利用均值算法求解凸函数极小值的收敛性分析

郑权等在[1]-[3]中提出了一种求解无约束优化问题的均值算法，若假设目标函数f(x)是连续的，还讨论了均值算法的收敛性。若假设f(x) 有界闭集Ω上的凸函数，本文证明了求解凸函数极小值的均值算法是线性收敛的。     

一类约束不可微优化问题的区间极大熵方法

本文研究求解不等式约束离散ｍｉｎｉｍａｘ问题的区间算法,其中目标函数和约束函数是 Ｃ~１类函数．利用罚函数法和极大熵函数思想将问题转化为无约束可微优化问题,讨论了极大熵函数的区间扩张,证明了收敛性等性质,提出了无解区域删除原则,建立了区间极大熵算法,并给出了数值算例．该算法是收敛、可靠和有效的．     

广义伪凸函数与非光滑优化不动点算法的收敛性

本文给出了一类较广泛的函数类——广义伪凸函数的概念,讨论了这类函数的性质,它与其它凸性函数的关系.然后证明了将 Merrill 不动点算法和Eaves-Saigal 单纯同伦算法用于目标函数和约束函数均为广义伪凸函数的约束优化问题的大范围收敛性.     

不等式约束优化一个基于滤子思想的广义梯度投影算法

讨论非线性不等式约束优化问题, 借鉴于滤子算法思想,提出了一个新型广义梯度投影算法.该方法既不使用罚函数又无真正意义下的滤子.每次迭代通过一个简单的显式广义投影法产生搜索方向,步长由目标函数值或者约束违反度函数值充分下降的Armijo型线搜索产生.算法的主要特点是: 不需要迭代序列的有界性假设;不需要传统滤子算法所必需的可行恢复阶段;使用了ε积极约束集减小计算量.在合适的假设条件下算法具有全局收敛性, 最后对算法进行了初步的数值实验.     

三项记忆梯度法及其投影算法的收敛性分析

对于无约束优化问题,提出了一类新的三项记忆梯度算法．这类算法是在参数满足某些假设的条件下,确定它的取值范围,从而保证三项记忆梯度方向是使目标函数充分下降的方向．在非单调步长搜索下讨论了算法的全局收敛性．为了得到具有更好收敛性质的算法,结合Solodov and Svaiter(2000)中的部分技巧,提出了一种新的记忆梯度投影算法,并证明了该算法在函数伪凸的情况下具有整体收敛性．     

非线性约束条件下一个超线性收敛的可行方法

在本文中，我们对非线性不等式约束条件下的非线性优化问题给出了一个新的ＳＱＰ类可行方法．此算法不但结构简单、易于计算，并且在适当的假设条件下，我们证明了算法具有全局收敛性及超线性收敛性     

无约束非光滑优化问题信赖域算法的收敛条件

1 引言 考虑下列无约束非光滑优化问题 minf(x),(1) x∈R~n,其中f为R~n上的局部Lipschitz函数,本文将‖·‖_2简记为‖·‖.记下列信赖域子问题为S∪B(x,△). min m(x,s)=φ(x,s)+1/2s~TBs, 其中φ:R~(2m)→R为f的迭代函数。 对于无约束非光滑优化问题(1),[11],[13],[3]、[4]和[5]分别在特殊的条件下给出了信赖域算法用以求解(1)的收敛性结果。最近,[10]、[2]和[6]在不同的假设条件下分别给出了信赖域算法求解无约束非光滑优化问题的一般模型,并在子问题的目标函数满足局部一致有界性条件时证明了算法模型的整体收敛性。在目标函数满足某种正则性条件时,[11]和[9]给出了当信赖域子问题的目标函数中二次项不满足一致有界性条件时的收敛性结果.本文则在目标函数仅为局部Lipschitz函数时得到了和[8]、[11]、[9]相同的收敛性结果。     

初始旋度ω0∈L（logL）^α（α〉1／2）的2—D Euler方程弱解的存在性

本文证明了当初始旋度ω０∈Ｌ（ｌｏｇＬ）＾α∩Ｌ＾α〉１／２时二维无粘Ｅｕｌｅｒ方程柯西问题的整体解的存在性，其中Ｌ（ｌｏｇＬ）＾α是Ｚｙｇｍｕｎｄ函数类，在有界区域，它包含任意的Ｌ＾ｐ∩Ｌ＾１空间。     

具有线性化技术的三块非凸不可分优化问题BregmanADMM收敛性分析

交替方向乘子法是求解两块可分离凸优化问题的有效方法,但是对于三块不可分的非凸优化问题的交替方向乘子法的收敛性可能无法保证.该文主要研究的是用线性化广义Bregman交替方向乘子法(L-G-BADMM)求解目标函数是三块不可分的非凸极小化问题的收敛性分析.在适当假设条件下,对算法中子问题进行求解并构建满足Kurdyka-Lojasiewicz性质的效益函数,经过理论证明可以得到该算法的收敛性.     

解约束优化的分段线性有理NCP函数

本文给出新的NCP函数,这些函数是分段线性有理正则伪光滑的,且具有良好的性质.把这些NCP函数应用到解非线性优化问题的方法中.例如,把求解非线性约束优化问题的KKT点问题分别用QP-free方法,乘子法转化为解半光滑方程组或无约束优化问题.然后再考虑用非精确牛顿法或者拟牛顿法来解决该半光滑方程组或无约束优化问题.这个方法是可实现的,且具有全局收敛性.可以证明在一定假设条件下,该算法具有局部超线性收敛性.     

基于模糊FNN-ELECTRE法的国有商业银行竞争力评价

随着金融业全方位开放,国有商业银行已成为国家经济命脉的核心,其竞争力强弱关系着国家经济的繁荣与衰退。本文基于模糊FNN-ELECTRE方法建立国有商业银行竞争力评价模型,以因素神经网络理论(FNN)与ELECTRE融合方法为基础,以现有商业银行竞争力评价指标为着眼,把国有商业银行竞争力的评价体系从现实竞争力和潜在竞争力两个因素抽取为一级指标,以规模、质量、业务结构、效率、成长性五个因素设为二级指标。从实证的角度对银行A、银行B、银行C、银行D、银行E五大国有银行进行分析验证,对其属性值进行和谐性与非和谐性检验得出,银行B竞争能力最强,银行E竞争能力最弱,五大国有银行均有改进空间。     

Formulation of linear problems and solution by a universal machine

Using the predicate language for ordered fields a class of problems referred to aslinear problems is defined. This class contains, for example, all systems of linear equations and inequalities, all linear programming problems, all integer programming problems with bounded variables, all linear complementarity problems, the testing of whether sets that are defined by linear inequalities are semilattices, all satisfiability problems in sentenial logic, the rank-computation of matrices, the computation of row-reduced echelon forms of matrices, and all quadratic programming problems with bounded variables. A single, one, algorithm, to which we refer as theUniversal Linear Machine, is described. It solves any instance of any linear problem. The Universal Linear Machine runs in two phases. Given a linear problem, in the first phase a Compiler running on a Turing Machine generates alinear algorithm for the problem. Then, given an instance of the linear problem, in the second phase the linear algorithm solves the particular instance of the linear problem. The linear algorithm is finite, deterministic, loopless and executes only the five ordered field operations — additions, multiplications, subtractions, divisions and comparisons. Conversely, we show that for each linear algorithm there is a linear problem which the linear algorithm solves uniquely. Finally, it is shown that with a linear algorithm for a linear problem, one can solve certain parametric instances of the linear problem.Research was supported in part by the National Science Foundation Grant DMS 92-07409, by the Department of Energy Grant DE-FG03-87-ER-25028, by the United States—Israel Binational Science Foundation Grant 90-00434 and by ONR Grant N00014-92-J1142.Corresponding author.     

A double-stage piecewise recursive filter for real-time applications

This paper examines the basic problem of estimating the state of a system described by a set of dynamical equations. The state estimation is performed by means of optimal filtering techniques in which the state observation is given by a set of nonlinear equations.The basic issue encountered in most engineering problems is the availability of more observable information than the processor can effectively process in real-time, while performing full optimal filtering on all available observable data. To alleviate this deficiency, an algorithm is developed in which the rate of information processing is kept at a high value, while the system statistics are evaluated at a much slower rate.The method utilizes a combination of fast and slow filtering loops, in which the observable data are processed at a high rate during the fast loop, while the system error covariance, gain computations, and all other system statistics are processed at a lower rate in the slow loop. Methods are provided to resolve any incompatibility in the system statistics resulting from this fast-loop/slow-loop processing combination.Applications describing problems pertaining to aircraft navigation are presented. Specifically, applications to aircraft navigation through a satellite network are studied. Appropriate simulation results from the above studies are shown.This work was performed under contract with the Department of the Air Force, Space and Missile Organization (SAMSO), Los Angeles, California, Contract No. F04-701-75-C-0180.The authors are indebted to Dr. C. Johnson, GPS System Engineer, Mr. G. Consolver, GPS Software Manager, and Mr. W. Riley, Communications/Navigation Department, Texas Instruments for their continuous encouragement in the course of this work. Also, they are indebted to Major M. Birnbaum, SAMSO, for his constructive criticism on the conceptual design of this work, and to Mr. A. Bierman, Aerospace Corporation, for kindly providing simulation data of aircraft trajectories. Finally, the authors would like to express their gratitude to Dr. N. Carlson, Intermetrics, for valuable consultation during the progress of this work.Portions of this paper were presented by the senior author at the GSP Conference, Plan 76, IEEE Position, Location, and Navigation Symposium, San Diego, California, 1976.     

Moment information for probability distributions,without solving the moment problem,II: Main-mass,tails and shape approximation

How much information does a small number of moments carry about the unknown distribution function? Is it possible to explicitly obtain from these moments some useful information, e.g., about the support, the modality, the general shape, or the tails of a distribution, without going into a detailed numerical solution of the moment problem? In this, previous and subsequent papers, clear and easy to implement answers will be given to some questions of this type. First, the question of how to distinguish between the main-mass interval and the tail regions, in the case we know only a number of moments of the target distribution function, will be addressed. The answer to this question is based on a version of the Chebyshev–Stieltjes–Markov inequality, which provides us with upper and lower, moment-based, bounds for the target distribution. Then, exploiting existing asymptotic results in the main-mass region, an explicit, moment-based approximation of the target probability density function is provided. Although the latter cannot be considered, in general, as a satisfactory solution, it can always serve as an initial approximation in any iterative scheme for the numerical solution of the moment problem. Numerical results illustrating all the theoretical statements are also presented.     

提高维吾尔文搜索质量的一些关键技术研究

维吾尔语和土耳其语、乌兹别克语、哈萨克语等同属于阿尔泰语系突厥语分支,是一种粘着型语言,具有丰富的构词合构形附加成分.主要分布于中国新疆维吾尔自治区,以及中西亚地区的维吾尔族使用.目前使用基于阿拉伯文字母的维吾尔文.当前用google,bing等搜索引擎搜索维吾尔文时搜索结果不包括该词语的构词合构形附加成分等其它形式.严重影响了搜索结果的准确性.通过词干提取和词缀附加生成词语组合的方法以解决提高维吾尔文搜索质量的问题.此问题解决,将在大幅度提高维吾尔文搜索的准确性,并针对提高其它突厥语族语言(土耳其语、乌兹别克语、哈萨克语等)的搜索质量、数据挖掘、数据分析、数据安全、自然语言理解等研究领域有一定的参考价值.     

绿色产品定价与市场入侵研究:基于市场权力结构的视角

为研究绿色产品定价与市场入侵问题,针对由两个制造商和一个零售商组成的系统,构建了制造商主导、零售商主导以及供需双方均势等三种权力结构模型,分析了权力结构、消费者绿色偏好等因素对绿色产品定价和市场入侵的影响。研究发现:绿色产品入侵市场存在可行条件,并且绿色产品市场入侵会导致普通产品制造商的利润损失;市场入侵对零售商有利,说明零售商会有动机引入绿色产品;绿色产品入侵市场和市场权力结构均会影响各个企业最优定价策略,其中拥有主导权的企业会选择高价策略;权力结构会影响供需双方的利润分配,以及导致系统利润的损失。     

金融风险在股票市场的传染效应及联动行为分析

近年来金融危机频发并表现出了易传染性,引起了众多学者的高度关注。以动态条件相关模型研究美欧股市与中、日、韩股市间的时变相关性,并结合内生多重结构突变模型划分危机传染阶段,选用溢出指数模型分析股市间的风险溢出特性;随后,定义股市间相互影响的联动模式并构建不同传染阶段的加权有向网络图分析股市间的联动行为。研究表明:美欧股市对中日韩股市有明显的传染效应,被传染的速度和持续时间均不相同;金融传染和风险溢出展现出一定的不一致性,危机期间日股的风险溢出效应强于美股;传染效应在联动网络中表现为联动模式的高聚类性和高联动性,相比欧债危机,次贷危机时期股市间展现出更强的联动行为;日股与美欧股市在两次危机中均表现出最强的联动性,其所受影响也最大。     

Personae at play. ‘Men of Mathematics’ in commentary

Mathematical writers, above all, Euclid, tend to present their theorems as decontextualized, abstract propositions, which has become the standard modus of textual presentation in theoretical mathematics. Mathematical commentators, however, provide their readers with personal names and historical facts in order to elucidate problems, provide contexts of discovery, or construct doxographies, among other things. Modern readers have used such information for the construction of histories of science. When we look at these passages, however, we see that personal names and information about mathematicians of the past can serve quite a range of different objectives, such as the strategic self-positioning of the commentator vis-à-vis the past or present of mathematics, the education of the reader, mathematical or moral, the construction of the history of the field, etc. Not only does the commentator present a persona of himself to the reader, he can also turn colleagues and predecessors into personae. This paper attempts to elucidate the practice, by offering four examples of such plays of and with personae, in Pappus, Eutocius, al-Nayrīzī, and Proclus.     

On the mechanical and thermomechanical properties of low-density polyethylene/ethylene-α-octene copolymer blends

Blends of low-density polyethylene (LDPE) and ethylene-octene copolymer (EOC) were obtained. The effect of EOC content and absorbed radiation dose on the mechanical and thermomechanical properties of LDPE/EOC blends are investigated. Particular attention is given to a tensile stress-strain analysis and the “form-memory” effect of the blends. With growing LDPE content, the elastic modulus, the yield stress, and the thermorelaxation and residual stresses of the blends increase, but the ultimate elongation at break decreases, which is caused by the higher crystallinity of polyethylene. As a result of radiation-induced cross-linking, the elastic modulus, the yield stress (at a 1% strain), the ultimate yield strength, and the thermorelaxation and residual stresses increase, while the ultimate elongation at break and the melt flow-behavior index decrease, which is confirmed by the growing gel fraction in the blend. __________ Translated from Mekhanika Kompozitnykh Materialov, Vol. 44, No. 2, pp. 279–286, March–April, 2008.     

Representation of the solution to a model problem in semiconductor physics

We study a mixed type problem for the Poisson equation arising in the modeling of charge transport in semiconductor devices [V. Romano, 2D simulation of a silicon MESFET with a non-parabolic hydrodynamical model based on the maximum entropy principle, J. Comput. Phys. 176 (2002) 70-92; A.M. Blokhin, R.S. Bushmanov, A.S. Rudometova, V. Romano, Linear asymptotic stability of the equilibrium state for the 2D MEP hydrodynamical model of charge transport in semiconductors, Nonlinear Anal. 65 (2006) 1018-1038]. Unlike well-studied elliptic boundary-value problems in domains with smooth boundaries (see, for example, [O.A. Ladyzhenskaya, N.N. Uralceva, Linear and Quasilinear Elliptic Equations, Nauka, Moscow, 1973; D. Gilbarg, N.S. Trudinger, Elliptic Partial Differential Equations of Second Order, Springer-Verlag, Berlin, 1983]), our problem has two significant features: firstly, the boundary is not a smooth curve and, secondly, the type of boundary conditions is mixed (the Dirichlet condition is satisfied on the one part of the boundary whereas the Neumann condition on the other part). The well-posedness of the problem in Hölder and Sobolev spaces is proved. The representation of the solution to the problem is obtained in an explicit form.