Professional Documents
Culture Documents
T
= (0, T), S
T
= (0, T).
We use notation:
= (0, ), S
= (0, ),
u = (u
1
, u
2
, ..., u
s
), u
t
k = (u
1t
k, u
2t
k, ..., u
st
k), u
jt
k =
k
u
j
/t
k
,
D
u = (D
u
1
, D
u
2
, ..., D
u
s
), D
u
j
=
||
u
j
x
1
1
x
2
2
x
n
n
,
where = (
1
,
2
, ...,
n
) and [[ =
1
+
2
+ +
n
.
Throughout the paper we will use the following functional spaces:
(1) H
m
() is the space of complex vector functions u = u(x), x having
generalized derivatives D
u L
2
() with the norm
|u|
H
m
()
=
_
_
||m
_
[D
u[
2
dx
_
_
1/2
.
(2) H
m,k
(
T
) is the space of complex vector functions u = u(x, t), (x, t)
T
having generalized derivatives D
u L
2
(
T
), [[ m and u
t
l L
2
(
T
),
1 l k with the norm
|u|
2
H
m,k
(
T
)
=
_
T
_
_
||m
[D
u[
2
+
k
l=1
[u
t
l[
2
_
_
dxdt.
In particular,
|u|
2
H
m,0
(
T
)
=
||m
_
T
[D
u[
2
dxdt.
(3)
H
m,k
(
T
) is the closure in H
m,k
(
T
) of the set of all infinitely differentiable
complex vector functions on
T
which vanish near S
T
.
Solvability of the First Mixed Problem 2603
(4) H
m,k
(e
t
;
u, [[ m, u
t
l , 1 l k such that
|u|
2
H
m,k
(e
t
,)
=
_
_
_
||m
[D
u[
2
+
k
l=1
[u
t
l[
2
_
_
e
2t
dxdt < ,
where is a positive constant.
(5)
H
m,k
(e
t
,
) is the closure in H
m,k
(e
t
,
.
(6) L
(0, ; L
2
()) is the space of measurable complex valued functions u :
(0, ) L
2
(); t u(, t) satisfying
|u|
L
(0,;L
2
())
= ess sup
t>0
|u(, t)|
L
2
()
< .
We introduce the differential operator
(2.1) L(x, t, D) :=
||,||m
D
(a
(x, t)D
) +
m
||=1
a
(x, t)D
+ a(x, t),
where a
= a
(x, t), a
= a
; a
= (1)
||+||
a
, a
are
complex conjugate transportation matrices of a
.
Let f belong to L
(0, ; L
2
()). We consider the following in the cylinder
:
(2.2) (1)
m1
L(x, t, D)u u
tt
= f,
with the initial condition
(2.3) u[
t=0
= 0, u
t
[
t=0
= 0,
and boundary conditions
(2.4)
j
u
j
[
S
= 0; j = 0, ..., (m1),
where
j
u
j
is derivative with respect to the outer unit normal of S
.
2604 Bui Trong Kim and Nguyen Manh Hung
Definition 2.1. The differential operator L is said to be strongly hyperbolic in
(x, t)
h
0
[[
2m
[[
2
,
for al R
n
0, C
s
0 and (x, t)
.
Definition 2.2. A function u(x, t) is called a generalized solution of the
problem (2.2)-(2.4) in the space H
m,1
(e
t
;
T
_
_
m
||,||=1
(1)
||
a
uD
+
m
||=1
a
u +au
_
_
dxdt
+
_
T
u
t
t
dxdt =
_
T
fdxdt.
holds for all
H
m,1
(
T
) satisfying (, T) = 0.
3. SOLVABILITY OF THE PROBLEM
In this section we establish theorems on the uniqueness and existence of a
generalized solution of problem (2.2)-(2.4). First we prove the following lemma
which is a generalization of the Garding inequality (see [8]).
Lemma 3.1. Assume that L is strongly hyperbolic in
and a
(x, t) are
continuous in x uniformly with respect to t [0, ) whenever [[ = [[ = m.
Then there exist constants
0
> 0 and
0
0 such that
(3.1) (1)
m
B(u, u)(t)
0
|u(, t)|
2
H
m
()
0
|u(, t)|
2
L
2
()
for all u
H
m,1
(e
t
;
), where B is defined by
(3.2) B(u, u)(t) =
m
||,||=1
(1)
||
_
uD
udx.
Proof. We use the notation
ls
= (1)
|l|+m
a
ls
for [l[ + [s[ < 2m and rewrite
(1)
m
B(u, u)(t) =
||=||=m
_
uD
udx +
|l|+|s|<2m
_
b
ls
D
s
uD
l
udx.
Solvability of the First Mixed Problem 2605
Since is bounded, there exists a cube R
n
such that . We now consider
the following cases.
Case 1. b
ls
= 0 for all l, s satisfying [l[ + [s[ < 2m and a
(x, t) = a
(t).
For each u
H
m
(), using the Fourier expansion, we have
u(x, t) = (2)
n/2
k=
c
k
e
ikx
,
where
c
k
= c
k
(t) = (2)
n/2
_
u(x, t)e
ikx
dx.
Hence
D
u = (2)
n/2
k=
i
||
k
c
k
e
ikx
,
a
u = (2)
n/2
k=
i
||
k
c
k
e
ikx
.
Using Parseval equality, the strongly hyperbolic condition of L and Friedrichs in-
equality we obtain
(1)
m
B(u, u)(t) = (2)
n/2
k=
||=||=m
a
c
k
c
k
h
0
k=
[k[
2m
[c
k
[
2
0
|u(, t)|
2
H
m
()
,
where
0
=
h
0
C
(mes)
2/m
, C is a constant depending on n only.
Case 2. We put
G := x : u(x, t) ,= 0
and assume that DiamG <
0
for some
0
> 0. Let (x
0
, t) G. than there exists
a positive constant C
1
which is independent of (x
0
, t) such that
(3.3) C
1
|u(, t)|
2
H
m
()
||=||=m
_
(x
0
, t)D
uD
udx.
2606 Bui Trong Kim and Nguyen Manh Hung
On the other hand we have
||=||=m
_
(x
0
, t)D
uD
udx
=
||=||=m
_
(x, t)D
uD
udx +
|l|+|s|<2m
_
b
ls
(x, t)D
s
uD
l
udx
+
||=||=m
_
[a
(x
0
, t) a
(x, t)]D
uD
udx
|l|+|s|<2m
_
b
ls
(x, t)D
s
uD
l
udx
(1)
m
B(u, u)(t) +C
2
|u(, t)|
H
m
()
|u(, t)|
H
m1
()
+C
3
max
xG
_
_
_
||=||=m
[a
(x
0
, t) a
(x, t)[
_
_
_
|u(, t)|
2
H
m
()
,
where C
2
and C
3
are positive constants. Since a
(x, t) is continuous in x
uniformly with respect to t (0, ), for
0
> 0 sufficiently small, we have
max
xG
_
_
_
||=||=m
[a
(x
0
, t) a
(x, t)[
_
_
_
< C
1
/2C
3
.
Combining this with (3.3) we get
C
1
2
|u(, t)|
2
H
m
()
(1)
m
B(u, u)(t) +C
2
|u(, t)|
H
m
()
|u(, t)|
H
m1
()
.
By the interpolation inequality (see [8]) we obtain the desired inequality.
Case 3. Consider the general case. We choose a partition of unity in ,
1hN
2
h
= 1,
h
C
||=||=m
[a
(x
1
, t) a
(x
2
, t)[
_
_
_
< C
1
/2C
3
for [x
1
x
2
[ <
0
.
Solvability of the First Mixed Problem 2607
Using the partition of unity above we have
(1)
m
B(u, u)(t)=
||=||=m
_
h=1
2
h
a
uD
udx
+
|l|+|s|<2m
_
b
ls
D
s
uD
l
udx
=
N
h=1
||=||=m
_
2
h
D
(
h
u)D
(
h
u)dx +
|l|+|s|<2m
_
b
ls
D
s
uD
l
udx
=
N
h=1
||=||=m
_
(
h
u)D
(
h
u)dx + O(|u|
H
m
()
|u|
H
m1
()
).
From the second case we have
|
h
u(, t)|
2
H
m
()
C
4
||=||=m
_
(
h
u)D
(
h
u)dx+C
5
|
h
u(, t)|
2
L
2
()
,
where C
4
, C
5
are positive constants. Therefore we have
(1)
m
B(u, u)(t)
N
h=1
_
C
6
|
h
u(, t)|
2
H
m
()
C
7
|
h
u|
2
L
2
()
_
C
8
|u|
H
m
()
|u|
H
m1
()
= C
6
N
h=1
_
_
||m
_
(D
h
u)
2
dx
_
_
C
7
N
h=1
_
(
h
u)
2
dxC
8
|u|
H
m
()
|u|
H
m1
()
= C
6
||m
_
_
N
h=1
_
(D
h
u)
2
dx
_
_
C
7
N
h=1
_
(
h
u)
2
dxC
8
|u|
H
m
()
|u|
H
m1
()
C
6
||m
1
N
_
_
N
h=1
D
h
u)
2
dx
_
C
7
_
[u[
2
dx C
8
|u|
H
m
()
|u|
H
m1
()
=
C
6
N
|u(, t)|
2
H
m
()
C
7
|u|
2
L
2
()
C
8
|u|
H
m
()
|u|
H
m1
()
where C
6
, C
7
, C
8
const > 0.
Using arguments as in the proof at the end of Case 2, we obtain the conclusion of
the lemma. The proof is complete.
Theorem 3.2. (Uniqueness of genrealized solution). Assume that the coefficients
of operator L satisfy conditions of lemma 3.1 and there exists a positive constant
2608 Bui Trong Kim and Nguyen Manh Hung
such that [a
[, [a
. Then
the problem (2.2)-(2.4) has at most generalized solution in H
m,1
(e
t
;
) for all
> 0.
Proof. Suppose that there exists > 0 such that the problem (2.2) -(2.4) has
two generalized u
1
and u
2
belong to H
m,1
(e
t
;
). Putting u = u
1
u
2
, we
see that u
H
m,1
(e
t
;
_
0 if b t < T,
t
_
b
u(x, )d if 0 < t b.
Applying to (2.6) and noting that
t
= u, we obtain
(3.4)
(1)
m1
_
b
(
m
||,||=1
a
(D
t
)D
+
m
||=1
a
(D
t
) + a
t
)dxdt +
_
tt
t
dxdt = 0.
Put
B
1
(u, u)(t) = B(u, u)(t) + 2Re
m
||=1
_
(D
u)udx.
From lemma 3.1 and Cauchys inequality we have
(1)
m
B
1
(u, u)(t)
1
|u|
2
H
m
()
1
|u|
2
L
2
()
,
where
1
and
1
are constants,
1
> 0. We denote by I the s s unit matrix and
define a
00
, a
1
by a
00
:= (1)
m
1
I and a
1
:= a a
00
.
From equality (3.4) we have
(1)
m1
_
b
_
_
m
||,||=0
a
(D
t
)D
+
m
||=1
a
(D
t
) + a
1
_
_
dxdt
+
_
tt
t
dxdt = 0.
Since a
= (1)
||+||
a
t
(
t
t
)dxdt+Re
_
b
_
_
m
||,||=0
(1)
||+m1
t
(a
)
_
_
dxdt
Re
_
b
_
_
m
||,||=0
(1)
||+m1
a
t
D
)
_
_
dxdt
(1)
m1
2Re
_
b
_
_
m
||=1
a
(D
t
) +a
1
_
_
dxdt = 0.
Since
_
b
a
(D
t
)dxdt =
_
t
(a
(D
))dxdt
b
a
t
(D
)dxdt
_
b
a
(D
)
t
dxdt
and
(3.6)
B
1
(, )(0) + (1)
m
1
|(, 0)|
2
L
2
()
=
m
||,||=0
(1)
||
_
[
t=0
dx+2Re
m
||=1
_
(D
)[
t=0
dx,
(3.5) implies
(3.7)
_
[
t
(x, t)[
2
dx + (1)
m
B
1
(, )(0) +
1
|(, 0)|
2
L
2
()
+ (1)
m
2Re
_
b
a
1
t
dxdt
Re
m
||,||=0
(1)
||+m1
_
b
a
t
D
dxdt+
+ (1)
m
2Re
m
||=1
_
b
(
a
t
(D
) +a
)
t
)dxdt = 0.
Put v
(x, t) =
0
_
t
D
(x, t) = v
(x, b) v
(x, t).
2610 Bui Trong Kim and Nguyen Manh Hung
Therefore we have
(1)
m
B
1
(, )(0) +
1
|(x, 0)|
2
L
2
()
1
m
||=0
_
[v
(x, b)[
2
dx.
From (3.7) we obtain
_
[
t
(x, b)[
2
dx +
1
m
||=0
_
[v
(x, b)[
2
dx
C
1
_
m
||=0
_
b
[v
(x, t)[
2
dxdt +
_
b
[
t
(x, t)[
2
dxdt
_
+C
2
b
_
0
_
m
||=0
_
[v
(x, b)[
2
dx
_
dt,
where C
1
and C
2
are positive constants. This is equivalent to
_
([
t
(x, b)[
2
+ (
1
bC
2
)
m
||=0
[v
(x, b)[
2
)dx
C
1
_
b
([
t
(x, t)[
2
+
m
||=0
[v
(x, t)[
2
)dxdt.
Putting y(t) =
_
([
t
(x, t)[
2
+
m
||=0
[v
(x, t)[
2
)dx, we obtain y(b) C
b
_
0
y(t)dt for
almost b (0,
1
2C
2
] (0,
1
C
2
). Hence from the Gronwall-Bellman inequality we
have y(b) = 0 for a.e. b (0,
1
2C
2
]. Consequently, u(x, b) = 0 for a.e. b [0,
1
2C
2
].
Using the similar arguments as the above we can prove that u(x, b) = 0 for a.e.
b [
1
2C
2
,
1
C
2
]. By the same procedure, after some steps we obtain u(x, b) = 0 for
a.e. b [0, T]. Since T is arbitrary, we obtain u
1
= u
2
. The proof of the theorem
is complete.
Theorem 3.3. (The existence of generalized solution). Suppose that f
L
(0, ; L
2
()) and hypothesis of theorem 3.2 are satisfied.
Then there exists
0
> 0 such that for all >
0
, problem (2.2)-(2.4) has at least
a generalized solution in H
m,1
(e
t
,
). Moreover
|u|
H
m,1
(e
t
,)
C|f|
L
(0,;L
2
())
,
where C is a positive constant which is independent of u and f.
Proof. We shall use the Galerkins approximate method to prove the existence
of generalized solutions.
Solvability of the First Mixed Problem 2611
Let
k
C
0
() be an orthogonal systemin L
2
() such that its linear closure
in H
m
() coincides with
H
m
(). For each natural number N, we consider the
function
u
N
(x, t) =
N
k=1
c
N
k
(t)
k
(x),
where c
N
k
(t) are the solution of the system of ordinary differential equations of
second order:
(3.8)
_
_
u
N
tt
l
+
m
||,||=0
(1)
m+||
a
u
N
D
l
_
dx +a
0
u
N
)
l
dx
+(1)
m
_
_
m
||=1
a
u
N
+a
0
u
N
_
l
dx
=
_
f
l
dx, l = 1, 2, . . . , N
with the initial conditions
(3.9) c
N
k
(0) =
d
dt
c
N
k
(0) = 0.
Here we use the notations a
00
= (1)
m
0
I, a
0
= a a
00
and
0
is the constant
as in Lemma 3.1.
Since (3.8) is a linear system with initial condition (3.9), it has unique solution
c
N
k
. Moreover,
d
2
c
N
k
dt
2
L
2
(0, T). Multiplying (3.8) by
dc
N
k
dt
, taking the sum with
respect to l and integrating the obtained equality with respect to t on (0, T] , we get
_
T
_
u
N
tt
u
N
t
+
m
||,||=0
(1)
m+||
a
u
N
D
u
N
t
_
dxdt+
+(1)
m
_
T
_
m
||=1
a
u
N
u
N
t
+ a
0
u
N
u
N
t
_
dxdt =
_
T
fu
N
t
dxdt.
Adding this equality to its complex conjugate and integrating by part we obtain
(3.10)
_
[u
N
t
(x, t)[
2
dx +
_
_
m
||,||=0
(1)
m+||
a
(x, t)D
u
N
D
u
N
_
dx
+(1)
m
2Re
_
T
_
m
||=1
a
u
N
u
N
t
+a
0
u
N
u
N
t
_
dxdt
Re
_
T
_
m
||,||=0
(1)
m+||
a
t
D
u
N
D
u
N
_
dxdt=2Re
_
T
fu
N
t
dxdt.
2612 Bui Trong Kim and Nguyen Manh Hung
By Lemma 3.1 we have
_
_
m
||,||=0
(1)
m+||
a
(x, t)D
u
N
D
u
N
_
dx
= (1)
m
B(u
N
, u
N
)(t) +
0
|u
N
(, t)|
2
L
2
()
0
|u
N
(, t)|
2
H
m
()
.
Combing this with (3.10) and using the Causchy inequality yields
(3.11)
|u
N
t
|
2
L
2
()
+
0
|u
N
|
2
H
m
()
2
_
T
_
m
||=1
[D
u
N
[[u
N
t
[ + [u
N
[[u
N
t
[
_
dxdt
+
_
T
_
m
||,||=0
[D
u
N
[[D
u
N
[
_
dxdt + 2
_
T
[f[[u
N
t
[dxdt
= 2
_
T
m
=0
1
[D
u
N
[
[u
N
t
[dxdt
+
_
T
_
m
||,||=0
[D
u
N
[[D
u
N
[
_
dxdt + 2
_
T
1
1
[f[
1
[u
N
t
[dxdt
_
T
_
m
||=0
(
1
[D
u
N
[
2
+[u
N
t
[
2
_
dxdt
+
T
_
0
|u
N
|
2
H
m
()
dt +
1
_
T
[u
N
t
[
2
dxdt +
1
1
_
T
[f[
2
dxdt
=
T
_
0
[(
1
+ 1)|u
N
|
2
H
m
()
+ (m
+
1
)|u
N
t
|
2
L
2
()
]dt
+
1
1
_
T
[f[
2
dxdt,
where m
=
m
||=0
1, > 0 and
1
> 0. This implies that
Solvability of the First Mixed Problem 2613
(3.12)
|u
N
t
|
2
L
2
()
+
0
|u
N
|
2
H
m
()
(m
+
1
)
T
_
0
_
|u
N
t
|
2
L
2
()
+
(1 +)
(m
+
1
)
|u
N
|
2
H
m
()
_
dt
+
T
1
|f|
2
L
(0,;L
2
())
.
Choosing such that
(1+)
(m
+
1
)
=
0
, we get
=
1
0
+
_
(
1
0
)
2
+ 4
2
0
m
2
0
m
.
Put
J
N
(t) = |u
N
t
(x, t)|
2
L
2
()
+
0
|u
N
(x, t)|
2
H
m
()
.
From (3.12) we have
J
N
(t) (m
+
1
)
_
T
0
J
N
(t)dt +
T
1
|f|
2
L
(0,;L
2
())
.
From this and the Gronwal- Bellman inequality, we obtain
(3.13)
|u
N
t
(x, t)|
2
L
2
()
+
0
|u
N
(x, t)|
2
H
m
()
1
exp((m
+
1
)T)|f|
2
L
(0,;L
2
())
.
If
0
1 then (3.13) implies
|u
N
t
(x, t)|
2
L
2
()
+ |u
N
(x, t)|
2
H
m
()
C
1
exp((m
+
1
)T)|f|
2
L
(0,;L
2
())
.
If 0 <
0
< 1 then we have
|u
N
t
(x, t)|
2
L
2
()
+|u
N
(x, t)|
2
H
m
()
C
1
exp((m
+
1
)T)|f|
2
L
(0,;L
2
())
.
Thus there exist a constant C > 0 such that
|u
N
t
(x, t)|
2
L
2
()
+|u
N
(x, t)|
2
H
m
()
C exp((m
+
1
)T)|f|
2
L
(0,;L
2
())
.
Put (
1
) =
m
+
1
2
. Multiplying both sides of the later inequality by e
(
1
)T
and
integrating with respect to t (0, ), we have
|u
N
|
2
H
m,1
(e
(
1
)T
,)
C|f|
2
L
(0,;L
2
())
,
2614 Bui Trong Kim and Nguyen Manh Hung
where C depends on
0
and
1
. Since
(
1
) =
_
(
1
0
)
2
+ 4
2
0
m
+
1
0
4
_
(
1
0
)
2
+ 4
2
0
m
> 0,
we have inf
>0
(
1
) =
0
:=
(1+
1+4
0
m
)
4
0
. Hence for all >
0
, there exists
1
> 0 such that > (
1
) >
0
. Consequently,
(3.14) |u
N
|
2
H
m,1
(e
t
,)
C|f|
2
L
(0,;L
2
())
,
where C depends on and
0
. Thus we have shown that u
N
is bounded in
0
H
m,1
(e
t
,
H
m,1
(e
t
;
) and u(x, 0) = 0 on , u
H
m,1
(e
t
;
).
It remains to show that u is a generalized solution of the problem. In fact, we
define the set M
N
by
M
N
= =
N
l=1
d
l
l
: d
l
H
1
(0, T), d
l
(T) = 0.
Taking any M
N
, we have =
N
l=1
d
l
l
. By multiplying (3.8) by d
l
, taking the
sum with respect to l and integrating in t (0, T), we obtain
(3.15)
_
T
u
N
tt
dxdt + (1)
m
_
T
_
m
||,||=1
(1)
||
a
u
N
D
+
m
||=1
a
u
N
+au
N
_
dxdt =
_
T
fdxdt.
Integrating by part with respective to t in the first term of (3.15) yields
(1)
m1
_
T
(
m
||,||=1
(1)
||
a
u
N
D
+
m
||=1
a
u
N
+au
N
)dxdt
+
_
T
u
N
t
t
)dxdt =
_
T
fdxdt.
By letting N , we obtain
(3.16)
(1)
m1
_
T
(
m
||,||=1
(1)
||
a
uD
+
m
||=1
a
u + au)dxdt
+
_
T
u
t
t
)dxdt =
_
T
fdxdt.
Solvability of the First Mixed Problem 2615
It is noted that the set M :=
N=1
M
N
is dense in the space of functions
H
m,1
(
T
), (x, T) = 0. Therefore (3.16) is valid for all
H
m,1
(
T
) satisfying
(x, T) = 0. This implies that u is a generalized solution of (2.2)-(2.4). Moreover,
from (3.14) we have
|u|
2
H
m,1
(e
t
,)
liminf
N
|u
N
|
2
H
m,1
(e
t
,)
C|f|
2
L
(0,;L
2
())
.
The proof is complete.
4. APPLICATIONS
In this section we will apply the previous results to the study of generalized
solution existence of elasticity problems.
Let us assume that u and f be real vector functions which defined on R
n
.
Consider the differential operators:
(4.1) L
s
(x, t, D)(u) =
n
j,h,k=1
x
h
(a
jk
sh
(x, t)
u
j
x
k
), s = 1, ..., n
where a
jh
sh
are continuous real functions on satisfying condition
(4.2) a
jk
sh
= a
jk
hs
= a
sh
jk
.
Put
(4.3) e
sh
=
1
2
(
u
s
x
h
+
u
h
x
s
)
and
(4.4) W(x, t, e) =
n
s,h,j,k=1
a
jk
sh
e
sh
e
jk
.
Assume that W(x, t, e) is an elastic potential which is a positive definite quadratic
form with respect to e
sh
, 1 s h n for every (x, t)
.
According to [4], for any u(x, t) satisfying u(, t) H
1
() for t (0, ) we
have
n
s,h=1
_
(
u
s
x
h
+
u
h
x
s
)
2
dx C|u(x, t)|
2
H
1
()
,
where C is a positive constant which is independent of u and t. From this inequality
we obtain
2616 Bui Trong Kim and Nguyen Manh Hung
(4.5)
n
s,h,j,k=1
_
a
jk
sh
u
s
x
h
u
j
x
k
C|u(x, t)|
2
H
1
()
,
where C is a positive constant which is independent of u and t. It follows that the
elastic potential satisfies the inequality (3.1) with
0
= 0.
We now consider the following problem in
.
(4.6) L
s
(x, t, D)u u
tt
= f(x, t), s = 1, 2, ..., n
with the initial conditions
(4.7) u [
t=0
= u
t
[
t=0
= 0
and boundary condition
(4.8) u [
S
= 0.
From Theorem 3.1, Theorem 3.2 and (4.5) we obtain the following result.
Theorem 4.1. Assume that W(x, t, e) is a positive definite quadratic form
with respect to the variable e
sh
, 1 s h n, for all (x, t)
. Assume
further the following conditions:
(i) [a
jk
sh
/t[
2
, (x, t)
(ii) f L
(0, ; L
2
()).
Then there exists
0
> 0 such that for all >
0
problem (4.6)-(4.8) has a
unique generalized solution u(x, t) in the space H
1,1
(e
t
,
). Moreover
|u|
H
1,1
(e
t
,)
C|f|
L
(0,;L
2
())
where C is independent of u and f.
Now we study the differential operator of the Lames type
+ ( +)grad(div)
where is the Laplace operator, > 0, and + > 0. Consider the following
problem in
.
(4.9) u + ( +)grad(divu) u
tt
= f(x, t)
(4.10) u [
t=0
= u
t
[
t=0
= 0
Solvability of the First Mixed Problem 2617
(4.11) u [
S
= 0.
Put a
ss
ss
= + , a
hs
sh
= a
sh
hs
= with s ,= h, a
j
j
ss
= 0 with s ,= j and a
hs
jk
= 0
with jk ,= sh. Hence we have
(4.12)
2W(x, t, e) =
n
s,h,j,k=1
a
jk
sh
e
sh
e
jk
= (2 +)
n
s=1
e
2
ss
+
1,n
h=s
e
ss
e
hh
+
1,n
h=s
e
2
hs
= 2
n
s=1
e
2
ss
+(
n
s=1
e
ss
)
2
+
1,n
h=s
e
2
hs
.
Since > 0 and + > 0 we can find > 0 such that > 0 and + > 0.
Hence (4.12) implies
(4.13)
2W(x, t, e) = 2
n
s=1
e
2
ss
+ 2( )
n
s=1
e
2
ss
+(
n
s=1
e
ss
)
2
+
1,n
h=s
e
2
hs
2
n
s=1
e
2
ss
+
1,n
h=s
e
2
hs
+
2
n
( )(
n
s=1
e
ss
)
2
+(
n
s=1
e
ss
)
2
= 2
n
s=1
e
2
ss
+
1,n
h=s
e
2
hs
+ (
2
n
( ) +)(
n
s=1
e
ss
)
2
.
This implies that b
(4.14) W(x, t, e)
0
n
s,h=1
e
2
sh
where
0
is a positive constant. Therefore system (4.9) satisfies conditions of
Theorem 3.1 and Theorem 3.2, where we can choose
2
= 0. From Theorem 3.1
and Theorem 3.2 we obtain
Theorem 4.2. Let f be a function which belongs to L
(0, ; L
2
()). Then
for each > 0 problem (4.9)-(4.11) has a unique generalized solution u = u(x, t)
in H
1,1
(e
t
,
). Moreover,
|u|
H
1,1
(e
t
,)
C|f|
L
(0,;L
2
())
,
where C is a positive constant which is independent of u and t.
2618 Bui Trong Kim and Nguyen Manh Hung
REFERENCES
1. R. A. Adams and J. J. F. Fourier, Sobolev Space, Academic Press, 2003.
2. S. Agmons, A. Douglis and L. Nirenberg, Estimates near the boundary, Comm. Pure
Appl. Math., 12(N
4) (1959), 623-727.
3. N. M. Hung, Asymtotic behaviour of solutions of the first boundary value problem
for strongly hyperbolic systems near a conical point at the boundary of the domain,
Sbornik: Mathematics, 190 (1999), 1035-1058.
4. V. A. Kondratiev, Boundary value problems for elliptic equation in domains with con-
ical or angned points. Transactions of Moscow Mathematical Society, (in Russian),
16 (1967), 209-292.
5. V. A. Kozlov, V. G. Mazya and J. Rossman, Elliptic boundary value problem in
domains with point singularities, Math. Surveys Monographs, 52 (1997), AMS.
6. R. Dautray and J.-L. Lion, Mthematical Analysis and Numerical Methods for Science
and Tecnology, Vol. 5, Springer, 1992.
7. V. G. Mazia and B. A. Plamenevsky. On the coefficients in the asymptotic of solutions
to the elliptic boundary value problems in domains with conical points. Math. Nachr.
(in Russian), 76 (1977), 29-60.
8. G. Fichera. Existense theorems in elasticity, Springer, New York-Berlin, 1972.
B. T. Kim
Department of Mathematics,
HaNam College of Education,
Vietnam
E-mail: buitrongkim@gmail.com
N. M. Hung
Department of Mathematics,
HaNoi National University of Education,
Vietnam
E-mail: hungnmmath@hnue.edu.vn