English

The essential m-dissipativity for degenerate infinite dimensional stochastic Hamiltonian systems and applications

Probability 2024-10-22 v1 Functional Analysis

Abstract

We consider a degenerate infinite dimensional stochastic Hamiltonian system with multiplicative noise and establish the essential m-dissipativity on L2(μΦ)L^2(\mu^{\Phi}) of the corresponding Kolmogorov (backwards) operator. Here, Φ\Phi is the potential and μΦ\mu^{\Phi} the invariant measure with density eΦe^{-\Phi} with respect to an infinite dimensional non-degenerate Gaussian measure. The main difficulty, besides the non-sectorality of the Kolmogorov operator, is the coverage of a large class of potentials. We include potentials that have neither a bounded nor a Lipschitz continuous gradient. The essential m-dissipativity is the starting point to establish the hypocoercivity of the strongly continuous contraction semigroup (Tt)t0(T_t)_{t\geq 0} generated by the Kolmogorov operator. By using the refined abstract Hilbert space hypocoercivity method of Grothaus and Stilgenbauer, originally introduced by Dolbeault, Mouhot and Schmeiser, we construct a μΦ\mu^{\Phi}-invariant Hunt process with weakly continuous paths and infinite lifetime, whose transition semigroup is associated with (Tt)t0(T_t)_{t\geq 0}. This process provides a stochastically and analytically weak solution to the degenerate infinite dimensional stochastic Hamiltonian system with multiplicative noise. The hypocoercivity of (Tt)t0(T_t)_{t\geq 0} and the identification of (Tt)t0(T_t)_{t\geq 0} with the transition semigroup of the process leads to the exponential ergodicity. Finally, we apply our results to degenerate second order in time stochastic reaction-diffusion equations with multiplicative noise. A discussion of the class of applicable potentials and coefficients governing these equations completes our analysis.

Keywords

Cite

@article{arxiv.2410.15993,
  title  = {The essential m-dissipativity for degenerate infinite dimensional stochastic Hamiltonian systems and applications},
  author = {Benedikt Eisenhuth and Martin Grothaus},
  journal= {arXiv preprint arXiv:2410.15993},
  year   = {2024}
}