多背包问题的半定松驰算法

The multiple knapsack problem denoted by MKP (B,S,rn,n) can be defined as follows. A set B of n items and a set S of rn knapsacks are given such that each item j has a profit pi and weight wj,and each knapsack i has a capacity Ci. The goal is to find a subset of items of maximum profit such that they have a feasible packing in the knapsacks. MKP (B,S,m,n) is strongly NP-Complete and no polynomial time approximation algorithm can have an approximation ratio better than 0.5. In the last ten years,semi-definite programming has been empolyed to solve some combinatorial problems successfully. This paper firstly presents a semi-definite relaxation algorithm (MKPS) for MKP (B,S,rn,n). It is proved that MKPS have a approximation ratio better than 0. 5 for a subclass of MKP (B,S,m,n) with n≤100, m≤5 and max^nj=1{wj}/min^mi=1={Ci}≤2/3.     

解线性不等式的神经网络

本文提出两个解线性不等式的Hopfiedl-Tank型的神经网络。第一个网络模拟同时松弛投影方法，第二个网络是二次规划方法。当线性不等式的解集非空时，这两个方法都给出该线性不等式的解。同时我们还给出了这两个网络的数值模拟。     

一种可压缩流动的高阶加权紧致非线性格式(WCNS)的加速收敛方法

以高超声速粘性绕流的数值模拟为例,研究LU-SGS、高斯-赛德尔点松弛、线松弛以及GMRES等隐式求解方法在空间项采用高阶精度格式WCNS离散时的收敛性,并对GMRES(generalized minimal residual)方法中的子迭代影响作了对比计算.结果表明,采用准确的解析雅克比矩阵的点、线松弛的收敛速度优于LU-SGS,以线松弛为预处理的GMRES算法具有良好的收敛特性.     

“例谈发展课本例习题的常用策略”中两个结论的订正

张俊老师在文[1]中给出了发展课本例习题的常用策略,本人受益良多,但其中的新题23与新题25的结论有误,现指出并予以证明.     

用等离子体粒子能谱拟合温度的新方法

用最小二乘法拟合等离子体粒子能谱实验数据时，通常使拟合函数与实验能谱数据之间的误差平方和极小化。如果将能谱实验数据的对数拟合成直线以求得粒子温度，则最好对温度误差的平方和极小化。给出了这种拟合方法的计算公式，并对两种不同方法得到的结果进行了比较。     

离差平方和分解的投影分析

以双因素协方差分析为例,介绍如何用代数中投影的方法进行离差平方和分解.结果表明:这种分解方法比通常采用的分析方法更简洁明了.     

研究岩石破裂等弱化行为的力－位移组合控制方法

本文介绍了一种用力和位移的线性组合控制岩石试验的新方法。该方法已在一台自制的伺服实验系统上和一台MTS实验系统上得到了实现。用这种方法能够简便地求出通常难以求出的具有Ⅱ类特性岩石的全应力-应变曲线,而且能够变化出如广义应力松驰等其它实验类型,从而能够对时间依赖性进行深入的研究。由于其易于采用的特点,它在研究岩石破裂乃至摩擦等弱化过程方面有着广泛的应用前景。     

ON SOLVABILITY AND WAVEFORM RELAXATION METHODS FOR LINEAR VARIABLE-COEFFICIENT DIFFERENTIAL-ALGEBRAIC EQUATIONS

This paper is concerned with the solvability and waveform relaxation methods of linear variable-coefficient differential-algebraic equations （DAEs）. Most of the previous works have been focused on linear variable-coefficient DAEs with smooth coefficients and data, yet no results related to the convergence rate of the corresponding waveform relaxation methods has been obtained. In this paper, we develope the solvability theory for the linear variable-coefficient DAEs on Legesgue square-integrable function space in both traditional and least squares senses, and determine the convergence rate of the waveform relaxation methods for solving linear variable-coefficient DAEs.