Recurrence and transience criteria for two cases of stable-like Markov chains (Q471514): Difference between revisions

The paper considers the recurrence and transience problems for two cases of time-homogeneous Markov chains on the real line. Their transition kernels have the form \(p(x,dy) = {f_x}(x - y)dy\), where \({f_x}(y)\) are probability densities of symmetric distributions. It is supposed that the function \(x \mapsto {f_x}(y)\) is a Borel measurable function for all \(y\) and \( {f_x}(y) \sim c(x){\left| y \right|^{ - \alpha (x) - 1}}\) for \(\left| y \right| \to \infty \). The corresponding Markov chain is said to belong to the class of stable-like Markov chains. The first case deals with a stable-like chain \({\{ X_n^{(\alpha ,\beta )}\} _{n \geqslant 0}}\) given by transition densities with characteristic functions \(\varphi (x;\xi ) = \exp ( - \gamma {\left| \xi \right|^\alpha }) \text{for}\;x < 0\), and \(\varphi (x,\xi ) = \exp ( - \delta {\left| \xi \right|^\beta }) \text{for}\;x \geqslant 0, \) where \(\alpha ,\beta \in (0,2)\) and \(\gamma ,\delta \in (0,\infty )\). The following theorem is proved: Theorem 1.1. The stable-like chain \({\{ X_n^{(\alpha ,\beta )}\} _{n \geqslant 0}}\) is recurrent if and only if \(\alpha + \beta \geqslant 2\). The second case concerns a chain \({\{ X_n^p\} _{n \geqslant 0}}\) with periodic function \(x \mapsto {f_x}(y)\). Under a uniformity condition on \({f_x}(y)\) and some mild technical conditions, the following result is proved: Theorem 1.2. If the set \(\{ x:\alpha (x) = {\alpha _0} = {\inf _x}\alpha (x)\} \) has positive Lebesgue measure, then the periodic stable-like chain \({\{ X_n^p\} _{n \geqslant 0}}\) is recurrent if and only if \({\alpha _0} \geqslant 1\).