Talk by Peter Caines (McGill)

Time and place: Thursday, May 13, 10:40-11:40, 202 Chernoff Hall
Title: Mean Field Social Dynamics with Egoistic and Altruistic Agents: A Nash Game Formulation (1 hour)
Abstract: We study mean field LQG problems with cost coupling. The cost of each agent is a convex combination of its own cost and the social cost, where the weight assigned to the latter reflects the willingness of the agent in question to contribute to the social objective. We use the Nash Certainty Equivalence (NCE) approach to construct a set of decentralized strategies for the agents. The connection with the so-called social optimality theorem in a model of all altruistic agents will be discussed. Also, we examine the impact of the interaction strength and the cooperation willingness of the agents on the formation of a stable mean field behavior.

Andrew D. Lewis (andrew at mast.queensu.ca)