11.08.2013 Views

Untitled - Cdm.unimo.it

Untitled - Cdm.unimo.it

Untitled - Cdm.unimo.it

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

230 Polynomial Approximation of Differential Equations<br />

where ζ > 0, U0 : [−1,1] → R, σ : [0,T] → R, are given. We assume that U0 and<br />

σ are continuous and σ(0) = U0(−1).<br />

All the standard spectral schemes can be considered to approximate U. Neverthe-<br />

less, the theoretical analysis of convergence is more difficult than the analysis of the<br />

problem in section 10.2. We discuss some examples.<br />

Take σ ≡ 0 . In this case, we can explic<strong>it</strong>ly compute the solution, given by U(x,t) =<br />

U0(x − ζt) if − 1 < x − ζt ≤ 1, U(x,t) = 0 if x − ζt ≤ −1. Consider the collocation<br />

method. We must find pn(·,t) ∈ Pn, n ≥ 1, t ∈]0,T], such that<br />

(10.3.4)<br />

∂pn<br />

∂t (η(n)<br />

i ,t) = −ζ ∂pn<br />

∂x (η(n)<br />

i ,t), 1 ≤ i ≤ n, ∀t ∈]0,T],<br />

(10.3.5) pn(η (n)<br />

i ,0) = U0(η (n)<br />

i ), 0 ≤ i ≤ n,<br />

(10.3.6) pn(η (n)<br />

0 ,t) = 0, ∀t ∈]0,T].<br />

Again, these equations can be reduced to a n ×n linear system of ordinary differential<br />

equations. For instance, using the notations of section 7.2, in the case n = 3, we have<br />

(10.3.7)<br />

d<br />

dt<br />

⎡<br />

pn(η<br />

⎢<br />

⎣<br />

(n)<br />

1 ,t)<br />

pn(η (n)<br />

⎤<br />

⎥<br />

2 ,t) ⎥<br />

⎦<br />

pn(η (n)<br />

3 ,t)<br />

⎡<br />

⎢<br />

= −ζ ⎢<br />

⎣<br />

˜d (1)<br />

11<br />

˜d (1)<br />

21<br />

˜d (1)<br />

31<br />

˜d (1)<br />

12<br />

˜d (1)<br />

22<br />

˜d (1)<br />

32<br />

˜d (1)<br />

13<br />

˜d (1)<br />

23<br />

˜d (1)<br />

33<br />

⎤⎡<br />

pn(η<br />

⎥⎢<br />

⎥⎢<br />

⎥⎢<br />

⎥⎢<br />

⎦⎣<br />

(n)<br />

1 ,t)<br />

pn(η (n)<br />

⎤<br />

⎥<br />

2 ,t) ⎥<br />

⎥,<br />

t ∈]0,T].<br />

⎦<br />

pn(η (n)<br />

3 ,t)<br />

The eigenvalues of the matrix in (10.3.7) (see also (7.4.3)) have been studied in section<br />

8.1. We will return to this point in section 10.6.<br />

In the Legendre case (w ≡ 1), we can provide a simple analysis of stabil<strong>it</strong>y. In fact,<br />

formula (3.5.1) allows us to wr<strong>it</strong>e for any t ∈]0,T]<br />

(10.3.8)<br />

d<br />

dt pn(·,t) 2 w,n = 2<br />

n<br />

j=0<br />

<br />

∂pn<br />

∂t pn<br />

<br />

(η (n)<br />

j ,t) w (n)<br />

j<br />

=

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

Saved successfully!

Ooh no, something went wrong!