A simple and practical adaptive trust-region method

TL;DR

This paper introduces a simple, adaptive trust-region method for unconstrained optimization that achieves near-optimal convergence bounds, improves practical performance over existing methods, and requires fewer evaluations on benchmark problems.

Contribution

A novel adaptive trust-region algorithm with inexact subproblem solutions that improves convergence bounds and practical efficiency over prior methods.

Findings

Achieves near-optimal convergence bounds for finding stationary points.

Uses fewer function, gradient, and Hessian evaluations than state-of-the-art methods.

Practical enhancements significantly reduce wall-clock time.

Abstract

We present an adaptive trust-region method for unconstrained optimization that allows inexact solutions to the trust-region subproblems. Our method is a simple variant of the classical trust-region method of \citet{sorensen1982newton}. The method achieves the best possible convergence bound up to an additive log factor, for finding an $\epsilon$-approximate stationary point, i.e., $O( \Delta_f L^{1/2} \epsilon^{-3/2}) + \tilde{O}(1)$ iterations where $L$ is the Lipschitz constant of the Hessian, $\Delta_f$ is the optimality gap, and $\epsilon$ is the termination tolerance for the gradient norm. This improves over existing trust-region methods whose worst-case bound is at least a factor of $L$ worse. We compare our performance with state-of-the-art trust-region (TRU) and cubic regularization (ARC) methods from the GALAHAD library on the CUTEst benchmark set on problems with more than 100 variables. We use fewer function, gradient, and Hessian evaluations than these methods. For instance, our algorithm's median number of gradient evaluations is $23$ compared to $36$ for TRU and $29$ for ARC. Compared to the conference version of this paper \cite{hamad2022consistently}, our revised method includes several practical enhancements. These modifications dramatically improved performance, including an order of magnitude reduction in the shifted geometric mean of wall-clock times. We also show it suffices for the second derivatives to be locally Lipschitz to guarantee that either the minimum gradient norm converges to zero or the objective value tends towards negative infinity, even when the iterates diverge.