Iterative Methods for Determining Derivatives of Stationary Distributions of Finite Markov Chains
Jose J. Cruz, Shaler Stidham · 2021
We present algorithms for obtaining derivatives of steady-state probability distributions of finite Markov chains. Our focus is on a conceptually simple approach for developing such algorithms from established methods of computing steady state probability distributions. We briefly discuss our prototype (an application of the approach to the power method) and present the results of applying the approach to LOPSI (lopsided iteration) and Neut’s matrix geometric methods.