Using Linearized Optimal Transport to Predict the Evolution of Stochastic Particle Systems

TL;DR

This paper introduces a novel Euler-type method based on linearized optimal transport to predict the evolution of probability measures in stochastic particle systems, reducing micro-scale computations while maintaining accuracy.

Contribution

It develops a macro-scale timestepper using linearized optimal transport that accurately predicts measure evolution without explicit operator learning, effective even with chaotic particle behavior.

Findings

Method achieves first-order accuracy for smooth measure evolution.

Algorithm reduces micro-scale simulation steps needed for prediction.

Validated on biological, PDE, and Langevin dynamics models.

Abstract

We develop an Euler-type method to predict the evolution of a time-dependent probability measure without explicitly learning an operator that governs its evolution. We use linearized optimal transport theory to prove that the measure-valued analog of Euler's method is first-order accurate when the measure evolves ``smoothly.'' In applications of interest, however, the measure is an empirical distribution of a system of stochastic particles whose behavior is only accessible through an agent-based micro-scale simulation. In such cases, this empirical measure does not evolve smoothly because the individual particles move chaotically on short time scales. However, we can still perform our Euler-type method, and when the particles' collective distribution approximates a measure that \emph{does} evolve smoothly, we observe that the algorithm still accurately predicts this collective behavior over relatively large Euler steps, thus reducing the number of micro-scale steps required to step forward in time. In this way, our algorithm provides a ``macro-scale timestepper'' that requires less micro-scale data to still maintain accuracy, which we demonstrate with three illustrative examples: a biological agent-based model, a model of a PDE, and a model of Langevin dynamics.