Tightness and continuity of a family of invariant measures for Markov chains depending on a parameter (Q1358018)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Tightness and continuity of a family of invariant measures for Markov chains depending on a parameter |
scientific article |
Statements
Tightness and continuity of a family of invariant measures for Markov chains depending on a parameter (English)
0 references
16 March 1998
0 references
A family of Markov chains with transition functions \((P^{(\delta)}(x,B))_{\delta\in R}\) (\(x\in S\): metric space, \(B\in{\mathcal B}(S)\)) is considered. Every kernel \(p^{(\delta)}\) has at least one invariant distribution \(\pi^{(\delta)}\): \(\pi^{(\delta)}(B)= \int_S P^{(\delta)}(x,B)\pi^{(\delta)}(dx)\). Two conditions are proved to be sufficient for \(\pi^{(\delta)}\) to converge weakly to \(\pi^{(0)}\) \((\delta\to 0)\). They are the following: 1) for all \(x\in S\) under \(y\to x\), \(P^{(\delta)}(y,\cdot)\) tends weakly to \(P^{(0)}(x,\cdot)\), 2) the family \((\pi^{(\delta)})\) is tight in \(S\). For investigation of tightness of the family a method of Lyapunov test functions is used. These are some real functions with a special property of their increments on the chain. In case of \(S=R_+\) the tightness is connected with distributions of increments of the chain itself.
0 references
Markov kernel
0 references
weak convergence
0 references
Lyapunov test functions
0 references
increments
0 references