块三对角线性方程组的一类二维区域分解并行不完全分解预条件

基于二维重叠区域分解,对每个子区域上局部不完全LU分解所得到的上、下三角因子分别进行组合,给出一类全局并行不完全分解型预条件.所给出的并行化方法适用于任何不完全LU分解型预条件.对采用二维区域分解与一维区域分解时所得并行预条件的并行计算性能进行分析比较.实验结果表明,提出的并行化方法普遍优于加性Schwarz并行化方法,且当处理器个数相对较多时采用二维区域分解优于一维区域分解.     

带阻尼项定常Navier-Stokes方程的并行两水平有限元算法

基于两重网格离散和区域分解技术，提出数值求解带阻尼项定常Navier-Stokes方程的三种并行两水平有限元算法。其基本思想是首先在粗网格上求解完全的非线性问题，以获得粗网格解，然后在重叠的局部细网格子区域上并行求解Stokes、 Oseen和Newton线性化的残差问题，最后在非重叠的局部细网格子区域上校正近似解。数值算例验证了算法的有效性。     

地球外核热对流运动的区域分解多重网格并行数值模拟

旋转球层中热对流运动的数值模拟是地球发电机模型的重要组成部分,对研究地球发电机作用机理具有重要意义.本文设计一个基于国产超级计算平台并行性能良好的地球外核热对流运动并行数值模型.时间积分方案采用与Crank-Nicolson格式和二阶Adams-Bashford公式相结合的近似分解分步法,空间离散基于立方球网格的二阶精度有限体积格式.所得到的两个大规模稀疏线性代数方程组采用带预处理的Krylov子空间迭代法进行求解.为加速迭代求解过程及提高并行性能,迭代过程采用区域分解多重网格的多层限制型加法Schwarz预处理子,减少了求解程序的计算时间,提高了数值模型的并行性能,模型被很好地扩展到上万处理器核数.数值模拟结果与基准模型算例0的参考值吻合得很好.     

二维抛物型问题的Strang型交替分段区域分裂格式

给出求解二维抛物型方程的Strang型的交替分段区域分裂格式。交替分段思想可以将区域分为一些不重叠的子区域，Strang型算子分裂技巧通过将高维问题的求解分解为几个低维问题的求解来降低其求解的复杂度。方法是无条件稳定的，理论分析了截断误差。数值算例说明格式的有效性及时空的二阶精度.     

二维抛物型问题的Strang型交替分段区域分裂格式（英文）

给出求解二维抛物型方程的Strang型的交替分段区域分裂格式。交替分段思想可以将区域分为一些不重叠的子区域,Strang型算子分裂技巧通过将高维问题的求解分解为几个低维问题的求解来降低其求解的复杂度。方法是无条件稳定的,理论分析了截断误差。数值算例说明格式的有效性及时空的二阶精度.     

基于界面修正的迭代并行计算方法

构造基于界面修正的迭代并行方法的一般途径是：将物理空间区域剖分成若干不重叠的块；在分块子区域的内边界上，采用某种显式格式计算出界面值作为预估值；然后采用某种隐式格式并行求解各个子块区域上的解，这里的隐式格式通常需要进行迭代求解(称为内迭代)；可在每一迭代步或几次迭代步结束时，利用已计算出的分块子区域内的(近似)解，在分块子区域内边界处利用隐式格式计算出在内边界处的校正值；随后再转入各个子块区域上的求解，该过程称为外迭代。与以往的并行差分格式不同，在求解的子区域上的定解问题时，可以仅仅在第一个(初始)迭代步求解时所需边界条件使用子区域内界面处的某种显式格式的解，在随后的迭代步中即可改用子区域内界面处的隐式修正格式的解。由此，至少可区分如下3类性质不完全相同的迭代并行格式。     

块三对角线性方程组不完全分解预条件的一种一维区域分解并行化方法

对块三对角线性方程组,不完全分解是最有效的预条件之一,但它本质上是一个串行计算过程,难以有效并行化.基于一维重叠区域分解,对局部不完全分解得到的上、下三角因子分别各自进行组合,构造一类全局的并行不完全分解型预条件.在具体实现时,给出两种具体途径,其中一种基于所有重叠部分对应分量的交换.之后,在仔细对其中的计算过程进行分析的基础上,给出一种只需要一条网格线上分量通信的实现算法,大大减少了通信量,且通信不随重叠度的增加而增加.这种并行化方法可以应用于块三对角线性方程组的任何不完全分解型预条件.实验结果表明,文中提出的并行化方法普遍优于加性Schwarz并行化方法.     

非定常Navier-Stokes方程基于两重网格离散的有限元并行算法

基于两重网格离散和区域分解技巧，提出三种求解非定常Navier-Stokes方程的有限元并行算法.算法的基本思想是在每一时间迭代步，在粗网格上采用Oseen迭代法求解非线性问题，在细网格上分别并行求解Oseen、Newton、Stokes线性问题以校正粗网格解.对于空间变量采用有限元离散，时间变量采用向后Euler格式离散.数值实验验证了算法的有效性.     

粒子输运离散纵标方程基于界面修正的并行计算方法

为了改造粒子输运方程求解的隐式格式,研究设计适应大型并行计算机的并行计算方法,介绍一类求解粒子输运方程离散纵标方程组的基于界面修正的源迭代并行计算方法.应用空间区域分解,在子区域内界面处首先采用迎风显式差分格式进行预估,构造子区域的入射边界条件,然后,在各个子区域内部进行源迭代求解隐式离散纵标方程组.在源迭代过程中,在内界面入射边界处采用隐式格式进行界面修正.数值算例表明该并行计算方法在精度、并行度、简单性诸方面均具有良好的性质.     

求解Boltzmann模型方程的气体运动论大规模并行算法

研究各流域三维流动问题的Boltzmann模型方程计算方法,建立直接求解分子速度分布函数的气体运动论耦合迭代数值格式;基于变量依赖关系、数据通信与并行可扩展性分析,使用区域分解并行化方法,建立气体运动论数值算法并行方案,发展求解各流域三维绕流问题的气体运动论并行算法.拟定高低不同马赫数下来自不同流域的三维球体及返回舱绕流算例,进行高性能Fortran(HPF)大规模并行计算,将计算结果与有关实验数据、相关理论预测等进行比较分析,研究揭示不同流区复杂绕流现象及流动机理.研究表明,所发展的气体运动论并行算法具有很好的并行独立性,基本达到线性加速的并行效果,显示出良好的并行可扩展性.     

Parallel multilevel methods for implicit solution of shallow water equations with nonsmooth topography on the cubed-sphere

High resolution and scalable parallel algorithms for the shallow water equations on the sphere are very important for modeling the global climate. In this paper, we introduce and study some highly scalable multilevel domain decomposition methods for the fully implicit solution of the nonlinear shallow water equations discretized with a second-order well-balanced finite volume method on the cubed-sphere. With the fully implicit approach, the time step size is no longer limited by the stability condition, and with the multilevel preconditioners, good scalabilities are obtained on computers with a large number of processors. The investigation focuses on the use of semismooth inexact Newton method for the case with nonsmooth topography and the use of two- and three-level overlapping Schwarz methods with different order of discretizations for the preconditioning of the Jacobian systems. We test the proposed algorithm for several benchmark cases and show numerically that this approach converges well with smooth and nonsmooth bottom topography, and scales perfectly in terms of the strong scalability and reasonably well in terms of the weak scalability on machines with thousands and tens of thousands of processors.     

二维柱几何中子输运方程的并行区域分解方法

分析不同的区域分解方法及优先级插入算法对二维柱几何下中子输运方程S_n间断有限元方程并行效率的影响,给出基于最小面体比的正方形区域分解方法及沿径向的优先级插入算法,并通过将正方形区域分解方法与径向优先级插入算法进行组合,形成新的算法.新算法更适应于二维柱几何下输运方程S_n间断有限元方法的并行计算.数值试验表明,在通信延迟较高的大型国产并行机上,新算法用数百个CPU还可以取得较好的并行效果,比已有方法具有更良好的可扩展性.     

Scalable parallel methods for monolithic coupling in fluid–structure interaction with application to blood flow modeling

We introduce and study numerically a scalable parallel finite element solver for the simulation of blood flow in compliant arteries. The incompressible Navier–Stokes equations are used to model the fluid and coupled to an incompressible linear elastic model for the blood vessel walls. Our method features an unstructured dynamic mesh capable of modeling complicated geometries, an arbitrary Lagrangian–Eulerian framework that allows for large displacements of the moving fluid domain, monolithic coupling between the fluid and structure equations, and fully implicit time discretization. Simulations based on blood vessel geometries derived from patient-specific clinical data are performed on large supercomputers using scalable Newton–Krylov algorithms preconditioned with an overlapping restricted additive Schwarz method that preconditions the entire fluid–structure system together. The algorithm is shown to be robust and scalable for a variety of physical parameters, scaling to hundreds of processors and millions of unknowns.     

Energy stable numerical methods for hyperbolic partial differential equations using overlapping domain decomposition

Overlapping domain decomposition methods, otherwise known as overset grid or chimera methods, are useful for simplifying the discretization of partial differential equations in or around complex geometries. Though in wide use, such methods are prone to numerical instability unless numerical diffusion or some other form of regularization is used, especially for higher-order methods. To address this shortcoming, high-order, provably energy stable, overlapping domain decomposition methods are derived for hyperbolic initial boundary value problems. The overlap is treated by splitting the domain into pieces and using generalized summation-by-parts derivative operators and polynomial interpolation. New implicit and explicit operators are derived that do not require regularization for stability in the linear limit. Applications to linear and nonlinear problems in one and two dimensions are presented, where it is found the explicit operators are preferred to the implicit ones.     

Discussion on Fresnel's mirrors and Young's double-slit interferometers

In this work, Fresnel's mirror and Young's double-slit experiments are compared. Numerical calculations show that Fresnel's experiment and Young's experiment present significant differences between their interference patterns when the optical source is extended rather than point-like. Those differences agree with the analogous between the overlapping of two quasi-monochromatic beams on the temporal domain (the analogous of Young's experiment) and these same beams superimposed previous modulation with sawtooth-wave (the analogous of Fresnel's interferometer). The implemented algorithms allow evaluating the cases of fully spatially coherence, fully spatially incoherence and spatially partially coherence. Computer simulations are presented to show the validity of our proposal.     

不可压N-S方程的基于粘接元的区域分裂解法

首先,简单介绍了基于粘接元的无重叠区域分裂方法.这种方法利用变分原理,非常适合有限元近似.然后,着重讨论了这种区域分裂方法在求解不可压Navier-Stokes方程中的应用,具体包括等价变分公式的建立、通过算子分裂的时间离散、区域分裂情形下广义Stokes问题的共轭梯度迭代求解方法、空间的有限元离散.最后,以数值实验结果验证了这种区域分裂方法应用于不可压Navier-Stokes方程求解时的可靠性.     

Non-conformal domain decomposition method with second-order transmission conditions for time-harmonic electromagnetics

Non-overlapping domain decomposition (DD) methods provide efficient algorithms for solving time-harmonic Maxwell equations. It has been shown that the convergence of DD algorithms can be improved significantly by using high order transmission conditions. In this paper, we extend a newly developed second-order transmission condition (SOTC), which involves two second-order transverse derivatives, to facilitate fast convergence in the non-conformal DD algorithms. However, the non-conformal nature of the DD methods introduces an additional technical difficulty, which results in poor convergence in many real-life applications. To mitigate the difficulty, a corner-edge penalty method is proposed and implemented in conjunction with the SOTC to obtain truly robust solver performance. Numerical results verify the analysis and demonstrate the effectiveness of the proposed methods on a few model problems. Finally, drastically improved convergence, compared to the conventional Robin transmission condition, was observed for an electrically large problem of practical interest.     

A domain decomposition method for solving the three-dimensional time-harmonic Maxwell equations discretized by discontinuous Galerkin methods

We present here a domain decomposition method for solving the three-dimensional time-harmonic Maxwell equations discretized by a discontinuous Galerkin method. In order to allow the treatment of irregularly shaped geometries, the discontinuous Galerkin method is formulated on unstructured tetrahedral meshes. The domain decomposition strategy takes the form of a Schwarz-type algorithm where a continuity condition on the incoming characteristic variables is imposed at the interfaces between neighboring subdomains. A multifrontal sparse direct solver is used at the subdomain level. The resulting domain decomposition strategy can be viewed as a hybrid iterative/direct solution method for the large, sparse and complex coefficients algebraic system resulting from the discretization of the time-harmonic Maxwell equations by a discontinuous Galerkin method.