Gauss-Newton Accuracy and Indefinite Hessians: Uniform Coexistence in Low-Cost Sets
Abstract
We study the accuracy of Gauss-Newton curvature in ridge-regularized nonlinear least squares. Under local regularity and persistence of level-set curvature magnitude along an exact-fit section, we prove uniform coexistence of two curvature regimes. Global minimizers exist, and every global minimizer has relative Hessian error below (1+√2)/8, while the same low-cost set contains a point with an indefinite Hessian and relative error at least 15/8. One positive ridge cap works for all independent center and label perturbations in fixed neighborhoods and every positive ridge weight up to the cap. These neighborhoods do not shrink as the ridge weight tends to zero. A pointwise certificate based on the current prediction level set controls the normal, mixed, and tangent parts of the Hessian correction. We prove a sharp relative-error bound over the stated pointwise class when the prediction map and ridge vary. Analytic examples describe the roles of output alignment, curvature orientation, and persistence. A separate structural result gives full Jacobian row rank throughout low-cost sets and exact interpolation near a rank-deficient reference.
Related
- Cyclotomic Newton Expansions and a Rank-Uniform Integer-Valued Newton Completion
- Time-uniform accuracy of ensemble Kalman filters with localization
- Optimal Parameter-Free Second-order Acceleration with Inexact Hessians under Generalized Smoothness
- CaCuTe Meets SESOP: Better Theory and Practice with Lipschitz Hessians
- Existence and multiplicity of solutions for a class of new nonlocal problem with indefinite potential
BibTeX
@misc{rhee2026gaussnewton,
title = {{Gauss-Newton Accuracy and Indefinite Hessians: Uniform Coexistence in Low-Cost Sets}},
author = {Kihun Rhee and Hanjoon Byun},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.09675},
eprint = {2610.09675},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.09675},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Gauss-Newton Accuracy and Indefinite Hessians: Uniform Coexistence in Low-Cost Sets
AU - Kihun Rhee
AU - Hanjoon Byun
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2610.09675
UR - https://arxiv.org/abs/2610.09675
AB - We study the accuracy of Gauss-Newton curvature in ridge-regularized nonlinear least squares. Under local regularity and persistence of level-set curvature magnitude along an exact-fit section, we prove uniform coexistence of two curvature regimes. Global minimizers exist, and every global minimizer has relative Hessian error below (1+√2)/8, while the same low-cost set contains a point with an indefinite Hessian and relative error at least 15/8. One positive ridge cap works for all independent center and label perturbations in fixed neighborhoods and every positive ridge weight up to the cap. These neighborhoods do not shrink as the ridge weight tends to zero. A pointwise certificate based on the current prediction level set controls the normal, mixed, and tangent parts of the Hessian correction. We prove a sharp relative-error bound over the stated pointwise class when the prediction map and ridge vary. Analytic examples describe the roles of output alignment, curvature orientation, and persistence. A separate structural result gives full Jacobian row rank throughout low-cost sets and exact interpolation near a rank-deficient reference.
ER -