Invariant probabilities for Feller-Markov chains (Q1908326): Difference between revisions

A homogeneous Feller Markov chain \(x_t\) valued in a \(\sigma\)-compact metric space \(X\) is studied. It is shown that the existence of an invariant probability measure is equivalent to each of the following statements: (a) There exist a probability measure \(\nu\) and a moment \(v\) such that \(\limsup_{n\to\infty}n^{-1}E_{\nu}[\sum_{t=0}^{n-1}v(x_t)]<\infty\) ; (b) there exist a probability measure \(\nu\) and a moment \(v\) such that \(\limsup_{n\to\infty}(1-\alpha)E_{\nu}[\sum_{t=0}^{n-1}\alpha^tv(x_t)]<\infty\) , where a moment \(v\) means a nonnegative Borel function with \(\inf\{v(x)\mid x\notin K_n\}\uparrow\infty\) for some compact sets \(K_n\uparrow X\), as \(n\to\infty\). This weakens assumptions, such as strong Feller property or nondegeneracy or uniform countable-additivity, presumed in references.