1: \documentclass[aps,tightenlines,preprint]{revtex4}
2: \usepackage{epsfig}
3: \newcommand{\n}{\noindent}
4: \begin{document}
5:
6: \title{Level velocity statistics of hyperbolic chaos}
7:
8: \author{R. Sankaranarayanan \footnote{Present Address: Center for Nonlinear
9: Dynamics, Department of Physics, Bharathidasan University, Trichy 620024,
10: India.}}
11: \affiliation{Physical Research Laboratory,\\ Navrangpura, Ahmedabad 380009,
12: India.}
13: %\date{\today}
14:
15: \begin{abstract}
16: A generalized version of standard map is quantized as a model of quantum
17: chaos. It is shown that, in hyperbolic chaotic regime, second moment of
18: quantum level velocity is $\sim 1/\hbar$ as predicted by the random matrix
19: theory.
20: \end{abstract}
21: \maketitle
22:
23: \section{Introduction}
24:
25: Chaotic systems are characterized by positive lyapunov exponents such that
26: in phase space neighbourhood trajectories diverge exponentially. Among the
27: chaotic systems there are special class of systems which are {\it completely}
28: chaotic or hyperbolic \cite{ott}. Phase space of hyperbolic systems has only
29: unstable orbits as there are expanding and contracting real directions with
30: positive and negative lyapunov exponents respectively. Area preserving maps
31: like cat map and baker map are examples of hyperbolic systems. Even the well
32: known standard map of kicked rotor, a text book paradigm of Hamiltonian chaos
33: \cite{lich}, is not proven to be hyperbolic even for strong external kick
34: strength.
35:
36: In the study of chaotic quantum systems, it is of fundamental interest to
37: characterize highly chaotic (but not known to be hyperbolic) and hyperbolic
38: chaotic regimes in quantum domain. This forms our motivation here, and to
39: pursue further it would be more appropriate to quantize a {\it single}
40: dynamical system which has parameters for highly chaotic {\it and}
41: hyperbolic chaotic regimes. In one of our earlier works, a generalized
42: version of standard map was introduced to study the dynamics of a
43: kicked particle trapped inside an one dimensional infinite square well
44: potential \cite{sankar01}. The generalization, arising from length scales
45: namely well width and field wave length, has parameters to fulfill the
46: present requirement. With our knowledge, there is no other single system
47: possessing parameters for the above mentioned classical regimes.
48:
49: In quantum domain, dynamics of levels in parameter space is known to have
50: manifestations of classical complexity. While quantum levels cross each other
51: for regular case, they exhibit avoided crossings when underlying classical
52: dynamics is chaotic. Level dynamics can be described by level velocity wherein
53: system parameter plays the role of pseudo-time. In Ref. \cite{gaspard} the
54: notion of curvature i.e., second derivative of levels with parameter, is
55: introduced to quantify the avoided crossings. It is known that the curvature
56: distribution of chaotic system follows an universal behaviour \cite{models}.
57:
58: In this connection, one another quantity of importance is the second moment
59: of level velocity. A semiclassical analysis of kicked system shows that
60: second moment is the sum of all classical time correlations of the kicking
61: potential, such that lowest (zeroth) order correlation being the Random Matrix
62: Theory (RMT) predicted second moment \cite{arul99}. It is also known for
63: quantized standard map that, in highly chaotic regime there are systematic
64: deviations between second moment and the corresponding RMT prediction. The
65: deviations are thus related to the non-vanishing higher order classical time
66: correlations \cite{arul99}.
67:
68: In the present work, we introduce quantum version of a generalized standard
69: map as a model of quantum chaos to study the level velocity statistics. In
70: particular, we compute the second moment and compare with the RMT prediction
71: in order to characterize different chaotic regimes.
72:
73: \section{The Model}
74: \subsection{Classical system}
75:
76: Considering a particle trapped in a one dimensional infinite square well
77: potential $V_0(q)$ of unit width (hard walls at $q=\pm 1/2$), which
78: experiences a periodic kick from an external pulsed field. The Hamiltonian is
79:
80: \begin{equation}
81: \tilde {H} = {p^2\over 2} + V_0(q) + {k\lambda\over 4{\pi}^2}
82: \cos\left({2\pi q\over\lambda}\right) \sum_j \delta (j-t)
83: \end{equation}
84:
85: \n and underlying kick to kick dynamics of the particle is {\it equivalent}
86: to discrete dynamics described by a dimensionless area-preserving mapping:
87:
88: \begin{eqnarray}
89: p_{j+1}&=&p_j+{k\over 2\pi}\sin(2\pi rq_j ) \nonumber \\
90: q_{j+1}&=&q_{j}+p_{j+1}
91: \label{gsm}
92: \end{eqnarray}
93:
94: \n which is defined on 2-torus i.e., a unit square $[-1/2,1/2)\times
95: [-1/2,1/2)$ with periodic boundaries. Here $r=1/\lambda$ is ratio of two
96: length scales of the system namely, well width and field wavelength;
97: $k$ is effective strength of the kick. This is the {\it Generalized Standard
98: Map} (GSM) which was introduced in our earlier studies on the above
99: Hamiltonian \cite{sankar01}.
100:
101: GSM is continuous when $r$ is integer and discontinuous otherwise. One can
102: immediately recognize that widely studied standard map of kicked rotor is a
103: special case ($r=1$) of GSM. Since the standard map is a continuous map,
104: for small $(k<1)$ dynamics is predominantly regular wherein many rotational
105: invariant circles (also called as KAM tori) are interspersed in the phase
106: space. They act as forbidden barriers for chaotic orbits to diffuse. Gradual
107: destruction of these invariant structures with the increase of $k$, leads to
108: onset of chaos; for $k\gg 1$, dynamics is highly chaotic. On the other hand,
109: when $r$ is non-integer no KAM tori exist in the phase space. In this case,
110: depends of the parameter $r$ the phase space is either mixed or fully chaotic
111: even for small $k$ values.
112:
113: The Jacobian ${\bf J}$ of GSM is such that
114:
115: \begin{equation}
116: |\hbox{Trace}\;{\bf J}| = |2+kr\cos(2\pi rq_j)| \, .
117: \end{equation}
118:
119: \n Since $|q_j|\le 1/2$, for $r\le 1/2 \;\; |\hbox{Trace}\;{\bf J}| > 2$.
120: That is to say Jacobian has real eigenvalues. In other words, the system is
121: {\it completely} chaotic or hyperbolic for $r\le 1/2$. In this regime there
122: are contracting and expanding real directions or alternatively stable and
123: unstable manifolds throughout the phase space. Thus GSM is realized as a
124: rare class of dynamical system as it has parameters for both highly chaotic
125: and hyperbolic chaotic regimes.
126:
127: \subsection{Quantum system}
128:
129: GSM arises from the equation of motion of free particle in presence of a field
130: $V(q)$ which is applied as time periodic impulse. The field is defined as:
131: $V(q) = k\cos(2\pi rq)/(4{\pi}^2r); V(q) = V(q+1)$ and the Hamiltonian is
132:
133: \begin{equation}
134: H = {p^2\over 2} + V(q)\sum_j \delta (j-t) \, .
135: \label{ham}
136: \end{equation}
137:
138: \n By integrating Shr\"{o}dinger equation over unit time we obtain
139: corresponding quantum propagator as
140:
141: \begin{equation}
142: \hat{U} = e^{-i{\hat{p}}^2/2\hbar} \,\,\, e^{-iV(\hat{q})/\hbar} \, .
143: \end{equation}
144:
145: \n Then the quantum dynamics can be described as $|\psi (t+1)\rangle =
146: \hat{U}| \psi (t)\rangle$, which is a quantum analogue of the classical map.
147:
148: On quantizing 2-torus phase space by introducing periodic boundary conditions
149: both in $q$ and $p$ \cite{ford91} we have: $\hat{q}|n\rangle = (n/N)|n\rangle
150: \; ; \; \hat{p}|m\rangle = (m/N)|m\rangle$ where $n,m = -N/2, -N/2+1\ldots
151: N/2-1$. Here $N={(2\pi\hbar)}^{-1}$ is the dimensionality of the Hilbert space
152: and the semiclassical limit is $N\rightarrow \infty$. The position and
153: momentum eigenstates obey the periodicity $|n+N\rangle = |n\rangle\; ;\;
154: |m+N\rangle = |m\rangle$ and the transformation function is
155:
156: \begin{equation}
157: \langle n|m\rangle = {1\over \sqrt{N}}
158: \exp\left[{i2\pi mn\over N}\right] \, .
159: \end{equation}
160:
161: \n Being a homogeneous linear system, Shr\"{o}dinger equation in finite $N$
162: dimensional space has solutions $|\phi_j\rangle ,\; j=1,2,\ldots N$, which
163: are linearly independent. Since the Hamiltonian is time periodic (unit period),
164: according to Floquet theory \cite{jor} the solutions satisfy eigenvalue
165: equation $\hat{U}|\phi_j\rangle = e^{-i\phi_j}|\phi_j\rangle$. Eigenstates
166: $|\phi_j\rangle$ are quasienergy states and eigenangles $\phi_j$ are
167: quasienergies. Then general solution at a given time is $|\psi\rangle =
168: \sum_j c_j|\phi_j\rangle$ where $c_j$ are constants. As a consequence of
169: hermiticity of Hamiltonian, quasienergy states are orthogonal and they form
170: a complete set in finite $N$ dimensional space.
171:
172: Matrix form of the propagator in discrete position representation
173: is \cite{arul97}
174:
175: \begin{equation}
176: U_{nn^\prime} \equiv \langle n|\hat{U}|n'\rangle = {1\over \sqrt{N}}
177: \exp\left[ -i\pi \left\{ {1\over 4} - {{(n-n')}^2\over N} +
178: 2NV\left({n'\over N}\right) \right\} \right] \, .
179: \end{equation}
180:
181: \n The Hamiltonian (\ref{ham}) has reflection symmetry about the origin
182: i.e., $H(q,p) = H(-q,-p)$. This symmetry is reflected in the quantum
183: propagator matrix
184:
185: \begin{equation}
186: U_{nn^\prime} = {e^{-i\pi/4}\over\sqrt{N}}\; e^{i\pi{(n-n')}^2/N}
187: \exp\left\{{-ikN\over 2\pi r} \cos \left[{2\pi r\over N}(n'+\alpha)
188: \right]\right\}
189: \end{equation}
190:
191: \n through the relation $[\hat{U},\hat{R}]=0$ where the hermitian operator
192: $\hat{R}$ is defined as
193:
194: \begin{equation}
195: \begin{array}{llll}
196: \hat{R}|n\rangle &=& |-n\rangle &\hbox{for} \;\;\alpha = 0 \\
197: &=& |-n-1\rangle &\hbox{for} \;\;\alpha = 0.5 \, .
198: \end{array}
199: \end{equation}
200:
201: \n Note that a phase factor $\alpha$ is introduced in the matrix element to
202: avoid exact quantum symmetry. Since $\hat{R}^2 = 1$ we may label the
203: eigenstates of $\hat{U}$ with eigenvalues $\pm 1$ of $\hat{R}$ i.e., the
204: states are $|\phi_\pm\rangle$. For $\alpha = 0.5$, symmetry matrix of order
205: $N$ is
206:
207: \begin{equation}
208: R_N = \langle n|\hat{R}|n^\prime \rangle =
209: \delta (n+n^\prime +1) \;\;\;\;\; (\hbox{mod}\; N)
210: \end{equation}
211:
212: \n which has ones along secondary diagonal and zeros elsewhere. Then the state
213: components have a relation $\langle -n-1|\phi\rangle = \pm\langle n
214: |\phi\rangle$ i.e.,
215:
216: \begin{equation}
217: |\phi_\pm\rangle = \left(\begin{array}{r} |v\rangle \\
218: \pm R_{N/2}|v\rangle \end{array}\right) \, .
219: \end{equation}
220:
221: Eigenstates can be numerically obtained by diagonalizing the matrix
222: $U_{nn^\prime}$ of order $N$. If $N$ is even integer, there are $N/2$ even
223: parity states \{$|\phi_+\rangle$\} and $N/2$ odd parity states
224: \{$|\phi_-\rangle$\}. On exploiting $R$-symmetry, the diagonalization can be
225: reduced to the matrix of order $N/2$ by standard procedure \cite{sar89}.
226: The reduced matrix is
227:
228: \begin{eqnarray}
229: {\cal U}_{nn^\prime} &=& {e^{-i\pi/4}\over\sqrt{N}}
230: \exp\left\{{-ikN\over 2\pi r} \cos \left[ {2\pi r\over N} \left(n+{1\over 2}-
231: {N\over 2}\right) \right]\right\} \nonumber \\[8pt]
232: && \;\;\;\;\;\;\;\;\;\;\;\;\;\;\; \times \left\{e^{i\pi{(n-n')}^2/N}
233: \pm e^{i\pi{(n+n'+1)}^2/N} \right\}
234: \end{eqnarray}
235:
236: \n where $n,n^\prime = 0,1,\; .\; .\; .\;N/2-1$. Now the separation of parity
237: states is obvious.
238:
239: \section{Spectral statistics}
240:
241: One of the standard statistical measures for a chaotic quantum system is the
242: nearest neighbour spacing distribution of quantum levels. For a regular system,
243: levels are clustered such that the spacings follow Poisson distribution. On
244: the other hand, in chaotic case the levels exhibit repulsion such that spacings
245: exhibit RMT predicted Wigner distribution. We expect the classical complexity
246: of GSM, arises due to the parameter $r$, will also have manifestation in
247: the spectral spacings. It is evident from Fig. \ref{nns} that, in contrast
248: to the predominantly regular case, for $r=0.5$ (hyperbolic) the spacings
249: follow Wigner distribution.
250:
251: \begin{figure}[h]
252: \centerline{\psfig{figure=nns_k.3.ps,height=10cm,width=12cm}}
253: \caption{Nearest neighbour spacing distributions of 1000 quasienergies which
254: correspond to even parity states for $k=0.3$ with $\alpha = 0.5$. Smooth
255: curves are the Poisson and Wigner distribution.}
256: \label{nns}
257: \end{figure}
258:
259: \subsection{Level velocities}
260:
261: Having seen the effect of $r$ in the spectral spacing, we further investigate
262: on dynamics of the quasienergies in parameter space i.e., level velocities.
263: To be specific, we study second moments of level velocities in different
264: classical regimes. We take $\alpha = 0.35$ so that $R$-symmetry is broken in
265: the quantum system, and the factor $\alpha$ is dropped out in following
266: expressions for the sake of convenience. Quasienergies $\phi_j \equiv
267: \phi_j(k,r)$ have scaled velocities:
268:
269: \begin{eqnarray}
270: x_j &=& \left({2\pi^2 r^2\over N}\right) {\partial\phi_j\over\partial k}
271: \nonumber \\[8pt]
272: &=& \left({2\pi^2 r^2\over N}\right) \left[ {1\over\hbar }
273: \langle\phi_j|\partial V/\partial k|\phi_j\rangle \right] \nonumber \\[8pt]
274: &=& \pi r\sum_n \cos (2\pi rn/N) {|\langle n|\phi_j\rangle |}^2
275: \end{eqnarray}
276:
277: \n and
278:
279: \begin{eqnarray}
280: y_j &=& \left({2\pi r^2\over Nk}\right) {\partial\phi_j\over \partial r}
281: \nonumber \\[8pt]
282: &=& \left({2\pi r^2\over Nk}\right) \left[ {1\over\hbar}
283: \langle\phi_j|\partial V/\partial r|\phi_j\rangle \right] \nonumber \\[8pt]
284: &=& -2\pi r\sum_n (n/N)\sin (2\pi rn/N) {|\langle n|\phi_j\rangle |}^2 -
285: \sum_n \cos (2\pi rn/N) {|\langle n|\phi_j\rangle |}^2 .
286: \end{eqnarray}
287:
288: \n Average velocities in semiclassical limit are
289:
290: \begin{eqnarray}
291: \langle x\rangle &=& {1\over N}\sum_j x_j = \sin(\pi r) \nonumber \\
292: \langle y\rangle &=& {1\over N}\sum_j y_j = \cos(\pi r) -
293: {2\sin(\pi r)\over\pi r} \, .
294: \end{eqnarray}
295:
296: Then the second moment of $x$ is given by
297:
298: \begin{eqnarray}
299: \langle x^2 \rangle &=& {1\over N} \sum_j x_j^2 \nonumber \\[8pt]
300: &=& {(\pi r)}^2 \left\{\sum_n \cos^2 (2\pi rn/N)
301: \left\langle {|\langle n|\phi_j\rangle|}^4\right\rangle \right .
302: \nonumber \\[8pt]
303: && ~~~~~~~~~~~ + \left . \sum_{n\neq n'} \cos (2\pi rn/N)
304: \cos (2\pi rn^\prime/N) \left\langle {|\langle n|\phi_j\rangle|}^2
305: {|\langle n'|\phi_j\rangle|}^2 \right\rangle \right\}
306: \label{x2}
307: \end{eqnarray}
308:
309: \n Assuming that spectral averaged eigenfunction components are independent
310: of specific position eigenvalues $n$, terms within the angle brackets can be
311: taken out of the sum. Then
312:
313: \begin{eqnarray}
314: \langle x^2\rangle &\sim& {(\pi r)}^2 \left\{\left[\left\langle
315: {|\langle n|\phi_j\rangle|}^4 \right\rangle - \left\langle
316: {|\langle n|\phi_j\rangle|}^2 {|\langle n'|\phi_j\rangle|}^2
317: \right\rangle \right] \sum_n \cos^2(2\pi rn/N) \right . \nonumber \\[8pt]
318: && ~~~~~~~~~~~+ \left . \left\langle {|\langle n|\phi_j\rangle|}^2
319: {|\langle n'|\phi_j\rangle|}^2 \right\rangle {\left[\sum_n
320: \cos (2\pi rn/N)\right]}^2 \right\} \, .
321: \end{eqnarray}
322:
323: \n In chaotic regimes, standard RMT results \cite{brody81}
324:
325: \begin{equation}
326: \begin{array}{rll}
327: \left\langle{|\langle n|\phi_j\rangle|}^4 \right\rangle &=&
328: 3{[N(N+2)]}^{-1} \simeq 3N^{-2} \nonumber \\[8pt]
329: \left\langle{|\langle n|\phi_j\rangle|}^2{|\langle n'|\phi_j\rangle|}^2
330: \right\rangle &=& {[N(N+2)]}^{-1} \simeq N^{-2}
331: \end{array}
332: \label{rmt}
333: \end{equation}
334:
335: \n which correspond to Gaussian orthogonal ensemble are applicable here as
336: well. It should be noted that application of RMT results essentially adopt
337: the assumption made above. Replacing the sum by integration in semiclassical
338: limit we arrive to
339:
340: \begin{equation}
341: {\langle x^2\rangle }_{\hbox{\tiny RMT}} = {{(\pi r)}^2\over N}
342: \left[1 + {\sin (2\pi r)\over 2\pi r}\right] + {\langle x\rangle}^2 \, .
343: \label{x2_rmt}
344: \end{equation}
345:
346: \n Similarly the second moment of $y$ is
347:
348: \begin{eqnarray}
349: \langle y^2 \rangle &=& {1\over N} \sum_j y_j^2 \\[8pt]
350: &=& \sum_n \{{\left[2\pi r(n/N)
351: \sin (2\pi rn/N)\right]}^2 + \cos^2 (2\pi rn/N) \nonumber \\[8pt]
352: && + \; 4\pi r(n/N)\sin(2\pi r n/N)\cos (2\pi rn/N)\}
353: \left\langle {|\langle n|\phi_j\rangle|}^4\right\rangle \nonumber \\[8pt]
354: && + \sum_{n\neq n^\prime} \{{(2\pi r/N)}^2nn^\prime\sin (2\pi rn/N)
355: \sin (2\pi rn^\prime /N) + \cos (2\pi rn/N) \cos (2\pi rn^\prime /N)
356: \nonumber \\[8pt]
357: && + \; 4\pi r(n/N)\sin (2\pi rn/N)\cos(2\pi rn^\prime /N)\} \left\langle
358: {|\langle n|\phi_j\rangle|}^2 {|\langle n'|\phi_j\rangle|}^2 \right\rangle
359: \label{y2}
360: \end{eqnarray}
361:
362: \n and the RMT prediction is
363:
364: \begin{equation}
365: {\langle y^2\rangle }_{\hbox{\tiny RMT}} = {1\over N} \left\{ 1 +
366: {{(\pi r)}^2\over 3} + {\sin (2\pi r)\over 2\pi r} \left[{5\over 2}-
367: {(\pi r)}^2\right] -{3\over 2}\cos (2\pi r)\right\} + {\langle y\rangle}^2\, .
368: \label{y2_rmt}
369: \end{equation}
370:
371: In chaotic regime, quantum states are such that the quantities in left hand
372: side of Eqn. (\ref{rmt}) fluctuate about the respective RMT values. These
373: fluctuations could lead to the failure of RMT in predicting the second moment.
374: In Ref. \cite{arul99} a semiclassical analysis on the systematic deviation
375: between the second moment and its RMT prediction has been made. Here we are
376: interested to see the validity of the RMT prediction in different classical
377: regimes. For that we define normalized deviations:
378:
379: \begin{equation}
380: \Delta_x = \left|{\langle x^2 \rangle - {\langle x^2\rangle}_{\hbox{\tiny RMT}}
381: \over\langle x^2\rangle}\right| \;\; ; \;\;
382: \Delta_y = \left|{\langle y^2 \rangle - {\langle y^2\rangle}_{\hbox{\tiny RMT}}
383: \over\langle y^2\rangle}\right|
384: \end{equation}
385:
386: \n and taking average of the two positive quantities as
387:
388: \begin{equation}
389: \Delta = {\Delta_x+\Delta_y\over 2} \, .
390: \end{equation}
391:
392: The deviation is calculated for various parameters and plotted in
393: Fig. \ref{dev1}. Let us first consider the data obtained for small values of
394: $k$ (0.3 and 0.1). As we see the deviation is nearly zero for $r\le 1/2$.
395: For further values of $r$, the deviation is found to be significantly large.
396: This is expected as the underlying classical dynamics is predominantly regular
397: or mixed and the RMT is not applicable. On the other hand, the data for large
398: $k$ (5 and 25) exhibit different behaviour. Though the classical phase space
399: is highly chaotic for these parameters, significant deviation observed for
400: various values of $r$ shows the failure of RMT prediction. However, it is
401: noticable from the four sets of observation that for $r \le 1/2$ the deviation
402: is nearly zero.
403:
404: \begin{figure}[h]
405: \centerline{\psfig{figure=rmt_dev1.ps,height=9cm,width=12cm}}
406: \caption{Deviation between level velocity second moments and RMT predictions.
407: In all the cases we have taken $\alpha = 0.35$ and $N=200$.}
408: \label{dev1}
409: \end{figure}
410:
411: In Fig. \ref{dev2}, the deviation is plotted for two cases by varying $k$.
412: For $r=1$ the deviation has strong fluctuations with $k$. The large
413: deviation for $k<5$ is due to the predominantly regular/mixed behaviour of
414: the underlying classical system. We also observe significant deviation when
415: $k$ is close to integer multiples of $2\pi$. This may be attributed to the
416: presence of ``accelerator modes'' \cite{zas97}, which are small regular regions
417: embedded in the sea of chaotic phase space. On the other hand, for $r=0.5$
418: the deviation remains negligible irrespective of $k$ values. This shows that
419: RMT predicts second moment of level velocity in hyperbolic chaotic regime. From
420: the Eqn. (\ref{x2_rmt}), we see that $\langle x^2 \rangle_{\hbox{\tiny RMT}}
421: \sim 1/N$ or $\langle {(\partial\phi/\partial k)}^2 \rangle_{\hbox{\tiny RMT}}
422: \sim N \sim 1/\hbar$. We obtain the same result for the velocity with respect
423: to $r$ as well.
424:
425: \begin{figure}
426: \centerline{\psfig{figure=rmt_dev2.ps,height=9cm,width=12cm}}
427: \caption{Deviation between level velocity second moments and RMT predictions.
428: We have taken $\alpha = 0.35$ and $N=200$.}
429: \label{dev2}
430: \end{figure}
431:
432: \subsection{Sawtooth map}
433:
434: In this section we study another quantum model whose classical counterpart
435: has hyperbolic regime. Consider a free particle that is subjected to a time
436: periodic impulsive potential $V(q)=-\lambda q^2/2;V(q+1)=V(q)$. Kick to kick
437: dynamics of such a particle is described by the sawtooth map \cite{cary81}:
438:
439: \begin{equation}
440: \begin{array}{lll}
441: p_{j+1} &=& p_j + \lambda q_j \\ q_{j+1} &=& q_j + p_{j+1}
442: \end{array}
443: \end{equation}
444:
445: \n defined on a unit torus. This map is stable for $-4 < \lambda < 0$ and
446: unstable (or hyperbolic) otherwise. Quantized version of this map can be
447: obtained, as usual, upon introducing periodic boundaries in $q$ and $p$.
448: Details of quantized sawtooth map are given elsewhere \cite{arul95}. Note
449: that here also the earlier symmetry arguments hold. From the corresponding
450: quasienergies and quasienergy states, we define the scaled level velocity as
451:
452: \begin{equation}
453: z_j = \left({-1\over N\pi}\right){d\phi_j\over d\lambda}
454: = -2\left\langle\phi_j\left|{dV\over d\lambda}\right|\phi_j\right\rangle
455: = \sum_n {(n/N)}^2 {|\langle n|\phi_j\rangle|}^2
456: \end{equation}
457:
458: \n with average $\langle z\rangle = 1/12$. Then the second moment is given by
459:
460: \begin{equation}
461: \langle z^2\rangle = {1\over N} \sum_j z_j^2 = \sum_n {(n/N)}^4
462: \left\langle{|\langle n|\phi_j\rangle|}^4\right\rangle + \sum_{n\neq n^\prime}
463: {(n/N)}^2 {(n^\prime/N)}^2 \left\langle {|\langle n|\phi_j\rangle|}^2
464: {|\langle n^\prime|\phi_j\rangle|}^2 \right\rangle \, .
465: \label{z2}
466: \end{equation}
467:
468: \n Repeating the earlier procedures, we find the RMT prediction as
469:
470: \begin{equation}
471: {\langle z^2\rangle}_{\hbox{\tiny RMT}} = {1\over 40N} +
472: {\langle z\rangle}^2 \, .
473: \label{z2_rmt}
474: \end{equation}
475:
476: \n Defining normalized deviation as
477:
478: \begin{equation}
479: \Delta '= \left|{\langle z^2 \rangle - {\langle z^2\rangle}_{\hbox{\tiny RMT}}
480: \over\langle z^2\rangle}\right|
481: \end{equation}
482:
483: \n in Fig. \ref{saw} we have plotted the deviation for different $\lambda$
484: values. In the stable region $-4<\lambda<0$, where the RMT result is not
485: applicable, the deviation is large. Evidently, RMT predicts level velocity
486: second moment in hyperbolic regime. Eqn. (\ref{z2_rmt}) shows that $\langle
487: z^2 \rangle_{\hbox{\tiny RMT}} \sim 1/N$ or $\langle {(d\phi/d\lambda)}^2
488: \rangle_{\hbox{\tiny RMT}} \sim N \sim 1/\hbar$. Here also we see that second
489: moment is inversely proportional to the Planck constant for hyperbolic chaos.
490:
491: \begin{figure}
492: \centerline{\psfig{figure=saw_dev.ps,height=9cm,width=12cm}}
493: \caption{Deviation $\Delta '$ for the quantum sawtooth map with $\alpha =
494: 0.35$ and $N=200$.}
495: \label{saw}
496: \end{figure}
497:
498: \section {conclusion}
499:
500: In this paper we have introduced quantum version of a generalized standard
501: map. The classical system corresponds to the dynamics of a particle, trapped
502: in an 1D infinite square well potential, in presence of time-periodic kicks.
503: For different classical regimes, the second moment of quantum level velocity
504: is computed and compared with the RMT prediction. We have shown that while
505: the prediction fails in highly chaotic regime, the second moment is well
506: predicted by the RMT as $\sim 1/\hbar$ in hyperbolic chaotic regime. By
507: considering another example viz., quantum sawtooth map, the RMT prediction
508: of second moment in hyperbolic chaotic regime has been reinforced. We hope
509: the presented result would shed new lights to explore more on quantum
510: hyperbolic chaos. \\
511:
512: \n {\bf Acknowledgement} \\
513:
514: The author is greatful to Dr. A. Lakshminarayan for useful discussions.
515:
516: \begin{thebibliography}{99}
517: \bibitem{ott}
518: Edward Ott, Chaos in Dynamical Systems, Cambridge University Press, 1993.
519: \bibitem{lich}
520: A.J. Lichtenberg and M.A. Lieberman, Regular and Chaotic Dynamics,
521: Springer-Verlag, 1992.
522: \bibitem{sankar01}
523: R. Sankaranarayanan, A. Lakshminarayan, and V.B. Sheorey, Phys. Lett. A
524: 279 (2001) 313.
525: \bibitem{gaspard}
526: P. Gaspard, S.A. Rice, and K. Nakamura, Phys. Rev. Lett. 63 (1989) 930;
527: P. Gaspard, S.A. Rice, H.J. Mikeska and K. Nakamura, Phys. Rev. A 42 (1990)
528: 4015.
529: \bibitem{models}
530: D. Saher and F. Haake, Phys. Rev. A 44 (1991) 7481;
531: T. Takami and H. Hasegawa, Phys. Rev. Lett. 68 (1992) 419;
532: J. Zakrzewski and D. Delande, Phys. Rev. E 47 (1993) 1650.
533: \bibitem{arul99}
534: A. Lakshminarayan, N.R. Cerruti and S. Tomsovic, Phys. Rev. E 60 (1999) 3992.
535: \bibitem{ford91}
536: J. Ford, G. Mantica and G.H. Ristow, Physica D 50 (1991) 493.
537: \bibitem{jor}
538: D.W. Jordan and P. Smith, Nonlinear Ordinary Differential Equations,
539: Oxford, 1977.
540: \bibitem{arul97}
541: A. Lakshminarayan, Pramana J. Phys. 48 (1997) 517.
542: \bibitem{sar89}
543: M. Saraceno, Ann. Phys. (N.Y.) 190 (1989) 1.
544: \bibitem{brody81}
545: T.A. Brody, J. Flores, J.B. French, P.A. Mello, A. Pandey and S.S.M. Wong,
546: Rev. Mod. Phys. 53 (1981) 385.
547: \bibitem{zas97}
548: G.M. Zaslavsky, M. Edelman and B.A. Niyazov, Chaos 7 (1997) 159.
549: \bibitem{cary81}
550: John R. Cary and James D. Meiss, Phys. Rev. A 24 (1981) 2664.
551: \bibitem{arul95}
552: A. Lakshminarayan and N.L. Balazs, Chaos Solitons \& Fractals 5 (7)
553: (1995) 1169.
554: \end{thebibliography}
555:
556: \end{document}
557: