An adaptive regularization algorithm for unconstrained nonconvex optimization is presented in which the objective function is never evaluated but only derivatives are used. This algorithm belongs to the class of adaptive regularization methods, for which optimal worst-case complexity results are known for the standard framework where the objective function is evaluated. It is shown in this paper that these excellent complexity bounds are also valid for the new algorithm despite the fact that significantly less information is used. In particular, it is shown that if derivatives of degree one to [Formula: see text] are used, the algorithm will find an [Formula: see text]-approximate first-order minimizer in at most [Formula: see text] iterations and an [Formula: see text]-approximate second-order minimizer in at most [Formula: see text] iterations. As a special case, the new algorithm using first and second derivatives, when applied to functions with Lipschitz continuous Hessian, will find an iterate [Formula: see text] at which the gradient’s norm is less than [Formula: see text] in at most [Formula: see text] iterations.
No takes yet. Share an insight, caveat, or question.
Gratton et al. (2023) studied this question.