New Heuristics and Adaptive Memory Procedures for Boolean Optimization Problems
Arne Løkketangen, Fred Glover, Lars Magnus Hvattum · Operations research series · 2005
We describe new constructive and iterative search methods for Boolean Optimization Problems. Extending previous work by the authors, we describe the use of adaptive clause weights and probabilistic move acceptance for the adaptive memory search. We also describe how the use of the concepts persistent attractiveness measures and marginal conditional validity as search guidance mechanisms lead to very good results. Computational tests on portfolios of benchmark problems taken from the literature disclose that our method obtains results that improve on those previously published. 1