An empirical comparison of simulated annealing and genetic algorithms on NK fitness landscapes

Abstract

The paper presents the features of GAs as static optimization techniques through an empirical comparison with SA on the NK fitness landscape model. This is done with consideration of the problem size and the ruggedness. Experimental results show that the genetic search is not comparable to simulated annealing on the NK fitness landscapes. Furthermore, the performance gap gets larger as the problem size increases and the landscape becomes rugged.