计算方法（五）：数值微分与数值积分

数值微分与数值积分

数值微分

根据函数在一些离散点上的函数值，求该函数导数的近似值称为数值微分。

利用插值多项式构造数值微分公式

插值函数法：设 $P_n(x)$ 是 $f(x)$ 的 $n$ 次插值多项式，在插值区间 $[a,b]$ 内，一种直观的数值微分公式是：
$$f^{(k)}(x)\approx P_n^{(k)}(x),1\le k\le n,a\le x\le b$$
称为插值型求导公式。

误差分析：
$$\begin{gathered} R_n(x)=f(x)-P_n(x)=\frac{f^{(n+1)}(\xi )}{(n+1)!}(x-x_0)(x-x_1)\cdots (x-x_n) \\ {f}^{\prime }(x)-P_n^{\prime }(x)=\frac{f^{(n+1)}(\xi )}{(n+1)!}{\omega }_{n+1}^{\prime }(x)+\frac{{\omega }_{n+1}(x)}{(n+1)!}\frac{d}{dx}{f}^{(n+1)}(\xi ),其中,{\omega }_{n+1}(x)=(x-x_0)(x-x_1)\cdots (x-x_n) \\ x=x_k时,f^{\prime }(x_k)-P_n^{\prime }(x_k)=\frac{f^{(n+1)}(\xi )}{(n+1)!}{\omega }_{n+1}^{\prime }(x_k) \\ {f}^{\prime \prime }(x)-P_n^{\prime \prime }(x)=\frac{f^{(n+1)}(\xi )}{(n+1)!}{\omega }_{n+1}^{\prime \prime }(x)+2\frac{{\omega }_{n+1}^{\prime }(x)}{(n+1)!}\frac{d}{dx}{f}^{(n+1)}(\xi )+\frac{{\omega }_{n+1}(x)}{(n+1)!}\frac{d^2}{d{x}^2}{f}^{(n+1)}(\xi ) \\ x=x_k时,f^{\prime \prime }(x_k)-P_n^{\prime \prime }(x_k)=\frac{f^{(n+1)}(\xi )}{(n+1)!}{\omega }_{n+1}^{\prime \prime }(x_k)+2\frac{{\omega }_{n+1}^{\prime }(x_k)}{(n+1)!}\frac{d}{dx}{f}^{(n+1)}(\xi ) \end{gathered}$$

等距结点处的数值微分公式

1. 两点公式：设 $x_0<x_1$，则
   $$\begin{gathered} P_1(x)=f(x_0)\frac{x-x_1}{x_0-x_1}+f(x_1)\frac{x-x_0}{x_1-x_0} \\ {f}^{\prime }(x)\approx P_1^{\prime }(x)=\frac{f(x_1)-f(x_0)}{h},h=x_1-x_0,x_0\le x\le x_1 \\ {f}^{\prime }(x_0)\approx \frac{f(x_1)-f(x_0)}{h} \\ {f}^{\prime }(x_1)\approx \frac{f(x_1)-f(x_0)}{h} \\ {\omega }_2(x)=(x-x_0)(x-x_1) \\ {f}^{\prime }(x_0)-P_1^{\prime }(x_0)=\frac{f^{\prime \prime }({\xi }_0)}{2}{\omega }_2^{\prime }(x_0)=\frac{f^{\prime \prime }({\xi }_0)}{2}(x_0-x_1)=-\frac{h}{2}{f}^{\prime \prime }({\xi }_0) \\ {f}^{\prime }(x_1)-P_1^{\prime }(x_1)=\frac{f^{\prime \prime }({\xi }_1)}{2}{\omega }_2^{\prime }(x_1)=\frac{f^{\prime \prime }({\xi }_1)}{2}(x_1-x_0)=\frac{h}{2}{f}^{\prime \prime }({\xi }_1),x_0<{\xi }_0,{\xi }_1<x_1 \end{gathered}$$

2. 三点公式：设 $x_0<x_1<x_2,x_1=x_0+h,x_2=x_0+2h$，则
   $$\begin{gathered} P_2(x)=\frac{(x-x_1)(x-x_2)}{(x_0-x_1)(x_0-x_2)}f(x_0)+\frac{(x-x_0)(x-x_2)}{(x_1-x_0)(x_1-x_2)}f(x_1)+\frac{(x-x_0)(x-x_1)}{(x_2-x_0)(x_2-x_1)}f(x_2) \\ {f}^{\prime }(x)\approx P_2^{\prime }(x)=\frac{2x-x_1-x_2}{2h^2}f(x_0)-\frac{2x-x_0-x_2}{h^2}f(x_1)+\frac{2x-x_0-x_1}{2h^2}f(x_2) \\ {f}^{\prime }(x_0)\approx \frac{-3f(x_0)+4f(x_1)-f(x_2)}{2h} \\ {f}^{\prime }(x_1)\approx \frac{-f(x_0)+f(x_2)}{2h} \\ {f}^{\prime }(x_2)=\frac{f(x_0)-4f(x_1)+3f(x_2)}{2h} \\ {\omega }_3(x)=(x-x_0)(x-x_1)(x-x_2),{\omega }_3^{\prime }(x)=(2x-x_0-x_1)(x-x_2)+(x-x_0)(x-x_1) \\ {f}^{\prime }(x_0)-P_2^{\prime }(x_0)=\frac{f^{\prime \prime \prime }({\xi }_0)}{6}{\omega }_3^{\prime }(x_0)=\frac{f^{\prime \prime \prime }({\xi }_0)}{6}2h^2=\frac{h^2}{3}{f}^{\prime \prime \prime }({\xi }_0) \\ {f}^{\prime }(x_1)-P_2^{\prime }(x_1)=\frac{f^{\prime \prime \prime }({\xi }_1)}{6}{\omega }_3^{\prime }(x_1)=-\frac{h^2}{6}{f}^{\prime \prime \prime }({\xi }_1) \\ {f}^{\prime }(x_2)-P_2^{\prime }(x_2)=\frac{f^{\prime \prime \prime }({\xi }_2)}{6}{\omega }_3^{\prime }(x_2)=\frac{f^{\prime \prime \prime }({\xi }_0)}{6}2h^2=\frac{h^2}{3}{f}^{\prime \prime \prime }({\xi }_2),x_0\le {\xi }_0,{\xi }_1,{\xi }_2\le x_2 \end{gathered}$$
   二阶数值微分公式：
   f ′ ′ ( x i ) ≈ P 2 ′ ′ ( x i ) = d d x [ 2 x − x 1 − x 2 2 h 2 f ( x 0 ) − 2 x − x 0 − x 2 h 2 f ( x 1 ) + 2 x − x 0 − x 1 2 h 2 f ( x 2 ) ] ∣ x = x i = f ( x 0 ) − 2 f ( x 1 ) + f ( x 2 ) h 2 , i = 0 , 1 , 2 w 3 ′ ′ ( x ) = 2 ( 3 x − x 0 − x 1 − x 2 ) f ′ ′ ( x 0 ) − P 2 ′ ′ ( x 0 ) = f ′ ′ ′ ( ξ 0 ) 6 ω 3 ′ ′ ( x 0 ) + 2 ω 3 ′ ( x 0 ) 6 d d x f ′ ′ ′ ( ξ 1 ) = − h f ′ ′ ′ ( ξ 0 ) + h 2 6 f ( 4 ) ( ξ 1 ) f ′ ′ ( x 1 ) − P 2 ′ ′ ( x 1 ) = f ′ ′ ′ ( ξ 2 ) 6 ω 3 ′ ′ ( x 1 ) + 2 ω 3 ′ ( x 1 ) 6 d d x f ′ ′ ′ ( ξ 3 ) = − 1 12 f ( 4 ) ( ξ 3 ) f ′ ′ ( x 2 ) − P 2 ′ ′ ( x 2 ) = f ′ ′ ′ ( ξ 4 ) 6 ω 3 ′ ′ ( x 2 ) + 2 ω 3 ′ ( x 2 ) 6 d d x f ′ ′ ′ ( ξ 5 ) = h f ′ ′ ′ ( ξ 4 ) − h 2 6 f ( 4 ) ( ξ 5 ) f''(x_i)\approx P_2''(x_i)=\frac{d}{dx}[\frac{2x-x_1-x_2}{2h^2}f(x_0)-\frac{2x-x_0-x_2}{h^2}f(x_1)+\frac{2x-x_0-x_1}{2h^2}f(x_2)]|_{x=x_i}\\=\frac{f(x_0)-2f(x_1)+f(x_2)}{h^2},i=0,1,2\\w_3''(x)=2(3x-x_0-x_1-x_2)\\f''(x_0)-P_2''(x_0)=\frac{f'''(\xi_0)}{6}\omega''_{3}(x_0)+2\frac{\omega'_{3}(x_0)}{6}\frac{d}{dx}f'''(\xi_1)=-hf'''(\xi_0)+\frac{h^2}{6}f^{(4)}(\xi_1)\\f''(x_1)-P_2''(x_1)=\frac{f'''(\xi_2)}{6}\omega''_{3}(x_1)+2\frac{\omega'_{3}(x_1)}{6}\frac{d}{dx}f'''(\xi_3)=-\frac{1}{12}f^{(4)}(\xi_3)\\f''(x_2)-P_2''(x_2)=\frac{f'''(\xi_4)}{6}\omega''_{3}(x_2)+2\frac{\omega'_{3}(x_2)}{6}\frac{d}{dx}f'''(\xi_5)=hf'''(\xi_4)-\frac{h^2}{6}f^{(4)}(\xi_5) f′′(xi​)≈P2′′​(xi​)=dxd​[2h22x−x1​−x2​​f(x0​)−h22x−x0​−x2​​f(x1​)+2h22x−x0​−x1​​f(x2​)]∣x=xi​​=h2f(x0​)−2f(x1​)+f(x2​)​,i=0,1,2w3′′​(x)=2(3x−x0​−x1​−x2​)f′′(x