THE REPULSION ALGORITHM, A NEW MULTISTART METHOD FOR GLOBAL OPTIMIZATION

This paper proposes a new multistart algorithm to find the global mini mum of constrained problems. This algorithm, which in this paper is ca lled the repulsion algorithm, efficiently selects initial design point s for local searches. A Bayesian approach provides the stopping rules. The method uses information from the previous sampling points and the corresponding sequences generated by local searches to select new ini tial points. This approach increases the probability of finding all lo cal minima with fewer local searches. Numerical example problems show that compared with traditional multistart methods, the repulsion algor ithm reduces significantly the number of local searches required to fi nd the global minimum.