We consider a cooperating two-species Lotka-Volterra model of degenerate
parabolic equations. We are interested in the coexistence of the species in a bounded domain. We establish the existence of global generalized solutions of the initial boundary value problem by means of parabolic regularization and also consider the existence of the nontrivial time-periodic solution for this system.
1. Introduction
In this paper, we consider the following two-species cooperative system:ut=Δum1+uα(a-bu+cv),(x,t)∈Ω×R+,vt=Δvm2+vβ(d+eu-fv),(x,t)∈Ω×R+,u(x,t)=0,v(x,t)=0,(x,t)∈∂Ω×R+,u(x,0)=u0(x),v(x,0)=v0(x),x∈Ω,
where m1,m2>1, 0<α<m1, 0<β<m2, 1≤(m1-α)(m2-β), a=a(x,t), b=b(x,t), c=c(x,t), d=d(x,t), e=e(x,t), f=f(x,t) are strictly positive smooth functions and periodic in time with period T>0 and u0(x) and v0(x) are nonnegative functions and satisfy u0m1,v0m2∈W01,2(Ω).
In dynamics of biological groups, the system (1.1)-(1.2) can be used to describe the interaction of two biological groups. The diffusion terms Δum1 and Δvm2 represent the effect of dispersion in the habitat, which models a tendency to avoid crowding and the speed of the diffusion is rather slow. The boundary conditions (1.3) indicate that the habitat is surrounded by a totally hostile environment. The functions u and v represent the spatial densities of the species at time t and a,d are their respective net birth rate. The functions b and f are intra-specific competitions, whereas c and e are those of interspecific competitions.
As famous models for dynamics of population, two-species cooperative systems like (1.1)-(1.2) have been studied extensively, and there have been many excellent results, for detail one can see [1–6] and references therein. As a special case, men studied the following two-species Lotka-Volterra cooperative system of ODEs:u′(t)=u(t)(a(t)-b(t)u(t)+c(t)v(t)),v′(t)=v(t)(d(t)+e(t)u(t)-f(t)v(t)).
For this system, Lu and Takeuchi [7] studied the stability of positive periodic solution and Cui [1] discussed the persistence and global stability of it.
When m1=m2=α=β=1, from (1.1)-(1.2) we get the following classical cooperative system:ut=Δu+u(a-bu+cv),vt=Δv+v(d+eu-fv).
For this system, Lin et al. [5] showed the existence and asymptotic behavior of T- periodic solutions when a,b,c,e,d,f are all smooth positive and periodic in time with period T>0. When a,b,c,e,d,f are all positive constants, Pao [6] proved that the Dirichlet boundary value problem of this system admits a unique solution which is uniformly bounded when ce<bf, while the blowup solutions are possible when the two species are strongly mutualistic (ce>bf). For the homogeneous Neumann boundary value problem of this system, Lou et al. [4] proved that the solution will blow up in finite time under a sufficient condition on the initial data. When c=e=0 and α=β=1, from (1.1) we get the single degenerate equation ut=Δum+u(a-bu).
For this equation, Sun et al. [8] established the existence of nontrivial nonnegative periodic solutions by monotonicity method and showed the attraction of nontrivial nonnegative periodic solutions.
In the recent years, much attention has been paid to the study of periodic boundary value problems for parabolic systems; for detail one can see [9–15] and the references therein. Furthermore, many researchers studied the periodic boundary value problem for degenerate parabolic systems, such as [16–19]. Taking into account the impact of periodic factors on the species dynamics, we are also interested in the existence of the nontrivial periodic solutions of the cooperative system (1.1)-(1.2). In this paper, we first show the existence of the global generalized solution of the initial boundary value problem (1.1)–(1.4). Then under the condition that blfl>cMeM,
where fM=sup{f(x,t)∣(x,t)∈Ω×ℝ},fl=inf{f(x,t)∣(x,t)∈Ω×ℝ}, we show that the generalized solution is uniformly bounded. At last, by the method of monotone iteration, we establish the existence of the nontrivial periodic solutions of the system (1.1)-(1.2), which follows from the existence of a pair of large periodic supersolution and small periodic subsolution. At last, we show the existence and the attractivity of the maximal periodic solution.
Our main efforts center on the discussion of generalized solutions, since the regularity follows from a quite standard approach. Hence we give the following definition of generalized solutions of the problem (1.1)–(1.4).
Definition 1.1.
A nonnegative and continuous vector-valued function (u,v) is said to be a generalized solution of the problem (1.1)–(1.4) if, for any 0≤τ<T and any functions φi∈C1(Q¯τ) with φi|∂Ω×[0,τ)=0(i=1,2),∇um1,∇vm2∈L2(Qτ), ∂um1/∂t,∂vm2/∂t∈L2(Qτ) and
∬Qτu∂φ1∂t-∇um1∇φ1+uα(a-bu+cv)φ1dxdt=∫Ωu(x,τ)φ1(x,τ)dx-∫Ωu0(x)φ1(x,0)dx,∬Qτv∂φ2∂t-∇vm2∇φ2+vβ(d+eu-fv)φ2dxdt=∫Ωv(x,τ)φ2(x,τ)dx-∫Ωv0(x)φ2(x,0)dx,
where Qτ=Ω×(0,τ).
Similarly, we can define a weak supersolution (u¯,v¯) (subsolution (u̲,v̲)) if they satisfy the inequalities obtained by replacing “=” with “≤” (“≥”) in (1.3), (1.4), and (1.9) and with an additional assumption φi≥0(i=1,2).
Definition 1.2.
A vector-valued function (u,v) is said to be a T-periodic solution of the problem (1.1)–(1.3) if it is a solution in [0,T] such that u(·,0)=u(·,T), v(·,0)=v(·,T) in Ω. A vector-valued function (u¯,v¯) is said to be a T-periodic supersolution of the problem (1.1)–(1.3) if it is a supersolution in [0,T] such that u¯(·,0)≥u¯(·,T),v¯(·,0)≥v¯(·,T) in Ω. A vector-valued function (u̲,v̲) is said to be a T-periodic subsolution of the problem (1.1)–(1.3), if it is a subsolution in [0,T] such that u̲(·,0)≤u̲(·,T),v̲(·,0)≤v̲(·,T) in Ω.
This paper is organized as follows. In Section 2, we show the existence of generalized solutions to the initial boundary value problem and also establish the comparison principle. Section 3 is devoted to the proof of the existence of the nonnegative nontrivial periodic solutions by using the monotone iteration technique.
2. The Initial Boundary Value Problem
To solve the problem (1.1)–(1.4), we consider the following regularized problem:∂uε∂t=div((muεm1-1+ε)∇uε)+uεα(a-buε+cvε),(x,t)∈QT,∂vε∂t=div((mvεm2-1+ε)∇vε)+vεβ(d+euε-fvε),(x,t)∈QT,uε(x,t)=0,vε(x,t)=0,(x,t)∈∂Ω×(0,T),uε(x,0)=u0ε(x),vε(x,0)=v0ε(x),x∈Ω,
where QT=Ω×(0,T), 0<ɛ<1,u0ɛ,v0ɛ∈C0∞(Ω) are nonnegative bounded smooth functions and satisfy 0≤u0ɛ≤‖u0‖L∞(Ω),0≤v0ɛ≤‖v0‖L∞(Ω),u0ɛm1⟶u0m1,v0ɛm2⟶v0m2,inW01,2(Ω)asɛ⟶0.
The standard parabolic theory (cf. [20, 21]) shows that (2.1)–(2.4) admits a nonnegative classical solution (uɛ,vɛ). So, the desired solution of the problem (1.1)–(1.4) will be obtained as a limit point of the solutions (uɛ,vɛ) of the problem (2.1)–(2.4). In the following, we show some important uniform estimates for (uɛ,vɛ).
Lemma 2.1.
Let (uɛ,vɛ) be a solution of the problem (2.1)–(2.4).
If 1<(m1-α)(m2-β), then there exist positive constants r and s large enough such that
1m2-β<m1+r-1m2+s-1<m1-α,‖uɛ‖Lr(QT)≤C,‖vɛ‖Ls(QT)≤C,
where C is a positive constant only depending on m1,m2,α,β,r,s, |Ω|, and T.
If 1=(m1-α)(m2-β), then (2.7) also holds when |Ω| is small enough.
Proof.
Multiplying (2.1) by uɛr-1(r>1) and integrating over Ω, we have that
∫Ω∂uɛr∂tdx=-4r(r-1)m1(m1+r-1)2∫Ω|∇uɛ(m1+r-1)/2|2dx+r∫Ωuɛα+r-1(a-buɛ+cvɛ)dx.
By Poincaré’s inequality, we have that
K∫Ωuɛm1+r-1dx≤∫Ω|∇uɛ(m1+r-1)/2|2dx,
where K is a constant depending only on |Ω| and N and becomes very large when the measure of the domain Ω becomes small. Since α<m1, Young's inequality shows that
auɛα+r-1≤Kr(r-1)m1(m1+r-1)2uɛm1+r-1+CK-(α+r-1)/(m1-α),cuɛα+r-1vɛ≤Kr(r-1)m1(m1+r-1)2uɛm1+r-1+CK-(α+r-1)/(m1-α)vɛ(m1+r-1)/(m1-α).
For convenience, here and below, C denotes a positive constant which is independent of ɛ and may take different values on different occasions. Complying (2.8) with (2.9) and (2.10), we obtain
∫Ω∂uɛr∂tdx≤-2Kr(r-1)m1(m1+r-1)2∫Ωuɛm1+r-1dx+CK-(α+r-1)/(m1-α)∫Ωvɛ(m1+r-1)/(m1-α)dx+CK-(α+r-1)/(m1-α).
As a similar argument as above, for vɛ and positive constant s>1, we have that
∫Ω∂vɛs∂tdx≤-2Ks(s-1)m2(m2+s-1)2∫Ωvɛm2+s-1dx+CK-(β+s-1)/(m2-β)∫Ωuɛ(m2+s-1)/(m2-β)dx+CK-(β+s-1)/(m2-β).
Thus we have that
∫Ω(∂uɛr∂t+∂vɛs∂t)dx≤-2Kr(r-1)m1(m1+r-1)2∫Ωuɛm1+r-1dx+CK-(β+s-1)/(m2-β)∫Ωuɛ(m2+s-1)/(m2-β)dx-2Ks(s-1)m2(m2+s-1)2∫Ωvɛm2+s-1dx+CK-(α+r-1)/(m1-α)∫Ωvɛ(m1+r-1)/(m1-α)dx+CK-(α+r-1)/(m1-α)+CK-(β+s-1)/(m2-β).
For the case of 1<(m1-α)(m2-β), there exist r,s large enough such that
1m1-α<m2+s-1m1+r-1<m2-β.
By Young's inequality, we have that
∫Ωuɛ(m2+s-1)/(m2-β)dx≤r(r-1)m1K(m2+s-1)/(m2-β)C(m1+r-1)2∫Ωuɛm1+r-1dx+CK-γ1,∫Ωvɛ(m1+r-1)/(m1-α)dx≤s(s-1)m2K(m1+r-1)/(m1-α)C(m2+s-1)p2∫Ωvɛm2+s-1dx+CK-γ2,
where
γ1=(m2+s-1)2[m2-β][(m2-β)(m1+r-1)-(m2+s-1)],γ2=(m1+r-1)2[m1-α][(m1-α)(m2+s-1)-(m1+r-1)].
Together with (2.13), we have that
∫Ω(∂uɛr∂t+∂vɛs∂t)dx≤-K∫Ω(uɛm1+r-1+vɛm2+s-1)dx+C(K-θ1+K-θ2)+CK-(α+r-1)/(m1-α)+CK-(β+s-1)/(m2-β),
where
θ1=(m2+s-1)+(m1+r-1)(β+s-1)(m2-β)(m1+r-1)-(m2+s-1),θ2=(m1+r-1)+(m2+s-1)(α+r-1)(m1-α)(m2+s-1)-(m1+r-1).
Furthermore, by Hölder's and Young's inequalities, from (2.17) we obtain
∫Ω(∂uɛr∂t+∂vɛs∂t)dx≤-K∫Ω(uɛr+vɛs)dx+C(K-θ1+K-θ2)+2K|Ω|+CK-(α+r-1)/(m1-α)+CK-(β+s-1)/(m2-β).
Then by Gronwall's inequality, we obtain
∫Ω(uɛr+vɛs)dx≤C.
Now we consider the case of 1=(m1-α)(m2-β). It is easy to see that there exist positive constants r,s large enough such that
1m1-α=m2+s-1m1+r-1=m2-β.
Due to the continuous dependence of K upon |Ω| in (2.9), from (2.13) we have that
∫Ω(∂uɛr∂t+∂vɛs∂t)dx≤-K∫Ω(uɛm1+r-1+vɛm2(p2-1)+s-1)dx+C
when |Ω| is small enough. Then by Young's and Gronwall's inequalities we can also obtain (2.20), and thus we complete the proof of this lemma.
Taking uɛm1, vɛm2 as the test functions, we can easily obtain the following lemma.
Lemma 2.2.
Let (uɛ,vɛ) be a solution of (2.1)–(2.4); then
∬QT|∇uɛm1|2dxdt≤C,∬QT|∇vɛm2|2dxdt≤C,
where C is a positive constant independent of ɛ.
Lemma 2.3.
Let (uɛ,vɛ) be a solution of (2.1)–(2.4), then
‖uɛ‖L∞(QT)≤C,‖vɛ‖L∞(QT)≤C,
where C is a positive constant independent of ɛ.
Proof.
For a positive constant k>∥u0ɛ∥L∞(Ω), multiplying (2.1) by (uɛ-k)+m1χ[t1,t2] and integrating the results over QT, we have that
1m1+1∬QT∂(uɛ-k)+m1+1χ[t1,t2]∂tdxdt+∬QT|∇(uɛ-k)+m1χ[t1,t2]|2dxdt≤∬QTauɛα+m1(a+cvɛ)dxdt,
where s+=max{0,s} and χ[t1,t2] is the characteristic function of [t1,t2](0≤t1<t2≤T). Let
Ik(t)=∫Ω(uɛ-k)+m1+1dx;
then Ik(t) is absolutely continuous on [0,T]. Denote by σ the point where Ik(t) takes its maximum. Assume that σ>0, for a sufficient small positive constant ϵ. Taking t1=σ-ϵ, t2=σ in (2.25), we obtain
1(m1+1)ϵ∫σ-ϵσ∫Ω∂(uɛ-k)+m1+1∂tdxdt+1ϵ∫σ-ϵσ∫Ω|∇(uɛ-k)+m1|2dxdt≤1ϵ∫σ-ϵσ∫Ωuɛα+m1(a+cvɛ)dxdt.
From
∫σ-ϵσ∫Ω∂(uɛ-k)+m1+1∂tdxdt=Ik(σ)-Ik(σ-ϵ)≥0,
we have that
1ϵ∫σ-ϵσ∫Ω|∇(uɛ-k)+m1|2dxdt≤1ϵ∫σ-ϵσ∫Ωuɛα+m1(a+cvɛ)dxdt.
Letting ϵ→0+, we have that
∫Ω|∇(uɛ(x,σ)-k)+m1|2dx≤∫Ωuɛα+m1(x,σ)(a+cvɛ(x,σ))dx.
Denote Ak(t)={x:uɛ(x,t)>k} and μk=supt∈(0,T)|Ak(t)|; then
∫Ak(σ)|∇(uɛ-k)+m1|2dx≤∫Ak(σ)uɛα+m1(a+cvɛ)dx.
By Sobolev's theorem,
(∫Ak(σ)((uɛ-k)+m1)pdx)1/p≤C(∫Ak(σ)|∇(uɛ-k)+m1|2dx)1/2,
with
2<p<{+∞,N≤2,2NN-2,N>2,
we obtain
(∫Ak(σ)((uɛ-k)+m1)pdx)2/p≤C∫Ak(σ)|∇(uɛ-k)+m1|2dx≤C∫Ak(σ)uɛα+m1(a+vɛ)dx≤C(∫Ak(σ)uɛrdx)(m1+α)/r(∫Ak(σ)(a+vɛ)r/(r-m1-α)dx)(r-m1-α)/r≤C(∫Ak(σ)(a+vɛ)r/(r-m1-α)dx)(r-m1-α)/r≤C(∫Ak(σ)(a+vɛ)sdx)1/s|Ak(σ)|(s(r-m1-α)-r)/sr≤Cμk(s(r-m1-α)-r)/sr,
where r>p(m1+α)/(p-2), s>pr/(p(r-m1-α)-2r) and C denotes various positive constants independent of ɛ. By Hölder’s inequality, it yields
Ik(σ)=∫Ω(uɛ-k)+m1+1dx=∫Ak(σ)(uɛ-k)+m1+1dx≤(∫Ak(σ)(uɛ-k)+m1pdx)(m1+1)/m1pμk1-(m1+1)/m1p≤Cμk1+[sp(r-m1-α)-pr-2sr](m1+1)/2psrm1.
Then
Ik(t)≤Ik(σ)≤Cμk1+[sp(r-m1-α)-pr-2sr](m1+1)/2psrm1,t∈[0,T].
On the other hand, for any h>k and t∈[0,T], we have that
Ik(t)≥∫Ak(t)(uɛ-k)+m1+1dx≥(h-k)m1+1|Ah(t)|.
Combined with (2.35), it yields
(h-k)m1+1μh≤Cμk1+[sp(r-m1-α)-pr-2sr](m1+1)/2psrm1,
that is,
μh≤C(h-k)m1+1μk1+[sp(r-m1-α)-pr-2sr](m1+1)/2psrm1.
It is easy to see that
γ=1+[sp(r-m1-α)-pr-2sr](m1+1)2psrm1>1.
Then by the De Giorgi iteration lemma [22], we have that
μl+d=sup|Al+d(t)|=0,
where d=C1/(m1+1)μl(γ-1)/(m1+1)2γ/(γ-1). That is,
uɛ≤l+da.e.inQT.
It is the same for the second inequality of (2.24). The proof is completed.
Lemma 2.4.
The solution (uɛ,vɛ) of (2.1)–(2.4) satisfies the following:
∬QT|∂uɛm1∂t|2dxdt≤C,∬QT|∂vɛm2∂t|2dxdt≤C,
where C is a positive constant independent of ɛ.
Proof.
Multiplying (2.1) by (∂/∂t)uɛm1 and integrating over Ω, by (2.3), (2.4) and Young's inequality we have that
4m1(m1+1)2∬QT|∂∂tuɛ(m1+1)/2|2dxdt=∬QT∂uɛ∂t∂uɛm1∂tdxdt=12∫Ω|∇uɛm1(x,0)|2dx-12∫Ω|∇uɛm1(x,T)|2dx+∬QTm1uɛα+m1-1(a-buɛ+cvɛ)∂uɛ∂tdxdt=12∫Ω|∇uɛm1(x,0)|2dx-12∫Ω|∇uɛm1(x,T)|2dx+∬QT2m1m1+1uɛ(2α+m1-1)/2(a-buɛ+cvɛ)∂uɛ(m1+1)/2∂tdxdt≤12∫Ω|∇uɛm1(x,0)|2dx+2m1∬QTuɛ2α+m1-1(a-buɛ+cvɛ)2dxdt+2m1(m1+1)2∬QT|∂∂tuɛ(m1+1)/2|2dxdt,
which together with the bound of a,b,c,uɛ,vɛ shows that
∬QT|∂uɛ(m1+1)/2∂t|2dxdt≤C,
where C is a positive constant independent of ɛ. Noticing the bound of uɛ, we have that
∬QT|∂uɛm1∂t|2dxdt=4m12(m1+1)2∬QTuɛm1-1|∂∂tuɛ(m1+1)/2|2dxdt≤C.
It is the same for the second inequality. The proof is completed.
From the above estimates of uɛ,vɛ, we have the following results.
Theorem 2.5.
The problem (1.1)–(1.4) admits a generalized solution.
Proof.
By Lemmas 2.2, 2.3, and 2.4, we can see that there exist subsequences of {uɛ},{vɛ} (denoted by themselves for simplicity) and functions u,v such that
uɛ⟶u,vɛ⟶v,a.einQT,∂uɛm1∂t⟶∂um1∂t,∂vɛm2∂t⟶∂vm2∂t,weaklyinL2(QT),∇uɛm1⟶∇um1,∇vɛm2⟶∇vm2,weaklyinL2(QT),
as ɛ→0. Then a rather standard argument as [23] shows that (u,v) is a generalized solution of (1.1)–(1.4) in the sense of Definition 1.1.
In order to prove that the generalized solution of (1.1)–(1.4) is uniformly bounded, we need the following comparison principle.
Lemma 2.6.
Let (u̲,v̲) be a subsolution of the problem (1.1)–(1.4) with the initial value (u̲0,v̲0) and (u¯,v¯) a supersolution with a positive lower bound of the problem (1.1)–(1.4) with the initial value (u¯0,v¯0). If u̲0≤v¯0, u̲0≤v¯0, then u̲(x,t)≤u¯(x,t), v̲(x,t)≤v¯(x,t) on QT.
Proof.
Without loss of generality, we might assume that ∥u̲(x,t)∥L∞(QT), ∥u¯(x,t)∥L∞(QT),∥v̲(x,t)∥L∞(QT), ∥v¯(x,t)∥L∞(QT)≤M, where M is a positive constant. By the definitions of subsolution and supersolution, we have that
∫0t∫Ω-u̲∂φ∂t+∇u̲m1∇φdxdτ+∫Ωu̲(x,t)φ(x,t)dx-∫Ωu̲0(x)φ(x,0)dx≤∫0t∫Ωu̲α(a-bu̲+cv̲)φdxdτ,∫0t∫Ω-u¯∂φ∂t+∇u¯m1∇φdxdτ+∫Ωu¯(x,t)φ(x,t)dx-∫Ωv¯0(x)φ(x,0)dx≥∫0t∫Ωu¯α(a-bu¯+cv¯)φdxdτ.
Take the test function as
φ(x,t)=Hɛ(u̲m1(x,t)-u¯m1(x,t)),
where Hɛ(s) is a monotone increasing smooth approximation of the function H(s) defined as follows:
H(s)={1,s>0,0,otherwise.
It is easy to see that Hɛ′(s)→δ(s) as ɛ→0. Since ∂u̲m1/∂t,∂u¯m1/∂t∈L2(QT), the test function φ(x,t) is suitable. By the positivity of a,b,c we have that
∫Ω(u̲-u¯)Hɛ(u̲m1-u¯m1)dx-∫0t∫Ω(u̲-u¯)∂Hɛ(u̲m1-u¯m1)∂tdxdτ+∫0t∫ΩHɛ′(u̲m1-u¯m1)|∇(u̲m1-u¯m1)|2dxdτ≤∫0t∫Ωa(u̲α-u¯α)Hɛ(u̲m1-u¯m1)+c(u̲αv̲-u¯αv¯)Hɛ(u̲m1-u¯m1)dxdτ,
where C is a positive constant depending on ∥a(x,t)∥C(Qt),∥c(x,t)∥C(Qt). Letting ɛ→0 and noticing that
∫0t∫ΩHɛ′(u̲m1-u¯m1)|∇(u̲m-u¯m)|2dxdτ≥0,
we arrive at
∫Ω[u̲(x,t)-u¯(x,t)]+dx≤C∫0t∫Ω(u̲α-u¯α)++v̲(u̲α-u¯α)++u¯α(v̲-v¯)+dxdτ.
Let (u¯,v¯) be a supsolution with a positive lower bound σ. Noticing that
(xα-yα)+≤C(α)(x-y)+,forα≥1,(xα-yα)+≤xα-1(x-y)+≤yα-1(x-y)+,forα<1,
with x,y>0, we have that
∫0t∫Ω(u̲α-u¯α)++v̲(u̲α-u¯α)++u¯α(v̲-v¯)+dxdτ≤C∫0t∫Ω(u̲-u¯)++(v̲-v¯)+dxdτ,
where C is a positive constant depending upon α,σ,M.
Similarly, we also have that
∫Ω[v̲(x,t)-v¯(x,t)]+dx≤C∫0t∫Ω(u̲-u¯)++(v̲-v¯)+dxdτ.
Combining the above two inequalities, we obtain
∫Ω[u̲(x,t)-u¯(x,t)]++[v̲(x,t)-v¯(x,t)]+dx≤C∫0t∫Ω(u̲-u¯)++(v̲-v¯)+dxdτ.
By Gronwall's lemma, we see that u̲≤u¯,v̲≤v¯. The proof is completed.
Corollary 2.7.
If blfl>cMeM, then the problem (1.1)–(1.4) admits at most one global solution which is uniformly bounded in Ω¯×[0,∞).
Proof.
The uniqueness comes from the comparison principle immediately. In order to prove that the solution is global, we just need to construct a bounded positive supersolution of (1.1)–(1.4).
Let ρ1=(aMfl+dMcM)/(blfl-cMeM) and ρ2=(aMeM+dMbl)/(blfl-cMeM), since blfl>cMeM; then ρ1,ρ2>0 and satisfy
aM-blρ1+cMρ2=0,dM+eMρ1-flρ2=0.
Let (u¯,v¯)=(ηρ1,ηρ2), where η>1 is a constant such that (u0,v0)≤(ηρ1,ηρ2); then we have that
u¯t-Δu¯m1=0≥u¯α(a-bu¯+cv¯),v¯t-Δv¯m2=0≥v¯β(d+eu¯-fv¯).
That is, (u¯,v¯)=(ηρ1,ηρ2) is a positive supersolution of (1.1)–(1.4). Since u¯,v¯ are global and uniformly bounded, so are u and v.
3. Periodic Solutions
In order to establish the existence of the nontrivial nonnegative periodic solutions of the problem (1.1)–(1.3), we need the following lemmas. Firstly, we construct a pair of T-periodic supersolution and T-periodic subsolution as follows.
Lemma 3.1.
In case of blfl>cMeM, there exists a pair of T-periodic supersolution and T-periodic subsolution of the problem (1.1)–(1.3).
Proof.
We first construct a T-periodic subsolution of (1.1)–(1.3). Let λ be the first eigenvalue and ϕ be the uniqueness solution of the following elliptic problem:
-Δϕ=λϕ,x∈Ω,ϕ=0,x∈∂Ω;
then we have that
λ>0,ϕ(x)>0inΩ,|∇ϕ|>0on∂Ω,M=maxx∈Ω¯ϕ(x)<∞.
Let
(u̲,v̲)=(ɛϕ2/m1(x),ɛϕ2/m2(x)),
where ɛ>0 is a small constant to be determined. We will show that (u̲,v̲) is a (time independent, hence T-periodic) subsolution of (1.1)–(1.3).
Taking the nonnegative function φ1(x,t)∈C1(Q¯T) as the test function, we have that
∬QT(u̲∂φ1∂t+Δu̲m1φ1+u̲α(a-bu̲+cv̲)φ1)dxdt+∫Ωu̲(x,0)φ1(x,0)-u̲(x,T)φ1(x,T)dx=∬QT(u̲α(a-bu̲+cv̲)+Δu̲m1)φ1dxdt=∬QTu̲α(a-bu̲+cv̲)φ1dxdt-∬QT∇u̲m1∇φ1dxdt=∬QTu̲α(a-bu̲+cv̲)φ1dxdt-2ɛm1∬QTϕ∇ϕ⋅∇φ1dxdt=∬QTu̲α(a-bu̲+cv̲)φ1dxdt-2ɛm1∬QT∇ϕ∇(ϕφ1)-|∇ϕ|2φ1dxdt=∬QTu̲α(a-bu̲+cv̲)φ1dxdt-2ɛm1∬QT-div(∇ϕ)ϕφ1-|∇ϕ|2φ1dxdt=∬QTu̲α(a-bu̲+cv̲)φ1dxdt-2ɛm1∬QT(λϕ2-|∇ϕ|2)φ1dxdt.
Similarly, for any nonnegative test function φ2(x,t)∈C1(Q¯T), we have that
∬QT(v̲∂φ2∂t+Δv̲m2φ2+v̲β(d+eu̲-fv̲)φ2)dxdt+∫Ωv̲(x,0)φ2(x,0)-v̲(x,T)φ2(x,T)dx=∬QTv̲β(d+eu̲-fv̲)φ2dxdt-2ɛm2∬QT(λϕ2-|∇ϕ|2)φ2dxdt.
We just need to prove the nonnegativity of the right-hand side of (3.4) and (3.5).
Since ϕ1=ϕ2=0,|∇ϕ1|,|∇ϕ2|>0 on ∂Ω, then there exists δ>0 such that
λϕ2-|∇ϕ|2≤0,x∈Ω¯δ,
where Ω¯δ={x∈Ω∣dist(x,∂Ω)≤δ}. Choosing
ɛ≤min{albMM2/m1,dlfMM2/m2},
then we have that
2ɛm1∫0T∫Ωδ(λϕ2-|∇ϕ|2)φ1dxdt≤0≤∫0T∫Ωδu̲α(a-bu̲+cv̲)φ1dxdt,2ɛm2∫0T∫Ωδ(λϕ2-|∇ϕ|2)φ2dxdt≤0≤∫0T∫Ωδv̲β(d+eu̲-fv̲)φ2dxdt,
which shows that (u̲,v̲) is a positive (time independent, hence T-periodic) subsolution of (1.1)–(1.3) on Ω¯δ×(0,T).
Moreover, we can see that, for some σ>0,
ϕ(x)≥σ>0,x∈Ω∖Ω¯δ.
Choosing
ɛ≤min{al2bMM2/m1,(al4λM2(m1-α)/m1)1/(m1-α),dl2fMM2/m2,(dl4λM2(m2-β)/m2)1/(m2-β)},
then
ɛαϕ2α/m1a-bɛα+1ϕ2(α+1)/m1+cɛαϕ2α/m1ɛϕ2/m2-2ɛm1λϕ2≥0,ɛβϕ2β/m2d+eɛϕ2/m1ɛβϕ2β/m2-fɛβ+1ϕ2(β+1)/m2-2ɛm2λϕ2≥0
on QT, that is
∬QTu̲α(a-bu̲+cv̲)φ1dxdt-2ɛm1∬QT(λϕ2-|∇ϕ|2)φ1dxdt≥0,∬QTv̲β(d+eu̲-fv̲)φ2dxdt-2ɛm2∬QT(λϕ2-|∇ϕ|2)φ2dxdt≥0.
These relations show that (u̲,v̲)=(ɛϕ12/m1(x),ɛϕ22/m2(x)) is a positive (time independent, hence T-periodic) subsolution of (1.1)–(1.3).
Letting (u¯,v¯)=(ηρ1,ηρ2), where η,ρ1,ρ2 are taken as those in Corollary 2.7, it is easy to see that (u¯,v¯) is a positive (time independent, hence T-periodic) subsolution of (1.1)–(1.3).
Obviously, we may assume that u̲(x,t)≤u¯(x,t), v̲(x,t)≤v¯(x,t) by changing η, ɛ appropriately.
Lemma 3.2 (see [24, 25]).
Let u be the solution of the following Dirichlet boundary value problem
∂u∂t=Δum+f(x,t),(x,t)∈Ω×(0,T),u(x,t)=0,(x,t)∈∂Ω×(0,T),
where f∈L∞(Ω×(0,T)); then there exist positive constants K and α∈(0,1) depending only upon τ∈(0,T) and ∥f∥L∞(Ω×(0,T)), such that, for any (xi,ti)∈Ω×[τ,T](i=1,2),
|u(x1,t1)-u(x2,t2)|≤K(|x1-x2|α+|t1-t2|α/2).
Lemma 3.3 (see [26]).
Define a Poincaré mapping
Pt:L∞(Ω)×L∞(Ω)⟶L∞(Ω)×L∞(Ω),Pt(u0(x),v0(x))∶=(u(x,t),v(x,t))(t>0),
where (u(x,t),v(x,t)) is the solution of (1.1)–(1.4) with initial value (u0(x),v0(x)). According to Lemmas 2.6 and 3.2 and Theorem 2.5, the map Pt has the following properties:
Pt is defined for any t>0 and order preserving;
Pt is order preserving;
Pt is compact.
Observe that the operator PT is the classical Poincaré map and thus a fixed point of the Poincaré map gives a T-periodic solution setting. This will be made by the following iteration procedure.
Theorem 3.4.
Assume that blfl>cMeM and there exists a pair of nontrivial nonnegative T-periodic subsolution (u̲(x,t),v̲(x,t)) and T-periodic supersolution (u¯(x,t),v¯(x,t)) of the problem (1.1)–(1.3) with u̲(x,0)≤u¯(x,0); then the problem (1.1)–(1.3) admits a pair of nontrivial nonnegative periodic solutions (u*(x,t),v*(x,t)),(u*(x,t),v*(x,t)) such that
u̲(x,t)≤u*(x,t)≤u*(x,t)≤u¯(x,t),v̲(x,t)≤v*(x,t)≤v*(x,t)≤v¯(x,t),inQT.
Proof.
Taking u¯(x,t), u̲(x,t) as those in Lemma 3.1 and choosing suitable B(x0,δ),B(x0,δ′),Ω′, k1, k2, and K, we can obtain u̲(x,0)≤u¯(x,0). By Lemma 2.6, we have that PT(u̲(·,0))≥u̲(·,T). Hence by Definition 1.2 we get PT(u̲(·,0))≥u̲(·,0), which implies P(k+1)T(u̲(·,0))≥PkT(u̲(·,0)) for any k∈ℕ. Similarly we have that PT(u¯(·,0))≤u¯(·,T)≤u¯(·,0), and hence P(k+1)T(u¯(·,0))≤PkT(u¯(·,0)) for any k∈ℕ. By Lemma 2.6, we have that PkT(u̲(·,0))≤PkT(u¯(·,0)) for any k∈ℕ. Then
u*(x,0)=limk→∞PkT(u̲(x,0)),u*(x,0)=limk→∞PkT(u¯(x,0))
exist for almost every x∈Ω. Since the operator PT is compact (see Lemma 3.3), the above limits exist in L∞(Ω), too. Moreover, both u*(x,0) and u*(x,0) are fixed points of PT. With the similar method as [26], it is easy to show that the even extension of the function u*(x,t), which is the solution of the problem (1.1)–(1.4) with the initial value u*(x,0), is indeed a nontrivial nonnegative periodic solution of the problem (1.1)–(1.3). It is the same for the existence of u*(x,t). Furthermore, by Lemma 2.6, we obtain (3.16) immediately, and thus we complete the proof.
Furthermore, by De Giorgi iteration technique, we can also establish a prior upper bound of all nonnegative periodic solutions of (1.1)–(1.3). Then with a similar method as [18], we have the following remark which shows the existence and attractivity of the maximal periodic solution.
Remark 3.5.
If blfl>cMeM, the problem (1.1)–(1.3) admits a maximal periodic solution (U,V). Moreover, if (u,v) is the solution of the initial boundary value problem (1.1)–(1.4) with nonnegative initial value (u0,v0), then, for any ɛ>0, there exists t depending on u0, v0, and ɛ, such that
0≤u≤U+ɛ,0≤v≤V+ɛ,forx∈Ω,t≥t.
Acknowledgments
This work was supported by NSFC (10801061), the Fundamental Research Funds for the Central Universities (Grant no. HIT. NSRIF. 2009049), Natural Sciences Foundation of Heilongjiang Province (Grant no. A200909), and also the 985 project of Harbin Institute of Technology.
CuiJ. A.Global asymptotic stability in n-species cooperative system with time delays19947145481263647HuD.ZhangZ.Four positive periodic solutions to a Lotka-Volterra cooperative system with harvesting terms20101121115112110.1016/j.nonrwa.2009.02.0022571282ZBL1187.34050KimK. I.LinZ.A degenerate parabolic system with self-diffusion for a mutualistic model in ecology20067459760910.1016/j.nonrwa.2005.03.0202235221ZBL1120.35055LouY.NagylakiT.NiW.-M.On diffusion-induced blowups in a mutualistic model200145332934210.1016/S0362-546X(99)00346-61835478ZBL0980.35059LinZ.LiuJ.PedersenM.Periodicity and blowup in a two-species cooperating model201112147948610.1016/j.nonrwa.2010.06.0332729036ZBL1202.35115PaoC. V.1992New York, NY, USAPlenum Pressxvi+7771212084LuZ. Y.TakeuchiY.Permanence and global stability for cooperative Lotka-Volterra diffusion systems1992191096397510.1016/0362-546X(92)90107-P1192275ZBL0784.93092SunJ.WuB.ZhangD.Asymptotic behavior of solutions of a periodic diffusion equation201020101210.1155/2010/5975695975692586602ZBL1190.35021AhmadS.LazerA. C.Asymptotic behaviour of solutions of periodic competition diffusion system198913326328410.1016/0362-546X(89)90054-0986448ZBL0686.35060TineoA.Existence of global coexistence state for periodic competition diffusion systems199219433534410.1016/0362-546X(92)90178-H1178407ZBL0779.35058GopalsamyK.Global asymptotic stability in a periodic Lotka-Volterra system1985271667210.1017/S0334270000004768792408ZBL0588.92019López-GómezJ.Positive periodic solutions of Lotka-Volterra reaction-diffusion systems19925155721141727ZBL0754.35065MorganJ. J.HollisS. L.The existence of periodic solutions to reaction-diffusion systems with periodic data19952651225123210.1137/S00361410932571791347418ZBL0849.35052PaoC. V.Periodic solutions of parabolic systems with nonlinear boundary conditions1999234269571610.1006/jmaa.1999.64121695336ZBL0932.35111TianC.LinZ.Periodic solutions of reaction diffusion systems in a half-space domain20089381182110.1016/j.nonrwa.2007.01.0012392376ZBL1146.35305BrownK. J.HessP.Positive periodic solutions of predator-prey reaction-diffusion systems199116121147115810.1016/0362-546X(91)90202-C1111625ZBL0743.35030SunJ.-B.JinC.-H.KeY.-Y.Existence of non-trivial nonnegative periodic solutions for a nonlinear diffusion system20072321671752339817ZBL1142.35322SunJ.-B.Asymptotic bounds for solutions of a periodic reaction diffusion system2010101281352725709YinJ.WangY.Asymptotic behaviour of solutions for nonlinear diffusion equation with periodic absorption1999River Edge, NJ, USAWorld Scientific3053081742042ZBL0987.35086FriedmanA.1964Englewood Cliffs, NJ, USAPrentice-Hallxiv+3470181836LadyženskajaO. A.SolonnikovV. A.Ural'cevaN. N.196723Providence, RI, USAAmerican Mathematical Societyxi+648Translations of Mathematical Monographs0241822WuZ. Q.YinJ. X.WangC. P.2003Beijing, ChinaScientific PublicationsWuZ.ZhaoJ.YinJ.LiH.2001River Edge, NJ, USAWorld Scientificxviii+5021881297DiBenedettoE.Continuity of weak solutions to a general porous medium equation19833218311810.1512/iumj.1983.32.32008684758ZBL0526.35042PorzioM. M.VespriV.Hölder estimates for local solutions of some doubly nonlinear degenerate parabolic equations1993103114617810.1006/jdeq.1993.10451218742ZBL0796.35089HessP.PozioM. A.TeseiA.Time periodic solutions for a class of degenerate parabolic problems19952123673941342417ZBL0837.35075