并行PCG算法在电法勘探中的应用研究

采用有限元法进行电法勘探时，会产生大型稀疏线性方程组，如何提高方程组的求解效率成为物探研究的关键。针对传统直接法难以实现并行求解的缺点，提出了在Beowulf集群环境下，采用并行PCG算法求解物探系统线性方程组。在集群环境下，该算法具有机器间相互通讯少、时间复杂度低等优点，并且易于并行实现。实验结果表明，采用PCG算法获得了良好的并行效果。     

并行PCG算法在电法勘探中的应用研究

采用有限元法进行电法勘探时,会产生大型稀疏线性方程组,如何提高方程组的求解效率成为物探研究的关键。针对传统直接法难以实现并行求解的缺点,提出了在Beowulf集群环境下,采用并行PCG算法求解物探系统线性方程组。在集群环境下,该算法具有机器间相互通讯少、时间复杂度低等优点,并且易于并行实现。实验结果表明,采用PCG算法获得了良好的并行效果。     

并行PCG算法在电法勘探中的应用研究

采用有限元法进行电法勘探时,会产生大型稀疏线性方程组,如何提高方程组的求解效率成为物探研究的关键。针对传统直接法难以实现并行求解的缺点,提出了在Beowulf集群环境下,采用并行PCG算法求解物探系统线性方程组。在集群环境下,该算法具有机器间相互通讯少、时间复杂度低等优点,并且易于并行实现。实验结果表明,采用PCG算法获得了良好的并行效果。     

Cobra: Parallel path following for computing the matrix pseudospectrum

The construction of an accurate approximation of the ε-pseudospectrum of a matrix by means of the standard grid method is a very demanding computational task. In this paper, we describe Cobra, a domain-based method for the computation of pseudospectra that combines predictor corrector path following with a one-dimensional grid. The algorithm offers large and medium grain parallelism and becomes particularly attractive when we seek fine resolution of the pseudospectrum boundary. We implement Cobra using standard LAPACK components and show that it is more robust than the existing path following technique and faster than it and the traditional grid method. Cobra is also combined with a partial SVD algorithm to produce an effective parallel method for computing the matrix pseudospectrum.     

New PCG based algorithms for the solution of Hermitian Toeplitz systems

In order to solve Toeplitz linear systems A_n(f)x=b generated by a nonnegative integrable function f, through use of the preconditioned conjugate gradient (PCG) method, several authors have proposed A_n(g) as preconditioner in the case where g is a trigonometric polynomial [10, 14, 27, 12, 28]. In preceding works, we studied the distribution and the extremal properties of the spectrum of the preconditioned matrix G=A _n ⁻¹ (g) A_n(f). In this paper we prove that the union of the spectra of all the G_n is dense on the essential range of f/g, i.e.,ER(f/g) and we obtain asymptotic information about the rate of convergence of the smallest eigenvalue λ _l ⁿ of G_n to r (and of λ _n ⁿ to R). As a consequence of this second order result, it is possible to handle the case where f has zeros of any order θ, through the PCG methods proposed in [10, 14]. This is a noteworthy extension since the techniques developed in [10, 14, 27, 12, 28] are shown to be effective only when f has zeros of even orders. The cost of this procedure is O(n^1+c(θ) log n) arithmetic operations (ops) where the quantity c(θ) belongs to interval [0,2⁻¹] and takes the maximum value 2⁻¹ when f has a zero of odd order. Finally, for the special case of zeros of odd orders, we propose a further algorithm which makes use of the PCG techniques proposed in [10, 14, 27, 12, 28] for theeven order case, reducing the cost to O(n long n) ops.     

A parallel preconditioned conjugate gradient solution method for finite element problems with coarse–fine mesh formulation

The object of this paper is a parallel preconditioned conjugate gradient iterative solver for finite element problems with coarse-mesh/fine-mesh formulation. An efficient preconditioner is easily derived from the multigrid stiffness matrix. The method has been implemented, for the sake of comparison, both on a IBM-RISC590 and on a Quadrics-QH1, a massive parallel SIMD machine with 128 processors. Examples of solutions of simple linear elastic problems on rectangular grids are presented and convergence and parallel performance are discussed.     

An execution time and energy model for an energy-aware execution of a conjugate gradient method with CPU/GPU collaboration

The parallel preconditioned conjugate gradient method (CGM) is used in many applications of scientific computing and often has a critical impact on their performance and energy consumption. This article investigates the energy-aware execution of the CGM on multi-core CPUs and GPUs used in an adaptive FEM. Based on experiments, an application-specific execution time and energy model is developed. The model considers the execution speed of the CPU and the GPU, their electrical power, voltage and frequency scaling, the energy consumption of the memory as well as the time and energy needed for transferring the data between main memory and GPU memory. The model makes it possible to predict how to distribute the data to the processing units for achieving the most energy efficient execution: the execution might deploy the CPU only, the GPU only or both simultaneously using a dynamic and adaptive collaboration scheme. The dynamic collaboration enables an execution minimising the execution time. By measuring execution times for every FEM iteration, the data distribution is adapted automatically to changing properties, e.g. the data sizes.     

多媒体相册中照片修补算法的研究与实现

照片的破损和丢失是照片数字化后常见的现象。用以AR模型为基础的预示算子进行修复时,边缘连续性差,最小最大函数插补法可以克服这一问题。论文介绍了一种利用最小最大函数插补法,并引入了共轭梯度法的照片修补算法,达到了较好的效果。     

计算对称带状矩阵特征值问题的并行二分／多分法

文中提出了在分布式环境下并行求解对称带状矩阵特征值问题的并行二分．多分法及其改进,该算法利用变形高斯消去法计算对称带状矩阵的Sturm序列,并利用Rayleigh商迭代对二分／多分法加以改进,在算法的并行执行过程中,各处理机间不需通信,特别适用在分布式环境下的并行计算,最后给出了数值实验结果。     

A new subspace minimization conjugate gradient method based on tensor model for unconstrained optimization

ABSTRACT

A new subspace minimization conjugate gradient method based on tensor model is proposed and analysed. If the objective function is close to a quadratic, we construct a quadratic approximation model in a two-dimensional subspace to generate the search direction; otherwise, we construct a tensor model. It is remarkable that the search direction satisfies the sufficient descent property. We prove the global convergence of the proposed method under mild assumptions. Numerical comparisons are given with well-known CGOPT and CG_DESCENT and show that the proposed algorithm is very promising.     

A projection method for solving nonsymmetric linear systems on multiprocessors

We consider the iterative solution of large sparse linear systems of equations arising from elliptic and parabolic partial differential equations in two or three space dimensions. Specifically, we focus our attention on nonsymmetric systems of equations whose eigenvalues lie on both sides of the imaginary axis, or whose symmetric part is not positive definite. This system of equation is solved using a block Kaczmarz projection method with conjugate gradient acceleration. The algorithm has been designed with special emphasis on its suitability for multiprocessors. In the first part of the paper, we study the numerical properties of the algorithm and compare its performance with other algorithms such as the conjugate gradient method on the normal equations, and conjugate gradient-like schemes such as ORTHOMIN(k), GCR(k) and GMRES(k). We also study the effect of using various preconditioners with these methods. In the second part of the paper, we describe the implementation of our algorithm on the CRAY X-MP/48 multiprocessor, and study its behavior as the number of processors is increased.     

Dirichlet feedback control for the stabilization of the wave equation: a numerical approach

In (J. Differential Equations 66 (1987) 340) a uniform stabilization method of the wave equation by boundary control à la Dirichlet has been discussed. In this article, we investigate the numerical implementation of the above stabilization process by a numerical scheme which mimics the energy decay properties of its continuous counterpart. The practical implementation of that scheme leads to a biharmonic problem of a new type which is solved by a method directly inspired by some related work of Glowinski and Pironneau on the solution of the Dirichlet problem for the biharmonic operator (SIAM Rev. 21(2) (1979) 167). Numerical experiments show that the decay properties of the energy are well-preserved by our numerical methodology.     

Parallel implementation of bisection for the calculation of eigenvalues of tridiagonal symmetric matrices

Kuck and Sameh [9], Huang [8], and Wallach [10] have investigated parallel implementations of Givens' bisection algorithm [4, 5]. On a MIMD (multiple instruction stream — multiple data stream) machine one could apply parallelism on any or all of three levels: within each Sturm sequence calculation, within each individual bisection, or on the outer level assigning intervals to be searched. We show that allocating bisections to individual processors on the outer level is more effective than having processors share the work of a bisection when the number of processors is smaller than the number of eigenvalues to be found.     

A parallel partition method for solving banded systems of linear equations

The partition method of Wang for tridiagonal equations is generalized to the arbitrary band case. A stability criterion is given. The algorithm is compared to Gaussian elimination and cyclic reduction.     

A method for calculating successive approximate solutions for a class of block banded M/G/1 type Markovian models

This paper presents an efficient equilibrium solution algorithm for a class of infinite block banded M/G/1 type Markov chains. By re-blocking the states, these are a class of the so-called quasi-birth-and-death (QBD) type chains. The proposed algorithm is not based on an iterative approach, so that the exact solution can be computed in a known finite number of steps. The key point on which the algorithm is based is the identification of a linear dependence among variables. This dependence is expressed in terms of a companion matrix. The equilibrium solution of the Markov chain is obtained operating on this matrix.

An attractive feature of the algorithm is that it allows the computation of a succession of approximate solutions with growing accuracy, until the exact solution is obtained in a finite number of steps. The class of block-banded M/G/1 type Markov chains we consider requires that the lower diagonal block is invertible and that the chain is ergodic. However, many models arising from telecommunication systems satisfy this restriction. Results for a case study show that the proposed algorithm is efficient and quite accurate, even when providing approximate solutions.     

Parallel machine earliness/tardiness scheduling problem under the effects of position based learning and linear/nonlinear deterioration

This paper considers a parallel machine earliness/tardiness (ET) scheduling problem with different penalties under the effects of position based learning and linear and nonlinear deterioration. The problem has common due-date for all jobs, and effects of learning and deterioration are considered simultaneously. By the effects of learning we mean that the job processing time decreases along the sequence of partly similar jobs, and by the effects of deterioration we mean slowing performance or time increases along the sequence of jobs. This study shows that optimal solution for ET scheduling problem under effects of learning and deterioration is V-shape schedule under certain agreeable conditions. Furthermore, we design a mathematical model for the problem under study and algorithm and lower bound procedure to solve larger test problems. The algorithm can solve problems of 1000 jobs and four machines within 3 s on average. The performance of the algorithm is evaluated using results of the mathematical model.     

Distributed Lagrange multipliers based on fictitious domain method for second order elliptic problems

In this article we further investigate the solution of linear second order elliptic boundary value problems by distributed Lagrange multipliers based fictitious domain methods. The following issues are addressed: (i) Derivation of the fictitious domain formulations. (ii) Finite element approximation. (iii) Iterative solution of the resulting finite dimensional problems (of the saddle-point type) by preconditioned conjugate gradient and Lanczos algorithms.     

A near-exact method for solving the log-truck scheduling problem

We propose a solution method for the log‐truck scheduling problem, which is a generalisation of the pick‐up and delivery problem with time windows. Our approach is based on column generation and pseudo branch and price. Each column in the proposed mathematical model represents one feasible route for one truck. We start by designing a priori an initial set of routes. Then, the subproblem, which is a constrained shortest path problem, is solved by applying a k‐shortest path algorithm. Numerical results from a case study are presented.     

A hybrid strategy for the time- and energy-efficient trajectory planning of parallel platform manipulators

In planning the trajectories of motor-driven parallel platform manipulators, the objective is to identify the trajectory which accomplishes the assigned motion with the minimal travel time and energy expenditure subject to the constraints imposed by the kinematics and dynamics of the manipulator structure. In this study, the possible trajectories of the manipulator are modeled using a parametric path representation, and the optimal trajectory is then obtained using a hybrid scheme comprising the particle swarm optimization method and the local conjugate gradient method. The numerical results confirm the feasibility of the optimized trajectories and show that the hybrid scheme is not only more computationally efficient than the standalone particle swarm optimization method, but also yields solutions of a higher quality.