矩阵方程AXB+CX~TD=E自反最佳逼近解的迭代算法

本文研究了Sylvester矩阵方程AXB+CXTD=E自反(或反自反)最佳逼近解.利用所提出的共轭方向法的迭代算法,获得了一个结果:不论矩阵方程AXB+CXTD=E是否相容,对于任给初始自反(或反自反)矩阵X1,在有限迭代步内,该算法都能够计算出该矩阵方程的自反(或反自反)最佳逼近解.最后,三个数值例子验证了该算法是有效性的.     

求解矩阵方程组A_1XB_1+C_1XD_1=E_1,A_2XB_2+C_2XD_2=E_2的迭代算法

应用共轭梯度方法和线性投影算子,给出迭代算法求解了线性矩阵方程组A_1XB_1+C_1XD_1=E_1,A_2XB_2+C_2XD_2=E_2在任意线性子空间上的约束解及其最佳逼近.可以证明,当矩阵方程组A_1XB_1+C_1XD_1=E_1,A_2XB_2+C_2XD_2=E_2相容时,所给迭代算法经过有限步迭代可得到矩阵方程组的约束解,极小范数解和最佳逼近.文中的数值例子证实了该算法的有效性.     

一类矩阵方程的埃尔米特自反最小二乘解

利用埃尔米特自反矩阵的表示定理和矩阵的拉直方法,研究了矩阵方程$AX+BY=C$的埃尔米特自反最小二乘问题,进一步,给出了方程在埃尔米特自反矩阵集合中可解的充分必要条件,得到解的一般表达式,最后,对任意给定的一对复矩阵,得到了其相关最佳逼近问题解的表达式.     

利用交替投影算法求解矩阵方程AXB=C的广义中心对称解

利用交替投影算法求解矩阵方程AXB=C的广义中心对称解,当矩阵方程AXB=C不相容时,利用Dykstra's交替投影算法来求其广义中心对称解的最佳逼近,数值结果表明该方法是行之有效的.     

一类Riccati矩阵方程广义自反解的双迭代算法

本文研究了一类Riccati矩阵方程广义自反解的数值计算问题.利用牛顿算法将Riccati矩阵方程的广义自反解问题转化为线性矩阵方程的广义自反解或者广义自反最小二乘解问题,再利用修正共轭梯度法计算后一问题,获得了求Riccati矩阵方程的广义自反解的双迭代算法.拓宽了求解非线性矩阵方程的迭代算法.数值算例表明双迭代算法是有效的.     

线性子空间上求解矩阵方程组A_1XB_1=C_1,A_2XB_2=C_2的迭代算法

应用共轭梯度方法,结合线性投影算子,给出迭代算法求解了线性矩阵方程组A_1XB_1=C_1,A_2XB_2=C_2在任意线性子空间上的约束解及其最佳逼近.当矩阵方程组A_1XB_1=C_1,A_2XB_2=C_2相容时,可以证明,所给迭代算法经过有限步迭代可得到矩阵方程组的约束解、极小范数解和最佳逼近.文中的数值例子证实了该算法的有效性.     

矩阵方程AXB+CXD=F对称解的迭代算法

在共轭梯度思想的启发下,本文给出了迭代算法求解约束矩阵方程AXB+CXD=F的对称解及其最佳逼近.应用迭代算法,矩阵方程AXB+CXD=F的相容性可以在迭代过程中自动判断.当矩阵方程AXB+CXD=F有对称解时,在有限的误差范围内,对任意初始对称矩阵X1,运用迭代算法,经过有限步可得到矩阵方程的对称解;选取合适的初始迭代矩阵,还可以迭代出极小范数对称解.而且,对任意给定的矩阵X0,矩阵方程AXB+CXD=F的最佳逼近对称解可以通过迭代求解新的矩阵方程A(X)B+C(X)D=(F)的极小范数对称解得到.文中的数值例子证实了该算法的有效性.     

矩阵方程ATXA=B的对称广义中心对称解

本文研究了一类矩阵方程AT XA=B的对称广义中心对称解.利用广义奇异值分解和广义逆矩阵,获得了该方程有对称广义中心对称解的充要条件及解的通式,并讨论了解对于已知矩阵的最佳逼近问题,得到了解的表达式.     

矩阵方程的反Hermite自反矩阵问题的解存在条件证明

通过求极值的方法,得到了求矩阵方程AHXA=B与已知矩阵最佳逼近的反H erm ite-自反解的存在的充要条件.     

求矩阵方程AXB+CXD=F的中心对称最小二乘解的迭代算法

该文建立了求矩阵方程AXB+CXD=F的中心对称最小二乘解的迭代算法.使用该算法不仅可以判断该矩阵方程的中心对称解的存在性,而且无论中心对称解是否存在,都能够在有限步迭代计算之后得到中心对称最小二乘解.选取特殊的初始矩阵时,可求得极小范数中心对称最小二乘解.同时,也能给出指定矩阵的最佳逼近中心对称矩阵.     

一四元数矩阵方程组的广义(反)反射解

该文给出了四元数矩阵方程组X_1B_1=C_1,X_2B_2=C2,A_1X_1B_3+A_2X_2B_4=C_b可解的充要条件及其通解的表达式,利用此结果建立了四元数矩阵方程组XB_a=C_a,A_bXB_b=C_b有广义(反)反射解的充要条件及其有此种解时通解的表达式.     

The least squares problem of the matrix equation A1X1B1^T ＋ A2X2B2^T = T

The matrix least squares （LS） problem minx ｜｜AXB^T--T｜｜F is trivial and its solution can be simply formulated in terms of the generalized inverse of A and B. Its generalized problem minx1,x2 ｜｜A1X1B1^T ＋ A2X2B2^T - T｜｜F can also be regarded as the constrained LS problem minx=diag（x1,x2） ｜｜AXB^T -T｜｜F with A = [A1, A2] and B = [B1, B2]. The authors transform T to T such that min x1,x2 ｜｜A1X1B1^T＋A2X2B2^T -T｜｜F is equivalent to min x=diag（x1 ,x2） ｜｜AXB^T - T｜｜F whose solutions are included in the solution set of unconstrained problem minx ｜｜AXB^T - T｜｜F. So the general solutions of min x1,x2 ｜｜A1X1B^T ＋ A2X2B2^T -T｜｜F are reconstructed by selecting the parameter matrix in that of minx ｜｜AXB^T - T｜｜F.     

线性流形上的广义反射矩阵反问题

设 R∈C^m×m 及 S∈C^n×n 是非平凡Hermitian酉矩阵, 即 R^H=R=R^-1≠±I_m ,S^H=S=S^-1≠±I_n.若矩阵 A∈C^m×n 满足 RAS=A, 则称矩阵 A 为广义反射矩阵.该文考虑线性流形上的广义反射矩阵反问题及相应的最佳逼近问题.给出了反问题解的一般表示, 得到了线性流形上矩阵方程AX₂=Z₂, Y₂^H A=W₂^H 具有广义反射矩阵解的充分必要条件, 导出了最佳逼近问题唯一解的显式表示.     

The left and right inverse eigenvalue problems of generalized reflexive and anti-reflexive matrices

Let n×n complex matrices R and S be nontrivial generalized reflection matrices, i.e., R^∗=R=R⁻¹≠±I_n, S^∗=S=S⁻¹≠±I_n. A complex matrix A with order n is said to be a generalized reflexive (or anti-reflexive ) matrix, if RAS=A (or RAS=−A). In this paper, the solvability conditions of the left and right inverse eigenvalue problems for generalized reflexive and anti-reflexive matrices are derived, and the general solutions are also given. In addition, the associated approximation solutions in the solution sets of the above problems are provided. The results in present paper extend some recent conclusions.     

Difference Families in Z2d+1⊕Z2d+1 and Infinite Translation Designs in Z⊕Z

We analyse 3-subset difference families of Z_2d+1⊕Z_2d+1 arising as reductions (mod 2d+1) of particular families of 3-subsets of Z⊕Z. The latter structures, namely perfect d-families, can be viewed as 2-dimensional analogues of difference triangle sets having the least scope. Indeed, every perfect d-family is a set of base blocks which, under the natural action of the translation group Z⊕Z, cover all edges {(x,y),(x′,y′)} such that |x−x′|, |y−y′|≤d. In particular, such a family realises a translation invariant (G,K₃)-design, where V(G)=Z⊕Z and the edges satisfy the above constraint. For that reason, we regard perfect families as part of the hereby defined translation designs, which comprise and slightly generalise many structures already existing in the literature. The geometric context allows some suggestive additional definitions. The main result of the paper is the construction of two infinite classes of d-families. Furthermore, we provide two sporadic examples and show that a d-family may exist only if d≡0,3,8,11 (mod 12).     

On Optimal Feedback Control for Stationary Linear Systems

We study linear-quadratic optimal control problems for finite dimensional stationary linear systems A X+B U=Z with output Y=C X+D U from the viewpoint of linear feedback solution. We interpret solutions in relation to system robustness with respect to disturbances Z and relate them to nonlinear matrix equations of Riccati type and eigenvalue-eigenvector problems for the corresponding Hamiltonian system. Examples are included along with an indication of extensions to continuous, i.e., infinite dimensional, systems, primarily of elliptic type.     

环F2+uF2+vF2上自对偶码的两种构造方法

本文研究环R=F2+uF2+vF2上的自对偶码问题.利用Rn到F3n2的Gray映射及R上的自对偶码C的Gray像为F2上自对偶码,获得了R上任何偶长度的自对偶码存在性的结论.最后,给出了R上两种构造自对偶码的方法.     

On K1 and K2 of Algebraic Surfaces

In this paper we study higher Chow groups of smooth, projective surfaces over a field k of characteristic zero, using some new Hodge theoretic methods which we develop for this purpose. In particular we investigate the subgroup of CH ^r+1 (X,r) with r = 1,2 consisting of cycles that are supported over a normal crossing divisor Z on X. In this case, the Hodge theory of the complement forms an interesting variation of mixed Hodge structures in any geometric deformation of the situation. Our main result is a structure theorem in the case where X is a very general hypersurface of degree d in projective 3-space for d sufficiently large and Z is a union of very general hypersurface sections of X. In this case we show that the subgroup of CH ^r+1 (X,r) we consider is generated by obvious cycles only arising from rational functions on X with poles along Z. This can be seen as a generalization of the Noether–Lefschetz theorem for r = 0. In the case r = 1 there is a similar generalization by Müller-Stach, but our result is more precise than it, since it is geometric and not only cohomological. The case r = 2 is entirely new and original in this paper. For small d, we construct some explicit examples for r = 1 and 2 where the corresponding higher Chow groups are indecomposable, i.e. not the image of certain products of lower order groups. In an appendix Alberto Collino constructs even more indecomposable examples in CH ³ (X,2) which move in a one-dimensional family on the surface X.Contribution to appendix.     

The Real Solutions to a System of Quaternion Matrix Equations with Applications

In this article we establish necessary and sufficient conditions for the existence and the expressions of the general real solutions to the classical system of quaternion matrix equations A ₁ XB ₁ = C ₁, A ₂ XB ₂ = C ₂. Moreover, formulas of the maximal and minimal ranks of four real matrices X ₁, X ₂, X ₃, and X ₄ in solution X = X ₁ + X ₂ i + X ₃ j + X ₄ k to the system mentioned above are derived. As applications, we give necessary and sufficient conditions for the quaternion matrix equations A ₁ XB ₁ = C ₁, A ₂ XB ₂ = C ₂, A ₃ XB ₃ = C ₃ to have common real solutions. In addition, the maximal and minimal ranks of four real matrices E, F, G, and H in the common generalized inverse of A ₁ + B ₁ i + C ₁ j + D ₁ k and A ₂ + B ₂ i + C ₂ j + D ₂ k, which can be expressed as E + Fi + Gj + Hk are also presented.