Quasilinear parabolic problems with nonlinear boundary conditions
Quasilinear parabolic problems with nonlinear boundary conditions
Quasilinear parabolic problems with nonlinear boundary conditions
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
The strategy for solving (4.1) is now as follows. First we determine the solution of<br />
(4.1) on the interval [0, T1]. Let us denote it by v1 . Now suppose we already know the<br />
solution vi of (4.1) on the interval [0, Ti], where 1 ≤ i ≤ M − 1. We then seek the<br />
solution vi+1 of (4.1) on the larger interval [0, Ti+1] which equals vi on [0, Ti]. The last<br />
step is repeated as long as i < M. Proceeding in this way we finally obtain the solution<br />
of (4.1) on the entire interval [0, T ]. This is the basic idea concerning the localization<br />
in time. Besides, we will also localize (4.1) <strong>with</strong> respect to the space variable. This will<br />
be done in each single time step by means of a partition of unity {ϕj} N j=0 ⊂ C∞ (Rn )<br />
which enjoys the properties �N j=0 ϕj ≡ 1, 0 ≤ ϕj(x) ≤ 1 and supp ϕj ⊂ Uj. We will<br />
also make use of a fixed family {ψj} N j=0 ⊂ C∞ (Rn ) that satisfies ψj ≡ 1 on an open set<br />
Vj containing supp ϕj, and supp ψj ⊂ Uj.<br />
Suppose we are in the (i + 1)th time step of the above procedure. Set (0)Zi+1 =<br />
(0)H α p ([0, Ti+1]; Lp(Rn )) ∩ Lp([0, Ti+1]; H2 p(Rn )). If i = 0 (initial time step), we have to<br />
find v1 in the space<br />
Z1(v 0 ) := {w ∈ Z1 : ∂ m t w|t=0 = ∂ m t f|t=0, if α > m + 1/p, m = 0, 1}.<br />
If i > 0, we assume that v i =: V −1<br />
i f lies in Zi. Here Vi refers to the operator I + k ∗<br />
A(·, x, Dx) on [0, Ti]. Using the notation<br />
Zi+1( ˜w) := {w ∈ Zi+1 : w| [0,Ti] = ˜w} for ˜w ∈ Zi, (4.3)<br />
our aim is then to determine vi+1 in the space Zi+1(vi ). To achieve this, observe that<br />
<strong>with</strong> d(w1, w2) = |w1 − w2|Zi+1 , (Zi+1(vi ), d) is a complete metric space. Our plan is<br />
to transform (4.1) to an appropriate fixed point equation in Zi+1(vi ) and to apply the<br />
contraction principle.<br />
We first derive the local equations associated <strong>with</strong> {ϕj} N j=0 . Note that (4.1) is equivalent<br />
to<br />
v + k ∗ A#(·, x, Dx)v = f − k ∗ AR(·, x, Dx)v,<br />
which when multiplied by ϕj becomes<br />
ϕjv + k ∗ A j<br />
# (·, x, Dx)ϕjv = ϕjf − k ∗ ϕjAR(·, x, Dx)v<br />
+k ∗ [A#(·, x, Dx), ϕj]v. (4.4)<br />
We freeze the coefficients of the local operator A j<br />
# (t, x, Dx) at the point (Ti, xj) to get the<br />
homogeneous differential operator <strong>with</strong> constant coefficients Aij (Dx) := A j<br />
# (Ti, xj, Dx).<br />
Then (4.4) can be written as<br />
ϕjv + k ∗ A ij (Dx)ϕjv = ϕjf − k ∗ ϕjAR(·, x, Dx)v + k ∗ [A#(·, x, Dx), ϕj]v<br />
+k ∗ (A j<br />
# (Ti, xj, Dx) − A j<br />
# (·, x, Dx))ϕjv. (4.5)<br />
Let A ij be the Lp -realization of the differential operator A ij (Dx). For l ∈ {1, . . . , M},<br />
we put Xl = Lp([0, Tl]; Lp(R n )) and define the space Ξl as the set of all functions g ∈<br />
H α p ([0, Tl]; Lp(R n )) satisfying g|t=0 ∈ γ0Z in case α > 1/p, and ∂tg|t=0 ∈ γ1Z in case<br />
α > 1 + 1/p. Ξl endowed <strong>with</strong> the norm<br />
|g|Ξl<br />
= |g|(k,1)<br />
H α p ([0,Tl];Lp(R n )) + χ ( 1<br />
p ,2)(α)|g|t=0|γ0Z + χ (1+ 1<br />
p ,2)(α)|∂tg|t=0|γ1Z<br />
59