Action Refinement Applied to Late Decisions

Abstract

Abstract In modular approaches to specify concurrent systems a system is built up from components using various operators as e.g. the sequential, the parallel, or the choice (+) operator. Usually the choice between two components, i.e. P 1 + P 2 , is taken in favor of that component that is first to start an action. We follow here the alternative view that a choice is taken in favor of that component that is the first to terminate an action (end-based choice). This alternative has various applications and interesting implications. In particular the different points of view lead to different action refinement operators. The contribution of this paper is to present here an action refinement operator for the end-based view in a suitable true concurrency setting and to establish two equivalences that are the coarsest congruences for this refinement operator with respect to trace (respectively bisimulation) equivalence.