Reachability in Fuzzy Game Graphs

Abstract

Two-player turn-based games on graphs (or game graphs for short) and their probabilistic versions have received increasing attention in computer science, especially in the formal verification of reactive systems. However, in the fuzzy setting, game graphs are yet to be addressed, although some practical applications, such as modeling fuzzy systems that interact with their environments, appeal to such models. To fill the gap, in this paper, we propose a fuzzy version of game graphs and focus on the fuzzy game graphs with reachability objectives, which we will refer to as fuzzy reachability games (FRGs). In an FRG, the goal of one player is to maximize her truth value of reaching a given target set, while the other player aims at the opposite. In this framework, we show that FRGs are determined in the sense that for every state, both of the two players have the same value, and there exist optimal memoryless strategies for both players. Moreover, we design algorithms, which achieve polynomial time complexity in the size of the FRG, to compute the values of all states and the optimal memoryless strategies for the players. For a special class of FRGs, we provide an improved algorithm that achieves linear-logarithmic running time to compute the values of states. In addition, several examples are given to illustrate our motivation and the theoretical development.