Evolutionary Global Optimization Survival of the Fittest Algorithm

This research presents the improved evolutionary Survival of the Fittest algorithm for global optimization. The algorithm uses a measure of probability concentration near the best solution to control the search process. It is strictly established in the paper that the sequence of points generated by the algorithm approaches the global optimum with the probability of unity. The improvements focus on enhancing the crossover operation, generalizing convergence conditions, and increasing overall algorithm efficiency in finite-dimensional spaces. Through testing with classical multidimensional and randomly generated functions, this method demonstrates superior performance compared to other evolutionary algorithms. Additionally, comparative analysis is conducted to select optimal hyperparameters for the method. The improved Survival of the Fittest algorithm offers a promising approach to efficiently solving complex optimization challenges across various domains.