Hybrid genetic algorithm for solving the p-median problem

Abstract

The p-median problem is an NP-complete combinatorial optimisation problem well investigated in the fields of facility location and more recently, clustering and knowledge discovery. We show that hybrid optimisation algorithms provide reasonable speed and high quality of solutions, allowing effective trade-off of quality of the solution with computational effort. Our approach to hybridisation is a tightly coupled approach rather than a serialisation of hill-climbers with genetic algorithms. Our hybrid algorithms use genetic operators that have some memory about how they operated in their last invocation.