An adaptive importance sampling methodology is proposed to compute the mult
idimensional integrals encountered in reliability analysis. It is based on
a Markov simulation algorithm due to Metropolis et al. (Metropolis, Rosenbl
uth, Rosenbluth and Teller, Equations of state calculatons by fast computin
g machines. Journal of Chemical Physics, 1953;21(6): 1087-1092). In the pro
posed methodology, samples are simulated as the states of a Markov chain an
d are distributed asymptotically according to the optimal importance sampli
ng density. A kernel sampling density is then constructed from these sample
s which is used as the sampling density in an importance sampling simulatio
n. The Markov chain samples populate the region of higher probability densi
ty in the failure region and so the kernel sampling density approximates th
e optimal importance sampling density for a large variety of shapes of the
failure region. This adaptive feature is insensitive to the probability lev
el to be estimated. A variety of numerical examples demonstrates the accura
cy, efficiency and robustness of the methodology, (C) 1999 Elsevier Science
Ltd. All rights reserved.