Computational analysis on a linkage between generalized logit dynamic and discounted mean field game

Abstract

Logit dynamics are dynamical systems describing transitions and equilibria of actions of interacting players under uncertainty. An uncertainty is embodied in logit dynamic as a softmax type function often called a logit function originating from a maximization problem subjected to an entropic penalization. This study provides another explanation for the generalized logit dynamic, particularly its logit function and player's heterogeneity, based on a discounted mean field game subjected to the costly decision making of a representative player. A large discount limit of the mean field game is argued to yield a logit dynamic. Further, mean field games that lead to classical and generalized logit dynamics are clarified and their well posedness is discussed. Additionally, numerical methods based on a finite difference discretization for computing generalized logit dynamics and corresponding mean field games are presented. Numerical methods are applied to two problems arising in the management of resources and environment; one involves an inland fisheries management problem with legal and illegal anglers, while the other is a sustainable tourism problem. Particularly, cases that possibly lack the regularity condition to be satisfied for the unique existence of stationary solutions are computationally discussed.