Deterministic particle method for Fokker-Planck equation with strong oscillations - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2023

Deterministic particle method for Fokker-Planck equation with strong oscillations

Résumé

The aim of this paper is to investigate a deterministic particle method for a model containing a Fokker-Planck collision operator in velocity and strong oscillations (characterized by a small parameter ε) induced by a space and velocity transport operator. First, we investigate the properties (collisional invariants and equilibrium) of the asymptotic model obtained when ε → 0. Second a numerical method is developed to approximate the solution of the multiscale Fokker-Planck model. To do so, a deterministic particle method (recently introduced for the Landau equation in [8]) is proposed for Fokker-Planck type operators. This particle method consists in reformulating the collision operator in an advective form and in regularizing the advection field in such a way that it conserves the geometric bracket structure. In the Fokker-Planck homogeneous case, the properties of the resulting method are analysed. In the non homogeneous case, the particle method is coupled with a uniformly accurate time discretization in ε that enables to capture numerically the solution of the asymptotic model. Numerous numerical results are displayed, illustrating the behavior of the method.
Fichier principal
Vignette du fichier
Deterministic_particle_method_for_Fokker-Planck_equation_with_strong_oscillations.pdf (2.38 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04343011 , version 1 (13-12-2023)

Licence

Paternité

Identifiants

  • HAL Id : hal-04343011 , version 1

Citer

Anais Crestetto, Nicolas Crouseilles, Damien Prel. Deterministic particle method for Fokker-Planck equation with strong oscillations. 2023. ⟨hal-04343011⟩
41 Consultations
33 Téléchargements

Partager

Gmail Facebook X LinkedIn More