Citation: Zhouhong Li, Wei Zhang, Chengdai Huang, Jianwen Zhou. Bifurcation for a fractional-order Lotka-Volterra predator-Cprey model with delay feedback control[J]. AIMS Mathematics, 2021, 6(1): 675-687. doi: 10.3934/math.2021040
[1] | Hongying Jiao, Shuhai Zhu, Jinguo Zhang . Existence of infinitely many solutions for critical sub-elliptic systems via genus theory. Communications in Analysis and Mechanics, 2024, 16(2): 237-261. doi: 10.3934/cam.2024011 |
[2] | Jizheng Huang, Shuangshuang Ying . Hardy-Sobolev spaces of higher order associated to Hermite operator. Communications in Analysis and Mechanics, 2024, 16(4): 858-871. doi: 10.3934/cam.2024037 |
[3] | Xiulan Wu, Yaxin Zhao, Xiaoxin Yang . On a singular parabolic p-Laplacian equation with logarithmic nonlinearity. Communications in Analysis and Mechanics, 2024, 16(3): 528-553. doi: 10.3934/cam.2024025 |
[4] | Erlend Grong, Irina Markina . Harmonic maps into sub-Riemannian Lie groups. Communications in Analysis and Mechanics, 2023, 15(3): 515-532. doi: 10.3934/cam.2023025 |
[5] | Leandro Tavares . Solutions for a class of problems driven by an anisotropic (p,q)-Laplacian type operator. Communications in Analysis and Mechanics, 2023, 15(3): 533-550. doi: 10.3934/cam.2023026 |
[6] | Velimir Jurdjevic . Time optimal problems on Lie groups and applications to quantum control. Communications in Analysis and Mechanics, 2024, 16(2): 345-387. doi: 10.3934/cam.2024017 |
[7] | Zhiyong Wang, Kai Zhao, Pengtao Li, Yu Liu . Boundedness of square functions related with fractional Schrödinger semigroups on stratified Lie groups. Communications in Analysis and Mechanics, 2023, 15(3): 410-435. doi: 10.3934/cam.2023020 |
[8] | Ming Liu, Binhua Feng . Grand weighted variable Herz-Morrey spaces estimate for some operators. Communications in Analysis and Mechanics, 2025, 17(1): 290-316. doi: 10.3934/cam.2025012 |
[9] | Xiaotian Hao, Lingzhong Zeng . Eigenvalues of the bi-Xin-Laplacian on complete Riemannian manifolds. Communications in Analysis and Mechanics, 2023, 15(2): 162-176. doi: 10.3934/cam.2023009 |
[10] | Eleonora Amoroso, Angela Sciammetta, Patrick Winkert . Anisotropic (→p,→q)-Laplacian problems with superlinear nonlinearities. Communications in Analysis and Mechanics, 2024, 16(1): 1-23. doi: 10.3934/cam.2024001 |
The fundamental work of Hansen and Spies [4] modeled a two-layer beam with a structural damping due to the interfacial slip through the following system
{ρφtt+G(ψ−φx)x=0,Iρ(3w−ψ)tt−D(3w−ψ)xx−G(ψ−φx)=0,Iρwtt−Dwxx+3G(ψ−φx)+4γw+4βwt=0, | (1.1) |
where φ=φ(x,t) is the transverse displacement, ψ=ψ(x,t) is the rotation angle, w=w(x,t) is proportional to the amount of slip along the interface, 3w−ψ denotes the effective rotation angle. The physical quantities ρ,Iρ,G,D,β and γ are respectively: the density, mass moment of inertia, shear stiffness, flexural rigidity, adhesive damping and adhesive stiffness. Equation (1.1)3 describes the dynamics of the slip. For β=0, system (1.1) describes the coupled laminated beams without structural damping at the interface. In the recent result [1], Apalara considered the thermoelastic-laminated beam system without structural damping, namely
{ρφtt+G(ψ−φx)x=0,Iρ(3s−ψ)tt−D(3s−ψ)xx−G(ψ−φx)=0,Iρstt−Dsxx+3G(ψ−φx)+4γs+δθx=0,ρ3θt−λθxx+δstx=0, | (1.2) |
where (x,t)∈(0,1)×(0,+∞), θ=θ(x,t) is the difference temperature. The positive quantities γ,β,k,λ are adhesive stiffness, adhesive damping, heat capacity and the diffusivity respectively. The author proved that (1.2) is exponential stable provided
Gρ=DIρ. | (1.3) |
When β>0, the adhesion at the interface supplies a restoring force proportion to the interfacial slip. But this is not enough to stabilize system (1.1), see for instance [2]. To achieve exponential or general stabilization of system (1.1), many authors in literature have used additional damping. In this direction, Gang et al. [9] studied the following memory-type laminated beam system
{ρφtt+G(ψ−φx)x=0,Iρ(3w−ψ)tt−D(3w−ψ)xx+∫t0g(t−s)(3w−ψ)xx(x,s)ds−G(ψ−φx)=03Iρwtt−3Dwxx+3G(ψ−φx)+4γw+4βwt=0 | (1.4) |
and established a general decay result for more regular solutions and Gρ≠DIρ. Mustafa [15] also considered the structural damped laminated beam system (1.4) and established a general decay result provided Gρ=DIρ. Feng et al. [8] investigated the following laminated beam system
{ρwtt+Gφx+g1(wt)+f1(w,ξ,s)=h1,Iρξtt−Gφ−Dξxx+g2(ξt)+f2(w,ξ,s)=h2,Iρstt+Gφ−Dsxx+g3(st)+f2(w,ξ,s)=h3 | (1.5) |
and established the well-posedness, smooth global attractor of finite fractal dimension as well as existence of generalized exponential attractors. See also, recent results by Enyi et al. [20]. We refer the reader to [5,6,7,11,13,14,17,18] and the references cited therein for more related results.
In this present paper, we consider a thermoelastic laminated beam problem with a viscoelastic damping
{ρwtt+G(ψ−wx)x=0,Iρ(3s−ψ)tt−D(3s−ψ)xx+∫t0g(t−τ)(3s−ψ)xx(x,τ)dτ−G(ψ−wx)=03Iρstt−3Dsxx+3G(ψ−wx)+4γs+δθx=0,kθt−λθxx+δsxt=0 | (1.6) |
under initial conditions
{w(x,0)=w0(x), ψ(x,0)=ψ0(x), s(x,0)=s0(x), θ(x,0)=θ0(x), x∈[0,1],wt(x,0)=w1(x), ψt(x,0)=ψ1(x), st(x,0)=s1(x), x∈[0,1] | (1.7) |
and boundary conditions
{w(0,t)=ψx(0,t)=sx(0,t)=θ(0,t)=0,t∈[0,+∞),wx(1,t)=ψ(1,t)=s(1,t)=θx(1,t)=0,t∈[0,+∞). | (1.8) |
In the system (1.6), the integral represents the viscoelastic damping, and g is the relaxation function satisfying some suitable assumptions specified in the next section. According to the Boltzmann Principle, the viscoelastic damping (see [21] for details) is represented by a memory term in the form of convolution. It acts as a damper to reduce the internal/external forces like the beam's weight, heavy loads, wind, etc., that cause undesirable vibrations.
In most of the above works, the authors have established their decay result by including the structural damping along with other dampings. So, the natural question that comes to mind.
Is it possible to obtain general/optimal decay result (decay rates that agrees with that of g) to the thermoelastic laminated beam system (1.6)–(1.8), in the absence of the structural damping.
The novelty of this article is to answer this question in a consenting way, by using the ideas developed in [10] to establish general and optimal decay results for Problem 1.6. Moreover, we establish a weaker decay result in the case of a non-equal wave of speed propagation. To the best of our knowledge, there is no stability result for the latter in the literature.
The rest of work is organized as follows: In Section 2, we recall some preliminaries and assumptions on the memory term. In Section 3, we state and prove the main stability result for the case equal-speed and in the case of non-equal-speed of propagation. We also give some examples to illustrate our findings. Finally, in Section 4, we give the proofs of the lemmas used our main results.
In this section, we recall some useful materials and conditions. Through out this paper, C is a positive constant that may change through lines, ⟨.,.⟩ and ‖.‖2 denote respectively the inner product and the norm in L2(0,1). We assume the relaxation function g obeys the assumptions:
(G1). g:[0,+∞)⟶(0,+∞) is a non-inecreasing C1− function such that
g(0)>0,D−∫∞0g(τ)dτ=l0>0. | (2.1) |
(G2). There exist a C1 function H:[0,+∞)→(0,+∞) which is linear or is strictly convex C2 function on (0,ϵ0), ϵ0≤g(0), with H(0)=H′(0)=0 and a positive nonincreasing differentiable function ξ:[0,+∞)→(0,+∞), such that
g′(t)≤−ξ(t)H(g(t)),t≥0, | (2.2) |
Remark 2.1. As in [10], we note here that, if H is a strictly increasing convex C2− function on (0,r], with H(0)=H′(0)=0, then H has an extension ˉH, which is strictly increasing and strictly convex C2-function on (0,+∞). For example, ˉH can be defined by
ˉH(s)=H″(r)2s2+(H′(r)−H″(r)r)s+H(r)−H′(r)r+H″(r)2r2, s>r. | (2.3) |
Let
H1∗(0,1)={u∈H1(0,1)/u(0)=0}, ˉH1∗(0,1)={u∈H1(0,1)/u(1)=0}, |
H2∗(0,1)={u∈H2(0,1)/ux∈H1∗(0,1)}, ˉH2∗(0,1)={u∈H2(0,1)/ux∈ˉH1∗(0,1)}. |
The existence and regularity result of problem (1.6) is the following
Theorem 2.1. Let (w0,3s0−ψ0,s0,θ0)∈H1∗(0,1)×ˉH1∗(0,1)×ˉH1∗(0,1)×H1∗(0,1) and (w1,3s1−ψ1,s1)∈L2(0,1)×L2(0,1)×L2(0,1) be given. Suppose (G1) and (G2) hold. Then problem (1.6) has a unique global weak solution (w,3s−ψ,s,θ) which satisfies
w∈C(R+,H1∗(0,1))∩C1(R+,L2(0,1)), (3s−ψ)∈C(R+,ˉH1∗(0,1))∩C1(R+,L2(0,1)), |
s∈C(R+,ˉH1∗(0,1))∩C1(R+,L2(0,1)), θ∈C(R+,L2(0,1))∩L2(R+,H1(0,1)). |
Furthermore, if (w0,(3s0−ψ0),s0,θ0)∈H2∗(0,1)×ˉH2∗(0,1)×ˉH2∗(0,1)×H2(0,1)∩H1∗(0,1) and (w1,(3s1−ψ1),s1)∈H1∗(0,1)×ˉH1∗(0,1)×ˉH1∗(0,1), then the solution of (1.6) satisfies
w∈C(R+,H2∗(0,1))∩C1(R+,H1∗(0,1))∩C2(R+,L2(0,1)), |
(3s−ψ)∈C(R+,ˉH2∗(0,1))∩C1(R+,ˉH1∗(0,1))∩C2(R+,L2(0,1)), |
s∈C(R+,ˉH2∗(0,1))∩C1(R+,ˉH1∗(0,1))∩C2(R+,L2(0,1)), |
θ∈C(R+,H2(0,1)∩H2∗(0,1))∩C1(R+,H1∗(0,1)). |
The proof of Theorem 2.1 can be established using the Galerkin approximation method as in [16]. Throughout this paper, we denote by ⋄ the binary operator, defined by
(g⋄ν)(t)=∫t0g(t−τ)‖ν(t)−ν(τ)‖22dτ,t≥0. |
We also define h(t) and Cα as follow
h(t)=αg(t)−g′(t) and Cα=∫+∞0g2(τ)αg(τ)−g′(τ)dτ. |
The following lemmas will be applied repeatedly throughout this paper
Lemma 2.1. For any function f∈L2loc([0,+∞),L2(0,1)), we have
∫10(∫t0g(t−s)(f(t)−f(s))ds)2dx≤(1−l0)(g⋄f)(t), | (2.4) |
∫10(∫x0f(y,t)dy)2dx≤‖f(t)‖22. | (2.5) |
Lemma 2.2. Let v∈H1∗(0,1) or ˉH1∗(0,1), we have
∫10(∫t0g(t−s)(v(t)−v(τ))dτ)2dx≤Cp(1−l0)(g⋄v)(t), | (2.6) |
where Cp>0 is the poincaré constant.
Lemma 2.3. Let (w,3s−ψ,s,θ) be the solution of (1.6). Then, for any 0<α<1 we have
∫10(∫t0g(t−τ)((3s−ψ)x(τ)−(3s−ψ)x(t))dτ)2dx≤Cα(h⋄(3s−ψ)x)(t). | (2.7) |
Proof. Using Cauchy-Schwarz inequality, we have
∫10(∫t0g(t−τ)((3s−ψ)x(τ)−(3s−ψ)x(t))dτ)2dx=∫10(∫t0g(t−τ)√h(t−τ)√h(t−τ)((3s−ψ)x(τ)−(3s−ψ)x(t))dτ)2dx≤(∫+∞0g2(τ)h(τ)ds)∫10∫t0h(t−τ)((3s−ψ)x(τ)−(3s−ψ)x(t))2dτdx=Cα(h⋄(3s−ψ)x)(t). | (2.8) |
Lemma 2.4. [12] Let F be a convex function on the close interval [a,b], f,j:Ω→[a,b] be integrable functions on Ω, such that j(x)≥0 and ∫Ωj(x)dx=α1>0. Then, we have the following Jensen inequality
F(1α1∫Ωf(y)j(y)dy)≤1α1∫ΩF(f(y))j(y)dy. | (2.9) |
In particular if F(y)=y1p, y≥0, p>1, then
(1α1∫Ωf(y)j(y)dy)1p≤1α1∫Ω(f(y))1pj(y)dy. | (2.10) |
Lemma 2.5. The energy functional E(t) of the system (1.6)-(1.8) defined by
E(t)=12[ρ‖wt‖22+3Iρ‖st‖22+Iρ‖3st−ψt‖22+3D‖sx‖22+G‖ψ−wx‖22]+12[(D−∫t0g(τ)dτ)‖3sx−ψx‖22+(g⋄(3sx−ψx))(t)+4γ‖s‖22+k‖θ‖22], | (2.11) |
satisfies
E′(t)=12(g′⋄(3sx−ψx))(t)−12g(t)‖3sx−ψx‖22−λ‖θx‖22≤12(g′⋄(3sx−ψx))(t)≤0, ∀ t≥0. | (2.12) |
Proof. Multiplying (1.6)1, (1.6)2, (1.6)3 and (1.6)4, respectively, by wt, (3st−ψt), st and θ, integrating over (0,1), and using integration by parts and the boundary conditions (1.7), we arrive at
12ddt(ρ‖wt‖22+G‖ψ−wx‖22)=G⟨(ψ−wx),ψt⟩, | (2.13) |
12ddt[Iρ‖3st−ψt‖22+(D−∫t0g(τ)dτ)‖3sx−ψx‖22+(g⋄(3sx−ψx))(t)]=G⟨(ψ−wx),(3s−ψ)t⟩+12(g′⋄(3sx−ψx))(t)−12g(t)‖3sx−ψx‖22, | (2.14) |
12ddt[3Iρ‖st‖22+3D‖sx‖22+4γ‖s‖22]=−3G⟨(ψ−wx),st⟩−δ⟨θx,st⟩, | (2.15) |
and
12ddt(k‖θ‖22)=−λ‖θx‖22+δ⟨θx,st⟩. | (2.16) |
Adding the equations (2.13)–(2.16), taking into account (G1) and (G2), we obtain (2.12) for regular solutions. The result remains valid for weak solutions by a density argument. This implies the energy functional is non-increasing and
E(t)≤E(0), ∀t≥0. |
This section is subdivided into two. In the first subsection, we prove the stability result for equal-wave-speed of propagation, whereas in the second subsection, we focus on the stability result for non-equal-wave-speed of propagation.
Our aim, in this subsection, is to prove an explicit, general and optimal decay rate of solutions for system (1.6)–(1.8). To achieve this, we define a Lyapunov functional
L(t)=NE(t)+6∑j=1NjIj(t), | (3.1) |
where N, Nj, j=1,2,3,4,5,6 are positive constants to be specified later and
I1(t)=−Iρ∫10(3s−ψ)t∫t0g(t−τ)((3s−ψ)(t)−(3s−ψ)(τ))dτdx,t≥0, |
I2(t)=3Iρ∫10sstdx+3ρ∫10wt∫x0s(y)dydx,I3(t)=−3kIρ∫10θ∫x0st(y)dydx,t≥0, |
I4(t)=−ρ∫10wtwdx,I5(t)=Iρ∫10(3s−ψ)(3s−ψ)tdx,t≥0, |
I6(t)=3IρG∫10(ψ−wx)stdx−3ρD∫10wtsxdx,I7(t)=∫10∫t0J(t−τ)(3sx−ψx)2(τ)dτdx,t≥0, |
where
J(t)=∫+∞tg(τ)dτ. |
The following lemma is very important in the proof of our stability result.
Lemma 3.1. Suppose Gρ=DIρ. Under suitable choice of t0,N, Nj, j=1,2,3,4,5,6, the Lyapunov functional L satisfies, along the solution of (1.6)−(1.8), the estimate
L′(t)≤−β(‖wt‖22+‖st‖22+‖3st−ψt‖22+‖sx‖22+‖3wx−ψx‖22+‖ψ−wx‖22)−β(‖s‖22+‖θx‖22)+12(g⋄(3sx−ψx))(t),∀ t≥t0 | (3.2) |
and the equivalence relation
α1E(t)≤L(t)≤α2E(t) | (3.3) |
holds for some β>0, α1, α2>0.
Proof. By virtue of assumption (3.1) and using h(t)=αg(t)−g′(t), it follows from Lemmas 2.5, 4.1-4.6 (see the Appendix for detailed derivations) that, for all t≥t0>0,
L′(t)≤−[N4ρ−N2δ4]‖wt‖22−[N3δIρ2−N2C(1+1ϵ2)−N6C(1+1ϵ1)]‖st‖22−3N2γ‖s‖22−[N1Iρg0−N5Iρ−N6ϵ1]‖3st−ψt‖22−[3DN2−N3ϵ3−N4C−N6C]‖sx‖22−[N6G2−N1ϵ2−N3ϵ3−N4Cϵ4−N5C]‖ψ−wx‖22−[N5l04−N1ϵ1−N4ϵ4]‖3sx−ψx‖22−[λN−N2C−N3C(1+1ϵ3)−N6C]‖θx‖22+Nα2(g⋄(3sx−ψx))(t)−[N2−CCα(N5+N1(1+1ϵ1+1ϵ2))](h⋄(3sx−ψx))(t). | (3.4) |
Now, we choose
N4=N5=1, ϵ4=l08 | (3.5) |
and select N1 large enough such that
μ1:=N1Iρg0−Iρ>0. | (3.6) |
Next, we choose N6 large so that
μ2:=N6G2−C>0. | (3.7) |
Also, we select N2 large enough so that
μ3:=3DN2−C−N6C>0. | (3.8) |
After fixing N1,N2,N6, and letting ϵ3=μ12N3, we then select ϵ1,ϵ2, and δ4 very small such that
ρ−N2δ4>0, μ1−N6ϵ1>0, μ4:=μ22−N1ϵ2>0 | (3.9) |
and select N3 large enough so that
N3δIρ2−N2C(1+1ϵ2)−N6C(1+1ϵ1)>0. | (3.10) |
Now, we note that αg2(s)h(s)=αg2(s)αg(s)−g′(s)<g(s); thus the dominated convergence theorem gives
αCα=∫+∞0αg2(s)αg(s)−g′(s)ds→0 as α→0. | (3.11) |
Therefore, we can choose some 0<α0<1 such that for all 0<α≤α0,
αCα<14C(1+N1(1+1ϵ1+1ϵ2)). | (3.12) |
Finally, we select N so large enough and take α=1N So that
λN−N2C−N3C(1+1ϵ3)−N6C>0,N2−CCα(1+N1(1+1ϵ1+1ϵ2))>0. | (3.13) |
Combination of (3.6) - (3.13) yields the estimate (3.2). The equivalent relation (3.3) can be obtain easily by using Young's, Cauchy-Schwarz, and Poincaré's inequalities.
Now, we state and prove our stability result for this subsection.
Theorem 3.1. Assume Gρ=DIρ and (G1) and (G2) hold. Then, there exist positive constants a1 and a2 such that the energy solution (2.11) satisfies
E(t)≤a2H−11(a1∫tt0ξ(τ)dτ), where H1(t)=∫rt1τH′(τ)dτ | (3.14) |
and H1 is a strictly decreasing and strictly convex function on (0,r], with limt→0H1(t)=+∞.
Proof. Using the fact that g and ξ are positive, non-increasing and continuous, and H is positive and continuous, we have that for all t∈[0,t0]
0<g(t0)≤g(t)≤g(0), 0<ξ(t0)≤ξ(t)≤ξ(0). |
Thus for some constants a,b>0, we obtain
a≤ξ(t)H(g(t))≤b. |
Therefore, for any t∈[0,t0], we get
g′(t)≤−ξ(t)H(g(t))≤−ag(0)g(0)≤−ag(0)g(t) | (3.15) |
and
ξ(t)g(t)≤−g(0)ag′(t). | (3.16) |
From (2.12) and (3.15), it follows that
∫t00g(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≤−g(0)a∫t00g′(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≤−CE′(t), ∀t≥t0. | (3.17) |
From (3.2) and (3.17), we have
L′(t)≤−βE(t)+12(g⋄(3sx−ψx))(t)=−βE(t)+12∫t00g(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ+12∫tt0g(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≤−βE(t)−CE′(t)+12∫tt0g(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ. |
Thus, we get
L′1(t)≤−βE(t)+12∫tt0g(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ, ∀t≥t0, | (3.18) |
where L1=L+CE∼E by virtue of (3.3). To finish our proof, we distinct two cases:
Case 1: H(t) is linear. In this case, we multiply (3.18) by ξ(t), keeping in mind (2.12) and (G2), to get
ξ(t)L′1(t)≤−βξ(t)E(t)+12ξ(t)∫tt0g(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≤−βξ(t)E(t)+12∫tt0ξ(τ)g(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≤−βξ(t)E(t)−12∫tt0g′(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≤−βξ(t)E(t)−CE′(t), ∀ t≥t0. | (3.19) |
Therefore
(ξL1+CE)′(t)≤−βξ(t)E(t), ∀ t≥t0. | (3.20) |
Since ξ is non-increasing and L1∼E, we have
L2=ξL1+CE∼E. | (3.21) |
Thus, from (3.20), we get for some positive constant α
L′2(t)≤−βξ(t)E(t)≤−αξ(t)L2(t), ∀ t≥t0. | (3.22) |
Integrating (3.22) over (t0,t) and recalling (3.21), we obtain
E(t)≤a1e−a2∫tt0ξ(s)ds=a1H−11(a2∫tt0ξ(s)ds). |
Case 2: H(t) is nonlinear. In this case, we consider the functional L(t)=L(t)+I7(t). From (3.2) and Lemma 4.7 (see the Appendix), we obtain
L′(t)≤−dE(t), ∀t≥t0, | (3.23) |
where d>0 is a positive constant. Therefore,
d∫tt0E(s)ds≤L(t0)−L(t)≤L(t0). |
Hence, we get
∫+∞0E(s)ds<∞. | (3.24) |
Using (3.24), we define p(t) by
p(t):=η∫tt0‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ, |
where 0<η<1 so that
p(t)<1,∀t≥t0. | (3.25) |
Moreover, we can assume p(t)>0 for all t≥t0; otherwise using (3.18), we obtain an exponential decay rate. We also define q(t) by
q(t)=−∫tt0g′(τ)‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ. |
Then q(t)≤−CE′(t), ∀t≥t0. Now, we have that H is strictly convex on (0,r] (where r=g(t0)) and H(0)=0. Thus,
H(στ)≤σH(τ), 0≤σ≤1 and τ∈(0,r]. | (3.26) |
Using (3.26), condition (G2), (3.25), and Jensen's inequality, we get
q(t)=1ηp(t)∫tt0p(t)(−g′(τ))η‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≥1ηp(t)∫tt0p(t)ξ(τ)H(g(τ))η‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≥ξ(t)ηp(t)∫tt0H(p(t)g(τ))η‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≥ξ(t)ηH(η∫tt0g(τ)η‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ)=ξ(t)ηˉH(η∫tt0g(τ)η‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ), | (3.27) |
where ˉH is the convex extention of H on (0,+∞) (see remark 2.1). From (3.27), we have
∫tt0g(τ)η‖(3sx−ψx)(t)−(3sx−ψx)(t−τ)‖22dτ≤1ηˉH−1(ηq(t)ξ(t)). |
Therefore, (3.18) yields
L′1(t)≤−βE(t)+CˉH−1(ηq(t)ξ(t)), ∀ t≥t0. | (3.28) |
For r0<r, we define L3(t) by
L3(t):=ˉH′(r0E(t)E(0))L1(t)+E(t)∼E(t) |
since L1∼E. From (3.28) and using the fact that
E′(t)≤0, ˉH′(t)>0, ˉH″(t)>0, |
we obtain for all t≥t0
L′3(t)=r0E′(t)E(0)ˉH″(r0E(t)E(0))L1(t)+ˉH′(r0E(t)E(0))L′1(t)+E′(t)≤−βE(t)ˉH′(r0E(t)E(0))+CˉH′(r0E(t)E(0))ˉH−1(ηq(t)ξ(t))+E′(t). | (3.29) |
Let us consider the convex conjugate of ˉH denoted by ˉH∗ in the sense of Young (see [3] page 61-64). Thus,
ˉH∗(τ)=τ(ˉH′)−1(τ)−ˉH[(ˉH′)(τ)] | (3.30) |
and ˉH∗ satisfies the generalized Young inequality
AB≤ˉH∗(A)+ˉH(B). | (3.31) |
Let A=ˉH′(r0E(t)E(0)) and B=ˉH−1(μz(t)ξ(t)), It follows from (2.12) and (3.29)-(3.31) that
L′3(t)≤−βE(t)ˉH′(r0E(t)E(0))+CˉH∗(ˉH′(r0E(t)E(0)))+Cηq(t)ξ(t)+E′(t)≤−βE(t)ˉH′(r0E(t)E(0))+Cr0E(t)E(0)ˉH′(r0E(t)E(0))+Cηq(t)ξ(t)+E′(t). | (3.32) |
Next, we multiply (3.32) by ξ(t) and recall that r0E(t)E(0)<r and
ˉH′(r0E(t)E(0))=H′(r0E(t)E(0)), |
we arrive at
ξ(t)L′3(t)≤−βξ(t)E(t)H′(r0E(t)E(0))+Cr0E(t)E(0)ξ(t)H′(r0E(t)E(0))+Cηq(t)+ξ(t)E′(t)≤−βξ(t)E(t)H′(r0E(t)E(0))+Cr0E(t)E(0)ξ(t)H′(r0E(t)E(0))−CE′(t). | (3.33) |
Let L4(t)=ξ(t)L3(t)+CE(t). Since L3∼E, it follows that
b0L4(t)≤E(t)≤b1L4(t), | (3.34) |
for some b0,b1>0. Thus (3.33) gives
L′4(t)≤−(βE(0)−Cr0)ξ(t)E(t)E(0)ξ(t)H′(r0E(t)E(0)), ∀t≥t0. |
We select r0<r small enough so that βE(0)−Cr0>0, we get
L′4(t)≤−mξ(t)E(t)E(0)ξ(t)H′(r0E(t)E(0))=−mξ(t)H2(E(t)E(0)), ∀t≥t0, | (3.35) |
for some constant m>0 and H2(τ)=τH′(r0τ). We note here that
H′2(τ)=H′(r0τ)+r0tH″(r0τ), |
thus the strict convexity of H on (0,r], yields H2(τ)>0,H′2(τ)>0 on (0,r]. Let
F(t)=b0L4(t)E(0). |
From (3.34) and (3.35), we obtain
F(t)∼E(t) | (3.36) |
and
F′(t)=a0L′4(t)(t)E(0)≤−m1ξ(t)H2(F(t)), ∀t≥t0. | (3.37) |
Integrating (3.37) over (t0,t), we arrive at
m1∫tt0ξ(τ)dτ≤−∫tt0F′(τ)H2(F(τ))dτ=1r0∫r0F(t0)r0F(t)1τH′(τ)dτ. | (3.38) |
This implies
F(t)≤1r0H−11(¯m1∫tt0ξ(τ)dτ), where H1(t)=∫rt1τH′(τ)dτ. | (3.39) |
Using the properties of H, we see easily that H1 is strictly decreasing function on (0,r] and
limt⟶0H1(t)=+∞. |
Hence, (3.14) follows from (3.36) and (3.39). This completes the proof.
Remark 3.1. The stability result in (3.1) is general and optimal in the sense that it agrees with the decay rate of g, see [10], Remark 2.3.
Corollary 3.2. Suppose Gρ=DIρ, and (G1), and (G2) hold. Let the function H in (G2) be defined by
H(τ)=τp, 1≤p<2, | (3.40) |
then the solution energy (2.11) satisfies
E(t)≤a2exp(−a1∫t0ξ(τ)dτ), for p=1,E(t)≤C(1+∫tt0ξ(τ)dτ)1p−1, for 1<p<2 | (3.41) |
for some positive constants a2,a1 and C.
In this subsection, we establish another stability result in the case non-equal speeds of wave propagation. To achieve this, we consider a stronger solution of (1.6). Let (w,3s−ψ,s,θ) be the strong solution of problem (1.6)–(1.8), then differentiation of 1.6 with respect to t gives
{ρwttt+G(ψ−wx)xt=0,Iρ(3s−ψ)ttt−D(3s−ψ)xxt+∫t0g(τ)(3s−ψ)xxt(x,t−τ)dτ+g(t)(3s0−ψ0)xx−G(ψ−wx)t=03Iρsttt−3Dsxxt+3G(ψ−wx)t+4γst+δθxt=0,kθtt−λθxxt+δsxtt=0, | (3.42) |
where (x,t)∈(0,1)×(0,+∞) and (3s−ψ)xx(x,0)=(3s0−ψ0)xx. The modified energy functional associated to (3.42) is defined by
E1(t)=12[ρ‖wtt‖22+3Iρ‖stt‖22+Iρ‖3stt−ψtt‖22+3D‖sxt‖22+G‖ψt−wxt‖22]+12[4γ‖st‖22+k‖θt‖22+(D−∫t0g(τ)dτ)‖3sxt−ψxt‖22+(g⋄(3sxt−ψxt))(t)]. | (3.43) |
Lemma 3.2. Let (w,3s−ψ,s,θ) be the strong solution of problem (1.6)-(1.8). Then, the energy functional (3.43) satisfies, for all t≥0
E′1(t)=12(g′⋄(3sxt−ψxt))(t)−12g(t)‖3sxt−ψxt‖22−g(t)⟨(3stt−ψtt),(3s0−ψ0)xx⟩−λ‖θxt‖22 | (3.44) |
and
E1(t)≤C(E1(0)+‖(3s0−ψ0)xx‖22). | (3.45) |
Proof. The proof of (3.44) follows the same steps as in the proof of Lemma 2.5. From (3.44), it is obvious that
E′1(t)≤−g(t)⟨(3stt−ψtt),(3s0−ψ0)xx⟩. |
So, using Cauchy-Schwarz inequality, we obtain
E′1(t)≤Iρg(t)2‖3stt−ψtt‖22+g(t)2Iρ‖(3s0−ψ0)xx‖22≤g(t)E1(t)+g(t)2Iρ‖(3s0−ψ0)xx‖22. | (3.46) |
This implies
ddt(E1(t)e−∫t0g(τ)dτ)≤e−∫t0g(τ)dτg(t)2Iρ‖(3s0−ψ0)xx‖22≤g(t)2Iρ‖(3s0−ψ0)xx‖22 | (3.47) |
Integrating (3.47) over (0,t) yields
E1(t)e−∫+∞0g(τ)dτ≤E1(t)e−∫t0g(τ)dτ≤E1(0)+12Iρ(∫t0g(τ)dτ)‖(3s0−ψ0)xx‖22≤E1(0)+12Iρ(∫+∞0g(τ)dτ)‖(3s0−ψ0)xx‖22. | (3.48) |
Hence, (3.45) follows.
Remark 3.2. Using Young's inequality, we observe from (3.44) and (3.45) that
λ‖θxt‖22=−E′1(t)+12(g′⋄(3sxt−ψxt))(t)−12g(t)‖3sxt−ψxt‖22−g(t)⟨(3stt−ψtt),(3s0−ψ0)xx⟩≤−E′1(t)−g(t)⟨(3stt−ψtt),(3s0−ψ0)xx⟩≤−E′1(t)+g(t)(‖3stt−ψtt‖22+‖(3s0−ψ0)xx‖22)≤−E′1(t)+g(t)(2IρE1(t)+‖(3s0−ψ0)xx‖22)≤C(−E′1(t)+c1g(t)) | (3.49) |
for some fixed positive constant c1. Similarly, we obtain
0≤−(g′⋄(3sxt−ψxt))(t)≤C(−E′1(t)+c1g(t)). | (3.50) |
As in the case of equal-wave-speed of propagation, we define a Lyapunov functional
˜L(t)=˜NE(t)+6∑j=1~NjIj(t)+~N6I8(t), | (3.51) |
where ˜N, ~Nj, j=1,2,3,4,5,6, are positive constants to be specified later and
I8(t)=3λδ(IρG−ρD)∫10θxwxdx. |
Lemma 3.3. Suppose Gρ≠DIρ. Then, under suitable choice of ˜N, ~Nj, j=1,2,3,4,5,6, the Lyapunov functional ˜L satisfies, along the solution of (1.6), the estimate
˜L′(t)≤−˜βE(t)+12(g⋄(3sx−ψx))(t)+C(−E′1(t)+c1g(t)),∀ t≥t0, | (3.52) |
for some positive constants ˜β and c1.
Proof. Following the proof of Lemma 3.1, we end up with (3.52).
Lemma 3.4. Suppose assumptions (G1) and (G2) hold and the function H in (G2) is linear. Let (w,3s−ψ,s,θ) be the strong solution of problem (1.6)-(1.8). Then,
ξ(t)(g⋄(3sxt−ψxt))(t)≤C(−E′1(t)+c1g(t)), ∀ t≥0, | (3.53) |
where c1 is a fixed positive constant.
Proof. Using (3.50) and the fact that ξ is decreasing, we have
ξ(t)(g⋄(3sxt−ψxt))(t)=ξ(t)∫t0g(t−τ)(‖(3sxt−ψxt)(t)−(3sxt−ψxt)(τ)‖22)dτ≤∫t0ξ(t−τ)g(t−τ)(‖(3sxt−ψxt)(t)−(3sxt−ψxt)(τ)‖22)dτ≤−∫t0g′(t−τ)(‖(3sxt−ψxt)(t)−(3sxt−ψxt)(τ)‖22)dτ=−(g′⋄(3sxt−ψxt))(t)≤C(−E′1(t)+c1g(t)). | (3.54) |
Our stability result of this subsection is
Theorem 3.3. Assume (G1) and (G2) hold and Gρ≠DIρ. Then, there exist positive constants a1,a2 and t2>t0 such that the energy solution (2.11) satisfies
E(t)≤a2(t−t0)H−12(a1(t−t0)∫tt2ξ(τ)dτ),∀t>t2, where H2(τ)=τH′(τ). | (3.55) |
Proof. Case 1: H is linear. Multiplying (3.52) by ξ(t) and using (G1), we get
ξ(t)˜L′(t)≤−˜βξ(t)E(t)+12ξ(t)(g⋄(3sx−ψx))(t)+Cξ(t)(−E′1(t)+c1g(t))≤−˜βξ(t)E(t)−CE′(t)−Cξ(0)E′1(t)+ξ(0)c1g(t), ∀ t≥t0 |
Using the fact that ξ non-increasing, we obtain
(ξ˜L+CE+E1)′(t)≤−˜βξ(t)E(t)+c2g(t), ∀ t≥t0. |
for some fixed positive constant c_2 . This implies
\begin{equation} \tilde{\beta} \xi(t) E(t)\leq - \left(\xi\tilde{L} + CE+ E_1 \right)'(t) +c_2 g(t), \ \forall \ t\geq t_0. \end{equation} | (3.56) |
Integrating (3.56) over (t_0, t) , using the fact that E is non-increasing and the inequality (3.45), we arrive at
\begin{equation} \begin{aligned} \tilde{\beta} E(t)\int_{t_0}^t\xi(\tau)d\tau &\leq \tilde{\beta}\int_{t_0}^t \xi(\tau) E(\tau)d\tau\\ &\leq -\left(\xi\tilde{L} + CE+ E_1 \right)(t)+\left(\xi\tilde{L} + CE+ E_1 \right)(t_0)+ c_2\int_{t_0}^tg(\tau)d\tau\\ &\leq \left(\xi\tilde{L} + CE+ E_1 \right)(0)+ C\|(3s_0-\psi_0)_{xx}\|_2^2 + c_2\int_{0}^{\infty}g(\tau)d\tau\\ & = \left(\xi\tilde{L} + CE+ E_1 \right)(0)+ C\|(3s_0-\psi_0)_{xx}\|_2^2 +c_2(D-l_0). \end{aligned} \end{equation} | (3.57) |
Thus, we have
\begin{equation} E(t)\leq \frac{C}{ \int_{t_0}^t\xi(\tau)d\tau} , \ \forall \ t\geq t_0. \end{equation} | (3.58) |
Case II: H is nonlinear. First, we observe from (3.52) that
\begin{equation} \begin{aligned} \tilde{L}'(t)\leq &-\tilde{\beta} E(t)+ \frac{1}{2}\left( g\diamond (3s_x -\psi_x)\right)(t)+ C \left( -E'_1(t) + c_1g(t)\right)\\ \leq & -\tilde{\beta} E(t)+ C\left( \left( g\diamond (3s_x -\psi_x)\right)(t) +\left( g\diamond (3s_{xt} -\psi_{xt})\right)(t)\right)+ C \left( -E'_1(t) + c_1g(t)\right), \ \forall \ t\geq t_0. \end{aligned} \end{equation} | (3.59) |
From (2.12), (3.16) and (3.50), we have for any t\geq t_0
\begin{equation} \begin{aligned} \int_0^{t_0}g(\tau)\|(3s_x -&\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2d\tau+\int_0^{t_0}g(\tau)\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2d\tau \\ & \quad \leq \frac{1}{\xi(t_0)}\int_0^{t_0}\xi(\tau)g(\tau)\|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2d\tau\\ & \quad +\frac{1}{\xi(t_0)}\int_0^{t_0}\xi(\tau)g(\tau)\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2d\tau\\ & \quad \leq - \frac{g(0)}{a\xi(t_0)}\int_0^{t_0}g'(\tau)\|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2d\tau\\ & \quad -\frac{g(0)}{a\xi(t_0)}\int_0^{t_0}g'(\tau)\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2d\tau\\ & \quad \leq-C\left(E'(t)+ E'_1(t) \right) +c_2 g(t), \end{aligned} \end{equation} | (3.60) |
where c_2 is a fixed positive constant. Substituting (3.60) into (3.59), we obtain for any t\geq t_0
\begin{equation} \begin{aligned} \tilde{L}'(t)\leq &-\tilde{\beta} E(t)-C\left(E'(t)+ E'_1(t) \right) + c_3g(t)+ C \int_{t_0}^{t}g(\tau) \|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2 d\tau\\ & +C \int_{t_0}^{t}g(\tau)\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 d\tau, \end{aligned} \end{equation} | (3.61) |
where c_3 is a fixed positive constant. Now, we define the functional \Phi by
\begin{equation} \begin{aligned} \Phi(t) = &\frac{\sigma}{t-t_0}\int_{t_0}^{t}\|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2 d\tau\\ &+\frac{\sigma}{t-t_0} \int_{t_0}^{t}\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 d\tau, \ \forall \ t \gt t_0. \end{aligned} \end{equation} | (3.62) |
Using (2.11), (2.12), (3.43) and (3.45), we easily get
\begin{equation} \begin{aligned} \frac{1}{t-t_0}\int_{t_0}^{t} \|(3s_x -\psi_x)(t)-&(3s_x -\psi_x)(t-\tau)\|_2^2 d\tau+\frac{1}{t-t_0} \int_{t_0}^{t}\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 d\tau\\ &\leq \frac{2}{t-t_0}\int_{t_0}^{t}\left( \|(3s_x -\psi_x)(t)\|_2^2+\|(3s_x -\psi_x)(t-\tau)\|_2^2\right) d\tau\\ & \quad +\frac{2}{t-t_0}\int_{t_0}^{t}\left( \|(3s_{xt} -\psi_{xt})(t)\|_2^2+\|(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2\right) d\tau\\ &\leq \frac{4}{l_0(t-t_0)}\int_{t_0}^{t}\left(E(t)+E(t-\tau)+E_1(t)+ E_1(t-\tau) \right)d\tau \\ &\leq \frac{8}{l_0(t-t_0)}\int_{t_0}^{t}\left(E(0)+ C\left( E_1(0)+ \|(3s_0-\psi_0)_{xx}\|_2^2 \right) \right)d\tau \\ &\leq \frac{8}{l_0}\left(E(0)+ C\left( E_1(0)+ \|(3s_0-\psi_0)_{xx}\|_2^2\right) \right) \lt \infty, \ \forall \ t \gt t_0. \end{aligned} \end{equation} | (3.63) |
This last inequality allows us to choose 0 < \sigma < 1 such that
\begin{equation} \Phi(t) \lt 1, \ \ \forall \ t \gt t_0. \end{equation} | (3.64) |
Hence forth, we assume \Phi(t) > 0 , otherwise, we get immediately from (3.61)
\begin{equation} \nonumber E(t)\leq \frac{C}{t-t_0}, \ \ \forall \ t \gt t_0. \end{equation} |
Next, we define the functional \mu by
\begin{equation} \begin{aligned} \mu(t) = -&\int_{t_0}^{t}g'(\tau)\|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2 d\tau\\ &-\int_{t_0}^{t}g'(\tau)\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 d\tau \end{aligned} \end{equation} | (3.65) |
and observe that
\begin{equation} \mu(t)\leq -C\left(E'(t)+ E'_1(t) \right) + c_4g(t), \ \ \ \ \forall \ t \gt t_0, \end{equation} | (3.66) |
where c_4 is a fixed positive constant. The fact that H is strictly convex and H(0) = 0 implies
\begin{equation} H(\nu \tau)\leq \nu H(\tau), \ \ 0\leq \nu\leq 1 \ {\rm and}\ \tau\in (0, r]. \end{equation} | (3.67) |
Using assumption (G1) , (3.67), Jensen’s inequality and (3.64), we get for any t > t_0
\begin{equation} \begin{aligned} \mu(t) = &-\frac{1}{\Phi(t)}\int_{t_0}^{t}\Phi(t)g'(\tau)\|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2 d\tau\\ & \quad -\frac{1}{\Phi(t)}\int_{t_0}^{t}\Phi(t)g'(\tau)\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 d\tau\\ &\geq \frac{1}{\Phi(t)}\int_{t_0}^{t}\Phi(t)\xi(\tau)H(g(\tau))\|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2 d\tau\\ & \quad +\frac{1}{\Phi(t)}\int_{t_0}^{t}\Phi(t)\xi(\tau)H(g(\tau))\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 d\tau\\ &\geq \frac{\xi(t)}{\Phi(t)}\int_{t_0}^{t}H(\Phi(t)g(\tau))\|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2 d\tau\\ & \quad +\frac{\xi(t)}{\Phi(t)}\int_{t_0}^{t}H(\Phi(t)g(\tau))\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 d\tau\\ &\geq \frac{\xi(t)(t-t_0)}{\sigma}H\left(\frac{\sigma}{t-t_0} \int_{t_0}^{t}g(\tau)(\Omega_1(t-\tau)+\Omega_2(t-\tau)) d\tau\right) \\ & = \frac{\xi(t)(t-t_0)}{\sigma}\bar{H}\left(\frac{\sigma}{t-t_0} \int_{t_0}^{t}\left( \Omega_1(t-\tau)+\Omega_2(t-\tau)\right) d\tau\right), \end{aligned} \end{equation} | (3.68) |
where
\begin{equation} \nonumber \begin{aligned} &\Omega_1(t-\tau) = \|(3s_x -\psi_x)(t)-(3s_x -\psi_x)(t-\tau)\|_2^2 , \\ &\Omega_2(t-\tau) = \|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 \end{aligned} \end{equation} |
and \bar{H} is the C^2- strictly increasing and convex extension of H on (0, +\infty). This implies
\begin{equation} \begin{aligned} \int_{t_0}^{t}g(\tau)\|(3s_x -\psi_x)(t)&-(3s_x -\psi_x)(t-\tau)\|_2^2 d\tau +\int_{t_0}^{t}g(\tau)\|(3s_{xt} -\psi_{xt})(t)-(3s_{xt} -\psi_{xt})(t-\tau)\|_2^2 d\tau\\ &\leq \frac{(t-t_0)}{\sigma}\bar{H}^{-1}\left(\frac{\sigma \mu(t)}{\xi(t)(t-t_0)} \right), \ \forall\ t \gt t_0. \end{aligned} \end{equation} | (3.69) |
Thus, the inequality (3.61) becomes
\begin{equation} \begin{aligned} \tilde{L}'(t)\leq &-\tilde{\beta} E(t)-C\left(E'(t)+ E'_1(t) \right) + c_3g(t)+ \frac{C(t-t_0)}{\sigma}\bar{H}^{-1}\left(\frac{\sigma \mu(t)}{\xi(t)(t-t_0)} \right), \ \forall\ t \gt t_0. \end{aligned} \end{equation} | (3.70) |
Let \tilde{L}_1(t): = \tilde{L}(t)+C\left(E(t)+ E_1(t) \right). Then (3.70) becomes
\begin{equation} \tilde{L}'_1(t)\leq -\tilde{\beta} E(t) + \frac{C(t-t_0)}{\sigma}\bar{H}^{-1}\left(\frac{\sigma \mu(t)}{\xi(t)(t-t_0)} \right) + c_3g(t), \ \forall\ t \gt t_0. \end{equation} | (3.71) |
For 0 < r_1 < r, we define the functional \tilde{L}_2 by
\begin{equation} \tilde{L}_2(t): = \bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\tilde{L}_1(t), , \ \forall\ t \gt t_0. \end{equation} | (3.72) |
From (3.71) and the fact that
E'(t)\leq 0, \ \bar{H}'(t) \gt 0, \ \bar{H}''(t) \gt 0, |
we obtain, for all t > t_0,
\begin{align} \tilde{L}'_2(t) = &\left(-\frac{r_1}{(t-t_0)^2} .\frac{E(t)}{E(0)}+\frac{r_1}{t-t_0}.\frac{E'(t)}{E(0)}\right) \bar{H}''\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\tilde{L}_1(t)+\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\tilde{L}'_1(t)\\ \leq&-\tilde{\beta} E(t)\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)+c_3g(t)\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)+\frac{C(t-t_0)}{\sigma}\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\bar{H}^{-1}\left(\frac{\sigma \mu(t)}{\xi(t)(t-t_0)} \right). \end{align} | (3.73) |
Let \bar{H}^{\ast} be the convex conjugate of \bar{H} as in (3.30) and let
A = \bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\ \ {\rm and}\ \ B = \bar{H}^{-1}\left(\frac{\sigma \mu(t)}{\xi(t)(t-t_0)} \right). |
Then, (3.30), (3.31) and (3.73) yield, for all t > t_0,
\begin{align} \tilde{L}'_2(t) \leq&-\tilde{\beta} E(t)\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)+c_3g(t)\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\\ &+\frac{C(t-t_0)}{\sigma}\bar{H}^{\ast}\left(\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\right)+\frac{C(t-t_0)}{\sigma}.\frac{\sigma \mu(t)}{\xi(t)(t-t_0)}\\ \leq &-\tilde{\beta} E(t)\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)+c_3g(t)\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)+Cr_1\frac{E(t)}{E(0)}\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)+C\frac{\mu(t)}{\xi(t)}\\ \leq&-(\tilde{\beta}E(0)-Cr_1)\frac{E(t)}{E(0)}\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)+C\frac{\mu(t)}{\xi(t)}+c_3g(t)\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right) \end{align} | (3.74) |
By selecting r_1 small enough so that (\tilde{\beta}E(0)-Cr_1) > 0 , we arrive at
\begin{equation} \begin{aligned} \tilde{L}'_2(t)\leq &-\tilde{\beta}_2\frac{E(t)}{E(0)}\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)+C\frac{\mu(t)}{\xi(t)}+c_3g(t)\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right), \ \forall\ t \gt t_0, \end{aligned} \end{equation} | (3.75) |
for some positive constant \tilde{\beta}_2.
Now, multiplying (3.75) by \xi(t) and recalling that r_1\frac{E(t)}{E(0)} < r , we arrive at
\begin{align} \xi(t)\tilde{L}'_2(t)&\leq-\tilde{\beta}_2 \xi(t)\frac{E(t)}{E(0)}\bar{H}'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right) + C\mu(t)+c_3g(t)\xi(t)H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\\ &\leq -\tilde{\beta}_2 \xi(t)\frac{E(t)}{E(0)}H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)-C(E'(t)+E'_1(t))+c_4g(t)+c_3g(t)H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right), \ \forall\ t \gt t_0. \end{align} | (3.76) |
Since \frac{r_1}{t-t_0}\longrightarrow 0 as t\longrightarrow \infty, there exists t_2 > t_0 such that \frac{r_1}{t-t_0} < r_1 , whenever t > t_2 . Using this fact and observing that H' strictly increasing, and E and \xi are non-decreasing, we get
\begin{equation} H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\leq H'(r_1), \ \forall\ t \gt t_2. \end{equation} | (3.77) |
Using (3.77), it follows from (3.76) that
\begin{equation} \tilde{L}_3'(t)\leq-\tilde{\beta}_2 \xi(t)\frac{E(t)}{E(0)}H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right) + c_5g(t), \ \forall\ t \gt t_2, \end{equation} | (3.78) |
where \tilde{L}_3 = (\xi\tilde{L}_2+CE+ CE_1) and c_5 > 0 is a constant. Using the non-increasing property of \xi, we have
\begin{equation} \tilde{\beta}_2 \xi(t)\frac{E(t)}{E(0)}H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\leq -\tilde{L}_3'(t) + c_5g(t), \ \forall\ t \gt t_2. \end{equation} | (3.79) |
Using the fact that E is non-increasing and H'' > 0 we conclude that the map
\begin{equation} \nonumber t\longmapsto E(t)H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right) \end{equation} |
is non-increasing. Therefore, integrating (3.79) over (t_2, t) yields
\begin{equation} \begin{aligned} \tilde{\beta}_2\frac{E(t)}{E(0)}H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\int_{t_2}^{t} \xi(\tau)d\tau &\leq \tilde{\beta}_2\int_{t_2}^{t} \xi(\tau)\frac{E(\tau)}{E(0)}H'\left( \frac{r_1}{\tau-t_0}.\frac{E(\tau)}{E(0)}\right)d\tau\\ &\leq -\tilde{L}_3(t)+\tilde{L}_3(t_2) + c_5\int_{t_2}^tg(\tau)d\tau\\ &\leq \tilde{L}_3(t_2) + c_5\int_{0}^{\infty} g(\tau)d\tau\\ & = \tilde{L}_3(t_2) + c_5(b-l_0), \ \forall\ t \gt t_2. \end{aligned} \end{equation} | (3.80) |
Next, we multiply both sides of (3.80) by \frac{1}{t-t_0} , for t > t_2, we get
\begin{equation} \frac{\tilde{\beta}_2}{(t-t_0)}.\frac{E(t)}{E(0)}H'\left( \frac{r_1}{t-t_0}.\frac{E(t)}{E(0)}\right)\int_{t_2}^{t} \xi(\tau)d\tau\leq \frac{\tilde{L}_3(t_2) + c_5(b-l_0)}{t-t_0}, \ \ \forall\ t \gt t_2. \end{equation} | (3.81) |
Since H' is strictly increasing, then H_2(\tau) = \tau H'(\tau) is a strictly increasing function. It follows from (3.81) that
\begin{equation} \nonumber E(t)\leq a_2(t-t_0) H_2^{-1} \left(\frac{a_1}{ (t-t_0)\int_{t_2}^t \xi (\tau)d\tau }\right), \ \forall\ t \gt t_2. \end{equation} |
for some positive constants a_1 and a_2. This completes the proof.
(1). Let g(t) = ae^{-bt}, \ t\geq 0, \ \ a, \ b > 0 are constants and a is chosen such that ( G_1) holds. Then
\begin{equation} \nonumber g'(t) = - abe^{-bt} = -bH(g(t)) \ \ {\rm with } \ \ H(t) = t. \end{equation} |
Therefore, from (3.14), the energy function (2.11) satisfies
\begin{equation} E(t)\leq a_2e^{-\alpha t}, \ \forall \ t\geq 0, \ {\rm where}\ \ \alpha = ba_1. \end{equation} | (3.82) |
Also, for H_2(\tau) = \tau , it follows from (3.55) that, there exists t_2 > 0 such that the energy function (2.11) satisfies
\begin{equation} E(t)\leq \frac{C}{t-t_2}, \ \ \forall\ t \gt t_2, \end{equation} | (3.83) |
for some positive constant C.
(2). Let g(t) = ae^{-(1+t)^{b}}, \ t\geq 0, \ \ a > 0, \ 0 < b < 1 are constants and a is chosen such that ( G_1) holds. Then,
\begin{equation} \nonumber g'(t) = -ab(1+t)^{b-1}e^{-(1+t)^{b}} = -\xi(t) H(g(t)), \end{equation} |
where \xi(t) = b(1+t)^{b-1} and H(t) = t. Thus, we get from (3.14) that
\begin{equation} E(t)\leq a_2e^{-a_1(1+t)^b}, \ \forall \ t\geq 0. \end{equation} | (3.84) |
Likewise, for H_2(t) = t , then estimate (3.55) implies there exists t_2 > 0 such that the energy function (2.11) satisfies
\begin{equation} E(t)\leq \frac{C}{(1+t)^{b}}, \ \ \forall\ t \gt t_2, \end{equation} | (3.85) |
for some positive constant C.
(3). Let g(t) = \frac{a}{(1+t)^{b}}, \ t\geq 0, \ \ a > 0, \ b > 1 are constants and a is chosen in such a way that ( G_1) holds. We have
\begin{equation} \nonumber g'(t) = \frac{-ab}{(1+t)^{b+1}} = -\xi\left(\frac{a}{(1+t)^b} \right)^{\frac{b+1}{b}} = -\xi g^q(t) = -\xi H(g(t)), \end{equation} |
where
\begin{equation} \nonumber H(t) = t^q, \ \ q = \frac{b +1}{b} \ \ {\rm satisfying} \ \ 1 \lt q \lt 2 \ \ {\rm and }\ \ \xi = \frac{b}{a^\frac{1}{b}} \gt 0. \end{equation} |
Hence, we deduce from (3.41) that
\begin{equation} E(t)\leq \frac{C}{(1+t)^b}, \ \forall \ t\geq 0. \end{equation} | (3.86) |
Furthermore, for H_2(t) = qt^q , estimate (3.55) implies there exists t_2 > 0 such that the energy function (2.11) satisfies
\begin{equation} E(t) \leq \frac{C}{(1+t)^{(b-1)/(b+1)}}, \ \forall\ t \gt t_2, \end{equation} | (3.87) |
for some positive constant C.
In this section, we prove the functionals L_i, i = 1\cdots8, used in the proof of our stability results.
Lemma 4.1. The functional I_1(t) satisfies, along the solution of (1.6)-(1.8), for all t\geq t_0 > 0 and for any \epsilon_1, \epsilon_2 > 0 , the estimate
\begin{align} I_1'(t)\leq -\frac{I_{\rho}g_0}{2} &\| 3s_t-\psi_t\|_2^2+\epsilon_1\|3s_x -\psi_x\|_2^2 + \epsilon_2\|\psi-w_x\|_2^2+ C C_{\alpha}\left( 1+\frac{1}{\epsilon_1}+\frac{1}{\epsilon_2}\right)\left(h\diamond (3s_x-\psi_x) \right)(t), \end{align} | (4.1) |
where g_0 = \int_0^{t_0}g(\tau)d\tau\leq \int_0^{t}g(\tau)d\tau.
Proof. Differentiating I_1(t) , using (1.6)_2 and integrating by part, we have
\begin{align} I_1'(t) = - &I_{\rho}\int_0^1(3s_t-\psi_t) \int_0^t g'(t-\tau)\left((3s-\psi)(t)- (3s-\psi)(\tau)\right)d\tau dx\\ & + D(t) \int_0^1(3s_x-\psi_x)\int_0^t g(t-\tau)\left((3s_x-\psi_x)(t)- (3s_x-\psi_x)(\tau)\right)d\tau dx\\ & +\int_0^1 \left(\int_0^t g(t-\tau)\left( (3s_x-\psi_x)(t)- (3s_x-\psi_x)(\tau)\right)d\tau \right)^2dx-I_{\rho}\left( \int_0^t g(\tau)d\tau \right) \int_0^1(3s_t-\psi_t)^2 dx \\ &-G\int_0^1(\psi-w_x) \int_0^t g(t-\tau)\left(( 3s-\psi)(t)- (3s-\psi)(\tau)\right)d\tau dx, \end{align} | (4.2) |
where D(t) = \left(D-\int_0^t g(\tau)d\tau\right). Now, we estimate the terms on the right hand-side of (4.2). Exploiting Young's and Poincaré's inequalities, Lemmas 2.1-2.6 and performing similar computations as in (2.8), we have for any \epsilon_1 > 0 ,
\begin{align} D(t)\int_0^1(3s_x-\psi_x)&\int_0^t g(t-\tau)\left((3s_x-\psi_x)(t)- (3s_x-\psi_x)(\tau)\right)d\tau dx\\ &\leq \epsilon_1 \|(3s_x-\psi_x\|_2^2 + \frac{C C_{\alpha}}{\epsilon_1}\left(h\diamond (3s_x-\psi_x)\right)(t) \end{align} | (4.3) |
and
\begin{equation} \int_0^1 \left(\int_0^t g(t-\tau)\left( (3s_x-\psi_x)(t)- (3s_x-\psi_x)(\tau)\right)d\tau \right)^2dx\leq C_{\alpha}\left(h\diamond (3s_x-\psi_x)\right)(t). \end{equation} | (4.4) |
Also, for \delta_1 > 0, we have
\begin{align} - I_{\rho}&\int_0^1(3s_t-\psi_t) \int_0^t g'(t-\tau)\left((3s-\psi)(t)- (3s-\psi)(\tau)\right)d\tau dx\\ = &I_{\rho}\int_0^1(3s_t-\psi_t) \int_0^t h(t-\tau)\left((3s-\psi)(t)- (3s-\psi)(\tau)\right)d\tau dx\\ &-I_{\rho}\alpha\int_0^1(3s_t-\psi_t) \int_0^t g(t-\tau)\left((3s-\psi)(t)- (3s-\psi)(\tau)\right)d\tau dx\\ \leq&\delta_1\|3s_t-\psi_t\|_2^2 + \frac{I_{\rho}^2}{2\delta_1} \int_0^1 \left(\int_0^t h(t-\tau)\left( (3s-\psi)(t)- (3s-\psi)(\tau)\right)d\tau \right)^2dx\\ &+\frac{\alpha^2 I_{\rho}^2}{2\delta_1}\int_0^1 \left(\int_0^t g(t-\tau)\left( (3s-\psi)(t)- (3s-\psi)(\tau)\right)d\tau\right)^2dx\\ \leq& \delta_1\|3s_t-\psi_t\|_2^2 + \frac{ I_{\rho}^2}{2\delta_1}\left( \int_0^t h(\tau)d\tau\right) \left(h\diamond (3s-\psi)\right)(t)+\frac{\alpha^2 I_{\rho}^2 C_{\alpha}}{2\delta_1} \left(h\diamond (3s-\psi)\right)(t)\\ \leq& \delta_1\|3s_t-\psi_t\|_2^2 + \frac{C(C_{\alpha}+1)}{\delta_1} \left(h\diamond (3s-\psi)_x\right)(t). \end{align} | (4.5) |
For the last term, we have
\begin{align} -G\int_0^1(\psi-w_x)& \int_0^t g(t-\tau)\left(( 3s-\psi)(t)- (3s-\psi)(\tau)\right)d\tau dx\leq \epsilon_2\|\psi-w_x\|_2^2 + \frac{G^2 C_{\alpha}}{4\epsilon_2}\left(h\diamond (3s-\psi)_x\right)(t). \end{align} | (4.6) |
Combination of (4.2)-(4.6) lead to
\begin{align} I_1'(t)\leq -&\left( I_{\rho} \int_0^t g(\tau)d\tau-\delta_1\right) \|3w_t-\psi_t\|_2^2 +\epsilon_1\|3s_x -\psi_x\|_2^2+\epsilon_2\|\psi-w_x\|_2^2 \\ & + CC_{\alpha}\left( 1+\frac{1}{\delta_1}+\frac{1}{\epsilon_1}+\frac{1}{\epsilon_2}\right)\left(h\diamond (3s_x-\psi_x) \right)(t). \end{align} | (4.7) |
Since g(0) > 0 and g is continuous. Thus for any t\geq t_0 > 0, we get
\begin{equation} \int_0^{t}g(\tau)d\tau\geq \int_0^{t_0}g(\tau)d\tau = g_0 \gt 0. \end{equation} | (4.8) |
We select \delta_1 = \dfrac{I_{\rho}g_0}{2} to get (4.1).
Lemma 4.2. The functional I_2(t) satisfies, along the solution of (1.6)-(1.8) and for any \delta_4 > 0 , the estimate
\begin{equation} \begin{aligned} I'_2(t)&\leq-3D\|s_x\|_2^2 -3\gamma \|s\|_2^2 + \delta_4\|w_t\|_2^2 + C\left(1+\frac{1}{\delta_4} \right) \|s_t\|_2^2 +C \|\theta_x\|_2^2, \ \ \forall t\geq 0. \end{aligned} \end{equation} | (4.9) |
Proof. Differentiation of I_2(t) , using (1.6)_1 and (1.6)_3 and integration by part, leads to
\begin{equation} \nonumber I_2'(t) = 3I_{\rho}\|s_t\|_2^2-3D\|s_x\|_2^2-4\gamma \|s\|_2^2-\delta\int_0^t s\theta_x dx + 3\rho\int_0^1w_t\int_0^x s_t (y)dy dx. \end{equation} |
Applying Cauchy-Schwarz and Young's inequalities and (2.5), we get for any \delta_4 > 0,
\begin{eqnarray} I_2'(t)&\leq & 3I_{\rho}\|s_t\|_2^2-3D\|s_x\|_2^2-4\gamma \|s\|_2^2 + \gamma \|s\|_2^2 +\frac{\delta^2}{4\gamma} \|\theta_x\|_2^2 +\delta_4\|w_t\|_2^2 + \frac{9\rho^2}{4\delta_4}\int_0^1\left(\int_0^x s_t(y)dy \right)^2dx\\ &\leq& -3D\|s_x\|_2^2 -3\gamma \|s\|_2^2 + \delta_4\|w_t\|_2^2 + C\left(1+\frac{1}{\delta_4} \right) \|s_t\|_2^2 + C \|\theta_x\|_2^2. \end{eqnarray} |
This completes the proof.
Lemma 4.3. The functional I_3(t) satisfies, along the solution of (1.6)-(1.8) and for any \epsilon_3 > 0, the estimate
\begin{equation} \begin{aligned} I_3'(t)\leq& - \frac{\delta I_{\rho}}{2}\|s_t\|_2^2 + \epsilon_3 \|s_x\|_2^2 + \epsilon_3\|\psi-w_x\|_2^2 + C\left(1+\frac{1}{\epsilon_3} \right)\|\theta_x\|_2^2, \ \ \forall t\geq 0. \end{aligned} \end{equation} | (4.10) |
Proof. Differentiation of I_3 , using (1.6)_3 , (1.6)_4 and integration by parts, yields
\begin{align*} \nonumber I_3'(t) = 3&\lambda I_{\rho}\int_0^1 \theta_x s_t dx -3I_{\rho}\delta\|s_t\|_2^2 -3kD\int_0^1 \theta s_x dx +k\delta\|\theta\|_2^2\\ \nonumber &+ 3kG\int_0^1\theta\int_0^x (\psi-w_y)(y)dy dx +4\gamma k\int_0^1\theta\int_0^t s(y)dydx . \end{align*} |
Using Cauchy-Schwarz, Young's and Poincaré's inequalities together with Lemmas 2.1-2.6, we have
\begin{eqnarray} I_3'(t)&\leq& \delta_2\|s_t\|_2^2 + C_{\delta_2}\|\theta_x\|_2^2 -3I_{\rho}\delta\|s_t\|_2^2 + \frac{\epsilon_3}{2}\|s_x\|_2^2 + C\left( 1+\frac{1}{\epsilon_3}\right) \|\theta\|_2^2 \\ &&+\epsilon_3\int_0^1\left(\int_0^x (\psi-w_y)(y)dy \right)^2 dx +\frac{\epsilon_3}{2}\int_0^1\left(\int_0^x s(y)dy \right)^2dx\\ &\leq & \delta_2\|s_t\|_2^2 + C_{\delta_2}\|\theta_x\|_2^2 -3I_{\rho}\delta\|s_t\|_2^2 + \epsilon_3\|s_x\|_2^2 +\epsilon_3\|\psi-w_x\|_2^2 +C\left( 1+\frac{1}{\epsilon_3}\right) \|\theta_x\|_2^2 . \end{eqnarray} |
We choose \delta_2 = \dfrac{5I_{\rho}\delta}{2} to get (4.10).
Lemma 4.4. The functional I_4(t) satisfies, along the solution of (1.6)-(1.8) and for any \epsilon_4 > 0, the estimate
\begin{equation} \begin{aligned} I_4'(t)\leq& -\rho\|w_t\|_2^2 + \epsilon_4\|3s_x-\psi_x\|_2^2+ C\|s_x\|_2^2 + C_{\epsilon_4} \|\psi-w_x\|_2^2 , \ \ \forall t\geq 0. \end{aligned} \end{equation} | (4.11) |
Proof. Using (1.6)_1 and integration by parts, we have
\begin{equation*} I_4'(t) = -\rho\|w_t\|_2^2- G\int_0^1(\psi-w_x) w_x dx. \end{equation*} |
We note that w_x = -(\psi-w_x)-(3s-\psi)+3s to arrive at
\begin{equation} \nonumber I_4'(t) = -\rho\|w_t\|_2^2 + G\|\psi-w_x\|^2_2 + G\int_0^1(\psi-w_x)(3s-\psi)dx -3G\int_0^1(\psi-w_x)s dx. \end{equation} |
It follows from Young's and Poincaré's inequalities that
\begin{align*} I_4'(t)&\leq -\rho\|w_t\|_2^2 + G\|\psi-w_x\|^2_2 + \epsilon_4\|3s-\psi\|_2^2 +\frac{C}{\epsilon_4}\|\psi-w_x\|^2_2+\frac{3G}{2} \|\psi-w_x\|^2_2 + \frac{3G}{2}\|s\|_2^2\\ \nonumber \leq&- \rho\|w_t\|_2^2 + G\|\psi-w_x\|^2_2 + \epsilon_4\|3s_x-\psi_x\|_2^2+ C \|s_x\|_2^2+C\left( 1+\frac{1}{\epsilon_4}\right)\|\psi-w_x\|^2_2. \end{align*} |
This completes the proof.
Lemma 4.5. The functional I_5(t) satisfies, along the solution of (1.6)-(1.8) and for any 0 < \alpha < 1 , the estimate
\begin{equation} I_5'(t)\leq -\frac{l_0}{4}\|3s_x-\psi_x\|_2^2 + I_{\rho} \|3s_t-\psi_t\|_2^2 + C\|\psi-w_x\|_2^2 + CC_{\alpha}\left(h\diamond (3s_x-\psi_x) \right)(t). \end{equation} | (4.12) |
Proof. Differentiating I_5 , using (1.6)_2 , we arrive at
\begin{eqnarray} I'_5(t)& = & I_{\rho} \|3s_t-\psi_t\|_2^2-\left(D-\int_0^t g(\tau)d\tau \right)\|3s_x-\psi_x\|_2^2+ G\int_0^1(3s-\psi)(\psi-w_x)dx \\ &&+\int_0^1(3s_x-\psi_x)\int_0^t g(t-\tau)\left(\left( 3s_x-\psi_x\right)(x, \tau)- \left(3s_x-\psi_x\right)(x, t)\right)d\tau dx. \end{eqnarray} |
Applying Lemmas 2.1-2.6, Cauchy-Schwarz, Young's and Poincaré's inequalities, we obtain any \delta_3 > 0
\begin{eqnarray} I'_5(t)&\leq & I_{\rho} \|3s_t-\psi_t\|_2^2 -l_0\|3s_x-\psi_x\|_2^2 + \delta_3\|3s_x-\psi_x\|_2^2 +\frac{G^2}{4\delta_3}\|\psi-w_x\|_2^2\\ && +\frac{l_0}{2}\|3s_x-\psi_x\|_2^2 +\frac{1}{2l_0}C_{\alpha}\left(h\diamond (3s_x-\psi_x) \right)(t). \end{eqnarray} | (4.13) |
We select \delta_3 = \dfrac{l_0}{4} and obtain the desired result.
Lemma 4.6. The functional I_6(t) satisfies, along the solution of (1.6)-(1.8) and for any for any \epsilon_1 , the estimate
\begin{align} I_6'(t)\leq-&G^2\|\psi-w_x\|_2^2 + \epsilon_1\|3s_t-\psi_t\|_2^2 + C\left(1+\frac{1}{\epsilon_1} \right)\|s_t\|_2^2\\ &+ C\|s_x\|^2_2 +C\|\theta_x\|_2^2+3(I_{\rho}G-\rho D)\int_0^1 w_ts_{xt} dx , \ \ \forall t\geq 0. \end{align} | (4.14) |
Proof. Differentiating I_6(t) , using (1.6)_1 and (1.6)_3 and integration by parts, we obtain
\begin{align} I'_6(t) = -&3G^2\|\psi-w_x\|_2^2-4\gamma G\int_0^1(\psi-w_x)s dx -\delta G \int_0^1 (\psi-w_x) \theta_x dx \\ &-3I_{\rho}G\int_0^t(3s_t-\psi_t)s_t dx + 9I_{\rho}G\|s_t\|_2^2 +3(I_{\rho}G-\rho D)\int_0^1 w_ts_{xt} dx. \end{align} | (4.15) |
Young's and Poincaré's inequalities give
\begin{eqnarray} &&-4\gamma G\int_0^1(\psi-w_x)s dx \leq G^2\|\psi-w_x\|_2^2 + 4\gamma^2 C_p\|s_x\|^2_2, \\ && -\delta G \int_0^1 (\psi-w_x) \theta_x dx\leq G^2\|\psi-w_x\|_2^2 +\frac{\delta^2}{4}\|\theta_x\|_2^2, \\ && -3I_{\rho}G\int_0^t(3s_t-\psi_t)s_t dx \leq \epsilon_1\|3s_t-\psi_t\|_2^2 +\frac{(3I_{\rho}G)^2}{\epsilon_1}\|s_t\|_2^2. \end{eqnarray} | (4.16) |
Substituting (4.16) into (4.15), we obtain (4.14). This completes the proof.
Lemma 4.7. The functional I_7(t) satisfies, along the solution of (1.6)-(1.8), the estimate
\begin{equation} I_7'(t)\leq 3(D-l_0)\|3s_x-\psi_x\|_2^2- \frac{1}{2} (g\diamond (3s_x-\psi_x))(t), \ \forall t\geq 0. \end{equation} | (4.17) |
Proof. Differentiate I_7(t) and use the fact that J'(t) = -g(t) to get
\begin{equation} \begin{aligned} I'_7(t) = & \int_0^1 \int_0^t J'(t-\tau)(3s_x-\psi_x)^2(\tau)d\tau dx + J(0)\|3s_x-\psi_x\|_2^2\\ = & -(g\diamond (3s_x-\psi_x))(t) + J(t)\|3s_x-\psi_x\|_2^2\\ & \ -2\int_0^1 (3s_x-\psi_x)\int_0^t g(t-\tau)\left((3s_x-\psi_x)(\tau)-(3s_x-\psi_x)(t) \right)dx. \end{aligned} \end{equation} | (4.18) |
Using Cauchy-Schwarz and (G1), we have
\begin{equation} \begin{aligned} -2\int_0^1 &(3s_x-\psi_x)\int_0^t g(t-\tau)\left((3s_x-\psi_x)(\tau)-(3s_x-\psi_x)(t) \right)\\ \leq& 2(D-l_0)\|3s_x-\psi_x\|_2^2 +\frac{\int_0^t g(\tau)d\tau}{2(D-l_0)}(g\diamond (3s_x-\psi_x))(t)\\ \leq& 2(D-l_0)\|3s_x-\psi_x\|_2^2 +\frac{1}{2}(g\diamond (3s_x-\psi_x))(t) \end{aligned} \end{equation} | (4.19) |
Thus, we get
\begin{equation} I'_7(t)\le 2(D-l_0)\|3s_x-\psi_x\|_2^2- \frac{1}{2}(g\diamond (3s_x-\psi_x))(t) + J(t)\|3s_x-\psi_x\|_2^2. \end{equation} | (4.20) |
Since J is decreasing (J'(t) = -g(t)\leq 0) , so J(t)\leq J(0) = D-l_0 . Hence, we arrive at
\begin{equation} \nonumber I_7'(t)\leq 3(D-l_0)\|3s_x-\psi_x\|_2^2- \frac{1}{2} (g\diamond (3s_x-\psi_x))(t). \end{equation} |
The next lemma is used only in the proof of the stability result for nonequal-wave-speed of propagation.
Lemma 4.8. Let (w, 3s-\psi, s, \theta) be the strong solution of problem (1.6). Then, for any positive numbers \sigma_1, \sigma_2, \sigma_3 , the functional I_8(t) satisfies
\begin{equation} \begin{aligned} I'_8(t)\leq &-3(I_{\rho}G-\rho D)\int_0^1 w_ts_{xt} dx +\sigma_1\|w_t\|_2^2 + \sigma_2\|\psi-w_x\|_2^2 +\sigma_3\|3s_x-\psi_x\|_2^2\\ &+C\|s_x\|_2^2+C\left(1+\frac{1}{\sigma_1}+\frac{1}{\sigma_2}+\frac{1}{\sigma_3} \right) \|\theta_{xt}\|_2^2, \ \forall \ t\geq t_0. \end{aligned} \end{equation} | (4.21) |
Proof. Differentiation of I_8 , using integration by part and the boundary condition give
\begin{equation} \begin{aligned} I'_8(t)& = \frac{3\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_x w_{xt} dx+\frac{3\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_{xt} w_x dx\\ & = \frac{3\lambda}{\delta}(I_{\rho} G-\rho D)\left[-\int_0^1\theta_{xx}w_t dx \right] +\frac{3\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_{xt} w_x dx. \end{aligned} \end{equation} | (4.22) |
We note that w_x = -(\psi-w_x)-(3s-\psi)+3s and from (1.6)_4, \ \lambda\theta_{xx} = k\theta_t +\delta s_{xt} . So, (4.22) becomes
\begin{equation} \begin{aligned} I'_8(t) = &-\frac{3}{\delta}(I_{\rho}G-\rho D)k\int_0^1\theta_t w_{t} dx- 3(I_{\rho}G-\rho D)\int_0^1s_{xt} w_t dx+ \frac{9\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_{xt}s dx\\ &-\frac{3\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_{xt}(\psi-w_x) dx -\frac{3\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_{xt} (3s-\psi)dx \end{aligned} \end{equation} | (4.23) |
Using Young's and Poincaré's inequalities, we have for any positive numbers \sigma_1, \sigma_2, \sigma_3 ,
\begin{equation} \begin{aligned} &-\frac{3}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_t w_{t} dx\leq \sigma_1\|w_t\|_2^2 +\frac{C}{\sigma_1} \|\theta_{xt}\|_2^2, \\ &-\frac{3\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_{xt}(\psi-w_x) dx\leq \sigma_2\|\psi-w_x\|_2^2 +\frac{C}{\sigma_2} \|\theta_{xt}\|_2^2, \\ &-\frac{3\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_{xt} (3s-\psi)dx\leq \sigma_3\|3s_x-\psi_x\|_2^2 + \frac{C}{\sigma_3} \|\theta_{xt}\|_2^2, \\ &\frac{9\lambda}{\delta}(I_{\rho}G-\rho D)\int_0^1\theta_{xt}s dx\leq C\|s_x\|_2^2 + C \|\theta_{xt}\|_2^2. \end{aligned} \end{equation} | (4.24) |
Substituting (4.24) into (4.23), we obtain (4.21).
In this paper, we have established a general and optimal stability estimates for a thermoelastic Laminated system, where the heat conduction is given by Fourier's Law and memory as the only source of damping. Our results are established under weaker conditions on the memory and physical parameters. From our results, we saw that the decay rate is faster provided the wave speeds of the first two equations of the system are equal (see (1.3)). A similar result was established recently in [19] when the heat conduction is given by Maxwell-Cattaneo's Law. An interesting case is when the kernel memory term is couple with the first or third equations in system (1.6). Our expectation is that the stability in both cases will depend on the speed of wave propagation.
The authors appreciate the continuous support of University of Hafr Al Batin, KFUPM and University of Sharjah. The first and second authors are supported by University of Hafr Al Batin under project #G-106-2020 . The third author is sponsored by KFUPM under project #S B181018.
The authors declare no conflict of interest
[1] |
N. Bairagi, D. Adak, Switching from simple to complex dynamics in a predator-prey-parasite model: An interplay between infection rate and incubation delay, Math. Biosci., 277 (2016), 1-14. doi: 10.1016/j.mbs.2016.03.014
![]() |
[2] |
R. M. Eide, A. L. Krause, N. T. Fadai, R. A. V. Gorder, Predator-prey-subsidy population dynamics on stepping-stone domains, J. Theor. Biol., 451 (2018), 19-34. doi: 10.1016/j.jtbi.2018.04.038
![]() |
[3] | M. Peng, Z. D. Zhang, X. D. Wang, Hybrid control of Hopf bifurcation in a Lotka-Volterra predator-prey model with two delays, Adv. Differ. Equ., 387 (2017), 1-12. |
[4] | Z. Li, D. Q. Jiang, D. O'Regan, T. Hayat, B. Ahmad, Ergodic property of a Lotka-Volterra predator-prey model with white noise higher order perturbation under regime switching, Appl. Math. Comput., 330 (2018), 93-102. |
[5] | Z. Z. Ma, F. D. Chen, C. Q. Wu, W. L. Chen, Dynamic behaviors of a Lotka-Volterra predator-prey model incorporating a prey refuge and predator mutual interference, Appl. Math. Comput., 219 (2013), 7945-7953. |
[6] | R. Q. Shi, L. S. Chen, Staged-structured Lotka-Volterra predator-prey models for pest management, Appl. Math. Comput., 203 (2008), 258-265. |
[7] |
S. X. Pan, Asymptotic spreading in a Lotka-Volterra predator-prey system, J. Math. Anal. Appl., 407 (2013), 230-236. doi: 10.1016/j.jmaa.2013.05.031
![]() |
[8] | C. J. Xu, Y. S. Wu, L. Lu, Permanence and global attractivity in a discrete Lotka-Volterra predator- prey model with delays, Adv. Differ. Equ., 208 (2014), 1-15. |
[9] |
Z. L. Luo, Y. P. Lin, Y. X. Dai, Rank one chaos in periodically kicked Lotka-Volterra predator-prey system with time delay, Nonlinear Dynam., 85 (2016), 797-811. doi: 10.1007/s11071-016-2723-3
![]() |
[10] |
J. Xia, Z. X. Yu, R. Yuan, Stability and Hopf bifurcation in a symmetric Lotka-Volterra predator- prey system with delays, Electron. J. Differ. Equ., 2013 (2013), 118-134. doi: 10.1186/1687-1847-2013-118
![]() |
[11] | L. Men, B. S. Chen, G. Wang, Z. W. Li, W. Liu, Hopf bifurcation and nonlinear state feedback control for a modified Lotka-Volterra differential algebraic predator-prey system, In: Fifth Int. Conference on Intelligent Control and Information Processing, 2015 (2015), 233-238. |
[12] | C. J. Xu, M. X. Liao, X. F. He, Stability and Hopf bifurcation analysis for a Lotka-Volterra predator-prey model with two delays, Int. J. Appl. Math. Comput. Sci., 21 (2011), 97-107. |
[13] |
G. M. Mahmoud, Periodic solutions of strongly non-linear Mathieu oscillators, Int. J. NonLinear Mechanics, 32 (1997), 1177-1185. doi: 10.1016/S0020-7462(96)00126-6
![]() |
[14] |
K. W. Chung, C. L. Chan, Z. Xu, G. M. Mahmoud, A perturbation-incremental method for strongly nonlinear autonomous oscillators with many degrees of freedom, Nonlinear Dyn., 28 (2002), 243- 259. doi: 10.1023/A:1015620928121
![]() |
[15] | X. W. Jiang, X. Y. Chen, T. W. Huang, H. C. Yang, Bifurcation and control for a predator-prey system with two delays, IEEE T. Circuits II, 99 (2020), 1-1. |
[16] |
X. P. Yan, C. H. Zhang, Hopf bifurcation in a delayed Lokta-Volterra predator-prey system, Nonlinear Anal., 9 (2008), 114-127. doi: 10.1016/j.nonrwa.2006.09.007
![]() |
[17] |
N. Laskin, Fractional quantum mechanics, Phys. Rev. E, 62 (2000), 3135-3145. doi: 10.1103/PhysRevE.62.3135
![]() |
[18] |
F. Wang, Y. Q. Yang, Quasi-synchronization for fractional-order delayed dynamical networks with heterogeneous nodes, Appl. Math. Comput., 339 (2018), 1-14. doi: 10.1016/j.cam.2018.03.008
![]() |
[19] |
R. Chinnathambi, F. A. Rihan, Stability of fractional-order prey-predator system with time-delay and Monod-Haldane functional response, Nonlinear Dyn., 92 (2018), 1637-1648. doi: 10.1007/s11071-018-4151-z
![]() |
[20] |
J. Alidousti, M. M. Ghahfarokhi, Stability and bifurcation for time delay fractional predator-prey system by incorporating the dispersal of prey, Appl. Math. Model., 72 (2019), 385-402. doi: 10.1016/j.apm.2019.03.029
![]() |
[21] |
V. Lakshmikantham, A. S. Vatsala, Basic theory of fractional differential equations, Nonlinear Anal., 69 (2008), 2677-2682. doi: 10.1016/j.na.2007.08.042
![]() |
[22] |
A. A. Elsadany, A. E. Matouk, Dynamical behaviors of fractional-order Lotka-Volterra predator- prey model and its discretization, J. Appl. Math. Comput., 49 (2015), 269-283. doi: 10.1007/s12190-014-0838-6
![]() |
[23] | M. Javidi, N. Nyamoradi, Dynamic analysis of a fractional order prey-predator interaction with harvesting, Appl. Math. Model., 37 (2015), 8946-8956. |
[24] |
F. A. Rihan, S. Lakshmanan, A. H. Hashish, R. Rakkiyappan, E. Ahmed, Fractional-order delayed predator-prey systems with Holling type-Ⅱ functional response, Nonlinear Dyn., 80 (2015), 777- 789. doi: 10.1007/s11071-015-1905-8
![]() |
[25] |
C. D. Huang, X. Y. Song, B. Fang, M. Xiao, J. D. Cao, Modeling, analysis and bifurcation control of a delayed fractional-order predator-prey model, Int. J. Bifurcat. Chaos, 28 (2018), 1850117. doi: 10.1142/S0218127418501171
![]() |
[26] | C. D. Huang, J. D. Cao, Comparative study on bifurcation control methods in a fractional-order delayed predator-prey system, Sci. China Technol. Sci., 62 (2018), 298-307. |
[27] | K. Baisad, S. Moonchai, Analysis of stability and Hopf bifurcation in a fractional Gauss-type predator-prey model with Allee effect and Holling type-Ⅲ functional response, Adv. Differ. Equ., 82 (2018), 1-20. |
[28] |
Z. H. Li, C. D. Huang, Y. Zhang, Comparative analysis on bifurcation of four-neuron fractional ring networks without or with leakage delays, Adv. Differ. Equ., 2019 (2019), 1-22. doi: 10.1186/s13662-018-1939-6
![]() |
[29] | L. Wu, Z. H. Li, Y. Zhang, B. G. Xie, Complex behavior analysis of a fractional-order land dynamical model with Holling-Ⅱ type land reclamation rate on time delay, Discrete Dyn. Nat. Soc., 2020 (2020), 1-10. |
[30] |
P. Song, H. Y. Zhao, X. B. Zhang, Dynamic analysis of a fractional order delayed predator-prey system with harvesting, Theor. Biosci., 135 (2016), 1-14. doi: 10.1007/s12064-016-0224-z
![]() |
[31] |
K. M. Owolabi, Mathematical modelling and analysis of two-component system with Caputo fractional derivative order, Chaos Soliton. Fractal., 103 (2017), 544-554. doi: 10.1016/j.chaos.2017.07.013
![]() |
[32] |
R. Chinnathambi, F. A. Rihan, Stability of fractional-order prey-predator system with time-delay and Monod-Haldane functional response, Nonlinear Dyn., 92 (2018), 1637-1648. doi: 10.1007/s11071-018-4151-z
![]() |
[33] |
M. Xiao, W. X. Zheng, J. X. Lin, G. P. Jiang, L. D. Zhao, J. D. Cao, Fractional-order PD control at Hopf bifurcations in delayed fractional-order small-world networks, J. Frankl. Inst., 354 (2017), 7643-7667. doi: 10.1016/j.jfranklin.2017.09.009
![]() |
[34] | C. D. Huang, H. Li, J. D. Cao, A novel strategy of bifurcation control for a delayed fractional predator-prey model, Appl. Math. Comput., 347 (2019), 808-838. |
[35] |
C. J. Xu, Y. S. Wu, Bifurcation and control of chaos in a chemical system, Appl. Math. Model., 39 (2015), 2295-2310. doi: 10.1016/j.apm.2014.10.030
![]() |
[36] |
D. W. Ding, X. Y. Zhang, J. D. Cao, N. Wang, D. Liang, Bifurcation control of complex networks model via PD controller, Neurocomputing, 175 (2016), 1-9. doi: 10.1016/j.neucom.2015.09.094
![]() |
[37] |
J. N. Luo, M. L. Li, X. Z. Liu, W. H. Tian, S. M. Zhong, K. B. Shi, Stabilization analysis for fuzzy systems with a switched sampled-data control, J. Frankl. Inst., 357 (2020), 39-58. doi: 10.1016/j.jfranklin.2019.09.029
![]() |
[38] |
K. B. Shi, J. Wang, S. M. Zhong, Y. Y. Tang, J. Cheng, Hybrid-driven finite-time H∞ sampling synchronization control for coupling memory complex networks with stochastic cyber attacks, Neurocomputing, 387 (2020), 241-254. doi: 10.1016/j.neucom.2020.01.022
![]() |
[39] |
K. B. Shi, J. Wang, Y. Y. Tang, S. M. Zhong, Reliable asynchronous sampled-data filtering of T-S fuzzy uncertain delayed neural networks with stochastic switched topologies, Fuzzy Set Syst., 381 (2020), 1-25. doi: 10.1016/j.fss.2018.11.017
![]() |
[40] |
K. B. Shi, J. Wang, S. M. Zhong, Y. Y. Tang, J. Cheng, Non-fragile memory filtering of T-S fuzzy delayed neural networks based on switched fuzzy sampled-data control, Fuzzy Set Syst., 394 (2020), 40-64. doi: 10.1016/j.fss.2019.09.001
![]() |
[41] | C. X. Zhu, Controlling hyperchaos in hyperchaotic Lorenz system using feedback controllers, Appl. Math. Comput., 216 (2010), 3126-3132. |
[42] |
C. D. Yang, C. H. Tao, P. Wang, Comparison of feedback control methods for a hyperchaotic lorenz system, Phys. Lett. A, 374 (2010), 729-732. doi: 10.1016/j.physleta.2009.11.064
![]() |
[43] |
G. M. Mahmoud, A. A. Arafa, T. M. Abed-Elhameed, E. E. Mahmoud, Chaos control of integer and fractional orders of chaotic Burke-Shaw system using time delayed feedback control, Chaos Soliton. Fractal., 104 (2017), 680-692. doi: 10.1016/j.chaos.2017.09.023
![]() |
[44] | I. Podlubny, Fractional differential equations, New York: Academic Press, 1999. |
[45] | D. Matignon, Stability results for fractional differential equations with applications to control processing, IEEE-SMC Pro., 2 (1996), 963-968. |
[46] | S. Bhalekar, D. Varsha, A predictor-corrector scheme for solving nonlinear delay differential equations of fractional order, Int. J. Fract. Calc. Appl., 1 (2011), 1-9. |
1. | Yuxia Guo, Yichen Hu, Infinitely many solutions for Hamiltonian system with critical growth, 2024, 13, 2191-950X, 10.1515/anona-2023-0134 | |
2. | Xingyue He, Chenghua Gao, Jingjing Wang, k-convex solutions for multiparameter Dirichlet systems with k-Hessian operator and Lane-Emden type nonlinearities, 2024, 13, 2191-950X, 10.1515/anona-2023-0136 | |
3. | Hongying Jiao, Shuhai Zhu, Jinguo Zhang, Existence of infinitely many solutions for critical sub-elliptic systems via genus theory, 2024, 16, 2836-3310, 237, 10.3934/cam.2024011 | |
4. | Jinli Yang, Jiajing Miao, Algebraic Schouten solitons of Lorentzian Lie groups with Yano connections, 2023, 15, 2836-3310, 763, 10.3934/cam.2023037 |