Parallel Genetic Algorithm for the Capacitated Lot-Sizing Problem

A parallel genetic algorithm is presented to solve the well-known ca-pacitated lot-sizing problem. The approach is implemented on a massively parallel single instruction multiple data architecture with 16384 4-bit processors. Based on a random keys representation a schedule is backward oriented obtained which enables us to apply a very simple capacity check. Genetic algorithms are a general purpose optimization technique inspired by population genetics. The elds in which genetic algorithms are used range from operations research problems 17] and learning classiier systemss7], 11] to training neural networks 3]. For a detailed introduction to genetic algorithms see e.g. 9] or 15]. A genetic algorithm models the development of a population over a number of generations as it happens in nature. The understanding is that in nature the 1