I'm trying to implement parallel evolutionary algorithm and I want to test it's performance. The best case is a heavy fitting function. What function should I choose for this test?
I've looked at multidimensional Rosenbrock function. But I wonder is there any heavier (for computation) function for such tests?
Try the Keano's bump m-dimensional function:
$$ f(\mathbf{x}) = - \left(\sum_1^m cos^4(x_i)- 2\prod_i^m cos^2(x_i) \right)/\left(\prod_i^m i x_i^2\right)^{0.5} $$
subject to
$$ 0.75 - \prod_i^m x_i < 0, \\ \sum_i^m x_i - 7.5m < 0,\\ 0 < x_i < 10 $$
Details can be found here: https://mpra.ub.uni-muenchen.de/3098/1/MPRA_paper_3098.pdf