cond-mat0506766/ams.tex
1: \documentclass[12pt]{iopart}
2: \usepackage[dvips]{graphicx}
3: %\usepackage{amsmath}
4: %\usepackage{times}
5: %\usepackage{psfrag}
6: 
7: 
8: \begin{document}
9: 
10: \title[Quantum lattice solitons]
11: {Quantum lattice solitons in ultracold bosons near the Feshbach resonance}
12: 
13: \author{K V Krutitsky$^1$ and D V Skryabin$^2$}
14: 
15: \address
16: {$^1$
17: Fachbereich Physik der Universit\"at Duisburg-Essen, Campus Essen,
18: Universit\"atsstr.~5, 45117 Essen, Germany
19: }
20: \address
21: {$^2$
22: Department of Physics, University of Bath, Bath, BA2~7AY, UK
23: }
24: \ead{kostya@theo-phys.uni-essen.de}
25: 
26: \begin{abstract}
27: Quantum lattice solitons in a system of two
28: ultracold bosons near Feshbach resonance are investigated.
29: It is shown that their binding energy,
30: effective mass, and spatial width, can be manipulated varying the detuning from
31: the Feshbach resonance.
32: In the case of attractive atomic interactions, the molecule creation stabilizes
33: the solitons.
34: In the case of repulsive interactions, the molecule creation leads
35: to the possibility of existence of bright solitons in some interval of detunings.
36: Due to quantum fluctuations the distance between the atoms is a random quantity
37: with the standard deviation larger than the mean value.
38: \end{abstract}
39: 
40: %\pacs{03.75.Lm,63.20.Pw,71.35.Lk}
41: 
42: %\submitto{\JPB}
43: 
44: \maketitle
45: 
46: %-----------------------------------------------------
47: \section{Introduction}
48: %-----------------------------------------------------
49: 
50: Many-body phenomena in ultracold quantum gases is a subject of extensive
51: ongoing research. Interaction between atoms plays a
52: crucial role in many situations and is responsible for the most
53: striking experimental observations of solitons in
54: Bose-Einstein condensates trapped by harmonic~\cite{becsol} and periodic~\cite{markus}
55: potentials. The typical number of atoms in such solitons
56: varies from several hundreds to tens of thousands.
57: Therefore these structures can be well described by the mean-field Gross-Pitaevsky
58: equation. However, the mean-field theory does never provide an exact description
59: of interacting quantum systems (e.g., due to unavoidable depletion) and it becomes
60: interesting to investigate quantum effects in the soliton propagation~\cite{bth}.
61: If the atoms are loaded into the optical lattice, the interaction effects become more
62: important and in the limit of the small number of atoms an exact quantum analysis 
63: of the solitons~\cite{Scott,el}
64: reveals strong deviations from the results provided by the
65: Gross-Pitaevsky equation with lattice potential~\cite{skryabin} or
66: its discrete version~\cite{Scott,discr}.
67: 
68: The use of Feshbach resonances to control interaction between
69: ultracold atoms in optical potentials is a widely spread technique allowing
70: transformation of atoms into molecules and changing  magnitude and
71: sign of the effective scattering length of the atoms (see, e.g.,~\cite{MVA,fesh}).
72: Coherent solitons in condensed atomic-molecular mixtures
73: without optical lattice were studied in several papers, see,
74: e.g.,~\cite{par1,par2}. The mathematical model previously used for the coupled
75: atomic-molecular condensates is equivalent to the model describing
76: parametric interaction of photons in quadratically nonlinear
77: crystals~\cite{rev}. It was demonstrated in the mean-field
78: limit that the resonant atomic-molecular interaction serves as a
79: mechanism responsible for supporting bright solitons in the case
80: of repulsive bosons and for preventing collapse in the case of
81: attractive bosons~\cite{par2,rev}. The quantum atomic-molecular
82: solitons in the system without periodic potential were also
83: studied~\cite{par1,quant}.
84: Optical parametric solitons in
85: the system of coupled waveguides, playing the role of a periodic potential
86: for photons, were recently observed experimentally~\cite{steg}.
87: Atomic-molecular solitons in a
88: deep optical lattice have been theoretically considered in the
89: mean-field approximation~\cite{konotop}, which is mathematically equivalent
90: to the system studied in~\cite{steg}.
91: 
92: In this work, we demonstrate existence and study
93: properties of the quantum atomic-molecular solitons in an optical
94: lattice near Feshbach resonance.
95: Under the quantum lattice soliton we understand the quantum state of the system
96: of interacting particles with the localized eigenfunction and the discrete
97: energy level belonging to a spectral interval forbidden for the spatially extended
98: periodic states~\cite{Scott}.
99: Note that the discrete energy levels belonging to the intervals forbidden
100: for the linear waves are also a generic feature of the classical lattice solitons.
101: Advances in manipulation of ultracold atomic systems with small number of particles per
102: lattice site~\cite{Greiner} as well as in cooling and trapping of single
103: atoms~\cite{single} allow one to hope that quantum lattice
104: solitons will soon become relevant for experimental research.
105: 
106: %-----------------------------------------------------
107: \section{Hamiltonian}
108: %-----------------------------------------------------
109: 
110: We consider two atoms of mass $m$ in an optical lattice created by a
111: far-detuned standing laser wave. If the laser wavelength is $\lambda_{\rm L}=2\pi/k_{\rm L}$,
112: then the lattice constant $d=\lambda_{\rm L}/2$. It is convenient
113: to represent the amplitude of the periodic potential in the form $\hbar\omega_{\rm R} s$,
114: where $\omega_{\rm R}=\hbar k_{\rm L}^2/(2 m)$ is the recoil frequency and
115: $s$ is a dimensionless parameter.
116: In the case of a deep optical lattice every lattice site can be described by a harmonic potential
117: with the frequency $\omega=2\omega_{\rm R}\sqrt{s}$
118: and the lowest-band atomic Wannier function
119: is well approximated by a Gaussian with the characteristic length
120: $l_{\rm a}=\sqrt{\hbar/m \omega}$.
121: The atoms in the lattice are subject to the magnetic field $B$,
122: with $B=B_0$ corresponding to the Feshbach resonance of the width $\Delta B$.
123: 
124: There are several processes which are to be
125: taken into account in such a system: atomic interaction, molecule production and
126: atomic and molecular hopping.
127: Taking into account only the hopping between the nearest lattice sites
128: as well as on-site atomic interactions and in the lowest-band approximation
129: the Hamiltonian of the system is given by~\cite{DKOS,comment}
130: \begin{eqnarray}
131: \label{bh}
132: H
133: &=&
134: -
135: t_{\rm a}
136: \sum_{\langle i,j \rangle}
137: a^{\dagger}_{i}
138: a^{\phantom \dagger}_{j}
139: -
140: t_{\rm m}
141: \sum_{\langle i,j \rangle}
142: b^{\dagger}_{i}
143: b^{\phantom \dagger}_{j}
144: +
145: \left(
146:     \delta-\frac{3}{2}\hbar\omega
147: \right)
148: \sum_{i}
149: b^{\dagger}_{i}
150: b^{\phantom \dagger}_{i}
151: +
152: \frac{U_{\rm bg}}{2}
153: \sum_{i}
154: a^{\dagger}_{i}
155: a^{\dagger}_{i}
156: a^{\phantom \dagger}_{i}
157: a^{\phantom \dagger}_{i}
158: \nonumber\\
159: &+&
160: \tilde g
161: \sum_{i}
162: \left(
163: b^{\dagger}_{i}
164: a^{\phantom \dagger}_{i}
165: a^{\phantom \dagger}_{i}
166: +
167: a^{\dagger}_{i}
168: a^{\dagger}_{i}
169: b^{\phantom \dagger}_{i}
170: \right)
171: \;,
172: \end{eqnarray}
173: where
174: $a^{\dagger}_{i}$~($b^{\dagger}_{i}$)
175: and
176: $a^{\phantom \dagger}_{i}$~($b^{\phantom \dagger}_{i}$)
177: are creation and annihilation operators of a single atom (molecule) at a lattice site $i$,
178: $\delta=\Delta\mu(B-B_0)$ is a detuning from the Feshbach resonance.
179: Here, $\Delta\mu$ is the difference in magnetic moments of the two atoms and a molecule.
180: The atom-molecule conversion is determined by
181: $
182: \tilde g
183: =
184: \hbar
185: \sqrt{2 \pi a_{\rm bg}
186: \Delta B \Delta \mu /m}
187: /(2 \pi l_{\rm a}^2)^{3/4}
188: $
189: and the background on-site atomic interaction parameter is
190: $
191: U_{\rm bg}
192: =
193: \sqrt{2/\pi}
194: \hbar \omega
195: \left(
196:     a_{\rm bg} / l_{\rm a}
197: \right)
198: $
199: with $a_{\rm bg}$ being the background scattering length.
200: In the Gaussian approximation, the atomic and molecular tunneling matrix elements are given by
201: %\begin{equation}
202: $
203: t_{\rm a,m}
204: =
205: \frac{\hbar \omega}{2}
206: \left[
207:     1 - \left( \frac{2}{\pi} \right)^{2}
208: \right]
209: \left(
210:     \frac{\lambda_{\rm L}}{4 l_{\rm a,m}}
211: \right)^{2}
212: e^{
213:     -
214:     \left(
215:         \lambda_{\rm L}/4 l_{\rm a,m}
216:     \right)^{2}
217:   }
218: $.
219: %\end{equation}
220: Since
221: $l_{\rm m} = l_{\rm a}/\sqrt{2}$,
222: the molecular tunneling rate is much smaller than the atomic one.
223: 
224: %-----------------------------------------------------
225: \section{Solution of the on-site problem}
226: %-----------------------------------------------------
227: 
228: The on-site problem for the Hamiltonian (\ref{bh}) can be easily solved
229: analitically. In the case when the atoms are on the same lattice site
230: there are two eigenmodes which are superpositions of the two-atom and
231: molecular states with the energies
232: \begin{equation}
233: \label{e}
234: E_\pm
235: =
236: \frac
237: {\delta'+U_{\rm bg}}
238: {2}
239: \pm
240: \sqrt
241: {
242:   \left(
243:       \frac
244:       {\delta'-U_{\rm bg}}
245:       {2}
246:   \right)^2
247:   +
248:   2
249:   \tilde g^2
250: }
251: \;,
252: \end{equation}
253: and the probability to find a molecule
254: \begin{equation}
255: \label{pm}
256: p_{{\rm m}\pm}
257: =
258: \frac{1}{2}
259: \left[
260:     1
261:     \pm
262:     \frac
263:     {\delta'-U_{\rm bg}}
264:     {
265:       \sqrt
266:       {
267:        \left(
268:            \delta'-U_{\rm bg}
269:        \right)^2
270:        +
271:        8
272:        \tilde g^2
273:       }
274:     }
275: \right]
276: \;,
277: \end{equation}
278: where
279: $\delta'=\delta-\frac{3}{2}\hbar\omega$
280: is an effective detuning.
281: 
282: The two-atoms on-site problem was exactly solved in Ref.~\cite{DKOS} for the infinite number of bands
283: neglecting the atom-atom interaction.
284: The eigenenergies $E$ are shown to be determined by the equation
285: \begin{equation}
286: \label{emo}
287: E - \delta'
288: =
289: \frac{2\sqrt{\pi}\tilde g^2}{\hbar \omega}
290: \frac
291: {\Gamma (- E / 2 \hbar \omega)}
292: {\Gamma (- E / 2 \hbar \omega - 1/2)}
293: \;.
294: \end{equation}
295: The eigenenergies given by Eq.~(\ref{e}) for $U_{\rm bg}=0$ and Eq.~(\ref{emo})
296: are plotted in Fig.~\ref{ee}. As we see, our lower-branch solution $E_-$ in Eq.~(\ref{e})
297: is in excellent agreement with the corresponding branch of Eq.~(\ref{emo}) for arbitrary $\delta$.
298: The upper-branch solution $E_+$ fails to reproduce the second branch of Eq.~(\ref{emo})
299: if $\delta$ is far above the Feshbach resonance where the contribution of the second band
300: becomes significant, remaning however in a very good agreement near the resonance
301: and below it. This implies
302: that the lowest-band approximation is valid if the effective detuning
303: $\delta'$ is less than
304: the gap between the two lowest Bloch bands, which is the quantity of the order of $\hbar\omega$,
305: and/or if we are interested in the eigenmodes of the Hamiltonian (\ref{bh})
306: with the energies less than the energy of the second Bloch band.
307: The latter is always the case in the present work.
308: In addition, the parameters $U_{\rm bg}$ and $\tilde g$ must be much smaller than
309: the bands separation which is also fulfilled.
310: 
311: %------------------------------------------------------------------------
312: \begin{figure}[tb]
313: \centering
314: 
315: %psfrag{e}[b]{$E/\hbar\omega$}
316: %psfrag{d}[c]{$\delta'/\hbar\omega$}
317: 
318: % \includegraphics[width=5cm]{eigenenergies.eps}
319: \hspace{-3cm}
320:   \includegraphics[width=8cm]{figure1.eps}
321: 
322: \caption{Eigenenergies in the case of two atoms on the same lattice site.
323:          Solid lines show the results given by Eq.~(\ref{e}) which corespond
324: 	 to the lowest-band approximation. The results for the infinite number of bands [Eq.~(\ref{emo})]
325: 	 are shown by dashed lines.
326: 	 $U_{\rm bg}=0$,
327: 	 $2 \sqrt{\pi} \tilde g^2 / \left(\hbar\omega\right)^2=0.1$.
328: 	}
329: \label{ee}
330: \end{figure}
331: %------------------------------------------------------------------------
332: 
333: %-----------------------------------------------------
334: \section{Eigenmodes of the complete Hamiltonian and the soliton band}
335: %-----------------------------------------------------
336: 
337: We consider a one-dimensional model with $L$ lattice sites
338: and assume that $L$ is odd\footnote{In the case of even $L$ there will be only unessenstial modifications
339: in the equations.}.
340: Under periodic boundary conditions the eigenstates of the Hamiltonian~(\ref{bh}) are
341: \begin{eqnarray}
342: \label{psi}
343: |\psi_k\rangle
344: &=&
345: c^{\rm m}_{k}
346: \sum_{j=1}^L
347: \left(
348:     \hat T/\tau_k
349: \right)^{j-1}
350: |1_{\rm m} 0 \dots 0\rangle
351: +
352: c^{\rm a}_{0k}
353: \sum_{j=1}^L
354: \left(
355:     \hat T/\tau_k
356: \right)^{j-1}
357: |2 0 \dots 0\rangle
358: \nonumber\\
359: &+&
360: c^{\rm a}_{1k}
361: \sum_{j=1}^L
362: \left(
363:     \hat T/\tau_k
364: \right)^{j-1}
365: |1 1 0 \dots 0\rangle
366: +
367: \dots
368: \nonumber\\
369: &+&
370: c^{\rm a}_{(L-1)/2,k}
371: \sum_{j=1}^L
372: \left(
373:     \hat T/\tau_k
374: \right)^{j-1}
375: |1 0 \dots 0 1 0 \dots 0\rangle
376: \;,
377: \end{eqnarray}
378: where $|1_{\rm m} 0 \dots 0\rangle$ is a state with one molecule
379: on the first lattice site and all the other sites being unoccupied,
380: $|n_1 \dots n_L\rangle$ is a state with $n_i$ atoms on site $i$, $i=1,\dots,L$.
381: $\hat T$ is the translation operator which has the eigenvalues
382: $
383:  \tau_k
384:  =
385:  \exp
386:  \left(
387:      i \pi k/k_{\rm L}
388:  \right)
389: $
390: with the wave number
391: $k=k_{\rm L} 2 \nu/L$, $\nu=0,\pm 1,\dots,\pm (L-1)/2$~\cite{Scott}.
392: The eigenvalue problem for the Hamiltonian~(\ref{bh}) can be written down in the matrix form
393: \begin{equation}
394: \label{evp}
395: \left(
396:     \begin{array}{cc}
397:        \epsilon_k^{\rm m} & A^T \\
398:        A & Q_k
399:     \end{array}
400: \right)
401: \left(
402:     \begin{array}{c}
403:        {c}_k^{\rm m} \\
404:        {\bf c}_k^{\rm a}
405:     \end{array}
406: \right)
407: =
408: E_k
409: \left(
410:     \begin{array}{c}
411:        {c}_k^{\rm m} \\
412:        {\bf c}_k^{\rm a}
413:     \end{array}
414: \right)
415: \;,
416: \end{equation}
417: where
418: $
419:  \epsilon_{k}^{\rm m}
420:  =
421:  \delta'
422:  -
423:  2 t_{\rm m}
424:  \cos
425:  \left(
426:      \pi k/k_{\rm L}
427:  \right)
428: $.
429: The vector $A$ has a length $(L+1)/2$ and its nonvanishing
430: element is $A_{1}=\sqrt{2}\tilde g$.
431: The nonvanishing elements of the tridiagonal $(L+1)/2 \times (L+1)/2$ matrix $Q_k$
432: are given by~\cite{Scott}
433: \begin{eqnarray}
434: &&
435: Q_{11}=U_{\rm bg}
436: \;,\;
437: Q_{21} = Q_{12}^* = - t_{\rm a} \sqrt{2} (1+\tau_k)
438: \;,
439: \\
440: &&
441: Q_{i+1,i} = Q_{i,i+1}^* = - t_{\rm a} (1+\tau_k)
442: \;,\;
443: i=2,\dots,(L-1)/2
444: \;,
445: \nonumber\\
446: &&
447: Q_{(L+1)/2,(L+1)/2}
448: =
449: - t_{\rm a}
450: \left[
451:     \tau_k^{(L+1)/2}
452:     +
453:     \tau_k^{(L-1)/2}
454: \right]
455: \;.
456: \nonumber
457: \end{eqnarray}
458: The eigenvectors in Eq.~(\ref{evp}) consist of two parts
459: ${c}_k^{\rm m}$,
460: $
461: {\bf c}_k^{\rm a}
462: =
463: {\rm col}
464: \left[
465:     c^{\rm a}_{0k},\dots,c^{\rm a}_{(L-1)/2,k}
466: \right]
467: $,
468: and satisfy the normalization condition
469: \begin{eqnarray}
470: \label{norm}
471: \left|
472:     c_{k}^{\rm m}
473: \right|^2
474: +
475: \sum_{i=0}^{(L-1)/2}
476: \left|
477:     c_{ik}^{\rm a}
478: \right|^2
479: &=&
480: 1
481: \;.
482: \end{eqnarray}
483: 
484: In the absence of the molecular mode, the eigenvalue problem~(\ref{evp}) reduces
485: to that one solved in Ref.~\cite{Scott}, where it was shown that in the case of attractive
486: interaction the energy spectrum consists always of (quasi)continuum band
487: and a discrete level below the (quasi)continuum which corresponds to the bright soliton.
488: Its characteristic feature is that
489: $
490:   \left|
491:       c_{0k}^{\rm a}
492:   \right|^2
493:   \gg
494:   \left|
495:       c_{ik}^{\rm a}
496:   \right|^2
497: $,
498: $i=1,\dots,(L-1)/2$, i.e., the probability of finding two atoms on the same lattice site
499: is much higher than all the other ones.
500: This localization corresponds to the soliton solution of the discrete nonlinear
501: Schr\"odinger equation and, therefore, the discrete level can be called
502: a "soliton band"~\cite{Scott}. Our aim is to investigate
503: the influence of the molecular mode on the soliton band.
504: 
505: After the eigenvalue problem~(\ref{evp}) is solved, one can calculate
506: the soliton binding energy $E_{\rm b}$
507: which is defined as the difference of the energy at the bottom
508: of the (quasi)continuum and the soliton level at $\nu=0$ which corresponds to $k=k_0=0$.
509: The effective mass $m^*$ can be worked out using a quadratic approximation for the eigenenergy
510: at some small value of $\nu$ (e.g., $\nu=1$)
511: %\begin{eqnarray}
512: $
513: E_{k_1}
514: =
515: E_{k_0}
516: +
517: \hbar^2 k_1^2
518: /
519: \left(
520:     2 m^*
521: \right)
522: $,
523: %\;,
524: %\end{eqnarray}
525: which leads to
526: \begin{eqnarray}
527: m^*
528: &=&
529:  2 \hbar^2 k_{\rm L}^2
530:  /
531: \left[
532:  \left(
533:      E_{k_1}
534:      -
535:      E_{k_0}
536:  \right)
537:  L^2
538: \right]
539: \;.
540: \end{eqnarray}
541: According to Eq.~(\ref{psi}) the distance between the atoms $w_k$
542: is a random variable which takes the values
543: $w_{ki}=0,1,\dots,(L-1)/2$,
544: with the probabilities
545: $
546: \left|
547:      c_{ik}^{\rm a}
548: \right|^2
549: /
550: \left(
551:     1
552:     -
553:     \left|
554:         c_{k}^{\rm m}
555:     \right|^2
556: \right)
557: $.
558: Thus, it is necessary to calculate
559: not only the mean interatomic distance $\langle w_k \rangle$ but also its standard deviation
560: \begin{eqnarray}
561: \sigma_{wk}
562: &=&
563: \sqrt
564: {
565: \langle w_k^2 \rangle
566: -
567: \langle w_k \rangle^2
568: }
569: \;,
570: \langle w_k^l \rangle
571: =
572: \sum_{i=0}^{(L-1)/2}
573: \frac
574: {
575:  i^l
576:  \left|
577:      c_{ik}^{\rm a}
578:  \right|^2
579: }
580: {
581:  1
582:  -
583:  \left|
584:      c_{k}^{\rm m}
585:  \right|^2
586: }
587: \;,
588: \end{eqnarray}
589: and the soliton width can be defined as
590: $
591: \sqrt
592: {
593:  \langle w_k^2 \rangle
594: }
595: %=
596: %\sqrt
597: %{
598: % \sigma_{wk}^2
599: % +
600: % \langle w_k \rangle^2
601: %}
602: $.
603: 
604: We have solved the eigenvalue problem~(\ref{evp}) numerically for finite values of $L$
605: and analytically in the limit of infinite lattice.
606: The results are presented below.
607: We consider the cases of attractive and repulsive atomic interactions and
608: concentrate on the properties of the lower-energy modes.
609: 
610: %-----------------------------------------------------
611: \subsection{Attractive atomic interaction}
612: %-----------------------------------------------------
613: 
614: %------------------------------------------------------------------------
615: \begin{figure}[tb]
616: \centering
617: 
618: %\psfrag{E}[b]{$E_k/\hbar\omega$}
619: %\psfrag{v}[c]{$k/k_L$}
620: %\psfrag{s}[lb]{soliton band}
621: 
622: %  \includegraphics[width=5cm]{attr-d3.eps}
623: \hspace{-3cm}
624:   \includegraphics[width=8cm]{figure2.eps}
625: 
626: \caption{Energy eigenvalues of the Hamiltonian~(\ref{bh}).
627:          The parameters are $s=5$,
628:          $2 \sqrt{\pi} \tilde g^2 / \left(\hbar\omega\right)^2=0.1$,
629:          $\delta'/\hbar\omega=3$,
630: 	 $a_{\rm bg}/\lambda_L=-0.005$.
631: 	 Dots are the results of numerical solution of Eq.~(\ref{evp}) for $L=41$ and
632: 	 the solid lines correspond to the limit $L\to\infty$.
633: 	 The spectrum is truncated from above in order to be consistent
634: 	 with the lowest-band approximation.
635: 	}
636: \label{s}
637: \end{figure}
638: %------------------------------------------------------------------------
639: 
640: %------------------------------------------------------------------------
641: \begin{figure}[tb]
642: \centering
643: 
644: %\psfrag{p}[b]{probability}
645: %\psfrag{d}[c]{$\delta'/\hbar\omega$}
646: 
647: %\psfrag{m}[c]{\small m}
648: %\psfrag{z}[c]{\small $0$}
649: %\psfrag{o}[c]{\small $1$}
650: %\psfrag{t}[c]{\small $2$}
651: 
652: %  \includegraphics[width=5cm]{attr-prob.eps}
653: \hspace{-3cm}
654:   \includegraphics[width=8cm]{figure3.eps}
655: 
656: \caption{Probabilities of molecular and atomic states corresponding to the soliton band:
657:          $
658: 	  \left|
659: 	      c_{0}^{\rm m}
660: 	  \right|^2
661: 	 $ (m),
662: 	 $
663: 	  \left|
664: 	      c_{i0}^{\rm a}
665: 	  \right|^2
666: 	 $ ($i$),
667: 	 $i=0,1,2$
668: 	 [$
669: 	   \left|
670: 	       c_{20}^{\rm a}
671: 	   \right|^2
672: 	  $
673: 	  is shown by the dashed line].
674:          The parameters are the same as in Fig.~\ref{s}.
675: 	}
676: \label{pn}
677: \end{figure}
678: %------------------------------------------------------------------------
679: 
680: %------------------------------------------------------------------------
681: \begin{figure}[tb]
682: \centering
683: 
684: %\psfrag{e}[b]{$E_b/\hbar\omega$}
685: %\psfrag{d}[c]{$\delta'/\hbar\omega$}
686: 
687: %  \includegraphics[width=5cm]{attr-eb.eps}
688: 
689: %  \vspace{-3.1cm}
690: 
691: %  \hspace{1.4cm}\includegraphics[width=3.5cm]{attr-eb-inset.eps}
692: 
693: %  \vspace{0.8cm}
694: \hspace{-3cm}
695:   \includegraphics[width=8cm]{figure4.eps}
696: 
697: \caption{Soliton binding energy.
698:          The parameters are the same as in Fig.~\ref{s} and $k=0$.
699: 	}
700: \label{ebn}
701: \end{figure}
702: %------------------------------------------------------------------------
703: 
704: %------------------------------------------------------------------------
705: \begin{figure}[tb]
706: \centering
707: 
708: %\psfrag{m}[b]{$m^*/\tilde m$}
709: %\psfrag{d}[c]{$\delta'/\hbar\omega$}
710: 
711: %  \includegraphics[width=5cm]{attr-mass.eps}
712: 
713: %  \vspace{-3.1cm}
714: 
715: %  \hspace{1.4cm}\includegraphics[width=3.5cm]{attr-mass-inset.eps}
716: 
717: %  \vspace{0.8cm}
718: \hspace{-3cm}
719:   \includegraphics[width=8cm]{figure5.eps}
720: 
721: \caption{The ratio of the soliton effective mass $m^*$ to the effective mass
722:          $\tilde m$ at the bottom of (quasi)continuum band.
723:          The parameters are the same as in Fig.~\ref{s}.
724: 	}
725: \label{mn}
726: \end{figure}
727: %------------------------------------------------------------------------
728: 
729: %------------------------------------------------------------------------
730: \begin{figure}[tb]
731: \centering
732: 
733: %\psfrag{w}[cr]{\rotatebox{100}{$\langle w_k \rangle$}}
734: %\psfrag{k}[c]{\rotatebox{-17}{$k/k_L$}}
735: %\psfrag{d}[c]{\hspace{2mm}\rotatebox{65}{$\delta'/\hbar\omega$}}
736: %\psfrag{s}[cr]{\rotatebox{100}{$\phantom{\langle}\sigma_{wk}\phantom{\rangle}$}}
737: 
738: %  \includegraphics[width=3.5cm]{attr-width.eps}
739: %  \hspace{0.5cm}
740: %  \includegraphics[width=3.5cm]{attr-deviation.eps}
741: \hspace{-3cm}
742:   \includegraphics[width=10cm]{figure6.eps}
743: 
744: \caption{Mean interatomic distance $\langle w_k \rangle$ (left panel)
745:          and its standard deviation $\sigma_{wk}$ (right panel).
746:          The parameters are the same as in Fig.~\ref{s}.
747: 	}
748: \label{wdn}
749: \end{figure}
750: %------------------------------------------------------------------------
751: 
752: We consider first bosons with attractive interactions ($U_{\rm bg}<0$).
753: If $\delta'$ is negative and its absolute value is very large,
754: the coupling between the molecular
755: mode and the atomic mode is negligible and we have two discrete levels below the (quasi)continuum.
756: The lower one corresponds to the pure molecule and another one to the atomic bright soliton.
757: If $\delta'$ increases, i.e, we come closer to the Feshbach resonance,
758: both discrete levels approach the (quasi)continuum.
759: At some critical value of $\delta'=\delta'_-$
760: the upper level merges with the (quasi)continuum.
761: In the numerical calculations it is not quite clear how to determine $\delta'_-$
762: exactly because there are several possibilities to define it.
763: Analytical analysis in the case of the infinite lattice shows that the mergence occurs
764: if at least one of the inequalities
765: \begin{equation}
766: \label{ineq}
767: \left|
768:     c_{1k}^{\rm a}
769: \right|
770: >
771: \left|
772:     c_{ik}^{\rm a}
773: \right|
774: \;,
775: i=2,3,\dots,
776: \end{equation}
777: is violated. We adopt this as a definition of $\delta'_-$ and by doing numerical diagonalization
778: for $k=0$ and for the values of parameters in the caption of Fig.~\ref{s} we obtain
779: $\delta'_-=-1.531\,\hbar\omega$. In order to have inequalities (\ref{ineq}) again fulfilled,
780: one has to increase $\delta'$ up to $\delta'_+$. Using the same values of the parameters
781: we get $\delta'_+=-1.490\,\hbar\omega$. If $\delta'>\delta'_+$, a discrete level
782: appears above the (quasi)continuum, while the lower one
783: which becomes a linear combination of atomic and molecular states remains below (see Fig.~\ref{s}).
784: If we increase $\delta'$ further and go far away from the Feshbach resonance
785: ($\delta' \gg \hbar\omega$),
786: the contribution of the molecular mode into the lowest-energy eigenstate becomes negligible~(Fig.\ref{pn})
787: and we have a pure atomic bright soliton below the (quasi)continuum~\cite{Scott}.
788: The upper discrete level is located very far above the (quasi)continuum and cannot be interpreted
789: within the lowest-band approximation.
790: 
791: The soliton binding energy $E_{\rm b}$ is shown in Fig.~\ref{ebn}.
792: Due to the large contribution of the molecular mode
793: near the resonance the binding energy is larger than its asymptotic value at $\delta'\to\infty$.
794: The effective mass $m^*$ is also larger at smaller values of $\delta'$ (see Fig.~\ref{mn})
795: because due to the fact that $t_{\rm m} \ll t_{\rm a}$
796: the effective mass of the molecule is much larger than the atomic effective mass.
797: The corresponding contributions of the molecular and atomic states into the soliton band
798: are shown in Fig.~\ref{pn}.
799: 
800: The mean interatomic distance $\langle w_k \rangle$ as well as its standard deviation $\sigma_{wk}$
801: are shown in Fig.~\ref{wdn}. The interatomic distance is well below the lattice constant $d$
802: and the maximal localization is achieved at the edges of the Brillouin zone.
803: However, quantum fluctuations are very strong and $\sigma_{wk}>\langle w_k \rangle$.
804: 
805: %-----------------------------------------------------
806: \subsection{Repulsive atomic interaction}
807: %-----------------------------------------------------
808: 
809: %------------------------------------------------------------------------
810: \begin{figure}[tb]
811: \centering
812: 
813: %\psfrag{p}[b]{probability}
814: %\psfrag{d}[c]{$\delta'/\hbar\omega$}
815: 
816: %\psfrag{m}[c]{\small m}
817: %\psfrag{z}[c]{\small $0$}
818: %\psfrag{o}[c]{\small $1$}
819: %\psfrag{t}[c]{\small $2$}
820: 
821: %  \includegraphics[width=5cm]{repul-prob.eps}
822: \hspace{-3cm}
823:   \includegraphics[width=8cm]{figure7.eps}
824: 
825: \caption{Probabilities of molecular and atomic states corresponding to the soliton band:
826:          $
827: 	  \left|
828: 	      c_{0}^{\rm m}
829: 	  \right|^2
830: 	 $ (m),
831: 	 $
832: 	  \left|
833: 	      c_{i0}^{\rm a}
834: 	  \right|^2
835: 	 $ ($i$),
836: 	 $i=0,1,2$
837: 	 [$
838: 	   \left|
839: 	       c_{20}^{\rm a}
840: 	   \right|^2
841: 	  $
842: 	  is shown by the dashed line].
843: 	 $a_{\rm bg}/\lambda_L=0.005$ and the other
844:          parameters are the same as in Fig.~\ref{s}.
845: 	}
846: \label{pp}
847: \end{figure}
848: %------------------------------------------------------------------------
849: 
850: %------------------------------------------------------------------------
851: \begin{figure}[tb]
852: \centering
853: 
854: %\psfrag{m}[b]{$m^*/\tilde m$}
855: %\psfrag{d}[c]{$\delta'/\hbar\omega$}
856: 
857: %  \includegraphics[width=5cm]{repul-mass.eps}
858: \hspace{-3cm}
859:   \includegraphics[width=8cm]{figure8.eps}
860: 
861: \caption{The ratio of the soliton effective mass $m^*$ to the effective mass
862:          $\tilde m$ at the bottom of (quasi)continuum band.
863:          The parameters are the same as in Fig.~\ref{pp}.
864: 	}
865: \label{mp}
866: \end{figure}
867: %------------------------------------------------------------------------
868: 
869: %------------------------------------------------------------------------
870: \begin{figure}[tb]
871: \centering
872: 
873: %\psfrag{w}[cr]{\rotatebox{100}{$\langle w_k \rangle$}}
874: %\psfrag{k}[c]{\rotatebox{-17}{$k/k_L$}}
875: %\psfrag{d}[c]{\hspace{5mm}\rotatebox{65}{$\delta'/\hbar\omega$}}
876: %\psfrag{s}[cr]{\rotatebox{100}{$\phantom{\langle}\sigma_{wk}\phantom{\rangle}$}}
877: 
878: %  \includegraphics[width=3.5cm]{repul-width.eps}
879: %  \hspace{0.5cm}
880: %  \includegraphics[width=3.5cm]{repul-deviation.eps}
881: \hspace{-3cm}
882:   \includegraphics[width=10cm]{figure9.eps}
883: 
884: \caption{Mean interatomic distance $\langle w_k \rangle$ (left panel)
885:          and its standard deviation $\sigma_{wk}$ (right panel).
886:          The parameters are the same as in Fig.~\ref{pp}.
887: 	}
888: \label{wdp}
889: \end{figure}
890: %------------------------------------------------------------------------
891: 
892: In the case of repulsive atomic interaction ($U_{\rm bg}>0$), the situation is quite different.
893: There can be only one discrete level below the (quasi)continuum which is occupied
894: by the molecule as long as $\delta'$ is negative and its absolute value remains very large.
895: Above the (quasi)continuum, there is another discrete level corresponding to the atomic
896: bright soliton. If we increase $\delta'$
897: the lower discrete level approaches the (quasi)continuum and the probabilities of the atomic
898: states become larger meaning that the system enters the bright soliton regime
899: supported by the molecule creation (see Fig.~\ref{pp}).
900: Inequalities (\ref{ineq}) are satisfied in this regime.
901: If we increase $\delta'$ further and reach the value $\delta'_-$,
902: the probability of the molecular state becomes very small.
903: Inequalities (\ref{ineq}) are violated and
904: the discrete level merges with the (quasi)continuum, i.e.,
905: the bright soliton is destroyed.
906: The probability
907: $
908: \left|
909:     c_{k}^{\rm m}
910: \right|^2
911: $
912: can be also interpreted as a relative population of the molecular component.
913: It is a decreasing function of $\delta'$ like in the case of classical
914: atomic-molecular solitons~\cite{konotop}.
915: 
916: If the detuning is further increased up to $\delta'_+$,
917: the soliton band appears above the (quasi)continuum.
918: For the values of parameters used in our numerical estimations,
919: $\delta_-'=1.479\,\hbar\omega$ and
920: $\delta_+'=1.521\,\hbar\omega$.
921: 
922: The soliton binding energy $E_{\rm b}$ is again
923: a decreasing function of $\delta'$ which vanishes at $\delta'=\delta_-'$.
924: The effective mass $m^*$ equals to the effective mass of the molecule for large
925: negative $\delta'$ and reaches the value $\tilde m$ at $\delta'=\delta_-'$ (Fig.~\ref{mp}).
926: If we come closer to $\delta_-'$ the solitons become less localized especially at $k=0$
927: and the interatomic-distance fluctuations increase (Fig.~\ref{wdp}).
928: According to our definition of the soliton width, its behavior is similar to that of
929: $\langle w_k \rangle$ and $\sigma_{wk}$ shown in Fig.~\ref{wdp}.
930: 
931: %-----------------------------------------------------
932: \subsection{The limit $L\to\infty$}
933: %-----------------------------------------------------
934: 
935: %------------------------------------------------------------------------
936: \begin{figure}[tb]
937: \centering
938: 
939: %\psfrag{ds}[b]{$\Delta_s$}
940: %\psfrag{dbt}[b]{$\Delta_t$, $\Delta_b$}
941: %\psfrag{L}[c]{$L$}
942: %
943: %  \includegraphics[width=5cm]{ds.eps}
944: %
945: %  \includegraphics[width=5cm]{dbt.eps}
946: \hspace{-3cm}
947:   \includegraphics[width=8cm]{figure10.eps}
948: 
949: \caption{Deviations of the eigenvalues of Eq.(\ref{evp}) for finite $L$ from that
950:          obtained in the limit $L\to\infty$.
951: 	 $\Delta_t$ is shown by the dashed line.
952: 	 The parameters are the same as in Fig.~\ref{s}.
953: 	}
954: \label{d}
955: \end{figure}
956: %------------------------------------------------------------------------
957: 
958: In the limit $L\to\infty$, the wave number $k$ becomes a continuous variable.
959: The energies of the continuum band are enclosed in the interval
960: $
961:  \left|
962:      E_k
963:  \right|
964:  \le
965:  q_k
966: $,
967: where
968: $
969:  q_k
970:  =
971:  4 t_{\rm a}
972:  \cos
973:  \left(
974:      \frac{\pi}{2}
975:      \frac{k}{k_{\rm L}}
976:  \right)
977: $,
978: and the coefficients
979: $c_{0k}^{\rm a}$, $c_{k}^{\rm m}$,
980: in Eq.~(\ref{evp}) become negligibly small.
981: Outside of the continuum band, the solutions have the form
982: $
983:  c_{jk}^{\rm a}
984:  =
985:  a_k b_k^j
986:  \exp
987:  \left(
988:      i
989:      \frac{\pi}{2}
990:      \frac{k}{k_{\rm L}}
991:      j
992:  \right)
993: $,
994: $j=1,2,\dots,\infty$.
995: Substituting this ansatz into Eq.~(\ref{evp}) we obtain the equation
996: for the eigenenergy ${\cal E}_k = \lim_{L\to\infty}E_k$
997: \begin{eqnarray}
998: \label{E}
999: {\cal E}_k^2
1000: &=&
1001: \left(
1002:     U_{\rm bg}
1003:     +
1004:     U_k
1005: \right)^2
1006: +
1007: q_k^2
1008: \;,\;
1009: U_k
1010: =
1011: 2
1012: \tilde g^2
1013: /
1014: \left(
1015:     {\cal E}_k-\epsilon_{k}^{\rm m}
1016: \right)
1017: \;.
1018: \end{eqnarray}
1019: Note that the quantity $U_{\rm bg} + U_k$ plays a role of the effective atomic
1020: interaction.
1021: The values of $a_k$ and $b_k$ corresponding to a certain ${\cal E}_k$ are given by
1022: %\begin{eqnarray}
1023: %\label{b}
1024: $
1025: a_k
1026: =
1027: \sqrt{2}
1028: c_{0k}^{\rm a}
1029: $,
1030: %\;,
1031: $
1032: b_k
1033: %&=&
1034: =
1035: \left(
1036:     U_{\rm bg}
1037:     +
1038:     U_k
1039:     -
1040:     {\cal E}_k
1041: \right)/q_k
1042: $,
1043: %\;,
1044: %\end{eqnarray}
1045: and the expressions for the probabilities of the state with two atoms
1046: on the same lattice site and the molecular state take the form
1047: \begin{eqnarray}
1048: \label{c0a}
1049: \left|
1050:     c_{0k}^{\rm a}
1051: \right|^2
1052: &=&
1053: \left(
1054:     1-b_k^2
1055: \right)
1056: /
1057: \left[
1058:  1
1059:  +
1060:  b_k^2
1061:  +
1062:  \left(
1063:      1 - b_k^2
1064:  \right)
1065:  S_k
1066: \right]
1067: \;,
1068: \nonumber\\
1069: \left|
1070:     c_{k}^{\rm m}
1071: \right|^2
1072: &=&
1073: 2
1074: \tilde g^2
1075: \left|
1076:     c_{0k}^{\rm a}
1077: \right|^2
1078: /
1079:  \left(
1080:      {\cal E}_k
1081:      -
1082:      \epsilon_{k}^{\rm m}
1083:  \right)^2
1084: \;,
1085: \end{eqnarray}
1086: where
1087: $
1088: S_k
1089: =
1090: 2
1091: \tilde g^2
1092: /
1093: {
1094: \left(
1095:     {\cal E}_k
1096:     -
1097:     \epsilon_{k}^{\rm m}
1098: \right)^2
1099: }
1100: $.
1101: 
1102: Eq.~(\ref{E}) can be multiplied by
1103: $
1104:  \left(
1105:      {\cal E}_k
1106:      -
1107:      \epsilon_{k}^{\rm m}
1108:  \right)^2
1109: $
1110: and treated as quartic equation for ${\cal E}_k$ which contains always four roots.
1111: However, depending on the values of the parameters only one or two roots are real
1112: and provide normalized eigenstates implying that the others are unphysical
1113: and should be rejected. The normalization condition (\ref{norm}) requires
1114: $
1115:  a_k^2/2<1
1116: $
1117: as well as
1118: $
1119:   \left|
1120:       b_k
1121:   \right|
1122:   < 1
1123: $.
1124: One can easily show that in the special case $t_{\rm a}=t_{\rm m}=0$ the physical solutions
1125: of Eq.~(\ref{E}) are given by (\ref{e}).
1126: 
1127: We substitute ${\cal E}_{k\pm} = \pm q_k$ corresponding to the edges of the continuum band
1128: into Eq.(\ref{E}) and get
1129: \begin{equation}
1130: \label{b}
1131: U_{\rm bg}+U_{k\pm}=0
1132: \;,
1133: \end{equation}
1134: which leads to the identity
1135: $
1136:   \left|
1137:       b_k
1138:   \right|
1139:   = 1
1140: $
1141: and as a consequence to the violation of inequalities~(\ref{ineq}).
1142: Eq.~(\ref{b}) allows to obtain the boundaries $\delta'_-$ and $\delta'_+$ of the interval
1143: of $\delta'$ within which there is only one physical solution:
1144: \begin{equation}
1145: \label{dpm}
1146: \delta'_\pm
1147: =
1148: \pm q_k
1149: +
1150: 2 t_{\rm m}
1151: \cos
1152: \left(
1153:     \pi k/k_{\rm L}
1154: \right)
1155: +
1156: 2 \tilde g^2/U_{\rm bg}
1157: \,.
1158: \end{equation}
1159: For the values of parameters used in the numerical diagonalization, we find
1160: $\delta'_-=-1.531\,\hbar\omega$
1161: and
1162: $\delta'_+=-1.479\,\hbar\omega$
1163: in the case of attractive interaction, and
1164: $\delta'_-=1.479\,\hbar\omega$
1165: and
1166: $\delta'_+=1.532\,\hbar\omega$
1167: in the case of repulsive interaction.
1168: The values of $\delta_-'$ are in perfect agreement with the results of numerical
1169: calculations for $L=41$, while $\delta_+'$ have small deviations from the corresponding
1170: numerical estimations.
1171: In the special case $t_{\rm a}=t_{\rm m}=0$, $\delta'_-=\delta'_+=\delta'_*$ and Eq.~(\ref{b})
1172: leads to the condition
1173: \begin{equation}
1174: \label{cond}
1175: U_{\rm bg}
1176: -
1177: 2
1178: \tilde g^2/\delta'_*
1179: =
1180: 0
1181: \;.
1182: \end{equation}
1183: This is equivalent to the requirement that the effective scattering length
1184: $
1185:  a_{\rm bg}(1-\Delta B \Delta\mu/\delta')
1186: $,
1187: which appears in the mean-field theory as a result of the adiabatic elimination
1188: of the molecular field~\cite{MVA}, vanishes.
1189: The calculations presented above show that in the interval of the detunings
1190: $\delta'_- < \delta' <\delta'_+$
1191: the effective atomic interaction is gradually switched from the attractive
1192: to the repulsive one.
1193: 
1194: The probabilities
1195: $
1196: \left|
1197:     c_{jk}^{\rm a}
1198: \right|^2
1199: $,
1200: $j=1,2,\dots$,
1201: of the atomic states in Eq.(\ref{psi}) decrease with $j$ and have the form
1202: \begin{eqnarray}
1203: \left|
1204:     c_{ik}^{\rm a}
1205: \right|^2
1206: =
1207: \left(
1208:     1 - b_k^2
1209: \right)
1210: b_k^{2(i-1)}
1211: \left[
1212:     1
1213:     -
1214:     \left|
1215:         c_{0k}^{\rm a}
1216:     \right|^2
1217:     \left(
1218:         1
1219: 	+
1220:         S_k
1221:     \right)
1222: \right]
1223: \;.
1224: \end{eqnarray}
1225: The soliton effective mass
1226: \begin{equation}
1227: m^*
1228: =
1229: \hbar^2
1230: \left(
1231:     \left.
1232:     \frac
1233:     {\partial^2 {\cal E}_k}
1234:     {\partial k^2}
1235:     \right|_{k=0}
1236: \right)^{-1}
1237: \\
1238: =
1239: \hbar^2
1240: \left.
1241: \frac
1242: {
1243:  {\cal E}_k
1244:  +
1245:  \left(
1246:      U_{\rm bg}
1247:      +
1248:      U_k
1249:  \right)
1250:  S_k
1251: }
1252: {
1253:  2
1254:  t_{\rm m}
1255:  \left(
1256:      U_{\rm bg}
1257:      +
1258:      U_k
1259:  \right)
1260:  S_k
1261:  - 4 t_{\rm a}^2
1262: }
1263: \right|_{k=0}
1264: \end{equation}
1265: is smaller than that at the bottom of the continuum
1266: \begin{equation}
1267: \tilde m
1268: =
1269: \hbar^2
1270: \left(
1271:     -
1272:     \left.
1273:     \frac
1274:     {\partial^2 q_k}
1275:     {\partial k^2}
1276:     \right|_{k=0}
1277: \right)^{-1}
1278: =
1279: \frac
1280: {\hbar^2 k_{\rm L}^2}
1281: {\pi^2 t_{\rm a}}
1282: \;.
1283: \end{equation}
1284: The first two moments of the interatomic-distance distribution can be shown to be
1285: \begin{eqnarray}
1286: \langle
1287:    w_k
1288: \rangle
1289: &=&
1290: 2
1291: \left|
1292:     c_{0k}^{\rm a}
1293: \right|^2
1294: b_k^2
1295: /
1296: \left[
1297:     \left(
1298:         1 - b_k^2
1299:     \right)^2
1300:     \left(
1301:         1
1302: 	-
1303:         \left|
1304:             c_{k}^{\rm m}
1305:         \right|^2
1306:     \right)
1307: \right]
1308: \;,
1309: \\
1310: \langle
1311:    w_k^2
1312: \rangle
1313: &=&
1314: 2
1315: \left|
1316:     c_{0k}^{\rm a}
1317: \right|^2
1318: b_k^2
1319: \left(
1320:     1 + b_k^2
1321: \right)
1322: /
1323: \left[
1324:     \left(
1325:         1 - b_k^2
1326:     \right)^3
1327:     \left(
1328:         1
1329: 	-
1330:         \left|
1331:             c_{k}^{\rm m}
1332:         \right|^2
1333:     \right)
1334: \right]
1335: \;.
1336: \nonumber
1337: \end{eqnarray}
1338: 
1339: In order to demonstrate the convergence to the limit $L\to\infty$,
1340: we have plotted in Fig.~\ref{d} the quantities
1341: $
1342:   \Delta_{\rm s}
1343:   =
1344:   \sup_k
1345:   \left|
1346:       E_k^{({\rm s})}
1347:       -
1348:       {\cal E}_k
1349:   \right|
1350:   /
1351:   \hbar\omega
1352: $
1353: as well as
1354: $
1355:   \Delta_{\rm t}
1356:   =
1357:   \sup_k
1358:   \left|
1359:       E_k^{({\rm t})}
1360:       -
1361:       q_k
1362:   \right|
1363:   /
1364:   \hbar\omega
1365: $
1366: and
1367: $
1368:   \Delta_{\rm b}
1369:   =
1370:   \sup_k
1371:   \left|
1372:       E_k^{({\rm b})}
1373:       +
1374:       q_k
1375:   \right|
1376:   /
1377:   \hbar\omega
1378: $
1379: for different $L$, where
1380: $E_k^{({\rm s})}$, $E_k^{({\rm t})}$, and $E_k^{({\rm b})}$
1381: are the eigenenergies of Eq.(\ref{evp}) corresponding to the soliton band,
1382: the top and the bottom of the quasi-continuum band, respectively.
1383: $\Delta_{\rm s}$ decreases exponentially with the increase of $L$ and it is very small
1384: even for low values of $L$. The convergence for the boundaries of the continuum
1385: band is slower, but the limit $L\to\infty$ describes quite well the results of
1386: the numerical diagonalization already for a few tens of the lattice sites.
1387: In addition, we have compared the results of the calculations
1388: obtained on the basis of numerical solution
1389: of the eigenvalue problem~(\ref{evp}) for $L=41$
1390: which are presented in Figs.~\ref{s}-\ref{wdp}
1391: with that worked out in the limit $L\to\infty$
1392: and did not find any noticeable discrepancies.
1393: This is consistent with the exponential decrease of $\Delta_{\rm s}(L)$.
1394: 
1395: In the absence of the magnetic field, the molecule creation is impossible
1396: and one has to put $\tilde g=0$, $\delta'=0$, $t_{\rm m}=0$, in all the equations.
1397: In this special case, the normalizable solution of Eq.~(\ref{E}) is given by
1398: $
1399: {\cal E}_k^{(0)}
1400: =
1401: {\rm sign}(U_{\rm bg})
1402: \sqrt{
1403:       U_{\rm bg}^2
1404:       +
1405:       q_k^2
1406:      }
1407: $,
1408: which leads to the following expression for the effective mass
1409: $
1410: m^{*(0)}
1411: =
1412: -
1413: \hbar^2
1414: {\rm sign}(U_{\rm bg})
1415: \sqrt{
1416:       U_{\rm bg}^2
1417:       +
1418:       16 t_{\rm a}^2
1419:      }
1420: /
1421: \left(
1422:  4 t_{\rm a}^2
1423: \right)
1424: $.
1425: These are exactly the results presented in Ref.~\cite{Scott}.
1426: The soliton band exists again for repulsive as well as attractive atomic
1427: interaction, but in the case of repulsive interaction it appears to be
1428: a highly excited mode with the energy above the continuum band.
1429: 
1430: %-----------------------------------------------------
1431: \section{Conclusion}
1432: %-----------------------------------------------------
1433: 
1434: Summarizing, we have investigated quantum lattice solitons in a system of two
1435: ultracold bosons near the Feshbach resonance.
1436: Binding energy, effective mass, and spatial width of the solitons,
1437: can be manipulated varying the detuning from
1438: the Feshbach resonance.
1439: In the case of attractive atomic interactions, the molecule creation stabilizes
1440: the solitons increasing their effective mass as well as the binding energy
1441: and decreasing the width.
1442: In the case of repulsive interactions, the molecule creation leads
1443: to the possibility of existence of bright solitons in some interval of detunings
1444: analogous to the corresponding classical system.
1445: The presence of quantum fluctuations leads to the fact that the interatomic distance
1446: is a random quantity. Its standard deviation is even larger than the mean value.
1447: 
1448: The classical limit of the problem studied in the present work was considered
1449: in~\cite{konotop}. Our results for the relative populations of the atomic
1450: and molecular components are in agreement with the corresponding classical results.
1451: In order to understand the transition from quantum to classical solitions
1452: it is necessary to perform analogous calculations for higher number of atoms.
1453: This can be done employing the same method as in the present study.
1454: However, one has to keep in mind
1455: that the dimension of the Hilbert space increases rapidly with the increase
1456: of the particle number and the number of lattice sites.
1457: 
1458: %-----------------------------------------------------
1459: \ack
1460: This work was partly supported by the INTAS (Project No. 01-855) and SFB/TR 12.
1461: K.V.K. would like to thank the University of Bath for kind hospitality.
1462: 
1463: %-------------------------------------------
1464: \section*{References}
1465: %-------------------------------------------
1466: 
1467: %---------------------------------------------------------------------
1468: \begin{thebibliography}{99}
1469: %---------------------------------------------------------------------
1470: 
1471: \bibitem{becsol}
1472: Burger S, Bongs K, Dettmer S, Ertmer W, Sengstock K, Sanpera A, Shlyapnikov G V 
1473: and Lewenstein M
1474: 1999 \PRL {\bf  83} 5198
1475: 
1476: \nonum
1477: Denschlag J et al 2000 {\it Science} {\bf 287} 97
1478: 
1479: \nonum
1480: Strecker K E, Partridge G B, Truscott A G and Hulet R G
1481: 2002 {\it Nature} {\bf 417} 150
1482: 
1483: \nonum
1484: Khaykovich L, Schreck F, Ferrari G, Bourdel T, Cubizolles J, Carr L D, Castin Y and Salomon C
1485: 2002 {\it Science} {\bf 296} 1290
1486: 
1487: \bibitem{markus}
1488: Eiermann B, Anker Th, Albiez M, Taglieber M, Treutlein P, Marzlin K P and Oberthaler M K
1489: 2004 \PRL {\bf 92}, 230401
1490: 
1491: \bibitem{bth}
1492: Dziarmaga J and Sacha K 2002 \PR A {\bf 66} 043620
1493: \nonum
1494: Law C K, Leung P T and Chu M C 2002 \jpb {\bf 35} 3583
1495: \nonum
1496: Law C K 2003 \PR A {\bf 68} 015602
1497: \nonum
1498: Dziarmaga J 2004 \PR A {\bf 70} 063616
1499: \nonum
1500: Buljan H, Segev M and Vardi A 2005 \PRL {\bf 95} 180401
1501: 
1502: \bibitem{Scott}
1503: Scott A C 1999 {\it Nonlinear science}
1504: (Oxford University Press) Ch.~8
1505: and references therein
1506: 
1507: \bibitem{el}
1508: Dorignac J, Eilbeck J C, Salerno M and Scott A C 2004 \PRL {\bf 93} 025504
1509: \nonum
1510: Eilbeck J C and Palmero F 2004 \PL A {\bf 331} 201
1511: 
1512: \bibitem{skryabin}
1513: Yulin A V, Skryabin D V and Russell P St J 2003 \PRL {\bf 91} 260402
1514: 
1515: \bibitem{discr}
1516: Trombettoni A, Smerzi A and Bishop A R 2002 \PRL {\bf 88} 173902
1517: 
1518: \bibitem{MVA}
1519: Moerdijk A J, Verhaar B J and Axelsson A 1995 \PR A {\bf 51} 4852
1520: 
1521: \bibitem{fesh}
1522: Zwierlein M W, Stan C A, Schunck C H, Raupach S M F, Gupta S, Hadzibabic Z and Ketterle W
1523: 2003 \PRL {\bf 91} 250401
1524: \nonum
1525: D\"urr S, Volz Th, Marte A and Rempe G 2004 \PRL {\bf 92} 020406
1526: 
1527: \bibitem{par1}
1528: Drummond P D, Kheruntsyan K V and He H 1998 \PRL {\bf 81} 3055
1529: 
1530: \bibitem{par2}
1531: Vaughan T G, Kheruntsyan K V and Drummond P D 2004 \PR A {\bf 70} 063611
1532: 
1533: \bibitem{rev}
1534: Buryak A V, Trapani P D, Skryabin D V and Trillo S
1535: 2002 {\it Physics Reports} {\bf 370} 63 and
1536: references therein
1537: 
1538: \bibitem{quant}
1539: Drummond P D and He H 1997 \PR A {\bf 56} R1107
1540: \nonum
1541: Kheruntsyan K V and Drummond P D 1998 \PR A {\bf 58} R2676
1542: \nonum
1543: Kheruntsyan K V and Drummond P D 1998 \PR A {\bf 58} 2488
1544: \nonum
1545: Kheruntsyan K V and Drummond P D 2000 \PR A {\bf 61} 063816
1546: \nonum
1547: Drummond P D and Kheruntsyan K V 2004 \PR A {\bf 70} 033609
1548: 
1549: \bibitem{steg}
1550: Iwanow R, Schiek R, Stegeman G I, Pertsch T, Lederer F, Min Y and Sohler W
1551: 2004 \PRL {\bf 93} 113902
1552: 
1553: \bibitem{konotop}
1554: Abdullaev F Kh and Konotop V V 2003 \PR A {\bf 68} 013605
1555: 
1556: \bibitem{Greiner}
1557: Greiner M, Mandel O, Esslinger T, H\"ansch Th W and Bloch I
1558: 2002 {\it Nature} {\bf 415} 39
1559: 
1560: \bibitem{single}
1561: Pinkse P W H, Fischer T, Maunz P and Rempe G
1562: 2000 {\it Nature} {\bf 404} 365
1563: \nonum
1564: Maunz P, Puppe T, Schuster I, Syassen N, Pinkse P W H and Rempe G
1565: 2004 {\it Nature} {\bf 428} 50
1566: 
1567: \bibitem{DKOS}
1568: Dickerscheid D B M, Khawaja U Al, van Oosten D and Stoof H T C 2005 \PR A {\bf 71} 043604
1569: 
1570: \bibitem{comment}
1571: Diener R B and Ho T L 2006 \PR A {\bf 73} 017601
1572: \nonum
1573: Dickerscheid D B M, van Oosten D and Stoof H T C 2006 \PR A {\bf 73} 017602
1574: 
1575: \end{thebibliography}
1576: 
1577: \end{document}
1578: