Prior normalization for certified likelihood-informed subspace detection of Bayesian inverse problems

Abstract

Markov chain Monte Carlo (MCMC) methods form one of the algorithmic foundations of Bayesian inverse problems. The recent development of likelihood-informed subspace (LIS) methods offers a viable route to designing efficient MCMC methods for exploring high-dimensional posterior distributions via exploiting the intrinsic low-dimensional structure of the underlying inverse problem. However, existing LIS methods and the associated performance analysis often assume that the prior distribution is Gaussian. This assumption is limited for inverse problems aiming to promote sparsity in the parameter estimation, as heavy-tailed priors, e.g., Laplace distribution or the elastic net commonly used in Bayesian LASSO, are often needed in this case. To overcome this limitation, we consider a prior normalization technique that transforms any non-Gaussian (e.g. heavy-tailed) priors into standard Gaussian distributions, which makes it possible to implement LIS methods to accelerate MCMC sampling via such transformations. We also rigorously investigate the integration of such transformations with several MCMC methods for high-dimensional problems. Finally, we demonstrate various aspects of our theoretical claims on two nonlinear inverse problems.

References in corpus (14)

• The pseudo-marginal approach for efficient Monte Carlo computations
• Dimension-independent likelihood-informed MCMC
• Likelihood-informed dimension reduction for nonlinear inverse problems
• Sampling Can Be Faster Than Optimization
• Transport map accelerated Markov chain Monte Carlo
• Convergence properties of pseudo-marginal Markov chain Monte Carlo algorithms
• Scalable posterior approximations for large-scale Bayesian inverse problems via likelihood-informed parameter and state reduction
• Solving Large-Scale PDE-constrained Bayesian Inverse Problems with Riemann Manifold Hamiltonian Monte Carlo
• A TV-Gaussian prior for infinite-dimensional Bayesian inverse problems and its numerical implementations
• Data-Free Likelihood-Informed Dimension Reduction of Bayesian Inverse Problems
• Learning non-Gaussian graphical models via Hessian scores and triangular transport
• Nonlinear dimension reduction for surrogate modeling using gradient information
• Simple Nonlinear Models with Rigorous Extreme Events and Heavy Tails
• Gradient-based data and parameter dimension reduction for Bayesian models: an information theoretic perspective

Cited by in corpus (3)

• On the representation and learning of monotone triangular transport maps
• Analysis of sloppiness in model simulations: unveiling parameter uncertainty when mathematical models are fitted to data
• Optimal low-rank posterior mean and distribution approximation in linear Gaussian inverse problems on Hilbert spaces