Title
Mixed Precision General Alternating-Direction Implicit Method for Solving Large Sparse Linear Systems

About the Speaker

Prof. Zhang Juan

Professor Zhang Juan, serves as Vice Dean of the School of Mathematics and Computational Science, Xiangtan University, and Executive Deputy Director of the Key Laboratory of Intelligent Computing and Information Processing, Ministry of Education. She was selected into the Huxiang Youth Talent Program and the Hunan Provincial Training Program for Young Backbone Teachers. She has been a visiting scholar at the National University of Singapore and the University of Macau on multiple occasions. She presides over the General Program of the National Natural Science Foundation of China, research projects of Kaiyuan International Research Institute, and projects of the 711 Research Institute. As the sub-project principal investigator, she undertakes the National Key R&D Program, projects of the Ministry of Industry and Information Technology. Her main research interests include numerical algebra and control theory. In the past five years, she has published more than 20 academic papers in top journals in computational mathematics, control theory and data science, including SINUM, SISC, Automatica, IEEE TKDE, JCP, JSC and CSIAM-AM.

Abstract

This talk introduces a novel mixed-precision formulation of the general alternating-direction implicit (GADI) method, specifically designed to accelerate the solution of large-scale sparse linear systems. By computing the computationally intensive subsystems in low precision (e.g., Bfloat16 or FP32) while performing residual calculations and solution updates in high precision, the proposed approach achieves a significant reduction in execution time without sacrificing the final solution accuracy. This paper presents a comprehensive rounding error analysis, which rigorously establishes the convergence rate and limiting accuracy of the method under mixed-precision arithmetic. Furthermore, the project team proposes a robust strategy based on Gaussian process regression (GPR) for the optimal selection of the regularization parameter. Performance benchmarks conducted on an NVIDIA A100 GPU demonstrate that the mixed-precision GADI method delivers speedups of up to 2.6× compared to standard double-precision implementations, with tests performed on large-scale convection-diffusion and reaction-diffusion problems.

Host

Prof. Chen Congliang

Date & Time

October 9, 2026, Friday,

2:30pm-3:30pm

Venue

Room B301, Shenzhen Loop Area Institute

(6 Hongmian Rd, Fubao Sub-Street, Futian District, Shenzhen, navigate to "Shenzhen Loop Area Institute (South Gate)" on the map.)

Online link

Join the Meeting Online

Tencent Meeting: 948-1135-4043