Fuzzy Recombination for the Breeder Genetic Algorithm

A new recombination operator called fuzzy recombination FR is introduced for contin uous genes The performance of the operator is analyzed by means of the equation describ ing the response to selection The operator is evaluated according to a new design cri terion maximizing the product of heritabil ity and standard deviation The breeder ge netic algorithm BGA with FR converges lin early for a test suite of benchmark functions The computational complexity is also com puted We believe that linear convergence is the optimum to be achieved by random search methods The question remains open Can a random search method be found which gives the best linear convergence i e the smallest constant for a well de ned class of functions