Improved Analysis of Quaternion Randomized Block Krylov Iterations for Low‐Rank Matrix Approximation
Authors:
- 6 August 2026
- journal article
- Published by Wiley in Numerical Linear Algebra with Applications
- Vol. 33 (4)
Abstract
This paper focuses on approximating a quaternion matrix and its singular values via the quaternion randomized block Krylov iteration (qRBKI) algorithm. Specifically, for qRBKI with block size , we establish novel gap‐based error bounds for the rank‐ matrix approximation in both Frobenius and spectral norms, explicitly demonstrating that qRBKI outperforms the quaternion randomized subspace iteration (qRSI) in approximation quality. For , leveraging simulated starting block vectors and quaternion polynomials, we show that the approximation error depends on the minimal relative singular value gap under the ‐order neighbouring structure. The near‐zero‐gap‐induced dependence can be eliminated by perturbing the input matrix with a low‐intensity random quaternion Gaussian or uniform matrix. Finally, numerical experiments on synthetic data and color image applications validate our theoretical findings and illustrate the effectiveness of qRBKI.Keywords
Funding Information
- National Natural Science Foundation of China (12471355)
- Natural Science Foundation of Shanghai Municipality (23ZR1422400)
This publication has 27 references indexed in Scilit:
- A subspace-orbit randomized algorithm for quaternion tensor singular value decomposition based on Qt-productApplied Mathematics and Computation, 2025
- Randomized Algorithm for Constrained Quaternion Singular Value Decomposition and Its ApplicationsJournal of Scientific Computing, 2025
- Efficient quaternion CUR decomposition based on discrete empirical interpolation methodBIT Numerical Mathematics, 2025
- A Comprehensive Analysis of Quaternion Deep Neural Networks: Architectures, Applications, Challenges, and Future ScopeArchives of Computational Methods in Engineering, 2024
- Randomized block Krylov subspace algorithms for low-rank quaternion matrix approximationsNumerical Algorithms, 2023
- Randomized Quaternion QLP Decomposition for Low-Rank ApproximationJournal of Scientific Computing, 2022
- Randomized block Krylov methods for approximating extreme eigenvaluesNumerische Mathematik, 2021
- Randomized block Krylov subspace methods for trace and log-determinant estimatorsBIT Numerical Mathematics, 2021
- Lanczos method for large-scale quaternion singular value decompositionNumerical Algorithms, 2018
- Quaternions and matrices of quaternionsLinear Algebra and its Applications, 1997