General framework of Artificial Physics Optimization Algorithm
Liping Xie, Jianchao Zeng, Zhihua Cui · 2009
This paper presents a general framework of physics-inspired method named Artificial Physics Optimization (APO) Algorithm, a population-based, stochastic for multidimensional search and optimization. APO invokes a gravitational metaphor in which the force of gravity may be attractive or repulsive, the aggregate effect of which is to move particles toward local and global optima. APO's particles (solutions to the optimization problem) are treated as physical individuals, each individual has a mass, position and velocity. The mass of each individual corresponds to a user-defined function of the value of an objective function to be optimized. Responding to virtual forces, APO's individuals move toward other particles with larger “masses” (better fitnesses) and away from lower mass particles (worse fitnesses). Each individual attracts all others whose mass is lower, and repels all others whose mass is greater. The individual with the greatest mass (“best” individual) attracts all other individuals, and it is neither attracted to nor repelled by all the others. The attraction-repulsion rule causes APO's population to search regions of the decision space with better fitnesses. Experimental simulations show that APO is tested against several benchmark functions with better results.