7 - MAX-SAT and MAX-CSP

Abstract

The maximum satisfiability problem (MAX-SAT) and the maximum constraint satisfaction problem (MAX-CSP) are the optimization variant of SAT and CSP. These problems are theoretically and practically interesting, because they are among the conceptually simplest combinatorial optimization problems, yet instances of optimization problems from many application domains can be represented as MAX-SAT or MAX-CSP instances in an easy and natural way. SLS algorithms are among the most powerful and successful methods for solving large and hard MAX-SAT and MAX-CSP instances. This chapter introduces MAX-SAT and MAX-CSP. It presents some of the best-performing SLS algorithms for various types of MAX-SAT and gives an overview of results on their behavior and relative performance. Moreover, the chapter discusses SLS methods for solving the general problem as well as the closely related over constrained pseudo-Boolean and integer optimization problems.