Generalised Reachability Games Revisited

Abstract: Classic reachability games on graphs are zero-sum games, where the goal of one player, Eve, is to visit a vertex from a given target set, and that of other player, Adam, is to prevent this. Generalised reachability games, studied by Fijalkow and Horn, are a generalisation of reachability objectives, where instead of a single target set, there is a family of target sets and Eve must visit all of them in any order. In this work, we further study the complexity of solving two-player games on graphs with generalised reachability objectives. Our results are twofold: first, we provide an improved complexity picture for generalised reachability games, expanding the known tractable class from games in which all target sets are singleton to additionally allowing a logarithmic number of target sets of arbitrary size. Second, we study optimisation variants of generalised reachability with a focus on the size of the target sets. For these problems, we show intractability for most interesting cases. Particularly, in contrast to the tractability in the classic variant for singleton target sets, the optimisation problem is NP-hard when Eve tries to maximise the number of singleton target sets that are visited. Tractability can be recovered in the optimisation setting when all target sets are singleton by requiring that Eve pledges a maximum sized subset of target sets that she can guarantee to visit.