Abstracts
Abstract
An Alternating Intensification/Diversification (AID) method is proposed to tackle global optimization problems, focusing here on global function minimization over continuous variables. Our method is a local search procedure that is particularly easy to implement, and can readily be embedded as a supporting strategy within more sophisticated methods that make use of population-based designs. We perform computational tests comparing the AID method to 20 other algorithms, many of them representing a similar or higher level of sophistication, on a total of 28 benchmark functions. The results show that the new approach generally obtains good quality solutions for unconstrained global optimization problems, suggesting the utility of its underlying notions and the potential value of exploiting its multiple avenues for generalization.
Keywords:
- Intensification/Diversification,
- global optimization,
- local search,
- scatter search,
- path relinking
Download the article in PDF to read it.
Download