Mean Field Games with Diverse Behavioral Patterns, Informational Constraints, and Learning

Abstract

This research program will address several fundamental problems in noncooperative nonzero-sum dynamic games with asymmetric information across a heterogeneous group of players (agents), with diverse behavioral patterns and misaligned objectives, beliefs and perceptions, operating under only partial or no modeling information and not even sharing a common probabilistic outlook, and with possibly multiple layers in decision making. The focus will be on stochastic dynamic games with a large population of networked players, and study of the precise relationship between such games and the corresponding ones with an infinite population (that is, mean-field games), such as the extent to which the mean-field equilibria (MFE) obtained for the latter (under various game-theoretic solution concepts, such as Nash, Stackelberg-Nash, team-optimal, or saddle-point, as appropriate for the underlying game) provide approximate equilibria of a similar type for the former.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 06, 2025
Source ID
FA95502410152

Entities

People

  • Tamer Başar

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of Illinois Urbana–Champaign

Tags

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Game Theory.
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.