Kolmogorov equation associated to the stochastic reflection problem ...
Kolmogorov equation associated to the stochastic reflection problem ...
Kolmogorov equation associated to the stochastic reflection problem ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Proof. Let (ϕ n ) ⊂ C 1 (K) be such that<br />
ϕ n → 0 in L 2 (K, ν), Dϕ n → F in L 2 (K, ν; H),<br />
as n → ∞. We have <strong>to</strong> show that F = 0. Let ψ ∈ C0(K) 1 and z ∈ Q 1/2 (H).<br />
Then by (2.10) with ϕ n ψ replacing ϕ (see Theorem 2.2) we have that<br />
∫<br />
∫<br />
〈Dϕ n (x), Q 1/2 z〉 ψ(x) ν(dx) = − 〈Dψ(x), Q 1/2 z〉 ϕ n (x) ν(dx)<br />
K<br />
+ 1<br />
2µ(K)<br />
∫<br />
Σ<br />
∫<br />
ϕ n (y) ψ(y) 〈n(y), Q 1/2 z〉µ Σ (dy) +<br />
∫<br />
= − 〈Dψ(x), Q 1/2 z〉 ϕ n (x) ν(dx) +<br />
K<br />
since ψ vanishes on Σ. Letting n → ∞ we find that<br />
∫<br />
〈F (x), Q 1/2 z〉ψ(x)µ(dx) = 0.<br />
H<br />
K<br />
∫<br />
K<br />
K<br />
W z (x)ϕ n (x)ψ(x)ν(dx)<br />
W z (x)ϕ n (x)ψ(x)ν(dx),<br />
(2.12)<br />
This implies F = 0 in view of <strong>the</strong> arbitrariness of ψ and z (recall Lemma 2.3<br />
and that Q 1/2 (H) is dense in H).<br />
We shall still denote by D <strong>the</strong> closure of D and by W 1,2 (K, ν) its domain<br />
of definition. W 1,2 (K, ν) is a Hilbert space with <strong>the</strong> scalar product<br />
∫<br />
〈ϕ, ψ〉 W 1,2 (K,ν) = [ϕψ + 〈Dϕ, Dψ〉]dν.<br />
2.3 The trace of a function of W 1,2 (K, ν)<br />
K<br />
In order <strong>to</strong> define <strong>the</strong> trace of a function ϕ ∈ W 1,2 (K, ν) we need a technical<br />
lemma.<br />
Lemma 2.5. Assume that ϕ ∈ Cb 1 (H). Then <strong>the</strong> following estimate holds,<br />
∫<br />
∫<br />
|Q 1/2 n(y)| 2 ϕ 2 (y)µ Σ (dy) + |x| 2 ϕ 2 (x)ν(dx)<br />
Σ<br />
≤ 2Tr Q<br />
∫<br />
K<br />
K<br />
∫<br />
ϕ 2 (x)ν(dx) + 4 Tr [Q 2 ]<br />
K<br />
|Dϕ(x)| 2 ν(dx).<br />
(2.13)<br />
11