We observe that the magnitude of~$\nabla \obj(\bar{\iter}) + \nabla h(\bar{\iter})^{\T} \bar{\lm}$ affects the quality of the model in general, as indicated by the error estimation in \cref{thm:sqp-path}.
-->
We observe that a smaller magnitude of~$\nabla \obj(\bar{\iter}) + \nabla h(\bar{\iter})^{\T} \bar{\lm}$ implies a better accuracy of the model, as indicated by the error estimation in \cref{thm:sqp-path}.
https://github.com/ragonneau/phd-thesis/blob/9da0514a4cf45b0bd2bd4b68d0af06b05af2d56b/content/sqp.tex#L366
-->