1: \documentclass[12pt, a4paper]{article}
2: \usepackage{graphicx}
3: \usepackage{wrapfig}
4: \oddsidemargin=0pt \textwidth=16cm \headheight=0pt \headsep=0pt
5: \topmargin=0pt \textheight=25cm
6: \author{V.P.Berezovoj$^\ast$, Yu.L.Bolotin$^\ast$, G.I.Ivashkevych$^\star$.}
7: \title{Geometrical approach for description \\of the mixed state in
8: multi-well potentials.}
9: \begin{document}
10: \maketitle
11: \begin{center}
12: {\footnotesize $^\ast$ National Scientific Center "Kharkov
13: Institute of Physics and
14: Technology"\\
15: Akademicheskaya str., 1, 61108, Kharkov, Ukraine\\ E-mail:
16: bolotin@kipt.kharkov.ua\\
17: $^\star$ Kharkov National University, Department of Physics and
18: Technology \\Svobody sqr., 4, 61077, Kharkov, Ukraine}
19: \end{center}
20: {\footnotesize We use so-called geometrical approach~\cite{Cerruti}
21: in description of transition from regular motion to chaotic in
22: Hamiltonian systems with potential energy surface that has several
23: local minima. Distinctive feature of such systems is coexistence of
24: different types of dynamics (regular or chaotic) in different wells
25: at the same energy ~\cite{bolotin1}. Mixed state reveals unique
26: opportunities in research of quantum manifestations of classical
27: stochasticity~\cite{berezovoj}. Application of traditional criteria
28: for transition to chaos (resonance overlap criterion, negative
29: curvature criterion and stochastic layer destruction criterion) is
30: inefficient in case of potentials with complex topology. Geometrical
31: approach allows considering only configuration space but not phase
32: space when investigating stability. Trajectories are viewed as
33: geodesics of configuration space equipped with suitable metric. In
34: this approach all information about chaos and regularity consists in
35: potential function. The aim of this work is to determine what
36: details of geometry of potential lead to chaos in Hamiltonian
37: systems using geometrical approach. Numerical calculations are
38: executed for potentials that are relevant with lowest umbilical
39: catastrophes.}
40:
41: \section {Mixed state. Phenomenological description.}
42:
43: Hamiltonian system with multi-well potential energy surface(PES)
44: represents a realistic model, describing the dynamics of
45: transition between different equilibrium states, including such
46: important cases as chemical reactions, nuclear fission and phase
47: transitions. It became known in 80-th that existence of mixed
48: state is an important feature of such systems~\cite{bolotin1}.
49: Mixed state means that there are different dynamical regimes in
50: different local minima at the same energy, either regular, or
51: chaotic. For example let’s demonstrate an existence of mixed state
52: for nuclear quadrupole oscillations Hamiltonian. It can be shown
53: that using only transformation properties of the interaction the
54: deformation potential of surface quadrupole oscillations of nuclei
55: takes on the form ~\cite{mosel}:
56: \begin{equation} U(a_0 ,a_2 ) = \sum\limits_{m,n} {C_{mn} (a_0^2 +
57: 2a_2^2 )^m a_0^n (6a_2^2 - a_0^2 )^n } \end{equation} where $a_0$
58: and $a_2$ are internal coordinates of the nuclear surface during
59: the quadrupole oscillations:
60:
61: \begin{equation}
62: R(\theta ,\varphi ) = R_0 \{ 1 + a_0 Y_{2,0} (\theta ,\varphi ) + a_2 [Y_{2,2} (\theta ,\varphi ) + Y_{2, - 2} (\theta ,\varphi )]\}
63: \end{equation}
64:
65: Constants $C_{mn}$ can be considered as phenomenological
66: parameters. Restricting with the members of the fourth degree in
67: the deformation and assuming the equality of mass parameters for
68: two independent directions, we get $C_{3v}$-symmetric Hamiltonian:
69: \begin{equation}
70: H = (p_x^2 + p_y^2 )/2m + U_{QO} (x,y;a,b,c)
71: \end{equation}
72: where
73: \begin{equation}
74: \begin{array}{c}
75: U_{QO} (x,y;a,b,c) = \frac{a}{2}(x^2 + y^2 ) + b(x^2 y - \frac{1}{3}y^3 ) + c(x^2 + y^2 )^2 \\
76: x = \sqrt 2 a_2 ,y = a_0 ,a = 2C_{10} ,b = 3C_{01} ,c = C_{20} \\
77: \end{array}
78: \end{equation}
79: Hamiltonian (3) and corresponding equations of motion depend only on
80: parameter $W=b^2/ac$, the unique dimensionless quantity we can build
81: from parameters a,b,c. The same parameter determines the geometry of
82: PES. Interval $0<W<16$ includes potentials with single extremum –
83: minimum in the origin that corresponds to spherical symmetric shape
84: of the nucleus. In the interval $W>16$ PES $U_{QO}$ contains seven
85: extremums: four minima (central, placed in the origin and three
86: peripheral, which correspond to deformated states of nuclei) and
87: three saddles, which separate peripheral minima from central one.
88: The distinctive feature of transition from regularity to chaos in
89: such a potential lies in the fact that energy of transition is not
90: the same in different local minima. Thus, $E_{cr} \sim E_{s}/2$
91: ($E_s$ - energy in the saddles) for the central minimum and
92: $E_{cr}\sim E_s$ for peripheral. Due to this in the interval
93: $E_s/2<E<E_s$ classical dynamics is mainly chaotic in the central
94: minimum and remains regular in peripheral minima (Fig.~\ref{allp},
95: right). Term “mixed state” is used for designation of such specific
96: dynamics.
97:
98: Mixed state is natural for multi-well potentials. This statement
99: is illustrated by Fig.~\ref{allp} (left and center), which
100: represents level lines and Poincare sections in different energies
101: for multi-well potentials from family of umbilical catastrophes
102: $D_5$ and $D_7$:
103: \begin{equation}
104: \begin{array}{c}
105: U_{D_5 } = 2y^2 - x^2 + xy^2 + \frac{1}{4}x^4 \\
106: U_{D_7 } = \sqrt 2 y^2 + \frac{3}{8}x^2 + xy^2 - \frac{1}{2}x^4 + \frac{1}{6}x^6 \\
107: \end{array}
108: \end{equation}
109:
110: One can see that there exists chaos in wells with three saddles, while in other wells motion is regular.
111: Let’s note the distinction of sections structure in different wells. In the low energy there exists a hyperbolic point in the section for wells with
112: chaotic motion. At the same time there is no such a point in the regular wells and structure of sections is similar at the different energies.
113:
114: \section {Importance of the mixed state for quantum chaos.}
115: \begin{wrapfigure}[13]{r}{5cm}\hspace*{1pt}
116: \includegraphics[scale=0.5]{waveD5}
117: \caption{\footnotesize Wave function structure in $D_5$ potential}\label{waveD5}
118: \end{wrapfigure}
119:
120: The mixed state represents optimal object for investigation of
121: quantum manifestations of classical stochasticity (QMCS) in wave
122: function structure. Indeed, usual procedure of search for QMCS in
123: wave functions implies distinction in its structure below and
124: above classical critical energy (or other parameters of
125: regularity-chaos transition). However, such procedure meets
126: difficulties connected with necessity to separate QMCS from
127: modifications of wave functions structure due to trivial changes
128: of its quantum numbers. Wave functions of the mixed state allow
129: finding QMCS in comparison not different eigenfunctions, but
130: different parts of the same eigenfunction, suited in different
131: regions of configuration space (different local minima of the
132: potential). By way of example, comparing the structure of the
133: eigenfunctions in central and peripheral minima of the QO
134: potential or in left and right minima of the $D_5$, it is evident
135: that nodal structure of the regular and chaotic parts is clearly
136: different, but correlating with the character of the classical
137: motion (see Fig.~\ref{waveD5}).
138: \begin{figure}
139: \begin{center}\includegraphics[scale=0.7]{allp}\end{center} \caption{\footnotesize
140: Level lines and Poincare sections for $D_5$(left), $D_7$(center) and
141: QO (right) potentials. Sections are presented at energies $E_s/4$,
142: $E_s/2$ and $E_s$.} \label{allp}
143: \end{figure}
144:
145:
146: \section {Stochastic criteria for mixed state.}
147:
148: As well known~\cite{Zaslavsky}, stochasticity is understood as a
149: rise of statistical properties in purely deterministic system due
150: to local instability. According to this idea values of parameters
151: of dynamical system, under which local instability arises, are
152: identified as regularity-chaos transition values. However,
153: stochasticity criteria of such a type are not sufficient (their
154: necessity offer a separate and complicated question), since loss
155: of stability could lead to transformation of one kind of regular
156: motion to another one. Although this serious limitation,
157: stochastic criteria in combination with numerical experiments
158: facilitate an analysis of motion and essentially extend efficacy
159: of numerical calculations.
160:
161: First among widely used stochasticity criteria is nonlinear
162: resonances overlap criterion presented by
163: Chirikov~\cite{Chirikov}. According to this criterion rise of
164: local instability is generated by contact of separatrixes of
165: neighboring nonlinear resonances. In this approach the scenario of
166: stochasticity is the following. The averaged motion of the system
167: in the neighborhood of the isolated nonlinear resonance on the
168: plane of the action-angle variables is similar to the particle
169: behavior in the potential well. Several resonances correspond to
170: several potential wells. The overlap of the resonances is
171: responsible for the possibility of the random walk of particle
172: between these wells. This method could be modified for the systems
173: with unique resonance~\cite{Doviel}. In this case the origin of
174: the large-scale stochasticity is connected with the destruction of
175: the stochastic layer near the separatrix of the isolated
176: resonance.
177:
178: Application of these criteria in presence of
179: strong nonlinearity (which is inevitable when considering
180: multi-well potentials) encounters an obstacle: action-angle
181: variables effectively work only in neighborhood of local minimum.
182: Because of this, an interest to methods, based on direct
183: estimation of trajectories moving away speed, arises. The
184: criterion of such a type is so-called negative curvature criterion
185: (NCC)~\cite{toda}. This criterion connects stochastisation of
186: motion with getting to part of configuration space, where Gaussian
187: curvature of PES is negative when energy increases (while in
188: neighborhood of minima curvature is always positive). Then energy
189: of transition to chaos is close to minimal energy on the
190: zero-curvature line. Negative curvature criterion allows getting a
191: number of interesting results both for classical motion and QMCS
192: in potentials with simple geometry – a single
193: minimum~\cite{berezovoj}. However, when passing on to the
194: multi-well potentials, NCC fails to work correct. In particular,
195: for above mentioned potentials ($D_5$ and $D_7$), structure of
196: Gaussian curvature is similar in different wells. For example, for
197: $D_5$ potential according to NCC we get the same value of critical
198: energy for both minima: $-5/9$, but chaotic motion is only
199: observed in the left well (Fig.~\ref{allp}, left). A natural
200: question immediately arises: is it possible, using only
201: geometrical properties of PES but not solving numerically
202: equations of motion, to formulate an algorithm for finding a
203: critical energy for single local minima in multi-well potential?
204: We'll try to answer on this question below in the framework of
205: geometrical approach.
206:
207: \section {Geometrical approach to Hamiltonian mechanics.}
208:
209: We will use so-called geometrical approach in consideration of
210: mixed state. Let’s recall the basics of this
211: method~\cite{Cerruti}. It is known that Hamiltonian dynamics could
212: be formulated in the terms of Riemannian geometry. In this
213: approach trajectories of the system are considered as geodesics of
214: some manifold. Grounds for such consideration lie in variational
215: base of Hamiltonian mechanics. Geodesics are determined by
216: condition:
217: \begin{equation}
218: \delta \int\limits_L {ds = 0}
219: \end{equation}
220: At the same time trajectories of dynamical system are determined
221: according to the Maupertuis principle:
222: \begin{equation}
223: \delta \int\limits_\gamma {2Tdt = 0}
224: \end{equation}
225: ($\gamma$ - all isoenergetic paths connecting end points) or to the
226: Hamilton’s principle:
227: \begin{equation}
228: \delta \int\limits_{t_1 }^{t_2 } {Ldt = 0}
229: \end{equation}
230:
231: Once chose a suitable metric action could be rewrote as a length
232: of the curve on the manifold. Then trajectories will be geodesics
233: on this manifold(configurational - CM). This approach has an
234: evident advantage: potential energy function includes all
235: information about the system, so one need to consider only
236: configurational space (CS) but not phase space. Equations of
237: motion in this case take on the form:
238: \begin{equation}
239: \frac{{d^2 q^i }}{{ds^2 }} + \Gamma _{jk}^i \frac{{dq^j
240: }}{{ds}}\frac{{dq^k }}{{ds}} = 0
241: \end{equation}
242: Christoffel symbols in this approach play as counterparts of
243: forces in ordinary mechanics. The most natural metric is a Jacobi
244: one. It has the form:
245: \begin{equation}
246: g_{ij} = 2[E - V(q)]\delta _{ij}
247: \end{equation}
248: By means of this metric Maupertuis principle could be rewrote in
249: the form equivalent to condition for geodesics.
250:
251: Let’s consider local instability in the framework of above
252: mentioned geometrical approach. Let $q$ and $q'$ be two nearby
253: at the $t=0$ trajectories:
254: \begin{equation}
255: q'^i (s) = q^i (s) + J^i (s)
256: \end{equation}
257: Separation vector $\overrightarrow{J}$ then satisfy the
258: Jacobi-Levi-Civita equation:
259: \begin{equation}
260: \frac{{d^2 J^i }}{{ds^2 }} + R_{jkl}^i \frac{{dq^j }}{{ds}}J^k
261: \frac{{dq^l }}{{ds}} = 0
262: \end{equation}
263:
264: It can be shown that dynamics of the deviation is determined only by
265: Riemannian curvature of the manifold. For two-degrees-of-freedom
266: systems Riemannian curvature has a form:
267: \begin{equation}
268: R = \frac{1}{{4(E - V)^2 }}[2(E - V)\Delta V + 2\left| {\nabla V}
269: \right|^2 ]
270: \end{equation}
271: Laplasian of $V$ is positive for considered potentials so
272: Riemannian curvature is positive too. Due to this we couldn't
273: connect divergence of trajectories with negative Riemannian
274: curvature. The one way to solve this problem consists in
275: introduction of higher-dimensional (than $N$) metrics. Let’s
276: examine this question closer. It can be shown that equation for
277: separation vector $\overrightarrow{J}$ could be reformulated in
278: form, which doesn't depend on dimensionality of manifold:
279: \begin{equation}
280: \frac{1}{2}\frac{{d^2 \left\| {\vec J} \right\|^2 }}{{ds^2 }} +
281: K^{(2)} (\vec J,\vec v)\left\| {\vec J} \right\|^2 - \left\|
282: {\frac{\nabla }{{ds}}\vec J} \right\|^2 = 0
283: \end{equation}
284: where $K^{(2)}$ is a sectional curvature in two-dimensional
285: direction:
286: \begin{equation}
287: K^{(2)} (\vec J,\vec v) = R_{iklm} \frac{{J^i }}{{\left\| {\vec J}
288: \right\|}}\frac{{dq^k }}{{ds}}\frac{{J^l }}{{\left\| {\vec J}
289: \right\|}}\frac{{dq^m }}{{ds}}
290: \end{equation}
291: and $ \left\langle {\vec J,\vec v} \right\rangle = 0 $. Note that
292: point where $K^{(2)}<0$ is an unstable one. Since there are more
293: than one sectional curvature for the case $N>2$, we could connect
294: instability with negative sign of some of them. One among the
295: enlarged metrics is an Eisenhart metric. Eisenhart metric is
296: $N+2$-dimentional and contains two additional coordinates. One of
297: these coordinates coincides with a time and second is connected
298: with action. Using Eisenhart metric, quantity $K^{(2)}$ could be
299: rewrote in the form:
300: \begin{equation}
301: K^{(2)} (\dot{\vec {q}} ,\vec q) = \frac{1}{{2(E -
302: V)}}(\frac{{\partial ^2 V}}{{\partial q_1^2 }}\mathop
303: {{\dot{q}_2}^2} + \frac{{\partial ^2 V}}{{\partial q_2^2 }}\mathop
304: {{\dot{q}_1}^2} - 2\frac{{\partial ^2 V}}{{\partial q_1
305: \partial q_2 }}\mathop {\dot{q}_1} \mathop {\dot{q}_{2}} )
306: \end{equation}
307:
308: Now, investigation of the $K^{(2)}$ – structure on the considered
309: manifold could be used for studying of chaotic regimes and, in
310: particular, the mixed state.
311: \section {Investigation of mixed state in the framework of geometrical
312: approach.}
313: As mentioned above, negative sign of $K^{(2)}$ is a
314: condition for rise of local instability. It is necessary to
315: clarify whether this condition is sufficient for development of
316: chaoticity or not, clearly speaking, one needs to answer the
317: question, does a presence of negative curvature parts on CM always
318: lead to chaos. Potentials with mixed state are a very convenient
319: model for investigation of this question, since there exist both
320: regimes of motion. So, we need to study, how differs the structure
321: of $K^{(2)}$ in different wells. For that we calculate a part of
322: phase space with negative curvature as a function of energy, i.e.
323: a volume of phase space where $K^{(2)}<0$ referred to the total
324: volume:
325: \begin{equation}
326: \mu (E) = {{\int {d\vec qd\vec p \Theta ( - K^{(2)} ) \delta
327: (H(\vec q,\vec p) - E)} } \mathord{\left/
328: {\vphantom {{\int {d\vec qd\vec p \Theta ( - K^{(2)} )\delta (H(\vec q,\vec p) - E)} } {\int {d\vec qd\vec p\delta (H(\vec q,\vec p) - E)} }}} \right.
329: \kern-\nulldelimiterspace} {\int {d\vec qd\vec p\delta (H(\vec q,\vec p) - E)} }}
330: \end{equation}
331: An advantage of this approach consists in necessity to calculate
332: only geometrical properties of system without solving equations of
333: motion.
334: \begin{figure}[h]
335: \hspace*{1pt}
336: \begin{center}\includegraphics[scale=0.9]{D57k2}\end{center}
337: \caption{\footnotesize Function $\mu(E)$ for $D_5 (a)$ and $D_7 (b)$ potentials. $\mu(E)$ for chaotic wells are represented by doted lines, for regular - by triangles.} \label{D57}
338: \end{figure}
339: \begin{figure}[h]
340: \hspace*{1pt}
341: \begin{center}\includegraphics[scale=0.9]{D57kg}\end{center}
342: \caption{\footnotesize Function $\mu_g(E)$ for $D_5 (a)$ and $D_7 (b)$ potentials.} \label{D57g}
343: \end{figure}
344:
345: We carried out calculations for two potentials: $D_5$ and $D_7$.
346: Calculations of $\mu(E)$ (Figure~\ref{D57}) show that there are
347: parts, where $K^{(2)}<0$, in all wells, but nevertheless chaos
348: exists only in one well. Moreover, for well with chaotic motion
349: function $\mu(E)$ gives correct value of critical energy (in the
350: sense, specified in Section 3). In this energy $\mu$ becomes
351: positive. Situation with regular wells is more complex. Although
352: part of phase space, where $K^{(2)}<0$, is nonzero, chaos in the
353: well doesn't exist. This can be viewed on the Poincare sections.
354: For comparison on the Fig.~\ref{D57g} part of CS with negative
355: Gaussian curvature is shown. One can see that structure of
356: negative Gaussian curvature is similar to the $K^{(2)}$-structure.
357:
358: \section {Conclusions.}
359:
360:
361: Investigation of curvature of manifold, as one can see from the
362: cited above data, doesn't give a plain method for identification
363: of chaos in any minimum, especially if there exist both regular
364: and chaotic regimes of motion. It is impossible to determine a
365: priori whether chaos existed in the system without using dynamical
366: description (in out case that are Poincare sections).
367: Nevertheless, one can efficiently use geometrical methods for
368: investigation of chaos in multi-well potentials. In considered
369: above potentials chaos exists only in wells, which have two
370: details: non-zero part of negative curvature on the manifold and
371: at least one hyperbolic point in the Poincare section. According
372: to this, one can use the following method for identification of
373: chaos and calculation of critical energy. At the first step the
374: Poincare section in low energy is drew for the well and the
375: presence of hyperbolic point is determined. If so, the quantity
376: $\mu$ must be calculated (or the part of CS with negative Gaussian
377: curvature). Value of energy, in which $\mu(E)$ become positive,
378: could be associated with critical energy. If there is no
379: hyperbolic point in the section than chaos doesn't exist in the
380: well. Consequently, geometrical methods could be efficiently used
381: for determination of critical energy in complex potentials and
382: identification of chaos in general. However, one must carefully
383: use these methods and combine them with qualitative methods, such
384: as Poincare sectioning method.
385: \begin{thebibliography}{10}
386: \bibitem{Cerruti}
387: M.Cerruti-Sola, M.Pettini, Phys.Rev. E 53, 179 (1996).
388: \bibitem{bolotin1}
389: Yu.L.Bolotin, V.Yu.Gonchar, E.V.Inopin, Yad. Fiz. V 45, 351
390: (1987).
391: \bibitem{berezovoj}
392: V.P.Berezovoj, Yu.L.Bolotin, V.A.Cherkaskiy, Phys. Lett. V. A 323,
393: 318 (2004).
394:
395: \bibitem{mosel}
396: V.Mosel, W.Greiner, Z.Phys. V 217, 256
397: (1968).
398: \bibitem{Zaslavsky}
399: G.M.Zaslavsky, Chaos in Dynamical Systems, N.Y.:Harwood, 1985.
400: \bibitem{Chirikov}
401: B.V.Chirikov, Atom.Energy, V 6, 630 (1959).
402: \bibitem{Doviel}
403: P.Doviel, D.Escande, J.Codacconi, Phys.Rev.Lett., V 49, 1879
404: (1983).
405: \bibitem{toda}
406: M.Toda,
407: Phys.Lett. A, V 48, 335 (1974).
408: \end{thebibliography}
409: {\footnotesize Report was made at XIII International Seminar
410: "Nonlinear phenomena in complex systems", Minsk, Belarus, May 16-19
411: 2006}
412: \newpage
413: \begin{figure}
414: \begin{center}\includegraphics[scale=1]{allp}\end{center} \caption{\footnotesize
415: Level lines and Poincare sections for $D_5$(left), $D_7$(center) and
416: QO (right) potentials. Sections are presented at energies $E_s/4$,
417: $E_s/2$ and $E_s$.} \label{allp}
418: \end{figure}
419: \end {document}
420: