Saved in:
Bibliographic Details
Main Authors: Shen, Ya, Li, Qing-Na, Dai, Yu-Hong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.16731
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908600256954368
author Shen, Ya
Li, Qing-Na
Dai, Yu-Hong
author_facet Shen, Ya
Li, Qing-Na
Dai, Yu-Hong
contents Gradient methods are among the simplest yet most widely used algorithms for unconstrained optimization. Motivated by a geometric property of the steepest descent (SD) method that can alleviate the zigzag behavior in quadratic problems, we develop a new gradient variant called the Triangle Steepest Descent (TSD) method. The TSD method introduces a cycle parameter $j$ that governs the periodic combination of past search directions, providing a geometry-driven mechanism to enhance convergence. To the best of our knowledge, TSD is the first formally established geometry-based gradient scheme since Akaike (1959). We prove that TSD is at least R-linearly convergent for strongly convex quadratic problems and demonstrate through extensive numerical experiments that it exhibits superlinear behavior, outperforming the Barzilai-Borwein (BB) method and monotone Dai-Yuan gradient method (DY) in quadratic cases. These results suggest that incorporating geometric information into gradient directions offers a promising avenue for developing efficient optimization algorithms.
format Preprint
id arxiv_https___arxiv_org_abs_2501_16731
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Triangle Steepest Descent: A Geometry-Based Gradient Algorithm with Guaranteed R-Linear Convergence
Shen, Ya
Li, Qing-Na
Dai, Yu-Hong
Optimization and Control
90C20 \and 90C25
Gradient methods are among the simplest yet most widely used algorithms for unconstrained optimization. Motivated by a geometric property of the steepest descent (SD) method that can alleviate the zigzag behavior in quadratic problems, we develop a new gradient variant called the Triangle Steepest Descent (TSD) method. The TSD method introduces a cycle parameter $j$ that governs the periodic combination of past search directions, providing a geometry-driven mechanism to enhance convergence. To the best of our knowledge, TSD is the first formally established geometry-based gradient scheme since Akaike (1959). We prove that TSD is at least R-linearly convergent for strongly convex quadratic problems and demonstrate through extensive numerical experiments that it exhibits superlinear behavior, outperforming the Barzilai-Borwein (BB) method and monotone Dai-Yuan gradient method (DY) in quadratic cases. These results suggest that incorporating geometric information into gradient directions offers a promising avenue for developing efficient optimization algorithms.
title Triangle Steepest Descent: A Geometry-Based Gradient Algorithm with Guaranteed R-Linear Convergence
topic Optimization and Control
90C20 \and 90C25
url https://arxiv.org/abs/2501.16731