When calling recursively nlls_iterate, "iteration 0" will call evaltensor_f witch triggers an expensive call to eval_hp regardless that the iterate, cost, (etc) do not change with respect to the outter iterate/iteration.
Is this the case?
If so, it would be convenient to avoid a call to eval_f (evaltensor_f) at iteration 0 of recursive call tonlls_iterateviasolve_tensor_newton`.
When calling recursively
nlls_iterate, "iteration 0" will callevaltensor_fwitch triggers an expensive call toeval_hpregardless that the iterate, cost, (etc) do not change with respect to the outter iterate/iteration.Is this the case? If so, it would be convenient to avoid a call to
eval_f(evaltensor_f) at iteration 0 of recursive call tonlls_iterateviasolve_tensor_newton`.