Partial Differential Equations - Modelling and ... - ResearchGate
Partial Differential Equations - Modelling and ... - ResearchGate
Partial Differential Equations - Modelling and ... - ResearchGate
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
170 B. Maury<br />
b(u ε ,u ε )/ε is bounded, so that b(u ε ,u ε )goesto0withε. Consequently, it<br />
holds 0 ≤ b(z,z) ≤ lim inf b(u ε ,u ε ) = 0, which implies z ∈ K, sothatz = u.<br />
To establish the strong character of the convergence, we show that u ε<br />
converges toward u for the norm associated to a(·, ·), which is equivalent<br />
to the original norm. As u ε converges weakly to u for this scalar product<br />
(a(u ε ,v) → a(u, v) for any v ∈ V ), it is sufficient to establish the convergence<br />
of |u ε | a<br />
= a(u ε ,u ε ) 1/2 towards |u| a<br />
. Firstly |u| a<br />
≤ lim inf |u ε | a<br />
, <strong>and</strong> the other<br />
inequality comes from (3):<br />
1<br />
2 a(uε ,u ε ) −〈ϕ, u ε 〉≤ 1 a(u, u) −〈ϕ, u〉,<br />
2<br />
so that lim sup |u ε | a<br />
≤|u| a<br />
.<br />
The proposition does not say anything about the rate of convergence, <strong>and</strong><br />
it can be very poor, as the following example illustrates.<br />
Example 1. Consider I =]0, 1[, V = H 1 (I), <strong>and</strong> the problem which consists<br />
in minimizing the functional<br />
J(v) = 1 ∫<br />
|u ′ | 2 ,<br />
2 I<br />
over K = {v ∈ V | v(x) = 0 a.e. in O = ]0, 1/2[}. The solution to that<br />
problem is obviously u =max{0, 2(x − 1/2)}. Now let us denote by u ε the<br />
minimum of the penalized functional<br />
J ε = 1 ∫<br />
|u ′ | 2 + 1 ∫<br />
|u| 2 ,<br />
2 I 2ε O<br />
The solution to the penalized problem can be computed exactly:<br />
( )<br />
( ( )<br />
x√ε x√ε<br />
u ε =k ε (x)sh<br />
in ]0, 1/2[ with k ε (x)= sh + 1 ( )) −1 x√ε<br />
2 √ ε ch ,<br />
<strong>and</strong> u ε affine in ]1/2, 1[, continuous at 1/2. This makes it possible to estimate<br />
|u ε − u|, which turns out to behave like ε 1/4 .<br />
Yet, in many situations, convergence can be shown to be of order 1, given<br />
some assumptions are fulfilled. Let us introduce ξ ∈ V ′ as the unique linear<br />
functional such that<br />
a(u, v)+〈ξ,v〉 = 〈ϕ, v〉 ∀v ∈ V. (4)<br />
Before stating the first order convergence result, we show here that the penalty<br />
method provides an approximation of ξ.<br />
Proposition 2. Let ξ ε ∈ V ′ be defined by<br />
v ∈ V ↦−→ 〈ξ ε ,v〉 = 1 ε b(uε ,v).<br />
Then ξ ε converges (strongly) to ξ in V ′ , at least as fast as u ε converges to u.