A Unified View on Hybrid Metaheuristics -- Packing Problems with Soft Rectangles -- A Multi-population Parallel Genetic Algorithm for Highly Constrained Continuous Galvanizing Line Scheduling -- Improvement in the Performance of Island Based Genetic Algorithms Through Path Relinking -- Using Datamining Techniques to Help Metaheuristics: A Short Survey -- An Iterated Local Search Heuristic for a Capacitated Hub Location Problem -- Using Memory to Improve the VNS Metaheuristic for the Design of SDH/WDM Networks -- Multi-level Ant Colony Optimization for DNA Sequencing by Hybridization -- Hybrid Approaches for Rostering: A Case Study in the Integration of Constraint Programming and Local Search -- A Reactive Greedy Randomized Variable Neighborhood Tabu Search for the Vehicle Routing Problem with Time Windows -- Incorporating Inference into Evolutionary Algorithms for Max-CSP -- Scheduling Social Golfers with Memetic Evolutionary Programming -- Colour Reassignment in Tabu Search for the Graph Set T-Colouring Problem -- Investigation of One-Go Evolution Strategy/Quasi-Newton Hybridizations.

The International Workshop on Hybrid Metaheuristics reached its third edition with HM 2006. The active and successful participation in the past editions was a clear indication that the research community on metaheuristics and related areas felt the need for a forum to discuss speci?c aspects of hybridization of metaheuristics. The selection of papers for HM 2006 consolidated some of the mainstream issues that have emerged from the past editions. Firstly, there are prominent examples of e?ective hybrid techniques whose design and implementation were motivated by challenging real-world applications. We believe this is particularly important for two reasons: on the one hand, researchers are conscious that the primary goal of developing algorithms is to solve relevant real-life problems; on the other hand, the path towarde?cient solving methods for practical problems is a source of new outstanding ideas and theories. A second important issue is that the research community on metaheur- tics has become increasingly interested in and open to techniques and methods known from arti?cial intelligence (AI) and operations research (OR). So far, the most representative examples of such integration have been the use of AI/OR techniques as subordinates of metaheuristic methods. As a historical and - ymological note, this is in perfect accordance with the original meaning of a metaheuristic as a “general strategy controlling a subordinate heuristic. ” The awareness of the need for a sound experimental methodology is a third keypoint.