Optimal transport over nonlinear systems via infinitesimal generators on graphs (Q1728239)

The paper deals with the problem of continuous-time optimal transport over nonlinear dynamical systems. The main aim is to compute controls that minimize the cost of transporting an initial measure to a final one over an assigned time-horizon. The solution of this problem is useful in the field of multi-agent systems or swarm control. The authors develop a set-oriented graph-based computational framework for this problem. In fact, using infinitesimal generators, the problem is reduced to a modified Monge-Kantorovich optimal transport problem, resulting in a convex Benamou-Brenier-type fluid dynamics formulation on a graph. They show that the well-posedness of the resulting problem on a graph is related to controllability of the underlying dynamical system and prove that the graph is strongly-connected if the underlying dynamical system is controllable. In conclusion, they prove that arbitrary final measures not lying on the boundary of the probability simplex can be reached in finite-time.