张量鲁棒主成分分析的增广拉格朗日交替方向方法

张量的鲁棒主成分分析是将未知的一个低秩张量与一个稀疏张量从已知的它们的和中分离出来.因为在计算机视觉与模式识别中有着广阔的应用前景,该问题在近期成为学者们的研究热点.本文提出了一种针对张量鲁棒主成分分析的新的模型,并给出交替方向极小化的求解算法,在求解过程中给出了两种秩的调整策略.针对低秩分量本文对其全部各阶展开矩阵进行低秩矩阵分解,针对稀疏分量采用软阈值收缩的策略.无论目标低秩张量为精确低秩或近似低秩,本文所提方法均可适用.本文对算法给出了一定程度上的收敛性分析,即算法迭代过程中产生的任意收敛点均满足KKT条件.如果目标低秩张量为精确低秩,当迭代终止时可对输出结果进行基于高阶奇异值分解的修正.针对人工数据和真实视频数据的数值实验表明,与同类型算法相比,本文所提方法可以得到更好的结果.     

Singular Value Decomposition Solution of the Schrödinger Equation in the Presence of Exchange Terms

A new numerical method of solving integro-differential equations appearing in the theory of atomic and nuclear scattering systems has been devised. It is termed Singular Value Decomposition Method (SVD). It consists in expanding the exchange kernel into a number of separable terms by means of the Singular Value Decomposition and then iterating over the remainder. In this paper, we extend our SVD method to the scattering of low energy electron-helium which has been the subject of interest, both theoretically and experimentally. We compare our results with the Moments Method which is widely used. The Moments Method consists of making an expansion of the solution into an especially favorable basis that takes care of the non-exchange part of the Hamiltonian.     

哈密顿矩阵的逆特征值问题

该文探讨了哈密顿矩阵的逆特征值问题, 得到了有解的充要条件、通解的表达式以及最小范数解.并给出了最佳逼近解的求法. 给出了相应的算法, 数值实例说明算法是可行的.     

Regularization of Discrete Ill-Posed Problems

The paper concerns conditioning aspects of finite-dimensional problems arising when the Tikhonov regularization is applied to discrete ill-posed problems. A relation between the regularization parameter and the sensitivity of the regularized solution is investigated. The main conclusion is that the condition number can be decreased only to the square root of that for the nonregularized problem. The convergence of solutions of regularized discrete problems to the exact generalized solution is analyzed just in the case when the regularization corresponds to the minimal condition number. The convergence theorem is proved under the assumption of the suitable relation between the discretization level and the data error. As an example the method of truncated singular value decomposition with regularization is considered. This revised version was published online in July 2006 with corrections to the Cover Date.     

应用LSI实现WEB图片的索引和查询

应用奇异值分解方法，分析相关文本词条和图片语义的关系，构造了一个图片“潜在语义索引”模型，用于缓解传统的许多WEB图片检索系统在索引和查询中遇到的同义词和多义词问题，实现语义索引和查询实验表明，该模型能有效地改善图片的索引和查询性能。     

Strong Semismoothness of the Fischer-Burmeister SDC and SOC Complementarity Functions

We show that the Fischer-Burmeister complementarity functions, associated to the semidefinite cone (SDC) and the second order cone (SOC), respectively, are strongly semismooth everywhere. Interestingly enough, the proof relys on a relationship between the singular value decomposition of a nonsymmetric matrix and the spectral decomposition of a symmetric matrix.The author’s research was partially supported by Grant R146-000-035-101 of National University of Singapore.The author’s research was partially supported by Grant R314-000-042/057-112 of National University of Singapore and a grant from the Singapore-MIT Alliance.Mathematics Subject Classification (2000): 90C33, 90C22, 65F15, 65F18     

Study of sawtooth oscillations on the HT-7 tokamak using 2D tomography of soft x-ray signal

It is the first time so far as we know that two arrays of multi-channel soft x-ray detectors are used to generate twodimensional （2D） images of sawtooth oscillation on the HT-7 tokamak using the Fourier-Bessel harmonic reconstruction method, and using the singular value decomposition to analyse the data from soft x-ray cameras. By these two arrays, 2D image reconstruction of soft x-ray emissivity can be obtained without assumption of plasma rigid rotation. Tomographic reconstruction of the m=1 mode structure is obtained during the precursor oscillation of the sawtooth crash. The crescent-shaped mode structure appearing on the contour map of the soft x-ray emissivity is consistent with the quasiinterchange mode. The characteristics of the m=1/n=1 mode structure observed in the soft x-ray tomography are as follows： the magnetic surface is made up of the crescent-shaped “hot core” and the circular “cold bubble”. The structure of the magnetic surface rotates in the direction of the electron diamagnetic drift and the rotation frequency is the oscillation frequency of soft x-ray signals.     

Analysis of Henrici's Transformation for Singular Problems

Henrici's transformation is the underlying scheme that generates, by cycling, Steffensen's method for the approximation of the solution of a nonlinear equation in several variables. The aim of this paper is to analyze the asymptotic behavior of the obtained sequence (s _n ^* ) by applying Henrici's transformation when the initial sequence (s _n ) behaves sublinearly. We extend the work done in the regular case by Sadok [17] to vector sequences in the singular case. Under suitable conditions, we show that the slowest convergence rate of (s _n ^* ) is to be expected in a certain subspace N of R ^p . More precisely, if we write s _n ^* =s _n ^* ,N+s _n ^* ,N, the orthogonal decomposition into N and N , then the convergence is linear for (s _n ^* ,N) but ( _n ^* ,N) converges to the same limit faster than the initial one. In certain cases, we can have N=R ^p and the convergence is linear everywhere.     

Explicit treatment of force contribution from alignment tensor using overdetermined linear equations and its application in NMR structure determination

Residual dipolar coupling (RDC) provides valuable information about the orientation of each internuclear vector in a macromolecule with respect to the static magnetic field. However, structure determination utilizing RDC still remains challenging without additional restraints such as NOE. In this context, a novel approach has been developed to efficiently extract structural information from RDC by successive application of singular value decomposition (SVD) method in the course of NMR structure determination. Force contribution from the alignment tensor is rigorously formulated in the context of SVD, and assessments have been made to verify its numerical accuracy. The efficacy of this approach is illustrated by showing that RDC restraints alone can restore a distorted beta-hairpin to native-like structure using the replica-exchange molecular dynamics simulations.     

Fast tensor method for summation of long‐range potentials on 3D lattices with defects

In this paper, we present a method for fast summation of long‐range potentials on 3D lattices with multiple defects and having non‐rectangular geometries, based on rank‐structured tensor representations. This is a significant generalization of our recent technique for the grid‐based summation of electrostatic potentials on the rectangular L × L × L lattices by using the canonical tensor decompositions and yielding the O(L) computational complexity instead of O(L³) by traditional approaches. The resulting lattice sum is calculated as a Tucker or canonical representation whose directional vectors are assembled by the 1D summation of the generating vectors for the shifted reference tensor, once precomputed on large N × N × N representation grid in a 3D bounding box. The tensor numerical treatment of defects is performed in an algebraic way by simple summation of tensors in the canonical or Tucker formats. To diminish the considerable increase in the tensor rank of the resulting potential sum, the ?‐rank reduction procedure is applied based on the generalized reduced higher‐order SVD scheme. For the reduced higher‐order SVD approximation to a sum of canonical/Tucker tensors, we prove the stable error bounds in the relative norm in terms of discarded singular values of the side matrices. The required storage scales linearly in the 1D grid‐size, O(N), while the numerical cost is estimated by O(NL). The approach applies to a general class of kernel functions including those for the Newton, Slater, Yukawa, Lennard‐Jones, and dipole‐dipole interactions. Numerical tests confirm the efficiency of the presented tensor summation method; we demonstrate that a sum of millions of Newton kernels on a 3D lattice with defects/impurities can be computed in seconds in Matlab implementation. The tensor approach is advantageous in further functional calculus with the lattice potential sums represented on a 3D grid, like integration or differentiation, using tensor arithmetics of 1D complexity. Copyright © 2015 John Wiley & Sons, Ltd.