stop_criterion requires knowledge of optima

[anandtrex]

Which is not known in most cases. It should be defined as the delta change in fitness where the optimization is stopped.

[maharjun]

The thing is even that criterion requires knowledge of the optimizee fitness scale. To really have an optimizee independent solution, we probably will need to find a solution to the fitness scaling issue.

[anandtrex]

I thought we already had the solution, with bounding_func also taking care of scaling after being appropriately renamed. I think @dsalaj is working on this.

[maharjun]

but bounding_func only has access to the parameters of an individual, what is responsible for scaling the fitness (apart from the scaling parameter which is again optimizee dependent)?

[anandtrex]

True, the fitness is not being scaled. But knowing the scale of the fitness is much much more likely than knowing the actual optima. Since, in many cases, the user defines the fitness.