probability theory

McKean-Vlasov SDEs with Bounded Measurable Interaction

arXiv:2302.05845

summary

The paper studies McKean‑Vlasov stochastic differential equations whose drift interaction is only bounded and measurable, deriving a variation‑norm regularity estimate for the associated nonlinear semigroup and proving exponential ergodicity in Wasserstein‑1 distance for a class of such systems.

Abstract

In this paper, McKean-Vlasov SDEs with bounded measurable interaction is investigated. The regularity estimate $$\|P_t^\astγ^1-P_t^\astγ^2\|_{var}\leq ct^{-\frac{1}{2}}\W_{1}(γ^1,γ^2),\ \ t\in(0,T]$$ for the nonlinear semigroup associated to McKean-Vlasov SDEs is derived. Two cases are considered respectively. The first case concentrates on the model where the interaction in the drift is merely assumed to be bounded measurable while the distribution dependent diffusion term is allowed to be Lipschitz continuous under ()-Wasserstein distance in the measure variable. In the second case, the diffusion is distribution free and the drift contain two parts: a bounded measurable interaction term plus a partially dissipative term. As an application of the regularity estimate, the exponential ergodicity in $\W_1$ is obtained in the second case.

30 pages

Topics & keywords

#stochastic differential equations#mean-field interaction#regularity estimates#ergodicity#Wasserstein distanceMcKean-Vlasov SDEbounded measurable driftnonlinear semigroupvariation normexponential ergodicity
McKean-Vlasov SDEs with Bounded Measurable Interaction · wovepaper