Spectrally accurate fully discrete schemes for some nonlocal and nonlinear integrable PDEs via explicit formulas
Yvonne Alama Bronsard, Xi Chen, Matthieu Dolbeault · 2024
Read Spectrally accurate fully discrete schemes for some nonlocal and nonlinear integrable PDEs via explicit formulas by Yvonne Alama Bronsard, Xi Chen, Matthieu Dolbeault (2024) 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-2412-13480.txt.
Introduction
Yvonne Alama Bronsard, Xi Chen, and Matthieu Dolbeault design computer schemes for three hard wave equations that are exact in time and highly accurate in space. The Benjamin-Ono, Calogero-Sutherland, and cubic Szegő equations are nonlinear and nonlocal, yet integrable, with rare exact solution formulas. The authors build fully discrete methods on those formulas. Each step costs the same no matter how far ahead the simulation runs.
They prove sharp convergence rates for the first two equations, with error growing only linearly in time rather than exploding. Tests at short and long times confirm the strength of the approach. Because cost ignores the final time, the schemes suit the study of long-term behavior. The authors present them as a new tool for integrable systems, where long horizons usually defeat standard methods.
Catalog record
| CREATOR | Yvonne Alama Bronsard, Xi Chen, Matthieu Dolbeault |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-13480
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Spectrally accurate fully discrete schemes for some nonlocal and nonlinear integrable PDEs via explicit formulas Yvonne Alama Bronsard, Xi Chen, Matthieu Dolbeault Abstract We construct fully-discrete schemes for the Benjamin–Ono, Calogero–Sutherland DNLS, and cubic Szegő equations on the torus, which are exact in time with spectral accuracy in space. We prove spectral convergence for the first two equations, of order K − s + 1 K^{-s+1} for initial data in H s ( 𝕋 ) H^{s}(\mathbb{T}) , with an error constant depending linearly on the final time instead of exponentially. These schemes are based on explicit formulas , which have recently emerged in the theory of nonlinear integrable equations. Numerical simulations show the strength of the newly designed methods both at short and long time scales. These schemes open doors for the understanding of the long-time dynamics of integrable equations. Keywords— Integrable systems, Benjamin–Ono equation, explicit formulas, Lax pairs, spectral accuracy, fully discrete error analysis, long-time dynamics 1 Introduction We consider fully discrete approximations to three nonlinear and nonlocal integrable equations. Important progress has recently been made on the theoretical level for these equations, opening the way to new numerical approaches that we present here. The first equation, central in the theory of integrable systems, is the Benjamin–Ono equation \partial_{t}u(t,x)=\partial_{x}\left(|D|u-u^{2}\right)(t,x),\quad u_{|t=0}(x)=u_{0}(x),\quad(t,x)\in\mathbb{R}\times\mathbb{T}, (BO) where u(t,x)\in\mathbb{R} is a real-valued solution, D=\frac{1}{i}\frac{d}{dx} and |D| is defined in Fourier space as \widehat{|D|f}(k)=|k|\widehat{f}(k),\quad f\in L^{2}(\mathbb{T}). This nonlocal quasilinear dispersive equation models long, unidirectional internal gravity waves in two-layered fluids [ 7 , 41 ] , as rigorously justified in the recent work [ 44 ] . Although the ( BO ) equation resembles the well-known Korteweg–de Vries equation (KdV), with Airy’s dispersive flow \partial_{t}+\partial_{xxx} replaced by a Schrödinger-type flow \partial_{t}-\partial_{x}|D| , the dispersion present in the equation is significantly reduced, thus rendering the control of the derivative in the nonlinearity a harder problem. Using techniques from the theory of integrable systems, and notably a Birkhoff normal form transformation, Gérard–Kappeler–Topalov [ 27 ] show global well-posedness of ( BO ) in H^{s}(\mathbb{T}) spaces if s>-\frac{1}{2} and ill-posedness otherwise, see also Killip–Laurens–Vişan [ 34 ] . For a recent survey of known results and open challenges we refer to the book of Klein–Saut [ 37 , Chapter 3] , and the references therein. The second equation considered is the focusing ( + sign) or defocusing ( - sign) Calogero–Sutherland derivative nonlinear Schrödinger (DNLS) equation i\partial_{t}u+\partial_{x}^{2}u\pm\frac{2}{i}u\partial_{x}\Pi(|u|^{2})=0,\quad u_{|t=0}(x)=u_{0}(x),\quad(t,x)\in\mathbb{R}\times\mathbb{T}, (CS) where the Riesz–Szegő projector \Pi is defined in Fourier space as \widehat{\Pi f}(k)=\mathbbm{1}_{k\geq 0}\,\widehat{f}(k),\quad f\in L^{2}(\mathbb{T}). ( \Pi ) This is a nonlocal nonlinear Schrödinger equation and is derived from the Calogero–Sutherland–Moser system in [ 12 , 48 , 49 ] . This physical model represents a system of N identical particles interacting pairwise. Abanov, Bettelheim and Wiegmann [ 1 ] formally show that taking the thermodynamic limit of such a model and applying a change of variables leads to the ( CS ) equation. One can also recover ( CS ) formally as a limit of the intermediate nonlinear Schrödinger equation introduced by Pelinovsky [ 43 ] . Badreddine [ 4 ] achieves global well-posedness in the Hardy–Sobolev space H_{+}^{s}(\mathbb{T})=\Pi(H^{s}) for s\geq 0 , by additionally requiring small initial data \|u_{0}\|_{L^{2}}^{2}=\frac{1}{2\pi}\int_{-\pi}^{\pi}|u_{0}|^{2}<1 in the focusing case. Remarkably, even though ( CS ) is a completely integrable equation, one can expect the existence of finite time blow-ups in the focusing case. Indeed, on the real line, Gérard–Lenzmann [ 28 ] prove global well-posedness in H^{1}_{+}(\mathbb{R}) if \|u_{0}\|_{L^{2}(\mathbb{R})}^{2}=\int_{\mathbb{R}}|u_{0}|^{2}\leq 2\pi , whereas Kim–Kim–Kwon [ 35 ] very recently show the existence of smooth solutions with mass arbitrarily close to 2\pi , whose H^{1} norm blows up in finite time. In the periodic setting, the dynamics of the focusing ( CS ) equation for initial data with mass greater or equal to one remains a compelling open problem. Finally, the third equation is the cubic Szegő equation i\partial_{t}u=\Pi(|u|^{2}u),\quad u_{|t=0}(x)=u_{0}(x),\quad(t,x)\in\mathbb{R}\times\mathbb{T}, (S) where \Pi is once again the Riesz–Szegő projector. This equation is introduced in [ 23 ] by Gérard and Grellier who show global well-posedness in H_{+}^{s}(\mathbb{T}) for s\geq{1}/{2} , using the fact that the norm H^{{1}/{2}}_{+} is conserved. As opposed to the last two equations, ( S ) is non-dispersive, and is used as a toy model for studying the NLS equation when there is a lack of dispersive smoothing due to the confining geometry of the domain. Another motivation comes from the study of wave turbulence , since the equation admits energy cascades from low to high frequencies, as well as energy transfers from high to low frequencies due to the almost time-periodicity of the solution [ 25 ] . A key feature of these integrable equations is the existence of Lax pairs [ 20 , 38 ] , from which an infinite number of conservation laws can be derived. Recently, ground-breaking results were obtained for the ( BO ), ( CS ) and ( S ) equations, proving the existence of an explicit formula for the solution u , based upon their Lax pair structure. On the torus, the first result is due to Gérard and Grellier [ 24 ] for the ( S ) equation, followed by Gérard [ 22 ] for the ( BO ) equation, and Badreddine [ 4 ] for the ( CS ) equation. The goal of this paper is to make a bridge between these new analytical results and the field of computational mathematics, by obtaining efficient approximations to the above equations via these explicit formulas and proving their convergence on the discrete level. Before presenting our methodology, we discuss previous numerical discretizations to the above equations. While ( CS ) and ( S ) are relatively recent, there exists a vast literature on the numerical approximation of ( BO ). We detail some of these works here, with an emphasis on results providing explicit convergence rates. Given the nonlocal nature of the linear operator \partial_{x}|D| and its diagonal expression ik|k| in the Fourier variables, pseudo-spectral methods are usually adopted due to their computational efficiency. This leads to spatial semi-discretizations, which are then coupled with suitable time approximations, such as finite differences. For a comparison of different efficient spectral numerical methods we refer to Boyd–Xu [ 11 ] , and to Deng–Ma [ 16 ] for a semi-discrete pseudo-spectral error analysis result. In the fully discrete case, Pelloni–Dougalis [ 45 ] prove convergence of a scheme combining leap-frog in time and spectral Galerkin method in space, whose error analysis is refined in Deng–Ma [ 15 ] , while Galtung [ 19 ] studies a Crank–Nicolson Galerkin scheme. On the full line \mathbb{R} , fully discrete approximations are also analyzed, where the authors consider a large torus in numerical implementations. We refer to Thomee [ 50 ] and Dutta–Holden–Koley–Risebro [ 17 ] for a finite difference approximation, and Dwivedi–Sarkar [ 18 ] for a local discontinuous Galerkin method. Unlike previous methods, which rely on discretizing the underlying PDE, we introduce novel schemes based on the explicit formulas of [ 22 , 4 , 24 ] . While these formulas give an explicit representation of the solution u(t) in terms of the initial data u_{0} and the time t , they involve taking the inverse of a