On a D-solution of a cooperative m-person discounted Markov game (Q1110464)

The authors consider cooperative m-person discounted Markov games with countable state space and compact metric action spaces. The authors introduce the concept of a D-solution (D for dominating) analogous to the domination concept for multiobjective decision problems [\textit{P. L. Yu}, J. Optimization Theory Appl. 14, 319-377 (1974; Zbl 0268.90057)]. For the existence of a D-solution, the authors show that the game may be replaced by a new decision process with a collective loss function. To this purpose side payments are introduced. Finally, the relation is treated between the D-solution of a game and the supergradient of a lower support function.