Dual-space posterior sampling for Bayesian inference in constrained inverse problems

2026/03/31 by Ali Siahkoohi, Kamal Aghazade, Ali Gholami · 5 citations
Computer Science · Mathematics · Physics and Astronomy · #cs.LG #physics.geo-ph #stat.ML

paper · pdf · doi:10.48550/arxiv.2603.00393

https://github.com/luqigroup/SVGDADMMSampler.jl

arxiv created 2026/08/05 · arxiv updated 2026/08/07

Abstract

Inverse problems constrained by partial differential equations are often ill-conditioned due to noisy, incomplete data or inherent non-uniqueness. A prominent example is full waveform inversion (FWI), which estimates Earth's subsurface properties by fitting seismic measurements subject to the wave equation, where ill-conditioning stems from noisy, band-limited, finite-aperture measurements and complex geological structures. A Bayesian framework describes the solution more comprehensively: instead of a single estimate, a posterior distribution of plausible solutions characterizes the non-uniqueness and can be sampled to quantify uncertainty. However, no clear procedure exists for translating hard physical constraints, such as the wave equation, into priors amenable to existing sampling techniques. We address this by sampling the posterior in the dual space via an augmented Lagrangian formulation, which converts hard constraints into penalties suited to sampling algorithms while enforcing them progressively through multiplier updates, so they are satisfied in the limit. We integrate the alternating direction method of multipliers (ADMM) with Stein variational gradient descent (SVGD), a particle-based sampler: the constraint is relaxed at each iteration and the multiplier updates progressively enforce its satisfaction. This enables posterior sampling under hard constraints while inheriting the favorable conditioning of dual-space solvers, where partial constraint relaxation permits productive updates even when the current model is far from the true solution. We validate the method on a stylized Rosenbrock conditional inference problem and on frequency-domain FWI for a Gaussian anomaly model and the Marmousi II benchmark, demonstrating physically consistent uncertainty estimates and posterior contraction with increasing data coverage.

Citations

• Automatic Penalty Parameter Selection by Residual Whiteness Principle (RWP) and GCV for Full Waveform Inversion
• Bayesian full waveform inversion with learned prior using deep convolutional autoencoder
• Full-waveform variational inference with full common-image gathers and diffusion network
• Fast and Automatic Full Waveform Inversion by Dual Augmented Lagrangian
• Conditional Image Prior for Uncertainty Quantification in Full Waveform Inversion
• ASPIRE: Iterative Amortized Posterior Inference for Bayesian Inverse Problems
• ASPIRE: iterative amortized posterior inference for Bayesian inverse problems
• Bayesian Variational Time-lapse Full-waveform Inversion
• Conditional score-based diffusion models for Bayesian inference in infinite dimensions
• From Optimization to Sampling Through Gradient Flows
• Reliable amortized variational inference with physics-based latent distribution correction
• Deep Bayesian inference for seismic imaging with tasks
• A dual formulation of wavefield reconstruction inversion for large-scale seismic inversion
• A Non-Asymptotic Analysis for Stein Variational Gradient Descent
• Weak deep priors for seismic imaging
• Uncertainty quantification in imaging and automatic horizon tracking: a Bayesian deep-prior based approach
• BayesFlow: Learning complex stochastic models with invertible neural networks
• Stein Variational Gradient Descent With Matrix-Valued Kernels
• Efficient MCMC Sampling with Dimension-Free Convergence Rate using\n ADMM-type Splitting
• Projected Stein Variational Newton: A Fast and Scalable Bayesian Inference Method in High Dimensions
• Improving full waveform inversion by wavefield reconstruction with the alternating direction method of multipliers
• Validating Bayesian Inference Algorithms with Simulation-Based Calibration
• Split-and-augmented Gibbs sampler - Application to large-scale inference problems
• Total-variation regularization strategies in full-waveform inversion
• Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithm
• Variational Inference: A Review for Statisticians
• Global Convergence of ADMM in Nonconvex Nonsmooth Optimization
• Variational Inference with Normalizing Flows
• A penalty method for PDE-constrained optimization in inverse problems
• Transport map accelerated Markov chain Monte Carlo
• Convergence Analysis of Alternating Direction Method of Multipliers for a Family of Nonconvex Problems
• A computational framework for infinite-dimensional Bayesian inverse problems. Part I: The linearized case, with application to global seismic inversion
• A review of the adjoint-state method for computing the gradient of a functional with geophysical applications
• On optimization techniques for solving nonlinear inverse problems
• Monte Carlo sampling of solutions to inverse problems
• Monte Carlo sampling methods using Markov chains and their applications