Importance Sampling for Multi-Constraints Rare Event Probability

Abstract

Improving Importance Sampling estimators for rare event probabilities requires sharp approximations of the optimal density leading to a nearly zero-variance estimator. This paper presents a new way to handle the estimation of the probability of a rare event defined as a finite intersection of subset. We provide a sharp approximation of the density of long runs of a random walk conditioned by multiples constraints, each of them defined by an average of a function of its summands as their number tends to infinity.