Central Force Optimization and NEOs – First Cousins?
Richard A. Formato
Central Force Optimization is a new deterministic multidimensional search and optimization algorithm based on the metaphor of gravitational kinematics. This paper describes CFO and suggests some possible directions for its future development. Because CFO is deterministic, it is computationally more efficient than stochastic algorithms and may lend itself well to “parameter tuning” implementations. But, like all deterministic algorithms, CFO is prone to local trapping. Oscillation in CFO’s Davg curve appears to be a reliable harbinger of trapping. And there seems to be a reasonable basis for believing that trapping can be handled deterministically using the theory of gravitationally trapped Near Earth Objects. Deterministic mitigation of local trapping would be a major step forward in optimization theory. Finally, CFO may be thought of as a “gradient-like” algorithm utilizing the Unit Step function as a critical element, and it is suggested that a useful, new derivative-like mathematical construct might be defined based on the Unit Step.
Keywords: CFO, Central Force Optimization, Gravitation, Deterministic, Multidimensional, Search, Optimization, Algorithm, Near Earth Object, NEO, Gradient-like.