Think about defaults for simple EA parameters. about mlrmbo HOT 9 CLOSED

Let us revive this issue. In my opinion default parameters for evolutionary algorithms are not suited very well. I like the self-adaptive approaches. However, I am amenable for suggestions :-)

from mlrmbo.

Agreed. We can later test whether this improves the method.

from mlrmbo.

I think we can relegate to my ecr package soon.

from mlrmbo.

That would be great.

from mlrmbo.

Well, a default population size of 10 is quite small for >= 2D problems I guess. We could

• increase the default, but this is not the best solution IMHO
• set the default population size based on the number of parameters

I prefer the second approach.

from mlrmbo.

just do it and doc it

from mlrmbo.

See PR #154

from mlrmbo.

Done

from mlrmbo.

we can later deal with that as an ECR test case . with maybe self adaptation

from mlrmbo.