Stats4Grads: Long Term Behaviour of Imprecise Markov Chains
2 May 2007 13:00 in CM221
Much work has been done regarding how Markov chains behave over time. Often we discuss the stationary distribution, or in the case of a Markov chain with a single absorbing state, we might discuss the quasi-stationary distribution. However, the vast bulk of such work makes the assumption that the Markov chain is time-homogeneous (that is that the transition probabilities do not vary over time). Further, practically all the work done assumes that at each step the transition probabilities are known precisely. In this talk we describe how the stationary distribution and quasi-stationary distribution might be generalised in situations where one or both of these assumptions no longer hold.
See the Stats4Grads page for more details about this series.