1: %%%% ivrs.tex
2: %%%% Written with REVTeX 4 on LaTeX2e
3: %%%% Modified apssamp.tex provided the REVTeX 4 package
4:
5: %%%
6: %%% styles
7: %%%
8: \documentclass[%
9: %%draft,%
10: preprint,
11: %%twocolumn,preprintnumbers,%
12: aps,pra,showpacs,amsmath,amssymb]{revtex4}
13: \usepackage{graphicx}
14: \bibliographystyle{apsrev}
15: %\usepackage{bm}% bold math
16:
17: %%%
18: %%% Private macros
19: %%%
20: \newcommand{\ket}[1]{| #1 \rangle}
21: \newcommand{\bra}[1]{\langle #1 |}
22: \newcommand{\bracket}[2]{\langle #1 | #2 \rangle}
23:
24:
25: %\nofiles
26:
27:
28: \begin{document}
29: %%%
30: %%% Title page
31: %%%
32: %\preprint{{\sf DRAFT (\today)}
33: % %% PhysTMU-NLIN-2005-XX
34: %}
35:
36: \title{Determination of the border between ``shallow'' and ``deep''
37: tunneling regions for Herman-Kluk method
38: by asymptotic approach}
39: \author{Atushi Tanaka}
40: \email{tanaka@phys.metro-u.ac.jp}
41: %%\homepage{http://www.comp.metro-u.ac.jp/~atanaka/}
42: \affiliation{Department of Physics, Tokyo Metropolitan University,
43: Minami-Osawa, Hachioji, Tokyo 192-0397, Japan}
44: %\date{\today}
45:
46: %%% (within 600 chars)
47: \begin{abstract}
48: The evaluation of a tunneling tail by the Herman-Kluk method, which is a
49: quasiclassical way to compute quantum dynamics, is examined by asymptotic
50: analysis. In the shallower part of the tail, as well as in the classically
51: allowed region, it is shown that the leading terms of semiclassical
52: evaluations of quantum
53: theory and the Herman-Kluk formula agree, which is known as an {\em asymptotic
54: equivalence}. In the deeper part, it is shown that the
55: asymptotic equivalence breaks down, due to the emergence of unusual
56: ``tunneling trajectory'', which is an artifact of the Herman-Kluk method.
57: \end{abstract}
58:
59: \pacs{03.65.Sq, 05.45.Mt, 02.70.-c}
60: %% PACS 2003
61: %% 03.65.Sq Semiclassical theories and applications
62: %% 05.45.Mt Quantum chaos; semiclassical methods
63: %% 02.70.-c Computational techniques
64:
65: \maketitle
66:
67:
68: %%%
69: %%% Start of the main text
70: %%%
71:
72: %%%
73: %%% Introduction
74: %%%
75: Even nowadays, it is still impossible to carry out serious
76: numerical investigation of quantum dynamics, even
77: for modest (e.g., 10--100) degrees of freedom systems,
78: with the present state of the art of computational technology,
79: unless we take drastic approximations that need to be based
80: on good physical insights.
81: As a starting point to invent such a method, it is often employed
82: the semiclassical approximation (i.e. asymptotic evaluation) of
83: the path integral representation of
84: time evolution operator (Feynman kernel)~\cite{Schulman:1981}.
85: The semiclassical approximation, however, has difficulties due to
86: the exponential proliferation of contributing trajectories and
87: caustics (see, e.g. Ref.\ \cite{Tomsovic:PRE-1993-282}) and
88: due to Stokes phenomenon that requires to remove
89: non-contributing complex-valued trajectories~\cite{Adachi:AP-195-45},
90: even in few degrees of freedom systems,
91: when the corresponding classical system is chaotic.
92: %
93: Furthermore, in the semiclassical method, the boundary conditions
94: of the contributing classical trajectories both for initial and
95: finial times, is troublesome (known as a root-search problem~%
96: \cite{Miller:JCP-53-3578,Miller:JCP-56-5668}),
97: in computations for realistic systems such as atoms and molecules.
98: %%
99: In order to avoid the root-search problem,
100: initial value representations (IVRs) of Feynman kernels are proposed.
101: The IVRs imposes only the initial conditions on the classical trajectories.
102: %%
103: In Ref.\ \cite{Miller:JCP-53-3578}, the earliest version of IVR is
104: introduced by a change of integral variables to semiclassical
105: Feynman kernel.
106: A general framework of IVR is proposed by Kay~\cite{Kay:JCP-100-4377},
107: who discussed
108: various integral expressions (IEs) of approximate Feynman kernel,
109: which are composed by classical trajectories that are emitted from
110: real-valued initial conditions. The important guiding principle of
111: Kay's theory is that the leading semiclassical expressions of both an
112: IE and exact quantum theory must agree. This is called
113: {\em the asymptotic equivalence}%
114: ~\cite{Campolieti:JCP-98-5969,Kay:JCP-100-4377}.
115: Furthermore, Kay argued that several
116: known IVRs (IEs), including
117: thawed Gaussian approximation~\cite{Heller:JCP-62-1544},
118: cellular dynamics~\cite{Heller:JCP-94-2723}, and
119: the Herman-Kluk (HK) formula~\cite{Herman:CP-91-27},
120: are asymptotically equivalent with quantum theory.
121: %%
122: %% Why HK?
123: %%
124: Nowadays, a lot of numerical investigations of quantum dynamics,
125: including rather realistic systems, employ IVRs, in particular,
126: the Herman-Kluk method~\cite{ReviewHK}.
127:
128:
129:
130: The limitation of IVRs, however, is not clear~\cite{HKdisscussion},
131: in particular, in the
132: descriptions of classically forbidden processes, e.g., tunneling processes,
133: whose conventional semiclassical treatments need to take into account
134: the contributions from complex-valued classical trajectories.
135: %%
136: Numerical experiments to reproduce tunneling tails by IVR approaches
137: suggest that the ``shallow'' side of tunneling tails is tractable~%
138: \cite{Keshavamurthy:CPL-218-189}.
139: %%
140: On the other hand, concerning to the ``deep'' side, the IVR approaches have
141: difficulties.
142: Kay's {\em semiclassical} analysis of $\mathcal{O}(\hbar^2)$ error term of
143: Herman-Kluk method reveals that the magnitude of the error is
144: controlled by the {\em complex-valued} classical
145: trajectories~\cite{Kay:JCP-107-2313}.
146: %%
147: However, these works are not conclusive.
148: %%
149: First, since the tunneling tails are exponentially small, the corresponding
150: error analysis also requires to treat {\em exponentially small errors}.
151: Hence, the $\mathcal{O}(\hbar^2)$ error term, which is satisfactory in
152: classically allowed region, is too large.
153: %%
154: Second, the border between the shallow side and the deep side of the tunneling
155: tail has been unknown.
156: In order to clarify the limitation of IVR, the identification of the border
157: is inevitable.
158: %%
159: We remind that asymptotic (i.e., semiclassical) analysis has an ability
160: to treat exponentially small quantities. This suggests some asymptotic
161: approach may reveal the limitation of IVRs with much better accuracy.
162:
163:
164: In this paper, we examine an evaluation of a tunneling tail, by the
165: Herman-Kluk formula, with asymptotic analysis.
166: %%
167: Here we
168: focus on the Feynman kernel, rather than energy spectra or correlation
169: functions. This facilitates to identify the origin of discrepancies.
170:
171: %%
172: %% End of introduction
173: %%
174:
175: We here examine a single degree of freedom system that is described by
176: a Hamiltonian $H = - g p^3 / 3$, where $q$ and $p$ are the position and
177: the momentum of the system, respectively. We assume that the strength
178: of folding $g$ is positive.
179: %%
180: This is a canonical model that describes a
181: nonlinear folding process in classical phase space
182: (see, Fig.\ \ref{fig:manifolds} (a))%
183: ~\cite{Adachi:AP-195-45}.
184: When both the initial and the final states of Feynman kernel are
185: eigenstates of the position operator, the nonlinear folding dynamics
186: produces a caustic.
187: %%
188: Note that in generic, nonlinear systems induce foldings in general.
189: %, even if we carefully choose initial and
190: %final states of Feynman kernel.
191: Dynamics locally around each foldings are described by the canonical
192: Hamiltonian with appropriate rotations and
193: scaling in phase space (see, e.g., Fig.~\ref{fig:manifolds}~(b)).
194:
195:
196: \begin{figure}
197: \includegraphics[width=8.6cm]{fig1a}
198: \hfill
199: \includegraphics[width=8.6cm]{fig1b}
200: \caption{\label{fig:manifolds}
201: Evolutions of classical manifolds in phase space.
202: (a)~With the folding Hamiltonian $H = - g p^3 / 3$~\cite{Adachi:AP-195-45}.
203: The initial manifold (thin line) is at $q = 0$. After a time
204: interval $\tau$, the manifold folds (thick line).
205: The caustic in the position representation is at $q = 0$.
206: (b)~With a nonlinear oscillator $H = \frac{1}{2} p^2 + V(q)$,
207: where
208: $V(q) = D \{(1 - e^{-\lambda q})^2 - 1\} + \frac{1}{2}(1-\epsilon) q^2$,
209: $\epsilon=0.975$, $\lambda = 1/\sqrt{12}$, and
210: $D = \epsilon / (2 \lambda^2)$ (Contour lines of $V(q)$ are thin)%
211: ~\cite{Brickmann:JCP-75-5744}.
212: The initial manifold $\{(q, p)| q = 9, -3 < p < 3\}$ (vertical line)
213: mimics an eigenstate of the position operator, with an energy cutoff.
214: The corresponding final manifold (thick line), at $t=18$,
215: has two prominent caustics in the position representation .
216: }
217: \end{figure}
218:
219:
220:
221: The Feynman kernel of a time evolution of interval $[0,\tau]$
222: ($\tau > 0$) in the position representation
223: $K(q) \equiv \bra{q}\exp(-i H\tau/\hbar)\ket{q=0}$
224: is expressed exactly with Airy function
225: \begin{equation}
226: K(q) = A_i (q/l) / l
227: \end{equation}
228: where $l\equiv(\hbar^2 g \tau)^{1/3}$ is a characteristic length for
229: a penetration into the classically forbidden region $q > 0$.
230: The asymptotic evaluation
231: of the integral representation of the Feynman kernel
232: \begin{equation}
233: \label{eq:exactIE}
234: K(q) = \int dp \bracket{q}{p}e^{+i g p^3 \tau/ (3\hbar)}\bracket{p}{q=0}
235: \end{equation}
236: gives a leading semiclassical approximation $K_{\rm SC}$, which are
237: composed by classical trajectories.
238: %%
239: On one hand, in the classically allowed region $q < 0$, we have
240: a superposition of incoming and outgoing waves:
241: \begin{equation}
242: K_{\rm SC}(q) =
243: \frac{1}{\sqrt{\pi} l (|q|/l)^{1/4}}
244: \cos\left\{\frac{2}{3}\left(\frac{|q|}{l}\right)^{3/2}-\frac{\pi}{4}
245: \right\}.
246: \end{equation}
247: The corresponding classical trajectories are characterized by their
248: momentum, which are conserved quantities:
249: $p_{\pm}(q) = \pm\sqrt{|q|/(\tau g)}$.
250: On the other hand, in the classically forbidden region $q > 0$,
251: we have a tunneling tail:
252: \begin{equation}
253: \label{eq:KSCtail}
254: K_{\rm SC}(q) =
255: \frac{1}{2\sqrt{\pi} l (q/l)^{1/4}}
256: \exp\left\{-\frac{2}{3}\left(\frac{q}{l}\right)^{3/2}\right\}.
257: \end{equation}
258: The momentum of the tunneling trajectory is pure imaginary:
259: $p_0(q) = i \sqrt{q/(\tau g)}$.
260: Between these regions, at $q=0$, these
261: classical trajectories merge to produce a caustic.
262: %%
263: The change of asymptotic expansions for different signs of $q$
264: is controlled by Stokes phenomenon~\cite{Berry:RPP-35-315}.
265:
266:
267: In the following, the corresponding Herman-Kluk kernel~%
268: \cite{Herman:CP-91-27} is examined:
269: \begin{eqnarray}
270: \label{eq:HK}
271: K^{\rm HK}(q) &=&
272: \int \int \frac{dq_0 dp_0}{2\pi\hbar}
273: \bracket{q}{\varphi^{\gamma}(q_{\tau}, p_{\tau})} C(q_0, p_0, \tau)
274: \nonumber \\
275: &&{}\times
276: e^{i S_{\tau}(q_0, p_0)/\hbar}
277: \bracket{\varphi^{\gamma}(q_0, p_0)}{q = 0},
278: \end{eqnarray}
279: where $\gamma$ ($>0$) determines the width of Gaussian packet
280: $\bracket{q}{\varphi^{\gamma}(q_0, p_0)}
281: = (2\gamma/\pi)^{1/4}\exp\{-\gamma (q - q_0)^2 + i p_0 (q - q_0)/\hbar\}$,
282: $(q_{\tau}, p_{\tau})$ is the classical trajectory at time
283: $t = \tau$, emitted from $(q_0, p_0)$ at $t =0$,
284: $S_{\tau}(q_0, p_0)$ is the classical action along the time evolution,
285: and
286: $C(q_0, p_0, \tau) =
287: \{\partial q_{\tau}/\partial q_0 + \partial p_{\tau}/\partial p_0
288: - 2i\hbar\gamma \partial q_{\tau}/\partial p_0
289: - (2i\hbar\gamma)^{-1} \partial p_{\tau}/\partial q_0\}^{1/2}/\sqrt{2}$.
290: For the folding Hamiltonian, $C(q_0, p_0, \tau)$ have a branch point
291: at $p_{\rm I} = i / (2\hbar\gamma\tau g)$. After the integration of
292: the variable
293: $q_0$, $K^{\rm HK}(q)$~(\ref{eq:HK}) becomes
294: \begin{equation}
295: \label{eq:HKp}
296: K^{\rm HK}(q)
297: = \int \frac{dp}{2\pi\hbar} C(p, \tau) e^{-\phi_{\tau}(p)},
298: \end{equation}
299: where $C(p,\tau) = (1 - p/p_{\rm I})^{1/2}$ and
300: $\phi_{\tau}(p) =
301: \gamma (q + \tau g p^2)^2/2 - i p q /\hbar -i \tau g p^3 / (3\hbar)$.
302: The integral~(\ref{eq:HKp}) has three saddle points
303: $p = \pm\sqrt{- q/(\tau g)}$ and $p_{\rm I}$. The former momenta
304: $p = \pm\sqrt{- q/(\tau g)}$ correspond to
305: the classical momenta $p_{\pm}(q)$ in the classically allowed region and
306: $p_0(q)$ in the classically forbidden region.
307: The latter momentum $p_{\rm I}$, which is
308: the branch
309: point of $C(p,\tau)$, appears only in the semiclassical
310: analysis of $K^{\rm HK}(q)$.
311: Note that all the saddle points need not to make contributions
312: to the semiclassical kernel, due to the Stokes phenomena.
313: Actually,
314: in the classically allowed region $q>0$, there are the contributions
315: only from $p = p_{\pm}(q)$ (FIG.~\ref{fig:integrationPath}(a)).
316: Hence the leading semiclassical
317: evaluation of $K^{\rm HK}(q)$ agree with $K_{\rm SC}(q)$. Thus the
318: asymptotic equivalence between Herman-Kluk kernel
319: and quantum theory holds for $q<0$~\cite{Kay:JCP-100-4377}.
320:
321:
322: \begin{figure}
323: \includegraphics[width=0.3\textwidth]{fig2a}
324: \hfill
325: \includegraphics[width=0.3\textwidth]{fig2b}
326: \hfill
327: \includegraphics[width=0.3\textwidth]{fig2c}
328: \caption{\label{fig:integrationPath}
329: Locations of saddle points (closed and open circles) and integration paths
330: (thick lines) for the asymptotic evaluations of
331: the integral~(\ref{eq:HKp}),
332: for (a)~the classical region $q<0$,
333: (b)~the shallow tunneling region $0<q<l_{\gamma}$, and,
334: (c)~the deep tunneling region $q>l_{\gamma}$.
335: Dashed lines emanating from the branch point $p_{\rm I}$ are
336: branch cuts.
337: Closed and open circles are contributing and non-contributing
338: saddle points, respectively.
339: }
340: \end{figure}
341:
342:
343: In the semiclassical evaluation of $K^{\rm HK}(q)$ in the classically
344: forbidden region $q>0$, there is a length scale
345: $l_{\gamma} = \gamma^{-2} l^{-3} / 4$ that divides the tunneling tail
346: into two regions: a ``shallow'' region $0 < q < l_{\gamma}$ and
347: a ``deep'' region $q > l_{\gamma}$.
348: In the shallow region, the semiclassical evaluation of $K^{\rm HK}(q)$
349: has only a single contribution from the classical trajectory
350: $p = p_0(q)$ (FIG.~\ref{fig:integrationPath}(b)).
351: Hence the asymptotic equivalence between
352: Herman-Kluk kernel and quantum theory holds both classically allowed
353: region and the shallow tunneling region $q < l_{\gamma}$.
354: This is a promising evidence that collections of real classical
355: trajectories can describe {\em shallow} tunneling tails, through
356: IVR techniques.
357: This is the first example to show that IVR technique can
358: describe a classically forbidden process with an analytical argument.
359:
360:
361: On the other hand, in the ``deep'' tunneling region, we found a
362: discrepancy: Due to Stokes phenomenon, the contribution from
363: ``conventional'' tunneling trajectory $p = p_0(q)$ disappears, and in
364: turn, the contribution from the classical trajectory $p = p_{\rm I}$,
365: which is an artifact of the Herman-Kluk kernel, appears
366: (FIG.~\ref{fig:integrationPath}(c)).
367: The resultant semiclassical evaluation of $K^{\rm HK}(q)$ is
368: \begin{eqnarray}
369: \label{eq:HKSCtail}
370: K^{\rm HK}_{\rm SC}(q) &=&
371: \frac{\Gamma(3/4)}{2\pi l (\gamma l^2)^{1/4}}
372: \left(\frac{l}{q - l_{\gamma}}\right)^{3/4}
373: \nonumber\\
374: &&{}\times
375: \exp\left\{-\frac{\gamma}{2}(q + l_{\gamma})^2 -
376: \frac{4}{3}l_{\gamma}^2\right\}.
377: \end{eqnarray}
378: Note that $q = l_{\gamma}$ is a caustic, which is an reminiscent of
379: the Stokes phenomena. At the same time, the asymptotic form of
380: tunneling tail $\sim \exp(-\gamma q^2 /2)$ for $q \gg l_{\gamma}$,
381: which sensitively depends on $\gamma$, is qualitatively different from
382: $K_{\rm SC}(q)\sim\exp\{-2 (q/l)^{3/2}/3\}$~(\ref{eq:KSCtail}).
383: Thus the breakdown of the asymptotic equivalence is evident.
384:
385:
386: \begin{figure}
387: \includegraphics[width=8.6cm]{fig3a}
388: \hfill
389: \includegraphics[width=8.6cm]{fig3b}
390: \caption{\label{fig:compareK}
391: Comparisons of exact, Herman-Kluk (HK), and semiclassical
392: Herman-Kluk (HKSC) evaluations of the Feynman kernel $K(q)$,
393: in particular,
394: its tunneling tail $q > 0$. The penetration depth is $l = 1$.
395: (a) With $l_{\gamma} = 1$, all three theories are shown.
396: At the (conventional) turning point $q=0$, the semiclassical
397: Herman-Kluk encounters caustic.
398: In the shallow region $0 < q < l_{\gamma}$, the
399: discrepancy between quantum theory and Herman-Kluk method is not
400: significant. On the other hand, at $q=l_{\gamma}$, the semiclassical
401: Herman-Kluk encounters another caustic. Note that the conventional
402: semiclassical theory meets the caustic at $q=0$, the conventional
403: turning point, only.
404: %%
405: In the deep region $q > l_{\gamma}$,
406: the tunneling tails of quantum theory and Herman-Kluk
407: formula take qualitatively different shape
408: (see, Eq.\ (\ref{eq:KSCtail}) and Eq.\ (\ref{eq:HKSCtail})).
409: (b) Comparison of quantum theory and Herman-Kluk formula, with several
410: values of $l_{\gamma}$.
411: In order to reproduce the shape of deep tunneling tail
412: by Herman-Kluk method, we need larger $l_{\gamma}$~\cite{Kay:JCP-107-2313}.
413: }
414: \end{figure}
415:
416:
417: In the argument above, the discrepancy between Herman-Kluk kernel and
418: Feynman kernel comes from (1) a nonlinear folding dynamics in the
419: corresponding classical phase space, and (2) the appearance of
420: artificial tunneling trajectory. When the folding dynamics is not
421: significant (this is the case before Ehrenfest time), there is a
422: workaround to remove the contribution from the artificial tunneling
423: trajectories, by
424: adjusting the value of $\gamma$ in the Herman-Kluk kernel
425: to push $p_{\rm I}$ deeper in the complex plane
426: (see., Fig.\ \ref{fig:integrationPath}).
427: %%
428: Indeed, this is a known strategy, which is proposed by Kay,
429: to reduce the magnitude of the error
430: of Herman-Kluk method, and the strategy succeeds to
431: a certain extent~\cite{Kay:JCP-107-2313,Maitra:JPC-112-531}.
432: However, in generic, nonlinear systems, it will become
433: difficult to carry out such workarounds in practice, due to
434: the emergence of multiple foldings
435: (see, e.g., Fig.\ \ref{fig:manifolds}~(b)).
436:
437: We summarize this paper.
438: With an exactly solvable
439: model that describes nonlinear folding process in corresponding
440: classical phase-space dynamics, we identified a boundary between a
441: shallow and deep tunneling regions for the Herman-Kluk kernel.
442: In the former region, the Herman-Kluk kernel and quantum theory are
443: asymptotically equivalent. Hence there remains a hope that Herman-Kluk
444: kernel describes classically forbidden process to a certain extent.
445: However, in the deep region, the breakdown of the asymptotic
446: equivalent is shown.
447: %%
448: Besides Herman-Kluk method, other IVR approaches, in particular, which
449: are based on Kay's framework~\cite{Kay:JCP-100-4377}, will have similar
450: scenario on successes and failures in descriptions of classically
451: forbidden phenomena.
452: %%
453: We expect that the nonlinear folding Hamiltonian employed
454: here is a good test-case not only for the Herman-Kluk method, but
455: also for various semiclassical method. We remind that this model was
456: employed to elucidate the limitation of single trajectory approximation of
457: semiclassical coherent-state path integrals~\cite{Adachi:AP-195-45,%
458: Klauder:PRL-1986-897}.
459: %%
460: At the same time, it is highly desirable to develop a convenient way to
461: find the boundary between the shallow and the deep regions of
462: Herman-Kluk method, though the present analysis requires a
463: full knowledge of the Stokes phenomena to determine the boundary.
464:
465:
466: \begin{acknowledgments}
467: This work is partially supported by the Grant-in-Aid for Young Scientists (B)
468: (No.~15740241) from the Ministry of Education, Culture, Sports, Science
469: and Technology, Japan.
470: %% A.~T. thanks Professor A.~Shudo for useful conversations.
471: \end{acknowledgments}
472:
473:
474: %% for preparation, I prefer to use BibTeX
475: %\bibliography{ivr}
476:
477: %% for submission, we need to embed the .bbl file here
478: \begin{thebibliography}{21}
479: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
480: \expandafter\ifx\csname bibnamefont\endcsname\relax
481: \def\bibnamefont#1{#1}\fi
482: \expandafter\ifx\csname bibfnamefont\endcsname\relax
483: \def\bibfnamefont#1{#1}\fi
484: \expandafter\ifx\csname citenamefont\endcsname\relax
485: \def\citenamefont#1{#1}\fi
486: \expandafter\ifx\csname url\endcsname\relax
487: \def\url#1{\texttt{#1}}\fi
488: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
489: \providecommand{\bibinfo}[2]{#2}
490: \providecommand{\eprint}[2][]{\url{#2}}
491:
492: \bibitem[{\citenamefont{Schulman}(1981)}]{Schulman:1981}
493: \bibinfo{author}{\bibfnamefont{L.~S.} \bibnamefont{Schulman}},
494: \emph{\bibinfo{title}{Techniques and applications of path integration}}
495: (\bibinfo{publisher}{John Wiley \& Sons}, \bibinfo{address}{New York},
496: \bibinfo{year}{1981}).
497:
498: \bibitem[{\citenamefont{Tomsovic and Heller}(1993)}]{Tomsovic:PRE-1993-282}
499: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Tomsovic}} \bibnamefont{and}
500: \bibinfo{author}{\bibfnamefont{E.~J.} \bibnamefont{Heller}},
501: \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{47}},
502: \bibinfo{pages}{282} (\bibinfo{year}{1993}).
503:
504: \bibitem[{\citenamefont{Adachi}(1989)}]{Adachi:AP-195-45}
505: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Adachi}},
506: \bibinfo{journal}{Ann. Phys. (N. Y.)} \textbf{\bibinfo{volume}{195}},
507: \bibinfo{pages}{45} (\bibinfo{year}{1989}).
508:
509: \bibitem[{\citenamefont{Miller}(1970)}]{Miller:JCP-53-3578}
510: \bibinfo{author}{\bibfnamefont{W.~H.} \bibnamefont{Miller}},
511: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{53}},
512: \bibinfo{pages}{3578} (\bibinfo{year}{1970}).
513:
514: \bibitem[{\citenamefont{Miller and George}(1972)}]{Miller:JCP-56-5668}
515: \bibinfo{author}{\bibfnamefont{W.~H.} \bibnamefont{Miller}} \bibnamefont{and}
516: \bibinfo{author}{\bibfnamefont{T.~F.} \bibnamefont{George}},
517: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{56}},
518: \bibinfo{pages}{5668} (\bibinfo{year}{1972});
519: %%
520: \bibinfo{author}{\bibfnamefont{W.~H.} \bibnamefont{Miller}},
521: \bibinfo{journal}{Adv. Chem. Phys.} \textbf{\bibinfo{volume}{25}},
522: \bibinfo{pages}{69} (\bibinfo{year}{1974}).
523:
524: \bibitem[{\citenamefont{Kay}(1994)}]{Kay:JCP-100-4377}
525: \bibinfo{author}{\bibfnamefont{K.~G.} \bibnamefont{Kay}}, \bibinfo{journal}{J.
526: Chem. Phys.} \textbf{\bibinfo{volume}{100}}, \bibinfo{pages}{4377}
527: (\bibinfo{year}{1994}).
528:
529: \bibitem[{\citenamefont{Campolieti and Brumer}(1992)}]{Campolieti:JCP-98-5969}
530: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Campolieti}} \bibnamefont{and}
531: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Brumer}},
532: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{96}},
533: \bibinfo{pages}{5969} (\bibinfo{year}{1992}).
534:
535: \bibitem[{\citenamefont{Heller}(1975)}]{Heller:JCP-62-1544}
536: \bibinfo{author}{\bibfnamefont{E.~J.} \bibnamefont{Heller}},
537: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{62}},
538: \bibinfo{pages}{1544} (\bibinfo{year}{1975}).
539:
540: \bibitem[{\citenamefont{Heller}(1991)}]{Heller:JCP-94-2723}
541: \bibinfo{author}{\bibfnamefont{E.~J.} \bibnamefont{Heller}},
542: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{94}},
543: \bibinfo{pages}{2723} (\bibinfo{year}{1991}).
544:
545: \bibitem[{\citenamefont{Herman and Kluk}(1984)}]{Herman:CP-91-27}
546: \bibinfo{author}{\bibfnamefont{M.~F.} \bibnamefont{Herman}} \bibnamefont{and}
547: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Kluk}},
548: \bibinfo{journal}{Chem. Phys.} \textbf{\bibinfo{volume}{91}},
549: \bibinfo{pages}{27} (\bibinfo{year}{1984}).
550:
551: %\bibitem[{\citenamefont{Kay}(2001)}]{Kay:JPCA-105-2535}
552: %\bibinfo{author}{\bibfnamefont{K.~G.} \bibnamefont{Kay}}, \bibinfo{journal}{J.
553: % Phys. Chem. A} \textbf{\bibinfo{volume}{105}}, \bibinfo{pages}{2535}
554: % (\bibinfo{year}{2001}).
555: \bibitem{ReviewHK}
556: \bibinfo{author}{\bibfnamefont{F.} \bibnamefont{Grossmann}},
557: \bibinfo{journal}{Comment At. Mol. Phys.} \textbf{\bibinfo{volume}{34}},
558: \bibinfo{pages}{3} (\bibinfo{year}{1999});
559: \bibinfo{author}{\bibfnamefont{K.~G.} \bibnamefont{Kay}},
560: \bibinfo{journal}{J. Phys. Chem. A} \textbf{\bibinfo{volume}{105}},
561: \bibinfo{pages}{2535} (\bibinfo{year}{2001});
562: \bibinfo{author}{\bibfnamefont{W.~H.} \bibnamefont{Miller}},
563: \bibinfo{journal}{J. Phys. Chem. A} \textbf{\bibinfo{volume}{105}},
564: \bibinfo{pages}{2942} (\bibinfo{year}{2001});
565:
566: \bibitem{HKdisscussion}
567: The argument presented in this paper is not immediately relevant
568: with the recent discussion on the validity of the Herman-Kluk method
569: [M. Baranger et al., J. Phys. A {\bf 34}, 7227 (2001);
570: F. Grossman and M. F. Herman, {\em ibid.} {\bf 35}, 9489 (2002);
571: M. Baranger et al., {\em ibid.} {\bf 35}, 9493 (2002);
572: M. Baranger, M. A. M. de Aguiar and H. J. Korsch,
573: {\em ibid.} {\bf 36}, 9795 (2002)].
574:
575: \bibitem[{\citenamefont{Keshavamurthy and
576: Miller}(1994)}]{Keshavamurthy:CPL-218-189}
577: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Keshavamurthy}}
578: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.~H.}
579: \bibnamefont{Miller}}, \bibinfo{journal}{Chem. Phys. Lett.}
580: \textbf{\bibinfo{volume}{218}}, \bibinfo{pages}{189} (\bibinfo{year}{1994});
581: %%\bibitem[{\citenamefont{Grossmann and Heller}(1995)}]{Grossmann:CPL-241-1995}
582: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Grossmann}} \bibnamefont{and}
583: \bibinfo{author}{\bibfnamefont{E.~J.} \bibnamefont{Heller}},
584: \bibinfo{journal}{Chem. Phys. Lett.} \textbf{\bibinfo{volume}{241}},
585: \bibinfo{pages}{45} (\bibinfo{year}{1995});
586: %%\bibitem[{\citenamefont{Zor and Kay}(1995)}]{Zor:PRL-76-1990}
587: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Zor}} \bibnamefont{and}
588: \bibinfo{author}{\bibfnamefont{K.~G.} \bibnamefont{Kay}},
589: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{76}},
590: \bibinfo{pages}{1990} (\bibinfo{year}{1995}).
591:
592: \bibitem[{\citenamefont{Kay}(1997)}]{Kay:JCP-107-2313}
593: \bibinfo{author}{\bibfnamefont{K.~G.} \bibnamefont{Kay}}, \bibinfo{journal}{J.
594: Chem. Phys.} \textbf{\bibinfo{volume}{107}}, \bibinfo{pages}{2313}
595: (\bibinfo{year}{1997}).
596:
597: \bibitem[{\citenamefont{Brickmann and Russegger}(1981)}]{Brickmann:JCP-75-5744}
598: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Brickmann}} \bibnamefont{and}
599: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Russegger}},
600: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{75}},
601: \bibinfo{pages}{5744} (\bibinfo{year}{1981});
602: %%\bibitem[{\citenamefont{Kluk et~al.}(1986)\citenamefont{Kluk, Herman, and Davis}}]{Kluk:JCP-84-326}
603: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Kluk}},
604: \bibinfo{author}{\bibfnamefont{M.~F.} \bibnamefont{Herman}},
605: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{H.~L.} \bibnamefont{Davis}},
606: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{84}},
607: \bibinfo{pages}{326} (\bibinfo{year}{1986}).
608:
609: \bibitem[{\citenamefont{Berry and Mount}(1972)}]{Berry:RPP-35-315}
610: \bibinfo{author}{\bibfnamefont{M.~V.} \bibnamefont{Berry}} \bibnamefont{and}
611: \bibinfo{author}{\bibfnamefont{K.~E.} \bibnamefont{Mount}},
612: \bibinfo{journal}{Rep. Prog. Phys.} \textbf{\bibinfo{volume}{35}},
613: \bibinfo{pages}{315} (\bibinfo{year}{1972}).
614:
615: \bibitem[{\citenamefont{Maitra}(2000)}]{Maitra:JPC-112-531}
616: \bibinfo{author}{\bibfnamefont{N.~T.} \bibnamefont{Maitra}},
617: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{112}},
618: \bibinfo{pages}{531} (\bibinfo{year}{2000}).
619:
620: \bibitem[{\citenamefont{Klauder}(1986)}]{Klauder:PRL-1986-897}
621: \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Klauder}},
622: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{56}},
623: \bibinfo{pages}{897} (\bibinfo{year}{1986}).
624:
625: \end{thebibliography}
626:
627: %% end of .bbl
628:
629: \end{document}
630: %%
631: %% End of ivrs.tex
632: %%
633: