1:
2: \documentclass[fleqn]{article}
3: \usepackage[dvips]{graphicx}% grafikebis chasasmelad
4:
5: \textwidth=16cm \textheight=22cm \oddsidemargin=0pt
6:
7: \author{L.Chotorlishvili, Z.Toklikishvili, V.Bochorishvili , A.Sagaradze}
8: \date{}
9: \title{\large \bf Investigation of Quantum Chaos in the Parametric Dependent System of
10: Interacting oscillators }
11:
12: \begin{document}
13: \maketitle
14: \begin{center}
15: Tbilisi State University, Georgia, 0128. Tbilisi, Chavchavadze av.
16: 3 \\
17: Email: lchotor@yahoo.com
18: \begin{abstract}
19: Formation of chaos in the parametric dependent system of
20: interacting oscillators for the both classical and quantum cases
21: has been investigated. Domain in which classical motion is chaotic
22: is defined. It has been shown that for certain values of the
23: parameters from this domain, form of the classical power spectrum
24: is in a good agreement with the quantum band profile. Local
25: density of states is calculated. The range in which application of
26: perturbation theory is correct has been defined.
27: \end{abstract}
28: Pacs number 05.45. -a
29: \end{center}
30: \section{ Introduction. Problem statement.}
31: At last time the investigation of parametric dependent
32: Hamiltonians like $H(Q,P,\lambda)$ is a subject of great interest.
33: Here $(Q,P)$ is a set of canonical coordinates and impulses, and
34: $\lambda$ is a parameter describing connection of the system with
35: the external field. Interest to such systems is motivated by
36: mesoscopic physics [1]. Most of the works devoted to the
37: parametric dependent systems are dealing with the investigation of
38: following situation: For $\lambda=0$ Hamiltonian is exactly
39: integrable. For $\lambda>0$ Hamiltonian $H(Q,P,\lambda)$ becomes
40: non-integrable and for some defined values of
41: $\lambda=\lambda_{0}$, solutions of classical equations generated
42: by the Hamiltonian $H(Q,P,\lambda_{0})$ display chaos. After this,
43: the little variation of the parameter
44: $\delta\lambda=\lambda-\lambda_{0}$ is of interest. Using the
45: basis $\left|\Phi_{n}\right\rangle$ in which
46: $\hat{H}(Q,P,\lambda_{0})$ is diagonal, for small values of
47: $\delta\lambda$, Hamiltonian $\hat{H}(Q,P,\lambda)$ can be written
48: in the form of sum of two matrix $$ \hat{H}=E_{0}+\delta\lambda
49: B.$$ Here $E_{0}$ is the diagonal matrix, and $B$ is the banded
50: matrix, off-diagonal elements of which are random numbers [2]. By
51: way of numerical diagonalization it is possible to define
52: eigenvectors
53: $\left|\Phi_{n}(\lambda_{0}+\delta\lambda)\right\rangle$,\,
54: $\left|\Phi_{n}(\lambda_{0})\right\rangle$ and eigenvalues
55: $E_{n}(\lambda_{0}+\delta\lambda)$,\,$E(\lambda_{0})$
56: corresponding to the Hamiltonians
57: $\hat{H}(Q,P,\lambda)$,\,$\hat{H_{0}}(Q,P,\lambda_{0})$
58: appropriately. The main interest at this is connected with the
59: dependence of parametric kernel
60: $$P(n|m)=\left|\left\langle\Phi_{n}(\lambda_{0}+\delta\lambda)|\Phi_{m}(\lambda_{0})\right\rangle\right|^{2}$$
61: from the parameter $\delta\lambda$. The value of
62: $P(r)=\overline{P(n|n+r)}$ averaged over the $n$ can be considered
63: as the local density of states. Descriptions of foregoing methods
64: of solving of problems of quantum chaos are given in [3-5]. Goal
65: of this paper is investigation of chaos in the system of
66: interacting oscillators when both of the Hamiltonians
67: $\hat{H_{0}}(Q,P,\lambda_{0})$, $\hat{H}(Q,P,\lambda)$ display
68: chaos. Situation when the Hamiltonian $\hat{H_{0}}$ is integrable
69: has been studied by us earlier (see papers [6,7]).
70: \newpage
71: \section{ Classical consideration. Model Hamiltonian.}
72: Investigation of the following Hamiltonian is of interest for us
73: \begin{eqnarray} \label{1}
74: H&=&H_{0}+\lambda V \nonumber ,\\
75: H_{0}&=&\frac{P_{1}^{2}}{2}+\frac{P_{2}^{2}}{2}+\frac{P_{3}^{2}}{2}+
76: \frac{1}{2}(q_{1}^{2}+q_{2}^{2}+q_{3}^{2}) ,\\
77: V&=&q_{1}^{2}q_{2}^{2}q_{3}^{2} \nonumber.
78: \end{eqnarray}
79:
80: Here and below we work with the dimensionless quantities.
81: Canonical equations for the Hamiltonian (\ref{1}) are of form
82: \begin{eqnarray}\label{2}
83: \dot{q_{1}}&=&p_{1} \nonumber ,\\
84: \dot{p_{1}}&=&-q_{1}-2\lambda q_{1}q_{2}^{2}q_{3}^{2} ,\nonumber \\
85: \dot{q_{2}}&=&p_{2} ,\\
86: \dot{p_{2}}&=&-q_{2}-2\lambda q_{2}q_{1}^{2}q_{3}^{2} ,\nonumber \\
87: \dot{q_{3}}&=&p_{3} \nonumber ,\\
88: \dot{p_{3}}&=&-q_{3}-2\lambda q_{3}q_{1}^{2}q_{2}^{2} \nonumber.
89: \end{eqnarray}
90: Solutions for the set of equations in case of $\lambda\neq0$ can
91: be obtained only numerically. As is shown by numerical analyze,
92: solutions of the set of equations (\ref{2}) for the values of the
93: parameters $\lambda=3.0$, $H=4.2$ are essentially chaotic (see
94: Fig.1). For further calculations we have to obtain classical power
95: spectrum with the purpose of its further comparison with the band
96: profile. Let us construct correlation function for
97: $V(q_{1}\,q_{2}\,q_{3})=-\frac{\partial H}{\partial\lambda}$
98: \begin{equation}\label{3}
99: G(\tau)=\left\langle\left(V(t+\tau)-\bar{V}\right)\left(V(t)-\bar{V}\right)\right\rangle
100: \end{equation}
101: where $\left\langle\ldots\right\rangle$ means averaging over the
102: time, $\bar{V}$ is the averaged values of $V$. We have to make
103: Fourier transformation of the expression (\ref{3}) to get
104: classical power spectrum
105: \begin{equation}\label{4}
106: C(\omega)=\int e^{-i\omega t}G(\tau)d\tau.
107: \end{equation}
108: In the practical calculation of (\ref{4}) we use the method
109: equivalent of (\ref{4})
110: \begin{equation}\label{5}
111: C(f)=\frac{1}{T}\left|\int_{-T/2}^{T/2}e^{-i\omega t}V(t)dt\right|,
112: \end{equation}
113: here $x \in \left[0,T \right]$ is time interval in which solution
114: of the set of equations (\ref{2}) is obtained,
115: $f=\frac{\omega}{2\pi}$. To get the power spectrum one needs to
116: use discrete form of expression (\ref{5})
117: \begin{equation} \label{6}
118: C(k)=\frac{T\Delta}{2l+1}\sum_{n=-l}^{l}\left|X(k+n)\right|^{2},
119: \end{equation}
120: here $X(k)=\sum_{j=0}^{N-1}V(j)e^\frac{i 2 \pi k}{N\Delta}$ ,
121: $k=fN\Delta$, $N$ is a number of dots dividing the interval $T$,
122: $\Delta=\frac{T}{N}$, $l$ is the averaging parameter [8,9]. Form
123: of the classical power spectrum
124: for the different values of the parameter $l$ is given on the Figures Fig. 2, Fig.3.
125: These data are confirming the fact that solutions of (\ref{2}) are chaotic.
126: \newpage
127: \section{Quantum mechanical consideration}
128: Our subsequent goal is the comparison of the band profile $\left|B_{nm}\right|^{2}$ considered as the
129: function of $\omega=\frac{E_{n}-E_{m}}{\hbar}$ with the classical band profile. Let us set about to the quantum-mechanical
130: analyze of the Hamiltonian (\ref{1}). Eigenvectors $ \left|\Psi_{n}\right\rangle$ of the basis, in which Hamiltonian
131: $H(P,Q,0)$ is diagonal can be found easily and are of the form [10]
132: \begin{equation}\label{7}
133: \Psi_{n_{1}n_{2}n_{3}}(q_{1}q_{2}q_{3})=\frac{H_{n_{1}}(q_{1})H_{n_{2}}(q_{2})H_{n_{3}}(q_{3})}{\sqrt{2^{n_{1}+n_{2}+n_{3}}n_{1}!n_{2}!n_{3}!\pi^{3}}}e^{-\frac{1}{2}(q_{1}^{2}+q_{2}^{2}+q_{3}^{2})},
134: \end{equation}
135: where $H_{n}(q)$ denotes the Hermite polynomials [11]. Eigenvalue
136: energy is given by the formula
137: \begin{equation}\label{8}
138: E_{n}=(n+\frac{3}{2}),
139: \end{equation}
140: where
141: \begin{equation}\label{9}
142: n=n_{1}+n_{2}+n_{3}.
143: \end{equation}
144: Condition (\ref{9}) implies existence of the
145: $\frac{(n+1)(n+2)}{2}$ fold degeneration for the given $n$.
146: \newline Diagonalization of the Hamiltonian (1) can be done only
147: numerically. On the Figures 4, and 5 dependence of the energy
148: eigenvalues $E_{n}(\lambda)$ from the parameter $\lambda$ are
149: plotted. According to Fig.4 for the $\lambda\neq0$ removal of
150: degeneration takes place. In case of further increasing of the
151: parameter $\lambda$ up to the values characterizing the domain of
152: chaotic motion $\lambda=3.0$, as easy to see from Fig.5,
153: repulsion of the energy terms happens. This fact is of great
154: importance since the repulsion of the energy terms is a sine of
155: emerging of quantum chaos [12].
156: For the defining of the band profile at first we need to find eigenvectors of the
157: Hamiltonian $H(P,Q,\lambda_{0})$. This problem is possible to resolve by way of numerical diagonalization
158: of the Hamiltonian $H(P,Q,\lambda_{0})$, at the values of
159: $\lambda_{0}=3.0$ in the basis of functions (\ref{7})
160: \begin{equation}\label{10}
161: \left\langle\Psi_{n_{1}n_{2}n_{3}}\left|\left(H_{0}+\lambda_{0}V\right)\right|\Psi_{m_{1}m_{2}m_{3}}\right\rangle
162: \end{equation}
163: Since the matrix elements are satisfying the condition
164: $$\left\langle n\left|V(q_{1}q_{2}q_{3})\right|m\right\rangle=\left\langle n_{1}n_{2}n_{2}\left|V(q_{1}q_{2}q_{3})\right|m_{1}m_{2}m_{2}\right\rangle=\left\langle n_{1}\left|q_{1}^{2}\right|m_{1}\right\rangle\left\langle n_{2}\left|q_{2}^{2}\right|m_{2}\right\rangle\left\langle
165: n_{3}\left|q_{3}^{2}\right|m_{3}\right\rangle,$$
166: taking into account that
167: \begin{equation}\label{11}
168: q_{nm}=\frac{1}{\sqrt{2}}(\sqrt{n}\delta_{n-1,m}+\sqrt{n+1}\delta_{n+1,m}),
169: \end{equation}
170: and using the condition
171: $\left(q^{2}\right)_{nm}=\sum_{k}q_{nk}q_{km}$ we get
172: \begin{equation}\label{12}
173: \left(q^{2}\right)_{nm}=\frac{1}{2}(\sqrt{n}\sqrt{n-1}\delta_{n-2,m}+(2n+1)\delta_{n,m}+\sqrt{n+1}\sqrt{n+2}\delta_{n+2,m}).
174: \end{equation}
175: In (\ref{11}) and (\ref{12}) $\delta_{n,m}$ stands for Kronecker
176: delta function.
177: Substituting (\ref{11}) and (\ref{12}) into (\ref{10}) after
178: straightforward calculations one can obtain
179: \begin{eqnarray}
180: \left\langle
181: n_{1}n_{2}n_{3}\left|V\right|m_{1}m_{2}m_{3}\right\rangle&=&\frac{1}{8}(\sqrt{n_{1}}\sqrt{n_{1}-1}\delta_{n_{1}-2,m_{1}}+(2n_{1}+1)\delta_{n_{1},m_{1}}+\sqrt{n_{1}+1}\sqrt{n_{1}+2}\delta_{n_{1}+2,m_{1}})\times \nonumber \\
182: & & \mbox{} \times(\sqrt{n_{2}}\sqrt{n_{2}-1}\delta_{n_{2}-2,m_{2}}+(2n_{2}+1)\delta_{n_{2},m_{2}}+\sqrt{n_{2}+1}\sqrt{n_{2}+2}\delta_{n_{2}+2,m_{2}})\times \nonumber \\
183: & & \mbox{} \times(\sqrt{n_{3}}\sqrt{n_{3}-1}\delta_{n_{3}-2,m_{3}}+(2n_{3}+1)\delta_{n_{3},m_{3}}+\sqrt{n_{3}+1}\sqrt{n_{3}+2}\delta_{n_{3}+2,m_{3}}) ,\nonumber
184: \end{eqnarray}
185: \begin{equation}\label{13}
186: \left\langle
187: n_{1}n_{2}n_{3}\left|H_{0}\right|m_{1}m_{2}m_{3}\right\rangle=(n_{1}+n_{2}+n_{3}+\frac{3}{2})\delta_{n_{1}m_{1}}\delta_{n_{2}m_{2}}\delta_{n_{3}m_{3}}.
188: \end{equation}
189: After forming from the matrix elements (\ref{13}) matrix and
190: making diagonalization one can obtain eigenvalues spectrum
191: $E_{n(\lambda_{0})}$ and basis of eigenvectors
192: $\Phi_{n}(\lambda_{0})$. Calculating matrix elements of
193: $V(q_{1},q_{2},q_{3})$ in the basis $\Phi_{n}(\lambda_{0})$ we can
194: get the band profile
195: \begin{equation} \label{14}
196: B_{nm}^{2}=\frac{4\pi^{2}\hbar^{2}}{\triangle_{0}}\left|\left\langle\Phi_{n}(\lambda_{0})\left|V(q_{1},q_{2},q_{3})\right|\Phi_{m}(\lambda_{0})\right\rangle\right|^{2}
197: \end{equation}
198: Here $\Delta_{0}=a \hbar^{d}$ is the mean level spacing, and $d$
199: is the system's dimension. Coefficient $a$ is determined from the
200: condition $a=\frac{1}{D(E)}$ , where $D(E)=\frac{\delta N}{\delta
201: E}$ is the density of states, and $\delta N$ is the number of
202: states in the energy interval $\delta E=E(P,Q,\lambda_{0}+\delta
203: \lambda)-E(P,Q,\lambda_{0})$. Density of states in it's turn is
204: possible to determine from the numerical diagonalization of the
205: Hamiltonian $H(P,Q,X_{0})$ , or from the integral
206: \begin{equation}
207: D(E)=\frac{1}{(2\pi\hbar)^{\alpha}}\frac{\partial}{\partial
208: E}\int_{H(P,Q,\lambda)=E}dPdQ .
209: \end{equation}
210: Here the integral must be taken over the phase space enclosed by
211: energy surface (see Fig.6) In our calculations we are interested
212: in the domain in which classical motion is essentially chaotic:
213: $E(P,Q,\lambda=3)=4.2;$ $\delta\lambda=0.1;$ $\delta E=0.1;$
214: $\hbar=0.1;$ $a\approx7.2$. Taking smaller values of $\hbar$ leads
215: to the necessity of diagonalization of matrix larger then our. In
216: our case ($\hbar=0.1$) matrix dimension is $455\times455$. Band
217: profile $B_{nm}^{2}$ as a function of
218: $\omega=\frac{E_{n}-E_{m}}{\hbar}$ is plotted on the Fig.7.
219: According to Fig.7 band profile is in a good qualitative agreement
220: with the classical power spectrum. Better quantitative agreement
221: is possible to achieve for smaller values of $\hbar$ [1] (since
222: the quasi-classical approximation implies $\hbar\rightarrow 0$).
223: Another quantity we are interested in is the parametric kernel
224: \begin{equation}\label{16}
225: P(r)=\overline{\left|\left\langle\Phi_{n+r}(\lambda_{0}+\delta\lambda)|\Phi_{n}(\lambda_{0})\right\rangle\right|^{2}},
226: \end{equation}
227: where the averaging is done over the several $n$ states. Evidently
228: (\ref{16}) determines correlation between the wave functions. The
229: parametric kernel (\ref{16}) frequently is named as the Local
230: Density of States (LDS) [3]. We are interested in $P(r)$, since
231: with its help, is possible to determine the boundaries in which
232: perturbation theory is valid. For that one has to compare
233: parametric kernel (\ref{16}) with the results for LDS obtained by
234: means of perturbation theory [5]
235: \begin{equation} \label{17}
236: P_{prt}(r)=\frac{\delta
237: x^{2}\left|B_{nm}\right|^{2}}{\Gamma^{2}(\delta
238: x)+(E_{n}-E_{m})^{2}},
239: \end{equation}
240: where the parameter $\Gamma(\delta x)$ for given $\delta x$ must
241: be defined from the normalization condition
242: \begin{equation} \label{18}
243: \sum_{r}P_{prt}(r)=1 .
244: \end{equation}
245: Results of numerical calculations for the values $\delta x=0.1$
246: and $\delta x=0.2$ are presented on the Figures Fig.8 and Fig.9.
247: On basis of obtained results we conclude that boundary in which
248: perturbation theory for our system is correct is limited by the
249: condition $\delta x\leq1$.
250: \newpage
251:
252: \begin{thebibliography}{99}
253:
254: \bibitem{1} 1. Y. Imry, "Introduction in Mesoscopic Physics", Oxford University Press, New York (1997)
255: \bibitem{2} 2. M. Feingold , A. Peres Phys. Rev. A.34 , 591 (1986)
256:
257: \bibitem{3} 3. D. Cohen, T. Kottos, arXiv: cond-mat / 0302319 v3 18 Mar (2004)
258: \bibitem{4} 4. D. Cohen, T. Kottos, Phys.Rev.Lett. 85, 4839 (2000)
259: \bibitem{5} 5. D. Cohen, T. Kottos, Phys.Rev.E 63, 36203 (2001)
260: \bibitem{6} 6. A. Ugulava, L. Chotorlishvili, and K. Nickoladze Phys.Rev.E 68, 026216 (2003)
261: \bibitem{7} 7. A. Ugulava, L. Chotorlishvili, and K. Nickoladze Phys. Rev. E 70, (2004)
262: \bibitem{8} 8. Gwilim M. Jenkins, Donald G. Watts " Spectral Analysis and its applications",
263: Holden-Day, San Francisco, Cambridge, London, Amsterdam 1969.
264: \bibitem{9} 9.Robert K. Otnes, Loren Enochson, " Applied Time Series Analysis", A Wiley-Interscience Publication
265: Jon Wiley and Sons New York Chichester Brisbane Toront 1978
266: \bibitem{10} 10. L.P.Landau, E.M.Lifshitz, Quantum Mechanics, Nonrelativistic Theory, Pergamon Press, Oxford, New York, 1977,
267: F. A. Kaempffer "Concepts in Quantum Mechanics" Academic Press.
268: New York and London. 1965
269:
270: \bibitem{11} 11. H.Bateman, A.Erdelyi, Higher Transcendental Functions, New York-Toronto-London, 1955,
271: Handbook of Mathematical Functions Edited by M. Abramovitz,
272: National Bureau of standarts Appl. Math.series.55, Issued
273: June 1964.
274:
275: \bibitem{12} 12. F. Haake " Quantum signature of chaos" , Springer Berlin, (2001)
276: \end{thebibliography}
277:
278: \newpage
279:
280: \begin{figure}
281: \begin{center}
282: \includegraphics[angle=0, width=0.5\textwidth ]{Fig1.eps}
283: \newline Fig.1 Phase trajectory projection on the plane $(p_{1},
284: q_{1})$ obtained for the values of the parameters: $E=4.2,
285: \lambda=3.0$. It is easy to see that trajectory is chaotic.
286: \end{center}
287: \end{figure}
288:
289: \begin{figure}
290: \begin{center}
291: \includegraphics[angle=0, width=0.5\textwidth ]{Fig2.eps}
292: \newline Fig.2 Classical Power Spectrum plotted for the values of the
293: parameters: $E=4.2, \lambda=3.0 ,l=10$
294: \end{center}
295: \end{figure}
296:
297: \newpage
298:
299: \begin{figure}
300: \begin{center}
301: \includegraphics[angle=0, width=0.5\textwidth ]{Fig3.eps}
302: \newline Fig.3 Classical Power Spectrum plotted for the values of the
303: parameters: $E=4.2, \lambda=3.0, l=3$
304: \end{center}
305: \end{figure}
306:
307: \begin{figure}
308: \begin{center}
309: \includegraphics[angle=0, width=0.5\textwidth ]{Fig4.eps}
310: \newline Fig. 4 Energy terms as a functions of the parameter $\lambda$
311: \end{center}
312: \end{figure}
313:
314: \newpage
315:
316: \begin{figure}
317: \begin{center}
318: \includegraphics[angle=0, width=0.5\textwidth ]{Fig5.eps}
319: \newline Fig. 5 Energy terms as a functions of the parameter $\lambda$. For
320: the values from the chaotic domain $\lambda\geq3.0$, repulsion of
321: the energy levels happens.
322: \end{center}
323: \end{figure}
324:
325:
326: \begin{figure}
327: \begin{center}
328: \includegraphics[angle=0, width=0.5\textwidth ]{Fig6.eps}
329: \newline Fig. 6 Equipotential surface constraint by the condition
330: $H(P,Q,\lambda)=E=4,2$
331: \end{center}
332: \end{figure}
333:
334: \newpage
335:
336: \begin{figure}
337: \begin{center}
338: \includegraphics[angle=0, width=0.5\textwidth ]{Fig7.eps}
339: \newline Fig. 7 Quantum band profile $B_{nm}^{2}$ as a function of
340: $\omega=\frac{E_{n}-E_{m}}{\hbar}$ for the values of parameters
341: $E=4.2, \lambda=3.0 ,\delta\lambda=0,1$. Evidently $B_{nm}^{2}$ is
342: in a good qualitative agreement with the Classical Power Spectrum
343: (see Fig. 2 )
344: \end{center}
345: \end{figure}
346:
347: \begin{figure}
348: \begin{center}
349: \includegraphics[angle=0, width=0.5\textwidth ]{Fig8.eps}
350: \newline Fig. 8 Quantum Profile $P(r)$ (Local Density of States) and
351: First-Order Perturbative Profile $P_{prt}(r)$ for the values of
352: the parameter $\lambda=0,1$.
353: \end{center}
354: \end{figure}
355:
356: \newpage
357:
358: \begin{figure}
359: \begin{center}
360: \includegraphics[angle=0, width=0.5\textwidth ]{Fig9.eps}
361: \newline Fig. 9 Quantum Profile $P(r)$ (Local Density of States) and
362: First-Order Perturbative Profile $P_{prt}(r)$ for the values of
363: $\lambda=0,2$. Comparing Fig.8 with Fig.9 one could say that
364: agreement between $P(r)$ and is better if $\lambda<0,1$ .
365: \end{center}
366: \end{figure}
367:
368:
369: \end{document}
370: