1:
2: \documentstyle[12pt]{article}
3: \input{epsf.tex}
4: \font\big = cmr12 scaled \magstep2
5: \font\twelverm = cmr12
6: \font\tbf = cmbx12
7:
8:
9: \begin{document}
10:
11:
12:
13: \begin{flushright} \begin{small}
14: DF/IST-4.2001 \\gr-qc/0105103
15: \end{small} \end{flushright}
16:
17:
18:
19:
20:
21:
22:
23: \vskip 0.5cm
24:
25:
26:
27:
28:
29:
30: \begin{center}
31: {\bf QUASI-NORMAL MODES \\
32: OF SCHWARZSCHILD ANTI-DE SITTER BLACK HOLES:\\
33: ELECTROMAGNETIC AND GRAVITATIONAL PERTURBATIONS
34: %Quasi-Normal Modes of Schwarzschild anti-de Sitter Black holes:
35: %electromagnetic and gravitational perturbations
36: } \\
37: \vskip 0.5cm
38: Vitor Cardoso\footnote{E-mail: vcardoso@fisica.ist.utl.pt} \\
39: \vskip 0.2cm
40: {\scriptsize CENTRA, Departamento de F\'{\i}sica,
41: Instituto Superior T\'ecnico,} \\
42:
43: {\scriptsize Av. Rovisco Pais 1, 1096 Lisboa, Portugal.} \\
44: \vskip 0.5cm
45:
46: Jos\'e P. S. Lemos\footnote{E-mail: lemos@kelvin.ist.utl.pt} \\
47: \vskip 0.2cm
48: {\scriptsize CENTRA, Departamento de F\'{\i}sica,
49: Instituto Superior T\'ecnico,} \\
50: {\scriptsize Av. Rovisco Pais 1, 1096 Lisboa, Portugal.}
51: %\&
52: %\\
53: %{\scriptsize Observat\'orio Nacional-MCT,} \\
54: %{\scriptsize Rua General Jos\'e Cristino 77, 20921
55: %Rio de Janeiro, Brazil.}
56: \end{center}
57:
58: \vskip 0.1cm
59: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
60:
61: \begin{abstract}
62: \noindent
63: We study the quasi-normal modes (QNM) of electromagnetic and
64: gravitational perturbations of a Schwarzschild black hole in an
65: asymptotically Anti-de Sitter (AdS) spacetime. Some of the
66: electromagnetic modes do not oscillate, they only decay, since they
67: have pure imaginary frequencies. The gravitational modes show
68: peculiar features: the odd and even gravitational perturbations no
69: longer have the same characteristic quasinormal frequencies. There is
70: a special mode for odd perturbations whose behavior differs completely
71: from the usual one in scalar and electromagnetic perturbation in an
72: AdS spacetime, but has a similar behavior to the Schwarzschild black
73: hole in an asymptotically flat spacetime: the imaginary part of the
74: frequency goes as $\frac{1}{r_+}$, where $r_+$ is the horizon radius.
75: We also investigate the small black hole limit showing that the
76: imaginary part of the frequency goes as $r_+^2$. These results are
77: important to the AdS/CFT conjecture since according to it the QNMs
78: describe the approach to equilibrium in the conformal field theory.
79: \strut
80: \newline
81: \end{abstract}
82:
83: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
84: \noindent
85: \section{ Introduction}
86: \vskip 3mm
87: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
88: QNMs of black holes play an important role in the study of the
89: dynamics outside black holes. They appear, for instance, when one
90: deals with the evolution of some field in the black hole spacetime, or
91: in black hole-black hole collision processes. Numerical simulations
92: ranging from the formation of a black hole in a gravitational collapse
93: \cite{Cunningham} to the collision of two black
94: holes \cite{Anninos} provide clear evidence that
95: no matter how one perturbs a black hole, its response will be
96: dominated by the QNMs. QNMs allow us not only to test the stability
97: of the event horizon against small perturbations, but also to probe
98: the black hole mass, electric charge and angular momentum, through
99: their characteristic waveform.
100:
101: A great deal of effort has been spent to calculate the QNMs and their
102: associated frequencies. New powerful methods, both analytical and
103: numerical have been developed. The main interest in these studies
104: is in the application to the analysis of the data from the gravitational
105: waves to be detected by the forthcoming gravitational wave
106: detectors. We refer the reader to \cite{Kokkotas,Andersson} for
107: reviews. In a different context, York \cite{York} tried to explain the
108: thermal quantum radiance of a Schwarzschild black hole in terms of
109: quantum zero-point fluctuations of zero mean in the QNMs.
110:
111:
112:
113:
114: All these previous works deal with
115: asymptotically flat spacetimes, but the recent
116: AdS/CFT correspondence conjecture \cite{maldacena}
117: makes the investigation of QNMs in anti-de
118: Sitter spacetimes more appealing. According to it, the black hole
119: corresponds to a thermal state in the conformal field theory, and the decay of
120: the test field in the black hole spacetime, corresponds to the decay of the perturbed state
121: in the CFT.
122: The dynamical timescale for the return to thermal equilibrium is very
123: hard to compute directly, but can be done relatively easily using the
124: AdS/CFT correspondence. Horowitz and Hubeny
125: \cite{Horo} (see also \cite{Horowitz}) began the study of QNMs
126: in AdS, by thoroughly investigating scalar perturbations in 4, 5
127: and 7 spacetime dimensions. Subsequently, Wang and et al
128: \cite{Wang1,Wang2} analyzed scalar QNMs in a Reissner-Nordst\"om AdS
129: geometry. Recently, Cardoso and Lemos \cite{Cardoso} found an exact
130: solution for the QNMs of scalar, electromagnetic and Weyl
131: perturbations of a BTZ black hole.
132: Another conjecture is related to the speculation \cite{Horo,Bir,Kim} that
133: there might be a connection between the critical exponent of Choptuik
134: \cite{Chop} and the imaginary part of the frequency, for small black
135: holes. This is still an open question.
136:
137: In this paper we shall go beyond the scalar perturbations
138: \cite{Horo,Wang1,Wang2}, and consider electromagnetic and
139: gravitational perturbations of a Schwarzschild black hole in an
140: asymptotically AdS spacetime.
141: Electromagnetic
142: perturbations are of interest due to the AdS/CFT conjecture since they can be seen
143: as perturbations for some generic supergravity gauge field.
144: In addition, the Maxwell field is an important field with different features
145: from scalar or gravitational fields, which makes it worth studying.
146: On the other hand, gravitational perturbations
147: have the additional interest of arising from any other type of
148: perturbation, be it scalar, electromagnetic, Weyl, etc., which in turn
149: disturb the background geometry. Therefore, questions like the
150: stability of spacetime for scalar or other perturbations, have a
151: direct dependence on the stability to gravitational perturbations.
152:
153: We will find that in the case of electromagnetic perturbations of
154: large black holes, the characteristic QNM frequencies have only an
155: imaginary part, and scale with the horizon radius. As for
156: gravitational perturbations, there are two novel features. First,
157: contrary to the asymptotically flat spacetime case, odd and even
158: perturbations no longer have the same spectra, although in certain
159: limits one can still prove that the frequencies are almost the same.
160: The second intriguing result is that, for odd perturbations, there is
161: a mode with a totally different behavior from that found in the
162: scalar and electromagnetic case: in this mode the frequency scales
163: with $\frac{1}{r_+}$, just as in asymptotically flat Schwarzschild
164: spacetime. We also investigate the small black hole limit (a problem
165: recently addressed by Zhu et al \cite{Wang3}), and find that the QNM
166: frequencies go as $r_+^2$.
167:
168:
169: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
170: %\vskip 0.5cm
171:
172: \noindent
173: \section{Electromagnetic and Gravitational perturbations in a
174: Schwarzschild AdS background}
175:
176: %\vskip 3mm
177: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
178:
179:
180: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
181: \subsection{Maxwell perturbations}
182: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
183: We consider the evolution of a Maxwell field in a
184: Schwarzschild-anti-de Sitter spacetime with metric given by
185: \begin{equation}
186: ds^{2}= f(r) dt^{2}- \frac{dr^{2}}{f(r)}-
187: r^{2}(d\theta^{2}+\sin^2\theta d\phi^{2})\,,
188: \label{lineelement}
189: \end{equation}
190: where, $f(r)=(\frac{ r^{2}}{R^2}+1-\frac{2M}{r})$,
191: $R$ is the AdS radius and $M$ the black hole mass.
192: The evolution is governed by Maxwell's equations:
193: \begin{equation}
194: {F^{\mu\nu}}_{;\nu}=0 \quad, F_{\mu\nu}=A_{\nu,\mu}-A_{\mu,\nu}\,,
195: \label{maxwell}
196: \end{equation}
197: where a comma stands for ordinary derivative and a semi-colon
198: for covariant derivative. As the background is spherically symmetric,
199: we can expand $A_{\mu}$ in 4-dimensional vector sphericall
200: harmonics (see \cite{Ruffini}):
201:
202: {\small
203: \begin{eqnarray}
204: A_{\mu}(t,r,\theta,\phi)=\sum_{l,m}\left( \begin{array}{cc}\left[
205: \begin{array}{c} 0 \\ 0 \\
206: \frac{a^{lm}(t,r)}{\sin\theta}\partial_\phi Y_{lm}\\
207: -a^{lm}(t,r)\sin\theta\partial_\theta Y_{lm}\end{array}\right] &
208: +\left[ \begin{array}{c}f^{lm}(t,r)Y_{lm}\\h^{lm}(t,r)Y_{lm} \\
209: k^{lm}(t,r) \partial_\theta Y_{lm}\\ k^{lm}(t,r) \partial_\phi
210: Y_{lm}\end{array}\right] \end{array}\right)\,,
211: \label{expansion}
212: \end{eqnarray}}
213:
214: \noindent where the first term in the right-hand side has parity $(-1)^{l+1}$
215: and the second term has parity $(-1)^{l}$, $m$ is the azimuthal number
216: and $l$ the angular quantum number. If we put this expansion into Maxwell's
217: equations (\ref{maxwell}) we get a second order differential
218: equation for the perturbation:
219: \begin{equation}
220: \frac{\partial^{2} \Psi(r)}{\partial r_*^{2}} +
221: \left\lbrack\omega^2-V(r)\right\rbrack\Psi(r)=0 \,,
222: \label{wavemaxwell}
223: \end{equation}
224: where the wavefunction $\Psi(r)$ is a linear combination of the functions
225: $f^{lm}$, $h^{lm}$, $k^{lm}$ and $a^{lm}$ as appearing in
226: (\ref{expansion}). $\Psi$ has a different functional dependence
227: according to the parity: for odd parity, i.e, $(-1)^{l+1}$, $\Psi$
228: is explicitly given by $\Psi=a^{lm}$ whereas for even parity $(-1)^l$
229: it is given by $\Psi=\frac{r^2}{l(l+1)}\left(-i\omega
230: h^{lm}-\frac{df^{lm}}{dr}\right)$, see \cite{Ruffini} for further details.
231: It is assumed that the time dependence is $\Psi(t,r)=e^{-i\omega t}\Psi(r)$.
232: The potential $V$ appearing in equation (\ref{wavemaxwell}) is given by
233: \begin{equation}
234: V(r)=f(r)\left\lbrack\frac{l(l+1)}{r^2}\right\rbrack \,,
235: \label{potentialmaxwell}
236: \end{equation}
237: and the tortoise coordinate $r_*$ is defined as
238: \begin{equation}
239: \frac{\partial r}{\partial r_*}= f(r)\,.
240: \end{equation}
241: We can of course rescale $r$, $r\rightarrow\frac{r}{R}$ and if we do
242: this, the wave equation again takes the form (\ref{wavemaxwell}) with
243: rescaled constants i.e., $r_+ \rightarrow \frac{r_+}{R}$, $\omega
244: \rightarrow \omega R $, where $r_+$ is the horizon radius. So, we can
245: take $R=1$ and measure everything in terms of $R$.
246: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
247: \subsection{ Gravitational perturbations}
248: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
249: When dealing with first order gravitational perturbations one supposes
250: that, at least in some restricted region of spacetime, the metric
251: functions can be written as
252: \begin{equation}
253: g_{ab}(x^\nu)= g^{(0)}_{ab}(x^\nu)+h_{ab}(x^\nu)\,,
254: \label{4.1}
255: \end{equation}
256: where the metric $g^{(0)}_{ab}(x^\nu)$ is the background metric,
257: given by some known solution of Einstein's equations, and $ h_{ab}(x^\nu)$ is
258: a small perturbation. Our background metric is a
259: Schwarszchild-anti-de Sitter metric (\ref{lineelement}) and the metric
260: $g_{ab}(x^\nu)$ will follow Einstein's equations in vacuum with a
261: cosmological constant:
262: \begin{equation}
263: G_{ab}-\Lambda g_{ab}=0\,.
264: \label{einstein}
265: \end{equation}
266: Upon substituting (\ref{4.1}) in (\ref{einstein}) we will obtain some
267: differential equations for the perturbations. We use the same
268: perturbations as originally given by Regge and Wheeler \cite{Regge},
269: retaining their notation. After a decomposition in tensorial spherical
270: harmonics (see Zerilli \cite{Zerilli1} and Mathews \cite{Mathews}),
271: these fall into two distinct classes - odd and even - with parities
272: $(-1)^{l+1}$ and $(-1)^l$ respectively, where $l$ is the angular
273: momentum of the particular mode. While working in general relativity
274: one has some gauge freedom in choosing the elements $ h_{ab}(x^\nu)$
275: and one should take advantage of that freedom in order to simplify
276: the rather lengthy calculations involved in computing
277: (\ref{einstein}). We shall therefore work with the classical
278: Regge-Wheeler gauge in which the canonical form for the perturbations
279: is (see also \cite{Vish1}):
280:
281: \bigskip
282: \noindent { odd \, parity:}
283: \begin{eqnarray}
284: h_{\mu \nu}= \left[
285: \begin{array}{cccc}
286: 0 & 0 &0 & h_0(r)
287: \\ 0 & 0 &0 & h_1(r)
288: \\ 0 & 0 &0 & 0
289: \\ h_0(r) & h_1(r) &0 &h_0(r)
290: \end{array}\right] e^{-i \omega t}
291: \left(\sin\theta\frac{\partial}{\partial\theta}\right)
292: P_l(\cos\theta)\,;
293: \label{odd}
294: \end{eqnarray}
295: \bigskip
296: { even \, parity:}
297: \begin{eqnarray}
298: h_{\mu \nu}= \left[
299: \begin{array}{cccc}
300: H_0(r) f(r) & H_1(r) &0 & 0
301: \\ H_1(r) & H_2(r)/f(r) &0 & 0
302: \\ 0 & 0 &r^2K(r) & 0
303: \\ 0 & 0 &0 & r^2K(r)\sin^2\theta
304: \end{array}\right] e^{-i \omega t}
305: P_l(\cos\theta).
306: \label{even}
307: \end{eqnarray}
308: Here $P_l(\cos\theta)$ is the Legendre polynomial with angular
309: momentum $l$.
310: If we put this decomposition into Einstein's equations we get ten
311: coupled second order differential equations that fully describe the
312: perturbations: three equations for odd perturbations and seven for
313: even perturbations. It is however possible to circumvent the
314: task of solving these coupled equations. Regge and Wheeler
315: \cite{Regge} and Zerilli \cite{Zerilli2} showed how to combine these
316: ten equations into two second order differential equations, one for
317: each parity. So following Regge and Wheeler \cite{Regge} (see also
318: \cite{Vish2} for more details) we define, for odd parity the wave
319: function $Q(r)$ given by
320: perturbations,
321: \begin{equation}
322: Q(r)= \frac{f(r)}{r} h_1(r) \,.
323: \label{qodd}
324: \end{equation}
325: After some work, Einstein's equations yield
326: \begin{equation}
327: \frac{\partial^{2} Q}{\partial r_*^{2}} +
328: \left\lbrack\omega^2 -V_{\rm odd}(r)\right\rbrack Q=0 \,,
329: \label{waveodd}
330: \end{equation}
331: where
332: \begin{equation}
333: V_{\rm odd}= f(r)
334: \left\lbrack\frac{l(l+1)}{r^2}-\frac{6m}{r^3}\right\rbrack\,.
335: \label{vodd}
336: \end{equation}
337: Likewise, following Zerilli \cite{Zerilli2} one can define for even modes the
338: wavefunction $T(r)$ implicitly in terms of $H_0$, $H_1$ and $K$, through the
339: equations
340: \begin{eqnarray}
341: K=
342: \frac{6m^2+c\left(1+c\right)r^2+m\left(3cr-3\frac{r^3}{R^2}\right)}
343: {r^2\left(3m+cr\right)} T
344: +\frac{dT}{dr_*} \,, \\
345: H_1= -\frac{i\omega\left(-3m^2-3cmr+cr^2-3m\frac{r^3}{R^2}\right)}
346: {r \left(3m+cr\right) f(r)} T -i \omega
347: \frac{r}{f(r)}\frac{dT}{dr_*}\,,
348: \label{4.10}
349: \end{eqnarray}
350: where $c=\frac{1}{2}\left\lbrack l(l+1)-2\right\rbrack$.
351: Then Einstein's equations for even parity perturbations can be written as
352: \begin{equation}
353: \frac{\partial^{2} T}{\partial r_*^{2}} +
354: \left\lbrack \omega^2 -V_{\rm even}(r)\right\rbrack T=0 \,,
355: \label{waveeven}
356: \end{equation}
357: with
358: \begin{equation}
359: V_{\rm even}= \frac{2f(r)}{r^3}
360: \frac{9m^3+3c^2mr^2+c^2\left(1+c\right)r^3+3m^2\left(3cr+3\frac{r^3}{R^2}\right)}
361: {\left(3m+cr\right)^2} \,.
362: \label{veven}
363: \end{equation}
364: Now, by defining
365: \begin{equation}
366: W=
367: \frac{2m}{r^2}+\frac{-3-2c}{3r}+\frac{3c^2+2c^2+27\frac{m^2}{R^2}}
368: {3c\left(3m+cr\right)}+j\,,
369: \label{W}
370: \end{equation}
371: where $j=-\frac{1}{3}\left(\frac{c}{m}+\frac{c^2}{m}+\frac{9m}{cR^2}\right)$, we obtain
372: \begin{equation}
373: V_{\rm odd}=W^2 {+} \frac{dW}{dr_*} +\beta \,,
374: \quad
375: V_{\rm even}=W^2 {-} \frac{dW}{dr_*} +\beta \,,
376: \label{V}
377: \end{equation}
378: where
379: $\beta=-\frac{c^2+2c^3+c^4}{9m^2}$. It is interesting to note that the two
380: potentials, odd and even, can be written in such a simple form, a fact
381: which seems to have been discovered by Chandrasekhar
382: \cite{Chandra2}. Potentials related in this manner are sometimes
383: called super-partner potentials \cite{Cooper}). We note that similar
384: equations were obtained by Mellor and Moss \cite{Mellor} for
385: Schwarzschild-de Sitter spacetime, using a different approach.
386:
387:
388: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
389: %\vskip 0.5cm
390:
391: \noindent
392: \section{Quasinormal modes and some of its properties}
393:
394: %\vskip 3mm
395: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
396:
397: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
398: \subsection{Analytical properties}
399: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
400: To solve equation (\ref{wavemaxwell}) for Maxwell fields
401: and equations (\ref{waveodd}-\ref{waveeven}) for gravitational
402: fields, one must
403: specify boundary conditions.
404: Consider first the case of a Schwarzschild black hole in an
405: asymptotically flat spacetime (see, e.g., \cite{Kokkotas}). Since
406: in this case the
407: potential vanishes at both infinity and horizon, the two
408: solutions near these points are plane waves of the type
409: $\Psi \sim e^{\pm i\omega r_*}$, where the $r_*$ coordinate
410: in this case ranges from $-\infty $ to $\infty$. Quasinormal modes
411: are defined by the condition that at the horizon there are only ingoing waves,
412: i.e., $\Psi_{\rm hor}\sim e^{-i\omega r_*} $ . Furthermore, one does not want to
413: have fields coming in from infinity (where the potential in this case
414: vanishes). So, there is only a purely outgoing wave at infinity, i.e.,
415: $\Psi_{\rm \infty}\sim e^{i\omega r_*} $. Only a discrete set of complex
416: frequencies $\omega$ meet these requirements.
417:
418: Consider now a Schwarzschild black hole in an
419: asymptotically AdS spacetime. The boundary condition at the horizon
420: is the same, we want that near the horizon
421: $\Psi_{\rm hor}\sim e^{-i\omega r_*} $. However, $r_*$ has a finite range,
422: so the second boundary condition needs to be modified. There have been
423: several papers discussing which boundary conditions one should impose at infinity
424: in AdS spacetimes (\cite{Avis,Breit,Burgess}). We
425: shall require energy conservation and thus adopt the reflective boundary
426: conditions at infinity \cite{Avis}.
427: This means that the wavefunction is zero at infinity. For
428: a different boundary condition see \cite{Dasgupta}.
429:
430: We now show that the imaginary part of the frequency $\omega$ is negative, for
431: waves satisfying these boundary conditions, provided the potential $V$ is positive.
432: The proof proceeds as for the scalar field perturbation case \cite{Horo},
433: although there are some steps we think are useful to display explicitly here.
434: Writing $\phi$ for a generic wavefunction as
435: \begin{equation}
436: \phi=e^{i\omega r_*} Z\,,
437: \label{generic}
438: \end{equation}
439: where, $Z$ can be $\Psi$, $Q$ or $T$, we find
440: \begin{equation}
441: f(r)\frac{\partial^{2} \phi}{\partial r^{2}}+
442: \left\lbrack f'-2i\omega \right\rbrack
443: \frac{\partial \phi}{\partial r} -\frac{V}{f}\phi=0\,,
444: \label{waveeq}
445: \end{equation}
446: where $f=( r^{2}+1-\frac{2M}{r})$.
447: In the proof, we are going to need the asymptotic behavior of
448: the solutions of equation (\ref{waveeq}).
449: For $r \rightarrow r_+$ we have $f \sim (3r_+ +\frac{1}{r_+})(r-r_+)$
450: and $\frac{V}{f}\sim C$, where C is a constant which takes different values
451: depending on the case, electromagnetic, odd or even gravitational perturbations.
452: So equation (\ref{waveeq}) becomes, in this limit,
453: \begin{equation}
454: Ay\frac{\partial^{2} \phi}{\partial y^{2}}+ [A-2i\omega]\frac{\partial
455: \phi}{\partial y} -C\phi=0\,,
456: \label{waveeq1}
457: \end{equation}
458: where $y=r-r_+$, and $A=3r_+ +\frac{1}{r_+}$. This equation has an exact
459: solution in terms of the modified Bessel functions $ I_{\nu}(z)$ \cite{Stegun},
460: \begin{equation}
461: \phi=C_1y^{i\frac{\omega}{A}}I_{-\frac{i\omega}{A}}
462: \left(2(\frac{C}{A}y)^{\frac{1}{2}}\right)
463: +C_2y^{i\frac{\omega}{A}}I_{\frac{i\omega}{A}}
464: \left(2(\frac{C}{A}y)^{\frac{1}{2}}\right).
465: \label{wavesol1}
466: \end{equation}
467: We want the asymptotic behavior of these functions when $y
468: \rightarrow 0$ which is given by $I_\nu(z)
469: \rightarrow \frac{(\frac{z}{2})^\nu}{\Gamma(\nu +1)}\,,z \rightarrow
470: 0$. So, near the horizon the wavefunction $\phi$ behaves
471: as
472: \begin{equation}
473: \phi_{r_+}=C_1\frac{
474: \left(\frac{C}{A}\right)^{-\frac{i\omega}{A}}}
475: {\Gamma\left(1-\frac{2i\omega}{A}\right)}
476: +C_2\frac{y^{\frac{2i\omega}{A}}
477: \left(\frac{C}{A}\right)^{\frac{i\omega}{A}}}
478: {\Gamma\left(1+\frac{2i\omega}{A}\right)}.
479: \label{waveeqhor}
480: \end{equation}
481: We can see that if one wants to rule out outgoing modes at the
482: horizon, we must have $C_2=0$, so that $\phi$ in equation
483: (\ref{generic}) does not depend on $y$. Let's now investigate the asymptotic
484: behavior at infinity. For $r \rightarrow \infty$ we have
485: $\frac{V}{f} \rightarrow \frac{l(l+1)}{r^2}$. Therefore near infinity
486: equation (\ref{waveeq}) becomes
487: \begin{equation}
488: r^2\frac{\partial^{2} \phi}{\partial r^{2}}+
489: [2r-2i\omega]\frac{\partial \phi}{\partial r}
490: -\frac{l(l+1)}{r^2}\phi=0 \,.
491: \label{waveeq2}
492: \end{equation}
493: Putting $x=\frac{1}{r}$ we have
494: \begin{equation}
495: \frac{\partial^{2} \phi}{\partial x^{2}}+ 2i\omega\frac{\partial
496: \phi}{\partial x} -l(l+1)\phi=0 \,,
497: \label{waveeq3}
498: \end{equation}
499: with solution $\phi=\phi_{\rm \infty}$ given by
500: \begin{equation}
501: \phi_{\rm \infty}(x)=A \,
502: {\rm e}^{\left\lbrack -i\omega+i\left(\omega^2-l(l+1)\right)^{1/2}\right\rbrack x}
503: + B\,
504: {\rm e}^{\left\lbrack-i\omega-i\left(\omega^2-l(l+1)\right)^{1/2}\right\rbrack x}
505: \label{wavesol2}
506: \end{equation}
507: Now, $\phi_{\rm \infty}(x=0)=0$, therefore $A=-B$, and thus,
508: \begin{equation}
509: \phi_{\infty}(x)=A\,{\rm e}^{-i\omega x}
510: \sin\left\lbrack \left(\omega^2-l(l+1)\right)^{\frac{1}{2}}x\right\rbrack
511: \label{wavesol3}
512: \end{equation}
513: We can now proceed in the proof. Multiplying equation (\ref{waveeq}) by $\bar{\phi}$
514: (the complex conjugate of $\phi$), and
515: integrating from $ r_+ $ to $\infty$ we obtain
516: \begin{equation}
517: \int_{r_+}^{\infty}dr\left\lbrack\bar{\phi}
518: \frac{d\left(f\frac{d\phi}{dr}\right)}{dr}-
519: 2i\omega\bar{\phi}\frac{d\phi}{dr}-\frac{V}{f}\bar{\phi}\phi \right\rbrack=0\,.
520: \label{1}
521: \end{equation}
522: Integrating by parts yields
523: \begin{equation}
524: \int_{r_+}^{\infty}dr \left\lbrack
525: \frac{d[\bar{\phi}f\frac{d\phi}{dr}]}{dr}-
526: f|\frac{d\phi}{dr}|^2 - 2i\omega\bar{\phi}\frac{d\phi}{dr}-\frac{V}{f}|\phi|^2
527: \right\rbrack=0\,.
528: \label{2}
529: \end{equation}
530: Now, one can show that $[\bar{\phi}f\frac{d\phi}{dr}]_{r_+}=0$,
531: in order to satisfy the boundary conditions. Indeed, at $r_+$,
532: $\phi(r_+)={\rm constant}$ and $f(r_+)=0$. Now, at infinity, even though
533: $\bar{\phi}(\infty)=0$, we have also $f(\infty)=\infty$, so we have to
534: show that $[\bar{\phi}f\frac{d\phi}{dr}]_{\infty}=0 $. From equation
535: (\ref{wavesol3}) we can check that this is indeed true.
536: Thus, equation (\ref{2}) gives
537: \begin{equation}
538: \int_{r_+}^{\infty}dr \left\lbrack
539: f|\frac{d\phi}{dr}|^2 + 2i\omega\bar{\phi}\frac{d\phi}{dr}+\frac{V}{f}|\phi|^2
540: \right\rbrack=0.
541: \label{3}
542: \end{equation}
543: Taking the imaginary part of (\ref{3}) we have
544: \begin{equation}
545: \int_{r_+}^{\infty}dr \left\lbrack
546: \omega\bar{\phi}\frac{d\phi}{dr}+ \bar{\omega}\phi\frac{d\bar{\phi}}{dr}
547: \right\rbrack=0\,,
548: \label{4}
549: \end{equation}
550: wich, after an integration by parts reduces to
551: \begin{equation}
552: (\omega-\bar{\omega})\int_{r+}^{\infty}dr \left\lbrack
553: \bar{\phi}\frac{d\phi}{dr}
554: \right\rbrack
555: =\bar{\omega}|\phi(r_+)|^2.
556: \label{5}
557: \end{equation}
558: Finally, inserting this back into (\ref{3}) yields
559: \begin{equation}
560: \int_{r_+}^{\infty}dr \left\lbrack
561: f|\frac{d\phi}{dr}|^2 +\frac{V}{f}|\phi|^2 \right\rbrack
562: =-\frac{|\omega|^2|\phi(r_+)|^2}{{\rm Im}\,
563: \omega}.
564: \label{6}
565: \end{equation}
566: From this relation, one can infer that, if V is positive definite then
567: ${\rm Im}\,\omega\,<0$ necessarily.
568: So, since electromagnetic and even gravitational perturbations have $V>0$
569: one always has ${\rm Im}\, \omega <0$. As for odd gravitational perturbations
570: there are instances where $V<0$, making this theorem unreliable for these cases.
571: However, for
572: $r_+<\left\lbrack\frac{l(l+1)}{3}-1\right\rbrack^{\frac{1}{2}}$,
573: i.e., small enough masses,
574: $V>0$ (see equation (\ref{vodd})), and the theorem applies.
575:
576: Another important point concerns the late time behavior of these fields,
577: and the existence or not of power-law tails. As shown by Ching et al \cite{Ching},
578: for potentials that vanish exponentially near the horizon, there are no power-law tails,
579: so there will be no such tails in our case.
580:
581: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
582: \subsection{Numerical Calculation of the QNM frequencies}
583: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
584: To find the frequencies $\omega$ that satisfy the
585: boundary conditions we first note that equation (\ref{waveeq}) has only regular
586: singularities in the range of interest. It has therefore, by Fuchs
587: theorem, a polynomial solution \cite{arfken}.
588: To deal with the point at infinity, we
589: first change the independent variable $r$ to $x=\frac{1}{r}$. Now we can use
590: Fr\"{o}benius method by looking for an indicial equation (for further
591: details see \cite{Horo}), and force it to obey the boundary condition
592: at the horizon ($x=\frac{1}{r_+}=h$). We get the following solution to equation
593: (\ref{waveeq}),
594: \begin{eqnarray} \phi(x)=
595: \sum_{n=0}^{\infty} a_{n(\omega)} (x-h)^n \,,
596: \label{frobenius}
597: \end{eqnarray}
598: where $a_{n(\omega)}$ is a function of the frequency. If we put
599: (\ref{frobenius}) into (\ref{waveeq}) and use the boundary condition
600: $\phi=0$ at infinity ($x=0$) we obtain
601: \begin{equation}
602: \sum_{n=0}^{\infty} a_{n(\omega)}(-h)^n=0 \,.
603: \label{numerico}
604: \end{equation}
605: Our problem is reduced to that of finding a
606: numerical solution of the polynomial equation (\ref{numerico}).
607: The numerical roots for $\omega$ of equation (\ref{numerico}) can be
608: evaluated resorting to numerical computation. Obviously, one cannot
609: determine the full sum in expression ($\ref{numerico}$), so we have to
610: determine a partial sum from $0$ to $N$, say and find the roots $\omega$
611: of the resulting polynomial expression. We then move onto the next
612: term $N+1$ and determine the roots. If the method is reliable, the
613: roots should converge. We stop our search when we have a 3 decimal digit
614: precision.
615:
616:
617: \subsubsection{Electromagnetic modes}
618: As long as the modes are decaying, it does not matter whether they're
619: oscillating or not. However,
620: as we will see there are frequencies in the electromagnetic case
621: with a vanishing real
622: part, which makes it possible to use an approximation, due to Liu \cite{Liu}, to
623: the highly damped modes. Although the method was
624: originally developed for the asymptotically flat space, it is quite
625: straightforward to apply it to our case. There is therefore a way to
626: test our results. Unfortunately, this method relies heavily
627: on having a frequency with a large pure imaginary part, so as we shall see
628: it will only work for electromagnetic perturbations.
629: We have computed the lowest frequencies for some
630: values of the horizon radius $r_+$, and $l$.
631: The frequency is written as $\omega = \omega_r +
632: i\omega_i$, where $\omega_r$ is the real part of the frequency and
633: $\omega_i$ is its imaginary part.
634: In tables 1 and 2 we list the numerical values of the lowest quasinormal
635: frequencies of electromagnetic perturbations
636: for $l=1$ and $l=2$ and for selected values of $r_+$ .
637: For frequencies with no real part, we list the values obtained
638: in Liu's aproximation.
639: \begin{center}
640: \begin{tabular}{|l|l|l|l|l|} \hline
641: \multicolumn{1}{|c|}{} &
642: \multicolumn{2}{c|}{ Numerical} &
643: \multicolumn{2}{c|}{ Liu's approximation} \\ \hline
644: $r_+$ & $-\omega_i$ & $\omega_r$ & $-\omega_i$ & $\omega_r$ \\ \hline
645: 0.8 & 1.287 & 2.175 & $-$ & $-$ \\ \hline
646: 1 & 1.699 & 2.163 & $-$ & $-$ \\ \hline
647: 5 & 8.795 & $\sim0$ & 7.6 & $\sim0$ \\ \hline
648: 10 & 15.506 & $\sim0$ & 15.05 & $\sim0$ \\ \hline
649: 50 & 75.096 & $\sim0$ & 75.01 & $\sim0$ \\ \hline
650: 100 & 150.048 & $\sim0$ & 150.005 & $\sim0$ \\ \hline
651: \end{tabular}
652: \end{center}
653: \vskip 1mm
654: {\noindent Table 1. Lowest QNM of electromagnetic perturbations for $l=1$.
655: The $-$ in Liu's approximation columns means that the method is not applicable.}
656: \vskip 8mm
657:
658: \begin{center}
659: \begin{tabular}{|l|l|l|l|l|} \hline
660: \multicolumn{1}{|c|}{} &
661: \multicolumn{2}{c|}{ Numerical} &
662: \multicolumn{2}{c|}{ Liu's approximation} \\ \hline
663: $r_+$ & $-\omega_i$ & $\omega_r$ & $-\omega_i$ & $\omega_r$ \\ \hline
664: 0.8 & 1.176 & 2.501 & $-$ & $-$ \\ \hline
665: 1 & 1.579 & 2.496 & $-$ & $-$ \\ \hline
666: 5 & 10.309 & 0.822 & 7.6 & $\sim0$ \\ \hline
667: 10 & 15.755 & $\sim0$ & 15.05 & $\sim0$ \\ \hline
668: 50 & 75.139 & $\sim0$ & 75.01 & $\sim0$ \\ \hline
669: 100 & 150.069 & $\sim0$ & 150.005 & $\sim0$ \\ \hline
670: \end{tabular}
671: \end{center}
672: \vskip 1mm
673: {\noindent Table 2. Lowest QNM of electromagnetic perturbations for $l=2$.}
674: \bigskip
675:
676: \noindent
677: As one can see, the imaginary part of the frequency, which determines
678: how damped the mode is, and which according to the AdS/CFT conjecture
679: is a measure of the characteristic time $\tau=\frac{1}{\omega_i}$ of
680: approach to thermal equilibrium, scales linearly (for large black
681: holes) with the horizon radius supporting the arguments given in
682: \cite{Horo}. Moreover, the frequencies do not seem to depend on the
683: angular quantum number $l$, and are in excellent agreement with the
684: analytical approximation for strongly damped modes.
685:
686: \noindent
687: For a better visualization we also plot $\omega_i \times r_+ $ in
688: Figure 1.
689: \vskip 3mm
690: \centerline{\epsffile{graph1.eps}}
691: \vskip 1mm
692: {\noindent Figure 1. Lowest electromagnetic
693: QNM for $l=1$ as function of $r+$. The black lozenges represent some frequencies numerically
694: calculated. The line connecting them is a linear fit.}
695: \vskip 3mm
696:
697:
698: \subsubsection{Gravitational modes}
699:
700: The numerical calculation of the quasinormal frequencies for gravitational
701: perturbations proceeds as outlined previously (the associated differential
702: equation has only regular singularities, so it is possible
703: to use an expansion such as (\ref{frobenius})).
704:
705:
706: \bigskip
707: \noindent {\bf (i) Odd modes:}
708: In tables 3 and 4 we show
709: the two lowest QNM frequencies for $l=2$ and $l=3$ odd gravitational
710: perturbations.
711: An important point in odd QNMs is that there is a mode for which the
712: frequencies are not only pure imaginary and very small, but also scale
713: with $\frac{1}{r_+}$! This is similar to the behavior of
714: Schwarzschild black holes in asymptotically flat spacetimes, as
715: mentioned. However, all frequencies have a negative imaginary part,
716: which indicates that the spacetime is stable for this kind of
717: perturbations.
718:
719: \begin{center}
720: \begin{tabular}{|l|l|l|l|l|} \hline
721: \multicolumn{1}{|c|}{} &
722: \multicolumn{2}{c|}{ lowest QNM} &
723: \multicolumn{2}{c|}{ second lowest QNM} \\ \hline
724: $r_+$ & $-\omega_i$ & $\omega_r$ & $-\omega_i$ & $\omega_r$ \\ \hline
725: 0.5 & 6.4 & 0 & 0.72 &3.037 \\ \hline
726: 1 &$\sim$ 2(?) & 0 & 2.404 & 3.033 \\ \hline
727: 2 & 0.728 & $\sim0$ &5.258 & 4.447 \\ \hline
728: 5 & 0.2703 & $\sim0$ &13.294 &9.577 \\ \hline
729: 10 & 0.13378 & $\sim0$ & 26.626 & 18.662 \\ \hline
730: 50 & 0.02667 & $\sim0$ & 133.19 & 92.505 \\ \hline
731: 100 & 0.0132 & $\sim0$ & 266.384 & 184.959 \\ \hline
732: \end{tabular}
733: \end{center}
734: \vskip 1mm
735: {\noindent Table 3. Lowest QNM of gravitational odd perturbations for $l=2$.}
736: \vskip 8mm
737:
738:
739: \begin{center}
740: \begin{tabular}{|l|l|l|l|l|} \hline
741: \multicolumn{1}{|c|}{} &
742: \multicolumn{2}{c|}{ lowest QNM} &
743: \multicolumn{2}{c|}{ second lowest QNM} \\ \hline
744: $r_+$ & $-\omega_i$ & $\omega_r$ & $-\omega_i$ & $\omega_r$ \\ \hline
745: 1 & 10 & 0 & 1.639 & 3.849 \\ \hline
746: 2 & 2.189 & 0 &5.080 & 4.615 \\ \hline
747: 5 & 0.690 & $\sim0$ &13.247 &9.735 \\ \hline
748: 10 & 0.336 & $\sim0$ & 26.603 & 18.742 \\ \hline
749: 50 & 0.0669 & $\sim0$ & 133.19 & 92.521 \\ \hline
750: 100 & 0.0333 & $\sim0$ & 266.382 & 184.967 \\ \hline
751: \end{tabular}
752: \end{center}
753: \vskip 1mm
754: {\noindent Table 4. Lowest QNM of gravitational odd perturbations for $l=3$.}
755: \vskip 3mm
756:
757:
758: The value $\omega=2$ in Table 3 marked
759: with a ``?'' is a somewhat dubious result. In fact, from
760: (\ref{waveeqhor}) it follows that if $1-\frac{2i\omega}{A}=-n$, with
761: $n$ an integer, then there is nothing going down the hole, so
762: perhaps it is not a QNM. It is
763: also an ``algebraically special value'' in the sense of Chandrasekhar
764: \cite{Chandrass}.
765:
766: \bigskip
767: \noindent {\bf (ii) Even modes:}
768: In table 5 we show the lowest QNM frequencies for $l=2$ and $l=3$ even
769: gravitational perturbations. We point out the remarkable resemblance
770: of the values in table 3 with those in \cite{Horo} for scalar
771: perturbations, even though the potentials are quite different. We have
772: performed calculations for higher values of the angular quantum number
773: $l$, and found that the QNM frequencies are indeed very similar
774: throughout all values of $l$.
775:
776:
777:
778: \begin{center}
779: \begin{tabular}{|l|l|l|l|l|} \hline
780: \multicolumn{1}{|c|}{} &
781: \multicolumn{2}{c|}{ lowest QNM, $l=2$} &
782: \multicolumn{2}{c|}{ lowest QNM, $l=3$} \\ \hline
783: $r_+$ & $-\omega_i$ & $\omega_r$ & $-\omega_i$ & $\omega_r$ \\ \hline
784: 1 &1.584 &3.018 & 1.392 & 3.909 \\ \hline
785: 2 & 3.974 &4.546 & 3.299 & 4.597 \\ \hline
786: 5 & 12.649 &9.83 & 11.642 &10.217 \\ \hline
787: 10 & 26.301 &18.806 & 25.788 &19.089 \\ \hline
788: 50 &133.125 &92.535 & 133.022 &92.596 \\ \hline
789: 100 & 266.351 & 184.959 & 266.300 &185.005 \\ \hline
790: \end{tabular}
791: \end{center}
792: \vskip 1mm
793: {\noindent Table 5. Lowest QNM of gravitational even perturbations.}
794: \vskip 3mm
795:
796:
797:
798:
799: \bigskip
800: \noindent {\bf (iii) Discussion:}
801: We first note that there is clearly a distinction between odd
802: and even perturbations: they no longer have the same spectra, contrary
803: to the asymptotically flat space case (see \cite{Chandras}), a problem
804: we shall consider in more detail in the next subsection.
805: We also remark that in electromagnetic and scalar perturbations the
806: frequency scales with $r_+$ (for large black holes at least). Since
807: the temperature scales also with
808: $r_+$ in the large black hole regime, this means that
809: the frequency scales with the temperature. Thus, in the dual CFT the
810: approach to thermal equilibrium is faster for higher temperatures.
811: This is a totally different
812: behavior from that of asymptotically flat space, in which the
813: frequency scales with $\frac{1}{r_+}$.
814: However, for odd modes there is one that scales
815: with $\frac{1}{r_+}$. This is a reflection of the different behavior of
816: the potential $V_{\rm odd}$ for odd perturbations (that was why we couldn't prove
817: stability for odd perturbations in the first place), and of the boundary conditions,
818: as we shall show in section 3.3.
819: The odd modes are therefore particularly long-lived.
820:
821:
822:
823: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
824: \noindent
825: \subsection{On the isospectrality breaking between odd and even perturbations }
826: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
827: As is well known \cite{Chandra2,Chandras} in the case of a
828: Schwarzschild black hole in an asymptotically flat space the two
829: potentials $V_{\rm even}$ and $ V_{\rm odd}$ give rise to the same quasinormal
830: frequencies (in fact to the same absolute value of the reflection
831: and transmission coefficients). This remarkable
832: property followed from a special relation (the equivalent for
833: asymptotically flat spacetimes of our equation (\ref{V})) between the
834: potentials and the behavior of W at the boundaries. However, as one
835: can see in tables 3, 4 and 5 there is a isospectrality breaking between odd and
836: even perturbations in Schwarzschild anti-de Sitter spacetime.
837:
838: We shall now treat this problem.
839: The breaking of the isospectrality is intimately
840: related to the behavior of $W$ at infinity. On taking
841: advantage of the machinery developed by Chandrasekhar, we seek a
842: relation between odd and even perturbations of the form
843: \begin{eqnarray}
844: Q= p_1T+q_1\frac{dT}{dr_*} \,, \\ T= p_2Q+q_2\frac{dQ}{dr_*}\,,
845: \label{relation1}
846: \end{eqnarray}
847: yielding (see \cite{Chandra2} for details),
848: $q_1^2=\frac{1}{\beta-\omega^2}\,$, $p_1=qW\,$, $p_2=-p_1\,$ and
849: $q_2=q_1=q$. Thus, we obtain
850: \begin{eqnarray}
851: Q= qWT+q\frac{dT}{dr_*} \,, \\ T= -qWQ+q\frac{dQ}{dr_*}\,.
852: \label{relation2}
853: \end{eqnarray}
854: Suppose now that $\omega$ is a QNM frequency of $T$, i.e., one for which
855: \begin{eqnarray}
856: T \rightarrow A_{\rm even} e^{-i\omega r_*} \,,\,\, r\rightarrow r_+\,, \\
857: T \rightarrow 0 \,,\,\, r\rightarrow \infty\,.
858: \label{Tasymptotic}
859: \end{eqnarray}
860: Substituting this into equation (\ref{relation2}) we see that
861: \begin{eqnarray}
862: Q \rightarrow A_{\rm even}
863: q\left\lbrack W(r+)-i\omega\right\rbrack
864: e^{-i\omega r_*} \,,\,\,r\rightarrow r_+\,, \\
865: Q \rightarrow q\left(\frac{dT}{dr_*}
866: \right)_{r=\infty} \,,\,\, r\rightarrow \infty\,.
867: \label{Qasymptotic}
868: \end{eqnarray}
869: However, from equation (\ref{wavesol3}),
870: $\left(\frac{dT}{dr_*}\right)_{r=\infty}$ is in general
871: not zero so that $\omega$ will in general fail to be a QNM frequency
872: for Q. Should $Q$ and $T$ be smooth functions of $\omega$,
873: one expects that if $q$
874: is ``almost zero'' then $\omega$ should ``almost'' be a QNM frequency
875: for $Q$. Now, the condition that $q$ is almost zero is that
876: $\beta-\omega^2$ be very large, and one expects this to be true either
877: when $\omega$ is very large or else when $\beta$ is very large. And in
878: fact, as one can see in tables 3, 4 and 5 for very large $\omega$ the
879: frequencies are indeed almost identical. On the other hand, for very
880: small black holes ($\beta$ very large) one expects the frequencies to
881: be exactly the same, since both potentials have the same asymptotic
882: behavior in this regime, as we shall see in the next section. One
883: would be tempted to account for the remarkable resemblances between QNM
884: frequencies of scalar and gravitational perturbations by a similar
885: approach, but the proof is still eluding us. Should such an approach
886: work, it could be of great importance not only to this specific
887: problem, but also to the more general problem of finding the
888: asymptotic distribution of eigenvalues, by studying a different
889: potential with (asymptotically) the same eigenvalues, but more easy to
890: handle.
891:
892: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
893: %\vskip 0.5cm
894: \noindent
895: \section{The limit m $\rightarrow$ 0}
896: %\vskip 3mm
897: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
898: Although it is not possible to solve exactly for the QNM frequencies,
899: it is possible to gain some analytical insight in the special case of
900: very small black holes. There has been some discussion
901: about this regime (see \cite{Wang3}, and also \cite{Horo,Hubeny}
902: and references therein). Here
903: we shall exploit the behavior of QNM frequencies in this regime a
904: little further.
905: For very small black holes one can easily see that both
906: potentials (electromagnetic and gravitational) look like,
907: in the $r_*$ coordinate, a barrier with
908: unequal heights:
909: \vskip 3mm
910: \centerline{\epsffile{barreira.eps}}
911: \vskip 1mm
912: {\noindent Figure 2. The potential for small black holes.
913: $V_{\rm max}=\frac{4l(l+1)}{27r_+^2}$ and $a \sim 3r_+$.}
914: \vskip 3mm
915: It is trivial to obtain equations for the quasinormal
916: frequencies in this limit. If $\Psi$ is a general wavefunction then
917: \begin{eqnarray}
918: \frac{d\Psi/dr_*}{\Psi}=-i\omega \;,\quad r_*<0\,, \\
919: \frac{d\Psi/dr_*}{\Psi}=\frac{ik_1Be^{ik_1r_*}-ik_1Ce^{-ik_1r_*}}
920: {Be^{ik_1r_*}+Ce^{-ik_1r_*}}\;, \quad 0<r_*<a\,, \\
921: \frac{d\Psi/dr_*}{\Psi}=\frac{ikDe^{ikr_*}-ik_1Ee^{-ikr_*}}
922: {De^{ikr_*}+Ee^{-ikr_*}}\;, \quad r_*>a \;,
923: \label{5.1}
924: \end{eqnarray}
925: where $k_1=(\omega^2-V_{max})^{1/2}$, and
926: $k=\left\lbrack\omega^2-l(l+1)\right\rbrack^{1/2}$. Imposing the continuity
927: of the logarithmic derivative and furthermore that $\Psi=0$ at
928: infinity ($r_*=\frac{\pi}{2}$), we get
929: \begin{eqnarray}
930: k_1\left\lbrack\frac{\frac{k_1-\omega}{k_1+\omega}e^{2ik_1a}-1}
931: {\frac{k_1-\omega}{k_1+\omega}e^{2ik_1a}+1}\right\rbrack
932: =
933: k\left\lbrack
934: \frac{e^{2ik(\frac{\pi}{2}-a)}+1}
935: {1-e^{2ik(\frac{\pi}{2}-a)}}\right\rbrack\;.
936: \label{5.2}
937: \end{eqnarray}
938: In the limit $a \rightarrow 0$, $k_1 \rightarrow \infty$
939: ($m \rightarrow 0$) we have, supposing that $\omega$ stays small,
940: the condition $e^{2ik\frac{\pi}{2}}=1$ which means that
941: \begin{equation}
942: \omega_0^2=4n^2+l(l+1)\;, \quad n=1,2,...\;,
943: \label{5.3}
944: \end{equation}
945: corresponding to a bound state. This gives for the lowest
946: QNM frequencies ($n=1$):
947: $\omega_0=2.45 \,$ for $l=1$ and $\omega_0=3.16 \,$ for $l=2$.
948: The above are to be compared with those in Tables 1-5.
949: The agreement seems excellent, and we can now go a step further
950: : If we linearize (\ref{5.2}) around the solution (\ref{5.3}), i.e,
951: if we write $\omega= \omega_0+i\delta$ and substitute back in (\ref{5.2}) we
952: obtain,
953: to third order in $\delta$, the values listed in Table 6 .
954: \begin{center}
955: \begin{tabular}{|l|l|l|l|l|} \hline
956: \multicolumn{1}{|c|}{} &
957: \multicolumn{1}{c|}{$ a=3/h$} &
958: \multicolumn{1}{c|}{$ a=6/h$} \\ \hline
959: \multicolumn{1}{|c|}{l} & \multicolumn{1}{|c|}{ $\delta$} &\multicolumn{1}{|c|}{ $\delta$}
960: \\ \hline
961: 1 &$-2.1/h-i(1.42/h^2)$ &$ -1.92/h-i(0.05/h^2)$
962: \\ \hline
963: 2 &$ -0.859/h-i(0.04/h^2) $ &
964: $ -0.85/h-i(1.4{\rm x}10^{-4}/h^2)$
965: \\ \hline
966: \end{tabular}
967: \end{center}
968: \vskip 1mm
969: {\noindent Table 6. The linearized frequency $\delta\,$ for selected values of the
970: angular quantum number $l$ and the potential width $a$. }
971: \vskip 3mm
972: We have chosen a typical value of $a\sim \frac{3}{h}$, but we can see
973: that, although the real part does not depend very much on $a$, the imaginary part
974: is strongly sensitive to $a$. Nevertheless, one can be sure that whatever
975: value of $a$, the imaginary part goes as $\frac{1}{h^2}$ and is always negative.
976: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
977: %\vskip 0.5cm
978:
979: \noindent
980: \section{Conclusions}
981: %\vskip 3mm
982: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
983: We have computed the electromagnetic and gravitational QNM
984: frequencies of Schwarzschild-AdS black holes in four dimensions. These
985: modes dictate the late time behavior of a minimally coupled
986: electromagnetic field and of small gravitational perturbations,
987: respectively. The conclusions are:
988: (i) The frequencies all have a negative imaginary part, which means that
989: the black hole is stable against these perturbations, since these will
990: decay exponentially with time;
991: (ii) Maxwell perturbations are strongly damped, so according
992: to the AdS/CFT conjecture, any electromagnetic perturbed
993: thermal state will rapidly approach equilibrium;
994: (iii) for odd gravitational perturbations in the large black hole regime, the
995: imaginary part of the frequency (decaying mode) goes to zero scaling with
996: $\frac{1}{r_+}$, just as in asymptotically flat space. In terms of the
997: AdS/CFT correspondence, this implies that the greater the mass, the
998: more time it takes to approach equilibrium, an unusual result;
999: (iv) scalar \cite{Horo} and gravitational even perturbations exhibit an amazing
1000: similarity for the characteristic time damping of the perturbations,
1001: but we have not been able to prove it analytically;
1002: (v) in the small black hole regime
1003: the imaginary part of the frequency (decaying mode) scales with $r_+^2$.
1004:
1005: \vskip 3mm
1006: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1007:
1008:
1009:
1010:
1011:
1012:
1013: \vskip .5cm
1014:
1015: \section*{Acknowledgments}
1016: This work was partially funded
1017: by Funda\c c\~ao para a Ci\^encia e Tecnologia (FCT)
1018: through project PESO/PRO/2000/4014. V.C. also
1019: acknowledges finantial support from FCT
1020: through PRAXIS XXI programme.
1021: J. P. S. L. thanks Observat\'orio Nacional do Rio de Janeiro for
1022: hospitality.
1023:
1024: \vskip .5cm
1025: \newpage
1026:
1027: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1028: \begin{thebibliography}{100}
1029:
1030: \bibitem{Cunningham} C. T. Cunningham, R. H. Price, and
1031: V. Moncrief,
1032: Ap. J. Lett. {\bf 224}, 643(1978).
1033: \bibitem{Anninos} P. Anninos, D. Hobill, E. Seidel, L. Smarr
1034: and W-M. Suen,
1035: Phys. Rev. Lett. {\bf 71}, 2851(1993).
1036: \bibitem{Kokkotas} K. D. Kokkotas, and B. G. Schmidt,
1037: Living Rev. Rel. {\bf2}, 2(1999).
1038: \bibitem{Andersson} N. Andersson, in
1039: {\it Black Holes, Gravitational radiation and the Universe},
1040: (Edited by Bala R. Iyer and Biplab Bhawall, 1999).
1041: \bibitem{York} J. W. York,
1042: Phys. Rev. {\bf D28}, 2929(1983).
1043: \bibitem{maldacena} J. Maldacena, Adv. Theor. Math. Phys.
1044: {\bf 2}, 253(1998).
1045: \bibitem{Horo} G. T. Horowitz, and V. Hubeny,
1046: Phys. Rev. {\bf D62}, 024027(2000).
1047: \bibitem{Horowitz} G. T. Horowitz, Class. Quantum Grav.
1048: {\bf D17}, 1107(2000).
1049: \bibitem{Wang1} B. Wang, C. Y. Lin, and E. Abdalla,
1050: Phys. Lett. {\bf B481}, 79(2000).
1051: \bibitem{Wang2} B. Wang, C. M. Mendes, and E. Abdalla,
1052: Phys. Rev. {\bf D63}, 084001(2001).
1053: \bibitem{Cardoso}
1054: V. Cardoso, J. P. S. Lemos, Phys. Rev. {\bf D63}, 124015(2001), gr-qc/0101052.
1055: \bibitem{Bir} D. Birmingham , hep-th/0101194.
1056: \bibitem{Chop} M. W. Choptuik,
1057: Phys. Rev. Lett. {\bf 70}, 9(1993).
1058: \bibitem{Kim} W. T. Kim, J. J. Oh, hep-th/0105103
1059: \bibitem{Wang3} J. Zhu, B. Wang, and E. Abdalla,
1060: Phys. Rev. {\bf D63}, 124004(2001).
1061: \bibitem{Ruffini} R. Ruffini, in
1062: {\it Black Holes: les Astres Occlus},
1063: (Gordon and Breach Science Publishers, 1973).
1064: \bibitem{Liu}
1065: H. Liu, Class. Quantum Grav. {\bf 12}, 543(1995).
1066: \bibitem{Regge} T. Regge, J. A. Wheeler, Phys. Rev. {\bf 108},
1067: 1063(1957).
1068: \bibitem{Zerilli1}
1069: F. Zerilli, J. Math. Phys. {\bf 11}, 2203(1970).
1070: \bibitem{Mathews}
1071: J. Mathews, J. Soc. Ind. Appl. Math. {\bf 10}, 768(1962).
1072: \bibitem{Vish1}
1073: C. V. Vishveshwara, Phys. Rev. {\bf D1}, 2870(1970).
1074: \bibitem{Vish2}
1075: L. A. Edelstein, and C. V. Vishveshwara,
1076: Phys. Rev. {\bf D1}, 3514(1970).
1077: \bibitem{Zerilli2}
1078: F. Zerilli,
1079: Phys. Rev. Lett. {\bf 24}, 737(1970).
1080: \bibitem{Cooper}
1081: F. Cooper, A. Khare and U. Sukhatme,
1082: Phys. Rep. {\bf 251}, 267(1995).
1083: \bibitem{Chandra2} S. Chandrasekhar, in
1084: {\it The Mathematical Theory of Black Holes},
1085: (Oxford University Press, New York, 1983).
1086: \bibitem{Mellor}
1087: F. Mellor, I. Moss ,
1088: Phys. Rev. {\bf D41}, 403(1990).
1089: \bibitem{Avis}
1090: S. J. Avis, C. J. Isham and D. Storey,
1091: Phys. Rev. {\bf D18}, 3565(1978).
1092: \bibitem{Breit}
1093: P. Breitenlohner and D. Z. Freedman,
1094: Phys. Lett. {\bf B115}, 197(1982).
1095: \bibitem{Burgess}
1096: C. P. Burgess and C. A. Lutken,
1097: Phys. Lett. {\bf B153}, 137(1985).
1098: \bibitem{Dasgupta}
1099: A. Dasgupta,
1100: Phys. Lett. {\bf B445}, 279(1999).
1101: \bibitem{Stegun} M. Abramowitz, and I. A. Stegun in
1102: {\it Handbook of Mathematical Functions},
1103: (Dover, New York, 1970).
1104: \bibitem{Ching} E. S. C. Ching, P. T. Leung, W. M. Suen, and K. Young,
1105: Phys. Rev. {\bf D52}, 2118(1995).
1106: \bibitem{arfken} G. B. Arfken, and H. J. Weber,
1107: {\it Mathematical Methods for Physicists},
1108: (Academic Press, 1995).
1109: \bibitem{Chandrass} S. Chandrasekhar,
1110: Proc. R. Soc. London, Ser. A {\bf 392}, 1(1984).
1111: \bibitem{Chandras} S. Chandrasekhar, and S. Detweiler,
1112: Proc. R. Soc. London, Ser. A {\bf 344}, 441(1975).
1113: \bibitem{Hubeny} V. Hubeny, L. Susskind, and N. Toumbas, hep-th/0011164.
1114: \end{thebibliography}
1115: \end{document}
1116:
1117:
1118:
1119:
1120:
1121: