PatWie / CppNumericalSolvers

a lightweight header-only C++17 library of numerical optimization methods for nonlinear functions based on Eigen
MIT License
873 stars 201 forks source link

Max number of variables #99

Closed xsedla0k closed 6 years ago

xsedla0k commented 6 years ago

Hi there,

I have a question how many variable could be solved with your method(s).

I must optimize objective function with about +10 000 parameters, and i am worrying about low memory error. Thanks for answer. ;)

PatWie commented 6 years ago

This obviously depends on your machine and objective function. I guess that for 10000+ parameter it will take some time.

If the objective function is simply convex I suggest to have a look at simpler algorithm implemented in CUDA. Or relying on TensorFlow or PyTorch only which does all on the GPU (excl. scipy-opt)