A Three Particles Algorithm for Solving Some Very High Dimensional Benchmark Unconstrained Optimization Problems Using a Fuzzy Fitness Criterium

Abstract

This work focuses predominantly on unconstrained optimization problems and presents an original algorithm (the code can be downloaded from Ref. 1), which is used for solving a variety of benchmark problems whose dimensions range from 2 to 2.5 millions, using only 3 particles. The algorithm was tested in 36 benchmark continuous unconstrained optimization problems, on a total of 312 instances. The results are presented comparing two fitness criteria: crisp and a fuzzy. The numerical results show that the proposed algorithm is able to reach the global optimum in every benchmark problem.