Particle filtering with path sampling and an application to a bimodal ocean current model

摘要: This paper introduces a recursive particle filtering algorithm designed to filter high dimensional systems with complicated non-linear and non-Gaussian effects. The method incorporates parallel marginalization (PMMC) step in conjunction the hybrid Monte Carlo (HMC) scheme improve samples generated by standard filters. Parallel is an efficient Markov chain (MCMC) strategy that uses lower approximate marginal distributions of target distribution accelerate equilibration. As validation tested on 2516 dimensional, bimodal, stochastic model motivated Kuroshio current runs along Japanese coast. results this test indicate attractive alternative for problems require generality but have been inaccessible due limitations strategies.