数学物理学报  2017, Vol. 37 Issue (3): 491-498   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
李金菊
张正杰
广义Choquard-Pekar方程两个非负解的存在性
李金菊, 张正杰     
华中师范大学数学与统计学学院 武汉 430079
摘要:运用集中紧致原理、变分方法以及局部极值方法,研究广义Choquard-Pekar方程$-\Delta u+a(x)u=\int _{{\Bbb R}^N}\frac{Q(x,y)u^2(y){\rm d}y}{|x|^h|x-y|^{r-2h}|y|^h}\cdot u(x)+g(x) , \quad x\in {\Bbb R}^N.$.作者得到一定条件下这类问题的两个非负解的存在性.其中一个解是通过局部极小得到的,另一个是运用山路引理得到的.
关键词广义Choquard-Pekar方程    非负解    山路引理    
The Existence of Two Non-Negative Solutions for the Generalized Choquard-Pekar Equation
Li Jinju, Zhang Zhengjie     
School of Mathematics and Statistics, Central China Normal University, Wuhan 430079
Abstract: In the paper, we used variational method to study the generalized Choquard-Pekar equation on ${{\mathbb{R}}^{N}}$. We get that there exists two non-negative solutions for our problem, one solution is a local minimum and the other is of the mountain pass type.
Key words: Choquard-Pekar     Non-negative solution     Mountian pass theorem    
1 引言

本文, 我们考虑如下问题

$\left\{\begin{array}{ll} -\Delta u+a(x)u=\int _{{\Bbb R}^N}\frac{Q(x,y)u^2(y)} {|y|^h|x-y|^{r-2h}}{\rm d}y\cdot \frac{u(x)}{|x|^h}+g(x),x\in {\Bbb R}^N,\\[2mm] u\in H^{1}({\Bbb R}^N),\end{array}\right.$ (1.1)

这里$N\geq 3$, 函数$a(x),Q(x,y)$适合如下条件.

(1) $a(x)$${\Bbb R}^N$中的连续函数, 且$a(x)\geq \lim\limits_{|x|\rightarrow +\infty }a(x)=\overline{a}>0$;

(2) $Q(x,y)=Q(y,x)$${\Bbb R}^N\times {\Bbb R}^N$上连续有界, 且$\lim\limits_{(|x|,|y|)\rightarrow (+\infty,+\infty )}Q(x,y)=\overline{Q}>0$.

$h=0$, $r=1$, $N=3$, $g(x)\equiv 0$时, 方程(1.1) 首先是物理学家Choquard引入的, 相关的问题在等离子体的Hartree-Fock理论以及Thomas-Fermi理论中都有出现, 在量子力学中也有很多的应用.研究这类问题的方法主要就是利用变分原理, 为了得到问题(1.1) 的弱解, 我们可以转化为研究与其相应的变分泛函

$I _{g}(u)=\frac{1}{2}\int _{{\Bbb R}^N}(|\bigtriangledown u|^2+a(x)u^2){\rm d}x-\\ \frac{1}{4}\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2(x)u^2(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y-\int _{{\Bbb R}^N}ug(x){\rm d}x$ (1.2)

临界点的存在性.

本文受文献[8], 及近期几篇研究多解问题的方法的启发(见文献[5, 7, 9-10]), 我们考虑一定条件下泛函$I_g(u)$存在一个局部极小的临界点和山路引理形式的另一个临界点, 然后证明这两个临界点的可达性, 得到问题(1.1) 至少有两个弱解的存在结果.

定理1.1 假设$g(x)\in H^{-1}({\Bbb R}^N)$, $g(x)\geq 0$$g(x)\not\equiv 0$, 在条件(1), (2) 下, 及$0<r<\min\{4,N\}$, $0<2h<r$, $0<h<2$.则存在常数$C$, 当$0<||g(x)||_{H^{-1}}<C $时, 问题(1.1) 在$H^1({\Bbb R}^N)$中至少存在两个非平凡解.

2 本文的几个重要引理及定义

在本节首先我们给出

定义2.1  $||u||=(\int_{{\Bbb R}^N}(|\bigtriangledown u|^2+a(x)u^2){\rm d}x)^{\frac{1}{2}}$是与普通$H^1({\Bbb R}^N)$范数等价的范数.记

$\begin{align}I (u)=\frac{1}{2}\int _{{\Bbb R}^N}(|\bigtriangledown u|^2+a(x)u^2){\rm d}x-\frac{1}{4}\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2(x)u^2(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y.\end{align}$ (2.1)
$I _{\infty }(u)=\frac{1}{2}\int _{{\Bbb R}^N}(|\bigtriangledown u|^2+\overline{a}u^2){\rm d}x-\frac{1}{4}\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{\overline{Q}u^2(x)u^2(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y.$ (2.2)

定义2.2

$F(u)=\left\{\begin{array}{ll} 0,&\mbox{ 若} u \equiv 0,\\ \frac{\int _{{\Bbb R}^N}(|\bigtriangledown u|^2+\overline{a}u^2){\rm d}x} {\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{\overline{Q}u^2(x)u^2(y)} {|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y},~ & \mbox{ 若} u\not\equiv 0,\end{array}\right.$ (2.3)

$I^{\infty }=\inf\{I_{\infty }(u) | u\in H^1({\Bbb R}^N),F(u)=1\}. $

由文献[2]可知, 存在$\overline{u} \in H^1({\Bbb R}^N)$$I^{\infty }$的达到函数, 且

$I^{\infty }=I_{\infty }(\overline{u})=\sup_{t>0}I_{\infty }(t\overline{u}).$ (2.4)

为了证明我们的结果, 我们需要如下引理.

引理2.1 设$\{ P_n\}_{n\geq 1}$$L^1({\Bbb R}^N)$中的序列, 满足在${\Bbb R}^N$$P_n\geq 0 $$\int _{{\Bbb R}^N}P_n=L$ ($L>0$为常数), 则存在子序列, 满足如下三者之一

(1) (compactness)存在$P_{n_k},$是紧的, 即在${\Bbb R}^N$中存在$y_k$满足

$\forall \epsilon >0,\exists R <\infty ,\begin{matrix} {} & {} \\ \end{matrix}\int_{{{y}_{k}}+{{B}_{R}}}{{{P}_{{{n}_{k}}}}}\text{d}x\ge L-\epsilon ;$

(2) (vanishing)

$\lim\limits_{R\to \infty }\sup_{y\in {\Bbb R}^N} \int _{y+B_R}P_{n_k}{\rm d}x=0,\begin{matrix} {} & {} \\ \end{matrix}\forall R<\infty;$

(3) (dichotomy)存在$a\in (0,L)$, $\forall \epsilon >0$存在$k_0>1$, $P^1_k$, $P^2_k$适合, 当$k\geq k_0$

$ ||P_{n_k}-P^1_k-P^2_k||_{L^1({\Bbb R}^N)}<\epsilon,$
$ {\rm dist(supp}\{P^1_k\},{\rm supp}\{P^2_k\})\to \infty,$
$ \Big|\int _{{\Bbb R}^N}P^1_k{\rm d}x-a \Big|<\epsilon,\begin{matrix} {} & {} \\ \end{matrix} \Big|\int _{{\Bbb R}^N}P^2_k{\rm d}x-(L-a) \Big|<\epsilon .$

引理2.2 设$1<p\leq \infty $, $1\leq q<\infty $, 当$N>p$$q\neq \frac{Np}{N-p}$, 如果$u_n$$L^q({\Bbb R}^N)$中有界, $\bigtriangledown u_n$$L^p({\Bbb R}^N)$中有界, 且$\sup\limits_{y\in {\Bbb R}^N}\int _{y+B_R}|u_n|^q\to 0$ (对某$R>0$成立), 则$u_n\to 0$$L^{\alpha }({\Bbb R}^N)$中, $\alpha $位于$q$$\frac{Np}{N-p}$之间.

这两个引理在文献[4]中有详细的证明, 在此我们省略其证明过程.

Hardy-Littlewood不等式:如果$u\in L^p({\Bbb R}^N)$, $v\in L^s({\Bbb R}^N)$, 则

$ \bigg|\iint _{{\Bbb R}^N\times {\Bbb R}^N}\frac{u(x)v(y){\rm d}x{\rm d}x} {|x|^h|y|^l|x-y|^{r-h-l}} \bigg|\leq C|u|_{L^p({\Bbb R}^N)}|v|_{L^s({\Bbb R}^N)},$

其中$k,s,r,h,l$满足如下关系

$ 0<h<N(1-\frac{1}{p}),0<l<N(1-\frac{1}{s}),$
$r=N(2-\frac{1}{p}-\frac{1}{s}),\frac{1}{s}+\frac{1}{p}\geq 1,p>1,s>1.$

如果$u\in H^1({\Bbb R}^N)$, 我们有convolution不等式

$\iint _{{\Bbb R}^N\times {\Bbb R}^N}\frac{u^2(x)u^2(y){\rm d}x{\rm d}x}{|x|^h|y|^l|x-y|^{r-h-l}}\leq C|u|^4_{L^{2p}({\Bbb R}^N)}\leq C|u|^{4-r}_{L^2({\Bbb R}^N)}|\bigtriangledown u|^r_{L^2({\Bbb R}^N)},$

这里$p=\frac{2N}{2N-r}$, 这里及以后的$C$表示各种常数.上述两不等式见文献[3].

引理2.3 设(1), (2) 成立, 则变分泛函$I_g(u)$$0<c<I^{\infty }$, 满足$(PS)_c$条件.

 首先由条件可知$I_g(u)\in C^1(H^1({\Bbb R}^N),R)$, 设$\{u_n\}_{n\geq 1}$$H^1({\Bbb R}^N)$中一序列, 适合$ I_g(u_n)\to C$, $I'_g(u_n)\to 0$$ H^{-1}({\Bbb R}^N)$.即

$ \frac{1}{2}\int _{{\Bbb R}^N}(|\bigtriangledown u_n|^2+a(x)u^{2}_n){\rm d}x -\frac{1}{4}\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y)} {|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y\\ -\int _{{\Bbb R}^N}g(x)u_n{\rm d}x=C+0(1),$ (2.5)
$\int _{{\Bbb R}^N}(\bigtriangledown u_n\bigtriangledown v+a(x)u_nv){\rm d}x -\frac{1}{2}\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u_n(y)v(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y \\ \quad -\frac{1}{2}\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u_n(x)u^2_n(y)v(x)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y-\int _{{\Bbb R}^N}g(x)v{\rm d}x =\langle \phi _n,v\rangle,$ (2.6)

这里$\phi _n\to 0$, 在$H^{-1}({\Bbb R}^N)$中.

在(2.6) 式中取$v=u_n$, 结合(2.5) 式可以导出$\{u_n\}$$H^1({\Bbb R}^N)$中有界, 从而存在弱收敛的子序列, 我们仍记其为$\{u_n\}$, 满足

$u_n\rightharpoonup u,\begin{matrix} {} & {} \\ \end{matrix}\mbox{在$ H^1({\Bbb R}^N)$中,} $
$u_n\to u,\begin{matrix} {} & {} \\ \end{matrix}\mbox{在$ {\Bbb R}^N$中几乎处处.} $

不妨设$||u_n||^2\to \mu \geq 0$, 若$\mu =0$, 由$Q(x,y)$是连续有界的, 用Convolution不等式, 可得

$ \bigg|\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y)} {|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y\bigg| \leq C|u_n|^{4-r }_{L^2({\Bbb R}^N)}|\bigtriangledown u_n|^{r}_{L^2({\Bbb R}^N)} \to 0,$

$I _{g}(u_n)=\frac{1}{2}\int _{{\Bbb R}^N}(|\bigtriangledown u_n|^2 +a(x)u^2_n){\rm d}x-\frac{1}{4}\iint_{{\Bbb R}^N\times {\Bbb R}^N} \frac{Q(x,y)u^2_n(x)u^2_n(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y\\ \quad \quad \quad -\int _{{\Bbb R}^N}u_ng(x){\rm d}x\to 0.$

这与$C\in (0,I^{\infty })$相矛盾.

$\rho _n=|\bigtriangledown u_n|^2+a(x)u^2_n$, 由前面引理知, 要证$\rho _n$是紧的, 只要排除下面两种情况.

(1) Vanishing不会发生.若不然, 则

$\lim\limits_{n\to \infty } \sup_{y\in {\Bbb R}^N}\int _{y+B_R} (|\bigtriangledown u_n|^2+a(x)u^2_n){\rm d}x=0,\begin{matrix} {} & {} \\ \end{matrix}\forall R<\infty .$

从而可得在$L^q({\Bbb R}^N)$$u_n\to 0$ ($2 <q<\frac{2N}{n-2}$).则

$\int _{{\Bbb R}^N}g(x)u_n{\rm d}x\to 0 $

$ \bigg|\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y \bigg|\leq C|u_n|^4_{L^{2p}({\Bbb R}^N)}\to 0,$

这里$p=\frac{4N}{2N-r}$, 显然$2<2p<\frac{2N}{N-2}.$

由于在$H^{-1}({\Bbb R}^N)$$I'_g(u_n)\to 0$, 从而可得

$\int _{{\Bbb R}^N}(|\bigtriangledown u_n|^2+a(x)u^2_n){\rm d}x=\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y-\int _{{\Bbb R}^N}u_ng(x){\rm d}x+0(1).$

则可以推得$I_g(u_n)\to 0$ ($n\to \infty $), 与$C>0$矛盾.

(2) Dichotomy不会发生.若不然, 设

$Q_n(t)=\sup_{y\in {\Bbb R}^N}\int _{y+B_t}(|\bigtriangledown u_n|^2+a(x)u^2_n){\rm d}x.$

由文献[4]知, 存在$Q_n(t)$的子列收敛, 在此我们任然记为$Q_n(t)$, 适合$Q_n(t)\to Q(t)$, 显然$Q(t)$$(0,\infty )$上的一个非负非减的有界函数, 则

$\lim\limits_{t\to \infty }Q(t)=\alpha \in (0,\mu ),$

所以, $\forall \epsilon >0$, $\exists R_0>0$, 当$R>R_0$$Q(R)\geq \alpha -\frac{\epsilon }{4}$.则, 当$n$充分大时有

$Q_n(R)\in (\alpha -\frac{\epsilon }{2},\alpha +\frac{\epsilon }{2}).$

选取$R_n$适当大, 使得

$Q(2R_n)\in (\alpha -\epsilon,\alpha +\epsilon ).$

$\omega,\varphi $为光滑函数, $0\leq \omega,\varphi \leq 1$,

$ \omega =0,|x|>2;\omega =1,|x|<1.$
$\varphi =1,|x|>2;\varphi =0,|x|<1.$

$\omega _n(x)=\omega (\frac{x-y_n}{R_1}),$ $\varphi _n(x)=\varphi (\frac{x-y_n}{R_n})$(其中$R_1$由后面确定).首先可得

$\bigg|\int _{{\Bbb R}^N}(\omega ^2_n|\bigtriangledown u_n|^2 -|\bigtriangledown (\omega _nu_n)|^2){\rm d}x\bigg|\leq \frac{C}{R_1},$

选取$R_1$使得$\frac{C}{R_1}<\epsilon $, 且$R_1>R_0$, 我们有

$ \bigg|\int _{{\Bbb R}^N}(\omega ^2_n|\bigtriangledown u_n|^2 -|\bigtriangledown (\omega _nu_n)|^2){\rm d}x\bigg|<\epsilon,$

$\begin{align} &\quad \iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y) \omega ^2_n(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y -\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y)\omega ^2_n(y)\omega ^2_n(x)}{|x|^h|x-y|^{r-2h}|y|^h} {\rm d}x{\rm d}y \\ &\leq \bigg[\iint_{|x-y_n|>2R_1,|y-y_n|<R_1}+\iint _{x-y_n|>2R_1,|y-y_n| \geq R_1}\\ & +\iint _{R_1<|x-y_n|\leq 2R_1,|y-y_n|\leq 2R_1}\bigg]\frac{|Q(x,y) |u^2_n(x)u^2_n(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y \\ &\leq \frac{C}{R^{r-2h}_1}\iint _{{\Bbb R}^N\times {\Bbb R}^N} \frac{u^2_n(x)u^2_n(y)}{|x|^h|y|^h}{\rm d}x{\rm d}y+C \bigg(\int _{R_1<|x-y_n|\leq 2R_1}|u_n|^{2p}{\rm d}x\bigg)^{\frac{1}{p}} \bigg (\int _{{\Bbb R}^N}|u_n|^{2p}{\rm d}y\bigg)^{\frac{1}{p}} \\ &\leq \frac{C}{R^{r-2h}_1}\bigg(\int _{B_1}\frac{u^2_n(x)}{|x|^h}{\rm d}x +\int _{{\Bbb R}^N}u^2_n{\rm d}x\bigg)^2+C(Q_n(2R_1)-Q_n(R_1))^{2r}. \end{align}$ (2.7)

显然

$\int _{B_1}\frac{u^2_n(x)}{|x|^h}{\rm d}x\leq \bigg(\int _{B_1}|u_n|^{\frac{2N}{N-2}}{\rm d}x\bigg)^{\frac{N-2}{N}} \bigg(\int _{B_1}\frac{{\rm d}x}{|x|^{\frac{hN}{2}}}\bigg)^{\frac{2}{N}}\leq C ,$

所以, 结合(2.7) 式, 我们可以得到

$\begin{align}&\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y) \omega ^2_n(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y\\ & -\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y) \omega ^2_n(y)\omega ^2_n(x)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y \leq \mu (\epsilon ).\end{align}$ (2.8)

同理可得

$\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y)\varphi ^2_n(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y\\ -\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2_n(x)u^2_n(y)\varphi ^2_n(y)\varphi ^2_n(x)}{|x|^h|x-y|^{r-2h}|y|^h} {\rm d}x{\rm d}y \leq\mu (\epsilon ).$ (2.9)

这里及以后, $\mu (\epsilon )$表示各种小量, 具有性质$\epsilon \to 0$时, $\mu (\epsilon )\to 0.$

$u^{(1)}_n=\omega _nu_n$, $u^{(2)}_n=\varphi _nu_n$.则我们可以推得

$I_g(u_n)\geq I_g(u^{(1)}_n)+I_g(u^{(2)}_n)-\mu (\epsilon ). $

利用(2.7), (2.8) 和(2.9) 式, 结合$\{u_n\}$$(PS)_c$, 我们可得$n\to \infty $时, $I_g(u_n)>I^{\infty }$与条件$0<c<I^{\infty }$矛盾, 即Dichotomy不会发生.从而引理得证.

3 主要定理的证明

在本节, 我们将证明本文的主要结果.

定义3.1

$\overline{B_R}=\{u\in H^1({\Bbb R}^N)\ \big|\ \|u\|\leq R\}$
$I_0=I_0(R)=\inf_{u\in \overline{B_R}}I_g(u).$

定理3.1 假设$a(x)$, $Q(x,y)$适合条件(1), (2), 且在$H^{-1}({\Bbb R}^N)$$g(x)\geq 0$, $g(x)\not\equiv 0$则存在常数$C,R$, 当$\|g\|_{H^{-1}}\leq C$

(a)当$u∈ \overline{B_R}= \{u\in H^1({\Bbb R}^N)\ \big|\ \|u\|=R\}$时, $I_g(u)\geq \delta ||u||^2$($\delta >0$为常数);

(b)存在$u_0\in H^1({\Bbb R}^N)$使得$I_g(u_0)=I_0=\inf\limits_{u\in \overline{B_R}}I_g(u)<0$, 且$u_0$是问题(1.1) 的一个解.

 由定义

$ I _{g}(u)=\frac{1}{2}\int _{{\Bbb R}^N}(|\bigtriangledown u|^2+a(x)u^2){\rm d}x-\frac{1}{4}\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2(x)u^2(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y-\int _{{\Bbb R}^N}ug(x){\rm d}x.$

考虑$a(x),Q(x,y)$适合的条件, 结合convolution不等式知

$\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2(x)u^2(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y\leq C|u|^{4-r}_{L^2}|\bigtriangledown u|^2_{L^2}\leq C||u||^4. $

运用Holder不等式, 有

$\int _{{\Bbb R}^N}g(x)u{\rm d}x\leq ||g||_{L^2}||u||. $

从而可得

$\begin{align}&I_g(u)\geq \frac{1}{2}||u||^2-C||u||^4-||g||_{L^2}||u|| \\&\quad \quad =\frac{1}{2}||u||(||u||-2C||u||^3-||g||_{L^2}) \\ &\quad \quad \geq \frac{1}{2}||u||(-2C||u||^3-||g||_{L^2}).\end{align}$ (3.1)

首先选取适当的$R$, 当$||u||=R$时, $1-2CR^2\geq \frac{1}{2}$, 然后可以选取$\overline{C}$, 使得$||g||_{L^2}\leq \overline{C}<\frac{||u||}{4}$, 由(3.1) 式, 可得$||u||=R$时, 存在$\delta >0$, 适合

$I_g(u)\geq \delta ||u||^2.$

即(a)得证.

由于$I_g(u)$是连续泛函, 由(a)及$I_g(0)=0$, 结合(3.1) 式可知$I_g(u)$$\overline{B_R}$上的下确界存在, 返回到$I_g(u)$的定义知

$\begin{align}&I _{g}(u)=\frac{1}{2}\int _{{\Bbb R}^N}(|\bigtriangledown u|^2+a(x)u^2){\rm d}x-\frac{1}{4}\iint_{{\Bbb R}^N\times {\Bbb R}^N}\frac{Q(x,y)u^2(x)u^2(y)}{|x|^h|x-y|^{r-2h}|y|^h}{\rm d}x{\rm d}y-\int _{{\Bbb R}^N}ug(x){\rm d}x \\ &\quad \quad \leq \frac{1}{2}\int _{{\Bbb R}^N}(|\bigtriangledown u|^2+a(x)u^2){\rm d}x-\int _{{\Bbb R}^N}ug(x){\rm d}x.\end{align}$ (3.2)

由于$g(x)\geq 0$, 且$g(x)\not\equiv 0$, 则存在$v\in H^1({\Bbb R}^N),v\geq 0$使得$\int_{{\Bbb R}^N}g(x)v{\rm d}x>0$.又

$I_{g}(tv)\leq t \bigg(\frac{t}{2}\int(|\bigtriangledown v|^2+a(x)v^2){\rm d}x-\int g(x)v{\rm d}x \bigg),$

从而可以选取适当小的$t$, 使得$I_g(tv)<0$.

也就是$I_g(u)$$\overline{B_R}$上的下确界小于0, 则由$I_g(u)$的弱下半连续性得证结论(b), 定理得证.

为了完成本文的主要定理得证明, 我们考虑$\overline{u}\in H^1({\Bbb R}^N)$为(2.4) 式给出, 结合$I_g(u)$的定义可知, 存在适当大的$\overline{t}$, 当$t\geq \overline{t}$时, 有$I_g(t\overline{u})<0$.

定义3.2

$\Gamma =\{\gamma \in C([0,1],H^1({\Bbb R}^N)):\gamma (0)=0,\gamma(1)=\overline{t}\overline{u} \},$ (3.3)
$C_g=\inf_{\gamma\in \Gamma}\sup_{u\in \gamma }I_g(u).$ (3.4)

定理3.2 假设条件$a(x)$, $Q(x,y)$适合条件(1), (2), 且在$H^{-1}({\Bbb R}^N)$$g(x)\geq 0$, $g(x)\not\equiv 0$则存在常数$C,R$, 当$||g||_{H^{-1}}\leq C$时, 存在$u_1\in H^1({\Bbb R}^N)$使得$I_g(u_1)=C_g$, 且$I'_g(u_1)=0$.即$u_1$是问题的解.

 首先由条件及$C_g$的定义, 可以推得$0<C_g<I^{\infty }$.由Ekeland变分原理知, 存在$\{u_n\}\in H^1({\Bbb R}^N)$适合

$I_g(u_n)\to C_g,I'_g(u_n)\to 0.$ (3.5)

$\{u_n\}$$I_g(u)$$(PS)_c$序列, 结合引理, 可知存在$u_1\in H^1({\Bbb R}^N)$, 在$H^1({\Bbb R}^N)$中有$\{u_n\}$的子序列强收敛到$u_1$.从而

$I_g(u_1)= C_g,I'_g(u_1)=0.$ (3.6)

$u_1$是问题的一个解.

最后回到本文定理1.1的证明:由定理3.1与定理3.2知, 问题(1.1) 有两个解$u_0$, $u_1$, 且$I_g(u_0)<0<I_g(u_1)$, 从而完成本文结论的证明.

参考文献
[1] Ambrosetti A, Badiale M. Homoclinics:Poincare-Melnikov type results via a variational approach. Ann Inst H Poincare Analyse Non Lineaire, 1998, 15: 233–252. DOI:10.1016/S0294-1449(97)89300-6
[2] Lieb E H. Existence and uniqueness of the minimizing solution of Choquard's nonlinear equation. Stud Appl Math, 1997, 57: 93–105.
[3] Lieb E H. Sharp constants in the Hardy-Littlewood-Sobolev inequality and related inequalities. Ann Math, 1983, 118: 349–374. DOI:10.2307/2007032
[4] Lions P L. The concentration-compactness principle in calculus of variations, the locally case, partI, partⅡ. Ann I H Anal Nonli, 1984, 1: 109–145, 223-283.
[5] Stuart C A. Two positive solutions of a quasilinear elliptic dirichlet problem. Milan J Math, 2011, 79: 327–341. DOI:10.1007/s00032-011-0140-0
[6] Stuart C A, Zhou H S. Existence of guided cylindrical TM-modes in an inhomogeneous self-focusing Dielectric. Math Models Meth Appl Sci, 2010, 20: 1681–1719. DOI:10.1142/S0218202510004751
[7] Li Y, Peng S J. Multiple solutions to an elliptic problem related to vortex pairs. J Differential Equations, 2011, 250: 3448–3472. DOI:10.1016/j.jde.2010.09.023
[8] 曹道民. 广义Choquard-Pekar方程非平凡解的存在性. 数学物理学报, 1989, 9A(1): 101–112.
Cao D M. The existence of nontrivial solutions to general Choquard-Pekar equation. Acta Math Sci, 1989, 9A(1): 101–112.
[9] 赵越. ${{\mathbb{R}}^{N}}$上一类拟线性椭圆方程解的存在性[D]. 武汉: 华中师范大学硕士学位论文, 2013
Zhao Y. The Existence of Solutions for a Class of Quasilinear Elliptic Equations on ${{\mathbb{R}}^{N}}$ [D]. Wuhan:Central China Normal University, 2013
[10] 张正杰, 张莹. ${{\mathbb{R}}^{N}}$上拟线性椭圆型方程两非负解的存在性. 数学物理学报, 2015, 35A(2): 225–233.
Zhang Z J, Zhang Y. Two non-negative solutions of a quasilinear elliptic equation on ${{\mathbb{R}}^{N}}$. Acta Math Sci, 2015, 35A(2): 225–233.