解二维对流扩散方程的块ADI方法

构造了解二维对流扩散方程的块交替方向隐式(ADI)方法。块ADI方法绝对稳定,并且具有计算和通讯的局部化特点,适合在分布存储的大规模并行计算系统上应用。文中给出了工作站机群系统上并行计算的数值例子     

利用改进的ADI方法实现单机千万节点的油水两相渗流数值模拟

针对实际油藏的非均质分布特征及其复杂的边界条件,本文通过引入迭代参数的压缩因子和放大因子,对现有的交替方向迭代法(ADI)进行改进,提出一种适用于大规模油藏数值模拟的新算法．改进的ADI 方法计算精度可靠,且与现有的算法相比计算效率有所提高．更为关键的是,ADI 算法将求解三维压力方程的七对角矩阵分解为三个方向的三对角矩阵依次迭代求解,所需的运算存储量大幅减少,最大的计算规模有了大幅的提升．使用改进的ADI 方法,在单机上成功实现了千万节点的油水两相渗流数值模拟．计算实例表明,在同等单机硬件条件下,改进ADI 算法的最大计算规模是现有算法的1.7 倍以上．     

Divergence preservation in the ADI algorithms for electromagnetics

The recent advances in alternating direct implicit (ADI) methods promise important new capability for time domain plasma simulations, namely the elimination of numerical stability limits on the time step. But the utility of these methods in simulations with charge and current sources, such as in electromagnetic particle-in-cell (EMPIC) computations, has been uncertain, as the methods introduced so far do not have the property of divergence preservation. This property is related to charge conservation and self-consistency, and is critical for accurate and robust EMPIC simulation. This paper contains a complete study of these ADI methods in the presence of charge and current sources. It is shown that there are four significantly distinct cases, with four more related by duality. Of those, only one preserves divergence and, thus, is guaranteed to be stable in the presence of moving charged particles. Computational verification of this property is accomplished by implementation in existing 3D-EMPIC simulation software. Of the other three cases, two are verified unstable, as expected, and one remains stable, despite the lack of divergence preservation. This other stable algorithm is shown to be related to the divergence preserving case by a similarity transformation, effectively providing the complement of the divergence preserving field in the finite-difference energy quantity.     

On the ADI method for Sylvester equations

This paper is concerned with the numerical solution of large scale Sylvester equations AX−XB=C, Lyapunov equations as a special case in particular included, with C having very small rank. For stable Lyapunov equations, Penzl (2000) [22] and Li and White (2002) [20] demonstrated that the so-called Cholesky factor ADI method with decent shift parameters can be very effective. In this paper we present a generalization of the Cholesky factor ADI method for Sylvester equations. An easily implementable extension of Penz’s shift strategy for the Lyapunov equation is presented for the current case. It is demonstrated that Galerkin projection via ADI subspaces often produces much more accurate solutions than ADI solutions.     

On optimal message vector length for block single parallel partition algorithm in a three-dimensional ADI solver

It has long been recognized that many direct parallel tridiagonal solvers are only efficient for solving a single tridiagonal equation of large sizes, and they become inefficient when naively used in a three-dimensional ADI solver. In order to improve the parallel efficiency of an ADI solver using a direct parallel solver, we implement the single parallel partition (SPP) algorithm in conjunction with message vectorization, which aggregates several communication messages into one to reduce the communication costs. The measured performances show that the longest allowable message vector length (MVL) is not necessarily the best choice. To understand this observation and optimize the performance, we propose an improved model that takes the cache effect into consideration. The optimal MVL for achieving the best performance is shown to depend on number of processors and grid sizes. Similar dependence of the optimal MVL is also found for the popular block pipelined method.     

ADI土壤水分反演方法

土壤水分是影响植被、土壤和大气之间能量和水分循环的重要因素,及时准确获取土壤湿度信息有利于提高作物估产精度和改善田间管理措施。本文基于红光与近红外光谱特征空间(NIR-RED)发展了一种新型土壤水分遥感监测模型ADI(angle dryness index),提高了可见光与近红外波段监测土壤水分的精度。经过研究表明,在红光与近红外(NIR-RED)特征空间中,存在一个中间角度变量θ,利用光谱反射率与土壤水分之间的经验关系式模型以及混合像元分解公式证明该变量能够表征土壤湿度情况,而不受植被覆盖度的影响,因此利用该原理构建了ADI方法。最后利用两组遥感数据(分别为TM5与MODIS产品数据)以及对应的地面观测数据进行验证,结果表明计算值与实测值均具有较高的一致性,R2分别达到0.74与0.64。同时,将MPDI的计算结果与实测值进行了比较,两组数据的R2均小于0.60,表明ADI方法的计算精度高于MPDI。在MPDI的计算过程中用到了植被覆盖度,这可能是引起计算结果误差的主要因素。此外,MPDI的计算结果表征土壤湿度的相对值,而ADI则能定量的获取土壤水分含量。MODIS像元除了具有植被与土壤两个端元,还有其他类型端元的概率高于TM数据,因而MODIS数据的计算精度低于TM。因此,ADI是一种简单可行且具有较大应用前景的土壤水分反演方法,适合于推广应用。     

Numerical approximation of Turing patterns in electrodeposition by ADI methods

In this paper we study the numerical approximation of Turing patterns corresponding to steady state solutions of a PDE system of reaction–diffusion equations modeling an electrodeposition process. We apply the Method of Lines (MOL) and describe the semi-discretization by high order finite differences in space given by the Extended Central Difference Formulas (ECDFs) that approximate Neumann boundary conditions (BCs) with the same accuracy. We introduce a test equation to describe the interplay between the diffusion and the reaction time scales. We present a stability analysis of a selection of time-integrators (IMEX 2-SBDF method, Crank–Nicolson (CN), Alternating Direction Implicit (ADI) method) for the test equation as well as for the Schnakenberg model, prototype of nonlinear reaction–diffusion systems with Turing patterns. Eventually, we apply the ADI-ECDF schemes to solve the electrodeposition model until the stationary patterns (spots & worms and only spots) are reached. We validate the model by comparison with experiments on Cu film growth by electrodeposition.     

Convergence of compact ADI method for solving linear Schrödinger equations

A compact ADI scheme of second‐order in time and fourth‐order in space is proposed for solving linear Schrödinger equations with periodic boundary conditions. By using the recently suggested discrete energy method, it is shown that the stable compact ADI method is unconditionally convergent in the maximum norm. Numerical experiments, including the comparisons with the second‐order ADI scheme and the time‐splitting Fourier pseudospectral method, are presented to support the theoretical results and show the effectiveness of our method. © 2011 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq 2012     

An unconditionally stable alternating direction implicit scheme for the two space dimensional linear hyperbolic equation

We report a new unconditionally stable implicit alternating direction implicit (ADI) scheme of O(k² + h²) for the difference solution of linear hyperbolic equation u_tt + 2αu_t + β²u = u_xx + u_yy + f(x, y, t), αβ ≥ 0, 0 < x, y < 1, t > 0 subject to appropriate initial and Dirichlet boundary conditions, where α > 0 and β ≥ 0 are real numbers. The resulting system of algebraic equations is solved by split method. Numerical results are provided to demonstrate the efficiency and accuracy of the method. © 2001 John Wiley & Sons, Inc. Numer Methods Partial Differential Eq 17: 684–688, 2001     

Fully HOC Scheme for Mixed Convection Flow in a Lid-Driven Cavity Filled with a Nanofluid

A fully higher-order compact (HOC) finite difference scheme on the 9-point two-dimensional (2D) stencil is formulated for solving the steady-state laminar mixed convection flow in a lid-driven inclined square enclosure filled with water-$Al_2O_3$ nanofluid. Two cases are considered depending on the direction of temperature gradient imposed (Case I, top and bottom; Case II, left and right). The developed equations are given in terms of the stream function-vorticity formulation and are non-dimensionalized and then solved numerically by a fourth-order accurate compact finite difference method. Unlike other compact solution procedure in literature for this physical configuration, the present method is fully compact and fully higher-order accurate. The fluid flow, heat transfer and heat transport characteristics were illustrated by streamlines, isotherms and averaged Nusselt number. Comparisons with previously published work are performed and found to be in excellent agreement. A parametric study is conducted and a set of graphical results is presented and discussed to elucidate that significant heat transfer enhancement can be obtained due to the presence of nanoparticles and that this is accentuated by inclination of the enclosure at moderate and large Richardson numbers.