On the effect of step width selection schemes on the performance of stochastic local search strategies
Lars Nolle · 2004
This paper examines and discusses the effects of different step width selection schemes on the effectiveness of stochastic local search strategies. It is shown that the success of these search strategies depends, among other things, heavily on the chosen neighbourhood definition for a particular application. Simulations have shown that the use of randomly selected steps with a defined upper limit increases the probability of finding the global optimum, in contrast of using a constant step width. It has also been demonstrated that the efficiency of stochastic local search strategies can be improved by reducing the maximum step width over time.