用自适应MG法,FMG法求解Helmholtz方程

本文针对Ｖ循环、Ｗ循环和多重网格法中最优光滑次数及循环体个数难以确定的缺点，以Ｈｅｌｍｈｏｌｔｚ方程为例给出自适应的多重网格算法和自适应的完全多重网格算法。     

一类非对称椭圆问题瀑布型多重网格法

本将瀑布型多重网格法用于求解非对称椭圆边值问题，数值结果表明算法是有效的。     

二次Lagrangian有限元方程的几何多重网格法

为了构造快速求解二次Lagrangian有限元方程的几何多重网格法,在选择二次Lagrangian有限元空间和一系列线性Lagrangian有限元空间分别作为最细网格层和其余粗网格层以及构造一种新限制算子的基础上,提出了一种新的几何多重网格法,并对它的计算量进行了估计.数值实验结果,与通常的几何多重网格法和AMG01法相比,表明了新算法计算量少且稳健性强.     

一类非线性椭圆问题的瀑布型多重网格法

本对二阶非线性椭圆问题提出一种瀑布型多重网格法，数值实验表明该算法非常有效，当d＝1时，给出了理论结果。     

求解三维高次拉格朗日有限元方程的代数多重网格法

本文针对带有间断系数的三维椭圆问题,讨论任意四面体剖分下的二次拉格朗日有限元方程的代数多重网格法．通过分析线性和高次有限元空间之间的关系,我们给出了一种新的网格粗化算法和构造提升算子的代数途径．进一步,我们还对新的代数多重网格法给出了收敛性分析．数值实验表明这种代数多重网格法对求解二次拉格朗日有限元方程是健壮和有效的。     

新外推完全多重网格法北大核心

使用新外推公式和高阶插值算子,为相邻细层提供好的初值,对初值使用磨光算子磨光几次后,再调用V型多重网格法求得该层数值解,构造了基于四阶紧致差分格式的新外推完全多重网格法.数值实验表明,与对比算法相比,新算法迭代次数少、计算时间短、稳健性强.     

新外推完全多重网格法

使用新外推公式和高阶插值算子,为相邻细层提供好的初值,对初值使用磨光算子磨光几次后,再调用V型多重网格法求得该层数值解,构造了基于四阶紧致差分格式的新外推完全多重网格法.数值实验表明,与对比算法相比,新算法迭代次数少、计算时间短、稳健性强.     

解二阶椭圆型变分不等式的瀑布型多重网格法

本文将瀑布型多重网格法推广应用于求解二阶椭圆型变分不等式并给出了一些数值例子。数值算例表明该算法是有效的。     

求解线性互补问题的乘性Schwarz算法的收敛速度估计

1．引言区域分解法是八十年代兴起并得到迅速发展及广泛应用的数值计算方法．和多重网格法一样，区域分解法用于求解椭圆边值问题时具有与剖分网格h无关的收敛速度［8］，因而是一种高效快速算法．八十年代末及九十年代初，这种区域分解思想也开始应用于障碍问题的求解［2－8，10。12，16］数值实验表明，该算法对于障碍问题也是有效的·但是，和多重网格法一样，用于求解障碍问题时，算法的收敛速度分析存在一定的困难［11，13，14］对于障碍问题，一般的收敛性证明都是建立在证明算法产生的序列为一个极小化序列的基础之上［‘，‘’，“…     

双障碍问题的多重网格法

本文研究双障碍问题的多重网格法,提出了两类算法,证明了其收敛性及对贴合分量的有限步收敛性,同时对其中一种算法的特款提出了一个 k无关收敛性定理。     

Artificial damping in multigrid methods

When the solution and problem coefficients are highly oscillatory, the computed solution may not show characteristics of the original physical problem unless the numerical mesh is sufficiently fine. In the case, the coarse grid problem of a multigrid (MG) algorithm must be still huge and poorly-conditioned, and therefore, it is hard to solve by either a direct method or an iterative scheme. This article suggests a MG algorithm for such problems in which the coarse grid problem is slightly modified by an artificial damping (compressibility) term. It has been numerically observed that the artificial damping, even if slight, makes the coarse grid problem much easier to solve, without deteriorating the overall convergence rate of the MG method. For most problems, 2–6 times speed up have been observed.     

Improved variational image registration model and a fast algorithm for its numerical approximation

In a multimodal image registration scenario, where two given images have similar features, but noncomparable intensity variations, the sum of squared differences is not suitable for inferring image similarities. In this article, we first propose a new variational model based on combining intensity and geometric transformations, as an alternative to use mutual information and an improvement to the work by Modersitzki and Wirtz (Modersitzki and Wirtz, Lect Notes Comput Sci 4057 (2006), 257–263), and then develop a fast multigrid (MG) algorithm for solving the underlying system of fourth‐order and nonlinear partial differential equations. We can demonstrate the effective smoothing property of the adopted primal‐dual smoother by a local Fourier analysis. Numerical tests will be presented to show both the improvements achieved in image registration quality and MG efficiency. © 2011 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq, 2011     

Truncated Newton-Based Multigrid Algorithm for Centroidal Voronoi Diagram Calculation

In a variety of modern applications there arises a need to tessellate the domain into representative regions, called Voronoi cells. A particular type of such tessellations, called centroidal Voronoi tessellations or CVTs, are in big demand due to their optimality properties important for many applications. The availability of fast and reliable algorithms for their construction is crucial for their successful use in practical settings. This paper introduces a new multigrid algorithm for constructing CVTs that is based on the MG/Opt algorithm that was originally designed to solve large nonlinear optimization problems. Uniform convergence of the new method and its speedup comparing to existing techniques are demonstrated for linear and nonlinear densities for several 1d and 2d problems, and $O(k)$ complexity estimation is provided for a problem with $k$ generators.     

Using inexact gradients in a multilevel optimization algorithm

Many optimization algorithms require gradients of the model functions, but computing accurate gradients can be computationally expensive. We study the implications of using inexact gradients in the context of the multilevel optimization algorithm MG/Opt. MG/Opt recursively uses (typically cheaper) coarse models to obtain search directions for finer-level models. However, MG/Opt requires the gradient on the fine level to define the recursion. Our primary focus here is the impact of the gradient errors on the multilevel recursion. We analyze, partly through model problems, how MG/Opt is affected under various assumptions about the source of the error in the gradients, and demonstrate that in many cases the effect of the errors is benign. Computational experiments are included.     

On θ-completions for Maximal Subgroups

For a maximal subgroup M of a group G,a θ-completion for M is a subgroup C such that C is not contained in M while Ma,the core of M in G,is contained in C and C/M_G has no proper normal subgroup of G/M_G.This concept was introduced by ZHAO Yao-qing in 1998.In this paper we characterize the solvability of finite groups by means of θ-completions and obtain some new results.     

Extrapolation Combined with Multigrid Method for Solving Finite Element Equations

An algorithm combining the MG method with two types of extrapolation is given for solving finite element equations with any initial triangulation. A high order approximation to the solution of PDEs can be obtained at the cost of order O(N) of computational work.     

Comparison of V‐cycle multigrid method for cell‐centered finite difference on triangular meshes

We consider a multigrid algorithm (MG) for the cell centered finite difference scheme (CCFD) on general triangular meshes using a new prolongation operator. This prolongation is designed to solve the diffusion equation with strongly discontinuous coefficient as well as with smooth one. We compare our new prolongation with the natural injection and the weighted operator in Kwak, Kwon, and Lee ( 8 ) and the behaviors of these three prolongation are discussed. Numerical experiments show that (i) for smooth problems, the multigrid with our new prolongation is fastest, the next is the weighted prolongation, and the third is the natural injection; and (ii) for nonsmooth problems, our new prolongation is again fastest, the next is the natural injection, and the third is the weighted prolongation. In conclusion, our new prolongation works better than the natural injection and the weighted operator for both smooth and nonsmooth problems. © 2006 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq, 2006     

MgNet: A unified framework of multigrid and convolutional neural network

We develop a unified model, known as MgNet, that simultaneously recovers some convolutional neural networks(CNN) for image classification and multigrid(MG) methods for solving discretized partial differential equations(PDEs). This model is based on close connections that we have observed and uncovered between the CNN and MG methodologies. For example, pooling operation and feature extraction in CNN correspond directly to restriction operation and iterative smoothers in MG, respectively. As the solution space is often the dual of the data space in PDEs, the analogous concept of feature space and data space(which are dual to each other) is introduced in CNN. With such connections and new concept in the unified model, the function of various convolution operations and pooling used in CNN can be better understood. As a result,modified CNN models(with fewer weights and hyperparameters) are developed that exhibit competitive and sometimes better performance in comparison with existing CNN models when applied to both CIFAR-10 and CIFAR-100 data sets.     

TIME-EXTRAPOLATION ALGORITHM （TEA） FOR LINEAR PARABOLIC PROBLEMS

The fast solutions of Crank-Nicolson scheme on quasi-uniform mesh for parabolic prob- lems are discussed. First, to decrease regularity requirements of solutions, some new error estimates are proved. Second, we analyze the two characteristics of parabolic discrete scheme, and find that the efficiency of Multigrid Method （MG） is greatly reduced. Nu- merical experiments compare the efficiency of Direct Conjugate Gradient Method （DCG） and Extrapolation Cascadic Multigrid Method （EXCMG）. Last, we propose a Time- Extrapolation Algorithm （TEA）, which takes a linear combination of previous several level solutions as good initial values to accelerate the rate of convergence. Some typical extrapolation formulas are compared numerically. And we find that under certain accuracy requirement, the CG iteration count for the 3-order and 7-level extrapolation formula is about 1/3 of that of DCG＇s. Since the TEA algorithm is independent of the space dimension, it is still valid for quasi-uniform meshes. As only the finest grid is needed, the proposed method is regarded very effective for nonlinear parabolic problems.     

Convergence of the multigrid full approximation scheme for a class of elliptic mildly nonlinear boundary value problems

Summary The multigrid full approximation scheme (FAS MG) is a well-known solver for nonlinear boundary value problems. In this paper we restrict ourselves to a class of second order elliptic mildly nonlinear problems and we give local conditions, e.g. a local Lipschitz condition on the derivative of the continuous operator, under which the FAS MG with suitably chosen parameters locally converges. We prove quantitative convergence statements and deduce explicit bounds for important quantities such as the radius of a ball of guaranteed convergence, the number of smoothings needed, the number of coarse grid corrections needed and the number of FAS MG iterations needed in a nested iteration. These bounds show well-known features of the FAS MG scheme.