Partial Information, Knowledge, and Uniform Strategies

We present two versions of alternating-time temporal logics with uniform strategies, interpreted in a novel way on transition systems for modelling situations in which agents with partial information interact to determine the way the system updates. The logic uATL allows us to model what properties agents can enforce when they act according to strategies based on their knowledge, but we show that agents’ knowledge cannot always be modelled based on the state of the system, so we also introduce the history based logic euATL which is stronger than uATL and allows us to model agents’ knowledge based on the history of a system’s execution. Apart from the usual memoryless strategies, we distinguish state-based memory, where agents recall the history of previous states, from perfect recall, where agents also recall their actions. We show that this makes a difference. Our logics both include three strategic operators for groups, representing the case where all the agents in the group cooperate actively, but do not share their knowledge, the case where some agents in the group may be passive, and the case where all the agents in the group share their knowledge. We include a detailed comparison to the literature on the subject.