Suppose a Markov chain leaves transient states and enters a closed two-state class which alternates deterministically. If entry is always into the first state, let and be the probabilities of entering at even and odd times. After the probability of still being transient tends to zero, the even-time probabilities at the two recurrent states tend to , while the odd-time probabilities tend to . Individual limits therefore exist exactly when . A periodic recurrent class can thus have convergent state probabilities from particular transient starting states even though convergence fails from its own states.
// Target: geometry-and-topology.bigb

Articles by others on the same topic (0)

There are currently no matching articles.