On the \(f\)-norm ergodicity of Markov processes in continuous time (Q727859): Difference between revisions

Let \(\{\Phi_t\}_{t\geq0}\) be a non-explosive Borel right process defined on a Polish space \(X\), and let \(f\geq1\) be a measurable function on \(X\). Under the assumptions that \(\{\Phi_t\}_{t\geq0}\) is \(\psi\)-irreducible and aperiodic, it has been shown in [\textit{S. P. Meyn} and \textit{R. L. Tweedie}, Adv. Appl. Probab. 25, No. 3, 518--548 (1993; Zbl 0781.60053)] that the following two properties are equivalent: {\parindent=0.7cm \begin{itemize}\item[(i)] \(\{\Phi_t\}_{t\geq0}\) admits an invariant probability measure \(\pi\) satisfying \(\int_Xf(x)\pi(dx)<\infty.\) \item[(ii)] There exists a closed small set \(C\subseteq X\) such that \[ \sup_{x\in C}\mathbb E^x \left[\int_0^{\tau_C(1)}f(\Phi_t)dt\right]<\infty, \] where \(\tau_C(1):=\inf\{t\geq1:\Phi_t\in C\}.\) \end{itemize}} Furthermore, in the same paper it has also been shown that (ii) implies: {\parindent=0.7cm \begin{itemize} \item[(iii)] There exist a closed small set \(C\subseteq X\), a constant \(b\in\mathbb{R}\) and a measurable nonnegative function \(V\) on \(X\) satisfying \(V(x_0)<\infty\) for some \(x_0\in X\) (Lyapunov function), such that \(\mathcal{D}V\leq-f+b1_C,\) where \(\mathcal{D}\) denotes the extended generator of \(\{\Phi_t\}_{t\geq0}\). \end{itemize}} In the paper under review, the authors show that (ii) and (iii) are actually equivalent. Furthermore, they also show that if the function \(V\) in (iii) is everywhere finite, then there is a constant \(C_f\in\mathbb{R}\) such that \[ \int_0^\infty\| \mathbb{P}^x(\Phi_t\in dz)-\mathbb{P}^y(\Phi_t\in dz)\|_fdt\leq C_f(1+V(x)+V(y)). \] If, in addition, \(\int_XV(x)\pi(dx)<\infty,\) then there is a constant \(C_f\in\mathbb{R}\) such that \[ \int_0^\infty\| \mathbb{P}^x(\Phi_t\in dz)-\pi(dz)\|_fdt\leq C_f(1+V(x)). \]