1: %--------------------------------------------------------------
2: %Quasi-normal mode analysis in BEC acoustic black holes
3: %--------------------------------------------------------------
4: \documentclass[prd,twocolumn,showpacs,eqsecnum,amsfonts,amssymb,
5: amsmath,a4paper,english,superscriptaddress, nofootinbib
6: %,draft
7: ]{revtex4}
8: \usepackage{graphicx,multirow}
9: %\usepackage[innercaption]{sidecap}
10: \addtolength{\oddsidemargin}{-3.5mm} %a4
11: \addtolength{\topmargin}{12mm} %a4
12:
13:
14: \newlength{\inda}
15: \newlength{\indb}
16: \def\nt{N_{_{\!\!\!\!\!\!\sim}}}
17: \def\euc{{\sc e}}
18: \def\ym{{\text{\textsc{ym}}}}
19: \def\eqand{\hspace{3mm}\mbox{and}\hspace{3mm}}
20: \settowidth{\inda}{$^{aa}$} \settowidth{\indb}{$^{aaa}$}
21:
22: \newcommand{\definition}{:=}
23: \newcommand{\sa}[1]{{^{#1}\!}}
24: \newcommand{\YM}[1]{\ensuremath{\mathbb{#1}}}
25: \newcommand{\ls}{\ell_*}
26:
27: \newcommand{\ins}{\text{\rmfamily in}}
28: \newcommand{\out}{\text{\rmfamily out}}
29:
30: %-------------------------------------------------------------------------
31: % Author-specific definitions
32: %-------------------------------------------------------------------------
33: \def\g{\text{\sl g}}
34: \def\half{{1\over2}}
35: \def\L{{\mathcal L}}
36: \def\S{{\mathcal S}}
37: \def\d{{\mathrm{d}}}
38: \def\x{{\mathbf x}}
39: \def\v{{\mathbf v}}
40: \def\k{{\mathbf k}}
41: \def\im{{\rm i}}
42: \def\etal{{\emph{et al\/}}}
43: \def\det{{\mathrm{det}}}
44: \def\tr{{\mathrm{tr}}}
45: \def\ie{{\emph{i.e.}}}
46: \def\bnabla{\mbox{\boldmath$\nabla$}}
47: \def\Box{\kern0.5pt{\lower0.1pt\vbox{\hrule height.5pt width 6.8pt
48: \hbox{\vrule width.5pt height6pt \kern6pt \vrule width.3pt}
49: \hrule height.3pt width 6.8pt} }\kern1.5pt}
50: \def\HRULE{{\bigskip\hrule\bigskip}}
51: %-------------------------------------------------------------------------
52:
53:
54: %--------------------------------------------------------------
55: \begin{document}
56: %--------------------------------------------------------------
57: \setcounter{topnumber}{1}
58:
59: \title{Quasi-normal mode analysis in BEC acoustic black holes}
60: %--------------------------------------------------------------
61: \author{C. Barcel\'{o}}
62: \affiliation{Instituto de Astrof\'{i}sica de Andaluc\'{i}a, CSIC, Camino Bajo de Hu\'{e}tor 50, 18008 Granada, Spain}
63: %--------------------------------------------------------------
64: \author{A. Cano}
65: \affiliation{Instituto de Astrof\'{i}sica de Andaluc\'{i}a, CSIC, Camino Bajo de Hu\'{e}tor 50, 18008 Granada, Spain}
66: %--------------------------------------------------------------
67: \author{L. J. Garay}
68: \affiliation{Departamento de F\'{i}sica Te\'{o}rica II, Universidad Complutense de Madrid, 28040 Madrid, Spain}
69: \affiliation{Instituto de Estructura de la Materia, CSIC, Serrano 121, 28006 Madrid, Spain}
70: %--------------------------------------------------------------
71: \author{G. Jannes}
72: \affiliation{Instituto de Astrof\'{i}sica de Andaluc\'{i}a, CSIC, Camino Bajo de Hu\'{e}tor 50, 18008 Granada, Spain}
73: \affiliation{Instituto de Estructura de la Materia, CSIC, Serrano 121, 28006 Madrid, Spain}
74:
75: \date{\today}
76:
77: %--------------------------------------------------------------
78: \begin{abstract}
79: %--------------------------------------------------------------
80:
81: We perform a quasi-normal mode analysis of black hole configurations
82: in Bose-Einstein condensates (BEC). In this analysis we use the full
83: Bogoliubov dispersion relation, not just the hydrodynamic or
84: geometric approximation. We restrict our attention to one-dimensional
85: flows in BEC with step-like discontinuities. For this case we show
86: that in the hydrodynamic approximation quasi-normal
87: modes do not exist. The full dispersion relation, however, allows the existence of quasi-normal modes. Remarkably, the spectrum of these modes is not discrete but continuous.
88: %When the modifications of the acoustic dispersion relation are taken into account, we find not only that quasi-normal modes exist, but that they appear in continuous bands.
89:
90:
91:
92:
93:
94: %--------------------------------------------------------------
95: \end{abstract}
96: %--------------------------------------------------------------
97: \pacs{04.70.Dy, 03.75.Kk, 04.80.-y} \maketitle
98:
99: %--------------------------------------------------------------
100: \section{Introduction}
101: \label{S:introduction}
102: %--------------------------------------------------------------
103: Analogue models of gravity \cite{review,analogue-book} allow the
104: study of certain geometrical aspects of linearized general relativity
105: (GR). When the analogue models are based on an intrinsically quantum
106: system, such as Bose-Einstein condensates (BECs) \cite{sonicBH1,
107: sonicBH2, barcelo-bec}, this opens up the possibility of studying some
108: corrections to this geometric description. These corrections ---
109: encoded in the form of non-relativistic or modified dispersion relations, similar to what
110: is expected in a wide range of scenarios for gravity at Planck-scale
111: energies, leading to possible violations of Lorentz invariance \cite{lorentz_violations} --- are then of a quantum origin, but
112: they express themselves in a classical manner. In general, these
113: corrections have to be taken into account in any treatment involving high energies (high
114: frequencies). In BECs in particular, in the presence of horizons, the deviations from the geometric or hydrodynamic regime are important
115: %a relevant correction is inherent
116: even at low energies \cite{instabilities}.
117: %This leads to a modification of the dispersion relation, similar to what is expected in a wide range of scenarios for gravity at Planck-scale energies\cite{lorentz_violations}.
118:
119: In a previous article \cite{instabilities}, the stability of several
120: one-dimensional configurations in BECs was studied. These
121: configurations were chosen in analogy with the geometries
122: associated with gravitational black holes.
123: %gravitational situations such us black holes.
124: Here we will elaborate further on black hole
125: configurations, which were shown to be devoid of instabilities,
126: %to be stable,
127: and examine their quasi-normal or relaxation modes. The analysis we
128: carry out is analogous to the standard quasi-normal mode (QNM)
129: analysis in gravity
130: \cite{nollert, kokkotas}. According to GR, when black holes are
131: perturbed, they emit gravitational waves in the following way. The
132: initial regime depends mainly on the concrete form of the
133: perturbation. Then, an oscillatory phase is attained in which the
134: precise form of the emitted wave depends only on the properties of the
135: black hole itself. Finally, a polynomial tail characterizes the return
136: to equilibrium. In the intermediate phase, a discrete set of complex
137: frequencies are excited: the quasi-normal modes of our concern. Since
138: quasi-normal modes are modes of decay, i.e. of energy dissipation,
139: they are outgoing. In GR this means that their group velocities are
140: directed outwards both in the asymptotic region and at the
141: horizon. Moreover, as we will illustrate, in GR this also
142: automatically implies that they are non-normalizable or divergent.
143:
144: A general investigation of quasi-normal modes in analogue black holes
145: was carried out in \cite{berti}. As we will briefly point out, in standard GR in $(1+1)$ dimensions, quasinormal modes do not exist. In higher dimensions, a discrete spectrum appears \cite{nollert, kokkotas}. Similar results were found in \cite{berti} for analogue black holes. Here the essential element that we add is the influence of a
146: non-relativistic contribution to the dispersion relation on these
147: QNMs\footnote{Modifications of the quasi-normal mode
148: spectrum of black holes due to Lorent violations have also
149: been investigated in \cite{chen} but from a different perspective. They
150: introduce spontaneous Lorentz-symmetry breaking
151: terms in an effective Lagrangian via non-zero expectation
152: values of different tensor fields in the vacuum, thereby restricting
153: attention to a modified Dirac equation in a Schwarzschild background.
154: However, these terms do not lead to modified dispersion relations
155: of the type analyzed here.}. With the full Bogoliubov dispersion relation in BEC, not only does a
156: QNM-spectrum appear even in the one-dimensional case studied here, but
157: this spectrum turns out to be continuous. This result suggests the straightforward speculation that in $(3+1)$ dimensions, for both analogue and GR black holes, there will also be continuous regions of QNMs when modifications of the dispersion relation with respect to the hydrodynamic or relativistic case are taken into account.
158:
159: %Here the essential element that we add is the influence of the modification of the dispersion relation on these QNMs. As we will briefly illustrate, with acoustic (unmodified) dispersion relations, there are no QNMs in one-dimensional BEC configurations. In higher dimensions, a discrete spectrum appears \cite{berti}. These results are analogous to what is found in GR \cite{nollert, kokkotas}. However, with the full modified dispersion relations, not only does a QNM-spectrum appear even in the one-dimensional case studied here, but this spectrum turns out to be continuous.
160:
161: The structure of this paper is as follows. In the next two sections we
162: will review the steps that lead from the general description of a BEC
163: to the simplified formalism for one-dimensional configurations with
164: abrupt discontinuities, and highlight the modified dispersion
165: relation. In Sec.~\ref{S:boundary_conditions} we will discuss the
166: boundary conditions for our problem, and compare with the standard
167: quasi-normal mode analysis of gravitational black holes. Then, in
168: Sec.~\ref{S:discussion} we present and discuss our main results.
169: % and discuss these.
170: %
171: %--------------------------------------------------------------
172: \section{Preliminaries}
173: \label{S:preliminaries}
174: %--------------------------------------------------------------
175: In this section we will briefly review the description of background
176: BEC configurations and their perturbations, with particular emphasis
177: on the modified dispersion relation. The main aim here is to fix
178: notation. General reviews of BECs can be found e.g. in
179: \cite{dalfovo,castin} while the type of configurations discussed here
180: were amply described in \cite{instabilities}.
181:
182: A dilute gas of interacting bosons can be described in terms of quantum field operators $\widehat \Psi({\mathbf x})$ [and $\widehat \Psi^\dag({\mathbf x})$] that annihilate [and create] particles. %Consider a quantum field operator $\widehat \Psi$ {\tt that annihilates particles in a} %associated to a dilute gas of interacting bosons.
183: When this gas condensates, $\widehat \Psi$ can be separated into a
184: macroscopic wave function $\psi$ and a field operator $\widehat
185: \varphi$ describing quantum fluctuations:
186: \mbox{${\widehat \Psi}=\psi+{\widehat \varphi}$}.
187: %
188: The macroscopic wave
189: function $\psi$ satisfies the Gross-Pitaevskii (GP) equation
190: %
191: \begin{align}
192: \im \hbar \; \frac{\partial }{\partial t} \psi(t,\x)= \left(
193: - \frac{\hbar^2}{2m} \nabla^2
194: + V_{\rm ext}(\x)
195: + \g \; |\psi|^2 \right) \psi(t,\x),
196: \label{GP}
197: \end{align}
198: %
199: where $m$ is the boson mass, $V_\text{ext}$ the external potential
200: and $\text{\sl g}$ a coupling constant which is related to the
201: corresponding scattering length $a$ through $\g ={4\pi \hbar^2 a
202: /m}$.
203:
204: This equation can be expressed in terms of hydrodynamic quantities
205: such as the local speed of sound $c$ and the velocity of the fluid
206: flow $\mathbf v$. This proceeds by first introducing the Madelung
207: representation for the order parameter in the GP equation:
208: \begin{eqnarray}
209: \psi = \sqrt{n}e^{\im \theta/\hbar} e^{- \im \mu t/\hbar }.
210: \label{madelung}
211: \end{eqnarray}
212: %
213: Here $n$ is the condensate density, $\mu$ the chemical
214: potential and $\theta$ a phase factor.
215: %Next, assume that $n$ and $\theta$ can be linearized:
216: Next, as we are interested in the acoustic perturbations of the
217: BEC, we linearize in $n$ and $\theta$:
218: %
219: \begin{subequations}
220: \begin{align}
221: n (\x, t) &= n_0(\x) + \g^{-1} \widetilde n_1(\x, t),
222: \\
223: \theta(\x, t) &= \theta_0(\x) + \theta_1(\x, t).
224: \end{align}
225: \end{subequations}
226: %
227: Then, the following set of equations for the linear
228: perturbations is obtained from the complex GP equation
229: %
230: \begin{subequations}
231: \label{GP_lin}
232: \begin{align}
233: \partial _t \widetilde n_1 &= - \nabla \cdot
234: \left( \widetilde n_1 \mathbf v + c^2 \nabla \theta_1 \right),
235: \label{GP1_lin}
236: \\
237: \partial _t \theta_1
238: &= -\mathbf v \cdot \nabla \theta_1 - \widetilde n_1 +\frac{1}{4} \xi^2
239: \nabla \cdot \left[c^2 \nabla \left( \frac{\widetilde n_1}{c^2}\right)\right],
240: \label{GP2_lin}
241: \end{align}
242: \end{subequations}
243: %
244: which are equivalent to the Bogoliubov equations, and where $\xi \equiv \frac{\hbar}{mc}$ is the healing length, and $c$ and $\mathbf v$ are
245: %
246: \begin{equation}
247: c^2 \equiv \g n_0/m, \qquad \mathbf v \equiv \nabla \theta_0 /m.
248: \end{equation}
249: The last term on the right-hand side in \eqref{GP2_lin} is the
250: so-called ``quantum potential'' term.
251:
252: At this point, let us briefly recall that in the hydrodynamic limit, which is obtained by neglecting the quantum potential term just mentioned, an effective Lorentzian metric can be constructed of
253: the type:
254: %
255: \begin{align}
256: (g_{\mu \nu}) \propto
257: \begin{pmatrix}
258: v^2-c^2 && -\mathbf v^\text{T}\\
259: -\mathbf v && \mathbf \openone
260: \end{pmatrix},
261: \label{metric}
262: \end{align}
263: %
264: which historically formed the onset of the interest in hydrodynamic analogue models for gravity (for an overview, see \cite{review} and references therein).
265:
266: From now on, consider one-dimensional profiles. The equations \eqref{GP_lin} can be solved in a homogeneous region ($c$ and $v$ constant) by writing the perturbations in plane-wave form
267: %
268: \begin{subequations}\label{plane-waves}\begin{align}
269: \widetilde n_1(x,t)&=
270: A e^{i(k x - \omega t)}, \\
271: \theta_1(x,t)&= B e^{i(k x - \omega t)}.
272: \end{align}\end{subequations}
273: %
274: This immediately leads to the local dispersion relation\footnote{The term `dispersion relation' is sometimes reserved strictly for an equation of the form $\omega=\omega(k)$, while an equation such as \eqref{dispersion_xi} would then have to be called e.g. the `characteristic equation'. For the sake of clarity, we will always use the term `dispersion relation' to refer to an equation of the form \eqref{dispersion_xi} or \eqref{quadr_dispersion}.}
275: %
276: \begin{align}\label{dispersion_xi}
277: (\omega - vk)^2= c^2k^2 + \frac{1}{4} c^2 \xi^2 k^4.
278: \end{align}
279: %
280: This dispersion relation is \textit{quartic} in $k$. It should be compared to the hydrodynamic dispersion relation that is obtained by neglecting the quantum potential:
281: %
282: \begin{align}\label{quadr_dispersion}
283: (\omega - vk)^2= c^2k^2.
284: \end{align}
285: %
286: This hydrodynamic dispersion relation is of the usual \textit{quadratic}
287: form.
288:
289: However, here we are not so much interested in this hydrodynamic limit, but precisely in deviations from it. Therefore we will always work with the full dispersion relation, and only use the hydrodynamic limit as a means of comparison. The key point is that, because the full dispersion relation is quartic, when writing the mode corresponding to a certain frequency $\omega$:
290: %
291: \begin{equation}\label{bogoliubov_modes}
292: u_\omega=e^{-i\omega t}\sum_j A_j e^{ik_jx},
293: \end{equation}
294: %
295: there are now four contributions to this $u_\omega$-mode, stemming from the four values of $k$ associated to each value of $\omega$. In the hydrodynamic limit there are obviously only two contributions to each $u_\omega$-mode:
296: %
297: \begin{equation}
298: k_{1,2} = \omega/(v \pm c).
299: \end{equation}
300: %
301: In the configurations analyzed in this paper, the additional,
302: ``non-hydrodynamic'' modes will turn out to play a crucial role even
303: at low frequencies.
304: %
305: %--------------------------------------------------------------
306: \section{Background configurations and numerical method}
307: \label{S:numerical}
308: %--------------------------------------------------------------
309: We are looking for quasi-normal modes of black hole configurations in
310: one-dimensional BECs of the type represented in fig.~\ref{F:profiles}. These profiles are piecewise homogeneous with a single discontinuity. In order to find these relaxation modes, first, we will need a way to describe the mode solutions that
311: %satisfy
312: characterize the profiles under study. The idealized form of these profiles
313: allows a simplified description in each homogeneous region. Then, the
314: various sections have to be linked through matching conditions at the
315: discontinuity. After finding all the mode solutions, the quasi-normal modes will be those that moreover satisfy the adequate boundary conditions. We will describe the matching conditions in the current section, together with a straightforward algorithm for
316: numerical implementation. The boundary conditions will be the topic of
317: the next section.
318: %
319: %=============================================================
320: \begin{figure}[tbp]
321: \includegraphics[width=.4\textwidth,clip]{figures/profiles}
322: \bigskip
323: \caption{(Color online) Flow and sound velocity profile consisting of two homogeneous regions with a step-like discontinuity at $x=0$ simulating a black hole-like configuration in a BEC. The solid blue line represents the speed of sound $c$, the dashed red line the fluid velocity $v$. In the upper part of the picture, the negative value of $v$ indicates that the fluid is left-moving. For $x>0$, the fluid is subsonic since $c>\lvert v \rvert$. At $x<0$ it has become supersonic. At $x=0$, there is a sonic horizon.}
324: \label{F:profiles}
325: \end{figure}
326: %=============================================================
327:
328: Integrating \eqref{GP_lin} in an infinitesimal neighbourhood of the discontinuity (say $x=0$) and simplifying with the use of the continuity equation ($vc^2 = \text{constant}$) leads, for our choice of a piecewise homogeneous background, to the following set of matching conditions \cite{instabilities}:
329: %
330: \begin{subequations}\label{matching}\begin{align}
331: [\theta_1]=0,\qquad [c^2 \partial_x \theta_1]=0,\label{matching_a}\\
332: \left[ \frac{\widetilde n_1}{c^2}\right]=0,
333: \qquad \left[\partial_x
334: \widetilde n_1\right]=0,
335: \end{align}\end{subequations}
336: %
337: where the brackets indicate, for example, \mbox{$[\theta_1]=\left.\theta_1\right|_{x=0^+} -\left.\theta_1\right|_{x=0^-}$}.
338:
339: In each homogeneous section, we look for a solution for a particular frequency $\omega$ as a sum of $k$-modes of the form
340: %
341: \begin{align}\label{general_solution}
342: \widetilde n_1 =&
343: \displaystyle\sum _j A_j e^{i(k_jx - \omega t)},\\
344: \theta_1 =&
345: \displaystyle\sum _j A_j \frac{\omega - vk_j}{ic^2k_j^2} e^{i(k_jx- \omega t)}.
346: \end{align}
347: %
348: where the sum is over the four $k_j$-values associated to $\omega$ in that homogeneous region.
349:
350: The conditions \eqref{matching} can be written in matrix form
351: \begin{equation}
352: \Lambda_{ij}A_j=0.
353: \end{equation}
354: Since there are eight free parameters $A_j$ for each discontinuity (four in each region),
355: and only four conditions, matching is in principle always
356: possible. However, there are additional conditions: the boundary
357: conditions which we will discuss in the next section. These boundary
358: conditions can be added to the matrix $\Lambda_{ij}$, to obtain a
359: $(4+N)\times 8$ matrix $\widetilde\Lambda_{ij}$, with $N$ the number
360: of constraints on the coefficients $A_j$ that follow from the boundary
361: conditions. Depending on the value of $N$, we can have the following
362: cases.
363: %
364: \begin{itemize}
365: \item If $N<4$, then there will always exist a non-trivial solution ${A_j}$, since there are more degrees of freedom than constraints. Therefore, a region of the complex $\omega$-plane where $N<4$ will represent a continuous region of eigenfrequencies.
366: \item If $N=4$, then we have a $8 \times 8$ system of equations and there is a non-trivial solution ${A_j}$ only if $\det(\widetilde\Lambda)=0$. Generally speaking, this means that in a region where $N=4$, we might at the very most expect isolated discrete eigenfrequencies, but no continuous zones of solutions.
367: \item If $N>4$ then the system can be split into two (or more) $8\times 8$ subsystems, each with a subdeterminant $\lambda_i$. For a non-trivial solution to exist, each of these subsystems should fulfill the previous condition $\det(\lambda_i)=0$.
368: \end{itemize}
369: %
370: To summarize, we can define a non-negative function $F(\omega)$ such that
371: %
372: \begin{itemize}
373: \item $F(\omega)=0$ if $N<4$,
374: \item $F(\omega)=|\det(\widetilde\Lambda)|$ if $N=4$,
375: \item $F(\omega)=\sum|\det(\lambda_i)|$ if $N>4$.
376: \end{itemize}
377: The eigenfrequencies of the system are those values of $\omega$ for which $F(\omega)=0$, and since the quasi-normal modes are relaxation modes they will be located in the lower half complex $\omega$ plane.
378: %
379: %--------------------------------------------------------------
380: \section{Quasi-normal mode analysis and boundary conditions}
381: \label{S:boundary_conditions}
382: %--------------------------------------------------------------
383: The essential ingredient that is still missing to calculate the quasi-normal modes in these BEC black hole configurations are the boundary conditions. We will first take a look at the boundary conditions that are used to calculate QNMs in GR. This can be immediately extrapolated to BECs in the hydrodynamic limit. However, as we will show, in this hydrodynamic limit in $(1+1)$ dimensions, there exist no QNMs. Finally, we will come to the boundary conditions with the full dispersion relation.
384: %
385: %- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
386: \subsection{Quasinormal modes of a gravitational black hole}
387: \label{SS:QNM_gravity}
388: %- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
389: Quasi-normal modes are simply the modes of energy dissipation of a perturbed object.
390: Gravitational black hole QNMs in particular are the relaxation modes that characterize the pulsations of the black hole after perturbations initiated internally in the spacetime surrounding the black hole.
391:
392: The fact that these modes decay in time means that they can be represented by complex frequencies $\omega$ with Im$(\omega) < 0$. The fact that they are relaxation modes means that they will have to be outgoing in the asymptotic region. At the horizon, QNMs are also required to be outgoing because in GR nothing can escape through the horizon (note that in QNM terminology ``outgoing'' means directed towards the exterior of the spacetime region connected to an asymptotic observer, in other words at the horizon this means directed towards the singularity). So the boundary conditions that are imposed in GR to find the QNMs of a black hole are simply that these modes should be outgoing both in the asymptotic region and at the horizon.
393:
394: Since these QNMs are outgoing, they are also divergent. Indeed, because of the quadratic dispersion relation, say $\omega^2=c^2k^2$, we have for the group velocity $v_g$ and for the imaginary part of $k$:
395: %
396: \[
397: v_g=\pm c, \qquad \text{Im}(k)=\text{Im}(\omega) /v_g%\pm c
398: \]
399: %
400: and therefore, since QNMs have Im$(\omega)<0$:
401: %
402: \[
403: \text{sign}[\text{Im}(k)]=-\text{sign}[v_g].
404: \]
405: %
406: To sum up, QNMs in a gravitational black hole are found by imposing the boundary condition that they should be outgoing, and because of the quadratic dispersion relation they are also automatically divergent.
407:
408: %- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
409: \subsection{BEC hydrodynamic limit: absence of quasi-normal modes in one-dimensional flows}
410: \label{SS:hydrodynamic_QNM}
411: %- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
412:
413: In BECs in the hydrodynamic limit, decaying outgoing modes are also automatically divergent, again because of the quadratic dispersion relation. Indeed,
414: \[
415: v_g=v \pm c,\qquad
416: \text{Im}(k)=\text{Im}(\omega)/v_g%(v \pm c),
417: \]
418: %
419: and therefore $\text{sign}[\text{Im}(k)]=-\text{sign}[v_g]$ as before. Let us see whether QNMs can exist in this case.
420:
421: In the hydrodynamic limit, the quantum potential term in \eqref{GP_lin} is neglected, and so upon integration of the remaining terms in these equations, only two matching conditions remain:
422: %
423: \begin{equation}
424: [\theta_1]=0, \qquad [v \widetilde n_1 + c^2\partial_x \theta_1]=0.
425: \end{equation}
426: %
427: Note that this second equation cannot be simplified as in \eqref{matching}, because this requires the use of a third matching condition which came precisely from integration of the quantum potential.
428:
429: The hydrodynamic solution for a particular \mbox{frequency $\omega$},
430: \begin{align}
431: \widetilde n_1 &= A_1 e^{i(k_1x - \omega t)} + A_2 e^{i(k_2x - \omega t)},\\
432: \theta_1 &= B_1 e^{i(k_1x- \omega t)} + B_2 e^{i(k_2x- \omega t)},
433: \end{align}
434: %
435: has to satisfy $A_j = i(\omega-vk_j)B_j$. Assume that the configuration is completely subsonic. Then, for $x<0$, \mbox{$k_1=\omega/(v+c)\equiv k_\text{in}^-$} is \textit{in}going, whereas \mbox{$k_2=\omega/(v-c)\equiv k_\text{out}^-$} is \textit{out}going, and vice versa for $x>0$, where we write $k_\text{in}^+$ and $k_\text{out}^+$ respectively. In terms of the associated
436: coefficients $B_j$, the matching conditions then become
437: %
438: \begin{align}
439: B_\text{out}^+ + B_\text{in}^+ &= B_\text{in}^- + B_\text{out}^-,\\
440: c^+(B_\text{out}^+ - B_\text{in}^+) &= c^-(B_\text{in}^- - B_\text{out}^-),
441: \end{align}
442: %
443: or in matrix form
444: %
445: \begin{align}
446: \begin{pmatrix}
447: B_\text{in}^+\\B_\text{out}^+
448: \end{pmatrix}
449: =
450: \frac{1}{2c^+}
451: \begin{pmatrix}
452: c^+ - c^- && c^+ + c^-\\
453: c^+ + c^- && c^+ - c^-
454: \end{pmatrix}
455: %
456: \begin{pmatrix}
457: B_\text{in}^-\\B_\text{out}^-
458: \end{pmatrix}.
459: \label{matrix}
460: \end{align}
461: %
462: For a solution to be outgoing in both regions, %$B_\text{in}^+$ should vanish when $B_\text{in}^-$ does.
463: both $B_\text{in}^+$ and $B_\text{in}^-$ should vanish. This means that the coefficient $c^+ + c^-$ connecting $B_\text{in}^+$ with $B_\text{out}^-$ must be zero. But since $c$ is always a positive integer, this is impossible. Therefore, there are no quasi-normal solutions in the hydrodynamic limit.
464:
465: %The above calculation is strictly valid only for completely subsonic
466: %profile. When the left-hand side is supersonic, the hydrodynamic
467: %approximation breaks down due to the presence of the sonic horizon
468: %\cite{instabilities}. Moreover, the identification of in- and outgoing
469: %modes in the supersonic region becomes
470: %observer-dependent. Nevertheless, we know that for such a
471: %configuration the outgoing hydrodynamic mode at the right-hand side
472: %cannot connect to any hydrodynamic mode at the left-hand side,
473: %precisely because of the presence of the horizon. So therecannot be
474: %any QNMs, regardless of the sub- or supersonic character of the
475: %profile.
476:
477: The above calculation is strictly valid only for completely
478: subsonic profiles. When the left-hand side is supersonic, the
479: hydrodynamic approximation breaks down in the passage through the
480: sonic horizon \cite{instabilities}. Both modes in the supersonic region become outgoing at $x\rightarrow -\infty$, while there is still one outgoing mode in the right asymptotic (subsonic) region. Since the matrix in \eqref{matrix} is well defined, it seems then that some QNMs exist for these configurations. However, this is a spurious effect caused by the subsonic to supersonic step-like discontinuity in combination with the hydrodynamic regime. Precisely because of the presence of the horizon, we know that none of the modes at the left-hand side can be connected to the outgoing mode at the right-hand side. So QNMs cannot exist. From a point of view more similar to GR, when the $(x<0)$ region is supersonic, it becomes disconnected from the subsonic region $(x>0)$. But then, just like in GR, the requirement that QNMs must be outgoing in the left asymptotic region should be replaced by the requirement of being outgoing at the horizon. Because of the homogeneity of the profile, none of the modes can be outgoing both at the right asymptotic region and at the horizon. So in any case, QNMs cannot exist in the hydrodynamic limit in $(1+1)$ dimensions, regardless of the sub- or supersonic character of the profile.
481:
482: Note that in GR in $(1+1)$ dimensions, a more general argument prohibits
483: the existence of QNMs. Indeed, all two-dimensional metrics are
484: conformally flat. In addition, the d'Alembertian equation in two
485: dimensions is conformally invariant \cite{birrell-davies}. Therefore,
486: all the solutions of the system are conformally equivalent to plane waves, and so they don't satisfy the requirement of being outgoing both in the asymptotic region and at the horizon. Hence QNMs cannot exist in $(1+1)$ GR. The situation is completely different in $(3+1)$ dimensions, in which GR black holes exhibit a discrete QNM spectrum.
487:
488:
489: %Note that in GR, a more general argument prohibits the existence of
490: %QNMs in any type of $(1+1)$-compactified situation. Indeed, all
491: %two-dimensional metrics are conformally flat. Moreover... (Birrell \&
492: %Davies) Finally, recall that in $(3+1)$, GR black holes have a
493: %discrete QNM spectrum.
494: %
495: %- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
496: \subsection{Boundary conditions with the full dispersion relation}
497: \label{SS:boundary_conditions}
498: %- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
499: As we just discussed, in BECs in the hydrodynamic limit, decaying
500: outgoing modes are automatically divergent. But when looking at the
501: full dispersion law, this relation is no longer valid. Note that for
502: our choice of profiles, we have two asymptotic regions, and moreover
503: no prohibition of crossing the horizon in any direction, since
504: %we explicitly do not want to prohibit crossing of the horizon in
505: %any direction, because
506: the permeability of the horizon is an essential
507: feature of modified dispersion relations \cite{instabilities}. An
508: outgoing mode is then one that has a group velocity $v_g<0$ for $x<0$ and $v_g>0$ for $x>0$, where $v_g$ is obtained from the full dispersion
509: relation:
510: %
511: \begin{equation}\label{group-velocity}
512: v_g(\omega, k) \equiv \text{Re} \left( \frac{d \omega}{dk}\right) =
513: \text{Re}\left(
514: \frac{c^2 k + {\frac{1}{2}} \xi^2 c^2 k^3}{\omega - vk} + v
515: \right).
516: \end{equation}
517: %
518: Clearly, there is no immediate connection with the sign of Im$(k)$ anymore, so the automatic divergent character of outgoing modes is lost.
519:
520: In any case, quasi-normal modes should be outgoing, and this remains the essential boundary condition. Additionally, one might wonder whether the QNMs will be convergent, or divergent after all (in spite of the modification of the dispersion relation). An easy way to check this is by applying convergence as an additional boundary condition and comparing the resulting spectrum with the one obtained without this additional condition.
521:
522: To sum up, quasi-normal modes should be outgoing. In GR and in the
523: hydrodynamic approximation, this automatically implies a
524: divergent behaviour in the asymptotic regions.
525: %divergence.
526: However, in both these cases, there are no QNMs in $(1+1)$ dimensions
527: (as opposed to $(3+1)$ GR black holes with their discrete QNM
528: spectrum). In BECs with the full modified dispersion relation,
529: outgoing modes are no longer automatically divergent. The boundary
530: condition to determine QNMs in the $(1+1)$ BEC black hole
531: configurations under study is therefore that they should be
532: outgoing. Nonetheless, we will check whether they are convergent or
533: divergent.
534: %
535: %--------------------------------------------------------------
536: \section{Results and discussion}
537: \label{S:discussion}
538: %--------------------------------------------------------------
539: In this section we present and discuss results for the quasi-normal mode spectrum of black holes in one-dimensional BECs obtained with the full dispersion relations.
540:
541: Fig.\ref{F:QNM_outgoing} illustrates the QNM spectrum for an acoustic black hole containing a single discontinuity, subsonic at the right-hand side and supersonic at the left. Remember from \cite{instabilities} that such a black hole did not possess unstable eigenfrequencies, so it indeed makes sense to look for stable modes.
542:
543: As discussed in the previous section, the boundary condition that determines the existence of QNMs is that they should be outgoing in both asymptotic regions. Surprisingly, as can be seen from Fig.\ref{F:QNM_outgoing}, not only does such a QNM spectrum indeed show up, but this spectrum is actually continuous.
544: %
545: %=============================================================
546: \begin{figure}[tbp]
547: \includegraphics[width=0.4\textwidth,clip]{figures/QNM_outgoing}
548: \bigskip
549: \caption{Plots illustrating the quasi-normal mode spectrum for a one-dimensional black hole configuration in BEC. The left-hand part of the picture represents the number $N$ of constraints following from the boundary condition that QNMs should be outgoing. The right-hand part shows the function log$[F(\omega)]$ defined in Sec.~\ref{S:numerical}, where $F(\omega)=0$ corresponds to quasi-normal modes. The QNM spectrum consists of the continuous region where $N<4$. [The numerical values used for these plots, in units such that the healing length $\xi=1$, are $c=1; v=0.7$ in the subsonic region and $v=1.8$ in the supersonic region.]}
550: \label{F:QNM_outgoing}
551: \end{figure}
552: %=============================================================
553:
554: %
555: %\includegraphics[width=0.4\textwidth,clip]{figures/QNM_outgoing}
556: %
557: %\newline
558: The next question is whether these modes are convergent or not. In Fig.\ref{F:QNM_convergence}, we have imposed convergence, respectively at both sides, in the right asymptotic region only, and in the left asymptotic region (the singularity) only.
559: %\newline
560: %=============================================================
561: \begin{figure}[tbp]
562: \includegraphics[width=0.4\textwidth,clip]{figures/QNM_convergence_both}
563: \includegraphics[width=0.4\textwidth,clip]{figures/QNM_convergence_right}
564: \includegraphics[width=0.4\textwidth,clip]{figures/QNM_convergence_left}
565: \bigskip
566: \caption{Alteration of QNM spectrum resulting from imposing the additional boundary conditions of (from top to bottom) convergence in both asymptotic regions, in the right asymptotic region only, and in the left asymptotic region only. Since $F(\omega)$ never reaches zero, it is as if the QNM spectrum of Fig.~\ref{F:QNM_outgoing} had disappeared. In other words, the QNMs are divergent in both asymptotic regions. [Same numerical values for $c$ and $v$ as in Fig.~\ref{F:QNM_outgoing}.]}
567: \label{F:QNM_convergence}
568: \end{figure}
569: %=============================================================
570: %
571: %\includegraphics[width=0.4\textwidth,clip]{figures/QNM_convergence_both.png}
572: %\includegraphics[width=0.4\textwidth,clip]{figures/QNM_convergence_right.png}
573: %\includegraphics[width=0.4\textwidth,clip]{figures/QNM_convergence_left.png}
574: %
575: The QNM spectrum disappears as soon as the convergence condition is imposed at either side. In other words, the \textit{quasi-normal modes are divergent}. Remember from the previous section that for outgoing hydrodynamic modes, this divergent character is automatic. However, we have also seen that when only retaining the hydrodynamic modes, there are no QNMs. So although the QNMs that we have just found are divergent, the inclusion of the non-hydrodynamic modes is essential for their existence, and so their divergent character is not ``automatic'', i.e. it is not due to a strict relation between the signs of $v_g$ and Im$(k)$. As a matter of fact, we have checked that the QNMs disappear when the convergent $k$-contributions are excluded.
576:
577: We have also calculated the QNM spectrum for other black-hole like configurations that were shown to be devoid of instabilities in \cite{instabilities}, and we have checked that a similar continuous spectrum of QNMs appeared in all of them. This indicates that the continuous character of the QNM spectrum is an essential consequence of the modified dispersion relation, and not just of the particular characteristics of a precise configuration.
578:
579: To sum up, there is a continuous region of \mbox{frequencies $\omega$} such that they have associated modes
580: \[
581: u_\omega=e^{-i\omega t}\sum_{j=1}^4 A_j e^{ik_jx},
582: \]
583: which are outgoing at both ends. These modes are divergent, just like in the hydrodynamic case, although the fact that here non-hydrodynamic modes come into play means that this relation cannot be derived directly from the dispersion relation.
584: %
585: %--------------------------------------------------------------
586: \section{Conclusions}
587: %--------------------------------------------------------------
588: We have examined the quasi-normal modes of one-dimensional black hole configurations in BECs, and laid particular emphasis on the importance of the deviation of the full dispersion relation with respect to a relativistic or acoustic one. The full dispersion relation is quartic in BECs, as compared to the usual second order of its hydrodynamic and general relativistic equivalent. With such quadratic dispersion relations, outgoing relaxation modes are automatically divergent. In particular, this argument is valid for QNMs in GR black holes. But the modification of the dispersion relation means that this relation is no longer automatic. Nevertheless, the quasi-normal modes in the systems discussed here are also divergent.
589:
590: More importantly, the QNM spectrum that was obtained in these systems with a modified dispersion relation consists of a continuous region in the complex frequency plane. The importance of this result is easy to see when taking into account that both in GR and in BECs in the hydrodynamic limit, QNMs simply do not exist in $(1+1)$ dimensions, while in higher dimensions the quasi-normal modes of a GR black hole form a discrete spectrum. So due to the modification of the dispersion relation, the QNM spectrum in one-dimensional BEC black holes changes from non-existent to a continuous region of frequencies.
591:
592: We would like to end this paper with the straightforward
593: speculation that the discrete spectrum which is obtained in
594: the usual quasi-normal mode analysis in standard $(3+1)$ GR, will
595: also develop continuous regions when taking the modification of the
596: dispersion relation at high energies into account. The existence of these
597: continuous bands could become a signal of trans-Planckian physics encoded in the
598: emitted spectrum of gravitational waves.
599: %A straightforward speculation is that the discrete spectrum which is obtained in the standard quasi-normal mode analysis in three-dimensionsal GR will also obtain continuous regions when taking the modification of the dispersion relation at high energies into account.
600: \acknowledgements
601: C.B. has been funded by the Spanish MEC under project
602: FIS2005-05736-C03-01 with a partial FEDER contribution. G.J. was
603: supported by CSIC grant I3P-BPD2005 of the I3P
604: programme, cofinanced by the European Social Fund,
605: and by the Spanish MEC under project FIS2005-05736-C03-01.
606: L.G. was supported by the Spanish MEC under project FIS2005-05736-C03-02.
607: The authors also acknowledge support from the Spanish MEC project FIS2006-26387-E.
608:
609:
610: %============================================================
611: \begin{thebibliography}{99}
612: %============================================================
613:
614: %-------------------------------------------------------------
615: \bibitem{review}
616: C.~Barcel\'o, S.~Liberati, M.~Visser,
617: %``Analogue Gravity,''
618: Living Rev. Relativity {\bf 8}, 12 (2005), http://www.livingreviews.org/lrr-2005-12.
619: %-------------------------------------------------------------
620: \bibitem{analogue-book}
621: M. Novello, M. Visser and G. Volovik (eds.),
622: {\it Artificial black holes},
623: World Scientific, Singapore (2002).
624: %-------------------------------------------------------------
625: \bibitem{barcelo-bec}
626: %\cite{Barcelo:2000tg}
627: %\bibitem{Barcelo:2000tg}
628: C.~Barcel\'o, S.~Liberati, M.~Visser,
629: % ``Analog gravity from Bose-Einstein condensates,''
630: Class.\ Quant.\ Grav.\ {\bf 18}, 1137 (2001).
631: % [arXiv:gr-qc/0011026].
632: %%CITATION = GR-QC 0011026;%%
633: %%Cited 40 times in SPIRES-HEP
634: %-------------------------------------------------------------
635: \bibitem{sonicBH1}
636: %\cite{Garay:1999sk}
637: %\bibitem{Garay:1999sk}
638: L.~J.~Garay, J.~R.~Anglin, J.~I.~Cirac, P.~Zoller,
639: %``Black holes in Bose-Einstein condensates,''
640: Phys.\ Rev.\ Lett.\ {\bf 85}, 4643 (2000).
641: % [arXiv:gr-qc/0002015].
642: %%CITATION = GR-QC 0002015;%%
643: %%Cited 62 times in SPIRES-HEP
644: %-------------------------------------------------------------
645: \bibitem{sonicBH2}
646: %\cite{Garay:2000jj}
647: %\bibitem{Garay:2000jj}
648: L.~J.~Garay, J.~R.~Anglin, J.~I.~Cirac, P.~Zoller,
649: %``Sonic black holes in dilute Bose-Einstein condensates,''
650: Phys.\ Rev.\ A {\bf 63}, 023611 (2001).
651: % [arXiv:gr-qc/0005131].
652: %%CITATION = GR-QC 0005131;%%
653: %%Cited 43 times in SPIRES-HEP
654: %-------------------------------------------------------------
655: \bibitem{lorentz_violations}
656: D.~Mattingly,
657: %"Modern Tests of Lorentz Invariance",
658: Living Rev. Relativity {\bf 8}, 5 (2005), http://www.livingreviews.org/lrr-2005-5
659: %-------------------------------------------------------------
660: \bibitem{instabilities}
661: C.~Barcel\'o, A.~Cano, L.~J.~Garay, G.~Jannes,
662: %``Stability analysis of sonic horizons in Bose-Einstein condensates,''
663: Phys.\ Rev.\ D {\bf 74}, 024008 (2006).
664: %-------------------------------------------------------------
665: \bibitem{kokkotas}
666: %\cite{Kokkotas:1999bd}
667: %\bibitem{Kokkotas:1999bd}
668: K.~D.~Kokkotas, B.~G.~Schmidt,
669: % ``Quasi-normal modes of stars and black holes,''
670: Living Rev.\ Relativity {\bf 2}, 2 (1999), http://www.livingreviews.org/lrr-1999-2
671: % [arXiv:gr-qc/9909058].
672: %%CITATION = GR-QC 9909058;%%
673: %%Cited 116 times in SPIRES-HEP
674: %-------------------------------------------------------------
675: \bibitem{nollert}
676: H.~P.~Nollert,
677: %``Quasinormal modes: the characteristic 'sound' of black holes and neutron stars,''
678: Class. Quantum Grav. {\bf 16}, R159-R216 (1999).
679: %-------------------------------------------------------------
680: \bibitem{berti}
681: E.~Berti, V.~Cardoso, J.~P.~S.~Lemos,
682: %``Quasinormal modes and classical wave propagation in analogue black holes,''
683: Phys.\ Rev.\ D {\bf 70}, 124006 (2004).
684: %-------------------------------------------------------------
685: \bibitem{chen}
686: Songbai Chen, Bin Wang, Rukeng Su,
687: %``Influence of Lorentz violation on Dirac quasinormal modes in the Schwarzschild black hole spacetime,''
688: Class. Quantum Grav. {\bf 23}, 7581 (2006).
689: %-------------------------------------------------------------
690: \bibitem{dalfovo}
691: F.~Dalfovo, S.~Giorgini, L.~Pitaevskii, S.~Stringari,
692: % ``Theory of Bose-Einstein condensation in trapped gases´´,
693: Rev.\ Mod.\ Phys.\ {\bf 71}, 463 (1999).
694: % [arXiv:cond-mat/9806038].
695: %-------------------------------------------------------------
696: \bibitem{castin} Y. Castin,
697: ``Bose-Einstein condensates in atomic gases: simple theoretical results,''
698: in: {\it Coherent atomic matter waves},
699: Lecture Notes of Les Houches Summer School,
700: edited by R. Kaiser, C. Westbrook, and F. David,
701: EDP Sciences and Springer-Verlag (2001).
702: %-------------------------------------------------------------
703: %\cite{Birrell:1982ix}
704: %\bibitem{Birrell:1982ix}
705: \bibitem{birrell-davies}
706: N.~D.~Birrell and P.~C.~W.~Davies,
707: ``Quantum Fields In Curved Space,''
708: Cambridge University Press (1982).
709: %\href{http://www.slac.stanford.edu/spires/find/hep/www?irn=998621}{SPIRES entry}
710: %============================================================
711: \end{thebibliography}
712: %============================================================
713:
714: %-------------------------------------------------------------
715: \end{document}
716: %----------------------------------------------