文档库 最新最全的文档下载
当前位置:文档库 › 数值分析与程序设计教学大纲

数值分析与程序设计教学大纲

上海交通大学致远学院2013年春季学期

《数值分析与程序设计》课程教学说明

一.课程基本信息

1.开课学院(系)和学科:致远学院

2.课程名称:《数值分析与程序设计》(Scientific Computing)

3.学时/学分:64学时/ 4学分

4.上课时间:周二(10:00 -- 11:40)、周五(8:00 -- 9:40)

5.上课地点:下院205

6.任课教师:应文俊wying@https://www.wendangku.net/doc/c517037586.html,

7.办公室及电话:包玉刚图书馆619 (021-********)

8.助教:待定

9.答疑时间:周四(16:00 -- 18:00)

课程成绩:由平时作业成绩,期末考试成绩及课堂表现综合组成

教材/教学参考书:

【1】《数值分析》(第七版影印版),Richard L. Burden and J. Douglas Faires, 高等教育出版社。

【2】《数值分析基础》关治,陆金甫编著高等教育出版社。

【3】《Numerical Linear Algebra》, L. N. Trefethen and David Bau, SIAM, 1997.

【4】《Numerical Recipes: The Art of Scientific Computing》, W. Press et al., Cambridge University Press, 2002.

二.课程主要内容(中文)

1.数值分析与程序设计介绍(4个学时)

科学计算的精神:“更快、更高、更强”

Linux操作系统,C/C++语言规范及程序基本结构

2.非线性方程求根(4个学时)

多项式求根,二分法,不动点迭代法,牛顿法,割线法与Muller方法

杂交法,Aitken加速技巧

3.线性方程组的直接法(8个学时)

Gauss消去法(含主元法),LU分解法,平方根法,追赶法

Gram-Schmidt正交化过程,Householder变换,QR分解法

最小二乘问题和高斯消去法不适用问题的数值解法

4.矩阵特征值的计算(8个学时)

幂法,逆幂法,Rayleigh商,QR迭代法

Jacobi方法,Sturm序列和二分法

奇异值分解(Singular Value Decomposition)

5.线性方程组的静态迭代法(4个学时)

Jacobi,Gauss-Seidel, 超松弛迭代法(SOR)

6.线性方程组的变分(动态)迭代法(6个学时)

最速下降法,共轭梯度法,最小残量法,广义最小残量法

7.非线性方程组的迭代法(6个学时)

非线性Jacobi方法,非线性Gauss-Seidel方法

非线性最速下降法,非线性共轭梯度法

牛顿法,拟牛顿法,非线性最小二乘,惩罚法,拉格朗日乘子法

8.多项式插值和函数逼近(6个学时)

拉格朗日插值,牛顿插值公式,厄米特插值,样条函数插值

正交多项式,周期函数的最佳平方逼近,函数的最佳一致逼近

9.数值积分和数值微分(6个学时)

Newton-Cotes求积公式,复合求积公式,Gauss,Romberg求积公式

奇异积分与振荡函数的积分,数值微分

10.常微分方程初值问题的数值解法(8个学时)

欧拉法,Runge-Kutta法,线性多步法,刚性方程组的稳定解法

11.常微分方程边值问题的数值解法(4个学时)

有限差分法,有限元方法

三.课程内容(英文)

1.Introduction to scientific computing(4 lectures)

Spirit of the numeric world: “faster, higher, stronger”

Linux operating system, programing in C/C++

2.Root finding of nonlinear scalar equation(4 lectures)

Bisection method, fixed point method, Newton method, secant method, Muller method, Aitken’s acceleration technique and hybrid methods

3.Direct methods for linear equations(8 lectures)

Gauss elimination (including the one with pivoting), LU decomposition, Cholesky decomposition and the Thomas algorithm

Gram-Schmidt orthogonalization process, Householder transform, QR decomposition method, solution of the normal equation (least squares)

https://www.wendangku.net/doc/c517037586.html,putation of eigenvalues and eigenvectors(8 lectures)

The power methods, the Rayleigh acceleration, the QR algorithm

Other methods for symmetric matrices: Jacobi method, the bisection method with Sturm sequence, Singular value decomposition

5.Stationary iterative methods for linear equations (4 lectures)

The Jacobi,Gauss-Seidel, successive over-relaxation (SOR) methods

6.Variational iterative methods for linear equations (6 lectures)

Optimization methods:the steepest descent method, the conjugate gradient method and its preconditioning

Projection methods: the Arnoldi method, the MINRES, GMRES methods

7.Iterative methods for system of nonlinear equations (6 lectures)

Nonlinear Jacobi, Gauss-Seidel methods

Nonlinear steepest descent and conjugate gradient methods

Newton method and quasi-Newton method

Nonlinear optimization, Levenberg-Marquardt method

The penalty method, the Lagrange multiplier method

8.Polynomial interpolation and function approximation (6 lectures)

Lagrange interpolation, Newton divided difference, Hermite interpolation, spline interpolation

Orthogonal polynomials, the best uniform approximation, trigonometric approximation

9.Numerical integration and numerical differentiation (6 lectures)

Newton-Cotes formula,composite rules,Gauss quadrature,Romberg quadrature

Integration of singular or oscillatory functions,numerical differentiation

10.Numerical solution of initial value problems (8 lectures)

Euler methods, Runge-Kutta methods, multistep methods

Stable methods for stiff ODEs

11.Numerical solution of boundary value problems (4 lectures)

Finite difference method and finite element method

《数值分析与程序设计》教学计划

相关文档
相关文档 最新文档