高数笔记 P02:一元函数微分学
概念:导数、微分\(dx,dy\)、高阶导数
1 导数
定义
\(\displaystyle \lim_{\Delta x \to 0} \cfrac {f(x_0 + \Delta x) - f(x_0)}{\Delta x} = f'(x_0) \ \iff\)
\(\displaystyle \lim_{x \to x_0} \cfrac {f(x) - f(x_0)}{x - x_0} = f'(x_0)\)
上述两个定义都是导数的定义,其中的变量满足一动一静。
\(f'(x)\)存在 \(\iff f'_+(x) = f'_-(x)\)
若\(f(x)\)是可导的偶函数,则\(f'(x)\)为奇函数;若\(f(x)\)为可导的奇函数,则\(f'(x)\)为可导的偶函数。
性质
- 若\(f(x)\)在\(x\)处可导,则\(f(x)\)在此处连续。
- \(dy = f'(x)dx\).
- \(\Delta y = dy + \omicron (\Delta x) = f'(x_0) + \omicron (\Delta x) \Rightarrow \Delta y - dy = \frac 12 f''(\xi)(\Delta x)^2\)
计算
-
\((uv)' = u'v + v'u\)
-
\((\cfrac uv)' = \cfrac {u'v -v'u}{v^2},(v \neq 0)\)
-
\((u \pm v)^{(n)} = u^{(n)} \pm v^{(n)}\)
-
莱布尼茨公式:\((uv)^{(n)} = u^{(n)} v + \mathrm{C}_n^1 u^{(n-1)} v' + \cdots + \mathrm{C}_n^k u^{(n-k)} v^{(k)} + \cdots + u v^{(n)}\)
-
\(\displaystyle (\int_{\varphi_1 (x)}^{\varphi_2 (x)}f(t)dt)' = f(\varphi_2 (x)) \varphi_2' (x) - f(\varphi_1 (x)) \varphi_1' (x)\)
-
初等函数的导数
- \(C' = 0\)
- \((x^\alpha)' = \alpha x^{\alpha - 1}\)
- \((a^x)' = a^x \ln a\)
- \((\ln x)' = \cfrac 1x, \ (\log_a x)' = \cfrac 1{x \ln a}\)
- \((x^x)' = (e^{x \ln x})' = x^x(\ln x + 1)\)
- \((\sin x)' = \cos x, \ (\cos x)' = -\sin x\)
- \((\tan x)' = \sec^2 x, \ (\cot x)' = -\csc^2 x\)
- \((\sec x)' = \sec x \tan x, \ (\csc x)' = -\csc x \cot x\)
- \((\arcsin x)' = \cfrac 1{\sqrt{1-x^2}}, \ (\arccos x)' = -\cfrac 1{\sqrt{1-x^2}}\)
- \((\arctan x)' = \cfrac 1{1+x^2}\)
-
常见的\(n\)阶导数
- \((e^{ax})^{(n)} = a^n e^{ax}\)
- \((\sin ax)^{(n)} = a^n \sin(\cfrac {n\pi}2 + ax)\)
- \((\cos ax)^{(n)} = a^n \cos(\cfrac {n\pi}2 + ax)\)
- \((\ln (1+x))^{(n)} = \cfrac {(-1)^{n-1}(n-1)!}{(1+x)^n}\)
- \(((1+x)^\alpha)^{(n)} = \alpha (\alpha - 1) \cdots (\alpha - n + 1)(1 + x)^{\alpha - n}\)
- \((\ln x)^{(n)} = \cfrac {(-1)^{n-1}(n-1)!}{x^n}\)
- \((a^x)^{(n)} = a^x \ln^n a\)
- \((\cfrac 1{x+a})^{(n)} = \cfrac {(-1)^n n!}{(x+a)^{n+1}}\)
-
复合函数求导:\([f(\varphi (x))]' = f'(\varphi (x)) \cdot \varphi' (x)\)
-
隐函数求导
-
对数求导法:\(u(x)^{v(x)} = e^{v(x) \ln u(x)}\)
-
反函数求导:
\(\cfrac {dx}{dy} = \cfrac 1{y'}\)
\(\cfrac {d^2x}{dy^2} = \cfrac {d \cfrac {dx}{dy}}{dy} = \cfrac {d \cfrac 1{y'}}{dx} \cdot \cfrac {dx}{dy} = - \cfrac 1{(y')^2} \cdot y'' \cdot \cfrac 1{y'} = -\cfrac {y''}{(y')^3}\)
\(\cfrac {d^3x}{dy^3} = \cfrac {3(y'')^2 -y'y'''}{(y')^5}\)
-
参数方程求导:对于\(\begin{cases} x = x(t) \\ y = y(t) \end{cases}\),有\(y_x' = \cfrac {y_t'}{x_t'}\)
应用
-
极值、最值
- 设\(f(x)\)在\(x = x_0\)处连续,在\(x = x_0\)的去心邻域内可导。左侧\(f'(x) \gt 0\),右侧\(f'(x) \lt 0\),则\(f(x_0)\)为极大值;左侧\(f'(x) \lt 0\),右侧\(f'(x) \gt 0\),则\(f(x_0)\)为极小值。
- 设\(f(x)\)在\(x = x_0\)处存在二阶导数,\(f'(x_0) = 0\),\(f''(x_0) \neq 0\)。\(f''(x_0) \lt 0\),则\(f(x_0)\)为极大值;\(f''(x_0) \gt 0\),则\(f(x_0)\)为极小值。
- 设\(f(x)\)在\(x = x_0\)处存在\(n\)阶导数,\(f'(x_0) = f''(x_0) = \cdots = f^{(n-1)} (x) = 0\),\(f^{(n)} (x_0) \neq 0\)。\(n\)为奇数时,\(f(x_0)\)不是极值点;\(n\)为偶数时,若\(f''(x_0) \lt 0\),则\(f(x_0)\)为极大值,若\(f''(x_0) \gt 0\),则\(f(x_0)\)为极小值。
-
单调性、凹凸性
-
弦在曲线上方为凹;反之为凸。
\(\cfrac {f(x_1) + f(x_2)}2 \gt f(\cfrac {x_1 + x_2}2)\)为凹;\(\cfrac {f(x_1) + f(x_2)}2 \lt f(\cfrac {x_1 + x_2}2)\)为凸。
-
任意区间上\(f''(x) \gt 0\)为凹;反之为凸。
-
-
拐点(凹凸的分界点)、驻点(导函数等于零的点)
- 设\(f(x)\)在\(x = x_0\)处连续,在\(x = x_0\)的某去心邻域内二阶可导,且在\(x = x_0\)的左右邻域\(f''(x)\)反号,则点\((x_0, f(x_0))\)是曲线\(y = f(x)\)的拐点。
- 设\(f(x)\)在\(x = x_0\)处\(n\)可导,且\(f'(x_0) = f''(x_0) = \cdots = f^{(n-1)} (x) = 0\),\(f^{(n)} (x_0) \neq 0\)。\(n\)为奇数时,点\((x_0, f(x_0))\)是曲线\(y = f(x)\)的拐点。
-
渐近线
- 水平渐近线:\(\displaystyle \lim_{x \to \infty} f(x) = b\),则\(y = b\)是一条水平渐近线。
- 铅直渐近线:\(\displaystyle \lim_{x \to x_0^+} = \infty \ or \ \lim_{x \to x_0^-} = \infty\),则\(x = x_0\)是一条铅直渐近线。\(x_0\)的取值一般是分母为零、对数的真数为零等。
- 斜渐近线:\(\displaystyle \lim_{x \to \infty} \cfrac {f(x)}x = a, \ \lim_{x \to \infty} (f(x) - ax) = b\),则\(y = ax + b\)是一条斜渐近线。
-
曲线
- 弧微分:\(ds = \sqrt{1+y'^2}dx\)
- 曲率:\(K = \cfrac {y''}{(1+y'^2)^{\frac 32}}\)
- 曲率圆与曲率半径:\(\rho = \cfrac 1K\)
方程近似求解
- 二分法:
- 寻找区间\([a, b]\)满足\(f(a) \cdot f(b) \lt 0\);
- 取中点\(\xi_1 = \cfrac {a+b}2\),计算\(f(\xi_1)\);
- 若\(f(\xi_1) = 0\),则\(\xi_1\)为所求解;否则根据符号异号减小区间,再次取中点计算,直到满足误差;
- 误差为\(\cfrac 1{2^n}(b - a)\)。
- 切线法:
- 寻找区间\([a, b]\)满足\(f(a) \cdot f(b) \lt 0\);
- 选取一个合适的区间端点做切线\(y - f(\xi_0) = f'(\xi_0)(x - \xi_0)\),与\(x\)轴交点\(\xi_1 = \xi_0 \cfrac {f(\xi_0)}{f'(\xi_0)}\),计算\(f(\xi_1)\);
- 若\(f(\xi_1) = 0\),则\(\xi_1\)为所求解;否则根据符号异号减小区间,利用\(\xi_n = \xi_{n-1} \cfrac {f(\xi_{n-1})}{f'(\xi_{n-1})}\)计算,直到满足误差;
- 误差为最后所取区间的大小。
- 割线法
- 寻找区间\([a, b]\)满足\(f(a) \cdot f(b) \lt 0\);
- 取\(\xi_{n+1} = \xi_n - \cfrac {\xi_n - \xi_{n-1}}{f(\xi_n) - f'(\xi_{n-1})} f(\xi_n)\),计算\(f(\xi_{n+1})\);
- \(f(\xi_{n+1}) = 0\),则\(\xi_{n+1}\)为所求解;否则根据符号异号减小区间,重复步骤 2、3;
- 误差为最后所取区间的大小。