Stochastic multisymplectic PDEs and their structure-preserving numerical methods

Ruiao Hu and Linyu Peng 1. Department of Mathematics, Imperial College London,London SW7 2AZ, UK2. Department of Mechanical Engineering, Keio University,Yokohama 223-8522, Japan · 2025

Read Stochastic multisymplectic PDEs and their structure-preserving numerical methods by Ruiao Hu and Linyu Peng 1. Department of Mathematics, Imperial College London,London SW7 2AZ, UK2. Department of Mechanical Engineering, Keio University,Yokohama 223-8522, Japan (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-2501-16913.txt.

Introduction

Ruiao Hu and Linyu Peng develop math methods for wave-like systems that include random noise, in a paper from 2025. They extend a classic frame that tracks how energy and shape are kept as a system evolves. The goal is both new theory and workable ways to compute with it.

The pair build a random version of the frame and prove matching rules for what stays fixed over time. They also prove a link between smooth shifts and kept amounts. Next they design step-by-step solvers that keep the same traits on a grid. Tests on a key wave equation with random effects show the solvers work well.

Catalog record

CREATORRuiao Hu and Linyu Peng 1. Department of Mathematics, Imperial College London,London SW7 2AZ, UK2. Department of Mechanical Engineering, Keio University,Yokohama 223-8522, Japan
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE84 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-16913

Browse the full library · About the Tower of Babel

Full text

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

Stochastic multisymplectic PDEs and their structure-preserving numerical methods Ruiao Hu and Linyu Peng 1. Department of Mathematics, Imperial College London,London SW7 2AZ, UK2. Department of Mechanical Engineering, Keio University,Yokohama 223-8522, Japan Abstract We construct stochastic multisymplectic systems by considering a stochastic extension to the variational formulation of multisymplectic partial differential equations proposed in [Hydon, Proc. R. Soc. A , 461, 1627–1637, 2005]. The stochastic variational principle implies the existence of stochastic 1 1 -form and 2 2 -form conservation laws, as well as conservation laws arising from continuous variational symmetries via a stochastic Noether’s theorem. These results are the stochastic analogues of those found in deterministic variational principles. Furthermore, we develop stochastic structure-preserving collocation methods for this class of stochastic multisymplectic systems. These integrators possess a discrete analogue of the stochastic 2 2 -form conservation law and, in the case of linear systems, also guarantee discrete momentum conservation. The effectiveness of the proposed methods is demonstrated through their application to stochastic nonlinear Schrödinger equations featuring either stochastic transport or stochastic dispersion. Keywords: Stochastic variational principle; multisymplectic system; multisymplectic integrator; conservation law 1 Introduction The inclusion of stochasticity into dynamical systems is natural in modelling the effects of unresolvable fluctuations or unknown physical processes. One important application is in weather/climate forecasting, where stochastic perturbations are introduced to increase the ensemble variance [ 2 , 10 ] . However, when the physical dynamics systems possess inherent structures, and in turn, conservation laws, special types of structure-preserving stochasticity are required such that the underlying structures are preserved in the presence of noise. For Hamiltonian dynamics in finite dimensions, canonical Hamilton’s equations possess conservation laws that include the total energy (Hamiltonian), symplectic form and Liouville form [ 24 , 35 ] . In addition to conservation laws, when the system possesses variational symmetries, reduction by symmetry can be performed to obtain Lie–Poisson equations that are equivalent to canonical Hamilton’s equations [ 23 , 33 , 34 ] . The dynamics of the reduced system lies on coadjoint orbits and they typically have Casimir functions that are invariant under the motion. One well-studied class of structure-preserving stochastic dynamics is the stochastic canonical Hamilton’s equations, see, e.g., [ 3 , 5 , 31 ] , where the conservation of symplectic form and Liouville Theorem are preserved under the stochastic motion. However, the conservation of energy is lost. In addition to the preservation of conservation laws, when the deterministic system and its stochastic perturbations possess variational symmetries, reduction by symmetry can be performed to obtain equivalent stochastic Lie–Poisson equations. In the pioneering work of Holm [ 20 ] , stochasticity of Hamiltonian type is derived for applications to ideal fluid dynamics which takes the form of an infinite-dimensional Lie–Poisson equation defined on \mathfrak{X}^{*} , the space of 1 -form densities. In this setting, the stochastic perturbations take the form of stochastic transport, exhibiting remarkable analytical properties. The aim of this paper is twofold. The first is to provide an extension of the stochastic canonical Hamilton’s equations to infinite dimensions by modifying the variational multisymplectic framework of [ 6 , 8 , 28 ] to include stochasticity. By working in the stochastic multisymplectic framework, the resulting stochastic multisymplectic partial differential equations (PDEs), also referred to as Hamiltonian PDEs, inherently possess stochastic analogues of the deterministic conservation laws. The second aim of this paper is to construct structure-preserving methods such that (at least some of) the stochastic conservation laws are preserved in numerical simulations. In the deterministic setting, these methods are known as multisymplectic methods first introduced in [ 9 ] which has seen extensions to the stochastic setting, see, e.g., [ 13 , 29 ] . Main contents of the paper. We summarise the main contributions and contents of this paper as follows. • In Section 2 , after a brief review of deterministic multisymplectic systems and their associated conservation laws, we construct a general class of stochastic multisymplectic systems by considering appropriate stochastic variational principles in Section 2.2 . These stochastic dynamics possess stochastic analogues of the deterministic multisymplectic 1 -form and 2 -form conservation laws, and they are sufficiently general to incorporate existing stochastic multisymplectic systems. To demonstrate the flexibility of the construction, we present the multisymplectic formulation of stochastic transport noise and stochastic nonlinear Schrödinger (NLS) equations in Sections 2.3 and 2.4 , respectively. • In Section 3 , we construct a general family of stochastic structure-preserving numerical schemes based on stochastic Runge–Kutta methods that preserve exactly, a discrete version of the continuous stochastic symplectic form conservation law. For quadratic Hamiltonians, we show that the methods constructed can additionally preserve exactly, a discrete version of the stochastic momentum conservation law. In Section 3.2 , we evaluate the behaviour of the numerical schemes applied to stochastic NLS equations obtained in Section 2.4 . • Section 4 contains concluding remarks and a discussion on directions for future work. 2 Stochastic multisymplectic systems 2.1 Review of deterministic multisymplectic systems We brief review the formulations of multisymplectic systems and their associated conservation laws following [ 8 , 15 , 28 ] . A system of PDEs is said to be multisymplectic if it can be derived from a variational principle with the Lagrangian L(x,[z]) of the following form \displaystyle L({x},[{z}])=L^{i}_{\alpha}({x},{z})z^{\alpha}_{,i}-H({x},{z})\,, (2.1) where {x}=(x^{0},x^{1},\ldots,x^{n})\in\mathbb{R}^{n+1} is the independent (base space) variable with the identification x^{0}=t , {z}=(z^{1},z^{2},\ldots,z^{m})\in\mathbb{R}^{m} is the dependent variable, and H({x},{z}) is the Hamiltonian of the system. Furthermore, total derivative D_{i} with respect to x^{i} is also denoted by the subscript i after a comma, [{z}] denotes {z} and finitely many of its derivatives, and the Einstein summation convention is assumed unless otherwise stated. Hence, the Lagrangian for a (deterministic) multisymplectic system is affine in the first derivatives of the dependent variables. To derive the Euler–Lagrange equations associated with the Lagrangian L(x,[z]) , we consider the corresponding variational principle, \displaystyle 0=\delta\mathcal{S}[z]=\delta\int L({x},[{z}])\,\mathrm{d}vol\,, where \mathrm{d}vol is the volume form on the base space locally defined by \mathrm{d}vol:=\mathrm{d}t\wedge\mathrm{d}x^{1}\wedge\cdots\wedge\mathrm{d}x^{n} and the variations \delta z are assumed to be arbitrary and vanishing at boundaries. This results the Euler–Lagrange equations \displaystyle K^{i}_{\alpha\beta}({x},{z})z^{\beta}_{,i}-\frac{\partial L^{i}_{\alpha}}{\partial x^{i}}=\frac{\partial H}{\partial z^{\alpha}}\,,\quad\text{where}\quad K^{i}_{\alpha\beta}:=\frac{\partial L^{i}_{\beta}}{\partial z^{\alpha}}-\frac{\partial L^{i}_{\alpha}}{\partial z^{\beta}}\,, (2.2) and K^{i}_{\alpha\beta} are known as the multisymplectic structure matrices. By construction, K^{i}_{\alpha\beta} are antisymmetric for all i and we can construct closed 2 -forms \kappa^{i} by \displaystyle\kappa^{i}=\frac{1}{2}K^{i}_{\alpha\beta}({x},{z})\,\mathrm{d}z^{\alpha}\wedge\mathrm{d}z^{\beta},\quad i=0,1,\ldots,n\,, (2.3) which satisfy \displaystyle\kappa^{i}_{,i}=0\,. (2.4) This is known as the structural conservation law [ 7 ] . Via the Poincaré Lemma for the variational bicomplex (see, e.g., [ 1 , 8 , 43 ] ), we can obtain \left(L_{\beta}^{j}(z)z_{,i}^{\beta}\right)_{,j}-\left(L^{j}_{\beta}(z)z^{\beta}_{,j}-H\right)_{,i}=\frac{\partial H}{\partial x^{i}}-\frac{\partial L_{\beta}^{j}}{\partial x^{i}}z^{\beta}_{,j}\,,\quad i=0,1,\ldots,n\,. (2.5) If the Lagrangian does not depend on some particular components of the independent variables, then conservation laws can be obtained as the RHS for these components vanishes. This also generalises the conservation laws obtained in [ 28 ] ; namely, when the Lagrangian is independent from all x^{i} (which is often true), the RHS of ( 2.5 ) always vanishes. In this special case, the Euler–Lagrange equations ( 2.2 ) become \displaystyle K^{i}_{\alpha\beta}({z})z^{\beta}_{,i}=\frac{\partial H}{\partial z^{\alpha}}\,, (2.6) and now the 1 -form \mathrm{d}L(x,[z])=\mathrm{d}(L^{i}_{\alpha}({z})z^{\alpha}_{,i}-H({z})) satisfies the 1 -form quasi-conservation law \displaystyle(L^{i}_{\alpha}({z})\mathrm{d}z^{\alpha})_{,i}=\mathrm{d}\left(L^{i}_{\alpha}({z})z^{\alpha}_{i}-H({z})\right)=\mathrm{d}L(x,[z])\,. (2.7) Specially, the j=0 case in equation ( 2.5 ) gives the energy conservation law and the other cases give

Read the complete text