Skip to the main content

Original scientific paper

https://doi.org/10.64785/mc.31.2.3

The randomized block steepest descent iterative method for solving linear systems and its extension to the tensor framework: Randomized Block Residual Norm Steepest Descent Method

Fatemeh Panjeh Ali Beik ; Department of Mathematics, Vali-e-Asr University of Rafsanjan, Rafsanjan, Iran *
Ke Zhang ; Department of Mathematics, Shanghai Maritime University, Shanghai, China

* Corresponding author.


Full text: english pdf 960 Kb

page 179-195

downloads: 0

cite


Abstract

We present a new randomized block variant of the residual norm steepest descent iterative method for solving a consistent linear system of equations. The proposed algorithm is also applied both in matrix form for solving a class of matrix equations and in the t-product framework for solving a class of tensor equations. We prove that the residual norm of the computed approximate solutions converges linearly in expectation. Numerical experiments are reported to illustrate the validity of established results and superiority of the proposed approach over three recently examined methods for solving a linear system of equations with randomly generated coefficient matrices, a matrix equation and a tensor equation arising from color image restoration problems in the literature.

Keywords

linear system, randomized iterative method, steepest descent, convergence, t-product

Hrčak ID:

353487

URI

https://hrcak.srce.hr/353487

Publication date:

17.11.2026.

Visits: 0 *