柯西矩阵的三角分解及应用

1 引 言 矩阵分解具有非常重要的应用.例如,可以利用矩阵的LU分解回代求解线性方程组Ax=b.对于在有理函数的计算中经常遇到的以柯西矩阵为系数的线性方程组的求解问题,需要做柯西矩阵的三角分解.     

混合边界条件下的三维定常 旋转Navier-Stokes方程的原始变量有限元计算

本文利用原始变量有限元法求解混合边界条件下的三维定常旋转Navier-Stokes方程,证明了离散问题解的存在唯一性,得到了有限元解的最优误差估计.给出了求解原始变量有限元逼近解的简单迭代算法,并证明了算法的收敛性.针对三维情况下计算资源的限制,采用压缩的行存储格式存储刚度矩阵的非零元素,并利用不完全的LU分解作预处理的GMRES方法求解线性方程组.最后分析了简单迭代和牛顿迭代的优劣对比,数值算例表明在同样精度下简单迭代更节约计算时间.     

Transputer上Cholesky分解的并行实现

§1.引言 对称正定矩阵A的Cholesky分解在求解线性系统Ax-b中非常重要,如果R是上三角矩阵,使得A=R~TR,则求解上述方程组可以通过向前及向后迭代来完成。然而求解一个线性系统,主要是计算系数矩阵的分解。这里主要是介绍如何有效地并行求矩阵R。在串行机上,已经有了很好的实现方法,如[1]至于如何在并行机上实现,是本文的目的。 众所周知,在并行机上求解大规模问题是今后科学与工程计算的必然发展方向。然     

保护私有信息的一般线性方程组计算协议

作为科学计算的一个重要问题,保护私有信息的线性方程组的求解在金融、机械及通信等领域有着广泛的应用.在不经意传输的意义下,利用有限域上计算Moore-Penrose伪逆矩阵的概率算法,设计新的安全协议,解决了隐私保护的一般线性方程组在有限域上的安全两方计算问题,并利用模拟范例证明该协议在半诚实模型下是安全的.     

不等式约束优化的一个超线性收敛FSSLE算法

本文针对不等式约束优化问题,提出了一个可行序列线性方程组(FSSLE)算法.该算法每次迭代只需求解四个具有相同系数矩阵的线性方程组,因而计算量较小.在没有假设算法产生的聚点是孤立点和近似乘子列有界的条件下,证明了算法具有全局收敛性.在一般条件下,证明了算法具有超线性收敛性.     

不等式约束优化全局收敛的滤子线性方程组算法

设计了求解不等式约束非线性规划问题的一种新的滤子序列线性方程组算法,该算法每步迭代由减小约束违反度和目标函数值两部分构成.利用约束函数在某个中介点线性化的方法产生搜索方向.每步迭代仅需求解两个线性方程组,计算量较小.在一般条件下,证明了算法产生的无穷迭代点列所有聚点都是可行点并且所有聚点都是所求解问题的KKT点.     

不等式约束优化一个可行序列线性方程组算法

提出了求解非线性不等式约束优化问题的一个可行序列线性方程组算法. 在每次迭代中, 可行下降方向通过求解两个线性方程组产生, 系数矩阵具有较好的稀疏性. 在较为温和的条件下, 算法具有全局收敛性和强收敛性, 数值试验表明算法是有效的.     

贯序和并行求解线性与非线性方程组的一类算法

1.引论 Abaffy,Broyden和spedicato在最近的论文中,提出了一类求解线性和非线性方程组的算法(有可能推广于求解其它问题,例如最优化问题).我们首先给出这类算法求解线性方程组时的基本形式.设线性方程组为 或把它写成矩阵形式 其中A=(a_1,…,a_m)是n×m阶矩阵,共秩q可以小于m.算法具有拟Newton型结构,其计算步骤如下:     

求解不等式约束优化问题无严格互补松弛条件的QP-Free新算法

本文针对不等式约束优化问题,结合Facchinei-Fischer-Kanzow精确有效集识别技术,给出—个新的线性方程组与辅助方向相结合的可行下降算法.算法每步迭代只需求解一个降维的线性方程组或计算一次辅助方向,且获取辅助方向的投影矩阵只涉及近似有效约束集中的元素,问题规模大为减少,且当迭代次数充分大时,只需求解一个降维的线性方程组.无需严格互补松弛条件,算法全局且一步超线性收敛.     

求解H-矩阵线性方程组的几种预处理迭代法

在求解H-矩阵线性方程组预处理Gauss-Seidel迭代法的基础上,提出了一种渐变预处理技术,提高了H-阵线性方程组的求解效率,加快了Gauss-Seidel迭代法的收敛速度.同时,讨论了两种特殊形式的预处理子:上Hessenberg预处理矩阵和下Hessenberg预处理矩阵,并证明了算法的收敛性.最后用数值实验验证了算法的可行性及有效性.     

一类代数方程组的异步并行计算的收敛性

异步并行算法是MIMD(多指令流多数据流)计算机系统给数值分析提出的新问题.这里所说的异步并行算法,是将一个数值计算问题分成有关的几个子问题,每个子问题在一台计算机(或一个处理器)上计算,各机共用的数据通过公共存贮器进行交换.每台计     

Cost-efficient numerical algorithm for solving the linear inverse problem of finding a variable magnetization

The paper is devoted to developing an original cost-efficient algorithm for solving the inverse problem of finding a variable magnetization in a rectangular parallelepiped. The problem is ill-posed and is described by the integral Fredholm equation. It is shown that after discretization of the area and approximation of the integral operator, this problem is reduced to solving a system of linear algebraic equations with the Toeplitz-block-Toeplitz matrix. We have constructed the memory efficient variant of the stabilized biconjugate gradient method BiCGSTABmem. This optimized algorithm exploits the special structure of the matrix to reduce the memory requirements and computing time. The efficient implementation is developed for multicore CPU and GPU. A series of the model problems with synthetic and real magnetic data are solved. Investigation of efficiency and speedup of parallel algorithm is performed.     

解线性代数方程组的新型二次PEk方法

1引言在电离层动力学和飞行器设计等工程领域,经常遇到具有周期边界条件的椭圆型或抛物型偏微分方程的求解问题．通过适当的离散逼近,此类问题可以转化为大型块状三对角线性方程组的求解问题．1977年,William S．Helliwell提出了一种(Pseudo- Elimination)方法来求解系数矩阵为块状三对角矩阵的线性代数方程组,这种方法具有迭代收敛快及存贮量少等优点．胡家赣等在系数矩阵为对称正定矩阵和对角优势L-矩阵的情况下证明了一次PE方法和一次PE_k方法的收敛性,指出了一次PE方法比     

Parallel algorithms for solving linear systems with block-tridiagonal matrices on multi-core CPU with GPU

For solving systems of linear algebraic equations with block-tridiagonal matrices arising in geoelectrics problems, the parallel matrix sweep algorithm, conjugate gradient method with preconditioner, and square root method are proposed and implemented numerically on multi-core CPU Intel with graphics processors NVIDIA. Investigation of efficiency and optimization of parallel algorithms for solving the problem with quasi-model data are performed.     

On the solution of boundary value problems with nonseparated multipoint and integral conditions

We suggest a numerical method for solving systems of linear nonautonomous ordinary differential equations with nonseparated multipoint and integral conditions. By using this method, which is based on the operation of convolution of integral conditions into local ones, one can reduce the solution of the original problem to the solution of a Cauchy problem for systems of ordinary differential equations and linear algebraic equations. We establish bounded linear growth of the error of the suggested numerical schemes. Numerical experiments were carried out for specially constructed test problems.     

On the numerical solution of loaded systems of ordinary differential equations with nonseparated multipoint and integral conditions

We propose a numerical method of solving systems of loaded linear nonautonomous ordinary differential equations with nonseparated multipoint and integral conditions. This method is based on the convolution of integral conditions to obtain local conditions. This approach allows one to reduce solving the original problem to solving a Cauchy problem for a system of ordinary differential equations and linear algebraic equations. Numerous computational experiments on several test problems with the formulas and schemes proposed for the numerical solution have been carried out. The results of the experiments show that the approach is reasonably efficient.     

Efficient numerical algorithm for solving the gravimetry problem of finding a lateral density in a layer: Parallel implementation

The paper is devoted to developing the new time- and memory-efficient algorithm BiCGSTABmem for solving the inverse gravimetry problem of determination of a variable density in a layer using the gravitational data. The problem is in solving the linear Fredholm integral equation of the first kind. After discretization of the domain and approximation of the integral operator, this problem is reduced to solving a large system of linear algebraic equations. It is shown that the matrix of coefficients is the Toeplitz-block-Toeplitz one in the case of the horizontal layer. For calculating and storing the elements of this matrix, we construct an efficient method, which significantly reduces the required memory and time. For the case of the curvilinear layer, we construct a method for approximating the parts of the matrix by a Toeplitz-block-Toeplitz one. This allows us to exploit the same efficient method for storing and processing the coefficient matrix in the case of a curvilinear layer. To solve the system of linear equations, we constructed the parallel algorithm on the basis of the stabilized biconjugated gradient method with using the Toeplitz-block-Toeplitz structure of the matrix. We implemented the BiCGSTAB and BiCGSTABmem algorithms for the Uran cluster supercomputer using the hybrid MPI + OpenMP technology. A model problem with synthetic data was solved for a large grid. It was shown that the new BiCGSTABmem algorithm reduces the computation time in comparison with the BiCGSTAB. Scalability of the parallel algorithm was studied.     

Remarks on the optimal convolution kernel for CSOR waveform relaxation

The convolution SOR waveform relaxation method is a numerical method for solving large-scale systems of ordinary differential equations on parallel computers. It is similar in spirit to the SOR acceleration method for solving linear systems of algebraic equations, but replaces the multiplication with an overrelaxation parameter by a convolution with a time-dependent overrelaxation function. Its convergence depends strongly on the particular choice of this function. In this paper, an analytic expression is presented for the optimal continuous-time convolution kernel and its relation to the optimal kernel for the discrete-time iteration is derived. We investigate whether this analytic expression can be used in actual computations. Also, the validity of the formulae that are currently used to determine the optimal continuous-time and discrete-time kernels is extended towards a larger class of ODE systems.     

广义异步并行多分裂松弛算法

在本文中，我们设计了求解大型线性代数方程组的适用于ＭＩＭＤ系统的异步并行多分裂松弛算法的一般模型，并在系数矩阵是Ｈ－矩阵的条件下，建立了该一般模型的收敛性理论。     

A new collocation method for solution of mixed linear integro-differential-difference equations

Numerical solution of mixed linear integro-differential-difference equation is presented using Chebyshev collocation method. The aim of this article is to present an efficient numerical procedure for solving mixed linear integro-differential-difference equations. Our method depends mainly on a Chebyshev expansion approach. This method transforms mixed linear integro-differential-difference equations and the given conditions into matrix equation which corresponds to a system of linear algebraic equation. The reliability and efficiency of the proposed scheme are demonstrated by some numerical experiments and performed on the computer algebraic system Maple10.