Correction to \PARTIAL EXACT CONTROLLABILITY
ANDEXPONENTIAL STABILITY IN
HIGHER-DIMENSIONAL LINEAR THERMOELASTICITY"
WEIJIU LIU
In 1], as a consequence of (3.16), we were assuming that
k
v
0(t
)k20e
1;w TE
(vT
)in lines 5 and 6 on page 41. But, the correct inequality is
k
v
0(t
)k20 2e
1;w (T;t)E
(vT
):
Consequently, the proof of Theorem 2.3 is not correct. We give here a corrected proof that needs however an additional smallness condition on
.We acknowledge Aissa Guesmia for pointing out this mistake to us.
We now give a corrected and weaker version of Theorem 2.3 with a com- plete proof. We adopt the notation and numbering of formulas, Theorems and Denitions from 1].
THEOREM 2.3. Let ;1 and ;2 be given by (1.4) and (1.5), respectively, satisfying (2.12). Assume that the function
a
(x
) satises (2.19) or (2.20).Suppose that
< !
2e:
Let
T
0 be large enough so that2
e
2(1;! T0)<
1; 4222e
2!
2 (2:
33) andT
T
0, where!
is the constant in Theorem 2.1. Then for any(u
0u
1)2W
, there exists a boundary control function (xt
) withp
m
2(L
2(2))nsuch that the solution of (2.30) satises (2.31). Moreover, there exists a positive constant
c
, independent of (u
0u
1), such thatk p
m
k(L2(2))nc
k(u
0u
1)kW:
(2:
34)Dept of Applied Mechanics and Engineering Sciences, Univ. of California San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0411, USA. E-mail: [email protected]
Received April 29, 1998. Accepted for publication May 27, 1998.
c
Societe de Mathematiques Appliquees et Industrielles. Typeset by TEX.
Proof of Theorem 2.3. The main idea of the proof is rst to construct a linear operator by using the Lame system and the system of thermoelasticity, and then to show that
I
; is an isomorphism by using the uniform stabilization of these systems.Given (
v
0v
1)2W
, we consider the Lame system8
>
>
>
<
>
>
>
:
v
00;v
;(+)rdivv
= 0 inQ
v
= 0 on 1@v @
+ (+)div(v
)+am
v
;m
v
0 = 0 on 2v
(T
) =v
0v
0(T
) =v
1 in(3
:
15)which has a unique solution with
(
v
(t
)v
0(t
))2C
(0T
]W
):
Moreover, by Theorem 2.2 (note that we are assuming that
a
(x
) satises (2.19) or (2.20), and in this case, Theorem 2.2 on the stabilization of the Lame system holds), there exists a positive constant!
such thatE
(vt
)E
(vT
)e
1;! (T;t) 8t
20T
] (3:
16) whereE
(vt
) = 12Zh
j
v
0(xt
)j2+jrv
(xt
)j2+ (+)jdivv
(xt
)j2idx
+ 12Z
;
2
am
jv
(t
)j2d
;:
Using the solution
v
of (3.15), we then consider8
<
:
0; =div(v
0) inQ
= 0 on (0) = 0 in (3:
17)and
8
>
>
>
>
>
>
>
<
>
>
>
>
>
>
>
:
w
00;w
;(+)rdivw
+r=;r inQ
0;+divw
0= 0 inQ
= 0 onw
= 0 on 1@w @
+ (+)div(w
)+am
w
+m
w
0 = 0 on 2w
(0) =v
(0)w
0(0) =v
0(0) (0) = 0 in:
(3
:
18)Since
div(
v
0)2(L
2(0T H
;1()))n it follows that (3.17) has a unique solution with 2C
(0T
]L
2())\L
2(0T H
01()):
In addition, multiplying (3.17) by
and integrating overQ
, we obtain 12k(T
)k20+Z
T
0
kr
(t
)k20dt
=
Z T0 Z
div(
v
0)dxdt
Z T0
kdiv(
v
0(t
))k;1kr(t
)k0dt
(use (2:
32))
12
Z
T
0
kr
(t
)k20dt
+ 22 2Z
T
0
k
v
0(t
)k20dt
(use (3:
16))
12
Z
T
0
kr
(t
)k20dt
+22E
(vT
)Z
T
0
e
1;! (T;t)dt
12
Z
T
0
kr
(t
)k20dt
+ 22e
! E
(vT
) which impliesk
(T
)k20+Z
T
0
kr
(t
)k20dt
222e
! E
(vT
):
(3:
19) On the other hand, since 2L
2(0T H
01()) andZ
r
dx
=Z
;
d
; = 0then (0
;r0) 2L
1(0T
H). Thus, by the classical theory of semi- groups, the nonhomogeneous problem (3.18) has a unique solution with(
ww
0)2C
(0T
]H):
Moreover, the solution can be expressed as(
ww
0) =S
(t
)(w
(0)w
0(0)(0))+Z
t
0
S
(t
;)(0;r0)d
whereS
(t
) denotes the strongly continuous semigroup of contractions gen- erated by the thermoelastic system. By Theorem 2.1, we havek
S
(t
)ke
(1;! t)=2 8t
0 then we deduce from (3.19)E
(wt
)2kS
(t
)k2E
(w
0)+ 2Z
t
0
k
S
(t
;)kkr()k0d
22
e
1;! tE
(w
0)+ 22Z t0
e
1;! (t;)d
Z t0
kr
()k20d
2
e
1;! tE
(w
0)+ 4222e
2!
2E
(vT
):
(3
:
20)Set
u
=w
;v
=+and =;
m
(w
0+v
0) (3:
21)then
u
satises8
>
>
>
>
>
>
>
>
>
<
>
>
>
>
>
>
>
>
>
:
u
00 ;u
;(+)rdivu
+r= 0 inQ
0;+divu
0= 0 inQ
= 0 onu
= 0 on 1@u @
+ (+)div(u
)+am
u
= on 2u
(0) =u
0(0) = 0 (0) = 0 inu
(T
) =w
(T
);v
0u
0(T
) =w
0(T
);v
1 in:
We dene an operator by(
v
0v
1) = (w
(T
)w
0(T
)):
Then it is clear that is a linear operator from
W
intoW
. Moreover, by (3.16) and (3.20), we havek(
v
0v
1)k2W(note denition (2
:
1) of the norm ofW
)E
(wT
)2
e
1;! TE
(w
0)+ 4222e
2!
2E
(vT
) (sincew
(0) =v
(0)w
0(0) =v
0(0) (0) = 0)= 2
e
1;! TE
(v
0)+ 4222e
2!
2E
(vT
)
h2
e
2(1;! T)+ 4222e
2!
2i
E
(vT
) (use (3:
16))=h2
e
2(1;! T)+ 4222e
2!
2i
k(
v
0v
1)k2W:
Therefore,kk22
e
2(1;! T)+ 4222e
2!
2:
Let
T
0 be large enough so that (2.33) holds ifT
T
0. Then ;I
is an isomorphism fromW
ontoW
. Thus, for any (u
0u
1) 2W
, there exists a unique (v
0v
1)2W
such that(
u
0u
1) = (v
0v
1);(v
0v
1)= (
w
(T
)w
0(T
));(v
0v
1)= (
u
(T
)u
0(T
)):
(3:
22)Consequently, we have constructed a control function =;
m
(w
0+v
0) solving the partial controllability problem (2.30).On the other hand, multiplying the rst equation of (3.15) by
v
i0 and integrating overQ
, we obtainZ
2
m
jv
0j2d
=E
(vT
);E
(v
0):
(3:
23) Multiplying the rst equation and second equation of (3.18) byw
i0 and , respectively, and integrating overQ
, we deduce from (3.19) and (3.20) that (the followingc
's denoting various constants that may depend onT
)E
(wT
)+Z
Q
jr
j2dxdt
+Z
2
m
jw
0j2d
=
E
(w
0);ZQ
w
0rdxdt
E
(w
0)+cE
(vT
)+Z
T
0
E
(w
)d
cE
(w
0)+cE
(vT
):
(3
:
24)Noting
E
(v
0) =E
(w
0), we deduce from (3.16), (3.23) and (3.24) thatZ
2 j j
2
m
d
=Z
2
m
jv
0+w
0j2d
cE
(vT
)which, combined with (3.22), implies (2.34) since ;
I
is an isomorphism fromW
onto itself.References
1] W.J. Liu: Partial exact controllability and exponential stability in higher dimensional linear thermoelasticity, ESAIM: Control Optim. Calc. Var.,3, 1998, 23{48.