Posterior exploration for computationally intensive forward models

2024/05/01 by Lykkegaard, Mikkel B., Fox, Colin, Higdon, Dave +2
#Computation (stat.CO) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2405.00397

Abstract

In this chapter, we address the challenge of exploring the posterior distributions of Bayesian inverse problems with computationally intensive forward models. We consider various multivariate proposal distributions, and compare them with single-site Metropolis updates. We show how fast, approximate models can be leveraged to improve the MCMC sampling efficiency.

Related

• The Gaussian Latent Machine: Efficient Prior and Posterior Sampling for Inverse Problems
• Variance-based sensitivity of Bayesian inverse problems to the prior distribution
• POPE: Post Optimization Posterior Evaluation of Likelihood Free Models
• Bayesian Posterior Perturbation Analysis with Integral Probability Metrics
• Localization for MCMC: sampling high-dimensional posterior distributions with local structure