The Journal of Geometric Analysis (2024) 34:151

https://doi.org/10.1007/s12220-024-01595-9

Exponential Decay for a Time-Varying Coefficients Wave

Equation with Dynamic Boundary Conditions

Jianghao Hao1 · Fangqing Du1

Received: 27 October 2023 / Accepted: 19 February 2024

© Mathematica Josephina, Inc. 2024

Abstract
We consider a wave equation with variable coefficients in time and space in a bounded
domain  which has the smooth boundary  = 0 ∪ 1 such that  0 ∩  1 = ∅. We
study this system that has a homogeneous Dirichlet boundary on 0 and a dynamic
boundary on 1 . The innovation of the paper lies in the coefficients which depends on
the time variable and the singularities generated by changing the boundary conditions
along the interface, thus we need some special techniques to deal with these difficulties.
Under some geometric assumptions, the exponential decay result of the system is
established by the Riemannian geometry method and the energy perturbation method.

Keywords Wave equation · Time-varying coefficients · Dynamic boundary

conditions · Exponential decay

Mathematics Subject Classification 35B37 · 35L55 · 74D05 · 93D15

1 Introduction

Let  be a bounded domain in R n (n ≥ 2) which has a boundary  = 0 ∪ 1 of

class C 2 . Here meas(0 ) and meas(1 ) are positive and  =  0 ∩  1 = ∅. Let ω be
an open neighborhood of the part 1 of the boundary that is supposed to be connected
and meas(ω ∩ 0 ) > 0 (see Fig. 1).
Here ν = (ν1 , · · · , νn ) represents the outward unit vector normal to . We denote
the gradient and the divergence by ∇ and div respectively, and the tangential-gradient
and the tangential-divergence by ∇T and divT respectively in the Eucliden metric. In

B Jianghao Hao
[email protected]
Fangqing Du
[email protected]

1 School of Mathematical Sciences, Shanxi University, Taiyuan 030006, Shanxi, China

0123456789().: V,-vol 123

151 Page 2 of 30 J. Hao, F. Du

Fig. 1 An example of domain  satisfying the required geometrical assumptions for n = 2

this paper, we study the following problem

⎧
⎪
⎪u tt + A(x, t)u + b(x)u t = 0
⎪ in  × R + ,
⎪
⎪
⎨u = 0 on 0 × R + ,
∂u (1.1)
⎪
⎪u tt + + AT (x, t)u + u t = 0 on 1 × R + ,
⎪
⎪ ∂ν A
⎪
⎩u(x, 0) = u (x), u (x, 0) = u (x)
0 t 1 in ,

where

b(x) = b0 , x ∈ ω,

and b0 > 0 is a constant. The initial data u 0 and u 1 are in suitable function spaces. The
second-order differential operators A(x, t) and AT (x, t) are given by

A(x, t)u := −β(t)div(A(x)∇u),

 
AT (x, t)u := −β(t)divT A(x)|1 ∇T u ,

in which β ∈ W 1,∞ (0, ∞) is a given function and A(x) = (ai j (x))n×n are symmetric
and positive definite matrices functions with ai j (x) ∈ C ∞ (R n ), and the operators also
satisfy the uniform ellipticity conditions


n 
n
ai j (x)ξi ξ j ≥ λ ξi2 , x ∈ , 0 = (ξ1 , ξ2 , ..., ξn )T ∈ R n ,
i, j=1 i=1

for some constant λ > 0. And

∂u 
n
∂u
= β(t) ai j (x) νi = β(t)A(x)∇u · ν
∂νA ∂x j
i, j=1

123
Exponential Decay for a Wave Equation Page 3 of 30 151

is the outer normal derivative.

Since the boundaries 0 and 1 satisfy  =  0 ∩  1 = ∅, the singularities occur
when the boundary conditions change from 0 to 1 . We cannot get the regularity
of solution by the elliptic results, so we have to use some technicalities to overcome
this difficulty. For the two-dimensional case, we can refer to the method in [1] to deal
with the problem of lack of regularity, and for more on this can be seen in [2, 3]. The
main idea in these papers is to divide the weak solution u corresponding to the elliptic
problem into two parts that the regular part and the singular part. More precisely, they
decompose the solution into

u := u 1 + u 2 ,

where u 1 ∈ H 2 () and u 2 is given by

 √ θ
u 2 := ρ(r , θ ) r sin ,
2
x∈

here (r , θ ) is a coordinate system centered on x ∈  and ρ is an appropriately smooth

function with a compact support satisfying 0 ≤ ρ ≤ 1. This decomposition of u allows
us to estimate some integrals that resulting from the existence of singularities. For the
case of higher dimensions (n ≥ 3), Bey et al. [4] extended the above results, and [4,
Theorem 4] is very important and helpful for the proof of our stability. Later, Cornilleau
et al. [5] further developed the results of [4] and considered the possible singularities
in which they changed the boundary conditions along the interface  =  0 ∩  1 .
They assumed a partition (0 , 1 ) of  = ∂ such that

•  =  0 ∩  1 is a C 3 -manifold of dimension n − 2,
• (x − x0 ) · ν = 0 on , where x0 ∈ R n is a fixed point,
•  ∩ is a C 3 -manifold of dimension n − 1,
• Hn−1 (0 ) > 0,

where n ≥ 2 is the dimension of , is a suitable neighborhood of  and Hn−1

denotes the usual (n−1)-dimensional Hausdorff measure. Under a simple geometrical
condition concerning the orientation of the boundary, they obtained the stability results
for systems with linear or nonlinear Neumann feedbacks.
It is worth noting that [4] and [5] mentioned here are the literature for the constant
coefficients. For the variable coefficients case, we can refer to [6] which extended the
stability results of [5] to a time-dependent coefficients case. In [6] Cavalcanti et al.
concerned the following hyperbolic equation with boundary damping

⎧
⎪
⎪ K (x, t)u tt − A(t)u + F(x, t, u, ∇u) = 0 in  × R + ,
⎨
u=0 on 0 × R + ,
⎪
⎪ ∂u
⎩ + β(x)u t = 0 on 1 × R + ,
∂ν A

123
151 Page 4 of 30 J. Hao, F. Du

where  ⊂ R n (n ≥ 2) is a bounded open set with the boundary  = 0 ∪ 1 such

that  0 ∩  1 = ∅. Let x0 ∈ R n be a fixed point, the sets 0 and 1 given by

0 = {x ∈  : (x − x0 ) · ν < 0} and 1 = {x ∈  : (x − x0 ) · ν ≥ 0}.

Under some assumptions about the functions F, K and A, the authors obtained the
exponential decay result by using the energy method.
Our work changes part of the boundary conditions in the above reference, that is,
we study the influence of the dynamic boundary on the stability of system. This type
of boundary condition takes acceleration into account on the boundary to affect the
stability and the exact controllability of elastic structures. In [7], Li et al. considered
the following one-dimensional system
⎧
⎪
⎪u tt − u x x = 0, x ∈ (0, 1), t ≥ 0,
⎨
u(0, t) = 0, t ≥ 0,
⎪
⎪ ∂u
⎩κu tt (1, t) + (1, t) + u t (1, t) = 0, t ≥ 0.
∂ν

At κ = 1, they got the optimal polynomial decay result, even though the system is
exponentially stable if κ = 0. Thus, for the study of dynamic boundary, it not only
is very important for theoretical significance, but also is a good reference for some
practical applications. This kind of boundary is suitable for dynamic vibration mod-
eling of linear viscoelastic rods and beams with attached masses at their free ends, we
refer to the reference [8–13]. These questions are common in analyzing the mechan-
ical behavior of any structure with elongated members attached to smaller, heavier
objects, for example, a structure consisting of robotic arms attached to satellites. For
early studies of the system with dynamic boundary conditions we can refer to [14–16].
[16] was devoted to study of the following damped Cauchy-Ventcel problem
⎧
⎪
⎪
⎪u tt + A(x)u + a(x)g1 (u t ) = 0 in  × R + ,
⎪
⎪
⎨u = v on  × R + ,
⎪u=0 on 0 × R + ,
⎪
⎪ ∂u
⎪
⎪
⎩vtt + + AT (x)v + g2 (vt ) = 0 on 1 × R + ,
∂νA

where there exists a vector field H such that

1 = {x ∈  : H · ν > 0} and  0 ∩  1 = ∅.

The uniform energy decay rate for the above problem was established by Riemannian
geometry method which was first introduced by Yao [17] to study the exact control-
lability of wave equation with variable coefficients. For more information about the
variable coefficients that are only related to the space variable, we can refer to [18–23]
and the references in them.

123
Exponential Decay for a Wave Equation Page 5 of 30 151

While for the time-varying coefficients case, we can refer to [24–27]. In [25], Liu
considered the mixed problem

u tt + A(x, t)u = 0,

with Neumann boundary control

∂u
= v on ,
∂νA

or Dirichlet boundary control

u = v on 0 and u = 0 on 1 ,

where  0 ∩  1 = ∅ and v is a suitable control function. The observability inequalities

were established by the Riemannian geometry method under some geometric condi-
tions. For more on time-dependent linear operators, evolution families, and evolution
equations and their applications, we refer the reader to [28, 29] and the references
therein.
Inspired by the above literature, in this paper we mainly study the system (1.1) with
the assumption

 0 ∩  1 = ∅.

In fact, there are two main difficulties in our work. First, because the coefficients are
related to the time variable, we cannot accurately estimate the positive and negative
properties of the derivative of the energy functional, which makes it impossible to get
the stability results of the system by traditional methods. Second, we lose the regularity
of solution due to the existence of singularities. So for these two main difficulties, we
need more skills to deal with system (1.1) to get the decay result that we want.
The paper is organized as follows. Section 2 presents the notations and some
assumptions that we need to follow. By using the semigroup method, we give the
well-posedness result in Sect. 3. In Sect. 4, we obtain the exponential decay estimate
of the energy. Finally, a concluding remark is stated in Sect. 5.

2 Preliminaries

In this section, we present some materials and assumptions used in this paper. L 2 (·)
and H 1 (·) denote the usual Sobolev spaces. · 2 and · 2,1 are the norms in the
L 2 () and L 2 (1 ), respectively. For simplicity, we write · and · 1 instead of
· 2 and · 2,1 , respectively. Let C denote various positive constants which may
be different at different occurrences.
Denote

H10 () := u ∈ H 1 () : u|0 = 0 .

123
 151 Page 6 of 30 J. Hao, F. Du

Since the Poincaré inequality holds in H10 (), then the apace H10 () can be endowed
with the norm ∇ · , which is equivalent to usual norm of H 1 ().

2.1 Riemannian Notations

We define

g(x) = (gi j (x)) = A−1 (x), x ∈ R n ,

as a Riemannian metric on R n and consider the couple (R n , g) as a Riemannian

manifold with the inner product and the norm

X , Y g = (A−1 (x)X , Y ) = A−1 (x)X · Y , |X |2g = X , X g , X , Y ∈ Rxn ,

where (·, ·) is the Euclidean product of R n . For any C 1 function w, we define


n
∇g w = A(x)∇w, |∇g w|2g = ai j (x)wxi wx j , x ∈ R n ,
i, j=1

where ∇g is the gradient of the metric g. Denote by D the Levi-Civita connection in

the Riemannian metric g and let H be a vector field on R n , then for each x ∈ R n , the
covariant differential D H of H determines a bilinear form on R n × R n :

D H (X , Y ) = DY H , X g , X , Y ∈ Rxn ,

where DY H is the covariant derivative of the vector field H with respect to Y . Next
we give the following lemma that provides some further relationships between the
Riemannian metric g and the Euclidean metric.
 
Lemma 2.1 [17] Let f ∈ C 2  and H be vector field. Then, with the references to
the above notations, we have

(i) H ( f ) = ∇g f , H g = ∇ f · H ,
   1   1
(ii) ∇g f , ∇g (H ( f )) g = D H ∇g f , ∇g f + div |∇g f |2g H − |∇g f |2g divH ,
2 2
where divH is the divergence of the vector field H in the Euclidean metric.

2.2 Assumptions

1,∞
(H1) Assume that β(t) ∈ Wloc (0, ∞), β  (t) ∈ L 1 (0, ∞) and

β(t) ≥ β0 > 0, t ≥ 0,

where β0 is some positive constant.

123
Exponential Decay for a Wave Equation Page 7 of 30 151

(H2) [30] Let H be a vector field on Riemannian manifold (R n , g), there exists a
continuous function σ (x) such that

D H (X , X ) = σ (x)|X |2g , X ∈ Rxn , x ∈ ,

and denote σ1 = min σ (x) > 0 and σ2 = max σ (x). Moreover, assuming that the
x∈ x∈
vector field H satisfies

H · ν ≤ 0 for x ∈ 0 ,

and

H · ν > 0 for x ∈ 1 .

Remark 2.2 The vector field H in assumption (H2), which called the escape vector
field and firstly introduced by [31] as a checkable assumption. The existence of the
vector field H depends on the Riemannian curvature of the metric g. In [32], we know
that if assumption (H2) holds, then GCC (Geometric Control Condition) holds. And
for the constant coefficients case i.e. considering A(x, t) = −, many papers always
take H = x −x0 where x0 ∈ R n is a fixed point and x ∈ .

2.3 Main Results

Consider the phase space

H := H10 () × L 2 () × H 1 (1 ) × L 2 (1 ),

endowed with the inner product

 
(w1 , w2 , w3 , w4 )T , (v1 , v2 , v3 , v4 )T
H
 
   
= β(t)∇g w1 ∇v1 + w2 v2 d x + β(t)(∇T )g w3 (∇T )v3 + w4 v4 d.
 1
(2.1)

where (∇T )g w = A(x)|1 (∇T )w for any C 1 function w. Taking U (t) =

(u, u t , γ1 (u), γ1 (u t ))T with the trace operator γ1 (·) = ·|1 , the system (1.1) can be
rewritten by ⎧
⎨ dU
= A(t)U , t > 0,
dt (2.2)
⎩ U (0) = U = (u , u , γ (u ), γ (u ))T ,
0 0 1 1 0 1 1

123
151 Page 8 of 30 J. Hao, F. Du

where the time-dependent linear operators A(t) have the form

⎛ ⎞ ⎛ ϕ
⎞
u
⎜ϕ ⎟ ⎜ −A(x, t)u − b(x)ϕ ⎟
A(t) ⎜ ⎟=⎜
⎜ ψ
⎟
⎟,
⎝v ⎠ ⎝ ⎠
∂v
ψ − − AT (x, t)v − ψ
∂νA

with domain
 
D(A(t)) = D(A(0)) = H 2 () ∩ H10 () × H10 ()
 
× H 2 (1 ) ∩ H 1 (1 ) × H 1 (1 ), t ≥ 0,

which means D(A(t)) do not depend on t.

Now, we state the well-posedness result for the Cauchy problem (2.2), which ensures
that the system (1.1) is globally well-posed.

Theorem 2.3 Suppose that (H1) and (H2) hold. Then for any U0 ∈ H, the problem
(2.2) admits a unique solution U (t) such that

U (t) ∈ C(R+ ; H).

Further, assuming  
A(x)|1 H , τ ≤ 0, x ∈ , (2.3)
where τ is the unit tangent vector pointing towards the exterior of 1 , from 1 to 0 .
In fact, for A(x) = I , this geometric assumption (2.3) is the same as in [4, 5] when
taking the vector field H = x − x0 where x0 ∈ R n is a fixed point and x ∈ .
Define the associated energy of system (1.1) by
   
1
E(t) := ut 2
+ β(t) |∇g u|2g d x + u t 2
1 + β(t) |(∇T )g u|2g d , (2.4)
2  1

according to the inner product of state space H. Our main decay result can be given
as follows

Theorem 2.4 Assume that (H1), (H2) and (2.3) hold and there exist positive constants
C2 , α and m such that for all t sufficiently large, it holds
 t
eC2 s |β  (s)|ds ≤ αt m . (2.5)
0

Then the energy decay exponentially, i.e.,

 
E(t) ≤ C E(0) + αt m e−C2 t .

123
Exponential Decay for a Wave Equation Page 9 of 30 151

Remark 2.5 For assumption (2.5), we give the following two examples for the function
β(s).
Example (i). Let β(s) = ae−bs + β0 with a > 0 and b ≥ C2 > 0. It is obviously
that taking β to satisfy (H1). By direct calculation, we have

|β  (s)| = abe−bs ,

and
 t  t
eC2 s |β  (s)|ds ≤ C e−(b−C2 )s ds.
0 0

Therefore, there exist positive constants α and m such that for all t sufficiently large,
(2.5) holds.
Example (ii). Let β(s) = se−as + β0 with a ≥ C2 > 0. It is obviously that taking
β to satisfy (H1). By direct calculation, we have

β  (s) = (1 − as)e−as ,

and
 t  t

e C2 s
|β (s)|ds ≤ C (1 + |s|)e−(a−C2 )s ds.
0 0

Therefore, there exist positive constants α and m such that for all t sufficiently large,
(2.5) holds.

3 Well-Posedness

In this section, we study the existence and uniqueness of the solution to system (1.1),
that is, using the semigroup method to prove Theorem 2.3.

Proof of Theorem 2.3 This proof is divided into four main steps.
Step 1. The first step is to prove that the linear operators A(t) are dissipative. Indeed, let
U = (v1 , v2 , v3 , v4 )T ∈ D(A(0)), using (2.1) and the fact of v1 = v3 , v2 = v4 on 1 ,
we have

 
A(t)U , U H = β(t)∇g v2 ∇v1 − (A(x, t)v1 + b(x)v2 )v2 d x

  
∂v3
+ β(t)(∇T )g v4 (∇T )v3 − + AT (x, t)v3 + v4 v4 d
1 ∂νA
 
= − b0 |v2 |2 d x − |v4 |2 d
ω 1
≤ 0,

123
151 Page 10 of 30 J. Hao, F. Du

which yields the operators A(t) are dissipative.

Step 2. In this step, we prove the surjection of the operators I − A(t), where I stands
for the identity operator. In fact, set F = ( f 1 , f 2 , f 3 , f 4 )T ∈ H, we will prove that
there exists U = (v1 , v2 , v3 , v4 )T such that

(I − A(t)) U = F,

which is equivalent to
⎧
⎪
⎪v1 − v2 = f 1
⎪
⎪
in H10 (),
⎪
⎨v2 + A(x, t)v1 + b(x)v2 = f 2 in L 2 (),
v3 − v4 = f 3 in H 1 (1 ), (3.1)
⎪
⎪
⎪
⎪ ∂v3
⎪
⎩v4 + + AT (x, t)v3 + v4 = f 4 in L 2 (1 ).
∂νA

From the first and third equations of (3.1), we obtain

v2 = v1 − f 1 ,
(3.2)
v4 = v3 − f 3 .

Substituting the equations of (3.2) into the second and the forth equations of (3.1), we
have ⎧
⎨v1 + A(x, t)v1 + b(x)v1 = f 1 + f 2 + b(x) f 1 ,
∂v3 (3.3)
⎩2v3 + + AT (x, t)v3 = 2 f 3 + f 4 .
∂νA
This is an elliptic system of two equations. For (ϕ1 , ψ1 ), (ϕ2 , ψ2 ) ∈ H10 () ×
H 1 (1 ), we introduce the following bilinear form

 
B((ϕ1 , ψ1 ), (ϕ2 , ψ2 )) = ϕ1 ϕ2 + β(t)∇g ϕ1 ∇ϕ2 + b(x)ϕ1 ϕ2 d x


 
+ 2ψ1 ψ2 + β(t)(∇T )g ψ1 (∇T )ψ2 d.
1

It is easy to show that B((ϕ1 , ψ1 ), (ϕ2 , ψ2 )) is a bounded bilinear form and

 ! "
B((ϕ1 , ψ1 ), (ϕ1 , ψ1 )) = |ϕ1 |2 + β(t)|∇g ϕ1 |2g + b(x)|ϕ1 |2 d x

 ! "
+ 2|ψ1 |2 + β(t)|(∇T )g ψ1 |2g d,
1

is coercive. Then by using the conditions of v1 = v3 , ϕ = ψ on 1 , we can find

(v1 , v3 ) ∈ H10 () × H 1 (1 ), such that for all (ϕ, ψ) ∈ H10 () × H 1 (1 ), the
following holds

123
Exponential Decay for a Wave Equation Page 11 of 30 151

 
B((v1 , v3 ), (ϕ, ψ)) = [ f 1 + f 2 + b(x) f 1 ]ϕd x + (2 f 3 + f 4 )ψd.
 1

Therefore, the system (3.3) admits a unique weak solution (v1 , v3 ) ∈ H10 () ×
H 1 (1 ) by the well-known Lax-Milgram theorem. And we deduce from (3.2) that
v2 ∈ H10 ()(→ L 2 ()) and v4 ∈ H 1 (1 )(→ L 2 (1 )). This implies that U ∈ H
which gives us the desired solution.
Step 3. Define a vector valued function h : R+ → H with h(t) = A(t)U . We will
prove in this step that h is differentiable and its Frechet derivative is the vector valued
function
⎛ ⎞
0
⎜ −A (x, t)u ⎟
h  (t) = ⎜
⎝
⎟,
⎠
0
−AT (x, t)v

where

A (x, t)u = −β  (t)div(A(x)∇u),

and

AT (x, t)v = −β  (t)divT (A(x)|1 ∇T v).

Indeed, it is quite obvious that h  (t) ∈ H for t ≥ 0. And for any t, τ ≥ 0 with t = τ ,
we have
⎛ ⎞
0
h(t) − h(τ ) 1 ⎜⎜ −[A(x, t) − A(x, τ )]u ⎟ .
⎟
=
t −τ t −τ ⎝ 0 ⎠
−[AT (x, t) − AT (x, τ )]v

Then
⎛ ⎞
 0 
⎜ ⎟
⎜ − A(x, t) − A(x, τ ) − A (x, t) u ⎟
h(t) − h(τ ) ⎜ − τ ⎟
− h  (t) = ⎜
⎜
t ⎟,
⎟
t −τ ⎜  0  ⎟
⎝ AT (x, t) − AT (x, τ ) ⎠
− − AT (x, t) v
t −τ

which yields
# # #   #
# h(t) − h(τ ) # # #
# − h 
(t) # = #− A(x, t) − A(x, τ ) − A (x, t) u #
# t −τ # # t −τ #
H

123
151 Page 12 of 30 J. Hao, F. Du

#   #
# AT (x, t) − AT (x, τ ) #
#
+ #− − AT (x, t) v #

t −τ #
1
# #
# β(t) − β(τ ) #
= #
# − β (t) div∇g u #

#
t −τ
# #
# β(t) − β(τ ) #
+## − β 
(t) div T (∇ #
T g # .
) v
t −τ 1

We have
# #
# h(t) − h(τ ) #
# − h (t)#

lim
t→τ # t −τ # = 0.
H

Hence according to the chapters 5 of Pazy’s book [33], we define the solution operators
of the initial value problem
⎧
⎨ dU$
= A(t)U $ 0 ≤ s < t ≤ T,
(3.4)
⎩ $dt $1 ,
U (s) = U

by

$1 = U
W (t, s)U $(t), 0 ≤ s < t ≤ T ,

where U$(t) is the solution of (3.4) and W (t, s) is a two parameter family of operators.
Then, the evolution equation (2.2) has a unique mild solution

U (t) = W (t, 0)U0 , t ∈ [0, Tmax ).

Step 4. Let us show that Tmax = ∞. From the definition of energy (2.4), we have
    
1 
E  (t) = β (t) |∇g u|2g d x + |(∇T )g u|2g d − b0 u 2t d x − u 2t d,
2  1 ω 1

and   
|β  (t)|
|E  (t)| ≤ |∇g u|2g d x + |(∇T )g u|2g d
2  1
   
|β (t)|β(t)
≤ |∇g u|2g d x + |(∇T )g u|2g d
2β0  1

|β (t)|
≤ E(t).
β0
The above inequality gives

|β  (t)| |β  (t)|
− E(t) ≤ E  (t) ≤ E(t). (3.5)
β0 β0

123
Exponential Decay for a Wave Equation Page 13 of 30 151

Let
 ∞ |β  (s)|
β1 = ds,
0 β0

we have
e−β1 E(0) ≤ E(t) ≤ eβ1 E(0), t ∈ [0, Tmax ). (3.6)
Then using (2.1), (2.4) and (3.6), we have

U 2H = 2E(t) ≤ 2eβ1 E(0), t ∈ [0, Tmax ),

where U = (u, u t , γ1 (u), γ1 (u t ))T ∈ H. Therefore, the local solution cannot blow-up
in finite time and it follows that Tmax = ∞. 


Remark 3.1 It is worth noting that in this paper we assumes that  0 ∩  1 = ∅, so we

cannot use the elliptic regularity argument to get u ∈ H 2 () and γ1 (u) ∈ H 2 (1 )
from

A(x, t)u ∈ L 2 () and AT (x, t)γ1 (u) ∈ L 2 (1 ).

Then, for the more regular initial data U0 ∈ D(A(0)), we do not have a more regular
solution U ∈ C(R+ ; D(A(0))) ∩ C 1 (R+ ; H).

4 Decay Result

Because of the existence of singularities, we need to avoid them in the following work.
As in [6], let δ > 0 be a small and fixed constant and consider
%
Bδ = B(x, δ),
x∈

where B(x, δ) = {y ∈  : |x − y| < δ}. Denote

δ =  \ Bδ .

Next, we will study the stability result of the corresponding system in δ (see
Fig. 2), whose boundary is defined as

∂δ = $
0 ∪ $
1 ∪ δ ,

where

0 = ∂δ ∩ 0 , $
$ 1 = ∂δ ∩ 1 ,

123
151 Page 14 of 30 J. Hao, F. Du

Fig. 2 The new domain δ for n = 2

and

δ = ∂ Bδ ∩ .

Then the system (1.1) is transformed into the following system

⎧
⎪
⎪ u tt + A(x, t)u + b(x)u t = 0 in δ × R + ,
⎨
u=0 on $
0 × R + , (4.1)
⎪
⎪ ∂u
⎩u tt + + AT (x, t)u + u t = 0 on $
1 × R + .
∂νA

Define the energy associated with the problem (4.1) in δ by


1
E δ (t) := |u t |2 d x
2 δ
   
+β(t) |∇g u|2g d x + |u t |2 d + β(t) |(∇T )g u|2g d , (4.2)
δ $
1 $
1

then by a simple calculation, we have

   
1 
E δ (t) = β (t) |∇g u|2g d x + |(∇T )g u|2g d − b0 u 2t d x
2 δ $
1 ω
 
∂u
− u t d +
2
u t d. (4.3)
$1 δ ∂νA

Lemma 4.1 The functional

 
δ (t) := u t ud x + u t ud (4.4)
δ $
1

123
Exponential Decay for a Wave Equation Page 15 of 30 151

satisfies

 
b0 C p ε1 Ct ε2
δ (t) ≤ − 1 − − β(t)|∇g u|2g d x − β(t)|(∇T )g u|2g d
λβ0 λβ0 δ $1
   (4.5)
b0 1 ∂u
+ 1+ |u t |2 d x + 1 + |u t |2 d + ud,
4ε1 δ 4ε2 $
1 δ ∂νA

where ε1 and ε2 are some positive constants, Ct is the smallest possible positive
constant produced by the trace theorem and Poincaré inequality and C p > 0 is the
Poincaré optimal embedding constant.

Proof Taking the derivative of (4.4) with respect to the variable t and using (4.1) and
Green formula yield
   
δ (t) = |u t |2 d x + u tt ud x + |u t |2 d + u tt ud
δ δ $
1 $
1
 
= |u t |2 d x + |u t |2 d
 $
1
δ   
∂u
− [A(x, t)u + b(x)u t ] ud x − + AT (x, t)u + u t ud
 1 ∂νA
$
 δ   
= |u t |2 d x + |u t |2 d − β(t)|∇g u|2g d x − β(t)|(∇T )g u|2g d
 $1 δ $
1
δ  
∂u
− b(x)u t ud x − u t ud + ud.
δ $1 δ ∂νA
(4.6)
Making use of Young’s inequality, Poincaré inequality, the trace theorem and (H1),
we have
& & 
& &
& b(x)u t ud x && ≤ b0 |u t u|d x
&
δ δ
 
b0
≤ |u t |2 d x + b0 ε1 |u|2 d x (4.7)
4ε1 δ δ
 
b0 b0 C p ε1
≤ |u t |2 d x + β(t)|∇g u|2g d x,
4ε1 δ λβ0 δ

and & & 

& &
& &
& $ u t ud & ≤ $ |u t u|d
1 1
  (4.8)
1 Ct ε2
≤ |u t | d +
2
β(t)|∇g u|2g d x.
4ε2 $ 1 λβ0 δ
Thus, taking (4.7) and (4.8) into (4.6), (4.5) is obtained. 


123
151 Page 16 of 30 J. Hao, F. Du

Lemma 4.2 Suppose that H is the vector field defined in assumption (H2), and define
HT = H − (H · ν)ν. Then, the functional
 
δ (t) := 2 u t H (u)d x + 2 u t HT (u)d (4.9)
δ $
1

satisfies

 
δ (t) ≤ − (nσ1 − b0 1 ) |u t |2 d x − (cσ − 2 ) |u t |2 d
δ $
1
' (
maxx∈¯ δ |H |2 b0
+ nσ2 − 2σ1 + β(t)|∇g u|2g d x
2λβ0 1 δ
' (
∇T A 1 ) max x∈$̄
L ∞ ($ 1
|HT | maxx∈$̄

|HT |2
+ + 1
− cσ β(t)|(∇T )g u|2g d
λ 2λβ0 2 $
1
 ) & & *
β(t) && ∂u &&2
+ |u t | + 2
H · νd
$
1 |νA |2g & ∂νA &
 
∂u
+ [|u t |2 − β(t)|∇g u|2g ]H · νd + 2 H (u)d,
δ δ ∂νA
(4.10)
where 1 , 2 and cσ are some positive constants.
Proof Taking the derivative of (4.9) with respect to the variable t and using (4.1) and
Green formula yield
 
δ (t) = 2 u t H (u t )d x + 2 u tt H (u)d x
δ δ
 
+2 u t HT (u t )d + 2 u tt HT (u)d
$
1 $
1
 
= H (u 2t )d x − 2 [A(x, t)u + b(x)u t ] H (u)d x
δ δ
   
∂u
+ HT (u 2t )d − 2 + AT (x, t)u + u t HT (u)d
$1 1 ∂νA
$
  
∂u
= H (u 2t )d x − 2 β(t)∇g u∇(H (u))d x + 2 H (u)d (4.11)
δ δ ∂δ ∂νA
  
∂u
−2 b(x)u t H (u)d x + HT (u 2t )d − 2 HT (u)d
δ $
1 $1 ∂ν A
 
−2 β(t)(∇T )g u∇T (HT (u))d − 2 u t HT (u)d.
$
1 $
1

We begin to deal with the terms on the right of (4.11). Since the assumption (H2), we
have

nσ1 ≤ divH ≤ nσ2 ,

123
Exponential Decay for a Wave Equation Page 17 of 30 151

then there are positive constants cσ and Cσ such that

cσ ≤ divT HT ≤ Cσ .

Using the divergence theorem, (H2) and the fact of HT · ν = 0, we have

 
H (u 2t )d x + HT (u 2t )d
δ $
1
  
= div(u t H )d x −
2
|u t |2 divH d x − |u t |2 divT HT d
δ δ $
1
  
= |u t |2 H · νd − |u t |2 divH d x − |u t |2 divT HT d
∂δ δ $
1
 
≤ |u t |2 H · νd + |u t |2 H · νd
$
1 δ
 
− nσ1 |u t |2 d x − cσ |u t |2 d. (4.12)
δ $
1

Using Lemma 2.1 and (H2), we get

 
−2 β(t)∇g u∇(H (u))d x − 2 β(t)(∇T )g u∇T (HT (u))d
δ $
1

= −2 β(t)∇g u, ∇g (H (u))g d x
δ
   
− β(t)A(x)|1 ∇T |∇T u|2 · HT d − 2 β(t)|(∇T )g u|2g divT HT d
$
1 $1
 
= −2 β(t)D H (∇g u, ∇g u)d x − β(t)|∇g u|2g H · νd
δ ∂δ

+ β(t)|∇g u|2g divH d x
δ
 
 
+ β(t) ∇T A(x)|1 |∇T u| · HT d −
2
β(t)|(∇T )g u|2g divT HT d
$
1 $
1
 
≤ (nσ2 − 2σ1 ) β(t)|∇g u|g d x −
2
β(t)|∇g u|2g H · νd
δ ∂δ
' (
∇T A(x)|1 L ∞ ($1 ) max x∈$̄
1
|HT |
+ − cσ β(t)|(∇T )g u|2g d.
λ $
1
(4.13)

123
151 Page 18 of 30 J. Hao, F. Du

And using Young’s inequality, we have

 
−2 b(x)u t H (u)d x ≤ 2b0 |u t H (u)|d x
δ δ
 
max x∈¯ δ |H |2 b0
≤ b0 1 |u t |2 d x + β(t)|∇g u|2g d x,
δ 2λβ0 1 δ
(4.14)
and
  max x∈$̄ |HT |2 

−2 u t HT (u)d x ≤ 2 |u t | d x +
2 1
β(t)|(∇T )g u|2g d,
$
1 $
1 2λβ0 2 $
1
(4.15)
where 1 > 0 and 2 > 0 are some constants. Substituting (4.12)–(4.15) into (4.11),
we obtain
 

δ (t) ≤ − (nσ1 − b0 1 ) |u t | d x − (cσ − 2 )
2
|u t |2 d
δ $1
' (
max x∈¯ δ |H |2 b0
+ nσ2 − 2σ1 + β(t)|∇g u|2g d x
2λβ0 1 δ
∇T A(x)|1 1 ) max x∈$̄
L ∞ ($ 1
|HT |
+
λ
max x∈$̄ |HT |2 

+ 1
− cσβ(t)|(∇T )g u|2g d
2λβ0 2 $
1
  
∂u
+ |u t |2 H · νd − 2 HT (u)d + |u t |2 H · νd
$
1 $1 ∂νA δ
 
∂u
− β(t)|∇g u|2g H · νd + 2 H (u)d. (4.16)
∂δ ∂δ ∂νA

Now, we are going to estimate the integrals over ∂δ . On $

0 ∪ $
1 , we have
& &
1 & ∂u &2
|∇g u|2g = & & + |(∇T )g u|2 ,
|νA |2g & ∂νA &g g

and
& &
∂u ∂u β(t) && ∂u &&2
H (u) = HT (u) + H · ν.
∂νA ∂νA |νA |2g & ∂νA &g

Since u = u t = 0 on $
0 , it hold that

∂u
H (u) = β(t)|∇g u|2g H · ν, x ∈ $
0 ,
∂νA

123
Exponential Decay for a Wave Equation Page 19 of 30 151

then
 
∂u
− β(t)|∇g u|2g H · νd + 2 H (u)d
∂δ ∂δ ∂νA

= β(t)|∇g u|2g H · νd
$
0
 ) & &2 *
∂u β(t) & ∂u &
+ 2 HT (u) − β(t)|(∇T )g u|2g H · ν + & & H · ν d
$ ∂νA |νA |2g & ∂ν &
1 A
 
∂u
− β(t)|∇g u|2g H · νd + 2 H (u)d.
δ δ ∂ν A
(4.17)
Combining (4.16) and (4.17), we infer that
 
δ (t) ≤ − (nσ1 − b0 1 ) |u t |2 d x − (cσ − 2 ) |u t |2 d
δ $
1
' (
max x∈¯ δ |H |2 b0
+ nσ2 − 2σ1 + β(t)|∇g u|2g d x
2λβ0 1 δ
∇T A(x)|1 1 ) max x∈$̄
L ∞ ($ 1
|HT |
+
λ
max x∈$̄ |HT |2  

+ 1
− cσ β(t)|(∇T )g u|2g d + β(t)|∇g u|2g H · νd
2λβ0 2 $
1 $
0
 ) & &2 *
&
β(t) & ∂u & &
+ |u t |2 − β(t)|(∇T )g u|2g + 2 & & H · νd
$
1 |ν A |g ∂νA
 ! " 
∂u
+ |u t |2 − β(t)|∇g u|2g H · νd + 2 H (u)d. (4.18)
δ δ ∂νA

Therefore, using (H2), we obtain (4.10). 



Next we define the perturbed energy associated with δ by

E δ,θ (t) := N E δ (t) + δ (t) + θ δ (t), (4.19)

where N and θ are some positive constants.

Lemma 4.3 If

E δ → E, δ →  and δ → , as δ → 0,

then

E δ,θ → E θ , as δ → 0,

123
151 Page 20 of 30 J. Hao, F. Du

where
 
(t) := u t ud x + u t ud,
 1
 
(t) := 2 u t H (u)d x + 2 u t HT (u)d,
 1

and

E θ (t) := N E(t) + (t) + θ (t).

Proof Using Lebesgue dominated convergence theorem, we can easily obtain Lemma
4.3. 


Obviously, for N large enough, we have

E δ ∼ E δ,θ .

Moreover, due to (3.6), there exists a positive constant β2 such that

E δ (t) ≤ β2 , t ≥ 0. (4.20)

Proof of Theorem 2.4 Differentiating (4.19) and taking (4.3), (4.5) and (4.10) into it,
we have


E δ,θ (t) = N E δ (t) + δ (t) + θ δ (t)
 
b0
≤ − N b0 + (nσ1 − b0 1 )θ − 1 + |u t |2 d x
4ε1 δ
 
1
− N + (cσ − 2 )θ − 1 + − CH θ |u t |2 d
4ε2 $
1

b0 C p ε1 Ct ε2
− 1− −
λβ0 λβ0
' ( 
max x∈¯ δ |H |2 b0
− nσ2 − 2σ1 + + CH θ β(t)|∇g u|2g d x
2λβ0 1 δ
 ∇T A L ∞ ($ |HT |
1 ) max x∈$̄ 1
− 1−
λ
max x∈$̄ |HT | 2 
1
+ − cσ θ β(t)|(∇T )g u|2g d
2λβ0 2 $
1
 
N |β  (t)| N |β  (t)|
+ β(t)|∇g u|2g d x + β(t)|(∇T )g u|2g d
2β0 δ 2β0 $
1
 
∂u ∂u
+N u t d + ud
δ ∂νA δ ∂νA

123
Exponential Decay for a Wave Equation Page 21 of 30 151

 ! " 
∂u
+θ |u t |2 − β(t)|∇g u|2g H · νd + 2θ H (u)d, (4.21)
δ δ ∂νA

where C H > 0 is some constant depending on H . Thus, we choose ε1 , ε2 , 1 and 2

small enough such that

nσ1 − b0 1 > 0,
cσ − 2 > 0,

and

b0 C p ε1 Ct C p ε2
1− − > 0.
λβ0 λβ0

Fixing them, choosing θ sufficiently small and N lager enough such that
' (
b0 C p ε1 Ct ε2 max x∈¯ δ |H |2 b0
c1 := 1 − − − nσ2 − 2σ1 + θ > 0,
λβ0 λβ0 2λβ0 1
' (
∇T A L ∞ ($1 ) max x∈$̄
1
|HT | max x∈$̄
1
|HT |2
c2 := 1 − + − cσ θ > 0,
λ 2λβ0 2
b0
c3 := N b0 + (nσ1 − b0 1 )θ − 1 + > 0,
4ε1

and

1
c4 := N + (cσ − 2 )θ − 1 + − C H θ > 0.
4ε2

Therefore, combining (4.20) and (4.21) we obtain


E δ,θ (t) ≤ − C1 E δ (t) + C|β  (t)| + θ δ (t) + δ (t), (4.22)

where

C1 := min {c1 , c2 , c3 , c4 } > 0,

  
∂u
δ (t) := 2 H (u) − β(t)|∇g u|g H · ν d,
2
δ ∂νA

and
  
∂u ∂u
δ (t) := u+N u t + θ |u t |2 H · ν d.
δ ∂νA ∂νA

123
 151 Page 22 of 30 J. Hao, F. Du

Fig. 3 The vectors ν, τ , ν̃ and τ̃ for n = 2

Then, integrating (4.22) from 0 to t, we obtain

  t  t 

E δ,θ (t) ≤ C E δ,θ (0) + e C2 s
|β (s)|ds + e C2 s
[δ (s) + δ (s)] ds e−C2 t ,
0 0
(4.23)
where C2 > 0 is some constant.
 t
• Estimate for L δ (t) := eC2 s δ (s)ds
0
(i) For n = 2, at each point x̃ ∈ , we can build the vectors ν̃ = ν(x̃) and τ̃ = τ (x̃)
which depend on x̃ and . ν̃ is the unit normal vector pointing towards the exterior
and τ̃ is the tangent vector pointing towards from 1 to 0 . If δ is small enough, each
point x ∈ δ belongs to one and only one coordinate system (x̃, ν̃, τ̃ ) and x belongs
to some arc of circle ∂ B(x̃, δ) contained in this coordinate system (see Fig. 3).
We decompose the solution

u(x) = u 1 (x) + u 2 (x) := u 1 (x) + η(x̃)U2 (x − x̃),

1
where u 1 ∈ H 2 (), η is locally in H 2 () and U2 is given by

√ θ
U2 (x − x̃) = U2 (r , θ ) = ρ(r ) r sin ,
2

where ρ is a C ∞ -function with compact support such that ρ(r ) = 1 in some neigh-
bourhood of 0 and supp(ρ) ⊂ [−, ] ⊂ (−1, 1), where  > 0 is as small as we want.
Then, using coordinate in (x̃, ν̃, τ̃ ), similar to [4], we have

∂U2 β(t)A(x) 1
2 H (U2 ) − β(t)|∇g U2 |2g H · ν = H̃ · τ̃ − ν · τ̃ , (4.24)
∂νA 4 δ

123
Exponential Decay for a Wave Equation Page 23 of 30 151

where H̃ = H (x̃) with H̃ · ν̃ = 0. When δ → 0, ∂ B(x̃, δ) behaves as a half-circle,

then we obtain

1
A(x)d → A(x̃) = Ã as δ → 0.
π δ ∂ B(x̃,δ)

Integrating (4.24) along ∂ B(x̃, δ), we get

  
∂U2 πβ(t)
2 H (U2 ) − β(t)|∇g U2 |2g H · ν d → Ã H̃ · τ̃ as δ → 0.
∂ B(x̃,δ) ∂νA 4

Therefore, integrating the regular part of solution u 1 on ∂ B(x̃, δ) and using Cauchy-
Schwarz inequality, Lebesgue dominated convergence theorem, (2.3) and the com-
pactness of function ρ, we obtain that L δ (t) has a limit of non-positive number as
δ → 0.
(ii) For n ≥ 3, when δ is small enough, we also know that each point x ∈ δ belongs
to one and only one plane defined by (x̃, ν̃, τ̃ ), and x belongs to some arc of circle
l(x̃, δ) contained in this plane (the figure is similar to Fig. 3).
Firstly, just like in the case of n = 2, we write

u(x) = u 1 (x) + u 2 (x),

then

  
∂u
2 H (u) − β(t)|∇g u|2g H · ν d
∂ B(x̃,δ) ∂νA
 ! "
= β(t)A(x) 2∇u · ν∇u · H − |∇u|2 H · ν d
∂ B(x̃,δ)
 ! "
= β(t)A(x) 2∇(u 1 + u 2 ) · ν∇(u 1 + u 2 ) · H − |∇(u 1 + u 2 )|2 H · ν d
∂ B(x̃,δ)
 ! "
= β(t)A(x) 2∇u 1 · ν∇u 1 · H − |∇u 1 |2 H · ν d
∂ B(x̃,δ)
+ ,- .
Iδ (∇u 1 )
 ! "
+ β(t)A(x) 2∇u 2 · ν∇u 2 · H − |∇u 2 |2 H · ν d
∂ B(x̃,δ)
+ ,- .
Iδ (∇u 2 )

+2 β(t)A(x) [∇u 1 · ν∇u 2 · H + ∇u 2 · ν∇u 1 · H − ∇u 1 · ∇u 2 H · ν] d .
∂ B(x̃,δ)
+ ,- .
Jδ (∇u 1 ,∇u 2 )

Since u 1 ∈ H 2 () and meas(∂ B(x̃, δ)) → 0 as δ → 0 and using the regularity of
β(t) and A(x), it is easy to get

Iδ (∇u 1 ) → 0 as δ → 0.

123
151 Page 24 of 30 J. Hao, F. Du

Further, let us do the decomposition of ∇u 2 :

∇u 2 = ∇ X u 2 + ∇2 u 2 ,

where ∇2 u 2 belongs to the plane (x̃, ν̃, τ̃ ) defined above and ∇ X u 2 is orthogonal to
∇2 u 2 , i.e.,

∇ X u 2 · ∇2 u 2 = 0 and |∇u 2 |2 = |∇ X u 2 |2 + |∇2 u 2 |2 .

Then
  
|∇u 2 |2 d = |∇ X u 2 |2 d + |∇2 u 2 |2 d.
∂ B(x̃,δ) ∂ B(x̃,δ) ∂ B(x̃,δ)

Thanks to the first part of Theorem 4 in [4] and the Lebesgue dominated convergence
theorem, we get

|∇ X u 2 |2 d → 0 as δ → 0.
∂ B(x̃,δ)

And as u 2 (x) = η(x̃)U2 (x − x̃), using the Fubini’s theorem, we have

  
|∇2 u 2 | d =
2
η 2
|∇2 U2 |2 dld(x̃),
∂ B(x̃,δ)  l(x̃,δ)

1
and know that this integral is bounded using η ∈ H 2 () and the definition of U2 . So
we can end up with

|∇u 2 |2 d ≤ C.
∂ B(x̃,δ)

Therefore, making use of Cauchy-Schwarz inequality, we obtain

 1
2
 1
2
|Jδ (∇u 1 , ∇u 2 )| ≤ C |∇u 1 |2 d |∇u 2 |2 d .
∂ B(x̃,δ) ∂ B(x̃,δ)

It tends to zero since the first term vanishes as δ → 0 and the second one is bounded.
And finally we start to deal with the term Iδ (∇u 2 ). Similar to the above process, we
decompose ∇u 2 and have

Iδ (∇u 2 ) = β(t)A(x)[2∇ X u 2 · ν∇ X u 2 · H − |∇ X u 2 |2 ]d
∂ B(x̃,δ)
+ ,- .
Iδ (∇ X u 2 )

123
Exponential Decay for a Wave Equation Page 25 of 30 151


+ β(t)A(x)[2∇2 u 2 · ν∇2 u 2 · H − |∇2 u 2 |2 H · ν]d
∂ B(x̃,δ)
+ ,- .
Iδ (∇2 u 2 )

+2 β(t)A(x)[∇ X u 2 · ν∇2 u 2 · H + ∇2 u 2 · ν∇ X u 2 · H ]d .
∂ B(x̃,δ)
+ ,- .
Jδ (∇ X u 2 ,∇2 u 2 )

As above, Iδ (∇ X u 2 ) → 0 as δ → 0. Since

 1
2
 1
2
|Jδ (∇ X u 2 , ∇2 u 2 )| ≤ C |∇ X u 2 |2 d |∇2 u 2 |2 d ,
∂ B(x̃,δ) ∂ B(x̃,δ)

we have the first term vanishes as δ → 0 and the second one is bounded, thus we
obtain Jδ (∇ X u 2 , ∇2 u 2 ) tends to zero. For Iδ (∇2 u 2 ), as the case of n = 2, we also
have

1 1
2∇2 U2 · ν∇2 U2 · H − |∇2 U2 |2 H · ν = H̃ · τ̃ − ν · τ̃ ,
4 δ

and


πβ(t)
β(t)A(x)[2∇2 U2 · ν∇2 U2 · H − |∇2 U2 |2 H · ν]dl → Ã H̃ · τ̃ as δ → 0.
l(x̃,δ) 4

In other words, this integral term on l(x̃, δ) is bounded. Then, the dominated conver-
gence theorem can be used to get

πβ(t)
Iδ (∇2 u 2 ) → η2 Ã H̃ · τ̃ d(x̃) as δ → 0.
4 

Therefor, using the assumption (2.3), we know that Iδ (∇2 u 2 ) converges to a non-
positive number.
In conclusion, we infer

δ (t) → ζ as δ → 0,

where ζ ≤ 0 is a real number. When ζ < 0, then there exists δ1 > 0 such that

δ (t) < 0, δ < δ1 ,

and
 t
L δ (t) = eC2 s δ (t)ds < 0.
0

123
 151 Page 26 of 30 J. Hao, F. Du

When ζ = 0, we need to talk about two cases. One case is that there exists a positive
constant δ2 such that

δ (t) < 0, δ < δ2 .

Then the process is the same as ζ < 0. The other case is that there exists a positive
constant δ3 such that

δ (t) > 0, δ < δ3 .

Then
 t  t
−C2 t −C2 t
0 ≤ L δ (t)e = e C2 s
δ (s)dse ≤ δ (s)ds → 0 as δ → 0,
0 0

i.e.,

L δ (t) ≡ 0 as δ → 0.

Thus, we get
L δ (t) ≤ 0 as δ → 0. (4.25)
 t
• Estimate for Mδ (t) := eC2 s δ (s)ds
0
Just like we did above, we split u into two parts, this is,

u(x) = u 1 (x) + u 2 (x),

where u 1 ∈ H 2 () is the regular part and u 2 = η · U2 is the singular part, then

∂u
ud
∂ B(x̃,δ) ∂νA

= β(t)A(x)(∇u 1 + ∇u 2 ) · ν(u 1 + u 2 )d
∂ B(x̃,δ)

= β(t)A(x)[∇u 1 · νu 1 + ∇u 2 · νu 2 + ∇u 1 · νu 2 + ∇u 2 · νu 1 ]d.
∂ B(x̃,δ)
(4.26)
Using Cauchy-Schwarz inequality and the above results, we have

∇u 1 · νu 1 d → 0 as δ → 0, (4.27)
∂ B(x̃,δ)

∇u 2 · νu 2 d → 0 as δ → 0, (4.28)
∂ B(x̃,δ)

∇u 1 · νu 2 d → 0 as δ → 0, (4.29)
∂ B(x̃,δ)

123
Exponential Decay for a Wave Equation Page 27 of 30 151

and 
∇u 2 · νu 1 d → 0 as δ → 0. (4.30)
∂ B(x̃,δ)

Taking (4.27)–(4.30) into (4.26) and using Lebesgue dominated convergence theorem
allow us to get  t 
∂u
eC2 s udds → 0 as δ → 0. (4.31)
0 ∂ B(x̃,δ) ∂νA

Analogously,
 
t ∂u
eC2 s u t dds → 0 as δ → 0. (4.32)
0 ∂ B(x̃,δ) ∂νA

On the other hand, using the decomposition of u into a regular part and a singular
part, we have
 t 
e C2 s
|u t |2 H · νdds
0 ∂ B(x̃,δ)
 t   t 
= eC2 s |u 1,t |2 H · νdds + eC2 s |u 2,t |2 H · νdds.
0 ∂ B(x̃,δ) 0 ∂ B(x̃,δ)

From the dominated convergence theorem, we have

 t 
e C2 s
|u 1,t |2 H · νdds → 0 as δ → 0.
0 ∂ B(x̃,δ)

As
  2π  δ 1 3
|u 2,t | H · νdl ≤ C
2
r 2 dr dθ ≤ Cδ 2 ,
l(x̃,δ) 0 0

we get
 t 
eC2 s |u 2,t |2 H · νdds → 0 as δ → 0.
0 ∂ B(x̃,δ)

Then it follows
 t 
eC2 s |u t |2 H · νdds → 0 as δ → 0. (4.33)
0 ∂ B(x̃,δ)

Therefore, combining (4.31)–(4.33), we get

 t
Mδ (t) = eC2 s δ (s)ds → 0 as δ → 0. (4.34)
0

123
 151 Page 28 of 30 J. Hao, F. Du

In conclusion, let δ → 0 and taking (4.25) and (4.34) into (4.23), we have
 t
Eθ (t) ≤ C Eθ (0) + eC2 s |β  (s)|ds e−C2 t .
0

Then, using the energy equivalence relation and (2.5), we get

 
E(t) ≤ C E(0) + αt m e−C2 t .

Thus, we complete the proof of Theorem 4.4. 



5 Conclusions

In this paper, we present a study on the stability of a time-varying coefficients wave

equation in the bounded domain . The smooth boundary of  is  = 0 ∪ 1 such
that  =  0 ∩  1 = ∅. We consider that a homogeneous Dirichlet boundary on 0
and a dynamic boundary with damping term on 1 . Since the coefficients depends
on the time variable and the singularities are generated by changing the boundary
conditions along the interface, these bring no small difficulty to our proof, so some
special techniques are needed to deal with these problems. Under the appropriate
geometric assumptions, the exponential decay result of the system is established by
the Riemannian geometry method and the energy perturbation method.
There are many other issues associated with this type of problem, but we have not
studied them here.

(i) The geometric conditions (H2) and (2.3) are essential in the proof of our expo-
nential stability result, but their necessity leads us to exclude many mathematical
models of interest that should also be uniformly stable.
(ii) We assume that there are no external forces acting on the system or its boundary
other than friction. If there is thermal force in system, can the energy still be
uniformly stable? What if there is a nonlinear negative source term on 1 ?
Acknowledgements This work was partially supported by National Natural Science Foundation of
China (12271315), the special fund for Science and Technology Innovation Teams of Shanxi Province
(202204051002015), Fundamental Research Program of Shanxi Province (202203021221018), the
Research Project Supported by Shanxi Scholarship Council of China (2021-008).

Author Contributions All authors contributed to the study conception and design. Material preparation,
data collection and analysis were performed by all authors. All authors wrote the first draft of the manuscript
and commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data Availability Statement Data sharing not applicable to this article as no datasets were generated or
analysed during the current study.

Declarations
Conflict of Interest The authors declare there is no conflict of interest.

123
Exponential Decay for a Wave Equation Page 29 of 30 151

Ethical Approval This article does not contain any studies with human participants or animals performed
by the authors.

References
1. Grisvard, P.: Exact controllability of solutions of the wave equation in the presence of singularities. J.
Math. Pures Appl. 68, 215–259 (1989)
2. Grisvard, P.: Elliptic Problems in Nonsmooth Domains. Pitman, London (1982)
3. Grisvard, P.: Singularities in Boundary Value Problems. Springer, Berlin (1992)
4. Bey, R., Lohéac, J.-P., Moussaoui, M.: Singularities of the solution of a mixed problem for a general
second order elliptic equation and boundary stabilization of the wave equation. J. Math. Pures Appl.
78(10), 1043–1067 (1999)
5. Cornilleau, P., Lohéac, J.-P., Osses, A.: Nonlinear Neumann boundary stabilization of the wave equation
using rotated multipliers. J. Dyn. Control Syst. 16(2), 163–188 (2010)
6. Cavalcanti, M.M., Domingos Cavalcanti, V.N., Vicente, A.: Stability for a nonlinear hyperbolic equa-
tion with time-dependent coefficients and boundary damping. Z. Angew. Math. Phys. 73(6), 1–20
(2022)
7. Li, C., Liang, J., Xiao, T.J.: Boundary stabilization for wave equations with damping only on the
nonlinear Wentzell boundary. Nonlinear Anal. 164, 155–175 (2017)
8. Krenk, S.: Dynamics of Structures. McGraw-Hill, New York (1975)
9. Banerjee, A.K., Kane, T.R.: Extrusion of a beam from a rotating base. J. Guidance Control Dyn. 12(2),
140–146 (1989)
10. Banks, H.T., Fabiano, R.H., Wang, Y.: Inverse problem techniques for beams with tip body and time
hysteresis damping. Mat. Apl. Comput. 8(2), 101–118 (1989)
11. Chang, T.P.: Forced vibration of a mass-loaded beam with a heavy tip body. J. Sound Vib. 164(3),
471–484 (1993)
12. Lee, H.P.: Vibration of an axially moving beam with a tip mass. Mech. Res. Commun. 20(5), 391–397
(1993)
13. Andrews, K.T., Kuttler, K.L., Shillor, M.: Second order evolution equations with dynamic boundary
conditions. J. Math. Anal. Appl. 197(3), 781–795 (1996)
14. Dalsen, G.V.: On fractional powers of a closed pair of operators and a damped wave equation with
dynamic boundary conditions. Appl. Anal. 53(1–2), 41–54 (1994)
15. Dalsen, G.V.: On the initial-boundary-value problem for the extensible beam with attached load. Math.
Methods Appl. Sci. 19(12), 943–957 (1996)
16. Cavalcanti, M.M., Khemmoudj, A., Medjden, M.: Uniform stabilization of the damped Cauchy-Ventcel
problem with variable coefficients and dynamic boundary conditions. J. Math. Anal. Appl. 328(2),
900–930 (2007)
17. Yao, P.F.: On the observability inequality for exact controllability of wave equations with variable
coefficients. SIAM J. Control Optim. 37(5), 1568–1599 (1999)
18. Hao, J.H., Du, F.Q.: Decay rate for viscoelastic wave equation of variable coefficients with delay and
dynamic boundary conditions. Math. Methods Appl. Sci. 44(1), 284–302 (2021)
19. Du, F.Q., Hao, J.H.: Energy decay for wave equation of variable coefficients with dynamic boundary
conditions and time-varying delay. J. Geom. Anal. 33(4), 1–26 (2023)
20. Mohamed, F., Ali, H.: Global existence and energy decay result for a weak viscoelastic wave equations
with a dynamic boundary and nonlinear delay term. Comput. Math. Appl. 71(3), 779–804 (2016)
21. Akram, B.A., Mohamed, F.: Stability result for viscoelastic wave equation with dynamic boundary
conditions. Z. Angew. Math. Phys. 69(4), 1–13 (2018)
22. Feng, B.W.: General decay rates for a viscoelastic wave equation with dynamic boundary conditions
and past history. Mediterr. J. Math. 15(3), 1–17 (2018)
23. Hao, J.H., Du, F.Q.: Decay and blow-up for a viscoelastic wave equation of variable coefficients with
logarithmic nonlinearity. J. Math. Anal. Appl. 506(1), 1–20 (2022)
24. Al-Khulaifi, W., Diagana, T., Guesmia, A.: Well-posedness and stability results for some nonau-
tonomous abstract linear hyperbolic equations with memory. Semigr. Forum. 105(2), 351–373 (2022)
25. Liu, Y.X.: Exact controllability of the wave equation with time-dependent and variable coefficients.
Nonlinear Anal. Real World Appl. 45, 226–238 (2019)

123
 151 Page 30 of 30 J. Hao, F. Du

26. Cavalcanti, M.M.: Exact controllability of the wave equation with mixed boundary condition and
time-dependent coefficients. Arch. Math. 35(1), 29–57 (1999)
27. Cavalcanti, M.M., Domingos Cavalcanti, V.N., Soriano, J.A.P.: Existence and boundary stabilization
of a nonlinear hyperbolic equation with time-dependent coefficients. Electron. J. Differ. Equ. 1998(08),
1–21 (1998)
28. Lunardi, A.: Analytic Semigroups and Optimal Regularity in Parabolic Problems, PNLDE Vol. 16.
Birkhäuser Verlag, Basel (1995)
29. Diagana, T.: Semilinear Evolution Equations and Their Applications. Springer, Cham (2018)
30. Cao, X.M., Yao, P.F.: General decay rate estimates for viscoelastic wave equation with variable coef-
ficients. J. Syst. Sci. Complex. 27, 836–852 (2014)
31. Yao, P.F.: Modeling and Control in Vibrational and Structural Dynamics—A Differential Geometric
Approach. Applied Mathematics and Nonlinear Science Series. CRC Press, Chapman and Hall/CRC
(2011)
32. Ning, Z.H.: Asymptotic behavior of the nonlinear Schrödinger equation on exterior domain. Math.
Res. Lett. 27(6), 1825–1866 (2020)
33. Pazy, A.: Semigroups of Linear Operators and Applications to Partial Differential Equations. Springer,
New York (1983)

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under
a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted
manuscript version of this article is solely governed by the terms of such publishing agreement and applicable
law.

123