
We consider the existence and stability of Weyl almost periodic solutions for a class of quaternion-valued shunting inhibitory cellular neural networks with time-varying delays. In order to overcome the incompleteness of the space composed of Weyl almost periodic functions, we first obtain the existence of a bounded continuous solution of the system under consideration by using the fixed point theorem, and then prove that the bounded solution is Weyl almost periodic by using a variant of Gronwall inequality. Then we study the global exponential stability of the Weyl almost periodic solution by using the inequality technique. Even when the system we consider degenerates into a real-valued one, our results are new. A numerical example is given to illustrate the feasibility of our results.
Citation: Yongkun Li, Xiaoli Huang, Xiaohui Wang. Weyl almost periodic solutions for quaternion-valued shunting inhibitory cellular neural networks with time-varying delays[J]. AIMS Mathematics, 2022, 7(4): 4861-4886. doi: 10.3934/math.2022271
[1] | Xiaofang Meng, Yongkun Li . Pseudo almost periodic solutions for quaternion-valued high-order Hopfield neural networks with time-varying delays and leakage delays on time scales. AIMS Mathematics, 2021, 6(9): 10070-10091. doi: 10.3934/math.2021585 |
[2] | Nina Huo, Bing Li, Yongkun Li . Global exponential stability and existence of almost periodic solutions in distribution for Clifford-valued stochastic high-order Hopfield neural networks with time-varying delays. AIMS Mathematics, 2022, 7(3): 3653-3679. doi: 10.3934/math.2022202 |
[3] | Yanshou Dong, Junfang Zhao, Xu Miao, Ming Kang . Piecewise pseudo almost periodic solutions of interval general BAM neural networks with mixed time-varying delays and impulsive perturbations. AIMS Mathematics, 2023, 8(9): 21828-21855. doi: 10.3934/math.20231113 |
[4] | Qi Shao, Yongkun Li . Almost periodic solutions for Clifford-valued stochastic shunting inhibitory cellular neural networks with mixed delays. AIMS Mathematics, 2024, 9(5): 13439-13461. doi: 10.3934/math.2024655 |
[5] | Ardak Kashkynbayev, Moldir Koptileuova, Alfarabi Issakhanov, Jinde Cao . Almost periodic solutions of fuzzy shunting inhibitory CNNs with delays. AIMS Mathematics, 2022, 7(7): 11813-11828. doi: 10.3934/math.2022659 |
[6] | R. Sriraman, P. Vignesh, V. C. Amritha, G. Rachakit, Prasanalakshmi Balaji . Direct quaternion method-based stability criteria for quaternion-valued Takagi-Sugeno fuzzy BAM delayed neural networks using quaternion-valued Wirtinger-based integral inequality. AIMS Mathematics, 2023, 8(5): 10486-10512. doi: 10.3934/math.2023532 |
[7] | Jin Gao, Lihua Dai . Weighted pseudo almost periodic solutions of octonion-valued neural networks with mixed time-varying delays and leakage delays. AIMS Mathematics, 2023, 8(6): 14867-14893. doi: 10.3934/math.2023760 |
[8] | Jin Gao, Lihua Dai . Anti-periodic synchronization of quaternion-valued high-order Hopfield neural networks with delays. AIMS Mathematics, 2022, 7(8): 14051-14075. doi: 10.3934/math.2022775 |
[9] | Ailing Li, Mengting Lv, Yifang Yan . Asymptotic stability for quaternion-valued BAM neural networks via a contradictory method and two Lyapunov functionals. AIMS Mathematics, 2022, 7(5): 8206-8223. doi: 10.3934/math.2022457 |
[10] | Huahai Qiu, Li Wan, Zhigang Zhou, Qunjiao Zhang, Qinghua Zhou . Global exponential periodicity of nonlinear neural networks with multiple time-varying delays. AIMS Mathematics, 2023, 8(5): 12472-12485. doi: 10.3934/math.2023626 |
We consider the existence and stability of Weyl almost periodic solutions for a class of quaternion-valued shunting inhibitory cellular neural networks with time-varying delays. In order to overcome the incompleteness of the space composed of Weyl almost periodic functions, we first obtain the existence of a bounded continuous solution of the system under consideration by using the fixed point theorem, and then prove that the bounded solution is Weyl almost periodic by using a variant of Gronwall inequality. Then we study the global exponential stability of the Weyl almost periodic solution by using the inequality technique. Even when the system we consider degenerates into a real-valued one, our results are new. A numerical example is given to illustrate the feasibility of our results.
Since shunting inhibitory cellular neural networks were proposed by Bouzerdoum and Pinter [1] as a new type of neural networks, they have received more and more attention and have been widely applied in optimisation, psychophysics, speech and other fields. At the same time, since time delays are ubiquitous, many research results have been obtained on the dynamics of shunting inhibitory cellular neural networks with time delays [2,3,4,5,6].
On the one hand, the quaternion is a generalization of real and complex numbers [7]. The skew field of quaternions is defined by
H:={q=qR+iqI+jqJ+kqK}, |
where qR,qI,qJ,qK∈R and the elements i,j and k obey the Hamilton's multiplication rules:
ij=−jk=k,jk=−kj=i,ki=−ik=j,i2=j2=k2=−1. |
For q=qR+iqI+jqJ+kqK, we denote ˇq=iqI+jqJ+kqK and qR=q−ˇq. The norm of q is defined by ‖q‖H=√(qR)2+(qI)2+(qJ)2+(qK)2. For y=(y1,y2,⋯,yn)T∈Hn, we define ‖y‖Hn=max1≤p≤n{‖y‖H}, then (Hn,‖⋅‖Hn) is a Banach space. As we all know, quaternion-valued neural networks include real-valued neural networks and complex-valued neural networks as their special cases. Compared with complex-valued neural networks, quaternion-valued neural networks only needs half of the connection weight parameters of complex-valued neural networks when dealing with multi-level information [8]. In recent years, quaternion-valued neural networks have attracted the attention of many researchers, and their various dynamic behaviors, including fractional-order and stochastic quaternion-valued neural networks, have been extensively studied [9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24].
On the other hand, because periodic and almost periodic oscillations are important dynamics of neural networks, the periodic and almost periodic oscillations of neural networks have been studied a lot in the past few decades [25,26,27,28,29,30,31,32,33]. Weyl almost periodicity is a generalization of Bohr almost periodicity and Stepanov almost periodicity [34,35,36,37]. It is a more complex recurrent oscillation. Because the spaces composed of Bohr almost periodic functions and Stepanov almost periodic functions are Banach spaces, it brings some convenience to study the existence of almost periodic solutions in these two senses of differential equations. Therefore, many results have been obtained on the Bohr almost periodic oscillation and Stepanov almost periodic oscillation of neural networks. However, the space composed of Weyl almost periodic functions is incomplete [38]. Therefore, the results of Weyl almost periodic solutions of neural networks are still very rare. Therefore, it is a meaningful and challenging work to study the existence of Weyl almost periodic solutions of neural networks.
Motivated by the above, in this paper, we consider the following shunting inhibitory cellular neural networks with time-varying delays:
˙xij(t)=−aij(t)xij(t)−∑Ckl∈Nr(i,j)Bklij(t)fij(xkl(t))xij(t)−∑Ckl∈Ns(i,j)Cklij(t)gij(xkl(t−τkl(t)))xij(t)+Iij(t), | (1.1) |
where ij∈{11,12,…,1n,…,m1,m2,…,mn}:=Λ, Cij denotes the cell at the (i,j) position of the lattice. The r-neighborhood Nr(i,j) of Cij is given as
Nr(i,j)={Ckl:max(|k−i|,|l−j|)≤r,1≤k≤m,1≤l≤n}, |
and Ns(i,j) is similarly specified; xij(t)∈H denotes the activity of the cell of Cij, Iij(t)∈H is the external input to Cij, aij(t)∈H is the coefficient of the leakage term, which represents the passive decay rate of the activity of the cell Cij, Bklij(t)≥0 and Cklij(t)≥0 represent the connection or coupling strength of postsynaptic of activity of the cell transmitted to the cell Cij, the activity functions fij,gij:H→H are continuous functions representing the output or firing rate of the cell Cij, τkl(t) corresponds to the transmission delay and satisfies 0≤τkl(t)≤τ.
The purpose of this paper is to use the fixed point theorem and a variant of Gronwall inequality to establish the existence and global exponential stability of Weyl almost periodic solutions for a class of quaternion-valued shunting inhibitory cellular neural networks whose coefficients of the leakage terms are quaternions. This is the first paper to study the existence and global exponential stability of Weyl almost periodic solutions of system (1.1) by using the fixed point theorem and a variant of Gronwall inequality. Our result of this paper is new, and our method can be used to study other types of quaternion-valued neural networks.
For convenience, we introduce the following notations:
am=minij∈Λ{inft∈R{aij(t)}},aM=maxij∈Λ{supt∈R{aij(t)}},ˇaMij=supt∈R‖ˇaij(t)‖H, |
BklMij=supt∈R{Bklij(t)},CklMij=supt∈R{Cklij(t)},τMij=supt∈R{τij(t)},τ′=maxkl∈Λ{supt∈R{τ′kl(t)}}. |
The initial condition of system (1.1) is given by
xij(s)=φij(s),s∈[−τ,0], |
where φij∈C(R,H),ij∈Λ.
Throughout this paper, we assume that:
(H1) For ij,kl∈Λ, functions aRij,Bklij,Cklij∈AP(R,R+), ˇaij∈AP(R,H), Iij∈APWp(R,H), τkl∈AP(R,R+)∩C1(R,R) and τ′<1.
(H2) For ij∈Λ, there exist positive constants Lfij and Lgij such that for all x,y∈H,
‖fij(x)−fij(y)‖H≤Lfij‖x−y‖H,‖gij(x)−gij(y)‖H≤Lgij‖x−y‖H, |
and fij(0)=gij(0)=0.
The rest of this paper is arranged as follows. In Section 2, we introduce some definitions and lemmas. In Section 3, we study the existence and global exponential stability of Weyl almost periodic solutions of (1.1). In Section 4, an example is given to verify the theoretical results. This paper ends with a brief conclusion in Section 5.
Let (X,‖⋅‖X) be a Banach space and BC(R,X) be the set of all bounded continuous functions from R to X.
Definition 2.1. [38] A function f∈BC(R,X) is said to be almost periodic, if for every ϵ>0, there exists a constant l=l(ϵ)>0 such that in every interval of length l(ϵ) contains at least one σ such that
‖f(t+σ)−f(t)‖X<ϵ,t∈R. |
Denote by AP(R,X) the set of all such functions.
For p∈[1,∞), we denote by Lploc(R,X) the space of all functions from R into X which are locally p-integrable. For f∈Lploc(R,X), we define the following seminorm:
‖f‖Wp=limr→+∞supβ∈R(1r∫β+rβ‖f(t)‖pXdt)1p. |
Definition 2.2. [38] A function f∈Lploc(R,X) is said to be p-th Weyl almost periodic (Wp-almost periodic for short), if for every ϵ>0, there exists a constant l=l(ϵ)>0 such that in every interval of length l(ϵ) contains at least one σ such that
‖f(t+σ)−f(t)‖Wp<ϵ. |
This σ is called on ϵ-translation number of f. The set of all such functions will be denoted by APWp(R,X).
Remark 2.1. By Definitions 2.1 and 2.2, it is easy to see that if f∈AP(R,X), then f∈APWp(R,X).
Similar to the proofs of the lemma on page 83 and the lemma on page 84 of [39], it is not difficult to prove the following two lemmas.
Lemma 2.1. If f∈APWp(R,X), then f is bounded and uniformly continuous on R withrespect to the seminorn ‖⋅‖Wp.
Using the argumentation contained in the proof of Proposition 3.21 in [38], one can easily prove the following.
Lemma 2.2. If fk∈APWp(R,X), k=1,2,…,n. Then, for every ϵ>0, there exist common ϵ-translation numbers for these functions.
Lemma 2.3. [40]Let g:R→R be a continuous function such that, for every t∈R,
0≤g(t)≤ρ(t)+γ1∫t−∞e−η1(t−s)g(s)ds+⋯+γn∫t−∞e−ηn(t−s)g(s)ds | (2.1) |
for some locally integrable function ρ:R→R, and for some constantsγ1,…,γn≥0, and some constants η1,…,ηn≥γ, whereγ=n∑p=1γp. We assume that the integrals in the right hand side of (2.1)are convergent. Let η=min{η1,…,ηn}. Then, for every ξ∈(0,η−γ]such that ∫0−∞eξsρ(s)ds converges, we have, for every t∈R,
g(t)≤ρ(t)+γ∫t−∞e−ξ(t−s)ρ(s)ds. |
In particular, if ρ(t) is constant, we have
g(t)≤ρηη−γ. |
Let BUC(R,Hm×n) be a collection of bounded and uniformly continuous functions from R to Hm×n, then, the space BUC(R,Hm×n) with the norm ‖x‖∞=supt∈R‖x(t)‖Hm×n is a Banach space, where x∈BUC(R,Hm×n).
Denote ϕ0=(ϕ011,⋯,ϕ01n,ϕ021,⋯,ϕ02n,⋯,ϕ0m1,⋯,ϕ0mn)T, where
ϕ0ij(t)=∫t−∞e−∫tsaRij(v)dvIij(s)ds,ij∈Λ. |
We will show that ϕ0 is well defined under assumption (H1). In fact, by Iij∈APWp(R,H) and Lemma 2.1, there exists a constant M>0 such that ‖Iij‖Wp≤M for all ij∈Λ. According to the Hölder inequality, one has
‖ϕ0ij(t)‖H≤‖∫t−∞e−am(t−s)Iij(s)ds‖H≤∞∑r=0(∫t−rt−(r+1)e−amq(t−s)ds)1q(∫t−rt−(r+1)‖Iij(s)‖pHds)1p≤∞∑r=0e−amrM<+∞, | (3.1) |
where 1p+1q=1, which means that ϕ0 is well defined.
Take a positive constant α≥‖ϕ0‖∞. Let
Ω={ϕ∈BUC(R,Hm×n)|‖ϕ−ϕ0‖∞≤α}. |
Then, for every ϕ∈Ω, one has
‖ϕ‖∞≤‖ϕ−ϕ0‖∞+‖ϕ0‖∞≤2α. |
Theorem 3.1. Assume (H1)–(H2) hold. Furthermore, suppose that
(H3)
κ=maxij∈Λ{2am[ˇaMij+∑Ckl∈Nr(i,j)2BklMijLfijα+∑Ckl∈Ns(i,j)2CklMijLgijα]}<1, |
(H4)for p>2,
maxij∈Λ{24(2p−4amp)p−2(4amp)2[2(ˇaMij)p+2(∑Ckl∈Nr(i,j)2BklMijLfijα)p+(1+2ep4amτ1−τ′)(∑Ckl∈Ns(i,j)2CklMijLgijα)p]}<1, |
and, for p=2,
maxij∈Λ{24(2am)2[2(ˇaMij)2+2(∑Ckl∈Nr(i,j)2BklMijLfijα)2+(1+2e12amτ1−τ′)(∑Ckl∈Ns(i,j)2CklMijLgijα)2]}<1, |
then system (1.1) has a unique Wp-almost periodic solution in Ω.
Proof. It is easy to check that if x=(x11,⋯,x1n,x21,⋯,x2n,⋯,xm1,⋯,xmn)T∈Ω is a solution of the integral equation
xij(t)=∫t−∞e−∫tsaRij(v)dv[−ˇaij(s)xij(s)−∑Ckl∈Nr(i,j)Bklij(s)fij(xkl(s))xij(s)−∑Ckl∈Ns(i,j)Cklij(s)gij(xkl(s−τkl(s)))xij(s)+Iij(s)]ds,ij∈Λ, | (3.2) |
then x is a solution of system (1.1).
Define an operator T:Ω→Hm×n by
(Tϕ)(t)=((T11ϕ)(t),⋯,(T1nϕ)(t),(T21ϕ)(t),⋯,(T2nϕ)(t),⋯,(Tm1ϕ)(t),⋯,(Tmnϕ)(t))T, |
where
(Tijϕ)(t)=∫t−∞e−∫tsaRij(v)dv[−ˇaij(s)ϕij(s)−∑Ckl∈Nr(i,j)Bklij(s)fij(ϕkl(s))ϕij(s)−∑Ckl∈Ns(i,j)Cklij(s)gij(ϕkl(s−τkl(s)))ϕij(s)+Iij(s)]ds,ij∈Λ. |
Now, we will prove that Tϕ is well defined. Actually, by (H1)–(H3) and (3.1), for ij∈Λ, one deduces that
‖(Tijϕ)(t)‖H≤∫t−∞e−∫tsaRij(v)dv‖−ˇaij(s)ϕij(s)−∑Ckl∈Nr(i,j)Bklij(s)(fij(ϕkl(s))−fij(0))ϕij(s)−∑Ckl∈Ns(i,j)Cklij(s)(gij(ϕkl(s−τkl(s)))−gij(0))ϕij(s)‖Hds+∫t−∞e−∫tsaRij(v)dv‖Iij(s)‖Hds≤1am(ˇaMij+∑Ckl∈Nr(i,j)BklMijLfij‖ϕ‖∞+∑Ckl∈Ns(i,j)CklMijLgij‖ϕ‖∞)‖ϕ‖∞+∫t−∞e−∫tsaRij(v)dv‖Iij(s)‖Hds<+∞. | (3.3) |
That is, Tϕ is well defined.
We will divide the rest of the proof into four steps.
Step 1, we will prove that Tϕ∈BUC(R,Hm×n), for every ϕ∈Ω.
In fact, by (3.3), we see that Tϕ is bounded on R. So, we only need to show that Tϕ is uniformly continuous on R. Based on the Hölder inequality for 0≤h≤1 and q≥1 with 1p+1q=1, one has
‖(Tijϕ)(t+h)−(Tijϕ)(t)‖H=‖∫t+h−∞e−∫t+hsaRij(v)dv(−ˇaij(s)ϕij(s)−∑Ckl∈Nr(i,j)Bklij(s)fij(ϕkl(s))ϕij(s)−∑Ckl∈Ns(i,j)Cklij(s)×gij(ϕkl(s−τkl(s)))ϕij(s)+Iij(s))ds−∫t−∞e−∫tsaRij(v)dv(−ˇaij(s)ϕij(s)−∑Ckl∈Nr(i,j)Bklij(s)×fij(ϕkl(s))ϕij(s)−∑Ckl∈Ns(i,j)Cklij(s)gij(ϕkl(s−τkl(s)))ϕij(s)+Iij(s))ds‖H≤∫t−∞|e−∫t+hsaRij(v)dv−e−∫tsaRij(v)dv|‖ˇaij(s)ϕij(s)+∑Ckl∈Nr(i,j)Bklij(s)fij(ϕkl(s))ϕij(s)+∑Ckl∈Ns(i,j)Cklij(s)×gij(ϕkl(s−τkl(s)))ϕij(s)‖Hds+∫t−∞|e−∫t+hsaRij(v)dv−e−∫tsaRij(v)dv|‖Iij(s)‖Hds+∫t+hte−∫t+hsaRij(v)dv‖ˇaij(s)ϕij(s)+∑Ckl∈Nr(i,j)Bklij(s)fij(ϕkl(s))ϕij(s)+∑Ckl∈Ns(i,j)Cklij(s)×gij(ϕkl(s−τkl(s)))ϕij(s)‖Hds+∫t+hte−∫t+hsaRij(v)dv‖Iij(s)‖Hds≤(ˇaMij+∑Ckl∈Nr(i,j)BklMijLfij‖ϕ‖∞+∑Ckl∈Ns(i,j)CklMijLgij‖ϕ‖∞)‖ϕ‖∞∫t−∞(e−∫tsaRij(v)dv∫t+htaRij(v)dv)ds+∫t−∞(e−∫tsaRij(v)dv∫t+htaRij(v)dv)‖Iij(s)‖Hds+(ˇaMij+∑Ckl∈Nr(i,j)BklMijLfij‖ϕ‖∞+∑Ckl∈Ns(i,j)CklMijLgij‖ϕ‖∞)‖ϕ‖∞∫t+hte−∫tsaRij(v)dvds+∫t+hte−∫tsaRij(v)dv‖Iij(s)‖Hds≤aMh(ˇaMij+∑Ckl∈Nr(i,j)BklMijLfij‖ϕ‖∞+∑Ckl∈Ns(i,j)CklMijLgij‖ϕ‖∞)‖ϕ‖∞∫t−∞e−am(t−s)ds+aMh∫t−∞e−∫tsaRij(v)dv‖Iij(s)‖Hds+(ˇaMij+∑Ckl∈Nr(i,j)BklMijLfij‖ϕ‖∞+∑Ckl∈Ns(i,j)CklMijLgij‖ϕ‖∞)‖ϕ‖∞×∫t+hte−am(t−s)ds+(∫t+hte−qam(t−s)ds)1q(∫t+ht‖Iij(s)‖pHds)1p≤aMam(ˇaMij+∑Ckl∈Nr(i,j)BklMijLfij‖ϕ‖∞+∑Ckl∈Ns(i,j)CklMijLgij‖ϕ‖∞)‖ϕ‖∞h+aMh∫t−∞e−∫tsaRij(v)dv‖Iij(s)‖Hds+eamh(ˇaMij+∑Ckl∈Nr(i,j)BklMijLfij‖ϕ‖∞+∑Ckl∈Ns(i,j)CklMijLgij‖ϕ‖∞)‖ϕ‖∞h+eamh‖Iij‖Wph, |
where ij∈Λ. Hence, letting h→0+, by (3.1), we have
‖(Tijϕ)(t+h)−(Tijϕ)(t)‖H→0, |
which means that (Tijϕ) is uniformly continuous on R, ij∈Λ. Therefore, Tϕ∈BUC(R,Hm×n).
Step 2, we will prove that T is a self-mapping from Ω to Ω.
Actually, for arbitrary ϕ∈Ω, from (H2)–(H3), we have
‖Tϕ−ϕ0‖∞≤supt∈R{maxij∈Λ[∫t−∞e−∫tsaRij(v)dv‖ˇaij(s)ϕij(s)+∑Ckl∈Nr(i,j)Bklij(s)(fij(ϕkl(s))−f(0))ϕij(s)+∑Ckl∈Ns(i,j)Cklij(s)(gij(ϕkl(s−τkl(s)))−gij(0))ϕij(s)‖Hds]}≤maxij∈Λ{1am[ˇaMij+∑Ckl∈Nr(i,j)2BklMijLfijα+∑Ckl∈Ns(i,j)2CklMijLgijα]}‖ϕ‖∞≤κα≤α, |
which implies that Tϕ∈Ω. Consequently, T is a self-mapping from Ω to Ω.
Step 3, we will prove T is a contraction mapping.
As a matter of fact, in view of (H1)–(H2), for any ϕ,ν∈Ω, we can get
‖Tϕ−Tν‖∞≤supt∈R{maxij∈Λ[∫t−∞e−am(t−s)(ˇaMij‖ϕij(s)−νij(s)‖H+∑Ckl∈Nr(i,j)BklMij‖fij(ϕkl(s))(ϕij(s)−νij(s))+(fij(ϕkl(s))−fij(νkl(s)))νij(s)‖H+∑Ckl∈Ns(i,j)CklMij‖gij(ϕkl(s−τkl(s)))(ϕij(s)−νij(s))+(gij(ϕkl(s−τkl(s)))−gij(νkl(s−τkl(s))))νij(s)‖H)ds]}≤supt∈R{maxij∈Λ[∫t−∞e−am(t−s)[ˇaMij‖ϕij(s)−νij(s)‖H+∑Ckl∈Nr(i,j)BklMij(2Lfijα‖ϕij(s)−νij(s)‖H+2Lfijα‖ϕkl(s)−νkl(s)‖H)+∑Ckl∈Ns(i,j)CklMij(2Lgijα‖ϕij(s)−νij(s)‖H2Lgijα‖ϕkl(s−τkl(s))−νkl(s−τkl(s))‖H)]ds]}≤maxij∈Λ{1am[ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)4CklMijLgijα]}‖ϕ−ν‖∞. |
From this and (H3), one has
||Tϕ−Tν||∞≤κ‖ϕ−ν‖∞. |
Noticing that κ<1, T is a contraction mapping. Consequently, system (1.1) has a unique solution x in Ω.
Step 4, we will prove that the unique solution x∈Ω is Wp-almost periodic.
Indeed, since x=(x11,⋯,x1n,x21,⋯,x2n,⋯,xm1,⋯,xmn)T∈Ω, x is bounded and uniformly continuous. Hence, for every ϵ>0, there exists a δ∈(0,ϵ) such that for any t1,t2∈R with |t1−t2|<δ and ij∈Λ, we have
‖xij(t1)−xij(t2)‖H<ϵ. | (3.4) |
Also, for this δ, in view of (H1) and Lemma 2.2, we see that there exists a common δ-translation number σ such that
limr→+∞supβ∈R(1r∫β+rβ‖Iij(t+σ)−Iij(t)‖pHdt)1p<δ<ϵ, | (3.5) |
|Bklij(t+σ)−Bklij(t)|<ϵ, | (3.6) |
|Cklij(t+σ)−Cklij(t)|<ϵ, | (3.7) |
|aRij(t+σ)−aRij(t)|<ϵ,‖ˇaij(t+σ)−ˇaij(t)‖H<ϵ | (3.8) |
and
|τij(t+σ)−τij(t)|<δ, | (3.9) |
where ij,kl∈Λ. Consequently, from (3.4) and (3.9), we get
‖xij(t−τij(t+σ))−xij(t−τij(t))‖H<ϵ. | (3.10) |
Since x is a solution of system (1.1), by (3.2), for ij∈Λ, we have
‖xij(t+σ)−xij(t)‖H≤‖∫t−∞e−∫tsaRij(v+σ)dv(ˇaij(s+σ)xij(s+σ)−ˇaij(s)xij(s))ds‖H+‖∫t−∞e−∫tsaRij(v+σ)dv∑Ckl∈Nr(i,j)[Bklij(s+σ)fij(xkl(s+σ))xij(s+σ)−Bklij(s)fij(xkl(s))xij(s)]ds‖H+‖∫t−∞e−∫tsaRij(v+σ)dv∑Ckl∈Ns(i,j)[Cklij(s+σ)×gij(xkl(s+σ−τkl(s+σ)))xij(s+σ)−Cklij(s)gij(xkl(s−τkl(s)))xij(s)]ds‖H+‖∫t−∞e−∫tsaRij(v+σ)dv[Iij(s+σ)−Iij(s)]ds‖H+‖∫t−∞(e−∫tsaRij(v+σ)dv−e−∫tsaRij(v)dv)ˇaij(s)xij(s)ds‖H+‖∫t−∞(e−∫tsaRij(v+σ)dv−e−∫tsaRij(v)dv)∑Ckl∈Nr(i,j)Bklij(s)fij(xkl(s))xij(s)ds‖H+‖∫t−∞(e−∫tsaRij(v+σ)dv−e−∫tsaRij(v)dv)∑Ckl∈Ns(i,j)Cklij(s)gij(xkl(s−τkl(s)))xij(s)ds‖H+‖∫t−∞(e−∫tsaRij(v+σ)dv−e−∫tsaRij(v)dv)Iij(s)ds‖H:=8∑l=1Blij(t). | (3.11) |
When p>2, it follows from Hölder's inequality (2p,p−2p), Hölder's inequality (12,12) and (H2) that
B2ij(t)≤∫t−∞e−am(t−s)∑Ckl∈Nr(i,j)‖Bklij(s+σ)fij(xkl(s+σ))(xij(s+σ)−xij(s))‖Hds+∫t−∞e−am(t−s)∑Ckl∈Nr(i,j)‖Bklij(s+σ)(fij(xkl(s+σ))−fij(xkl(s)))xij(s)‖Hds+∫t−∞e−am(t−s)∑Ckl∈Nr(i,j)‖(Bklij(s+σ)−Bklij(s))fij(xkl(s))xij(s)‖Hds≤(∫t−∞e−p2p−4am(t−s)ds)p−2p[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xij(s+σ)−xij(s))‖H)p2ds]2p+(∫t−∞e−p2p−4am(t−s)ds)p−2p[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xkl(s+σ)−xkl(s)‖H)p2ds]2p+(∫t−∞e−p2p−4am(t−s)ds)p−2p[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)4Lfijα2|Bklij(s+σ)−Bklij(s)|ds)p2ds]2p≤(2p−4amp)p−2p{[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xij(s+σ)−xij(s))‖H)p2ds]2p+[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xkl(s+σ))−xkl(s)‖H)p2ds]2p+[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)4Lfijα2|Bklij(s+σ)−Bklij(s)|ds)p2ds]2p}≤(2p−4amp)p−2p{[(∫t−∞e−p4am(t−s)ds)12(∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xij(s+σ)−xij(s))‖H)pds)12]2p+[(∫t−∞e−p4am(t−s)ds)12(∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα×‖xkl(s+σ))−xkl(s)‖H)pds)12]2p+[(∫t−∞e−p4am(t−s)ds)12(∫t−∞e−p4am(t−s)×(∑Ckl∈Nr(i,j)4Lfijα2|Bklij(s+σ)−Bklij(s)|)pds)12]2p}≤(2p−4amp)p−2p(4amp)1p{[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xij(s+σ)−xij(s))‖H)pds]1p+[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xkl(s+σ))−xkl(s)‖H)pds]1p+[∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)4Lfijα2|Bklij(s+σ)−Bklij(s)|)pds]1p}, | (3.12) |
for ij∈Λ. Similarly, we have
B1ij(t)≤(2p−4amp)p−2p(4amp)1p{ˇaMij[∫t−∞e−p4am(t−s)‖xij(s+σ)−xij(s)‖pHds]1p+2α[∫t−∞e−p4am(t−s)‖ˇaij(s+σ)−˘aij(s)‖pHds]1p}, | (3.13) |
B3ij(t)≤(2p−4amp)p−2p(4amp)1p{[∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xij(s+σ)−xij(s))‖H)pds]1p+[∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xkl(s+σ−τkl(s+σ))−xkl(s−τkl(s))‖H)pds]1p+[∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)4Lgijα2|Cklij(s+σ)−Cklij(s))|)pds]1p} | (3.14) |
and
B4ij(t)≤(2p−4amp)p−2p(4amp)1p(∫t−∞e−p4am(t−s)‖Iij(s+σ)−Iij(s)‖pHds)1p, | (3.15) |
for ij∈Λ.
Besides, combining with Hölder's inequality (2p,p−2p), Hölder's inequality (12,12) and (H2) that
B5ij(t)≤2ˇaMijα∫t−∞e−am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)ds≤2ˇaMijα(∫t−∞e−p2p−4am(t−s)ds)p−2p[∫t−∞e−p4am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)p2ds]2p≤(2p−4amp)p−2p2ˇaMijα[(∫t−∞e−p4am(t−s)ds)12(∫t−∞e−p4am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)pds)12]2p≤(2p−4amp)p−2p(4amp)1p2ˇaMijα[∫t−∞e−p4am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)pds]1p, | (3.16) |
for ij∈Λ. In a similar way, one can get that
B6ij(t)≤(2p−4amp)p−2p(4amp)1p∑Ckl∈Nr(i,j)4BklMijLfijα2[∫t−∞e−p4am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)pds]1p, | (3.17) |
B7ij(t)≤(2p−4amp)p−2p(4amp)1p∑Ckl∈Ns(i,j)4CklMijLgijα2[∫t−∞e−p4am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)pds]1p | (3.18) |
and
B8ij(t)≤(2p−4amp)p−2p(4amp)1p[∫t−∞e−p4am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)p‖Iij(s)‖pHds]1p, | (3.19) |
for ij∈Λ. Hence, together with a change of variables, Fubini's theorem, Hölder's inequality, (3.8) and (3.13), we derive that
1r∫β+rβBp1ij(t)dt≤(2p−4amp)p−24amp{1r∫β+rβ[[ˇaMij∫t−∞e−p4am(t−s)‖xij(s+σ)−xij(s)‖pHds]1p+2α[∫t−∞e−p4am(t−s)‖ˇaij(s+σ)−˘aij(s)‖pHds]1p]pdt}≤(2p−4amp)p−28amp{(ˇaMij)p1r∫β+rβ∫t−∞e−p4am(t−s)‖xij(s+σ)−xij(s)‖pHdsdt+(2α)p1r∫β+rβ∫t−∞e−p4am(t−s)‖ˇaij(s+σ)−˘aij(s)‖pHdsdt}≤(2p−4amp)p−28amp{(ˇaMij)p∫β−∞e−p4am(β−s)(1r∫s+rs‖xij(t+σ)−xij(t)‖pHdt)ds+(2α)p∫β−∞e−p4am(β−s)(1r∫s+rs‖ˇaij(t+σ)−˘aij(t)‖pHdt)ds}=(ˇaMij)p(2p−4amp)p−28amp∫β−∞e−p4am(β−s)Θσ,r(s)ds+ρ1ij, |
where
Θσ,r(s):=1r∫s+rs‖x(t+σ)−x(t)‖pHndt |
and
ρ1ij:=2(2p−4amp)p−2(4amp)2(2αϵ)p, | (3.20) |
and, together with a change of variables, Fubini's theorem, Hölder's inequality, (3.6) and (3.12), we derive that
1r∫β+rβBp2ij(t)dt≤(2p−4amp)p−24amp{1r∫β+rβ[(∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xij(s+σ)−xij(s))‖H)pds)1p+(∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xkl(s+σ)−xkl(s)‖H)pds)1p+(∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)4Lfijα2|Bklij(s+σ)−Bklij(s)|)pds)1p]pdt}≤(2p−4amp)p−212amp{1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xij(s+σ)−xij(s)‖H)pdsdt+1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)2BklMijLfijα‖xkl(s+σ)−xkl(s)‖H)pdsdt+1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Nr(i,j)4Lfijα2|Bklij(s+σ)−Bklij(s)|)pdsdt}≤(2p−4amp)p−212amp[(∑Ckl∈Nr(i,j)2BklMijLfijα)p1r∫β+rβ∫t−∞e−p4am(t−s)‖x(s+σ)−x(s)‖pHm×ndsdt+(∑Ckl∈Nr(i,j)2BklMijLfijα)p1r∫β+rβ∫t−∞e−p4am(t−s)‖x(s+σ)−x(s)‖Hm×ndsdt+1r∫β+rβ∫t−∞e−p4am(t−s)(4mnLfijα2ϵ)pdsdt]≤(2p−4amp)p−212amp[(∑Ckl∈Nr(i,j)2BklMijLfijα)p∫β−∞e−p4am(β−s)(1r∫s+rs‖x(t+σ)−x(t)‖pHm×ndt)ds+(∑Ckl∈Nr(i,j)2BklMijLfijα)p∫β−∞e−p4am(β−s)(1r∫s+rs‖x(t+σ)−x(t)‖Hm×ndt)ds+4amp(4mnLfijα2ϵ)p]=(2p−4amp)p−224amp(∑Ckl∈Nr(i,j)2BklMijLfijα)p∫β−∞e−p4am(β−s)Θσ,r(s)ds+ρ2ij, |
where
ρ2ij:=3(2p−4amp)p−2(4amp)2(4mnLfijα2ϵ)p. | (3.21) |
Moreover, based on a change of variables, Fubini's theorem, Hölder's inequality, (3.7), (3.10) and (3.14), we deduce that
1r∫β+rβBp3ij(t)dt≤(2p−4amp)p−24amp{1r∫β+rβ[(∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xij(s+σ)−xij(s))‖H)pds)1p+(∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xkl(s+σ−τkl(s+σ))−xkl(s−τkl(s))‖H)pds)1p+(∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)4Lgijα2‖(Cklij(s+σ)−Cklij(s))‖H)pds)1p]pdt}≤(2p−4amp)p−212amp{1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xij(s+σ)−xij(s))‖H)pdsdt+1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xkl(s+σ−τkl(s+σ))−xkl(s−τkl(s))‖H)pdsdt+1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)4Lgijα2|Cklij(s+σ)−Cklij(s))|)pdsdt}≤(2p−4amp)p−212amp{1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xij(s+σ)−xij(s))‖H)pdsdt+1r∫β+rβ[2∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xkl(s+σ−τkl(s+σ))−xkl(s−τkl(s+σ))‖H)pds+2∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)4CklMijLgijα2‖xkl(s−τkl(s+σ))−xkl(s−τkl(s))‖H)pds]dt+1r∫β+rβ∫t−∞e−p4am(t−s)(4mnLgijα2ϵ)pdsdt}≤(2p−4amp)p−212amp{1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xij(s+σ)−xij(s))‖H)pdsdt+1r∫β+rβ[21−τ′∫t−τkl(t+σ)−∞e−p4am(t−u−τ)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xkl(u+σ)−xkl(u)‖H)pdu+2(∑Ckl∈Ns(i,j)2CklMijLgijαϵ)p∫t−∞e−p4am(t−s)ds]dt+4amp(4mnLgijα2ϵ)p}≤(2p−4amp)p−212amp{1r∫β+rβ∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xij(s+σ)−xij(s))‖H)pdsdt+1r∫β+rβ[2ep4amτ1−τ′∫t−∞e−p4am(t−s)(∑Ckl∈Ns(i,j)2CklMijLgijα‖xkl(s+σ)−xkl(s)‖H)pds]dt+8amp(∑Ckl∈Ns(i,j)2CklMijLgijαϵ)p+4amp(4mnLgijα2ϵ)p}≤(2p−4amp)p−212amp[(∑Ckl∈Ns(i,j)2CklMijLgijα)p∫β−∞e−p4am(β−s)(1r∫s+rs‖x(t+σ)−x(t))‖Hm×ndt)ds+2ep4amτ1−τ′(∑Ckl∈Ns(i,j)2CklMijLgijα)p∫β−∞e−p4am(β−s)(1r∫s+rs‖x(t+σ)−x(t)‖pHm×ndt)ds+8amp(∑Ckl∈Ns(i,j)2CklMijLgijαϵ)p+4amp(4mnLgijα2ϵ)p]=(2p−4amp)p−212amp(1+2ep4amτ1−τ′)(∑Ckl∈Ns(i,j)2CklMijLgijα)p∫β−∞e−p4am(β−s)Θσ,r(s)ds+ρ3ij, |
where
ρ3ij:=3(2p−4amp)p−2(4amp)2[2(∑Ckl∈Ns(i,j)2CklMijLgijα)p+(4mnLgijα2)p]ϵp. | (3.22) |
In view of (3.5), (3.8) and (3.15)–(3.19), we can easily obtain that
1r∫β+rβBp4ij(t)dt≤(2p−4amp)p−24amp∫β−∞e−p4am(β−s)(1r∫s+rs‖Iij(t+σ)−Iij(t)‖pHdt)ds≤(2p−4amp)p−2(4amp)2ϵp:=ρ4ij, | (3.23) |
1r∫β+rβBp5ij(t)dt≤(2p−4amp)p−24amp(2ˇaMijα)p[1r∫β+rβ∫t−∞e−p4am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)pdsdt]≤(2p−4amp)p−24amp(2ˇaMijαϵ)p∫∞0e−p4amsspds:=ρ5ij, | (3.24) |
1r∫β+rβBp6ij(t)dt≤(2p−4amp)p−24amp(∑Ckl∈Nr(i,j)4BklMijLfijα2)p[1r∫β+rβ∫t−∞e−p4am(t−s)×(∫ts|aRij(v+σ)−aRij(v)|dv)pdsdt]≤(2p−4amp)p−24amp(∑Ckl∈Nr(i,j)4BklMijLfijα2ϵ)p∫∞0e−p4amsspds:=ρ6ij, | (3.25) |
1r∫β+rβBp7ij(t)dt≤(2p−4amp)p−24amp(∑Ckl∈Ns(i,j)4CklMijLgijα2)p[1r∫β+rβ∫t−∞e−p4am(t−s)×(∫ts|aRij(v+σ)−aRij(v)|dv)pdsdt]≤(2p−4amp)p−24amp(∑Ckl∈Ns(i,j)4CklMijLgijα2ϵ)p∫∞0e−p4amsspds:=ρ7ij | (3.26) |
and
1r∫β+rβBp8ij(t)dt≤(2p−4amp)p−24amp[1r∫β+rβ∫t−∞e−p4am(t−s)(∫ts|aRij(v+σ)−aRij(v)|dv)p‖Iij(s)‖pHdsdt]≤ϵp(2p−4amp)p−24amp[1r∫β+rβ∫t−∞e−p4am(t−s)(t−s)p‖Iij(s)‖pHdsdt]≤(2p−4amp)p−24amp‖Iij‖pWpϵp∫∞0e−p4amsspds:=ρ8ij, | (3.27) |
for ij∈Λ. Consequently, combining with (3.11) and (3.20)–(3.27), we obtain
Θσ,r(β)≤maxij∈Λ{88∑l=11r∫β+rβBplij(t)dt}≤ρ+γ∫β−∞e−η(β−s)Θσ,r(s)ds, |
where η=p4am,
ρ=86∑l=1maxij∈Λ{ρlij}=8maxij∈Λ{(2p−4amp)p−24amp[8amp(2α)p+12amp(4mnLfijα2)p+24amp(∑Ckl∈Ns(i,j)2CklMijLgijα)P+12amp(4mnLgijα2)p+4amp+[(2ˇaMijα)p+(∑Ckl∈Nr(i,j)4BklMijLfijα2)p+(∑Ckl∈Ns(i,j)4CklMijLgijα2)p+‖Iij‖pWp]∫∞0e−p4amsspds]}ϵp |
and
γ=8maxij∈Λ{(2p−4amp)p−212amp[23(ˇaMij)p+2(∑Ckl∈Nr(i,j)2BklMijLfijα)p+(1+2ep4amτ1−τ′)(∑Ckl∈Ns(i,j)2CklMijLgijα)p]}. |
By (H4), we have γ<η. Thus, it follows from Lemma 2.3 that
1r∫β+rβ‖x(t+σ)−x(t)‖pHm×ndt≤ρηη−γ. |
Hence, x∈APWp(R,Hm×n).
When p=2, similar to the proof of the case of p>2, one can obtain
Θσ,r(β)≤˜ρ+˜γ∫β−∞e−˜η(β−s)Θσ,r(s)ds, |
where ˜η=am2,
˜ρ=88∑l=1maxij∈Λ{˜ρlij}=maxij∈Λ{16am[4am(2α)2+6am(4mnLfijα2ϵ)2+12am(2∑Ckl∈Ns(i,j)CklMijLgijα)2+6am(4mnLgijα2)2+2am+[(2ˇaMijα)2+(∑Ckl∈Nr(i,j)4BklMijLfijα2)2+(∑Ckl∈Ns(i,j)4CklMijLgijα2)2+‖Iij‖2W2]∫∞0e−12amss2ds]}ϵ2 |
and
˜γ=maxij∈Λ{48am[23(ˇaMij)2+2(∑Ckl∈Nr(i,j)2BklMijLfijα)2+(1+2e12amτ1−τ′)(∑Ckl∈Ns(i,j)2CklMijLgijα)2]}. |
By (H4), we have ˜γ<˜η. Thus, it follows from Lemma 2.3 that
1r∫β+rβ‖x(t+σ)−x(t)‖pHm×ndt≤˜ρ˜η˜η−˜γ, |
which means that x∈APW2(R,Hm×n). The proof is complete.
Definition 3.1. [14] Let x be a solution of system (1.1) with the initial value φ and y be an arbitrary solution of system (1.1) with the initial value ψ, respectively. If there exist positive constants λ and M such that
‖x(t)−y(t)‖Hm×n≤Me−λt‖φ−ψ‖τ,t∈R+, |
where ‖φ−ψ‖τ=supt∈[−τ,0]‖φ(t)−ψ(t)‖Hm×n. Then the solution x of system (1.1) is said to be globally exponentially stable.
Theorem 3.2. Assume that (H1)–(H3) hold, then system (1.1) has a unique Wp-almost periodic solution that is globallyexponentially stable.
Proof. Let x(t) be the Wp-almost periodic solution with the initial value φ(t) and y(t) be an arbitrary solution with the initial value ψ(t). Taking
zij(t)=xi(t)−yij(t),ϕij(t)=φij(t)−ψij(t),ij∈Λ, |
we have
˙zij(t)=−aij(t)zij(t)−∑Ckl∈Nr(i,j)Bklij(t)(fij(xkl(t))xij(t)−fij(ykl(t))yij(t)−∑Ckl∈Ns(i,j)Cklij(t)(gij(xkl(t−τkl(t)))xij(t)−gij(ykl(t−τkl(t)))yij(t)),ij∈Λ. | (3.28) |
For ij∈Λ, we define the following functions:
Πij(u)=am−u−(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)2CklijLgijαeuτMij+∑Ckl∈Ns(i,j)2CklijLgijα). |
From (H3), we get
Πij(0)=am−(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)4CklMijLgijα)>0,ij∈Λ. |
Since Πij(u) is continuous on [0,+∞) and Πij(u)→−∞ as u→∞, there exists ζij>0 such that Πij(ζij)=0 and Πij(u)>0, for u∈(0,ζij), ij∈Λ. Let ς=minij∈Λ{ζij}, then we have Πij(ς)≥0,ij∈Λ. Hence, we can choose a positive constant λ such that 0<λ<min{ς,am} and Πij(λ)>0. Thus, one has
1am−λ(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)2CklijLgijαeλτMij+∑Ckl∈Ns(i,j)2CklijLgijα)<1, | (3.29) |
where ij∈Λ. Take a constant M=maxi∈Λ{amˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)4CklMijLgijα}, then by (H3), we have M>1. Thus,
1M−1am−λ(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)2CklijLgijαeλτMij+∑Ckl∈Ns(i,j)2CklijLgijα)<0,ij∈Λ. | (3.30) |
From (3.28), we have
˙zij(t)+aRij(t)zij(t)=−ˇaij(t)zij(t)−∑Ckl∈Nr(i,j)Bklij(t)(fij(xkl(t))xij(t)−fij(ykl(t))yij(t))−∑Ckl∈Ns(i,j)Cklij(t)(gij(xkl(t−τkl(t)))xij(t)−gij(ykl(t−τkl(t)))yij(t)),ij∈Λ. | (3.31) |
Multiplying both sides of (3.31) by e∫t0aRij(v)dv and integrating over [0,t], we have
zij(t)=ϕij(0)e−∫t0aRij(v)dv+∫t0e−∫tsaRij(v)dv[−ˇaij(t)zij(s)−∑Ckl∈Nr(i,j)Bklij(t)(fij(xkl(t))xij(t)−fij(ykl(t))yij(t))−∑Ckl∈Ns(i,j)Cklij(t)(gij(xkl(t−τkl(t)))xij(t)−gij(ykl(t−τkl(t)))yij(t))]ds,ij∈Λ. |
Hence, for any ϵ>0, it is easy to see that
‖z(t)‖Hm×n<(‖ϕ‖τ+ϵ)e−λt<M(‖ϕ‖τ+ϵ)e−λt,∀t∈(−τ,0]. |
We claim that
‖z(t)‖Hm×n<M(‖ϕ‖τ+ϵ)e−λt,∀t∈[0,+∞). | (3.32) |
Otherwise, there exists t∗>0 such that
‖z(t∗)‖Hm×n=M(‖ϕ‖τ+ϵ)e−λt∗ | (3.33) |
and
‖z(t)‖Hm×n<M(‖ϕ‖τ+ϵ)e−λt,t<t∗. | (3.34) |
From (3.29), (3.30) and (3.34), we get
‖zij(t∗)‖H≤‖ϕij(0)‖He−t∗am+∫t∗0e−t∗am[ˇaMij‖zij(s)‖H+∑Ckl∈Nr(i,j)Bklij(t)×(‖fij(xkl(t))(xij(t)−yij(t))‖H+‖(fij(xkl(t))−fij(ykl(t)))yij(t)‖H)+∑Ckl∈Ns(i,j)Cklij(t)(‖gij(xkl(t−τkl(t)))(xij(t)−yij(t))‖H+‖(gij(xkl(t−τkl(t)))−gij(ykl(t−τkl(t))))yij(t)‖H)]ds≤(‖ϕ‖τ+ϵ)e−amt∗+M(‖ϕ‖τ+ϵ)∫t∗0e−(t∗−s)am(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)2CklijLgijαeλτMij+∑Ckl∈Ns(i,j)2CklijLgijα)e−λsds≤(‖ϕ‖τ+ϵ)e−λt∗e(λ−am)t∗+M(‖ϕ‖τ+ϵ)e−λt∗(1−e(λ−am)t∗)am−λ(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)2CklijLgijαeλτMij+∑Ckl∈Ns(i,j)2CklijLgijα)=M(‖ϕ‖τ+ϵ)e−λt∗[e(λ−am)t∗M+1−e(λ−am)t∗am−λ(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)2CklijLgijαeλτMij+∑Ckl∈Ns(i,j)2CklijLgijα)]=M(‖ϕ‖τ+ϵ)e−λt∗[(1M−1am−λ(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)2CklijLgijαeλτMij+∑Ckl∈Ns(i,j)2CklijLgijα)e(λ−am)t∗+1am−λ(ˇaMij+∑Ckl∈Nr(i,j)4BklMijLfijα+∑Ckl∈Ns(i,j)2CklijLgijαeλτMij+∑Ckl∈Ns(i,j)2CklijLgijα)]<M(‖ϕ‖τ+ϵ)e−λt∗,ij∈Λ, |
which contradicts the Eq (3.33). Hence, (3.32) holds. Letting ϵ→0+, from (3.32), we have
‖z(t)‖Hm×n≤M‖ϕ‖τe−λt,∀t>0. |
Therefore, the Wp-almost periodic solution of system (1.1) is globally exponentially stable. This completes the proof.
Example 4.1. In system (1.1), let i,j=1,2, r=s=1 and take
xij(t)=xRij(t)+ixIij(t)+jxJij(t)+kxKij(t)∈H, |
a11(t)=32sin2(√3t)+i√25cos(2t)−j25sin(√5t)+k1100cos2(3t), |
a21(t)=39|cos(√2t)|−i210sin(2t)+j310sin2(√7t)+k√3100cos(t), |
a12(t)=33cos2(t)−i√36sin(5t)−j14cos3(√5t)+k16cos2(3t) |
a22(t)=37sin4(√3t)+i14cos(7t)−j√2916|sin(√11t)|+k18cos2(5t), |
τ11(t)=√232sin8(√2t),τ12(t)=116cos2(√3t),τ21(t)=141sin2(√25t),τ22(t)=129sin4(√37t), |
B11(t)=13|cos(√2t)|,B12(t)=16sin(4t)+12B21(t)=34sin(2t)+1,B22(t)=14sin2(√5t), |
C11(t)=cos(√3t)+3,C12(t)=|sin(√5t)|,C21(t)=2sin2(√7t),C22(t)=14sin(2t)+74, |
f11(x)=f12(x)=15sin(14xR+√36xK)−i15|3√24xI+√35xJ|+k14sin(15xR), |
f21(x)=f22(x)=14|√23xI+√37xK|−i√28sin(15xR+√25xK)+k√23sin(15xIj), |
g11(x)=g12(x)=120sin(9√22xR)−i√340sin(75xK)+j√220sin(13xJ+34xI), |
g21(x)=g22(x)=√225|13xJ+23xI|−i√340sin(75xK+xR)+j121sin(4√2xR), |
I11(t)=I21(t)=√2sint+i43e−|t|+k√23cos(12t), |
I12(t)=I22(t)=i√23sin√2t−j15cos2t+ke−|t|. |
By computing, we obtain
Lf11=Lf12=310,Lf21=Lf22=16,Lg11=Lg12=920,Lg21=Lg22=821, |
am=32,ˇaM11=12,ˇaM12=25,ˇaM21=512,ˇaM22=716,τ=116,τ′=12, |
∑Ckl∈N1(1,1)BklM11=∑Ckl∈N1(1,2)BklM12=∑Ckl∈N1(2,1)BklM21=∑Ckl∈N1(2,2)BklM22=4, |
∑Ckl∈N1(1,1)CklM11=∑Ckl∈N1(1,2)CklM12=∑Ckl∈N1(2,1)CklM21=∑Ckl∈N1(2,2)CklM22=8. |
Choose a constant α=116≥‖ϕ0‖∞, for i,j=1,2, we obtain that
κ=maxij∈Λ{2am[ˇaMij+∑Ckl∈Nr(i,j)2BklMijLfijα+∑Ckl∈Ns(i,j)2CklMijLgijα]}=0.06875<1. |
For i,j=1,2, take p=3, it is easy to obtain that
maxij∈Λ{96(23am)3[23(ˇaMij)3+2(∑Ckl∈Nr(i,j)2BklMijLfijα)3+(1+2e34amτ1−τ′)(∑Ckl∈Ns(i,j)2CklMijLgijα)3]}=0.0016<1. |
Take p=2, we have
maxij∈Λ{24(2am)2[23(ˇaMij)2+2(∑Ckl∈Nr(i,j)2BklMijLfijα)2+(1+2e12amτ1−τ′)(∑Ckl∈Ns(i,j)2CklMijLgijα)2]}=0.2832<1. |
Thus, conditions (H1)–(H4) of Theorems 3.1 and Theorems 3.2 are satisfied. Hence, system (1.1) has a unique Wp-almost periodic solution that is globally exponentially stable (see Figures 1, 2).
Remark 4.1. No known results are available to give the results of Example 4.1.
In this paper, the existence and global exponential stability of Weyl almost periodic solutions for a class of quaternion-valued neural networks with time-varying delays are established. Even when the system we consider is a real-valued system, our results are brand-new. In addition, the method in this paper can be used to study the existence of Weyl almost periodic solutions for other types of neural networks.
This work is supported by the National Natural Science Foundation of China under Grant 11861072.
The authors declare no conflict of interest in this paper.
[1] |
A. Bouzerdoum, R. B. Pinter, Shunting inhibitory cellular neural networks: derivation and stability analysis, IEEE Trans. Circuits Syst., 40 (1993), 215–221. https://doi.org/10.1109/81.222804 doi: 10.1109/81.222804
![]() |
[2] |
X. Huang, J. Cao, Almost periodic solution of shunting inhibitory cellular neural networks with time-varying delay, Phys. Lett. A, 314 (2003), 222–231. https://doi.org/10.1016/S0375-9601(03)00918-6 doi: 10.1016/S0375-9601(03)00918-6
![]() |
[3] | B. Liu, Stability of shunting inhibitory cellular neural networks with unbounded time-varying delays, Appl. Math. Lett., 22 (2009), 1–5. |
[4] |
C. Huang, S. Wen, L. Huang, Dynamics of anti-periodic solutions on shunting inhibitory cellular neural networks with multi-proportional delays, Neurocomputing, 357 (2019), 47–52. https://doi.org/10.1016/j.neucom.2019.05.022 doi: 10.1016/j.neucom.2019.05.022
![]() |
[5] |
Y. Li, X. Meng, Almost periodic solutions for quaternion-valued shunting inhibitory cellular neural networks of neutral type with time delays in the leakage term, Int. J. Syst. Sci., 49 (2018), 2490–2505. https://doi.org/10.1080/00207721.2018.1505006 doi: 10.1080/00207721.2018.1505006
![]() |
[6] | N. Huo, Y. Li, Antiperiodic solutions for quaternion-valued shunting inhibitory cellular neural networks with distributed delays and impulses, Complexity, 2018 (2018), 6420256. |
[7] |
A. Sudbery, Quaternionic analysis, Math. Proc. Cambridge, 85 (1979), 199–225. https://doi.org/10.1017/S0305004100055638 doi: 10.1017/S0305004100055638
![]() |
[8] |
M. Kobayashi, Quaternionic Hopfield neural networks with twin-multistate activation function, Neurocomputing, 267 (2017), 304–310. https://doi.org/10.1016/j.neucom.2017.06.013 doi: 10.1016/j.neucom.2017.06.013
![]() |
[9] |
Y. Liu, D. Zhang, J. Lou, J. Lu, J. Cao, Stability analysis of quaternion-valued neural networks: decomposition and direct approaches, IEEE Trans. Neural Netw. Learn. Syst., 29 (2017), 4201–4211. https://doi.org/10.1109/TNNLS.2017.2755697 doi: 10.1109/TNNLS.2017.2755697
![]() |
[10] | C. A. Popa, E. Kaslik, Multistability and multiperiodicity in impulsive hybrid quaternion-valued neural networks with mixed delays, Neural Netw., 99 (2018), 1–18. |
[11] | Y. Li, J. Qin, B. Li, Existence and global exponential stability of anti-periodic solutions for delayed quaternion-valued cellular neural networks with impulsive effects, Math. Meth. Appl. Sci., 42 (2019), 5–23. |
[12] |
Z. Tu, Y. Zhao, N. Ding, Y. Feng, W. Zhang, Stability analysis of quaternion-valued neural networks with both discrete and distributed delays, Math. Meth. Appl. Sci., 343 (2019), 342–353. https://doi.org/10.1016/j.amc.2018.09.049 doi: 10.1016/j.amc.2018.09.049
![]() |
[13] |
X. Qi, H. Bao, J. Cao, Exponential input-to-state stability of quaternion-valued neural networks with time delay, Appl. Math. Comput., 358 (2019), 382–393. https://doi.org/10.1016/j.amc.2019.04.045 doi: 10.1016/j.amc.2019.04.045
![]() |
[14] |
J. Xiang, Y. Li, Pseudo almost automorphic solutions of quaternion-valued neural networks with infinitely distributed delays via a non-decomposing method, Adv. Differ. Equ., 2019 (2019), 356. https://doi.org/10.1186/s13662-019-2295-x doi: 10.1186/s13662-019-2295-x
![]() |
[15] |
Y. Li, J. Xiang, Existence and global exponential stability of anti-periodic solutions for quaternion-valued cellular neural networks with time-varying delays, Adv. Differ. Equ., 2020 (2020), 47. https://doi.org/10.1186/s13662-020-2523-4 doi: 10.1186/s13662-020-2523-4
![]() |
[16] |
H. Wang, G. Wei, S. Wen, T. Huang, Impulsive disturbance on stability analysis of delayed quaternion-valued neural networks, Appl. Math. Comput., 390 (2021), 125680. https://doi.org/10.1016/j.amc.2020.125680 doi: 10.1016/j.amc.2020.125680
![]() |
[17] |
Y. Li, H. Wang, X. Meng, Almost automorphic synchronization of quaternion-valued high-order Hopfield neural networks with time-varying and distributed delays, IMA J. Math. Control Inform., 36 (2019), 983–1013. https://doi.org/10.1093/imamci/dny015 doi: 10.1093/imamci/dny015
![]() |
[18] |
Y. Li, X. Meng, Almost automorphic solutions for quaternion-valued Hopfield neural networks with mixed time-varying delays and leakage delays, J. Syst. Sci. Complexity, 33 (2020), 100–121. https://doi.org/10.1007/s11424-019-8051-1 doi: 10.1007/s11424-019-8051-1
![]() |
[19] |
A. Pratap, R. Raja, J. Alzabut, J. Cao, G. Rajchakit, C. Huang, et al. {Mittag-Leffler stability and adaptive impulsive synchronization of fractional order neural networks in quaternion field}, Math. Methods Appl. Sci., 43 (2020), 6223–6253. https://doi.org/10.1002/mma.6367 doi: 10.1002/mma.6367
![]() |
[20] | R. Sriraman, G. Rajchakit, C. P. Lim, P. Chanthorn, R. Samidurai, Discrete-time stochastic quaternion-valued neural networks with time delays: an asymptotic stability analysis, Symmetry, 12 (2020), 936. |
[21] | U. Humphries, G. Rajchakit, P. Kaewmesri, P. Chanthorn, R. Sriraman, R. Samidurai, et al. {Stochastic memristive quaternion-valued neural networks with time delays: an analysis on mean square exponential input-to-state stability}, Mathematics, 8 (2020), 815. |
[22] | U. Humphries, G. Rajchakit, P. Kaewmesri, P. Chanthorn, R. Sriraman, Global stability analysis of fractional-order quaternion-valued bidirectional associative memory neural networks, Mathematics, 8 (2020), 801. |
[23] |
A. Pratap, R. Raja, J. Alzabut, J. Dianavinnarasi, G. Rajchakit, Finite-time Mittag-Leffler stability of fractional-order quaternion-valued memristive neural networks with impulses, Neural Process. Lett., 51 (2020), 1485–1526. https://doi.org/10.1007/s11063-019-10154-1 doi: 10.1007/s11063-019-10154-1
![]() |
[24] | G. Rajchakit, P. Chanthorn, P. Kaewmesri, R. Sriraman, C. P. Lim, Global Mittag-Leffler stability and stabilization analysis of fractional-order quaternion-valued memristive neural networks, Mathematics, 8 (2020), 422. |
[25] |
G. T. Stamov, I. M. Stamova, Almost periodic solutions for impulsive neural networks with delay, Appl. Math. Modell., 31 (2007), 1263–1270. https://doi.org/10.1016/j.apm.2006.04.008 doi: 10.1016/j.apm.2006.04.008
![]() |
[26] |
A. Arbi, Dynamics of BAM neural networks with mixed delays and leakage time-varying delays in the weighted pseudo-almost periodic on time-space scales, Math. Meth. Appl. Sci., 41 (2018), 1230–1255. https://doi.org/10.1002/mma.4661 doi: 10.1002/mma.4661
![]() |
[27] |
D. Li, Z. Zhang, X. Zhang, Periodic solutions of discrete-time quaternion-valued BAM neural networks, Chaos Solitons Fractals, 138 (2020), 110144. https://doi.org/10.1016/j.chaos.2020.110144 doi: 10.1016/j.chaos.2020.110144
![]() |
[28] |
S. Shen, Y. Li, Sp-almost periodic solutions of Clifford-valued fuzzy cellular neural networks with time-varying delays, Neural Process. Lett., 51 (2020), 1749–1769. https://doi.org/10.1007/s11063-019-10176-9 doi: 10.1007/s11063-019-10176-9
![]() |
[29] | Y. Li, N. Huo, B. Li, On μ-pseudo almost periodic solutions for Clifford-valued neutral type neural networks with delays in the leakage term, IEEE Trans. Neural Netw. Learn. Syst., 32 (2021), 1365–1374. |
[30] |
G. Stamov, I. Stamova, A. Martynyuk, T. Stamov, Almost periodic dynamics in a new class of impulsive reaction-diffusion neural networks with fractional-like derivatives, Chaos Solitons Fractals, 143 (2021), 110647. https://doi.org/10.1016/j.chaos.2020.110647 doi: 10.1016/j.chaos.2020.110647
![]() |
[31] |
S. Chen, K. Wang, J. Liu, X. Lin, Periodic solutions of Cohen-Grossberg-type Bidirectional associative memory neural networks with neutral delays and impulses, AIMS Math., 6 (2021), 2539–2558. https://doi.org/10.3934/math.2021154 doi: 10.3934/math.2021154
![]() |
[32] |
Y. Li, J. Qin, Existence and global exponential stability of periodic solutions for quaternion-valued cellular neural networks with time-varying delays, Neurocomputing, 292 (2018), 91–103. https://doi.org/10.1016/j.neucom.2018.02.077 doi: 10.1016/j.neucom.2018.02.077
![]() |
[33] |
Y. Li, J. Xiang, B. Li, Almost periodic solutions of quaternion-valued neutral type high-order Hopfield neural networks with state-dependent delays and leakage delays, Appl. Intell., 50 (2020), 2067–2078. https://doi.org/10.1007/s10489-020-01634-2 doi: 10.1007/s10489-020-01634-2
![]() |
[34] |
H. Weyl, Integralgleichungen und fastperiodische Funktionen, Math Ann., 97 (1927), 338–356. https://doi.org/10.1007/BF01447871 doi: 10.1007/BF01447871
![]() |
[35] |
F. Bedouhene, Y. Ibaouene, O. Mellah, P. R. de Fitte, Weyl almost periodic solutions to abstract linear and semilinear equations with Weyl almost periodic coefficients, Math. Meth. Appl. Sci., 41 (2018), 9546–9566. https://doi.org/10.1002/mma.5312 doi: 10.1002/mma.5312
![]() |
[36] | M. Kostić, Almost Periodic and Almost Automorphic Solutions to Integro-Differential Equations, W. de Gruyter, Berlin, 2019. |
[37] | M. Kostić, Selected Topics in Almost Periodicity, W. de Gruyter, Berlin, 2022. |
[38] | C. Corduneanu, Almost periodic oscillations and waves, Springer, Berlin, 2009. |
[39] | A. S. Besicovitch, Almost periodic function, Dover Publications, New York, 1954. |
[40] |
M. Kamenskii, O. Mellah, P. R. de Fitte, Weak averaging of semilinear stochastic differential equations with almost periodic coefficients, J. Math. Anal. Appl., 427 (2015), 336–364. https://doi.org/10.1016/j.jmaa.2015.02.036 doi: 10.1016/j.jmaa.2015.02.036
![]() |
1. | Adnène Arbi, Najeh Tahri, New results on time scales of pseudo Weyl almost periodic solution of delayed QVSICNNs, 2022, 41, 2238-3603, 10.1007/s40314-022-02003-0 | |
2. | Marko Kostić, Weyl ρ-Almost Periodic Functions in General Metric, 2023, 73, 1337-2211, 465, 10.1515/ms-2023-0035 | |
3. | Qi Shao, Yongkun Li, Almost periodic solutions for Clifford-valued stochastic shunting inhibitory cellular neural networks with mixed delays, 2024, 9, 2473-6988, 13439, 10.3934/math.2024655 | |
4. | Adnène Arbi, Najeh Tahri, Synchronization analysis of novel delayed dynamical Clifford-valued neural networks on timescales, 2024, 18, 1748-3018, 10.1177/17483026241241492 |