Ensemble Kalman Inversion for nonlinear problems: weights, consistency, and variance bounds

Zhiyan Ding, Qin Li, Jianfeng Lu

Published 2020-03-04Version 1

Ensemble Kalman Inversion (EnKI), originally derived from Enseble Kalman Filter, is a popular sampling method for obtaining a target posterior distribution. It is, however, inconsistent when the forward map is nonlinear. Important Sampling (IS), on the other hand, ensures consistency at the expense of large variance of weights, leading to slow convergence of high moments. We propose a WEnKI, a weighted version of EnKI in this paper. It follows the same gradient flow as that of EnKI with a weight correction. Compared to EnKI, the new method is consistent, and compared with IS, the method has bounded weight variance. Both properties will be proved rigorously in this paper. We further discuss the stability of the underlying Fokker-Planck equation. This partially explains why EnKI, despite being inconsistent, performs well occasionally in nonlinear settings. Numerical evidence will be demonstrated at the end.