Hindawi Publishing Corporation

Journal of Applied Mathematics

Volume 2014, Article ID 827698, 6 pages
http://dx.doi.org/10.1155/2014/827698

Research Article
Controllability for a Wave Equation with Moving Boundary

Lizhi Cui1,2 and Libo Song3

1
College of Applied Mathematics, Jilin University of Finance and Economics, Changchun 130117, China
2
School of Mathematics and Statistics, Northeast Normal University, Changchun 130024, China
3
Educational Administration, Jilin University of Finance and Economics, Changchun 130117, China

Correspondence should be addressed to Lizhi Cui; cuilz829@nenu.edu.cn

Received 8 November 2013; Accepted 15 February 2014; Published 27 February 2014

Academic Editor: Laurent Gosse

Copyright © 2014 L. Cui and L. Song. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We investigate the controllability for a one-dimensional wave equation in domains with moving boundary. This model characterizes
small vibrations of a stretched elastic string when one of the two endpoints varies. When the speed of the moving endpoint is less
than 1 − 1/√𝑒, by Hilbert uniqueness method, sidewise energy estimates method, and multiplier method, we get partial Dirichlet
boundary controllability. Moreover, we will give a sharper estimate on controllability time that only depends on the speed of the
moving endpoint.

1. Introduction and Main Results The exact controllability problem for (3) is formulated
as follows: given 𝑇 > 0 large enough. For each (𝑢0 , 𝑢1 ) ∈
This paper concerns a finite vibrating string described by a 𝐿2 (0, 1) × 𝐻−1 (0, 1) and for each (𝑢𝑑0 , 𝑢𝑑1 ) ∈ 𝐿2 (0, 𝛼𝑘 (𝑇)) ×
wave equation. The left boundary endpoint of the string is
𝐻−1 (0, 𝛼𝑘 (𝑇)), is it possible to find a control V ∈ 𝐿2 (0, 𝑇) such
fixed, while the right boundary endpoint is moving. Given
̂𝑘 for the following noncylindrical domain: that the corresponding solution of (3) satisfies
that 𝑇 > 0, write 𝑄𝑇
𝑢 (𝑇) = 𝑢𝑑0 , 𝑢𝑡 (𝑇) = 𝑢𝑑1 ? (4)
{(𝑦, 𝑡) ∈ R2 ; 0 < 𝑦 < 𝛼𝑘 (𝑡) , ∀𝑡 ∈ (0, 𝑇)} , (1)
The main purpose of this paper is to study the exact
where controllability of (3). As we all know, there exist pieces of
𝛼𝑘 (𝑡) = 1 + 𝑘𝑡, 0 < 𝑘 < 1. (2) literature on the controllability problems of wave equations
in a cylindrical domain. However, as far as we know, there
Consider the following wave equation in the noncylindri- are only a few works on the exact controllability for wave
̂𝑘 :
cal domain 𝑄 equations defined in noncylindrical domains. We refer to [1–
𝑇
3] for some known results in this respect. In [1], the boundary
̂𝑘 ,
𝑢𝑡𝑡 − 𝑢𝑦𝑦 = 0 in 𝑄 controllability problem for a multidimensional wave equation
𝑇
with constant coefficients in a noncylindrical domain was
𝑢 (0, 𝑡) = V (𝑡) , 𝑢 (𝛼𝑘 (𝑡) , 𝑡) = 0 on (0, 𝑇) , (3) discussed. However, in [1] in the one-dimensional case, the
following condition seems necessary:
𝑢 (0) = 𝑢0 , 𝑢𝑡 (0) = 𝑢1 in (0, 1) ,
∞
󵄨 󵄨
where V is the control variable and is put on fixed endpoint. ∫ 󵄨󵄨󵄨󵄨𝛼𝑘󸀠 (𝑡)󵄨󵄨󵄨󵄨 𝑑𝑡 < ∞. (5)
The constant 𝑘 is called to be the speed of the moving 0

endpoint. By [1], for 0 < 𝑘 < 1, any (𝑢0 , 𝑢1 ) ∈ 𝐿2 (0, 1) × It is easy to check that this condition is not satisfied for the
𝐻−1 (0, 1), and V ∈ 𝐿2 (0, 𝑇), (3) admits a unique transposition moving boundary in (3). In [2], the exact controllability of
solution. a multidimensional wave equation with constant coefficients
2 Journal of Applied Mathematics

in a noncylindrical domain was established, while a con- Theorem 4. Suppose that 0 < 𝑘 < 1 − 1/√𝑒. Let 𝑇 > 𝑇𝑘∗ .
trol entered the system through the whole noncylindrical Then, for any initial value (𝑤0 , 𝑤1 ) ∈ 𝐿2 (0, 1) × 𝐻−1 (0, 1) and
domain. In [3], the exact controllability of the following target (𝑤𝑑0 , 𝑤𝑑1 ) ∈ 𝐿2 (0, 1) × 𝐻−1 (0, 1), there exists a control
system is studied: V ∈ 𝐿2 (0, 𝑇) such that the corresponding solution 𝑤 of (8) in
̂𝑘 , the sense of transposition satisfies
𝑢𝑡𝑡 − 𝑢𝑦𝑦 = 0 in 𝑄𝑇
𝑤 (𝑇) = 𝑤𝑑0 , 𝑤𝑡 (𝑇) = 𝑤𝑑1 . (11)
𝑢 (0, 𝑡) = 0, 𝑢 (𝛼𝑘 (𝑡) , 𝑡) = V (𝑡) on (0, 𝑇) , (6)
In order to obtain Theorem 4, we will use Hilbert
𝑢 (0) = 𝑢0 , 𝑢𝑡 (0) = 𝑢1 in (0, 1) . uniqueness method. The main idea is to define a weighted
energy function for the following wave equation with variable
But the control is put on moving endpoint. In order to over- coefficients in cylindrical domains:
come these difficulties and drop the additional conditions
for the moving boundary, we use sidewise energy estimates 𝛼𝑘 (𝑡) 𝑧𝑡𝑡 − [𝛽𝑘 (𝑥, 𝑡) 𝑧𝑥 ]𝑥 + 𝛾𝑘 (𝑥) 𝑧𝑡𝑥 = 0 in 𝑄,
method to obtain observability inequality.
𝑧 (0, 𝑡) = 0, 𝑧 (1, 𝑡) = 0 on (0, 𝑇) , (12)
The main result of this paper is stated as follows.

Theorem 1. Suppose that 0 < 𝑘 < 1 − 1/√𝑒. For any given 𝑧 (0) = 𝑧0 , 𝑧𝑡 (0) = 𝑧1 in (0, 1) ,
𝑇 > 𝑇𝑘∗ , (3) is exactly controllable at time 𝑇. where 𝑘 ∈ (0, 1), (𝑧 , 𝑧 ) ∈ 0 1
𝐻01 (0, 1)
× 𝐿2 (0, 1) is any given
initial value, and 𝛼𝑘 , 𝛽𝑘 , and 𝛾𝑘 are the functions given in (8).
Remark 2. 𝑇𝑘∗ will be defined during the course of the proof. Similar to Theorem 3.2 in [4], we have that (12) has a unique
weak solution
Remark 3. It seems natural to expect that the exact control-
lability of (3) holds when 𝑘 ∈ (0, 1). However, we do not 𝑧 ∈ 𝐶 ([0, 𝑇] ; 𝐻01 (0, 1)) ∩ 𝐶1 ([0, 𝑇] ; 𝐿2 (0, 1)) . (13)
succeed in extending the approach developed in Theorem 1
to this case. In the sequel, we denote by 𝐶 a positive constant depend-
ing only on 𝑇 and 𝑘, which may be different from one place
In order to prove Theorem 1, we first transform (3) into to another.
a wave equation with variable coefficients in a cylindrical The energy function of system (12) is defined as
domain. For this aim, set
1 1 󵄨 󵄨2 󵄨 󵄨2
𝑦 𝐸 (𝑡) = ∫ [𝛼𝑘 (𝑡) 󵄨󵄨󵄨󵄨𝑧󸀠 (𝑥, 𝑡)󵄨󵄨󵄨󵄨 + 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑥
𝑥= , 𝑤 (𝑥, 𝑡) = 𝑢 (𝑦, 𝑡) = 𝑢 (𝛼𝑘 (𝑡) 𝑥, 𝑡) 2 0
𝛼𝑘 (𝑡)
(7) (𝑡 ≥ 0) .
̂ .
for (𝑦, 𝑡) ∈ 𝑄 𝑘
(14)
𝑇

Then, it is easy to check that (𝑥, 𝑡) varies in 𝑄 := (0, 1)×(0, 𝑇). In particular
Also, (3) is transformed into the following equivalent wave 1 1 󵄨󵄨 1 󵄨󵄨2 󵄨 󵄨2
equation in the cylindrical domain 𝑄: 𝐸0 = 𝐸 (0) = ∫ [󵄨󵄨𝑧 󵄨󵄨 + 𝛽𝑘 (𝑥, 0) 󵄨󵄨󵄨󵄨𝑧𝑥0 󵄨󵄨󵄨󵄨 ] 𝑑𝑥. (15)
2 0 󵄨 󵄨
𝛽𝑘 (𝑥, 𝑡) 𝛾 (𝑥)
𝑤𝑡𝑡 − [ 𝑤𝑥 ] + 𝑘 𝑤 = 0 in 𝑄, Note that this weighted energy is different from the usual
𝛼𝑘 (𝑡) 𝑥 𝛼𝑘 (𝑡) 𝑡𝑥 one, but they are equivalent. We will obtain explicit energy
(8) equality. Using this energy equality, we will first prove the
𝑤 (0, 𝑡) = V (𝑡) , 𝑤 (1, 𝑡) = 0 on (0, 𝑇) , following observability result.
𝑤 (0) = 𝑤0 , 𝑤𝑡 (0) = 𝑤1 in (0, 1) , Theorem 5. Let 𝑇 > 𝑇𝑘∗ . For any (𝑧0 , 𝑧1 ) ∈ 𝐻01 (0, 1)×𝐿2 (0, 1),
where there exists a constant 𝐶 > 0 such that the corresponding
solution 𝑧 of (12) satisfies
1 − 𝑘2 𝑥2
𝛽𝑘 (𝑥, 𝑡) = , 𝛾𝑘 (𝑥) = −2𝑘𝑥, 󵄨 󵄨2 󵄨 󵄨2
𝛼𝑘 (𝑡) (9) 𝐶 (󵄨󵄨󵄨󵄨𝑧0 󵄨󵄨󵄨󵄨𝐻1 (0,1) + 󵄨󵄨󵄨󵄨𝑧1 󵄨󵄨󵄨󵄨𝐿2 (0,1) )
0

0 0 1 1
𝑤 =𝑢 , 𝑤 =𝑢 + 𝑘𝑥𝑢𝑥0 . 𝑇
󵄨 󵄨2
≤ ∫ 𝛽𝑘 (0, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (0, 𝑡)󵄨󵄨󵄨 𝑑𝑡 (16)
Equation (8) admits a unique solution in the sense of 0
transposition 󵄨 󵄨2 󵄨 󵄨2
≤ 𝐶 (󵄨󵄨󵄨󵄨𝑧0 󵄨󵄨󵄨󵄨𝐻1 (0,1) + 󵄨󵄨󵄨󵄨𝑧1 󵄨󵄨󵄨󵄨𝐿2 (0,1) ) .
𝑤 ∈ 𝐶 ([0, 𝑇] ; 𝐿2 (0, 1)) ∩ 𝐶1 ([0, 𝑇] ; 𝐻−1 (0, 1))
0
(10)
Then, applying Hilbert uniqueness method, we will
(see [4]). deduce Theorem 4.
Therefore, the exact controllability of (3) (Theorem 1) is The rest of this paper is organized as follows. In Section 2,
reduced to the following controllability result for the wave we derive Theorem 5. Section 3 is devoted to a proof of
equation (8). Theorem 4.
Journal of Applied Mathematics 3

2. Observability: Proof of Theorem 5 where

𝑇−𝐴 1 (𝑥)
In this section, first we give the following two lemmas (see the ∫ 𝛼𝑘 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑡,𝑥 (𝑥, 𝑡) 𝑑𝑡
detailed proof in [3]). 𝐴 1 (𝑥)

󵄨𝑇−𝐴 (𝑥)
Lemma 6. For any (𝑧0 , 𝑧1 ) ∈ 𝐻01 (0, 1) × 𝐿2 (0, 1) and 𝑡 ∈ = [𝛼𝑘 (𝑡)𝑧𝑡 (𝑥, 𝑡)𝑧𝑥 (𝑥, 𝑡)]󵄨󵄨󵄨𝐴 1 (𝑥)1
[0, 𝑇], any solution 𝑧 of (12) satisfies the following estimate: (22)
𝑇−𝐴 1 (𝑥)
1 −∫ [𝛼𝑘,𝑡 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
𝐸 (𝑡) = 𝐸. (17)
𝛼𝑘 (𝑡) 0 𝐴 1 (𝑥)

Lemma 7. Suppose that 𝑞 ∈ 𝐶1 ([0, 1]) is any given function. + 𝛼𝑘 (𝑡) 𝑧𝑡𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)] 𝑑𝑡.
Then any solution 𝑧 of (12) satisfies the following estimate:
By (12), it follows that
󵄨1
1 𝑇 󵄨 󵄨2 󵄨󵄨
[ ∫ 𝛽𝑘 (𝑥, 𝑡)𝑞(𝑥)󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑡]󵄨󵄨󵄨󵄨 𝛼𝑘 (𝑡) 𝑧𝑡𝑡 (𝑥, 𝑡)
2 0 󵄨󵄨0
= [𝛽𝑘 (𝑥, 𝑡)𝑧𝑥 ]𝑥 (𝑥, 𝑡) − 𝛾𝑘 (𝑥) 𝑧𝑡𝑥 (𝑥, 𝑡)
1 𝑇 1 󵄨 󵄨2 (23)
= ∫ ∫ [𝑞𝑥 (𝑥) 𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 = 𝛽𝑘,𝑥 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡) + 𝛽𝑘 (𝑥, 𝑡) 𝑧𝑥𝑥 (𝑥, 𝑡)
2 0 0
+ (𝑞𝑥 (𝑥) 𝛽𝑘 (𝑥, 𝑡) − 𝛽𝑘,𝑥 (𝑥, 𝑡) 𝑞 (𝑥)) − 𝛾𝑘 (𝑥) 𝑧𝑡𝑥 (𝑥, 𝑡) ,
󵄨 󵄨2 from which and using integrating by parts, we have that
× 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑥 𝑑𝑡 (18)
𝑇−𝐴 1 (𝑥)
𝑇 1 ∫ 𝛼𝑘 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑡,𝑥 (𝑥, 𝑡) 𝑑𝑡
− ∫ ∫ 𝛼𝑘,𝑡 (𝑡) 𝑞 (𝑥) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡) 𝑑𝑥 𝑑𝑡 𝐴 1 (𝑥)
0 0
󵄨𝑇−𝐴 (𝑥)
1 = [𝛼𝑘 (𝑡)𝑧𝑡 (𝑥, 𝑡)𝑧𝑥 (𝑥, 𝑡)]󵄨󵄨󵄨𝐴 1 (𝑥)1
+ ∫ [𝛼𝑘 (𝑡) 𝑞 (𝑥) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
0 𝑇−𝐴 1 (𝑥)

󵄨󵄨𝑇 −∫ [𝛼𝑘,𝑡 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)

1 󵄨 󵄨2
+ 𝛾𝑘 (𝑥)𝑞(𝑥)󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑥󵄨󵄨󵄨󵄨 .
𝐴 1 (𝑥)
2 󵄨0 󵄨 󵄨2
+ 𝛽𝑘,𝑥 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨
Now, we give a proof of Theorem 5.
+ 𝛽𝑘 (𝑥, 𝑡) 𝑧𝑥𝑥 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
Proof. We first give the proof of the second inequality in (16).
Define − 𝛾𝑘 (𝑥) 𝑧𝑡𝑥 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)] 𝑑𝑡
𝑇−𝐴 1 (𝑥)
1 󵄨 󵄨2 󵄨𝑇−𝐴 (𝑥)
𝐺 (𝑥) = ∫ [𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 = [𝛼𝑘 (𝑡)𝑧𝑡 (𝑥, 𝑡)𝑧𝑥 (𝑥, 𝑡)]󵄨󵄨󵄨𝐴 1 (𝑥)1
2 𝐴 1 (𝑥) (19)
𝑇−𝐴 1 (𝑥)
󵄨 󵄨2
+ 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑡. −∫ [𝛼𝑘,𝑡 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
𝐴 1 (𝑥)
Note that 󵄨 󵄨2
+ 𝛽𝑘,𝑥 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨
1 𝑇−𝐴 1 (𝑥) 󵄨 󵄨2
𝐺 (0) = ∫ 𝛽𝑘 (0, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (0, 𝑡)󵄨󵄨󵄨 𝑑𝑡. (20)
2 𝐴 1 (𝑥) + 𝛽𝑘 (𝑥, 𝑡) 𝑧𝑥𝑥 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)] 𝑑𝑡
The derivative of the functional of 𝐺 is
󵄨2 󵄨󵄨 1
𝑇−𝐴 (𝑥)
1 󵄨
𝐺󸀠 (𝑥) + [ 𝛾𝑘 (𝑥)󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ]󵄨󵄨󵄨󵄨
2 󵄨𝐴 1 (𝑥)
𝑇−𝐴 1 (𝑥)
=∫ [𝛼𝑘 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑡,𝑥 (𝑥, 𝑡) = [𝛼𝑘 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
𝐴 1 (𝑥)

󵄨2 󵄨󵄨 1
𝑇−𝐴 (𝑥)
+ 𝛽𝑘 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡) 𝑧𝑥𝑥 (𝑥, 𝑡) 1 󵄨
+ 𝛾𝑘 (𝑥)󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ]󵄨󵄨󵄨󵄨
2 󵄨𝐴 1 (𝑥)
1 󵄨 󵄨2
+ 𝛽 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑡
2 𝑘,𝑥 𝑇−𝐴 1 (𝑥)
−∫ [𝛼𝑘,𝑡 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
1 𝐴 1 (𝑥)
+ (− 𝐴󸀠1 (𝑥))
2 󵄨 󵄨2
+ 𝛽𝑘,𝑥 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨
󵄨 󵄨2 󵄨 󵄨2
× ∑ [𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 + 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] ,
𝑡=𝑇−𝐴 1 (𝑥),𝐴 1 (𝑥) + 𝛽𝑘 (𝑥, 𝑡) 𝑧𝑥𝑥 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)] 𝑑𝑡.
(21) (24)
4 Journal of Applied Mathematics

We conclude, using (24), that 1 𝑇−𝐴 1 (𝑥) 󵄨 󵄨2

≤ ∫ 𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑡
2𝜀 𝐴 1 (𝑥)
𝐺󸀠 (𝑥) 𝜀 𝑇−𝐴 1 (𝑥) 𝑘2 󵄨 󵄨2
+ ∫ 𝛽 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑡
𝑇−𝐴 1 (𝑥) 2 𝐴 1 (𝑥) 1 − 𝑘2 𝑥2 𝑘
= −∫ [𝛼𝑘,𝑡 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
𝐴 1 (𝑥)
𝑇−𝐴 1 (𝑥)
𝑘2 𝑥 󵄨 󵄨2
+∫ 𝛽 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑡
1 𝐴 1 (𝑥) 1 − 𝑘2 𝑥2 𝑘
󵄨 󵄨2
+ 𝛽𝑘,𝑥 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑡
2 1 1 𝑇−𝐴 1 (𝑥) 󵄨 󵄨2
≤ ∫ 𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑡
2 𝜀 𝐴 1 (𝑥)
+ [𝛼𝑘 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
1 𝑘2 (𝜀 + 2) 𝑇−𝐴 1 (𝑥) 󵄨 󵄨2
1 󵄨 󵄨2 󵄨󵄨 1
𝑇−𝐴 (𝑥) + ∫ 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑡.
+ 𝛾𝑘 (𝑥)󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ]󵄨󵄨󵄨󵄨 2 1 − 𝑘2 𝐴 1 (𝑥)
2 󵄨𝐴 1 (𝑥) (27)
1
+ (− 𝐴󸀠1 (𝑥))
2 Take 𝜀 = (1 − 𝑘)/𝑘; then it is easy to check that
󵄨 󵄨󵄨2 󵄨 󵄨󵄨2
× ∑ [𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨 + 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨 ] .
𝑡=𝑇−𝐴 1 (𝑥),𝐴 1 (𝑥) 1 𝑘2 (𝜀 + 2)
= . (28)
(25) 𝜀 1 − 𝑘2

Hence
We will choose 𝐴 1 (𝑥) later which satisfies
𝑘
𝐺󸀠 (𝑥) ≤ 𝐺 (𝑥) 𝑥 ∈ (0, 1) . (29)
󵄨2 󵄨󵄨 1
𝑇−𝐴 (𝑥)
1 󵄨 1−𝑘
[𝛼𝑘 (𝑡)𝑧𝑡 (𝑥, 𝑡)𝑧𝑥 (𝑥, 𝑡) + 𝛾𝑘 (𝑥)󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ]󵄨󵄨󵄨󵄨
2 󵄨𝐴 1 (𝑥)
By Gronwall inequality, there exists 𝐶 > 0 such that
1
+ (− 𝐴󸀠1 (𝑥))
2
𝐺 (𝑥) ≤ 𝐶𝐺 (0) . (30)
󵄨 󵄨2 󵄨 󵄨2
× ∑ [𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 + 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ]
𝑡=𝑇−𝐴 1 (𝑥),𝐴 1 (𝑥)
Integrating (30) in [0, 1], we have
≤ 0.
(26) 1
∫ 𝐺 (𝑥) 𝑑𝑥 ≤ 𝐶𝐺 (0) . (31)
0
From (25) and (26), for 𝜀 > 0, it concludes that
By (17), we deduce that
󸀠
𝐺 (𝑥)
[𝑇 − 2𝐴 1 (1)] 𝐸0
𝑇−𝐴 1 (𝑥)
≤ −∫ [𝛼𝑘,𝑡 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡) 𝑇−𝐴 1 (1)
𝐴 1 (𝑥) =∫ 𝐸0 𝑑𝑡
𝐴 1 (1)
1 󵄨 󵄨2
+ 𝛽𝑘,𝑥 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑡 𝑇−𝐴 1 (1)
2 ≤ (1 + 𝑘𝑇) ∫ 𝐸 (𝑡) 𝑑𝑡
𝑇−𝐴 1 (𝑥) 𝐴 1 (1)
=∫ [ − 𝑘𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
1 𝑇−𝐴 1 (1) 1 󵄨 󵄨2
𝐴 1 (𝑥)
= (1 + 𝑘𝑇) ∫ ∫ [𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑡
2 𝐴 1 (1) 0
𝑘2 𝑥 󵄨󵄨 󵄨2
+ 󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑡 󵄨 󵄨2
𝛼𝑘 (𝑡) + 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑡 𝑑𝑥.
󵄨󵄨 𝑇−𝐴 1 (𝑥) 󵄨󵄨 (32)
󵄨 𝑘 󵄨
≤ 󵄨󵄨󵄨󵄨∫ √𝛼𝑘 (𝑡)𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡) 𝑑𝑡󵄨󵄨󵄨󵄨
󵄨󵄨 𝐴 1 (𝑥) √𝛼𝑘 (𝑡) 󵄨󵄨
Now choose that 𝐴 1 (𝑥) also satisfies
󵄨󵄨 𝑇−𝐴 1 (𝑥) 󵄨
󵄨 𝑘2 𝑥 󵄨󵄨 󵄨󵄨2 󵄨󵄨󵄨󵄨
+ 󵄨󵄨󵄨󵄨∫ 𝛽 (𝑥, 𝑡) 󵄨𝑧
󵄨 𝑥 (𝑥, 𝑡) 󵄨 𝑑𝑡
󵄨 󵄨󵄨󵄨
󵄨󵄨 𝐴 1 (𝑥) 1 − 𝑘2 𝑥2 𝑘 󵄨 𝐴󸀠1 (𝑥) > 0, 𝐴 1 (1) ≥ 𝐴 1 (𝑥) ≥ 𝐴 1 (0) ≥ 0. (33)
Journal of Applied Mathematics 5

Then from (19), (31), and (32), it follows that Assume that
󵄨2 󵄨󵄨 1
𝑇−𝐴 (𝑥)
1 󵄨
[𝛼𝑘 (𝑡)𝑧𝑡 (𝑥, 𝑡)𝑧𝑥 (𝑥, 𝑡) + 𝛾𝑘 (𝑥)󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ]󵄨󵄨󵄨󵄨
[𝑇 − 2𝐴 1 (1)] 𝐸0 2 󵄨𝐴 1 (𝑥)
1
1 𝑇−𝐴 1 (𝑥) 1 󵄨 󵄨2 + (− 𝐴󸀠1 (𝑥))
≤ (1 + 𝑘𝑇) ∫ ∫ [𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑡 2
2 𝐴 1 (𝑥) 0
󵄨 󵄨2 󵄨 󵄨2
󵄨 󵄨2 × ∑ [𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 + 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ]
+ 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑡 𝑑𝑥 𝑡=𝑇−𝐴 1 (𝑥),𝐴 1 (𝑥)

1 1
≤ (1 + 𝑘𝑇) ∫ 𝐺 (𝑥) 𝑑𝑥 ≤ 𝐴󸀠1 (𝑥)
0 2

≤ 𝐶 (1 + 𝑘𝑇) 𝐺 (0) , 󵄨 󵄨2
× ∑ {(𝜀 − 1) 𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨
(34) 𝑡=𝑇−𝐴 1 (𝑥),𝐴 1 (𝑥)

󵄨󵄨 󵄨
1 𝛼𝑘 (𝑡) 󵄨󵄨𝛾𝑘 (𝑥)󵄨󵄨󵄨
from which and from (20), we have that +[ + − 𝛽𝑘 (𝑥, 𝑡)]
𝜀 (𝐴󸀠 (𝑥))2 𝐴󸀠1 (𝑥)
1

󵄨 󵄨2
1 𝑇−𝐴 1 (𝑥) 󵄨 󵄨2 × 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 } ≤ 0.
𝐺 (0) = ∫ 𝛽𝑘 (0, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (0, 𝑡)󵄨󵄨󵄨 𝑑𝑡
2 𝐴 1 (𝑥)
(35) (38)
1
≥ [𝑇 − 2𝐴 1 (1)] 𝐸0 . We must take 0 < 𝜀 ≤ 1 and 𝐴 1 (𝑥) satisfies
𝐶 (1 + 𝑘𝑇)
󵄨󵄨 󵄨
1 𝛼𝑘 (𝑡) 󵄨󵄨𝛾𝑘 (𝑥)󵄨󵄨󵄨
+ − 𝛽𝑘 (𝑥, 𝑡) ≤ 0. (39)
Let 𝜀 (𝐴󸀠 (𝑥))2 𝐴󸀠1 (𝑥)
1

From (39), we derive

𝑇𝑘∗ ≜ 2𝐴 1 (1) . (36) 1 2𝑘𝑥 𝑘2 𝑥 2 1
+ + ≤ . (40)
𝜀(𝐴󸀠1 (𝑥))
2 𝐴󸀠1 (𝑥) 𝛼𝑘 (𝑡) 𝛼𝑘2 (𝑡) 𝛼𝑘2 (𝑡)

When 𝑇 > 𝑇𝑘∗ , by (35), (16) follows. Let 𝜀 = 1. Then it follows that
In the following, when 𝑇 > 𝑇𝑘∗ , we choose 𝐴 1 (𝑥) which
1 𝑘𝑥 2 1
satisfies (33) and (26). In (26), for 0 < 𝜀 ≤ 1, we have that [ + ] ≤ 2 ; (41)
𝐴󸀠1 (𝑥) 𝛼𝑘 (𝑡) 𝛼𝑘 (𝑡)

󵄨󵄨 󵄨 that is,
󵄨󵄨𝛼𝑘 (𝑡) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨
1 𝑥 1
󵄨󵄨 󵄨󵄨 + ≤ . (42)
= 󵄨󵄨󵄨󵄨√𝛼𝑘 (𝑡)𝑧𝑡 (𝑥, 𝑡) √𝛼𝑘 (𝑡)𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨󵄨 𝐴󸀠1 (𝑥) 𝛼𝑘 (𝑡) 𝛼𝑘 (𝑡)
󵄨 󵄨
󵄨󵄨 󵄨󵄨 By (42), we deduce that 𝑇 > 𝑇𝑘∗ ,
󵄨󵄨 √𝛼𝑘 (𝑡) 󵄨󵄨
󵄨󵄨 󵄨
= 󵄨󵄨󵄨√𝛼𝑘 (𝑡)√𝐴 1 (𝑥)𝑧𝑡 (𝑥, 𝑡)
󸀠 𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨󵄨 𝛼𝑘 (𝑡) 1 + 𝑘𝑇𝑘∗
󵄨󵄨 󸀠
√𝐴 1 (𝑥) 󵄨󵄨 𝐴󸀠1 (𝑥) ≥ ≥ . (43)
󵄨󵄨 󵄨󵄨 1 − 𝑘𝑥 1 − 𝑘𝑥
𝜀 󵄨 󵄨2 1 𝛼𝑘 (𝑡) 󵄨󵄨 󵄨2 Integrating into (0, 𝑥), we have
≤ 𝛼𝑘 (𝑡) 𝐴󸀠1 (𝑥) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 + 󵄨𝑧 (𝑥, 𝑡)󵄨󵄨󵄨
2 2𝜀 𝐴󸀠1 (𝑥) 󵄨 𝑥
1 + 𝑘𝑇𝑘∗
𝐴 1 (𝑥) ≥ [− ln (1 − 𝑘𝑥)] . (44)
𝐴󸀠 (𝑥) 󵄨 󵄨2 𝑘
= 1 [𝜀𝛼𝑘 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 ]
2 Hence, we choose
𝐴󸀠1 (𝑥) 1 𝛼𝑘 (𝑡) 󵄨󵄨 󵄨2 1 + 𝑘𝑇𝑘∗
+ [ 󵄨𝑧 (𝑥, 𝑡)󵄨󵄨󵄨 ] , 𝐴 1 (𝑥) = [− ln (1 − 𝑘𝑥)] , (45)
2 𝜀 (𝐴󸀠 (𝑥))2 󵄨 𝑥 𝑘
1

󵄨󵄨 1 󵄨 󵄨󵄨 󵄨󵄨 which satisfies (33) and (26). It follows that

󵄨󵄨 𝛾 (𝑥) 󵄨󵄨󵄨𝑧 (𝑥, 𝑡)󵄨󵄨󵄨2 󵄨󵄨󵄨 ≤ 1 𝐴󸀠 (𝑥) 󵄨󵄨𝛾𝑘 (𝑥)󵄨󵄨 󵄨󵄨󵄨𝑧 (𝑥, 𝑡)󵄨󵄨󵄨2 .
󵄨󵄨 𝑘 󵄨 𝑥 󵄨
󵄨 󵄨󵄨 2 1
󵄨2 𝐴󸀠1 (𝑥) 󵄨 𝑥 󵄨 1 + 𝑘𝑇𝑘∗
(37) 𝐴 1 (1) = [− ln (1 − 𝑘)] . (46)
𝑘
6 Journal of Applied Mathematics

From the definition of 𝑇𝑘∗ (see (36)) and (46), we deduce that Therefore, by (48)–(50), we have
𝑘 ∈ (0, 1 − 1/√𝑒),
1 𝑇 󵄨 󵄨2
−2 ln (1 − 𝑘) ∫ 𝛽 (0, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (0, 𝑡)󵄨󵄨󵄨 𝑑𝑡 ≤ 𝐶𝐸0
𝑇𝑘∗ = = 2𝐴 1 (1) . (47) 2 0 𝑘
𝑘 [1 + 2 ln (1 − 𝑘)]
󵄨 󵄨2 󵄨 󵄨2
In the following, we give the proof of the second inequal- ≤ 𝐶 (󵄨󵄨󵄨󵄨𝑧0 󵄨󵄨󵄨󵄨𝐻1 (0,1) + 󵄨󵄨󵄨󵄨𝑧1 󵄨󵄨󵄨󵄨𝐿2 (0,1) ) .
0
ity in (16). (51)
We choose 𝑞(𝑥) = 𝑥 − 1 for 𝑥 ∈ [0, 1] in (18). Noting that
𝛼𝑘󸀠 (𝑡) = 𝑘, 𝛽𝑘,𝑥 (𝑥, 𝑡) = −2𝑘2 𝑥/(1 + 𝑘𝑡), and 𝛾𝑘 (𝑥) = −2𝑘𝑥, it
follows that
Remark 8. Theorem 5 implies that, for any (𝑧0 , 𝑧1 ) ∈
1 𝑇 󵄨 󵄨2 𝐻01 (0, 1) × 𝐿2 (0, 1), the corresponding solution 𝑧 of (12)
∫ 𝛽 (0, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (0, 𝑡)󵄨󵄨󵄨 𝑑𝑡
2 0 𝑘 satisfies 𝑧𝑥 (0, ⋅) ∈ 𝐿2 (0, 𝑇).
𝑇
= ∫ 𝐸 (𝑡) 𝑑𝑡 3. Controllability: Proof of Theorem 4
0
𝑇 1 In this section, we prove the exact controllability for the wave
− ∫ ∫ 𝑘 (𝑥 − 1) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡) 𝑑𝑥 𝑑𝑡 equation (8) in the cylindrical domain 𝑄 (Theorem 4) by
0 0
(48) Hilbert uniqueness method. The main idea is to seek a control
𝑇 1 2
𝑘 𝑥 (𝑥 − 1) 󵄨󵄨 󵄨2 in the form V = 𝑧𝑥 (0, 𝑡), where 𝑧 is the solution of (12) for
+∫ ∫ 󵄨𝑧 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑥 𝑑𝑡
0 0 1 + 𝑘𝑡 󵄨 𝑥 some suitable initial data. The other proof is similar to the
proof of Theorem 2.1 in [3].
1
+ ∫ [𝛼𝑘 (𝑡) (𝑥 − 1) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡) Remark 9. It is easy to check that
0

󵄨𝑇 −2 ln (1 − 𝑘)
− 𝑘𝑥(𝑥 − 1)|𝑧𝑥 (𝑥, 𝑡)|2 ] 𝑑𝑥󵄨󵄨󵄨󵄨0 . 𝑇0∗ ≜ lim 𝑇𝑘∗ = lim = 2. (52)
𝑘→0 𝑘 → 0 𝑘 [1 + 2 ln (1 − 𝑘)]
Next, we estimate every term in the right side of (48).
It is well known that the wave equation (3) in the cylindrical
Notice that 1 ≤ 𝛼𝑘 (𝑡) ≤ 1 + 𝑘𝑇 and 0 < (1 − 𝑘2 )/(1 + 𝑘𝑇) ≤ domain is null controllable at any time 𝑇 > 𝑇0∗ . However, we
𝛽𝑘 (𝑥, 𝑡) ≤ 1 for any (𝑥, 𝑡) ∈ 𝑄. By (17), we have do not know whether the controllability time 𝑇𝑘∗ is sharp.
𝑇 𝑇 1
∫ 𝐸 (𝑡) 𝑑𝑡 − ∫ ∫ 𝑘 (𝑥 − 1) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡) 𝑑𝑥 𝑑𝑡 Conflict of Interests
0 0 0

𝑇 1
𝑘2 𝑥 (𝑥 − 1) 󵄨󵄨 󵄨2 The authors declare that there is no conflict of interests
+∫ ∫ 󵄨𝑧 (𝑥, 𝑡)󵄨󵄨󵄨 𝑑𝑥 𝑑𝑡
0 0 1 + 𝑘𝑡 󵄨 𝑥 regarding the publication of this paper.
𝑇 𝑇 1
󵄨 󵄨2 󵄨 󵄨2
≤ ∫ 𝐸 (𝑡) 𝑑𝑡 + 𝐶 ∫ ∫ [󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 + 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑥 𝑑𝑡 Acknowledgments
0 0 0
𝑇 This work is partially supported by the NSF of China under
≤ ∫ 𝐸 (𝑡) 𝑑𝑡 Grants 11171060 and 11371084 and Department of Education
0 Program of Jilin Province under Grants 2012187 and 2013287.
𝑇 1
󵄨 󵄨2 󵄨 󵄨2
+ 𝐶 ∫ ∫ [𝛼𝑡 (𝑡) 󵄨󵄨󵄨𝑧𝑡 (𝑥, 𝑡)󵄨󵄨󵄨 + 𝛽𝑘 (𝑥, 𝑡) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑥 𝑑𝑡 References
0 0
𝑇 [1] M. M. Miranda, “Exact controllability for the wave equation
≤ 𝐶 ∫ 𝐸 (𝑡) 𝑑𝑡 ≤ 𝐶𝐸0 . in domains with variable boundary,” Revista Matemática de la
0
Universidad Complutense de Madrid, vol. 9, no. 2, pp. 435–457,
(49)
1996.
On the other hand, for each 𝑡 ∈ [0, 𝑇], it holds that [2] C. Bardos and G. Chen, “Control and stabilization for the
wave equation—part III: domain with moving boundary,” SIAM
󵄨󵄨 1
󵄨󵄨 Journal on Control and Optimization, vol. 19, no. 1, pp. 123–138,
󵄨󵄨∫ [𝛼𝑘 (𝑡) (𝑥 − 1) 𝑧𝑡 (𝑥, 𝑡) 𝑧𝑥 (𝑥, 𝑡)
󵄨󵄨 0 1981.
󵄨
[3] L. Cui, X. Liu, and H. Gao, “Exact controllability for a one-
󵄨󵄨
󵄨 󵄨2 󵄨 dimensional wave equation in non-cylindrical domains,” Jour-
− 𝑘𝑥 (𝑥 − 1) 󵄨󵄨󵄨𝑧𝑥 (𝑥, 𝑡)󵄨󵄨󵄨 ] 𝑑𝑥󵄨󵄨󵄨󵄨 (50) nal of Mathematical Analysis and Applications, vol. 402, no. 2,
󵄨󵄨
pp. 612–625, 2013.
𝑇 [4] M. M. Miranda, “HUM and the wave equation with variable
≤ 𝐶 ∫ 𝐸 (𝑡) 𝑑𝑡 ≤ 𝐶𝐸0 . coefficients,” Asymptotic Analysis, vol. 11, no. 4, pp. 317–341, 1995.
0
Copyright of Journal of Applied Mathematics is the property of Hindawi Publishing
Corporation and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.