前言
本文将介绍如何用定积分计算空间中一段光滑曲线的弧长。首先我们会给出光滑曲线以及曲线弧长的定义,然后从定义出发,用求黎曼和的思想推导出弧长的计算公式。
光滑曲线的定义
设平面曲线的参数方程为
{ x = x ( t ) , y = y ( t ) , t ∈ [ T 1 , T 2 ] . \begin{cases}x=x(t), \\ y=y(t),\end{cases} t\in [T_1,T_2]. {x=x(t),y=y(t),t∈[T1,T2].
若 x ′ ( t ) x'(t) x′(t)与 y ′ ( t ) y'(t) y′(t)在 [ T 1 , T 2 ] [T_1, T_2] [T1,T2]上连续,且 [ x ′ ( t ) ] 2 + [ y ′ ( t ) ] 2 ≠ 0 [x'(t)]^2+[y'(t)]^2 \neq 0 [x′(t)]2+[y′(t)]2=0,则该曲线被称为光滑曲线。
从光滑曲线的定义能看出,其切线向量在任意方向都是连续变动的,且在任意一点的切线向量都不是零向量,即有特定的方向。
弧长的定义
对曲线参数方程的参数区间 [ T 1 , T 2 ] [T_1,T_2] [T1,T2]做如下划分:
T 1 = t 0 < t 1 < . . . < t n = T 2 T_1=t_0<t_1<...<t_n=T_2 T1=t0<t1<...<tn=T2 这些划分在曲线上形成了 n + 1 n+1 n+1个点 P 0 , P 1 , . . . , P n P_0, P_1, ..., P_n P0,P1,...,Pn,如下图所示
定义连接两个相邻点之间的线段长度为 P i − 1 P i ‾ \overline{P_{i-1}P_{i}} Pi−1Pi。若当 λ = m a x 1 ≤ i ≤ n ( Δ t i ) → 0 \lambda=\mathop{max}\limits_{1\leq i \leq n}(\Delta t_i)\rightarrow 0 λ=1≤i≤nmax(Δti)→0时,极限 lim λ → 0 ∑ i = 1 n P i − 1 P i ‾ \displaystyle \lim_{\lambda \rightarrow 0}\sum_{i=1}^n \overline{P_{i-1}P_{i}} λ→0limi=1∑nPi−1Pi存在,且极限值与区间 [ T 1 , T 2 ] [T_1,T_2] [T1,T2]的划分无关,则称这条曲线是可求长的,并将极限值
l = lim λ → 0 ∑ i = 1 n P i − 1 P i ‾ l=\displaystyle \lim_{\lambda \rightarrow 0}\sum_{i=1}^n \overline{P_{i-1}P_{i}} l=λ→0limi=1∑nPi−1Pi
定义为曲线的弧长。
从曲线弧长的定义来看,它和Riemann和的定义非常相似,如果我们能将 l = lim λ → 0 ∑ i = 1 n P i − 1 P i ‾ l=\displaystyle \lim_{\lambda \rightarrow 0}\sum_{i=1}^n \overline{P_{i-1}P_{i}} l=λ→0limi=1∑nPi−1Pi写成Riemann和" lim λ → 0 ∑ i − 1 n f ( ξ i ) Δ t i \displaystyle \lim_{\lambda \rightarrow 0}\sum_{i-1}^n f(\xi_i)\Delta t_{i} λ→0limi−1∑nf(ξi)Δti"的形式,那我们就可以通过定积分来求解弧长公式了。下面我们将推导出弧长公式。
光滑曲线弧长公式的推导
首先将线段长度写成坐标形式:
P i − 1 P i ‾ = [ x ( t i ) − x ( t i − 1 ) ] 2 + [ y ( t i ) − y ( t i − 1 ) ] 2 \begin{equation} \overline{P_{i-1}P_{i}}=\sqrt{[x(t_i)-x(t_{i-1})]^2+[y(t_i)-y(t_{i-1})]^2} \end{equation} Pi−1Pi=[x(ti)−x(ti−1)]2+[y(ti)−y(ti−1)]2
由于曲线光滑,因此根据Lagrange中值定理,在 ( t i − 1 , t i ) (t_{i-1}, t_i) (ti−1,ti)上存在 η i , μ i \eta_i , \mu_i ηi,μi,使得
x ( t i ) − x ( t i − 1 ) = x ′ ( η i ) Δ t i , y ( t i ) − y ( t i − 1 ) = y ′ ( μ i ) Δ t i , \begin{equation} x(t_i)-x(t_{i-1})=x'(\eta_i) \Delta t_i, y(t_i)-y(t_{i-1})=y'(\mu_i) \Delta t_i, \end{equation} x(ti)−x(ti−1)=x′(ηi)Δti,y(ti)−y(ti−1)=y′(μi)Δti,
则 ( 1 ) (1) (1)式可以写成
P i − 1 P i ‾ = [ x ′ ( η i ) ] 2 + [ y ′ ( μ i ) ] 2 ⋅ Δ t i \begin{equation} \overline{P_{i-1}P_{i}}= \sqrt{[x'(\eta_i)]^2+[y'(\mu_i)]^2} \cdot \Delta t_i \end{equation} Pi−1Pi=[x′(ηi)]2+[y′(μi)]2⋅Δti
因此有
∑ i = 1 n P i − 1 P i ‾ = ∑ i = 1 n [ x ′ ( η i ) ] 2 + [ y ′ ( μ i ) ] 2 ⋅ Δ t i \begin{equation} \sum_{i=1}^n \overline{P_{i-1}P_{i}}=\displaystyle\sum_{i=1}^n \sqrt{[x'(\eta_i)]^2+[y'(\mu_i)]^2} \cdot \Delta t_i \end{equation} i=1∑nPi−1Pi=i=1∑n[x′(ηi)]2+[y′(μi)]2⋅Δti
由于 η i , μ i \eta_i , \mu_i ηi,μi并不一定相同,因此此时 ( 4 ) (4) (4)式并不是一个Riemann和的形式:
∑ i = 1 n [ x ′ ( ξ i ) ] 2 + [ y ′ ( ξ i ) ] 2 ⋅ Δ t i \begin{equation} \sum_{i=1}^n \sqrt{[x'(\xi_i)]^2+[y'(\xi_i)]^2} \cdot \Delta t_i \end{equation} i=1∑n[x′(ξi)]2+[y′(ξi)]2⋅Δti
接下来我们证明 ( 4 ) (4) (4)式和 ( 5 ) (5) (5)式在 λ → 0 \lambda \rightarrow 0 λ→0时是相等的。将 ( 4 ) (4) (4)式减去 ( 5 ) (5) (5)式,取绝对值,有
∣ ∑ i = 1 n P i − 1 P i ‾ − ∑ i = 1 n [ x ′ ( ξ i ) ] 2 + [ y ′ ( ξ i ) ] 2 ⋅ Δ t i ∣ = ∣ ∑ i = 1 n [ x ′ ( η i ) ] 2 + [ y ′ ( μ i ) ] 2 ⋅ Δ t i − ∑ i = 1 n [ x ′ ( ξ i ) ] 2 + [ y ′ ( ξ i ) ] 2 ⋅ Δ t i ∣ = ∣ ∑ i = 1 n ( [ x ′ ( η i ) ] 2 + [ y ′ ( μ i ) ] 2 − [ x ′ ( ξ i ) ] 2 + [ y ′ ( ξ i ) ] 2 ) Δ t i ∣ ≤ ∑ i = 1 n ∣ ( [ x ′ ( η i ) ] 2 + [ y ′ ( μ i ) ] 2 − [ x ′ ( ξ i ) ] 2 + [ y ′ ( ξ i ) ] 2 ) ∣ Δ t i \begin{equation} \begin{align} &\quad \left| \sum_{i=1}^n \overline{P_{i-1}P_{i}}-\sum_{i=1}^n \sqrt{[x'(\xi_i)]^2+[y'(\xi_i)]^2} \cdot \Delta t_i \right| \nonumber \\ &=\left|\sum_{i=1}^n \sqrt{[x'(\eta_i)]^2+[y'(\mu_i)]^2} \cdot \Delta t_i- \sum_{i=1}^n \sqrt{[x'(\xi_i)]^2+[y'(\xi_i)]^2} \cdot \Delta t_i \right| \nonumber \\ &=\left|\sum_{i=1}^n \left( \sqrt{[x'(\eta_i)]^2+[y'(\mu_i)]^2} - \sqrt{[x'(\xi_i)]^2+[y'(\xi_i)]^2}\right) \Delta t_i\right| \nonumber \\ & \leq \sum_{i=1}^n \left|\left( \sqrt{[x'(\eta_i)]^2+[y'(\mu_i)]^2} - \sqrt{[x'(\xi_i)]^2+[y'(\xi_i)]^2}\right) \right| \Delta t_i \nonumber \\ \end{align} \end{equation} i=1∑nPi−1Pi−i=1∑n[x′(ξi)]2+[y′(ξi)]2⋅Δti = i=1∑n[x′(ηi)]2+[y′(μi)]2⋅Δti−i=1∑n[x′(ξi)]2+[y′(ξi)]2⋅Δti = i=1∑n([x′(ηi)]2+[y′(μi)]2−[x′(ξi)]2+[y′(ξi)]2)Δti ≤i=1∑n ([x′(ηi)]2+[y′(μi)]2−[x′(ξi)]2+[y′(ξi)]2) Δti
根据向量形式三角不等式
∣ x 1 2 + x 2 2 − y 1 2 + y 2 2 ∣ ≤ ( x 1 − y 1 ) 2 + ( x 2 − y 2 ) 2 ≤ ∣ x 1 − y 1 ∣ + ∣ x 2 − y 2 ∣ \begin{equation} \left| \sqrt{x_1^2+x_2^2} - \sqrt{y_1^2+y_2^2}\right | \leq \sqrt{(x_1 - y_1)^2 +(x_2 - y_2)^2} \leq |x_1-y_1|+|x_2-y_2| \end{equation} x12+x22−y12+y22 ≤(x1−y1)2+(x2−y2)2≤∣x1−y1∣+∣x2−y2∣
(这里简单说明一下 ( 7 ) (7) (7)式的证明方法:在 O x y Oxy Oxy坐标平面上取三个点: A ( x 1 , x 2 ) , B ( y 1 , y 2 ) , C ( y 1 , x 2 ) A(x_1, x_2), B(y_1, y_2), C(y_1, x_2) A(x1,x2),B(y1,y2),C(y1,x2)。在 O A B OAB OAB和 A B C ABC ABC这两个三角形中利用三角形不等式有: ∣ A B ∣ ≥ ∣ O A ∣ − ∣ O B ∣ , ∣ A B ∣ ≤ ∣ A C ∣ + ∣ A B ∣ |AB|\geq |OA|-|OB|, |AB|\leq |AC|+|AB| ∣AB∣≥∣OA∣−∣OB∣,∣AB∣≤∣AC∣+∣AB∣)
利用 ( 7 ) (7) (7)式结论, ( 6 ) (6) (6)式可以写为
∣ ∑ i = 1 n P i − 1 P i ‾ − ∑ i = 1 n [ x ′ ( ξ i ) ] 2 + [ y ′ ( ξ i ) ] 2 ⋅ Δ t i ∣ ≤ ∑ i = 1 n ∣ ( [ x ′ ( η i ) ] 2 + [ y ′ ( μ i ) ] 2 − [ x ′ ( ξ i ) ] 2 + [ y ′ ( ξ i ) ] 2 ) ∣ Δ t i ≤ ∑ i = 1 n ∣ x ′ ( η i ) − x ′ ( ξ i ) ∣ Δ t i + ∑ i = 1 n ∣ y ′ ( μ i ) − y ′ ( ξ i ) ∣ Δ t i ≤ ∑ i = 1 n ω ‾ i Δ t i + ∑ i = 1 n ω ~ i Δ t i \begin{equation} \begin{align} &\quad \left| \sum_{i=1}^n \overline{P_{i-1}P_{i}}-\sum_{i=1}^n \sqrt{[x'(\xi_i)]^2+[y'(\xi_i)]^2} \cdot \Delta t_i \right| \nonumber \\ & \leq \sum_{i=1}^n \left|\left( \sqrt{[x'(\eta_i)]^2+[y'(\mu_i)]^2} - \sqrt{[x'(\xi_i)]^2+[y'(\xi_i)]^2}\right) \right| \Delta t_i \nonumber \\ &\leq \sum_{i=1}^n \left| x'(\eta_i) - x'(\xi_i)\right| \Delta t_i + \sum_{i=1}^n \left| y'(\mu_i) - y'(\xi_i)\right| \Delta t_i \nonumber \\ &\leq \sum_{i=1}^n \overline{\omega}_i \Delta t_i + \sum_{i=1}^n \widetilde{\omega}_i \Delta t_i \nonumber \\ \end{align} \end{equation} i=1∑nPi−1Pi−i=1∑n[x′(ξi)]2+[y′(ξi)]2⋅Δti ≤i=1∑n ([x′(ηi)]2+[y′(μi)]2−[x′(ξi)]2+[y′(ξi)]2) Δti≤i=1∑n∣x′(ηi)−x′(ξi)∣Δti+i=1∑n∣y′(μi)−y′(ξi)∣Δti≤i=1∑nωiΔti+i=1∑nω iΔti
其中 ω ‾ i , ω ~ i \overline{\omega}_i, \widetilde{\omega}_i ωi,ω i分别是 x ′ ( t ) , y ′ ( t ) x'(t), y'(t) x′(t),y′(t)在 [ t i − 1 , t i ] [t_{i-1}, t_i] [ti−1,ti]上的振幅。由于曲线光滑,因此 x ′ ( t ) , y ′ ( t ) x'(t), y'(t) x′(t),y′(t)在 [ t i − 1 , t i ] [t_{i-1}, t_i] [ti−1,ti]上可积,根据黎曼可积的充要条件,当 λ = m a x 1 ≤ i ≤ n ( Δ t i ) → 0 \lambda=\mathop{max}\limits_{1\leq i \leq n}(\Delta t_i)\rightarrow 0 λ=1≤i≤nmax(Δti)→0时,振幅满足: ∑ i = 1 n ω ‾ i Δ t i → 0 \displaystyle \sum_{i=1}^n \overline{\omega}_i \Delta t_i \rightarrow 0 i=1∑nωiΔti→0, ∑ i = 1 n ω ~ i Δ t i → 0 \displaystyle \sum_{i=1}^n \widetilde{\omega}_i \Delta t_i \rightarrow 0 i=1∑nω iΔti→0。因此有:
l = lim λ → 0 ∑ i = 1 n P i − 1 P i ‾ = lim λ → 0 ∑ i = 1 n [ x ′ ( ξ i ) ] 2 + [ y ′ ( ξ i ) ] 2 ⋅ Δ t i = ∫ T 1 T 2 [ x ′ ( t ) ] 2 + [ y ′ ( t ) ] 2 d t \begin{equation} l=\displaystyle \lim_{\lambda \rightarrow 0}\sum_{i=1}^n \overline{P_{i-1}P_{i}}=\lim_{\lambda \rightarrow 0}\sum_{i=1}^n \sqrt{[x'(\xi_i)]^2+[y'(\xi_i)]^2} \cdot \Delta t_i =\int_{T_1}^{T_2}\sqrt{[x'(t)]^2+[y'(t)]^2} \mathrm{d}t \end{equation} l=λ→0limi=1∑nPi−1Pi=λ→0limi=1∑n[x′(ξi)]2+[y′(ξi)]2⋅Δti=∫T1T2[x′(t)]2+[y′(t)]2dt
上式即为弧长的计算公式。对上式求微分即得到弧长的微分:
d l = [ x ′ ( t ) ] 2 + [ y ′ ( t ) ] 2 d t \begin{equation} \mathrm{d}l=\sqrt{[x'(t)]^2+[y'(t)]^2} \mathrm{d}t \end{equation} dl=[x′(t)]2+[y′(t)]2dt
若曲线采用显式方程 y = f ( x ) , x ∈ [ a , b ] y=f(x), x\in[a,b] y=f(x),x∈[a,b]来表示时,我们利用 f ′ ( x ) = y ′ ( t ) x ′ ( t ) f'(x)=\dfrac{y'(t)}{x'(t)} f′(x)=x′(t)y′(t)进行换元,即可得到显式方程下的弧长公式:
l = ∫ a b 1 + [ f ′ ( x ) ] 2 d x \begin{equation} l=\int_{a}^{b}\sqrt{1+[f'(x)]^2}\mathrm{d}x \end{equation} l=∫ab1+[f′(x)]2dx
当曲线采用极坐标方程 r = r ( θ ) , θ ∈ [ α , β ] r=r(\theta),\theta\in[\alpha, \beta] r=r(θ),θ∈[α,β]来表示时,由于 x = r ( θ ) cos θ , y = r ( θ ) sin θ x=r(\theta)\cos \theta, y=r(\theta)\sin \theta x=r(θ)cosθ,y=r(θ)sinθ,对 θ \theta θ求导后有
x ′ ( θ ) = r ′ ( θ ) cos θ − r ( θ ) sin θ , y ′ ( θ ) = r ′ ( θ ) sin θ + r ( θ ) cos θ \begin{equation} x'(\theta)=r'(\theta)\cos \theta-r(\theta)\sin \theta, y'(\theta)=r'(\theta)\sin \theta+r(\theta)\cos \theta \end{equation} x′(θ)=r′(θ)cosθ−r(θ)sinθ,y′(θ)=r′(θ)sinθ+r(θ)cosθ
将其代入 ( 9 ) (9) (9)式进行换元后得到极坐标下的弧长公式:
l = ∫ α β [ r ( θ ) ] 2 + [ r ′ ( θ ) ] 2 d θ \begin{equation} l=\int_{\alpha}^{\beta}\sqrt{[r(\theta)]^2+[r'(\theta)]^2}\mathrm{d}\theta \end{equation} l=∫αβ[r(θ)]2+[r′(θ)]2dθ
采用本文同样的方法,可以证明在三维欧式空间中由参数方程
{ x = x ( t ) , y = y ( t ) , z = z ( t ) , t ∈ [ T 1 , T 2 ] . \begin{cases}x=x(t), \\ y=y(t),\\ z=z(t),\end{cases} t\in [T_1,T_2]. ⎩ ⎨ ⎧x=x(t),y=y(t),z=z(t),t∈[T1,T2].
确定的光滑曲线的弧长为
l = ∫ T 1 T 2 [ x ′ ( t ) ] 2 + [ y ′ ( t ) ] 2 + [ z ′ ( t ) ] 2 d t \begin{equation} l=\int_{T_1}^{T_2}\sqrt{[x'(t)]^2+[y'(t)]^2+[z'(t)]^2} \mathrm{d}t \end{equation} l=∫T1T2[x′(t)]2+[y′(t)]2+[z′(t)]2dt