nlin0008005/000.tex
1: \documentstyle[12pt,prb,aps,psfig]{revtex}
2: \begin{document}
3:  \draft
4:  \title{ Stability of the Ground State of a Harmonic Oscillator \\
5: in  a Monochromatic Wave}
6:  \author{Gennady P. Berman, Daniel F.V. James, and Dimitry I. Kamenev
7:  \footnote{On leave from Nizhny Novgorod State University, Nizhny Novgorod,
8:  603600, Russia}}
9:  \address{Theoretical Division and CNLS, Los Alamos National Laboratory,
10:  Los Alamos, NM 87545}
11: \maketitle
12: % \address{$^2$ Nizhny Novgorod State University, Nizhny Novgorod, 603600,
13: % Russia}
14: \vspace{20mm}
15: 
16: \begin{abstract}
17: 
18: Classical and quantum dynamics of a harmonic oscillator 
19: in a monochromatic wave is studied 
20: in the exact resonance and near resonance cases.  This 
21: model describes, in particular, a dynamics of a cold 
22: ion trapped in a linear ion trap and interacting with 
23: two lasers fields with close frequencies.
24: Analytically and numerically a stability of the
25: ``classical ground state'' (CGS) --  the vicinity
26: of the point ($x=0, p=0$) -- is analyzed. In the quantum 
27: case, the method for studying a stability of the quantum 
28: ground state (QGS) is suggested, based on the quasienergy representation. 
29: The dynamics depends on four parameters: the detuning from the resonance, 
30: $\delta=\ell-\Omega/\omega$, where
31: $\Omega$ and $\omega$ are, respectively, the wave and the 
32: oscillator's frequencies;
33: the positive integer (resonance) number, $\ell$;
34: the dimensionless Planck constant, $h$, 
35: and the dimensionless wave amplitude, $\epsilon$. For $\delta=0$,
36: the CGS and the QGS are unstable for resonance numbers $\ell=1,\,2$.
37: For small $\epsilon$, the QGS becomes more stable
38: with increasing $\delta$ and decreasing $h$.
39: When $\epsilon$ increases, the influence
40: of chaos on the stability of the QGS is analyzed for different parameters of
41: the model, $\ell$, $\delta$ and $h$.
42: \end{abstract}
43: 
44: \section{Introduction}
45: One of the major difficulties in developing quantum technologies,
46: such as quantum computers, \cite{fortsch,bdmt} are different kinds of
47: specifically quantum dynamical instabilities that can occur due
48: to interactions between different degrees of freedom and resonant 
49: interaction with the external fields. Instabilities in quantum systems 
50: have different nature than instabilities in classical systems, in which
51: such kind of phenomenon as dynamical
52: chaos occurs as a result of exponential divergence of initially close
53: trajectories. In quantum systems, the notion of a trajectory is
54: not well defined. This is one of the  main reasons why most of
55: well-developed methods for stability analysis can not be directly applied to
56: quantum systems.
57: 
58: In this paper the dynamics of a harmonic oscillator in a monochromatic
59: wave field is studied.
60: This system describes, in particular, a quantum dynamics of a cold ion
61: trapped in a linear ion trap and interacting with two laser fields with
62: close frequencies.\cite{BD}
63: Our attention is focused mainly on classical and quantum  behavior in the
64: region of parameters close to the quantum ground state (QGS) of the harmonic
65: oscillator. This case corresponds to classical dynamics in the vicinity of
66: the point ($x=0,\,p=0$) in the phase space. We shall call the corresponding
67: region of parameters for the classical system a ``classical ground state''
68: (CGS) of a harmonic oscillator.
69: The stability of the CGS and the QGS at stable and chaotic
70: classical regimes is analyzed for different parameters of the model.
71: In Sec. II, we consider classical dynamics in the vicinity of the
72: CGS in the exact resonance case, $\delta=0$.
73: In this situation the system we study is degenerate,\cite{Z} and an
74: infinitely small perturbation generates in the classical phase space an
75: infinite number of the ``resonance cells'' separated by the infinite
76: stochastic web. A classical dynamics inside the resonance cells at small
77: $\epsilon$ is described by using the resonance perturbation theory.\cite{L}
78: It is shown, that in the vicinity of the CGS (the ``central
79: cell''), and for small enough $\epsilon$, the dynamics in the case
80: $\delta=0$ is unstable for
81: resonance numbers $\ell=1,2$ and stable for $\ell\ge 3$.
82: 
83: We show that in spite of the  ``resonant Hamiltonian'' describes many
84: features of the dynamics inside the resonance cells, it can not be used
85: for describing the stability of the system in the central cell, at
86: $\ell\ge 3$. The classical dynamics in this region is determined,
87: by the Mathieu equation. It is shown that at small enough value of
88: $\epsilon$ the area of the central cell
89: increases with increasing the resonance number, $\ell$, and increasing
90: the perturbation amplitude, $\epsilon$.
91: 
92: The dynamics of the system near the CGS
93: in the near resonance case, when $\delta\ne 0$,
94: is considered in Sec. III. It is shown that in this case
95: at small $\epsilon$ the classical
96: dynamics near the CGS is stable at any value of
97: the resonance number $\ell$. The cases $\ell=1$ and $\ell=2$ are
98: considered in detail. It is shown that the CGS in these two 
99: cases becomes unstable at much less values of $\epsilon$,
100: than for large $\ell$ ($\ell=4,5$...).  
101: 
102: A stability of the quantum system is considered in Sec. 4. In this case, an 
103: additional parameter, a dimensionless Planck constant, $h$, influences
104: significantly the behavior of the system.
105: Because the Hamiltonian is time-periodic, with the period $2\pi/\Omega$, 
106: we use the Floquet theory to study localization properties of quantum 
107: system in the region of the QGS. For $\epsilon\ll 1$, and small enough 
108: $h$, the quasienergy (QE) states can be conventionally divided into the
109: following groups. The first group includes the QE states which belong to
110: some particular ``resonance cells'' in the Hilbert space.
111: These QE states are well-localized inside a given resonance cell. The second
112: group includes the delocalized ``separatrix'' QE states which correspond to
113: the classical stochastic web.
114: Usually, these ``separtrix'' QE states are responsible for tunneling effects
115: between different ``resonance cells'',  even for small $\epsilon$.\cite{1}
116: 
117: We show that the QGS is stable when the following conditions are
118: satisfied: (a) an existence of the QE state mainly localized in 
119: the QGS of the harmonic oscillator, 
120: (b) when $\epsilon\ll 1$ and chaos is weak, (c) small enough $h$, when 
121: one can neglect the tunneling phenomenon. 
122: When $h$ is larger than the size of the 
123: central cell, no QE state localized in the QGS of the
124: harmonic oscillator was found. 
125: We show, that at small enough values of $\epsilon$,
126: the stability of the QGS can be improved by choosing the non-resonant
127: frequency of the wave, so that $\delta=\ell-\Omega/\omega\ne 0$.
128: 
129: For small enough $\epsilon$, a stability of the QGS is mainly determined
130: by the structure of the QE state mostly localized at the QGS 
131: of the harmonic oscillator (QGS QE state). In the exact resonance case 
132: this  particular QE state has a complicated structure.  
133: On the one hand, it is mostly localized on the QGS,
134: but, on the other hand, it has the  ``separatrix'' structure\cite{3} 
135: and provides a tunneling from the region of the QGS to other 
136: resonance cells. In order to improve stability of the QGS, the separatrix 
137: structure should be destroyed by choosing $\delta\ne 0$.  
138: 
139: In the exact and near resonance cases
140: we study numerically the probability, $P_0$, for a particle to remain 
141: in the QGS, depending on $\epsilon$ and for different values of $h$, 
142: $\ell$ and $\delta$. 
143: When $\delta=0$ and $\epsilon$ is small,
144: the dynamics depends on the value of the 
145: quantum parameter, $h$, which controls degree of delocalization
146: of the QGS QE state. When $h$ is small,
147: the probability of tunneling to other cells is small too, and
148: $P_0$ decreases with $\epsilon$ increasing, because chaos makes the 
149: QGS QE state more delocalized.
150: For large enough values of $h$, 
151: the dependence of $P_0$ on $\epsilon$ 
152: becomes more complicated because in this case we have 
153: two superimposing effects: 
154: the influence of chaos on the dynamics, and
155: the effect of tunneling to the classically unacceptable cells. 
156: These two features make the dependence $P_0=P_0(\epsilon)$ non-monotonic. 
157: In the region of small $\epsilon$, $P_0$ increases 
158: with $\epsilon$ increasing.  This behavior of $P_0$ can be explained 
159: as a result of 
160: destroying the delocalized QE states with increasing $\epsilon$. Further 
161: increase of $\epsilon$ leads to decreasing $P_0$. This behavior of 
162: $P_0$ is connected with delocalization of QE states due to increasing
163: of the chaotic component. At large enough values of $\epsilon$, the
164: dependence of $P_0$ on $\epsilon$ becomes more complicated, and generally
165: must be considered using many QE states. It is shown that the value of
166: $\delta$ influence significantly the stability of the QGS only at small
167: $\epsilon\le 1$.
168: The stability of the QGS at $\ell=1,\,2$
169: and $\delta\ne 0$ is explored. It is shown that the QGS becomes unstable
170: at much less values of the wave amplitude than
171: in the case when $\ell$ is large. The derived in this paper results
172: are important for understanding a stability of quantum dynamics in the
173: vicinity of the ground state.
174: 
175: \section{Classical dynamics near the CGS in the case of exact resonance}
176: The Hamiltonian of the harmonic oscillator in a monochromatic
177: wave is,
178: \begin{equation}
179: \label{cl_H}
180: H=\frac{p^2}{2M}+\frac{M\omega^2x^2}2
181: +v_0\cos(kx-\Omega t)=H_0+V(x,t),
182: \end{equation}
183: where $M$ is the mass of the
184: particle, $p$ is the momentum, $k$ is the wave vector,
185: $v_0$ is the amplitude of the perturbation, and
186: $H_0$ is the Hamiltonian of the harmonic oscillator.
187: In the first part of this section we discuss the case of the resonance,
188: when $\Omega=\ell\omega$. It is known
189: (see for example Ref.~\cite{Z}) that under the resonance condition 
190: the infinitely small perturbation, $v_0$, is enough to generate 
191: in the classical phase space the infinite stochastic web. 
192: The web is inhomogeneous, and its width decays with decreasing 
193: the perturbation amplitude, $v_0$, and increasing $|p|$ and $|x|$.             
194: Inside the cells of the web a particle moves along stable 
195: closed trajectories.
196: 
197: To analyze the 
198: dynamics of the harmonic oscillator in a monochromatic wave,
199: described by the Hamiltonian (\ref{cl_H}), 
200: it is convenient to use
201: the resonance perturbation theory \cite{L}
202: discussed below.   
203: Let us perform a transformation  from the variables ($p,x$)
204: to the canonically
205: conjugated variables ($\bar J_\varphi,\,\varphi$),
206: \begin{equation}
207: \label{xtheta}
208: x=(2\bar J_\varphi/M\omega)^{1/2}\sin\varphi=
209: r(\bar J_\varphi)\sin\varphi,
210: \end{equation}
211: \begin{equation}
212: \label{ptheta}
213: p_x=(2\bar J_\varphi M\omega)^{1/2}\cos\varphi=
214: M\omega r(\bar J_\varphi)\cos\varphi,
215: \end{equation}
216: where $r(\bar J_\varphi)=(2\bar J_\varphi/M\omega)^{1/2}$
217: is the amplitude of oscillations.
218: It is more convenient to work with
219: the dimensionless coordinate,
220: $X=kx$, and the dimensionless momentum, $P=kp/M\omega$, which are related to
221:  the variables ($\bar J_\varphi,\,\varphi$)
222: by the formulas,
223: \begin{equation}
224: \label{Xtheta}
225: X=\rho(\bar J_\varphi)\sin\varphi,
226: \end{equation}
227: \begin{equation}
228: \label{Ptheta}
229: P=\rho(\bar J_\varphi)\cos\varphi,
230: \end{equation}
231: where $\rho(J_\varphi)=\sqrt{X^2+P^2}=kr(J_\varphi)$. 
232: In order to treat the time on the same ground as the phase 
233: $\varphi$ let us introduce the new pair of canonically conjugated 
234: variables, $(\bar J_\beta,\,\beta)$, where $\beta=\Omega t$.  
235: The initial Hamiltonian (\ref{cl_H}) expressed through the new variables
236: takes the form,  
237: \begin{equation}
238: \label{cl_H0}
239: H=\bar J_\varphi\omega+\bar J_\beta\Omega+
240: v_0\cos\left(\rho\sin\varphi-\beta\right).
241: \end{equation}
242: It is independent of time, but describes the motion in 
243: the two-dimensional space.  
244: The nonlinear perturbation in Eq. (\ref{cl_H0})
245: can be expressed in a series,
246: \begin{equation}
247: \label{wave_decomposition}
248: v_0\cos\left(\rho\sin\varphi-\beta\right)=v_0\sum_{m=-\infty}^\infty
249: J_m(\rho)\cos\left(m\varphi-\beta\right),
250: \end{equation}
251: where $J_m(\rho)$ is the Bessel function.  
252: Under the resonance condition, 
253: $\Omega=\ell\omega$, all terms
254: in the sum (\ref{wave_decomposition}) quickly oscillate
255: and can be averaged out, except for one term with $m=\ell$.
256: In this approximation, the
257: Hamiltonian (\ref{cl_H0}) is reduced to,
258: \begin{equation}
259: \label{cl_H1}
260: H=\bar J_\varphi\omega+\bar J_\beta\Omega+v_0J_\ell(\rho)
261: \cos\left(\ell\varphi-\beta\right).
262: \end{equation}
263: It is convenient to introduce new, resonance, variables,
264: ($\tilde I,\,\theta$), ($\tilde J,\,\tilde\beta$), 
265: by using the generating function,
266: $$
267: F=\tilde I\left(\ell\varphi-\beta\right)+\tilde J\beta,
268: $$
269:  
270: The new Hamiltonian,
271: \begin{equation}
272: \label{cl_H3}
273: H=\tilde I(\ell\omega-\Omega)+\tilde J\omega+v_0J_\ell(\rho)\cos\theta,
274: \end{equation}
275: where $\theta=\ell\varphi-\beta$,
276: is independent of the variable $\tilde\beta$. Hence,
277: $\tilde J=const$. The resonance Hamiltonian,
278: \begin{equation}
279: \label{cl_HR}
280: H_\ell(\rho,\theta)=H-\tilde J\omega=v_0J_\ell(\rho)\cos\theta,
281: \end{equation}
282: (where we used the resonance condition $\ell\omega=\Omega$)
283: is independent of time, unlike the initial Hamiltonian (\ref{cl_H}).
284: 
285: The Poincar\`e surfaces of section of the system
286: described by the 
287: Hamiltonian (\ref{cl_H}) in variables ($X,\,P$) 
288: are shown in Figs. 1 (a) - 1 (e), for the cases 
289: $\ell=1,2,3,4,5$. The phase 
290: 
291:  \begin{figure}[tb]
292:  \begin{center}
293: %\setcounter{figure}{5}
294:  \mbox{\psfig{file=px1.ps,width=7cm,height=7cm}
295:        \psfig{file=px2.ps,width=7cm,height=7cm}}
296:  \mbox{\psfig{file=px3.ps,width=7cm,height=7cm}
297:        \psfig{file=px4.ps,width=7cm,height=7cm}}
298:  \psfig{file=px5.ps,width=7cm,height=7cm}
299:  \vspace{0.5cm}
300:  \caption{The resonance cells in the phase space for 
301:  $\epsilon=0.05$, $\delta=0$ and: (a) $\ell=1$, (b) $\ell=2$,
302:  (c) $\ell=3$, (d) $\ell=4$, (e) $\ell=5$.}
303:  \end{center}
304:  \label{fig:1}
305:  \end{figure}
306: 
307: \noindent
308: points are plotted at the moments
309: $t_j=2\pi j/\Omega$, where $j=0,\,1,\,2,\dots$.  
310: One can see that the phase space  
311: has an axial symmetry of the order $\ell$. The phase 
312: space is divided into the cells. 
313: A particle moves along closed trajectories inside the cells. 
314: (In Figs. 1 (a) - 1 (e) only the boundaries of the cells are demonstrated.) 
315: At small values of $v_0$, the motion inside the resonant cells, 
316: illustrated in Figs. 1 (a) - 1 (e), 
317: can be considered in the resonance approximation. 
318: The next order approximation 
319: is required only for analysis of the motion inside the exponentially 
320: small chaotic regions near the separatrices. It is shown below that 
321: the resonance approximation also fails to describe the motion 
322: in the region near the point $(X=0,\,P=0)$.   
323: 
324: The symmetry of the phase space in Figs. 1 (a) - 1 (e)
325: follows from the form of the resonance
326: Hamiltonian (\ref{cl_HR}), which is invariant under the transformation,
327: \begin{equation}
328: \label{transform1}
329: \varphi\rightarrow\varphi + 2\pi/\ell. 
330: \end{equation}
331: In the phase space there are $\ell$ cells connected
332: by the transformation (\ref{transform1}), and $\ell$ cells  
333: connected by the same transformation but described by the 
334: Hamiltonian $H_\ell(\rho,\theta+\pi)$ in 
335: Eq. (\ref{cl_HR}). The total number of cells at given $\rho$
336: is  $2\ell$.  
337: Thus, in Fig. 1 (a)
338: the upper cell corresponds to the Hamiltonian $H_1(\rho,\theta)$, while
339: the lower cell corresponds to the Hamiltonian $H_1(\rho,\theta+\pi)$ in 
340: Eq. (\ref{cl_HR}); the upper and lower cells in Fig. 1 (b) 
341: correspond to $H_2(\rho,\theta)$, and the right and the left cells 
342: correspond to $H_2(\rho,\theta+\pi)$, and so on. 
343: 
344: It is easy to see, the resonance Hamiltonian    
345: (\ref{cl_HR}) yields unstable solution near the CGS
346: (the point ($X=0,\,P=0$)). 
347: To show this, we present the Hamiltonian (\ref{cl_HR}) in the form,
348: \begin{equation}
349: \label{cl_HR1}
350: H_\ell=v_0\frac{\rho^\ell}{2^\ell\ell !}\cos\ell\varphi=E_\ell,
351: \end{equation}
352: where $E_\ell=const$ (because the resonance Hamiltonian is independent 
353: of time). Also, we took into account that at $\rho\ll 1$ the Bessel 
354: function can be expressed in the form, 
355: \begin{equation}
356: \label{Bessel1}
357: J_\ell(\rho)\approx \rho^\ell/2^\ell\ell !,   
358: \end{equation}
359: and the fact that in the Poincar\`e surfaces of section the position
360: of the particle is taken at the moments 
361: $\Omega t_k=2\pi k$, $k=0,\,1,\,2,\dots$. 
362: It follows from Eq. (\ref{cl_HR1}) that at the 
363: angles $\varphi_k=\frac\pi{2\ell}(2k-1)$, $k=1,\,2,\dots,\,2\ell$,
364: the radius $\rho$ should sharply increase or decrease. 
365: It is seen from Figs. 1 (a) - 1 (e) that
366: at angles $\varphi_k$ the particle moves in the radial 
367: direction. The grows of $\rho$ is limited due to nonlinear effects. 
368: 
369: However, the resonance perturbation theory does not adequately describe
370: the motion of a particle in the vicinity of the point ($X=0,\,P=0$)
371: at large values of the resonance number $\ell$, because the amplitude 
372: of the resonance term due to Eq. (\ref{Bessel1}) quickly decays when 
373: the radius, $\rho$, decreases. Indeed, at $\ell=4$ and $\rho=0.1$ the 
374: amplitude 
375: of the resonance term with $m=\ell=4$ in Eq. (\ref{wave_decomposition}) 
376: is $80$ times less than the amplitude of the non-resonant term with
377: $m=\ell-1=3$. In order to describe the motion in the region near
378: the CGS,   we 
379: consider the initial Hamiltonian (\ref{cl_H}) under the condition $X\ll 1$. 
380: The exact classical equation of motion reads,
381: \begin{equation}
382: \label{e_motion1}
383: \frac{d^2}{dt^2}X+\omega^2 X=\frac{v_0k^2}{M}\sin(X-\Omega t).   
384: \end{equation}
385: 
386:  \begin{figure}[tb]
387:  \begin{center}
388: %\setcounter{figure}{5}
389:  \mbox{\psfig{file=px3a.ps,width=5cm,height=5cm}
390:        \psfig{file=px4a.ps,width=5cm,height=5cm}
391:        \psfig{file=px5a.ps,width=5cm,height=5cm}}
392:  \vspace{0.5cm}
393:  \caption{The trajectories in the central resonance
394:  cell in the phase space for $\delta=0$ and (a) $\ell=3$,
395:  $\epsilon=5\times 10^{-4}$, (b) $\ell=4$, $\epsilon=0.05$,
396:  (c) $\ell=5$, $\epsilon=0.05$.}
397:  \end{center}
398:  \label{fig:2}
399:  \end{figure}
400: 
401: \noindent
402: Up to the first order in $X$ Eq. (\ref{e_motion1}) is,
403: \begin{equation}
404: \label{e_motion2}
405: \frac{d^2}{dt^2}X+\omega^2(1-\epsilon\cos(\Omega t))X=
406: \frac{v_0k^2}{M}\sin(\Omega t),
407: \end{equation}
408: where $\epsilon=v_0k^2/M\omega^2$ is the dimensionless
409: perturbation amplitude. 
410: If we introduce a new dimensionless time, $2\tau=\Omega t$,
411: then from Eq. (\ref{e_motion2}) we obtain the Mathieu equation 
412: with the additional right-hand side term  
413: in the form, 
414: \begin{equation}
415: \label{Mathieu}
416: \frac{d^2}{d\tau^2}X+
417: a_\ell(1-\epsilon\cos(2\tau))X=a_\ell\epsilon\sin(2\tau),
418: \end{equation}
419: where $a_\ell=\left(2/\ell\right)^2$. 
420: From the theory of Mathieu 
421: functions \cite{Mathieu} it is known 
422: that for small $\epsilon$,
423: Eq.~(\ref{Mathieu}) has unstable general solutions  
424: at $a_\ell=1$ and $a_\ell=4$ which correspond to the resonance numbers, 
425: $\ell=2$ and $\ell=1$. The additional term in the 
426: right-hand side of Eq. (\ref{Mathieu}) does not influence
427: the stability of trajectories (see Ref. \cite{Mathieu}, \S 6.22).
428: At $a_\ell<1$ and small enough values
429: of $\epsilon$, the Mathieu equation has periodic solutions which
430: correspond to the stable dynamics for resonance numbers $\ell>2$.
431: In Fig. 2 stable trajectories in the system described by the
432: Hamiltonian (\ref{cl_H}) are shown for $\ell=3,\,4,\,5$.
433: Stable region in the vicinity of the CGS can be
434: considered as additional ``central cells'' to those resonance cells shown
435: in Figs. 1~(c) - 1~(e). The motion in the central cell is characterized
436: by the following features: i) the size of the cell increases
437: with increasing the resonance number, $\ell$;
438: ii) trajectories in the cell have the
439: axial symmetry of the order $\ell$;  iii) as follows from  
440: numerical calculations, the period of the motion in the central 
441: cell along all trajectories in Figs. 2 (a) - 2 (c) is 
442:  \vspace{0.3cm}
443: 
444:  \begin{figure}[tb]
445:  \begin{center}
446: %\setcounter{figure}{5}
447:  \mbox{\psfig{file=par5a.ps,width=8cm,height=8cm}
448:        \psfig{file=par5d.ps,width=8cm,height=8cm}}
449: % \vspace{0.2cm}
450:  \end{center}
451: % \caption{ 
452: %}
453:  \label{fig:3}
454:  \end{figure}
455: 
456:  FIG.3. The trajectories obtained by solution of Eq.
457:  (\ref{e_motion1}) when only the terms up to (a) the
458:  first order (Eq. (\ref{Mathieu})) and (b) the fourth
459:  order in $X$ are taken into 
460:  account; $\ell=5$, $\delta=0$,
461:  $\epsilon=0.05$.
462: 
463: \noindent
464: approximately the same (in each figure), and very large.
465: For example, the period of motion along the trajectories in Fig. 2 (b)
466: for $\ell=4$ is $(1-1.2)\times 10^5T$, where $T=2\pi/\Omega$,
467: while the period of motion along the
468: trajectories in Fig. 1 (d) is of the order $\sim 5\times 10^3T$.
469: The property i) can be described by the fact
470: that the resonance term, which leads to unstable solution near the
471: point ($X=0,\,P=0$), for
472: larger values of $\ell$ has less influence on the dynamics
473: because its amplitude at $\rho\ll 1$ quickly decreases with
474: $\ell$ increasing, due to Eq. (\ref{Bessel1}).
475: 
476: 
477: In order to treat the properties ii), iii), we have considered
478: the influence of the terms of higher order in $X$ in Eq. (\ref{e_motion1})
479: on the dynamics described by Mathieu equation (\ref{Mathieu}). The results
480: for the case $\ell=5$ are shown in Figs. 3 (a), 3 (b).
481: From Figs. 3 (a), 3 (b) it is seen that
482: the terms of high order in $X$
483: change the shape of trajectories, and make
484: them symmetrical with the axial symmetry of the order $\ell$.    
485: This fact follows also from comparison of different trajectories
486: in Fig. 2 (c). Indeed, internal trajectories with small values
487: of $\rho$ have the round form, unlike external trajectories with 
488: larger values of $\rho$, which have the form of the pentagon.  
489: From comparison of Figs. 2 (c) and 3 (b) it is clear that the 
490: terms of high order in $X$ (the terms of the order $X^i$ where
491: $i>\ell-1$) in Eq.(\ref{e_motion1}) do not influence significantly
492: the dynamics,
493: because the form of trajectories in Fig. 3 (b), computed for
494: the approximate model, is practically the same
495: as the form of trajectories
496: in the exact model shown  in Fig. 2 (c).
497: 
498: It was established numerically that slowness of the motion
499: in the vicinity of
500: the CGS is also result of influence of the terms of high order in $X$. Thus,
501: the period of motion along the external trajectory described by the
502: approximate equation (\ref{Mathieu}) in Fig. 3 (a) is $\sim 4\times 10^4T$.
503: Including into consideration the second order term in $X$ increases the
504: period of motion up to $\sim 3.3\times 10^5T$. If we include,
505: as well, the third order term, the period becomes $\sim 3.6\times 10^5T$.
506: Finally, including the 
507: resonance term leads to increase of the period up to    
508: the value $\sim 3.8\times 10^5T$ shown in Figs. 2 (c) and 3 (b).
509: 
510: Next, we shall analyze the dynamics in the 
511: central cell depending on $\epsilon$. Modification of trajectories 
512: at increasing the wave amplitude, $\epsilon$, for the resonance
513: number $\ell=4$ is shown in Figs. 4 (a) - 4 (d). 
514: Two features in the structure of the trajectories of the central cell 
515: can be observed. i) Increase in $\epsilon$ shifts the central cell up. 
516: ii) The size of the cell increases considerably in 
517: Figs. 4 (a) - 4 (c) with increasing $\epsilon$ in comparison with 
518: the case shown in Fig. 2 (b), when the value of $\epsilon$ is very small.
519: The cell shrinks when $\epsilon$ increases, which is shown in Fig. 4 (d)
520: for $\epsilon=10$.
521: Further increase of $\epsilon$ destroys the central cell entirely.
522: 
523:  \vspace{0.5cm}
524:  \begin{figure}[tb]
525: % \begin{center}
526: \setcounter{figure}{3}
527: \centerline{\mbox{\psfig{file=pcao4a.ps,width=8cm,height=8cm}
528:        \psfig{file=pcao4b.ps,width=8cm,height=8cm}}}
529:  \vspace{0.4cm}
530: \centerline{\mbox{\psfig{file=pcao4c.ps,width=8cm,height=8cm}
531:        \psfig{file=pcao4d.ps,width=8cm,height=8cm}}}
532:  \vspace{0.4cm}
533:  \caption{Trajectories in the central resonance
534:  cell at $\ell=4$, $\delta=0$ under influence of the perturbation
535: with the  amplitude, $\epsilon$: a) $\epsilon=0.5$, b) $\epsilon=2$,
536:  c) $\epsilon=5$, d) $\epsilon=10$.}
537: % \end{center}
538:  \label{fig:4}
539:  \end{figure}
540: 
541: Similar features were observed in dependence of the dynamics
542: in the central cell
543: on $\epsilon$ for the case $\ell=5$, shown in Figs. 5 (a) - 5 (d). 
544: Comparison of the data for $\ell=5$ in Figs. 5 (a) - 5 (d)
545: with those for $\ell=4$ in Figs. 4 (a) - 4 (d) allows us to 
546: conclude that the area of the 
547: central cell increases with increasing $\ell$, and chaotization 
548: of the motion in the central cell at 
549: 
550:  \vspace{0.9cm}
551:  \begin{figure}[tb]
552:  \begin{center}
553: %\setcounter{figure}{5}
554:  \mbox{\psfig{file=pcao5a.ps,width=8cm,height=8cm}
555:        \psfig{file=pcao5b.ps,width=8cm,height=8cm}}
556:  \vspace{0.7cm}
557:  \mbox{\psfig{file=pcao5c.ps,width=8cm,height=8cm}
558:        \psfig{file=pcao5d.ps,width=8cm,height=8cm}}
559: % \vspace{0.2cm}
560:  \caption{Trajectories in the central resonance
561:  cell at $\ell=5$, $\delta=0$ under influence of the perturbation
562:  with the amplitude, $\epsilon$: a) $\epsilon=5$, b) $\epsilon=10$,
563:  c) $\epsilon=15$, d) $\epsilon=19$.}
564:  \end{center}
565:  \label{fig:5}
566:  \end{figure}
567: 
568: \noindent
569: larger values of $\ell$ requires
570: larger values of $\epsilon$. In other words, the motion in the 
571: central cell becomes more stable with increasing the resonance 
572: number, $\ell$. 
573: In order to understand the observed dynamics, we 
574: again included into consideration only the 
575: first order terms in $X$ in Eq. (\ref{e_motion1}),
576: and computed the dynamics for large values of $\epsilon$. 
577: The trajectories described by the approximate equation (\ref{Mathieu})
578: for $\epsilon=5$, $\ell=4$ and for $\epsilon=9.5$, $\ell=5$ are
579: shown in Figs. 6 (a), 6 (b). The following features can be observed.
580: i) As follows from our calculations, 
581: % \vspace{0.2cm}
582: 
583:  \begin{figure}[tb]
584:  \begin{center}
585: %\setcounter{figure}{5}
586:  \mbox{\psfig{file=par4.ps,width=8cm,height=8cm}
587:        \psfig{file=par5.ps,width=8cm,height=8cm}}
588: % \caption{
589: %}
590:  \end{center}
591:  \label{fig:6}
592:  \end{figure}
593: 
594:  FIG. 6. The phase trajectories described by the approximate 
595:  equation (\ref{Mathieu}) for $\delta=0$ and (a) $\ell=4$, $\epsilon=5$,
596:  (b) $\ell=5$, $\epsilon=9.5$.
597: \vspace{0.5cm}
598: 
599: \noindent
600: the phase portrait shifts up 
601: from the point $(X=0,\,P=0)$
602: under the
influence of the term in the right-hand side of the approximate
603: equation (\ref{Mathieu}). ii) Comparison of Fig. 6 (a) with Fig. 4 (c)
604: and Fig. 6 (b) with Fig. 5~(b) allows us to conclude that the terms 
605: of the high order in $X$ in Eq. (\ref{e_motion1}) change the 
606: shape of trajectories and restrict the region of stable motion. 
607: iii) The motion described by the approximate equation (\ref{Mathieu})
608: becomes unstable 
609: at values of $\epsilon>\epsilon_\ell$ where $\epsilon_\ell$ lies
610: in the interval $2.3<\epsilon_3<2.4$ for $\ell=3$;
611: $5.5<\epsilon_4<5.6$ for 
612: $\ell=4$; and $9.6<\epsilon_5<9.7$ for $\ell=5$. One can see from Fig. 4 (d) 
613: and Figs. 5 (c), 5 (d) that the motion described by 
614: exact equation (\ref{e_motion1}) 
615: in the region $\epsilon>\epsilon_\ell$ remains stable.
616: Thus, the high order terms in $X$ in Eq. (\ref{e_motion1})    
617: stabilize the dynamics at large values of the perturbation amplitude, 
618: $\epsilon$.    
619: 
620:  \begin{figure}[tb]
621:  \begin{center}
622: \setcounter{figure}{6}
623:  \mbox{\psfig{file=px4_1.ps,width=8cm,height=8cm}
624:        \psfig{file=px4_5.ps,width=8cm,height=8cm}}
625:  \mbox{\psfig{file=px5_1.ps,width=8cm,height=8cm}
626:        \psfig{file=px5_5.ps,width=8cm,height=8cm}}
627:  \vspace{0.2cm}
628:  \caption{Influence of chaos on different resonant cells, $\delta=0$,
629:  (a) $\ell=4$, $\epsilon=1$; (b) $\ell=4$, $\epsilon=5$;
630:  (c) $\ell=5$, $\epsilon=1$; (d) $\ell=5$, $\epsilon=5$.}
631:  \end{center}
632:  \label{fig:7}
633:  \end{figure}
634: 
635: % \vspace{0.2cm}
636: Let us compare the classical dynamics in the central 
637: cell with the dynamics in the other cells
638: when the perturbation parameter $\epsilon$ is not small. The results of 
639: calculation of the dynamics in several cells are shown in
640: Figs. 7 (a) - 7 (d).
641: From comparison Fig. 7 (b) with Figs. 4 (c), 4 (d) and Fig. 7 (d) with
642: Figs. 5 (a) - 5 (d) one can see that the trajectories in the central
643: cell remain stable, while other nearest cells are completely destroyed
644: by chaos. The extremely high stability of trajectories in the central
645: cell can be explained by relatively small influence on the
646: dynamics of the 
647: terms of high order in $X$, oscillating with different frequencies, because 
648: their amplitudes are small at small $X$ (or $\rho$).      
649: 
650: 
651: 
\section{The classical dynamics near the CGS in the near resonance case}
652: Now, let us consider the CGS in the near resonance case, when $\delta\ne 0$. 
653: The resonant Hamiltonian (\ref{cl_H3}) takes form 
654: \begin{equation}
655: \label{dw_H}
656: H_\ell=\tilde I(\delta\omega)+v_0J_\ell(\rho)\cos\theta.
657: \end{equation}
658: The stationary points for the dynamics generated by the 
659: Hamiltonian (\ref{dw_H}) 
660: are defined by the conditions,
661: $$
662: \dot\theta=\partial H_\ell/\partial\tilde I=0,\qquad
663: \dot{\tilde I}=-\partial H_\ell/\partial\theta=0.
664: $$
665: Positions of the elliptic stationary points are given by the expressions,
666: \begin{equation}
667: \label{elliptic1}
668: v_0{\partial J_\ell[kr(\tilde I_e)]\over\partial\tilde I}=\mp \delta\omega,
669: \qquad \theta_e=0,\pi,
670: \end{equation}
671: where the sing $``-''$ corresponds to the stable point, with the angle 
672: $\theta_e=0$ with the $P$-axis, and the sing $``+''$ corresponds 
673: to the stable point with the angle $\theta_e=\pi$. 
674: In the dimensionless form Eq. (\ref{elliptic1}) is, 
675: \begin{equation}
676: \label{elliptic}
677: \frac 1{\rho_e}{\partial J_\ell(\rho_e)\over\partial\rho}=
678: \mp{\delta\over\ell\epsilon},
679: \end{equation}
680: where $\rho_e=kr(\tilde I_e)$. 
681: For the positions of the hyperbolic stationary points one has,
682: \begin{equation}
683: \label{hyperbolic}
684: J_\ell\left[ kr(\tilde I_h)\right]=0, \qquad \theta_h=\pm\frac\pi 2.
685: \end{equation}
686: As one can see from Eq. (\ref{elliptic}), the number of the elliptic stable
687: points in the near resonance case, when $\delta\ne 0$,
688: is finite because the right-hand side of Eq. (\ref{elliptic}) is constant
689: while the left-hand side oscillates, and decreases on average.
690: As a consequence, there is a finite number of the resonance cells. 
691: 
692: The motion near the CGS can be described by the approximate equation 
693: (\ref{Mathieu}) with the parameter $a_\ell$ equal to: 
694: $a_\ell=[2/(\ell-\delta)]^2$. It is known \cite{Mathieu} that
695: at small $\epsilon$ and $\delta\ne 0$ the Mathieu equations have the stable 
696: solutions at any $\ell$  including 
697: the cases $\ell=1$ and $\ell=2$. 
698: 
699: Let us consider the cases $\ell=1$ and $\ell=2$ at $\delta\ne 0$ in 
700: detail. In the case $\ell=1$, one may use the results of the resonance 
701: theory at arbitrary small $X$ and $P$, because the term of the lowest order 
702: in $X$ (proportional to $X$) in the Hamiltonian (\ref{cl_H}) is resonant.
703: Let us suppose that the dimensionless radius $\rho_e$ in Eq. (\ref{elliptic}) 
704: is small, $\rho_e\ll 1$. Then $J_1'(\rho_e)\approx 1/2$, and  
705: Eq. (\ref{elliptic}) yields, 
706: \begin{equation}
707: \label{roe}
708: \rho_e=\mp \epsilon/(2\delta\ell).
709: \end{equation}
710: Thus, shift of the stable elliptic point from the CGS is small,
711: $\rho_e\ll 1$, when the condition $\epsilon\ll 2|\delta|\ell$ is satisfied.
712: At small values of the wave amplitude,
713: $\epsilon$, shift of the elliptic stable point from the point $X=0,\,P=0$ is
714: proportional to $\epsilon$. One can see from Eq. (\ref{roe}) that at
715: $\ell=1$ one elliptic stable point exists at arbitrary small value of
716: $\epsilon$ (which follows also from the theory of Mathieu functions).
717: 
718:  \begin{figure}[tb]
719:  \begin{center}
720: %\setcounter{figure}{5}
721:  \mbox{\psfig{file=e04d01.ps,width=5.4cm,height=5.4cm}
722:        \psfig{file=e07d01.ps,width=5.4cm,height=5.4cm}
723:        \psfig{file=e12d01.ps,width=5.4cm,height=5.4cm}}
724: % \vspace{0.2cm}
725:  \caption{The phase space for the near resonance case, $\delta=0.1$,
726:  $\ell=1$, and (a) $\epsilon=0.4$, (b) $\epsilon=0.7$, (c) $\epsilon=1.2$}
727:  \end{center}
728:  \label{fig:8}
729:  \end{figure}
730: 
731: When $\epsilon$ is small, the phase trajectories 
732: are the circles with the center located near the CGS.
733: Figuratively speaking, in the case $\delta\ne 0$  and  $\epsilon\ll 1$
734: there is only one resonant (central) cell with an infinite 
735: area, because at small enough value of $\epsilon$ the
736: equation (\ref{elliptic}) has no other solutions, except for 
737: Eq. (\ref{roe}), and in the phase space there are 
738: no other cells, except for the central one. 
739: When $\epsilon$ increases (we suppose $\epsilon>0$)
740: and $\delta>0$, the stable point shifts down, as shown in Fig. 8 (a),
741: because the left-hand side of Eq. (\ref{elliptic}) is positive
742: and we should take the sign ``$+$'' in the right-hand side,
743: which corresponds to the shift in the direction $\theta=\pi$.  
744: 
745: When $\epsilon$ increases (see Fig. 8 (b), 8 (c)), the dynamical
746: chaos appears, and the
747: area of the central cell decreases. As before, we have considered
748: the influence of the  high order terms in $X$ in the exact equation of motion
749: (\ref{e_motion1}) on the dynamics described by the approximate equation
750: (\ref{Mathieu}). The approximate equation
751: (\ref{Mathieu}) yields unstable solutions when $\epsilon>\epsilon_1$, where
752: $|\epsilon_1|=\sqrt{24|\delta|/5}$ if $\delta>0$, and
753: $\epsilon_1=\sqrt{24|\delta|}$ if $\delta<0$.\cite{Landau}
754: The parameter $\delta=0.1$ yields $\epsilon_1=0.69$.\footnote{We
755: also checked these criterion numerically.} As one can see from
756: Figs. 8 (b), 8 (c) the central cell remains undestroyed. Thus,
757: the nonlinear terms stabilize the dynamics in the near resonance case,
758: similar to the case of exact resonance.
759: At $\epsilon=1.2$ in Fig. 8 (c)
760: one more cell is generated, because the condition (\ref{elliptic}) is
761: satisfied for two values of $kr$.
762: 
763: 
764:  \begin{figure}[tb]
765:  \begin{center}
766: %\setcounter{figure}{5}
767:  \mbox{\psfig{file=l2a.ps,width=8cm,height=8cm}
768:        \psfig{file=l2b.ps,width=8cm,height=8cm}}
769:  \vspace{0.5cm}
770:  \caption{The phase space for the near resonance case, $\delta=0.1$,
771:  $\ell=2$, and (a) $\epsilon=0.17$, (b) $\epsilon=0.22$}
772:  \end{center}
773:  \label{fig:9}
774:  \end{figure}
775: 
776: 
777: Unlike the case $\ell=1$, 
778: when $\ell=2$ and $\epsilon$ is small enough (see Figs. 9 (a)), the 
779: stable point does not shift from the point $(X=0,\,P=0)$. Instead, 
780: in Fig. 9 (b) we
781: observe bifurcation at the value $\epsilon=\epsilon_2$, where 
782: $\epsilon_2$ can be estimated from the solution of the approximate
783: equation (\ref{Mathieu}). Namely, up to the second order in $\delta$,  
784: the dynamics becomes unstable at
785: $\epsilon_2=2\delta-\delta^2/2$.\cite{Landau} Our computed value
786: of $\epsilon_2$ lies in the interval $0.185<\epsilon_2<0.186$ which 
787: is slightly less than the estimated quantity due to the influence 
788: of nonlinear in $X$ terms, which are neglected in the approximate equation
789: (\ref{Mathieu}).        
790:     
791: \begin{figure}[tb]
792: %\begin{center}
793: %\setcounter{figure}{5}
794: \centerline{\psfig{file=vader.ps,width=13cm,height=13cm}} 
795: \caption{The same as in Figs. 9 (a), 9 (b) but for $\epsilon=0.8$} 
796: \label{fig:10}
797: \end{figure}
798: 
799: As shown in Fig. 10, 
800: at further increase of $\epsilon$, two stable points, formed
801: after bifurcation, diverge 
802: at larger distance from each other, the chaotic area increases, 
803: and  additional cells appear because the condition (\ref{elliptic})
804: is satisfied for more number of points ($kr(\tilde I_e)$).      
805: 
806: \section{Stability of the quantum ground state}
807: Now we consider a stability of the ground state of the quantum
808: harmonic oscillator (QGS) under the same conditions as in the classical model.
809: The quantum Hamiltonian is,
810: \begin{equation}
811: \label{q_H}
812: \hat H=\frac{\hat p^2}{2M}+\frac{M\omega^2}2x^2
813: +v_0\cos(kx-\Omega t)=\hat H_0+\hat V(x,t),
814: \end{equation}
815: where $\hat p=-i\hbar\partial/\partial x$, and the same notation as in Eq. (\ref{cl_H}) where used. Since the 
816: Hamiltonian (\ref{q_H}) is periodic in time, we can use the Floquet 
817: theorem, and write the solution of the Schr\"odinger equation in the form, 
818: \begin{equation}
819: \label{qef}
820: \psi_q(x,t)=\exp(-iE_qt/\hbar)u_q(x,t),
821: \end{equation}
822: where $E_q$ is the quasienergy, $\psi_q(x,t)$ is the quasienergy (QE) 
823: eigenfunction, and the function $u_q(x,t)$ is periodic in time, 
824: $u_q(x,t)=u_q(x,t+T)$, where $T=2\pi/\Omega$. 
825: We expand $u_q(x,t)$ in the basis of the unperturbed harmonic 
826: oscillator,   
827: \begin{equation}
828: \label{qe_dec}
829: u_q(x,t)=\sum_{n=0}^\infty C_n^q(t)\psi_n(x),
830: \end{equation}
831: where coefficients, $C_n^q(t)$, are periodic in time, 
832: $C_n^q(t)=C_n^q(t+T)$. Due to periodicity of $C_n^q(t)$, the approach 
833: based on  Floquet states is very convenient for investigation 
834: of localization properties of the 
835: quantum system. Namely, if some initial state coincides with the
836: quasienergy function localized in some region of the Hilbert space,
837: $C_n(0)=C_n^q(0)$, then it will remain localized in this region for any time.   
838: 
839: We used the following numerical procedure to calculate the 
840: QE states. \cite{Ber1,Ber2,Reichl_a} The QE states are the 
841: eigenstates of the evolution operator for one period of the wave 
842: field, $\hat U(T)$. 
843: In order to build the matrix $U_{nm}$ of the operator  
844: $\hat U(T)$ we choose the 
845: representation of the Hamiltonian $\hat H_0$.   
846: Let us act with the
847: evolution operator on the wave function $\psi(x,0)$,
848: \begin{equation}
849: \label{qefind1}
850: \hat U(T)\psi(x,0)=\psi(x,T),
851: \end{equation}
852: and choose the initial state in the form $C_n(0)=\delta_{n,n_0}$.  
853: In this way we obtain a column in the evolution operator matrix,  
854: 
855:  \begin{figure}[tb]
856:  \begin{center}
857:  \mbox{\psfig{file=qe1.ps,width=7cm,height=7cm}\hspace{0.5cm}
858:        \psfig{file=qe2.ps,width=7cm,height=7cm}}
859:  \end{center}
860:  
861:  \begin{center}
862:  \mbox{\psfig{file=qe3.ps,width=7cm,height=7cm}
863:        \hspace{0.5cm}\psfig{file=qe4.ps,width=7cm,height=7cm}}
864:  \end{center}
865:  \centerline{\psfig{file=qe5.ps,width=7cm,height=7cm}}
866:  \vspace{0.4cm}  
867:  \caption{ The plots of averages versus dispersions for 
868:   $h=0.5$, $\delta=0$ and
869:  (a) $\ell=1$, $\epsilon=0.05$, (b) $\ell=2$, $\epsilon=0.05$,
870:  (c) $\ell=3$, $\epsilon=5\times 10^{-4}$, (d) $\ell=4$, $\epsilon=0.05$,
871:  (e) $\ell=5$, $\epsilon=0.05$.}
872:  \label{fig:11}
873:  \end{figure}
874: \noindent
875: \begin{equation}
876: \label{qefind2}
877: U_{n,n_0}=C_n^{(n_0)}(T),
878: \end{equation}
879: where the coefficients, $C_n^{(n_0)}(T)$, can be obtained by numerical solution 
880: of the Schr\"odinger equation (for more detailed discussion see 
881: Ref. \cite{3}). After diagonalization  of $U_{n,m}$, we obtain the QE
882: functions, $C_n^q\equiv C_n^q(mT)$, $m=0,\,1,\,2,\dots$ and the
883: quasienergies, $E_q$. The values of matrix elements, $U_{n,m}$,
884: depend on three dimensionless parameters: the wave amplitude, 
885: $\epsilon$, the quantum parameter,
886: $h=k^2\hbar/M\omega$, which can be treated as a dimensionless Planck constant,
887: and from the ratio $\Omega/\omega=\ell-\delta$.
888: %
889:  \begin{figure}[tb]
890:  \begin{center}
891: %\setcounter{figure}{5}
892:  \mbox{\psfig{file=f4.ps,width=8cm,height=8cm}\hspace{0.3cm}
893:        \psfig{file=f5.ps,width=8cm,height=8cm}}
894: % \vspace{0.1cm}
895:  \end{center}
896:  \caption{The probability distribution for the QGS QE states 
897:  with the smallest average, $n_q$, for the cases (a) $\ell=4$,
898:  (b) $\ell=5$; $\epsilon=0.05$, $h=0.5$, $\delta=0$.}
899:  \label{fig:12}
900:  \end{figure}
901: %
902: When $\delta=0$ and 
903: the amplitude of the wave is small, $\epsilon\ll 1$, most of the QE states 
904: are divided into almost independent groups, each located in one resonance cell 
905: of the Hilbert space. \cite{1} In order to show this let us characterize 
906: each QE state by its average, 
907: $n_q=\sum_nn|C_n^q|^2$, and a dispersion, 
908: $\sigma_q=\left[\sum_n(n-n_q)^2|C_n^q|^2\right]^{1/2}$ and plot 
909: $n_q$ versus $\sigma_q$ (see also Ref.\cite{bvi}). Such plots for 
910: different values of the resonance number, $\ell$,
911: and for small value of the wave amplitude, $\epsilon$, 
912: are shown in Figs. 11 (a) - 11 (e). The boundaries of the cells are 
913: marked by arrows. The radius of the external boundary of 
914: classical cells in Figs.  1~(a)~-~1~(e) corresponds to the position of the 
915: boundary of the first quantum cell, respectively,
916: in Figs. 11 (a) - 11 (e) with the quantized radius, $\rho_n=\sqrt{2nh}$.
917: One can see from Figs. 11 (a) - 11 (e) that the QE states
918: are mostly located within the quantum resonance cells, because their
919: averages, $n_q$, are situated inside the cells, and their widths, 
920: $\sigma_q$, do not exceed the width of the corresponding resonance
921: cell. Such states  form rows in Figs. 11 (a) - 11 (e). 
922: Each cell in the Hilbert space in the 
923: quasiclassical limit corresponds to $2\ell$ classical cells. \cite{4} 
924: There also
925: exist the QE states which do not belong to a particular resonance cell, 
926: but rather they belong to the stochastic web. These QE states have a 
927: ``separatrix'' structure, i.e. they delocalized over several resonance cells
928: and have large dispersion, $\sigma$. Such states are represented by scattered 
929: points on the diagrams $n_q=n_q(\sigma_q)$ in Figs. 11 (a) - 11 (e).  
930: The structure of these QE states was in details discussed
931: in Refs. \cite{1,3,2}.
932: %
933:  \begin{figure}[tb]
934:  \begin{center}
935: %\setcounter{figure}{5}
936:  \mbox{\psfig{file=f4a.ps,width=7.5cm,height=7.5cm}\hspace{0.5cm}
937:        \psfig{file=f5a.ps,width=7.5cm,height=7.5cm}}
938: % \vspace{0.2cm}
939:  \caption{ The same as in Figs. 12 (a), 12 (b) but in the
940:  logarithmic scale.}
941:  \end{center}
942:  \label{fig:13}
943:  \end{figure}
944: %
945:  In this paper, we focus on the QE states which belong to 
946: the central resonance cell (QGS QE states), because these 
947: states are mainly responsible for stability properties of the QGS. 
948: Such QGS QE states are characterized by 
949: small average, $n_q$, and are located in the low
950: part of the plots, $n_q(\sigma_q)$, in Figs. 11 (d), 11 (e). In the 
951: cases $\ell=1,2$ in Figs. 11 (a) - 11 (b) these states are absent,  
952: which corresponds to unstable dynamics near the CGS in 
953: the phase classical space, shown, respectively, in Figs. 1 (a) - 1 (b). 
954: In the case $\ell=3$, the area of the stable island in Fig. 2 (a) is much 
955:  less than the value of the dimensionless Planck constant $h$ ($h=0.5$).
956:  So, in this case, the QGS QE state is absent, too.      
957: 
958:  The plots of the probability distribution
959: for the QGS QE states $q'$ with 
960: the smallest average, $n_{q'}$, marked in Figs. 11 (d), 11 (e) by arrows, are 
961: shown in Figs. 12~(a), 12~(b), for the cases $\ell=4$ and $\ell=5$.  
962: In the logarithmic scale, these states  are illustrated in Figs.
963: 13 (a), 13 (b). The Husimi functions of these states are 
964: shown in Figs. 14 (a) ($\ell=4$) and 14~(b) ($\ell=5$).      
965: As one can see from Figs. 12 (a), 12 (b) the QGS QE state is mainly localized 
966: in the CGS of the harmonic oscillator.
967: One can see from Figs. 12 (a), 12 (b) and 13 (a), 13 (b)  that the small part 
968: of the probability distribution is located at the levels with the 
969: numbers $n=\ell m$, where $m=1,\,2,\dots$. This 
970: can be explained by influence of the resonance terms in the
971: quantum equations of motion.\cite{1} In the resonance approximation 
972: the QE states can be defined from the set of algebraic equations
973: which in the dimensionless form can be written as,
974: \begin{equation}
975: \label{al_eq}
976: \left({E_q\over \hbar\omega}-\delta n\right)C_n^q=
977: \frac\epsilon h(V_{n,n+\ell}C_{n+\ell}^q+V_{n,n-\ell}C_{n-\ell}^q).
978: \end{equation}
979: 
980:  \begin{figure}[tb]
981:  \begin{center}
982: %\setcounter{figure}{5}
983:  \mbox{\psfig{file=hus4.ps,width=8cm,height=8cm}
984:        \psfig{file=hus5.ps,width=8cm,height=8cm}}
985:  \vspace{0.2cm}
986:  \caption{(a) The Husimi functions of the QGS QE state shown in Fig. 12 (a).
987:  (b) The Husimi functions of the QGS QE state shown in Fig. 12 (b).
988:  The cross-sections are plotted
989:  from the level $0.047$ with the increment $0.042$, $h=0.5$, $\epsilon=0.05$.} 
990:  \end{center}
991:  \label{fig:14}
992:  \end{figure}
993: 
994: \noindent
995: The matrix elements for $n\gg 1$ can be approximated by the 
996: Bessel functions $J_m$,\footnote{More precise form of matrix elements 
997: see in Ref.\cite{1}}  
998: \begin{mathletters}
999: \begin{equation}
1000: \label{m.elements1}
1001: V_{n,n+2m+1}=\frac{1}2
1002: \frac{(-1)^m n^{m+1/2} e^{-\frac{h}4}}
1003: {\sqrt{(n+1)\dots (n+2m+1)}}J_{2m+1}(\sqrt{2nh}) ,
1004: \end{equation}
1005: \begin{equation}
1006: \label{m.elements2}
1007: V_{n,n+2m}=\frac{1}2\frac{(-1)^m n^m e^{-\frac{h}4}}
1008: {\sqrt{(n+1)\dots (n+2m)}}J_{2m}(\sqrt{2nh}).
1009: \end{equation}
1010: \end{mathletters}
1011: 
1012: As one can see from Eq. (\ref{al_eq}), in the resonance approximation 
1013: the QE functions have the form $C_n^q=C_{\ell m}^q$ 
1014: with $m=1,\,2,\dots$. In this case, the particle is allowed to move only 
1015: between the states 
1016: with the numbers $n=\ell m$. As shown in Ref. \cite{4}, 
1017: a particular form 
1018: of the QE function, $C_n^q=C_{\ell m}^q$, makes the 
1019: Husimi functions, illustrated in Figs. 14 (a) and 14 (b), 
1020: symmetric with the
1021: axial symmetry of the order $\ell$.  
1022: One can see that the width of the Husimi distribution 
1023: in Figs. 14 (a) and 14 (b), 
1024: is, $\Delta P\sim\Delta X\sim\sqrt h\sim 0.7$.  
1025: 
1026: 
1027:    \begin{figure}[tb]
1028:  \begin{center}
1029: %\setcounter{figure}{5}
1030:  \mbox{\psfig{file=5b.ps,width=7.8cm,height=7.8cm}\hspace{0.2cm}
1031:        \psfig{file=fun5b.ps,width=7.8cm,height=7.8cm}}
1032:  \vspace{0.2cm}
1033:  \caption{(a) The Husimi functions of the shifted from the 
1034:  ground state QE function, shown in Fig. 15 (b).
1035:  The cross-sections are plotted
1036:  from the level $0.047$ with the increment $0.042$,
1037:  $\epsilon=5.0$, $h=0.5$.} 
1038:  \end{center}
1039:  \label{fig:15}
1040:  \end{figure}
1041: 
1042: Now, let us consider influence of dynamical chaos on the QE states,
1043: when $\epsilon$ increases.
1044: There are several QE states localized 
1045: near the QGS of the harmonic oscillator. Some of them 
1046: are shifted from the level with the number $n=0$. They can be associated 
1047: with shifted up classical central resonance cell, when $\epsilon$ increases.  
1048: In this case, the Husimi function in Fig. 15 (a) of the QE state with the 
1049: probability 
1050: distribution illustrated in Fig. 15 (b) (for $\ell=5$ and $\epsilon=5$)
1051: has a similar form to the form of the trajectories in the classical phase space 
1052: shown in Fig. 5 (a). 
1053: At large enough values of the wave amplitude, 
1054: ($\epsilon=5$ in Fig. 15 (a)), there are no 
1055: QE states localized in the nearest quantum cells, except for the 
1056: QE states localized in the central cell. 
1057: This corresponds to the chaotic classical dynamics  
1058: shown in Figs. 7 (b), 7 (d) with the stable island in the center
1059: of the phase space.  
1060: 
1061: \begin{figure}[tb]
1062: %\begin{center}
1063: %\setcounter{figure}{5}
1064: \centerline{\psfig{file=n0l5.ps,width=15cm,height=10cm}} 
1065: \caption{The probability to find the system 
1066: in the QGS of the 
1067: harmonic oscillator, defined 
1068: by the values $|C_{n=0}^{q'}|^4$ of the QE state $q'$ mostly localized at the 
1069: state with $n=0$, versus the wave amplitude, $\epsilon$, for three values 
1070: of the effective Planck constant, $h$; $\ell=5$.}
1071: %\end{center}
1072: \label{fig:16}
1073: \end{figure}
1074: 
1075: The QGS QE states, mostly localized at the CGS of the harmonic oscillator, 
1076: are of the most interest for us, because they mainly define the dynamics of the 
1077: quantum state initially located at the level with $n=0$. The 
1078: time-evolution of the system with the initial 
1079: state $C_n(0)=\delta_{n,n_0}$ is defined by the equation,  
1080: \begin{equation}
1081: \label{cnmt}
1082: C_n(mT)=\sum_q C_n^{q*} C_{n_0}^q\exp(-iE_q mT/\hbar),
1083: \end{equation}
1084: where $m=0,\,1,\,2,\dots$.
1085: The amplitude of probability to find the system in the initial state,  
1086: $n_0$, is, 
1087: \begin{equation}
1088: \label{cn0mt}
1089: C_{n_0}(mT)=\sum_q |C_{n_0}^{q}|^2 \exp(-iE_q mT/\hbar).
1090: \end{equation}
1091: Suppose that some QE state with the number $q'$ 
1092: is mostly localized at the level $n=n_0$, i.e. 
1093: $|C_{n_0}^{q'}|^2\gg |C_{n_0}^{q}|^2$ for all $q\ne q'$. Then, 
1094: the term with $q=q'$ dominates in the sum in the right-hand side 
1095: of Eq. (\ref{cn0mt}), and we can write, 
1096: \begin{equation}
1097: \label{cn0mt1}
1098: C_{n_0}(mT)\approx |C_{n_0}^{q'}|^2 \exp(-iE_{q'} mT/\hbar).
1099: \end{equation}
1100: The probability, $P_{n_0}(mT)$, to find the system at the moments $t_m=mT$ 
1101: in the state with $n=n_0$ is given by,
1102: \begin{equation}
1103: \label{cn0mt2}
1104: P_{n_0}(mT)=|C_{n_0}(mT)|^2\approx |C_{n_0}^{q'}|^4.
1105: \end{equation}
1106: The value of $P_{n_0}(mT)$ in this approximation
1107: is independent of the number of periods passed, 
1108: $P_{n_0}(mT)\equiv P_{n_0}$. In the next approximation, 
1109: the neglected terms in Eq. (\ref{cn0mt}) cause the probability $P_{n_0}$
1110: slightly oscillate with time.  
1111: 
1112: In Fig. 16, we present a plot of the probability, $P_0=|C_{n_0=0}^q|^4$,
1113: to find the system in the QGS
1114: as a function of the wave amplitude, $\epsilon$,
1115: if the initial state is the QGS of the harmonic oscillator. 
1116: One can see that the dynamical chaos (the range of large enough $\epsilon$) decreases this probability. 
1117: However, the process of
1118: delocalization of the QGS QE state 
1119: is extremely slow when $\epsilon$ increases,
1120: in comparison 
1121: with that in other nearest cells. For example, at $\epsilon=5$ all QE states
1122: in the nearest cells are chaotic (delocalized) which corresponds 
1123: to the chaotic classical 
1124: dynamics in Fig. 7 (d), while the QE state located 
1125: at the QGS remains localized with the probability 
1126: $P_0\approx 0.56$ when $h=0.5$, and $P_0\approx 0.6$ when $h=1.0$.  
1127: From comparison   
1128: 
1129:  \begin{figure}[tb]
1130:  \begin{center}
1131: %\setcounter{figure}{5}
1132:  \mbox{\psfig{file=f005b.ps,width=7.8cm,height=6.8cm}\hspace{0.5cm}
1133:        \psfig{file=f05b.ps,width=7.8cm,height=6.8cm}}
1134:  \mbox{\psfig{file=f55b.ps,width=7.8cm,height=6.8cm}\hspace{0.5cm}
1135:        \psfig{file=f6b.ps,width=7.8cm,height=6.8cm}}
1136:  \mbox{\psfig{file=f76b.ps,width=7.8cm,height=6.8cm}\hspace{0.5cm}
1137:        \psfig{file=f92b.ps,width=7.8cm,height=6.8cm}}
1138:  \vspace{0.2cm}
1139:  \caption{(a) The QE states mostly localized at the
1140:  QGS of the harmonic oscillator (QGS QE states), $h=0.5$, 
1141:  (a) $\epsilon=0.05$, (b) $\epsilon=0.5$, (c) $\epsilon=5.5$
1142:  (d) $\epsilon=6.0$, (e) $\epsilon=7.6$, (f) $\epsilon=9.2$.
1143:  The boundaries of the quantum cell are marked by arrows.}   
1144:  \end{center}
1145:  \label{fig:17}
1146:  \end{figure}
1147: 
1148: \noindent
1149: of different curves, for h=0.1,
1150: $h=0.5$ and $h=1.0$, one can note the following features. 
1151: i) At small values of $\epsilon$,  
1152: increase of $h$ leads, on average, to decrease of stability
1153: of the QGS. ii) The QGS, at large values of $h$ ($h=1$), 
1154: is more stable
1155: under the influence of chaos (the range of large enough $\epsilon$) 
1156: than that at small values of $h$ ($h=0.1$).  
1157: We should note, that oscillations of $P_0(mT)$ in time should increase when  
1158: $P_0(mT)$ decreases. This happens in the region of 
1159: large enough $\epsilon$ in Fig. 16, 
1160: because in this case the influence of 
1161: neglected terms in Eq. (\ref{cn0mt}) becomes more significant.   
1162: 
1163: More information about the stability of the QGS can be 
1164: extracted from the analysis of the 
1165: structure of QE states located at the QGS 
1166: at different values of the wave amplitude, $\epsilon$,
1167: shown in Fig. 17 (a) - 17 (f) (for $h=0.5$). The QGS QE state 
1168: shown in Fig. 17 (a) for small value of $\epsilon$   
1169: ($\epsilon=0.05$) is similar
1170: to the separatrix QE states, \cite{3} because it has the regular
1171: structure (compare, for example, with Fig. 17 (c)),
1172: and its maxima are located 
1173: near the quantum separatrices, indicated in Fig. 17~(a) by arrows. 
1174: Thus, the QGS QE state also possesses the properties of the 
1175: ``separatrix'' QE states, considered in Ref. \cite{3}. 
1176: 
1177: The separatrix QE states are of the quantum nature \cite{3} because
1178: they are delocalized over several resonance cells. These QE states
1179: provide the tunneling between the cells 
1180: when chaotic regions in the phase space are negligible small, and 
1181: the classical particle can not practically penetrate from one resonance cell to
1182: another. The separatrix QE function mostly localized in the QGS
1183: is a particular one, and it is different from other separatrix states 
1184: studied before.\cite{3} On the one hand, it is delocalized over several
1185: resonance cells (see Fig. 17 (a)) as other separatrix QE functions. 
1186: On the other hand, this particular QE function is mostly concentrated 
1187: on the QGS of the harmonic oscillator, unlike the other separatrix QE 
1188: functions. These ``contradictory'' features of the QGS QE state
1189: define the dynamics: on the one hand, the system mainly
1190: remains localized in the QGS, but on the other hand, a small part 
1191: of the probability distribution can tunnel to the states with large $n$, 
1192: located in the other resonant cells.
1193: When we increase the parameter $h$,
1194: the separatrix
1195: QE states become more delocalized, and the probability to tunnel to
1196: other cells increases, which explain decrease of $P_0$ with increasing $h$
1197: in Fig. 16 (compare the different curves in the region of small $\epsilon$).
1198: 
1199: At intermediate values of $\epsilon$, when
1200: $1<\epsilon<7$, a stability of the QGS increases with increasing $h$.
1201: However, we can not increase $h$ indefinitely because when
1202: $h$ becomes larger than the resonance area in the phase space, the 
1203: QGS becomes unstable. Thus, at $h=5$, $\ell=4$, $\epsilon=2$ 
1204: (see the classical phase space in Fig. 4 (b)) and at
1205: $h=5$, $\ell=5$, $\epsilon=5$ (Fig. 5 (a)) no localized QGS was
1206: found. 
1207: 
1208: An increase of $P_0$ with increasing the 
1209: wave amplitude, $\epsilon$, when $\epsilon$ is small 
1210: ($\epsilon<0.5$ for $h=0.5$ and $\epsilon<1.0$ for $h=1.0$),
1211: shown in Fig. 16, is a consequence of a partial localization of the separatrix
1212: QE state (see Fig. 17 (b)) under the influence of chaos   
1213: explored in Ref. \cite{3} In this case, the QGS QE function, 
1214: shown in Fig. 17 (b), loses its ``separatrix'' features. 
1215: Further increase of $\epsilon$, makes 
1216: the QGS QE state more delocalized. However, as one can see from 
1217: Figs. 17 (c) ($\epsilon=5.5,\,h=0.5$)
1218: and 17 (d) ($\epsilon=6.0,\,h=0.5$)
1219: delocalization takes place mainly over the nearest oscillator states
1220: with small numbers, $n$. At $\epsilon>5.5$ ($h=0.5$) the  
1221: oscillations appear in the dependence $P_0=P_0(\epsilon)$, in Fig. 16. 
1222: Thus, the QE function 
1223: at $\epsilon=6$ in Fig. 17 (c) is less localized than the 
1224: QE function at $\epsilon=7.6$ shown in Fig. 17 (d). 
1225: At large values of $\epsilon$ ($\epsilon>7.6$), practically 
1226: all QE states are delocalized, which
1227: is the quantum 
1228: manifestation of chaotization of the classical central cell
1229: in the phase space. 
1230: 
1231:  \begin{figure}[tb]
1232:  \begin{center}
1233: %\setcounter{figure}{5}
1234:  \mbox{\psfig{file=e005l5.ps,width=5.4cm,height=5.4cm}
1235:        \psfig{file=e05l5.ps,width=5.4cm,height=5.4cm}
1236:        \psfig{file=e76l5.ps,width=5.4cm,height=5.4cm}}
1237: % \vspace{0.2cm}
1238:  \caption{Time-evolution of the dispersion $\sigma=\sigma(m)$,
1239:  where $m=t/T$, for three values of the wave amplitudes:
1240:  (a) $\epsilon=0.05$,
1241:  (b) $\epsilon=0.5$, (c) $\epsilon=7.6$, and for $h=0.5$, $\ell=5$,
1242:  $\delta=0$.}
1243:  \end{center}
1244:  \label{fig:18}
1245:  \end{figure}
1246: 
1247: 
1248: In order to illustrate  a non-monotonic character of 
1249: localization of the QGS as a function of the wave amplitude at small $\epsilon$, 
1250: we computed the dynamics of the quantum state initially concentrated 
1251: on the ground state of the harmonic oscillator, $C_n(0)=\delta_{n,0}$,
1252: using Eq (\ref{cnmt}). Time-evolution of the dispersion, 
1253: \begin{equation}
1254: \label{sigma}
1255: \sigma(mT)=\sqrt{\sum_n|C_n(mT)|^2(n-\bar n(mT))^2}, 
1256: \end{equation}
1257: where $\bar n(mT)=\sum_n|C_n(mT)|^2 n$ is the average, 
1258: is presented in Figs. 18 (a) - 18 (c), for three values of 
1259: $\epsilon$. When the wave amplitude, $\epsilon$, is small (Fig. 18 (a)), 
1260: a small 
1261: part of the wave packet can propagate to large values of $n$ due to 
1262: effect of diffusion via the separatrices as shown in Fig. 19 (a). 
1263: Similar tunneling effect of  the wave packet between the resonance cells 
1264: via the separatrices was explored in Ref. \cite{3} 
1265: In spite of a small probability of tunneling to other cells,
1266: contribution of this part to the dispersion, $\sigma$, is significant because it is
1267: proportional to $(n-\bar n)^2$, where $n-\bar n\gg 1$.
1268: At $\epsilon=0.5$ the separatrix QE 
1269: states are destroyed by chaos, as shown in Fig. 17 (b), and QGS 
1270: becomes more localized (see Fig. 19 (b)). This leads to a significant 
1271: decrease of $\sigma$ in Fig. 18 (b) in comparison with the case of 
1272: small $\epsilon$, shown in Fig. 18 (a). Further increase of $\epsilon$, 
1273: up to the value $\epsilon=7.6$, results in delocalization of the 
1274: QGS, as shown in Fig. 19 (c), and the dispersion, $\sigma=\sigma(mT)$,
1275: in Fig. 18 (c) becomes large again. 
1276: 
1277:  \begin{figure}[tb]
1278: % \begin{center}
1279: %\setcounter{figure}{5}
1280:  \mbox{\psfig{file=a005l5.ps,width=5cm,height=5cm}\hspace{0.2cm}
1281:        \psfig{file=a05l5.ps,width=5cm,height=5cm}\hspace{0.2cm}
1282:        \psfig{file=a76l5.ps,width=5cm,height=5cm}}
1283: % \vspace{0.2cm}
1284:  \caption{Probability distribution in the system with
1285:  initial condition $C_n(0)=\delta_{n,n_0}$ and
1286:  (a) $\epsilon=0.05$, $mT=3\times 10^5T$; here $|C_0|^2=0.746$,
1287:  (b) $\epsilon=0.5$, $mT=10^4T$; $|C_0|^2=0.982$,
1288:  (c) $\epsilon=7.6$, $mT=5000T$; $|C_0|^2=0.137$, $|C_1|^2=0.423$.
1289:  The boundaries of quantum cells (quantum separatrices)
1290:  are marked by arrows; $h=0.5$, $\ell=5$, $\delta=0$.}
1291: % \end{center}
1292:  \label{fig:19}
1293:  \end{figure}
1294: 
1295: Comparison of Figs. 19 (a) and 
1296: 19 (c) allows us to conclude that delocalization of the QGS at very small, 
1297: and at large values of $\epsilon$ is of a different nature. In the former 
1298: case, the diffusion is caused by the separatrix QE states. These states 
1299: are the quantum objects because they are delocalized,
1300: and provide tunneling between the resonant cells. \cite{3}
1301: As a consequence of the quantum nature of the separatrix QE states, 
1302: increase of the dimensionless Planck constant, $h$, leads to  
1303: delocalization of the separatrix QE states 
1304: and decrease of localization of QGS at small $\epsilon$,
1305: shown in Fig. 16. On the other hand, when $\epsilon$ is large we 
1306: observe delocalization 
1307: caused by chaos, which is manifested in the irregular form of the 
1308: probability distribution, shown in Fig. 19 (c). 
1309: 
1310: 
1311: As follows from Fig. 16, in order to make the QGS more stable at 
1312: small $\epsilon$ one should decrease the Planck constant, $h$.
1313: A plot of the dispersion, $\sigma$, versus time for $h=0.1$ is presented in 
1314: Fig. 20 (a). 
1315: The system remains in the ground state with the probability 
1316: $P_0=0.996$. However the dispersion is still large enough because the particle
1317: can propagate with small probability to the levels with large $n\gg 1$, due to the 
1318: diffusion via the separatrices which can be seen from the plot 
1319: of the probability distribution presented in Fig. 20 (b).  
1320: 
1321: Another way to increase the stability of the ground state at small 
1322: $\epsilon$, is to destroy the separatrix QE functions by choosing
1323: the non-resonant value of the wave frequency, so that
1324: $\delta=\ell-\Omega/\omega\ne 0$. 
1325: For a non-resonant case ($\delta=0.01$),
1326: the plot of the dispersion as a function of time is presented 
1327: in Fig. 21 (a), and
1328: the probability distribution at time $m=t/T=3\times 10^6$ is illustrated
1329: in Fig. 21 (b). 
1330: One can see from Fig 21 (a)
1331: that introducing a detuning,
1332: $\delta\ne 0$, results in considerable improvement of the stability of the 
1333: QGS in comparison with  
1334: the case of the exact resonance, illustrated in Fig. 18 (a). 
1335: Thus, in order to make the ground state more stable
1336: at small values of $\epsilon$ one must
1337: detune the system from the exact resonance.  
1338: 
1339: The minimal value of detuning, $\delta$, required to 
1340: destroy the separatrix structure, and
1341: to make the ground state more 
1342: stable, can be estimated from the quantum equations of motions 
1343: in the resonance approximation (\ref{al_eq}). 
1344: The term, proportional to $\delta n$ destroys the 
1345: separatrix QE states as shown in Ref.\cite{1}  
1346: This term becomes significant when it becomes of the order 
1347: of the relation, $\epsilon/h$.
1348: On the other hand, the separatrix 
1349: QE function must, at least, occupy the two separatrices. Thus, we can 
1350: estimate the number, $n=n_\delta$, of the oscillator's state at which the 
1351: separatrix structure is destroyed, namely $n_\delta=\epsilon/\delta h$.
1352: The separatrix QE functions decay exponentially
1353:  with increasing $n$
1354: in the region $n>n_\delta$.\cite{1}  
1355: For the parameters 
1356: in Fig. 21, $n_\delta=50$, which is much less than the position of the
1357: first separatrix $n_1=385$ shown in Fig. 20 (b) by arrow. 
1358: So, the separatrix QE states at these parameters do not exist, and we 
1359: do not observe tunneling from the QGS to other resonance cells. 
1360: 
1361:  \begin{figure}[tb]
1362: % \begin{center}
1363: %\setcounter{figure}{5}
1364: \centerline{\mbox{\psfig{file=h01s.ps,width=8cm,height=8cm}\hspace{0.4cm}
1365:        \psfig{file=h01p.ps,width=8cm,height=8cm}}}
1366: % \end{center}
1367:  \vspace{0.4cm}
1368:  \caption{(a) Time-evolution of the dispersion, $\sigma=\sigma(m)$,
1369:  where $m=t/T$, and (b) probability distribution at time 
1370:  $m=3\times 10^6$ for small value of $h$, $\delta=0$, $h=0.1$, $\epsilon=0.05$}
1371:  \label{fig:20}
1372:  \end{figure}
1373: \begin{figure}[tb]
1374: % \begin{center}
1375: %\setcounter{figure}{5}
1376: \centerline{\mbox{\psfig{file=e005l5dw.ps,width=8cm,height=8cm}\hspace{0.4cm}
1377:        \psfig{file=a005l5dw.ps,width=8cm,height=8cm}}}
1378: % \end{center}
1379: % \vspace{0.2cm}
1380:  \vspace{0.4cm}
1381:  \caption{(a) Time-evolution of the dispersion $\sigma=\sigma(m)$,
1382:  where $m=t/T$, and (b) probability distribution at time 
1383:  $m=3\times 10^6$ for the near resonance case, $\delta=0.01$; $h=0.1$, $\epsilon=0.05$}
1384:  \label{fig:21}
1385:  \end{figure}
1386: Decreasing the value of $\epsilon$, in the near resonance case, 
1387: makes the QGS more stable. 
1388: %Thus, probability to find 
1389: %the system in the ground state at $\epsilon=0.05$ and $\delta=0.01$ is 
1390: %$P_0=0.9979$ at $h=1$, $P_0=0.999546$ at $h=0.5$, $P_0=0.99897$ at $h=0.1$.
1391: Unlike the near resonance case, in the case of the exact resonance,
1392: a stability of the QGS at $\epsilon\ll 1$ is independent 
1393: of the wave amplitude, because in this case the localization 
1394: properties of the QGS are defined by the structure of separatrix 
1395: QE function, which
1396: in the resonance approximation (when $\epsilon$ is small) is 
1397: independent of $\epsilon$.\cite{1} 
1398: One can see from a comparison of Fig. 21 (a) with Fig. 20 (a) 
1399: ($h=0.1$, $\epsilon=0.05$)
1400:  that the dispersion in the near resonance case (Fig. 21 (a))
1401: is much less than that in the exact resonance case (Fig. 20 (a)), in spite of
1402: the probability to stay in the ground state, $P_0$, for these two cases 
1403: does not differ significantly ($P_0=0.99897$ at $\delta=0.01$ and 
1404: $P_0=0.996$ at $\delta=0$). The reason is that the dynamics in the exact 
1405: resonance case is mainly determined by the separatrix QE function, and 
1406: in this case there is a small probability for particle to tunnel 
1407: to the high oscillator levels with $n\gg 1$, as shown in Fig. 20 (b).  
1408: %   
1409: \begin{figure}[tb]
1410: %\begin{center}
1411: %\setcounter{figure}{5}
1412: \centerline{\psfig{file=n0l5dw.ps,width=15cm,height=10cm}} 
1413: \caption{The same as in Fig. 16, but for the near resonance case,
1414: $\delta=0.01$.}
1415: %\end{center}
1416: \label{fig:22}
1417: \end{figure}
1418: %
1419: When the wave amplitude $\epsilon$ increases, the condition
1420: $\delta\ne 0$ becomes less significant. This can be explained by 
1421: less influence of the term proportional to $\delta$ on the
1422: dynamics in comparison with influence of the wave with the amplitude
1423: $\epsilon$ in the region where the value of $kr$ is relatively small
1424: (see the classical Hamiltonian in Eqs. (\ref{cl_H0}),
1425: (\ref{wave_decomposition}), (\ref{cl_H3}) and
1426: quantum Eq. (\ref{al_eq})).
1427: In Fig. 22 we plot the function
1428: $P_0=P_0(\epsilon)$ for the near resonance case.
1429: In Figs. 23 (a) - 23 (c) we compare the results for
1430: the exact resonance case
1431: ($\delta=0$) with those for the near resonance case,
1432: when $\delta=0.01$ and $\delta=0.1$.
1433: One can see from  these figures that the dynamics in the vicinity
1434: of the QGS in the near resonance
1435: case is similar to that in the exact resonance case,
1436: except for the region of small $\epsilon$, which was discussed above.
1437: %
1438:  \vspace{0.4cm}
1439:  \begin{figure}[tb]
1440: % \begin{center}
1441: %\setcounter{figure}{5}
1442: \centerline{\mbox{\psfig{file=n0l5h1.ps,width=5.4cm,height=5.4cm}
1443:        \psfig{file=n0l5h05.ps,width=5.4cm,height=5.4cm}
1444:        \psfig{file=n0l5h01.ps,width=5.4cm,height=5.4cm}}}
1445: %\end{center}
1446:  \vspace{0.5cm}
1447:  \caption{Plot of $P_0$ versus $\epsilon$ for 
1448:  the exact resonance case,  when $\delta=0$ (solid line and open circles),
1449:  and two curves for the near resonance cases:
1450:  $\delta=0.01$ (dashed line and filled squares), 
1451:  $\delta=0.1$ (dot-dashed line and crosses);   
1452:  $\ell=5$, (a) $h=1$, (b) $h=0.5$,
1453:  (c) $h=0.1$.}   
1454:  \label{fig:23}
1455:  \end{figure}      
1456: %
1457: In the quantum case, similar to the classical one, at very small $\epsilon$
1458: and finite $\delta$, there always exists the QGS QE state at any $\ell$
1459: including the cases $\ell=1$ and $\ell=2$. In Fig. 24 we present a plot
1460: $P_0=P_0(\epsilon)$ for the cases $\ell=1$ and $\ell=2$ when
1461: $\delta=0.1$ and $h=0.5$. As one can see from Fig. 24 the QGS QE state
1462: in these two cases becomes unstable at considerably less
1463: values of $\epsilon$ than that in the case $\ell=5$ in Fig. 23, which
1464: corresponds to the classical dynamics in Figs. 8 - 10. In the case 
1465: $\ell=1$ the stable point shifts down from the region $X=0,\,P=0$ in 
1466: Figs. 8 (a)- 8 (c), and in the case $\ell=2$ the stable point 
1467: at  $\epsilon\approx 0.2$ becomes unstable (see Figs. 9 (a), 9 (b)) and 10).
1468: The quantum manifestation of this process is a rapid decay of the value
1469: of $P_0$ in the region $\epsilon\ge 0.2$ in Fig. 24.
1470: 
1471: In conclusion, the classical dynamics in the vicinity of the point ($x=0,\,p=0$) in
1472: the classical phase space and quantum dynamics in the vicinity of the ground state of the
1473: harmonic oscillator in a field of a monochromatic wave, is explored.
1474: Both resonance and near resonance cases are analyzed.
1475: It is shown that at small $\epsilon$ and finite detuning from the resonance,
1476: $\delta$, the quantum ground state is always stable.
1477: In the case $\delta=0$, the dynamics is
1478: unstable for the resonance
1479: numbers $\ell=1,2$.
1480: %For small enough wave amplitude, $\epsilon$
1481: %and small quantum parameter, $h$, the stability of the ground state 
1482: %can be significantly improved by choosing nonresonant value of the 
1483: %wave with frequency $\Omega\ne\ell\omega$, where $\ell=1,2\dots$.
1484: Stability of the classical dynamics in the central cell 
1485: and stability of the quantum dynamics near the ground state of the harmonic 
1486: oscillator under the influence of chaos is analyzed. It is shown, 
1487: that under certain conditions 
1488: ($\ell>2,\,\delta=0,\,\epsilon\ll 1,\,h\sim 1$)
1489: the presence of chaos makes the quantum ground state more localized. 
1490: Increase of the quantum parameter, $h$,  
1491: at intermediate values of $\epsilon$, enhances localization of the 
1492: QGS considerably.   
1493: Experimental observation of discussed results in the system of an 
1494: ion trapped in a linear ion trap and interacting with laser
1495: fields represents a significant interest for understanding
1496: the stability properties of this system.
1497: 
1498: \begin{figure}[tb]
1499: %\begin{center}
1500: %\setcounter{figure}{5}
1501: \centerline{\psfig{file=l1l2.ps,width=15cm,height=10cm}} 
1502: \caption{Plot $P_0=P_0(\epsilon)$ for the resonance numbers 
1503: $\ell=1$ (open circles and solid line) and 
1504: $\ell=2$ (filled squares and dashed line) in the near resonance case, 
1505: $\delta=0.1$; $h=0.5$.}
1506: %\end{center}
1507: \label{fig:24}
1508: \end{figure}
1509: 
1510: 
1511: %
1512: \section{Acknowledgments}
1513: We are thankful to R.J. Hughes and G.D. Doolen for useful discussions.
1514: This work was  supported by the National Security Agency, and by the
1515: Department of Energy under the contract W-7405-ENG-36.
1516: Work of D.I.K. was partly supported by Russian Foundation for Basic Research
1517: (Grants No. 98-02-16412 and 98-02-16237).
1518: %\newpage
1519: %
1520: \begin{thebibliography}{}
1521: 
1522: \bibitem{fortsch}
1523: R.J. Hughes, D.F.V. James, J.J. Gomez, M.S. Gulley, M.H. Holzscheiter,
1524: P.G. Kwiat, S.K. Lamoreaux, C.G. Peterson, V. D. Sandberg, M.M.
1525: Schauer, C.M. Simmons, C.E. Thorburn, D. Tupa, P.Z. Wang and A.G.
1526: White, {\it Fortschritte der
1527: Physik},
1528: {\bf 46}, 329 (1998).
1529: 
1530: \bibitem{bdmt}
1531: G.P. Berman, G.D. Doolen, R. Mainieri and V.I. Tsifrinovich, {\it
1532: Introduction to Quantum Computers}, World Scientific Publishing Company,
1533: Singapore, 1998.
1534: 
1535: \bibitem{BD}
1536: G.P. Berman, D.F.V. James, R.J. Hughes, M.S. Gulley, M.H. Holzscheiter,
1537: G.V. L\'opez, quant-ph/9903063.
1538: 
1539: \bibitem{Z}
1540: G.M. Zaslavsky, R.Z. Sagdeev, D.A. Usikov,
1541: and A.A. Chernikov,
1542: {\it Weak Chaos and Quasiregular Patterns}, (Cambridge Univ. Press,
1543: Cambridge, 1991)
1544: 
1545: \bibitem{L} A.J. Lichtenberg and M.A. Lieberman,
1546:      {\it Regular and Stochastic Motion}, (Springer, New York, 1983).
1547: 
1548: \bibitem{1} V.Ya. Demikhovskii, D.I. Kamenev, and G.A.
1549:     Luna-Acosta, {\it Phys. Rev. E}, {\bf 52}, (1995) 3351.
1550: 
1551: \bibitem{3} V.Ya. Demikhovskii, D.I. Kamenev, and G.A.
1552:     Luna-Acosta, {\it Phys. Rev. E}, {\bf 59}, (1999) 294.
1553: 
1554: \bibitem{Mathieu} N.W. McLachlan,
1555:      {\it Theory and application of Mathieu functions}, 
1556:      (Oxford University Press, Oxford, 1947).
1557: 
1558: \bibitem{Landau} L.D. Landau, E.M. Lifshits,
1559:      {\it Mechanics}, (Moscow, 1958) (in Russian).
1560: 
1561: \bibitem{Ber1}
1562: G.P. Berman, A.R. Kolovsky, G.M. Zaslavsky, {\it Phys. Lett. A}, {\bf 111}, (1982) 17.
1563: 
1564: \bibitem{Ber2}
1565: G.P. Berman, A.R. Kolovsky, {\it Sov. Phys. Usp.}, {\bf 35} (1992) 303.
1566: 
1567: \bibitem{Reichl_a} L.E. Reichl and W.A. Lin, {\it Phys. Rev. A}, {\bf 40},
1568:      (1989) 1055.
1569: 
1570: 
1571: \bibitem{bvi}
1572: G.P. Berman, O.F. Vlasova, F.M. Izrailev, {\it Sov. Phys. JETP}, {\bf 66}, (1987) 269.
1573: 
1574: \bibitem{4} G.P. Berman, V.Ya. Demikhovskii, D.I. Kamenev, quant-ph/9906079.
1575: 
1576: \bibitem{2} V.Ya. Demikhovskii, D.I. Kamenev,
1577:       {\it Phys. Lett. A}, {\bf 228}, (1997) 391.
1578: 
1579: \end{thebibliography}{}
1580: 
1581: \end{document}
1582: 
1583: