Hypocoercivity for infinite-dimensional non-linear degenerate stochastic differential equations with multiplicative noise

TL;DR

This paper establishes hypocoercivity and exponential ergodicity for infinite-dimensional non-linear degenerate stochastic PDEs with multiplicative noise, providing explicit convergence rates and solutions construction.

Contribution

It develops a hypocoercivity framework for infinite-dimensional stochastic equations, constructing solutions and proving exponential convergence to equilibrium with explicit constants.

Findings

Proves essential m-dissipativity of Kolmogorov generators.

Constructs weak solutions with infinite lifetime.

Derives explicit exponential convergence rates.

Abstract

We analyze infinite-dimensional non-linear degenerate stochastic differential equations with multiplicative noise. First, essential m-dissipativity of their associated Kolmogorov backward generators on $L^2(\mu^{\Phi})$ defined on smooth finitely based functions is established. Here $\Phi$ is an appropriate potential and $\mu^{\Phi}$ is the invariant measure with density $e^{-\Phi}$ w.r.t. an infinite-dimensional non-degenerate Gaussian measure. Second, we use resolvent methods to construct corresponding right processes with infinite lifetime, solving the martingale problem for the Kolmogorov backward generators. They provide weak solutions, with weakly continuous paths, to the non-linear degenerate stochastic partial differential equations. Moreover, we identify the transition semigroup of such a process with the strongly continuous contraction semigroup $(T_t)_{t\geq 0}$ generated by the corresponding Kolmogorov backwards generator. Afterwards, we apply a refinement of the abstract Hilbert space hypocoercivity method, developed by Dolbeaut, Mouhot and Schmeiser, to derive hypocoercivity of $(T_t)_{t\geq 0}$. I.e. we take domain issues into account and use the formulation in the Kolmogorov backwards setting worked out by Grothaus and Stilgenbauer. The method enables us to explicitly compute the constants determining the exponential convergence rate to equilibrium of $(T_t)_{t\geq 0}$. The identification between $(T_t)_{t\geq 0}$ and the transition semigroup of the process yields exponential ergodicity of the latter. Finally, we apply our results to second order in time stochastic reaction-diffusion and Cahn-Hilliard type equations with multiplicative noise. More generally, we analyze corresponding couplings of infinite-dimensional deterministic with stochastic differential equations.