CMA-ES is very common approach for real-valued optimization problems. We can rely on a third part package for implementation of the actual CMA-ES where LEAP would essentially be a wrapper for that package.
Nick Hansen's cma package is as authoritative as it gets, and it has an ask() and tell() interface that should make it easy to wrap into a LEAP operator.
I suspect this is the same package that Jeff had demoed a proof-of-concept with a while back.
CMA-ES is very common approach for real-valued optimization problems. We can rely on a third part package for implementation of the actual CMA-ES where LEAP would essentially be a wrapper for that package.