This PR allows the user to keep only the best agents for each iteration instead of all information stored by all implemented meta-heuristics. Storing all this information can become a burden while performing extensive optimization tasks, such as hyper-complex optimization on higher-dimensional spaces.
Additionally the user can supply a custom function that is executed before evaluating the target function, allowing to alter the meta-heuristic behavior. I intend to add an example of this possible application quite soon.
This PR allows the user to keep only the best agents for each iteration instead of all information stored by all implemented meta-heuristics. Storing all this information can become a burden while performing extensive optimization tasks, such as hyper-complex optimization on higher-dimensional spaces.
Additionally the user can supply a custom function that is executed before evaluating the target function, allowing to alter the meta-heuristic behavior. I intend to add an example of this possible application quite soon.
Regards, Felix