Estimating the failure probability of a Markovian system during regeneration period by the essential sampling method (Q1280942)

A repairable system is modelled by a discrete time Markov chain with state space \(\{1,\dots,N\}\); faulty states comprise \(F\subset\{2, \dots, N\}\). For large systems, exact calculation of \(q\), the probability that the chain with initial state 1 will hit \(F\) before returning to 1, is often cumbersome. Therefore attention has focussed on fast simulation techniques: in particular, the essential sampling method [\textit{P. Heidelberger}, ACM Trans. Model. Comput. Simul. 5, No. 1, 43-85 (1995; Zbl 0843.62096)] provides estimators of \(q\) through simulating a related Markov chain. This paper proposes an algorithm for creating the transition probabilities of the new chain. Its performance is quantified favourably both analytically, and numerically via an example from \textit{M. K. Nakayama} [Adv. Appl. Probab. 28, No. 3, 687-727 (1996; Zbl 0863.65095)], wherein previous schemes are cited.