微机网络并行环境下杆壳组合结构动力特性并行计算

回顾了有限元并行计算发展的历史,阐述了微机网络并行计算环境的意义,给出了基于微机网络并行环境的杆壳组合结构动力分析并行算法,该算法包括杆壳组合结构总刚度矩阵和总质量矩阵的并行计算以及求解广义特征值问题的并行子空间迭代法的并行计算,在多台微机上安装PVM．使用Linux操作系统．构成分布式微机网络并行计算环境,将上述算法用于某型号飞机机翼及某型号挂架动力特性的并行计算,在该并行环境下的教值试验表明所给算法是非常有效的。     

话说超级计算

简介了超级计算(大规模并行计算)的基本概念及软硬件体系,结合笔者经历回顾了近20余年我国超级计算的发展.简要介绍了异构并行计算的基本概念以及并行计算的基本编程方式.     

有限元并行计算的预处理

首先指出了有限元并行计算预处理的重要性和影响有限元并行计算效率的关键因素，然后介绍了两类有限元并行计算的任务划分方法，并在此基础上提出了一种改进的划分方法，经过验证这种方法是有效的。最后基于网络机群并行环境和消息传递编程模式，给出了一种能够降低通信成本的消息传递次序调整方法，并对调整前后的通信成本进行了比较。     

三维Euler方程的分区和并行计算

三维全机绕流区域分解成多块子区域,多块区域之间采用迎风型通量守恒内边界耦合条件,分区计算总体区域,形成总体耦合流场的分区数值解。利用PVM并行环境,采用纯结点并行计算编程方式和“先进先出”的同步控制等待机制,对三维复杂流动跨音速流场相应分区实现了多区域并行计算。分析了影响并行效率的主要因素,将并行计算结果与串行计算结果和实验结果作了比较,讨论了多种区域分解数目的并行计算效率。在负载平衡程度较好时,可得到较高的并行效率。     

基于并行计算和遗传算法的钢-UHPC华夫板组合梁优化设计

为实现钢-超高性能混凝土(UHPC)华夫板组合梁结构快速经济合理的设计,提出了基于并行计算与遗传算法的结构优化设计方法。通过Python建立了并行计算平台,使Abaqus和Python能够执行同步数值模拟和数据处理,以成本最小化为目标,采用遗传算法对钢-UHPC华夫板组合梁进行了优化,验证了所提方法的可行性。结果表明,遗传算法中密集的分析任务可以并行化并分配给不同的计算资源以提高计算效率;使用并行计算可以提高8.6倍的优化效率;并行计算和串行计算的CPU平均使用率分别为82%和18%。本文方法的成功应用可为其他类型结构的优化设计提供参考。     

无单元伽辽金法的并行计算

对无单元伽辽金法的并行计算进行了详细研究,并将其应用于弹性动力学问题。使用并行桶搜索算法进行节点搜索,使用并行几何搜索算法进行样点搜索,讨论了移动最小二乘MLS(Moving Least Squares)形函数及其导数的并行计算和方程组的并行求解,并利用多层图形划分实现负载平衡。最后给出了并行无单元伽辽金法应用于弹性动力学的计算流程和实例。计算结果表明无单元伽辽金法具有很高的并行性和很好的并行效率,对其进行并行计算具有非常重要的意义。     

闪光X射线辐射照相蒙特卡罗程序并行设计

研究了闪光X射线辐射照相蒙特卡罗程序在MPI平台下的并行计算实现,给出了实现过程中并行随机数的产生方法。通过算例,采用多CPU并行计算,可以成比例地提高加速比和计算效率。     

基于GPU并行计算的浅水波运动数值模拟

利用有限体积法求解描述水流运动的二维浅水方程组,模拟洪水波运动传播过程,并通过GPU并行计算技术对程序进行加速,建立了浅水运动高效模拟方法。数值模拟结果表明,基于本文提出的GPU并行策略以及通用并行计算架构(CUDA)支持,能够实现相比CPU单核心最高112倍的加速比,为利用单机实现快速洪水预测以及防灾减灾决策提供有效支撑。此外,对基于GPU并行计算的浅水模拟计算精度进行了论证,并对并行性能优化进行了分析。利用所建模型模拟了溃坝洪水在三维障碍物间的传播过程。     

基于PVM的网络并行有限元面向对象的设计

子结构是有限元并行计算常用的一种方法,本文采用面向对象的方法,首先对子结构进行了面向对象的设计,得到了其类层次结构图;然后针对工作站网络有限元并行计算环境。提出了基于PVM消息传递平台上的Shadow—Mirror数据传输模型,该模型在有限元并行计算数据传输时,充分发挥数据面向对象的特性,采用设置数据缓冲区、短消息合并等方法以缩短数据通信时间,并据此编制了相应的程序。计算结果表明,使用文中提出的面向对象的Shadow—Mirror数据传输模型可以得到较为理想的并行加速比,而且随着问题规模增大,并行加速比增高。本文研究内容为进一步开展基于工作站网络的并行有限元研究提供了一个可参考的基础。     

应用基于GPU的SPH方法模拟二维楔形体入水砰击问题

光滑粒子流体动力学(SPH)法是一种无网格的拉格朗日效值方法,广泛应用于计算流体领域模拟复杂自由表面流问题.SPH方法的主要缺点就是计算量过大,而基于GPU的并行计算方法可使SPH计算得到有效加速.本文应用基于GPU的SPH并行计算方法研究了二维楔形体的入水砰击问题.数值计算结果与文献中对应的解析解比较一致,验证了基于GPU的SPH方法的精度和可靠性.仿真结果同时显示基于GPU的并行计算方法可使SPH计算速度得到显著提高.     

A CLASS OF ALTERNATING GROUP METHOD OF BURGERS' EQUATION

IntroductionWeconsiderthefollowingnonlinearBurgers’equation : u t u u x=ε 2 u x2 ,　　 0 0 ,( 1 )withtheinitialandtheboundaryconditions　　　　　u(x,0 ) =f(x) ,　　 0 相似文献   

A CLASS OF ALTERNATING GROUP METHODOF BURGERS＇ EQUATION

Some new Saul‘ yev type asymmetric difference schemes for Burgers’ equation isgiven, by the use of the schemes, a kind of alternating group four points method for solvingnonlinear Burgers‘ equation is constructed here. The basic idea of the method is that thegrid points on the same time level is divided into a number of groups, the differenceequations of each group can be solved independently, hence the method with intrinsicparallelism can be used directly on parallel computer. The method is unconditionally stableby analysis of linearization procedure. The numerical experiments show that the method has good stability and accuracy.     

一类非线性发展方程的交替分段显隐并行数值方法

给出了一类非线性发展方程的交替分段显隐并行数值方法 ,得到了方法的无条件稳定性和并行性兼顾的结果。数值例子说明理论分析的正确性和格式的有效性     

地下水流并行有限层方法及同伦反演研究

根据有限层求解格式存在的解耦性,实现了地下水三维流问题的高效并行化计算。在此基础上,结合非线性同伦方法,提出了地下水参数反演分析的并行同伦算法,利用MATLAB编译了相应的正反演计算程序。与已有解析解和有限差分解的对比以及数值算例,验证了并行化正反演方法及程序的正确性,探讨了并行算法的计算效率。研究表明,并行方法可以有效提高计算速度,较串行方法具有明显优势,同时同伦反演方法具有大范围收敛的特点,不依赖于参数值的初始选取。     

Further improved F-expansion and new exact solutions for nonlinear evolution equations

The improved F-expansion method with a computerized symbolic computation is used to construct the exact traveling wave solutions of four nonlinear evolution equations in physics. As a result, many exact traveling wave solutions are obtained which include new soliton-like solutions, trigonometric function solutions, and rational solutions. The method is straightforward and concise, and it holds promise for many applications.     

New explicit and exact traveling wave solutions for two nonlinear evolution equations

In this paper, with the aid of computer symbolic computation system such as Maple, an algebraic method is firstly applied to two nonlinear evolution equations, namely, nonlinear Schrodinger equation and Pochhammer–Chree (PC) equation. As a consequence, some new types of exact traveling wave solutions are obtained, which include bell and kink profile solitary wave solutions, triangular periodic wave solutions, and singular solutions. The method is straightforward and concise, and it can also be applied to other nonlinear evolution equations in mathematical physics.     

Nonlinear damage model for seismic damage assessment of reinforced concrete frame members and structures

A nonlinear damage model based on the combination of deformation and hysteretic energy and its validation with experiments are presented. Also, a combination parameter is defined to consider the mutual effect of deformation and hysteretic energy for different types of components in different loading stages. Four reinforced concrete (RC) columns are simulated and analyzed using the nonlinear damage model. The results indicate that the damage evolution evaluated by the model agrees well with the experimental phenomenon. Furthermore, the seismic damage evolution of a six-story RC frame was analyzed, revealing four typical failure modes according to the interstory drift distribution of the structure; the damage values calculated using the nonlinear damage model agree well with the four typical failure modes.     

Alternating segment explicit-implicit scheme for nonlinear third-order KdV equation

A group of asymmetric difference schemes to approach the Korteweg-de Vries(KdV)equation is given here.According to such schemes,the full explicit difference scheme and the fun implicit one,an alternating segment explicit-implicit difference scheme for solving the KdV equation is constructed.The scheme is linear unconditionally stable by the analysis of linearization procedure,and is used directly on the parallel computer. The numerical experiments show that the method has high accuracy.     

Differential-algebraic approach to large deformation analysis of frame structures subjected to dynamic loads

A nonlinear mathematical model for the analysis of large deformation of frame structures with discontinuity conditions and initial displacements, subject to dynamic loads is formulated with arc-coordinates. The differential quadrature element method （DQEM） is then applied to discretize the nonlinear mathematical model in the spatial domain, An effective method is presented to deal with discontinuity conditions of multivariables in the application of DQEM. A set of DQEM discretization equations are obtained, which are a set of nonlinear differential-algebraic equations with singularity in the time domain. This paper also presents a method to solve nonlinear differential-algebra equations. As application, static and dynamical analyses of large deformation of frames and combined frame structures, subjected to concentrated and distributed forces, are presented. The obtained results are compared with those in the literatures. Numerical results show that the proposed method is general, and effective in dealing with disconti- nuity conditions of multi-variables and solving differential-algebraic equations. It requires only a small number of nodes and has low computation complexity with high precision and a good convergence property.     

Poincare-Lighthill-Kuo Method and Symbolic Computation

ＩｎｔｒｏｄｕｃｔｉｏｎＴｈｅＰＬＫｍｅｔｈｏｄｉｓａｎｅｆｆｅｃｔｉｖｅｐｅｒｔｕｒｂａｔｉｏｎｍｅｔｈｏｄａｎｄｈａｓａｈｉｓｔｏｒｙｏｆｍｏｒｅｔｈａｎ 10 0ｙｅａｒｓｉｎｉｔｓａｄｖｅｎｔａｎｄｄｅｖｅｌｏｐｍｅｎｔ (ｃｆ.Ｒｅｆｓ.[1- 6 ]) .Ｉｎｔｈｅ 80ｓｏｆｔｈｅｎｉｎｅｔｅｅｎｔｈｃｅｎｔｕｒｙ ,Ｐｏｉｎｃａｒｅ ,Ｌｉｎｄｓｔｅｄｔｅｔａｌ.ｐｒｏｐｏｓｅｄａｔｅｃｈｎｉｑｕｅｏｆｓｔｒａｉｎｅｄｐａｒａｍｅｔｅｒｓｉｎｔｈｅｉｒｓｔｕｄｙｏｆｃｅｌｅｓｔｉａｌｍｅｃｈａｎｉｃｓ.Ｔｈ…