WorksMathematics
A Graph Theory Assisted Method for Determining Limiting Probability Distributions of Higher Order Markov Chains
Abstract
This paper presents a graph theory assisted method that checks a sufficient condition for the existence of the limiting probability distribution of a higher order Markov chain and identifies this distribution when the condition is satisfied. This method is more effective as compared to the direct verification of the condition since the latter hinges wholly on computations of the full spectrum on a large-size transition matrix. Even though the graph theoretic tools employed here are well known, their utilization in the context of a higher order Markov chain is new. This method is also implemented as an easy-to-use MATLAB function gtam. Several illustrative examples are provided too.