Continuous Optimization with two Evolutionary Strategies: Comma (mu, lambda) and Plus (eta

Abstract:

In the present work we have carried out an experimental study of two Evolutionary Strategies: Comma (mu, lambda) and Plus (mu + lambda) with Auto-Adaptive mutation, independent and uncorrelated steps for each variable, in an evolutionary algorithm that solves the optimization problem of the Sphere, minimizing the objective function to an optimal value of f((x) over right arrow) <= 10(-6); with the aim of know the required computational effort (measured in number of evaluations) to achieve the optimal value, and define significant differences between the results of both strategies. To the generation of new solutions, two types of recombinations were applied: Discreet and Arithmetic. The results indicated that mostly there is not significant differences between the two strategies on the different size of the problem, so based on the computational effort, the Plus strategy converges faster to the optimal value than …

Año de publicación:

2016

Keywords:

Fuente:

google