1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: \documentstyle[aps,amssymb,12pt]{revtex}
3: %\documentclass{article}
4: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
5: %\documentstyle[prb,aps,twocolumn,amssymb]{revtex}
6: \begin{document}
7: \title{ Quantum and classical solutions for a \\
8: free particle in wedge billiards}
9: \author{ A. G\'ongora-T$^{1,2}$, Jorge V.
10: Jos\'{e}$^{2}$, S. Schaffner $^{2}$ and P. H. E. Tiesinga$^{3}$}
11: \address{$^{1}$ Centro de Ciencias F\'\i sicas, Universidad Nacional
12: Aut\'onoma de M\'exico\\ Apartado Postal 48-3, 62250 Cuernavaca,
13: Morelos, MEXICO. \\ $^{2}$Physics Department and Center for Interdisciplinary
14: Research on Complex Systems, \\ Northeastern University, Boston,
15: MA~02115, USA.\\ $^{3}$Salk Institute, La Jolla, California, USA.
16: \\ }
17: \date{\today}
18: \preprint
19: \draft
20: \maketitle
21: %
22: \begin{abstract}
23: We have studied the quantum and classical solutions of a particle
24: constrained to move inside a sector circular billiard with angle
25: $\theta_w$ and its pacman complement with angle $2\pi-\theta_w$.
26: In these billiards rotational invariance is broken and angular momentum
27: is no longer a conserved quantum number. The ``fractional" angular momentum
28: quantum solutions are given in terms of Bessel functions of fractional
29: order, with indices $\lambda_p={p\pi \over {\theta_w}}$, $p=1,2,...$ for the
30: sector and $\mu_q={q\pi \over {2\pi - \theta_w}}$, $q=1,2...$ for the
31: pacman. We derive a ``duality'' relation between both fractional indices
32: given by $\lambda_p={{p\mu_q} \over {2\mu_q - q}}$ and $\mu_q = {{q\lambda_p}
33: \over {2\lambda_p - p}}$. We find that the average of the angular momentum
34: $\hat L_z$
35: is zero but the average of $\hat L^2_z$ has as eigenvalues $\lambda_p^2$ and
36: $\mu_q^2$.
37: We also make a connection of some classical solutions to their quantum
38: wave eigenfunction counterparts.
39:
40: \end{abstract}
41: {Pacs 05.45.+b, 03.65.-w,72.20.Ht}
42: \newpage
43: Significant progress has been made in the last few years in
44: understanding the connection between classical and quantum solutions
45: for problems that show chaotic behavior. A particularly important role
46: in this progress has been played by studies
47: in billiards with different geometries \cite{casati,bohigas,houches2,stock}.
48: The dynamics of a
49: free particle in a circular billiard is
50: completely integrable since energy and angular momentum are conserved.
51: A simple change of the circular billiard to the stadium billiard
52: immediately leads to
53: chaotic particle dynamics \cite{stock}.
54: The analysis of these billiards eigenvalue spectra and eigenfunctions
55: have yielded a
56: number of clearly defined quantum manifestations of classically chaotic
57: Hamiltonians \cite{casati,bohigas,stock}. There are also other
58: types of billiards that,
59: although not being
60: explicitly chaotic,
61: can yield interesting novel quantum and classical behavior. A case in
62: point considered in this paper has a free particle that moves inside a boundary
63: defined by a wedge shaped section of a circular billiard. In this case
64: regular $2\pi$-rotational invariant
65: angular momentum is no longer conserved. However, {\it fractional} quantum angular
66: momentum
67: is well defined. It is this fractional angular momentum that makes this
68: problem interesting. There have been other studies of modified circular
69: billiards, like a wedge sector in the presence of a constant
70: linear field that shows chaotic behavior \cite{wedge}, and circular chaotic billiards
71: with a straight cut \cite{linda}. These billiards are different from the
72: ones we study in this paper.
73: Here we consider sector billiards with angle $\theta_s=\theta_w$
74: and their pacman complement of angle $\theta_p=2\pi -\theta_w$.
75: These two types of
76: billiards can be considered special cases of simple
77: wedge billiards that have corner discontinuities. For our
78: quantum analysis it is, however, relevant to separate them this
79: way since the angular momentum spectral
80: properties of the sector and its corresponding pacman
81: can be related to each other.
82:
83: Our original motivation to study this problem actually came from experiments carried
84: out in pacman type mesoscopic billiards \cite{berry1,berry2}. These billiards
85: were studied in the presence of a magnetic field where the breaking of
86: rotational invariance precludes the full rotational closing of Aharonov-Bohm loops.
87: This significantly affects their contribution to the magnetoresistence.
88: As a first step towards understanding the magnetic field problem, we consider in this paper
89: the quantum and classical zero field cases. The quantum problem can be studied
90: to some extent analytically,
91: leading to interesting duality spectral relations. We also make a connection between the
92: quantum and classical solutions that help to understand the quantum results.
93:
94: The quantum Hamiltonian for a free particle of mass $M$ in a wedge billiard is
95: $\hat H = {\hat P^2 \over 2M} + V(r )$, with
96:
97: \[
98: V(r)=\left\{
99: \begin{array}{ll}
100: 0 & r\varepsilon D, \\
101: \infty & r\notin D.
102: \end{array}
103: \right.
104: \]
105: Here $D$ is the domain of the wedge billiard.
106: The corresponding time-independent Schr\"{o}dinger equation is
107: \begin{equation}
108: (\nabla^2 + k^2)\Psi_{D} = 0,
109: \end{equation}
110: \noindent
111: with boundary condition $\Psi_{C}=\Psi(r\in C)= 0 $, with $C$ the
112: boundary of the domain $D$. Here $k^2={2ME \over {\hbar ^2}}$,
113: $E$ is the energy and
114: $2\pi \hbar =h$ is Planck's constant.
115: We can immediately write the full set of solutions for the sector and
116: pacman wave-functions that satisfy the radial and angular boundary
117: conditions. The general normalized sector wave function is then given by
118: \begin{equation}
119: \Psi_{s}=\Sigma_{p}^{\infty} \Sigma_{n>1}^{\infty}
120: {J_{\lambda _p}({\alpha_{\lambda_p , n} \over r_w} \rho) \sin\lambda_p
121: \theta \over {\sqrt{\theta_s}\over 2}{r_w}
122: J_{\lambda_{p+1}}(\alpha_{\lambda_p,n})}.
123: \end{equation}
124: \noindent
125: Note that this wave function vanishes when $\theta =0$ and
126: $\theta=\theta_s$, and
127: $\rho={r_w}$, with $r_w$ the sector radius. The angular boundary
128: condition determines the values for the indices $\lambda_p$ that are
129: \begin{equation}
130: \lambda_p ={ p\pi \over \theta_s} ,\; p=1,2,3...
131: \end{equation}
132: \noindent
133: Consider, for example, the case
134: where the angular momenta are the integers
135: $\lambda_p=ap$ that correspond to the
136: sector angles
137: $\theta_s={\pi/a}$, with $a$ an integer.
138: Depending on the value of $a$ we will have a set of integers
139: that will be a subset of the index values
140: for the angular momenta
141: of the full circle. For example, if we take
142: $\theta_s={\pi/4}$, we get $\lambda_p=4,8,12,16,...$ or if
143: $\theta_s={\pi/3}$ we get $\lambda_p=3,6,9,12,15,...$.
144: If the index is even we get a subset of even integer angular momenta
145: while for the odd case we get a subset of even and odd values for the
146: angular momentum. We can instead chose $\theta_s={b\over a}\pi$, which
147: will give $\lambda_p=p\frac{b}{a}$, with $b$ and $a$ prime numbers. In
148: this case $\lambda_p$ will generally be fractional and no full
149: circle angular momentum values will be present.
150: In the case where the angle is irrational, say
151: $\theta_s=\alpha\pi$, with $\alpha$ irrational,
152: the situation radically changes since $\lambda_p=p/\alpha$
153: has no corresponding analog in the angular momenta for the complete
154: circle or for the rational angles.
155:
156: In the pacman case we can also write the complete wave function as
157: \begin{equation}
158: \Psi_{p}=\Sigma_{\mu_q}^{\infty} \Sigma_{m>1}^{\infty}
159: {J_{\mu_q}({\alpha_{\mu_q , m} \over {r_w}} \rho)
160: \sin\mu_q \theta
161: \over
162: {\sqrt{2\pi-\theta_w
163: }\over 2}{r_w} J_{\mu_{q+1}}(\alpha_{\mu_q , m})},
164: \end{equation}
165: \noindent
166: where $\mu_q$ is the order of the Bessel function and $m=1,2,3....$,
167: with $\theta_p=2\pi-\theta_w$, and
168: \begin{equation}
169: \mu_q={q\pi \over (2\pi - \theta_w)},\; q=1,2,3...,
170: \end{equation}
171: which is of the same form
172: as in the sector case. Again as in that case this wave function
173: satisfies the imposed angular and radial boundary conditions. Note
174: that if we take the pacman angle equal to ${\pi/4}$,
175: or equivalently as a ${7\pi/4}$ wedge, then
176: $\mu_q=4/7,8/7,12/7,16/7,...$ that has even numerators
177: as in the sector case but with a fractional angular
178: momentum index that has no counterpart in the sector
179: nor in the circle cases. For an angle $\theta_w=\pi/a$, with $a$ an integer, we
180: have $\mu_q=q/(2a-1)$, $a\geq 1$. Or, more generally,
181: for $\theta_w={b\over a}\pi$,
182: with $b$ and $a$ prime numbers, we have $\mu_q={q\over (2a-b)}$, and
183: we need to have that $2a\not=b$, which is satisfied for prime numbers.
184: This is an interesting result since these fractional angular momentum
185: cases do not correspond to cases previously studied in
186: group theory \cite{groups},
187: at least not to the best of our knowledge.
188: In the irrational case
189: with $\theta_w=2\pi-\alpha\pi$, we have $\mu_q=q/\alpha$, which has the
190: same value as in the sector case. In fact, we show from geometric relations between
191: the sector and the pacman wedges and for the same $\theta_w$,
192: the ``duality'' relations between the corresponding
193: Bessel function fractional angular momentum indices given by
194: \begin{equation}
195: \mu_q={\lambda_p \over {2\lambda_p - p}}q\Leftrightarrow \lambda_p =
196: {\mu_q \over {2\mu_q - q}}p.
197: \end{equation}
198: \noindent
199: We note that, although the energy eigenvalues for the sector and the
200: pacman given by
201: $E_{\mu{_q}m}={\hbar \alpha^2_{\mu_q,m}\over 2M}$
202: and
203: $E_{\lambda{_p}n}={\hbar \alpha^2_{\lambda_p,n}\over 2M}$,
204: \noindent
205: are not the same, this duality relation gives a nontrivial
206: connection between fractional Bessel functions and the corresponding
207: ``fractional'' angular momenta for the sector and its pacman complement.
208: In the wedge billiards rotational invariant
209: angular momentum is not a good quantum number. We calculate
210: then the average of the $z$-component of the angular momentum
211: $\hat L_z=(\frac{\hbar}{i}){\partial \over \partial \theta}$,
212: using the wave function given in Eq.(2), and we get
213: \begin{equation}
214: <\Psi_{s}\mid{\hat L_z}\mid \Psi_{s}>=
215: (\frac{\hbar}{i}){\sin^2\lambda_p\theta_w \over \theta_w}=0,
216: \end{equation}
217: and also for the pacman
218: \begin{equation}
219: <\Psi_{p}\mid{\hat L_z}\mid \Psi_{p}>=
220: (\frac{\hbar}{i}){\sin^2\mu_q(2\pi-\theta_w) \over (2\pi- \theta_w)}=0.
221: \end{equation}
222: \noindent
223: These results can be physically understood from a semi-classical
224: analysis. In a full circle we have that the particle motion
225: completes a full rotation between $0$ and $2\pi$. For the wedges
226: the angular motion is limited to be between $\theta \in [0,\theta_w]$
227: for the sectors or $\theta \in [\theta_w,2\pi]$ for the
228: pacmen. Since in the sector the
229: particle moves from $0$ to $\theta_w$ and then bounces back to move from
230: $\theta_w$ to $0$, the motion is librational rather than
231: rotational.
232: One result of this is that for the separable eigenfunctions given above all have
233: zero average $L_z$. When the particle bounces off the outer periphery, only $v_r$
234: changes and $L_z$ is unchanged. When it bounces off either wedge
235: wall, $v_\theta$ changes sign and thus ${L_z}^2$ is unchanged.
236: This quantum result comes from our use of a specific coordinate system
237: with origin of rotation at the apex of the billiard. However, in the classical analysis
238: described below we also find internal closed orbits in the billiards
239: where the angular momentum is not zero. To represent the latter ones
240: we need to write a linear combination of the apex centered eigenfunctions
241: and carry out a coordinate transformation to define the new angular
242: momentum with center of rotation away from the apex.
243:
244: Because of the $v_r$ cancellations with the boundary collisions
245: we are led to consider instead $\hat L_z^2$ that
246: would take care of these cancellations. Using the full sector
247: and pacman wave-functions we get
248: \begin{equation}
249: <\Psi_{s}\mid (\hat L_z)^2\mid \Psi_{s}>=
250: -(\frac{\hbar}{i})^2 \lambda_p^2,
251: \end{equation}
252: \noindent
253: and
254: \begin{equation}
255: <\Psi_{p}\mid (\hat L_z)^2\mid \Psi_{p}>=
256: -(\frac{\hbar}{i})^2 \mu_q^2.
257: \end{equation}
258: \noindent
259: We then see that the square of the Bessel function indices are good
260: quantum numbers. Note that this result applies even in the
261: irrational $\theta_w$ angle case.
262: Additionally, we also evaluated the expectation value of the quantum
263: mechanical currents defined by
264: $J\sim Im [(\Psi\nabla)^\ast\Psi]$. We found that the angular
265: component of the current
266: is zero, contrary to what happens in
267: the circle case, while the radial current component
268: is zero in
269: both billiard types, as in the circular case.
270:
271: \noindent
272: We now make a qualitative connection of these quantum results to their classical counterparts.
273: Consider a wedge cut from a thin disk with radius $r_{w}$. A particle
274: moves freely within the wedge, bouncing elastically from the walls.
275: Define a polar coordinate system $(\rho ,\theta)$
276: with origin at the apex of the wedge billiard and with $\theta=0$
277: pointing along the
278: right hand boundary of the wedge. The particle is
279: constrained to move in the ranges $0\leq\theta\leq\theta _{w}$
280: and $0<\rho<r_{w}$. The classical motion of the particle is complicated in polar coordinates,
281: but the elastic collisions are simple to describe. Suppose the particle has velocity $(v_{\rho},v_{\theta})$ when it
282: encounters the maximum radius of the wedge. The outer circle normal
283: vector is perpendicular to $\hat \theta$,
284: so $v_{\theta}$ will be unchanged. Because the collisions are elastic,
285: the energy ${m\over 2}(v_{\rho}^{2}+(r{v}_{\theta })^{2})$ must also be constant.
286: This requires that $v_{\rho}\Rightarrow-v_{\rho} $. Similarly, for a collision
287: with a wedge radial wall,
288: $(v_{\rho},v_{\theta })\Rightarrow(v_{\rho},-v_{\theta}) $.
289: We can use dimensionless coordinates where $r_{w}=1$
290: and $\left|\vec v\right|=1$, since the geometry of
291: the trajectory is not changed by the scaling of time and radius.
292: In fact, the collisions with the radial walls can be removed by lifting
293: copies of the wedge onto an infinite spiral in $\theta$.
294: The wedge copies are connected on the spiral in such a way that the motion of a particle
295: The spiral
296: is then projected onto the wedge by folding it like a fan.
297: The mapping from the spiral to the wedge is done in two stages. First,
298: define $\sigma\, \doteq\, \theta \,mod\, 2\theta_{w}$. Then
299:
300: \[
301: \theta=\left\{
302: \begin{array}{ll}
303: \sigma, & 0\leq \sigma \leq \theta _{w} \\
304: 2\theta _{w}-\sigma, &
305: \theta _{w}\leq \sigma <2\theta _{w}.
306: \end{array}
307: \right.
308: \]
309: In order for an orbit to close, the
310: particle must bounce off exactly the same point on the periphery,
311: going in the same direction. This requires that the total angle
312: traversed must
313: be $b 2 \theta_w$, for some positive integer $b$.
314: Combining the two, a closed orbit is possible when $a \theta_c= b 2
315: \theta_w$. It
316: is easy to show that orbits will be closed or
317: open independently of the starting position of the first orbit. Orbits
318: can then
319: be characterized by the pair of positive integers
320: $(a,b)$ \cite{robinett}. The chord angle $\theta_c$ is related to the angle of incidence
321: $\phi$
322: by $\theta_c =2\phi$. In order for the chord to stay
323: within the wedge, $ \theta _c\leq \pi $,
324: with $n \geq {{ 2 \theta_w\over\pi}b}$. This implies, in particular, that the trajectory
325: will be repeating if and only if $\theta_{w}$ and $\theta_{c}$ are commensurate.
326: The particle makes a series of collisions with the outer perimeter
327: at regular intervals $a \theta_c$.
328:
329:
330: We can now compare the spectra of two wedges to see if they both have
331: orbits for certain common values of $\theta_c$. In particular, consider a
332: wedge sector with angle $\theta_s$ and its $\theta_p = 2\pi-\theta_s$
333: complement pacman. In order for an orbit in each to share a common $\theta_c$,
334: we need to have $
335: \theta _c={{b_s2\theta _s}
336: \over n_s}
337: ={{b_{p}2\theta_p}
338: \over n_{p}}
339: $ or
340: $
341: {n_p\over b_p}
342: ={\theta _p\over \theta _s}
343: \, {n_s\over b_s}
344: $. Thus, if the pacman complement angle is an integer multiple $m$ of the
345: angle of the sector, any orbit $(a,b)$ in the wedge
346: will have an orbit $(m*a,b)$ in the pacman for exactly the same value of
347: $\theta_c$.
348:
349:
350: In Fig. (\ref{fig1}) we show an example of a pacman rational angle case
351: with angle of $\pi/4$. In this figure we show the orbit $(16,3)$, where
352: the particle strikes the outer circle boundary 16 times before the orbit closes. During
353: such a traversal, the total rotation of the particle on the lifted spiral is equal
354: to $6\cdot \frac{7\pi}{4}$. We can also generate a whole family of classical
355: orbits for this pacman with indices like $(7,1)$, $(8,1)$, $(28,1)$, $(28,3)$,
356: $(28,5)$ and so on.
357: In Fig. (\ref{fig2}) we show the corresponding eigenfunction for this pacman
358: case. Note that the caustic radius and the number of small and large
359: triangles correlate to the wave function amplitude densities shown in this
360: figure. The eigenfunction calculations were carried out by solving directly
361: the Shr$\ddot o$dinger equation using the finite element method.
362: In order to make a direct correlation between classical orbits and
363: quantum eigenfunctions we need to consider all the possible allowed
364: classical and quantum solutions for different values of the caustic
365: radius and set of parameter orbits $(a,b)$ and their corresponding quantum
366: counterparts. We do not carry out this analysis since we already know
367: the exact quantum solutions and we have described some of the classical
368: solutions to further understand their corresponding quantum
369: counterparts.
370:
371:
372: In conclusion, we have considered the quantum and classical problems of a
373: free particle in wedge billiards that exhibit fractional angular momentum due to
374: the breaking of rotational invariance. We found
375: a new duality relation between the ``fractional'' indices for the
376: angular momentum Bessel functions for the sector and its complement
377: pacman billiard. We showed that $\hat L_z^2$
378: becomes a good quantum operator that can be used to characterize
379: the fractional angular momentum eigenvalues of these billiards.
380: We will treat elsewhere the classical and quantum dynamics of a charged
381: particle in wedge billiards in the presence of a
382: constant homogeneous magnetic field. In that case there is a full transition from
383: integrable to chaotic behavior as the magnetic field increases and new states
384: of chaotic whispering gallery modes appear in the large field limit.
385:
386: \thanks
387:
388: The work by AGT has been supported in part by CONACYT 3047P and UNAM sabbatical
389: grants, M\'exico. JVJ thanks NSF for partial financial support.
390: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
391: \newpage
392: %%%Bibliography%%%%%%%%%
393: \begin{references}
394: \bibitem{casati} G. Casati and C. Chirikov, ``Quantum Chaos, between order and disorder".
395: Cambridge University Press (1995), Cambridge, UK.
396:
397: \bibitem{bohigas} O. Bohigas, M. J. Giannoni and C. Schmit, {\sl Phys.
398: Rev. Lett.} {\bf 52}, 1 (1984).
399:
400:
401: \bibitem{houches2}
402: {M. J. Giannoni, A. Voros, and J. Zinn-Justin}.
403: \newblock In {\em { ``Chaos and Quantum Physics"}}, {\bf 52}, (Les Houches
404: Lecture Series). North-Holland, Amsterdam, (1991)
405:
406: \bibitem{stock} H.J. St\"{o}ckman. Quantum Chaos an introduction.
407: Cambridge University Press, (1999), Cambridge, UK.
408:
409: \bibitem{berry1}
410: {M. J. Berry II,}~{J. A. Katine} and {A. C. Gossard}.
411: \newblock {\em Phys. Rev. B}, {\bf 50}, 17721 (1994).
412:
413: \bibitem{berry2}
414: {M. J. Berry II},
415: \newblock {\em {``Mesoscopic Transport and quantum chaos in ballistic quantum
416: dots"}}.
417: \newblock PhD thesis, Harvard University, (1994).
418:
419: \bibitem{wedge} T. Zereadi and D. A. Goodings, Phys. Rev. {\bf E48},
420: 3518 (1993).
421:
422: \bibitem{linda} R. Suhan and L.E. Reichl, Phys. Rev. {\bf E60}, 1607
423: (1999).
424:
425:
426: \bibitem{groups} A. Kirillov, {\em {``\'El\'ements de la th\'eorie des
427: repr\'esentations"}}. Editions MIR, Moscu, (1974).
428:
429: \bibitem{robinett} R. W. Robinett, J. Math. Phys. {\bf 39}, 278 (1998).
430:
431:
432: \end{references}
433: \newpage
434:
435: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
436: %% %%
437: %% Figures %%
438: %% %%
439: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
440: %%
441: %%
442: %%
443: \vskip 1cm
444: \begin{figure}[hp]\begin{center}
445: \caption{
446: Here we show the orbit $(16,3)$ for the $\frac{\pi}{4}$ pacman
447: described in the text. The particle strikes the outer circle 16
448: times in 3 rotations to close the orbit completely. The total angle
449: covered in this orbit is equal to $6\cdot \frac{7\pi}{4}$.
450: \label{fig1}}
451: \end{center}\end{figure}
452:
453:
454:
455: \begin{figure}[hp]\begin{center}
456: \caption{
457: In this figure we show the wave function density for the
458: $\frac{\pi}{4}$ pacman that geometrically corresponds to the classical
459: orbit of Fig.(1). Note that the number of small and large triangles,
460: as well as the radius of the inner caustic,
461: agree with the ones shown in Fig. (1).
462: \label{fig2}}
463: \end{center}\end{figure}
464:
465:
466:
467:
468:
469:
470:
471:
472:
473: \end{document}
474: