1: %march 3, 04: fig 1 revised
2: %jan 8, 04 (RB)
3: %jan6, 04
4: %Dec1, 03
5: % Oct 22, 03
6: \documentclass[prl,aps,twocolumn,showpacs,floatfix]{revtex4}
7: \usepackage{graphicx}
8: \begin{document}
9: \title{Geometric Phase and Classical-Quantum Correspondence}
10: \author{Indubala I. Satija}
11: \affiliation{Department of Physics, George Mason University, Fairfax, VA 22030}
12: \author{Radha Balakrishnan}
13: \affiliation{The Institute of Mathematical Sciences, Chennai 600 113, India}
14: \date{\today}
15: \begin{abstract}
16: { We study the geometric phase factors underlying
17: the classical and the corresponding quantum dynamics of a
18: driven nonlinear oscillator
19: exhibiting chaotic dynamics. For the classical problem, we
20: compute the geometric phase factors associated with the
21: phase space trajectories using Frenet-Serret formulation. For the
22: corresponding quantum problem,
23: the geometric phase
24: associated with the time evolution of the wave function is computed.
25: Our studies suggest that the classical geometric phase may be related to the
26: the difference in the quantum geometric phases between two neighboring eigenstates.}
27: \end{abstract}
28: \pacs{02.40-k, 05-45.Ac }
29: \maketitle
30:
31: Since the discovery of Berry phase\cite{berry}, the study of geometric phases based on the common
32: mathematical theme of anholonomy,
33: has emerged in a variety of fields.\cite{wilczek}
34: The Berry phase is a path-dependent
35: geometric phase associated with the adiabatic time evolution of the wave function,
36: associated with circuits in parameter space.
37: This concept has been extended\cite{aharanov,MS,sam} to
38: non-adiabatic cases and also
39: to non-cyclic circuits, since
40: the phase acquired by the wave function in {\it any}
41: type of time evolution,
42: may have a component that is of purely geometric origin.
43: This phase is a gauge invariant quantity and is equal to the
44: difference between the
45: total phase and the dynamical phase acquired by the wave function.
46:
47: One of the open questions has been the classical limit of the Berry phase or its generalization
48: describing the geometric part of the phase of the quantum wave function.
49: For the special case of integrable Hamiltonian systems, described in terms of action-angle
50: variables, the so-called Hannay angle $\theta_h$\cite{hannay}
51: represents the semiclassical limit of the
52: Berry phase. Berry gave an explicit formula relating the classical angle, called the Hannay angle and the $n^{th}$
53: quantum eigenstate $\phi_n$ as,
54: $\theta_h = -\partial_n \phi_n$.
55: There have been some attempts to describe the classical limit of Berry phase for chaotic systems
56: where the effort has been focused on finding a generalization of the {\it 2-form}
57: within Wigner-Weyl formalism.\cite{RB}
58:
59: Viewing the Berry phase as an {\it anholonomy effect} underlying dynamical evolution described by
60: Schr\"{o}dinger equation,
61: we seek a classical analog of anholonomy, underlying the corresponding classical evolution
62: described by the Newton's equations of motion.
63: Here, we are concerned with the geometric phases
64: associated with periodic, quasiperiodic and
65: chaotic dynamics of a driven nonlinear oscillator.
66: We compute the geometric component of the phase of the wave function
67: using a kinematic formulation\cite{MS} of the Berry phase\cite{MS} as given by
68: Mukunda and Simon.
69: In the corresponding classical
70: problem, we find the anholonomy underlying nonplanar periodic phase space trajectories and then extend this formulation
71: to quasiperiodic and chaotic trajectories.
72: It should be emphasized that unlike in Berry phase, our circuits
73: are not in parameter space but are in phase space.
74: By treating a classical phase space trajectory as a space curve,
75: we show that a
76: geometric phase can be associated with every trajectory, by
77: using a Frenet-Serret (FS) formulation\cite{FS}.
78: The classical geometric phase is the integrated torsion
79: of the phase trajectory, and it bears a strong analogy to the geometric
80: phase factor associated with the wave function.
81:
82: To illustrate this correspondence, we begin with the Frenet-Serret equations,
83: describing the time evolution of the orthonormal FS triad, consisting of the tangent ${\bf T}$, the normal ${\bf N}$, and the
84: binormal ${\bf B}$ to the space curve ${\bf r}(t)$,
85: \begin{equation}
86: \dot{\bf T} = v\kappa {\bf N}\,,\,
87: \dot{\bf N} = - v\,\kappa \,{\bf T} + v\,\tau \,{\bf B}\,,\,
88: \dot{\bf B} = - v\,\tau \,{\bf N}\,,
89: \label{fseqns}
90: \end{equation}
91: where $\kappa$ and $\tau$ are respectively the curvature and
92: torsion of the curve
93: and $v = |\dot{\bf r}|$. Here, the overdot denotes derivative with
94: respect to time.
95: The above equations can be written as,
96: $\dot{\bf F}= \xi \times {\bf F}$, where
97: ${\bf F} =
98: {\bf T},\,{\bf N},$ or ${\bf B},$ and
99: $\xi =-v\,\kappa {\bf B} + v\tau {\bf T}$.
100: That is, the FS triad rotates around ${\bf T}$ and ${\bf B}$. One way to
101: quantify this rotation is to work in a frame in which ${\bf T}$ is
102: parallel transported, and measure the angle of rotation around the
103: tangent
104: ${\bf T}$. This will be given by the angle
105: $\phi_c(t)=\int_{0}^{t} \tau v dt^{\prime}$. Thus $\phi_c(t)$ is
106: the geometric phase characterizing
107: the anholonomy associated with the corresponding phase space orbit.
108:
109: If we define a complex vector ${\bf M}= ({\bf N}+i{\bf B})/\sqrt {2}$,
110: the {\it classical} geometric phase can be written as\\
111: $\phi_c(t)= i\int_{0}^{t}
112: {\bf M}^{*}\cdot \dot {\bf M}~ v dt^{\prime} $.\\
113: This expression has exactly the same form as the {\it quantum}
114: geometric phase found by Berry \cite{berry}, when the classical
115: vector ${\bf M}$ is replaced by a quantum eigenstate $\psi_n$.
116: Additionally, by mapping the closed phase space trajectory to a circuit on a unit sphere
117: traced by the tip of the tangent vector ( tangent indicatrix), $\phi_c$ can be shown to be the
118: solid angle subtended by this tangent indicatrix at the center of the
119: sphere.\cite{radha}
120: These results are valid also
121: for a non-periodic trajectory,
122: since it can always be closed using a geodesic on the sphere.\cite{MS}
123:
124: Motivated by this close analogy between the geometric phase in
125: a quantum system and the FS geometric phase,
126: we investigate any possible relationship between the two.
127: The underlying key question
128: is whether the classical
129: anholonomy is related to the corresponding quantum one.
130: Here we calculate the classical and the quantum geometric phases
131: for a periodically driven
132: nonlinear oscillator exhibiting complex dynamics.
133: The system under investigation is an "impact oscillator"\cite{bs},
134: the oscillator
135: that rebounds elastically whenever its displacement $x$ drops to zero.
136: For $ x > 0$,
137: the system is described by,
138: \begin{equation}
139: H = \frac{1}{2}p^2+\frac{1}{2} \omega_0^2 x^2 -f\cos(\omega t) x
140: \label{ham}
141: \end{equation}
142: The system is piecewise
143: linear, and the analytic solutions can be obtained for $x > 0$.
144: The discontinuity at the origin makes it essentially nonlinear.
145: Without loss of generality, we
146: choose the units of $t$ and $x$ such that $\omega=1$ and $f=1$.
147: The phase space trajectory of the dynamical system
148: can be viewed as a
149: space curve generated by the three-dimensional
150: vector ${\bf r}(t)= ( x, \dot{x}, \ddot{x})$
151: parameterized by the time $t$.
152:
153: As we discuss below, the impact oscillator exhibits very rich and
154: complex dynamics,
155: where periodic, quasiperiodic and chaotic dynamics coexist.
156: Our choice of this example was motivated by the fact that in addition to the
157: simplicity underlying the classical analysis of the oscillator,
158: the quantum
159: wave functions for the driven oscillator are known in terms of the classical solution as,
160: \cite{Kerner},
161: \begin{equation}
162: \psi(x,t)=\chi(x^{\prime}, t)\exp\frac{1}{\hbar}[{\dot{x_c}}(t) x^{\prime}+\int_{0}^{t}L(t^{\prime}) dt^{\prime}]
163: \end{equation}
164:
165: Here $x^{\prime} = x-x_c(t)$, with $x_c(t)$ being the solution of the classical equation of motion
166: and $L$ is the Lagrangian of the driven system. $\chi(x,t)$ is the wave function of the oscillator
167: in the absence of driving.
168: Note that the wave functions of the driven oscillator
169: are centered on the position of the classical forced oscillator.
170:
171: If we take $\chi$ to be an eigenstate of the undriven oscillator,
172: $\chi_n(x,t)= u_n(x) \exp-i(E_n t/\hbar)$, the eigenfunction
173: can be written in terms
174: of Hermite polynomials as
175: $u_n(x)=\exp-[\omega_0/(2\hbar) x^2] H_n(\sqrt{\omega_0/\hbar}x)$.
176: It should be noted that corresponding to the eigenstate,
177: $|\psi(x,t)|^2=|\chi_n(x^{\prime},t)|^2$.
178: Therefore, the center of the wave packet $x_c(t)$ obeys
179: classical equation
180: of motion, and
181: the shape of the wave packet ( the density distribution with respect to the center $x_c$)
182: is unaffected by the driving force.
183:
184: In view of the fact that we have the analytic solution for the
185: classical system for $x > 0$,
186: and {\it also} a closed form solution for the quantum wave function,
187: we can compute the classical
188: and the quantum geometric phases with extreme precision.
189: These will be presented below.
190:
191: Figures ~1 shows richness and complexity underlying the
192: classical dynamics of the oscillator
193: as we vary the initial energy (initial conditions) of the oscillator.
194: We see periodic orbits and quasiperiodic tori, in addition to
195: chaotic trajectories.
196:
197: All results are for a fixed $\omega_0=1.6$
198: which corresponds to the oscillator frequency of $3.2$,
199: in units of $\omega$, the frequency of the driver, putting
200: our analysis is close to the adiabatic regime.
201: %Furthermore, initial conditions used in our analysis are such that
202: %probability of inducing a transition ( due to driving) is rather low.
203:
204: Furthermore, we believe that the probability of inducing a
205: transition (due to driving) is rather small because the classical
206: energies of the particle under consideration here are
207: far below the threshold for the transition between two quantum
208: states $n_{1}$ and $n_{2}$ in units of $\hbar \omega_{0}$.
209:
210: \begin{figure}[htbp]
211: \includegraphics[width=3.5in]{fig1new.eps}
212: \leavevmode
213: \caption{ For $p_0=0$, the figure shows the
214: possible $x$ and $p$
215: values (once every period of the driver), as a function
216: of $x_0$, the initial position
217: of the oscillator.
218: For $x_0 \approx 1.0416$, there is a period-$11$ orbit.
219: For $x_0 > 1.046$,
220: we get invariant quasiperiodic tori or cantori as seen in
221: the lower figure. For $x_0 < 1.035$, almost all initial
222: conditions result in chaotic dynamics.}
223: \label{fig1}
224: \end{figure}
225:
226: Using the classical solution, the FS geometric phase $\phi_c$
227: can be computed as the integral of the torsion, given by
228: $\tau= {\bf r}_{t} \cdot({\bf r}_{tt}
229: \times {\bf r}_{ttt}) /|{\bf r}_{t}\times {\bf r}_{tt}|^2$
230: For the corresponding quantum problem,
231: we can obtain the geometric phase,
232: using kinematic formulation by Mukunda and Simon\cite{MS}.
233: For a given wave function $\psi(x,t)$, the quantum geometric phase $\phi_q$is a gauge invariant quantity is given by
234: $\phi_q(t)=arg\int[\psi^*(x,0) \psi(x,t)]dx-Im\int_{0}^{t}
235: \psi^*(x,t^{\prime})\partial_t\psi(x,t^{\prime})dt^{\prime}$.
236: It is clear that the first term describes total phase while the second term describes the dynamical phase
237: accumulated by the wave function in time $t$.
238: Note that this formulation does not require any circuits to define geometric phase.
239:
240: For the driven impact oscillator, if we substitute the solution for the wave function, given by equation (3).
241: , the geometric phase can be written in terms of the expectation values of $<x>$ and the classical solution $x_c$ ,
242: \begin{equation}
243: -\hbar \phi_{q,n}=\int_{0}^{t} [L(t^{\prime})-<x> \ddot{x_c}(t^{\prime})]dt^{\prime}+G(t)
244: \end{equation}
245: where $G(t)$ is given by,
246: $G(t)=(\dot{x_c}(t)x_c(t)-\dot{x_c}(0)x_c(0))+arg[\int[(u_n(x^{\prime}(0)u_n(x^{\prime}(t)
247: \exp\frac{ix}{\hbar}(\dot{x_c}(t)-\dot{x_c}(0))]dx$.
248: For periodic evolution, $G(t)=0$. Also, $G(t)=0$ if we consider $t$ values where the classical
249: particle is at the turning points. In view of this, we will confine our calculations of geometric phases
250: to only such values of $t$.
251:
252: Since the classical impact oscillator is confined to $x \ge 0$, the eigenstates of the corresponding quantum
253: system are restricted to {\it odd} $n$ values only. Substituting the explicit expression for the Hermite polynomials,
254: $<x>$ can be expressed in terms
255: of parabolic cylindrical functions which are functions of $x_c$.
256: \begin{equation}
257: <x>_{2n-1} = \frac{\sum_{l=1}^{4n} A_{4n-l}(y_c) D_{-l}}{\sum_{l=1}^{4n-1} B_{4n-l}(y_c) D_{-l}}
258: \end{equation}
259: where $D_{-m}(y_c)$ is a parabolic cylinder functions of order $m$.
260: Here $y_c=\alpha x_c$ where $\alpha=\sqrt{2\omega_0/\hbar}$.
261: $A_l(y_c)$ and $B_l(y_c)$ are polynomials of $y_c$ of degree $l$ that are uniquely determined by $n$.
262: For example, for $n=1$, we have
263: \begin{equation}
264: <x>_{1}=\frac{1}{\alpha}\frac{6D_{-4}(y_c)-4y_cD_{-3}(y_c)+y_c^2D_{-2}(y_c)}{2D_{-3}(y_c)-2y_cD_{-2}(y_c)+y_c^2D_{-1}(y_c)}
265: \end{equation}
266:
267: For higher values of $n$, the expressions are very complicated.
268: Our analysis has been confined to the ground state $n=1$, and the
269: first excited state $n=3$. For all values of $n$,
270: $\hbar\phi_{q,n}$ reduces to the classical action as $\hbar \to 0$.
271: We factor out this part
272: of the phase factor $\phi_0$ and
273: write $\phi_{q,n}=\phi_0+\phi_n$. As we show below, it is the
274: difference in the phases between the two neighboring eigenstates that exhibit quantum fingerprints of classical dynamics.
275:
276: Figures ~2 and ~3 show geometric phases for a fixed time
277: interval $T$ , equal to the driving period, for a periodic and a chaotic trajectory.
278: These results as well as other similar analysis suggest that $\phi_c$ may be related to $\phi_{n-1}-\phi_n$.
279:
280: For dynamical evolution involving arbitrary time $t$, our detailed analysis shows that the integrated quantum phase factors
281: can be expressed as a sum of linear and oscillatory functions of $t$ and hence can be written as
282: , $\phi_n(t)=v_nt+f_n(t)$. Here
283: $v_n$ are constants, independent of $t$ and $f_n(t)$ are oscillatory functions which are found to
284: retain the fingerprints of the corresponding classical dynamics for all times.
285:
286: In contrast to quantum phases, the classical $\phi_c(t)$ is found to be an oscillatory function
287: for periodic , quasiperiodic as well as for the chaotic trajectories.
288: This should be contrasted with the the driven and damped oscillator
289: phases\cite{QP1} where the classical phase
290: averaged over the period of the driver was finite.
291: \begin{figure}[htbp]
292: \includegraphics[width=3.5in]{fig2.eps}
293: \leavevmode
294: \caption{ For a period-$11$ trajectory, the four figures
295: (from top to bottom) show the quantum geometric phases for
296: the ground state( $n=1$), the first excited state ( $n=3$),
297: classical geometric phase $\phi_c$ and $d\phi=\phi_1-\phi_3$,
298: respectively.}
299: \label{fig2}
300: \end{figure}
301:
302: \begin{figure}[htbp]
303: \includegraphics[width=3.5in]{fig3.eps}
304: \leavevmode
305: \caption{For a chaotic trajectory, same plots as in Fig. ~2
306: In the third figure, we superimpose $d\phi$ with $\phi_c$.}
307: \label{fig3}
308: \end{figure}
309:
310: The results for arbirary time evolution are shown in
311: figures ~4 and ~5. Here we factor out the linearly increasing
312: parts of the phases
313: and show the fluctuating parts along with the
314: classical phase. For the parameter values corresponding to
315: the figures 2 to 5,
316: $v_1/(2\pi)=.4218$ and $v_3/(2\pi)=.3956$ for
317: the periodic orbit and $v_1/(2\pi)=.3723$ and $v_3/(2\pi)=.4021$
318: for the chaotic orbits.
319: We notice that $\phi_c$ is of the same order of magnitude
320: as $f_1-f_3$ with $\phi_c$
321: exhibiting intermittent fluctuations.
322: In view of the fact that $v_1 \approx v_3$, it is
323: conceivable that $v_n$ approaches a constant, independent
324: of $n$ as $n \rightarrow \infty$
325: and therefore $\phi_c$ may be related to the $d\phi$ for
326: arbitrary time $t$.
327: This is analogous to Berry's relation $\theta_n = -\partial_n
328: \psi_n$ which was true for integrable systems in semiclassical limit.
329:
330: \begin{figure}[htbp]
331: \includegraphics[width=3.5in]{fig4.eps}
332: \leavevmode
333: \caption{For a period-11 trajectory, the top two figures
334: show the oscillatory
335: components $f_1$ and $f_3$ of the net accumulated phases
336: for the ground state and the excited state.
337: The third figure shows the classical phase
338: and the differnce $df=f_1-f_3$. The lowest figure
339: shows the position of the
340: oscillator. }
341: \label{fig4}
342: \end{figure}
343:
344: \begin{figure}[htbp]
345: \includegraphics[width=3.5in]{fig5.eps}
346: \leavevmode
347: \caption{For a chaotic trajectory, the same plots as in
348: Fig. ~4. }
349: \label{fig5}
350: \end{figure}
351:
352: In summary, our results as depicted above describe preliminary
353: studies of classical and quantum phases
354: underlying a driven nonlinear oscillator. An interesting
355: result is the possible relationship
356: between the classical phases and the difference in the
357: quantum phases between the two
358: neighboring states, reminiscent of the Berry relation
359: $\theta_h=-\partial_n \phi_n$.
360: As a consequence, the smalless of the classical phase
361: will have its origin in
362: the relative stability of the quantum phase with respect to
363: the quantum number $n$.
364: It is rather intriguing that the fluctuations in the quantum
365: geometric phases retain the finger prints of the corresponding
366: classical dynamics.
367: Further detailed studies involving higher excited states
368: are needed to confirm these speculative views.
369:
370: In physical applications such as atom optics, Hamiltonians of the form,
371: $H(x,t)=H_0+\lambda x \sin(\omega t)$ are relevant where $H_0$
372: is the time-independent
373: Hamiltonian and
374: the time-dependent term describes
375: the interaction with a single mode radiation field in
376: dipole approximation.
377: For nonlinear $H_0$, nontrivial dynamics may lead to
378: many surprises. The impact oscillators
379: exhibit dynamics with many features that are typical of a
380: nonlinear oscillator.
381: This suggests that the driven impact oscillator may provide
382: an interesting theoretical model
383: to explore various issues relevant to quantum chaos.
384:
385: The research of IIS is supported by National Science
386: Foundation Grant No. DMR~0072813.
387: \begin{references}
388: \bibitem{berry}M. V. Berry, Proc. Roy. Soc. London A {\bf 392}, 45 (1984).
389: \bibitem{wilczek}See, for example,
390: {\it Geometrical Phases in Physics}, edited by A. Shapere and
391: F. Wilczek ( World Scientific, Singapore, 1989), and references therein.
392: \bibitem{aharanov} Y. Aharanov and J. Anandan, Phys. Rev. Lett. {\bf 58}, 1593 (1987);
393: \bibitem{MS} N. Mukunda and R. Simon, Ann. Phys. (N.Y.) {\bf 228}, 205 (1993).
394: \bibitem{sam}J. Samuel and R. Bhandari, Phys. Rev. Lett. {\bf 60},
395: 2339 (1988).
396: \bibitem{hannay} J. H. Hannay, J. Phys. A: Math. Gen. {\bf 18}, 221 (1985).
397: \bibitem{RB} J. M. Robbins and M. V. Berry, Proc. R. Soc. Lond. A (1992) 436, 631.
398: \bibitem{FS}D. J. Struik, {\it Lectures on Classical Differential
399: Geometry} (Addison-Wesley, Reading, Mass., 1961).
400: \bibitem{bs} See, for example, J.M.T. Thompson and H. B. Stewart,
401: {\it Nonlinear Dynamics and Chaos} (Wiley, New York, 1986), Chapter
402: 14, and references therein.
403: \bibitem{radha}Radha Balakrishnan, A. R. Bishop, and R. Dandoloff, Phys. Rev. Lett.
404: {\bf 64}, 2107 (1990).
405: \bibitem{Kerner} See Y. Nogamin, Am. J. Phys, {\bf 59}, 64
406: (1991).
407: \bibitem{QP1} "Anholonomy and Geometrical localization in
408: Nonlinear oscillator", Radha Balakrishnan
409: and Indubala Satija, (preprint)nlin.CD/0303071.
410: \end{references}
411: \end{document}
412:
413:
414:
415:
416:
417:
418:
419:
420:
421:
422:
423: