Approximation Theory

Let f (x) be a continuous function on an interval [a, b]. If P(x) is a

polynomial, we are interested in nding
E[P] = max
axb
|f (x) P(x)|
the maximum possible error in the approximation of f (x) by P(x) on [a, b].
For each degree n dene

n
(f ) = min
deg(P)n
E[P] = min
deg(P)n

max
axb
|f (x) P(x)|

The minimax error,

n
(f ), is the smallest value for E[P] that can be
obtained with a polynomial of degree n.
Dana Mackey (DIT) Maths 1 / 10
Minimax polynomial
It can be shown that the minimax error
n
(f ) on [a, b] is achieved for a
unique polynomial of degree n called the minimax polynomial
approximation of order n, denoted by M
n
(x).
Example: Let f (x) = e
x
on [1, 1] and consider linear polynomial
approximations to f . The Taylor polynomial for this function is
T
1
(x) = 1+x
and the maximum possible error
E[T
1
] = max
1x1
|e
x
T
1
(x)| = 0.718
On the other hand, it can be shown that the linear minimax polynomial is
M
1
(x) = 1.2643+1.1752x
for which the maximum possible error is
E[M
1
] = max
1x1
|e
x
M
1
(x)| = 0.279 < E[T
1
]
Dana Mackey (DIT) Maths 2 / 10
Chebyshev polynomials
For any integer n 0 dene the function
T
n
(x) = cos(ncos
1
(x)), 1 x 1
We need to show that T
n
(x) is a polynomial of degree n. We calculate the
functions T
n
(x) recursively.
Let = cos
1
(x) so cos() = x. Then
T
n
(x) = cos(n)
Easy to see that:
n = 0 =T
0
(x) = cos(0) = 1
n = 1 =T
1
(x) = cos() = x
n = 2 =T
2
(x) = cos(2) = 2cos
2
() 1 = 2x
2
1
Dana Mackey (DIT) Maths 3 / 10
Recurrence relations for Chebyshev polynomials
Using trigonometric formulas we can prove that
T
n+m
(x) +T
nm
(x) = 2T
n
(x)T
m
(x)
for all n m 0 and all x [1, 1].
Hence, for m = 1 we get
T
n+1
(x) +T
n1
(x) = 2xT
n
(x)
which is then used to calculate the Chebyshev polynomials of higher order.
Example: Calculate T
3
(x), T
4
(x) and T
5
(x).
Dana Mackey (DIT) Maths 4 / 10
More properties of Chebyshev polynomials
Note that
|T
n
(x)| 1
and
T
n
(x) = 2
n1
x
n
+lower degree terms
for all n 0 and all x in [1, 1].
If we dene the modied Chebyshev polynomial:

T
n
(x) =
T
n
(x)
2
n1
then we have
|

T
n
(x)|
1
2
n1
and

T
n
(x) = x
n
+lower degree terms
for all n 0 and all x in [1, 1].
Dana Mackey (DIT) Maths 5 / 10
Zeros of Chebyshev polynomials
We have
T
n
(x) = cos(n), = cos
1
(x)
so
T
n
(x) = 0 = n =

2
,
3
2
,
5
2
, . . .
which implies
=
(2k +1)
2n
, k = 0, 1, 2, . . .
and hence the zeros of T
n
(x) are given by
x
k
= cos

(2k +1)
2n

, k = 0, 1, 2, . . . n 1.
Dana Mackey (DIT) Maths 6 / 10
The minimum size property
Let n 1 be an integer and consider all possible monic polynomials (that
is, polynomials whose highest-degree term has coecient equal to 1) of
degree n.
Then the degree n monic polynomial with the smallest maximum absolute
value on [1, 1] is the modied Chebyshev polynomial

T
n
(x) and its
maximum value is 1/2
n1
.
Dana Mackey (DIT) Maths 7 / 10
A near-minimax approximation method
Let f (x) be a continuous function on [1, 1]. We are looking for an
approximation given by an interpolating polynomial of degree 3, C
3
(x).
Let x
0
, x
1
, x
2
, x
3
be the interpolating nodes.
Recall the formula for the interpolation error:
f (x) C
3
(x) =
(x x
0
)(x x
1
)(x x
2
)(x x
3
)
4!
f
(4)
()
where is in [1, 1].
We need to nd the interpolating points so that we minimize
E[C
3
] = max
1x1
|f (x) C
3
(x)|
Dana Mackey (DIT) Maths 8 / 10
This is equivalent to minimizing
max
1x1
|(x x
0
)(x x
1
)(x x
2
)(x x
3
)|
But we know that the minimum value of a monic polynomial is obtained
for the modied Chebyshev polynomial

T
4
(x) hence
(x x
0
)(x x
1
)(x x
2
)(x x
3
) =
T
4
(x)
2
3
=
1
8
(8x
4
8x
2
+1)
hence the interpolating points x
0
, x
1
, x
2
, x
3
are the zeros of T
4
(x), that is
cos(

8
), cos(
3
8
), cos(
5
8
), cos(
7
8
)
Dana Mackey (DIT) Maths 9 / 10
Example
Let f (x) = e
x
on [1, 1]. In order to get the interpolating polynomial of
degree 3 which approximates f (x) such that the maximum error is
minimized, the interpolation nodes x
0
, x
1
, x
2
, x
3
have to be chosen as the
zeros of T
4
(x).
We use Newtons divided dierence formula for the interpolating
polynomial
P
3
(x) = f (x
0
) +(x x
0
)f [x
0
, x
1
] +(x x
0
)(x x
1
)f [x
0
, x
1
, x
2
]
+(x x
0
)(x x
1
)(x x
2
)f [x
0
, x
1
, x
2
, x
3
]
where
f (x
0
) = e
cos(

8
)
2.5190
f [x
0
, x
1
] =
f (x
1
) f (x
0
)
x
1
x
0
1.94538
.
.
.
Dana Mackey (DIT) Maths 10 / 10
Recall that the degree n monic polynomial with the smallest maximum
absolute value on [1, 1] is the modied Chebyshev polynomial

T
n
(x) and
its maximum value is 1/2
n1
.
Hence, the Chebyshev polynomials can be used to minimize approximation
error by providing optimal interpolation points.
The Chebyshev polynomials also provide a method for reducing the degree
of an approximating polynomial with minimal loss of accuracy.
Dana Mackey (DIT) Maths 11 / 10
Economization of power series
Consider approximating a polynomial of degree n
P
n
(x) = a
n
x
n
+a
n1
x
n1
+ +a
1
x +a
0
on [1, 1] with a polynomial of degree at most n 1. We need to choose
P
n1
(x) which minimizes
max
x[1,1]
|P
n
(x) P
n1
(x)|
We know that
max
x[1,1]
|

T
n
(x)| =
1
2
n1
max
x[1,1]
|
1
a
n
(P
n
(x) P
n1
(x))|
Dana Mackey (DIT) Maths 12 / 10
Hence we have
1
a
n
(P
n
(x) P
n1
(x)) =

T
n
(x)
so
P
n1
(x) = P
n
(x) a
n

T
n
(x)
and this choice gives
max
x[1,1]
|P
n
(x) P
n1
(x)| =
|a
n
|
2
n1
Dana Mackey (DIT) Maths 13 / 10
Examples
1
Starting with the fourth-order MacLaurin polynomial, nd the
polynomial of least degree which best approximates the function
f (x) = e
x
on [1, 1] while keeping the error less than 0.05.
2
Find the sixth order MacLaurin polynomial for sin(x) and use
Chebyshev polynomials to obtain a lesser degree polynomial
approximation while keeping the error less than 0.01 on [1, 1].
3
Use the zeros of

T
3
to construct an interpolating polynomial of
degree 2 for the functions (i) sin(x) and (ii) ln(x +2), on [-1,1].
Dana Mackey (DIT) Maths 14 / 10