SAT Modulo Theories: Getting the Best of SAT and Global Constraint Filtering

Abstract

The propositional satisfiability problem (SAT) is one of the simplest instances of Constraint Programming (CP): variables are bi-valued (can only take values 0 or 1), and all constraints are clauses (disjunctions of literals) like $x \bigvee \bar{y} \bigvee \bar{z}$ (meaning that x = 1 or y = 0 or z = 0).

In spite of its simplicity, SAT has become very important for practical applications, especially in the multi-billion industry of electronic design automation (EDA), and, in general, hardware and software verification. Research on SAT has been pushed by these huge industrial needs and resources, in a very pragmatic way: prestigious conferences are eager to publish papers describing how to improve performance on their real-world problems, even if these improvements are not based on highly original techniques (in contrast with conferences like CP, which tend to prefer new ideas, even if they are tested only on academic random or artificial problem instances).