Subspace Diffusion Posterior Sampling for Travel-Time Tomography

TL;DR

This paper introduces a subspace diffusion generative model for travel-time tomography, improving reconstruction quality and sampling speed in PDE-based inverse problems through dimension reduction and adjoint-state equations.

Contribution

It proposes a novel posterior sampling method using adjoint-state equations and employs subspace dimension reduction to accelerate PDE-based inverse problem solving.

Findings

Enhanced travel-time imaging quality

Reduced sampling time for reconstructions

Effective dimension reduction technique

Abstract

Diffusion models have been widely studied as effective generative tools for solving inverse problems. The main ideas focus on performing the reverse sampling process conditioned on noisy measurements, using well-established numerical solvers for gradient updates. Although diffusion-based sampling methods can produce high-quality reconstructions, challenges persist in nonlinear PDE-based inverse problems and sampling speed. In this work, we explore solving PDE-based travel-time tomography based on subspace diffusion generative models. Our main contributions are twofold: First, we propose a posterior sampling process for PDE-based inverse problems by solving the associated adjoint-state equation. Second, we resorted to the subspace-based dimension reduction technique for conditional sampling acceleration, enabling solving the PDE-based inverse problems from coarse to refined grids. Our numerical experiments showed satisfactory advancements in improving the travel-time imaging quality and reducing the sampling time for reconstruction.