01.05.2013 Views

Modélisation, analyse mathématique et simulations numériques de ...

Modélisation, analyse mathématique et simulations numériques de ...

Modélisation, analyse mathématique et simulations numériques de ...

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.

tel-00656013, version 1 - 3 Jan 2012<br />

5.3 Time Fourier analysis and boundary layer approximations 149<br />

in the second case one should lift the Dirichl<strong>et</strong> error [40]. Here, we chose to extend û0,k by<br />

a linear function in Ωε \Ω0:<br />

<br />

û0,k in Ω0,<br />

û0,k :=<br />

, where Mk :=<br />

Mkx2 in Ωε \Ω0,<br />

∂<br />

û0,k(x1,0) =<br />

∂x2<br />

Ĉkr(2−e r −e−r )<br />

ik(e−r −er .<br />

)<br />

5.3.2 Zero or<strong>de</strong>r error estimates<br />

We d<strong>et</strong>ail here the error estimates. I<strong>de</strong>ntical proofs should also be used for higher or<strong>de</strong>r<br />

approximations below: we d<strong>et</strong>ail here every step. Denote χΩ the characteristic function of<br />

the domain Ω, δΓ0 the Dirac measure concentrated on Γ0 .<br />

Proposition 1. There exist two positive constants c1 and c2, <strong>de</strong>pending only of the mo<strong>de</strong><br />

Ĉk and the Sobolev’s inequalities, such that:<br />

√<br />

ε , ûǫ,k −û0,kL 2 (Ω0) ≤ c2ε. (5.3.4)<br />

ûǫ,k −û0,k H 1 (Ωε) ≤ c1<br />

Proof. The first part of the proof is based on standard a priori estimates. The existence<br />

and uniqueness of ûǫ,k are well known and <strong>de</strong>rive from the Lax-Milgram theorem. We focus<br />

on the error, namely we s<strong>et</strong> R ε 0 := ûǫ,k−û0,k. Since the extension û0,k of û0,k in the rough<br />

domain satisfies: ⎧⎨<br />

⎩<br />

Lk û0,k = ĈkχΩ0 +ikMkx2χ Ωε\Ω0 in Ωε,<br />

û0,k = 0 on Γ1,<br />

û0,k = Mkx2<br />

on Γε.<br />

Then the zeroth or<strong>de</strong>r error solves:<br />

⎧<br />

⎨ Lk R<br />

⎩<br />

ε 0 = ĈkχΩε\Ω0 −ikMkx2χ Ωε\Ω0 in Ωε,<br />

Rε 0 = 0 on Γ1,<br />

= −Mkx2<br />

on Γε.<br />

R ε 0<br />

(5.3.5)<br />

(5.3.6)<br />

We remark that a part of the error comes from the source term localized in Ωε \Ω0, and<br />

another part comes from the non homogeneous boundary term on Γε. We s<strong>et</strong> the lift:<br />

Then: ⎧ ⎨<br />

⎩<br />

s = −Mkx2χ Ωε\Ω0 , ˜ R ε 0 = R ε 0 −s.<br />

Lk ˜ Rε 0 = ĈkχΩε\Ω0 +MkδΓ0 in Ωε,<br />

˜R ε 0 = 0 on Γ1,<br />

˜R ε 0 = 0 on Γε,<br />

(5.3.7)<br />

where the <strong>de</strong>rivatives are computed in the sense of distributions. Then, on the one hand,<br />

using Poincaré inequality, we have:<br />

<br />

<br />

<br />

Lk ˜ R ε 0 ˜ Rε 0dx <br />

2<br />

<br />

<br />

= ik ˜ R ε 0 2 L2 (Ωε) +∇˜ R ε 0 2 L2 <br />

<br />

(Ωε) 2<br />

Ωε<br />

= k 2 ˜ R ε 0 4 L 2 (Ωε) +∇˜ R ε 0 4 L 2 (Ωε)<br />

≥ c ˜ R ε 0 4 H 1 (Ωε) .<br />

(5.3.8)

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

Saved successfully!

Ooh no, something went wrong!