Several accelerated gradient-based iteration methods for solving AXB = C with application to tensor surface fitting

Zhaolu Tian

doi:10.1007/s11075-025-02220-8

Several accelerated gradient-based iteration methods for solving AXB = C with application to tensor surface fitting

2025-10-01

0

PRE

AI

DOI：10.1007/s11075-025-02220-8

Original

Original request for help

Save

Abstract

En 中文

The gradient-based iteration (GBI) method is widely utilized for solving large-scale matrix equations AXB = C due to its simplicity and efficiency. This paper proposes significant enhancements to the convergence rate of the GBI method by integrating preconditioned technique, momentum acceleration, and Chebyshev semi-iterative scheme. We give rigorous convergence analyses for these accelerated methods and provide detailed investigations into optimal parameter selection. Finally, comprehensive numerical experiments are carried out to demonstrate the superior efficiency of the accelerated methods with the corresponding optimal parameters. In addition, their potential utility is illustrated through the real-world applications, such as tensor surface fitting in computer-aided geometric design.

Keywords:

GBI method

Optimal parameter

Momentum

Chebyshev polynomial

Several accelerated gradient-based iteration methods for solving AXB = C with application to tensor surface fitting

Abstract

Journal

Organization