Newton’s backward interpolation





Newton’s backward interpolation

Suppose \(y=f( x )\)is tabulated for equally spaced valued
\( x=x_0,x_0+h,x_0+2h,\ldots ,x_0+nh \)
given by
\( y=y_0,y_1,y_2,\ldots ,y_n \)
Here,
\( x_i=x_0+ih \) for \( i=0,1,2,\ldots ,n \)
Now, assuming \(y=f( x )\) as a polynomial of degree n in x, we write
\( y=a_n+a_{n-1}( x-x_n )+a_{n-2}( x-x_n )( x-x_{n-1} )+\ldots +a_0( x-x_n )\ldots ( x-x_1 )\)


Putting, \( x=x_n\), we get
\( y=a_n+a_{n-1}( x-x_n )+a_{n-2}( x-x_n )( x-x_{n-1} )+\ldots +a_0( x-x_n )\ldots ( x-x_1 )\)
or \( y_n=a_n+a_{n-1}( \) \(x_n\) \(-x_n )+a_{n-2}( \) \(x_n\) \( -x_n )( \) \(x_n\) \(-x_{n-1} )+\ldots +a_0( \) \(x_n\) \( -x_n )\ldots ( \) \(x_n\) \(-x_1 )\)
or \(y_n=a_n+0+...\)
or \(y_n=a_n\)


Putting, \( x=x_{n-1}\), we get
\( y=a_n+a_{n-1}( x-x_n )+a_{n-2}( x-x_n )( x-x_{n-1} )+\ldots +a_0( x-x_n )\ldots ( x-x_1 )\)
or \( y_{n-1}=a_n+a_{n-1}( \) \(x_{n-1}\) \(-x_n )+a_{n-2}( \) \(x_{n-1}\) \( -x_n )(\) \(x_{n-1}\) \(-x_{n-1} )+\ldots \)
\(+a_0( \) \(x_{n-1}\) \( -x_n ) ( \) \(x_{n-1}\) \(-x_{n-1} )\)

or \(y_{n-1}=a_n+a_{n-1}( x_{n-1}-x_n )+0+...\)
or \(y_{n-1}=a_n+a_{n-1}( -h)\)
or \(y_{n-1}=y_n-a_{n-1}h\)
or \(a_{n-1}=\frac{y_n-y_{n-1}}{h}\)
or \(a_{n-1}=\frac{\nabla {y_n}}{h}\)


Putting, \( x=x_{n-2}\), we get
\( y=a_n+a_{n-1}( x-x_n )+a_{n-2}( x-x_n )( x-x_{n-1} )+\ldots +a_0( x-x_n )\ldots ( x-x_1 )\)
or \( y_{n-2}=a_n+a_{n-1}( \) \(x_{n-2}\) \(-x_n )+a_{n-2}( \) \(x_{n-2}\) \( -x_n )(\) \(x_{n-2}\) \(-x_{n-1} )\)
\(+a_{n-3}( \) \(x_{n-2}\) \( -x_n ) ( \) \(x_{n-2}\) \(-x_{n-1} )(\)
\( x_{n-2}\) \( - x_{n-2})+...\)
or \(y_{n-2}=a_n+a_{n-1}( x_{n-2}-x_n )+ a_{n-2}( x_{n-2}-x_n )( x_{n-2}-x_{n-1} )+0+...\)
or \(y_{n-2}=a_n+a_{n-1}( x_{n-2}-x_n )+ a_{n-2}( x_{n-2}-x_n )( x_{n-2}-x_{n-1} )\)
or \(y_{n-2}=a_n+a_{n-1}( 2h )+ a_{n-2}(2h)(h)\)
or \(y_{n-2}=2a_n+a_{n-1}( h )+2 a_{n-2}(h^2)\)
or \(a_{n-2}=\frac{\nabla ^2y_n}{2!h^2}\)


Similarly, we get
\(a_{n-i}=\frac{\nabla^i y_n}{i!h^i}\) for \( i=0,1,2,\ldots ,n \)


Setting , \( x={x_n}+ph\) and substituting, \( a_i\) for \( i=0,1,2,\ldots ,n \) we get
\( y=a_n+a_{n-1}( x-x_n )+a_{n-2}( x-x_n )( x-x_{n-1} )+\ldots +a_0( x-x_n )\ldots ( x-x_1 )\)
or \( y=y_n+\frac{\nabla y_n}{h}[ ph ]+ \frac{\nabla ^2y_n}{2!h^2} [ p( p+1 )h^2 ]+...\)
or \( y=y_n+\nabla y_n[ p ]+ \frac{\nabla ^2y_n}{2!} [ p( p+1 )]+...\)


This is Newton’s backward difference interpolation formula and it is useful for interpolation near the end of a set of tabular values.




Example 1

The population of a town was as given below. Estimate the population for the year 1925.

x 1891 1901 1911 1921 1931
y 46 66 81 93 101

Solution
According to given set of data values, we form difference table as below

X Y 1st diff 2nd diff 3rd diff 4th diff
1891 46
66-46=20
1901 66 15-20=-5
81-66=15 (-3)-(-5)=2
1911 81 12-15=-3 (-1)-(2)=-3
93-81=12
(-4)-(-3)=-1
1921 93 8-12=-4
101-93=8
1931 101

Here,
\( h=10\) and \( x_n=1931\)
Thus, we have
\( x=x_n+ph\)
or \( 1925=1931+p\times 10\)
or \( p=-0.6\)
Using formula, we get
\( y=y_n+\nabla y_n[ p ]+ \frac{\nabla ^2y_n}{2!} [ p( p+1 )]+...\)
or \( y_{1925}= 101+ 8 [ -0.6 ]+ \frac{(-4)}{2!} [ (-0.6)( 0.4 )]+\frac{(-1)}{3!} [ (-0.6)( 0.4 )(1.4)]+\frac{(-3)}{4!} [ (-0.6)( 0.4 )(1.4)(2.4)]\)

or \( y_{1925}=101-4.8+0.48+0.056+0.1008\)

or \( y_{1925}=96.83\)
This completes.

Example 2

In the table below, values of y are consecutive terms of a series are given. Find tenth term of the series.

3 4 5 6 7 8 9
2.7 6.4 12.5 21.6 34.3 51.2 72.9

Solution
According to given set of data values, the difference table is

X Y 1st diff 2nd diff 3rd diff
3 2.7

3.7
4 6.4 2.4
6.1 0.6
5 12.5 3
9.1 0.6
6 21.6 3.6
12.7 0.6
7 34.3 4.2
16.9 0.6
8 51.2 4.8
21.7
9 72.9

Here,
\( h=1\) and \({x_n}=9\)
Thus, we have
\(x={x_n}+ph\)
or \(10=9+p\)
or \(p=1\)
Now, tenth term of the series is
\(y=y_n+p\nabla y_n+\frac{p( p+1 )}{2!}\nabla ^2y_n+\ldots \)
or \(y=72.9+21.7+4.8\frac2{2!}+0.6\frac{2\times 3}{3!}\)
or \(y=100\)




Example 3

Find cubic polynomial which takes the following values.
\( y( 0 )=1,y( 1 )=0,y( 2 )=1\) and \(y( 3 )=10\)

Solution
According to given set of data values, we form difference table as below

X Y 1st diff 2nd diff 3rd diff
0 1
-1
1 0
2
1
6
2 1 8
9
3 10

Here,
\( h=1\) and \(x_n=3\)
Thus, we have
\( x=x_n+ph\)
or \( p=(x-3)\)
Now, cubic polynomial for given data value is
\(y=y_n+p\nabla y_n+\frac{p( p+1 )}{2!}\nabla ^2y_n+\ldots \)
or \( y=10+(x-3)(9 )+\frac{(x-3)(x-2)}{2!}(8)+\frac{(x-3)(x-2)(x-1)}{3!}(6)\)
or \( y=10+9x-27+4(x^2-5x+6)+(x-1)(x^2-5x+6)\)
or \( y=10+9x-27+(4x^2-20x+24)+(x^3-6x^2+11x-6)\)
or \( y=x^3-2x^2+1\)
This completes.




Exercise

Use Newton's Backward difference formula to compute followings.
  1. Find the value when x=4 from the following table [Answer: 33]
    x0123
    y10110
  2. Find solution of f(11) of an equation \(x^2-x+1\) using a = 2 and b= 10, step value h = 1.
  3. Estimate the number of students who obtained marks between 75 and 80
    Marks30-4040-5050-6060-7070-80
    Number3142513531
  4. Find cubic polynomial which takes the following values: y( 0 )=1,y( 1 )=2, y( 2 )=1 and y( 3 )=10

No comments:

Post a Comment