Ensemble Kalman filter based sequential Monte Carlo sampler for sequential Bayesian inference

Abstract Many real-world problems require one to estimate parameters of interest, in a Bayesian framework, from data that are collected sequentially time. Conventional methods for sampling posterior distributions, such as Markov chain Monte Carlo cannot efficiently address they do not take advantage the data’s sequential structure. To this end, which seek update distribution whenever new collection become available often used solve these types problems. Two popular choices method ensemble Kalman filter (EnKF) and sampler (SMCS). While EnKF only computes Gaussian approximation distribution, SMCS can draw samples directly posterior. Its performance, however, depends critically upon kernels used. In work, we present constructs using an formulation, demonstrate performance with numerical examples.