Thèse en cours

Collisions en théorie cinétique à des échelles de temps longues

FR  |  
EN
Auteur / Autrice : Florent Fougères
Direction : Isabelle GallagherSergio Simonella
Type : Projet de thèse
Discipline(s) : Mathématiques
Date : Inscription en doctorat le 01/09/2023
Etablissement(s) : Université Paris sciences et lettres
Ecole(s) doctorale(s) : École doctorale Sciences mathématiques de Paris centre (Paris ; 2000-....)
Partenaire(s) de recherche : Laboratoire : École normale supérieure (Paris ; 1985-....). Département de mathématiques et applications (1998-....)
établissement opérateur d'inscription : Ecole normale supérieure

Résumé

FR  |  
EN

This project thesis is devoted to the mathematical analysis of scaling limits for large particle systems. The foundations of the kinetic theory of gases can be given a solid mathematical structure as predicted already in Hilbert's VIth problem and after a long series of works initiated by H. Grad and O. Lanford. Still, the theory is far from a mature state and the most important problems remain open. In this thesis we focus on two of such widely open problems: 1) large deviations on long time scales; 2) microscopic derivation of Landau's theory. The main goal of problem 1 is to describe the macroscopic fluctuations in a Hamiltonian system of classical particles (for instance hard spheres) with random initial conditions, in a limit of low-density (Boltzmann-Grad limit). The average behaviour is governed by the Boltzmann equation, and the most important problem in kinetic theory is constructing physical solutions of such equation globally in time. This problem seems to be exceptionally difficult. Even locally in time, the link with microscopic systems is very delicate. A completely alternative take on this problem, focused on large deviations, has been developed recently by T. Bodineau, I. Gallagher, L. Saint-Raymond and S. Simonella. Here the Boltzmann equation is replaced by a much richer structure based on a Hamilton-Jacobi equation, a variational principle and connections with gradient flows. Such a refined analysis is still restricted to very short times. In this project, we tackle the problem of extending the large deviations results to longer times. In order to do so, in absence of any previous result in this direction, we shall start from the simplest possible case. We will therefore study the small perturbations of a gas at equilibrium, for instance induced by a tagged particle in a Rayleigh gas. The main goal of problem 2 is to derive kinetic equations describing collisions in plasma physics. The common theme with project 1 is the control of the collisional dynamics up to long time scales, and here up to diffusive scales. It is indeed known from Landau's work that, for charged particles, the Boltzmann equation needs to be replaced by a diffusive variant of it. Landau argued from the Boltzmann equation itself, truncating divergencies and applying a delicate procedure in which binary collisions transform into a weak coupling process. This famous approximation argument has been the subject of intense mathematical investigation. However, a derivation of Landau's equation from the Newton laws is totally open. This problem requires the development of new, highly nontrivial techniques. A good model system to approach this problem is provided again by the Rayleigh gas, or by the even simpler Lorentz gas (a tracer particle moving in a random distribution of scatterers): the expected kinetic equation is in these cases a Landau equation of linear type. Some results are available in such cases, but only for unrealistic potentials or with no quantitative convergence bounds.