矩阵方程AXB=C的最小二乘解的定秩研究
2013-04-29孟纯军李桃珍
孟纯军 李桃珍
摘 要:研究了矩阵方程AXB=C最小二乘解的秩的范围,利用矩阵的奇异值分解以及Frobenius范数的特征,得到了秩约束下最小二乘解的表达式,并得到了最大秩和最小秩最小二乘解.
关键词:最优控制;最小二乘解;秩约束;奇异值分解;Frobenius范数
中图分类号:O241.6 文献标识码:A
On the Rank Range of the Leastsquares
Solutions of the Matrix Equation AXB=C
MENG Chunjun,LI Taozhen
(College of Mathematics and Econometrics, Hunan Univ, Changsha, Hunan 410082, China)
Abstract:This paper, we considered the rank range of the leastsquares solutions of matrix equation AXB=C. By applying the singular value decomposition of matrix and the properties of Frobenius matrix norm, we have obtained the range of the rank and the leastsquares solution expression of under rank constrained. Finally, we have provided the expressions of the leastsquares solutions with maximal and minimum rank respectively.
Key words:optimal control; leastsquares solutions; rank constrained; SVD decomposition; Frobenius norm
1 引 言
约束矩阵方程的定秩求解问题与非线性规划中的半定规划有着密切的联系[1-2],为解决最优控制、鲁棒优化,以及组合优化中的问题提供了一种有效的工具. 因此,越来越多的国内外学者致力于矩阵方程问题的定秩研究,使得约束矩阵方程的定秩求解问题成为了数值代数的热门研究课题之一.研究矩阵表达式的秩和矩阵方程解的秩有很多的文献,如[3-8],但是矩阵方程的最小二乘解的定秩研究还很少.本文着重研究阵方程AXB=C最小二乘解的秩,得到最小二乘解的最大秩、最小秩以及相應的最大(小)秩解,并给出了具有给定秩的最小二乘解的表达式.
本文研究的问题数学描述如下:
参考文献
[1] 何旭初,孙文瑜.广义逆矩阵引论[M].南京:江苏科学技术出版社,1992.
HE Xuchu, SUN Wenyu. The generalized mverse of matrix theory[M].Nanjing:Jiangsu Science and Technology Press, 1992.(In Chinese)
[2] GENE H G,CHARLES F V L, Matrix computations[M].Third Edition. Baltimore, Maryland:The Johns Hopkins University Press, 1996.
[3] MARSAGLIA G, STYAN G P H. Equalities and inequalities for ranks of matrices[J]. Linear and Multilinear Algebra,1974,2:269-292.
[4] WOODGATE K G. Leastsquare solution of F=PG over positive semidefinite symmetric[J]. Linear Algebra and its Applications, 1996, 145: 171-190.
[5] ZHANG X, CHENG M Y. The rankconstrained Hermitian nonnegativedefinite and positivedefine solutions to matrix equation AXA*=B[J].Linear Algebra and its Applications, 2003, 370: 163-174.
[6] XIAO Q F, HU X Y, ZHANG L. The symmetric minimal rank solution of the matrix equation AX=B and the optimal approximation[J].Electronic Journal of Linear Algebra, 2009, 18: 264-273.
[7] LIU Yonghui. Ranks of solutions of the linear matrix equation AX+YB=C[J]. Computers and Mathematics with Applications, 2006, 52: 861-872.
[8] TIAN Yongge. Maximization and minimization of the rank and inertia of the Hermitian matrix expression A-BX-BX* with applications[J]. Linear Algebra and its Applications,2011, 434: 2109-2139.