Compact Game Representations: Reasoning and

Computation A proliferation of compact representation schemes have been been proposed for various classes of games, along with efficient computations of solution concepts on such representations. The two main schemes have been Graphical games [7] and Multi-agent Influence Diagrams [14]. However, various other representations exist such as Game Networks [13] and Action-Graph Games [2]. This work will present a general overview of these different representation schemes and concentrate on two ideas associated with them; what kind of reasoning and visualization can be performed with each scheme, and what sort of computations can be made on them. We focus on correlated equilibria [1] calculations on such representations. The results of [12, 11], which provided a general framework for correlated equilibria computations on various suitable compact representations, will be extended to show a polynomial time algorithm to compute correlated equilibria on Action-Graph Games