Jin Feng (University of Kansas)
Title: Hydrodynamic limit large deviation for stochastic Carleman particles, a Hamilton-Jacobi approach
Abstract: The deterministic Carleman equation can be considered as an one dimensional two speed fictitious gas model. Its associated hydrodynamic limit gives a nonlinear heat equation. The first rigorous derivation of such limit was given by Kurtz in 1973. In this talk, starting from a more refined stochastic model giving the Carleman equation as the mean field, we derive a macroscopic fluctuation structure associated with the hydrodynamic limit.
The large deviation result is established through an abstract Hamilton-Jacobi method applied to this specific setting. The principal idea is to identify a two scale averaging structure in the context of Hamiltonian convergence in the space of probability measures. This is conceptually achieved through a change of coordinate to the density-flux description of the problem. We also extend a method in the weak KAM theory to the infinite particle context for explicitly identifying the effective Hamiltonian. In the end, we conclude by establishing a comparison principle for a set of Hamilton-Jacobi equations in the space of measures.
I will present some subtle issues involved and put the method into perspective regarding challenges we face when applying to other hydrodynamic issues.
This is a joint work with Toshio Mikami and Johannes Zimmer.