含缺失数据线性模型的线性不等式约束EM算法

研究具有缺失数据的线性模型的回归参数在一般线性不等式A₀β≥a约束下的极大似然估计问题;提出了一般线性不等式约束下的EM算法并且证明了此算法的收敛性.     

二阶超线性差分方程周期解与次调和解的存在性

应用临界点理论, 为研究差分方程周期解与次调和解的存在性和多重性提供了一种新方法. 对二阶差分方程 D²x_n-1+f (n, x_n)=0, 当f(t, z)在0点及无穷远点为超线性增长时, 上述问题得到某些新结果.     

拟线性抛物方程组具有界面外推的并行本性差分方法

构造了拟线性抛物型方程组初边值问题的一类具有界面外推的并行本性差分格式. 为给出子区域间界面上的值或者与界面相邻点处的值,给出了两类时间外推的方式, 得到了二阶精度无条件稳定的并行差分格式. 并且不作启示性假定,证明了所构造的并行差分格式的离散向量解的存在性和 唯一性. 而且在格式的离散向量解对原始问题的已知离散数据连续依赖的意义下, 证明了并行差分格式的解按离散W^(2,1)₂(Q_Δ)范数是无条件稳定的.最后证明了具有界面外推的并行本性差分格式的离散向量解收敛到原始拟线性抛物问题的唯一广义解. 给出了数值例子,数值结果表明所构造的格式是无条件稳定的, 具有二阶精度,且具有高度并行性.     

乘法酉算子的Kac-系统的刻画

设V为可分Hilbert空间H上的乘法酉算子,V对应B(H)的两个子代数A(V)和V在满足V₂=I的条件下,得到Baaj与Skandalis主要定理的充要条件：即V有Kac-系统当且仅当这两个C^*-代数乘积的线性闭包为紧算子空间;同时还得到一对量子群.     

计算Henon方程多个正解的分歧方法

首先应用分歧方法给出计算Henon方程边值问题D₄对称正解的3种算法, 然后以Henon方程中的参数r为分歧参数, 在D₄对称正解解枝上用扩张系统方法求出对称破缺分歧点, 进而用解枝转接方法计算出其他具有不同对称性质的正解.     

交换环上的典型群在一般线性群中的扩群

对任一有1的交换环R, 给出了R上的酉群U_2nR(n≥ 5,含辛群, 正交群和标准酉群) 在R上一般线性群GL_2n R 中扩群的完整刻画.     

G-旋模型观测量代数间的C*-指标

在取值于有限群G的二维格子旋系统模型中, 可以定义场代数F. 群G的Double代数D(G), 进而由子群H决定的子Hopf代数D(G;H), 在F上有自然作用, 使得F成为模代数. 给出F的D(G; H)-不变子空间A_H的具体结构, 通过构造A_H到A_G的条件期望γ_G的拟基, 得到γ_G的C^*-指标, 等于子群H在G中的指标.     

带强迫项的二阶非线性椭圆方程的区域振动准则

对二阶非线性椭圆型方程∑ _i,j=1ⁿ D_i[A_ij(x)D_jy]+∑_i=1ⁿ b_i(x)D_iy+q(x)f(y)=e(x)建立了若干新的振动准则, 所得结果仅依赖于方程在外区域Ω С Rⁿ的一个子区域序列的信息而有别于已知的大多数结论.     

Sobolev圆盘代数上的乘法算子

研究一类由单位圆盘D上的Sobolev空间W^2,2(D)中的解析函数构成的代数, 称之为Sobolev圆盘代数, 给出了其上的有界线性乘法算子M_f的基本性质, 刻画了乘法算子M_f的换位子代数, 证明了A′(M_f)是交换的当且仅当M_f^*是指标为1的Cowen-Douglas算子.     

作用在有限线性空间上基柱为3D4(q) 的几乎单群

旗传递线性空间的分类完成以后, 人们开始关注线传递线性空间. 线传递线性空间可以分为非点本原和点本原两种情形. 根据 Delandtsheer-Doyen 理论, 非点本原线传递分类比较容易解决. 而点本原的情形, 根据 O''Nan-Scott 理论和 Camina 的一些前期工作, 又可以分成基柱为初等交换群或非交换单群两种情形. 本文考虑 T 是非交换单群, T≤ G≤ Aut(T) 且 G 线传递作用在有限线性空间上的情形. 并获得了一些有用的引理. 特别地, 证明了当 T 同构于 ³D₄(q) 时, T 是线传递的, 这里 q 是素数 p 的方幂.     

Hilbert空间上线性算子广义逆A(2)T,S的存在性及其表示式

设H₁和H₂是两个Hilbert空间, B(H₁,H₂)表示从H₁到H₂的所有有界线性算子的集合, T和S分别是H₁和H₂的两个闭子空间. 如果存在线性算子X ∈ B(H₂,H₁)满足XAX=X, R(X)=T, N(X)=S,则称X为线性算子$A$的具有指定像空间T和零空间S的外逆,记为A⁽²⁾_T,S. 该文进一步研究了线性算子广义逆A⁽²⁾_T,S存在的若干等价条件及其性质,建立了算子广义逆A⁽²⁾_T,S的表示形式.     

一般线性混合模型中的最佳线性无偏估计和谱分解估计

该文在一般线性混合模型中, 研究了固定和随机效应线性组合的估计问题.对观测向量的协方差阵可以为奇异矩阵情形下,导出了该组合的最佳线性无偏估计,并证明了它的唯一性.在一般线性混合模型的特例, 三个小域模型下, 得到了小域均值u_i 和方差分量的谱分解估计. 进而, 获得了基于谱分解估计的两步估计均方误差的二阶逼近.     

Minimal Determinants and Lattice Inequalities

Some results of P. McMullen on determinants of sublattices ofZ^d induced by rational subspaces are generalized to arbitrarylattices. As an application, we obtain an equality for the minimaldeterminants introduced by J. M. Wills, namely D_t(L) = D_d(L)D_d–1((L*).Using an inequality of Lagarias, Lenstra and Schnorr, we generalizetwo isoperimetric inequalities withlattice constraints by Bokowski,Hadwiger and Wills, and Hadwiger, respectively, to arbitrarylattices.     

On exact D-optimal designs for regression models with correlated observations

Let ^* be an exact D-optimal design for a given regression model Y = X + Z . In this paper sufficient conditions are given for sesigning how the covariance matrix of Z may be changed so that not only ^* remains D-optimal but also that the best linear unbiased estimator (BLUE) of stays fixed for the design ^*, although the covariance matrix of Z * is changed. Hence under these conditions a best, according to D-optimality, BLUE of is known for the model with the changed covariance matrix. The results may also be considered as determination of exact D-optimal designs for regression models with special correlated observations where the covariance matrices are not fully known. Various examples are given, especially for regression with intercept term, polynomial regression, and straight-line regression. A real example in electrocardiography is treated shortly.     

Similarity Solutions of a Non-linear Diffusion Equation

In several physical contexts the equations for the dispersionof a buoyant contaminant can be approximated by the Erdogan-Chatwin(1967) equation {dot}c = {dot}_y{[D_o + ({dot}_yc)²D₂]{dot}_yc}. Here it is shown that in the limit of strong non-linearity (i.e.D_o = 0) there are similarity solutions for a concentration jumpand for a finite discharge. A stability analysis for the latterproblem involves a new family of orthogonal polynomials Y_n(z)where (1 – z⁴)Y – 6z³Y + n(n + 5)z² Y_n = 0 and the degree n is restricted to the values 0, 1, 4, 5, 8,9,.... A numerical solution of the Erdogan-Chatwin equationis given which describes the transition between the non-linearand linear (Gaussian) similarity solutions.     

Generalized Ramsey Theory for Graphs V. the Ramsey Number of a Digraph

Continuing this series of papers on generalized Ramsey theoryfor graphs, we define the Ramsey number r(D_l, D₂) of two digraphsD₁ and D₂ as the minimum p such that every 2-colouring of thearcs (directed lines) of DK_P (the complete symmetric digraphof order p) contains a monochromatic D₁ or D₂. It is shown(Theorem1) that this number exists if and only if D₁ or D₂ is acyclic.Then r(D), the diagonal Ramsey number of a given acyclic digraphD, is defined as r(D, D). Notation: D' is the converse of D,GD is the underlying graph of D, DG is the symmetric digraphof G, and T_p is the transitive tournament of order p. Let r(m,n) be the traditional Ramsey number of the two complete graphsK_m and K_n. Finally, let S_n be the star with n arcs from onepoint u to n points v_i. Assuming the Ramsey numbers under discussionexist, we prove the following results: THEOREM 2. r(D₁, D₂) = r(D₁' D₂'). THEOREM 3. r(D₁, D₂) r(GD₁, GD₂). THEOREM 4. r(D₁, D₂) r(T_P1, T_P2) if both D₁ and D₂ (with p₁and p₂ points respectively) are acyclic. THEOREM 5. r(T_m, T_n) = r(m, n). THEOREM 6. r(m, ri) r(T_m, DK_n) r(2^m–1, n). THEOREM 7. r(S_m, S_n) = r(S_m, S_n') = m+n. Finally, we establish all Ramsey numbers r(D₁, D₂) for digraphswithout isolates and with less than four points, and all diagonalRamsey numbers r(D) of acyclic digraphs without isolates withless than five points.     

Reduced-order filtering with energy-to-peak performance for discrete-time Markovian jumping systems

In this paper, the l₂–l (energy-to-peak) performanceof the discrete-time Markovian jump linear system is investigated.The jump parameters are modelled by a discrete-time Markov process.Furthermore, we study the l₂–l reduced-order filteringproblem for the Markovian jump linear system. A reduced-orderfilter with the same randomly jumping parameters is proposedwhich can make the error systems with Markovian jump parametersstochastically stable with a prescribed l₂–lperformance.Sufficient conditions in terms of linear matrix inequalities(LMIs) and a coupling non-convex rank constraint are derivedfor the existence of a solution to the reduced-order filteringproblems. A numerical example is given to illustrate the designprocedures.     

Estimation of an unobservable parameters in the type II constraints in singular linear regression model

In regular linear regression model with the type II constraints the unobservable parameters in the constraints can be estimated by the BLUE of the observable parameters. The BLUE respects the constraints. If the estimator of the observable parameters which is the BLUE in the model without constraints is used, then the estimator of the unobservable parameters is the same.     

A Class of Pencils of Matrices

A linear machine is one in which the time dependent input yis related to the output z by P(D). z = S(D). y where P andS are polynomials in D = d/dt with constant coefficients. Fornumerical computation it is necessary to replace this relationby a set of simultaneous first order differential equationsand this paper shows how to construct such equations by methodswhich extend the results of Gilder (1961). Attention is restrictedto those sets of equations that are of a special form (see (1))which is characterized by the matrix operating on the dependentvariables. This matrix forms a pencil, being linear in D, andthree theorems are given to show how such matrix pencils maybe constructed from the polynomials. The theorems also statethat any matrix pencil with the required properties can be transformedinto the canonical forms given in the theorems by pre- and post-multiplicationby suitable constant non-singular matrices. Thus the variablesof any set of equations having the required properties are linearcombinations of the variables of the equations given by thetheorems. In the paper it is assumed that the degree of P(D)is greater than that of S(D), as otherwise z would be replacedby z₁+Q(D) . y, where Q is the quotient of S(D)/P(D). Also,as the algebriac manipulations are independent of the natureof the polynomials, D is replaced by an indeterminate x andthe coefficients considered to be from an arbitrary field. Fortechnical reasons we rename y and z, y_o and y_n–_m respectively.