Combining graph decomposition techniques and metaheuristics for solving PCSPs. Application to MI-FAP

摘要

This paper presents a study towards a framework for solving discrete optimisation problems modelled as partial constraint satisfaction problems (PCSPs). These studies follow two approaches, namely a bottom-up, and a top-down one. Three decomposition methods and an adaptive genetic algorithm (AGA) are associated with these approaches. The experimental results obtained for MI-FAP problems show a good trade-off between the quality of the solution and the execution time of the different algorithms.