基于新的拟牛顿方程的Broyden-Fletcher-Goldfarb-Shanno算法

通过对函数的泰勒展开式进行误差分析,提出了对二次模型进行改进的新模型,在此基础上得到了改进的拟牛顿条件,并得到了与其相应的Broyden-Fletcher-Goldfarb-Shanno(BFGS)算法.证明了在适当条件下该算法全局收敛.从试验函数库中选择标准测试函数,对经典的BFGS算法与改进的BFGS算法进行数值试验,试验结果表明改进的算法优于经典的BFGS算法.     

一种基于训练数据的迭代改进核函数

为提高支持向量机性能,提出一种支持向量机核函数的迭代改进新算法.利用与数据有关的保角映射,使核函数包含了全部学习样本的信息,即核函数具有数据依赖性.基本核函数的参数可取随机初值,通过对核函数进行多次迭代改进,直至得到满意的学习效果.与传统方法相比,新算法不需要筛选核函数的参数.对一元连续函数和强地震事件的仿真计算结果表明,改进SVR(support vector regression)的学习效果优于传统方法,并且随着迭代次数的增加,学习风险下降收敛,收敛速度依赖于传统方法的基本参数和改进方法的参数.     

基于粒子群优化的神经网络预测模型

BP学习算法多采用梯度下降法调整权值,针对其易陷入局部极小、收敛速度慢和易引起振荡的固有缺陷,提出了一种改进粒子群神经网络算法.其基本思想是:首先采用改进粒子群优化算法反复优化BP神经网络模型的权值参数组合,再用BP算法对得到的网络参数进一步精确优化,最后用得到精确的最优参数组合进行预测.实验结果表明,该算法在股指预测中的预测性能明显提高.     

基于蚁群神经网络的财务危机预警方法

为了克服神经网络财务危机预警方法收敛慢、不收敛和网络结构难以确定等缺陷,提出了基于蚁群算法的改进神经网络财务危机预警方法。将神经网络模型的结构和参数进行编码,利用蚁群算法确定若干个神经网络模型的结构和参数,然后通过评价函数得到神经网络的最佳结构,最后通过BP算法训练该神经网络,得到神经网络财务危机预警模型。验证结果表明,该模型结构简单、预警精度高。     

Banach空间中关于m-增生算子零点的一种新的粘性隐式迭代算法

在Banach空间中研究关于两个逆强增生算子的一般变分不等式问题和m-增生算子零点的粘性隐式迭代算法,对参数的适当限制下,利用超梯度方法,得到了若干强收敛定理,推广和改进了其他相关作者的主要结果.     

等式约束优化一个修正的投影变尺度法

本文研究了等式约束优化问题.利用罚函数和投影变尺度方法,得到了一个修正的算法及其全局收敛与超线性收敛率.改进了文献[J]中的方法.     

三维Navier-Stokes方程加罚有限元的共轭梯度法和分块迭代法

本文对Navier-Stokes问题加罚变分形成有限元解给出了共轭梯度算法和分块迭代算法,由于共轭梯度算法中,求解单变量极小值问题得到简化,使得计算时间大为节约. 本文还给出了计算实例.     

同伦摄动稀疏正则化方法及其应用

稀疏正则化方法在参数重构中起到了越来越重要的作用.与传统的正则化方法相比,稀疏正则化方法能较好地重构稀疏变量.由于稀疏正则化的不可微性,需要对已有的经典算法进行改进.本文构建同伦摄动稀疏正则化方法克服标准稀疏正则化的不可微性,并将该方法应用到基于布莱克一斯科尔斯期权定价模型重构隐含波动率和基于托达罗模型重构政策参数.数值实验表明,所提出的方法是收敛和稳定的.     

凸可行问题的一种强收敛算法

无限维Hilbert空间中,解凸可行问题的平行投影算法通常是弱收敛的.本文对一般的平行投影算法进行改进,设计了一种解凸可行问题的具有强收敛性的新算法.该算法主要是在原有算法基础上引入了一个参数序列,在参数序列满足一定的控制条件下保证了算法的强收敛性.为了简单证明算法的强收敛性,我们构建了一个新的积空间,然后把原空间的这种改进平行投影算法转换为积空间中的交替投影算法.这样,改进的平行投影算法的强收敛性就可以通过交替投影算法的收敛性证明得到.     

Banach空间中可数拟-φ-非扩张映像族的公共不动点的收敛定理

在某些Banach空间中针对一类闭的拟-φ-非扩张映像的可数无限族,修正经典的正规Mann迭代算法以达到强收敛的目标,所得结果改进并扩展了Matsushita和Takahashi等人的相关结果.     

Penalty finite element approximations for the Stokes equations by L2 projection

This paper considers the penalty finite element method for the Stokes equations, based on some stable finite elements space pair (X_h, M_h) that do satisfy the discrete inf–sup condition. Theoretical results show that the penalty error converges as fast as one should expect from the order of the elements. Moreover, the penalty finite element method by L² projection can improve the penalty error estimates. Finally, we confirm these results by a series of numerical experiments. Copyright © 2008 John Wiley & Sons, Ltd.     

Composite penalty method of a low order anisotropic nonconforming finite element for the Stokes problem

Composite penalty method of a low order anisotropic nonconforming quadrilateral finite element for the Stokes problem is presented. This method with a large penalty parameter can achieve the same accuracy as the stand method with a small penalty parameter and the convergence rate of this method is two times as that of the standard method under the condition of the same order penalty parameter. The superconvergence for velocity is established as well. The results of this paper are also valid to the most of the known nonconforming finite element methods.     

Stabilized Nonconforming Mixed Finite Element Method for Linear Elasticity on Rectangular or Cubic Meshes

Based on the primal mixed variational formulation, a stabilized nonconforming mixed finite element method is proposed for the linear elasticity on rectangular and cubic meshes. Two kinds of penalty terms are introduced in the stabilized mixed formulation, which are the jump penalty term for the displacement and the divergence penalty term for the stress. We use the classical nonconforming rectangular and cubic elements for the displacement and the discontinuous piecewise polynomial space for the stress, where the discrete space for stress are carefully chosen to guarantee the well-posedness of discrete formulation. The stabilized mixed method is locking-free. The optimal convergence order is derived in the $L^2$-norm for stress and in the broken $H^1$-norm and $L^2$-norm for displacement. A numerical test is carried out to verify the optimal convergence of the stabilized method.     

A discontinuous Galerkin method for Stokes equation by divergence-free patch reconstruction

A discontinuous Galerkin method by patch reconstruction is proposed for Stokes flows. A locally divergence-free reconstruction space is employed as the approximation space, and the interior penalty method is adopted which imposes the normal component penalty terms to cancel out the pressure term. Consequently, the Stokes equation can be solved as an elliptic system instead of a saddle-point problem due to such weak form. The number of degree of freedoms of our method is the same as the number of elements in the mesh for different order of accuracy. The error estimations of the proposed method are given in a classical style, which are then verified by some numerical examples.     

The p-version of the FEM for computational contact mechanics

Contact analyses are being performed in various engineering applications. Here, like in most other fields, FE codes are based on low order elements using linear or quadratic shape functions. The intention of this paper is to show that finite elements with shape functions of high polynomial degree (p-FEM) are a very attractive alternative to low order elements, even for computational contact mechanics. One of the advantages is the possibility to enhance the element formulation with the blending function method in order to accurately discretize the given geometry, which leads in combination with high convergence rates to very efficient computations. In order to solve the problem of frictionless contact, a penalty formulation is applied in this work. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Optimal Error Estimate of the Penalty Finite Element Method for the Micropolar Fluid Equations

We present an optimal error estimate of the numerical velocity, pressure, and angular velocity for the fully discrete penalty finite element method of the micropolar equations when the parameters ?, Δ t, and h are sufficiently small. In order to obtain this estimate, we present the time discretization of the penalty micropolar equation that is based on the backward Euler scheme; the spatial discretization of the time discretized penalty micropolar equation is based on a finite elements space pair (X _h , M _h ) that satisfies some approximations properties.     

An unfitted interface penalty method for the numerical approximation of contrast problems

We aim to approximate contrast problems by means of a numerical scheme which does not require that the computational mesh conforms with the discontinuity between coefficients. We focus on the approximation of diffusion-reaction equations in the framework of finite elements. In order to improve the unsatisfactory behavior of Lagrangian elements for this particular problem, we resort to an enriched approximation space, which involves elements cut by the interface. Firstly, we analyze the H¹-stability of the finite element space with respect to the position of the interface. This analysis, applied to the conditioning of the discrete system of equations, shows that the scheme may be ill posed for some configurations of the interface. Secondly, we propose a stabilization strategy, based on a scaling technique, which restores the standard properties of a Lagrangian finite element space and results to be very easily implemented. We also address the behavior of the scheme with respect to large contrast problems ending up with a choice of Nitsche?s penalty terms such that the extended finite element scheme with penalty is robust for the worst case among small sub-elements and large contrast problems. The theoretical results are finally illustrated by means of numerical experiments.     

A geometric‐based algebraic multigrid method for higher‐order finite element equations in two‐dimensional linear elasticity

In this paper, we will discuss the geometric‐based algebraic multigrid (AMG) method for two‐dimensional linear elasticity problems discretized using quadratic and cubic elements. First, a two‐level method is proposed by analyzing the relationship between the linear finite element space and higher‐order finite element space. And then a geometric‐based AMG method is obtained with the existing solver used as a solver on the first coarse level. The resulting AMG method is applied to some typical elasticity problems including the plane strain problem with jumps in Young's modulus. The results of various numerical experiments show that the proposed AMG method is much more robust and efficient than a classical AMG solver that is applied directly to the high‐order systems alone. Moreover, we present the corresponding theoretical analysis for the convergence of the proposed AMG algorithms. These theoretical results are also confirmed by some numerical tests. Copyright © 2008 John Wiley & Sons, Ltd.     

Analysis of some mixed elements for the Stokes problem

In this paper we discuss some mixed finite element methods related to the reduced integration penalty method for solving the Stokes problem. We prove optimal order error estimates for bilinear-constant and biquadratic-bilinear velocity-pressure finite element solutions. The result for the biquadratic-bilinear element is new, while that for the bilinear-constant element improves the convergence analysis of Johnson and Pitkäranta (1982). In the degenerate case when the penalty parameter is set to be zero, our results reduce to some related known results proved in by Brezzi and Fortin (1991) for the bilinear-constant element, and Bercovier and Pironneau (1979) for the biquadratic-bilinear element. Our theoretical results are consistent with the numerical results reported by Carey and Krishnan (1982) and Oden et al. (1982).     

A finite element time relaxation method

We discuss a finite element time-relaxation method for high Reynolds number flows. The method uses local projections on polynomials defined on macroelements of each pair of two elements sharing a face. We prove that this method shares the optimal stability and convergence properties of the continuous interior penalty (CIP) method. We give the formulation both for the scalar convection–diffusion equation and the time-dependent incompressible Euler equations and the associated convergence results. This note finishes with some numerical illustrations.