数学物理学报, 2024, 44(1): 101-119

一类非局部临界椭圆方程组高能量解的多重性

付培源,, 夏阿亮,*

江西师范大学数学与统计学院 南昌 330022

Multiplicity of High Energy Solutions for a Class of Nonlocal Critical Elliptic System

Fu Peiyuan,, Xia Aliang,*

School of Mathematics and Statistics, Jiangxi Normal University, Nanchang 330022

通讯作者: 夏阿亮, E-mail:aliang_xia@jxnu.edu.cn

收稿日期: 2023-01-5   修回日期: 2023-04-13  

基金资助: 国家自然科学基金(12161044)
江西省自然科学基金(20224ACB218001)
江西省自然科学基金(20212BAB211013)
江西省自然科学基金(20224BCD41001)

Received: 2023-01-5   Revised: 2023-04-13  

Fund supported: NSFC(12161044)
Jiangxi Province Natural Science Foundation(20224ACB218001)
Jiangxi Province Natural Science Foundation(20212BAB211013)
Jiangxi Province Natural Science Foundation(20224BCD41001)

作者简介 About authors

付培源,E-mail:2398570050@qq.com

摘要

利用变分方法, 结合拓扑度理论, 该文证明了一类带有 Hardy-Littlewood-Sobolev 临界指标的椭圆方程组至少存在两个正的高能量解.

关键词: 非局部椭圆方程组; Hardy-Littlewood-Sobolev 临界指标; 变分法; 拓扑度

Abstract

By using variational methods and topological degree theory, this paper proved a class of coupled nonlocal elliptic system involving the Hardy-Littlewood-Sobolev critical exponents has at least two positive high energy solutions.

Keywords: Nonlocal elliptic system; Hardy-Littlewood-Sobolev critical exponents; Variational methods; Topological degree theory

PDF (638KB) 元数据 多维度评价 相关文章 导出 EndNote| Ris| Bibtex  收藏本文

本文引用格式

付培源, 夏阿亮. 一类非局部临界椭圆方程组高能量解的多重性[J]. 数学物理学报, 2024, 44(1): 101-119

Fu Peiyuan, Xia Aliang. Multiplicity of High Energy Solutions for a Class of Nonlocal Critical Elliptic System[J]. Acta Mathematica Scientia, 2024, 44(1): 101-119

1 引言

本文主要研究了一类带有 Hardy-Littlewood-Sobolev 临界指标的非局部椭圆方程组

$\begin{equation}\left\{\begin{array}{ll} -\Delta u+V_1(x)u=\alpha_1(|x|^{-\mu}*|u|^{2^*_{\mu}})|u|^{2^*_{\mu}-2}u+\beta(|x|^{-\mu}*|v|^{2^*_{\mu}})|u|^{2^*_{\mu}-2}u,\ & x\in \mathbb{R}^N,\\ -\Delta v+V_2(x)v=\alpha_2(|x|^{-\mu}*|v|^{2^*_{\mu}})|v|^{2^*_{\mu}-2}v+\beta(|x|^{-\mu}*|u|^{2^*_{\mu}})|v|^{2^*_{\mu}-2}v,& x\in \mathbb{R}^N \end{array} \right. \end{equation}$

高能解的存在性, 其中$*$表示卷积,$N>4$,$\alpha_{1}>0$,$\alpha_{2}>0$,$\beta>0$,$0<\mu\le4$,$2_\mu^*=(2N-\mu)/(N-2)$,$V_{i}(x)\,(i=1,2)$是位势函数. 方程组(1.1)主要来源于非线性光学[14]和双分量玻色-爱因斯坦凝聚态[6,8].

早在上世纪九十年代, Benci 和 Cerami[1]研究了带有 Sobolev 临界指数的椭圆方程

$\begin{equation}-\Delta u+a(x)u=u^{\frac{N+2}{N-2}},\quad x\in \mathbb{R}^{N}\end{equation}$

高能解的存在性, 其中$a(x)\ge0$并且$\|a\|_{L^{\frac{N}{2}}(\mathbb{R}^{N})}$足够小. 最近, Liu 等[13]将文献[1]中的结果推广到了4维非线性临界 Schrödinger 系统

$\begin{equation}\left\{\begin{array}{ll}-\Delta u+V_1(x)u=\alpha_1u^{3}+\beta uv^{2},&\quad x\in \mathbb{R}^{4},\\-\Delta v+V_2(x)v=\alpha_2v^{4}+\beta u^{2}v,&\quad x\in \mathbb{R}^{4}. \end{array} \right.\end{equation}$

在$\beta>\max\{\alpha_{1},\alpha_{2}\}\ge\min\{\alpha_{1},\alpha_{2}\}\ge0$和$\|V_{1}\|_{L^{2}(\mathbb{R}^{N})}+\|V_{2}\|_{L^{2}(\mathbb{R}^{N})}>0$且适当小的假设下, 他们证明了问题(1.3)至少存在一个正的高能量解. 随后, Gao 等[9]还将文献[13]中的结果推广到了带 Hardy-Littlewood-Sobolev 临界指数的非局部椭圆方程组(1.1). 当$N\ge5$和$\mu=4$时, 克服非局部项所带来的困难, 他们证明了(1.1)式至少存在一个正高能量解, 其中$\alpha_{1},\alpha_{2}, \beta, V_{1}(x)$和$V_{2}(x)$类似地满足$\beta>\max\{\alpha_{1},\alpha_{2}\}\ge\min\{\alpha_{1},\alpha_{2}\}\ge0$和$\|V_{1}\|_{L^{\frac{N}{2}}(\mathbb{R}^{N})}+\|V_{2}\|_{L^{\frac{N}{2}}(\mathbb{R}^{N})}>0$且适当小.

注意到上述文献均假设了位势函数$a(x)$或$V_{i}(x)\in L^{\frac{N}{2}}(\mathbb{R}^{N})\,(i=1,2)$, 也就是说位势函数$a(x)$或$V_{i}(x)$$(i=1,2)$可能在无穷远处消失. 很自然地, 我们会想知道如果位势函数在无穷远处不会消失, 能否得到类似结果.

当位势函数$V(x)$满足$\lim\limits_{|x|\to+\infty}V(x)=V_{\infty}>0$,$V(x)\ge V_{\infty}$和$(V-V_{\infty})\in L^{\frac{3}{2}}(\mathbb{R}^N)$时, 最近 Cerami 和 Molle[2]证明了 3 维 Schrödinger-Poisson 方程

$\begin{equation*}\left\{\begin{array}{ll}-\Delta u+V(x)u+K(x)\phi u=u^{5},&\quad x\in \mathbb{R}^{3},\\-\Delta \phi=K(x)u^{2},&\quad x\in \mathbb{R}^{3} \end{array} \right.\end{equation*}$

高能量解的存在性. 同时利用拓扑度理论, 他们还得到了该方程高能量解的多重性. 注意此时$V(x)\ge V_{\infty}>0$不会在无穷远处消失并且$V(x)\not\in L^{\frac{3}{2}}(\mathbb{R}^N)$. Guo 和 Li[11]还将文献[2]中的结果推广到了方程(1.2)的分数阶情形 (即将$-\Delta$替换成分数阶拉普拉斯算子$(-\Delta)^{s}\,(0<s<1)$).

受到上述文章的启发, 本文将利用文献[2]中的方法和技巧, 在文献[9]的基础上, 继续讨论方程组(1.1)高能量解的存在性和多重性. 具体地来说, 我们假设$N>4$,$0<\mu\le4$, 以及位势函数$V_i(x)\,(i=1,2)$满足

(V1)$\lim\limits_{|x|\to+\infty}V_i(x)=V_{i\infty}>0$;

(V2)$V_i(x)\ge V_{i\infty}$,$V_i(x)\not\equiv V_{i\infty}$,$\forall \,\,x\in \mathbb{R}^N$;

(V3)$(V_i-V_{i\infty})\in L^{\frac{N}{2}}(\mathbb{R}^N)$.

本文主要结果如下.

定理 1.1 假设$N>4$,$0<\mu\le4$以及$0<\beta\le\sqrt{\alpha_1\alpha_2}$. 如果条件 (V1)-(V3) 成立, 则

(i) 若$\|V_i-V_{i\infty}\|_{L^\frac{N}{2}(\mathbb{R}^N)}\not=0$,$i=1,2$, 则存在正常数$\tilde{V}_i$使得当$V_{i\infty}\in(0,\tilde{V}_i)$,$i=1,2$时, 方程(1.1)至少存在一个正的高能量解.

(ii) 若

$\begin{equation}0<\|V_i-V_{i\infty}\|_{L^\frac{N}{2}(\mathbb{R}^N)}<\left(2^{\frac{N-\mu+2}{2N-\mu}}-1\right)S,\quad i=1,2, \end{equation}$

其中$S$是 Sobolev 嵌入$D^{1,2}(\mathbb{R}^N)\hookrightarrow L^{2^*}(\mathbb{R}^N)\,(2^{*}=2N/(N-2))$的最佳常数,则存在正常数$\tilde{V}_i^{*}$使得当$V_{i\infty}\in(0,\tilde{V}_i^{*})$,$i=1,2$时, 方程(1.1)至少存在两个正的高能量解.

由假设 (V1), 容易验证$V_{i}(x)\not\in L^{\frac{N}{2}}(\mathbb{R}^{N})$, 所以本文与文献[9]中的假设完全不同. 此外, 在定理 1.1 中, 本文还可以得到方程组(1.1)高能量解的多重性, 所以除了借鉴文献[9]中的变分技巧外, 我们还需利用形变引理 (见文献[16, 引理 2.3]) 和拓扑度理论. 最后, 在利用变分法时, 本文的主要困难是紧性缺失 ($V_{i}(x)>0\,(i=1,2)$), 所以我们还需证明方程组(1.1)的全局紧性结果.

在本文中我们用$\|(\cdot,\cdot)\|$表示$\mathcal{H}$空间中的范数, 即

$\|(u,v)\|:=\left(\int_{\mathbb{R}^N}\left(|\nabla u|^2+|\nabla v|^2+V_{1\infty}u^2+V_{2\infty}v^2\right){\rm d}x\right)^{\frac{1}{2}},\ \mbox{当}\,\mathcal{H}=H^1(\mathbb{R}^N)\times H^1(\mathbb{R}^N)\,\mbox{时}$,

$\|(u,v)\|:=\left(\int_{\mathbb{R}^N}\left(|\nabla u|^2+|\nabla v|^2\right){\rm d}x\right)^{\frac{1}{2}},\ \mbox{当}\,\mathcal{H}=D^{1,2}(\mathbb{R}^N)\times D^{1,2}(\mathbb{R}^N)\,\mbox{时}.$

2 预备知识

我们首先回顾 Hardy-Littlewood-Sobolev 不等式.

命题 2.1 (Hardy-Littlewood-Sobolev 不等式) 假设$p,q>1$,$0<\mu<N$且满足$\frac{1}{p}+\frac{\mu}{N}+\frac{1}{q}=2$, 则存在一个常数$C(N,\mu,p)$, 使得对$f\in L^p(\mathbb{R}^N)$,$g\in L^q(\mathbb{R}^N)$, 都有

$\left|\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{f(x)g(y)}{|x-y|^\mu}{\rm d}x{\rm d}y\right|\le C(N,\mu,p)\|f\|_{L^p(\mathbb{R}^N)}\|g\|_{L^q(\mathbb{R}^N)}.$

由命题 2.1, 我们可以定义

$S_{H,L}:=\inf_{u\in D^{1,2}(\mathbb{R}^N)\setminus{\{0\}}}\frac{\int_{\mathbb{R}^N}|\nabla u|^2{\rm d}x}{\left(\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|u(x)|^{2_\mu^*}|u(y)|^{2_\mu^*}}{|x-y|^\mu}\,{\rm d}x{\rm d}y\right)^\frac{1}{2_\mu^*}},$

并且它是可达的 (见文献[7]) 当且仅当

$u(x)=C\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{N-2}{2}},$

其中$C>0$是常数,$a\in \mathbb{R}^N$和$b\in(0,\infty)$都为参数. 此外, 存在仅依赖于$N$和$\mu$的常数$\mathcal{C}(N,\mu)$使得

$S_{H,L}=\mathcal{C}(N,\mu)^{-\frac{1}{2_\mu^*}}S.$

令$u^+=\max\{u,0\}$,$v^+=\max\{v,0\}$, 我们考虑下面的修正问题

$\begin{equation} \left\{\begin{array}{ll} -\Delta u+V_1(x)u=\alpha_1(|x|^{-\mu}*|u^+|^{2^*_{\mu}})|u^+|^{2^*_{\mu}-1}+\beta(|x|^{-\mu}*|v^+|^{2^*_{\mu}})|u^+|^{2^*_{\mu}-1},& x\in \mathbb{R}^N,\\ -\Delta v+V_2(x)v=\alpha_2(|x|^{-\mu}*|v^+|^{2^*_{\mu}})|v^+|^{2^*_{\mu}-1}+\beta(|x|^{-\mu}*|u^+|^{2^*_{\mu}})|v^+|^{2^*_{\mu}-1},& x\in \mathbb{R}^N. \end{array} \right.\end{equation}$

注意到, 若$(u,v)$是方程(2.5)的非平凡解, 那么由强极值原理可得在$\mathbb{R}^N$中,$u>0$,$v>0$, 即$(u,v)$是方程(1.1)的正解. 因此, 下面我们只需要证明问题 (2.5)具有非平凡解.

我们先讨论标量方程

$\begin{equation} -\Delta u+V_i(x)u=\alpha_i\big(|x|^{-\mu}*|u^+|^{2^*_{\mu}}\big)|u^+|^{2^*_{\mu}-1}\quad x\in \mathbb{R}^N,\end{equation}$

以及所对应的能量泛函$\mathcal{I}_i:H^{1}(\mathbb{R}^N)\to \mathbb{R}$,

$\mathcal{I}_i(u)=\frac{1}{2}\int_{\mathbb{R}^N}\left(|\nabla u|^2+V_i(x)u^2\right){\rm d}x-\frac{\alpha_i}{2\cdot 2^*_\mu}\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|u^+(x)|^{2^*_\mu}|u^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y.$

我们知道, 方程(2.6)的解包含在 Nehari 流形$\mathcal{M}_i=\left\{u\in H^{1}(\mathbb{R}^N)\setminus\{0\}:\langle\mathcal{I}_i^\prime(u),u\rangle=0\right\}$中. 考虑极值问题

$c_i=\inf_{u\in\mathcal{M}_i}\mathcal{I}_i(u).$

若$V_i(x)\equiv0$, 我们分别用$\mathcal{I}_i^\infty,\,c_i^\infty,\mathcal{M}_i^\infty$来替代$\mathcal{I}_i,\,c_i,\mathcal{M}_i$.

在文献[7,定理 1.3]中证明了方程

$-\Delta u=\left(|x|^{-\mu}*u^{2_\mu^*}\right)u^{2_\mu^*-1},\quad x\in\mathbb{R}^N$

的正解一定形如

$\begin{equation}U_{a,b}(x)=C_{N,\mu}\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{N-2}{2}},\end{equation}$

其中$b>0$,$a\in \mathbb{R}^N$, 正常数$C_{N,\mu}$只依赖于$N$和$\mu$. 所以, 直接计算可得

$\begin{equation}\int_{\mathbb{R}^N}|\nabla U_{a,b}|^2{\rm d}x=\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|U_{a,b}(x)|^{2_\mu^*}|U_{a,b}(y)|^{2_\mu^*}}{|x-y|^\mu}\,{\rm d}x{\rm d}y=S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\end{equation}$

而且可以验证

$c_{i}^\infty=\frac{N-\mu+2}{2(2N-\mu)}\alpha_i^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}.$

下面我们定义

$V_i(x)=(V_i(x)-V_{i\infty})+V_{i\infty}:=W_i(x)+V_{i\infty},$

其中$i=1,2$以及$x\in \mathbb{R}^N$. 由 (V3), 可知$W_{i}(x)\in L^{\frac{N}{2}}( \mathbb{R}^N)$,$i=1,2$.

引理 2.1 假设$V_i(x)\,(i=1,2)$满足 (V1)-(V3), 则$c_i=c_i^\infty=\frac{N-\mu+2}{2(2N-\mu)}\alpha_i^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\,i=1,2.$

取$u\in\mathcal{M}_i$, 则存在$t_u>0$使得$t_uu\in\mathcal{m}_i^\infty$. 由 (V1) 和 (V2), 我们有$V_i(x)\ge0$,$x\in \mathbb{R}^N$, 所以

$\begin{align*} t_u^{2\left(2_\mu^*-1\right)}=\frac{\int_{\mathbb{R}^N}|\nabla u|^2{\rm d}x}{\alpha_i \int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|u^+(x)|^{2^*_\mu}|u^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y}\le\frac{\int_{\mathbb{R}^N}\left(|\nabla u|^2+V_i(x)u^2\right){\rm d}x}{\alpha_i \int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|u^+(x)|^{2^*_\mu}|u^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y}=1, \end{align*}$

并且

$\begin{align*} c_i^\infty\le\mathcal{I}_i^\infty(t_uu)&=\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)t_u^2\int_{\mathbb{R}^N}|\nabla u|^2{\rm d}x\\&\le \left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\int_{\mathbb{R}^N}\left(|\nabla u|^2+V_i(x)u^2\right){\rm d}x =\mathcal{I}_i(u).\end{align*}$

因此, 可得$c_i^\infty\le c_i$.

另一方面, 定义$t_n>0$使得

$t_n^{2(2_\mu^*-1)}=\frac{\int_{\mathbb{R}^N}\left(|\nabla U_{0,b_n}|^2+V_i(x)U_{0,b_n}^2\right){\rm d}x}{\alpha_i \int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|U_{0,b_n}(x)|^{2^*_\mu}|U_{0,b_n}(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y},$

其中当$n\to+\infty$时,$b_n \to 0$. 由$U_{a,b}$的定义, 可以验证$t_nU_{0,b_n}\in\mathcal{M}_i$. 所以

$c_i\le \mathcal{I}_i(t_nU_{0,b_n})=\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)t_n^2\int_{\mathbb{R}^N}\left(|\nabla U_{0,b_n}|^2+V_i(x)U_{0,b_n}^2\right){\rm d}x.$

回顾$V_i(x)=V_{i\infty}+W_i(x)$,$W_i(x)\in L^{\frac{N}{2}}(\mathbb{R}^N)$,$i=1,2$. 我们断言

$\begin{equation}\int_{\mathbb{R}^N}V_i(x)U_{0,b_n}^2{\rm d}x=\int_{\mathbb{R}^N}V_{i\infty}U_{0,b_n}^2{\rm d}x+\int_{\mathbb{R}^N}W_i(x)U_{0,b_n}^2{\rm d}x=o_n(1),\quad i=1,2.\end{equation}$

事实上, 首先通过 Hölder 不等式, 我们有

$\begin{matrix}\int_{\mathbb{R}^N}W_i(x)U_{0,b_n}^2{\rm d}x&=\int_{B_\rho(0)}W_i(x)U_{0,b_n}^2{\rm d}x+\int_{\mathbb{R}^N\setminus B_\rho(0)}W_i(x)U_{0,b_n}^2{\rm d}x \\&\le\|W_{i}\|_{L^{\frac{N}{2}}(B_\rho(0))}\|U_{0,b_n}\|^2_{L^{2^*}(B_\rho(0))}\!+\!\|W_{i}\|_{L^{\frac{N}{2}}(\mathbb{R}^N\setminus B_\rho(0))}\|U_{0,b_n}\|^2_{L^{2^*}(\mathbb{R}^N\setminus B_\rho(0))}.\end{matrix}$

由于$\lim\limits_{n\to+\infty}\|U_{0,b_n}\|_{L^{2^*}}=\|U_{0,1}\|_{L^{2^*}}=\mbox{常数},$则对任意的$\rho>0$,

$\begin{equation}\lim\limits_{n\to+\infty}\|W_{i}\|_{L^{\frac{N}{2}}(B_\rho(0))}\|U_{0,b_n}\|^2_{L^{2^*}(B_\rho(0))}\le C\|W_{i}\|_{L^{\frac{N}{2}}(B_\rho(0))}.\end{equation}$

再令$\rho\to$, 由(2.11)式可得

$\begin{equation}\lim\limits_{n\to+\infty}\|W_{i}\|_{L^{\frac{N}{2}}(B_\rho(0))}\|U_{0,b_n}\|^2_{L^{2^*}(B_\rho(0))}=0.\end{equation}$

通过直接计算, 可得

$\begin{matrix} \lim\limits_{n\to+\infty}\|U_{0,b_n}\|^{2^*}_{L^{2^*}(\mathbb{R}^N\setminus B_\rho(0))}&=\lim\limits_{n\to+\infty}C_{N,\mu}^{2^*}\int_{\mathbb{R}^N\setminus B _\rho(0)}\frac{b_n^N}{\left(b_n^2+|x|^2\right)^N}{\rm d}x \\ &\le \lim\limits_{n\to+\infty}C_{N,\mu}^{2^*}b_n^N\omega_N\int_\rho^\infty r^{-(N+1)}{\rm d}r \\&\le\lim\limits_{n\to+\infty}C\rho^{-N}b_n^N=0,\end{matrix}$

其中$\omega_n$是$\mathbb{R}^N$中单位球的表面积. 所以, 由(2.10), (2.12) 和 (2.13) 式, 我们有

$\begin{equation}\lim\limits_{n\to+\infty}\int_{\mathbb{R}^N}W_i(x)U_{0,b_n}^2{\rm d}x=0.\end{equation}$

另一方面, 因为$V_{i\infty}>0\,(i=1,2)$和$N>4$, 所以有

$\begin{matrix}\int_{\mathbb{R}^N}V_{i\infty}U_{0,b_n}^2{\rm d}x&=\int_{\mathbb{R}^N}V_{i\infty}\left[b_n^{-\frac{N-2}{2}}U_{0,1}\left(\frac{x}{b_n}\right)\right]^2{\rm d}x \\&=b_n^2\int_{\mathbb{R}^N}V_{i\infty}U_{0,1}^2(x){\rm d}x,\quad i=1,2\end{matrix}$

$\begin{matrix}\int_{\mathbb{R}^N}U_{0,1}^2(x){\rm d}x&=C_{N,\mu}^2\int_{\mathbb{R}^N}\frac{1}{(1+|x|^2)^{N-2}}{\rm d}x \\ &\le C\int_{B_{r_0}(0)}\frac{1}{(1+|x|^2)^{N-2}}{\rm d}x+C\int_{\mathbb{R}^N\setminus B_{r_0}(0)}\frac{1}{(1+|x|^2)^{N-2}}{\rm d}x\\ &\le C_1+C\int_{\mathbb{R}^N\setminus B_{r_0}(0)}\frac{1}{|x|^{2N-4}}{\rm d}x=C_1+C\int_{r_0}^{+\infty}\frac{1}{r^{N-3}}{\rm d}r<C_2.\end{matrix}$

结合(2.15)和(2.16)式, 可得

$\begin{equation}\lim\limits_{n\to+\infty}\int_{\mathbb{R}^N}V_{i\infty}U_{0,b_n}^2{\rm d}x=0.\end{equation}$

所以, 由(2.14)和(2.17)式, 可知断言(2.9)式成立.

根据$t_n$的定义, 我们知道当$n\to+\infty$时,$t^2_n=\alpha_i^{-\frac{1}{2^*_\mu-1}}+o_n(1).$所以

$\begin{align*}c_i\le \mathcal{I}_i(t_nU_{0,b_n})&=\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\left(\alpha_i^{-\frac{1}{2^*_\mu-1}}+o_n(1)\right)\left(S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}+o_n(1)\right)\\&=\frac{N-\mu+2}{2(2N-\mu)}\alpha_i^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}+o_n(1),\end{align*}$

因此$c_i\le\frac{N-\mu+2}{2(2N-\mu)}\alpha_i^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}=c_i^\infty,\,i=1,2$.证毕.

为了利用变分方法去研究方程组(2.5), 考虑它所对应的能量泛函$\Phi:\mathcal{H} \to \mathbb{R}$,

$\begin{align*} &\Phi(u,v)=\frac{1}{2}\int_{\mathbb{R}^N}\left(|\nabla u|^2+|\nabla v|^2+V_1(x)u^2+V_2(x)v^2\right){\rm d}x\\ &-\frac{1}{2\cdot 2^*_\mu}\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{\alpha_1|u^+(x)|^{2^*_\mu}|u^+(y)|^{2^*_\mu}+\alpha_2|v^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}+2\beta|u^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y.\end{align*}$

由 Hardy-Littlewood-Sobolev 不等式 (见命题 2.1), 可知泛函$\Phi\in C^1(\mathcal{H},\mathbb{R})$. 此外, 我们知道$(u,v)$为$\Phi$的临界点当且仅当$(u,v)$是方程组(2.5) 的弱解.

类似地, 研究极值问题

$c=\inf_{(u,v)\in\mathcal{N}}\Phi(u,v),$

其中$\mathcal{N}=\{(u,v)\in\mathcal{H}\setminus\{(0,0)\}:\,\langle\Phi^\prime(u,v),(u,v)\rangle=0\}.$当$V_1\equiv V_2\equiv0$时, 同样地把$\Phi,\,c,\,\mathcal{N}$分别记为$\Phi^\infty,\,c^\infty,\,\mathcal{N}^\infty$.注意到泛函$\Phi^\infty$的临界点与下列耦合系统

$\begin{equation} \left\{\begin{array}{ll} -\Delta u=\alpha_1(|x|^{-\mu}*|u^+|^{2^*_{\mu}})|u^+|^{2^*_{\mu}-1}+\beta(|x|^{-\mu}*|v^+|^{2^*_{\mu}})|u^+|^{2^*_{\mu}-1},& x\in \mathbb{R}^N,\\ -\Delta v=\alpha_2(|x|^{-\mu}*|v^+|^{2^*_{\mu}})|v^+|^{2^*_{\mu}-1}+\beta(|x|^{-\mu}*|u^+|^{2^*_{\mu}})|v^+|^{2^*_{\mu}-1},& x\in \mathbb{R}^N \end{array} \right.\end{equation}$

的弱解有关. 参考文献[15](或文献[4]), 定义集合

$\widetilde{\mathcal{N}}=\{(u,v)\in\mathcal{H}:\,u\not\equiv0,\,v\not\equiv0,\,\langle(\Phi^\infty)^\prime(u,v),(u,0)\rangle=\langle(\Phi^\infty)^\prime(u,v),(0,v)\rangle=0\}\not=\emptyset,$

则方程(2.18)的任何非平凡解都属于$\widetilde{\mathcal{N}}$. 同理, 考虑极值问题

$\widetilde{c^\infty}=\inf_{(u,v)\in\widetilde{\mathcal{N}}}\Phi^\infty(u,v).$

另一方面, 如果$(kU_{a,b},lU_{a,b})$是方程(2.18)的正解, 则$(k,l)\in \mathbb{R}^{2}$满足非线性问题

$\begin{equation} \left\{\begin{array}{ll} \alpha_1k^{2\cdot2^*_\mu-2}+\beta k^{2^*_\mu-2}l^{2_\mu^*}=1,\\ \alpha_2l^{2\cdot2^*_\mu-2}+\beta k^{2^*_\mu}l^{2^*_\mu-2}=1,\\ k>0,\quad l>0. \end{array} \right.\end{equation}$

引理 2.2 如果$0<\beta\le\sqrt{\alpha_1\alpha_2}$, 则(2.19)式有唯一的正解$(k_0,l_0)$. 此外,

$c^\infty=\widetilde{c^\infty}=\Phi^\infty(k_0U_{a,b},l_0U_{a,b})=\frac{N-\mu+2}{2(2N-\mu)}\left(k_0^2+l_0^2\right)S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},$

并且方程(2.18)的最小能量解都可以表示为$\left(k_0U_{a,b},l_0U_{a,b}\right),$其中$a\in\mathbb{R}^N$,$b>0$.

引理的证明过程与文献[15,定理1.4-(ii)] 类似, 所以我们这里省略. 还可以见文献[4].

引理 2.3 假设$0<\beta\le\sqrt{\alpha_1\alpha_2}$,$V_1(x)$和$V_2(x)$满足 (V1)-(V3), 则$c=c^\infty$并且$c$是不可达的.

类似于引理2.1的证明, 我们可证$c\ge c^\infty=\frac{N-\mu+2}{2(2N-\mu)}\left(k_0^2+l_0^2\right)S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}$, 其中$(k_0,l_0)$由引理2.2中给出, 以及

$\begin{equation} \lim\limits_{n\to+\infty}\int_{\mathbb{R}^N}k_0^2V_{1}(x)U_{0,b_n}^2+l_0^2V_{2}(x)U_{0,b_n}^2{\rm d}x=0,\end{equation}$

其中当$n\to+\infty$时,$b_n\to 0$.

定义

$t_n^{2\cdot 2_\mu^*-2}=\frac{\int_{\mathbb{R}^N}\left[k_0^2\left(|\nabla U_{0,b_n}|^2+V_1(x)U_{0,b_n}^2\right)+l_0^2\left(|\nabla U_{0,b_n}|^2+V_2(x)U_{0,b_n}^2\right)\right]{\rm d}x}{(k_0^2+l_0^2)\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{U_{0,b_n}^{2_\mu^*}(x)U_{0,b_n}^{2_\mu^*}(y)}{|x-y|^\mu}\,{\rm d}x{\rm d}y},$

则由(2.19)式可知$\left(t_nk_0U_{0,b_n},t_nl_0U_{0,b_n}\right)\in\mathcal{N}$以及当$n\to+\infty$时,$t_n^{2\cdot 2_\mu^*-2}=1+o_n(1)$. 因此,

$\begin{align*}c&\le \Phi\left(t_nk_0U_{0,b_n},t_nl_0U_{0,b_n}\right)\\& =\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)t_n^2\int_{\mathbb{R}^N}\left[k_0^2\left(|\nabla U_{0,b_n}|^2+V_1(x)U_{0,b_n}^2\right)+l_0^2\left(|\nabla U_{0,b_n}|^2+V_2(x)U_{0,b_n}^2\right)\right]{\rm d}x\\&=\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\left(1+o_n(1)\right)\left((k_0^2+l_0^2)S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}+o_n(1)\right)\\&=\frac{N-\mu+2}{2(2N-\mu)}(k_0^2+l_0^2)S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}+o_n(1),\end{align*}$

即$c\le\frac{N-\mu+2}{2(2N-\mu)}(k_0^2+l_0^2)S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}=c^\infty$. 所以有$c=c^\infty$.

下面证明$c$不可达. 反之, 假设$(u,v)\in\mathcal{N}$满足

$\Phi(u,v)=\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\int_{\mathbb{R}^N}\left[\left(|\nabla u|^2+V_1(x)u^2\right)+\left(|\nabla v|^2+V_2(x)v^2\right)\right]{\rm d}x=c.$

取$t^*>0$使得$(t^*u,t^*v)\in \mathcal{N}^\infty$, 则有$t^*\le 1$(见引理 2.1 的证明). 又因为$V_i(x)\ge V_{i\infty}>0$,$i=1,2$, 则有

$\begin{align*}c^\infty&\le \Phi^\infty(t^*u,t^*v)=\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)(t^*)^2\int_{\mathbb{R}^N}|\nabla u|^2+|\nabla v|^2{\rm d}x\\&<\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\int_{\mathbb{R}^N}\left[\left(|\nabla u|^2+V_1(x)u^2\right)+\left(|\nabla v|^2+V_2(x)v^2\right)\right]{\rm d}x\\&=c=c^\infty,\end{align*}$

矛盾. 所以,$c$不可达. 证毕.

上述引理表明极小值$c$是不可达的, 即方程(1.1)没有基态解, 所以方程(1.1)的正解只有在高能量级下才有可能存在.

下面我们还需对临界耦合

$\begin{equation} \left\{\begin{array}{ll} -\Delta u=\alpha_1(|x|^{-\mu}*u^{2^*_{\mu}})u^{2^*_{\mu}-1}+\beta(|x|^{-\mu}*v^{2^*_{\mu}})u^{2^*_{\mu}-1},&\quad \,\,x\in \mathbb{R}^N,\\ -\Delta v=\alpha_2(|x|^{-\mu}*v^{2^*_{\mu}})v^{2^*_{\mu}-1}+\beta(|x|^{-\mu}*u^{2^*_{\mu}})v^{2^*_{\mu}-1},&\quad \,\,x\in \mathbb{R}^N \end{array} \right.\end{equation}$

的正解进行分类.

定义$R_N=\frac{\Gamma\left(\frac{N-2}{2}\right)}{4\pi^{\frac{N}{2}}}$以及$I(s)=\frac{\pi^\frac{N}{2}\Gamma\left(\frac{N-2s}{2}\right)}{\Gamma\left(N-s\right)},$其中$0<s<\frac{N}{2}$,$\Gamma(\cdot)$是 Gamma 函数,则问题 (2.21) 等价于积分方程组 (见文献[3])

$\begin{equation} \left\{\begin{array}{ll} u(x)=\alpha_1R_N\int_{\mathbb{R}^N}\frac{u^{2_\mu^*-1}(y)f(y)}{|x-y|^{N-2}}{\rm d}y+\beta R_N\int_{\mathbb{R}^N}\frac{u^{2_\mu^*-1}(y)g(y)}{|x-y|^{N-2}}{\rm d}y,\\[3mm] v(x)=\alpha_2R_N\int_{\mathbb{R}^N}\frac{v^{2_\mu^*-1}(y)g(y)}{|x-y|^{N-2}}{\rm d}y+\beta R_N\int_{\mathbb{R}^N}\frac{v^{2_\mu^*-1}(y)f(y)}{|x-y|^{N-2}}{\rm d}y,\\[3mm] f(y)=\int_{\mathbb{R}^N}\frac{u^{2_\mu^*}(y)}{|x-y|^\mu}{\rm d}y,\\[3mm] g(y)=\int_{\mathbb{R}^N}\frac{v^{2_\mu^*}(y)}{|x-y|^\mu}{\rm d}y. \end{array} \right.\end{equation}$

命题 2.2 假设$0<\beta\le\sqrt{\alpha_1\alpha_2}$,$(u,v)\in\mathcal{H}$是方程组 (2.21)的正经典解, 则有

$u(x)=C_1\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{N-2}{2}}, \quad v(x)=C_2\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{N-2}{2}},$

其中$a\in\mathbb{R}^N$,$b>0$,$C_1=\frac{k_0}{\left(R_NI\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)\right)^{\frac{N-2}{2(N-\mu+2)}}}$,$C_2=\frac{l_0}{\left(R_NI\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)\right)^{\frac{N-2}{2(N-\mu+2)}}},$$(k_0,l_0)$由引理 2.2 给出.

利用移动球面, 类似于文献[9]($\mu=4$以及$N\ge5$情形) 或文献[7], 可以证明$(u,v)$具有如下形式

$\begin{equation}u(x)=C_1\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{N-2}{2}}\quad \mbox{和}\quad v(x)=C_2\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{N-2}{2}},\end{equation}$

其中$a\in \mathbb{R}^N$,$C_1,\,C_2,\,b>0$.

由恒等式 (见文献[5,(37)式])

$\begin{equation}\int_{\mathbb{R}^N}\frac{1}{|x-y|^{2s}}\left(\frac{1}{1+|y|^2}\right)^{N-s}{\rm d}y=I(s)\left(\frac{1}{1+|x|^2}\right)^s,\quad \,0<s<\frac{N}{2},\end{equation}$

再将(2.23)式代入(2.22)式中, 则有

$f(y)=\int_{\mathbb{R}^N}\frac{u^{2_\mu^*}(y)}{|x-y|^\mu}{\rm d}y=C_1^{2_\mu^*}I\left(\frac{\mu}{2}\right)\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{\mu}{2}},$
$g(y)=\int_{\mathbb{R}^N}\frac{v^{2_\mu^*}(y)}{|x-y|^\mu}{\rm d}y=C_2^{2_\mu^*}I\left(\frac{\mu}{2}\right)\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{\mu}{2}}$

$\begin{align*} u(x)&=\alpha_1R_N\int_{\mathbb{R}^N}\frac{u^{2_\mu^*-1}(y)f(y)}{|x-y|^\mu}{\rm d}y+\beta R_N\int_{\mathbb{R}^N}\frac{u^{2_\mu^*-1}(y)g(y)}{|x-y|^\mu}{\rm d}y\\ &=\alpha_1R_NC_1^{2\cdot 2_\mu^*-1}I\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{N-2}{2}}\\ &\qquad+\beta R_NC_1^{2_\mu^*-1}C_2^{2_\mu^*}I\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)\left(\frac{b}{b^2+|x-a|^2}\right)^{\frac{N-2}{2}}.\end{align*}$

再由(2.23)式可得$\alpha_1R_NC_1^{2\cdot 2_\mu^*-2}I\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)+\beta R_NC_1^{2_\mu^*-2}C_2^{2_\mu^*}I\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)=1.$类似地, 也有$\alpha_2R_NC_2^{2\cdot 2_\mu^*-2}I\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)+\beta R_NC_1^{2_\mu^*}C_2^{2_\mu^*-1}I\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)=1.$最后, 通过引理2.2, 直接计算可得$C_1=\frac{k_0}{\left(R_NI\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)\right)^{\frac{N-2}{2(N-\mu+2)}}}$,$C_2=\frac{l_0}{\left(R_NI\left(\frac{\mu}{2}\right)I\left(\frac{N-2}{2}\right)\right)^{\frac{N-2}{2(N-\mu+2)}}}.$

由命题 2.2 和引理 2.2, 可得

推论2.1 假设$0<\beta\le\sqrt{\alpha_1\alpha_2}$,$(u,v)\in\mathcal{H}$是方程(2.18)的正经典解, 则有

$(u,v)=(k_0U_{a,b},\,l_0U_{a,b}),$

其中$a\in\mathbb{R}^N$,$b>0$.此外, 方程(2.18)的每一个非平凡正的经典解$(u,v)\in\mathcal{H}$都是基态解.

3 全局集中紧和一些估计

由于本文讨论的问题紧性缺失, 所以通过全局紧性结果来对泛函$\Phi$的 Palais-Smale 序列进行描述. 首先需要下面的 Brezis-Lieb 型引理.

引理 3.1 假设$N\ge3$,$0<\mu<N$. 若$\{(u_n,v_n)\}$是$L^{\frac{2N}{N-2}}(\mathbb{R}^N)\times L^{\frac{2N}{N-2}}(\mathbb{R}^N)$中的有界序列, 且当$n\to+\infty$时,$(u_n,v_n)\to(u,v)$a.e., 则当$n\to +\infty$时, 有下面式子成立

$\begin{align*}& \int_{\mathbb{R}^N}\left(|x|^{-\mu}*|u_n^+|^{2_\mu^*}\right)|u_n^+|^{2_\mu^*}{\rm d}x-\int_{\mathbb{R}^N}\left(|x|^{-\mu}*|(u_n-u)^+|^{2_\mu^*}\right)|(u_n-u)^+|^{2_\mu^*}{\rm d}x\\&\qquad\to\int_{\mathbb{R}^N}\left(|x|^{-\mu}*|u^+|^{2_\mu^*}\right)|u^+|^{2_\mu^*}{\rm d}x{,}\\&\int_{\mathbb{R}^N}\left(|x|^{-\mu}*|u_n^+|^{2_\mu^*}\right)|v_n^+|^{2_\mu^*}{\rm d}x-\int_{\mathbb{R}^N}\left(|x|^{-\mu}*|(u_n-u)^+|^{2_\mu^*}\right)|(v_n-v)^+|^{2_\mu^*}{\rm d}x\\&\qquad\to\int_{\mathbb{R}^N}\left(|x|^{-\mu}*|u^+|^{2_\mu^*}\right)|v^+|^{2_\mu^*}{\rm d}x{.}\end{align*}$

类似于文献[10,引理2.2](或文献[17])和文献[9,引理4.1]的证明, 通过 Hardy-Littlewood-Sobolev 不等式(见命题2.1), 容易得到引理 3.1 中的结论, 所以证明省略.

命题 3.1 设$V_{1}(x)$和$V_{2}(x)$满足 (V1)-(V3),$\{(u_n,v_n)\}$是泛函$\Phi$的一个$(PS)_d$序列,$(u_0,v_0)$是方程(2.5)的解, 并且在$\mathcal{H}$中$(u_n,v_n)$弱收敛于$(u_0,v_0)$,$n \to+\infty$, 则$\{(u_n,v_n)\}$(或$\{(u_n,v_n)\}$的子列) 满足

(a) 在$\mathcal{H}$中,$(u_n,v_n)\rightarrow(u_0,v_0)$或

(b) 存在$k\in\mathbb{N}$和方程(2.18)的非零解$(u_1,v_1), (u_2,v_2),\cdots (u_k,v_k)$, 使得当$n\to+\infty$时, 有

$\|(u_n,v_n)\|^2\to\|(u_0,v_0)\|^2+\sum_{i=1}^k\|(u_j,v_j)\|^2, \Phi(u_n,v_n)\to\Phi(u_0,v_0)+\sum_{i=1}^k\Phi^\infty(u_j,v_j).$

在文献[9,命题4.2] 中, 作者证明了当$V_i\in L^{\frac{N}{2}}(\mathbb{R}^N)$($i=1,2$) 和$\mu=4$时, 命题 3.1 中结论成立. 当$0<\mu\le 4$和$V_{i}(x)\,(i=1,2)$满足 (V1)-(V3) 时, 利用引理3.1以及文献[2,命题2.9] (或文献[11,引理2.3]) 中的一些技巧和想法, 也容易验证命题 3.1 中结论成立. 证明略.

根据文献[15,引理4.6](或文献[4,引理2.3]) 的证明过程, 我们知道如下事实

$c^\infty=c<\min\left\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}\right\},$

则我们可证明

推论 3.1 设$V_{1}(x)$和$V_{2}(x)$满足 (V1)-(V3),$\{(u_n,v_n)\}\subset \mathcal{N}$是约束泛函$\Phi|_{\mathcal{N}}$的$(PS)_d$序列, 其中

$d\in\left(c^\infty,\min\left\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\right\}\right),$

则$\{(u_n,v_n)\}$在$\mathcal{H}$中是相对紧的.

由命题 3.1 可知, 我们只需要排除命题 3.1 中的情形 (b). 假设命题 3.1 中的 (b) 成立, 则存在$k\in\mathbb{N}$, 方程(2.5)的解$(u_0,v_0)$和 (2.18)的非零解$(u_1,v_1), (u_2,v_2),\cdots (u_k,v_k)$,$\{(u_n,v_n)\}$(或$\{(u_n,v_n)\}$的子列) 满足

$\begin{align*}&\lim\limits_{n\to+\infty}\|(u_n,v_n)\|^2=\|(u_0,v_0)\|^2+\sum_{i=1}^k\|(u_j,v_j)\|^2,\\&d=\lim\limits_{n\to+\infty}\Phi(u_n,v_n)=\Phi(u_0,v_0)+\sum_{i=1}^k\Phi^\infty(u_j,v_j).\end{align*}$

下面我们断言,$(u_0,v_0)\not=(0,0)$.否则, 由$d<2c^\infty$和$\Phi^\infty(u_j,v_j)\ge c^\infty$, 有$k=1$.所以由推论2.1和 Choquard 方程

$-\Delta u=\alpha_i(|x|^{-\mu}*u^{2^*_\mu})u^{2^*_\mu-1},\quad x\in \mathbb{R}^N$

正解的唯一性[7],则通过平移和伸缩,$(u_1,v_1)$一定形如

$(k_0U_{0,1},l_0U_{0,1}) \quad\mbox{或}\quad \Big(\alpha_1^{-\frac{N-2}{2(N-\mu+2)}}U_{0,1},0\Big)\quad\mbox{或}\quad\Big(0,\alpha_2^{-\frac{N-2}{2(N-\mu+2)}}U_{0,1}\Big).$

因此,$d=c^\infty$或$\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}$或$\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}$, 这与$d$的假设矛盾. 所以$(u_0,v_0)\not=(0,0)$. 再由引理2.2和$d<2c^\infty$, 可知$k=0$, 这与$k\ge 1$矛盾. 证毕.

我们还需要下面这些估计. 类似于文献[9], 定义映射$\kappa:\mathcal{H}\setminus\{(0,0)\}\to \mathbb{R}^N$,

$\begin{align*}&\kappa(u,v)=\frac{\int_{\mathbb{R}^N}\frac{x}{1+|x|}\Big(\int_{\mathbb{R}^N}\frac{\alpha_1|u^+(x)|^{2^*_\mu}|u^+(y)|^{2^*_\mu}+\alpha_2|v^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}+2\beta|u^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}}{|x-y|^\mu}{\rm d}y\Big){\rm d}x}{\|(u,v)\|^{2\cdot 2_\mu^*}_{NL}},\end{align*}$

其中

$\begin{align*}&\|(u,v)\|^{2\cdot 2_\mu^*}_{NL}\\&:=\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{\alpha_1|u^+(x)|^{2^*_\mu}|u^+(y)|^{2^*_\mu}+\alpha_2|v^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}+2\beta|u^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y.\end{align*}$

同时, 定义泛函

$\begin{align*}&\gamma(u,v)\\&=\frac{\int_{\mathbb{R}^N}\left|\frac{x}{1+|x|}\!-\!\kappa(u,v)\right|\Big(\int_{\mathbb{R}^N}\frac{\alpha_1|u^+(x)|^{2^*_\mu}|u^+(y)|^{2^*_\mu}+\alpha_2|v^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}+2\beta|u^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}}{|x-y|^\mu}{\rm d}y\Big){\rm d}x}{\|(u,v)\|^{2\cdot 2_\mu^*}_{NL}}.\end{align*}$

注意到映射$\kappa(u,v)$和$\gamma(u,v)$是连续的且满足对任意的$t\in \mathbb{R}\setminus\{0\}$,$(u,v)\in\mathcal{H}\setminus\{(0,0)\}$都有

$\begin{equation}\kappa(tu,tv)=\kappa(u,v),\qquad \gamma(tu,tv)=\gamma(u,v).\end{equation}$

引理 3.2 假设$0<\beta\le\sqrt{\alpha_1\alpha_2}$以及$V_1(x)$和$V_2(x)$满足 (V1)-(V3), 则

$\begin{equation}\mathcal{L}^\infty:=\inf\left\{\Phi(u,v):\,\,(u,v)\in\mathcal{N},\,\kappa(u,v)=0,\,\gamma(u,v)=\frac{1}{2}\right\}>c^\infty.\end{equation}$

引理 3.2 的证明过程类似于文献[9,引理 5.1], 所以这里我们省略.

引理 3.3 假设$0<\beta\le\sqrt{\alpha_1\alpha_2}$, 以及$V_1(x)$和$V_2(x)$满足 (V1)-(V3), 则

$\begin{equation} \widetilde{\mathcal{L}^\infty}:=\inf\left\{\Phi(u,v):\,\,(u,v)\in\mathcal{N},\,\kappa(u,v)=0,\,\gamma(u,v)\ge\frac{1}{2}\right\}>c^\infty. \end{equation}$

首先, 显然有

$\begin{equation}\widetilde{\mathcal{L}^\infty}\ge c^\infty.\end{equation}$

下面只需证明(3.4)式中的不等式是严格的. 反之, 假设(3.4)式中的等式成立, 则存在一个序列$(u_n,v_n)$使得

$\begin{equation} (u_n,v_n)\in\mathcal{N},\quad\kappa(u_n,v_n)=0,\quad\gamma(u_n,v_n)\ge\frac{1}{2}\quad\mbox{和}\quad\lim\limits_{n\to+\infty}\Phi(u_n,v_n)=c^\infty. \end{equation}$

因此, 类似于文献[9,引理5.1]的证明, 由引理 2.3和命题 3.1, 可以有

$(u_n,v_n)=\left(k_0U_{a_n,b_n},l_0U_{a_n,b_n}\right)+(\varphi_n,\psi_n),$

其中$a_n\in \mathbb{R}^N$,$b_n>0$以及$(\varphi_n,\psi_n)\in\mathcal{H}$且在$\mathcal{H}$中满足当$n\to+\infty$时,$(\varphi_n,\psi_n)\to(0,0)$.

下证$\{a_n\}$有界. 否则, 若当$n\to+\infty$时,$a_n\to\infty$. 由(3.5)式, 我们有

$\begin{align*} c^\infty&=\lim\limits_{n\to+\infty}\Phi(u_n,v_n)\\ &=\lim\limits_{n\to+\infty}\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\int_{\mathbb{R}^N}\left(|\nabla u_n|^2+|\nabla v_n|^2+V_1(x)u_n^2+V_2(x)v_n^2\right){\rm d}x\\ &\ge\lim\limits_{n\to+\infty}\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\bigg[\int_{\mathbb{R}^N}\left(|\nabla u_n|^2+|\nabla v_n|^2\right){\rm d}x\\ & \ +\int_{B_{b_n}(a_n)}\left((W_1(x)+V_{1\infty})u_n^2+(W_2(x)+V_{2\infty})v_n^2\right){\rm d}x\bigg]\\ &\ge\liminf_{n\to+\infty}\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\\ & \times\bigg[\int_{\mathbb{R}^N}\left(|\nabla u_n|^2+|\nabla v_n|^2\right){\rm d}x+\int_{B_{b_n}(a_n)}\left(V_{1\infty}u_n^2+V_{2\infty}v_n^2\right){\rm d}x+o_n(1)\bigg]\\ &\ge\liminf_{n\to+\infty}\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)a_n^2\int_{B_{1}(0)}\left(k_0^2V_{1\infty}U_{0,1}^2+l_0^2V_{2\infty}U_{0,1}^2\right){\rm d}x=\infty, \end{align*}$

矛盾. 不失一般性, 假设$\{a_{n}\}$(或子列) 满足$\lim\limits_{n\to+\infty}a_n=a^*$. 并且, 类似于文献[9,引理5.1]的证明, 还可得$a^*>0$. 同理, 也可以证明$\{b_n\}$是有界的, 并且存在$b^*\in\mathbb{R}$, 使得$\lim\limits_{n\to+\infty}b_n=b^*$.

因此, 在$\mathcal{H}$和$L^2_{loc}(\mathbb{R}^N)\times L^2_{loc}(\mathbb{R}^N)$中, 有$\left(k_0U_{a_n,b_n},l_0U_{a_n,b_n}\right)\to\left(k_0U_{a^*,b^*},l_0U_{a^*,b^*}\right).$并且,

$\begin{align*} c^\infty& \ge\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)\bigg[\int_{\mathbb{R}^N}\left(k_0^2|\nabla U_{a^*,b^*}|^2+l_0^2|\nabla U_{a^*,b^*}|^2\right){\rm d}x\\ & \ +\int_{B_{b^*}(a^*)}\left(k^2_0V_{1\infty}U_{a^*,b^*}^2+l_0^2V_{2\infty}U_{a^*,b^*}^2\right){\rm d}x\bigg]\\ & >\left(\frac{1}{2}-\frac{1}{2\cdot 2_\mu^*}\right)(k_0^2+l_0^2)\int_{\mathbb{R}^N}|\nabla U_{a^*,b^*}|^2{\rm d}x=c^\infty, \end{align*}$

矛盾. 引理 3.3 得证.

令常数$\theta$满足

$0<\theta<\min\Bigg\{1,\log_2 {\frac{\min\Big\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}\Big\}}{c^\infty}}\Bigg\},$

使得

$\begin{equation} \|V_i-V_{i\infty}\|_{L^\frac{N}{2}(\mathbb{R}^N)}=\|W_{i}\|_{L^\frac{N}{2}(\mathbb{R}^N)}=\left(2^{\frac{N-\mu+2}{2N-\mu}\theta}-1\right)S,\quad i=1,2. \end{equation}$

取常数$\hat{c}$满足

$\begin{matrix}\nonumber & c^\infty<\hat{c}\\\ &<\min\left\{\frac{\mathcal{L^\infty}+c^\infty}{2},\frac{N-\mu+2}{2^{1+\theta}(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2^{1+\theta}(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2^{1-\theta}c^\infty\right\}. \end{matrix}$

由引理 3.2, 可知$\hat{c}\in(c^\infty,2c^\infty)$.

设$\eta\in C_0^\infty(B_1(0))$是一个非负径向递减函数, 且存在$0<r<1$, 使得在$B_r(0)$中,$\eta\equiv1$. 定义$u_\varepsilon(x)=\eta(x)U_{0,\varepsilon}(x)$,$\varepsilon>0$. 令

$ t^2_\varepsilon=\frac{\int_{\mathbb{R}^N}|\nabla u_\varepsilon|^2{\rm d}x}{\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{u_\varepsilon(x)^{2^*_\mu}u_\varepsilon(y)^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y}. $

定义

$\begin{equation}w(x)=t_\varepsilon u_\varepsilon(x),\end{equation}$

其中$\varepsilon>0$且足够小. 此时, 非负径向函数$w\in C_0^\infty(\mathbb{R}^N)$满足: supp$w\subset B_1(0)$, 且关于$r=|x|$,$w(r)$是非增的. 此外, 文献[9]证明了

$\begin{equation} \|w\|_{D^{1,2}(\mathbb{R}^N)}^2=\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{w(x)^{2^*_\mu}w(y)^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y>S_{H,L}^{\frac{2N-\mu}{N-\mu+2}}, \end{equation}$
$\begin{equation} \frac{N-\mu+2}{2(2N-\mu)}(k_0^2+l_0^2)\|w\|^2_{D^{1,2}(\mathbb{R}^N)}<\hat{c}. \end{equation}$

文献[2,引理 3.6](或文献[12,引理 4.2]) 中证明了

引理 3.4 对于$\delta>0$,$z\in \mathbb{R}^N$, 定义

$\begin{equation}\omega_{\delta,z}(x)=\delta^{-\frac{N-2}{2}}w\left(\frac{x-z}{\delta}\right),\end{equation}$

其中$w(\cdot)$由(3.8)式给出. 对于$i=1,2$, 则有下列估计式成立

(a)$\lim\limits_{\delta\to 0}\sup\left\{\int_{\mathbb{R}^N}W_{i}(x)|\omega_{\delta,z}|^2{\rm d}x:\,\,z\in \mathbb{R}^N\right\}=0;$

(b)$\lim\limits_{\delta\to+\infty}\sup\left\{\int_{\mathbb{R}^N}W_{i}(x)|\omega_{\delta,z}|^2{\rm d}x:\,\,z\in \mathbb{R}^N\right\}=0;$

(c)$\lim\limits_{r\to+\infty}\sup\left\{\int_{\mathbb{R}^N}W_{i}(x)|\omega_{\delta,z}|^2{\rm d}x:\,\,|z|=r,\,\,z\in \mathbb{R}^N,\,\,\delta>0\right\}=0.$

类似于文献[9,引理 5.3]中的证明, 我们可以证明

引理 3.5 下列关系式成立

(a)$\lim\limits_{\delta\to 0}\sup\left\{\gamma\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right):\,\,z\in \mathbb{R}^N\right\}=0$;

(b)$\lim\limits_{\delta\to+\infty}\inf\left\{\gamma\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right):\,\,z\in \mathbb{R}^N,\,\,|z|\le r\right\}=1$,$\forall r>0$;

(c)$\langle\kappa\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right),z\rangle>0$,$\forall z\in \mathbb{R}^N$,$\forall \delta>0$, 其中$\langle x,z\rangle$表示向量$x,z\in \mathbb{R}^N$的内积.

4 高能量解的存在性

本节将证明高能量解的存在性, 即定理1.1. 我们首先定义泛函$\mathcal{J}:\mathcal{H}\to\mathbb{R}$,

$\begin{align*} & \mathcal {J}(u,v)=\frac{1}{2}\int_{\mathbb{R}^N}\left(|\nabla u|^2+|\nabla v|^2+W_1(x)u^2+W_2(x)v^2\right){\rm d}x\\ &-\frac{1}{2\cdot 2^*_\mu}\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{\alpha_1|u^+(x)|^{2^*_\mu}|u^+(y)|^{2^*_\mu}+\alpha_2|v^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}+2\beta|u^+(x)|^{2^*_\mu}|v^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y, \end{align*}$

其中$W_i(x)=V_i(x)-V_{i\infty}\in L^\frac{N}{2}(\mathbb{R}^N)$,$i=1,2$. 定义 Nehari 流形$\mathcal{N}^0:=\{(u,v)\in\mathcal{H}\setminus\{(0,0)\}:\,\langle\mathcal{J}^\prime(u,v),(u,v)\rangle=0\},$并且令

$\mathcal{L}^0:=\left\{\mathcal{J}(u,v):\,\,(u,v)\in\mathcal{N}^0,\,\kappa(u,v)=0,\,\gamma(u,v)=\frac{1}{2}\right\}.$

对于每个$\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)$($(k_{0},l_{0})$见引理2.2, 我们分别记$\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}$和$\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}$为$\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)$在$\mathcal{N}^0$和$\mathcal{N}$上的投影,即

$\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}:=\widehat{t_{\delta,z}}\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)\in \mathcal{N}^0\quad\mbox{和}\quad\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}:=\widetilde{t_{\delta,z}}\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)\in \mathcal{N},$

则我们可以证明

引理 4.1 下述结论成立

(a)$\lim\limits_{\delta\to 0}\sup\left\{|\widehat{t_{\delta,z}}-1|:\,\,z\in \mathbb{R}^{N}\right\}=0$;

(b)$\lim\limits_{\delta\to+\infty}\sup\left\{|\widehat{t_{\delta,z}}-1|:\,\,z\in \mathbb{R}^{N}\right\}=0$;

(c)$\lim\limits_{r\to 0}\sup\left\{|\widehat{t_{\delta,z}}-1|:\,\,z\in \mathbb{R}^{N},\,\,|z|=r,\,\,\delta>0\right\}=0$.

首先, 由(3.9)式和$\big(\widehat{t_{\delta,z}}k_0\omega_{\delta,z},\widehat{t_{\delta,z}}l_0\omega_{\delta,z}\big)\in \mathcal{N}^0$, 我们有

$\begin{align*} 1&=\frac{\|\omega_{\delta,z}\|_{D^{1,2}(\mathbb{R}^N)}^2}{\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{\omega_{\delta,z}(x)^{2^*_\mu}\omega_{\delta,z}(y)^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y}\\ &=\frac{\widehat{t_{\delta,z}}^{{2\cdot 2_{\mu}^{*}-2}}\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{\omega_{\delta,z}(x)^{2^*_\mu}\omega_{\delta,z}(y)^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y-\frac{1}{k_{0}^{2}+l_{0}^{2}}\int_{\mathbb{R}^{N}}k_{0}^{2}W_{1}(x)\omega_{\delta,z}^{2}+l_{0}^{2}W_{2}(x)\omega_{\delta,z}^{2}{\rm d}x}{\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{\omega_{\delta,z}(x)^{2^*_\mu}\omega_{\delta,z}(y)^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y}. \end{align*}$

所以, 由上式和引理3.4, 可以得到结论.

引理 4.2 存在常数$\bar{r}>0$,$0<\delta_{1}<\frac{1}{2}<\delta_{2}$, 使得对任意的$z\in \mathbb{R}^{N}$满足$|z|<\bar{r}$, 有

$\begin{equation} \gamma\widehat{\left(k_0\omega_{\delta_{1},z},l_0\omega_{\delta_{1},z}\right)}<\frac{1}{2},\quad\gamma\widehat{\left(k_0\omega_{\delta_{2},z},l_0\omega_{\delta_{2},z}\right)}>\frac{1}{2} \end{equation}$

$\begin{equation} \sup\{\mathcal{J}\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}:\,\,(\delta,z)\in\partial \Gamma\}<\hat{c} \end{equation}$

成立, 其中$\Gamma:=\{(\delta,z)\in \mathbb{R}r^{+}\times \mathbb{R}^{N}:\,\,\delta\in[\delta_{1},\delta_{2}],\,\,|z|<\bar{r}\}.$

直接计算, 可知

$\begin{align*} \mathcal{J}\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}&=\frac{\widehat{t_{\delta,z}}^{2}}{2}(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x+\frac{\widehat{t_{\delta,z}}^{2}}{2}\int_{\mathbb{R}^N}\left(W_1(x)k_{0}^{2}\omega_{\delta,z}^2+W_2(x)l_{0}^{2}\omega_{\delta,z}^2\right){\rm d}x\\ &\quad-\frac{\widehat{t_{\delta,z}}^{2\cdot2^{*}_{\mu}}}{2\cdot 2^*_\mu}(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|\omega_{\delta,z}^+(x)|^{2^*_\mu}|\omega_{\delta,z}^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y. \end{align*}$

下面, 我们通过三步来证明结论. 首先, 通过(3.1)式, 引理3.4(a),(3.10) 式, 引理 3.5(a) 和引理 4.1(a), 我们知道存在$\delta_{1}\in\left(0,\frac{1}{2}\right)$, 使得对所有$z\in\mathbb{R}^{N}$, 有$\gamma\widehat{\left(k_0\omega_{\delta_{1},z},l_0\omega_{\delta_{1},z}\right)}<\frac{1}{2}$和$\mathcal{J}\widehat{\left(k_0\omega_{\delta_{1},z},l_0\omega_{\delta_{1},z}\right)}<\hat{c}.$其次, 由引理3.4(c), (3.10) 式以及引理4.1(c), 我们可以证明存在$\bar{r}>0$, 并且当$|z|=\bar{r}$, 对所有的$\delta>0$都有$\mathcal{J}\widehat{\left(k_0\omega_{\delta_{1},z},l_0\omega_{\delta_{1},z}\right)}<\hat{c}.$最后, 固定$\bar{r}$, 然后再利用(3.1)式, 引理3.4(b), (3.10)式, 引理3.5(b) 和引理4.1(b), 我们可知存在$\delta_{2}>\frac{1}{2}$, 使得当$|z|\le\bar{r}$时, 有$\gamma\widehat{\left(k_0\omega_{\delta_{2},z},l_0\omega_{\delta_{2},z}\right)}>\frac{1}{2}$和$\mathcal{J}\widehat{\left(k_0\omega_{\delta_{2},z},l_0\omega_{\delta_{2},z}\right)}<\hat{c}.$证毕.

引理 4.3 假设$\Gamma$由引理4.2中的给定, 则存在$(\tilde{\delta},\tilde{z})\in\partial\Gamma$以及$(\bar{\delta},\bar{z})\in\mathring{\Gamma}$使得

$\begin{equation} \kappa(k_{0}\omega_{\tilde{\delta},\tilde{z}},l_{0}\omega_{\tilde{\delta},\tilde{z}})=0,\quad\gamma(k_{0}\omega_{\tilde{\delta},\tilde{z}},l_{0}\omega_{\tilde{\delta},\tilde{z}})\ge\frac{1}{2}, \end{equation}$

以及

$\begin{equation} \kappa(k_{0}\omega_{\bar{\delta},\bar{z}},l_{0}\omega_{\bar{\delta},\bar{z}})=0,\quad\gamma(k_{0}\omega_{\bar{\delta},\bar{z}},l_{0}\omega_{\bar{\delta},\bar{z}})=\frac{1}{2}. \end{equation}$

首先, 当$(\tilde{\delta},\tilde{z})=(\delta_{2},0)$时, (4.3) 式显然成立. 事实上, 根据$\omega_{\delta_{2},0}$的对称性, 我们知道$\kappa(k_{0}\omega_{\delta_{2},0},l_{0}\omega_{\delta_{2},0})=0$. 此外, 再由(3.1) 式和引理 4.2, 则有$\gamma(k_{0}\omega_{\delta_{2},0},l_{0}\omega_{\delta_{2},0})\ge\frac{1}{2}$.

对于$(\delta,z)\in\Gamma$, 令$\Xi(\delta,z)=\left(\gamma(k_{0}\omega_{\delta,z},l_{0}\omega_{\delta,z}),\kappa(k_{0}\omega_{\delta,z},l_{0}\omega_{\delta,z})\right),$并且定义映射$\mathcal{T}:[0,1] \times\partial\Gamma\to \mathbb{R}\times \mathbb{R}^{N}$,

$\begin{equation}\mathcal{T}(t,\delta,z):=(1-t)(\delta,z)+t\Xi(\delta,z).\end{equation}$

为了证明(4.4)式, 只需证明

$\begin{equation} \deg\left(\Xi,\mathring{\Gamma},\left(\frac{1}{2},0\right)\right)=1. \end{equation}$

我们断言, 对任意的$(\delta,z)\in\partial\Gamma$,$t\in[0,1]$, 都有

$\begin{equation} \left((1-t)\delta+t\gamma(k_{0}\omega_{\delta,z},l_{0}\omega_{\delta,z}),(1-t)z+t\kappa(k_{0}\omega_{\delta,z},l_{0}\omega_{\delta,z})\right)\not=\left(\frac{1}{2},0\right). \end{equation}$

注意到$\deg\left(Id,\mathring{\Gamma},\left(\frac{1}{2},0\right)\right)=1$, 再由拓扑度同伦不变性, 可以得到(4.6)式.

所以下面证明(4.7)式成立. 将$\partial\Gamma$分为三部分来讨论,$\partial\Gamma=\Gamma_{1}\cup\Gamma_{2}\cup\Gamma_{3}$, 其中

$\Gamma_{1}=\{(\delta,z)\in\partial\Gamma:\,\,|z|\le\bar{r},\,\,\delta=\delta_{1}\},$
$\Gamma_{2}=\{(\delta,z)\in\partial\Gamma:\,\,|z|\le\bar{r},\,\,\delta=\delta_{2}\},$
$\Gamma_{3}=\{(\delta,z)\in\partial\Gamma:\,\,|z|=\bar{r},\,\,\delta\in[\delta_{1},\delta_{2}]\}.$

若$(\delta,z)\in\Gamma_{1}$, 由(3.1)和(4.1)式, 有$(1-t)\delta_{1}+t\gamma(k_{0}\omega_{\delta_{1},z},l_{0}\omega_{\delta_{1},z})<\frac{1}{2}(1-t)+\frac{1}{2}t=\frac{1}{2}.$如果$(\delta,z)\in\Gamma_{2}$, 由(3.1)和(4.1)式, 有$(1-t)\delta_{2}+t\gamma(k_{0}\omega_{\delta_{2},z},l_{0}\omega_{\delta_{2},z})>\frac{1}{2}(1-t)+\frac{1}{2}t=\frac{1}{2}.$如果$(\delta,z)\in\Gamma_{3}$, 由引理 3.5(c) 有$\langle(1-t)z+t\kappa(k_{0}\omega_{\delta,z},l_{0}\omega_{\delta,z}),z\rangle=(1-t)|z|^{2}+t\langle(\kappa(k_{0}\omega_{\delta,z},l_{0}\omega_{\delta,z}),z\rangle>0,$即$(1-t)z+t\kappa(k_{0}\omega_{\delta,z},l_{0}\omega_{\delta,z})\not=0$. 因此, (4.7)式成立. 证毕.

引理 4.4 假设$\Gamma$由引理4.2中的给定. 若$W_{1}(x)$和$W_{2}(x)$满足(3.6)式, 则有

$\begin{matrix} \mathcal{K}&=\sup\left\{\mathcal{J}\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}:\,\,(\delta,z)\in\Gamma\right\} \\ & <\min\left\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\right\}. \end{matrix}$

因为$\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}=\widehat{t_{\delta,z}}\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)\in \mathcal{N}^0$, 通过直接计算, 我们有

$\begin{matrix}& \widehat{t_{\delta,z}}^{2}(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x+\widehat{t_{\delta,z}}^{2}\int_{\mathbb{R}^N}\left(W_1(x)k_{0}^{2}\omega_{\delta,z}^2+W_2(x)l_{0}^{2}\omega_{\delta,z}^2\right){\rm d}x \\ =\ &\widehat{t_{\delta,z}}^{2\cdot2^{*}_{\mu}}(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|\omega_{\delta,z}^+(x)|^{2^*_\mu}|\omega_{\delta,z}^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y. \end{matrix}$

所以, 由 Hölder 不等式, (3.6)和(3.9)式,

$\begin{matrix}\widehat{t_{\delta,z}}^{2\cdot2^{*}_{\mu}-2}&=\frac{(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x+\int_{\mathbb{R}^N}\left(W_1(x)k_{0}^{2}\omega_{\delta,z}^2+W_2(x)l_{0}^{2}\omega_{\delta,z}^2\right){\rm d}x}{(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|\omega_{\delta,z}^+(x)|^{2^*_\mu}|\omega_{\delta,z}^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y} \\&\le1+\frac{\left(k_{0}^{2}\|W_{1}\|_{L^\frac{N}{2}}+l_{0}^{2}\|W_{2}\|_{L^\frac{N}{2}}\right)\|\omega_{\delta,z}\|_{L^{2^{*}}}^{2}}{(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x} \\&\le 2^{\frac{N-\mu+2}{2N-\mu}\theta},\end{matrix}$

其中$\theta\in(0,1)$由(3.6)式中给出.再由(4.9), (3.6)式和 Hölder 不等式, 得

$\begin{matrix}& \ \mathcal{J}\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}\\&=\frac{\widehat{t_{\delta,z}}^{2}}{2}(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x+\frac{\widehat{t_{\delta,z}}^{2}}{2}\int_{\mathbb{R}^N}\left(W_1(x)k_{0}^{2}\omega_{\delta,z}^2+W_2(x)l_{0}^{2}\omega_{\delta,z}^2\right){\rm d}x \\ &\quad\ -\frac{\widehat{t_{\delta,z}}^{2\cdot2^{*}_{\mu}}}{2\cdot 2^*_\mu}(k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{|\omega_{\delta,z}^+(x)|^{2^*_\mu}|\omega_{\delta,z}^+(y)|^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y \\&=\left(\frac{1}{2}-\frac{1}{2\cdot 2^*_\mu}\right)\widehat{t_{\delta,z}}^{2}\left((k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x+\int_{\mathbb{R}^N}\left(W_1(x)k_{0}^{2}\omega_{\delta,z}^2+W_2(x)l_{0}^{2}\omega_{\delta,z}^2\right){\rm d}x\right) \\&\le\left(\frac{1}{2}-\frac{1}{2\cdot 2^*_\mu}\right)\widehat{t_{\delta,z}}^{2}\left((k_{0}^{2}+l_{0}^{2})\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x+\left(k_{0}^{2}\|W_{1}\|_{L^\frac{N}{2}}+l_{0}^{2}\|W_{2}\|_{L^\frac{N}{2}}\right)\|\omega_{\delta,z}\|_{L^{2^{*}}}^{2}\right) \\&\le\frac{N-\mu+2}{2(2N-\mu)}2^{\frac{N-\mu+2}{2N-\mu}\theta}\widehat{t_{\delta,z}}^{2}\left(k_{0}^{2}+l_{0}^{2}\right)\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x.\end{matrix}$

最后, 结合(4.10), (4.11), (3.10)和(3.7)式, 有

$\begin{align*}\mathcal{J}\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}&\le\frac{N-\mu+2}{2(2N-\mu)}2^{\frac{N-\mu+2}{2N-\mu}\theta\cdot\left(\frac{1}{2^*_{\mu}-1}+1\right)}\left(k_{0}^{2}+l_{0}^{2}\right)\int_{\mathbb{R}^N}|\nabla \omega_{\delta,z}|^2{\rm d}x\\& <2^{\theta}\hat{c}<\min\left\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\right\}.\end{align*}$

因此, (4.8)式成立.

引理 4.5 假设$\delta_{1},\,\delta_{2},\,\bar{r}$和$\Gamma$由引理 4.2 中的给定, 则存在常数$\widetilde{V}_{i}>0$, 使得当对$V_{i\infty}\in(0,\widetilde{V}_{i})$时,$i=1,2$, 有

$\begin{equation}\gamma\widetilde{\left(k_0\omega_{\delta_{1},z},l_0\omega_{\delta_{1},z}\right)}<\frac{1}{2},\quad\gamma\widetilde{\left(k_0\omega_{\delta_{2},z},l_0\omega_{\delta_{2},z}\right)}>\frac{1}{2},\quad\forall z\in \mathbb{R}^{N},\quad |z|<\bar{r}\end{equation}$

$\begin{equation}\tilde{l}:=\sup\{\Phi\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}:\,\,(\delta,z)\in\partial \Gamma\}<\hat{c}\end{equation}$

成立. 此外, 如果(1.4)式成立, 则存在$V_{i}^{*}$, 使得当$V_{i\infty}\in(0,V_{i}^{*})$时,$i=1,2$, 则有(4.12)和(4.13)式成立, 并且

$\begin{matrix}\widetilde{\mathcal{K}}&:=\sup\{\Phi\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}:\,\,(\delta,z)\in \Gamma\} \\& <\min\left\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\right\}.\end{matrix}$

也成立.

由(3.1)式知,

$\gamma\left(k_0\omega_{\delta_{1},z},l_0\omega_{\delta_{1},z}\right)=\gamma\widehat{\left(k_0\omega_{\delta_{1},z},l_0\omega_{\delta_{1},z}\right)}=\gamma\widetilde{\left(k_0\omega_{\delta_{1},z},l_0\omega_{\delta_{1},z}\right)},\quad\forall (\delta,z)\in\mathbb{R}^{+}\times\mathbb{R}^{N}.$

所以, (4.1)式可推出(4.12)式成立. 另一方面, 通过直接计算, 有

$\begin{align*}1&=\frac{\|\omega_{\delta,z}\|_{D^{1,2}(\mathbb{R}^N)}^2}{\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{\omega_{\delta,z}(x)^{2^*_\mu}\omega_{\delta,z}(y)^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y}\\&=\widetilde{t_{\delta,z}}^{{2\cdot 2_{\mu^{*}}-2}}-\frac{1}{k_{0}^{2}+l_{0}^{2}}\frac{\int_{\mathbb{R}^{N}}[k_{0}^{2}\left(V_{1\infty}+W_{1}(x)\right)\omega_{\delta,z}^{2}+l_{0}^{2}\left(V_{2\infty}+W_{2}(x)\right)\omega_{\delta,z}^{2}]{\rm d}x}{\int_{\mathbb{R}^N}\int_{\mathbb{R}^N}\frac{\omega_{\delta,z}(x)^{2^*_\mu}\omega_{\delta,z}(y)^{2^*_\mu}}{|x-y|^\mu}\,{\rm d}x{\rm d}y}\end{align*}$

和$\int_{\mathbb{R}^{N}}V_{1\infty}\omega_{\delta,z}^{2}{\rm d}x=V_{1\infty}\delta^2\int_{B_{1}(0)}\omega^{2}{\rm d}x$,$i=1,2$, 也就是说$\lim\limits_{V_{i\infty}\to 0}\sup_{(\delta,z)\in\Gamma}|\widetilde{t_{\delta,z}}-\widehat{t_{\delta,z}}|=0.$所以, 当$V_{1\infty}$和$V_{2\infty}$充分小时, 由 (4.2)和 (4.8) 式可知 (4.13) 和 (4.14) 式成立.

下面, 通过上述引理和形变引理[16,引理 2.3]来证明高能量解的存在性. 首先, 定义集合$\Phi^{c}=\{(u,v)\in\mathcal{N}:\,\,\Phi(u,v)\le c\}.$

先证明定理 1.1(i).

我们将证明泛函$\Phi|_{\mathcal{N}}$在$(c^{\infty},\hat{c})$有一个临界值. 令$V_{i\infty}\in(0,\widetilde{V}_{i})$,$i=1,2$, 其中$\widetilde{V}_{i}$由引理 4.5 给出, 则由 (3.2), (3.3), (3.7), (4.3) 和 (4.13) 式, 有

$\begin{equation}c^{\infty}<\widetilde{\mathcal{L}^\infty}\le \Phi(k_{0}\omega_{\tilde{\delta},\tilde{z}},l_{0}\omega_{\tilde{\delta},\tilde{z}})\le\tilde{l}<\hat{c}<\mathcal{L}^\infty.\end{equation}$

断言泛函$\Phi|_{\mathcal{N}}$在$(\widetilde{\mathcal{L}^\infty},\tilde{l})$中有一个临界值. 如若不然,由推论 3.1, 已经知道泛函$\Phi$在$\bigg(c^\infty,\min\Big\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\Big\}\bigg)$中满足 Palais-Smale 条件, 则由形变引理, 存在正常数$\mu_{1}$满足$\widetilde{\mathcal{L}^\infty}-\mu_{1}>c^{\infty}$和$\tilde{l}+\mu_{1}<\hat{c}.$并且, 还存在一个连续函数$\eta:[0,1]\times\Phi^{\tilde{l}+\mu_{1}}\to\Phi^{\tilde{l}+\mu_{1}}$使得

$\begin{equation}\eta(0,u,v)=(u,v),\quad\forall u\in\Phi^{\tilde{l}+\mu_{1}},\end{equation}$
$\begin{equation}\eta(s,u,v)=(u,v),\quad\forall u\in\Phi^{\widetilde{\mathcal{L}^\infty}-\mu_{1}},\,\,\forall s\in[0,1],\end{equation}$
$\begin{equation}\Phi \circ\eta(s,u,v)\le \Phi(u,v),\quad\forall u\in\Phi^{\tilde{l}+\mu_{1}},\,\,\forall s\in[0,1],\end{equation}$
$\begin{equation}\eta\left(1,\Phi^{\tilde{l}+\mu_{1}}\right)\subset\Phi^{\widetilde{\mathcal{L}^\infty}-\mu_{1}}.\end{equation}$

由(4.13)和(4.15)式, 有

$\begin{equation}(\delta,z)\in\partial\Gamma\Longrightarrow\Phi\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}\le\tilde{l}\Longrightarrow\Phi\circ\eta\left(1,\widetilde{t_{\delta,z}}k_0\omega_{\delta,z},\widetilde{t_{\delta,z}}l_0\omega_{\delta,z}\right)\le\widetilde{\mathcal{L}^\infty}-\mu_{1}.\end{equation}$

假设$s\in[0,1]$,$(\delta,y)\in\Gamma$, 定义

$\begin{align*}\Theta(t,\delta,z)=\left\{\begin{array}{ll}\mathcal{T}(2t,\delta,z),& t\in\left[\frac{1}{2}\right],\\[3mm]\left(\gamma\circ\eta(2t\!-\!1,\widetilde{t_{\delta,z}}k_0\omega_{\delta,z},\widetilde{t_{\delta,z}}l_0\omega_{\delta,z}),\beta\circ\eta(2t\!-\!1,\widetilde{t_{\delta,z}}k_0\omega_{\delta,z},\widetilde{t_{\delta,z}}l_0\omega_{\delta,z}\right),& t\in\left[\frac{1}{2},1\right], \end{array} \right.\end{align*}$

其中$\mathcal{T}$定义见(4.5)式. 由(4.7)式得

$\begin{equation}\forall t\in\left[\frac{1}{2}\right],\quad(\delta,y)\in\partial\Gamma,\quad\Theta(t,\delta,z)\not=\left(\frac{1}{2},0\right).\end{equation}$

由(4.20), (4.18) 和 (4.15) 式, 有

$\begin{equation*}\Phi\circ\eta\left(2t-1,\widetilde{t_{\delta,z}}k_0\omega_{\delta,z},\widetilde{t_{\delta,z}}l_0\omega_{\delta,z}\right)\le\Phi\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}<\hat{c}<\mathcal{L}^\infty \quad\forall t\in\left[\frac{1}{2},1\right],\,\,(\delta,z)\in\partial\Gamma,\end{equation*}$

$\begin{equation}\forall t\in\left[\frac{1}{2},0\right],\quad(\delta,y)\in\partial\Gamma,\quad\Theta(t,\delta,z)\not=\left(\frac{1}{2},0\right).\end{equation}$

所以, 根据(4.21), (4.22)式和$\Theta$的连续性, 存在$\left(\hat{\delta},\hat{z}\right)\in\partial\Gamma$使得

$\begin{equation}\beta\circ\eta\left(1,\widetilde{t_{\hat{\delta},\hat{z}}}k_0\omega_{\hat{\delta},\hat{z}},\widetilde{t_{\hat{\delta},\hat{z}}}l_0\omega_{\hat{\delta},\hat{z}}\right)=0,\quad \gamma\circ\eta\left(1,\widetilde{t_{\hat{\delta},\hat{z}}}k_0\omega_{\hat{\delta},\hat{z}},\widetilde{t_{\hat{\delta},\hat{z}}}l_0\omega_{\hat{\delta},\hat{z}}\right)\ge\frac{1}{2}.\end{equation}$

事实上, 由于$\partial \Gamma$与$\mathbb{R}^{N+1}$中的球面同伦, 并且$\left(\frac{1}{2},0\right)\in\mathring{\Gamma}$, 所以直线$\left(\frac{1}{2},+\infty\right)\times\{0\}$与$\partial\Gamma$相交. 又因为$\partial\Gamma$与$\Theta(1,\partial\Gamma)$同伦, 所以(4.23)式不成立当且仅当$\Theta(1,\delta,z)=\left(\frac{1}{2},0\right)$, 其中$t\in[0,1]$,$(\delta,z)\in\partial\Gamma$. 但是, 因为(4.21)和(4.22)式, 这是不可能发生的. 结合(4.20), (4.17), (3.3) 和 (4.23) 式, 有

$\Phi\circ \eta\left(1,\widetilde{t_{\hat{\delta},\hat{z}}}k_0\omega_{\hat{\delta},\hat{z}},\widetilde{t_{\hat{\delta},\hat{z}}}l_0\omega_{\hat{\delta},\hat{z}}\right)\ge \widetilde{\mathcal{L}^\infty},$

这与(4.20)式矛盾.

所以, 当$V_{i\infty}\in(0,\tilde{V}_{i})$时,$i=1,2$, 泛函$\Phi|_{\mathcal{N}}$有一个临界点$(u_{l},v_{l})\in\mathcal{N}$满足$\Phi(u_{l},v_{l})\in(c^{\infty},\hat{c})$. 再由引理2.5和

$\Phi\left(u_l, v_l\right)<\hat{c}<\min \left\{\frac{N-\mu+2}{2(2 N-\mu)} \alpha_1^{\frac{N-2}{\mu-2-N}} S_{H, L}^{\frac{2 N-\mu}{N-\mu+2}}, \frac{N-\mu+2}{2(2 N-\mu)} \alpha_2^{\frac{N-2}{\mu-2-N}} S_{H, L}^{\frac{2 N-\mu}{N-\mu-2}}\right\},$

很容易知道$u_{l}\not=0$和$v_{l}\not=0$. 所以,$(u_{l},v_{l})$是泛函$\Phi$的临界点. 最后, 由极大值原理, 我们可知$(u_{l},v_{l})$是方程(1.1)的正解. 证毕.

最后, 证明定理1.1(ii).

假设 (1.4) 成立, 且$V_{i\infty}\in(0,V^*_i)$,$i=1,2$, 其中$V^*_i$由引理 4.5 中给出. 由 (3.2), (3.3), (3.7), (4.4) 和 (4.14) 式, 有

$\begin{matrix}\hat{c}&<\mathcal{L}^\infty\le\Phi(k_{0}\omega_{\bar{\delta},\bar{z}},l_{0}\omega_{\bar{\delta},\bar{z}})\le\tilde{\mathcal{K}} \\&<\min\left\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\right\}.\end{matrix}$

我们断言泛函$\Phi|_{\mathcal{N}}$在$(\mathcal{L}^\infty,\tilde{\mathcal{K}})$中有一个临界值. 类似于上面的证明, 我们也用反证法进行论证, 假设其不成立.

由推论 3.1 可知$\Phi$在$\bigg(c^\infty,\min\Big\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\Big\}\bigg)$中满足 Palais-Smale 条件, 则再由形变引理, 存在一个正常数$\mu_{2}$满足$\mathcal{L^\infty}-\mu_{2}>\hat{c}$和

$\tilde{\mathcal{K}}+\mu_{2}<\min\left\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\right\},$

以及连续函数$\eta: \Phi^{\tilde{\mathcal{K}}+\mu_{2}}\to\Phi^{\mathcal{L^\infty}-\mu_{2}}$满足

$\eta(u,v)=(u,v),\quad\forall (u,v)\in\Phi^{\mathcal{L^\infty}-\mu_{2}}$

以及

$\Phi\circ\eta\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}\le \mathcal{L^\infty}-\mu_{2},\quad \forall (\delta,z)\in\Gamma.$

因此, 由引理 4.5, 有

$\begin{equation}\Pi(\delta,z):=\left(\gamma\circ\eta\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)},\,\beta\circ\eta\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}\right)\not=\left(\frac{1}{2},0\right),\quad \forall (\delta,z)\in\Gamma.\end{equation}$

另一方面, 在引理 4.5 的证明中, 我们已经证得了这一事实: 若$V_{i\infty}\in(0,V_i^*)$,$i=1,2$, 则

$\sup_{(\delta,z)\in\Gamma}|\widetilde{t_{\delta,z}}-\widehat{t_{\delta,z}}|=o(1).$

因此,

$\begin{equation}\sup\{\Phi\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}:\,\,(\delta,z)\in\partial\Gamma\}=\sup\{\mathcal{J}\widehat{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}:\,\,(\delta,z)\in\partial\Gamma\}+o(1).\end{equation}$

由(4.2)和(4.26)式, 可知若$V_{i\infty}\in(0,V_i^*)$,$i=1,2$, 则

$\Phi\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}<\hat{c}<\mathcal{L^\infty}-\mu_{2},\quad \forall (\delta,z)\in\partial\Gamma,$

这意味着

$\eta\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}=\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)},\quad \forall (\delta,z)\in\partial\Gamma.$

此外,

$\Pi(\delta,z):=\Xi(\delta,z)=\left(\gamma\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)},\,\beta\widetilde{\left(k_0\omega_{\delta,z},l_0\omega_{\delta,z}\right)}\right),\quad \forall (\delta,z)\in\partial\Gamma.$

再利用(4.6)式和拓扑度的同伦不变性, 有

$1=\deg\left(\Xi,\mathring{\Gamma},\left(\frac{1}{2},0\right)\right)=\deg\left(\Pi,\mathring{\Gamma},\left(\frac{1}{2},0\right)\right).$

所以, 存在$(\check{\delta},\check{z})$使得$\Pi(\check{\delta},\check{z})=\left(\frac{1}{2},0\right)$, 这与 (4.25) 式矛盾. 因此,$\Phi(u,v)$在$\mathcal{N}$的限制下至少存在一个正的临界点$(u_h,v_h)\in\mathcal{N}$, 并且满足

$\hat{c}<\Phi(u_h,v_h)<\min\left\{\frac{N-\mu+2}{2(2N-\mu)}\alpha_1^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},\frac{N-\mu+2}{2(2N-\mu)}\alpha_2^{\frac{N-2}{\mu-2-N}}S_{H,L}^{\frac{2N-\mu}{N-\mu+2}},2c^\infty\right\}.$

证毕.

参考文献

Benci V, Cerami G.

Existence of positive solutions of the equation$-\Delta u+a(x)u=u^{\frac{N+2}{N-2}}$in$\mathbb{R}^{N}$

J Funct Anal, 1990, 88: 90-117

DOI:10.1016/0022-1236(90)90120-A      URL     [本文引用: 2]

Cerami G, Molle R.

Multiple positive bound states for critical Schrödinger-Poisson systems

ESAIM Control Optim Calc Var, 2019, 25: Paper No. 73

[本文引用: 5]

Chen W, Li C, Ou B.

Classification of solutions for an integral equation

Comm Pure Appl Math, 2006, 59(3): 330-343

DOI:10.1002/cpa.v59:3      URL     [本文引用: 1]

Chen Z, Zou W.

Positive least energy solutions and phase separation for coupled Schrödinger equations with critical exponent: Higher dimensional case

Calc Var Partial Differential Equations, 2015, 52(1/2): 423-467

DOI:10.1007/s00526-014-0717-x      URL     [本文引用: 3]

Dai W, Huang J, Qin Y, et al.

Regularity and classification of solutions to static Hartree equations involving fractional Laplacians

Discrete Contin Dyn Syst, 2019, 39(3): 1389-1403

DOI:10.3934/dcds.2018117      URL     [本文引用: 1]

Dalfovo F, Giorgini S, Pitaevskii L P, Stringari S.

Theory of Bose-Einstein condensation in trapped gases

Rev Mod Phys, 1999, 71: 463-512

DOI:10.1103/RevModPhys.71.463      URL     [本文引用: 1]

Du L, Yang M.

Uniqueness and nondegeneracy of solutions for a critical nonlocal equation

Discrete Contin Dyn Syst, 2019, 39(10): 5847-5866

DOI:10.3934/dcds.2019219      URL     [本文引用: 4]

Esry B, Greene C, Burke J, Bohn J.

Hartree-Fock theory for double condensates

Phys Rev Lett, 1997, 78: 3594-3597

DOI:10.1103/PhysRevLett.78.3594      URL     [本文引用: 1]

Gao F, Liu H, Moroz V, Yang M.

High energy positive solutions for a coupled Hartree system with Hardy-Littlewood-Sobolev critical exponents

J Differential Equations, 2021, 287: 329-375

DOI:10.1016/j.jde.2021.03.051      URL     [本文引用: 13]

Gao F, Yang M.

The Brezis-Nirenberg type critical problem for the nonlinear Choquard equation

Sci China Math, 2018, 61(7): 1219-1242

DOI:10.1007/s11425-016-9067-5      [本文引用: 1]

Guo L, Li Q.

Multiple high energy solutions for fractional Schrödinger equation with critical growth

Calc Var Partial Differential Equations, 2022, 61(1): Paper No. 15

[本文引用: 2]

Liu H, Liu Z.

Positive solutions of a nonlinear Schrödinger system with nonconstant potentials

Discrete Contin Dyn Syst, 2016, 36: 1431-1464

DOI:10.3934/dcdsa      URL     [本文引用: 1]

Liu H, Liu Z.

A coupled Schrödinger system with critical exponent

Calc Var Partial Differential Equations, 2020, 59(5): Paper No. 145

[本文引用: 2]

Mitchell M, Segev M.

Self-trapping of incoherent white light

Nature, 1997, 387: 880-883

DOI:10.1038/43136      URL     [本文引用: 1]

Wang J.

Qualitative analysis for the nonlinear fractional Hartree type system with nonlocal interaction

Adv Nonlinear Anal, 2022, 11(1): 385-416

[本文引用: 3]

Willem M. Minimax Theorems.

Progress in Nonlinear Differential Equations and Their Applications, 24

Boston: Birkhäuser, 1996

[本文引用: 2]

Zheng Y, Gao F, Shen Z, Yang M.

On a class of coupled critical Hartree system with deepening potential

Math Methods Appl Sci, 2021, 44(1): 772-798

DOI:10.1002/mma.v44.1      URL     [本文引用: 1]

/