1: \documentstyle[epsfig, 11pt] {article}
2: \textwidth 160mm
3: \textheight 230mm
4: \parskip 12pt
5: \oddsidemargin 0pt
6: \evensidemargin 0pt
7: \topmargin 0pt
8: \begin{document}
9: \title {Singularity in classical and quantum Kepler Problem with
10: Weak Anisotropy}
11: \author {Zai-Qiao Bai and Wei-Mou Zheng \\
12: Institute of Theoretical Physics, Academia Sinica,
13: Beijing 100080, China }
14: \date{}
15: \maketitle
16: \baselineskip 24pt \vskip 5mm
17: \begin {minipage}{140mm}
18: \begin {center} {\bf Abstract} \end {center}
19: \baselineskip 24pt
20: Anisotropic Kepler problem is investigated by
21: perturbation method in both classical
22: and quantum mechanics.
23: In classical mechanics, due to the singularity
24: of the potential, global diffusion in phase
25: space occurs at an arbitrarily small perturbation
26: parameter. In quantum mechanics, the singularity
27: induces a large transition amplitude between
28: quasi degenerate eigen states, which generically decays as
29: $\hbar$ in the semi-classical limit.
30: \end {minipage}
31: \vskip 10mm
32: PACS number: 03.20.+i; 03.65.-w; 03.65.sq
33: \\
34: Keywords: Anisotropic Kepler problem, Singularity, Quantum-classical correspondence
35:
36: \newpage
37: The anisotropic Kelper problem (AKP)
38: is a 2-dimensional Hamiltonian system defined by
39: $$
40: H=\frac{1}{2}(p_x^2+p_y^2)-\frac{1}{\sqrt{\nu x^2+\mu y^2}}
41: \eqno (1)
42: $$
43: with $\mu>\nu>0$ and $\mu \nu=1$ (For a detailed discussion
44: of its dynamics, see \cite {Gut1} ). While the system
45: reduces to the Kepler problem when $\mu=1$,
46: the existence of chaos was rigorously
47: proved by Gutzwiller and Devaney in the case of
48: $\mu/\nu >9/8$ \cite{Gut2, Dev}.
49: The emergence of irregular dynamics dose not
50: follow the conventional KAM scenario due to
51: the singularity at the origin. In fact,
52: the extremely unstable motion in the vicinity of the
53: origin played a central role in Gutzwiller
54: and Devaney's proof. The dynamical implication
55: of the singularity when $\mu \rightarrow 1$ is, however,
56: still not clear. When $\mu \approx 1$, (1) can be
57: rewritten as
58: $$
59: H=\frac{1}{2}(p_x^2+p_y^2)-\frac{1}{r}
60: -\epsilon\frac{\cos 2\theta}{r}=
61: H_{0}+\epsilon F,
62: \eqno{(2)}
63: $$
64: where $r=\sqrt{x^2+y^2}$ and $\theta=\tan^{-1}(y/x)$
65: while $\epsilon=(\mu-1)/2$ appears as a natural
66: perturbation parameter. In this letter we shall
67: study the classical and quantum AKP by perturbation
68: method. We hope this will provide a deeper understanding
69: of the quantum-classic correspondence in non-KAM system.
70:
71: We begin with the classical mechanics.
72: Kepler problem is a maximally integrable system.
73: All its orbits are closed with period
74: $T=2\pi(-2E)^{-3/2}$, where $E=H_0<0$.
75: This global periodic motion will be destroyed
76: by an infinitely small perturbation. The first
77: order effect of a slight perturbation can be described
78: by the drift of closed orbit, i.e., the slow variation
79: of its parameters. Therefore, at this level of
80: approximation, we can consider the motion in the
81: space of all closed Kepler orbits (orbital space ).
82:
83: The orbital space can be conveniently constructed
84: by taking advantage of the $so(3)$ dynamical symmetry.
85: Let
86: $$
87: J_1=(\cos\theta-p_yJ_3)/\sqrt{-2E},~~J_2=(\sin\theta+p_xJ_3)/\sqrt{-2E},~~
88: {\rm and}~~J_3=xp_y-yp_x.
89: \eqno (3)
90: $$
91: $(J_1,J_2,J_3)$ form a
92: $so(3)$ algebra, i.e., $\{J_i,J_j\}=\varepsilon_{ijk}J_k,i,j=1,2,3$
93: and $H_0=\frac{-1}{2(J_1^2+J_2^2+J_3^2)}.$
94: There is a one-to-one correspondence
95: between ${\mathcal J}=(J_1,J_2,J_3)$ and close Kepler orbits.
96: Specifically, write $(J_1,J_2,J_3)=\frac{1}{\sqrt{-2E}}
97: (\sin \alpha \cos\beta,\sin \alpha \sin\beta, \cos\alpha)$,
98: the corresponding Kepler orbit
99: in configuration space is defined as
100: $$
101: r=\frac{ J_3^2}{1-\sin \alpha \cos(\theta-\beta)}
102: \eqno (4)
103: $$
104: while the sign of $J_3$ determines its direction.
105: Therefore, the orbital space is coordinated by
106: $(J_1,J_2,J_3)$ ( or $(E, \alpha, \beta$)).
107:
108: The variation of $J_i$ in one Kepler period is given by
109: $$
110: J_i(t+T)=J_i(t)+\epsilon T \{J_i,F_0({\mathcal J})\}+o(\epsilon),
111: ~~~i=1,2,3,
112: \eqno(5)
113: $$
114: where $F_0$ is the time average of
115: $F$ over closed Kepler orbit,
116: $$
117: F_0=F_0({\mathcal J})=\frac{1}{T}\int _0^{2\pi}
118: F(r,\theta)\frac{r^2}{|J_3|}d\theta=
119: 2E\frac {1-|\cos\alpha|}{1+|\cos\alpha|} \cos 2\beta.
120: \eqno (6)
121: $$
122: When $\epsilon\rightarrow 0$, Eq. (5) can be approximated by
123: a differential equation,
124: $$
125: \frac{d J_i}{d t}=\epsilon\{J_i,F_0({\mathcal J})\},~~~i=1,2,3.
126: \eqno (7)
127: $$
128: $\{H_0,F_0\}=0$ implies that the motion in the orbital space is
129: confined within a sphere ${\mathcal S}_E$ with $H_0=E=const.$.
130: Restricting $so(3)$ Poisson structure on ${\mathcal S}_E$ induces a
131: natural sympletic form
132: $
133: \omega_2=\frac{\sin \alpha}{\sqrt{-2E}} d\beta \wedge d\alpha,
134: $
135: which is, up to a constant, the ordinary area element.
136:
137: As a two-dimensional Hamiltonian system, the dynamics on ${\mathcal S}_E$
138: can be easily determined by the contour chart of the
139: effective Hamiltonian $F_0$. There exist
140: six fixed points on the sphere.
141: The poles $(\alpha=0,\pi)$, which represent the two
142: circular Kepler orbits, are unstable while the
143: four on equator $(\alpha=\pi/2,\beta=0,\pm \pi/2,\pi)$,
144: which corresponds to the linear orbits on
145: the $x$ or $y$ axis, are stable.
146: The remaining orbits are either the
147: heteroclinic orbits ($\beta=\pm \pi/4,\pm 3\pi/4$)
148: that connect the poles or periodic orbits
149: surrounding one of the stable fixed points.
150: In other words, besides separatrixs, the classical
151: motion consists of four islands centered respectively
152: at the degenerate Kepler orbits on the $x$
153: and $y$ axis.
154:
155: It should be pointed out that
156: the simple picture given by
157: perturbation analysis is not correct at
158: the vicinity of $\alpha=\pi/2$, i.e., the
159: collision orbits, where $\epsilon F$ is not bounded.
160: In fact, the first order derivative of $F_0$ is in general not
161: continuous at equator. This non-smoothness
162: manifests the non-perturbative
163: nature of the motion near the origin.
164: Noticing that all orbits on ${\mathcal S}_E$
165: except for the poles cross $\alpha=\pi/2$, we conclude that
166: the global dynamics of AKP cannot be described by
167: perturbation with respect to Kepler problem even at
168: the limit $\epsilon\rightarrow 0$.
169: Our numerical study show that the
170: collision orbits provide a passage-way
171: for global diffusion in the
172: orbital (and hence phase) space (Fig. 1).
173:
174: Now we turn to quantum mechanics.
175: The $so(3)$ symmetry of quantum planar Kepler problem
176: is constructed in a way similar to its classical analog.
177: Specifically,
178: $$
179: \left\{\begin{array}{cl}
180: J_1= & [\cos\theta-\frac{1}{2}(p_yJ_3+J_3p_y)]/\sqrt{-2H_0},
181: \\\\
182: J_2= & [\sin\theta+\frac{1}{2}(p_xJ_3+J_3p_x)]/\sqrt{-2H_0},\\\\
183: J_3= & xp_y-yp_x,
184: \end{array}
185: \right.
186: \eqno (8)
187: $$
188: $[J_j,J_k]=i\varepsilon_{jkl}J_l$ and
189: $H_0=\frac{-1}{2(J^2+\frac{1}{4})}$ ($\hbar\equiv 1$).
190: Let $\{ |n,m>: m\leq |n|, n=0,1,... \}$ be
191: the standard $so(3)$ orthonormal set, i.e.,
192: $J_3 |n,m>=m|n,m> $ and $J^2|n,m>=n(n+1)|n,m>$.
193: $|n,m>$ is an eigenstate of $H_0$ with energy
194: $E=E_n=\frac{-1}{2(n+\frac{1}{2})^2}$ and wave function
195: in coordinate representation given by
196: $$
197: \Psi_{n,m}(r,\theta)=(-1)^{m}\frac{1}{|2m|!}\sqrt{\frac{(n+|m|)!}
198: {(n-|m|)!}}\frac{4 s^{|m|}}{(2n+1)^{3/2}}
199: e^{-\frac{1}{2}s }F(-n+|m|,2|m|+1,s)
200: \frac{e^{im\theta}}{\sqrt{2\pi}}
201: \eqno(9)
202: $$
203: where $s=4r/(2n+1)$ and $F(a,b,x)$ is the confluent hypergeometric
204: function.
205:
206: The first order effect of a slight perturbation
207: is the mixing of states with definite $n$. In the
208: interaction representation, the long-time evolution
209: when $\epsilon \rightarrow 0$ is given by an effective
210: Sh$\ddot o$rdinger equation
211: $$
212: i \frac{d}{d\tau}\Psi=\overline {F} \Psi
213: \eqno (10)
214: $$
215: where $\overline {F}$ is the restriction of $F$ in eigen spaces
216: of $H_0$, i.e.,
217: $$
218: <n',m'|\overline {F}|n,m>=\delta_{n,n'}<n',m'|F|n,m>.
219: \eqno (11)
220: $$
221: Eq. (10) can be regarded as the quantum counterpart of
222: Eq. (6) and hence $F_0$ is the classical correspondence of
223: $\overline{F}$. The matrix elements of $\overline {F}$
224: is readily evaluated in coordinate representation,
225: $$
226: <n,m'|\overline{F}|n,m>=
227: E_n\sqrt{\frac{(n+m_1)!(n-m_1)!}{(n+m_2)!(n-m_2)!}}\delta_{|m'-m|,2},
228: \eqno(12)
229: $$
230: where $m_1=\min \{|m|,|m'|\}$ and $m_2=\max \{|m|,|m'|\}$.
231:
232: By diagonalizing $\overline{F}$ in each
233: $2n+1$-dimensional subspace, we can study
234: the classical-quantum correspondence
235: in the framework of perturbation theory.
236: Notice that $\exp(i\frac{\pi}{2} J_3) \overline {F}+
237: \overline {F}\exp(i\frac{\pi}{2} J_3)=0$,
238: the spectrum of $\overline {F}$ is symmetric with respect to $0$.
239: We shall focus on the positive part, which corresponds to
240: the classical islands centered at fixed
241: points $(\alpha,\beta)=(\pi/2,\pi/2)$
242: and $(\pi/2,3\pi/2)$.
243:
244: The classical orbits centered at fixed point $(\pi/2,\pi/2)$
245: contribute to the spectrum of $\overline {F}$
246: according to the semi-classical quantization rule,
247: $$
248: \frac{1}{2\pi\hbar}\int_{\sigma(\lambda_k)}
249: \omega_2=k+\mu_s/4, ~~~k=0,1,...
250: \eqno(13)
251: $$
252: where $\sigma(\lambda)\subset S_{E}$ is the region
253: enclosed by orbit with $F_0(E,\alpha,\beta)=\lambda$
254: and the Maslov index $\mu_s=2$.
255: For convenience, we rescale $E=-\frac{1}{2}$ and
256: $\hbar=\frac{2}{2n+1}$ so that $|F_0|\le 1$ and
257: Eq. (13) yields
258: $$
259: S(n,\lambda_k)\equiv \frac{n+\frac{1}{2}}{\pi}(\cos^{-1}\lambda_k+
260: \frac{2\lambda_k\ln \lambda_k}{\sqrt{1-\lambda_k^2}})=k+\frac{1}{2}.
261: \eqno(14)
262: $$
263: The orbits surrounding point $(\pi/2,3\pi/2)$ gives
264: the same contribution so that the semi-classical
265: spectrum consists of 2-fold degeneracies.
266: Eq. (14) is a good approximation of the exact spectrum (Table 1).
267:
268:
269: The degeneracy predicted by the semi-classical rule
270: is not exact due to quantum tunneling between the
271: two quasi static states each of them classically
272: corresponds to one periodic orbit. The splitting of
273: spectrum $(\Delta \lambda)$
274: is connected with the transition amplitude accumulated
275: in one Kepler period ($A_T$) by
276: $A_T=\epsilon \Delta \lambda T/2\hbar$.
277: In conventional quantum system, such as a particle confined in
278: double-well potential, $\Delta\lambda$ decays as
279: $\sim \hbar^{\gamma}\exp(-S/\hbar)$ when
280: $\hbar\rightarrow 0$, which vanishes faster
281: than any power of $\hbar$.
282: Fig. 2 shows $\Delta\lambda_{sc}\equiv\Delta\lambda/\hbar^2$ for
283: $n=100$ and $1000$, from which we can see that
284: $\Delta\lambda\sim \hbar^2$ (or $A_T\sim \hbar$) in the
285: semi-classical limit.
286: As a characteristic phenomenon in
287: in non-smooth system\cite{Bai}, this power-law reflects the
288: non-smoothness in $F_0$ and hence the singularity in AKP.
289:
290: In summary, we have studied in this letter the Anisotropic
291: Kepler problem by perturbation method. By using the dynamical
292: symmetry of the unperturbed system, we obtained
293: an effective Hamiltonian in both classical and
294: quantum mechanics.
295: We shown that the long-time evolution of
296: most of the classical orbits is non-perturbative
297: due to the singularity at origin and
298: global diffusion in phase space occurs even at
299: an arbitrarily small perturbation parameter.
300: In quantum mechanics, we shown that
301: the characteristic phenomenon attributed
302: the singularity is the the power-law $\hbar$-dependence of
303: tunneling amplitude between states
304: which classically correspond to
305: separated periodic orbits.
306:
307: \begin{thebibliography}{9}
308: \bibitem{Gut1} M. C. Gutzwiller, Chaos in Classical and Quantum Mechanics
309: (Springer, New York, 1990).
310: \bibitem{Gut2} M. C. Gutzwiller, J. Math. Phys. 18, 806 (1977).
311: \bibitem{Dev} R. L. Devaney, Invent. Math. 45, 221 (1978).
312: \bibitem{Bai} Z. Q. Bai preprint (quant-ph/0110126)
313:
314: \end{thebibliography}
315:
316: \newpage
317: \begin {center}
318: Table
319:
320: \begin{tabular}{|c|c|c|c|c|}
321: \hline
322: $k$& $\lambda$ & $S(n,\lambda)$ & $\lambda$ &$S(n,\lambda)$ \\
323: \hline
324: 0& 0.6883 & 0.34 & 0.6512 & 0.42 \\
325: 1& 0.3335 & 1.52 & 0.3264 & 1.55 \\
326: 2& 0.1944 & 2.43 & 0.1875 & 2.48 \\
327: 3& 0.0898 & 3.50 & 0.0856 & 3.55 \\
328: 4& 0.0307 & 4.43 & 0.0293 & 4.46 \\
329: \hline
330: \end{tabular}
331:
332: Tab.1 Spectrum of $\overline {F}$ at $n=10$.
333: \end{center}
334:
335: \newpage
336: \begin{figure} [htbp]
337: \vspace{-9cm}
338: \epsfig{file=f1.eps,width=16cm}
339: \vspace{-8cm}
340: \caption{(a) Numerical calculation of one orbit at $\mu=1.04$.
341: $5000$ intersections are plotted on the Poincar$\acute{e}$
342: surface of section defined at $\theta=0$.
343: with coordinate $(R,P)={\rm sign} (p_y)(\sqrt{r},\sqrt{r}p_x)$,
344: comparing with the regular structure implied by the first-order
345: perturbation (b).
346: }
347: \end{figure}
348:
349: \begin{figure} [htbp]
350: \vspace{-8cm}
351: \epsfig{file=f2.eps,width=16cm}
352: \vspace{-8cm}
353: \caption{ Scaled spectrum splitting ($\Delta \lambda_{sc}$) for $n=100$ and
354: $1000$. Except for few points near $\lambda=0$,
355: $(\lambda,\Delta \lambda_{sc})$ is approximately located
356: at two curves (determined by whether $k$ is even or odd)
357: irrespective of $n$.}
358: \end{figure}
359: \end {document}
360: