In Markov chain theory a stochastic matrix $P$ is regular if some matrix power $P^n$ contains only strictly positive elements. Regularity of transition matrix of a Markov chain guarantees the existence of a unique invariant distribution which is also the limiting distribution. In the present paper a similar result is shown for the generalized Markov chain model that replaces classical probabilities with interval probabilities. We generalize the concept of regularity and show that for a regular interval transition matrix sets of probabilities corresponding to consecutive steps of a Markov chain converge to a unique limiting set of distributions that only depends on transition matrix and is independent of the initial distribution. A similar convergence result is also shown for approximations of the invariant set.
Keywords. Markov chains, interval probabilities
Paper Download
The paper is availabe in the following formats:
Authors addresses:
Kardeljeva ploscad 5
1000 Ljubljana
E-mail addresses:
| Damjan Škulj | damjan.skulj@fdv.uni-lj.si |