Abstract Cyclic Communicating Processes: A Logical View

Abstract

Temporal logics have been very successful as tools for specifying and verifying dynamic properties of finite state distributed systems. An intriguing fact is that these logics, in order to be effective, must semantically filter out at least one of the two basic features of the behavior of distributed systems: indeterminacy and concurrency. To be precise, linear time temporal logics use as models the interleaved runs of a system in which both indeterminacy (i.e. the choices presented to the system) and concurrency (i.e. causally independent occurrences of actions) have been, in some sense, defined away. In branching time temporal logics, the (interleaved) runs glued together into a single object-usually called a computation tree-serves as a model. Here, speaking again loosely, indeterminacy is present through the branching nature of the computation tree but information regarding concurrency has been filtered out. There are also families of linear time temporal logics which are interpreted over the partially ordered runs (often represented as Mazurkiewicz traces) of a system. In this setting, concurrency is in but indeterminacy is out.