Low Rank Independence Samplers in Bayesian Inverse Problems

TL;DR

This paper introduces a computationally efficient low-rank independence sampler for high-dimensional Gaussian distributions in Bayesian inverse problems, improving sampling efficiency in ill-posed scenarios.

Contribution

It proposes a novel low-rank approximation-based independence sampler for Bayesian inverse problems, reducing computational costs in high-dimensional settings.

Findings

Acceptance rate depends on the number of eigenvalues retained.

The sampler performs well in image deblurring, CT, and NMR relaxometry.

High acceptance rates are achievable under certain conditions.

Abstract

In Bayesian inverse problems, the posterior distribution is used to quantify uncertainty about the reconstructed solution. In practice, Markov chain Monte Carlo algorithms often are used to draw samples from the posterior distribution. However, implementations of such algorithms can be computationally expensive. We present a computationally efficient scheme for sampling high-dimensional Gaussian distributions in ill-posed Bayesian linear inverse problems. Our approach uses Metropolis-Hastings independence sampling with a proposal distribution based on a low-rank approximation of the prior-preconditioned Hessian. We show the dependence of the acceptance rate on the number of eigenvalues retained and discuss conditions under which the acceptance rate is high. We demonstrate our proposed sampler by using it with Metropolis-Hastings-within-Gibbs sampling in numerical experiments in image deblurring, computerized tomography, and NMR relaxometry.