Exactly satisfying initial conditions neural network models for numerical treatment of first Painlevé equation

In this paper, novel computing approach using three different models of feed-forward artificial neural networks (ANNs) are presented for the solution of initial value problem (IVP) based on first Painlevé equation. These mathematical models of ANNs are developed in an unsupervised manner with capability to satisfy the initial conditions exactly using log-sigmoid, radial basis and tan-sigmoid transfer functions in hidden layers to approximate the solution of the problem. The training of design parameters in each model is performed with sequential quadratic programming technique. The accuracy, convergence and effectiveness of the proposed schemes are evaluated on the basis of the results of statistical analyses through sufficient large number of independent runs with different number of neurons in each model as well. The comparisons of these results of proposed schemes with standard numerical and analytical solutions validate the correctness of the design models.     

Reduced modified quadratures for quadratic membrane finite elements

Reduced integration is frequently used in evaluating the element stiffness matrix of quadratically interpolated finite elements. Typical examples are the serendipity (Q8) and Lagrangian (Q9) membrane finite elements, for which a reduced 2 × 2 Gauss–Legendre integration rule is frequently used, as opposed to full 3 × 3 Gauss–Legendre integration. This ‘softens’ these element, thereby increasing accuracy, albeit at the introduction of spurious zero energy modes on the element level. This is in general not considered problematic for the ‘hourglass’ mode common to Q8 and Q9 elements, since this spurious mode is non‐communicable. The remaining two zero energy modes occurring in the Q9 element are indeed communicable. However, in topology optimization for instance, conditions may arise where the non‐communicable spurious mode associated with the elements becomes activated. To effectively suppress these modes altogether in elements employing quadratic interpolation fields, two modified quadratures are employed herein. For the Q8 and Q9 membrane elements, the respective rules are a five and an eight point rule. As compared to fully integrated elements, the new rules enhance element accuracy due to the introduction of soft, higher‐order deformation modes. A number of standard test problems reveal that element accuracy remains comparable to that of the under‐integrated counterparts. Copyright © 2004 John Wiley & Sons, Ltd.     

多目标二次规划非劣解集的理论生成及应用

本文对多目标二次规划问题非劣解集的理论生成进行了探讨,在介绍二次规划单目标问题优化解求法的基础上,提出了在决策平面上互切点,共交点轨迹的概念.由此导出了非劣解集的一般解析式(对切点解问题)和另一类交点解问题时的非劣解集求法,然后介绍了在目标平面非劣解集的解析表达式.使多目标二次规划问题整个非劣解集的推求严格、明确和方便,有理论和实际意义。文中以一个综合利用水库参数的多目标优选作为例子阐述了应用,并求出了最终的最佳协调解.     

关于布尔函数的二次逼近

本文首先利用矩阵形式引入了二阶ｗａｌｓｈ谱的概念并讨论了其性质，其次引入了二次ｂｅｎｔ函数的概念，同时论述了二次ｂｅｎｔ函数的密码学意义。     

Class normalization in centroid-based text categorization

Centroid-based categorization is one of the most popular algorithms in text classification. In this approach, normalization is an important factor to improve performance of a centroid-based classifier when documents in text collection have quite different sizes and/or the numbers of documents in classes are unbalanced. In the past, most researchers applied document normalization, e.g., document-length normalization, while some consider a simple kind of class normalization, so-called class-length normalization, to solve the unbalancedness problem. However, there is no intensive work that clarifies how these normalizations affect classification performance and whether there are any other useful normalizations. The purpose of this paper is three folds; (1) to investigate the effectiveness of document- and class-length normalizations on several data sets, (2) to evaluate a number of commonly used normalization functions and (3) to introduce a new type of class normalization, called term-length normalization, which exploits term distribution among documents in the class. The experimental results show that a classifier with weight-merge-normalize approach (class-length normalization) performs better than one with weight-normalize-merge approach (document-length normalization) for the data sets with unbalanced numbers of documents in classes, and is quite competitive for those with balanced numbers of documents. For normalization functions, the normalization based on term weighting performs better than the others on average. For term-length normalization, it is useful for improving classification accuracy. The combination of term- and class-length normalizations outperforms pure class-length normalization and pure term-length normalization as well as unnormalization with the gaps of 4.29%, 11.50%, 30.09%, respectively.     

Existence of discrete-time LQG-controllers

Existence results for the LQG-controller are investigated. An infimal Riccati equation based controller may potentially give closed loop eigenvalues on the unit circle. Assuming left and right invertibility it is shown that there exists an optimal controller if and only if the Riccati equation based controller stabilizes the closed loop system after removal of all its unobservable and uncontrollable modes. Furthermore this reduced controller is the optimal controller, and its transfer function is unique. This existence condition is a considerable simplification of the more general geometric condition recently derived by Trentelman and Stoorvogel.     

Metaheuristic downhill simplex method in combinatorial optimization

The concept of combinatorial objects is formalized. It allows strict definition of a combinatorial optimization problem (COP). An efficient metaheuristic method to solve COPs (H-method) is considered. It includes stochastic local search algorithms as a built-in procedure. A parallel implementation of the H-method is set forth and analyzed. The results from a numerical experiment and solution of some well-known COPs on personal computers and on the SKIT cluster supercomputer are presented. The study was supported by INTAS (Project 06-1000017-8909). __________ Translated from Kibernetika i Sistemnyi Analiz, No. 6, pp. 70–79, November–December 2007.     

基于SPOT-5多光谱影像的矿区塌塘水体提取方法研究

针对基于遥感影像的水体提取方法存在水体提取不完整和误提的现象,提出了一种基于SPOT-5多光谱影像的矿区塌塘水体提取方法。在利用波段合成增加一个可用波段的基础上对已有的水体提取方法进行适当的改进,并基于决策树分类器和改进后的方法进行矿区水体的四级提取,保证了水体提取的完整性,同时减少了误提率;最后利用实测数据对水体提取的精度进行了评定。试验结果表明,基于决策树分类器的水体提取方法具有较高的精度,能满足矿区实际应用的需要。     

gain bounds for systems with sector bounded and slope‐restricted nonlinearities

This paper proposes a new method for calculating a bound on the gain of a system comprising a linear time invariant part and a static nonlinear part, which is odd, bounded, zero at zero and has a restriction on its slope. The nonlinear part is also assumed to be sector bounded, with the sector bound being (possibly) different from that implied by the slope restriction. The computation of the gain bound is found by solving a set of linear matrix inequalities, which arise from an integral quadratic constraint formulation of a multiplier problem involving both Zames‐Falb and Popov multipliers. Examples illustrate the effectiveness of the results, and comparisons are made against the state‐of‐the‐art. Copyright © 2011 John Wiley & Sons, Ltd.     

不确定离散系统的鲁棒Gl2滤波器设计

对于凸多面体不确定离散系统,研究了满足误差估计系统鲁棒稳定且满足一定Gl₂性能指标的线性全阶稳定滤波器设计方法.分别给出了确定性和不确定性离散系统具有Gl₂性能的线性矩阵不等式（linear matrix inequality, LMI)充要条件和充分条件.在此基础上,通过使用滤波器参数化方法,分别推导出了以LMI描述的确定性和不确定性系统的Gl₂滤波器存在的充要条件和充分条件,进而可以使用LMI凸优化程序求解滤波器.通过对算例的计算,结果表明,在噪声信号为独立范数有界情况下,鲁棒Gl₂滤波比H_∞滤波具有更少的保守性.