Skip to content

Evolutionary Computing 2017, Vrije Universiteit Amsterdam.

Notifications You must be signed in to change notification settings

sunsistemo/evolutamine

Repository files navigation

evolutamine

Evolutionary Computing 2017, Vrije Universiteit Amsterdam

An evolutionary algorithm (EA) to maximise three continuous optimisation problems in 10 dimensions within a restricted evaluation budget: Bent Cigar, Katsuura and Schaffers F7.

A fitness function is provided that scales the performance of the algorithm between 0 and 10. A score of 0 means that the algorithm scores as good as a random search. A value of 10 means that the algorithm found the global optimum.

The search space is [-5,5]^10 (meaning that every variable must have a value between [-5,5]). One can check whether the functions have the following properties to adapt the EA in the right way: whether it is multimodal, whether it is separable, whether it has a strong structure and the number of available evaluations.