An adaptive Gaussian process method for multi-modal Bayesian inverse problems (2409.15307v1)

Abstract: Inverse problems are prevalent in both scientific research and engineering applications. In the context of Bayesian inverse problems, sampling from the posterior distribution is particularly challenging when the forward models are computationally expensive. This challenge escalates further when the posterior distribution is multimodal. To address this, we propose a Gaussian process (GP) based method to indirectly build surrogates for the forward model. Specifically, the unnormalized posterior density is expressed as a product of an auxiliary density and an exponential GP surrogate. In an iterative way, the auxiliary density will converge to the posterior distribution starting from an arbitrary initial density. However, the efficiency of the GP regression is highly influenced by the quality of the training data. Therefore, we utilize the iterative local updating ensemble smoother (ILUES) to generate high-quality samples that are concentrated in regions with high posterior probability. Subsequently, based on the surrogate model and the mode information that is extracted by using a clustering method, MCMC with a Gaussian mixed (GM) proposal is used to draw samples from the auxiliary density. Through numerical examples, we demonstrate that the proposed method can accurately and efficiently represent the posterior with a limited number of forward simulations.