Restricted Decision Diagrams

Abstract

This chapter presents a general-purpose methodology for obtaining a set of feasible solutions to a discrete optimization problems using restricted decision diagrams. A restricted diagram can be perceived as a counterpart of the concept of relaxed diagrams introduced in previous chapters, and represents an under approximation of the feasible set, the objective function, or both. We first show how to modify the top-down compilation approach to generate restricted diagrams that observe an input-specified width. Next, we provide a computational study of the bound provided by restricted diagrams, particularly focusing on the set covering and set packing problems.