MMODELYST
Papers/Two-Parameter Flows for Learning Population Dynamics of Physical Systems
PAP

Two-Parameter Flows for Learning Population Dynamics of Physical Systems

May 25, 2026

arXiv
Abstract

This work addresses the problem of learning the dynamics of high-dimensional probability densities over time using unlabeled samples, without assuming access to trajectory information. We introduce two-parameter flows that learn only sampling-time transports from a base distribution to each marginal and then extract a physics-time velocity by regressing on coupled synthetic trajectories. We prove that the resulting physics-time dynamics are unique and inherit regularity from the sampling-time transports. Because we can build on standard, well-developed conditional flow matching techniques for learning the base-to-marginal transports, our approach scales to high dimensions and avoids per-step optimal-transport couplings, while allowing admissible non-gradient dynamics that can naturally explain rotational or circulating physics phenomena.

Select text to highlight · click a highlight to remove · saved in this browser only
Authors
Paul Schwerdtner, Tobias Blickhan, Benjamin Peherstorfer
Your notes (browser-local)
saved
arXiv:2605.26285