Topographical Differential Evolution Using Pre-calculated Differentials
M. Montaz Ali, Aimo Törn · Kluwer Academic Publishers eBooks · 2005
We present an algorithm for finding the global minimum of multimodal functions. The proposed algorithm is based on differential evolution (DE). Its distinguishing features are that it implements pre-calculated differentials and that it suitably utilizes topographical information on the objective function in deciding local search. These features are implemented in a periodic fashion. The algorithm has been tested on easy, moderately difficult test problems as well as on the difficult Lennard-Jones (LJ) potential function. Computational results using problems of dimensions upto 24 are reported. A robust computational behavior of the algorithm is shown.