非线性最小二乘问题收敛性的证明

主要讨论了无约束最优化中非线性最小二乘问题的收敛性.侧重于收敛的速率和整体、局部分析.改变了Gauss—Newton方法收敛性定理的条件,分两种情况证明了:(1)目标函数的海赛矩阵正定(函数严格凸)时为强整体二阶收敛;(2)目标函数不保证严格凸性,但海赛矩阵的逆存在时为局部收敛,敛速仍为二阶,同时给出了J(X)~(-1)和Q(X)~(-1)之间存在、有界性的等价条件.     

关于Newton法的Kantorovich型定理及其优函数

本文研究Newton法的Kantorovich型定理的特点及其对Newton法的半局部收敛性研究的思想方法,论述广义Lipschitz条件下的Kantorovich型定理的概括性和统一性.同时,在理论上当x_0取定时,针对每一个满足广义Lipschitz条件的光滑算子,给出优函数的一个构造方法.     

非线性互补问题的一种全局收敛的显式光滑Newton方法

本针对Po函数非线性互补问题，给出了一种显式光滑Newton方法，该方法将光滑参数μ进行显式迭代而不依赖于Newton方向的搜索过程，并在适当的假设条件下，证明了算法的全局收敛性。     

一阶HÖlder条件下带参数的修正型Euler-Halley迭代族的局部收敛性

1引言设X和Y为实或复Banach空间,Ω■X是开凸子集,F:Ω■X→Y是一阶连续可微的非线性算子.非线性算子方程F(x)=0 (1.1) 的求解及收敛域问题是现代科学计算理论的基本问题.解方程(1.1)的最著名的迭代方法是Newton法,在适当的条件下,它是二阶收敛的,此即著名的Kantorovich定理.关于Newton法收敛球半径的估计由Traub和王兴华分别给出,见[2]和[3],而收敛性研究的进一步发展可参看[4,5,6]及综述文章[7].     

m步非线性多重分裂Newton法的收敛性

本文讨论了多重分裂算法在求解一类非线性方程组的全局收敛性和单侧收敛性．当用研步Newton法来代替求得每个非线性多重分裂子问题的近似解时，同样给出相应收敛性结论．数值算例证实了算法的有效性．     

求解半光滑方程组的近似Newton法

本文提出了求解半光滑方程组的近似Newton法，并证明了该算法的局部超线性收敛性。数值结果表明 该算法是有效的。     

求解非线性不适定的混合Newton-Tikhonov迭代法

鉴于Newton型方法在实际计算中计算量可能非常大,因此提出了一种一步Newton结合若干步简化Newton的混合Newton-Tikhonov方法,并且在一定条件下证明了该方法的收敛性和稳定性.数值试验表明,在减少计算量方面该方法相对于经典的Newton方法有明显的改善.     

二阶锥规划两个新的预估-校正算法

基于不可行内点法和预估-校正算法的思想,提出两个新的求解二阶锥规划的内点预估-校正算法．其预估方向分别是Newton方向和Euler方向,校正方向属于Alizadeh-Haeberly-Overton(AHO)方向的范畴．算法对于迭代点可行或不可行的情形都适用．主要构造了一个更简单的中心路径的邻域,这是有别于其它内点预估-校正算法的关键．在一些假设条件下,算法具有全局收敛性、线性和二次收敛速度,并获得了O(rln(ε0/ε))的迭代复杂性界,其中r表示二阶锥规划问题所包含的二阶锥约束的个数．数值实验结果表明提出的两个算法是有效的．     

线性无关假设如何影响非线性约束最优化算法

首先综述非线性约束最优化最近的一些进展. 首次定义了约束最优化算法的全局收敛性. 注意到最优性条件的精确性和算法近似性之间的差异, 并回顾等式约束最优化的原始的Newton 型算法框架, 即可理解为什么约束梯度的线性无关假设应该而且可以被弱化. 这些讨论被扩展到不等式约束最优化问题. 然后在没有线性无关假设条件下, 证明了一个使用精确罚函数和二阶校正技术的算法可具有超线性收敛性. 这些认知有助于接下来开发求解包括非线性半定规划和锥规划等约束最优化问题的更加有效的新算法.     

MIMD机上解非线性方程组的同步化并行拟Newton法

本文给出了适于在MIMD机上解非线性方程组的同步化并行Broyden方法和换列修正拟Newton法的迭代格式,以及它们的局部收敛性定理.数值试验结果也验证了收敛性.     

An enhanced finite volume method to model 2D linear elastic structures

This paper details the evaluation and enhancement of the vertex-centred finite volume method for the purpose of modelling linear elastic structures undergoing bending. A matrix-free edge-based finite volume procedure is discussed and compared with the traditional isoparametric finite element method via application to a number of test-cases. It is demonstrated that the standard finite volume approach exhibits similar disadvantages to the linear Q4 finite element formulation when modelling bending. An enhanced finite volume approach is proposed to circumvent this and a rigorous error analysis conducted. It is demonstrated that the developed finite volume method is superior to both standard finite volume and Q4 finite element methods, and provides a practical alternative to the analysis of bending-dominated solid mechanics problems.     

A wavelet-based stochastic finite element method of thin plate bending

A wavelet-based stochastic finite element method is presented for the bending analysis of thin plates. The wavelet scaling functions of spline wavelets are selected to construct the displacement interpolation functions of a rectangular thin plate element and the displacement shape functions are expressed by the spline wavelets. A new wavelet-based finite element formulation of thin plate bending is developed by using the virtual work principle. A wavelet-based stochastic finite element method that combines the proposed wavelet-based finite element method with Monte Carlo method is further formulated. With the aid of the wavelet-based stochastic finite element method, the present paper can deal with the problem of thin plate response variability resulting from the spatial variability of the material properties when it is subjected to static loads of uncertain nature. Numerical examples of thin plate bending have demonstrated that the proposed wavelet-based stochastic finite element method can achieve a high numerical accuracy and converges fast.     

The influence of integration rule accuracy on the calculation of surface subsidence by the nucleus of strain method in conjunction with a finite element reservoir simulator

Relative accuracy of numerical quadrature rules when applied to the simulation of underground petroleum reservoirs by means of the finite element method is investigated. Fluid flow within the reservoir is calculated via the finite element method and the resulting deformation by the nucleus of strain technique. By analysing a simple problem it was found that the solution method was susceptible to changes in numerical quadrature for reservoirs that were positioned near the ground surface and that care is required when solving such problems due to the singularities occurring in the integrands which appear in the nucleus of strain approach.     

A Numerical Study of Blowup in the Harmonic Map Heat Flow Using the MMPDE Moving Mesh Method

The numerical solution of the harmonic heat map flow problems with blowup in finite or infinite time is considered using an adaptive moving mesh method. A properly chosen monitor function is derived so that the moving mesh method can be used to simulate blowup and produce accurate blowup profiles which agree with formal asymptotic analysis. Moreover, the moving mesh method has finite time blowup when the underlying continuous problem does. In situations where the continuous problem has infinite time blowup, the moving mesh method exhibits finite time blowup with a blowup time tending to infinity as the number of mesh points increases. The inadequacy of a uniform mesh solution is clearly demonstrated.     

M-P逆和加权M-P逆的有限迭代计算公式

本文，对于任意给定的矩阵A，我们给出了计算其M—P逆和加权M—P逆的有限迭代计算公式．根据这一迭代公式，当我们选取初始矩阵为X0=A^#，则矩阵A的加权M—P逆A^＋MN在不考虑舍入误差的情况下，可以在有限迭代的情况得到，同样当我们选取初始矩阵X0=A^＊，其M—P逆A^＋亦可以在有限迭代下获得．最后我们用数值例子检验了我们算法的正确性。     

FLIM - A Variant of FEM based on Line Integration

A variant of the finite element method (FEM) for modelling and solving partial differential equations based on triangular and tetrahedral meshes is proposed. While FEM is based on integration over finite elements, the new approach - briefly denoted as FLIM hereafter - uses integration along edges (finite lines). The stiffness matrix, which - for linear triangles and tetrahedra - is identical with the one obtained with FEM, as well as the load vector can solely be obtained by summing up the edge contributions. This new variant requires much lower storage than FEM, especially for three-dimensional problems, but yields the same approximation error and convergence rate as the finite element method. It is shown that its performance, when applied to linear problems, is in close agreement with the performance of the finite element method. (© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Error estimates for the finite point method

In this paper, error estimates for the finite point method are presented in Sobolev spaces in multiple dimensions when nodes and shape functions satisfy certain conditions. From the error analysis of the finite point method, the error bound of the numerical solution is directly related to the radii of the weight functions and the condition number of the coefficient matrix.     

A BALANCED OVERSAMPLING FINITE ELEMENT METHOD FOR ELLIPTIC PROBLEMS WITH OBSERVATIONAL BOUNDARY DATA

In this paper we propose a finite element method for solving elliptic equations with observational Dirichlet boundary data which may subject to random noises. The method is based on the weak formulation of Lagrangian multiplier and requires balanced oversampling of the measurements of the boundary data to control the random noises. We show the convergence of the random finite element error in expectation and, when the noise is sub-Gaussian, in the Orlicz $\psi_2$-norm which implies the probability that the finite element error estimates are violated decays exponentially. Numerical examples are included.     

一类基于小波基函数插值的有限元方法

在分析具有大的梯度问题中,将具有紧支集的小波基函数引入到传统的有限元插值函数的构造中,对传统的插值方法进行修正。对新的插值模式进行了数值稳定性（解的唯一存在性）分析并通过分片分析讨论了解的收敛性,新的插值模式所引入的附加自由度通过静力凝聚法来消除,最后得到了基于变分原理的小波有限元列式。     

On the Number of Conserved Quantities for the Two-Layer Shallow-Water Equations

The shallow-water equations for two-layer inviscid flow with a free surface overlying a rigid horizontal bottom subject to gravitational forcing only are examined to determine the possible forms of conservation laws that the equations permit. In the case of a single layer with flow in only one horizontal direction, it is known that there are an infinite number of associated equations in conservation form, where the conserved quantity is a multinomial in the layer variables. The method used to determine this result is generalized to show that in the two-layer case, the result does not generalize, and it is discovered that only a finite number of conservation equations exist when the density difference between the layers is nonzero. The subsequent conservation equations are given explicitly, and a systematic method for deriving conservation laws from an arbitrary first-order system is described. For the case when the flow is in both horizontal dimensions, the method of analysis is straightforward in the one-layer case, and the finite number of conservation equations are derived. The two-layer case is similar, and the finite number of generalized conserved quantities are stated, although the question of whether or not there are only a finite number is posed as an open question.