组合算法及其应用

提出了在组合算法设计和组合算法选择方面所应当遵循的三个原则,即通用性、可计算性和较少的信息冗余量,并初步分析了它们之间的相互关系.通过分析,提示了组合算法的设计方法,改进方向和优化技术.     

Gowers norms and pseudorandom measures of subsets

Let $A \subset {{\Bbb Z}_N}$, and ${f_A}(s) = \left\{ {\begin{array}{*{20}{l}}{1 - \frac{{|A|}}{N},}&{{\rm{for}}\;s \in A,}\\{ - \frac{{|A|}}{N},}&{{\rm{for}}\;s \notin A.}\end{array}} \right.$ We define the pseudorandom measure of order k of the subset A as follows, P_k(A, N) = $\begin{array}{*{20}{c}}{\max }\\D\end{array}$|$\mathop \Sigma \limits_{n \in {\mathbb{Z}_N}}$f_A(n + c₁)f_A(n + c₂) … f_A(n + c_k)|, where the maximum is taken over all D = (c₁, c₂, . . . , c_k) ∈ ${\mathbb{Z}^k}$ with 0 ≤ c₁ < c₂ < … < c_k ≤ N - 1. The subset A ⊂ ${{\mathbb{Z}_N}}$ is considered as a pseudorandom subset of degree k if P_k(A, N) is “small” in terms of N. We establish a link between the Gowers norm and our pseudorandom measure, and show that “good” pseudorandom subsets must have “small” Gowers norm. We give an example to suggest that subsets with “small” Gowers norm may have large pseudorandom measure. Finally, we prove that the pseudorandom subset of degree L(k) contains an arithmetic progression of length k, where L(k) = 2·lcm(2, 4, . . . , 2|$\frac{k}{2}$|), for k ≥ 4, and lcm(a₁, a₂, . . . , a_l) denotes the least common multiple of a₁, a₂, . . . , a_l.     

Toader型平均的算术与调和平均界

给出了Toader型平均T[A(a,b),G(a,b)]关于调和平均H(a,b)与算术平均A(a,b)组合的精确界.作为应用,发现了几个关于第二类完全椭圆积分的精确不等式.     

线性互补问题的异步并行多分裂松弛迭代算法

将求解线性方程组的异步并行多分裂松弛迭代算法推广到线性互补问题.当问题的系数矩阵为H-矩阵类时,证明了算法的全局收敛性.     

线性互补问题的并行多分裂松弛迭代算法

运用矩阵多重分裂理论,同时考虑并行计算与松弛迭代法,得到一类求解线性互补问题的高效数值算法．当问题的系数矩阵为对角元为正的H-矩阵或对称半正定矩阵时,证明了算法的全局收敛性；该算法与已有算法相比,具有计算量小、计算速度快等特点,因而特别适于求解大规模问题．数值试验的结果说明了算法的有效性．     

线性方程组异步并行迭代法的舍入误差分析

本文对求解大型线性方程组的异步并行迭代法进行了浮点运算的舍入误差分析,给出了算法是向前稳定的充分条件.     

分类问题的在线顺序双并联快速学习机

双并联前馈神经网络模型是单层感知机和单隐层前馈神经网络的混合结构,本文构造了一种双并联快速学习机算法,与其他类似算法比较,提出的算法能利用较少的隐层单元及更少的待定参数,获得近似的学习性能.数值实验表明,对很多实际分类问题,提出的算法具备更佳的泛化能力,因而可以作为快速学习机算法的有益补充.     

面向对象的串并行有限元分析系统

研究了面向对象的串并行有限元分析系统的关键实现技术。首先介绍的有限元系统类库分析,重点阐明了面向对象方法应用于串并行有限元分析所具有的优势,基于此将集成系统归纳成多个组成类库:包括数学计算类库、数据处理类库、有限元基本类库、数值分析模型类库、图形类库以及应用框架工具类库等,进而介绍分析程序的面向对象实现方法及过程,最后形成一个集成的交互式并行有限元分析框架OOParaFEA(Object OrientedParallel FEA framework)。系统快速高效的实现,表明面向对象的方法的确从软件的可维护性、可扩充性以及减少复杂性等方面带来前所未有的优势,并且由于其固有的继承特性而使其特别适合团队协作开发大型有限元分析软件,从而给我国大规模科学计算软件的研制提供一个新思路。     

An Efficient Parallel Reverse Conversion of Residue Code to Mixed-Radix Representation Based on the Chinese Remainder Theorem

In this paper, we deal with the critical problems in residue arithmetic. The reverse conversion from a Residue Number System (RNS) to positional notation is a main non-modular operation, and it constitutes a basis of other non-modular procedures used to implement various computational algorithms. We present a novel approach to the parallel reverse conversion from the residue code into a weighted number representation in the Mixed-Radix System (MRS). In our proposed method, the calculation of mixed-radix digits reduces to a parallel summation of the small word-length residues in the independent modular channels corresponding to the primary RNS moduli. The computational complexity of the developed method concerning both required modular addition operations and one-input lookup tables is estimated as , where k equals the number of used moduli. The time complexity is modular clock cycles. In pipeline mode, the throughput rate of the proposed algorithm is one reverse conversion in one modular clock cycle.     

算术运算的VHDL描述方法探讨

针对在VHDL语言教学过程中,如何进行算术运算的VHDL描述问题进行了探讨.文中首先论述了VHDL语言描述算术运算的局限性;重点阐述了加运算、减运算、乘运算、除运算、取模运算、取余运算的VHDL描述方法;由于设计的结果是一个专用的算式电路,不运行任何程序,工作的可靠性和稳定性都优于用CPU设计.