virtual fluctuation is usually considered to be short lived because it “violates energy conservation”.Quantum Fields in Rindler Space 41 ω= 8 (a) (b) (c) ω=− Fig. A natural question to ask is whether the thermal effects are “real”. From the viewpoint of the Fidos. 8 . a particle is injected into the system at ω = −∞ and ρ = 0. the horizon behaves like a hot membrane radiating and reabsorbing thermal energy. 3. A thermometer at rest in an inertial frame in the Minkowski vacuum will record zero temperature. The particle travels to some distance and then falls back towards ρ = 0 and ω = +∞. Obviously the answer is no. For example.

By contrast. although it is completely unseen by observers who fall freely through it. it is best to avoid the metaphysical question of whose description is closer to reality. a Frefo carrying similar apparatuses will see only the zero temperature vacuum state. We will see later that the stretched horizon has many other physical properties besides temperature. . We see that the correct condition must be that at some small distance ρo . Information. or other apparatus. 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. 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. will discover all the physical 1 phenomena associated with a local proper temperature T (ρ) = 2π ρ . Since Rindler space has a boundary at ρ = 0. and the String Theory Revolution However. Thus for the moment.42 Black Holes. particle detector. We can now state the sense in which a self contained description of Rindler space is possible in ordinary quantum field theory. 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. In particular a Fido equipped with a standard thermometer. by yielding to this temptation we risk prejudicing ourselves too much toward the viewpoint of the Frefos. a boundary condition of some sort must be provided. 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. an effective “membrane” is kept at a fixed temperature T (ρo ) = 2π1ρo by an infinite heat reservoir. In particular. Such a fictitious membrane at Planckian distance from the horizon is called the stretched horizon.

− ∂ 2 χk + k 2 e2u χk = λ2 χk ∂u2 (4.18. and neutrinos.0. a wide variety of different phenomena take place at different temperature scales. the Planck temperature is reached where totally new phenomena of an as yet unimagined kind take place. 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 space-time background. The free gas is replaced by a plasma. We consider the field theory defined by equation 2. Nevertheless. up the scale of energies.1. the study of a free quantum field in Rindler space is a useful starting point. At the lowest temperatures where only massless quanta are produced by thermal fluctuations. photons. There is serious danger in extrapolating far reaching conclusions from so oversimplified a situation. Finally. and statistical thermodynamics that these lectures are concerned with are largely a consequence of such unjustified extrapolation. Fourier decomposing the field χ in equation 2.19 leads to the wave equation in equation 2. In fact.1) In order to quantize the field χ it is necessary to provide a boundary condition when u → −∞.24. and so it goes. At higher temperatures. Increase the temperature to the e+ . All of these phenomena have their place in the Fido’s description of the region near a horizon. the paradoxes and contradictions associated with black holes. quantum mechanics. e− threshold and electron-positron pairs are produced. 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.Chapter 4 Entropy of the Free Quantum Field in Rindler Space In the real world. pions are produced which eventually dissociate into quarks and gluons.1.1. one expects to find a very weakly interacting gas of gravitons. The parameter represents the proper 43 .

However it will prove interesting to separate physical quantities into those which are sensitive to and those which are not. k) N (n. k) 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. k) has frequency λ(n. The total length of the box depends on k and according to L(k) = −log( k) (4. k) a(n.k (u) (4. Those things which depend on are sensitive to the behavior of the physical theory at temperatures of order 1 2π and greater. k) fn.0. Each transverse Fourier mode χk can be thought of as a free 1+1 dimensional quantum field confined to a box.0. It is by no means obvious that a reflecting boundary condition very near the horizon is a physically reasonable way to regularize the theory.0. is that .3) where the mode (n.0. k) (4. k) (4.44 Black Holes. Information. Physically we are introducing a perfectly reflecting mirror just outside the horizon at a distance . Thus we may approximate the potential by a second wall at u = u1 = −log k. One end of the box is at the reflecting boundary at u = uo = log . The new and unusual feature of Rindler quantization. k) = a† (n. k) fn. The Rindler Hamiltonian is given by HR = d2 k = where N (n. and the String Theory Revolution distance of the cutoff point to the horizon.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). Later we will remove the cutoff by allowing uo → −∞. encountered in Chapter 3.5) n λ(n. k) Thus far the quantization rules are quite conventional.k (u) + a− (n.4) n λ(n. k) a(n.

u ) = T r ρ χ(X. is not a translationally invariant system. u) χ(X . Y . As an example of insensitive quantities. k)] − 1 (4. Y . a careful computation of the expectation value of T µν in this state reveals a singular behavior at the horizon. The reader might wonder what goes wrong if we choose the state which is annihilated by the a’s. on the other hand. This however would clearly violate the fourth guiding principle stated in the introduction: To a freely falling observer. k) a(n.6) with ρR (n.8) (4.0.9) are found to have smooth limits as → 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 ρR (n.0. k) (4. k) ∼ exp −2πλ(n.0. Y. Y.0. the horizon of a black hole should in no way appear special.7) These particles constitute the thermal atmosphere. Y. . A black hole. Therefore such quantities can be said to decouple from the degrees of freedom within a distance of the horizon.Entropy of the Free Quantum Field in Rindler Space 45 we do not identifiy the vacuum with the state annihilated by the a(n. Y . k) a† (n. the field correlation functions such as χ(X. (4. u ) are kept away from the horizon. Moreover. 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. Physical quantities in Rindler space can be divided into those which are sensitive to the cutoff at and those which are not. u ). u) χ(X . Certainly this is not a good candidate to represent the original Minkowski vacuum. k) Thus the average occupation number of each mode is N (n. In fact. as long as the points (X. Such a state is not at all invariant under translations of the original Minkowski coordinates Z and T . but rather with the thermal density matrix ρR = n. u) and (X . k) = 1 exp[2πλ(n. k). A much more singular quantity which will be of great concern in future lectures is the entropy of the vacuum state.

16 and 3. .4.13 and 4.0.0.10) ∂ N ρ ∂N (4.0. Y plane by a finite torus with periodic boundary conditions. Information.28 is to produce the thermal density matrix in equation 3. and the String Theory Revolution Since the relevant density matrix has the Maxwell–Boltzmann form.4.14 can be thought of as both entanglement and thermal entropy in the special case of the Rindler space density matrix.13) (4. This is because the effect of integrating over the fields χL in equation 3.0.17 to obtain the entropy. To compute the total entropy we begin by replacing the infinite transverse X.0.14) The entropy S in equations 4.11) N =1 = +T r βH + lnZ (4.12) = 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 (4.3. 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. This has the effect of discretizing the values of k.0. kY = 2nY π B (4.0.0. Thus the computation of the entropy of Rindler space is reduced to ordinary thermodynamic methods. we can use equations 3.30. 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.15) where B is the size of the torus.46 Black Holes.3. Thus kX = 2nX π B .

2 for (4. Substituting T = the length L gives the entropy of χk S(k) = 1 |log k | 6 1 2π (4.0.0.20) Now consider the entropy stored in a layer of thickness δρ and area B 2 at . 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. One might have expected it to diverge as the volume of space. The entropy is stored in the vicinity of the stretched horizon and therefore grows only like the area.0.15 and let B → ∞ ST otal = B2 24π 2 d2 k |log k | (4. We find that S is approximately given by ST otal ≈ 1 B2 96π 2 2 (4. but this is not the case.19 we see two important features of the entropy of Rindler space. This is because when k = 1 the potential is already large at u = uo so that the entire contribution of χk is supressed.19) From equation 4. 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 .0.0.18) In evaluating equation 4.17) To sum over the values of k we use equation 4.0. the integral must be cut off when k > 1 . 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. A S standard calculation gives the entropy density L to be given by S π = T L 3 where T is the temperature.18.0.16) and equation 4. B 2 .0. As we shall see.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 . The first is that it is proportional to the transverse area of the horizon. The second feature which should alarm us is that the entropy per unit area diverges like 1 2 .

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 (4.1. Information. 4.1.22) = Now we see that the entropy is mainly found near the horizon because that is where the temperature gets large. However.0.0.24) The implication of this fact is that the temperature of the thermal atmosphere is reckoned to be red shifted also.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.7.22 by observing two facts: 1) The Rindler time ω is related to the Schwarzschild time t by the equation ω = t 4M G (4. A good qualitative understanding of the process can be obtained from the Rindler quantum field theory in equation 2. Thus the temperature as . Unlike the Rindler case.1. the thermal atmosphere is not absolutely confined by the centrifugal potential in equation 2.48 Black Holes. The particles of the thermal atmosphere will gradually leak through the barrier and carry off energy in the form of thermal radiation. and the String Theory Revolution a distance ρ from the horizon δS(ρ) = = 2π 2 45 T 3 (ρ) δρ B 2 (4.0. in the case of a black hole a new phenomenon can take place – evaporation.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.

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. In particular the maximum value that V takes on for angular momentum zero is 1 27 (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 × = . At the end of the next lecture.25) 2π 4M G 8πM G a form first calculated by Stephen Hawking.1. and ultimately lead to absurd results. .1. 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.25 indicates that some of the s-wave particles will easily escape to infinity. Indeed it only makes sense if there are interactions of sufficient strength to keep the system in equilibrium during the course of the evaporation. equation 4.1. Since the average energies of massless particles in thermal equilibrium at temperature T is of course of order T . The above description of Hawking radiation does not depend in any essential way on the free field approximation. A small fraction of the particles carry very low angular momenta. In fact. we will discuss one such absurdity. 1/2 √ Particles of angular momenta higher than s-waves cannot easily escape because the potential barrier is higher than the thermal scale. For these particles. most discussions of Hawking radiation rely in an essential way on the free field approximation. the walls are semi-transparent and the cavity slowly radiates its energy. (4.1. Unless the black hole is kept in equilibrium by incoming radiation it will lose energy to its surroundings.26) Vmax ( = 0) = 2 G2 1024 M T = 3 3 Any s-wave quanta with frequencies of order (Vmax ) = 32MG or greater will easily escape the barrier.

50 Black Holes. and the String Theory Revolution . Information.

0.1 dM = from which we deduce S = 4πM 2 G (5. This. It follows thermodynamically that it must also have an entropy.2) where E. Evidently something cuts off the modes which are very close to the horizon. Presumably this entropy is part of the entropy of the black hole. is replaced by M . 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.4) This is the famous Bekenstein–Hawking entropy. as we have seen.0. but unfortunately it is infinite as → 0.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. Using equation 5.0.0.0. To find the entropy we use the first law of thermodynamics in the form dE = T dS (5.3) 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. the black hole energy. It is gratifying that it is proportional to the area of the horizon. is where all the infalling matter accumulates according to external observers. To get an 51 .1) and energy M .

ω] = i (5. We assume that for a large black hole the Rindler energy is a function of the mass of the black hole. Another way to express this is through the level density of the black hole dN ∼ exp 4πM 2 G dM (5.52 Black Holes.5) G 15 (5. = exp 4πM 2 G .8) Here ER is the Rindler energy which is of course the eigenvalue of the Rindler Hamiltonian. 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 local geometry of a limiting black hole horizon is of course Rindler space. Information.19 not exceed the entropy of the black hole SBH 1 96π 2 or √ > ∼ < 2 ∼ 1 4G (5. where the Planck length is given in terms of Newton’s constant as P = c3 G.6) In other words. 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. Let us consider the Rindler energy of the horizon.0.H. This suggests that we can understand the entropy in terms of the local properties of a limiting black hole of infinite mass and area.0. The entropy of a large black hole is an extensive quantity in the sense that it is proportional to the horizon area.7) where dN is the number of distinct quantum states with mass M in the interval dM . the cutoff must not be much smaller than the Planck length.0.0. but the entropy per unit area is finite. Accordingly we write [ER (M ). and the String Theory Revolution idea of where the cut off must occur. we can require that the contribution in equation 4. The entropy diverges.0. This is of course not surprising. The number of such states is of order exp SB. By definition it is conjugate to the Rindler time ω.

0. The s-wave quanta are described 1 in terms of a 1+1 dimensional quantum field at Rindler temperature 2π .0.10) Finally. For simplicity.0. the stretched horizon also radiates like a black body. the barrier height for the s-wave quanta is comparable . In the same units. the conjugate character of M and t allows us to write equation 5. We first recall that only the very low angular momentum quanta can escape the barrier.13) The Rindler energy and entropy satisfy the first law of thermodynamics dER = (5. t] = i Now use ω = t 4MG (5.14) 1 where 2π is the Rindler temperature. while the Rindler energy is the (dimensionless) energy as defined by observers near the horizon using ω-clocks.10 in the form ∂ER = 4M G ∂M and ER = 2M 2 G (5.0.11) The Rindler energy and the Schwarzschild mass are both just the energy of the black hole. suppose that only the s-wave quanta get out.Thermodynamics of Black Holes 53 The mass and Schwarzschild time are also conjugate [M.0.12) (5. The Schwarzschild mass is the energy as reckoned by observers at infinity using t-clocks. As we have seen. The area density of Rindler energy is 1 ER = A 8πG 1 dS 2π (5.0.0. The exact rate of evaporation of the black hole is sensitive to many details.9) to obtain ER (M ). t] = 4M G i (5. t 4MG = i or [ER (M ). It is of interest that the Rindler energy is also extensive. but it can easily be estimated. 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.

the luminosity.0. . In terms of the Schwarzschild time.0. We call this L. It is therefore not really constant. unlike the image of the sun.15 gives equation 5.0. When the mass of the black hole is large and the temperature low. Evidently energy conservation requires the black hole to lose mass at just this rate dM C = −L = − 2 2 (5.0. It follows that approximately one quantum per unit Rindler time will excape.54 Black Holes.0. If we ignore the mass dependence of C. Furthermore each quantum carries an energy at 1 infinity of order the Schwarzschild temperature 8πMG .15 is essentially the Stephan–Boltzmann law L ∼ T 4 · Area (5.15 can be integrated to find the time that the black hole survives before evaporating to zero mass. 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.15) dt M G where C is a constant of order unity.17. only a few species of massless particles contribute and C is constant. Observing a black hole by means of its Hawking radiation will always produce a fuzzy image. which is about equal to the Schwarzschild radius. In that case the temperature and size of the system are related in an entirely different way. Information.0. while the radius of the surface of the sun is ∼ 1011 cm. equation 5. This evaporation time is evidently of order tevaporation ∼ M 3 G2 (5. and the String Theory Revolution to the temperature. The sun is for all intents and purposes infinite on the scale of the emitted photon wavelengths.0. However the physics is very different from that of a radiating star.17) 1 Using T ∼ MG and Area ∼ M 2 G2 in equation 5.16) It is interesting that luminosity in equation 5. The resulting rate 1 of energy loss is of order M 2 G2 . the flux of 1 quanta is of order MG . The black hole on the other hand emits quanta of wavelength 1 ∼ T ∼ M G. The typical wavelength of a photon radiated from the sun is ∼ 10−5 cm.

Typically black strings and branes are studied as systems of infinite extent. In this lecture we will describe the main facts about charged black holes.0. Generally the horizon occurs at r ∼ M G. Thus.1) The electric field is given by the familiar Coulomb law Q Er = r2 Eθ. The most important fact about them is that they cannot evaporate away completely. Pair production is exponentially e 55 .0. By going to higher dimensions we can consider not only black holes. black membranes. They have ground states with very special and simplifying features. and therefore have infinite entropy. where me is the electron mass.Chapter 6 Charged Black Holes There are a variety of ways to generalize the conventional Schwarzschild black hole. which will discharge the black hole in the same manner as a nucleus with Z >> 137 is discharged. but black strings. let us consider electrically charged black holes. For this reason they can store infinite amounts of information.φ = 0 (6. 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. Higher dimensional Schwarzschild black holes are quite similar to their four-dimensional counterparts. it will cause pair production of electrons.2) If the electric field is too strong at the horizon. and the threshold field for unsupressed pair production is E ∼ m2 . Another way to generalize the ordinary black hole is to allow it to carry gauge charge and/or angular momentum. and so forth.

0.6 .6) Note that in the extremal limit M 2 = Q the inner and outer horizons G merge at r± = M G.5) (6. Such “naked singularities” indicate a breakdown of classical relativity visible to a distant observer. Q e For Q2 > M 2 G the metric in equation 6.56 Black Holes. Using equation 6. The question is whether they can be described by classical general relativity. an outer one and an inner one.0. A Reissner–Nordstrom G Q G black hole that saturates this relationship M 2 = black hole.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. For practical purposes we regard them as stable. Clearly they cannot. Thus equation 6.0. To examine the geometry near the outer horizon. Clearly they can. The Reissner–Nordstrom solution has two horizons.3) or alternativley M >> m21G2 . Information. let us begin by computing the distance from r+ to an arbitrary point r > r+ . The electron is such an object. The question is not whether objects with Q2 > M 2 G can exist.0.3 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. and the String Theory Revolution suppressed if Q << m2 e M 2 G2 2 (6. Accordingly we restrict 2 2 our attention to the case M 2 > Q or M > Q .0. They are defined by 1− Q2 G 2M G + 2 r± r± = 0 (6.1 has a time-like singularity with no horizon to cloak it.

11) Note that the proper distance becomes infinite for extremal black holes.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. Since r+ ∼ M G the + .0.0.0.8) dy (6.0.0.0.9) −1 2 ∆y = y2 − + cosh The radial-time metric is given by ds = − 2 y2 − y+ ∆2 4 Σ 2 2 dt2 + dρ2 (6.12) ∼ ρ2 dω 2 − dρ2 = where ω ≡ ∆ 2 t 2r+ (6. The charge density on the horizon is 4πr2 .9 near the horizon r+ one finds ρ ≈ y − ∆ 2 1/2 2r+ ∆1/2 .0.10) Expanding equation 6.0.13) Evidently the horizon geometry is again well approximated by Rindler Q space.10 becomes ds2 ∼ = ∆2 4 4r+ ρ2 dt2 − dρ2 (6. For non-extremal black holes. (6.7) (6.0. equation 6.

0.0.0. and the String Theory Revolution Q charge density is ∼ 4πM 2 G2 . is approximately Rindler.14) 2 ∼ 4πM 2 G2 = 4πr+ 4πM G3/2 For very massive black holes the charge density becomes vanishingly small.0. From equation 6. although infinitely far from any fiducial observer with r = r+ . the charge density is of the form √ M G ∼ 1 Q (6.13 we can compute the temperature as seen at infinity. Thus for near extremal black holes. Therefore the local properties of the horizon cannot be distinguished from those of a Schwarzschild black hole.19) which.16) 1− Q2 M2G 1− Q2 M2G 2 (6.9 we see that as ∆ → 0 ρ → y + Σ log(2y) − log∆ 2 (6.15) (6. From equation 6. In particular.17) As the black hole tends to extremality. .0. 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 (6.0.18) Thus for a fiducial observer at a fixed value of r the horizon recedes to infinite proper distance as ∆ → 0. Information.0. the horizon becomes progressively more removed from any fiducial observer. the temperature at a 1 small distance ρo from the horizon is 2πρo .58 Black Holes.0. 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.

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 near-horizon 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

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. and j is a conserved current in the usual sense ∂µ j µ = 0. Instead of being light-like.62 Black Holes. or even nonexistent. To see that the horizon of a black hole has electrical properties. substituting the form of the metric ˙  1  A + ∇φ   8π ρ   2 (7. Stretching the horizon has another benefit.0. First let us define the stretched horizon.1) (7. the stretched horizon has dynamics of its own that includes such phenomena as viscosity and electrical conductivity. Information.0.0. 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. or “stretched” horizon at a distance of roughtly one Planck length from the mathematical horizon. a system at the stretched horizon is time-like. 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.3)  − ρ ∇×A 2  W =   + j · A dω dρ d2 x⊥ (7. it is sufficient to study electrodynamics in Rindler space. As we will see.0. This means that real dynamics and evolution can take place on the stretched horizon.2) (7.4) ˙ ∂A where A means ∂ω and φ = −A0 . As usual ˙ E = −∇φ − A B = ∇×A . This leads to the idea that the mathematical horizon should be replaced by an effective membrane.

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 (7.0.The Stretched Horizon 63 With these definitions. Thus 2 ∂ρ φ − 1 ∂ρ φ = −∇2 φ ⊥ ρ (7. or more precisely σ = =− 1 4πρ 1 4πρ Eρ ∂ρ φ ρ=ρo (7. The surface charge density on the stretched horizon is easily defined.5) ˙ E − ∇ × (ρB) = −4πj ˙ B + ∇×E = 0 (7.0.6 becomes ∇· 1 E ρ = −∇ · 1 ∇φ ρ = 0 (7. the distance of the charges from the stretched horizon is macroscopic.0.6) ∇· 1 ρ E = 4πj 0 ∇·B = 0 We begin by considering electrostatics.0. Since the charges are slowly moving in Rindler coordinates. By electrostatics we mean the study of fields due to stationary or slowly moving charges placed outside the horizon. In particular.0.7) ρ=ρo Working in the Coulomb gauge. the third expression in equation 7. it means that they are experiencing proper acceleration.0.8) near the stretched horizon. We also assume all length scales associated with the charges are much larger than ρo . It is just the component of the electric field perpendicular to the stretched horizon.9) .

14) 4π jy = ρ Bx Now let us consider an electromagnetic wave propagating toward the stretched horizon along the ρ axis.13) ρ Evidently this is a continuity equation if we define: 4π jx = −ρ By (7. equation 7. Since the black hole horizon is compact.0. We can easily identify the surface current density.0.0.0.0. and the String Theory Revolution We can solve this equation near the horizon by the ansatz φ ∼ ρα .0.7 and using Maxwell’s equations 7.12 proves that φ = constant on the horizon.64 Black Holes. From Maxwell’s equations we obtain ˙ Bx = ∂ρ Ey ˙ By = −∂ρ Ex 1 ˙ ρ Ex 1 ˙ ρ Ey (7. Thus we assume φ = F (x⊥ ) + ρ2 G(x⊥ ) + terms higher order in ρ (7.15) = −∂ρ (ρ By ) = ∂ρ (ρ Bx ) .12) A similar equation can also be derived for the finite mass black hole.9 and evaluating at ρ = ρo gives ∇2 F + ρ2 ∇2 G = 0 ⊥ o ⊥ (7.6 gives 4π σ = ˙ 1 ˙ Eρ = ρo ∇ × ρB (7. This is an interesting result.10 into equation 7.11 is simplified to ∇2 F = 0 ⊥ (7.10) Plugging equation 7.0. Information. We easily find that α = 0 or α = 2. which proves that the horizon behaves like an electrical conductor.0. then equation 7.0. The right hand side will be smaller than the left hand side by 2 powers of ρ and can therefore be ignored.11) If ρo is much smaller than all other length scales. Taking the time derivative of equation 7.0.0.0.

∗ For example.0.1.0. .0. Since the horizon is just ordinary flat space.The Stretched Horizon 65 To make the equation more familiar.15 then becomes ˙ βx = ∂u Ey ˙ βy = −∂u Ex (7. we can redefine the magnetic field ρB = β and use tortoise coordinates u = log ρ Equation 7. the Maxwell equations 7. one might conclude that the point charge just asymptotically approaches the horizon ∗ The unit Ω/square is not a misprint.18) βy = −Ex or from equation 7.16) Ex Ey Evidently the horizon is an ohmic conductor with a resistivity of 4π. if a circuit is constructed as in Figure 7. However the physics only makes sense for waves propagating toward the horizon from outside the black hole.0. That corresponds to a surface resistance of 377Ω/square. The resistance of a two-dimensional resistor is scale invariant and only depends on the shape. For such waves.17) ˙ Ex = ∂u βy ˙ Ey = −∂u βx The mathematical equations allow solutions in which the wave propagates in either direction along the u-axis. As a last example let us consider dropping a charged point particle into the horizon.17 give βx = Ey (7.14 jx = jy = 1 4π 1 4π (7. a current will flow precisely as if the horizon were a conducting surface.0.0.

ammeter attached to horizon with the transverse charge density remaining point-like.0. we can take the charge to be at rest at position zo in Minkowski coordinates.1 Battery. The calculation is easy because at any given time the Rindler coordinates are related to the Minkowski coordinates by a boost along the z-axis.66 Black Holes.21) .0. 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. Since the component of electric field along the boost direction is invariant. This process is shown in Figure 7.20) Now let’s consider the surface density for large Rindler time. Information. 7. Without loss of generality. and the String Theory Revolution A A Horizon Fig.0. we need to determine the field component Eρ . To compute the surface charge density on the stretched horizon. However this is not at all what happens on the stretched horizon.2. σ = e ρo e ω 4πρo [ρ2 e2ω + x2 ]3/2 o ⊥ (7.19) 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.

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. 7.0. For a real black hole. it is convenient to rescale x⊥ using x⊥ = eω y⊥ to obtain σ = e−2ω e 4π (ρ2 + y 2 )3/2 o ⊥ (7. 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 ) .2 Charge falling past the stretched horizon To better understand this expression.

68 Black Holes. it exhibits dissipative effects such as electrical resistivity and viscosity. and conservation of charge will cause it to spread exponentially. Taking the divergence gives ∇ · j ∼ ∇ · E ∼ σ. and energy. In addition to temperature. Information. Now use the continuity equation get the relation σ ∼ −σ. To see this we use Ohm’s law j = conductivity E. Thus we see that the horizon has the properties of a more or less conventional hot conducting membrane. The surprising and puzzling thing is that they are completely unnoticed by a freely falling observer who falls through the horizon! . Evidently the surface charge density will exponentially ˙ decrease. and the String Theory Revolution The exponential spreading of the charge is characteristic of an Ohm’s law conductor. entropy.

Now we let the system evolve. The region Γ(0) = Γ evolves into the region Γ(t). which indeed grows.1. In classical physics the principle is embodied in Liouville’s theorem: the conservation of phase space volume. the conservation of information is expressed as 69 . These laws are: 1) The principle of information conservation 2) The equivalence principle 3) The quantum xerox principle 8. information is lost because for most cases of interest the region Γ becomes very complicated like a fractal. we want to review three fundamental laws of nature whose compatibility has been challenged. we might represent this by specifying an initial region Γ(0) in the system’s phase space.Chapter 8 The Laws of Nature In this chapter. and if we coarse grain the phase space.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 we begin following a system with some limited knowledge of its exact state. This is illustrated in Figure 8. Liouville’s theorem tells us that the volume of Γ(t) is exactly the same as that of Γ. This is the origin of the second law of thermodynamics. The region Γ(0) has a volume VΓ in the phase space. In this sense the amount of information is conserved. if one takes every point in the phase space and surrounds it by solid spheres of fixed volume. As a definition of coarse graining. it will appear that Γ is growing. In quantum mechanics. In a practical sense.

then S = log VΓ . isolated system. we instead specify a probability density ρ(p.1 Evolution of a fixed volume in phase space the unitarity of the S-matrix. instead of a definite state. for quantum mechanics the sharp projector P can be replaced by a density matrix ρ.1. In this case the fine grained or Von Neumann entropy is S = −T r ρ log ρ (8. A more refined definition of information is provided by the concept of entropy. . and the String Theory Revolution Fig. Suppose that instead of specifying a region Γ in phase space. If we again approach a system with limited knowledge. P . Information. In both quantum mechanics and classical mechanics the equations of motion insure the exact conservation of S for a closed.2) For the case ρ = T P P the entropy is log N . q) (8. Thus the entropy is an estir mate of the logarithm of the number of quantum states that make up the initial ensemble. 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. A generalization of the volume is given by the exponential of the entropy VΓ → exp S. q) in phase space. 8.70 Black Holes. q) log ρ(p. we might express this by a projection operator onto a subspace. where S = − dp dq ρ(p.1) 1 It is easy to check that if ρ = VΓ inside Γ and zero outside.1. Similarly.

it involves separating a system into two or more subsystems. This means its eigenvalues are all either positive or zero. If one of the eigenvalues of ρA is equal to 1. all the others must vanish. Entanglement refers to quantum correlations between the system under investigation and a second system.2. 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. The subsystem A(B) is described by some complete set of commuting observables α (β). More precisely. β).5) It follows that all the eigenvalues are between zero and one. The result is an “entanglement entropy” for the subsystems. This only happens if the composite wave function factorizes Ψ(α. (ρA )αα = (ρB )ββ = β Ψ∗ (α.The Laws of Nature 71 8. there is another reason. 1) The density matrix is Hermitian (ρA )αα = (ρA )∗ α α (8. β) Ψ(α . For definiteness. β) Ψ(α. entanglement. In this case the subsystem A is in a pure state. consider ρA .2.6) .2.2. 3) The density matrix is normalized to 1. In quantum mechanics. Consider a composite system composed of 2 subsystems A and B. β) (8.3) ∗ α Ψ (α. Let us consider some properties of the density matrix.2 Entanglement Entropy In classical physics. Consider now the subsystems separately. All measurements performed on A(B) are describable in terms of a density matrix ρA (ρB ). Let us assume that the composite system is in a pure state with wave function Ψ(α. the only reason for introducing a phase space probability is a lack of detailed knowledge of the state. the constituent subsystems are generally not described by pure states. Trρ = 1 (8. but we could equally well focus on ρB . Even in the case of a pure state. β) = ψA (α) ψB (β) (8.4) 2) The density matrix is positive semidefinite.

From the equality of the non-vanishing eigenvalues of ρA and ρB an important property of entanglement entropy follows: SA = −T r ρA log ρA = SB (8. Each subsystem on the average will have energy . The equality of SA and SB is only true if the combined state is pure. Next. β) φ(α ) = λ φ(α) We assume λ = 0. we start with the eigenvalue equation for ρA . Let us suppose the subsystems weakly interact. β ) φ∗ (α ) = λ α Ψ∗ (α.2. let us consider a large system Σ that is composed of many similar small subsystems σi .2. 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 (8.7) Thus we can just refer to the entanglement entropy as SE . To prove this. Then the eigenvalue condition is βα Ψ∗ (α. β) Ψ(α. and the entire system is in a pure state with total energy E. β ) Ψ∗ (α . Now we define a candidate eigenvector of ρB by χ(β ) ≡ Then β α Ψ∗ (α . β ) φ∗ (α ). In that case. β ) χ(β ) Ψ∗ (α.8) . β) Ψ(α. β) φ∗ (α) = λ χ(β) Thus χ(β) is an eigenvector of ρB with eigenvalue λ. Information. and the String Theory Revolution In this case B is also in a pure state. (ρB )ββ χ(β ) = = αα β αβ Ψ∗ (α. Call φ the eigenvector of ρA . β) Ψ(α .72 Black Holes. 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.

The entropy of each subsystem Si as well as the entropy of the whole system Σ given by σΣ . The wave function develops correlations. suppose we start with the subsystems in a product state with no corelations. In this case.2.2. 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. Typically the fine grained entropy of Σ1 is defined as the entanglement entropy S(Σ1 ) of Σ1 with the remaining subsystem Σ − Σ1 . the subsystem entropies become nonzero Si = 0 and the coarse grained entropy also becomes nonzero ST hermal = i Si = 0 However. This will always be less than the coarse grained entropy of Σ1 ST hermal (Σ1 ) > S(Σ1 ) (8. meaning that it now fails to factorize.9) By definition it is additive. Another concept that we can now make precise is the information in a subsystem. Let us consider an arbitrary subsystem Σ1 of Σ which may consist of one. In fact. The information can be defined by I = ST hermal − S (8. To see why. or all of the subsystems σi . many.2. Now the subsystems interact. we will assume that the entire system Σ is in a pure state with vanishing entropy. the fine grained entropy S(Σ1 ) will tend to zero.10) For example. The thermal density matrix maximizes the entropy for a given average energy . and the coarse grained entropy of Σ all vanish.11) . In general large subsystems or the entire system will not be thermal. as Σ1 approaches Σ.The Laws of Nature 73 where Hi is the energy of the subsystem. It is not conserved. The coarse grained entropy is what we usually think of in the context of thermodynamics. the “fine grained” entropy of Σ is exactly conserved and therefore remains zero.

A bit is the entropy of a two state system if nothing is known[2]. we use two facts: S(Σ − Σ1 ) = S(Σ1 ) S(Σ − Σ1 ) ∼ ST hermal (Σ − Σ1 ) = Thus S(Σ1 ) ∼ ST hermal (Σ − Σ1 ) = (8. Thus for Σ1 < 1 Σ 2 S(Σ1 ) ∼ ST hermal (Σ1 ) = I(Σ1 ) ≈ 0 Next consider a subsystem with Σ1 > 1 Σ. The subsystem 1 Σ has about 1 bit of 2 information. 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). How much information does 2 it have? To compute it. Entropy and information are naturally measured in “bits”. where f is the fraction of the total degrees of freedom contained in Σ1 .12) The coarse grained entropy of Σ − Σ1 will be of order (1 − f ) ST hermal (Σ). the information in a small subsystem vanishes.2. However. The numerical value of a bit is log 2. At the opposite extreme the information of the combined system Σ is just its total thermal entropy.74 Black Holes. However for 2 .13) (8. Typically for subsystems less than half the size of Σ the information is smaller than 1 bit. so that the information is the difference between coarse grained and fine grained entropy. Since typical small subsystems have a thermal density matrix.2. Thus. What actually happens is that for subsystems smaller than about 1/2 of the total system. the information is negligible. 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. for Σ1 < 1 Σ the information in Σ1 is essentially zero. Information. this is not so.

entropy of the box increases. When the bomb explodes. photons slowly leak out. Consider a box with perfectly reflecting walls. The entire system Σ consists of the subsystem B that includes everything in the box. begins to increase. no photons have yet escaped.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.15) (8. The result is that the interior and exterior of the box become entangled.2. The entanglement entropy. which is equal for A and B. in this case. so S(A) = 0 at this time. Furthermore. The thermal. Initially the bomb is in its ground state. (8.2.14) . The box has a small hole that allows the thermal radiation to slowly leak out.2. or coarse grained. all of the photons escape the box. The thermal entropy in the box decreases: SEntanglement = 0 ST hermal (B) = 0 ST hermal (A) = 0 Eventually. The subsystem A consists of everyting outside of the box. it fills the box with thermal radiation. and B has vanishing entropy. Inside the box we have a bomb which can explode and fill the box with radiation. its ground state. The thermal or coarse grained entropy as well as the fine grained entropy in the box tends to zero. ST hermal (B) = 0 S(B) = 0 S(A) = 0 Next. but its fine grained entropy does not. outgoing photons. For now let’s consider a conventional system which is described by known laws of physics.16) (8. The box is in a pure state.

But the fine grained entropy of A must vanish. The second law of thermodynamics insures that ST hermal (A) is larger than ST hermal (B) just after the explosion.2 Top: Von Neumann entropy of Σ1 vs fraction f.76 Black Holes. the thermal or coarse grained entropy of the exterior radiation has increased to its final value. . and the String Theory Revolution S (Σ 1 ) Pure state 1 2 - 1 f SThermal (Σ 1 ) Additive coarse-grained entropy 1 2 - 1 f I (Σ 1 ) 1 2 - 1 f Fig. since the entanglement has gone to zero. 8. Middle: coarse grained entropy of Σ1 vs fraction f. Bottom: information vs fraction of total degrees of freedom in Σ1 At this time. Information.

the curvature components.The Laws of Nature 77 The actual entanglement entropy must be less than the thermal entropy of A or B. In its simplest form the equivalence principle says that a gravitational field is locally equivalent to an accelerated frame. This time is called the information retention time. Before that point. However. . The consequence of this principle is the final radiation field outside the box must be in a pure state. The description of the evolution of the various kinds of entropy follow from very general principles. but no information. Note that it applies to observations made from outside the black hole. Thus a plot of the various entropies looks like Figure 8. a good deal of energy has escaped. Roughly the point where information appears outside of the box is the point where half of the final entropy of the photons has emerged. Thus we see how information conservation works for a conventional quantum system. this does not mean that localized regions containing a small fraction of the photons cannot be extremely thermal. it says that a freely falling observer or system will not experience the effects of gravity except through the tidal forces. More exactly. 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 M G will not be distorted or otherwise disrupted by the presence of the horizon.3 Equivalence Principle A second law of nature concerns the nature of gravitation. It is the amount of time that it takes to retrieve a single bit of information about the initial state of the box. 8. It is useful to define this time at which information begins to emerge. 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. or equivalently. They typically carry negligible information. Thus we regard the conservation of information in black hole evaporation as a fundamental law of nature.3.

78 Black Holes. Information. Bottom: evolution of entanglement entropy and information .3 Top: evolution of the thermal entropies of box and exterior. 8. 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.

To see why such a system is not possible.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. if it is in the down state.e.19) quantum Xerox principle is sometimes called the no-cloning principle . 8.4 Schematic diagram of quantum Xerox machine respect to the z-axis.4.4. it is duplicated |↑ → |↑ |↑ Similarly. i. imagine that we insert a spin in the input port as in Figure 8.18) (8. We call it a Quantum Xerox Machine. producing the original and a duplicate. no effect at all. If the spin is in the up state with In Out Fig.4. 8. namely.17) Now suppose that the spin is inserted with its polarization along the x-axis. and which will copy that system.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.∗ It is a machine into which any system can be inserted. it is duplicated |↓ → |↓ |↓ (8.4. 1 √ (|↑ + |↓ ) 2 ∗ The (8. What it says is that a particular kind of apparatus cannot be built.

20) On the other hand.4. we would be able to violate the Heisenberg uncertainty principle by a set of measurements on those states.80 Black Holes.21.21) 1 2 = |↑ |↑ + 1 2 |↑ |↓ + |↓ |↑ + |↓ |↓ The state in equation 8. a true quantum Xerox machine is required to duplicate the spin in equation 8. Thus we see that the principle of linearity forbids the existence of quantum Xerox machines. If we could construct Xeroxed quantum states. Information.18 1 1 √ (|↑ + |↓ ) → √ (|↑ |↑ + |↓ |↓ ) 2 2 (8.19 1 √ 2 (|↑ + |↓ ) → 1 2 1 √ 2 (|↑ + |↓ ) 1 2 1 √ 2 (|↑ + |↓ ) (8. and the String Theory Revolution The general principles of quantum mechanics require the state to evolve linearly.17 and 8.4. Thus from equations 8.4.20 is obviously not the same as that in equation 8. .4.4.4.4.

it played a central role in replacing the old ideas of locality with a new paradigm. so that the final state is given by |ψout = S |ψin (9. instead of the correct Bekenstein–Hawking c3 value of 4G .1) One way of stating the principle of information conservation is through the unitarity of S. quantum field theory has a serious defect when it comes to describing systems with horizons. The argument was simple and persuasive. He meant that in a fine grained sense. the first of the laws of nature described in the Introduction would be violated. quantum field theory must be replaced with an entirely new paradigm in which the concept of locality is radically altered. It gives rise to an infinite entropy density on the horizon. According to the principles of quantum mechanics. Although Hawking’s conclusion is undoubtedly wrong.0. Let us concentrate on a theory of massless particles. namely local quantum field theory in the fixed background of a black hole. so that 81 . By information loss. Hawking did not mean the practical loss of information such as would occur in the bomb-in-the-box experiment in Chapter 8.1. It was based on the only available tool of that time.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. Following Hawking. let’s begin with a black hole that has been formed during a collapse. information would be lost. the evolution of an initial state |ψin is governed by a unitary S-matrix. The Penrose diagram is shown in Figure 9. The point is that a unitary matrix has an inverse. As we will see. we think of the geometry as a background on which we can formulate quantum field theory. In other words. As we have already seen. To state the problem.

3) In arguing that the final Hilbert space is a tensor product.2) Now let us consider the S-matrix in a black hole background. 9.82 Black Holes.2 that some of the final particles will escape to I + and some will be lost behind the horizon. Evidently the final Hilbert space.0. These incoming particles will interact and scatter by means of Feynman diagrams in the black hole background. and the String Theory Revolution Singularity H or iz on + Collapsing energy - Fig. Hawking . hout is a tensor product of the states on I+ and those at the singularity S. Thus hin = hI− hout = hI+ ⊗ hS (9.1 Penrose diagram of black hole formed by collapse in principle the initial state can be recovered from the final state |ψin = S † |ψout (9. It is evident from Figure 9. The Hilbert space of initial states hin is clearly associated with quanta coming in from I − .0. Information.

9.0.0.4) where T rsingularity means a trace over the states on the singularity. Since every point on the singularity S is space-like relative to every point on I + .2 Feynman diagrams in black hole background relied on an important property of quantum field theory: locality. . Thus. 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 S-matrix which connects hin and hout as in equation 9.1. all experiments on the outgoing particles are described by a density matrix ρout = T rsingularity |ψout ψout | (9. This was central to Hawking’s analysis. Since they have no access to S.The Puzzle of Information Conservation in Black Hole Environments 83 S + - Fig. Let us consider the description of the final state from the viewpoint of the observers at I + . the operators on S all commute with those on I + .

Once the entropy of the black hole has evaporated past 1 its original value. This is a fundamental law of quantum mechanics. is that it badly violates the equivalence principle. Such remnants would be extremely pathological. Instead they end their lives as stable Planck-mass remnants that contain all the lost information. a law of nature would be violated from the viewpoint of the external observer. In this case the state on I + will not be pure. or even infinite entropy. Information. A freely falling observer would encounter a “brick wall” just above the horizon. and we will not pursue that line further. The situation is even worse if the information is not emitted at all. the only possibilities would seem to be that either information is lost during the entire process of formation and evaporation. especially by relativists. Obviously such remnants would have to have an enormous. A final possibility that was advocated by some authors is that black holes never completely evaporate. However. By correlated. Hawking went on to make arguments that the purity of the state would not be restored if the black hole evaporated. One is that the horizon is not penetrable. The reason that this was never seriously entertained. 9. and the String Theory Revolution A simple example might involve a pair of correlated particles which are thrown in from I − . we have seen in Chapter 8 that the maximum amount of information that can be hidden in a system is its entropy. By the time the black hole has small mass and entropy. In other words. If one particle ends up on S and the other ends up on I + . from the viewpoints of an infalling system. we mean that the two particles wave function is entangled.1 A Brick Wall? There are two more possibilities worth pointing out. it must begin to 2 come out in the emitted radiation. the horizon bounces everything out. Since the near horizon region of a Schwarzschild black hole is essentially flat space-time. Thus. . even if all information were emitted at the very end of the evaporation process.84 Black Holes. the entanglement entropy of the radiation cannot be larger than the black hole’s remaining entropy. or the information is restored to the outside world at the very end of the evaporation process. then the final state will be entangled. when the singularity is “exposed” at Planckian temperature. In fact.

the horizon is seen as flat featureless space-time. it tells us that the micro-degrees of freedom can absorb. the horizon cannot duplicate information. since the infalling observer cannot send reports from behind the horizon. But a potential contradiction can occur for the infalling observer. and eventually re-emit all information in the form of 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. requires information to freely pass through the horizon. . the fine grained entropy of the radiation field (due to entanglement) cannot exceed the entropy of the black hole. That number is of order the area of the horizon in Planck units. The information conservation principle requires all information to be returned to the outside in Hawking radiation. black hole complementarity just says that no observer ever witnesses a violation of a law of nature. It does not tell us what those micro-degrees of freedom are. In other words.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.2 Black Hole Complementarity In its simplest form. No obvious contradiction is posed for the external observer. the quantum Xerox principle closes out the last possibility. For a freely falling observer. 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. The equivalence principle. No high temperatures or other anomolies are encountered.12). Furthermore. At any given time. and send one copy into the black hole while sending a second copy out. At the end of evaporation. all information is carried off in Hawking radiation. This means that as long as the black hole is much larger than the infalling system. but it allows us to estimate their number. Thus. on the other hand.10 and 1. It seems that some law of nature must break down. Even more convincing is the fact that the horizon of a black hole formed by a light-like shell forms before the shell gets to the center (see Figures 1. 9. Evidently we have come to an impasse. Finally. The quantum Xerox principle precludes both happening. thermalize. black hole complementarity tells us that the equivalence principle is respected. at least for some observer.

After collecting some information about A (from the Hawking radiation ). it passes through the horizon without incident. and the String Theory Revolution Singularity C M es sa g In fo rm at io n e B A Fig.86 Black Holes.3 Information exchange from External to infalling observer Consider a black hole. Information. According to assumption. System A is assumed to contain some information. We don’t actually need observer B to decode the information. shown in Figure 9. 9. Next. .3 along with an infalling system A. According to observations done by A. All we really need is a mirror outside the black hole horizon to reflect the Hawking radiation back into the black hole. observer B then jumps into the black hole. the photons recorded by observer B encode the information carried in by system A. consider an observer B who hovers above the black hole monitoring the Hawking radiation.

2. we need to remember that no information will be emitted until about 1 the entropy of 2 the black hole has evaporated. since observer B has discovered a quantum Xeroxing of information from observer A.The Puzzle of Information Conservation in Black Hole Environments 87 We now have two copies of the original information carried by A.2. Observer B might be forced to the singularity before a message can arrive. Now we have a contradiction.7) From Figure 9.2.2. This means that A has a time of order ∗ ∆t ≈ (M G) e−ω to send the message.2. this takes a time of order t∗ ≈ M 3 G2 (9.4 we can see why there might be a problem with the experiment. . To see why the experiment fails.8) Thus. We can imagine A sending a signal to observer B so that observer B discovers the duplicate information at point C in Figure 9. it must ∗ all occur at x− < (M G) e−ω . From equation 5.6) In terms of the light cone coordinates x± = ρe±ω we have x+ x− > ∗ ∗ x+ ∗ > ≈ P 2 P exp (ω ∗ ) (9. observer B will hit the singularity at a point with x− < ∼ (M G) e−ω ∗ (9.16. then black hole complementarity is not self-consistent. the singularity is given by x+ x− = (M G) 2 (9.9) The implication is that if A is to send a signal that B can receive. This means that observer B’s jump off point must correspond to Rindler coordinates satisfying ω∗ ≥ ρ∗ ≥ t∗ 4MG P ≈ M2 G (9.5) 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 fact. observer B hovers above the stretched horizon.3.0. In other words. If this experiment is possible.

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 . and that the experiment cannot be done.88 Black Holes.= (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 (9.2.4 Now. the energy carried by observer A must be many orders of magnitude larger than the black hole mass.10) . there are no limits on how much information can be sent in an arbitrarily small time with arbitrarily small energy.6 we see that this energy is exponential in the square of the black hole mass (in Planck units) E > eM G . However in quantum mechanics.2. Since that quantum must be emitted between x− = 0 and ∗ x− = M G e−ω . to send a single bit requires at least one quantum. and the String Theory Revolution M es sa ge H or fro iz m on A Fig. Information. for observer A to be able to signal observer B before observer B hits the singularity. In other words. in the classical theory. its energy must satisfy E > ∗ 1 eω MG From equation 9. 9. It is obvious that A cannot fit into the horizon.

It is formed by the collapse of roughly 1057 nucleons. the mass of the black hole at this point is about 1010 kilograms. its temperature increases. None of these carry off baryon number.5. and its temperature is 10−8 electron volts. and the transition evidently violates baryon conservation. This is a tiny fraction of the original black hole mass. unlike electric charge. The obvious difference between baryon number and electric charge is that baryon number is not the source of a long range gauge field. It can only begin to radiate baryons when its temperature has increased to about 1 GeV. It is far too cold to emit anything but very low energy photons and gravitons. as well as between two positrons. Baryon number must be violated by quantum gravitational effects. . Using the connection between mass and temperature in equation 4. and even if it were to decay into nothing but protons. Consider a typical black hole of stellar mass.0.22. it cannot carry any quantum numbers. there are powerful reasons to believe that baryon number. Thus it can disappear without some flux having to suddenly change at infinity. 9. As it radiates. 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. As a simplified example. it could produce only about 1037 of them. and pions. can at best be an approximate conservation law. Since the X-boson is described by a real field. most modern theories predict baryon violation by ordinary quantum field theoretic processes. In fact. Nevertheless. as in Figure 9. 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. The proton could then decay into a positron and an electronpositron pair. and at some point it is hot enough to emit massive neutrinos and anti-neutrinos.3 Baryon Number Violation The conservation of baryon number is the basis for the incredible stability of ordinary matter. then electrons. muons. let’s suppose there is a heavy scalar particle X which can mediate a transition between an elementary proton and a prositron.

The baryon number is lost. Now let us consider what happens when a proton falls into a black hole. 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. The complicated history of the proton is described by Feynman diagrams such as shown in Figure 9. there is nothing that would stimulate it to decay before that. the time along the proton trajectory between horizon crossing and the singularity can be enormous. So the external observer will conclude that baryon violation can take place at the horizon. and will not be radiated back out in the Hawking radiation. On the other hand. For a very large black hole. Information. that sounds absurd. Surely the event of proton decay takes place in some definite place. Temperatures higher than MX can certainly excite the proton to decay. The proton will have a lifetime in excess of 1032 years.5 Feynman diagrams of p→ Xe+ and e+ → Xe+ If the mass MX is sufficiently large. it is difficult to understand how there can be an ambiguity. the proton encounters enormously high temperatures as it approaches the horizon. Who is right? The answer that black hole complementarity implies is that they are both right. From the proton’s viewpoint. 9. On the face of it. baryon conservation will be a very good symmetry at ordinary energies. The interactions cause it to make virtual transitions from the bare proton to a state with an X-boson and a positron. Thus. from the vantage point of the external observer. The real proton propagating through space-time is not the simple structureless bare proton. The diagrams make evident that the real proton is .6. and the String Theory Revolution e+ e+ X X p e+ Fig.90 Black Holes. just as it is in the real world.

and MX of the order 1016 GeV. The transitions 1 take place on a time scale of order δt ∼ MX . κ of the order of the Planck mass. As long as the X-boson is sufficiently massive. The quantity that we normally call baryon number is really the time averaged baryon number normalized to unity for the proton. the intermediate state has vanishing baryon number.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. 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. e+ system is not small. this is not a short-lived intermediate state. 9. Now let us consider a proton passing through a horizon. the rate for real proton decay will be very small. since the proton is so stable.7 that from the viewpoint of an external observer. Nothing about virtual baryon non-conservation is especially surprising. This probability is closely related to the wave function renormalization constant of the proton. and is of the order P robability ∼ g2 4π κ log MX where κ is the cutoff in the field theory. But it is clear from Figure 9. its instantaneous baryon number is zero. 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. The resolution of the paradox is that the proton is continuously making extremely rapid transitions between baryon number states. in the particular processes shown in Figure 9. Since the probability that the proton is actually an X. For example. as shown in Figure 9. and the proton will be stable for long times. it is not unlikely that when it passes the horizon. . the probability that the proton has the “wrong” baryon number is of order unity.7. This might seem paradoxical.6. for g ∼ 1.

7 Proton fluctuations while falling through horizon This. Baryon violating processes are expected at that temperature. the proton is falling into a −1 region of increasing temperature. As a proton or any other system approaches the horizon.92 Black Holes. so that a short-lived virtual fluctuation becomes stretched out into a real process. To him. internal oscillations or fluctuations appear to indefinitely slow down. is nothing but the usual time dilation near the horizon. The process of baryon violation near the horizon should not be totally surprising to the external observer. the temperature becomes of order MX . At a proper distance MX from the horizon. and the String Theory Revolution Time resolution becomes very dense ∆t ∆t ∆t ∆t ∆t ∆t ∆t t=0 Fig. Information. of course. An interesting question is whether an observer falling with the proton . 9.

Thus the observer cannot measure and report the absence of baryon violation at the horizon without himself causing it. . In the proton’s frame. the time spent in the hot region before 1 crossing the horizon is MX . 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. frequencies so high that the ordinary low energy observer has no chance of detecting them. 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. The answer is interesting. to obtain a single bit of information about the state of the proton. and send a message to the outside world that the proton has not decayed. In order to make an observation while the proton is in a region of temperature MX the observer must do so very quickly. the observer has to probe it with at least one quantum with energy of order MX .The Puzzle of Information Conservation in Black Hole Environments 93 can observe the baryon number just before crossing the horizon.

and the String Theory Revolution .94 Black Holes. Information.

By analyzing the highest energy fragments. We expect to discover high energy collision products flying out at all angles. Thus we see that a giant “Super Plancketron” Collider (SPC) would fail to discover fundamental length scales smaller than the Planck scale P . we hope to reconstruct very short distance events. 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. the collision will create a black hole of mass ∼ E∆x . The problem with this thinking is that at energies far above the Planck mass. head on. To make matters worse. a size ∆x can be probed with a quantum of energy E∆x ≈ c ∆x (10. no matter how high the energy. The typical Hawking particle has energy ∼ Rc which decreases with the S incident energy.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 .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. but rather low energy Hawking radiation. Let’s begin with a traditional attempt to study interactions at length scales smaller than the Planck scale.0. In fact. particles with center of mass energy E∆x . According to conventional thinking. what we need to do is to collide. According to conventional relativistic quantum mechanics. the products of collision will not be high energy particles.

and the String Theory Revolution increases we would be probing ever larger scales ∆x ∼ 2G E . Thus in order to observer the proton before it crosses the horizon. More generally. Thus. Call the proper time ∆τ . Let’s consider further: what happens to a proton falling into the black hole? The proton carries some information with it. 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. the amount of proper time that it has before crossing the horizon tends to zero as the Schwarzschild time tends to infinity.4) ∆τ ∼ 8M G δR e− 4M G . . in Chapter 7. The only reasonable expectation is that the constituents of the proton “melt” and diffuse throughout the horizon.0. the observer who follows the proton does not see it spread at all as it falls. As the proton is observed over shorter and shorter time intervals. its charge. the region over which it is localized must grow.2) We might express this in another form as a kind of space-time uncertainty principle: ∆x ∆t ∼ 2G c4 ≈ 2 P (10. From the viewpoint of the external observer. In fact.96 Black Holes. we have to do it in a time which is exponentially small as t → ∞. Following the proton as it falls. Information. what we need in order to be consistent with the thermal spreading of information is now clear.0. location on the horizon through which it falls. c2 c2 (10. On the other hand. we saw just such a phenomenon involving a charge falling onto the stretched horizon. Then ∆τ varies with Schwarzschild time like √ ∆t (10. all of the information stored in the incident proton should quickly be spread over the horizon. The proton is like a tiny piece of ice thrown into a tub of very hot water.3) This is the simplest example of the ultraviolet/infrared connection that will control the relation between frequency and spatial size. momentum. Very high frequency is related to large size scale. etc. As we saw in the previous lecture. particle type. The need to reconcile the complementary descriptions gives us an important clue to the behavior of matter at high frequencies.0. The UV/IR connection is deeply connected to black hole complementarity. the proton is falling into an increasingly hot region.

but it is at the heart of complementarity. Fortunately it is also built into the mathematical framework of string theory. if we assume that it grows consistently with the uncertainty relation in equation 10. it will spread over the horizon exponentially fast.3. if the proton size depends on the time resolution according to equation 10.3 (substituting ∆τ for ∆t) we obtain ∆x ∼ ∆τ 2 P ∼ 2 P MG et/4MG Thus. It is implicit in the modern idea of the UV/IR connection. .0. The idea of information spreading as the resolving time goes to zero is very unfamiliar.Horizons and the UV/IR Connection 97 In fact.0.

Information. and the String Theory Revolution .98 Black Holes.

PART 2 Entropy Bounds and Holography .

.

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 . We begin by considering a large region of space Γ. for example. But they also carry very small Rindler energy and therefore correspond to very long times to the external observer. Suppose we are dealing with a lattice of discrete spins. Now consider the space of states that describe arbitrary systems that can fit into Γ such that the region outside Γ is empty space. Let the lattice spacing be a and the volume of Γ be V .Chapter 11 Entropy Bounds 11. One might think that this is just an indication that a more or less conventional ultraviolet regulator is needed to render the theory consistent. For simplicity we take the region to be a sphere. The Holographic Principle is about the counting of quantum states of a system. Our goal is to determine the dimensionality of that state-space. This is an example of the weirdness of the Ultraviolet/Infrared connection.1 Maximum Entropy Quantum field theory has too many degrees of freedom to consistently describe a gravitational theory. The divergence arises from modes very close to the horizon. it would be in lattice field theories. But the divergent horizon entropy is not an ordinary ultraviolet phenomenon. 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. The modes which account for the divergence are very close to the horizon and would appear to be ultra short distance modes. Let us consider some preliminary examples. The main indication that we have seen of this overabundance of degrees of freedom is the fact that the horizon entropy-density is infinite in quantum field theory.

102 Black Holes. and the String Theory Revolution and the number of orthogonal states supported in Γ is given by Nstates = 2V /a 3 (11.3) (11. the energy within certain limits. In general entropy is not really a property of a given system but also involves one’s state of knowledge of the system. The total entropy is S = s(ρ)V The total number of states is of order Nstates ∼ exp S = exp s(ρmax )V (11. One begins by computing the thermodynamic entropy density s as a function of the energy density ρ.1. The entropy is essentially the logarithm of the number of quantum states that satisfy the given restrictions. Information. In this case the states can be counted using some concepts from thermodynamics.1. We can limit the states. In this case the number of quantum states will diverge for obvious reasons. In other words it is the entropy given that we know nothing about the state of the system. For definiteness we will take spacetime to be four-dimensional. Let us now consider a system that includes gravity. for example. More precisely it is proportional to the number of simple degrees of freedom that it takes to describe the system. To define entropy we begin with some restrictions that express what we know.4) This is typical of the maximum entropy. for example by requiring the energy density to be no larger than some bound ρmax . the angular momentum and whatever else we may know. For the spin system the maximum entropy is Smax = V log 2 a3 (11. 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. There is another concept that we will call the maximum entropy.1. Again we focus on a spherical .2) In each case the number of distinct states is exponential in the volume V . This is a property of the system. Whenever it exists it is proportional to the volume.1. It is the logarithm of the total number of states. In counting the states of a system the entropy plays a central role.1) A second example is a continuum quantum field theory.

1. 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. 11. This is shown in Figure 11. Thus we see a radical difference between the number of states of any (regulated) quantum field theory and a theory that includes gravity.1 In-moving (zero entropy) spherical shell of photons result of this process is a system of known entropy. The area of the boundary is A.Entropy Bounds 103 region of space Γ with a boundary ∂Γ. The Eg EM Γ R Fig. In other words the maximum entropy of a region of space is proportional to its area measured in Planck units. In other words the area of the horizon of the black hole is A. Now imagine collapsing a spherically symmetric light-like 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. 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. . Suppose we have a thermodynamic system with entropy S that is completely contained within Γ. S = A/4G.

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. just outside of the shell the geometry becomes increasingly curved as the Schwarzschild radius is approached (see Figure 11.1. As the photonic shell approaches the center. and a singularity forms at a later time.104 Black Holes. Information. 11. However. The horizon grows until the collapsing Light beams get more vertical as the energy becomes more concentrated Horizon is a light-like surface R Low curvature Infalling Shell E γ Fig. the geometry is Schwarzschild with low curvature (no black hole) prior to the shell crossing the Schwarzschild radius.2 Space-time depiction of radially in-moving shell of photons shell crosses. the horizon forms prior to the actual crossing of the shell.1. Inside of the shell. However. at that time the system winds up with entropy S= A 4G .2). . and the String Theory Revolution Space-time depiction of horizon formation Consider the collapsing spherically symmetric shell of light-like energy depicted in Figure 11.

4G (11.6) 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.1. We will see that it is most natural to define holographic entropy bounds on light-like surfaces[3] as opposed to space-like surfaces.2 Entropy on Light-like Surfaces So far we have considered the entropy that passes through space-like surfaces. Entropy is proportional to the number of photons. so this “holographic” limit must be satisfied. (11. R ∼ 1051 cm. consider thermal radiation at a temperature of 1000◦K. The number of photons Nγ in a volume of radius R satis3 V fies Nγ ∼ λ3 ∼ R(cm) ⊗ 105 .5) The coarse grained volume in phase space cannot decrease. 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. Under .Entropy Bounds 105 Therefore. which corresponds to photons of wavelength ∼10−5 cm. and thus one expects S ∝ R(cm) ⊗ 105 . 11. considerable larger than the observable universe 1028 cm. the entropy of any system is limited by Smax ≤ A .1. Compare this with the maximum entropy calculated using the holographic limit Smax ≈ R lP lanck 2 3 ∼ R(cm) ⊗ 1033 = 2 .

106

Black Holes, Information, and the String Theory Revolution

certain circumstances the entropy bounds of light-like surfaces can be translated to space-like surfaces, but not always. The case described above is one of those cases where a space-like bound is derivable. Let us start with an example in asymptotically flat space-time. 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 three-dimensional light-like surface. In general, they define a light-like 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 space-time except for those points that lie behind black hole horizons. Now consider a space-time 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 two-dimensional 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 light-like 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 two-dimensional 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 three-dimensional 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,

2. da/dλ > 0. demonstrated in Figures 11.e. This will play an important role in generalizing the holographic bound to more general geometries.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. i.8. 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. One sees that due to gravitational lensing.7 Initial and later path and image during “slow” motion near caustic . Thus far we have assumed asymptotically flat boundary conditions.7. 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. if we attempt to use screens for which the light rays are converging then the argument fails.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. the image of the particle can move at arbitrarily large speeds! produce large. 11. This allowed us to choose the screen so that the light rays forming the light sheet intersect the screen at right angles. However. 11. rapid motions on the screen Small motions near caustic CAUSTIC CAUSTIC Fig.

The light sheet is now defined by the backward light cone formed by light rays that propagate from ∂Γ into the past (See Figure 11.9) where the index m runs over the d spatial directions. In particular we assume that the spatial entropy density (per unit d volume) is homogeneous. and the String Theory Revolution 11.R. ∂Γ. t Screen Past light cone Projection from screen onto light cone t>0 Fig.10) Let’s also make the usual simplifying cosmological assumptions of homogeneity.W.W. where it is called the Fischler–Susskind (FS) bound[5] and to more general geometries by Bousso[6].R. a(t) = a0 tp (11. The function a(t) is assumed to grow as a power of t. The boundary (d − 1)-sphere.3.110 Black Holes. 11. geometries. Later. The metric has the form dτ 2 = dt2 − a(t)2 dxm dxm (11. At time t we consider a spherical region Γ of volume V and area A. we will relax these assumptions. case.8 Holographic surface for calculating entropy bound with a spherical surface as the screen .3. will play the same role as the screen in the previous discussion.8). Consider the general case of d + 1 dimensions. Information. First we will review the F.3 Friedman Robertson Walker Geometry The holographic bound can be generalized to flat F.

This bound states that the total entropy passing through the light sheet does not exceed A/4G. the resulting density satisfies the holographic bound. If we only count that portion of the entropy which did pass through the light sheet. However if Γ is larger than the particle horizon at time t the light sheet is not a cone. It is now easy to see why we concentrate on light sheets instead of spacelike surfaces. it will scale like the area A. 11. In other words the light sheet has positive expansion into the future and negative expansion into the past. We will return to the question of space-like bounds after discussing Bousso’s generalization[6] of the FS bound. but rather a truncated cone as in Figure 11. which is cut off by the big bang at t = 0. We observe that at the screen the area of the outgoing bundle of light rays is increasing as we go to later times. The key to a proof is again the focusing theorem. Obviously if the spatial entropy density is uniform and we choose Γ big enough.9. t Sphere Past light cone Entropy flux through light cone bounded Big Bang t=0 Entropy not bounded Fig. Thus a portion of the entropy present at time t never passed through the light sheet. The focusing theorem again tells us that if we map the entropy of black holes passing through the light sheet to the screen.Entropy Bounds 111 As in the previous case the holographic bound applies to the entropy passing through the light sheet.9 Region of space causally connected to particle horizon . the entropy will exceed the area. It is believed that the bound is very general.

11) An outgoing light ray (null geodesic) which would generate the particle horizon satisfies dt = a(t)dx. The proper size of the horizon is approximately given by 1018 (light) seconds. and the Planck time is approximately 10−43 seconds.112 Black Holes.3. and d represent the number of spatial dimensions. we will determine if the entropy per horizon-area contained within the particle horizon is increasing or decreasing. 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 . which gives the form for the time dependence .W. Nγ ≈ 1090 . 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: SHorizon ≤ AHorizon 4G ??? We can check by recognizing that the entropy is primarily given by the number of black body cosmic background photons.R.10). Let σ be the entropy volume density. This gives a proper size for the horizon of about 1061 Planck units. We want to check whether d S ∼ σ RHorizon <? (a(t) RHorizon ) 4G d−1 (11. 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. using the F. geometry. Let R Horizon represent the coordinate size of the particle horizon (Figure 11. Next.

and the entropy bound will eventually be contradicted.10 Particle Horizon of the size of the particle horizon: t RHorizon (t) = 0 dt a(t ) (11.3.3.W.12) If we assume the form a(t) = ao tp then the particle horizon evolves according to the formula RHorizon (t) = t1−p .R.15) where P is the pressure and u is the energy density and w is a constant. Suppose that the matter in the F.3. 11.13) Therefore for the entropy bound to continue to be valid.14) One sees that if the expansion rate is too slow. d (11. the time dependence must satisfy t(1−p)d < td−1 . cosmology satisfies the equation of state P = wu (11. This bounds the expansion rate coefficient to satisfy p > 1 . ao (11. Given w and the scale factor for the expansion p. then the coordinate volume will grow faster than the area. one can use the Einstein .3.Entropy Bounds 113 t Past light cone RHorizon Entropy flux Big Bang t=0 Fig.

18) A 4G is reached when tdecoupling ∼ 10−44 sec 1056 (11. Take a space-like region Γ bounded by the space-like boundary ∂Γ. one gets S = 10−28 A/(4G) The entropy bound S = tdecoupling t 1 2 tdecoupling t 1 2 (11. We will . 11.3. and the String Theory Revolution field equation to calculate a relationship between them: p= 1 2 > d (1 + w) d (11. Recall that the speed of sound within a medium is given by 2 vs = ∂P ∂u =w Therefore. As one moves forward in time. the bound will always be satisfied.114 Black Holes.3.19) = 1028 ⇒ t = This time is comparable to the Planck time (by coincidence??).17) This is an interesting result. (11. Next. Therefore the entropy bound is not exceeded after the Planck time.4 Bousso’s Generalization Consider an arbitrary cosmology.16) We see that the number of spatial dimensions cancels and that the entropy bound is satisfied so long as w ≤ 1.3. in the future. 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. The relation satisfied by 2 the scale factor vs = w ≤ 1 is just the usual causality requirement. At any point on the boundary we can construct four light rays that are perpendicular to the boundary[6]. Information. since the speed of sound is always less than the speed of light. go back in time using the black body radiation background as the dominant entropy. not less.3. the entropy bound then becomes more satisfied.

In Region I of Figure 11. These light surfaces terminate on the initial boundary of the . However we are only interested in the branches of 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. Each point of such a diagram really stands for a 2-sphere (more generally a (d − 1)-sphere).12 the expansion of the universe causes both past branches to have negative expansion. In general as we move around in the Penrose diagram the particular branches which have negative expansion may change.12 we draw the Penrose–Bousso (PB) diagram for a positive curvature.11 we illustrate flat space-time and the negative expansion branches of a typical local 2-sphere. If we consider general boundaries the sign of the expansion of a given branch may vary as we move over the surface. together with its neighbors define a positive or negative expansion as we move away from the boundary. It consists of four distinct regions. Two branches go toward the future. One of them is composed of outgoing rays and the other is ingoing. However in non-static universes other combinations are possible. matter dominated universe that begins with a bang and ends with a crunch. Thus in Figure 11.Entropy Bounds 115 call these the four branches. 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. In each region the negative expansion branches are indicated by their directions at a typical point. Bousso’s bound states that the entropy passing through these light sheets is less than A/4G where A is the boundary of ∂Γ. For example if the cosmology initially expands and then collapses. The Penrose diagram represents the radial and time directions. In ordinary flat space-time. For example in a rapidly contracting universe both future branches have negative expansion while the past branches have positive expansion. the outgoing future branch will go from positive to negative expansion. For simplicity we restrict attention to those regions for which a given branch has a unique sign. Thus we draw light surfaces into the past. On any of these branches a light ray. We can now state Bousso’s rule: From the boundary ∂Γ construct all light sheets which have negative expansion as we move away. Bousso introduced a notation to indicate this. Similarly two branches go to the past. The four branches at a given point on the Penrose diagram are represented by a pair of 45 degree lines passing through that point. The Penrose diagram is divided into a number of regions depending on which branches have negative expansion.

W. The relevant light sheets in this region begin on the 2-sphere q and both terminate at the spatial origin. one condition for a space-like bound is that the entropy is bounded by a corresponding future light sheet. These are untruncated cones and the entropy on both of them is holographically bounded. In Region III the entropy bounds are both on future light sheets. The reason is that not all matter . We find that the entropy is bounded on a future light sheet. Therefore. There is something new in this case. Bousso’s rule tells us nothing in this case about the entropy on space-like surfaces bounded by p. By the second law of thermodynamics the entropy on the space-like surface can not exceed the entropy on the future light sheet. 11. The holographic bound in this case says that the entropy passing through either backward light surface is bounded by the area of the 2-sphere at point p. Thus in this case the entropy in a space-like region can be holographically bounded. It is evident that any matter which passes through the space-like surface must also pass through the future light sheet.13).116 Black Holes. and the String Theory Revolution Fig. and therefore the entropy is not bounded on space-like regions with boundary p. Nevertheless there is no space-like bound. case. Information.11 Negative expansion branches of 2-sphere in flat space-time geometry and are similar to the truncated cones that we discussed in the flat F. With this in mind we return to Region I.R. Now consider a space-like surface bounded by q and extending to the spatial origin (shown in Figure 11. For Region I there is no future bound. Now move on to Region II.

Region IV is identical to Region II with the spatial origin being replaced by the diametrically opposed antipode.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.14 demonstrates a region that does not satisfy an entropy bound for this cosmology.12 Penrose–Bousso diagram for matter dominated universe which pass through space-like surfaces are forced to pass through the future light sheets.) Figure 11. 11. (See Figure 11.13. . Thus we see that there are light-like bounds in all four regions but only in II and IV are there holographic bounds on space-like regions.

Information.14 Light-like surface which does not satisfy entropy bound .118 Black Holes. 11. The entropy thru such a surface is NOT bounded in a Robertson–Walker cosmology t = 0 Fig.13 Bounding surfaces for inflowing entropy t final Area is INCREASING in part of this light-like surface. 11. and the String Theory Revolution Entropy on this space like surface is also bounded by flux through light-like surface q Fig.

22) where dΩ2 is the metric of a unit 3-sphere. and the scale factor a is given 3 by a(t) = R cosh(t/R). 3 2 (11.5.Entropy Bounds 119 11.5. First it is believed that the early universe underwent an epoch of rapid inflation during which the space-time geometry was very close to de Sitter.5. (11.5.5.24) (11.21) de Sitter space can also be written in the form dτ 2 = dt2 − a(t)2 dΩ2 3 (11. R2 = 3 Gλ (11. It is the hyperboloid given by 4 i=1 (xi )2 − (x0 )2 = R2 . The model of cosmology which is presently gaining the status of a “standard model” involves de Sitter space in two ways.20) The radius of curvature R is related to the cosmological constant.5 de Sitter Cosmology de Sitter space holds special interest because of its close connection with observational cosmology.23) For our purposes we want to put this in a form that will allow us to read off the Penrose diagram.25) .5. Four-dimensional de Sitter space may be defined by embedding it in (4 + 1) dimensional flat Minkowski space. The inflationary theory is more than two decades old and by now has been well tested. To that end define the conformal time T by dT = dt/a(t). λ. (11. More recently de Sitter space has entered cosmology as the likely candidate for the final fate of the universe. One easily finds that the geometry has the form dτ 2 = (a(t)) (dT 2 − dΩ2 ). 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.

and the String Theory Revolution 0 −π Fig. 11. case. The reason is that the geometry rapidly expands as we move toward the upper and lower boundary of the de Sitter space.16. We leave it as an exercise to show that the entropy is . Once again the Bousso construction divides the Penrose diagram into four quadrants. the conformal time (T = tan−1 (tanh(t/R)) − π ) varies from −π to 0.15.120 Black Holes. 11.W.15 0 π Penrose diagram for de Sitter space-time S Fig. Information. Since the polar angle on 2 the sphere also varies from 0 to π the Penrose diagram is a square as in Figure 11.16 Extremal space-like surface in de Sitter space-time Furthermore when t varies between ±∞.R. However. the contracting light sheets in the upper and lower quadrants are oriented oppositely to the usual F. Of particular interest is the bound on a space-like surface beginning at the center of the Penrose diagram and extending to the left edge as in Figure 11.

Inflationary cosmology is illustrated in Figure 11.R. The dotted line where the two geometries are joined is the reheating surface where the entropy of the universe is created. the universe began with a large vacuum energy which mimicked the effects of a positive cosmological constant.17.Entropy Bounds 121 t Post-inflation particle horizon Larger area Period of inflation Fig.W.19. In fact the maximum entropy on any such surface is given by S= 4πR2 . According to the inflationary hypothesis. During that period the geometry was de Sitter space.R. space as in Figure 11.18. . But then at some time the vacuum energy began to decrease and the universe made a transition to an F. universe.R. The Penrose–Bousso diagram for de Sitter space is shown in Figure 11.17 Inflationary universe bounded by the area of the 2-sphere at the center of the diagram. 4G (11.26) de Sitter space has a special importance because of its role during the early evolution of the universe.5.W. In order to describe inflationary cosmology we must terminate the de Sitter space at some late time and attach it to a conventional F. 11. and is called reheating.W. The transition was accompanied by the production of a large amount of entropy. 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.

11. It is easy to see that in an ordinary inflationary cosmology p can be chosen so that the entropy on the space-like 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. Fig. 11. Information. and the String Theory Revolution Point corresponds to finite spherical area Fig.122 Black Holes.19 Joining of inflationary and post-inflationary geometries Let us focus on the point p in Figure 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. . There is no entropy prior.

Anti de Sitter space is important for an entirely different reason. 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. 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. 11. It is the background in which the holographic principle is best understood. Typical inflationary cosmologies can be studied to see that the past light sheet bound is not violated. .6 Anti de Sitter Space de Sitter space is important because it may describe the early and late time behavior of the real universe.Entropy Bounds 123 Entropy flux only after reheating p q Robertson– Walker No entropy flux here REHEATING Inflation Backwards light cone Fig.

The vacuum of AdS is a genuine zero temperature state. The space-time in appropriate coordinates is static but unlike de Sitter space the static coordinates cover the entire space. These two properties imply a very far . In other words the entire region to the left of L can be foliated with space-like surfaces such that the maximum entropy on each surface is A/4G. 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. Consider an arbitrary point p on L. 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. 11.124 Black Holes. AdS space is an example of a special class of geometries which have time-like Killing vectors and which can be foliated by space-like surfaces that satisfy the Holographic bound. The PB diagram consists of a single region in which both negative expansion light sheets point toward the origin.21. The PB diagram consists of an infinite strip bounded on the left by the spatial origin and on the right by a boundary.21 Static surface of large area in AdS space AdS space is the vacuum of theories with a negative cosmological constant. Therefore the entropy on any space-like surface bounded by p and including the origin is also holographically bounded. Information. We will think of L as an infrared cutoff.

.6.Entropy Bounds 125 reaching conclusion.27) The holographic description of AdS space is the subject of the next chapter. 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.

and the String Theory Revolution . Information.126 Black Holes.

Chapter 12 The Holographic Principle and Anti de Sitter Space 12. The entropy bound tells us that the maximum number of non-redundant degrees of freedom is proportional to the area. the maximum entropy is a measure of the number of simple degrees of freedom∗ that it takes to completely describe the region. as in the case of a lattice of spins. This is almost always proportional to the volume of Γ. measured in Planck units. 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. Typically. 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. 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. The exception is gravitational systems. 127 . For a large macroscopic region this is an enormous reduction in the required degrees of freedom. 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. One of the elements of this paradigm is the Holographic Principle and its embodiment in AdS space. time and locality. Let us consider a region of flat space Γ. 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.1 The Holographic Principle As we have seen in Chapter 11. A simple degree of freedom represents a single bit of information.

The remarkable precision is due to the unusually high degree of symmetry of the theory which includes a powerful version of supersymmetry. 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. 12. This is a 10-dimensional product space consisting of two factors. 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. Why the S(5)? The reason involves the high degree of supersymmetry enjoyed by superstring theory.128 Black Holes. three-dimensional space described by a two-dimensional hologram at its boundary! That this is possible is called the Holographic Principle. For our purposes 5-dimensional AdS space may be considered to be a solid 4-dimensional . Information. In this lecture we will describe a very precise version of AdS “Holography” which grew out of the mathematics of string theory. The particular space that we will be interested in is not simple 5dimensional AdS but rather AdS(5)⊗S(5)[7][8][9]. 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 analogy with a hologram is obvious. Ordinarily. and the String Theory Revolution Planck area. Thus AdS is a natural framework in which to study the Holographic Principle. But there is one special situation which is naturally ball-like. it does not make sense to consider a ball-like region with a boundary in the general theory of relativity. From a mathematical point of view it is essential if we want to be able to make precision statements. However we will downplay the mathematical aspects of the theory and concentrate on those physical principles which are likely to be general. We will begin with a brief review of AdS geometry.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. From a conceptual point of view the extra internal 5-sphere is not important. the 5-dimensional AdS and a 5-sphere S(5). It occurs when there is a negative cosmological constant. Anti de Sitter space. Generally the kind of supergravity theories that emerge from string theory don’t have cosmological constants.

The boundary is at z = ∞. The geometry can be described by dimensionless coordinates t.1 and 12.2.2. The metric in equation 12. For all practical purposes AdS space behaves . r is the radial coordinate (0 ≤ r < 1) and Ω parameterizes the unit 3-sphere.2.2 may be expressed in terms of the coordinate z = 1/y.2. It has a horizon at y = ∞. For example.2.2 are not the same. light can travel to the boundary and back in a finite time.2.2.2 is related to 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.2.1 in the vicinity of a point on the boundary at r = 1.2 is not geodesically complete. time coordinates appearing in 12. To construct the space AdS(5) ⊗ S(5) all we have to do is define 5 more coordinates ω5 describing the unit 5-sphere and add a term to the metric 2 ds2 = R2 dω5 5 (12.2.2. 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. First of all it is an approximation to equation 12.1 and 12. The geometry has uniform curvature R−2 where R is the radius of curvature.1) There is another form of the metric which is in common use.The Holographic Principle and Anti de Sitter Space 129 spatial ball times the infinite time axis.2. When interpreted in this manner. dτ 2 = R2 z 2 (dt2 − dxi dxi ) − 1 2 dz z2 (12.2. The 3-sphere is replaced by the flat tangent plane parameterized by xi and the radial coordinate is replaced by y.4) Although the boundary of AdS is an infinite proper distance from any point in the interior of the ball. A time-like geodesic can get to y = ∞ in a finite proper time so that the space in equation 12. The second way that equations 12. The metric 12.2.2 are related is that 12. dτ 2 = (12. r.1 in two different ways.2) where i runs from 1 to 3. Ω where t is time. with y = (1 − r).2 is the exact metric of an incomplete patch of AdS space.2. 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.

To see what this entails. the metric is approximately conformal. The size of the cavity is of order R. 4) Generally. Information. 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. 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. a hyperbolic plane (or the Poincar´ disk ). let us compute the area of the boundary. A unit coordinate time interval corresponds to increasingly large proper time intervals. A radial null geodesic satisfies 1 + r2 dt2 = 4dr2 . and the String Theory Revolution like a finite cavity with reflecting walls. However the Holographic Principle requires more than that.3 Holography in AdS Space We will refer to AdS(5) ⊗ S(5) as the bulk space and the 4-dimensional boundary of AdS at y = 0 as the boundary. the spatial metric is that of a uniformly (negatively) curved space. which means that light rays move at 45◦ angles near 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. which makes the space causally finite. 2) The metric is singular at r = 1 in all components. e 12.2. From equation 12.130 Black Holes. 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. 3) Near r = 1. . It requires that the boundary theory has no more than 1 degree of freedom per Planck area. 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 = 0.1 we see that the metric diverges near the boundary.

As we mentioned.2. That suggests that the boundary theory might be a quantum field theory.5) On the other hand if we consider the representation of AdS in 12. Let us consider the metric in the form 12. The full symmetry group of AdS(5) is the group O(4|2). AdS(5) ⊗ S(5) is a solution of the 10-dimensional supergravity that describes low energy superstring theory. All of the transformations act on the boundary as conformal transformations which preserve light-like directions on the boundary. and so it is. x. In addition there is a “dilatation” symmetry t → λt xi → λxi y → λy (12. the 4-dimensional Poincar´ symmetry e acts on the boundary straightforwardly. In fact the full AdS symmetry group. The additional symmetry is the so-called N = 4 supersymmetry. The implication of this symmetry of the boundary is that the holographic boundary theory must be invariant under the conformal group. This symmetry must also be realized by the holographic theory. 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.3. Indeed the space has more symmetry than just the conformal group and the O(6) symmetry of the internal 5-sphere. when acting on the boundary at y = 0 is precisely the conformal group of 4dimensional Minkowski space. All of this leads us to the remarkable conclusion that quantum gravity in AdS(5) ⊗ S(5) should be exactly described by an appropriate .1 we can see additional symmetry. and that is in fact the case. The dilatation symmetry also acts as a simple dilatation of the coordinates t. but for the time being we can assume that the number of degrees of freedom per unit coordinate area is infinite. xi . Another important fact involves the symmetry of AdS. Obviously.The Holographic Principle and Anti de Sitter Space 131 Later we will regulate this divergence by moving in a little way from y = 0.2. 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. It is obvious that the geometry is invariant under ordinary Poincar´ transformations of the 4-dimensional Minkowski e coordinates t.2. In addition there is also the symmetry O(6) associated with rotations of the internal 5-sphere. For example the rotations of the sphere Ω are symmetries.

The total 9-dimensional spatial volume in the interior of L is easily computed using the metric 12.2.9) 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. To do so we introduce a surface L at r = 1 − δ. A conventional cutoff would involve a microscopic length scale such as the 10dimensional Planck length lp . 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.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. We find Smax ∼ 1 R8 8 δ 3 lp (12. To say it slightly differently. 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. and is seen to be critically divergent.8) where A is the 8-dimensional area of the boundary L.3.7) In such a theory we also will find that the maximum entropy is of the same order of magnitude. This is also easily computed.132 Black Holes. Information. V (δ) ∼ R9 .3. However the Holographic Principle suggests that this entropy is overestimated. the Holographic Principle implies a complete description of all physics in the bulk of a very large AdS .2. 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. One way to do this would be to introduce a spatial lattice in 9-dimensional space. and the String Theory Revolution superconformal Lorentz invariant quantum field theory associated with the AdS boundary. δ3 (12.3.3. but a random lattice with an average spacing lp is possible. It is not generally possible to make a regular lattice.

It is well known that SYM is conformally invariant and therefore has no dimensional parameters. Since AdS(5) ⊗ S(5) is a 10-dimensional spacetime one might think that its boundary is (8 + 1) dimensional.4 The AdS/CFT Correspondence The search for a holographic description of AdS(5) ⊗ S(5) is considerably narrowed by the symmetries. x. It will be convenient to define the theory to live on the boundary parametrized by the dimensionless coordinates t. The correspondence states that there is a complete equivalence between superstring theory in the bulk of AdS(5) ⊗ S(5) and N = 4.The Holographic Principle and Anti de Sitter Space 133 space in terms of only lp /R degrees of freedom per spatial Planck volume. Thus if ESYM and M represent the energy in the SYM and bulk theories ESYM = M R. the conversion factor is R. Ω or t. SU (N ). SYM theory on the boundary of the AdS space[7][8][9]. The corresponding momenta are also dimensionless. length and temperature carry their usual dimensions. In fact there is only one known class of systems with the appropriate N = 4 supersymmetry. 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. There is one question that may be puzzling to the reader. Nonetheless the theory must be able to describe microscopic events in the bulk even when R becomes extremely large. 3 + 1-dimensional. 12. the SU (N ) Supersymmetric Yang–Mills (SYM) theories. To see this let us Weyl rescale the metric by a . N represents the number of supersymmetries. Similarly bulk time intervals are given by multiplying the t interval by R. In fact we will use the convention that all SYM quantities are dimensionless. We will only review some of the salient features. and N is the dimension of the Yang–Mills gauge theory. To convert from SYM to bulk variables. But there is an important sense in which it is 3 + 1 dimensional. On the other hand the bulk gravity theory quantities such as mass. In these lectures SYM theory will always refer to this particular version of supersymmetric gauge theory.

134 2 Black Holes.4.11) On the other side of the correspondence. Let us return to the correspondence between the bulk and boundary theories. The boundary of the geometry is therefore 3+1 dimensional. The 10-dimensional bulk theory has two dimensionless parameters. R s 2 = (N gym ) 4 1 2 g = gym (12. The AdS/CFT correspondence has been widely studied as a tool for learning about the behavior of gauge theories in the .4.4. and the String Theory Revolution R factor (1−r2 )2 so that the rescaled metric at the boundary is finite. The new metric is 2 dS 2 = (1 + r2 )2 dt2 − 4dr2 − 4r2 dΩ2 + (1 − r2 )2 dω5 (12.13) There are two distinct limits that are especially interesting. 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 . the gauge theory also has two constants. These are: 1) The radius of curvature of the AdS space measured in string units R/ s . depending on one’s motivation. 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 (12.10) Notice that the size of the 5-sphere shrinks to zero as the boundary at r = 1 is approached.4. Alternately we could measure R in 10-dimensional Planck units. Information.12) We can also write the 10-dimensional Newton constant in the form G = R8 /N 2 (12. The relation between string and Planck lengths is given by g2 8 s 8 = lp 2) The dimensionless string coupling constant g. In these lectures we will assume the relation that was originally derived in [7].

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 12.5 The Infrared Ultraviolet Connection In either of the metrics in equation 12.2 the proper area of any finite coordinate patch tends to ∞ as the boundary of AdS is approached. This means we want the limit g = constant s → ∞ R/ On the gauge theory side this is (12.14) (12.4. This limit is dominated by classical supergravity theory.2.4. In order to do a more refined counting[9] .16) gym = constant N →∞ (12.8. 12.2. 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.4.3.4. This is consistent with the fact that a continuum quantum field theory such as SYM has an infinity of modes in any finite three-dimensional patch.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 gym → 0 N →∞ 2 gym N = constant From the bulk string point of view the limit is g→0 R = constant s (12.1 or 12. Thus we expect that the number of degrees of freedom associated with such a patch should diverge.15) Thus the strongly coupled ’t Hooft limit is also the classical string theory limit in a fixed and large AdS space.

As we will see. these apparently different regulators are really two sides of the same coin. Such objects fill 3 dimensions of space and also time.2. We are especially interested in D3-branes. Thus we can think of the z location as a scalar field living on the D-brane. 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.136 Black Holes.18) . Our main concern will be with the z(x. Let us assume that they fill time and the 3 spatial coordinates xi . A Dp-brane is a p-dimensional object. 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.2. The most important property of D3-branes is that they are embedded in a 10-dimensional space.3 and 12. It is in many ways similar to the behavior of strings as we study them at progressively shorter time scales. In fact the geometry defined by 12. Let the other 6 coordinates be called √ z m and let z ≡ z m z m . For example the location in z may fluctuate. 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]. 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.5. 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 D-branes. D-branes are objects which occur in superstring theory. 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 D-brane has local degrees of freedom. 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. They are stable “impurities” of various dimensionality that can appear in the vacuum. What makes the D-brane stack interesting to us is that the geometry sourced by the stack is exactly that of AdS(5) ⊗ S(5). In addition there are modes of the brane which are described by vector fields with components in the t. We will place a “stack” of N D3-branes at z = 0. Specifically the geometry sourced by the D-branes is a particular solution of the supergravity equations of motion: ds2 = F (z)(dt2 − dx2 ) − F (z)−1 dzdz (12. D-branes can also be juxtaposed to form stacks of D-branes.4 is closely related to that of a D-brane stack. A stack of N D-branes has a mass and D-brane charge which grow with N . Information.

19) cgs N z4 and c is a numerical constant.5.2.23) . + (z 6 )2 (12. 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 .The Holographic Principle and Anti de Sitter Space 137 where F (z) = 1+ cgs N z4 −1/2 (12..2. . 12.5.. √ However the radial coordinate z = z m z m can be defined by z2 = gym 2 s R2 2 1 T rφ2 N (12. ω5 .5.. The location of a point on a 3-brane is defined by six coordinates z.. then the precise connection between the z s and φ s is z= gym 2 s φ R2 (12. Furthermore the theory of the fluctuations of the stack is N = 4 SYM. The geometry is noncommutative and only configurations in which the six matrix valued fields commute have a classical interpretation.5.20) It is then a simple exercise to see that the D-brane metric is of the form 12.5.5.22 does not make sense because the fields φ in SU (N ) are N × N matrices.3.5.21) As we indicated.. All of the fields in this theory form a single supermultiplet belonging to the adjoint (N × N matrix) representation of SU (N ). We may also choose the six coordinates to be Cartesian coordinates (z 1 .2.3. In this lecture we give an argument for the IR/UV connection based on the quantum fluctuations of the positions of the D3-branes which are nominally located at the origin of the coordinate z in equation 12. The original coordinate z is defined by z 2 = (z 1 )2 + .4. the coordinates z m are represented in the SYM theory by six scalar fields on the world volume of the branes. z 6 ). where we identify the N eigenvalues of the matrices in equation 12.22) Strictly speaking equation 12. If the six scalar fields φn are canonically normalized.18 to be the coordinates z m of the N D3branes[10]. (12.

25) In terms of the coordinate y which vanishes at the boundary of AdS < y >2 ∼ δ 2 .5. It follows from equation 12. From this it follows that any of the N 2 components of φ satisfies φ2 ∼ δ −2 ab (12. 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 .5.23. 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 = ∞. and the String Theory Revolution A question which is often asked is: where are the D3-branes located in the AdS space? The usual answer is that they are at the horizon z = 0.27) Here it is seen that the location of the brane is given by the ultraviolet cutoff of the field theory on the boundary.5.12 < z >2 ∼ δ −2 (12.5.24) where δ is the ultraviolet regulator of the field theory. To answer the corresponding question about D3-branes 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 . The precise connection between the UV SYM cutoff and the bulk theory IR cutoff is given by equation 12.26) gym 2 s R2 2 N δ2 (12. (12. However our experiences in Chapter 14 with similar questions will warn us that the answer may be more subtle.5. Information. High frequency or short time probes see the string widely spread in space while low frequency probes see a well localized string.5.6 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]. What we will find there is that the way information is localized in space depends on what frequency range it is probed with. using equation 12.4.138 Black Holes.20 that the average value of z satisfies < z >2 ∼ or.

13 (12.28) δ3 On the other hand the 8-dimensional 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 12.4. Thus the number of degrees of freedom on the unit area is N2 (12.The Holographic Principle and Anti de Sitter Space 139 unit coordinate area. Within each such cell the fields are constant in a cutoff theory. .6.30) Ndof 1 ∼ (12.6.31) A G This is very gratifying because it is exactly what is required by the Holographic Principle.6.6. Within that patch there are 1/δ 3 cutoff cells of size δ.29) (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 ).

and the String Theory Revolution .140 Black Holes. Information.

suppose a fluctuation occurs in which the black hole absorbs an extra bit of energy from the surrounding heat bath. They are long-lived objects.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. Any object with this property is thermodynamically unstable. to see how the holographic description of AdS(5) ⊗ S(5) provides a description of black holes. When the black hole absorbs some energy it cools but so does the finite heat bath. The black hole will grow indefinitely. If the box is not too big the heat bath will cool more than the black hole and the flow of heat will 141 . their temperature decreases as their energy or mass increases. The reason is that their specific heat is negative. but of course they are not. The fluctuations are self-regulating. but eventually they evaporate. Again a runaway will occur that leads the black hole to disappear. If on the other hand the black hole gives up some energy to the environment it will become hotter than the bath. We can try to prevent their evaporation by placing them in a thermal heat bath at their Hawking temperature but that does not work. 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. A well known way to stabilize the black hole is to put it in a box so that the environmental heat bath is finite. 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. Now we will return. This in turn will favor an additional flow of energy from the bath to the black hole and a runaway will occur. full circle. We have treated Schwarzschild black holes as if they were states of thermal equilibrium.

let us write the metric of AdS in a form which is convenient for generalization. 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.4) . The black holes which are stable have Schwarzschild radii as large or larger than the radius of curvature R. Note that the coordinates r. Before writing the AdS–Schwarzschild metric. 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.2. 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.1.1. In this lecture we will consider the properties of black holes which are stabilized by the natural box provided by Anti de Sitter space.142 Black Holes. Information. dτ 2 = 1+ r2 R2 dt2 − 1 + r2 R2 −1 dr2 − r2 dΩ2 (13. t are not the same as in equation 12.0. 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 (13.3) where c is a numerical constant. They homogeneously fill the 5-sphere and are solutions of the dimensionally reduced 5-dimensional Einstein equations with a negative cosmological constant.2) where the parameter µ is proportional to the mass of the black hole and G is the 10-dimensional Newton constant.0.0.0. Using the thermodynamic relation dM = T dS we can compute the relation between mass and temperature: M =c R11 T 4 G (13.1) where in this formula r runs from 0 to the boundary at r = ∞. One finds that the entropy is related to the mass by S = c M 3 R11 G−1 1 4 (13.

Black Holes in a Box 143 Fig.3 can be computed in two ways.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 13.0.0. The second way is to calculate it from the boundary quantum field theory in the free field approximation.0.0. but the free field gives too big a coefficient by a factor of 4/3.5 is nothing but the Stephan–Boltzmann law for black body radiation. . In the present case the relevant volume is the dimensionless 3-area of the unit boundary sphere. Recall that in 3 + 1 dimensions the Stephan–Boltzmann law for the energy density of radiation is E ∼ T 4V (13. The constant c in equation 13. The first is from the black hole solution and the Bekenstein–Hawking formula.1 Penrose diagram of the AdS black hole or in terms of dimensionless SYM quantities Esym = c R8 4 T G sym (13. 13.5) 4 = cN 2 Tsym Equation 13. a black body thermal gas of N 2 species of quanta propagating on the boundary hologram.6) where V is the volume. This is not too surprising because the classical gravity approximation is only valid 2 when gY M N is large.5 has a surprisingly simple interpretation.

13. Let us return to equation 12. Roughly speaking this is insured by the renormalization group as applied to the SYM description of the branes.5. instead of using equation 12. All probes sense thermal fluctuations of the brane locations.25 we use the thermal field fluctuations of φ. An example is the thermal fluctuations of fields at high temperature. In that case the action for φ is the minimally coupled scalar action .1 The Horizon The high frequency quantum fluctuation of the location of the D3-branes are invisible to a low frequency probe. A simple example is the minimally coupled scalar dilaton field φ.5.8) It is clear that the thermal fluctuations will be strongly felt out to a coordinate distance z = Tsym . For each of the N 2 components the thermal fluctuations have the form 2 < φ2 >= Tsym (13.5.9) In fact this coincides with the location of the horizon of the AdS black hole. The renormalization group is what insures that our bodies are not severely damaged by constant exposure to high frequency vacuum fluctuations. and the String Theory Revolution 13.144 Black Holes.24 but now. We will only consider dilaton fields which are constant on the 5sphere.1.1. Information. Such objects are 10-dimensional fields and should not be confused with the 4-dimensional quantum fields associated with the boundary theory.7) and we find equations 12.5. In terms of r the corresponding position is 1 − r ∼ 1/Tsym (13. We are not protected in the same way from classical fluctuations.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.27 replaced by 2 < z >2 ∼ Tsym 2 −2 < y > ∼ Tsym (13.26 and 12. Let us consider some field that appears in the supergravity description of the bulk.1.

(13.11) Equation 13. Plugging in the black hole metric given in equation 13.11 has implications for quantum correlation functions in the black hole background. 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.e. r → ∞.2.11 requires it to behave like φ(t)φ(t ) ∼ exp−γ|t − t |. (13. However. Consider the correlator φ(t)φ(t ) . Indeed .10) The appropriate boundary conditions are φ → 0 at the boundary of the AdS. 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.13) where H is a dimensionless increasing function of its argument. in the pure AdS background φ is periodic in time. Secondly.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. we find a number of things. In other words the system does not preserve any memory of the initial perturbation.2. This is true both in pure AdS as well as in the AdS– Schwarzschild metrics.2. Since AdS is exactly described by a conventional quantum system it follows that the correlator should not go to zero. i. We shall now prove this assertion. 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.0. I= √ d5 x −gg µν ∂µ φ∂ν φ.2.2.2. Equation 13. This type of behavior is characteristic of large thermal systems where γ would correspond to some dissipation coefficient. but in the black hole metric φ goes to zero exponentially with time: φ → exp−γt. Any quantum system with finite entropy preserves some memory of a perturbation.Black Holes in a Box 145 in 5-dimensional AdS. The essential point is that any quantum system with finite thermal entropy must have a discrete spectrum. (13.2.

the long time behavior of correlation functions.2.2. The long time average L is obviously nonzero and positive. Inserting a complete set of (discrete) energy eigenstates gives F (t) = 1 Z e−βEi ei(Ej −Ei )t |Aij |2 . In studying QFT in the vicinity of a horizon we have seen that the entropy is UV divergent. 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 the long time average of F (t)F ∗ (t). 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. This observation allows us to understand why it tends to zero in the (bulk) QFT approximation. Z (13.2. This is due to the enormous number of short wave length modes near the horizon. ij (13.146 Black Holes.14) By finite we simply mean that the spectrum is discrete and the entropy finite.16) Using equation 13. L = lim 1 T →∞ 2T +T −T dtF (t)F ∗ (t) (13.15) For simplicity we will assume that the operator A has no matrix elements connecting states of equal energy. 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. . Information. This includes questions of information loss and of particular interest in this lecture.17) where the delta function is defined to be zero if the argument is nonzero and 1 if it is zero. ijkl (13.15 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 ) . and it is typically of the order e−S where S is the entropy of the system. 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. This means that the time average of F vanishes.

These fluctuations are analogous to the classical phenomenon of Poincar´ recurrences. . Thus we can expect large fluctuations in the correlators at intervals of order eS .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.2.18) This long time behavior. (13. missed by bulk quantum field theory. 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. A large fluctuation which reduces the entropy by amount ∆S has probability e−∆S . The expected behavior is that they fluctuate chaotically.

Information.148 Black Holes. and the String Theory Revolution .

PART 3 Black Holes and Strings .

.

in an elementary way. As a quantum field theory. 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 earliest evidence that QFT is too rich in degrees of freedom was the uncontrollable short distance divergences in gravitational perturbation theory. Nevertheless.Chapter 14 Strings We have learned a great deal about black holes by considering the behavior of quantum fields near horizons. Later in Chapter 12 we quantified just how over-rich QFT is. 151 . with a theory that seems to have just the right number of degrees of freedom: string theory. 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. The remaining portion of this book deals. But ultimately local quantum field theory fails in a number of ways. string theory was mostly defined by a set of perturbation rules. Until relatively recently. The problems posed by black holes for a fundamental theory of quantum gravity are non-perturbative. imagine a freely falling whistle that emits a sound of such high frequency that it cannot be heard by the human ear. In general the failures can be attributed to a common cause – quantum field theory has too many degrees of freedom. Entropy is a direct measure of the number of active degrees of freedom of a system. As an example. Einstein’s general relativity is very badly behaved in the ultraviolet. Evidently there are far too many degrees of freedom very close to a horizon in QFT. even in perturbation theory. Even more relevant for our purposes is the divergence in the entropy per unit area of horizons that was found in Chapter 4. we will see certain trends that are more consistent with black hole complementarity than the corresponding trends in QFT.

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 light-like in the particle’s rest frame. In other words, the particle and the Rindler observer are boosted relative to one another by an ever-increasing 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 σ-reparameterizations. 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 σ-reparameterizations. 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 non-relativistic 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 non-relativistic 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

1.11.3.1. For the more interesting case of an interacting quantum field theory.1.13) Finally. Since the interaction is spread over the time interval in equation 14.1. Let us also assume the very high momentum components of the wave function are negligible. the effective probability distribution for Xδ is concentrated in a well localized region of fixed extent.Strings 155 experiment which takes place over a short time interval δ just before horizon crossing. For example. we use the non-relativistic equations of motion X i (σ) = X i (0) + eP i σ to give i Xδ = X i (0) + (14.12) eP i δ . In this kind of environment the information stored in the infalling system would be thermalized and spread . δX. No matter how small δ is.10.11) To evaluate equation 14. Under these conditions nothing singular happens i to the probability distribution for Xδ as δ → 0. except that the probe carries out information about the transverse location of the particle instead of its baryon number. and it too shows no sign of spreading as the sampling interval δ tends to zero. Complementarity requires the probe to report that the particle fell into a very high temperature environment in which it repeatedly suffered high energy collisions.1. This experiment is similar to the δ one discussed in Chapter 9. Why is this a problem? The reason is that it conflicts with the complementarity principle. All of this is exactly what is expected for an ordinary particle in free quantum field theory. let us suppose that the particle wave function is initially a smooth wave packet well localized in transverse position with uncertainty ∆X i . the particle is probed over the time interval −δ < X + < 0 (14. There is no tendency for information to transversely spread over a stretched horizon.10) by a quantum of (Minkowski) energy ∼ 1 . a time averaged relative coordinate or charge density can be defined. the instantaneous transverse position should be replaced by the time averi aged coordinate Xδ i Xδ = 1 δ 0 −δ X i (σ) dσ (14.1. we could study the transverse properties of an interacting or composite particle such as a hydrogen atom. In other words. 2 (14.

which is identified with light cone time X + . There are a number of excellent textbooks on string theory that the reader who is interested in technical details can consult. Thus X i (σ 0 .2. X(σ 0 .2. A free string is a generalization of a free particle. H = 1 P− 2π 0 dσ 2 |Pi (σ )| + 2 ∂X i ∂σ 2 (14. the tendency is already there in the theory of free strings.15 is a simple wave equation ∂2X i (∂σ 0 ) 2 − ∂2X i (∂σ 1 ) 2 = 0.2. 0).16) Quantization of the string is straightforward. The equation of motion following from equations 14. 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. σ 1 ) is a field defined on a 1+1 dimensional parameter space (σ a ).2 Light Cone String Theory Although naive pertubative string theory cannot capture this effect completely correctly. where σ 1 runs from 0 to 2π.16 with periodic boundary conditions in σ 1 . .2. only the most elementary aspects of string theory will be needed. For our purposes. X i (σ) becomes a free scalar field in 1+1 dimensions satisfying equation 14. the canonical momentum density Pi (σ) can also be defined. It is also a function of a time-like parameter σ 0 . The transverse coordinates of the point at σ 1 are labeled X i (σ).2. In addition to X i (σ). and the String Theory Revolution over the horizon.2.14 and 14. 14. Pj (σ ) = i δj δ(σ − σ ) (14. A string is a one-dimensional continuum whose points are parameterized by a continuous parameter σ 1 .14) The light cone Hamiltonian for the free string is a natural generalization of that for a free particle. 2π) = X(σ 0 . Information. At equal times X and P satisfy i X i (σ).15) We have used units in which the string tension (energy per unit length in the rest frame) is unity.156 Black Holes.

Finally.18 is the ground state string.2.2.2. Let us consider the analog of the question that we addressed about the time averaged location of the point particle.2. . that is 2 (∆X)2 = Xδ − Xδ 2 (14.2. it is the short time behavior that is key to complementarity.2. 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. This quantity is easily calculated and diverges logarithmically as δ → 0. the inverse string tension.2. it doesn’t matter what value σ 1 takes on the right hand side when we evaluate Xδ . especially in its short time behavior. As we have repeatedly emphasized.0.20 we have restored the units by including the factor α .4 (∆X)2 ∼ | log (2Le−2τ ) | ∼ 2τ.20) In equation 14. as the string approaches the horizon. we can use the relation betwen Rindler time and Schwarzschild time given by τ = t/4M G to obtain (∆X)2 ∼ αt . (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 |.Strings 157 The string differs in important ways from the free particle. Thus. Here we see the beginnings of an explanation of complementarity.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. define Xδ = 1 δ δ 0 dσ 0 X(σ) (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 . 4M G (14. The observer outside the black hole will find the string diffusing over an increasing area of the horizon as time progresses.17) Since all points σ 1 are equivalent. Now we consider the time averaged location of a point on the string.19) Another way to write equation 14. In other words.

2. To see how interactions influence the evolution. The true exponential asymptotic growth is undoubtedly a non-perturbative phenomenon that involves string interactions in an essential way.0. A completely consistent theory would require these growth patterns to match.2.158 Black Holes. and the String Theory Revolution The linear growth of the area in equation 14. i If however ∂X is averaged over the time interval δ.2. 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. In that case inspection of 7. and the result is quadratically divergent.4 ≈ 1 2τ e . To see this.2. projected onto the two-dimensional transverse plane 2π = 0 dσ 1 ∂X i ∂σ 1 2 (14. .21 indicates that the growth is exponential.2.22) is Gaussian in free Using the fact that the probability distribution for field theory. This ∂σ is another exercise in free scalar quantum field theory.20 is much slower than the growth of a charged particle described in Chapter 7. let us consider the ground state average of ∂X1 . 2L ∂X 2 ∂σ1 or scales as (14. let’s determine the average total length of string.21) i As a preliminary.24) In other words.0. it grows exponentially in length.1 to derive ∂ ∂ρ = 1 √ 2 −e−τ ∂ ∂X + − + eτ ∂ ∂X − . we use the transformation in equation 14. we can conclude that ≈ or using equation 14. Information.0.23) 1 δ (14. as the string falls toward the horizon. Another quantity which exponentially grows is the ρ component of the Rindler momentum.

3. √ Roughly speaking. The quantity g 2 α in string theory also governs the gravitational interaction between masses.Strings 159 or in terms of momenta 1 Pρ = √ eτ P− − e−τ P+ 2 (14. String interactions are governed by a dimensionless coupling constant g which determines the amplitude for strings to rearrange when they cross. It is clear that this behavior cannot continue indefinitely. When g 2 ρ becomes large. √ Sometimes α is replaced by a length s = α . It is . There is only one dimensional constant in string theory. when a piece of string gets within a distance of order α of another piece. Let ρ be the number of such crossings per unit horizon area. As the string falls toward the horizon. Evidently then the ratio of the string length to its total radial momentum is fixed. this not the end of the story. Now. they can interact. the form g 2 ρ is not dimensionless. P± are conserved.2. the inverse string tension α with units of area. The the total length of the string grows exponentially with τ . The dimensionally correct statement is that string interactions become important when g2ρ ≥ 1 2 s ≈ 1 g2 α (14.25) In the Rindler approximation to a black hole horizon. The number of such string encounters will obviously increase without bound as τ → ∞. 14. Obviously the importance of interactions is governed not only by g. 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. but also by the local density of string crossings. interactions can no longer be ignored.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. and therefore as τ → ∞ the radial momentum Pρ grows like eτ .26) This criterion has a profound significance. However. its radial momentum increases by the mechanism of its physical length increasing.

3. the area density of horizon entropy. string theory is not a 4-dimensional theory. The implication is that interactions become important when the 1 area density of the string approaches G . A 1 good guess is that the density of string saturates at order G . we can be sure that something new will happen. Although we cannot follow the string past the point where interactions become important. This is reminiscent of the pattern of growth that we encountered in Chapter 7.3. It is important to check if the same logic applies in higher dimensions.3. Since the spreading is associated with a decreasing time of averaging it is not seen by a freely falling observer.3. provides a phenomenological description of how information spreads over the horizon.30) .29) so that the perturbative limit is again reached when the density is of order ρ ≈ 1 G (14.27) (14.26 is replaced by g2 ρ ≥ or ρ ≈ 1 D−2 g2 s 1 D−2 s = 1 (α ) D−2 2 (14. Let D be the spacetime dimension. and that the repulsion prevents the density from increasing indefinitely. and the String Theory Revolution none other than the gravitational coupling constant (in units with c = = 1). Information. But the simple assumption that splitting and joining interactions cause effective short range repulsion. Verifying that as δ → 0 the string grows as if the density saturates is beyond the current technology of string theory. Since the τ total length of string grows like ≈ e the area that it occupies must also grow exponentially.3. The general case goes as follows: Since the number of transverse directions is D-2.160 Black Holes.28) In D dimensions. equation 14. In general. This is the essence of complementarity. the gravitational and string couplings are related by G ≈ g2 D−2 s (14.

4. It is a curious property of string theory that X − (σ) is not an independent degree of freedom.32 is to observe that δ is an averaging time in light cone coordinates.4.4. local operators can be characterized by a mass dimension. neither the stretched nor the mathematical horizon appears real.4. The only degrees of freedom carried by the string are the transverse coordinates X i (σ) which lie in the plane of the horizon. to a Frefo observer. Equation 14. located a microscopic distance above the mathematical horizon. Let us draw the motion of the fluctuating string in the X ± plane as it falls towards . Thus it is apparent that X − (σ) has dimension 1.32) The factor 2 is needed for dimensional reasons. In such a theory. ∂X i ∂X i ∂X − = ∂σ1 ∂σ1 ∂σ0 (14. The right ∂σ side of equation 14. s Another way to write equation 14.4. It immediately follows that the fluctuation in X − satisfies ∆X − ∼ 2 s δ (14.33) Now the geometric meaning of equation 14.32 then takes the form of an uncertainty principle ∆X − ∆X + ≈ 2 s. The derivation is provided in the supplement that follows this discussion. In other words δ = ∆X + . we have repeatedly run into the idea of the stretched horizon.31) Recall that the quantum theory of X(σ) is a simple (1+1 dimensional) quantum field theory of (D-2) free scalar fields defined on a unit circle.4. the degrees of freedom of an infalling object should get deposited in this layer of finite thickness. X i has dimension zero.31 has dimension 2.4 Longitudinal Motion In discussing the properties of horizons. (14. Thus. 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. i For example.4. It is natural to ask if the stretched horizon has any reality or whether it is just a mathematical fiction? Of course.Strings 161 14. consistency requires that in some appropriate sense. To study this question we must examine how strings move in the X − direction.33 is very interesting. The longitudinal location is defined by an equation whose origin is in the gauge fixing to the light cone gauge. while ∂X has dimension 1.

14.162 Black Holes.33 must increase. unlike a point particle. Note that if the string coupling satisfies gs << 1. Evidently as X + tends to zero the fluctuation in X − required by equation 14. P = s g.2. Information. This is shown in Figure 14. the stringy substance is seen by a probe to hover at a distance ∼ s above the horizon. In other words. the string length s can be much greater than the Planck length.4. and the String Theory Revolution ∆x - ∆τ} x+ Region of increasing fluctuation Fig. 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 + = 0.2 Near horizon Rindler time slices the horizon. 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. Once again this surprising result is a direct consequence of arbitrarily high frequency fluctuations implicit in the stringy structure of matter. .

3 String parameter translational invariance .31 by 1 = ∂X . but is + given by equation 14. Setting this quantity to zero insures that the physical spectrum consists only of states which are invariant under σ reparameterization. namely shifting σ by a constant. Under the transformation σ → σ + δσ the X s transform as X→X+ ∂X δσ. However even in the light cone gauge there is a bit of residual gauge invariance.Strings 163 Supplement: Light cone gauge fixing of longitudinal string motions The coordinate X− (τ .σ) is not an independent degree of freedom.4.4. To see this multiply 14. In going to the light cone frame the constraint serves to define X − in terms of the transverse coordinates. which ∂τ is true for light cone coordinates. ∂τ ∂σ ∂τ ∂σ (14.4.31. 14.34) The content of this equation expresses the underlying invariance of string theory to a reparameterization of the σ coordinate.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. Thus we must check the validity of the following equation ∂X j ∂X j ∂X + ∂X − − = 0.34.4. ∂σ (14.4.

164 Black Holes. . Information.4.37) Setting this to zero simultaneously insures invariance under shifts of σ and periodicity of X − (σ). and the String Theory Revolution dσ ∂X j dX j ∂σ ∂τ ∂X − .4.4. ∂σ (14.31 is equal to dσ (14.36) which by equation 14.

What is more. Let us begin with the formula for the entropy of a Schwarzschild black hole in an arbitrary number of dimensions. the statistical mechanics of strings allows us to compute the entropy up to numerical factors of order unity. 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. but it doesn’t tell us what they are. All of this is in cases where quantum field theory would give an infinite result. some in 3 + 1 dimensions. String theory has many different kinds of black holes.0. some in higher dimensions. counting microstates. 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. In every case the results nontrivially agree with the Bekenstein– Hawking formula.1) 165 . it is worth exploring the connections between strings and black holes in any dimension. 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. The black holes can be neutral or be charged with the various charges that string theory permits. A real theory of quantum gravity should tell us and also allow us to compute the entropy by quantum statistical mechanics. that is.Chapter 15 Entropy of Strings and Black Holes The Bekenstein–Hawking entropy of black holes points to some kind of microphysical degrees of freedom. Call the number of spacetime dimensions D. We will see that for the entire range of such black holes. Because string theory is not necessarily a 4-dimensional theory.

6) From Einstein’s equation for ideal pressureless matter. Supplement: Schwarzschild geometry in D = d + 1 dimensions To extend a static spherically symmetric geometry to D = d + 1 dimensions.0.0.0.0.4) The entropy in equation 15. and the String Theory Revolution The horizon is defined by RS = 16π(D − 3)GM ΩD−2 (D − 2) 1 D−3 (15.3) Finally. 4G 4G D−2 (15.0...4 is what is required by black hole thermodynamics. Gtt = κρ.. 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. + sin2 θ1 . the metric can be assumed to be of the form ds2 = −e2Φ dt2 + e2∆ dr2 (15.8) .2) and its D-2 dimensional “area” is given by D−2 A = RS D−2 dΩD−2 = RS ΩD−2 .sin2 θd−2 dθd−1 .5) 2 2 2 + r2 dθ1 + sin2 θ1 dθ2 + .166 Black Holes. Information..0. the entropy is given by S = (2GM ) D−3 ΩD−2 A = .0. ˆ Using orthonormal coordinates.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 (15.0. This ˆˆ means Gt = − ˆ t ˆ (D − 2) d D−3 r (1 − e−2∆ ) = −κρ 2rD−2 dr (15. (15.

0.0.0.9) r + κρ For pressureless matter in the exterior region (ρ = 0 = P ).0.13) Thus. . we can immediately conclude that Φ = −∆. Near the horizon.0. a useful shortcut for calculating the solution to Einstein’s equation 15.0.0. F (r) = 1 + 2φN ewton . 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 (15.14) Using the first law of thermodynamics.Entropy of Strings and Black Holes 2π (D−1)/2 Γ((D−1)/2) .0. 2π 2RS (15.12) which gives the relation between Rindler time/temperature units and Schwarzschild time/temperature units dω = (D − 3) dt 2RS 1 (D − 3) . the proper distance to the horizon is given by ρ = 2RS (D − 3) r Rs D−3 −1 (15.15) Substituting the form κ = 8π(D − 3)G for the gravitational coupling gives the previous results in D-dimensions. the entropy can be directly calculated to be of the form S = 2π(D − 3)A κ (15. the Hawking temperature of the black hole is given by THawking = (15.11) The Hawking temperature can be calculated by determining the dimensional factor between the Rindler time and Schwarzschild time. (15.7 is to note its equivalence to the Newtonian Poisson equation in the exterior region ∇2 F (r) = −κρ . 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 (15.10) If we write F (r) ≡ e2Φ .

Entropy is an adiabatic invariant. On the other hand the quantization of the string defines a 1+1 dimensional quantum field theory in which the (D-2) transverse coordinates X i (σ) play the role of free scalar fields. Very highly excited free strings have an enormously rich spectrum. They can be thought of as a mass of tangled string that forms a time-varying random walk in space. Information. energy levels are neither created nor destroyed. and the String Theory Revolution All calculations of entropy in string theory make use of a well known trick of quantum mechanics. The entropy of a string of mass m can be calculated by returning to the light cone quantization of the previous lecture. Therefore a neutral black hole must evolve into a collection of free strings. It must turn into a collection of free strings. In the process of adiabatic variation. A very massive black hole might evolve into a large number of low mass strings or. and it runs from 0 to 2π. 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 counting of the states of a free string is best done in the light . Thus we begin with a black hole of mass Mo in a theory with string coupling go . 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. But such a variation will not alter its entropy. The spatial coordinate of this field theory is σ1 . Basically we are using the quantum analog of the method of adiabatic invariants. What happens to the black hole as g tends to zero? The answer is obvious. But only the free strings have finite energy in the limit g → 0. The trick consists of identifying some kind of control parameter that can be adiabtically varied. D-branes. Let us imagine decreasing the string coupling g.168 Black Holes. branes of various dimensionality such as membranes. For any eigenstate of the Hamiltonian with vanishing transverse momentum and unit P− the light cone energy is m2 /2. at the opposite extreme. Such random walking strings have a large entropy and can be studied statistically. String theory has all kinds of non-perturbative objects. and so on. Adiabatically varying a control parameter like go will cause a change in the black hole mass and other internal structural features. monopoles. a single very highly excited string.

Entropy of Strings and Black Holes 169 cone version of the theory that we discussed in the last lecture.19 is correct for the simplest bosonic string. Call this entropy Sn (m). Recall that the free string is described by means of a 1 + 1 dimensional quantum theory containing D − 2 fields X i .17) Subleading corrections can also be calculated to give S = 2(D − 2)π m s − c log (m s) (15. a formal light cone temperature T can be defined. each carrying mass m .21) Obviously for large n the single string is favored.0. Now let us compare the entropy of the single string with that of n strings. For a given total mass.0.0. The leading contribution for large energy is (setting s = 1) E = π T 2 (D − 2) (15. This is actually quite general.20) − n c log m s n (15.0. the statistically most likely state in free string theory is a single excited string.0. Then n Sn (m) = n S(m/n) or Sn (m) = 2(D − 2)π m s (15. The entropy is the log of the density of states. Therefore the number of states with mass m is Nm = 1 m s c exp 2π(D − 2) m s (15.0. most of the black hole states will evolve into a single excited string. In order to describe the highly excited string spectrum. The entropy and energy of such a quantum field theory can be calculated by standard means.18) where c is a positive constant.0.19) The formula 15. but similar formulae exist for the various versions of superstring theory.16) S = 2πT (D − 2) Using E = m and eliminating the temperature yields S = 2 or. 2 2(D − 2)π m (15. restoring the units S= 2(D − 2)π m s. . Thus it is expected that when the string coupling goes to zero.

This is the point at which the transition from black hole to string occurs.0.24) (15.0. The Schwarzschild radius is of order RS ∼ (Mo G) D−3 .0. and using G ≈ g2 we find RS s D−2 s 1 (15. the string length will exceed the Schwarzschild radius of the black hole. The string and Planck length scales are related by g2 D−2 s = D−2 .170 Black Holes.22) • At some point as g → 0 the black hole will make a transition to a string. (15. Information.0. at some value of the coupling that depends on the mass of the black hole. But no matter how massive the black hole is.0. as we decrease g a point will come where the gravitational constant is too weak to matter. Eventually. Let us begin with a black hole of mass Mo in a string theory with coupling constant go .25) ≈ s 2 Mo go 1 D−3 . 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. p (15. In what follows we will vary the g while keeping fixed the string length s .23) Evidently as g decreases the string length scale becomes increasingly big in Planck units. and the String Theory Revolution These observations allow us to estimate the entropy of a black hole. This implies that the Planck length varies.26) . That is the point where the black hole makes a transition and begins to act like a string. 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. To understand this last assumption begin with a massive black hole. Gravity is clearly important and cannot be ignored. The point at which this happens is when the horizon radius is of the order of the string scale.

Thus we see that a black hole of mass Mo will evolve into a free string satisfying equation 15.33) This is a very pleasing result in that it agrees with the Bekenstein–Hawking entropy in equation 15.0. D−2 1 (15.29) Now as g → 0 the ratio of the g-dependent horizon radius to the string scale decreases.32.2 it becomes of order unity at M (g) which can be written M (g) s D−2 P ≈ D−3 s (15. But now we can compute the entropy of the free string. Thus. (15.0. (15.32) As we continue to decrease the coupling.0.0.0.0.Entropy of Strings and Black Holes 171 Thus for fixed go if the mass is large enough. The evolution from black hole to string can be pictorially represented by starting with a large black hole.0. However. From equation 15.30) ≈ 1 .28) ≈ s g 2 D−2 we can write equation 15. M (g) Since P P = constant. as long as the system remains a black hole. Now start to decrease g.22 we find D−3 S ≈ MoD−3 Go . the weakly coupled string mass will not change significantly.4. D−2 1 (15.0. Let us call the g-dependent mass M (g). In general the mass will vary during an adiabatic process. Note M (go ) = Mo (15.0. The stretched horizon is composed of a . From equation 15.27) The entropy of a Schwarzschild black hole (in any dimension) is a function of the dimensionless variable M P .0.31 we find M (g) s D−3 ≈ MoD−3 Go .0.0. in this calculation the entropy is calculated as the microscopic entropy of fundamental strings. the horizon radius will be much bigger than s .0.28 as M (g) = Mo 2 go g2 1 D−2 . g2 (15.29 and 15.31) Combining equations 15.0.

In this case the extremal black hole is absolutely stable and in addition. (a) A black hole with stringy stretched horizon smaller than Schwarzschild radius. its mass is completely determined by supersymmetry.1b. Finally the horizon radius is no larger than s (Figure 15. the area of the horizon decreases but the depth of the stretched horizon stays fixed as in Figure 15. (b) with stretched horizon and string scale comparable to radius scale. the mass is fixed.1a. and the String Theory Revolution il d sch arz hw Sc R S Fig. Information. The method is always the same. the maximum temperature that a string can achieve. 15.1c) and the black hole turns into a string. As the Schwarzschild radius is decreased (in string units). A particularly interesting situation is that of charged extremal black holes which may be supersymmetric configurations of a supersymmetric theory. Another important property of the stretched horizon is its proper temperature. the temperature of the stringy mass will be TStretched ≈ 1 s This temperature is close to the so-called Hagedorn temperature. Since the proper temperature 1 of a Rindler horizon is 2πρ . When this occurs there is no need to follow the mass of the black hole as g varies.172 Black Holes.1 Evolution from black hole to string. Under these conditions the black hole can be compared . and (c) turned into a string stringy mass to a depth of ρ = s as in the diagram Figure 15. The area 1 density of string is saturated at ∼ G . By now a wide variety of black holes that occur in string theory have been analyzed in this manner. We adiabatically allow g to go to zero and identify the appropriate string configuration that the black hole evolves into.

Nevertheless the principles are that same as those that we used to study the Schwarzschild black hole in D dimensions. recall the behavior of the entropy given by S∼log(density of states). the Hagedorn temperature is the .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. The results in these more complicated examples are in precise agreement with the Hawking–Bekenstein entropy. 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 . If one examines multi-string fluctuations as a function of temperature. we have seen that the entropy of the string scales like Sstring ∼ df Ms ls . where d f is the number of internal degrees of freedom available. Using dimensional considerations. Thus. Hagedorn Temperature Supplement On general grounds. the density of states behaves like √ eSstring ∼ e df Ms ls ∼ e1/THagedorn which gives the scale T Hagedorn ∼ 1/l s . To get a feel for the scale of the Hagedorn temperature. In the cases where exact calculations are possible the charges carried by black holes are associated not with fundamental strings but D-branes.

15. Information.2.2 String “percolation” . and the String Theory Revolution “percolation” temperature for multiple strings fluctuations to coalesce into fluctuations of a single string as represented in Figure 15.174 Black Holes. Increase Temp Increase Temp T<TH T=TH T>TH Fig.

It was never adequate to deal with the combination of quantum mechanics and general relativity. The most fundamental object was the space-time point or better yet. The horizon is harmless and she 175 . The first sign of this was the failure of standard quantum field theory methods when applied to the Einstein action. namely Alice herself. call her Alice. 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. Although quantum mechanics made the event probabilistic and relativity made simultaneity non-absolute.Conclusions The views of space and time that held sway during most of the 20th century were based on locality and field theory. nothing special is felt at the horizon. falling into an enormous black hole with Schwarzschild radius of a billion years. The first is Black Hole Complementarity. 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. But the dilemma of apparent information loss in black hole physics that was uncovered by Hawking in 1976 said otherwise. the event. According to the low frequency observer. But the paradigm is gradually shifting. Radical changes are called for. it was assumed that all observers would agree on the usual invariant relationships between events. 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. or someone falling with her. first classical field theory and later quantum field theory. The new paradigm that is gradually emerging is based on four closely related concepts.

. In many ways this is the most surprising ingredient. At energies above the Planck scale any possible short distance physics that we might look for is shrouded behind a black hole horizon. being thermalized. Very closely related to Black Hole Complementarity. If one wished to study progressively smaller and smaller objects one had to use higher and higher energy probes. It does not tell us what they are. provides an existence proof for a consistent microscopic theory of black hole entropy. there are vastly fewer degrees of freedom than in a field theory. The non-redundant degrees of freedom that describe a region of space are in some sense on its boundary. At one per Planck area. and the String Theory Revolution or her descendants can live for a billion years before being crushed at the singularity. Obviously this has to do with more than just a modification of the short distance physics. and eventually re-emitted as Hawking radiation. the high frequency observer who stays outside the black hole finds that his description involves Alice falling into a hellish region of extreme temperature. Finally. Information. especially for cosmology are undoubtedly profound but still unknown. Third is the Holographic Principle. 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. The number of degrees of freedom per unit volume becomes arbitrarily small as the volume gets large. String theory does provide a microscopic framework for the use of statistical mechanics. All of this takes place just outside the mathematical horizon. But once gravity is involved that trend is reversed. As we have seen. In all cases the entropy of the appropriate string system agrees with the Bekenstein–Hawking entropy. the IR/UV connection reverses one of the most fundamental trends of 20th century physics. This. 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 second new idea is the Infrared/Ultraviolet connection. cutoff at the Planck volume. What is less clear is the dictionary for decoding the CFT hologram. In apparent complete contradiction. In this framework the Holographic Principle. As we raise the energy we wind up probing larger and larger distance scales. if nothing else. The ultimate implications of this. Throughout that century a close connection between energy and size prevailed. the key to black hole complementarity is the extreme red shift of the quantum fluctuations as seen by the external observer. Black Hole Complementarity and the IR/UV connection are completely manifest. not its interior as they would be in field theory.176 Black Holes.

Conclusions 177 The theory of black hole entropy is incomplete. . is used to determine the relation between entropy and mass for the specific string-theoretic object that is believed to represent a particular black hole. an observer in a cosmological setting is very much like one behind a horizon. In each case a trick. Is this important? It is if you are that observer. 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. 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. Then classical general relativity is used to determine the area–mass relation and the Bekenstein–Hawking entropy. And in some ways. Hopefully our next book will have more to say about this. Nor for that matter is there any good idea of how to connect the new paradigm of quantum gravity to cosmology. In this sense the complete universality of the area–entropy relation is still not fully understood.

Information.178 Black Holes. and the String Theory Revolution .

36 (1995) 6377–6396. 2 (1998) 253–291. Phys. MacDonald. hep-th/9806039. Thorne. 7. Price and Douglas A. D55 (1997) 5112–5128. M Theory as a Matrix Model: A Conjecture. J. hep-th/0203101. 3. Black Holes: The Membrane Paradigm (Yale University Press. Susskind and Edward Witten. Phys. 5. Raphael Bousso. Information in Black Hole Radiation. Theor. D53 (1996) 6720–6724. Jacobson. hep-th/9409089. Maldacena. Banks. J. hep-th/9610043. Lett. 6. 10.H. grqc/9602043. Theor. 179 . Math. Phys. Fischler and L. The World as a Hologram. Rev. Leonard Susskind.Bibliography 1. 1986). Rev. hep-th/9805114. Adv. Adv. The Holographic Bound in Anti-de Sitter Space. W. The Holographic Principle. Edward Witten. Corley and T. hep-th/9711200. Richard H. hep-th/9306083. Rev. Math. Anti De Sitter Space and Holography. Phys. hep-th/9802150. Don N. 4. 2 (1998) 231– 252. 2. T. Rev. 74 (2002) 825–874. Focusing and the Holographic Hypothesis. Theor. 71 (1993) 3743–3746. L. Phys. Kip S. 9. Mod. Phys. S. Math. Holography and Cosmology. Phys. Susskind. Shenker and L. Phys. Int. Susskind. Juan M. 38 (1999) 1113–1133. S. 8. The Large N Limit of Superconformal Field Theories and Supergravity. W. Fischler. Page.

180 Black Holes. Information. and the String Theory Revolution .

70. 81 entropy (strings). 119 degrees of freedom. 103 complementarity. 31 coarse graining. 165 entropy (bounds). 89 Birkoff’s theorem. 71 equivalence principle. 55 classical fields. 97. 76 entropy (fine grained). 141 ’t Hooft limit. 119 curvature. 71 effective potential. 15 Boltzmann factor. 165 entropy (Bekenstein–Hawking). 27 anti de Sitter space. 121 brick wall. 168 entropy (coarse grained). 7 cosmological constant. 35. 168 adiabatic variation (string coupling). 34. 165. 138 density matrix. 160 conductivity. 102. 85 entanglement entropy (equality). 35. 106 charge. 81 entropy (maximum). 52 entanglement. 62 electromagnetic field. 77 evaporation. 28 baryon number violation. 52. 72 entropy. 123. 40 Bousso’s construction. 63 energy (horizon). 128 barrier penetration. 102 entropy (quantum field theory). 5 181 D-branes. 69 collapsing light-like shell. 131 conformally flat. 114. 45 entropy (Von Neumann). 54 expansion rate. 73 entropy (infinite). 101 entropy (calculation). 11 . 55 charged black holes. 76 entropy of entanglement. 73. 46. 73 entropy (vacuum). 113 extended horizon. 17 black hole formation. 46. 120. 36.Index AdS(5) ⊗ S(5). 71. 61. 85. 45. 168 entropy (thermal). 48 evaporation time. 136 de Sitter space. 157. 21. 35. 144 AdS/CFT correspondence. 69. 128. 135 adiabatic invariants. 51. 133 angular momentum. 26 electrical properties. 68 conformal field theory. 70. 84 caustic lines. 62 electrostatics. 32. 170 AdS black hole. 31. 43.

79 red shift. 55 resistance. 39. 32. 58. 41. 20. 77 Kruskal–Szekeres coordinates. 112 path integral. 142 Penrose diagram (de Sitter space). 129 ground state (charged black hole). 21.R. 110 fluctuations. 23. and the String Theory Revolution extremal black hole. 107 free field approximation. 161 fiducial observer (see also Fidos). Information. 127 Poincar´ disk. 85 high frequency phenomena. 3 .W. 170 Planck units. 37 Penrose diagram (AdS black hole). 25. 8. 127. 44. 55 particle horizon. 81 information retention time. 74. 22 geodesic completeness. 32. 81 Hawking radiation. 5 inflation. 44 quantum field theory. 49 freely falling observer (see also Frefos). 43 quantum fields. 59 Hagedorn temperature. 151 holographic principle. 41. 31. 152 hyperbolic plane. 31 quantum Xerox principle. 127 holography (AdS space). 151 quantum fields (Rindler space). 101 laws of nature. 69. 121 Reissner–Nordstrom black hole. 10 lattice of discrete spins. 120 Penrose diagram (F. 49. 93 focusing theorem. 52. 152. 8 near horizon wave equation. 48 reheating. 89 quantization rules. 9. 81 scalar wave equation. 69 longitudinal string motions. 22 Frefos. 44 Rindler space. 21 Fischler–Susskind bound. 134. 97. 54 Minkowski space. 32 near horizon coordinates. 8. 41. 156 Liouville’s theorem.182 Black Holes. 52. space). 144. 101. 167 S-matrix. 167 proton decay. 101. 85 Friedman–Robertson–Walker geometry. 172 Hawking. 153 light cone string theory. 134 no-cloning principle. 123 Planck distance. 37 Fidos. 130 infalling observer. 56 Feynman–Hellman theorem. 44 momentum. 130 holography. 62. 69 level density. 130 e proper distance. 25 Schwarzschild black hole. 115 Penrose diagrams. 53 Rindler Hamiltonian. 23 Planck length. 110 gauge fixing. 121 information. 163 light cone quantum mechanics. 52 light cone gauge fixing. 4. 21. 79 pair production. 102. 8. 106 mirror boundary condition. 14 Penrose–Bousso diagram. 144 information conservation. 48. 61. 130 horizon. 69. 28 Newton’s constant. 65 Rindler energy. 25. 57. 161 luminosity. 60. 117 Penrose–Bousso diagram (AdS).

39 thermal atmosphere. 42. 48. 7. 135 vacuum. 28 transfer matrix. 155 Unruh. 39 thermal fluctuations. 134 string interactions. 21 stretched horizon. 133 supersymmetry. 174 strings. 161 string coupling constant. 5 tortoise coordinates. 101. 141 thermodynamics.Index 183 Schwarzschild coordinates. 61. 39 temperature (Rindler). 166 second law of thermodynamics. 39 UV/IR connection. 38 transverse spreading. 63 temperature (Hawking). 36. 48 thermal ensemble. 95. 39 static observers. 167 temperature (proper). 151 Supersymmetric Yang–Mills (SYM). 131. 103 solid angle. 62. 144 thermodynamic instability. 3 Schwarzschild geometry (D-dimensions). 45. 167 standard thermometer. 133 surface charge density. 3. 159 string percolation. 65 tortoise-like coordinates. 51 tidal forces. 45 .

Sign up to vote on this title
UsefulNot useful