mirror of
https://github.com/Didnelpsun/Math.git
synced 2026-02-08 21:14:50 +08:00
592 lines
24 KiB
TeX
592 lines
24 KiB
TeX
\documentclass[UTF8]{ctexart}
|
||
% UTF8编码,ctexart现实中文
|
||
\usepackage{xcolor}
|
||
% 使用颜色
|
||
\definecolor{orange}{RGB}{255,127,0}
|
||
\definecolor{violet}{RGB}{192,0,255}
|
||
\definecolor{aqua}{RGB}{0,255,255}
|
||
\usepackage{geometry}
|
||
\setcounter{tocdepth}{4}
|
||
\setcounter{secnumdepth}{4}
|
||
% 设置四级目录与标题
|
||
\geometry{papersize={21cm,29.7cm}}
|
||
% 默认大小为A4
|
||
\geometry{left=3.18cm,right=3.18cm,top=2.54cm,bottom=2.54cm}
|
||
% 默认页边距为1英尺与1.25英尺
|
||
\usepackage{indentfirst}
|
||
\setlength{\parindent}{2.45em}
|
||
% 首行缩进2个中文字符
|
||
\usepackage{amssymb}
|
||
% 因为所以与其他数学拓展
|
||
\usepackage{amsmath}
|
||
% 数学公式
|
||
\usepackage{setspace}
|
||
\renewcommand{\baselinestretch}{1.5}
|
||
% 1.5倍行距
|
||
\usepackage{pifont}
|
||
% 圆圈序号
|
||
\usepackage{tikz}
|
||
% 绘图
|
||
\usepackage{array}
|
||
% 设置表格行距
|
||
\author{Didnelpsun}
|
||
\title{一元函数微分学}
|
||
\date{}
|
||
\begin{document}
|
||
\renewcommand{\arraystretch}{1.5}
|
||
% 表格高1.5倍
|
||
\maketitle
|
||
\thispagestyle{empty}
|
||
\tableofcontents
|
||
\thispagestyle{empty}
|
||
\newpage
|
||
\pagestyle{plain}
|
||
\setcounter{page}{1}
|
||
\section{概念}
|
||
\subsection{引例}
|
||
|
||
设$f(x)$下$x$在$x_0$的邻域内,$\alpha$为切线所成夹角。
|
||
|
||
$\tan\alpha=f'(x_0)=\lim_{x\to x_0}\dfrac{f(x)-f(x_0)}{x-x_0}=k$
|
||
|
||
\subsection{导数}
|
||
|
||
设$y=f(x)$定义在区间$I$上,让自变量在$x=x_0$处加一个增量$\Delta x$,其中$x_0\in I$,$x_0+\Delta x\in I$,则可得函数的增量$\Delta y=f(x_0+\Delta x)-f(x_0)$。若函数增量$\Delta y$与自变量增量$\Delta x$的比值在$\Delta x\to 0$时的极限存在,则称函数$y=f(x)$在$x_0$处可导,并称这个极限为$y=f(x)$在点$x_0$处的导数,记作$f'(x)$,即$f'(x)=\lim_{\Delta x\to 0}\dfrac{\Delta y}{\Delta x}=\lim_{\Delta x\to 0}\dfrac{f(x_0+\Delta x)-f(x_0)}{\Delta x}$。
|
||
|
||
下面三句话等价:
|
||
|
||
\begin{enumerate}
|
||
\item $y=f(x)$在$x_0$处可导。
|
||
\item $y=f(x)$在$x_0$处导数存在。
|
||
\item $f'(x)=A$。($A$为有限数)
|
||
\end{enumerate}
|
||
|
||
单侧导数分为左导数和右导数。
|
||
|
||
$f'_-(x)=\lim_{\Delta x\to 0^-}\dfrac{\Delta y}{\Delta x}=\lim_{\Delta x\to 0}\dfrac{f(x_0+\Delta x)-f(x_0)}{\Delta x}$
|
||
|
||
$f'_+(x)=\lim_{\Delta x\to 0^+}\dfrac{\Delta y}{\Delta x}=\lim_{\Delta x\to 0}\dfrac{f(x_0+\Delta x)-f(x_0)}{\Delta x}$
|
||
|
||
所以$f(x)$在$x_0$处可导的充要条件是其左导数和右导数存在且相等。
|
||
|
||
若$f(x)$在$x_0$的左右,如$y=\vert x\vert$在$0$的左右出现了单侧的不同的切线,那这个$x_0$就是一个\textbf{角点},该角点处不可导。
|
||
|
||
若$f(x)$在$x_0$处导数为无穷,如$y=x^{\frac{1}{3}}$在$0$处利用导数的极限定义计算得到为正无穷,那么该点的导数为无穷导数,在考研中被认为是不存在的。
|
||
|
||
\textbf{例题1:}证明若$f(x)$为可导的偶函数,则$f'(x)$为奇函数,若$f(x)$为可导的奇函数,则$f'(x)$为偶函数。
|
||
|
||
该证明是准备部分的定理。
|
||
|
||
首先已知$f(-x)=f(x)$,证明$f'(-x)=-f'(x)$。
|
||
|
||
$\therefore$
|
||
|
||
$
|
||
\begin{aligned}
|
||
f'(-x) &=\lim_{\Delta x\to 0}\dfrac{f(-x+\Delta x)-f(-x)}{\Delta x} \\
|
||
& =\lim_{\Delta x\to 0}\dfrac{f(x+(-\Delta x))}{\Delta x} \\
|
||
& =-\lim_{-\Delta x\to 0}\dfrac{f(x+(-\Delta x))}{-\Delta x} \\
|
||
& =-f'(x)
|
||
\end{aligned}
|
||
$
|
||
|
||
同理得证$f(-x)=-f(x)\Rightarrow f'(-x)=f'(x)$。
|
||
|
||
\textbf{例题2:}证明$f(x)$为可导的周期为$T$的周期函数,则$f'(x)$也是以$T$为周期的周期函数。
|
||
|
||
已知$f(x+T)=f(x)$,求证$f'(x+T)=f'(x)$。
|
||
|
||
$\therefore f'(x+T)=\lim_{\Delta x\to 0}\dfrac{f(x+T+\Delta x)-f(x+T)}{\Delta x}=\lim_{\Delta x\to 0}\dfrac{f(x+\Delta x)-f(x)}{\Delta x}=f'(x)$。
|
||
|
||
\textbf{例题3:}设$f(x)$是二阶可导的以2为周期的奇函数,且$f(\dfrac{1}{2})>0$,$f'(\dfrac{1}{2})>0$,比较$f(-\dfrac{1}{2})$、$f'(\dfrac{3}{2})$、$f''(0)$的大小。
|
||
|
||
$\because f(x)$为二阶奇函数,$\therefore f(x)\text{奇函数}\Rightarrow f'(x)\text{偶函数}\Rightarrow f''(x)\text{奇函数}\Rightarrow f''(0)=0$。
|
||
|
||
$\therefore f(-\dfrac{1}{2})=-f(\dfrac{1}{2})<0$。
|
||
|
||
$\because f(x)T=2\Rightarrow f'(x)T=2$,$\therefore f'(\dfrac{3}{2})=f'(\dfrac{3}{2}-2)=f'(-\dfrac{1}{2})=f'(\dfrac{1}{2})>0$。
|
||
|
||
$\therefore f'(\dfrac{3}{2})>f''(0)>f(-\dfrac{1}{2})$。
|
||
|
||
\textbf{例题4:}证明$(uv)'=u'v+uv'$。
|
||
|
||
令$f(x)=u(x)v(x)$。
|
||
|
||
$
|
||
\begin{aligned}
|
||
& (u\cdot v)' \\
|
||
& =f'(x) \\
|
||
& =\lim_{\Delta x\to 0}\dfrac{f(x+\Delta x)-f(x)}{\Delta x} \\
|
||
& =\lim_{\Delta x\to 0}\dfrac{u(x+\Delta x)v(x+\Delta x)-u(x)v(x)}{\Delta x} \\
|
||
& =\lim_{\Delta x\to 0}\dfrac{u(x+\Delta x)v(x+\Delta x)-u(x)v(x+\Delta x)+u(x)v(x+\Delta x)-u(x)v(x)}{\Delta x} \\
|
||
& =\lim_{\Delta x\to 0}\dfrac{u(x+\Delta x)-u(x)}{\Delta x}v(x+\Delta x) +\lim_{\Delta x\to 0}\dfrac{v(x+\Delta x)-v(x)}{\Delta x}u(x) \\
|
||
& =u'(x)v(x)+v'(x)u(x)
|
||
\end{aligned}
|
||
$
|
||
|
||
\textbf{例题5:}证明可导必连续。
|
||
|
||
已知连续定义:$\lim_{\Delta x\to 0}f(x+\Delta x)=f(x)$,即$\lim_{\Delta x\to 0}f(x+\Delta x)-f(x)=0$。
|
||
|
||
可导定义:$f'(x)=\lim_{\Delta x\to 0}\dfrac{f(x+\Delta x)-f(x)}{\Delta x} = A$
|
||
|
||
$
|
||
\begin{aligned}
|
||
& \lim_{\Delta x\to 0}f(x+\Delta x)-f(x) \\
|
||
& =\lim_{\Delta x\to 0}\dfrac{f(x+\Delta x)-f(x)}{\Delta x}\cdot\Delta x \\
|
||
& =A\cdot 0 \\
|
||
& =0
|
||
\end{aligned}
|
||
$
|
||
|
||
\textbf{例题6:}若$f(x)$在$x=x_0$处连续,且$\lim_{x\to x_0}\dfrac{f(x)}{x-x_0}=A$,则$f(x_0)=0$且$f'(x_0)=A$。
|
||
|
||
证明:$\because\text{连续,}\therefore f(x_0)=\lim_{x\to x_0}f(x)=\lim_{x\to x_0}\dfrac{f(x)}{x-x_0}(x-x_0)=A\cdot 0=0$。
|
||
|
||
又$f'(x_0)=\lim_{x\to x_0\dfrac{f(x)-f(x_0)}{x-x_0}}=\lim_{x\to x_0}\dfrac{f(x)}{x-x_0}=A$。
|
||
|
||
如$\lim_{x\to 1}\dfrac{f(x)}{x-1}=2$且$f(x)$连续,可以推出$f(1)=0$与$f'(1)=2$。
|
||
|
||
高阶导数\textcolor{violet}{\textbf{定义:}}$f^n(x_0)=\lim_{\Delta x\to 0}\dfrac{f^{n-1}(x_0+\Delta x)-f^{n-1}(x_0)}{\Delta x}$,其中$n\geqslant 2$且$n\in N^+$,$f^{n-1}(x)$在$x_0$的某领域内有定义,$x_0+\Delta x$也在该邻域内。
|
||
|
||
\subsection{微分}
|
||
|
||
有一个边长为$x$的正方形,变化了$\Delta x$,其面积$\Delta S=(x+\Delta x)^2-x^2=2x\Delta x+(\Delta x)^2$。
|
||
|
||
当$\Delta x\to 0$时,将这个变化定义为$2x\cdot\Delta x+o(\Delta x)$,前项为线性主部,后面为误差。这个就是$S$的微分。
|
||
|
||
增量$\Delta y=f(x_0+\Delta)-f(x_0)=A\Delta x+o(\Delta x)$,这个$A\Delta x$定义为$\rm{d}y$,叫做$y$的微分。
|
||
|
||
$\therefore \rm{d}y\vert_{x=x_0}=A\Delta x=y'(x_0)\cdot\Delta x=y'(x_0)\cdot\rm{d}x$
|
||
|
||
由此,可导必可微,可微必可导。
|
||
|
||
\begin{tikzpicture}[scale=0.9]
|
||
\draw[-latex](-0.5,0) -- (4.5,0) node[below]{$x$};
|
||
\draw[-latex](0,-0.5) -- (0,4) node[above]{$y$};
|
||
\draw[black, thick, domain=1.5:3] plot (\x,{pow(\x-1,2)/2+1}) node[above]{$y(x)$};
|
||
\filldraw[black] (0,0) node[below]{$O$};
|
||
\draw[black, densely dashed](1.5,1.125) -- (1.5,0) node[below]{$x_0$};
|
||
\draw[black, densely dashed](1.5,1.125) -- (0,1.125) node[left]{$y_0$};
|
||
\draw[black, densely dashed](3,3) -- (3,0) node[below]{$x_0+\Delta x$};
|
||
\draw[black, densely dashed](3,3) -- (0,3) node[left]{$y_0+\Delta x$};
|
||
\draw[black, densely dashed](3,1.875) -- (0,0.375) node[left]{$\rm{d}yx+b$};
|
||
\draw[<->, black](1.5,1.125) -- (3,1.125);
|
||
\draw[<->, black](4,1.125) -- (4,3);
|
||
\draw[<->, black](3.25,1.125) -- (3.25,1.875);
|
||
\draw[<->, black](3.25,3) -- (3.25,1.875);
|
||
\draw[black](3,3) -- (4.5,3);
|
||
\draw[black](3,1.125) -- (4.5,1.125);
|
||
\draw[black](3,1.875) -- (3.75,1.875);
|
||
\filldraw[black] (2.25,0.75) node{$\Delta x$};
|
||
\filldraw[black] (4.3,2) node{$\Delta y$};
|
||
\filldraw[black] (3.5,1.5) node{\scriptsize{$\rm{d}y$}};
|
||
\filldraw[black] (3.5,2.5) node{\scriptsize{$o(\Delta x)$}};
|
||
\end{tikzpicture}
|
||
|
||
所以可微就是用简单线性取代复杂线性,如图用直线取替代曲线。
|
||
|
||
\section{导数与微分计算}
|
||
\subsection{四则运算}
|
||
|
||
若函数可导:
|
||
|
||
\begin{enumerate}
|
||
\item 和差的导数或微分:$[u(x)\pm v(x)]'=u'(x)\pm v'(x)$,$\rm{d}[u(x)\pm v(x)]=\rm{d}u(x)\pm\rm{d}v(x)$。
|
||
\item 积的导数或微分:$[u(x)v(x)]'=u'(x)v(x)+u(x)v'(x)$,$\rm{d}[u(x)v(x)]=u(x)\rm{d}v(x)+v(x)\rm{d}u(x)$,$[u(x)v(x)w(x)]'=u'(x)v(x)w(x)+u(x)v'(x)w(x)+u(x)v(x)+w'(x)$。
|
||
\item 商的导数:$\left[\dfrac{u(x)}{v(x)}\right]'=\dfrac{u'(x)v(x)-u(x)v'(x)}{[v(x)]^2}$,$\rm{d}\left[\dfrac{u(x)}{v(x)}\right]=\dfrac{v(x)\rm{d}u(x)-u(x)\rm{d}v(x)}{[v(x)]^2}$,$v(x)\neq 0$。
|
||
\item 复合函数的导数:链式求导法则$\dfrac{\rm{d}u}{\rm{d}x}=\dfrac{\rm{d}u}{\rm{d}y}\cdot\dfrac{\rm{d}y}{\rm{d}x}$。
|
||
\end{enumerate}
|
||
|
||
\subsection{分段函数的导数}
|
||
|
||
设$f(x)=\left\{
|
||
\begin{array}{lcl}
|
||
f_1(x), & & x\geqslant x_0 \\
|
||
f_2(x), & & x<x_0 \\
|
||
\end{array}
|
||
\right.$。
|
||
|
||
在分段点用定义:判断$f'_+(x_0)=\lim_{x\to x_0^+}\dfrac{f_1(x)-f(x_0)}{x-x_0}\overset{?}{=}\lim_{x\to x_0^-}\dfrac{f_2(x)-f(x_0)}{x-x_0}$。
|
||
|
||
非分段点使用导数公式求导:$x>x_0,f'(x)=f_1'(x),x<0,f'(x)=f_2'(x)$。
|
||
|
||
\subsection{复合函数的导数与微分形式不变性}
|
||
|
||
$u=g(x)$在$x$可导,$y=f(u)$在$u=g(x)$处可导,则$\{f[g(x)]\}'=f'[g(x)]g'(x)$,$\rm{d}\{f[g(x)]\}=f'[g(x)]g'(x)\rm{d}x$。
|
||
|
||
一阶微分形式不变性指:$\rm{d}f(\varsigma)=f'(\varsigma)\rm{d}\varsigma$,无论$\varsigma$是什么。(类似导数的链式求导法则)
|
||
|
||
\textbf{例题7:}设$f(x)=\Pi_{n=1}^{100}\left(\tan\dfrac{\pi x^a}{4}-n\right)$,则$f'(1)$为?
|
||
|
||
原式=$\left(\tan\dfrac{\pi x}{4}-1\right)\left(\tan\dfrac{\pi x^2}{4}-2\right)\cdots\left(\tan\dfrac{\pi x^100}{4}-100\right)$。
|
||
|
||
令$\left(\tan\dfrac{\pi x^2}{4}-2\right)\cdots\left(\tan\dfrac{\pi x^100}{4}-100\right)=g(x)$。
|
||
|
||
$\therefore f(x)=\left(\tan\dfrac{\pi x}{4}-1\right)\cdot g(x)$。
|
||
|
||
$\therefore f'(x)=\sec^2\dfrac{\pi x}{4}\cdot\dfrac{\pi}{4}\cdot g(x)+\left(\tan\dfrac{\pi x}{4}-1\right)\cdot g'(x)$。
|
||
|
||
$\therefore$根据导数的四则运算,需要导数的乘积为每一项求导乘以其他不求导项的和,而$\tan\dfrac{\pi x}{4}-1$当$x=1$时为0,只要它不求导,其他的项都必然是0,所以原式的后面的结果都是0。
|
||
|
||
$\therefore$
|
||
|
||
$
|
||
\begin{aligned}
|
||
f'(1) & =f'(x)\vert_{x=1} \\
|
||
& =\dfrac{\pi}{2}\cdot g(1)+0\cdot g'(x) \\
|
||
& =\dfrac{\pi}{2}\cdot g(1) \\
|
||
& =\dfrac{\pi}{2}(-1)(-2)\cdots(-99) \\
|
||
& =-\dfrac{\pi}{2}\cdot 99!
|
||
\end{aligned}
|
||
$
|
||
|
||
\textbf{例题8:}设$y=e^{\sin(\ln x)}$,求$\rm{d}y$。
|
||
|
||
$\because y=e^{\sin(\ln x)} \therefore$
|
||
|
||
$
|
||
\begin{aligned}
|
||
\rm{d}y &=\rm{d}e^{\sin(\ln x)} \\
|
||
& =e^{\sin(\ln x)}\cdot\rm{d}(\sin(\ln x)) \\
|
||
& =e^{\sin(\ln x)}\cdot\cos(\ln x)\cdot\rm{d}\ln x \\
|
||
& =e^{\sin(\ln x)}\cdot\cos(\ln x)\cdot\dfrac{1}{x}\rm{d}x
|
||
\end{aligned}
|
||
$
|
||
|
||
\subsection{反函数导数}
|
||
|
||
\textcolor{aqua}{\textbf{定理:}}$y=f(x)$可导,且$f'(x)\neq 0$,则存在反函数$x=\varphi(y)$,且$\dfrac{\rm{d}x}{\rm{d}y}=\dfrac{1}{\dfrac{\rm{d}y}{\rm{d}x}}$,即$\varphi'(x)=\dfrac{1}{f'(x)}$。
|
||
|
||
$y=f(x)$可导,且$f'(x)\neq 0$就是指严格单调,而严格单调必有反函数。
|
||
|
||
\textbf{例题9:}求$y=\arcsin x,x\in(-1,1)$与$y=\arctan x$的导数。
|
||
|
||
首先反三角函数就是三角函数的反函数、
|
||
|
||
求$y=\arcsin x$,即$x=\sin y$。
|
||
|
||
$\therefore\dfrac{\rm{d}\arcsin x}{\rm{d}x}=\dfrac{1}{\dfrac{\rm{d}\sin y}{\rm{d}y}}=\dfrac{1}{\cos y}=\dfrac{1}{\sqrt{1-\sin^2y}}=\dfrac{1}{\sqrt{1-x^2}}$。
|
||
|
||
求$y=\arctan x$,就$x=\tan y$。
|
||
|
||
$\therefore\dfrac{\rm{d}\arctan x}{\rm{d}x}=\dfrac{1}{\dfrac{\rm{d}\tan y}{\rm{d}y}}=\dfrac{1}{\sec^2y}=\dfrac{1}{1+\tan^2y}=\dfrac{1}{1+x^2}$。
|
||
|
||
二阶反函数导数\textcolor{aqua}{\textbf{定理:}}:
|
||
|
||
$
|
||
\begin{aligned}
|
||
f''(x) &=y''_{xx} \\
|
||
& =\dfrac{\rm{d}\left(\dfrac{\rm{d}y}{\rm{d}x}\right)}{\rm{d}x} \\
|
||
& =\dfrac{\rm{d}^2y}{\rm{d}x^2} \\
|
||
& =\dfrac{\rm{d}\left(\dfrac{1}{\varphi'(y)}\right)}{\rm{d}x} \\
|
||
& =\dfrac{\rm{d}\left(\dfrac{1}{\varphi'(y)}\right)}{\rm{d}y}\cdot\dfrac{\rm{d}y}{\rm{d}x} \\
|
||
& =-\dfrac{x_{yy}''}{(x_y')^2}\cdot\dfrac{1}{x_y'} \\
|
||
& =-\dfrac{x_{yy}''}{(x_y')^3}
|
||
\end{aligned}
|
||
$
|
||
|
||
其中$\rm{d}x\cdot\rm{d}x=(\rm{d}x)^2=\rm{d}x^2$称为微分的幂,而$\rm{d}(x^2)$叫幂的微分。
|
||
|
||
\textbf{例题10:}设$y=f(x)$的反函数是$x=\varphi(y)$,且$f(x)=\int_1^{2x}e^{t^2}\rm{d}t+1$,求$\varphi''(1)$。
|
||
|
||
$\because y=f(x)$,$\therefore x=\varphi(y)$,$x_{yy}''=\varphi''(y)=-\dfrac{y_{xx}''}{(y_x')^3}=-\dfrac{f''(x)}{[f'(x)]^3}$。
|
||
|
||
其中根据变限积分求导公式:$f'(x)=2e^{4x^2}$,$f''(x)=2e^{4x^2}\cdot 8x=16xe^{4x^2}$
|
||
|
||
又$y=1\Rightarrow x=\dfrac{1}{2}\Rightarrow\varphi''(1)=-\dfrac{f''\left(\dfrac{1}{2}\right)}{\left[f'\left(\dfrac{1}{2}\right)\right]^3}=-\dfrac{1}{e^2}$。
|
||
|
||
\subsection{参数方程函数导数}
|
||
|
||
设函数$y=y(x)$由参数方程$\left\{
|
||
\begin{array}{l}
|
||
x=\varphi(t) \\
|
||
y=\psi(t)
|
||
\end{array}
|
||
\}\right.$确定,其中$t$为参数,且$\varphi(t)\psi(t)$对于$t$都可导,$\varphi(t)\neq 0$,则:
|
||
|
||
\bigskip
|
||
|
||
一阶导数:$\dfrac{\rm{d}y}{\rm{d}x}=\dfrac{\rm{d}y/\rm{d}t}{\rm{d}x/\rm{d}t}=\dfrac{\psi'(t)}{\varphi'(t)}=u(t)$。
|
||
|
||
二阶导数:$\dfrac{\rm{d}^2y}{\rm{d}x^2}=\dfrac{\rm{d}\left(\dfrac{\rm{d}y}{\rm{d}x}\right)}{\rm{d}x}=\dfrac{\rm{d}\left(\dfrac{\rm{d}y}{\rm{d}x}\right)/\rm{d}t}{\rm{d}x/\rm{d}t}=\dfrac{\rm{d}u/\rm{d}t}{\rm{d}x/\rm{d}t}=\dfrac{u'_t}{x'_t}$
|
||
|
||
\textbf{例题11:}设$y=y(x)$由方程$\left\{
|
||
\begin{array}{l}
|
||
x=\sin t \\
|
||
y=t\sin t+\cos t
|
||
\end{array}
|
||
\right.
|
||
$($t$为参数)确定,求$\dfrac{\rm{d}^2y}{\rm{d}x^2}\vert_{t=\frac{\pi}{4}}$。
|
||
|
||
求参数方程的二阶导数首先就要求出其一阶导数:
|
||
|
||
$\dfrac{\rm{d}y}{\rm{d}x}=\dfrac{y_t'}{x_t'}=\dfrac{t\cos t}{\cos t}=t$。
|
||
|
||
$\therefore\dfrac{\rm{d}^2y}{\rm{d}x^2}=\dfrac{\rm{d}\left(\dfrac{\rm{d}y}{\rm{d}x}\right)}{\rm{d}x}=\dfrac{t_t'}{(\sin t)_t'}=\dfrac{1}{\cos t}$
|
||
|
||
$\therefore \sqrt{2}$。
|
||
|
||
\subsection{隐函数求导法}
|
||
|
||
设函数$y=y(x)$由方程$F(x,y)=0$确定的可导函数,则\ding{172}方程两边对自变量$x$求导,($y=y(x)$就是将$y$看作中间变量)得到一个关于$y'$的方程。\ding{173}解该方程就可以得出$y'$。
|
||
|
||
\textbf{例题12:}设$y=y(x)$是由方程$\sin(xy)=\ln\dfrac{x+e}{y}+1$确定的隐函数,求$y'(0)$。
|
||
|
||
两边求导:
|
||
|
||
$
|
||
\begin{aligned}
|
||
\sin(xy) &=\ln(x+e)-\ln(y)+1 \\
|
||
\cos(xy)(y+xy') &=\dfrac{1}{x+e}-\dfrac{y'}{y} \\
|
||
\because\text{将0代入} & x=0, y=e^2 \\
|
||
e^2&=\dfrac{1}{e}-\dfrac{y'(0)}{e^2} \\
|
||
y'(0) & =e-e^4
|
||
\end{aligned}
|
||
$
|
||
|
||
\subsection{对数求导法}
|
||
|
||
对于多项相乘、相除、开方、乘方的式子,一般先取对数再求导,设$y=f(x)(f(x)>0)$,则\ding{172}等式两边取对数:$\ln y=\ln f(x)$。\ding{173}两边对自变量$x$求导,得$\dfrac{1}{y}y'=[\ln f(x)]'\Rightarrow y'=y[\ln f(x)]'$。
|
||
|
||
\textbf{例题13:}求$y=\sqrt[3]{\dfrac{(x+1)(2x-1)^2}{(4-3x)^5}}$的导数。
|
||
|
||
取对数:$\ln\vert y\vert=\dfrac{1}{3}[\ln\vert x+1\vert+2\ln\vert 2x-1\vert-5\ln\vert 4-3x\vert]$。
|
||
|
||
$\because \ln\vert y\vert'=\ln y'$。
|
||
|
||
两边对x求导:
|
||
|
||
$\dfrac{y'}{y}=\dfrac{1}{3}\left(\dfrac{1}{x+1}+\dfrac{4}{2x-1}-\dfrac{5}{4-3x}\cdot(-3)\right)$
|
||
|
||
$\therefore y'=\dfrac{1}{3}\left(\dfrac{1}{x+1}+\dfrac{4}{2x-1}-\dfrac{5}{4-3x}\cdot(-3)\right)y$
|
||
|
||
\subsection{幂指函数求导法}
|
||
|
||
非常重要。
|
||
|
||
对于$u(x)^{v(x)}(u(x)>0,u(x)\neq 1)$除了对数求导法外还可以使用指数函数$u(x)^{v(x)}=e^{v(x)\ln u(x)}$,然后求导得到$[u(x)^{v(x)}]'=[e^{v(x)\ln u(x)}]'=u(x)^{v(x)}\left[v'(x)\ln u(x)+v(x)\cdot\dfrac{u'(x)}{u(x)}\right]$。
|
||
|
||
\textbf{例题14:}求$y=x^x(x>0)$的导数。
|
||
|
||
$\because x^x=e^{x\ln x}$,$\therefore (x^x)'=(e^{x\ln x})'=x^x\cdot(\ln x+1)$。
|
||
|
||
\textbf{例题15:}求解$y=x^{\frac{1}{x}}(x>0)$的整数最大值。
|
||
|
||
$\because y=x^{\frac{1}{x}}=e^{\frac{1}{x}\ln x}$。
|
||
|
||
$\therefore y'=\left(x^{\frac{1}{x}}\right)=\left(e^{\frac{1}{x}\ln x}\right)'=x^{\frac{1}{x}}\cdot\dfrac{1-\ln x}{x^2}$。
|
||
|
||
令导数结果为0,因为$x^{\frac{1}{x}}$与$x^2$在$x>0$时都不为0,所以只有一个驻点$x=e$。
|
||
|
||
$0<x<e$时$1-\ln x$大于0,所以导数大于0,函数在该区间增。相反$x>e$时函数在区间减。
|
||
|
||
研究驻点左侧情况,求对应的极限:$e^{\lim_{x\to 0^+}\frac{\ln x}{x}}=e^{-\infty}\to 0$。
|
||
|
||
研究驻点右侧情况,求对应的极限:$e^{\lim_{x\to+\infty}\frac{\ln x}{x}}=e^0\to 1$。
|
||
|
||
\begin{tikzpicture}[scale=0.5]
|
||
\draw[-latex](-0.5,0) -- (10,0) node[below]{$x$};
|
||
\draw[-latex](0,-0.5) -- (0,3) node[above]{$y$};
|
||
\draw[black, thick, domain=0.1:10] plot (\x,{pow(\x,pow(\x,-1))});
|
||
\filldraw[black] (8,2.5) node{$y=x^{\frac{1}{x}}$};
|
||
\filldraw[white, draw=black, line width=1pt] (0,0) circle (4pt);
|
||
\filldraw[black] (0,0) node[below]{$O$};
|
||
\filldraw[black] (e,1.5) circle (4pt);
|
||
\filldraw[black] (e,1.5) node[above]{$(e,\sqrt[e]{e})$};
|
||
\draw[black, densely dashed](e,1.5) -- (e,0) node[below]{$e$};
|
||
\draw[black, densely dashed](10,1.5) -- (0,1.5) node[left]{$\sqrt[e]{e})$};
|
||
\end{tikzpicture}
|
||
|
||
所以必然在$\sqrt{2}$与$\sqrt[3]{3}$两点取得整数最大值,而全部六次方后$\sqrt{2}^6=8<\sqrt[3]{3}=9$,所以$\sqrt[3]{3}$为最大整数解。
|
||
|
||
\subsection{高阶导数}
|
||
|
||
即导数阶数在2以及以上的导数计算。
|
||
|
||
\subsubsection{归纳法}
|
||
|
||
即依次求导得出规律。
|
||
|
||
$(a^x)^n=a^x(\ln a)^{(n)}$,如$y=2^x$,则$y'=2^x\ln 2$,$y''=2^x(\ln 2)^2\cdots$得到$y^{(n)}=2^x(\ln 2)^n,n\in N$。
|
||
|
||
\textbf{例题16:}求$\sin x$的$n$阶导数。
|
||
|
||
$\because \sin x'=\cos x$而不断求导会发现正负号会++--++--地变化而难以归纳为公式,所以需要另想办法。
|
||
|
||
使用诱导公式:
|
||
|
||
$
|
||
\begin{aligned}
|
||
y'& =\cos x=\sin(x+\dfrac{\pi}{2}) \\
|
||
y''& =\cos(x+\dfrac{\pi}{2})=\sin(x+\dfrac{\pi}{2}+\dfrac{\pi}{2}) \\
|
||
\cdots & \\
|
||
y^{(n)}& =\sin(x+\dfrac{\pi}{2}\cdot n)
|
||
\end{aligned}
|
||
$
|
||
|
||
\subsubsection{莱布尼茨公式}
|
||
|
||
设$u=u(x)$,$v=v(x)$均$n$阶可导,则$(uv)^{(n)}=\sum_{k=0}^nC_n^ku^{(n-k)}v^{(k)}$。
|
||
|
||
展开:$(uv)^{(n)}=C_n^0u^{(n)}v^{(0)}+C_n^1u^{(n-1)}v'+\cdots+C_n^nu^{(0)}v^{(n)}$。
|
||
|
||
莱布尼兹公式里的系数与考研数学准备章节的因式分解公式的二次项公式的系数一致,可以使用杨辉三角形来记忆:
|
||
|
||
\begin{tikzpicture}[scale=0.9]
|
||
\node[black] at (0,0) {$C_0^0$};
|
||
\node[black] at (-1,-1) {$C_1^0$};
|
||
\node[black] at (0,-1) {$C_1^1$};
|
||
\node[black] at (-2,-2) {$C_2^0$};
|
||
\node[black] at (-1,-2) {$C_2^1$};
|
||
\node[black] at (-0,-2) {$C_2^2$};
|
||
\node[black] at (-3,-3) {$C_3^0$};
|
||
\node[black] at (-2,-3) {$C_3^1$};
|
||
\node[black] at (-1,-3) {$C_3^2$};
|
||
\node[black] at (-0,-3) {$C_3^3$};
|
||
\node[black] at (-4,-4) {$C_4^0$};
|
||
\node[black] at (-3,-4) {$C_4^1$};
|
||
\node[black] at (-2,-4) {$C_4^2$};
|
||
\node[black] at (-1,-4) {$C_4^3$};
|
||
\node[black] at (-0,-4) {$C_4^4$};
|
||
\end{tikzpicture}
|
||
\hspace{2.5em}
|
||
\begin{tikzpicture}[scale=0.9]
|
||
\node[black] (0) at (0,0) {1};
|
||
\node[black] (1) at (-1,-1) {1};
|
||
\node[black] (2) at (1,-1) {1};
|
||
\node[black] (3) at (-2,-2) {1};
|
||
\node[black] (4) at (0,-2) {2};
|
||
\node[black] (5) at (2,-2) {1};
|
||
\node[black] (6) at (-3,-3) {1};
|
||
\node[black] (7) at (-1,-3) {3};
|
||
\node[black] (8) at (1,-3) {3};
|
||
\node[black] (9) at (3,-3) {1};
|
||
\node[black] (10) at (-4,-4) {1};
|
||
\node[black] (11) at (-2,-4) {4};
|
||
\node[black] (12) at (0,-4) {6};
|
||
\node[black] (13) at (2,-4) {4};
|
||
\node[black] (14) at (4,-4) {1};
|
||
\draw[-,thick] (0) to (1);
|
||
\draw[-,thick] (0) to (2);
|
||
\draw[-,thick] (1) to (3);
|
||
\draw[-,thick] (1) to (4);
|
||
\draw[-,thick] (2) to (4);
|
||
\draw[-,thick] (2) to (5);
|
||
\draw[-,thick] (3) to (6);
|
||
\draw[-,thick] (3) to (7);
|
||
\draw[-,thick] (4) to (7);
|
||
\draw[-,thick] (4) to (8);
|
||
\draw[-,thick] (5) to (8);
|
||
\draw[-,thick] (5) to (9);
|
||
\draw[-,thick] (6) to (10);
|
||
\draw[-,thick] (6) to (11);
|
||
\draw[-,thick] (7) to (11);
|
||
\draw[-,thick] (7) to (12);
|
||
\draw[-,thick] (8) to (12);
|
||
\draw[-,thick] (8) to (13);
|
||
\draw[-,thick] (9) to (13);
|
||
\draw[-,thick] (9) to (14);
|
||
\end{tikzpicture}
|
||
|
||
\textbf{例题17:}已知函数$y=e^x\cos x$,求$y^{(4)}$。
|
||
|
||
根据莱布尼兹公式:
|
||
|
||
$
|
||
\begin{aligned}
|
||
& (e^x\cos x)^{(4)} \\
|
||
& =C_4^0e^x\cos x+C_4^1e^x(-\sin x)+C_4^2e^x(-\cos x)+C_4^3e^x(\sin x)+C_4^4e^x(\cos x) \\
|
||
& =e^x\cos x+4e^x(-\sin x)+6e^x(-\cos x)+4e^x\sin x+e^x\cos x \\
|
||
& =-4e^x\cos x
|
||
\end{aligned}
|
||
$
|
||
|
||
\subsubsection{泰勒公式}
|
||
|
||
先写出$y=f(x)$的泰勒公式或麦克劳林公式,再通过比较系数来获得$f^{(n)}(x_0)$:
|
||
|
||
\begin{enumerate}
|
||
\item 任何一个无穷阶可导的函数(在收敛的情况下)都可以写为 \\
|
||
$y=f(x)=\sum_{n=0}^\infty\dfrac{f^{(n)}(x_0)}{n!}(x-x_0)^n$ 或 $y=f(x)=\sum_{n=0}^\infty\dfrac{f^{(n)}(0)}{n!}x^n$。
|
||
\item 给出的任意一个具体的无穷阶可导函数$y=f(x)$都可以通过已知的公式展开为幂级数。
|
||
\item 而函数的展开式具有唯一性,比较步骤一步骤二的公式的系数就可以获取倒$f^{(n)}(x_0)$或$f^{(n)}(0)$。
|
||
\end{enumerate}
|
||
|
||
\textbf{例题18:}设$y=x^3\sin x$,求$y^{(6)}(0)$。
|
||
|
||
\ding{172}$y=\sum_{n=0}^\infty\dfrac{y^{(n)}(0)}{n!}x^n$。
|
||
|
||
$\because$需要结果的导数阶数为6,所以最后得到的次数为6就可以了。
|
||
|
||
\ding{173}$\therefore y=x^3\left(x-\dfrac{1}{6}x^3+\cdots\right)=x^4-\dfrac{1}{6}x^6+\cdots$(不要写$o(x^n)$,因为这里$x$并不是趋向0的)。
|
||
|
||
\ding{174}步骤一的抽象函数当$n=6$时为$\dfrac{y^{(6)}(0)}{6!}x^6$,它应该与步骤二得到的$x^4-\dfrac{1}{6}x^6+\cdots$的6阶项的系数相等。
|
||
|
||
$\therefore \dfrac{y^{(6)}(0)}{6!}=-\dfrac{1}{6}\Rightarrow y^{(6)}(0)=-5!=-120$。
|
||
|
||
\subsection{变限积分求导公式}
|
||
|
||
必然会考。
|
||
|
||
已知更改区间限制的积分$s(x)=\int_{\varphi_1(x)}^{\varphi_2(x)}g(t)\rm{d}x$,$s'(x)=g[\varphi_2(x)]\cdot\varphi_2'(x)-g[\varphi_1(x)]\cdot\varphi_1'(x)$。
|
||
|
||
\subsection{基本求导公式}
|
||
|
||
\subsubsection{对幂指函数}
|
||
|
||
\begin{center}
|
||
\begin{tabular}{|c|c|c|c|}
|
||
\hline
|
||
原函数 & 导函数 & 原函数 & 导函数\\ \hline
|
||
$C$ & $0$ & $n^x$ & $n^x\ln n$ \\ \hline
|
||
$\log_ax$ & $\dfrac{1}{x\ln a}$ & $\ln x=\ln\vert x\vert$ & $\dfrac{1}{x}$ \\ \hline
|
||
$x^n$ & $nx^{n-1}$ & $\sqrt[n]{x}$ & $\dfrac{x^{-\frac{n-1}{n}}}{n}$ \\ \hline
|
||
$\dfrac{1}{x^n}$ & $-\dfrac{n}{x^{n+1}}$ & & \\
|
||
\hline
|
||
\end{tabular}
|
||
\end{center}
|
||
|
||
\subsubsection{三角与反三角函数}
|
||
|
||
\begin{center}
|
||
\begin{tabular}{|c|c|c|c|}
|
||
\hline
|
||
原函数 & 导函数 & 原函数 & 导函数\\ \hline
|
||
$\sin x$ & $\cos x$ & $\cos x$ & $-\sin x$ \\ \hline
|
||
$\tan x$ & $\dfrac{1}{\cos^2x}=\sec^2x$ & $\cot x$ & $\dfrac{1}{\sin^2x}=\csc^2x$ \\ \hline
|
||
$\sec x$ & $\sec x\tan x$ & $\csc x$ & $-\csc x\cot x$ \\ \hline
|
||
$\arcsin x$ & $\dfrac{1}{1-x^2}$ & $\arccos x$ & $-\dfrac{1}{1-x^2}$ \\ \hline
|
||
$\arctan x$ & $\dfrac{1}{1+x^2}$ & $\rm{arccot}\,\textit{x}$ & $-\dfrac{1}{1+x^2}$ \\ \hline
|
||
$\rm{arcsec}\,\textit{x}$ & $\dfrac{1}{x\sqrt{x^2-1}}$ & $\rm{arccsc}\,\textit{x}$ & $-\dfrac{1}{x\sqrt{x^2-1}}$ \\ \hline
|
||
\hline
|
||
\end{tabular}
|
||
\end{center}
|
||
|
||
\subsubsection{双曲与反双曲函数}
|
||
|
||
\begin{itemize}
|
||
\item 双曲正弦:$\rm{sinh}\,\textit{x}=\rm{sh}\,\textit{x}=\dfrac{e^x-e^{-x}}{2}$。
|
||
\item 双曲余弦:$\rm{cosh}\,\textit{x}=\rm{ch}\,\textit{x}=\dfrac{e^x+e^{-x}}{2}$。
|
||
\item 双曲正切:$\rm{tanh}\,\textit{x}=\rm{th}\,\textit{x}=\dfrac{\rm{sinh}\,\textit{x}}{\rm{cosh}\,\textit{x}}=\dfrac{e^x-e^{-x}}{e^x+e^{-x}}$。
|
||
\item 双曲余切:$\rm{coth}\,\textit{x}=\dfrac{\rm{cosh}\,\textit{x}}{\rm{sinh}\,\textit{x}}=\dfrac{e^x+e^{-x}}{e^x-e^{-x}}$。
|
||
\item 双曲正割:$\rm{sech}\,\textit{x}=\dfrac{1}{\rm{cosh}\,\textit{x}}=\dfrac{2}{e^x+e^{-x}}$。
|
||
\item 双曲余割:$\rm{csch}\,\textit{x}=\dfrac{1}{\rm{sinh}\,\textit{x}}=\dfrac{2}{e^x-e^{-x}}$。
|
||
\item 反双曲正弦:$\rm{arcsinh}\,\textit{x}=\ln\left(x+\sqrt{x^2+1}\right)$。
|
||
\item 反双曲余弦:$\rm{arccosh}\,\textit{x}=\ln\left(x+\sqrt{x^2-1}\right)$。
|
||
\item 反双曲正切:$\rm{arctanh}\,\textit{x}=\dfrac{1}{2}\ln\left(\dfrac{1+x}{1-x}\right)$。
|
||
\end{itemize}
|
||
|
||
\begin{center}
|
||
\begin{tabular}{|c|c|c|c|}
|
||
\hline
|
||
原函数 & 导函数 & 原函数 & 导函数\\ \hline
|
||
$\rm{sinh}\,\textit{x}$ & $\rm{cosh}\,\textit{x}$ & $\rm{cosh}\,\textit{x}$ & $\rm{sinh}\,\textit{x}$ \\ \hline
|
||
$\rm{tanh}\,\textit{x}$ & $\dfrac{1}{\rm{cosh}\,\textit{x}^2}$ & $\rm{arcsinh}\,\textit{x}$ & $\dfrac{1}{\sqrt{x^2+1}}$ \\ \hline
|
||
$\rm{arccosh}\,\textit{x}$ & $\dfrac{1}{\sqrt{x^2-1}}$ & $\rm{arctan}\,\textit{x}$ & $\dfrac{1}{1-x^2}$ \\
|
||
\hline
|
||
\end{tabular}
|
||
\end{center}
|
||
|
||
\end{document}
|