Capacity of invariant measures related to Poisson-driven stochastic differential equations (Q4415728)

In the paper under review, a stochastic differential equation of the form \[ d\xi(t)= a(\xi(t)) dt+ \int_\Theta\sigma(\xi(\tau), \Theta)N_P(dt, d\Theta),\quad t\geq 0\tag{1} \] with the initial condition \(\xi(0)= \xi_0\), is considered. Here \((\xi(t))_{t\geq 0}\) is a stochastic process with values in a separable Banach space \(X\), \(a: \mathbb{R}_+\to X\) is a Lipschitz function, \(\sigma:X\times\Theta\to X\) is a perturbation term also satisfying the Lipschitz condition, \((\Theta,{\mathcal G},k)\) is a probability space and \(N_P\) is the Poisson random counting measure. The problem (1) defines a semigroup of Markov operators \((P^t)_{t\geq 0}\) acting on the space of Borel measures on \(X\). Let \({\mathcal M}\) be the family of all finite Borel measures on \(X\), \(M_1= \{\mu\in{\mathcal M}:\mu(X)= 1\}\). An operator \(P:{\mathcal M}\to{\mathcal M}\) is called nonexpansive with respect to the norm \(\|\cdot\|_q\), \(q\in (0,1]\), if \(\|P\mu_1- P\mu_2\|_q\leq \|\mu_1- \mu_2\|_q\), \(\mu_1,\mu_2\in{\mathcal M}\). An operator \(P:{\mathcal M}\to{\mathcal M}\) is called essentially nonexpansive if \(\exists q \in (0,1]\) such that \(P\) is nonexpansive with respect to the norm \(\|\cdot\|_q\). An operator \(P\) is called semi-concentrating if \(\forall\varepsilon> 0 \exists C\subset X\), contained in a finite union of balls of radius \(\varepsilon\), such that \(\inf_{\mu\in M_1} \liminf_{n\to\infty} P^n\mu(C)> 0\). The main results of the present paper are: 1. If a Markov operator \(P:{\mathcal M}\to{\mathcal M}\) is essentially nonexpansive and semi-concentrating, then \(P\) admits an invariant measure. 2. Let \(X\) be a separable Banach space. Under some hypotheses, the semigroup \((P^t)_{t\geq 0}\) admits a unique invariant distribution. 3. The lower capacity of a unique invariant measure with respect to the semigroup generated by the equation (1) is \(\geq 1\).