Perturbation results for comparison of Markov models

Abstract: Markov chains are useful to model various complex systems. In numerous situations, the underlying Markov chain is subject to changes. For example, states may be added or deleted and transition probabilities perturbed. It is therefore, necessary to ensure the robustness of the system and to estimate the resulting deviation in the characteristics. In this paper we study the sensitivity of finite Markov chains subject to changes in their state space and propose updating formulas and perturbation bounds