Question on on num_samplers and num_evaluators and distributed approach

[fraspin]

Hi,

thanks a lot for providing this useful tool!

I have a question regarding the num_samplers and num_evaluators variables. For both, It is written in the code that "A value larger than 1 only has an effect when the samplers are able to execute in parallel, e.g. on different machines of a distributed system."

Could you explain a bit more in detail what you mean?

Also, if for instance, I have a cluster with 10 servers, how can I configure the funsearch application to use all of them?

[Rubiel1]

Hey, If the samplers are in a loop, only when one ai finishes will the second ai start. Instead you can just have one ai running continuously and the effect will be the same. So to see an improvement you want the samplers to work in parallel. As for the actual implementation on a cluster, I wonder the same.