关于GQR因子分解和Cholesky分解块修改的扰动分析

近来,广义QR分解引起了数值代数界的广泛兴趣.Anderson等研究了GQR的若干性质并讨论了在广义最小二乘等问题上的应用;Paige研究了GQR的数值性质;Hammer-ling用GQR处理一般的Gauss-Markov线性模型参数估计问题;Barrlund给出了GQR分解因子的扰动界.我们注意到Barrlund的论证方法和所得结果都比较复杂.     

一类迭代法的舍入误差分析

本文给出了解线性方程组的最速下降法(钭量法)和残量法的舍入误差分析,在此基础上证明了这两类方法都是数值稳定的。文中所得出的结论优于Wosniakowski的相应结果。     

多项式插值算法的舍入误差分析

本文定义了多项式插值算子的条件数和多项式插值算法的数值稳定性等概念.主要研究结果是:若n和Y_max不太大,当结点等距分布时,Lagrange插值和Newton插值算法都是数值稳定的.但是不论结点如何分布,上述两法的外推计算可能是数值不稳定的.文中数值例子验证了这些理论结果.     

矩阵LU分解定理的简单证明及其数值实现

用简单的方法证明了矩阵LU分解定理,讨论了定理的推广以及定理相应的数值实现,并对《数值分析》课程教学方法改革进行了思考.     

关于TLS问题

1.引 言考虑观测线性系统AX=B,（1.1a）其中A∈Cm×n,B∈Cm×d（本文通篇假设m≥n d）,分别是精确但不可观测的A0∈Cm×n,B0∈Cm×d的近似,即精确线性系统是A0X=B0.（1.1b）Golub和Van Loan于1980年提出的总体最小二乘问题（以下简称TLS问题）就是求解线性系统AX=B（1.2）     

一些矩阵分解的敏度分析

１．引言 代数Ｒｉｃｃａｔｉ方程是线性系统理论与设计的核心课题之一．矩阵的Ｈｅｓｓｅｎｂｅｒｇ分解、Ｈａｍｉｌｔｏｎ矩阵的平方约化分解、辛矩阵的ＱＴ分解是数值求解代数Ｒｉｃｃａｔｉ 方程的基本工具．关于 Ｈｅｓｓｅｎｂｅｒｇ分解的研究工作有很多（参阅 ［４］及其参考文献）．最近, Ｓｕｎ［４］利用矩阵分裂算子研究了Ｈｅｓｓｅｎｂｅｒｇ分解因子的扰动分析,并根据所得的扰动上界定义了分解因子的条件数．本文第 ２节将运用局部展开方法引入 Ｈｅｓｓｅｎｂｅｒｇ分解因子的条件数．有趣的是所定义的条件数与Ｓｕｎ引…     

TLS和LS问题的比较

There are a number of articles discussing the total least squares(TLS) and the least squares(LS) problems.M.Wei(M.Wei, Mathematica Numerica Sinica 20(3)(1998),267-278) proposed a new orthogonal projection method to improve existing perturbation bounds of the TLS and LS problems.In this paper,wecontinue to improve existing bounds of differences between the squared residuals,the weighted squared residuals and the minimum norm correction matrices of the TLS and LS problems.     

用Prony方法计算信号数据参数的误差分析

对于用Prony方法计算信号数据参数的误差分析,指出了提高计算精度的方法和影响精度的因素.我们分析了秩亏LS和TLS方法和误差界,指出用秩亏的Prony方法可大大提高计算精度.同时也指出了T的变化和η的变化对结果精度的影响,以及精确计算重奇点的方法,     

最小二乘估计关于误差分布的稳健性

对于设计矩阵$X$是列降秩的线性统计模型, 本文讨论了最小二乘估计关于误差分布的稳健性, 给出了误差分布的最大类, 使得误差项的分布在此范围内变动时, 最小二乘估计在均方误差矩阵准则下是最优估计.     

非负张量分解的不平衡乘性更新

针对非负张量分解的乘性更新算法,讨论了其元素形式与矩阵形式的一致性,并给出了不平衡的乘性更新算法．数值试验表明,新的算法具有更快的收敛性．     

ON GROWTH FACTORS OF THE LU FACTORIZATION

In this article, we derive upper bounds of different growth factors for the LU factorization, which are dominated by A11(k)-1A12(k),A21(k)A11(k)-1, where A11(k), A12(k), A21(k), A22(k) are sub-matrices of A. We also derive upper bounds of growth factors for the Cholesky factorization. Numerical examples are presented to verify our findings.     

LU分解的增长因子

In this article, we derive upper bounds of different growth factors for the LU factorization, which are dominated by A11(k)-1A12(k),A21(k)A11(k)-1, where A11(k), A12(k), A21(k), A22(k) are sub-matrices of A. We also derive upper bounds of growth factors for the Cholesky factorization. Numerical examples are presented to verify our findings.     

广义Cholesky分解乘法扰动分析

0引言关于实对称矩阵的广义Cholesky分解和扰动问题是矩阵计算的重要问题,可参考文献[1-2].本文首先介绍已有的采用加法扰动的角度得到的广义Cholesky分解的一阶相对     

COMBINATIVE PRECONDITIONERS OF MODIFIED INCOMPLETE CHOLESKY FACTORIZATION AND SHERMAN-MORRISON-WOODBURY UPDATE FOR SELF-ADJOINT ELLIPTIC DIRICHLET-PERIODIC BOUNDARY VALUE PROBLEMS

For the system of linear equations arising from discretization of the second-order self-adjoint elliptic Dirichlet-periodic boundary value problems,by making use of the specialstructure of the coefficient matrix we present a class of combinative preconditioners whichare technical combinations of modified incomplete Cholesky factorizations and Sherman-Morrison-Woodbury update.Theoretical analyses show that the condition numbers of thepreconditioned matrices can be reduced to(?)(h~(-1)),one order smaller than the conditionnumber(?)(h~(-2))of the original matrix.Numerical implementations show that the resultingpreconditioned conjugate gradient methods are feasible,robust and efficient for solving thisclass of linear systems.     

Stability and Sensitivity of Tridiagonal LU Factorization without Pivoting

In this paper the accuracy of LU factorization of tridiagonal matrices without pivoting is considered. Two types of componentwise condition numbers for the L and U factors of tridiadonal matrices are presented and compared. One type is a condition number with respect to small relative perturbations of each entry of the matrix. The other type is a condition number with respect to small componentwise perturbations of the kind appearing in the backward error analysis of the usual algorithm for the LU factorization. We show that both condition numbers are of similar magnitude. This means that the algorithm is componentwise forward stable, i.e., the forward errors are of similar magnitude to those produced by a componentwise backward stable method. Moreover the presented condition numbers can be computed in O(n) flops, which allows to estimate with low cost the forward errors. AMS subject classification (2000) 65F35, 65F50, 15A12, 15A23, 65G50.Received October 2003. Accepted August 2004. Communicated by Per Christian Hansen.Froilán M. Dopico: This research has been partially supported by the Ministerio de Ciencia y Tecnología of Spain through grants BFM2003-06335-C03-02 (M. I. Bueno) and BFM2000-0008 (F. M. Dopico).     

非线性抛物方程耦合的离散化及其误差分析

0　引　　言对于线性抛物型初边值问题的有限元与边界元耦合法,我们已作过研究(可参见[2],[3]).然而,许多实际问题,如流体、对流扩散、热辐射及热传输等涉及到非线性问题,因此研究非线性问题的数值方法显得尤为重要.近年来,G.N.Gatica与G.C.Hsiao已将有限元(FEM)与边界元(BEM)耦合法拓广到非线性椭圆问题(如[5],[6]),但如何应用FEM与BEM耦合法来处理非线性抛物型初边值问题,就作者所知迄今为止尚属空白.这里我们试图对此进行研究.设Ω为R2中一有界单连通区域,其边界为Γ:=Ω.Ωc:=R2\Ω为闭区域Ω的补区域,I:=(0,T].我们考虑如…     

块三对角阵分解因子的估值与应用

1.引 言 许多物理应用问题归结为求微分方程数值解,而这可以通过离散化为求解稀疏线性方程组,所以稀疏线性方程组求解的有效性在很大程度上决定了原问题求解算法的有效性.直接     

Backus—Gilbert方法对不精确数据解的误差估计

1 引言 1968年地质学家G Backus and F Gilbert给出了求解线性(非线性)矩问题的一种方法,用来求解地球物理反问题。后来称这种方法为B-G方法。过去几年,数学家们从理论和应用方面研究了B-G方法。从理论上分析了其收敛性并给出了误差估计。第一类算子方程在不同函数空间的离散化得到不同形式的矩问题。[4]、[5]研究了再生核空间和小波空间的B-G方法。并用于信号恢复问题。由于矩问题的不适定性,有必要分析B-G方法解的正则性,本文得到了B-G方法对不精确数据的误差估计,从而证明了B-G方法是—种正则化方法。 考虑线性矩问题     

二阶扩张状态观测器的误差分析

本文利用分片光滑的Ｌｙａｐｕｎｏｖ函数来进行扩张状态观测器的误差分析,指出了非线性扩张状态观测器有更好的估计精度能力,给出了为提高估计精度扩张状态观测器的参数所满足的条件．     

A PRIORI ERROR ESTIMATE AND SUPERCONVERGENCE ANALYSIS FOR AN OPTIMAL CONTROL PROBLEM OF BILINEAR TYPE

In this paper, we investigate a priori error estimates and superconvergence properties for a model optimal control problem of bilinear type, which includes some parameter estimation application. The state and co-state are discretized by piecewise linear functions and control is approximated by piecewise constant functions. We derive a priori error estimates and superconvergence analysis for both the control and the state approximations. We also give the optimal L^2-norm error estimates and the almost optimal L^∞-norm estimates about the state and co-state. The results can be readily used for constructing a posteriori error estimators in adaptive finite element approximation of such optimal control problems.