A comparative numerical study of stochastic Hamiltonian Camassa-Holm equations

Darryl D. Holm, Maneesh Kumar Singh, Oliver D. Street · 2025

Read A comparative numerical study of stochastic Hamiltonian Camassa-Holm equations by Darryl D. Holm, Maneesh Kumar Singh, Oliver D. Street (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-2507-14264.txt.

Introduction

Holm, Singh, and Street compare two ways to add random noise to a known wave equation in a 2025 math paper. The base equation, the Camassa-Holm equation, is known for forming sharp peaked waves called peakons. Past noisy versions changed key traits of the flow, so the authors build one version that keeps total energy fixed.

They set that version against a second approach that keeps a different set of values, called Casimirs, fixed. Computer runs show the energy keeping version stays spread around the plain solution. The other version can send peakons drifting much farther from the plain path. The contrast helps users pick a noisy model with care.

Catalog record

CREATORDarryl D. Holm, Maneesh Kumar Singh, Oliver D. Street
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-14264

Browse the full library · About the Tower of Babel

Full text

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

A comparative numerical study of stochastic Hamiltonian Camassa-Holm equations Darryl D. Holm, Maneesh Kumar Singh, Oliver D. Street Abstract We introduce a stochastic perturbation of the Camassa-Holm equation such that, unlike previous formulations, energy is conserved by the stochastic flow. We compare this to a complementary approach which preserves Casimirs of the Poisson bracket. Through an energy preserving numerical implementation of the model, we study the influence of noise on the well-known ‘peakon’ formation behaviour of the solution. The energy conserving stochastic approach generates an ensemble of solutions which are spread around the deterministic Camassa-Holm solution, whereas the Casimir conserving alternative develops peakons which may propagate away from the deterministic solution more dramatically. 1 Introduction The word ‘ peakon ’ had not yet appeared in the literature as a scientific term before the publication of the paper ( Camassa and Holm, 1993 ) . However, the success of that paper has resulted in the words ‘peakon’ and ‘Camassa–Holm (CH) equation’ becoming common terms in the literature of emergent singular solutions of nonlinear wave equations. An interest in stochastic equations governing the dynamics of a physical system, derived using the geometric structure of the corresponding deterministic model, has recently been developing. This interest has grown in popularity following the inclusion of transport noise in ideal fluid dynamics, developed by Holm, 2015 using a Lie group invariant variational principle. As illustrated in ( Street and Takao, 2023 ) , this approach is intimately related to the seminal work of Bismut, 1981 on finite-dimensional stochastic Hamiltonian systems, the theory of which was later studied by Lázaro-Camí and Ortega, 2008 . More specifically to this paper, these methods have also been applied to derive stochastic wave models ( Crisan and Holm, 2018 ; Street, 2023 ) . Several studies of the properties of a Stochastic Camassa-Holm (SCH) equation perturbed with transport noise have recently been undertaken. The formulation of the transport noise perturbation is called Stochastic Advection by Lie Transport (SALT) and is based on stochastic variational principles with Lie group symmetry. As shown in Crisan and Holm, 2018 , wave breaking in the form of peakon emergence occurs with positive probability for the stochastic CH equation with SALT. It was conjectured in Crisan and Holm, 2018 that the time-asymptotic solutions of their stochastic CH equation would comprise emergent wave trains of peakons moving along stochastic space-time paths. This was illustrated numerically in the paper ( Bendall et al., 2021 ) , which presented a finite-element discretisation for this model, and explored the formation of peakons. The simulations presented in Bendall et al., 2021 using this discretisation with the SALT approach revealed that peakons can still form in the presence of stochastic perturbations. Peakons can emerge both through wave breaking, as the slope turns vertical, and also without wave breaking as the inflection points of the velocity profile rise to meet at the summit. The transport noise form of the stochastic CH equation was also analysed in Albeverio et al., 2021 , whose main result is the proof of the existence and uniqueness of local strong solutions in the Sobolev spaces H^{2,q} for 1<q<\infty . As mentioned in Albeverio et al., 2021 some results have also been shown in Tang, 2018 and Chen et al., 2021 for a different class of stochastic perturbations of the CH equation. Namely, the results in this other class are that pathwise solutions of the CH equation exist under perturbation by stochastic linear multiplicative forcing with Itô noise, provided certain initial conditions are sufficiently smooth. However, those studies do not cover the transport noise case, where the diffusion coefficient is generated by an unbounded linear operator. The same may be said about Chen et al., 2021 ; Chen and Li, 2022 , although transport noise is mentioned for the stochastic two-component CH equation in an appendix of Chen et al., 2023 . Moreover, Cruzeiro and Liu, 2017 use an alternative stochastic variational approach to derive a different stochastic Camassa-Holm equation. Finally, Albeverio et al., 2021 re-emphasise the importance of studying the stochastic CH equation because of its geometric and physical motivations, and its relevance in nonlinear wave theory and geophysical applications. 1.1 Objectives of the paper The primary objective of this paper is the comparison of solutions of stochastic Camassa–Holm (SCH) equations for two different structure-preserving types of stochasticity. The development of structure-preserving stochastic models of Hamiltonian fluid dynamics may be split into two approaches, commonly referred to in the literature as Stochastic Forcing by Lie Transport (SFLT) and Stochastic Advection by Lie Transport (SALT). Stochastic Option 1 (SFLT) preserves the energy (Hamiltonian), whilst stochasticity perturbing the Poisson operator. Stochastic Option 2 (SALT) preserves the Poisson operator, while it introduces stochasticity into the energy Hamiltonian. The comparison between these two alternatives is based on how introducing either option affects the coherent structures of the deterministic CH equation; namely, its solitary wave solutions known as ‘peakons’ for the peaked shape of their profiles. Comparing the SCH solution behaviour for these two choices of stochasticity reveals the roles of the energy Hamiltonian and the Poisson operator. Said briefly, the metric in the kinetic-energy Hamiltonian is responsible for the shape of the coherent structures in the solitary wave solution (peakons) and the Poisson operator is responsible for creating these coherent structures. 2 The Model 2.1 Physical derivation of the deterministic Camassa-Holm equation An asymptotic expansion in the dimension-free scaling parameters \epsilon_{1}^{2}<\epsilon_{2}<\epsilon_{1}\ll 1 of the free-surface Euler fluid equation for long wave-length, small-amplitude nonlinear water waves of elevation, \eta , yields ( Dullin et al., 2004 ) \displaystyle 0 \displaystyle=\eta_{t}+\eta_{x}+\frac{3}{2}\epsilon_{1}\,\eta\,\eta_{x}+\frac{1}{6}\epsilon_{2}\,\eta_{xxx}-\frac{3}{8}\epsilon_{1}^{2}\,\eta^{2}\,\eta_{x} \displaystyle+\,\epsilon_{1}\epsilon_{2}\left(\frac{23}{24}\eta_{x}\,\eta_{xx}+\frac{5}{12}\eta\,\eta_{xxx}\right)+\epsilon_{2}^{2}\frac{19}{360}\eta_{xxxxx}+o(\epsilon_{1}^{2},\epsilon_{2})\,. At this order of asymptotic expansion, certain choices of the three free constant parameters \alpha_{1} , \alpha_{2} , and \beta in the (non-local) Kodama normal form transformation ( Kodama and Mikhailov, 1997 ) , \displaystyle\eta=u+\epsilon_{1}\left(\alpha_{1}u^{2}+\alpha_{2}u_{x}\partial_{x}^{-1}u\right)+\epsilon_{2}\,\beta u_{xx}\,, result in the deterministic CH equation, expressed as ( Camassa and Holm, 1993 ) u_{t}-\alpha^{2}u_{txx}=-3uu_{x}+2\alpha^{2}u_{x}u_{xx}+\alpha^{2}uu_{xxx}\,, (2.1) or, equivalently, m_{t}=-(\partial m+m\partial)u\quad\hbox{where}\quad m:=(1-\alpha^{2}\partial^{2})u\,. (2.2) The ‘ hydrodynamic ’ form of the CH equation is ( Camassa and Holm, 1993 ) \displaystyle\begin{split}u_{t}+uu_{x}&=-\,\partial_{x}\int_{\mathbb{R}}\frac{1}{2}\,e^{-|x-y|/\alpha}\left(u^{2}(y,t)+\frac{1}{2}u_{y}^{2}(y,t)\right)\,dy\\ &=:-\,K\ast\Big(u^{2}+\frac{1}{2}u_{x}^{2}\Big)_{x}\,.\end{split} (2.3) Here, the peaked kernal K(|x-y|)=\tfrac{1}{2}\exp(-|x-y|/\alpha) in the convolution integral in ( 2.3 ) is the Green function for the Helmholz operator (1-\alpha^{2}\partial^{2}) appearing in equation ( 2.2 ). Solutions of equation ( 2.3 ) determine the vector field u=\dot{g}_{t}{g}_{t}^{-1} which generates the smooth invertible flow g_{t}\in{\rm Diff}(S^{1}) governing the solution of the CH equation ( 2.3 ). The geometric content of this statement may be illustrated by writing the CH equation in ( 2.2 ) equivalently in terms of the tangent to the push-forward {g_{t}}_{*} of the right action of g_{t} on the 1-form density m(x,t)dx\otimes dx . Namely, \frac{d}{dt}{g_{t}}_{*}\big(m(x,t)dx\otimes dx\big)=-\,{g_{t}}_{*}{\cal L}_{\dot{g}_{t}{g}_{t}^{-1}}(mdx\otimes dx)\,. (2.4) This tangent relation also defines the Lie derivative {\cal L}_{u} along the characteristic curves of the vector field u=\dot{g}_{t}{g}_{t}^{-1} . Upon evaluating {g}_{t} in ( 2.4 ) at the identity, {g}_{t}|_{t=0}={Id} , the CH equation may finally be written geometrically as, \partial_{t}\big(m(x,t)\,dx\otimes dx\big)=-\,{\cal L}_{u}\big(m(x,t)\,dx\otimes dx\big)\,. (2.5) Solitary wave solutions The CH equation ( 2.1 ) is known for its solitary wave solutions. The deterministic CH equation produces trains of peaked solitary wave solutions called ‘peakons’, given by the formula u(x,t)=\frac{1}{2}\sum_{a=1}^{M}p_{a}(t)\mathrm{e}^{-|x-q_{a}(t)|/\alpha}=\sum_{a=1}^{M}p_{a}(t)K(x-q_{a}(t))\,. (2.6) Since u=K*m , the formula ( 2.6 ) for the

Read the complete text