1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %%% (May 2003)
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4:
5: \documentclass{iopart}
6: \usepackage{epsfig}
7: \begin{document}
8:
9: \newcommand{\tbox}[1]{\mbox{\tiny #1}}
10: \newcommand{\half}{\mbox{\small $\frac{1}{2}$}}
11: \newcommand{\mbf}[1]{{\mathbf #1}}
12: \newcommand{\hide}[1]{}
13:
14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
15:
16: \title{Non-perturbative response: chaos versus disorder}
17:
18: %\shorttitle{Non-perturbative response}
19:
20: %\author{Doron Cohen\inst{1} and Tsampikos Kottos\inst{2}}
21: \author{Doron Cohen$^{1}$ and Tsampikos Kottos$^{2}$}
22:
23: \address{
24: \mbox{$^{1}$ Department of Physics, Ben-Gurion University, Beer-Sheva 84105, Israel} \\
25: \mbox{$^{2}$ Max-Planck-Institut f\"ur Str\"omungsforschung, Bunsenstra\ss e 10, D-37073 G\"ottingen, Germany}
26: }
27:
28:
29: %\pacs{03.65.-w}{Quantum mechanics}
30: %\pacs{05.45.Mt}{Quantum chaos}
31: %\pacs{73.23.-b}{Mesoscopic systems}
32:
33:
34: %\maketitle
35:
36:
37: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
38:
39: \begin{abstract}
40: Quantized chaotic systems are generically characterized by
41: two energy scales: the mean level spacing $\Delta$,
42: and the bandwidth $\Delta_b\propto\hbar$.
43: This implies that with respect to driving such systems
44: have an adiabatic, a perturbative, and a non-perturbative regimes.
45: A ``strong" non-linearity in the response,
46: due to a quantal non-perturbative effect,
47: is found for {\em disordered} systems that
48: are described by random matrix theory models.
49: Is there a similar effect for quantized {\em chaotic} systems?
50: Theoretical arguments cannot exclude the existence
51: of an analogous ``weak" version of the above mentioned
52: non-linear response effect, but our numerics demonstrates
53: an unexpected degree of semiclassical correspondence.
54: \end{abstract}
55:
56:
57: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
58:
59: \section{Introduction}
60:
61:
62: \subsection{The two energy scales in Quantum chaos}
63:
64: The name ``Quantum Mechanics" is associated with the idea
65: that the energy is quantized. For generic (chaotic) system
66: the mean level spacing is $\Delta\propto\hbar^d$,
67: where $d$ is the dimensionality of the system.
68: However, one should recognize that there is a second
69: energy scale $\Delta_b\propto\hbar$ which is introduced
70: by Quantum Mechanics. This $\hbar$ energy scale is
71: related to the chaos implied decay of the classical correlations.
72: It is known in the literature as the ``non-universal" energy
73: scale \cite{berry}, or as the ``bandwidth" \cite{mario}.
74: The dimensionless bandwidth is defined as $b=\Delta_b/\Delta$.
75: For reasonably small $\hbar$ one has $b\gg1$.
76:
77:
78: This observation, of having two energy scales, has motivated
79: the study of Wigner model \cite{wigner} within
80: the framework of random matrix theory (RMT). This model,
81: which is defined in terms of $\Delta$ and $\Delta_b$,
82: is totally artificial: it does not possess any classical limit.
83: Still note that it can be re-interpreted as a model for the motion
84: of a particle in a quasi one-dimensional {\em disordered} lattice~\cite{quasi}.
85:
86:
87: The main focus of Quantum chaos studies (so far) was on
88: issues of spectral statistics \cite{books}. In that context it turns
89: out that the sub-$\hbar$ statistical features of the
90: energy spectrum are ``universal", and obey the predictions
91: of random matrix theory. Non universal (system specific)
92: features are reflected only in the large scale properties
93: of the spectrum (analyzing energy intervals $>\Delta_b$).
94:
95:
96:
97: %%%%%%%%%%%%%%%%%%%%
98: %%%%%%%%%%%%%%%%%%%%
99: % driving, regimes
100:
101: \subsection{Regimes in the theory of driven systems}
102:
103: In recent years we have made some progress in understanding
104: the theory of driven quantized chaotic systems \cite{crs,rsp,revw}.
105: Driven systems are described by Hamiltonian ${\cal H}(Q,P,x(t))$,
106: where~$x(t)$ is a time dependent parameter.
107: Such systems are of interest in mesoscopic physics (quantum dots),
108: as well as in nuclear, atomic and molecular physics.
109: The time dependent parameter $x(t)$ may have the significance
110: of external electric field or magnetic flux or gate voltage.
111: %
112: Linear driving $x(t)=Vt$ is characterized by one parameter ($V$),
113: while more generally a periodic driving $x(t)=A f(t)$ is characterized
114: by both amplitude ($A$) and frequency ($\Omega$).
115: Due to the time dependence of $x(t)$, the energy of the system
116: is not a constant of motion. Rather the system makes ``transitions"
117: between energy levels, and therefore absorbs energy.
118:
119:
120:
121: The main object of our studies is the energy spreading
122: kernel $P_t(n|m)$. Regarded as a function of the level
123: index $n$, it gives the energy distribution after time $t$,
124: where $m$ is the initial level.
125: %
126: Having two quantal energy scales \mbox{($\Delta$, $\Delta_b$)}
127: implies the existence of different quantum-mechanical (QM)
128: $V$~{\em regimes} \cite{crs}, or more generally
129: $(A,\Omega)$~{\em regimes} \cite{rsp}, in the theory of $P_t(n|m)$.
130: %
131: Most familiar is the QM adiabatic regime (very very small~$V$),
132: whose existence is associated with having finite~$\Delta$.
133: Outside of the adiabatic regime we are used to the idea
134: that there is a perturbative regime, where the Fermi golden rule
135: applies, leading to a Markovian picture of the dynamics,
136: with well defined transition rates between levels.
137: %
138: Less familiar \cite{crs,rsp} is the QM non-perturbative regime
139: ($V$ is quantum mechanically large, but still classically small)
140: whose existence is associated with the energy scale~$\Delta_b$.
141: %
142: %
143: As implied by the terminology, in the QM non-perturbative regime
144: perturbation theory (to any order) is not a valid tool for
145: the analysis of the energy spreading.
146: Consequently the Fermi golden rule picture of the dynamics
147: does not apply there.
148:
149:
150: %%%%%%%%%%%%%%%%%%
151: %%%%%%%%%%%%%%%%%%
152: % LRT, response
153:
154: \subsection{Linear response theory}
155:
156: Of special importance (see discussion below) is the
157: theory for the variance $\delta E(t)^2=\sum_n P_t(n|m) (E_n-E_m)^2$
158: of the energy spreading. Having $\delta E(t) \propto A$
159: means {\em linear response}. If $\delta E(t)/A$ depends on $A$
160: we call it ``non-linear response". In this paragraph we explain
161: that linear response theory (LRT) is based on the ``LRT formula"
162: %
163: \begin{eqnarray} \label{e1}
164: \delta E(t)^2 \ = \ A^2 \times
165: \int_{-\infty}^{\infty} \frac{d\omega}{2\pi}
166: \tilde{F}_t(\omega) \tilde{C}(\omega)
167: \end{eqnarray}
168: %
169: Two spectral functions are involved: One is the power
170: spectrum $\tilde{C}(\omega)$ of the fluctuations,
171: and the other $\tilde{F}_t(\omega)$ is the spectral
172: content of the driving. See Eq.(\ref{e4}) and Eq.(\ref{e5})
173: for exact definitions.
174: %
175: %
176: %
177: A special case of Eq.(\ref{e1}) is the sudden limit
178: ($V=\infty$) for which $f(t)$ is a step function,
179: hence $F_t(\omega)=1$, and accordingly
180: %
181: \begin{eqnarray} \label{e2}
182: \delta E \ = \ \sqrt{C(0)} \times A
183: \hspace*{3cm} \mbox{\small [``sudden" case]}
184: \end{eqnarray}
185: %
186: Another special case is the response for
187: persistent (either linear or periodic) driving.
188: In such case the long time limit of
189: $F_t(\omega)$ is linear in time
190: [e.g. for linear driving ($f(t)=t$) we get
191: $F_t(\omega) = t \times 2\pi\delta(\omega)$].
192: This implies diffusive behavior:
193: %
194: \begin{eqnarray} \label{e3}
195: \delta E(t) = \sqrt{2 D_E t}
196: \hspace*{3cm} \mbox{\small [``Kubo" case]}
197: \end{eqnarray}
198: %
199: In the latter case the expression for $D_E\propto A^2$
200: is known as Kubo (or Kubo-Greenwood) formula,
201: leading to a fluctuation-dissipation relation \cite{revw}.
202:
203:
204:
205:
206: The LRT formula Eq.(\ref{e1}) has a simple classical
207: derivation\cite{crs}.
208: From now on it goes without saying that we
209: assume that the {\em classical} conditions
210: on $(A,\Omega)$ for the validity of Eq.(\ref{e1})
211: are satisfied (no $\hbar$ involved in such conditions).
212: The question is {\em what happens to the validity of
213: LRT once we ``quantize" the system}.
214: Can we trust Eq.(\ref{e1}) for any $(A,\Omega)$?
215: Or maybe we can trust it only in a restricted regime?
216: %
217: In previous publications\cite{crs,rsp,revw},
218: we were able to argue the following: \\
219: %
220: \begin{minipage}{\hsize}
221: \vspace*{0.2cm}
222: \begin{itemize}
223: \setlength{\itemsep}{0cm}
224: \item[(A)]
225: The LRT formula can be trusted
226: in the perturbative regime, with the exclusion
227: of the adiabatic regime.
228: \item[(B)]
229: In the sudden limit the LRT formula can
230: be trusted also in the non-perturbative regime.
231: \item[(C)]
232: In general the LRT formula cannot be
233: trusted in the non-perturbative regime.
234: \item[(D)]
235: The LRT formula can be trusted deep in the non-perturbative
236: regime, provided the system has a classical limit.
237: \end{itemize}
238: \vspace*{0.0cm}
239: \end{minipage}
240: %
241: For a system that does not have a classical limit
242: (Wigner model) we were able to demonstrate \cite{rsp}
243: that LRT fails in the non-perturbative regime.
244: Namely, for Wigner model the response $\delta E(t)/A$
245: becomes $A$ dependent for large $A$, meaning that
246: the response is non-linear.
247: Hence the statement in item (C) above has been established.
248: We had argued that the observed non-linear response
249: is the result of a quantal non-perturbative effect.
250: %
251: {\em Do we have a similar type of non-linear response
252: in case of quantized chaotic systems?}
253: The statement in item (D) above seems to suggest that
254: the observation of such non-linearity is not likely.
255: Still, we argue below that this does not exclude
256: the possibility of observing a ``weak" non-linearity.
257:
258:
259:
260: %%%%%%%%%%%%%%%%%%%%
261: %%%%%%%%%%%%%%%%%%%%
262: % linear response?
263:
264: \section{Perturbation theory and linear response}
265:
266: The immediate (naive) tendency is to regard LRT as the outcome
267: of first order perturbation theory (FOPT).
268: In fact the regimes of validity of FOPT and of LRT
269: do not coincide. On the one hand we have the
270: adiabatic regime where FOPT is valid as a leading order
271: description, but not for response calculation
272: (see further details below).
273: On the other hand, the validity of Eq.(\ref{e1})
274: goes well beyond FOPT. This leads to the (correct) identification
275: \cite{crs,rsp} of what we call the ``perturbative regime".
276: The border of this regime [in $(A,\Omega)$ space]
277: is determined by the energy scale $\Delta_b$,
278: while $\Delta$ is not involved.
279: Outside of the perturbative regime we cannot
280: trust the LRT formula. However, as we further explain below,
281: the fact that Eq.(\ref{e1}) is not valid in the non-perturbative regime,
282: does not imply that it {\em fails} there.
283:
284:
285: We stress again that one should distinguish
286: between ``non-perturbative response" and ``non-linear response".
287: These are not synonyms. As we explain in the next paragraph,
288: the adiabatic regime is ``perturbative" but ``non-linear",
289: while the semiclassical limit is ``non-perturbative" but ``linear".
290:
291:
292: In the {\em adiabatic regime}, FOPT implies zero probability
293: to make a transitions to other levels. Therefore, to the
294: extend that we can trust the adiabatic approximation,
295: all the probability remains concentrated in the initial level.
296: Thus, in the adiabatic regime, Eq.(\ref{e1}) is not a
297: valid formula: It is essential to use higher orders
298: of perturbation theory, and possibly non-perturbative
299: corrections (Landau-Zener \cite{wilk}), in order to
300: calculate the response. Still, FOPT provides a meaningful
301: leading order description of the dynamics (i.e. having no transitions),
302: and therefore we do not regard the adiabatic
303: non-linear regime as ``non-perturbative".
304:
305: In the {\em non-perturbative regime} the evolution
306: of $P_t(n|m)$ cannot be extracted from perturbation theory:
307: not in leading order, neither in any order.
308: Still it does not necessarily imply a non-linear response.
309: On the contrary: the semiclassical limit is contained in
310: the (deep) non-perturbative regime\cite{rsp}.
311: There, the LRT formula Eq.(\ref{e1}) is in fact valid. But its
312: validity is {\em not} a consequence of perturbation theory,
313: but rather the consequence of {\em quantal-classical correspondence}.
314:
315:
316:
317: %%%%%%%%%%%%%%%%
318: %%%%%%%%%%%%%%%%
319: % weak effect?
320:
321: \section{The quest for non perturbative response}
322:
323: As stated above, an effect of non-linear response
324: due to the {\em quantum mechanical} non-perturbative nature
325: of the dynamics, has been demonstrated so far only
326: for Wigner model \cite{rsp}.
327: There, its existence is related to the {\em disordered} RMT nature
328: of the model (see discussion below).
329: Semiclassical correspondence considerations
330: seem to exclude the manifestation of this disorder-related
331: non-linearity in case of quantized {\em chaotic} systems.
332: In this Letter we explain that this does not exclude
333: the possibility of having a ``weak" version of this effect.
334: We also report the results of an intense numerical effort
335: aimed in finding a ``weak" non-linearity in the case of a simple
336: low-dimensional quantized chaotic systems. To our surprise,
337: an unexpected degree of semiclassical correspondence is observed.
338:
339:
340:
341: It is appropriate here to clarify the notions of ``weak"
342: and ``strong" effects. In the literature regarding the
343: dynamics in disordered lattices one distinguishes between ``weak"
344: and ``strong" localization effects. The former term implies
345: that while the leading behavior is classical (diffusion),
346: there are ``on top" quantum mechanical corrections
347: (enhanced return probability). In contrast to that the
348: term ``strong" implies that the classical description fails
349: even as a leading order description. In the literature
350: regarding quantum chaos we have the effect of ``scarring",
351: which should be regarded as ``weak" effect.
352: "Strong" quantum mechanical effects (e.g. dynamical localization
353: in 1D kicked systems \cite{qkr}) are non-generic: The leading order
354: behavior of generic quantized chaotic systems is typically classical.
355: In the present context of driven systems, we use the
356: terms ``weak" and ``strong" in the same sense:
357: The adjective ``weak" is associated with the (conjectured)
358: non-linear response of quantized driven chaotic systems,
359: while the adjective ``strong" is associated
360: with the (established) non-linear response
361: in the corresponding RMT (Wigner) model.
362:
363:
364: %%%%%%%%%%%%%%%%%%%%%%%%
365: %%%%%%%%%%%%%%%%%%%%%%%%
366: % the numerics
367:
368: \section{The numerical findings}
369:
370: How do we detect non-linear response?
371: The most straightforward way is to fix the pulse shape
372: $f(t)$ and to plot $\delta E/A$ versus $A$.
373: A deviation from constant value means
374: "non-linear" response. The simulations below
375: are done for a quantized chaotic system.
376: Due to obvious numerical limitations
377: we will consider the response to one-pulse driving,
378: rather than to persistent (periodic) driving.
379: %
380: {\em The central question is whether the observed non-linear
381: effect is of semiclassical origin, or of novel
382: quantum mechanical origin}. We deal with this issue below.
383:
384:
385: In our numerical simulations (Fig.1) we have considered
386: a particle in a two dimensional ($d=2$) anharmonic well.
387: This model (with deformation parameter $x=\mbox{const}$)
388: is defined in \cite{lds,wpk}.
389: In the energy region of interest ($E\sim3$), the classical
390: motion inside the two dimensional well (2DW) is chaotic,
391: with characteristic correlation time $\tau_{cl}\sim 1$.
392: %
393: For the following presentation it is enough to say that
394: the quantum mechanical Hamiltonian is represented
395: by a matrix ${\cal H}=\mbf{E}+x(t)\mbf{B}$,
396: where $\mbf{E}$ is a diagonal matrix with mean
397: level spacing $\Delta\approx 4.3\times\hbar^{d}$,
398: and $\mbf{B}$ is a banded matrix.
399: The bandwidth (in energy units)
400: is $\Delta_b=2\pi\hbar/\tau_{cl}$.
401: The bandprofile (see Fig.2 of \cite{lds}) is described
402: by a spectral function which is defined as follows:
403: %
404: \begin{eqnarray} \label{e4}
405: \tilde{C}(\omega) \ = \
406: \sum_{n(\ne m)}
407: \left|\mbf{B}_{nm}\right|^2
408: \ 2\pi\delta\left(\omega-\frac{E_n{-}E_m}{\hbar}\right)
409: \end{eqnarray}
410: %
411: with implicit average over the reference state $m$.
412: The bandprofile, as defined above, can be determined
413: from the the classical dynamics. This means
414: that $\tilde{C}(\omega)\approx\tilde{C}^{cl}(\omega)$
415: where $\tilde{C}^{cl}(\omega)$ is the Fourier transform
416: of a classical correlation function $C^{cl}(\tau)$.
417: The $\hbar$ dependence of $\tilde{C}(\omega)$ is relatively weak.
418:
419:
420:
421: The driving pulse in our numerical simulations
422: has a rectangular shape. This means $f(0)=f(T)=0$
423: and $f(0<t<T)=A$, where $T=0.375$.
424: The spectral content of the driving is defined as:
425: %
426: \begin{eqnarray} \label{e5}
427: \tilde{F}_t(\omega) \ = \
428: \left| \int_0^t \dot{f}(t') \mbox{e}^{i\omega t'} dt' \right|^2
429: \end{eqnarray}
430: %
431: The spectral content of the driving after a rectangular pulse
432: is $F_t(\omega) = |1-\mbox{e}^{i\omega T}|^2$.
433: %
434: We have also made simulations (not presented) with a driving
435: scheme that involves a positive pulse $+A$ followed by
436: a negative pulse $-A$, with the intention of considering
437: eventually a persistent (multi cycle) periodic driving.
438: However, we have realized that all the relevant
439: physics is observed already in the single pulse case.
440: %
441: Note that the regime diagram for either
442: linear or (as in the following simulations)
443: rectangular driving pulse, is greatly simplified,
444: because the driving is characterized by only one parameter
445: ($V$ in the former case, $A$ in the latter case).
446:
447:
448: Let us look carefully at the results of the 2DW simulations (Fig.1).
449: For small $A$ we see as expected ``linear response"
450: meaning $\delta E /A = \mbox{const}$, as implied by Eq.(\ref{e1}).
451: Note that the ``constant" has a weak $\hbar$ dependence (a $10\%$ effect).
452: This is due to the above mentioned weak dependence of $\tilde{C}(\omega)$
453: on $\hbar$. So this quantum-mechanical effect is quite
454: trivial, and has a simple explanation within LRT.
455: Now let us look what happens for large $A$.
456: We clearly see a $2\%$~deviation from linear response.
457: In what follows we discuss the reason for this non-linear effect.
458:
459:
460: For sake of comparison we also perform simulations with an
461: effective RMT model that corresponds to the 2DW model Hamiltonian.
462: The effective RMT model is obtained by randomizing the signs
463: of the off-diagonal elements of the $\mbf{B}$ matrix.
464: The effective RMT Hamiltonian has the {\em same} bandprofile
465: $\tilde{C}(\omega)$ as the original (2DW) Hamiltonian.
466: Therefore, as far as LRT Eq.(\ref{e1}) is concerned, the response should
467: be the {\em same}. Still we see that at the same $A$ regime,
468: as in the case of the 2DW simulations, we have deviation
469: from linear response. However, this non-linear deviation
470: is much much stronger.
471:
472:
473: Looking at the curves of Fig.1, it is very tempting to regard the observed
474: non-linear $2\%$ effect in the 2DW simulations as a ``weak" version
475: of the ``strong" effect which is observed in the corresponding RMT simulations.
476: However, the careful analysis below indicates that apparently
477: this is not the case.
478:
479:
480: %%%%%%%%%%%%%%%%
481: %%%%%%%%%%%%%%%%
482: % validity of LRT
483:
484: \section{Discussion and analaysis}
485:
486: In analyzing the validity of the LRT formula,
487: it is instructive to consider first the
488: sudden limit Eq.(\ref{e1}). This limit has been
489: studied in \cite{lds}. The spreading profile
490: $P(n|m)$, after the sudden change in $x$,
491: depends on the amplitude $A$ of the
492: perturbation. [We omit the time index $t$,
493: which is of no relevance in this limit].
494: %
495: The perturbative regime is $A<A_{\tbox{prt}}$,
496: where $A_{\tbox{prt}}=2\pi\hbar/(\tau_{cl}\sqrt{C(0)})$.
497: For the 2DW simulations $A_{\tbox{prt}} = 5.3\times\hbar$.
498: In the perturbative regime $P(n|m)$
499: has a core-tail structure
500: (the generalization of Wigner Lorentzian),
501: and the variance $\delta E^2$ is determined
502: by the first order tail component of the energy distribution.
503: For $A>A_{\tbox{prt}}$ the spreading profile
504: $P(n|m)$ becomes non-perturbative.
505: This means that the perturbative tail (if survives)
506: is no longer the predominant feature. Thus the variance
507: is determined by the non-perturbative component
508: (the ``core") of the energy distribution.
509: %
510: The remarkable fact is that
511: the crossover from the perturbative $A$ regime
512: to the non-perturbative $A$ regime is {\em not}
513: reflected in the variance (see Fig.5 of \cite{lds}).
514: The agreement with Eq.(\ref{e2}) is perfect.
515: Taking into account the ``dramatic" differences
516: in the appearance of $P(n|m)$, this looks
517: quite surprising. In fact (see Sec.12 of \cite{lds})
518: there is a simple proof \cite{felix} that
519: Eq.(\ref{e2}) remains exact beyond any order
520: of perturbation theory, which means that it
521: is exact even in the non-perturbative regime
522: where perturbation theory is not applicable.
523:
524:
525:
526: We turn back to our simulations,
527: where we have a rectangular pulse
528: (rather than step function). Here
529: the sudden limit does not apply,
530: and the dynamics within the time
531: interval $0<t<T$ should be taken
532: into account.
533: %
534: If we take an eigenstate of $\mbf{E}$ and propagate
535: it using $\mbf{E}+A\mbf{B}$, then we get in the classical case
536: {\em ballistic} spreading followed by saturation.
537: ["Eigenstate" in the classical case means microcanonical
538: distribution]. This is true for any $A$.
539: %
540: Quantum mechanically we observe in the 2DW model
541: simulations a similar ballistic behavior \cite{wpk},
542: whereas in the corresponding RMT model there is
543: an intermediate stage of {\em diffusion} \cite{wpk}.
544: This diffusion is of non-perturbative nature,
545: and it is related to the ``disorder" which is artificially
546: introduced via the sign-randomization procedure.
547: The strong non-linear response effect \cite{rsp}
548: is a consequence of this diffusion.
549:
550:
551:
552: Coming back to the 2DW model, we realize that there is
553: no ``disorder" build into the model, and therefore
554: no diffusion. Still, looking at Fig.1,
555: it is tempting to interpret the observed
556: $2\%$~non-linear deviation as a ``weak" version of the
557: strong non-linear effect.
558: Moreover, regarded as such, it vanishes, as expected,
559: in the deep non-perturbative regime,
560: which had been argued on the basis of
561: semiclassical correspondence considerations\cite{rsp}.
562:
563:
564: In order to properly determine whether the dips
565: in Fig.1 are the result of the QM non-perturbative
566: nature of the dynamics, we have rescaled the vertical axis,
567: and plotted the response once (Fig.2a) versus $A$,
568: and once (Fig.2b) versus $A/\hbar$.
569: On the basis of the scaling we see that
570: the strong non-linear response in the RMT
571: case is indeed the result of the quantal ($\hbar$-dependent)
572: non-perturbative effect. In contrary to that
573: the $\hbar$-independent scaling in the 2DW case,
574: indicates that the non-linear deviation there is of ``semiclassical"
575: rather than ``quantal non-perturbative" nature.
576:
577:
578:
579: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
580: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
581:
582: \section{Conclusion}
583:
584: Theoretical arguments cannot exclude
585: the existence of a ``weak" non-linearity
586: in the response of a driven quantized chaotic system,
587: which is due to a quantum mechanical non-perturbative effect.
588: But our careful numerics, regarding a simple
589: low-dimensional system, demonstrates an unexpected
590: degree of semiclassical correspondence.
591: %
592: %
593: Our findings should be regarded as the outcome
594: of an ongoing quest, which has not ended,
595: that is aimed in finding novel quantum mechanical
596: deviations from linear response theory.
597:
598:
599: \ \\
600: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
601: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
602:
603: This work was supported by a grant from the GIF,
604: the German-Israeli Foundation for Scientific Research and Development,
605: and by the Israel Science Foundation (grant No.11/02).
606:
607:
608: \newpage
609: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
610: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
611: \vspace*{0.4cm}
612:
613: \begin{thebibliography}{99}
614:
615: \bibitem{berry}
616: M.V. Berry in {\it Chaos and Quantum Systems},
617: ed. M.-J.~Giannoni, A.~Voros, J.~Zinn-Justin
618: (Elsevier, Amsterdam, 1991).
619:
620: \bibitem{mario} M. Feingold and A. Peres, Phys. Rev. A {\bf 34} 591, (1986); M.
621: Feingold, D. Leitner, M. Wilkinson, Phys. Rev. Lett. {\bf 66}, 986 (1991); M.
622: Feingold, A. Gioletta, F. M. Izrailev, L. Molinari, ibid. {\bf 70}, 2936 (1993);
623: M. Wilkinson, M. Feingold, D. Leitner, J. Phys. A {\bf 24}, 175 (1991).
624: Y.V.Fyodorov and A.D.Mirlin Phys.Rev.Lett. {\bf 67}, 2405 (1991).
625:
626: \bibitem{wigner} E. Wigner, Ann. Math {\bf 62} 548 (1955); {\bf 65} 203 (1957);
627: V.V. Flambaum, A.A. Gribakina, G.F. Gribakin and M.G. Kozlov, Phys. Rev. A {\bf 50}
628: 267 (1994); G. Casati, B.V. Chirikov, I. Guarneri, F.M. Izrailev, Phys. Rev. E
629: {\bf 48}, R1613 (1993); \ Phys. Lett. A {\bf 223}, 430 (1996).
630: D. Cohen, F. Izrailev and T. Kottos, Phys. Rev. Lett., {\bf 84} 2052 (2000).
631: Y.V. Fyodorov, O.A.Chubykalo, F.M.Izrailev, and G.Casati
632: Phys.Rev.Lett. {\bf 76}, 1603 (1996).
633:
634:
635: \bibitem{quasi}
636: Y.V. Fyodorov and A.D. Mirlin,
637: Int. J. Mod Phys. B {\bf 8} (27), 3795 (1994).
638:
639:
640: \bibitem{books}
641: F. Haake, ``Quantum Signatures of Chaos" (Springer 2000).
642: H.J. Stockmann, ``Quantum Chaos: An Introduction" (Cambridge Univ Pr 1999).
643:
644:
645:
646: \bibitem{crs}
647: D. Cohen, Phys. Rev. Lett. {\bf 82}, 4951 (1999);
648: Annals of Physics {\bf 283}, 175 (2000).
649: For an elementary classical and quantum mechanical derivations
650: of Eq.(\ref{e1}), using either fixed or adiabatic basis,
651: see Appendices C and D of D. Cohen and D.A. Wisniacki,
652: nlin.CD/0204025, Phys. Rev. E (February 2003).
653:
654:
655: \bibitem{rsp}
656: D. Cohen and T. Kottos, Phys. Rev. Lett. {\bf 85}, 4839 (2000).
657:
658:
659: \bibitem{revw} For an updated review see D. Cohen
660: in Proceedings of the 38th Karpacz Winter School
661: of Theoretical Physics {\em Dynamics of Dissipation},
662: Ed. P. Garbaczewski and R. Olkiewicz (Springer, 2002,
663: http://link.springer.de/link/service/series/2669/tocs/t2597.htm).
664: See also D.~Cohen in
665: {\em Proceedings of the International School of Physics Enrico Fermi Course CXLIII},
666: Edited by G. Casati, I. Guarneri and U. Smilansky, IOS Press, Amsterdam (2000).
667:
668:
669: \bibitem{wilk}
670: M. Wilkinson, J. Phys. A {\bf 21}, 4021 (1988);
671: J. Phys. A {\bf 20}, 2415 (1987).
672:
673:
674: \bibitem{lds}
675: D. Cohen and T. Kottos, Phys. Rev. E {\bf 63}, 36203 (2001).
676:
677:
678: \bibitem{wpk}
679: T. Kottos and D. Cohen, Phys. Rev. E {\bf 64}, R-065202 (2001).
680:
681: \bibitem{qkr}
682: For review and references see
683: S. Fishman in ``Quantum Chaos",
684: {\em Proceedings of the International School
685: of Physics ``Enrico Fermi", Course CXIX},
686: Ed. G. Casati, I. Guarneri and U. Smilansky
687: (North Holland 1991).
688:
689:
690: \bibitem{felix}
691: We thank F.M. Izrailev for pointing out the
692: existence of this proof in \cite{wigner}.
693:
694:
695: \end{thebibliography}
696: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
697: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
698:
699: \newpage
700:
701: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
702: %\begin{figure}[h]
703: \centerline{\epsfig{figure=lrt_fig1, width=0.7\hsize}}
704: %\caption
705: {\footnotesize FIG.1:
706: The response $\delta E/A$ as a result of a rectangular
707: pulse ($T=0.375$). Deviation from $\delta E/A=\mbox{const}$
708: implies non-linear response. All the data are averaged
709: over a number of different initial conditions. The simulations
710: are done with the 2DW Hamiltonian (circles), and also with the
711: associated RMT model (stars). See text for explanations.}
712: %\end{figure}
713: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
714:
715:
716: \ \\
717:
718: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
719: %\begin{figure}[h]
720: \centerline{\epsfig{figure=lrt_fig5, width=0.8\hsize}}
721: %\caption
722: {\footnotesize FIG.2: Scaled versions of Fig.1. The vertical
723: scaling is aimed in removing the weak $\hbar$ dependence
724: of the bandprofile. In (b) the horizontal scaling is aimed in
725: checking whether the deviation from linear response
726: is in fact a quantal non-perturbative effect.}
727: %\end{figure}
728: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
729:
730:
731:
732: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
733: \end{document}
734: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
735: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
736:
737:
738:
739:
740:
741:
742:
743:
744:
745:
746:
747:
748:
749:
750:
751:
752:
753: