1: % -------------------------------------------------------------

2: % Surzhykov, Lubasch, Zinn-Justin and Jentschura 

3: % -------------------------------------------------------------

4: % Quantum Dot Potentials: 

5: % Symanzik Scaling, Resurgent Expansions and Quantum Dynamics

6: % -------------------------------------------------------------

7: % Submission to Physical Review B 

8: % ------------------------------------------------------------

9: %\documentclass[preprint,superscriptaddress,aps,prb,showpacs,floatfix]{revtex4}

10: \documentclass[twocolumn,superscriptaddress,aps,prb,showpacs,floatfix]{revtex4}

11: 

12: 

13: %\usepackage{showkeys}

14: 

15: \usepackage{amsmath}

16: \usepackage{amssymb}

17: \usepackage{amsfonts}

18: \usepackage{pifont}

19: \usepackage{colordvi}

20: \usepackage{bm}

21: \usepackage{bbm}

22: \usepackage{epsf}

23: \usepackage{epsfig}

24: \usepackage{times}

25: \usepackage{nicefrac}

26: \usepackage{colordvi}

27: \usepackage{color}

28: \usepackage{graphicx}

29: \usepackage{dcolumn}

30: %\newcolumntype{.}{D{x}{}{-1}}

31: 

32: %

33: % Our definitions

34: %

35: 

36: \def\sprm#1#2{  \left\langle #1 \left\vert \right. #2 \right\rangle   }

37: \def\mem#1#2#3{  \left\langle #1 \left\vert  #2 \right\vert #3 \right\rangle   }

38: 

39: \bibliographystyle{myprsty}

40: 

41: \begin{document}

42: 

43: \preprint{Version 3.1}

44: 

45: \title{Quantum Dot Potentials: Symanzik Scaling, 

46: Resurgent Expansions and Quantum Dynamics}

47: 

48: \author{Andrey Surzhykov}

49: \affiliation{Max--Planck--Institut f\"ur Kernphysik,

50: Saupfercheckweg 1, 69117 Heidelberg, Germany}

51: 

52: \author{Michael Lubasch}

53: \affiliation{Max--Planck--Institut f\"ur Kernphysik,

54: Saupfercheckweg 1, 69117 Heidelberg, Germany}

55: 

56: \author{Jean Zinn--Justin}

57: \affiliation{DAPNIA, Commissariat \`{a} l'\'{E}nergie Atomique,

58: Centre de Saclay, 91191 Gif--Sur--Yvette, France}

59: 

60: \author{Ulrich D.~Jentschura\footnote{e-mail: ulj@mpi-hd.mpg.de}$^{,}$}

61: \affiliation{Max--Planck--Institut f\"ur Kernphysik,

62: Saupfercheckweg 1, 69117 Heidelberg, Germany}

63: 

64: \date{\today}

65: 

66: \begin{abstract}

67: This article is concerned with a special class of the ``double--well--like'' 

68: potentials that occur naturally in the analysis of finite

69: quantum systems. Special attention is paid, in particular, to 

70: the so--called Fokker--Planck potential, which has a particular property:

71: the perturbation series for the ground--state energy vanishes to all orders 

72: in the coupling parameter, but the actual ground-state energy

73: is positive and dominated by instanton configurations of the form 

74: $\exp(-a/g)$, where $a$ is the instanton action. The instanton

75: effects are most naturally taken into account within 

76: the modified Bohr--Sommerfeld quantization conditions whose 

77: expansion leads to the generalized perturbative expansions 

78: (so-called resurgent expansions) for the energy eigenvalues of the 

79: Fokker--Planck potential. Until now, these

80: resurgent expansions have been mainly applied for small values

81: of coupling parameter $g$, while much less attention has been paid to

82: the strong-coupling regime. In this contribution, we compare the 

83: energy values, obtained by directly resumming 

84: generalized Bohr--Sommerfeld quantization conditions, to the strong-coupling 

85: expansion, for which we determine the first few expansion coefficients 

86: in powers of $g^{-2/3}$. Detailed calculations are performed

87: for a wide range of coupling parameters $g$ and indicate a considerable 

88: overlap between the regions of validity of the 

89: weak-coupling resurgent series and 

90: of the strong--coupling expansion. Apart from the analysis of the 

91: energy spectrum of the Fokker--Planck Hamiltonian, we also briefly 

92: discuss the computation of its eigenfunctions. These

93: eigenfunctions may be utilized for the numerical integration of the 

94: (single-particle) time-dependent Schr\"odinger equation and, hence, for 

95: studying the dynamical evolution of the

96: wavepackets in the double--well--like potentials.     

97: \end{abstract}

98: 

99: % 11.15.Bt General properties of perturbation theory

100: % 11.10.Jj Asymptotic problems and properties

101: % 68.65.Hb Quantum dots

102: % 73.21.La Quantum dots

103: % 03.67.-a Quantum information

104: % 03.67.Lx Quantum computation

105: % 85.25.Cp Josephson devices

106: \pacs{11.15.Bt, 11.10.Jj, 68.65.Hb, 73.21.La, 03.67.-a, 03.67.Lx, 85.25.Cp}

107: 

108: \maketitle

109: 

110: % \tableofcontents

111: 

112: \section{Introduction}

113: \label{intro}

114: 

115: In this paper, we study the all-order summation of instanton contributions

116: to the energy eigenvalues of anharmonic quantum mechanical oscillators

117: which involve (almost) degenerate minima.

118: The Euclidean path integral of quantum mechanical systems of

119: this kind with one space-

120: and one time-dimension is dominated by instanton configurations whose

121: action remains finite in the limit of large Euclidean (imaginary) times.

122: In order to find the energy eigenvalues, instanton configurations

123: have to be taken into account.

124: Modified quantization conditions have been conjectured for various

125: classes of potentials (for a review see

126: Refs.~\onlinecite{ZJJe2004i,ZJJe2004ii}), 

127: and accurate numerical calculations have been

128: verified against analytic expansions in the regime of small 

129: coupling~\cite{JeZJ2001,JeZJ2004plb}.

130: Indeed, for small coupling, the energy eigenvalues are dominated

131: by one-, two-, and three-instanton effects which correspond to

132: trajectories of the classical particle with

133: a small number of oscillations between the (almost) degenerate minima

134: of the potential. Here, we are concerned with the all-order

135: summation of the instanton contributions, which is applicable

136: to intermediate and strong coupling. Also, the transition

137: from small to strong coupling, and strong-coupling expansions,

138: will be discussed. Finally, we consider applications in 

139: the quantum dynamical simulation of finite systems.

140: 

141: The purpose of this paper thus is threefold:

142: first, to find generalized perturbative expansions

143: (so-called ``resurgent expansions'') for excited states

144: of certain classes of notoriously problematic~\cite{HeSi1978} 

145: quantum mechanical potentials, second, to derive large-coupling

146: asymptotics for these potentials and to investigate

147: overlap regions between small- and large-coupling asymptotics,

148: and third, to outline applications of the considerations

149: for the quantum dynamical simulation of a single particle 

150: in a double-well-like potential.

151: The first of these purposes is connected with mathematical physics,

152: the second is tied to Symanzik scaling in the 

153: ``poor man's'' variant and therefore, to a basic implementation 

154: of the renormalization group, and the third one is 

155: rather application-oriented.

156: 

157: Within the first and second aims of our investigation, 

158: we also investigate the fundamental question 

159: whether small-coupling perturbative expansions 

160: can be continued analytically to the regime of 

161: large coupling, if one includes instantons into the 

162: formalism. Instantons can be considered 

163: either on the level of a resurgent expansion,

164: augmented by an optimized (generalized) Borel-Pad\'{e} resummation,

165: or on the level of a generalized quantization condition,

166: which entails an all-order resummation of the 

167: instanton expansion. 

168: 

169: The third application is mainly tied to the semiconductor 

170: ``double quantum dot'' structures~\cite{LuEtAl1989,LoDV1998,HoEtAl2002,HuEtAl2005}, 

171: which are formed from two quantum dots coupled by quantum 

172: mechanical tunneling. Nowadays, these 

173: structures are generally accepted to belong to one of the most hopeful 

174: candidates for the realization of quantum bits (qubits), because

175: a single electron state in a double-well potential obviously can 

176: be localized in either of the two wells and, in that sense, represents a 

177: two-quantum-state system needed for quantum computing. 

178: Indeed, the theoretical analysis of the

179: structural and the dynamical properties of such (single-electron) 

180: double quantum dots can be traced back to double-well-like 

181: potentials. In this context, quantum dynamical calculations for a 

182: tunneling of a single particle between the two localized 

183: states nowadays attract special interest \cite{GrDiJuHa1991, Op1999} but 

184: require a detailed knowledge on the eigenvalues and eigenfunctions 

185: of the (double-well-potentials) potentials for a wide range

186: of coupling parameters. 

187: 

188: Another question may as well be asked: The effective 

189: instanton-related expansion parameter,

190: which reads $\Xi_1(g) = \sqrt{2/\pi}\, {\rm e}^{-1/6g}/g$

191: for the first two excited states of the Fokker--Planck

192: potential (as discussed below), is nonperturbatively

193: small for $g \to 0$, but numerically not very

194: small for some very moderate $g$. Specifically,

195: $\Xi_1(g)$ reaches its maximum $\Xi_1(1/6) = 1.76115\dots$ 

196: already at a rather small coupling parameter $g = 1/6$.

197: So, one may ask how the ``instanton expansion''

198: in powers of $\Xi_1(g)$ should be resummed, in addition to the

199: perturbative expansions about each instanton.

200: This latter step has never been accomplished, and we pursue its completion

201: via a direct resummation of generalized quantization conditions.

202: 

203: This paper is organized as follows.

204: In Sec.~\ref{frame}, basic definitions

205: related to the Fokker--Planck and the double-well

206: potential are recalled. Calculations are described

207: in Sec.~\ref{resumm}. Specifically, we consider

208: the resummation of the resurgent expansion

209: in Sec.~\ref{resurg}, the resummation of the 

210: quantization condition in Sec.~\ref{resumq}, 

211: large-coupling asymptotics in Sec.~\ref{largec},

212: and quantum dynamic simulations in Sec.~\ref{quadyn}.

213: Conclusions are drawn in Sec.~\ref{conclu}.

214: 

215: %

216: % Basic framework and numerical procedure

217: %

218: \section{Basic framework and numerical procedure}

219: \label{frame}

220: 

221: %

222: % Basic formulation

223: %

224: \subsection{Basic formulation}

225: \label{basic}

226: 

227: In this manuscript, we discuss the determination of the eigenvalues of the 

228: one--dimension Fokker--Planck (FP) Hamiltonian  

229: %

230: \begin{equation}

231: \label{hamfp}

232: H_{\rm FP} = -\frac{1}{2} \left( \frac{d}{dq} \right)^2 \, + \,

233: \frac{1}{2} q^2 \left( 1 - \sqrt{g}q \right)^2 \, + \, \sqrt{g} q \, - \, 

234: \frac{1}{2} \,,

235: \end{equation}

236: %

237: where $g$ is a positive coupling constant. 

238: For $g \, = \, 0$, Eq.~(\ref{hamfp}) represents

239: the Hamiltonian of the quantum harmonic oscillator whose eigenvalues are

240: given by the well known formula $E^{(K)} \, = \, K$, where 

241: $K = 0, 1, 2,\dots$ is the ``principal'' quantum number. For 

242: nonvanishing coupling, in contrast, no 

243: closed-form analytic expressions have been derived so far,

244: and approximations have to be used (for a classification

245: of the Fokker--Planck Hamiltonian in terms of a SUSY

246: algebra, see App.~\ref{appa}). The usefulness

247: of the notation $K$ instead of $N$ will become clear

248: in the following. If one considers the operator 

249: $V(g) = \sqrt{g} \, q - \sqrt{g} \, q^3 + g \, q^4/2$ in (\ref{hamfp}) 

250: as a perturbation and formally applies 

251: Rayleigh--Schr\"odinger perturbative expansion

252: to the $K$th harmonic oscillator state, then one finds the following

253: result for the first terms,

254: %

255: \begin{equation}

256: \label{epert}

257: E_{\rm FP, pert}^{(K)}(g) = K - 3K^2 g - 

258: \left(17 \, K^3 + \frac{5}{2} \, K \right) g^2   + 

259: \mathcal{O}(g^3) \,.

260: \end{equation}

261: %

262: All coefficients up to order $g^{80}$ 

263: are available for download~\cite{JeHomeHD}.

264: This perturbation expansion~\cite{HeSi1978}

265: fails to reproduce the spectrum of the Hamiltonian 

266: (\ref{hamfp}) even qualitatively. For instance, while the 

267: true ground--state energy $E^{(K = 0)}_{\rm FP}$ is manifestly nonvanishing

268: and positive, the perturbation series (\ref{epert}),

269: for $K = 0$, vanishes identically to all orders in the coupling $g$

270: and is thus formally converging to a zero energy eigenvalue.

271: A generalization of perturbation theory is required, therefore, 

272: in order to correctly describe the physical properties of the 

273: Fokker--Planck Hamiltonian, including its energy spectrum.

274: 

275: A complete description 

276: of the eigenvalues of the Hamiltonian (\ref{hamfp}) 

277: has been proposed recently~\cite{JeZJ2004plb,ZJJe2004i,ZJJe2004ii} by 

278: using a generalized perturbation series involving instanton 

279: contributions. Since the concept of instantons in quantum mechanics has 

280: been presented in a number of places 

281: \cite{JeZJ2004plb,ZJJe2004i,ZJJe2004ii,ZJ1984jmp}, we may here restrict 

282: ourselves to a rather short account of the basic formulas. In the 

283: semi--classical framework, the eigenvalues of the Fokker--Planck 

284: Hamiltonian can be found by solving the generalized Bohr--Sommerfeld 

285: quantization condition \cite{JeZJ2004plb,ZJ1984jmp}

286: %

287: %

288: \begin{eqnarray}

289: \label{qcond}

290: \lefteqn{\frac{1}{\Gamma(-B_{\rm FP}(E, g)) \,\,

291: \Gamma(1-B_{\rm FP}(E, g))}} \nonumber\\[2ex]

292: & & + \, \left(-\frac{2}{g}\right)^{2B_{\rm FP}(E, g)} 

293: \frac{{\rm e}^{-A_{\rm FP}(E, g)}}{2 \pi} \, = \, 0 \,.

294: \end{eqnarray}

295: %

296: %

297: In this expression, the functions $B_{\rm FP}(E, g)$ and 

298: $A_{\rm FP}(E, g)$ determine the perturbative expansion and the 

299: perturbative expansion about the instantons, correspondingly. 

300: 

301: The evaluation of these functions in terms of series in variables $E$ and $g$ 

302: has been described in detail elsewhere \cite{ZJJe2004i,ZJJe2004ii,ZJ1984jmp} 

303: for rather general classes of 

304: potentials. In the particular case of the Fokker--Planck potential, for 

305: example, the function $B_{\rm FP}(E, g)$ has the following expansion

306: [see Eq.~(14a) of Ref.~\onlinecite{JeZJ2004plb}]:

307: %

308: \begin{eqnarray}

309: \label{B_expansion}

310: B_{\rm FP}(E, g) \, =  E + 3E^2g + 

311: \left(35E^3 + \frac{5}{2}E\right)g^2 + \mathcal{O}(g^3) \, . 

312: \end{eqnarray}

313: %

314: The function

315: $B_{\rm FP}(E, g)$ alone

316: defines the perturbation expansion (\ref{epert}) which 

317: can be easily found by inverting the equation $B_{\rm FP}(E, g) \, = \, 

318: K$. The instanton contributions to the eigenvalues of the Fokker--Planck 

319: Hamiltonian are described by the function

320: [see Eq.~(14b) of Ref.~\onlinecite{JeZJ2004plb}]:

321: %

322: \begin{eqnarray}

323: \label{A_expansion}

324: \lefteqn{A_{\rm FP}(E, g) \, = \, \frac{1}{3g} \, + \, 

325: \left( 17E^2 +\frac{5}{6}\right) \, g } \nonumber\\[2ex]

326: & & + \, \left(227E^3 + \frac{55}{2}E\right)g^2 \, + \, 

327: \mathcal{O}(g^3) \, . 

328: \end{eqnarray}

329: %

330: Extensive numerical checks of the generalized quantization condition 

331: (\ref{qcond}) and the expansions (\ref{B_expansion})--(\ref{A_expansion}) have 

332: recently been performed for the ground state of the Fokker--Planck potential 

333: in the weak coupling regime \cite{JeZJ2004plb}. However, to the best of our 

334: knowledge, a numerical verification of these formulae 

335: (i) for excited states and (ii) for the large values of $g$

336: is still missing. The numerical checks will be presented in Sec.~\ref{resumq}.

337: 

338: While, of course, the present work is mainly devoted to the investigation of 

339: the energies and the corresponding 

340: wave functions of the Fokker--Planck potential, we will 

341: also briefly recall the properties of the well--known double-well potential

342: which is characterized by the Hamiltonian

343: %

344: %

345: \begin{equation}

346: \label{hamdw}

347: H_{\rm dw} = -\frac{1}{2} \left( \frac{d}{dq} \right)^2 \, + \,

348: \frac{1}{2} q^2 \left( 1 - \sqrt{g}q \right)^2 \, .

349: \end{equation}

350: %

351: %

352: Moreover, for the analysis of the energy spectra of the Hamiltonians 

353: (\ref{hamfp}) and (\ref{hamdw}) it is very convenient to introduce 

354: the interpolating potential:

355: %

356: %

357: \begin{equation}

358: \label{hami}

359: H_{\rm I} = -\frac{1}{2} \left( \frac{d}{dq} \right)^2 \, + \,

360: \frac{1}{2} q^2 \left( 1 - \sqrt{g}q \right)^2 \, + 

361: \eta \left( \sqrt{g} q - \frac{1}{2} \right)\, ,

362: \end{equation}

363: %

364: which corresponds to the double-well potential if $\eta = 0$,

365: whereas $\eta = 1$ gives the Fokker--Planck case.

366: 

367: %

368: % Numerical calculation of eigenenergies

369: %

370: \subsection{Numerical calculation of eigenenergies}

371: \label{numer}

372: 

373: In order to numerically calculate energy eigenvalues of the 

374: Fokker--Planck and of the double--well potential,

375: it is sufficient to consider matrix elements of these

376: potentials in the basis of harmonic oscillator eigenfunctions,

377: and to perform matrix diagonalization in a large basis

378: spanned by harmonic oscillator eigenfunctions

379: (typically, a number of $\gtrsim 5000$ basis functions is sufficient

380: for all calculations reported in the current article).

381: One then observes the apparent convergence of the 

382: eigenenergies as the size of the basis is increased.

383: This procedure is numerically stable

384: provided one uses quadrupole precision (128-bit, 32 decimal figure)

385: arithmetic. 

386: 

387: %

388: %

389: \begin{figure}[t]

390: %\vspace*{-1cm}

391: \begin{center}

392: \includegraphics[height=0.9\linewidth,angle=270]{Fig1.eps}

393: \end{center}

394: %\vspace*{-5cm}

395: \caption{\label{fig1} 

396: (Color online) Eigenvalues $E_{\rm FP}^{(M = 0)}$ (left panel) and 

397: $E_{\rm FP}^{(M = 1,2)}$ (right panel) of the Fokker--Planck Hamiltonian 

398: as a function of the coupling parameter $g$. Results have been computed by the

399: diagonalization of the Fokker--Planck Hamiltonian 

400: (\ref{hamfp}) in the basis of the harmonic 

401: oscillator wavefunctions.}

402: \end{figure}

403: %

404: %

405: 

406: Fokker--Planck energies of the lowest three eigenstates

407: [of the potential (\ref{hamfp})]

408: found by matrix diagonalization are displayed in Fig.~\ref{fig1} 

409: as a function of the coupling $g$ (we use the notation

410: $M = 0,1,2$ in order to denote these three energy levels). As seen 

411: from this Figure, the states $M = 1,2$ are degenerate

412: in the limit $g \to 0$. A similar energy level 

413: splitting is well known for the symmetric double--well potential 

414: \cite{ZJJe2004i,ZJJe2004ii,JeZJ2001} and may be explained in terms of 

415: nonperturbative instanton contributions. In contrast to the 

416: double--well case, the Fokker--Planck potential contains a linear 

417: symmetry--breaking term [cf. Eq.~(\ref{hamfp})], and this term

418: might be expected to lift any degeneracy. 

419: However, excited states can still develop a degeneracy

420: for $g \to 0$ in view of the (only perturbatively broken) parity 

421: $\varepsilon \, = \, \pm 1$ of the quantum eigenstates.

422: The ground state of the Fokker--Planck potential, however,

423: is located in one of the wells and does not develop

424: any degeneracy due to parity (see also Fig.~\ref{fig9} below).

425: 

426: It is interesting to investigate the adiabatic following 

427: of eigenvalues for the interpolating potential (\ref{hami})

428: as a function of the parameter $\eta$ for fixed $g$.

429: This calculation (see Fig.~\ref{fig2}) reveals that the 

430: identification of the double-well 

431: energy eigenvalues~\cite{JeZJ2001,ZJJe2004i,ZJJe2004ii} with 

432: quantum numbers $(N,\varepsilon)$ for the double-well 

433: with the quantum number $M$

434: for the Fokker--Planck potential should proceed as follows:

435: %

436: \begin{subequations}

437: \label{ident}

438: \begin{align}

439: (N=0,+) \Leftrightarrow & \; M=0\,, \\

440: (N=0,-) \Leftrightarrow & \; M=1\,, \\

441: (N=1,+) \Leftrightarrow & \; M=2\,, \\

442: (N=1,-) \Leftrightarrow & \; M=3\,.

443: \end{align}

444: \end{subequations}

445: %

446: The general relation is $M = 2 N + (1 - \varepsilon)/2$.

447: However, the asymptotic behavior of the eigenenergies 

448: for $g \to 0$ is different in the two cases:

449: %

450: \begin{subequations}

451: \begin{align}

452: \label{NtoN}

453: E_{\rm dw}^{(N,\varepsilon)}(g) \to & \; N + \frac12\,, 

454: \qquad g \to 0 \,, \\

455: \label{MtoK}

456: E_{\rm FP}^{(M)}(g) \to & \; 

457: [\mkern - 2.5 mu  [ (M+1)/2 ] \mkern - 2.5 mu ] \,, 

458: \qquad g \to 0 \,,

459: \end{align}

460: \end{subequations}

461: %

462: where $[\mkern - 2.5 mu  [x] \mkern - 2.5 mu ]$ is the integral part of

463: $x$, i.e., the largest integer $m$ satisfying $m \le x$.

464: Equation (\ref{MtoK}) implies that the perturbative contribution

465: to the Fokker--Planck energy level with quantum number $M$ is 

466: given by Eq.~(\ref{epert}) with 

467: $K = [\mkern - 2.5 mu  [ (M+1)/2 ] \mkern - 2.5 mu ]$.

468: 

469: Apart from the degeneracy introduced by the instanton contributions, the 

470: eigenvalues $E^{(M = 1,2)}_{\rm FP}(g)$ also have a 

471: qualitatively different dependence on $g$ when compared to the 

472: ground--state energy $E^{(M = 0)}_{\rm FP}(g)$. As seen from Fig.~\ref{fig1}, while the 

473: energy $E^{(M = 0)}_{\rm FP}(g)$ increases monotonically as a function of the 

474: coupling constant $g$, the energies $E^{(M = 1,2)}_{\rm FP}(g)$ 

475: have minima at

476: $g_0 \, \approx \, 0.07$ and $g_0 \, \approx \, 0.025$, respectively.

477: 

478: %

479: %

480: \begin{figure}[ht]

481: %\vspace*{-1.0cm}

482: \begin{center}

483: \includegraphics[width=0.7\linewidth,angle=270]{Fig2.eps}

484: \end{center}

485: %\vspace*{1.0cm}

486: \caption{\label{fig2} (Color online)

487: Adiabatic following of the lowest four eigenvalues 

488: of the interpolating potential (\ref{hami}) from $\eta = 0$

489: (double-well) to $\eta = 1$ (Fokker--Planck potential).

490: The value of the coupling parameter is held constant at

491: $g = 0.007$.}

492: \end{figure}

493: %

494: %

495: 

496: %

497: % Resummations

498: %

499: \section{Resummations, Energy Eigenvalues and Quantum Dynamics}

500: \label{resumm}

501: 

502: %

503: % Resummation of the resurgent expansion

504: %

505: \subsection{Resummation of the resurgent expansion}

506: \label{resurg}

507: 

508: The generalized Bohr--Sommerfeld quantization condition (\ref{qcond}) together 

509: with the expansions (\ref{B_expansion})---(\ref{A_expansion}) of the 

510: $A_{\rm FP}$ and $B_{\rm FP}$ functions uniquely determines the eigenvalues 

511: of the Fokker--Planck Hamiltonian. For the ground state,

512: the energy eigenvalue can be found by systematic expansion

513: of (\ref{qcond}) in powers of the two small parameters $\exp(-1/3g)$ and $g$,

514: whereas for excited states, the parameters are $\exp(-1/6g)$ and $g$.

515: Thus, for the Fokker--Planck potential, the particular form of the

516: expansion differs for the ground vs.~excited states.

517: 

518: Explicitly, the ground-state energy ($M=0$) is given by the resurgent

519: expansion \cite{JeZJ2004plb}: 

520: %

521: %

522: \begin{equation}

523: \label{eground}

524: E^{(0)}_{\rm FP}(g) =

525: \sum^{\infty}_{n=1} \left( {{\rm e}^{-1/3g} \over 2 \pi} \right)^{n} \,

526: \sum^{n-1}_{k=0} \left\{ \ln\left(-\frac{2}{g}\right) \right\}^k \,

527: \sum^{\infty}_{l=0} f^{(0)}_{nkl} \, g^{l} \, ,

528: \end{equation}

529: %

530: %

531: where the index $n$ denotes the order of the ``instanton contribution'':

532: $n = 1$ is a one-instanton, $n=2$ is a two-instanton, etc. 

533: (as noted below, the one-instanton configuration involves a 

534: back-tunneling of the particle to the lower well for the 

535: ground state and thus has twice the action of the characteristic 

536: one-instanton effect for excited states). Another subtle point which 

537: should be recalled here is that the leading one-instanton term 

538: involves a summation over all possible $n$--instanton configurations but

539: neglects instanton interactions \cite{ZJJe2004i}. As seen from 

540: Eq.~(\ref{eground}), the evaluation

541: of the ground state energy within such a (first--order) approximation, 

542: also referred as a ``dilute instanton gas'' approximation, requires the 

543: knowledge of the $f^{(0)}_{10l}$ coefficients. Since these coefficients are 

544: available for download~\cite{JeHomeHD}, we only recall the six leading 

545: ones~\cite{JeZJ2004plb}:

546: %

547: %

548: \begin{eqnarray}

549: f^{(0)}_{100} &=& 1 \,, \quad 

550: f^{(0)}_{101} = -\frac{5}{6} \,, \quad 

551: f^{(0)}_{102} = -\frac{155}{72} \,, \nonumber \\[1ex]

552: f^{(0)}_{103} &=& -\frac{17315}{1296} \,, \quad

553: f^{(0)}_{104} = -\frac{3924815}{31104} \,, \\[1ex]

554: f^{(0)}_{105} &=& -\frac{294332125}{186624} \,, \quad 

555: f^{(0)}_{106} = -\frac{163968231175}{6718464} \,. \nonumber

556: \end{eqnarray}

557: %

558: %

559: By inserting these coefficients in Eq.~(\ref{eground}), we are able

560: to perform now the numerical check of the validity of the one-instanton 

561: expansion for the ground state at small coupling. In Table~\ref{table1},

562: for example, the energy $E^{(0)}_{\rm FP}(g)$ is displayed for coupling 

563: parameters in the range 0.005 $\le g \le$ 0.03 and is compared to the 

564: ``true'' eigenvalues as obtained from the diagonalization of the 

565: Fokker--Planck potential in the basis of harmonic oscillator wavefunctions.

566: As seen from Table~\ref{table1}, the ground state energy is 

567: dominated by the one-instanton effect for relatively small 

568: values of the coupling parameter $g \le$ 0.01. 

569: For stronger coupling, however,

570: large discrepancies between the ``true'' energies and the results of

571: resummation of Eq.~(\ref{eground}) at $n$ = 1 are found indicating 

572: the importance of the higher--instanton terms which take into account

573: the instanton interactions. The evaluation of the higher--order corrections 

574: ($n \ge$ 2) to the ground--state energy $E^{(0)}_{\rm FP}(g)$ is, however, 

575: a very difficult task since it requires a double resummation of the resurgent 

576: expansion, in powers of both $g$ and $\exp(-1/3g)$. In the 

577: present work such a double summation based on sequentially

578: adding higher-order instanton terms will not be performed.

579: Still, for the sake of completeness, we here indicate the leading 

580: two--instanton \cite{JeZJ2004plb} and three--instanton coefficients

581: for the Fokker--Planck ground state:  

582: %

583: %

584: \begin{eqnarray}

585: f^{(0)}_{210} &=& 2 \,, \quad

586: f^{(0)}_{200} = 2 \gamma \,, \nonumber \\[1ex]

587: f^{(0)}_{211} &=& -\frac{10}{3} \,, \quad

588: f^{(0)}_{201} = -\frac{10}{3} \gamma - 3 \,, \\[1ex]

589: f^{(0)}_{310} &=& 8 \, \gamma \,, \quad

590: f^{(0)}_{300} = 6 \gamma^2 + \frac{\pi^2}{6}\,, \nonumber \\[1ex]

591: f^{(0)}_{311} &=& -\frac{80}{3} \gamma - 6 \,, \quad

592: f^{(0)}_{301} = -15 \gamma^2 - 12 \gamma -17 + \frac{5}{12} \pi^2 \,, \nonumber

593: \end{eqnarray}

594: %

595: %         

596: where $\gamma = 0.577216\dots$ is Euler's constant. 

597: 

598: %

599: % Table 1

600: %

601: \begin{table}[t]

602: \caption{\label{table1} Ground--state energy of the Fokker--Plank 

603: Hamiltonian. Results have been computed by the diagonalization of the

604: Hamiltonian in the basis of the harmonic oscillator wavefunctions ("true" 

605: energy) and by resummation of Eq.~(\ref{eground}) within the one-instanton

606: approximation ($n$ = 1). The numerical uncertainty of the entries 

607: in the right column is estimated on the basis of the apparent 

608: convergence of results under an appropriate

609: increase of the number of $f^{(0)}_{10l}$ parameters, which corresponds

610: to the number of terms in the perturbative expansion about the leading 

611: instanton. Numerical discrepancies between the left and right column

612: are due to higher-order instanton contributions, as described in the text.

613: We underline those decimal figures in the one-instanton results 

614: which are equal to the corresponding ones in the complete 

615: numerical solution.}

616: \vspace*{0.3cm}

617: \begin{scriptsize}

618: \begin{center}

619: \begin{tabular}{l@{\hspace*{0.5cm}}l@{\hspace*{0.5cm}}l} 

620: \hline

621: \hline

622: \rule[-3mm]{0mm}{8mm}

623: $g$ & $E^{(M = 0)}_{\rm FP}$ (diagonalization) 

624:     & $n = 1$ term of Eq.~(\ref{eground}) \\

625: \hline

626: \rule[-3mm]{0mm}{8mm}

627: 0.005 &             1.766 107 332 563      $\times$ 10$^{-30}$  &

628:          \underline{1.766 107 332 563}     $\times$ 10$^{-30}$ \\

629: \rule[-3mm]{0mm}{8mm}

630: 0.010 &             5.267 473 259 637      $\times$ 10$^{-16}$ & 

631:          \underline{5.267 473 259 637}     $\times$ 10$^{-16}$ \\

632: \rule[-3mm]{0mm}{8mm}

633: 0.015 &             3.508 587 565 372      $\times$ 10$^{-11}$  &

634:          \underline{3.508 587 56}4 030     $\times$ 10$^{-11}$ \\

635: \rule[-3mm]{0mm}{8mm}

636: 0.020 &             9.033 155 571 641      $\times$ 10$^{-09}$  &

637:          \underline{9.033 15}4 730 920     $\times$ 10$^{-09}$ \\

638: \rule[-3mm]{0mm}{8mm}

639: 0.025 &             2.519 767 018 258      $\times$ 10$^{-07}$  &

640:          \underline{2.519 76}0 770 755(1)  $\times$ 10$^{-07}$ \\

641: \rule[-3mm]{0mm}{8mm}

642: 0.030 &             2.313 302 179 961      $\times$ 10$^{-06}$  &

643:          \underline{2.313} 251 574 075(2)  $\times$ 10$^{-06}$ \\

644: \hline

645: \hline

646: \end{tabular}

647: \end{center}

648: \end{scriptsize}

649: \end{table}

650: 

651: As seen from Eq.~(\ref{eground}), no splitting into levels with 

652: positive and negative parity arises for the ground state of the 

653: Fokker--Planck potential due to the linear symmetry--breaking term 

654: in Eq.~(\ref{hamfp}). This term modifies the potential in such a way that the

655: leading, one-instanton ($n$ = 1) shift of the ground state energy

656: results from a back--tunneling (instanton--antiinstanton configuration)    

657: of the particle to the lower well \cite{JeZJ2004plb}. For excited states, 

658: in contrast, the one-instanton configuration is a trajectory which starts in

659: one well and ends in the other, restoring the broken symmetry.

660: Therefore, any excited state ($M >$ 0) of the Fokker--Planck Hamiltonian can 

661: be characterized by its principal quantum number 

662: %

663: %

664: \begin{equation}

665:   \label{FP_M_to_K}

666:   K = [\mkern - 2.5 mu  [ (M+1)/2 ] \mkern - 2.5 mu ]

667: \end{equation} 

668: %

669: %

670: and the parity 

671: %

672: %

673: \begin{equation}

674:    \label{FP_M_to_epsilon}

675:    \varepsilon = 2 \left( 2 K - M - \frac12 \right) \, .

676: \end{equation}

677: %

678: %

679: In fact, this classification is very

680: similar to the double-well potential (\ref{hamdw}) except,

681: of course, the particular case of the ground state. It follows

682: naturally that the resurgent expansion for

683: the excited states of the Fokker--Planck potential is very close to the

684: analogous expansion for the double--well potential and reads 

685: \cite{JeZJ2004plb}:

686: %

687: %

688: \begin{align}

689: \label{efpgen}

690: & E^{(M > 0)}_{\rm FP}(g) = 

691: E^{(K, \varepsilon)}_{\rm FP}(g) =

692: \sum_{l = 0}^\infty E_{K, l} \, g^l 

693: \nonumber\\[2ex]

694: & + \sum^{\infty}_{n=1} \left[-\varepsilon \, \Xi_K(g) \right]^n \,

695: \sum^{n-1}_{k=0} \left\{ \ln\left(-\frac{2}{g}\right) \right\}^k \,

696: \sum^{\infty}_{l=0} f^{(K)}_{nkl} \, g^{l} \, ,

697: \end{align}

698: %

699: %

700: where $E_{K, l}$ are perturbative coefficients and

701: $\Xi_K(g)$ is given by

702: %

703: %

704: \begin{equation}

705: \Xi_K(g) = \frac{2^{K-1/2}}{g^K\, \sqrt{ \pi \, K! \, (K-1)!}}\,

706: \rm{e}^{ - 1/6 g} \, .

707: \end{equation}

708: %

709: %

710: The power of $\Xi$ can again 

711: be associated with the order of the instanton

712: ($K = 1$: one-instanton, $K = 2$ means two-instanton, etc.).

713: One should note that two intricacies are associated to 

714: the precise meaning of the quantities that enter Eq.~(\ref{efpgen}):

715: %

716: \begin{itemize}

717: \item In analogy to the double-well potential, the

718: imaginary part which is generated by the resummation of the 

719: perturbation series about the leading instanton

720: (the ``discontinuity'' of the distributional Borel sum

721: in the terminology of Ref.~\onlinecite{CaGrMa1986})

722: is compensated by an explicit imaginary part that stems from the

723: two-instanton effect [from the factor $\ln(-2/g)$].

724: Related questions have been discussed at length in 

725: Refs.~\onlinecite{ZJJe2004i,ZJJe2004ii}.

726: %

727: \item In contrast to the ground--state energy (\ref{eground}), the leading

728: contribution to the 

729: energies $E^{(\varepsilon,K > 0)}_{\rm FP}(g)$ for small coupling

730: arises from the perturbation expansion (\ref{epert}) which is manifestly

731: nonvanishing to all orders in $g$. However, since this perturbation expansion

732: is independent of the parity $\varepsilon$, the energy splitting of 

733: the levels with the same principal quantum number $K$ is again dominated by the

734: one-instanton contribution ($n = 1$). 

735: \end{itemize}

736: %

737: Similar to the ground state 

738: (\ref{eground}), we may compute such a contribution and, hence, a

739: splitting of an arbitrary excited state $K$ by making use of 

740: the $f^{(K)}_{10l}$ coefficients, which for $K > 0$ read

741: ($l = 0,1,2,3$): 

742: %

743: %

744: \begin{subequations}

745: \begin{align}

746: f^{(K)}_{100} =& 1 \,, \qquad

747: f^{(K)}_{101} =  -\frac{17}{2} K^2 - 6 K - \frac{5}{12} \,,

748: \\[1ex]

749: f^{(K)}_{102} =& \frac{289}{8} K^4

750: - \frac{23}{2} K^3 - \frac{1139}{24} K^2 - \frac{45}{4} K 

751: - \frac{695}{288} \,,

752: \\[1ex]

753: f^{(K)}_{103} =& -\frac{4913}{48} K^6 

754: + \frac{629}{2} K^5 + \frac{1637}{32} K^4 \nonumber\\[2ex]

755: & - \frac{1885}{3} K^3 

756: - \frac{155825}{576} K^2 - \frac{3835}{24} K - \frac{68885}{10368} \,.

757: \end{align}

758: \end{subequations}

759: %

760: %

761: Results for $K \leq 28$ are available for download~\cite{JeHomeHD}.

762: In Table~\ref{table2}, for example, the splitting 

763: $E^{(\varepsilon = -1,K = 1)}_{\rm FP}(g) - 

764: E^{(\varepsilon = +1,K = 1)}_{\rm FP}(g)$ of the first 

765: two excited states $M = 1,2$ due to the one-instanton effect 

766: ($n$ = 1) is displayed as 

767: a function of the coupling parameter $g$. Again, a comparison of the results 

768: obtained by the resummation of Eq.~(\ref{efpgen}) and by the diagonalization of 

769: the Fokker--Planck Hamiltonian indicates the importance of the 

770: higher--instanton

771: effects ($n >$ 1) and, hence, the necessity of a double resummation of the 

772: resurgent expansion, in powers of both $g$ and $\exp(-1/6g)$. Instead of

773: performing such a double summation explicitly, it is more

774: convenient to enter directly into the 

775: quantization condition (\ref{qcond}), with resummed 

776: quantities as defined by the 

777: $A_{\rm FP}(E, g)$ and $B_{\rm FP}(E, g)$ functions. We discuss this

778: alternative approach in the next Section.

779: 

780: %

781: % Table 2

782: %

783: \begin{table}[t]

784: \caption{\label{table2} Energy difference between the excited states

785: $E^{(M = 1, 2)}_{\rm FP}(g)$ of the Fokker--Planck 

786: Hamiltonian. Results have been computed by

787: diagonalizing the Hamiltonian in the basis of the harmonic oscillator 

788: wavefunctions (left column) and by resummation of Eq.~(\ref{efpgen}) within 

789: the one-instanton approximation (right column). The numerical uncertainty is

790: estimated on the basis of the apparent 

791: convergence of results under an appropriate

792: increase of the number of $f^{(K)}_{10l}$ parameters.

793: As in Table~\ref{table1},

794: we underline those decimal figures in the one-instanton results

795: which are equal to the corresponding ones in the complete 

796: numerical solution.}

797: \vspace*{0.3cm}

798: \begin{scriptsize}

799: \begin{center}

800: \begin{tabular}{l@{\hspace*{0.5cm}}l@{\hspace*{0.5cm}}l} 

801: \hline

802: \hline

803: \rule[-3mm]{0mm}{8mm}

804: $g$  & $E^{(M = 2)}_{\rm FP} - E^{(M = 1)}_{\rm FP}$ (diag.) 

805:      & $n = 1$ term of Eq.~(\ref{efpgen}) \\

806: \hline

807: \rule[-3mm]{0mm}{8mm}

808: 0.005 &             9.848 553 978 903       $\times$ 10$^{-13}$  &

809:          \underline{9.848 553 978 903}      $\times$ 10$^{-13}$ \\

810: \rule[-3mm]{0mm}{8mm}

811: 0.010 &             7.801 059 663 554       $\times$ 10$^{-06}$ & 

812:          \underline{7.801 059 65}9 99(1)    $\times$ 10$^{-06}$ \\

813: \rule[-3mm]{0mm}{8mm}

814: 0.015 &             1.213 924 539 483       $\times$ 10$^{-03}$  &

815:          \underline{1.213 9}1 452(2)        $\times$ 10$^{-03}$ \\

816: \rule[-3mm]{0mm}{8mm}

817: 0.020 &             1.289 613 568 640       $\times$ 10$^{-02}$  &

818:          \underline{1.28}8 765(1)           $\times$ 10$^{-02}$ \\

819: \rule[-3mm]{0mm}{8mm}

820: 0.025 &             4.633 794 364 814       $\times$ 10$^{-02}$  &

821:          \underline{4.6}11 6(1)             $\times$ 10$^{-02}$ \\

822: \rule[-3mm]{0mm}{8mm}

823: 0.030 &             9.699 341 140 782       $\times$ 10$^{-02}$  &

824:          \underline{9.6}10 2(6)             $\times$ 10$^{-02}$ \\

825: \hline

826: \hline

827: \end{tabular}

828: \end{center}

829: \end{scriptsize}

830: \end{table}

831: %

832: %

833: %

834: 

835: 

836: %

837: % Resummation of the quantization condition

838: %

839: \subsection{Resummation of the quantization condition}

840: \label{resumq}

841: 

842: The resurgent expansions (\ref{eground}) and (\ref{efpgen}) for the energies of

843: the ground and excited states of the Fokker--Planck Hamiltonian

844: follow as a direct consequence of the quantization condition (\ref{qcond}).

845: As seen from our calculations summarized in Tables \ref{table1} and 

846: \ref{table2}, these expansions are very useful for small coupling,

847: but not of particular usefulness even for rather

848: moderate values of $g$, because of the necessity of their double resummation.

849: Here, we would like to investigate whether it is possible to 

850: resum the divergent series that gives rise to $A_{\rm FP}(E, g)$ and 

851: $B_{\rm FP}(E, g)$ directly and look for solutions of the 

852: quantization condition (\ref{qcond}) without any

853: intermediate recourse to the resurgent 

854: expansion. In fact, this approach currently

855: appears to be the only feasible way to evaluate the 

856: multi--instanton expansion (in powers of $n$),

857: because the quantization condition incorporates all instanton orders. 

858: 

859: In order to introduce such a ``direct summation'' approach, we recall

860: that the solution of the generalized Bohr--Sommerfeld quantization condition 

861: (\ref{qcond}) for a particular coupling parameter $g$ must provide the 

862: energy spectrum of the Fokker--Planck Hamiltonian. In other words, if 

863: one defines the left--hand side of the quantization condition 

864: (\ref{qcond}) as a function of two variables $E$ and $g$:

865: %

866: \begin{align}

867: \label{Q_function}

868: Q(E, g) =& \, 

869: \frac{1}{\Gamma(-B_{\rm FP}(E, g)) \, \Gamma(1-B_{\rm FP}(E, g))} 

870: \nonumber\\[1ex]

871: & + \left(-\frac{2}{g}\right)^{2B_{\rm FP}(E, g)} 

872: \frac{{\rm e}^{-A_{\rm FP}(E, g)}}{2 \pi} \,,

873: \end{align}

874: %

875: then the zeros of this function at fixed $g$ determine the energy spectrum of 

876: (\ref{hamfp}):

877: %

878: \begin{equation}

879:    \label{Q_zeros}

880:    Q(E^{(M)}_{\rm FP}(g), g) \, = \, 0, \qquad M = 0, 1, 2, .... \, .

881: \end{equation}

882: %

883: A numerical analysis of the function $Q(E, g)$ can

884: be used, therefore, in order to examine

885: the validity and applicability of the generalized

886: quantization condition given by

887: Eqs.~(\ref{qcond})---(\ref{A_expansion}) 

888: for the case of strong coupling.

889: 

890: \begin{figure}[t]

891: \begin{center}

892: \vspace*{-0.5cm}

893: \includegraphics[width=0.5\linewidth,angle=270]{Fig3.eps}

894: \end{center}

895: \vspace*{-0.7cm}

896: \caption{\label{fig3} (Color online) 

897: Energy dependence of (the real part of) the 

898: function $Q(E, g)$. Calculations have been performed for 

899: fixed $g = 0.03$.}

900: \end{figure}

901: 

902: As seen from Eq.~(\ref{Q_function}), any analysis of the function 

903: $Q(E, g)$ can be traced back to the evaluation of the functions 

904: $A_{\rm FP}(E, g)$ and $B_{\rm FP}(E, g)$ which constitute series in two

905: variables, namely

906: $E$ and $g$ [cf. Eqs.~(\ref{B_expansion})---(\ref{A_expansion})].

907: In order to compute these series, it is convenient to re--write the 

908: functions $A_{\rm FP}(E, g) \equiv A_{\rm FP}(E, g x) |_{x=1}$ and 

909: $B_{\rm FP}(E, g) \equiv B_{\rm FP}(E, g x) |_{x=1}$ 

910: as (formal) power series in terms of a variable $x$, taken at $x$ = 1

911: [cf. \S 8.5 of Ref.~\onlinecite{Ha1949}]:

912: %

913: \begin{eqnarray}

914: \label{a_x_series}

915: B_{\rm FP}(E, g) \, = \, \sum\limits_{n = 0}^{N_{max}} 

916: \left. b^{(n)}_{\rm FP}(E, g) \, x^n \right|_{x=1}\, ,		

917: \end{eqnarray}

918: %

919: %

920: \begin{eqnarray}

921: \label{b_x_series}

922: A_{\rm FP}(E, g) - \frac{1}{3 g} \, = \, \sum\limits_{n = 0}^{N_{max}} 

923: \left. a^{(n)}_{\rm FP}(E, g) \, x^n \right|_{x=1}\, ,

924: \end{eqnarray}

925: %

926: %

927: where the coefficients 

928: $a^{(n)}_{\rm FP}(E, g)$ and $b^{(n)}_{\rm FP}(E, g)$ 

929: are uniquely determined by Eqs.~(\ref{B_expansion}) and (\ref{A_expansion}): 

930: $b^{(0)}_{\rm FP}(E, g) \, = \, E$, 

931: $b^{(1)}_{\rm FP}(E, g) \, = \, 3E^2g$,

932: etc. In the 

933: computations, the power series (\ref{a_x_series}) and (\ref{b_x_series}) 

934: allow one to use a unified computer algebra routine for the Borel--like 

935: summations, which simply takes as input the variables $a^{(n)}_{\rm FP}(E, g)$

936: and $b^{(n)}_{\rm FP}(E, g)$, as a function of $g$ and returns the

937: value of the resummed series at $x=1$. Indeed, this routine can be universally 

938: used for different values of $g$ and is therefore convenient for 

939: further numerical computations which are discussed below.   

940: 

941: Making use of Eqs.~(\ref{a_x_series}) and (\ref{b_x_series}), we 

942: may now perform a simultaneous summation of the perturbation series as 

943: well as of the perturbation series about each of the instantons and find the 

944: functions $B_{\rm FP}(E, g)$ and $A_{\rm FP}(E, g)$, correspondingly,

945: where we use the same notation for a function and its Borel sum.

946: There is a small subtlety because 

947: for positive $g$, the power series (\ref{a_x_series}) and 

948: (\ref{b_x_series}) are nonalternating and divergent and, hence, special 

949: resummation techniques are required to calculate the 

950: Borel sums. In our present 

951: calculations, for example, we apply a generalized Borel--Pad\'e 

952: method \cite{Je2000prd,JeSo2001}. The discussion of this method is

953: beyond the scope of the present work, and the reader is referred to 

954: Refs.~\onlinecite{FrGrSi1985,Je2000prd,JeSo2001} for a more detailed discussion.

955: Because of their nonalternating property,

956: the perturbation series defining the functions

957: $A_{\rm FP}(E, g)$ and $B_{\rm FP}(E, g)$ are Borel summable

958: only in the distributional sense~\cite{CaGrMa1986}.

959: The evaluation of the Borel--Laplace integral thus requires an

960: integration along a contour which is tilted with respect to the real

961: axis (for details see 

962: Refs.~\onlinecite{FrGrSi1985,Je2001pra,JeSo2001}

963: and the contours $C_{+1}$ and $C_{-1}$ in Ref.~\onlinecite{Je2000prd}).

964: The resummation of the divergent series 

965: (\ref{a_x_series}) and (\ref{b_x_series}) may be carried out along each of 

966: these contours, but it is important to characterize the perturbative and

967: instanton contributions in the same way, i.e. to deform the contours

968: for $B_{\rm FP}(E, g)$ and $A_{\rm FP}(E, g)$

969: either above or below the real axis, consistently. In the 

970: terminology of Ref.~\onlinecite{CaGrMa1986}, one should 

971: exclusively use either ``upper sums'' or ``lower sums,'' but

972: mixed prescriptions are forbidden. From a historical perspective, it is 

973: interesting to remark that the possibility of deforming the  

974: Borel integration contour had already been anticipated in

975: a remark near the end of Chap.~8 of the classic Ref.~\onlinecite{Ha1949}.

976: 

977: %

978: %

979: \begin{table}[t]

980: \caption{\label{table3} Ground--state energy of the Fokker--Plank 

981: Hamiltonian. Results have been computed by the diagonalization of the

982: Hamiltonian in the basis of the harmonic oscillator wavefunctions 

983: (left column) and by solving Eq.~(\ref{Q_zeros}), 

984: as indicated in the right column.}

985: \vspace*{0.3cm}

986: \begin{scriptsize}

987: \begin{center}

988: \begin{tabular}{l@{\hspace*{0.5cm}}l@{\hspace*{0.5cm}}l} 

989: \hline

990: \hline

991: \rule[-3mm]{0mm}{8mm}

992: $g$   & $E^{(0)}_{\rm FP}$ (diagonalization)& Zero of ${\rm Re}[Q(E, g)]$ \\

993: \hline

994: \rule[-3mm]{0mm}{8mm}

995: 0.010 &             5.267 473 259 637     $\times$ 10$^{-16}$ & 

996:          \underline{5.267 473 259 637}    $\times$ 10$^{-16}$ \\

997: \rule[-3mm]{0mm}{8mm}

998: 0.030 &             2.313 302 179 961     $\times$ 10$^{-06}$  &

999:          \underline{2.313 302 17}(2)      $\times$ 10$^{-06}$ \\

1000: \rule[-3mm]{0mm}{8mm}

1001: 0.070 &             1.267 755 797 982     $\times$ 10$^{-03}$  &

1002:          \underline{1.267 7}4(6)          $\times$ 10$^{-03}$ \\

1003: \rule[-3mm]{0mm}{8mm}

1004: 0.100 &             5.199 138 696 222     $\times$ 10$^{-03}$ & 

1005:          \underline{5.199} 3(2)           $\times$ 10$^{-03}$ \\

1006: \rule[-3mm]{0mm}{8mm}

1007: 0.170 &             2.079 244 408 360     $\times$ 10$^{-02}$ & 

1008:          \underline{2.07}8(1)             $\times$ 10$^{-02}$ \\

1009: \rule[-3mm]{0mm}{8mm}

1010: 0.300 &             5.318 357 438 655     $\times$ 10$^{-02}$ & 

1011:          \underline{5.3}23(9)             $\times$ 10$^{-02}$ \\

1012: \hline

1013: \hline

1014: \end{tabular}

1015: \end{center}

1016: \end{scriptsize}

1017: \end{table}

1018: %

1019: %

1020: 

1021: We are now in the position 

1022: to analyze the properties of the function $Q(E, g)$ and,

1023: hence, to extract the energy spectrum of the Fokker--Planck Hamiltonian.

1024: As mentioned above, to perform such an analysis for any particular $g$ we have 

1025: to (i) resum the (divergent) series for the functions $A_{\rm FP}(E, g)$ and 

1026: $B_{\rm FP}(E, g)$ and (ii) insert the resulting generalized Borel sums 

1027: into Eq.~(\ref{Q_function}). We may then interpret 

1028: the $Q(E, g)$ as a function of 

1029: $E$ (at fixed $g$) and (iii) numerically determine the zeros of this 

1030: function which correspond to the energy values $E^{(M)}_{\rm FP}$ of the

1031: Fokker--Plank Hamiltonian, according to Eq.~(\ref{Q_zeros}).   

1032: In Fig.~\ref{fig3}, for instance, we display the energy dependence of the

1033: real part of the function $Q(E, g)$ taken at $g = 0.03$. 

1034: In the energy range $0 < E < 1$, this function has three zeros which

1035: obviously correspond to the ground $E^{(0)}_{\rm FP}$ and to the excited 

1036: $E^{(1, 2)}_{\rm FP}$ states. The ground-state energy 

1037: $E^{(0)}_{\rm FP}$ determined in such a way is presented in Table~\ref{table3} 

1038: and compared to reference values obtained by the diagonalization of the

1039: Hamiltonian matrix in the basis of harmonic oscillator wavefunctions. Moreover, 

1040: apart from the particular case of $g = 0.03$, we also display 

1041: the energy $E^{(0)}_{\rm FP}$ for other coupling parameters spanning 

1042: the range from $g = 0.01$ to $g = 0.3$ (see also Fig.~\ref{fig4}).

1043: This rather wide range of coupling parameters $g$ considered here allows us to 

1044: investigate the behavior of the generalized Bohr--Sommerfeld quantization

1045: condition in the transition from weak to strong coupling. 

1046: As seen from Table~\ref{table3}, the ground--state energy is well reproduced 

1047: at $g = 0.01$ (up to 14 decimal digits).

1048: Alternatively, a highly accurate value of the ground 

1049: state energy (at $g = 0.01$) can be obtained from the  

1050: the one-instanton contribution to the resurgent expansion (\ref{eground})

1051: for $g < 0.01$, as indicated in Table~\ref{table1}. 

1052: The accuracy of the one-instanton approximation is 

1053: rapidly decreasing for higher $g$. For instance, at the moderate value of 

1054: $g = 0.03$, the one-instanton term of the 

1055: resurgent expansion (\ref{eground}) reproduces the ground-state

1056: energy only to four decimal digits (see the last row of 

1057: Table~\ref{table1}), while a 

1058: total of eight digits can be obtained from solving 

1059: Eq.~(\ref{Q_zeros}), as indicated in the second row of Table~\ref{table3}. 

1060: For even stronger coupling, one observes a much larger 

1061: numerical uncertainty in the determination of the zeros of the 

1062: function $Q(E, g)$, because the convergence of the 

1063: generalized and optimized Borel--Pad\'{e} methods employed in 

1064: the resummation of the 

1065: $A_{\rm FP}(E, g)$ and $B_{\rm FP}(E, g)$ functions 

1066: is empirically observed to reach fundamental

1067: limits for larger values of the coupling, which cannot be overcome

1068: by the use of multiprecision arithmetic and might indicate a

1069: fundamental limitation for the convergence of the transforms

1070: and are not due to numerical cancellations. It might be interesting

1071: to explore these limits also from a mathematical point of view.

1072: Specifically, we have determined the numerical uncertainty

1073: of the $A_{\rm FP}(E, g)$ and $B_{\rm FP}(E, g)$ functions 

1074: on the basis of the apparent convergence of the Borel--Pad\'{e}

1075: approximants, integrated in the complex plane and accelerated 

1076: according to Ref.~\onlinecite{JeSo2001}, using an optimal truncation

1077: of the order of the transforms. We found that as the order of

1078: the Borel--Pad\'{e} transformation was increased, the apparent 

1079: convergence of the transforms stopped at around order 40 for 

1080: $g = 0.03$ and higher. Despite these difficulties, the generalized 

1081: Bohr--Sommerfeld quantization formula (\ref{Q_zeros})

1082: determines the ground--state energy of the 

1083: Fokker--Planck potential with an accuracy of about 0.01 \% 

1084: up to $g \leq 0.3$ (cf. Table~\ref{table3}).

1085: 

1086: \begin{figure}[t]

1087: %\vspace*{-2.0cm}

1088: \begin{center}

1089: \includegraphics[height=0.9\linewidth,angle=270]{Fig4.eps}

1090: \end{center}

1091: %\vspace*{-4.5cm}

1092: \caption{\label{fig4} (Color online)

1093: Energy dependence of (the real part of) the 

1094: function $Q(E, g)$. As in Fig.~\ref{fig3},

1095: the results of linear regression analysis of the 

1096: function $Q(E, g)$ are depicted by the solid line. Calculations have been 

1097: performed around the ``true'' energies of the ground state and for the 

1098: different values of the coupling parameter: $g$ = 0.03 (left panel), 

1099: $g = 0.1$ (middle panel) and $g = 0.3$ (right panel). The energy

1100: eigenvalues obtained using the displayed graphs determine the 

1101: corresponding entries in the right column of Table~\ref{table3}.}

1102: \end{figure}

1103: 

1104: Until now we have discussed the computation of the ground--state 

1105: energy $E^{(0)}_{\rm FP}$ of the Fokker--Planck Hamiltonian. Of course, the 

1106: function $Q(E, g)$ may also help to determine the energies of 

1107: excited states. In contrast to the ground state, however, the

1108: numerical analysis of the function $Q(E, g)$ for excited states is more 

1109: complicated due to bad convergence of the Borel sums

1110: for the $A_{\rm FP}(E, g)$ and $B_{\rm FP}(E, g)$ functions

1111: in the energy range relevant for the excited states.

1112: As seen from Fig.~\ref{fig5}, the convergence problems lead 

1113: to relatively large numerical 

1114: uncertainties for the numerical calculation of the function 

1115: $Q(E, g)$ already for a relatively mild coupling parameter 

1116: $g = 0.07$. We recall that this value of $g$ corresponds to the 

1117: minimum of the energy $E_{\rm FP}^{(M = 1)}(g)$ as a 

1118: function of $g$ and thus can be naturally identified

1119: as marking the transition from weak to strong coupling. 

1120: As a result of the numerical

1121: uncertainties, the energy $E_{\rm FP}^{(M = 1)}(g = 0.07)$ of the 

1122: first excited state may be reproduced only up to 2 decimal digits 

1123: (see Table~\ref{table4}). 

1124: For even larger values of parameter $g$, the maximal accuracy of 

1125: calculations, based on Eq.~(\ref{Q_zeros}),

1126: is only a single significant digit, even though the 

1127: double resummation of the instanton expansion, and of the

1128: perturbative expansion about each instanton, is implicitly

1129: contained in the cited Equation.

1130: 

1131: \begin{figure}[t]

1132: \begin{center}

1133: %\vspace*{-1.5cm}

1134: \includegraphics[height=0.9\linewidth,angle=270]{Fig5.eps}

1135: \end{center}

1136: %\vspace*{-4.5cm}

1137: \caption{\label{fig5} (Color online) 

1138: Energy dependence of (the real part of) the 

1139: function $Q(E, g)$. The results for the zeroes of 

1140: $Q(E, g)$ obtained by a quadratic regression 

1141: analysis are depicted by solid lines. Calculations have been performed 

1142: around the "true" energies of the first excited state with 

1143: $K$ = 1, $\varepsilon$ = +1 ($M$ = 1)

1144: and for three different 

1145: values of the coupling parameter: $g = 0.01$ (left panel), 

1146: $g = 0.03$ (middle panel) and $g = 0.05$ (right panel).

1147: The energy eigenvalues obtained using the displayed graphs determine the

1148: corresponding entries in the right column of Table~\ref{table4}.}

1149: \end{figure}

1150: 

1151: Supplementing the the specific Fokker--Planck energies of the $M = 0,1$ states

1152: presented in Tables~\ref{table3} and \ref{table4}, 

1153: we indicate in Fig.~\ref{fig6} 

1154: the $g$--dependence of the energies $E^{(0)}_{\rm FP}$ and $E^{(1)}_{\rm FP}$

1155: as obtained from the analysis of the function $Q(E, g)$ and 

1156: compare to reference values obtained from the 

1157: diagonalization of the Hamiltonian matrix. As indicated in Fig.~\ref{fig6},

1158: the results obtained by a direct resummation of the 

1159: quantization condition give a good quantitative picture 

1160: of the ground--state and the first excited--state eigenvalues in the 

1161: ranges $0 \le g \le 0.7$ and $0 \le g \le 0.3$, respectively. 

1162: The behavior of the numerical uncertainty as a function of $g$

1163: is explicitly shown in Fig.~\ref{fig7}, where we plot the quantity

1164: %

1165: \begin{equation}

1166: \label{relative_error}

1167: \Delta(g) \, = \, 

1168: \left| \frac{E_{\rm FP, \, resum}^{(M)}(g) - 

1169: E_{\rm FP, \, diag}^{(M)}(g)}{E_{\rm FP, \, diag}^{(M)}(g)} 

1170: \right| \, ,

1171: \end{equation}

1172: %

1173: with $E^{(M)}_{\rm FP, \, resum}(g)$ and $E_{\rm FP, \, diag}^{(M)}(g)$ being 

1174: the eigenvalues as obtained from the direct 

1175: resummation of the quantization condition given in 

1176: Eq.~(\ref{Q_zeros}) and the 

1177: diagonalization of the Hamiltonian (\ref{hamfp}), 

1178: respectively (the latter values, which are numerically 

1179: more accurate, are taken as the reference values). While the accuracy of 

1180: resummations for the ground--state energy remains 

1181: satisfactory even for (relatively) strong coupling, the relative error 

1182: for the first excited is numerically much more significant.

1183: 

1184: \begin{table}[t]

1185: \caption{\label{table4} The energy of the first excited state 

1186: with $M=1$

1187: of the Fokker--Planck Hamiltonian. Results have been 

1188: computed by the diagonalization of the Hamiltonian in the basis of the 

1189: harmonic oscillator wavefunctions (exact energy) and by solving 

1190: Eq.~(\ref{Q_zeros}).}

1191: \vspace*{0.3cm}

1192: \begin{scriptsize}

1193: \begin{center}

1194: \begin{tabular}{l@{\hspace*{0.5cm}}l@{\hspace*{0.5cm}}l} 

1195: \hline

1196: \hline

1197: \rule[-3mm]{0mm}{8mm}

1198: $g$   & $E^{(M = 1)}_{\rm FP}$ (diagonalization) & zero of the Re[$Q(E, g)$]  \\

1199: \hline

1200: \rule[-3mm]{0mm}{8mm}

1201: 0.010 &             9.677 074 461 352   $\times$ 10$^{-01}$  &

1202:          \underline{9.677 074 461 352}  $\times$ 10$^{-01}$ \\

1203: \rule[-3mm]{0mm}{8mm}

1204: 0.020 &             9.219 489 780 495   $\times$ 10$^{-01}$  &

1205:          \underline{9.219 4}90(3)  $\times$ 10$^{-01}$ \\

1206: \rule[-3mm]{0mm}{8mm}

1207: 0.030 &             8.354 795 860 905   $\times$ 10$^{-01}$  &

1208:          \underline{8.354} 4(6)         $\times$ 10$^{-01}$ \\

1209: \rule[-3mm]{0mm}{8mm}

1210: 0.070 &             6.828 548 309 058   $\times$ 10$^{-01}$  &

1211:          \underline{6.8}33(4)           $\times$ 10$^{-01}$ \\

1212: \rule[-3mm]{0mm}{8mm}

1213: 0.200 &             8.710 869 037 634   $\times$ 10$^{-01}$ & 

1214:          \underline{8.7}6(5)            $\times$ 10$^{-01}$ \\

1215: \rule[-3mm]{0mm}{8mm}

1216: 0.250 &             9.508 936 793 119   $\times$ 10$^{-01}$ & 

1217:          \underline{9}.4(2)             $\times$ 10$^{-01}$ \\

1218: \hline

1219: \hline

1220: \end{tabular}

1221: \end{center}

1222: \end{scriptsize}

1223: \end{table}

1224: 

1225: 

1226: %

1227: % Strong-coupling expansion

1228: %

1229: \subsection{Strong--coupling expansion}

1230: \label{largec}

1231: 

1232: As discussed in the previous Section, a numerical procedure based on

1233: the generalized Bohr--Sommerfeld formulae may provide relatively accurate

1234: estimates of the ground as well as the (first

1235: two) excited--state energies for the 

1236: coupling parameters in the range $0 \le g \le 0.3$. 

1237: The question arises whether $g \approx 0.3$ can be considered as 

1238: belonging to the strong coupling regime. Since the minima of the 

1239: first two excited-state energies occur at $g = 0.07$ and $g = 0.025$,

1240: respectively, one might be tempted to answer the question affirmatively.

1241: However, one could devise a different

1242: criterion for the transition to the strong

1243: coupling regime. For instance, one might define the strong--coupling region 

1244: as a region of an appropriately specified

1245: large-coupling asymptotic behavior of the eigenvalues 

1246: $E_{\rm FP}^{(M)}(g)$. 

1247: 

1248: The large--coupling asymptotics of the Fokker--Planck potential thus 

1249: represents a natural next aim in the current investigation. 

1250: To this end, we apply a

1251: so--called Symanzik scaling $q \rightarrow g^{-1/6} q$ in Eq.~(\ref{hamfp}) 

1252: and rewrite the Fokker--Planck potential into another one 

1253: with the same eigenvalues but a fundamentally different structure 

1254: \cite{SiDi1970, We1996b}: 

1255: %

1256: %

1257: \begin{eqnarray}

1258: \label{scaled_hamiltonian}

1259: H_{\rm FP} = g^{1/3} 

1260: \left[ H_{\rm L} + 

1261: \left( -q^3 - \frac{1}{2}\right)g^{-1/3} + \frac{q^2}{2} g^{-2/3}

1262: \right] \, ,

1263: \end{eqnarray}

1264: %

1265: %

1266: where the Hamiltonian $H_{\rm L}$ does not depend on $g$:

1267: %

1268: %

1269: \begin{eqnarray}

1270: \label{HL}

1271: H_{\rm L} = -\frac{1}{2} \left( \frac{d}{dq} \right)^2 + 

1272: q + \frac{q^4}{2} \, .

1273: \end{eqnarray}

1274: %

1275: % 

1276: We conclude that the $M$th eigenvalue of the Fokker--Planck Hamiltonian

1277: for the $g \rightarrow \infty$ is determined in leading order by the 

1278: $M$th eigenvalue $E_{\rm L}^{(M)}$ of the Hamiltonian (\ref{HL}):

1279: %

1280: %

1281: \begin{equation}

1282:    \label{leading_asymptotics}

1283:    E_{\rm FP}^{(M)}(g) \approx g^{1/3} E_{\rm L}^{(M)} \, .

1284: \end{equation}

1285: %

1286: % 

1287: Moreover, Eq.~(\ref{leading_asymptotics}) also indicates that 

1288: the classifications of the levels $M = 0, 1, 2, \dots$ of the

1289: Fokker--Planck and the $H_{\rm L}$ Hamiltonians are obviously identical 

1290: in the strong--coupling regime.

1291: 

1292: \begin{figure}[t]

1293: %\vspace*{-2.0cm}

1294: \begin{center}

1295: \includegraphics[height=0.9\linewidth,angle=270]{Fig6.eps}

1296: \end{center}

1297: %\vspace*{-5.5cm}

1298: \caption{\label{fig6} (Color online) Eigenvalues $E^{(M = 0)}_{\rm FP}$ (left panel) and 

1299: $E^{(M = 1)}_{\rm FP} \equiv E^{(K = 1, \varepsilon = +1)}_{\rm FP}$ (right panel) 

1300: of the Fokker--Planck Hamiltonian 

1301: as a functions of the coupling parameter $g$. Results have been computed by the

1302: diagonalization of the Fokker--Planck Hamiltonian in the basis of the harmonic 

1303: oscillator wavefunctions (solid lines) and by solving Eq.~(\ref{Q_zeros}),

1304: where the latter correspond the the data points with the error bars.}

1305: \end{figure}

1306: 

1307: Based on Eq.~(\ref{leading_asymptotics}), we now wish to compute the leading

1308: asymptotics of the ground $E_{\rm FP}^{(M = 0)}(g)$ and excited--state 

1309: $E_{\rm FP}^{(M = 1, 2)}(g)$ energies of the Fokker--Planck

1310: Hamiltonian. This computation obviously requires information about the

1311: corresponding eigenvalues of the Hamiltonian $H_{\rm L}$. The

1312: energies $E_{\rm L}^{(M)}$ have again been determined by a

1313: diagonalization of the 

1314: Hamiltonian matrix within a basis of up to 1000 harmonic oscillator

1315: wavefunctions and than utilized in Eq.~(\ref{leading_asymptotics}). 

1316: As seen from Fig.~\ref{fig8}, the leading asymptotics of the eigenvalues 

1317: $E_{\rm FP}^{(M = 0, 1, 2)}(g)$ (dotted line), 

1318: calculated in such a way, significantly 

1319: overestimate the energies of the Fokker--Planck Hamiltonian for the

1320: region $0 \le g \le 0.3$. Higher--order corrections

1321: to the large-coupling asymptotics are therefore required,

1322: in order to reproduce more accurately the asymptotics of the eigenvalues

1323: $E_{\rm FP}^{(M = 0, 1, 2)}(g)$. We observe that 

1324: we may apply standard Rayleigh--Schr\"{o}dinger perturbation theory 

1325: to Eq.~(\ref{scaled_hamiltonian}) and use the fact that 

1326: the perturbative with respect to $H_{\rm L}$, which is 

1327: $V(g) = \left( -q^3 - 1/2 \right)g^{-1/3} + q^2 g^{-2/3}/2$,

1328: remains Kato--bounded

1329: with respect to the unperturbed Hamiltonian $H_{\rm L}$ for large $g$.

1330: Within such an approach, a strong--coupling perturbation expansion can be

1331: written for each energy $E_{\rm FP}^{(M)}(g)$:

1332: %

1333: %

1334: \begin{equation}

1335:    \label{strong_coupling_expansion}

1336:    E_{\rm FP}^{(M)}(g) = g^{1/3} \sum\limits_{k = 0}^{\infty} L^{(M)}_k 

1337:    g^{-2k/3} \,

1338: \end{equation}

1339: %

1340: % 

1341: where $L^{(M)}_0 \equiv E_{\rm L}^{(M)}$ and the higher perturbation

1342: coefficients $L^{(M)}_{k > 0}$ are calculated in the basis of the wavefunctions

1343: of the unperturbed Hamiltonian (\ref{HL}). The first six $L^{(M)}_{k}$

1344: coefficients for the ground $M$ = 0 as well as the first excited  $M$ = 1, 2

1345: states are given in Table~\ref{table5}. 

1346: 

1347: \begin{figure}[t]

1348: \begin{center}

1349: %\vspace*{-0.3cm}

1350: \includegraphics[width=0.6\linewidth,angle=270]{Fig7.eps}

1351: \end{center}

1352: %\vspace*{-0.6cm}

1353: \caption{\label{fig7} (Color online) Relative error (\ref{relative_error}) of 

1354: calculations of the eigenvalues as a function of coupling constant $g$. 

1355: Results are presented for the ground ($M$ = 0) and the first excited 

1356: ($M$ = 1) states of the Fokker--Planck potential.}

1357: \end{figure}

1358: 

1359: In Fig.~\ref{fig8}, we implemented the first few 

1360: expansion coefficients as listed in Table~\ref{table5}, in order to calculate 

1361: strong-coupling asymptotics 

1362: for the three lowest levels of the Fokker--Planck Hamiltonian 

1363: (dashed lines). Obviously, the 

1364: partial sum of the expansion (\ref{strong_coupling_expansion}) as defined by

1365: only six coefficients provides much better agreement with 

1366: numerically determined energy eigenvalues 

1367: than the leading asymptotics (\ref{leading_asymptotics}). For instance, 

1368: the energy of the first excited state $M = 1$ is very well described by the

1369: (first six terms of the) strong--coupling expansion already at $g = 0.07$

1370: (the agreement is better than one percent). 

1371: This point, which

1372: also corresponds to the minimum of the function $E_{\rm FP}^{(M = 1)}(g)$, 

1373: can thus naturally be identified as the transition region

1374: between the regimes of weak and strong coupling. By using such a 

1375: definition for the transitory regime, we

1376: may finally conclude that the resummation of the Bohr--Sommerfeld quantization

1377: formulae (\ref{qcond})--(\ref{A_expansion}),

1378: using the condition (\ref{Q_zeros}), may provide reasonable estimates

1379: for the energy levels $E_{\rm FP}^{(M)}(g)$ even in a limited 

1380: subregion of the strong-coupling regime. 

1381: 

1382: 

1383: %

1384: % Quantum dynamics

1385: %

1386: \subsection{Quantum dynamics}

1387: \label{quadyn}

1388: 

1389: In previous Sections of the current paper, 

1390: we have presented a systematic study of the energy

1391: spectrum of the Fokker--Planck potential. In particular, two methods for

1392: computing the eigenvalues $E_{\rm FP}^{(M)}(g)$ 

1393: have been discussed in detail: (i) 

1394: a ``brute--force'' method which is based on the diagonalization of the 

1395: Hamiltonian matrix and (ii) a generalized perturbative approach which 

1396: accounts for the instanton effects.  

1397: While, of course, the precise computation of the energy levels is 

1398: a very important

1399: task, the complete description of the properties of the particular 

1400: Hamiltonian also requires an access to its wavefunctions 

1401: $\Phi^{(M)}_{\rm FP}(q)$. Most naturally, these wavefunctions may be obtained

1402: together with the eigenvalues $E_{\rm FP}^{(M)}(g)$ by 

1403: matrix diagonalization.

1404: In our calculations, the basis of the standard harmonic

1405: oscillator wavefunction $\{ \phi_n(q) \}_{n=0}^{\infty}$ is used for

1406: the construction of a Hamiltonian matrix 

1407: $\mem{\phi_n}{H_{\rm FP}}{\phi_m}$. The 

1408: Fokker--Planck eigenfunctions are thus given by:

1409: %

1410: %

1411: \begin{equation}

1412:    \label{FP_wavefunctions}

1413:    \Phi^{(M)}_{\rm FP}(q) = 

1414:    \sum\limits_{n = 0}^{\infty} c^{(M)}_n \phi_n(q) \, ,

1415: \end{equation}

1416: %

1417: % 

1418: where the coefficients $c^{(M)}_n$ are found by the diagonalization procedure.

1419: In Fig.~\ref{fig9} we display, for example, the wavefunctions 

1420: (\ref{FP_wavefunctions}) of the ground $M$ = 0 and the first excited 

1421: $M = 1, 2$ states as calculated for a coupling parameter $g = 0.05$. 

1422: As is evident from Fig.~\ref{fig9}, the 

1423: symmetry--breaking term of the Hamiltonian 

1424: (\ref{hamfp}) leads to a ground--state wavefunction 

1425: $\Phi^{(M = 0)}_{\rm FP}(q)$ (dashed line) which is neither a symmetric nor 

1426: antisymmetric combination of the wavefunctions of the right and left wells,

1427: but localized in the lower well. For the first

1428: excited states, in contrast, the symmetry is partially restored, and we may

1429: attribute the wavefunctions $\Phi^{(M = 1)}_{\rm FP}(q)$ (dashed and dotted 

1430: line) and $\Phi^{(M = 2)}_{\rm FP}(q)$ (dotted line) to states with 

1431: positive ($\varepsilon$ = +1) and negative ($\varepsilon$ = -1) parities 

1432: [see also Eqs.~(\ref{FP_M_to_K})---(\ref{FP_M_to_epsilon})].

1433: 

1434: %

1435: %

1436: %\begin{widetext}

1437: \begin{center}

1438: \begin{table}

1439: \begin{center}

1440: %\begin{minipage}{8.6cm}

1441: \caption{\label{table5} $L^{(M)}_{k}$ coefficients of the strong--coupling

1442: perturbation expansion (\ref{strong_coupling_expansion}) of the eigenvalues

1443: $E_{\rm FP}^{(M = 0, 1, 2)}$ of the Fokker--Planck Hamiltonian.}

1444: \begin{tabular}{c@{\hspace{0.6cm}}r@{\hspace{0.3cm}}r@{\hspace{0.3cm}}r}

1445: \hline

1446: \hline

1447: %\rule[-3mm]{0mm}{8mm}

1448: $j$ &

1449: \multicolumn{1}{c}{$M=0$} &

1450: \multicolumn{1}{c}{$M=1$} &

1451: \multicolumn{1}{c}{$M=2$} \\

1452: \hline

1453: 0 &  $0.281\,067\,805$ &  $1.854\,587\,292$ &  $3.686\,419\,624$ \\

1454: 1 & $-0.132\,985\,313$ & $-0.209\,853\,650$ & $-0.307\,985\,031$ \\

1455: 2 &  $0.021\,367\,333$ &  $0.028\,174\,214$ &  $0.025\,765\,237$ \\

1456: 3 & $-0.000\,876\,935$ &  $0.000\,220\,875$ & $-0.000\,099\,791$ \\

1457: 4 & $-0.000\,060\,335$ &  $0.000\,031\,789$ & $-0.000\,002\,135$ \\

1458: 5 & $-0.000\,001\,557$ &  $0.000\,000\,739$ & $-0.000\,000\,385$ \\

1459: \hline

1460: \hline

1461: \end{tabular}

1462: %\end{minipage}

1463: \end{center}

1464: \end{table}

1465: \end{center}

1466: %\end{widetext}

1467: %

1468: %

1469: 

1470: Until now, we have only discussed the evaluation of the eigenstates

1471: and eigenenergies 

1472: of the Fokker--Planck Hamiltonian. In theoretical studies of 

1473: double quantum-dot nanostructures \cite{GrDiJuHa1991, Op1999, GrHa1998, TsOp2004}

1474: and of the quantum tunneling phenomena in atomic physics

1475: \cite{LiBa1990, LiBa1992}, a large number of problems arise, 

1476: in which the time-propagation of some (specially prepared) wavepacket in 

1477: double-well-like potentials has to be considered. The methods

1478: for such a time-propagation, such as the well-known Crank-Nicolson method, 

1479: the split-operator technique, and approaches based on the Floquet 

1480: formalism and many others, are discussed in detail 

1481: in the literature (see, e.g., 

1482: Refs.~\onlinecite{GrDiJuHa1991, GrHa1998,CrNi1947, FlMoFe1976, PrKeKn1997}). 

1483: As a supplement to our previous considerations,

1484: we will now consider the time-propagation of an initial wavepacket in a

1485: double-well-like potential, recalling the adiabatic

1486: approach as one of the most simple and best-known techniques

1487: (see, e.g., the book~\onlinecite{Te2003} and references therein) 

1488: for the integration

1489: of the (time--dependent) single particle Schr\"odinger equation.

1490: Within such a technique, in which we can naturally make use of the 

1491: results previously derived for the eigenstates

1492: and eigenenergies, the propagation of a wavepacket $\Psi(q, t)$ in 

1493: the (time--independent) Fokker--Planck potential (\ref{hamfp}) is 

1494: given by:

1495: %

1496: %

1497: \begin{equation}

1498: \label{propagation_FP}

1499: \Psi(q, t) = \sum\limits_{M=0}^{\infty} 

1500: b^{(M)} \,

1501: \exp\left(-{\rm i}\, E^{(M)}_{\rm FP} t\right) \, 

1502: \Phi^{(M)}_{\rm FP}(q) \,.

1503: \end{equation}

1504: %

1505: %   

1506: Here the coefficients $b^{(M)} = \sprm{\Phi^{(M)}_{\rm FP}}{\Psi(t=0)}$

1507: determine the decomposition of the initial wavepacket (at $t$ = 0) in the basis

1508: of the eigenfunctions (\ref{FP_wavefunctions}).

1509: 

1510: Equation~(\ref{propagation_FP}) provides an exact solution for

1511: the wavefunction $\Psi(q, t)$ at an arbitrary moment of time only 

1512: in the limit of an infinitely large basis of harmonic oscillator 

1513: $\{ \phi_n(q) \}_{n=0}^{\infty}$ and Fokker--Planck 

1514: $\{ \Phi^{(M)}_{\rm FP}(q) \}_{M=0}^{\infty}$ wavefunctions. For computational

1515: reasons, however, summations over basis functions have to be 

1516: restricted to a finite number. In our calculations,

1517: basis sets of 300--1000 wavefunctions have been applied depending on the

1518: parameters of the initial wavepacket and the coupling parameter $g$. The

1519: actual size of the basis has been chosen according to the numerical checks of

1520: the Ehrenfest theorem or the normalization of the wavepacket.    

1521: 

1522: \begin{figure}[t]

1523: %\vspace*{-2.0cm}

1524: \begin{center}

1525: \includegraphics[width=0.8\linewidth,angle=270]{Fig8.eps}

1526: \end{center}

1527: %\vspace*{-0.5cm}

1528: \caption{\label{fig8} (Color online)

1529: Exact values (solid line) for the ground state 

1530: and the two lowest excited energy levels 

1531: for the Fokker--Planck potential as a function of $g$, together with the

1532: leading asymptotics (dotted line) and the  

1533: partial sum of the strong-coupling expansion (dashed line) as defined by the 

1534: first six nonvanishing terms in powers of $ g^{-2k/3}$

1535: [see Eq.~(\ref{strong_coupling_expansion})], which are

1536: listed in Table~\ref{table3}. Note that the leading strong-coupling 

1537: asymptotics alone cannot satisfactorily 

1538: describe the true energy eigenvalues for 

1539: moderate and small coupling. By contrast, the partial sum

1540: of the first six nonvanishing terms 

1541: of the strong-coupling asymptotics yields numerical values

1542: which are indistinguishable from the true eigenvalues on the level 

1543: of the line thickness used in the plots, down to rather small

1544: values of the coupling (dashed vs.~solid lines).}

1545: \end{figure}

1546: 

1547: Within the adiabatic approximation, which is valid for 

1548: slowly varying potentials~\cite{Te2003,CN}, we may divide the 

1549: time evolution of the potential into small intervals $\Delta t$ and assume that

1550: for every $k$th interval the Hamiltonian is time--independent and the 

1551: propagation of the wavepacket is governed again by Eq.~(\ref{propagation_FP})

1552: where, of course, eigenvalues $E^{(M)}_{\rm FP}$ and 

1553: eigenfunctions $\Phi^{(M)}_{\rm FP}(q)$ should be replaced with the eigenvalues

1554: $E^{(M)}_{k}$ and eigenfunctions $\Phi^{(M)}_{k}(q)$ of the 

1555: Hamiltonian $H_k \equiv H(t_k)$. We have applied this adiabatic 

1556: time--propagation method, 

1557: whose variations are well known from the 

1558: literature \cite{HwPe1977,Ve2000,Su2003,Te2003} and which is 

1559: equivalent to an exponentiation of the instantaneous

1560: Hamiltonian for each time interval $\Delta t$,

1561: to investigate the evolution of an (initially) Gaussian wavepacket in 

1562: a time--dependent potential (\ref{hami}) which oscillates 

1563: sinusoidally between the Fokker--Planck and the 

1564: double-well cases. Since the animated results of this simulation are 

1565: available for download\cite{JeHomeHD}, we just present a

1566: small series of snapshots in Fig.~\ref{fig10}. 

1567: As seen from these pictures, the wavepacket, which 

1568: is initially located in the right well, performs 

1569: oscillations between the wells.

1570: These oscillations are controlled by the temporal change of 

1571: the potential (\ref{hami}). We have checked empirically that the adiabatic

1572: approximation employed here does not represent an obstacle for an accurate

1573: time evolution in even rapidly oscillating potentials,

1574: because of the calculational efficiency of the other steps in our time

1575: propagation algorithm (notably, the diagonalization including the 

1576: determination of eigenvectors can be implemented 

1577: in a computationally very favorable way on modern computers).

1578: It is thus possible to perform quantum dynamical simulations 

1579: in potentials which oscillate between two limiting forms with 

1580: two fundamentally different characteristic ground-state configurations,

1581: each of which is governed by instantons, though in a different way.

1582: A generalization of our approach to two-dimensional potentials

1583: appears to be feasible and is currently being studied.

1584: 

1585: \begin{figure}[t]

1586: %\vspace*{-1.5cm}

1587: \begin{center}

1588: \includegraphics[width=0.5\linewidth,angle=270]{Fig9.eps}

1589: \end{center}

1590: %\vspace*{-0.6cm}

1591: \caption{\label{fig9} (Color online)

1592: The wavefunctions $\Phi^{(M)}_{\rm FP}(q)$ of the ground $M$ = 0 and the first

1593: excited $M$ = 1, 2 states of the Fokker--Planck Hamiltonian calculated at

1594: coupling parameter $g$ = 0.05. The base lines for the plots of the wave 

1595: functions correspond to their energies $E_{\rm FP}^{(M)}(g)$.}

1596: \end{figure}

1597: 

1598: %

1599: % Conclusions

1600: %

1601: \section{Conclusions}

1602: \label{conclu}

1603: 

1604: In this paper, 

1605: we have investigated the Fokker--Planck [Eq.~(\ref{hamfp})]

1606: and the double-well [Eq.~(\ref{hamdw})] potential from the point of 

1607: view of large-order perturbation theory (resurgent expansion and 

1608: generalized quantization condition), in order to map out the regimes 

1609: of validity of the instanton-related resurgent expansion for the 

1610: lowest energy levels, and in order to explore the possibility of 

1611: reaching the strong-coupling regime via a direct resummation

1612: of the $A_{\rm FP}$ and $B_{\rm FP}$ functions

1613: given in Eqs.~(\ref{B_expansion}) and (\ref{A_expansion}), which enter the 

1614: generalized quantization condition (\ref{qcond}).

1615: The latter approach entails a complete double resummation

1616: of the resurgent expansions (\ref{eground}) and 

1617: (\ref{efpgen}) both in powers of the 

1618: instanton coupling $\exp(-1/6g)$ and in powers of the coupling $g$

1619: (perturbative expansion about each instanton). The quest has been

1620: to explore the applicability of resummed expansions for medium

1621: and large coupling parameters, in the transitory regime to large coupling.

1622: 

1623: It is quite natural to identify the transition region for the 

1624: first excited state as defined 

1625: by the minimum of the energy level $E_{\rm FP}^{(M=1)}(g)$ as a 

1626: function of $g$ (see the right panel of Fig.~\ref{fig6}),

1627: which occurs near $g \approx 0.07$.

1628: As is evident from Figs.~\ref{fig6} and~\ref{fig8}, it is possible to reach 

1629: convergence for both the resummed secular equation (\ref{Q_zeros}) 

1630: as well as convergence of the strong-coupling 

1631: expansion~(\ref{strong_coupling_expansion}) 

1632: in a somewhat restricted overlap region $0.04 \lesssim g \lesssim 0.3$.

1633: The question whether it is possible to use the resummed instanton

1634: expansion for large coupling, cannot universally be answered affirmatively, 

1635: although it is reassuring to find at 

1636: least a restricted region of overlap. In order to interpret

1637: the overlap, one should remember that for 

1638: typical Borel summable series

1639: as they originate in various contexts in field theory,

1640: it is much easier to perform resummations at moderate and even large

1641: coupling parameters than for the instanton-related case considered

1642: here. An example is the 30-loop resummation

1643: of the anomalous dimension $\gamma$ function of six-dimensional

1644: $\phi^3$-theories and of Yukawa model theories~\cite{BrKr2000},

1645: which lead to excellent convergence for couplings as large

1646: as $g = 10$ and higher~\cite{JeSo2001}. 

1647: 

1648: %

1649: %

1650: \begin{figure}[t]

1651: %\vspace*{-1.5cm}

1652: \begin{center}

1653: \includegraphics[width=0.7\linewidth,angle=270]{Fig10.eps}

1654: \end{center}

1655: %\vspace*{-0.8cm}

1656: \caption{\label{fig10} (Color online)

1657: Time propagation of the (initially) Gaussian wavepacket in the time--dependent

1658: potential (\ref{hami}) with $\eta = \sin(t/60)$ which oscillates between the 

1659: double--well and Fokker--Planck potentials. The base line

1660: for the plot of the wave function corresponds to its instantaneous 

1661: energy.}

1662: \end{figure}

1663: %

1664: %

1665: 

1666: For the latter case, it is even possible to obtain, numerically,

1667: the strong-coupling asymptotics on the basis of a resummed

1668: weak-coupling perturbation theory (see, e.g., Ref.~\onlinecite{Su2001}

1669: for a remarkable realization of this idea in an 

1670: extremely nontrivial context). The general notion is that any perturbation

1671: (a potential in the case of quantum mechanics and 

1672: an interaction Lagrangian in the case of field theory)

1673: determines the large-order behavior of perturbation series

1674: describing a specific physical quantity.

1675: However, the potential or interaction Lagrangian 

1676: also determine the large-coupling expansion for the 

1677: physical quantity under investigation.

1678: This means that there is a connection between the large-coupling expansion

1679: and the large-order behavior of the perturbation series generated

1680: in each theory, and this correspondence can be exploited in order

1681: to infer strong-coupling asymptotics even in cases where only 

1682: a few perturbative terms are known~\cite{Su2001}.

1683: According to our numerical 

1684: results, the corresponding calculation of strong-coupling asymptotics

1685: on the basis of only a few perturbative terms would be much more difficult in

1686: cases where instantons are present, even if like in our case,

1687: additional information is present in the form of the generalized quantization

1688: condition (\ref{qcond}).

1689: 

1690: We deliberately refrain from speculating about further implications of this

1691: observation and continue with a summary of the application-oriented 

1692: results gained in the current investigation.  

1693: Potentials of the double-well type are important for a number

1694: of application--oriented calculations, including

1695: Josephson junction qubits~\cite{BeEtAl2003, WeEtAl2005}, 

1696: inversion doubling in molecular physics \cite{Hu1927,Xu2005},

1697: semiconductor double 

1698: quantum dots~\cite{LoDV1998, HoEtAl2002, HuEtAl2005},

1699: as well as, in a wider context, Bose--Einstein condensates in 

1700: multi--well traps~\cite{StCeMo2004, AlStCe2005, StAlCe2006, AlEtAl2005}.

1701: For the latter case, 

1702: a number of theoretical works have been performed recently in 

1703: order to explore the time propagation of (one particle) 

1704: wavepackets in driven double--well potentials

1705: \cite{GrDiJuHa1991,Op1999,TsOp2004}. In Sec.~\ref{numer}

1706: we discuss a numerical procedure for an accurate description of 

1707: energy levels and of the corresponding wave functions, which can thus be 

1708: used in order to construct basis sets for an accurate quantum dynamical time

1709: evolution of wave packets in both static

1710: as well as time--dependent potentials. This well-known adiabatic technique 

1711: for the integration of the (single particle) time-dependent 

1712: Schr\"odinger equation is briefly recalled in Sec.~\ref{quadyn}.

1713: We illustrate this technique by a calculation for a potential

1714: which oscillates between the Fokker--Planck and the double-well

1715: cases, governed by a time-dependent

1716: interpolating parameter $\eta = \cos(\omega t)$ 

1717: as given in Eq.~(\ref{hami}). Similar calculations can be done for 

1718: cases where the potential admits resonances as in the 

1719: case of a cubic anharmonic oscillator. In this case, the 

1720: method of complex scaling leads to a basis of states which can be used 

1721: in order to start quantum dynamical simulations. 

1722: Related work is currently in progress.

1723: 

1724: %

1725: % Acknowledgments

1726: %

1727: \section{Acknowledgments}

1728: 

1729: U.D.J.~acknowledges support from the Deutsche

1730: Forschungsgemeinschaft (Heisenberg program).

1731: M.L.~is grateful to Max--Planck--Institute for Nuclear Physics

1732: for the stimulating atmosphere during a guest researcher

1733: appointment, on the occasion of which the current work was

1734: completed. 

1735: 

1736: \appendix

1737: 

1738: %

1739: % Supersymmetry and the Fokker--Planck Potential

1740: %

1741: \section{Supersymmetry and the Fokker--Planck Potential}

1742: \label{appa}

1743: 

1744: This appendix is meant to provide a brief identification 

1745: of the Fokker--Planck potential (\ref{hamfp}) in terms

1746: of a supersymmetric (SUSY) algebra~\cite{CoKhSu1995,KaSo1997}. 

1747: We define the operators

1748: %

1749: \begin{equation}

1750: B^\pm = \frac{1}{\sqrt{2}} \, 

1751: \left( W(q) \mp \frac{d}{dq} \right)

1752: \end{equation}

1753: %

1754: with $W(q) = q \, (1- \sqrt{g} \, q)$ and the SUSY Hamiltonian 

1755: %

1756: \begin{equation}

1757: H_{\rm SUSY} = 

1758: \left( \begin{array}{cc} B^+ B^- & 0 \\ 0 & B^- B^+ \end{array} \right) =

1759: \left( \begin{array}{cc} H_1 & 0 \\ 0 & H_2 \end{array} \right) \,.

1760: \end{equation}

1761: %

1762: with 

1763: %

1764: \begin{subequations}

1765: \begin{align}

1766: H_1 =& -\frac{1}{2} \left( \frac{d}{dq} \right)^2 + 

1767: \frac{1}{2} q^2 \left( 1 - \sqrt{g}q \right)^2 + 

1768: \sqrt{g} q - \frac{1}{2} \,, \\

1769: %

1770: H_2 =& -\frac{1}{2} \left( \frac{d}{dq} \right)^2 + 

1771: \frac{1}{2} q^2 \left( 1 - \sqrt{g}q \right)^2 - 

1772: \sqrt{g} q + \frac{1}{2} \,.

1773: \end{align}

1774: \end{subequations}

1775: %

1776: Notice that $W(q)$ finds a natural interpretation as a 

1777: ``superpotential'' in the sense of Refs.~\onlinecite{CoKhSu1995,KaSo1997}.

1778: The Hamiltonians $H_1$ and $H_2$ are ``superpartners.''

1779: They are related to each other by a simple reflection and 

1780: translation, $q \to 1/\sqrt{g} - q$, and have the same 

1781: spectra. In that sense, one may say that the Fokker--Planck 

1782: potential is its own superpartner. The Fokker--Planck potential

1783: therefore constitutes 

1784: a case of ``broken supersymmetry'' with zero Witten index

1785: [see, e.g., Eq.~(2.88) of Ref.~\onlinecite{KaSo1997}].

1786: The construction of the supersymmetric partner thus does not 

1787: help in the analysis of the Fokker--Planck Hamiltonian. One is 

1788: forced into the instanton-inspired analysis presented in the current 

1789: study. 

1790: 

1791: As a last remark, we recall that the Fokker--Planck potential

1792: %

1793: \begin{equation}

1794: V_{\rm FP}(q) = \frac{1}{2} q^2 \left( 1 - \sqrt{g}q \right)^2 + 

1795: \sqrt{g} q - \frac{1}{2} = \frac12\, \left[W^2(q) - W'(q) \right] 

1796: \end{equation}

1797: %

1798: could be assumed to admit a zero eigenvalue. However, 

1799: as is evident from the 

1800: discussion following Eq.~(7.28) of Ref.~\onlinecite{ZJJe2004ii},

1801: the corresponding eigenfunction is not normalizable, 

1802: and thus cannot be interpreted as a physical state vector.

1803: Neither the Fokker--Planck potential nor its isospectral

1804: supersymmetric partner admit a zero eigenvalue.

1805: 

1806: \begin{thebibliography}{10}

1807: 

1808: \bibitem{ZJJe2004i}

1809: J. Zinn-Justin and U.~D. Jentschura, Ann. Phys. (N.Y.) {\bf 313},  197  (2004).

1810: 

1811: \bibitem{ZJJe2004ii}

1812: J. Zinn-Justin and U.~D. Jentschura, Ann. Phys. (N.Y.) {\bf 313},  269  (2004).

1813: 

1814: \bibitem{JeZJ2001}

1815: U.~D. Jentschura and J. Zinn-Justin, J. Phys. A {\bf 34},  L253  (2001).

1816: 

1817: \bibitem{JeZJ2004plb}

1818: U.~D. Jentschura and J. Zinn-Justin, Phys. Lett. B {\bf 596},  138  (2004).

1819: 

1820: \bibitem{HeSi1978}

1821: I.~W. Herbst and B. Simon, Phys. Rev. Lett. {\bf 41},  67  (1978).

1822: 

1823: \bibitem{LuEtAl1989}

1824: M. Luban, J.~H. Luscombe, M.~A. Reed, and D.~L. Pursey, Appl. Phys. Lett. {\bf

1825:   54},  1997  (1989).

1826: 

1827: \bibitem{LoDV1998}

1828: D. Loss and D.~P. DiVincenzo, Phys. Rev. A {\bf 57},  120  (1998).

1829: 

1830: \bibitem{HoEtAl2002}

1831: A.~W. Holleitner, R.~H. Blick, A.~K. H\"uttel, K. Eberl, and J.~P. Kotthaus,

1832:   Science {\bf 297},  70  (2002).

1833: 

1834: \bibitem{HuEtAl2005}

1835: A.~K. H\"uttel, S. Ludwig, H. Lorenz, K. Eberl, and J.~P. Kotthaus, Phys. Rev.

1836:   B {\bf 72},  R081310  (2005).

1837: 

1838: \bibitem{GrDiJuHa1991}

1839: F. Grossmann, T. Dittrich, P. Jung, and P. H\"anggi, Phys. Rev. Lett. {\bf 67},

1840:    516  (1991).

1841: 

1842: \bibitem{Op1999}

1843: L.~A. Openov, Phys. Rev. B {\bf 60},  8798  (1999).

1844: 

1845: \bibitem{JeHomeHD}

1846: See the URL http://www.mpi-hd.mpg.de/\~{}ulj.

1847: 

1848: \bibitem{ZJ1984jmp}

1849: J. Zinn-Justin, J. Math. Phys. {\bf 25},  549  (1984).

1850: 

1851: \bibitem{CaGrMa1986}

1852: E. Caliceti, V. Grecchi, and M. Maioli, Commun. Math. Phys. {\bf 104},  163

1853:   (1986).

1854: 

1855: \bibitem{Ha1949}

1856: G.~H. Hardy, {\em Divergent Series} (Clarendon Press, Oxford, UK, 1949).

1857: 

1858: \bibitem{Je2000prd}

1859: U.~D. Jentschura, Phys. Rev. D {\bf 62},  076001  (2000).

1860: 

1861: \bibitem{JeSo2001}

1862: U.~D. Jentschura and G. Soff, J. Phys. A {\bf 34},  1451  (2001).

1863: 

1864: \bibitem{FrGrSi1985}

1865: V. Franceschini, V. Grecchi, and H.~J. Silverstone, Phys. Rev. A {\bf 32},

1866:   1338  (1985).

1867: 

1868: \bibitem{Je2001pra}

1869: U.~D. Jentschura, Phys. Rev. A {\bf 64},  013403  (2001).

1870: 

1871: \bibitem{SiDi1970}

1872: B. Simon and A. Dicke, Ann. Phys. (N.Y.) {\bf 58},  76  (1970).

1873: 

1874: \bibitem{We1996b}

1875: E.~J. Weniger, Ann. Phys. (N.Y.) {\bf 246},  133  (1996).

1876: 

1877: \bibitem{GrHa1998}

1878: M. Grifoni and P. H\"anggi, Phys. Rep. {\bf 304},  229  (1998).

1879: 

1880: \bibitem{TsOp2004}

1881: A.~V. Tsukanov and L.~A. Openov, Semiconductors {\bf 38},  91  (2004).

1882: 

1883: \bibitem{LiBa1990}

1884: W.~A. Lin and L.~E. Ballentine, Phys. Rev. Lett. {\bf 65},  2927  (1990).

1885: 

1886: \bibitem{LiBa1992}

1887: W.~A. Lin and L.~E. Ballentine, Phys. Rev. A {\bf 45},  3637  (1992).

1888: 

1889: \bibitem{CrNi1947}

1890: J. Crank and P. Nicolson, Proc. Camb. Phil. Soc. {\bf 43},  50  (1947).

1891: 

1892: \bibitem{FlMoFe1976}

1893: J.~A. Fleck, J.~R. Morris, and M.~D. Feit, Appl. Phys. A {\bf 10},  129

1894:   (1992).

1895: 

1896: \bibitem{PrKeKn1997}

1897: M. Protopapas, C.~H. Keitel, and P.~L. Knight, Rep. Prog. Phys. {\bf 60},  389

1898:   (1997).

1899: 

1900: \bibitem{Te2003}

1901: S. Teufel, {\em Adiabatic Perturbation Theory in Quantum Dynamics --- Lecture

1902:   Notes in Mathematica Vol. 1821} (Springer, Berlin, Heidelberg, New York,

1903:   2003).

1904: 

1905: \bibitem{CN}

1906: We note that our adiabatic approach, for slowly varying potentials, is limited

1907:   essentially by the incompleteness of the basis used in each time propagation

1908:   step, while the principal restriction of the familiar Crank--Nicolson time

1909:   propagation algorithm (see Ref.~\cite{CrNi1947}) lies in the truncation of

1910:   the exponential of the instantaneous Hamiltonian. Of course, for the

1911:   one-dimensional problems studied here, an implementation of the

1912:   Crank--Nicolson scheme also leads to a feasible alternative approach to the

1913:   investigation of the quantum dynamics. In any case, the continuous monitoring

1914:   of the Ehrenfest theorem and of the conservation of the normalization of the

1915:   wavefunction provide important cross-checks for the computational consistency

1916:   of the time propagation.

1917: 

1918: \bibitem{HwPe1977}

1919: J.~T. Hwang and P. Pechukas, J. Comp. Phys. {\bf 67},  4640  (1977).

1920: 

1921: \bibitem{Ve2000}

1922: E.~P. Velicheva, Phys. At. Nucl. {\bf 63},  661  (2000).

1923: 

1924: \bibitem{Su2003}

1925: A.~A. Suzko, Phys. Lett. A {\bf 308},  267  (2003).

1926: 

1927: \bibitem{BrKr2000}

1928: D. Broadhurst and D. Kreimer, Phys. Lett. B {\bf 475},  63  (2000).

1929: 

1930: \bibitem{Su2001}

1931: I.~M. Suslov, Pis'ma v. Zh. \'{E}ksp. Teor. Fiz. {\bf 74},  211  (2001), [JETP

1932:   {\bf 74}, 191 (2001)].

1933: 

1934: \bibitem{BeEtAl2003}

1935: A.~J. Berkley, H. Xu, R.~C. Ramos, M.~A. Gubrud, F.~W. Strauch, P.~R. Johnson,

1936:   J.~R. Anderson, A.~J. Dragt, C.~J. Lobb, and F.~C. Wellstood, Science {\bf

1937:   300},  1548  (2003).

1938: 

1939: \bibitem{WeEtAl2005}

1940: L.~F. Wei, Y.~X. Liu, and F. Nori, Phys. Rev. B {\bf 71},  134506  (2005).

1941: 

1942: \bibitem{Hu1927}

1943: F. Hund, Z. Phys. {\bf 43},  803  (1927).

1944: 

1945: \bibitem{Xu2005}

1946: C.~N. Xuan, Chem. Phys. Lett. {\bf 406},  415  (2005).

1947: 

1948: \bibitem{StCeMo2004}

1949: A.~I. Streltsov, L.~S. Cederbaum, and N. Moiseyev, Phys. Rev. A {\bf 70},

1950:   053607  (2004).

1951: 

1952: \bibitem{AlStCe2005}

1953: O.~E. Alon, A.~I. Streltsov, and L.~S. Cederbaum, Phys. Lett. A {\bf 347},  88

1954:   (2005).

1955: 

1956: \bibitem{StAlCe2006}

1957: A.~I. Streltsov, O.~E. Alon, and L.~S. Cederbaum, Phys. Rev. A {\bf 73},

1958:   063626  (2006).

1959: 

1960: \bibitem{AlEtAl2005}

1961: M. Albiez, R. Gati, J. F\"olling, S. Hunsmann, M. Cristiani, and M.~K.

1962:   Oberthaler, Phys. Rev. Lett. {\bf 95},  010402  (2005).

1963: 

1964: \bibitem{CoKhSu1995}

1965: F. Cooper, A. Khare, and U. Sukhatme, Phys. Rep. {\bf 251},  267  (1995).

1966: 

1967: \bibitem{KaSo1997}

1968: H. Kalka and G. Soff, {\em Supersymmetrie} (B. G. Teubner, Stuttgart, 1997).

1969: 

1970: \end{thebibliography}

1971: \end{document}

1972: 

1973: