Computational Discrete Time Markov Chain with Correlated Transition Probabilities

Abstract

This study presents a computational procedure for analyzing statistics of steady state probabilities in a discrete time Markov chain with correlations among their transition probabilities. The proposed model simply uses the first order Taylor’s series expansion and statistical expected value properties to obtain the resulting linear matrix equations system. Computationally, the bottleneck is O(n⁴) but can be improved by distributed and parallel processing. A preliminary computational experience is reported.