满秩仿Hermite多项式矩阵的J一谱分解计算

基于仿等价变换实现了满秩仿Hermite多项式矩阵的J-谱分解计算.对于一个满秩的仿Hermite多项式矩阵,首先用仿等价变换将其变换为单模的满秩仿Hermite多项式矩阵,进一步再用仿等价变换将其变换为常数满秩矩阵,最后变换为J-矩阵.将这些变换矩阵积累起来,得到满秩仿Hermite多项式矩阵的J-谱分解.在作者发展的多项式矩阵运算程序库的基础上,给出了实现所提出的计算方法的算法.数例表明,该方法是有效的.     

酉变换步进搜索求根MUSIC算法的研究与仿真

求根MUSIC算法具有比基本MUSIC算法更好的性能.但其实现难度极大而基本MUSIC算法的计算量很大,时延较长.酉变换步进搜索求根MUSIC算法依据求根MUSIC算法的原理采用步进搜索的方法,将摹本MUSIC算法的复对称矩阵分解转化为实对称矩阵分解,并将复数域的矩阵计算转化为复数域的只有乘法和加法的多项式运算,从而大大降低算法的计算量和实现难度.在matlab上的仿真也表明:酉变换步进搜索求根MUSIC算法是一种实现容易,计算量较小,效果较好的一种DOA估计方法.     

渐进迭代逼近方法的数值分析

基于一类特殊矩阵的幂级数展开,推导了渐进迭代逼近方法和代数插值方法的等价性.在此基础上,针对PIA方法中因病态矩阵而导致收敛速度过慢的问题,通过矩阵QR分解引入变换矩阵,再优化迭代矩阵的谱半径,来加速PIA方法收敛;相对于因不同的参数化而导致计算效率的不确定性问题,采用向心加速参数化、优化配置矩阵来确保计算效率.最后通过数值实例验证了理论推导的正确性和文中方法的有效性.     

奇异线性系统最优估计的多项式方法

基于多项式ARMA新息模型方法提出了随机奇异线性离散时间系统的稳态最优估计.估值器的增益矩阵是通过新息分析和射影方法推得;其计算归结为求解一个多项式方程和谱分解.这一结果是最优估计多项式方法在奇异系统中的应用.     

解决文本聚类集成问题的两个谱算法

聚类集成中的关键问题是如何根据不同的聚类器组合为最终的更好的聚类结果. 本文引入谱聚类思想解决文本聚类集成问题, 然而谱聚类算法需要计算大规模矩阵的特征值分解问题来获得文本的低维嵌入, 并用于后续聚类. 本文首先提出了一个集成算法, 该算法使用代数变换将大规模矩阵的特征值分解问题转化为等价的奇异值分解问题, 并继续转化为规模更小的特征值分解问题; 然后进一步研究了谱聚类算法的特性, 提出了另一个集成算法, 该算法通过求解超边的低维嵌入, 间接得到文本的低维嵌入. 在TREC和Reuters文本数据集上的实验结果表明, 本文提出的两个谱聚类算法比其他基于图划分的集成算法鲁棒, 是解决文本聚类集成问题行之有效的方法.     

线性最优控制系统的频域综合方法(多输入-多输出系统)

本文用频域方法讨论了二次性能指标下的多输入-多输出线性系统最优控制的综合问题. 利用多项式矩阵的谱分解方法,把求解最优综合函数的问题归结为求解两个多项式矩阵的 Diophantine方程,从而给出了该问题解的频域形式.     

基于多项式变换的二维整型离散余弦变换快速算法

提出了一种基于多项式变换的二维整型离散余弦变换（DCT）快速算法,利用多项式变换将二维DCT变换的计算转化为一系列一维DCT变换及其变换系数的求和运算,减少了乘法和加法的计算量;利用提升矩阵,实现了整型DCT变换,进一步提高了运算效率的同时,使信号可精确重构。     

基于QR分解的方阵特征多项式数值算法

借鉴QR分解的概念,探讨如何利用QR分解法求一个方阵的特征多项式的数值算法。这就为用Matlab编程求解矩阵的特征多项式提供了条件。通过对三个不同类型矩阵的实例计算验证了该数值方法能够较好地求解一般方阵的特征多项式,比用通常的求带变量的行列式的方法和基于迹的算法要相对简单些,特别是对于高阶矩阵,其优势更加明显,且易在计算机上用Matlab编程实现,从而说明了该算法具有较高的实用价值。     

基于Toeplitz矩阵的酉变换波达角估计算法

为提高Toeplitz矩阵重构算法的估计性能、降低计算量,提出了基于Toeplitz矩阵的酉变换DOA估计算法UHT-MUSIC。该算法在保持估计性能不变的前提下,先将Toeplitz型协方差矩阵变换为Hermition矩阵,然后利用酉变换将其转换为实数矩阵。在此基础上利用MUSIC算法进行DOA估计,其特征值分解及谱峰搜索的计算量降低到同条件下TOEP-MUSIC算法的1/4。同时该算法还有效降低了信源的相关系数,从而提高了算法的分辨性能。仿真实验验证了该算法的正确性。     

求解弱非比例阻尼系统实模态解的阻尼矩阵摄动法

提出了一种求解弱非比例阻尼振动系统实模态解的摄动方法和将非比拟阻尼矩阵分解为比例阻尼矩阵和余项阻尼矩阵的方法.对于弱非比例阻尼振动系统,通过同时对阻尼矩阵和响应矢量进行小参数摄动,将原非比例阻尼系统分解为一系列的比例阻尼振动系统,在此基础上用正则模态变换将各阶比例阻尼的摄动方程解耦,从而求得原非比例阻尼振动系统的近似解析解.计算实例表明,此方法的结果与数值计算结果十分吻合.     

The operational matrix of polynomial series transformation

This paper introduces the operational matrix of polynomial series transformation T that may be applied to transform any polynomial series basis vector to the Taylor polynomials. The matrix is determined for most commonly used polynomial series expansions, such as the Chebyshev, the Laguerre, the Legendre and the Hermite. Using the polynomial series transformation matrix, the corresponding operational matrix of integration of a polynomial series, may easily be determined. Finally, it is shown that all the approximate methods using polynomial-based operational matrices of integration may be connected to the Taylor-series method.     

The equivalence of the constrained Rayleigh quotient and Newton methods for matrix polynomials expressed in different polynomial bases along with the confluent case

We show that using the constrained Rayleigh quotient method to find the eigenvalues of matrix polynomials in different polynomial bases is equivalent to applying the Newton method to certain functions. We find those functions explicitly for a variety of polynomial bases including monomial, orthogonal, Newton, Lagrange and Bernstein bases. In order to do so, we provide explicit symbolic formulas for the right and left eigenvectors of the generalized companion matrix pencils for matrix polynomials expressed in those bases. Using the properties of the Newton basis, we also find two different formulas for the companion matrix pencil corresponding to the Hermite interpolation. We give pairs of explicit LU factors associated with these pencils. Additionally, we explicitly find the right and left eigenvectors for each of these pencils.     

Non-negative and sparse spectral clustering

Spectral clustering aims to partition a data set into several groups by using the Laplacian of the graph such that data points in the same group are similar while data points in different groups are dissimilar to each other. Spectral clustering is very simple to implement and has many advantages over the traditional clustering algorithms such as k-means. Non-negative matrix factorization (NMF) factorizes a non-negative data matrix into a product of two non-negative (lower rank) matrices so as to achieve dimension reduction and part-based data representation. In this work, we proved that the spectral clustering under some conditions is equivalent to NMF. Unlike the previous work, we formulate the spectral clustering as a factorization of data matrix (or scaled data matrix) rather than the symmetrical factorization of the symmetrical pairwise similarity matrix as the previous study did. Under the NMF framework, where regularization can be easily incorporated into the spectral clustering, we propose several non-negative and sparse spectral clustering algorithms. Empirical studies on real world data show much better clustering accuracy of the proposed algorithms than some state-of-the-art methods such as ratio cut and normalized cut spectral clustering and non-negative Laplacian embedding.     

A recursive method for solving unconstrained tangentialinterpolation problems

An efficient recursive solution is presented for the one-sided unconstrained tangential interpolation problem. The method relies on the triangular factorization of a certain structured matrix that is implicitly defined by the interpolation data. The recursive procedure admits a physical interpretation in terms of discretized transmission lines. In this framework the generating system is constructed as a cascade of first-order sections. Singular steps occur only when the input data is contradictory, i.e., only when the interpolation problem does not have a solution. Various pivoting schemes can be used to improve numerical accuracy or to impose additional constraints on the interpolants. The algorithm also provides coprime factorizations for all rational interpolants and can be used to solve polynomial interpolation problems such as the general Hermite matrix interpolation problem. A recursive method is proposed to compute a column-reduced generating system that can be used to solve the minimal tangential interpolation problem     

A new method for computing a column reduced polynomial matrix

A new algorithm is presented for computing a column reduced form of a given full column rank polynomial matrix. The method is based on reformulating the problem as a problem of constructing a minimal polynomial basis for the right nullspace of a polynomial matrix closely related to the original one. The latter problem can easily be solved in a numerically reliable way. Three examples illustrating the method are included.     

Estimating 3D shape from degenerate sequences with missing data

Reconstructing a 3D scene from a moving camera is one of the most important issues in the field of computer vision. In this scenario, not all points are known in all images (e.g. due to occlusion), thus generating missing data. On the other hand, successful 3D reconstruction algorithms like Tomasi & Kanade’s factorization method, require an orthographic model for the data, which is adequate in close-up views. The state-of-the-art handles the missing points in this context by enforcing rank constraints on the point track matrix. However, quite frequently, close-up views tend to capture planar surfaces producing degenerate data. Estimating missing data using the rank constraint requires that all known measurements are “full rank” in all images of the sequence. If one single frame is degenerate, the whole sequence will produce high errors on the reconstructed shape, even though the observation matrix verifies the rank 4 constraint. In this paper, we propose to solve the structure from motion problem with degenerate data, introducing a new factorization algorithm that imposes the full scaled-orthographic model in one single optimization procedure. By imposing all model constraints, a unique (correct) 3D shape is estimated regardless of the data degeneracies. Experiments show that remarkably good reconstructions are obtained with an approximate models such as orthography.     

The Kalman-Yakubovich-Popov Lemma in a behavioural framework

The classical Kalman-Yakubovich-Popov Lemma provides a link between dissipativity of a system in state-space form and the solution to a linear matrix inequality. In this paper we derive the KYP Lemma for linear systems described by higher-order differential equations. The result is an LMI in terms of the original coefficients in which the dissipativity problem is posed. Subsequently we study the connection between dissipativity and spectral factorization of polynomial matrices. This enables us to derive a new algorithm for polynomial spectral factorization in terms of an LMI in the coefficients of a polynomial matrix.     

Incorporating non-motion cues into 3D motion segmentation

We address the problem of segmenting an image sequence into rigidly moving 3D objects. An elegant solution to this problem in the case of orthographic projection is the multibody factorization approach in which the measurement matrix is factored into lower rank matrices. Despite progress in factorization algorithms, their performance is still far from satisfactory and in scenes with missing data and noise, most existing algorithms fail.In this paper we propose a method for incorporating 2D non-motion cues (such as spatial coherence) into multibody factorization. We show the similarity of the problem to constrained factor analysis and use the EM algorithm to find the segmentation. We show that adding these cues improves performance in real and synthetic sequences.