Optimal Coarse Correlated Equilibria in Mean Field Games: Linear Programming and No-Regret Learning
Researchers have introduced a new concept called optimal coarse correlated equilibria for continuous-time mean field games. This approach involves a moderator selecting equilibria that optimize a specific performance criterion, which may differ from the individual player's objective. The study presents a linear programming formulation for this problem, proves the existence of optimal equilibria, and designs a no-regret primal-dual algorithm for learning these equilibria, complete with convergence rates and numerical examples. AI
IMPACT Introduces novel theoretical frameworks for analyzing complex game dynamics, potentially applicable to multi-agent AI systems.