一种求解非线性方程组的3p阶迭代方法

本文将一种改进的二步迭代算法作为预测,将高斯-勒让德求积公式作为校正,提出了一种求解非线性方程组的具有3p收敛阶的迭代方法.最后给出了一些数值实例,将本文的实验结果与现有的几种迭代方法的实验结果作了比较分析,验证了本文所提出的结果.     

一种基于Chebyshev迭代解非线性方程组的方法

本文根据牛顿迭代和Chebyshev迭代法给出了一种新的迭代方法,该方法有较高的收敛阶,并在理论上给予了证明.最后给出了四个实例,将本文的实验结果与现有的几种方法的实验结果进行比较,表明我们的方法迭代次数少,有明显的优势.     

用于求根问题的一个基于Thiele连分式的四阶收敛的迭代方法

利用截断的Thiele连分式,本文给出了一个求解非线性单变量方程的单步迭代方法,并证明了所提出的迭代方法具有四阶收敛性.最后,本文通过一些数值例子说明了所提出的方法的有效性和表现.     

求解时谐涡流模型鞍点问题的分块交替分裂隐式迭代算法的改进

分块交替分裂隐式迭代方法是求解具有鞍点结构的复线性代数方程组的一类高效迭代法.本文通过预处理技巧得到原方法的一种加速改进方法,称之为预处理分块交替分裂隐式迭代方法·理论分析给出了新方法的收敛性结果.对于一类时谐涡旋电流模型问题,我们给出了若干满足收敛条件的迭代格式.数值实验验证了新型算法是对原方法的有效改进.     

两类五阶解非线性方程组的迭代算法

本文首先根据Runge-Kutta方法的思想,结合Newton迭代法,提出了一类带参数的解非线性方程组F(x)=0的迭代算法,然后基于解非线性方程f(x)=0的King算法,给出第二类解非线性方程组的迭代算法,收敛性分析表明这两类算法都是五阶收敛的.其次给出了本文两类算法的效率指数,以及一些已知算法的效率指数,并且将本文算法的效率指数与其它方法进行详细的比较,通过效率比率R_(i,j)可知本文算法具有较高的计算效率.最后给出了四个数值实例,将本文两类算法与现有的几种算法进行比较,实验结果说明本文算法收敛速度快,迭代次数少,有明显的优势.     

求解非对称代数Riccati 方程几个新的预估-校正法

来源于输运理论的非对称代数Riccati 方程可等价地转化成向量方程组来求解. 本文提出了求解该向量方程组的几个预估-校正迭代格式,证明了这些迭代格式所产生的序列是严格单调递增且有上界,并收敛于向量方程 组的最小正解. 最后,给出了一些数值实验,实验结果表明,本文所提出的算法是有效的.     

外推的MHSS迭代法求解一类大型稀疏的复对称线性系统

本文对改良的Hermitian和反Hermitian分裂迭代方法 (MHSS)使用了外推技术,构造了外推的MHSS(EMHSS)迭代法.从理论上给出了EMHSS迭代方法的迭代矩阵与MHSS迭代方法的迭代矩阵之间的关系,并讨论了EMHSS迭代方法的收敛条件.最后用数值实验验证了所提方法的有效性.     

无约束最优化的微分下降法

本文利用曲线线性搜索法和最优化的微分梯度法的特点,提出了一种一般的曲线搜索方式:微分下降法。这种方法通过下降方向对确定迭代矩阵,由初值微分方程的解析解确定迭代搜索曲线。本文给出了算法的整体收敛性证明,并给出了满意的数值实验结果。     

一种不精确分裂迭代方法中外迭代步数的估计方法（英文）

单步分裂迭代方法用于求解大型稀疏线性方程组时,迭代解的精度对迭代过程的收敛和方程组解的精度有很大影响.基于文献(参见[Bai Z Z,Rozlozník M.On the numerical behavior of matrix splitting iteration methods for solving linear systems.SIAM J Numer Anal,2015,53(4):1716-1737.])的结果,对给定的精度,给出了一个估计最大外迭代步数的方法.数值实验结果表明,本文所给出的最大外迭代步数的估计与实际计算过程中达到相同精度所需的迭代步数非常接近.     

在线Logistic回归

针对连续数据流分类问题,基于在线学习理论,提出一种在线logistic回归算法.研究带有正则项的在线logistic回归,提出了在线logistic-l2回归模型,并给出了理论界估计.最终实验结果表明,随着在线迭代次数的增加,提出的模型与算法能够达到离线预测的分类结果.本文工作为处理海量流数据分类问题提供了一种新的有效方法.     

Convergence of Conjugate Gradient Methods with a Closed-Form Stepsize Formula

Conjugate gradient methods are efficient methods for minimizing differentiable objective functions in large dimension spaces. However, converging line search strategies are usually not easy to choose, nor to implement. Sun and colleagues (Ann. Oper. Res. 103:161–173, 2001; J. Comput. Appl. Math. 146:37–45, 2002) introduced a simple stepsize formula. However, the associated convergence domain happens to be overrestrictive, since it precludes the optimal stepsize in the convex quadratic case. Here, we identify this stepsize formula with one iteration of the Weiszfeld algorithm in the scalar case. More generally, we propose to make use of a finite number of iterates of such an algorithm to compute the stepsize. In this framework, we establish a new convergence domain, that incorporates the optimal stepsize in the convex quadratic case. The authors thank the associate editor and the reviewer for helpful comments and suggestions. C. Labat is now in postdoctoral position, Johns Hopkins University, Baltimore, MD, United States.     

Symbolic operational images and decomposition formulas for hypergeometric functions

Based upon the classical derivative and integral operators we introduce a new operator which allows the derivation of new symbolic operational images for hypergeometric functions. By means of these symbolic operational images a number of decomposition formulas involving quadruple series are then found. Other closely-related results are also considered.     

Some new problems in numerical integration

In this paper, we examine the numerical computation of the (multiple) integrals generated by Galerkin methods applied to two nonstandard hypersingular integral equations, which are of interest by their own. These equations are used to solve two classical electromagnetic problems that are briefly described.     

On families of quadrature formulas based on Euler identities

A family consisting of quadrature formulas which are exact for all polynomials of order ?5 is studied. Changing the coefficients, a second family of quadrature formulas, with the degree of exactness higher than that of the formulas from the first family, is produced. These formulas contain values of the first derivative at the end points of the interval and are sometimes called “corrected”.     

Hyperfunctions, formal groups and generalized Lipschitz summation formulas

A construction relating the theory of hyperfunctions with the theory of formal groups and generalizations of the classical Lipschitz summation formula is proposed. It involves new polylogarithmic rational functions constructed via the Fourier expansion of certain sequences of Bernoulli-type polynomials, related to the Lazard formal group. Related families of one-dimensional hyperfunctions are also constructed.     

Numerical integration formulas of degree two

Numerical integration formulas in n-dimensional nonsymmetric Euclidean space of degree two, consisting of n+1 equally weighted points, are discussed, for a class of integrals often encountered in statistics. This is an extension of Stroud's theory [A.H. Stroud, Remarks on the disposition of points in numerical integration formulas, Math. Comput. 11 (60) (1957) 257–261; A.H. Stroud, Numerical integration formulas of degree two, Math. Comput. 14 (69) (1960) 21–26]. Explicit formulas are given for integrals with nonsymmetric weights. These appear to be new results and include the Stroud's degree two formula as a special case.     

一类新共轭梯度法的全局收敛性

共轭梯度法是求解无约束最优化问题的有效方法.本文在βkDY的基础上对βk引入参数,提出了一类新共轭梯度法,并证明其在强Wolfe线性搜索条件下具有充分下降性和全局收敛性.     

Fibonacci求积公式对具有有界混合差分的光滑函数类的误差估计

考虑对具有有界混合差分的二元光滑函数类Ｂ＾γ，ｐ，θ的求积公式，本文证明了Ｆｉｂｏｎａｃｃｉ求积公式是渐近最优的，并求出了春误差的渐近最优价。