cond-mat0405064/lwi.tex
1: %\documentstyle[aps,preprint,pra,epsfig]{revtex} 
2: \documentstyle[aps,multicol,pre,epsfig]{revtex} 
3:  
4: %\documentstyle[aps,multicol,epsfig]{revtex}  
5: \pagenumbering{arabic}  
6: \newcommand{\beq}{\begin{equation}}  
7: \newcommand{\eeq}{\end{equation}}   
8:  
9: \begin{document}  
10: \begin{title}
11: {\bf Avoiding Infrared Catastrophes in Trapped 
12: Bose-Einstein Condensates} 
13: \end{title} 
14:  
15: \author{P.G. Kevrekidis$^{1}$, G. Theocharis$^2$, 
16: D.J. Frantzeskakis$^{2}$, and A. Trombettoni$^3$}  
17: \address{$^{1}$ Department of Mathematics and Statistics, 
18: University of Massachusetts, Amherst MA 01003-4515, USA \\  
19: $^{2}$ Department of Physics, University of Athens, Panepistimiopolis,  
20: Zografos, Athens 15784, Greece \\  
21: $^3$ Istituto Nazionale per la Fisica della Materia and 
22: Dipartimento di Fisica, Universita' di Parma, 
23: parco Area delle Scienze 7A, I-43100 Parma, Italy} 
24: \maketitle   
25:  
26: \begin{abstract}  
27: This paper is concerned with the long wavelength instabilities 
28: (infrared catastrophes) occurring in Bose-Einstein condensates (BECs). 
29: We examine the modulational instability in 
30: ``cigar-shaped'' (1D) attractive BECs and the  
31: transverse instability of dark solitons in ``pancake'' (2D)  
32: repulsive BECs. We suggest mechanisms, and give explicit estimates, 
33: on how to  ``engineer'' the trapping conditions of the condensate to avoid  
34: such instabilities: the main result being that a tight enough 
35: trapping potential suppresses the instabilities present  
36: in the homogeneous limit. We compare the obtained estimates  
37: with numerical results and we  
38: highlight the relevant regimes of dynamical behavior. 
39: % in the presence 
40: %as well as in the absence of the instability.   
41: \end{abstract}   
42:  
43: %\begin{multicols}{2} 
44:  
45: \vspace{2mm}   
46:  
47: \section{Introduction and Setup}
48:  
49: Infrared catastrophes, or long-wavelength instabilities (LWI) as 
50: they are otherwise known, are ubiquitous in physical phenomena.  
51: From Magneto-Hydro-Dynamics  \cite{MHD} to chemical models  
52: \cite{RD}, and from quantum systems \cite{wire}  
53: to fluid mechanics \cite{fluids}, polymer  
54: physics \cite{polymers} and plasmas \cite{plasma}, large  
55: scale modulations may destabilize the system of interest. 
56: On the other hand,  
57: many of these instabilities (and their thresholds) have been  
58: quantified in the context of nonlinear models. 
59:  
60: In the past few years, another context that can be accurately  
61: modeled by nonlinear partial differential  
62: equations that are known to possess LWI, has become experimentally tractable. 
63: This setting is, in particular, the one  
64: of Bose-Einstein condensates (BECs) \cite{review}  
65: whose experimental realization has led to an explosion of  
66: interest in the field of atomic matter waves and of nonlinear excitations  
67: in them. Such nonlinear waves have been recently experimentally  
68: generated in BECs, namely dark \cite{dark}  
69: and bright \cite {bright,paris} solitons. Also,  
70: two dimensional excitations, such as vortices \cite{vortex} and  
71: lattice patterns thereof \cite{vl} have also been obtained   
72: experimentally, while other nonlinear waves,  
73: such as Faraday waves \cite{stal}, ring   
74: solitons and vortex necklaces \cite{theo} have been theoretically predicted.  
75: It is worth noting here that the nature of Bose-Einstein nonlinear matter  
76: waves depends crucially on the type of the interatomic interactions:  
77: dark (bright) solitons can be created in BECs with repulsive  
78: (attractive)  
79: interatomic interactions, resulting from positive (negative)  
80: scattering length. 
81:  
82: In the present work, our scope is to re-examine the LWI  
83: in the context of Bose-Einstein condensates and  
84: exploit the consequences on the LWI of the  
85: inhomogeneity induced by the trapping potential.  
86: It is well-known \cite{review} in this setting that the  
87: condensates are formed under appropriate confining conditions 
88: that typically consist of magnetic traps modeled by  
89: parabolic potentials. Our scope is  
90: then two-fold. On the one hand, we aim at illustrating the potential for  
91: the instabilities and at examining their dynamical development when  
92: they exist. On the other hand, the magnetic confinement provides an  
93: additional ``trapping'' length scale to the problem whose  
94: {\it competition} with the instability length scale may disallow  
95: the dynamical manifestation of the LWI. In this way, we will propose 
96: how to {\it engineer} trapping conditions so as to prevent  
97: unstable dynamical evolution. 
98:  
99: We will examine two benchmark examples of long-wavelength  
100: instabilities in BECs: the modulational instability  
101:  occurring in attractive, one-dimensional (1D) BECs and  
102: giving rise to the formation of bright matter-wave solitons, and  
103: the transverse (``snaking'') instability of dark soliton stripes  
104: formed in repulsive two-dimensional (2D) BECs. It should be noted  
105: here that the genuinely three-dimensional (3D) condensate can be  
106: considered as approximately 1D if the nonlinear inter-atomic interaction  
107: is weak relative to the trapping potential force in the transverse  
108: directions; then, the transverse size of the condensates is much smaller than  
109: their length, i.e., the BEC is ``cigar-shaped'' and can  
110: be effectively described by 1D models \cite{GPE1d,VVK}. Similarly, if 
111: the transverse confinement is strong along one direction and weak 
112: along the others, then for this ``pancake-shaped'' BEC, 2D model equations 
113: are relevant \cite{GPE2d}. 
114:  
115: Close to zero temperature, it is well-known that the 3D Gross-Pitaevskii (GP)  
116: equation \cite{review} accurately captures the dynamics of the condensate. 
117: For cigar-shaped BECs, the model equation is effectively 1D  
118: and can be expressed in the following dimensionless form, 
119: \begin{eqnarray} 
120: i \frac{\partial u}{\partial t}= -  
121: \frac{1}{2} \frac{\partial^2 u}{\partial x^2} + \alpha |u|^2 u + V(x) u, 
122: \label{eqn1} 
123: \end{eqnarray} 
124: where $u$ is the macroscopic wave function  
125: (normalized to $1$), and $\alpha=a/|a|=\pm 1$ is the renormalized  
126: scattering length,  
127: which is positive (negative) for repulsive (attractive) condensates.  
128: In this equation, $t$ and $x$ are respectively measured in units of  
129: $9/16\epsilon^{2}\omega_{\perp}$ and  
130: $3 a_{\perp}/4 \epsilon$,  
131: where $\omega_{\perp}$ is the confining frequency in the transverse  
132: direction, $a_{\perp}=\sqrt{\hbar/m \omega_{\perp}}$ is the transverse  
133: harmonic-oscillator length and $\epsilon \equiv N |a|/a_{\perp}$ is a  
134: small dimensionless parameter, $N$ being the total number of atoms  
135: \cite{VVK}. Finally,   
136: \begin{equation}
137: V(x)=\frac{\Omega ^{2}}{2} x^{2}
138: \label{V}
139: \end{equation} 
140: where $\Omega=(9/16\epsilon^{2}) \cdot  
141: (\omega_x/\omega_{\perp})$ is the frequency of the magnetic trap 
142: potential $V$ in our dimensionless units, and $\omega_x$ is the axial  
143: confining frequency. Similarly the 2D model for the pancake-shaped  
144: condensate assumes the form: 
145: \begin{eqnarray} 
146: i \frac{\partial u}{\partial t}= -  
147: \frac{1}{2} \Delta u + \alpha |u|^2 u + V(r) u 
148: \label{eqn2} 
149: \end{eqnarray} 
150: where now the role of the axial frequency/spatial variable  
151: is played by the radial one ($r\equiv \sqrt{x^{2}+y^{2}}$)  
152: and the normalized variables are connected to the dimensional  
153: ones similarly to the 1D case, but with the role of $\omega_{x}$  
154: ($\omega_{\perp}$) now played by $\omega_{\perp}$ ($\omega_{z}$). 
155: 
156: We structure our presentation as follows: 
157: in Section II we study the suppression 
158: of the modulational instability for 1D cigar-shaped BECs 
159: in presence of a tight confining magnetic potential, while in 
160: Section III we investigate the transverse instability for 2D trapped BECs. 
161: In both cases we provide a criterion (in terms of the trap parameters) 
162: giving the conditions under which LWI can be avoided. 
163: Section IV is devoted to our concluding remarks.  
164: 
165:  
166: \section{Modulational Instability}
167:    
168: It is well-known (see, e.g., the recent work \cite{zoi}  
169: and references therein) that, in the absence of external potential,  
170: the continuous-wave (cw) solution $u=u_{0}\exp(-i\alpha u_{0}^{2}t)$  
171: of amplitude $u_0$, of Eq. (\ref{eqn1}) becomes modulationally unstable 
172: when perturbations of wavenumber $k < k_{cr} \equiv 2 |\alpha|^{1/2} u_0$ 
173: are imposed.  
174: This can be equivalently interpreted as follows: when length scales 
175: \begin{eqnarray} 
176: \lambda > \lambda_{cr} \equiv \frac{\pi}{u_0 \sqrt{|\alpha|}} 
177: \label{eqn3} 
178: \end{eqnarray} 
179: become ``available'' to the system, then the modulation over these scales 
180: leads the solution to instability.  
181:  
182: However, in the presence of the magnetic trap, 
183: there is a characteristic scale 
184: set by the trap, namely the BEC axial size, $\lambda_{BEC}$,  
185: which depends on the trapping frequency $\Omega$. When  
186: $\lambda_{BEC} < \lambda_{cr}$ suppression of the modulational  
187: instability is expected. To estimate $\lambda_{BEC}$ in a specific setup,  
188: we will examine a protocol relevant to the recent experiments conducted  
189: by the Rice \cite{bright} and Paris \cite{paris} groups, which has stimulated a considerable amount of theoretical attention  
190: \cite{brand}. In particular, we start with a 1D repulsive condensate,  
191: whose ground state wavefunction  
192: is approximately in the so-called Thomas-Fermi (TF)  
193: regime \cite{review}, and subsequently change the interaction  
194: into an attractive one.  
195: This is experimentally achieved using the so-called Feshbach resonance:  
196: an external magnetic field is used to modify the scattering length of the  
197: interatomic interactions \cite{inouye}.  
198: In our example, we use as initial condition  
199: the ground state of Eq. (\ref{eqn1}) with $\alpha=1$,  
200: which is in the TF approximation $|u|^2 \approx \mu - V(x)$;  
201: then at $t=0$ we change the sign of the scattering length,  
202: i.e., we set $\alpha=-1$.  
203: Therefore in this situation $\lambda_{BEC} \approx 2\sqrt{2\mu}/\Omega$. From  
204: the normalization condition $\int dx |u|^2=1$  
205: one gets 
206: \begin{equation}
207: \mu=\Bigg(\frac{3 \Omega}{4\sqrt{2}} \Bigg)^{2/3}.
208: \label{mu_1D}
209: \end{equation}
210:   
211: The condition for the suppression of the modulational instability 
212: $\lambda_{BEC} < \lambda_{cr}$ [where $\lambda_{cr}$ is given by 
213: Eq. (\ref{eqn3})] gives $\Omega> 2^{3/2} (|\alpha| \mu)^{1/2} u_0/\pi$. 
214: In this context, the amplitude $u_{0}$ in Eq. (\ref{eqn5})  
215: can be well approximated as $u_0 \approx \sqrt{\mu}$,  
216: since the TF approximation is most accurate close to the center  
217: of the condensate. Therefore if $\Omega > \Omega_{cr}$, where 
218: \begin{eqnarray} 
219: \Omega_{cr} = \frac{9}{\sqrt{2} \pi^3} \approx 0.2, 
220: \label{eqn5} 
221: \end{eqnarray} 
222: then the trapping conditions are ``engineered'' in such a way that  
223: the modulational instability cannot manifest itself: 
224: for a tight enough trapping potential the modulational instability 
225: does not occur. This is one of the key aims of this paper, 
226: namely to quantitatively highlight how the infrared catastrophes 
227: can be avoided in the presence of  
228: sufficiently ``tight'' trapping of the condensate. We remark 
229: that if one evaluates $\lambda_{BEC}$ by using for the ground state  
230: of Eq. (\ref{eqn1}) (with $\alpha=1$) 
231: a gaussian with variationally determined  
232: width, one has an estimate of $\Omega_{cr}$ in good agreement  
233: with Eq. (\ref{eqn5}). We also notice that the critical length 
234: $\lambda_{cr}$ is a few healing lengths $\xi$: indeed, when the 
235: the density grows from $0$ to $u_{0}^{2}$ within a distance $\xi$, 
236: the quantum pressure and interaction energy terms are equal when
237: $1/(2\xi^{2}) \approx u_{0}^{2}$, which means that
238: $\xi \approx (\sqrt{2}u_{0})^{-1}$ and therefore 
239: $\lambda_{cr} \approx \pi\sqrt{2}\xi = 4.44 \xi$. 
240: 
241: Typical experimental values for a $^{7}$Li BEC are  
242: $\omega_{x}=2\pi \times 5$Hz, $\omega_{\perp}=2\pi \times 500$Hz,  
243: $a=-3a_0$ and $N \approx 10^{3}$; these yield  
244: $\epsilon \sim 0.1$ and $\Omega \sim 0.5$, which should be  
245: a sufficiently 
246: large value to avoid the emergence of the modulational instability. 
247:  
248: We tested this prediction by means of direct numerical simulations  
249: shown in Fig. \ref{fig1}. The panels (a), (b), (c) show respectively  
250: the case of $\Omega=0.3$, $0.1$, and $0.02$. 
251: It is seen that the supercritical, ``tight'' trap 
252: $\Omega=0.3$ does not allow the development of the instability 
253: for the condensate [Fig. \ref{fig1} (a)] .  
254: The only consequence of the change of the sign of the scattering length is  
255: the excitation of an internal mode of the condensate, due to the fact  
256: that the TF cloud is not the ground-state  
257: for $\alpha=-1$. This results in nearly periodic oscillations of  
258: the width of the wavefunction  
259: $W=[\int x^2 |u|^2 dx - (\int x |u|^2)]^{1/2}$,  
260: which we monitor as a diagnostic in our numerical examples.  
261: For $\Omega=0.1$ [Fig. \ref{fig1} (b)], the width oscillations  
262: are no longer periodic: the result of the 
263: modulational instability can be clearly discerned in the addition of  
264: an extra frequency to the motion of the condensate for subcritical trapping.  
265: We have identified this to be a rather smooth transition,  
266: becoming more pronounced close to the theoretically predicted  
267: critical point of $\Omega \approx 0.2$.  
268: We have also examined the development of the 
269: instability for weaker trapping conditions  
270: [$\Omega=0.02$, see Figs. \ref{fig1} (c), (d)].  
271: Then an additional frequency emerges,  
272: as the condensate is now cleaved in two pieces during  
273: the time evolution [see Fig. \ref{fig1} (d)].  
274: For even weaker trappings, we found  
275: that the larger the ratio of $\lambda_{BEC}/\lambda_{cr}$,  
276: the more frequencies appear in the condensate density evolution,  
277: rendering it increasingly chaotic. Hence, we have demonstrated that a  
278: cascade of frequencies arises in the BEC dynamics  due to the modulational  
279: instability and subsequent higher resonances that allow the breakup of the 
280: condensate and make its motion less regular.  
281: This mechanism should clearly be experimentally detectable;  
282: in fact we surmise that the observation of a single matter-wave 
283: bright soliton \cite{paris}, rather than a soliton train \cite{bright}, 
284: is due to the fact that $\lambda_{BEC}$ was smaller in the former case, 
285: as the number of atoms was an order of magnitude 
286: fewer in the Paris experiment, while the ratio of the trapping 
287: frequencies was approximately the same.  
288: %To conclude this Section, we mention that if one starts  
289: %already with attractive interactions and perturb the ground-state  
290: %(e.g., suddenly displacing the magnetic trap), one has yet suppression  
291: %of the modulational instability during the dynamics for a tight enough  
292: %trap confinement, similarly to the previous analysis. 
293:  
294: % in  
295: %the formation of bright matter-wave soliton trains. 
296: %and theoretical results \cite{hulet,brand} for the  
297: %formation of bright matter-wave soliton trains. 
298:  
299: \section{Transverse Instability}
300:   
301: One of the infrared catastrophes that occur in two spatial  
302: dimensions is the transverse instability of  
303: dark-soliton stripes in repulsive ($\alpha > 0$) BECs.  
304: As a result, a dark-soliton undergoes a  
305: transverse snake deformation \cite{luther},  
306: %where transverse modulations cause  
307: causing the nodal plane to decay into vortex pairs.  
308: This instability has been examined in the context of BECs  
309: in \cite{shlyap}, and the Bogoliubov spectrum of the dark soliton  
310: has been obtained; in this context, the relevant imaginary modes  
311: were identified to transfer the energy of the condensate to collective  
312: excitations parallel to the nodal plane destroying the configuration.  
313: However, as is highlighted in \cite{feder}, ``the explicit connection  
314: between the existence of imaginary excitations and a dynamical snake  
315: instability remains unclear''. Our scope here is to illustrate  
316: the criterion for the transverse instability and to 
317: test it against direct numerical simulations, exposing the possible  
318: dynamical scenarios and quantifying their dependence 
319: on the trapping parameters. 
320:  
321: In the absence of the potential, the transverse instability occurs  
322: for perturbation wavenumbers   
323: \begin{equation}
324: k < k_{cr} \equiv  
325: \left[ 2 \sqrt{\sin^2{\phi}+ u_{0}^{-2}\sin{\phi}+
326: u_{0}^{-4}}-\left(2\sin{\phi}+u_{0}^{-2} \right)\right]^{1/2},
327: \label{k_cr_2D}
328: \end{equation}  
329: where $\sin\phi$ is the dark-soliton velocity 
330: \cite{luther} and $u_{0}$ is the amplitude of the homogeneous 
331: background, connected with the chemical potential through 
332: $u_{0}^{2}=\mu$ similarly to the previous ($1$D) setting.  
333: In the case of stationary (black) solitons, of interest here,  
334: $\sin\phi=0$, hence $k_{cr}=u_{0}^{-1}$. On the other hand, for  
335: $V(r)=\Omega^2 r^2/2$, a similar calculation as for the  
336: 1D problem yields the characteristic length scale of the BEC  
337: (i.e., the diameter of the TF cloud) as  
338: $\lambda_{BEC} \approx 2 \sqrt{2\mu}/\Omega$.  
339: Then, the criterion for the suppression of the transverse instability  
340: is that the scale of the BEC is shorter than the minimal one  
341: for the instability. The corresponding condition reads 
342: \begin{eqnarray} 
343: \Omega > \frac{\sqrt{2\mu}}{\pi u_{0}}. 
344: \label{eqn6} 
345: \end{eqnarray} 
346: To obtain the minimum value of $\Omega$ we need to know how 
347: $\mu$ is connected with $u_{0}$. As a first guess, 
348: in the absence of the dark soliton, 
349: one can assume $u_{0}^{2} \approx \mu$ 
350: (close to the center of the BEC), which yields $\Omega > \sqrt{2}/\pi = 
351: 0.45$. Hence, stronger trapping should ``drown'' the transverse instability 
352: and preserve dark soliton stripes on top of the Thomas-Fermi cloud  
353: %These can be otherwise viewed as  
354: (i.e., stable ``dipole'' solutions). Note that in terms of real  
355: physical units, the above mentioned critical value of $\Omega$  
356: may correspond, e.g., to a weakly interacting $^{87}$Rb pancake  
357: condensate, containing $\approx 10^{3}$ atoms, confined in a trap  
358: with $\omega_{r}=2\pi \times 5$Hz and $\omega_{z}=2\pi \times 50$Hz. 
359:  
360: We have numerically tested this condition, finding it to be an  
361: {\it overestimate} of the critical trapping frequency  
362: for the transverse instability, which is $\Omega_{cr}  
363: \approx 0.31$.     
364: %note that in terms of perturbation wavenumbers, 
365: This result shows that the numerically found  
366: %stability band is $k>0.69$ (instead of $k>1$).  
367: Fig. \ref{fig2} shows a dynamical evolution example  
368: of the dipolar solution, initialized with a $\tanh(y)$ imposed on the  
369: TF cloud, for $\Omega=0.35$ (left) and $\Omega=0.15$ (right).  
370: Both snapshots show the contour plot of the square modulus of the wave  
371: function at $t=1000$ (left) and $t=200$ (right).  
372: Clearly, in the former case the transverse instability is suppressed,  
373: while in the latter a vortex pair has been formed demonstrating  
374: the dynamical instability of the configuration.  
375: We have monitored the asymptotic, long time evolution of the  
376: dipole and have observed the following interesting phenomenology:  
377: for $0.18 \lesssim \Omega \lesssim 0.31$, while the stripe is dynamically  
378: unstable, there is not sufficient space  
379: for the instability-induced vortices to fully develop; as a result,  
380: after their formation, they subsequently recombine and disappear.  
381: This behavior is shown in Fig. \ref{fig3} (for $\Omega=0.2$):  
382: the two vortices formed are shown at $t=190$ (left panel),  
383: while at $t=210$ they recombine to form a transient dark stripe  
384: (right panel). This configuration is unstable and it subsequently  
385: breaks up to a new vortex pair (not shown here), which eventually  
386: recombines at longer times ($t\approx 400$). It is interesting  
387: to note that as, in the present case, the available size of the  
388: condensate is of the order of a few healing lengths $\xi$  
389: (in our units, $\xi=1/\sqrt{2}$), the two vortices formed  
390: hardly fit the condensate size (recall that the vortex core  
391: is of order of $\xi$ \cite{fetter}). This is a possible  
392: qualitative explanation of this recombination, whose origin is 
393: the competition of the length scales available in the BEC.  
394: Finally, it should be noted that for $\Omega \lesssim 0.18$, 
395: the vortices will 
396: survive in the asymptotic evolution of the system, and naturally the weaker  
397: the trapping the larger the number of ``engulfed'' vortices  
398: generated due to the stripe breakup.  
399:  
400: %%%%%%%%%%%%%%%%%%%%%%%%%%% 
401: A question that naturally arises in the results 
402: above concerns the disparity between the critical 
403: point theoretical estimate for the transverse instability 
404: and the corresponding numerical finding. 
405: We believe that the disparity is justified by the fact  
406: that the theoretical stability analysis of \cite{luther} is  
407: performed for the infinite homogeneous medium (the dark-soliton pedestal),  
408: in the absence of a magnetic trap. In the presence of the trap   
409: on the one hand, the background is inhomogeneous, 
410: %but also, in the case under consideration, i.e., 
411: while, on the other hand,
412: for tight traps resulting in small condensate sizes, 
413: the presence of the dark soliton at the BEC center, 
414: significantly modifies the maximum density. 
415: Thus, one should expect that the relation $u_{0}^{2}=\mu$ 
416: should be modified as $\delta u_{0}^{2}=\mu$, where 
417: the ``rescaling'' factor $\delta<1$. Based on the results 
418: of the numerical simulations (see Figs. \ref{fig2}-\ref{fig3}) 
419: we have observed that embedding the dark soliton in the BEC 
420: center reduces the maximum density of the TF cloud to 
421: the half of its initial value. 
422: This suggests that $\delta=1/2$, which, according to Eq. 
423: (\ref{eqn6}), leads to a new value for $\Omega$ minimum, 
424: i.e., $\Omega=1/\pi\approx 0.318$. 
425: This modified criterion for the suppression of the 
426: transverse instability, namely $\Omega>0.318$ is in 
427: very good agreement with the above mentioned numerically 
428: found condition $\Omega>0.31$. 
429:  
430: \section{Conclusions}
431: 
432: We have examined case examples of long  
433: wavelength instabilities in Bose-Einstein condensates.  
434: We have revisited the modulational instability  
435: in quasi-1D, cigar-shaped BECs, and the transverse instability 
436: in quasi-2D, pancake-shaped BECs. We have advocated that trapping  
437: conditions can be  
438: engineered to avoid or induce such instabilities at will.  
439: Using the length scale competition of the entrainment due to trapping and of  
440: critical instability wavelength, we have given explicit estimates for  
441: critical values of the trapping frequency beyond which the instabilities  
442: will be absent.  
443: We have tested these criteria  
444: in both cases and have found good agreement with the numerical results  
445: in the 1D case. 
446: In the 2D setting, we have explained the overestimation of the  
447: critical point, on the basis of the  
448: homogeneous background assumed in the theoretical estimate, 
449: as well as the modification of the value of the maximum density 
450: of the condensate due to the presence of the dark soliton. 
451: %(which is significant for tight traps).  
452: Our results demonstrate how to engineer 
453: the trapping conditions, in order to achieve supercritical regimes, 
454: devoid of long  
455: wavelength instabilities. On the other hand, for the subcritical regimes,  
456: we have illustrated the relevant phenomenology through direct numerical  
457: simulations and the cascade that leads to 
458: the instability and eventual destruction of the original coherent  
459: structure. Such results can be particularly useful in quantifying the  
460: selection of external  
461: conditions so as to achieve or avoid a given  experimental outcome.  
462: %An analytical understanding of how the inhomogeneity of the background can be taken into  
463: %consideration to revise  
464: %the critical point prediction would definitely be of interest  
465: %for future investigations. 
466:  
467: This work was supported by the Eppley Foundation and 
468: NSF-DMS-0204585 (PGK), the ``A.S. Onasis'' 
469: Public Benefit Foundation (GT) and the Special Research Account of the  
470: University of Athens (GT, DJF), and by M.I.U.R. through grant No. 
471: 2001028294 (AT). 
472:  
473: \begin{references}  
474:  
475: \bibitem{MHD} V.P. Lakhin and V.D. Levchenko, Plasma Phys. Rep.  
476: {\bf 29} 328 (2003). 
477:  
478: \bibitem{RD} T.K. Callahan and E. Knobloch, Phys. Rev. E {\bf 64},  
479: 036214 (2001); L.Q. Zhou and Q. Ouyang, 
480: Phys. Rev. Lett. {\bf 85}, 1650 (2000). 
481:  
482: \bibitem{wire} R.K. Moudgil, J. Phys.: Cond. Mat. {\bf 12}, 1781 (2000); 
483: J.A. Holyst and L.A. Turski, \newblock Phys. Rev. A {\bf 45}, 6180 (1992). 
484:  
485: \bibitem{fluids} W. Boos and A. Thess, Phys. Fluids {\bf 11}, 1484 (1999); 
486: M. Lee {\it et al.}, Phys. Rev. Lett. {\bf 76}, 2702 (1996). 
487:  
488: \bibitem{polymers} Y. Shiwa, \newblock Phys. Lett. A {\bf 228}, 279 (1997). 
489:  
490: \bibitem{plasma} S.P. Gary, Plasma Phys. Contr. F. {\bf 15}, 399 (1973); 
491: R. Bingham, Phys. Fluids {\bf 7}, 1001 (1965). 
492:   
493: \bibitem{review}  
494: L.P. Pitaevskii and S. Stringari, {\it Bose-Einstein Condensation}, 
495: (Oxford University Press, Oxford, 2003). 
496: %F. Dalfovo, {\it et al.},  
497: %S. Giorgini, L. P. Pitaevskii, and S. Stringari,  
498: %Theory of Bose-Einstein condensation in trapped gases,  
499: %Rev. Mod. Phys. {\bf 71}, 463%-521 (1999).   
500: %\bibitem{Kett}  M.H. Anderson {\it et al}, Science {\bf 269}, 198 (1995);  
501: %K.B. Davis {\it et al}, Phys. Rev. Lett. {\bf 75}, 3969 (1995); C.C. Bradley  
502: %{\it et al}, Phys. Rev. Lett. {\bf 75}, 1687 (1995).   
503:  
504: \bibitem{dark}  S. Burger {\it et al.}, Phys. Rev. Lett. {\bf 83},  
505: 5198 (1999); J. Denschlag {\it et al.}, Science {\bf 287}, 97 (2000);  
506: B.P. Anderson {\it et al.}, Phys. Rev. Lett. {\bf 86}, 2926 (2001).  
507:  
508: \bibitem{bright}  K.E. Strecker {\em et al.}, Nature {\bf 417}, 150 (2002); 
509: K.E. Strecker {\it et al.}, New. J. Phys. {\bf 5}, 73 (2003). 
510:  
511: \bibitem{paris} L. Khaykovich {\em et al.}, Science {\bf 296}, 1290 (2002).   
512:  
513: \bibitem{vortex} M.R. Matthews {\it et al.}, Phys. Rev. Lett.  
514: {\bf 83}, 2498 (1999); K.W. Madison {\it et al.}  
515: Phys. Rev. Lett. {\bf 84}, 806 (2000);  
516: S. Inouye {\em et al.}, Phys. Rev. Lett.{\bf 87}, 080402 (2001). 
517:  
518: \bibitem{vl} J.R. Abo-Shaeer {\it et al.}, Science {\bf 292}, 476 (2001);  
519: J.R. Abo-Shaeer, C. Raman, and W. Ketterle, Phys. Rev. Lett. {\bf{88}},  
520: 070409 (2002); P. Engels {\it et al}, Phys. Rev. Lett. {\bf 89}, 100403 (2002). 
521:  
522: \bibitem{stal} K. Staliunas, S. Longhi, and G. J. de Valc{a'}rcel, 
523: \newblock Phys. Rev. Lett. {\bf 89}, 210406 (2002). 
524:  
525: \bibitem{theo} G. Theocharis {\it et al.},   
526: %D.J. Frantzeskakis, P.G. Kevrekidis, B.A. Malomed and Yu.S. Kivshar,  
527: \newblock Phys. Rev. Lett. {\bf 90}, 120403 (2003). 
528:  
529: \bibitem{GPE1d}  V.M. P\'{e}rez-Garc\'{i}a,  
530: H. Michinel, and H. Herrero, Phys. Rev. A {\bf 57}, 3837 (1998);  
531: Yu.S. Kivshar, T.J. Alexander, and S.K. Turitsyn,  
532: Phys. Lett. A {\bf 278}, 225 (2001).  
533:  
534: \bibitem{VVK} F.Kh. Abdullaev {\it et al.},  
535: \newblock Phys. Rev. Lett. {\bf 90}, 230402 (2003).  
536: %V.V. Konotop and M. Salerno, Phys. Rev. A {\bf 65}, 021602 (2002). 
537:  
538: \bibitem{GPE2d} L. Salasnich, A. Parola,  
539: and L. Reatto, Phys. Rev. A {\bf 65}, 043614 (2002).  
540: %Y.B.\ Band, I.\ Towers, and B.A.\ Malomed, Phys.\ Rev.\ A \textbf{67}, 023602 (2003). 
541:  
542: \bibitem{zoi} G. Theocharis {\it et al.},  
543: Phys. Rev. A {\bf 67}, 063610 (2003).   
544:   
545: %\bibitem{hulet} K.E. Strecker {\it et al.}, New. J. Phys. {\bf 5}, 73 (2003). 
546:  
547: \bibitem{brand} U. Al Khawaja {\it et al.}, Phys. Rev. Lett.  
548: {\bf 89}, 200404 (2002); L.D. Carr and J. Brand, 
549: Phys. Rev. Lett. {\bf 92}, 040401 (2004);
550: %cond-matt/0303257; 
551: L. Salasnich, A. Parola, and L. Reatto, Phys. Rev. Lett. {\bf  91},  
552: 080405 (2003). 
553:  
554: \bibitem{inouye} S. Inouye {\it et al.}, Nature {\bf 392}, 151 (1998);  
555: E.A. Donley {\it et al.}, Nature {\bf 412}, 295 (2001). 
556:   
557: %\bibitem{inouye}  S. Inouye {\it et al.}, Nature {\bf 392}, 151 (1998);  
558: %J. Stenger {\it et al.}, Phys. Rev. Lett. {\bf 82}, 2422 (1999); 
559: %J.L. Roberts {\it et al.}, Phys. Rev. Lett. {\bf 81}, 5109 (1998);  
560: %S.L. Cornish {\it et al.}, Phys. Rev. Lett. {\bf 85}, 1795 (2000); 
561: %E.A. Donley {\it et al.}, Nature {\bf 412}, 295 (2001). 
562:  
563: \bibitem{luther} Yu.S. Kivshar and B. Luther-Davies, 
564: Phys. Rep. {\bf 298}, 81 (1998). 
565:  
566: \bibitem{shlyap} A.E. Muryshev {\it et al.}, Phys. Rev. A {\bf 60}, R2665 (1999);  
567: P.O. Fedichev {\it et al.}, 
568: %A.E. Muryshev, and G.V. Shlyapnikov, 
569: {\it ibid.}, {\bf 60}, 3220 (1999). 
570:  
571: \bibitem{feder} D.L. Feder {\it et al.}, 
572: Phys. Rev. A {\bf 62}, 053606 (2000). 
573:  
574: \bibitem{fetter} A.L. Fetter and A.A. Svidzinsky, 
575: J. Phys.: Cond. Matt. {\bf 13}, R135 (2001). 
576:  
577: \end{references} 
578:    
579: %\end{multicols} 
580:  
581: %\begin{figure}[tbp]  
582: %\epsfxsize=5cm  
583: %\epsffile{lwimi1.ps} 
584: %\epsfxsize=5cm  
585: %\epsffile{lwimi4.ps} 
586: %\epsfxsize=5cm  
587: %\epsffile{lwimi5.ps} 
588: %\center{ 
589: %\centerline{\epsffile{frmfig2b.ps}} 
590: %\caption{Left panel: The amplitude (dashed line) and width  
591: %(solid line) of the condensate as a function of time for  
592: %$\Omega=1$ (top subplot). In the same case, the space-time, contour 
593: %plot evolution of the density $|u(x,t)|^2$ is also given (bottom subplot).  
594: %Middle panel: The width of the condensate above (top subplot) and below  
595: %(bottom subplot) the critical point  
596: %(for $\Omega=0.95$ and 0.84 respectively).  
597: %Right panel: The same for  
598: %$\Omega=0.63$ (top subplot), $0.45$ (middle) and $0.3$ (bottom subplot).}  
599: %\label{fig1} 
600: %\end{figure} 
601:  
602: \begin{figure} 
603: \centerline{\psfig{figure=fig1.ps,width=60mm,angle=270}} 
604: \caption{Condensate width as a function of time for  
605: $\Omega=0.3$ (a), $\Omega=0.1$ (b), and $\Omega=0.02$ (c).  
606: In (d) we plot the atomic density with $\Omega=0.02$  
607: at three different times:  
608: $t=0$ (solid line), $t=80$ (dotted line, corresponding  
609: to point $1$ in (c)) and $t=160$ (dashed line, corresponding  
610: to point $2$ in (c)).} 
611: \label{fig1} 
612: \end{figure} 
613:  
614: \begin{figure}[tbp]  
615: \epsfxsize=6.5cm  
616: \epsffile{fig2a.ps} 
617: \epsfxsize=6.5cm  
618: \epsffile{fig2b.ps} 
619: \caption{The panels show the contour plots of the density  
620: $|u|^2$ for $\Omega=0.35$ at $t=1000$ (left) and  
621: $\Omega=0.15$ at $t=200$ (right). In the first case the  
622: transverse instability is clearly suppressed, while  
623: in the second it sets in, giving rise to a formation of a vortex pair. }  
624: \label{fig2} 
625: \end{figure} 
626:  
627: \begin{figure}[tbp]  
628: \epsfxsize=6.5cm  
629: \epsffile{fig3a.ps} 
630: \epsfxsize=6.5cm  
631: \epsffile{fig3b.ps} 
632: \caption{Snapshots of a vortex-pair evolution in a case  
633: where snaking instability has set in ($\Omega=0.2$).  
634: In the left panel ($t=190$) the formed vortex pair is shown,  
635: while the right panel ($t=210$) shows the recombination  
636: of the two vortices, resulting in the re-generation of a dark stripe 
637: structure.  
638: The latter is unstable and decays at longer times ($t\approx 400$). }  
639: \label{fig3} 
640: \end{figure} 
641:  
642: %\begin{figure}[tbp]  
643: %\epsfxsize=7.5cm  
644: %\epsffile{homo1000.ps} 
645: %\epsfxsize=7.5cm  
646: %\epsffile{homo500.ps} 
647: %\caption{Same as in Fig \ref{fig2}, but for a trapping potential  
648: %V=(r/R)^{16}$: A stable dark soliton at $t=1000$ for $R=3.7$  
649: %(left panel) and a vortex pair at $t=500$ created  
650: %as a result of the instability of a dark soliton for $R=3.5$ (right panel).}  
651: %\label{fig4} 
652: %\end{figure} 
653:  
654: \end{document} 
655:   
656:  
657: \bibitem{DM} C. Kurtzke, IEEE Photon. Technol. Lett. {\bf 5}, 1250 (1993);  
658:  
659: %\bibitem{sulem} C. Sulem and P.L. Sulem, \newblock {\it The Nonlinear Schr{\"o}dinger Equation}, Springer-Verlag (New York, 1999).% 
660:  
661: \bibitem{DMS} N.J. Smith {\it et al}, Electron. Lett. {\bf 32}, 54 (1996);  
662: I. Gabitov and S. Turitsyn, Opt. Lett. {\bf 21}, 327 (1996);  
663: JETP Lett. {\bf 63}, 861 (1996).  
664:  
665: \bibitem{smerzi}  A. Smerzi, A. Trombettoni,  
666: P.G. Kevrekidis, and  A.R. Bishop, Phys. Rev. Lett. {\bf 89}, 170402 (2002). 
667:  
668: \bibitem{catal} F.S. Cataliotti {\it et al.},  cond-mat/0207139. 
669:  
670: \bibitem{rapti} Z. Rapti, P.G. Kevrekidis, and V.V. Konotop, 
671: to appear in the Proceedings of the El-Escorial meeting (2003). 
672:  
673: \bibitem{theoch1} G. Theocharis, Z. Rapti, P.G. Kevrekidis, 
674: D.J. Frantzeskakis and V.V. Konotop (unpublished). 
675:  
676: \bibitem{Kittel} C. Kittel, {\it Introduction to Solid State Physics}, John Wiley and Sons, 1986. 
677:  
678:  
679: \bibitem{winkler} Magnus and Winkler, Hill's equation. 
680:  
681: \bibitem{abdullaev} F.Kh. Abdullaev {\it et al.} cond-mat. 
682:  
683: \bibitem{goldman} see e.g., Goldman and Krivchenkov, 
684: {\it Problems in Quantum Mechanics}, Dover. 
685:     
686: