STAIRS: A NOVEL MULTIVARIATE OPTIMIZATION METHOD BASED ON A UNIVARIATE APPROACH

Niño Elias D., Garrido Johan, Encinales Luis. STAIRS: A Novel Multivariate Optimization Method Based On A Univariate Approach. Proceedings of the International Conference on Computer and Computational Intelligence, ASME, ISBN: 9780791859926, Bangkok – Thailand, December 2011.

ABSTRACT

This article proposes a novel method for multivariate optimization unconstrained named Stairs, based on optimization in one variable. The proposed method is compared against methods from the specialized literature such as the Multivariate Newton-Raphson and the Multivariate Fletcher-Powell. The instances of the problems were taken from real life situations. For a real comparison, metrics such as Number of Iterations, Number of Instructions and Processing time were taken into account. Stairs showed a speed improvement relative to the compared methods in problems that include difficult differentiation because it does not use matrix operations.

http://www.asme.org/products/books/international-conference-on-computer-and-computati