11.08.2013 Views

Untitled - Cdm.unimo.it

Untitled - Cdm.unimo.it

Untitled - Cdm.unimo.it

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

36 Polynomial Approximation of Differential Equations<br />

one considers the polynomial (x − ξ1)(x − ξ2)un and, if n > 2, deduces the existence<br />

of a third zero, and so on. The procedure ends when n zeroes are finally obtained.<br />

For any n ≥ 1 we denote the n zeroes of un by ξ (n)<br />

k , 1 ≤ k ≤ n. We assume that these<br />

are in increasing order (although many authors prefer the reverse order). It is obvious<br />

that the polynomial u ′ n has n − 1 real distinct zeroes in I. We denote these zeroes by<br />

η (n)<br />

k , 1 ≤ k ≤ n − 1. One can prove that un and un−1 do not have common zeroes.<br />

Moreover, between any two consecutive zeroes of un−1, there exists one and only one<br />

zero of un.<br />

Many general theorems characterize the zeroes of orthogonal polynomials. It suf-<br />

fices to recall the following statement.<br />

Theorem 3.1.2 - Let {un} n∈N be a sequence of orthogonal polynomials in I. Then,<br />

for any interval [a,b] ⊂ I, a < b, <strong>it</strong> is possible to find m ∈ N such that um has at least<br />

one zero in [a,b].<br />

In other words, this theorem states that the set J = <br />

n≥1<br />

n k=1 {ξ(n)<br />

k<br />

} is dense in Ī.<br />

We are going to review some properties of the zeroes of classical orthogonal poly-<br />

nomials. We refer for instance to szegö(1939) for proofs and further results.<br />

Zeroes in the Jacobi case - We restrict ourselves to the case where − 1<br />

2<br />

− 1<br />

2<br />

1 ≤ β ≤ 2 . Under these cond<strong>it</strong>ions, for n ≥ 1 we have the estimates<br />

(3.1.1) −1 ≤ −cos<br />

≤ α ≤ 1<br />

2 and<br />

k + (α + β − 1)/2<br />

k<br />

π ≤ ξ(n)<br />

k ≤ −cos<br />

π ≤ 1,<br />

n + (α + β + 1)/2 n + (α + β + 1)/2<br />

1 ≤ k ≤ n.<br />

This shows that asymptotically the distance between two consecutive zeroes is propor-<br />

tional to 1/n for points located in the central part of the interval I =] − 1,1[, and<br />

proportional to 1/n 2 for points located near the endpoints of the same interval.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!