计算物理 ›› 2024, Vol. 41 ›› Issue (1): 64-74.DOI: 10.19596/j.cnki.1001-246x.8787

• 面向超级计算机的性能优化技术与数值并行算法专刊 • 上一篇    下一篇

特征修正并行预条件算法框架

徐小文1,2(), 莫则尧1,2, 胡少亮1,2, 安恒斌1,2   

  1. 1. 北京应用物理与计算数学研究所, 计算物理全国重点实验室, 北京 100094
    2. 中国工程物理研究院高性能数值模拟软件中心, 北京 100088
  • 收稿日期:2023-06-30 出版日期:2024-01-25 发布日期:2024-02-05
  • 作者简介:徐小文, 男, 博士, 研究员, 博士生导师, 研究方向为大规模并行计算与并行算法, E-mail: xwxu@iapcm.ac.cn
  • 基金资助:
    国家自然科学基金项目(62032023);国防科工局挑战专题项目和CAE软件专项

Feature-modified Algorithm Framework for Parallel Preconditioning in Sparse Linear Solvers

Xiaowen XU1,2(), Zeyao MO1,2, Shaoliang HU1,2, Hengbin AN1,2   

  1. 1. Institute of Applied Physics and Computational Mathematics, Laboratory of Computational Physics, Beijing 100094, China
    2. Software Center for High Performance Numerical Simulation, China Academy of Engineering Physics, Beijing 100088, China
  • Received:2023-06-30 Online:2024-01-25 Published:2024-02-05

摘要:

针对实际应用中稀疏线性解法器计算复杂度偏离线性扩展的瓶颈问题, 提出特征修正预条件算法统一框架, 通过凝练物理特征中影响算法效率的代数特征, 结合多层次特征分析, 构造特征修正组件。通过几类典型特征修正预条件算法及应用成效, 展示了该框架的有效性。

关键词: 稀疏线性代数方程组, 特征修正, 迭代方法, 预条件算法, 并行算法

Abstract:

To address the high computational complexity of sparse linear solvers caused by complex physical characteristics in practical applications, this paper presents a unified framework for feature-modified preconditioning algorithms. By refining the algebraic features affecting the efficiency from physical characteristics and combining multilevel feature analysis, we construct feature-modified components. The effectiveness of this framework is demonstrated through several typical feature-modified preconditioning algorithms and their application results.

Key words: sparse linear algebraic equations, feature-modified, iterative methods, preconditioning algorithms, parallel algorithms

中图分类号: