Approximation of sojourn-times via maximal couplings: motif frequency distributions (Q2510394): Difference between revisions

Markov chains in Polish state spaces \(S\), which encompass discrete state spaces, are considered. Given a measurable set \(T \in S\), a sojourn time \(T_n\) is a random number of times a Markov chain visits \(T\) in its \(n\) first transitions. For relatively small values of \(n\), distributions of \(T_n\) can be obtained explicitly using generating functions. If \(n\) is large enough, then normal approximation or compound Poisson approximation are used to describe distributions of \(T_n\). In this article, approximations to sojourn-time distributions in Harris recurrent Markov chains are applied to such \(n\) that are too long for exact calculations, but too short to rely on normal or compound Poisson approximations. Distributions of \(T_n\) are approximated for \(n\) of order \((1-\alpha)^{-m}\), where \(m\) is the largest integer for which the exact distribution of \(T_m\) is accessible and \(0 \leq \alpha \leq 1\) is an ergodicity coefficient associated with the probability transition kernel of the chain.