1: \documentstyle[cite,11pt]{article}
2: %\documentstyle[cite,multicol]{article}
3: \topmargin=-10mm
4: \evensidemargin=0mm
5: \oddsidemargin=0mm
6: %\textheight=20cm
7: %\textwidth=16cm
8: \textheight=23cm
9: \textwidth=165mm
10: \newcommand{\be}{\begin{equation}}
11: \newcommand{\ee}{\end{equation}}
12: \newcommand{\beqn}{\begin{eqnarray}}
13: \newcommand{\eeqn}{\end{eqnarray}}
14: \newcommand{\beqnn}{\begin{eqnarray*}}
15: \newcommand{\eeqnn}{\end{eqnarray*}}
16: \newcommand{\hr}{\hat\rho}
17: \def\a{\alpha}
18: \def\g{\gamma}
19: \def\l{\lambda}
20: \def\s{\sigma}
21: \def\vf{\varphi}
22: \def\ep{\epsilon}
23: \def\vep{\varepsilon}
24: %\renewcommand {\baselinestretch}{2}
25:
26: \begin{document}
27: %\draft
28:
29: \title{Nonstationary Casimir effect in cavities with two resonantly
30: coupled modes}
31:
32: \author{A. V. Dodonov and
33: V. V. Dodonov
34: \thanks{ e-mail: vdodonov@df.ufscar.br}
35: \thanks{%
36: on leave from Lebedev Physics Institute and Moscow Institute of Physics and
37: Technology, Russia}
38: %}
39: \\
40: %\address{
41: Departamento de F\'{\i}sica, Universidade Federal de
42: S\~ao Carlos,\\
43: Via Washington Luiz, km 235, 13565-905 S\~ao Carlos, SP, Brasil}
44: \date{}
45: \maketitle
46:
47: %\small
48:
49: \begin{abstract}
50: We study the peculiarities of the nonstationary Casimir effect
51: (creation of photons in cavities with moving boundaries)
52: in the special case of two resonantly coupled modes with frequencies
53: $\omega_0$ and $(3+\Delta)\omega_0$, parametrically excited due to small
54: amplitude oscillations of the ideal cavity wall at the frequency
55: $2\omega_0(1+\delta)$ (with $|\delta|,|\Delta|\ll 1$).
56: The effects of thermally induced oscillations in time dependences of the
57: mean numbers of created photons and the exchange of quantum purities
58: between the modes are discovered. Squeezing and photon
59: distributions in each modes are calculated for initial vacuum and thermal
60: states. A possibility of compensation of detunings is shown.
61:
62: \end{abstract}
63:
64: \vspace{5mm}
65:
66: {\it PACS}: {42.50.Lc; 42.50.Dv; 46.40.Ff; 03.70.+k; 03.65.-w}
67:
68: {\it Key words\/}:
69: Dynamical Casimir effect; Vibrating boundary; Parametric resonance;
70: Coupled modes; Squeezing; Photon distribution
71:
72: \newpage %\twocolumn
73: \setlength{\baselineskip}{18pt}
74: %\renewcommand{\baselinestretch}{0.5}
75:
76: %\narrowtext
77: %\twocolumn
78: %\begin{multicols}{2}
79:
80: %\normalbaselines
81:
82:
83: \section{Introduction}
84:
85: Classical and quantum phenomena in cavities with moving boundaries attracted
86: attention of many researchers for a long time (see review \cite{review}).
87: Especially popular this topic
88: became in the last decade, being known now under the names {\it %
89: nonstationary Casimir effect\/} \cite{DKM89},
90: {\it dynamical Casimir effect\/} \cite{Sch},
91: or {\it mirror (motion) induced radiation\/} \cite{BE,Lamb}.
92: One of several theoretical results obtained in the last years was the
93: prediction of the exponential growth of the energy of the field under the
94: resonance conditions, when the wall performs vibrations at the frequency
95: which is a multiple of the unperturbed field eigenfrequency
96: \cite{Lamb,Law,D95,DK96}.
97:
98: In most papers, the special case of the one-dimensional
99: cavity with an equidistant spectrum of unperturbed field modes
100: was studied.
101: Then an infinite number of modes are excited due to an intermode
102: interaction (resulting from the Doppler effect on the moving boundary),
103: and the total number of created photons depends on time as $t^2$
104: (but the total energy increases exponentially)
105: \cite{review,DK96,D98,DA,AD00}.
106: A more realistic case of a three-dimensional cavity
107: was considered, e.g., in \cite{ViHaJa95,Mund,Ji98a}.
108: In particular, simple
109: analytical solutions describing the resonance creation of photons
110: in cavities with totally nonequidistant spectra,
111: when only one mode is in resonance with the vibrating wall,
112: were given in \cite{review,D95,DK96}.
113: However, the case of a few resonantly interacting modes is also of a
114: great interest. In particular, the interaction between the excited field
115: mode and the detector, approximated either by a harmonic oscillator
116: or by a two-level atom, was considered in
117: \cite{D95,DK96,Law95,JaVi96,TaKo98,Janow98,Fedot00}.
118:
119: It was shown recently \cite{Maz01}, that
120: for rectangular three- and two-dimensional cavities,
121: there exist special configurations,
122: when two (or more) modes can occur in resonance with the moving wall and
123: between themselves.
124: For example, the eigenmode spectrum in a cubical cavity of length $L$
125: is given by the formula $\omega_{klm}=(\pi c/L)\sqrt{k^2 +l^2 +m^2}$.
126: Since $\omega_{511} =3\omega_{111}$, both the modes, $\{111\}$ and $\{511\}$,
127: will be excited if the wall oscillates at the frequency $2\omega_{111}$.
128: In such a case, it appears \cite{Maz01} that the energies of both the modes
129: increase in time exponentially (with some additional oscillations), but
130: the increment of the exponential growth is twice smaller than in the case
131: of a single resonance mode.
132:
133: Our goal is to give analytical solutions describing the case
134: of two resonantly interacting modes in a more or less generic situation,
135: without specializing the form of the cavity.
136: We take into account the
137: possibility of detuning between the frequency of the vibrating wall and
138: the twice fundamental field eigenfrequency, and detuning between the
139: frequency of the second mode and the triple fundamental eigenfrequency
140: (however, we do not consider the effects of damping).
141: Using the method of slowly varying amplitudes, we obtain explicit solutions
142: of equations of motion
143: for the canonical operators describing the fields in coupled modes,
144: which enable us to calculate the mean energy and the invariant uncertainty
145: product for each mode.
146: We analyze the evolution of the photon distribution functions
147: for each mode and show the possibility of strong {\em squeezing\/} in the
148: long-time limit (for other configurations this was done in
149: \cite{D95,DK96,DA}).
150:
151:
152: \section{General solution for two coupled modes}
153:
154: We use the Hamiltonian approach proposed by Law \cite{Law} and
155: developed in \cite{Plun} (for other references see \cite{review}).
156: Consider the scalar massless field $\Phi({\bf r},t)$, satisfying
157: the wave equation $\Phi_{tt}=\nabla^2\Phi$ inside the cavity and the
158: Dirichlet boundary condition $\Phi=0$ on the boundary.
159: We assume that we know the complete orthonormalized set of eigenfunctions
160: (and eigenfrequencies) of the Laplace equation
161: $\nabla^2 f_{\a}({\bf r}) +\omega_{\a}^2 f_{\a}({\bf r})=0$
162: in the case of stationary cavity.
163: Now suppose that a part of the boundary is a plane surface moving
164: according to a {\em prescribed\/} law of motion $L(t)$ (for the most recent
165: study of the case when $L(t)$ is a {\em dynamical variable\/} due to
166: the back reaction of the field see \cite{Cole01}).
167: Expanding the field $\Phi({\bf r},t)$ over ``instantaneous''
168: eigenfunctions $f_{\a}({\bf r};L(t))$,
169: \be
170: \Phi({\bf r},t)=\sum_{\a} q_{\a}(t) f_{\a}({\bf r};L(t)),
171: \label{expansion}
172: \ee
173: we satisfy automatically the boundary conditions. Then the dynamics of the
174: field is described completely by the dynamics of the generalized
175: coordinates $q_{\a}(t)$, which, in turn,
176: can be derived from the
177: {\em time-dependent Hamiltonian\/} \cite{Plun}
178: \be
179: H(t) = \frac12\sum_{\a}\left[p_{\a}^2 +\omega_{\a}^2(L(t)) q_{\a}^2\right]
180: +\frac{\dot{L}(t)}{L(t)}\sum_{\a \neq \beta} p_{\a} m_{\a\beta}q_{\beta}
181: \label{genHam}
182: \ee
183: with antisymmetrical time-independent coefficients
184: \be
185: m_{\a\beta}= - m_{\beta\a} =L \int dV
186: \frac{\partial f_{\a}({\bf r};L)}{\partial L}f_{\beta}({\bf r};L).
187: \label{defmab}
188: \ee
189: For example, in the case of a
190: rectangular three-dimensional cavity, the eigenmodes are the well known
191: products of three sine functions like $\sin\left(\pi k_x x/L_x\right)$,
192: labeled by three
193: natural numbers $k_x,k_y,k_z$. If one surface of the parallelepiped,
194: perpendicular to the $x$-axis, moves in the $x$-direction
195: (so that the $L_x$-dimension of the cavity is a function of time), then
196: \cite{Maz01}
197: \be
198: m_{{\bf k}{\bf j}} = (-1)^{k_x+j_x} \frac{2 k_x j_x}{j_x^2 - k_x^2}
199: \delta_{k_y j_y}\delta_{k_z j_z}.
200: \label{m3D}
201: \ee
202:
203: We are interested in the case when one of the cavity walls performs
204: small oscillations with the frequency $\Omega$ close to the double frequency
205: of some unperturbed mode $\omega_1^{(0)} \equiv 1$ (i.e., we normalize all
206: frequencies by $\omega_1^{(0)}$), so that the time-dependent frequency
207: $\omega_1(t)$ reads
208: \be
209: \omega_1(t) = 1 + 2\ep \cos(2\overline{\omega}t),
210: \quad \overline{\omega}=1 +\delta,
211: \label{Lt}
212: \ee
213: where we assume that $|\delta|\ll 1$ and $|\ep|\ll 1$.
214: Also we suppose that the unperturbed field frequency spectrum includes
215: the frequency $\omega_3^{(0)} =3 +\Delta$ with $|\Delta|\ll 1$,
216: but it does not
217: contain frequencies close to $5\omega_1^{(0)}$. Then we have the case of two
218: resonantly interacting modes, and it is sufficient to consider only the
219: part of the total Hamiltonian (\ref{genHam}) related to these modes:
220: \beqn
221: H_{13} &=&\frac12\left(p_1^2 + p_3^2\right) + \frac12
222: \left[1 + 4\ep \cos(2\overline{\omega}t)\right] x_1^2
223: \nonumber \\ &&
224: +\frac12 \left[9+6\Delta +\tilde{\ep} \cos(2\overline{\omega}t)\right]x_3^2
225: \nonumber \\ &&
226: +3\mu\ep\sin(2\overline{\omega}t)\left(p_1 x_3 -p_3 x_1\right).
227: \label{H2}
228: \eeqn
229: The constant parameter $\mu$
230: is proportional to the coefficient $m_{12}$ in (\ref{genHam}).
231: Writing (\ref{H2}) we have neglected the
232: second order terms with respect to $\ep$ and $\Delta$.
233: Since the time-dependent frequency shift
234: $\omega_1(t)-\omega_1^{(0)} \sim L(t)-L_0$ is chosen
235: to be proportional to $\cos(2\overline{\omega}t)$, the last term in
236: (\ref{H2}), which is proportional to $\dot{L}(t)$, must depend on time
237: as $\sin(2\overline{\omega}t)$.
238: All numerical coefficients
239: in (\ref{H2}) are chosen in order to avoid an appearance of fractions in
240: formulas below.
241: Parameter $\tilde{\ep}$
242: %in the time-dependent frequency squared $\omega_3^2(t)$
243: has the same order of magnitude as $\ep$,
244: but it does not affect the solution in the
245: zeroth order approximation (in which we are interested here),
246: as will be shown below.
247:
248: Hamiltonian (\ref{H2}) results in the following differential
249: equations for the generalized coordinates $x_1$ and $x_3$
250: (we neglect corrections of the second order):
251: \beqn
252: \ddot{x}_1 &=& -\left[1 + 4\ep \cos(2\overline{\omega}t)\right] x_1
253: \nonumber \\ && +
254: 24\mu\ep\left[ \cos(2\overline{\omega}t) x_3
255: + \sin(2\overline{\omega}t) \dot{x}_3\right],
256: \label{ddotx1}
257: \eeqn
258: \beqn
259: \ddot{x}_3 &=&
260: - \left[9+6\Delta +\tilde{\ep} \cos(2\overline{\omega}t)\right] x_3
261: \nonumber \\ && -
262: 24\mu\ep\left[ \cos(2\overline{\omega}t) x_1
263: + \sin(2\overline{\omega}t) \dot{x}_1\right].
264: \label{ddotx2}
265: \eeqn
266: We solve equations (\ref{ddotx1}) and (\ref{ddotx2}), using the method
267: of slowly varying amplitudes \cite{Louis,Land,Bogol})
268: (which was applied earlier in studies
269: \cite{D95,DK96,D98,D98a}).
270: Namely, we look for the solutions in the form
271: \be
272: x_{k}(t)= \xi_{k}^{+}(t) e^{ik\overline{\omega}t}
273: +\xi_{k}^{-}(t) e^{-ik\overline{\omega}t}, \quad k=1,3
274: \label{formx1}
275: \ee
276: where each coefficient $\xi_{k}^{\pm}$ is a {\em slowly varying function
277: of time\/}, whose derivatives are proportional to the small parameters
278: $\ep,\delta,\Delta$, so that we can neglect the second-order derivatives
279: $\ddot\xi_{k}^{\pm}$. Then, equating the coefficients at
280: $\exp[\pm i\overline{\omega}t]$ in equation (\ref{ddotx1}) and
281: coefficients at $\exp[\pm 3i\overline{\omega}t]$ in equation (\ref{ddotx2})
282: (and neglecting the terms proportional to squares and products of small
283: coefficients $\ep,\delta,\Delta$),
284: we obtain a set of equations with {\em constant\/} coefficients
285: for the slowly varying amplitudes. It is convenient to write this set in
286: the matrix form $d{\bf v}/dt={\cal A}{\bf v}$, introducing vector
287: ${\bf v}=\left(\xi_{1}^{+},\xi_{1}^{-},\xi_{3}^{+},\xi_{3}^{-}\right)$
288: and matrix
289: \be
290: {\cal A}=\left\Vert
291: \begin{array}{cccc}
292: -i\delta & i\ep & 12i\mu\ep & 0 \\
293: -i\ep & i\delta & 0 & -12i\mu\ep \\
294: 4i\mu\ep & 0 & i(\Delta - 3\delta) & 0 \\
295: 0 & -4i\mu\ep & 0 & -i(\Delta - 3\delta)
296: \end{array}
297: \right\Vert .
298: \label{matrA}
299: \ee
300: Strictly speaking, function $x_1(t)$ contains also terms proportional
301: to $\exp[\pm 3i\overline{\omega}t]$, as well as
302: function $x_3(t)$ contains terms proportional
303: to $\exp[\pm i\overline{\omega}t]$. However,
304: the amplitudes of these additional terms are
305: proportional to the small parameters $\ep,\delta,\Delta$, so we neglect
306: them, confining ourselves to the principal terms whose amplitudes are
307: of the zeroth order with respect to small parameters. Just for this
308: reason, the parameter $\tilde\ep$ does not appear in matrix ${\cal A}$:
309: while the term $4\ep \cos(2\overline{\omega}t)$ standing at $x_1$ in
310: equation (\ref{ddotx1}) connects the slowly varying amplitudes
311: $\xi_{1}^{+}$ and $\xi_{1}^{-}$, a similar term
312: $\tilde\ep \cos(2\overline{\omega}t)$ standing at $x_3$ in (\ref{ddotx2})
313: does not connect the amplitudes $\xi_{3}^{+}$ and $\xi_{3}^{-}$
314: between themselves,
315: but it gives small corrections to the amplitudes of higher-order
316: terms in $x_3$, oscillating with frequencies $\overline{\omega}$ and
317: $5\overline{\omega}$. A more strict (from the mathematical
318: point of view) method, based on the multiple scale analysis, was
319: exposed in \cite{Maz01}. In the zeroth-order approximation it leads to
320: the same results.
321:
322: Matrix ${\cal A}$ has four eigenvalues, $\pm\lambda_{\pm}$, where
323: \beqn
324: \lambda_{\pm} &=&\frac1{\sqrt2}\sqrt{a \pm \sqrt{c}}
325: \label{lam1}\\ &\equiv&
326: \frac12\left(\sqrt{a+\sqrt{b}} \pm \sqrt{a-\sqrt{b}}\right),
327: \label{lam2}
328: \eeqn
329: \be
330: a= \ep^2(1-\nu) -\delta^2 -(\Delta - 3\delta)^2,
331: \label{def-a}
332: \ee
333: \beqn
334: b&=&\left[2(\delta-\ep)(\Delta-3\delta)+\nu\ep^2\right]
335: \nonumber\\&\times&
336: \left[2(\delta+\ep)(\Delta-3\delta)+\nu\ep^2\right],
337: \label{def-b}
338: \eeqn
339: \beqn
340: c&\equiv& a^2-b = 2\ep^2 \nu\left[(\Delta - 4\delta)^2 -\ep^2\right]
341: \nonumber \\ &&
342: +\left[\ep^2 + (\Delta - 2\delta)(\Delta - 4\delta)\right]^2,
343: \label{def-c}
344: \eeqn
345: \be
346: \nu \equiv 96\mu^2 .
347: \label{def-nu}
348: \ee
349: Comparing our matrix ${\cal A}$ (\ref{matrA}) with a similar matrix
350: found in \cite{Maz01} for the rectangular cavity, one can verify that
351: if modes $\{k_x,m,n\}$ and $\{j_x,m,n\}$ are in resonance, then
352: $\mu=j_x/(12k_x)$ (different phases of elements of matrices in \cite{Maz01}
353: and here are due to different choices of trigonometrical dependences
354: in function $L(t)$: we use $\cos$-function instead of $\sin$-function
355: in \cite{Maz01}). In particular, for the modes $\{111\}$ and $\{511\}$
356: of the cubical cavity we have $\nu=50/3$. Due to this explicit example
357: (and other examples related to the rectangular cavities), we assume
358: hereafter that parameter $\nu$ is large, so that it satisfies
359: the inequality $\nu \gg 1 $.
360:
361: \section{Exact resonance}
362:
363: We start with the simplest case of the exact
364: resonance, $\delta=\Delta=0$. Then formula (\ref{lam2}) yields
365: \be
366: \lambda_{\pm}^{(\nu)}=\frac{\ep}{2}\left(1 \pm \sqrt{1-2\nu}\right).
367: \label{lam0}
368: \ee
369: If $\nu=0$ (no intermode coupling), then we obtain the increment of the
370: exponential growth $\lambda_+^{(0)}=\ep$, in accordance with the solutions
371: of the single-mode problem found in \cite{D95,DK96,D98a}.
372: On the contrary, for $\nu > 1/2$ the increment
373: (real part of $\lambda_{\pm}$) does not depend on the form of the cavity
374: (which is ``hidden'' in the value of $\nu$), being
375: exactly twice smaller than in the single-mode case.
376: In the special case of rectangular cavities this result was obtained
377: in \cite{Maz01}.
378:
379: %Calculating eigenvectors of matrix (\ref{matrA}),
380: After some straightforward calculations we arrive at the following
381: explicit expressions for the time dependences of
382: the generalized coordinates
383: %and the canonically conjugated momenta
384: (as far as we neglect all corrections of the order of $\ep$ in the amplitude
385: coefficients, we may identify the canonical momenta with velocities,
386: $p_k=\dot{x}_k$, $k=1,2$):
387: \beqn
388: &&x_1(t) = x_1(0)\left[C_{1}^{-}\,\cos(\rho\tau)
389: +S_{1}^{-}\,\frac{\sin(\rho\tau)}{\rho}\right]
390: \nonumber\\&&
391: - p_1(0)\left[S_{1}^{-}\,\cos(\rho\tau)
392: +C_{1}^{-}\,\frac{\sin(\rho\tau)}{\rho}\right]
393: \nonumber\\&&
394: +8\mu\frac{\sin(\rho\tau)}{\rho}
395: \left[3S_{1}^{-}\, x_3(0) +C_{1}^{-}\, p_3(0)\right],
396: \label{x1t}
397: \eeqn
398: \beqn
399: &&x_3(t) = x_3(0)\left[C_{3}^{+}\,\cos(\rho\tau)
400: -S_{3}^{+}\,\frac{\sin(\rho\tau)}{\rho}\right]
401: \nonumber\\&&
402: +\frac13 p_3(0)\left[S_{3}^{+}\,\cos(\rho\tau)
403: -C_{3}^{+}\,\frac{\sin(\rho\tau)}{\rho}\right]
404: \nonumber\\&&
405: -8\mu\frac{\sin(\rho\tau)}{\rho}
406: \left[S_{3}^{+}\, x_1(0) -C_{3}^{+}\, p_1(0)\right],
407: \label{x2t}
408: \eeqn
409: where
410: \be
411: \tau \equiv \frac12 \ep t, \quad \rho=\sqrt{2\nu-1},
412: \label{deftaurho}
413: \ee
414: \beqnn
415: &&C_{k}^{\pm}(\tau;t) = \cosh\tau \cos(k\overline{\omega}t)
416: \pm \sinh\tau \sin(k\overline{\omega}t),
417: %\label{defCpm}
418: \\
419: &&S_{k}^{\pm}(\tau;t) = \sinh\tau \cos(k\overline{\omega}t)
420: \pm \cosh\tau \sin(k\overline{\omega}t).
421: %\label{defSpm}
422: \eeqnn
423: Similar expressions for the time dependent canonical momenta can be
424: easily obtained by differentiating equations (\ref{x1t})-(\ref{x2t})
425: with respect to
426: the ``fast time'' $t$, considering the ``slow time'' variable $\tau$ as
427: an independent parameter and neglecting corrections of the order of
428: $\ep,\delta,\Delta$ in the amplitude coefficients. The following relations
429: hold in this approximation:
430: \[
431: \frac{\partial C_{k}^{\pm}}{\partial t}= \pm k S_{k}^{\mp}, \quad
432: \frac{\partial S_{k}^{\pm}}{\partial t}= \pm k C_{k}^{\mp}.
433: \]
434: Symbols $x_k,p_k$ in equations (\ref{x1t})-(\ref{x2t}) can be considered
435: both as classical variables and quantum operators in the Heisenberg
436: picture, due to the linearity of the problem (or due to the quadratic
437: nature of Hamiltonian (\ref{genHam})). Using equations
438: (\ref{x1t})-(\ref{x2t}) and their momentum counterparts,
439: one can calculate mean values of squares and
440: products of canonical variables (operators) at any moment of time,
441: provided such mean values were known at the initial moment $t=0$.
442: We confine ourselves to the simplest case, when initially the field
443: modes were in thermal states with the mean photon numbers
444: $(\theta_1-1)/2$ and $(\theta_3-1)/2$, where $\theta_k=\coth(k\beta/2)$,
445: $\beta$ being inverse absolute temperature in dimensionless units.
446: One can check the relations
447: \[
448: \theta_{31}\equiv \frac{\theta_{3}}{\theta_{1}} =\theta_{13}^{-1}
449: =\frac{\theta_{1}^2 +3}{3\theta_{1}^2 +1}, \quad
450: 1\ge \theta_{31} \ge \frac13.
451: \]
452: The mean energies in each mode,
453: ${\cal E}_k =\frac12\langle p_k^2 + \omega_k^2 x_k^2\rangle$,
454: depend on time as follows,
455: \beqn
456: {\cal E}_1 &=& \frac{\theta_1}{2}\Bigg\{
457: \cosh(2\tau)\left[
458: \frac{\sin^2(\rho\tau)}{\rho^2}
459: \left(1 +2\nu\,\theta_{31}\right)\right.
460: \nonumber\\ && \left.
461: +\cos^2(\rho\tau)\right]
462: + \sinh(2\tau) \frac{\sin(2\rho\tau)}{\rho}\Bigg\},
463: \label{E1t}
464: \eeqn
465: \beqn
466: {\cal E}_3 &=& \frac{3\theta_3}{2}\Bigg\{
467: \cosh(2\tau)\left[
468: \frac{\sin^2(\rho\tau)}{\rho^2}
469: \left(1 +2\nu\,\theta_{13}\right) \right.
470: \nonumber\\ && \left.
471: +\cos^2(\rho\tau)\right]
472: - \sinh(2\tau) \frac{\sin(2\rho\tau)}{\rho}\Bigg\}.
473: \label{E2t}
474: \eeqn
475: For rectangular cavities (and, perhaps, for others),
476: $\nu\gg 1$ in the case of intermode resonance.
477: Then $\rho^2\approx 2\nu \gg 1$, so that equations (\ref{E1t})
478: and (\ref{E2t}) can be simplified:
479: \beqnn
480: {\cal E}_1 &\approx& \frac12 \cosh(2\tau)
481: \left[\theta_1\cos^2(\rho\tau) +\theta_3 \sin^2(\rho\tau)\right],
482: %\label{E1app}
483: %\ee
484: %\be
485: \\
486: {\cal E}_3 &\approx& \frac32 \cosh(2\tau)
487: \left[\theta_3\cos^2(\rho\tau) +\theta_1 \sin^2(\rho\tau)\right].
488: %\label{E2app}
489: \eeqnn
490: For the initial vacuum states ($\theta_1=\theta_3=1$) we have
491: a monotonous growth of energy and number of photons in each mode:
492: ${\cal E}_k\approx (\omega_k^{(0)}/2)\cosh(2\tau)$, $k=1,2$.
493: On the contrary, for high-temperature initial states (with equal
494: temperatures), $\theta_1=3\theta_3 \gg 1$, which results in
495: strong oscillations of mean energies and numbers of photons:
496: \beqnn
497: {\cal E}_1 &\approx& \frac{\theta_1}{2} \cosh(2\tau)
498: \left[ 1 -\frac23 \sin^2(\rho\tau)\right],
499: \\
500: {\cal E}_3 &\approx& \frac{\theta_1}{2} \cosh(2\tau)
501: \left[ 1 +2 \sin^2(\rho\tau)\right].
502: \eeqnn
503: In Fig.~1 we show normalized time dependences of the mean energies
504: (calculated with the aid of complete formulas (\ref{E1t}) and (\ref{E2t}))
505: in each resonant mode in the low- and high-temperature limits.
506: Note that one has $\theta_1 \approx 140$,
507: if $L_0=1\,$cm and $T=300\,$K.
508:
509: Since Hamiltonian (\ref{genHam}) is {\em quadratic\/} with respect to
510: canonical coordinates and momenta), the initial thermal state
511: is transformed with time to a generic {\em Gaussian quantum state\/}
512: (whose density matrix or Wigner function is a Gaussian exponential).
513: The single-mode density matrices of such states are completely characterized
514: (in the case of the mean values of quadrature components), besides the
515: mean energy, by the {\em invariant uncertainty product\/} (IUP)
516: \be
517: {\cal D}\equiv \langle x^2\rangle \langle p^2\rangle
518: -\langle (xp+px)/2\rangle^2.
519: \label{IUP}
520: \ee
521: For $\nu>1/2$ this quantity varies in time periodically. For the first mode,
522: \beqn
523: {\cal D}_1 &=& \frac{\theta_1^2}{4}\left[\cos^4(\rho\tau) +
524: \sin^2(2\rho\tau)\frac{2\nu\,\theta_{31}-1}{2(2\nu-1)}
525: \right.\nonumber \\&& \left. +
526: \sin^4(\rho\tau)\left(\frac{2\nu\,\theta_{31}+1}{2\nu-1}
527: \right)^2\right].
528: \label{Dtau}
529: \eeqn
530: For another excited mode
531: one should interchange indices $1$ and $3$ in (\ref{Dtau}).
532:
533: The {\em purity\/} of quantum Gaussian states is expressed through IUP as
534: $\mbox{Tr}(\hat\rho^2)=(4{\cal D})^{-1/2}$ (here $\hat\rho$ is the
535: statistical operator of the state).
536: For initial vacuum states,
537: \be
538: {\cal D}_{1,3}=
539: \frac14\left[1 + \frac{8\nu}{(2\nu-1)^2}\sin^4(\rho\tau)\right],
540: \label{Dvac}
541: \ee
542: so that the states remain practically pure, if $\nu \gg 1$.
543: This is a significant difference from the case of a single resonance
544: mode coupled to a harmonic oscillator detector, studied in \cite{D95,DK96}.
545: If $\theta_1 \neq \theta_3$ (nonzero initial temperature) and $\nu \gg 1$,
546: then
547: \[
548: {\cal D}_1 \approx \frac14\left[ \theta_1 \cos^2(\rho\tau)+
549: \theta_3 \sin^2(\rho\tau)\right]^2,
550: \]
551: so that coupled modes {\em exchange their purities\/} at the moments
552: when $\sin(\rho\tau)=1$.
553:
554:
555: The {\em squeezing coefficient\/}, defined as the ratio of the minimal
556: value of any canonical variance ($\sigma_x$ or $\sigma_p$) for the
557: period of fast oscillations (with frequencies $\overline{\omega}$ or
558: $3\overline{\omega}$) to the vacuum value ($(2\omega_k^{(0)})^{-1}$ or
559: $\omega_k^{(0)}/2$), can be expressed through
560: $\tilde{\cal E}\equiv {\cal E}/\omega_k^{(0)}$ and ${\cal D}$ as
561: \cite{princ,Pavel}
562: \be
563: s=\frac{2{\cal D}}{\tilde{\cal E} +\sqrt{\tilde{\cal E}^2 -{\cal D}}}.
564: \label{s}
565: \ee
566: For thermal states, $s(0)=\theta$, but
567: asymptotically one can obtain any desired degree of squeezing
568: (even for initial high-temperature states) in each resonant mode,
569: because $s \approx {\cal D}(\tau)/\tilde{\cal E}(\tau)$ for $\tau \gg 1$.
570: Evidently, $\tilde{\cal E}$ is nothing but the mean number of photons
571: in the mode, if $\tilde{\cal E} \gg 1$. The variance of the photon
572: distribution in the Gaussian states is given by the formula \cite{1mod}
573: \[
574: \sigma_n= 2 \tilde{\cal E}^2 - {\cal D} -1/4.
575: \]
576: The photon statistics is strongly super-Poissonian for
577: $\tau \gg 1$, when
578: $\sigma_n/\langle n\rangle \approx 2 \tilde{\cal E}\gg 1$.
579: Nonetheless, the quantum states of each excited mode become highly
580: nonclassical for $\tau\gg 1$.
581: It is seen from the photon distribution function (PDF),
582: which can be expressed in terms of two parameters,
583: $\tilde{\cal E}$ and ${\cal D}$,
584: for any Gaussian state
585: with zero mean values of quadrature variables \cite{1mod}):
586: \begin{eqnarray}
587: {\cal P}_n&=&\frac{2}{\sqrt{1+4\tilde{\cal E}+4{\cal D}}}
588: \left(\frac{1+4{\cal D} -4\tilde{\cal E}}{1+4{\cal D}+4\tilde{\cal E}}
589: \right)^{n/2}\nonumber\\
590: &\times&P_n\left(\frac{4{\cal D}-1}
591: {\sqrt{(4{\cal D}+1)^2-16\tilde{\cal E}^2}}\right).
592: \label{dist}
593: \end{eqnarray}
594: Here $P_n(z)$ is the Legendre polynomial.
595: Initially $\tilde{\cal E}^2(0) ={\cal D}(0)$, and (\ref{dist})
596: is a monotonous geometric distribution,
597: ${\cal P}_n(0)= 2(\theta-1)^n/(\theta+1)^{n+1}$.
598: Since we are interested in the cases of large numbers of photons
599: created due to the parametric resonance,
600: it is convenient to use asymptotical forms of the exact distribution
601: (\ref{dist}) for $n\gg 1$. Note that the argument of the Legendre
602: polynomial in (\ref{dist}) is always outside the interval $(-1,1)$,
603: being equal to $1$ only at the initial moment (for thermal states).
604: With the course of time this argument
605: increases to $\infty$, becomes pure imaginary, and asymptotically
606: goes to zero, when $\tilde{\cal E}\gg {\cal D}$.
607: Therefore it is convenient to use the asymptotical formula for $n\gg 1$
608: \cite{Olver}
609: \begin{equation}
610: P_n(\cosh\xi )\approx
611: \left(\frac {\xi}{\sinh\xi}\right)^{1/2}
612: I_0\left(\left[n+1/2\right]\xi\right)
613: \label{as-Olv}
614: \end{equation}
615: (where $I_0(z)$ is the modified Bessel function), because it
616: holds even for complex values of variable $\xi$, provided
617: $\mbox{Re}\xi \ge 0$ and $|\mbox{Im}\xi |<\pi$. We are interested in the
618: case when the mean energy of each mode significantly exceeds its initial
619: value, i.e., $\tilde{\cal E} \gg \sqrt{{\cal D}}$. If ${\cal D} \gg 1$
620: (high temperature initial states), then it is possible that
621: $\tilde{\cal E} <{\cal D} +1/4$. In such a case, $\xi$ is real and large,
622: so that one can replace the modified Bessel function $I_0(x)$ by
623: its asymptotical expression for $x \gg 1$,
624: $I_0(x) \approx (2\pi x)^{-1/2}\exp(x)$.
625: Moreover, one can use either of approximate equalities
626: $\exp(x)\approx 2\cosh(x) \approx 2\sinh(x)$.
627:
628: When $\tilde{\cal E} > {\cal D} +1/4$ (this is just the regime when
629: {\em squeezing\/} happens), the argument of the Legendre
630: polynomial becomes pure imaginary, and the variable $\xi$ becomes complex:
631: $\xi =-i\pi/2 +y$, with $y>0$. Then one should replace the modified
632: Bessel function in (\ref{as-Olv}) by the usual Bessel function
633: $J_0\left(\left[n+1/2\right][\pi/2 +iy]\right)$. Since the absolute value
634: of the argument of this function is also large, one can replace
635: $J_0(x)$ by $(\pi x/2)^{-1/2}\cos(x -\pi/4)$. In this case, we have
636: different expressions for even and odd values of index $n$
637: (this is especially clear from the initial formula (\ref{dist}), when the
638: argument of the Legendre polynomial is close to zero for
639: $\tilde{\cal E} \gg {\cal D}$).
640: Finally, we arrive at the following asymptotical expression,
641: which holds both for real and imaginary values of the argument
642: of the Legendre polynomial in (\ref{dist})
643: (provided $\tilde{\cal E}^2 \gg {\cal D}$):
644: \beqn
645: {\cal P}_n &\approx& \frac{\sqrt2
646: \left|1+4{\cal D} -4\tilde{\cal E}\right|^{n/2}}
647: {\sqrt{\pi n \tilde{\cal E}}
648: \left(1+4{\cal D} +4\tilde{\cal E}\right)^{(n+1)/2}}
649: \nonumber\\&&\times
650: \left\{
651: \begin{array}{ll}
652: \cosh\left([n+1/2]\ln|\chi|\right),
653: & n \;\; \mbox{even}\\
654: \sinh \left([n+1/2]\ln|\chi|\right),
655: & n \;\; \mbox{odd}
656: \end{array}
657: \right.
658: \label{as-unif}
659: \eeqn
660: where
661: \be
662: \chi=\frac{4{\cal D}-1 +4\sqrt{\tilde{\cal E}^2 - {\cal D}}}
663: {\sqrt{(4{\cal D}+1)^2-16\tilde{\cal E}^2}}.
664: \label{defchi}
665: \ee
666: If the ratio $\tilde{\cal E} / {\cal D}$ is not very large, then
667: $\ln|\chi|$ is not small, and one can replace $\cosh$ and $\sinh$
668: functions in (\ref{as-unif}) by exponentials. In this case we have
669: a nonoscillating ``quasigeometric'' distribution, i.e.,
670: a geometric distribution modified by a slowly decreasing factor $n^{-1/2}$:
671: \be
672: {\cal P}_n \approx \frac{1}{\sqrt{2\pi n \tilde{\cal E}}}
673: \left(\frac{4{\cal D} -1 +4\sqrt{\tilde{\cal E}^2 - {\cal D}}}
674: {4{\cal D} +1 +4\tilde{\cal E}}\right)^{n+1/2}.
675: \label{dist-mon}
676: \ee
677: On the contrary, if $\tilde{\cal E} \gg {\cal D}$ (then $\chi$ is pure
678: imaginary and small), we have a strongly oscillating distribution
679: for $n < \tilde{\cal E} / {\cal D}$
680: (typical for squeezed states):
681: \beqn
682: {\cal P}_n &\approx& \sqrt{\frac{2}{\pi n \tilde{\cal E}}}
683: \left(1- \frac{4{\cal D} +1}{2\tilde{\cal E}}\right)^{n/2}
684: \nonumber\\ && \times
685: \left\{
686: \begin{array}{ll}
687: \cosh\left(n \frac{4{\cal D} -1}{4\tilde{\cal E}}\right),
688: & n \;\; \mbox{even}\\
689: \sinh\left(n \frac{4{\cal D} -1}{4\tilde{\cal E}}\right)
690: & n \;\; \mbox{odd}
691: \end{array}
692: \right.
693: \label{asPn}
694: \eeqn
695: Only the ``tail'' of distribution (\ref{asPn}) does not oscillate
696: (and does not depend on ${\cal D}$):
697: \be
698: {\cal P}_n \approx \frac{\exp\left[-n/(2\tilde{\cal E})\right]}
699: {\sqrt{2\pi n \tilde{\cal E}}}, \quad
700: n \frac{4{\cal D} -1}{4\tilde{\cal E}} \gg 1.
701: \label{tail}
702: \ee
703: For the vacuum initial state, the PDF
704: oscillates from the beginning. For $\tau\gg 1$ and $\nu\gg 1$
705: we can write (for each excited mode)
706: \beqnn
707: {\cal P}_n &\approx& e^{-\tau}\sqrt{\frac{8}{\pi n}}
708: \left(1-4e^{-2\tau}\right)^{n/2}
709: \\ && \times
710: \left\{
711: \begin{array}{ll}
712: \cosh\left( \frac{2n}{\nu}\sin^4(\rho\tau) e^{-2\tau}\right),
713: & n \;\; \mbox{even}\\
714: \sinh\left( \frac{2n}{\nu}\sin^4(\rho\tau) e^{-2\tau}\right),
715: & n \;\; \mbox{odd}
716: \end{array}
717: \right.
718: \eeqnn
719: Oscillations exist for any $n$ at the moments of time when
720: $\sin(\rho\tau)=0$.
721: Two examples of the photon distribution functions
722: for the initial vacuum and thermal states
723: (calculated using the exact formula (\ref{dist}))
724: are given in figures 2 and 3.
725:
726:
727: \section{Influence of detunings}
728:
729: Now let us consider the generic case of nonzero detunings.
730: Hereafter we use the normalized parameters
731: $\tilde{\delta}\equiv \delta/\ep$ and $\tilde{\Delta}\equiv \Delta/\ep$.
732: The condition of the parametric resonance is that at least one of the
733: eigenvalues $\lambda_{\pm}$, given by equations
734: (\ref{lam1}) or (\ref{lam2}), has nonzero real part. Since we assume
735: that $\nu>1$, the coefficient $a$ (\ref{def-a}) is always negative.
736: Looking at the eigenvalues in the form (\ref{lam1}), we see that there are
737: two possibilities.
738:
739: The first one is to choose $c<0$. Then complex number $a+\sqrt{c}$
740: has nonzero imaginary part, so its square root inevitably
741: has nonzero real part.
742: Designating $\tilde{\Delta}- 4\tilde{\delta}=\eta$, we see that $c<0$
743: for $|\eta| <\eta_c$, where the critical value $\eta_c$ (which
744: depends on $\nu$ and $\tilde{\delta}$) must be obviously less than $1$.
745: If $\nu\gg 1$ and $|\delta|\sim 1$, then $\eta_c$
746: is close to $1$, with corrections of the order of $\nu^{-1}$.
747: If $|\tilde{\delta}| \gg 1$, then for $1\sim|\eta|\ll|\tilde{\delta}|$
748: one can write
749: $c\approx 2\nu\left(\eta^2-1\right) + 4\tilde{\delta}^2\eta^2$
750: (taking into account that
751: $\tilde{\Delta}- 2\tilde{\delta} = 2\tilde{\delta} +\eta
752: \approx 2\tilde{\delta}$). In this way we obtain the following approximate
753: inequality, describing the region of resonance in the parameter plane
754: $\tilde{\delta},\tilde{\Delta}$ (see figure 4):
755: \be
756: |\tilde{\Delta}- 4\tilde{\delta}| < \eta_c\approx
757: \sqrt{\frac{\nu}{\nu +2\tilde{\delta}^2}}.
758: \label{cond-}
759: \ee
760: Consequently, one can always excite both the modes, compensating
761: one detuning by another.
762: For example, if $\tilde{\Delta}= 4\tilde{\delta}$
763: and $\nu \gg 1$, then
764: $b=a^2-c$ with $|c|\ll a^2$, so that
765: $\sqrt{b} \approx |a|-c/(2|a|)$.
766: In this case equation (\ref{lam2}) yields
767: \be
768: \lambda_{\pm} \approx \frac{\ep}{2}\left[\sqrt{\frac{\nu}
769: {\nu +2\tilde{\delta}^2}} \pm
770: i\sqrt{2\left(\nu+2\tilde{\delta}^2\right)}\right].
771: \label{Relam}
772: \ee
773: As far as $\tilde{\delta}^2 \ll \nu$, the maximal value of the increment
774: coefficient $\mbox{Re}(\lambda_{\pm})$ is practically the same as in the case
775: of strict resonance (\ref{lam0}).
776: Energies of both modes have the same order of magnitude (as in the
777: strict resonance case), therefore
778: this regime of excitation can be named ``symmetrical''.
779: However, with increase of $|\tilde{\delta}|$ the increment of energy growth
780: decreases (by the same law as the resonance width $\eta_c$), and for
781: $\tilde{\delta}^2\gg \nu$ we have
782: $ \mbox{Re}(\lambda_{\pm}) \approx \ep
783: \left[\nu/(8\tilde{\delta}^2)\right]^{1/2} \ll \ep
784: $.
785:
786: The second possibility to swing the modes is to choose $b<0$ in
787: (\ref{def-b}). Then $c > a^2$, and the eigenvalue $\lambda_+$ is real
788: (whereas $\lambda_-$ is pure imaginary). There are two symmetrical
789: regions of swinging in the
790: parameter plane $\tilde{\delta},\tilde{\Delta}$, which are located
791: between the branches of hyperbolas (see figure 4):
792: \beqn
793: \frac{\nu}{2(1-\tilde{\delta})} < \gamma
794: < -\,\frac{\nu}{2(1+\tilde{\delta})}, && |\tilde{\delta}| >1,
795: \label{cond-2} \\
796: \frac{\nu}{2(1-\tilde{\delta})} < \gamma
797: \;\; \mbox{or} \;\;
798: \gamma < -\,\frac{\nu}{2(1+\tilde{\delta})},
799: && |\tilde{\delta}| \le 1,
800: \label{cond-2a}
801: \eeqn
802: where $\gamma \equiv \tilde{\Delta}- 3\tilde{\delta}$.
803: If $|\tilde{\delta}|\le 1$, then we may parametrize $\gamma$ as
804: $\gamma= -\nu\xi/4$, where $|\xi|\ge 1$. In this case, the
805: coefficient $c$ has the following structure:
806: $c=\gamma^4\left(1 +2\nu/\gamma^2 +c_2/\gamma^2 +\ldots\right)$,
807: where $c_2\sim 1$.
808: Using formula (\ref{lam1}) and expanding $\sqrt{c}$ as
809: $\sqrt{c}= \gamma^2\left[1 +\nu/\gamma^2 +c_2/(2\gamma^2)
810: -\nu^2/(2\gamma^4) +\ldots\right]$, we find, neglecting corrections
811: of the order of $\nu^{-1}$,
812: \be
813: \lambda_+^2/\ep^2 = 1- \tilde{\delta}^2 +
814: 4(\tilde{\delta}\xi -1)/\xi^2.
815: \label{lamplus0}
816: \ee
817: If $|\xi| \gg 1$, then we have the same expression for $\lambda_+$
818: as in the case of a single resonance mode \cite{D95,DK96}:
819: $\lambda_+= \sqrt{\ep^2-\delta^2}$. This means that the second mode
820: goes out of resonance, if its detuning $\Delta$ satisfies the inequality
821: $|\Delta|\gg \nu$. It is interesting, however, that adjusting two
822: detunings, one can achieve the maximal possible value
823: $\lambda_+^{(max)}\approx\ep$ even for $\delta\sim\ep$, when the photon
824: generation becomes impossible in the single-mode case: this happens for
825: $\xi_*=2/\tilde{\delta}$, i.e., $\gamma_* =-\nu/(2\tilde{\delta})$.
826:
827: For $|\tilde{\delta}| >1$ we write
828: $\gamma =-\,\frac12 \nu/(\tilde{\delta} +\chi)$.
829: Then the resonance exists for $-1 <\chi < 1$.
830: Under the condition $\nu\gg 1$, the real eigenvalue $\lambda_+$
831: can be written as (we neglect corrections of the relative order of
832: $\nu^{-1}$)
833: \be
834: \lambda_{+} \approx
835: \frac{\ep\,\nu\, \sqrt{1-\chi^2}}{\nu +2\tilde{\delta}^2} .
836: \label{lamplus}
837: \ee
838: Again, one can achieve the maximal possible value
839: $\lambda_+^{(max)}\approx\ep$,
840: if $\nu \gg \tilde{\delta}^2$, but for $\nu \ll \tilde{\delta}^2$
841: the maximal eigenvalue decreases as $\ep\,\nu/(2\tilde{\delta}^2)$.
842:
843: The resonance width $\Gamma_{\Delta}$ (the distance between two points
844: of intersections between the vertical line $\tilde{\delta}=const$
845: and the hyperbolas limiting the region of resonance in figure 4) equals
846: $\Gamma_{\Delta}=\nu/(\tilde{\delta}^2 -1)$ (for $|\tilde{\delta}|> 1$),
847: so it decreases rapidly for $|\tilde{\delta}|\gg 1$. Similar widths
848: $\Gamma_{\delta}^{(l,r)}$ (determined in an obvious way from the points
849: of intersections between the horizontal line $\tilde{\Delta}=const$
850: and the hyperbolas) are given by the expressions
851: $\Gamma_{\delta}^{(l,r)}=1 \pm \sigma$, where
852: \[
853: \sigma=\frac16\left[\sqrt{(\tilde{\Delta}+3)^2 +6\nu}
854: -\sqrt{(\tilde{\Delta}-3)^2 +6\nu}\right].
855: \]
856: The sum of these widths does not depend on $\tilde{\Delta}$, and
857: the smallest of them decreases as
858: $\Gamma_{\delta}^{(small)}\approx 3\nu/(4\tilde{\Delta}^2)$ for
859: $|\tilde{\Delta}|\gg\nu$.
860:
861: Simple explicit solutions of the equations of motion can be found in the
862: special case $\tilde{\delta}=1$, $\gamma=-\nu/2$ (when the generation stops
863: for the single mode, but the increment takes its maximal value for coupled
864: two modes). In this case, the eigenvalues of matrix ${\cal A}$ (\ref{matrA})
865: are as follows,
866: \[
867: \lambda_+=R= 1-\frac{2}{\nu} +{\cal O}(\nu^{-2}),
868: \]
869: \[
870: \lambda_-= iJ, \quad J= \frac{\nu}{2} +1 +{\cal O}(\nu^{-2}).
871: \]
872: With an accuracy up to terms of the order of $\nu^{-1}$,
873: the solutions read as
874: \beqn
875: &&x_1(t) = x_1(0)
876: \left[\left(1-\frac{2}{\nu}\right) C_{1}^{-}(2R\tau;t)
877: +\frac{2}{\nu}\cos\phi_1 \right]
878: \nonumber\\&&
879: - p_1(0)
880: \left[\left(1-\frac{2}{\nu}\right) S_{1}^{-}(2R\tau;t)
881: -\frac{2}{\nu}\sin\phi_1 \right]
882: \nonumber\\&&
883: +\frac{x_3(0)}{4\mu}
884: \left[C_{1}^{-}(2R\tau;t) -\cos\phi_1 \right]
885: \nonumber\\&&
886: -\frac{p_3(0)}{12\mu}
887: \left[S_{1}^{-}(2R\tau;t) +\sin\phi_1 \right]
888: \label{x1t1}
889: \eeqn
890: \beqn
891: &&x_3(t) = x_3(0)
892: \left[\left(1-\frac{2}{\nu}\right)\cos\phi_3
893: +\frac{2}{\nu} C_{3}^{-}(2R\tau;t) \right]
894: \nonumber\\&&
895: +\frac13 p_3(0)
896: \left[\left(1-\frac{2}{\nu}\right)\sin\phi_3
897: -\frac{2}{\nu} S_{3}^{-}(2R\tau;t) \right]
898: \nonumber\\&&
899: +\frac{x_1(0)}{12\mu}
900: \left[C_{3}^{-}(2R\tau;t) -\cos\phi_3 \right]
901: \nonumber\\&&
902: -\frac{p_1(0)}{12\mu}
903: \left[S_{3}^{-}(2R\tau;t) +\sin\phi_3 \right],
904: \label{x2t1}
905: \eeqn
906: where the notation is the same as in the preceding section, and
907: \[
908: \phi_{k}(\tau;t)=k\overline{\omega} t - 2J\tau.
909: \]
910:
911: The mean energies of each mode depend on time as follows
912: (we neglect corrections
913: of the order of $\nu^{-2}$, because they are small; moreover, they
914: increase with time slower than the preserved terms of the order of
915: $\nu^{-1}$):
916: \beqn
917: {\cal E}_1 &=& \frac{\theta_1}{2}\left[
918: \left(1-\frac{4}{\nu}\right)\cosh(4R\tau) + \frac{4}{\nu}\psi(\tau)
919: \right]
920: \nonumber\\&& +
921: \frac{\theta_3}{\nu}\left[\cosh(4R\tau) + 1 -2\psi(\tau)\right],
922: \label{E1ofr}
923: \eeqn
924: \beqn
925: {\cal E}_3 &=& \frac32{\theta_3}\left[
926: 1-\frac{4}{\nu} + \frac{4}{\nu}\psi(\tau)\right]
927: \nonumber\\&& +
928: \frac{3\theta_1}{\nu}\left[\cosh(4R\tau) + 1 -2\psi(\tau)\right],
929: \label{E3ofr}
930: \eeqn
931: where
932: \[
933: \psi(\tau) \equiv \cosh(2R\tau)\cos(2J\tau).
934: \]
935: We see that the energy of the first mode is almost the same
936: that it would be in the case of a single mode under the condition
937: of the strict resonance, i.e., $\theta_1\cosh(4\tau)/2$.
938: The energy of the third mode is significantly less than the energy of the
939: first mode, if $\nu\gg 1$.
940: Therefore this regime of excitation can be named ``asymmetrical''.
941: For $\tau>1$, ${\cal E}_3/ {\cal E}_1\approx 6/\nu$.
942: Note, however, that for the cubical
943: cavity with $\nu=50/3$, the energy of the third mode is only three times
944: less than that of the first one.
945: It is important, nonetheless, that
946: the rates of increase of the energies of each mode are
947: almost twice bigger than they were in the case of the strict resonance
948: discussed in the preceding section.
949:
950: For the invariant uncertainty products we obtain, with the same accuracy,
951: almost identical expressions:
952: \beqn
953: {\cal D}_{1,3} &=& \frac{\theta_{1,3}^2}{4}\left[
954: 1-\frac{8}{\nu} + \frac{8}{\nu}\psi(\tau)\right]
955: \nonumber\\&& +
956: \frac{\theta_1\theta_3}{\nu}\left[\cosh(4R\tau) + 1 -2\psi(\tau)\right].
957: \label{IUP1}
958: \eeqn
959: We see a drastic difference from the strict resonance case: now the
960: IUP of each mode increases (asymptotically) exponentially with time.
961: For $\tau\gg 1$, each mode appears in a highly mixed
962: quantum state, with
963: ${\cal D}_{1}= {\cal D}_{3}=\theta_1\theta_3\exp(4R\tau)/(2\nu)$.
964:
965: The photon distributions in the first and the
966: third modes turn out essentially different (and different from the
967: distributions in the strict resonance case). We have shown in the
968: preceding section, that the form of the PDF (oscillating or not) depends
969: on the ratio $\tilde{\cal E}/{\cal D}$. For the first mode,
970: $\tilde{\cal E}_1/{\cal D}_1 \approx \nu/(2\theta_3)$ for $\tau\gg 1$.
971: Consequently, the asymptotical PDF oscillates for the initial vacuum
972: state and for initial thermal states with not very large mean numbers
973: of photons (less than, approximately, $\nu/4$). Also, the squeezing
974: coefficient (\ref{s}) tends to the finite asymptotical value
975: $s_1^{\infty}=
976: \lim_{\tau\to \infty}({\cal D}_1/\tilde{\cal E}_1) = 2\theta_3/\nu$
977: (typical dependences $s_1(\tau)$ are given in figure~5).
978: This resembles (qualitatively) the behavior of the squeezing
979: coefficient in the one-dimensional cavity with an infinite number
980: of coupled modes \cite{DA}.
981: On the contrary, since ${\cal D}_3/\tilde{\cal E}_3 \to \theta_3$ for
982: $\tau\to \infty$, the final squeezing coefficient of the third mode
983: coincides with the initial value:
984: $s_3(0)=s_3^{\infty}= \theta_3 \ge 1$, with very small deviations from this
985: value in the interval $0<\tau<\infty$.
986: The photon distribution function of the third mode does not oscillate,
987: being close to the ``quasigeometric'' distribution (\ref{dist-mon}).
988:
989:
990:
991:
992:
993: \section{Conclusion}
994:
995: We have studied the problem of photon creation due to the nonstationary
996: (dynamical) Casimir effect in a three- or
997: two-dimensional ideal cavity with an oscillating boundary, in the case when
998: the spectrum of eigenfrequencies contains two frequencies, whose
999: ratio is close to $3$, and the boundary oscillates at the frequency
1000: close to the double lowest field eigenfrequency. We have calculated the mean
1001: energy, squeezing coefficient, invariant uncertainty product, and the
1002: photon distribution in each of two resonance modes, both under the strict
1003: and approximate resonance conditions. We have shown that,
1004: due to a strong coupling between the resonance modes, it is possible
1005: to obtain the photon generation even for relatively large detunings between
1006: the frequencies of the wall and the field,
1007: compensating one detuning by another.
1008: However, the resonance width decreases with increase of
1009: detuning, as well as the increment of the exponential growth of the energy.
1010: We have considered both the vacuum and thermal initial states, having
1011: demonstrated not only an amplification of the number of created photons
1012: due to the initial thermal fluctuations, but also such effects as the
1013: ``purity exchange'' between the modes and the transformation of the
1014: initial smooth photon distribution to an oscillating distribution,
1015: typical for strongly squeezed states.
1016: Our results show that choosing a proper set of parameters
1017: of realistic three-dimensional cavities,
1018: one could facilitate observing
1019: the nonstationary Casimir effect, which is a challenge for
1020: experimentalists. However, a correct account of the influence of
1021: damping is still an unsolved problem.
1022:
1023: \section*{Acknowledgement}
1024: The authors are grateful to the Brazilian agency CNPq for the support.
1025:
1026: \newpage
1027: \begin{thebibliography}{99}
1028:
1029: \bibitem{review} V.V. Dodonov,
1030: in: M.W. Evans (Ed.), Modern Nonlinear Optics,
1031: Advances in Chem. Phys. Series, vol. 119, Wiley, New York, 2001,
1032: Part 3, p.309.
1033:
1034: \bibitem{DKM89} V.V. Dodonov, A.B. Klimov, V.I. Man'ko,
1035: Phys. Lett. A 142 (1989) 511.
1036:
1037:
1038: \bibitem{Sch} J. Schwinger,
1039: Proc. Nat. Acad. Sci. USA 90 (1993) 958.
1040:
1041: \bibitem{BE} G. Barton, C. Eberlein,
1042: Ann. Phys. (NY) 227 (1993) 222.
1043:
1044: \bibitem{Lamb} A. Lambrecht, M.-T. Jaekel, S. Reynaud,
1045: Phys. Rev. Lett. 77 (1996) 615.
1046:
1047: \bibitem{Law} C.K. Law,
1048: Phys. Rev. A 49 (1994) 433; 51 (1995) 2537.
1049:
1050: \bibitem{D95} V.V. Dodonov,
1051: Phys. Lett. A 207 (1995) 126.
1052:
1053: \bibitem{DK96} V.V. Dodonov, A.B. Klimov,
1054: Phys. Rev. A 53 (1996) 2664.
1055:
1056: \bibitem{D98} V.V. Dodonov,
1057: J. Phys. A 31 (1998) 9835.
1058:
1059: \bibitem{DA}
1060: V.V. Dodonov, M.A. Andreata,
1061: J. Phys. A 32 (1999) 6711.
1062:
1063: \bibitem{AD00}
1064: M.A. Andreata, V.V. Dodonov,
1065: J. Phys. A 33 (2000) 3209.
1066:
1067: \bibitem{ViHaJa95} C. Villarreal, S. Hacyan, R. J\'auregui,
1068: Phys. Rev. A 52 (1995) 594.
1069:
1070: \bibitem{Mund} D.F. Mundarain, P.A. Maia Neto,
1071: Phys. Rev. A 57 (1998) 1379.
1072:
1073: \bibitem{Ji98a} J.-Y. Ji, K.-S. Soh, R.-G. Cai, S.P. Kim,
1074: J. Phys. A 31 (1998) L457.
1075:
1076: \bibitem{Law95} C.K. Law, S.-Y. Zhu, M.S. Zubairy,
1077: Phys. Rev. A 52 (1995) 4095.
1078:
1079: \bibitem{JaVi96} R. J\'auregui, C. Villarreal,
1080: Phys. Rev. A 54 (1996) 3480.
1081:
1082: \bibitem{TaKo98} T. Taneichi, T. Kobayashi,
1083: J. Phys. Soc. Japan 67 (1998) 1594.
1084:
1085: \bibitem{Janow98} M. Janowicz,
1086: Phys. Rev. A 57 (1998) 5016.
1087:
1088: \bibitem{Fedot00} A.M. Fedotov, N.B. Narozhny, Y.E. Lozovik,
1089: Phys. Lett. A 274 (2000) 213.
1090:
1091: \bibitem{Maz01}
1092: M. Crocce, D.A.R. Dalvit, F.D. Mazzitelli,
1093: Phys. Rev. A 64 (2001) 013808.
1094:
1095: \bibitem{Plun} R. Sch\"utzhold, G. Plunien, G. Soff,
1096: Phys. Rev. A 57 (1998) 2311.
1097:
1098: \bibitem{Cole01} C.K. Cole, W.C. Schieve,
1099: Phys. Rev. A 64 (2001) 023813.
1100:
1101: \bibitem{Louis} W.H. Louisell,
1102: Coupled Mode and Parametric Electronics, Wiley, New York, 1960.
1103:
1104: \bibitem{Land} L.D. Landau, E.M. Lifshitz, Mechanics,
1105: Pergamon, Oxford, 1969.
1106:
1107: \bibitem{Bogol} N.N. Bogoliubov, Y.A. Mitropolsky,
1108: Asymptotic Methods in the Theory of Non-Linear Oscillations,
1109: Gordon \& Breach, New York, 1985.
1110:
1111: \bibitem{D98a} V.V. Dodonov,
1112: Phys. Lett. A 244 (1998) 517;
1113: Phys. Rev. A 58 (1998) 4147.
1114:
1115: \bibitem{princ} A. Luk\v s, V. Pe\v rinov\'a, Z. Hradil,
1116: Acta Phys. Polon. A 74 (1988) 713.
1117:
1118: \bibitem{Pavel}
1119: V.V. Dodonov, V.I. Man'ko, P G. Polynkin,
1120: Phys. Lett. A 188 (1994) 232.
1121:
1122: \bibitem{1mod} V.V. Dodonov, O.V. Man'ko, V.I. Man'ko,
1123: Phys. Rev. A 49 (1994) 2993.
1124:
1125: \bibitem{Olver} F.W.J. Olver, Asymptotics and Special Functions,
1126: Academic Press, New York, 1974.
1127:
1128:
1129: \end{thebibliography}
1130:
1131: %\end{multicols}
1132: %\vfill
1133: \newpage
1134:
1135: %\newpage
1136:
1137: \begin{figure}
1138: \caption{
1139: Mean energies (normalized by their initial values at $\tau=0$)
1140: of the first and the third modes of the cubic cavity with $\nu=50/3$
1141: in the case of the strict resonance, versus the dimensionless
1142: ``slow time'' $\tau$ (\ref{deftaurho}). The order of the curves
1143: in the interval of time between 1.4 and 1.6 is as follows (from bottom
1144: to top):
1145: the first mode for $\theta_{31}=1/3$
1146: (high-temperature initial states);
1147: the first mode for $\theta_1=\theta_3=1$ (vacuum initial state);
1148: the third mode for $\theta_1=\theta_3=1$;
1149: the third mode for $\theta_{13}=3$.
1150: }
1151: \end{figure}
1152:
1153: \begin{figure}
1154: \caption{
1155: The photon distribution function ${\cal P}(n)$ for the first mode
1156: and the initial vacuum state, under the condition of the strict resonance,
1157: for the ``slow time'' $\tau=5\pi/(2\rho)$ and $\nu=50/3$.
1158: }
1159: \end{figure}
1160:
1161: \begin{figure}
1162: \caption{
1163: The photon distribution function ${\cal P}(n)$ for the first mode
1164: and the initial thermal state with $\theta_1=5$
1165: (i.e., for $\langle n(0)\rangle= 2$ in the first mode),
1166: under the condition of the strict resonance,
1167: for the ``slow time'' $\tau=5\pi/(2\rho)$ and $\nu=50/3$.
1168: }
1169: \end{figure}
1170:
1171: \begin{figure}
1172: \caption{
1173: Regions of photon generation in the plane $(\tilde{\delta},\tilde{\Delta})$
1174: of the detuning parameters (tildes have been omitted).
1175: The region of ``symmetrical generation''
1176: $|\tilde{\Delta} -4\tilde{\delta}|<\eta_c$
1177: is between two dashed (almost straight) lines.
1178: Two regions of ``asymmetrical generation'' are confined with
1179: hyperbolas drawn as continuous curves.
1180: }
1181: \end{figure}
1182:
1183: \begin{figure}
1184: \caption{
1185: The squeezing coefficient (\ref{s}) versus ``slow time'' $\tau$
1186: (\ref{deftaurho}) for the first mode in the case of ``asymmetrical
1187: generation'', for $\tilde{\delta}=1$, $\gamma=-\nu/2$, and
1188: $\nu=50/3$. The lower curve corresponds to the case $\theta_1=1$
1189: (vacuum initial state). The upper curve corresponds to the case
1190: $\theta_1=5$.
1191: }
1192: \end{figure}
1193:
1194:
1195: \end{document}
1196:
1197: