Variational Optimality of F\"ollmer Processes in Generative Diffusions

TL;DR

This paper introduces a variational framework for F"ollmer processes in generative diffusions, enabling simulation-free estimation and optimal tuning of diffusion coefficients for improved probabilistic modeling.

Contribution

It provides a new variational characterization of F"ollmer processes, linking them to path-space KL divergence minimization and offering a data-driven, simulation-free estimation method.

Findings

F"ollmer process minimizes relative entropy among diffusions with the same interpolation schedule.

Optimal diffusion coefficient makes path-space KL divergence schedule-independent.

Numerical experiments demonstrate benefits in forecasting and data assimilation.

Abstract

We construct and analyze generative diffusions that transport a point mass to a prescribed target distribution over a finite time horizon using the stochastic interpolant framework. The drift is expressed as a conditional expectation that can be estimated from independent samples without simulating stochastic processes. We show that the diffusion coefficient can be tuned \emph{a~posteriori} without changing the time-marginal distributions. Among all such tunings, we prove that minimizing the impact of estimation error on the path-space Kullback--Leibler divergence selects, in closed form, a F\"ollmer process -- a diffusion whose path measure minimizes relative entropy with respect to a reference process determined by the interpolation schedules alone. This yields a new variational characterization of F\"ollmer processes, complementing classical formulations via Schr\"odinger bridges and stochastic control, and provides a conditional-expectation representation of the F\"ollmer drift that enables simulation-free estimation from data. We further establish that, under this optimal diffusion coefficient, the path-space Kullback--Leibler divergence becomes independent of the interpolation schedule, rendering different schedules statistically equivalent in this variational sense. We provide numerical experiments to illustrate the impact of path-space variational optimality of F\"ollmer's processes in probabilistic forecasting and data assimilation applications.