1: %\usepackage{
2: %amsmath,
3: %amstext,
4: %amsgen,amsbsy,amsopn,
5: %amsfonts,
6: %graphicx,overcite,
7: %theorem,amssymb}
8: %\setcounter{MaxMatrixCols}{10}
9: %\renewcommand{\baselinestretch}{1.5}
10: %{\theorembodyfont{\rmfamily} \theoremheaderfont{\itshape}
11: %\newtheorem{thm}{Theorem}
12: %}
13: %\documentclass[12pt]{article}
14: %\usepackage{amsmath}
15:
16: %last updated 05/03/04 by MAP
17:
18:
19: \documentstyle[aps,amsmath,multicol,prl,epsfig]{revtex}
20:
21: \begin{document}
22: \title{Resonant and Non-Resonant Modulated Amplitude Waves for Binary
23: Bose-Einstein Condensates in Optical Lattices}
24: \author{Mason A. Porter$^1$, P.G. Kevrekidis$^2$, and Boris A. Malomed$^3$}
25: \address{$^1$School of Mathematics and Center for Nonlinear Science, \\
26: Georgia Institute of Technology, Atlanta GA 30332, USA \\
27: $^2$Department of Mathematics and Statistics,\\
28: University of Massachusetts, Amherst MA 01003-4515, USA \\
29: $^3$Department of Interdisciplinary Studies, Faculty of Engineering, \\
30: Tel Aviv University, Tel Aviv 69978, Israel}
31: \date{\today }
32: \maketitle
33:
34: \begin{abstract}
35: We consider a system of two Gross-Pitaevskii (GP) equations, in the presence
36: of an optical-lattice (OL) potential, coupled by both nonlinear and linear
37: terms. This system describes a Bose-Einstein condensate (BEC) composed of
38: two different spin states of the same atomic species, which interact
39: linearly through a resonant electromagnetic field. In the absence of the OL,
40: we find plane-wave solutions and examine their stability. In the
41: presence of the OL, we derive a system of amplitude equations for spatially
42: modulated states which are coupled to the periodic potential through the
43: lowest-order subharmonic resonance. We determine this averaged system's
44: equilibria, which represent spatially periodic solutions, and subsequently
45: examine the stability of the corresponding solutions with direct simulations
46: of the coupled GP equations. We find that symmetric (equal-amplitude) and
47: asymmetric (unequal-amplitude) dual-mode resonant states are, respectively,
48: stable and unstable. The unstable states generate periodic oscillations
49: between the two condensate components, which is possible only because of the
50: linear coupling between them. We also find four-mode states, but they are
51: always unstable. Finally, we briefly consider ternary (three-component)
52: condensates.
53: \end{abstract}
54:
55:
56: \vspace{2mm}
57:
58: PACS: 05.45.-a, 03.75.Lm, 05.30.Jp, 05.45.Ac
59:
60:
61:
62: \section{Introduction}
63:
64: At sufficiently low temperatures, particles in a dilute boson gas condense in
65: the ground state, forming a Bose-Einstein condensate (BEC).
66: This was first observed experimentally in 1995 in Na and Rb
67: vapors\cite {pethick,stringari,ketter,edwards}.
68:
69: In the mean-field approximation, a dilute BEC is described by the
70: nonlinear Schr\"{o}dinger (NLS) equation with an external
71: potential, which is also called the Gross-Pitaevskii (GP)
72: equation. In particular, BECs may be considered in the
73: quasi-one-dimensional (quasi-1D) regime, with the transverse
74: dimensions of the condensate on the order of its mean healing
75: length $\chi $ (given by $\chi ^{2}=(8\pi n|a|)^{-1}$) and a much larger
76: longitudinal dimension \cite{bronski,bronskirep,bronskiatt,stringari}.
77: The length $\chi $ is determined by the mean atomic density $n$ and the
78: two-body $s$-wave scattering length $a$, where the interactions between
79: atoms are repulsive if $a>0$ and attractive if
80: $a<0$ \cite{pethick,stringari,kohler,baiz}.
81:
82: The quasi-1D regime, which corresponds to \textquotedblleft
83: cigar-shaped\textquotedblright\ BECs, is described by the 1D limit of the 3D
84: mean-field theory (rather than by a 1D mean-field theory proper, which would
85: only be appropriate for extremely small transverse dimensions of order
86: $\sim a$) \cite{bronski,bronskiatt,bronskirep,salasnich,towers}. In this
87: situation, the condensate wave function $\psi (x,t)$ obeys the effective 1D GP
88: equation,
89: \[
90: i\hbar \psi _{t}=-\frac{\hbar ^{2}}{2m}\psi _{xx}+g|\psi |^{2}\psi
91: +V(x)\psi ,
92: \]
93: where $m$ is the atomic mass, $V(x)$ is an external potential,
94: $g=[4\pi \hbar ^{2}a/m][1+O(\eta ^{2})]$, and $\eta
95: =\sqrt{n|a|^{3}}$ is the dilute-gas
96: parameter\cite{stringari,kohler,baiz,band}. Experimentally
97: relevant potentials $V(x)$ include harmonic traps and periodic
98: potentials (created as optical lattices, which are denoted OLs and arise as
99: interference
100: patterns produced by coherent counterpropagating
101: laser beams illuminating the condensate). In the presence of both potentials,
102: $ V(x)=V_{0}\cos \left[
103: 2\kappa (x-x_{0})\right] +V_{1}x^{2}/2$, where $x_{0}$ is the
104: offset of the periodic potential relative to the center of the the
105: parabolic trap. When $\left( 2\pi /\kappa \right) ^{2}V_{1}\ll
106: V_{0}$, the potential is dominated by its periodic component over
107: many periods \cite {kutz,promislow,lattice}; for example, when
108: $V_{0}/V_{1}=500$ and $\kappa =10 $, the parabolic component in
109: $V(x)$ is negligible for the 10 periods closest to the trap's center.
110: In this work, we set $V_{1}=0$ and focus entirely on OL
111: potentials. This assumption is motivated by numerous recent
112: experimental studies of BECs in OLs \cite{hagley,anderson} and is widely
113: adopted in theoretical studies \cite
114: {bronski,bronskiatt,bronskirep,space1,space2,promislow,kutz,malopt,alf,smer,mapbecprl,mapbec,mueller,wu4,machholm,pethick2}.
115:
116: Multiple-component BECs, which constitute the subject of this
117: work, have been considered in {a number of} theoretical works
118: \cite {couple1,couple2,couple3,couple4,couple5,couple6,dec}.
119: Mixtures of two different species (such as $^{85}$Rb and
120: $^{87}$Rb) are described by nonlinearly coupled GP equations:
121: \begin{align}
122: i\hbar \frac{\partial \psi _{1}}{\partial t}& =-\frac{\hbar
123: ^{2}}{2m_{1}} \nabla ^{2}\psi _{1}+g_{1}|\psi _{1}|^{2}\psi
124: _{1}+V(x)\psi _{1}+h|\psi
125: _{2}|^{2}\psi _{1}, \nonumber \\
126: i\hbar \frac{\partial \psi _{2}}{\partial t}& =-\frac{\hbar
127: ^{2}}{2m_{2}} \nabla ^{2}\psi _{2}+g_{2}|\psi _{2}|^{2}\psi
128: _{2}+V(x)\psi _{2}+h|\psi _{1}|^{2}\psi _{2}, \label{cnls1}
129: \end{align}
130: where $m_{1,2}$ are the atomic masses of the species, $g_{j}\equiv 4\pi
131: \hbar ^{2}a_{j}/m_{j}$ corresponds to the self-scattering length $a_{j}$,
132: and
133: \begin{equation}
134: h\equiv g_{12}=2\pi \hbar ^{2}a_{12}\left( m_{1}+m_{2}\right) /\left(
135: m_{1}m_{2}\right) \label{h}
136: \end{equation}
137: depends on the cross-scattering length $a_{12}$
138: \cite{couple2}. There are numerous subcases of Eqs. (\ref{cnls1})
139: to consider, as various combinations of signs for the scattering
140: coefficients $g_{1}$, $g_{2}$, and $ h$ may occur. It is important
141: to note, however, that if $g_{1,2}$ are positive (repulsion
142: between the atoms), then $h$ is normally positive as well. However,
143: if $g_{1,2}$ are negative (corresponding to the less typical
144: case of attraction
145: between atoms belonging to a single species), then $h$ [see Eq. (\ref{h})]
146: may be {\it either} positive or negative.
147:
148: The system (\ref{cnls1}) resembles a well-known model
149: describing the nonlinear self-phase-modulation (SPM) and
150: cross-phase-modulation (XPM) interactions of light waves with
151: different polarizations or carried by different wavelengths
152: in nonlinear optics \cite{Agrawal}. In the case of optical fibers,
153: the evolution variable is the propagation distance $z$ (rather
154: than time), and the role of $x$ is played by the reduced temporal
155: variable $\tau $ \cite{Agrawal}. In optical models, however, the
156: choice of the nonlinear coefficients is limited to the
157: combinations $g_{1}=g_{2}=3h/2$ for orthogonal linear
158: polarizations in a birefringent fiber and $ g_{1}=g_{2}=h/2$ for
159: circular polarizations or different carrier wavelengths. In fact,
160: the latter case is quite important in the application to
161: two-component BECs as well, as it occurs if one assumes that the
162: collision lengths for interactions between all the atoms are the
163: same.
164:
165: Another physically interesting feature, which we include in the model to be
166: considered below, is linear coupling between the two wave functions. This
167: occurs in a mixture of two different spin states of the
168: same isotope, which arises through a resonant microwave field that
169: induces transitions between the
170: states\cite{linear-coupling1,linear-coupling2}. Condensates
171: containing two different spin states of $^{87}$Rb have been created
172: experimentally via sympathetic cooling \cite{myatt}. In this
173: situation, the normalized coupled GP equations take the form
174: \begin{align}
175: i\frac{\partial \psi _{1}}{\partial t}& =-\nabla ^{2}\psi _{1}+g|\psi
176: _{1}|^{2}\psi _{1}+V(x)\psi _{1}+h|\psi _{2}|^{2}\psi _{1}+\alpha \psi _{2},
177: \nonumber \\
178: i\frac{\partial \psi _{2}}{\partial t}& =-\nabla ^{2}\psi _{2}+g|\psi
179: _{2}|^{2}\psi _{2}+V(x)\psi _{2}+h|\psi _{1}|^{2}\psi _{2}+\alpha \psi _{1},
180: \label{cnls2}
181: \end{align}
182: where the self- and cross-scattering coefficients are ${ g_1 =
183: g_2\equiv g}$ and $h$, and the linear coupling coefficient is
184: $\alpha $, which can always be made positive without loss of
185: generality.
186:
187: Experimental studies of mixtures of two interconvertible
188: condensates (with positive scattering lengths) loaded in an OL
189: have not yet been reported. However, all the necessary
190: experimental ingredients for such a work are currently available.
191: Moreover, in a very recent paper \cite{kuklov}, an experimental
192: procedure, based on Ramsey spectroscopy and adjusted exactly
193: for such a system, was elaborated. Experiments in this setting
194: would be quite interesting, as they would allow the study of the
195: direct interplay between two crucially important physical
196: factors used as tools in the current experimental work---namely,
197: the OL potential and inter-conversion between two different spin
198: states in the BEC, controlled by the resonant field. Furthermore,
199: there are now recent experimental results with linearly
200: coupled BECs \cite{dsh}.
201: The use of an optical potential in the latter setting is a rather
202: straightforward extension.
203:
204: The model combining nonlinear XPM and linear couplings, as in Eqs.
205: (\ref {cnls2}), occurs in fiber optics as well. In that case, the
206: linear coupling is generated by a twist applied to the fiber in
207: the case of two linear polarizations, and by an elliptic
208: deformation of the fiber's core in the case of circular
209: polarizations (see, for example, \cite{old}). However,\ linear
210: coupling is impossible in the case of two different wavelengths.
211: Another optical model, with only linear coupling between two
212: modes, applies to dual-core nonlinear fibers, as discussed in
213: \cite{UNSW} (and references therein). In the context of BECs, it
214: may correspond to a special case in which the cross-scattering
215: length is made (very close to) zero using a Feshbach resonance \cite{inouye}.
216:
217: In this work, we aim to investigate modulated-wave states in the binary BEC
218: described by Eqs. (\ref{cnls2}), which include both nonlinear and linear
219: couplings and an OL potential. We stress that the interplay between the
220: microwave-induced linear coupling in the binary model and the OL-induced
221: periodic potential has never before been considered. As both features
222: represent important laboratory tools, the results reported here suggest
223: possibilities for new experiments. The model we study predicts new dynamical
224: effects, such as oscillations of matter between the linearly-coupled components
225: trapped in the potential wells of the OL.
226:
227: In our study, we begin by examining plane-wave solutions with
228: $V(x)\equiv 0$. When $V(x)\neq 0$, we apply a
229: standing-wave ansatz to Eq. (\ref{cnls2}), which leads to a system of coupled
230: parametrically forced Duffing equations describing the spatial evolution of
231: the fields. Using the method of averaging \cite{675,rand}, we study periodic
232: solutions of the latter system (called \textquotedblleft modulated amplitude
233: waves\textquotedblright and denoted MAWs). The stability of MAWs
234: (and the ensuing dynamics, in the case of instability) is then
235: tested by numerically simulating the underlying system of coupled
236: GP equations. This approach, though simpler than the more
237: ``rigorous'' computation of linear stability eigenvalues for
238: infinitesimal perturbations, provides a more realistic emulation of
239: physical experiments. Note additionally that although our stability
240: results will be illustrated by a few selected examples, we have
241: checked---by exploring different parameter regions---that these
242: examples represent the MAW stability features rather generally.
243:
244: The MAW solutions are especially interesting when the system
245: exhibits a spatial resonance. In this work, we consider both
246: non-resonant solutions and solutions featuring a subharmonic
247: resonance of the $2\!:\!1\!:\!1$ form. The latter situation has
248: been studied in the context of period-doubling in {\em
249: single-component} BECs in an OL potential
250: \cite{pethick2,mapbecprl,mapbec}, but---to the best of our
251: knowledge---spatial-resonance states in models of composite BECs
252: have not been considered previously.
253:
254: An alternative (but less general) approach to the study of binary BECs with
255: linear coupling, loaded into an OL, would be to seek exact
256: elliptic-function solutions to Eqs. (\ref{cnls2}) for the case of
257: elliptic-function potentials, $V(x)=-V_{0}{\rm sn}^{2}(\kappa x,k)$, as
258: has been done earlier in the two-component model without linear coupling
259: \cite{dec}. In that work, stable standing-wave solutions were found under the
260: assumption that the interaction matrix is positive definite. This occurs, for
261: instance, when all the interactions are repulsive, although small negative
262: cross-interactions are compatible with this condition as well.
263:
264: The rest of this paper is structured as follows: In section II, we derive
265: plane-wave solutions and analyze their stability. In section III, we introduce
266: modulated amplitude waves, and in section IV, we derive and solve averaged
267: equations that describe them in both non-resonant and resonant situations.
268: We corroborate our results and test the stability of the MAWs using
269: numerical simulations. In section V, we briefly examine a more general model
270: of a ternary (three-component) BEC with linear couplings. Finally, we
271: summarize our results in section VI.
272:
273:
274: \section{Plane-Wave Solutions}
275:
276: In the absence of the external potential ($V=0$), we find plane-wave
277: solutions of the form
278: \begin{equation}
279: \psi _{j}=R_{j}\exp \left[ i(k_{j}x-\mu _{j}t)\right] \,,\quad j=1,2\,.
280: \label{plane}
281: \end{equation}
282: For the linearly coupled GP equation (\ref{cnls2}), it is
283: necessary that $ k_{1}=k_{2}\equiv k$ and $\mu _{1}=\mu
284: _{2}\equiv \mu $. Without linear coupling, [as in Eqs.
285: (\ref{cnls1})], one may seek a broader class of solutions with
286: independent frequencies (which correspond to chemical potentials in the
287: physical context of BECs). It is important to note that the results obtained
288: in the study of optical models suggest that the addition of linear
289: coupling terms to a system of coupled NLS equations drastically alters
290: the dynamical behavior \cite{old}. In this section, we study the
291: model without the OL, which will be included in subsequent sections.
292:
293: Inserting Eq. (\ref{plane}) into Eqs. (\ref{cnls2}) yields
294: \begin{align}
295: \mu R_{1}& =k^{2}R_{1}+gR_{1}^{3}+hR_{1}R_{2}^{2}+\alpha R_{2}\,, \\
296: \mu R_{2}& =k^{2}R_{2}+gR_{2}^{3}+hR_{1}^{2}R_{2}+\alpha R_{1}\,,
297: \end{align}
298: which implies that
299: \[
300: \lbrack (g-h)R_{1}R_{2}-\alpha ][R_{1}^{2}-R_{2}^{2}]=0\,.
301: \]
302: One of the following two relations must then be satisfied:
303: \begin{equation}
304: R_{1}=\pm R_{2}\,; \label{plane1}
305: \end{equation}
306: \begin{equation}
307: R_{1}R_{2}=\frac{\alpha }{g-h}\,. \label{plane2}
308: \end{equation}
309:
310: With Eq. (\ref{plane1}), a nonzero solution satisfies $R_{2}=\pm
311: \sqrt{\left( k^{2}+\mu \mp \alpha \right) /\left( g+h\right)
312: }$. It exists when $g+h>0$, provided $k^{2}+\mu \mp \alpha
313: >0$, and when $g+h<0$, if $ k^{2}+\mu \mp \alpha <0$. When
314: $g=-h$, one obtains solutions of the form $ \left(
315: R_{1},R_{2}\right) =\left( \pm R,R\right) $ with arbitrary $R$ and
316: $ k^{2}=\mu \mp \alpha $.
317:
318: From Eq. (\ref{plane2}), one finds that
319: \begin{equation}
320: R_{2}^{2}=\frac{\mu -k^{2}}{2g}\pm \frac{1}{2}\sqrt{\left( \frac{\mu
321: -k^{2}}{g}\right) ^{2}-\frac{4\alpha ^{2}}{(g-h)^{2}}}\,, \label{tip2}
322: \end{equation}
323: under the restriction that this expression must be positive. When
324: $\alpha \neq 0$, the term inside the square root is smaller in
325: magnitude than the one outside, so solutions of this type exist as
326: long as $\left( \mu -k^{2}\right) g\geq 0$ and the argument of
327: the square root in (\ref{tip2}) is non-negative. Hence, for
328: repulsive and attractive BECs, respectively, the first condition implies,
329: $\mu >k^{2}$ and $\mu <k^{2}$. When $ h=0$,
330: Eq. (\ref{tip2}) takes the form
331: \[
332: R_{2}^{2}=\left( 2g\right) ^{-1}\left( \mu -k^{2}\pm \sqrt{(\mu
333: -k^{2})^{2}-4\alpha ^{2}}\right) \,.
334: \]
335: For both $h=0$ and $h=2g$, the condition on the argument of the square root
336: implies that to obtain real solutions, it is necessary to impose the
337: condition $|\mu -k^{2}|~\geq 2\alpha $.
338:
339: To examine the stability of the plane waves, we substitute
340: \[
341: \psi _{j}(x,t)=\phi _{j}(x,t)[1+\epsilon _{j}(x,t)]\,,\quad |\epsilon
342: _{j}|^{2}\ll 1\,,
343: \]
344: into Eqs. (\ref{cnls2}). This yields coupled linearized equations
345: for $ \epsilon _{1}(x,t)$ and $\epsilon _{2}(x,t)$. Assuming that
346: $\epsilon _{j}$ is periodic in $x$, it can be expanded in a
347: Fourier series,
348: \[
349: \epsilon _{j}(x,t)=\sum_{n=-\infty }^{\infty }\hat{\epsilon}_{jn}(t)\exp
350: (i\nu _{n}x)\,,
351: \]
352: where the $n$th mode has wavenumber $\nu _{n}$. The perturbation growth
353: rates that determine the stability of the $n$th mode are then given by
354: \begin{equation}
355: \lambda _{n}=-2ik\nu _{n}\pm \nu _{n}\sqrt{-\nu
356: _{n}^{2}-g(|R_{1}|^{2}+|R_{2}|^{2})\pm \sqrt{
357: g^{2}(|R_{1}|^{2}-|R_{2}|^{2})^{2}+4h^{2}|R_{1}|^{2}|R_{2}|^{2}}}\,,
358: \label{lambda}
359: \end{equation}
360: where the two sign combinations $\pm $ are independent (so there
361: are four distinct eigenvalues). Instability occurs when the
362: expression under the square root in Eq. (\ref{lambda}) has a
363: positive real part, causing the side-band modes $k+\nu _{n}$,
364: $k-\nu _{n}$ of the perturbed solution to grow exponentially. In
365: single-component condensates, this can occur only for $ g<0$
366: \cite{split}.
367:
368: Eigenvalues whose interior square root in Eq. (\ref{lambda}) has a $+$ sign
369: will produce the instability before ones with a $-$ sign, so we only need to
370: check the former case. For example, if $h=2g$, the instability occurs if
371: \begin{equation}
372: \nu _{n}^{2}+g\left( |R_{1}|^{2}+|R_{2}|^{2}-\sqrt{
373: |R_{1}|^{4}+14|R_{1}^{2}R_{2}^{2}|+|R_{2}|^{2}}\right) <0. \label{side}
374: \end{equation}
375: Stability conditions for the plane-wave solutions to Eqs. (\ref{cnls2}) with
376: $V=0$ can be obtained for all the possible sign combinations of $g$ and $h$.
377: We do not display them here, as they are rather cumbersome to write (although
378: straightforward to compute).
379:
380:
381: \section{Modulated Amplitude Waves}
382:
383: We now generalize the above analysis to consider the two-component GP system
384: in the presence of an OL potential. Toward this aim, we introduce solutions to
385: Eqs. (\ref{cnls2}) that describe coherent structures of the form
386: \begin{equation}
387: \psi _{j}(x,t)=R_{j}(x)\exp \left[ i(\theta (x)-\mu t)\right] \,,\quad
388: j=1,2\,. \label{coher}
389: \end{equation}
390: Inserting Eq. (\ref{coher}) into Eqs. (\ref{cnls2}) and equating real and
391: imaginary parts of the resulting equations yields
392: \begin{align}
393: \mu R_{1}& =-R_{1}^{\prime \prime }+ R_1 \left( \theta ^{\prime }\right)
394: ^{2}{}+gR_{1}^{3}+V(x)R_{1}+hR_{2}^{2}R_{1}+\alpha R_{2}\,, \nonumber \\
395: \mu R_{2}& =-R_{2}^{\prime \prime }+ R_2 \left( \theta ^{\prime }\right)
396: ^{2}+gR_{2}^{3}+V(x)R_{2}+hR_{1}^{2}R_{2}+\alpha R_{1}\,, \label{R} \\
397: 0& =R_{1}\theta ^{\prime \prime }+R_{1}^{\prime }\theta ^{\prime },
398: \nonumber \\
399: 0& =R_{2}\theta ^{\prime \prime }+R_{2}^{\prime }\theta ^{\prime },
400: \label{theta}
401: \end{align}
402: where the prime stands for $d/dx$. Equations (\ref{theta}) imply that
403: \begin{equation}
404: \theta (x)=c_{1}\int \frac{dx^{\prime }}{R_{1}^{2}(x^{\prime })}=c_{2}\int
405: \frac{dx^{\prime }}{R_{2}^{2}(x^{\prime })}\,, \label{strange}
406: \end{equation}
407: with arbitrary integration constants $c_{1}$ and $c_{2}$, so $
408: R_{1}(x)=bR_{2}(x)$ for some constant $b$ unless $c_{1}=c_{2}=0$.
409: In other words, $R_{1}(x)$ and $R_{2}(x)$ are different in this
410: context only when one considers solutions with null
411: \textquotedblleft angular momenta.\textquotedblright \hspace{.1 cm} In the
412: latter situation, Eqs. (\ref{R}) assume the form
413: \begin{align}
414: R_{1}^{\prime }& =S_{1}\,, \nonumber \\
415: S_{1}^{\prime }& =-\mu R_{1}+gR_{1}^{3}+hR_{1}R_{2}^{2}+\alpha
416: R_{2}+V(x)R_{1}\,, \nonumber \\
417: R_{2}^{\prime }& =S_{2}\,, \nonumber \\
418: S_{2}^{\prime }& =-\mu R_{2}+gR_{2}^{3}+hR_{1}^{2}R_{2}+\alpha
419: R_{1}+V(x)R_{2}\,. \label{mat}
420: \end{align}
421: When the potential $V(x)$ is sinusoidal, Eqs. (\ref{mat}) are (linearly and
422: nonlinearly) coupled cubic Mathieu equations.
423:
424:
425: \section{Averaged Equations and Spatial Subharmonic Resonances}
426:
427: To achieve some analytical understanding of the spatial resonances in
428: linearly coupled BECs, we average equations (\ref{mat}) in the physically
429: relevant case of the OL potential,
430: \begin{equation}
431: V(x)=V_{0}\cos (2\kappa x)\,. \label{OL}
432: \end{equation}
433: Defining $V_{0}\equiv -\epsilon \tilde{V}_{0}$, $g\equiv \epsilon
434: \tilde{g}$, $h\equiv \epsilon \tilde{h}$, and $\alpha \equiv
435: \epsilon \tilde{\alpha}$, Eqs. (\ref{mat}) may be written
436: \begin{align}
437: R_{1}^{\prime \prime }+\mu R_{1}& =-\epsilon
438: \tilde{V}_{0}R_{1}\cos (2\kappa x)+\epsilon
439: \tilde{g}R_{1}^{3}+\epsilon \tilde{h}
440: R_{1}R_{2}^{2}+\epsilon \tilde{\alpha}R_{2}\,, \nonumber \\
441: R_{2}^{\prime \prime }+\mu R_{2}& =-\epsilon
442: \tilde{V}_{0}R_{2}\cos (2\kappa x)+\epsilon
443: \tilde{g}R_{2}^{3}+\epsilon \tilde{h} R_{1}^{2}R_{2}+\epsilon
444: \tilde{\alpha}R_{1}\,. \label{mat2}
445: \end{align}
446: Assuming $\mu >0$, we insert the ansatz
447: \begin{align}
448: R_{j}(x)& =A_{j}(x)\cos \left( \sqrt{\mu }x\right) +B_{j}(x)\sin \left(
449: \sqrt{\mu }x\right) \,, \nonumber \\
450: R_{j}^{\prime }(x)& =-\sqrt{\mu }A_{j}(x)\sin \left(
451: \sqrt{\mu } x\right) +\sqrt{\mu }B_{j}(x)\cos \left(
452: \sqrt{\mu }x\right) \label{ansatz}
453: \end{align}
454: (with $j=1,2$) into Eqs. (\ref{mat2}). Differentiating the first equation of
455: (\ref{ansatz}) and comparing it with the second yields a consistency
456: condition,
457: \[
458: A_{j}^{\prime }\cos (\sqrt{\mu }x)+B_{j}^{\prime }\sin
459: (\sqrt{\mu } x)=0\,,\quad j=1,2\,,
460: \]
461: that must be satisfied for this procedure to be valid. Inserting
462: these equations into Eqs. (\ref{mat2}) yields a set of coupled
463: differential equations for $A_{j}$ and $B_{j}$, whose right-hand
464: sides are expanded as truncated Fourier series to isolate
465: contributions from different harmonics \cite{675,rand}. The
466: leading contribution in these equations is of $ O(\epsilon )$, so
467: the equations assume a general form
468: \begin{align}
469: A_{j}^{\prime }& =\epsilon F_{A_{j}}(A_{1},A_{2},B_{1},B_{2},x)+O(\epsilon
470: ^{2})\,, \nonumber \\
471: B_{j}^{\prime }& =\epsilon F_{B_{j}}(A_{1},A_{2},B_{1},B_{2},x)+O(\epsilon
472: ^{2})\,. \label{preavg}
473: \end{align}
474: When $\epsilon =0$, Eqs. (\ref{mat2}) decompose into two uncoupled harmonic
475: oscillators. We have computed the exact functions $F_{A_{j}}$ and $F_{B_{j}}$
476: in Eqs. (\ref{preavg}) and provide them in the Appendix.
477:
478: Our objective is to isolate the parts of the functions $A_{j}(x)$
479: and $ B_{j}(x)$ that vary slowly in comparison with the fast
480: oscillations of $\cos (\sqrt{\mu }x)$ and $\sin (\sqrt{\mu
481: }x)$ and to derive averaged equations governing their slow
482: evolution. To commence averaging, we decompose $A_{j}$ and $B_{j}$
483: into the sum of slowly varying parts and small rapidly oscillating
484: ones (which are written as power-series expansions in $ \epsilon
485: $):
486: \begin{align}
487: A_{j}& =\bar{A}_{j}+\epsilon
488: W_{A_{j}}(\bar{A}_{1},\bar{A}_{2},\bar{B}_{1},
489: \bar{B}_{2},x)+O(\epsilon ^{2}), \nonumber \\
490: B_{j}& =\bar{B}_{j}+\epsilon
491: W_{B_{j}}(\bar{A}_{1},\bar{A}_{2},\bar{B}_{1},
492: \bar{B}_{2},x)+O(\epsilon ^{2})\,. \label{av}
493: \end{align}
494: Here, the {\it generating functions} $W_{A_{j}}$, $W_{B_{j}}$ are chosen so
495: as to eliminate all the rapidly oscillating terms in Eqs. (\ref{preavg})
496: after the substitution of Eqs. (\ref{av}).
497:
498: This procedure yields evolution equations for the averaged
499: quantities $\bar{A}_{j}$ and $\bar{B}_{j}$\cite{675}, which we
500: henceforth denote simply as $ A_{j}$ and $B_{j}$. (All other terms
501: in the originally defined $A_{j}$ and $ B_{j}$ are cancelled out
502: by the choice of the generating functions.) As we shall see, the
503: slow-flow equations so derived are different in resonant and
504: non-resonant situations.
505:
506:
507: \subsection{The Non-Resonant Case}
508:
509: When $\sqrt{\mu }\neq \kappa $ [recall that $\kappa $ is half the wave
510: number of the OL potential; see Eq. (\ref{OL})], which is the non-resonant
511: case, effective equations governing the slow evolution are
512: \begin{align}
513: A_{1}^{\prime }& =\frac{\epsilon }{\sqrt{\mu }}\left[
514: -\frac{3\tilde{g}}{8
515: }B_{1}(A_{1}^{2}+B_{1}^{2})-\frac{\tilde{\alpha}}{2}B_{2}-\frac{\tilde{h}}{4}
516: A_{1}A_{2}B_{2}-\frac{\tilde{h}}{8}B_{1}(A_{2}^{2}+3B_{2}^{2})\right]
517: +O(\epsilon ^{2})\,, \nonumber \\
518: A_{2}^{\prime }& =\frac{\epsilon }{\sqrt{\mu }}\left[
519: -\frac{3\tilde{g}}{8
520: }B_{2}(A_{2}^{2}+B_{2}^{2})-\frac{\tilde{\alpha}}{2}B_{1}-\frac{\tilde{h}}{4}
521: A_{1}A_{2}B_{1}-\frac{\tilde{h}}{8}B_{2}(A_{1}^{2}+3B_{1}^{2})\right]
522: +O(\epsilon ^{2})\,, \nonumber \\
523: B_{1}^{\prime }& =\frac{\epsilon }{\sqrt{\mu }}\left[
524: \frac{3\tilde{g}}{8}
525: A_{1}(A_{1}^{2}+B_{1}^{2})+\frac{\tilde{\alpha}}{2}A_{2}+\frac{\tilde{h}}{4}
526: A_{2}B_{1}B_{2}+\frac{\tilde{h}}{8}A_{1}(3A_{2}^{2}+B_{2}^{2})\right]
527: +O(\epsilon ^{2})\,, \nonumber \\
528: B_{2}^{\prime }& =\frac{\epsilon }{\sqrt{\mu }}\left[
529: \frac{3\tilde{g}}{8}
530: A_{2}(A_{2}^{2}+B_{2}^{2})+\frac{\tilde{\alpha}}{2}A_{1}+\frac{\tilde{h}}{4}
531: A_{1}B_{1}B_{2}+\frac{\tilde{h}}{8}A_{2}(3A_{1}^{2}+B_{1}^{2})\right]
532: +O(\epsilon ^{2})\,. \label{nonres}
533: \end{align}
534: In this case, the OL does not contribute to $O(\epsilon )$ terms, so the
535: terms explicitly written in Eqs. (\ref{nonres}) correspond to what one would
536: obtain from coupled Duffing equations, as Eqs. (\ref{mat2}) reduce to
537: coupled Duffing oscillators in the absence of the OL potential \cite{param}.
538: These contributions yield the wavenumber-amplitude relations for decoupled
539: condensates,\cite{mapbecprl,mapbec} as well as mode-wavenumber relations
540: produced by the coupling terms \cite{675}.
541:
542: The non-resonant equations (\ref{nonres}) give rise to three types
543: of equilibria, at which $A_{1}^{\prime }=A_{2}^{\prime
544: }=B_{1}^{\prime }=B_{2}^{\prime }=0$: the trivial (zero)
545: equilibrium and those which we will call double modes and
546: quadruple modes. These have, respectively, two and four nonzero
547: amplitudes $A_{j}$, $B_{j}$. Single-mode and triple-mode
548: equilibria do not exist. Different double modes that can be found
549: are $\pi /2 $ phase shifts of each other: these are
550: \textquotedblleft $A_{1}A_{2}$\textquotedblright\ equilibria
551: with $A_{1},\,A_{2}\neq 0$ and $ B_{1}=B_{2}=0 $, and
552: \textquotedblleft $B_{1}B_{2}$\textquotedblright\ ones with
553: $A_{1}=A_{2}=0$ and $B_{1},\,B_{2}\neq 0$.
554:
555: The $A_{1}A_{2}$ equilibria satisfy
556: \begin{equation}
557: A_{1}^{2}=A_{2}^{2}=\mp \frac{4\alpha }{3(g+h)}\,, \label{eqref1}
558: \end{equation}
559: where the signs $-$ and $+$ arise, respectively, when $(g+h)<0$,
560: and $ (g+h)>0 $ (recall that $\alpha >0$). In the former and
561: latter cases, we find that $A_{1}=A_{2}$ and $A_{1}=-A_{2}$,
562: respectively. This yields the following two $A_{1}A_{2}$
563: equilibria:
564: \begin{align}
565: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( \sqrt{\frac{4\alpha
566: }{3(g+h)}},-\sqrt{
567: \frac{4\alpha }{3(g+h)}},0,0\right) ,\quad {\rm if}~~g+h>0\,; \nonumber \\
568: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( \sqrt{\frac{-4\alpha
569: }{3(g+h)}},\sqrt{ \frac{-4\alpha }{3(g+h)}},0,0\right) ,\quad {\rm
570: if}~~g+h<0\,. \label{aa}
571: \end{align}
572: Similar expressions for the $B_{1}B_{2}$ equilibria are obtained
573: by phase-shifting the $A_{1}A_{2}$ modes by $\pi /2$.
574:
575: We have examined the stability of the approximate stationary solutions
576: corresponding to the double-mode equilibria obtained above with direct
577: simulations of the coupled GP equations (\ref{cnls2}). Typically, the
578: simulations generate solutions that oscillate in time (as the initial
579: configurations are not exact stationary solutions) without
580: developing any apparent instability.
581:
582: One can also find four sets of quadruple-mode equilibria in which
583: $ A_{1}^{2}=A_{2}^{2}$ and $B_{1}^{2}=B_{2}^{2}$. In the first two
584: sets, $ A_{2} $ is arbitrary:
585: \begin{align}
586: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( -A_{2},A_{2},\pm
587: \sqrt{-A_{2}^{2}+ \frac{4\alpha }{3(g+h)}},\mp
588: \sqrt{-A_{2}^{2}+\frac{4\alpha }{3(g+h)}}\right)
589: ,~{\rm if}~~g+h>0\,, \nonumber \\
590: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( A_{2},A_{2},\pm
591: \sqrt{-A_{2}^{2}- \frac{4\alpha }{3(g+h)}},\pm
592: \sqrt{-A_{2}^{2}-\frac{4\alpha }{3(g+h)}}\right) ,~{\rm
593: if}~~g+h<0\,. \label{quad1}
594: \end{align}
595: In the second two sets, $B_{2}$ is arbitrary:
596: \begin{align}
597: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( \pm
598: \sqrt{-B_{2}^{2}+\frac{4\alpha } {3(g+h)}},\mp
599: \sqrt{-B_{2}^{2}+\frac{4\alpha }{3(g+h)}},-B_{2},B_{2}\right) ,~
600: {\rm if}~~g+h>0\,, \nonumber \\
601: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( \pm
602: \sqrt{-B_{2}^{2}-\frac{4\alpha } {3(g+h)}},\pm
603: \sqrt{-B_{2}^{2}-\frac{4\alpha }{3(g+h)}},B_{2},B_{2}\right) ,~
604: {\rm if}~~g+h<0\,. \label{quad2}
605: \end{align}
606: Each of the expressions (\ref{quad1}) and (\ref{quad2}) includes four
607: equilibria, as there are two possible choices of the exterior signs. The
608: presence of the arbitrary amplitudes in these expressions means that the
609: quadruple-mode stationary solutions are obtained as rotations
610: of the above double-mode ones given, respectively, by Eqs. (\ref{aa}) and by
611: those same equations with an additional $\pi /2$ phase shift. Accordingly,
612: direct simulations of Eqs. (\ref{cnls2}) starting with the approximate
613: quadruple-mode stationary states reveal only oscillations but no instability
614: growth, just as with simulations initiated by the approximate dual-mode
615: stationary solutions.
616:
617:
618: \subsection{Subharmonic Resonances}
619:
620: The most fundamental spatial resonance is a subharmonic one, of
621: type $ 2\!:\!1\!:\!1$\cite{675,rand,gucken}. In this
622: situation, the parameter $ \mu $ from the initial plane-wave
623: approximation (\ref{plane}) [recall that $\mu _{1}=\mu
624: _{2}\equiv \mu $] is of the form
625: \begin{equation}
626: \mu =\kappa ^{2}+\epsilon \tilde{\mu}_{1}+O(\epsilon ^{2})\,,
627: \label{detune}
628: \end{equation}
629: where $\epsilon \tilde{\mu}_{1}$ is the {\it detuning}
630: constant\cite {rand,675,param}. [Recall that $\epsilon $ is a small
631: parameter; we assume $ \tilde{\mu}_{1}=O(1)$.] In this
632: situation, new terms occur in Eqs. (\ref {nonres}). This leads to
633: equations that include a contribution from the OL potential,
634: \begin{align}
635: A_{1}^{\prime }& =\frac{\epsilon }{\kappa }\left[ \left(
636: \frac{\tilde{\mu} _{1}}{2}-\frac{\tilde{V}_{0}}{4}\right)
637: B_{1}-\frac{3\tilde{g}}{8}
638: B_{1}(A_{1}^{2}+B_{1}^{2})-\frac{\tilde{\alpha}}{2}B_{2}-\frac{\tilde{h}}{4}
639: A_{1}A_{2}B_{2}-\frac{\tilde{h}}{8}B_{1}(A_{2}^{2}+3B_{2}^{2})\right]
640: +O(\epsilon ^{2})\,, \nonumber \\
641: A_{2}^{\prime }& =\frac{\epsilon }{\kappa }\left[ \left(
642: \frac{\tilde{\mu} _{1}}{2}-\frac{\tilde{V}_{0}}{4}\right)
643: B_{2}-\frac{3\tilde{g}}{8}
644: B_{2}(A_{2}^{2}+B_{2}^{2})-\frac{\tilde{\alpha}}{2}B_{1}-\frac{\tilde{h}}{4}
645: A_{1}A_{2}B_{1}-\frac{\tilde{h}}{8}B_{2}(A_{1}^{2}+3B_{1}^{2})\right]
646: +O(\epsilon ^{2})\,, \nonumber \\
647: B_{1}^{\prime }& =\frac{\epsilon }{\kappa }\left[ -\left(
648: \frac{\tilde{\mu }_{1}}{2}+\frac{\tilde{V}_{0}}{4}\right)
649: A_{1}+\frac{3\tilde{g}}{8}
650: A_{1}(A_{1}^{2}+B_{1}^{2})+\frac{\tilde{\alpha}}{2}A_{2}+\frac{\tilde{h}}{4}
651: A_{2}B_{1}B_{2}+\frac{\tilde{h}}{8}A_{1}(3A_{2}^{2}+B_{2}^{2})\right]
652: +O(\epsilon ^{2})\,, \nonumber \\
653: B_{2}^{\prime }& =\frac{\epsilon }{\kappa }\left[ -\left(
654: \frac{\tilde{\mu }_{1}}{2}+\frac{\tilde{V}_{0}}{4}\right)
655: A_{2}+\frac{3\tilde{g}}{8}
656: A_{2}(A_{2}^{2}+B_{2}^{2})+\frac{\tilde{\alpha}}{2}A_{1}+\frac{\tilde{h}}{4}
657: A_{1}B_{1}B_{2}+\frac{\tilde{h}}{8}A_{2}(3A_{1}^{2}+B_{1}^{2})\right]
658: +O(\epsilon ^{2})\,. \label{res}
659: \end{align}
660:
661: Equations (\ref{res}) have three types of equilibria when $\alpha \neq 0$:
662: the trivial one, double modes, and quadruple modes. When $\alpha =0$, we
663: also find single-mode equilibria and extra double-mode ones. Triple-mode
664: stationary solutions never appear. All the equilibria of Eqs. (\ref{res}),
665: except for the trivial one, correspond to spatially periodic stationary
666: solutions of the underlying system (\ref{mat2}).
667:
668: There are two kinds of $A_{1}A_{2}$ (double-mode) equilibria. The first
669: satisfies $A_{1}^{2}=A_{2}^{2}$, so that the two components have equal
670: amplitudes:
671: \begin{align}
672: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( \sqrt{\frac{-4\alpha +2(2\mu
673: _{1}+V_{0})}{3(g+h)}},\sqrt{\frac{-4\alpha +2(2\mu _{1}+V_{0})}{3(g+h)}} ,0,0\right) \,, \nonumber \\
674: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( \sqrt{\frac{4\alpha
675: +2(2\mu _{1}+V_{0})}{3(g+h)}},-\sqrt{\frac{4\alpha +2(2\mu
676: _{1}+V_{0})}{3(g+h)}} ,0,0\right) \,. \label{a1a2type1}
677: \end{align}
678: A crucial issue is the dynamical stability of
679: these solutions, which we tested with direct simulations of the
680: underlying equations (\ref{cnls2}). We found that they are {\em
681: stable}, as exemplified in Fig. \ref{mpfig3} for $\kappa =\mu
682: =1$.
683:
684:
685: \begin{figure}[tbp]
686: {\epsfig{file=mpnew3.ps, width=7.5cm,angle=0, clip=}}
687: {\epsfig{file=mpnew3a.ps, width=7.5cm,angle=0, clip=}}
688: \caption{An
689: example of evolution of the $A_{1}A_{2}$ double mode with {\it
690: equal} amplitudes $\left\vert A_{1}\right\vert $ and $\left\vert
691: A_{2}\right\vert $ in the case of the $2\!:\!1\!:\!1$ subharmonic
692: resonance, constructed as per Eq. (\protect\ref{a1a2type1}) for
693: $\protect\kappa = \protect\mu =1$, $V_{0}=0.1$, $g=h=0.025$,
694: and $\protect\alpha =0.02$. The left subplot shows the
695: spatio-temporal evolution of $\left\vert \psi _{1}\right\vert
696: ^{2}$ by means of grayscale contour plots ($\left\vert
697: \protect\psi _{2}\right\vert ^{2}$ behaves similarly). The right
698: subplot displays snapshots of the field $\left\vert \protect\psi
699: _{1}\right\vert ^{2} $ for $t=99$ (upper panel) and $t=105$ (lower
700: panel). In these panels, the optical-lattice potential is
701: also shown by a dashed line. The results have been obtained
702: through numerical integration of Eqs. (\protect \ref{cnls2}) in
703: time.} \label{mpfig3}
704: \end{figure}
705:
706:
707: The other $A_{1}A_{2}$ double-mode equilibrium has
708: {\it unequal} components $A_{1}$ and $A_{2}$ [note that in the
709: non-resonant case considered above, the double-mode equilibria,
710: which are given by Eqs. (\ref {aa}) and by a $\pi /2$ phase shift
711: thereof, always have equal nonzero components]:
712: \[
713: A_{1}^{2}+A_{2}^{2}=\frac{2(2\mu _{1}+V_{0})}{3g}>0\,,
714: \]
715: \begin{align}
716: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( \sqrt{\frac{2\mu
717: _{1}+V_{0}}{3g} \pm \frac{1}{3}\sqrt{\frac{(2\mu
718: _{1}+V_{0})^{2}}{g^{2}}-\frac{16\alpha
719: ^{2}}{(g-h)^{2}}}}\,,\right. \nonumber \\
720: & \qquad \left. \sqrt{\frac{2\mu _{1}+V_{0}}{3g}\mp
721: \frac{1}{3}\sqrt{ \frac{(2\mu
722: _{1}+V_{0})^{2}}{g^{2}}-\frac{16\alpha ^{2}}{(g-h)^{2}}}}
723: ,0,0\right) \,, \label{unequal}
724: \end{align}
725: where the interior $+$ sign in the first component corresponds to the $-$
726: sign in the second, and vice versa. The exterior sign $\pm $ is independent
727: of the interior one. A necessary condition for the existence of this
728: solution is
729: \[
730: \left\vert \frac{2\mu _{1}+V_{0}}{g}\right\vert \geq
731: \frac{4\alpha }{|g-h| }\,.
732: \]
733: In particular, when $h=2g$, which is a case of special physical relevance
734: (as explained above), the solution becomes
735: \begin{align}
736: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( \sqrt{\frac{1}{3g}\left[
737: 2\mu _{1}+V_{0}\pm \sqrt{(2\mu _{1}+V_{0})^{2}-16\alpha
738: ^{2}}\right] }
739: \,,\right. \\
740: & \qquad \left. \sqrt{\frac{1}{3g}\left[ 2\mu _{1}+V_{0}\mp
741: \sqrt{ (2\mu _{1}+V_{0})^{2}-16\alpha ^{2}}\right] },0,0\right)
742: \,,
743: \end{align}
744: provided $|2\mu _{1}+V_{0}|\geq 4\alpha $.
745:
746: In fact, the existence of pairs of equilibria in which the two
747: components have unequal amplitudes that are mirror images of each
748: other is a manifestation of {\it spontaneous symmetry breaking} in
749: the present model, which is described by the symmetric system of
750: coupled equations (\ref{cnls2}). A similar phenomenon was studied
751: in detail (in terms of soliton solutions) in the aforementioned
752: model of dual-core nonlinear optical fibers, which includes only
753: linear coupling between two equations \cite {UNSW}.
754:
755: The stability of the asymmetric stationary solutions, which
756: correspond to equilibria with unequal components, was also
757: simulated in the framework of Eqs. (\ref{cnls2}). We show the results
758: of a typical simulation in Fig. \ref {mpfig4}. As seen in
759: the figure, these states are subject to
760: %{\em modulational
761: %instability}, leading to the onset of
762: periodic oscillations between the two
763: components (which is possible only in the presence of
764: the linear coupling between them).
765:
766:
767: \begin{figure}[tbp]
768: {\epsfig{file=mpsub4b.ps, width=7.5cm,angle=0, clip=}}
769: {\epsfig{file=mpsub4c.ps, width=7.5cm,angle=0, clip=}}
770: \caption{The same as in Fig. \protect\ref{mpfig3}, but for the
771: $A_{1}A_{2}$ double mode with {\it unequal} amplitudes $\left\vert
772: A_{1}\right\vert $ and $\left\vert A_{2}\right\vert $, given by
773: Eq. (\protect\ref{unequal}). The top panel in the right subplot
774: shows the time evolution of the numbers of atoms in the two
775: components, $N_{1,2}=\protect\int |\psi_{1,2}|^{2}dx$, by dashed and
776: dash-dotted lines (the sum of the two, $N=N_{1}+N_{2}$, is shown
777: by the solid line). The right panel also shows the fields
778: $\left\vert \protect\psi _{1}\right\vert ^{2}$ and $\left\vert
779: \protect\psi_{2}\right\vert ^{2}$ (by solid and dash-dotted lines,
780: respectively) at $ t=105$ and $t=210$. The OL potential is shown
781: by the dashed line. Oscillations of matter between the two
782: components are clearly discernible. The parameters are $g=0.025$,
783: $h=0.005$, $\protect\alpha =-0.02$, $V_{0}=0.3$, and
784: $\protect\kappa =\protect\mu =1$. (Recall that the sign of $\alpha$
785: can be chosen arbitrarily.)} \label{mpfig4}
786: \end{figure}
787:
788:
789: The resonant equations (\ref{res}) give rise to two types of $B_{1}B_{2}$
790: dual-mode equilibria. The first satisfies $B_{1}^{2}=B_{2}^{2}$ and
791: \begin{align}
792: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( 0,0,\sqrt{\frac{-4\alpha +2(2\mu
793: _{1}-V_{0})}{3(g+h)}},\sqrt{\frac{-4\alpha +2(2\mu _{1}-V_{0})}{3(g+h)}} \right) \,, \nonumber \\
794: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( 0,0,\sqrt{\frac{4\alpha
795: +2(2\mu _{1}-V_{0})}{3(g+h)}},-\sqrt{\frac{4\alpha +2(2\mu
796: _{1}-V_{0})}{3(g+h)}} \right) \,. \label{b1b2type1}
797: \end{align}
798: The second type satisfies
799: \[
800: B_{1}^{2}+B_{2}^{2}=\frac{2(2\mu _{1}-V_{0})}{3g}>0\,,
801: \]
802: \begin{align}
803: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( 0,0,\sqrt{\frac{2\mu
804: _{1}-V_{0}} {3g}\pm \frac{1}{3}\sqrt{\frac{(2\mu
805: _{1}-V_{0})^{2}}{g^{2}}-\frac{16\alpha
806: ^{2}}{(g-h)^{2}}}},\right. \nonumber \\
807: & \qquad \left. \sqrt{\frac{2\mu _{1}-V_{0}}{3g}\mp
808: \frac{1}{3}\sqrt{ \frac{(2\mu
809: _{1}-V_{0})^{2}}{g^{2}}-\frac{16\alpha ^{2}}{(g-h)^{2}}}} \right)
810: \,. \label{unequalBB}
811: \end{align}
812: As above, the interior $+$ sign in the first component is paired to the $-$
813: sign in the second, and vice versa, whereas the exterior $\pm $ is
814: independent. A necessary condition for the existence of this solution is
815: \begin{equation}
816: \left\vert \frac{2\mu _{1}-V_{0}}{g}\right\vert \geq
817: \frac{4\alpha }{|g-h| }\,. \label{cond}
818: \end{equation}
819: When $h=2g$, the present solution becomes
820: \begin{align}
821: (A_{1},A_{2},B_{1},B_{2})& =\pm \left( 0,0,\sqrt{\frac{1}{3g}\left[ 2\mu
822: _{1}-V_{0}\pm \sqrt{(2\mu _{1}-V_{0})^{2}-16\alpha ^{2}}\right] },\right.
823: \\
824: & \qquad \left. \sqrt{\frac{1}{3g}\left[ 2\mu _{1}-V_{0}\mp
825: \sqrt{ (2\mu _{1}-V_{0})^{2}-16\alpha ^{2}}\right] }\right) \,,
826: \end{align}
827: provided $|2\mu _{1}-V_{0}|\geq 4\alpha $.
828:
829: Unlike the non-resonant case, the resonant $B_{1}B_{2}$ modes are {\em
830: not} precise phase shifts of the $A_{1}A_{2}$ modes,
831: as the
832: %as the
833: %OL breaks the symmetry that causes this
834: spatial parametric excitation resulting from the OL has only the
835: cosine harmonic. Nevertheless, the equations describing these two
836: classes of modes are similar, differing only in the sign of $V_{0}$.
837: Direct simulations demonstrate that the stability of stationary solutions
838: corresponding to the $B_{1}B_{2}$ equilibria is the same as in the
839: case of the $A_{1}A_{2}$ double-mode equilibria considered above:
840: the symmetric ones with $\left\vert B_{1}\right\vert =\left\vert
841: B_{2}\right\vert $ are {\it stable}, and the asymmetric solutions
842: with $\left\vert B_{1}\right\vert \neq \left\vert B_{2}\right\vert
843: $ are {\it unstable}.
844:
845: We have also found two sets of quadruple modes in the resonant case. The
846: first set satisfies $B_{1}=B_{2}$, $A_{1}=-A_{2}$, and
847: \begin{align}
848: A_{1}^{2}& =\frac{V_{0}}{2h}+\frac{\alpha }{h}+\frac{2\mu _{1}}{3g+h}, \\
849: B_{1}^{2}& =-\frac{V_{0}}{2h}-\frac{\alpha }{h}+\frac{2\mu _{1}}{3g+h}\,.
850: \end{align}
851: A necessary condition for its existence is
852: \[
853: \frac{2\mu _{1}}{3g+h}>\left\vert \frac{V_{0}}{2h}+\frac{\alpha
854: }{h} \right\vert \,,
855: \]
856: and hence it is necessary that $\mu _{1}/(3g+h)>0$. The second set of
857: quadruple modes satisfies $B_{1}=-B_{2}$, $A_{1}=A_{2}$, and
858: \begin{align}
859: A_{1}^{2}& =\frac{V_{0}}{2h}-\frac{\alpha }{h}+\frac{2\mu _{1}}{3g+h}\,,
860: \\
861: B_{1}^{2}& =-\frac{V_{0}}{2h}+\frac{\alpha }{h}+\frac{2\mu _{1}}{3g+h}\,.
862: \end{align}
863: A necessary existence condition for this mode to exist is
864: \[
865: \frac{2\mu _{1}}{3g+h}>\left\vert \frac{V_{0}}{2h}-\frac{\alpha
866: }{h} \right\vert \,,
867: \]
868: which also implies that $\mu _{1}/(3g+h)>0$.
869:
870: We considered quadruple modes in the presence of detuning, so
871: $ \mu _{1}\neq 0$. This is rather difficult to implement
872: numerically, as---in view of the periodic boundary conditions in
873: $x$ employed in the numerical integration scheme---it is necessary
874: to match {\it both} the potential and initial condition to the
875: size of the integration domain. Nevertheless, we were able to
876: perform stability simulations in this case too. We show an example of
877: these simulations in Fig. \ref{mpfig5}. We observe that the quadruple
878: mode is {\it unstable} against long-wave perturbations, even if
879: the simulations are run with $\alpha =0$ (no linear coupling). We
880: have not observed stable quadruple states.
881:
882:
883: \begin{figure}[tbp]
884: \centerline{ {\epsfig{file=mpnew5.ps,width=7.5cm,angle=0, clip=}}}
885: \caption{A typical example of the long-wave instability of a
886: quadruple (four-amplitude) stationary state in the resonant case.
887: The parameters are $ \protect\mu =1.5791$, $\protect\kappa
888: =1.2320$, $V_{0}=0.1$, $g=0.025$, $ h=0.05$ and $\protect\alpha
889: =0$, and the size of the integration box is $ L=255$.}
890: \label{mpfig5}
891: \end{figure}
892:
893:
894: When $\alpha =0$ (no linear coupling between BEC components), one
895: can find additional double-mode equilibria and four single-mode
896: ones, the latter of which take the form $(A_{1},0,0,0)$,
897: $(0,A_{2},0,0)$, $(0,0,B_{1},0)$, $ (0,0,0,B_{2})$, with
898: \begin{align}
899: A_{1}^{2}& =A_{2}^{2}=\frac{2(2\mu _{1}+V_{0})}{3g}>0\,, \\
900: B_{1}^{2}& =B_{2}^{2}=-\frac{2(2\mu _{1}-V_{0})}{3g}>0\,.
901: \end{align}
902: The $A_{j}$- and $B_{j}$-modes both exist when $V_{0}/g>0$. In
903: this case ($\alpha =0$), matter cannot be exchanged between the
904: components. In this same situation, there is also an $A_{1}B_{2}$
905: double-mode equilibrium [of the form $ (A_{1},0,0,B_{2})$], which
906: satisfies
907: \begin{align}
908: A_{1}^{2}& =\frac{4\mu _{1}}{3g+h}+\frac{2V_{0}}{3g-h}>0\,, \nonumber \\
909: B_{2}^{2}& =\frac{4\mu _{1}}{3g+h}-\frac{2V_{0}}{3g-h}>0\,. \label{a1b2}
910: \end{align}
911: From Eq. (\ref{a1b2}), it follows that a necessary condition for
912: this mode to exist is $8\mu _{1}/(3g+h)>0$. Its counterpart is
913: an $A_{2}B_{1}$ equilibrium, in which the subscripts $1$ and $2$
914: are swapped in Eq. (\ref {a1b2}).
915:
916: One can extend the analysis to higher-order spatial resonances in BECs (from
917: the lowest subharmonic resonance considered here) either by considering
918: higher-order corrections to the averaged equations, or by employing a
919: perturbative scheme based on elliptic functions, as has been done for
920: single-component BECs in OLs\cite{mapbec,mapbecprl}. Toward this aim, it may
921: be fruitful to utilize an action-angle formulation and the elliptic-function
922: structure of solutions to Eqs. (\ref{mat}) when $V_{0}=0$. However, detailed
923: consideration of higher-order resonances is beyond the scope of this work.
924:
925:
926: \section{Ternary BECs in Optical Lattices}
927:
928: To evince the generality of the above analysis, we briefly
929: consider its extension to a BEC model of three hyperfine states
930: coupled by two different microwave fields, which is
931: also a physically relevant situation \cite{pengels}. The corresponding
932: coupled GP equations (with $\hbar =1$ and $ m=1/2$) are
933: \begin{align}
934: i\frac{\partial \psi _{1}}{\partial t}& =-\nabla ^{2}\psi _{1}+g|\psi
935: _{1}|^{2}\psi _{1}+V(x)\psi _{1}+h_{12}|\psi _{2}|^{2}\psi _{1}+h_{13}|\psi
936: _{3}|^{2}\psi _{1}+\alpha _{12}\psi _{2}+\alpha _{13}\psi _{3}\,, \nonumber
937: \\
938: i\frac{\partial \psi _{2}}{\partial t}& =-\nabla ^{2}\psi _{2}+g|\psi
939: _{2}|^{2}\psi _{2}+V(x)\psi _{2}+h_{12}|\psi _{1}|^{2}\psi _{2}+h_{23}|\psi
940: _{3}|^{2}\psi _{2}+\alpha _{12}\psi _{1}+\alpha _{23}\psi _{3}\,, \nonumber
941: \\
942: i\frac{\partial \psi _{3}}{\partial t}& =-\nabla ^{2}\psi _{3}+g|\psi
943: _{3}|^{3}\psi _{3}+V(x)\psi _{3}+h_{13}|\psi _{1}|^{2}\psi _{3}+h_{23}|\psi
944: _{2}|^{2}\psi _{3}+\alpha _{13}\psi _{1}+\alpha _{23}\psi _{2}\,,
945: \label{cnls3}
946: \end{align}
947: where the self- and cross-scattering coefficients are
948: $g := g_{1} = g_{2} = g_{3}$ and $h_{jk}$, and the linear coupling constants
949: are $\alpha _{jk}$.
950:
951: As in the binary case, we start with the general form
952: (\ref{coher}) for stationary solutions, with $\theta
953: _{1}(x)=\theta _{2}(x)=\theta _{3}(x)\equiv \theta (x)$ and
954: $\mu _{1}=\mu _{2}=\mu _{3}\equiv \mu $. Then, as done above,
955: we set $c_{j}=0$ (i.e., $\theta =0$) in Eqs. (\ref{strange}) to consider
956: standing wave solutions and arrive at the following equations:
957: \begin{align}
958: R_{1}^{\prime }& =S_{1}\,, \nonumber \\
959: S_{1}^{\prime }& =-\mu
960: R_{1}+gR_{1}^{3}+h_{12}R_{1}R_{2}^{2}+h_{13}R_{1}R_{3}^{2}+\alpha
961: _{12}R_{2}+\alpha _{13}R_{3}+V(x)R_{1}\,, \nonumber \\
962: R_{2}^{\prime }& =S_{2}\,, \nonumber \\
963: S_{2}^{\prime }& =-\mu
964: R_{2}+gR_{2}^{3}+h_{12}R_{1}^{2}R_{2}+h_{23}R_{2}R_{3}^{2}+\alpha
965: _{12}R_{1}+\alpha _{23}R_{3}+V(x)R_{2}\,, \nonumber \\
966: R_{3}^{\prime }& =S_{3}\,, \nonumber \\
967: S_{3}^{\prime }& =-\mu
968: R_{3}+gR_{3}^{3}+h_{13}R_{1}^{2}R_{3}+h_{23}R_{2}^{2}R_{3}+\alpha
969: _{13}R_{1}+\alpha _{23}R_{2}+V(x)R_{3}\,, \label{mat3}
970: \end{align}
971: where $V(x)$ is the sinusoidal OL potential, as before.
972:
973: One can average Eqs. (\ref{mat3}) with the same procedure that we applied to
974: Eqs. (\ref{mat}) and thereby derive both resonant and non-resonant equations
975: describing the system's slow dynamics. In particular, for the most
976: fundamental resonant case (the lowest-order, $2\!:\!1\!:\!1\!:\!1$,
977: resonance), the averaged equations are
978: \begin{align}
979: A_{1}^{\prime }& =\frac{\epsilon }{\kappa }\left[ \left(
980: \frac{\tilde{\mu} _{1}}{2}-\frac{\tilde{V}_{0}}{4}\right)
981: B_{1}-\frac{3\tilde{g}}{8}
982: B_{1}(A_{1}^{2}+B_{1}^{2})-\frac{\tilde{\alpha}_{12}}{2}B_{2}-\frac{\tilde{
983: \alpha}_{13}}{2}B_{3}-\frac{\tilde{h}_{12}}{4}A_{1}A_{2}B_{2}-\frac{\tilde{h} _{13}}{4}A_{1}A_{3}B_{3}\right. \nonumber \\
984: & \qquad \left.
985: -\frac{\tilde{h}_{12}}{8}B_{1}(A_{2}^{2}+3B_{2}^{2})-\frac{
986: \tilde{h}_{13}}{8}B_{1}(A_{3}^{2}+3B_{3}^{2})\right] +O(\epsilon
987: ^{2})\,,
988: \nonumber \\
989: A_{2}^{\prime }& =\frac{\epsilon }{\kappa }\left[ \left(
990: \frac{\tilde{\mu} _{1}}{2}-\frac{\tilde{V}_{0}}{4}\right)
991: B_{2}-\frac{3\tilde{g}}{8}
992: B_{2}(A_{2}^{2}+B_{2}^{2})-\frac{\tilde{\alpha}_{12}}{2}B_{1}-\frac{\tilde{
993: \alpha}_{23}}{2}B_{3}-\frac{\tilde{h}_{12}}{4}A_{1}A_{2}B_{1}-\frac{\tilde{h} _{23}}{4}A_{2}A_{3}B_{3}\right. \nonumber \\
994: & \qquad \left.
995: -\frac{\tilde{h}_{12}}{8}B_{2}(A_{1}^{2}+3B_{1}^{2})-\frac{
996: \tilde{h}_{23}}{8}B_{2}(A_{3}^{2}+3B_{3}^{2})\right] +O(\epsilon
997: ^{2})\,,
998: \nonumber \\
999: A_{3}^{\prime }& =\frac{\epsilon }{\kappa }\left[ \left(
1000: \frac{\tilde{\mu} _{1}}{2}-\frac{\tilde{V}_{0}}{4}\right)
1001: B_{3}-\frac{3\tilde{g}}{8}
1002: B_{3}(A_{3}^{2}+B_{3}^{2})-\frac{\tilde{\alpha}_{13}}{2}B_{1}-\frac{\tilde{
1003: \alpha}_{23}}{2}B_{2}-\frac{\tilde{h}_{13}}{4}A_{1}A_{3}B_{1}-\frac{\tilde{h} _{23}}{4}A_{2}A_{3}B_{2}\right. \nonumber \\
1004: & \qquad \left.
1005: -\frac{\tilde{h}_{13}}{8}B_{3}(A_{1}^{2}+3B_{1}^{2})-\frac{
1006: \tilde{h}_{23}}{8}B_{3}(A_{2}^{2}+3B_{2}^{2})\right] +O(\epsilon
1007: ^{2})\,,
1008: \nonumber \\
1009: B_{1}^{\prime }& =\frac{\epsilon }{\kappa }\left[ -\left(
1010: \frac{\tilde{\mu }_{1}}{2}+\frac{\tilde{V}_{0}}{4}\right)
1011: A_{1}+\frac{3\tilde{g}}{8}
1012: A_{1}(A_{1}^{2}+B_{1}^{2})+\frac{\tilde{\alpha}_{12}}{2}A_{2}+\frac{\tilde{
1013: \alpha}_{13}}{2}A_{3}+\frac{\tilde{h}_{12}}{4}A_{2}B_{1}B_{2}+\frac{\tilde{h} _{13}}{4}A_{3}B_{1}B_{3}\right. \nonumber \\
1014: & \qquad \left.
1015: +\frac{\tilde{h}_{12}}{8}A_{1}(3A_{2}^{2}+B_{2}^{2})+\frac{
1016: \tilde{h}_{13}}{8}A_{1}(3A_{3}^{2}+B_{3}^{2})\right] +O(\epsilon
1017: ^{2})\,,
1018: \nonumber \\
1019: B_{2}^{\prime }& =\frac{\epsilon }{\kappa }\left[ -\left(
1020: \frac{\tilde{\mu }_{1}}{2}+\frac{\tilde{V}_{0}}{4}\right)
1021: A_{2}+\frac{3\tilde{g}}{8}
1022: A_{2}(A_{2}^{2}+B_{2}^{2})+\frac{\tilde{\alpha}_{12}}{2}A_{1}+\frac{\tilde{
1023: \alpha}_{23}}{2}A_{3}+\frac{\tilde{h}_{12}}{4}A_{1}B_{1}B_{2}+\frac{\tilde{h} _{23}}{4}A_{3}B_{2}B_{3}\right. \nonumber \\
1024: & \qquad \left.
1025: +\frac{\tilde{h}_{12}}{8}A_{2}(3A_{1}^{2}+B_{1}^{2})+\frac{
1026: \tilde{h}_{23}}{8}A_{2}(3A_{3}^{2}+B_{3}^{2})\right] +O(\epsilon
1027: ^{2})\,,
1028: \nonumber \\
1029: B_{3}^{\prime }& =\frac{\epsilon }{\kappa }\left[ -\left(
1030: \frac{\tilde{\mu }_{1}}{2}+\frac{\tilde{V}_{0}}{4}\right)
1031: A_{3}+\frac{3\tilde{g}}{8}
1032: A_{3}(A_{3}^{2}+B_{3}^{2})+\frac{\tilde{\alpha}_{13}}{2}A_{1}+\frac{\tilde{
1033: \alpha}_{23}}{2}A_{2}+\frac{\tilde{h}_{13}}{4}A_{1}B_{1}B_{3}+\frac{\tilde{h} _{23}}{4}A_{2}B_{2}B_{3}\right. \nonumber \\
1034: & \qquad \left.
1035: +\frac{\tilde{h}_{13}}{8}A_{3}(3A_{1}^{2}+B_{1}^{2})+\frac{
1036: \tilde{h}_{23}}{8}A_{3}(3A_{2}^{2}+B_{2}^{2})\right] +O(\epsilon
1037: ^{2})\,. \label{res3}
1038: \end{align}
1039:
1040: One can find double-mode solutions to (\ref{res3}) that are
1041: analogous to those of Eq. (\ref{res}). For example, if $|\alpha
1042: _{13}|=|\alpha _{23}|$, so that the first and second components in
1043: the ternary condensate have the same strength in their linear
1044: coupling to the third component, there exists a double-mode
1045: equilibrium with $A_{1}^{2}=A_{2}^{2}$ and $
1046: A_{3}=B_{1}=B_{2}=B_{3}=0$. The values of $A_{1}$ and $A_{2}$ are
1047: exactly as for binary BECs [see Eq. (\ref{a1a2type1})], except
1048: that $\alpha $ and $h $ in the solution are replaced by $\alpha
1049: _{12}$ and $h_{12}$. Further, in this case, one finds
1050: $A_{1}=-A_{2}$ for $\alpha _{13}=\alpha _{23}$ and $ A_{1}=A_{2}$
1051: for $\alpha _{13}=-\alpha _{23}$. In fact, these modes are a
1052: straightforward extension of their two-component counterparts, as
1053: the third component is absent in the stationary solution.
1054: Furthermore, the stability of the symmetric double-mode
1055: equilibria, reported above, ensures the stability of these
1056: solutions in the ternary model.
1057:
1058: The situation is more interesting for {\em asymmetric} two-mode
1059: solutions, such as the ones corresponding to Eq. (\ref{unequal}),
1060: which are, simultaneously, solutions to Eqs. (\ref{cnls3}) with
1061: $A_{3}=B_{3}=0$, provided $ A_{1}=-(\alpha _{23}/\alpha
1062: _{13})A_{2}$. [Note that this relation is used to determine
1063: $\alpha_{23}/\alpha_{13}$, as $A_1$ and $A_2$ are determined from
1064: Eq. (\ref{unequal}).] Direct simulations of the three-component GP
1065: equations (\ref{cnls3}) with $h := h_{12} = h_{13} = h_{23}$ show that these
1066: asymmetric solutions are unstable,
1067: just as in the two-component model. The instability
1068: development, illustrated by Fig. \ref{mpfig6}, leads to an
1069: interesting dynamical interplay between the components. In
1070: particular, as a result of the instability, the third component is
1071: eventually excited, which leads to periodic oscillation of matter
1072: between all three components; i.e., in this case, we observe a
1073: true example of three-component dynamics.
1074:
1075:
1076: \begin{figure}[tbp]
1077: {\epsfig{file=mpsub6.ps, width=7.5cm,angle=0, clip=}}
1078: {\epsfig{file=mpsub6a.ps,width=7.5cm,angle=0, clip=}}
1079: \caption{The same as in Fig. \protect\ref{mpfig4}, but for the
1080: three-component (ternary) model. The bottom panel of the left plot
1081: shows the spatio-temporal evolution of the third component (which
1082: is absent in the unperturbed, unstable two-mode solution), and
1083: the thick solid line on the right shows the evolution of the
1084: number of atoms in this component (top subplot). Its spatial
1085: profile is shown as well (middle and bottom subplots, for $ t=35.1$
1086: and $t=70.2$, respectively). Periodic oscillations of matter
1087: between {\em all three components} are evident, cf. the
1088: oscillations in the two-compoment model, displayed in Fig.
1089: \protect\ref{mpfig4}. The parameters are $g=0.025$, $h=0.005$,
1090: $\protect\alpha _{12}=\protect\alpha _{13}=-0.02$, $V_{0}=0.3$,
1091: and $\protect\kappa =\protect\mu =1$. The quantity $\protect\alpha
1092: _{23}\approx 0.117$ is determined from the values of $\protect\alpha
1093: _{12}=\protect\alpha _{13}=-0.02$ (see text).} \label{mpfig6}
1094: \end{figure}
1095:
1096:
1097: \section{Conclusions}
1098:
1099: In this work, we analyzed spatial structures in coupled Gross-Pitaevskii
1100: (coupled GP) equations, which include both nonlinear and linear
1101: interactions, in an optical-lattice (OL) potential. The model describes a
1102: BEC consisting of a mixture of two different hyperfine states of one atomic
1103: species, which are linearly coupled by a resonant electromagnetic field. In
1104: the absence of the OL, we found plane-wave solutions and examined their
1105: stability. In the presence of the OL, we derived a system of averaged
1106: equations to describe a spatially modulated state which is coupled to the
1107: periodic potential through a subharmonic resonance. We found equilibria of
1108: the latter system and examined the stability of the corresponding spatially
1109: periodic solutions to the coupled GP equations using direct simulations.
1110: We demonstrated that symmetric dual-mode resonant states with two equal
1111: amplitudes are stable, whereas asymmetric ones (with unequal amplitudes) are
1112: unstable, generating solutions that oscillate periodically in time. The latter
1113: type of dynamical behavior is only possible in the presence of linear
1114: coupling between BEC components. We found four-mode stationary solutions as
1115: well, but they are always unstable. Finally, a three-component generalization
1116: of the model was introduced and briefly considered. In this case, we found
1117: that the unstable asymmetric two-mode solution, with one component originally
1118: empty, develops time-periodic oscillations in which the initially empty
1119: component becomes populated.
1120:
1121:
1122: \section*{Acknowledgements}
1123:
1124: We appreciate useful discussions with Bernard Deconinck, Alex Kuzmich,
1125: Alexandru Nicolin, and Richard Rand. P.G.K. gratefully acknowledges support
1126: from NSF-DMS-0204585, from the Eppley Foundation for Research and from an
1127: NSF-CAREER award. The work of B.A.M. was supported in a part by the grant
1128: No. 8006/03 from the Israel Science Foundation.
1129:
1130:
1131: \section*{Appendix}
1132:
1133: The functions $F_{A_{j}}$ and $F_{B_{j}}$, which appear in Eqs.
1134: (\ref{preavg}), can be written as a sum of harmonic
1135: contributions. To simplify the notation, we write $\tilde{g}$,
1136: $\tilde{h}$, $\tilde{\alpha}$, and $\tilde{V} _{0}$ simply as $g$,
1137: $h$, $\alpha $, and $V$.
1138:
1139: In the non-resonant case, $F_{A_{1}}=G_{A_{1}}/\sqrt{\mu }$, where
1140: \begin{align}
1141: G_{A_{1}}(A_{1},A_{2},A_{3},A_{4},x)& =\left[ -\frac{\alpha }{2}-\frac{3g}{8} B_{1}(A_{1}^{2}+B_{1}^{2})-\frac{h}{4}A_{1}A_{2}B_{2}-\frac{h}{8} B_{1}(A_{2}^{2}+3B_{2}^{2})\right] \nonumber \\
1142: & \quad +\left[ -\frac{\alpha
1143: }{2}-\frac{g}{4}A_{1}(A_{1}^{2}+3B_{1}^{2})-
1144: \frac{h}{2}A_{2}B_{1}B_{2}-\frac{h}{4}A_{1}(A_{2}^{2}+B_{2}^{2})\right]
1145: \sin
1146: (2\sqrt{\mu }x) \nonumber \\
1147: & \quad +\left[ \frac{g}{8}A_{1}(3B_{1}^{2}-A_{1}^{2})+\frac{h}{4}
1148: A_{1}B_{1}B_{2}+\frac{h}{8}A_{1}(B_{2}^{2}-A_{2}^{2})\right] \sin
1149: (4\sqrt{
1150: \mu }x) \nonumber \\
1151: & \quad +\left[ -\frac{V}{4}A_{1}\right] \sin (2[\kappa
1152: -\sqrt{\mu }]x)+
1153: \left[ \frac{V}{4}\right] \sin (2[\kappa +\sqrt{\mu }]x) \nonumber \\
1154: & \quad +\left[ \frac{g}{2}B_{1}^{3}+\frac{\alpha }{2}B_{2}+\frac{h}{2} B_{1}B_{2}^{2}\right] \cos (2\sqrt{\mu }x) \nonumber \\
1155: & \quad +\left[
1156: \frac{g}{8}B_{1}(3A_{1}^{2}-B_{1}^{2})+\frac{h}{4}A_{1}A_{2}B_{2}+
1157: \frac{h}{8}B_{1}(A_{2}^{2}-B_{2}^{2})\right] \cos (4\sqrt{\mu
1158: }x)
1159: \nonumber \\
1160: & \quad +\left[ \frac{V}{2}B_{1}\right] \cos (2\kappa x)+\left[
1161: -\frac{V}{4} \right] \cos (2[\kappa -\sqrt{\mu }]x)+\left[
1162: -\frac{V}{4}\right] \cos (2[\kappa +\sqrt{\mu }]x)\,,
1163: \label{harm1}
1164: \end{align}
1165: and $F_{B_{1}}=G_{B_{1}}/\sqrt{\mu }$, where
1166: \begin{align}
1167: G_{B_{1}}(A_{1},A_{2},A_{3},A_{4},x)& =\left[ \frac{\alpha
1168: }{2}A_{2}+\frac{3g
1169: }{8}A_{1}(A_{1}^{2}+B_{1}^{2})+\frac{h}{4}A_{2}B_{1}B_{2}+\frac{h}{8} A_{1}(A_{2}^{2}+B_{2}^{2})\right] \nonumber \\
1170: & \quad +\left[ \frac{\alpha }{2}B_{2}+\frac{g}{4} B_{1}(3A_{1}^{2}+B_{1}^{2})+\frac{h}{2}A_{1}A_{2}B_{2}+\frac{h}{4} B_{1}(A_{2}^{2}+B_{2}^{2})\right] \sin (2\sqrt{\mu }x) \nonumber \\
1171: & \quad +\left[ \frac{g}{8}B_{1}(3A_{1}^{2}-B_{1}^{2})+\frac{h}{4}
1172: A_{1}A_{2}B_{2}+\frac{h}{8}B_{1}(A_{2}^{2}-B_{2}^{2})\right] \sin
1173: (4\sqrt{
1174: \mu }x) \nonumber \\
1175: & \quad +\left[ \frac{V}{4}B_{1}\right] \sin (2[\kappa
1176: -\sqrt{\mu }]x)+
1177: \left[ -\frac{V}{4}\right] \sin (2[\kappa +\sqrt{\mu }]x) \nonumber \\
1178: & \quad +\left[ \frac{\alpha }{2}A_{2}+\frac{g}{2}A_{1}^{3}+\frac{h}{2} A_{1}A_{2}^{2}\right] \cos (2\sqrt{\mu }x) \nonumber \\
1179: & \quad +\left[
1180: \frac{g}{8}A_{1}(A_{1}^{2}-3B_{1}^{2})-\frac{h}{4}A_{2}B_{1}B_{2}+
1181: \frac{h}{8}A_{1}(A_{2}^{2}-B_{2}^{2})\right] \cos (4\sqrt{\mu
1182: }x)
1183: \nonumber \\
1184: & \quad +\left[ -\frac{V}{2}A_{1}\right] \cos (2\kappa x)+\left[
1185: -\frac{V}{4} A_{1}\right] \cos (2[\kappa -\sqrt{\mu }]x)+\left[
1186: -\frac{V}{4}A_{1} \right] \cos (2[\kappa +\sqrt{\mu }]x)\,.
1187: \label{harm2}
1188: \end{align}
1189: Only $O(1)$ [i.e., constant harmonic] terms remain after averaging.
1190:
1191: In the resonant case, one obtains, after averaging, an extra term
1192: depending on the periodic potential $V$, because a term that was a
1193: prefactor of a non-constant harmonic in (\ref{harm1}) and
1194: (\ref{harm2}) has become a coefficient in front of the $O(1)$
1195: term. Other harmonic terms are also simplified due to the
1196: resonance, but they nevertheless do not contribute to the averaged
1197: equations because they are still prefactors of
1198: non-constant harmonics. The extra terms with $\mu _{1}$ arise
1199: from Taylor expanding in powers of $\epsilon $ and keeping the
1200: leading-order terms. In the resonant case,
1201: $F_{A_{1}}=G_{A_{1}}/\kappa $ and $F_{B_{1}}=G_{B_{1}}/ \kappa $.
1202:
1203: In both the resonant and non-resonant cases, the expressions for $F_{A_2}$
1204: and $F_{B_2}$ are obtained by switching the subscripts $1
1205: \longleftrightarrow 2$ in the equations above.
1206:
1207:
1208: %\bibliographystyle{plain}
1209: %\bibliography{ref,bec}
1210:
1211:
1212: \begin{thebibliography}{10}
1213:
1214: \bibitem{Agrawal}
1215: G.~P. Agrawal.
1216: \newblock {\em Nonlinear Fiber Optics}.
1217: \newblock Academic Press, San Diego, CA, 1995.
1218:
1219: \bibitem{alf}
1220: G.~L. Alfimov, P.~G. Kevrekidis, V.~V. Konotop, and M.~Salerno.
1221: \newblock Wannier functions analysis of the nonlinear {S}chrodinger equation
1222: with a periodic potential.
1223: \newblock {\em Physical Review E}, 66(046608), October 2002.
1224:
1225: \bibitem{anderson}
1226: B.~P. Anderson and M.~A. Kasevich.
1227: \newblock Macroscopic quantum interference from atomic tunnel arrays.
1228: \newblock {\em Science}, 282(5394):1686--1689, November 1998.
1229:
1230: \bibitem{baiz}
1231: B.~B. Baizakov, V.~V. Konotop, and M.~Salerno.
1232: \newblock Regular spatial structures in arrays of {B}ose-{E}instein condensates
1233: induced by modulational instability.
1234: \newblock {\em Journal of Physics B: Atomic Molecular and Optical Physics},
1235: 35:5105--5119, 2002.
1236:
1237: \bibitem{linear-coupling1}
1238: R.~J. Ballagh, K.~Burnett, and Scott~T. F.
1239: \newblock Theory of an output coupler for {B}ose-{E}instein condensed atoms.
1240: \newblock {\em Physical Review Letters}, 78:1607--1611, March 1997.
1241:
1242: \bibitem{towers}
1243: Y.~B. Band, I.~Towers, and Boris~A. Malomed.
1244: \newblock Unified semiclassical approximation for {B}ose-{E}instein
1245: condensates: Application to a {BEC} in an optical potential.
1246: \newblock {\em Physical Review A}, 67(023602), February 2003.
1247:
1248: \bibitem{space1}
1249: Kirstine Berg-S{\o}rensen and Klaus M{\o}lmer.
1250: \newblock Bose-{E}instein condensates in spatially periodic potentials.
1251: \newblock {\em Physical Review A}, 58(2):1480--1484, August 1998.
1252:
1253: \bibitem{bronskiatt}
1254: Jared~C. Bronski, Lincoln~D. Carr, Ricardo Carretero-Gonz\'alez, Bernard
1255: Deconinck, J.~Nathan Kutz, and Keith Promislow.
1256: \newblock Stability of attractive {B}ose-{E}instein condensates in a periodic
1257: potential.
1258: \newblock {\em Physical Review E}, 64(056615), 2001.
1259:
1260: \bibitem{bronski}
1261: Jared~C. Bronski, Lincoln~D. Carr, Bernard Deconinck, and J.~Nathan Kutz.
1262: \newblock Bose-{E}instein condensates in standing waves: The cubic nonlinear
1263: {S}chr\"odinger equation with a periodic potential.
1264: \newblock {\em Physical Review Letters}, 86(8):1402--1405, February 2001.
1265:
1266: \bibitem{bronskirep}
1267: Jared~C. Bronski, Lincoln~D. Carr, Bernard Deconinck, J.~Nathan Kutz, and Keith
1268: Promislow.
1269: \newblock Stability of repulsive {B}ose-{E}instein condensates in a periodic
1270: potential.
1271: \newblock {\em Physical Review E}, 63(036612), 2001.
1272:
1273: \bibitem{lattice}
1274: S.~Burger, F.~S. Cataliotti, C.~Fort, F.~Minardi, and M.~Inguscio.
1275: \newblock Superfluid and dissipative dynamics of a {B}ose-{E}instein condensate
1276: in a periodic optical potential.
1277: \newblock {\em Physical Review Letters}, 86(20):4447--4450, May 2001.
1278:
1279: \bibitem{edwards}
1280: Keith Burnett, Mark Edwards, and Charles~W. Clark.
1281: \newblock The theory of {B}ose-{E}instein condensation of dilute gases.
1282: \newblock {\em Physics Today}, 52(12):37--42, December 1999.
1283:
1284: \bibitem{couple5}
1285: Th. Busch, J.~I. Cirac, V.~M. P\'erez-Garc\'ia, and P.~Zoller.
1286: \newblock Stability and collective excitations of a two-component
1287: {B}ose-{E}instein condensed gas: A moment approach.
1288: \newblock {\em Physical Review A}, 56(4):2978--2983, October 1997.
1289:
1290: \bibitem{promislow}
1291: Ricardo Carretero-Gonz\'alez and Keith Promislow.
1292: \newblock Localized breathing oscillations of {B}ose-{E}instein condensates in
1293: periodic traps.
1294: \newblock {\em Physical Review A}, 66(033610), September 2002.
1295:
1296: \bibitem{space2}
1297: Dae-Il Choi and Qian Niu.
1298: \newblock Bose-{E}instein condensates in an optical lattice.
1299: \newblock {\em Physical Review Letters}, 82(10):2022--2025, March 1999.
1300:
1301: \bibitem{stringari}
1302: Franco Dalfovo, Stefano Giorgini, Lev~P. Pitaevskii, and Sandro Stringari.
1303: \newblock Theory of {B}ose-{E}instein condensation on trapped gases.
1304: \newblock {\em Reviews of Modern Physics}, 71(3):463--512, April 1999.
1305:
1306: \bibitem{kutz}
1307: Bernard Deconinck, B.~A. Frigyik, and J.~Nathan Kutz.
1308: \newblock Dynamics and stability of {B}ose-{E}instein condensates: The
1309: nonlinear {S}chr\"odinger equation with periodic potential.
1310: \newblock {\em Journal of Nonlinear Science}, 12(3):169--205, 2002.
1311:
1312: \bibitem{dec}
1313: Bernard Deconinck, J.~Nathan Kutz, Matthew~S. Patterson, and Brandon~W. Warner.
1314: \newblock Dynamics of periodic multi-component {B}ose-{E}instein condenates.
1315: \newblock {\em Journal of Physics A--Mathematics and General},
1316: 36(20):5431--5447, May 2003.
1317:
1318: \bibitem{pengels}
1319: P.~Engels.
1320: \newblock Personal communication.
1321:
1322: \bibitem{couple4}
1323: B.~D. Esry, Chris~H. Greene, James~P. Burke~Jr., and John~L. Bohn.
1324: \newblock Hartree-{F}ock theory for double condensates.
1325: \newblock {\em Physical Review Letters}, 78(19):3594--3597, May 1997.
1326:
1327: \bibitem{gucken}
1328: John Guckenheimer and Philip Holmes.
1329: \newblock {\em Nonlinear Oscillations, Dynamical Systems, and Bifurcations of
1330: Vector Fields}.
1331: \newblock Number~42 in Applied Mathematical Sciences. Springer-Verlag, New
1332: York, NY, 1983.
1333:
1334: \bibitem{hagley}
1335: E.~W. Hagley, L.~Deng, M.~Kozuma, J.~Wen, K.~Helmerson, S.~L. Rolston, and
1336: W.~D. Phillips.
1337: \newblock A well-collimated quasi-continuous atom laser.
1338: \newblock {\em Science}, 283(5408):1706--1709, March 1999.
1339:
1340: \bibitem{dsh}
1341: D.~S. Hall.
1342: \newblock Personal communication.
1343:
1344: \bibitem{couple1}
1345: Tin-Lun Ho and V.~B. Shenoy.
1346: \newblock Binary mixtures of {B}ose condensates of alkali atoms.
1347: \newblock {\em Physical Review Letters}, 77(16):3276--3279, October 1996.
1348:
1349: \bibitem{inouye}
1350: S.~Inouye, M.~R. Andrews, J.~Stenger, H.~J. Miesner, D.~M. Stamper-Kurn, and
1351: W.~Ketterle.
1352: \newblock Observation of {F}eshbach resonances in a {B}ose-{E}instein
1353: condensate.
1354: \newblock {\em Nature}, 392(6672):151--154, March 1998.
1355:
1356: \bibitem{ketter}
1357: Wolfgang Ketterle.
1358: \newblock Experimental studies of {B}ose-{E}instein condensates.
1359: \newblock {\em Physics Today}, 52(12):30--35, December 1999.
1360:
1361: \bibitem{kohler}
1362: Thorsten K\"ohler.
1363: \newblock Three-body problem in a dilute {B}ose-{E}instein condensate.
1364: \newblock {\em Physical Review Letters}, 89(21):210404, 2002.
1365:
1366: \bibitem{kuklov}
1367: Anatoly Kuklov, Nikolay Prokof'ev, and Boris Svistunov.
1368: \newblock Detecting supercounterfluidity by {R}amsey spectroscopy.
1369: \newblock {\em Physical Review A}, 69(025601), February 2004.
1370:
1371: \bibitem{band}
1372: Pearl J.~Y. Louis, Elena~A. Ostrovskaya, Craig~M. Savage, and Yuri~S. Kivshar.
1373: \newblock Bose-einstein condensates in optical lattices: {B}and-gap structure
1374: and solitons.
1375: \newblock {\em Physical Review A}, 67(013602), 2003.
1376:
1377: \bibitem{pethick2}
1378: M.~Machholm, A.~Nicolin, C.~J. Pethick, and H.~Smith.
1379: \newblock Spatial period-doubling in {B}ose-{E}instein condensates in an
1380: optical lattice.
1381: \newblock {\em Physical Review A}, 69(043604), 2004.
1382: \newblock ArXiv:cond-mat/0307183.
1383:
1384: \bibitem{machholm}
1385: M.~Machholm, C.~J. Pethick, and H.~Smith.
1386: \newblock Band structure, elementary excitations, and stability of a
1387: {B}ose-{E}instein condensate in a periodic potential.
1388: \newblock {\em Physical Review A}, 67(053613), 2003.
1389:
1390: \bibitem{old}
1391: Boris~A. Malomed.
1392: \newblock Polarization dynamics and interactions of solitons in a birefringent
1393: optical fiber.
1394: \newblock {\em Physical Review A}, 43(1):410--423, January 1991.
1395:
1396: \bibitem{UNSW}
1397: Boris~A. Malomed, I.~M. Skinner, P.~L. Chu, and G.~D. Peng.
1398: \newblock Symmetric and asymmetric solitons in twin-core nonlinear optical
1399: fibers.
1400: \newblock {\em Physical Review E}, 53(4):4084--4091, April 1996.
1401:
1402: \bibitem{malopt}
1403: Boris~A. Malomed, Z.~H. Wang, P.~L. Chu, and G.~D. Peng.
1404: \newblock Multichannel switchable system for spatial solitons.
1405: \newblock {\em Journal of the Optical Society of America B}, 16(8):1197--1203,
1406: August 1999.
1407:
1408: \bibitem{mueller}
1409: Erich~J. Mueller.
1410: \newblock Superfluidity and mean-field energy loops; hysteretic behavior in
1411: {B}ose-{E}instein condensates.
1412: \newblock {\em Physical Review A}, 66(063603), 2002.
1413:
1414: \bibitem{myatt}
1415: C.~J. Myatt, E.~A. Burt, R.~W. Ghrist, E.~A. Cornell, and C.~E. Wieman.
1416: \newblock Production of two overlapping {B}ose-{E}instein condensates by
1417: sympathetic cooling.
1418: \newblock {\em Physical Review Letters}, 78:586--589, January 1997.
1419:
1420: \bibitem{pethick}
1421: C.~J. Pethick and H.~Smith.
1422: \newblock {\em Bose-Einstein Condensation in Dilute Gases}.
1423: \newblock Cambridge University Press, Cambridge, United Kingdom, 2002.
1424:
1425: \bibitem{mapbecprl}
1426: Mason~A. Porter and Predrag Cvitanovi\'c.
1427: \newblock Modulated amplitude waves in {B}ose-{E}instein condensates.
1428: \newblock {\em Physical Review E}, 69(047201), 2004.
1429: \newblock Ar{X}iv: nlin.CD/0307032.
1430:
1431: \bibitem{mapbec}
1432: Mason~A. Porter and Predrag Cvitanovi\'c.
1433: \newblock A perturbative analysis of modulated amplitude waves in
1434: {B}ose-{E}instein condensates.
1435: \newblock {\em Chaos}, To appear.
1436: \newblock Ar{X}iv: nlin.CD/0308024.
1437:
1438: \bibitem{couple3}
1439: H.~Pu and N.~P. Bigelow.
1440: \newblock Collective excitations, metastability, and nonlinear response of a
1441: trapped two-species {B}ose-{E}instein condensate.
1442: \newblock {\em Physical Review Letters}, 80(6):1134--1137, February 1998.
1443:
1444: \bibitem{couple2}
1445: H.~Pu and N.~P. Bigelow.
1446: \newblock Properties of two-species {B}ose condensates.
1447: \newblock {\em Physical Review Letters}, 80(6):1130--1133, February 1998.
1448:
1449: \bibitem{rand}
1450: Richard~H. Rand.
1451: \newblock {\em Topics in Nonlinear Dynamics with Computer Algebra}, volume~1 of
1452: {\em Computation in Education: Mathematics, Science and Engineering}.
1453: \newblock Gordon and Breach Science Publishers, USA, 1994.
1454:
1455: \bibitem{param}
1456: Richard~H. Rand.
1457: \newblock Dynamics of a nonlinear parametrically-excited pde: 2-term
1458: truncation.
1459: \newblock {\em Mechanics Research Communications}, 23(3):283--289, 1996.
1460:
1461: \bibitem{675}
1462: Richard~H. Rand.
1463: \newblock Lecture notes on nonlinear vibrations.
1464: \newblock a free online book available at {\it
1465: http://www.tam.cornell.edu/randdocs/nlvibe45.pdf}, 2003.
1466:
1467: \bibitem{salasnich}
1468: L.~Salasnich, A.~Parola, and L.~Reatto.
1469: \newblock Periodic quantum tunnelling and parametric resonance with
1470: cigar-shaped {B}ose-{E}instein condensates.
1471: \newblock {\em Journal of Physics B: Atomic Molecular and Optical Physics},
1472: 35(14):3205--3216, July 2002.
1473:
1474: \bibitem{linear-coupling2}
1475: D.~T. Son and M.~A. Stephanov.
1476: \newblock Domain walls of relative phase in two-component {B}ose-{E}instein
1477: condensates.
1478: \newblock {\em Physical Review A}, 65:063621, June 2002.
1479:
1480: \bibitem{couple6}
1481: Marek Trippenbach, Krzysztof G\'oral, Kazimierz Rz\c{a}\.{z}ewski, Boris
1482: Malomed, and Y.~B. Band.
1483: \newblock Structure of binary {B}ose-{E}instein condensates.
1484: \newblock {\em Journal of Physics B: Atomic, Molecular, and Optical Physics},
1485: 33:4017--4031, 2000.
1486:
1487: \bibitem{smer}
1488: A.~Trombettoni and A.~Smerzi.
1489: \newblock Discrete solitons and breathers with dilute {B}ose-{E}instein
1490: condensates.
1491: \newblock {\em Physical Review Letters}, 86(11):2353--2356, March 2001.
1492:
1493: \bibitem{split}
1494: J.~A.~C. Weideman and B.~M. Herbst.
1495: \newblock Split-step methods for the solution of the nonlinear {S}chr\"odinger
1496: equation.
1497: \newblock {\em SIAM Journal on Numerical Analysis}, 23(3):485--507, June 1986.
1498:
1499: \bibitem{wu4}
1500: Biao Wu, Roberto~B. Diener, and Qian Niu.
1501: \newblock Bloch waves and {B}loch {B}ands of {B}ose-{E}instein condensates in
1502: optical lattices.
1503: \newblock {\em Physical Review A}, 65(025601), 2002.
1504:
1505: \end{thebibliography}
1506:
1507:
1508: \end{document}
1509: