Search papers, labs, and topics across Lattice.
This paper introduces a novel advective Fisher-Rao metric tailored for optimizing paths of probability measures governed by the continuity equation, demonstrating that it leads to optimal descent directions in these optimization tasks. The metric is derived from three distinct perspectives, including its emergence as the rescaled zero-noise limit of the Fisher-Rao metric and as the Hessian of the Benamou鈥揃renier action functional. Empirical results confirm that this new metric effectively fits probability densities better than traditional methods like the Gauss鈥揘ewton approach, which focuses on velocity fields.
The advective Fisher-Rao metric not only optimizes probability measure paths but also outperforms conventional methods in fitting probability densities.
A novel advective Fisher-Rao metric is introduced for optimization tasks on paths of probability measures governed by the continuity equation. This metric is shown to lead to optimal descent directions. It is then shown that this metric arises naturally from three different perspectives: As the rescaled zero-noise limit of the Fisher-Rao metric on path measures, as the expected value of the second variation of the Freidlin--Wentzell large deviation rate functional, and as the Hessian of the Benamou--Brenier action functional from dynamic optimal transport. We supplement this geometric construction with computational experiments. Here, we demonstrate empirically that the advective Fisher-Rao metric yields the desired optimal fitting of probability densities, whereas the Gauss--Newton method yields optimal fitting of velocity fields.
