摘要
针对二维与三维二阶偏微分方程,在矩形网格上分别采用五点与七点差分离散,并采用自然排序得到的矩阵,证明了比已有结论更弱的可对称化的充分条件.实验数值表明:文中方法优于传统的、直接应用到原线性方程组的BICG、CGS、BICGSTAB、GMRES及QMR等Krylov子空间迭代法.
For matrices derived from partial differential equations on rectangular region and in natural node ordering, the paper proves some much weaker symmetrizable conditions than the exist ones. The objective matrices include the ones from 2-D PDEs with 5-point difference scheme and the ones from 3-D PDEs with 7-point difference scheme. The results show that the method used in the paper is better than the traditional Krylov subspace methods including BICG, CGS, BICGSTAB, GMRES and QMR.
出处
《装备指挥技术学院学报》
2003年第3期89-92,共4页
Journal of the Academy of Equipment Command & Technology
基金
部委级资助项目
