Title: Central force optimisation: a new gradient-like metaheuristic for multidimensional search and optimisation

Authors: Richard A. Formato

Addresses: P.O. Box 1714 Harwich, MA 02645, USA

Abstract: This paper introduces central force optimisation, a novel, nature-inspired, deterministic search metaheuristic for constrained multidimensional optimisation in highly multimodal, smooth, or discontinuous decision spaces. CFO is based on the metaphor of gravitational kinematics. The algorithm searches a decision space by |flying| its |probes| through the space by analogy to masses moving through physical space under the influence of gravity. Equations are developed for the probes| positions and accelerations using the gravitational metaphor. Small objects in our universe can become trapped in close orbits around highly gravitating masses. In |CFO space| probes are attracted to |masses| created by a user-defined function of the value of an objective function to be maximised. CFO may be thought of in terms of a vector |force field| or, loosely, as a |generalised gradient| methodology because the force of gravity can be computed as the gradient of a scalar potential. The CFO algorithm is simple and easily implemented in a compact computer program. Its effectiveness is demonstrated by running CFO against several widely used benchmark functions. The algorithm exhibits very good performance, suggesting that it merits further study.

Keywords: central force optimisation; CFO; optimisation; multidimensional search; deterministic metaheuristics; generalised gradient; bio-inspired computation; nature-inspired computing; evolutionary algorithms; gravitational kinematics.

DOI: 10.1504/IJBIC.2009.024721

International Journal of Bio-Inspired Computation, 2009 Vol.1 No.4, pp.217 - 238

Published online: 28 Apr 2009 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article