Prove optimization of nonconvex function is NP-hard