A central limit theorem for Fleming-Viot particle systems (Q2179253): Difference between revisions

Fleming-Viot-type particle systems are considered, which consist of independently moving particles that are killed on the boundary of a domain. At the time of death of a particle, another particle branches so that the particles population size in principle might be kept constant. In [\textit{M. Bieniek} et al., Probab. Theory Relat. Fields 153, No. 1--2, 293--332 (2012; Zbl 1253.60089)], a proof has been given of a generic non-extinction of the Fleming-Viot model. The fact that particle population does not approach the boundary simultaneously in a finite time, in some Lipschitz diomains, has been employed to prove a limit theorem for the empirical distribution of the particle family. An earlier paper [\textit{K. Burdzy} et al., J. Phys. A, Math. Gen. 29, No. 11, 2633--2642 (1996; Zbl 0901.60054)] has been devoted to the link of Fleming-Viot models with Laplacian eigenfunctions, where the link with the concept of quasi-stationary distributions naturally appears [\textit{P. Collet} et al., Quasi-stationary distributions. Markov chains, diffusions and dynamical systems. Berlin: Springer (2013; Zbl 1261.60002)]. Accordingly, Fleming-Viot type particle systems may be viewed as representing a classical way to approximate the distribution of a Markov process (not necessarily Brownian motion) with killing, given that it is still alive at a final deterministic time. The model follows a simple algorithm for systems of independently evolving particles. Namely, each particle is presumed to follow to the law of the underlying Markov process until its killing, and then branches instantaneously from the state of another randomly chosen particle. The consistency of this algorithm in the large population limit has been recently studied in several articles. The major purpose of the present paper is to prove central limit theorems under more general assumptions. For this, the key suppositions are that the particle system does not explode in finite time, and that the jump and killing times have atomless distributions. The convergence to a stationary distribution for large particle systems is thereby established. A comparative study is performed with sequential Monte Carlo simulations for discrete time algorithms.