Newton--Krylov methods are variants of inexact Newton methods where the approximate Newton direction is taken from a subspace of small dimension. Here we introduce a new hybrid Newton-GMRES method where a global strategy restricted to a low-dimensional subspace generated by GMRES is performed. The obtained process is consistent with preconditioning and with matrix-free implementation. Computational results indicate that our proposal enhances the classical backtracking inexact method.

We propose a stochastic first-order trust-region method with inexact function and gradient evaluations for solving finite-sum minimization problems. Using a suitable reformulation of the given problem, our method combines the inexact restoration approach for constrained optimization with the trust-region procedure and random models. Differently from other recent stochastic trust-region schemes, our proposed algorithm improves feasibility and optimality in a modular way. We provide the expected number of iterations for reaching a near-stationary point by imposing some probability accuracy requirements on random functions and gradients which are, in general, less stringent than the corresponding ones in literature. We validate the proposed algorithm on some nonconvex optimization problems arising in binary classification and regression, showing that it performs well in terms of cost and accuracy, and allows to reduce the burdensome tuning of the hyper-parameters involved.