Resumen de A Masked Hybrid Genetic Algorithm for the Quadratic Assignment Problem
This paper presents a Masked Hybrid Genetic Algorithm (MHGA) for the Quadratic Assignment Problem (QAP) and reports its computational behavior. The algorithm incorporates a hybrid design consisting of a Genetic Algorithm (GA) and a local search, the Tabu Search (TS). The GA is designed by generating the initial population of size x using a randomized construction heuristic. Chromosomes are then subject to the fitness function. The fittest are selected and submitted to the crossover operator. For this process, a new operator the Mask is introduced, which randomly selects k times the number of crossover points and therefore generates k parallel populations of chromosomes. This crossover scheme promotes diversity. Mutation is then applied and the TS is performed onto the k populations. Finally, the best x chromosomes are reconsidered to form the next generation. The algorithm is simulated and tested on all instances of QAPLIB, the library of QAP. The best known solutions are obtained for approximately 80% of instances of different sizes and the solution found for the remaining 20% instances do not exceed a 1% gap to the best known solution. The computational testing shows that the MHGA developed behaves very well in terms of the quality of the solution.
