Conditional Optimal Transport on Function Spaces

TL;DR

This paper develops a theoretical framework for conditional optimal transport maps in infinite-dimensional function spaces, with applications to Bayesian inference and functional parameter estimation.

Contribution

It generalizes optimal triangular transport theory to infinite-dimensional spaces and provides regularity estimates for Bayesian conditioning maps.

Findings

Theoretical development of triangular transport maps in function spaces.

Regularity estimates for prior-to-posterior maps in Bayesian inference.

Numerical experiments demonstrating applicability to likelihood-free inference.

Abstract

We present a systematic study of conditional triangular transport maps in function spaces from the perspective of optimal transportation and with a view towards amortized Bayesian inference. More specifically, we develop a theory of constrained optimal transport problems that describe block-triangular Monge maps that characterize conditional measures along with their Kantorovich relaxations. This generalizes the theory of optimal triangular transport to separable infinite-dimensional function spaces with general cost functions. We further tailor our results to the case of Bayesian inference problems and obtain regularity estimates on the conditioning maps from the prior to the posterior. Finally, we present numerical experiments that demonstrate the computational applicability of our theoretical results for amortized and likelihood-free inference of functional parameters.