1: \documentclass[runningheads]{llncs}

2: 

3: \RequirePackage[english]{babel}

4: \usepackage{graphicx,epsfig}

5: 

6: \pagestyle{headings}

7: 

8: \begin{document}

9: 

10: \title{Nodal two-dimensional solitons in

11: nonlinear parametric resonance}

12: 

13: \author{N.V. Alexeeva\inst{1} \and

14: E.V. Zemlyanaya\inst{2} }

15: 

16: \institute{Department of Mathematics, University of Cape Town,

17: South Africa; \email{nora@figaro.mrh.uct.ac.za} \and Joint

18: Institute for Nuclear Research, Dubna 141980, Russia;

19: \email{elena@jinr.ru} }

20: 

21: 

22: \titlerunning{Nodal two-dimensional solitons in nonlinear parametric resonance}

23: \toctitle{Nodal two-dimensional solitons in nonlinear parametric

24: resonance}

25: 

26: \authorrunning{N.V. Alexeeva and E.V. Zemlyanaya}

27: \tocauthor{N.V. Alexeeva and E.V. Zemlyanaya}

28: 

29: \maketitle

30: 

31: \begin{abstract}

32: The parametrically driven damped nonlinear Schr\"odinger equation

33: serves as an amplitude equation for a variety of resonantly forced

34: oscillatory systems on the plane. In this note, we consider its

35: nodal soliton solutions. We show that although the nodal solitons

36: are stable against radially-symmetric perturbations for

37: sufficiently large damping coefficients, they are always unstable

38: to azimuthal perturbations. The corresponding break-up scenarios

39: are studied using direct numerical simulations. Typically, the

40: nodal solutions break into symmetric ``necklaces" of stable

41: nodeless solitons.

42: 

43: \end{abstract}

44: 

45: {\bf 1.} Two-dimensional  localised oscillating structures,

46: commonly referred to as oscillons, have been detected in

47: experiments on vertically vibrated layers of granular material

48: \cite{Swinney}, Newtonian fluids  and suspensions

49: \cite{Faraday,Astruc}. Numerical simulations  established the

50: existence of stable oscillons in a variety of pattern-forming

51: systems, including the Swift-Hohenberg and Ginsburg-Landau

52: equations, period-doubling maps with continuous spatial coupling,

53:  semicontinuum theories and hydrodynamic models

54: \cite{Astruc,numerical}. These simulations provided a great deal

55: of insight into the phenomenology of the oscillons; however, the

56: mechanism by which they acquire or loose their stability remained

57: poorly understood.

58: 

59: In order to elucidate this mechanism, a simple model of a

60: parametrically forced oscillatory medium was proposed recently

61: \cite{us}. The model comprises a two-dimensional lattice of

62: diffusively coupled, vertically vibrated pendula. When driven at

63: the frequency close to their double natural frequency, the pendula

64: execute almost synchronous librations whose slowly varying

65: amplitude satisfies the 2D parametrically driven, damped nonlinear

66: Schr\"odinger (NLS) equation. The NLS equation was shown to

67: support radially-symmetric, bell-shaped (i.e. nodeless) solitons

68: which turned out to be stable for sufficiently large values of the

69: damping coefficient. These stationary solitons of the amplitude

70:  equation correspond to the spatio-temporal envelopes of

71: the oscillons in the original lattice system. By reducing the NLS

72: to a finite-dimensional system in the vicinity of the soliton, its

73: stabilisation mechanism (and hence, the oscillon's stabilisation

74: mechanism) was clarified \cite{us}.

75: 

76: In the present note we consider a more general class of

77: radially-symmetric solitons of the parametrically driven, damped

78: NLS on the plane, namely solitons with nodes. We will demonstrate

79: that these solitons are unstable against azimuthal modes, and

80: analyse the evolution of this instability.

81: 

82: 

83: 

84: {\bf 2.} The  parametrically driven, damped NLS equation has the

85: form:

86: \begin{equation}

87: i \psi_t + \nabla^2 \psi+ 2|\psi|^2 \psi - \psi = h \psi^* -i

88: \gamma \psi. \label{2Dnls}

89: \end{equation}

90: Here $\nabla^2=\partial^2 /\partial x^2 +\partial^2 /\partial

91: y^2.$ Eq.(\ref{2Dnls}) serves as an amplitude equation for a wide

92: range of nearly-conservative two-dimensional oscillatory systems

93: under parametric forcing. This equation was also used as a

94: phenomenological model of nonlinear Faraday resonance in water

95: \cite{Astruc}. The coefficient $h>0$ plays the role of the

96: driver's strength and $\gamma>0$ is the damping coefficient.

97: 

98: \begin{figure}[t]

99: \begin{center}

100: %%%\psfig{file=fff.eps,width=8cm,height=5.cm}

101: \psfig{file=fig0.ps,width=8cm,height=5.cm}

102: \end{center}

103: \caption{\sf Solutions of eq.(\ref{master}):

104:  ${\cal

105: R}_0(r)$ (thin continuous line), ${\cal R}_1(r)$ (thick  line),

106: ${\cal R}_2(r)$ (dashed). } \label{fig0}

107: \end{figure}

108: 

109: We start with the discussion of its nodeless solitons and their

110: stability. The exact (though not explicit) stationary

111: radially-symmetric solution is  given by

112: \begin{equation}

113: \psi_0= {\cal A} e^{- i \theta} \, {\cal R}_0({\cal A} r),

114: \label{soliton}

115: \end{equation}

116: where $r^2=x^2+y^2$,

117:  $${\cal A}^2=1 + \sqrt{h^2-\gamma^2}, \quad

118: \theta= \frac12 \arcsin\left(\frac{\gamma}{h}\right),$$ and ${\cal

119: R}_0(r)$ is the bell-shaped nodeless solution of the equation

120: \begin{equation}

121:  {\cal R}_{rr} +\frac 1 r {\cal R} - {\cal R} + 2 {\cal R}^3 =0

122:   \label{master}

123: \end{equation}

124: with the boundary conditions ${\cal R}_r(0)={\cal R}(\infty)=0$.

125: (Below we simply write ${\cal R}$ for ${\cal R}_0$.)

126: %Here $ \nabla_r^2=\partial_r^2+ \frac{D-1}{r} \partial_r$.

127: Solutions of eq.(\ref{master}) are well documented in literature

128: \cite{Rypdal}; see Fig.\ref{fig0}.

129: 

130: \begin{figure}[t]

131: \begin{center}

132: \psfig{file=spr_f1.ps,width=8cm,height=5.cm}

133:   %previous file nn.ps

134: \end{center}

135: \caption{\sf Stability diagram for two-dimensional solitons. The

136: $(\gamma, h-\gamma)$-plane is used for visual clarity. No

137: localised or periodic attractors exist for $h< \gamma$ (below the

138: horisontal axis).  The region of stability of the soliton

139: $\psi_{0}$  lies  to the right of the

140:  solid curve marked ``$n=0$". Also shown are the regions of stability of

141:  the solitons $\psi_1$ and $\psi_2$ with respect to the radially-symmetric

142:  perturbations. (These lie to the right of the corresponding curves in the

143:  figure.)

144: } \label{chart}

145: \end{figure}

146: 

147: {\bf 3.} To examine the stability of the solution (\ref{soliton})

148:  with nonzero $h$ and $\gamma$, we

149: linearise eq.(\ref{2Dnls}) in the small perturbation

150: $$

151: \delta \psi({\bf x},t)= e^{(\mu-\Gamma) {\tilde t}-i

152: \theta_{\pm}}[u({\tilde {\bf x}})+i v({\tilde {\bf x}})],

153: %\label{pert}

154: $$

155: where ${\tilde {\bf x}}={\cal A} {\bf x}$, ${\tilde t}={\cal A}^2

156: t$. This yields an eigenvalue problem

157: \begin{equation}

158:  L_1 u =- (\mu+ \Gamma )v, \quad

159: (L_0  - \epsilon)v = (\mu- \Gamma )u, \label{EV_gamma}

160: \end{equation}

161: where $\Gamma = \gamma /{\cal A}^2$ and the operators

162: \begin{equation}

163: L_0  \equiv - {\tilde \nabla}^2+1 -2 {\cal R}^2({\tilde r}), \quad

164: L_1 \equiv L_0 -4 {\cal R}^2({\tilde r}), \label{op_nabla}

165: \end{equation}

166: with ${\tilde \nabla}^2= \partial^2/\partial {\tilde

167: x}^2+\partial^2/\partial {\tilde y}^2$. (We are dropping the

168: tildas below.) For further convenience, we introduce the positive

169: quantity $\epsilon=  2 \sqrt{h^2-\gamma^2}/{\cal A}^2$.

170:  Fixing $\epsilon$ defines a curve

171: on the $(\gamma,h)$-plane:

172: \begin{equation}

173: h= \sqrt{ \epsilon^2/(2- \epsilon)^2 + \gamma^2}.

174: \label{h_of_gamma}

175: \end{equation}

176: Introducing

177: \begin{equation}

178: \lambda^2=\mu^2-\Gamma^2 \label{mugamma}

179: \end{equation}

180: and performing the transformation \cite{BBK} $$v({\bf x}) \to (\mu

181: +\Gamma) \lambda^{-1} v({\bf x}),$$ reduces eq.(\ref{EV_gamma})

182: to a {\it one-}parameter eigenvalue problem:

183: \begin{equation}

184: (L_0-\epsilon) v= \lambda u, \quad L_1 u= -\lambda v. \label{EV}

185: \end{equation}

186: 

187: We first consider the stability with respect to radially symmetric

188: perturbations $u=u(r), \, v=v(r)$. In this case the operators

189: (\ref{op_nabla}) become

190: \begin{equation}

191: L_0  =- \frac{d^2}{dr^2} -\frac{1}{r} \frac{d}{dr}+1 -2 {\cal

192: R}^2(r), \quad L_1 = L_0 -4 {\cal R}^2(r). \label{op_radsym}

193: \end{equation}

194: In the absence of the damping and driving, all localised initial

195: conditions in the unperturbed 2D NLS equation are known to either

196: disperse or blow-up in finite time \cite{Rypdal,Malkin,collapse}.

197: It turned out, however, that the soliton $\psi_0$ stabilises as

198: the damping $\gamma$ is increased above a certain value \cite{us}.

199: The stability condition

200:  is

201: $\gamma \ge \gamma_c$, where

202:  \begin{equation}

203: \gamma_c=\gamma_c(\epsilon) \equiv \frac{2}{2-\epsilon} \cdot

204: \frac{{\rm Re\/} \lambda(\epsilon)

205:   \, {\rm Im\/} \lambda(\epsilon)}

206:  {\sqrt{({\rm Im\/} \lambda)^2- ({\rm Re\/} \lambda)^2}}.

207:  \label{gce}

208:  \end{equation}

209: We obtained $\lambda(\epsilon)$ by solving the eigenvalue problem

210: (\ref{EV}) directly.   Expressing $\epsilon$ via $\gamma_c$ from

211: (\ref{gce}) and feeding into (\ref{h_of_gamma}), we get the

212: stability boundary on the $(\gamma,h)$-plane (Fig.\ref{chart}).

213: 

214: {\bf 4.} To study the stability to asymmetric perturbations we

215: factorise, in (\ref{EV}),

216: $$u({\bf x})={\tilde u}(r) e^{i m \varphi}, \quad v({\bf x})={\tilde v}(r)

217: e^{i m \varphi},$$ where $\tan \varphi= y/x$ and $m$ is an

218: integer. The functions ${\tilde u}(r)$ and ${\tilde v}(r)$ satisfy

219: the  eigenproblem (\ref{EV}) where the operators (\ref{op_radsym})

220: should be replaced by

221: \begin{equation}

222: L_0^{(m)}

223:  \equiv L_0

224:   +{m^2}/{r^2},

225: \quad L_1^{(m)} \equiv L_1+ {m^2}/{r^2}, \label{m2r2}

226: \end{equation}

227: respectively. This modified eigenvalue problem can be analysed  in

228: a similar way to eqs.(\ref{EV}). It is not difficult to

229: demonstrate  that all discrete eigenvalues of (\ref{EV}) (if any

230: exist) have to be pure imaginary in this case, and hence the

231: azimuthal perturbations do not lead to any instabilities of the

232: solution in question \cite{us}.

233: 

234: \begin{figure}[t]

235: \begin{center}

236: \psfig{file=f3a.ps,width=6cm,height=4.7cm}\\

237: \psfig{file=f3b.ps,width=6cm,height=4.7cm}

238: \psfig{file=f3c.ps,width=6cm,height=4.7cm}

239: \end{center}

240: \caption{\sf The discrete eigenvalues  of the linearised operator

241: (\ref{EV}) for the one-node soliton, $\psi_1$. Panels (a) and (b)

242: show the {\it complex\/} eigenvalues, $\mbox{Im} \, \lambda$ vs

243: $\mbox{Re} \, \lambda$. Arrows indicate the direction of increase

244: of $\epsilon$.

245:  Panel (c) shows the {\it real\/} eigenvalues,

246: as functions of $\epsilon$.} \label{fig2}

247: \end{figure}

248: 

249: 

250: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

251: \begin{figure}[t]

252: \begin{center}

253: \psfig{file=f4a.ps,width=6cm,height=4.7cm}

254: \psfig{file=f4b.ps,width=6cm,height=4.7cm}\\

255: \psfig{file=f4c.ps,width=6cm,height=4.7cm}

256: \psfig{file=f4d.ps,width=6cm,height=4.7cm}\end{center}

257: \caption{\sf The discrete eigenvalues  of the linearised operator

258: (\ref{EV}) for the two-node soliton, $\psi_2$. Panels (a),(b), and

259: (c) show the complex eigenvalues, $\mbox{Im} \, \lambda$ vs

260: $\mbox{Re} \, \lambda$. Panel (d) shows the real eigenvalues, as

261: functions of $\epsilon$.} \label{fig3}

262: \end{figure}

263: 

264: 

265: \begin{figure}[t]

266: \begin{center}

267: \psfig{file=fig5n.eps}

268: %\psfig{file=fig5a.ps,width=6cm,height=4.cm}

269: %\psfig{file=fig5b.ps,width=6cm,height=4.cm}\\

270: %\psfig{file=fig5c.ps,width=6cm,height=4.cm}

271: %\psfig{file=fig5d.ps,width=6cm,height=4.cm}

272: \end{center}

273: \caption{\sf Evolution of the azimuthal instability of the

274: one-node soliton. (a): the initial condition, soliton $\psi_1$ ;

275: (b) and (c):

276:  dissociation of the ring-like ``valley'' into 4 nodeless solitons;

277:  (d): divergence of the fragments.

278:  Here $\gamma=3.5$ and $h=3.6$;  shown is $\mbox{Re} \, \psi$.

279:  Note the change of the

280: vertical scale in (d).} \label{fig4}

281: \end{figure}

282: 

283: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

284: 

285: \begin{figure}

286: \begin{center}

287: \psfig{file=fig6n.eps}

288: %\psfig{file=fig6a.ps,width=6cm,height=4.cm}

289: %\psfig{file=fig6b.ps,width=6cm,height=4.cm}\\

290: %\psfig{file=fig6c.ps,width=6cm,height=4.cm}

291: %\psfig{file=fig6d.ps,width=6cm,height=4.cm}\\

292: %\psfig{file=fig6e.ps,width=6cm,height=4.cm}

293: %\psfig{file=fig6f.ps,width=6cm,height=4.cm}

294: \end{center}

295: \caption{\sf The evolution of the azimuthal instability of the

296: two-node soliton $\psi_2$. (a): the initial condition; (b)-(c):

297: the rapid  dissociation of the ``valley" into 4 nodeless solitons;

298: (c)-(d): a slower decay of the ``ridge" into 8 solitons $\psi_0$;

299: (e)-(f): the annihilation of the internal ring and the central

300: soliton, and the repulsion of the persisting 8 solitons. Here

301: $\gamma=4.5$ and $h=4.53$; shown is $\mbox{Re} \, \psi$. Note the

302: change of the vertical scale in (e)-(f) w.r.t. that in (a)-(d).}

303: \label{fig5}

304: \end{figure}

305: 

306: {\bf 5.} Besides the nodeless solution ${\cal R}_0(r)$, the

307: ``master" equation (\ref{master}) has solutions ${\cal R}_n(r)$

308: with $n$ nodes, $n \ge 1$. (See Fig.\ref{fig0}). These give rise

309: to a sequence of nodal solutions $\psi_n$ of the damped-driven NLS

310: (\ref{2Dnls}), defined by eq.(\ref{soliton}) with ${\cal R}_0 \to

311: {\cal R}_n$. To examine the stability of the

312: $\psi_{n}$, %  solution,

313: we  solved the eigenvalue problem (\ref{EV}) numerically,

314: with % the

315:  operators $L_{0,1}^{(m)}$ as in

316: (\ref{m2r2}). The {\it radial\/} stability properties of the nodal

317: solitons turned out to be similar to those of the nodeless soliton

318: $\psi_0$. Namely, the $\psi_n$ solutions are stable against

319: radially-symmetric perturbations  for sufficiently large $\gamma$.

320: The corresponding stability regions for $\psi_1$ and $\psi_2$ are

321: depicted in Fig.(\ref{chart}). However, the {\it azimuthal\/}

322: stability properties of the nodal solitons have turned out to be

323: quite different.

324: 

325: Both $\psi_1$ and $\psi_2$ solutions do have  eigenvalues

326: $\lambda$ with nonzero real parts  for

327:  orbital numbers $m \ge 1$.

328: (See Fig.\ref{fig2} and \ref{fig3}.) Having found  eigenvalues

329: $\lambda$  for each $\epsilon$, one still has to identify those

330: giving rise to the largest growth rates

331: \begin{equation}

332: \nu= \mbox{Re} \, \mu -\Gamma \label{nu}

333: \end{equation}

334: for each pair $(\epsilon, \gamma)$ [or, equivalently, for each

335: $(h,\gamma)$]. In (\ref{nu}), $\mu$ is reconstructed using

336: eq.(\ref{mugamma}). The selection of {\it real\/} eigenvalues is

337: straightforward; in this case we

338: have the following two simple rules:\\

339: $\bullet$ If, for some $\epsilon$, there are eigenvalues

340: $\lambda_1>0$, $\lambda_2>0$ such that $\lambda_1>\lambda_2$, then

341: $\nu_1>\nu_2$ for this $\epsilon$ and all $\gamma$. That is, of

342: all real eigenvalues $\lambda$ one  has to

343: consider only the  largest one.\\

344: $\bullet$ If, for some $\epsilon$, there is a real eigenvalue

345: $\lambda_1>0$ and a complex eigenvalue $\lambda_2$, with

346: $\mbox{Re}\, \lambda_2>0$ and $\lambda_1 > \mbox{Re} \,

347: \lambda_2$, then $\nu_1>\nu_2$ for this $\epsilon$ and all

348: $\gamma$. That is, one can ignore all complex eigenvalues with

349: real parts smaller than a real eigenvalue --- if there is one.

350: 

351: The comparison of two complex eigenvalues is not so

352: straightforward. In particular, the fact that $\mbox{Re} \,

353: \lambda_1

354: > \mbox{Re} \, \lambda_2$ does not necessarily imply that

355: $\nu_1>\nu_2$. Which  of the two growth rates, $\nu_1$ or $\nu_2$,

356: is larger will depend on the imaginary parts of $\lambda$, as well

357: as on $\gamma$.

358: 

359: In figures \ref{fig2} and \ref{fig3}, we illustrate the real and

360: imaginary parts of the eigenvalues, arising for different $m$, for

361: the solitons $\psi_1$ and $\psi_2$. The soliton $\psi_1$ has

362: discrete eigenvalues $\lambda$ associated with orbital numbers

363: $m=0,1,...,5$ and the soliton $\psi_2$ with

364:  $m=0,1,...,10$.

365: 

366: \begin{table}

367: \begin{center}

368: %\begin{tabular}{cc}

369: %$\qquad\qquad\qquad n=1\qquad\qquad\qquad $ & $\qquad\qquad\qquad

370: %n=2\qquad\qquad\qquad $ \\

371: %\end{tabular}

372: \begin{tabular}{|c|c|c|c|c|c|c|c|c|} \cline{1-4} \cline{6-9}

373: m & $\nu$ & Re$\lambda$ & Im$\lambda$&$\qquad$&  m & $\nu$ & Re$\lambda$ & Im$\lambda$ \\

374: \cline{1-4} \cline{6-9}

375:          0 & -0.1620 &   1.5797   & 4.2181  &&   0 &   -0.3361  &  2.3531 &   6.1585 \\

376:          0 & -1.4827 &   0.0609   & 1.8743 && 0 &   -0.5877  &  0.1819 &

377:          1.7847 \\

378:                   & & & &&   0 &  -0.4168   &  0.3093 &   1.5572 \\

379: \cline{1-4} \cline{6-9}

380:     1 &  -0.8255    & 0.2272    & 1.6198  && 1 &   -0.8089 &   0.3818  &  2.1021 \\

381:     1 &    2.79e-6   & 0.0033    & 0.0000 &&   1 &   -0.4891 &   0.1111  &  1.6352 \\

382:     & & & & &  1 &   1.07e-5 & 0.0079 &   0.0000 \\

383: \cline{1-4} \cline{6-9}

384:     2 &   -0.3012   & 0.2213    & 1.0602  &&   2 &   -0.3497 &   0.3737  &  1.4597 \\

385: & & & & & 2 &   -0.0328 &   0.2602  &  0.5128 \\

386: \cline{1-4} \cline{6-9}

387:     3 &    0.0872   & 0.5821    & 0.0000 && 3 &    0.5406 &   1.8686  &  0.0000 \\

388: \cline{1-4} \cline{6-9}

389:     4 &    0.3689   & 1.2399    & 0.0000 &&   4 &    0.5286 &   1.8462  &  0.0000 \\

390: \cline{1-4} \cline{6-9}

391:     5 &    0.2057   & 0.9076    & 0.0000 &&    5 &    0.0263 &   0.3958  &  0.0000 \\

392: \cline{1-4} \cline{6-9}

393: &&&&&    6 &    0.1611 &   0.9898  &  0.0000 \\

394:  \cline{6-9}

395:  &&&&&    7 &    0.2490 &   1.2392  &  0.0000  \\

396:  \cline{6-9}

397:  &&&&&    8 &    0.2783 &   1.3133  &  0.0000 \\

398:  \cline{6-9}

399:  &&&&&    9 &    0.2288 &   1.1861  &  0.0000  \\

400:  \cline{6-9}

401:  &&&&&   10 &    0.0720 &   0.6567  &  0.0000 \\

402: \cline{1-4}  \cline{6-9}

403:    \end{tabular}

404:    \end{center}

405:    \caption{\sf Eigenvalues $\lambda$

406:    and corresponding growth rates

407: $\nu$ for the

408:  solitons $\psi_1$ (left panel) and $\psi_2$ (right panel).

409:    We included only the eigenvalues which can,

410:  potentially, give rise to the largest growth rate in each

411:  ``symmetry class" $m$. Some other eigenvalues have been

412:  filtered out using the above selection rules.}

413: \end{table}

414: 

415: In order to compare the conclusions based on the linearised

416: analysis with direct numerical simulations of the unstable

417: solitons $\psi_1$ and $\psi_2$, we fix some $h$ and $\gamma$ and

418: identify the eigenvalue with the maximum growth rate in each case.

419: In the case of the soliton $\psi_1$, we choose $\gamma=3.5$ and

420: $h=3.6$; the corresponding $\epsilon=0.9146$. The real and

421: imaginary parts of $\lambda$ for each $m$ as well as the resulting

422: growth rates $\nu$ are given in Table 1 (left panel). The

423: eigenvalue with the largest ${\rm Re} \lambda$ is associated with

424: $m=0$; however, for the given $\epsilon$ and $\gamma$ the

425: resulting $\nu <0$. (This is because we have chosen a

426:  point in the ``radially stable'' part of the $(\gamma, h)$-plane,

427: to the right of the ``$n=1$'' curve in Fig.{\ref{chart}}.) On the

428: contrary, the growth rates corresponding to the real eigenvalues

429: associated with

430:  $m=3,4,5$ are

431: positive for all $\gamma$. The maximum growth rate is associated

432: with $m=4$. The corresponding eigenfunctions $u(r)$ and $v(r)$

433: have a single maximum near the position of the minimum of the

434: function ${\cal R}_1(r)$; that is, the perturbation is

435: concentrated near the circular ``valley" in the relief of

436: $\psi(x,y)|^2$.

437:  This

438: observation suggests that for $\gamma =3.4$ and $h=3.5$, the

439: soliton $\psi_1$ should break into a symmetric pattern of 5

440: solitons $\psi_0$: one at the origin and four around it.

441: 

442: Next, in the case of the soliton $\psi_2$ we fix $\gamma=4.5$ and

443: $h=4.53$; this gives $\epsilon=0.6846$. The corresponding

444: eigenvalues, for each $m$, are presented in Table 1 (right panel).

445: Again, the eigenvalue with the largest $\mbox{Re} \lambda$ is the

446: one for $m=0$ but  the resulting $\nu$ is negative. The largest

447: growth rates ($\nu_3=0.54$ and $\nu_4=0.53$, respectively) are

448: those pertaining to $m=3$ and $m=4$. The corresponding

449: eigenfunctions have their maxima near the position of the minimum

450: of the function ${\cal R}_2(r)$. Therefore, the circular ``valley"

451: of the soliton $\psi_2$ is expected to break into three or four

452: nodeless solitons $\psi_0$. (Since $\nu_3$ is so close to $\nu_4$,

453: the actual number of resulting solitons

454: --- three or four --- will be very sensitive

455: to the choice of the initial perturbation.) Next, eigenfunctions

456: pertaining to $m=5,6,...10$ have their maxima near the second,

457: lateral, maximum of the function ${\cal R}_2(r)$. The largest

458: growth rate in this group of eigenvalues arises for $m=8$. Hence,

459: the circular ``ridge" of the soliton $\psi_2$ should break into 8

460: nodeless solitons, with this process taking longer than the

461: bunching of the ``valley" into the ``internal ring" of solitons.

462: 

463: The direct numerical simulations corroborate the above scenarios.

464: The $\psi_1$ soliton with $\gamma=3.5$ and $h=3.6$ splits into a

465: constellation of 5 nodeless solitons: one at the origin and  four

466: solitons of opposite polarity at the vertices of the square

467: centered at the origin. The emerging nodeless solitons are stable

468: but  repelling each other, see Fig.\ref{fig4}. Hence, no

469: stationary nonsymmetric configurations are possible; the

470: peripheral solitons escape to infinity. The $\psi_2$ soliton with

471: $\gamma=4.5$ and $h=4.53$ has a more complicated evolution. As

472: predicted by the linear stability analysis, it dissociates into 13

473: nodeless solitons: one at the origin, four solitons of opposite

474: polarity forming a square around it and  eight more solitons (of

475: the same polarity as the one at the origin) forming an outer ring.

476: (The fact that the inner ring consists of four and not three

477: solitons, is due to the square symmetry of our domain of

478: integration which favours perturbations with $m=4$ over those with

479: $m=3$.) In the subsequent evolution the central soliton and the

480: nearest four annihilate

481:  and only the eight outer solitons persist. They

482: repel each other and eventually escape to infinity, see

483: Fig.\ref{fig5}.

484: 

485: In conclusion, our analysis suggests the interpretation of  the

486: nodal solutions as degenerate, unstable coaxial complexes of the

487: nodeless solitons $\psi_0$.

488: 

489: 

490: {\bf Acknowledgements.} This project was supported by the NRF of

491: South Africa under grant  2053723. The work of E.Z.    was

492: supported by   an RFBR grant 03-01-00657.

493: 

494: 

495: \begin{thebibliography}{99}

496: \bibitem{Swinney} P.B. Umbanhowar, F. Melo, and H.L. Swinney,

497: Nature {\bf 382}, 793 (1996)

498: 

499: \bibitem{Faraday} O. Lioubashevski {\it et al\/},

500: %Y. Hamiel, A. Agnon, Z. Reches, and J. Fineberg,

501: Phys. Rev. Lett. {\bf 83}, 3190 (1999);

502:  H. Arbell and J. Fineberg,

503: {\it ibid.\/} {\bf 85}, 756 (2000)

504: 

505: \bibitem{Astruc} D. Astruc and S. Fauve, in: Fluid Mechanics and Its

506: Applications, vol.62, p. 39-46

507: %in: IUTAM Symposium on

508:  %Free Surface Flows (A.C.King, Y.D.Shikhmurzaev, eds.,

509:  (Kluwer, 2001)

510: 

511: 

512: \bibitem{numerical} L.S. Tsimring and I.S. Aranson, Phys. Rev. Lett.

513: {\bf 79}, 213 (1997);

514:  E. Cerda, F. Melo, and S. Rica, {\it ibid.\/} {\bf 79}, 4570 (1997);

515:            % discrete-time map with spatial coupling

516:       S.C. Venkataramani and E. Ott, {\it ibid.\/}

517: {\bf 80}, 3495 (1998); % oscillons in discrete time model

518: D. Rothman, Phys. Rev. E {\bf 57}, 1239 (1998);

519:    % \semicontinuum model for granular Faraday resonance

520:     J. Eggers and H. Riecke, {\it ibid.\/} {\bf 59}, 4476 (1999);

521:     % hydrodynamic-type equations for granular materials

522: C. Crawford, H. Riecke, Physica D {\bf 129}, 83 (1999); % oscillons in

523:                                                         %Swift-Hohenberg

524: % including provisionally:

525: H. Sakaguchi, H.R. Brand, Europhys. Lett. {\bf 38}, 341 (1997);

526: Physica D {\bf 117}, 95 (1998)

527: 

528: 

529: \bibitem{us} I.V. Barashenkov, N.V. Alexeeva, and E.V. Zemlyanaya, Phys.

530: Rev. Lett. {\bf 89}, 104101 (2002)

531: 

532: \bibitem{Rypdal} See e.g. K. Rypdal, J.J. Rasmussen and K. Thomsen,

533: Physica {\bf D 16}, 339 (1985) and references therein.

534: \bibitem{Malkin} E.A. Kuznetsov and S.K. Turitsyn,

535: Phys. Lett. {\bf 112A}, 273 (1985); V.M. Malkin and E.G. Shapiro,

536: Physica D {\bf 53}, 25 (1991)

537: 

538: \bibitem{collapse} For  a recent review  and references on the blowup

539: in  2D and 3D NLS equations, see e.g. L. Berge, Phys. Rep. {\bf

540: 303}, 259 (1998);

541:  G. Fibich and G. Papanicolaou,

542: SIAM J. Appl. Math. {\bf 60}, 183 (1999)

543: 

544: 

545: \bibitem{BBK} I.V. Barashenkov, M.M. Bogdan and V.I. Korobov,

546: Europhys. Lett. {\bf 15}, 113 (1991)

547: 

548: 

549: \end{thebibliography}

550: 

551: 

552: \end{document}

553: