Score: 1

Splitting-based randomised dynamical low-rank approximations for stiff matrix differential equations

Published: June 18, 2025 | arXiv ID: 2506.15259v1

By: Zi Wu, Yong-Liang Zhao

BigTech Affiliations: Weibo

Potential Business Impact:

Solves hard math problems faster by breaking them down.

Business Areas:
A/B Testing Data and Analytics

In the fields of control theory and machine learning, the dynamic low-rank approximation for large-scale matrices has received substantial attention. Considering the large-scale semilinear stiff matrix differential equations, we propose a dynamic numerical integrator for obtaining low-rank approximations of solutions. We first decompose the differential equation into a stiff linear component and a nonstiff nonlinear term, then employ an exponential integrator along with a dynamic low-rank approach to resolve these subsystems, respectively. Furthermore, the proposed framework naturally extends to rank-adaptation scenarios. Through rigorous validation on canonical stiff matrix differential problems, including spatially discretized Allen-Cahn equations and differential Riccati equations, we demonstrate that the method achieves the theoretically predicted convergence orders. Numerical evidence confirms the robustness and accuracy of the proposed methods.

Country of Origin
🇨🇳 China

Page Count
21 pages

Category
Mathematics:
Numerical Analysis (Math)