A Novel Differential Evolution Incorporated with Parallel Processing Mechanism
Dong Hyun Lim, Ngoc Hoang Luong, Chang Wook Ahn · 2010
In this paper, we introduce the parallel mechanism to the Differential Evolution (DE). Differential Evolution algorithm suffers from the problem that individuals gather in a single point and cannot escape the basin of a local optimum. We propose the parallel processing as a solution to this problem. Parallel processing helps DE maintain more than one best individual (attractor). Different attractors help DE expand its search towards various directions. We name our approach a differential evolution algorithm processed in the parallel fashion (P-DE). P-DE, when compared with the original DE in various test problems, exhibits its superior performance.