第四章 数值积分与数值微分
4.1 Newton-cotes求积
将有限区间 [ a , b ] [a,b] [a,b]分成 n n n等份,取等距基点: a ≤ x 0 < x 1 < ⋅ ⋅ ⋅ < x n ≤ b a \leq x_0 <x_1<···<x_n \leq b a≤x0<x1<⋅⋅⋅<xn≤b,则可以先根据这些点求得 f ( x ) f(x) f(x)的插值函数,然后对插值函数进行求导(这样会方便一点的)。
当n=1时:
f
(
x
)
=
x
−
x
1
x
0
−
x
1
f
0
+
x
−
x
0
x
1
−
x
0
f
1
I
=
h
2
(
f
0
+
f
1
)
E
=
−
1
12
h
3
f
′
′
\begin{align} f(x)&=\frac{x-x_1}{x_0-x_1}f_0+\frac{x-x_0}{x_1-x_0}f_1\\ I &= \frac{h}{2}(f_0+f_1)\\ E &= -\frac{1}{12}h^3f^{\prime\prime} \end{align}
f(x)IE=x0−x1x−x1f0+x1−x0x−x0f1=2h(f0+f1)=−121h3f′′
当n = 2时:
f
(
x
)
=
(
x
−
x
1
)
(
x
−
x
2
)
(
x
0
−
x
1
)
(
x
0
−
x
2
)
f
0
+
(
x
−
x
0
)
(
x
−
x
2
)
(
x
1
−
x
0
)
(
x
1
−
x
2
)
f
1
+
(
x
−
x
0
)
(
x
−
x
1
)
(
x
2
−
x
0
)
(
x
2
−
x
1
)
f
2
I
=
h
6
(
f
(
x
0
)
+
4
f
(
x
1
)
+
f
(
x
2
)
)
E
=
−
1
90
h
5
f
(
4
)
(
δ
)
\begin{align} f(x)&=\frac{(x-x_1)(x-x_2)}{(x_0-x_1)(x_0-x_2)}f_0+\frac{(x-x_0)(x-x_2)}{(x_1-x_0)(x_1-x_2)}f_1+\frac{(x-x_0)(x-x_1)}{(x_2-x_0)(x_2-x_1)f_2}\\ I &= \frac{h}{6}(f(x_0)+4f(x_1)+f(x_2))\\ E &= -\frac{1}{90}h^5f^{(4)}(\delta) \end{align}
f(x)IE=(x0−x1)(x0−x2)(x−x1)(x−x2)f0+(x1−x0)(x1−x2)(x−x0)(x−x2)f1+(x2−x0)(x2−x1)f2(x−x0)(x−x1)=6h(f(x0)+4f(x1)+f(x2))=−901h5f(4)(δ)
当n= 4时:
I
=
h
90
[
7
f
(
x
0
)
+
32
f
(
x
1
)
+
12
f
(
x
2
)
+
32
f
(
x
3
)
+
7
f
(
x
4
)
]
E
=
−
8
945
h
7
f
(
6
)
(
δ
)
\begin{align} I &= \frac{h}{90}[7f(x_0)+32f(x_1)+12f(x_2)+32f(x_3)+7f(x_4)]\\ E &= -\frac{8}{945}h^7f^{(6)}(\delta) \end{align}
IE=90h[7f(x0)+32f(x1)+12f(x2)+32f(x3)+7f(x4)]=−9458h7f(6)(δ)
4.2 复合积分
在区间 [ a , b ] [a,b] [a,b]中的小区间里使用积分公式进行积分,然后得到结果
4.3 变步长的求积公式
选定了某种积分方法之后,如何选择适当的 n n n,使得计算结果达到预先选定的精度?
针对于n=2的梯形公式
R
(
f
,
T
n
)
=
−
b
−
a
12
h
2
f
′
′
(
η
n
)
R
(
f
,
T
2
n
)
=
−
b
−
a
12
(
h
2
)
2
f
′
′
(
η
2
n
)
T
2
n
−
T
n
=
−
b
−
a
12
(
h
2
)
2
[
4
f
′
′
(
η
n
)
−
f
′
′
(
η
2
n
)
]
若充分大
,
f
′
(
η
)
在
(
a
,
b
)
上连续且
f
′
′
(
η
n
)
≈
f
′
′
(
η
2
n
)
则
T
2
n
−
T
n
=
3
(
I
−
T
n
)
I
−
T
2
n
=
1
3
(
T
2
n
−
T
n
)
\begin{align} R(f,T_n)&=-\frac{b-a}{12}h^2f^{\prime\prime}(\eta_n)\\ R(f,T_{2n})&=-\frac{b-a}{12}(\frac{h}{2})^2f^{\prime\prime}(\eta_{2n})\\ T_2n - T_n&=-\frac{b-a}{12}(\frac{h}{2})^2[4f^{\prime\prime}(\eta_n)-f^{\prime\prime}(\eta_{2n})]\\ 若充分大&,f^{\prime}(\eta)在(a,b)上连续且f^{\prime\prime}(\eta_n)\thickapprox f^{\prime\prime}(\eta_{2n}) 则\\ T_{2n}-T_n&=3(I-T_n)\\ I-T_{2n}&=\frac{1}{3}(T_{2n}-T_n) \end{align}
R(f,Tn)R(f,T2n)T2n−Tn若充分大T2n−TnI−T2n=−12b−ah2f′′(ηn)=−12b−a(2h)2f′′(η2n)=−12b−a(2h)2[4f′′(ηn)−f′′(η2n)],f′(η)在(a,b)上连续且f′′(ηn)≈f′′(η2n)则=3(I−Tn)=31(T2n−Tn)
同理对于n=3的simpson公式有
I
−
T
2
n
=
1
15
(
T
2
n
−
T
n
)
I-T_{2n} = \frac{1}{15}(T_{2n}-T_n)
I−T2n=151(T2n−Tn)
n=4的cotes公式有
I
−
T
2
n
=
1
63
(
T
2
n
−
T
n
)
I-T_{2n}=\frac{1}{63}(T_{2n}-T_n)
I−T2n=631(T2n−Tn)
4.4 龙贝格公式
4.5 Gauss-Legendre求积公式
定义:
求积公式具有 2 n + 1 2n+1 2n+1次的代数精度,则称该求积公式为Gauss型求积公式。
节点的来源: G a u s s − L e g e n d r e Gauss-Legendre Gauss−Legendre公式的根
G
u
a
s
s
−
L
e
g
e
n
d
r
e
Guass-Legendre
Guass−Legendre公式:
P
0
(
x
)
=
1
P
1
(
x
)
=
x
P
2
(
x
)
=
1
2
(
3
x
2
−
1
)
…
…
P
n
(
x
)
=
2
n
−
1
n
x
P
n
−
1
(
x
)
−
n
−
1
n
P
n
−
2
(
x
)
\begin{align} P_0(x)&=1\\ P_1(x)&=x\\ P_2(x)&=\frac{1}{2}(3x^2-1)\\ &……\\ P_n(x)&=\frac{2n-1}{n}xP_{n-1}(x)-\frac{n-1}{n}P_{n-2}(x) \end{align}
P0(x)P1(x)P2(x)Pn(x)=1=x=21(3x2−1)……=n2n−1xPn−1(x)−nn−1Pn−2(x)