Abstract

We study determinacy, definability and complexity issues of path games on finite and infinite graphs. Compared to the usual format of infinite games on graphs (such as Gale-Stewart games) we consider here a different variant where the players select in each move a path of arbitrary finite length, rather than just an edge. The outcome of a play is an infinite path, the winning condition hence is a set of infinite paths, possibly given by a formula from S1S, LTL, or first-order logic. Such games have a long tradition in descriptive set theory (in the form of Banach-Mazur games) and have recently been shown to have interesting application for planning in nondeterministic domains. It turns out that path games behave quite differently than classical graph games. For instance, path games with Muller conditions always admit positional winning strategies which are computable in polynomial time. With any logic on infinite paths (defining a winning condition) we can associate a logic on graphs, defining the winning regions of the associated path games. We explore the relationships between these logics. For instance, the winning regions of path games with an S1S-winning condition are definable in the modal mu-calculus. Further, if the winning condition is first-order (on paths), then the winning regions are definable in monadic path logic, or, for a large class of games, even in first-order logic. As a consequence, winning regions of LTL path games are definable in CTL.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call