nlin0510004/etd.tex
1: \documentclass[twocolumn,showpacs,epsfig]{revtex4}
2: \usepackage{graphicx}
3: %\usepackage{dcolumn}
4: %\usepackage{bm}
5: 
6: \begin{document}
7: 
8: \draft
9: \title{Survival probability time distribution in dielectric cavities}
10: \author{Jung-Wan Ryu$^{1,2}$}
11: \author{Soo-Young Lee$^1$}
12: \author{Chil-Min Kim$^1$}
13: \author{Young-Jai Park$^2$}
14: \affiliation{$^1$ National Creative Research Initiative Center for Controlling Optical Chaos,\\
15: Pai-Chai University, Daejeon 302-735, Korea}
16: \affiliation{$^2$ Department of Physics, Sogang University, Seoul 121-742, Korea}
17: 
18: 
19: \begin{abstract}
20: We study the survival probability time distribution (SPTD) in dielectric cavities.
21: In a circular dielectric cavity the SPTD has an algebraic long time behavior, $\sim t^{-2}$
22: in both TM and TE cases, but shows different short time behaviors due to the existence of 
23: the Brewster angle in TE case where the short time behavior is exponential.
24:  The SPTD for a stadium-shaped cavity decays exponentially, and
25: the exponent shows a relation of $\gamma \sim n^{-2}$, $n$ is the refractive index,
26:  and the proportional coefficient 
27: is obtained from a simple model of the steady probability distribution. 
28:  We also discuss about the SPTD for a quadrupolar deformed cavity and
29: show that the long time behavior can be algebraic or exponential depending on
30:  the location of islands.
31: \end{abstract}
32: \pacs{05.60.Cd, 42.15.-i, 05.45.-a}
33: \maketitle
34: \narrowtext
35: 
36: \section{Introduction}
37: 
38: Recently, lasing modes from dielectric microcavities have attracted much attention
39: due to its potential application to optoelectric circuits and optical communications \cite{Cha96}.
40: In particular, there was a lot of theoretical and experimental effort 
41: to excite directional lasing modes in deformed microcavities \cite{Noc97}.
42: It is now well known that the lasing pattern has a very close relationship with
43: the internal ray dynamics given by the boundary geometry of cavity.
44: It is also reported that the property of the openness of the dielectric cavity 
45: plays an important role in the resonance pattern analysis \cite{Lee04,Har04}. 
46: 
47: 
48: For a general open system, the survival probability time distribution (SPTD)
49: or its time derivative, the escape time distribution, is a basic physical quantity. 
50: Many studies are focused on the relation between the long time behavior of the SPTD and
51: the internal dynamics, and it is known that the SPTD has algebraic and
52: exponential decays in integrable and chaotic systems, respectively
53: \cite{Bau90,Dor90,Fri01,Bun05,Alt96,Sch02}.
54: In mixed systems, having both integrable islands and chaotic sea in phase space,
55: the SPTD has algebraic long time behavior which originates from the slow
56: escape mechanism due to the stickiness of KAM tori \cite{Fen94}.
57: 
58: 
59: 
60: The property of openness of the dielectric cavity is different from the
61: open systems previously studied \cite{Bau90,Dor90,Fri01,Bun05,Alt96,Sch02}.
62: Rays can escape through any boundary
63: point, and  partial escapes, depending on the incident angle, are possible.
64: This unique property can be reflected on the long time behavior of
65: the SPTD. 
66: 
67: 
68: In this paper, we study the SPTD  in dielectric cavities of various
69: boundary geometries such as circle, stadium, and quadrupole, which are
70: typical examples of integrable, chaotic, and mixed systems, respectively.
71: The SPTDs in these dielectric cavities show basically similar behavior to
72: the open cavity with a small hole on the cavity boundary, but the exponents
73: are different. In particular, we show that the ergodic property cannot be
74: applied for the stadium-shaped dielectric cavity even in the small opening
75: limit, $n\rightarrow \infty$, $n$ is the refractive index. 
76: 
77: The paper is organized as follows. In Sec. II the algebraic long time behavior of
78: the SPTD in the circular dielectric cavity is derived analytically  and confirmed
79: numerically for both TM and TE waves. 
80: It is shown in Sec. III that the SPTD for a stadium-shaped cavity decays exponentially, 
81: and the exponent $\gamma$  has $\sim n^{-2}$ dependence and the proportional coefficient  
82: can be understood from a simple model of the steady probability distribution (SPD).
83: The SPTD in the quadrupole-deformed dielectric cavity is discussed in Sec. IV and
84: we finally summarize the results in Sec. V.  
85: 
86: 
87: 
88: 
89: 
90: \section{Circular dielectric cavity - integrable system}
91: 
92: 
93: Many authors have studied the SPTD
94: for the open billiard with a small hole on boundary \cite{Bau90,Dor90,Fri01,Bun05}.
95: It is known that the SPTD in a circular billiard decays algebraically,
96: $P_{sv}(t) \propto t^{-1}$.
97: In this section we study the SPTD for the circular dielectric cavity,
98: and it will be shown that the SPTD shows still an algebraic decay but the exponent
99: is different.
100: 
101: For simplicity, we focus on TM  wave first and for TE wave we will mention
102: only difference later. In the circular geometry, ray dynamics is integrable and
103: rays in the open area of the phase space, i.e., $-1/n < p < 1/n$, $p=\sin \theta$,
104:  $\theta$ being the incident angle,  can partially 
105: escape from the cavity. 
106: The formal expression of the SPTD in the circular dielectric cavity is given by
107: \begin{eqnarray}
108: P_{sv}(t) =\frac{n}{2L} \int_{0}^{L}ds  \int_{-p_c}^{p_c}dp \,\, R(p)^{N(t)},
109: \label{psv}
110: \end{eqnarray}
111: where $L$ is the boundary length, and $p_c$ is the critical line 
112: for total internal reflection, i.e., 
113: $p_c=\sin \theta_c=1/n$, and  $R(p)$ is the reflection coefficient for TM wave \cite{Haw95},
114: 
115: \begin{equation}
116: R(\theta)=\left({{n \cos{\theta} - \cos{\theta_t}}
117: \over{{n\cos{\theta} + \cos{\theta_t}}}}\right)^2,
118: \label{Rtheta}
119: \end{equation}
120: where $n \sin \theta = \sin \theta_t$, and
121: $N(t)$ is the number of bounce on the boundary. Since $N(t)=t/2\cos \theta$ in the
122: circular geometry, when considering a unit circle and
123: a time scale as the length of ray trajectory, Eq.(\ref{psv}) can
124: be rewritten as   
125: \begin{eqnarray}
126: P_{sv}(t) = n\int_{0}^{\theta_c} d\theta \,\, \cos{\theta}  \exp\left[{-G(\theta)t}\right],
127: \label{pG}
128: \end{eqnarray}
129: where
130: \begin{equation}
131: G(\theta)\equiv {{1} \over {\cos{\theta}}}
132:  \ln \left({1+ {{2\cos{\theta_t}} \over {n\cos{\theta} - \cos{\theta_t}}}}\right).
133: \end{equation}
134: 
135: 
136: 
137: Note that the rays near the critical line $p_c$ can survive longer time and
138: dominate long time tail behavior. Therefore, we can expand $G(\theta)$ from
139: $\theta_c$ by changing variable, $\theta=\theta_c - \chi$, as
140: \begin{eqnarray}
141: G(\theta) \approx \alpha \chi^{1/2} + \beta \chi^{3/2}+ \cdots ,
142: \label{G}
143: \end{eqnarray}
144: where
145: \begin{equation}
146: \alpha = {{2n\sqrt{2\sqrt{n^2 - 1}}}\over{n^2 - 1}}
147: \end{equation}
148: and
149: \begin{equation}
150: \beta =-\alpha (\frac{n^2-6}{4\sqrt{n^2-1}}-\frac{2}{n}). 
151: \end{equation}
152: Substituting the lowest term in Eq.(\ref{G}) into Eq.(\ref{pG}), we can obtain the long time
153: behavior of the SPTD as
154: \begin{eqnarray}
155: P_{sv}(t) &\simeq& 
156:   \frac{2\sqrt{n^2-1}}{\alpha^2} t^{-2}  \left[ 1-(1+ \alpha \sqrt{\theta_c}t)
157: e^{-\alpha \sqrt{\theta_c}t} \right] \nonumber \\
158:  &\simeq& \frac{2\sqrt{n^2-1}}{\alpha^2} \, t^{-2}.
159: \label{t2}
160: \end{eqnarray}
161: We emphasize that  the SPTD for the circular dielectric cavity decays as $t^{-2}$
162: as shown in Eq.(\ref{t2}), different from 
163: the open billiard with a small hole where decays as $t^{-1}$. This means
164: that the property of openness can change the exponent of
165: the algebraic decaying SPTD.  
166: 
167: 
168: 
169:  
170: For TE wave case the reflection coefficient  is given by
171: \begin{equation}
172: R_{TE}(\theta)=\left({{n \cos{\theta_t} - \cos{\theta}}\over{{n\cos{\theta_t} + \cos{\theta}}}}\right)^2,
173: \label{RTE}
174: \end{equation}
175: and the expansion of $G(\theta)$ and the SPTD at a long time are
176: the same as Eq.(\ref{G}) and Eq.(\ref{t2}) with different
177: expansion coefficients, i.e.,
178: \begin{equation}
179: \alpha_{TE}={{2n^3 \sqrt{2 \sqrt{n^2 - 1}}}\over{n^2 - 1}}=n^2 \alpha
180: \end{equation}
181: and
182: \begin{equation}
183: \beta_{TE}=-\frac{\alpha_{TE}}{4\sqrt{n^2-1}} (8n^4 + n^2 + 6).
184: \end{equation}
185: We note that the dependence of the SPTD on the refractive index $n$ in TM and TE waves is
186: quite different, i.e., $P_{sv}(t)\simeq n^2 t^{-2}$ for
187: TM case, but $P_{sv}(t)\simeq n^{-2} t^{-2}$ for TE case.
188: The proportionality of  $n^{-2}$ of the TE case  does not mean that
189: the circular cavity with a higher $n$ is more leaky, since we take into
190: account of only the open region in the phase space,$-1/n <p<1/n$ (see Eq.(\ref{psv})).
191: 
192: 
193: \begin{figure}
194: \begin{center}
195: \includegraphics[width=0.4\textwidth]{fig1.eps}
196: %\includegraphics[width=0.4\textwidth]{}
197: \caption{(color online) The SPTDs of the circular dielectric cavity 
198: for (a) TM and (b) TE waves.
199: Black circle, red square, green diamond, blue triangle (up), and brown triangle (down) are
200: for n=2, 4, 6, 8, and 10, respectively.
201: These shows $t^{-2}$ long time behavior and a very good agreement with the solid lines
202: which represent the $n=2$ case in Eq.(\ref{t2}). 
203: }
204: \end{center}
205: \end{figure}
206: 
207: 
208: In order to perform numerical calculation for the SPTD in the circular cavity,
209:  we take $10^8$ random initial
210: points in the open region of the phase space. We then trace each point
211: with a weight determined by $R(p)$ when bouncing from the boundary, 
212: and sum the weights between $t$ and $t+\Delta$, we take $\Delta=1$ in the calculations,
213: for all points in the ensemble, and finally normalize to be unit when $t=0$.
214: Figure 1 shows the numerical results of the SPTD in the circular cavity for TM and TE cases.
215: It is clear that  the SPTD for both cases shows
216: an algebraic long time behavior, $\sim t^{-2}$, and the dependence on $n$ 
217: is correctly described by Eq.(\ref{t2}) which is indicated by the solid lines for $n=2$
218: in Fig.1.
219: 
220: 
221: 
222: \begin{figure}
223: \begin{center}
224: \includegraphics[width=0.4\textwidth]{fig2.eps}
225: %\includegraphics[width=0.4\textwidth]{}
226: \caption{(color online) (a) The exponential short time behavior of the SPTDs for TE case.
227: The colors for different $n$ is the same as in Fig. 1. These can be fitted as
228:  $\exp(-\gamma(n) t)$ in a short time range.
229: (b) The exponents $\gamma(n)$. The solid line represents the result of a simple 
230: approximation of Eq.(\ref{gamma}) .}
231: \end{center}
232: \end{figure}
233: 
234:  A substantial difference between the TM and TE cases appears in the short time
235: behavior. As shown in Fig. 1 (a) the short time behavior of the TM SPTD is smoothly
236: connected to the $ t^{-2}$ long time tail, on the other hand that of the TE SPTD
237: shows rather an abrupt transition to the algebraic long time tail and the detail of
238: the short time behavior seems to be characterized by an exponential decay.
239: This exponential short time decay is clear in Fig. 2 (a) and the exponent
240: $\gamma(n)$, when fitted as $\exp(-\gamma(n) t)$, appears as the solid dots in Fig. 2 (b).
241: These exponents are well described by a simple approximation for the reflection
242: coefficient $R_{TE} (\theta)$. If we expand  $R_{TE} (\theta)$ at $\theta=0$ and take
243: only the lowest term, then
244: \begin{equation} 
245: P_{sv}(t)\simeq \exp(-\ln(\frac{n+1}{n-1}) t) =\exp(-\gamma(n) t).
246: \label{expand0}
247: \end{equation}
248: The solid line in Fig. 2 (b) represents the relation
249: \begin{equation}
250: \gamma(n)=\ln(\frac{n+1}{n-1}).
251: \label{gamma}
252: \end{equation} 
253: Here we emphasize that although the lowest term of the expansion of Eq.(\ref{expand0})
254: is the same for both TM and TE cases, only TE case allows the exponential short time
255: decay. The reason for this is the existence of the Brewster angle in the TE case, 
256: $\theta_B=\arctan(1/n)$ where rays can escape without reflection, i.e., $R(\theta_B)=0$. 
257: The rays with the incident angle in the range of $-\theta_B < \theta < \theta_B$
258: dominate the short time behavior, while the other parts, $\theta_B < |\theta| < \theta_c$,
259: mainly contribute to the long time algebraic tail.
260: 
261: 
262: 
263: It is important to know when the algebraic decay starts to appear in both TM and TE cases.
264: In TM case, the main factor for the deviation from the $t^{-2}$ decay comes from the
265: finite integral bound, and it corresponds to the terms containing the upper bound
266:  $\theta_c$ in Eq.(\ref{t2}). We then estimate the transition time when the deviation from
267: the $t^{-2}$ decay is about 10 \%, and the result is
268: \begin{equation}
269: t_c \simeq \frac{1.38 (n^2-1)^{3/4}}{n\sqrt{\arcsin(1/n)}} \propto n.
270: \end{equation}
271: In Fig. 1 (a) the corresponding transition times are indicated by arrows and
272: show a good agreement with the numerical calculations for various $n$.
273: 
274: 
275: Due to the existence of the Brewster angle, the transition time for TE SPTD
276: can be determined by a different way. As mentioned above, the TE SPTD shows
277: a short time exponential behavior and a long time algebraic behavior.
278: Therefore we can estimate the transition time by finding the intersection time
279: for both different behaviors.
280: From Eq.(\ref{t2}) with $\alpha_{TE}$ and Eq.(\ref{expand0}),
281:  for a large $n$ we can get an implicit equation for the transition time
282: as
283: \begin{equation}
284: \frac{t_c}{n}\exp(-\frac{t_c}{n})= \frac{1}{2n^2}.
285: \label{tcte}
286: \end{equation} 
287: The transition times for various refractive indices are indicated by arrows in Fig. 1 (b)
288: and well represent the transition
289: times of the numerical results. The solution $t_c$ of the above equation
290: cannot be described by a simple power of $n$, but we can show 
291: \begin{equation}
292: t_c (n) \propto n^{\mu(n)}, \,\,\,\, \mu(n) > 1.
293: \end{equation}
294: If we take a logarithm of Eq.(\ref{tcte}), then we get
295: \begin{equation}
296: \frac{t_c}{n}-\ln \frac{t_c}{n} = \ln 2 n^2,
297: \end{equation}
298: which generally has two solutions and the larger solution is relevant.
299: The point $t_0$, at which the slopes of the two functions in the left hand side of
300: the above equation are identical, should locate between the two solutions.
301: By differentiating the above equation, we get $t_0=n$.
302: Therefore,
303: \begin{equation}
304: t_c > t_0 = n.
305: \end{equation}
306: 
307: Even though both TM and TE cases show the same $t^{-2}$ long time
308: decay in the circular cavity,
309:  the short time behavior and the $n$ dependence of the transition time
310: are quite different. We emphasize that these differences originate
311: from the existence of the Brewster angle in the TE case. 
312: 
313: 
314: 
315: 
316: 
317: \section{Stadium-shaped dielectric cavity - chaotic system}
318: 
319: As an example of chaotic dielectric cavities, we take a stadium-shaped one 
320: with parallel linear segments of a length $l$ and two semicircles of a radius $R$.
321: The stadium-shaped billiard has been a typical chaotic system in the research of   
322: classical and quantum chaos. The escape property through a small hole on the
323: boundary of the stadium-shaped billiard has been investigated by many authors \cite{Alt96}.  
324: They have shown that the escape time distribution exponentially decays first and later
325: becomes algebraic, and the transition time $t_c$ increases as the hole size
326: decreases. The algebraic decay at long times comes from the stickiness near the marginally
327: stable line in phase space corresponding to the bouncing ball trajectories.
328: On the other hand, in the stadium-shaped dielectric cavity the ray trajectories
329: of the bouncing ball type cannot contribute to the long time behavior due to
330: the property of openness, i.e., rays with almost vertical incidence escape
331: easily and contribute to the short time behavior. As a result the SPTD shows only
332: exponential decay (see Fig. 4, 5).
333: 
334: 
335: The exponential decay in the dielectric chaotic cavity implies the existence
336: of the {\it steady probability distribution} (SPD), $P_s(s,p)$ which is defined as the
337: spatial part of the survival probability distribution $\tilde{P}_{sv}(s,p,t)$ \cite{Lee04}.
338: With this SPD, we can express the SPTD as
339: \begin{equation}
340: P_{sv}(t)=\int_0^L ds \int_{-1}^{1} dp \,\, \tilde{P}_{sv}(s,p,t)
341:  \simeq C \exp(-\gamma t),
342: \end{equation}
343: where $C$ is a constant and
344: \begin{equation}
345: \gamma =\int_0^L ds \int_{-1}^{1} dp \,\, P_s (s,p) T(p),
346: \label{gammaspd}
347: \end{equation} 
348: where the transmission coefficient $T(p)$ is given as $T(p)=1-R(p)$.
349: 
350: \begin{figure}
351: \begin{center}
352: \includegraphics[width=0.4\textwidth]{fig3.eps}
353: %\includegraphics[width=0.4\textwidth]{}
354: \caption{The steady probability distributions. (a) $n=2$ case with system parameters $(R,l)=(1,2)$
355:  for TM wave.  (b) $n=10$ case with system parameters $(R,l)=(1,2)$ for TM wave.
356: The black points indicate the rays traced from a random ensemble with weights higher than 0.1
357: at about $t=12$ in $n=2$ case and $t=37$ in $n=10$ case.
358: The solid lines denote the critical line $\pm p_c$ for total internal reflection.
359: }
360: \end{center}
361: \end{figure}
362: 
363: 
364: Note that the above equation is satisfied in the exponential decay region and
365: cannot describe the nonexponential very short time behavior. 
366: From Eq.(\ref{gammaspd}), if we know the SPD, we can estimate the decay rate $\gamma$.
367: However, the structure of the SPD is usually very complicated because it depends
368: on the openness as well as the boundary geometry of the cavity. 
369: Figure 3 (a) shows the approximate of the SPD when $n=2$ which is a
370: snap shot of the $\tilde{P}_{sv}(s,p,t)$ captured at about $t=12$.
371: The partial escape property of the dielectric cavity  allows for rays
372: to distribute on unstable manifold structure in the open region, $-1/n < p < 1/n$. 
373: 
374: 
375: 
376: \begin{figure}
377: \label{sttm}
378: \begin{center}
379: \includegraphics[width=0.4\textwidth]{fig4.eps}
380: %\includegraphics[width=0.4\textwidth]{}
381: \caption{ (color online) The SPTD and the decay rates for TM wave.
382: (a) The exponential SPTD for a stadium dielectric cavity with $(R,l)=(1,1)$.
383: Black, red, green, blue, and brown lines are for n=2, 4, 6, 8, and 10, respectively.
384: (b) The decay rates with increasing $n$. Black circle and red square represent
385: results for system parameters $(R,l)=(1,1)$ and $(1,2)$, respectively.
386: The solid line shows $n^{-2}$ behavior. The coefficient $D$ is given as
387:  $D=\frac{\sqrt{A}L}{2\pi^2R}$. 
388: }
389: \end{center}
390: \end{figure}
391: 
392: 
393: 
394: 
395: Even though it is difficult to estimate the SPD in usual cases, for the large $n$ case,
396: the small opening case, we can simplify the SPD by assuming a uniform distribution
397: over the whole phase space except the open regions related to the linear segments
398: of the stadium boundary. The approximate of the SPD for $n=10$ shown in Fig. 3 (b)
399: supports this assumption. We note that this is a substantial difference from the
400: escape through a small hole on boundary where entirely uniform distribution is
401: assumed due to the ergodic property \cite{Bau90}.
402:  Based on the assumption of the partial ergodicity,
403: we can rewritten the decay rate as
404: \begin{equation}
405: \gamma=\frac{\pi R}{\sqrt{A}(L-2l/n)}\int dp \,\, T(p),
406: \label{gammaT}
407: \end{equation}
408: where we insert the factor $1/\sqrt{A}$, $A=\pi R^2+2Rl$ being the area of the stadium,
409: from the consideration of time scale. 
410: The integral of the above equation means the degree of openness and in the large $n$ limit
411: decreases as $\sim 2\nu \pi n^{-2}$ for both TM ($\nu=1$) and TE ($\nu=2$) cases (see Appendix).
412: Therefore, for the large $n$ limit the decay rate becomes
413: \begin{equation}
414: \gamma \simeq  \frac{2\nu\pi^2 R}{\sqrt{A}L} n^{-2}.
415: \label{gammafinal}
416: \end{equation}
417: 
418: 
419: 
420: Numerical results for the SPTD in the chaotic stadium-shaped dielectric cavity are shown
421: in Figs. 4 and 5 for TM and TE cases, respectively. We take two systems; one is $(R,l)=(1,1)$
422: and the other is $(R,l)=(1,2)$. In calculation, we use a random ensemble of $10^4$ initial
423: points spread over the whole phase space and trace the survival probability with time, the time
424: is scaled as the length of trajectory in the spatial space as before.
425: The exponential behavior of the SPTD is clear even at long time limit in both TM and TE cases.
426: This means that the sticky region locating on the center of the open region in phase space 
427: dose not contribute long time decay due to its easy escape.  The dependence of the decay rate
428: $\gamma$ on the refractive index $n$ shows a very good agreement with Eq.(\ref{gammafinal})
429: for large refractive indices, in both systems with different area $A$.
430:  This implies that even in the small opening limit, $n \rightarrow \infty$,
431: we cannot use the ergodic property over the whole phase space. Instead, we have to consider
432: structure of the SPD even in the small opening limit. 
433: 
434: 
435: 
436: 
437: \begin{figure}
438: \label{stte}
439: \begin{center}
440: \includegraphics[width=0.4\textwidth]{fig5.eps}
441: %\includegraphics[width=0.4\textwidth]{}
442: \caption{(color online) The SPTD and the decay rates for TE wave.
443: Detail of caption is the same as Fig. 4 except $D=\frac{\sqrt{A}L}{4\pi^2R}$ .
444: }
445: \end{center}
446: \end{figure}
447: 
448: \section{Quadrupole-deformed dielectric cavity - mixed system}
449: 
450: The escape property in generic mixed systems, showing a mixed phase space portrait:
451: integrable islands in a chaotic sea, has been extensively studied.
452: It is well known that the long time behavior of the SPTD is
453: algebraic due to the stickiness of the KAM tori surrounding islands
454: \cite{Fen94},
455: \begin{equation}
456: P_{sv} (t) \sim t^{-\eta}.
457: \end{equation}
458: However, there is no rigorous theory expecting the value of the exponent $\eta$
459: which has been estimated based on numerical calculations and seems to be nonuniversal.
460:  
461: In this section, we consider a quadrupolar dielectric cavity which is
462: the typical example of a deformed microcavity and shows a mixed dynamics. 
463: The boundary equation is, in the polar coordinates,
464: \begin{equation}
465: r(\phi)=1+\varepsilon \, \cos2\phi,
466: \end{equation}
467: where $\varepsilon$ is the deformation parameter. 
468: Here, we present numerical results of the SPTD and show that the long time behavior
469: of the SPTD is determined by whether islands locate in the closed region, $p_c < |p| < 1$,
470: or not.
471: 
472: 
473: \begin{figure}
474: \begin{center}
475: \includegraphics[height=0.35\textwidth,width=0.4\textwidth]{fig6.eps}
476: %\includegraphics[width=0.4\textwidth]{}
477: \caption{(color online) The SPTDs in the quadrupolar deformed dielectric cavity.
478: The SPTDs are algebraic in the  $\varepsilon=0.1$ cases while exponential in the  $\varepsilon=0.2$ cases.
479: The black and red lines denote the case of  TM and TE waves, respectively.}
480: \end{center}
481: \end{figure}
482: 
483: For $n=2$ case, we numerically calculate the SPTDs at two deformation parameter values, 
484: $\varepsilon=0.1$ and $0.2$, which are shown in Fig. 6.
485: In the $\varepsilon=0.1$ case, the SPTD shows an algebraic decay, i.e., 
486: $P_{sv} (t) \sim t^{-0.2}$, which is consistent with the previous studies on mixed systems.
487: However, in the $\varepsilon=0.2$ case, the long time behavior of the SPTD is exponential,
488: i.e., $P_{sv} (t) \sim \exp (-0.05t)$. 
489: This clear difference of the SPTD between $\varepsilon=0.1$ and $0.2$ cases can be
490: explained by the phase space portraits.
491: Figure 7 (a) shows the phase space portrait for $\varepsilon=0.1$ case.
492: There are many islands in the closed region, $p_c < p < 1$, so the stickiness of 
493: the KAM tori delays the ray escape and results in the algebraic tail. 
494: On the other hand, as shown in Fig. 7 (b) there is no island in the closed region
495: for $\varepsilon=0.2$ case and all islands exist in the open region.
496: The rays trapped by the stickiness of the KAM tori contribute to the short time 
497: escape behavior and the resulting SPTD shows exponential long time decay.
498: Therefore, the position of islands plays important role to understand the SPTD of mixed systems.  
499: 
500: 
501: \begin{figure}
502: \begin{center}
503: \includegraphics[width=0.4\textwidth]{fig7.eps}
504: %\includegraphics[width=0.4\textwidth]{}
505: \caption{ The phase space portraits for the quadrupolar deformed cavity.
506: (a) The $\varepsilon = 0.1$ case. 
507: (b) The $\varepsilon = 0.2$ case.
508: The dashed lines denote the critical line $p_c=1/2$ for total internal reflection
509: for $n=2$.}
510: \end{center}
511: \end{figure}
512: 
513: 
514: 
515: 
516: \section{Summary}
517: 
518: We have investigated the survival probability time distribution (SPTD) in three dielectric cavities
519: showing different ray dynamics; circle (integrable), stadium (chaotic), quadrupole (mixed) shapes.
520: In the circular dielectric cavity the SPTD has an algebraic long time behavior, $\sim t^{-2}$
521: in both TM and TE cases, but shows very different short time behavior due to the existence 
522: of the Brewster angle in TE case where exponential short time behavior is shown. 
523: The SPTD for a stadium-shaped cavity decays exponentially, and
524: the exponent has a close relation to the steady probability distribution (SPD). 
525: In the large $n$ limit, the SPD can be approximated by an assumption of a partial ergodicity, 
526: a uniform distribution over a specific part of phase space, which gives a correct description of
527: the exponent in both TM and TE cases. 
528: We have also discussed about the SPTD for the quadrupolar deformed cavity and
529: shown that the long time behavior can be algebraic or exponential, depending on the location of islands.
530: 
531: 
532: 
533: 
534: \section*{Acknowledgments}
535: 
536: This work is supported by Creative Research Initiatives of the Korean Ministry
537: of Science and Technology.
538: 
539: 
540: \section*{Appendix}
541: 
542: Here we present the analytical expression of the degree of openness (see Eq.(\ref{gammaT}))
543: for TM wave.
544: The degree of openness is defined by
545: \begin{eqnarray}
546: I \equiv \int_{-1/n}^{1/n} dp \,\,T(p),
547: \end{eqnarray}
548: where $T(p)=1-R(p)$, $R(p)$ is given in Eq.(\ref{Rtheta}). This integral can be
549: expressed by an analytical function as 
550: \begin{eqnarray}
551: I &=& {{4} \over {(n^2 - 1)^2}}
552: [B({{1} \over {2}},{{3} \over {2}})F(-{{3} \over {2}},{{1} \over {2}};2;{{1} \over {n^2}}) n^2 \nonumber\\
553: &&+B({{1} \over {2}},{{5} \over {2}})F(-{{1} \over {2}},{{1} \over {2}};3;{{1} \over {n^2}})  
554: -{{40} \over {15}}n +{{8} \over {15}}{{1} \over {n}}] \nonumber \\
555: &\simeq& 2 \pi n^{-2},
556: \end{eqnarray}
557: where $B(x,y)$ is the beta function and  $F(\alpha,\beta;\gamma;z)$ 
558: the Gauss hypergeometric function \cite{Gra00}.
559: 
560: For TE wave, only difference is the replacement of $R(p)$ by  $R_{TE}(p)$ 
561: of Eq.(\ref{RTE}), and the result is
562: \begin{equation}
563: I\simeq 4\pi n^{-2}
564: \end{equation}
565: for the large $n$ limit based on a numerical calculation.
566: 
567: 
568: 
569: \begin{thebibliography}{150}
570: 
571: \bibitem{Cha96} {\it Optical Processes in Microcavities},
572: 								edited by R. K. Chang and A. J. Campillo (World Scientific, Singapore, 1996). 
573: \bibitem{Noc97}	J. U. N\"{o}ckel and A. D. Stone, Nature {\bf 385}, 45 (1997);
574: 								C. Gmachl, F. Capasso, E. E. Narimanov, J. U. N\"{o}ckel, A. D. Stone,
575: 								J. Faist, D. L. Sivco, and A. Y. Cho, Science {\bf 280}, 1556 (1998);
576: 								G. D. Chern, H. E. Tureci, A. D. Stone, R. K. Chang, M. Kneissl, and N. M. Johnson,
577: 								Appl. Phys. Lett. {\bf 83}, 1710 (2003);
578: 								M. Kneissl, M. Teepe, N. Miyashita, N. M. Johnson, G. D. Chern, and R. K. Chang,
579: 								Appl. Phys. Lett. {\bf 84}, 2485 (2004);
580: 								T. Ben-Messaoud and J. Zyss, Appl. Phys. Lett. {\bf 86}, 241110 (2005);
581: 								M. S. Kurdoglyan, S. -Y. Lee, S. Rim, and C. -M. Kim, Opt. Lett. {\bf 29}, 2758 (2004).
582: \bibitem{Lee04} S. -Y. Lee, S. Rim, J. -W. Ryu, T. -Y. Kwon, M. Choi, and C. -M. Kim,
583: 								Phys. Rev. Lett. {\bf93}, 164102 (2004);
584: 								S, -Y. Lee, J. -W. Ryu, T. -Y. Kwon, S. Rim, and C. -M. Kim, arXiv:nlin.CD/0505040 (2005).
585: \bibitem{Har04} {\it Cavity-Enhanced Spectroscopies}, edited by R. D. van Zee and J. P. Looney
586: 								(Academic Press, San Diego, 2002);
587: 								H. G. L. Schwefel, N. B. Rex, H. E. Tureci, R. K. Chang, A. D. Stone,
588: 								T. Ben-Messaoud, and J. Zyss, J. Opt. Soc. Am. B {\bf 21}, 923 (2004).
589: \bibitem{Bau90} W. Bauer and G. F. Bertsch, Phys. Rev. Lett. \textbf{65}, 2213 (1990);
590: 								O. Legrand and D. Sornette, Phys. Rev. Lett. \textbf{66}, 2172 (1991);
591: 								W. Bauer and G. F. Bertsh, Phys. Rev. Lett. \textbf{66}, 2173 (1991);
592: 								O. Legrand and D. Sornette, Physica D \textbf{44}, 229 (1990).
593: \bibitem{Dor90}	E. Doron, U. Smilansky, and A. Frenkel, Phys. Rev. Lett. \textbf{65}, 3072 (1990).
594: \bibitem{Fri01}	N. Friedman, A. Kaplan, D. Carasso, and N. Davidson,
595: 								Phys. Rev. Lett. \textbf{86}, 1518 (2001).
596: \bibitem{Bun05}	L. A. Bunimovich and C. P. Dettmann, Phys. Rev. Lett. \textbf{94}, 100201 (2005).
597: \bibitem{Alt96} H. Alt, H. -D. Gr\"{a}f, H. L. Harney, R. Hofferbert, H. Rehfeld, A. Richter, and P. Schardt,
598: 								Phys. Rev. E \textbf{53}, 2217 (1996);
599: 								F. Vivaldi, G. Casati, and I. Guarneri, Phys. Rev. Lett. \textbf{51}, 727 (1983);
600: 								K. -C. Lee, Phys. Rev. Lett. \textbf{60}, 1991 (1988).
601: \bibitem{Sch02} J. Schneider, T. T\'{e}l, and Z. Neufeld, Phys. Rev. E \textbf{66}, 066218 (2002).								
602: \bibitem{Fen94} A. J. Fendrik,  A. M. F. Rivas, and M. J. S\'{a}nchez, Phys. Rev. E \textbf{50}, 1948 (1994);
603: 								A. J. Fendrik and M. J. S\'{a}nchez, Phys. Rev. E \textbf{51}, 2996 (1995);
604:  								C. F. F. Karney, Physica D \textbf{8}, 360 (1983);
605:  								B. V. Chirikov and D. L. Shepelyansky, Physica D \textbf{13}, 395 (1984);
606:  								P. Grassberger and H. Kantz, Phys. Lett. A \textbf{113}, 167 (1985);
607:  								Y. C. Lai, M. Ding, C. Grebogi, and R. Bl\"{u}mel, Phys. Rev. A \textbf{46}, 4661 (1992);
608:  								M. Weiss, L. Hufnagel, and R. Ketzmerick, Phys. Rev. E \textbf{67}, 046209 (2003).
609: \bibitem{Haw95} J. Hawkes and I. Latimer, \textit{Lasers; Theory and Practice}
610: 								(Prentice Hall, Englewood Cliffs, NJ, 1995).
611: \bibitem{Gra00} I.S. Gradshteyn, and I. M. Ryzbik, \textit{Table of Integrals, Series, and Products},
612: 									6th Edition (Academic Press, San Diego, 2000).
613: 
614: \end{thebibliography}
615: 
616: \end{document}
617: