
In this paper, we investigate a predator-prey system with fractional type cross-diffusion incorporating the Beddington-DeAngelis functional response subjected to the homogeneous Neumann boundary condition. First, by using the maximum principle and the Harnack inequality, we establish a priori estimate for the positive stationary solution. Second, we study the non-existence of non-constant positive solutions mainly by employing the energy integral method and the Poincaré inequality. Finally, we discuss the existence of non-constant positive steady states for suitable large self-diffusion d2 or cross-diffusion d4 by using the Leray-Schauder degree theory, and the results reveal that the diffusion d2 and the fractional type cross-diffusion d4 can create spatial patterns.
Citation: Pan Xue, Cuiping Ren. Spatial patterns for a predator-prey system with Beddington-DeAngelis functional response and fractional cross-diffusion[J]. AIMS Mathematics, 2023, 8(8): 19413-19426. doi: 10.3934/math.2023990
[1] | Fahad Alsharari, Ahmed O. M. Abubaker, Islam M. Taha . On r-fuzzy soft γ-open sets and fuzzy soft γ-continuous functions with some applications. AIMS Mathematics, 2025, 10(3): 5285-5306. doi: 10.3934/math.2025244 |
[2] | Jinguo Zhang, Dengyun Yang, Yadong Wu . Existence results for a Kirchhoff-type equation involving fractional p(x)-Laplacian. AIMS Mathematics, 2021, 6(8): 8390-8403. doi: 10.3934/math.2021486 |
[3] | Cenap Ozel, M. A. Al Shumrani, Aynur Keskin Kaymakci, Choonkil Park, Dong Yun Shin . On δb-open continuous functions. AIMS Mathematics, 2021, 6(3): 2947-2955. doi: 10.3934/math.2021178 |
[4] | Ibtesam Alshammari, Islam M. Taha . On fuzzy soft β-continuity and β-irresoluteness: some new results. AIMS Mathematics, 2024, 9(5): 11304-11319. doi: 10.3934/math.2024554 |
[5] | Sang-Eon Han . Semi-topological properties of the Marcus-Wyse topological spaces. AIMS Mathematics, 2022, 7(7): 12742-12759. doi: 10.3934/math.2022705 |
[6] | Samirah Alzahrani, A. A. Nasef, N. Youns, A. I. EL-Maghrabi, M. S. Badr . Soft topological approaches via soft γ-open sets. AIMS Mathematics, 2022, 7(7): 12144-12153. doi: 10.3934/math.2022675 |
[7] | Nicky K. Tumalun, Philotheus E. A. Tuerah, Marvel G. Maukar, Anetha L. F. Tilaar, Patricia V. J. Runtu . An application of generalized Morrey spaces to unique continuation property of the quasilinear elliptic equations. AIMS Mathematics, 2023, 8(11): 26007-26020. doi: 10.3934/math.20231325 |
[8] | Dina Abuzaid, Samer Al-Ghour . Supra soft Omega-open sets and supra soft Omega-regularity. AIMS Mathematics, 2025, 10(3): 6636-6651. doi: 10.3934/math.2025303 |
[9] | Syafrizal Sy, Rinovia Simanjuntak, Tamaro Nadeak, Kiki Ariyanti Sugeng, Tulus Tulus . Distance antimagic labeling of circulant graphs. AIMS Mathematics, 2024, 9(8): 21177-21188. doi: 10.3934/math.20241028 |
[10] | Mesfer H. Alqahtani, Alaa M. Abd El-latif . Separation axioms via novel operators in the frame of topological spaces and applications. AIMS Mathematics, 2024, 9(6): 14213-14227. doi: 10.3934/math.2024690 |
In this paper, we investigate a predator-prey system with fractional type cross-diffusion incorporating the Beddington-DeAngelis functional response subjected to the homogeneous Neumann boundary condition. First, by using the maximum principle and the Harnack inequality, we establish a priori estimate for the positive stationary solution. Second, we study the non-existence of non-constant positive solutions mainly by employing the energy integral method and the Poincaré inequality. Finally, we discuss the existence of non-constant positive steady states for suitable large self-diffusion d2 or cross-diffusion d4 by using the Leray-Schauder degree theory, and the results reveal that the diffusion d2 and the fractional type cross-diffusion d4 can create spatial patterns.
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] |
W. Chen, M. Wang, Qualitative analysis of predator-prey models with Beddington-DeAngelis functional response and diffusion, Math. Comput. Model., 42 (2005), 31–44. http://dx.doi.org/10.1016/j.na.2006.08.022 doi: 10.1016/j.na.2006.08.022
![]() |
[2] |
R. S. Cantrell, C. Cosner, On the dynamics of predator-prey models with the Beddington-DeAngelis functional response, J. Math. Anal. Appl., 257 (2001), 206–222. https://doi.org/10.1006/jmaa.2000.7343 doi: 10.1006/jmaa.2000.7343
![]() |
[3] |
D. T. Dimitrov, H. V. Kojouharov, Complete mathematical analysis of predator-prey models with linear prey growth and Beddington-DeAngelis, Appl. Math. Comput., 162 (2005), 523–538. https://doi.org/10.1016/j.amc.2003.12.106 doi: 10.1016/j.amc.2003.12.106
![]() |
[4] |
D. Luo, Q. Wang, Global bifurcation and pattern formation for a reaction-diffusion predator-prey model with prey-taxis and double Beddington-DeAngelis functional responses, Nonlinear Anal.-Real, 67 (2022), 103638. https://doi.org/10.1016/j.nonrwa.2022.103638 doi: 10.1016/j.nonrwa.2022.103638
![]() |
[5] |
Y. Lou, W. M. Ni, Diffusion, self-diffusion and cross-diffusion, J. Differ. Equations, 131 (1996), 79–131. https://doi.org/10.1006/jdeq.1996.0157 doi: 10.1006/jdeq.1996.0157
![]() |
[6] |
Y. Lou, W. M. Ni, Diffusion vs cross-diffusion: An elliptic approach, J. Differ. Equations, 154 (1999), 157–190. https://doi.org/10.1006/jdeq.1998.3559 doi: 10.1006/jdeq.1998.3559
![]() |
[7] | F. Yi, Turing instability of the periodic solutions for reaction-diffusion systems with cross-diffusion and the patch model with cross-diffusion-like coupling, 281 (2021), 379–410. https://doi.org/10.1016/j.jde.2021.02.006 |
[8] |
Q. Meng, L. Yang, Steady state in a cross-diffusion predator-prey model with the Beddington-DeAngelis functional response, Nonlinear Anal.-Real, 45 (2019), 401–413. https://doi.org/10.1016/j.nonrwa.2018.07.012 doi: 10.1016/j.nonrwa.2018.07.012
![]() |
[9] |
J. F. Zhang, Spatial patterns of a fractional type cross-diffusion Holling-Tanner model, Comput. Math. Appl., 76 (2018), 957–965. https://doi.org/10.1016/j.camwa.2018.05.033 doi: 10.1016/j.camwa.2018.05.033
![]() |
[10] |
P. Xue, Y. Jia, C. Ren, X. Li, Non-constant positive solutions of a general Gause-type predator-prey system with self- and cross- diffusions, Math. Model. Nat. Pheno., 16 (2021), 1–25. https://doi.org/10.1051/mmnp/2021017 doi: 10.1051/mmnp/2021017
![]() |
[11] |
Y. Jia, P. Xue, Effects of the self- and cross-diffusion on positive steady states for a generalized predator-prey system, Nonlinear Anal.-Real, 32 (2016), 229–241. https://doi.org/10.1016/j.nonrwa.2016.04.012 doi: 10.1016/j.nonrwa.2016.04.012
![]() |
[12] |
C. Li, Existence of positive solution for a cross-diffusion predator-prey system with Holling type-II functional response, Chaos Soliton. Fract., 99 (2017), 226–232. https://doi.org/10.1016/j.chaos.2017.04.001 doi: 10.1016/j.chaos.2017.04.001
![]() |
[13] |
S. Li, Y. Yamada, Effect of cross-diffusion prey-preydator model with a protection zone II, J. Math. Anal. Appl., 461 (2018), 971–992. https://doi.org/10.1016/j.jmaa.2017.12.029 doi: 10.1016/j.jmaa.2017.12.029
![]() |
[14] | A. Okubo, Diffusion and ecological problems: Mathematical models, Springer-Verlag, Berlin, 1980. |
[15] |
J. Cao, H. Sun, P. Hao, P. Wang, Bifurcation and turing instability for a predator-prey model with nonlinear reaction cross-diffusion, Appl. Math. Model., 89 (2021), 1663–1677. https://doi.org/10.1016/j.apm.2020.08.030 doi: 10.1016/j.apm.2020.08.030
![]() |
[16] |
X. Chen, Y. Qi, M. Wang, A strongly coupled predator-prey system with non-monotonic functional response, Nonlinear Anal., 67 (2007), 1966–1979. https://doi.org/10.1016/j.na.2006.08.022 doi: 10.1016/j.na.2006.08.022
![]() |
[17] | Y. Lou, W. M. Ni, S. Yotsutani, On a limiting system in the Lotka-Volterra competition with cross-diffusion, Discrete Cont. Dyn. Syst., 10 (2004), 435–458. |
[18] |
J. H. P. Dawes, M. O. Souza, A derivation of Holling's type I, II and III functional responses in predator-prey systems, J. Theor. Biol., 327 (2013), 11–22. https://doi.org/10.1016/j.jtbi.2013.02.017 doi: 10.1016/j.jtbi.2013.02.017
![]() |
[19] | D. Gilbarg, N. S. Trudinger, Elliptic partial differential equation of second order, Springer-Verlag, Berlin, 2001. |
[20] | X. Zeng, Non-constant positive steady states of a prey-predator system with cross-diffusions, J. Math. Anal. Appl., 332 (2007), 989–1009. https://doi.org/10.1006/jmaa.2000.7343 |
[21] |
R. K. Ghaziani, W. Govaerts, C. Sonck, Resonance and bifurcation in a discrete-time predator-prey system with Holling functional response, Nonlinear Anal.-Real, 13 (2012), 1451–1465. https://doi.org/10.1016/j.nonrwa.2011.11.009 doi: 10.1016/j.nonrwa.2011.11.009
![]() |
[22] |
R. Peng, J. P. Shi, Non-existence of non-constant positive steady states of two Holling-type II predator-prey systems: Strong interaction case, J. Differ. Equations, 247 (2009), 866–886. https://doi.org/10.1016/j.jde.2009.03.008 doi: 10.1016/j.jde.2009.03.008
![]() |
[23] |
Y. Jia, A sufficient and necessary condition for the existence of positive solutions for a prey-predator system with Ivlev-type functional response, Appl. Math. Lett., 24 (2011), 1084–1088. https://doi.org/10.1016/j.aml.2011.01.027 doi: 10.1016/j.aml.2011.01.027
![]() |
[24] |
J. Sugie, Two-parameter bifurcation in a predator-prey system of Ivlev type, J. Math. Anal. Appl., 217 (1998), 349–371. https://doi.org/10.1006/jmaa.1997.5700 doi: 10.1006/jmaa.1997.5700
![]() |
[25] |
Y. Jia, J. Wu, H. K. Xu, Positive solutions of a Lotka-Volterra competition model with cross-diffusion, Comput. Math. Appl., 68 (2014), 1220–1228. https://doi.org/10.1016/j.camwa.2014.08.016 doi: 10.1016/j.camwa.2014.08.016
![]() |
[26] |
W. Ko, K. Ryu, A qualitative study on general Gause-type predator-prey models with constant diffusion rates, J. Math. Anal. Appl., 344 (2008), 217–230. https://doi.org/10.1016/j.jmaa.2008.03.006 doi: 10.1016/j.jmaa.2008.03.006
![]() |
[27] |
K. Oeda, Effect of cross-diffusion on the stationary problem of a prey-predator model with a protection zone, J. Differ. Equations, 250 (2011), 3988–4009. https://doi.org/10.1016/j.jde.2011.01.026 doi: 10.1016/j.jde.2011.01.026
![]() |
[28] |
Y. Peng, T. Zhang, Turing instability and pattern induced by cross-diffusion in a predator-prey system with Allee effect, Appl. Math. Comput., 275 (2016), 1–12. https://doi.org/10.1016/j.amc.2015.11.067 doi: 10.1016/j.amc.2015.11.067
![]() |
[29] | L. Nirenberg, Topics in nonlinear functional analysis, American Mathematical Socity, Providence, Rhode Island, 2001. |
[30] |
H. Wang, P. Liu, Pattern dynamics of a predator-prey system with cross-diffusion, Allee effect and generalized Holling IV functional response, Chaos Soliton. Fract., 171 (2023), 113456. https://doi.org/10.1016/j.chaos.2023.113456 doi: 10.1016/j.chaos.2023.113456
![]() |
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 |