本文主要是介绍导数的基本法则与常用导数公式的推导,希望对大家解决编程问题提供一定的参考价值,需要的开发者们随着小编来一起学习吧!
目录
- n 次幂函数导数公式的推导
- 导数和的运算法则的证明
- 正弦、余弦函数导数公式的推导
- 代数证明
- 两个重要极限(引理)及证明
- 具体推导
- 几何直观
- 导数积的运算法则的证明
- 导数商的法则的证明
- 链式法则的证明
- 有理幂函数求导法则的证明
- 反函数求导法则的证明
- 反正切函数导数公式的推导
- 指数函数导数公式的推导
- 引入
- 证明
- 第一种方法:代入消元
- 第二种方法:对数微分
- 幂指函数导数公式的推导
- 幂法则(The Power Rule)的证明
- 第一种方法:e 的代换
- 第二种方法:对数微分
- e 的极限本质
- 双曲函数导数公式的推导
- 重要性质的证明
非专业、适用于计算机科学的微积分简要提纲,用于汇总梳理知识。如有错误恳请批评指正!
n 次幂函数导数公式的推导
d d x x n = lim Δ x → 0 Δ y Δ x = lim Δ x → 0 ( x + Δ x ) n − x n Δ x \begin{aligned} \frac{d}{dx} x^n & = \lim_{\Delta x\to 0} \frac{\Delta y}{\Delta x} \\\\ & = \lim_{\Delta x\to 0} \frac{(x+\Delta x)^n - x^n}{\Delta x} \end{aligned} dxdxn=Δx→0limΔxΔy=Δx→0limΔx(x+Δx)n−xn
对于微小量 Δ x \Delta x Δx,根据牛顿二项式定理
( x + y ) n = ∑ k = 0 n ( n k ) x n − k y k (x+y)^n = \sum_{k=0}^{n} \binom{n}{k} x^{n-k} y^k (x+y)n=k=0∑n(kn)xn−kyk
我们有(略去无穷小量余项)
( x + Δ x ) n = x n + n ( Δ x ) x n − 1 + o ( Δ x ) 2 (x+\Delta x)^n = x^n + n(\Delta x) x^{n-1} + o(\Delta x)^2 (x+Δx)n=xn+n(Δx)xn−1+o(Δx)2
则有
Δ y Δ x = [ x n + n ( Δ x ) x n − 1 + o ( Δ x ) 2 ] − x n Δ x = n x n − 1 + o ( Δ x ) \begin{aligned} \frac{\Delta y}{\Delta x} & = \frac{[x^n + n(\Delta x) x^{n-1} + o(\Delta x)^2] -x^n}{\Delta x} \\\\ &= nx^{n-1} + o(\Delta x) \end{aligned} ΔxΔy=Δx[xn+n(Δx)xn−1+o(Δx)2]−xn=nxn−1+o(Δx)
当 Δ x → 0 \Delta x \to 0 Δx→0 时,我们有
lim Δ x → 0 Δ y Δ x = n x n − 1 \lim_{\Delta x\to 0} \frac{\Delta y}{\Delta x} = nx^{n-1} Δx→0limΔxΔy=nxn−1
即
d d x x n = n x n − 1 \boxed{\frac{d}{dx} x^n = nx^{n-1}} dxdxn=nxn−1
导数和的运算法则的证明
( u + v ) ′ ( x ) = u ′ ( x ) + v ′ ( x ) \boxed{(u+v)'(x) = u'(x) + v'(x)} (u+v)′(x)=u′(x)+v′(x)
( u + v ) ′ ( x ) = lim Δ x → 0 ( u + v ) ( x + Δ x ) − ( u + v ) ( x ) Δ x = lim Δ x → 0 u ( x + Δ x ) − u ( x ) + v ( x + Δ x ) − v ( x ) Δ x = lim Δ x → 0 u ( x + Δ x ) − u ( x ) Δ x + lim Δ x → 0 v ( x + Δ x ) − v ( x ) Δ x = u ′ ( x ) + v ′ ( x ) . \begin{aligned} (u+v)'(x) & = \lim_{\Delta x \to 0} \frac{(u+v)(x+\Delta x)-(u+v)(x)}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \frac{u(x+\Delta x)-u(x)+v(x+\Delta x)-v(x)}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \frac{u(x+\Delta x)-u(x)}{\Delta x} + \lim_{\Delta x \to 0} \frac{v(x+\Delta x)-v(x)}{\Delta x} \\\\ &= u'(x)+v'(x). \end{aligned} (u+v)′(x)=Δx→0limΔx(u+v)(x+Δx)−(u+v)(x)=Δx→0limΔxu(x+Δx)−u(x)+v(x+Δx)−v(x)=Δx→0limΔxu(x+Δx)−u(x)+Δx→0limΔxv(x+Δx)−v(x)=u′(x)+v′(x).
正弦、余弦函数导数公式的推导
代数证明
两个重要极限(引理)及证明
lim Δ x → 0 c o s Δ x − 1 Δ x = 0 \boxed{\lim_{\Delta x \to 0} \frac{cos\Delta x - 1}{\Delta x} = 0 } Δx→0limΔxcosΔx−1=0
设 Δ x = θ \Delta x = \theta Δx=θ,我们考虑一个半径 r = 1 \ r = 1 r=1 的单位圆,以弧度为度量单位, θ \theta θ 扫过的弧线长度 l = θ r = θ \ l = \theta r = \theta l=θr=θ。过 θ \theta θ 角的终止角向其起始角作垂线,则垂点到圆心的水平距离为 c o s θ \ cos \ \theta cos θ,垂点到弧的一端的水平距离为 1 − c o s θ \ 1-cos\ \theta 1−cos θ。
当 θ → 0 \theta \to 0 θ→0 时, 1 − c o s θ \ 1-cos\ \theta 1−cos θ 递减比 θ \theta θ 更快,因此比值 1 − c o s θ θ → 0 \frac{1-cos\ \theta}{\theta} \to 0 θ1−cos θ→0,则可以证明 lim Δ x → 0 c o s Δ x − 1 Δ x = 0 \lim_{\Delta x \to 0} \frac{cos\Delta x - 1}{\Delta x} = 0 Δx→0limΔxcosΔx−1=0 成立。
lim Δ x → 0 s i n Δ x Δ x = 1 \boxed{\lim_{\Delta x \to 0} \frac{sin\Delta x}{\Delta x} = 1} Δx→0limΔxsinΔx=1
同理,如图
当 θ → 0 \theta \to 0 θ→0 时, s i n θ ≈ θ \ sin\ \theta \approx \theta sin θ≈θ 逐步趋向于精确,比值 s i n θ θ → 1 \frac{sin\ \theta}{\theta} \to 1 θsin θ→1,因此极限 lim Δ x → 0 s i n Δ x Δ x = 1 \lim_{\Delta x \to 0} \frac{sin\Delta x}{\Delta x} = 1 Δx→0limΔxsinΔx=1 成立。
具体推导
d d x s i n x = lim Δ x → 0 s i n ( x + Δ x ) − s i n x Δ x = lim Δ x → 0 s i n x c o s Δ x + c o s x s i n Δ x − s i n x Δ x = lim Δ x → 0 [ s i n x c o s Δ x − s i n x Δ x + c o s x s i n Δ x Δ x ] = lim Δ x → 0 [ s i n x ( c o s Δ x − 1 Δ x ) + c o s x ( s i n Δ x Δ x ) ] \begin{aligned} \frac{d}{dx} sin\ x &= \lim_{\Delta x \to 0} \frac{sin(x+\Delta x)-sin\ x}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \frac{sin\ x\ cos\Delta x + cos\ x \ sin\Delta x - sin \ x}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \left[\frac{sin\ x\ cos\Delta x - sin \ x}{\Delta x} + \frac{cos\ x \ sin\Delta x }{\Delta x}\right] \\\\ &= \lim_{\Delta x \to 0} \left[sin\ x\left(\frac{cos\Delta x - 1}{\Delta x}\right) + cos\ x\left( \frac{ \ sin\Delta x }{\Delta x}\right)\right] \\\\ \end{aligned} dxdsin x=Δx→0limΔxsin(x+Δx)−sin x=Δx→0limΔxsin x cosΔx+cos x sinΔx−sin x=Δx→0lim[Δxsin x cosΔx−sin x+Δxcos x sinΔx]=Δx→0lim[sin x(ΔxcosΔx−1)+cos x(Δx sinΔx)]
利用两个重要极限,可得
d d x s i n x = c o s x \begin{aligned} \boxed{\frac{d}{dx} sin\ x = cos\ x} \end{aligned} dxdsin x=cos x
同样地
d d x c o s x = lim Δ x → 0 c o s ( x + Δ x ) − c o s x Δ x = lim Δ x → 0 c o s x c o s Δ x − s i n x s i n Δ x − c o s x Δ x = lim Δ x → 0 [ c o s x c o s Δ x − c o s x Δ x − s i n x s i n Δ x Δ x ] = lim Δ x → 0 [ c o s x ( c o s Δ x − 1 Δ x ) − s i n x ( s i n Δ x Δ x ) ] \begin{aligned} \frac{d}{dx} cos\ x &= \lim_{\Delta x \to 0} \frac{cos(x+\Delta x)-cos\ x}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \frac{cos\ x\ cos\Delta x - sin\ x \ sin\Delta x - cos \ x}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \left[\frac{cos\ x\ cos\Delta x - cos \ x}{\Delta x} - \frac{sin\ x \ sin\Delta x }{\Delta x}\right] \\\\ &= \lim_{\Delta x \to 0} \left[cos\ x\left(\frac{cos\Delta x - 1}{\Delta x}\right) - sin\ x\left( \frac{ \ sin\Delta x }{\Delta x}\right)\right] \\\\ \end{aligned} dxdcos x=Δx→0limΔxcos(x+Δx)−cos x=Δx→0limΔxcos x cosΔx−sin x sinΔx−cos x=Δx→0lim[Δxcos x cosΔx−cos x−Δxsin x sinΔx]=Δx→0lim[cos x(ΔxcosΔx−1)−sin x(Δx sinΔx)]
利用两个重要极限,可得
d d x c o s x = − s i n x \begin{aligned} \boxed{\frac{d}{dx} cos\ x = -sin\ x} \end{aligned} dxdcos x=−sin x
几何直观
我们在单位圆上给予 θ \theta θ 微小增量 Δ θ \Delta \theta Δθ,则 P 的纵坐标为 s i n θ \ sin\ \theta sin θ,Q 的纵坐标为 s i n ( θ + Δ θ ) \ sin(\theta + \Delta \theta) sin(θ+Δθ)。
如图,纵坐标增量 Δ y = ∣ P R ∣ \Delta y = |PR| Δy=∣PR∣,线段 PQ 则是对弧线 P Q ⌢ \overset{\LARGE{\frown}}{PQ} PQ⌢的近似,我们可以很容易得到 ∣ P Q ∣ ≈ Δ θ \ |PQ| \approx \Delta\theta ∣PQ∣≈Δθ。考虑到 Δ θ \Delta \theta Δθ 为微小量,线段 PQ 几乎是圆的一条切线,以至于 ∠ O P Q \angle OPQ ∠OPQ 近似为直角,由几何关系, ∠ R P Q ≈ θ \angle RPQ \approx \theta ∠RPQ≈θ,则 c o s θ ≈ ∣ P R ∣ Δ θ = s i n ( θ + Δ θ ) − s i n θ Δ θ . \ cos\ \theta \approx \frac{|PR|}{\Delta\theta}=\frac{sin(\theta+\Delta\theta)-sin\ \theta}{\Delta\theta}. cos θ≈Δθ∣PR∣=Δθsin(θ+Δθ)−sin θ.
那么,当 Δ θ → 0 \Delta\theta \to 0 Δθ→0 时,这一近似结果就越准确,直到 lim Δ θ → 0 s i n ( θ + Δ θ ) − s i n θ Δ θ = c o s θ \lim_{\Delta\theta \to 0} \frac{sin(\theta+\Delta\theta)-sin\ \theta}{\Delta\theta} = cos\ \theta Δθ→0limΔθsin(θ+Δθ)−sin θ=cos θ
成立。
y = c o s θ y = cos\ \theta y=cos θ 同理易证,但是需要注意余弦函数的单调性(单调递减,增量为负)。
导数积的运算法则的证明
( u v ) ′ = u ′ v + u v ′ \boxed{(uv)'=u'v+uv'} (uv)′=u′v+uv′
( u v ) ′ = lim Δ x → 0 u ( x + Δ x ) v ( x + Δ x ) − u ( x ) v ( x ) Δ x (uv)' = \lim_{\Delta x \to 0} \frac{u(x+\Delta x)v(x+\Delta x)-u(x)v(x)}{\Delta x} (uv)′=Δx→0limΔxu(x+Δx)v(x+Δx)−u(x)v(x)
我们希望构造出 [ u ( x + Δ x ) − u ( x ) ] v ( x ) \left[u(x+\Delta x) - u(x)\right]v(x) [u(x+Δx)−u(x)]v(x) 这一项以期凑出 u ′ ( x ) u'(x) u′(x),于是我们注意到
u ( x + Δ x ) v ( x ) − u ( x + Δ x ) v ( x ) = 0 u(x+\Delta x)v(x)-u(x+\Delta x)v(x)=0 u(x+Δx)v(x)−u(x+Δx)v(x)=0
代入得
( u v ) ′ = lim Δ x → 0 u ( x + Δ x ) v ( x ) − u ( x ) v ( x ) + u ( x + Δ x ) v ( x + Δ x ) − u ( x + Δ x ) v ( x ) Δ x = lim Δ x → 0 { [ u ( x + Δ x ) − u ( x ) Δ x ] v ( x ) + u ( x + Δ x ) [ v ( x + Δ x ) − v ( x ) Δ x ] } = u ′ ( x ) v ( x ) + lim Δ x → 0 { u ( x + Δ x ) [ v ( x + Δ x ) − v ( x ) Δ x ] } = u ′ ( x ) v ( x ) + u ( x ) v ′ ( x ) . \begin{aligned} (uv)'&= \lim_{\Delta x \to 0} \frac{u(x+\Delta x)v(x)-u(x)v(x)+u(x+\Delta x)v(x+\Delta x)-u(x+\Delta x)v(x)}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \left\{ \left[\frac{u(x+\Delta x)-u(x)}{\Delta x}\right]v(x)+u(x+\Delta x) \left[\frac{v(x+\Delta x)-v(x)}{\Delta x}\right] \right\} \\\\ &= u'(x)v(x)+\lim_{\Delta x \to 0} \left\{u(x+\Delta x) \left[\frac{v(x+\Delta x)-v(x)}{\Delta x}\right] \right\} \\\\ &= u'(x)v(x)+u(x)v'(x). \end{aligned} (uv)′=Δx→0limΔxu(x+Δx)v(x)−u(x)v(x)+u(x+Δx)v(x+Δx)−u(x+Δx)v(x)=Δx→0lim{[Δxu(x+Δx)−u(x)]v(x)+u(x+Δx)[Δxv(x+Δx)−v(x)]}=u′(x)v(x)+Δx→0lim{u(x+Δx)[Δxv(x+Δx)−v(x)]}=u′(x)v(x)+u(x)v′(x).
导数商的法则的证明
( u v ) ′ = u ′ v − u v ′ v 2 \boxed{\left(\frac{u}{v}\right)'=\frac{u'v-uv'}{v^2}} (vu)′=v2u′v−uv′
( u v ) ′ = lim Δ x → 0 u ( x + Δ x ) v ( x + Δ x ) − u ( x ) v ( x ) Δ x \begin{aligned} \left(\frac{u}{v}\right)' = \lim_{\Delta x \to 0} \frac{\frac{u(x+\Delta x)}{v(x+\Delta x)}-\frac{u(x)}{v(x)}}{\Delta x} \end{aligned} (vu)′=Δx→0limΔxv(x+Δx)u(x+Δx)−v(x)u(x)
设 Δ u = u ( x + Δ x ) − u ( x ) \Delta u = u(x+\Delta x)-u(x) Δu=u(x+Δx)−u(x) , Δ v = v ( x + Δ x ) − v ( x ) \Delta v = v(x+\Delta x)-v(x) Δv=v(x+Δx)−v(x) ,则有
u ( x + Δ x ) v ( x + Δ x ) − u ( x ) v ( x ) = u + Δ u v + Δ v − u v = ( u + Δ u ) v − u ( v + Δ v ) ( v + Δ v ) v = u v + ( Δ u ) v − u v + u ( Δ v ) ( v + Δ v ) v = ( Δ u ) v + u ( Δ v ) ( v + Δ v ) v \begin{aligned} \frac{u(x+\Delta x)}{v(x+\Delta x)}-\frac{u(x)}{v(x)} &= \frac{u+\Delta u}{v+\Delta v} - \frac{u}{v} \\\\ &= \frac{(u+\Delta u)v-u(v+\Delta v)}{(v+\Delta v)v} \\\\ &= \frac{uv+(\Delta u)v-uv+u(\Delta v)}{(v+\Delta v)v} \\\\ &= \frac{(\Delta u)v+u(\Delta v)}{(v+\Delta v)v} \end{aligned} v(x+Δx)u(x+Δx)−v(x)u(x)=v+Δvu+Δu−vu=(v+Δv)v(u+Δu)v−u(v+Δv)=(v+Δv)vuv+(Δu)v−uv+u(Δv)=(v+Δv)v(Δu)v+u(Δv)
我们有
( u v ) ′ = lim Δ x → 0 ( Δ u ) v + u ( Δ v ) ( v + Δ v ) u Δ x = lim Δ x → 0 1 Δ x ( Δ u ) v + u ( Δ v ) ( v + Δ v ) u = lim Δ x → 0 ( Δ u Δ x ) v − u ( Δ v Δ x ) ( v + Δ v ) v = ( d u d x ) v − u ( d v d x ) v 2 = u ′ v − u v ′ v 2 . \begin{aligned} \left(\frac{u}{v}\right)' &= \lim_{\Delta x \to 0} \frac{\frac{(\Delta u)v+u(\Delta v)}{(v+\Delta v)u}}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \frac{1}{\Delta x} \frac{(\Delta u)v+u(\Delta v)}{(v+\Delta v)u} \\\\ &= \lim_{\Delta x \to 0} \frac{\left(\frac{\Delta u}{\Delta x}\right) v-u\left(\frac{\Delta v}{\Delta x}\right)}{(v+\Delta v)v} \\\\ &= \frac{ \left( \frac{du}{dx}\right)v-u \left( \frac{dv}{dx}\right)}{v^2} \\\\ &= \frac{u'v-uv'}{v^2}. \end{aligned} (vu)′=Δx→0limΔx(v+Δv)u(Δu)v+u(Δv)=Δx→0limΔx1(v+Δv)u(Δu)v+u(Δv)=Δx→0lim(v+Δv)v(ΔxΔu)v−u(ΔxΔv)=v2(dxdu)v−u(dxdv)=v2u′v−uv′.
链式法则的证明
d y d t = d y d x d x d t \boxed{\frac{dy}{dt}= \frac{dy}{dx} \frac{dx}{dt}} dtdy=dxdydtdx
d y = d y d x d x = d y d x d x d t d t . \begin{aligned} dy &= \frac{dy}{dx} dx \\\\ &= \frac{dy}{dx} \frac{dx}{dt} dt. \end{aligned} dy=dxdydx=dxdydtdxdt.
有理幂函数求导法则的证明
若 a ∈ Q a \in Q a∈Q,则有
d d x ( x a ) = a x a − 1 \boxed{\frac{d}{dx} (x^a)=ax^{a-1}} dxd(xa)=axa−1
假设 a = m n a=\frac{m}{n} a=nm,其中 m ∈ N m \in N m∈N, n ∈ N n \in N n∈N,我们有 y = x m n y = x^{\frac{m}{n}} y=xnm,则
y n = ( x m n ) n y n = x m \begin{aligned} y^n &= \left(x^{\frac{m}{n}}\right)^n \\ y^n &= x^m \end{aligned} ynyn=(xnm)n=xm
等式两边同时求导
d d x y n = d d x x m ( d d y y n ) d y d x = m x m − 1 n y n − 1 d y d x = m x m − 1 d y d x = m n x m − 1 y n − 1 \begin{aligned} \frac{d}{dx} y^n &= \frac{d}{dx} x^m \\\\ \left(\frac{d}{dy} y^n\right) \frac{dy}{dx} &= mx^{m-1} \\\\ ny^{n-1}\ \frac{dy}{dx} &= mx^{m-1} \\\\ \frac{dy}{dx} &= \frac{m}{n} \frac{x^{m-1}}{y^{n-1}} \end{aligned} dxdyn(dydyn)dxdynyn−1 dxdydxdy=dxdxm=mxm−1=mxm−1=nmyn−1xm−1
将 y = x m n y = x^{\frac{m}{n}} y=xnm 代入方程得
d y d x = m n [ x m − 1 ( x m n ) n − 1 ] = m n [ x m − 1 x m ( n − 1 ) n ] = m n x [ ( m − 1 ) − m ( n − 1 ) n ] = m n x [ n ( m − 1 ) − m ( n − 1 ) n ] = m n x m − n n = m n x m n − 1 = a x a − 1 . \begin{aligned} \frac{dy}{dx} &= \frac{m}{n} \left[ \frac{x^{m-1}}{(x^{\frac{m}{n}})^{n-1}}\right] \\\\ &= \frac{m}{n} \left[ \frac{x^{m-1}}{x^{\frac{m(n-1)}{n}}}\right] \\\\ &= \frac{m}{n} x^{\left[(m-1)-\frac{m(n-1)}{n}\right]} \\\\ &= \frac{m}{n} x^{\left[\frac{n(m-1)-m(n-1)}{n}\right]} \\\\ &= \frac{m}{n} x^{\frac{m-n}{n}} \\\\ &= \frac{m}{n} x^{\frac{m}{n}-1} \\\\ &= ax^{a-1}. \end{aligned} dxdy=nm[(xnm)n−1xm−1]=nm[xnm(n−1)xm−1]=nmx[(m−1)−nm(n−1)]=nmx[nn(m−1)−m(n−1)]=nmxnm−n=nmxnm−1=axa−1.
反函数求导法则的证明
d x d y = 1 d y d x \boxed{\frac{dx}{dy}=\frac{1}{\frac{dy}{dx}}} dydx=dxdy1
我们有 y = f ( x ) y = f(x) y=f(x), f − 1 ( y ) = x f^{-1}(y)=x f−1(y)=x,则
d d x [ f − 1 ( y ) ] = d d x ( x ) d d x [ f − 1 ( y ) ] = 1 \begin{aligned} \frac{d}{dx}[f^{-1}(y)] &= \frac{d}{dx} (x) \\\\ \frac{d}{dx}[f^{-1}(y)] &= 1 \\\\ \end{aligned} dxd[f−1(y)]dxd[f−1(y)]=dxd(x)=1
由链式法则
d d y [ f − 1 ( x ) ] d y d x = 1 d d y [ f − 1 ( x ) ] = 1 d y d x . \begin{aligned} \frac{d}{dy}[f^{-1}(x)]\ \frac{dy}{dx} &= 1 \\\\ \frac{d}{dy}[f^{-1}(x)] &= \frac{1}{\frac{dy}{dx}}. \\\\ \end{aligned} dyd[f−1(x)] dxdydyd[f−1(x)]=1=dxdy1.
反正切函数导数公式的推导
d d x a r c t a n ( x ) = 1 1 + x 2 \boxed{\frac{d}{dx}arctan(x) = \frac{1}{1+x^2}} dxdarctan(x)=1+x21
我们有 y = a r c t a n ( x ) = t a n − 1 ( x ) y = arctan(x)=tan^{-1}(x) y=arctan(x)=tan−1(x) ,等式两边同时取正切以简化方程:
t a n y = t a n [ t a n − 1 ( x ) ] t a n y = x \begin{aligned} tan\ y &= tan[tan^{-1}(x)] \\ tan \ y &= x \end{aligned} tan ytan y=tan[tan−1(x)]=x
如图,我们限制函数 t a n ( x ) tan(x) tan(x) 的定义域为 ( − π 2 , π 2 ) \left(-\frac{\pi}{2}, \frac{\pi}{2}\right) (−2π,2π) ,我们沿 y = x y=x y=x 将 t a n ( x ) tan(x) tan(x) 翻转即可得到反函数 t a n − 1 ( x ) tan^{-1}(x) tan−1(x) :
我们可以知道, lim x → ∞ t a n − 1 ( x ) = π 2 \lim_{x\to \infty} tan^{-1}(x) = \frac{\pi}{2} limx→∞tan−1(x)=2π ,而我们显然可以推导出
d d y t a n y = d d y s i n y c o s y = c o s 2 y − s i n 2 y c o s 2 y = 1 c o s 2 y = s e c 2 y \begin{aligned} \frac{d}{dy} tan\ y &= \frac{d}{dy} \frac{sin\ y}{cos\ y} \\\\ &= \frac{cos^2 y \ -sin^2y}{cos^2 y} \\\\ &= \frac{1}{cos^2 y} \\\\ &= sec^2 y \end{aligned} dydtan y=dydcos ysin y=cos2ycos2y −sin2y=cos2y1=sec2y
我们对 t a n y = x tan\ y = x tan y=x 两边同时求导
d d x [ t a n ( y ) ] = d d x x d d y [ t a n ( y ) ] d y d x = 1 1 c o s 2 y d y d x = 1 d y d x = c o s 2 y \begin{aligned} \frac{d}{dx}[tan(y)] &= \frac{d}{dx} x \\\\ \frac{d}{dy}[tan(y)]\ \frac{dy}{dx} &= 1 \\\\ \frac{1}{cos^2y}\ \frac{dy}{dx} &= 1 \\\\ \frac{dy}{dx} &= cos^2y \end{aligned} dxd[tan(y)]dyd[tan(y)] dxdycos2y1 dxdydxdy=dxdx=1=1=cos2y
我们仍需要以 x x x 消去 y y y,由三角关系, c o s ( y ) = 1 1 + x 2 cos(y)= \frac{1}{\sqrt{1+x^2}} cos(y)=1+x21,那么 c o s 2 ( y ) = 1 1 + x 2 cos^2(y) = \frac{1}{1+x^2} cos2(y)=1+x21,则
d d x a r c t a n ( x ) = 1 1 + x 2 . \frac{d}{dx} arctan(x) = \frac{1}{1+x^2}. dxdarctan(x)=1+x21.
指数函数导数公式的推导
d d x a x = a x ln a \boxed{\frac{d}{dx}a^x=a^x\ln a} dxdax=axlna
引入
由导数的定义
d d x a x = lim Δ x → 0 a x + Δ x − a x Δ x = lim Δ x → 0 a x a Δ x − a x Δ x = lim Δ x → 0 a x a Δ x − 1 Δ x = a x lim Δ x → 0 a Δ x − 1 Δ x \begin{aligned} \frac{d}{dx} a^x &= \lim_{\Delta x \to 0} \frac{a^{x+\Delta x}-a^x}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} \frac{a^x a^{\Delta x} - a^x}{\Delta x} \\\\ &= \lim_{\Delta x \to 0} a^x \frac{a^{\Delta x}-1}{\Delta x} \\\\ &= a^x\lim_{\Delta x \to 0} \frac{a^{\Delta x}-1}{\Delta x} \\\\ \end{aligned} dxdax=Δx→0limΔxax+Δx−ax=Δx→0limΔxaxaΔx−ax=Δx→0limaxΔxaΔx−1=axΔx→0limΔxaΔx−1
设 lim Δ x → 0 a Δ x − 1 Δ x = M ( a ) \lim_{\Delta x \to 0} \frac{a^{\Delta x}-1}{\Delta x} = M(a) limΔx→0ΔxaΔx−1=M(a) ,则有
d d x a x = M ( a ) a x \frac{d}{dx}a^x = M(a)\ a^x dxdax=M(a) ax
由于 d d x a x ∣ x = 0 = M ( a ) \frac{d}{dx} a^x |_{x=0}=M(a) dxdax∣x=0=M(a) ,我们只需要求解出 y = a x y=a^x y=ax 图象在 x = 0 x=0 x=0 处切线的斜率即可得到指数函数导数公式。假设一个常数 e e e ,使得 M ( e ) = 1 M(e)=1 M(e)=1 ,那么有
d d x e x = e x \boxed{\frac{d}{dx} e^x=e^x} dxdex=ex
同时, y = e x y=e^x y=ex 图象在 x = 0 x=0 x=0 的切线斜率为 1. 下面证 e e e 的唯一性(不完全):
随着底数 a a a 增长,指数函数的图象愈来愈陡峭,则其导数单调递增,当 a = 1 a=1 a=1 时, a x = 1 a^x=1 ax=1 恒成立且对应指数函数图象斜率恒为 0;我们用割线来逼近 a = 2 a=2 a=2 以及 a = 4 a=4 a=4 的情况:
如图,从 ( 0 , 1 ) (0,1) (0,1) 到 ( 1 , 2 ) (1,2) (1,2) 关于 y = 2 x y=2^x y=2x 的割线斜率为 1,我们很容易可以看出, M ( 2 ) < 1 M(2)<1 M(2)<1 ;
如下图,从 ( − 1 2 , 1 2 ) (-\frac{1}{2}, \frac{1}{2}) (−21,21) 到 ( 1 , 0 ) (1,0) (1,0) 关于 y = 4 x y=4^x y=4x 的割线斜率为 1,我们也可以得到 M ( 4 ) > 1 M(4)>1 M(4)>1 。
由于 y = a x y=a^x y=ax 的导数必定连续且单增,则 M ( a ) M(a) M(a) 也连续且单增,由上述推理可知 a ∈ ( 2 , 4 ) a \in (2, 4) a∈(2,4) 有且只有一个 e e e 满足 M ( a ) = 1 M(a) =1 M(a)=1 ,唯一性得证。
自然对数是以 e e e 为底的指数函数的反函数:
y = e x , ln ( y ) = x y=e^x, \ln (y)=x y=ex,ln(y)=x
设 w = ln x w=\ln x w=lnx,则有
d d x e w = d d x x d d w ( e w ) d w d x = 1 e w d w d x = 1 d w d x = 1 e w \begin{aligned} \frac{d}{dx}e^w &= \frac{d}{dx} x \\\\ \frac{d}{dw}(e^w)\ \frac{dw}{dx} &= 1 \\\\ e^w\ \frac{dw}{dx} &= 1 \\\\ \frac{dw}{dx} &= \frac{1}{e^w}\\\\ \end{aligned} dxdewdwd(ew) dxdwew dxdwdxdw=dxdx=1=1=ew1
由假设,我们有
d d x ln x = 1 x \boxed{\frac{d}{dx} \ln x=\frac{1}{x}} dxdlnx=x1
证明
第一种方法:代入消元
我们使用 e e e 重新表达 a x a^x ax :
a x = ( e ln a ) x = e x ln a a^x=\left(e^{\ln a}\right)^x=e^{x\ln a} ax=(elna)x=exlna
由链式法则,我们可以得到
d d x e x ln a = ( ln a ) e x ln a . \frac{d}{dx} e^{x \ln a}=(\ln a)\ e^{x\ln a}. dxdexlna=(lna) exlna.
即
d d x a x = a x ln a \boxed{\frac{d}{dx}a^x=a^x\ln a} dxdax=axlna
那么, M ( a ) = ln a M(a) = \ln a M(a)=lna .
第二种方法:对数微分
设 u = a x u=a^x u=ax ,而由链式法则, d d x ln u = 1 u d u d x \frac{d}{dx}\ln u=\frac{1}{u} \frac{du}{dx} dxdlnu=u1dxdu,即 ( ln u ) ′ = u ′ u . (\ln u)'=\frac{u'}{u}. (lnu)′=uu′. 则有
ln u = ln ( a x ) ln u = x ln a ( ln u ) ′ = ln a \begin{aligned} \ln u &= \ln (a^x) \\ \ln u &= x\ln a \\ (\ln u)' &= \ln a \end{aligned} lnulnu(lnu)′=ln(ax)=xlna=lna
由 ( ln u ) ′ = u ′ u (\ln u)'=\frac{u'}{u} (lnu)′=uu′ 可得 u ′ = u ( ln u ) ′ u'=u(\ln u)' u′=u(lnu)′ ,即 ( a x ) ′ = a x ln a . (a^x)' = a^x \ln a. (ax)′=axlna.
幂指函数导数公式的推导
d d x x x = x x ( 1 + ln x ) \boxed{\frac{d}{dx}x^x=x^x(1+\ln x)} dxdxx=xx(1+lnx)
之前我们已然推导出 ( ln u ) ′ = u ′ u (\ln u)'=\frac{u'}{u} (lnu)′=uu′,设 v = x x v=x^x v=xx ,我们利用对数微分的技巧求解 v ′ v' v′ :
ln v = x ln x ( ln v ) ′ = ln x + x ⋅ 1 x = v ′ v \begin{aligned} \ln v &= x \ln x \\ (\ln v)' &= \ln x + x \cdot \frac{1}{x} \\ &= \frac{v'}{v} \end{aligned} lnv(lnv)′=xlnx=lnx+x⋅x1=vv′
则有
v ′ x x = 1 + ln x v ′ = x x ( 1 + ln x ) . \begin{aligned} \frac{v'}{x^x} &= 1+\ln x \\\\ v' &=x^x(1+\ln x). \end{aligned} xxv′v′=1+lnx=xx(1+lnx).
幂法则(The Power Rule)的证明
我们先前证明过有理数幂函数的导数公式,现在我们将其拓展到实数域:
d d x x r = r x r − 1 \boxed{\frac{d}{dx}x^r=rx^{r-1}} dxdxr=rxr−1
第一种方法:e 的代换
显然 x r = e r ln x x^r=e^{r\ln x} xr=erlnx ,则
d d x x r = d d x e r ln x = e r ln x d d x ( r ln x ) = e r ln x ( r x ) = x r ( r x ) = r x r − 1 . \begin{aligned} \frac{d}{dx}x^r &= \frac{d}{dx} e^{r\ln x} = e^{r\ln x}\ \frac{d}{dx}(r\ln x) \\\\ &= e^{r\ln x}\ \left(\frac{r}{x}\right) \\\\ &= x^r \left(\frac{r}{x}\right) = rx^{r-1}. \end{aligned} dxdxr=dxderlnx=erlnx dxd(rlnx)=erlnx (xr)=xr(xr)=rxr−1.
第二种方法:对数微分
我们定义 f ( x ) = x r f(x)=x^r f(x)=xr,则有
ln f = r ln x ( ln f ) ′ = f ′ f = r x f ′ = f ( ln f ) ′ = x r ( r x ) = r x r − 1 . \begin{aligned} \ln f &= r\ln x \\ (\ln f)' &= \frac{f'}{f} = \frac{r}{x} \\ f' &= f(\ln f)' = x^r \left(\frac{r}{x}\right) \\ &= rx^{r-1}. \end{aligned} lnf(lnf)′f′=rlnx=ff′=xr=f(lnf)′=xr(xr)=rxr−1.
e 的极限本质
lim n → ∞ ( 1 + 1 n ) n = e \boxed{\lim_{n\to \infty}\left(1+\frac{1}{n}\right)^n = e} n→∞lim(1+n1)n=e
取自然对数
ln [ ( 1 + 1 n ) n ] = n ln ( 1 + 1 n ) \ln \left[\left(1+\frac{1}{n}\right)^n\right] = n\ln \left(1+\frac{1}{n}\right) ln[(1+n1)n]=nln(1+n1)
当 n → ∞ n\to \infty n→∞ 时, Δ x = 1 n → 0 \Delta x =\frac{1}{n} \to 0 Δx=n1→0, n = 1 Δ x n=\frac{1}{\Delta x} n=Δx1,这样我们就可以将其转换为趋近于 0 的极限问题:
lim n → ∞ n ln ( 1 + 1 n ) = lim Δ x → 0 [ 1 Δ x ln ( 1 + Δ x ) ] \begin{aligned} \lim_{n\to \infty}n\ln \left(1+\frac{1}{n}\right) &= \lim_{\Delta x \to 0} \left[\frac{1}{\Delta x} \ln (1+\Delta x)\right] \\\\ \end{aligned} n→∞limnln(1+n1)=Δx→0lim[Δx1ln(1+Δx)]
利用 ln 1 = 0 \ln 1 = 0 ln1=0 这一性质,我们可以构造出熟悉的导数极限定义式的形式:
lim Δ x → 0 [ 1 Δ x ln ( 1 + Δ x ) ] = lim Δ x → 0 { 1 Δ x [ ln ( 1 + Δ x ) − ln 1 ] } = lim Δ x → 0 ln ( 1 + Δ x ) − ln 1 Δ x = d d x ln x ∣ x = 1 = 1 x ∣ x = 1 = 1 \begin{aligned} \lim_{\Delta x \to 0} \left[\frac{1}{\Delta x} \ln (1+\Delta x)\right] &= \lim_{\Delta x\to 0} \left\{ \frac{1}{\Delta x} \big [ \ln (1+\Delta x)-\ln 1 \big ]\right\} \\\\ &= \lim_{\Delta x \to 0} \frac{\ln (1+\Delta x)-\ln 1}{\Delta x} \\\\ &= \frac{d}{dx} \ln x\ \bigg|_{x=1} = \frac{1}{x}\ \bigg|_{x=1} \\\\ &= 1 \end{aligned} Δx→0lim[Δx1ln(1+Δx)]=Δx→0lim{Δx1[ln(1+Δx)−ln1]}=Δx→0limΔxln(1+Δx)−ln1=dxdlnx x=1=x1 x=1=1
这是原极限取自然对数的结果,意味着
lim n → ∞ ( 1 + 1 n ) n = lim n → ∞ e ln [ ( 1 + 1 n ) n ] = e lim n → ∞ ln [ ( 1 + 1 n ) n ] = e 1 = e . \begin{aligned} \lim_{n\to \infty}\left(1+\frac{1}{n}\right)^n &= \lim_{n\to \infty} e^{\ln\left[ \left(1+\frac{1}{n}\right)^n\right]} \\\\ &= e^{\lim_{n\to \infty}\ln\left[ \left(1+\frac{1}{n}\right)^n\right]} \\\\ &= e^1 = e. \end{aligned} n→∞lim(1+n1)n=n→∞limeln[(1+n1)n]=elimn→∞ln[(1+n1)n]=e1=e.
双曲函数导数公式的推导
双曲正弦函数
sinh ( x ) = e x − e − x 2 \sinh(x) = \frac{e^x-e^{-x}}{2} sinh(x)=2ex−e−x
双曲余弦函数
cosh ( x ) = e x + e − x 2 \cosh(x) = \frac{e^x+e^{-x}}{2} cosh(x)=2ex+e−x
则有
d d x sinh ( x ) = cosh ( x ) \boxed{\frac{d}{dx}\sinh(x) = \cosh(x)} dxdsinh(x)=cosh(x)
d d x cosh ( x ) = sinh ( x ) \boxed{\frac{d}{dx}\cosh(x) = \sinh(x)} dxdcosh(x)=sinh(x)
d d x sinh ( x ) = e x − ( − e − x ) 2 = cosh ( x ) . \frac{d}{dx} \sinh(x) = \frac{e^x-(-e^{-x})}{2} = \cosh(x). dxdsinh(x)=2ex−(−e−x)=cosh(x).
cosh ( x ) \cosh(x) cosh(x) 同理可证。
重要性质的证明
cosh 2 ( x ) − sinh 2 ( x ) = 1 \boxed{\cosh^2(x)-\sinh^2(x)=1} cosh2(x)−sinh2(x)=1
cosh 2 ( x ) − sinh 2 ( x ) = ( e x + e − x 2 ) 2 − ( e x − e − x 2 ) 2 = 1 4 ( e 2 x + 2 e x e − x + e − 2 x ) − 1 4 ( e 2 x − 2 + e − 2 x ) = 1 4 ( 2 + 2 ) = 1. \begin{aligned} \cosh^2(x)-\sinh^2(x) &= \left(\frac{e^x+e^{-x}}{2}\right)^2-\left(\frac{e^x-e^{-x}}{2}\right)^2 \\\\ &= \frac{1}{4} (e^{2x}+2e^xe^{-x}+e^{-2x}) - \frac{1}{4} (e^{2x}-2+e^{-2x}) \\\\ &= \frac{1}{4} (2+2) = 1. \end{aligned} cosh2(x)−sinh2(x)=(2ex+e−x)2−(2ex−e−x)2=41(e2x+2exe−x+e−2x)−41(e2x−2+e−2x)=41(2+2)=1.
这篇关于导数的基本法则与常用导数公式的推导的文章就介绍到这儿,希望我们推荐的文章对编程师们有所帮助!