A Hybrid Global Optimization Algorithm Involving Simplex and Inductive Search
Chetan P. Offord, Željko Bajzer · Lecture notes in computer science · 2001
We combine the recently proposed inductive search and the Nelder-Mead simplex method to obtain a hybrid global optimizer, which is not based on random searches. The global search is performed by line minimizations (Brent’s method) and plane minimizations (simplex method), while the local multidimensional search employs the standard and a modified simplex method. Results for the test bed of the Second International Competition of Evolutionary Optimization and for another larger test bed show remarkable success. The algorithm was also efficient in minimizing functions related to energy of protein folding.