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.

is a Banach space, cp. Corollary 2.8.2. Recall that for g ∈ 0Ξl := 0H α p ([0, Tl]; Lp(R n )),<br />

one has |g| (k,1)<br />

H α p ([0,Tl];Lp(R n )) = |Bkg|Xl , Bk denoting the inverse convolution operator in Xl<br />

associated <strong>with</strong> the kernel k. For the spaces Zl, l = 1, . . . , M, we choose the norm<br />

|w|Zl = |w|(k,1)<br />

Hα p ([0,Tl];Lp(Rn )) + |∇2xw| Xn2 l<br />

Claim 1: Let i ∈ {0, . . . , M − 1}, j ∈ {0, . . . , N}, and l ∈ {1, . . . , M}. Then<br />

w + k ∗ A ij w = g, t ∈ [0, Tl], x ∈ R n , (4.6)<br />

possesses a unique solution w =: L ij<br />

l g in the space Zl if and only if g ∈ Ξl. Further,<br />

there exists a constant C > 0 not depending on i, j, l such that<br />

|L ij<br />

l g|Zl ≤ C|g|Ξl , ∀g ∈ 0Ξl. (4.7)<br />

Claim 1 is an immediate consequence of Theorem 3.1.4. Note that after a rotation<br />

and a stretch of the spatial coordinates, the elliptic operator Aij becomes the negative<br />

Laplacian. The constant C in (4.7) can be selected to be independent of i and j, due to<br />

the uniform ellipticity assumption (H3); the independence on l is clear, because in (4.7),<br />

functions g in the subspace 0Ξl are considered, only.<br />

By applying the solution operator L ij<br />

i+1 to (4.5) we get<br />

where<br />

and<br />

(I − S ij )ϕjv = L ij<br />

i+1 (ϕjf) + L ij<br />

i+1 k ∗ Cj(·, x, Dx)v =: h ij (f, v), (4.8)<br />

S ij w = L ij<br />

i+1 k ∗ (Aj # (Ti, xj, Dx) − A j<br />

# (·, x, Dx))w, (4.9)<br />

Cj(t, x, Dx) = [A#(t, x, Dx), ϕj] − ϕjAR(t, x, Dx). (4.10)<br />

One immediately verifies that S ij ∈ B(Zi+1). Furthermore, S ij enjoys the subsequent<br />

properties, which will be shown at the end of this proof.<br />

Claim 2: There exists η0 > 0 such that whenever η ≤ η0, i ∈ {0, . . . , M − 1}, and<br />

j ∈ {0, . . . , N},<br />

(a) |S ij w|Zi+1<br />

≤ 1<br />

2 |w|Zi+1 for all w ∈ Zi+1(0) (Z1(0) := 0Z1);<br />

(b) if w ∈ Zi+1(0) and w0 = (I − Sij )w, then |w|Zi+1 ≤ 2|w0|Zi+1 ;<br />

(c) for each g ∈ Ξi+1 and v ∈ Zi+1(V −1<br />

i g), the equation (I − Sij )w = h ij (g, v) ad-<br />

mits a unique solution w =: (I − S ij )| −1<br />

Zi+1(ϕjV −1<br />

.<br />

i g)hij (g, v) in Zi+1(ϕjV −1<br />

i g). Here<br />

Z1(V −1<br />

0 g) := {v ∈ Z1 : ∂ m t v|t=0 = ∂ m t g|t=0, if α > m + 1/p, m = 0, 1}.<br />

Let η ≤ η0. By employing the operators (I − Sij )| −1<br />

Zi+1(ϕjv i desribed in Claim 2 we infer<br />

)<br />

from (4.8) that<br />

ϕjv = (I − S ij )| −1<br />

Zi+1(ϕjv i ) hij (f, v).<br />

60

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

Saved successfully!

Ooh no, something went wrong!