1: \input harvmac
2: \input epsf
3: \newcount\figno
4: \figno=0
5: \def\fig#1#2#3{
6: \par\begingroup\parindent=0pt\leftskip=1cm\rightskip=1cm
7: \parindent=0pt
8: \baselineskip=11pt
9: \global\advance\figno by 1
10: \midinsert
11: \epsfxsize=#3
12: \centerline{\epsfbox{#2}}
13: \vskip 12pt
14: {\bf Fig. \the\figno:} #1\par
15: \endinsert\endgroup\par
16: }
17: \def\figlabel#1{\xdef#1{\the\figno}}
18: \def\encadremath#1{\vbox{\hrule\hbox{\vrule\kern8pt\vbox{\kern8pt
19: \hbox{$\displaystyle #1$}\kern8pt}
20: \kern8pt\vrule}\hrule}}
21:
22: \overfullrule=0pt
23: \def\identity{{\rlap{1} \hskip 1.6pt \hbox{1}}}
24:
25:
26:
27:
28: \Title{UK/02-01,TIFR-TH/02-02}
29: {\centerline{Thermality in de Sitter and Holography}}
30: \smallskip
31: \centerline{Sumit R. Das
32: \foot{das@pa.uky.edu, das@theory.tifr.res.in}}
33: \smallskip
34: \centerline{{\it Department of Physics and Astronomy,}}
35: \centerline{\it University of Kentucky,
36: Lexington, KY 40506, U.S.A.}
37: \smallskip
38: \centerline{and}
39: \smallskip
40: \centerline{{\it Tata Institute of Fundamental Research,}}
41: \centerline{\it Homi Bhabha Road, Bombay 400 005, INDIA.}
42:
43: \bigskip
44: %\baselineskip 18pt
45:
46: Assuming the existence of a $dS/CFT$ correspondence we study the
47: holograms of sources moving along geodesics in the bulk by calculating
48: the one point functions they induce in the boundary theory. In
49: analogy with a similar study of uniformly accelerated sources in $AdS$
50: spacetime, we argue that comoving geodesic observers correspond to a
51: coordinate system on the boundary in which the one point function is
52: {\it constant}. For $dS_3$ we show that the conformal transformations
53: on the boundary which achieve this - when continued suitably to
54: Lorentzian signature - induce nontrivial Bogoliubov transformations
55: between modes, leading to a thermal spectrum. This may be regarded as
56: a holographic signature of thermality detected by bulk geodesic
57: observers.
58:
59:
60:
61: \medskip
62:
63: \noindent
64:
65: \Date{January 2002}
66:
67:
68: \def\scri{{\cal I}}
69: \def\vx{{\vec x}}
70: \def\cO{{\cal O}}
71: \def\cQ{{\cal Q}}
72: \def\ba{{\bf a}}
73: \def\bb{{\bf b}}
74: \def\tbb{{\tilde{\bf b}}}
75: \def\bPhi{{\bf \Phi}}
76: \def\vk{{\vec k}}
77: \def\bu{{\bar u}}
78:
79: \newsec{Introduction and summary}
80:
81: There are several situations where quantum fields on curved
82: space-times lead to thermal behavior \ref\birell{See e.g. N.D. Birell
83: and P.C.W. Davies, {\it Quantum fields in curved space} (Cambridge
84: University Press, 1982)}. The most dramatic example is Hawking
85: radiation from black holes \ref\hawking{S. Hawking, {\it Nature}
86: {\bf 248} (1974) 30; S. Hawking, {\it Comm. Math. Phys.} {\bf 43} (1975)
87: 199.}.
88: A second example is the thermal spectrum observed by uniformly
89: accelerated detectors in flat space (the Unruh effect)
90: \ref\unruh{W. Unruh, {\it Phys. Rev.} {\bf D14} (1976) 870.} -
91: which is intimately related to black hole
92: radiation. Unruh radiation has generalizations to other space-times,
93: e.g. anti-de-Sitter (AdS) spacetimes \ref\deser{S. Deser and O. Levin,
94: {\it Class. Quant. Gr.} {\bf 14} L163; S. Deser and O. Levin,
95: {\it Class. Quant. Gr.} {\bf 15} (1998) L85, {\tt hep-th/9806223};
96: S. Deser and O. Levin,
97: {\it Phys. Rev.} {\bf D59} (1999) 064004, {\tt hep-th/9809159}
98: ; T. Jacobson, {\it Class.Quant.Grav} {\bf 15} (1998)
99: 251, {\tt gr-qc/9709048}}. Another set of
100: examples are cosmological
101: space-times \birell, the most well-studied instance being de-Sitter (dS)
102: spacetime. In this spacetime any geodesic observer perceives the
103: invariant vacuum as a thermal state \ref\desitter{G.W. Gibbons and
104: S. Hawking, {\it Phys. Rev.} {\bf D15} (1977) 2738.}.
105:
106: While recent developments in string theory have thrown valuable light
107: on the microscopic origin of black hole radiation
108: \ref\blackhole{A. Strominger and C. Vafa, {\it Phys. Lett.} {\bf
109: B379} (1996) 99, {\tt hep-th/9601029}; C. Callan and J. Maldacena
110: {\it Nucl. Phys.} {\bf B472} (1996) 591, {\tt hep-th/9602043};
111: A. Dhar, G. Mandal and S.R. Wadia, {\it Phys. Lett.} {\bf B388} (1996)
112: 51, {\tt hep-th/9605234} ;
113: S.R. Das and S.D. Mathur, {\it Nucl. Phys.} {\bf B478} (1996) 561,
114: {\tt hep-th/9606185 };
115: J. Maldacena and A. Strominger, {\it Phys. Rev.} {\bf D55} (1997) 861,
116: {\tt hep-th/9609026}.}, we know very little about thermal behavior in
117: cosmological spacetimes. This note is a attempt to throw some light on
118: this important issue for de Sitter spacetimes.
119:
120: The microscopic origin of black hole thermodynamics led to a concrete
121: realization of the holographic principle \ref\holographic{
122: G. 't Hooft, in ``{\it Salamfest}'' (1993) 0284,
123: {\tt gr-qc/9310026}; L. Susskind,
124: {\it J. Math. Phys.} {\bf 36} (1995) 6377, {\tt hep-th/9409089}.} -
125: the $AdS/CFT$ correspondence \ref\adscft{J. Maldacena, {\it Adv. Theo. Math. Phys.} {\bf 2}
126: (1998) 231, {\tt hep-th/9711200}; S. Gubser, I. Klebanov and
127: A. Polyakov, {\it Phys. Lett.} {\bf B428} (1998) 105, {\tt
128: hep-th/ 9802109}; E. Witten, {\it Adv. Theo. Math. Phys.} {\bf
129: 2} (1998) 253, {\tt hep-th/9802150}.}.
130: Conversely $AdS/CFT$ duality has provided a physical
131: understanding of Hawking radiation and related phenomena in terms of
132: the dual field theory. If the holographic principle is correct in this
133: context,
134: quantum gravity in de Sitter space-time should have a holographic dual
135: which is some theory living on one of the spacelike boundaries ${\cal
136: I}^\pm$. Our experience with $AdS/CFT$ duality suggests that
137: understanding this dual theory would throw valuable light on bulk
138: behavior in de Sitter space-times.
139:
140: Unfortunately, we do not know how to obtain de Sitter space-time
141: from string theory in a fully satisfactory manner. In fact, under
142: some assumptions there appears to be a no-go theorem \ref\nogo{B. de Wit,
143: D. Smit and N,D, Hari Dass,
144: {\it Nucl.Phys.} {\bf B283} (1987) 165;J. Maldacena
145: Nucl.Phys.B283:165,1987
146: and C. Nunez, {\it Int.J.Mod.Phys.} {\bf A16} (2001) 822.},
147: though there have been several proposals in the past as well as in
148: recent years
149: \ref\attempts{C. Hull, {\it JHEP} 9807 (1998) 021, {\tt hep-th/9806146},
150: E. Silverstein, hep-th/0106209;
151: C.M. Hull, {\it JHEP 0111} (2001) 012, {\tt hep-th/0109213};
152: and {\it JHEP} {\bf 0111} (2001) 061, {\tt hep-th/0110048};
153: C. Madeiros, C. Hull and B. Spence, {\tt hep-th/0111190};
154: G. Gibbons and C. Hull, {\tt hep-th/0111072}.}
155: Nevertheless it is
156: important to figure out what would be the holographic signature of
157: bulk phenomena {\it assuming} that such a correspondence exists. In
158: this spirit \ref\balahorava{V. Balasubramanian, P. Horava and
159: D. Minic, JHEP {\bf 0105} (2001) 043, {\tt hep-th/0103171}}
160: \ref\witten{E. Witten, hep-th/0106109}
161: \ref\strom{A. Strominger, hep-th/0106113}
162: propose various versions of this correspondence, in direct analogy
163: with the $AdS/CFT$ correspondence. For other work in this direction,
164: see \ref\other{
165: M. Li, {\tt hep-th/0106184};
166: D. Klemm, {\tt hep-th/0106247};
167: S. Nojiri and S.D. Odinstov, {\tt hep-th/0106191, hep-th/0107134};
168: A. Tolley and N. Turok, {\tt hep-th/0108119};
169: T. Shiromizu, S. Ida and T. Torii, {\tt hep-th/0109057};
170: B. McInnes, {\tt hep-th/0110062};
171: A. Strominger, {\tt hep-th/0110087};
172: V. Balasubramanian, J. de Boer and D. Minic, {\tt hep-th/0110108};
173: B. Carniero de Cunha, {\tt hep-th/0110169};
174: R. Cai, Y. Myung and Y. Zhang, {\tt hep-th/0110234};
175: U. Danielsson, {\tt hep-th/0110265};
176: S. Ogushi, {\tt hep-th/0111008};
177: A. Petkou and G. Siopsis, {\tt hep-th/0111085};A.M. Ghezelbash and
178: R.B. Mann, {\it JHEP} {\bf 0201} (2002) 005, {\tt hep-th/0111217};
179: A.M. Ghezelbash, D. Ida, R.B. Mann, T. Shiromizu,
180: {\tt hep-th/0201004}.}.
181: General aspects of holography in de Sitter spaces are discussed in
182: \ref\boussobanks{R. Bousso, hep-th/0012052, JHEP {\bf 9906} (1999) 028
183: [{\tt hep-th/9906022}] and JHEP {\bf 0011} (2000) 038
184: [{\tt hep-th/0010252}]; T. Banks, {\tt hep-th/0007146}}.
185: While various aspects of these proposals
186: lead to interesting insights into the nature of the holographic
187: theory, a holographic explanation of the thermal behavior observed by
188: a geodesic observer is still a mystery.
189:
190: In this paper we throw some light on this issue. We use an earlier work
191: \ref\dz{S.R. Das and A. Zelnikov, {\it Phys. Rev.} {\bf D64} (2001) 104001
192: [{\tt hep-th/0104198}]}, which addressed a similar question
193: in AdS spacetimes. In the latter situation, {\it uniformly
194: accelerated} observers measure a thermal spectrum, provided the
195: acceleration exceeds a critical bound \deser. The purpose of \dz\ was
196: to ask : how does one understand this thermality in the holographic
197: theory ? To answer this question, \dz\ considered an external source
198: moving with a uniform acceleration. The source couples to one of the
199: supergravity fields in the bulk, e.g. the dilaton $\Phi$. According
200: to the standard $AdS/CFT$ correspondence the value of the field $\Phi$
201: produced by this source is equal to the one point function of the
202: operator dual to this field in the boundary CFT
203: \ref\bulkboundary{V. Balasubramanian, P. Kraus and A. Lawrence,
204: {\it Phys. Rev.} {\bf D59} (1999) 046003, {\tt hep-th/9805171};
205: V. Balasubramanian, P. Kraus, A. Lawrence and
206: S. Trivedi, {\it Phys. Rev.} {\bf D59} (1999) 104021,
207: {\tt hep-th/9808017}; T. Banks, G. Horowitz and
208: E. Martinec, {\tt hep-th/9808}; E. Keski-Vakkuri,
209: {\it Phys. Rev.} {\bf D59} (1999) 104001, {\tt hep-th/9808037};
210: U. Danielsson, E. Keski-Vakkuri and M. Kruczenski,
211: {\it JHEP} {\bf 9901} (1999) 002, {\tt hep-th/9812007}.},
212: \ref\dasghosh{S.R. Das and B. Ghosh, {\bf JHEP} {\bf 0006} (2000) 043,
213: [{\tt hep-th/0005007}]}. This provides a
214: ``hologram'' of the moving source. Consider such a source which moves
215: normal to the boundary, away from it. Generically, at some given time,
216: the hologram has a profile which is peaked at the point where the bulk
217: trajectory intersects the boundary, dying off away from it. The width
218: of this profile is related to the distance of the source from the
219: boundary in accordance with the IR/UV connection. Thus the profile is
220: time dependent, spreading as the source in the bulk moves deeper into
221: $AdS$ space. We now need to understand how to describe a {\it comoving} bulk
222: observer holographically. This may be done by performing a conformal
223: coordinate transformation {\it on the boundary} such that the
224: transformed one point function is {\it time independent}. Such a
225: coordinate transformation on the boundary would generically mix up
226: positive and negative frequency modes of any field in the boundary
227: CFT. In \dz\ it was found that such mixing occurs only when the bulk
228: acceleration exceeds the critical value. In fact the metric of the
229: boundary generically now becomes time-dependent, i.e. a cosmological
230: space-time. Some of the cosmologies obtained in this way are well known.
231: The final
232: result is that there is a holographic relationship between
233: acceleration radiation in the bulk and thermal behavior in cosmologies
234: defined on the boundary.
235:
236: In the following, we adopt the same strategy for geodesic trajectories
237: in de Sitter space. However, now there are crucial differences which
238: make the physical interpretation a bit confusing. The boundaries are
239: now space-like, $\scri^\pm$, and the dual theory is euclidean. In
240: planar coordinates, time evolution in the bulk maps into {\it
241: decreasing} radial distance on the boundary.
242: Furthermore, unlike the $AdS/CFT$ correspondence it is not yet clear
243: whether there is an operator correspondence in $dS/CFT$. In fact, as
244: we shall see below, an operator correspondence is not necessarily
245: equivalent to a correspondence between the bulk effective action and
246: the CFT free energy in the presence of sources. Nevertheless, in this
247: work we will assume an operator correspondence. Then one point
248: functions of dual operators are related to the value of the bulk field
249: on the boundary - apart from the standard factor involving a UV
250: cutoff.
251:
252: We consider a source for some scalar field $\Phi$ of mass $m < 1$ (in
253: units where the de Sitter scale is set to unity) moving along a
254: geodesic in three dimensional de Sitter space. We calculate the value
255: of the field on a cutoff boundary, and hence the one point function of
256: the dual operator, on the boundary $\scri^+$. While the field $\Phi$
257: is a scalar, the definition of the cutoff boundary, and hence the one
258: point function, depends on the coordinate system used. In planar
259: coordinates, the one point function peaks at the point where the
260: geodesic intersects $\scri^+$ and decays as a power of the radial
261: distance. In analogy with \dz\ we then ask whether there is a
262: coordinate transformation {\it on the boundary} which renders the
263: transformed one point function {\it constant} over the entire
264: boundary. This is indeed possible since the dual operator has
265: nontrivial conformal dimensions. Boundary (euclidean) correlation
266: functions which are single valued in planar coordinates are {\it
267: periodic} in one of the new coordinates, and can therefore be
268: interpreted as thermal Green's functions of a Lorentzian signature
269: theory. After an analytic continuation to Lorentzian signature, this
270: new coordinate system is the holographic interpretation of a comoving
271: observer along a geodesic. The coordinate transformation, when
272: analytically continued in this fashion, mixes positive and negative
273: frequency modes of fields in the boundary theory - leading to the
274: correct temperature. An entirely analogous story holds for holograms
275: in global coordinates as well.
276:
277: The coordinate transformation involved is once again a conformal
278: transformation and is in fact the restriction of the bulk
279: transformation which takes the planar or global coordinates to
280: ``static'' coordinates to the boundary. In fact, the field produced on a
281: cutoff boundary defined in terms of a ``static'' coordinate system is
282: constant. As emphasized above, while the field $\Phi$ is a scalar, the
283: one point function transforms nontrivially as a conformal field -
284: which explains the result. The final offshoot is that thermality in
285: the dual description appears because of a nontrivial Bogoliubov
286: transformation involved in the passage to the natural holographic analogs
287: of ``comoving'' bulk observers.
288:
289: While the above results pertain to $dS_3$ we believe that the picture
290: is similar for other $dS_d$, though we do not present explicit
291: computations.
292:
293: In this work we have used retarded Green's functions to determine the
294: field due to the geodesic source which is then identified with the
295: one point function in the boundary theory. While the rationale for this
296: is clear in the AdS/CFT correspondence, it is not so in the present
297: case. We comment, without explicit calculations, the relevance of
298: this issue to recent results which concern the behavior of geodesic
299: detectors in the bulk in the one parameter class of de Sitter invariant
300: states.
301:
302: Section 2 contains definitions of various coordinate systems used in
303: this work. In Section 3 we discuss an operator version of the dS/CFT
304: correspondence. Section 4 contains the calculation of the hologram
305: of a geodesic source in planar coordinates. Section 5 discusses
306: interpretations of the hologram in both planar and global coordinates
307: and possible extensions to $dS_d$ for $d \neq 3$. Section 6 contains
308: conclusions and comments.
309:
310: \newsec{Coordinate systems in de Sitter}
311:
312: Throughout the paper, all dimensional quantities are measured in
313: units of the de Sitter length scale.
314:
315: $d+1$ dimensional de-Sitter space is a hyerboloid in $d+2$ dimensional
316: flat space with signature $(-1,1,1 \cdots 1)$ defined by the equation
317: \eqn\aone{-(Y^0)^2 + (Y^1)^2 + \cdots (Y^{d+1})^2 = 1}
318: Various coordinate systems are given by different ways of solving this
319: equation.
320:
321: Global coordinates are defined by the embedding
322: \eqn\atwo{\eqalign{ Y^0 & = \tan T \cr
323: Y^1 & = \sec T ~\cos \theta_1 \cr
324: Y^2 & = \sec T ~\sin \theta_1 \cos \theta_2 \cr
325: & \cdots \cr
326: Y^{d+1} & = \sec T~\sin \theta_1 \cdots \sin \theta_d}}
327: where
328: \eqn\athree{\eqalign{& -{\pi\over 2} \leq T \leq {\pi \over 2} \cr
329: & 0 \leq \theta_i \leq \pi ~~~~~~i = (1, \cdots (d-1))\cr
330: & 0 \leq \theta_d \leq 2\pi}}
331: The de-Sitter metric is then
332: \eqn\athree{ds^2 = \sec^2 T[-dT^2 + d\Omega_d^2]}
333: where $d\Omega_d^2$ is the metric on a unit $S^d$ whose coordinates are
334: $\theta_1 \cdots \theta_d$. The penrose diagram may be drawn using
335: \athree\ directly, as shown in Fig. 1. The future infinity $\scri^+$ is
336: at $T = {\pi \over 2}$ while the past infinity $\scri^-$ is at
337: $T = -{\pi \over 2}$ The diagram supresses the angles $\theta_2 \cdots
338: \theta_{d+1}$ and north pole corresponds to $\theta_1 =0 $ while the
339: south pole is at $\theta_1 = \pi$.
340:
341: \fig{Penrose diagram of de Sitter space. The curved line is a constant
342: ${\hat t}$ surface.}
343: {desitter.eps}{3.8in}
344:
345:
346: Planar or steady-state coordinates which cover regions I and II in Fig. 1
347: are defined by
348: \eqn\afour{\eqalign{ Y^0 & = {1\over 2} e^{{\hat t}} \rho^2 + \sinh {\hat t} \cr
349: Y^1 & = {1\over 2} e^{{\hat t}} \rho^2 - \cosh {\hat t} \cr
350: Y^2 & = e^{\hat t} \rho \cos \theta_2 \cr
351: Y^3 & = e^{\hat t} \rho \sin \theta_2 \cos \theta_3 \cr
352: & \cdots \cr
353: Y^{d+1} & = e^{\hat t} \rho \sin \theta_2 \cdots \sin \theta_{d}}}
354: The angles $\theta_i$ in \afour\ are the same as in \atwo. $\rho$ is a
355: radial coordinate in $R^d$, $0 \leq \rho \leq \infty$
356: which is formed by $\rho$ and the $d-1$ angles
357: $\theta_2 \cdots \theta_d$. The metric now reads
358: \eqn\afive{ ds^2 = - d{\hat t}^2 + e^{2{\hat t}}(d\rho^2 +
359: \rho^2 d\Omega_{d-1}^2)}
360: It is sometimes convenient to introduce cartesian coordinates on the $R^d$
361: which we denote by $x^i$, and also introduce a time coordinate $y$
362: \eqn\afiveb{ y = e^{\hat t}}
363: in terms of these the metric becomes
364: \eqn\afivea{ds^2 = {1\over y^2}[-dy^2 + dx^i dx^j \delta_{ij}]}
365: Comparing \atwo\ and \afour\ it is easy to see that this coordinate system
366: covers only regions I and II. This is because \afour\ implies that
367: $Y^0 - Y^1 = e^{{\hat t}} > 0$, while from \atwo\ we get
368: \eqn\asix{Y^0 - Y^1 = 2 \sec T~\cos [{1\over 2}(T+{\pi \over 2}-\theta_1)]
369: \sin [{1\over 2}(T-{\pi \over 2}+\theta_1)]}
370: Since both $[{1\over 2}(T+{\pi \over 2}-\theta_1)]$ and
371: $[{1\over 2}(T-{\pi \over 2}+\theta_1)]$ range from $-{\pi \over 2}$ to
372: ${\pi\over 2}$, the first two factors in \asix\ are always positive, so that
373: the sign is determined by the sign of $[{1\over 2}(T-{\pi \over 2}+\theta_1)]$.
374: The latter is positive in regions I and II. Planar coordinates which cover
375: regions III and IV may be defined in an analogous fashion.
376:
377: A third coordinate system will be called ``static'' coordinates. In Region
378: I we have
379: \eqn\aseven{\eqalign{Y^0 & = {\sqrt{1-r^2}}~\sinh t \cr
380: Y^1 & = -{\sqrt{1-r^2}}~\cosh t \cr
381: Y^2 & = r~ \cos \theta_2 ~~~~~~~~~~~~~~~~~~~{\rm Region~I}\cr
382: Y^3 & = r~ \sin \theta_2 \cos \theta_3 \cr
383: & \cdots \cr
384: Y^{d+1} & = r~\sin \theta_2 \cdots \sin \theta_{d}}}
385: where $-\infty \leq t \leq \infty$ and $0 \leq r \leq 1$. The angles
386: $\theta_2 \cdots \theta_{d-1}$ are the same in \atwo\ and \afour.
387: The metric is
388: \eqn\aseven{ds^2 = -(1-r^2) dt^2 + {dr^2 \over 1-r^2} + r^2 d\Omega_{d-1}^2}
389: The south pole
390: is given by $ r = 0$ while the past and future horizons are given by
391: $r = 1$. The metric is time independent in this region, which is why
392: these are called static coordinates.
393:
394: In region II we have and $1 \leq r \leq \infty$
395: \eqn\aeight{\eqalign{Y^0 & = {\sqrt{r^2-1}}~\sinh t \cr
396: Y^1 & = -{\sqrt{r^2-1}}~\cosh t \cr
397: Y^2 & = r~ \cos \theta_2 ~~~~~~~~~~~~~~~~~~~{\rm Region~II}\cr
398: Y^3 & = r~ \sin \theta_2 \cos \theta_3 \cr
399: & \cdots \cr
400: Y^{d+1} & = r~\sin \theta_2 \cdots \sin \theta_{d}}}
401: The metric is again given by \aseven, but $r$ rather than $t$ is a timelike
402: coordinate in this region. Thus the metric is not stationary any more.
403: However we will retain the nomenclature ``static coordinates'' even in this
404: region.
405: The future infinity $\scri^+$ is given by $r = \infty$.
406: It is possible to introduce Kruskal coordinates which cover the entire
407: space-time. These are denoted by $(U,V, \theta_2 \cdots \theta_d)$ where
408: \eqn\anine{\eqalign{& r = {1+UV \over 1-UV}~~~~~~
409: t ={1\over 2} {\rm log}~(-{U\over V}) ~~~~~~{\rm Region~I}\cr
410: & r = {1+UV \over 1-UV}~~~~~~
411: t ={1\over 2} {\rm log}~({U\over V}) ~~~~~~{\rm Region~II}}}
412: Thus in region I $U > 0, V < 0$ while in region II $U,V > 0$.
413: In terms of these Kruskal coordinates we have in both regions I and II
414: \eqn\aten{\eqalign{Y^0 & = {U+V \over 1-UV} \cr
415: Y^1 & = {V-U \over 1-UV} \cr
416: Y^2 & = {1+UV \over 1-UV} \cos \theta_2 \cr
417: Y^3 & ={1+UV \over 1-UV} ~ \sin \theta_2 \cos \theta_3 \cr
418: & \cdots \cr
419: Y^{d+1} & = {1+UV \over 1-UV}~\sin \theta_2 \cdots \sin \theta_{d}}}
420:
421: \newsec{An operator $dS/CFT$ correspondence}
422:
423: According to the $dS/CFT$ correspondence, physics in the bulk of $dS_d$ has
424: a holographic dual which is a conformal field theory living on the boundary
425: of the space-time. In global coordinates the boundaries could be either
426: $\scri^+$ or $\scri^-$, but not both \strom.
427: For bulk fields which satisfy standard wave equations with two
428: derivatives, the boundary data are the values of the field
429: on $\scri^+$ and $\scri^-$. Equivalently, one considers the two
430: independent solutions of the equations of motion and specifies the
431: data in terms of these, as in \strom. From the point of view of
432: a formulation of the $dS/CFT$ correspondence which relates
433: the bulk effective action to the free energy
434: of the CFT in the presence of sources, {\it both} the solutions
435: have to be retained - in contrast to the $AdS/CFT$ correspondence. As
436: a result there are {\it two} dual CFT operators for each bulk field.
437:
438: In planar coordinates which cover regions I and II, and static
439: coordinates which cover region II, the boundary is at
440: $\scri^+$. However, since these coordinates do not cover the entire
441: space-time, one has to specify the values of various bulk fields along
442: the horizons. Equivalently one has to again consider {\it both} the
443: independent solutions of the equations of motion.
444:
445: In the $AdS/CFT$ correspondence there is an operator formulation
446: \bulkboundary. We assume that there is a similar operator version of the
447: $dS/CFT$ correspondence. We will spell out this operator
448: correspondence in planar coordinates, though similar considerations
449: are valid for global coordinates as well.
450:
451: Consider a massive scalar field $\Phi$ of mass $m$ in $dS_{d+1}$. We will
452: consider the case $m < d/2$ in $dS$ units.
453: The free Klein-Gordon equation is
454: \eqn\bone{ (\nabla^2 - m^2) \Phi = 0}
455: In planar coordinates given by \afivea, the two independent solutions may
456: be easily written down
457: \eqn\btwo{\eqalign{\Phi^{(1)}_k ({\vec x},y)
458: & = {1\over 2 (2\pi)^{d-1\over 2}}
459: y^{d/2}~H_\nu^{(1)}(|k|y)~e^{i \vk \cdot \vx} \cr
460: \Phi^{(2)}_k ({\vec x},y) & = {1\over 2 (2\pi)^{d-1\over 2}}
461: y^{d/2}~H_\nu^{(2)}(|k|y)~e^{i \vk \cdot \vx}}}
462: where $H_\nu^{(i)}$ are Hankel functions and
463: \eqn\bthree{\nu = +{\sqrt{(d/2)^2-m^2}}}
464:
465: The modes have been chosen so that they are complex conjugates of each
466: other and normalized according to the standard Klein Gordon inner
467: product. Therefore the mode expansion which defines the
468: creation/annihilation operators is
469: \eqn\bfour{\Phi({\vec x},y) = \int {d^d k \over (2\pi)^{d}}[
470: \Phi_k^{(1)} ({\vec x},y)~ {\bf a} (\vk) + {\rm h.c.}]}
471: The operators $\ba (k)$ satisfy
472: \eqn\bfoura{ [ \ba (\vk), \ba^\dagger (\vk ') ] = \delta^d (\vk - \vk ')}
473: The Fock vaccuum in these coordinates is then defined by
474: \eqn\bfive{ {\bf a} (\vk) |0> = 0~~~~~~~{\rm for~all}~\vk}
475: and the states are labelled as usual by the values of the momenta $\vk$.
476: Single
477: particle states are
478: \eqn\bsix{ |\vk> = {\bf a}^\dagger (\vk) |0>}
479: Consider the mode expansion \bfour\ close to $\scri^+$, i.e. $y = y_0
480: \rightarrow 0$.
481: The leading order result for the field operator is then, upto
482: numerical factors
483: \eqn\bseven{\Phi (\vx, y_0) \sim {y_0^{{d\over 2}-\nu}\over i\Gamma (1-\nu)
484: \sin~\pi\nu}\int d^d k~k^\nu~[\ba (\vk) - \ba^\dagger (-\vk)]
485: e^{i\vk \cdot \vx}}
486: where $k \equiv |\vk|$.
487: Thus we can define a boundary operator $\cO_- (k)$ by
488: \eqn\beight{ \cO_-(\vk) \equiv k^\nu[\ba (\vk) - \ba^\dagger (-\vk)]}
489: which is the fourier transform of some local operator $\cO_-(\vx)$ on the
490: boundary. The power of $y_0$ clearly indicates that the conformal dimension
491: of this operator is
492: \eqn\bnine{\Delta_- = {d\over 2} - \nu}
493: This is an operator form of $dS/CFT$ correspondence.
494:
495: Note that roughly {\it half} of the bulk operators are related to the
496: specific boundary operator which arises from restricting the field to
497: the boundary. The Hankel functions have two pieces $J_\nu (|k|y)$ and
498: $J_{-\nu}(|k|y)$. Near $\scri^+$ the latter dominates and the operator
499: which comes with it is what we have identified above. There is another
500: operator $\cO_+ (k)$, which comes with the other Bessel function
501: $J_\nu (|k|y)$ and is independent of $\cO_-(k)$. In position space
502: one has
503: \eqn\bninea{{\rm Lim}_{y_0 \rightarrow 0}~\Phi (y_0, \vx)
504: \sim (y_0)^{d/2-\nu} \cO_- (\vx) + \sim (y_0)^{d/2+\nu} \cO_+ (\vx)}
505: This is the operator
506: manifestation of the appearance of {\it two} dual operators for a
507: single bulk field, as observed in \strom.
508:
509: This is in sharp contrast with the situation in $AdS/CFT$ correspondence.
510: In Poincare coordinates of $AdS_{d+1}$
511: \eqn\bten{ds^2 = {1\over z^2} [-dt^2 + dz^2 + d \vx \cdot d \vx]}
512: the mode expansion of a massive scalar field is
513: \eqn\beleven{\Phi (z,x,t) \sim z^{d/2}\int_0^\infty d\alpha\int d^{d-1}k~
514: ({\alpha \over \omega})^{1/2}~J_\mu (\alpha z)[\bb (\vk,\alpha) e^{-i(\omega
515: t - \vk \cdot \vx)} + h.c.]}
516: where
517: \eqn\btwelve{\omega^2 = \vk^2 + \alpha^2~~~~~~~\mu= +{\sqrt{(d/2)^2+m^2}}}
518: The mode expansion involves only one of the independent solutions of the
519: Klein Gordon expansion since the other solution is not normalizable.
520: In this case, the field operator near the boundary $z = z_0 \rightarrow 0$
521: gets indentified with a local boundary operator $\cQ (x,t)$ by
522: \bulkboundary, \dasghosh\
523: \eqn\bthirteen{ {\rm Lim}_{z_0 \rightarrow 0}~\Phi (z_0,x,t) =
524: (z_0)^{\mu + d/2}~\cQ(x,t)}
525: where the fourier transform of $\cQ(x,t)$ on the boundary, $\cQ (k,\omega)$
526: is given in terms of the annihilation and creation operators in
527: \beleven by \dasghosh\
528: \eqn\bfourteen{\cQ (\omega, \vk) \sim (\omega^2 - k^2)^{\mu / 2}[
529: \theta (\omega)~\tbb (\omega, \vk) +
530: \theta (-\omega)~\tbb^\dagger (-\omega, - \vk)]}
531: where
532: \eqn\bfourteena{\tbb (\omega, \vk) = ({\omega \over \alpha})^{1/2}
533: \bb (\alpha, \vk)}
534: In this case {\it all} of the annihilation and creation operators of the bulk
535: field are necessary to construct the boundary operator. Consequently, for
536: a given bulk field there is only one dual operator.
537:
538: In the $AdS/CFT$ correspondence, the bulk and the boundary share the
539: same time. If there is a $dS/CFT$ correspondence, the dual theory is
540: euclidean. From the form of the metric it is clear that rescaling time
541: is equivalent to rescaling distances on $\scri^+$. In the above correspondence
542: we have used the momenta $\vk$ to label states of the CFT. This means
543: we are considering one of the coordinates on $\scri^+$, $x_2$ or $x_3$
544: as the euclidean time. A proper interpretation would however involve
545: a continuation to lorentzian signature on the boundary.
546:
547: Assuming that there is such a $dS/CFT$ correspondence, it is clear
548: that the correlation functions of boundary operators can be written in
549: terms of the Green's functions of the bulk fields. Such an assumption
550: has been made in e.g. \ref\bms{R. Bousso, A. Maloney and
551: A. Strominger, {\tt hep-th/ 0112218}} and \ref\sv{M. Spradlin and
552: A. Volovich, {\tt hep-th/0112223}}. However it is not clear how
553: this is related to a $dS/CFT$ correspondence based on the equality of
554: the effective action of the bulk theory and conformal field theory
555: free energy in the presence of sources. For example, \bms,\sv
556: show that an operator correspondence leads to different results for
557: CFT correlators for different members of the one parameter family of
558: de Sitter invariant vacua found in \ref\mot{E. Mottola, {\it
559: Phys. Rev.} {\bf D 31} (1985) 754; B. Allen, {\it Phys. Rev.} {\bf D
560: 32} (1985) 3136.}. However if these CFT correlators are calculated
561: according to the prescription of \strom, they are independent of the
562: value of the parameter. This is because the Green's functions
563: (satisfying the inhomogeneous equation) for different values of this
564: vacuum parameter differ from one another by solutions of the
565: homogeneous equation and this addition does not change the field
566: evaluated on the boundary according to the procedure of
567: \ref\giddings{S. Giddings, {\it Phys. Rev. Lett.} {\bf 83} (1999)
568: 2707 [{\tt hep-th/9903048}]}.
569:
570: In the following we will assume an operator correspondence. While we
571: have described this in the planar coordinate system, it is clear that
572: this can be done in global coordinates using the modes derived in
573: \ref\dsmodes{N.A. Chernikov and E.A. Tagirov, {\it Ann. Poinc. Ohys. Theor.}
574: {\bf A 9} (1968) 109; E.A. Tagirov, {\it Annals. Phys.} {\bf 76} (1973) 561;
575: R. Figari, R. Hoegh-Krohn and C. Nappi, {\it Comm. Math. Phys.} {\bf 44}
576: (1975) 265; H. Rumpf and H.K. Urbantke, {\it Ann. Phys.} {\bf 114} (1978)
577: 332; L. Abbott and S. Deser, {\it Nucl. Phys.} {\bf B 195} (1982) 76;
578: A.H. Hajmi and A. Ottewill, {\it Phys. Rev} {\bf D 30} (1984) 1733;
579: L. Ford, {\it Phys. Rev.} {\bf D 31} (1985) 710; B. Allen and A. Folacci,
580: {\it Phys. Rev.} {\bf D 35} (1987) 3331; C.J. Burgess, {\it Nucl. Phys.}
581: {\bf B 247} (1984) 533.} and \mot, or for any other coordinate system.
582:
583: \newsec{Holograms of Geodesics in $dS_3$}
584:
585: The simplest geodesic trajectory is the worldline of the south
586: pole. In planar coordinates this is described by $\rho = 0$ while in
587: static coordinates in region I this corresponds to $r = 0$. All other
588: geodesics are obtained from this by isometries of de Sitter space. It
589: is clear that the trajectory of any point of the spatial $S^d$ in
590: global coordinates is a geodesic. Similarly any point on the $R^d$ in
591: planar coordinates is a geodesic as well. Because of the maximal
592: symmetry of the space it is sufficient to consider the geodesic at the
593: south pole.
594:
595: Consider a source moving along the geodesic in $dS_3$ which couples to
596: a bulk scalar field $\Phi$ of mass $m$. We work in planar coordinates which
597: cover regions I and II of the Penrose diagram.
598: According to the previous section, the leading value \foot{We work in
599: the leading order of the semiclassical expansion of the bulk theory.} of the
600: one point function of the dual operator $\cO_- (\vx)$ in this state is
601: given in terms of the value of the scalar field by
602: \eqn\cone{<\cO_- (\vx) >
603: \sim {\rm Lim}_{y_0 \rightarrow}~(y_0)^{1-\nu}~\Phi (y_0, \vx)}
604: The field $\Phi (y, \vx)$ is produced by the source.
605: When the source is at $(y,\vx) = (y'(\lambda),0)$ where $\lambda$ is the
606: proper time along the geodesic, this is given by
607: \eqn\ctwo{\Phi (y, \vx) = \int d\lambda~ G_R (y,\vx; y'(\lambda),0)}
608: where $G_R(y,\vx;y',\vx')$ is the retarded Green's function with $y < y'$.
609: The latter is given by
610: \eqn\cthree{G_R(y,\vx;y',\vx ') = i \theta (y' - y)~
611: <0| [ \bPhi (y,\vx), \bPhi (y',\vx ')]|0>}
612: where $\bPhi$ denotes the field operator. This may be readily calculated
613: using the mode expansion \bfour. For $d = 3$ one has $\nu = {\sqrt{1-m^2}}$
614: so that $\nu < 1$ and non-integral. We can then use the expressions for
615: the Hankel functions
616: \eqn\cfour{\eqalign{H^{(1)}_\nu(z) & = {1\over i~\sin\pi\nu}[J_{-\nu}(z)
617: - e^{-i\pi\nu} J_\nu (z)]\cr
618: H^{(2)}(z) & = {1\over i~\sin\pi\nu}[e^{i\pi\nu} J_{\nu}(z)
619: - J_\nu (z)]}}
620: The result for $G_R$ is then
621: \eqn\cfive{G_R(y,\vx;y',\vx ')= {2\theta (y' - y)\over~
622: \sin \pi\nu}\int {d^2 k \over 16 \pi^4} (y y')
623: e^{i \vk \cdot (\vx - \vx')}[J_{-\nu}(|k|y)J_{\nu}(|k|y') -
624: J_{-\nu}(|k|y')J_{\nu}(|k|y)]}
625: Along the trajectory $\vx = 0$ the proper time interval $d\lambda$ is
626: related to the increment in the coordinate time $dy'$ by
627: \eqn\csix{ d\lambda = - {dy' \over y'}}
628: Combining \ctwo, \cfive\ and \csix\ we finally get
629: \eqn\cseven{ \Phi (y,\vx) \sim
630: \int_\infty^y {dy' \over y'}\int d^2 k ~(y y')e^{i\vk \cdot \vx}
631: [J_{-\nu}(|k|y)J_{\nu}(|k|y') - J_{-\nu}(|k|y')J_{\nu}(|k|y)]}
632: where we have ignored inessential constants.
633: The field $\Phi$ is of course a scalar. Therefore the expression in any
634: other coordinate system may be obtained by simply reexpressing the right
635: hand side of \cseven\ in terms of the new coordinates. Alternatively one
636: can start out with a mode decomposition in the new coordinates.
637:
638: To extract the hologram, i.e. the one point function of the dual operator
639: we have to take a limit of \cseven\ when $y = y_0 \rightarrow 0$. In this
640: limit the dominant contribution comes from the term $J_{-\nu}(|k|y)
641: J_{\nu}(|k|y')$ in \cseven, which behaves as $|k|^{-\nu}(y_0)^{-\nu}
642: J_{\nu}(|k|y')$. Peforming the angular integral in momentum space we get
643: the result
644: \eqn\ceight{{\rm Lim}_{y_0 \rightarrow 0}~\Phi(y_0,\vx)
645: = (y_0)^{1-\nu}\int_\infty^{y_0} dy''\int_0^\infty d|k|~|k|^{1-\nu}~J_0(|k|\rho)
646: J_\nu(|k|y'')}
647: where $\rho = |\vx|$ as before.
648: The integral over $y'$ may be performed using
649: \eqn\cnine{\int_\infty^{y_0} dy'~ J_\nu (|k| y')
650: = {1\over |k|} [-1 + 2 \sum_{n=0}^\infty J_{2n+1-\nu}(|k|y_0)]}
651: Since we have $\nu < 1$ we get
652: \eqn\cten{{\rm Lim}_{y_0 \rightarrow 0}~\Phi(y_0,\vx)
653: = (y_0)^{1-\nu}\int_0^\infty d|k|~|k|^{-\nu}~J_0(|k|\rho)
654: \sim ( {y_0 \over \rho} ) ^{1-\nu}}
655: which leads to a one point function
656: \eqn\celeven{< \cO_- (\vx)>_{planar} \sim {1\over \rho^{1-\nu}}}
657: The power is appropriate for that of an operator with dimensions
658: $(\half (1-\nu), \half (1-\nu))$ in the two dimensional euclidean CFT
659: on the boundary.
660:
661: \newsec{Interpreting the hologram}
662:
663: In the bulk, an observer comoving with the geodesic will perceive the
664: vacuum as a thermal state with a temperature $T = {1/2\pi}$ in our
665: units. We want to see how is this reflected in the holographic
666: dual.
667:
668: \subsec{Accelerated objects in $AdS_3$ (Poincare boundary)}
669:
670: Before we start to interpret the hologram of a geodesic source in $dS_3$ let
671: us recall the main results of a similar calculation of holograms of
672: {\it accelerated} objects in $AdS_3$ coupled to a bulk massless scalar field
673: \dz. Consider the following trajectory
674: \eqn\done{t = \alpha ~z}
675: in a Poincare coordinate system
676: \eqn\dones{ds^2 = {1\over z^2}[-dt^2 +dz^2 +dx^2]}
677: This has a uniform invariant acceleration $a$ given by
678: \eqn\dtwo{a^2 = {\alpha^2 \over \alpha^2 -1}}
679: The parameter $\alpha$ labels the specific trajectory. A set of observers
680: comoving with this class of objects records a Unruh temperature
681: $T_U = {1\over 2\pi}$. The local temperature measured by a particular
682: trajectory is related to $T_U$ by a redshift factor, leading to
683: $T = {1\over 2\pi{\sqrt{\alpha^2-1}}}$
684:
685: When such an accelerated object couples to a massless scalar field in the
686: bulk, the one point function of the dual operator in the boundary CFT defined
687: on a cutoff boundary at $z=z_B \rightarrow 0$ can
688: be calculated along the lines of the previous section, with the result
689: \dz\
690: \eqn\dthree{<\cO> \sim \alpha {\sqrt{\alpha^2 -1}}{t \over [(\alpha^2-1)
691: x^2 + t^2]^{3/2}}}
692: As the Poincare time $t$ increases the object moves deeper into the
693: bulk of $AdS$ spacetime. The hologram, given by \dthree, reflects this :
694: the support of the one point function spreads with time.
695:
696: Consider an observer {\it on the boundary} according to whom the
697: one point function is {\it time-independent}. This would be the hologram of
698: a bulk observer co-moving with the accelerated object. To find such observers
699: one must therefore look for a new coordinate system in which the one point
700: function is time independent. It turns out that this is in fact a
701: conformal transformation on the boundary. Recalling the fact that the
702: conformal dimensions of $\cO$ are $(1,1)$ it is easy to see that
703: the new time $\eta$ and the new
704: space $\psi$ are related to the original coordinates $(t,x)$ by
705: \eqn\dfour{t \pm x = {1\over \beta}e^{-\beta(\eta \mp \psi)}}
706: where $\beta$ is determined by requiring that the proper time interval
707: for any section of the trajectory is equal to the interval in terms of the
708: new time $\eta$,
709: \eqn\dfive{\beta = {1\over {\sqrt{\alpha^2-1}}}}
710: Then the transformed one point function is
711: \eqn\dsix{ <\cO_{\eta,\psi}> \sim \alpha {\sqrt{\alpha^2 -1}}
712: {\cosh \beta \psi \over [\alpha^2 \sinh^2 \beta\psi + 1]^{3/2}}}
713:
714: The transformation \dfour\ covers only one wedge of the full Minkowski
715: space $(t,x)$ - the wedge which corresponds to the future light cone of
716: the point $t=x=0$. The original boundary metric becomes
717: \eqn\dseven{e^{-2\beta \eta}[-d\eta^2 + d\psi^2]}
718: This is the metric of a Milne universe. It is well known that the
719: Minkowski vacuum appears as a thermal state in terms of particles defined
720: according to postive frequency using the time $\eta$, with a temperature
721: $T = {\beta \over 2\pi}$ \birell, \ref\milne{T. Tanaka and M. Sasaki,
722: {\it Phys. Rev.} {\bf D55} (1997) 6061.}
723: . This is exactly the bulk temperature.
724: The upshot is that acceleration radiation in the bulk is interpreted as
725: a {\it cosmological} radiation in the boundary theory.
726:
727: There is in fact a good reason why \dfour\ is the correct transformation.
728: To see this we consider a different coordinate system in $AdS_3$, which
729: we call BTZ coordinates. The spatial coordinates are $\rho$ with a
730: range $1 < \rho < \infty$ and $\psi$ with range $0 < \psi < \infty$
731: while the time coordinate $\eta$ has a range $-\infty < \eta < +\infty$.
732: The metric now reads
733: \eqn\dseven{ds^2 = - (\rho^2 -1) d\eta^2 + {d\rho^2 \over \rho^2 -1}
734: + \rho^2 d\psi^2}
735: The boundary is now at $\rho = \infty$, and the boundary coordinates on
736: a cutoff boundary $\rho = \rho_B$ are $(\eta,\psi)$. In these coordinates
737: the accelerated trajectory \done\ simply corresponds to
738: \eqn\deight{\rho = \alpha}
739: It is clear that if we calculate the one point function on a cutoff
740: boundary using these coordinates the result will be independent of the
741: time $\eta$.
742:
743: This explains why \dfour is the correct transformation.
744: The point is that the transformations \dfour\
745: are precisely the coordinate transformation between
746: the Poincare coordinates $(t,z,x)$ and BTZ coordinates $(\eta,\rho,\psi)$
747: when restricted to a cutoff boundary at $z = z_0$or equivalently $\rho
748: = {1\over z_0}$.
749:
750: It is important to realize that while $\Phi$ is scalar under coordinate
751: transformations, the defintion of the cutoff boundary depends on the
752: specific coordinate system used \foot{This feature has been also observed
753: in computations of the Casimir energy, see Ghezelbash et.al. in last
754: reference in \other.}. This makes the one point function
755: coordinate dependent - in fact it just transforms as a conformal field
756: with the appropriate conformal dimension.
757:
758: The treatment for other coordinates in $AdS$ is entirely analogous and
759: has been discussed in detail in \dz.
760:
761:
762: \subsec{Geodesics in $dS_3$ : Planar boundary}
763:
764: The dual theory for $dS_3$ is euclidean. It is clear from the form
765: of the metric that in planar coordinates time evolution in the bulk
766: maps into scale evolution on the boundary.
767: In terms of the complex coordinate
768: \eqn\eone{z = x^2 + i x^3 = \rho e^{i\theta_2}}
769: the scale is represented by $\rho$.
770:
771: In analogy with the case of accelerated objects in $AdS_3$ we therefore
772: ask : is there a conformal transformation
773: \eqn\etwo{ z \rightarrow w = w(z)}
774: which renders the one point
775: function \celeven\ independent of $|w|$ ? Because of the symmetry of the
776: problem this means that the transformed one point function is in fact
777: a constant.
778:
779: This is indeed possible. Using the fact that the operator $\cO_-$ has
780: dimensions $({\Delta_- \over 2},{ \Delta_-\over 2})$ where $\Delta_- =
781: 1 - \nu$ it is easy to see that the transformation is
782: \eqn\ethree{z = e^w}
783: Our discussion of a possible operator correspondence in planar
784: coordinates imply that we can label the states of this euclidean
785: theory by the spatial momenta of the bulk theory. In other words, one
786: of the coordinates $x^2$ or $x^3$, e.g. $x^3$ can be regarded as an
787: euclidean time. Analytic continuation in $x^3$ then provides one definition
788: of a quatum theory on the boundary.
789:
790: Upon analytic continuation $x^3 \rightarrow i x^3$
791: the one point function \celeven\ becomes
792: \eqn\eeone{< \cO_- (\vx)>_{planar} \sim {1\over [(x^2)^2 - (x^3)^2]^
793: {({1-\nu \over 2 })}}}
794: Now make the transformation to coordinates $(\sigma_1,\sigma_2)$
795: \eqn\eetwo{ x^2-x^3 = -e^{-(\sigma_2-\sigma_3)}~~~~~
796: x^2+x^3 = e^{-(\sigma_2+\sigma_3)}}
797: The transformed one point function is then independent of $\sigma_i$.
798: The transformation \eetwo\ is precisely the transformation between
799: Minkowski and Rindler coordinates in two dimensional flat space. As is well
800: known this induces a nontrivial Bogoliubov transformation between modes
801: and the Miknowski vacuum appears as a thermal state in terms of Rindler
802: particles with the temperature given by
803: \eqn\efour{T = {1\over 2\pi}.}
804:
805: For the special case ($AdS_3$) we are considering, there is another way to
806: understand this. Define
807: \eqn\ethreeb{ w = \sigma_1 + i \sigma_2}
808: It is then clear that correlation functions which are single valued in
809: $(x^2,x^3)$ would be periodic in $\sigma_2$. This means that in an
810: alternative definition of a quantum theory on the boundary in which
811: $\sigma_2$ is regarded as an euclidean time, these correlators would
812: be {\it thermal} with a temperature given by \efour.
813: This is exactly the temperature measured by a geodesic observer in the
814: bulk. Note this is not the usual way a field theory on a cylinder is
815: defined. As we will see later, this argument needs
816: a modification in
817: higher dimensions.
818:
819: Once again there is a good reason why this is the right conformal
820: transformation. The geodesic $\vx = 0$ corresponds to the point
821: $r = 0$ in the static coordinate system in the patch I. This coordinate
822: patch does not contain $\scri^+$, but the coordinate $r$ can be continued
823: to the coordinate $r$ in the static patch in region II containing $\scri^+$.
824: Our experience with $AdS$ leads us to expect that the one point function
825: in these coordinates is in fact a constant.
826:
827: Let us check this explicitly. Since $\Phi$ is a scalar field, all we
828: have to do is to find the coordinate transformation relating the
829: static and planar coordinates and express the expression for $\Phi$ at
830: some general point, equation \cseven, in terms of static
831: coordinates. It is important that we do this before we take any limit
832: which takes us to a boundary since cutoff boundaries in different
833: coordinate systems do not coincide.
834:
835: The coordinate transformations can be read off from \afour,\aseven,
836: \aeight\ and \aten. In Kruskal coordinates we have
837: \eqn\efive{ \rho = {1 + UV \over 2U}~~~~~~~y = {1-UV \over 2U}}
838: in both regions I and II, while in terms of the $(r,t)$ coordinates
839: we have in region I
840: \eqn\esix{ \rho = {r \over {\sqrt{1-r^2}}}~e^{-t}~~~~~~~~
841: y = {1 \over {\sqrt{1-r^2}}}~e^{-t}~~~~~~~~({\rm Region~I})}
842: while in region II
843: \eqn\esixa{ \rho = {r \over {\sqrt{r^2-1}}}~e^{-t}~~~~~~~~
844: y = {1 \over {\sqrt{r^2-1}}}~e^{-t}~~~~~~~~({\rm Region~I})}
845:
846: In \cseven, the point $(y,\vx)$ is in region II while the points labelled
847: by $y'$ are all in region I.
848: Since the trajectory has $r = r' = 0$ we have
849: $y' = e^{-t'}$. For a point $(r,t,\theta_2)$ in region II the field is
850: \eqn\eseven{ \eqalign{ \Phi (r,t,\theta_2)
851: = \int_{-\infty}^{t + \log~{\tilde r}}dt'\int
852: dk~k~d\theta~(e^{-t'}{1\over {\tilde r}}e^{-t})~
853: e^{ik{\tilde \rho} \cos\theta}~
854: [J_{-\nu} & ({|k|\over {\tilde r}} e^{-t})J_{\nu}(|k|e^{-t'})\cr
855: & - J_{-\nu}(|k|e^{-t'})J_{\nu}({|k|\over {\tilde r}}e^{-t})]}}
856: where we have defined
857: \eqn\eeight{ {\tilde r} = {\sqrt{r^2-1}}~~~~~~~
858: {\tilde \rho} = {r \over {\sqrt{r^2-1}}}~e^{-t}}
859: To determine the one point function in static coordinates we have to now
860: go to the boundary at $r = r_0 >> 1$. Then ${\tilde r} \sim r_0$ and
861: ${\tilde \rho} \sim e^{-t}$. Peforming (in order) the angular integration over
862: $\theta$, the integral over $t'$ and finally the integral over $|k|$ exactly
863: as in Section 4 it is easy to see that
864: \eqn\enine{
865: < \cO_- (t,\theta_2)>_{static} =
866: {\rm Lim}_{r_0 \rightarrow \infty} [(r_0)^{\nu - 1}~
867: \Phi(r,t,\theta_2)] \sim {\rm constant}}
868:
869: Indeed the restriction of the coordinate transformations \esixa\ to
870: $\scri^+$ is precisely the conformal transformation \ethree, with the
871: identifications
872: \eqn\eten{w = t + i\theta_2}
873: This explains why the conformal transformation \ethree\ render the one
874: point function constant.
875:
876: \subsec{Geodesics in $dS_3$ : Global boundary}
877:
878: The story is similar in global coordinates. Now the cutoff boundary
879: is at $T = T_0 = {\pi \over 2} - \epsilon$ with $\epsilon << 1$. The
880: transformation between the global coordinates and planar coordinates
881: of regions I and II are given by directly comparing the formulae in
882: Section 2,
883: \eqn\fone{\eqalign{& y = {\cos T \over \sin T - \cos \theta_1} \cr
884: & \rho = {\sin \theta \over \sin T - \cos \theta_1}}}
885: The worldline of the geodesic is now $\theta_1' = \pi$. The one
886: point function on the boundary may be now calculated by evaluating
887: the field in \cseven\ by substituting \fone\ and finally performing
888: the limit $\epsilon \rightarrow 0$. The final result is
889: \eqn\ftwo{
890: < \cO_- (\theta_1,\theta_2)>_{global} =
891: {\rm Lim}_{(T_0 \rightarrow {\pi \over 2})}
892: [ (\cos~T_0)^{\nu - 1}~
893: \Phi(T, \theta_1 ,\theta_2)] \sim ({1 \over \sin \theta_1})^{1-\nu}}
894:
895: To understand this result, consider the transformation between global
896: coordinates and static coordinates in Region II. These are
897: \eqn\fthree{r = \sec T~\sin \theta_1~~~~~~~~~
898: \tanh t = - {\rm cosec}~ T~\cos \theta_1}
899: On $\scri^+$ this becomes
900: \eqn\ffour{ \tanh t = -\cos \theta_1}
901: or equivalently
902: \eqn\ffive{ \tan {\theta_1 \over 2} = e^t}
903: In terms of standard complex coordinates on $S^2$
904: \eqn\fsix{ u = \tan {\theta_1 \over 2}~e^{i\theta_2}}
905: we therefore have the conformal transformation
906: \eqn\fseven{ u = e^w}
907: where $w$ has been defined in \eten. This is exactly the conformal
908: transformation between the planar complex coordinate $z$ and
909: $w$. Thus one might have expected that on the sphere we should have
910: a one point function
911: \eqn\feight{ ({1 \over u \bu})^{1-\nu \over 2} =
912: ({1\over \tan {\theta_1 \over 2}})^{1-\nu}}
913: which is not the same as \ftwo.
914:
915: However there is a Weyl anomaly here since the metric on the sphere is
916: given in terms of $u,\bu$ by
917: \eqn\fnine{ds^2 = {4 du d\bu\over (1 + u\bu)^2}}
918: This means that while performing the conformal transformation we must
919: account for this conformal factor which is not a product of a function
920: of $u$ and a function of $\bu$. This is exactly what has to be done to
921: calculate two point functions of operators on the sphere. Taking this
922: into account we should get
923: \eqn\fine{ ({(1 + u\bu)^2 \over u \bu})^{1-\nu \over 2}}
924: which is precisely \ftwo.
925:
926: \subsec{Higher dimensions}
927:
928: Many of the above considerations would generalize to other dimensions
929: as well. For similar reasons, one point functions measured on the boundary
930: defined in terms of static coordinates would be a constant. For planar
931: and global coordinates the transformations to static coordinates are
932: in fact exactly the ones given above for all dimensions, and so would
933: be their restrictions to the boundary. The transformation laws for
934: one point functions would be however different.
935:
936: For $AdS_{d+1}$ the boundary metric in static coordinates may be written as
937: \eqn\none{ds^2 = r_0^2[dt^2 + (1-\mu^2)d\phi^2 + {d\mu^2 \over 1-\mu^2}
938: + \mu^2 d\Omega_{d-3}^2]}
939: The normal way to interpret this theory would be to consider $t$ as the
940: euclidean time.
941: The previous discussion suggests that there is another way to interpret
942: the theory, viz via the analytic continuation
943: \eqn\ntwo{\phi = i\eta}
944: and regarding $\eta$ as the time.
945: Then the metric \none\ becomes,
946: \eqn\nthree{ds^2 = dt^2 - (1-\mu^2)d\eta^2 + {d\mu^2 \over 1-\mu^2}
947: + \mu^2 d\Omega_{d-3}^2]}
948: This is the metric on $dS_{d-1} \times R$. Constant $\mu$ observers
949: will perceive the invariant vacuum as a thermal bath. In fact one
950: often uses the reverse of this argument to understand why there is
951: thermality in de Sitter spacetimes from the bulk viewpoint. It is
952: interesting that the holographic signature of thermality gets related
953: to the question of thermality in de Sitter spacetimes again, albeit
954: in two less dimensions.
955:
956:
957:
958: \newsec{Conclusions}
959:
960: We have offered a {\it signature} of the thermal properties of geodesic
961: observers in the holographic theory. This is not quite an {\it explanation}.
962: However we believe that this insight will be useful in a proper holographic
963: understanding of cosmological spacetimes.
964:
965: One important assumption in our work is that the one point function of
966: a CFT operator is given by the value of the bulk field on the
967: boundary, with suitable powers of the cutoff stripped off. We
968: calculated the field on the boundary in a standard fashion using
969: retarded Green's functions. This is entirely analogous to treatments
970: in the AdS/CFT correspondence. In the AdS/CFT correspondence this is
971: almost forced upon us since the bulk and the boundary share the same
972: ``time'', and one would like to maintain causality. Things are less
973: clear in the present case. Here the boundary is either $\scri^+$ or
974: $\scri^-$. Our procedure gives the one point functions in the boundary
975: on $\scri^+$ and a zero value on $\scri^-$. Presumably to get one
976: point functions on $\scri^-$ one should use {\it advanced} Green's
977: functions in the bulk. Another possiblity is to use a symmetric
978: Green's function. In fact the latter is suggested by an interesting
979: result of \bms. These authors consider the one parameter class of
980: invariant vacua in the bulk \mot, or their planar coordinate analogs
981: \sv. They show that a bulk geodesic
982: observer would detect particles in all these vacua. However the
983: spectrum is thermal only for a preferred value of the parameter - the
984: one which leads to the analytic continuation of the euclidean bulk
985: vacuum. This result could be reconciled with our results if one uses a
986: symmetric Green's function to obtain the field in the presence of a
987: geodesic source, since while the retarded or advanced Green's function
988: does not depend on the vacuum parameter, the symmetric Green's
989: function does.
990:
991: Our holograms are obtained from the fields produced by point sources.
992: In a natural extension of terminology, this is a ${1\over N}$ effect.
993: In the $AdS/CFT$ correspondence one needs nonlocal operators to probe
994: local physics in the bulk \ref\sussk{J. Polchinski, L. Susskind and
995: N. Toumbas, {\it Phys.Rev.} {\bf D60} (1999) 084006,
996: {\tt hep-th/9903228}; N. Toumbas and L. Susskind,
997: {\it Phys.Rev.} {\bf D61} (2000) 044001,
998: {\tt hep-th/9909013}.}. Since we are probing the entire trajectory of
999: a particle one should have a better description in terms of these
1000: objects and this would be a leading effect. It would be interesting
1001: to see analogs of these in the $dS/CFT$ correspondence.
1002:
1003: Finally, the presence of an external source in our discussion is not
1004: quite natural. The proper formulation of the problem would be to
1005: consider a wavepacket made out of bulk fields whose center follows
1006: a geodesic, as in \bulkboundary. The tails of these wavepackets
1007: then provide the one point functions necessary for the hologram.
1008: However one has to redo the analysis of thermal behavior by
1009: ``comoving'' set of observers in the bulk. Because of the nontrivial
1010: profile of the wavepacket one would get rather complicated transformations
1011: in the boundary theory.
1012:
1013: \newsec{Acknowledgements}
1014:
1015: I would like to thank Sandip Trivedi for numerous discussions and
1016: collaboration in the early stages of this work. I also thank Samir
1017: Mathur, Shiraz Minwalla and Al Shapere for discussions on various
1018: aspects of this work and the referee for a helpful comment. This work
1019: is partially supported by DOE Grant No. DE-FG01-00ER45832.
1020:
1021:
1022: \listrefs
1023: \end
1024:
1025:
1026:
1027:
1028:
1029:
1030:
1031:
1032:
1033:
1034:
1035:
1036:
1037:
1038: