GA=PSO: Realizing Equivalent Particle Swarm Optimization by Genetic Algorithm

Evolutionary computation algorithms (EC) are a kind of powerful stochastic optimization approaches inspired by biological phenomena, such as natural selection, genetic information transmission, and intelligent swarm behaviors. After years of development, EC has many branches. Among them, the most classical one is the genetic algorithm (GA), which is also the earliest EC method. Though a lot of novel EC methods have been developed in the literature, they actually are all on the foundation of GA. However, little attention is put on the research on the connection between GA and the other EC algorithms. To fill this gap, this paper takes an in-depth study to investigate the relation between GA and the particle swarm optimization algorithm (PSO). By analyzing the operators in GA and PSO, this paper designs a new GA by devising novel crossover and mutation operators to simulate the evolution of particles in PSO. In this way, this paper gives the first transformation of GA to PSO and demonstrates that PSO is actually a special version of GA. Experiments have conducted on the CEC2017 single-objective benchmark set. Experimental results show that the designed new GA achieves very similar performance with the classical PSO.