Search

The Online Encyclopedia and Dictionary

 
     
 

Encyclopedia

Dictionary

Quotes

 

Orthogonal polynomials

In mathematics, two polynomials f and g are orthogonal to each other with respect to a nonnegative "weight function" w precisely if

\int_{x_1}^{x_2} f(x)g(x)w(x)\,dx=0.

In other words, if polynomials are treated as vectors and the inner product of two polynomials f(x) and g(x) is defined as

\langle f,g \rangle=\int_{x_1}^{x_2} f(x)g(x)\,w(x)\,dx

then the orthogonal polynomials are simply orthogonal vectors in this inner product space.

A polynomial sequence pn(x) for n = 0, 1, 2, ... , where pn(x) has degree n, is said to be a sequence of orthogonal polynomials with respect to a "weight function" w when any two of them are orthogonal with respect to that weight function, i.e.,

\langle p_n, p_m \rangle=\int_{x_1}^{x_2} p_n(x) p_m(x)\,w(x)\,dx=0\ \mbox{whenever}\ n\neq m.

The sequence of orthogonal polynomials can be successively constructed by carrying out the Gram-Schmidt process with the sequence of powers x^k, \; k \ge 0, where the positive-definite inner product \langle p,q \rangle on the space of polynomials is given by the integral above.

Contents

General properties of orthogonal polynomials

  • Three-term recurrence relations

For each non-negative weight function the corresponding orthogonal polynomials obey a recurrence relation

fn + 1 = (an + xbn)fn - cnfn - 1

where the constants an, bn and cn are given by

b_n=\frac{k_{n+1}}{k_n}
a_n=b_n \left(\frac{k_{n+1}'}{k_{n+1}} - \frac{k_n'}{k_n} \right)
c_n=\frac{k_{n+1}k_{n-1}h_n} {k_n^2 h_{n-1}}

and kn and kn' are the leading terms in the expansion of the polynomial:

fn(x) = knxn + kn'xn - 1 + ...

and hn is the normalization, defined below.


The classical orthogonal polynomials

The collective name "classical orthogonal polynomials" refers to a class of orthogonal polynomials which are distinguished by several characteristic properties. They occur in many applications including mathematical physics, interpolation theory , the theory of random matrices and many others and have been therefore studied in mathematics since a long time. Some of their characteristic properties will be outlined in the following subsections:

Differential equation

The classical orthogonal polynomials satisfy a second-order differential equation

g2(x)fn''(x) + g1(x)fn'(x) + dnfn(x) = 0

where g1(x) and g2(x) are independent of n and dn is a constant that depends only on n. The coefficient function g2(x) is a polynomial of degree \le 2, the coefficient g1(x) is a polynomial of degree \le 1.

Existence of a Rodrigues formula

Every classical orthogonal polynomial can be obtained via a so-called Rodrigues formula:

f_n(x)=\frac{1}{e_n w(x)}\, \frac{d^n}{dx^n} w(x)[g(x)]^n

where w(x) is the defining weight function of the series of orthogonal polynomials (defined in the list below) and en is a constant depending only on n, and g(x) is a polynomial independent of n.


The orthogonality relationship is

\int_{x_1}^{x_2}p_n(x)p_m(x)w(x)\,dx=\delta_{mn}h_n

where δmn is the Kronecker delta and hn is defined in the table below for each weight function.


Table Of Classical Orthogonal Polynomials
Namex1x2w(x)hn
Chebyshev polynomials (first kind) - 1 1 (1 - x2) - 1 / 2 \left\{ \begin{matrix} \pi   &:~n=0 \\ \pi/2 &:~n\ne 0 \end{matrix}\right.
Chebyshev polynomials (second kind) - 1 1 (1 - x2)1 / 2 π / 2
Legendre polynomials - 1 1 1 \frac{2}{2n+1}
Laguerre polynomials 0 \infty e - x 1
Hermite polynomials -\infty \infty e^{-x^2} n!\,\sqrt{2\pi}

See also

References

The contents of this article are licensed from Wikipedia.org under the GNU Free Documentation License. How to see transparent copy