Coherent structures in long range FPUT lattices, Part I: Solitary Waves

Udoh Akpan, J. Douglas Wright · 2025

Read Coherent structures in long range FPUT lattices, Part I: Solitary Waves by Udoh Akpan, J. Douglas Wright (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-24828.txt.

Introduction

Douglas Wright and Udoh Akpan study solitary waves in long-range versions of a classic lattice model, in part one of a 2025 series. In the Fermi-Pasta-Ulam-Tsingou lattice, particles link along a chain, and the long-range variants let them interact over any distance. The authors develop conditions strong enough to guarantee that solitary wave solutions, waves that hold their shape as they travel, can be built in these lattices.

Catalog record

CREATORUdoh Akpan, J. Douglas Wright
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE51 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-24828

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Coherent structures in long range FPUT lattices, Part I: Solitary Waves Udoh Akpan, J. Douglas Wright Abstract 1. Preamble We put W^{s,p}:=W^{s,p}({\bf{R}}) , the usual Sobolev spaces as defined in (say) [ 1 ] . We also set H^{s}:=W^{s,2} , L^{2}:=H^{0} , L^{\infty}:=W^{0,\infty} and E^{s}:=H^{s}\cap\{\text{even functions}\} . We write “ f(x) is C^{r} on I ” if f^{(r)}(x) is uniformly continuous on the set I\subset{\bf{R}} . We write “ f(x) is C^{r,q} on I ” if f^{(r)}(x) is uniformly q -Hölder continuous on the set I\subset{\bf{R}} . We use the following conventions for the Fourier transform and its inverse: \displaystyle{\mathfrak{F}}[f](k):=\widehat{f}(k)={1\over 2\pi}\int_{\bf{R}}e^{-ikx}f(x)dx and \displaystyle{\mathfrak{F}}^{-1}[\widehat{f}](x):=\int_{\bf{R}}e^{ikx}\widehat{f}(k)dk . 2. Introduction We consider a general version of the Fermi-Pasta-Ulam-Tsingou (FPUT) lattice with long range interactions. Specifically: (1) \ddot{u}_{j}=\sum_{m\geq 1}\Phi_{m}^{\prime}(u_{j+m}-u_{j})-\Phi_{m}^{\prime}(u_{j}-u_{j-m}). Here j\in{\bf{Z}} , t\in{\bf{R}} , u_{j}(t)\in{\bf{R}} and \Phi_{m}:{\bf{R}}\to{\bf{R}} . The usual interpretation is that these are the equations of motion for a mechanical system consisting of infinitely many particles, indexed by j , arranged on a line and interacting pairwise. The position of the j^{th} particle at time t is u_{j}(t) and \Phi_{m}(r) gives the potential energy associated with the interaction between a particle and the particles m spots down the line, in either direction. In this setting ( 1 ) is simply Newton’s law. There are multiple other interpretations, ranging from power transmission and molecular dynamics to automotive traffic, but here we stick with the usual mechanical point of view. Our goal in this paper, and its sequel, is to understand the existence of traveling wave solutions to systems of the form ( 1 ). There are many situations where such solutions of ( 1 ) have been found: • When \Phi_{m}(r)=0 for m>1 , ( 1 ) corresponds to a classical FPUT problem and there is a rather vast literature on traveling wave solutions: see [ 12 , 3 , 2 , 13 , 14 , 10 , 9 , 8 ] . • The “next nearest neighbor” (NNN) problem, where \Phi_{m}(r)=0 for m>2 , has been studied in [ 3 , 15 ] . • The case where \Phi_{m}(r)=0 for m\geq N , for some N>1 , (that is to say interactions which have long, but finite, range) was studied in [ 4 ] 1 1 1 [ 4 ] , by Michael Herrmann and Alice Mikikits-Leitner, served as a major inspiration for this work. Michael Herrmann, a phenomenal mathematician and even more phenomenal colleague, passed away in Summer 2024. This article is dedicated to his memory. . We will call this situation the “finite range” (FR) problem. • Putting \Phi_{m}(r)=1/r^{a} corresponds to an analysis of generalized Calogero-Moser lattices. Solitary waves have recently been constructed for a\in(4/3,3) in [ 6 , 5 ] . Our work in this paper extends the results of [ 15 , 4 ] to interactions with infinite range, i.e. to situations where \Phi_{m}\neq 0 for infinitely many m . In particular we develop necessary conditions on the potentials \Phi_{m} which lead to the existence of solitary wave solutions. A notable application will be to generalized Calogero-Moser lattices with a>3 . In a subsequent paper we will construct periodic and generalized traveling wave solutions. The lattice ( 1 ) is in equilibrium when u_{j}(t)=r_{*}j , where r_{*} is a constant 2 2 2 This constant r_{*} can be any real number (or at least any real number for which r_{*}m is in the domain of \Phi_{m}^{\prime} ). Many works, though hardly all, make us of the fact that a simple coordinate change and modification of the potentials can put r_{*}=0 . We don’t do that and instead take r_{*} as being some specified, fixed value throughout. . We make the following assumption on the regularity of the potentials \Phi_{m} so that we can expand them in a series about this steady state. Assumption 1 . There exists \delta_{*}>0 so that, for all m\geq 1 , \Phi_{m}(r_{*}m+\eta) is C^{3,1} on |\eta|\leq m\delta_{*} . Consequently we have (2) \Phi^{\prime}_{m}(r_{*}m+\eta)=\varsigma_{m}+\alpha_{m}\eta+\beta_{m}\eta^{2}+\Psi^{\prime}_{m}(\eta) with \Psi^{\prime}_{m}(\eta)={\mathcal{O}}(\eta^{3}) . Precisely, \Psi^{\prime}_{m}(0)=0 and there are constants \gamma_{m}\geq 0 so that (3) |\eta|\leq m\delta_{*}\implies|\Psi^{\prime}_{m}(\eta)|\leq\gamma_{m}|\eta|^{3}\quad\text{and}\quad|\Psi^{\prime\prime}_{m}(\eta)|\leq 3\gamma_{m}|\eta|^{2}. Of course \varsigma_{m}:=\Phi^{\prime}(r_{*}m) , \alpha_{m}:=\Phi^{\prime\prime}(r_{*}m) and \beta_{m}:=\Phi^{\prime\prime\prime}(r_{*}m)/2 . Likewise \gamma_{m}:={\textrm{Lip }}\Phi^{\prime\prime\prime}_{m}/6, where by {\textrm{Lip }}\Phi^{\prime\prime\prime}_{m} we mean the Lipschitz constant of \Phi^{\prime\prime\prime}_{m}(r_{*}m+\eta) on |\eta|\leq m\delta_{*} . The constants \varsigma_{m} will play almost no role. Many aspects of our main results are best phrased in terms of the the dispersion relation, which we compute presently. 2.1. The dispersion relation, phase speed and classification thereof Linearizing ( 1 ) about the equilibrium r_{*}j gives (4) \ddot{u}_{j}=\sum_{m\geq 1}\alpha_{m}(u_{j+m}-2u_{j}+u_{j-m}). Plugging the plane wave Ansatz u_{j}(t)=e^{i(kj-\omega t)} into ( 4 ) and some routine algebra/trigonometry results in the dispersion relation (5) \omega^{2}=\theta(k):=\sum_{m\geq 1}4\alpha_{m}\sin^{2}(mk/2). We also define (6) \lambda(k):={\theta(k)\over k^{2}}=\sum_{m\geq 1}\alpha_{m}m^{2}\sinc^{2}(mk/2), along with (7) c_{0}^{2}:={\lambda(0)}=\sum_{m\geq 1}\alpha_{m}m^{2}. \sqrt{\lambda(k)} is the phase speed and c_{0} is variously referred to as the “speed of sound” or the “critical speed.” We make a few observations about \theta(k) : (a) it is 2\pi -periodic in the wavenumber k , (b) it is even in k , and (c) \theta(0)=0 . Indeed, given that we have not specified the coefficients \alpha_{m} at this stage, we see something rather surprising: \theta(k) can be any function meeting (a), (b) and (c). We codify this: Lemma 1 . Let \theta(k) be a piecewise continuously differentiable, even, 2\pi -periodic function which vanishes at k=0 . Then there are coefficients \alpha_{m} so that the linear dispersion relation for ( 1 ) is given by \omega^{2}=\theta(k) . Proof. The hypotheses on \theta(k) imply that it is equal to its Fourier cosine series. That is to say \theta(k)=\sum_{m=0}^{\infty}b_{m}\cos(mk) for some coefficients b_{m} . The condition that \theta(0)=0 tells us that b_{0}=-\sum_{m=1}^{\infty}b_{m} which in turn tells us that \theta(k)=\sum_{m=1}^{\infty}b_{m}(-1+\cos(mk)) . Letting \alpha_{m}=-b_{m}/2 along with the half angle formula finishes the proof. ∎ We will not make direct use of Lemma 1 in what follows, but that there is so much flexibility in \theta(k) means that it is often easier to work in terms of it and \lambda(k) as opposed to conditions on the coefficients \alpha_{m} . We now distinguish two special types of \lambda(k) : Definition 1 . duh • We say \lambda(k) is Type I if the following hold. (i) \lambda(k) is bounded below. (ii) \lambda^{\prime\prime}(0)<0 . (iii) There exists \mu_{*}>0 , k_{*}>0 and \sigma\in(0,2] so that |k|\leq k_{*} implies \lambda(k)-\lambda(0)\leq-\mu_{*}k^{2}\quad\text{and}\quad|\lambda(k)-\lambda(0)-{1\over 2}\lambda^{\prime\prime}(0)k^{2}|\leq\mu_{*}|k|^{2+\sigma}. (iv) \sup_{|k|\geq k_{*}}\lambda(k)<\lambda(0) . • We say \lambda(k) is Type II if the following hold. (1) \lambda(k) is bounded below. (2) \lambda^{\prime\prime}(0)>0 . (3) \lambda(k) is analytic for |\Im(k)|<\rho where \rho>0 . (4) There exist 0\leq c_{-}<c_{0} so that the equation \lambda(k)=c^{2} has a unique positive solution k=k_{c} when c\in(c_{-},c_{0}) and \lim_{c\to c_{0}^{-}}\lambda^{\prime}(k_{c})\neq 0 . In Figure 1 we sketch examples of \lambda(k) for both types. See the caption there for more details. We are not saying that all lattices fall into one of these two types. Indeed Lemma 1 precludes such an easy categorization. But a routine classical FPUT lattice is of Type I, as are the FR lattices studied in [ 4 ] . And the NNN lattices studied in [ 15 ] are of Type II. More on this below. (a) \lambda(k) is Type I. (b) \lambda(k) is Type II. Figure 1 . \lambda(k) vs k . Roughly speaking,

Read the complete text