Semivariogram methods for modeling Whittle-Matérn priors in Bayesian inverse problems

DOI10.1088/1361-6420/AB762EMaRDI QIDQ5000601zbMATHOpenAlexFDO

Authors Richard Brown, Johnathan M. Bardsley, Tiangang Cui

Publication date 14 July 2021

Published in Inverse Problems (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1811.09446

zbMATH Keywords

stochastic partial differential equationsBayesian methodsinverse problemsvariogramboundary conditionsGaussian fieldWhittle-Matérn

Mathematics Subject Classification ID

Bayesian inference (62F15) Stochastic partial differential equations (aspects of stochastic analysis) (60H15) Image analysis in multivariate analysis (62H35) Ill-posedness and regularization problems in numerical linear algebra (65F22)

Abstract: We present a new technique, based on semivariogram methodology, for obtaining point estimates for use in prior modeling for solving Bayesian inverse problems. This method requires a connection between Gaussian processes with covariance operators defined by the Mat'ern covariance function and Gaussian processes with precision (inverse-covariance) operators defined by the Green's functions of a class of elliptic stochastic partial differential equations (SPDEs). We present a detailed mathematical description of this connection. We will show that there is an equivalence between these two Gaussian processes when the domain is infinite -- for us, $mathbb{R}^2$ -- which breaks down when the domain is finite due to the effect of boundary conditions on Green's functions of PDEs. We show how this connection can be re-established using extended domains. We then introduce the semivariogram method for estimating the Mat'ern covariance parameters, which specify the Gaussian prior needed for stabilizing the inverse problem. Results are extended from the isotropic case to the anisotropic case where the correlation length in one direction is larger than another. Finally, we consider the situation where the correlation length is spatially dependent rather than constant. We implement each method in two-dimensional image inpainting test cases to show that it works on practical examples.

Recommendations

• Whittle-Matérn priors for Bayesian statistical inversion with applications in electrical impedance tomography
• Hyperpriors for Matérn fields with applications in Bayesian inversion
• Discretization-invariant Bayesian inversion and Besov space priors
• Besov priors for Bayesian inverse problems
• Bayesian approach to inverse problems for functions with a variable-index Besov prior

Cites work

• scientific article; zbMATH DE number 3984248 (Why is no real title available?)
• scientific article; zbMATH DE number 3027842 (Why is no real title available?)
• scientific article; zbMATH DE number 3088518 (Why is no real title available?)
• An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach
• An introduction to computational stochastic PDEs
• Analysis of Boundary Effects on PDE-Based Sampling of Whittle--Matérn Random Fields
• Computational Methods for Inverse Problems
• Computational uncertainty quantification for inverse problems
• Constructing continuous stationary covariances as limits of the second-order stochastic difference equations
• Empirical Bayes spatial prediction using a Monte Carlo EM algorithm
• Fast and Exact Simulation of Stationary Gaussian Processes through Circulant Embedding of the Covariance Matrix
• Gaussian Markov Random Fields
• Green's functions and boundary value problems.
• Hyperpriors for Matérn fields with applications in Bayesian inversion
• ON STATIONARY PROCESSES IN THE PLANE
• On Fourier transforms of radial functions and distributions
• Statistical and computational inverse problems.
• Statistics for spatial data
• Studies in the history of probability and statistics XLIX: On the Matérn correlation family
• Ten equivalent definitions of the fractional Laplace operator
• Whittle-Matérn priors for Bayesian statistical inversion with applications in electrical impedance tomography

Cited in

(7)

• Optimization-Based Markov Chain Monte Carlo Methods for Nonlinear Hierarchical Statistical Inverse Problems
• Efficient algorithms for Bayesian Inverse Problems with Whittle--Mat\'ern Priors
• Hybrid projection methods for large-scale inverse problems with mixed Gaussian priors
• The Matérn model: a journey through statistics, numerical analysis and machine learning
• Hyperpriors for Matérn fields with applications in Bayesian inversion
• Whittle-Matérn priors for Bayesian statistical inversion with applications in electrical impedance tomography
• Efficient iterative methods for hyperparameter estimation in large-scale linear inverse problems

Describes a project that uses

Uses Software

• UTV
• GMRFLib

This page was built for publication: Semivariogram methods for modeling Whittle-Matérn priors in Bayesian inverse problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5000601)