IfP is a transition matrix of a Markov chain, and\(\tilde P\) is derived by perturbing the elements ofP, then we find conditions such that\(\tilde P\) is also positive recurrent whenP is, and relate the invariant probability measures for the two. Similar results are found for recurrence of chains, and the methods then yield analogues for continuous time processes also.
Seneta, E. (1968). Finite approximations to non-negative matrices II: refinements and applications,Proc. Camb. Phil. Soc.,64, 465–470.MathSciNetCrossRefGoogle Scholar
Takahashi, Y. (1973). On the effects of small deviations in the transition matrix of a finite Markov chain,J. Operat. Res. Soc. Japan,16, 104–129.MathSciNetzbMATHGoogle Scholar
Tweedie, R. L. (1973). The calculation of limit probabilities for denumerable Markov processes from infinitesimal properties,J. Appl. Prob.,10, 84–99.MathSciNetCrossRefGoogle Scholar
Tweedie, R. L. (1975). The robustness of positive recurrence and recurrence of Markov chains under perturbations of the transition probabilities.J. Appl. Prob.,12, 744–752.MathSciNetCrossRefGoogle Scholar