Sei sulla pagina 1di 3

Lecture -3

Simpson’s one-third Rule:


Putting n=2 in the formula (A) and similar to n=1, neglecting third and higher order
differences in (A), we get
x2 x2  s  s  1 2 
 f ( x) dx    f 0  sf 0  2!  f 0  dxs
x0 x0  
2 s2  s 2 
 h   f 0  s f 0   f0 ds
0  2! 
2
 s2 1  s3 s 2  
 h  sf 0   f 0      2 f0 
 2 2!  3 2  0
h
  f 0  4 f1  f2 
3
Similarly,
x4 2
 s  s 1 2  h
 f ( x ) dx  h   f 2  s f 2  2!  f 2  ds  3  f 2  4 f 3  f 4  and so on
x2 0  
xn h
 f  x  dx  3  fn2  4 f n1  fn 
xn2
If the number of intervals is even i.e. divisible by 2, we see that after adding these integrals
xn h
 f  x  dx  3  f0  4  f1  f 3  fn 1   2  f 2  f 4  ...  f n  2   f n ]
x0
This is known as Simpson’s one-third rule (1/3 rd S.R.)
x2
 c3 h f1 "' 1  dx, 1   x0 , x2 
3
Now the error term of rd S. R. in one step = S

x0
2 s(s 1)(s  2)
 h f "'1  
4
ds
0 3!
 h f "' 1   zero  zero
4

So, to get the error, we must integrate the succeeding term i.e.
x2
 h  sc4 h f iv 1  dx , x0  1  x2
4
Local error
x0
2 s( s 1)( s  2)( s  3)
 h f iv 1  
5
ds
0 4!
5
h iv
 f 1   O(h )
5

90
i.e. Local error of rd S. R. is of order (h5 ).
Global error of rd S. R means, the sum of all local errors, which is

 h5
90 
    f ( )  ...  f   , xi1  
iv iv iv
f 1 3 n 1 i  xi 1 , i  1,3,5,..., n  1

5
h n iv
   f   , x0    xn
90 2
b  a  h f  , using nh = b - a, i.e. O  h  .
  
4 iv 4

180
1
Note : rd S.R. requires the even number of intervals or odd number of ordinates.
3

Simpson’s Three-eighth Rule:


Putting = 3 in the formula (A) i.e. considering four points in one step or three intervals in a
step. Obviously, one can fit a third degree polynomial through four points or we can say only
third order differences can be obtained from these four points. So, neglecting all differences
above the third, we get
x3 3
 s(s 1) 2 s  s 1 s  2  3 
 f ( x ) dx  h  0
 f  s  f0   f0   f0 ds
x0 0   2! 3! 

 h  f0  3  f1  f2   f3 
3
8
Similarly,
x6
 f ( x) dx  8 h  f3  3  f4  f5   f6  and so on
3
x3

xn
 f ( x) dx  8 h  f n3  3  f n2  f n1   f n 
3
xn3
If the number of intervals is divisible by 3, we see that the sum of these integrals gives
xn 3
 f ( x) dx  8 h  f0  3 f1  f2  f4  f5  ...  fn2  fn1   2 f3  f6  ...  fn3   fn 
x 0

3 
Which is known as Simpson’s three-eighth rule  th S.R. 
8 
3 
The error terms of  th S.R.  in one step
8 
x3 s 4 iv
  c4 h f 1  dx, x0  1  x3
x0
3
s  s 1 s  2  s  3
 h f 1  
5 iv
ds
0 4!
5
3h iv
 f 1   O  h 
5

80
3
i.e. local error of th S.R. is of order (h5).
8
3
Now, to get the global error of th S.R. , we have to add all the local errors i.e.
8

Global error   3 h5  f iv 1   f iv  4   f iv  7   ...  f iv  n  2  , xi 1  i  xi  2 , i  1,3,..., n  2


80
5
3h n iv
  f   , x0    xn
80 3

b  a  h4 f iv ( )  O(h4 )
80
3
Note: th S.R. requires the number of divisions of the interval [a, b.] in a multiple of 3.
8

Potrebbero piacerti anche