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.

for all t, τ ∈ J, and a.a. x ∈ Ω. Therefore,<br />

Finally,<br />

� T � T �<br />

(<br />

0 0 Ω<br />

|I5(t, τ, x)| p<br />

1<br />

dx dτ dt) p ≤ CT<br />

|t − τ| 1+s1p r−s1 [f − g]X m 1 .<br />

|I6(t, τ, x)| ≤ |bξξ| ∞, m 2|f − g|∞, m|ft(t, x)−ft(τ, x)| ≤ C[f − g]Y m|ft(t, x)−ft(τ, x)|.<br />

for all t, τ ∈ J, and a.a. x ∈ Ω. So we deduce<br />

� T � T �<br />

(<br />

0 0<br />

The assertion follows now from<br />

Ω<br />

[b(·, ·, f) − b(·, ·, g)] Y 1,s 1<br />

1<br />

|I6(t, τ, x)| p<br />

1<br />

dx dτ dt)<br />

|t − τ| 1+s1p<br />

p ≤ C|f − g|Y m[f] (Y 1,s1 1 ) m<br />

≤ C(ρ + [w] 1,s<br />

(Y 1<br />

1 ) m)|f − g|Y m ≤ C(ρ + µ(T ))|f − g|Y m.<br />

≤ I1 +<br />

6�<br />

(<br />

j=3<br />

� T � T<br />

0<br />

0<br />

�<br />

Ω<br />

|Ij(t, τ, x)| p<br />

1<br />

dx dτ dt) p . �<br />

|t − τ| 1+s1p<br />

Under the corresponding assumptions, cf. (H4) in Section 6.1, we can repeat the above<br />

steps <strong>with</strong> the roles of J and Ω being reversed to obtain the estimate<br />

�<br />

�<br />

≤ C ρ + µ(T ) + |bξ(·, ·, w)|∞, m<br />

[b(·, ·, f) − b(·, ·, g)] Y 1,s 2<br />

2<br />

|f − g| (Y 1,s 1<br />

1<br />

It is further not difficult to check that the same line of arguments also yields<br />

�<br />

�<br />

≤ C ρ + µ(T ) + |bξ(·, ·, w)|∞, m<br />

[b(·, ·, f) − b(·, ·, g)] Y 1,s 2<br />

2<br />

|f − g| (Y 0,s 1<br />

1<br />

∩Y 1,s2 2 ) m, f, g ∈ Σ.<br />

∩Y 1,s2 2 ) m, f, g ∈ Σ,<br />

in the case (k1, k2) = (0, 1). Here, the reader should recall that in the proof of Lemma<br />

6.2.2, we employed the embedding Y 1,s1<br />

1 ∩ Y 1,s2<br />

2 ↩→ Cr (J; C( ¯ Ω)) <strong>with</strong> some r ∈ (s1, 1).<br />

In the case (k1, k2) = (0, 1) the situation is more comfortable since, by assumption, we<br />

even have the embedding Y 0,s1<br />

1 ∩ Y 1,s2<br />

2 ↩→ C(J; C1 (Ω)).<br />

To conclude, we see that the estimate (6.19) holds true for F = Y k1,s1<br />

1 ∩ Y k2,s2<br />

2 . As<br />

already mentioned at the beginning of this section, this result can be transferred to the<br />

spaces Y T D and Y T N considered in Section 6.1 by means of the well-known method of local<br />

coordinates. If we apply the corresponding result to the function b defined by<br />

b(t, x, ξ) = b D (t, x, ξ)−b D (t, x, φ(t, x))−bξ(t, x, φ(t, x))(ξ−φ(t, x)), t ∈ J, x ∈ ΓD, ξ ∈ U0,<br />

then we get, owing to bξ(t, x, ξ) = bD ξ (t, x, ξ) − bD ξ (t, x, φ(t, x)), an estimate of the form<br />

|R φ<br />

D (u) − Rφ<br />

D (v)| Y T D ≤ M(µ(T ) + ρ + |bDξ (·, ·, w) − bDξ (·, ·, φ)|∞)|u − v| ZT for all u, v ∈ Σ(ρ, T, φ) ∪ {φ|J}. Since |b D ξ (·, ·, w) − bD ξ (·, ·, φ)|∞ → 0 as T → 0, the<br />

inequality (6.17) <strong>with</strong> K = D follows. In the same way one can also show the validity<br />

of (6.17) <strong>with</strong> K = N.<br />

We turn now to the spaces X m 1, s = Hs p(J; Lp(Ω)), 0 < s < 1.<br />

108

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

Saved successfully!

Ooh no, something went wrong!