0% found this document useful (0 votes)
127 views6 pages

Cheb Ident

The document defines and provides properties of Chebyshev polynomials of the first and second kind. It defines the polynomials recursively and provides closed-form expressions. The polynomials are orthogonal over specific domains and the document discusses their roots, special values, relationships between the two kinds of Chebyshev polynomials, and references further resources on the topic.

Uploaded by

Shanti Naidu
Copyright
© Attribution Non-Commercial (BY-NC)
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
0% found this document useful (0 votes)
127 views6 pages

Cheb Ident

The document defines and provides properties of Chebyshev polynomials of the first and second kind. It defines the polynomials recursively and provides closed-form expressions. The polynomials are orthogonal over specific domains and the document discusses their roots, special values, relationships between the two kinds of Chebyshev polynomials, and references further resources on the topic.

Uploaded by

Shanti Naidu
Copyright
© Attribution Non-Commercial (BY-NC)
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

Chebyshev Polynomials

Stefan Hollos and Richard Hollos June 19, 2006


Exstrom Laboratories LLC, 662 Nelson Park Dr, Longmont, Colorado, 80503, USA http://www.exstrom.com

1 Polynomials of the First Kind


There are many ways to dene the Chebyshev polynomials. We will dene the Chebyshev polynomials of the rst kind as solutions to the following recurrence equation.

Tn+1 (x) 2xTn (x) + Tn1 (x) T0 (x) T1 (x)

= 0 = 1 = x

The following table lists the rst 13 polynomials along with their factorization over the integers.
n 0 1 2 3 4 5 6 7 8 9 10 11 12 Tn (x) 1 x 2x2 1 4x3 3x 4 8x 8x2 + 1 16x5 20x3 + 5x 32x6 48x4 + 18x2 1 64x7 112x5 + 56x3 7x 128x8 256x6 + 160x4 32x2 + 1 256x9 576x7 + 432x5 120x3 + 9x 512x10 1280x8 + 1120x6 400x4 + 50x2 1 1024x11 2816x9 + 2816x7 1232x5 + 220x3 11x 2048x12 6144x10 + 6912x8 3584x6 + 840x4 72x2 + 1

Factored 1 x 2x2 1 x 4x2 3 8x4 8x2 + 1 x 16x4 20x2 + 5 2 2x 1 16x4 16x2 + 1 x 64x6 112x4 + 56x2 7 128x8 256x6 + 160x4 32x2 + 1 x 4x2 3 64x6 96x4 + 36x2 3 2 2x 1 256x8 512x6 + 304x4 48x2 + 1 x 1024x10 2816x8 + 2816x6 1232x4 + 220x2 11 8x4 8x2 + 1 256x8 512x6 + 320x4 64x2 + 1

The recurrence equation can be solved in closed form. This gives the following formula for Tn (x)

Tn (x) =

n n z1 + z2 2

z1 z2
where z1 and z2 satisfy the following relationships

= =

x+ x

x2 1 x2 1

z1 z2 = 1 z1 + z2 = 2x z1 z2 = 2 x2 1

These relationships can be useful in deriving properties of the Tn (x). If |x| 1 then we can set x = cos and z1 and z2 become

z1 = cos + i sin = ei z2 = cos i sin = ei


So that for x = cos we have

ein + ein = cos n 2 Using this form for Tn (x) it is very easy to derive the orthogonality relationship for these polynomials. Starting with the following well known integral m=n 0 m=n=0 cos m cos nd = 0 m=n=0 2 Tn (x) = 0 Tm (x)Tn (x) dx = 1 x2
2

and making the substitution x = cos we get


1 1

m=n m=n=0 m=n=0


1

The Tn (x) are therefore orthogonal under the weight function w(x) = (1 x2 ) 2 however note that they are not orthonormal. Now if |x| > 1 then we can set x = cosh and z1 and z2 become

z1 = cosh + sinh = e z2 = cosh sinh = e en + en = cosh n 2 The Tn (x) are not orthogonal outside the interval 1 x 1. We will now look at the roots of Tn (x). First note that all the roots are in the interval 1 x 1 and they can therefore easily be found using the expression Tn (x) = cos n with x = cos . The roots are at Tn (x) = k =
or So that for x = cosh we have

(2k 1) 2n

k = 1, 2, . . . n

xk = cos k = cos

(2k 1) 2n
2

k = 1, 2, . . . n

With the roots in hand, we can write Tn (x) in product form as


n

Tn (x) = 2n1

(x xk )
k=1

Other special values of Tn (x) which are easily derived from the above relations are

Tn (x) = Tn (1) Tn (1) T2n (0) T2n1 (0)

(1)n Tn (x)

= 1 = (1)n = (1)n = 0

It is possible to extend the denition of the Tn (x) to negative index.

Tn (x) =

n n n z1 + z2 z n + z1 = 2 = Tn (x) 2 2

Tn (x) is therefore even with respect to the index. The formula for the product of two polynomials is easy to derive using the trigonometric form Tn (x)Tm (x) = cos n cos m = cos(n + m) + cos(n m) Tn+m (x) + Tnm (x) = 2 2

The polynomials can also be combined iteratively as follows

Tn (Tm (x)) = Tn (cos m) = cos nm = Tnm (x)


This property can be used to explain some of the patterns in the factorization of the polynomials that can be seen in the table above. The Tn (x) can also be dened by the following generating function

1 tx = Tn (x)tn 1 2tx + t2 n=0

2 Polynomials of the Second Kind


We will dene the Chebyshev polynomials of the second kind as solutions to the following recurrence equation.

Un+1 (x) 2xUn (x) + Un1 (x) = 0 U0 (x) = 1 U1 (x) = 2x


The following table lists the rst 13 polynomials along with their factorization over the integers.

n 0 1 2 3 4 5 6 7 8 9

Un (x) 1 2x 4x2 1 8x3 4x 4 16x 12x2 + 1 32x5 32x3 + 6x 64x6 80x4 + 24x2 1 128x7 192x5 + 80x3 8x 256x8 448x6 + 240x4 40x2 + 1 512x9 1024x7 + 672x5 160x3 + 10x 1024x10 2304x8 + 1792x6 560x4 + 60x2 1 2048x11 5120x9 + 4608x7 1792x5 + 280x3 12x 4096x12 11264x10 + 11520x8 5376x6 + 1120x4 84x2 + 1

10 11 12

Factored 1 2x (2x 1) (2x + 1) 4x 2x2 1 2 4x 2x 1 4x2 + 2x 1 2x (2x 1) (2x + 1) 4x2 3 3 8x 4x2 4x + 1 8x3 + 4x2 4x 1 8x 2x2 1 8x4 8x2 + 1 (2x 1) (2x + 1) 8x3 6x 1 8x3 6x + 1 2x 4x2 2x 1 4x2 + 2x 1 16x4 20x2 + 5 32x5 16x4 32x3 + 12x2 + 6x 1 32x5 + 16x4 32x3 12x2 + 6x + 1 4x (2x 1) (2x + 1) 2x2 1 4x2 3 16x4 16x2 + 1 64x6 32x5 80x4 + 32x3 + 24x2 6x 1 64x6 + 32x5 80x4 32x3 + 24x2 + 6x 1

The recurrence equation can be solved in closed form. This gives the following formula for Un (x)

Un (x) z1 z2

= = =

n+1 n+1 z1 z2 z1 z2

x+ x

x2 1 x2 1

The z 's are the same as those dened for Tn (x). If |x| 1 then we can set x = cos so that z1 = ei , z2 = ei and then Un (x) becomes ei(n+1) ei(n+1) sin(n + 1) Un (x) = = i ei e sin From this trigonometric form, it is easy to derive the orthogonality relationship. First start with the integral

sin(n + 1) sin(m + 1) d =
0

n,m 2

then make the substitution x = cos to get


1

1 x2 Un (x) Um (x) dx =
1
1

n,m 2

The Un (x) are orthogonal under the weight function w(x) = (1 x2 ) 2 , note however that they are not orthonormal. For |x| > 1 we can set x = cosh to get z1 = e , z2 = e and

Un (x) =

e(n+1) e(n+1) sinh(n + 1) = e e sinh

The Un (x) are not orthogonal outside the interval 1 x 1. 4

We now look at the roots of Un (x) which are all in the interval 1 x 1 so we can nd them using the trigonometric form of Un (x). The roots are at

k =
or

k n+1

k = 1, 2, . . . n k n+1
n

xk = cos k = cos
We can then write Un (x) in product form as

k = 1, 2, . . . n

Un (x) = 2n
k=1

(x xk )

Other special values of Un (x) are

Un (x) Un (1) Un (1) U2n (0) U2n1 (0)

= (1)n Un (x) = n+1 = (1)n (n + 1) = (1)n = 0

It is possible to extend the denition of the Un (x) to negative index.

Un (x) =

n+1 n+1 n1 z1 z2 z n1 z1 = 2 = Un2 (x) z1 z2 z1 z2

The formula for the product of two polynomials is

Un (x)Um (x) =

Tnm (x) Tn+m+2 (x) 2(1 x2 )

The Un (x) can also be dened by the following generating function

1 = Un (x)tn 1 2tx + t2 n=0

3 Relationships Between Tn (x) and Un (x)


It is easy to derive hundreds of relationships between the Tn (x) and Un (x) by using their trigonometric forms or the formulas in terms of the z 's. Some of the relationships are listed below.

d Tn (x) = nUn1 (x) dx (1 x2 )Un1 (x) = xTn (x) Tn+1 (x) Tn (x) = Un (x) xUn1 (x) Unm1 (x) Un1 (Tm (x)) = Um1 (x)
5

References
[1] George Arfken. Mathematical Methods for Physicists. Academic Press, third edition, 1985. [2] J.C. Mason and D.C. Handscomb. Chebyshev Polynomials. Chapman and Hall/CRC, 2003. [3] Theodore J. Rivlin. Chebyshev Polynomials : From Approximation Theory to Algebra and Number Theory. John Wiley & Sons, 1990. [1, 2, 3]

You might also like