Cost-Bounded Binary Decision Diagrams for 0-1 Programming

Abstract. In recent work binary decision diagrams (BDDs) were introduced as a technique for postoptimality analysis for integer programming. In this paper we show that much smaller BDDs can be used for the same analysis by employing cost bounding techniques in their construction. Binary decision diagrams (BDDs) have seen widespread application in logic circuit design and product configuration. They also have potential application to optimization, particularly to postoptimality analysis. A BDD can represent, often in compact form, the entire feasible set of an optimization problem. Optimal solutions correspond to shortest paths in the BDD. Due to the efficiency of this representation, one can rapidly extract a good deal of information about a problem and its solutions by querying the BDD. This opens the door to fast, in-depth postoptimality analysis without having to re-solve the problem repeatedly—provided the BDD is of manageable size. For instance, one can identify all optimal or ...

• Type: Conference
• Year: 2007
• Where: CPAIOR
• Authors: Tarik Hadzic, John N. Hooker
• Keywords: Binary Decision Diagrams, CPAIOR 2007, Near-optimal Solutions, Postoptimality Analysis