English

应用科学学报 ›› 2000, Vol. 18 ›› Issue (2): 143-147.

• 论文 • 上一篇    下一篇

MR-CGS:一种最小剩余CGS方法在DDM求解中的应用

赵佳宝1, 盛昭瀚1, 魏同立2   

  1. 1. 东南大学经济管理学院, 江苏南京 210096;
    2. 东南大学微电子中心, 江苏南京 210096
  • 收稿日期:1998-10-24 修回日期:1999-05-06 出版日期:2000-06-30 发布日期:2000-06-30
  • 作者简介:赵佳宝(1972-),男,湖南桃源人,博士生;盛昭瀚(1944-),男,江苏镇江人,教授,博导;魏同立(1934-),男,河北清苑人,教授,博导.

MR-CGS: A Minimal Residual CGS Method and Its Application in DDM

ZHAO Jia-bao1, SHENG Zhao-han1, WEI Tong-li2   

  1. 1. School of Economic and Management, Southeast University, Nanjing 210096, China;
    2. Microelectronic Center, Southeast University, Nanjing 210096, China
  • Received:1998-10-24 Revised:1999-05-06 Online:2000-06-30 Published:2000-06-30

PDF

69

摘要： 从CGS方法入手,针对该算法在收敛过程中出现的残差不平稳的缺点,对CGS算法进行改进,给出了一种最小残差CGS(MR-CGS)算法.算例给出了该改进算法在求解DD模型中的应用.

关键词: 双正交共轭梯度法, 半导体器件, 稀疏矩阵, 数值模拟, 迭代算法, 共轭梯度平方法, 最小剩余共轭梯度平方法

Abstract: Due to the instability of the residuals aroused in the conjugate gradient square method, we present a new method named MR-CGS, which has the minimal residual property while retaining the square converging property. Finally, an example is given to exhibit the application of the improved algorithm in solving the DDM model.

Key words: semiconductor device, sparse matrix, BiCG, iterative algorithm, MR-CGS, numerical simulation, CGS

中图分类号: