1: \documentclass[showpacs,pre]{revtex4}
2: \usepackage{graphicx}
3:
4: \begin{document}
5:
6: \title{Two-dimensional solitons in the Gross-Pitaevskii equation with
7: spatially modulated nonlinearity}
8: \author{Hidetsugu Sakaguchi$^{1}$ and Boris A. Malomed$^{2}$}
9:
10: \address{$^{1}$Department of Applied Science for Electronics and Materials,\\
11: Interdisciplinary Graduate School of Engineering Sciences,\\
12: Kyushu University, Kasuga, Fukuoka 816-8580, Japan\\
13: $^{2}$Department of Interdisciplinary Studies,\\
14: School of Electrical Engineering, Faculty of Engineering,\\
15: Tel Aviv University, Tel Aviv 69978, Israel}
16:
17: \begin{abstract}
18: We introduce a dynamical model of a Bose-Einstein condensate based
19: on the 2D Gross-Pitaevskii equation, in which the nonlinear
20: coefficient is a function of radius. The model describes a
21: situation with spatial modulation of the the negative atomic
22: scattering length, via the Feshbach resonance controlled by a
23: properly shaped magnetic of optical field. We focus on the
24: configuration with the nonlinear coefficient different from zero
25: in a circle or annulus, including the case of a narrow ring.
26: Two-dimensional solitons are found in a numerical form, and also
27: by means of a variational approximation; for an infinitely narrow
28: ring, the soliton is found in an exact form. A stability region
29: for the axisymmetric solitons is identified by means of numerical
30: and analytical methods. In particular, if the nonlinearity is
31: supported on the annulus, the upper stability border is determined
32: by azimuthal perturbations; the stability region disappears if the
33: ratio of the inner and outer radii of the annulus exceeds a
34: critical value $\approx 0.47$. The model gives rise to
35: bistability, as the stationary solitons coexist with stable
36: breathers, whose stability region extends to higher values of the
37: norm than that of the static solitons. The collapse threshold
38: strongly increases with the radius of the inner hole of the
39: annulus. Vortex solitons are found too, but they are unstable.
40: \end{abstract}
41:
42: \pacs{03.75.Lm, 05.45.Yv, 42.65.Tg}
43: \maketitle
44:
45: \section{Introduction}
46:
47: Matter-wave solitons have been created in Bose-Einstein
48: condensates (BECs) in various effectively one-dimensional (1D)
49: settings. First, these were dark solitons in repulsive condensates
50: \cite{dark}. Then, bright solitons were created in an attractive
51: BEC (lithium) \cite{bright}. This was followed by the making of
52: gap solitons in a repulsive rubidium condensate loaded in a
53: periodic potential, which was induced by the optical lattice (OL),
54: i.e., interference pattern between two laser beams illuminating
55: the medium \cite{Oberthaler}.
56:
57: A challenge to the experiment is creation of 2D matter-wave
58: solitons. A natural problem in this case is the trend of solitons
59: in multidimensional attractive condensates to be unstable because
60: of the possibility of collapse in this setting \cite{collapse}. In
61: theoretical works, several approaches were proposed to stabilize
62: 2D solitons. One of them relies on the use of a full
63: two-dimensional OL \cite{2D-OL}, or its low-dimensional (quasi-1D)
64: counterpart \cite{Q1D-OL}, which can stabilize fundamental
65: solitons. In addition, 2D lattices lend stability to vortical
66: solitons \cite{2D-OL}, including higher-order vortices, and
67: ``super-vortex" complexes \cite{we}; the latter are built as
68: circular chains of compact vortices, with global vorticity imposed
69: on top of the chain. Another theoretically elaborated approach
70: relies upon the use of a nonlocal anisotropic nonlinearity induced
71: by the long-range interactions between atoms with a magnetic
72: momentum (chromium), polarized by an external field
73: \cite{dipolar}.
74:
75: An alternative mechanism proposed for the stabilization of 2D matter-wave
76: solitons is based on the Feshbach resonance (FR), which makes it possible to
77: control the value of the scattering length, i.e., as a matter of fact, an
78: effective nonlinear coefficient in the corresponding Gross-Pitaevskii
79: equation (GPE), by means of an external magnetic field \cite{FR}. Moreover,
80: the FR may switch the sign of the nonlinearity (in particular, the
81: FR-induced switch from repulsion to weak attraction was instrumental to the
82: creation of bright solitons in lithium \cite{bright}). Application of a
83: low-frequency ac magnetic field may provide for periodic alternation of the
84: nonlinearity sign in the GPE via the FR. It was predicted that the FR
85: technique based on the ac field gives rise to novel states in the 1D
86: geometry \cite{FRM}, and can stabilize 2D solitons, even in the absence of
87: the external trap \cite{2Dstabilization}. The same technique, if applied in
88: combination with a quasi-1D OL potential, may also stabilize matter-wave
89: solitons in the 3D geometry \cite{Michal}.
90:
91: It has been predicted \cite{Optical-theory}, and demonstrated in
92: experiment \cite{Optical-experiment}, that the FR can also be
93: induced by a properly tuned optical field. Then, illuminating the
94: condensate by two counterpropagating coherent laser beams, one can
95: build an OL that will provide for periodic modulation of the
96: nonlinearity coefficient along the respective spatial coordinate.
97: Solitons in the corresponding one-dimensional GPE with the
98: \emph{nonlinear} OL were recently investigated in Ref. \cite{1d},
99: where stability regions for static solitons and breathers were
100: found (motion of free solitons in the same model was recently
101: studied in Ref. \cite{Fatkhulla}, and rigorous proofs concerning
102: the stability of static solutions in this setting were reported in
103: work \cite{MIW}). The soliton dynamics in the 1D model with other
104: configurations of the spatial modulation of the nonlinearity
105: coefficient was studied in Refs. \cite{DimPan} (unlike Ref.
106: \cite{1d}, the nonlinearity coefficient did not change its sign in
107: the models considered in the latter works).
108:
109: Static spatial modulation of the nonlinearity through the FR,
110: controlled by the properly shaped magnetic or optical field, may
111: be tried as another means for the stabilization of 2D solitons,
112: which is the subject of the present work. A natural form of am
113: axisymmetric OL in the 2D geometry corresponds to the Bessel beam,
114: i.e., a nondiffracting light signal in a bulk linear medium. In
115: the case when the Bessel beam creates an effective linear
116: potential in the equation of the GPE type with self-attraction, it
117: has been shown that the corresponding radial lattice can readily
118: stabilize various types of 2D solitons \cite{Barcelona}. However,
119: our results show that, within a broad parameter region that we
120: were able to explore, stabilization of 2D solitons by means of a
121: \emph{nonlinear} Bessel lattice, i.e., within the framework of the
122: GPE whose nonlinear coefficient is $g(r)=g_{0}J_{n}(ar)$, where
123: $r$ is the radial coordinate, $g_{0}$ and $a$ are constants, and
124: $J_{n}$ is the Bessel function with $n=0,1,...$, appears to be
125: \emph{impossible} -- stationary axisymmetric soliton solutions can
126: be easily constructed, but in simulations they all suffer either
127: decay or collapse.
128:
129: Nevertheless, in this work we demonstrate that a simpler shape of the radial
130: modulation of the nonlinearity, in which it takes a constant value,
131: corresponding to self-attraction, inside a finite circle or annulus, and is
132: zero (or corresponds to self-repulsion) outside this region, is able to
133: stabilize axisymmetric 2D solitons. In addition to that, we will demonstrate
134: that the model gives rise to \emph{bistability}: the stationary solitons
135: coexist with stable breathers, that feature persistent oscillations in the
136: radial direction. In fact, the stability region of the breathers is larger
137: than that of the static solitons, extending to higher values of the norm
138: (number of atoms in the BEC).
139:
140: It should be said that, in the case of the nonlinearity controlled
141: by the optical beam through the FR mechanism, the beam with the
142: cross section in the form of a circle or annulus is not
143: divergence-free, unlike its Bessel-shaped counterpart. However,
144: this circumstance does not impede the physical realization of the
145: model, as an effectively 2D condensate can be easily trapped
146: between two blue-detuned light sheets, which strongly repel the
147: atoms, as demonstrated in the experiment \cite{pancake}. The
148: thickness of the corresponding ``pancake" is a few microns, while
149: its diameter is measured in hundreds of microns (at least), hence
150: the diffraction of the light beam within this range is completely
151: negligible.
152:
153: The paper is organized as follows. In Section 2, we give the formulation of
154: the model, and present numerical and analytical solutions for static
155: solitons. The analytical part includes a variational approximation for the
156: solutions in the general case, an exact solution for solitons supported by
157: an infinitely narrow annulus carrying the nonlinearity, and predictions for
158: the stability against radial perturbations, based on the Vakhitov-Kolokolov
159: (VK) \cite{VK,Berge'} criterion. The stability threshold for azimuthal
160: perturbations is determined by a solution of the corresponding eigenvalue
161: problem. An inference is that stability borders in the model with the
162: nonlinearity supported on the circle are completely determined by radial
163: perturbations, while in the annular model the upper stability border (in
164: terms of the soliton's norm) is controlled by azimuthal perturbations. No
165: stable solitons are possible if the annulus is relatively narrow, with the
166: ratio of inner and outer radii exceeding a critical value $\approx 0.47$. In
167: Section 3, we summarize results of direct numerical simulations of the
168: stability of fundamental stationary solutions, which precisely confirm the
169: existence of a well-defined stability region of the 2D solitons in the
170: model's parameter space, predicted in Section 2. The bistability
171: (coexistence of the stable stationary solitons and breathers) and the
172: extended stability region for the breathers are also reported in Section 3.
173: In Section 4, we briefly consider solitons with intrinsic vorticity, and
174: conclude that all the vortices are unstable (the vortex splits in two
175: fundamental solitons, each one then collapsing intrinsically). The paper is
176: concluded by Section 5.
177:
178: \section{Stationary solitons}
179:
180: \subsection{The model and numerical solutions}
181:
182: The GPE for the single-atom wave function $\psi $ in the normalized form is
183: \begin{equation}
184: i\frac{\partial \psi }{\partial t}=-\frac{1}{2}\nabla ^{2}\psi -g(r)|\psi
185: |^{2}\psi , \label{GPE}
186: \end{equation}
187: with $t$ time, $\nabla ^{2}$ the 2D Laplacian, and the nonlinearity
188: coefficient shaped, by means of the external magnetic or optical field, as
189: said above:
190: \begin{equation}
191: g(r)=\left\{
192: \begin{array}{cc}
193: 1, & \rho <r<R, \\
194: 0, & r<\rho ~\mathrm{or}~r>R.\end{array}\right. \label{g}
195: \end{equation}
196: The number of atoms is determined by the norm of the wave function,
197: \begin{equation}
198: N=2\pi \int_{0}^{\infty }|\psi (r)|^{2}rdr. \label{N}
199: \end{equation}
200: Using the scaling invariance of Eq. (\ref{GPE}), we set $R=2$, keeping $\rho
201: $ as a free parameter. Note that the model without the inner orifice, $\rho
202: =0$, is a universal one, as it contains no parameters.
203:
204: We also considered a model with the nonlinearity switched to self-repulsion,
205: i.e., $g(r)<0$, in the regions of $r<\rho $ and $r>R$. However, we focus on
206: the case with $g=0$ in these regions, as such a case is least favorable for
207: the existence of solitons, hence it provides for results which are most
208: relevant to the experimental realization of the scheme.
209:
210: Stationary solutions for fundamental solitons are looked for as $\psi =\phi
211: (r)e^{-i\mu t}$, $\ $with a real chemical potential $\mu $, and a real
212: function $\phi $ obeying the equation
213: \begin{equation}
214: 2\mu \phi +\phi ^{\prime \prime }+r^{-1}\phi ^{\prime }+2g(r)\phi ^{3}=0
215: \label{phi}
216: \end{equation}
217: (the prime stands for $d/dr$). Equation (\ref{phi}) is to be solved with the
218: boundary conditions $\phi ^{\prime }(r=0)=0$ and $\phi (r=\infty )=0$ (the
219: latter one implies that $\mu $ must be negative). The solution was searched
220: for numerically by selecting the value of $\phi (r=0)$ with which the
221: boundary condition at $r=\infty $ could be met.
222:
223: Two examples of the solution are displayed in Fig. 1(a), one for
224: $\rho =0$, i.e., the configuration with no inner ``hole", and the
225: other one with the ``hole" corresponding to $\rho =0.5$; in the
226: latter case, the solution attains a maximum at $r=\rho $, having a
227: shallow minimum at $r=0$. Families of the soliton solutions are
228: characterized by dependences $\mu (N)$, which are displayed in
229: Fig. 1(b) for $\rho =0$ and two nonzero values of $\rho $. These
230: dependences predict a necessary stability condition as per the VK
231: criterion \cite{VK}, $dN/d\mu <0$, i.e., parts of the solution
232: families beneath the turning points in Fig. 1(b) may be stable
233: (below, the turning point will be denoted as
234: $N=N_{\mathrm{cr}}^{(\mathrm{lower)}}$). In fact, the stability
235: region exists due to the fact that the attractive nonlinearity
236: acts in a finite region of space, $r<R$.
237: \begin{figure}[tbp]
238: \includegraphics[height=5cm]{fig1.eps}
239: \caption{(a) Examples of stable soliton solutions with
240: $\protect\rho =0$, $\protect\mu =-0.0399$, and $N=5.59$ (solid
241: curve) and $\protect\rho =0.5$, $\protect\mu =-0.0648$, and
242: $N=6.721$ (dashed curve). (b) Chemical potential $\protect\mu $
243: vs. norm $N$ for soliton families found numerically with
244: $\protect\rho =0$, $0.2$, and $0.5$. For $\protect\rho =0$, the
245: \textit{VK-stable} portion of the solution, i.e., one with
246: $dN/d\protect\mu <0$, is found in the interval
247: $N_{\mathrm{cr}}^{(\mathrm{lower)}}\approx
248: 5.449<N<N_{\mathrm{Townes}}\approx 5.85$. (c) $\protect\mu (N)$
249: curves predicted by the variational approximation for the same
250: cases, $\protect\rho =0$, $0.2$, and $0.5$.} \label{fig1}
251: \end{figure}
252:
253: In the absence of the inner orifice ($\rho =0$), the soliton becomes narrow
254: as $\mu $ takes large negative values. In this case, the medium seems nearly
255: uniform for the soliton, hence it approaches the shape of the well-known
256: \textit{Townes soliton}, which is a universal weakly unstable localized
257: solution of the 2D nonlinear Schr\"{o}dinger (NLS) equation with the
258: spatially uniform self-focusing nonlinearity \cite{Berge'}. Accordingly, the
259: soliton's norm approaches the value $N_{\mathrm{Townes}}\approx 5.85$, which
260: plays a critical role in the radial dynamics, being equal to the norm of the
261: Townes soliton.
262:
263: \subsection{Variational approximation}
264:
265: The fundamental soliton solutions in the present model can also be obtained
266: by means of the variational approximation (see a review of the method in
267: Ref. \cite{Progress}). To this end, we adopt the ansatz
268: \begin{equation}
269: \phi =A\exp \left( -\frac{r^{2}}{2w^{2}}\right) , \label{ansatz}
270: \end{equation}
271: with an amplitude $A$ and width $w$. The substitution of the
272: ansatz in norm (\ref{N}) and Lagrangian of Eq. (\ref{phi}),\begin{equation}
273: L=2\pi \int_{0}^{\infty }\left[ 2\mu \phi ^{2}-\left( \frac{d\phi
274: }{dr}\right) ^{2}+g(r)\phi ^{4}\right] rdr, \label{L}
275: \end{equation}
276: yields $N=\pi A^{2}w^{2}$ (we use this relation to eliminate $A$ in favor of
277: $N$), and
278: \[
279: L=2\mu N-\frac{N}{w^{2}}+\frac{N^{2}}{2\pi w^{2}}\left(
280: 1-e^{-2R^{2}/w^{2}}\right) .
281: \]Then, the variational equations, $\partial L/\partial N=0$ and $\partial
282: L/\partial w=0$, predict the following relations between the norm, width and
283: chemical potential of the soliton,
284: \begin{eqnarray}
285: \frac{2\pi }{N} &=&\left( 1-2\frac{\rho ^{2}}{w^{2}}\right) e^{-2\rho
286: ^{2}/w^{2}}-\left( 1-2\frac{R^{2}}{w^{2}}\right) e^{-2R^{2}/w^{2}},
287: \nonumber \\
288: \mu w^{2} &=&1-\frac{N}{2\pi \left( e^{-2\rho
289: ^{2}/w^{2}}-e^{-2R^{2}/w^{2}}\right) }. \label{VA}
290: \end{eqnarray}
291: The $\mu (N)$ dependence, predicted by Eq. (\ref{VA}), is shown in Fig. 1(c)
292: for several values of $\rho $. It is consistent with the numerical results
293: displayed in Fig.~1(b), although the variational approximation predicts
294: somewhat larger values of $N$.
295:
296: \subsection{The narrow-ring model}
297:
298: The simple ansatz (\ref{ansatz}) cannot predict the shape of the
299: solution with the local minimum at $r=0$, such as the one shown in
300: Fig. 1(a) for $\rho \neq 0$. The minimum becomes deeper as the
301: nonlinearity-supporting annulus narrows, which corresponds to
302: $\left( R-\rho \right) /R\rightarrow 0$. As a limit form, one can
303: take the GPE with the $\delta $-functional nonlinearity support,
304: \begin{equation}
305: i\frac{\partial \psi }{\partial t}=-\frac{1}{2}\nabla ^{2}\psi -\delta
306: (r-R)|\psi |^{2}\psi \label{delta}
307: \end{equation}
308: [the coefficient in front of the $\delta $-function is scaled to be $1$, cf.
309: Eq. (\ref{g})]. By final rescaling, one can again set $R=2$ in Eq.
310: (\ref{delta}), as was done above in Eq. (\ref{GPE}), so as to cast
311: Eq. (\ref{delta}) in a parameter-free form. It is relevant to
312: mention that a BEC configuration in the form of a narrow ring was
313: recently created in the experiment by means of an accordingly
314: shaped magnetic trap \cite{ring}.
315:
316: In the present case, the stationary wave function $\phi (r)$ obeys a linear
317: equation,
318: \begin{equation}
319: \frac{d^{2}\phi }{dr^{2}}+\frac{1}{r}\frac{d\phi }{dr}+2\mu \phi =0,
320: \label{linear}
321: \end{equation}
322: which must be solved separately for $r<R$ and $r>R$. The inner and outer
323: solutions, one with $\phi ^{\prime }(r=0)=0$ and the other
324: vanishing at $r\rightarrow \infty $, are to be linked by the
325: conditions of the continuity of $\phi (r)$ and jump of $\phi
326: ^{\prime }(r)$ at $r=R$, which follows from Eq. (\ref{delta}):
327: \begin{equation}
328: \phi ^{\prime }(r=R+0)-\phi ^{\prime }(r=R-0)=-2\left[ \phi (r=R)\right]
329: ^{3}. \label{r=R}
330: \end{equation}
331: Appropriate solutions to Eq. (\ref{linear}) are\begin{equation}
332: \phi (r)=A\left\{
333: \begin{array}{cc}
334: I_{0}\left( \sqrt{-2\mu }r\right) /I_{0}\left( \sqrt{-2\mu }R\right) , & r<R,
335: \\
336: K_{0}\left( \sqrt{-2\mu }r\right) /K_{0}\left( \sqrt{-2\mu
337: }R\right) , & r>R,\end{array}\right. \label{Bessel}
338: \end{equation}
339: where $I_{0}$ and $K_{0}$ are the modified Bessel and Hankel functions, $A$
340: is a constant, and the continuity of $\phi (r)$ at $r=R$ is
341: provided automatically. The substitution of expressions
342: (\ref{Bessel}) in Eq. (\ref{r=R}) yields
343: \begin{equation}
344: A^{2}=\sqrt{-\frac{\mu }{2}}\left[ \frac{K_{1}\left( z\right)
345: }{K_{0}\left( z\right) }+\frac{I_{1}\left( z\right) }{I_{0}\left(
346: z\right) }\right] |_{z=\sqrt{-2\mu }R}~. \label{A^2}
347: \end{equation}
348: The norm (\ref{N}) of the exact solution given by
349: Eqs. (\ref{Bessel}) and (\ref{A^2}) can also be calculated in an explicit form:
350: \begin{equation}
351: N=\pi \sqrt{-\frac{\mu }{2}}\left[ \frac{K_{1}\left( z\right)
352: }{K_{0}\left( z\right) }+\frac{I_{1}\left( z\right) }{I_{0}\left(
353: z\right) }\right] \left[ \frac{K_{1}^{2}\left( z\right)
354: }{K_{0}^{2}\left( z\right) }-\frac{I_{1}^{2}\left( z\right)
355: }{I_{0}^{2}\left( z\right) }\right] |_{z=\sqrt{-2\mu }R}~.
356: \label{NBessel}
357: \end{equation}
358: Figures 2(a) and (b) display, respectively, an example of the solution, and
359: the $\mu (N)$ dependence plotted as per the exact expression
360: (\ref{NBessel}).
361: \begin{figure}[tbp]
362: \includegraphics[height=3.5cm]{fig2.eps}
363: \caption{(a) An example of the solution of Eq. (\protect\ref{delta}). (b)
364: The $\protect\mu (N)$ dependence for the $\protect\delta $-functional model,
365: according to Eq. (\protect\ref{NBessel}).}
366: \label{fig2}
367: \end{figure}
368:
369: \subsection{Stability diagram for stationary solitons}
370:
371: Figure 2(b) shows the existence of solutions with $dN/d\mu <0$ in the model
372: with the radial $\delta $-function, which may be stable according to the VK
373: criterion. However, it can only guarantee the stability against radial
374: perturbations that do not break the axial symmetry of the solutions. On the
375: other hand, it is well known that axisymmetric ring-shaped states may be
376: easily subject to instability against azimuthal perturbations (see, e.g.,
377: Refs. \cite{azimuthal}).
378:
379: To study the stability against angular modulations in the general case [with
380: $g(x)$ taken as per Eq. (\ref{g})], including the $\delta $-functional
381: limit, as in Eq. (\ref{delta}), we take a perturbed solution as
382: \begin{equation}
383: \psi (r,\theta ,t)=e^{-i\mu t}[\phi (r)+\delta \phi _{+}(r)e^{-i\chi
384: t+im\theta }+\delta \phi _{-}(r)e^{i\chi ^{\ast }t-im\theta }], \label{pert}
385: \end{equation}
386: where $\theta $ is the angular variable, $m$ is an integer perturbation
387: index, $\chi $ is a perturbation eigenfrequency, with $\ast $
388: standing for the complex conjugation ($\chi $ may be complex
389: \cite{azimuthal}), and $\delta \phi _{\pm }(r)$ are components of
390: the respective eigenfunction. In particular, the instability
391: threshold may correspond to $\chi =0$, then the eigenfunction has
392: $\delta \phi _{+}=\delta \phi _{-}\equiv \delta \phi _{0}(r)$, and
393: the substitution of expression (\ref{pert}) in Eq. (\ref{delta})
394: and subsequent linearization lead to an equation for the
395: \textit{zero mode},
396: \begin{equation}
397: \left[ \mu +\frac{1}{2}\left(
398: \frac{d^{2}}{dr^{2}}+\frac{1}{r}\frac{d}{dr}-\frac{m^{2}}{r^{2}}\right)
399: +3g(r)\left( \phi (r)\right) ^{2}\right] \delta \phi _{0}=0.
400: \label{delta-phi}
401: \end{equation}
402: The instability threshold is achieved when real $\mu $, found as an
403: eigenvalue of Eq. (\ref{delta-phi}), coincides with the actual
404: value of the chemical potential of the unperturbed solution $\phi
405: (r)$. This way, the threshold was identified for the lowest
406: azimuthal perturbation mode, with $m=1$ (in direct simulations
407: presented in the next section, instability was observed solely
408: against the azimuthal modulations with $m=1$).
409:
410: The result of the analysis is summarized in Fig. 3, in the form of a
411: stability diagram in the $(\rho ,N)$ parameter plane. The upper dotted
412: border is the critical curve for the azimuthal instability with $m=1$, found
413: as described above, while the lower dashed curve is the existence and
414: stability border for the soliton solutions, which is identified as a set of
415: turning points of the $\mu (N)$ curves in Fig. 1. Soliton solutions
416: satisfying the VK criterion, $dN/d\mu <0$, exist above the lower border.
417: Below the upper border, they are stable against the $m=1$ azimuthal
418: disturbances, i.e., the solitons are expected to be completely stable
419: between the two curves. This expectation was verified by direct simulations,
420: see the next section.
421:
422: \begin{figure}[tbp]
423: \includegraphics[height=3.5cm]{fig3.eps}
424: \caption{The stability diagram for the soliton solutions. In the region
425: between the two borders, the stationary solitons are stable --
426: simultaneously according to the VK criterion, i.e., against radial
427: perturbations (above the lower border), and against azimuthal modulations
428: (below the upper border). }
429: \label{fig3}
430: \end{figure}
431:
432: Note that the border of the azimuthal instability in Fig. 3 is
433: located, for $\rho =0$, at a value of $N$ which is identical to
434: $N_{\mathrm{Townes}}\approx 5.85$, i.e., in the case of $\rho =0$
435: (no inner orifice), the thresholds for the collapse in the radial
436: direction, and for the breakup of the axial symmetry in the
437: azimuthal direction, are identical. The coincidence of the two
438: thresholds for $\rho =0$ can be explained. Indeed, differentiation
439: of Eq. (\ref{phi}) in $r$ shows that, for given $\phi (r)$, the
440: function $\phi ^{\prime }(r)$ solves the following linear
441: equation:
442: \begin{equation}
443: \left[ 2\mu
444: +\frac{d^{2}}{dr^{2}}+\frac{1}{r}\frac{d}{dr}-\frac{1}{r^{2}}+6g(r)\left(
445: \phi (r)\right) ^{2}\right] \phi ^{\prime }=-2g^{\prime }(r)\left(
446: \phi (r)\right) ^{3}. \label{phi-prime}
447: \end{equation}
448: If $g^{\prime }=0$, Eq. (\ref{phi-prime}) exactly
449: coincides with Eq. (\ref{delta-phi}) for $m=1$, hence the function $\phi ^{\prime }(r)$ may be
450: identified as the corresponding zero mode. Of course, when $g$ is
451: a function of $x$ defined by Eq. (\ref{g}), which means $g^{\prime
452: }(x)=\delta (r-R)$, the term on the right-hand side of Eq.
453: (\ref{phi-prime}) does not allow $\phi ^{\prime }(x)$ to be the
454: zero mode; nevertheless, in the limit of $N\rightarrow
455: N_{\mathrm{Townes}}$, the soliton shrinks to a size much smaller
456: than $R$, hence $\left( \phi (R)\right) ^{3}$ becomes vanishingly
457: small, along with the above-mentioned term. Thus, in the limit of
458: $N=N_{\mathrm{Townes}}$, the function $\phi ^{\prime }(r)$
459: provides for a solution to Eq. (\ref{delta-phi}) with $m=1$,
460: making $N=N_{\mathrm{Townes}}$ the threshold of instability to the
461: azimuthal perturbations with $m=1$.
462:
463: A notable feature of the stability diagram in Fig. 3 is that the
464: lower and upper stability borders meet and \emph{close down} the
465: stability region at $\rho =\rho _{\mathrm{\max }}\approx 0.95$,
466: which means that the nonlinearity-carrying annulus with the ratio
467: of the inner and outer radii exceeding the critical value, $\rho
468: _{\max }/R\approx \allowbreak 0.47$, cannot support stable
469: solitons. This conclusion implies that solitons cannot be stable
470: either in model (\ref{delta}) with the radial $\delta $-function.
471: Indeed, detailed consideration of that model reveals the region of
472: the azimuthal stability at $N>11.0$ and $\mu >-0.0116$, which
473: entirely belongs to the upper branch of the $\mu (N)$ curve in
474: Fig. 2(b), with $dN/d\mu >0$, i.e., the region is VK-unstable.
475:
476: \section{Direct simulations}
477:
478: To check the predictions for the stability of the solitons, and examine the
479: evolution of unstable ones, we have performed direct 2D simulations by dint
480: of the split-step Fourier method, employing a basis composed of $512\times
481: 512$ modes. The size of the integration domain was $L\times L=60\times 60$,
482: with the center of the circle or annulus set at point $\left( x,y\right)
483: =(L/2,L/2)$, and the timestep $\Delta t=0.005$.
484:
485: The simulations have confirmed the stability of the solitons in the region
486: between the lower and upper borders in Fig. 3, and instability outside of
487: this region. Figure 4(a) displays an example of the time evolution of $|\psi
488: (x,L/2)|$ (i.e., the profile of the cross section through the central point
489: along the $x$ axis) in a perturbed stable soliton, for $\rho =0$. On the
490: other hand, Figs. 4(b) and (c) demonstrate that (for the same case of $\rho
491: =0$) unstable solitons suffer collapse.
492: \begin{figure}[bp]
493: \includegraphics[height=3.5cm]{fig4.eps}
494: \caption{(a) The evolution of $|\protect\psi (x,L/2)|$ (central
495: cross section) in a stable soliton for $\protect\rho =0$,
496: $\protect\mu =-0.189$ and $N=5.62$. (b) An example of collapse of
497: an unstable soliton, for $\protect\rho =0$, $\protect\mu =-0.0179$
498: and $N=6.175$. (c) The time dependence of the field amplitude,
499: i.e., maximum value of $|\protect\phi (x,y)|$, for the same case
500: as in (b).} \label{fig4}
501: \end{figure}
502:
503: However, unstable solitons [ones belonging to the upper,
504: VK-unstable, part of the $\mu (N)$ curve in Fig. 1(b), with
505: $dN/d\mu >0$] whose norm is taken below a critical value,
506: $N_{\mathrm{cr}}^{(\mathrm{upper)}}\approx 5.99$ (for $\rho =0$),
507: which is \emph{higher} than the norm $N_{\mathrm{Townes}}\approx
508: 5.85$ of the Townes soliton in the two-dimensional NLS\ equation,
509: neither collapse nor decay into radiation (in the NLS equation, a
510: pulse with $N<N_{\mathrm{Townes}}$ is bound to decay in the 2D
511: uniform space). Instead, the unstable soliton rearranges itself
512: into a \emph{stable breather}. Figures 5(a) and (b) display an
513: example of the evolution of breathers. In the simulations, the
514: breathers remain stable indefinitely long, their oscillations
515: getting more regular as $N$ decreases. The amplitude of the
516: oscillations, which we define as the root-mean square of the
517: variation of the soliton's amplitude, $A(t)\equiv |u(x=y=L/2,t)|$,
518: decreases with $N$, and it vanishes at another critical value,
519: $N_{\mathrm{cr}}^{(\mathrm{lower)}}\approx 5.449$. Up to the
520: numerical accuracy, the latter one is precisely the smallest value
521: of $N$ at which the stationary solitons exist for $\rho =0 $, see
522: Fig. 1(b). Thus, $N=N_{\mathrm{cr}}^{(\mathrm{lower)}}$ is not
523: only the point of the merger of the VK-stable and VK-unstable
524: branches of the solutions, but also the one at which the breathers
525: merge into the static solitons.
526: \begin{figure}[tbp]
527: \includegraphics[height=3.5cm]{fig5.eps}
528: \caption{(a) The time evolution of the cross-section profile,
529: $|\protect\psi (x,L/2)|$, of a breather, for $\protect\rho =0$,
530: $\protect\mu =-0.0399$, and $N=5.59$. The initial amplitude of the
531: soliton is $A(0)=0.6$. (b) Evolution of the amplitude of the
532: breathing soliton for the same case. (c) The amplitude of
533: intrinsic oscillations of the breather as a function of $A(0)$.}
534: \label{fig5}
535: \end{figure}
536:
537: We stress that the existence of the stable \emph{axisymmetric} breathers up
538: to $N_{\mathrm{cr}}^{(\mathrm{upper)}}\approx 5.99$ does not contradict the
539: fact that the symmetry-breaking azimuthal instability occurs, for $\rho =0$,
540: at $N>N_{\mathrm{Townes}}\approx 5.85$, as explained above. Indeed, the
541: latter pertains to the angular instability of the static solitons, but not
542: breathers.
543:
544: A noteworthy consequence of these results is the
545: \emph{bistability}: in the entire interval of values of the norm,
546: $N_{\mathrm{cr}}^{(\mathrm{lower)}}\approx
547: 5.449<N<N_{\mathrm{Townes}}\approx 5.85$, where stable stationary
548: solitons are found (for $\rho =0$), they coexist with breathers.
549: On the other hand, in the adjacent interval,
550: $5.85<N<N_{\mathrm{cr}}^{(\mathrm{upper)}}\approx 5.99$, only
551: stable breathers are possible (and no stable objects exist for
552: $N>5.99$).
553:
554: Stable breathers and bistability were found for $\rho >0$ as well. We note
555: that stable breathers were also found in the model based on the
556: one-dimensional GPE with a nonlinear OL [i.e., the nonlinearity coefficient
557: modulated in space as $\cos (kx)$] \cite{1d}. In the latter model,
558: bistability was observed too, as the breathers exist at the same values of
559: the norm at which stable stationary solitons are found.
560:
561: As said above, all the solitons which are stable against collapse
562: in the model with $\rho =0$, are stable too against the azimuthal
563: perturbations. Actual instability against the azimuthal mode
564: (\ref{pert}) with $m=1$ occurs at $\rho >0$. To study the
565: azimuthal instability in direct simulations, we used an initial
566: condition in the form of a stationary soliton subjected to a weak
567: angular deformation. Figure 6 displays a typical example of the
568: development of the azimuthal instability for $\rho =0.55$. As a
569: result, the soliton does not split into fragments, which is a
570: generic result of the azimuthal instability of vortex-ring
571: solitons in uniform media \cite{ring}, but rather shifts from the
572: central point, $(x,y)=(30,30)$, to a position centered at
573: $(x,y)\approx (29,30)$. Because the norm of the soliton exceeds
574: $N_{\mathrm{Townes}}$, it then develops intrinsic collapse at the
575: new position, where the hole does not essentially affect its
576: dynamics. The shift of the soliton off the center and subsequent
577: collapse were found to be a generic outcome of the development of
578: the azimuthal instability. This feature can be easily explained by
579: the fact obvious in Fig. 3: all the solitons which are subject to
580: the azimuthal instability have $N>N_{\mathrm{Townes}}$, hence they
581: should collapse after being displaced away from the hole.
582: \begin{figure}[tbp]
583: \includegraphics[height=3.5cm]{fig6.eps}
584: \caption{Snapshots of contour maps of $|\protect\psi (x,y)|$ for
585: an azimuthally unstable soliton, taken at $t=5$ (a), $t=100$ (b)
586: and $t=122$ (c). In this case, $\protect\rho =0.55$, $\protect\mu
587: =-0.0782$ and $N=7.055$.} \label{fig6}
588: \end{figure}
589:
590: As said above, direct simulations corroborate the stability of the
591: stationary solitons in the region between the two borders in Fig.~3. We
592: illustrate this conclusion in Fig. 7, which displays the time evolution of
593: the field amplitude (maximum value of $|\psi (x,y)|$) for $\rho =0.2$ and
594: three different values of the norm. The first soliton, with $N=5.702$,
595: belongs to the stability region in Fig. 3, and it is seen to be stable
596: indeed. Two other solitons, with $N=5.994$ and $N=6.01$, are azimuthally
597: unstable, which eventually leads to the collapse (after the spontaneous
598: off-center shift, as shown in Fig. 6). Note that, as $N=5.994$ is close to
599: the border of the azimuthal instability, the respective instability
600: development time is large.
601: \begin{figure}[tbp]
602: \includegraphics[height=3.5cm]{fig7.eps}
603: \caption{The field amplitude vs. time, for weakly perturbed
604: solitons with $\protect\rho =0.2$ and (a) $N=5.702$, $\protect\mu
605: =-0.0758$, (b) $N=5.994$, $\protect\mu =-0.164$, and (c) $N=6.01$,
606: $\protect\mu =-0.169$.} \label{fig7}
607: \end{figure}
608:
609: \section{Vortex solitons}
610:
611: Besides the fundamental solitons considered above, Eq. (\ref{GPE}) also
612: gives rise to vortex solitons, in the form of $\psi =\phi _{S}(r)e^{-i\mu
613: t+iS\theta }$, with integer vorticity $S$ and real function $\phi (r)$
614: satisfying the equation [cf. Eq. (\ref{phi})]
615: \begin{equation}
616: \phi _{S}^{\prime \prime }+r^{-1}\phi _{S}^{\prime }-S^{2}r^{-2}\phi
617: _{S}+2g(r)\phi _{S}^{3}+2\mu \phi _{S}=0. \label{phi-S}
618: \end{equation}
619: In particular, in the model with the radial $\delta $-function,
620: see Eq. (\ref{delta}), the vortex solution can be found in an exact form,
621: cf. Eqs. (\ref{Bessel}) and (\ref{A^2}):
622: \[
623: \phi _{S}(r)=A\left\{
624: \begin{array}{cc}
625: I_{S}\left( \sqrt{-2\mu }r\right) /I_{S}\left( \sqrt{-2\mu }R\right) , & r<R,
626: \\
627: K_{S}\left( \sqrt{-2\mu }r\right) /K_{S}\left( \sqrt{-2\mu
628: }R\right) , & r>R,\end{array}\right.
629: \]
630: \[
631: A^{2}=\frac{1}{2}\sqrt{-\frac{\mu }{2}}\left[ \frac{I_{S+1}\left(
632: x\right) +I_{S-1}(x)}{I_{S}\left( x\right) }+\frac{K_{S+1}\left(
633: x\right) +K_{S-1}(x)}{K_{S}\left( x\right) }\right]
634: |_{x=\sqrt{-2\mu }R}~.
635: \]
636: The norm of this solution can also be calculated in an analytical form.
637:
638: An example of a vortex soliton, and the dependence $\mu (N)$ for these
639: solutions, are displayed in Figs. 8(a) and (b), for $\rho =0$ and $S=1$. The
640: figures show that a part of the solution family has $dN/d\mu <0$, hence it
641: is stable against radial perturbations, pursuant to the VK criterion.
642: \begin{figure}[tbp]
643: \includegraphics[height=3.5cm]{fig8.eps}
644: \caption{(a) A typical example of profile $\protect\phi (r)$ for
645: the vortex soliton with $\protect\rho =0$, $S=1$, and
646: $N=23.9$,~$\protect\mu =-1.33$. (b) The $\protect\mu (N)$
647: dependence for the vortex-soliton family with $\protect\rho =0$
648: and $S=1$.}
649: \end{figure}
650:
651: Comparing Fig. 8(b) to Fig. 1(b), one observes that the norm of
652: the vortices is much larger than the norm of the fundamental
653: solitons, which suggest that the vortex soliton may break up into
654: a set of fundamental ones (as said above, this is a typical
655: outcome of the development of azimuthal instability of vortex
656: solitons in uniform media \cite{ring}). Indeed, further analysis
657: demonstrates that the vortex solitons with $S=1$ are unstable
658: against azimuthal disturbances with $m=2$ [cf. Eq. (\ref{pert})].
659: An example, displayed in Fig. 9 for $\rho =0$, shows that the
660: instability splits the vortex into a set of two zero-vorticity
661: solitons, each then collapsing intrinsically, as its norm exceeds
662: the critical value, $N_{\mathrm{Townes}}\approx 5.85$. Before the
663: collapse, the soliton pair rotates in the counter-clockwise
664: direction. No example of a stable vortex soliton was found in the
665: model.
666: \begin{figure}[tbp]
667: \includegraphics[height=3.5cm]{fig9.eps}
668: \caption{Instability of vortex solitons is illustrated by a set of three
669: snapshots of the contour map of $|\protect\psi (x,y)|$ for a vortex with $S=1
670: $, $\protect\rho =0$, $N=23.9$ and $\protect\mu =-1.33$, taken at $t=50$
671: (a), $85$ (b), and $90$ (c).}
672: \label{fig9}
673: \end{figure}
674:
675: \section{Conclusion}
676:
677: The purpose of the work was to investigate the two-dimensional
678: Gross-Pitaevskii equation in which the attractive nonlinearity is limited to
679: a finite region in the form of a circle or annulus, including the case of a
680: narrow ring. In Bose-Einstein condensates trapped between a pair of
681: blue-detuned light sheets, this configuration can be implemented through the
682: Feshbach resonance by means of a properly configured magnetic or optical
683: field which controls the scattering length of collisions between atoms.
684: Using numerical and analytical methods, we have found a stability region for
685: axisymmetric fundamental (zero-vorticity) solitons in the model, which is
686: impossible in the case of the spatially uniform nonlinearity. It is
687: noteworthy that the stability borders of the solitons in the model with the
688: nonlinearity supported on the circle are completely determined by radial
689: perturbations, while in the annular model the upper stability border is set
690: by azimuthal modulations. The stability is limited to relatively broad
691: annuli, with the ratio of the inner and outer radii smaller than a critical
692: value, $\rho _{\max }/R\approx 0.47$. Moreover, the model gives rise to
693: bistability, as the stationary solitons coexist with stable axisymmetric
694: breathers. The stability region of the breathers extends, in terms of their
695: norm, to values exceeding the critical value corresponding to the Townes
696: soliton. The collapse threshold strongly increases with the radius of the
697: inner hole. Vortex solitons were constructed too, but they are unstable.
698: Essentially the same results were obtained also for a model in which,
699: outside of the circle or annulus, the nonlinearity is not zero but rather
700: repulsive (that case is not considered in the paper, as the configuration
701: with the zero nonlinearity is the most challenging one, as concerns the
702: stability of solitons). The results reported in this work suggest a
703: straightforward possibility to create stable two-dimensional matter-wave
704: solitons in Bose-Einstein condensates.
705:
706: \section*{Acknowledgements}
707:
708: We acknowledge valuable discussions with D. Frantzeskakis and P. G.
709: Kevrekidis. B. A. M. appreciates hospitality of the Department of Applied
710: Science for Electronics and Materials at the Interdisciplinary Graduate
711: School of Engineering Sciences, Kyushu University (Fukuoka, Japan). This
712: work was partly supported by the Grant-in-Aid for Scientific Research
713: No.17540358 from the Ministry of Education, Culture, Sports, Science and
714: Technology of Japan, and by the Israel Science Foundation through the
715: Center-of-Excellence in Research grant No. 8006/03.
716:
717: \begin{thebibliography}{99}
718: \bibitem{dark} S.~Burger, K.~Bongs, S.~Dettmer, W.~Ertmer, and K.~Sengstock,
719: Phys. Rev. Lett. \textbf{83}, 5198 (1999); J.~Denschlag, J.~E.~Simsarian,
720: D.~L.~Feder, C. W. Clark, L. A. Collins, J. Cubizolles, L. Deng, E. W.
721: Hagley, K. Helmerson, W. P. Reinhardt, S. L. Rolston, B. I. Schneider, and
722: W. D. Phillips, Science \textbf{287}, 97 (2000).
723:
724: \bibitem{bright} L.~Khaykovich, F. Scherck, G. Ferrari, T. Bourdel, J.
725: Cubizolles, L. D. Carr, Y. Castin, C. Salomon, Science \textbf{296}, 1290
726: (2002), K.~E.~Strecker, G. B. Partridge, A. G. Truscott and R. G. Hulet,
727: Nature \textbf{417}, 153 (2002).
728:
729: \bibitem{Oberthaler} B. Eiermann, Th. Anker, M. Albiez, M. Taglieber, P.
730: Treutlein, K.-P. Marzlin, and M. K. Oberthaler, Phys. Rev. Lett.
731: \textbf{92}, 230401 (2004).
732:
733: \bibitem{collapse} J. M. Gerton, D. Strekalov, I. Prodan, and R. G. Hulet,
734: %%Direct observation of growth and collapse of a Bose-Einstein condensate
735: %%with attractive interactions,
736: Nature \textbf{408}, 692%%-695
737: (2000); E. A. Donley, N. R. Claussen, S. L. Cornish, J. L. Roberts, E. A.
738: Cornell, and C. E. Wieman, %%Dynamics of collapsing and exploding
739: %%Bose-Einstein condensates,
740: \textit{ibid}. \textbf{412}, 295%%-299
741: (2001).
742:
743: \bibitem{2D-OL} B. B. Baizakov, B. A. Malomed and M. Salerno, Europhys.
744: Lett. \textbf{63}, 642 (2003); J. Yang and Z. Musslimani, Opt.
745: Lett. \textbf{23}, 2094 (2003).
746:
747: \bibitem{Q1D-OL} B. B. Baizakov, B. A. Malomed and M. Salerno, Phys. Rev. A
748: \textbf{70}, 053613 (2004); B. B. Baizakov, M. Salerno, and B. A. Malomed,
749: in \textit{Nonlinear Waves: Classical and Quantum Aspects}, (Kluwer Academic
750: Publishers, Dordrecht, 2004) edited by F. Kh. Abdullaev and V. V. Konotop,
751: p. 61; also available at \newline
752: http://rsphy2.anu.edu.au/\symbol{126}asd124/Baizakov\_2004\_61\_Nonlinear
753: Waves.pdf
754:
755: \bibitem{we} H. Sakaguchi and B. A. Malomed, Europhys. Lett., in press
756: (2005).
757:
758: \bibitem{dipolar} P. Pedri and L. Santos, Phys. Rev. Lett. \textbf{95},
759: 200404 (2005).
760:
761: \bibitem{FR} S.~Inouye, M. R. Andrews, J. Stenger, H.-J. Miesner, D. M.
762: Stamper-Kurn, and W. Ketterle, Nature \textbf{392}, 151 (1998);
763: E.~A.~Donley, N. R. Claussen, S. L. Cornish, J. L. Roberts, E. A. Cornell,
764: and C. E. Wieman, Nature \textbf{412}, 295 (2001); H.~Saito and M.~Ueda,
765: Phys. Rev. A \textbf{65}, 033624 (2002).
766:
767: \bibitem{FRM} P. G. Kevrekidis, G. Theocharis, D. J. Frantzeskakis and B. A.
768: Malomed, Phys. Rev. Lett. \textbf{90}, 230401 (2003).
769:
770: \bibitem{2Dstabilization} H. Saito and M. Ueda, Phys. Rev. Lett. \textbf{90}, 040403 (2003);
771: F. Kh. Abdullaev, J. G. Caputo, R. A. Kraenkel, and B. A.
772: Malomed, Phys. Rev. A \textbf{67}, 013605 (2003); G. D.
773: Montesinos, V. M. P\'{e}rez-Garc\'{\i}a, and P. J. Torres, Physica
774: D \textbf{191}, 193 (2004).
775:
776: \bibitem{Michal} M. Trippenbach, M. Matuszewski, and B. A. Malomed,
777: Europhys. Lett. \textbf{70}, 8 (2005); M.~Matuszewski, E.~Infeld,
778: B.~A.~Malomed, and M.~Trippenbach, Phys. Rev. Lett. \textbf{95}, 050403
779: (2005).
780:
781: \bibitem{Optical-theory} P. O. Fedichev, Yu. Kagan, G. V. Shlyapnikov, and
782: J. T. M. Walraven, Phys. Rev. Lett. \textbf{77}, 2913 (1996).
783:
784: \bibitem{Optical-experiment} M. Theis, G. Thalhammer, K. Winkler, M.
785: Hellwig, G. Ruff, R. Grimm, and J. H. Denschlag, Phys. Rev. Lett.
786: \textbf{93}, 123001 (2004).
787:
788: \bibitem{1d} H. Sakaguchi and B. A. Malomed, Phys. Rev. E \textbf{72},
789: 046610 (2005).
790:
791: \bibitem{Fatkhulla} F. Kh. Abdullaev and J. Garnier, e-print
792: cond-mat/0511264.
793:
794: \bibitem{MIW} G. Fibich, Y. Sivan, and M. I. Weinstein, preprint
795: (http://www.columbia.edu/\symbol{126}miw2103/recentpub.html).
796:
797: \bibitem{DimPan} G. Theocharis, P. Schmelcher, P. G. Kevrekidis, and D. J.
798: Frantzeskakis, Phys. Rev. A \textbf{72}, 033614 (2005); e-print
799: cond-mat/0509471 (2005).
800:
801: \bibitem{Barcelona} Y. V. Kartashov, V. A. Vysloukh, and L. Torner, Phys.
802: Rev. Lett. \textbf{93}, 093904 (2004); \textit{ibid}. \textbf{94}, 043902
803: (2005); Y. V. Kartashov, A. A. Egorov, V. A. Vysloukh, and L. Torner, J.
804: Opt. B: Quantum Semiclass. Opt. \textbf{6}, 444 (2004).
805:
806: \bibitem{pancake} A. G\"{o}rlitz, J. M. Vogels, A. E. Leanhardt, C. Raman,
807: T. L. Gustavson, J. R. Abo-Shaeer, A. P. Chikkatur, S. Gupta, S. Inouye, T.
808: Rosenband, and W. Ketterle, Phys. Rev. Lett. \textbf{87}, 130402 (2001).
809:
810: \bibitem{VK} M. G. Vakhitov and A. A. Kolokolov, Izv. Vuz. Radiofiz.
811: \textbf{16}, 1020 (1973) [in Russian; English translation: Sov. J. Radiophys.
812: Quantum Electr. \textbf{16}, 783 (1973)].
813:
814: \bibitem{Berge'} L. Berg{\'{e}}, Phys. Rep. \textbf{303}, 260 (1998).
815:
816: \bibitem{Progress} B. A. Malomed, in \textit{Progress in Optics}, ed. by E.
817: Wolf (North-Holland, Amsterdam, 2002), Vol. 43, p. 71.
818:
819: \bibitem{ring} S. Gupta, K. W. Murch, K. L. Moore, T. P. Purdy, and D. M.
820: Stamper-Kurn, Phys. Rev. Lett. \textbf{95}, 143201 (2005).
821:
822: \bibitem{azimuthal} I. Towers, A. V. Buryak, R. A. Sammut, B. A. Malomed, L.
823: C. Crasovan, and D. Mihalache, Phys. Lett. A \textbf{288}, 292
824: (2001); a short review of the topic was given in: B. A. Malomed,
825: G. D. Peng, P. L. Chu, I. Towers, A. V. Buryak, and R. A. Sammut,
826: Pramana J. Phys. \textbf{57}, 1061 (2001).
827: \end{thebibliography}
828:
829: \end{document}
830: