A multi dynamics algorithm for global optimization

In this paper a new global optimization algorithm for real valued functions, MAGO, is introduced. MAGO (Multi Dynamics Algorithm for Global Optimization) has been inspired by ideas from Estimation of Distribution Algorithms, Differential Evolution Algorithms and Statistical Quality Control. MAGO makes use of three different population dynamics: a changing uniform distribution to explore the searching space, a mechanism of propagating differences related to the best individuals, and a strategy for diversity preservation. Only the population size and the number of generations must be provided by the final user. The algorithm’s success in achieving global optima in the presence of multimodality has been shown through its application on a set of standard test functions.