1: \documentclass[pre,aps,twocolumn]{revtex4}

2: \usepackage{graphicx}

3: 

4: \newcommand{\beq}{\begin{equation}}

5: \newcommand{\eeq}{\end{equation}}

6: \newcommand{\ba}{\begin{array}}

7: \newcommand{\ea}{\end{array}}

8: \newcommand{\bean}{\begin{eqnarray*}}

9: \newcommand{\eean}{\end{eqnarray}}

10: \newcommand{\bea}{\begin{eqnarray}}

11: \newcommand{\eea}{\end{eqnarray}}

12: \newcommand{\bc}{\begin{center}}

13: \newcommand{\ec}{\end{center}}

14: \newcommand{\bt}{\begin{table}}

15: \newcommand{\et}{\end{table}}

16: \newcommand{\eps}{\epsilon}

17: \newcommand{\bpsi}{{\bf \psi}}

18: \newcommand{\bphi}{{\bf \phi}}

19: \newcommand{\la}[1]{\label{#1}}

20: \newcommand{\p}{\partial}

21: \renewcommand{\d}{{\rm d}}

22: \newcommand{\de}{\delta}

23: \newcommand{\vp}{\varphi}

24: \newcommand{\ds}{\displaystyle}

25: \newcommand{\no}{\noindent}

26: \newcommand{\pp}[2]{{\partial #1 \over \partial #2}}

27: \newcommand{\ppn}[3]{{\partial^{#1} #2 \over \partial #3^{#1}}}

28: \newcommand{\Pain}{Painlev\'{e} }

29: \newcommand{\mbf}[1]{\mbox{\boldmath {$#1$}}}

30: \newcommand{\binomial}[2]{\left(\ba{c}#1 \\ #2 \ea\right)}

31: \newcommand{\rf}[1]{(\ref{#1})}

32: \newcommand{\beqno}{\begin{displaymath}}

33: \newcommand{\eeqno}{\end{displaymath}}

34: \newcommand{\dd}[2]{{\delta #1 \over \delta #2}}

35: \newcommand{\ddn}[3]{{\delta^{#1} #2 \over \delta #3^{#1}}}

36: \newcommand{\ie}{{\em i.e.}}

37: \newcommand{\been}{\begin{enumerate}}

38: \newcommand{\een}{\end{enumerate}}

39: \renewcommand{\aa}[2]{\alpha_{#1}^{(#2)}(r)}

40: \newcommand{\al}[1]{\alpha_{#1}(r)}

41: \renewcommand{\iff}{\Leftrightarrow}

42: \newcommand{\since}[1]{\stackrel{\mbox {using \rf{#1}}}{=}}

43: \newcommand{\used}[2]{#1 \dotfill #2\\}

44: \newcommand{\call}{{\cal L}}

45: \newcommand{\sn}{{\rm sn}}

46: \newcommand{\cn}{{\rm cn}}

47: \newcommand{\dn}{{\rm dn}}

48: 

49: \begin{document}

50: 

51: \title{X,Y,Z-Waves: Extended Structures in Nonlinear Lattices}

52: \author{P. G.\ Kevrekidis$^{1}$, J. Gagnon$^{1}$, D. J.\ Frantzeskakis$^{2}$

53: and B. A.\ Malomed$^{3}$}

54: \affiliation{$^{1}$ Department of Mathematics and Statistics, University of

55: Massachusetts, Amherst MA 01003-4515\\

56: $^2$ Department of Physics, University of Athens, Panepistimiopolis,

57: Zografos, Athens 15784, Greece \\

58: $^3$ Department of Interdisciplinary Studies, Faculty of Engineering, Tel

59: Aviv University, Tel Aviv 69978, Israel}

60: 

61: \begin{abstract}

62: Motivated by recent experimental and theoretical results on

63: optical X-waves, we propose a new type of waveforms in 

64: %two- and three-dimensional 

65: 2D and 3D discrete media -- multi-legged extended

66: nonlinear structures (ENS), built as arrays of lattice solitons

67: ({\it tiles} or {\it stones}, in the 2D and 3D cases,

68: respectively). First, we study the stability of the tiles and

69: stones analytically, and then extend them numerically to complete

70: ENS forms 

71: %. Results are reported 

72: for both 2D and 3D lattices.

73: %including the first analytical stability calculations in the latter case. 

74: The predicted 

75: %extended 

76: patterns are relevant to a variety of physical settings, such as Bose-Einstein condensates in

77: deep optical lattices, lattices built of microresonators, 

78: photorefractive crystals with optically induced lattices 

79: (in the 2D case) 

80: and others.

81: \end{abstract}

82: 

83: \maketitle

84: 

85: \textit{Introduction and setup}. In the past few years, studies of dynamics

86: in discrete media, as well as in their continuum periodically modulated

87: counterparts, 

88: have enjoyed very rapid progress \cite{general_review}. 

89: This is fueled by a need to

90: tackle diverse physical contexts such as nonlinear optics of waveguide

91: arrays and photorefractive crystals \cite{review_opt}, the physics of

92: Bose-Einstein condensates (BECs) in optical lattices \cite{review_BEC}, 

93: %and

94: the denaturation of the DNA double-strand \cite{review_DNA}, and so on.

95: %among many others \cite{general_review}. 

96: These efforts have resulted in the creation of

97: a variety of novel nonlinear structures, such as discrete dipole \cite{dip},

98: quadrupole \cite{quad}, necklace \cite{neck} and other

99: multi-pulse/multi-pole localized patterns \cite{multi}, discrete vortices 

100: \cite{vortex}, 

101: %and 

102: ring solitons \cite{rings}, and others \cite{us_yuri}.

103: Such structures have a potential to be used as carriers and conduits for

104: data transmission and processing, in the context of all-optical 

105: and quantum-information schemes.

106: 

107: The present work is motivated by the above-mentioned achievements, 

108: %in studies of the lattice solitons, 

109: as well as by recent 

110: developments in studies of \textit{X-waves}, which are

111: quasi-linear extended structures generated in three-dimensions (3D) \cite{ditrapani},

112: two-dimensions (2D) \cite{boris}, and even quasi-discrete \cite{hiza} media. These

113: waves arise when the linear second-order differential operator in

114: the 

115: %corresponding 

116: relevant wave equation is genuinely, or effectively \cite{ditrapani2},  

117: sign-indefinite (roughly speaking, a D'Alembertian instead of the Laplacian). 

118: A similar shape is featured by cross configurations formed by two intersecting domain

119: walls in a system of two 

120: %nonlinearity 

121: coupled 2D Gross-Pitaevskii equations \cite{cross}, with a difference that these structure are

122: essentially nonlinear ones. 

123: 

124: In the present work, we are dealing with the sign-definite

125: discrete Laplacian (nevertheless, one can

126: easily induce the sign-indefiniteness in the discrete wave

127: equation, applying the well-known \textit{staggering

128: transformation} \cite{general_review} along one coordinate only).

129: Instead of emulating the mathematical structure 

130: of the continuous wave equations supporting X-waves in the lattice medium, 

131: our aim is to construct stable extended nonlinear structures (ENSs) in \emph{discrete} media.

132: These wave structures are partly delocalized, by their nature,

133: ranging from a few sites to infinitely extended ones. An essential

134: difference between the quasi-linear X-waves in continua and the lattice

135: entities presented here 

136: is that the latter are \emph{strongly nonlinear} structures, being built

137: as arrays of individual lattice solitons (``tiles"); in that

138: sense, they somewhat resemble weakly localized 

139: \textit{hypersolitons} \cite{hm} and 

140: \textit{supervortices} \cite{HSakaguchi}, constructed as finite ensembles

141: of individual solitons. 

142: %in continuum models combining cubic or saturable nonlinearity and a square lattice). 

143: In fact, we initialize our analysis (see below) from a purely nonlinear

144: (\textit{anti-continuum}, AC) limit, without linear intersite

145: coupling on the lattice. The eventual shape of the extended

146: lattice waves features a great variety, including X-waves,

147: Y-waves, Z-waves and other forms, provided that the underlying

148: core (``tile") is dynamically stable. Each species is stable in a

149: finite interval of values of the intersite coupling constant. 

150: %(for general reasons similar to those analyzed in Ref. \cite{dep}).

151: Some of those patterns may definitely be of interest (for

152: instance, as star-shaped waveguiding configurations) to

153: applications in both nonlinear-optical and matter-wave contexts,

154: where they can potentially be created. 

155: %be (and, to some extent, have been \cite{multi}) created.

156: 

157: The analysis is performed in terms of the 

158: %paradigmatic 

159: discrete nonlinear Schr{\"{o}}dinger (DNLS) equation in 2D or 3D,

160: %two or three dimensions,

161: %

162: \begin{equation}

163: i\dot{u}_{\mathbf{n}}=-\epsilon \left( \Delta _{2}u\right)

164: _{\mathbf{n}}-|u_{\mathbf{n}}|^{2}u_{\mathbf{n}},  

165: \label{DNLS}

166: \end{equation}

167: %

168: where $u_{\mathbf{n}}$ is a complex amplitude of the electromagnetic wave in

169: an optical waveguide array (in the 2D case)

170: \cite{review_opt,DNLS_opt}, or the BEC\ wave function at nodes of

171: a deep (2D or 3D) optical lattice \cite{DNLS_BEC}, $\mathbf{n}\

172: $being the vectorial lattice index, and $\Delta _{2} $ the

173: standard discrete Laplacian. Further, $\epsilon $ is the constant

174: of the intersite coupling, and the overdot stands for the

175: derivative with respect to the evolution variable; the latter is $z$ in optical

176: arrays, or $t$ in the BEC model (or in 

177: %. Another physical realization of both 3D and 2D DNLS equations with the evolution in $t$ is provided by 

178: a crystal built of microresonators \cite{photons}).

179: 

180: The presentation of our results on ENS solutions of Eq.

181: (\ref{DNLS}) proceeds as follows. First, 

182: in the AC limit ($\epsilon =0$), 

183: %in Eq. (\ref{DNLS})], 

184: we specify main types of the 

185: %above-mentioned 

186: ``tiles'', of which the 

187: %extended structures 

188: ENSs are to be composed. Then, we numerically extend these compositions into full

189: ENS solutions for $\epsilon >0$. Finally, we summarize our

190: findings and discuss further problems.

191: 

192: \textit{Analytical results: tiles, stones, and their stability}. We

193: look for standing-wave solutions to Eq. (\ref{DNLS}), in the form

194: of $u_{\mathbf{n}}=\exp (i\Lambda t)\phi _{\mathbf{n}}$, with

195: $\phi _{\mathbf{n}}$ satisfying the 

196: %stationary 

197: equation,

198: %

199: \begin{equation}

200: f(\phi _{\mathbf{n}},\epsilon )=\Lambda \phi _{\mathbf{n}}-\epsilon \Delta

201: _{2}\phi _{\mathbf{n}}-|\phi _{\mathbf{n}}|^{2}\phi _{\mathbf{n}}=0.

202: \label{steady}

203: \end{equation}

204: %

205: %Linearization of 

206: Perturbation of Eq. (\ref{DNLS}) around the solutions of Eq. (\ref{steady}) 

207: %shows that a set of a small perturbation

208: %of the solution and its complex conjugate is a zero mode of the following

209: leads to the linearization operator,

210: %

211: \begin{eqnarray}

212: \mathcal{H}_{\mathbf{n}}^{(\epsilon )} = \left(

213: \begin{array}{cc}

214: \Lambda -2|\phi _{\mathbf{n}}|^{2} & -\phi _{\mathbf{n}}^{2} \\

215: -\bar{\phi}_{\mathbf{n}}^{2} & \Lambda -2|\phi _{\mathbf{n}}|^{2}\end{array}\right)

216: %\nonumber \\

217: -\epsilon \Delta _{2}\left(

218: \begin{array}{cc}

219: 1 & 0 \\

220: 0 & 1\end{array}\right) ,  

221: \label{oper}

222: \end{eqnarray}

223: %

224: the overbar standing for complex conjugate. By means of a rescaling, we

225: fix $\Lambda \equiv 1$, keeping $\epsilon $ as a free parameter.

226: %where $s_{n'} \phi_n=\phi_{n+n'}$ is the shift operator (written

227: %in 3d in Eq. (\ref{energy})).

228: 

229: In the AC limit of $\epsilon =0$, solutions to Eq. (\ref{steady})

230: are 

231: %obvious, 

232: $u_{\mathbf{n}}=re^{i\theta _{\mathbf{n}}}$, where

233: real amplitude $r $ is $0$ or $\sqrt{\Lambda }$, and $\theta _{\mathbf{n}}$ 

234: are arbitrary constants.\ To continue such a

235: solution to $\epsilon \neq 0$, the Lyapunov-Schmidt condition

236: needs to be satisfied \cite{dep}, \textit{viz}., the projection of

237: eigenvectors of the kernel of $\mathcal{H}_{\mathbf{n}}^{(0)}$

238: (i.e., zero modes of the operator) onto the system of stationary

239: equations should be null. This condition gives rise to a

240: solvability condition at\ each ``AC-filled" site, i.e., one with

241: $r\neq 0 $ in the AC limit,

242: %

243: \begin{equation}

244: -2ig_{\mathbf{n}}(\theta ,\epsilon )=\epsilon e^{-i\theta

245: _{\mathbf{n}}}\Delta _{2}\phi _{\mathbf{n}}-\epsilon e^{i\theta

246: _{\mathbf{n}}}\left( \Delta _{2}\bar{\phi}\right) _{\mathbf{n}}

247: \label{inter}

248: \end{equation}

249: %

250: (the factor of $-2i$ is introduced for convenience). 

251: %Which is quite important for our present purposes, 

252: Importantly, the eigenvalues $\gamma $ of

253: the Jacobian, $\mathcal{M}_{ij}=\partial g_{i}/\partial \theta

254: _{j}$, are intimately related to the leading-order approximation

255: for eigenvalues ($\lambda $) of the linearization around a

256: stationary solution, $\lambda =\pm \sqrt{2\gamma }$. We use this

257: relation, alongside a perturbative expansion in the solution,

258: $\phi _{\mathbf{n}}=\phi _{\mathbf{n}}^{(0)}+\epsilon \phi

259: _{\mathbf{n}}^{(1)}+\dots $, to derive leading-order bifurcation

260: conditions for any given configuration, and subsequently to find

261: the corresponding linear-stability eigenvalues.

262: 

263: 2D configurations can be categorized by the number of their ``legs\textquotedblright\ 

264: %(i.e., quasi-1D lines of AC-filled sites it is composed of).

265: (i.e., the number of quasi-1D lines of excited sites it contains).

266: Simplest is the one-leg structure assembled of

267: two- or three-site tiles, 

268: %This case is particularly simple as it allows to borrow 

269: for which 1D stability results are applicable 

270: %from Ref. 

271: \cite{dep}; here, the resulting Jacobian is

272: \[

273: (\mathcal{M}_{1})_{i,j}=\left\{

274: \begin{array}{lcl}

275: \cos (\theta _{j+1}-\theta _{j})+\cos (\theta _{j-1}-\theta _{j}), 

276: \quad 

277: i=j, \\

278: -\cos (\theta _{j}-\theta _{i}), 

279: \quad  

280: i=j\pm 1, 

281: \\

282: 0, 

283:  \quad  

284: |i-j|\geq 2,\end{array}\right.

285: \]

286: from which 

287: %important 

288: conclusions for the stability of ENSs may be drawn. In

289: particular, a fundamental stability condition extensively used in

290: our analysis is that the phase shift between fields at adjacent

291: sites must be $\pi $ \cite{todd,dep}. For the two-site tile

292: satisfying this condition, the stability eigenvalues are

293: \cite{dep} $\lambda =\sqrt{2\gamma }=\pm 2\sqrt{\epsilon }i$,

294: while for its three-site counterpart, $(+1,-1,+1)$, they are

295: $\lambda =\pm \sqrt{2\epsilon }i$ and $\lambda =\pm

296: \sqrt{6\epsilon }i$. The full one-leg configuration assembled of

297: these tiles is a chain, $(\dots ,-1,+1,-1,+1,-1,\dots )$,

298: hereafter termed configuration 1.

299: 

300: \begin{figure}[tbp]

301: \begin{center}

302: \hskip-0.2cm\includegraphics[height=6cm,width=6cm]{full1.eps} ~~

303: %%\includegraphics[width=3.5cm]{skyrfig3d_v.jpg.ps}

304: %%\includegraphics[width=3.5cm]{skyrfig3d_v_lighted.jpg.ps}

305: \end{center}

306: \par

307: \vskip-0.5cm \caption{(Color online) A family of single-leg

308: patterns in the 

309: %two-dimensional 

310: 2D DNLS equation. The top left and

311: right panels show, respectively, the norm and the largest

312: instability growth rate vs. $\protect\epsilon $. The contour plots

313: in the middle panels show the stationary solutions  for

314: $\protect\epsilon =0.05$ (stable) and $\protect\epsilon =0.1$

315: (unstable).  The spectral planes $(\protect\lambda

316: _{r},\protect\lambda _{i}) $ of numerically computed eigenvalues

317: for these solutions are displayed in the bottom panels.}

318: \label{Fig1}

319: \end{figure}

320: 

321: 

322: \begin{figure}[tbp]

323: \begin{center}

324: \hskip-0.15cm

325: \begin{tabular}{cc}

326: \includegraphics[height=6cm,width=2.75cm]{full2a.eps}

327: \includegraphics[height=6cm,width=2.75cm]{full2b.eps}

328: \includegraphics[height=6cm,width=2.75cm]{full2c.eps} &  \\

329: \includegraphics[height=4cm,width=2.75cm]{tile2a.eps}

330: \includegraphics[height=4cm,width=2.75cm]{tile2b.eps}

331: \includegraphics[height=4cm,width=2.75cm]{tile2c.eps} &

332: \end{tabular}\end{center}

333: \par

334: \vskip-0.7cm 

335: \caption{Same as in Fig. 1 but for configurations 2a, 2b and 2c. 

336: %Also included in 

337: Bottom panels present the stability

338: eigenvalues for individual tiles of which these patterns are

339: built, numerically found ones (solid 

340: %blue 

341: lines) versus the

342: analytical prediction (dashed 

343: %red 

344: lines). Examples of species 2a

345: and 2b are displayed in two left and two middle panels for

346: $\protect\epsilon =0.15$ and $0.25$, and an example of 2c in the

347: right panels pertain to $\protect\epsilon =0.05$ and

348: $\protect\epsilon =0.1$.} 

349: \label{Fig2}

350: \end{figure}

351: 

352: %

353: \begin{figure}[tbp]

354: \begin{center}

355: \hskip-0.15cm

356: \begin{tabular}{cc}

357: \includegraphics[height=6cm,width=4cm]{full3a.eps}

358: \includegraphics[height=6cm,width=4cm]{full3b.eps} &  \\

359: \includegraphics[width=4cm]{tile3a.eps}

360: \includegraphics[width=4cm]{tile3b.eps} &

361: \end{tabular}\end{center}

362: \par

363: \vskip-0.7cm

364: \caption{Same as Fig. 2, but for configurations 3a and 3b. Examples of both

365: species are shown for $\protect\epsilon =0.05$ and $0.1$.}

366: \label{Fig3}

367: \end{figure}

368: 

369: \begin{figure}[tbp]

370: \begin{center}

371: \hskip-0.15cm

372: \begin{tabular}{cc}

373: \includegraphics[height=12cm,width=4cm]{full4a.eps}

374: \includegraphics[height=12cm,width=4cm]{full4b.eps} &  \\

375: \includegraphics[width=4cm]{tile4a.eps}

376: \includegraphics[width=4cm]{tile4b.eps} &

377: \end{tabular}\end{center}

378: \par

379: %\vskip-0.7cm

380: \caption{Same as in above figures, but for 3D configurations 4a

381: and 4b. Their examples are plotted, in two left and two right

382: panels, respectively, for $\protect\epsilon =0.03$ and

383: $\protect\epsilon =0.1$. In configuration 4b, the imaginary parts

384: are shown by different colors, green for phase $\protect\pi /2$

385: and yellow for $3\protect\pi /2$.} \label{Fig4}

386: \end{figure}

387: 

388: 

389: Proceeding to two-tile configurations allows us to examine more

390: complex (in particular, X- and Y-like) 2D structures.

391: X-configurations can be built of square- or cross-shaped tiles,

392: which, in the AC\ limit, are

393: %

394: \begin{equation} \left(

395: \begin{array}{ccc}

396: 1 & 0 & -1 \\

397: 0 & 0 & 0 \\

398: -1 & 0 & 1\end{array}\right) ~\mathrm{and~}\left(

399: \begin{array}{ccc}

400: 1 & 0 & 1 \\

401: 0 & -1 & 0 \\

402: 1 & 0 & 1\end{array}\right)   \label{C2a}

403: \end{equation}

404: %

405: However, the latter one gives rise to an unstable

406: eigenvalue pair, $\lambda =\pm 2\epsilon $ (to leading order),

407: therefore it is not considered further. The former tile has a pair

408: of $\lambda =\pm 2\sqrt{2}\epsilon i$, and a double one, $\lambda

409: =\pm 2\epsilon i$, hence stable X-configurations (to be called 2a)

410: can be constructed at finite $\epsilon $ as arrays of such tiles

411: with alternating signs.

412: 

413: Other 

414: %relevant 

415: two-leg configurations are Y-shaped and ``skew-X"

416: ones, composed, respectively, of the 

417: %following 

418: tiles:

419: %

420: \begin{equation}

421: \left(

422: \begin{array}{ccc}

423: 1 & 0 & -1 \\

424: 0 & 0 & 0 \\

425: 0 & 0 & 1\end{array}\right) ~\mathrm{and~}\left(

426: \begin{array}{ccc}

427: 0 & -1 & 1 \\

428: 0 & 1 & 0 \\

429: 1 & -1 & 0\end{array}\right) .  \label{C2bc}

430: \end{equation}

431: %

432: Both these tiles are stable, with respective eigenvalues

433: $\lambda =\pm \sqrt{2}\epsilon i$ and $\lambda =\pm \sqrt{6}\epsilon i$

434: for the former, and $\lambda =\pm 0.874\sqrt{\epsilon }i$,

435: $\lambda =\pm 1.663\sqrt{\epsilon }i$,

436: $\lambda =\pm 2.882\sqrt{\epsilon }i$ and $\lambda =\pm

437: 2.690\sqrt{\epsilon }i$ for the latter. Stable sign-alternate ENSs

438: assembled of them are termed 2b and 2c, respectively.

439: 

440: Three-leg ENSs can be built of other (generally, denser filled) \emph{stable}

441: tiles, such as

442: \begin{equation}

443: \left(

444: \begin{array}{ccc}

445: 1 & -1 & 1 \\

446: -1 & 1 & -1 \\

447: 1 & -1 & 1\end{array}\right) ~\mathrm{and~}\left(

448: \begin{array}{ccc}

449: 1 & -1 & 1 \\

450: 0 & 1 & 0 \\

451: -1 & 0 & 0\end{array}\right) .  \label{C3}

452: \end{equation}

453: The former tile possesses three pairs of double eigenvalues,

454: $\pm \sqrt{2\epsilon }i$, $\pm \sqrt{6\epsilon }i$, and $\pm \sqrt{8\epsilon }i$, and

455: two pairs of single ones, $\pm 2\sqrt{\epsilon }i$ and $\pm \sqrt{12\epsilon

456: }i$. The latter has a double eigenvalue $\pm \sqrt{2\epsilon }i$ and a

457: single one $\pm \sqrt{8\epsilon }i$. Three-leg configurations constructed of

458: these tiles with alternating signs are called 3a and 3b, see below. In fact,

459: the latter one will be a Z-shaped array.

460: 

461: Finally, we consider two configurations as a proof-of-principle of

462: the extension of the ENS\ concept to the 3D space. In particular,

463: augmenting the first tile of (\ref{C3}) by two $-1$ sites, adjacent

464: to middle $1$ along the third direction, we fabricate a ``stone".

465: The stone is stable, with single eigenvalues 

466: $\lambda =\pm 1.248\sqrt{\epsilon }i$, 

467: $\lambda =\pm 2\sqrt{\epsilon }i$,

468: $\lambda =\pm \sqrt{6\epsilon }i$, 

469: $\lambda =\pm 2.763\sqrt{\epsilon }i$ and 

470: $\lambda =\pm 3.848\sqrt{\epsilon }i$, and double and triple ones, 

471: $\lambda =\pm \sqrt{8\epsilon }i$ and 

472: $\lambda =\pm \sqrt{2\epsilon }i$, respectively. These stones

473: will be used to build a stable 3D pattern (again, with sign

474: alternations), called 4a in our nomenclature.

475: 

476: %Perhaps 

477: The simplest stable stone 

478: has a ``diamond"-like shape, 

479: %is shaped as a ``diamond", 

480: i.e., it consists of six AC-filled sites surrounding, as nearest

481: neighbors, an empty one. 

482: %While 

483: Although some configurations of this type

484: are unstable \cite{us_yuri}, we have found a stable one, 

485: %that we have found to be stable

486: carrying a phase distribution that corresponds to a quadrupole in the

487: plane, with phases $\pi /2$ and $3\pi /2 $ lent to the two

488: out-of-plane sites. This stone has triple and single eigenvalues,

489: $\lambda =\pm 2\epsilon i$ and $\lambda =\pm 4\epsilon i$,

490: respectively, and a higher-order one, $\mathcal{O}(\epsilon ^{2})$. 

491: Augmenting stones of this type by alternating phase pulses, we

492: will build a stable 3D structure labeled 4b.

493: 

494: \textit{Numerical results}. In Figs. \ref{Fig1}-\ref{Fig4} we

495: present, in a unified format, results of the numerical

496: continuation of various ENS configurations from the AC limit.

497: Figure \ref{Fig1} shows configuration 1 (a straight chain of

498: sign-alternating tiles). The norm of the configuration,

499: $N=\sum_{n}\phi _{n}^{2}$, and its maximum instability growth rate

500: are shown, as a function of $\epsilon $, in the top left and right

501: panel, respectively. In the 21x21 lattice used in our 2D

502: computation, the configuration becomes unstable for $\epsilon

503: >0.074$. The middle panels show typical examples of stable and

504: unstable configurations, and the bottom panels show the spectral

505: plane $(\lambda _{r},\lambda _{i})$ of their numerically found

506: (in)stability eigenvalues, $\lambda \equiv \lambda _{r}+i\lambda_{i}$, 

507: the instability being ushered by $\lambda _{r}\neq 0$.

508: %

509: 

510: Figure \ref{Fig2} presents three different varieties of the two-leg

511: configurations: 2a (X-waves, left panels), 2b (Y-waves, middle panels) and

512: 2c (skewed X-waves, right panels). In addition, numerically computed

513: stability eigenvalues of the corresponding tiles, i.e., the first one in Eq.

514: (\ref{C2a}) and both in Eq. (\ref{C2bc}), are shown by solid blue lines for

515: comparison with the analytical predictions of the previous section (dashed

516: red lines), indicating good agreement between them. Configurations 2a, 2b,

517: and 2c become unstable at $\epsilon >0.202$, $\epsilon \geq

518: 0.206$ and $\epsilon \geq 0.068$ respectively.

519: %

520: 

521: Similarly, Fig. \ref{Fig3} shows two three-leg configurations, 3a and 3b

522: (the latter one may be naturally called a Z-wave), and their stability

523: characteristics. These configurations become unstable at $\epsilon >0.053$

524: and $\epsilon \geq 0.075$, respectively.

525: 

526: Finally, figure \ref{Fig4} shows the continuation to $\epsilon

527: \neq 0$ of 3D configurations 4a and 4b (defined above), which are

528: stable, respectively, at $\epsilon \leq 0.043$ and $\epsilon \leq 0.075$.

529: Note that the corresponding numerically computed eigenvalues 

530: (for the respective stones) again

531: agree well with the analytical predictions.

532: 

533: \textit{Conclusion}. We have presented a systematic approach towards

534: constructing a variety of extended states on 2D and 3D nonlinear lattices.

535: They are composed of building blocks, namely, tiles (2D) or stones (3D) with

536: alternating signs, that are originally defined in the anti-continuum limit

537: (the sign alternation is necessary for stability of the patterns). Stability

538: intervals for the structures were revealed by numerical continuation of the

539: corresponding solution families to finite values of the coupling constant.

540: Examination of nonlinear evolution of unstable patterns is another relevant

541: problem, to be considered elsewhere.

542: 

543: Another interesting topic is a discrete model such as Eq.

544: (\ref{DNLS}), but with a negative intersite coupling in one

545: spatial direction (say, $x$); as mentioned above, this model can

546: be generated from Eq. (\ref{DNLS}) by means of the staggering

547: transformation applied in this direction, i.e., 

548: $u_{n_{x},n_{y},n_{z}}\equiv (-1)^{n_{x}}v_{n_{x},n_{y},n_{z}}$.

549: In the 2D version of such a model, one can immediately find two

550: exact \emph{linear} single-leg solutions oriented along the

551: diagonals, \textit{viz}., $u_{n_{x},n_{y}}=\exp (-|n_{x}\mp

552: n_{y}|)$. Their superposition can be used to construct exact

553: linear X-waves. A potential extension of the latter states into the 

554: nonlinear model would also be worth considering.

555: 

556: \begin{thebibliography}{99}

557: 

558: \bibitem{general_review} S.\ Aubry, \newblock Physica \textbf{103D}, 201

559: (1997); S.\ Flach and C. R.\ Willis, \newblock Phys.\ Rep.\ \textbf{295},

560: 181 (1998); D. K.\ Campbell, S. Flach, and Y. S. Kivshar, Phys. Today,

561: January 2004, p. 43.

562: 

563: 

564: \bibitem{review_opt} D. N.\ Christodoulides \textit{et al.},

565: %F.\ Lederer, and Y.\ Silberberg,

566: Nature \textbf{424}, 817 (2003); A. A.\ Sukhorukov \textit{et al.},

567: %%Y. S.\ Kivshar, H. S.\ Eisenberg, and Y.\ Silberberg,

568: IEEE J. Quant. Elect. \textbf{39}, 31 (2003). J. W. Fleischer

569: \textit{et al.}, Opt. Express \textbf{13}, 1780 (2005).

570: 

571: %; U. Peschel

572: %\textit{et al.}, J.\ Opt.\ Soc.\ Am.\ B \textbf{19}, 2637 (2002).

573: 

574: \bibitem{review_BEC} V. A. Brazhnyi and V. V. Konotop, Mod. Phys. Lett. B

575: \textbf{18}, 627 (2004); P. G. Kevrekidis and D. J. Frantzeskakis, Mod.

576: Phys. Lett. B \textbf{18}, 173 (2004).

577: 

578: \bibitem{review_DNA} M. Peyrard, Nonlinearity \textbf{17}, R1 (2004).

579: 

580: 

581: \bibitem{dip} J. Yang \textit{et al.}, Opt. Lett. \textbf{29}, 1662 (2004).

582: 

583: \bibitem{quad} J. Yang \textit{et al.}, Stud. Appl. Math. \textbf{113}, 389

584: (2004).

585: 

586: \bibitem{neck} J. Yang \textit{et al.}, Phys. Rev. Lett. \textbf{94}, 113902

587: (2005).

588: 

589: \bibitem{multi} Z. Chen \textit{et al.}, Phys. Rev. Lett. \textbf{92},

590: 143902 (2004).

591: 

592: \bibitem{vortex} D. N. Neshev \textit{et al}., Phys. Rev. Lett. 92, 123903

593: (2004); J. W. Fleischer \textit{et al}., Phys. Rev. Lett. \textbf{92},

594: 123904 (2004).

595: 

596: \bibitem{rings} X. Wang, Z. Chen, and P. G. Kevrekidis, Phys. Rev. Lett.

597: \textbf{96}, 083904 (2006).

598: 

599: \bibitem{us_yuri} R. Carretero-Gonz{\'{a}}lez \textit{et al.}, Phys. Rev.

600: Lett. \textbf{94}, 203901 (2005); T. J. Alexander \textit{et al.}, Phys.

601: Rev. A \textbf{72}, 043603 (2005).

602: 

603: \bibitem{ditrapani} J. Y. Lu and J. F. Greenleaf, IEEE Trans. Ultrason.

604: Ferroelectr. Freq. Control \textbf{39}, 19 (1992); P. Di Trapani \textit{et

605: al.}, Phys. Rev. Lett. \textbf{91}, 093904 (2003). C. Conti \textit{et al.},

606: Phys. Rev. Lett. \textbf{90}, 093904 (2003).

607: 

608: \bibitem{boris} D. N. Christodoulides \textit{et al.}, Opt. Lett. \textbf{29}, 1446 (2004).

609: 

610: \bibitem{hiza} S. Droulias \textit{et al.}, Opt. Express \textbf{13}, 1827

611: (2005).

612: 

613: \bibitem{ditrapani2} C. Conti and S. Trillo, Phys. Rev. Lett. \textbf{92},

614: 120404 (2004).

615: 

616: \bibitem{cross} B. A. Malomed \textit{et al.}, 

617: %D. J. Frantzeskakis, and P. G. Kevrekidis,

618: Phys. Rev. A \textbf{70}, 043616 (2004).

619: 

620: \bibitem{hm} J. R. Salgueiro, H. Michinel, and M. I. Rodas-Verde\textit{.},

621: e-print nlin.PS/0603011.

622: 

623: \bibitem{HSakaguchi} H. Sakaguchi and B. A. Malomed, Europhys. Lett. \textbf{72}, 698 (2005).

624: 

625: 

626: \bibitem{DNLS_opt} P. G. Kevrekidis, K. {\O }. Rasmussen and A. R. Bishop,

627: Int. J. Mod. Phys. B \textbf{15}, 2833 (2001).

628: 

629: \bibitem{DNLS_BEC} A.\ Trombettoni and A.\ Smerzi, Phys.\ Rev.\ Lett.\

630: \textbf{86}, 2353 (2001); %F. Kh.\ Abdullaev \textit{et al.}, Phys.\ Rev.\

631: %\textbf{A64}, 043606 (2001)

632: G. L.\ Alfimov \textit{et al.}, Phys.\ Rev.\ E \textbf{66}, 046608 (2002)

633: 

634: \bibitem{photons} %'Slow' and 'fast' light in resonator-coupled waveguides

635: J. E.\ Heebner and R. W.\ Boyd, J.\ Mod.\ Opt.\ \textbf{49}, 2629 (2002);

636: %Depositing light in a photonic stop gap by use of Kerr nonlinear

637: %microresonators

638: P.\ Chak \textit{et al.}, %J.E.\ Sipe, and S.\ Pereira,

639: Opt.\ Lett.\ \textbf{28}, 1966 (2003). %%\bibitem{polar}

640: %%D. M. Whittaker and J. S.\ Roberts, Phys.\ Rev.\ B \textbf{62},

641: %%R16247 (2000); P. G.\ Savvidis and P. G.\ Lagoudakis, Sem.\ Sci.\

642: %%Tech.\ \textbf{18} , S311 (2003).

643: %\bibitem{atkinson}

644: 

645: \bibitem{dep} D. E. Pelinovsky \textit{et al.}, Physica D \textbf{212}, 1

646: (2005); \textit{ibid.} \textbf{212}, 20 (2005).

647: 

648: 

649: \bibitem{todd} T. Kapitula, P. G. Kevrekidis, and B.A. Malomed, Phys. Rev. E

650: \textbf{63}, 036604 (2001).

651: \end{thebibliography}

652: 

653: \end{document}

654: