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

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Lemma 6.2.4 Let 0 < s1 < r < 1. Then there exists a constant C > 0 not depending<br />

on T such that<br />

[fg] Y 1,s 1<br />

1<br />

for all f, g ∈ Y 1,s1<br />

1<br />

Proof. Clearly Y 1,s1<br />

1<br />

≤ C([f] 1,s<br />

Y 1 |g|∞ + |f|∞[g] 1,s<br />

1<br />

Y 1 + T<br />

1<br />

r−s1 [f]X1 [g]C r 1 + T r−s1 [f]C r[g]X1 )<br />

1<br />

∩ C r (J; C(Ω)).<br />

↩→ X1. Suppose f, g ∈ Y 1,s1<br />

1<br />

∩ C r (J; L∞(Ω)). We estimate<br />

[fg] 1,s<br />

Y 1 =<br />

1<br />

� T � T �<br />

|(ftg)(t, x) + (fgt)(t, x) − (ftg)(τ, x) − (fgt)(τ, x)|<br />

= (<br />

0 0 Ω<br />

p<br />

|t − τ| 1+s1p<br />

dx dτ dt) 1<br />

p<br />

� T � T �<br />

(|ft(t, x) − ft(τ, x)||g(t, x)|)<br />

≤ (<br />

0 0 Ω<br />

p<br />

|t − τ| 1+s1p dx dτ dt) 1<br />

p +<br />

� T � T �<br />

(|ft(τ, x)||g(t, x) − g(τ, x)|)<br />

+(<br />

0 0 Ω<br />

p<br />

|t − τ| 1+s1p dx dτ dt) 1<br />

p +<br />

� T � T �<br />

(|f(t, x)||gt(t, x) − gt(τ, x)|)<br />

+(<br />

0 0 Ω<br />

p<br />

|t − τ| 1+s1p dx dτ dt) 1<br />

p +<br />

� T � T �<br />

(|f(t, x) − f(τ, x)||gt(τ, x)|)<br />

+(<br />

0 0 Ω<br />

p<br />

|t − τ| 1+s1p dx dτ dt) 1<br />

p<br />

≤ [f] 1,s<br />

Y 1 |g|∞ + [g]C<br />

1<br />

r 1 (<br />

� T �<br />

|ft(τ, x)|<br />

0 Ω<br />

p � T dt<br />

(<br />

) dx dτ)1 p +<br />

0 |t − τ| 1+(s1−r)p<br />

+|f|∞[g] 1,s<br />

Y 1 + [f]C<br />

1<br />

r 1 (<br />

� T �<br />

|gt(τ, x)|<br />

0 Ω<br />

p � T dt<br />

(<br />

) dx dτ)1 p<br />

0 |t − τ| 1+(s1−r)p<br />

≤ [f] 1,s<br />

Y 1 |g|∞ + C1T<br />

1<br />

r−s1 [f]X1 [g]C r 1 + |f|∞[g] 1,s<br />

Y 1 + C1T<br />

1<br />

r−s1 [f]C r[g]X1 ,<br />

1<br />

where C1 = [2/(r − s1)p] 1/p . �<br />

A corresponding estimate can be obtained for [fg] 1,s<br />

Y 2 , so together <strong>with</strong> the inequalities<br />

2<br />

from Section 4.2.2, we see that Y k1,s1<br />

1 ∩ Y 1,s2<br />

2 <strong>with</strong> k1 ∈ {0, 1} is a multiplication algebra<br />

provided the above assumptions on p are fulfilled.<br />

We conclude this section by justifying the normalization step which we carried<br />

through in Section 6.1 for the coefficients on the <strong>boundary</strong>. Let f, g ∈ Y T N and f(t, x) ><br />

0, t ∈ J, x ∈ ΓN. By compactness of ΓN, we even have f(t, x) ≥ c, t ∈ J, x ∈ ΓN, for<br />

some positive constant c. Set K = (c/2, ∞) and consider the function b : K → R defined<br />

by b(ξ) = 1/ξ. Clearly f is K-valued, b ∈ C ∞ (K) and b, b ′ are bounded. Therefore<br />

1/f ∈ Y T N . Since Y T N is a multiplication algebra, we deduce that g/f ∈ Y T N , too. By an<br />

analogous argument, this property can also be proved for the space Y T D .<br />

110

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

Saved successfully!

Ooh no, something went wrong!