Translating alloy using Boolean circuits

Abstract

Alloy is a automatically analyzable modelling language based on first-order logic. An Alloy model can be translated into a Boolean formula whose satisfying assignments correspond to instances in the model. Currently, the translation procedure mechanically converts each piece of the Alloy model individually into its most straightforward Boolean representation. This thesis proposes a more efficient approach to translating Alloy models. The key is to take advantage of the fact that an Alloy model contains patterns that are used repeatedly. This makes it natural to give a model a more structured Boolean representation, namely a Boolean circuit. Reusable pieces in the model correspond to circuit components. By identifying the most frequently used components and optimizing their corresponding Boolean formulas, the size of the overall formula for the model would be reduced without significant additional work. A smaller formula would potentially decrease the time required to determine satisfiability, resulting in faster analysis overall.