1: \documentclass[aps,twocolumn,showpacs]{revtex4}
2: \usepackage{graphicx}
3:
4: \begin{document}
5:
6: \title{Effect of interactions on the localization of a Bose-Einstein condensate in a quasi-periodic lattice.}
7:
8: \author{J. E. Lye$^{1}$, L. Fallani$^{1}$, C. Fort$^{1}$, V. Guarrera$^{1}$, M. Modugno$^{2}$,
9: D. S. Wiersma$^{1}$, and M. Inguscio$^{1}$}
10:
11: \affiliation{$^{1}$ LENS, Dipartimento di Fisica and INFM
12: Universit\`a di Firenze via Nello
13: Carrara 1, I-50019 Sesto Fiorentino (FI), Italy
14: \\
15: $^{2}$ LENS, Dipartimento di Matematica Applicata, Universit\`a di
16: Firenze and BEC-INFM Center, Universit\`a di Trento, I-38050 Povo
17: (TN), Italy }
18:
19: %\affiliation{$^{1}$ LENS, Dipartimento di Fisica and INFM,
20: %Universit\`{a} di Firenze, via Nello Carrara 1, I-50019 Sesto
21: %Fiorentino (FI), Italy \\
22: % $^{2}$ LENS, Dipartimento di Matematica
23: %Applicata, Universit\`{a}
24: %di Firenze and BEC-INFM Center, Universit\`{a} di Trento, Povo, Italy \\
25: %$^{3}$INFM-MATIS, Catania, Italy}
26:
27: \begin{abstract}
28: The transport properties of a Bose-Einstein condensate in a 1D
29: incommensurate bichromatic lattice are investigated both
30: theoretically and experimentally. We observe a blockage of the
31: center of mass motion with low atom number, and a return of
32: motion when the atom number is increased. Solutions of
33: the Gross-Pitaevskii equation show how the localization due to the
34: quasi-disorder introduced by the incommensurate bichromatic
35: lattice is affected by the interactions.
36: \end{abstract}
37:
38: \pacs{03.75.Kk, 32.80.Pj, 42.25.Dd}
39:
40: %\date{\today}
41:
42: \maketitle
43:
44: The intrinsic perfection of lattices made from a standing wave
45: of light surprisingly makes them an excellent candidate for the
46: investigation of disorder in atomic systems. Free from
47: uncontrollable or undesirable defects, precise disorder can be
48: added simply in the form of additional optical lattices
49: \cite{Damski,RothandBurnett,Sanchez,Schulte,quasiperiodic,Scarola}
50: or with an optical speckle potential
51: \cite{Schulte,Lye,Clement,Fort}. The combination of optical
52: lattices with a Bose-Einstein condensate (BEC) offers the
53: stimulating complexity of interactions in a setting unimpeded by
54: large thermal fluctuations. The significant observation of the
55: Mott-Insulator phase was realized utilizing a strongly interacting
56: BEC produced in a three-dimensional ordered crystal of light
57: \cite{MIone}. Extending this work to the strongly disordered
58: regime with the inclusion of quasi-disorder from a bichromatic
59: lattice, recently led to initial experimental evidence of a
60: Bose-glass phase \cite{boseglass}. Another open question remains
61: as to the intermediate behavior between the non-interacting
62: disordered Anderson localized phase and the strongly interacting
63: Bose-glass phase \cite{logan}.
64:
65: To begin to address this cross-over regime we have carried out
66: investigations of the transport properties of a condensate in a
67: quasi-periodic lattice. The very complexity provided by the
68: interplay of disorder and interactions that makes disordered
69: condensates a stimulating topic, also can introduce instability
70: inherent in non-linear transport \cite{Fallani2004}. In fact we
71: find that dynamical instability does occur for transport in a
72: quasi-periodic lattice when the center of mass motion reaches a
73: critical velocity, and has a non-trivial dependence on the
74: interaction strength.
75:
76: %\textbf{Anderson-like localization in a bicolor lattice}
77:
78: Anderson's seminal paper in 1958 showed that the wave function of
79: a particle placed in a lattice with disordered on-site energies
80: remains localized when the range of the on-site energies is
81: sufficiently large compared to the hopping energy between
82: neighboring sites \cite{Anderson58}.
83: %The large body of work that
84: %followed in subsequent years showed that the precise distribution
85: %of the disorder, for example gaussian or box distributions, does
86: %not greatly effect the fundamental behavior of localization
87: %\cite{Kramer93}. Extending this idea further,
88: Anderson's hopping model can be approximated using a
89: quasi-periodic bichromatic lattice, obtained by superimposing a
90: primary optical lattice with a weak secondary lattice of a
91: different incommensurate wavelength. The secondary lattice serves
92: to break the discrete translational invariance of the system, thus
93: favoring localization of the wave functions, however the effect of
94: the quasi \textit{order} may be important depending on the exact
95: parameters of the bichromatic lattice
96: \cite{quasiperiodic,Sanchez}.
97:
98: In our system the 1D incommensurate bichromatic lattice is
99: produced combining the primary optical lattice derived from a
100: Titanium:Sapphire laser operating at a wavelength $\lambda_{1} =
101: 830.7(1)$ nm with a secondary lattice obtained from a
102: fiber-amplified diode laser emitting at $\lambda_{2} = 1076.8(1)$
103: nm. Our $^{87}$Rb BEC is produced in a Ioffe-Pritchard magnetic
104: trap, elongated in the direction of the bichromatic lattice. The
105: trapping frequencies are $\omega_x=2\pi \times 8.7$ Hz axially and
106: $\omega_\perp=2\pi \times 90$ Hz radially. The BEC can be produced
107: in the range of $\simeq 1.5 \times 10^4$ - $ 2 \times 10^5$ atoms.
108: The resulting potential along the lattice axis can be expressed as
109: \begin{equation}
110: V(x)=s_{1}E_{R1}\sin^2(k_{1}x)+s_{2}E_{R2}\sin^2(k_{2}x)+\frac{m
111: }{2}\omega_x^2 x^2
112: \end{equation}
113: where $s_{1}$ and $s_{2}$ measure the height of the lattice
114: potentials in units of the respective recoil energies
115: $E_{R1}=h^2/(2m\lambda_{1}^2)\simeq h \times 3.33$ kHz and $
116: E_{R2}=h^2/(2m\lambda_{2}^2)\simeq h \times 1.98$ kHz, $k_1$ and
117: $k_2$ are the wave numbers of the two lasers, $h$ is the Planck
118: constant and $m$ the mass of $^{87}$Rb.
119:
120: \begin{figure}[t!]
121: \begin{center}
122: \includegraphics[width=0.8\columnwidth]{fig1.eps}
123: \end{center}
124: \caption{Non-interacting density profiles in log scale of the
125: ground state with increasing disorder in (a) a bichromatic lattice
126: and (b) a lattice with random on-site energies. The thin line in
127: each graph represents the ground state with only the primary
128: lattice. The last graph in each column shows the on-site energies
129: in the respective potentials with the amount of disorder shown by
130: $\Delta/J \simeq 26$. In all cases the height of the primary
131: lattice is $s_1=10$ giving a tunnelling energy of $J/h=75$ Hz.}
132: \label{setup}
133: \end{figure}
134:
135: The possibility of localization with our bichromatic lattice in
136: the absence of interactions is demonstrated by numerical
137: diagonalization of the stationary 1D Schr\"{o}dinger equation
138: using the potential defined by Eq. (1). A strong primary lattice
139: is chosen, $s_1=10$, $s_1 E_{R1}/h=33$ kHz, and a perturbing
140: secondary lattice of maximum height $s_2=2$, $s_2 E_{R2}/h=4$ kHz.
141: The ground state resulting from the bichromatic lattice is shown
142: in Fig. 1(a) and is contrasted with the ground state of a pure
143: random case in (b). The random potential is simulated using only
144: the primary lattice, $s_1=10$, with additional random on-site
145: energies in a box distribution in the range [0, $\Delta$], where
146: $\Delta/h \leq 4$ kHz. The amount of the disorder is given by
147: either the height of the secondary lattice in (a) or by the
148: maximum on-site energy in the random case (b), and is denoted by
149: $\Delta$ in both (a) and (b). We define $J$ as the tunnelling in
150: the primary lattice. The thin line is the ground state with only
151: the primary lattice, showing the total length of the system in the
152: harmonic trap. The localized states may have a different center
153: from the harmonic trap depending on the disorder realization. The
154: last graph in each column is an enlargement of the actual
155: potentials. The quasi-periodic system mimics true disorder to a
156: certain extent, showing localized states characterized by an
157: exponential decay in the envelope of the density moving away from
158: the localization center, $|\Psi(x)|^2 \propto \exp(-|x-x_0|/l)$
159: where $l$ is the localization length. The exponential localization
160: occurs only above a threshold level at approximately $\Delta/J
161: \simeq 6$ in the bichromatic case. This is in clear contrast with
162: random disorder where in a one-dimensional infinite system the
163: localization behavior persists for any infinitesimal amount of
164: disorder.
165:
166:
167:
168: %\textbf{Anderson Glass}
169:
170: In the presence of weak interactions one expects localization
171: effects to persist. In fact an Anderson glass phase has been
172: identified when increasing interactions from an Anderson localized
173: phase to the Bose-glass state, using the superfluid density as the
174: order parameter for the phase transition. Unlike the Bose-glass
175: phase, where both interactions and disorder cooperate to block
176: diffusion, in the Anderson glass phase the interactions tend to
177: delocalize the atoms and the disorder must compete against the
178: interactions \cite{Scalettar91,logan}.
179:
180: \begin{figure}[t!]
181: \begin{center}
182: \includegraphics[width=0.8\columnwidth]{fig2.eps}
183: \end{center}
184: \caption{Density profile in log scale of the ground state with
185: interactions for (a)-(c) increasing atom number and fixed
186: $\Delta/J = 6.5$, and for (d)-(f) increasing $\Delta/J$ and fixed
187: atom number $N = 1.5 \times 10^4$. In all cases the height of the
188: primary lattice is $s_1=10$ giving a tunnelling energy of $J/h=75$
189: Hz.} \label{AG}
190: \end{figure}
191:
192: Such behavior can be seen in Fig. 2, showing the ground states in
193: our incommensurate bichromatic lattice with interactions
194: calculated by means of a 1D effective Gross-Pitaevskii equation,
195: namely the nonpolynomial Schr\"{o}dinger equation (NPSE)
196: \cite{NPSE}.
197:
198: In the presence of interactions the strongly localized ground
199: state of (a) transforms into a state with multiple peaks with
200: partially overlapping tails (b). Upon increasing the interactions,
201: the overlap between these peaks increases until the state
202: eventually becomes extended (c). These results show similar
203: behavior to previous simulations of a BEC in a three-color lattice
204: \cite{Schulte}. Even with the increased interactions, a state with
205: multiple peaks can be recovered by increasing disorder (d)-(f).
206: The cross-over behavior between an extended superfluid state and a
207: possible Anderson glass state is very difficult to quantify in
208: this picture. We take a pragmatic approach and characterize the
209: state of the system by investigating the transport properties.
210:
211:
212: %\textbf{Dipole Oscillations}
213:
214: In the experiment, to set in motion our harmonically-trapped
215: condensate we excite dipole oscillations by abruptly shifting the
216: center of the magnetic trap. With a single color lattice, the
217: superfluid BEC oscillates freely, with the dipole frequency
218: modified by the effective mass \cite{note1,Cataliotti}. Instead,
219: adding an incommensurate bichromatic lattice, we expect the
220: oscillations to be blocked by localization effects. Fig. 3(a)
221: shows the center of mass motion after a shift in the magnetic trap
222: of 6 $\mu$m, with a fixed number of atoms $N = 1.5 \times 10^4$, a
223: fixed height of the primary lattice $s_1 = 10$, and a variable
224: height of the secondary lattice. At $s_2 = 0.1$ ($\Delta/J=2.6$),
225: the BEC oscillates with some damping. Increasing $s_2$ the motion
226: is strongly damped, until at $s_2 = 0.5$ ($\Delta/J=13$), the BEC
227: does not move. Interestingly, the height of the secondary lattice
228: at which the condensate becomes localized is slightly greater than
229: the non-interacting threshold for localization in the bichromatic
230: lattice shown in Fig 1. The increase of this threshold could be
231: due to the screening effect of interactions shown in Fig. 2.
232:
233: We extended the investigations to a variable interaction energy
234: effectuated by changing the number of atoms in the condensate.
235: Fig. 3(b) shows the measured center of mass motion of the atoms
236: for two different atom numbers $N=1.5 \times 10^4$ and $N=2 \times
237: 10^5$, after an initial harmonic trap shift of 6 $\mu$m, $s_1=10$
238: and $s_2=0.5$ ($\Delta/J=13$). Transport is stopped only for the
239: low number of atoms, although damping is seen with the higher atom
240: number.
241:
242: \begin{figure}[t!]
243: \begin{center}
244: \includegraphics[width=0.95\columnwidth]{fig3.eps}
245: \end{center}
246: \caption{(a) Measured dipole oscillations with increasing
247: intensity of the secondary lattice. The height of the primary
248: lattice and atom number are fixed at $s_1=10$ and $N=1.5 \times
249: 10^4$ respectively. (b) Measured dipole oscillations with fixed
250: $s_1=10$, $\Delta/J=13$ and two different atom number, $N=1.5
251: \times 10^4$ and $N=2 \times 10^5$. The center of mass motion of
252: the atoms was measured after 20 ms of expansion, fitting to the
253: central peak. The dotted lines show damped sinusoidals fitted to
254: the data with the frequency fixed.} \label{blocked}
255: \end{figure}
256:
257: %\textbf{Dynamical Instability and Bragg reflections}
258:
259: Localization due to disorder is not the only physical effect which
260: can block the motion of the atoms. In the case of an
261: incommensurate bichromatic lattice the Bloch theorem cannot be
262: applied, nevertheless the energy spectrum still shows the
263: existence of energy bands. The band structure for the single
264: lattice is complicated by the emergence of 'mini-gaps' opening up
265: almost everywhere across the spectrum \cite{quasiperiodic}.
266: However the dominant modification to the first Brillouin zone of
267: the primary lattice, for a weak addition of the secondary lattice,
268: are the extra energy gaps at $k_{b}$ and $k_1 - k_b$, where
269: $k_1=2\pi/\lambda_1$ gives the boundary of the first Brillouin
270: zone of the primary lattice, and $k_b=2\pi/\lambda_b$ corresponds
271: to the quasi-periodicity introduced on the larger length scale
272: $\lambda_{b}=\lambda_{1}\lambda_{2}/(\lambda_{2}-\lambda_{1})=4.38\lambda_1$
273: from the beating between the two colors. This simplification to
274: the energy spectrum is particularly true when interactions are
275: introduced, since they can effectively screen the potential
276: varying on length scales larger than the healing length, washing
277: away the energy gaps at smaller $k$. As a consequence, in the
278: presence of interactions one should carefully investigate the
279: possible contribution of dynamical instability that in a single
280: lattice has been observed to block dipole oscillations
281: \cite{Cataliotti, Fallani2004} when the quasi-momentum becomes
282: greater than $\simeq 0.5 k_1$. In the bichromatic lattice
283: dynamical instability may occur at a small quasi-momentum $\gtrsim
284: 0.5 k_{b}$ corresponding to the beat periodicity, much smaller
285: than the onset of instability with only the primary lattice.
286:
287: \begin{figure}[t!]
288: \begin{center}
289: \includegraphics[width=0.95\columnwidth]{fig4.eps}
290: \end{center}
291: \caption{(a) The center of mass motion from NPSE simulations of
292: the dipole oscillations with $s_1=10$, $\Delta/J=13$ and two
293: different $N$. The dotted line shows the expected superfluid
294: oscillation in a one color lattice with $s_1 = 10$. (b) The
295: corresponding momentum spectra in log scale at different times for
296: $N=1.5\times 10^4$. The vertical arrows show the time at which
297: each momentum spectrum is taken. (c) The momentum spectra in log
298: scale at different times for $N=2\times 10^5$.}\label{NPSE}
299: \end{figure}
300:
301: To better understand the contribution of the various physical
302: effects instigating localization, the dipole oscillations in our
303: actual system were simulated using the time-dependent NPSE. Fig.
304: 4(a) shows the center of mass motion using the same parameters as
305: the measurements taken in Fig. 3(b); an initial magnetic trap
306: shift of 6 $\mu$m, $s_1=10$, $s_2=0.5$, and for both $N=1.5 \times
307: 10^4$ and $N=2.0 \times 10^5$. The strongly damped center of mass
308: motion with low atom number and the increased movement with high
309: atom number seen in the experimental measurements is confirmed by
310: the solutions of the NPSE.
311:
312: From the momentum spectrum it is possible to distinguish when
313: dynamical instability is present. Fig. 4(b) shows the progression
314: of the momentum spectra with $N=1.5 \times 10^4$, and Fig. 4(c)
315: shows the momentum spectra with $N=2 \times 10^5$. At $t=25$ ms,
316: the main components of the momentum spectra correspond to the
317: peaks of the primary lattice at integer multiples of $\pm 2 \hbar
318: k_1$, and peaks of the beat periodicity at integer multiples of
319: $\pm 2 \hbar k_b$ around the primary peaks. For both high and low
320: atom number, at $t = 50$ ms extra momentum components begin to
321: rapidly grow signifying the onset of dynamical instability
322: \cite{Fallani2004}. We observe that the instability occurs when
323: the quasimomentum becomes greater than $0.5 k_{b}$. By 75 ms there
324: is a marked difference between the cases of high and low atom
325: number. With $N=1.5 \times 10^4$ atoms, the initial spectrum has
326: been obscured by the additional momentum components. In contrast,
327: with $N=2 \times 10^5$ atoms, the momentum spectrum retains much
328: of the original spectrum structure. Intriguingly, the increased
329: non-linearities inhibit the growth of the instability.
330:
331: Augmenting the interactions reduces the healing length of the
332: condensate. In previous experiments, with a single lattice of
333: spacing 0.4 $\mu$m and typically a healing length of $\simeq 0.3$
334: $\mu$m, the lattice spacing is not significantly greater than the
335: healing length, and therefore measurements are largely indifferent
336: to the atom number \cite{Fallani2004,Cataliotti}. However, in the
337: case of the bichromatic lattice the large spacing of the beating
338: (1.8$\mu$m) allows interactions to effectively smooth over the
339: large scale beat periodicity. The growth of dynamical instability
340: in our bichromatic lattice is governed by the competition between
341: augmentation by increased non-linearity and diminution by
342: screening of the beat periodicity \cite{note2}.
343:
344: Consideration of the center of mass motion shown in Fig. 4(a)
345: together with the momentum spectrum helps to unravel the different
346: contributions to localization. The contaminated momenta at 75 ms
347: for low atom number, shown in Fig. 4(b), is reflected in the
348: complete blockage of the center of mass motion. However, most
349: importantly, in the absence of dynamical instability, which is
350: macroscopically observed only after 50 ms, the movement of the
351: atoms is strongly damped with respect to the superfluid case (see
352: dotted line in Fig. 4(a)), suggesting that strong damping of the
353: oscillations is not only due to dynamical instability originating
354: from the beat periodicity of the bichromatic lattice. The presence
355: of interactions and the possible screening of the minigaps at
356: small $k$ from the incommensurate bichromatic lattice renders
357: interpretation of the blocked motion at short times difficult. As
358: already pointed out, with sufficiently strong interactions the
359: energy spectrum could be dominated by the periodicity at
360: $\lambda_b$, and correspondingly the strongly damped motion at
361: short times could be described by a high effective mass. Similarly
362: to the screening behavior seen in Fig. 2, the contribution of the
363: disorder will depend on the relative strength of interactions and
364: the intensity of the disordering potential.
365:
366:
367: In conclusion, maintaining a minimal number of atoms we have
368: observed a transition from oscillations to blocked motion with
369: increasing intensity of the incommensurate bichromatic lattice.
370: Simulations of our experimental parameters using the NPSE show
371: that the quasi-order inherent in the quasi-periodic bichromatic
372: lattice leads to the onset of dynamical instability, that
373: contributes to the blocked motion only after a critical time.
374: Screening of both localization due to the disorder and dynamical
375: instability due to the beat periodicity was observed with
376: strengthened interactions in the simulations. Increasing the
377: number of atoms in the experiment, we observed a return of
378: oscillating motion.
379:
380: This work shows that the exact choice of parameters is crucial to
381: separate and isolate the effect of disorder-induced localization,
382: or the non-trivial onset of dynamical instability in a bichromatic
383: lattice. In future measurements one could focus on measurements at
384: small quasimomentum to explore the transport behavior without the
385: complication of dynamical instability, however in an interacting
386: system the contribution of the disorder to blocked center of mass
387: motion cannot be easily distinguished from small amplitude
388: oscillations resulting from an initial high effective mass when
389: the energy spectrum is dominated by the beat periodicity. In the
390: non-interacting limit, the bichromatic lattice is also a very
391: promising tool to investigate Anderson-like localization that
392: could be accessed utilizing fermions or Feshbach resonances, which
393: also provide the important possibility of tuning the interactions.
394:
395: This work has been funded by the EU Contracts No.
396: HPRN-CT-2000-00125, IST-NoE-Phoremost, MIUR FIRB 2001, MIUR PRIN
397: 2005 and Ente Cassa di Risparmio di Firenze. We thank G. Modugno
398: for useful discussions and all the LENS Quantum Gases group.
399:
400: %We would like to acknowledge G. Modugno for critical reading of a
401: %reliminary version of the manuscript.
402:
403: \begin{thebibliography}{99}
404:
405: %\bibitem[*]{mail}
406: %Electronic address: \verb"lye@lens.unifi.it"
407:
408: \bibitem{Damski}
409: B. Damski \textit{et al.}, Phys. Rev. Lett. \textbf{91}, 080403
410: (2003);
411:
412: %B. Damski, J. Zakrzewski, L. Santos, P. Zoller, and M. Lewenstein,
413:
414: \bibitem{RothandBurnett}
415: R. Roth and K. Burnett, Phys. Rev. A \textbf{68}, 023604 (2003).
416:
417: \bibitem{Sanchez}
418: L. Sanchez-Palencia and L. Santos, Phys. Rev. A. \textbf{72}
419: 053607 (2005).
420:
421: \bibitem{Scarola}
422: V. W. Scarola and S. Das Sarma, Phys. Rev. A. \textbf{73}
423: 041609(R).
424:
425: \bibitem{quasiperiodic}
426: R. B. Diener \textit{et al.}, Phys. Rev. A. \textbf{64} 033416
427: (2001);
428:
429: \bibitem{Schulte}
430: T. Schulte \textit{et al.}, Phys. Rev. Lett. \textbf{95}, 170411
431: (2005). T. Schulte \textit{et al.}, arXiv:cond-mat/0609774 (2006).
432:
433:
434: \bibitem{Lye}
435: J. E. Lye \textit{et al.}, Phys. Rev. Lett. \textbf{95}, 070401
436: (2005).
437: %J. E. Lye, L. Fallani, M. Modugno, D. S. Wiersma, C. Fort, and M.
438: %Inguscio
439:
440: \bibitem{Clement}
441: D. Cl\'{e}ment \textit{et al.}, Phys. Rev. Lett. \textbf{95},
442: 170409 (2005).
443:
444: \bibitem{Fort}
445: C. Fort \textit{et al.}, Phys. Rev. Lett. \textbf{95}, 170410
446: (2005).
447: %C. Fort, L.Fallani, V. Guarrera, J. E. Lye, M. Modugno, D. S.
448: %Wiersma and M. Inguscio
449:
450: \bibitem{MIone}
451: M. Greiner \textit{et al.}, Nature \textbf{415}, 39 (2002).
452: %M. Greiner, O. Mandel, T. Esslinger, T. W. H¨ansch, and I. Bloch,
453:
454: \bibitem{boseglass}
455: L. Fallani \textit{et al.}, arXiv:cond-mat/0603655 (2006)
456: %L. Fallani, J. E. Lye, V. Guarrera, C. Fort, M. Inguscio
457:
458: \bibitem{logan}
459: P. Lugan \textit{et al.}, arXiv:cond-mat/0610389 (2006)
460:
461: \bibitem{Fallani2004}
462: L. Fallani \textit{et al.}, Phys. Rev. Lett. \textbf{93} 140406
463: (2004).
464: %L. Fallani, L. De Sarlo, J. E. Lye, M. Modugno, R. Saers, C. Fort,
465: %and M. Inguscio
466:
467: \bibitem{Anderson58}
468: P. W. Anderson, Phys. Rev. \textbf{109}, 1492 (1958).
469:
470: \bibitem{Scalettar91}
471: R. T. Scalettar, G. G. Batrouni, and G. T. Zimanyi, Phys. Rev.
472: Lett. \textbf{66}, 3144 (1991).
473:
474: \bibitem{NPSE}
475: L. Salasnich, Laser Phys. \textbf{12}, 198 (2002); L. Salasnich,
476: A. Parola, and L. Reatto, Phys. Rev. A \textbf{65}, 043614 (2002).
477:
478: \bibitem{note1}
479: This was confirmed in the experiment for $s_1=10$ and a trap shift
480: of 6 $\mu$m.
481:
482: \bibitem{Cataliotti}
483: F. S. Cataliotti \textit{et al.}, New J. Phys. \textbf{5} 71
484: (2003)
485:
486: \bibitem{note2}
487: A detailed analysis will be presented in a future theoretical
488: work.
489:
490:
491:
492:
493:
494: \end{thebibliography}
495:
496: \end{document}
497: