1: %\documentstyle[aps,multicol,prl,epsf]{revtex}

2: %\input{psfig}

3: %\renewcommand{\baselinestretch}{2.0}

4: %\draft

5: 

6: 

7: \documentstyle[aps,multicol,prl,epsfig]{revtex}

8: 

9: \pagenumbering{arabic}

10: 

11: \begin{document}

12: \begin{title}

13: {\bf 

14: Feedback Loops Between Fields and Underlying Space Curvature: an Augmented

15: Lagrangian Approach}

16: \end{title}

17: 

18: \author{P.G. Kevrekidis$^1$, F.L. Williams$^1$, A.R. Bishop$^2$, 

19: I.G. Kevrekidis$^3$, and B.A. Malomed$^4$}

20: \address{$^1$ Department of Mathematics and Statistics, 

21: University of Massachusetts, Amherst, MA 01003-4515, USA \\ 

22: $^2$ 

23: Theoretical Division, MS B210,

24: Los Alamos National Laboratory, Los Alamos, NM 87545, USA \\

25: $^3$ Department of Chemical Engineering, Princeton University,

26: 6 Olden Str. Princeton, NJ 08544 \\

27: $^4$ Department of Interdisciplinary Studies, Faculty of Engineering, Tel

28: Aviv University, Tel Aviv 69978, Israel \\

29: }

30: %\title{}

31: \maketitle

32: 

33: \begin{abstract}

34: We demonstrate a systematic implementation of coupling between

35: a scalar field and the geometry of the space

36: (curve, surface, etc.) which carries the field. This naturally

37: gives rise to a feedback mechanism

38: between the field and the geometry. We develop a systematic model for the

39: feedback in a general form, inspired by a specific implementation in the

40: context of molecular dynamics (the so-called Rahman-Parrinello molecular

41: dynamics, or RP-MD). We use a generalized Lagrangian that allows for the

42: coupling of the space's metric tensor (the first fundamental form) to the

43: scalar field, and add terms motivated by RP-MD. We present two

44: implementations of the scheme: one in which the metric is only

45: time-dependent [which gives rise to ordinary differential equation (ODE) for

46: its temporal evolution], and one with spatio-temporal dependence [wherein

47: the metric's evolution is governed by a partial differential equation (PDE)].

48: Numerical results are reported for the (1+1)-dimensional model with a

49: nonlinearity of the sine-Gordon type.

50: \end{abstract}

51: 

52: %\date{\today}

53: 

54: \vspace{5mm}

55: 

56: %\section{Introduction}

57: 

58: \begin{multicols}{2}

59: 

60: Recently, much attention has been focused on soft-condensed-matter objects,

61: such as vesicles, microtubules, and membranes \cite{av1,av2,gaid1,gaid2}.

62: Many nanoscale physical systems, including nanotubes and electronic and

63: photonic waveguide structures \cite{gaid3,gaid4}, have nontrivial geometry

64: and are influenced by substrate effects. These classes of systems, many of

65: which are inherently nonlinear, raise the question of the interplay between

66: nonlinearity and a substrate with variable curvature. Of particular interest

67: is a possibility of developing curvature in the substrate due to forces

68: generated by the nonlinear field. The resulting curvature can in turn affect

69: the field.

70: 

71: There is an increasing body of literature dealing with the interplay of

72: nonlinearity and a curved substrate. Usually, however, the substrate

73: geometry is assumed to be {\it fixed}, see, e.g.,  \cite{curv}.

74: Nevertheless, for many applications, ranging from condensed matter to optics

75: to biophysics, it is relevant to introduce  models that admit

76: a flexible substrate, which is affected by the field(s) that it carries, as

77: well as feeding back into the field dynamics. In this situation, equations for

78: the fields in a nonlinear system abutting on the flexible substrate should

79: include both the field dynamics proper and the feedback coupling to the

80: substrate. Equations for the evolution of the substrate should in turn be

81: affected by the evolution of the field. A prototypical physical example of

82: this type is Euler buckling \cite{antm}, where the evolution of a thermal

83: profile causes the underlying surface to buckle (and hence locally modify

84: its curvature).

85: 

86: In a discrete setting, a model of this type has recently been presented in

87:  \cite{kmb}. However, it was limited to a system of masses coupled by

88: nonlinear springs. Some studies have also been performed in a special case

89: of the continuum limit of classical spin systems (such as the Heisenberg

90: chain) coupled to the curvature; geometric frustration was found to arise in

91: such settings \cite{dand}.

92: 

93: About twenty years ago, a problem similar to the theme of our study was

94: examined in the context of molecular dynamics studies of

95: structural transitions in crystals. 

96: In particular, in a series of papers \cite{RP}, Rahman and

97: Parrinello introduced a new idea for studying such transitions by means of

98: an augmented Lagrangian that would account for the degrees of freedom of the

99: ``box'' (the cell) in which the MD particles lie. In

100: studying the time evolution of the box dynamics (naturally obtained through

101: the Euler-Lagrange equations of the augmented Lagrangian for the box degrees

102: of freedom), they were able to identify structural transitions (under

103: external shear) from square to hexagonal patterns, fcc to bcc etc. We will

104: hereafter refer to this technology as the RP-MD method. The relevant

105: Lagrangian for the particles and the box in this case reads: 

106: \begin{equation}

107: L=\frac{1}{2}\sum_{i}m_{i}{\bf \dot{s}}_{i}G{\bf \dot{s}}_{i}-

108: \sum_{i,j>i}V(r_{ij})+\frac{1}{2}W{\rm Tr}(\dot{f}^{{\rm T}}\dot{f}),

109: \label{rpeq1}

110: \end{equation}

111: where $m_{i}$ is the mass of the $i$-th particle, ${\bf \dot{s}}_{i}$ is its

112: vectorial velocity, the spatial part $G$ of the spatiotemporal metric tensor

113: may be represented in terms of another matrix $f$ as $G=f^{{\rm T}}f$ ($G$

114: is positive definite). $^{{\rm T}}$ and ${\rm Tr}$ denote the transposition

115: and trace respectively, 

116: $r_{ij}$ and $V$ are distances between the particles and the

117: potential of interaction between them, and $W$ is an effective mass of the

118: box.

119: 

120: Our purpose in this work is to extend the RP-MD methodology

121: to the case of a continuum scalar field, 

122: coupled to either a spatially averaged geometric

123: characteristic (``average curvature''), which will give rise to an ODE, or

124: to a spatiotemporal curvature field, that will generate a PDE. 

125: The continuum field

126: may represent, e.g., a chemical concentration propagating over a membrane, 

127: or a salt solution, causing the swelling of a polymer gel \cite{yy}, 

128: or an envelope

129: wave of the electric

130: field in nanosystems.

131: The

132: spatiotemporal metric is assumed to have the simple form, 

133: \begin{equation}

134: g=\left( 

135: \begin{array}{cc}

136: -1 & 0 \\ 

137: 0 & G

138: \end{array}

139: \right) .  \label{rpeq2}

140: \end{equation}

141: We will first consider the general case, where $G$ is a $d\times d$ matrix, 

142: $d$ being the space dimension.

143: 

144: One can define a field-type generalization of the RP-MD model, with a scalar

145: field $\phi $, as follows: 

146: \begin{equation}

147: L=\int d^{d}x\left[ -g^{ij}\frac{\partial \phi }{\partial x^{i}}\frac{

148: \partial \phi }{\partial x^{j}}-V(\phi )\right] +\frac{1}{2}W{\rm Tr}(\dot{f}

149: ^{{\rm T}}\dot{f}),  \label{rpeq3}

150: \end{equation}

151: where $V(\phi )$ is a potential governing the nonlinear evolution of the field 

152: $\phi $ and $f=f(t)$ (only).

153: If $f$ (and hence $G$) is a function of both spatial coordinates and

154: time, an elastic-energy term \cite{elastic} should be added to the Lagrangian 

155: (\ref{rpeq3}), so that it becomes 

156: \[

157: L=\int d^{d}x\left[ -g^{ij}\frac{\partial \phi }{\partial x^{i}}\frac{

158: \partial \phi }{\partial x^{j}}-V(\phi )\right] + 

159: \]

160: \begin{equation}

161: \int d^{d}x\left\{ \frac{1}{2}W\left[ {\rm Tr}(\dot{f}^{{\rm T}}\dot{f})-

162: \frac{1}{2}{\rm Tr}(\frac{\partial f^{{\rm T}}}{\partial x_{i}}\frac{

163: \partial f}{\partial x^{i}})\right] \right\} .  \label{rpeq4}

164: \end{equation}

165: 

166: %This general formulation and can be adapted to the cases $d=1$, $2$ or $3$.

167: %Here we aim to put forward the simplest one-dimensional (1d) model, with the

168: %single component of the metric tensor, $G\equiv f^{2}$, where the scalar $f$

169: %can be a function of time $t$ only or of $x$ and $t$. 

170: 

171: An objective of this brief report is to propose two models that include the

172: feedback to the curvature, and, simultaneously,  admit as particular

173: solutions the (unperturbed) solutions for the flat 

174: (original) metric. In particular, we choose $G=1+f^2$ 

175: (so that the metric is positive definite and for $f=0$ has as a special

176: case the original Minkowskian metric). Notice a slight deviation in

177: our choice from the RP case of $G=f^2$ \cite{choice}. Although the

178: formulation of Eqs. (\ref{rpeq3})-(\ref{rpeq4}) is very general, we

179: hereafter focus on the $d=1$ case that we will examine

180: in more detail.

181: 

182: Assuming initially that $f=f(t)$ only 

183: (e.g., including only an effect of the ``mean

184: curvature'' on the scalar-field dynamics), the Lagrangian (\ref{rpeq3})

185: becomes 

186: \begin{equation}

187: L=\frac{1}{2}W\dot{f}^{2}+\int dx\left[ \frac{1}{2}\left( \frac{\partial

188: \phi }{\partial t}\right) ^{2}-\frac{1 + f^{2}}{2}\left( \frac{\partial \phi }

189: {\partial x}\right) ^{2}-V(\phi )\right] .  \label{rpeq6}

190: \end{equation}

191: %The equations of motion following from it are 

192: %\begin{eqnarray}

193: %\phi _{tt} &=&f^{2}\phi _{xx}-\frac{\partial V}{\partial \phi }

194: %\label{rpeq7} \\

195: %Wf_{tt} &=&-f\int \phi _{x}^{2}dx,  \label{rpeq8}

196: %\end{eqnarray}

197: %where the subscript stands for the corresponding derivative.

198: Then, the resulting equations of motion (to which we will hereafter refer

199: as model A) are

200: %we will refer to the following variation of Eqs. (\ref{rpeq7})-

201: %(\ref{rpeq8}) as a model A, 

202: \begin{eqnarray}

203: \phi _{tt} &=&(1+f^{2})\phi _{xx}-\frac{\partial V}{\partial \phi }

204: \label{rpeq13} \\

205: Wf_{tt} &=&-f\int \phi _{x}^{2}dx,  \label{rpeq14}

206: \end{eqnarray}

207: where the subscripts stand for the corresponding partial derivatives.

208: 

209: Notice that the function $f$ is directly related to the

210: scalar curvature of the 1-d space. In particular, the Ricci scalar, 

211: which is $R=2R_{1212}/det(g)$ \cite{weinberg} 

212: in the general case, in the 1-d case

213: is $R=-2 \tilde{f}_{tt}/\tilde{f}$, where $\tilde{f}=\sqrt{1+f^2}$.

214: %where the

215: %relevant entry of the curvature tensor is given by:

216: %$$

217: %R_{1212}=\frac{1}{2} 

218: %\left( \frac{\partial^2 g_{22}} 

219: %{\partial t^2} - \frac{\partial^2 g_{22}} 

220: %{\partial x^2} \right)-\frac{1}{4 g_{22}}

221: %\left[ \left(\frac{\partial g_{22}}{\partial t} \right)^2

222: %+\frac{\partial g_{11}}{\partial x} 

223: %\frac{\partial g_{22}}{\partial x} \right] 

224: %$$

225: %\begin{eqnarray}

226: %-\frac{1}{4 g_{11}}

227: % \left[ \left(\frac{\partial g_{11}}{\partial x} \right)^2

228: %+\frac{\partial g_{11}}{\partial t} 

229: %\frac{\partial g_{22}}{\partial t} \right]=f f_{tt}

230: %\label{rpeq12}

231: %\end{eqnarray}

232: %and hence the scalar curvature is directly 

233: %related to $f$ through $R=-2 f_{tt}/f$.

234: 

235: 

236: On the other hand, for a metric with both spatial and temporal

237: dependence (e.g., for $f=f(x,t)$), 

238: one arrives at the following Lagrangian: 

239: \[

240: L=\int dx\left[ \frac{1}{2}\left( \frac{\partial \phi }{\partial t}\right)

241: ^{2}-\frac{1 + f^{2}}{2}\left( \frac{\partial \phi }{\partial x}\right)

242: ^{2}-V(\phi )\right] + 

243: \]

244: %\[

245: \begin{equation}

246: \int dx\left[ \frac{W}{2}\left( \frac{\partial f}{\partial t}\right) ^{2}-

247: \frac{W}{2}\left( \frac{\partial f}{\partial x}\right) ^{2}\right] . 

248: \label{add1}

249: \end{equation}

250: %\]

251: The ensuing coupled equations for the scalar field and the curvature 

252: (to which we will refer as model B) are 

253: \begin{eqnarray}

254: \phi _{tt} &=&\left( (1+f^{2})\phi _{x}\right) _{x}-\frac{\partial V}

255: {\partial \phi }  \label{rpeq15} \\

256: Wf_{tt} &=&Wf_{xx}-f\phi _{x}^{2}.  \label{rpeq16}

257: \end{eqnarray}

258: 

259: %\begin{eqnarray}

260: %\phi _{tt} &=&(f^{2}\phi _{x})_{x}-\frac{\partial V}{\partial \phi }

261: %\label{rpeq10} \\

262: %W\left( f_{xx}-f_{tt}\right) &=&f\phi _{x}^{2}.  \label{rpeq11}

263: %\end{eqnarray}

264: 

265: %It is then particularly interesting to examine the effects of the 

266: %curvature for a specific nonlinear potential. 

267: As a particular application of models A and B, 

268: we examine the physically ubiquitous

269: sine-Gordon (sG) potential, $V(\phi )=1-\cos (\phi )$ \cite{eilbeck}. It is

270: clear that Eqs. (\ref{rpeq13})-(\ref{rpeq14}) and

271: (\ref{rpeq15})-(\ref{rpeq16}) have particular

272: solutions with $f=0$, for which the latter equation of each pair is satisfied

273: trivially, while the former reduces exactly to the sG equation. Basic

274: solitary-wave solutions of the sG equation are the topological soliton

275: (kink), 

276: \begin{equation}

277: \phi _{{\rm k}}(x,t)=4\tan ^{-1}\left[ \exp \left( \gamma

278: (x-x_{0}-vt)\right) \right] ,  \label{kink}

279: \end{equation}

280: where $v$ is its velocity, $\gamma =(1-v^{2})^{-1/2}$ is the Lorentz factor,

281: and $x_{0}$ is the initial position of the kink's center, and the breather, 

282: $\phi _{{\rm br}}(x,t)=4\tan^{-1} \{ \sqrt{(1-\omega ^{2})/\omega ^{2}}

283: \sin \left[ \omega \gamma (t-v(x-x_{0}))\right]  

284: \times \\

285: {\rm sech}\left[

286: \gamma \sqrt{1-\omega ^{2}}(x-x_{0}-vt)\right] \} $, where $\omega $

287: is the frequency of its internal oscillations ($0<\omega <1$).

288: 

289: %However, Eqs. (\ref{rpeq7}) and (\ref{rpeq8}), as well as (\ref{rpeq10}) and

290: %(\ref{rpeq11}), do not offer an appropriate model to study the coupling of

291: %the field dynamics to curvature -- first of all, because the flat-metric

292: %state (the one with $f={\rm const}$) containing a soliton is not a solution

293: %in this case. In fact, the kink and breather are bound to be destroyed in

294: %such a model. This occurs (in all the cases that we have examined, for both

295: %the kinks and breathers) through the formation of sharp gradients and the

296: %trapping of the solitons, while the geometrical factor $f$ develops

297: %oscillations around $f=0$ (see an example in Fig. \ref{rfig1}).

298: 

299: The results of the interaction of the kink with the curvature in model A

300: are shown in Fig. \ref{rfig2} \cite{ww}. 

301: The curvature variable $f$, initialized with

302: a small random value, performs smooth oscillations with a frequency of 

303: $\omega \approx 2.866$. Notice that this is natural in this case, since the

304: kink has an approximately fixed ``mass'', $M_{k}=\int_{-\infty }^{+\infty

305: }u_{x}^{2}dx$, which can be found to be $M_{k}=8.247$ for the velocity $%

306: v\approx 0.25$). Then, Eq. (\ref{rpeq14}) predicts the frequency of these

307: oscillations $\omega \approx \sqrt{M_{k}}=2.871$, which is very close to the

308: above-mentioned numerically exact value. Fast small-amplitude oscillations

309: of the kink's velocity, observed in Fig. \ref{rfig2} both with and without

310: the curvature, are due to ``hopping'' over sites of a lattice (with spacing 

311: $h=0.1$) employed in the numerical scheme which solves Eq. (\ref{rpeq14}).

312: Notice, however, that in the top panel the mean velocity is $\approx 0.2503$,

313: while in the bottom panel it is $\approx 0.2499$, hence the curvature

314: oscillations increase the kink's velocity. This may be anticipated due to

315: the presence of the positive definite factor $1+f^{2}$ in front of $\phi

316: _{xx}$ in Eq. (\ref{rpeq13}), which is expected to renormalize $v^{2}$.

317: 

318: The curvature-breather interaction in model A is shown in Fig. \ref

319: {rfig3}. The frequency of the breather does not change significantly (it

320: fluctuates between $0.87$ and $0.93$), but its amplitude decreases

321: substantially (by more than $50\%$), resulting in its becoming much more

322: mobile (the velocity increases to $\approx 0.425$ from the initial value 

323: $0.25$).

324: 

325: In model B, we examine collision of a kink with a localized pulse of the

326: substrate field $f$. For the breather, we have obtained results which are

327: qualitatively similar to those presented below for the kink. We create the

328: curvature pulse to the left or to the right of the kink. As $\phi _{x}$

329: vanishes far from the kink, the equation (\ref{rpeq16}) for $f$ becomes a

330: linear wave equation. Hence, we observe splitting of the pulse into left-

331: and one right-traveling ones. In the case where the kink is initially to the

332: left of the pulse, it collides with the left-propagating fragment of

333: the (split) pulse. In the opposite case,

334: the kink is eventually caught by the co-propagating right fragment of

335: the (split) pulse. Numerical

336: simulations shown in Fig. \ref{rfig4} demonstrate that the collision with

337: the counter-propagating pulse reduces the kink's velocity, while the

338: interaction with the co-propagating pulse gives rise to an increase of the

339: velocity. In particular, in the former case, the mean speed of the kink

340: after the collision is $\approx 0.2476$, while in the latter one, it

341: increases to $\approx 0.2527$. Notice that in both cases a small fragment of

342: the pulse that collides with the kink passes through it, while a larger

343: fraction of the pulse is reflected by it. The latter feature may be

344: explained by a momentum-balance analysis.

345: 

346: In conclusion, we have extended field theory in the spirit of

347: the Rahman-Parrinello

348: Molecular Dynamics technique. The resulting

349: equations couple the spatio-temporal evolution of the field to that of the

350: underlying curvature of the space which carries the field. Coupled equations

351: for the temporal or spatiotemporal evolution of the metric are obtained in a

352: general setting, and, as an example,

353: are solved together with the sine-Gordon field

354: equation. The purely temporal evolution of the metric has been found to

355: increase the velocity of the field solitons, while the model allowing

356: spatio-temporal evolution of the metric can induce both increase and

357: decrease of the velocity.

358: 

359: It would be particularly interesting to extend the models A and B to higher

360: dimensions. It is also worth studying how the local evolution of the curvature

361: affects kinematics and dynamics of the solitons, and to correlate such

362: observations with the behavior of reactant chemical concentrations in

363: chemical or biological environments with non-trivial geometry \cite{yannis}.

364: 

365: {\it Acknowledgements}: The authors acknowledge D. Maroudas for 

366: stimulating discussions. The support of NSF and AFOSR (IGK) and

367: NSF (DMS-0204585), UMass and the Clay Institute (PGK) is gratefully

368: acknowledged. Work at Los Alamos is supported by the US DoE.

369: 

370: \begin{references}

371: \bibitem{av1}  E. Sackmann, in R. Lipowsky and E. Sackmann (Eds.), The

372: Structure and Dynamics of Membranes: Handbook of Biological Physics, vol. 1

373: (Elsevier, Amsterdam, 1995).

374: 

375: \bibitem{av2}  U. Seifert, \newblock Adv. in Phys. {\bf 46}, 13 (1997).

376: 

377: \bibitem{gaid1}  W. Saenger. 

378: \newblock{\it Principles of Nucleic Acid

379: Structure} (Springer-Verlag: Berlin, 1984).

380: 

381: \bibitem{gaid2}  J. Meunier, D.

382: Langevin, and D. Boccara (Eds.),

383: {\it Physics of Amphiphilic Layers},  Springer-Verlag (Berlin, 1997).

384: 

385: \bibitem{gaid3}  Y. Imry. {\it Introduction to Mesoscopic Physics} (Oxford

386: University Press: New York, 1997).

387: 

388: \bibitem{gaid4}  C.M. Soukoulis

389: (Ed.), {\it Photonic Band Gaps and Localization}, Plenum Press 

390: (New York, 1993).

391: 

392: \bibitem{curv}  See e.g., Yu. B. Gaididei, S. F. Mingaleev, and P. L.

393: Christiansen, Phys. Rev. E {\bf 62}, R53 (2000); M. Ibanes, J. M. Sancho,

394: and G. P. Tsironis, Phys. Rev. E {\bf 65}, 041902 (2002);

395: 

396: \bibitem{antm} S.P. Timoshenko and S. Woinowsky-Krieger,

397: {\it Theory of plates and shells}, McGraw-Hill (New York, 1959)

398: S.S. Antman,

399: \newblock {\it Nonlinear problems of elasticity}, Springer-Verlag 

400: (New York 1995).

401: 

402: \bibitem{kmb}  P.G. Kevrekidis, B.A. Malomed and A.R. Bishop, \newblock

403: Phys. Rev. E {\bf 66}, 046621 (2002).

404: 

405: \bibitem{dand}  S. Villain-Guillot, R. Dandoloff, A. Saxena and A.R. Bishop,

406: \newblock Phys. Rev. B {\bf 52}, 6712 (1995); R. Dandoloff, S.

407: Villain-Guillot, A. Saxena and A.R. Bishop, \newblock Phys. Rev. Lett. {\bf

408: 74}, 813 (1995).

409: 

410: 

411: \bibitem{RP}  see e.g., M. Parrinello and A. Rahman, \newblock Phys. Rev.

412: Lett. {\bf 45}, 1196 (1980). M. Parrinello and A. Rahman, \newblock J. Appl.

413: Phys. {\bf 52}, 7182 (1981).

414: 

415: \bibitem{yy} E. C. Achilleos, R. K. Prud'homme, I. G. Kevrekidis, K. N. Christodoulou and

416: K.R. Gee,

417: { AIChE Journal}, {\bf 46}, 2128 (2000).

418: 

419: \bibitem{elastic} Elastic dynamics is a natural first approximation,

420: even though it is possible for the substrate  to have more 

421: complicated dynamical evolution laws (which can also be incorporated

422: in this setting).

423: 

424: \bibitem{choice} We also ran detailed simulations of the case with 

425: $G=f^2$. Here, the non-existence of the original steady state

426: of the field and its competition with the oscillatory behavior of $f$

427: around $f=0$ yields rather unphysical behavior involving large gradients

428: of the field. For these reasons, this case is not presented in detail here.

429: 

430: \bibitem{eilbeck}  R. K. Dodd, J. C. Eilbeck, J. D. Gibbon, and H. C.

431: Morris, Solitons and Nonlinear Wave Equations (Academic Press, London, 1982).

432: 

433: \bibitem{weinberg}  B. O'Neill, Semi-Riemannian Geometry with Applications

434: to Relativity (Academic Press, London, 1983).

435: 

436: \bibitem{ww} In the numerical computations $W$ has been fixed to $W=1$.

437: By varying $W$, we have found, similarly to the original works of RP,

438: that its increase slows down the rate at which the presented 

439: phenomenology occurs.

440: 

441: \bibitem{yannis}  J. Wolff, A.G. Papathanasiou, I.G. Kevrekidis, H.H.

442: Rotermund, and G. Ertl, Science {\bf 294}, 134 (2001).

443: \end{references}

444: 

445: \end{multicols}

446: 

447: %\centerline{{\bf Figure Captions}}

448: 

449: %\begin{figure}[tbp]

450: %\centering

451: %{\epsfig{file=cn1.ps, width=6.4cm,angle=0, clip=}}

452: %\caption{Results of simulations of Eqs. (\ref{rpeq7}) and (\ref{rpeq8}) with

453: %the initial configuration in the form of the sine-Gordon kink (\ref{kink})

454: %with the velocity $v=0.25$ in the $\protect\phi$) field. The metric variable 

455: %$f$ is initialized at a small random value. The top left and right panel

456: %show, respectively, the $\protect\phi$-field profile at $t=28.5$, and the

457: %actual position of the kink's center (solid line) vs. the position that it

458: %would have, moving at the constant velocity equal to the initial value.

459: %(dashed line). The bottom panel shows the time evolution of $f$.}

460: %\label{rfig1}

461: %\end{figure}

462: 

463: \begin{figure}[tbp]

464: \centering

465: {\epsfig{file=rpfig1a.ps, width=6.4cm,angle=0, clip=}}

466: \centering

467: {\epsfig{file=rpfig1b.ps, width=6.4cm,angle=0, clip=}}

468: \caption{The top left panel shows the kink's spatial profile at $t=60$ in

469: model A. The initial kink profile (centered around the new center

470: position) is shown by the dashed line, and is practically indistinguishable

471: from the solution at $t=60$, indicating that the kink maintains its shape.

472: The bottom left panel shows smooth oscillations of $f(t)$. The top and

473: bottom right panels show, respectively, the kink's velocity $v(t)$ vs. $t$,

474: and the same quantity but for $f=0$.}

475: \label{rfig2}

476: \end{figure}

477: 

478: \begin{figure}[tbp]

479: \centering

480: {\epsfig{file=rpfig2.ps, width=6.4cm,angle=0, clip=}}

481: \caption{The top left and right panels show, respectively, the breather in

482: model A at the end of the simulation period, $t=120$, and its position

483: vs. time (solid), as compared to that which it would have moving at the

484: initial velocity, $v=0.25$ (dashed). The middle left panel shows the time

485: evolution of $f(t)$, while the middle right and bottom left panels present

486: exchange between the curvature-mode's energy, $E_f= f_t^2/2 + (f^2/4) \int 

487: \protect\phi_x^2 dx$, and the rest of the energy, $E_{\protect\phi}=E-E_f$.

488: Finally, the bottom right panel shows the time evolution of the field 

489: $\protect\phi$ at the center of the breather.}

490: \label{rfig3}

491: \end{figure}

492: 

493: \begin{figure}[tbp]

494: \centering

495: {\epsfig{file=rpfig3a.ps, width=6.4cm,angle=0, clip=}}

496: \centering

497: {\epsfig{file=rpfig3b.ps, width=6.4cm,angle=0, clip=}}

498: \centering

499: {\epsfig{file=rpfig3c.ps, width=6.4cm,angle=0, clip=}}

500: \centering

501: {\epsfig{file=rpfig3d.ps, width=6.4cm,angle=0, clip=}}

502: \caption{The left part of the figure shows spatial profiles of the fields 

503: $\protect\phi$ and $f$ at $t=60$ in model B. It also shows the time

504: evolution of the energies $E_{\protect\phi}=E-E_f$ and $E_f=\int (1/2)

505: (f_t^2+f_x^2+ (f \protect\phi_x)^2/2 ) dx$, as well as the kink's velocity

506: as a function of time. The kink's center is initially at $x=90$, while the

507: center of the curvature pulse is at $x=110$. The right subplots show

508: counterparts of those in the left part, but for the case of the kink and

509: curvature pulse initially centered at $x=110$ and $x=90$, respectively.}

510: \label{rfig4}

511: \end{figure}

512: 

513: \end{document}

514: