02.12.2012 Views

Quasilinear parabolic problems with nonlinear boundary conditions

Quasilinear parabolic problems with nonlinear boundary conditions

Quasilinear parabolic problems with nonlinear boundary conditions

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

Here the operator D is pseudo-sectorial in X, and we assume that D A 1/2 ↩→ DD. As<br />

above we seek a solution in<br />

Z = H α p (J; Lp(R+; X)) ∩ Lp(J; H 2 p(R+; X)) ∩ Lp(J; Lp(R+; DA)).<br />

Before we state the theorem concerning (3.45) we note that any function v in the space<br />

Z automatically satisfies v|y=0 ∈ Lp(J; DD) and<br />

Dv(·, 0) ∈ B<br />

1 1<br />

α( − 2 2p )<br />

pp<br />

(J; X) ∩ Lp(J; DA( 1<br />

In fact, if v ∈ Z, then it follows, by Theorem 3.5.1, that<br />

v|y=0 ∈ Lp(J; DA(1 − 1<br />

2p , p)),<br />

which entails the first claim, for we have the embeddings<br />

DA(1 − 1<br />

2p , p) ↩→ D A 1 2<br />

2<br />

↩→ DD.<br />

1 − 2p , p)). (3.46)<br />

As for (3.46), we see <strong>with</strong> the aid of the mixed derivative theorem (Proposition 2.3.2)<br />

that for w := A 1/2 v,<br />

w ∈ H α<br />

2<br />

p (J; Lp(R+; X)) ∩ Lp(J; H 1 p(R+; X)) ∩ Lp(J; Lp(R+; D A 1 2 )). (3.47)<br />

We extend w w.r.t. t to all of R such that (3.47) holds true, <strong>with</strong> J replaced by R.<br />

As above, set Y = Lp(R; X) and define the operator F by (3.32). Then we know that<br />

F is invertible and lies in the class BIP(Y ) <strong>with</strong> power angle θF < π/2. Further,<br />

w ∈ H 1 p(R+; Y ) ∩ Lp(R+; DF ). Thus, Theorem 3.4.1 yields w|y=0 ∈ DF (1 − 1/p, p),<br />

which by restriction to t ∈ J implies<br />

A 1<br />

2 v(·, 0) ∈ B<br />

1 1<br />

α( − 2 2p )<br />

pp (J; X) ∩ Lp(J; DA( 1 1<br />

2 − 2p , p)),<br />

in view of (3.33). Assertion (3.46) now follows on account of D A 1/2 ↩→ DD.<br />

Theorem 3.5.2 Let p ∈ (1, ∞), X be a Banach space of class HT , a ∈ K 1 (α, θa) <strong>with</strong><br />

α ∈ (0, 2) \<br />

� 1<br />

p , 2<br />

p−1<br />

, 1 + 1<br />

p<br />

�<br />

, and A ∈ BIP(X) <strong>with</strong> power angle θA. Let further D be a<br />

pseudo-sectorial operator in X such that D ∈ BIP(R(D)) <strong>with</strong> power angle θD. Suppose<br />

that A and D commute in the resolvent sense, and that D A 1/2 ↩→ DD. Assume further<br />

the angle <strong>conditions</strong> θa + θA < π and θD < π − max{θa, θA}/2. Then the problem (3.45)<br />

has a unique solution in Z if and only if the data f and φ are subject to the following<br />

<strong>conditions</strong>.<br />

(i) f ∈ H α p (J; Lp(R+; X));<br />

(ii) φ ∈ B<br />

1 1<br />

α( − 2 2p )<br />

pp (J; X) ∩ Lp(J; DA( 1 1<br />

2 − 2p , p));<br />

2<br />

2− pα<br />

pp<br />

(iii) f|t=0 ∈ B (R+; X) ∩ Lp(R+; DA(1 − 1<br />

1<br />

pα , p)), if α > p ;<br />

(iv) f|t=0, y=0 ∈ D(D) and − ∂yf|t=0, y=0 + Df|t=0, y=0 = φ|t=0, if α > 2<br />

p−1 ;<br />

(v) ∂tf|t=0 ∈ B<br />

1 1<br />

2(1− − α pα )<br />

pp<br />

If this is the case, then additionally<br />

(R+; X) ∩ Lp(R+; DA(1 − 1<br />

α<br />

1<br />

1<br />

− pα , p)), if α > 1 + p .<br />

(vi) f|t=0, y=0 ∈ DA(1 − 1 1<br />

2p − pα , p), if α > 2<br />

2p−1 ;<br />

(vii) Df|t=0, y=0, φ|t=0 ∈ DA( 1 1 1<br />

2 − 2p − pα , p), if α > 2<br />

p−1 .<br />

53

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

Saved successfully!

Ooh no, something went wrong!