An evolutionary strategy for global minimization and its Markov chain analysis

摘要: The mutation-or-selection evolutionary strategy (MOSES) is presented. goal of this to solve complex discrete optimization problems. MOSES evolves a constant sized population labeled solutions. dynamics employ mechanisms mutation and selection. At each generation, the best solution selected from current population. A random binomial variable N which represents number offspring by sampled. Therefore first solutions are replaced offspring, other replicas solution. relationships between convergence, parameters strategy, geometry problem theoretically studied. As result, explicit parametrizations proposed.