Joris Gillis edited this page Apr 2, 2012 · 10 revisions

This page lists some experiences that users might want to share about the usage of ipopt.

Step 0

Even when max_iter == 0, it is not guaranteed that input(NLP_X_INIT) == output(NLP_X_OPT). Indeed if bounds on X or constraints are unmet, they will differ.


Ipopt conveniently offers a derivative_check. The point at which derivatives are check is formed by input(XLP_X_INIT) plus a pertubation. By default, this perturbation is large (10). You may wish to set Option point_perturbation_radius to 0 to avoid this perturbation.