Zeroth-Order Riemannian Adaptive Regularized Proximal Quasi-Newton Optimization Method
Axioms · 10.03.2026 · 15 (3) · 203
DOI: 10.3390/axioms15030203 ↗Kurz gefasst
Fachartikel in der Zeitschrift Axioms, online erschienen am 10.03.2026. Verfasst von Yinpu Ma und 3 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Natural Science Foundation of Ningxia, National Social Science Fund of China, North Minzu University Research Initiative.
Abstract (Originalsprache)
Recently, the adaptive regularized proximal quasi-Newton (ARPQN) method has demonstrated a strong performance in solving composite optimization problems over the Stiefel manifold. However, its reliance on first-order information limits its applicability to scenarios where gradient and Hessian evaluations are unavailable or costly. In this paper, we propose a zeroth-order adaptive regularized proximal quasi-Newton method (ZO-ARPQN) for black-box composite optimization over Riemannian manifolds, particularly the Stiefel and symmetric positive definite (SPD) manifolds. The proposed method estimates the Riemannian gradient and curvature information through randomized one-point finite-difference approximations and adaptively updates a regularized quasi-Newton matrix to capture the local manifold geometry. Theoretically, we established global convergence and complex analyses under mild assumptions. More importantly, by incorporating curvature-aware regularization and random perturbations in the proximal quasi-Newton framework, we proved that ZO-ARPQN can escape strict saddle points with a high probability. This guarantees convergence to a stationary point, even in the absence of explicit gradients. Extensive numerical experiments were conducted on manifold-constrained problems, including sparse PCA and robot stiffness tuning. These demonstrated that ZO-ARPQN shows a competitive convergence behavior compared with other state-of-the-art Riemannian optimization methods, while requiring only function evaluations.
Förderung
- Natural Science Foundation of Ningxia
- National Social Science Fund of China
- North Minzu University Research Initiative
- First-Class Discipline Construction Program of Ningxia
- Ningxia Youth Talent Support Program
- Leading Talent Program of North Minzu University
- Governance and Social Management Research Center of Northwest Ethnic Regions
Zitieren
APA
Ma, Y., Li, C., Wang, Z., & Li, Q. (2026). Zeroth-Order Riemannian Adaptive Regularized Proximal Quasi-Newton Optimization Method. Axioms, 15(3), 203. https://doi.org/10.3390/axioms15030203
Vancouver
Ma Y, Li C, Wang Z, Li Q. Zeroth-Order Riemannian Adaptive Regularized Proximal Quasi-Newton Optimization Method. Axioms. 2026;15(3):203. doi:10.3390/axioms15030203
BibTeX
@article{Ma2026,
author = {Ma, Yinpu and Li, Cunlin and Wang, Zhichao and Li, Qian},
title = {Zeroth-Order Riemannian Adaptive Regularized Proximal Quasi-Newton Optimization Method},
journal = {Axioms},
year = {2026},
volume = {15},
number = {3},
pages = {203},
doi = {10.3390/axioms15030203}
}
EUVIMED