Graph-based Prior and Forward Models for Inverse Problems on Manifolds with Boundaries

Abstract

This paper develops manifold learning techniques for the numerical solution of PDE-constrained Bayesian inverse problems on manifolds with boundaries. We introduce graphical Matérn-type Gaussian field priors that enable flexible modeling near the boundaries, representing boundary values by superposition of harmonic functions with appropriate Dirichlet boundary conditions. We also investigate the graph-based approximation of forward models from PDE parameters to observed quantities. In the construction of graph-based prior and forward models, we leverage the ghost point diffusion map algorithm to approximate second-order elliptic operators with classical boundary conditions. Numerical results validate our graph-based approach and demonstrate the need to design prior covariance models that account for boundary conditions.

References in corpus (8)

• Galerkin Approximation of Dynamical Quantities using Trajectory Data
• Numerical solution of fractional elliptic stochastic PDEs with spatial white noise
• The rational SPDE approach for Gaussian random fields with general smoothness
• A finite area scheme for shallow granular flows on three-dimensional surfaces
• The Bayesian Formulation and Well-Posedness of Fractional Elliptic Inverse Problems
• Weak convergence of Galerkin approximations for fractional elliptic stochastic PDEs with spatial white noise
• Data-Driven Forward Discretizations for Bayesian Inversion
• Unlabeled Data Help in Graph-Based Semi-Supervised Learning: A Bayesian Nonparametrics Perspective

Cited by in corpus (2)

• Spectral methods for solving elliptic PDEs on unknown manifolds
• Spectral Convergence of Symmetrized Graph Laplacian on manifolds with boundary