This was part of New Directions in Reinforcement Learning and Control

Controlled dynamical systems on the space of probability measures

Max Raginsky, University of Illinois at Urbana-Champaign