Efficient Operations On MDDs for Building Constraint Programming Models / 374
Guillaume Perez, Jean-Charles Régin
We propose improved algorithms for defining the most common operations on Multi-Valued Decision Diagrams (MDDs): creation, reduction, complement, intersection, union, difference, symmetric difference, complement of union and complement of intersection. Then, we show that with these algorithms and thanks to the recent development of an efficient algorithm establishing arc consistency for MDD based constraints (MDD4R), we can simply solve some problems by modeling them as a set of operations between MDDs. We apply our approach to the regular constraint and obtain competitive results with dedicated algorithms. We also experiment our technique on a large scale problem: the phrase generation problem and we show that our approach gives equivalent results to those of a specific algorithm computing a complex automaton.