Efficient, Multimodal, and Derivative-Free Bayesian Inference With Fisher-Rao Gradient Flows

TL;DR

This paper introduces GMKI, a derivative-free Bayesian sampling method using Fisher-Rao gradient flows, capable of efficiently handling large-scale, multi-modal inverse problems without requiring gradient evaluations.

Contribution

It develops a novel framework combining Fisher-Rao flows, Gaussian mixtures, and Kalman updates for efficient, derivative-free sampling in complex Bayesian inverse problems.

Findings

Achieves rapid convergence to target distributions.

Effectively captures multiple modes with Gaussian mixtures.

Demonstrates success in large-scale Navier-Stokes inverse problem.

Abstract

In this paper, we study efficient approximate sampling for probability distributions known up to normalization constants. We specifically focus on a problem class arising in Bayesian inference for large-scale inverse problems in science and engineering applications. The computational challenges we address with the proposed methodology are: (i) the need for repeated evaluations of expensive forward models; (ii) the potential existence of multiple modes; and (iii) the fact that gradient of, or adjoint solver for, the forward model might not be feasible. While existing Bayesian inference methods meet some of these challenges individually, we propose a framework that tackles all three systematically. Our approach builds upon the Fisher-Rao gradient flow in probability space, yielding a dynamical system for probability densities that converges towards the target distribution at a uniform exponential rate. This rapid convergence is advantageous for the computational burden outlined in (i). We apply Gaussian mixture approximations with operator splitting techniques to simulate the flow numerically; the resulting approximation can capture multiple modes thus addressing (ii). Furthermore, we employ the Kalman methodology to facilitate a derivative-free update of these Gaussian components and their respective weights, addressing the issue in (iii). The proposed methodology results in an efficient derivative-free sampler flexible enough to handle multi-modal distributions: Gaussian Mixture Kalman Inversion (GMKI). The effectiveness of GMKI is demonstrated both theoretically and numerically in several experiments with multimodal target distributions, including proof-of-concept and two-dimensional examples, as well as a large-scale application: recovering the Navier-Stokes initial condition from solution data at positive times.