Gravitational search algorithm: R is better than R2?
Mohamad Nizam Aliman, Khairul Hamimah Abas, Muhammad Sharfi Najib, Nor Azlina Ab. Aziz, Mohd Saberi Mohamad, Zuwairie Ibrahim · Siti Hasmah Digital Library-MMU Institutiona Repository (Multimedia University) · 2016
Gravitational Search Algorithm (GSA) is a metaheuristic population-based optimization alg orithm inspired by the Newtonian law of gravity and law of motion. Ever since it was introd uced in 2009, GSA has been employed to solve various optimization problems. Despite its superior performance, GSA has a f undamental problem. It has been revealed that the force calculation in GSA is not genuinely based on the Newtonia n law of gravity. Based on the Newtonian law of gravity, force between two masses in the universe is inversely pr oportional to the square of the distance between them. However, in the original GSA, R is used instead of R 2 . In this paper, the performance of GSA is re-evaluated considerin g the square of the distance between masses, R 2 . The CEC2014 benchmark functions for real-parameter single objec tive optimization problems are employed in the evaluation. An important finding is that by considering the square of the distance between masses, R 2 , significant improvement over the original GSA is observed provid ed a large gravitational constant should be used at the beginning of the optimization process.