Id: arxiv/2608.12665 Url: https://arxiv.org/abs/2608.12665 Tags: math.OC, cs.LG, stat.ML Score: 0 Title: A Local-Linearly Convergent Algorithm for Nonconvex Equality-Constrained Optimization Domain: math.OC Source: arxiv Authors: Frank E. Curtis, Lingjun Guo, Daniel P. Robinson Reviews: 0 Subject: math.OC Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-08-12 Page: https://delphi.science/r/arxiv/2608.12665 For solving nonconvex equality-constrained optimization problems, a recent Gradient-Eigenstep Algorithm by Goyens et al.~is an iteration-efficient approach, based on minimizing Fletcher's augmented Lagrangian function, for finding an approximate second-order stationary point from an arbitrary starting point. In this paper, the analysis of this algorithm is extended, offering a two-fold contribution. First, it is shown that a local-linear rate of convergence can be obtained by this method if it is initiated sufficiently close to a strong second-order stationary point and employs a…