We are concerned with the uniqueness of mild solutions in the critical Lebesgue space Ln2(Rn) for the parabolic-elliptic Keller-Segel system, n≥4. For that, we prove the bicontinuity of the bilinear term of the mild formulation in the critical weak-Ln2 space, without using Kato time-weighted norms, time-spatial mixed Lebesgue norms (i.e., Lq((0,T);Lp)-norms with q≠∞), and any other auxiliary norms. Our proofs are based on Yamazaki's estimate, duality and Hölder's inequality, as well as an adapted Meyer-type argument. Since they are different from those of Kozono, Sugiyama and Yahagi [J. Diff. Eq. 253 (2012)] and it is not clear whether mild solutions are weak solutions in the critical C([0,T);Ln2), our results complement theirs in a twofold way. Moreover, the bilinear estimate together heat semigroup estimates yield a well-posedness result whose dependence with respect to the decay rate γ of the chemoattractant is also analyzed.
Citation: Lucas C. F. Ferreira. On the uniqueness of mild solutions for the parabolic-elliptic Keller-Segel system in the critical Lp-space[J]. Mathematics in Engineering, 2022, 4(6): 1-14. doi: 10.3934/mine.2022048
[1] | Lingzheng Kong, Haibo Chen . Normalized solutions for nonlinear Kirchhoff type equations in high dimensions. Electronic Research Archive, 2022, 30(4): 1282-1295. doi: 10.3934/era.2022067 |
[2] | Zhiwei Hao, Huiqin Zheng . Existence and multiplicity of solutions for fractional p(x)-Kirchhoff-type problems. Electronic Research Archive, 2023, 31(6): 3309-3321. doi: 10.3934/era.2023167 |
[3] | Guaiqi Tian, Hongmin Suo, Yucheng An . Multiple positive solutions for a bi-nonlocal Kirchhoff-Schr¨odinger-Poisson system with critical growth. Electronic Research Archive, 2022, 30(12): 4493-4506. doi: 10.3934/era.2022228 |
[4] | Tao Zhang, Tingzhi Cheng . A priori estimates of solutions to nonlinear fractional Laplacian equation. Electronic Research Archive, 2023, 31(2): 1119-1133. doi: 10.3934/era.2023056 |
[5] | Hui Jian, Min Gong, Meixia Cai . Global existence, blow-up and mass concentration for the inhomogeneous nonlinear Schrödinger equation with inverse-square potential. Electronic Research Archive, 2023, 31(12): 7427-7451. doi: 10.3934/era.2023375 |
[6] | Shiyong Zhang, Qiongfen Zhang . Normalized solution for a kind of coupled Kirchhoff systems. Electronic Research Archive, 2025, 33(2): 600-612. doi: 10.3934/era.2025028 |
[7] | Jiayi Fei, Qiongfen Zhang . On solutions for a class of Klein–Gordon equations coupled with Born–Infeld theory with Berestycki–Lions conditions on R3. Electronic Research Archive, 2024, 32(4): 2363-2379. doi: 10.3934/era.2024108 |
[8] | Christos Sourdis . A Liouville theorem for ancient solutions to a semilinear heat equation and its elliptic counterpart. Electronic Research Archive, 2021, 29(5): 2829-2839. doi: 10.3934/era.2021016 |
[9] | Xiaoyong Qian, Jun Wang, Maochun Zhu . Existence of solutions for a coupled Schrödinger equations with critical exponent. Electronic Research Archive, 2022, 30(7): 2730-2747. doi: 10.3934/era.2022140 |
[10] | Mohammad Kafini, Maher Noor . Delayed wave equation with logarithmic variable-exponent nonlinearity. Electronic Research Archive, 2023, 31(5): 2974-2993. doi: 10.3934/era.2023150 |
We are concerned with the uniqueness of mild solutions in the critical Lebesgue space Ln2(Rn) for the parabolic-elliptic Keller-Segel system, n≥4. For that, we prove the bicontinuity of the bilinear term of the mild formulation in the critical weak-Ln2 space, without using Kato time-weighted norms, time-spatial mixed Lebesgue norms (i.e., Lq((0,T);Lp)-norms with q≠∞), and any other auxiliary norms. Our proofs are based on Yamazaki's estimate, duality and Hölder's inequality, as well as an adapted Meyer-type argument. Since they are different from those of Kozono, Sugiyama and Yahagi [J. Diff. Eq. 253 (2012)] and it is not clear whether mild solutions are weak solutions in the critical C([0,T);Ln2), our results complement theirs in a twofold way. Moreover, the bilinear estimate together heat semigroup estimates yield a well-posedness result whose dependence with respect to the decay rate γ of the chemoattractant is also analyzed.
In this paper, we mainly study the following Kirchhoff-type system
{−(a1+b1∫R3|∇u|2dx)Δu=2αα+β|u|α−2u|v|β+εf(x),−(a2+b2∫R3|∇v|2dx)Δv=2βα+β|u|α|v|β−2v+εg(x),(u,v)∈D1,2(R3)×D1,2(R3), | (1.1) |
under the condition of (C.1),(C.2) where (C.1): a1,a2≥0,b1,b2>0,α,β>1,α+β=2∗=2×3/(3−2)=6,ε>0; (C.2): f(x),g(x)≥0,f(x),g(x)≢0,f(x),g(x)∈L65(R3). 2∗=2NN−2 is the critical Sobolev exponent where N=3.
It is well known that the critical problem
{−Δu=|u|2∗−2u,x∈Ω,u=0,x∈∂Ω | (1.2) |
does not have a minimizing energy solution when Ω is a domain different from RN. This result is derived from the sharp Sobolev inequality on RN [1,2,3]. However, the situation is different if there is a nonhomogeneous term f(x):
{−Δu=|u|2∗−2u+μf(x),x∈Ω,u=0,x∈∂Ω. | (1.3) |
Tarantello [4] showed that problem (1.3) has at least two solutions in bounded domains. The main idea is to use the Ekeland variational principle to get one solution u0 which is a local minimum solution and use the mountain pass theorem to obtain the second solution u1 with the energy
I(u1)<I(u0)+1NSN2 |
where S is the best constant for the Sobolev embedding D1,2(RN)↪L2NN−2(RN), namely
S=infu∈D1,2(RN)∖{0}∫RN|∇u|2dx(∫RN|u|2NN−2dx)N−2N. |
Let U be a solution for the following problem:
{−Δu=|u|2∗−2u,x∈RN,u∈D1,2(RN). | (1.4) |
Then, we know
Uε,y(x)=(N(N−2))N−24εN−22(ε2+|x−y|2)N−22, |
which are all positive solutions of (1.4) (see [5,6]) for any ε>0 and y∈RN. Moreover, we know that U satisfies
‖U‖2=|U|2∗2∗=SN2 | (1.5) |
where ‖U‖=(∫RN|∇u|2dx)12 and |U|s=(∫RN|u|sdx)1s are the norms of the Sobolev space D1,2(RN) and Lebesgue space Ls(RN),s∈[2,2∗], respectively. Han [7] extended (1.3) to the following elliptic system
{−Δu=2αα+βu|u|α−2|v|β+εf(x),inΩ,−Δv=2βα+β|u|αv|v|β−2+εg(x),inΩ,u=v=0,on∂Ω | (1.6) |
in bounded domains. The author used upper and lower solution methods and variational methods to prove that the problem (1.6) has at least two solutions for both subcritical and critical cases. The first solution (˜u0,˜v0) of (1.6) is also a local minimizer of the associated functional I and they obtained the second solution (˜u1,ˉv1) with
I(˜u1,˜v1)<I(˜u0,˜v0)+2N(Sα,β2)N2 |
where
Sα,β=inf(u,v)∈D1,2(RN)×D1,2(RN)∖{(0,0)}∫RN|∇u|2+|∇v|2dx(∫RN|u|α|v|βdx)2α+β. |
Indeed, according to [7] we know that
u0=kU,v0=lU |
is a solution of the following system
{−Δu=2αα+β|u|α−2u|v|β,−Δv=2βα+β|u|α|v|β−2v,(u,v)∈D1,2(RN)×D1,2(RN), | (1.7) |
where U is a solution of (1.4),
k=[(2αα+β)β(α+β2α)β−2]−12(α+β−2),l=[(2αα+β)α(α+β2α)α−2]−12(α+β−2) |
and
Sα,β=[(αβ)βα+β+(αβ)−αα+β]S. |
We are certain that the existence of solutions will be affected by the nonhomogeneous term f(x) or g(x) and the existence of a second solution for (1.3) or (1.6) will also be affected by Eq (1.2) or system (1.7).
When a1=a2, b1=b2, f=g and u=v, system (1.1) transforms into the following individual equation.
{−(a1+b1∫RN|∇u|2dx)Δu=|u|2∗−2u+εf(x),inRN,u∈D1,2(RN) | (1.8) |
which is related to the stationary analogue of the equation
ρ∂2u∂t2−(P0h+E2L∫L0|∂u∂t|2 dx)∂2u∂t2=0 |
presented by Kirchhoff in [8]. Here, the parameters in (1.8) carry the following interpretations: ρ represents the mass density, P0 the initial tension, h the cross-sectional area, E the Young's modulus of the material and L the length of the string. It was underscored in [9] that the Kirchhoff-type problem models a variety of physical and biological systems where u characterizes a process that relies on its own average (e.g., population density). Early investigations into the Kirchhoff-type problem can be traced back to the work of Bernstein [10] and Pohoẑaev [11]. Nevertheless, Eq (1.8) only attracted significant attention after Lions [12] introduced an abstract framework for such problems.
Liu et al. [5] discovered that problem (1.8) has at least two positive solutions when N=3,4 under certain assumptions for f(x). Specifically, to consider the existence of a second solution for problem (1.8), they needed to establish the existence of a unique positive solution for (1.8) when ε=0. In fact, they found that the positive solutions of
{−(a+b∫RN|∇u|2dx)Δu=|u|2∗−2u,inRN,u∈D1,2(RN) |
can be expressed as
Vε,λ,y(x)=λN−24Uε,y(x) |
with
λ=a+bSN2λN−22. |
The solvability or multiplicity of the Kirchhoff type equation with critical exponent has been extensively studied in recent years; see, for instance, [5,13,14,15,16,17,18,19,20,21,22,23,24,25] and references therein.
Inspired by the ideas presented in [5] and [4], we discuss system (1.1) with ε>0 small enough. First, we establish the existence of solutions for problem (1.1) when ε=0:
Theorem 1.1. Assume that ε=0 and (u0,v0) is a positive solution of (1.7) and
S1=∫R3|∇u0|2dx=α3(Sα,β2)32,S2=∫R3|∇v0|2dx=β3(Sα,β2)32. |
Then, we have
(i) If a1=a2=0, problem (1.1) has a unique positive solution z=(u′,v′) where
u′=(b1S1)α−28(b2S2)β8u0,v′=(b1S1)α8(b2S2)β−28v0. |
(ii) If a1 and a2 are not equal to 0 at the same time, problem (1.1) has a unique positive solution
z=(u′,v′)=(λ1u0,λ2v0). |
Indeed, if a1=0,a2≠0,
λα−21λβ2=(λα1λβ−22)β2+β(b1S1)4β+2,λα1λβ−22=a2+(b2S2)(b1S1)α2+β(λα1λβ−22)β−22+β. |
(iii) If a1≠0(>0),a2≠0(>0), problem (1.1) has at least a positive solution z=(u′,v′)=(λ1u0,λ2v0) where
λ1=C2−β81Cβ82,λ2=Cα81C2−α82 |
and
C1=a1+C2−β41Cβ42b1S1,C2=a2+Cα41C2−α42b2S2. |
Remark 1.1. In fact, from the following equations
λ1=C2−β81Cβ82,λ2=Cα81C2−α82 |
and
C1=a1+C2−β41Cβ42b1S1,C2=a2+Cα41C2−α42b2S2, |
we find that the unique or explicit form of (λ1,λ2) is affected by a1,a2,b1,b2. Thus, we can only obtain the uniqueness of (λ1,λ2) in cases (i)−(ii) and the explicit form of (λ1,λ2) in case (i). Furthermore, we believe that if a1,a2,b1,b2 satisfy certain assumptions, (λ1,λ2) will be unique, and we can also obtain the explicit form of (λ1,λ2).
We define ε∗=g(√9bS3α,β20)√S|f|65+|g|65 where g(√9bS3α,β20)=b4(√9bS3α,β20)3−13S3(√9bS3α,β20)5, b=12min(b1,b2). Next, we consider the existence of a local minimum solution for problem (1.1) by applying the Ekeland variational principle.
Theorem 1.2. Assuming conditions (C.1) and (C.2) hold, system (1.1) has a local minimum solution for any ε∈(0,ε∗).
Remark 1.2. First, we demonstrate that the minimal value of the set minimization problem can be attained by (u,v) and then we prove that (u,v) is a solution of (1.1). Unlike with single equations, due to the mutual interaction of (u,v), it is challenging to obtain ‖u‖≥‖un‖+o(1),‖v‖≥‖vn‖+o(1) where (un,vn) represents the minimizing sequence of c0. The definitions of un,vn,u,v,c0 can be found in the proof of Theorem 1.2 in Section 4. By drawing upon the proof method from Theorem 1.3 in [5] and employing meticulous estimates, we can overcome this difficulty.
Finally, we investigate the existence of a second solution for problem (1.1) by applying the mountain pass lemma and the concentration compactness principle. To obtain the energy estimation of the associated functional Φε for problem (1.1), we will need the explicit form of (λ1,λ2). Therefore, when a1=a2=0 we have:
Theorem 1.3. Assume α=β=3 and b1=b2, there exists ε∗∗∈(0,ε∗] such that for any ε∈(0,ε∗∗), problem (1.1) has another solution. The value of ε∗ is defined in Theorem 1.2.
Remark 1.3. First, we prove that the associated functional Φε for problem (1.1) satisfies the mountain pass structure, from which we obtain a (PS) sequence. Then, we establish the (PS) condition by using the concentration compactness principle. From this, we obtain another solution for (1.1). Owing to the lack of compactness (α+β=2∗), the mutual action of (u,v) and the influence of the nonlocal term, there arises a new challenge in employing the concentration compactness principle. Moreover, it is difficult to derive an explicit expression when the values of a1 and a2 are non-zero. The reason for only considering the case where α=β=3 is that after extensive estimation, it is only when α=β=3 that Λ reaches its minimum value Λmin. Therefore, the estimate satisfied by m contradicts Lemma 4.4(i). The definitions of Λ, Λmin, and m as well as the details of their related proofs can be found in Remark 4.2 and the Proof of Theorem 1.3.
Remark 1.4. The innovation of this paper lies in overcoming the lack of compactness (α+β=2∗) and the mutual interaction of (u,v) to demonstrate the existence of at least two solutions for systems (1.1). This extends the results from single equations in [5] to a system of equations. We accomplished this by applying the Ekeland variational principle, the mountain pass lemma and the concentration compactness principle as well as through some precise estimates.
The structure of this paper is as follows: Section 2 provides some preliminary background knowledge. Section 3 is dedicated to the proof of Theorem 1.1. Finally, we present the proofs for Theorem 1.2 and Theorem 1.3.
First, we introduce the following notations, which will be useful for proving the upcoming theorems in this section.
∙ The function space corresponding to problem (1.1) is E=D1,2(R3)×D1,2(R3) with the space norm defined as ‖(u,v)‖=(‖u‖2+‖v‖2)12. E∗ is the dual space of E.
∙(u,v)∈E,Bρ={u∈E:‖(u,v)‖<ρ}.
∙¯Bρ={(u,v)∈E:‖(u,v)‖≤ρ},∂Bρ={(u,v)∈E:‖(u,v)‖=ρ}.
∙ The following elliptic system
{−Δu=2αα+β|u|α−2u|v|β,−Δv=2βα+β|u|α|v|β−2v,(u,v)∈D1,2(R3)×D1,2(R3) | (2.1) |
has a positive radial vector solution z0=(u0,v0) under the condition (C.1) (see[7]).
∙u+=max{0,u},u−=max{0,−u}.
Let us denote the energy functional Φε:E→R corresponding to (1.1) by
Φε(u,v)=12(a1‖u‖2+a2‖v‖2)+14(b1‖u‖4+b2‖v‖4)−13∫R3(u+)α(v+)βdx−ε∫R3(fu+gv)dx. |
Obviously, Φε is of C1 and has the derivative given by
⟨Φ′ε(u,v),(φ,ψ)⟩=(a1+b1‖u‖2)∫R3∇u∇φdx+(a2+b2‖v‖2)∫R3∇v∇ψdx−α3∫R3(v+)β(u+)α−1φdx−β3∫R3(u+)α(v+)β−1ψdx−ε∫R3(fφ+gψ)dx. |
Next, we present the following lemma which can be utilized in the proof of Theorem 1.3.
Lemma 2.1. Assume α,β>1,α+β=6 and define
S=infu∈D1,2(R3)∖{0}∫R3|∇u|2dx(∫R3|u|α+βdx)2α+β,Sα,β=inf(u,v)∈E∖{(0,0)}∫R3|∇u|2+|∇v|2dx(∫R3|u|α|v|βdx)2α+β,˜Sα,β=inf(u,v)∈E∖{(0,0)}∫R3a|∇u|2+b|∇v|2dx(∫R3|u|α|v|βdx)2α+β. | (2.2) |
Then,
Sα,β=[(αβ)βα+β+(αβ)−αα+β]S, |
˜Sα,β=aαα+βbβα+β[(αβ)βα+β+(αβ)−αα+β]S |
where a, b is any real number.
Proof. Refer to [26] for the proof of Sα,β=[(αβ)β6+(αβ)−α6]S; we will provide the proof of ˜Sα,β later. Assume that ωn is a minimizing sequence for S, let s,t>0 to be chosen. Taking un=s√aωn,vn=t√bωn in (2.2), we have that
aαα+βbβα+βs2+t2(sαtβ)2α+β∫R3|∇ωn|2dx(∫R3|ωn|α+βdx)2α+β≥˜Sα,β. | (2.3) |
Noting that
s2+t2(sαtβ)2α+β=(st)2βα+β+(ts)2αα+β, | (2.4) |
we can proceed to define the function as follows.
g(x)=x2βα+β+x−2αα+β,x>0. |
When x=√αβ, there exists the minimum value
g(x)min=g(√αβ)=(αβ)βα+β+(αβ)−αα+β. | (2.5) |
Considering (2.3)–(2.5), we get
aαα+βbβα+β[(αβ)βα+β+(αβ)−αα+β]S≥˜Sα,β. | (2.6) |
Then, we need to prove that
aαα+βbβα+β[(αβ)βα+β+(αβ)−αα+β]S≤˜Sα,β. | (2.7) |
Let (un,vn) be a minimizing sequence for ˜Sα,β. Define zn=snvn, for some sn>0 such that
∫R3|un|α+βdx=∫R3|zn|α+βdx. | (2.8) |
By Young's inequality
∫R3|un|α|zn|βdx≤αα+β∫R3|un|α+βdx+βα+β∫R3|zn|α+βdx. | (2.9) |
By (2.8), we have
(∫R3|un|α|zn|βdx)2α+β≤(∫R3|un|α+βdx)2α+β=(∫R3|zn|α+βdx)2α+β. | (2.10) |
Using (2.10), we have
∫R3a|∇un|2+b|∇vn|2dx(∫R3|un|α|vn|βdx)2α+β=s2βα+βn∫R3a|∇un|2+b|∇vn|2dx(∫R3|un|α|zn|βdx)2α+β≥as2βα+βn∫R3|∇un|2dx(∫R3|un|α+βdx)2α+β+bs2βα+βns−2n∫R3|∇zn|2dx(∫R3|zn|α+βdx)2α+β≥h(sn)S, |
where h(sn)=as2βα+βn+bs−2αα+βn. Then, we get
h(sn)min=a(√bαaβ)2βα+β+b(√bαaβ)−2αα+β=aαα+βbβα+β[(αβ)βα+β+(αβ)−αα+β]. |
Therefore, (2.7) is proved. Combining (2.6), we get
˜Sα,β=aαα+βbβα+β[(αβ)β6+(αβ)−α6]S. |
Theorem 2.1. ([27], Theorem 4.1) Let M be a complete metric space with metric d and let I : M↦(−∞,+∞] be a lower semicontinuous function, bounded from below and not identical to +∞. Let ϵ>0 be given and u∈M be such that
I(u)≤infMI+ϵ. |
Then, there exists v∈M such that
I(v)≤I(u),d(u,v)≤1. |
For each w∈M, one has
I(v)≤I(w)+ϵd(v,w). |
In this section, we provide the proof of Theorem 1.1. The key idea is to observe that the right-hand side of problem (1.1), namely a1+b1∫R3|∇u|2dx and a2+b2∫R3|∇v|2dx, can be regarded as two constants. This insight guides us to construct the solution for problem (1.1) by utilizing the solution for problem (2.1) and the method of undetermined coefficients.
Proof of Theorem 1.1.The proof of Theorem 1.1 is inspired by the idea presented in [5]. For any C1,C2>0, let (u′,v′)=(λ1u0,λ2v0) where (u′,v′) is a vector solution of (3.1), λ1>0, λ2>0.
{−C1Δu=2αα+β|u|α−2u|v|β,−C2Δv=2βα+β|u|α|v|β−2v. | (3.1) |
By (3.1) and the fact that (u0,v0) satisfies (2.1), we can obtain
{C1=λα−21λβ2,C2=λα1λβ−22, | (3.2) |
which implies
{λ1=C2−β81Cβ82,λ2=Cα81C2−α82. | (3.3) |
Now, we consider the equations
C1=a1+b1∫R3|∇u′|2dx,C2=a2+b2∫R3|∇v′|2dx. |
Let ∫R3|∇u0|2dx=S1,∫R3|∇v0|2dx=S2. Thus, C1,C2 satisfy
{C1=a1+C2−β41Cβ42b1S1,C2=a2+Cα41C2−α42b2S2. | (3.4) |
Next, we consider the existence and uniqueness of the positive solution from (1.1) when ε=0.
(i) If a1=0,a2=0, we deduce from (3.4) that
{C1=(b1S1)2+α4(b2S2)β4,C2=(b1S1)α4(b2S2)2+β4. |
Combining with (3.3), we have
{λ1=(b1S1)α−28(b2S2)β8,λ2=(b1S1)α8(b2S2)β−28. |
Hence, we have
{u′=(b1S1)α−28(b2S2)β8u0,v′=(b1S1)α8(b2S2)β−28v0. | (3.5) |
(ii) If a1 and a2 are not equal to 0 at the same time, we can assume a1=0 and a2≠0 which implies that
C1=Cβ2+β2(b1S1)4β+2,C2=a2+(b2S2)(b1S1)α2+βCβ−22+β2. |
According to the above, we define
A=(b2S2)(b1S1)α2+β>0,−13<k=β−22+β<37,B=a2>0
where x>B. We want to determine the number of C2; we only need to find the solution of
f(x)=Axk−x+B=0. |
(1′) When β∈(1,2), k∈(−13,0), we can get
f′(x)=Akxk−1−1<0. |
Considering f(x)>0 as x→B, f(x)→−∞ as x→+∞. Thus, there exists a unique C2>B such that f(C2)=0.
(2′) When β=2,k=0, from (3.6)
{C1=(b2S2b1S1+a2)12b1S1,C2=b2S2b1S1+a2. |
Considering (3.2), (3.3), we have
{u′=(b2S2b1S1+a2)14u0,v′=(b1S1)12v0. |
(3′) When β∈(2,5),k∈(0,37), we know x0=(Ak)−1k−1 is the only maximum from
f′(x)=Akxk−1−1=0. |
Considering f(x)>0 as x→B, f(x)→−∞ as x→+∞. Thus, there exists a unique C2>B such that f(C2)=0. So, we can prove that there exists a unique (u′,v′) as a result of the only λ1,λ2. If a1≠0 and a2=0, we can get a unique (u′,v′) in the same way.
(iii) If a1≠0(>0),a2≠0(>0), we deduce from (3.4) that
C1−a1C2−a2=b1S1b2S2×C2C1. | (3.7) |
Let C1=λC2. By the above equality, we have
(b2S2C2)λ2−(a1b2S2)λ−(C2−a2)b1S1=0. |
Because λ=C1C2>0, we deduce from (3.7) that
λ=a1b2S2+√(a1b2S2)2+4(C2−a2)b1S1b2S2C22b2S2C2. |
Then, we have
C2−a2=(b2S2)4−α4C2−α42(a1b2S2+√(a1b2S2)2+4(C2−a2)b1S1b2S2C22)α4. |
Let A′=(12)α4(b2S2)4−α4,B′=a1b2S2,C′=a2,D′=4b1b2S1S2 and
h(x)=A′x2−α4(B′+√(B′)2+D′x(x−C′))α4−x+C′. |
As a result of
x→+∞,h(x)→−∞;x→C′,h(x)>0. |
Then, there exists a C2>0 such that h(C2)=0. Because the uniqueness is not clear, we have some difficulties in considering the existence of the second solution of problem (1.1). Hence, the Theorem 1.1 is proved.
In this section, we establish the existence of two solutions for (1.1) by using some variational methods. First, we will present the proofs of Theorem 1.2 and Theorem 1.3 utilizing various Lemmas for each proof, respectively. We consider the existence of a local minimum solution for problem (1.1) by applying the Ekeland variational principle.
Lemma 4.1. Assume that (C.1),(C.2) hold. Then, there exists ρ>0 such that for any ε∈(0,ε∗), one has Φε|∂Bρ≥˜α for some ˜α>0. For the definition of ε∗, please refer to Page 4, line 26.
Proof. By the Hölder inequalities, b1>0, b2>0, the complete square formula and Sobolev inequalities, one has
Φε(u,v)=12(a1‖u‖2+a2‖v‖2)+14(b1‖u‖4+b2‖v‖4)−13∫R3(u+)α(v+)βdx−ε∫R3(fu+gv)dx≥12(a1‖u‖2+a2‖v‖2)+14(b1‖u‖4+b2‖v‖4)−13×‖(u,v)‖6S3α,β−ε√S(|f|65+|g|65)‖(u,v)‖≥[b4‖(u,v)‖3−13‖(u,v)‖5S3α,β−ε√S(|f|65+|g|65)]‖(u,v)‖ |
where b=12min(b1,b2). Let g(t)=b4t3−13S3t5, then maxt≥0g(t)=g(ρ)>0 with ρ=√9bS3α,β20. Owing to ε∗=g(ρ)√S|f|65+|g|65, for any ε∈(0,ε∗), we have
Φε(u,v)≥[g(ρ)−ε√S(|f|65+|g|65)]ρ=˜α>0(∀(u,v)∈∂Bρ). |
Hence, we conclude the proof.
Lemma 4.2. Suppose that (C.1),(C.2) hold. Then, for any ε∈(0,ε∗) one has c0=infu∈¯BρΦε(u,v)∈(−∞,0) where ρ, ε∗ is given by Lemma 4.1.
Proof. Firstly, we choose a (u,v)∈E such that ∫R3(fu+gv)dx>0. Then, for any t>0 we have
Φε(tu,tv)=12(a1‖u‖2+a2‖v‖2)t2+14(b1‖u‖4+b2‖v‖4)t4−13t6∫R3(u+)α(v+)βdx−εt∫R3(fu+gv)dx. |
Hence, there exists a sufficiently small t>0 such that ‖t(u,v)‖≤ρ and Φε(tu,tv)<0 which leads to
c0≤Φε(tu,tv)<0. |
As ‖(u,v)‖≤ρ, we obtain
Φε(u,v)>−13∫R3(u+)α(v+)βdx−ε∫R3(fu+gv)dx≥−13‖(u,v)‖6S3α,β−ε√S(|f|65+|g|65)‖(u,v)‖>−∞. |
Consequently, we can establish that c0∈(−∞,0), thereby completing the proof.
Proof of Theorem 1.2. By applying Lemma 4.2, we find that c0=infu∈¯BρΦε(u,v)∈(−∞,0). Moreover, we know that Φε|∂Bρ>0 from Lemma 4.1. Therefore, we deduce that the minimum cannot be attained on ∂Bρ. According to Lemma 4.1, Lemma 4.2 and Theorem 2.1 (Ekeland variational principle), there exists (un,vn)∈Bρ such that Φε(un,vn)→c0 and Φ′ε(un,vn)→0 in E∗. The above proof can be referred to [[28], pp. 534-535]. Consequently, there exists (u,v)∈E satisfying:
un⇀u,vn⇀vinE,un→u,vn→v,inLsloc(R3)×Lsloc(R3),(1≤s<2∗)un→u,vn→v,a.e.onR3×R3. |
First, we prove that (u,v) is a minimizer for c0. Noting that ¯Bρ is closed and convex,
(u,v)∈¯Bρ,Φε(u,v)≥c0. |
Therefore, what we need to prove next is that Φε(u,v)≤c0. The key idea here is that since Φε(un,vn) converges to c0, we need to establish the inequality relationship between Φε(u,v) and Φε(un,vn).
In order to eliminate 14(b1‖u‖4+b2‖v‖4), we have the following estimates. From Φ′ε(un,vn)→0 in E∗, it holds that
⟨Φ′ε(un,vn),(u,v)⟩=(a1+b1‖un‖2)∫R3∇un∇udx+(a2+b2‖vn‖2)∫R3∇vn∇vdx −α3∫R3(u+n)α−1(v+n)βudx−β3∫R3(u+n)α(v+n)β−1vdx−ε∫R3(fu+gv)dx+o(1)=(a1+b1‖un‖2)‖u‖2+(a2+b2‖vn‖2)‖v‖2−2∫R3(u+)α(v+)βdx−ε∫R3(fu+gv)dx+o(1)=o(1). |
Combining above equality and weakly lower semi-continuity of norm, we have
Φε(u,v)=12(a1‖u‖2+a2‖v‖2)+14(b1‖u‖4+b2‖v‖4)−13∫R3(u+)α(v+)βdx−ε∫R3(fu+gv)dx≤12(a1‖u‖2+a2‖v‖2)+14(b1‖un‖2‖u‖2+b2‖vn‖2‖v‖2)−13∫R3(u+)α(v+)βdx−ε∫R3(fu+gv)dx−14(a1+b1‖un‖2)‖u‖2−14(a2+b2‖vn‖2)‖v‖2+12∫R3(u+)α(v+)βdx+ε4∫R3(fu+gv)dx+o(1)≤a14‖u‖2+a24‖v‖2+16∫R3(u+)α(v+)βdx−3ε4∫R3(fu+gv)dx+o(1)≤a14‖un‖2+a24‖vn‖2+16∫R3(u+n)α(v+n)βdx−3ε4∫R3(fun+gvn)dx+o(1)≤Φε(un,vn)−14⟨Φ′ε(un,vn),(un,vn)⟩+o(1)=c0. |
Hence, we get (u,v) is a minimizer for c0.
Now, we need to prove (u,v) is a solution of (1.1) with Φε(u,v)=c0. On one hand, we have
c0=Φε(u,v)−16⟨Φ′ε(un,vn),(u,v)⟩+o(1)=12(a1‖u‖2+a2‖v‖2)+14(b1‖u‖4+b2‖v‖4)−13∫R3(u+)α(v+)βdx−ε∫R3(fu+gv)dx−a16∫R3∇un∇udx−a26∫R3∇vn∇vdx−b16∫R3|∇un|2dx∫R3∇un∇udx−b26∫R3|∇vn|2dx∫R3∇vn∇vdx+α18∫R3(v+n)β(u+n)α−1udx+β18∫R3(u+n)α(v+n)β−1vdx+ε6∫R3(fu+gv)dx+o(1)=13(a1‖u‖2+a2‖v‖2)+14(b1‖u‖4+b2‖v‖4)−b16‖un‖2‖u‖2−b26‖vn‖2‖v‖2−5ε6∫R3(fu+gv)dx+o(1). |
On the other hand, we have
c0=Φε(un,vn)−16⟨Φ′ε(un,vn),(un,vn)⟩+o(1)=12(a1‖un‖2+a2‖vn‖2)+14(b1‖un‖4+b2‖vn‖4)−13∫R3(u+n)α(v+n)βdx−ε∫R3(fun+gvn)dx−a16‖un‖2−a26‖vn‖2−b16‖un‖4−b26‖vn‖4+13∫R3(v+n)β(u+n)αdx+ε6∫R3(fun+gvn)dx+o(1)=a13‖un‖2+a23‖vn‖2+b112‖un‖4+b212‖vn‖4−5ε6∫R3(fu+gv)dx. |
Hence, we have
13a1‖u‖+13a2‖v‖2+14b1‖u‖4+14b2‖v‖4−b16‖un‖2‖u‖2−b26‖vn‖2‖v‖2+o(1)=13a1‖un‖+13a2‖vn‖2+112b1‖un‖4+112b2‖vn‖4+o(1)≥13a1‖u‖+13a2‖v‖2+(14−16)b1‖u‖4+(14−16)b2‖v‖4+o(1)≥13a1‖u‖+13a2‖v‖2+(14−16)b1‖u‖4+14b2‖v‖4−16b2‖vn‖2‖v‖2+o(1). |
So, we obtain
‖u‖2≥‖un‖2+o(1). |
Following the above steps, we have
13a1‖u‖+13a2‖v‖2+14b1‖u‖4+14b2‖v‖4−b16‖un‖2‖u‖2−b26‖vn‖2‖v‖2+o(1)=13a1‖un‖+13a2‖vn‖2+112b1‖un‖4+112b2‖vn‖4+o(1)≥13a1‖u‖+13a2‖v‖2+(14−16)b1‖u‖4+(14−16)b2‖v‖4+o(1)≥13a1‖u‖+13a2‖v‖2+14b2‖u‖4−16b2‖un‖2‖u‖2+(14−16)b1‖v‖4+o(1). |
Thus, we obtain
‖v‖2≥‖vn‖2+o(1). |
Using again the weakly lower semi-continuity of norm, we get ‖(u,v)‖=‖(un,vn)‖+o(1). Combining un⇀u, vn⇀v in E, we have un→u,vn→v in E and then Φε(un,vn)→ c0=Φε(u,v),Φ′ε(un,vn)→Φ′ε(u,v)=0 in E∗. We complete the proof.
Now, we give proof of the second positive solution by the mountain pass lemma and the concentration compactness principle.
Lemma 4.3. Suppose that (C.1),(C.2) hold. Then, there exists (u∗,v∗)∈E such that ‖(u∗,v∗)‖>ρ and Φε(u∗,v∗)<0 where ρ is given by Lemma 2.
Proof. Let u0=kU,v0=lU, we can obtain kl=√αβ and
{k=314αβ−28β−β8=(αβ)β8(3α)14,l=314α−α8βα−28=(βα)α8(3β)14, |
where U is a solution of (1.4). Then, for any t>0 it holds that
Φε(tu0,tv0)=12(a1‖tu0‖2+a2‖tv0‖2)+14(b1‖tu0‖4+b2‖tv0‖4)−t63∫R3(u+0)α(v+0)β−εt∫R3(fu0+gv0)dx=12(a1k2+a2l2)S32t2+14(b1k4+b2l4)S3t4−13kαlβS32t6−εt∫R3(fu0+gv0)dx≤12(a1k2+a2l2)S32t2+14(b1k4+b2l4)S3t4−13kαlβS32t6. |
Hence, there exists a sufficiently large t0>0 such that
‖t0(u0,v0‖>ρandΦε(t0u0,t0v0)<0. |
Let (u∗,v∗)=(t0u0,t0v0). This completes the proof.
According to lemma 4.1, lemma 4.3, we can find (u,v) such that
inf∂BρΦε(u,v)≜d>Φε(0,0)=0,(u∗,v∗)∉¯BρsatisfyΦε(u∗,v∗)<d. |
Then, we define
m≜infP∈Amaxu∈PΦε(u,v)≥d, |
where A is the set of all passes which connect 0 and e=(u∗,v∗),i,e.,
A={P∈C([0,1],X)|P(0)=0,P(1)=e}. |
Remark 4.1. For any ε∈(0,ε∗), we can obtain a nonnegative bounded (PS) sequence.
Proof. By the mountain pass theorem [6], there exists (un,vn)∈E such that I(un,vn)→m and I′(un,vn)→0 in E∗. Thus, we can get
5ε6∫R3(fun+gvn)dx+m+o(‖un,vn‖)=Φε(un,vn)−16⟨Φ′ε(un,vn),(un,vn)⟩=a13‖un‖2+a23‖vn‖2+112‖un‖4+112‖vn‖4≥Q1‖(un,vn)‖2. | (4.1) |
where Q1 is a positive constant with a sufficiently small value.
By the Hölder inequality and the Sobolev inequality, we get
5ε6∫R3(fun+gvn)dx≤5ε6|f|2|un|2+5ε6|g|2|vn|2≤C(ϵ)‖(un,vn)‖ | (4.2) |
where C(ε) is a a sufficiently small constant that depends only on ε. Combining (4.1) with (4.2), we conclude that (un,vn) is bounded in E.
Since u=u+−u−,v=v+−v−, we have
o(1)=−⟨Φ′ε(un,vn),(u−n,v−n)⟩=−a1∫R3∇un∇u−ndx−a2∫R3∇vn∇v−ndx−b1‖un‖2∫R3∇un∇u−ndx−b2‖vn‖2∫R3∇vn∇v−ndx+ε∫R3(fu−n+gv−n)dx=a1‖u−n‖2+a2‖v−n‖2+b1‖un‖2‖u−n‖2+b2‖vn‖2‖v−n‖2+ε∫R3(fu−n+gv−n)dx≥b1‖un‖2‖u−n‖2+b2‖vn‖2‖v−n‖2≥b1‖u−n‖4+b2‖v−n‖4 |
which implies ‖u−n‖=0,‖v−n‖=0,n→∞. According to Hölder and Sobolev inequality, we have
0≤ε∫R3(fu−n+gv−n)dx≤Cε(|f|65‖u−n‖+|g|65‖v−n‖). |
Therefore,
ε∫R3(fu−n+gv−n)dx=0,n→∞. |
Next, we need to verify that
Φε(u+n,v+n)→m,⟨Φ′ε(u+n,v+n),(φ,ψ)⟩→0. |
Given that ‖u−n‖=0,‖v−n‖=0,n→∞ and ε∫R3(fu−n+gv−n)dx=0,n→∞, we have
Φε(un,vn)=12(a1‖un‖2+a2‖vn‖2)+14(b1‖un‖4+b2‖vn‖4)−13∫R3(u+)α(v+)βdx−ε∫R3(fun+gvn)dx=12[a1(‖u+n‖2+‖u−n‖2)+a2(‖v+n‖2+‖v−n‖2)]−13∫R3(u+n)α(v+n)βdx+14[b1(‖u+n‖4+‖v+n‖4)+b2(‖v+n‖4+‖v−n‖4)]−ε∫R3g(v+n−v−n)dx−ε∫R3f(u+n−u−n)dx=12(a1(‖u+n‖2+a2(‖v+n‖2)+14(b1(‖u+n‖4+b2(‖v+n‖4)−13∫R3(u+n)α(v+n)βdx−ε∫R3(fu+n+gv+n)dx+o(1)=Φε(u+n,v+n)+o(1). |
Given that ‖u−n‖=0,‖v−n‖=0,n→∞ and ε∫R3(fu−n+gv−n)dx=0,n→∞, we have
⟨Φ′ε(un,vn),(φ,ψ)⟩=a1∫R3∇un∇φdx+a2∫R3∇vn∇ψdx+b1‖un‖2∫R3∇un∇φdx+b2‖vn‖2∫R3∇vn∇ψdx−α3∫R3(v+n)β(u+n)α−1φdx−β3∫R3(u+n)α(v+n)β−1ψdx−ε∫R3(fφ+gψ)dx=a1∫R3∇(u+n−u−n)∇φdx+a2∫R3∇(v+n−v−n)∇ψdx+b1(‖u+n‖2+‖u−n‖2)∫R3∇(u+n−u−n)∇φdx+b2(‖v+n‖2+‖v−n‖2)∫R3(∇v+n−v−n)∇ψdx−α3∫R3(v+n)β(u+n)α−1φdx−β3∫R3(u+n)α(v+n)β−1ψdx−ε∫R3(fφ+gψ)dx=a1∫R3∇u+n∇φdx+a2∫R3∇v+n∇ψdx+b1‖u+n‖2∫R3∇u+n∇φdx+b2‖v+n‖2∫R3∇v+n∇ψdx−α3∫R3(v+n)β(u+n)α−1φdx−β3∫R3(u+n)α(v+n)β−1ψdx−ε∫R3(fφ+gψ)dx+o(1)=⟨Φ′ε(u+n,v+n),(φ,ψ)⟩+o(1). |
Then, we can obtain a nonnegative bounded sequence for Φε. We complete the proof.
Lemma 4.4. Suppose that (C.1),(C.2) hold. Then, there exists ε∗∗∈(0,ε∗) such that for any ε∈(0,ε∗∗) where Λ is the Maximum value for p(t), the following statements hold:
(i)a1=a2=0,m≤supt≥0Φε(tu′,tv′)<Λ−ε2(|f|65+|g|65)S14;(ii)a1=0,a2≠0,m≤supt≥0Φε(tu′,tv′)<Λ−9ε2|g|26516a2S−ε2|f|65S14;(iii)a1≠0,a2≠0,m≤supt≥0Φε(tu′,tv′)<Λ−9ε2|f|26516a1S−9ε2|g|26516a2S. |
Proof. Let
h(t)=Φε(tu′,tv′)=12(a1‖u′‖2+a2‖v′‖2)t2+14(b1‖u′‖4+b2‖v′‖4)t4−13t6∫R3(u′)α(v′)βdx−εt∫R3(fu′+gv′)dx, |
and
p(t)=12(a1‖u′‖2+a2‖v′‖2)t2+14(b1‖u′‖4+b2‖v′‖4)t4−13t6∫R3(u′)α(v′)βdx. |
Then, there exists t1>0 such that p′(t1)=0. In this case, we have
t21=14(b1‖u′‖4+b2‖v′‖4)+√116(b1‖u′‖4+b2‖v′‖4)2+12(a1‖u′‖2+a2‖v′‖2)∫R3(u′)α(v′)βdx∫R3(u′)α(v′)βdx. | (4.3) |
On the other hand, we know that (u′,v′) satisfies
a1‖u′‖2+a2‖v′‖2+b1‖u′‖4+b2‖v′‖4=2∫R3(u′)α(v′)βdx. | (4.4) |
Combining (4.3) and (4.4), we obtain t1=1 and
Λ=maxt>0p(t)=p(t1)=13(a1‖u′‖2+a2‖v′‖2)+112(b1‖u′‖4+b2‖v′‖4). |
Let ε1∈(0,ε∗]. Then, for t2∈(0,t1) and ε∈(0,ε1), we have:
(i) when a1=a2=0,
max0≤t≤t2h(t)≤max0≤t≤t2(12(a1‖u′‖2+a2‖v′‖2)t2+14(b1‖u′‖4+b2‖v′‖4)t4<Λ−ε2(|f|65+|g|65)S14. |
(ii) when a1=0,a2≠0,
max0≤t≤t2h(t)≤max0≤t≤t2(12(a1‖u′‖2+a2‖v′‖2)t2+14(b1‖u′‖4+b2‖v′‖4)t4<Λ−9ε2|g|26516a2S−ε2|f|65S14. |
(iii) when a1≠0,a2≠0,
max0≤t≤t2h(t)≤max0≤t≤t2(12(a1‖u′‖2+a2‖v′‖2)t2+14(b1‖u′‖4+b2‖v′‖4)t4<Λ−9ε2|f|26516a1S−9ε2|g|26516a2S. |
Choosing ε∗∗∈(0,ε1] for any ε∈(0,ε∗∗), we can deduce that for all t≥t2,
maxt≥t2h(t)≤maxt≥t2p(t)−εt2∫R3(fu′+gv′)dx=Λ−εt2∫R3(fu′+gv′)dx. |
Furthermore, we obtain the following inequalities:
(i) when a1=a2=0,
maxt≥t2h(t)<Λ−ε2(|f|65+|g|65)S14. |
(ii) when a1=0,a2≠0,
maxt≥t2h(t)<Λ−9ε2|g|26516a2S−ε2|f|65S14. |
(iii) when a1≠0,a2≠0,
maxt≥t2h(t)<Λ−9ε2|f|26516a1S−9ε2|g|26516a2S. |
Therefore, we complete the proof.
Proof of Theorem 1.3. According to Remark 4.1, we can get that {(un,vn)} is bounded and nonnegative. Up to a subsequence, there exists (u,v)⊂E such that un⇀u, vn⇀v in E, un→u, vn→v in Lsloc(R3)×Lsloc(R3)(1≤s<2∗) and un→u,vn→va.einR3. By applying the concentration compactness principle (see Proposition 2.2 in [29]), we can find non-negative measures μ and ν on R3, a vector function (u,v) and an at most countable set Γ such that as n→∞,
|∇un|2+|∇vn|2⇀μ,|un|α|vn|β⇀ν | (4.5) |
in the sense of measure and
(i)ν=|u|α|v|β+∑i∈Γνiδxi,μ≥(|∇u|2+|∇v|2)+∑i∈Γμiδxi;(ii)μi≥Sα,β(νi)2α+β,i∈Γ. | (4.6) |
Here, δxi is the Dirac delta measure concentrated xi. We claim that Γ=∅. Suppose by contradiction that Γ≠∅. To obtain a contradiction, we estimate m=limn→∞Φε(un,vn) by utilizing the assumption Γ≠∅ and the concentration compactness principle. By comparing this estimation of m with the one provided in Lemma 4.1, we deduce a contradiction. To do this, we first present the following relevant estimates. Fix k∈Γ. For ρ>0, assume that φkρ∈C∞0(R3) satisfies φkρ∈[0,1],
φkρ(x)=1,for|x−ak|≤ρ2;φkρ(x)=0,for|x−ak|≥ρ |
and |∇φkρ|≤ρ2. It follows from ⟨(Φ′ρ(un,vn),(φkρun,0)⟩→0 that
(a1+b1‖un‖2)(∫R3un∇un∇φkρdx+∫R3|∇un|2φkρdx)=α3∫R3uαnvβnφkρdx+ε∫R3fφkρundx+o(1). | (4.7) |
In the same way, it follows from ⟨(Φ′ρ(un,vn),(0,φkρvn)⟩→0 that
(a2+b2‖vn‖2)(∫R3vn∇vn∇φkρdx+∫R3|∇vn|2φkρdx)=β3∫R3uαnvβnφkρdx+ε∫R3gφρvkndx+o(1). | (4.8) |
First, we need to solve the lack of compactness problem from the critical Sobolev exponent which causes the invariance of dilation. Combining (4.7), (4.8) and Hölder inequality, we have
A1=limρ→0lim supn→∞(a1+b1‖un‖2)|∫R3un∇un∇φkρdx|≤limρ→0lim supn→∞C(∫Bρ(ak)|∇un|2dx)12(∫Bρ(ak)|∇φkρ|2|un|2dx)12≤limρ→0C(∫Bρ(ak)|u|6dx)16=0 | (4.9) |
and
A2=limρ→0lim supn→∞(a2+b2‖vn‖2)|∫R3vn∇vn∇φkρdx|=0 | (4.10) |
where Bρ(ak)={x∈R3:|x−ak|<ρ}. By (4.5) and (4.6), we have
limρ→0lim supn→∞(a1+b1‖un‖2)∫R3|∇un|2φkρdx+(a2+b2‖vn‖2)∫R3|∇vn|2φkρdx≥limρ→0lim supn→∞(a1∫R3|∇un|2φkρdx+a2∫R3|∇vn|2φkρdx)+[b1(∫R3|∇un|2φkρdx)2+b2(∫R3|∇vn|2φkρdx)2]≥limρ→0lim supn→∞(a1∫R3|∇un|2φkρdx+a2∫R3|∇vn|2φkρdx)+12(√b1∫R3|∇un|2φkρdx+√b2∫R3|∇vn|2φkρdx)2≥min(a1,a2)Sα,βν13i+12min(b1,b2)S2α,βν23i, | (4.11) |
limρ→0lim supn→∞(α3+β3)∫R3uαnvβnφkρdx=2limρ→0∫R3uαvβφkρ+2νi=2νi, | (4.12) |
limρ→0lim supn→∞∫R3(fφkρun+gφkρvn)dx=limρ→0∫R3(fφkρu+gφkρv)dx=0. | (4.13) |
We can deduce from (4.7)–(4.13) that
νi≥12min(a1,a2)Sα,βν13i+14min(b1,b2)S2α,βν23i. |
So, we have
νi≥(min(b1,b2)S2α,β+√[min(b1,b2)]2S4α,β+32min(a1,a2)Sα,β8)3,μi≥min(a1,a2)min(b1,b2)S3α,β+√[min(b1,b2)]2S6α,β+32min(a1,a2)S3α,β8. |
For R>0, assume that φR∈C∞0(R3) satisfies φR∈[0,1],
φR(x)=1,for|x|<R,φR(x)=0,for|x|>2R, |
and |∇φR|<2R. By applying the concentration compactness principle, we obtain
m=limn→∞Φε(un,vn)−14⟨Φ′ε(un,vn),(un,vn)⟩=limn→∞14(a1‖un‖2+a2‖vn‖2)+16∫R3uαnvβndx−3ε4∫R3(fun+gvn)dx≥limR→∞limn→∞14(a1∫R3|∇un|2φRdx+a2∫R3|∇vn|2φRdx)+16∫R3uαnvβnφRdx−3ε4∫R3(fu+gv)dx≥a14∫R3|∇u|2dx+a24∫R3|∇v|2dx+14μi+16νi−3ε4∫R3(fu+gv)dx. |
Hence, we can infer that
m≥a14∫R3|∇u|2dx+a24∫R3|∇v|2dx+14μi+16νi−3ε4∫R3(fu+gv)dx. | (4.14) |
(i) If a1=a2=0, by (4.14), we need to demonstrate that
m≥14μi+16νi−ε2(|f|65+|g|65)S14≥Λ−ε2(|f|65+|g|65)S14. | (4.15) |
By Lemma 2.1 and the fact that (u0,v0) satisfies (2.1), we can obtain:
∫R3|u0|α|v0|βdx=12∫R3|∇u0|2+|∇v0|2dx=(Sα,β2)32 | (4.16) |
and
S1=∫R3|∇u0|2dx=α3(Sα,β2)32,S2=∫R3|∇v0|2dx=β3(Sα,β2)32. | (4.17) |
Combing with (4.16), (4.17) and (3.5), (u′,v′) satisfies
∫R3|∇u′|2dx=172bα−241bβ42αα+24ββ4S3α,β,∫R3|∇v′|2dx=172bα41bβ−242αα4ββ+24S3α,β,∫R3(u′)α(v′)βdx=11728bα21bβ22αα2ββ2S6α,β. | (4.18) |
Consequently, we have
p(t)=120736bα21bβ22αα2ββ2S6α,β(α+β)t4−15184bα21bβ22αα2ββ2S6α,βt6=15184bα21bβ22αα2ββ2S6α,β(32t4−t6). |
Based on p′(t)=0, we can determine that t=1. Therefore, there exists
Λ=maxt>0p(t)=110368bα21bβ22αα2ββ2S6α,β. | (4.19) |
On one hand, considering
f(α)=αα2ββ2=αα2(6−α)6−α2 |
we have
f(α)min=f(3). |
Hence,
Λmin=1384bα21bβ22S6α,β. |
On the other hand, we can derive
νi6≥16×(min(b1,b2)S2α,β4)3=1384[min(b1,b2)]3S6α,β. |
Therefore, it is only when α=β=3 and b1=b2 that
m≥16νi−ε2(|f|65+|g|65)S14≥Λmin−ε2(|f|65+|g|65)S14 | (4.20) |
which contradicts Lemma 4.4 (i).
Remark 4.2. The reason for only considering the case where α=β=3 is that after extensive estimation, it is only when α=β=3 that Λ reaches its minimum value Λmin. Therefore, the estimate satisfied by m contradicts Lemma 4.4 (i). For the cases in Lemma 4.4 (ii) and (iii), obtaining the result is challenging due to the mutual interaction of (u,v) adding complexity to our computations. This will be our main task in the following work.
Moving forward, we will only consider the case where a1=a2=0, α=β=3 and b1=b2. We need to solve the lack of compactness problem from the region R3 which causes the invariance of translation.
For R>0, define
μ∞=limR→∞lim supn→∞∫|x|>R|∇un|2+|∇vn|2dx,ν∞=limR→∞lim supn→∞∫|x|>Ruαnvβndx. | (4.21) |
By concentration compactness principle, we obtain
lim supn→∞∫R3|∇un|2+|∇vn|2dx=∫R3dμ+μ∞,lim supn→∞∫R3uαnvβndx=∫R3dν+ν∞, | (4.22) |
and Sα,βμ13∞≤ν∞. Next, we estimate ν∞ and μ∞. Assume that χR∈C∞0(R3) satisfy χR∈[0,1], we have
χR(x)=0,for|x|<R2,χR(x)=1,for|x|>R |
where |∇χR|<3R. It follows from ⟨(Φ′ε(un,vn),(χRun,0)⟩→0 that
(a1+b1‖un‖2)(∫R3un∇un∇χRdx+∫R3|∇un|2χRdx)=α3∫R3uαnvβnχRdx+ε∫R3fχRundx. | (4.23) |
In this way, we can also have from ⟨(Φ′ε(un,vn),(0,χRvn)⟩→0 that
(a2+b2‖vn‖2)(∫R3vn∇vn∇χRdx+∫R3|∇vn|2χRdx)=β3∫R3uαnvβnχRdx+ε∫R3gχRvndx. | (4.24) |
By Hölder inequality, we have
B1=limR→∞lim supn→∞(a1+b1‖un‖2)|∫R3un∇un∇χRdx|≤limR→∞lim supn→∞C(∫R2≤|x|≤R|∇un|2dx)12(∫R2≤|x|≤R|∇χR|2|un|2dx)12≤limR→∞C(∫R2≤|x|≤R|∇χR|3dx)13(∫R2≤|x|≤R|u|6dx)16≤limR→∞C(∫R2≤|x|≤R|u|6dx)16=0 | (4.25) |
and
B2=limR→∞lim supn→∞(a2+b2‖vn‖2)|∫R3vn∇vn∇χRdx|=0. | (4.26) |
Combining (4.21), we have
limR→∞lim supn→∞(a1+b1‖un‖2)∫R3|∇un|2χRdx+(a2+b2‖vn‖2)∫R3|∇vn|2χRdx≥limR→∞lim supn→∞(a1∫R3|∇un|2χRdx+a2∫R3|∇vn|2χRdx)+[b1(∫R3|∇un|2χRdx)2+b2(∫R3|∇vn|2χRdx)2]≥limR→∞lim supn→∞(a1∫|x|>R|∇un|2χRdx+a2∫|x|>R|∇vn|2χRdx)+12(√b1∫|x|>R|∇un|2χRdx+√b2∫|x|>R|∇vn|2χRdx)2≥12b1S2α,βν23∞ | (4.27) |
and
(α3+β3)limR→∞lim supn→∞∫R3uαnvβnχRdx=limR→∞lim supn→∞2∫|x|≥R2uαnvβnχRdx≤limR→∞2∫|x|≥R2uαnvβndx=2ν∞. | (4.28) |
Otherwise, we get
limR→∞lim supn→∞∫R3(fχRun+gχRvn)dx=limR→∞∫R3(fχRu+gχRv)dx=0. | (4.29) |
Combining (4.23)–(4.29), we have
ν∞≥14b1S2α,βν23∞. |
We obtain one of the following two cases holds:
(1) ν∞=0;μ∞=0.
(2)
ν∞≥(b1S2α,β+√b21S4α,β8)3,μ∞≥0. |
Suppose that case (2) holds. We deduce that
m=limn→∞Φε(un,vn)−14⟨Φ′ε(un,vn),(un,vn)⟩≥a14∫R3|∇u|2dx+a24∫R3|∇v|2dx+14μ∞+16ν∞−3ε4∫R3(fu+gv)dx. |
Considering as the same as (4.15)–(4.20), we get
m≥14μ∞+16ν∞−ε2(|f|65+|g|65)S14≥Λ−ε2(|f|65+|g|65)S14 |
which is a contradiction. Thus, case (1) holds.
Combining (4.5), (4.22) with Γ=∅, we have:
lim supn→∞∫R3uαnvβndx=∫R3uαvβdx. |
Applying Fatou's lemma, we obtain:
∫R3uαvβdx≤lim infn→∞∫R3uαnvβndx≤lim supn→∞∫R3uαnvβndx=∫R3uαvβdx. |
Thus, we have
limn→∞∫R3uαnvβndx=∫R3uαvβdx. |
Set ‖un‖→d. Then, by limn→∞∫R3uαnvβndx=∫R3uαvβdx, we have
0=(Φ′ε(un,vn),(un,0))+o(1)=(a1+b1‖un‖2)‖un‖2−∫R3uαnvβndx−ε∫R3fundx+o(1)=(a1+b1d2)d2−∫R3uαvβdx−ε∫R3fudx |
and
0=(Φ′ε(un,vn),(u,0))+o(1)=(a1+b1‖un‖2)∫R3∇un∇udx−∫R3uα−1nvβnudx−ε∫R3fudx+o(1)=(a1+b1d2)‖u‖2−∫R3uαvβdx−ε∫R3fudx |
which deduces d=‖u‖. Combining un⇀u in D1,2(R3), we obtain un→u in D1,2(R3). Following the same approach and steps, we can also establish that vn→v in D1,2(R3). According to Remark 4.1 and Lemma 4.4, there exists a non-negative bounded sequence (un,vn)⊂E that satisfies
Φε(un,vn)→m<Λ,Φ′ε(un,vn)→0. |
Consequently, by un→u in D1,2(R3) and vn→v in D1,2(R3) we have
Φε(un,vn)→m=Φε(u,v),Φ′ε(un,vn)→0=Φ′ε(u,v). |
This completes the proof of the existence of the second solution.
In this paper, we first consider the existence of a local minimum solution for problem (1.1) by applying the Ekeland variational principle. Next, we investigate the existence of a second solution for problem (1.1) by applying the mountain pass lemma and the concentration compactness principle. To obtain the energy estimation of the associated functional Φε for problem (1.1), we will need the explicit form of (λ1,λ2). Therefore, when a1=a2=0 we have: Assume α=β=3 and b1=b2, there exists ε∗∗∈(0,ε∗] such that for any ε∈(0,ε∗∗), problem (1.1) has another solution. The value of ε∗ is defined in Theorem 1.2. The reason for only considering the case where α=β=3 is that after extensive estimation, it is only when α=β=3 that Λ reaches its minimum value Λmin. Therefore, the estimate satisfied by m contradicts Lemma 4.4 (i). For the cases in Lemma 4.4 (ii) and (iii), obtaining the result is challenging due to the mutual interaction of (u,v) adding complexity to our computations. This will be our main task in the following work.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
The authors would like to express their sincere gratitude to Prof. HaiYang He for her invaluable guidance throughout this research. Her expertise and insightful comments have greatly contributed to the quality of this work.
The authors declare there is no conflicts of interest.
[1] | M. F. de Almeida, L. C. F. Ferreira, L. S. M. Lima, Uniform global well-posedness of the Navier-Stokes-Coriolis system in a new critical space, Math. Z., 287 (2017), 735–750. |
[2] | P. Biler, The Cauchy problem and self-similar solutions for a nonlinear parabolic equation, Studia Math., 114 (1995), 181–205. |
[3] | P. Biler, M. Cannone, I. A. Guerra, G. Karch, Global regular and singular solutions for a model of gravitating particles, Math. Ann., 330 (2004), 693–708. |
[4] | P. Benilan, H. Brezis, M. Crandall, A semilinear equation in L1(Rn), Ann. Scuola Norm. Sup. Pisa, Ser. 4 (1975), 523–555. |
[5] | J. Berg, J. Lofstrom, Interpolation spaces, Berlin-Heidelberg-New York: Springer, 1976. |
[6] | A. Blanchet, J. Dolbeault, B. Perthame, Two-dimensional Keller-Segel model: optimal critical mass and qualitative properties of the solutions, Electron. J. Differ. Eq., 2006 (2006), 1–33. |
[7] | V. Calvez, L. Corrias, The parabolic-parabolic Keller-Segel model in R2, Commun. Math. Sci., 6 (2008), 417–447. |
[8] | K. Carrapatoso, S. Mischler, Uniqueness and long time asymptotics for the parabolic-parabolic Keller-Segel equation, Commun. Part. Diff. Eq., 42 (2017), 291–345. |
[9] | X. Chen, Well-posedness of the Keller-Segel system in Fourier-Besov-Morrey spaces, Z. Anal. Anwend., 37 (2018), 417–433. |
[10] | L. Corrias, B. Perthame, H. Zaag, Global solutions of some chemotaxis and angiogenesis systems in high space dimensions, Milan J. Math., 72 (2004), 1–28. |
[11] | M. Cannone, Harmonic analysis tools for solving the incompressible Navier-Stokes equations, In: Handbook of mathematical fluid dynamics, Amsterdam: North-Holland, 3 (2004), 161–244. |
[12] | G. E. Fernández, S. Mischler, Uniqueness and long time asymptotic for the Keller-Segel equation: the parabolic-elliptic case, Arch. Rational Mech. Anal., 220 (2016), 1159–1194. |
[13] | L. C. F. Ferreira, J. C. Precioso, Existence and asymptotic behaviour for the parabolic-parabolic Keller-Segel system with singular data, Nonlinearity, 24 (2011), 1433–1449. |
[14] | L. C. F. Ferreira, On a bilinear estimate in weak-Morrey spaces and uniqueness for Navier-Stokes equations, J. Math. Pure. Appl., 105 (2016), 228–247. |
[15] | Y. Giga, T. Miyakawa, Navier-Stokes flow in R3 with measures as initial vorticity and Morrey spaces, Commun Part. Diff. Eq., 14 (1989), 577–618. |
[16] | L. Grafakos, Classical and modern Fourier analysis, Upper Saddle River, NJ: Pearson Education, Inc., 2004. |
[17] | R. Hunt, On L(p,q) spaces, L 'Enseignement Mathématique, 12 (1966), 249–276. |
[18] | T. Iwabuchi, Global well-posedness for Keller-Segel system in Besov type spaces, J. Math. Anal. Appl., 379 (2011), 930–948. |
[19] | T. Iwabuchi, M. Nakamura, Small solutions for nonlinear heat equations, the Navier-Stokes equation, and the Keller-Segel system in Besov and Triebel-Lizorkin spaces, Adv. Differential Equ., 18 (2013), 687–736. |
[20] | T. Kato, Strong Lp-solutions of the Navier-Stokes equation in Rm, with applications to weak solutions, Math. Z., 187 (1984), 471–480. |
[21] | T. Kato, Strong solutions of the Navier-Stokes equations in Morrey spaces, Bol. Soc. Brasil Mat., 22 (1992), 127–155. |
[22] | H. Kozono, Y. Sugiyama, Y. Yahagi, Existence and uniqueness theorem on weak solutions to the parabolic-elliptic Keller-Segel system, J. Differ. Equations, 253 (2012), 2295–2313. |
[23] | H. Kozono, Y. Sugiyama, Strong solutions to the Keller-Segel system with the weak−Ln2 initial data and its application to the blow-up rate, Math. Nachr., 283 (2010), 732–751. |
[24] | H. Kozono, M. Yamazaki, Semilinear heat equations and the Navier-Stokes equation with distributions in new function spaces as initial data, Commun. Part. Diff. Eq., 19 (1994), 959–1014. |
[25] | P. G. Lemarie-Rieusset, Recent developments in the Navier-Stokes equations, Chapman and Hall, 2002. |
[26] | P. G. Lemarié-Rieusset, On some classes of time-periodic solutions for the Navier-Stokes equations in the whole space, SIAM J. Math. Anal., 47 (2015), 1022–1043. |
[27] | J.-G. Liu, J. Wang, Refined hyper-contractivity and uniqueness for the Keller-Segel equations, Appl. Math. Lett., 52 (2016), 212–219. |
[28] | Y. Meyer, Wavelets, paraproducts and Navier-Stokes equations, In: Current developments in mathematics 1996, Cambridge: International Press, 1999,105–212. |
[29] | R. O'Neil, Convolution operators and L(p,q) spaces, Duke Math. J., 30 (1963), 129–142. |
[30] | M. Yamazaki, The Navier-Stokes equations in the weak-Ln space with time-dependent external force, Math. Ann., 317 (2000), 635–675. |