1: %\documentclass[aps,rmp,eqsecnum,amssymb,preprint,draft]{revtex4}
2: \documentclass[aps,rmp,twocolumn,eqsecnum,amssymb]{revtex4}
3: \usepackage{graphicx}
4: %\usepackage{showkeys}
5: \hyphenation{Fisch-ler Schwarz-schild Suss-kind}
6: \begin{document}
7: \title{The holographic principle}
8: \author{Raphael Bousso}
9: \affiliation{Institute for Theoretical Physics,
10: University of California, Santa Barbara, California 93106, U.S.A.}
11: \email{bousso@itp.ucsb.edu}
12: \begin{abstract}
13:
14: There is strong evidence that the area of any surface limits the
15: information content of adjacent spacetime regions, at $1.4 \times
16: 10^{69}$ bits per square meter. We review the developments that have
17: led to the recognition of this entropy bound, placing special emphasis
18: on the quantum properties of black holes. The construction of
19: light-sheets, which associate relevant spacetime regions to any given
20: surface, is discussed in detail. We explain how the bound is tested
21: and demonstrate its validity in a wide range of examples.
22:
23: A universal relation between geometry and information is thus
24: uncovered. It has yet to be explained. The holographic principle
25: asserts that its origin must lie in the number of fundamental degrees
26: of freedom involved in a unified description of spacetime and matter.
27: It must be manifest in an underlying quantum theory of gravity. We
28: survey some successes and challenges in implementing the holographic
29: principle.
30:
31: \end{abstract}
32: \maketitle
33: \tableofcontents
34:
35:
36:
37:
38:
39:
40: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
41: \section{Introduction}
42: \label{sec-intro}
43: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
44:
45:
46: \subsection{A principle for quantum gravity}
47: \label{sec-ia}
48:
49: Progress in fundamental physics has often been driven by the
50: recognition of a new principle, a key insight to guide the search for
51: a successful theory. Examples include the principles of relativity,
52: the equivalence principle, and the gauge principle. Such principles
53: lay down general properties that must be incorporated into the laws of
54: physics.
55:
56: A principle can be sparked by contradictions between existing
57: theories. By judiciously declaring which theory contains the elements
58: of a unified framework, a principle may force other theories to be
59: adapted or superceded. The special theory of relativity, for example,
60: reconciles electrodynamics with Galilean kinematics at the expense of
61: the latter.
62:
63: A principle can also arise from some newly recognized pattern, an
64: apparent law of physics that stands by itself, both uncontradicted and
65: unexplained by existing theories. A principle may declare this
66: pattern to be at the core of a new theory altogether.
67:
68: In Newtonian gravity, for example, the proportionality of
69: gravitational and inertial mass in all bodies seems a curious
70: coincidence that is far from inevitable. The equivalence principle
71: demands that this pattern must be made manifest in a new theory. This
72: led Einstein to the general theory of relativity, in which the
73: equality of gravitational and inertial mass is built in from the
74: start. Because all bodies follow geodesics in a curved spacetime,
75: things simply couldn't be otherwise.
76:
77: The {\em holographic principle\/} belongs in the latter class. The
78: unexplained ``pattern'', in this case, is the existence of a precise,
79: general, and surprisingly strong limit on the information content of
80: spacetime regions. This pattern has come to be recognized in stages;
81: its present, most general form is called the {\em covariant entropy
82: bound}. The holographic principle asserts that this bound is not a
83: coincidence, but that its origin must be found in a new theory.
84:
85: The covariant entropy bound relates aspects of spacetime geometry to
86: the number of quantum states of matter. This suggests that any theory
87: that incorporates the holographic principle must unify matter,
88: gravity, and quantum mechanics. It will be a quantum theory of
89: gravity, a framework that transcends general relativity and quantum
90: field theory.
91:
92: This expectation is supported by the close ties between
93: the covariant entropy bound and the semi-classical properties of black
94: holes. It has been confirmed---albeit in a limited context---by
95: recent results in string theory.
96:
97: The holographic principle conflicts with received wisdom; in this
98: sense, it also belongs in the former class. Conventional theories are
99: local; quantum field theory, for example, contains degrees of freedom
100: at every point in space. Even with a short distance cutoff, the
101: information content of a spatial region would appear to grow with the
102: volume. The holographic principle, on the other hand, implies that
103: the number of fundamental degrees of freedom is related to the area of
104: surfaces in spacetime. Typically, this number is drastically smaller
105: than the field theory estimate.
106:
107: Thus, the holographic principle calls into question not only the
108: fundamental status of field theory but the very notion of locality.
109: It gives preference, as we shall see, to the preservation of
110: quantum-mechanical unitarity.
111:
112: In physics, information can be encoded in a variety of ways: by the
113: quantum states, say, of a conformal field theory, or by a lattice of
114: spins. Unfortunately, for all its precise predictions about the {\em
115: number\/} of fundamental degrees of freedom in spacetime, the
116: holographic principle betrays little about their character. The
117: amount of information is strictly determined, but not its form. It is
118: interesting to contemplate the notion that pure, abstract information
119: may underlie all of physics. But for now, this austerity frustrates
120: the design of concrete models incorporating the holographic principle.
121:
122: Indeed, a broader caveat is called for. The covariant entropy bound
123: is a compelling pattern, but it may still prove incorrect or merely
124: accidental, signifying no deeper origin. If the bound does stem from
125: a fundamental theory, that relation could be indirect or peripheral,
126: in which case the holographic principle would be unlikely to guide us
127: to the core ideas of the theory. All that aside, the holographic
128: principle is likely only one of several independent conceptual
129: advances needed for progress in quantum gravity.
130:
131: At present, however, quantum gravity poses an immense problem tackled
132: with little guidance. Quantum gravity has imprinted few traces on
133: physics below the Planck energy. Among them, the information content
134: of spacetime may well be the most profound.
135:
136: The direction offered by the holographic principle is impacting
137: existing frameworks and provoking new approaches. In particular, it
138: may prove beneficial to the further development of string theory,
139: widely (and, in our view, justly) considered the most compelling of
140: present approaches.
141:
142: This article will outline the case for the holographic principle whilst
143: providing a starting point for further study of the literature. The
144: material is not, for the most part, technical. The main mathematical
145: aspect, the construction of light-sheets, is rather straightforward.
146: In order to achieve a self-contained presentation, some basic material
147: on general relativity has been included in an appendix.
148:
149: In demonstrating the scope and power of the holographic correspondence
150: between areas and information, our ultimate task is to convey its
151: character as a law of physics that captures one of the most intriguing
152: aspects of quantum gravity. If the reader is led to contemplate the
153: origin of this particular pattern nature has laid out, our review will
154: have succeeded.
155:
156:
157: \subsection{Notation and conventions}
158: \label{sec-not}
159:
160: Throughout this paper, Planck units will be used:
161: \begin{equation}
162: \hbar = G=c=k=1,
163: \end{equation}
164: where $G$ is Newton's constant, $\hbar$ is Planck's constant, $c$
165: is the speed of light, and $k$ is Boltzmann's constant. In particular, all
166: areas are measured in multiples of the square of the Planck length,
167: \begin{equation}
168: l_{\rm P}^2 = {G\hbar\over c^3} = 2.59 \times 10^{-66} \mbox{cm}^2.
169: \label{eq-lpl}
170: \end{equation}
171: The Planck units of energy density, mass, temperature, and other
172: quantities are converted to cgs units, e.g., in Wald (1984), whose
173: conventions we follow in general. For a small number of key formulas,
174: we will provide an alternate expression in which all constants are
175: given explicitly.
176:
177: We consider spacetimes of arbitrary dimension $D\geq 4$, unless noted
178: otherwise. In explicit examples we often take $D=4$ for definiteness.
179: The appendix fixes the metric signature and defines ``surface'',
180: ``hypersurface'', ``null'', and many other terms from general
181: relativity. The term ``light-sheet'' is defined in Sec.~\ref{sec-bb}.
182:
183: ``GSL'' stands for the generalized second law of thermodynamics
184: (Sec.~\ref{sec-gsl}). The number of degrees of freedom of a quantum
185: system, $N$, is defined as the logarithm of the dimension, ${\cal N}$,
186: of its Hilbert space in Sec.~\ref{sec-ndof}. Equivalently, $N$ can be
187: defined as the number of bits of information times $\ln 2$.
188:
189: \subsection{Outline}
190: \label{sec-outline}
191:
192: In Sec.~\ref{sec-bh}, we review Bekenstein's (1972) notion of black
193: hole entropy and the related discovery of upper bounds on the entropy
194: of matter systems. Assuming weak gravity, spherical symmetry, and
195: other conditions, one finds that the entropy in a region of space is
196: limited by the area of its boundary.%
197: %
198: \footnote{The metaphorical name of the principle ('t~Hooft, 1993)
199: originates here. In many situations, the covariant entropy bound
200: dictates that all physics in a region of space is described by data
201: that fit on its boundary surface, at one bit per Planck area
202: (Sec.~\ref{sec-spt}). This is reminiscent of a hologram. Holography
203: is an optical technology by which a three-dimensional image is stored
204: on a two-dimensional surface via a diffraction pattern. (To avoid any
205: confusion: this linguistic remark will remain our only usage of the
206: term in its original sense.) From the present point of view, the
207: analogy has proven particularly apt. In both kinds of ``holography'',
208: light rays play a key role for the imaging (Sec.~\ref{sec-bb}).
209: Moreover, the holographic code is not a straightforward projection, as
210: in ordinary photography; its relation to the three-dimensional image
211: is rather complicated. (Most of our intuition in this regard has come
212: from the AdS/CFT correspondence, Sec.~\ref{sec-ads}.) Susskind's
213: (1995) quip that the world is a ``hologram'' is justified by the
214: existence of preferred surfaces in spacetime, on which all of the
215: information in the universe can be stored at no more than one bit per
216: Planck area (Sec.~\ref{sec-screens}).}
217: %
218: Based on this ``spherical entropy bound'', 't~Hooft (1993) and
219: Susskind (1995b) formulated a holographic principle. We discuss
220: motivations for this radical step.
221:
222: The spherical entropy bound depends on assumptions that are clearly
223: violated by realistic physical systems. {\em A priori\/} there is no
224: reason to expect that the bound has universal validity, nor that it
225: admits a reformulation that does. Yet, if the number of degrees of
226: freedom in nature is as small as 't~Hooft and Susskind asserted, one
227: would expect wider implications for the maximal entropy of matter.
228:
229: In Sec.~\ref{sec-seb}, however, we demonstrate that a naive
230: generalization of the spherical entropy bound is unsuccessful. The
231: ``spacelike entropy bound'' states that the entropy in a given spatial
232: volume, irrespective of shape and location, is always less than the
233: surface area of its boundary. We consider four examples of realistic,
234: commonplace physical systems, and find that the spacelike entropy
235: bound is violated in each one of them.
236:
237: In light of these difficulties, some authors, forgoing complete
238: generality, searched instead for reliable conditions under which the
239: spacelike entropy bound holds. We review the difficulties faced in
240: making such conditions precise even in simple cosmological models.
241:
242: Thus, the idea that the area of surfaces generally bounds the entropy
243: in enclosed spatial volumes has proven wrong; it can be neither the
244: basis nor the consequence of a fundamental principle. This review
245: would be incomplete if it failed to stress this point. Moreover, the
246: ease with which the spacelike entropy bound (and several of its
247: modifications) can be excluded underscores that a general entropy
248: bound, if found, is no triviality. The counterexamples to the
249: spacelike bound later provide a useful testing ground for the
250: covariant bound.
251:
252:
253: Inadequacies of the spacelike entropy bound led Fischler and Susskind
254: (1998) to a bound involving light cones. The covariant entropy bound
255: (Bousso, 1999a), presented in Sec.~\ref{sec-bb}, refines and
256: generalizes this approach. Given any surface $B$, the bound states
257: that the entropy on any light-sheet of $B$ will not exceed the area of
258: $B$. Light-sheets are particular hypersurfaces generated by light
259: rays orthogonal to $B$. The light rays may only be followed as long
260: as they are not expanding. We explain this construction in detail.
261:
262: After discussing how to define the entropy on a light-sheet, we spell
263: out known limitations of the covariant entropy bound. The bound is
264: presently formulated only for approximately classical geometries, and
265: one must exclude unphysical matter content, such as large negative
266: energy. We conclude that the covariant entropy bound is well-defined
267: and testable in a vast class of solutions. This includes all
268: thermodynamic systems and cosmologies presently known or considered
269: realistic.
270:
271: In Sec.~\ref{sec-lsd} we review the geometric properties of
272: light-sheets, which are central to the operation of the covariant
273: entropy bound. Raychaudhuri's equation is used to analyse the effects
274: of entropy on light-sheet evolution. By construction, a light-sheet
275: is generated by light rays that are initially either parallel or
276: contracting. Entropic matter systems carry mass, which causes the
277: bending of light.
278:
279: This means that the light rays generating a light-sheet will be
280: focussed towards each other when they encounter entropy. Eventually
281: they self-intersect in a caustic, where they must be terminated
282: because they would begin to expand. This mechanism would provide an
283: ``explanation'' of the covariant entropy bound if one could show that
284: the mass associated with entropy is necessarily so large that
285: light-sheets focus and terminate before they encounter more entropy
286: than their initial area.
287:
288: Unfortunately, present theories do not impose an independent,
289: fundamental lower bound on the energetic price of entropy. However,
290: Flanagan, Marolf, and Wald (2000) were able to identify conditions on
291: entropy density which are widely satisfied in nature and which are
292: sufficient to guarantee the validity of the covariant entropy bound.
293: We review these conditions.
294:
295: The covariant bound can also be used to obtain sufficient criteria
296: under which the spacelike entropy bound holds. Roughly, these
297: criteria can be summarized by demanding that gravity be weak.
298: However, the precise condition requires the construction of
299: light-sheets; it cannot be formulated in terms of intrinsic properties
300: of spatial volumes.
301:
302: The event horizon of a black hole is a light-sheet of its final
303: surface area. Thus, the covariant entropy bound includes to the
304: generalized second law of thermodynamics in black hole formation as a
305: special case. More broadly, the generalized second law, as well as
306: the Bekenstein entropy bound, follow from a strengthened version of
307: the covariant entropy bound.
308:
309: In Sec.~\ref{sec-tests}, the covariant entropy bound is applied to a
310: variety of thermodynamic systems and cosmological spacetimes. This
311: includes all of the examples in which the spacelike entropy bound is
312: violated. We find that the covariant bound is satisfied in each case.
313:
314: In particular, the bound is found to hold in strongly gravitating
315: regions, such as cosmological spacetimes and collapsing objects.
316: Aside from providing evidence for the general validity of the bound,
317: this demonstrates that the bound (unlike the spherical entropy bound)
318: holds in a regime where it cannot be derived from black hole
319: thermodynamics.
320:
321: In Sec.~\ref{sec-hp}, we arrive at the holographic principle. We note
322: that the covariant entropy bound holds with remarkable generality but
323: is not logically implied by known laws of physics. We conclude that
324: the bound has a fundamental origin. As a universal limitation on the
325: information content of Lorentzian geometry, the bound should be
326: manifest in a quantum theory of gravity. We formulate the holographic
327: principle and list some of its implications. The principle poses a
328: challenge for local theories. It suggests a preferred role for null
329: hypersurfaces in the classical limit of quantum gravity.
330:
331: In Sec.~\ref{sec-hsht} we analyze an example of a holographic theory.
332: Quantum gravity in certain asymptotically Anti-de~Sitter spacetimes is
333: fully defined by a conformal field theory. The latter theory contains
334: the correct number of degrees of freedom demanded by the holographic
335: principle. It can be thought of as living on a kind of holographic
336: screen at the boundary of spacetime and containing one bit of
337: information per Planck area.
338:
339: Holographic screens with this information density can be constructed
340: for arbitrary spacetimes---in this sense, the world is a hologram. In
341: most other respects, however, global holographic screens do not
342: generally support the notion that a holographic theory is a
343: conventional field theory living at the boundary of a spacetime.
344:
345: At present, there is much interest in finding more general holographic
346: theories. We discuss the extent to which string theory, and a number
347: of other approaches, have realized this goal. A particular area of
348: focus is de~Sitter space, which exhibits an absolute entropy bound.
349: We review the implications of the holographic principle in such
350: spacetimes.
351:
352:
353: \subsection{Related subjects and further reading}
354: \label{sec-further}
355:
356: The holographic principle has developed from a large set of ideas and
357: results, not all of which seemed mutually related at first. This is
358: not a historical review; we have aimed mainly at achieving a coherent,
359: modern perspective on the holographic principle. We do not give equal
360: emphasis to all developments, and we respect the historical order only
361: where it serves the clarity of exposition. Along with length
362: constraints, however, this approach has led to some omissions and
363: shortcomings, for which we apologize.
364:
365: We have chosen to focus on the covariant entropy bound because it can
366: be tested using only quantum field theory and general relativity. Its
367: universality motivates the holographic principle independently of any
368: particular ansatz for quantum gravity (say, string theory) and without
369: additional assumptions (such as unitarity). It yields a precise and
370: general formulation.
371:
372: Historically, the idea of the holographic principle was tied, in part,
373: to the debate about information loss in black holes\footnote{See, for
374: example, Hawking (1976b, 1982), Page (1980, 1993), Banks, Susskind,
375: and Peskin (1984), 't~Hooft (1985, 1988, 1990), Polchinski and
376: Strominger (1994), Strominger (1994).} and to the notion of black
377: hole complementarity.\footnote{See, e.g., 't~Hooft (1991), Susskind,
378: Thorlacius, and Uglum (1993), Susskind (1993b), Stephens, 't~Hooft,
379: and Whiting (1994), Susskind and Thorlacius (1994). For recent
380: criticism, see Jacobson (1999).} Although we identify some of the
381: connections, our treatment of these issues is far from comprehensive.
382: Reviews include Thorlacius (1995), Verlinde (1995), Susskind and Uglum
383: (1996), Bigatti and Susskind (2000), and Wald (2001).
384:
385: Some aspects of what we now recognize as the holographic principle
386: were encountered, at an early stage, as features of string theory.
387: (This is as it should be, since string theory is a quantum theory of
388: gravity.) In the infinite momentum frame, the theory admits a
389: lower-dimensional description from which the gravitational dynamics of
390: the full spacetime arises non-trivially (Giles and Thorn, 1977; Giles,
391: McLerran, and Thorn, 1978; Thorn, 1979, 1991, 1995, 1996; Klebanov and
392: Susskind, 1988). Susskind (1995b) placed this property of string
393: theory in the context of the holographic principle and related it to
394: black hole thermodynamics and entropy limitations.
395:
396: Some authors have traced the emergence of the holographic principle
397: also to other approaches to quantum gravity; see Smolin (2001) for a
398: discussion and further references.
399:
400: By tracing over a region of space one obtains a density matrix.
401: Bombelli {\em et al.} (1986) showed that the resulting entropy is
402: proportional to the boundary area of the region. A more general
403: argument was given by Srednicki (1993). Gravity does not enter in
404: this consideration. Moreover, the entanglement entropy is generally
405: unrelated to the size of the Hilbert space describing either side of
406: the boundary. Thus, it is not clear to what extent this suggestive
407: result is related to the holographic principle.
408:
409: This is not a review of the AdS/CFT correspondence (Maldacena, 1998;
410: see also Gubser, Klebanov, and Polyakov, 1998; Witten, 1998). This
411: rich and beautiful duality can be regarded (among its many interesting
412: aspects) as an implementation of the holographic principle in a
413: concrete model. Unfortunately, it applies only to a narrow class of
414: spacetimes of limited physical relevance. By contrast, the
415: holographic principle claims a far greater level of generality---a
416: level at which it continues to lack a concrete implementation.
417:
418: We will broadly discuss the relation between the AdS/CFT
419: correspondence and the holographic principle, but we will not dwell on
420: aspects that seem particular to AdS/CFT. (In particular, this means
421: that the reader should not expect a discussion of every paper
422: containing the word ``holographic'' in the title!) A detailed
423: treatment of AdS/CFT would go beyond the purpose of the present text.
424: An extensive review has been given by Aharony {\em et al.} (2000).
425:
426: The AdS/CFT correspondence is closely related to some recent models of
427: our 3+1 dimensional world as a defect, or brane, in a 4+1 dimensional
428: AdS space. In the models of Randall and Sundrum (1999a,b), the
429: gravitational degrees of freedom of the extra dimension appear on the
430: brane as a dual field theory under the AdS/CFT correspondence. While
431: the holographic principle can be considered a prerequisite for the
432: existence of such models, their detailed discussion would not
433: significantly strengthen our discourse. Earlier seminal papers in
434: this area include Ho\v{r}ava and Witten (1996a,b).
435:
436: A number of authors (e.g., Brustein and Veneziano, 2000; Verlinde,
437: 2000; Brustein, Foffa, and Veneziano, 2001; see Cai, Myung, and Ohta,
438: 2001, for additional references) have discussed interesting bounds
439: which are not directly based on the area of surfaces. Not all of
440: these bounds appear to be universal. Because their relation to the
441: holographic principle is not entirely clear, we will not attempt to
442: discuss them here. Applications of entropy bounds to string cosmology
443: (e.g., Veneziano, 1999a; Bak and Rey, 2000b; Brustein, Foffa, and
444: Sturani, 2000) are reviewed by Veneziano (2000).
445:
446: The holographic principle has sometimes been said to exclude certain
447: physically acceptable solutions of Einstein's equations because they
448: appeared to conflict with an entropy bound. The covariant bound has
449: exposed these tensions as artifacts of the limitations of earlier
450: entropy bounds. Indeed, this review bases the case for a holographic
451: principle to a large part on the very generality of the covariant
452: bound. However, the holographic principle does limit the
453: applicability of quantum field theory on cosmologically large scales.
454: It calls into question the conventional analysis of the cosmological
455: constant problem (Cohen, Kaplan, and Nelson, 1999; Ho\v{r}ava, 1999;
456: Banks, 2000a; Ho\v{r}ava and Minic, 2000; Thomas, 2000). It has also
457: been applied to the calculation of anisotropies in the cosmic
458: microwave background (Hogan, 2002a,b). The study of cosmological
459: signatures of the holographic principle may be of great value, since
460: it is not clear whether more conventional imprints of short-distance
461: physics on the early universe are observable even in principle (see,
462: e.g., Kaloper {\em et al.}, 2002, and references therein).
463:
464: Most attempts at implementing the holographic principle in a unified
465: theory are still in their infancy. It would be premature to attempt a
466: detailed review; some references are given in Sec.~\ref{sec-toe}.
467:
468: Other recent reviews overlapping with some or all of the topics
469: covered here are Bigatti and Susskind (2000), Bousso (2000a), 't~Hooft
470: (2000b), Bekenstein (2001) and Wald (2001). Relevant textbooks include
471: Hawking and Ellis (1973); Misner, Thorne, and Wheeler (1973); Wald
472: (1984, 1994); Green, Schwarz, and Witten (1987); and Polchinski
473: (1998).
474:
475:
476:
477:
478: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
479: \section{Entropy bounds from black holes}
480: \label{sec-bh}
481: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
482:
483: This section reviews black hole entropy, some of the entropy bounds
484: that have been inferred from it, and their relation to 't~Hooft's
485: (1993) and Susskind's (1995b) proposal of a holographic principle.
486:
487: The entropy bounds discussed in this section are ``universal''
488: (Bekenstein, 1981) in the sense that they are independent of the
489: specific characteristics and composition of matter systems. Their
490: validity is not truly universal, however, because they apply only when
491: gravity is weak.
492:
493: We consider only Einstein gravity. For black hole thermodynamics in
494: higher-derivative gravity, see, e.g., Myers and Simon (1988), Jacobson
495: and Myers (1993), Wald (1993), Iyer and Wald (1994, 1995), Jacobson,
496: Kang, and Myers (1994), and the review by Myers
497: (1998).\footnote{Abdalla and Correa-Borbonet (2001) have commented on
498: entropy bounds in this context.}
499:
500:
501: \subsection{Black hole thermodynamics}
502: \label{sec-bhtd}
503:
504: The notion of black hole entropy is motivated by two results in
505: general relativity.
506:
507: \subsubsection{Area theorem}
508: \label{sec-areathm}
509:
510: The {\sl area theorem\/} (Hawking, 1971) states that {\em the area of
511: a black hole event horizon never decreases with time:}
512: \begin{equation}
513: dA \geq 0.
514: \label{eq-areathm}
515: \end{equation}
516: Moreover, if two black holes merge, the area of the new black hole
517: will exceed the total area of the original black holes.
518:
519: For example, an object falling into a Schwarzschild black hole will
520: increase the mass of the black hole, $M$.%
521: %
522: \footnote{This assumes that the object has positive mass. Indeed, the
523: assumptions in the proof of the theorem include the null energy
524: condition. This condition is given in the Appendix, where the
525: Schwarzschild metric is also found.}
526: %
527: Hence the horizon area, $A=16\pi M^2$ in $D=4$, increases. On the
528: other hand, one would not expect the area to decrease in any classical
529: process, because the black hole cannot emit particles.
530:
531: The theorem suggests an analogy between black hole area and
532: thermodynamic entropy.
533:
534: \subsubsection{No-hair theorem}
535: \label{sec-no-hair}
536:
537: Work of Israel (1967, 1968), Carter (1970), Hawking (1971, 1972), and
538: others, implies the curiously named {\sl no-hair theorem}: {\em A
539: stationary black hole is characterized by only three quantities: mass,
540: angular momentum, and charge.}\footnote{Proofs and further details can
541: be found, e.g., in Hawking and Ellis (1973), or Wald (1984). This
542: form of the theorem holds only in $D=4$. Gibbons, Ida, and Shiromizu
543: (2002) have recently given a uniqueness proof for static black holes
544: in $D>4$. Remarkably, Emparan and Reall (2001) have found a
545: counterexample to the stationary case in $D=5$. This does not affect
546: the present argument, in which the no hair theorem plays a heuristic
547: role.}
548:
549: Consider a complex matter system, such as a star, that collapses to
550: form a black hole. The black hole will eventually settle down into a
551: final, stationary state. The no-hair theorem implies that this state
552: is unique.
553:
554: From an outside observer's point of view, the formation of a black
555: hole appears to violate the second law of thermodynamics. The phase
556: space appears to be drastically reduced. The collapsing system may
557: have arbitrarily large entropy, but the final state has none at all.
558: Different initial conditions will lead to indistinguishable results.
559:
560: A similar problem arises when a matter system is dropped into an
561: existing black hole. Geroch has proposed a further method for
562: violating the second law, which exploits a classical black hole to
563: transform heat into work; see Bekenstein (1972) for details.
564:
565:
566: \subsubsection{Bekenstein entropy and the generalized second law}
567: \label{sec-gsl}
568:
569: Thus, the no-hair theorem poses a paradox, to which the area theorem
570: suggests a resolution. When a thermodynamic system disappears behind
571: a black hole's event horizon, its entropy is lost to an outside
572: observer. The area of the event horizon will typically grow when the
573: black hole swallows the system. Perhaps one could regard this area
574: increase as a kind of compensation for the loss of matter entropy?
575:
576: Based on this reasoning, Bekenstein (1972, 1973, 1974) suggested that
577: a black hole actually carries an entropy equal to its horizon area,
578: $S_{\rm BH} = \eta A$, where $\eta$ is a number of order unity. In
579: Sec.~\ref{sec-bh-radiation} it will be seen that $\eta={1\over 4}$
580: (Hawking, 1974):
581: \begin{equation}
582: S_{\rm BH} = {A\over 4}.
583: \label{eq-eta}
584: \end{equation}
585: [In full, $S_{\rm BH} = kA c^3 / (4G \hbar)$.] The entropy of a black
586: hole is given by a quarter of the area of its horizon in Planck units.
587: In ordinary units, it is the horizon area divided by about
588: $10^{-69}$\/m$^2$.
589:
590: Moreover, Bekenstein (1972, 1973, 1974) proposed that the second law
591: of thermodynamics holds only for the {\em sum\/} of black hole entropy
592: and matter entropy:
593: \begin{equation}
594: dS_{\rm total} \geq 0.
595: \label{eq-gsl}
596: \end{equation}
597: For ordinary matter systems alone, the second law need not hold. But
598: if the entropy of black holes, Eq.~(\ref{eq-eta}), is included in the
599: balance, the total entropy will never decrease. This is referred to
600: as the {\em generalized second law\/} or {\em GSL}.
601:
602: The content of this statement may be illustrated as follows. Consider
603: a thermodynamic system ${\cal T}$, consisting of well-separated,
604: non-interacting components. Some components, labeled ${\cal C}_i$,
605: may be thermodynamic systems made from ordinary matter, with entropy
606: $S({\cal C}_i)$. The other components, ${\cal B}_j$, are black holes,
607: with horizon areas $A_j$. The total entropy of ${\cal T}$ is given by
608: \begin{equation}
609: S_{\rm total}^{\rm initial} = S_{\rm matter} + S_{\rm BH}.
610: \end{equation}
611: Here, $S_{\rm matter} = \sum S({\cal C}_i)$ is the total entropy of
612: all ordinary matter. $S_{\rm BH} = \sum {A_j\over 4}$ is the total
613: entropy of all black holes present in ${\cal T}$.
614:
615: Now suppose the components of ${\cal T}$ are allowed to interact until a
616: new equilibrium is established. For example, some of the matter
617: components may fall into some of the black holes. Other matter
618: components might collapse to form new black holes. Two or more black
619: holes may merge. In the end, the system ${\cal T}$ will consist of a
620: new set of components $\hat{\cal C}_i$ and $\hat{\cal B}_j$, for which
621: one can again compute a total entropy, $S_{\rm total}^{\rm final}$.
622: The GSL states that
623: \begin{equation}
624: S_{\rm total}^{\rm final} \geq S_{\rm total}^{\rm initial}.
625: \end{equation}
626:
627: What is the microscopic, statistical origin of black hole entropy? We
628: have learned that a black hole, viewed from the outside, is unique
629: classically. The Bekenstein-Hawking formula, however, suggests that
630: it is compatible with $e^{S_{\rm BH}}$ independent quantum states.
631: The nature of these quantum states remains largely mysterious. This
632: problem has sparked sustained activity through various different
633: approaches, too vast in scope to sketch in this review.
634:
635: However, one result stands out because of its quantitative accuracy.
636: Recent developments in string theory have led to models of limited
637: classes of black holes in which the microstates can be identified and
638: counted (Strominger and Vafa, 1996; for a review, see, e.g., Peet,
639: 2000). The formula $S=A/4$ was precisely confirmed by this
640: calculation.
641:
642:
643: \subsubsection{Hawking radiation}
644: \label{sec-bh-radiation}
645:
646: Black holes clearly have a mass, $M$. If Bekenstein entropy, $S_{\rm
647: BH}$, is to be taken seriously, then the first law of thermodynamics
648: dictates that black holes must have a temperature, $T$:
649: \begin{equation}
650: dM = T dS_{\rm BH}.
651: \label{eq-fl}
652: \end{equation}
653: Indeed, Einstein's equations imply an analogous ``first law of black
654: hole mechanics'' (Bardeen, Carter, and Hawking, 1973). The entropy is
655: the horizon area, and the surface gravity of the black hole, $\kappa$,
656: plays the role of the temperature:
657: \begin{equation}
658: dM = \frac{\kappa}{8\pi} dA.
659: \end{equation}
660: For a definition of $\kappa$, see Wald (1984); e.g., a Schwarzschild
661: black hole in $D=4$ has $\kappa = (4M)^{-1}$.
662:
663: It may seem that this has taken the thermodynamic analogy a step too
664: far. After all, a blackbody with non-zero temperature must radiate.
665: But for a black hole this would seem impossible. Classically, no
666: matter can escape from it, so its temperature must be exactly zero.
667:
668: This paradox was resolved by the discovery that black holes do in fact
669: radiate via a quantum process. Hawking (1974, 1975) showed by a
670: semi-classical calculation that a distant observer will detect a
671: thermal spectrum of particles coming from the black hole, at a
672: temperature
673: \begin{equation}
674: T = \frac{\kappa}{2\pi}.
675: \end{equation}
676: For a Schwarzschild black hole in $D=4$, this temperature is $\hbar
677: c^3 / (8 \pi G k M)$, or about $10^{26}$ Kelvin divided by the mass of
678: the black hole in grams. Note that such black holes have negative
679: specific heat.
680:
681: The discovery of Hawking radiation clarified the interpretation of the
682: thermodynamic description of black holes. What might otherwise have
683: been viewed as a mere analogy (Bardeen, Carter, and Hawking, 1973) was
684: understood to be a true physical property. The entropy and
685: temperature of a black hole are no less real than its mass.
686:
687: In particular, Hawking's result affirmed that the entropy of black
688: holes should be considered a genuine contribution to the total entropy
689: content of the universe, as Bekenstein (1972, 1973, 1974) had
690: anticipated. Via the first law of thermodynamics, Eq.~(\ref{eq-fl}),
691: Hawking's calculation fixes the coefficient $\eta$ in the Bekenstein
692: entropy formula, Eq.~(\ref{eq-eta}), to be $1/4$.
693:
694: A radiating black hole loses mass, shrinks, and eventually disappears
695: unless it is stabilized by charge or a steady influx of energy. Over
696: a long time of order $M^{D-1\over D-3}$, this process converts the
697: black hole into a cloud of radiation. (See Sec.~\ref{sec-bhc} for the
698: question of unitarity in this process.)
699:
700: It is natural to study the operation of the GSL in the two types of
701: processes discussed in Sec.~\ref{sec-no-hair}. We will first discuss
702: the case in which a matter system is dropped into an existing black
703: hole. Then we will turn to the process in which a black hole is
704: formed by the collapse of ordinary matter. In both cases, ordinary
705: entropy is converted into horizon entropy.
706:
707: A third process, which we will not discuss in detail, is the Hawking
708: evaporation of a black hole. In this case, the horizon entropy is
709: converted back into radiation entropy. This type of process was not
710: anticipated when Bekenstein (1972) proposed black hole entropy and the
711: GSL. It is all the more impressive that the GSL holds also in this
712: case (Bekenstein, 1975; Hawking, 1976a). Page (1976) has estimated
713: that the entropy of Hawking radiation exceeds that of the evaporated
714: black hole by 62\%.
715:
716:
717: \subsection{Bekenstein bound}
718: \label{sec-bekbound}
719:
720: When a matter system is dropped into a black hole, its entropy is lost
721: to an outside observer. That is, the entropy $S_{\rm matter}$ starts
722: at some finite value and ends up at zero. But the entropy of the
723: black hole increases, because the black hole gains mass, and so its
724: area $A$ will grow. Thus it is at least conceivable that the total
725: entropy, $S_{\rm matter} + \frac{A}{4}$, does not decrease in the
726: process, and that therefore the generalized second law of
727: thermodynamics, Eq.~(\ref{eq-gsl}), is obeyed.
728:
729: Yet it is by no means obvious that the generalized second law will
730: hold. The growth of the horizon area depends essentially on the mass
731: that is added to the black hole; it does not seem to care about the
732: entropy of the matter system. If it were possible to have matter
733: systems with arbitrarily large entropy at a given mass and size, the
734: generalized second law could still be violated.
735:
736: The thermodynamic properties of black holes developed in the previous
737: subsection, including the assignment of entropy to the horizon, are
738: sufficiently compelling to be considered laws of nature. Then one may
739: turn the above considerations around and demand that the generalized
740: second law hold in all processes. One would expect that this would
741: lead to a universal bound on the entropy of matter systems in terms of
742: their extensive parameters.
743:
744: For any weakly gravitating matter system in asymptotically flat space,
745: Bekenstein (1981) has argued that the GSL implies the following bound:
746: \begin{equation}
747: S_{\rm matter} \leq 2 \pi E R.
748: \label{eq-bekbound}
749: \end{equation}
750: [In full, $S \leq 2 \pi k E R / (\hbar c)$; note that Newton's
751: constant does not enter.] Here, $E$ is the total mass-energy of the
752: matter system. The circumferential radius $R$ is the radius of the
753: smallest sphere that fits around the matter system (assuming that
754: gravity is sufficiently weak to allow for a choice of time slicing
755: such that the matter system is at rest and space is almost Euclidean).
756:
757: We will begin with an argument for this bound in arbitrary spacetime
758: dimension $D$ that involves a strictly classical analysis of the {\em
759: Geroch process}, by which a system is dropped into a black hole from
760: the vicinity of the horizon. We will then show, however, that a
761: purely classical treatment is not tenable. The extent to which
762: quantum effects modify, or perhaps invalidate, the derivation of the
763: Bekenstein bound from the GSL is controversial. The gist of some of
764: the pertinent arguments will be given here, but the reader is referred
765: to the literature for the subtleties.
766:
767: \subsubsection{Geroch process}
768: \label{sec-geroch}
769:
770: Consider a weakly gravitating stable thermodynamic system of total
771: energy $E$. Let $R$ be the radius of the smallest $D-2$ sphere
772: circumscribing the system. To obtain an entropy bound, one may move
773: the system from infinity into a Schwarzschild black hole whose radius,
774: $b$, is much larger than $R$ but otherwise arbitrary. One would like
775: to add as little energy as possible to the black hole, so as to
776: minimize the increase of the black hole's horizon area and thus to
777: optimize the tightness of the entropy bound. Therefore, the strategy
778: is to extract work from the system by lowering it slowly until it is
779: just outside the black hole horizon, before one finally drops it in.
780:
781: The mass added to the black hole is given by the energy $E$ of the
782: system, redshifted according to the position of the center of mass at
783: the drop-off point, at which the circumscribing sphere almost touches
784: the horizon. Within its circumscribing sphere, one may orient the
785: system so that its center of mass is ``down'', i.e., on the side of the
786: black hole. Thus the center of mass can be brought to within a proper
787: distance $R$ from the horizon, while all parts of the system remain
788: outside the horizon. Hence, one must calculate the redshift factor at
789: radial proper distance $R$ from the horizon.
790:
791: The Schwarzschild metric is given by
792: \begin{equation}
793: ds^2 = -V(r) dt^2 + V(r)^{-1} dr^2 + r^2 d\Omega_{D-2}^2,
794: \label{eq-schsch}
795: \end{equation}
796: where
797: \begin{equation}
798: V(r) = 1 - \left( \frac{b}{r} \right)^{D-3} \equiv \left[ \chi(r)
799: \right]^2
800: \label{eq-vofr3}
801: \end{equation}
802: defines the redshift factor, $\chi$ (Myers and Perry, 1986). The
803: black hole radius is related to the mass at infinity, $M$, by
804: \begin{equation}
805: b^{D-3} = \frac{16 \pi M}{(D-2) {\cal A}_{D-2}},
806: \label{eq-bm}
807: \end{equation}
808: where ${\cal A}_{D-2} = 2 \pi^{D-1\over 2}/\Gamma({D-1\over 2})$ is
809: the area of a unit $D-2$ sphere. The black hole has horizon area
810: \begin{equation}
811: A = {\cal A}_{D-2} b^{D-2}.
812: \label{eq-ab}
813: \end{equation}
814:
815: Let $c$ be the radial coordinate distance from the horizon:
816: \begin{equation}
817: c = r-b.
818: \end{equation}
819: Near the horizon, the redshift
820: factor is given by
821: \begin{equation}
822: \chi^2(c) = (D-3) \frac{c}{b},
823: \end{equation}
824: to leading order in $c/b$. The proper distance $l$ is related to the
825: coordinate distance $c$ as follows:
826: \begin{equation}
827: l(c) = \int_0^c \frac{dc}{\chi(c)} = 2 \left( \frac{bc}{D-3}
828: \right)^{1/2}.
829: \end{equation}
830: Hence,
831: \begin{equation}
832: \chi(l) = \frac{D-3}{2b}\, l.
833: \end{equation}
834:
835: The mass added to the black hole is
836: \begin{equation}
837: \delta M \leq E\, \chi(l)\left|_R = \frac{D-3}{2b}\, ER. \right.
838: \end{equation}
839: By Eqs.~(\ref{eq-bm}), (\ref{eq-ab}), and (\ref{eq-eta}), the black
840: hole entropy increases by
841: \begin{equation}
842: \delta S_{\rm BH} = \frac{dS_{\rm BH}}{dM}\, \delta M \leq 2 \pi ER.
843: \end{equation}
844: By the generalized second law, this increase must at least compensate
845: for the lost matter entropy: $\delta S_{\rm BH} - S_{\rm matter} \geq
846: 0$. Hence,
847: \begin{equation}
848: S_{\rm matter} \leq 2 \pi ER.
849: \label{eq-bekbound2}
850: \end{equation}
851:
852: \subsubsection{Unruh radiation}
853: \label{sec-unruh}
854:
855: The above derivation of the Bekenstein bound, by a purely classical
856: treatment of the Geroch process, suffers from the problem that it can
857: be strengthened to a point where it yields an obviously false
858: conclusion. Consider a system in a rectangular box whose height, $h$,
859: is much smaller than its other dimensions. Orient the system so that
860: the small dimension is aligned with the radial direction, and the long
861: dimensions are parallel to the horizon. The minimal distance between
862: the center of mass and the black hole horizon is then set by the
863: height of the box, and will be much smaller than the circumferential
864: radius. In this way, one can ``derive'' a bound of the form
865: \begin{equation}
866: S_{\rm matter} \leq \pi E h.
867: \label{eq-bekbound-wrong}
868: \end{equation}
869: The right hand side goes to zero in the limit of vanishing height, at
870: fixed energy of the box. But the entropy of the box does not go to
871: zero unless {\em all\/} of its dimensions vanish. If only the height
872: goes to zero, the vertical modes become heavy and have to be excluded.
873: But entropy will still be carried by light modes living in the other
874: spatial directions.
875:
876: Unruh and Wald (1982, 1983) have pointed out that a system held at
877: fixed radius just outside a black hole horizon undergoes acceleration,
878: and hence experiences Unruh radiation (Unruh, 1976). They argued that
879: this quantum effect will change both the energetics (because the
880: system will be buoyed by the radiation) and the entropy balance in the
881: Geroch process (because the volume occupied by the system will be
882: replaced by entropic quantum radiation after the system is dropped
883: into the black hole). Unruh and Wald concluded that the Bekenstein
884: bound is neither necessary nor sufficient for the operation of the
885: GSL. Instead, they suggested that the GSL is automatically protected
886: by Unruh radiation as long as the entropy of the matter system does
887: not exceed the entropy of unconstrained thermal radiation of the same
888: energy and volume. This is plausible if the system is indeed weakly
889: gravitating and if its dimensions are not extremely unequal.
890:
891: Bekenstein (1983, 1994a), on the other hand, has argued that Unruh
892: radiation merely affects the lowest layer of the system and is
893: typically negligible. Only for very flat systems, Bekenstein (1994a)
894: claims that the Unruh-Wald effect may be important. This would
895: invalidate the derivation of Eq.~(\ref{eq-bekbound-wrong}) in the
896: limit where this bound is clearly incorrect. At the same time, it
897: would leave the classical argument for the Bekenstein bound,
898: Eq.~(\ref{eq-bekbound2}), essentially intact. As there would be an
899: intermediate regime where Eq.~(\ref{eq-bekbound-wrong}) applies,
900: however, one would not expect the Bekenstein bound to be optimally
901: tight for non-spherical systems.
902:
903: The question of whether the GSL implies the Bekenstein bound remains
904: controversial (see, e.g., Bekenstein, 1999, 2001; Pelath and Wald,
905: 1999; Wald, 2001; Marolf and Sorkin, 2002).
906:
907: The arguments described here can also be applied to other kinds of
908: horizons. Davies (1984) and Schiffer (1992) considered a Geroch
909: process in de~Sitter space, respectively extending the Unruh-Wald and
910: the Bekenstein analysis to the cosmological horizon. Bousso (2001)
911: has shown that the GSL implies a Bekenstein-type bound for dilute
912: systems in asymptotically de~Sitter space, with the assumption of
913: spherical symmetry but not necessarily of weak gravity. In this case
914: one would not expect quantum buoyancy to play a crucial role.
915:
916:
917: \subsubsection{Empirical status}
918:
919: Independently of its logical relation to the GSL, one can ask whether
920: the Bekenstein bound actually holds in nature. Bekenstein (1981,
921: 1984) and Schiffer and Bekenstein (1989) have made a strong case that
922: all physically reasonable, weakly gravitating matter systems satisfy
923: Eq.~(\ref{eq-bekbound}); some come within an order of magnitude of
924: saturation. This empirical argument has been called into question by
925: claims that certain systems violate the Bekenstein bound; see, e.g.,
926: Page (2000) and references therein. Many of these counter-examples,
927: however, fail to include the whole gravitating mass of the system in
928: $E$. Others involve questionable matter content, such as a very large
929: number of species (Sec.~\ref{sec-species}). Bekenstein (2000c) gives
930: a summary of alleged counter-examples and their refutations, along
931: with a list of references to more detailed discussions. If the
932: Bekenstein bound is taken to apply only to complete, weakly
933: gravitating systems that can actually be constructed in nature, it has
934: not been ruled out (Flanagan, Marolf, and Wald, 2000; Wald, 2001).
935:
936: The application of the bound to strongly gravitating systems is
937: complicated by the difficulty of defining the radius of the system in
938: a highly curved geometry. At least for spherically symmetric systems,
939: however, this is not a problem, as one may define $R$ in terms of the
940: surface area. A Schwarzschild black hole in four dimensions has
941: $R=2E$. Hence, its Bekenstein entropy, $S=A/4=\pi R^2$, exactly
942: saturates the Bekenstein bound (Bekenstein, 1981). In $D>4$, black
943: holes come to within a factor $\frac{2}{D-2}$ of saturating the bound
944: (Bousso, 2001).
945:
946:
947:
948: \subsection{Spherical entropy bound}
949: \label{sec-spheb}
950:
951: Instead of dropping a thermodynamic system into an existing black hole
952: via the Geroch process, one may also consider the {\em Susskind
953: process}, in which the system is {\em converted\/} to a black hole.
954: Susskind (1995b) has argued that the GSL, applied to this
955: transformation, yields the {\sl spherical entropy bound}
956: \begin{equation}
957: S_{\rm matter} \leq \frac{A}{4},
958: \label{eq-spheb}
959: \end{equation}
960: where $A$ is a suitably defined area enclosing the matter system.
961:
962: The description of the Susskind process below is influenced by the
963: analysis of Wald (2001).
964:
965: \subsubsection{Susskind process}
966: \label{sec-susskind}
967:
968: Let us consider an isolated matter system of mass $E$ and entropy
969: $S_{\rm matter}$ residing in a spacetime ${\cal M}$. We require that
970: the asymptotic structure of ${\cal M}$ permits the formation of black
971: holes. For definiteness, let us assume that ${\cal M}$ is
972: asymptotically flat. We define $A$ to be the area of the
973: circumscribing sphere, i.e., the smallest sphere that fits around the
974: system. Note that $A$ is well-defined only if the metric near the
975: system is at least approximately spherically symmetric. This will be
976: the case for all spherically symmetric systems, and for all weakly
977: gravitating systems, but not for strongly gravitating systems lacking
978: spherical symmetry. Let us further assume that the matter system is
979: stable on a timescale much greater than $A^{1/2}$. That is, it
980: persists and does not expand or collapse rapidly, so that the
981: time-dependence of $A$ will be negligible.
982:
983: The system's mass must be less than the mass $M$ of a black hole of
984: the same surface area. Otherwise, the system could not be
985: gravitationally stable, and from the outside point of view it would
986: already be a black hole. One would expect that the system can be
987: converted into a black hole of area $A$ by collapsing a shell of mass
988: $M-E$ onto the system.%
989: \footnote{This assumes that the shell can actually be brought to
990: within $A$ without radiating or ejecting shell mass or system mass.
991: For two large classes of systems, Bekenstein (2000a,b) obtains
992: Eq.~(\ref{eq-spheb}) under weaker assumptions.}
993:
994: Let the shell be well-separated from the system initially. Its
995: entropy, $S_{\rm shell}$, is non-negative. The total initial entropy
996: in this thermodynamic process is given by
997: \begin{equation}
998: S_{\rm total}^{\rm initial} = S_{\rm matter}+S_{\rm shell}.
999: \end{equation}
1000: The final state is a black hole, with entropy
1001: \begin{equation}
1002: S_{\rm total}^{\rm final} = S_{\rm BH} = \frac{A}{4}.
1003: \end{equation}
1004: By the generalized second law of thermodynamics, Eq.~(\ref{eq-gsl}),
1005: the initial entropy must not exceed the final entropy. Since $S_{\rm
1006: shell}$ is obviously non-negative, Eq.~(\ref{eq-spheb}) follows.
1007:
1008:
1009: \subsubsection{Relation to the Bekenstein bound}
1010:
1011: Thus, the spherical entropy bound is obtained directly from the GSL
1012: via the Susskind process. Alternatively, and with similar
1013: limitations, one can obtain the same result from the Bekenstein bound,
1014: if the latter is assumed to hold for strongly gravitating systems.
1015: The requirement that the system be gravitationally stable implies $2M
1016: \leq R$ in four dimensions. From Eq.~(\ref{eq-bekbound}), one thus
1017: obtains:
1018: \begin{equation}
1019: S \leq 2\pi M R \leq \pi R^2 = \frac{A}{4}.
1020: \end{equation}
1021: This shows that the spherical entropy bound is weaker than the
1022: Bekenstein bound, in situations where both can be applied.
1023:
1024: The spherical entropy bound, however, is more closely related to the
1025: holographic principle. It can be cast in a covariant and general form
1026: (Sec.~\ref{sec-bb}). An interesting open question is whether one can
1027: reverse the logical direction and derive the Bekenstein bound from the
1028: covariant entropy bound under suitable assumptions
1029: (Sec.~\ref{sec-gia}).
1030:
1031: In $D>4$, gravitational stability and the Bekenstein bound imply only
1032: $S \leq \frac{D-2}{8} A$ (Bousso, 2001). The discrepancy may stem
1033: from the extrapolation to strong gravity and/or the lack of a reliable
1034: calibration of the prefactor in the Bekenstein bound.
1035:
1036:
1037: \subsubsection{Examples}
1038: \label{sec-gb}
1039:
1040: The spherical entropy bound is best understood by studying a number of
1041: examples in four spacetime dimensions. We follow 't~Hooft (1993) and
1042: Wald (2001).
1043:
1044: It is easy to see that the bound holds for black holes. By
1045: definition, the entropy of a single Schwarzschild black hole, $S_{\rm
1046: BH} = A/4$, precisely saturates the bound. In this sense, a black
1047: hole is the most entropic object one can put inside a given spherical
1048: surface ('t~Hooft, 1993).
1049:
1050: Consider a system of several black holes of masses $M_i$, in $D=4$.
1051: Their total entropy will be given by
1052: \begin{equation}
1053: S = 4 \pi \sum M_i^2.
1054: \end{equation}
1055: From the point of view of a distant observer, the system must not
1056: already be a larger black hole of mass $\sum M_i$. Hence, it must be
1057: circumscribed by a spherical area
1058: \begin{equation}
1059: A \geq 16 \pi \left( \sum M_i \right)^2 > 16 \pi \sum M_i^2 = 4S.
1060: \end{equation}
1061: Hence, the spherical entropy bound is satisfied with room to spare.
1062:
1063: Using ordinary matter instead of black holes, it turns out to be
1064: difficult even to approach saturation of the bound. In order to
1065: obtain a stable, highly entropic system, a good strategy is to make it
1066: from massless particles. Rest mass only enhances gravitational
1067: instability without contributing to the entropy. Consider, therefore,
1068: a gas of radiation at temperature $T$, with energy $E$, confined in a
1069: spherical box of radius $R$. We must demand that the system is not a
1070: black hole: $R \geq 2E$. For an order-of-magnitude estimate of the
1071: entropy, we may neglect the effects of self-gravity and treat the
1072: system as if it lived on a flat background.
1073:
1074: The energy of the ball is related to its temperature as
1075: \begin{equation}
1076: E \sim Z R^3 T^4,
1077: \end{equation}
1078: where $Z$ is the number of different species of particles in the gas.
1079: The entropy of the system is given by
1080: \begin{equation}
1081: S \sim Z R^3 T^3.
1082: \end{equation}
1083: Hence, the entropy is related to the size and energy as
1084: \begin{equation}
1085: S \sim Z^{1/4} R^{3/4} E^{3/4}.
1086: \end{equation}
1087: Gravitational stability then implies that
1088: \begin{equation}
1089: S \lesssim Z^{1/4} A^{3/4}.
1090: \label{eq-radball}
1091: \end{equation}
1092: In order to compare this result to the spherical entropy bound, $S
1093: \leq A/4$, recall that we are using Planck units. For any geometric
1094: description to be valid, the system must be much larger than the
1095: Planck scale:
1096: \begin{equation}
1097: A \gg 1.
1098: \end{equation}
1099: A generous estimate for the number of species in nature is $Z \sim
1100: O(10^3)$. Hence, $Z^{1/4} A^{3/4}$ is much smaller than $A$ for all
1101: but the smallest, nearly Planck size systems, in which the present
1102: approximations cannot be trusted in any case. For a gas ball of size
1103: $R\gg 1$, the spherical entropy bound will be satisfied with a large
1104: factor, $R^{1/2}$, to spare.
1105:
1106: \subsubsection{The species problem}
1107: \label{sec-species}
1108:
1109: An interesting objection to entropy bounds is that one can write down
1110: perfectly well-defined field theory Lagrangians with an arbitrarily
1111: large number of particle species (Sorkin, Wald, and Zhang, 1981; Unruh
1112: and Wald, 1982). In the example of Eq.~(\ref{eq-radball}), a
1113: violation of the spherical entropy bound for systems up to size $A$
1114: would require
1115: \begin{equation}
1116: Z \gtrsim A.
1117: \end{equation}
1118: For example, to construct a counterexample of the size of a proton,
1119: one would require $Z \gtrsim 10^{40}$. It is trivial to write down a
1120: Lagrangian with this number of fields. But this does not mean that
1121: the entropy bound is wrong.
1122:
1123: In nature, the effective number of matter fields is whatever it is; it
1124: cannot be tailored to the specifications of one's favorite
1125: counterexample. The spherical bound is a statement about nature. If
1126: it requires that the number of species is not exponentially large,
1127: then this implication is certainly in good agreement with observation.
1128: At any rate it is more plausible than the assumption of an
1129: exponentially large number of light fields.
1130:
1131: Indeed, an important lesson learned from black holes and the
1132: holographic principle is that nature, at a fundamental level, will not
1133: be described by a local field theory living on some background
1134: geometry (Susskind, Thorlacius, and Uglum, 1993).
1135:
1136:
1137: The spherical entropy bound was derived from the generalized second
1138: law of thermodynamics (under a set of assumptions). Could one not,
1139: therefore, use the GSL to rule out large $Z$? Consider a radiation
1140: ball with $Z\gtrsim A$ massless species, so that $S>A$. The system is
1141: transformed to a black hole of area $A$ by a Susskind process.
1142: However, Wald (2001) has shown that the apparent entropy decrease is
1143: irrelevant, because the black hole is catastrophically unstable. In
1144: Sec.~\ref{sec-bh-radiation}, the time for the Hawking evaporation of a
1145: black hole was estimated to be $A^{3/2}$ in $D=4$. This implicitly
1146: assumed a small number of radiated species. But for large $Z$, one
1147: must take into account that the radiation rate is actually
1148: proportional to $Z$. Hence, the evaporation time is given by
1149: \begin{equation}
1150: t_0 \sim \frac{A^{3/2}}{Z}.
1151: \end{equation}
1152: With $Z \gtrsim A$, one has $t_0 \lesssim A^{1/2}$. The time needed
1153: to form a black hole of area $A$ is at least of order $A^{1/2}$, so
1154: the black hole in question evaporates faster than it forms.
1155:
1156: Wald's analysis eliminates the possibility of using the GSL to exclude
1157: large $Z$ for the process at hand. But it produces a different,
1158: additional argument against proliferating the number of species.
1159: Exponentially large $Z$ would render black holes much bigger than the
1160: Planck scale completely unstable. Let us demand, therefore, that
1161: super-Planckian black holes be at least metastable. Then $Z$ cannot
1162: be made large enough to construct a counterexample from
1163: Eq.~(\ref{eq-radball}). From a physical point of view, the
1164: metastability of large black holes seems a far more natural assumption
1165: than the existence of an extremely large number of particle species.
1166:
1167: Further arguments on the species problem (of which the
1168: possible renormalization of Newton's constant with $Z$ has received
1169: particular attention) are found in Bombelli {\em et al.} (1986),
1170: Bekenstein (1994b, 1999, 2000c), Jacobson (1994), Susskind and Uglum
1171: (1994, 1996), Frolov (1995), Brustein, Eichler, and Foffa (2000),
1172: Veneziano (2001), Wald (2001), and Marolf and Sorkin (2002).
1173:
1174:
1175: \section{Towards a holographic principle}
1176: \label{sec-towards}
1177:
1178: \subsection{Degrees of freedom}
1179: \label{sec-ndof}
1180:
1181: How many degrees of freedom are there in nature, at the most
1182: fundamental level? The holographic principle answers this question in
1183: terms of the area of surfaces in spacetime. Before reaching this
1184: rather surprising answer, we will discuss a more traditional way one
1185: might have approached the question. Parts of this analysis follow
1186: 't~Hooft (1993) and Susskind (1995b).
1187:
1188: For the question to have meaning, let us restrict to a finite region
1189: of volume $V$ and boundary area $A$. Assume, for now, that gravity is
1190: not strong enough to blur the definition of these quantities, and that
1191: spacetime is asymptotically flat. Application of the spherical
1192: entropy bound, Eq.~(\ref{eq-spheb}), will force us to consider the
1193: circumscribing sphere of the region. This surface will coincide with
1194: the boundary of the region only if the boundary is a sphere, which we
1195: shall assume.
1196:
1197: In order to satisfy the assumptions of the spherical entropy bound we
1198: also demand that the metric of the enclosed region is not strongly
1199: time-dependent, in the sense described at the beginning of
1200: Sec.~\ref{sec-susskind}. In particular, this means that $A$ will not
1201: be a trapped surface in the interior of a black hole.
1202:
1203: Let us define the {\em number of degrees of freedom\/} of a
1204: quantum-mechanical system, $N$, to be the logarithm of the dimension
1205: ${\cal N}$ of its Hilbert space ${\cal H}$:
1206: \begin{equation}
1207: N = \ln {\cal N} = \ln \dim( {\cal H} ).
1208: \end{equation}
1209: Note that a harmonic oscillator has $N=\infty$ with this definition.
1210: The number of degrees of freedom is equal (up to a factor of $\ln 2$)
1211: to the number of bits of information needed to characterize a state.
1212: For example, a system with 100 spins has ${\cal N}=2^{100}$ states,
1213: $N=100 \ln 2$ degrees of freedom, and can store 100 bits of
1214: information.
1215:
1216: \subsection{Fundamental system}
1217: \label{sec-funds}
1218:
1219: Consider a spherical region of space with no particular restrictions
1220: on matter content. One can regard this region as a quantum-mechanical
1221: system and ask how many different states it can be in. In other
1222: words, what is the dimension of the quantum Hilbert space describing
1223: all possible physics confined to the specified region, down to the
1224: deepest level?
1225:
1226: Thus, our question is not about the Hilbert space of a specific
1227: system, such as a hydrogen atom or an elephant. Ultimately, all these
1228: systems should reduce to the constituents of a fundamental theory.
1229: The question refers directly to these constituents, given only the
1230: size\footnote{The precise nature of the geometric boundary conditions
1231: is discussed further in Sec.~\ref{sec-edef}.} of a region. Let us
1232: call this system the {\em fundamental system}.
1233:
1234: How much complexity, in other words, lies at the deepest level of
1235: nature? How much information is required to specify {\em any\/}
1236: physical configuration completely, as long as it is contained in a
1237: prescribed region?
1238:
1239:
1240: \subsection{Complexity according to local field theory}
1241: \label{sec-fromft}
1242:
1243: In the absence of a unified theory of gravity and quantum fields, it
1244: is natural to seek an answer from an approximate framework. Suppose
1245: that the ``fundamental system'' is local quantum field theory on a
1246: classical background spacetime satisfying Einstein's equations
1247: (Birrell and Davies, 1982; Wald, 1994). A quantum field theory
1248: consists of one or more oscillators at every point in space. Even a
1249: single harmonic oscillator has an infinite-dimensional Hilbert space.
1250: Moreover, there are infinitely many points in any volume of space, no
1251: matter how small. Thus, the answer to our question appears to be
1252: $N = \infty$. However, so far we have disregarded the
1253: effects of gravity altogether.
1254:
1255: A finite estimate is obtained by including gravity at least in a
1256: crude, minimal way. One might expect that distances smaller than the
1257: Planck length, $l_{\rm P} = 1.6 \times 10^{-33}$\/cm, cannot be
1258: resolved in quantum gravity. So let us discretize space into a Planck
1259: grid and assume that there is one oscillator per Planck volume.
1260: Moreover, the oscillator spectrum is discrete and bounded from below
1261: by finite volume effects. It is bounded from above because it must be
1262: cut off at the Planck energy, $M_{\rm P} = 1.3 \times 10^{19}$\/GeV.
1263: This is the largest amount of energy that can be localized to a Planck
1264: cube without producing a black hole. Thus, the total number of
1265: oscillators is $V$ (in Planck units), and each has a finite number of
1266: states, $n$. (A minimal model one might think of is a Planckian
1267: lattice of spins, with $n=2$.) Hence, the total number of independent
1268: quantum states in the specified region is
1269: \begin{equation}
1270: {\cal N} \sim n^V.
1271: \label{eq-states-volume}
1272: \end{equation}
1273: The number of degrees of freedom is given by
1274: \begin{equation}
1275: N \sim V \ln n \gtrsim V.
1276: \label{eq-ndof-volume}
1277: \end{equation}
1278: This result successfully captures our prejudice that the degrees of
1279: freedom in the world are local in space, and that, therefore,
1280: complexity grows with volume. It turns out, however, that this view
1281: conflicts with the laws of gravity.
1282:
1283:
1284: \subsection{Complexity according to the spherical entropy bound}
1285: \label{sec-fromspheb}
1286:
1287: Thermodynamic entropy has a statistical interpretation. Let $S$ be
1288: the thermodynamic entropy of an isolated system at some specified
1289: value of macroscopic parameters such as energy and volume. Then $e^S$
1290: is the number of independent quantum states compatible with these
1291: macroscopic parameters. Thus, entropy is a measure of our ignorance
1292: about the detailed microscopic state of a system. One could relax the
1293: macroscopic parameters, for example by requiring only that the energy
1294: lie in some finite interval. Then more states will be allowed, and
1295: the entropy will be larger.
1296:
1297: The question at the beginning of this section was ``How many
1298: independent states are required to describe all the physics in a
1299: region bounded by an area $A$?'' Recall that all thermodynamic
1300: systems should ultimately be described by the same underlying theory,
1301: and that we are interested in the properties of this ``fundamental
1302: system''. We are now able to rephrase the question as follows: ``What
1303: is the entropy, $S$, of the `fundamental system', given that only the
1304: boundary area is specified?'' Once this question is answered, the
1305: number of states will simply be ${\cal N} = e^S$, by the argument
1306: given in the previous paragraph.
1307:
1308: In Sec.~\ref{sec-spheb} we obtained the spherical entropy bound,
1309: Eq.~(\ref{eq-spheb}), from which the entropy can be determined without
1310: any knowledge of the nature of the ``fundamental system''. The bound,
1311: \begin{equation}
1312: S \leq \frac{A}{4},
1313: \end{equation}
1314: makes reference only to the boundary area; it does not care about the
1315: microscopic properties of the thermodynamic system. Hence, it applies
1316: to the ``fundamental system'' in particular. A black hole that just
1317: fits inside the area $A$ has entropy
1318: \begin{equation}
1319: S_{\rm BH} = \frac{A}{4},
1320: \end{equation}
1321: so the bound can clearly be saturated with the given boundary
1322: conditions. Therefore, the number of degrees of freedom in a region
1323: bounded by a sphere of area $A$ is given by
1324: \begin{equation}
1325: N = \frac{A}{4};
1326: \label{eq-ndof-area}
1327: \end{equation}
1328: the number of states is
1329: \begin{equation}
1330: {\cal N} = e^{A/4}.
1331: \label{eq-states-area}
1332: \end{equation}
1333:
1334: We assume that all physical systems are larger than the Planck scale.
1335: Hence, their volume will exceed their surface area, in Planck units.
1336: (For a proton, the volume is larger than the area by a factor of
1337: $10^{20}$; for the earth, by $10^{41}$.) The result obtained from the
1338: spherical entropy bound is thus at odds with the much larger number
1339: of degrees of freedom estimated from local field theory. Which of the
1340: two conclusions should we believe?
1341:
1342:
1343: \subsection{Why local field theory gives the wrong answer}
1344: \label{sec-qftwrong}
1345:
1346: We will now argue that the field theory analysis overcounted available
1347: degrees of freedom, because it failed to include properly the effects
1348: of gravitation. We assume $D=4$ and neglect factors of order unity.
1349: (In $D>4$ the gist of the discussion is unchanged though some of the
1350: powers are modified.)
1351:
1352: The restriction to a finite spatial region provides an infrared
1353: cut-off, precluding the generation of entropy by long wavelength
1354: modes. Hence, most of the entropy in the field theory estimate comes
1355: from states of very high energy. But a spherical surface cannot
1356: contain more mass than a black hole of the same area. According to
1357: the Schwarzschild solution, Eq.~(\ref{eq-schsch}), the mass of a black
1358: hole is given by its radius. Hence, the mass $M$ contained within a
1359: sphere of radius $R$ obeys
1360: \begin{equation}
1361: M \lesssim R
1362: \label{eq-mlessthanr}.
1363: \end{equation}
1364:
1365: The ultra-violet cutoff imposed in Sec.~\ref{sec-fromft} reflected
1366: this, but only on the smallest scale ($R=1$). It demanded only that
1367: each Planck volume must not contain more than one Planck mass. For
1368: larger regions this cutoff would permit $M \sim R^3$, in violation of
1369: Eq.~(\ref{eq-mlessthanr}). Hence our cut-off was too lenient to
1370: prevent black hole formation on larger scales.
1371:
1372: For example, consider a sphere of radius $R=1\,$cm, or $10^{33}$ in
1373: Planck units. Suppose that the field energy in the enclosed region
1374: saturated the naive cut-off in each of the $\sim 10^{99}$ Planck
1375: cells. Then the mass within the sphere would be $\sim 10^{99}$. But
1376: the most massive object that can be localized to the sphere is a black
1377: hole, of radius and mass $10^{33}$.
1378:
1379: Thus, most of the states included by the field theory estimate are too
1380: massive to be gravitationally stable. Long before the quantum fields
1381: can be excited to such a level, a black hole would form.%
1382: %
1383: \footnote{Thus, black holes provide a natural covariant cut-off which
1384: becomes stronger at larger distances. It differs greatly from the
1385: fixed distance or fixed energy cutoffs usually considered in quantum
1386: field theory.}
1387: %
1388: If this black hole is still to be contained within a specified sphere
1389: of area $A$, its entropy may saturate but not exceed the spherical
1390: entropy bound.
1391:
1392: Because of gravity, not all degrees of freedom that field theory
1393: apparently supplies can be used for generating entropy, or storing
1394: information. This invalidates the field theory estimate,
1395: Eq.~(\ref{eq-ndof-volume}), and thus resolves the apparent
1396: contradiction with the holographic result, Eq.~(\ref{eq-ndof-area}).
1397:
1398: Note that the present argument does not provide independent
1399: quantitative confirmation that the maximal entropy is given by the
1400: area. This would require a detailed understanding of the relation
1401: between entropy, energy, and gravitational back-reaction in a given
1402: system.
1403:
1404:
1405:
1406: \subsection{Unitarity and a holographic interpretation}
1407: \label{sec-hsi}
1408:
1409: Using the spherical entropy bound, we have concluded that $A/4$
1410: degrees of freedom are sufficient to fully describe any stable region
1411: in asymptotically flat space enclosed by a sphere of area $A$. In a
1412: field theory description, there are far more degrees of freedom.
1413: However, we have argued that any attempt to excite more than $A/4$ of
1414: these degrees of freedom is thwarted by gravitational collapse. From
1415: the outside point of view, the most entropic object that fits in the
1416: specified region is a black hole of area $A$, with $A/4$ degrees of
1417: freedom.
1418:
1419: A conservative interpretation of this result is that the demand for
1420: gravitational stability merely imposes a practical limitation for the
1421: information content of a spatial region. If we are willing to pay the
1422: price of gravitational collapse, we can excite more than $A/4$ degrees
1423: of freedom---though we will have to jump into a black hole to verify
1424: that we have succeeded. With this interpretation, all the degrees of
1425: freedom of field theory should be retained. The region will be
1426: described by a quantum Hilbert space of dimension $e^V$.
1427:
1428: The following two considerations motivate a rejection of this
1429: interpretation. Both arise from the point of view that physics in
1430: asymptotically flat space can be consistently described by a
1431: scattering matrix. The S-matrix provides amplitudes between initial
1432: and final asymptotic states defined at infinity. Intermediate black
1433: holes may form and evaporate, but as long as one is not interested in
1434: the description of an observer falling into the black hole, an
1435: S-matrix description should be satisfactory from the point of view of
1436: an observer at infinity.
1437:
1438: One consideration concerns economy. A fundamental theory should not
1439: contain more than the necessary ingredients. If $A/4$ is the amount
1440: of data needed to describe a region completely, that should be the
1441: amount of data used. This argument is suggestive; however, it could be
1442: rejected as merely aesthetical and gratuitously radical.
1443:
1444: A more compelling consideration is based on unitarity.
1445: Quantum-mechanical evolution preserves information; it takes a pure
1446: state to a pure state. But suppose a region was described by a
1447: Hilbert space of dimension $e^V$, and suppose that region was
1448: converted to a black hole. According to the Bekenstein entropy of a
1449: black hole, the region is now described by a Hilbert space of
1450: dimension $e^{A/4}$. The number of states would have decreased, and
1451: it would be impossible to recover the initial state from the final
1452: state. Thus, unitarity would be violated. Hence, the Hilbert space
1453: must have had dimension $e^{A/4}$ to start with.
1454:
1455: The insistence on unitarity in the presence of black holes led
1456: 't~Hooft (1993) and Susskind (1995b) to embrace a more radical,
1457: ``holographic'' interpretation of Eq.~(\ref{eq-ndof-area}).
1458:
1459: {\sl Holographic principle (preliminary formulation)}. {\em
1460: A region with boundary of area $A$ is fully described by no more than
1461: $A/4$ degrees of freedom, or about 1 bit of information per Planck
1462: area. A fundamental theory, unlike local field theory, should
1463: incorporate this counterintuitive result.}
1464:
1465:
1466: \subsection{Unitarity and black hole complementarity}
1467: \label{sec-bhc}
1468:
1469: The unitarity argument would be invalidated if it turned out that
1470: unitarity is not preserved in the presence of black holes. Indeed,
1471: Hawking (1976b) has claimed that the evaporation of a black hole---its
1472: slow conversion into a cloud of radiation---is not a unitary process.
1473: In semi-classical calculations, Hawking radiation is found to be
1474: exactly thermal, and all information about the ingoing state appears
1475: lost. Others (see Secs.~\ref{sec-further}, \ref{sec-strings}) argued,
1476: however, that unitarity must be restored in a complete quantum gravity
1477: theory.
1478:
1479: The question of unitarity of the S-matrix arises not only when a black
1480: hole forms, but again, and essentially independently, when the black
1481: hole evaporates. The holographic principle is necessary for unitarity
1482: at the first stage. But if unitarity were later violated during
1483: evaporation, it would have to be abandoned, and the holographic
1484: principle would lose its basis.
1485:
1486: It is not understood in detail how Hawking radiation carries away
1487: information. Indeed, the assumption that it does seems to lead to a
1488: paradox, which was pointed out and resolved by Susskind, Thorlacius,
1489: and Uglum (1993). When a black hole evaporates unitarily, the same
1490: quantum information would seem to be present both inside the black
1491: hole (as the original matter system that collapsed) and outside, in
1492: the form of Hawking radiation. The simultaneous presence of two
1493: copies appears to violate the linearity of quantum mechanics, which
1494: forbids the ``xeroxing'' of information.
1495:
1496: One can demonstrate, however, that no single observer can see both
1497: copies of the information. Obviously an infalling observer cannot
1498: escape the black hole to record the outgoing radiation. But what
1499: prevents an outside observer from first obtaining, say, one bit of
1500: information from the Hawking radiation, only to jump into the black
1501: hole to collect a second copy?
1502:
1503: Page (1993) has shown that more than half of a system has to be
1504: observed to extract one bit of information. This means that an
1505: outside observer has to linger for a time compared to the evaporation
1506: time scale of the black hole ($M^3$ in $D=4$) in order to gather a
1507: piece of the ``outside data'', before jumping into the black hole to
1508: verify the presence of the same data inside.
1509:
1510: However, the second copy can only be observed if it has not already
1511: hit the singularity inside the black hole by the time the observer
1512: crosses the horizon. One can show that the energy required for a
1513: single photon to evade the singularity for so long is exponential in
1514: the square of the black hole mass. In other words, there is far too
1515: little energy in the black hole to communicate even one bit of
1516: information to an infalling observer in possession of outside data.
1517:
1518: The apparent paradox is thus exposed as the artifact of an
1519: operationally meaningless, global point of view. There are two
1520: complementary descriptions of black hole formation, corresponding to
1521: an infalling and and an outside observer. Each point of view is
1522: self-consistent, but a simultaneous description of both is neither
1523: logically consistent nor practically testable. Black hole
1524: complementarity thus assigns a new role to the observer in quantum
1525: gravity, abandoning a global description of spacetimes with horizons.
1526:
1527: Further work on black hole complementarity includes 't~Hooft (1991),
1528: Susskind (1993a,b, 1994), Stephens, 't~Hooft, and Whiting (1994),
1529: Susskind and Thorlacius (1994), Susskind and Uglum, (1994). Aspects
1530: realized in string theory are also discussed by Lowe, Susskind, and
1531: Uglum (1994), Lowe {\em et al.} (1995); see Sec.~\ref{sec-strings}.
1532: For a review, see, e.g., Thorlacius (1995), Verlinde (1995), Susskind
1533: and Uglum (1996), and Bigatti and Susskind (2000).
1534:
1535: Together, the holographic principle and black hole complementarity
1536: form the conceptual core of a new framework for black hole formation
1537: and evaporation, in which the unitarity of the S-matrix is retained at
1538: the expense of locality.\footnote{In this sense, the holographic
1539: principle, as it was originally proposed, belongs in the first class
1540: discussed in Sec.~\ref{sec-ia}. However, one cannot obtain its modern
1541: form (Sec.~\ref{sec-hp}) from unitarity. Hence we resort to the
1542: covariant entropy bound in this review. Because the bound can be
1543: tested using conventional theories, this also obviates the need to
1544: assume particular properties of quantum gravity in order to induce the
1545: holographic principle.}
1546:
1547: In the intervening years, much positive evidence for unitarity has
1548: accumulated. String theory has provided a microscopic, unitary
1549: quantum description of some black holes (Strominger and Vafa, 1996;
1550: see also Callan and Maldacena, 1996; Sec.~\ref{sec-strings}).
1551: Moreover, there is overwhelming evidence that certain asymptotically
1552: Anti-de Sitter spacetimes, in which black holes can form and
1553: evaporate, are fully described by a unitary conformal field theory
1554: (Sec.~\ref{sec-ads}).
1555:
1556: Thus, a strong case has been made that the formation and evaporation
1557: of a black hole is a unitary process, at least in asymptotically flat
1558: or AdS spacetimes.
1559:
1560:
1561: \subsection{Discussion}
1562: \label{sec-limitations}
1563:
1564: In the absence of a generally valid entropy bound, the arguments for a
1565: holographic principle were incomplete, and its meaning remained
1566: somewhat unclear. Neither the spherical entropy bound,
1567: nor the unitarity argument which motivates its elevation to a
1568: holographic principle, are applicable in general spacetimes.
1569:
1570: An S-matrix description is justified in a particle accelerator, but
1571: not in gravitational physics. In particular, realistic universes do
1572: not permit an S-matrix description. (For recent discussions see,
1573: e.g., Banks, 2000a; Fischler, 2000a,b; Bousso, 2001a; Fischler {\em et
1574: al.}, 2001; Hellerman, Kaloper, and Susskind, 2001.) Even in
1575: spacetimes that do, observers don't all live at infinity. Then the
1576: question is not so much whether unitarity holds, but how it can be
1577: defined.
1578:
1579: As black hole complementarity itself insists, the laws of physics must
1580: also describe the experience of an observer who falls into a black
1581: hole. The spherical entropy bound, however, need not apply inside
1582: black holes. Moreover, it need not hold in many other important
1583: cases, in view of the assumptions involved in its derivation. For
1584: example, it does not apply in cosmology, and it cannot be used when
1585: spherical symmetry is lacking. In fact, it will be seen in
1586: Sec.~\ref{sec-seb} that the entropy in spatial volumes can exceed the
1587: boundary area in all of these cases.
1588:
1589: Thus, the holographic principle could not, at first, establish a
1590: general correspondence between areas and the number of fundamental
1591: degrees of freedom. But how can it point the way to quantum gravity,
1592: if it apparently does not apply to many important solutions of the
1593: classical theory?
1594:
1595: The AdS/CFT correspondence (Sec.~\ref{sec-ads}), holography's most
1596: explicit manifestation to date, was a thing of the future when the
1597: holographic principle was first proposed. So was the covariant
1598: entropy bound (Secs.~\ref{sec-bb}--\ref{sec-tests}), which exposes the
1599: apparent limitations noted above as artifacts of the original,
1600: geometrically crude formulation. The surprising universality of the
1601: covariant bound significantly strengthens the case for a holographic
1602: principle (Sec.~\ref{sec-hp}).
1603:
1604: As 't~Hooft and Susskind anticipated, the conceptual revisions
1605: required by the unitarity of the S-matrix have proven too profound to
1606: be confined to the narrow context in which they were first recognized.
1607: We now understand that areas should generally be associated with
1608: degrees of freedom in adjacent spacetime regions. Geometric
1609: constructs that precisely define this relation---light-sheets---have
1610: been identified (Fischler and Susskind, 1998; Bousso, 1999a). The
1611: holographic principle may have been an audacious concept to propose.
1612: In light of the intervening developments, it has become a difficult
1613: one to reject.
1614:
1615:
1616:
1617:
1618: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1619: \section{A spacelike entropy bound?}
1620: \label{sec-seb}
1621: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1622:
1623: The heuristic derivation of the spherical entropy bound rests on a
1624: large number of fairly strong assumptions. Aside from suitable
1625: asymptotic conditions, the surface $A$ has to be spherical, and the
1626: enclosed region must be gravitationally stable so that it can be
1627: converted to a black hole.
1628:
1629: Let us explore whether the spherical entropy bound, despite these
1630: apparent limitations, is a special case of a more general entropy
1631: bound. We will present two conjectures for such a bound. In this
1632: section, we will discuss the spacelike entropy bound, perhaps the most
1633: straightforward and intuitive generalization of
1634: Eq.~(\ref{eq-spheb}). We will present several counterexamples to
1635: this bound and conclude that it does not have general validity.
1636: Turning to a case of special interest, we will find that it is
1637: difficult to precisely define the range of validity of the spacelike
1638: entropy bound even in simple cosmological spacetimes.
1639:
1640: \subsection{Formulation}
1641: \label{sec-seb-form}
1642:
1643: One may attempt to extend the scope of Eq.~(\ref{eq-spheb}) simply by
1644: dropping the assumptions under which it was derived (asymptotic
1645: structure, gravitational stability, and spherical symmetry). Let us
1646: call the resulting conjecture the {\em spacelike entropy bound}: {\em
1647: The entropy contained in any spatial region will not exceed the area
1648: of the region's boundary.} More precisely, the {\sl spacelike entropy
1649: bound} is the following statement (Fig.~\ref{fig-spheb}):
1650: %%***************************************************************
1651: \begin{figure}
1652: \includegraphics[width=7cm]{fig-spheb}
1653: \caption{A hypersurface of equal time. The spacelike entropy bound
1654: attempts to relate the entropy in a spatial region, $V$, to the area
1655: of its boundary, $B$. This is not successful.}
1656: \label{fig-spheb}
1657: \end{figure}
1658: %%***************************************************************
1659:
1660: Let $V$ be a compact portion of a hypersurface of equal time in the
1661: spacetime ${\cal M}$.\footnote{Here $V$ is used both to denote a
1662: spatial region, and its volume. Note that we use more careful
1663: notation to distinguish a surface ($B$) from its area ($A$).} Let
1664: $S(V)$ be the entropy of all matter systems in $V$. Let $B$ be the
1665: boundary of $V$ and let $A$ be the area of the boundary of $V$. Then
1666: \begin{equation}
1667: S(V) \leq {A[B(V)]\over 4}.
1668: \label{eq-seb}
1669: \end{equation}
1670:
1671:
1672: \subsection{Inadequacies}
1673: \label{sec-fail}
1674:
1675: The spacelike entropy bound is not a successful conjecture.
1676: Eq.~(\ref{eq-seb}) is contradicted by a large variety of
1677: counterexamples. We will begin by discussing two examples from
1678: cosmology. Then we will turn to the case of a collapsing star.
1679: Finally, we will expose violations of Eq.~(\ref{eq-seb}) even for all
1680: isolated, spherical, weakly gravitating matter systems.
1681:
1682:
1683: \subsubsection{Closed spaces}
1684: \label{sec-fail1}
1685:
1686: It is hardly necessary to describe a closed universe in detail to see
1687: that it will lead to a violation of the spacelike holographic
1688: principle. It suffices to assume that the spacetime ${\cal M}$
1689: contains a closed spacelike hypersurface, ${\cal V}$. (For example,
1690: there are realistic cosmological solutions in which ${\cal V}$ has the
1691: topology of a three-sphere.) We further assume that ${\cal V}$
1692: contains a matter system that does not occupy all of ${\cal V}$, and
1693: that this system has non-zero entropy $S_0$.
1694:
1695: Let us define the volume $V$ to be the whole hypersurface, except for
1696: a small compact region $Q$ outside the matter system. Thus, $S_{\rm
1697: matter}(V) = S_0 >0$. The boundary $B$ of $V$ coincides with the
1698: boundary of $Q$. Its area can be made arbitrarily small by
1699: contracting $Q$ to a point. Thus one obtains $S_{\rm
1700: matter}(V)>A[B(V)])$, and the spacelike entropy bound,
1701: Eq.~(\ref{eq-seb}), is violated.
1702:
1703:
1704: \subsubsection{The Universe}
1705: \label{sec-fail2}
1706:
1707: On large scales, the universe we inhabit is well approximated as a
1708: three-dimensional, flat, homogeneous and isotropic space, expanding in
1709: time. Let us pick one homogeneous hypersurface of equal time, ${\cal
1710: V}$. Its entropy content can be characterized by an average ``entropy
1711: density'', $\sigma$, which is a positive constant on ${\cal V}$.
1712: Flatness implies that the geometry of ${\cal V}$ is Euclidean
1713: ${\mathbb R}^3$. Hence, the volume and area of a two-sphere grow in
1714: the usual way with the radius:
1715: \begin{equation}
1716: V={4\pi\over 3} R^3,~~~A[B(V)] = 4\pi R^2.
1717: \end{equation}
1718:
1719: The entropy in the volume $V$ is given by
1720: \begin{equation}
1721: S_{\rm matter}(V) = \sigma V = {\sigma\over 6\sqrt{\pi}} A^{3/2}.
1722: \end{equation}
1723: Recall that we are working in Planck units. By taking the radius of
1724: the sphere to be large enough,
1725: \begin{equation}
1726: R \geq {3\over 4\sigma},
1727: \end{equation}
1728: one finds a volume for which the spacelike entropy bound,
1729: Eq.~(\ref{eq-seb}), is violated (Fischler and Susskind, 1998).
1730:
1731:
1732: \subsubsection{Collapsing star}
1733: \label{sec-fail3}
1734:
1735: Next, consider a spherical star with non-zero entropy
1736: $S_0$. Suppose the star burns out and undergoes catastrophic
1737: gravitational collapse. From an outside observer's point of view, the
1738: star will form a black hole whose surface area will be at least
1739: $4S_0$, in accordance with the generalized second law of
1740: thermodynamics.
1741:
1742: However, we can follow the star as it falls through its own horizon.
1743: From collapse solutions (see, e.g., Misner, Thorne, and Wheeler,
1744: 1973), it is known that the star will shrink to zero radius and end in
1745: a singularity. In particular, its surface area becomes arbitrarily
1746: small: $A \to 0$. By the second law of thermodynamics, the entropy in
1747: the enclosed volume, i.e., the entropy of the star, must still be at
1748: least $S_0$. Once more, the spacelike entropy bound fails (Easther
1749: and Lowe, 1999).
1750:
1751: As in the previous two examples, this failure does not concern the
1752: spherical entropy bound, even though spherical symmetry may hold. We
1753: are considering a regime of dominant gravity, in violation of the
1754: assumptions of the spherical bound. In the interior of a black hole,
1755: both the curvature and the time-dependence of the metric are large.
1756:
1757:
1758: \subsubsection{Weakly gravitating system}
1759: \label{sec-fail4}
1760:
1761: The final example is the most subtle. It shows that the spacelike
1762: entropy bound can be violated by the very systems for which the
1763: spherical entropy bound is believed to hold: spherical, weakly
1764: gravitating systems. This is achieved merely by a non-standard
1765: coordinate choice that breaks spherical symmetry and measures a
1766: smaller surface area.
1767:
1768: Consider a weakly gravitating spherical thermodynamic system in
1769: asymptotically flat space. Note that this class includes most
1770: thermodynamic systems studied experimentally; if they are not
1771: spherical, one redefines their boundary to be the circumscribing
1772: sphere.
1773:
1774: A coordinate-independent property of the system is its world volume,
1775: $W$. For a stable system with the spatial topology of a
1776: three-dimensional ball ($\mathbf{D}^3$), the topology of $W$ is given
1777: by ${\mathbb R}\times \mathbf{D}^3$ (Fig.~\ref{fig-ball}).
1778: %%***************************************************************
1779: \begin{figure}
1780: \includegraphics[width=7.5cm]{fig-ball}
1781: \caption{The worldvolume of a ball of gas, with one spatial dimension
1782: suppressed. (a) A time slice in the rest frame of the system is shown
1783: as a flat plane. It intersects the boundary of system on a spherical
1784: surface, whose area exceeds the system's entropy. (b) In a different
1785: coordinate system, however, a time slice intersects the boundary on
1786: Lorentz-contracted surfaces whose area can be made arbitrarily small.
1787: Thus the spacelike entropy bound is violated. (c) The light-sheet of a
1788: spherical surface is shown for later reference (Sec.~\ref{sec-els}).
1789: Light-sheets of wiggly surfaces may not penetrate the entire system
1790: (Sec.~\ref{sec-nearnull}).---The solid cylinder depicted here can also
1791: be used to illustrate the conformal shape of Anti-de~Sitter space
1792: (Sec.~\ref{sec-ads}).}
1793: \label{fig-ball}
1794: \end{figure}
1795: %%***************************************************************
1796:
1797: The volume of the ball of gas, at an instant of time, is geometrically
1798: the intersection of the world volume $W$ with an equal time
1799: hypersurface $t=0$:
1800: \begin{equation}
1801: V \equiv W \cap \{t=0\}.
1802: \label{eq-vwt}
1803: \end{equation}
1804: The boundary of the volume $V$ is a surface $B$ given by
1805: \begin{equation}
1806: B = \partial W \cap \{t=0\}.
1807: \label{eq-bvwt}
1808: \end{equation}
1809:
1810: The time coordinate $t$, however, is not uniquely defined. One
1811: possible choice for $t$ is the proper time in the rest frame of the
1812: weakly gravitating system (Fig.~\ref{fig-ball}a). With this choice,
1813: $V$ and $B$ are metrically a ball and a sphere, respectively. The
1814: area $A(B)$ and the entropy $S_{\rm matter}(V)$ were calculated in
1815: Sec.~\ref{sec-gb} for the example of a ball of gas. They were found
1816: to satisfy the spacelike entropy bound, Eq.~(\ref{eq-seb}).
1817:
1818: From the point of view of general relativity, there is nothing special
1819: about this choice of time coordinate. The laws of physics must be {\em
1820: covariant}, i.e., invariant under general coordinate transformations.
1821: Thus Eq.~(\ref{eq-seb}) must hold also for a volume $V'$ associated
1822: with a different choice of time coordinate, $t'$. In particular, one
1823: may choose the $t'=$const hypersurface to be rippled like a fan. Then
1824: its intersection with $\partial W$, $B'$, will be almost null almost
1825: everywhere, like the zigzag line circling the worldvolume in
1826: Fig.~\ref{fig-ball}b. The boundary area so defined can be made
1827: arbitrarily small (Jacobson, 1999; Flanagan, Marolf, and Wald, 2000;
1828: Smolin, 2001).%
1829: %
1830: \footnote{The following construction exemplifies this for a
1831: spherical system. Consider the spatial $D-2$ sphere $B$ defined by
1832: $t=0$ and parametrized by standard spherical coordinates
1833: $(\theta_1,\ldots,\theta_{D-3},\varphi)$. Divide $B$ into $2n$
1834: segments of longitude defined by ${k\over 2n} \leq {\varphi\over
1835: 2\pi} < {k+1\over 2n}$ with $k=0\ldots 2n-1$. By translation of $t$
1836: this segmentation carries over to $\partial W$. For each even (odd)
1837: segment, consider a Lorentz observer boosted with velocity $\beta$
1838: in the positive (negative) $\varphi$ direction at the midpoint of
1839: the segment on the equator of $B$. The time foliations of these
1840: $2n$ observers, restricted respectively to each segment and joined
1841: at the segment boundaries, define global equal time hypersurfaces.
1842: The slices can be smoothed at the segment boundaries and in the
1843: interior of $W$ without affecting the conclusions. After picking a
1844: particular slice, $t'=0$, a volume $V'$ and its boundary $B'$ can be
1845: defined in analogy with Eqs.~(\ref{eq-vwt}) and (\ref{eq-bvwt}).
1846: %
1847: Since $V'$ contains the entire thermodynamic system, the entropy is
1848: not affected by the new coordinate choice:
1849: %
1850: $S_{\rm matter}(V') = S_{\rm matter}(V)$.
1851: %
1852: Because of Lorentz contraction, the proper area $A(B')$ is
1853: smaller than $A(B)$. Indeed, by taking $\beta \to 1$ and
1854: $n\to\infty$ one can make $A(B')$ arbitrarily small:
1855: %
1856: $A(B')~~\stackrel{n\to\infty}{\longrightarrow}~~
1857: A(B) \sqrt{1-\beta^2}~~
1858: \stackrel{\beta\to 1}{\longrightarrow}~~ 0.$
1859: %
1860: An analogous construction for a square system takes a simpler form;
1861: see Sec.~\ref{sec-nearnull}.}
1862: %
1863: This construction has shown that a spherical system with non-zero
1864: entropy $S_{\rm matter}$ can be enclosed within a surface of area
1865: $A(B')<S_{\rm matter}$, and the spacelike entropy bound,
1866: Eq.~(\ref{eq-seb}), is again violated.
1867:
1868: How is this possible? After all, the spherical entropy bound should
1869: hold for this system, because it can be converted into a spherical
1870: black hole of the same area. However, this argument implicitly
1871: assumed that the boundary of a spherically symmetric system is a
1872: sphere (and therefore agrees with the horizon area of the black hole
1873: after the conversion). With the non-standard time coordinate $t'$,
1874: however, the boundary is not spherically symmetric, and its area is
1875: much smaller than the final black hole area. (The latter is unaffected
1876: by slicing ambiguities because a black hole horizon is a null
1877: hypersurface.)
1878:
1879:
1880: \subsection{Range of validity}
1881: \label{sec-range}
1882:
1883: In view of these problems, it is clear that the spacelike entropy
1884: bound cannot be maintained as a fully general conjecture holding for
1885: all volumes and areas in all spacetimes. Still, the spherical entropy
1886: bound, Eq.~(\ref{eq-spheb}), clearly holds for many systems that do
1887: not satisfy its assumptions, suggesting that those assumptions may be
1888: unnecessarily restrictive.
1889:
1890: For example, the earth is part of a cosmological spacetime that is
1891: not, as far as we know, asymptotically flat. However, the earth does
1892: not curve space significantly. It is well separated from other matter
1893: systems. On time and distance scales comparable to the earth's
1894: diameter, the universe is effectively static and flat. In short, it
1895: is clear that the earth will obey the spacelike entropy bound.%
1896: %
1897: \footnote{Pathological slicings such as the one in
1898: Sec.~\ref{sec-fail4} must still be avoided. Here we define the
1899: earth's surface area by the natural slicing in its approximate Lorentz
1900: frame.}
1901: %
1902:
1903: The same argument can be made for the solar system, and even for the
1904: milky way. As we consider larger regions, however, the effects of
1905: cosmological expansion become more noticable, and the flat space
1906: approximation is less adequate. An important question is whether a
1907: definite line can be drawn. In cosmology, is there a largest region
1908: to which the spacelike entropy bound can be reliably applied? If so,
1909: how is this region defined? Or does the bound gradually become less
1910: accurate at larger and larger scales?%
1911: %
1912: \footnote{The same questions can be asked of the Bekenstein bound,
1913: Eq.~(\ref{eq-bekbound}). Indeed, Bekenstein (1989), who proposed its
1914: application to the past light cone of an observer, was the first to
1915: raise the issue of the validity of entropy bounds in cosmology.}
1916:
1917: Let us consider homogeneous, isotropic universes, known as
1918: Friedmann-Robertson-Walker (FRW) universes (Sec.~\ref{sec-cosmo}).
1919: Fischler and Susskind (1998) abandoned the spacelike formulation
1920: altogether (Sec.~\ref{sec-mot}). For adiabatic FRW universes,
1921: however, their proposal implied that the spacelike entropy bound
1922: should hold for spherical regions smaller than the particle horizon
1923: (the future light cone of a point at the big bang).
1924:
1925: Restriction to the particle horizon turns out to be sufficient for the
1926: validity of the spacelike entropy bound in simple flat and open
1927: models; thus, the problem in Sec.~\ref{sec-fail2} is resolved.
1928: However, it does not prevent violations in closed or collapsing
1929: universes. The particle horizon area vanishes when the light cone
1930: reaches the far end of a closed universe---this is a special case of
1931: the problem discussed in Sec.~\ref{sec-fail1}. An analogue of the
1932: problem of Sec.~\ref{sec-fail3} can arise also. Generally, closed
1933: universes and collapsing regions exhibit the greatest difficulties for
1934: the formulation of entropy bounds, and many authors have given them
1935: special attention.
1936:
1937: Davies (1988) and Brustein (2000) proposed a generalized second law
1938: for cosmological spacetimes. They suggested that contradictions in
1939: collapsing universes may be resolved by augmenting the area law with
1940: additional terms. Easther and Lowe (1999) argued that the second law
1941: of thermodynamics implies a holographic entropy bound, at least for
1942: flat and open universes, in regions not exceeding the Hubble
1943: horizon.\footnote{The Hubble radius is defined to be ${a\over da/dt}$,
1944: where $a$ is the scale factor of the universe; see Eq.~(\ref{eq-FRW1})
1945: below.} Similar conclusions were reached by Veneziano (1999b),
1946: Kaloper and Linde (1999), and Brustein (2000).
1947:
1948: Bak and Rey (2000a) argued that the relevant surface is the apparent
1949: horizon, defined in Sec.~\ref{sec-ah}. This is a minor distinction
1950: for typical flat and open universes, but it avoids some of the
1951: difficulties with closed universes.\footnote{Related discussions also
1952: appear in Dawid (1999) and Kalyana Rama (1999). The continued debate
1953: of the difficulties of the Fischler-Susskind proposal in closed
1954: universes (Wang and Abdalla, 1999, 2000; Cruz and Lepe, 2001) is, in
1955: our view, rendered nugatory by the covariant entropy bound.}
1956:
1957: The arguments for bounds of this type return to the Susskind process,
1958: the gedankenexperiment by which the spherical entropy bound was
1959: derived (Sec.~\ref{sec-susskind}). A portion of the universe is
1960: converted to a black hole; the second law of thermodynamics is
1961: applied. One then tries to understand what might prevent this
1962: gedankenexperiment from being carried out.
1963:
1964: For example, regions larger than the horizon are expanding too rapidly
1965: to be converted to a black hole---they cannot be ``held together''
1966: (Veneziano, 1999b). Also, if a system is already inside a black hole,
1967: it can no longer be converted to one. Hence, one would not expect the
1968: bound to hold in collapsing regions, such as the interior of black
1969: holes or a collapsing universe (Easther and Lowe, 1999; Kaloper and
1970: Linde, 1999).
1971:
1972: This reasoning does expose some of the limitations of the spacelike
1973: entropy bound (namely those that are illustrated by the explicit
1974: counterexamples given in Sec.~\ref{sec-fail1} and \ref{sec-fail3}).
1975: However, it fails to identify sufficient conditions under which the
1976: bound is actually reliable. Kaloper and Linde (1999) give
1977: counterexamples to any statement of the type ``The area of the
1978: particle (apparent, Hubble) horizon always exceeds the entropy
1979: enclosed in it'' (Sec.~\ref{sec-na}).
1980:
1981: In the following section we will introduce the covariant entropy
1982: bound, which is formulated in terms of light-sheets. In
1983: Sec.~\ref{sec-tests} we will present evidence that this bound has
1984: universal validity. Starting from this general bound, one can find
1985: sufficient conditions under which a spacelike formulation is valid
1986: (Sec.~\ref{sec-spt}, \ref{sec-cosmocor}). However, the conditions
1987: themselves will involve the light-sheet concept in an essential way.
1988: Not only is the spacelike formulation less general than the
1989: light-sheet formulation; the range of validity of the former cannot be
1990: reliably identified without the latter.
1991:
1992: We conclude that the spacelike entropy bound is violated by realistic
1993: matter systems. In cosmology, its range of validity cannot be
1994: intrinsically defined.
1995:
1996:
1997:
1998: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1999: \section{The covariant entropy bound}
2000: \label{sec-bb}
2001: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2002:
2003: In this section we present a more successful generalization of
2004: Eq.~(\ref{eq-spheb}): the covariant entropy bound.
2005:
2006: There are two significant formal differences between the covariant
2007: bound and the spacelike bound, Eq.~(\ref{eq-seb}). The spacelike
2008: formulation starts with a choice of spatial volume $V$. The volume,
2009: in turn, defines a boundary $B =\partial V$, whose area $A$ is then
2010: claimed to be an upper bound on $S(V)$, the entropy in $V$. The
2011: covariant bound proceeds in the opposite direction. A codimension 2
2012: surface $B$ serves as the starting point for the construction of a
2013: codimension 1 region $L$. This is the first formal difference. The
2014: second is that $L$ is a null hypersurface, unlike $V$ which is
2015: spacelike.
2016:
2017: More precisely, $L$ is a {\em light-sheet}. It is constructed by
2018: following light rays that emanate from the surface $B$, as long as
2019: they are not expanding. There are always at least two suitable
2020: directions away from $B$ (Fig.~\ref{fig-spheresheets}). When
2021: light rays self-intersect, they start to expand. Hence, light-sheets
2022: terminate at focal points.
2023:
2024: The {\sl covariant entropy bound} states that {\em the entropy on any
2025: light-sheet of a surface $B$ will not exceed the area of $B$}:
2026: \begin{equation}
2027: S[L(B)]\leq {A(B)\over 4}.
2028: \label{eq-bb}
2029: \end{equation}
2030: We will give a more formal definition at the end of this section.
2031: %%***************************************************************
2032: \begin{figure}[h] \centering
2033: \includegraphics[width=7cm]{fig-spheresheets}
2034: \caption{The four null hypersurfaces orthogonal to a spherical surface
2035: $B$. The two cones $F_1$, $F_3$ have negative expansion and hence
2036: correspond to light-sheets. The covariant entropy bound states that
2037: the entropy on each light-sheet will not exceed the area of $B$. The
2038: other two families of light rays, $F_2$ and $F_4$, generate the skirts
2039: drawn in thin outline. Their cross-sectional area is increasing, so
2040: they are not light-sheets. The entropy of the skirts is not related
2041: to the area of $B$.---Compare this figure to Fig.~\ref{fig-spheb}.}
2042: \label{fig-spheresheets}
2043: \end{figure}
2044: %%***************************************************************
2045:
2046: We begin with some remarks on the conjectural nature of the bound, and
2047: we mention related earlier proposals. We will explain the geometric
2048: construction of light-sheets in detail, giving special attention to
2049: the considerations that motivate the condition of non-expansion
2050: ($\theta\leq 0$). We give a definition of entropy on light-sheets,
2051: and we discuss the extent to which the limitations of classical
2052: general relativity are inherited by the covariant entropy bound. We
2053: then summarize how the bound is formulated, applied, and tested.
2054: Parts of this section follow Bousso (1999a).
2055:
2056:
2057: \subsection{Motivation and background}
2058: \label{sec-mot}
2059:
2060: There is no fundamental derivation of the covariant entropy bound. We
2061: present the bound because there is strong evidence that it holds
2062: universally in nature. The geometric construction is well-defined and
2063: covariant. The resulting entropy bound can be saturated, but no
2064: example is known where it is exceeded.
2065:
2066: In Sec.~\ref{sec-fmw} plausible relations between entropy and energy
2067: are shown to be sufficient for the validity of the bound. But these
2068: relations do not at present appear to be universal or fundamental. In
2069: special situations, the covariant entropy bound reduces to the
2070: spherical entropy bound, which is arguably a consequence of black hole
2071: thermodynamics. But in general, the covariant entropy bound cannot be
2072: inferred from black hole physics; quite conversely, the generalized
2073: second law of thermodynamics may be more appropriately regarded as a
2074: consequence of the covariant bound (Sec.~\ref{sec-gia}).
2075:
2076: The origin of the bound remains mysterious. As discussed in the
2077: introduction, this puzzle forms the basis of the holographic
2078: principle, which asserts that the covariant entropy bound betrays the
2079: number of degrees of freedom of quantum gravity (Sec.~\ref{sec-hp}).
2080:
2081: Aside from its success, little motivation for a light-like formulation
2082: can be offered. Under the presupposition that {\em some\/} general
2083: entropy bound waits to be discovered, one is guided to light rays by
2084: circumstantial evidence. This includes the failure of the spacelike
2085: entropy bound (Sec.~\ref{sec-seb}), the properties of the Raychaudhuri
2086: equation (Sec.~\ref{sec-ray}), and the loss of a dynamical dimension
2087: in the light cone formulation of string theory
2088: (Sec.~\ref{sec-further}).
2089:
2090: Whatever the reasons, the idea that light rays might be involved in
2091: relating a region to its surface area---or, rather, relating a surface
2092: area to a light-like ``region''---arose in discussions of the
2093: holographic principle from the beginning.
2094:
2095: Susskind (1995b) suggested that the horizon of a black hole can be
2096: mapped, via light rays, to a distant, flat holographic screen, citing
2097: the focussing theorem (Sec.~\ref{sec-ray}) to argue that the
2098: information thus projected would satisfy the holographic bound.
2099: Corley and Jacobson (1996) pointed out that the occurrence of focal
2100: points, or {\em caustics}, could invalidate this argument, but showed
2101: that one caustic-free family of light rays existed in Susskind's
2102: example. They further noted that both past and future directed
2103: families of light rays can be considered.
2104:
2105: Fischler and Susskind (1998) recognized that a light-like formulation
2106: is crucial in cosmological spacetimes, because the spacelike entropy
2107: bound fails. They proposed that any spherical surface $B$ in FRW
2108: cosmologies (see Sec.~\ref{sec-cosmo}) be related to (a portion of) a
2109: light cone that comes from the past and ends on $B$. This solved the
2110: problem discussed in Sec.~\ref{sec-fail2} for flat and open universes but
2111: not the problem of small areas in closed or recollapsing universes
2112: (see Secs.~\ref{sec-fail1}, \ref{sec-fail3}).
2113:
2114: The covariant entropy bound (Bousso, 1999a) can be regarded as a
2115: refinement and generalization of the Fischler-Susskind proposal. It
2116: can be applied in arbitrary spacetimes, to any surface $B$ regardless
2117: of shape, topology, and location. It considers all four null
2118: directions orthogonal to $B$ without prejudice. It introduces a new
2119: criterion, the contraction of light rays, both to select among the
2120: possible light-like directions and to determine how far the light rays
2121: may be followed. For any $B$, there will be at least two ``allowed''
2122: directions and hence two light-sheets, to each of which the bound applies
2123: individually.
2124:
2125:
2126:
2127: \subsection{Light-sheet kinematics}
2128: \label{sec-kin}
2129:
2130: Compared to the previously discussed bounds, Eqs.~(\ref{eq-spheb}) and
2131: (\ref{eq-seb}), the non-trivial ingredient of the covariant entropy
2132: bound lies in the concept of light-sheets. Given a surface, a
2133: light-sheet defines an adjacent spacetime region whose entropy should
2134: be considered. What has changed is not the formula, $S\leq A/4$, but
2135: the prescription that determines where to look for the entropy $S$
2136: that enters that formula. Let us discuss in detail how light-sheets
2137: are constructed.
2138:
2139: \subsubsection{Orthogonal null hypersurfaces}
2140: \label{sec-kin1}
2141:
2142: A given surface $B$ possesses precisely four orthogonal null
2143: directions (Fig.~\ref{fig-spheresheets}). They are sometimes referred
2144: to as {\em future directed ingoing}, {\em future directed outgoing},
2145: {\em past directed ingoing}, and {\em past directed outgoing}, though
2146: ``in'' and ``out'' are not always useful labels. Locally, these
2147: directions can be represented by null hypersurfaces $F_1,\ldots,F_4$
2148: that border on $B$. The $F_i$ are generated by the past and the
2149: future directed light rays orthogonal to $B$, on either side of $B$.
2150:
2151: For example, suppose that $B$ is the wall of a (spherical) room in
2152: approximately flat space, as shown in Fig.~\ref{fig-spheresheets}, at
2153: $t=0$. (We must keep in mind that $B$ denotes a surface at some
2154: instant of time.) Then the future directed light rays towards the
2155: center of the room generate a null hypersurface $F_1$, which looks
2156: like a light cone. A physical way of describing $F_1$ is to imagine
2157: that the wall is lined with light bulbs that all flash up at $t=0$.
2158: As the light rays travel towards the center of the room they generate
2159: $F_1$.
2160:
2161: Similarly, one can line the outside of the wall with light bulbs.
2162: Future directed light rays going to the outside will generate a second
2163: null hypersurface $F_2$. Finally, one can also send light rays
2164: towards the past. (We might prefer to think of these as arriving from
2165: the past, i.e., a light bulb in the center of the room flashed at an
2166: appropriate time for its rays to reach the wall at $t=0$.) In any
2167: case, the past directed light rays orthogonal to $B$ will generate two
2168: more null hypersurfaces $F_3$ and $F_4$.
2169:
2170: In Fig.~\ref{fig-spheresheets}, the two ingoing cones $F_1$ and $F_3$,
2171: and the two outgoing ``skirts'', $F_2$ and $F_4$, are easily seen to
2172: be null and orthogonal to $B$. However, the existence of four null
2173: hypersurfaces bordering on $B$ is guaranteed in Lorentzian geometry
2174: independently of the shape and location of $B$. They are always
2175: uniquely generated by the four sets of surface-orthogonal light rays.
2176:
2177: At least two of the four null hypersurfaces $F_1,\ldots,F_4$ will be
2178: selected as light-sheets, according to the condition of non-positive
2179: expansion discussed next.
2180:
2181:
2182: \subsubsection{Light-sheet selection}
2183: \label{sec-kin2}
2184:
2185: Let us return to the example where $B$ is the wall of a spherical
2186: room. If gravity is weak, one would expect that the area $A$ of $B$
2187: will be a bound on the entropy in the room (Sec.~\ref{sec-susskind}).
2188: Clearly, $A$ cannot be related in any way to the entropy in the
2189: infinite region outside the room; that entropy could be arbitrarily
2190: large. It appears that we should select $F_1$ or $F_3$ as
2191: light-sheets in this example, because they correspond to our intuitive
2192: notion of ``inside''.
2193:
2194: The question is how to generalize this notion. It is obvious that one
2195: should compare an area only to entropy that is in some sense
2196: ``inside'' the area. However, consider a closed universe, in which
2197: space is a three-sphere. As Sec.~\ref{sec-fail1} has illustrated, we
2198: need a criterion that prevents us from considering the large part of
2199: the three-sphere to be ``inside'' a small two-sphere $B$.
2200:
2201: What we seek is a local condition, which will select whether some
2202: direction away from $B$ is an inside direction. This condition should
2203: reduce to the intuitive, global notion---inside is where infinity is
2204: not---where applicable. An analogy in Euclidean space leads to a
2205: useful criterion, the {\em contraction condition}.
2206:
2207: %%***************************************************************
2208: \begin{figure}[h] \centering
2209: \includegraphics[width=8.5cm]{fig-inside}
2210: \caption{Local definition of ``inside''. (a) Ingoing rays
2211: perpendicular to a convex surface in a Euclidean geometry span
2212: decreasing area. This motivates the following local definition. (b)
2213: Inside is the direction in which the cross-sectional area decreases
2214: ($A'\leq A$). This criterion can be applied to light rays orthogonal
2215: to any surface. After light rays locally intersect, they begin to
2216: expand. Hence, light-sheets must be terminated at caustics.}
2217: \label{fig-inside}
2218: \end{figure}
2219: %%***************************************************************
2220: Consider a convex closed surface $B$ of codimension one and area $A$
2221: in flat Euclidean space, as shown in Fig.~\ref{fig-inside}a. Now
2222: construct all the geodesics intersecting $B$ orthogonally. Follow
2223: each geodesic an infinitesimal proper distance $dl$ to one of the two
2224: sides of $B$. The set of points thus obtained will span a similarly
2225: shaped surface of area $A'$. If $A'<A$, let us call the chosen
2226: direction the ``inside''. If $A'>A$, we have gone ``outside''.
2227:
2228: Unlike the standard notion of ``inside'', the contraction criterion
2229: does not depend on any knowledge of the global properties of $B$ and
2230: of the space it is embedded in. It can be applied independently to
2231: arbitrarily small pieces of the surface. One can always construct
2232: orthogonal geodesics and ask in which direction they contract. It is
2233: local also in the orthogonal direction; the procedure can be repeated
2234: after each infinitesimal step.
2235:
2236: Let us return to Lorentzian signature, and consider a codimension 2
2237: spatial surface $B$. The contraction criterion cannot be used to find
2238: a {\em spatial\/} region ``inside'' $B$. There are infinitely many
2239: different spacelike hypersurfaces $\Sigma$ containing $B$. Which side
2240: has contracting area could be influenced by the arbitrary choice of
2241: $\Sigma$.
2242:
2243: However, the four {\em null\/} directions $F_1,\ldots,F_4$ away from
2244: $B$ are uniquely defined. It is straightforward to adapt the
2245: contraction criterion to this case. Displacement by an infinitesimal
2246: spatial distance is meaningless for light rays, because two points on
2247: the same light ray always have distance zero. Rather, an appropriate
2248: analogue to length is the affine parameter $\lambda$ along the
2249: light ray (see the Appendix). Pick a particular direction $F_i$.
2250: Follow the orthogonal null geodesics away from $B$ for an
2251: infinitesimal affine distance $d\lambda$. The points thus constructed
2252: span a new surface of area $A'$. If $A' \leq A$, then the direction
2253: $F_i$ will be considered an ``inside'' direction, or {\em light-sheet
2254: direction}.
2255:
2256: By repeating this procedure for $i=1,\ldots,4$, one finds all null
2257: directions that point to the ``inside'' of $B$ in this technical
2258: sense. Because the light rays generating opposite pairs of null
2259: directions (e.g., $F_1$ and $F_4$) are continuations of each other, it
2260: is clear that at least one member of each pair will be considered
2261: inside. If the light rays are locally neither expanding nor
2262: contracting, both members of a pair will be called ``inside''. Hence,
2263: there will always be at least two light-sheet directions. In
2264: degenerate cases, there may be three or even four.
2265:
2266: Mathematically, the contraction condition can be formulated thus:
2267: \begin{equation}
2268: \theta(\lambda)\leq 0~~~~\mbox{for}~\lambda=\lambda_0,
2269: \label{eq-theta0}
2270: \end{equation}
2271: where $\lambda$ is an affine parameter for the light rays generating
2272: $F_i$ and we assume that $\lambda$ increases in the direction away
2273: from $B$. $\lambda_0$ is the value of $\lambda$ on $B$. The {\em
2274: expansion}, $\theta$, of a family of light rays is discussed in detail
2275: in Sec.~\ref{sec-ray}. It can be understood as follows. Consider a
2276: bunch of infinitesimally neighboring light rays spanning a surface
2277: area ${\cal A}$. Then
2278: \begin{equation}
2279: \theta(\lambda) \equiv \frac{d{\cal A}/d\lambda}{\cal A}.
2280: \label{eq-tda}
2281: \end{equation}
2282:
2283: As in the Euclidean analogy, this condition can be applied to each
2284: infinitesimal surface element separately and so is local. Crucially,
2285: it applies to open surfaces as well as to closed ones. This
2286: represents a significant advance in the generality of the formulation.
2287:
2288: For oddly shaped surfaces or very dynamical spacetimes, it is possible
2289: for the expansion to change sign along some $F_i$. For example, this
2290: will happen for smooth concave surfaces in flat space. Because of the
2291: locality of the contraction criterion, one may split such surfaces
2292: into pieces with constant sign, and continue the analysis for each
2293: piece separately. This permits us to assume henceforth without loss
2294: of generality that the surfaces we consider have continuous
2295: light-sheet directions.
2296:
2297: For the simple case of the spherical surface in Minkowski space, the
2298: condition (\ref{eq-tda}) reproduces the intuitive answer. The area is
2299: decreasing in the $F_1$ and $F_3$ directions---the past and future
2300: directed light rays going to the center of the sphere. We will call
2301: any such surface, with two light-sheet directions on the same spatial
2302: side, {\em normal}.
2303:
2304: In highly dynamical geometries, the expansion or contraction of space
2305: can be the more important effect on the expansion of light rays. Then
2306: it will not matter which spatial side they are directed at. For
2307: example, in an expanding universe, areas get small towards the past,
2308: because the big bang is approached. A sufficiently large sphere will
2309: have two past directed light-sheets, but no future directed ones. A
2310: surface of this type is called {\em anti-trapped}. Similarly, in a
2311: collapsing universe or inside a black hole, space can shrink so
2312: rapidly that both light-sheets are future directed. Surfaces with
2313: this property are {\em trapped}.
2314:
2315: In a Penrose diagram (Appendix), a sphere is represented by a point.
2316: The four orthogonal null directions correspond to the four legs of an
2317: ``X'' centered on this point. Light-sheet directions can be indicated
2318: by drawing only the corresponding legs (Bousso 1999a). Normal,
2319: trapped, and anti-trapped surfaces are thus denoted by wedges of
2320: different orientation (see Figs.~\ref{fig-flatfrw}, \ref{fig-clos}a,
2321: and \ref{fig-star}).
2322: %%***************************************************************
2323: \begin{figure}[h] \centering
2324: \includegraphics[width=5cm]{fig-flatfrw}
2325: \caption{Penrose diagram for an expanding universe (a flat or open FRW
2326: universe, see Sec.~\ref{sec-cosmo}). The thin curve is a slice of
2327: constant time. Each point in the interior of the diagram represents a
2328: sphere. The wedges indicate light-sheet directions. The apparent
2329: horizon (shown here for equation of state $p=\rho$) divides the normal
2330: spheres near the origin from the anti-trapped spheres near the big
2331: bang. The light-sheets of any sphere $B$ can be represented by
2332: inspecting the wedge that characterizes the local domain and drawing
2333: lines away from the point representing $B$, in the direction of the
2334: wedge's legs.}
2335: \label{fig-flatfrw}
2336: \end{figure}
2337: %%***************************************************************
2338:
2339:
2340: \subsubsection{Light-sheet termination}
2341: \label{sec-kin3}
2342:
2343: From now on we will consider only inside directions, $F_j$, where $j$
2344: runs over two or more elements of $\{1,2,3,4\}$. For each $F_j$, a
2345: light-sheet is generated by the corresponding family of light rays. In
2346: the example of the spherical surface in flat space, the light-sheets
2347: are cones bounded by $B$, as shown in Fig.~\ref{fig-spheresheets}.
2348:
2349: Strictly speaking, however, there was no particular reason to stop at
2350: the tip of the cone, where all light rays intersect. On the other
2351: hand, it would clearly be desastrous to follow the light rays
2352: arbitrarily far. They would generate another cone which would grow
2353: indefinitely, containing unbounded entropy. One must enforce, by some
2354: condition, that the light-sheet is terminated before this happens. In
2355: all but the most special cases, the light rays generating a
2356: light-sheet will not intersect in a single point, so the condition
2357: must be more general.
2358:
2359: A suitable condition is to demand that the expansion be non-positive
2360: {\em everywhere\/} on the light-sheet, and not only near $B$:
2361: \begin{equation}
2362: \theta(\lambda)\leq 0,
2363: \label{eq-theta}
2364: \end{equation}
2365: for all values of the affine parameter on the light-sheet.
2366:
2367: By construction (Sec.~\ref{sec-kin2}) the expansion is initially
2368: negative or zero on any light-sheet. Raychaudhuri's equation
2369: guarantees that the expansion can only decrease.. (This will be shown
2370: explicitly in Sec.~\ref{sec-ray}.) The only way $\theta$ can become
2371: positive is if light rays intersect, for example at the tip of the
2372: light cone. However, it is not necessary for all light rays to
2373: intersect in the same point. By Eq.~(\ref{eq-tda}), the expansion
2374: becomes positive at any {\em caustic}, that is, any place where a
2375: light ray crosses an infinitesimally neighboring light ray in the
2376: light-sheet (Fig.~\ref{fig-inside}b).
2377:
2378: Thus, Eq.~(\ref{eq-theta}) operates independently of any symmetries in
2379: the setup. It implies that light-sheets end at {\em
2380: caustics}.\footnote{If the null energy condition (Appendix) is
2381: violated, the condition (\ref{eq-theta}) can also terminate
2382: light-shees at non-caustic points.} In general, each light ray in a
2383: light-sheet will have a different caustic point, and the resulting
2384: caustic surfaces can be very complicated. The case of a light cone is
2385: special in that all light rays share the same caustic point at the
2386: tip. An ellipsoid in flat space will have a self-intersecting
2387: light-sheet that may contain the same object more than once (at two
2388: different times). Gravitational back-reaction of matter will make the
2389: caustic surfaces even more involved.
2390:
2391: Non-local self-intersections of light rays do not lead to violations
2392: of the contraction condition, Eq.~(\ref{eq-theta}). That is, the
2393: light-sheet must be terminated only where a light ray intersects its
2394: neighbor, but not necessarily when it intersects another light ray
2395: coming from a different portion of the surface $B$. One can consider
2396: modifications of the light-sheet definition where any
2397: self-intersection terminates the light-sheet (Tavakol and Ellis, 1999;
2398: Flanagan, Marolf, and Wald, 2000). Since this modification can only
2399: make light-sheets shorter, it can weaken the resulting bound.
2400: However, in most applications, the resulting light-sheets are easier
2401: to calculate (as Tavakol and Ellis, in particular, have stressed) and
2402: still give useful bounds.%
2403: %
2404: \footnote{Low (2002) has argued that the future directed light-sheets
2405: in cosmological spacetimes can be made arbitrarily extensive by
2406: choosing a closed surface containing sufficiently flat pieces. Low
2407: concludes that the covariant entropy bound is violated in standard
2408: cosmological solutions, unless it is modified to terminate
2409: light-sheets also at non-local self-intersections.---This reasoning
2410: overlooks that any surface element with local curvature radius larger
2411: than the apparent horizon possesses only past directed light-sheets
2412: (Bousso, 1999a; see Sec.~\ref{sec-ah}). Independently of the
2413: particular flaw in Low's argument, the conclusion is also directly
2414: invalidated by the proof of Flanagan, Marolf, and Wald (2000). (This
2415: is just as well, as the modification advocated by Low would not have
2416: solved the problem; non-local intersections can be suppressed by
2417: considering open surfaces.)}
2418:
2419: The condition, Eq.~(\ref{eq-theta}), subsumes Eq.~(\ref{eq-theta0}),
2420: which applied only to the initial value of $\lambda$. It is
2421: satisfying that both the direction and the extent of light-sheets are
2422: determined by the same simple condition, Eq.~(\ref{eq-theta}).
2423:
2424:
2425:
2426: \subsection{Defining entropy}
2427: \label{sec-edef}
2428:
2429: \subsubsection{Entropy on a fixed light-sheet}
2430: \label{sec-els}
2431:
2432: The geometric construction of light-sheets is well-defined.
2433: But how is ``the entropy on a light-sheet'', $S_{\rm matter}$,
2434: determined? Let us begin with an example where the definition of
2435: $S_{\rm matter}$ is obvious. Suppose that $B$ is a sphere around an
2436: isolated, weakly gravitating thermodynamic system. Given certain
2437: macroscopic constraints, for example an energy or energy range,
2438: pressure, volume, etc., the entropy of the system can be computed
2439: either thermodynamically, or statistically as the logarithm of the
2440: number of accessible quantum states.
2441:
2442: To good approximation, the two light-sheets of $B$ are a past and a
2443: future light cone. Let us consider the future directed light-sheet.
2444: The cone contains the matter system completely (Fig.~\ref{fig-ball}c),
2445: in the same sense in which a $t=\mbox{const}$ surface contains the
2446: system completely (Fig.~\ref{fig-ball}a). A light-sheet is just a
2447: different way of taking a snapshot of a matter system---in light cone
2448: time. (In fact, this comes much closer to how the system is actually
2449: observed in practice.) Hence, the entropy on the light-sheet is
2450: simply given by the entropy of the matter system.
2451:
2452: A more problematic case arises when the light-sheet intersects only a
2453: portion of an isolated matter system, or if there simply are no
2454: isolated systems in the spacetime. A reasonable (statistical) working
2455: definition was given by Flanagan, Marolf, and Wald (2000), who
2456: demanded that long wavelength modes which are not fully contained on
2457: the light-sheet should not be included in the entropy.
2458:
2459: In cosmological spacetimes, entropy is well approximated as a
2460: continuous fluid. In this case, $S_{\rm matter}$ is the integral of
2461: the entropy density over the light-sheet (Secs.~\ref{sec-fmw},
2462: \ref{sec-cosmo}).
2463:
2464: One would expect that the gravitational field itself can encode
2465: information perturbatively, in the form of gravitational waves.
2466: Because it is difficult to separate such structure from a ``background
2467: metric'', we will not discuss this case here.%
2468: %
2469: \footnote{Flanagan, Marolf, and Wald (2000) pointed out that
2470: perturbative gravitational entropy affects the light-sheet by
2471: producing shear, which in turn accelerates the focussing of light rays
2472: (Sec.~\ref{sec-lsd}). This suggests that the inclusion of such
2473: entropy will not lead to violations of the bound. Related research is
2474: currently pursued by Bhattacharya, Chamblin, and Erlich
2475: (2002).}
2476:
2477: We have formulated the covariant entropy bound for matter systems in
2478: classical geometry and have not made provisions for the inclusion of
2479: the semi-classical Bekenstein entropy of black holes. There is
2480: evidence, however, that the area of event horizons can be included in
2481: $S_{\rm matter}$. However, in this case the light-sheet must not be
2482: continued to the interior of the black hole. The Bekenstein entropy
2483: of the black hole already contains the information about objects that
2484: fell inside; it must not be counted twice (Sec.~\ref{sec-bhc}).
2485:
2486:
2487: \subsubsection{Entropy on an arbitrary light-sheet}
2488: \label{sec-es}
2489:
2490: So far we have treated the light-sheet of $B$ as a fixed null
2491: hypersurface, e.g., in the example of an isolated thermodynamic
2492: system. Different microstates of the system, however, correspond to
2493: different distributions of energy. This is a small effect on average,
2494: but it does imply that the geometry of light-sheets will vary with the
2495: state of the system in principle.
2496:
2497: In many examples, such as cosmological spacetimes, one can calculate
2498: light-sheets in a large-scale, averaged geometry. In this
2499: approximation, one can estimate $S_{\rm matter}$ while holding the
2500: light-sheet geometry fixed.
2501:
2502: In general, however, one can at best hold the surface $B$ fixed,%
2503: %
2504: \footnote{We shall take this to mean that the internal metric of the
2505: surface $B$ is held fixed. It may be possible to relax this further,
2506: for example by specifying only the area $A$ along with suitable
2507: additional restrictions.}
2508: %
2509: but not the light-sheet of $B$. We must consider $S_{\rm matter}$ to
2510: be the entropy on {\em any\/} light-sheet of $B$. Sec.~\ref{sec-csh},
2511: for example, discusses the collapse of a shell onto an apparent black
2512: hole horizon. In this example, a part of the spacetime metric is
2513: known, including $B$ and the initial expansions $\theta_i$ of its
2514: orthogonal light rays. However, the geometry to the future of $B$ is
2515: not presumed, and different configurations contributing to the entropy
2516: lead to macroscopically different future light-sheets.
2517:
2518: In a static, asymptotically flat space the specification of a
2519: spherical surface reduces to the specification of an energy range.
2520: The enclosed energy must lie between zero and the mass of a black hole
2521: that fills in the sphere. Unlike most other thermodynamic quantities
2522: such as energy, however, the area of surfaces is well-defined in
2523: arbitrary geometries.
2524:
2525: In the most general case, one may specify only a surface $B$ but no
2526: information about the embedding of $B$ in any spacetime. One is
2527: interested in the entropy of the ``fundamental system''
2528: (Sec.~\ref{sec-funds}), i.e., the number of quantum states associated
2529: with the light-sheets of $B$ in {\em any\/} geometry containing $B$.
2530: This leaves too much freedom for Eq.~(\ref{eq-bb}) to be checked
2531: explicitly. The covariant entropy bound essentially becomes the full
2532: statement of the holographic principle (Sec.~\ref{sec-hp}) in this
2533: limit.
2534:
2535:
2536:
2537: \subsection{Limitations}
2538: \label{sec-schlim}
2539:
2540: Here we discuss how the covariant entropy bound is tied to a regime of
2541: approximately classical spacetimes with reasonable matter content.
2542: The discussion of the ``species problem'' (Sec.~\ref{sec-species})
2543: carries over without significant changes and will not be repeated.
2544:
2545: \subsubsection{Energy conditions}
2546: \label{sec-econds}
2547:
2548: In Sec.~\ref{sec-gb} we showed that the entropy of a ball of radiation
2549: is bounded by $A^{3/4}$, and hence is less than its surface area. For
2550: larger values of the entropy, the mass of the ball would exceed its
2551: radius, so it would collapse to form a black hole. But what if matter
2552: of negative energy was added to the system? This would offset the
2553: gravitational backreaction of the gas without decreasing its entropy.
2554: The entropy in any region could be increased at will while keeping the
2555: geometry flat.
2556:
2557: This does not automatically mean that the holographic principle (and
2558: indeed, the generalized second law of thermodynamics) is wrong. A way
2559: around the problem might be to show that instabilities develop that
2560: will invalidate the setup we have just suggested. But more to the
2561: point, the holographic principle is expected to be a property of the
2562: real world. And to a good approximation, matter with negative mass
2563: does not exist in the real world.\footnote{We discuss quantum effects
2564: and a negative cosmological constant below.}
2565:
2566: Einstein's general relativity does not restrict matter content, but
2567: tells us only how matter affects the shape of spacetime. Yet, of all
2568: the types of matter that could be added to a Lagrangian, few actually
2569: occur in nature. Many would have pathological properties or
2570: catastrophic implications, such as the instability of flat space.
2571:
2572: In a unified theory underlying gravity and all other forces, one would
2573: expect that the matter content is dictated by the theory. String
2574: theory, for example, comes packaged with a particular field content in
2575: its perturbative limits. However, there are many physically
2576: interesting spacetimes that have yet to be described in string theory
2577: (Sec.~\ref{sec-strings}), so it would be premature to consider only
2578: fields arising in this framework.
2579:
2580: One would like to test the covariant entropy bound in a broad class of
2581: systems, but we are not interested whether the bound holds for matter
2582: that is entirely unphysical. It is reasonable to exclude matter whose
2583: energy density appears negative to a light ray, or which permits the
2584: superluminal transport of energy.%
2585: %
2586: \footnote{This demand applies to every matter component separately
2587: (Bousso, 1999a). This differs from the role of energy conditions in
2588: the singularity theorems (Hawking and Ellis, 1973), whose proofs are
2589: sensitive only to the total stress tensor. The above example shows
2590: that the total stress tensor can be innocuous when components of
2591: negative and positive mass are superimposed. An interesting question
2592: is whether instabilities lead to a separation of components, and thus
2593: to an eventual violation of energy conditions on the total stress
2594: tensor. We would like to thank J.~Bekenstein and A.~Mayo for raising
2595: this question.}
2596: %
2597: In other words, let us demand the null energy condition as well as the
2598: causal energy condition. Both conditions are spelled out in the
2599: Appendix, Eqs.~(\ref{eq-nec}) and (\ref{eq-cec}). They are believed
2600: to be satisfied classically by all physically reasonable forms of
2601: matter.%
2602: %
2603: \footnote{The dominant energy condition has sometimes been demanded
2604: instead of Eq.~(\ref{eq-nec}) and Eq.~(\ref{eq-cec}). It is a
2605: stronger condition that has the disadvantage of excluding a negative
2606: cosmological constant (Bousso, 1999a).---One can also ask whether, in
2607: a reversal of the logical direction, entropy bounds can be used to
2608: infer energy conditions that characterize physically acceptable matter
2609: (Brustein, Foffa, and Mayo, 2002).}
2610: %
2611:
2612: Negative energy density is generally disallowed by these conditions,
2613: with the exception of a negative cosmological constant. This is
2614: desirable, because a negative cosmological constant does not lead to
2615: instabilities or other pathologies. It may well occur in the
2616: universe, though it is not currently favored by observation. Unlike
2617: other forms of negative energy, a negative cosmological constant
2618: cannot be used to cancel the gravitational field of ordinary
2619: thermodynamic systems, so it should not lead to difficulties with the
2620: holographic principle.
2621:
2622: Quantum effects can violate the above energy conditions. Casimir
2623: energy, for example, can be negative. However, the relation between
2624: the magnitude, size, and duration of such violations is severely
2625: constrained (see, e.g., Ford and Roman, 1995, 1997, 1999; Flanagan,
2626: 1997; Fewster and Eveson, 1998; Fewster, 1999; further references are
2627: found in Borde, Ford, and Roman, 2001). Even where they occur, their
2628: gravitational effects may be overcompensated by those of ordinary
2629: matter. It has not been possible so far to construct a counterexample
2630: to the covariant entropy bound using quantum effects in ordinary
2631: matter systems.
2632:
2633:
2634: \subsubsection{Quantum fluctuations}
2635: \label{sec-qfluc}
2636:
2637: What about quantum effects in the geometry itself? The holographic
2638: principle refers to geometric concepts such as area, and orthogonal
2639: light rays. As such, it can be applied only where spacetime is
2640: approximately classical. This contradicts in no way its deep relation
2641: to quantum gravity, as inferred from the quantum aspects of black
2642: holes (Sec.~\ref{sec-bh}) and demonstrated by the AdS/CFT
2643: correspondence (Sec.~\ref{sec-ads}).
2644:
2645: In the real world, $\hbar$ is fixed, so the regime of classical
2646: geometry is generically found in the limit of low curvature and large
2647: distances compared to the Planck scale, Eq.~(\ref{eq-lpl}). Setting
2648: $\hbar$ to $0$ would not only be unphysical; as Lowe (1999) points
2649: out, it would render the holographic bound, $Akc^3/4G\hbar$, trivial.
2650:
2651: Lowe (1999) has argued that a naive application of the bound
2652: encounters difficulties when effects of quantum gravity become
2653: important. With sufficient fine tuning, one can arrange for an
2654: evaporating black hole to remain in equilibrium with ingoing radiation
2655: for an arbitarily long time. Consider the future directed outgoing
2656: light-sheet of an area on the black hole horizon. Lowe claims that
2657: this light-sheet will have exactly vanishing expansion and will
2658: continue to generate the horizon in the future, as it would in a
2659: classical spacetime. This would allow an arbitrarily large amount of
2660: ingoing radiation entropy to pass through the light-sheet, in
2661: violation of the covariant entropy bound.
2662:
2663: If a light-sheet lingers in a region that cannot be described by
2664: classical general relativity without violating energy conditions for
2665: portions of the matter, then it is outside the scope of the present
2666: formulation of the covariant entropy bound. The study of light-sheets
2667: of this type may guide the exploration of semi-classical
2668: generalizations of the covariant entropy bound. For example, it may
2669: be appropriate to associate the outgoing Hawking radiation with a
2670: negative entropy flux on this light-sheet (Flanagan, Marolf, and Wald,
2671: 2000).\footnote{More radical extensions have been proposed by
2672: Markopoulou and Smolin (1999) and by Smolin (2001).}
2673:
2674: However, Bousso (2000a) argued that a violation of the covariant
2675: entropy bound has not been demonstrated in Lowe's example. In any
2676: realistic situation small fluctuations in the energy density of
2677: radiation will occur. They are indeed inevitable if information is to
2678: be transported through the light-sheet. Thus the expansion along the
2679: light-sheet will fluctuate. If it becomes positive, the light-sheet
2680: must be terminated. If it fluctuates but never becomes positive, then
2681: it will be negative on average. In that case an averaged version of
2682: the focussing theorem implies that the light rays will focus within a
2683: finite affine parameter.
2684:
2685: The focussing is enhanced by the $-\theta^2/2$ term in Raychaudhuri's
2686: equation, (\ref{eq-ray}), which contributes to focussing whenever
2687: $\theta$ fluctuates about zero. Because of these effects, the
2688: light-sheets considered by Lowe (1999) will not remain on the horizon,
2689: but will collapse into the black hole. New families of light rays
2690: continually move inside to generate the event horizon. It is possible
2691: to transport unlimited entropy through the black hole horizon in this
2692: case, but not through any particular light-sheet.
2693:
2694:
2695: \subsection{Summary}
2696: \label{sec-presc}
2697:
2698:
2699:
2700: In any $D$-dimensional Lorentzian spacetime $M$, the covariant
2701: entropy bound can be stated as follows.
2702:
2703: {\em Let $A(B)$ be the area of an arbitrary $D-2$ dimensional spatial
2704: surface $B$ (which need not be closed). A $D-1$ dimensional
2705: hypersurface $L$ is called a light-sheet of $B$ if $L$ is generated by
2706: light rays which begin at $B$, extend orthogonally away from $B$, and
2707: have non-positive expansion,
2708: \begin{equation}
2709: \theta\leq 0,
2710: \end{equation}
2711: everywhere on $L$. Let $S$ be the entropy on any light-sheet of $B$.
2712: Then}
2713: \begin{equation}
2714: S \leq {A(B)\over 4}.
2715: \label{eq-bb2}
2716: \end{equation}
2717:
2718:
2719: Let us restate the covariant entropy bound one more time, in a
2720: constructive form most suitable for applying and testing the bound, as
2721: we will in Sec.~\ref{sec-tests}.
2722:
2723: \begin{enumerate}
2724:
2725: \item{Pick any $D-2$ dimensional spatial surface $B$, and determine
2726: its area $A(B)$. There will be four families of light rays projecting
2727: orthogonally away from $B$: $F_1 \ldots F_4$.}
2728:
2729: \item{Usually additional information is available, such as the
2730: macroscopic spacetime metric everywhere or in a neighborhood of
2731: $B$.\footnote{The case where no such information is presumed seems too
2732: general to be practally testable; see the end of Sec.~\ref{sec-es}.}
2733: Then the expansion $\theta$ of the orthogonal light rays can be
2734: calculated for each family. Of the four families, at least two will
2735: not expand ($\theta\leq 0$). Determine which.}
2736:
2737: \item{Pick one of the non-expanding families, $F_j$. Follow each
2738: light ray no further than to a caustic, a place where it intersects
2739: with neighboring light rays. The light rays form a $D-1$ dimensional
2740: null hypersurface, a light-sheet $L(B)$.}
2741:
2742: \item{Determine the entropy $S[L(B)]$ of matter on the light-sheet
2743: $L$, as described in Sec.~\ref{sec-els}.\footnote{In particular, one
2744: may wish to include in $S$ quantum states which do not all give rise
2745: to the same macroscopic spacetime geometry, keeping fixed only the
2746: intrinsic geometry of $B$. In this case, step 3 has to be repeated
2747: for each state or class of states with different geometry. Then
2748: $L(B)$ denotes the collection of all the different light-sheets
2749: emanating in the $j$-th direction.}}
2750:
2751: \item{The quantities $S[L(B)]$ and $A(B)$ can then be compared. The
2752: covariant entropy bound states that the entropy on the light-sheet
2753: will not exceed a quarter of the area: $S[L(B)] \leq \frac{A(B)}{4}$.
2754: This must hold for any surface $B$, and it applies to each
2755: non-expanding null direction, $F_j$, separately.}
2756:
2757: \end{enumerate}
2758:
2759: The first three steps can be carried out most systematically by using
2760: geometric tools which will be introduced at the beginning of
2761: Sec.~\ref{sec-ray}. In simple geometries, however, they often require
2762: little more than inspection of the metric.
2763:
2764: The light-sheet construction is well-defined in the limit where
2765: geometry can be described classically. It is conjectured to be valid
2766: for all physically realistic matter systems. In the absence of a
2767: fundamental theory with definite matter content, the energy conditions
2768: given in Sec.~\ref{sec-econds} approximately delineate the boundaries
2769: of an enormous arena of spacetimes and matter systems, in which the
2770: covariant entropy bound implies falsifiable, highly non-trivial
2771: limitations on information content.
2772:
2773: In particular, the bound is predictive and can be tested by
2774: observation, in the sense that the entropy and geometry of real matter
2775: systems can be determined (or, as in the case of large cosmological
2776: regions, at least estimated) from experimental measurements.
2777:
2778:
2779:
2780:
2781: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2782: \section{The dynamics of light-sheets}
2783: \label{sec-lsd}
2784: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2785:
2786: Entropy requires energy. In Sec.~\ref{sec-qftwrong}, this notion gave
2787: us some insight into a mechanism underlying the spherical entropy
2788: bound. Let us briefly repeat the idea. When one tries to excite too
2789: many degrees of freedom in a spherical region of fixed boundary area
2790: $A$, the region becomes very massive and eventually forms a black hole
2791: of area no larger than $A$. Because of the second law of
2792: thermodynamics, this collapse must set in before the entropy exceeds
2793: $A/4$. Of course, it can be difficult to verify this quantitatively
2794: for a specific system; one would have to know its detailed properties
2795: and gravitational back-reaction.
2796:
2797: In this section, we identify a related mechanism underlying the
2798: covariant entropy bound. Entropy costs energy, energy focusses light,
2799: focussing leads to the formation of caustics, and caustics prevent
2800: light-sheets from going on forever. As before, the critical link in
2801: this argument is the relation between entropy and energy.
2802: Quantitatively, it depends on the details of specific matter systems
2803: and cannot be calculated in general. Indeed, this is one of the
2804: puzzles that make the generality of the covariant entropy bound so
2805: striking.
2806:
2807: In many situations, however, entropy can be approximated by a local
2808: flow of entropy density. With plausible assumptions on the relation
2809: between the entropy and energy density, which we review, Flanagan,
2810: Marolf, and Wald (2000) proved the covariant entropy bound.
2811:
2812: We also present the spacelike projection theorem, which identifies
2813: conditions under which the covariant bound implies a spacelike bound
2814: (Bousso, 1999a).
2815:
2816:
2817: \subsection{Raychaudhuri's equation and the focussing theorem}
2818: \label{sec-ray}
2819:
2820: A family of light rays, such as the ones generating a light-sheet, is
2821: locally characterized by its expansion, shear, and twist, which are
2822: defined as follows.
2823:
2824: Let $B$ be a surface of $D-2$ spatial dimensions, parametrized by
2825: coordinates $x^\alpha$, $\alpha=1,\ldots,D-2$. Pick one of the four
2826: families of light rays $F_1,\ldots,F_4$ that emanate from $B$ into the
2827: past and future directions to either side of $B$
2828: (Fig.~\ref{fig-spheresheets}). Each light ray satisfies the equation
2829: for geodesics (Appendix):
2830: \begin{equation}
2831: \frac{dk^a}{d\lambda} + \Gamma^a_{~bc} k^b k^c =0,
2832: \label{eq-geodesic}
2833: \end{equation}
2834: where $\lambda$ is an affine parameter. The tangent vector $k^a$ is
2835: defined by
2836: \begin{equation}
2837: k^a = \frac{dx^a}{d\lambda}
2838: \end{equation}
2839: and satisfies the null condition $k^a k_a=0$. The light rays generate
2840: a null hypersurface $L$ parametrized by coordinates
2841: $(x^\alpha,\lambda)$. This can be rephrased as follows. In a
2842: neighborhood of $B$, each point on $L$ is unambiguously defined by the
2843: light ray on which it lies ($x^\alpha$) and the affine distance from
2844: $B$ ($\lambda$).
2845:
2846: Let $l^a$ be the null vector field on $B$ that is orthogonal to $B$
2847: and satisfies $k^a l_a=-2$. (This means that $l^a$ has the same time
2848: direction as $k^a$ and is tangent to the orthogonal light rays
2849: constructed on the other side of $B$.) The induced $D-2$ dimensional
2850: metric on the surface $B$ is given by
2851: \begin{equation}
2852: h_{ab} = g_{ab} + \frac{1}{2} \left(k_a l_b + k_b l_a\right).
2853: \label{eq-induce}
2854: \end{equation}
2855: In a similar manner, an induced metric can be found for all other
2856: spatial cross-sections of $L$.
2857:
2858: The {\em null extrinsic curvature},
2859: \begin{equation}
2860: B_{ab} = h^c_{~a} h^d_{~b} \nabla_c k_d,
2861: \label{eq-bab}
2862: \end{equation}
2863: contains information about the {\em expansion}, $\theta$, {\em shear},
2864: $\sigma_{ab}$, and {\em twist}, $\omega_{ab}$, of the family of
2865: light rays, $L$:
2866: \begin{eqnarray}
2867: \label{eq-thetadef}
2868: \theta &=& h^{ab} B_{ab}, \\
2869: \sigma_{ab} &=& \frac{1}{2} \left( B_{ab}+B_{ba} \right) -
2870: \frac{1}{D-2}\theta h_{ab}, \\
2871: \omega_{ab} &=& \frac{1}{2} \left( B_{ab}-B_{ba} \right).
2872: \end{eqnarray}
2873: Note that all of these quantities are functions of
2874: $(x^\alpha,\lambda)$.
2875:
2876: At this point, one can inspect the initial values of $\theta$ on $B$.
2877: Where they are positive, one must discard $L$ and choose a different
2878: null direction for the construction of a light-sheet.
2879:
2880: The Raychaudhuri equation describes the change of the expansion along
2881: the light rays:
2882: \begin{equation}
2883: \frac{d\theta}{d\lambda} = -\frac{1}{D-2} \theta^2 -
2884: \sigma_{ab}\sigma^{ab} + \omega_{ab}\omega^{ab} - 8\pi T_{ab} k^a k^b.
2885: \label{eq-ray}
2886: \end{equation}
2887: For a surface-orthogonal family of light rays, such as $L$, the twist
2888: vanishes (Wald, 1984). The final term, $-T_{ab} k^a k^b$, will be
2889: non-positive if the null energy condition is satisfied by matter,
2890: which we assume (Sec.~\ref{sec-econds}). Then the right hand side of
2891: the Raychaudhuri equation is manifestly non-positive. It follows that
2892: the expansion never increases.
2893:
2894: By solving the differential inequality
2895: \begin{equation}
2896: \frac{d\theta}{d\lambda} \leq -\frac{1}{D-2} \theta^2,
2897: \label{eq-rayn}
2898: \end{equation}
2899: one arrives at the {\em focussing theorem}:\footnote{In the context of
2900: the AdS/CFT correspondence (Sec.~\ref{sec-ads}), the role of focussing
2901: theorem in the construction of light-sheets has been related to the
2902: c-theorem (Balasubramanian, Gimon, and Minic, 2000; Sahakian,
2903: 2000a,b).} If the expansion of a family of light rays takes the
2904: negative value $\theta_1$ at any point $\lambda_1$, then $\theta$ will
2905: diverge to $-\infty$ at some affine parameter $\lambda_2\leq\lambda_1
2906: +{D-2\over|\theta_1|}$.
2907:
2908: The divergence of $\theta$ indicates that the cross-sectional area is
2909: locally vanishing, as can be seen from Eq.~(\ref{eq-tda}). As
2910: discussed in Sec.~\ref{sec-kin3}, this is a caustic point, at which
2911: infinitesimally neighboring light rays intersect.
2912:
2913: By construction, the expansion on light-sheets is zero or negative.
2914: If it is zero, the focussing theorem does not apply. For example,
2915: suppose that $B$ is a portion of the $xy$ plane in Minkowski space:
2916: $z=t=0,~~x^2+y^2\leq 1$. Then each light-sheet is infinitely large,
2917: with everywhere vanishing expansion: $z=\pm t,~~x^2+y^2\leq 1$.
2918: However, this is correct only if the spacetime is exactly Minkowski,
2919: with no matter or gravitational waves. In this case the light-sheets
2920: contain no entropy in any case, so their infinite size leads to no
2921: difficulties with the covariant entropy bound.
2922:
2923: If a light-sheet encounters any matter (or more precisely, if $T_{ab}
2924: k^a k^b >0$ anywhere on the light-sheet), then the light rays will be
2925: focussed according to Eq.~(\ref{eq-ray}). Then the focussing theorem
2926: applies, and it follows that the light rays will eventually form
2927: caustics, forcing the light-sheet to end. This will happen even if no
2928: further energy is encountered by the light rays, though it will occur
2929: sooner if there is additional matter.
2930:
2931: If we accept that entropy requires energy, we thus see at a
2932: qualitative level that entropy causes light rays to focus. Thus, the
2933: presence of entropy hastens the termination of light-sheets.
2934: Quantitatively, it appears to do so at a sufficient rate to protect
2935: the covariant entropy bound, but slowly enough to allow saturation of
2936: the bound. This is seen in many examples, including those studied in
2937: Sec.~\ref{sec-tests}. The reason for this quantitative behavior is
2938: not yet fundamentally understood. (This just reformulates, in terms
2939: of light-sheet dynamics, the central puzzle laid out in the
2940: introduction and reiterated in Sec.~\ref{sec-hp}.)
2941:
2942:
2943:
2944:
2945: \subsection{Sufficient conditions for the covariant entropy bound}
2946: \label{sec-fmw}
2947:
2948: Flanagan, Marolf, and Wald (2000; henceforth in this section, FMW)
2949: showed that the covariant entropy bound is always satisfied if certain
2950: assumptions about the relation between entropy density and energy
2951: density are made. In fact, they proved the bound under either one of
2952: two sets of assumptions. We will state these assumptions and discuss
2953: their plausibility and physical significance. We will not reproduce
2954: the two proofs here.
2955:
2956: The first set of conditions are no easier to verify, in any given
2957: spacetime, than the covariant entropy bound itself. Light-sheets
2958: have to be constructed, their endpoints found, and entropy can be
2959: defined only by an analysis of modes. The first set of conditions
2960: should therefore be regarded as an interesting reformulation of the
2961: covariant entropy bound, which may shed some light on its relation to
2962: the Bekenstein bound, Eq.~(\ref{eq-bekbound}).
2963:
2964: The second set of conditions involves relations between locally
2965: defined energy and entropy densities only. As long as the entropy
2966: content of a spacetime admits a fluid approximation, one can easily
2967: check whether these conditions hold. In such spacetimes, the second
2968: FMW theorem obviates the need to construct all light-sheets and verify
2969: the bound for each one.
2970:
2971: Neither set of conditions is implied by any fundamental law of
2972: physics. The conditions do not apply to some physically realistic
2973: systems (which nevertheless obey the covariant entropy bound).
2974: Furthermore, they do not permit macroscopic variations of spacetime,
2975: precluding a verification of the bound in its strongest sense
2976: (Sec.~\ref{sec-es}).
2977:
2978: Thus, as FMW point out, the two theorems do not constitute a
2979: fundamental explanation of the covariant entropy bound. By
2980: eliminating a large class of potential counter-examples, they do
2981: provide important evidence for the validity of the covariant entropy
2982: bound. The second set can significantly shortcut the verification of
2983: the bound in cosmological spacetimes. Moreover, the broad validity of
2984: the FMW hypotheses may itself betray an aspect of an underlying
2985: theory.
2986:
2987:
2988: \subsubsection{The first FMW theorem}
2989: \label{sec-fmw1}
2990:
2991: The first set of assumptions is
2992:
2993: \begin{itemize}
2994:
2995: \item{Associated with each light-sheet $L$ in spacetime there is an
2996: entropy flux 4-vector $s^a_L$ whose integral over $L$ is the
2997: entropy flux through $L$.}
2998:
2999: \item{The inequality
3000: \begin{equation}
3001: \left|s_L^a k^a\right|\leq
3002: {\pi}(\lambda_\infty-\lambda) T_{ab} k^a k^b
3003: \label{eq-fmw1}
3004: \end{equation}
3005: holds everywhere on $L$. Here $\lambda_\infty$ is the value of the
3006: affine parameter at the endpoint of the light-sheet.}
3007:
3008: \end{itemize}
3009:
3010: The entropy flux vector $s_L^a$ is defined non-locally by demanding
3011: that only modes that are fully captured on $L$ contribute to the
3012: entropy on $L$. Modes that are partially contained on $L$ do not
3013: contribute. This convention recognizes that entropy is a non-local
3014: phenomenon. It is particularly useful when light-sheets penetrate a
3015: thermodynamic system only partially, as discussed in
3016: Sec.~\ref{sec-els}.
3017:
3018: This set of assumptions can be viewed as a kind of ``light ray
3019: equivalent'' of Bekenstein's bound, Eq.~(\ref{eq-bekbound}), with the
3020: affine parameter playing the role of the circumferential radius.
3021: However, it is not clear whether one should expect this condition to
3022: be satisfied in regions of dominant gravity. Indeed, it does not
3023: apply to some weakly gravitating systems (Sec.~\ref{sec-gia}).
3024:
3025: FMW were actually able to prove a stronger form of the covariant
3026: entropy bound from the above hypotheses. Namely, suppose that the
3027: light-sheet of a surface of area $A$ is constructed, but the
3028: light rays are not followed all the way to the caustics. The
3029: resulting light-sheet is, in a sense, shorter than necessary, and one
3030: would expect that the entropy on it, $S$, will not saturate the bound.
3031: The final area spanned by the light rays, $A'$, will be less than $A$
3032: but non-zero (Fig.~\ref{fig-inside}b).
3033:
3034: FMW showed, with the above assumptions, that a tightened bound results
3035: in this case:
3036: \begin{equation}
3037: S\leq\frac{A-A'}{4}.
3038: \label{eq-strongbound}
3039: \end{equation}
3040: Note that this expression behaves correctly in the limit where the
3041: light-sheet is maximized [$A'\to 0$; one recovers Eq.~(\ref{eq-bb2})]
3042: and minimized ($A'\to A$; there is no light-sheet and hence no
3043: entropy).
3044:
3045: The strengthened form, Eq.~(\ref{eq-strongbound}), of the covariant
3046: entropy bound, Eq.~(\ref{eq-bb2}), appears to have broad, but not
3047: completely general validity (Sec.~\ref{sec-gia}).
3048:
3049:
3050: \subsubsection{The second FMW theorem}
3051: \label{sec-fmw2}
3052:
3053: Through a rather non-trivial proof, FMW showed that the covariant
3054: entropy bound can also be derived from a second set of assumptions,
3055: namely:
3056:
3057: \begin{itemize}
3058:
3059: \item{The entropy content of spacetime is well approximated by an
3060: absolute entropy flux vector field $s^a$.}
3061:
3062: \item{For any null vector $k^a$, the inequalities
3063: \begin{eqnarray}
3064: (s_a k^a)^2 & \leq & {1\over 16\pi} T_{ab} k^a k^b, \\
3065: \left| k^a k^b \nabla_a s_b \right| & \leq &
3066: {\pi\over 4} T_{ab} k^a k^b
3067: \end{eqnarray}
3068: hold at everywhere in the spacetime.}
3069:
3070: \end{itemize}
3071:
3072: These assumptions are satisfied by a wide range of matter systems,
3073: including Bose and Fermi gases below the Planck temperature. It is
3074: straightforward to check that all of the adiabatically evolving
3075: cosmologies investigated in Sec.~\ref{sec-cosmo} conform to the above
3076: conditions. Thus, the second FMW theorem rules out an enormous class
3077: of potential counterexamples, obviating the hard work of calculating
3078: light-sheets. (We will find light-sheets in simple cosmologies
3079: anyway, both in order to gain intuition about how the light-sheet
3080: formulation works in cosmology, and also because this analysis is
3081: needed for the discussion of holographic screens in
3082: Sec.~\ref{sec-screens}.)
3083:
3084: Generally speaking, the notion of an entropy flux assumes that entropy
3085: can be treated as a kind of local fluid. This is often a good
3086: approximation, but it ignores the non-local character of entropy and
3087: does not hold at a fundamental level.
3088:
3089:
3090: \subsection{Relation to other bounds and to the GSL}
3091: \label{sec-rob}
3092:
3093: \subsubsection{Spacelike projection theorem}
3094: \label{sec-spt}
3095:
3096: We have seen in Sec.~\ref{sec-fail} that the spacelike entropy bound
3097: does not hold in general. Taking the covariant entropy bound as a
3098: general starting point, one may derive other, more limited
3099: formulations, whose regimes of validity are defined by the assumptions
3100: entering the derivation. Here we use the light-sheet formulation to
3101: recover the spacelike entropy bound, Eq.~(\ref{eq-seb}), along with
3102: precise conditions under which it holds. By imposing further
3103: conditions, even more specialized bounds can be obtained; an example
3104: valid for certain regions in cosmological spacetimes is discussed in
3105: Sec.~\ref{sec-cosmocor} below.
3106:
3107: %%***************************************************************
3108: \begin{figure}[h] \centering
3109: \includegraphics[width=7cm]{fig-spt}
3110: \caption{Spacelike projection theorem. If the surface $B$ has a
3111: complete future directed light-sheet $L$, then the spacelike entropy
3112: bound applies to any spatial region $V$ enclosed by $B$.}
3113: \label{fig-spt}
3114: \end{figure}
3115: %%***************************************************************
3116: {\sl Spacelike projection theorem} (Bousso, 1999a). {\em Let $B$ be a
3117: closed surface. Assume that $B$ permits at least one future directed
3118: light-sheet $L$. Moreover, assume that $L$ is {\em complete}, i.e.,
3119: $B$ is its only boundary (Fig.~\ref{fig-spt}). Let $S(V)$ be the
3120: entropy in a spatial region $V$ enclosed by $B$ on the same side as
3121: $L$. Then}
3122: \begin{equation}
3123: S(V) \leq S(L) \leq \frac{A}{4}.
3124: \label{eq-spt}
3125: \end{equation}
3126: {\sl Proof.} Independently of the choice of $V$ (i.e., the choice of a
3127: time coordinate), all matter present on $V$ will pass through $L$.
3128: The second law of thermodynamics implies the first inequality, the
3129: covariant entropy bound implies the second.
3130:
3131: What is the physical significance of the assumptions made in the
3132: theorem? Suppose that the region enclosed by $B$ is weakly
3133: gravitating. Then we may expect that all assumptions of the theorem
3134: are satisfied. Namely, if $B$ did not have a future directed
3135: light-sheet, it would be anti-trapped---a sign of strong gravity. If
3136: $L$ had other boundaries, this would indicate the presence of a future
3137: singularity less than one light-crossing time from $B$---again, a sign
3138: of strong gravity.
3139:
3140: Thus, for a closed, weakly gravitating, smooth surface $B$ we may
3141: expect the spacelike entropy bound to be valid. In particular, the
3142: spherical entropy bound, deemed necessary for the validity of the GSL
3143: in the Susskind process, follows from the covariant bound. This can
3144: be see by inspecting the assumptions in (Sec.~\ref{sec-spheb}), which
3145: guarantee that the conditions of the spacelike projection theorem are
3146: satisfied.
3147:
3148:
3149: \subsubsection{Generalized second law and Bekenstein bound}
3150: \label{sec-gia}
3151:
3152: In fact, FMW showed that the covariant bound implies the GSL directly
3153: for any process of black hole formation, such as the Susskind process
3154: (Sec.~\ref{sec-susskind}).
3155:
3156: Consider a surface $B$ of area $A$ on the event horizon of a black
3157: hole. The past directed ingoing light rays will have non-positive
3158: expansion; they generate a light-sheet. The light-sheet contains all
3159: the matter that formed the black hole. The covariant bound implies
3160: that $S_{\rm matter} \leq A(B)/4 = S_{\rm BH}$. Hence, the
3161: generalized second law is satisfied for the process in which a black
3162: hole is newly formed from matter.
3163:
3164: Next, let us consider a more general process, the absorption of a
3165: matter system by an existing black hole. This includes the Geroch
3166: process (Sec.~\ref{sec-geroch}). Does the covariant bound
3167: also imply the GSL in this case?
3168:
3169: Consider a surface $B$ on the event horizon after the matter system,
3170: of entropy $S_{\rm matter}$, has fallen in, and follow the
3171: past-ingoing light-rays again. The light-rays are focussed by the
3172: energy momentum of the matter. ``After'' proceeding through the
3173: matter system, let us terminate the light-sheet. Thus the light-sheet
3174: contains precisely the entropy $S_{\rm matter}$. The rays will span a
3175: final area $A'$ (which is really the initial area of the event horizon
3176: before the matter fell in).
3177:
3178: According to an outside observer, the Bekenstein entropy of the black
3179: hole has increased by $(A-A')/4$, while the matter entropy $S_{\rm
3180: matter}$ has been lost. According to the ``strengthened form'' of the
3181: covariant entropy bound considered by FMW, Eq.~(\ref{eq-strongbound}),
3182: the total entropy has not decreased. The original covariant bound,
3183: Eq.~(\ref{eq-bb2}), does not by itself imply the generalized second
3184: law of thermodynamics, Eq.~(\ref{eq-gsl}), in this process.
3185:
3186: Eq.~(\ref{eq-strongbound}) can also be used to derive a version of
3187: Bekenstein's bound, Eq.~(\ref{eq-bekbound})---though, unfortunately, a
3188: version that is too strong. Consider the light-sheet of an
3189: approximately flat surface of area $A$, bounding one side of a
3190: rectangular thermodynamic system. With suitable time-slicing, the
3191: surface can be chosen to have vanishing null expansion, $\theta$.
3192:
3193: With assumptions on the average energy density and the equation of
3194: state, Raychaudhuri's equation can be used to estimate the final area
3195: $A'$ of the light-sheet where it exits the opposite side of the matter
3196: system. The strengthened form of the covariant entropy bound,
3197: Eq.~(\ref{eq-strongbound}) then implies the bound given in
3198: Eq.~(\ref{eq-bekbound-wrong}). However, for very flat systems this
3199: bound can be violated (Sec.~\ref{sec-unruh})!
3200:
3201: Hence, (\ref{eq-strongbound}) cannot hold in the same generality that
3202: is claimed for the original covariant entropy bound,
3203: Eq.~(\ref{eq-bb2}).\footnote{It follows that the first FMW hypotheses
3204: do not hold in general. An earlier counterexample to
3205: Eq.~(\ref{eq-strongbound}), and hence to these hypotheses, was given
3206: by Guedens (2000).} However, the range of validity of
3207: Eq.~(\ref{eq-strongbound}) does appear to be extremely broad. In view
3208: of the significance of its implications, it will be important to
3209: better understand its scope.
3210:
3211: We conclude that the covariant entropy bound implies the spherical
3212: bound in its regime of validity, defines a range of validity for the
3213: spacelike bound, and implies the GSL for black hole formation
3214: processes. The strengthened form of the covariant bound given by FMW,
3215: Eq.~(\ref{eq-strongbound}), implies the GSL for absorption processes
3216: and, under suitable assumptions, yields Bekenstein's bound [though in
3217: a form that demonstates that Eq.~(\ref{eq-strongbound}) cannot be
3218: universally valid].
3219:
3220: The result of this section suggest that the holographic principle
3221: (Sec.~\ref{sec-hp}) will take a primary role in the complex of ideas
3222: we have surveyed. It may come to be viewed as the logical origin not
3223: only of the covariant entropy bound, but also of more particular laws
3224: that hold under suitable conditions, such as the spherical entropy
3225: bound, Bekenstein's bound, and the generalized second law of
3226: thermodynamics.
3227:
3228:
3229:
3230:
3231: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3232: \section{Applications and examples}
3233: \label{sec-tests}
3234: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3235:
3236:
3237: In this section, the covariant entropy bound is applied to a variety
3238: of matter systems and spacetimes. We demonstrate how the light-sheet
3239: formulation evades the various difficulties encountered by the
3240: spacelike entropy bound (Sec.~\ref{sec-fail}).
3241:
3242: We apply the bound to cosmology and verify explicitly that it is
3243: satisfied in a wide class of universes. No violations are found
3244: during the gravitational collapse of a star, a shell, or the whole
3245: universe, though the bound can be saturated.
3246:
3247:
3248:
3249: \subsection{Cosmology}
3250: \label{sec-cosmo}
3251:
3252: \subsubsection{FRW metric and entropy density}
3253:
3254: Friedmann-Robertson-Walker (FRW) metrics describe homogeneous,
3255: isotropic universes, including, to a good degree of approximation, the
3256: portion we have seen of our own universe. Often the metric is
3257: expressed in the form
3258: \begin{equation}
3259: ds^2 = -dt^2 + a^2(t) \left( \frac{dr^2}{1-kr^2} + r^2 d\Omega^2
3260: \right).
3261: \label{eq-FRW1}
3262: \end{equation}
3263:
3264: We will find it more useful to use the conformal time $\eta$ and the
3265: comoving coordinate $\chi$:
3266: \begin{equation}
3267: d\eta = {dt\over a(t)},~~~d\chi = {dr\over\sqrt{1-kr^2}}.
3268: \end{equation}
3269: In these coordinates the FRW metric takes the form
3270: \begin{equation}
3271: ds^2 = a^2(\eta) \left[ -d\eta^2 + d\chi^2 + f^2(\chi) d\Omega^2
3272: \right].
3273: \label{eq-FRW2}
3274: \end{equation}
3275: Here $k = -1$, $0$, $1$ and $f(\chi) = \sinh \chi$, $\chi$, $\sin
3276: \chi$ correspond to open, flat, and closed universes respectively.
3277: Relevant Penrose diagrams are shown in Figs.~\ref{fig-flatfrw} and
3278: \ref{fig-clos}a.
3279: %%***************************************************************
3280: \begin{figure}[h] \centering
3281: \includegraphics[width=8.5cm]{fig-clos}
3282: \caption{(a) Penrose diagram for a closed FRW universe filled with
3283: pressureless dust. The three-sphere time slices are represented by
3284: horizontal lines (not shown). Two apparent horizons divide the
3285: diagram into four wedge domains: normal spheres are found near the
3286: poles, trapped (anti-trapped) spheres near the big bang (big crunch).
3287: (b) The construction of a global holographic screen
3288: (Sec.~\ref{sec-screens}) proceeds by foliating the spacetime into a
3289: stack of light cones. The information on each slice can be stored on
3290: the maximal sphere, which lies on the apparent horizon.}
3291: \label{fig-clos}
3292: \end{figure}
3293: %%***************************************************************
3294:
3295: In cosmology, the entropy is usually described by an entropy density
3296: $\sigma$, the entropy per physical volume:
3297: \begin{equation}
3298: S(V) = \int_V d^3x \sqrt{h}\, \sigma.
3299: \end{equation}
3300: For FRW universes, $\sigma$ depends only on time. We will assume, for
3301: now, that the universe evolves adiabatically. Thus, the physical
3302: entropy density is diluted by cosmological expansion:
3303: \begin{equation}
3304: \sigma(\eta) = {s\over a(\eta)^3}.
3305: \end{equation}
3306: The comoving entropy density $s$ is constant in space and time.
3307:
3308:
3309: \subsubsection{Expansion and apparent horizons}
3310: \label{sec-ah}
3311:
3312: Let us verify that the covariant entropy bound is satisfied for each
3313: light-sheet of any spherical surface $A$. The first step is to
3314: identify the light-sheet directions. We must classify each sphere as
3315: trapped, normal, or anti-trapped (Sec.~\ref{sec-kin2}). Let us
3316: therefore compute the initial expansion of the four families of
3317: light rays orthogonal to an arbitrary sphere characterized by some
3318: value of $(\eta,\chi)$.
3319:
3320: We take the affine parameter to agree locally with $\pm 2\eta$ and use
3321: Eq.~(\ref{eq-tda}). Differentiation with respect to $\eta$ ($\chi$)
3322: is denoted by a dot (prime). Instead of labelling the families
3323: $F_1,\ldots,F_4$, it will be more convenient to use the notation
3324: $(\pm\pm)$, where the first sign refers to the time ($\eta$) direction
3325: of the light rays and the second sign denotes whether they are
3326: directed at larger or smaller values of $\chi$.
3327:
3328: For the future directed families one finds
3329: \begin{equation}
3330: \theta_{+\pm} = {\dot{a}\over a} \pm {f'\over f}.
3331: \label{eq-thp}
3332: \end{equation}
3333: The expansion of the past directed families is given by
3334: \begin{equation}
3335: \theta_{-\pm} = - {\dot{a}\over a} \pm {f'\over f}.
3336: \label{eq-thm}
3337: \end{equation}
3338:
3339: Note that the first term in Eq.~(\ref{eq-thp}) is positive when the
3340: universe expands and negative if it contracts. The term diverges when
3341: $a\to 0$, i.e., near singularities. The second term is given by
3342: $\cot\chi$ ($1/\chi$; $\coth\chi$) for a closed (flat; open) universe.
3343: It diverges at the origin ($\chi\to 0$), and for a closed universe it
3344: also diverges at the opposite pole ($\chi\to\pi$).
3345:
3346: The signs of the four quantities $\theta_{\pm\pm}$ depend on the
3347: relative strength of the two terms. The quickest way to classify
3348: surfaces is to identify marginal spheres, where the two terms are of
3349: equal magnitude.
3350:
3351: The {\em apparent horizon\/} is defined geometrically as a sphere at
3352: which at least one pair of orthogonal null congruences have zero
3353: expansion. It satisfies the condition
3354: \begin{equation}
3355: {\dot{a}\over a} = \pm {f'\over f},
3356: \label{eq-ahcond}
3357: \end{equation}
3358: which can be used to identify its location $\chi_{\rm AH}(\eta)$ as a
3359: function of time. There is one solution for open and flat universes.
3360: For a closed universe, there are generally two solutions, which are
3361: symmetric about the equator [$\chi_{{\rm AH}'}(\eta) =
3362: \pi-\chi_{\rm AH} (\eta)$].
3363:
3364: The proper area of the apparent horizon is given by
3365: \begin{equation}
3366: A_{\rm AH}(\eta) = 4 \pi a(\eta)^2 f[\chi_{\rm AH}(\eta)]^2 =
3367: {4\pi a^2\over \left({\dot a\over a}\right)^2+k}.
3368: \end{equation}
3369: Using Friedmann's equation,
3370: \begin{equation}
3371: {\dot a^2\over a^2} = \frac{8\pi \rho a^2}{3} - k,
3372: \end{equation}
3373: one finds
3374: \begin{equation}
3375: A_{\rm AH}(\eta) = {3\over 2\rho(\eta)},
3376: \label{eq-ah-rho}
3377: \end{equation}
3378: where $\rho$ is the energy density of matter.
3379:
3380: At any time $\eta$, the spheres that are smaller than the apparent
3381: horizon,
3382: \begin{equation}
3383: A< A_{\rm AH},
3384: \end{equation}
3385: are normal. (See the end of Sec.~\ref{sec-kin2} for the definitions
3386: of normal, trapped, and anti-trapped surfaces.) Because the second
3387: term $f'/f$ dominates in the expressions for the expansion, the
3388: cosmological evolution has no effect on the light-sheet directions.
3389: The two light-sheets will be a past and a future directed family
3390: going to the same spatial side. In a flat or open universe, they will
3391: be directed towards $\chi=0$ (Fig.~\ref{fig-flatfrw}). In a closed
3392: universe, the light-sheets of a normal sphere will be directed towards
3393: the nearest pole, $\chi=0$ or $\chi=\pi$ (Fig.~\ref{fig-clos}a).
3394:
3395: For spheres greater than the apparent horizon
3396: \begin{equation}
3397: A> A_{\rm AH},
3398: \end{equation}
3399: the cosmological term $\dot a/a$ dominates in the expressions for the
3400: expansion. Then there are two cases. Suppose that $\dot a>0$, i.e.,
3401: the universe is expanding. Then the spheres are anti-trapped. Both
3402: light-sheets are past directed, as indicated by a wedge opening to the
3403: bottom in the Penrose diagram. If $A>A_{\rm AH}$ and $\dot a<0$, then
3404: both future directed families will have negative expansion. This case
3405: describes trapped spheres in a collapsing universe. They are denoted
3406: by a wedge opening to the top (Fig.~\ref{fig-clos}a).
3407:
3408:
3409: \subsubsection{Light-sheets vs.~spatial volumes}
3410: \label{sec-lvs}
3411:
3412: We have now classified all spherical surfaces in all FRW universes
3413: according to their light-sheet directions. Before proceeding to a
3414: detailed calculation of the entropy contained on the light-sheets, we
3415: note that the violations of the spacelike entropy bound identified in
3416: Secs.~\ref{sec-fail1} and \ref{sec-fail2} do not apply to the
3417: covariant bound.
3418:
3419: The area of a sphere at $\eta_0,\chi_0$ is given by
3420: \begin{equation}
3421: A(\eta_0,\chi_0) = 4\pi a(\eta_0)^2 f(\chi_0)^2.
3422: \label{eq-afrw}
3423: \end{equation}
3424: To remind ourselves that the spacelike entropy bound fails in
3425: cosmology, let us begin by comparing this area to the entropy enclosed
3426: in the spatial volume $V(\chi_0)$ defined by $\chi\leq\chi_0$ at equal
3427: time $\eta=\eta_0$. With our assumption of adiabaticity, this depends
3428: only on $\chi_0$:
3429: \begin{equation}
3430: S[V(\chi_0)] = 4\pi s \int_0^{\chi_0} d\chi f(\chi)^2.
3431: \end{equation}
3432:
3433: For a flat universe [$f(\chi) = \chi$], the area grows like $\chi_0^2$
3434: but the entropy grows like $\chi_0^3$. Thus, $S[V(\chi_0)]>A$ for
3435: sufficiently large $\chi_0$. (This was pointed out earlier in
3436: Sec.~\ref{sec-fail2}.) For a closed universe ($f(\chi) =\sin\chi$),
3437: $\chi$ ranges only from $0$ to $\pi$. $S[V(\chi_0)]$ is monotonically
3438: increasing in this range, but $A\to 0$ for $\chi_0\to\pi$. Again, one
3439: has $S[V(\chi_0)]>A$. This is a special case of the problem discussed
3440: in Sec.~\ref{sec-fail1}.
3441:
3442: Why don't light-sheets run into the same difficulties? Consider first
3443: a large sphere in a flat universe (Fig.~\ref{fig-flatfrw}). The
3444: future-ingoing light rays cover the same amount of entropy as the
3445: enclosed spatial volume. However, for spheres greater than the
3446: apparent horizon, the future-ingoing light rays are expanding and
3447: hence do not form a light-sheet. Only past directed light-sheets are
3448: permitted. The past-ingoing light rays, for example, will proceed
3449: towards the origin. However, if the sphere is greater than the
3450: particle horizon ($\chi>\eta$), they will terminate at the big bang
3451: ($\eta=0$) and will not get all the way to $\chi=0$. Instead of a
3452: comoving ball $0\leq\chi'\leq\chi$, they will sweep out only a shell
3453: of width $\eta$: $\chi-\eta\leq\chi'\leq\chi$. Thus the entropy to
3454: area ratio does not diverge for large $\chi$, but approaches a
3455: constant value.
3456:
3457: Small spheres ($A<A_{\rm AH}$) in a closed universe
3458: (Fig.~\ref{fig-clos}a) permit only light-sheets that are directed to
3459: the smaller enclosed region. The light rays directed towards the
3460: larger portion of the universe will be initially expanding and hence
3461: do not form light-sheets. Both in the flat and the closed case, we
3462: see that the $\theta\leq 0$ contraction condition is of crucial
3463: importance.
3464:
3465:
3466: \subsubsection{Solutions with fixed equation of state}
3467:
3468: The matter content of FRW universes is most generally described by a
3469: perfect fluid, with stress tensor
3470: \begin{equation}
3471: T^a_{~b} = \mbox{diag}(-\rho,p,p,p).
3472: \end{equation}
3473: Let us assume that the pressure $p$ and energy density $\rho$ are
3474: related by a fixed equation of state
3475: \begin{equation}
3476: p = w \rho.
3477: \end{equation}
3478: Our universe and many other more general solutions can be pieced
3479: together from solutions obtained via this ansatz, because the
3480: transitions between different effective equations of state are very
3481: rapid.
3482:
3483: For most of its lifetime, our universe was dominated by pressureless
3484: dust and hence was characterized by $w=0$. The early universe was
3485: dominated by radiation, which is described by $w={1\over 3}$. A
3486: cosmological constant, which may have been present at very early
3487: times and perhaps again today, corresponds to $w=-1$.
3488:
3489: With this ansatz for the matter content and the FRW ansatz for the
3490: metric, Einstein's equation can be solved. This determines the scale
3491: factor in Eq.~(\ref{eq-FRW2}):
3492: \begin{equation}
3493: a(\eta) = a_0 \left[f({\eta\over q})\right]^q,
3494: \label{eq-a}
3495: \end{equation}
3496: where
3497: \begin{equation}
3498: q = {2\over 1+3w},
3499: \label{eq-q}
3500: \end{equation}
3501: and $f$ is the $\sin$ (the identity, $\sinh$) for a closed (flat,
3502: open) universe, as in Eq.~(\ref{eq-FRW2}). From Eq.~(\ref{eq-ahcond})
3503: it follows that an apparent horizon is located at
3504: \begin{equation}
3505: \chi_{\rm AH}(\eta) = {\eta\over q}
3506: \label{eq-ahq}
3507: \end{equation}
3508: in all cases. An additional mirror horizon lies at $\pi-{\eta\over
3509: q}$ in the closed case.
3510:
3511:
3512:
3513:
3514: Having established the light-sheet directions as a function of $t$ and
3515: $r$, we will now check whether the covariant entropy bound is
3516: satisfied on all light-sheets. The present treatment concentrates on
3517: flat and closed ($k=0,1$) universes with $w\geq 0$. However, we will
3518: quote results for $w<0$, i.e., negative pressure (Kaloper and Linde,
3519: 1999), which involves additional subtleties. We will also comment on
3520: the inflationary case ($w=-1$). We omit the open universes ($k=-1$)
3521: because they do not give rise to qualitatively new features (Fischler
3522: and Susskind, 1998). Bousso (1999a) discusses closed universes in
3523: detail. The main additional features beyond the flat case are covered
3524: in Secs.~\ref{sec-lvs} and \ref{sec-collapses}. We will comment on
3525: the inflationary case ($w=-1$) separately.
3526:
3527:
3528: \subsubsection{Flat universe}
3529: \label{sec-ff}
3530:
3531: Let us consider all possible light-sheets of all spherical areas
3532: ($0<\chi<\infty$) at the time $\eta$ in a flat FRW universe,
3533: \begin{equation}
3534: A(\eta,\chi) = 4\pi a(\eta)^2 \chi^2.
3535: \end{equation}
3536: If $\chi\leq\chi_{\rm AH}(\eta)$, the sphere is normal, and the
3537: light-sheet directions are ($+-$) and ($--$). If $\chi\geq\chi_{\rm
3538: AH}$, the sphere is anti-trapped, with light-sheets ($-+$) and ($--$)
3539: (Fig.~\ref{fig-flatfrw}).
3540:
3541: We begin with the future-ingoing ($+-$) light rays. They contract
3542: towards the origin and generate a conical light-sheet whose
3543: coordinates $(\chi',\eta')$ obey
3544: \begin{equation}
3545: \chi'+\eta' = \chi+\eta,
3546: \end{equation}
3547: This light-sheet contains the comoving entropy in the region
3548: $0\leq\chi'\leq\chi$, which is given by
3549: \begin{equation}
3550: S_{+-} = {4\pi\over 3} s \chi^3.
3551: \end{equation}
3552: The ratio of entropy to area,
3553: \begin{equation}
3554: {S_{+-}\over A} = {s\chi\over 3a(\eta)^2},
3555: \end{equation}
3556: is maximized by the outermost normal
3557: surface at any given time $\eta$, the sphere on the apparent horizon.
3558: Thus we obtain the bound
3559: \begin{equation}
3560: {S_{+-}\over A} \leq {s \chi_{\rm AH}(\eta)\over 3 a(\eta)^2}.
3561: \end{equation}
3562:
3563: The past-ingoing ($--$) light-sheet of any surface with $\chi<\eta$
3564: also reaches a caustic at $\chi=0$. If $\chi>\eta$, then the
3565: light-sheet is truncated instead by the big bang singularity at
3566: $\eta=0$. Then it will contain the comoving entropy in the region
3567: $\chi-\eta\leq\chi'\leq\chi$. The entropy to area ratio is given by
3568: \begin{equation}
3569: {S_{--}\over A} = {s\, \eta\over a(\eta)^2} \left(1-{\eta\over\chi}
3570: +{\eta^2\over 3\chi^2} \right).
3571: \end{equation}
3572: This ratio is maximized for large spheres ($\chi\to\infty$), yielding
3573: the bound
3574: \begin{equation}
3575: {S_{--}\over A} \leq {s\, \eta\over a(\eta)^2}
3576: \end{equation}
3577: for the ($--$) light-sheets at time $\eta$.
3578:
3579: Finally, we must consider the past-outgoing ($-+$) light-sheet of any
3580: surface with $\chi>\chi_{\rm AH}$. It is truncated by the big bang
3581: singularity and contains the entropy within
3582: $\chi\leq\chi'\leq\chi+\eta$. The ratio of the entropy to the area,
3583: \begin{equation}
3584: {S_{-+}\over A} = {s\, \eta\over a(\eta)^2} \left(1+{\eta\over\chi}
3585: +{\eta^2\over 3\chi^2} \right),
3586: \end{equation}
3587: is maximized for the smallest possible value of $\chi$, the apparent
3588: horizon. We find the bound
3589: \begin{equation}
3590: {S_{-+}\over A} \leq {s\, \eta\over a(\eta)^2} \left(1+{\eta\over
3591: \chi_{\rm AH}} +{\eta^2\over 3 \chi_{\rm AH}^2} \right).
3592: \end{equation}
3593:
3594: We now use the solution for fixed equation of state, setting $a_0=1$
3595: for convenience:
3596: \begin{equation}
3597: a(\eta) = \left({\eta\over q}\right)^q,~~~
3598: \chi_{\rm AH}(\eta)={\eta\over q}.
3599: \end{equation}
3600: Up to factors of order unity, the bounds for all three types of
3601: light-sheets at time $\eta$ agree:
3602: \begin{equation}
3603: {S\over A} \leq s\, \eta^{1-2q}.
3604: \label{eq-soaflat}
3605: \end{equation}
3606: Note that one Planck distance corresponds to the comoving coordinate
3607: distance $\Delta\chi=a(\eta)^{-1}$. At the Planck time, $\eta \sim
3608: a(\eta) \sim O(1)$. Hence, $s$ is roughly the amount of entropy
3609: contained in a single Planck volume at one Planck time after the big
3610: bang. This is the earliest time and shortest distance scale one can
3611: hope to discuss without a full quantum gravity description. It is
3612: reasonable to assume that a Planck volume contains no more than one
3613: bit of information:
3614: \begin{equation}
3615: s\lesssim 1.
3616: \end{equation}
3617:
3618: Eq.~(\ref{eq-soaflat}) then implies that the covariant entropy bound,
3619: Eq.~(\ref{eq-bb2}), is satisfied at the Planck time. Moreover, the
3620: bound will continue to be satisfied by all light-sheets of all spheres
3621: at later times ($\eta>1$), if $q \geq {1\over 2}$. In terms of the
3622: parameter $w$, this corresponds to the condition
3623: \begin{equation}
3624: w\leq 1.
3625: \end{equation}
3626: This result was obtained by Fischler and Susskind (1998) who also
3627: assumed $w\geq 0$.
3628:
3629: Kaloper and Linde (1999) showed more generally that the entropy bound
3630: will be satisfied at all times if $-1<w\leq 1$, provided that the
3631: bound is satisfied at the Planck time.\footnote{Like Fischler and
3632: Susskind (1998), this work precedes the covariant entropy bound
3633: (Bousso, 1999a). Hence it considers only the ($--$) case, which
3634: corresponds to the Fischler-Susskind proposal. We have seen that the
3635: entropy range on other light-sheets does not differ significantly in
3636: the flat case. Of course, the {\em absence\/} of a ($--$) light-sheet
3637: for some surfaces in other universes is crucial for the validity of
3638: the covariant entropy bound (see, e.g., Secs.~\ref{sec-lvs},
3639: \ref{sec-collapses}).---Davies (1987) obtained $w \geq -1$ as a
3640: condition for the growth of the apparent horizon in an inflating
3641: universe.} The case $w=-1$ corresponds to de~Sitter space, in which
3642: there is no initial singularity. Since a cosmological constant does
3643: not carry entropy, the bound is trivially satisfied in this case. In
3644: summary, all light-sheets of all surfaces in any flat FRW universe
3645: with equation of state satisfying
3646: \begin{equation}
3647: -1\leq w\leq 1
3648: \label{eq-frwcond}
3649: \end{equation}
3650: satisfy the covariant entropy bound, Eq.~(\ref{eq-bb2}).
3651:
3652: This condition is physically very reasonable. It follows from the
3653: causal energy condition, which prohibits the superluminal flow of
3654: energy. We assumed in Sec.~\ref{sec-econds} that this condition holds
3655: along with the null energy condition. The definitions of all relevant
3656: energy conditions are reviewed in the Appendix.
3657:
3658: \subsubsection{Non-adiabatic evolution and mixed equations of state}
3659: \label{sec-na}
3660:
3661: So far, we have assumed that the universe evolves adiabatically. In
3662: order to relax this assumption, one generally has to abandon the FRW
3663: solution given above and find the exact geometry describing a
3664: cosmology with increasing entropy. However, the global solution will
3665: not change significantly if we rearrange matter on scales smaller than
3666: the apparent horizon.
3667:
3668: Consider the future-ingoing light-sheet of the present apparent
3669: horizon, $L_{+-}[\eta_0, \chi_{\rm AH}(\eta_0)]$. All
3670: entropy we generate using the matter available to us inside the
3671: apparent horizon, will have to pass through this light-sheet. An
3672: efficient way to generate entropy is to form black holes. Building on
3673: a related discussion by Bak and Rey (2000a), Bousso (1999a) showed
3674: that the highest entropy is obtained in the limit where all matter is
3675: converted into a few big black holes. In this limit,
3676: $S_{+-}/A[\eta_0,\chi_{\rm AH}(\eta_0)]$ approaches $1/4$ from below.
3677: Hence the covariant bound is satisfied and can be saturated.
3678:
3679: According to the inflationary model of the early universe (see, e.g.,
3680: Linde, 1990), a different non-adiabatic process occurred at the end of
3681: inflation. At the time of reheating, matter is produced and a large
3682: amount of entropy is generated. One might be concerned that the
3683: holographic principle is violated by inflation (Easther and Lowe,
3684: 1999), or that it places severe constraints on acceptable models
3685: (Kalyana Rama and Sarkar, 1999).
3686:
3687: Before inflation ended, however, there was almost no entropy. Hence,
3688: all past directed light-sheets can be truncated at the reheating
3689: surface, $\eta=\eta_{\rm reheat}$. The energy density at reheating is
3690: expected to be significantly below the Planck density. The
3691: light-sheets will be cut shorter than in our above discussion, which
3692: assumed that standard cosmology extended all the way back to the
3693: Planck era. Hence, inflation leads to no difficulties with the
3694: holographic principle.\footnote{Fabinger (2001) has suggested a bound
3695: on entanglement entropy, assuming certain inflationary models apply.}
3696:
3697: Kaloper and Linde (1999) studied a particularly interesting cosmology,
3698: a flat FRW universe with ordinary matter, $w_1\geq 0, \rho_1>0$, as
3699: well as a small negative cosmological constant, $w_2=-1, \rho_2<0$.
3700: The universe starts matter dominated, but the cosmological constant
3701: eventually takes over the evolution. It slows down and eventually
3702: reverses the expansion. In a time symmetric fashion, matter
3703: eventually dominates and the universe ends in a future singularity.
3704:
3705: The Kaloper-Linde universe provides a tough testing ground for
3706: proposals for a cosmological holographic principle. As in any flat
3707: FRW universe, spacelike holography breaks down for sufficiently large
3708: surfaces. Moreover, as in any collapsing universe, this occurs even
3709: if one restricts to surfaces within the particle horizon, or the
3710: Hubble horizon. Most interestingly, the ``apparent horizon'' proposal
3711: of Bak and Rey (2000a) fails in this cosmology. This can be understood
3712: by applying the spacelike projection theorem to cosmology, as we
3713: discuss next.
3714:
3715: The holographic principle in anisotropic models was discussed by
3716: Fischler and Susskind (1998) and by Cataldo {\em et al.} (2001).
3717: Inhomogeneous universes have been considered by Tavakol and Ellis
3718: (1999); see also Wang, Abdallah, and Osada (2000).
3719:
3720:
3721: \subsubsection{A cosmological corollary}
3722: \label{sec-cosmocor}
3723:
3724: Let us return to a question first raised in Sec.~\ref{sec-range}.
3725: What is the largest volume in a cosmological spacetime to which the
3726: spacelike holographic principle can be applied? The spacelike
3727: projection theorem (Sec.~\ref{sec-spt}) guarantees that the spacelike
3728: entropy bound will hold for surfaces that admit a future directed,
3729: complete light-sheet. Let us apply this to cosmology. Surfaces on or
3730: within the apparent horizon are normal and hence admit a future
3731: directed light-sheet. However, the completeness condition is not
3732: trivial and must be demanded separately. In the Kaloper-Linde
3733: universe, for example, the future light-sheets of sufficiently late
3734: surfaces on the apparent horizon are truncated by the future
3735: singularity.
3736:
3737: We thus arrive at the following corollary to the spacelike projection
3738: theorem (Bousso, 1999a): {\em The area of any sphere within the
3739: apparent horizon exceeds the entropy enclosed in it, if the future
3740: light-sheet of the sphere is complete.}
3741:
3742:
3743: \subsection{Gravitational collapse}
3744: \label{sec-collapses}
3745:
3746: Any argument for an entropy bound based on the generalized second law
3747: of thermodynamics must surely become invalid in a collapse regime.
3748: When a system is already inside its own Schwarzschild radius, it can
3749: no longer be converted into a black hole of equal surface area.
3750:
3751: Indeed, the example of the collapsing star (Sec.~\ref{sec-fail3}), and
3752: the conclusions reached by various analyses of collapsing universes
3753: (Sec.~\ref{sec-range}) would seem to discourage hopes of finding a
3754: non-trivial holographic entropy bound that continues to hold in
3755: regions undergoing gravitational collapse. Surprisingly, to the
3756: extent that it has been tested, the covariant entropy bound does
3757: remain valid in such regions.
3758:
3759: Whenever possible, the validity of the bound is most easily verified
3760: by showing that a given solution satisfies the local hypotheses of
3761: Flanagan, Marolf, and Wald (2000). Otherwise, light-sheets must be
3762: found explicitly.
3763:
3764: Ideally, one would like to investigate systems with high entropy, in
3765: dynamical, collapsing spacetime regions. Generically, such regions
3766: will be extremely inhomogeneous, which makes the practical calculation
3767: of light-sheets difficult. However, one should keep in mind that
3768: other proposals for general entropy bounds, such as the spacelike
3769: entropy bound, are quickly invalidated by simple, easily tractable
3770: counterexamples that make use of gravitational collapse.
3771:
3772: It is remarkable, from this point of view, that the covariant bound
3773: has not met its demise by any of the standard collapse solutions that
3774: are readily available in the literature. To illustrate how the
3775: covariant bound evades violation, we will review its application to
3776: two simple examples, a collapsing star and a closed universe.
3777:
3778: We will also consider a particular setup that allows the calculation
3779: of light-sheets deep inside a black hole formed by the collapse of a
3780: spherical shell. In this example one has good quantitative control
3781: over the collapse of a system of arbitrarily high entropy.
3782:
3783:
3784: \subsubsection{Collapsing universe}
3785: \label{sec-cun}
3786:
3787: Let us begin with a very simple example, the adiabatic recollapse of a
3788: closed FRW universe. In this case the recollapsing phase is just the
3789: time-reversal of the expanding phase. The light-sheet directions are
3790: similarly reversed (Fig.~\ref{fig-clos}a). Small spheres near the
3791: poles are normal, but larger spheres, which are anti-trapped during
3792: expansion will be trapped during collapse. Their light-sheets are
3793: future directed and hence are typically truncated by the future (big
3794: crunch) singularity.
3795:
3796: Because the solution is symmetric under time reversal, the validity of
3797: the covariant entropy bound in the collapse phase follows from its
3798: validity in the expanding phase. The latter can be verified
3799: straightforwardly. For anti-trapped spheres, the calculation
3800: (Fischler and Susskind, 1998) is similar to the analysis of the flat
3801: case (Sec.~\ref{sec-ff}). For small spheres one needs to pay special
3802: attention to choosing the correct inside directions (see
3803: Sec.~\ref{sec-lvs}).
3804:
3805:
3806: \subsubsection{Collapsing star}
3807: \label{sec-cst}
3808:
3809: Next, we return to the collapsing star of Sec.~\ref{sec-fail3}. Why
3810: don't the arguments demonstrating the break-down of other entropy
3811: bounds extend to the covariant entropy bound?
3812:
3813: The metric in and around a collapsing star is well described by the
3814: Oppenheimer-Snyder solution (Misner, Thorne, and Wheeler, 1973). In
3815: this solution, the star is modelled by a suitable portion of a
3816: collapsing closed FRW universe. That is, one considers the coordinate
3817: range
3818: \begin{equation}
3819: 0\leq\chi\leq\chi_0,~~\eta>q{\pi\over 2},
3820: \end{equation}
3821: in the metric of Eq.~(\ref{eq-FRW2}). Here, $q$ depends on the
3822: equation of state in the star according to Eq.~(\ref{eq-q}). Also,
3823: $\chi_0<\pi/2$, so that the star does not overclose the universe.
3824: Outside the star, space is empty. Birkhoff's theorem dictates that
3825: the metric will be given by a portion of the Schwarzschild solution,
3826: Eq.~(\ref{eq-schsch}).
3827:
3828: %%***************************************************************
3829: \begin{figure}[h] \centering
3830: \includegraphics[width=8.5cm]{fig-star}
3831: \caption{Penrose diagram of a collapsing star (shaded). At late
3832: times, the area of the star's surface becomes very small ($B$). The
3833: enclosed entropy (in the spatial region $V$) stays finite, so that the
3834: spacelike entropy bound is violated. The covariant entropy bound
3835: avoids this difficulty because only future directed light-sheets are
3836: allowed. $L$ is truncated by the future singularity; it does not
3837: contain the entire star.}
3838: \label{fig-star}
3839: \end{figure}
3840: %%***************************************************************
3841: The corresponding Penrose diagram is shown in Fig.~\ref{fig-star}.
3842: The light-sheet directions are obtained from the corresponding
3843: portions of the Penrose diagrams for the closed universe
3844: (Fig.~\ref{fig-clos}a) and for the Schwarzschild solution. At
3845: sufficiently late times, the apparent horizon reaches the surface of
3846: the star. At this moment, the star forms a black hole. The surface
3847: of the star is trapped at all later times. Hence, it admits only
3848: future directed light-sheets near the future singularity.
3849:
3850: According to Eq.~(\ref{eq-afrw}), the surface area of the star is
3851: given by
3852: \begin{equation}
3853: A(\chi_0,\eta) = A_{\rm max} \left( \sin{\eta\over q} \right)^q.
3854: \end{equation}
3855: Recall that $q$ is positive and of order unity for realistic equations
3856: of state. At the time of maximum expansion, $A=A_{\rm max} \equiv
3857: 4\pi a_0^2 \sin^2 \chi_0$. The future singularity corresponds to the
3858: time $\eta=q\pi$.
3859:
3860: Let $B$ be the star's surface at a time $\eta_0>q\pi-\chi_0$. The
3861: future directed ingoing light-sheet will be truncated by the future
3862: singularity at $\chi=\chi_0-(q\pi-\eta_0)$, i.e., it will not traverse
3863: the star completely (Fig.~\ref{fig-star}). Hence, it will not contain
3864: the full entropy of the star. For very late times, $\eta_0\to\pi$,
3865: the surface area approaches zero, $A(\chi_0,\eta_0) \to 0$. The
3866: spacelike entropy bound is violated, $S(V)>A(B)$, because the entropy
3867: of the star does not decrease (Sec.~\ref{sec-fail3}). But the entropy
3868: $S(L)$ on the ingoing light-sheet, $L$, vanishes in this limit,
3869: because $L$ probes only a shallow outer shell, rather than the
3870: complete star.
3871:
3872: Light-sheet truncation by future singularities is but one of several
3873: mechanisms that conspire to protect the covariant entropy bound during
3874: gravitational collapse (see Bousso, 1999a).
3875:
3876:
3877: \subsubsection{Collapsing shell}
3878: \label{sec-csh}
3879:
3880:
3881: Consider a small black hole of radius $r_0= 2m$. In the future of the
3882: collapse event that formed this black hole, the apparent black hole
3883: horizon is a null hypersurface with spacelike, spherical
3884: cross-sections of area $A=4\pi r_0^2$.
3885:
3886: Let us pick a particular sphere $B$ of area $A$ on the apparent
3887: horizon. By definition, the expansion of the past directed ingoing
3888: and the future directed outgoing light rays vanishes near $B$, so both
3889: are allowed light-sheet directions.
3890:
3891: The former light-sheet contains all of the infalling matter that
3892: formed the black hole, with entropy $S_{\rm orig}$. The covariant
3893: entropy bound, in this case, is the statement of the generalized
3894: second law: the horizon entropy, $A/4$, is greater than the lost
3895: matter entropy, $S_{\rm orig}$. The future directed ingoing
3896: light rays will be contracting. They will contain entropy $S_{\rm
3897: orig}$ or less, so the covariant bound is satisfied once more.
3898:
3899: We will be interested in the future directed outgoing light-sheet,
3900: $L$. It will continue to generate the apparent horizon of the black
3901: hole. Indeed, if no more matter ever enters the black hole, this
3902: apparent horizon coincides with the event horizon, and the light-sheet
3903: will continue forever at zero expansion.
3904:
3905: Suppose, however, that more matter eventually falls into the black
3906: hole. When this happens, the apparent horizon moves out to a larger
3907: value $r>r_0$. (It will be generated by a new set of light rays that
3908: were formerly expanding.) The light-sheet $L$, however, will begin to
3909: collapse, according to Eq.~(\ref{eq-ray}). The covariant entropy
3910: bound predicts that the light rays will reach a singularity, or a
3911: caustic, before encountering more entropy than $A/4$.
3912:
3913: This is a remarkable prediction. It claims that one cannot collapse
3914: more entropy through a (temporary) black hole horizon than it already
3915: has. This claim has been tested (Bousso, 1999a). Here we summarize
3916: only the method and results.
3917:
3918: Far outside the black hole, one can assemble a shell of matter
3919: concentric with the black hole. By choosing the initial radius of
3920: this shell to be sufficiently large, one can suppress local
3921: gravitational effects and give the shell arbitary total mass, $M$, and
3922: width, $w$.
3923:
3924: Let us assume that the shell is exactly spherically symmetric, even at
3925: the microscopic level. This suppresses the deflection of radial
3926: light rays into angular directions, rendering the eventual calculation
3927: of $L$ tractable. Moreover, it permits an estimate of the entropy of
3928: the shell.
3929:
3930: In weakly gravitating systems, Bekenstein's bound,
3931: Eq.~(\ref{eq-bekbound}), has much empirical support (Bekenstein, 1981,
3932: 1984; Schiffer and Bekenstein, 1989). There is independent evidence
3933: that the bound is always obeyed and can be nearly saturated by
3934: realistic, weakly gravitating matter systems.
3935:
3936: Because all excitations are carried by radial modes, the shell can be
3937: divided along radial walls. This yields several weakly gravitating
3938: systems of largest length scale $w$. To each, Bekenstein's bound
3939: applies. After reassembling the shell, one finds that its total
3940: entropy is bounded by
3941: \begin{equation}
3942: S\leq 2\pi Mw.
3943: \label{eq-shell}
3944: \end{equation}
3945: In principle, there are no restrictions on either $M$ or $w$, so the
3946: amount of entropy that can be collapsed onto the black hole is
3947: unlimited.
3948:
3949: Now consider the adiabatic collapse of the shell. When the inner
3950: surface of the shell has shrunk to area $A$, the shell will first be
3951: reached by the light rays generating $L$. As the light rays penetrate
3952: the collapsing shell, they are focussed by the shell's stress tensor.
3953: Their expansion becomes negative. Eventually they reach a caustic.
3954:
3955: In order to violate the bound with a shell of large entropy, one would
3956: like to ensure that all of the shell's entropy, $S$, will actually be
3957: contained on $L$. Thus, one should demand that the light rays must
3958: not reach a caustic before they have fully crossed the shell and
3959: reemerged on the outer surface of the shell.
3960:
3961: Inspection of the collapse solution, however, reveals that this
3962: requirement restricts the shell's mass and width,
3963: \begin{equation}
3964: M w \leq r_0^2/2.
3965: \end{equation}
3966: By Eq.~(\ref{eq-shell}), this also limits the entropy of the shell:
3967: \begin{equation}
3968: S \leq \pi r_0^2= {A\over 4}.
3969: \end{equation}
3970: The entropy on the light-sheet $L$ may saturate the covariant bound,
3971: but it will not violate it.
3972:
3973:
3974:
3975: \subsection{Nearly null boundaries}
3976: \label{sec-nearnull}
3977:
3978: In Sec.~\ref{sec-fail4} it was shown that any isolated, weakly
3979: gravitating matter system can be surrounded with a closed surface of
3980: arbitarily small area, in violation of the spacelike entropy bound,
3981: Eq.~(\ref{eq-seb}).
3982:
3983: In order to capture the key advantage of the light-sheet formulation,
3984: Eq.~(\ref{eq-bb2}), we find it simplest to consider a square-shaped
3985: system occupying the region $0\leq x,y \leq a$ in 2+1 dimensional
3986: Minkowski space; $t$ is the time coordinate in the system's rest frame
3987: (Fig.~\ref{fig-wiggly}a). The boundary length of the system at $t=0$
3988: is
3989: \begin{equation}
3990: A_0 = 4a.
3991: \end{equation}
3992: %%***************************************************************
3993: \begin{figure}[h] \centering
3994: \includegraphics[width=8cm]{fig-wiggly}
3995: \caption{(a) A square system in 2+1 dimensions, surrounded by a
3996: surface $B$ of almost vanishing length $A$. (b) [Here the time
3997: dimension is projected out.] The light-sheet of $B$ intersects only
3998: with a negligible (shaded) fraction of the system.}
3999: \label{fig-wiggly}
4000: \end{figure}
4001: %%***************************************************************
4002:
4003: Let us define a new boundary $B$ by a zig-zag curve consisting of the
4004: following four segments: $y=0$, $t=\beta x$ for $0\leq x\leq a$;
4005: $x=a$, $t=\beta (a-y)$ for $0\leq y\leq a$; $y=a$, $t=\beta (a-x)$ for
4006: $0\leq x \leq a$; and $x=0$, $t=\beta y$ for $0\leq y \leq a$. This
4007: can be regarded as the boundary of the system in some non-standard
4008: time-slicing. Its length is Lorentz-contracted relative to the
4009: boundary in the rest frame:
4010: \begin{equation}
4011: A(B) = A_0 \sqrt{1-\beta^2}.
4012: \end{equation}
4013: The length of $B$ vanishes in the limit as $\beta\to 1$.
4014:
4015: The future-ingoing light-sheet $L(B)$ can be computed by piecing
4016: together the light-sheets of all four segments. The light-sheet of
4017: the first segment is obtained by translating the segment in the
4018: direction $(1, \beta, \sqrt{1-\beta^2})$. (It is instructive to
4019: verify that this generates an orthogonal null hypersurface of
4020: vanishing expansion. The curvature of spacetime is neglected in order
4021: to isolate the effect of ``wiggling'' the boundary.) For
4022: $\beta^2>{1\over 2}$, this light-sheet covers a fraction
4023: $\frac{\sqrt{1-\beta^2}}{2\beta}$ of the total system
4024: (Fig.~\ref{fig-wiggly}b). The light-sheets of the other segments are
4025: similarly computed.
4026:
4027: To leading order in $(1-\beta)$, the total fraction of the system
4028: covered by $L(B)$,
4029: \begin{equation}
4030: {V(\beta)\over V_0} = \frac{2\sqrt{1-\beta^2}}{\beta},
4031: \end{equation}
4032: vanishes at
4033: the same rate as the boundary length.
4034:
4035: The future-ingoing light-sheet is not complete in this case; it has
4036: boundaries running through the interior of the system. Hence, the
4037: assumptions of the spacelike projection theorem are not satisfied.
4038: (This is not just an artifact of the sharp edges of $B$. If $B$ was
4039: smoothed at the edges, it would contain a segment on which only past
4040: directed light rays would be contracting. Thus, $B$ would not admit a
4041: future directed light-sheet everywhere, and the spacelike projection
4042: theorem would still not apply.)
4043:
4044:
4045:
4046:
4047: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4048: \section{The holographic principle}
4049: \label{sec-hp}
4050: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4051:
4052:
4053: \subsection{Assessment}
4054:
4055: The previous sections have built a strong case for a holographic
4056: principle.\footnote{All of the following points are
4057: independent of the considerations of economy and unitarity that
4058: motivated 't~Hooft's and Susskind's holographic principle
4059: (Sec.~\ref{sec-hsi}). However, those arguments emerge strengthened,
4060: since a key difficulty, the absence of a general entropy bound, has
4061: been overcome (Sec.~\ref{sec-limitations}). One can no longer object
4062: that more than $A/4$ degrees of freedom might be needed to describe
4063: the physics, say, in strongly gravitating regions.}
4064:
4065:
4066:
4067: \begin{itemize}
4068:
4069: \item{The covariant entropy bound is well-defined
4070: (Sec.~\ref{sec-bb}). The light-sheet construction establishes a
4071: precise relation between surfaces and adjacent hypersurfaces. The
4072: area of the former must be compared to the entropy contained on the
4073: latter. Thus the bound is testable, in an arena limited only by the
4074: range of semi-classical gravity, the approximate framework we are
4075: compelled to use until a general quantum theory of gravity becomes
4076: available. Like any law of physics, it can of course be tested only
4077: to the extent that the relevant quantities and constructs (here, area,
4078: light-sheets, and entropy) are practically computable. But
4079: importantly, the bound will not become ill-defined in a regime which
4080: is otherwise physically well-understood.}
4081:
4082: \item{The bound has been examined and found to hold in a wide range
4083: of examples, some of which we reviewed in Sec.~\ref{sec-tests}. No
4084: physically realistic counterexample has been found. This is
4085: remarkable especially in view of the ease with which the general
4086: validity of some alternative proposals can be excluded
4087: (Sec.~\ref{sec-seb}).}
4088:
4089: \item{The bound is non-trivial. Naively one would expect the maximal
4090: entropy to grow with the volume of spatial regions. Instead, it is
4091: set by the area of surfaces.}
4092:
4093: \item{The bound refers to statistical entropy.\footnote{A conventional
4094: thermodynamic interpretation is clearly not tenable. Most
4095: thermodynamic quantities are not defined in general spacetimes.
4096: Moreover, the time direction imprinted on thermodynamic entropy
4097: conflicts with the invariance of the covariant entropy bound under
4098: reversal of time (Bousso, 1999a).} Since it involves no assumptions
4099: about the microscopic properties of matter, it places a fundamental
4100: limit on the number of degrees of freedom in nature.}
4101:
4102: \item{The bound is not explained by other laws of physics that are
4103: presently known. Unlike its less general predecessors (e.g., the
4104: spherical entropy bound, Sec.~\ref{sec-spheb}), the covariant bound
4105: cannot be regarded merely as a consequence of black hole
4106: thermodynamics. Arguments involving the formation of black holes
4107: cannot explain an entropy bound whose scope extends to the deep
4108: interior of black holes and to cosmology.---We conclude that {\em the
4109: bound is an imprint of a more fundamental theory.}}
4110:
4111: \item{Yet, the covariant bound is closely related to the black hole
4112: entropy and the generalized second law, long considered important
4113: clues to quantum gravity. Though the bound does not itself follow
4114: from thermodynamics, it implies other bounds which have been argued to
4115: be necessary for upholding the second law (Sec.~\ref{sec-rob}). We
4116: also note that the bound essentially involves the quantum states of
4117: matter.---We conclude that {\em the fundamental theory responsible for
4118: the bound unifies matter, gravity, and quantum mechanics.}}
4119:
4120: \item{The bound relates information to a single geometric quantity
4121: (area). The bound's simplicity, in addition to its generality, makes
4122: the case for its fundamental significance compelling.---We conclude
4123: that {\em the area of any surface $B$ measures the information content
4124: of an underlying theory describing all possible physics on the
4125: light-sheets of $B$}.\footnote{An entropy bound in terms of a more
4126: complex combination of physical quantities (e.g., Brustein and
4127: Veneziano, 2000), even if it holds generally, would not seem to betray
4128: a concrete relation of this kind.}}
4129:
4130: \end{itemize}
4131:
4132:
4133: \subsection{Formulation}
4134:
4135: Let us combine the three conclusions drawn above (in italics) and
4136: formulate the {\sl holographic principle} (Bousso, 1999a,b).
4137:
4138: {\em The covariant entropy bound is a law of
4139: physics which must be manifest in an underlying theory. This theory
4140: must be a unified quantum theory of matter and spacetime. From it,
4141: Lorentzian geometries and their matter content must emerge in such a
4142: way that the number of independent quantum states describing the
4143: light-sheets of any surface $B$ is manifestly bounded by the
4144: exponential of the surface area:}
4145: \begin{equation}
4146: {\cal N}[L(B)] \leq e^{A(B)/4}.
4147: \end{equation}
4148: (See Secs.~\ref{sec-not}, \ref{sec-presc} for notation.)
4149:
4150: Implicit in the phrase ``quantum states'' is the equivalence, in
4151: quantum theory, of the logarithm of the dimension ${\cal N}$ of
4152: Hilbert space and the amount of information stored in the quantum
4153: system. As it is not obvious that quantum mechanics will be primary
4154: in a unified theory, a more neutral formulation of the holographic
4155: principle may be preferable:
4156:
4157: {\em $N$, the number of degrees of freedom (or the number of bits
4158: times $\ln 2$) involved in the description of $L(B)$, must not exceed
4159: $A(B)/4$.}
4160:
4161:
4162: \subsection{Implications}
4163: \label{sec-imp}
4164:
4165: The holographic principle implies a radical reduction in the number of
4166: degrees of freedom we use to describe nature. It exposes quantum
4167: field theory, which has degrees of freedom at every point in space, as
4168: a highly redundant effective description, in which the true number of
4169: degrees of freedom is obscured (Sec.~\ref{sec-qftwrong}).
4170:
4171: The holographic principle challenges us to formulate a theory in which
4172: the covariant entropy bound is manifest. How can a {\em holographic
4173: theory\/} be constructed? Physics appears to be local to a good
4174: approximation. The number of degrees of freedom in any local theory
4175: is extensive in the volume. Yet, the holographic principle dictates
4176: that the information content is in correspondence with the area of
4177: surfaces. How can this tension be resolved? There appear to be two
4178: main lines of approach, each casting the challenge in a different
4179: form.
4180:
4181: One type of approach aims to retain locality. A local theory could be
4182: rendered holographic if an explicit gauge invariance was identified,
4183: leaving only as many physical degrees of freedom as dictated by the
4184: covariant entropy bound. The challenge, in this case, is to implement
4185: such an enormous and rather peculiar gauge invariance.
4186:
4187: For example, 't~Hooft (1999, 2000a, 2001a,b,c) is pursuing a local
4188: approach in which quantum states arise as limit cycles of a classical
4189: dissipative system (see also van de Bruck, 2000). The emergence of an
4190: area's worth of physical degrees of freedom has yet to be demonstrated
4191: in such models.
4192:
4193: A second type of approach regards locality as an emergent phenomenon
4194: without fundamental significance. In this case, the holographic
4195: data are primary. The challenge is not only to understand their
4196: generation and evolution. One must also explain how to translate
4197: underlying data, in a suitable regime, into a classical spacetime
4198: inhabited by local quantum fields. In a successful construction, the
4199: geometry must be shaped and the matter distributed so as to satisfy
4200: the covariant entropy bound. Because holographic data are most
4201: naturally associated with the area of surfaces, a serious difficulty
4202: arises in understanding how locality can emerge in this type of
4203: approach.
4204:
4205: The AdS/CFT correspondence (Sec.~\ref{sec-ads}) lends credence to the
4206: second type of approach. However, because it benefits from several
4207: peculiarities of the asympotically AdS universes to which it applies
4208: (Sec.~\ref{sec-screens}), it has offered little help to researchers
4209: pursuing such approaches more broadly.
4210:
4211: Some of the proposals and investigations discussed in Secs.~\ref{sec-toe}
4212: and \ref{sec-ds} can be associated to the second type.
4213:
4214: Which type of approach one prefers will depend, to a great extent, on
4215: which difficulty one abhors more: the elimination of most degrees of
4216: freedom, or the recovery of locality. The dichotomy is hardly strict;
4217: the two alternatives are not mutually exclusive. A successful theory
4218: may admit several equivalent formulations, thus reconciling both
4219: points of view.
4220:
4221: Since light-sheets are central to the formulation of the holographic
4222: principle, one would expect null hypersurfaces to play a primary role
4223: in the classical limit of an underlying holographic theory (though
4224: this may not be apparent in descriptions of weakly time-dependent
4225: geometries; see Sec.~\ref{sec-ads}).
4226:
4227:
4228: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4229: \section{Holographic screens and holographic theories}
4230: \label{sec-hsht}
4231: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4232:
4233:
4234:
4235: We will begin this section by discussing which aspects of the
4236: holographic principle have already been realized in string theory. We
4237: assess how general the class of universes is in which the holographic
4238: principle is thus implemented. In this context, we will present the
4239: most explicit example of a holographic theory presently known. The
4240: AdS/CFT correspondence defines quantum gravity---albeit in a limited
4241: set of spacetimes. Anti-de Sitter space contains a kind of
4242: holographic screen, a distant hypersurface on which holographic data
4243: can be stored and evolved forward using a conformal field theory.
4244:
4245: We will then review the construction of holographic screens in general
4246: spacetimes, including those without boundary. Using light-sheets, it
4247: is always possible to find such screens. However, a theory that
4248: generally describes the generation and evolution of holographic data
4249: remains elusive. The structure of screens offers some clues about the
4250: difficulties that must be addressed. We will list a number of
4251: approaches.
4252:
4253: We will also discuss the application of the covariant entropy bound to
4254: universes with positive vacuum energy. In this class of spacetimes
4255: the holographic principle appears to place a particularly strong
4256: constraint on an underlying description.
4257:
4258:
4259: \subsection{String theory and the holographic principle}
4260: \label{sec-strings}
4261:
4262: \subsubsection{A work in progress}
4263:
4264: String theory naturally produces a unified quantum description of
4265: gravity and matter fields. Its framework has proven self-consistent
4266: in remarkably non-trivial ways, given rise to powerful mathematical
4267: structures, and solved numerous physical problems. One might wonder
4268: what the holographic principle is still needed for. If a good theory
4269: is available, why search further? What is left to do?
4270:
4271: String theory has developed in an unconventional way. It began as a
4272: formula whose physical interpretation in terms of strings was
4273: understood only later. The theory was first misunderstood as a
4274: description of hadrons, and only later recognized as a quantum theory
4275: of gravity. It forms part of a rigid mathematical structure whose
4276: content and physical implications continue to be explored.
4277:
4278: String theory\footnote{We shall take related eleven-dimensional
4279: theories to be included in this term.} has yet to address many of the
4280: most pressing questions one would like to ask of a fundamental theory.
4281: These include phenomenological issues: Why does the world have four
4282: large dimensions? What is the origin of the stardard model? How is
4283: supersymmetry broken? More importantly, there are conceptual
4284: difficulties. It is unclear how the theory can be applied to
4285: realistic cosmological spacetimes, and how it might describe most
4286: black holes and singularities of general relativity.
4287:
4288: String theory's most notable recent successes hinged on the discovery
4289: of a new set of objects in the theory, D-branes (Polchinksi, 1995).
4290: Before D-branes, string theory's list of open questions was longer
4291: than it is today. This serves as a reminder that unsolved problems
4292: need not signal the failure of string theory. Neither should they be
4293: dismissed as mere technical difficulties. Instead, they may indicate
4294: that there are still crucial parts of the theory that have not been
4295: discovered.
4296:
4297: There is little evidence that string theory, in its current form,
4298: represents more than a small portion, or a limiting case, of a bigger
4299: theoretical structure. Nor is it clear that the exploration of this
4300: structure will continue to proceed most efficiently from
4301: within.\footnote{In particular, Banks (2000b) has argued that there
4302: may be no sense in which all isolated ``vacua'' of the theory can be
4303: smoothly connected.}
4304:
4305: An intriguing success of the covariant entropy bound is its validity
4306: in highly dynamical geometries, whose description has proven
4307: especially difficult in string theory. This suggests that the
4308: holographic principle may offer useful guidance to the further
4309: development of the theory.
4310:
4311: Its present limitations prevent string theory from explaining the
4312: general validity of the covariant entropy bound. The theory is not
4313: under control in many situations of interest, for example when
4314: supersymmetry is broken. Moreover, many solutions of physical
4315: relevance, including most of the examples in this text, do not appear
4316: to be admitted by string theory in its current form.
4317:
4318:
4319: \subsubsection{Is string theory holographic?}
4320:
4321: These restrictions aside, one may ask whether the holographic
4322: principle is manifest in string theory. Let us consider, for a
4323: moment, only spacetimes that string theory can describe, and in which
4324: the holographic principle is also well-defined (i.e., geometry is
4325: approximately classical). Is the number of degrees of freedom
4326: involved in the string theory description set by the area of surfaces?
4327:
4328: In perturbative string theory, the holographic principle is only
4329: partly realized. Effects associated with holography include the
4330: independence of the wave function on the longitudinal coordinate in
4331: the light cone frame, and the growth of the size of states with their
4332: momentum (see the reviews cited in Sec.~\ref{sec-further}; Thorn, {\em
4333: op.\ cit.}; Klebanov and Susskind, 1988; Susskind, 1995b; see also
4334: Susskind, 1995a).
4335:
4336: A number of authors have studied the extent to which string theory
4337: exhibits the non-locality implied by the holographic principle (Lowe,
4338: Susskind, and Uglum, 1994; Lowe {\em et al.}, 1995). These
4339: investigations are closely related to the problem of understanding of
4340: the unitarity of black hole evaporation from the point of view of
4341: string theory, in particular through the principle of black hole
4342: complementarity (Sec.~\ref{sec-bhc}).
4343:
4344: The entropy bound of one bit per Planck area, however, is not explicit
4345: in perturbative string theory. Susskind (1995b) showed that the
4346: perturbative expansion breaks down before the bound is violated (see
4347: also Banks and Susskind, 1996). One would expect the holographic
4348: principle to be fully manifest only in a non-perturbative formulation
4349: of the theory.
4350:
4351: Since the holographic principle was conceived, non-perturbative
4352: definitions of string theory have indeed become available for two
4353: special classes of spacetimes. Remarkably, in the AdS/CFT
4354: correspondence, the number of degrees of freedom agrees manifestly
4355: with the holographic principle, as we discuss below. In Matrix theory
4356: (Banks {\em et al.}, 1997) the corresponding arguments are somewhat
4357: less precise. This is discussed, e.g., by Bigatti and Susskind
4358: (1997), and by Banks (1998, 1999), where further references can be
4359: found.\footnote{A significant non-perturbative result closely related
4360: to the holographic principle is the microscopic derivation of the
4361: entropy of certain black holes in string theory (Strominger and Vafa,
4362: 1996).}
4363:
4364: The holographic principle may not only aid the search for other
4365: non-perturbative definitions of string theory. It could also
4366: contribute to a background-independent formulation that would
4367: illuminate the conceptual foundation of string theory.
4368:
4369:
4370: \subsection{AdS/CFT correspondence}
4371: \label{sec-ads}
4372:
4373: An example of the AdS/CFT correspondence concerns type IIB string
4374: theory in an asymptotically AdS$_5\times \mathbf{S}^5$ spacetime (the
4375: {\em bulk}), with $n$ units of five-form flux on the
4376: five-sphere\footnote{There is a notational conflict with most of the
4377: literature, where $N$ denotes the size of the gauge group. In this
4378: review, $N$ is reserved for the number of degrees of freedom
4379: (Sec.~\ref{sec-ndof}).} (Maldacena, 1998; Gubser, Klebanov, and
4380: Polyakov, 1998; Witten, 1998). This theory, which includes gravity,
4381: is claimed to be non-perturbatively defined by a particular conformal
4382: field theory without gravity, namely 3+1 dimensional supersymmetric
4383: Yang-Mills theory with gauge group $U(n)$ and 16 real supercharges.
4384: We will refer to this theory as the {\em dual CFT}.
4385:
4386: The metric of AdS$_5\times \mathbf{S}^5$ is
4387: \begin{equation}
4388: ds^2 = R^2 \left[ - \frac{1+r^2}{1-r^2} dt^2 + \frac{4}{(1-r^2)^2}
4389: \left( dr^2 + r^2 d\Omega_3^2 \right) + d\Omega_5^2 \right],
4390: \label{eq-ads-metric}
4391: \end{equation}
4392: where $d\Omega_d$ denotes the metric of a $d$-dimensional unit sphere.
4393: The radius of curvature is related to the flux by the formula
4394: \begin{equation}
4395: R = n^{1/4},
4396: \label{eq-rflux}
4397: \end{equation}
4398: in units of the ten-dimensional Planck length.
4399:
4400: The proper area of the three-spheres diverges as $r \rightarrow 1$.
4401: After conformal rescaling (Hawking and Ellis, 1973), the spacelike
4402: hypersurface, $t=\mbox{const}, 0 \leq r < 1$ is an open ball, times a
4403: five-sphere. (The conformal picture for AdS space thus resembles the
4404: worldvolume occupied by a spherical system, as in Fig.~\ref{fig-ball}.)
4405: Because the five-sphere factor has constant physical radius, and the
4406: scale factor vanishes as $r\to 1$, the five-sphere is scaled to a
4407: point in this limit. Thus, the conformal boundary of space is a
4408: three-sphere residing at $r=1$.
4409:
4410: It follows that the conformal boundary of the spacetime is ${\mathbb
4411: R} \times \mathbf{S}^3$. This agrees with the dimension of the CFT.
4412: Hence, it is often said that the dual CFT ``lives'' on the boundary of
4413: AdS space.
4414:
4415: The idea that data given on the boundary of space completely describe
4416: all physics in the interior is suggestive of the holographic
4417: principle. It would appear that the dual CFT achieves what local
4418: field theory in the interior could not do. It contains an area's
4419: worth of degrees of freedom, avoiding the redundancy of a local
4420: description. However, to check quantitatively whether the holographic
4421: bound really manifests itself in the dual CFT, one must compute the
4422: CFT's number of degrees of freedom, $N$. This must not exceed the
4423: boundary area, $A$, in ten-dimensional Planck units.
4424:
4425: The proper area of the boundary is divergent. The number of degrees
4426: of freedom of a conformal field theory on a sphere is also divergent,
4427: since there are modes at arbitrarily small scales. In order to make a
4428: sensible comparison, Susskind and Witten (1998) regularized the bulk
4429: spacetime by removing the region $1-\delta<r<1$, where $\delta\ll 1$.
4430: This corresponds to an infrared cutoff. The idea is that a modified
4431: version of the AdS/CFT correspondence still holds for this truncated
4432: spacetime.
4433:
4434: The area of the $\mathbf{S}^3 \times \mathbf{S}^5$ boundary surface%
4435: %
4436: \footnote{Unlike Susskind and Witten (1998), we do not compactify the
4437: bulk to five dimensions in this discussion; all quantities refer to a
4438: ten-dimensional bulk. Hence the area is eight-dimensional.}
4439: %
4440: is approximately given by
4441: \begin{equation}
4442: A\approx {R^8\over\delta^3}
4443: \end{equation}
4444: In order to find the number of degrees of freedom of the dual CFT, one
4445: has to understand how the truncation of the bulk modifies the CFT.
4446: For this purpose, Susskind and Witten (1998) identified and exploited
4447: a peculiar property of the AdS/CFT correspondence: infrared effects in
4448: the bulk correspond to ultraviolet effects on the boundary.
4449:
4450: There are many detailed arguments supporting this so-called {\em UV/IR
4451: relation\/} (see also, e.g., Balasubramanian and Kraus, 1999; Peet and
4452: Polchinski, 1999). Here we give just one example. A string stretched
4453: across the bulk is represented by a point charge in the dual CFT. The
4454: energy of the string is linearly divergent near the boundary. In the
4455: dual CFT this is reflected in the divergent self-energy of a point
4456: charge. The bulk divergence is regularized by an infrared cut-off,
4457: which renders the string length finite, with energy proportional to
4458: $\delta^{-1}$. In the dual CFT, the same finite result for the
4459: self-energy is achieved by an ultraviolet cutoff at the short distance
4460: $\delta$.
4461:
4462: We have scaled the radius of the three-dimensional conformal sphere to
4463: unity. A short distance cut-off $\delta$ thus partitions the sphere
4464: into $\delta^{-3}$ cells. For each quantum field, one may expect to
4465: store a single bit of information per cell. A $U(n)$ gauge theory
4466: comprises roughly $n^2$ independent quantum fields, so the total
4467: number of degrees of freedom is given by
4468: \begin{equation}
4469: N \approx \frac{n^2}{\delta^3}.
4470: \end{equation}
4471: Using Eq.~(\ref{eq-rflux}) we find that the CFT number of degrees of
4472: freedom saturates the holographic bound,
4473: \begin{equation}
4474: N \approx A,
4475: \end{equation}
4476: where we must keep in mind that this estimate is only valid to within
4477: factors of order unity.
4478:
4479: Thus, the number of CFT degrees of freedom agrees with the number of
4480: physical degrees of freedom contained on any light-sheet of the
4481: boundary surface $\mathbf{S}^3 \times \mathbf{S}^5$. One must also
4482: verify that there is a light-sheet that contains all of the entropy in
4483: the spacetime. If all light-sheets terminated before reaching $r=0$,
4484: this would leave the possibility that there is additional information
4485: in the center of the universe which is not encoded by the CFT. In
4486: that case, the CFT would not provide a complete description of the
4487: full bulk geometry---which is, after all, the claim of the AdS/CFT
4488: correspondence.
4489:
4490: The boundary surface is normal (Bousso, 1999b), so that both past and
4491: future ingoing light-sheets exist. In an asymptotically AdS$_5\times
4492: \mathbf{S}^5$ spacetime without past or future singularities, either
4493: of these light-sheets will be complete. Thus one may expect the CFT
4494: to describe the entire spacetime.%
4495: %
4496: \footnote{If there are black holes in the spacetime, then the
4497: future directed light-sheet may cross the black hole horizon and end
4498: on the future singularity. Then the light-sheet may miss part of the
4499: interior of the black hole. One can still argue that the CFT
4500: completely describes all physics accessible to an observer at
4501: infinity. A light-sheet can be terminated at the black hole horizon,
4502: with the horizon area added to its entropy content. The data on a
4503: horizon, in turn, are complementary to the information in the black
4504: hole interior (Susskind, Thorlacius, and Uglum, 1993).}
4505: %
4506:
4507: Thus, the CFT state on the boundary (at one instant of time) contains
4508: holographic data for a complete slice of the spacetime. The full
4509: boundary of the spacetime includes a time dimension and is given by
4510: $\mathbb R\times \mathbf{S}^3\times \mathbf{S}^5$. Each moment of
4511: time defines an $\mathbf{S}^3\times \mathbf{S}^5$ boundary area, and
4512: each such area admits a complete future directed light-sheet. The
4513: resulting sequence of light-sheets foliate the spacetime into a stack
4514: of light cones (each of which looks like the cone in
4515: Fig.~\ref{fig-ball}c). There is a slice-by-slice holographic
4516: correspondence between bulk physics and dual CFT data. By the
4517: spacelike projection theorem (Sec.~\ref{sec-spt}), the same
4518: correspondence holds for the spacelike slicing shown in
4519: Fig.~\ref{fig-ball}a.
4520:
4521: Thus, a spacelike formulation of the holographic principle is mostly
4522: adequate in AdS. In recent years there has been great interest in
4523: models of our universe in which four-dimensional gauge fields are
4524: holographic duals to the physics of an extra spatial dimension (see,
4525: e.g., Randall and Sundrum, 1999a,b)---a kind of ``inverse
4526: holography''. Such models can be realized by introducing codimension
4527: one objects, {\em branes}, into a five-dimensional bulk spacetime.
4528: If the bulk is Anti-de Sitter space, the holographic correspondence is
4529: expected to be a version of the AdS/CFT correspondence. In a very
4530: general class of models (Karch and Randall, 2001), the brane fields
4531: are dual only to a portion of the bulk. Attempts to apply the
4532: spacelike holographic principle lead to contradictions in this case,
4533: and the use of the light-sheet formulation is essential (Bousso and
4534: Randall, 2001).
4535:
4536: To summarize, the AdS/CFT correspondence exhibits the following
4537: features:
4538: \begin{itemize}
4539: \item{There exists a slicing of the spacetime such that the state of
4540: the bulk on each slice is fully described by data not exceeding $A$
4541: bits, where $A$ is the area of the boundary of the slice.}
4542: \item{There exists a theory without redundant degrees of freedom, the
4543: CFT, which generates the unitary evolution of boundary data from slice
4544: to slice.}
4545: \end{itemize}
4546:
4547: Perhaps due to the intense focus on the AdS/CFT correspondence in
4548: recent years, the holographic principle has come to be widely regarded
4549: as synonymous with these two properties. Their partial failure to
4550: generalize to other spacetimes has sometimes been confused with a
4551: failure of the holographic principle. We emphasize, therefore, that
4552: neither property is sufficient or necessary for the holographic
4553: principle, as defined in Sec.~\ref{sec-hp}.
4554:
4555: Assuming the validity of the covariant entropy bound in arbitrary
4556: spacetimes, Bousso (1999b) showed that a close analogue of the first
4557: property always holds. The second, however, is not straightforwardly
4558: generalized. It should not be regarded as a universal consequence of
4559: the holographic principle, but as a peculiarity of Anti-de Sitter
4560: space.
4561:
4562:
4563:
4564: \subsection{Holographic screens for general spacetimes}
4565: \label{sec-screens}
4566:
4567: \subsubsection{Construction}
4568:
4569: Any spacetime, including closed universes, contains a kind of
4570: holographic boundary, or screen. It is most easily obtained by
4571: slicing the spacetime into light cones. The total entropy on each
4572: light cone can be holographically stored on the largest surface
4573: embedded in the cone. Our construction follows Bousso (1999b); see
4574: also Bigatti and Susskind (2000), Bousso (2000a).
4575:
4576: Consider the past light cone, ${\cal L}^-$ (technically, the boundary
4577: of the past), of a point $p$ in any spacetime satisfying the null
4578: energy condition. The following considerations will show that ${\cal
4579: L}^-$ consists of one or two light-sheets.
4580:
4581: The area spanned by the light rays will initially increase with affine
4582: parameter distance $\lambda$ from $p$. In some cases, for example
4583: AdS, the area keeps increasing indefinitely. For any surface
4584: $B(\lambda_1)$ the holographic principle implies that the total number
4585: of degrees of freedom on the portion $0\leq\lambda\leq\lambda_1$ is
4586: bounded by $A(B(\lambda_1))/4$. One can express this by saying that
4587: $B(\lambda_1)$ is a {\em holographic screen}, a surface on which the
4588: information describing all physics on the enclosed light cone portion
4589: can be encoded at less than one bit per Planck area. If the light
4590: cone is extended indefinitely, it will reach the conformal boundary of
4591: spacetime, where its area diverges. In this limit one obtains a
4592: holographic screen for the entire light cone.
4593:
4594: A second possibility is that the area does not increase forever with
4595: the affine parameter. Instead, it may reach a maximum, after which it
4596: starts to contract. The focussing theorem (Sec.~\ref{sec-ray})
4597: implies that contracting light rays will eventually reach caustics or
4598: a singularity of the spacetime. Let us continue the light cone until
4599: such points are reached.
4600:
4601: Let $B$ be the apparent horizon, i.e., the spatial surface with
4602: maximum area on the light cone. $B$ divides the light cone into two
4603: portions. By construction, the expansion of light rays in both
4604: directions away from $B$ vanishes locally and is non-positive
4605: everywhere. (We will not be concerned with the second pair of null
4606: directions, which does not coincide with the light cone.) Hence, both
4607: portions are light-sheets of $B$. It follows that the total number of
4608: degrees of freedom on the light cone is bounded by the area of its
4609: largest spatial surface:
4610: \begin{equation}
4611: N \leq {A(B)\over 2}.
4612: \end{equation}
4613: The denominator is 2 because the holographic bound ($A/4$) applies
4614: separately to each light-sheet, and the light cone consists of two
4615: light-sheets.
4616:
4617: Consider, for example, a universe that starts with an initial
4618: singularity, a big bang. Following light rays backwards in time, our
4619: past light cone grows at first. Eventually, however, it must shrink,
4620: because all areas vanish as the big bang is approached.
4621:
4622: One can summarize both cases by the statement that a holographic
4623: screen for all the data on a light cone is the surface where its
4624: spatial area is largest. A global holographic screen for the entire
4625: spacetime can now be constructed as follows.
4626:
4627: One picks a worldline $P(t)$ and finds the past light cone ${\cal
4628: L}^-(t)$ of each point. The resulting stack of light cones foliates
4629: the spacetime.%
4630: %
4631: \footnote{A few remarks are in order. 1.~A foliation can also be
4632: obtained from future light cones, or from more general null
4633: hypersurfaces. 2.~ Depending on global structure, the past light
4634: cones may foliate only the portion of the spacetime visible to the
4635: observer. Suitable extensions permit a global foliation by other null
4636: hypersurfaces. 3.~If light rays generating the past light cone of $p$
4637: intersect, they leave the boundary of the past of $p$ and become
4638: timelike separated from $p$. To obtain a good foliation, one should
4639: terminate such light rays even if they intersect with non-neighboring
4640: light rays, as suggested by Tavakol and Ellis (1999). This can only
4641: shorten the light-sheet and will not affect our conclusions.}
4642: %
4643: Each cone has a surface of maximal area, $B(t)$. These surfaces form
4644: a hypersurface in the spacetime or on its boundary. Cone by cone, the
4645: information in the spacetime bulk can be represented by no more than
4646: $A(t)$ bits on the screen, where $A(t)$ is the area of $B(t)$.
4647:
4648: In suitably symmetric spacetimes, the construction of holographic
4649: screens is simplified by a Penrose diagram. The spacetime must first
4650: be divided into ``wedge domains'', as shown in Fig.~\ref{fig-clos}a
4651: for a closed universe. (A light cone foliation corresponds to a set
4652: of parallel lines at 45 degrees to the vertical. The remaining
4653: ambiguity corresponds to the choice of past or future light cones.)
4654: In order to get to a holographic screen, one has to follow each line
4655: in the direction of the tip of the wedge. Either one ends up at a
4656: boundary, or at an apparent horizon, where the wedge flips.
4657:
4658: The example shown in Fig.~\ref{fig-clos}b is remarkable because it
4659: demonstrates that holographic screens can be constructed for closed
4660: universes. Thus, an explanation of the origin of the holographic
4661: principle should not ultimately hinge upon the presence of a boundary
4662: of spacetime, as it does in the AdS/CFT correspondence.
4663:
4664: Using the general method given above, global holographic screens have
4665: been constructed explicitly for various other spacetimes (Bousso,
4666: 1999b), including Minkowski space, de~Sitter space, and various FRW
4667: universes. In many cases, they do form a part of the boundary of
4668: spacetime, for example in asymptotically AdS, Minkowski, and
4669: de~Sitter spacetimes.%
4670: %
4671: \footnote{Some subtleties arise in the de~Sitter case which allow,
4672: alternatively, the use of a finite area apparent horizon as a screen
4673: (Bousso 1999b). See also Sec.~\ref{sec-ds}}
4674: %
4675: For several examples, Penrose diagrams with wedges and screens are
4676: found in Bousso (1999b) and Bigatti and Susskind (2000).
4677:
4678:
4679: \subsubsection{Properties and implications}
4680:
4681: Some of the properties of the boundary of AdS, such as its area and
4682: its behavior under conformal transformations, can be used to infer
4683: features of the dual CFT. Properties of global holographic screens
4684: can similarly provide clues about holographic theories underlying
4685: other classes of spacetimes (Bousso, 1999b).
4686:
4687: In AdS, the global holographic screen is unique. It is the direct
4688: product of a spatial sphere at infinity with the real time axis. If
4689: the sphere is regulated, as in Sec.~\ref{sec-ads} above, its area can
4690: be taken to be constant in time. None of these properties are
4691: necessarily shared by the global screens of other spacetimes. Let us
4692: identify some key differences and discuss possible implications.
4693:
4694: \begin{itemize}
4695:
4696: \item{In general, global holographic screens are highly non-unique. For
4697: example, observers following different worldlines correspond to
4698: different stacks of light cones; their screens do not usually agree.}
4699:
4700:
4701: \item{One finds that spacetimes with horizons can have disconnected
4702: screen-hypersurfaces. This occurs, for example, in the collapse of a
4703: star to form a black hole (Bousso, 1999b). Consider light cones
4704: centered at $r=0$. The past light cones are all maximal on ${\cal
4705: I}^-$. The future light cones are maximal on ${\cal I}^+$ only if
4706: they start outside the event horizon. Future light cones from points
4707: inside the black hole are maximal on an apparent horizon in the black
4708: hole interior. Thus, there is one screen in the past, but two
4709: disconnected screens in the future.
4710:
4711: These two features may be related to black hole complementarity
4712: (Sec.~\ref{sec-limitations}), which suggests that the choice of an
4713: observer (i.e., a causally connected region) is a kind of gauge choice
4714: in quantum gravity. Related questions have recently been raised in
4715: the context of de~Sitter space, where black hole complementarity
4716: suggests a restriction to one causal region (Sec.~\ref{sec-ds}). They
4717: also play a central role in the framework for a holographic theory of
4718: cosmology pursued by Banks (2000c) and Banks and Fischler (2001a,b).}
4719:
4720: \item{The area of the maximal surface generically varies from cone to
4721: cone: $A(t) \neq \mbox{const}$. For example, the area of the apparent
4722: horizon in a flat FRW universe vanishes at the big bang and increases
4723: monotonically, diverging for late-time cones (Fig.~\ref{fig-flatfrw}).
4724: In a closed FRW universe, the area of the apparent horizon increases
4725: while the universe expands and decreases during the collapsing phase
4726: (Fig.~\ref{fig-clos}b).
4727:
4728: This behavior poses a challenge, because it would seem that the number
4729: of degrees of freedom of a holographic theory can vary with
4730: time.\footnote{Strominger (2001b) has recently suggested that the
4731: growth of a screen might be understood as inverse RG flow in a dual
4732: field theory.} The shrinking of a screen raises concerns about a
4733: conflict with the second law (Kaloper and Linde, 1999). However, the
4734: following observation suggests that the parameter $t$ should not be
4735: uncritically given a temporal interpretation on a screen
4736: hypersurface.}
4737:
4738: \item{The maximal surfaces do not necessarily form timelike
4739: (i.e. Lorentzian signature) hypersurfaces. In de Sitter space, for
4740: example, the global screens are the two conformal spheres at past and
4741: future infinite time. Both of these screens have Euclidean
4742: signature.\footnote{This does not mean that holography reduces to
4743: ordinary Cauchy evolution. Holographic encoding does not make use of
4744: equations of motion. There is always a projection, slice by slice, of
4745: holographic data onto the screen. Moreover, the limit of 1 bit per
4746: Planck area, central to holography, plays no role in Cauchy
4747: evolution.} The same is true for the apparent horizons in spacetimes
4748: with a $w>1/3$ equation of state (Fig.~\ref{fig-flatfrw}).}
4749:
4750: \item{Screens can be located in the spacetime interior. Screens near
4751: the boundary have the advantage that metric perturbations and quantum
4752: fluctuations fall off in a controlled way. The common large distance
4753: structure of different asymptotically Anti-de Sitter spacetimes, for
4754: example, makes it possible to describe a whole class of universes as
4755: different states in the same theory.
4756:
4757: The shape of interior screens, on the other hand, is affected by small
4758: variations of the spacetime. The apparent horizon in cosmological
4759: solutions, for example, will depend on the details of the matter
4760: distribution. Thus it is not clear how to group cosmological
4761: spacetimes into related classes (see, however, Sec.~\ref{sec-ds}).}
4762:
4763: \end{itemize}
4764:
4765: The AdS/CFT correspondence realizes the holographic principle
4766: explicitly in a quantum gravity theory. The points just mentioned
4767: show that, intricate though it may be, this success benefits from
4768: serendipidous simplifications. In more general spacetimes, it remains
4769: unclear how the holographic principle can be made manifest through a
4770: theory with explicitly holographic degrees of freedom. In particular,
4771: one can argue that the screen should not be presumed; all information
4772: about the geometry should come out of the theory itself.
4773:
4774: Nevertheless, the existence of global holographic screens in general
4775: spacetimes is an encouraging result. It demonstrates that there is
4776: always a way of projecting holographic data, and it provides novel
4777: structures. The understanding of their significance remains an
4778: important challenge.
4779:
4780:
4781: \subsection{Towards a holographic theory}
4782: \label{sec-toe}
4783:
4784: We have convinced ourselves of a universal relation between areas,
4785: light-sheets, and information. The holographic principle instructs us
4786: to embed this relation in a suitable quantum theory of gravity. It
4787: suggests that null hypersurfaces, and possibly global screens, will be
4788: given a special role in the regime where classical geometry emerges.
4789: How far have we come in this endeavor?
4790:
4791: The extent to which holography is explicit in string theory and
4792: related frameworks has been discussed in Secs.~\ref{sec-strings} and
4793: \ref{sec-ads}. We have also mentioned the local approach being
4794: developed by 't~Hooft (Sec.~\ref{sec-imp}).
4795:
4796: An effectively lower-dimensional description is evident in the quantum
4797: gravity of $2+1$ dimensional spacetimes (Witten, 1988; see also van
4798: Nieuwenhuizen, 1985; Achucarro and Townsend, 1986; Brown and Henneaux,
4799: 1986. See Carlip, 1995, for a review of $2+1$ gravity). As in the
4800: light cone formulation of string theory, however, the entropy bound is
4801: not manifest. Ho\v{r}ava (1999) has proposed a Chern-Simons
4802: formulation of (eleven-dimensional) M-theory, arguing that the
4803: holographic entropy bound is thus implemented. Light-like directions
4804: do not appear to play a special role in present Chern-Simons
4805: approaches.
4806:
4807: The importance of null hypersurfaces in holography resonates with the
4808: twistor approach to quantum gravity (see the review by Penrose and
4809: MacCallum, 1972), but this connection has not yet been substantiated.
4810: Jacobson (1995) has investigated how Einstein's equation can be
4811: recovered from the geometric entropy of local Rindler horizons.
4812: Markopoulou and Smolin (1998, 1999) have proposed to construct a
4813: manifestly holographic quantum theory of gravity based on the
4814: formalism of spin networks. Smolin (2001) discusses related
4815: approaches to an implementation of the holographic principle and
4816: provides further references.
4817:
4818: Banks (2000c) and Banks and Fischler (2001a,b) have sketched a
4819: preliminary framework for holographic theories of cosmological
4820: spacetimes. After discretizing time, one considers a network of
4821: screens obtained from a discrete family of observers. In other words,
4822: one constructs the past light cones of a discrete set of points spread
4823: throughout the spacetime. The maximal area on each light cone
4824: determines the dimension of a Hilbert space describing the enclosed
4825: portion of the spacetime. light cone intersections and inclusion
4826: relations give rise to a complicated network of Hilbert spaces, whose
4827: dimensions encode geometric information. A theory is sought which
4828: will give rise to spacetime geometry by inverting these steps. The
4829: rules for the generation of Hilbert space networks, and the
4830: construction of a suitable time evolution operator, are not yet
4831: understood.
4832:
4833: Banks and Fischler (2001a) have also argued that considerations of
4834: entropy determine the inital state of a big bang universe. By
4835: Eqs.~(\ref{eq-soaflat}) and (\ref{eq-frwcond}), maximally stiff
4836: matter, with equation of state $p=\rho$, has marginal properties under
4837: the holographic principle. This motivates a model based on the
4838: initial domination of a $p=\rho$ fluid, from which Banks and Fischler
4839: are aiming to obtain new perspectives on a number of standard
4840: cosmological problems.
4841:
4842: It has recently been noticed (Banks, 2000a; Fischler, 2000a,b) that
4843: the holographic principle has particularly strong implications in
4844: certain universes with a positive cosmological constant. As we
4845: discuss next, this could be of help in characterizing a holographic
4846: theory for a class of spacetimes that may include our universe.
4847:
4848:
4849: \subsection{Holography in de~Sitter space}
4850: \label{sec-ds}
4851:
4852: Generally the holographic principle restricts the number of degrees of
4853: freedom, $N$, only relative to some specified surface. There are
4854: spacetimes, however, where the holographic principle implies an
4855: absolute upper limit on $N$. This follows in particular if it is
4856: possible to find a global holographic screen whose area never exceeds
4857: $N$. Physically, there is not ``enough room'' in such universes to
4858: generate entropy greater than $N$. In particular, they cannot
4859: accommodate black holes with area greater than of order $N$.
4860:
4861: An absolute entropy bound could be viewed as a hint about
4862: characteristics of the quantum description of a whole class of
4863: spacetimes. The most radical conclusion would be to look for theories
4864: that come with only $e^N$ of states (Banks, 2000a; Bousso, 2000b;
4865: Fischler, 2000a,b; Dyson, Lindesay, and Susskind, 2002).\footnote{For
4866: a speculation on the origin of the number $N$, see Mena Marugan and
4867: Carneiro (2001).} This is quite unusual; even the Hilbert space of a
4868: single harmonic oscillator contains infinitely many states.
4869:
4870: If a continuous deformation of Cauchy data can take a universe with
4871: maximal entropy $N$ to one with $N'\neq N$, it is hard to argue that
4872: they should be described by two entirely different theories. Hence,
4873: this approach will be compelling only if physical criteria can be
4874: found which characterize a class of spacetimes with finite $N$,
4875: independently of initial data.
4876:
4877: As we discuss below, a suitable class may be the universes that become
4878: similar to de~Sitter space asymptotically in the future. However, we
4879: will not find this criterion entirely satisfactory. We will comment
4880: on its problems and possible generalizations.
4881:
4882:
4883: \subsubsection{de~Sitter space}
4884:
4885: The maximally symmetric spacetime with positive curvature is de~Sitter
4886: space. It is a solution to Einstein's equation with a positive
4887: cosmological constant, $\Lambda$, and no other matter. Using $w=-1$
4888: in Eqs.~(\ref{eq-FRW2}), (\ref{eq-a}), and (\ref{eq-q}), the metric
4889: can be written as a closed FRW universe,
4890: \begin{equation}
4891: ds^2 = \frac{a_0^2}{\sin^2 \eta}\left( -d\eta^2+ d\chi^2 +
4892: \sin^2\chi\, d\Omega_{D-2}^2\right).
4893: \label{eq-dsmetric}
4894: \end{equation}
4895: The curvature radius is related to the cosmological constant by
4896: \begin{equation}
4897: a_0^2 = {(D-1)(D-2)\over 2\Lambda}.
4898: \label{eq-alambda}
4899: \end{equation}
4900: For simplicity, we will take $D=4$ unless stated otherwise.
4901:
4902: The spatial three-spheres contract from infinite size to size $a_0$
4903: ($0<\eta\leq \pi/2$), then re-expand ($\pi/2\leq\eta<\pi$). The
4904: Penrose diagram is a square, with spacelike conformal boundaries at
4905: $\eta=0$, $\pi$. A light ray emitted on the north pole ($\chi=0$) at
4906: early times ($\eta\ll 1$) barely fails to reach the south pole
4907: ($\chi=\pi$) in the infinite future (Fig.~\ref{fig-ds}a).
4908:
4909: The light rays at $\eta=\chi$ reach neither the north nor the south
4910: pole in finite affine time. They generate a null hypersurface $H$, of
4911: constant cross-sectional area. (All spatial sections of $H$ are
4912: spheres of radius $a_0$.) $H$ is the future event horizon of an
4913: observer at the south pole. It bounds the region from which signals
4914: can reach the observer. There is a past event horizon
4915: ($\eta=\pi-\chi$) which bounds the region to which the southern
4916: observer can send a signal.
4917: %%***************************************************************
4918: \begin{figure}[h] \centering
4919: \includegraphics[width=7cm]{fig-ds}
4920: \caption{(a) Penrose diagram for empty de~Sitter space. $H$ is the
4921: future event horizon of an observer on the south pole ($\chi=\pi$).
4922: The shaded region is the ``southern diamond''. (b) Penrose diagram
4923: for a generic solution that asymptotes to de~Sitter in the past and
4924: future (dS$^\pm$). The future event horizon has complete time slices
4925: in its past, such as $Y$.}
4926: \label{fig-ds}
4927: \end{figure}
4928: %%***************************************************************
4929:
4930: The intersection of both regions forms the ``southern diamond'', the
4931: region that can be probed by the observer. It is covered by a static
4932: coordinate system:\footnote{The coordinates $r$ and $t$ defined here
4933: differ from those defined at the beginning of Sec.~\ref{sec-cosmo}.}
4934: \begin{equation}
4935: ds^2 = a_0^2 \left[-(1-r^2) dt^2 + {dr^2\over 1-r^2}
4936: + r^2 d\Omega^2 \right].
4937: \end{equation}
4938:
4939: Note that the location of event horizons in de~Sitter space depends on
4940: a choice of observer ($r=0$). Despite this difference, Gibbons and
4941: Hawking (1977) showed that the future event horizon of de~Sitter space
4942: shares many properties with the event horizons of black holes.
4943: Classically, objects that fall across the event horizon cannot be
4944: recovered. This would seem to endanger the second law of
4945: thermodynamics, in the sense discussed in Sec.~\ref{sec-gsl}.
4946:
4947: Mirroring the reasoning of Sec.~\ref{sec-gsl}, one concludes that the
4948: horizon must be assigned a semi-classical Bekenstein-Hawking entropy
4949: equal to a quarter of its area,
4950: \begin{equation}
4951: S_{\rm dS} = \pi a_0^2 = \frac{3\pi}{\Lambda}.
4952: \label{eq-dsvac}
4953: \end{equation}
4954: Gibbons and Hawking (1977) showed that an observer in de~Sitter space
4955: will detect thermal radiation coming from the horizon, at a
4956: temperature $T=1/2\pi a_0$.\footnote{See the end of
4957: Sec.~\ref{sec-unruh} for references to Bekenstein and Unruh-Wald
4958: bounds arising in de~Sitter space.}
4959:
4960: In pure de~Sitter space, there is no matter entropy, so the total
4961: entropy is given by Eq.~(\ref{eq-dsvac}).
4962:
4963:
4964: \subsubsection{dS$^\pm$ spacetimes}
4965:
4966: So far we have discussed empty de Sitter space. Generally one is
4967: interested in describing a larger class of spacetimes, which might be
4968: characterized by asymptotic conditions. Let us consider spacetimes
4969: that approach de~Sitter space asymptotically both in the past and in
4970: the future. We denote this class by dS$^\pm$. Its quantum
4971: description has recently attracted much attention (e.g.,
4972: Balasubramanian, Ho\v{r}ava, and Minic, 2001; Strominger, 2001a;
4973: Witten, 2001; for extensive lists of references, see, e.g.,
4974: Balasubramanian, de Boer, and Minic, 2001; Spradlin and Volovich,
4975: 2001). Implications of the holographic principle in other
4976: accelerating universes have been considered by Hellerman, Kaloper, and
4977: Susskind (2001) and Fischler {\em et al.} (2001); see also Banks and
4978: Dine (2001), Carneiro da Cunha (2002).
4979:
4980: If de~Sitter space is not completely empty, the Penrose diagram will
4981: be deformed. In the asymptotic regions matter is diluted, but in the
4982: interior of the spacetime it can have significant density. Gao and
4983: Wald (2000) showed under generic assumptions%
4984: %
4985: \footnote{Among other technical requirements,
4986: the spacetime must be geodesically complete and non-empty. Strictly,
4987: the presence of both asymptotic regions is not sufficient to guarantee
4988: geodesic completeness, because black holes can form. One would not
4989: expect the geodesic incompleteness due to black hole singularities to
4990: invalidate the above conclusions, however.}
4991: %
4992: that the back-reaction of matter makes the height of the diagram
4993: greater than its width. Then the future event horizon will cross the
4994: entire space and converge in the north (Fig.~\ref{fig-ds}b). Because
4995: the spacetime approaches empty de~Sitter space in the future, the
4996: horizon will asymptote to a surface $B$, a sphere of radius $a_0$
4997: surrounding the south pole. There will be no matter inside this
4998: sphere at late times. All matter will have passed through the future
4999: event horizon.
5000:
5001: The future event horizon can be regarded as a light-sheet of the
5002: surface $B$. This implies that the entropy of all matter on any
5003: earlier Cauchy slices cannot exceed a quarter of the area $A(B) = 4\pi
5004: a_0^2$. With Eq.~(\ref{eq-alambda}) we find that
5005: \begin{equation}
5006: S_{\rm global} \leq \frac{3\pi}{\Lambda}.
5007: \end{equation}
5008: In particular, this holds for the total entropy in the asymptotic
5009: past.
5010:
5011: We will not be concerned with the unobservable future region behind
5012: the event horizon. We conclude that {\em in a dS$^\pm$ spacetime, the
5013: global entropy cannot exceed ($3\pi$ times) the inverse cosmological
5014: constant.}
5015:
5016: This may seem a surprising result, since the initial equal-time slices
5017: can be taken arbitrarily large, and an arbitrary amount of entropy can
5018: be placed on them. However, if the matter density becomes larger than
5019: the energy density of the cosmological constant during the collapsing
5020: phase, the universe will collapse to a big crunch. Then there will be
5021: no future infinity, in contradiction to our assumption.
5022:
5023:
5024:
5025: \subsubsection{dS$^+$ spacetimes}
5026:
5027: An even larger class of spacetimes is characterized by the condition
5028: that they approach de~Sitter space in the asymptotic future. No
5029: restrictions are made on the behavior in the past. This class will be
5030: labelled dS$^+$. In addition to all of the dS$^\pm$ universes, it
5031: includes, for example, flat FRW universes that start with a big bang
5032: singularity and are dominated by matter or radiation for some time.
5033: At late times, all matter is diluted, only a cosmological constant
5034: remains, and the metric approaches that of empty de~Sitter space.
5035:
5036: Recent astronomical data (Riess {\em et al.}, 1998; Perlmutter {\em et
5037: al.}, 1999) favor a non-zero value of $\Lambda\sim 10^{-120}$. If
5038: this really corresponds to a fixed cosmological constant, our own
5039: universe is in the dS$^+$ class. This makes the study of
5040: de~Sitter-like spacetimes, in particular the dS$^+$ class of
5041: universes, especially significant.
5042:
5043: The global entropy at early times is unbounded in this class. In the
5044: dS$^\pm$ class, constraints arise, roughly speaking, because all
5045: matter has to ``fit'' through a throat three-sphere at $\eta=\pi/2$.
5046: In the dS$^+$ class, there is no need for a contracting phase. The
5047: universe can be everywhere expanding, with non-compact spacelike
5048: hypersurfaces of infinite total entropy.
5049:
5050: However, an observer's vision is cloaked by the de~Sitter event
5051: horizon that forms at late times. Let us ask only how much entropy
5052: can be detected by any single observer (Banks, 2000a; Fischler,
5053: 2000a,b). This is easy to answer because the final entropy is known.
5054: At late times, there is no matter and only a de~Sitter event horizon,
5055: so the total entropy will be given by Eq.~(\ref{eq-dsvac}). By the
5056: generalized second law of thermodynamics, the entropy at all other
5057: times will be less or equal.
5058:
5059: It follows that {\em in a dS$^+$ spacetime, the entropy available to
5060: any observer cannot exceed ($3\pi$ times) the inverse cosmological
5061: constant.} The restriction to a single observer is natural in view of
5062: black hole complementarity (Sec.~\ref{sec-bhc}).
5063:
5064:
5065: \subsubsection{Other universes with positive $\Lambda$}
5066:
5067: Although they comprise a broad class, it still somewhat unnatural to
5068: restrict one's attention to dS$^+$ universes. Because of exposure
5069: to thermal radiation, an observer in de~Sitter space cannot last
5070: forever. It is as unphysical to talk about arbitrarily long times as
5071: it is to compare the observations of causally disconnected observers.
5072:
5073: Moreover, fluctuations in the Gibbons-Hawking radiation cause black
5074: holes to form. If they are too big, they can cause a big crunch---a
5075: collapse of the entire spacetime. But even the persistent production
5076: of ordinary black holes means that any observer who is not otherwise
5077: thermalized will fall into a black hole. In short, quantum effects
5078: will prevent any observer from reaching ${\cal I}^+$.
5079:
5080: So how can spacetimes with an absolute entropy bound be usefully
5081: characterized? With assumptions involving spherical symmetry, the
5082: covariant entropy bound implies that that the observable entropy in
5083: any universe with $\Lambda>0$ is bounded by $3\pi/\Lambda$ (Bousso,
5084: 2000b). In addition to all dS$^+$ spacetimes, this class includes,
5085: for example, closed recollapsing FRW universes in which the
5086: cosmological constant is subdominant at all times. This result relies
5087: on the ``causal diamond'' definition of an observable region. It
5088: would seem to suggest that $\Lambda>0$ may be a sufficient condition
5089: for the absolute entropy bound, $S\leq 3\pi/\Lambda$.
5090:
5091: At least in $D>4$, however, one can construct product manifolds with
5092: fluxes, which admit entropy greater than that of $D$-dimensional
5093: de~Sitter space with the same cosmological constant (Bousso, DeWolfe,
5094: and Myers, 2002). A fully satisfactory classification of spacetimes
5095: with finite entropy remains an outstanding problem.
5096:
5097: \section*{Acknowledgments}
5098:
5099: I would like to thank S.~Adler, M.~Aganagic, T.~Banks, J.~Bekenstein,
5100: W.~Fischler, E.~Flanagan, S.~Hughes, T.~Jacobson, D.~Marolf, R.~Myers,
5101: A.~Peet, J.~Polchinski, M.~Srednicki, and L.~Susskind for helpful
5102: comments on drafts of this text. This work was supported in part by
5103: the National Science Foundation under Grant No.\ PHY99-07949.
5104:
5105: \appendix \section{General relativity}
5106:
5107: In this Appendix, we summarize most of the geometric terminology that
5108: pervades this paper. No attempts at completeness and precision are
5109: made; in particular, we will ignore issues of smoothness. The
5110: textbooks of Hawking and Ellis (1973), Misner, Thorne, and Wheeler
5111: (1973), and Wald (1984) may be consulted for a more thorough
5112: discussion of this material.
5113:
5114: \paragraph{Metric, examples, and Einstein's equation}
5115:
5116: General relativity describes the world as a classical spacetime ${\cal
5117: M}$ with $D-1$ spatial dimensions and one time dimension.
5118: Mathematically, ${\cal M}$ is a manifold whose shape is described by a
5119: metric $g_{ab}$ of Lorentzian signature $(-,+,\ldots,+)$. In a
5120: coordinate system $(x^0,\ldots,x^{D-1})$, the invariant distance $ds$
5121: between infinitesimally neighboring points is given by
5122: \begin{equation}
5123: ds^2 = g_{ab}(x^0,\ldots,x^{D-1}) dx^a dx^b.
5124: \end{equation}
5125: Summation over like indices is always implied.
5126:
5127: For example, the flat spacetime of special relativity (Minkowski
5128: space) in $D=4$ has the metric
5129: \begin{eqnarray}
5130: ds^2 & = & -dt^2 + dx^2 + dy^2 + dz^2\\
5131: & = & -dt^2 + dr^2 + r^2 d\Omega^2
5132: \end{eqnarray}
5133: in Cartesian or spherical coordinates, respectively. A Schwarzschild
5134: black hole of mass $M$ is described by the metric
5135: \begin{equation}
5136: ds^2 = -\left(1-{2M\over r}\right) dt^2 + \left(1-{2M\over
5137: r}\right)^{-1} dr^2 + r^2 d\Omega^2.
5138: \end{equation}
5139: The black hole horizon, $r=2M$, is a regular hypersurface, though this
5140: is not explicit in these coordinates. There is a singularity at
5141: $r=0$.
5142:
5143: Einstein's equation,
5144: \begin{equation}
5145: G_{ab} = 8\pi T_{ab},
5146: \end{equation}
5147: relates the shape of space to its matter content. The Einstein
5148: tensor, $G_{ab}$, is a nonlinear construct involving the metric and
5149: its first and second partial derivatives. The stress tensor,
5150: $T_{ab}$, is discussed further below.
5151:
5152: \paragraph{Timelike, spacelike, and null curves}
5153:
5154: A curve is a map from (a portion of) $\mathbb R$ into ${\cal M}$. In
5155: a coordinate system it is defined by a set of functions
5156: $x^a(\lambda)$, $\lambda \in \mathbb R$. At each point the curve has
5157: a tangent vector, ${dx^a\over d\lambda}$.
5158:
5159: A vector $v^a$ pointing up or down in time is called {\em timelike}.
5160: It has negative norm, $g_{ab} v^a v^b <0$. Massive particles (such as
5161: observers) cannot attain or exceed the speed of light. They follow
5162: {\em timelike curves}, or {\em worldlines}, i.e., their tangent vector
5163: is everywhere timelike. A vector $k^a$ is called {\em null\/} or {\em
5164: light-like\/} if its norm vanishes. light rays follow {\em null
5165: curves\/} through spacetime; their tangent vector is everywhere null.
5166: {\em Spacelike vectors\/} have positive norm. Spacelike curves
5167: connect points that can be regarded as simultaneous (in some
5168: coordinate system). No physical object or information follows
5169: spacelike curves; this would require superluminal speed.
5170:
5171: \paragraph{Geodesic curves}
5172:
5173: Curves that are ``as straight as is possible'' in a given curved
5174: geometry are called {\em geodesics}. They satisfy the {\em geodesic
5175: equation},
5176: \begin{equation}
5177: {d^2x^a\over d\lambda^2} + \Gamma^a_{~bc} {dx^b\over d\lambda}
5178: {dx^c\over d\lambda} = \alpha {dx^a\over d\lambda}.
5179: \end{equation}
5180: (The {\em Christoffel symbols}, $\Gamma^a_{~bc}$, are obtained from
5181: the metric and its first derivatives.) Any geodesic can be
5182: reparametrized ($\lambda \to \lambda'$) so that $\alpha$ vanishes. A
5183: parameter with which $\alpha=0$ is called {\em affine}.
5184:
5185: Unless non-gravitational forces act, a massive particle follows a
5186: timelike geodesic. Similarly, light rays don't just follow any null
5187: curve; they generate a {\em null geodesic}. We use the terms
5188: ``light ray'' and ``null geodesic'' interchangeably.
5189:
5190: Two points are {\em timelike separated\/} if there exists a timelike
5191: curve connecting them. Then they can be regarded as subsequent events
5192: on an observer's worldline. Two points are {\em null separated\/} if
5193: they are connected only by a light ray. Two points are {\em spacelike
5194: separated\/} if it is impossible for any object or signal to travel
5195: from one point from the other, i.e., if they are connected only by
5196: spacelike curves.
5197:
5198: \paragraph{Visualization and light cones}
5199:
5200: In all depictions of spacetime geometry in this paper, the time
5201: direction goes up, and light rays travel at 45 degrees. The light
5202: rays emanating from a given event $P$ (e.g., when a bulb flashes) thus
5203: form a cone, the {\em future light cone}. light rays arriving at $P$
5204: from the past form the {\em past light cone\/} of $P$. They limit the
5205: spacetime regions that an observer at $P$ can send a signal to, or
5206: receive a signal from.
5207:
5208: Events that are timelike separated from $P$ are in the interior of the
5209: light cones. Null separated events are on one of the light cones, and
5210: spacelike separated events are outside the light cones. The worldline
5211: of a massive particle is always at an angle of less than 45 degrees
5212: with the vertical axis. A moment of time can be visualized as a
5213: horizontal plane.
5214:
5215: \paragraph{Surfaces and hypersurfaces}
5216:
5217: In this text, the term {\em surface\/} always denotes a $D-2$
5218: dimensional set of points, all of which are spacelike separated from
5219: each other. For example, a soap bubble at an instant of time is a
5220: surface. Its whole history in time, however, is not a surface.
5221:
5222: A {\em hypersurface\/} $H$ is a $D-1$ dimensional subset of the
5223: spacetime (with suitable smoothness conditions). $H$ has $D-1$
5224: linearly independent tangent vectors, and one normal vector, at every
5225: point. If the normal vector is everywhere timelike (null, spacelike),
5226: then $H$ is called a {\em spacelike (null, timelike) hypersurface}.
5227:
5228: Physically, a spacelike hypersurface can be interpreted as ``the world
5229: at some instant of time''; hence, it is also called a {\em
5230: hypersurface of equal time}, or simply, a {\em time slice}
5231: (Fig.~\ref{fig-spheb}). A timelike hypersurface can be interpreted as
5232: the history of a surface. A soap bubble, for example, inevitably
5233: moves forward in time. Each point on the bubble follows a timelike
5234: curve. Together, these curves form a timelike hypersurface.
5235:
5236: Null hypersurfaces play a central role in this review, because the
5237: holographic principle relates the area of a surface to the number of
5238: degrees of freedom on a light-sheet, and light-sheets are null
5239: hypersurfaces. If a soap bubble could travel at the speed of light,
5240: each point would follow a light ray. Together, the light rays would
5241: form a null hypersurface. A particularly simple example of a null
5242: hypersurface is a light cone.
5243:
5244: More generally, a null hypersurface is generated by the light rays
5245: orthogonal to a surface. This is discussed in detail in
5246: Sec.~\ref{sec-kin}. As before, ``null'' is borderline between
5247: ``spacelike'' and ``timelike''. This gives null hypersurfaces great
5248: rigidity; under small deformations, they lose their causal character.
5249: This is why any surface has only four orthogonal null hypersurfaces,
5250: but a continuous set of timelike or spacelike hypersurfaces.
5251:
5252: \paragraph{Penrose diagrams}
5253:
5254: Many spacetimes contain infinite distances in time, or in space, or
5255: both. They have four or more dimensions, and they are generally not
5256: flat. All of these features make it difficult to draw a spacetime on
5257: a piece of paper.
5258:
5259: However, often one is less interested in the details of a spacetime's
5260: shape than in global questions. Are there observers that can see the
5261: whole spacetime if they wait long enough? Are parts of the spacetime
5262: hidden behind horizons, unable to send signals to an asymptotic region
5263: (i.e., are there black holes)? Does the spacetime contain
5264: singularities, places where Einstein's equation predicts its own
5265: breakdown? If so, are they timelike, so that they can be probed, or
5266: spacelike, so that they lie entirely in the past or in the future?
5267:
5268: Penrose diagrams are two-dimensional figures that capture certain
5269: global features of a geometry while discarding some metric
5270: information. The ground rules are those of all spacetime diagrams:
5271: time goes up, and light rays travel at 45 degrees. An important new
5272: rule is that (almost) every point represents a sphere. This arises as
5273: follows.
5274:
5275: We assume that the spacetime ${\cal M}$ is at least approximately spherically
5276: symmetric. Then the only non-trivial coordinates are radius and time,
5277: which facilitates the representation in a planar diagram. Usually
5278: there is a vertical edge on one side of the diagram where the radius
5279: of spheres goes to zero. This edge is the worldline of the origin of
5280: the spherical coordinate system. All other points in the diagram
5281: represent $(D-2)$ spheres. (In a closed universe, the spheres shrink
5282: to zero size on two opposite poles, and the diagram will have two such
5283: edges. There are also universes where the spheres do not shrink to
5284: zero anywhere.)
5285:
5286: A {\em conformal transformation\/} takes the physical metric, $g_{ab}$,
5287: to an {\em unphysical metric}, $\tilde g_{ab}$:
5288: \begin{equation}
5289: g_{ab}\to \tilde g_{ab} = \Omega^2 g_{ab}.
5290: \end{equation}
5291: The conformal factor, $\Omega$, is a function on the spacetime
5292: manifold ${\cal M}$. The unphysical metric defines an unphysical
5293: spacetime $\tilde M$.
5294:
5295: A conformal transformation changes distances between points. However,
5296: it is easy to check that it preserves causal relations. Two points
5297: that are spacelike (null, timelike) separated in the spacetime ${\cal
5298: M}$ will have the same relation in the unphysical spacetime $\tilde
5299: {\cal M}$.
5300:
5301: Penrose diagrams exploit these properties. A Penrose diagram of
5302: ${\cal M}$ is really a picture of an unphysical spacetime $\tilde
5303: {\cal M}$ obtained by a suitable conformal transformation. The idea
5304: is to pick a transformation that will remove inconvenient aspects of
5305: the metric. The causal structure is guaranteed to survive. Here are
5306: two examples.
5307:
5308: A judicious choice of the function $\Omega$ will map asymptotic
5309: regions in ${\cal M}$, where distances diverge, to finite regions in
5310: $\tilde {\cal M}$. An explicit example is given by
5311: Eq.~(\ref{eq-dsmetric}). By dropping the overall conformal factor and
5312: suppressing the trivial directions along the $(D-2)$ sphere, one
5313: obtains the unphysical metric depicted in the Penrose diagram
5314: (Fig.~\ref{fig-ds}). The asymptotic infinities of de~Sitter space are
5315: thus shown to be spacelike. Moreover, the spacetime can now be
5316: represented by a finite diagram.
5317:
5318: A neighborhood of a singularity in the spacetime ${\cal M}$ can be
5319: ``blown up'' by the conformal factor, thus exposing the causal
5320: structure of the singularity. An example is the closed FRW universe,
5321: Eq.~(\ref{eq-FRW2}); let us take $w\geq 0$ in Eq.~(\ref{eq-q}) and
5322: (\ref{eq-a}). Again, the prefactor can be removed by a conformal
5323: transformation, which shows that the big bang and big crunch
5324: singularities are spacelike (Fig.~\ref{fig-clos}).
5325:
5326: Conformal transformations yielding Penrose diagrams of other
5327: spacetimes are found, e.g., in Hawking and Ellis (1973), and Wald
5328: (1984).
5329:
5330:
5331:
5332: \paragraph{Energy conditions}
5333:
5334: The stress tensor, $T_{ab}$, is assumed to satisfy certain conditions
5335: that are deemed physically reasonable. The {\em null energy
5336: conditon\/}\footnote{``null convergence condition'' in Hawking and
5337: Ellis (1973)} demands that
5338: \begin{equation}
5339: T_{ab} k^a k^b \geq 0 \mbox{~for all null vectors~} k^a.
5340: \label{eq-nec}
5341: \end{equation}
5342: This means that light rays are focussed, not anti-focussed, by
5343: matter (Sec.~\ref{sec-ray}).
5344: The {\em causal energy condition\/} is
5345: \begin{equation}
5346: T_{ab} v^b T^{ac} v_c \leq 0 \mbox{~for all timelike vectors~} v^a.
5347: \label{eq-cec}
5348: \end{equation}
5349: This means that energy cannot flow faster than the speed of light.
5350:
5351: In Sec.~\ref{sec-econds}, the null and causal conditions are both
5352: demanded to hold for any component of matter, in order to outline a
5353: classical, physically acceptable regime of spacetimes in which the
5354: covariant entropy bound is expected to hold. The {\em dominant energy
5355: condition\/} is somewhat stronger; it combines the causal energy
5356: condition, Eq.~(\ref{eq-cec}), with the {\em weak energy condition},
5357: \begin{equation}
5358: T_{ab} v^a v^b \geq 0 \mbox{~for all timelike vectors~} v^a.
5359: \end{equation}
5360:
5361: In cosmology and in many other situations, the stress tensor takes the
5362: form of a perfect fluid with energy density $\rho$ and pressure $p$:
5363: \begin{equation}
5364: T_{ab} = \rho u_a u_b + p (g_{ab} +u_a u_b),
5365: \end{equation}
5366: where the unit timelike vector field $u^a$ indicates the direction of
5367: flow. In a perfect fluid, the above energy conditions are equivalent
5368: to the following conditions on $p$ and $\rho$.
5369: \begin{eqnarray}
5370: \mbox{null e.~c.:~~} & \rho \geq -p \\
5371: \mbox{causal e.~c.:~~} & |\rho| \geq |p| \\
5372: \mbox{null and causal:~~} & |\rho| \geq |p| \mbox{~and} \\
5373: & \rho <0 \mbox{~only if~} \rho=-p \\
5374: \mbox{weak e.~c.:~~} & \rho \geq -p \mbox{~and~} \rho\geq 0 \\
5375: \mbox{dominant e.~c.:~~} & \rho \geq |p|
5376: \end{eqnarray}
5377: With the further assumption of a fixed equation of state, $p=w\rho$,
5378: conditions on $w$ can be derived.
5379:
5380:
5381: \bibliographystyle{myrmp} \bibliography{all}
5382:
5383:
5384: \end{document}
5385:
5386:
5387:
5388:
5389:
5390:
5391:
5392:
5393: