preprint

Inserted: 24 feb 2025

Year: 2025

Abstract:

This is the first of a series of papers aimed at constructing a general theory of optimal transport to describe the time evolution of complex systems of interacting particles at the kinetic level. We argue that this theory may have applications also to biology (RNA sequencing and cell development), computer vision (continuous interpolation of images), and engineering (optimal steering). Firstly, we define our main object: a new second-order discrepancy between probability measures, analogous to the $2$-Wasserstein distance, but based on the minimisation of the squared acceleration. Secondly, we prove existence for the associated Kantorovich problem, and, under suitable conditions, of an optimal transport map. We also provide an equivalent time-continuous characterisation (Benamou--Brenier formula), where the admissible curves of measures are solutions to an adapted class of Vlasov's continuity equations. Thirdly, we define a class of absolutely continuous curves of measures and prove that they coincide exactly with time reparametrisations of solutions to Vlasov's equations.