Stopping rules effect on a derivative-free filter multistart algorithm for multilocal programming

Multilocal programming aims to identify all the local solutions of constrained optimization problems. The purpose of this paper is to analyze the effect of stopping rules on the performance of a particular multistart method, which relies on a derivative-free local search procedure to converge to a solution, when solving multilocal optimization problems. The method herein presented implements the approximate descent direction method combined with a filter methodology to handle the constraints by forcing the local search towards the feasible region. Two stopping rules are tested on five classical multimodal problems.

Stopping rules effect on a derivative-free filter multistart algorithm for multilocal programming Artigo de Conferência