Dual methods and hybrid algorithms for set partitioning problems

In this paper the authors study the behaviour of some dual methods in the design of hybrid algorithms for set partitioning (SP) problems. Dual techniques become very important to solve problems with a combinatorial structure, not only because they provide lower bounds but also because employing them, along with heuristics and cutting plane procedures in the hybrid algorithms, allows evaluating the quality of upper bounds. The dual methods that have been studied are surrogate and lagrangean relaxation and a variant of BISA’s dual strengthening method. Computational results obtained with these techniques in a hybrid algorithm for SP are reported.