线性约束规划内点法及其修正算法

凸规划的内点算法是目前较热门的课题之一，参考资料「２」，「３」等均给出了较深入的研究，本文在参考前人的工作前提下，提出了带线性约束凸规划的内点算法结论及相应算法，另外，本文定义了偏移因子，偏移因子对的概念，对下降方向作出了修正，并给出了相关算法。     

一个新的SQP方法及其超线性收敛性

由Ｗｉｌｓｏｎ,Ｈａｎ,Ｐｏｗｅｌｌ发展的ＳＱＰ技术是解非线性规划的最有效的方法之一,但是,如果其中的二次子规划问题无可行解或者其搜索方向向量无界,该方法ａｎ和Ｂｕｒｋｅ「３」,周广路「２」分别对二次规划问题作了修正,克服了上述矛盾,本文在「２」的基础上,进上步修正,证明在Ａｒｍｉｊｏ搜索下算法具有全局收敛性,并通过解一辅助线性方程组,利用弧式搜索,得出该方法具有超线性收敛性。     

带球（椭球）约束的不定二次规划问题

本文证明了带球（椭球）约束的不定二次规划问题具有强Ｌａｇｒａｎｇｅ对偶性，设计了一个求解这类问题的算法，本语文的结论比文「７」强，所设计的算法比文「７」简洁。     

格上Ising模型的临界失真估计

本文是我们先前工作「１」，「２」的继续，对格上Ｉｓｉｎｇ模型的临界失真ｄｃ的估计，Ｎｅｗｍａｎ和Ｂａｋｅｒ「６」证明了ｄｃ和Ｉｓｉｎｇ模型的Ｍａｙｅｒ级数之收敛半径Ｒ有以下关系：ｄｃ＝Ｒ／（１＋Ｒ），在「１」中匀提出了估计Ｒ及ｄｃ的新方法，并它计算了二维矩形格Ｚ＾２上Ｉｓｉｎｇ模型的临界失真ｄｃ，此文中我们继续应用此方法首次计算了定义在其它二维和三维格上Ｉｓｉｎｇ模型的临界失真ｄｃ，数值计算的结果     

关于一类变分不等式的迭代算法

本文通过对「１」中相应结构的讨论与改进,利用变分不等式求解技巧,讨论了Ｈｉｌｂｅｒｔ空间中一类广义变分不等式的解的迭代算法及其收敛性,指出了文「１」中主要结果的证明的一些不足之处,本文所得结果推广并改进了「１」的相应结论。     

平面弹性力学中带孔洞的自由边界含裂纹问题

本文在文献「１」、「２」的基础上提出了一类平面弹性力学中带多边形孔洞的自由边界含裂纹问题，我们先将其分解为一个解析函数的黎曼－希尔伯特边值问题和一个复方程的混合边值问题，然后利用边值问题的理论和方法，讨论了这个问题的可解性。     

关于拟合值方改变敏感度的一个下界

本文对线性回归ＧＭ－估计的一个子类，给出拟合值方差改变敏感度的一个下界，并说明「１」中相应于ｎ的估计量达到此下界。     

关于迭代序列的子列极限点

本文讨论了度量空间上一个连续自映射在一点处的迭代序列的子列极限点集的结构，所得的结果统一和推广了Ｄｉａｚ和Ｍｅｔｃａｌｆ「１」，Ｍａｉｔｉ和Ｂａｂｕ「２，４」和Ｐａｒｋ「３」的若干结果。     

两类求解y“=f（x,y）具有极小相位延迟的显式方法

本文提出了两类数值积分二阶周期性初值问题ｙ〃＝ｆ（ｘ，ｙ），ｙ（ｘ０）＝ｙ０，ｙ（ｘ０）＝ｙ０具有检小相位延迟的显式两步法。这些方法推广和改进了文献「１」１－「７」中的某些方法。数值试验表明本文中的某些方法优于「１」－「７」中的某些方法。     

希尔伯特空间中半线性随机发展方程适度解的存在性

本文讨论如下形式的希尔伯特空间中半线性随机发展方程Ｃａｕｃｈｙ问题｛ｄｙ（ｔ）「Ａｙ（ｔ）＝ｆ（ｔ，ｙ（ｔ））」ｄｔ＋Ｇ（ｔ，ｙ（ｔ））ｄｗ（ｔ）ｙ（０）＝ｙ０适度解的存在性。在一组条件下得到了解的整体存在性，推广了文「１」的存在性定理。     

离散点集上的向量有理插值算法与特征性质

1引言 给出R~2中离散点集     

Schensted Algorithms for Dual Graded Graphs

This paper is a sequel to [3]. We keep the notation and terminology and extend the numbering of sections, propositions, and formulae of [3].The main result of this paper is a generalization of the Robinson-Schensted correspondence to the class of dual graded graphs introduced in [3], This class extends the class of Y-graphs, or differential posets [22], for which a generalized Schensted correspondence was constructed earlier in [2].The main construction leads to unified bijective proofs of various identities related to path counting, including those obtained in [3]. It is also applied to permutation enumeration, including rook placements on Ferrers boards and enumeration of involutions.As particular cases of the general construction, we re-derive the classical algorithm of Robinson, Schensted, and Knuth [19, 12], the Sagan-Stanley [18], Sagan-Worley [16, 29] and Haiman's [11] algorithms and the author's algorithm for the Young-Fibonacci graph [2]. Some new applications are suggested.The rim hook correspondence of Stanton and White [23] and Viennot's bijection [28] are also special cases of the general construction of this paper.In [5], the results of this paper and the previous paper [3] were presented in a form of extended abstract.     

Counterexamples to the Connectivity Conjecture of the Mixed Cells

In [4] a conjecture concerning the connectivity of mixed cells of subdivisions for Minkowski sums of polytopes was formulated. This conjecture was, in fact, originally proposed by Pedersen [3]. It turns out that a positive confirmation of this conjecture can substantially speed up the algorithm for the ``dynamical lifting' developed in [4]. In the mean time, when the polyhedral method is used for solving polynomial systems by homotopy continuation methods [2], the positiveness of this conjecture also plays an important role in the efficiency of the algorithm. Very unfortunately, we found that this conjecture is inaccurate in general. In Section 1 a counterexample is presented for a general subdivision. In Section 2 another counterexample shows that even restricted to ``regular' subdivisions induced by liftings, this conjecture still fails to be true. Received September 18, 1996, and in revised form February 17, 1997.     

The asymptotic behavior of an inertial alternating proximal algorithm for monotone inclusions

The aim of this paper is to investigate the asymptotic behavior of an inertial alternating algorithm based on the composition of resolvents of monotone operators. The proposed algorithm is a generalization of those proposed in Attouch et al. (2007) [3] and Bauschke et al. (2005) [1]. As a special case, we also recover the classical alternating minimization algorithm (Acker, 1980) [2], which itself is a natural extension of the alternating projection algorithm of von Neumann (1950) [4]. An application to equilibrium problems is also proposed.     

An H‐matrix Type Preconditioner For Frictional Contact Problems

The bilateral or unilateral contact problem with Coulomb friction between two elastic bodies is considered [1]. An algorithm is introduced to solve the resulting finite element system by a non‐overlapping domain decomposition method [2, 3]. The global problem is transformed to a independant local problems posed in each bodie and a problem posed on the contact surface (the interface problem). The solution is obtained by using a successive approximation method, in each step of this algorithm we solve two intermediate problems the first with prescribed tangential pressure and the second with prescribed normal pressure [8]. Our preconditioner construction is based on the application of the H‐matrix technique [6, 7] together with the representation of the H^1/2 seminorm by a sum of partial seminorms [4]. (© 2004 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

sum from n=1 to +∞(1/n~(2m))的初等解法

本文用初等的方法研究sum from n=1 to(1/n~(2m))(m∈N)的求和问题。这个问题最先由Euler[8]解决。文献[1][6]给出了另两种求解方法。特别地,对于m=1的情形,即sum from n=1 to ∞(1/n~2)=((π~2)/6),已有许多不同的证明方法,可见文献[2][3][4][5]以及那里的参考文献。本文的想法,主要受文献[5][6]的启发而来的。     

正定可对称化矩阵与预对称迭代算法

１．问题的提出 我们引入正定可对称化矩阵定义的背景是为了研究求解二阶椭圆型非自共轭方程的离散迭代有效算法、这类方程的椭圆型是本质的分析性质。是由二阶项决定的,在离散方程中表现为正定性;非自共轭性则是由方程中的一阶项引起的,在相当广泛一类问题中可通过变量代换化为自共轭。因此,我们称这类问题为正定可对称化问题。 例１．高维二阶常系数椭圆型方程其中 Ａ为常系数正定对称（ｓ．ｐ．ｄ）阵, 为正交阵, Ｄ是对角元素为正的对角阵。 先作变量代换,通过演算,偏微分方程对于新变量变成这里进而令可将原非自共轭偏微分算子…     

Energy-consistent time-integration for a dynamic finite deformation thermoviscoelastic continuum

The main goal of the present work is to develop a structure-preserving time integrator, which is based on the works of Öttinger [2] (GENERIC formalism) and Romero [4] (Thermodynamically consistent (TC) algorithm). This intergator is called the enhanced TC algorithm for thermoviscoelastic continuums. The underlying structure is then called enhanced GENERIC, analogously to the earlier puplished thermoviscoelastic double pendulum of Krüger et al. [1]. (© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Rosen's Gradient Projection with discrete steps

In 1960, J. B. Rosen gave a famous Gradient Projection Method in [1]. But the convergence of the algorithm has not been proved for a long time. Many authors paid much attention to this problem, such as X.S. Zhang proved in [2] (1984) that the limit point of {x _k} which is generated by Rosen's algorithm is a K-T piont for a 3-dimensional caes, if {x _k} is convergent. D. Z. Du proved in [3] (1986) that Rosen's algorithm is convergent for 4-dimensional. In [4] (1986), the author of this paper gave a general proof of the convergence of Rosen's Gradient Projection Method for ann-dimensional case. As Rosen's method requires exact line search, we know that exact line search is very difficult on computer. In this paper a line search method of discrete steps are presented and the convergence of the algorithm is proved.     

On 2-dimensional insertion-deletion Reed-Solomon codes with optimal asymptotic error-correcting capability

Reed-Solomon codes have gained a lot of interest due to its encoding simplicity, well structuredness and list-decoding capability [6] in the classical setting. This interest also translates to other metric setting, including the insertion and deletion (insdel for short) setting which is used to model synchronization errors caused by positional information loss in communication systems. Such interest is supported by the construction of a deletion correcting algorithm of insdel Reed-Solomon code in [22] which is based on the Guruswami-Sudan decoding algorithm [6]. Nevertheless, there have been few studies [3] on the insdel error-correcting capability of Reed-Solomon codes.In this paper, we discuss a criterion for a 2-dimensional insdel Reed-Solomon codes to have optimal asymptotic error-correcting capabilities, which are up to their respective lengths. Then we provide explicit constructions of 2-dimensional insdel Reed-Solomon codes that satisfy the established criteria. The family of such constructed codes can then be shown to extend the family of codes with asymptotic error-correcting capability reaching their respective lengths provided in [3, Theorem 2] which provide larger error-correcting capability compared to those defined in [25].