Local stationarity and time-inhomogeneous Markov chains (Q2313278)

In the paper, a probabilistic framework is introduced which allows us to consider a large class of Markov chain models on arbitrary state spaces, including most of the locally stationary autoregressive processes. Let \((E,d)\) be a metric space with its corresponding Borel \(\sigma\)-field \(\mathcal{B}(E)\), and let \(\{Q_u, u\in[0,1]\}\) be a family of Markov kernels on \((E,\mathcal{B}(E))\). The author considers triangular arrays \(\{X_{n,j},n\in\mathbb{N}, 1\leqslant j\leqslant n\}\) such that for all \(n\), the sequence \(\{X_{n,1}, X_{n,2},\dots, X_{n,n}\}\) is a nonhomogeneous Markov chain under condition \[ \mathbb{P}(X_{n,k}\in A\,|\,X_{n,k-1}=x)=Q_{k/n}(x,A),\ k\in\{ 1,2,\dots, n\}. \] In particular, he derives an array of interesting properties related with the local stationarity of the described nonhomogeneous Markov chain. In order to satisfy various properties the generating Markov kernels family should satisfy some regularity conditions and some contraction assumptions.