Abstract
This article adopts a direct approach to systematically solve a class of linear quadratic social team and mixed game problems with one major agent and numerous minor agents in the presence of mean field interactions. For the team problem, the major and minor agents minimize a social cost which equals the sum of all agents’ costs. However, for the mixed game problem, there is a conflict of interest between the major agent and the other minor agents (the minor agents still work as a team). A direct approach is applied to solve these two problems, and the corresponding decentralized optimal controls are achieved by applying the maximum principle. This method can keep our analysis in a low-dimensional environment after directly decoupling a high-dimensional forward–backward stochastic differential equation in the team problem. The proofs of asymptotic social optimality and $\epsilon$ -Nash equilibrium are also provided, respectively. Finally, we summarize the differences between the team and mixed game problems, and a numerical simulation is presented.

This publication has 40 references indexed in Scilit: