Speed up scipy functions performance using multi_processing/multi_threading

I have an idea about speed up functions performance (cpu bound) using either multiprocessing or multi-threading. According to the conversations i had with @rgommers , providing support for multiprocessing with scipy functions can be done. for example differential_evolution (using workers keyword) It would be nice if we can build a multiprocessor system that can generate parallelized instance of functions.

Thoughts?

[rgommers]

For something like khatri_rao there’s nothing to parallelize at the SciPy level, all it is is a few numpy calls.

[rgommers]

I’ll close this, since there’s no specific issue, we’ll put the general “we’d like performance improvements” on the roadmap.