Thanks to visit codestin.com
Credit goes to www.scribd.com

Scribd
Upload
194 views2 pages

MA-201:Error in Interpolation

This document discusses error bounds in linear and quadratic interpolation. It shows that for linear interpolation between two points, the absolute truncation error is bounded above by h^2M2/8, where h is the step size and M2 is the maximum value of the function. Similarly, for quadratic interpolation between three points, the absolute truncation error is bounded above by h^3M3/6, where M3 is the maximum of the third derivative of the function. The document provides formulas to calculate the maximum allowable step size h to ensure the truncation error is below a given threshold.

Uploaded by

Ankit Sharma
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
194 views2 pages

MA-201:Error in Interpolation

This document discusses error bounds in linear and quadratic interpolation. It shows that for linear interpolation between two points, the absolute truncation error is bounded above by h^2M2/8, where h is the step size and M2 is the maximum value of the function. Similarly, for quadratic interpolation between three points, the absolute truncation error is bounded above by h^3M3/6, where M3 is the maximum of the third derivative of the function. The document provides formulas to calculate the maximum allowable step size h to ensure the truncation error is below a given threshold.

Uploaded by

Ankit Sharma
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

MA-201:Error in Interpolation

Let f (x) be a function with its value given at n + 1 points say, x0 , x1 , . . . , xn such that x0 < x1 < · · · < xn . We
determine a unique polynomial P (x) called as interpolating polynomial of degree n which satisfies the conditions:

P (xi ) = f (xi ), i = 0, 1, 2, . . . , n (1)

Then the truncation error in the interpolation is defined as

E(f ; x) = f (x) − P (x). (2)

Clearly, the polynomial coincides with the function f(x) at node points and E(f;x)=0 at x = xi , for i = 1,2,...n

Error in linear polynomial


Let P(x) be a linear polynomial approximtion to function f(x) for x[x0 , x1 ].
Given two points x0 and x1 , then absolute error will be given as:

|E1 (f ; x)| = |f (x) − P (x)|



(x − x0 )(x − x1 ) 00
= f (ξ) , x0 < ξ < x1
2

Also, maximum error = M22 max |(x − x0 )(x − x1 )| where M2 = max |f (x)| .
x0 <ξ<x1 x0 <ξ<x1
Proof: We notice that if x = x0 ,or x = x1 ,then E1 (f ; x) = 0. If x(x0 , x1 ) then for this x we define a function G(t)
as

(t − x0 )(t − x1 )
G(t) = f (t) − P (t) − [f (x) − P (x)] (3)
(x − x0 )(x − x1 )
We can verify that

G(t)|t=x0 = 0
G(t)|t=x1 = 0
G(t)|t=x = 0,

We recall here Rolle’s theorem:


If g(x) is continuous function on some interval [a,b] and differentiable on (a,b) and if g(a)=0, g(b)=0,
then there exists at least one point ξ inside (a,b) for which g 0 (ξ) = 0.
Hence, by Rolle’s theorem on G(t) for (x0 , x) and (x, x1 ), we get

G0 (ξ1 ) = 0, x0 < ξ < x

G0 (ξ2 ) = 0, x < ξ2 < x1


Now, G0 (t) also satisfies conditions of Rolle’s theorem. Applying again Rolle’s theorem for G0 (t) on the interval
(ξ1 , ξ2 ), we get
G00 (ξ) = 0, x0 < ξ1 < ξ < ξ2 < x1 (4)
Since, P (x) is linear polynomial, which implies
P 00 (t) = 0 (5)
Hence, from equations (4) and (5), we get
2(f (x) − P (x))
G00 (t) = f 00 (t) −
(x − x0 )(x − x1 )

1
Setting G00 (ξ) = 0, we get
(x − x0 )(x − x1 ) 00
E1 (f ; x) = f (ξ)
2
(x−x0 )(x−x1 )
For maximum of error, we will find maximum of 2

Let
(x − x0 )(x − x1 )
H(x) =
2
which implies
(x − x1 ) (x − x0 )
H 0 (x) = +
2 2
For critical point,
x0 + x1
H 0 (x) = 0 ⇒ x =
2
which gives maximum value of H(x) as
x0 +x1
( 2 − x0 )( x0 +x − x1 ) (x1 − x0 )2 h2
1

2
= =
2 4 4

where h is the step size.


Hence, maximum value of truncation error in linear interpolation will be

h2
|E1 (x)| < M2
8
If the acceptable error is  > 0, we must have

|E1 (x)| < 


h2
⇒ M2 < 
8 √
2 2
⇒h<
M2
Above gives the bound in stepsize h in linear interpolation.

Task 1: Similarly prove that for quadratic interpolation for three nodes, the truncation error is given
by:

(x − x0 )(x − x1 )(x − x3 ) 000


E2 (f ; x) = f (ξ)
3!
Therefore,
M3
|E2 (x)| = max |(x − x0 )(x − x1 )(x − x2 )|
6 x0 <ξ<x2
where
M3 = max f 000 (ξ)
x0 <ξ<x2

Further, if the stepsize is h, then the quadratic interpolation with error bound  must satisfy the folllowing condition:
√ !1/3
9 3
h<
M3

Task 2:
1. Determine the step size h that can be used in the tabulation of f (x) = sin(x) in [0, π4 ] of equally spaced points
so that the truncation error of quadratic interpolation is less that 5 × 10−8 .
2. Determine the maximum value of uniform step size h that can be used in the tabulation of f (x) = (2 + x)4 in
[1, 2] of equally spaced points so that the truncation error of quadratic interpolation is less that 10−4 .

You might also like