1: \documentclass[multicol,aps,pre,epsf,amsmath,citesort,twocolumn]{revtex4}

2: \usepackage{graphicx,amsfonts,amsmath}

3: 

4: \def \BEA {\begin{eqnarray}}

5: \def \EEA {\end{eqnarray}}

6: \def \BC {\begin{cases}}

7: \def \EC {\end{cases}}

8: \def \ve {\varepsilon}

9: \def \d {\delta}

10: \def \l {\lambda}

11: \def \wt {\tilde{\omega}}

12: \def \Tt {\tilde{T}}

13: \def \tt {\tilde{t}}

14: \def \w {\omega}

15: \def \dw {\Delta\omega}

16: \def \a {\alpha}

17: \def \b {\beta}

18: \def \g {\gamma}

19: \def \at {\tilde{a}}

20: \def \af {\hat{a}}

21: \def \la {\langle}

22: \def \ra {\rangle}

23: \def \D {\Delta}

24: \def \tth {\tilde{\theta}}

25: \def \ffi {\varphi}

26: \def \i {\imath}

27: \def \eqN {\overset{N}{=}}

28: \def \eq1 {\overset{1}{=}}

29: \def \eqpi {\overset{\pi}{=}}

30: \def \p {\partial}

31: \def \kb {\bar{k}}

32: \def \eb {\bar{e}}

33: \def \ub {\bar{u}_2}

34: \def \T {\theta}

35: \def \tinf {\theta_{\infty}}

36: 

37: \begin{document}

38: 

39: \title{Interactions of renormalized waves in thermalized Fermi-Pasta-Ulam chains}

40: %

41: %Four-wave resonant interactions of renormalized waves in thermalized

42: %$\beta$-Fermi-Pasta-Ulam chains}

43: \author{Boris Gershgorin$^1$, Yuri V. Lvov$^1$ and David Cai$^2$, \\ \ \\

44: %

45: $^1$ \textit{Department of Mathematical Sciences, Rensselaer

46: Polytechnic

47: Institute, Troy, NY 12180}\\

48: %

49: $^2$ \textit{Courant Institute of Mathematical Sciences,

50:         New York University, New York, NY 10012}\\

51: }

52: \begin{abstract}

53: The dispersive interacting waves in Fermi-Pasta-Ulam (FPU) chains of

54: particles in \textit{thermal equilibrium} are studied from both

55: statistical and wave resonance perspectives. It is shown that, even

56: in a strongly nonlinear regime, the chain in thermal equilibrium can

57: be effectively described by a system of weakly interacting

58: \textit{renormalized} nonlinear waves that possess (i) the

59: Rayleigh-Jeans distribution and (ii) zero correlations between

60: waves, just as noninteracting free waves would. This renormalization

61: is achieved through a set of canonical transformations. The

62: renormalized linear dispersion of these renormalized waves is

63: obtained and shown to be in excellent agreement with numerical

64: experiments. Moreover, a dynamical interpretation of the

65: renormalization of the dispersion relation is provided via a

66: self-consistency, mean-field argument. It turns out that this

67: renormalization arises mainly from the trivial resonant wave

68: interactions, i.e., interactions with no momentum exchange.

69: Furthermore, using a multiple time-scale, statistical averaging

70: method, we show that the interactions of near-resonant waves give

71: rise to the broadening of the resonance peaks in the frequency

72: spectrum of renormalized modes. The theoretical prediction for the

73: resonance width for the thermalized $\beta$-FPU chain is found to be

74: in very good agreement with its numerically measured value.

75: \end{abstract}

76: \maketitle

77: \section{Introduction}

78: The study of discrete one-dimensional chains of particles with the

79: nearest-neighbor interactions provides insight to the dynamics of

80: various physical and biological systems, such as crystals, wave

81: systems, and biopolymers~\cite{fpu50, Peyrard, Toda}. In the thermal

82: equilibrium state, such nonlinear chains can be described by the

83: canonical Gibbs measure~\cite{LL5} with the Hamiltonian \BEA

84: H=\sum_j

85: \frac{p_j^2}{2}+\frac{(q_j-q_{j+1})^2}{2}+V(q_j-q_{j+1}),\label{intro_Hpq}

86: \EEA where $p_j$ and $q_j$ are the momentum and the displacement

87: from the equilibrium position of the $j$-th particle, respectively,

88: $V(q_j-q_{j+1})$ is the anharmonic part of the potential, and the

89: mass of each particle and the linear spring constant are scaled to

90: unity. In this article, we only consider the potentials of the

91: \textit{restoring} type, i.e., the potentials for which the Gibbs

92: measure exists. In order to study interactions of waves in such

93: systems, one usually introduces the canonical complex normal

94: variables $a_k$ via \BEA

95: a_k=\frac{P_k-\i\w_kQ_k}{\sqrt{2\w_k}},\label{intro_a} \EEA where

96: $P_k$ and $Q_k$ are the Fourier transforms of $p_j$ and $q_j$,

97: respectively, and $\w_k=2\sin(\pi k/N)$ is the linear dispersion

98: relation of the waves represented by $a_k$. In terms of the $a_k$,

99: the Hamiltonian~(\ref{intro_Hpq}) becomes \BEA

100: H=\sum\w_k|a_k|^2+V(a)\label{Ham_a}, \EEA where $V(a)$ is the

101: combination of various products of $a_k$ and $a_k^*$ corresponding

102: to various wave-wave interactions. If the potential in

103: Eq.~(\ref{intro_Hpq}) is harmonic, i.e., $V\equiv 0$, then $a_k$

104: correspond to ideal, \textit{free} waves, which have no energy

105: exchanges among different $k$ modes. In thermal equilibrium, the

106: Boltzmann distribution $ \exp(-\theta^{-1}\sum\w_k|a_k|^2) $ with

107: temperature $\T$, gives rise to the following properties of free

108: waves \BEA

109: \la a_k^*a_l\ra&=&n_k\delta^k_l,\label{intro_asa}\\

110: \la a_ka_l\ra&=&0\label{intro_aa}, \EEA for any $k$ and $l$, where

111: $n_k\equiv\la|a_k|^2\ra=\theta/\w_k$ is the power spectrum. If the

112: anharmonic part of the potential is sufficiently weak, then

113: corresponding waves $a_k$ remain almost free, and

114: Eqs.~(\ref{intro_asa}) and~(\ref{intro_aa}) would be approximately

115: satisfied in the weakly nonlinear regime. However, when the

116: nonlinearity becomes stronger, waves $a_k$ become strongly

117: correlated, and, in general, the correlations between waves

118: [Eq.~(\ref{intro_aa})] no longer vanish. In particular, $\la

119: a_ka_{N-k}\ra\neq 0$, as will be shown below. Naturally, the

120: question arises: can the strongly nonlinear system in thermal

121: equilibrium still be viewed as a system of almost free waves in some

122: statistical sense? In this article, we address this question with an

123: affirmative answer: it turns out that the system~(\ref{intro_Hpq})

124: can be described by a complete set of \textit{renormalized}

125: canonical variables $\at_k$, which still possess the wave properties

126: given by Eqs.~(\ref{intro_asa}) and~(\ref{intro_aa}) with a

127: renormalized linear dispersion. The waves that correspond to these

128: new variables $\at_k$ will be referred to as \textit{renormalized}

129: waves. Since these renormalized waves possess the equilibrium

130: Rayleigh-Jeans distribution~\cite{ZLF} and vanishing correlations

131: between waves, they resemble free, non-interacting waves, and can be

132: viewed as statistical normal modes. Furthermore, it will be

133: demonstrated that the renormalized linear dispersion for these

134: renormalized waves has the form $\wt_k=\eta(k)\w_k$, where $\eta(k)$

135: is the linear frequency renormalization factor, and is

136: \textit{independent} of $k$ as a consequence of the Gibbs measure.

137: 

138: In our method, the construction of the renormalized variables

139: $\at_k$ does not depend on a particular form or strength of the

140: anharmonic potential, as long as it is of the restoring type with

141: only the nearest neighbor interactions, as in Eq.~(\ref{intro_Hpq}).

142: Therefore, our approach is non-perturbative and can be applied to a

143: large class of systems with strong nonlinearity. However, in this

144: article, we will focus on the $\beta$-FPU chain to illustrate the

145: theoretical framework of the renormalized waves. We will verify that

146: $\at_k$ effectively constitute normal modes for the $\beta$-FPU

147: chain in thermal equilibrium by showing that (i) the theoretically

148: obtained renormalized linear dispersion relationship is in excellent

149: agreement with its dynamical manifestation in our numerical

150: simulation, and (ii) the equilibrium distribution of $\at_k$ is

151: still a Rayleigh-Jeans distribution and $\at_k$'s are uncorrelated.

152: Note that similar expressions for the renormalization factor $\eta$

153: have been previously discussed in the framework of an approximate

154: virial theorem~\cite{Alabiso} or effective long wave dynamics via

155: the Zwanzig-Mori projection~\cite{Lepri_renorm}. However, in our

156: theory, the exact formula for the renormalization factor is derived

157: from a \textit{precise} mathematical construction of statistical

158: normal modes, and is valid for all wave modes $k$ --- no longer

159: restricted to long waves.

160: 

161: Next, we address how renormalization arises from the dynamical wave

162: interaction in the $\beta$-FPU chain.

163: %Here, we will invoke the formalism of \textit{wave turbulence}~\cite{WT}.

164: %Wave turbulence theory provides a convenient way of describing resonant spectral energy transfer in a system of weakly interacting waves.

165: %This transfer occurs via the wave interactions on the resonance manifold, which is

166: %one of the main objects of study in wave turbulence.

167: We will show that the $\beta$-FPU chain can be \textit{effectively}

168: described as a four-wave interacting Hamiltonian system of the

169: renormalized resonant waves $\at_k$. We will study the resonance

170: structure of the $\beta$-FPU chain and find that most of the exact

171: resonant interactions are \textit{trivial}, i.e., the interactions

172: with no momentum exchange among different wave modes. In what

173: follows, the renormalization of the linear dispersion will be

174: explained as a collective effect of these trivial resonant

175: interactions of the renormalized waves $\at_k$. We will use a

176: self-consistency argument to find an approximation, $\eta_{sc}$, of

177: the renormalization factor $\eta$. As will be seen below, the

178: self-consistency argument essentially is of a mean-field type, i.e.,

179: the renormalization arises from the scattering of a wave by a

180: mean-background of waves in thermal equilibrium via trivial resonant

181: interactions. We note that our self-consistency, mean-field argument

182: is not limited to the weak nonlinearity. Very good agreement of the

183: renormalization factor $\eta$ and its dynamical approximation

184: $\eta_{sc}$ --- for weakly as well as strongly nonlinear waves ---

185: confirms that the renormalization is, indeed, \textit{a direct

186: consequence} of the trivial resonances.

187: 

188: 

189: We will further study the properties of these renormalized waves by

190: investigating how long these waves are coherent, i.e., what their

191: frequency widths are. Therefore, we consider \textit{near-resonant}

192: interactions of the renormalized waves $\at_k$, i.e., interactions

193: that occur in the vicinity of the resonance manifold, since most of

194: the exact resonant interactions are trivial, i.e., with no momentum

195: exchanges, and they, cannot effectively redistribute energy among

196: the wave modes.

197: 

198: We will demonstrate that near-resonant interactions of the

199: renormalized waves $\at_k$ provide a mechanism for effective energy

200: exchanges among different wave modes. Taking into account the

201: near-resonant interactions, we will study analytically the frequency

202: peak broadening of the renormalized waves $\at_k$ by employing a

203: multiple time-scale, statistical averaging method. Here, we will

204: arrive at a theoretical prediction of the spatiotemporal spectrum

205: $|\af_k(\w)|^2$, where $\af_k(\w)$ is the Fourier transform of the

206: normal variable $\at_k(t)$, and $\w$ is the frequency. The predicted

207: width of frequency peaks is found to be in good agreement with its

208: numerically measured values.

209: 

210: In addition, for a finite $\beta$-FPU chain, we will mention the

211: consequence, to the correlation times of waves, of the momentum

212: exchanges that cross over the first Brillouin zone. This process is

213: known as the umklapp scattering in the setting of phonon

214: scattering~\cite{Umklapp}. Note that, in the previous

215: studies~\cite{Kramer} of the FPU chain from the wave turbulence

216: point of view, the effects arising from the finite  nature of the

217: chain were not taken into account, i.e., only the limiting case of

218: $N\rightarrow\infty$, where $N$ is the system size, was considered.

219: 

220: The article is organized as follows. In

221: Section~\ref{sect_Hformalism}, we discuss a chain of particles with

222: the nearest-neighbor nonlinear interactions. We demonstrate how to

223: describe a strongly nonlinear system as a system of waves that

224: resemble free waves in terms of the power spectrum and vanishing

225: correlations between waves. We show how to construct the

226: corresponding renormalized variables with the renormalized linear

227: dispersion. In Section~\ref{sect_FPU}, we rewrite the $\beta$-FPU

228: chain as an interacting four-wave Hamiltonian system. We study the

229: dynamics of the chain numerically and find excellent agreement

230: between the renormalized dispersion, obtained analytically and

231: numerically. In Section~\ref{sect_Dispersion}, we describe the

232: resonance manifold analytically and illustrate its controlling role

233: in long-time averaged dynamics using numerical simulation. In

234: Section~\ref{sect_Renormalization}, we derive an approximation for

235: the renormalization factor for the linear dispersion using a

236: self-consistency condition. In Section~\ref{sect_Width}, we study

237: the broadening effect of frequency peaks  and predict analytically

238: the form of the spatiotemporal spectrum for the $\beta$-FPU chain.

239: We provide the comparison of our prediction with the numerical

240: experiment. We present the conclusions in

241: Section~\ref{sect_Conclusions}.

242: \section{Renormalized waves}

243: \label{sect_Hformalism} Consider a chain of particles coupled via

244: nonlinear springs. Suppose the total number of particles is $N$ and

245: the momentum and displacement from the equilibrium position of the

246: $j$-th particle are $p_j$ and $q_j$, respectively. If only the

247: nearest-neighbor interactions are present, then the chain can be

248: described by the Hamiltonian \BEA H=H_2+V,\label{Ham_pq} \EEA where

249: the quadratic part of the Hamiltonian takes the form \BEA

250: H_2=\frac{1}{2}\sum_{j=1}^N{p_j^2}+(q_j-q_{j+1})^2,\label{H2} \EEA

251: and the anharmonic potential $V$ is the function of the relative

252: displacement $q_j-q_{j+1}$. Here periodic boundary conditions

253: $q_{N+1}\equiv q_1$ and $p_{N+1}\equiv p_1$ are imposed. Since the

254: total momentum of the system is conserved, it can be set to zero.

255: 

256: In order to study the distribution of energy among the wave modes,

257: we transform the Hamiltonian to the Fourier variables $Q_k$, $P_k$

258: via \BEA \BC

259: Q_k=\displaystyle{\frac{1}{\sqrt{N}}}\sum_{j=0}^{N-1}q_je^{\frac{2\pi \i kj}{N}},\\

260: P_k=\displaystyle{\frac{1}{\sqrt{N}}}\sum_{j=0}^{N-1}p_je^{\frac{2\pi

261: \i kj}{N}}. \EC\label{PQ} \EEA This transformation is

262: canonical~\cite{LL1,Licht} and the Hamiltonian~(\ref{Ham_pq})

263: becomes \BEA

264: H&=&\frac{1}{2}\sum_{k=1}^{N-1}|P_k|^2+\omega_k^2|Q_k|^2+V(Q),\label{Ham_PQ}

265: \EEA where $\w_k=2\sin(\pi k/N)$ is the linear dispersion relation.

266: Note that, throughout the paper, for the simplicity of notation, we

267: denote the periodic wave number space by the set of integers in the

268: range $[0,N-1]$, i.e., we drop the conventional factor, $2\pi/N$.

269: The zeroth mode vanishes due to the fact that the total momentum is

270: zero.

271: 

272: If the system~(\ref{Ham_PQ}) is in thermal equilibrium, then the

273: canonical Gibbs measure, with the corresponding partition function

274: \BEA Z=\int_{-\infty}^{\infty}e^{-H(p,q)/\theta}dpdq,\label{Zpq}

275: \EEA with the temperature $\theta$, can be used to describe the

276: statistical behavior of the system. We consider the systems with the

277: anharmonic potential of the restoring type, i.e., the potential for

278: which the integral in Eq.~(\ref{Zpq}) converges. It can be easily

279: shown that for system~(\ref{Ham_PQ}) the average kinetic energy

280: $K_k$ of each mode is independent of the wave number \BEA \la

281: K_k\ra=\la K_l\ra,\label{Kkind} \EEA where $k$ and $l$ are wave

282: numbers, $K_k\equiv|P_k|^2/2$, and $\la\dots\ra$ denotes averaging

283: over the Gibbs measure. Similarly, the average quadratic potential

284: $U_k$ of each mode is independent of the wave number \BEA \la

285: U_k\ra=\la U_l\ra,\label{Qkind} \EEA where

286: $U_k\equiv\w_k^2|Q_k|^2/2$.

287: 

288: If the nonlinear interactions are weak, then it is convenient to

289: further transform the Hamiltonian~(\ref{Ham_PQ}) to the complex

290: normal variables defined by Eq.~(\ref{intro_a}). This transformation

291: is canonical, i.e., the dynamical equation of motion becomes \BEA

292: \i\dot{a}_k=\frac{\partial H}{\partial a_k^*}.\label{a_canonical}

293: \EEA In terms of these normal variables, the

294: Hamiltonian~(\ref{Ham_PQ}) takes the form~(\ref{Ham_a}). For the

295: system of noninteracting waves, i.e.,

296: $H=\sum_{k=1}^{N-1}\w_k|a_k|^2$, we obtain a standard virial theorem

297: in the form \BEA \la K_k\ra\vline_{V=0}=\la

298: U_k\ra\vline_{V=0}.\label{free} \EEA As a consequence of this virial

299: theorem, we have the properties of free waves, which were already

300: mentioned above [Eqs.~(\ref{intro_asa}) and~(\ref{intro_aa})], i.e.,

301: \BEA

302: \la a_k^*a_l\ra&=&\frac{1}{2\w_k}(\la|P_k|^2\ra+\w_k^2\la|Q_k|^2\ra)\delta^k_l=\frac{\theta}{\w_k}\delta^k_l,\label{asa}\\

303: \la

304: a_ka_l\ra&=&\frac{1}{2\w_k}(\la|P_k|^2\ra-\w_k^2\la|Q_k|^2\ra)\delta^{k+l}_N=0,\label{aa}

305: \EEA for all wave numbers $k$ and $l$. Note that

306: equation~(\ref{asa}) gives the classical Rayleigh-Jeans distribution

307: for the power spectrum of free waves~\cite{ZLF} \BEA

308: n_k=\frac{\T}{\w_k}.\label{PSa} \EEA However, if the nonlinearity is

309: present, the waves $a_k$ and $a_{N-k}$ become correlated, i.e., \BEA

310: \la

311: a_ka_{N-k}\ra=\frac{1}{2\w_k}(\la|P_k|^2\ra-\w_k^2\la|Q_k|^2\ra)\neq

312: 0,\label{akNk} \EEA since the property~(\ref{free}) is no longer

313: valid.

314: 

315: As we mentioned before, a complete set of new renormalized variables

316: $\at_k$ can be constructed, so that the strongly nonlinear system

317: can be viewed as a system of ``free'' waves in the sense of

318: vanishing correlations and the power spectrum, i.e., the new

319: variables $\at_k$ satisfy the properties of free waves given in

320: Eqs.~(\ref{asa}) and~(\ref{aa}). Next, we show how to construct

321: these renormalized variables $\at_k$.

322: 

323: Consider the generalization of the transformation~(\ref{intro_a}),

324: namely, the transformation from the Fourier variables $Q_k$ and

325: $P_k$ to the renormalized variables $\at_k$ by \BEA

326: \at_k=\frac{P_k-\i\wt_kQ_k}{\sqrt{2\wt_k}},\label{at} \EEA where

327: $\wt_k$ is an arbitrary function with the only restrictions \BEA

328: \wt_k>0,~~\wt_k=\wt_{N-k}.\label{canonical} \EEA One can show that,

329: these restrictions~(\ref{canonical}) provide a necessary and

330: sufficient condition for the transformation~(\ref{at}) to be

331: canonical. For the renormalized waves $\at_k$, we can compute \BEA

332: \la \at_k^*\at_l\ra&=&\frac{1}{2\wt_k}(\la|P_k|^2\ra+\wt_k^2\la|Q_k|^2\ra)\delta^k_l,\label{asat}\\

333: \la

334: \at_k\at_l\ra&=&\frac{1}{2\wt_k}(\la|P_k|^2\ra-\wt_k^2\la|Q_k|^2\ra)\delta^{k+l}_N.\label{aat}

335: \EEA Since we have the freedom of choosing any $\wt_k$ (with the

336: only restrictions~(\ref{canonical})), we can chose $\wt_k$ such that

337: $\la \at_k\at_{N-k}\ra$ vanishes. Thus, the renormalized variables

338: $\at_k$ for a strongly nonlinear system will behave like the bare

339: variables $a_k$ for a noninteracting system in terms of vanishing

340: correlations between waves. Therefore, we determine $\wt_k$ via \BEA

341: \la|P_k|^2\ra-\wt_k^2\la|Q_k|^2\ra=0.\label{vanish} \EEA Note that

342: the requirement~(\ref{vanish}) has the form of the virial theorem

343: for the free waves but with the renormalized linear dispersion

344: $\wt_k$. We rewrite Eq.~(\ref{vanish}) in terms of the kinetic and

345: quadratic potential parts of the energy of the mode $k$ as \BEA

346: \frac{\wt_k}{\w_k}=\sqrt{\frac{\la K_k\ra}{\la

347: U_k\ra}}.\label{vanish2} \EEA The  in Eqs.~(\ref{Kkind})

348: and~(\ref{Qkind}) leads to the $k$ independence of the right-hand

349: side of Eq.~(\ref{vanish2}). This allows us to define \textit{the

350: renormalization factor} $\eta$ for \textit{all} $k$'s by \BEA

351: \eta\equiv\frac{\wt_k}{\w_k}=\sqrt{\frac{\la K\ra}{\la

352: U\ra}}.\label{eta} \EEA for dispersion $\w_k$. Here

353: $K=\sum_{k=1}^{N-1}K_k$ and $U=\sum_{k=1}^{N-1}U_k$ are the kinetic

354: and the quadratic potential parts of the total energy of the

355: system~(\ref{Ham_PQ}), respectively. Note that the way of

356: constructing the renormalized variables $\at_k$ via the precise

357: requirement of vanishing correlations between waves yields the exact

358: expression for the renormalization factor, which is valid for all

359: wave numbers $k$ and any strength of nonlinearity. The independence

360: of $\eta$ of the wave number $k$ is a consequence of the Gibbs

361: measure. This $k$ independence phenomenon has been observed in

362: previous numerical experiments~\cite{our_prl,Alabiso}. We will

363: elaborate on this point in the results of the  numerical experiment

364: presented in Section~\ref{sect_FPU}.

365: 

366: The immediate consequence of the fact that $\eta$ is independent of

367: $k$ is that the power spectrum of the renormalized waves possesses

368: the precise Rayleigh-Jeans distribution, i.e., \BEA

369: \tilde{n}_k=\frac{\T}{\wt_k},\label{nt} \EEA from Eq.~(\ref{asat}),

370: where $\tilde{n}_k=\la |\at_k|^2\ra$. Combining Eqs.~(\ref{intro_a})

371: and~(\ref{at}), we find the relation between the ``bare'' waves

372: $a_k$ and the renormalized waves $\at_k$ to be \BEA

373: a_k=\frac{1}{2}\left(\sqrt{\eta}+\frac{1}{\sqrt{\eta}}\right)\at_k+\frac{1}{2}\left(\sqrt{\eta}-\frac{1}{\sqrt{\eta}}\right)\at_{N-k}.\label{a2at}

374: \EEA Using Eq.~(\ref{a2at}), we obtain the following form of the

375: power spectrum for the bare waves $a_k$ \BEA

376: n_k=\frac{1}{2}\left(1+\frac{1}{\eta^2}\right)\frac{\T}{\w_k},\label{nkbare}

377: \EEA which is a modified Rayleigh-Jeans distribution due to the

378: renormalization factor $(1+1/\eta^2)/2$. Naturally, if the

379: nonlinearity becomes weak, we have $\eta\rightarrow 1$, and,

380: therefore, all the variables and parameters with tildes reduce to

381: the corresponding ``bare'' quantities, in particular,

382: $\wt_k\rightarrow\w_k$, $\at_k\rightarrow a_k$,

383: $\tilde{n}_k\rightarrow n_k$. It is interesting to point out that,

384: even in a strongly nonlinear regime, the ``free-wave'' form of the

385: Rayleigh-Jeans distribution is satisfied

386: \textit{exactly}~[Eq.~(\ref{nt})] by the renormalized waves. Thus,

387: we have demonstrated that even in the presence of strong

388: nonlinearity, the system in thermal equilibrium can still be viewed

389: statistically as a system of ``free'' waves in the sense of

390: vanishing correlations between waves and the power spectrum.

391: 

392: Note that, in the derivation of the formula for the renormalization

393: factor [Eq.~(\ref{eta})], we only assumed the nearest-neighbor

394: interactions, i.e., the potential is the function of $q_j-q_{j+1}$.

395: One of the well-known examples of such a system is the $\beta$-FPU

396: chain, where only the forth order nonlinear term in $V$ is present.

397: In the remainder of the article, we will focus on the $\beta$-FPU to

398: illustrate the framework of the renormalized waves $\at_k$.

399: \section{Numerical study of the $\beta$-FPU chain}

400: \label{sect_FPU} Since its introduction in the early 1950s, the

401: study of the FPU lattice~\cite{FPU} has led to many great

402: discoveries in mathematics and physics, such as soliton

403: theory~\cite{Toda}. Being non-integrable, the FPU system also became

404: intertwined with the celebrated Kolmogorov-Arnold-Moser

405: theorem~\cite{Licht}. Here, we extend our results of the thermalized

406: $\beta$-FPU chain, which were briefly reported in~\cite{our_prl}.

407: 

408: The Hamiltonian of the $\beta$-FPU chain is of the form \BEA

409: H=\sum_{j=1}^N\frac{1}{2}p_j^2+\frac{1}{2}(q_j-q_{j+1})^2+\frac{\beta}{4}(q_j-q_{j+1})^4,\label{H_FPU}

410: \EEA where $\beta$ is a parameter that characterizes the strength of

411: nonlinearity.

412: 

413: The canonical equations of motion of the $\beta$-FPU chain are \BEA

414: \dot{q}_j&=&\frac{\partial H}{\partial p_j}=p_j,\label{dyn_pq_1}\\

415: \dot{p}_j&=&-\frac{\partial H}{\partial q_j}=(q_{j-1}-2q_j+q_{j+1})\nonumber\\

416: &+&\beta\big[(q_{j+1}-q_j)^3-(q_j-q_{j-1})^3\big].\nonumber \EEA To

417: investigate the dynamical manifestation of the renormalized

418: dispersion $\wt_k$ of $\at_k$, we numerically integrate

419: Eq.~(\ref{dyn_pq_1}). Since we study the thermal equilibrium

420: state~\cite{Ford,Alabiso_fpu,Lepri_fpu,Carati} of the $\beta$-FPU

421: chain, we use random initial conditions, i.e., $p_j$ and $q_j$ are

422: selected at random from the uniform distribution in the intervals

423: $(-p_{\rm{max}},p_{\rm{max}})$ and $(-q_{\rm{max}},q_{\rm{max}})$,

424: respectively, with the two constraints that (i) the total momentum

425: of the system is zero and (ii) the total energy of the system $E$ is

426: set to be a specified constant. We have verified that the results

427: discussed in the paper do not depend on details of the initial data.

428: Note that the behavior of $\beta$-FPU for fixed $N$ is fully

429: characterized by only one parameter $\beta E$~\cite{Poggy}. We use

430: the sixth order symplectic Yoshida algorithm~\cite{Yoshida} with the

431: time step $dt=0.01$, which ensures the conservation of the total

432: system energy up to the ninth significant digit for a runtime

433: $\tau=10^6$ time units. In order to confirm that the system has

434: reached the thermal equilibrium state~\cite{Parisi}, the value of

435: the energy localization~\cite{Cretegny} was monitored via

436: $L(t)\equiv{N\sum_{j=1}^{N}G_j^2}/{(\sum_{j=1}^{N}G_j)^2}$, where

437: $G_j$ is the energy of the $j$-th particle defined as \BEA

438: G_j&=&\frac{1}{2}p_j^2+\frac{1}{4}\big[(q_j-q_{j+1})^2+(q_{j-1}-q_j)^2\big]\nonumber\\

439: &+&\frac{\beta}{8}\big[(q_j-q_{j+1})^4+(q_{j-1}-q_j)^4\big]. \EEA If

440: the energy of the system is concentrated around one site, then

441: $L(t)=O(N)$. Whereas, if the energy is uniformly distributed along

442: the chain, then $L(t)=O(1)$. In our simulations, in thermal

443: equilibrium states, $L(t)$ is fluctuating in the range of $1$-$3$.

444: Since our simulation is of microcanonical ensemble, we have

445: monitored various statistics of the system to verify that the

446: thermal equilibrium state that is consistent with the Gibbs

447: distribution~(canonical ensemble) has been reached. Moreover, we

448: verified that, for $N$ as small as 32 and up to as large as 1024,

449: the equilibrium distribution in the thermalized state in our

450: microcanonical ensemble simulation is consistent with the Gibbs

451: measure. We compared the renormalization factor~(\ref{eta}) by

452: computing the values of $\la K\ra$ and $\la U\ra$ numerically and

453: theoretically using the Gibbs measure and found the discrepancy of

454: $\eta$ to be within $0.1\%$ for $\beta=1$ and the energy density

455: $E/N=0.5$ for $N$ from 32 to 1024.

456: 

457: We now address numerically how the renormalized linear dispersion

458: $\wt_k$ manifests itself in the dynamics of the $\beta$-FPU system.

459: We compute the spatiotemporal spectrum $|\af_k(\w)|^2$, where

460: $\af_k(\w)$ is the Fourier transform of $\at_k(t)$. (Note that, for

461: simplicity of notation, we drop a tilde in $\af_k$.)

462: \begin{figure}

463: \includegraphics[width=3in, height=2.5in]{awk_num}

464: \caption{The spatiotemporal spectrum $|\af_k(\w)|^2$ in thermal

465: equilibrium. The chain was modeled for $N=256$, $\beta=0.5$, and

466: $E=100$. [$\max\{-8,\ln{|\af_k(\w)|^2}\}$, with corresponding gray

467: scale, is plotted for a clear presentation]. The solid curve

468: corresponds to the usual linear dispersion $\w_k=2\sin(\pi k/N)$.

469: The dashed curve shows the locations of the actual frequency peaks

470: of $|\af_k(\w)|^2$.} \label{Fig_awk_num}

471: \end{figure}

472: Figure~\ref{Fig_awk_num} displays the spatiotemporal spectrum of

473: $\at_k$, obtained from the simulation of the $\beta$-FPU chain for

474: $N=256$, $\beta=0.5$, and $E=100$. In order to measure the value of

475: $\eta$ from the spatiotemporal spectrum, we use the following

476: procedure. For the fixed wave number $k$, the corresponding

477: renormalization factor $\eta(k)$ is determined by the location of

478: the center of the frequency spectrum $|\af_k(\w)|^2$, i.e., \BEA

479: \eta(k)=\frac{\w_{c}(k)}{\w_k},~~\mbox{with}~~

480: \w_{c}(k)=\frac{{\int}

481: \w|\af_k(\w)|^2~d\w}{\int|\af_k(\w)|^2~d\w}.\nonumber \EEA The

482: renormalization factor $\eta(k)$ of each wave mode $k$ is shown in

483: Fig.~\ref{Fig_et}~(inset). The numerical approximation $\bar{\eta}$

484: to the value of $\eta$ is obtained by averaging all $\eta(k)$, i.e.,

485: \BEA \bar{\eta}=\frac{1}{N-1}\sum_{k=1}^{N-1}\eta(k)\nonumber. \EEA

486: The renormalization factor for the case shown in

487: Fig.~\ref{Fig_awk_num} is measured to be $\bar{\eta}\approx1.1824$.

488: It can be clearly seen in Fig.~\ref{Fig_et}~(inset) that $\eta(k)$

489: is nearly independent of $k$ and its variations around $\bar{\eta}$

490: are less than $0.3\%$. We also compare the renormalization factor

491: $\eta$ obtained from Eq.~(\ref{eta}) (solid line in

492: Fig.~\ref{Fig_et}~(inset)) with its numerically computed

493: approximation $\bar{\eta}$ (dashed line in

494: Fig.~\ref{Fig_et}~(inset)). Equation~(\ref{eta}) gives the value

495: $\eta\approx 1.1812$ and the difference between $\eta$ and

496: $\bar{\eta}$ is less then $0.1\%$, which can be attributed to the

497: statistical errors in the numerical measurement.

498: \begin{figure}

499: \includegraphics[width=3in, height=2.5in]{etak_in}

500: \caption{The renormalization factor as a function of the

501: nonlinearity strength $\beta$. The analytical prediction

502: [Eq.~(\ref{eta})] is depicted with a solid line and the numerical

503: measurement is shown with circles. The chain was modeled for

504: $N=256$, and $E=100$. Inset: Independence of $k$ of the

505: renormalization factor $\eta(k)$. The circles correspond to

506: $\eta(k)$  obtained from the spatiotemporal spectrum shown in

507: Fig.~\ref{Fig_awk_num} [only even values of $k$ are shown for

508: clarity of presentation]. The dashed line corresponds to the mean

509: value $\bar{\eta}$. For $\beta=0.5$, the mean value of the

510: renormalization factor is found to be $\bar{\eta}\approx1.1824$. The

511: variations of $\eta_k$ around $\bar{\eta}$ are less then $0.3\%$.

512: [Note the scale of the ordinate.] The solid line corresponds to the

513: renormalization factor $\eta$ obtained from Eq.~(\ref{eta}). For the

514: given parameters $\eta\approx 1.1812$. } \label{Fig_et}

515: \end{figure}

516: In Fig.~\ref{Fig_et}, we plot the value of $\eta$ as a function of

517: $\beta$ for the system with $N=256$ particles and the total energy

518: $E=100$. The solid curve was obtained using Eq.~(\ref{eta}) while

519: the circles correspond to the value of $\eta$ determined via the

520: numerical spectrum $|\af_k(\w)|^2$ as discussed above. It can be

521: observed that there is excellent agreement between the theoretic

522: prediction and numerically measured values for a wide range of the

523: nonlinearity strength $\beta$.

524: 

525: In the following Sections, we will discuss how the renormalization

526: of the linear dispersion of the $\beta$-FPU chain in thermal

527: equilibrium can be explained from the wave resonance point of view.

528: In order to give a wave description of the $\beta$-FPU chain, we

529: rewrite the Hamiltonian~(\ref{H_FPU}) in terms of the renormalized

530: variables $\at_k$~[Eq.~(\ref{at})] with $\wt_k=\eta\w_k$, \BEA

531: H&=&\sum_{k=1}^{N-1}\frac{\w_k}{2}\left(\eta+\frac{1}{\eta}\right)|\at_k|^2\nonumber\\

532: &+&\frac{\w_k}{4}\left(\eta-\frac{1}{\eta}\right)(\at_k^*\at_{N-k}^*+\at_k\at_{N-k})\nonumber\\

533: &+&\sum_{k,l,m,s=1}^{N-1}T^{kl}_{ms}\Big[\D_{ms}^{kl}\at_k\at_l\at_m^*\at_s^*\nonumber\\

534: &+&\left(\frac{2}{3}\D^{klm}_{s}\at_k\at_l\at_m\at_s^*+c.c.\right)\nonumber\\

535: &+&\left(\frac{1}{6}\D^{klms}_{0}\at_k\at_l\at_m\at_s+c.c.\right)\Big],\label{H_a}

536: \EEA where c.c. stands for complex conjugate, and \BEA

537: T^{kl}_{ms}=\frac{3\beta}{8N\eta^2}\sqrt{\w_k\w_l\w_m\w_s}\label{T_inter}

538: \EEA is the interaction tensor coefficient. Note that, due to the

539: discrete nature of the system of finite size, the wave space is

540: periodic and, therefore, the ``momentum'' conservation is guaranteed

541: by the following ``periodic'' Kronecker delta functions \BEA

542: \D^{kl}_{ms}&\equiv&\delta_{ms}^{kl}-\delta^{klN}_{ms}-\delta^{kl}_{msN},\label{D22}\\

543: \D^{klm}_{s}&\equiv&\delta^{klm}_{s}-\delta^{klm}_{sN}+\delta^{klm}_{sNN},\label{D31}\\

544: \D^{klms}_{0}&\equiv&\delta^{klms}_{NN}-\delta^{klms}_{N}-\delta^{klms}_{NNN}.\label{D40}

545: \EEA Here, the Kronecker $\delta$-function is equal to 1, if the sum

546: of all superscripts is equal to the sum of all subscripts, and 0,

547: otherwise.

548: \section{Dispersion relation and resonances}

549: \label{sect_Dispersion} In order to address how the renormalized

550: dispersion arises from wave interactions, we study the resonance

551: structure of our nonlinear waves. Since the system~(\ref{H_a}) is a

552: Hamiltonian system with four-wave interactions, we will discuss the

553: properties of the resonance manifold associated with the $\beta$-FPU

554: system described by Eq.~(\ref{H_a}) as a first step towards the

555: understanding of its long time statistical behavior. We comment that

556: the resonance structure is one of the main objects of investigation

557: in wave turbulence

558: theory~\cite{ZLF,Newell,Lvov,Cai,MMT,Rink1,Rink2}. The theory of

559: wave turbulence focuses on the specific type of interactions, namely

560: resonant interactions, which dominate long time statistical

561: properties of the system. On the other hand, the non-resonant

562: interactions are usually shown to have a total vanishing average

563: contribution to a long time dynamics.

564: 

565: In analogy with quantum mechanics, where $a^+$ and $a$ are creation

566: and annihilation operators, we can view $\at_k^*$ as the

567: \textit{outgoing} wave with frequency $\wt_k$ and $\at_k$ as the

568: \textit{incoming} wave with frequency $\wt_k$. Then, the nonlinear

569: term $\at_k^*\at_l^*\at_m\at_s\Delta^{kl}_{ms}$ in

570: system~(\ref{H_a}) can be interpreted as the interaction process of

571: the type $(2\rightarrow 2)$, namely, two outgoing waves with wave

572: numbers $k$ and $l$ are ``created'' as a result of interaction of

573: the two incoming waves with wave numbers $m$ and $s$. Similarly,

574: $\at_k^*\at_l\at_m\at_s\Delta_{k}^{lms}$ in system~(\ref{H_a})

575: describes the interaction process of the type $(3\rightarrow 1)$,

576: that is, one outgoing wave with wave number $k$ is ``created'' as a

577: result of interaction of the three incoming waves with wave numbers

578: $l$, $m$, and $s$, respectively. Finally,

579: $\at_k\at_l\at_m\at_s\Delta_{0}^{klms}$ describes the interaction

580: process of the type $(4\rightarrow 0)$, i.e., all four incoming

581: waves interact and annihilate themselves. Furthermore, the complex

582: conjugate terms $\at_k\at_l^*\at_m^*\at_s^*\Delta_{lms}^{k}$ and

583: $\at_k^*\at_l^*\at_m^*\at_s^*\Delta^{0}_{klms}$ describe the

584: interaction processes of the type $(1\rightarrow 3)$ and

585: $(0\rightarrow 4)$, respectively.

586: 

587: Instead of the processes with the ``momentum'' conservation given

588: via the usual $\delta^{kl}_{ms}$, $\delta^{klm}_s$, or

589: $\delta^{klms}_0$ functions for an infinite discrete system, the

590: resonant processes of the $\beta$-FPU chain of a \textit{finite}

591: size are constrained to the manifold given by $\D^{kl}_{ms}$,

592: $\D^{klm}_s$, or $\D^{klms}_0$, respectively. Next, we describe

593: these resonant manifolds in detail. As will be pointed out in

594: Section~\ref{sect_Width}, there is a consequence of this

595: \textit{finite} size effect to the properties of the renormalized

596: waves.

597: 

598: The resonance manifold that corresponds to the $(2\rightarrow 2)$

599: resonant processes in the discrete periodic system, therefore, is

600: described by \BEA \BC

601: k+l\eqN m+s,\\

602: \wt_k+\wt_l=\wt_m+\wt_s, \EC\label{2to2} \EEA where we have

603: introduced the notation $g\eqN h$, which means that $g=h$, $g=h+N$,

604: or $g=h-N$ for any $g$ and $h$. The first equation in

605: system~(\ref{2to2}) is the ``momentum'' conservation condition in

606: the periodic wave number space. This ``momentum'' conservation comes

607: from $|\D^{kl}_{ms}|=1$. (Note that $|\D^{kl}_{ms}|$ can assume only

608: the value of $1$ or $0$.) Similarly, from $|\D^{klm}_{s}|=1$ and

609: $|\D^{klms}_{0}|=1$, the resonance manifolds corresponding to the

610: resonant processes of types $(3\rightarrow 1)$ and $(4\rightarrow

611: 0)$ are given by \BEA \BC

612: k+l+m\eqN s,\\

613: \wt_k+\wt_l+\wt_m=\wt_s, \EC\label{3to1} \EEA and \BEA \BC

614: k+l+m+s\eqN 0,\\

615: \wt_k+\wt_l+\wt_m+\wt_s=0, \EC\label{4to0} \EEA respectively. For

616: the processes of type $(3\rightarrow 1)$, the notation $g\eqN h$

617: means that $g=h$, $g=h+N$, or $g=h+2N$. For the $(4\rightarrow 0)$

618: processes, $g\eqN h$ means that $g=h+N$, $g=h+2N$, or $g=h+3N$.

619: 

620: To solve system (\ref{2to2}), we rewrite it in a continuous form

621: with $x={k}/{N}$, $y={l}/{N}$, $z={m}/{N}$, $v={s}/{N}$, which are

622: real numbers in the interval $(0,1)$. By recalling that

623: $\wt_k=2\eta\sin({\pi k}/{N})$, we have \BEA \BC

624: x+y\eq1 z+v,\\

625: \sin(\pi x)+\sin(\pi y)=\sin(\pi z)+\sin(\pi v). \EC\label{2to2cont}

626: \EEA Thus, any rational quartet that satisfies Eq.~(\ref{2to2cont})

627: yields a solution for Eq.~(\ref{2to2}). There are two distinct types

628: of the solutions of Eq.~(\ref{2to2cont}). The first one corresponds

629: to the case \BEA x+y=z+v,\nonumber \EEA whose only solution is given

630: by \BEA \BC

631: x=z,\\

632: y=v, \EC \mbox{or}~~ \BC

633: x=v,\\

634: y=z,\label{trivial} \EC \EEA i.e., these are \textit{trivial}

635: resonances, as we mentioned above. The second type of the resonance

636: manifold of the $(2\rightarrow 2)$-type interaction processes

637: corresponds to \BEA x+y=z+v\pm 1,\nonumber \EEA the solution of

638: which can be described by the following two branches \BEA

639: z_1&=&\frac{x+y}{2}+\frac{1}{\pi}\arcsin(A)+2j,\label{folded1}\\

640: z_2&=&\frac{x+y}{2}-1-\frac{1}{\pi}\arcsin(A)+2j,\label{folded2}

641: \EEA where

642: $A\equiv\tan\left(\pi({x+y})/{2}\right)\cos\left(\pi({x-y})/{2}\right)$

643: and $j$ is an integer. The second type of resonances arises from the

644: discreteness of our model of a finite length, leading to

645: \textit{non-trivial} resonances. For our linear dispersion here,

646: non-trivial resonances are only those resonances that involve wave

647: numbers crossing the first Brillouin zone. As mentioned above, in

648: the setting of the phonon physics, these non-trivial resonant

649: processes are also known as the umklapp scattering processes. In

650: Fig.~\ref{Fig_resonances}, we plot the solution of

651: Eq.~(\ref{2to2cont}) for $x={k}/{N}$ with the wave number $k=90$ for

652: the system with $N=256$ particles (the values of $k$ and $N$ are

653: chosen merely for the purpose of illustration). We stress that

654: \textit{all} the solutions of the system~(\ref{2to2cont}) are given

655: by the Eqs.~(\ref{trivial}),~(\ref{folded1}), and~(\ref{folded2}),

656: and that the non-trivial solutions arise only as a consequence of

657: discreteness of the finite chain. The curves in

658: Fig.~\ref{Fig_resonances} represent the loci of $(z,y)$,

659: parametrized by the fourth wave number $v$, i.e., $x$, $y$, $z$, and

660: $v$ form a resonant quartet, where $z={m}/{N}$, and $y={l}/{N}$.

661: Note that the fourth wave number $v$ is specified by the

662: ``momentum'' conservation, i.e., the first equation in

663: Eq.~(\ref{2to2cont}). The two straight lines in

664: Fig.~\ref{Fig_resonances} correspond to the trivial solutions, as

665: given by Eq.~(\ref{trivial}). The two curves (dotted and dashed)

666: depict the non-trivial resonances. Note that the dotted part of

667: non-trivial resonance curves corresponds to the

668: branch~(\ref{folded1}), and the dashed part corresponds to the

669: branch~(\ref{folded2}), respectively. An immediate question arises:

670: how do these resonant structures manifest themselves in the FPU

671: dynamics in the thermal equilibrium? By examining the

672: Hamiltonian~(\ref{H_a}), we notice that the resonance will control

673: the contribution of terms like

674: $\at_k^*\at_l^*\at_m\at_s\D^{kl}_{ms}$ in the long time limit.

675: Therefore, we address the effect of resonance by computing long time

676: average, i.e., $\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}$, and

677: comparing this average (Fig.~\ref{Fig_resonances_num}) with

678: Fig.~\ref{Fig_resonances}.

679: \begin{figure}

680: \includegraphics[width=2.5in, height=2.5in]{resonances}

681: \caption{The solutions of Eq.~(\ref{2to2cont}). The solid straight

682: lines correspond to the trivial resonances [solutions of

683: Eq.~(\ref{trivial})]. The solutions are shown for fixed $x=k/N$,

684: $k=90$, $N=256$ as the fourth wave number $v$ scans from ${1}/{N}$

685: to $(N-1)/{N}$ in the resonant quartet Eq.~(\ref{2to2cont}). The

686: non-trivial resonances are described by the dotted or dashed curves.

687: The dotted branch of the curves corresponds to the non-trivial

688: resonances described by Eq.~(\ref{folded1}) and and the dashed

689: branch corresponds to the non-trivial resonances described by

690: Eq.~(\ref{folded2}).} \label{Fig_resonances}

691: \end{figure}

692: \begin{figure}

693: \includegraphics[width=3in, height=2.5in]{numerical_resonance}

694: \caption{The long time average

695: $|\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}|$ of the $\beta$-FPU

696: system in thermal equilibrium. The parameters for the FPU chain are

697: $N=256$, $\beta=0.5$, and $E=100$.

698: $\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}$ was computed for fixed

699: $k=90$. The darker grayscale corresponds to the larger value of

700: $\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}$. The exact solutions of

701: Eq.~(\ref{2to2cont}), which are shown in Fig.~\ref{Fig_resonances},

702: coincide with the locations of the peaks of

703: $|\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}|$. Therefore, the

704: darker areas represent the near-resonance structure of the finite

705: $\beta$-FPU chain. (The two white lines show the locations, where

706: $s=0$ and, therefore, $\at_k^*\at_l^*\at_m\at_s\D^{kl}_{ms}=0$.)

707: [$\max\{2,\ln(|\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}|)\}$ with

708: the corresponding grayscale is plotted for a clean presentation].}

709: \label{Fig_resonances_num}

710: \end{figure}

711: To obtain Fig.~\ref{Fig_resonances_num}, the $\beta$-FPU system was

712: simulated with the following parameters: $N=256$, $\beta=0.5$,

713: $E=100$, and the averaging time window $\tau=400\tilde{t}_1$, where

714: $\tilde{t}_1$ is the longest linear period, i.e.,

715: $\tilde{t}_1={2\pi}/{\wt_1}$. In Fig.~\ref{Fig_resonances_num}, mode

716: $k$ was fixed with $k=90$ and the mode $s$, a function of $k$, $l$,

717: and $m$, is obtained from the constraint $k+l\eqN m+s$, i.e.,

718: $|\D^{kl}_{ms}|=1$. Note that we do not impose here the condition

719: $\wt_k+\wt_l=\wt_m+\wt_s$, therefore,

720: $|\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}|$ is a function of $l$

721: and $m$. By comparing Figs.~\ref{Fig_resonances} and

722: \ref{Fig_resonances_num}, it can be observed that the locations of

723: the peaks of the long time average

724: $|\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}|$ coincide with the

725: loci of the $(2\rightarrow 2)$-type resonances. This observation

726: demonstrates that, indeed, there are nontrivial $(2\rightarrow

727: 2)$-type resonances in the finite $\beta$-FPU chain in thermal

728: equilibrium. Furthermore, it can be observed in

729: Fig.~\ref{Fig_resonances_num} that, in addition to the fact that the

730: resonances manifest themselves as the locations of the peaks of

731: $|\la\at_k^*\at_l^*\at_m\at_s\ra\D^{kl}_{ms}|$, the structure of

732: \textit{near-resonances} is reflected in the finite width of the

733: peaks around the loci of the exact resonances. Note that, due to the

734: discrete nature of the finite $\beta$-FPU system, only those

735: solutions $x$, $y$, $z$, and $v$ of Eq.~(\ref{2to2cont}), for which

736: $Nx$, $Ny$, $Nz$, and $Nv$ are integers, yield solutions $k$, $l$,

737: $m$, and $s$ for Eq.~(\ref{2to2}). In general, the rigorous

738: treatment of the exact integer solutions of Eq.~(\ref{2to2}) is not

739: straightforward. For example, for $N=256$, we have the following two

740: exact quartets $\vec{k}=\{k,l,m,s\}$:

741: $\vec{k}=\{k,N/2-k,N/2+k,N-k\}$, $\vec{k}=\{k,N/2-k,N-k,N/2+k\}$ for

742: $k<N/2$, and $\vec{k}=\{k,3N/2-k,k-N/2,N-k\}$,

743: $\vec{k}=\{k,3N/2-k,N-k,k-N/2\}$ for $k>N/2$. We have verified

744: numerically that for $N=256$ there are no other exact integer

745: solutions of Eq.~(\ref{2to2}). In the analysis of the resonance

746: width in Section~\ref{sect_Width}, we will use the fact that the

747: number of exact non-trivial resonances [Eq.~(\ref{2to2})] is

748: significantly smaller than the total number of modes.

749: 

750: The broadening of the resonance peaks in

751: Fig.~\ref{Fig_resonances_num} suggests that, to capture the

752: near-resonances for characterizing long time statistical behavior of

753: the $\beta$-FPU system in thermal equilibrium, instead of

754: Eq.~(\ref{2to2}), one  needs to consider the following effective

755: system \BEA \BC

756: k+l\eqN m+s,\\

757: |\wt_k+\wt_l-\wt_m-\wt_s|<\Delta\w, \EC\label{2to2near} \EEA where

758: $0<\Delta\w\ll\wt_k$ for any $k$, and $\Delta\w$ characterizes the

759: resonance width, which results from the near-resonace structure.

760: Clearly, $\D\w$ is related to the broadening of the spectral peak of

761: each wave $\at_{\a}(t)$ with $\a=k,l,m$, or $s$ in the quartet, and

762: this broadening effect will be studied in detail in

763: Section~\ref{sect_Width}. Note that the structure of near-resonances

764: is a common characteristic of many periodic discrete nonlinear wave

765: systems~\cite{Lvov2,Colm, Pushkarev}.

766: 

767: Further, it is easy to show that the dispersion relation of the

768: $\beta$-FPU chain does not allow for the occurrence of

769: $(3\rightarrow 1)$-type resonances, i.e., there are no solutions for

770: Eq.~(\ref{3to1}), and, therefore, all the nonlinear terms

771: $\at_k^*\at_l\at_m\at_s\D^{k}_{lms}$ are non-resonant and their long

772: time average $\la\at_k^*\at_l\at_m\at_s\ra\D^{k}_{lms}$ vanishes. As

773: for the resonances of type $(4\rightarrow 0)$, since the dispersion

774: relation is non-negative, one can immediately conclude that the

775: solution of the system (\ref{4to0}) consists only of zero modes.

776: Therefore, the processes of type $(4\rightarrow 0)$ are also

777: non-resonant, giving rise to

778: $\la\at_k\at_l\at_m\at_s\ra\D_{0}^{klms}=0$. In this article, we

779: will neglect the higher order effects of the near-resonances of the

780: types $(3\rightarrow 1)$ and $(4\rightarrow 0)$.

781: 

782: In the following sections, we will study the effects of the resonant

783: terms of type $(2\rightarrow 2)$, namely, the linear dispersion

784: renormalization and the broadening of the frequency peaks of

785: $\at_k(t)$. It turns out that, the former is related to the trivial

786: resonance of type $(2\rightarrow 2)$ and the latter is related to

787: the near-resonances, as will be seen below.

788: \section{Self-consistency approach to frequency renormalization}

789: \label{sect_Renormalization} We now turn to the discussion of how

790: the trivial resonances give rise to the dispersion renormalization.

791: This question was examined in~\cite{our_prl} before. There, it was

792: shown that the renormalization of the linear dispersion of the

793: $\beta$-FPU chain arises due to the collective effect of the

794: nonlinearity. In particular, the trivial resonant interactions of

795: type $(2\rightarrow 2)$, i.e., the solutions of Eq.~(\ref{trivial}),

796: enhance the linear dispersion (the renormalized dispersion relation

797: takes the form $\wt_k=\eta\w_k$ with $\eta>1$), and effectively

798: weaken the nonlinear interactions. Here, we further address this

799: issue and present a self-consistency argument to arrive at an

800: approximation for the renormalization factor $\eta$. As it was

801: mentioned above, the contribution of the non-resonant terms have a

802: vanishing long time effect to the statistical properties of the

803: system, therefore, in our self-consistent approach, we ignore these

804: non-resonant terms. By removing the non-resonant terms and using the

805: canonical transformation \BEA \at_k=\frac{P_k-\i\eta_{sc}\w_k

806: Q_k}{\sqrt{2\eta_{sc}\w_k}},\label{a_sc} \EEA where $\eta_{sc}$ is a

807: factor to be determined, we arrive at a simplified effective

808: Hamiltonian from Eq.~(\ref{H_a}) for the finite $\beta$-FPU system

809: \BEA

810: H_{\rm{eff}}&=&\sum_{k=1}^{N-1}\frac{\w_k}{2}\left(\eta_{sc}+\frac{1}{\eta_{sc}}\right)|\at_k|^2\label{H_a_res}\\

811: &+&\sum_{k,l,m,s=1}^{N-1}T^{kl}_{ms}\Delta^{kl}_{ms}\at_k^*\at_l^*\at_m\at_s.\nonumber

812: \EEA The ``off-diagonal'' quadratic terms $\at_k\at_{N-k}$ from

813: Eq.~(\ref{H_a}) are not present in Eq.~(\ref{H_a_res}), since

814: $\at_k$ are chosen so that $\la\at_k\at_{N-k}\ra=0$ (see

815: Section~\ref{sect_Hformalism}). The contribution of the

816: \textit{trivial} resonances in $H_{\rm{eff}}$ is \BEA

817: H_4^{\rm{tr}}=4\sum_{k,l=1}^{N-1}T^{kl}_{kl}|\at_l|^2|\at_k|^2,\label{H4_diag}

818: \EEA which can be ``linearized'' in the sense that averaging the

819: coefficient in front of $|\at_k|^2$ in $H_4^{\rm{tr}}$ gives rise to

820: a quadratic form \BEA

821: H_2^{\rm{tr}}\equiv\sum_{k=1}^{N-1}\left(4\sum_{l=1}^{N-1}T^{kl}_{kl}\la|\at_l|^2\ra\right)|\at_k|^2.\nonumber

822: \EEA Note that the subscript $2$ in $H_2^{\rm{tr}}$ emphasizes the

823: fact that $H_2^{\rm{tr}}$ now can be viewed as a Hamiltonian for the

824: free waves with the familiar effective linear dispersion

825: $\Omega_k=4\sum_{l=1}^{N-1}T^{kl}_{kl}\la|\at_l|^2\ra$~\cite{our_prl,ZLF}.

826: This linearization is essentially a mean-field approximation, since

827: the long-time average of trivial resonances in Eq.~(\ref{H4_diag})

828: is approximated by the interaction of waves $\at_k$ with background

829: waves $\la |\at_l|\ra$. The self consistency condition, which

830: determines $\eta_{sc}$, can be imposed as follows: the quadratic

831: part of the Hamiltonian (\ref{H_a_res}), combined with the

832: ``linearized'' quadratic part, $H_2^{\rm{tr}}$, of the quartic

833: $H_4^{\rm{tr}}$, should be equal to an effective quadratic

834: Hamiltonian $\tilde{H}_2=\sum_{k=1}^{N-1}\wt_k|\at_k|^2$ for the

835: renormalized waves, i.e., \BEA

836: & &\sum_{k=1}^{N-1}\frac{\w_k}{2}\left(\eta_{sc}+\frac{1}{\eta_{sc}}\right)|\at_k|^2\nonumber\\

837: &

838: &+\sum_{k=1}^{N-1}\left(4\sum_{l=1}^{N-1}T^{kl}_{kl}\la|\at_l|^2\ra\right)|\at_k|^2=\sum_{k=1}^{N-1}\wt_k|\at_k|^2,\nonumber\\\label{self}

839: \EEA where $\wt_k$ is the renormalized linear dispersion, which is

840: used in the definition of our renormalized wave, Eq.~(\ref{a_sc}),

841: and $\wt_k=\eta_{sc}\w_k$. Equating the coefficients of

842: $\w_k|\at_k|^2$ on both sides for every wave number $k$ yields \BEA

843: \frac{1}{2}\left(\eta_{sc}+\frac{1}{\eta_{sc}}\right)+4\sum_{l=1}^{N-1}\frac{3\beta}{8N\eta_{sc}^2}\w_l\langle

844: |\at_l|^2\rangle=\eta_{sc},\nonumber \EEA where use is made of

845: Eq.~(\ref{T_inter}). After algebraic simplification, we have the

846: following equation for $\eta_{sc}$ \BEA

847: \eta_{sc}^3-\eta_{sc}=\frac{3\beta}{N}\sum_{l=1}^{N-1}\w_l\langle|\at_l|^2\rangle.\label{eq_eta}

848: \EEA Using the property~(\ref{asat}) of the renormalized normal

849: variables $\at_k$, we find the following dependence of

850: $\la|\at_k|^2\ra$ on $\eta_{sc}$, \BEA

851: \la|\at_l|^2\ra=\frac{1}{2\eta_{sc}\w_l}\left(\la|P_l|^2\ra+\eta_{sc}^2\w_l^2\la|Q_l|^2\ra\right).\label{rhs}

852: \EEA Combining Eqs.~(\ref{eq_eta}) and (\ref{rhs}) leads to \BEA

853: \eta_{sc}^4-A\eta_{sc}^2-B=0,\label{eq_eta2} \EEA where \BEA

854: A&=&1+\frac{3\beta}{2N}\sum_{l=1}^{N-1}\w_l^2\langle|Q_l|^2\rangle=1+\frac{3\beta}{N}\la U\ra,\nonumber\\

855: B&=&\frac{3\beta}{2N}\sum_{l=1}^{N-1}\langle|P_l|^2\rangle=\frac{3\beta}{N}\la

856: K\ra.\nonumber \EEA The only physically relevant solution of

857: Eq.~(\ref{eq_eta2}) is \BEA

858: \eta_{sc}=\sqrt{\frac{A+\sqrt{A^2+4B}}{2}}.\label{new_eta} \EEA The

859: constants $A$ and $B$ can be easily derived using the Gibbs measure.

860: 

861: Next, we compare the renormalization factor $\eta$ [Eq.~(\ref{eta})]

862: with its approximation $\eta_{sc}$ [Eq.~(\ref{new_eta})] from the

863: self-consistency argument. In Appendix~\ref{app_FPU}, we study in

864: detail the behavior of both $\eta$ and $\eta_{sc}$ in the two

865: limiting cases, i.e., when nonlinearity is small ($\beta\rightarrow

866: 0$ with fixed total energy $E$), and when nonlinearity is large

867: ($\beta\rightarrow\infty$ with fixed total energy $E$). As is shown

868: in Appendix~\ref{app_FPU}, for the case of small nonlinearity, both

869: $\eta$ and $\eta_{sc}$ have the same asymptotic behavior in the

870: first order of the small parameter $\beta$, \BEA

871: \eta&=&1+\frac{3E}{2N}\beta+O(\beta^2),\label{small_eta}\\

872: \eta_{sc}&=&1+\frac{3E}{2N}\beta+O(\beta^2).\nonumber \EEA Moreover,

873: in the case of strong nonlinearity $\beta\rightarrow\infty$, both

874: $\eta$ and $\eta_{sc}$ scale as $\beta^\frac{1}{4}$, i.e., \BEA

875: \eta\sim\eta_{sc}\sim\beta^{\frac{1}{4}}\label{large_eta} \EEA (see

876: Appendix~\ref{app_FPU} for details). Note that, in~\cite{our_prl},

877: we numerically obtained the scaling $\eta\sim\beta^{0.2}$, which

878: differs from the exact analytical result~(\ref{large_eta}) due to

879: statistical errors in the numerical estimate of the power.

880: \begin{figure}

881: \includegraphics[width=3in, height=3in]{eta_in}

882: \caption{The renormalization factor as a function of the

883: nonlinearity strength $\beta$ for small values of $\beta$. The

884: renormalization factor $\eta$ [Eq.~(\ref{eta})] is shown with the

885: solid line. The approximation $\eta_{sc}$ [Eq.~(\ref{new_eta})](via

886: the self-consistency argument) is depicted with diamonds connected

887: with the dashed line. The small-$\beta$ limit

888: [Eq.~(\ref{small_eta})] is shown with the solid circles connected

889: with the dotted line. Note that, abscissa is of logarithmic scale.

890: Inset: The renormalization factor as a function of the nonlinearity

891: strength $\beta$ for large values of $\beta$. The renormalization

892: factor $\eta$ [Eq.~(\ref{eta})] is shown with the solid line.

893: $\eta_{sc}$ [Eq.(\ref{new_eta})] is depicted with diamonds connected

894: with the dashed line. The large-$\beta$ scaling

895: [Eq.~(\ref{large_eta})] is shown with the dashed-dotted line. Note

896: that, the plot is of log-log scale with base 10.}

897: \label{Fig_eta_small}

898: \end{figure}

899: In Fig.~\ref{Fig_eta_small}, we plot the renormalization factor

900: $\eta$ and its approximation $\eta_{sc}$ for the case of small

901: nonlinearity $\beta$ for the system with $N=256$ particles and total

902: energy $E=100$. The solid line shows $\eta$ computed via

903: Eq.~(\ref{eta}), the diamonds with the dashed line represent the

904: approximation via Eq.~(\ref{new_eta}), and the solid circles with

905: the dotted line correspond to the small-$\beta$

906: limit~(\ref{small_eta}). In Fig.~\ref{Fig_eta_small}~(inset), we

907: plot the renormalization factor $\eta$ and its approximation

908: $\eta_{sc}$ for the case of large nonlinearity $\beta$ for the

909: system with $N=256$ particles and total energy $E=100$. The solid

910: line shows $\eta$ computed via Eq.~(\ref{eta}), the diamonds with

911: the dashed line represent the approximation via Eq.~(\ref{new_eta}),

912: and the dashed-dotted line correspond to the large-$\beta$

913: scaling~(\ref{large_eta}). Figure~\ref{Fig_eta_small} shows good

914: agreement between the renormalization factor $\eta$ and its

915: approximation $\eta_{sc}$ from the self consistency argument for a

916: wide range of nonlinearity, from $\beta\sim 10^{-3}$ to $\beta\sim

917: 10^4$. This agreement demonstrates, that (i) the effect of the

918: linear dispersion renormalization, indeed, arises mainly from the

919: trivial four-wave resonant interactions, and (ii) our

920: self-consistency, mean-field argument is not restricted to small

921: nonlinearity.

922: \section{Resonance width}

923: \label{sect_Width} Finally, we address the question of how coherent

924: these renormalized waves are, i.e., we study how the nonlinear

925: interactions of waves in thermal equilibrium broaden the

926: renormalized dispersion. We will obtain an analytical formula for

927: the spatiotemporal spectrum $|\af_k(\w)|^2$ for the $\beta$-FPU

928: chain and compare the numerically measured width of the frequency

929: peaks with the predicted width.

930: 

931: In the Hamiltonian~(\ref{H_a_res}), the nonlinear terms

932: corresponding to the trivial resonances have been absorbed into the

933: quadratic part via the effective renormalized dispersion $\wt_k$.

934: Therefore, the new effective Hamiltonian is \BEA

935: \bar{H}=\sum_{k=1}^{N-1}\wt_k|\at_k|^2+\sum_{k,l,m,s=1}^{N-1}\Tt^{kl}_{ms}\Delta^{kl}_{ms}\at_k^*\at_l^*\at_m\at_s,\label{Hef}

936: \EEA where \BEA \BC

937: \Tt^{kl}_{ms}=T^{kl}_{ms}=\displaystyle{\frac{3\beta}{8N\eta^2}}\sqrt{\w_k\w_l\w_m\w_s},~k\neq m,~\mbox{and}~k\neq s\\

938: \Tt^{kl}_{ms}=0,~~\mbox{otherwise.} \EC\label{Tt} \EEA The new

939: interaction coefficient $\Tt^{kl}_{ms}$ ensures that the terms that

940: correspond to the interactions with trivial resonances are not

941: doubly counted in the Hamiltonian~(\ref{Hef}) and are removed from

942: the quartic interaction. This new interactions in the quartic terms

943: include the exact non-trivial resonant and non-trivial near-resonant

944: as well as non-resonant interactions of the $(2\rightarrow 2)$-type.

945: 

946: We change the variables to the interaction picture by defining the

947: corresponding variables $b_k$ via \BEA

948: b_k=\at_ke^{\i\wt_kt},\nonumber \EEA so that, the dynamics governed

949: by the Hamiltonian (\ref{Hef}) takes the familiar form \BEA \i\dot

950: b_k=2\sum_{l,m,s=1}^{N-1}\Tt^{kl}_{ms}\Delta^{kl}_{ms}b_l^*b_mb_se^{\i\wt^{kl}_{ms}

951: t},\label{b_dyn} \EEA where

952: $\wt^{kl}_{ms}=\wt_k+\wt_l-\wt_m-\wt_s$~\cite{Lvov}. Without loss of

953: generality, we consider only the case of $k<N/2$. As we have noted

954: before, only for a very small number of quartets does

955: $\wt^{kl}_{ms}$ vanish \textit{exactly}, i.e., $\wt^{kl}_{ms}=0$. We

956: separate the terms on the RHS of Eq.~(\ref{b_dyn}) into two kinds

957: --- the first kind with $\wt^{kl}_{ms}=0$ that corresponds to exact

958: non-trivial resonances, and the second kind that corresponds to

959: non-trivial near-resonances and non-resonances. Since, in the

960: summation, the first kind contains far fewer terms than the second

961: kind, and all the terms are of the same order of magnitude, we will

962: neglect the first kind in our analysis. Therefore, Eq.~(\ref{b_dyn})

963: becomes \BEA \i\dot b_k&=&2{\sum_{l,m,s=1}^{N-1}}^{\prime}

964: \Tt^{kl}_{ms}\Delta^{kl}_{ms}b_l^*b_mb_se^{\i\wt^{kl}_{ms}

965: t},\label{b_dyn2} \EEA where the prime denotes the summation that

966: neglects the exact non-trivial resonances.

967: 

968: The problem of broadening of spectral peaks now becomes the study of

969: the frequency spectrum of the dynamical variables $b_k(t)$ in

970: thermal equilibrium. This is equivalent to study the  two-point

971: correlation in time of $b_k(t)$ \BEA C_k(t)=\la b_k(t)b_k^*(0)\ra,

972: \EEA where the angular brackets denote the thermal average, since,

973: by Wiener-Khinchin theorem, the frequency spectrum \BEA

974: |b(\w)|^2=\mathfrak{F}^{-1}[C(t)](\w),\label{WKh} \EEA where

975: $\mathfrak{F}^{-1}$ is the inverse Fourier transform in time. Under

976: the dynamics (\ref{b_dyn2}), time derivative of the two-point

977: correlation becomes \BEA

978: \dot{C}_k(t)&=&\la \dot{b}_k(t)b_k^*(0)\ra\nonumber \\

979: &=&\la -2\i{\sum_{l,m,s}}^{\prime}\Tt^{kl}_{ms}b_l^*(t)b_m(t)b_s(t)e^{\i\wt^{kl}_{ms}t}\Delta^{kl}_{ms}b_k^*(0)\ra\nonumber\\

980: &=&-2\i{\sum_{l,m,s}}^{\prime}\Tt^{kl}_{ms}e^{\i\wt^{kl}_{ms}t}J^{kl}_{ms}(t)\Delta^{kl}_{ms}\label{cdot},

981: \EEA where \BEA J^{kl}_{ms}(t)\equiv\la

982: b_l^*(t)b_k^*(0)b_m(t)b_s(t)\ra.\nonumber \EEA In order to obtain a

983: closed equation for $C_k(t)$, we need to study the evolution of the

984: fourth order correlator $J^{kl}_{ms}(t)$. We utilize the weak

985: effective nonlinearity in Eq.~(\ref{Hef})~\cite{our_prl} as the

986: small parameter in the following perturbation analysis and obtain a

987: closure for $C_k(t)$, similar to the traditional way of deriving

988: kinetic equation, as in~\cite{ZLF,Benney}. We note that the

989: effective interactions of renormalized waves can be weak, as we have

990: shown in~\cite{our_prl}, even if the $\beta$-FPU chain is in a

991: strongly nonlinear regime. Our perturbation analysis is a multiple

992: time-scale, statistical averaging method. Under the near-Gaussian

993: assumption, which is applicable for the weakly nonlinear wave fields

994: in thermal equilibrium, for the four-point correlator, we obtain

995: \BEA

996: J^{kl}_{ms}(t)\Delta^{kl}_{ms}=C_k(t)C_l(0)(\delta^k_m\delta^l_s+\delta^k_s\delta^l_m).~~\label{4corr}

997: \EEA Combining Eqs.~(\ref{Tt}) and~(\ref{4corr}), we find that the

998: right-hand side of Eq.~(\ref{cdot}) vanishes because \BEA

999: \Tt^{kl}_{ms}J^{kl}_{ms}(t)\Delta^{kl}_{ms}=0. \EEA Therefore, we

1000: need to proceed to the higher order contribution of

1001: $J^{kl}_{ms}(t)$. Taking its time derivative yields \BEA

1002: \dot{J}^{kl}_{ms}(t)\D^{kl}_{ms}&=&\la[\dot{b}_l^*(t)b_m(t)b_s(t)+b_l^*(t)\dot{b}_m(t)b_s(t)\nonumber\\

1003:                 &+&b_l^*(t)b_m(t)\dot{b}_s(t)]b_k^*(0)\ra\D^{kl}_{ms}.\label{Jdot}

1004: \EEA Considering the right-hand side of Eq.~(\ref{Jdot}) term by

1005: term, for the first term, we have \BEA

1006: & &\la \dot{b}_l^*(t)b_m(t)b_s(t)b_k^*(0)\ra\D^{kl}_{ms}\nonumber\\

1007: & &=\Bigg\la \Big[2\i{\sum_{\a,\b,\g}}^{\prime}\Tt^{l\a}_{\b\g}b_{\a}(t)b_{\b}^*(t)b_{\g}^*(t)\nonumber\\

1008: & &\times

1009: e^{-\i\w^{l\a}_{\b\g}}\D^{l\a}_{\b\g}\Big]b_m(t)b_s(t)b_k^*(0)\Bigg\ra\D^{kl}_{ms}.\label{4b}

1010: \EEA We can use the near-Gaussian assumption to split the correlator

1011: of the sixth order in Eq.~(\ref{4b}) into the product of three

1012: correlators of the second order, namely, \BEA

1013: & &\la b_k^*(0)b_m(t)b_s(t)b_{\a}(t)b_{\b}^*(t)b_{\g}^*\ra \D^{kl}_{ms}\nonumber\\

1014: & &=C_k(t)n_m

1015: n_s\delta^k_{\a}(\delta_{\b}^m\delta^s_{\g}+\delta^m_{\g}\delta^s_{\b}),\nonumber

1016: \EEA where we have used that $n_m=C_m(0)$. Then, Eq.~(\ref{4b})

1017: becomes \BEA

1018: & &\la \dot{b}_l^*(t)b_m(t)b_s(t)b_k^*(0)\ra\D^{kl}_{ms}\nonumber\\

1019: & &=4\i\Tt^{lk}_{ms}C_k(t)n_m n_s

1020: e^{-\i\wt^{lk}_{ms}}\D^{kl}_{ms}.\label{4b1} \EEA Similarly, for the

1021: remaining two terms in Eq.~(\ref{Jdot}), we have \BEA

1022: & &\la b_l^*(t)\dot{b}_m(t)b_s(t)b_k^*(0)\ra\D^{kl}_{ms}\nonumber\\

1023: & &=-4\i\Tt^{ms}_{kl}C_k(t)n_l n_s

1024: e^{\i\wt^{ms}_{kl}}\D^{kl}_{ms},\label{4b2} \EEA and \BEA

1025: & &\la b_l^*(t)b_m(t)\dot{b}_s(t)b_k^*(0)\ra\D^{kl}_{ms}\nonumber\\

1026: & &=-4\i\Tt^{ms}_{kl}C_k(t)n_l n_m

1027: e^{\i\wt^{ms}_{kl}}\D^{kl}_{ms},\label{4b3} \EEA respectively.

1028: Combining Eqs.~(\ref{4b1}), (\ref{4b2}), and (\ref{4b3}) with

1029: Eq.~(\ref{Jdot}), we obtain \BEA

1030: \dot{J}^{kl}_{ms}(t)\D^{kl}_{ms}&=&4\i\Tt^{kl}_{ms}C_k(t)e^{-\i\wt^{kl}_{ms}t}\D^{kl}_{ms}\nonumber\\

1031: & &\times(n_m n_s-n_l n_m-n_l n_s).\label{Jdot2} \EEA

1032: Equation~(\ref{Jdot2}) can be solved for $J^{kl}_{ms}(t)$ under the

1033: assumption that the term $e^{-\i\wt^{kl}_{ms}t}$ oscillates much

1034: faster than $C_k(t)$. We numerically verify [Fig.~\ref{Fig_corr}

1035: below] the validity of this assumption of time-scale separation.

1036: Under this approximation, the solution of Eq.~(\ref{Jdot2})  becomes

1037: \BEA

1038: J^{kl}_{ms}(t)\D^{kl}_{ms}&=&4\Tt^{kl}_{ms}C_k(t)\D^{kl}_{ms}\frac{e^{-\i\wt^{kl}_{ms}t}-1}{-\wt^{kl}_{ms}}\nonumber\\

1039: & &\times(n_m n_s-n_l n_m-n_l n_s )\label{J} \EEA Plugging

1040: Eq.~(\ref{J}) into Eq.~(\ref{cdot}), we obtain the following

1041: equation for $C_k(t)$ \BEA

1042: \dot{C}_k(t)&=&8\i C_k(t) {\sum_{l,m,s}}^{\prime} \left(\Tt^{kl}_{ms}\right)^2\D^{kl}_{ms}\frac{1-e^{\i\w^{kl}_{ms}t}}{\w^{kl}_{ms}}\nonumber\\

1043: & &\times(n_m n_s-n_l n_s-n_l n_m).\label{cdot2} \EEA Since in the

1044: thermal equilibrium $n_k$ is known, i.e., $n_k=\la

1045: |b_k(t)|^2\ra=\theta/\wt_k$ [Eq.~(\ref{nt})], Eq.~(\ref{cdot2})

1046: becomes a closed equation for $C_k(t)$. The solution of

1047: Eq.~(\ref{cdot2}) yields the autocorrelation function $C_k(t)$ \BEA

1048: \ln\frac{C_k(t)}{C_k(0)}&=&8{\sum_{l,m,s}}^{\prime} \left(\Tt^{kl}_{ms}\right)^2\frac{e^{\i\w^{kl}_{ms}t}-1-\i\w^{kl}_{ms}t}{(\w^{kl}_{ms})^2}\nonumber\\

1049: & &\times(n_l n_s+n_l n_m-n_m n_s)\D^{kl}_{ms}.\label{C} \EEA Using

1050: this observation, together with Eq.~(\ref{Tt}), finally, we obtain

1051: for the thermalized $\beta$-FPU chain \BEA

1052: \ln\frac{C_k(t)}{C_k(0)}&=&\frac{9\beta^2\theta^2}{8N^2\eta^6}\w_k{\sum_{l,m,s}}^{\prime}(\w_m+\w_s-\w_l)\D^{kl}_{ms}\nonumber\\

1053: &

1054: &\times\frac{e^{\i\w^{kl}_{ms}t}-1-\i\w^{kl}_{ms}t}{(\w^{kl}_{ms})^2}.\label{C}

1055: \EEA Equation~(\ref{C}) gives a direct way of computing the

1056: correlation function of the renormalized waves $\at_k$, which, in

1057: turn, allows us to predict the spatiotemporal spectrum

1058: $|\af_k(\w)|^2$. In Fig.~\ref{Fig_awk_both}(a), we plot the

1059: analytical prediction (via Eq.~(\ref{C})) of the spatiotemporal

1060: spectrum

1061: $|\af_k(\w)|^2\equiv|b_k(\w-\wt_k)|^2=\mathfrak{F}^{-1}[C(t)](\w-\wt_k)$.

1062: By comparing this plot with the one presented in

1063: Fig.~\ref{Fig_awk_both}(b), in which the corresponding numerically

1064: measured spatiotemporal spectrum is shown, it can be seen that the

1065: analytical prediction of the frequency spectrum via Eq.~(\ref{C}) is

1066: in good qualitative agreement with the numerically measured one.

1067: However, to obtain a more detailed comparison of the analytical

1068: prediction with the numerical observation, we show, in

1069: Fig.~\ref{Fig_aw}, the numerical frequency spectra of selected wave

1070: modes with the corresponding analytical predictions. It can be

1071: clearly observed that the agreement is rather good.

1072: \begin{figure}

1073: \includegraphics[width=3in, height=2.5in]{awk_both}

1074: \caption{ (a) Plot of the analytical prediction for the

1075: spatiotemporal spectrum $|\af_k(\w)|^2$ via Eq.~(\ref{C}). (b) Plot

1076: of the numerically measured spatiotemporal spectrum $|\af_k(\w)|^2$.

1077: The parameters in both plots were $N=256$, $\beta=0.125$, $E=100$

1078: and $\eta=1.06$, $\theta=0.401$. $\eta$ and  $\theta$ were computed

1079: analytically via Gibbs measure. The darker gray scale correspond to

1080: larger values of $|\af_k(\w)|^2$ in $\w$-$k$ space.

1081: [$\max\{-8,\ln{|\af_k(\w)|^2}\}$ is plotted for clear

1082: presentation].} \label{Fig_awk_both}

1083: \end{figure}

1084: \begin{figure}

1085: \includegraphics[width=3in, height=2.5in]{aw}

1086: \caption{Temporal frequency spectrum $|\af_k(\w)|^2$ for $k=30$

1087: (left peak) and $k=50$ (right peak). The numerical spectrum is shown

1088: with pluses and the analytical prediction [via Eq.~(\ref{C})] is

1089: shown with solid line. The parameters were $N=256$, $\beta=0.125$,

1090: $E=100$.

1091: %Both analytical and numerical peaks for each $k$ are centered at the same frequencies $\w$, which are equal to the numerically observed dispersion $\wt_k$.

1092: } \label{Fig_aw}

1093: \end{figure}

1094: One of the important characteristics of the frequency spectrum is

1095: the width of the spectrum. We compute the width $W(f)$ of the

1096: spectrum $f(\w)$ by \BEA W(f)=\frac{\int f(\w)~d\w}{\max_\w f(\w)}.

1097: \EEA In Fig.~\ref{Fig_widths}, we compare the width, as a function

1098: of the wave number $k$, of the frequency peaks from the numerical

1099: observation with that obtained from the analytical predictions. We

1100: observe that, for weak nonlinearity ($\beta=0.125$), the analytical

1101: prediction and the numerical observation are in excellent agreement.

1102: In the weakly nonlinear regime, this agreement can be attributed to

1103: the validity of (i) the near-Gaussian assumption, and (ii) the

1104: separation between the linear dispersion time scale and the time

1105: scale of the correlation $C_k(t)$. This separation was used in

1106: deriving the analytical prediction [Eq.~(\ref{C})]. However, when

1107: the nonlinearity becomes larger ($\beta=0.25$ and $\beta=0.5$), the

1108: discrepancy between the numerical measurements and the analytical

1109: prediction increases, as can be seen in Fig.~\ref{Fig_widths}.

1110: Nevertheless, it is important to emphasize that, even for very

1111: strong nonlinearity, our prediction is still qualitatively valid, as

1112: seen in Fig.~\ref{Fig_widths}. In order to find out the effect of

1113: the umklapp scattering due to the finite size of the chain, we also

1114: computed the correlation [Eq.~(\ref{C})] with the ``conventional''

1115: $\delta$-function $\delta^{kl}_{ms}$ (i.e., without taking into

1116: account the umklapp processes) instead of our ``periodic'' delta

1117: function $\D^{kl}_{ms}$. It turns out that the correlation time is

1118: approximately $30\%$ larger if it is computed without umklapp

1119: processes taken into account for the case $N=256$, $\beta=0.5$,

1120: $E=100$. It demonstrates that the influence of the non-trivial

1121: umklapp resonances is important and should be considered when one

1122: describes the dynamics of the finite length chain of particles.

1123: \begin{figure}

1124: \includegraphics[width=3in, height=2.5in]{widths}

1125: \caption{Frequency peak width $W(|\af_k(\w)|^2)$ as a function of

1126: the wave number $k$. The analytical prediction via Eq.~(\ref{C}) is

1127: shown with a dashed line and the numerical observation is plotted

1128: with solid circles. The parameters were $N=256$, $E=100$. The upper

1129: thick lines correspond to $\beta=0.5$, the middle fine lines

1130: correspond to $\beta=0.25$, and the lower solid circle and dashed

1131: line (almost overlap) correspond to $\beta=0.125$.}

1132: \label{Fig_widths}

1133: \end{figure}

1134: Finally, in Fig.~\ref{Fig_corr}, we verify the time scale separation

1135: assumption used in our derivation, i.e., the correlation time of the

1136: wave mode $k$ is sufficiently larger than the corresponding linear

1137: dispersion period $\tt_k=2\pi/\wt_k$. In the case of small

1138: nonlinearity ($\beta=0.125$), the two-point correlation changes over

1139: much slower time scale than the corresponding linear oscillations

1140: --- the correlation time is nearly two orders of magnitude larger than the corresponding linear oscillations for weak nonlinearity $\beta=0.125$, and

1141: nearly one order of magnitude larger than the corresponding linear

1142: oscillations for stronger nonlinearity $\beta=0.25$ and $\beta=0.5$.

1143: This demonstrates that the renormalized waves have long lifetimes,

1144: i.e., they are coherent over time-scales that are much longer than

1145: their oscillation time-scales.

1146: \begin{figure}

1147: \includegraphics[width=3in, height=2.5in]{awk_corr}

1148: \caption{ Ratio, as a function of $k$, of the correlation time

1149: $\tau_k$ of the mode $k$ to the corresponding linear period

1150: $\tt_k=2\pi/\wt_k$. Circles, squares, and diamonds represent the

1151: analytical prediction for $\beta=0.5$, $\beta=0.25$, and

1152: $\beta=0.125$ respectively. Stars, pentagrams, and triangles

1153: correspond to the numerical observation for $\beta=0.5$,

1154: $\beta=0.25$, and $\beta=0.125$ respectively. The parameters were

1155: $N=256$, $E=100$. The ratio is sufficiently large for all wave

1156: numbers $k$ even for relatively large $\beta=0.5$, which validates

1157: the time-scale separation assumption used in deriving Eq.~(\ref{J}).

1158: The comparison also suggests that for smaller $\beta$ the analytical

1159: prediction should be closer to the numerical observation, as is

1160: confirmed in Fig.~\ref{Fig_widths}.} \label{Fig_corr}

1161: \end{figure}

1162: \section{Conclusions}

1163: \label{sect_Conclusions} We have studied the statistical behavior of

1164: the nonlinear periodic lattice with the nearest neighbor

1165: interactions in thermal equilibrium. We have extended the notion of

1166: normal modes to the nonlinear system by showing that regardless of

1167: the strength of nonlinearity, the system in thermal equilibrium can

1168: still be effectively characterized by a complete set of renormalized

1169: waves, in the sense that those renormalized waves possess the

1170: Rayleigh-Jeans distribution and vanishing correlations between

1171: different wave modes. In addition, we have studied the property of

1172: dispersion relation of the renormalized waves. The results we

1173: obtained in Section~\ref{sect_Hformalism} are general and can be

1174: applied to the large class of nonlinear systems with the nearest

1175: neighbor interactions in thermal equilibrium.

1176: 

1177: We have further focused our attention on the particular system with

1178: the nearest neighbor interactions --- the famous FPU chain. We have

1179: confirmed that the general renormalization framework that we

1180: discussed above is consistent with the numerical observations. In

1181: particular, we have shown that the renormalized dispersion of the

1182: thermalized $\beta$-FPU chain is in excellent agreement with the

1183: numerical one for a wide range of the nonlinearity strength. We have

1184: further demonstrated that the renormalized dispersion is a direct

1185: consequence of the trivial resonant interactions of the renormalized

1186: waves. Using a self-consistency argument, we have found an

1187: approximation of the renormalization factor via a mean-field

1188: approximation. In addition, we have used the multiple time-scale,

1189: statistical averaging method to obtain the theoretical prediction of

1190: the spatiotemporal spectrum and demonstrated that the renormalized

1191: waves have long lifetimes. We note that the results obtained here

1192: can be extended to general nonlinear potentials with the nearest

1193: neighbor interactions.

1194: 

1195: The scenario of the wave behavior in the thermal equilibrium we

1196: obtained here may suggest a theoretical framework for the

1197: application of the wave turbulence to $\beta$-FPU in the situation

1198: of \textit{near}-equilibrium.

1199: \begin{acknowledgments}

1200: We thank Sergey Nazarenko and Naoto Yokoyama for discussions. Y.L.

1201: was supported by NSF CAREER DMS 0134955 and D.C. was supported by

1202: NSF DMS 0507901.

1203: \end{acknowledgments}

1204: \appendix

1205: \section{Limiting behaviors of $\eta$ for the thermalized $\beta$-FPU chain}

1206: \label{app_FPU} We change variables $y_j=q_j-q_{j+1}$ in the

1207: Hamiltonian~(\ref{H_FPU}) for $\beta$-FPU to obtain \BEA

1208: H(p,y)=\sum_{j=1}^N\left[\frac{1}{2}p_j^2+\frac{1}{2}y_j^2+\frac{\beta}{4}y_j^4\right].\label{app2_Ham_py}

1209: \EEA Next, we compute the pdf's for the momentum and displacement.

1210: Any $p_j$ is distributed with the Gaussian pdf

1211: $z_p=C_p\exp(-{\T}^{-1}p^2/2)$ and any $y_j$ is distributed with the

1212: pdf $z_y=C_y\exp(-{\T}^{-1}(y^2+\beta y^4/2)/2)$, where $C_p$ and

1213: $C_y$ are the normalizing constants. As we have discussed, the

1214: renormalization factor $\eta$ of the $\beta$-FPU system in thermal

1215: equilibrium is given by Eq.~(\ref{eta}), and its approximation via

1216: the self-consistency argument $\eta_{sc}$ is given by

1217: Eq.~(\ref{new_eta}). Here, we compare the behavior of both formulas

1218: in two limiting cases, i.e., the case of small nonlinearity

1219: $\beta\rightarrow 0$ and the case of strong nonlinearity

1220: $\beta\rightarrow\infty$. We will use the following expressions for

1221: the average density of kinetic, quadratic potential and quartic

1222: potential parts of the total energy of the system \BEA

1223: \frac{\la K\ra}{N}&=&\frac{1}{N}\sum_{j=1}^N\frac{\la p_j^2\ra}{2}=\frac{1}{2}\T,\label{app2_Hp2}\\

1224: \frac{\la U\ra}{N}&=&\frac{1}{N}\sum_{j=1}^N\frac{\la y_j^2\ra}{2}=\frac{1}{2}\la y^2\ra,\label{app2_Hy2}\\

1225: \frac{\la V\ra}{N}&=&\frac{\beta}{4N}\sum_{j=1}^N\la

1226: y_j^4\ra=\frac{\beta}{4}\la y^4\ra.\label{app2_Hy4} \EEA In a

1227: canonical ensemble, the temperature of a system is given by the

1228: temperature of the heat bath. By identifying the average energy

1229: density of the system with $\eb=E/N$ in our simulation (a

1230: microcanonical ensemble), we can determine $\T$ as a function of

1231: $\eb$ and $\beta$ by the following equation \BEA \frac{1}{N}\big(\la

1232: K\ra+\la U\ra+\la V\ra\big)=\eb.\label{app2_eqT} \EEA We start with

1233: the case of small nonlinearity $\beta\rightarrow 0$. Suppose in the

1234: first order of the small parameter $\beta$ the temperature has the

1235: following form \BEA \T(\beta)=\T_0+\beta \T_1\label{app2_Tsm}, \EEA

1236: where $\T_0=O(1)$ and $\T_1=O(1)$. We find the values of $\T_0$ and

1237: $\T_1$ using the constraint~(\ref{app2_eqT}). We use the following

1238: expansions in the small parameter $\beta$ \BEA

1239: & &\int_{-\infty}^{\infty}e^{-\frac{1}{2\T(\beta)}(y^2+\beta\frac{y^4}{2})}~dy\nonumber\\

1240: &

1241: &=\sqrt\frac{\pi}{8}\sqrt{\T_0}\left(4+\Big(\frac{2\T_1}{\T_0}-3\T_0\Big)\beta\right)+O(\beta^2),\label{app2_y0}

1242: \EEA \BEA

1243: & &\int_{-\infty}^{\infty}y^2e^{-\frac{1}{2\T(\beta)}(y^2+\beta\frac{y^4}{2})}~dy\nonumber\\

1244: &

1245: &=\sqrt\frac{\pi}{8}\sqrt{\T_0}\left(4\T_0+(6\T_1-15\T_0^2)\beta\right)+O(\beta^2),\label{app2_y2}

1246: \EEA \BEA

1247: & &\int_{-\infty}^{\infty}(y^2+\frac{\beta}{2}y^4)e^{-\frac{1}{2\T(\beta)}(y^2+\beta\frac{y^4}{2})}~dy\nonumber\\

1248: &

1249: &=\sqrt\frac{\pi}{8}\sqrt{\T_0}\left(4\T_0+(6\T_1-9\T_0^2)\beta\right)+O(\beta^2).\label{app2_y4}

1250: \EEA Then, in the first order in $\beta$, Eq.~(\ref{app2_eqT})

1251: becomes \BEA & &\T_0+\beta

1252: \T_1+\frac{\sqrt\frac{\pi}{8}\sqrt{\T_0}\left(4\T_0+(6\T_1-9\T_0^2)\beta\right)}{\sqrt\frac{\pi}{8}\sqrt{\T_0}\left(4+

1253: \Big(\frac{2\T_1}{\T_0}-3\T_0\Big)\beta\right)}=2\eb,\nonumber%\label{app2_eqT2}

1254: \EEA and we obtain $\T_0=\eb$ and $\T_1=(3/4)\eb^2$. Therefore, for

1255: the average kinetic energy density, we have \BEA \frac{\la

1256: K\ra}{N}=\frac{1}{2}\eb+\frac{3}{8}\eb^2\beta+O(\beta^2),\label{app2_Kb}

1257: \EEA and, for the average quadratic potential energy density, we

1258: have \BEA \frac{\la

1259: U\ra}{N}&=&\frac{1}{2}\frac{\sqrt\frac{\pi}{8}\sqrt{\T_0}\left(4\T_0+(6\T_1-15\T_0^2)\beta\right)}{\sqrt\frac{\pi}{8}\sqrt{\T_0}

1260: \left(4+\Big(\frac{2\T_1}{\T_0}-3\T_0\Big)\beta\right)}\nonumber\\

1261: &=&\frac{1}{2}\eb-\frac{9}{8}\eb^2\beta+O(\beta^2).\label{app2_U2b}

1262: \EEA Finally, we obtain Eq.~(\ref{small_eta}), i.e., for small

1263: $\beta$ \BEA

1264: \eta=1+\frac{3}{2}\eb\beta+O(\beta^2).\label{app2_eta_approx} \EEA

1265: Similarly, from Eq.~(\ref{new_eta}), we find the small $\beta$ limit

1266: of the approximation $\eta_{sc}$ \BEA

1267: \eta_{sc}=1+\frac{3}{2}\eb\beta+O(\beta^2).\nonumber \EEA Now, we

1268: consider the case of strong nonlinearity $\beta\rightarrow\infty$.

1269: From Eq.~(\ref{app2_eqT}), we conclude that temperature in the large

1270: $\beta$ limit, which we denote as $\tinf$, stays bounded, i.e., $

1271: 0<\tinf<2\eb, $ and, in the limit of large $\beta$, we obtain for

1272: Eq.~(\ref{app2_eqT}) \BEA

1273: \tinf+\frac{\int_{-\infty}^{\infty}\frac{\beta}{2}y^4e^{-\frac{\beta}{4\tinf}y^4}~dy}{\int_{-\infty}^{\infty}e^{-\frac{\beta}{4\tinf}y^4}~dy}=2\eb.

1274: \label{app2_eqT3} \EEA After performing the integration, we obtain $

1275: \tinf=(4/3)\eb, $ and the average kinetic energy density becomes $

1276: \la K\ra/N=(2/3)\eb $. For the average quadratic potential energy

1277: density, we have \BEA

1278: \frac{\la U\ra}{N}=\frac{1}{2}\frac{\int_{-\infty}^{\infty}y^2e^{-\frac{\beta}{4\tinf}y^4}~dy}{\int_{-\infty}^{\infty}e^{-\frac{\beta}{4\tinf}y^4}~dy}%=\nonumber\\

1279: %\frac{1}{2}\frac{\sqrt{2}\Gamma(\frac{3}{4})/\left(\frac{\beta}{\tinf}\right)^{\frac{3}{4}}}

1280: %{\Gamma(\frac{1}{4})/\Big(\sqrt{2}\left(\frac{\beta}{\tinf}\right)^{\frac{1}{4}}\Big)}

1281: =\frac{\Gamma(\frac{3}{4})}{\Gamma(\frac{1}{4})}\left(\frac{4\eb}{3\beta}\right)^{\frac{1}{2}}

1282: \EEA For the renormalization factor, we obtain the following large

1283: $\beta$ scaling \BEA

1284: \eta=\sqrt{\frac{\Gamma(\frac{3}{4})}{\sqrt{3}\Gamma(\frac{1}{4})}}\eb^\frac{1}{4}\beta^\frac{1}{4}.\label{app2_etainf}

1285: \EEA Similarly, for the approximation of $\eta_{sc}$, we obtain

1286: $A=C\sqrt{\eb\beta}$, $B=4\eb\beta$, and

1287: $C=2\sqrt{3}\Gamma(3/4)/\Gamma(1/4)$. Therefore, the large $\beta$

1288: scaling of $\eta_{sc}$ becomes \BEA

1289: \eta_{sc}=\sqrt{\frac{C+\sqrt{C^2+16}}{2}}\eb^\frac{1}{4}\beta^\frac{1}{4},\label{app2_etascinf}

1290: \EEA which yields Eq.~(\ref{large_eta}).

1291: \begin{thebibliography}{2000}

1292: 

1293: \bibitem{fpu50} Focus issue: The Fermi-Pasta-Ulam problem, the first 50 years, Chaos {\bf 15} (2005).

1294: 

1295: \bibitem{Peyrard} M. Peyrard, Nonlinearity {\bf 17}, R1 (2004), and references therein.

1296: 

1297: \bibitem{Toda} M. Toda, Theory of Nonlinear Lattice (Springer-Verlag, New York, 1989).

1298: 

1299: \bibitem{LL5} L. Landau and E. Lifshitz, {\it Statistical Physics}, Course of Theoretical Physics Vol. 5 (Pergamon, Oxford, 1981).

1300: 

1301: \bibitem{ZLF} V.E. Zakharov, V.S. Lvov,  and G. Falkovich, Kolmogorov Spectra of Turbulence (Springer-Verlag, Berlin, 1992).

1302: 

1303: \bibitem{Alabiso} C. Alabiso, M. Casartelli, P. Marenzoni, J. Stat. Phys. {\bf 79}, 451 (1995).

1304: 

1305: \bibitem{Lepri_renorm} S. Lepri, Phys. Rev. E {\bf 58}, 7165 (1998)

1306: 

1307: \bibitem{Umklapp} U. Z\"ulicke and A. H. MacDonald, Physica (Amsterdam) {\bf 6E}, 104 (2000) and references therein.

1308: 

1309: \bibitem{Kramer} J. Biello, P. Kramer, Y. Lvov, Proc. of the fourth international conf. on dyn. sys. and diff. eqns., 113 (2001);

1310: P. Kramer, J. Biello, Y. Lvov, \textit{ibid.}, 482 (2003).

1311: 

1312: \bibitem{LL1} L. Landau and E. Lifshitz, {\it Mechanics}, Course of Theoretical Physics Vol. 1 (Pergamon, Oxford, 1976).

1313: 

1314: \bibitem{Licht} A. J. Lichtenberg and M. A. Lieberman, {\it Regular and Chaotic Dynamics}, 2nd ed. (Springer-Verlag, Berlin, 1992).

1315: 

1316: \bibitem{our_prl} B. Gershgorin, Y.V. Lvov, D. Cai, Phys. Rev. Lett. {\bf 95}, 264302 (2005).

1317: 

1318: \bibitem{FPU} E. Fermi, J. Pasta, and S. Ulam, Los Alamos Scientific Laboratory Report No. LA-1940 (reprinted

1319: in Fermi E. Collected papers by University of Chicago Press,

1320: Chicago, 1965, Vol II, p 978).

1321: 

1322: \bibitem{Ford} J. Ford, Phys. Rep., {\bf 213}, 271, (1992).

1323: 

1324: \bibitem{Alabiso_fpu} C. Alabiso and M. Casartelli, J. Phys. A: Math. Gen. {\bf 34}, 1223 (2001).

1325: 

1326: \bibitem{Lepri_fpu} S. Lepri, R. Livi, A. Politi, Chaos {\bf 15}, 015118 (2005).

1327: 

1328: \bibitem{Carati} A. Carati, L. Galgani, A. Giorgilli, Chaos {\bf 15}, 015105 (2005).

1329: 

1330: \bibitem{Poggy} P. Poggy and S. Ruffo, Physica (Amsterdam) {\bf 103D}, 251 (1997).

1331: 

1332: \bibitem{Yoshida} H. Yoshida, Phys. Lett. A {\bf 150}, 262 (1990).

1333: 

1334: \bibitem{Parisi} G. Parisi, Europhysics Lett. 40, 357 (1997).

1335: 

1336: \bibitem{Cretegny} T. Cretegny et al., Physica (Amsterdam) {\bf 121D}, 109 (1998).

1337: 

1338: \bibitem{Newell} A. Newell, S. Nazarenko, and L. Biven, Physica (Amsterdam) {\bf 152D}, 520 (2001)

1339: % Wave turbulence and intermittency

1340: 

1341: \bibitem{Lvov} Y.V. Lvov and S. Nazarenko, Phys. Rev. E {\bf 69}, 066608 (2004);

1342:                Y. Choi, Y.V. Lvov, S. Nazarenko, and B. Pokorni, Phys. Lett. A {\bf 339}, 361 (2005).

1343: 

1344: \bibitem{Cai} D. Cai et al., Physica (Amsterdam) {\bf 152D}, 551 (2001).

1345: %Dispersive wave turbulence in one dimension

1346: 

1347: \bibitem{MMT} A.J. Majda, D.W. McLaughlin, and E.G. Tabak, J. Nonlinear Sci. {\bf 6}, 9 (1997).

1348: 

1349: \bibitem{Rink1} B. Rink, Comm. Math. Phys. {\bf 218(3)}, 665 (2001).

1350: % Symmetry and resonances in periodic FPU chains

1351: 

1352: \bibitem{Rink2} B. Rink,  Physica (Amsterdam) {\bf 175D}, 31 (2003).

1353: %Symmetric invariant manifolds in the Fermi-Pasta-Ulam lattice

1354: 

1355: \bibitem{Lvov2} Y.V. Lvov, S. Nazarenko, B. Pokorni, Physica (Amsterdam) {\bf 218D}, 24 (2006).

1356: 

1357: \bibitem{Colm} C. Connaughton, S.V. Nazarenko, and A.N. Pushkarev, Phys. Rev. E. {\bf 63}, 046306 (2001).

1358: 

1359: \bibitem{Pushkarev} A. Pushkarev, Eur. J. Mech. B/fluids {\bf 18(3)}, 345 (1999).

1360: %On the Kolmogorov frozen turbulence in numerical simulation of capillary waves

1361: 

1362: \bibitem{Benney} J. Benney and A. Newell, Stud. in Appl. Math. {\bf 48}, 29 (1969).

1363: % random wave closure

1364: \end{thebibliography}

1365: \end{document}

1366: