Smart Reduction

Abstract

Abstract.Compositional aggregation is a technique to palliate state explosion -the phenomenon that the behaviour graph of a parallel composition of asynchronous processes grows exponentially with the number of processes -which is the main drawback of explicit-state verification. It consists in building the behaviour graph by incrementally composing and minimizing parts of the composition modulo an equivalence relation. Heuristics have been proposed for finding an appropriate composition order that keeps the size of the largest intermediate graph small enough. Yet the underlying composition models are not general enough for systems involving elaborate forms of synchronization, such as multiway and/or nondeterministic synchronizations. We overcome this by proposing a generalization of compositional aggregation that applies to an expressive composition model based on synchronization vectors, subsuming many composition operators. Unlike some algebraic composition models, this model enables any composition order to be used. We also present an implementation of this approach within the Cadp verification toolbox in the form of a new operator called smart reduction, as well as experimental results assessing the efficiency of smart reduction.

Reduction and Abstraction Techniques for BIP

“…Several heuristics for selecting candidate components have been presented for LTS systems with no data transfer [19,9,10]. These heuristics consider merging pairs of components and favor the ones that result in smaller components.…”

“…Much work has been done on the automatic compositional reduction of communicating processes [1,10,9,19]. The techniques revolve around incrementally composing and minimizing individual components of an input system modulo an equivalence relation.…”

“…The work in [10] uses the concept of networks of LTSs introduced in [18] to support compositional reduction using different compositional operators. The authors use a heuristic similar to the ones presented in [9] to estimate the number of internal transitions that can be removed after the applying the composition operators, and compare the obtained metric for possible compositions.…”

“…The authors use a heuristic similar to the ones presented in [9] to estimate the number of internal transitions that can be removed after the applying the composition operators, and compare the obtained metric for possible compositions. Our technique differs from the work in [10] in that our transformations are solely targeted towards BIP systems, and need not be as general as the techniques presented in [10].…”

“…Given a BIP system consisting of several atomic components, our method automatically selects a pair of candidate components that have a high reduction potential based on their data dependencies and their synchronization. Our methods uses component selection heuristics similar to [19,9,10] and provides a user-defined selection API. Then, our method computes the product of the selected pair of atomic components which renders the interaction data transfer operations into transition data transfer operations.…”