張華民,殷紅彩
(1.蚌埠學院數理系,安徽蚌埠 233030;2.江南大學控制科學與工程研究中心,江蘇無錫 214122; 3.安徽財經大學管理科學與工程學院,安徽蚌埠 233000)
向量交換矩陣一種新的定義及應用
張華民1,2,殷紅彩3
(1.蚌埠學院數理系,安徽蚌埠 233030;2.江南大學控制科學與工程研究中心,江蘇無錫 214122; 3.安徽財經大學管理科學與工程學院,安徽蚌埠 233000)
利用單位矩陣和基本向量給出了向量交換矩陣的一種較以往表述簡單的新的定義.基于新的定義證明了向量交換矩陣的性質.給出了新定義與原有定義的等價性的證明.最后給出了矩陣克羅內克積奇異值的一個新的結論.
克羅內克積;向量交換矩陣;向量化算子;奇異值
克羅內克積(Kronecker product)是用數學家Leopold Kronecker(1823-1891)的名子命名的一個概念.事實上它應該被稱為Zehfuss product,因為是Johann Georg Zehfuss在1858年發表的一篇論文中給出了關于n階方陣的公式[12]:

克羅內克積被廣泛應用在系統理論[36],矩陣微分計算[79],線性矩陣方程[1015],系統辨識[16-19],及其它領域[20-25].
本文在總結已有表述的基礎上,提出了在克羅內克積的應用中起重要作用的向量交換矩陣(vec-permutation matrix)一種新的定義,并用新的定義證明了和克羅內克積,向量交換矩陣及向量化算子(vector operator)相關的結論,給出了新定義和原定義間的等價性證明,最后建立了關于矩陣克羅內克積奇異值的一個新結論.
設F是一個數域,例如是實數域R或復數域C.矩陣A=[aij]∈Fm×n和B∈Fp×q的克羅內克積(直積或張量積),記為A?B,定義如下

由定義可得兩個對角矩陣(上三角矩陣或下三角矩陣)的克羅內克積仍是對角矩陣(上三角矩陣或下三角矩陣).設AT和AH分別表示矩陣A的轉置和共軛轉置,Im是m階的單位矩陣.由定義可直接驗證下面的克羅內克積的一些性質:

其中性質1表明列向量α和行向量βT的矩陣乘積等價于二者的克羅內克積且α和βT是可交換的,這一性質在后面的證明中經常用到,性質4表明多個矩陣的克羅內克積適用結合律.
對于克羅內克積和矩陣乘法,下面稱為混合積(mixed products)的定理是許多有用結論的基礎[7,20,26].
引理1若矩陣A,B,C,D維數的選取能讓下面的運算都有意義,則有

向量交換矩陣在矩陣微分計算和解線性矩陣方程的理論中有重要的應用.在以往不同的文獻中向量交換矩陣常被被表述為不同的形式[7,20,2526],較為常用的表述如下:
定義1約定基本向量ein表示第i個位置上是1其他位置全為0的n維列向量,即有

向量交換矩陣定義如下:

下面給出它的一種新的定義.
定義2基本向量ein的意義如定義1,則向量交換矩陣定義如下:

即向量交換矩陣Pmn是一個mn×mn方陣,以往定義多是采用雙重求和是一種立體的形式,而本文給出新的定義是一個平面的形式,避開了雙重求和符號的使用,顯然較原有定義簡單.基于此定義,有如下的結論:
定理1根據向量交換矩陣Pmn的定義2,下面的兩個結論成立

推證過程中等號由上到下,依次用到了克羅內克積的性質3,性質2,性質1,性質2.結論1證畢.下面驗證結論2,根據克羅內克積的定義及混合積定理可得,

下面給出向量化算子(vector operator)的定義.如果A=[a1,a2,···,an]∈Fm×n,其中aj∈Fm,j=1,2,···,n,將矩陣A從左到右的n個列向量按從上到下的排成堆棧,形成一個mn維的列向量,記為col[A],定義如下:

對于任意的矩陣A∈Fm×n,容易驗證下面的結論col[A]=Pmncol[AT].這也是Pmn命名為向量交換矩陣的原因.
定理2對于矩陣A∈Fm×n,B∈Fp×q,由向量交換矩陣如定義2,可得如下結論:


其中Bi∈F1×q,i=1,2,···,p,j=1,2,···,q表示矩陣B的第i行.根據Pmn定義,性質2和混合積定理,有

下面證明這兩種定義等價性,即有結論:
定理3相關符號約定如上,則有

即這兩種定義是等價的.上面的證明中等式從上至下依次用到性質3,混合積定理,性質1和性質4及克羅內克積的定義.
面給出酉矩陣的定義.如果方陣A滿足AHA=AAH=I,則稱其是酉矩陣.直接計算可驗證下面的結論.如果A和B是酉矩陣(正交矩陣),則A?B也是酉矩陣(正交矩陣).約定σ[B]:={σ1,σ2,···,σn}是矩陣B∈Fm×n奇異值集合.由奇異值的定義和定理2,對矩陣A,B,有下面的結論成立.
定理4若矩陣A∈Cm×n和B∈Cp×q的奇異值集合是

則有σ[A?B]={σiρj|i=1,2,···,n,j=1,2,···,q}=σ[B?A].

本文討論了與克羅內克積相關的向量交換矩陣,給出了它的一個新的定義,并基于新定義證明了向量交換矩陣的一些性質,最后給出了矩陣克羅內克積奇異值的一個新的結論.值得指出的是矩陣方程的求解一直是數值線性代數的一個核心問題,求解線性矩陣方程的新方法也不斷出現[2730],但是如何利用矩陣的克羅內克積本身的豐富結構,來求解相關的線性矩陣方程,例如求解西爾維斯特矩陣方程(Sylvester matrix equation),仍然是一個需要研究的課題[3033].
參考文獻
[1]Jemderson H,Pukelsheim F,Searle S.On the history of the Kronecker product[J].Linear and Multilinear Algebra,1983,14(2):113-120.
[2]Higham N J.Accuracy and Stability of Numerical Algorithms[M].Philadelphia:Siam,1996.
[3]Ding F.Transformations between some special matrices[J].Computers&Mathematics with Applications, 2010,59(8):2676-2695.
[4]Ding J,Liu Y,Ding F.Iterative solutions to matrix equations of the form AiXBi=Fi[J].Computers& Mathematics with Applications,2010,59(11):3500-3507.
[5]Shi Y,Yu B.Output feedback stabilization of networked control systems with random delays modeled by Markov chains[J].IEEE Transactions on Automatic Control,2009,l54(7):1668-1674.
[6]Shi Y,Fang H,Yan M.Kalman filter-based adaptive control for networked systems with unknown parameters and randomly missing outputs[J].International Journal of Robust and Nonlinear Control,2009,19(18):1976-1992.
[7]Graham A.Kronecker Products and Matrix Calculus:With Applications[M].New York:John Wiley&Sons Inc.,1982.
[8]Bentler P,Lee S.Matrix derivatives with chain rule and rules for simple,Hadamard and Kronecker products[J].Journal of Mathematical Psychology,1978,17(3):255-262.
[9]Magnus J,Neudecker H.Matrix differential calculus with applications to simple,Hadamard and Kronecker products[J].Journal of Mathematical Psychology,1985,29(4):474-492.
[10]Ding F,Chen T.Iterative least-squares solutions of coupled sylvester matrix equations[J].Systems&Control Letters,2005,54(2):95-107.
[11]Ding F,Chen T.On iterative solutions of general coupled matrix equations[J].SIAM Journal on Control and Optimization,2006,44(6):2269-2284.
[12]Ding F,Liu P,Ding J.Iterative solutions of the generalized sylvester matrix equations by using the hierarchical identification principle[J].Applied Mathematics and Computation,2008,197(1):41-50.
[13]Xie L,Liu Y,Yang H.Gradient based and least squares based iterative algorithms for matrix equations AXB+CXD=F[J].Applied Mathematics and Computation,2010,217(5):2191-2199.
[14]戴華.矩陣論[M].北京:科學出版社,2001.
[15]姚國柱,段雪峰,廖安平.矩陣方程X=Q+A?(In?X-C)-1A的Hermitian正定解[J].純粹數學與應用數學,2012(2):257-261.
[16]Jodar L,Abou-Kandil H.Kronecker products and coupled matrix Riccati differential systems[J].Linear Algebra and Its Applications,1989,121(2/3):39-51.
[17]Bahuguna D,Ujlayan A,Pandey D N.Advanced type coupled matrix Riccati differential equation systems with Kronecker product[J].Applied Mathematics and Computation,2007,194(1):46-53.
[18]Dehghan M,Hajarian M.An iterative algorithm for solving a pair of matrix equationsAY B=E,CY D=F over generalized centro-symmetric matrices[J].Computers&Mathematics with Applications,2008,56(12): 3246-3260.
[19]Dehghan M,Hajarian M.An iterative algorithm for the reflexive solutions of the generalized coupled Sylvester matrix equations and its optimal approximation[J].Applied Mathematics and Computation, 2008,202(2):571-588.
[19]Steeb W,Hardy Y.Matrix Calculus and Kronecker Product:A Practical Approach to Linear and Multilinear Algebra[M].Singapore:World Scientific Publishing Company,2011.
[20]Loan C.The ubiquitous kronecker product[J].Journal of Computational and Applied Mathematics,2000, 123(1):85-100.
[21]Huhtanen Marko.Real linear Kronecker product operations[J].Linear Algebra and Its Applications, 2006,418(1):347-361.
[22]Steven D,Barel M V.Rank-deficient submatrices of Kronecker products of Fourier matrices[J].Linear Algebra and Its Applications,2007,426(2,3):349-367.
[23]Deo S G,Murty K N,Turner J.Qualitative properties of adjoint Kronecker product boundary value problems[J].Applied Mathematics and Computation,2002,133(2/3):287-295.
[24]張賢達.矩陣分析與應用[M].北京:清華大學出版社,2004.
[25]詹興致.矩陣論[M].北京:高等教育出版社,2008.
[26]鄧勇,黃敬頻.四元數體上一類矩陣方程解的數值方法[J].純粹數學與應用數學,2010,26,(5):706-709.
[27]鄧勇,黃敬頻,杜剛.四元數體上一類矩陣方程的極小范數最小二乘解[J].純粹數學與應用數學, 2010,26,(2):191-192,220.
[28]黃敬頻,于艷.四元數矩陣方程的復轉化及保結構算法[J].純粹數學與應用數學,2008,24,(2):321-326.
[29]黃敬頻.四元數矩陣方程AX+Y A=C的兩種最佳逼近解[J].純粹數學與應用數學,2004,20(2):109-115.
[30]Ding F,Chen T.Gradient based iterative algorithms for solving a class of matrix equations[J].IEEE Transactions on Automatic Control,2005,50(8):1216-1221.
[31]Ding F,Chen T.Iterative least squares solutions of coupled Sylvester matrix equations[J].Systems&Control Letters,2005,54(2):95-107.
[32]Xie L,Ding J,Ding F.Gradient based iterative solutions for general linear matrix equations[J].Computers &Mathematics with Applications,2009,58(7):1441-1448.
[33]Ding J,Liu Y,Ding F.Iterative solutions to matrix equations of form AiXBi=Fi[J].Computers& Mathematics with Applications,2010,59(11):3500-3507.
A new definition of vec-permutation matrix and its applications
Zhang Huamin1,2,Yin Hongcai3
(1.Department of Mathematics and Physics,Bengbu College,Bengbu 233030,China;
2.Control Science and Engineering Research Center,Jiangnan University,Wuxi214122,China;
3.School of Management Science and Engineering,Anhui University of Finance&Economics, Bengbu233000,China)
By using the Kronecker product of the identity matrix and the fundamental vector,a new definition of vec-permutation matrix is presented,which is simpler than the original one.Based on the new definition,the properties of the vec-permutation matrix are discussed.The proof of the equivalence between the new definition and the original one is given.At last,a new result on the singular values of Kroneker products of several matrices is established.
Kronecker product,vec-permutation matrix,vector operator,singular values
O151.2
A
1008-5513(2013)03-0246-09
10.3969/j.issn.1008-5513.2013.03.005
2012-12-01.
國家自然科學基金(60973043);111引智計劃(B12018);蚌埠學院自然科學基金(2011ZR17);安徽高等學校省級自然科學研究項目(KJ2013A183).
張華民(1972-),博士生,講師,研究方向:矩陣方程理論.
2010 MSC:15A69,16A18