Sampling per mode for rare event simulation in switching diffusions (Q432507)

A switching diffusion process \(X_t\) is considered which is described by a stochastic differential equation with coefficients depending on a Markov jump process. The problem is to estimate the probabilities connected with reaching a critical region by \(X_t\) in the case when it is a rare event. A Monte Carlo technique with multilevel splitting is considered for this purpose. Using the Feyman-Kac flows and interacting particles systems theory, the authors establish a law of large numbers and a central limit theorem for their estimate in the case when the number of particles tends to infinity. It is demonstrated that the proposed adaptive algorithm of particles resampling improves the asymptotic variance of the estimate.