AN ALTERNATIVE APPROACH FOR DIFFERENTIAL EVOLUTION ALGORITHM
The differential evolution algorithm is an effective method for global numerical optimization that is straightforward to understand, easy to implement, reliable, and efficient. As one of the evolutionary algorithms, Differential Evolution addresses global optimization problems by iteratively refining candidate solutions through an evolutionary process. It serves as a heuristic technique for minimizing potentially nonlinear and non-differentiable continuous functions. Global minimization algorithms require efficient computational time, which motivates the use of parallel computational approaches. This new technique utilizes multiple independent parallel computing units that occasionally share the best solutions they have discovered. This study proposes an improved approach designed to minimize the number of function calls, thereby strengthening the method's efficiency in exploring the objective function's search space. The proposed parallelizing Differential Evolution algorithm has been tested on several of optimization problems, and it appears from the experimental results that the executing time of the proposed parallelizing Differential Evolution becomes faster compared with Classic Differential Evolution and the other previous modifications of Differential Evolution. Also, results indicate significant improvements in function evaluations and solution quality compared to existing methodologies, demonstrating the effectiveness of the proposed approach.
Global Optimization, Metaheuristics, Parallel Computing, Differential Evolution Algorithm.