Stable Dynamical Mass Transportation via port-Hamiltonian Structure

Abstract

Dynamical optimal mass transportation over nonlinear systems is often hindered by the curse of dimensionality and open-loop implementation. We propose an alternative framework that sacrifices exact optimality in favor of stability, robustness, and scalability. We use a stochastic port-Hamiltonian system as a reference model. Under suitable structural assumptions, we show that its density converges asymptotically to a prescribed steady-state distribution. We then shape a deterministic nonlinear system into a compatible port-Hamiltonian form and add a local density-feedback term so that its closed-loop continuity equation matches the Fokker-Planck-Kolmogorov equation of the reference model. The resulting feedback drives a broad class of initial distributions to the target distribution. Because the density term is local at the particle level, the controller admits decentralized implementation and avoids solving a global finite-horizon transport problem online. Nonlinear examples and simulations illustrate the method. For linear systems with Gaussian targets, we derive necessary and sufficient conditions for stabilizing density feedback and give explicit controller constructions.