A Survey of Meta-heuristics Used for Computing Maximin Latin Hypercube
Finding maximin latin hypercube is a discrete optimization problem believed to be NP-hard. In this paper, we compare different meta-heuristics used to tackle this problem: genetic algorithm, simulated annealing and iterated local search. We also measure the importance of the choice of the mutation operator and the evaluation function. All the experiments are done using a fixed number of evaluations to allow future comparisons. Simulated annealing is the algorithm that performed the best. By using it, we obtained new highscores for a very large number of latin hypercubes.
Unable to display preview. Download preview PDF.
- 1.Bates, S.J., Sienz, J., Toropov, V.V.: Formulation of the optimal latin hypercube design of experiments using a permutation genetic algorithm. AIAA 2011, 1–7 (2004)Google Scholar
- 5.Holland, J.H.: Adaptation in natural and artificial systems: An introductory analysis with applications to biology, control, and artificial intelligence. U. Michigan Press (1975)Google Scholar
- 6.Husslage, B., Rennen, G., Van Dam, E.R., Den Hertog, D.: Space-filling Latin hypercube designs for computer experiments. Tilburg University (2006)Google Scholar
- 11.Lourenço, H.R., Martin, O.C., Stützle, T.: Iterated local search. International series in operations research and management science, pp. 321–354 (2003)Google Scholar