A Berry-Esseen theorem for Feynman-Kac and interacting particle models (Q1774203)

Feynman-Kac distribution flows and their particle interpretations represent a useful technique for a variety of applications, which include the spectral analysis of Schroedinger operators, modeling particle absorption, directed polymer simulations in biology and chemistry, nonlinear filtering in signal processing, rare event estimation in telecommunication etc. The present article extends earlier results of the authors concerning limiting variance computations in terms of Feynman-Kac semigroups in the context of particle density models. Here the study is focused on the speed of convergence of fluctuations. A new approach is proposed, which is based on Berry-Esseen type estimates for abstract martingale sequences, combined with original exponential concentration estimates of interacting processes. The results extend the classical theory and apply to a class of branching and genealogical path-particle models from statistical physics and biology.