1:
2:
3: \documentclass[twocolumn,showpacs,preprintnumbers,amssymb,nofootinbib]{revtex4}
4: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb,nofootinbib]{revtex4}
5:
6:
7: \usepackage{graphicx}% Include figure files
8: \usepackage{bm} % Include bold math: \bm{} creates bold letters and symbols in math mode
9:
10: \begin{document}
11:
12: %\preprint{DF/IST-04.2002}
13:
14: \title{The black hole bomb and superradiant instabilities}
15:
16:
17: \author{Vitor Cardoso}
18: \email{vcardoso@teor.fis.uc.pt} \affiliation{Centro de F\'{\i}sica
19: Computacional, Universidade de Coimbra, P-3004-516 Coimbra,
20: Portugal}
21: \author{\'Oscar J. C. Dias}
22: \email{odias@ualg.pt} \affiliation{ Centro
23: Multidisciplinar de Astrof\'{\i}sica - CENTRA, Departamento de
24: F\'{\i}sica, F.C.T., Universidade do Algarve, Campus de Gambelas,
25: 8005-139 Faro, Portugal}
26: \author{Jos\'e P. S. Lemos}
27: \email{lemos@kelvin.ist.utl.pt} \affiliation{ Centro
28: Multidisciplinar de Astrof\'{\i}sica - CENTRA, Departamento de
29: F\'{\i}sica, Instituto Superior T\'ecnico, Av. Rovisco Pais 1,
30: 1049-001 Lisboa, Portugal}
31: \author{Shijun Yoshida}
32: \email{shijun@waseda.jp}
33: \affiliation{
34: Centro Multidisciplinar de Astrof\'{\i}sica - CENTRA,
35: Departamento de F\'{\i}sica, Instituto Superior T\'ecnico,
36: Av. Rovisco Pais 1, 1049-001 Lisboa, Portugal \footnote{Present address:
37: Science and Engineering, Waseda University, Okubo, Shinjuku,
38: Tokyo 169-8555, Japan}}
39:
40:
41: \date{\today}
42:
43: \begin{abstract}
44:
45: A wave impinging on a Kerr black hole can be amplified as it scatters
46: off the hole if certain conditions are satisfied giving rise to
47: superradiant scattering. By placing a mirror around the black hole one
48: can make the system unstable. This is the black hole bomb of Press
49: and Teukolsky. We investigate in detail this process and compute the
50: growing timescales and oscillation frequencies as a function of the
51: mirror's location. It is found that in order for the system black hole
52: plus mirror to become unstable there is a minimum distance at which
53: the mirror must be located. We also give an explicit example showing
54: that such a bomb can be built. In addition, our arguments enable us
55: to justify why large Kerr-AdS black holes are stable and small
56: Kerr-AdS black holes should be unstable.
57:
58: \end{abstract}
59:
60: \pacs{04.70.-s}
61:
62: \maketitle
63: \newpage
64: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
65: \section{Introduction}
66: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
67: Superradiant scattering is known in quantum systems
68: for a long time, after the problems raised by Klein's paradox \cite{manogue,greiner}.
69: However, for classical systems superradiant scattering
70: was considered only much later in a paper by
71: Zel'dovich \cite{zel1} where it was examined what happens when scalar
72: waves hit a rotating cylindrical absorbing
73: object. Considering a wave of the form $e^{-i\omega t + i m \phi}$
74: incident upon such a rotating object, one concludes that if the
75: frequency $\omega$ of the incident wave satisfies
76: $\omega < m \Omega$,
77: where $\Omega$ is the angular velocity of the body, then the
78: scattered wave is amplified.
79: It was also anticipated in \cite{zel1} that by surrounding the
80: rotating cylinder by a mirror one could make the system unstable.
81:
82: A Kerr black hole is one of the most interesting rotating objects for
83: superradiant phenomena, where the condition $\omega< m \Omega$ also
84: leads to superradiant scattering, with $\Omega$ being now the angular
85: velocity of the black hole \cite{bardeen,staro1,teu}. Feeding back
86: the amplified scattered wave, one can extract as much rotational
87: energy as one likes from the black hole. Indeed, if one surrounds the
88: black hole by a reflecting mirror, the wave will bounce back and forth
89: between the mirror and the black hole amplifying itself each
90: time. Then the total extracted energy should grow exponentially until
91: finally the radiation pressure destroys the mirror. This is Press and
92: Teukolsky's black hole bomb \cite{press}. Nature sometimes provides
93: its own mirror. For example, if one considers a massive scalar field
94: with mass $\mu$ scattering off a Kerr black hole, then for $\omega <
95: \mu$ the mass $\mu$ effectively works as a mirror
96: \cite{damour,detweiler}.
97:
98: Here we investigate in detail the black hole bomb by using a scalar
99: field model. Specifically, the black hole bomb consists of a Kerr
100: black hole surrounded by a mirror placed at a constant $r$, $r=r_0$,
101: where $r$ is the Boyer-Lindquist radial coordinate. We study the
102: oscillation frequencies and growing timescales as a function of the
103: mirror's location, and as a function of the black hole rotation.
104: A spacetime with a ``mirror'' naturally incorporated in it is
105: anti-de Sitter (AdS) spacetime, which has attracted a great deal
106: of attention recently. It could therefore be expected that
107: Kerr-AdS black holes would be unstable. Fortunately, Hawking and
108: Reall \cite{hawking} have given a simple argument showing that, at
109: least large Kerr-AdS black holes are stable. As we shall show,
110: this is basically because superradiant modes are not excited for
111: these black holes. Furthermore, we suggest it is not only
112: possible but in fact highly likely that small Kerr-AdS black holes
113: are unstable.
114:
115: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
116: \section{The black hole bomb}
117: \label{bhbomb}
118: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
119: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
120: \subsection{Formulation of the problem and basic equations}
121: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
122: We shall consider a massless scalar field in the vicinity of a
123: Kerr black hole, with an exterior geometry described by the line
124: element:
125: \begin{eqnarray}
126: ds^2 \!\!&=&\!\! -\left ( 1-\frac{2Mr}{\rho^2}\right )dt^2
127: -\frac{2Mr a\sin^2\theta}{\rho^2}\, 2 dt d\phi +\frac{\rho^2}{\Delta}\,dr^2 \nonumber \\
128: & & +\rho^2 d\theta^2+\left ( r^2+a^2+\frac{2Mr
129: a^2\sin^2\theta}{\rho^2}\right )\sin^2\theta\, d\phi^2 \,, \nonumber \\
130: & &
131: \label{metric}
132: \end{eqnarray}
133: with
134: \begin{eqnarray}
135: \Delta=r^2+a^2-2Mr\,, \qquad \rho^2=r^2+a^2 \cos^2\theta \,.
136: \label{metric parameters}
137: \end{eqnarray}
138: This metric describes the gravitational field of the Kerr black
139: hole, with mass $M$, angular momentum $J=M a$, and has an event
140: horizon at $r=r_+=M+\sqrt{M^2-a^2}$.
141: A characteristic and important parameter of a Kerr black hole is the angular velocity
142: of its event horizon $\Omega$ given by
143: \begin{equation}
144: \Omega=\frac{a}{2Mr_+}\,.
145: \label{Omega}
146: \end{equation}
147: In absence of sources, which we consider to be our case, the
148: evolution of the scalar field is dictated by the
149: Klein-Gordon equation in curved spacetime, $\nabla_{\mu}\nabla^{\mu}\Phi=0$. To make
150: the whole problem more tractable, it is convenient to separate the
151: field as \cite{brill}
152: \begin{equation}
153: \Phi(t,r,\theta,\phi)=e^{-i\omega t + i m \phi} S^m _l(\theta)
154: R(r)\,, \label{separation}
155: \end{equation}
156: where $S^m _l(\theta)$ are spheroidal angular functions, and the
157: azimuthal number $m$ takes on integer (positive or negative)
158: values. For our purposes, it is enough to consider positive
159: $\omega$'s in (\ref{separation}) \cite{bardeen}. Inserting this in
160: Klein-Gordon equation, we get the following angular and radial
161: wave equations for $R(r)$ and $S^m _l(\theta)$:
162: \begin{eqnarray}
163: & & \frac{1}{\sin \theta}\partial_{\theta}\left ( \sin \theta
164: \partial_{\theta} S^m _l \right ) \nonumber \\
165: & &\hspace{1cm}+
166: \left [ a^2 \omega^2 \cos^2
167: \theta-\frac{m^2}{\sin ^2{\theta}}+A_{lm} \right ]S^m _l =0\,,
168: \label{wave eq separated1}
169: \\
170: & &
171: \Delta\partial_r \left (
172: \Delta \partial_r R \right )+ {\bigl [} \omega^2(r^2+a^2)^2-2M a m
173: \omega r +a^2 m^2 \nonumber \\
174: & & \hspace{3.2cm} - \Delta (a^2\omega^2+ A_{lm}) {\bigr ]} R=0\,,
175: \label{wave eq separated}
176: \end{eqnarray}
177: where $A_{lm}$ is the separation constant that allows the split of
178: the wave equation, and is found as an eigenvalue of
179: (\ref{wave eq separated1}). For small $a \omega$, the regime we shall be interested
180: on in the next subsection, one has \cite{staro1,seidel}
181: \begin{eqnarray}
182: A_{lm}=l(l+1)+{\cal O}(a^2\omega^2)\,.
183: \label{eigenvalues}
184: \end{eqnarray}
185: Near the boundaries of interest, which are the horizon, $r=r_+$,
186: and spatial infinity, $r=\infty$, the scalar field as given by
187: (\ref{separation}) behaves as
188: \begin{eqnarray}
189: \Phi \sim \frac{e^{-i\omega t}}{r}e^{\pm i\omega
190: r_*}\,\,,\,\,r\rightarrow \infty \label{bc1}
191: \\
192: \Phi \sim e^{-i\omega t}e^{\pm i(\omega-m\Omega)
193: r_*}\,\,,\,\,r\rightarrow r_+\,, \label{bc2}
194: \end{eqnarray}
195: where the tortoise $r_*$ coordinate is defined implicitly by
196: $\frac{dr_*}{dr}=\frac{r^2+a^2}{\Delta}$.
197: Requiring ingoing waves at
198: the horizon, which is the physically acceptable solution, one must
199: impose a negative group velocity $ v_{\rm gr}$ for the wave
200: packet. Thus, we choose $\Phi \sim
201: e^{-i\omega t}e^{- i(\omega-m\Omega) r_*}$.
202: However, notice that if
203: \begin{equation}
204: \omega < m \Omega \,,
205: \label{super}
206: \end{equation}
207: the phase velocity$\,$
208: --$\frac{\omega-m\Omega}{\omega}$ will be positive.
209: Thus, in this superradiance regime, waves appear as
210: outgoing to an inertial observer at spatial infinity,
211: and energy is in fact being extracted. Notice that, since
212: we are working with positive $\omega$, superradiance will occur only
213: for positive $m$, i.e., for waves that are co-rotating with the black
214: hole. This follows from the time and angular dependence of the wave
215: function, $\Phi \sim e^{i(-\omega t+m\phi)}$. The phase velocity along the
216: angle $\phi$ is then $v_{\phi}=\frac{\omega}{m}$, which for
217: $\omega>0$ and $m>0$ is positive, i.e., is in the same sense as the
218: angular velocity of the black hole.
219:
220: Here we shall consider a Kerr black hole surrounded by a
221: mirror placed at a constant Boyer-Lindquist radial $r$
222: coordinate with a radius $r_0$, so that the scalar field will be
223: required to vanish at the mirror's location, i.e., $\Phi(r=r_0)=0$.
224: With these two boundary conditions, ingoing waves at the horizon and a
225: vanishing field at the mirror, the problem is transformed into an
226: eigenvalue equation for $\omega$.
227:
228: The frequencies
229: satisfying both boundary conditions will be called Boxed Quasi-Normal
230: frequencies (BQN frequencies, $\omega_{BQN}$) and the associated modes
231: will accordingly be termed Boxed Quasi-Normal Modes (BQNMs). The quasi
232: stems from the fact that they are not stationary modes, and that BQN
233: frequencies are not real numbers. Instead they are complex quantities,
234: describing the decaying or amplification of the field. One expects
235: that for a mirror located at large distances, or for small black
236: holes, the imaginary part of the BQNs will be negligibly small and thus the
237: modes will be stationary, corresponding to the pure normal modes of
238: the mirror in the absence of the black hole.
239: The BQNMs are of course different from the usual quasinormal modes
240: (QNMs) in asymptotically flat spacetimes, because the latter have no
241: mirror and satisfy outgoing wave boundary conditions near spatial
242: infinity, they describe the free oscillations of the black hole
243: spacetime.
244: In the following we shall compute these modes
245: analytically in a certain limit, and numerically by directly
246: integrating the radial equation (\ref{wave eq separated}).
247:
248: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
249: \subsection{\label{sec:BH bomb analytical}The black hole bomb: analytical calculation of the unstable
250: modes}
251: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
252:
253: In this section, we will compute
254: analytically, within some approximations, the unstable modes
255: of a scalar field in a black hole mirror system.
256: Due to the presence of a reflecting mirror
257: around the black hole, the scalar wave is successively impinging back on the
258: black hole, and amplified.
259:
260: We assume that $1/\omega \gg M$, i.e., that the Compton wavelength
261: of the scalar particle is much larger than the typical size
262: of the black hole. We will also assume, for simplicity, that $a\ll
263: M$. Following \cite{staro1,unruh},
264: we divide the space outside the event horizon in two regions,
265: namely, the near-region, $r-r_+ \ll 1/\omega$, and the far-region,
266: $r-r_+ \gg M$. We will solve the radial equation
267: (\ref{wave eq separated}) in each one of these two regions.
268: Then, we will match the near-region and the far-region solutions
269: in the overlapping region where $M \ll r-r_+ \ll 1/\omega$ is satisfied.
270: Finally, we will insert a mirror around the black hole, and
271: we will find the properties of the unstable modes.
272:
273: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
274: \subsubsection{\label{sec:BH Near region}Near-region wave equation and solution}
275: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
276: In the near-region, $r-r_+ \ll 1/\omega$,
277: the radial wave equation can be written as
278: \begin{eqnarray}
279: \Delta\partial_r \left ( \Delta\partial_r R \right )+ r_+^4
280: (\omega-m\Omega)^2\,R - l(l+1)\Delta\,R=0 \, .
281: \label{near wave eq}
282: \end{eqnarray}
283: To find the analytical solution of this equation, one first
284: introduces a new radial coordinate,
285: \begin{eqnarray}
286: z=\frac{r-r_+}{r-r_-}\, , \qquad 0\leq z \leq 1\,,
287: \label{new radial coordinate}
288: \end{eqnarray}
289: with the event horizon being at $z=0$. Then, one has $\Delta
290: \partial_r=(r_+-r_-)z\partial_z$, and the near-region radial wave
291: equation can be written as
292: \begin{eqnarray}
293: & & \hspace{-0.5cm} z(1\!-\!z)\partial_z^2 R+ (1\!-\!z)\partial_z
294: R+ \varpi^2
295: \frac{1\!-\!z}{z} R - \frac{l(l+1)}{1\!-\!z} R=0\,, \nonumber \\
296: & &
297: \label{near wave eq with z}
298: \end{eqnarray}
299: where we have defined the superradiant factor
300: \begin{eqnarray}
301: \varpi \equiv(\omega-m\Omega)\frac{r_+^2}{r_+-r_-} \,.
302: \label{superradiance factor}
303: \end{eqnarray}
304: Through the definition
305: \begin{eqnarray}
306: R=z^{i \,\varpi} (1-z)^{l+1}\,F \,,
307: \label{hypergeometric function}
308: \end{eqnarray}
309: the near-region radial wave equation becomes
310: \begin{eqnarray}
311: & & \hspace{-0.5cm} z(1\!-\!z)\partial_z^2 F+ {\biggl [} (1+i\,
312: 2\varpi)-\left [ 1+2(l+1)+ i\, 2\varpi \right ]\,z {\biggr ]}
313: \partial_z F \nonumber \\
314: & & \hspace{1.5cm}- \left [ (l+1)^2+ i \,2\varpi (l+1)\right ]
315: F=0\,.
316: \label{near wave hypergeometric}
317: \end{eqnarray}
318: This wave equation is a standard hypergeometric equation
319: \cite{abramowitz}, $z(1\!-\!z)\partial_z^2
320: F+[c-(a+b+1)z]\partial_z F-ab F=0$, with
321: \begin{eqnarray}
322: & & \hspace{-0.5cm} a=l+1+i\,2\varpi \,, \qquad b=l+1 \,, \qquad
323: c=1+ i\,2\varpi \,, \nonumber \\
324: & &
325: \label{hypergeometric parameters}
326: \end{eqnarray}
327: and its most general solution in the neighborhood of $z=0$ is $A\,
328: z^{1-c} F(a-c+1,b-c+1,2-c,z)+B\, F(a,b,c,z)$. Using
329: (\ref{hypergeometric function}), one finds that the most general
330: solution of the near-region equation is
331: \begin{eqnarray}
332: \hspace{-0.5cm} R &=& A\, z^{-i\,\varpi}(1-z)^{l+1}
333: F(a-c+1,b-c+1,2-c,z)\nonumber \\
334: & & +B\,z^{i\,\varpi}(1-z)^{l+1} F(a,b,c,z) \,.
335: \label{hypergeometric solution}
336: \end{eqnarray}
337: The first term represents an ingoing wave at the horizon $z=0$,
338: while the second term represents an outgoing wave at the horizon.
339: We are working at the classical level, so there can be no outgoing
340: flux across the horizon, and thus one sets $B=0$ in
341: (\ref{hypergeometric solution}). One is now interested in the
342: large $r$, $z\rightarrow 1$, behavior of the ingoing near-region
343: solution. To achieve this aim one uses the $z \rightarrow 1-z$
344: transformation law for the hypergeometric function
345: \cite{abramowitz},
346: \begin{eqnarray}
347: & \hspace{-2cm} F(a\!-\!c\!+\!1,b\!-\!c\!+\!1,2\!-\!c,z)=
348: (1\!-\!z)^{c-a-b} & \nonumber \\
349: &\times
350: \frac{\Gamma(2-c)\Gamma(a+b-c)}{\Gamma(a-c+1)\Gamma(b-c+1)}
351: \,F(1\!-\!a,1\!-\!b,c\!-\!a\!-\!b\!+\!1,1\!-\!z) & \nonumber \\
352: & \hspace{-0.2cm}+
353: \frac{\Gamma(2-c)\Gamma(c-a-b)}{\Gamma(1-a)\Gamma(1-b)}
354: \,F(a\!-\!c\!+\!1,b\!-\!c\!+\!1,-c\!+\!a\!+\!b\!+\!1,1\!-\!z), & \nonumber \\
355: &
356: \label{transformation law}
357: \end{eqnarray}
358: and the property $F(a,b,c,0)=1$. Finally, noting that when
359: $r\rightarrow \infty$ one has $1-z\rightarrow (r_+-r_-)/r$, one
360: obtains the large $r$ behavior of the ingoing wave solution in the
361: near-region,
362: \begin{eqnarray}
363: R &\sim& A\,\Gamma(1-i\,2\varpi){\biggl [}
364: \frac{(r_+-r_-)^{-l}\,\Gamma(2l+1)}{\Gamma(l+1)\Gamma(l+1-i\,2\varpi)}\:
365: r^{l}\nonumber \\
366: & &
367: +\frac{(r_+-r_-)^{l+1}\,\Gamma(-2l-1)}{\Gamma(-l)\Gamma(-l-i\,2\varpi)}\: r^{-l-1}
368: {\biggr ]}.
369: \label{near field-large r}
370: \end{eqnarray}
371: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
372: \subsubsection{\label{sec:BH Far region}Far-region wave equation and solution}
373: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
374:
375: In the far-region, $r-r_+ \gg M$, the effects induced by the black
376: hole can be neglected ($a\sim 0$, $M \sim 0$, $\Delta \sim r^2$)
377: and the radial wave equation reduces to the wave equation of a
378: massless scalar field of frequency $\omega$ and angular momentum
379: $l$ in a flat background,
380: \begin{eqnarray}
381: \partial_r^2 (r R)+ \left [ \omega^2-l(l+1)/r^2
382: \right ] (r R)=0\,.
383: \label{far wave eq}
384: \end{eqnarray}
385: The most general solution of this equation is a linear combination
386: of Bessel functions \cite{abramowitz},
387: \begin{eqnarray}
388: R=r^{-1/2}\left [ \alpha J_{\,l+1/2}(\omega r)+ \beta
389: J_{-l-1/2}(\omega r)\right ]\,.
390: \label{far field}
391: \end{eqnarray}
392: For large $r$ this solution can be written as \cite{abramowitz}
393: \begin{eqnarray}
394: & & R \sim \sqrt{\frac{2}{\pi \omega}}\frac{1}{r}{\biggl [} \alpha
395: \sin(\omega r-l\pi/2)+ \beta \cos(\omega r+l\pi/2){\biggr ]},\nonumber \\
396: & &
397: \label{far field-large r}
398: \end{eqnarray}
399: while for small $r$ it reduces to \cite{abramowitz}
400: \begin{eqnarray}
401: R \sim \alpha\, \frac{(\omega/2)^{l+1/2}}{\Gamma(l+3/2)}\: r^{l} +
402: \beta\, \frac{(\omega/2)^{-l-1/2}}{\Gamma(-l+1/2)}\: r^{-l-1}.
403: \label{far field-small r}
404: \end{eqnarray}
405:
406: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
407: \subsubsection{\label{sec:BH Matching}Matching the near-region and the far-region solutions}
408: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
409:
410: When $M \ll r-r_+ \ll 1/\omega$, the near-region solution and the
411: far-region solution overlap, and thus one can match the large $r$
412: near-region solution (\ref{near field-large r}) with the small $r$
413: far-region solution (\ref{far field-small r}). This matching
414: yields
415: \begin{eqnarray}
416: & & \hspace{-0.5cm} A= \frac{(r_+-r_-)^l}{\Gamma(l+3/2)}
417: \frac{\Gamma(l+1)}{\Gamma(2l+1)}
418: \frac{\Gamma(l+1-i\,2\varpi)}{\Gamma(1-i\,2\varpi)}
419: \left (\frac{\omega}{2} \right )^{l+1/2}\!\!\alpha ,\nonumber \\
420: & &
421: \label{relation A-alpha}
422: \end{eqnarray}
423: \begin{eqnarray}
424: \frac{\beta}{\alpha} &=& \frac{\Gamma(-l+1/2)}{\Gamma(l+3/2)}
425: \frac{\Gamma(l+1)}{\Gamma(2l+1)}
426: \frac{\Gamma(-2l-1)}{\Gamma(-l)}
427: \frac{\Gamma(l+1-i\,2\varpi)}{\Gamma(-l-i\,2\varpi)}\nonumber \\
428: & &
429: \times \left (\frac{\omega}{2} \right
430: )^{2l+1}\!\!(r_+-r_-)^{2l+1}\,.
431: \label{relation beta-alpha}
432: \end{eqnarray}
433: Using the property of the gamma function,
434: $\Gamma(1+x)=x\Gamma(x)$, one can show that
435: $\frac{\Gamma(-l+1/2)}{\Gamma(1/2)}=\frac{(-1)^l 2^l}{(2l-1)!!}$,
436: $\frac{\Gamma(l+3/2)}{\Gamma(1/2)}=\frac{(2l+1)!!}{2^{l+1}}$,
437: $\frac{\Gamma(-2l-1\!)}{\Gamma(-l)}=\frac{(-1)^{l+1}l!}{(2l+1)!}$
438: and
439: $\frac{\Gamma(l+1-i\,2\varpi)}{\Gamma(-l-i\,2\varpi)}=i\,(-1)^{l+1}2\varpi\prod_{k=1}^l
440: (k^2+4\varpi^2)$. Then, the matching condition
441: (\ref{relation beta-alpha}) yields
442: \begin{eqnarray}
443: \frac{\beta}{\alpha} &=& i\, 2\varpi\, \frac{(-1)^l}{2l+1} \left (
444: \frac{l!}{(2l-1)!!} \right )^2
445: \frac{(r_+ -r_-)^{2l+1}}{(2l)! (2l+1)!} \nonumber \\
446: & & \times \, \left ( \prod_{k=1}^l (k^2+4\varpi^2) \right )
447: \omega^{2l+1}\,.
448: \label{relation beta-alpha 2}
449: \end{eqnarray}
450:
451: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
452: \subsubsection{\label{sec:BH Mirror}Mirror condition. Properties of the unstable modes}
453: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
454:
455: If one puts a mirror near infinity at a radius $r=r_0$, the scalar
456: field must vanish at the mirror surface. Thus, setting
457: the radial field (\ref{far field}) to zero yields the
458: extra condition between the amplitudes $\alpha$ and $\beta$, and
459: the position of the mirror $r_0$,
460: \begin{eqnarray}
461: \frac{\beta}{\alpha}=- \frac{J_{l+1/2}(\omega r_0)}{J_{-l-1/2}(\omega r_0)} \,.
462: \label{relation beta-alpha mirror}
463: \end{eqnarray}
464: This mirror condition together with the matching condition
465: (\ref{relation beta-alpha 2}) yield a condition
466: between the position of the mirror and the frequency of the scalar
467: wave,
468: \begin{eqnarray}
469: & & \hspace{-0.7cm}\frac{J_{l+1/2}(\omega r_0)}{J_{-l-1/2}(\omega r_0)}=
470: i(-1)^{l+1} \,\varpi\, \frac{2}{2l+1}
471: \left ( \frac{l!}{(2l-1)!!} \right )^2 \nonumber \\
472: & &
473: \times \, \frac{(r_+ -r_-)^{2l+1}}{(2l)! (2l+1)!}\left ( \prod_{k=1}^l (k^2+4\varpi^2)
474: \right ) \omega^{2l+1} \,.
475: \label{mirror eigen}
476: \end{eqnarray}
477: The solution of (\ref{mirror eigen}) can be
478: found in the approximation that applies suitably to this problem,
479: namely, $\omega\ll 1$, and ${\rm Re}(\omega)\gg {\rm Im}(\omega)$.
480: For very small $\omega$, the r.h.s. of (\ref{mirror eigen}) is very small and can be
481: assumed to be zero in the first approximation for $\omega$.
482: This yields
483: \begin{equation}
484: J_{l+1/2}(\omega r_0)=0\,,
485: \label{zerob}
486: \end{equation}
487: which has well studied (real) solutions \cite{abramowitz}.
488: We shall label the solutions of (\ref{zerob}) as $j_{l+1/2,n}$:
489: \begin{equation}
490: J_{l+1/2}(\omega r_0)=0 \Leftrightarrow \omega r_0=j_{l+1/2,n}\,,
491: \label{solzerob}
492: \end{equation}
493: where $n$ is a non-negative integer number.
494: We now assume that the complete solution to (\ref{mirror eigen}) can
495: be written as $\omega\sim j_{l+1/2,n}/r_0+i\tilde{\delta}/r_0$, where we
496: have inserted a small imaginary part proportional to
497: $\tilde{\delta}\ll 1$.
498: One then has, from (\ref{mirror eigen})
499: \begin{eqnarray}
500: & & \hspace{-1cm} \frac{J_{l+1/2}(j_{l+1/2,n}+i\tilde{\delta})}
501: {J_{-l-1/2}(j_{l+1/2,n}+i\tilde{\delta})}= i(-1)^{l+1} \,\varpi\,
502: \left ( \frac{l!}{(2l-1)!!} \right )^2 \nonumber \\
503: & & \hspace{-0.5cm}
504: \times \,\frac{2}{2l+1}\, \frac{(r_+ -r_-)^{2l+1}}{(2l)! (2l+1)!}\left ( \prod_{k=1}^l (k^2+4\varpi^2)
505: \right ) \omega^{2l+1} \,.
506: \label{mirror eigen2}
507: \end{eqnarray}
508: Now, we can use, for small $\tilde{\delta}$ the Taylor expansion of the l.h.s.,
509: \begin{equation}
510: \frac{J_{l+1/2}(j_{l+1/2,n}+i\tilde{\delta})}{J_{-l-1/2}(j_{l+1/2,n}+i\tilde{\delta})}
511: \sim i\,\tilde{\delta}\, \frac{J^{\prime}_{l+1/2}(j_{l+1/2,n})}{J_{-l-1/2}(j_{l+1/2,n})}
512: \label{app}
513: \end{equation}
514: The quantities $j_{l+1/2,n}\,,J^{\prime}_{l+1/2}(j_{l+1/2,n})\,,J_{-l-1/2}(j_{l+1/2,n})$
515: are tabulated in \cite{abramowitz}, and can also easily be extracted using Mathematica.
516: Here it is important to note that
517: $J'_{l+1/2}(j_{l+1/2,n})$ and $(-1)^{l}J_{-l-1/2}(j_{l+1/2,n})$ always have the same sign.
518: Furthermore, for large overtone $n$, $j_{l+1/2,n} \sim (n+l/2)\pi$.
519: The frequencies of the scalar wave that are allowed by
520: the presence of the mirror located at $r=r_0$ (BQN frequencies) are then
521: \begin{eqnarray}
522: \omega_{BQN} \simeq \frac{j_{l+1/2,n}}{r_0}+i\delta \,,
523: \label{mirror frequencies}
524: \end{eqnarray}
525: where $n$ is a non-negative integer number, labelling the mode overtone number.
526: For example, the fundamental mode corresponds to $n=0$.
527: In (\ref{mirror frequencies}), $\delta={\rm Im}[\omega_{BQN}]$ is obtained by substituting
528: (\ref{app}) in (\ref{mirror eigen2}),
529: \begin{eqnarray}
530: \delta \simeq -\gamma
531: \frac{(-1)^{l}J_{-l-1/2}(j_{l+1/2,n})}{J^{\prime}_{l+1/2}(j_{l+1/2,n})}
532: \frac{j_{l+1/2,n}/r_0-m\Omega}{r_0^{\:2(l+1)}}\,,
533: \label{mirror frequencies imaginary}
534: \end{eqnarray}
535: where
536: \begin{eqnarray}
537: \gamma &\equiv &
538: \left ( \frac{l!}{(2l-1)!!} \right )^2
539: \frac{r_+^2(r_+ -r_-)^{2l}}{(2l)! (2l+1)!} \nonumber \\
540: & &
541: \times \,\frac{2}{2l+1} \left ( \prod_{k=1}^l (k^2+4\varpi^2)
542: \right ) \left [j_{l+1/2,n}\right ]^{2l+1}.
543: \label{mirror frequencies imaginary2}
544: \end{eqnarray}
545: Notice that $\delta$ is very small for large $r_0$ and thus satisfies
546: the conditions that go with the approximation used, ${\rm Re}(\omega)\gg {\rm Im}(\omega)$.
547: Equations (\ref{mirror frequencies}) and
548: (\ref{mirror frequencies imaginary}) constitute the main results of this section.
549: Two important features of this system, black hole plus mirror, can already be
550: read from the equations above:
551: first, from equations (\ref{mirror frequencies}) and (\ref{mirror frequencies imaginary})
552: one has,
553: \begin{equation}
554: \delta \propto -({\rm Re}[\omega_{BQN}]-m\Omega)\,.
555: \label{imag}
556: \end{equation}
557: Therefore,
558: $\delta>0$ for ${\rm Re}[\omega_{BQN}]<m \Omega$, and $\delta<0$ for
559: ${\rm Re}[\omega_{BQN}]>m \Omega$. The scalar field $\Phi$ has the
560: time dependence $e^{-i\omega t}=e^{-i{\rm Re}(\omega) t}e^{\delta t}$,
561: which implies that for ${\rm Re}[\omega_{BQN}]<m \Omega$, the
562: amplitude of the field grows exponentially and the BQNM becomes
563: unstable, with a growth timescale given by $\tau=1/\delta$.
564: In this case, we see that the system behaves in fact as a bomb,
565: a black hole bomb.
566: Second,
567: \begin{equation}
568: {\rm Re}[\omega_{BQN}] = \frac{j_{l+1/2,n}}{r_0} \,,
569: \label{re}
570: \end{equation}
571: showing that the wave frequency is proportional to the
572: inverse of the mirror's radius. As one decreases the
573: distance at which the mirror is located, the allowed wave
574: frequency increases, and there will thus be a critical radius
575: at which the BQN frequency no longer satisfies the superradiant
576: condition (\ref{super}).
577: Notice also that ${\rm Re}[\omega_{BQN}]$ as given by (\ref{re})
578: is equal to the normal mode frequencies of a spherical
579: mirror in a flat spacetime \cite{landau}.
580: In the next subsection, when the
581: numerical results will also be available, we will return to this
582: discussion.
583:
584: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
585: \subsection{\label{sec:BH bomb numerical}The black hole bomb. Numerical
586: approach}
587: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
588:
589: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
590: \subsubsection{Numerical procedure}
591: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
592: In the numerical calculation for determining oscillation
593: frequencies of the modes, we use the same function as that defined
594: by Teukolsky \cite{teu2} given by (see also \cite{hughes})
595: %
596: \begin{equation}
597: Y=(r^2+a^2)^{1/2}R\, .
598: \end{equation}
599: %
600: Then, Teukolsky equation becomes a canonical equation, given by
601: %
602: \begin{equation}
603: {d^2\over dr_*^2}Y+VY=0\,, \label{mT-eq}
604: \end{equation}
605: %
606: where
607: %
608: \begin{eqnarray}
609: V={K^2-\lambda\Delta\over(r^2+a^2)^2}-G^2-{d\over dr_*}G\,,
610: \end{eqnarray}
611: %
612: with $K=(r^2+a^2)\omega-am$, and $G=r\Delta(r^2+a^2)^{-2}$.
613: For the separation constant $\lambda=A_{lm}+a^2 \omega
614: ^2-2am\omega$, we make use of a well known series expansion in
615: $a\omega$, given by
616: %
617: \begin{eqnarray}
618: \lambda=a^2\omega^2-2am\omega+
619: \sum_{i=0}^{\infty}\,_0f^{lm}_i(a\omega)^i\,,
620: \end{eqnarray}
621: %
622: where $_0f^{lm}_i$ is the expansion coefficient (for the explicit
623: form, see, e.g., \cite{seidel}). In this study, we keep the terms
624: in the expansion up to an order of $(a\omega)^2$. As mentioned,
625: near the horizon, the physically acceptable solution of equation
626: (\ref{mT-eq}) is the incoming wave solution, given by
627: %
628: \begin{eqnarray}
629: Y=e^{-i(\omega-m\Omega) r_*}[y_0+y_1 (r-r_+)+y_2 (r-r_+)^2+\cdots]\,,
630: \label{asyp-sol}
631: \end{eqnarray}
632: %
633: where $y_i$ is the expansion coefficient
634: determined by $\omega$ and $y_0$. Here, we do not show explicit
635: expression for the $y_i$'s because it is straightforward to derive it.
636:
637:
638: In order to obtain the proper solutions numerically, by using a
639: Runge-Kutta method, we start integrating the differential equation
640: (\ref{mT-eq}) outward from $r=r_+(1+10^{-5})$ with the asymptotic
641: solution (\ref{asyp-sol}). We then stop the integration at the
642: radius of the mirror, $r=r_0$, and get the value of the wave
643: function at $r=r_0$, which is considered a function of the
644: frequency, $Y(r_0,\omega)$. If $Y(r_0,\omega)=0$, the solution
645: satisfies the boundary condition of perfect reflection due to
646: the mirror and the frequency $\omega$ is a BQN frequency, which we
647: label as $\omega _{BQN}$. In other words, the dispersion relation
648: of our problem is given by the equation $Y(r_0,\omega _{BQN})=0$.
649: In order to solve the algebraic equation $Y(r_0,\omega _{BQN})=0$
650: iteratively, we use a secant method. Here, it is important to note
651: that if the mode is stable or ${\rm Im}(\omega _{BQN}) <0$, the
652: asymptotic solution (\ref{asyp-sol}) diverges exponentially and
653: another independent solution, which is unphysical, damps
654: exponentially as $r_*\rightarrow -\infty$.
655: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
656: \subsubsection{Numerical results}
657: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
658: Our numerical results are summarized in
659: Figs. \ref{fig:BHBa}-\ref{fig:BHBg}. As we remarked earlier, we
660: only show the data corresponding to the unstable BQNMs.
661: We have also found the
662: stable modes, but since they lead to no bomb we refrain from presenting them.
663: >From the figures we confirm the analytical expectations. In addition we can
664: discuss growing timescales, oscillation frequencies, energy extracted and efficiencies
665: with great accuracy.
666: Figure \ref{fig:BHBa} plots the imaginary part of the BQN frequency for the
667: fundamental BQNM as a function of the mirror's location $r_0$ and
668: the rotation parameter $a$.
669: In figure \ref{fig:BHBb}, we show the real part of the BQN frequency for the
670: fundamental BQNM also as a function of $r_0$ and $a$.
671: Supporting the analytical results, figure \ref{fig:BHBa} shows that:
672: (i) The instability is weaker (the
673: growing timescale $\tau=\frac{1}{{\rm Im}[\omega _{BQN}]}$ is larger)
674: for larger mirror radius, meaning that ${\rm Im}[\omega_{BQN}]$ decreases
675: as $r_0$ increases. This is also expected on physical grounds, as
676: was remarked in \cite{press}, if one views the
677: process as one of successive amplifications and reflections on the mirror.
678: (ii) As one decreases $r_0$ the instability gets stronger, as
679: expected, but surprisingly, suddenly the BQNM is no longer
680: unstable. The imaginary component of $\omega_{BQN}$ drops from its
681: maximum value to zero, and the mode becomes stable at a critical
682: radius $r_0^{\rm crit}$.
683: Physically, this happens because superradiance generates wavelengths with
684: $\lambda > 1/\Omega$. So the mirror at a distance $r_0$ will ``see'' these wavelengths
685: if $r_0>r_0^{\rm crit}\sim \lambda\sim 1/\Omega$.
686: One can improve the estimate for $r_0^{\rm crit}$ using equations
687: (\ref{super}), (\ref{imag}), and (\ref{re}), yielding
688: $r_0^{\rm crit} \sim \frac{j_{l+1/2,n}}{m\Omega}$.
689: This estimate for the critical radius matches very well
690: with our numerical data, even though the analytical calculation
691: is a large wavelength approximation.
692: In fact, to a great accuracy $r_0^{\rm crit}$
693: is given by the root of ${\rm Re}[\omega(r_0^{\rm crit})]-m\Omega=0$,
694: as is shown in figure \ref{fig:BHBc}.
695: Also supporting the analytical results, figure \ref{fig:BHBb} shows that
696: ${\rm Re}[\omega_{BQN}]$ behaves as $\frac{1}{r_0}$, which is consistent with
697: equation (\ref{mirror frequencies}). This means that it is
698: indeed the mirror which selects the allowed vibrating frequencies.
699: The results for higher mode number $n$ is shown in figures
700: \ref{fig:BHBd} and \ref{fig:BHBe}, and the behaviour agrees with the
701: picture provided by the analytical approximation.
702: The agreement between the analytical and numerical results is best seen in
703: Table \ref{tab:comp}, where we show the lowest BQN frequencies obtained
704: using both methods.
705: In figures \ref{fig:BHBf} and \ref{fig:BHBg} we show the numerical results referring
706: to different values of the angular number $l$ and $m$.
707: Our numerical results indicate
708: that $\frac{1}{{\rm Im}[\omega _{BQN}]} \sim r_0^{-2(l+1)}$, in
709: agreement with the analytical result, equation (\ref{mirror
710: frequencies imaginary}). The numerical results also indicate that the
711: oscillating frequencies (${\rm Re}[\omega_{BQN}]$) scale with $l$,
712: more precisely, ${\rm Re[\omega_{BQN}]}$ behaves as ${\rm
713: Re[\omega_{BQN}]} \sim \pi/r_0(n+l/2)$. This behavior is also
714: predicted by the analytical study.
715:
716: \begin{figure}
717: \centerline{\includegraphics[width=7 cm,height=7 cm] {BHBa.eps}}
718: \caption{The imaginary part of the fundamental $(n=0)$ BQN frequency
719: ($\omega_{BQN}$) as a function of the mirror's location $r_0$ is plotted.
720: The plot refers to a $l=m=1$ wave. It is also
721: shown the dependence on the rotation parameter $a$. One sees that for
722: $r_0$ greater than a critical value, $r_{0}^{\rm crit}$, depending
723: on $a$, there is the possibility of building a bomb. Moreover, the
724: imaginary component of the BQN frequency decreases abruptly from its
725: maximum value to zero at $r_{0}^{\rm crit}$. For $r_0<r_{0}^{\rm
726: crit}$ the BQNM is stable. Tracking the mode to yet smaller distances
727: shows that indeed it remains stable (the imaginary part of $\omega_{BQN}$
728: is negative).}
729: \label{fig:BHBa}
730: \end{figure}
731: \vskip 1mm
732: \begin{figure}
733: \centerline{\includegraphics[width=7 cm,height=7 cm] {BHBb.eps}}
734: \caption{The real part of the fundamental $(n=0)$ BQN frequency
735: ($\omega_{BQN}$) as a function of the mirror's location $r_0$ is
736: plotted. The plot refers to a $l=m=1$ wave.
737: There is no perceptive $a$-dependence (as matter of fact
738: there is a very small $a$-dependence but too small to be noticeable).
739: Thus, the oscillation frequency basically depends only on $r_0$, and
740: for large $r_0$ goes as $1/r_0$, as predicted by the analytical
741: formula (\ref{mirror frequencies}).
742: The dots indicate $r_{0}^{\rm crit}$ (cf. Fig. \ref{fig:BHBa}).}
743: \label{fig:BHBb}
744: \end{figure}
745: \vskip 1mm
746: \begin{figure}
747: \centerline{\includegraphics[width=7 cm,height=7 cm] {BHBc.eps}}
748: \caption{This figure helps in understanding why the instability
749: disappears for $r_0$ smaller than a certain critical value. The
750: condition for superradiance is $\omega-m\Omega <0$.
751: Since $\omega$ goes as $1/r_0$ (check Fig. \ref{fig:BHBb}), then it
752: is expected that the condition will stop to hold at a critical
753: $r_0$. This is clearly seen here. Note that the critical value
754: of $r_0$ is in excellent agreement with that shown in
755: Fig. \ref{fig:BHBa}.} \label{fig:BHBc}
756: \end{figure}
757: \vskip 1mm
758:
759:
760: \begin{figure}
761: \centerline{\includegraphics[width=7 cm,height=7 cm] {BHBd.eps}}
762: \caption{The imaginary part of the BQN frequency as a function of
763: $r_0$, for $a=0.4$ and for the three lowest overtones $n$, for
764: $l=m=1$. As expected from the general arguments presented, higher
765: overtones get stable at larger distances, and attain a smaller
766: maximum growing rate. } \label{fig:BHBd}
767: \end{figure}
768: \vskip 1mm
769: \begin{figure}
770: \centerline{\includegraphics[width=7 cm,height=7 cm] {BHBe.eps}}
771: \caption{Same as Fig. \ref{fig:BHBd}, but for the real part of
772: $\omega_{BQN}$.} \label{fig:BHBe}
773: \end{figure}
774: \vskip 1mm
775:
776:
777: \begin{figure}
778: \centerline{\includegraphics[width=7 cm,height=7 cm] {BHBf.eps}}
779: \caption{The imaginary part of the fundamental $\omega _{BQN}$ for an
780: $a=0.4$ black hole, as a function of the mirror's location $r_0$ here
781: shown for some values of $l,m$. Furthermore, as is evident from this figure and also
782: as could be anticipated, the larger $m$ the smaller $r_0$ can be,
783: still displaying instability. Note however that the maximum
784: instability is larger for the $m=1$ mode. This is a general feature.
785: The imaginary part of the frequency seems to behave as ${\rm
786: Im}[\omega_{BQN}] \sim r_{0}^{-2(l+1)}$, which agrees with the
787: analytical prediction (\ref{mirror frequencies imaginary}).
788: } \label{fig:BHBf}
789: \end{figure}
790: \vskip 1mm
791: \begin{figure}
792: \centerline{\includegraphics[width=7 cm,height=7 cm] {BHBg.eps}}
793: \caption{Same as Fig. \ref{fig:BHBf} but for the real part. Note that
794: there is an $l$-dependence of the real part of the frequency. On the
795: other hand, there is no noticeable $m$-dependence, in accordance with
796: the analytical result (\ref{mirror frequencies}).}
797: \label{fig:BHBg}
798: \end{figure}
799: \vskip 1mm
800:
801: \begin{table}
802: \caption{\label{tab:comp} The fundamental BQN frequencies
803: for a black hole with $a=0.8M$ and a mirror placed at $r_0=100M$.
804: The data corresponds to the $l=m$ modes, and the frequency is measured in units of
805: the mass $M$ of the black hole. We present both
806: the numerical ($\omega_{BQN}^N$) and analytical ($\omega_{BQN}^A$)
807: results. Notice that the agreement between the two is very good, and
808: it gets better as $l$ increases. Also, we have checked that for very
809: large $r_0$ the two yield basically the same results. }
810: \begin{ruledtabular}
811: \begin{tabular}{l|l|l} \hline
812: \multicolumn{1}{c}{} &
813: \multicolumn{2}{c}{ $a=0.8M\,,r_0=100M$}\\ \hline
814: $l$ &$\omega_{BQN}^N$:&$\omega_{BQN}^A$:\\ \hline
815: 1 &$8.75\times 10^{-2}+1.19\times 10^{-7}i$ &$8.99\times 10^{-2}+1.41\times 10^{-7}i$ \\ \hline
816: 2 &$1.13\times 10^{-1}+6.77\times 10^{-12}i$ &$1.15\times 10^{-1}+6.89\times 10^{-12}i$ \\ \hline
817: 3 &$1.37\times 10^{-1}+2.45\times 10^{-16}i$ &$1.39\times 10^{-1}+2.26\times 10^{-16}i$ \\ \hline
818: \end{tabular}
819: \end{ruledtabular}
820: \end{table}
821: \vskip 1mm
822: Let us now take Press and Teukolsky example of a black hole with mass $M=1
823: M_{\odot}$ \cite{press}. We are now in a position to make a much
824: improved quantitative analysis. We take $a=0.8\,M$, a large angular
825: momentum so that we make a full use of our results. In addition, to
826: better take advantage of the whole process, one should place the
827: mirror at a position near the point of maximum growing rate, but
828: farther. Thus, for the example, $r_0 \sim 22M \sim 33 \,{\rm Km}$ (see
829: Fig. \ref{fig:BHBa}). This gives a growing timescale of about $\tau
830: \sim 0.6 \,{\rm s}$ (see also Fig. \ref{fig:BHBa}), which means that
831: every $0.6 \,{\rm s}$ the amplitude of the field gets approximately
832: doubled. This means that at the end of 13 seconds the initial
833: amplitude of the wave has grown to $10^7$ of its initial value, and
834: that thus the energy content is $10^{14}$ times greater than the
835: initial perturbation. We consider there are no losses through the
836: mirror and assume the process to be adiabatic. Using the first law of
837: thermodynamics one can then set $\Delta M\sim \Omega \Delta J$, where
838: $\Delta M$ and $\Delta J$ are the changes in mass and angular momentum
839: of the black hole in this process, respectively. Now, the black hole
840: is losing angular momentum in each superradiant scattering. Thus $a$
841: decreases. If we go to figure \ref{fig:BHBa} we see that $r_0^{\rm
842: crit}$ increases with decreasing $a$. At a certain stage $r_0^{\rm
843: crit}$ coincides with the position of the mirror at $r_0$, at which
844: point there is no more possibility of superradiance. The process is
845: finished. Thus from figure \ref{fig:BHBa}, or more accurately from our
846: numerics, one can find $\Delta\,a$, and thus $\Delta\,J$. Then
847: $\Delta\,M$ follows from $\Delta\,M\sim \Omega \Delta\,J$. In the
848: example this gives a total amount $\Delta\, M\sim 0.01 M$ of extracted
849: energy before the bomb stops functioning. The process has thus an
850: efficiency of 1\%, about the same order of magnitude as the efficiency
851: of nuclear fusion of hydrogen burning into helium ($\sim 0.7$\%). If,
852: instead, the mirror is placed at $r_0 \sim 200 M \sim \, 300{\rm Km}$
853: one still gets a good growing timescale of about $15$ min. This means
854: that at the end of $6$ hours the initial amplitude of the wave has
855: grown to $10^7$ of its initial value. In this case
856: $\Delta\,M \sim 0.1\,M$, with a 10\% efficiency, and one can show that the
857: efficiency grows with mirror radius $r_0$. Since the cost of mirror
858: construction scales as $r_0^2$, we see that small mirrors are more
859: effective. One can give other interesting examples. For a black hole
860: at the center of a galaxy, with mass $M \sim 10^8 M_{\odot}$, $a=0.8
861: M$, and $r_0=22\,M$, the maximum growing timescale is of the order of
862: $2$ years. Another interesting situation happens when the black hole
863: has a mass of the order of the Earth mass. In this case, by placing
864: the mirror at $r_0= 1\,{\rm m}$ one gets a growing timescale of about
865: $0.02 \, {\rm s}$. At the other end of the black hole spectrum one
866: can consider Planck size black holes.
867: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
868: \subsection{\label{sec:Zeldovich cylinder}Zel'dovich's cylinder surrounded by a mirror}
869: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
870: As a corollary, we discuss here electromagnetic
871: superradiance in the presence of a cylindrical rotating body, a situation
872: first discussed by Zel'dovich \cite{zel1}.
873: He noted that by surrounding this rotating body with a reflecting mirror one
874: could amplify the radiation, much as the black hole bomb process just
875: described. Bekenstein and Schiffer \cite{bekenstein} have recently
876: elaborated on this.
877: An independent analytical approximation, similar in all respects to the one we
878: discussed earlier in the black hole bomb context, can also be applied here
879: for finding the BQN frequencies of this system (conducting cylinder plus reflecting mirror),
880: and leads to almost the same results as for the black hole bomb.
881: The imaginary component of $\omega_{BQN}$ is $\delta\propto -({\rm
882: Re}[\omega_{BQN}]-m\Omega)$. The electromagnetic field has the time
883: dependence $e^{-i\omega t}=e^{-i{\rm Re}(\omega) t}e^{\delta t}$ and
884: thus, for ${\rm Re}[\omega_{BQN}]<m \Omega$, the amplitude of the
885: field grows exponentially with time and the mode becomes unstable,
886: with a growth timescale given by $\tau=1/\delta$. Second, ${\rm
887: Re}[\omega_{BQN}] \propto 1/r_0$, i.e., the wave frequency is
888: proportional to the inverse of the mirror's radius, as it was for the
889: black hole bomb. Therefore, as one decreases the distance at which
890: the mirror is located, the allowed wave frequency increases, and again there
891: will be a critical radius at which the frequency no longer
892: satisfies the superradiant condition (\ref{super}). If one tries to
893: use the system as it is, it would be almost impossible to observe superradiance in the
894: laboratory. Take as an example a cylinder with a radius $R=10\, {\rm
895: cm}$, rotating at a frequency $\Omega=2\pi \times 10^2\, {\rm
896: s}^{-1}$, and a surrounding mirror with radius $r_0= 20 \, {\rm
897: cm}$. For the system to be unstable and experimentally detectable, the
898: minimum mirror radius is given by $r_{\rm crit} \sim
899: \frac{c}{m\Omega}$ (where we have reinstated the velocity of light
900: $c$), which yields $r_{\rm crit} \sim 1000 \,{\rm Km}$, for a $m=1$
901: wave. It seems impossible to use this apparatus to measure
902: superradiance experimentally. A way out of this problem may be the
903: one suggested in \cite{bekenstein}: to surround the conducting mirror
904: by a material with a low velocity of light. In this case the critical
905: radius would certainly decrease, although further investigation is
906: needed in order to ascertain what kind of material should be used.
907:
908: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
909: \section{Are Kerr-AdS black holes unstable?}
910: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
911: A spacetime with a naturally incorporated mirror in it is anti-de
912: Sitter (AdS) spacetime, which has attracted a great deal of attention
913: recently due to the AdS/CFT correspondence and other matters. As is
914: well known, anti-de Sitter (AdS) space behaves effectively as a box,
915: in other words, the AdS infinity works as a mirror wall. Thus, one
916: might worry that Kerr-AdS black holes could behave as the black hole
917: bomb just described, and that they would be unstable. Hawking and
918: Reall \cite{hawking} have shown that, at least for large Kerr-AdS
919: black holes, this instability is not present. The stability of large
920: Kerr-AdS black holes in four and higher dimensions can be understood
921: in yet another way, using the knowledge one acquired from the black
922: hole bomb study. The black hole rotation is constrained to be
923: $a<\ell$, where $\ell$ is the AdS radius \cite{hawking}.
924: Large black holes are the ones for which $r_+>>\ell$. In
925: this case the angular velocity of the horizon $\Omega=
926: \frac{a}{r_+^2+a^2}(1-a^2/\ell^2)$ goes to zero and one expects that
927: the rotation plays a neglecting role in this regime, with the results
928: found for the non-rotating AdS black hole \cite{horo,cardoso} still
929: holding approximately when the rotation is non-zero. The
930: characteristic quasinormal frequencies for large, non-rotating AdS
931: black holes were computed in \cite{horo,cardoso} showing that the real
932: part scales with $r_+$. Now, since $\Omega \rightarrow 0$ for large
933: black holes and the QNMs have a very large real part, one can
934: understand why there is no instability: superradiant modes are simply
935: not excited, as the condition for superradiance, $\omega<m\Omega$,
936: cannot be fulfilled. We could try to evade this by going to higher
937: values of $m$, but then $l$ has also to be large ($l\geq m$). However,
938: for large $l$'s, the real part of the QNMs is known to scale with $l$
939: \cite{cardosol}, thus the condition for superradiance is never
940: fulfilled.
941: What about small Kerr-AdS black holes, $r_+<< \ell\,$? Considering
942: the case $a$ small, $a<<r_+$ say, is enough for our purposes.
943: In this situation the
944: horizon's angular velocity scales as the inverse of $r_+$, and it
945: can be made arbitrarily large.
946: Although the effect of rotation cannot be neglected in this case,
947: the results of the QNM analysis for non-rotating AdS black holes
948: give some hints on what may happen.
949: For small non-rotating AdS black holes, the QN
950: frequencies have a real part that goes to a constant, independent of $r_+$,
951: whereas the imaginary part goes to zero \cite{cardoso}.
952: If we add a small angular momentum per unit mass $a$
953: to the black hole, we do not expect
954: the real part of the QN frequency to grow significantly.
955: But, since $\Omega$ is very large anyway, the superradiance condition
956: $\omega<m\Omega$ will most likely be fulfilled.
957: Therefore we expect to be
958: possible to excite the superradiant instability in these
959: spacetimes.
960: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
961: \section{Conclusions}
962: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
963: To conclude, we have investigated the black hole bomb thoroughly, by
964: analytical means in the long wavelength limit, and numerically. We
965: have provided both analytical and numerical accurate estimates for growing
966: timescales and oscillation frequencies of the corresponding unstable
967: Boxed Quasinormal Modes (BQNMs). Both
968: results agree and yield consistent answers. An important feature born
969: out in this work is that there is a minimum distance at which the
970: mirror must be located in order for the system to become unstable and
971: for the bomb to work. Basically this is because the mirror selects
972: the frequencies that may be excited. For distances smaller than this,
973: the system is stable and the perturbation dies off exponentially.
974: This minimum distance increases as the rotation parameter decreases.
975: We have given an explicit example where such a system works yielding a
976: reliable source of energy. By using appropriately this extracted energy
977: one could perhaps build a black hole power plant.
978: Although we have worked only with zero spin (scalar) waves,
979: we expect that the general features for other spins will be the
980: same. Moreover, it is known that a charged black hole, even in the
981: absence of rotation, provides a background for superradiance, as long
982: as the impinging wave is a bosonic charged wave (fermions do not
983: exhibit superradiance). In this case, the critical
984: radius should be of order $r_{\rm crit}\sim \frac{1}{e\Phi}$, where
985: $e$ is the charge of the scalar particle and $\Phi$ is the black hole's
986: electromagnetic potential.
987:
988: We have also shown that a mirror
989: surrounding Zel'dovich's rotating cylinder leads to a system that
990: displays the same instabilities as the black hole bomb. However, for the
991: instability to be triggered in an Earth based experiment, some improvments
992: must be made. In particular the cylinder should be surrounded by a material
993: with a low light velocity, since otherwise it would require huge mirror radius or huge
994: rotating frequencies.
995:
996: The study of the black hole bomb, and of the associated instabilities
997: allows one to better understand the absence of superradiance in large
998: Kerr-AdS black holes \cite{hawking} and moreover to expect that small
999: Kerr-AdS black holes will be unstable.
1000:
1001: Finally, it seems worth investigating whether or not this kind of
1002: black hole bomb is possible in TeV-scale gravity. In these scenarios,
1003: one has four non-compact dimensions and $n$ extra compact
1004: dimensions. It might be possible that these extra compactified
1005: dimensions work as a reflecting mirror, and therefore rotating black
1006: holes in $4+n$ dimensions could turn out to be unstable.
1007:
1008:
1009: We would also like to make a remark on a possible astrophysical
1010: application of this
1011: black hole-reflecting wall system. It has been proposed in
1012: \cite{putten}, and further discussed in \cite{aguirre}, that the
1013: superradiant amplification process might provide the energy necessary
1014: to feed the highly energetic gamma-ray burst. Magnetosonic plasma
1015: waves, generated in the accreting plasma around an astrophysical black
1016: hole, might enter in the waveguide cavity located between the
1017: gravitational potential barrier of the black hole and the inner edge
1018: of the accretion disk. Once there, the inner boundary of the accretion
1019: disk might act as a reflecting wall, and waves can then suffer
1020: multiple reflection and superradiant amplification, increasing their
1021: energy. This energy increase will continue until the magnetosphere
1022: that surrounds the system is no longer capable of supporting the
1023: energy pressure in the waveguide cavity. The wave energy would then be
1024: released in a burst, and collimated into a relativistic jet by, e.g.,
1025: the Blandford-Znajek process \cite{blandford}, and finally transferred
1026: into the observed $\gamma$ -ray photons, as described by the Fireball
1027: model \cite{piran}.
1028: In order to better compare the energy and timescales derived from the
1029: superradiant model with the observational data taken from gamma-ray bursts,
1030: it is appropriate to apply the methods of this paper to superradiant
1031: cavities that have an accreting matter configuration of a torus or disk.
1032:
1033:
1034:
1035:
1036: \vskip 2mm
1037:
1038:
1039: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1040: \section*{Acknowledgements}
1041: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1042: The authors acknowledge stimulating discussions with Ana Sousa
1043: and Jorge Dias de Deus, which have inspired us to do this work, and
1044: thank Emanuele Berti and Jacob Bekenstein for a critical reading of
1045: the manuscript and for useful suggestions.
1046: This work was partially funded by Funda\c c\~ao para a Ci\^encia e
1047: Tecnologia (FCT) -- Portugal through project CERN/FNU/43797/2001.
1048: V.C. and O.J.C.D. acknowledge finantial support from FCT
1049: through grant SFRH/BPD/2003. S.Y. acknowledges financial support
1050: from FCT through project SAPIENS 36280/99.
1051: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1052: \begin{thebibliography}{99}
1053:
1054: \bibitem{manogue} For a very good review on the subject as well as a
1055: careful explanation of various misinterpretations that have
1056: appeared over the years concerning the Klein paradox, we refer the
1057: reader to: C. A. Manogue, Annals of Physics {\bf 181}, 261 (1988).
1058:
1059: \bibitem{greiner} W. Greiner, B. M\"uller and J. Rafelski,
1060: {\it Quantum electrodynamics of strong fields}, (Springer-Verlag,
1061: Berlin, 1985).
1062:
1063: \bibitem{zel1} Ya. B. Zel'dovich,
1064: Pis'ma Zh. Eksp. Teor. Fiz. {\bf 14}, 270 (1971)
1065: [JETP Lett. {\bf 14}, 180 (1971)];
1066: Zh. Eksp. Teor. Fiz {\bf 62}, 2076 (1972)
1067: [Sov. Phys. JETP {\bf 35}, 1085 (1972)].
1068:
1069:
1070: \bibitem{bardeen} J. M. Bardeen, W. H. Press and S. A. Teukolsky,
1071: Astrophys. J. {\bf 178}, 347 (1972).
1072:
1073: \bibitem{staro1} A. A. Starobinsky,
1074: Zh. Eksp. Teor. Fiz {\bf 64}, 48 (1973)
1075: [Sov. Phys. JETP {\bf 37}, 28 (1973)];
1076: A. A. Starobinsky and S. M. Churilov,
1077: Zh. Eksp. Teor. Fiz {\bf 65}, 3 (1973)
1078: [Sov. Phys. JETP {\bf 38}, 1 (1973)].
1079:
1080: \bibitem{teu} S. A. Teukolsky and W. H. Press,
1081: Astrophys. J. {\bf 193}, 443 (1974).
1082:
1083: \bibitem{press} W. H. Press and S. A. Teukolsky,
1084: Nature {\bf 238}, 211 (1972).
1085:
1086: \bibitem{damour} T. Damour, N. Deruelle and R. Ruffini,
1087: Lett. Nuovo Cimento {\bf 15}, 257 (1976).
1088:
1089: \bibitem{detweiler} S. Detweiler,
1090: Phys. Rev. D {\bf 22}, 2323 (1980);
1091: T. M. Zouros and D. M. Eardley,
1092: Annals of Physics {\bf 118}, 139 (1979);
1093: H. Furuhashi and Y. Nambu, gr-qc/0402037.
1094:
1095: \bibitem{hawking} S. W. Hawking and H. S. Reall,
1096: Phys. Rev. D {\bf 61}, 024014 (1999).
1097:
1098: \bibitem{brill} D. R. Brill, P. L. Chrzanowski, C. M. Pereira, E. D. Fackerell and J. R. Ipser,
1099: Phys. Rev. D {\bf 5}, 1913 (1972); S. A. Teukolsky, Phys. Rev.
1100: Lett {\bf 29}, 1114 (1972).
1101:
1102: \bibitem{seidel} E. Seidel,
1103: Class. Quantum Grav. {\bf 6}, 1057 (1989).
1104:
1105: \bibitem{unruh} W. G. Unruh,
1106: Phys. Rev. D {\bf 14}, 3251 (1976).
1107:
1108: \bibitem{abramowitz} M. Abramowitz and
1109: A. Stegun, {\it Handbook of mathematical functions}, (Dover
1110: Publications, New York, 1970).
1111:
1112: \bibitem{landau} L. Landau and E. Lifshitz,
1113: {\it Quantum mechanics, non-relativistic theory}
1114: (Mir, Moscow, 1974).
1115:
1116: \bibitem{teu2} S.A. Teukolsky,
1117: Astrophys. J. {\bf 185}, 635 (1973).
1118:
1119: \bibitem{hughes} S. A. Hughes,
1120: Phys. Rev. D {\bf 62}, 044029 (2000); Erratum-ibid. D {\bf 67},
1121: 089902 (2003).
1122:
1123: \bibitem{leaver} E. W. Leaver,
1124: Proc. R. Soc. London A{\bf 402}, 285 (1985).
1125:
1126: \bibitem{bekenstein} J. D. Bekenstein and M. Schiffer,
1127: Phys. Rev. D {\bf 58}, 064014 (1998).
1128:
1129: \bibitem{horo} G. T. Horowitz, and V. Hubeny,
1130: Phys. Rev. D {\bf 62}, 024027(2000);
1131: V. Cardoso and J. P. S. Lemos, Phys. Rev. D {\bf 64}, 084017 (2001);
1132: E. Berti, K. D. Kokkotas, Phys. Rev. D {\bf 67}, 064020 (2003).
1133:
1134: \bibitem{cardoso} V. Cardoso, R. Konoplya, J. P. S. Lemos,
1135: Phys. Rev. D {\bf 68}, 044024 (2003).
1136:
1137: \bibitem{cardosol} V. Cardoso, unpublished.
1138:
1139: \bibitem{putten} M. H. P. M. Putten,
1140: Science {\bf 284}, 115 (1999).
1141:
1142: \bibitem{aguirre} A. Aguirre,
1143: Astrophys. J. {\bf 529}, L9 (2000).
1144:
1145: \bibitem{blandford} R. D. Blandford and R. L. Znajek,
1146: Mon. Not. Roy. Astron. Soc. {\bf 179}, 433 (1997).
1147:
1148: \bibitem{piran} T. Piran,
1149: Phys. Rep. {\bf 314}, 575 (1999);
1150: P. M\'esz\'aros,
1151: Ann. Rev. of Astron. and Astrophys. {\bf 40}, 137 (2002).
1152:
1153:
1154: \end{thebibliography}
1155:
1156: \end{document}
1157:
1158:
1159:
1160:
1161:
1162:
1163:
1164: