提问于:
浏览数:
2451
```
\documentclass{article}
\usepackage{subfiles}
\usepackage{tikz}
\usetikzlibrary{positioning}
\usetikzlibrary{positioning,shapes}
\tikzset{special/.style={fill=cyan,ellipse}}
\usepackage[natbibapa]{apacite}
\bibliographystyle{apacite}
% From https://en.wikibooks.org/wiki/LaTeX/Modular_Documents#Subfiles_and_bibtex
% To get citations working in both the main and sub files
\def\biblio{\bibliography{ref}}
\begin{document}
\def\biblio{}
This is the main file!!
\subfile{Chap1/Chap1}
\newpage
\bibliography{ref}
\end{document}
```
1 回答
0
```
\date{}
\begin{document}
\begin{multicols}{3}
\subparagraph{created by wth}\label{header-c7}
\subparagraph{cause}\label{header-c7}
Truncation:计算量有关
Round-off:bit数相关
\subparagraph{不动点法}\label{header-c23}
收敛的充分条件:{[}a,b{]}内所有x满足\textbar{}g'(x)\textbar{}
\textless{} k, (0 \textless{} k \textless{} 1)
收敛速度: \( |p_{n+1} - p_n| <= k * |p_n - p_{n-1}|\) , k越小越快
\subparagraph{牛顿法}\label{header-c30}
用泰勒展开的前两项作为函数近似,求该近似函数的零点,在零点处再求函数近似,迭代。
\(p \approx p_0 - \frac{f(p_0)}{f'(p_0)}\)
牛顿法的k在p点处是0,所以收敛非常快
\subparagraph{迭代法误差分析}\label{header-c38}
\(lim \frac{p_{n+1} - p}{|p_n - p|^a} = \lambda\)
pn收敛于p, a越大,收敛速度越快
a = 1, linearly convergent
a = 2, quadratically convergent
\( g'(p) \neq 0\) 则最少是 linear convergent,拉格朗日中值定理可退
\( g'(p) = 0\) 时(如牛顿法) g不等于0的最高阶导数阶数a
牛顿法是二阶收敛的
\subparagraph{quadratica newtom method}\label{header-c54}
有重根(几重根都可以)的时候令 \( \mu(x) = \frac{f(x)}{f'(x)} \)
然后再做牛顿法 \( g(x) = x - \frac{\mu(x)}{\mu'(x)} \)
\subparagraph{Aitken's Method (Steffensen's Method)}\label{header-c61}
原理
\(p_{n+2} = p_n - \frac{( \Delta p_n )^2}{ \Delta ^ 2 p_n }\)
\( p_1 = g(p_0),\ p2=g(p_1)\)
\(p=p_0-\frac{(p_1-p_0)^2}{(p_2-2*p_1+p_0)}\)
\(p_0=p\)
```
你的回答
请登录后回答
你的回答将会帮助更多人,请务必认真回答问题。