On the distribution and expectation of success runs in nonhomogeneous Markov dependent trials (Q1767306)

The author considers a nonhomogeneous two-state Markov chain and derives the one-dimensional probability mass function \(P(\eta_n= k)\) of length \(\eta_n\) \((n\geq 1)\) of the success run at the \(n\)th step of the chain. He then expresses, by means of \(\eta_j\) \((j= k\) \((k\geq 1),\dots, n+1)\), the number of success runs \(S_n(k)\) \((G_n(k))\) of size exactly \(k\) (greater than or equal to \(k\)) as the sum of dependent Bernoulli trials having different success probabilities, and derives its expectation. Finally, the author considers the longest success run \(L_n=\max_{k\leq j\leq n}\{\eta_j\}\) and derives the exact cumulative distribution function \(P(L_n< k)\) of \(L_n\). Upper and lower bounds are also given for \(P(L_n< k)\), and the results are illustrated by means of a numerical example.