dso-org/deep-symbolic-optimization

Deap fitness objective

Closed this issue · 1 comments

DYSIM commented

Is there a reason why in Deap the fitness objective is to minimize

creator.create("FitnessMin", base.Fitness, weights=(-1.0,))

and the reward is set as -p.r , when is already possible to set FitnessMax from the start like

creator.create("FitnessMax", base.Fitness, weights=(1.0,))

which lets the reward be used as it is, without negating.

Thanks!

Nope! That would probably be cleaner. I’d do a parity test and if it passes them replace it as you suggest. Don’t think I knew about FitnessMax at the time.