求解线性互补问题的一种新的势下降内点算法

本文针对具有半正定矩阵的线性互补问题提出了一个新的内点方法———势函数下降内点方法.采用部分校正技术和Sherman-Morrison-Woodbury准则获得问题的近似最优解.讨论了该算法的收敛性,并证明了该算法为多项式算法.     

一类随机线性互补问题的求法

考虑一类随机互线性补问题的求解方法,目的是通过定义NCP函数来使正则化期望残差最小化.通过拟蒙洛包洛方法产生一系列观察值并且证得离散近似问题最小值解的聚点就是相应随机线性互补问题的期望残差最小值ERM,同时得到利用ERM到解为有界的充分条件.进一步证明ERM法能够得到具有稳定性和最小灵敏度的稳健解.     

反射的椭圆随机偏微分方程的网格逼近

研究了由可乘噪声驱动的反射的椭圆随机偏微分方程网格近似解的收敛性,其中考虑区域D:=(0,1)~d,d=1,2,3.此外,还研究确定的椭圆障碍问题离散格式的解存在唯一性,并得到解关于障碍函数的连续依赖性和收敛性.     

随机非线性互补问题的条件风险价值模型及其求解方法

随机非线性互补问题(SNCP)在交通运输,工程力学,金融等许多方面都有着非常广泛的应用,由于随机因素的存在,SNCP通常无解.为解决这个问题,考虑构造一个合理的确定性模型,并将这个确定性模型的解作为SNCP的解.文章利用限定非线性互补函数(NCP函数)来构造投资组合优化中的损失函数,提出求解随机非线性互补问题(SNCP)的条件风险价值(CVaR)模型.由于该模型中含有数学期望及非光滑函数,为求解此模型,文章应用样本均值近似方法和光滑化方法,给出此模型的近似问题并进一步给出求解算法.在理论上,文章还考虑了条件风险价值模型水平集的有界性及该模型近似问题全局最优解序列以及稳定点的收敛性结果.以上结果从理论上保证了文章所提求解SNCP的新模型及其近似问题的可行性.此外,数值结果表明上述方法是有效的.     

对流-扩散问题的特征──块中心差分法

１．引言１９８２年,Ｄｏｕｇｌａｓ和Ｒｕｓｓｅｌｌ［１］提出解对流一扩散问题的特征一差分方法,网格节点为均匀分布,求解区域为直线Ｒ．文中讨论了基于二次插值的特征一差分格式,但其近似解按离散Ｌ２模未达到最优阶误差估计．１９８８年Ｗｅｉｓｅｒ和Ｗｈｅｅｌｅｒ［２］提出解线性椭圆型和线性抛物型方程的块中心差分法,１９９１年王申林［３］讨论了解拟线性双曲型积分微分方程的块中心差分方法,其共同特点为近似解按离散的Ｌ２模达到最优阶误差估计,解的一阶导数的近似解达到超收敛误差估计．１９９３年由同顺［４］讨论了…     

线性互补问题均衡解的存在形式与识别方法

研究了线性互补问题均衡解的存在形式与判定方法,给出了线性互补问题有解的充要条件,得到了带有几类特殊系数矩阵的线性互补问题的解的性质.在此基础上设计了求解线性互补问题均衡解的直接算法.     

线性常微分方程的全局误差估计和优化求解方法

线性常微分方程初值问题求解在许多应用中起着重要作用.目前,已存在很多的数值方法和求解器用于计算离散网格点上的近似解,但很少有对全局误差(global error)进行估计和优化的方法.本文首先通过将离散数值解插值成为可微函数用来定义方程的残差;再给出残差与近似解的关系定理并推导出全局误差的上界;然后以最小化残差的二范数为目标将方程求解问题转化为优化求解问题;最后通过分析导出矩阵的结构,提出利用共轭梯度法对其进行求解.之后将该方法应用于滤波电路和汽车悬架系统等实际问题.实验分析表明,本文估计方法对线性常微分方程的初值问题的全局误差具有比较好的估计效果,优化求解方法能够在不增加网格点的情形下求解出线性常微分方程在插值解空间中的全局最优解.     

离散变量结构优化设计的组合算法*

本文首先给出了离散变量优化设计局部最优解的定义,然后提出了一种综合的组合算法.该算法采用分级优化的方法,第一级优化首先采用计算效率很高且经过随机抽样性能实验表明性能较高的启发式算法─—相对差商法,求解离散变量结构优化设计问题近似最优解 X ;第二级采用组合算法,在 X 的离散邻集内建立离散变量结构优化设计问题的(-1,0.1)规划模型,再进一步将其化为(0,1)规划模型,应用定界组合算法或相对差商法求解该(0,1)规划模型,求得局部最优解.解决了采用启发式算法无法判断近似最优解是否为局部最优解这一长期未得到解决的问题,提高了计算精度,同时,由于相对差商法的高效率与高精度,以上综合的组合算法的计算效率也还是较高的.     

线性互补问题与多目标优化

本文研究了求解线性互补问题的一类新方法:把线性互补问题转化为多目标优化问题,利用多目标优化有效解的定义,给出了零有效解的概念;进而获得多目标优化问题的零有效解就是线性互补问题的最优解.最后给出了有解、无解线性互补问题,并分别把这些问题转化为多目标优化,采用极大极小方法求解转化后的多目标优化问题.数值实验结果表明了该方法的正确性和有效性,完善了文献[19]的数值结果.     

区间自适应遗传算法优化无约束非线性规划问题

针对无约束非线性规划传统优化方法存在的问题,将区间自适应遗传算法引入无约束非线性规划优化中,算法可以利用当前进化信息,自适应移动搜索区间,找到全局最优解,故可缩短搜索区间长度,提高编码精度,降低算法计算量,解决了传统遗传算法处理优化问题时,给定区间必须包含最优解这一问题,这也是本算法有别于其他优化算法的独特优势,为某些最优解所在区间难以估计的无约束非线性规划问题的优化提供了一条有效可行的途径.系统阐述了区间自适应遗传算法的原理,给出了算法优化无约束非线性规划问题的步骤,以MatlabR2016b仿真方式对算法进行了实例测试,结果表明,方法是一种计算稳定、正确、有效、可靠实用的无约束非线性规划优化方法.     

The behavior of a finite difference approximation to a singular initial boundary value problem

We develop difference approximations to a singular parabolic initial-boundary value problem and its corresponding steady-state problem. A critical value for the existence of nonnegative solutions to the discrete steady state system is established. Convergence of the computed critical values is obtained. The long time behavior for the approximated solution of the parabolic problem is investigated. It is shown that the behavior of the discrete system is consistent with that of the continuous one     

Homoclinic solutions of a class of periodic difference equations with asymptotically linear nonlinearities

By using critical point theory and periodic approximations, new sufficient conditions are obtained on the existence and nonexistence of homoclinic solutions for a class of discrete nonlinear periodic equations with asymptotically linear nonlinearities. These results partially answer an open problem proposed by Pankov (2006) [2] under rather weaker conditions and greatly improve the related results before.     

On the Reconstruction of a String from Spectral Data

We consider the problem to reconstruct the mass distribution of a string where the mass is concentrated in a finite number of points, or, equivalently, the problem to reconstruct a simply connected mass spring system with unknown masses and stiffness parameters if the following data are given. Problem 1: The spectra of the string and of a modification of the string, or. Problem 2: The spectra of two different modifications of the string. Here a modification of the string is a string which appears if we link the unknown string with another string of known mass distribution. The paper contains a necessary condition for the existence of a solution of Problem 1, and explicit formulas and an algorithm for the solutions of the Problems 1 and 2 under the condition that there exists a solution. For the case that the mass distribution of the unknown string is not discrete we consider the problem to find discrete approximations of this distribution from the respective spectral data. The methods are based on the spectral theory of generalized second order differential operators as developed by M. G. Krein     

反对称偏对称矩阵反问题的最小二乘解

该文研究了反对称偏对称矩阵反问题的最小二乘解,得到了该问题解的表达式以及该问题有解的充分必要条件.证明了其最佳逼近解的存在性和唯一性,建立了其最佳逼近解的表达式,并给出了求最佳逼近解的数值算法和算例.     

Boundary value problems of discrete generalized Emden-Fowler equation

By using the critical point theory, some sufficient conditions for the existence of the solutions to the boundary value problems of a discrete generalized Emden-Fowler equation are obtained. In a special case, a sharp condition is obtained for the existence of the boundary value problems of the above equation. For a linear case, by the discrete variational theory, a necessary and sufficient condition for the existence, uniqueness and multiplicity of the solutions is also established.     

On the existence of homoclinic solutions of a class of discrete nonlinear periodic systems

In this paper, we obtain a new sufficient condition on the existence of homoclinic solutions of a class of discrete nonlinear periodic systems by using critical point theory in combination with periodic approximations. We prove that it is also necessary in some special cases.     

Full discretization of a nonlinear parabolic problem containing volterra operators and an unknown dirichlet boundary condition

The reconstruction of an unknown solely time‐dependent Dirichlet boundary condition in a nonlinear parabolic problem containing a linear and a nonlinear Volterra operator is considered. The inverse problem is converted into a variational problem in which the unknown Dirichlet condition is eliminated using a given integral overdetermination. A time‐discrete recurrent approximation scheme is designed, using Backward Euler's method. The convergence of the approximations towards a solution of the variational problem is proved under appropriate assumptions on the data and on the Volterra operators. The uniqueness of this solution is shown in the case that the nonlinear Volterra operator satisfies a particular inequality. Moreover, the Finite Element Method is used to discretize the time‐discrete approximation scheme in space. Finally, full‐discrete error estimates are derived for a particular choice of the finite elements. The corresponding convergence rates are supported by a numerical experiment. © 2015 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq 31: 1444–1460, 2015     

Approximation and optimization of discrete and differential inclusions described by inequality constraints

In the first part of this article optimization of polyhedral discrete and differential inclusions is considered, the problem is reduced to convex minimization problem and the necessary and sufficient condition for optimality is derived. The optimality conditions for polyhedral differential inclusions based on discrete-approximation problem according to continuous problems are formulated. In particular, boundedness of the set of adjoint discrete solutions and upper semi-continuity of the locally adjoint mapping are proved. In the second part of this article an optimization problem described by convex inequality constraint is studied. By using the equivalence theorem concerning the subdifferential calculus and approximating method necessary and sufficient condition for discrete-approximation problem with inequality constraint is established.     

一般Lyness方程的周期性与严格振动性

研究了一般的Lyness方程 其中a、b∈[0,∞)且a+b>0,初值x_-1、x₀为任意正数。得到了一些新的结果;方程(*)解的周期性的一个必要充分条件;方程(*)的所有解严格振动的充分条件。作为应用,解决了G.Ladas提出的一个公开问题。