AN AUGMENTED LAGRANGIAN-BASED PARALLEL SPLITTING METHOD FOR A ONE-LEADER-TWO-FOLLOWER GAME

摘要

In this paper, we exploit a new parallel splitting method for the typical structured variational inequality problems which can be interpreted as a game with a leader and two followers. In the framework of this method, two followers decide their strategies simultaneously based on the instruction of the leader. Then, the leader improves his instruction by revising his own variable value according to the feedback information from the followers. The convergence of the method is established under some suitable conditions. Finally, we apply the proposed method to solve some application problems. Computational studies show that the method is reliable and efficient.