Tuning of the optimization strategies on the problem size

This paper is concerned with the problem of adapting the optimization algorithms to the dimension of the problem at the hand. As test case, the problem of designing the structure of a magnetic resonance imaging system is considered with different number of optimization variables. A greedy genetic algorithm for continuous variables electromagnetic optimization problems, already presented in literature, has been modified to consider different problem sizes. A detailed comparison of performances obtained implementing several strategies and tuning of optimization parameters is eventually presented.