This paper considers the singularity properties of positive solutions for a reaction-diffusion system with nonlocal boundary condition. The conditions on the existence and nonexistence of global positive solutions are given. Moreover, we establish the blow-up rate estimate for the blow-up solution.
1. Introduction
This paper studies the singularity properties of the following reaction-diffusion system with nonlocal boundary condition:
(1)ut=△u+uαvp,vt=△v+uqvβ,x∈Ω,t>0,u(x,t)=∫Ωf(x,y)u(x,y)dy,v(x,t)=∫Ωg(x,y)v(x,y)dy,x∈∂Ω,t>0,u(x,0)=u0(x),v(x,0)=v0(x),x∈Ω¯,
where Ω is a bounded domain of ℝN, N≥1, with smooth boundary ∂Ω and Ω¯ is the closure of Ω. α,β,p, and q are positive numbers which ensure that the equations in (1) are completely coupled with the nonlinear reaction terms. The functions f(x,y),g(x,y) defined for x∈∂Ω, y∈Ω¯ are nonnegative and continuous. The initial values u0(x) and v0(x) are nonnegative, which are mathematically convenient and currently followed throughout the paper. We also assume that (u0,v0) satisfies the compatibility condition on ∂Ω, and that f(x,·)≡0 and g(x,·)≡0 for any x∈∂Ω for the sake of the meaning of nonlocal boundary.
Denote that QT=Ω×(0,T), ST=∂Ω×(0,T), and Q¯T=Ω¯×[0,T), where 0<T≤+∞. A pair of functions (u(x,t),v(x,t)) is called a classical solution of problem (1) if (u,v)∈[C2,1(QT)∩C(Q¯T)]2 for some T, 0<T≤+∞, and satisfies (1). The local existence of classical solution of (1) is standard (see [1, 2]). If T<+∞, it is easy to see limt→T[maxx∈Ω¯u(x,t)+maxx∈Ω¯v(x,t)]=+∞, and we say that the solution (u(x,t),v(x,t)) of problem (1) blows up at finite time T. If T=+∞, (u(x,t),v(x,t)) is called a global solution of problem (1).
Over the past few years, many physical phenomena were formulated as nonlocal mathematical models (see [3, 4]). It has also being suggested that nonlocal growth terms present a more realistic model in physics for compressible reactive gases. Problem (1) arises in the study of the heat transfer with local source (see [5, 6]) and in the study of population dynamics (see [7, 8]).
In recent years many authors have investigated the following initial boundary value problem of reaction-diffusion system:
(2)ut-△u=f(u,v),vt-△v=g(u,v),x∈Ω,t>0
with Dirichlet, Neumanns or Robin boundary condition, which can be used to describe heat propagation on the boundary of container (see [2, 4, 9–23] and the literatures cited therein). Specially, when f(u,v),g(u,v) have the form
(3)f(u,v)=uαvp,g(u,v)=uqvβ,
a classical result is (see [9, 10, 12, 20]).
Theorem A.
The system (2) (f,g is of the form (3)) with homogenous Dirichlet boundary condition
(4)u(x,t)=v(x,t)=0,x∈Ω,t>0
admits a unique global solution for any nonnegative initial data u(x,t)=u0(x)≡0,v(x,t)=v0(x)≡0 if and only if α<1, β<1, and pq≤(1-α)(1-β).
However, there are some important phenomena formulated as parabolic equations which are coupled with nonlocal boundary conditions in mathematical modeling such as thermoelasticity theory (see [24–26]). In this case, the solution could be used to describe the entropy per volume of the material. The problem of nonlocal boundary conditions for linear parabolic equation of the type
(5)ut-Au=0,x∈Ω,t>0,u(x,t)=∫Ωf(x,y)u(y,t)dy,x∈∂Ω,t>0,u(x,0)=u0(x),x∈Ω¯
with uniformly elliptic operator
(6)Au=∑i,j=1Nai,j(x)∂2∂xi∂xj+∑i=1Nbi(x)∂∂xi+c(x)
and c(x)≤0 was studied by Friedman [26]. It was proved that the unique solution of (5) tends to 0 monotonically and exponentially as t→+∞ provided ∫Ωf(x,y)dy≤ρ<1 for any x∈∂Ω.
As for more general discussions on the dynamic of parabolic problem with nonlocal boundary conditions, one can see Pao [27], where the following problem:
(7)ut-Au=h(x,u),x∈Ω,t>0,α0∂u∂ν+u=∫Ωf(x,y)u(y,t)dy,x∈∂Ω,t>0,u(x,0)=u0(x),x∈Ω¯
was considered, and recently Pao [28] gave the numerical solutions for diffusion equation with nonlocal conditions.
In particular, the following single equation:
(8)ut-△u=uα,x∈Ω,t>0,u=∫Ωf(x,y)u(y,t)dy,x∈∂Ω,t>0,u(x,0)=u0(x),x∈Ω¯
under the assumption the ∫Ωf(x,y)dy=1 for any x∈∂Ω was consider, by Seo [29] and the following blow-up rate estimate is established:
(9)(α-1)-1/(α-1)(T*-t)≤maxx∈Ω¯u(x,t)≤C(T*-t)-1/(γ-1),
where T*<+∞ is the blow-up time, and γ is any constant satisfying 1<γ<α.
Recently, Kong and Wang [30] obtained the blow-up conditions and blow-up profiles of the following system by using some ideas of Souplet [4]:
(10)ut=△u+∫Ωuαvpdx,vt=△v+∫Ωuqvβdx,x∈Ω,t>0,u(x,t)=∫Ωf(x,y)u(x,y)dy,v(x,t)=∫Ωg(x,y)v(x,y)dy,x∈∂Ω,t>0,u(x,0)=u0(x),v(x,0)=v0(x),x∈Ω¯.
Furthermore, Zheng and Kong [31] gave the condition for global existence or nonexistence of solution to the following system:
(11)ut=△u+uα∫Ωvpdx,vt=△v+vβ∫Ωuqdx,x∈Ω,t>0,u(x,t)=∫Ωf(x,y)u(x,y)dy,v(x,t)=∫Ωg(x,y)v(x,y)dy,x∈∂Ω,t>0,u(x,0)=u0(x),v(x,0)=v0(x),x∈Ω¯.
Motivated by the above cited works, in this paper, we deal with singularity analysis of the parabolic system (1) with nonlocal boundary condition and it is seems that there is no work dealing with this type of systems except the single equations case, although this is a very classical model. Our main results read as follows.
Theorem 1.
If α<1, β<1, and pq≤(1-α)(1-β), then every nonnegative solution of (1) is global.
Theorem 2.
Suppose α>1 or β>1 or pq>(1-α)(1-β).
For any nonnegative functions f(x,y) and g(x,y), the nonnegative solution of (1) blows up in finite time provided the initial values are large enough.
If ∫Ωf(x,y)dy≥1 and ∫Ωg(x,y)dy≥1 for any x∈∂Ω, the nonnegative solution of (1) blows up in finite time with any positive initial values.
If ∫Ωf(x,y)dy<1 and ∫Ωg(x,y)dy<1 for any x∈∂Ω, the nonnegative solution of (1) is global with small initial values.
To estimate the blow-up rate of the blow-up solution of (1), we need to add some assumptions for initial data as follows.
u0(x),v0(x)∈C2+γ(Ω)∩C(Ω¯) for some 0<γ<1.
If p-β≥q-α≥1, then there exists a a sufficient small constant ɛ0>0 (which will be given in Section 4) such that △u0(x)+ɛ0u0λ≥0 and △v0(x)+ɛ0u0μ>0 on Ω¯, where
(12)λ=p(q+1)+α(1-β)p-β+1,μ=q(p+1)+β(1-α)p-β+1.
If q-α≥p-β≥1, then there exists a sufficient small constant ɛ0>0 (which will be given in Section 4) such that △u0(x)+ɛ0v0l≥0 and △v0(x)+ɛ0v0m>0 on Ω¯, where
(13)l=p(q+1)+α(1-β)q-α+1,m=q(p+1)+β(1-α)q-α+1.
Theorem 3.
Suppose p-β≥q-α≥1, and f(x,y)=g(x,y), ∫Ωf(x,y)dy≤1 for all x∈∂Ω and assumptions (H1)-(H2) hold. If (u(x,t),v(x,t)) is the smooth solution of (1) and blows up in finite time T*, then there exist positive constants Ci(i=1,2,3,4) such that
(14)C1≤maxx∈Ω¯u(x,t)(T*-t)(p-β+1)/(pq-(1-α)(1-β))≤C2,C3≤maxx∈Ω¯v(x,t)(T*-t)(q-α+1)/(pq-(1-α)(1-β))≤C4,
for 0<t<T*.
Theorem 4.
Suppose q-α≥p-β≥1, and f(x,y)=g(x,y), ∫Ωf(x,y)dy≤1 for all x∈∂Ω and assumptions (H1) and (H3) hold. If (u(x,t),v(x,t)) is the smooth solution of (1) and blows up in finite time T*, then there exist positive constants Ci(i=1,2,3,4) such that
(15)C1≤maxx∈Ω¯u(x,t)(T*-t)(p-β+1)/(pq-(1-α)(1-β))≤C2,C3≤maxx∈Ω¯v(x,t)(T*-t)(q-α+1)/(pq-(1-α)(1-β))≤C4,
for 0<t<T*.
The rest of this paper is organized as follows. In the next section, we give some preliminaries, which include the comparison principle related to system (1). In Section 3, we will study the conditions for the solution to blowup and exist globally and prove Theorems 1 and 2. In Section 4, we will establish the precise blowup rate estimate for small weighted nonlocal boundary and prove Theorems 3 and 4.
2. Preliminaries
In this section, we give some basic preliminaries. We begin with the definition of upper and lower solutions of (1).
Definition 5.
A pair of nonnegative functions (u¯(x,t),v¯(x,t)) is called an upper solution of problem (1) if (u¯,v¯)∈[C2,1(QT)∩C(Q¯T)]2 and satisfies
(16)u¯t≥△u¯+u¯αv¯p,v¯t≥△v¯+u¯qv¯β,(x,t)∈QT,u¯(x,t)≥∫Ωf(x,y)u¯(x,y)dy,v¯(x,t)≥∫Ωg(x,y)v¯(x,y)dy,(x,t)∈ST,u¯(x,0)≥u0(x),v¯(x,0)≥v0(x),x∈Ω¯.
Similarly, (u_,v_)∈[C2,1(QT)∩C(Q¯T)]2 is called a lower solution of (1) if it satisfied all the reversed inequalities in (16).
Lemma 6.
Let λi(x,t),Θi(x,t),i=1,2,f, and g be continuous and nonnegative functions, and let (w(x,t),z(x,t))∈[C2,1(QT)∩C(Q¯T)]2 satisfy
(17)wt≥△w+λ1(x,t)w(x,t)+Θ1(x,t)z(x,t),(x,t)∈QT,zt≥△z+λ2(x,t)w(x,t)+Θ2(x,t)z(x,t),(x,t)∈QT,w(x,t)≥∫Ωf(x,y)w(x,y)dy,z(x,t)≥∫Ωg(x,y)z(x,y)dy,(x,t)∈ST,w(x,0)≥0,z(x,0)≥0,x∈Ω¯,
and then w(x,t)>0,z(x,t)>0 on Q¯T.
Proof.
Set t1=sup{t∈(0,T):w(x,t)>0,z(x,t)>0,x∈Ω¯}. Since w(x,0)>0,z(x,0)>0, by continuous, there exists δ>0 such that w(x,t)>0,z(x,t)>0 for all (x,t)∈Ω¯×[0,δ). Thus, t1∈(δ,T].
We claim that t1<T will lead to a contradiction. Indeed, t1<T suggests that w(x1,t1)=0 or z(x1,t1)=0 for some x1∈Ω¯. Without loss of generality, we suppose that w(x1,t1)=0=inf(x,t)∈Q¯t1w(x,t).
If x1∈Ω, we first notice that
(18)wt-△w≥λ1(x,t)w(x,t)wt-△w≥+Θ1(x,t)z(x,t)≥0,(x,t)∈Qt1.
In addition, it is clear that w≥0 on the boundary ∂Ω and at initial state t=0. Then it follows from the strong maximum principle that w≡0 in Qt1, which contradicts w(x,0)>0.
If x1∈∂Ω, we will have a contradiction
(19)0=w(x1,t1)≥∫Ωf(x1,y)w(y,t1)>0.
In the last inequality, we have used the facts that f(x,·)≡0 for any x∈∂Ω and w(y,t1)>0 for any y∈Ω, which is a direct result of the previous case.
Therefore, the claim is true and thus t1=T, which implies that w>0,z>0 on Q¯T. The proof is complete.
Remark 7.
If ∫Ωf(x,y)dy≤1 and ∫Ωg(x,y)dy≤1 for any x∈∂Ω in Lemma 6, we can obtain w(x,t)≥0,z(x,t)≥0 on Q¯T under the assumption that w(x,0)≥0,z(x,0)≥0 for any x∈Ω¯. Indeed, for any ϵ>0, we can conclude that (w(x,t)+ϵet,z(x,t)+ϵet)>(0,0) on Q¯T as the proof of Lemma 6. Then the desired result follows from the limit procedure ϵ→0.
Lemma 8.
Let (u¯(x,t),v¯(x,t)) and (u_(x,t),v_(x,t)) be a upper and lower solution of (1) in QT, respectively. If (u¯(x,0),v¯(x,0))>(u_(x,0),v_(x,0)) for x∈Ω¯, then (u¯(x,t),v¯(x,t))>(u_(x,t),v_(x,t)) on Q¯T.
Proof.
Let w(x,t)=u¯(x,t)-u_(x,t) and z(x,t)=v¯(x,t)-v_(x,t), and then
(20)wt-△w≥u¯αv¯p-u_αv_p=v¯p(u¯α-u_α)+u_α(v¯p-v_p)wt-Δw=v¯pϕ1w+u_αϕ2z≔λ1(x,t)w(x,t)wt-Δw=+Θ1(x,t)z(x,t),zt-△z≥u¯qv¯β-u_qv_βzt-Δz=v¯β(u¯q-u_q)+u_q(v¯β-v_β)zt-Δz=v¯βϕ3w+u_qϕ4z≔λ2(x,t)w(x,t)zt-Δz=+Θ2(x,t)z(x,t),
where
(21)λ1(x,t)=v¯pϕ1=v¯p∫01α(λu¯+(1-λ)u_)α-1dλ,λ2(x,t)=v¯βϕ3=v¯β∫01q(λu¯+(1-λ)u_)q-1dλ,Θ1(x,t)=u_αϕ2=u_α∫01p(λv¯+(1-λ)v_)p-1dλ,Θ2(x,t)=u_qϕ4=u_q∫01β(λv¯+(1-λ)v_)β-1dλ.
So, the functions w(x,t) and z(x,t) satisfy
(22)wt≥△w+λ1(x,t)w(x,t)+Θ1(x,t)z(x,t),(x,t)∈QT,zt≥△z+λ2(x,t)w(x,t)+Θ2(x,t)z(x,t),(x,t)∈QT,w(x,t)≥∫Ωf(x,y)w(x,y)dy,z(x,t)≥∫Ωg(x,y)z(x,y)dy,(x,t)∈ST,w(x,0)≥0,z(x,0)≥0,x∈Ω¯.
Lemma 6 ensures that w(x,t)>0,z(x,t)>0 on Q¯T, that is, (u¯(x,t),v¯(x,t))>(u_(x,t),v_(x,t)) on Q¯T. The proof is complete.
3. Global Existence and Blowup
In this section, we will use the upper and lower solutions and their corresponding comparison principle developed in Section 2 to get the global existence or finite blowup of the solution to (1). Let us first give the proof of Theorem 1.
Proof of Theorem 1.
Using the condition α<1,β<1,pq≤(1-α)(1-β), and p>0,q>0, we have ((1-α)/p)·((1-β)/q)≥1. Thus, we can choose two positive constant and m,l>1 such that
(23)1-αp≥lm,1-βq≥ml.
Then, let ϕ(x,y)(x∈∂Ω,y∈Ω¯) be a continuous function such that ϕ(x,y)≥max{f(x,y),g(x,y)} and set
(24)a(x)=[∫Ωϕ(x,y)dy](1-m)/m,b(x)=[∫Ωϕ(x,y)dy](1-l)/l,x∈∂Ω.
We consider the following auxiliary problem:
(25)wt=△w+kw,x∈Ω,t>0,w(x,t)=(a(x)+b(x)+1)w(x,t)=×[∫Ω(ϕ(x,y)+1|Ω|)w(y,t)dy],x∈∂Ω,t>0,w(x,0)=1+u01/m(x)+v01/l(x),x∈Ω¯,
where |Ω| is the measure of Ω and k=(1/m)+(1/l). It follows from [32, Theorem 4.2] that w(x,t) exists globally and indeed w(x,t)>1 on Ω¯×[0,+∞) [32, Theorem 2.1].
Our aim is to show that (u¯,v¯)=(wm,wl) is a global upper solution of (1). Indeed, a direct computation yields
(26)u¯t-△u¯≥wm=(wm)αwm(1-α)=u¯α(wl)m(1-α)/l≥u¯αv¯p,x∈Ω,t>0.
Here, we have used that conclusion w>1 and the inequality (23). We still have to consider the boundary and initial conditions. When x∈∂Ω, we have
(27)u¯(x,t)=(a(x)+b(x)+1)mu¯(x,t)×[∫Ω(ϕ(x,y)+1|Ω|)w(y,t)dy]mu¯(x,t)≥(a(x))m[∫Ωϕ(x,y)w(y,t)dy]mu¯(x,t)=[∫Ωϕ(x,y)dy]1-m[∫Ωϕ(x,y)w(y,t)dy]mu¯(x,t)≥[∫Ωf(x,y)dy]1-m[∫Ωf(x,y)w(y,t)dy]mu¯(x,t)=[∫Ω(f1-m(x,y))1/(1-m)dy]1-mu¯(x,t)×[∫Ω(fm(x,y)wm(y,t))1/mdy]mu¯(x,t)≥∫Ωf1-m(x,y)(f(x,y)w(x,y))mdyu¯(x,t)=∫Ωf(x,y)wm(x,y)dyu¯(x,t)=∫Ωf(x,y)u¯(x,y)dy,x∈∂Ω,t>0.
Similarly, we have
(28)v¯t-△v¯≥u¯qvβ,x∈Ω,t>0,v¯≥∫Ωg(x,t)v¯(y,t)dy,x∈∂Ω,t>0.
It is clear that u0(x)<u¯(x,0) and v0(x)<v¯(x,0). Therefore, we get that (u¯,v¯) is a global upper solution of (1) and hence the solution of (1) exists globally by Lemma 8. The proof is complete.
Proof of Theorem 2.
(i) Let (u_,v_) be the solution of (2) and (3) with homogeneous Dirichlet boundary. Then it is well known for sufficiently large initial data that the solution (u_,v_) blows up in finite time when α>1 or β>1 or pq>(1-α)(1-β) (Theorem A). On the other hand, it is obvious that (u_,v_) is a lower solution of (1). Hence, the solution of (1) with large initial data blows up in finite time.
(ii) We consider the following ODE system:
(29)s′(t)=sαhp,h′(t)=sqhβ,t>0,s(0)=a=12minx∈Ω¯u0(x),h(0)=b=12minx∈Ω¯v0(x).
If α>1 or β>1, it is clear that the solution (s(t),h(t)) of (29) blows up in finite time. For the case 0<α,β<1 and pq>(1-α)(1-β), it follows that
(30)1q-α+1[sq-α+1(t)-aq-α+1].=1p-β+1[hp-β+1(t)-bp-β+1].
Thus, we get
(31)h′(t)=[q-α+1p-β+1hp-β+1h′(t)=D+(aq-α+1-q-α+1p-β+1bp-β+1)]q/(p-β+1)hβt>0,h(0)=b>0.
Then pq>(1-α)(1-β) implies that h(t) blows up in finite time, and so does f(t). From the above analysis, we see that α>1 or β>1 or pq>(1-α)(1-β) implies that (s,h) blows up in finite time. Under the assumption ∫Ωf(x,y)dy≥1 and ∫Ωg(x,y)dy≥1 for any x∈∂Ω, (s,h) is a lower solution of problem (1). Therefore, by Lemma 8, we see that the solution (u,v) of (1) satisfies (u,v)≥(s,h) and then (u,v) blows up in finite time.
(iii) Let ψ1(x) be the positive solution of the linear elliptic problem
(32)-Δψ1(x)=ɛ0,x∈Ω,ψ1(x)=∫Ωf(x,y)dy,x∈∂Ω,
and ψ2(x) be the positive solution of the linear elliptic problem
(33)-Δψ2(x)=ɛ0,x∈Ω,ψ2(x)=∫Ωg(x,y)dy,x∈∂Ω.
Since ∫Ωf(x,y)dy<1 and ∫Ωg(x,y)dy<1 for any x∈∂Ω, we can choose ɛ0>0 such that 0≤ψi(x)≤1, i=1,2.
Let u¯(x)=aψ1(x) and v¯(x)=bψ2(x), where a,b are positive constants which satisfy aɛ0≥aαbp,bɛ0≥aqbβ. We remark that under the assumption α>1 or β>1 or pq>(1-α)(1-β), we can choose such a,b easily. We now show that (u¯,v¯) is an upper solution of (1) for small initial data (u0,v0). Indeed, for any x∈Ω, we have
(34)u¯t-△u¯=aɛ0≥aαbp≥u¯αv¯p,v¯t-△v¯=bɛ0≥aqbβ≥u¯qv¯β.
When x∈∂Ω,
(35)u¯(x)=a∫Ωf(x,y)dy≥∫Ωf(x,y)aψ1(y)dyu¯(x)=∫Ωf(x,y)u¯(y)dy,v¯(x)=b∫Ωg(x,y)dy≥∫Ωg(x,y)bψ2(y)dyv¯(x)=∫Ωg(x,y)v¯(y)dy.
Here, we have used ψi≤1, i=1,2. The above inequalities show that (u¯,v¯) is an upper solution of (1) whenever u0(x)<aψ1(x) and v0(x)<bψ2(x). The proof is complete.
4. Blowup Rate
In this section, we will estimate the blow-up rate of (1). By the standard methods (see [1, 2, 6]), we can show that system (1) has a smooth solution (u,v) provided that u0,v0 satisfy the hypotheses (H1). We thus assume that the smooth solution (u,v) of (1) blows up at finite time T* and set M1(t)=maxx∈Ω¯u(x,t), M2(t)=maxx∈Ω¯v(x,t). We can obtain the blow-up rate from the following lemmas.
Lemma 9.
Suppose that u0(x),v0(x) satisfy (H1), and then there exists a positive constant K1 such that
(36)M1q-α+1(t)+M2p-β+1(t)≥K1(T*-t)-(q-α+1)(p-β+1)/(pq-(1-α)(1-β)).
Proof.
By the equations in (1), we have [33, Theorem 4.5]
(37)M1′≤M1αM2p,M2′≤M1qM2β,a.e.
Noticing that q-α+1>0 and p-β+1>0, we have
(38)(M1q-α+1(t)+M2p-β+1(t))′≤(p+q-α-β+2)M1β(t)M2α(t)≤K2(M1q-α+1(t)≤K2D+M2p-β+1(t))((q-α+1)q+(p-β+1)p)/((q-α+1)(p-β+1))
by virtue of Young's inequality. Integrating (38) from t to T*, we can get (36). The proof is complete.
Lemma 10.
If p-β≥q-α≥1, f(x,y)=g(x,y), and ∫Ωf(x,y)dy≤1 for any x∈∂Ω, then there exists a positive constant c0 such that the solution (u,v) of (1) with positive initial value (u0,v0) satisfies
(39)u(x,t)≥c0v(p-β+1)/(q-α+1)(x,t),(x,t)∈Ω×[0,T*).
Proof.
Let J(x,t)=u(x,t)-c0v(p-β+1)/(q-α+1)(x,t), where c0 is a positive constant to be chosen. For (x,t)∈Ω×(0,T*), a series of calculations shows that
(40)Jt-△J=ut-c0p-β+1q-α+1v(p-q+α-β)/(q-α+1)vt-△u+c0(p-β+1)(p-q+α-β)q-α+1v(p-2α-2q-β-1)/(q-α+1)×|∇u|2+c0p-β+1q-α+1v(p-q+α-β)/(q-α+1)△v≥uαvp-c0p-β+1q-α+1v(p-q+α-β)/(q-α+1)uqvβ=uαvβ+(p-q+α-β)/(q-α+1)×[v(p-β+1)(q-α)/(q-α+1)-c0p-β+1q-α+1uq-α]=uαvβ+(p-q+α-β)/(q-α+1)×[c0α-q(u-J)q-α-c0p-β+1q-α+1uq-α].
If we choose c0≤[(q-α+1)/(p-β+1)]1/(q-α+1), then c0α-q≥c0((p-β+1)/(q-α+1)). So, we have
(41)Jt-△J+c0(q-α)p-β+1q-α+1uαvβ+(p-q+α-β)/(q-α+1)θq-α+1(u,v)J≥0,
where θ(u,v) is a function of u and v, which lines between u-J and u.
When (x,t)∈∂Ω×(0,T*), we have
(42)J(x,t)=∫Ωf(x,y)u(y,t)dyJ(x,t)-c0[∫Ωf(x,y)v(y,t)dy](p-β+1)/(q-α+1).
Denote that H(x)=∫Ωf(x,y)dy≥0 for any x∈∂Ω. Since f(x,·)≡0 for any x∈∂Ω, H(x)>0. It follows from Jensen's inequality H(x)≤1 and (p-β+1)/(q-α+1)≥1 that
(43)∫Ωf(x,y)v(p-β+1)/(q-α+1)(y,t)dy-[∫Ωf(x,y)v(y,t)dy](p-β+1)/(q-α+1)≥H(x)[∫Ωf(x,y)v(y,t)H(x)dy](p-β+1)/(q-α+1)-[∫Ωf(x,y)v(y,t)dy](p-β+1)/(q-α+1)≥0.
Combining the above inequality with (42), we obtain
(44)J(x,t)≥∫Ωf(x,y)u(y,t)dyJ(x,t)≥-c0∫Ωf(x,y)v(p-β+1)/(q-α+1)(y,t)dyJ(x,t)=∫Ωf(x,y)J(y,t)dy.
For the initial condition, we have
(45)J(x,0)=u0(x)-c0v0(p-β+1)/(q-α+1)(x)≥0
on Ω¯ provided that c0≤infx∈Ω¯u0(x)v0-(p-β+1)/(q-α+1)(x).
Summarily, if we take c0 small enough such that
(46)c0≤min{[q-α+1p-β+1]1/(q-α+1)infx∈Ω¯u0(x)v0-(p-β+1)/(q-α+1)(x),c0≤min[q-α+1p-β+1]1/(q-α+1)},
it follows from (41), (44), (45), and [32, Theorem 2.1] that J(x,t)≥0, which implies (39). The proof is complete.
Combining (39) with (1), we know that the solution (u,v) of (1) satisfies
(47)ut≤△u+auλ,vt≤△v+bvμ,x∈Ω,0<t<T*,u(x,t)=∫Ωf(x,y)u(x,y)dy,v(x,t)=∫Ωg(x,y)v(x,y)d,x∈∂Ω,0<t<T*,u(x,0)=u0(x),v(x,0)=v0(x),x∈Ω¯
if f(x,y)=g(x,y), where a=[1/c0](p(q-α+1))/(p-β+1),b=[1/c0](β(q-α+1))/(p-β+1), λ=(p(q+1)+α(1-β))/(p-β+1), and μ=(q(p+1)+β(1-α))/(p-β+1). It is easy to see that λ>1 and μ>1 if p-β≥q-α≥1.
Let (w,z) be the solution of the following system:
(48)wt=△w+awλ,zt=△z+bwμ,x∈Ω,0<t<T*,w(x,t)=∫Ωf(x,y)w(x,y)dy,z(x,t)=∫Ωf(x,y)z(x,y)dy,x∈∂Ω,0<t<T*,w(x,0)=u0(x),z(x,0)=v0(x),x∈Ω¯.
It is easy to see that (w,z)≥(u,v) by Remark 7 if ∫Ωf(x,y)dy≤1 for any x∈∂Ω.
Lemma 11.
Suppose that u0(x) and v0(x) satisfy (H1)-(H2) and that the assumptions in Lemma 10 hold; then the solution (w,z) of (48) satisfies
(49)wt-aδwλ≥0,zt-bδwμ≥0,(x,t)∈Ω×(0,T*)
if ɛ0 is small enough such that 0<ɛ0<min{a,b} and δ≤min{1-ɛ0/a,1-ɛ0/b}.
Proof.
Let J1(x,t)=wt-aδwλ and J2(x,t)=zt-bδwμ. For (x,t)∈Ω×(0,T*), a series of calculations shows that
(50)J1t-△J1≥aλwλ-1J1,J2t-△J2≥bμwμ-1J1.
For (x,t)∈∂Ω×(0,T*), using the boundary conditions we have
(51)J1(x,t)=wt-aδwλJ1(x,t)=∫Ωf(x,y)wt(y,t)dyJ1(x,t)=-aδ[∫Ωf(x,y)w(y,t)dy]λJ1(x,t)=∫Ωf(x,y)(J1+aδwλ)(y,t)dyJ1(x,t)=-aδ[∫Ωf(x,y)w(y,t)dy]λJ1(x,t)=∫Ωf(x,y)J1(x,y)dyJ1(x,t)=+aδ[[∫Ωf(x,y)w(y,t)dy]λ∫Ωf(x,y)wλ(y,t)dyJ1(x,t)=+aδ-[∫Ωf(x,y)w(y,t)dy]λ].
It follows from ∫Ωf(x,y)dy≤1 for any x∈∂Ω and Jensen's inequality that
(52)∫Ωf(x,y)wλ(y,t)dy≥[∫Ωf(x,y)w(y,t)dy]λ.
Similarly, we have
(54)J2(x,t)≥∫Ωf(x,y)J2(x,y)dy,(x,t)∈∂Ω×(0,T*).
For the initial condition, under assumption (H2), we have
(55)J1(x,0)≥0,J2(x,t)≥0,x∈Ω¯
if ɛ0 is small enough such that 0<ɛ0<min{a,b} and δ≤min{1-ɛ0/a,1-ɛ0/b}. Then (49) follows from (50)–(55). The proof is complete.
Proof of Theorem 3.
Integrating the inequality for w in (49) on [t,T*) yields
(56)w(x,t)≤c1(T*-t)1/(1-λ),(x,t)∈Ω¯×[0,T*),
where c1=(aδ(λ-1))1/(1-λ). Since u(x,t)≤w(x,t), we obtain
(57)M1(t)≤c1(T*-t)1/(1-λ)=c1(T*-t)-(p-β+1)/(pq-(1-α)(1-β)).
Combining (36) and (57), we get
(58)C1≤maxx∈Ω¯u(x,t)(T*-t)(p-β+1)/(pq-(1-α)(1-β))≤C2,
where C1 and C2 are two positive constants.
Since M2′≤M1qM2β, it follows from (57) that
(59)M2-βM2′≤c1q(T*-t)-q(p-β+1)/(pq-(1-α)(1-β)).
Integrating this inequality from 0 to t, we get
(60)M2(t)≤c2(T*-t)-(q-α+1)/(pq-(1-α)(1-β)),
where c2 is a positive constant. Combining (36) and (60), we get
(61)C3≤maxx∈Ω¯v(x,t)(T*-t)(q-α+1)/(pq-(1-α)(1-β))≤C4,
where C3 and C4 are two positive constants. We completed the proof of Theorem 3.
Proof of Theorem 4.
The proof is similar to the proof of Theorem 3, and so we omit it.
Acknowledgments
This study is partially supported by the NSFC Grant 11201380, the Fundamental Research Funds for the Central Universities Grant XDJK2012B007, Doctor Fund of Southwest University Grant SWU111021, and Educational Fund of Southwest University Grant 2010JY053.
LadyženskajaO. A.SolonikV. A.Ural'cevaN. N.196723Providence, RI, USAAmerican Mathematical SocietyTranslations of Mathematical MonographsPaoC. V.1992New York, NY, USAPlenum Pressxvi+777MR1212084Bimpong-BotaK.OrtolevaP.RossJ.Far-from-equilibrium phenomena at local sites of reaction1974608312431332-s2.0-000155976710.1063/1.1681498SoupletP.Uniform blow-up profiles and boundary behavior for diffusion equations with nonlocal nonlinear source1999153237440610.1006/jdeq.1998.3535MR1683627ZBL0923.35077GalaktionovV. A.On asymptotic self-similar behaviour for a quasilinear heat equation: single point blow-up199526367569310.1137/S0036141093223419MR1325909ZBL0828.35067SamarskiiA. A.GalaktionovV. A.KurdyumovS. P.MikhailoiA. P.1987Moscow, RussiaNaukaCantrellR. S.CosnerC.Diffusive logistic equations with indefinite weights: population models in disrupted environments. II19892241043106410.1137/0522068MR1112065ZBL0726.92024FurterJ.GrinfeldM.Local vs. nonlocal interactions in population dynamics1989271658010.1007/BF00276081MR984226ChadamJ. M.PeirceA.YinH.-M.The blowup property of solutions to some diffusion equations with localized nonlinear reactions1992169231332810.1016/0022-247X(92)90081-NMR1180894ZBL0792.35095ChenH.Global existence and blow-up for a nonlinear reaction-diffusion system1997212248149210.1006/jmaa.1997.5522MR1464892ZBL0884.35068DicksteinF.EscobedoM.A maximum principle for semilinear parabolic systems and applications200145782583710.1016/S0362-546X(99)00419-8MR1845028ZBL0986.35044EscobedoM.HerreroM. A.A semilinear parabolic system in a bounded domain199316531533610.1007/BF01765854MR1271424ZBL0806.35088GalaktionovV. A.VázquezJ. L.The problem of blow-up in nonlinear parabolic equations20028239943310.3934/dcds.2002.8.399MR1897690ZBL1010.35057LiF.-C.HuangS.-X.XieC.-H.Global existence and blow-up of solutions to a nonlocal reaction-diffusion system2003961519153210.3934/dcds.2003.9.1519MR2017679ZBL1043.35069GangL.SleemanB. D.Non-existence of global solutions to systems of semi-linear parabolic equations199310411471682-s2.0-000111346910.1006/jdeq.1993.1066ZBL0816.35060MerleF.ZaagH.O.D.E. type behavior of blow-up solutions of nonlinear heat equations20028243545010.3934/dcds.2002.8.435MR1897691ZBL1009.35039RossiJ. D.WolanskiN.Blow-up vs. global existence for a semilinear reaction-diffusion system in a bounded domain19952011-121991200410.1080/03605309508821160MR1361728ZBL0851.35064SoupletP.Blow-up in nonlocal reaction-diffusion equations19982961301133410.1137/S0036141097318900MR1638054ZBL0909.35073WangM.WangY.Properties of positive solutions for non-local reaction-diffusion problems1996191411411156MR1409544WangM.Global existence and finite time blow up for a reaction-diffusion system200051116016710.1007/PL00001504MR1745297ZBL0984.35088ZhangJ.Boundedness and blow-up behavior for reaction-diffusion systems in a bounded domain199935783384410.1016/S0362-546X(97)00706-2MR1664896ZBL0932.35125ZhengS.Global boundedness of solutions to a reaction-diffusion system19992214354MR1669568ZBL0919.35060ZhengS.Global existence and global non-existence of solutions to a reaction-diffusion system200039332734010.1016/S0362-546X(98)00171-0MR1722823ZBL0955.35039DayW. A.A decreasing property of solutions of parabolic equations with applications to thermoelasticity1983404468475MR693879DayW. A.198530New York, NY, USASpringerviii+83Springer Tracts in Natural Philosophy10.1007/978-1-4613-9555-3MR804043FriedmanA.Monotonic decay of solutions of parabolic equations with nonlocal boundary conditions1986443401407MR860893ZBL0631.35041PaoC. V.Asymptotic behavior of solutions of reaction-diffusion equations with nonlocal boundary conditions199888122523810.1016/S0377-0427(97)00215-XMR1609090ZBL0920.35030PaoC. V.Numerical solutions of reaction-diffusion equations with nonlocal boundary conditions20011361-222724310.1016/S0377-0427(00)00614-2MR1855894ZBL0993.65094SeoS.Blowup of solutions to heat equations with nonlocal boundary conditions1996132123132MR1442200ZBL0874.35018KongL.-H.WangM.-X.Global existence and blow-up of solutions to a parabolic system with nonlocal sources and boundaries20075091251126610.1007/s11425-007-0105-5MR2370613ZBL1138.35031ZhengS.KongL.Roles of weight functions in a nonlinear nonlocal parabolic system20086882406241610.1016/j.na.2007.01.067MR2398660ZBL1138.35340DengK.Comparison principle for some nonlocal problems1992503517522MR1178431ZBL0777.35006FriedmanA.McLeodB.Blow-up of positive solutions of semilinear heat equations198534242544710.1512/iumj.1985.34.34025MR783924ZBL0576.35068