岭回归中确定K值的一种方法

本文给出了岭估计中确定K值的一种新方法,这种方法改进了Hoerl和Kennard的相应方法。     

具有四个不同特征值的竞赛矩阵

设A是n阶竞赛矩阵，k是非负整数。文[3]刻划了恰好有三个不同特征值的n阶竞赛矩阵，文[4]刻划了恰好有四个不同特征值并且0作为一个一重特征值的n阶竞赛矩阵。在这篇文章中我们主要研究了两个问题：（1）讨论当k是A的特征值时A的性质。（2）刻划恰好有四个不同特征值并且k作为一个一重特征值的全部n阶竞赛矩阵。     

带比例关系的实带状矩阵特征值反问题

研究了通过矩阵A的顺序主子矩阵A_((k))=(aij)_(i,j=1)^(n-k+1)的特征值{λ_i(n-k+1)的特征值{λ_i^((k)))}_(i=1)((k)))}_(i=1)^(n-k+1)k=1,2,…,r+1来构造一个带比例关系的实带状矩阵的特征值反问题.对当特征值{λ_i(n-k+1)k=1,2,…,r+1来构造一个带比例关系的实带状矩阵的特征值反问题.对当特征值{λ_i^((k))}_(i=1)((k))}_(i=1)^(n-k+1)中有多重特征值出现时,应当如何来构造这类矩阵进行了讨论,并给出了问题的具体算法及数值例子.     

AN INVERSE EIGENVALUE PROBLEM FOR JACOBI MATRICES

Let T1,n be an n x n unreduced symmetric tridiagonal matrix with eigenvaluesand is an (n - 1) x (n - 1) submatrix by deleting the kth row and kth column, k = 1, 2,be the eigenvalues of T1,k andbe the eigenvalues of Tk+1,nA new inverse eigenvalues problem has put forward as follows: How do we construct anunreduced symmetric tridiagonal matrix T1,n, if we only know the spectral data: theeigenvalues of T1,n, the eigenvalues of Ti,k-1 and the eigenvalues of Tk+1,n?Namely if we only know the data: A1, A2, An,how do we find the matrix T1,n? A necessary and sufficient condition and an algorithm ofsolving such problem, are given in this paper.     

THE OPPENHEIM-TYPE INEQUALITIES FOR THE HADAMARD PRODUCT OF M-MATRIX AND POSITIVE DEFINITE MATRIX

For the lower bound about the determinant of Hadamard product of A and B, where A is a n x n real positive definite matrix and B is a n x n M-matrix, Jianzhou Liu [SIAM J. Matrix Anal. Appl., 18(2)(1997).. 305-311] obtained the estimated inequality as follows     

Principal Submatrices of a Hermitian Matrix

Suppose k 1 ,<, k m and n are positive integers such that k 1 + … + k m h n . We characterize those k i × k i Hermitian matrices A i , i = 1, < , m that can appear as diagonal blocks of an n × n Hermitian matrix C with prescribed eigenvalues. The characterization will be given in terms of the eigenvalues of C and A i , i = 1, <, m . Our results extend those of Thompson and Freede, Horn, Fan and Pall.     

矩阵方程AX=B的双反对称最佳逼近解

本文主要讨论下而两个问题并得到相关结果:问题Ⅰ:给定A ∈ R~(k×n),B ∈ R~(k×n),求X ∈ BASR~(n×n),使得AX=B.问题Ⅱ:给定X* ∈R~(n×n),求X使得‖X-X~*‖=minX∈S_E‖X-X~*‖,其中S_E是问题Ⅰ的解集合,‖·‖是Frobenius范数.通过对上述问题的讨论给出了问题Ⅰ解存在的充分必要条件和其解的一般表达式同时给出了问题Ⅱ的解,算法,和数值例子.     

Hermitian型分块矩阵的极值秩

证明了可选取矩阵X和Hermitian矩阵Z,使得下面的Hermitian型分块矩阵(A XX*Z)取得它的极大秩和极小秩,这里A*=A∈Cm×m是一个已知的复矩阵,X∈Cm×k和Z*=Z∈Ck×k是两个任意的复矩阵.     

A Class of Constrained Inverse Eigenproblem and Associated Approximation Problem for Symmetric Reflexive Matrices

Let S∈Rn×n be a symmetric and nontrival involution matrix. We say that A∈E R n×n is a symmetric reflexive matrix if AT = A and SAS = A. Let S R r n×n(S)={A|A= AT,A = SAS, A∈Rn×n}. This paper discusses the following two problems. The first one is as follows. Given Z∈Rn×m (m < n),∧= diag(λ1,...,λm)∈Rm×m, andα,β∈R withα<β. Find a subset (?)(Z,∧,α,β) of SRrn×n(S) such that AZ = Z∧holds for any A∈(?)(Z,∧,α,β) and the remaining eigenvaluesλm 1 ,...,λn of A are located in the interval [α,β], Moreover, for a given B∈Rn×n, the second problem is to find AB∈(?)(Z,∧,α,β) such that where ||.|| is the Frobenius norm. Using the properties of symmetric reflexive matrices, the two problems are essentially decomposed into the same kind of subproblems for two real symmetric matrices with smaller dimensions, and then the expressions of the general solution for the two problems are derived.     

The role of eigenvalues in linear feature selection theory

A family of examples is constructed to show that if B is the k × n matrix (I_k|Z)U, where U is an n × n orthogonal matrix, then the eigenvalues of U do not affect the value of divergence D_B in the space of reduced dimension.     

NOTE ON REGULAR D—OPTIMAL MATRICES

Let A be a j x d (0,1) matrix. It is known that if j = 2k - 1 is odd, then det(AAT) ≤ (j+1)((j+1)d/4j)j; if j is even, then det(AAT) ≤ (j+1)((j+2)d/4(j+1))j. A is called a regular D-optimal matrix if it satisfies the equality of the above bounds. In this note, it is proved that if j = 2k - 1 is odd, then A is a regular D-optimal matrix if and only if A is the adjacent matrix of a (2k - 1, k, (j + l)d/4j)-BIBD; if j = 2k is even, then A is a regular D-optimal matrix if and only if A can be obtained from the adjacent matrix B of a (2k + 1,k + 1,(j + 2)d/4(j +1))-BIBD by deleting any one row from B. Three 21 x 42 regular D-optimal matrices, which were unknown in [11], are also provided.     

ON THE MULTIPLICATIVE INVERSE EIGENVALUE PROBLEMS

We present new sufficient conditions on the solvability and numerical methods for the following multiplicative inverse eigenvalue problem: Given an n x n real matrix A and n real numbers λ1 , λ2,..., λn, find n real numbers c1, c2,..., cn such that the matrix diag(c1,c2,...,cn)A has eigenvalues λ1,λ2,..., λn.     

三次幻方的构作

给出2k维m阶t次幻方及m模方阵,m模列满秩矩阵,模线,m经典模线集和t次m模基因阵的概念,并用矩阵法和组合法初步研究了t次幻方特别是三次幻方的构作.证明:(i)若存在2k阶t次m模基因阵,则存在2k维m阶t次幻方;(ii)若N=P1α1P2α2…PSαS为N的标准分解式,iα≥3,Piiα≥16(1≤i≤S),则存在二维N阶三次幻方;(iii)若存在二维偶m阶2t+1次幻方和二维n阶2t次幻方,则存在二维mn阶2t+1次幻方;(iv)若存在二维m阶和n阶t次幻方,则存在二维mn阶t次幻方;(v)当t≥3时,不存在二维单偶数阶t次幻方.     

Integral and distance integral Cayley graphs over generalized dihedral groups

Journal of Algebraic Combinatorics - A graph is said to be integral (resp. distance integral) if all the eigenvalues of its adjacency matrix (resp. distance matrix) are integers. Let H be a finite...     

Gaussian fluctuations of eigenvalues in log-gas ensemble: Bulk case I

We study the central limit theorem of the k-th eigenvalue of a random matrix in the log-gas ensemble with an external potential V = q2mx2 m. More precisely, let Pn(d H) = Cne-nTrV(H)dH be the distribution of n × n Hermitian random matrices, ρV(x)dx the equilibrium measure, where Cnis a normalization constant, V(x) = q2mx2m with q2m=Γ(m)Γ(12)/Γ(2m+1/2), and m ≥ 1. Let x1 ≤···≤ xnbe the eigenvalues of H. Let k := k(n) be such that k(n)/n∈ [a, 1- a] for n large enough, where a ∈(0,12).Define G(s) :=∫s-1ρV(x)dx,- 1 ≤ s ≤ 1,and set t := G-1(k/n). We prove that, as n →∞,xk- t log n1/2 2π21/2nρV(t)→ N(0, 1)in distribution. Multi-dimensional central limit theorem is also proved. Our results can be viewed as natural extensions of the bulk central limit theorems for GUE ensemble established by J. Gustavsson in 2005.     

k-控制阵

给出k-控制阵的定义,讨论k-控制阵与k-圈控制阵的关系,指出k-控制阵的周期是k的一个因子,指数不大于(n-1)k m.     

限制性集合矩阵的计数公式

In this paper, we consider the counting problem of matrix of set (Aij)k×n which satisfies k∪i=1 n∪j=1 Aij={a1,a2,…,am} and other conditions, and obtain some computational formulas which extend all the results in [1].     

基于矩阵幂运算的重特征值存在性定理

对于判断矩阵重特征值的存在性问题,运用“若λ是矩阵A的特征值,则入“是Ak的特征值”这一性质,通过矩阵的迹与特征值的关系,得到了实数域上矩阵重特征值的存在性定理并给出了证明．定理实现了“由矩阵幂运算来判断矩阵重特征值的存在性”这样一个计算过程,对讨论矩阵特征值问题具有一定的启示意义．     

On the General Algebraic Inverse Eigenvalue Problems

A number of new results on sufficient conditions for the solvability and numerical algorithms of the following general algebraic inverse eigenvalue problem are obtained: Given $n+1$ real $n\times n$ matrices $A=(a_{ij}),A_k=(a_{ij}^{(k)})(k=1,2,\cdots,n)$ and $n$ distinct real numbers $\lambda_1,\lambda_2,\cdots,\lambda_n,$ find $n$ real number $c_1,c_2,\cdots,c_n$ such that the matrix $A(c)=A+\sum\limits_{k=1}^{n}c_k A_k$ has eigenvalues $\lambda_1,\lambda_2,\cdots,\lambda_n.$     

代数特征值反问题Hermite情形可解的充分条件

１　引言及主要结果 本论文将要讨论如下问题［２,４］： 问题ＨＧ给定ｎ＋１个Ｈｅｒｍｉｔｅ矩阵Ａ＝（ａｉｊ）ｎ×ｎ和Ａｋ＝Ｓ和ｎ个实数　,求个实数ｃ１,…,ｃｎ,使得Ａ（ｃ）＝ ．的特征值为 对于上述问题,有解的充分条件已有许多研究结果,如［２,４,６］．下面将利用Ｂｒｏｕｗｅｒ不动点定理给出新的充分条件． 本文的符号和定义如下： 对任意ｎ阶Ｈｅｒｍｉｔｅ矩阵Ｂ＝（ｂｉｊ）,记Ｂ（０）＝Ｂ－ｄｉａｇ（ｂ１１,ｂ２２,…,ｂｎｎ）,ρ（Ｂ）表示Ｂ的谱半径, ｛λ（Ｂ）｝表示Ｂ的特征值（谱）集合,且设　表…