Activity-local symbolic state graph generation for high-level stochastic models

This paper introduces a new, efficient method for deriving compact symbolic representations of very large (labelled) Markov chains resulting from high-level model specifications such as stochastic Petri nets, stochastic process algebras, etc.. This so called “activity-local” scheme is combined with a new data structure, called zero-suppressed multi-terminal binary decision diagram, and a new efficient “activityoriented” scheme for symbolic reachability analysis. Several standard benchmark models from the literature are analyzed in order to show the superiority of our approach.