A Modified Electromagnetism-Like Algorithm Based on a Pattern Search Method
The Electromagnetism-like (EM) algorithm, developed by Birbil and Fang (J Global Optim 25(3):263–282, 2003) is a population-based stochastic global optimization algorithm that uses an attraction-repulsion mechanism to move sample points towards optimality. A typical EM algorithm for solving continuous bound constrained optimization problems performs a local search in order to gather information for a point, in the population. Here, we propose a new local search procedure based on the original pattern search method of Hooke and Jeeves, which is simple to implement and does not require any derivative information. The proposed method is applied to different test problems from the literature and compared with the original EM algorithm.
KeywordsLocal Search Pattern Search Local Search Procedure Trial Point Pattern Search Method
- 1.Birbil SI (2002) Stochastic global optimization techniques. PhD Thesis, North Carolina State UniversityGoogle Scholar
- 3.Dixon LCW, Szegö GP (1978) The global optimization problem: an introduction. In: Dixon LCW, Szegö GP (eds) Towards Global Optimisation 2. North-Holland, Amsterdam, pp 1–15Google Scholar