Scribd
Upload
26 views6 pages

S4 4

This document discusses approximating functions with polynomials and finding the best approximation. The best approximation is called the minimax approximation and minimizes the maximum error between the function and polynomial over an interval. For the function f(x)=e^x on [-1,1], tables show the minimax approximation has smaller error than the Taylor polynomial for a given degree. Graphs also illustrate how the minimax approximation fits the function more closely than the linear Taylor polynomial. While minimax approximations are most accurate, they are difficult to construct, so near-minimax approximations that are easier to construct but still near-optimal will be discussed.

Uploaded by

Florin
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
26 views6 pages

S4 4

This document discusses approximating functions with polynomials and finding the best approximation. The best approximation is called the minimax approximation and minimizes the maximum error between the function and polynomial over an interval. For the function f(x)=e^x on [-1,1], tables show the minimax approximation has smaller error than the Taylor polynomial for a given degree. Graphs also illustrate how the minimax approximation fits the function more closely than the linear Taylor polynomial. While minimax approximations are most accurate, they are difficult to construct, so near-minimax approximations that are easier to construct but still near-optimal will be discussed.

Uploaded by

Florin
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/ 6

BEST APPROXIMATION

Given a function f (x) that is continuous on a given interval [a, b],


consider approximating it by some polynomial p(x). To measure
the error in p(x) as an approximation, introduce
E (p) = max |f (x) − p(x)|
a≤x≤b

This is called the maximum error or uniform error of


approximation of f (x) by p(x) on [a, b].
We want to find the ‘best’ possible approximation for a given
degree n. With this in mind, introduce
 
ρn (f ) = min E (p) = min max |f (x) − p(x)|
deg(p)≤n deg(p)≤n a≤x≤b

The number ρn (f ) will be the smallest possible uniform error, or


minimax error, when approximating f (x) by polynomials of degree
at most n. If there is a polynomial giving this smallest error, we
denote it by mn (x); thus E (mn ) = ρn (f ).
Example
Let f (x) = e x on [−1, 1]. In the following table, we give the values
of E (tn ), tn (x) the Taylor polynomial of degree n for e x about
x = 0, and E (mn ).

Maximum Error in:


n tn (x) mn (x)
1 7.18E − 1 2.79E − 1
2 2.18E − 1 4.50E − 2
3 5.16E − 2 5.53E − 3
4 9.95E − 3 5.47E − 4
5 1.62E − 3 4.52E − 5
6 2.26E − 4 3.21E − 6
7 2.79E − 5 2.00E − 7
8 3.06E − 6 1.11E − 8
9 3.01E − 7 5.52E − 10
Consider graphically how we can improve on the Taylor polynomial
t1 (x) = 1 + x
as a uniform approximation to e x on the interval [−1, 1].
The linear minimax approximation is
m1 (x) = 1.2643 + 1.1752x

y
x
y=e

y=t (x)
1

y=m (x) 1
1

x
−1 1

Figure: Linear Taylor and minimax approximations to e x


y

0.0516

x
−1 1

Figure: Error in cubic Taylor approximation to e x

y
0.00553

x
−1 1

−0.00553

Figure: Error in cubic minimax approximation to e x


Accuracy of the minimax approximation

[(b − a)/2]n+1
ρn (f ) ≤ max f (n+1) (x)
(n + 1)!2n a≤x≤b
This error bound does not always become smaller with increasing
n, but it will give a fairly accurate bound for many functions f (x).
Example. Let f (x) = e x for −1 ≤ x ≤ 1. Then
e
ρn (e x ) ≤ (*)
(n + 1)!2n

n Bound (*) ρn (f )
1 6.80E − 1 2.79E − 1
2 1.13E − 1 4.50E − 2
3 1.42E − 2 5.53E − 3
4 1.42E − 3 5.47E − 4
5 1.18E − 4 4.52E − 5
6 8.43E − 6 3.21E − 6
7 5.27E − 7 2.00E − 7
Remark

Minimax approximation is more accurate than the Taylor


polynomial or an interpolating polynomial of a same degree.
However, it is difficult to construct the minimax approximation.
Therefore, we turn to a discussion of near-minimax approximation,
which provides a near optimal approximation and at the same
time, is easy to construct.
As a preparation for the near-minimax approximation, we will
introduce Chebyshev polynomials next.

You might also like