Chapter 10 Numerical Integration PDF

5/15/2016
Numerical Integration
Trapezoidal and Simpson Rule

Romberg Formula
prepared by Razana Alwee
Many definite integrals of interest can’t be evaluated

analytically.
Probably the best-known example is the integral that
gives the area under the standard bell-shaped curve,
given by x
1 1 z 2 2
F ( x)    e dz
2 2 0
This integral appears very frequently in probability and

statistics and is extensively tabulated.
prepared by Razana Alwee 1

5/15/2016
The integral is tabulated because it is a fact that there is

no way to express the antiderivative of exp(-z2) in terms
of elementary functions.
Because of cases like this we need methods to perform
approximate integration; for other cases it may be more
convenient to use a numerical method than a symbolic
one.
Often we’re given values of f(x) at various points but not
a formula for f and so have no choice but to use a
numerical method (for example, data from an
experiment)
The Newton-Cotes formula

Trapezoidal Rule
Simpson’s Rule
Romberg Integration
Gaussian Quadrature

5/15/2016
Trapezoidal Rule
The trapezoidal rule is equivalent to approximating the

area of the trapezoid under the straight line connecting
f(a) and f(b).
f(b)
f(a)
a b
Trapezoidal Rule
One way to improve the accuracy of the trapezoid rule is

to divide the integration interval from a to b into a
number of segments (N).
The areas of individual segment can then be added to
yield the integral for the entire interval.
There are N+1 equally spaced nodes
x0, x1, …., xN

5/15/2016
Trapezoidal Rule
Let, x0=a and xN=b
Assume
xk  x0  kh, k  0,1,2,..., N
are equally spaced nodes

ba
h
N
Trapezoidal Rule
Therefore
b x1 x2 xN
 f ( x) dx   f ( x) dx   f ( x) dx  ...   f ( x) dx
a x0 x1 x N 1
x k 1 x k 1
with
 xk
f ( x ) dx  
xk
p n ( x ) dx

5/15/2016
Trapezoidal Rule
Let
pn ( x)  f k  rf k
with
x  xk  rh (dx  hdr)
Trapezoidal Rule
x k 1 xk 1
 f ( x) dx   p n ( x) dx
xk xk
xk  h

xk
 f k  r f k  dx
1
 h
0
 f k  rf k  dr
10

5/15/2016
Trapezoidal Rule
1
 r2   1 
 h rf k  f k   h f k  f k 
 2 0  2 
 1 
 h  f k   f k 1  f k 
 2 
h
  f k  f k 1 
2
11
Trapezoidal Rule
h
 f 0  f 1   h  f 1  f 2   ...  h  f N 1  f N 
b
 f ( x) dx 
a 2 2 2
h
  f 0  2 f1  2 f 2  ...  2 f N 1  f N 
2
N 1
b h 
 f ( x) dx   f 0  f N  2 f i 
a 2 i 1 
12

5/15/2016
Example
Approximate the following integral using the Trapezoidal

rule with h=0.5 and h=0.25.
4 x
 dx
1
x4
13
Example
For h=0.5, a=1 and b=4
b  a 4 1
N  6
h 0.5
14

5/15/2016
Example
xi
i xi f ( xi ) 
xi  4
0 1 .0 0.4472
1 1 .5 0.6396
2 2. 0 0.8165
3 2 .5 0.9806
4 3. 0 1.1339
5 3 .5 1.2780
6 4. 0 1.4142
Total 1.8614
4.8486
15
Example
4 x 4
 1
x4
dx   f ( x) dx
1
h
 f  f 6  2 f 1  f 2  f 3  f 4  f 5 
2 0

0 .5
 1.8614  24.8486
2
 2.8896
16

5/15/2016
Example
For h=0.25, a=1 and b=4
b  a 4 1
N   12
h 0.25
17
xi
i xi f ( xi ) 
Example
xi  4
0 1 . 00 0 . 4472
1 1 . 25 0 . 5455
2 1 . 50 0 . 6396
3 1 . 75 0 . 7298
4 2 . 00 0 . 8165
5 2 . 25 0 . 9000
6 2 . 50 0 . 9806
7 2 . 75 1 . 0585
8 3 . 00 1 . 1339
9 3 . 25 1 . 2070
10 3 . 50 1 . 2780
11 3 . 75 1 . 3470
12 4prepared
. 00 by Razana Alwee 1 . 4142
Total 1 . 8614 10 . 6364 18

5/15/2016
Example
4 x 4
 1
x4
dx   f ( x) dx
1
11
h 
  f 0  f 12  2 f i 
2 i 1 
0.25
 1.8614  210.6364
2
 2.8918
(Note: The exact value of the integral is 2.8925)

19
Example
For h=0.5, N=6, the error is 0.0029

For h=0.25, N=12, the error is 0.0007
The error decreases as the number of segments (N)

increases.
When h is reduced by a factor of ½ the successive errors
are diminished by approximately ¼.
20

5/15/2016
Exercise
Approximate the following integrals using the

Trapezoidal rule with N=10.
2 2
(a)  e x dx (b)  x 3  1 dx
0 1
21
Simpson’s Rule
The trapezoidal rule usually requires a large number of

function evaluations to achieve an accurate answer.
Another way to obtain a more accurate estimate of an
integral is to use higher-order polynomials to connect the
points.
22

5/15/2016
Simpson’s Rule
Let
1
p2 ( x )  f k  rf k  r (r  1) 2 f k
2
23
Simpson’s Rule
xk  2 xk  2 h  1 2 
 f ( x) dx    f k  rf k  r ( r  1)  f k  dx
xk xk
 2 
2  1 2 
 h  f k  rf k  r ( r  1)  f k  dr
0
 2 
2
 r2 1  r3 r2  
 h rf k  f k     2 f k 
 2 2 3 2  0
24

5/15/2016
Simpson’s Rule
 1 
 h 2 f k  2 f k  2 f k 
 3 
 1 
 h 2 f k  2 f k 1  f k    f k 2  2 f k 1  f k 
 3 
h
  f  4 f k 1  f k  2 
3 k
25
Simpson’s Rule
Subdivide the interval [a,b] into N subintervals (N is

even).
b x2 x4
 f ( x) dx   f ( x) dx   f ( x ) dx
a x0 x2
xN
 ...   f ( x) dx
xN 2
h
  f  4 f 1  2 f 2  4 f 3  2 f 4 ...4 f N 1  f N 
3 0
26

5/15/2016
Simpson’s Rule
This equation is known as Simpson’s 1/3 rule.
N /2 N / 2 1
b h 
 f ( x) dx   f 0  f N   4  f 2 i 1  2  f 2i 
a 3 i 1 i 1 
Where N is even.
27
Simpson’s 3/8 Rule
Let,
1 1
p3 ( x)  f k  rf k  r ( r  1) 2 f k  r (r  1)( r  2) 3 f k
2 6
xk  3 xk  3
 xk
f ( x ) dx  
xk
p3 ( x ) dx
28

5/15/2016
Simpson’s 3/8 Rule
N /3 N / 3 1
b 3h  
 f ( x) dx   f 0  f N   3  f 3i  2  f 3i 1   2  f 3i 
a 8  i 1 i 1 
where N= 3, 6, 9, 12, ..…
This equation is called Simpson’s 3/8 rule because h is

multiplied by 3/8.
29
Example
Approximate the following integrals using the Simpson’s

1/3 rule with h=0.5 and h=0.25.
4 x
 dx
1
x4
30

5/15/2016
Example
For h=0.5, a=1 and b=4
b  a 4 1 (N is even)
N  6
h 0 .5
31
Example
xi
i xi f i  f ( xi ) 
xi  4
0 1.0 0.4472
1 1.5 0.6396
2 2.0 0.8165
3 2.5 0.9806
4 3.0 1.1339
5 3.5 1.2780
6 4.0 1.4142
Total 1.8614 prepared by Razana Alwee 2.8982 1.9504
32

5/15/2016
Example
3 2
4 x h 
 dx   f 0  f 6   4 f 2 i 1  2 f 2i 
1
x4 3 i 1 i 1 
0.5
 1.8614  4(2.8982)  2(1.9504)
3
 2.8925
33
Example
For h=0.25, a=1 and b=4
b  a 4 1
N   12 (N is even)
h 0.25
34

5/15/2016
xi
i xi f i  f ( xi ) 
xi  4
Example
0 1 .00 0 .4472
1 1 .25 0 .5455
2 1 .50 0 .6396
3 1 .75 0 .7298
4 2 .00 0 . 8165
5 2 . 25 0 . 9000
6 2 .50 0 .9806
7 2 . 75 1 . 0585
8 3 .00 1 .1339
9 3 . 25 1 .2070
10 3 .50 1 .2780
11 3 . 75 1 .3470
12 4 .00 1 .4142
Total 1 .8614 5 . 7878 4 . 8486
35
Example
6 5
4 x h 
 dx   f  f   4  f  2 f
3  
1 0 12 2 i 1 2 i
x4 i 1 i 1 
0.25
 1.8614  4(5.7878)  2(4.8486)
3
 2.8925
36

5/15/2016
Example
Approximate the following integrals using the Simpson’s

3/8 rule with h=0.25.
4 x
 dx
1
x4
37
Example
For h=0.25, a=1 and b=4
b  a 4 1
N   12
h 0.25
38

5/15/2016
xi
i xi f i  f ( xi ) 
xi  4
0 1 . 00 0 .4472 Example
1 1 .25 0 .5455
2 1 . 50 0 .6396
3 1 .75 0 . 7298
4 2 . 00 0 .8165
5 2 .25 0 .9000
6 2 . 50 0 .9806
7 2 .75 1 . 0586
8 3 . 00 1 . 1339
9 3 .25 1 . 2070
10 3 . 50 1 . 2780
11 3 .75 1 . 3470
12 4 . 00 1 .4142
Total 1 .8614prepared by Razana Alwee
7 .7191 2 . 9174
39
Example
4 3
4 x 3h  
 dx   f  f   3  f  f   2  f
8  
1 0 12 3i  2 3i 1 3i
x4 i 1 i 1 
3(0.25)
 1.8614  3(7.7191)  2(2.9174)
8
 2.8925
40

5/15/2016
Simpson’s Rule
Simpson’s 1/3 rule is usually the method of preference

because it attains third-order accuracy with three points
rather than the four points required for the 3/8 version.
41
Exercise
Approximate the following integral
3 1
 dx
0
x3 1
using
a) The Simpson’s 1/3 rule
b) The Simpson’s 3/8 rule
with N=12.
42

5/15/2016
Romberg Integration
Romberg integration uses the Trapezoidal rule to give

preliminary approximations and then applies the
Richardson extrapolation process to improve the
approximations.
43
Romberg Integration
The first step in the Romberg process obtains the

Trapezoidal rule approximations.
Start with 1 subinterval,
h1  b  a
h1
R1,1   f 0  f1 
2
44

5/15/2016
Romberg Integration
For Ni subintervals, Ni=2i-1 (eg. N2=2, N3=4, N4=8)
1 ba
hi  hi 1 or hi 
2 2i 1
i 2
1 2 
Ri ,1   Ri 1,1  hi 1  f 2 k 1 
2 k 1 
i=2, 3, …
45
Romberg Integration
Example: for i=2, N2=22-1=2
1
h2  h
2 1
1 
R2 ,1  h2   f 0  f 2   f 1 
2 
46

5/15/2016
Romberg Integration
1  h1 1 
R2 ,1  h2   f 0  f 2   f 1    2  f 0  f 2   f 1 
2  2
1  h1  1
 
2  2 0
f  f 2   h f
1 1
 R1,1  h1 f 1
 2
 
1
1 
  R1,1  h1  f 2 k 1 
2 k 1 
47
Romberg Integration
Example: for i=3, N3=23-1=4
1
h3  h
2 2
1 
R3,1  h3   f 0  f 4   f 1  f 2  f 3 
2 
48

5/15/2016
Romberg Integration
1 
R3,1  h3   f 0  f 4   f 1  f 2  f 3 
2 
h 1 
 2   f 0  f 4   f1  f 2  f 3 
2 2 
1 h 
  2  f 0  f 4   h2 f 2  h2  f 1  f 3 
22 
2
1 1 
2

 R2 ,1  h2  f 1  f 3   
2  R 2 ,1  h 2  f 2 k 1 
k 1 
49
Romberg Integration
i2
1 2 
Ri ,1   Ri 1,1  hi 1  f 2 k 1 
2 k 1 
For i=2, 3, …
1
hi 1  h
2 i 2
50

5/15/2016
Romberg Integration
Richardson’s improvement,
4 j 1 Ri , j 1  Ri 1, j 1
Ri , j 
4 j 1  1
for i = 2, 3, …, N and j = 2, 3 …, i.
51
Romberg Integration
4 R2 ,1  R1,1 4 R3,1  R2 ,1
R2 ,2  R3,2 
3 3
16 R3,2  R2 ,2 64 R4 ,3  R3,3
R3,3  R4 ,4 
15 63
52

5/15/2016
Romberg Table
ba
i hi  Ri ,1 Ri , 2 Ri , 3 ... Ri , N
2 i 1
1 h1 R1,1
2 h2 R2,1 R 2, 2
3 h3 R3,1 R3, 2 R3,3
... ... ... ... ... ...
N hN R N ,1 RN ,2 R N ,3 ... RN , N
Compute the Romberg table until |Ri,j - Ri,j-1|<

53
Example
Use Romberg integration to approximate
4 x
 dx
1
x4
Compute the Romberg table until |Ri,j  Ri,j-1|<0.0005
54

5/15/2016
Example
h1  b  a  4  1  3
h1
R1,1   f 0  f1 
2
3
 0.4472  1.4142
2
 2.7921.
55
Example
1
1 
R2 ,1   R1,1  h1  f 2 k 1 
2 k 1 
1
 R1,1  h1 ( f1 )  1 R1,1  h1 ( f (2.5))
2 2
1
 2.7921  30.9806  2.8670
2
56

5/15/2016
Example
4 R2 ,1  R1,1 42.8670  2.7921

R2 , 2    2.8920
3 3
R2 , 2  R2,1  2.8920  2.8670  0.025  0.0005
57
Example
2
1  1
R3,1   R 2 ,1  h2  f 2 k 1   R2,1  h2 ( f1  f 3 )
2 k 1  2
1
 R2,1  h2 ( f (1.75)  f (3.25))
2
1
 2.8670  1.50.7298  1.2070  2.8861
2
58

5/15/2016
Example
4 R3,1  R2,1 42.8861  2.8670

R3, 2    2.8925
3 3
R3, 2  R3,1  2.8925  2.8861  0.0064  0.0005
59
Example
16 R3, 2  R2, 2 162.8925  2.8920

R3,3    2.8925
15 15
R3,3  R3, 2  2.8925  2.8925  0.0000  0.0005
60

5/15/2016
Example
The solution is,
4 x
 1
x4
dx  R3,3  2.8925
61
Example
The Romberg Table,
ba
i hi  Ri ,1 Ri , 2 Ri ,3
2i 1
1 3 2.7921
2 1 .5 2.8670 2.8920
3 0.75 2.8861 2.8925 2.8925
62

5/15/2016
Exercise
Use Romberg integration to approximate:
2 
(a)  ( 4  x 2 ) 1 2 dx (b)  sin x dx
0 0
Compute the Romberg table until |Ri,j  Ri,j-1|<0.005
63
Exercise
Use the two point and three point Gaussian Quadrature

to approximate:
2 1 1
(a)  x 3  1 dx (b)  dx
1 1 1 x2
Do calculations in 4 decimal points.
64

Chapter 10 Numerical Integration PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chapter 10 Numerical Integration PDF

Uploaded by

Copyright:

Available Formats

5/15/2016

Trapezoidal and Simpson Rule

prepared by Razana Alwee

Many definite integrals of interest can’t be evaluated

This integral appears very frequently in probability and

prepared by Razana Alwee 1

The integral is tabulated because it is a fact that there is

prepared by Razana Alwee

The Newton-Cotes formula

prepared by Razana Alwee

prepared by Razana Alwee 2

The trapezoidal rule is equivalent to approximating the

One way to improve the accuracy of the trapezoid rule is

prepared by Razana Alwee

prepared by Razana Alwee 3

Let, x0=a and xN=b

are equally spaced nodes

prepared by Razana Alwee

prepared by Razana Alwee

prepared by Razana Alwee 4

prepared by Razana Alwee

prepared by Razana Alwee

prepared by Razana Alwee 5

prepared by Razana Alwee

prepared by Razana Alwee

prepared by Razana Alwee 6

Approximate the following integral using the Trapezoidal

prepared by Razana Alwee

For h=0.5, a=1 and b=4

prepared by Razana Alwee

prepared by Razana Alwee 7

prepared by Razana Alwee

prepared by Razana Alwee 8

For h=0.25, a=1 and b=4

prepared by Razana Alwee

prepared by Razana Alwee 9

(Note: The exact value of the integral is 2.8925)

For h=0.5, N=6, the error is 0.0029

The error decreases as the number of segments (N)

prepared by Razana Alwee

prepared by Razana Alwee 10

Approximate the following integrals using the

prepared by Razana Alwee

The trapezoidal rule usually requires a large number of

prepared by Razana Alwee

prepared by Razana Alwee 11

prepared by Razana Alwee

prepared by Razana Alwee

prepared by Razana Alwee 12

prepared by Razana Alwee

Subdivide the interval [a,b] into N subintervals (N is

prepared by Razana Alwee

prepared by Razana Alwee 13

This equation is known as Simpson’s 1/3 rule.

prepared by Razana Alwee

Simpson’s 3/8 Rule

prepared by Razana Alwee

prepared by Razana Alwee 14

Simpson’s 3/8 Rule

where N= 3, 6, 9, 12, ..…

This equation is called Simpson’s 3/8 rule because h is

prepared by Razana Alwee

Approximate the following integrals using the Simpson’s