一类线性矩阵方程的数值求解方法
发布时间:2018-08-05 17:18
【摘要】:在科学研究、工程计算、经济控制等领域中,许多问题的数学模型可以用线性矩阵方程来描述,因此,研究线性矩阵方程的求解方法具有重要意义。采用直接法求解大型矩阵方程,由于规模大、计算变元多、计算误差不易控制,因而,如何通过行之有效的数值迭代方法来求解线性矩阵方程的最优解成为诸多数学工作者的研究方向之一。本文分多层次从易而难研究了线性矩阵方程(?)Ai(j)XiBi(j)=F(j)(j=1,2,…,M)的数值求解方法,其中Ai(j)∈Rm×n,Bi(j)∈Rn×p,F(i)∈Rm×p。首先讨论了最简单最基本线性矩阵方程AX=B的几种迭代解法,即把线性方程组数值迭代方法推广运用到线性矩阵方程数值求解方法上来,得到诸如雅克比迭代法及基于此方法的方阵乘幂求和方法、高斯赛德尔迭代法、SOR迭代法等,并给出了这些数值算法在一定条件下的收敛性,最后举例证明算法的可行性。其次讨论了诸如矩阵方程AX+XB=F的求解问题,给出了这种线性矩阵方程五种计算方法——特征多项式法、特征向量法、级数法、上三角形法及小参数迭代法。然后对线性矩阵方程AXB + CXD = F,给出了诸如雅克比、高斯赛德尔以及拟高斯赛德尔等分组迭代解法,给出相关收敛条件,并对相关的收敛性定理给出证明,最后举例说明这些算法的有效性。最后针对一般线性矩阵方程或方程组,讨论了其变形共轭梯度算法,给出了单变量和多变量线性矩阵方程组的变形共轭梯度算法,数值算例说明了这两类矩阵方程在求某些特殊解时所给算法的正确性。本文第一章从应用领域,对线性矩阵方程数值求解的研究背景及意义进行了概述,对国内外相关研究文献进行了综述,对本文的主要工作安排作一阐述。第二章对求解线性矩阵方程AX=B的基本迭代算法以及这些算法的收敛性问题进行了讨论,并用数值例子证明了所给算法的收敛性及正确性。第三章主要讨论了简单线性矩阵方程AX+XB = F的直接求解方法和数值求解方法,包括特征多项式法,特征向量法,级数法(大参数方法),上三角形方法及小参数迭代法等。第四章对一般线性方程组,给出了变形共轭梯度数值迭代算法,举例说明这些算法的正确性。最后一章总结全文工作,并简单的展望未来的研究方向。
[Abstract]:In the fields of scientific research, engineering calculation, economic control and so on, many mathematical models of problems can be described by linear matrix equations. Therefore, it is of great significance to study the solving methods of linear matrix equations. By using direct method to solve large matrix equations, because of the large scale, the number of calculation variables and the calculation error are not easy to control, therefore, How to solve the optimal solution of linear matrix equation by effective numerical iterative method has become one of the research directions of many mathematics workers. In this paper, the linear matrix equation (?) Ai (j) XiBi (j) F (j) (JJ) is studied from the point of view of easy and difficult. Where Ai (j) 鈭,
本文编号:2166457
[Abstract]:In the fields of scientific research, engineering calculation, economic control and so on, many mathematical models of problems can be described by linear matrix equations. Therefore, it is of great significance to study the solving methods of linear matrix equations. By using direct method to solve large matrix equations, because of the large scale, the number of calculation variables and the calculation error are not easy to control, therefore, How to solve the optimal solution of linear matrix equation by effective numerical iterative method has become one of the research directions of many mathematics workers. In this paper, the linear matrix equation (?) Ai (j) XiBi (j) F (j) (JJ) is studied from the point of view of easy and difficult. Where Ai (j) 鈭,
本文编号:2166457
本文链接:https://www.wllwen.com/kejilunwen/yysx/2166457.html