A class of asynchronous parallel nonlinear accelerated overrelaxation methods for the nonlinear complementarity problems

In accordance with the principle of using sufficiently the delayed information, and by making use of the nonlinear multisplitting and the nonlinear relaxation techniques, we present in this paper a class of asynchronous parallel nonlinear multisplitting accelerated overrelaxation (AOR) methods for solving the large sparse nonlinear complementarity problems on the high-speed MIMD multiprocessor systems. These new methods, in particular, include the so-called asynchronous parallel nonlinear multisplitting AOR-Newton method, the asynchronous parallel nonlinear multisplitting AOR-chord method and the asynchronous parallel nonlinear multisplitting AOR-Steffensen method. Under suitable constraints on the nonlinear multisplitting and the relaxation parameters, we establish the local convergence theory of this class of new methods when the Jacobi matrix of the involved nonlinear mapping at the solution point of the nonlinear complementarity problem is an H-matrix.     

Analysis of two new parareal algorithms based on the Dirichlet-Neumann/Neumann-Neumann waveform relaxation method for the heat equation

The Dirichlet-Neumann and Neumann-Neumann waveform relaxation methods are nonoverlapping spatial domain decomposition methods to solve evolution problems, while the parareal algorithm is in time parallel fashion. Based on the combinations of these space and time parallel strategies, we present and analyze two parareal algorithms based on the Dirichlet-Neumann and the Neumann-Neumann waveform relaxation method for the heat equation by choosing Dirichlet-Neumann/Neumann-Neumann waveform relaxation as two new kinds of fine propagators instead of the classical fine propagator. Both new proposed algorithms could be viewed as a space-time parallel algorithm, which increases the parallelism both in space and in time. We derive for the heat equation the convergence results for both algorithms in one spatial dimension. We also illustrate our theoretical results with numerical experiments finally.

     

A parallel algorithm for large systems of Volterra integral equations of Abel type

A significative number of recent applications require numerical solution of large systems of Abel–Volterra integral equations. Here we propose a parallel algorithm to numerically solve a class of these systems, designed for a distributed-memory MIMD architecture. In order to achieve a good efficiency we employ a fully parallel and fast convergent waveform relaxation (WR) method and evaluate the lag term by using FFT techniques. To accelerate the convergence of the WR method and to best exploit the parallel architecture we develop special strategies. The performances of the resulting code, NSWR4, are illustrated on some examples.     

一种新的局部时间积分的区域分解波形松弛算法

提出一种新的区域分解波形松弛算法, 使得可以在不同的子域采用不同的时间步长来并行求解线性抛物方程的初边值问题. 与传统的区域分解波形松弛算法相比, 该算法可以通过预条件子来加快收敛速度, 并且对内存的需求大大降低. 给出了局部时间步长一种具体的实现方法, 证明了离散解的存在唯一性, 并在时间连续水平分析了预条件系统. 数值实验显示了新算法的有效性.     

Analysis of a new parareal algorithm based on waveform relaxation method for time-periodic problems

We present a new parallel algorithm for time-periodic problems by combining the waveform relaxation method and the parareal algorithm, which performs the parallelism both in sub-systems and in time. In the new algorithm, the waveform relaxation propagator is chosen as a new fine propagator instead of the classical fine propagator. And because of the characteristic of time-periodic problems, the new parareal waveform relaxation algorithm needs to solve a periodic coarse problem at the coarse level in each iteration. The new algorithm is proved to converge linearly at most. Then the theoretic parallel efficiency of the new algorithm is also considered. Numerical experiments confirm our analysis finally.     

基于自适应遗传算法的逐次超松驰迭代法

确定逐次超松驰迭代法中的最佳松驰因子,迄今,人们还没有给出一可行实用的方法.利用自适应遗传算法全局搜索性能、并行性及其遗传操作,构造出近似确定最佳松驰因子的一种自适应进化方法,并由此得到一近似确定ω功能的自适应逐次超松驰迭代算法.数值算例表明,该算法在求解线性方程组中是可行的,实用和快捷的.     

Overlapping Schwarz waveform relaxation method for the solution of the reaction-diffusion equation

We are interested in solving time dependent problems using domain decomposition methods. In the classical approach, one discretizes first the time dimension and then one solves a sequence of steady problems by a domain decomposition method. In this article, we treat directly the time dependent problem and we study a Schwarz waveform relaxation algorithm for the convection diffusion equation. We study the convergence of the overlapping Schwarz waveform relaxation method for solving the reaction-diffusion equation over multi-overlapped subdomains. Also we will show that the method converges linearly and superlinearly over long and short time intervals, and the convergence depends on the size of overlap. Numerical results are presented from solutions of a specific model problems to demonstrate the convergence, linear and superlinear, and the role of the overlap size.     

The Waveform Relaxation method for systems of differential/algebraic equations

This paper reports efforts towards establishing a parallel numerical algorithm known as Waveform Relaxation (WR) for simulating large systems of differential/algebraic equations. The WR algorithm was established as a relaxation based iterative method for the numerical integration of systems of ODEs over a finite time interval. In the WR approach, the system is broken into subsystems which are solved independently, with each subsystem using the previous iterate waveform as “guesses” about the behavior of the state variables in other subsystems. Waveforms are then exchanged between subsystems, and the subsystems are then resolved repeatedly with this improved information about the other subsystems until convergence is achieved.

In this paper, a WR algorithm is introduced for the simulation of generalized high-index DAE systems. As with ODEs, DAE systems often exhibit a multirate behavior in which the states vary as differing speeds. This can be exploited by partitioning the system into subsystems as in the WR for ODEs. One additional benefit of partitioning the DAE system into subsystems is that some of the resulting subsystems may be of lower index and, therefore, do not suffer from the numerical complications that high-index systems do. These lower index subsystems may therefore be solved by less specialized simulations. This increases the efficiency of the simulation since only a portion of the problem must be solved with specially tailored code. In addition, this paper established solvability requirements and convergence theorems for varying index DAE systems for WR simulation.     

A note on the H 1-convergence of the overlapping Schwarz waveform relaxation method for the heat equation

The overlapping Schwarz waveform relaxation method is a parallel iterative method for solving time-dependent PDEs. Convergence of the method for the linear heat equation has been studied under infinity norm but it was unknown under the energy norm at the continuous level. The question is interesting for applications concerning fluxes or gradients of the solutions. In this work, we show that the energy norm of the errors of iterates is bounded by their infinity norm. Therefore, we give an affirmative answer to this question for the first time.     

Convergence analysis of the parallel multisplitting TOR method

In this paper, we propose the parallel multisplitting TOR method, for solving a large nonsingular systems of linear equations Ax = b. These new methods are a generalization and an improvement of the relaxed parallel multisplitting method (Formmer and Mager, 1989) and parallel multisplitting AOR Algorithm (Wang Deren, 1991). The convergence theorem of this new algorithm is established under the condition that the coefficient matrix A of linear systems is an H-matrix. Some results also yield new convergence theorem for TOR method.     

An Optimization Problem with a Separable Non-Convex Objective Function and a Linear Constraint

For a class of global optimization (maximization) problems, with a separable non-concave objective function and a linear constraint a computationally efficient heuristic has been developed.The concave relaxation of a global optimization problem is introduced. An algorithm for solving this problem to optimality is presented. The optimal solution of the relaxation problem is shown to provide an upper bound for the optimal value of the objective function of the original global optimization problem. An easily checked sufficient optimality condition is formulated under which the optimal solution of concave relaxation problem is optimal for the corresponding non-concave problem. An heuristic algorithm for solving the considered global optimization problem is developed.The considered global optimization problem models a wide class of optimal distribution of a unidimensional resource over subsystems to provide maximum total output in a multicomponent systems.In the presented computational experiments the developed heuristic algorithm generated solutions, which either met optimality conditions or had objective function values with a negligible deviation from optimality (less than 1/10 of a percent over entire range of problems tested).     

An inexact parallel splitting augmented Lagrangian method for large system of linear equations

Parallel iterative methods are powerful in solving large systems of linear equations (LEs). The existing parallel computing research results focus mainly on sparse systems or others with particular structure. Most are based on parallel implementation of the classical relaxation methods such as Gauss-Seidel, SOR, and AOR methods which can be efficiently carried out on multiprocessor system. In this paper, we propose a novel parallel splitting operator method in which we divide the coefficient matrix into two or three parts. Then we convert the original problem (LEs) into a monotone (linear) variational inequality problem (VI) with separable structure. Finally, an inexact parallel splitting augmented Lagrangian method is proposed to solve the variational inequality problem (VI). To avoid dealing with the matrix inverse operator, we introduce proper inexact terms in subproblems such that the complexity of each iteration of the proposed method is O(n²). In addition, the proposed method does not require any special structure of system of LEs under consideration. Convergence of the proposed methods in dealing with two and three separable operators respectively, is proved. Numerical computations are provided to show the applicability and robustness of the proposed methods.     

Waveform relaxation methods for implicit differential equations

We apply a Runge-Kutta-based waveform relaxation method to initial-value problems for implicit differential equations. In the implementation of such methods, a sequence of nonlinear systems has to be solved iteratively in each step of the integration process. The size of these systems increases linearly with the number of stages of the underlying Runge-Kutta method, resulting in high linear algebra costs in the iterative process for high-order Runge-Kutta methods. In our earlier investigations of iterative solvers for implicit initial-value problems, we designed an iteration method in which the linear algebra costs are almost independent of the number of stages when implemented on a parallel computer system. In this paper, we use this parallel iteration process in the Runge-Kutta waveform relaxation method. In particular, we analyse the convergence of the method. The theoretical results are illustrated by a few numerical examples.     

Parallel multisplitting two-stage iterative methods for large sparse systems of weakly nonlinear equations

The finite difference or the finite element discretizations of many differential or integral equations often result in a class of systems of weakly nonlinear equations. In this paper, by reasonably applying both the multisplitting and the two-stage iteration techniques, and in accordance with the special properties of this system of weakly nonlinear equations, we first propose a general multisplitting two-stage iteration method through the two-stage multiple splittings of the system matrix. Then, by applying the accelerated overrelaxation (AOR) technique of the linear iterative methods, we present a multisplitting two-stage AOR method, which particularly uses the AOR-like iteration as inner iteration and is substantially a relaxed variant of the afore-presented method. These two methods have a forceful parallel computing function and are much more suitable to the high-speed multiprocessor systems. For these two classes of methods, we establish their local convergence theories, and precisely estimate their asymptotic convergence factors under some suitable assumptions when the involved nonlinear mapping is only directionally differentiable. When the system matrix is either an H-matrix or a monotone matrix, and the nonlinear mapping is a P-bounded mapping, we thoroughly set up the global convergence theories of these new methods. Moreover, under the assumptions that the system matrix is monotone and the nonlinear mapping is isotone, we discuss the monotone convergence properties of the new multisplitting two-stage iteration methods, and investigate the influence of the multiple splittings as well as the relaxation parameters upon the convergence behaviours of these methods. Numerical computations show that our new methods are feasible and efficient for parallel solving of the system of weakly nonlinear equations. This revised version was published online in August 2006 with corrections to the Cover Date.     

ON MONOTONE CONVERGENCE OF NONLINEARMULTISPLITTING RELAXATION METHODS

ＯＮＭＯＮＯＴＯＮＥＣＯＮＶＥＲＧＥＮＣＥＯＦＮＯＮＬＩＮＥＡＲＭＵＬＴＩＳＰＬＩＴＴＩＮＧＲＥＬＡＸＡＴＩＯＮＭＥＴＨＯＤＳ￥ＷＡＮＧＤＥＲＥＮ；ＢＡＩＺＨＯＮＧＺＨＩ（ＤｅｐａｚａｔｍｅｎｔｏｆＭａｔｈｅｍａｔｉｃｓ，ＳｈｓｌｌｇｈａｉＵｎｉｖｅ...     

关于多分裂方法收敛性的一些结果

1引言 考虑在并行计算机上解大型线性方程组AX=b假定有K台处理机可供使用，并且对于局部数据，处理机能执行不同的指令序列，毗邻的处理机之间能自然地通讯。     

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.     

手数约束和凹交易费下的离散投资组合模型及算法

本文建立带手数约束和凹交易费的离散投资组合模型,给出求解该模型的一种精确算法。该算法是一个基于拉格朗日松弛和次梯度对偶搜索的分枝定界算法。为测试算法的有效性,用随机产生的数据对模型进行数值实验。作为其应用,用沪深300指数的真实数据实证检验该模型,并与不含交易费用的离散投资组合模型进行数值比较分析。数值分析表明算法能在合理的时间内给出模型的投资组合策略, 对解决中小规模的离散投资组合问题是有效的。     

Convergence of the parallel chaotic waveform relaxation method for stiff systems

In this paper, we establish several algorithms for parallel chaotic waveform relaxation methods for solving linear ordinary differential systems based on some given models. Under some different assumptions on the coefficient matrix A and its multisplittings we obtain corresponding sufficient conditions of convergence of the algorithms. Also a discussion on convergence speed comparison of synchronous and asynchronous algorithms is given.     

一种迭代格式的有限元并行算法*

本文提出了一种求解有限元方程的迭代格式的并行算法.该方法在线性代数方程迭代解法的基础上,引进并行运算步骤;并且运用加权残数方法,通过选择适当的权函数,推导了该并行算法的有限元基本格式.该方法在西安交通大学BLXSI-6400并行计算机上程序实现.计算结果表明它能有效地提高运算速度,减少计算时间,是一种有效的求解大型结构有限元方程的并行算法.