Effective branching splitting method under cost contraint

Agnes Lagnoux (1)

1 IMT - Institut de Mathématiques de Toulouse UMR5219

This paper deals with the splitting method first introduced in rare event analysis. In this technique, the sample paths are split into R multiple copies at various stages during the simulation. Given the cost, the optimization of the algorithm suggests to sample a number of subtrials which may be non-integer and even unknown but estimated. In this paper, we present three different approaches to face this problem which provide precise estimates of the relative error between P(A) and its estimator.

Mots clés: splitting method; simulation; cost function; Laplace transform; Galton-Watson; branching processes; iterated functions; rare event

Domaines: Probabilités [math.PR]

Agnes Lagnoux. Effective branching splitting method under cost contraint. Stochastic Processes and their Applications, 2008, 118 (10), pp.1820-1851. ⟨10.1016/j.spa.2007.10.009⟩. ⟨hal-00644141⟩