POLONICI MATHEMATICI LIX.2 (1994)
On local motion of a general compressible
viscous heat conducting fluid bounded by a free surface by Ewa Zadrzy´ nska ( L´ od´ z)
and Wojciech M. Zaja ¸czkowski (Warszawa)
Abstract. The motion of a viscous compressible heat conducting fluid in a domain in R3bounded by a free surface is considered. We prove local existence and uniqueness of solutions in Sobolev–Slobodetski˘ı spaces in two cases: with surface tension and without it.
1. Introduction. In this paper we consider the motion of a viscous heat conducting fluid in a bounded domain Ω
t⊂ R
3with a free boundary S
t. Let v = v(x, t) be the velocity of the fluid (i.e. v = (v
1, v
2, v
3)), % = %(x, t) the density, ϑ = ϑ(x, t) the temperature, f = f (x, t) the external force field per unit mass, r = r(x, t) the efficiency of heat sources per unit mass, p = p(%, ϑ) the pressure, µ and ν the constant viscosity coefficients, σ the constant coefficient of surface tension, κ the constant coefficient of heat conductivity, c
v= c
v(%, ϑ) the specific heat at constant volume, p
0the external (constant) pressure. Then the problem is described by the following system (see [4], Chs.
2 and 5):
(1.1)
%(v
t+ v · ∇v) + ∇p − µ∆v − ν∇ div v = %f in e Ω
T,
%
t+ div(%v) = 0 in e Ω
T,
%c
v(ϑ
t+ v · ∇ϑ) + ϑp
ϑdiv v − κ∆ϑ
− µ 2
3
X
i,j=1
(v
i,xj+ v
j,xi)
2− (ν − µ)(div v)
2= %r in e Ω
T,
T · n − σHn = −p
0n on e S
T,
1991 Mathematics Subject Classification: 35A05, 35R35, 76N10.
Key words and phrases: free boundary, compressible viscuous heat conducting fluid, local existence, anisotropic Sobolev spaces.
Research supported by KBN grant no. 211419101.
(1.1)
[cont.]
v · n = − φ
t|∇φ| on e S
T,
∂ϑ
∂n = e ϑ on e S
T, v|
t=0= v
0in Ω ,
%|
t=0= %
0in Ω , ϑ|
t=0= ϑ
0in Ω ,
where φ(x, t) = 0 describes S
t, n is the unit outward vector normal to the boundary (i.e. n = ∇φ/|∇φ|), e Ω
T= S
t∈(0,T )
Ω
t× {t}, Ω
tis the domain of the drop at time t and Ω
0= Ω is its initial domain, e S
T= S
t∈(0,T )
S
t× {t}.
Finally, T = T(v, p) denotes the stress tensor of the form
T = {T
ij} = {−pδ
ij+ µ(v
i,xj+ v
j,xi) + (ν − µ)δ
ijdiv v}
(1.2)
≡ {−pδ
ij+ D
ij(v)} ,
where i, j = 1, 2, 3, D = D(v) = {D
ij} is the deformation tensor. Moreover, thermodynamic considerations imply that c
v> 0, κ > 0, ν ≥
13µ > 0. By H we denote the double mean curvature of S
twhich is negative for convex domains and can be expressed in the form
(1.3) Hn = ∆
St(t)x, x = (x
1, x
2, x
3) ,
where ∆
St(t) is the Laplace–Beltrami operator on S
t. Let S
tbe given by x = x(s
1, s
2, t), (s
1, s
2) ∈ U ⊂ R
2, where U is an open set. Then
∆
St(t) = g
−1/2∂
∂s
αg
−1/2b g
αβ∂
∂s
β(1.4)
= g
−1/2∂
∂s
αg
1/2g
αβ∂
∂s
β(α, β = 1, 2) ,
where the summation convention over repeated indices is assumed, g = det{g
αβ}
α,β=1,2, g
αβ= x
α· x
β(x
α= ∂x/∂s
α), {g
αβ} is the inverse matrix to {g
αβ} and { b g
αβ} is the matrix of algebraic complements of {g
αβ}.
Let the domain Ω be given. Then by (1.1)
5, Ω
t= {x ∈ R
3: x = x(ξ, t), ξ ∈ Ω}, where x = x(ξ, t) is the solution of the Cauchy problem
(1.5) ∂x
∂t = v(x, t), x|
t=0= ξ ∈ Ω, ξ = (ξ
1, ξ
2, ξ
3) .
Therefore the transformation x = x(ξ, t) connects Eulerian x and Lagran- gian ξ coordinates of the same fluid particle. Hence
(1.6) x = ξ +
t
R
0
u(ξ, s) ds ≡ X
u(ξ, t) ,
where u(ξ, t) = v(X
u(ξ, t), t). Moreover, the kinematic boundary condition (1.1)
5implies that the boundary S
tis a material surface. Thus, if ξ ∈ S = S
0then X
u(ξ, t) ∈ S
tand S
t= {x : x = X
u(ξ, t), ξ ∈ S}.
By the equation of continuity (1.1)
2and (1.1)
5the total mass M of the drop is conserved and the following relation between % and Ω
tholds:
(1.7) R
Ωt
%(x, t) dx = M .
The aim of this paper is to prove the local-in-time existence and unique- ness of solutions to problem (1.1) in Sobolev–Slobodetski˘ı spaces (see def- inition in Sect. 2). In the case of compressible barotropic fluid the corre- sponding drop problem has been considered by W. M. Zaj¸ aczkowski in [13]
and [16], while papers [14] and [15] refer to the global existence of solution to the same drop problem. Local existence of solutions in the compressible barotropic case was also considered in [5], [6], [12], while in the incompress- ible barotropic case local existence is proved in [2] and [10].
This paper consists of four sections. In Section 2 notation and auxiliary results are presented. In Section 3 we prove the local existence and unique- ness of solution to problem (1.1) in the case σ = 0. In this case there is no surface tension. Finally, Section 4 concerns the local existence and unique- ness of solution to problem (1.1) in the case σ 6= 0, i.e. when the shape of the free boundary S
tof Ω
tis governed by surface tension.
2. Notations and auxiliary results. In Sections 3 and 4 of this paper we use the anisotropic Sobolev–Slobodetski˘ı spaces W
2l,l/2(Q
T), l ∈ R
1+(see [3]), of functions defined in Q
Twhere Q
T= Ω
T≡ Ω × (0, T ) (Ω ⊂ R
3is a domain, T ≤ ∞) or Q
T= S
T≡ S × (0, T ), S = ∂Ω.
We define W
2l,l/2(Ω
T) as the space of functions u such that kuk
W2l,l/2(ΩT)=
X
|α|+2i≤[l]
kD
ξα∂
tiuk
2L2(ΩT)(2.1)
+ X
|α|+2i=[l]
TR
0
R
Ω
R
Ω
|D
ξα∂
itu(ξ, t) − D
αξ0∂
tiu(ξ
0, t)|
2|ξ − ξ
0|
3+2(l−[l])dξ dξ
0dt
+ R
Ω T
R
0 T
R
0
|D
ξα∂
tiu(ξ, t) − D
αξ∂
ti0u(ξ, t
0)|
2|t − t
0|
1+2(l/2−[l/2])dt dt
0dξ
1/2≡ h X
|α|+2i≤[l]
|D
ξα∂
tiu|
22,ΩT+ X
|α|+2i=[l]
([D
ξα∂
tiu]
2l−[l],2,ΩT,ξ+ [D
αξ∂
tiu]
2l/2−[l/2],2,ΩT,t) i
1/2< ∞ ,
where we use generalized (Sobolev) derivatives, D
αξ= ∂
ξα11
∂
ξα22
∂
ξα33
, ∂
ξαjj
=
∂
αj/∂ξ
jαj(j = 1, 2, 3), α = (α
1, α
2, α
3) is a multiindex, |α| = α
1+ α
2+ α
3,
∂
ti= ∂
i/∂t
iand [l] is the integer part of l. In the case when l is an integer the second terms in the above formulae must be omitted, while in the case of l/2 being integer the last terms in the above formulae must be omitted as well.
Similarly to W
2l,l/2(Ω
T), using local mappings and a partition of unity we introduce the normed space W
2l,l/2(S
T) of functions defined on S
T= S × (0, T ), where S = ∂Ω.
We also use the usual Sobolev spaces W
2l(Q), where l ∈ R
+, Q = Ω (Ω ⊂ R
3is a bounded domain) or Q = S. In the case Q = Ω the norm in W
2l(Ω) is defined as follows:
kuk
Wl2(Ω)
=
X
|α|≤[l]
kD
αξuk
2L2(Ω)+ X
|α|=[l]
R
Ω
R
Ω
|D
αξu(ξ) − D
ξα0u(ξ
0)|
2|ξ − ξ
0|
3+2(l−[l])dξ dξ
0 1/2≡ X
|α|≤[l]
|D
αξu|
22,Ω+ [D
ξαu]
2l−[l],2,Ω1/2,
where the last term is omitted when l is an integer. Similarly, by using local mappings and a partition of unity we define W
2l(S).
To simplify notation we write kuk
l,Q= kuk
Wl,l/22 (Q)
if Q = Ω
Tor Q = S
T, l ≥ 0 ; kuk
l,Q= kuk
Wl2(Q)
if Q = Ω or Q = S, l ≥ 0 , and W
20,0(Q) = W
20(Q) = L
2(Q). Moreover,
kuk
Lp(Q)= |u|
p,Q, 1 ≤ p ≤ ∞ . Next introduce the space Γ
0l,l/2(Ω) with the norm
kuk
Γ0l,l/2(Ω)= X
|α|+2i≤l
kD
ξα∂
tiuk
0,Ω≡ |u|
l,0,Ωand the space L
p(0, T ; Γ
0l,l/2(Ω)) with the norm
kuk
Lp(0,T ;Γ0l,l/2(Ω))≡ u
l,0,p,ΩT, where 1 ≤ p ≤ ∞ .
Moreover, let C
α(Ω
T) (α ∈ (0, 1)) denote the H¨ older space with the norm
kuk
Cα(ΩT)= sup
ΩT
|u(ξ, t) − u(ξ
0, t
0)|
(p|ξ − ξ
0|
2+ |t − t
0|
2)
α;
let C
B0(Ω
T) be the space of continuous bounded functions on Ω
Twith the norm
kuk
C0B(ΩT)
= sup
ΩT
|u(ξ, t)|
and let C
2,1(Q) (Q ⊂ R
3× (0, T )) denote the space of functions u such that D
αξ∂
tiu ∈ C
0(Q) for |α| + 2i ≤ 2.
Finally, the following seminorms are used:
u
κ,QT=
TR
0
|u|
22,Qt
2κdt
1/2,
where Q = Ω (Ω ⊂ R
3is a bounded domain) or Q = S, and κ ∈ (0, 1);
[u]
l,2,Q= [u]
l,2,Q,ξ+ [u]
l,2,Q,t, where
[u]
l,2,Q,ξ= X
|α|+2i=[l]
[D
ξα∂
tiu]
l−[l],2,Q,ξ, [u]
l,2,Q,t= X
|α|+2i=[l]
[D
ξα∂
tiu]
l/2−[l/2],2,Q,t,
Q = Ω × J (Ω ⊂ R
3is a domain, J = (0, T ) or J = (−∞, T )) or Q = S × J . In the case when J = (0, T ) the seminorms [D
αξ∂
tiu]
l−[l],2,Q,ξand [D
ξα∂
tiu]
l/2−[l/2],2,Q,tare defined in (2.1). In the case when J = (−∞, T ) we define the above seminorms in the same way.
Let X be whichever of the function spaces mentioned above. We say that a vector-valued function u = (u
1, . . . , u
ν) belongs to X if u
i∈ X for any 1 ≤ i ≤ ν.
In the sequel we shall use various notations for derivatives of u (where u is a scalar- or vector-valued function u = (u
1, u
2, u
3)). If u is a scalar-valued function we denote by D
ξku (where ξ ∈ Ω ⊂ R
3) the vector of all derivatives of u of order k, i.e. D
kξu = (D
αξu)
|α|=k. Similarly, if u = (u
1, u
2, u
3) we denote by D
ξku the vector (D
ξαu
j)
|α|=k, j=1,2,3. By D
kξ,tu we denote the vector (D
ξα∂
tiu
j)
|α|+2i=k,j=1,2,3in the case when u = (u
1, u
2, u
3) and the vector (D
ξα∂
tiu)
|α|+2i=kin the scalar case. Hence
|D
kξu| = X
|α|=k
|D
ξαu| and |D
kξ,tu| = X
|α|+2i=k
|D
ξα∂
tiu| .
We also use the notation ∇
ξu ≡ D
1ξu or u
ξ≡ D
ξ1u.
Next, we denote by u · v either the scalar product of vectors u and v, or
the product of matrices u and v.
Finally, we denote by D
kξ,tuD
lξ,tv the following number:
(2.2) D
ξ,tkuD
ξ,tlv = X
|α|+2i=k
|β|+2j=l p
X
m=1 s
X
n=1
D
αξ∂
tiu
mD
ξβ∂
tjv
n,
where u = (u
1, . . . , u
p), v = (v
1, . . . , v
s) (k ≥ 0, l ≥ 0, p > 1, s > 1). The product of more than two such factors is defined similarly.
We use the following lemmas.
Lemma 2.1. The following imbedding holds:
(2.3) W
rl(Ω) ⊂ L
αp(Ω) (Ω ⊂ R
3) ,
where |α| + 3/r − 3 ≤ l, l ∈ Z, 1 ≤ p, r ≤ ∞; L
αp(Ω) is the space of functions u such that |D
ξαu|
p,Ω< ∞; W
rl(Ω) is the Sobolev space.
Moreover , the following interpolation inequalities hold : (2.4) |D
αξu|
p,Ω≤ cε
1−κ|D
lξu|
r,Ω+ cε
−κ|u|
r,Ω,
where κ = |α|/l + 3/(lr) − 3/(lp) < 1, ε is a parameter and c > 0 is a constant independent of u and ε;
(2.5) |D
ξαu|
q,S≤ cε
1−κ|D
ξlu|
r,Ω+ cε
−κ|u|
r,Ω,
where κ = |α|/l + 3/(lr) − 2/(lq) < 1, ε is a parameter and c > 0 is a constant independent of u and ε.
Lemma 2.1 follows from Theorem 10.2 of [3].
Lemma 2.2 (see [7]). For sufficiently regular u we have k∂
tiu(t)k
2l−1−2i,Ω≤ c(kuk
2l,ΩT+ k∂
tiu(0)k
2l−1−2i,Ω) , where 0 ≤ 2i ≤ 2l − 1, l ∈ N and c > 0 is a constant independent of T .
Lemma 2.3. Let u(ξ, t) = 0 for t ≤ 0. Then
T
R
−∞
dt
T
R
−∞
dt
0|u(ξ, t) − u(ξ, t
0)|
22,Q|t − t
0|
1+2α≤
T
R
0
dt
T
R
0
dt
0|u(ξ, t) − u(ξ, t
0)|
22,Q|t − t
0|
1+2α+ 1
α
T
R
0
|u(t)|
22,Qt
2αdt , where Q = Ω (Ω ⊂ R
3is a domain) or Q = S = ∂Ω, and α ∈ R.
Lemma 2.4. Let τ ∈ (0, 1). Then for all u ∈ W
20,τ /2(Ω
T), (2.6)
T
R
0
|u|
22,Ωdt t
τ≤ c
1T
R
0
dt
T
R
0
dt
0|u(·, t) − u(·, t
0)|
22,Ω|t − t
0|
1+τ+ c
2T
−τT
R
0
|u|
22,Ωdt ,
where c
1, c
2do not depend on T and u. For T = ∞ the last term in (2.6) vanishes.
This was shown in [11], Lemma 6.3.
3. Local existence in the case σ = 0. In order to prove local existence of solutions of (1.1) we rewrite it in the Lagrangian coordinates introduced by (1.5) and (1.6):
(3.1)
ηu
t− µ∇
2uu − ν∇
u∇
u· u + ∇
up(η, γ) = ηg in Ω
T,
η
t+ η∇
u· u = 0 in Ω
T,
ηc
v(η, γ)γ
t− κ∇
2uγ = −γp
γ(η, γ)∇
u· u + µ
2
3
X
i,j=1
(ξ
xi· ∇
ξu
j+ ξ
xj· ∇
ξu
i)
2+ (ν − µ)(∇
u· u)
2+ ηk in Ω
T,
T
u(u, p) · n = −p
0n on S
T,
n · ∇
uγ = e γ on S
T,
u|
t=0= v
0in Ω ,
η|
t=0= ρ
0in Ω ,
γ|
t=0= ϑ
0in Ω ,
where u(ξ, t) = v(X
u(ξ, t), t), γ(ξ, t) = ϑ(X
u(ξ, t), t), η(ξ, t) = ρ(X
u(ξ, t), t), g(ξ, t) = f (X
u(ξ, t), t), k(ξ, t) = r(X
u(ξ, t), t), ∇
u= ξ
x∇
ξ≡ {ξ
ix∂
ξi},
T
u(u, p) = −p1 + D
u(u) ,
D
u(u) = {µ(ξ
kxi∂
ξku
j+ ξ
kxj∂
ξku
i) + (ν − µ)δ
ij∇
u· u}
(here the summation convention over repeated indices is assumed and 1 is the unit matrix) and γ(ξ, t) = e e ϑ(X
u(ξ, t), t).
Let A = {a
ij} be the Jacobi matrix of the transformation x = X
u(ξ, t), where a
ij= δ
ij+ R
t0
∂
ξju
i(ξ, τ ) dτ . Assuming that |∇
ξu|
∞,ΩT≤ M we obtain (3.2) 0 < c
1(1 − M t)
3≤ det{x
ξ} ≤ c
2(1 + M t)
3, t ≤ T ,
where c
1, c
2> 0 are constants and T > 0 is sufficiently small. Moreover, det A = exp
tR
0
∇
u· u dτ
= ρ
0/η .
Let S
tbe determined (at least locally) by the equation φ(x, t) = 0. Then S is described by φ(x(ξ, t), t)|
t=0≡ e φ(ξ) = 0. Thus, we have
n(x(ξ, t), t) = − ∇
xφ(x, t)
|∇
xφ(x, t)|
x=x(ξ,t)
and n
0(ξ) = − ∇
ξφ(ξ) e
|∇
ξφ(ξ)| e . First we consider the linear problems
(3.3)
Lu ≡ u
t− µ∇
2ξu − ν∇
ξ∇
ξ· u = F in Ω
T,
D
ξ(u) · n
0= G on S
T,
u|
t=0= u
0in Ω
(where D
ξ(u) = {µ(∂
ξiu
j+ ∂
ξju
i) + (ν − µ)δ
ij∂
ξku
k}) and
(3.4)
γ
t− κ∇
2ξγ = K in Ω
T, n
0· ∇
ξγ = e γ on S
T, γ|
t=0= γ
0in Ω . We assume
(3.5) F ∈ W
22,1(Ω
T), G ∈ W
3−1/2,3/2−1/42
(S
T),
D
ξ,t2G
1/4,ST< ∞, u
0∈ W
23(Ω) and
(3.6) K ∈ W
22,1(Ω
T), e γ ∈ W
3−1/2,3/2−1/42
(S
T),
D
2ξ,te γ
1/4,ST< ∞, γ
0∈ W
23(Ω) . Moreover, we assume the following compatibility conditions:
(3.7) D
ξα(D
ξ(u(0)) · n
0− G(0)) = 0, |α| ≤ 1, on S , and
(3.8) D
αξ(n
0· ∇
ξγ(0) − e γ(0)) = 0, |α| ≤ 1, on S .
First we consider problem (3.3). Define functions ψ
ifor i = 0, 1 by (3.9) ψ
i= ∂
tiu|
t=0in Ω .
Hence
(3.10) ψ
0= u
0, ψ
1= µ∆u
0+ ν∇ div u
0+ F (0) .
Lemma 3.1. Let Ω ⊂ R
3be either a halfspace or a bounded domain with smooth boundary ∂Ω and let T ≤ ∞. Assume that
(3.11) ψ
0∈ W
23(Ω), ψ
1∈ W
21(Ω) . Then there exists v ∈ W
24,2(Ω
T) such that
(3.12) ∂
itv|
t=0= ψ
iin Ω (i = 0, 1)
and
(3.13) kvk
4,ΩT≤ c(kψ
0k
3,Ω+ kψ
1k
1,Ω) , where c > 0 is a constant independent of T .
P r o o f. Using the Hestenes–Whitney method (see [1]) we can extend ψ
0and ψ
1to functions e ψ
0∈ W
23(R
3) and e ψ
1∈ W
21(R
3) such that k e ψ
0k
3,R3≤ ckψ
0k
3,Ω, k e ψ
1k
1,R3≤ ckψ
1k
1,Ω,
where c = c(Ω). Then from [16] (Lemma 6.5) we deduce that there exists e v ∈ W
24,2(R
3× R
1+) such that
(3.14) ∂
tie v|
t=0= e ψ
i, i = 0, 1 , and
(3.15) k e vk
4,R3×R1+
≤ c(k e ψ
0k
3,R3+ k e ψ
1k
1,R3) ≤ c(kψ
0k
3,Ω+ kψ
1k
1,Ω) . Therefore v = e v|
ΩTsatisfies conditions (3.12) and (3.13).
Now we prove the following theorem.
Theorem 3.1. Let S ∈ W
24−1/2and let assumptions (3.5) and (3.7) be satisfied (T < ∞). Then there exists a solution of (3.3) such that u ∈ W
24,2(Ω
T) and
(3.16) kuk
4,ΩT≤ c(T )(kF k
2,ΩT+ kGk
3−1/2,ST+ D
ξ,t2G
1/4,ST+ |u(0)|
3,0,Ω) , where c is an increasing continuous function of T and
(3.17) |u(0)|
3,0,Ω= ku
0k
3,Ω+ k∂
tu|
t=0k
1,Ω. P r o o f. Introduce the function
(3.18) w = u − v ,
where v is the function from Lemma 3.1. Then instead of (3.3) we obtain the problem
(3.19)
w
t− µ∆w − ν∇ div w = F − Lv ≡ f in Ω
T, D
ξ(w) · n
0= G − D
ξ(v) · n
0≡ g on S
T,
w|
t=0= 0 in Ω ,
where in view of the compatibility condition (3.7) we have
(3.20) f (0) = g(0) = 0 .
It is sufficient to consider problem (3.19) in Ω
T≡ R
T≡ R
3+× [0, T ] because using a partition of unity and appropriate norms (see [8], Sects.
20, 21) we obtain the existence and the appropriate estimate of solutions of
(3.19) in a bounded domain Ω.
Thus, consider problem (3.19) in R
Tand extend functions f , g, w by zero for t < 0 to functions f
1, g
1, w
1. Then instead of (3.19) we get the following boundary value problem:
(3.21) Lw
1= f
1in e D
4(T ) = R
3+× (−∞, T ] , D
ξ(w
1) · n
0= g
1in e E
3(T ) = R
2× (−∞, T ] .
Next using the Hestenes–Whitney method extend f
1and g
1to functions f
2and g
2defined on R
3+× (−∞, ∞) and R
2× (−∞, ∞), respectively. Then instead of (3.21) we have the problem
(3.22)
Lw
2= f
2in e D
4= R
3+× (−∞, ∞) , D
ξ(w
2) · n
0= g
2in e E
3= R
2× (−∞, ∞) , and the estimate
(3.23) kf
2k
2,D
e
4≤ ckf
1k
2,D
e
4(T ).
Next we extend f
2by the Hestenes–Whitney method to a function f
3∈ W
24,2(R
4) such that
(3.24) kf
3k
2,R4≤ ckf
2k
2,D
e
4. Consider now the system
(3.25) Lw
3= f
3in R
4.
By potential techniques (see [8], Sections 12 and 21) and (3.23), (3.24) there exists a solution w
3∈ W
24,2(R
4) of (3.25) and
(3.26) kw
3k
4,R4≤ c(T )kf k
2,RT, where c(T ) is an increasing function of T .
Introduce the function
(3.27) w
4= w
2− w
3.
By (3.22) and (3.25) we have (3.28)
Lw
4= 0 in e D
4,
D
ξ(w
4) · n
0= g
2− D
ξ(w
3) · n
0≡ g
3in e E
3.
Again by potential techniques there exists a solution w
4∈ W
24,2( e D
4) of (3.28) and
kw
4k
4,D
e
4≤ c[g
3]
3−1/2,2,E
e
3≤ c([g
2]
3−1/2,2,E
e
3+ kw
3k
4,D
e
4)
≤ c(T )([g
1]
3−1/2,2,E e
3(T ) + kf k
2,RT) ,
where we used (3.26) and the Hestenes–Whitney method for g
1. Hence
(3.29) kw
4k
4,RT≤ c(T )([g
1]
3−1/2,2,E
e
3(T )+ kf k
2,RT) . Further, using Lemma 2.3 we have
[g
1]
3−1/2,2,E
e
3(T )= X
|α|=2
[D
αξg
1]
1/2,2,E
e
3(T ),ξ+ [∂
tg
1]
1/2,2,E
e
3(T ),ξ(3.30)
+ X
|α|=2
[D
αξg
1]
1/4,2,E
e
3(T ),t+ [∂
tg
1]
1/4,2,E
e
3(T ),t≤ c
X
|α|=2
[D
αξg]
1/2,2,R2×[0,T ],ξ+ [∂
tg]
1/2,2,R2×[0,T ],ξ+ X
|α|=2
[D
αξg]
1/4,2,R2×[0,T ],t+ [∂
tg]
1/4,2,R2×[0,T ],t+
TR
0
|D
ξ,t2g|
22,R2t
1/2dt
1/2≤ c(kgk
3−1/2,ST+ D
ξ,t2g
1/4,ST) , where S
T= R
2× [0, T ].
Taking into account the right-hand side of (3.19) and (3.18), (3.21), (3.22), (3.25)–(3.30) we conclude that there exists a solution u ∈ W
24,2(Ω
T) of (3.3) satisfying
(3.31) kuk
4,RT≤ c(T )(kF k
2,RT+ kvk
4,RT+ kGk
3−1/2,ST+ D
2ξ,tG
1/4,ST+ kv
ξn
0k
3−1/2,ST+ D
2ξ,t(v
ξn
0)
1/4,ST) , where
v
ξn
0=
3
X
i,j,k=1
v
iξjn
0k. By Lemma 3.1 we have
(3.32) kvk
4,RT≤ c(kφ
0k
3,R3+
+ kφ
1k
1,R3+
) and
kv
ξn
0k
3−1/2,ST≤ k e v
ξn
0k
3−1/2,S×(0,∞)≤ ck v e
ξk
3,R3 +×(0,∞)(3.33)
≤ ckvk
4,RT≤ c(kφ
0k
3,R3+
+ kφ
1k
1,R3 +) , where we have used the fact that S ∈ W
24−1/2.
It remains to estimate D
ξ,t2(v
ξn
0)
1/4,ST. We have
(3.34) D
2ξ,t(v
ξn
0)
1/4,ST=
TR
0
|D
2ξn
0D
1ξv + D
1ξn
0D
ξ2v + n
0D
ξ3v + n
0∂
tD
1ξv|
22,St
1/2dt
1/2≤ c
TR
0
|D
2ξn
0|
24,S|D
ξ1v|
24,S+ |D
2ξv|
22,S+ |D
3ξv|
22,S+ |∂
tD
ξ1v|
22,St
1/2dt
1/2, where we have used the fact that S ∈ W
24−1/2and Lemma 2.1, and where the products are understood in the sense of (2.2).
Next, by Lemma 2.1 we get
T
R
0
|D
2ξn
0|
24,S|D
ξ1v|
24,St
1/2dt ≤ c
T
R
0
kD
1ξvk
21,St
1/2dt . Hence in view of Lemma 2.4, (3.34) yields
D
ξ,t2(v
ξn
0)
1/4,ST≤ c
∞R
0
|D
ξ1e v|
22,S+ |D
2ξe v|
22,S+ |D
3ξe v|
22,S+ |D
1ξ∂
te v|
22,St
1/2dt
1/2≤ c
1 ∞R
0
dt
∞
R
0
dt
0|D
ξ1e v(ξ, t) − D
1ξe v(ξ, t
0)|
22,S|t − t
0|
1+1/2+ |D
2ξe v(ξ, t) − D
ξ2e v(ξ, t
0)|
22,S|t − t
0|
1+1/2+ |∂
tD
ξ1e v(ξ, t) − ∂
t0D
1ξv(ξ, t e
0)|
22,S|t − t
0|
1+1/2+ |D
3ξe v(ξ, t) − D
ξ3e v(ξ, t
0)|
22,S|t − t
0|
1+1/2 1/2. Therefore
(3.35) D
ξ,t2(v
ξn
0)
1/4,ST≤ ckvk
4,RT, where we have used Theorem 5.1 from [8].
Taking into account (3.31)–(3.33) and (3.35) we get (3.16). This com- pletes the proof of the theorem.
In the same way we can prove
Theorem 3.2. Let S ∈ W
24−1/2and let assumptions (3.6) and (3.8) be satisfied (T < ∞). Then there exists a solution of (3.4) such that γ ∈ W
24,2(Ω
T) and
(3.36) kγk
4,ΩT≤ c(T )(kKk
2,ΩT+ k e γk
3−1/2,ST+ D
2ξ,te γ
1/4,ST+ |γ(0)|
3,0,Ω) ,
where c is an increasing function of T and
(3.37) |γ(0)|
3,0,Ω= kγ
0k
3,Ω+ k∂
t1γ|
t=0k
1,Ω. Now we have to consider the following problems:
(3.38)
ηu
t− µ∇
2ξu − ν∇
ξ∇
ξ· u = F in Ω
T, D
ξ(u) · n
0= G on S
T,
u|
t=0= u
0in Ω
and
(3.39)
ηc
v(η, β)γ
t− κ∇
2ξγ = K in Ω
T, n
0· ∇
ξγ = e γ on S
T,
γ|
t=0= γ
0in Ω .
First we consider (3.38). The following theorem is proved in [15].
Theorem 3.3. Assume that
S ∈ W
24−1/2, F ∈ W
22,1(Ω
T), G ∈ W
3−1/2,3/2−1/42
(S
T),
D
2ξ,tG
1/4,ST< ∞, u
0∈ W
23(Ω),
η ∈ L
∞(0, T ; Γ
02,1(Ω)) ∩ C
α(Ω
T) (α ∈ (0, 1)), 1/η ∈ L
∞(Ω
T).
Moreover , let the compatibility condition (3.7) be satisfied. Then there exists a unique solution u ∈ W
24,2(Ω
T) to problem (3.38) satisfying the estimate
kuk
4,ΩT≤ φ
1(|1/η|
∞,ΩT, |η|
∞,ΩT, T )[kF k
2,ΩT+ kGk
3−1/2,ST(3.40)
+ D
ξ,t2G
1/4,ST+ φ
2( η
2,0,∞,ΩT, kηk
Cα(ΩT))kuk
2,ΩT] + φ
3(|1/η|
∞,ΩT, T )|u(0)|
3,0,Ω,
where φ
i(i = 1, 2, 3) are nonnegative increasing continuous functions of their arguments.
Now we consider problem (3.39). Assume that η ∈ L
∞(0, T ; Γ
02,1(Ω)) and β ∈ W
24,2(Ω
T). Applying Theorems 10.2 and 10.4 of [3] we find that η ∈ C
B0(Ω
T) and β ∈ C
B0(Ω
T). Therefore, there exists a bounded domain V ⊂ R
2such that (η(ξ, t), β(ξ, t)) ∈ V for any (ξ, t) ∈ Ω
T.
Lemma 3.2. Assume that η ∈ L
∞(0, T ; Γ
02,1(Ω)) ∩ C
α(Ω
T) (where α ∈ (0, 1/2)), 1/η ∈ L
∞(Ω
T), η > 0, β ∈ W
24,2(Ω
T), 1/β ∈ L
∞(Ω
T), β > 0, c
v∈ C
2(R
2+), c
v> 0. Then ηc
v(η, β) ∈ L
∞(0, T ; Γ
02,1(Ω)), ηc
v∈ C
α(Ω
T), 1/(ηc
v(η, β)) ∈ L
∞(Ω
T) and
(3.41) ηc
v(η, β)
2,0,∞,ΩT≤ ψ
1(kc
vk
C2(V ), k1/c
vk
C0(V ), η
2,0,∞,ΩT, kβk
4,ΩT, |β(0)|
3,0,Ω) ,
(3.42) |ηc
v(η, β)|
∞,ΩT≤ kc
vk
C0(V )|η|
∞,ΩT,
(3.43) |1/(ηc
v(η, β))|
∞,ΩT≤ k1/c
vk
C0(V )|1/η|
∞,ΩT, (3.44) kηc
v(η, β)k
Cα(ΩT)≤ ψ
2(kc
vk
C2(V ), kηk
Cα(ΩT), kβk
4,ΩT, |β(0)|
3,0,Ω, T ) , where ψ
i(i = 1, 2) are nonnegative increasing continuous functions of their arguments.
P r o o f. By the assumptions we can choose V such that V ⊂ R
2+. Thus (3.42) and (3.43) are obviously satisfied. In order to obtain (3.41) we cal- culate the derivatives D
αξ∂
ti(ηc
v) (where |α| + 2i ≤ 2) and next we apply Lemmas 2.1 and 2.2. To prove (3.44) we also use Lemmas 2.1 and 2.2 and the Sobolev imbedding theorem.
Theorem 3.4. Assume that
S ∈ W
24−1/2, K ∈ W
22,1(Ω
T), e γ ∈ W
3−1/2,3/2−1/42