Bounded Normal Approximation in Highly Reliable Markovian Systems

Bruno Tuffin (1)

1 MODEL - Modeling Random Systems

In this paper, we give a necessary and sufficient condition to perform a good normal approximation for the Monte Carlo evaluation of highly reliable Markovian systems. We have recourse to simulation because of the frequent huge state space in practical systems. Literature has focused on the property of bounded relative error. In the same way, we can focus on bounded normal approximation. We see that the set of systems with bounded normal approximation is (strictly) included in the set of systems with bounded relative error.

Mots clés: SIMULATION; NORMAL APPROXIMATION; MARKOV CHAINS; HIGHLY RELIABLE SYSTEMS

Domaines: Autre [cs.OH]

Bruno Tuffin. Bounded Normal Approximation in Highly Reliable Markovian Systems. [Research Report] RR-3020, INRIA. 1996. ⟨inria-00073674⟩