Evolutionary algorithms with a coarse-to-fine function smoothing

A coarse-to-fine function smoothing method is presented for improving traditional evolutionary algorithms in function optimization. The method is motivated by the need of suppressing the local optima without distorting the location of the global optimum. By embedding the method into a traditional evolutionary algorithm, optimization is performed by running the evolutionary algorithm in each smoothed function from coarse level to fine level in which the output of the coarse level is used to guide the search process in the finer level. Simulations show that the method can improve traditional evolutionary algorithms in locating the global optimum of a given function