ICB-DCM/NOODLES

No Hessian

yannikschaelte opened this issue · 1 comments

To save time and also space for large-dimensional problems (i.e. make the algorithm efficient for large dimension), add option to use and store no hessian. This is useful if the subproblem solver does not need the Hessian, like for (perturbed) gradient descent.

implemented, and adapted in subproblems.