Hi all, I hacked a new global optimizer that is based on Differential Evolution and a local search algorithm. I can make it to solve 1000+ dimensional problems on a cluster within a very reasonable error and time. It's using Akka's actor system, so it scales very well and very easily just by tweaking configuration files. I'm wondering if it is good enough to make it to Spark's ecosystem.
Need some feedbacks from Spark/Scala lovers :) You can find it here: https://github.com/EstusDev/Estus/tree/master/estus-optimization Details is documented in its wiki: https://github.com/EstusDev/Estus/wiki/SolverMOS Best, Shouheng
