The Impact of Implied Constraints on MaxSAT B2B Instances

Abstract

The B2B scheduling optimization problem consists of finding a schedule of a set of meetings between pairs of participants, minimizing their number of idle time periods. Recent works have shown that SAT-based approaches are state-of-the-art on this problem. One interesting feature of such approaches is the use of implied constraints. In this work, we provide an experimental setting to study the impact of using these implied constraints in MaxSAT B2B instances. To this purpose and due to the reduced number of existing real-world B2B instances, we propose a random B2B instance generation model, which reproduces certain features of these problems. In our experimental analysis, we show that the impact of using some implied constraints in the MaxSAT encodings depends on the characteristics of the problem, and we also analyze the benefits of combining them. Finally, we give some insights on how a MaxSAT solver is able to exploit these implied constraints.