摘要
有限元求解的大型线性方程组,其系数矩阵一般具有稀疏对称的特点,为了减少系数矩阵的存贮规模,同时便于求解,采用全稀疏存贮策略:在求解过程中只存贮对称部分的非零元素,用链表式管理算法,既减少存贮空间,又便于存贮结构的动态更改。这里给出的数值算例是使用ICCG迭代法与不同的存贮方式配合求解,该方案在时间和存贮上都较为占优,更好的提高了求解效率,能够应用于有限元大型稀疏线性方程组的求解。
In solving systems of large scale sparse linear equations from FEM, it＇s coefficient matrix is sparse and symmetrical. Fully sparse strategy is introduced for dropping storage of coefficient matrix and solving conveniently. In the solution, nonzero elements of symmetrical part only are stored and managed by chain pattern arithmetic. So the storage structure is small and convenient for dynamic change. ICCG method with different storage models solved numerical illustrations. From the results, fully sparse strategy is predominant for time and storage, and can improve efficiency of solution. Therefore fully sparse strategy can be applied to solve systems of large scale sparse linear equations from FEM.
出处
《陕西理工学院学报（自然科学版）》
2005年第4期67-68,共2页
Journal of Shananxi University of Technology:Natural Science Edition
