Figure 5.

The comparison of convergence speed for four migration strategies in BBODE. In Figure 5, (A) four kinds of migration strategies (i.e., cosine model, quadratic model, exponential model, linear model) were tested by griewank function with the dimension 5. From the fitness curve in (A), quadratic model has the better convergence speed and accuracy. In (B), four kinds of migration strategies were tested by rosenbrock function with the dimension 2. From (B), considering convergence speed and accuracy, we can conclude that cosine model has a superior performance. Similarly, in (C–F), four kinds of migration strategies were tested by rastrigin function with the dimension 5, sphere function with the dimension 10, shaffer function with the dimension 5 and ackley function with the dimension 5, respectively. From (C–F), compared with quadratic, exponential, linear models, the convergence speed and accuracy of cosine model remarkably illustrate its excellent performance. Summarizing, cosine model is a superior migration strategy in BBODE algorithm.