A combined importance splitting and sampling algorithm for rare event estimation

Damien Jacquemart-Tomi (1) , Jérôme Morio (1) , François Le Gland (2)

1 ONERA - The French Aerospace Lab [Palaiseau]
2 ASPI - Applications of interacting particle systems to statistics

We propose some methodological basis for an improvement to the splitting method for a Markov process that evolves over a deterministic time horizon. Our algorithm is based on a decomposition of the selection functions that gives more importance to some well-chosen trajectories, typically those trajectories that manage to move earlier than others towards the critical region. Central limit theorem is established and numerical experiments are provided.

Keywords: SPLITTING; SAMPLING; RARE EVENT; STATISTICS

Domains: Engineering Sciences [physics]

Damien Jacquemart-Tomi, Jérôme Morio, François Le Gland. A combined importance splitting and sampling algorithm for rare event estimation. Proceedings of the 2013 Winter Simulation Conference, Washington 2013, Dec 2013, WASHINGTON, United States. pp.1035-1046, ⟨10.1109/WSC.2013.6721493⟩. ⟨hal-01058411⟩