The game-theoretic approach to multi-agent systems has been incorporated into the model-checking agenda by using temporal and dynamic logic to characterize notions such as Nash equilibria. Recent efforts concentrate on pure-strategy games, where intelligent agents act deterministically guided by utility functions. We build upon this tradition by incorporating stochastic actions. First, we present an extension of the Probabilistic Computation-Tree Logic (PCTL) to quantify and compare expected costs. Next, we give a discrete-time Markov chain codification for mixed-strategy games. Finally, we characterize mixed-strategy Nash equilibria.
展开▼
