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.

54 Polynomial Approximation of Differential Equations<br />

integral of polynomials of degree 2n−1, knowing their values only at n+1 points. W<strong>it</strong>h<br />

respect to Gauss formulas there is a slight loss in accuracy. Nevertheless, the fact that<br />

the points x = −1 and x = 1 are included in the nodes is important when imposing<br />

boundary cond<strong>it</strong>ions in the approximation of differential equations (see section 7.4).<br />

3.6 Gauss-Radau integration formulas<br />

Other integration formulas, known as Gauss-Radau formulas, are based on the nodes<br />

given in (3.1.12), (3.1.13) and (3.1.14) respectively. The first case actually corresponds<br />

to a Gauss type integration formula since, thanks to (1.3.6), the nodes given in (3.1.12)<br />

are the zeroes of P (3/2,3/2)<br />

n−1 . Therefore, the relative formula is exact for polynomials up<br />

to degree 2n − 3. The other two sets of n nodes include the points x = 1 or x = −1<br />

respectively. Appropriate weights can be defined so that the associated integration<br />

formulas are valid for polynomials of degree up to 2n − 2. Some details are discussed<br />

for instance in ralston(1965) or davis and rabinow<strong>it</strong>z(1984).<br />

In this section we are concerned w<strong>it</strong>h an add<strong>it</strong>ional formula, based on the zeroes of<br />

d<br />

dx L(α)<br />

n plus the point x = 0 (the only boundary point of the interval [0,+∞[). This is<br />

(3.6.1)<br />

+∞<br />

0<br />

pw dx =<br />

n−1 <br />

j=0<br />

p(η (n)<br />

j ) ˜w (n)<br />

j .<br />

˜(n) l j wdx, where the Lagrange<br />

Here w is the Laguerre weight function and ˜w (n)<br />

j := +∞<br />

0<br />

polynomials are defined in (3.2.11). W<strong>it</strong>h a proof similar to that of theorem 3.5.1,<br />

equation (3.6.1) turns out to be true for any polynomial p of degree at most 2n −2. For<br />

the weights, we obtain<br />

(3.6.2) ˜w (n)<br />

j =<br />

⎧<br />

(α + 1) Γ<br />

⎪⎨<br />

⎪⎩<br />

2 (α + 1) (n − 1)!<br />

Γ(n + α + 1)<br />

Γ(n + α)<br />

n!<br />

<br />

L (α)<br />

n (η (n)<br />

j ) d<br />

dx L(α)<br />

n−1 (η(n)<br />

j )<br />

−1<br />

if j = 0,<br />

if 1 ≤ j ≤ n − 1.<br />

These formulas are proven w<strong>it</strong>h an argument similar to that used in theorem 3.5.2.

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

Saved successfully!

Ooh no, something went wrong!