\documentclass[reqno]{amsart} \usepackage{hyperref} \usepackage{amssymb} \AtBeginDocument{{\noindent\small \emph{Electronic Journal of Differential Equations}, Vol. 2012 (2012), No. 169, pp. 1--18.\newline ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu \newline ftp ejde.math.txstate.edu} \thanks{\copyright 2012 Texas State University - San Marcos.} \vspace{9mm}} \begin{document} \title[\hfilneg EJDE-2012/169\hfil Existence of solutions] {Existence of solutions to quasilinear elliptic systems with combined critical Sobolev-Hardy terms} \author[N. Nyamoradi, M. Javidi \hfil EJDE-2012/169\hfilneg] {Nemat Nyamoradi, Mohamad Javidi} \address{Nemat Nyamoradi \newline Department of Mathematics, Faculty of Sciences\\ Razi University, 67149 Kermanshah, Iran} \email{nyamoradi@razi.ac.ir, neamat80@yahoo.com} \address{Mohamad Javidi \newline Department of Mathematics, Faculty of Sciences\\ Razi University, 67149 Kermanshah, Iran} \email{mo\_javidi@yahoo.com} \thanks{Submitted May 17, 2012. Published October 4, 2012} \subjclass[2000]{35B33, 35J60, 35J65} \keywords{Ekeland variational principle; critical Hardy-Sobolev exponent; \hfill\break\indent concentration-compactness principle} \begin{abstract} This article is devoted to the study of multiple positive solutions to a singular elliptic system where the nonlinearity involves a combination of concave and convex terms. Using the effect of the coefficient of the critical nonlinearity, and a variational method, we establish the main result which is based on a compactness argument. \end{abstract} \maketitle \numberwithin{equation}{section} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} %\newtheorem{remark}[theorem]{Remark} \allowdisplaybreaks \section{Introduction} The aim of this paper is to establish the existence of nontrivial solution to the elliptic system \begin{equation}\label{1} \begin{gathered} - \Delta_p u - \mu \frac{|u|^{p - 2} u}{|x|^p} = \frac{|u|^{p^* (s_1) - 2} u}{|x|^{s_1}} + \frac{\alpha}{\alpha + \beta} Q (x ) \frac{ |u|^{\alpha - 2} |v|^\beta u}{|x - x_0|^t} + \lambda h (x) \frac{ |u|^{q - 2} u}{|x|^s}, \\ % \quad x \in \Omega,\\ - \Delta_p v - \mu \frac{|v|^{p - 2} v}{|x|^p} = \frac{|v|^{p^* (s_2) - 2} v}{|x|^{s_2}} + \frac{\beta}{\alpha + \beta} Q (x ) \frac{ |u|^\alpha |v|^{\beta -2} v}{|x- x_0|^t} + \lambda h (x) \frac{ |v|^{q - 2} v}{|x|^s},\\ x \in \Omega,\\ u = v = 0, \quad x \in \partial \Omega \end{gathered} \end{equation} where $\Delta_p u = \operatorname{div} (|\nabla u|^{p - 2} \nabla u)$, $0 \in \Omega$ is a bounded domain in $\mathbb{R}^N$ ($N \geq 3$) with smooth boundary $\partial \Omega$, $\lambda > 0$ is a parameter, $1 \leq q < p$, $1 < p < N$, $0 \leq \mu < \overline{\mu} \triangleq \big(\frac{N - p}{p} \big)^p$; $Q (x)$ is nonnegative and continuous on $\overline{\Omega}$ satisfying some additional conditions which will be given later, $Q (x_0) = \|Q\|_\infty$ for $0 \ne x_0 \ne \Omega$, $h (x) \in C (\overline{\Omega})$; $\alpha, \beta > 1$, $\alpha + \beta = p^*(t) \triangleq \frac{p (N - t)}{N - p}$, $p^*(s) \triangleq \frac{p (N - s)}{N - p}$ ($0 < s, s_1, s_2 \leq t < p $) are critical Sobolev-Hardy exponents. Note that $ p^*(0) = p^* : = \frac{N p}{N - p}$ is the critical Sobolev exponent. We denote by $W_0^{1,p} (\Omega)$ the completion of $C_0^\infty (\Omega)$ with respect to the norm $\big(\int_\Omega |\nabla \cdot|^p dx \big)^{1/p}$. Problem \eqref{1} is related to the well known Caffarelli-Kohn-Nirenberg inequality in \cite{L. Caffarelli}: \begin{equation} \label{ine1} \Big(\int_\Omega \frac{|u|^r}{|x|^t } dx \Big)^{p/r} \leq C_{r,t, p} \int_\Omega |\nabla u|^p dx, \quad \text{for all } u \in W_0^{1,p} (\Omega), \end{equation} where $p \leq r < p^*(t)$. When $t = r = p$, the above inequality becomes the well known Hardy inequality \cite{L. Caffarelli,J. Garcia Azorero,N. Ghoussoub}: \begin{equation} \label{ine11} \int_\Omega \frac{|u|^p}{|x|^p } dx \leq \frac{1}{\overline{\mu}} \int_\Omega |\nabla u|^p dx, \quad \text{for all } u \in W_0^{1,p} (\Omega). \end{equation} In the space $W_0^{1,p} (\Omega)$ we use the norm \begin{align*} \|u\|_\mu = \|u\|_{D^{1,p} (\Omega)} : = \Big(\int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } \Big) dx \Big)^{1/p}, \quad \mu \in [0, \overline{\mu}). \end{align*} By using the Hardy inequality \eqref{ine11} this norm is equivalent to the usual norm $\big(\int_\Omega |\nabla u|^p dx \big)^{1/p}$. The elliptic operator $L := \big(|\nabla \cdot|^{p - 2} \nabla \cdot - \mu \frac{|\cdot|^{p - 2}}{|x|^p } \big)$ is positive in $W_0^{1,p} (\Omega)$ if $0 \leq \mu < \overline{\mu}$. Now, we define the space $W = W_0^{1,p}(\Omega) \times W_0^{1,p}(\Omega)$ with the norm \[ \|(u, v)\|^p = \|u\|_\mu^p + \|v\|_\mu^p. \] Also, by Hardy inequality and Hardy-Sobolev inequality, for $0 \leq \mu < \overline{\mu}$, $0 \leq t < p$ and $p \leq r \leq p^*(t)$ we can define the best Hardy-Sobolev constant: \begin{equation} \label{hardy} A_{\mu, t, r} (\Omega) = \inf_{u \in W_0^{1,p}(\Omega) \setminus \{0\}} \frac{\int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } \Big) dx}{\Big(\int_\Omega \frac{|u|^r}{|x|^t} dx \Big)^{p/r}}. \end{equation} In the important case when $r = p^*(t)$, we simply denote $A_{\mu, t, p^*(t)}$ as $A_{\mu, t}$. For any $0 \leq \mu < \overline{\mu}$, $\alpha, \beta > 1$ and $\alpha + \beta = p^* (t)$, by \eqref{ine1}, \eqref{ine11}, $0 < s_1, s_2 \leq t < p$, Set \begin{gather} A_{\mu, s} := \inf_{u \in W_0^{1,p} (\Omega) \setminus \{0\}} \frac{\int_\Omega \big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } \big) dx}{\big(\int_\Omega \frac{|u|^{p^* (s)}}{|x|^s} dx \big)^{\frac{p}{p^* (s)}}}\,, \label{san} \\ S_{s,\alpha, \beta} := \inf_{(u, v) \in W \setminus \{(0, 0)\}} \frac{\int_\Omega \big(|\nabla u|^p + |\nabla v|^p - \mu \frac{|u|^p + |v|^p}{|x|^p } \big) dx}{\big(\int_\Omega \frac{|u|^\alpha |v|^\beta}{|x|^s} dx \big)^{\frac{p}{\alpha + \beta}}}\,. \label{hardy2} \end{gather} Then we have the following equality (whose proof is the same as that of Theorem 5 in \cite{C. Alves}) \begin{align*} S_{s,\alpha, \beta} (\mu) = \Big(\big (\frac{\alpha}{\beta} \big)^\frac{\beta}{\alpha + \beta} + \big (\frac{\beta}{\alpha} \big)^\frac{\alpha}{\alpha + \beta} \Big)A_{\mu, s}\,. \end{align*} Throughout this paper, let $R_0$ be the positive constant such that $\Omega \subset B(0; R_0)$, where $B(0; R_0) = \{x \in \mathbb{R}^N : |x| < R_0 \}$. By Holder and Sobolev-Hardy inequalities, for all $u \in W^{1,p}_ 0 (\Omega)$, we obtain \begin{equation} \label{hold} \begin{split} \int_\Omega \frac{ |u|^q}{|x|^s} & \leq \Big(\int_{B(0; R_0)} |x|^{-s} \Big)^\frac{p^* (s) - q}{p^* (s)} \Big(\int_\Omega \frac{ |u|^{p^*(s)}}{|x|^s} \Big)^\frac{q}{p^* (s)} \\ &\leq \Big(\int_0^ {R_0} r^{N -s + 1} dr \Big)^\frac{p^* (s) - q}{p^* (s)} A_{\mu, s}^{-\frac{q}{p}} \|u\|^q \\ &\leq \Big(\frac{N \omega_N R_0^{N - s}}{N - s} \Big)^\frac{p^* (s) - q}{p^* (s)} A_{\mu, s}^{-\frac{q}{p}} \|u\|^q, \end{split} \end{equation} where $\omega_N = \frac{2 \pi^{N/2}}{N \Gamma(N/2)}$ is the volume of the unit ball in $\mathbb{R}^N$ . Existence of nontrivial non-negative solutions for elliptic equations with singular potentials were recently studied by several authors, but, essentially, only with a solely critical exponent. We refer, e.g., in bounded domains and for $p = 2$ to \cite{D. Cao, N. Ghoussoub,N. Ghoussoub1, Han1, Han2, Han4, Han3}, and for general $p > 1$ to \cite{ D. Cao1, Degiovanni, F. Demengel, R. Filippucci, N. Ghoussoub2, P. Han, Kang, S. Peng, Xuan1} and the references therein. For example, Han and Liu \cite{Han3} studied the problem \begin{equation}\label{ne2} \begin{gathered} - \Delta u - \mu \frac{ u}{|x|^2} = \lambda u + Q (x ) |u|^{2^* - 2} u, \quad x \in \Omega,\\ u = 0, \quad x \in \partial \Omega \end{gathered} \end{equation} where $0 \in \Omega$ is a smooth bounded domain in $\mathbb{R}^N$ ($N \geq 5$), $\lambda > 0$, $0 \leq \mu < \overline{\mu} \triangleq \big(\frac{N - 2}{2} \big)^2$, $2^* = \frac{2 N}{N -2}$ and $Q (x)$ is nonnegative and continuous on $\overline{\Omega}$ satisfying some suitable conditions. using critical point theory, the authors proved the existence of nontrivial solutions to problem \eqref{ne2}. Also, by investigating the effect of the coefficient $Q$, Han \cite{Han2} studied problem \eqref{ne2} and proved that there exists $\lambda_0 > 0$ such that \eqref{ne2} has at least $k$ positive solutions for $\lambda \in (0, \lambda_0)$. Kang in \cite{Kang} studied the following elliptic equation via the generalized Mountain-Pass theorem \cite{P. Rabinowitz}, \begin{equation}\label{ne1} \begin{gathered} - \Delta_p u - \mu \frac{|u|^{p - 2} u}{|x|^p} = \frac{|u|^{p^* (t) - 2} u}{|x|^t} + \lambda \frac{ |u|^{p - 2} u}{|x|^s}, \quad x \in \Omega,\\ u = 0, \quad x \in \partial \Omega \end{gathered} \end{equation} where $\Omega \subset \mathbb{R}^N$ is a bounded domain, $1 < p < N$, $0 \leq s, t < p$ and $0 \leq \mu < \overline{\mu} \triangleq \big(\frac{N - p}{p} \big)^p$. Degiovanni and Lancelotti \cite{Degiovanni} studied problem \eqref{ne1} with $\mu = s = t = 0$ and proved that \eqref{ne1} has at least one positive solutions for $\lambda \geq \lambda_1 : = A_{0,0}$ ($A_{0,0}$ is defined in \eqref{san}). Indeed, in \cite{Degiovanni} the much more difficult case $\lambda \geq \lambda_1$ is treated. The authors in \cite{R. Filippucci}, via the Mountain-Pass Theorem of Ambrosetti and Rabinowitz \cite{A. Ambrosetti}, proved that \begin{equation*} - \Delta_p u - \mu \frac{u^{p - 1} }{|x|^p} = |u|^{p^* - 1} + \frac{u^{p^* (s) - 1}}{|x|^s} , \quad \text{in } \mathbb{R}^N, \end{equation*} admits a positive solution in $\mathbb{R}^N$, whenever $\mu < \overline{\mu} \triangleq \big(\frac{N - p}{p} \big)^p$ and $0 < s < p$. Recently, in \cite{Xuan1} the author studied the following equation via the Mountain-Pass theorem, \begin{equation*} - \operatorname{div} \Big(\frac{|D u|^{p - 2} D u}{|x|^{a p}} \Big) - \mu \frac{|u|^{p - 2} u}{|x|^{(a + 1)p}} = \frac{|u|^{p^* (b) - 2} u}{|x|^{b p^*}} + \frac{|u|^{p^* (c) - 2} u}{|x|^{c p^*}}, \quad \text{in } \mathbb{R}^N \end{equation*} where $1 < p < N$, $0 \leq \mu < \overline{\mu} \triangleq \big(\frac{N - (a + 1) p}{p} \big)^p$, $0 \leq a < \frac{N - p}{p}$, $a \leq b, c < a + 1$, $p^* (b) = \frac{N p}{N - (a + 1 - b) p}$ and $p^* (c) = \frac{N p}{N - (a + 1 - c) p}$. Zhang and Wei \cite{J. Zhang} studied the existence of multiple positive solutions for \eqref{1} with $t = s = 0$, $Q(x) = f (x)$ and $h (x) = 1$. Set $s_1 = s_2 = t$, $s = t$, $x_0 = 0$ and $Q(x) = h (x) \equiv 1$, then problem \eqref{1} reduces to the quasilinear elliptic system \begin{equation}\label{1nia} \begin{gathered} - \Delta_p u - \mu \frac{|u|^{p - 2} u}{|x|^p} = \frac{|u|^{p^* (t) - 2} u}{|x|^t}+ \frac{\eta \alpha}{\alpha + \beta} \frac{ |u|^{\alpha - 2} |v|^\beta u}{|x|^t} + \lambda \frac{ |u|^{q - 2} u}{|x|^s},\\ - \Delta_p v - \mu \frac{|v|^{p - 2} v}{|x|^p} = \frac{|v|^{p^* (t) - 2} v}{|x|^t} + \frac{\eta \beta}{\alpha + \beta} \frac{ |u|^\alpha |v|^{\beta -2} v}{|x|^t} + \theta \frac{ |v|^{q - 2} v}{|x|^s},\\ x \in \Omega,\\ u = v = 0, \quad x \in \partial \Omega \end{gathered} \end{equation} where $\lambda > 0$, $\theta > 0$, $0 < \eta < \infty$, $1 < p < N$, $0 \leq \mu < \overline{\mu} \triangleq \big(\frac{N - p}{p} \big)^p$, $0 \leq s, t < p$, $1 \leq q < p$, $\alpha + \beta = p^*(t) \triangleq \frac{p (N - t)}{N - p}$ is the Hardy- Sobolev critical exponent. The author \cite{Nia1} have studied \eqref{1nia} via the Nehari manifold. In \cite{Y. Li}, Li et al. studied the following quasilinear elliptic problem \begin{equation}\label{li} \begin{gathered} - \Delta_p u - \mu \frac{|u|^{p - 2} u}{|x|^p} = K (x ) \frac{|u|^{p^* (s) - 2} u}{|x|^{s}} + Q (x ) \frac{ |u|^{p^* (t) - 2}}{|x - x_0|^t} + \lambda f (x, u), \quad x \in \Omega,\\ u = 0, \quad x \in \partial \Omega \end{gathered} \end{equation} where $1 < p < N$, $K (x), Q (x)$ are nonnegative continuous functions on $\overline{\Omega}$, $f$ satisfying some suitable conditions and obtained the existence of solutions via variational methods. For $p = 2$, $x_0 = 0$, $ K (x ) \equiv 1$ and $ Q (x ) \equiv 0$, the problem \eqref{li} has been studied. Motivated by the above works we study problem \eqref{1} by using the Mountain-Pass Theorem of Ambrosetti and Rabinowitz. We shall show that system \eqref{1} has at least two positive weak solutions. In this article, we assume that $0 < s_1, s_2 \leq t < p$, $\alpha, \beta > 1$ and $\alpha + \beta = p^*(t)$. For $0 \leq \mu < \overline{\mu} $, we set \begin{gather*} \theta (\mu, s) : = \frac{p - s}{p (N - s)} A_{\mu, s}^\frac{N -s}{p - s}, \\ \theta^* : = \big\{\theta (\mu, s_1), \theta (\mu, s_2), \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p- t}} S_{t, \alpha, \beta}^\frac{N - t}{p - t} \big\}. \end{gather*} Moreover, we assume that $Q (x)$ satisfies some of the following assumptions: \begin{itemize} \item[(H1)] $Q \in C(\overline{\Omega})$, $Q (x) \geq 0$ and $\text{meas}(\{x \in \Omega, \; h (x) > 0 \}) > 0$. \item[(H2)] There exist $\vartheta > 0$ such that $Q (x_0) = \|Q\|_\infty > 0$ and $Q (x) = Q (x_0) + O(|x - x_0|^\varrho)$, as $x \to x_0$. \item[(H3)] There exist $\beta_0$ and $\rho > 0$ such that $B_{2 \rho_0} (x_0) \subset \Omega$ and $h (x) \geq \beta_0$ for all $x \in B_{2 \rho_0} (x_0)$. \end{itemize} Set \begin{align*} h_+ : = \max \{h, 0 \}, \quad h_- : = \max \{-h, 0 \}. \end{align*} The main results of this article are stated in the following two theorems. \begin{theorem}\label{thmA} Assume that $N \geq 3, \mu \in [0,\overline{\mu}),\;1 0$, such that for $0 < \lambda < \Lambda_{11}^*$ problem \eqref{1} has at lest one positive solutions. \end{theorem} \begin{theorem}\label{the1} Assume that $N \geq p^2$, $0 \leq \mu < \overline{\mu}$, $\theta^* = \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p- t}} S_{t, \alpha,\beta}^\frac{N - t}{p - t}$, (H1)-(H3), $Q (0) = 0$, $\varrho > b (\mu) p + p - N + t$ and $\frac{N -s}{b (\mu)} < q < p$ hold, and $b(\mu)$ is the constant defined as in Lemma \ref{lem kang}. Then there exists $\Lambda^{**}> 0$, such that for $0 < \lambda < \Lambda^{**}$, problem \eqref{1} has at least two positive solutions. \end{theorem} This article is divided into three sections, organized as follows. In Section 2, we establish some elementary results. In Section 3, we prove our main results (Theorems \ref{thmA} and \ref{the1}). \section{Preliminary lemmas} The corresponding energy functional of problem \eqref{1} is defined by \begin{align*} J (u, v) &= \frac{1}{p} \int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } + |\nabla v|^p - \mu \frac{|v|^p}{|x|^p } \Big) d x - \frac{\lambda}{q} \int_\Omega h (x) (\frac{|u|^q}{|x|^s} + \frac{|v|^q}{|x|^s}) dx\\ &\quad - \frac{1}{p^*(s_1)} \int_\Omega \frac{|u|^{p^*(s_1)} }{|x|^{s_1}} dx - \frac{1}{p^*(s_2)} \int_\Omega \frac{|v|^{p^*(s_2)} }{|x|^{s_2}} dx \\ &\quad - \frac{1}{\alpha + \beta} \int_\Omega Q (x) \frac{|u|^\alpha |v|^\beta }{|x - x_0|^t} dx, \end{align*} for each $(u, v) \in W$. Then $J \in C^1 (W, \mathbb{R})$. \begin{lemma}\label{ps0} Assume that $N \geq 3$, $0 \leq \mu < \overline{\mu}$, (H1), $h_+\ne 0$ and $(u, v)$ is a weak solution of problem \eqref{1}. Then there exists a positive constant $d$ depending on $N, |\Omega|, |h_+|_\infty, A_{\mu, s}, s_1, s_2$ and $q$ such that \[ J(u, v) \geq - d \lambda^\frac{p}{p - q}. \] \end{lemma} \begin{proof} Since $(u, v)$ is a weak solution of \eqref{1}, then, Note that $\langle J'(u, v), (u, v) \rangle = 0$, we have \begin{equation} \label{j0} \begin{split} &\langle J'(u, v), (u, v) \rangle\\ & = \int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } + |\nabla v|^p - \mu \frac{|v|^p}{|x|^p } \Big) d x - \lambda \int_\Omega h (x) (\frac{|u|^q}{|x|^s} + \frac{|v|^q}{|x|^s}) dx \\ &\quad - \int_\Omega \frac{|u|^{p^*(s_1)} }{|x|^{s_1}} dx - \int_\Omega \frac{|v|^{p^*(s_2)} }{|x|^{s_2}} dx - \int_\Omega Q (x) \frac{|u|^\alpha |v|^\beta }{|x - x_0|^t} dx = 0. \end{split} \end{equation} Now, by using $h_+ \ne 0$, \eqref{j0}, \eqref{hold}, the H\"{o}lder inequality and the Sobolev-Hardy inequality, we have \begin{align*} J (u, v) & \geq \frac{1}{p} \int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } + |\nabla v|^p - \mu \frac{|v|^p}{|x|^p } \Big) d x - \frac{\lambda}{q} \int_\Omega h (x) (\frac{|u|^q}{|x|^s} + \frac{|v|^q}{|x|^s}) dx\\ &\quad - \frac{1}{p^*(t)} \Big[ \int_\Omega \frac{|u|^{p^*(s_1)} }{|x|^{s_1}} dx - \int_\Omega \frac{|v|^{p^*(s_2)} }{|x|^{s_2}} dx - \int_\Omega Q (x) \frac{|u|^\alpha |v|^\beta }{|x - x_0|^t} dx \Big] \\ & = \frac{1}{p} \int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } + |\nabla v|^p - \mu \frac{|v|^p}{|x|^p } \Big) d x - \frac{\lambda}{q} \int_\Omega h (x) (\frac{|u|^q}{|x|^s} + \frac{|v|^q}{|x|^s}) dx\\ &\quad - \frac{1}{p^*(t)} \Big[ \int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } + |\nabla v|^p - \mu \frac{|v|^p}{|x|^p } \Big) d x \\ &\quad - \lambda \int_\Omega h (x) (\frac{|u|^q}{|x|^s} + \frac{|v|^q}{|x|^s}) dx \Big]\\ & \geq \Big (\frac{1}{p} - \frac{1}{p^*(t)} \Big) \int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } + |\nabla v|^p - \mu \frac{|v|^p}{|x|^p } \Big) d x \\ &\quad - \lambda \Big (\frac{1}{q} - \frac{1}{p^*(t)} \Big ) \int_\Omega h (x) (\frac{|u|^q}{|x|^s} + \frac{|v|^q}{|x|^s}) dx \\ & \geq \Big (\frac{1}{p} - \frac{1}{p^*(t)} \Big) (\|u\|_\mu^p + \|v\|_\mu^p) \\ &\quad - \lambda \Big (\frac{1}{q} - \frac{1}{p^*(t)} \Big) \Big(\frac{N \omega_N R_0^{N - s}}{N - s} \Big)^\frac{p^* (s) - q}{p^* (s)} A_{\mu, s}^{-\frac{q}{p}} |h_+|_\infty(\|u\|_\mu^q + \|v\|_\mu^q) \\ & \geq 2 \inf_{t \geq 0} \Big[\Big (\frac{1}{p} - \frac{1}{p^*(s)} \Big) t^p - \lambda \Big (\frac{1}{q} - \frac{1}{p^*(s)} \Big) \Big(\frac{N \omega_N R_0^{N - s}}{N - s} \Big)^\frac{p^* (s) - q}{p^* (s)} A_{\mu, s}^{-\frac{q}{p}} |h_+|_\infty t^q \Big]\\ & \geq - d \lambda^\frac{p}{p - q}. \end{align*} Here $d_\Omega : = \sup_{x, y \in \Omega} |x - y|$ is the diameter of $\Omega$ and $d$ is a positive constant depending on $N, |\Omega|, |h_+|_\infty, A_{\mu, s}, s_1, s_2$ and $q$. \end{proof} Recall that a sequence $(u_n, v_n)_{n \in \mathbb{N}}$ is a $(PS)_c$ sequence for the functional $J$ if $J (u_n, v_n) \to c$ and $J' (u_n, v_n) \to 0$. If any $(PS)_c$ sequence $(u_n, v_n)_{n \in \mathbb{N}}$ has a convergent subsequence, we say that $J$ satisfies the $(PS)_c$ condition. \begin{lemma}\label{ps2} Assume that $N \geq 3$, $0 \leq \mu < \overline{\mu}$, {\rm (H1)}, $h_+ \ne 0$ and $Q (0) = 0$. Then $J(u, v)$ satisfies the $(PS)_c$ condition with $c$ satisfying \begin{equation} \label{c} \begin{split} c < c_* : = \min \Big\{&\frac{p - s_1}{p (N - s_1)} A_{\mu, s_1}^\frac{N - s_1}{p - s_1}, \frac{p - s_2}{p (N - s_2)} A_{\mu, s_2}^\frac{N - s_2}{p - s_2}, \\ & \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p- t}} S_{t, \alpha, \beta}^\frac{N - t}{p - t} \Big\} - d \lambda^\frac{p}{p - q}. \end{split} \end{equation} \end{lemma} \begin{proof} It is easy to see that the $(PS)_c$ sequence $(u_n, v_n)_{n \in \mathbb{N}}$ of $J (u, v)$ is bounded in $W$. Then $(u_n, v_n) \rightharpoonup (u, v)$ weakly in $W$ as $n \to \infty$, which implies $u_n \rightharpoonup u$ weakly and $v_n \rightharpoonup v$ weakly in $W^{1,p}_0 (\Omega)$ as $n \to \infty$. Passing to a subsequence we may assume that \begin{gather*} |\nabla u_n|^p dx \rightharpoonup \overline{\alpha}, \quad |\nabla v_n|^p dx \rightharpoonup \widetilde{\alpha}, \\ \frac{|u_n|^p}{|x|^p} dx \rightharpoonup \overline{\beta}, \quad \frac{|v_n|^p}{|x|^p} dx \rightharpoonup \widetilde{\beta}, \\ \frac{|u_n|^{p^*(s_1)}}{|x|^{s_1}} dx \rightharpoonup \overline{\gamma}, \quad \frac{|v_n|^{p^*(s_2)}}{|x|^{s_2}} dx \rightharpoonup \widetilde{\gamma}, \\ Q(x) \frac{|u_n|^\alpha |v_n|^\beta }{|x - x_0|^t} dx \rightharpoonup \nu \end{gather*} weakly in the sense of measures. Using the concentration-compactness principle in \cite{P.L. Lions}, there exist an at most countable set $I$, a set of points $\{x_i \}_{i \in I} \in \Omega \setminus \{ 0 \}$, real numbers $\overline{a}_{x_i} , \widetilde{a}_{x_i} , d_{x_i}$, $i \in I$, $\overline{a}_0, \widetilde{a}_0, \overline{b}_0, \widetilde{b}_0, \overline{c}_0, \widetilde{c}_0$ and $d_0$, such that \begin{gather} \overline{\alpha} \geq |\nabla u|^p dx +\sum_{i \in I}\overline{a}_{x_i}\delta_{x_i}+\overline{a}_0\delta_0, \label{n1}\\ \widetilde{\alpha} \geq |\nabla u|^p dx +\sum_{i \in I}\widetilde{a}_{x_i}\delta_{x_i}+\widetilde{a}_0\delta_0, \label{n2}\\ \overline{\beta} = \frac{|u|^p} {|x|^p} dx + \overline{b}_0\delta_0, \label{n3}\\ \widetilde{\beta} = \frac{|v|^p} {|x|^p} dx + \widetilde{b}_0\delta_0, \label{n4}\\ \overline{\gamma}= \frac{|u|^{p^*(s_1)}}{|x|^{s_1}}+\overline{c}_0\delta_0, \label{n5}\\ \widetilde{\gamma} = \frac{|v|^{p^*(s_2)}}{|x|^{s_2}} dx +\widetilde{c}_0\delta_0, \label{n6} \\ \nu = Q(x) \frac{|u|^\alpha |v|^\beta }{|x - x_0|^t} dx +\sum_{i \in I}Q (x_i) d_{x_i}\delta_{x_i}+ Q (0) d_0\delta_0,\label{n7} \end{gather} where $\delta_x$ is the Dirac-mass of mass 1 concentrated at the point $x$. First, we consider the possibility of the concentration at $\{x_i\}_{i \in I}\in\Omega \setminus \{0\}$. Let $\epsilon>0$ be small enough, take $\eta_{x_i} \in C_c^\infty (B_{2 \varepsilon} (x_i))$, such that $\eta_{x_i}|_{B_{\varepsilon} (x_i)} = 1$, $0 \leq \eta_{x_i} \leq 1$ and $|\nabla \eta_{x_i} (x)| \leq \frac{C}{\varepsilon}$. Then \begin{align*} o (1)&= \langle J' (u_n, v_n), (\eta_{x_i}^p u_n, \eta_{x_i}^p v_n) \rangle\\ &= \int_{\Omega}\left(|\nabla u_n|^{p-2}\nabla u_n \nabla(\eta_{x_i}^p u_n)+ |\nabla v_n|^{p-2}\nabla v_n \nabla(\eta_{x_i}^p v_n)\right)dx\\ &\quad - \int_{\Omega} Q(x) \frac{|u_n|^\alpha |v_n|^\beta}{|x - x_0|^t} \eta_{x_i}^p dx -\mu \int_{\Omega}\Big(\frac{|u_n|^{p}}{|x|^p} \eta_{x_i}^p + \frac{|v_n|^{p}}{|x|^p} \eta_{x_i}^p \Big)dx\\ &\quad - \lambda\int_{\Omega} h (x) \Big( \frac{|u_n|^q}{|x|^s} \eta_{x_i}^p + \frac{|v_n|^q}{|x|^s} \eta_{x_i}^p \Big)dx\\ &\quad -\int_{\Omega}\frac{|u_n|^{p^*(s_1)}}{|x|^{s_1}} \eta_{x_i}^p dx - \int_{\Omega} \frac{|v_n|^{p^*(s_2)}}{|x|^{s_2}}\eta_{x_i}^p dx. \end{align*} From \eqref{n3}-\eqref{n7}, one can obtain \begin{gather} \lim_{\varepsilon \to 0} \lim_{n \to \infty} \int_{\Omega}\Big(\frac{|u_n|^{p}} {|x|^p} \eta_{x_i}^p +\frac{|v_n|^{p}}{|x|^p} \eta_{x_i}^p \Big)dx = \lim_{\varepsilon \to 0}\Big(\int_{\Omega} \eta_{x_i}^p d \overline{\beta} + \int_{\Omega} \eta_{x_i}^p d\widetilde{\beta} \Big) = 0, \label{s3}\\ \lim_{\varepsilon \to 0} \lim_{n\to\infty}\int_{\Omega} \Big(\frac{|u_n|^{p^*(s_1)}} {|x|^{s_1}} \eta_{x_i}^p +\frac{|v_n|^{p^*(s_2)}}{|x|^{s_2}} \eta_{x_i}^p \Big)dx = \lim_{\varepsilon \to 0}\Big(\int_{\Omega}\eta_{x_i}^p d \overline{\gamma} + \int_{\Omega} \eta_{x_i}^p d \widetilde{\gamma}\Big) = 0, \nonumber\\ \lim_{\varepsilon \to 0} \lim_{n\to\infty}\int_{\Omega} h(x) \Big(\frac{|u_n|^q}{|x|^s} \eta_{x_i}^p + \frac{|v_n|^q}{|x|^s} \eta_{x_i}^p \Big) dx = 0, \nonumber\\ \lim_{\varepsilon \to 0}\lim_{n \to \infty} \int_{\Omega} Q(x) \frac{|u_n|^\alpha |v_n|^\beta}{|x - x_0|^t} \eta_{x_i}^p dx =\lim_{\varepsilon \to 0}\int_{\Omega}\eta_{x_i}^p d \nu = Q (x_i) d x_i. \nonumber \end{gather} Thus, \begin{equation}\label{e2.3} \begin{split} 0&=\lim_{\varepsilon \to 0} \lim_{n\to\infty}\int_{\Omega} \Big(|\nabla u_n|^{p-2}\nabla u_n\nabla(\eta_{x_i}^p u_n)\\ &\quad + |\nabla v_n|^{p-2}\nabla v_n\nabla(\eta_{x_i}^p v_n)\Big) dx - Q (x_i) d x_i. \end{split} \end{equation} Moreover, we have \begin{equation} \label{e2.4} \begin{split} &\lim_{\varepsilon\to0}\lim_{n\to\infty}\big|\int_{\Omega}u_n |\nabla u_n|^{p-2}\nabla u_n \nabla \eta_{x_i}^p dx\big|\\ &\leq \lim_{\varepsilon\to0} \lim_{n\to\infty}\Big(\int_{\Omega}|u_n|^p |\nabla \eta_{x_i}^p|^p dx \Big)^{1/p} \Big(\int_{\Omega}|\nabla u_n|^p dx \Big)^{\frac{p-1}{p}} \\ &\leq C \lim_{\varepsilon\to0}\lim_{n\to\infty} \Big(\int_{\Omega}|u_n|^p |\nabla \eta_{x_i}^p|^p dx \Big)^{1/p} \\ & = C \lim_{\varepsilon\to0} \Big(\int_{\Omega}|u|^p |\nabla \eta_{x_i}^p|^p dx \Big)^{1/p} \\ &\leq C\lim_{\varepsilon \to 0} \Big(\int_{B_\varepsilon (x_i)} |\nabla \eta_{x_i}^p |^N dx\Big)^{1/N} \Big(\int_{B_\varepsilon (x_i)}|u|^{p^*} dx\Big)^{1/P^*} \\ &\leq C\lim_{\varepsilon \to 0} \Big(\int_{B_\varepsilon (x_i)} |u|^{p^*} dx\Big)^{1/P^*} = 0.\\ \end{split} \end{equation} Similarly, \begin{equation}\label{e2.5} \lim_{\varepsilon \to0}\lim_{n\to\infty}\big|\int_{\Omega} v_n |\nabla v_n|^{p-2}\nabla v_n \nabla \eta_{x_i}^p dx\big|=0. \end{equation} Combining \eqref{e2.3}-\eqref{e2.5}, there holds \begin{equation}\label{e2.6} \begin{split} 0 & = \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega}(|\eta_{x_i} \nabla u_n|^p + |\eta_{x_i} \nabla v_n|^p) dx - Q (x_i) d_{x_i} \\ & = \lim_{\varepsilon \to 0} \int_{\Omega}\left(\eta_{x_i} ^p d\overline{\alpha}+ \eta_{x_i} d \widetilde{\alpha}\right)- Q (x_i) d_{x_i}. \end{split} \end{equation} On the other hand, \eqref{hardy2} implies \begin{equation} \label{s4} \begin{split} & \frac{1}{\|Q\|_\infty^\frac{p}{p^* (t)}} S_{t,\alpha, \beta} \Big(\int_\Omega Q (x) \frac{|\eta_{x_i} u_n|^\alpha |\eta_{x_i} v_n|^\beta}{|x - x_0|^t} dx \Big)^{\frac{p}{p^* (t)}} \\ & \leq \int_\Omega \Big(|\nabla (\eta_{x_i} u_n)|^p + |\nabla (\eta_{x_i} v_n)|^p - \mu \frac{|\eta_{x_i} u_n|^p + |\eta_{x_i} v_n|^p}{|x|^p } \Big) dx. \end{split} \end{equation} Note that \[ \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} |\nabla \eta_{x_i}|^p |u_n|^p dx = \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} |\nabla \eta_{x_i}|^p |v_n|^p dx = 0. \] From this equality, \eqref{e2.4} and \eqref{e2.5}, we obtain \begin{gather}\label{s1} \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} |\eta_{x_i} \nabla u_n|^p dx = \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} |\nabla( \eta_{x_i} u_n)|^p dx, \\ \label{s2} \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} |\eta_{x_i} \nabla v_n|^p dx = \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} |\nabla( \eta_{x_i} v_n)|^p dx. \end{gather} Relations \eqref{n7}, \eqref{s3} and \eqref{s4}-\eqref{s2} imply \begin{equation} \label{s5} \frac{1}{\|Q\|_\infty^\frac{p}{p^* (t)}} S_{t,\alpha, \beta} \big(Q (x_i) d_{x_i} \big)^{\frac{p}{p^* (t)}} \leq \lim_{\varepsilon \to0} \int_{\Omega} |\eta_{x_i}|^p d \overline{\alpha} + \lim_{\varepsilon \to0} \int_{\Omega} |\eta_{x_i}|^p d \widetilde{\alpha}. \end{equation} Combining \eqref{e2.6} and \eqref{s5}, \begin{equation}\label{e2.7} \frac{1}{\|Q\|_\infty^\frac{p}{p^* (t)}} S_{t,\alpha, \beta} \big(Q (x_i) d_{x_i} \big)^{\frac{p}{p^* (t)}} \leq Q (x_i) d_{x_i}, \end{equation} which implies that either \begin{equation}\label{e2.9} Q (x_i) d_{x_i} = 0,\quad\text{or}\quad Q (x_i) d_{x_i} \geq \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} S^{\frac{N - t}{p - t}}_{t,\alpha, \beta} . \end{equation} Now, we consider the possibility of the concentration at 0. For $\epsilon>0$ be small enough, take $\eta_0 \in C_c^\infty (B_{2 \varepsilon} (0))$, such that $\eta_0|_{B_{\varepsilon} (0)} = 1$, $0 \leq \eta_0 \leq 1$ and $|\nabla \eta_0 (x)| \leq \frac{C}{\varepsilon}$. Then \begin{align*} o (1)&= \langle J' (u_n, v_n), (\eta_0^p u_n, 0) \rangle\\ &= \int_{\Omega} |\nabla u_n|^{p-2}\nabla u_n \nabla(\eta_0^p u_n)dx -\mu \int_{\Omega}\frac{|u_n|^{p}}{|x|^p} \eta_0^p dx - \lambda \int_{\Omega} h (x) \frac{|u_n|^q}{|x|^s} \eta_0^p dx \\ &\quad -\int_{\Omega}\frac{|u_n|^{p^*(s_1)}}{|x|^{s_1}} \eta_0^p dx - \frac{\alpha}{\alpha + \beta} \int_{\Omega} Q(x) \frac{|u_n|^\alpha |v_n|^\beta}{|x - x_0|^t} \eta_0^p dx. \end{align*} From \eqref{n3}, \eqref{n5}, \eqref{n7} and $Q (0) = 0$, we obtain \begin{gather*} \lim_{\varepsilon \to 0} \lim_{n \to \infty} \int_{\Omega} \frac{|u_n|^{p}} {|x|^p} \eta_0^p dx = \overline{b}_0, \;\;\;\;\; \lim_{\varepsilon \to 0} \lim_{n \to \infty} \int_{\Omega} \frac{|u_n|^{p^* (s_1)}} {|x|^{s_1}} \eta_0^p dx = \overline{c}_0, \\ \lim_{\varepsilon \to 0}\lim_{n \to \infty} \int_{\Omega} Q(x) \frac{|u_n|^\alpha |v_n|^\beta}{|x - x_0|^t} \eta_0^p dx = \lim_{\varepsilon \to 0} \lim_{n \to \infty} \int_{\Omega} h (x) \frac{|u_n|^q}{|x|^s} \eta_0^p dx = 0. \end{gather*} Thus, \begin{equation}\label{s6} 0=\lim_{\varepsilon \to 0} \lim_{n\to\infty}\int_{\Omega} |\nabla u_n|^{p-2}\nabla u_n\nabla(\eta_0^p u_n) dx - \mu \overline{b}_0 - \overline{c}_0. \end{equation} Note that \begin{equation*} \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} u_n |\nabla u_n|^{p-2}\nabla u_n \nabla \eta_0^p dx =0. \end{equation*} This equality and \eqref{s6} yield \begin{equation}\label{s7} \lim_{\varepsilon \to 0} \int_{\Omega} \eta_0^p d\overline{\alpha} - \mu \overline{b}_0 = \overline{c}_0. \end{equation} On the other hand, \eqref{san} implies \[ A_{\mu, s_1} \Big(\int_\Omega \frac{|\eta_0 u_n|^{p^* (s_1)}}{|x|^{s_1}} dx \Big)^{\frac{p}{p^* (s_1)}} \leq \int_\Omega \Big(| \nabla (\eta_0 u_n)|^p - \mu \frac{|\eta_0 u_n|^p}{|x|^p } \Big) dx. \] Thus \begin{equation} \label{s11} A_{\mu, s_1} \overline{c}_0^{\frac{p}{p^* (s_1)}} \leq \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_\Omega | \nabla (\eta_0 u_n)|^p d x - \mu \overline{b}_0. \end{equation} Note that \[ \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} |\eta_0 \nabla u_n|^p dx = \lim_{\varepsilon \to0}\lim_{n\to\infty} \int_{\Omega} |\nabla( \eta_0 u_n)|^p dx, \] which together with \eqref{s11} imply \begin{equation} \label{s12} A_{\mu, s_1} \overline{c}_0^{\frac{p}{p^* (s_1)}} \leq \lim_{\varepsilon \to0} \int_\Omega |\eta_0|^p d \overline{\alpha} - \mu \overline{b}_0. \end{equation} Therefore, from \eqref{s7} and \eqref{s12}, \begin{equation} \label{s13} A_{\mu, s_1} \overline{c}_0^{\frac{p}{p^* (s_1)}} \leq \overline{c}_0, \end{equation} which implies that either \begin{equation}\label{s21} \overline{c}_0 = 0,\quad\text{or}\quad \overline{c}_0 \geq A_{\mu, s_1} ^{\frac{N - s_1}{p - s_1}}. \end{equation} Similarly, either \begin{equation}\label{s22} \overline{c}_0 = 0,\quad\text{or}\quad \overline{c}_0 \geq A_{\mu, s_2} ^{\frac{N - s_2}{p - s_2}}. \end{equation} Recall that $u_n \rightharpoonup u$ weakly and $v_n \rightharpoonup v$ weakly in $W^{1,p}_0 (\Omega)$, we have \begin{equation} \label{s14} \begin{split} &c + o (1) \\ &= J (u_n, v_n) \\ &= \frac{1}{p} \int_\Omega \Big(|\nabla u_n - \nabla u|^p - \mu \frac{|u_n - u|^p}{|x|^p } + |\nabla v_n - \nabla v|^p - \mu \frac{|v_n - v|^p}{|x|^p } \Big) d x \\ &\quad - \frac{1}{p^*(s_1)} \int_\Omega \frac{|u_n - u|^{p^*(s_1)} }{|x|^{s_1}} dx - \frac{1}{p^*(s_2)} \int_\Omega \frac{|v_n - v|^{p^*(s_2)} }{|x|^{s_2}} dx \\ &\quad - \frac{1}{p^* (t)} \int_\Omega Q (x) \frac{|u_n - u|^\alpha |v_n - v|^\beta }{|x - x_0|^t} dx + J (u, v). \end{split} \end{equation} On the other hand, from $o(1) = J' (u_n, v_n)$, we obtain \begin{equation} \label{s15} J' (u_n, v_n) = 0. \end{equation} Thus, $0 = \langle J '(u, v), (u, v) \rangle$, which together with $o(1) = \langle J '(u_n, v_n), (u_n, v_n) \rangle$ imply \begin{equation} \label{s16} \begin{split} o (1) &= \int_\Omega \Big(|\nabla u_n - \nabla u|^p - \mu \frac{|u_n - u|^p}{|x|^p } + |\nabla v_n - \nabla v|^p - \mu \frac{|v_n - v|^p}{|x|^p } \Big) dx \\ &\quad - \int_\Omega \frac{|u_n - u|^{p^*(s_1)} }{|x|^{s_1}} dx - \int_\Omega \frac{|v_n - v|^{p^*(s_2)} }{|x|^{s_2}} dx \\ &\quad -\int_\Omega Q (x) \frac{|u_n - u|^\alpha |v_n - v|^\beta }{|x - x_0|^t} dx. \end{split} \end{equation} From \eqref{s14}-\eqref{s16} and Lemma \ref{ps0}, \begin{equation} \label{s17} \begin{split} c + o (1) & \geq \frac{p - s_1}{p (N - s_1)} \int_\Omega \frac{|u_n - u|^{p^*(s_1)} }{|x|^{s_1}} dx + \frac{p - s_2}{p (N - s_2)} \int_\Omega \frac{|v_n - v|^{p^*(s_2)} }{|x|^{s_2}} dx \\ &\quad + \frac{p - t}{p(N - t)} \int_\Omega Q (x) \frac{|u_n - u|^\alpha |v_n - v|^\beta }{|x - x_0|^t} dx - d \lambda^\frac{p}{p- q}. \end{split} \end{equation} Passing to the limit in \eqref{s17} as $n \to \infty$, we have \begin{equation} \label{s18} c \geq \frac{p - s_1}{2 (N - s_1)} \overline{c}_0 + \frac{p - s_2}{p (N - s_2)} \widetilde{c}_0 + \frac{p - t}{p(N - t)} \sum_{i \in I} Q (x_i) d_{x_i} - d \lambda^\frac{p}{p - q}. \end{equation} By the assumption $c < c_*$ and in view of \eqref{e2.9}, \eqref{s21} and \eqref{s22}, there holds $\overline{c}_0 = \widetilde{c}_0 = 0$, $Q (x_i) d_{x_i} = 0$, $i \in I$. Up to a subsequence, $(u_n, v_n) \to (u, v)$ strongly in $W$ as $n \to \infty$. \end{proof} When the restriction $Q (0) = 0$ is removed, we establish the following version of Lemma \ref{ps2}. \begin{lemma}\label{ps3} Assume that $N \geq 3$, $0 \leq \mu < \overline{\mu}$, (H1) and $h_+ \ne 0$. Then $J(u, v)$ satisfies the $(PS)_c$ condition with $c$ satisfying \begin{equation} \label{c2} \begin{split} c < c_0 : = \min \Big\{&\frac{p - s_1}{p (N - s_1)} \Big (\frac{1}{p} A_{\mu, s_1} \Big)^\frac{N - s_1}{p - s_1}, \frac{p -s_2}{p (N - s_2)} \Big ( \frac{1}{p} A_{\mu, s_2} \Big)^\frac{N - s_2}{p - s_2}, \\ & \frac{p - t}{p(N - t)} \frac{\Big ( \frac{1}{p} S_{t, \alpha, \beta} \Big )^\frac{N - t}{p - t}}{\|Q\|_\infty^\frac{N - p}{p- t}} \Big\} - d \lambda^\frac{p}{p - q}. \end{split} \end{equation} \end{lemma} The proof of the above lemma is similar to Lemma \ref{ps2} and is omitted. \begin{lemma}[\cite{Kang}] \label{lem kang} Assume that $1 < p < N$, $0 \leq t < p$ and $0 \leq \mu < \overline{\mu}$. Then the limiting problem \begin{gather*} - \Delta_p u - \mu \frac{|u|^{p - 1} }{|x|^p} = \frac{|u|^{p^* (t) - 1} }{|x|^t}, \quad \text{in } \mathbb{R}^N \setminus \{0 \},\\ u \in D^{1, p} (\mathbb{R}^N), \quad u > 0, \quad \text{in } \mathbb{R}^N \setminus \{0 \}, \end{gather*} has positive radial ground states \begin{equation} \label{ve} V_\epsilon (x) \triangleq \epsilon^\frac{p - N}{p} U_{p, \mu}(\frac{x}{\epsilon}) = \epsilon^\frac{p - N}{p} U_{p, \mu} (\frac{|x|}{\epsilon}), \quad \forall \epsilon > 0, \end{equation} that satisfy \[ \int_\Omega \Big(|\nabla V_\epsilon (x)|^p - \mu \frac{|V_\epsilon (x)|^p}{|x|^p } \Big) dx = \int_\Omega \frac{|V_\epsilon (x)|^{p^*(t)}}{|x|^t } dx = (A_{\mu, t})^\frac{N - t}{p - t}, \] where $U_{p, \mu} (x) = U_{p, \mu} (|x|)$ is the unique radial solution of the limiting problem with \[ U_{p, \mu}(1) = \Big ( \frac{(N - t) (\overline{\mu} - \mu)}{N - p} \Big)^\frac{1}{p^*(t) - p}. \] Furthermore, $U_{p, \mu}$ has the following properties: \begin{gather*} \lim_{r \to 0} r^{a(\mu)} U_{p, \mu}(r) = C_1 > 0,\quad \lim_{r \to + \infty} r^{b(\mu)} U_{p, \mu}(r) = C_2 > 0,\\ \lim_{r \to 0} r^{a(\mu) + 1} |U'_{p, \mu}(r)| = C_1 a(\mu) \geq 0,\\ \lim_{r \to + \infty} r^{b(\mu) + 1} |U'_{p, \mu}(r)| = C_2 b (\mu) > 0, \end{gather*} where $C_i$ $(i = 1, 2)$ are positive constants and $a(\mu)$ and $b(\mu)$ are zeros of the function \[ f(\zeta) = (p - 1) \zeta^p - (N - p) \zeta^{p - 1} + \mu, \quad \zeta \geq 0, \; 0 \leq \mu < \overline{\mu}, \] that satisfy \[ 0 \leq a(\mu) < \frac{N - p}{p} < b(\mu) \leq \frac{N - p}{p - 1}. \] \end{lemma} \begin{lemma}\label{lem11} Under the assumptions of Theorem \ref{the1}, there exists $(u_1, v_1) \in W \setminus \{(0, 0)\}$ and $\Lambda_1 > 0$, such that for $0 < \lambda < \Lambda_1$, there holds \begin{equation} \label{3.11} \sup_{t \geq 0} J (t u_1, t v_1) < \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} S^\frac{N - t}{p - t}_{t,\alpha, \beta} (\mu) - d \lambda^\frac{p}{p - q}. \end{equation} \end{lemma} \begin{proof} First, we give some estimates on the extremal function $V_\epsilon (x)$ defined in \eqref{ve}. For $m \in \mathbb{N}$ large, choose $\varphi (x) \in C_0^\infty (\mathbb{R}^N)$, $0 \leq \varphi (x) \leq 1$, $\varphi (x) = 1$ for $|x| \leq \frac{1}{2m}$, $\varphi (x) = 0$ for $|x| \geq \frac{1}{m}$, $\|\nabla \varphi (x)\|_{L^p (\Omega)} \leq 4 m$, set $u_\epsilon (x) = \varphi (x) V_\epsilon (x)$. For $\epsilon \to 0$, the behavior of $u_\epsilon$ has to be the same as that of $V_\epsilon$, but we need precise estimates of the error terms. For $1 < p < N$, $0 \leq t < p$ and $1 < q < p^*(s)$, we have the following estimates \cite{D. Cao}: \begin{gather} \int_\Omega \Big(|\nabla u_\epsilon|^p - \mu \frac{|u_\epsilon|^p}{|x|^p } \Big) dx = (A_{\mu, t})^\frac{N-t}{p - t} + O (\epsilon ^{b(\mu) p + p - N}), \label{e8} \\ \int_\Omega \frac{|u_\epsilon|^{p^*(t)}}{|x|^t } dx = (A_{\mu, t})^\frac{N - t}{p - t} + O (\epsilon ^{b(\mu) p^*(t) - N + t}), \label{e9} \\ \int_\Omega \frac{|u_\epsilon|^q}{|x|^s } dx \geq \begin{cases} C \epsilon^{N - s + (1 - \frac{N}{p})q}, \ & q > \frac{N - s}{b (\mu)},\\ C \epsilon^{N - s + (1 - \frac{N}{p})q} |\ln \epsilon|, & q = \frac{N - s}{b (\mu)},\\ C \epsilon^{q (b(\mu) + 1 - \frac{N}{p})q}, & q < \frac{N - s}{b (\mu)}. \end{cases}\label{e10} \end{gather} Now, we consider the functional $I : W \to \mathbb{R}$ defined by \[ I (u, v) = \frac{1}{p} \int_\Omega \Big(|\nabla u|^p - \mu \frac{|u|^p}{|x|^p } + |\nabla v|^p - \mu \frac{|v|^p}{|x|^p }\Big) d x - \frac{1}{p^* (t)} \int_\Omega Q (x) \frac{|u|^\alpha |v|^\beta }{|x - x_0|^t} dx. \] Let $u_1 = \alpha^\frac{1}{p} u_\epsilon$, $v_1 = \beta^\frac{1}{p} u_\epsilon$ and define the function $g_1 (t) : = J (t u_1, t v_1)$, $t \geq 0$. Note that $\lim_{t \to + \infty} g_1(t) = - \infty$ and $g_1(t) > 0$ as $t$ is close to $0$. Thus $\sup_{t \geq 0} g_1(t)$ is attained at some finite $t_\epsilon > 0$ with $g'_1(t_\epsilon) = 0$. Furthermore, $C' < t_\epsilon < C''$; where $C'$ and $C''$ are the positive constants independent of $\epsilon$. We have \begin{equation}\label{sa3} I(t u_1, t v_1) = y (t u_1, t v_1) - \frac{t^{p^*(t)}}{p^*(t)} (\alpha^\frac{\alpha}{p} \beta^\frac{\beta}{p}) \int_\Omega (Q (x) - Q (x_0)) \frac{|u_\epsilon|^{p^*(t)}}{|x - x_0|^t } dx. \end{equation} where \begin{align*} &y (t u_1, t v_1)\\ &:= \Big [\frac{t^p}{p} (\alpha + \beta) \int_\Omega \Big(|\nabla u_\epsilon|^p - \mu \frac{|u_\epsilon|^p}{|x|^p } \Big) dx - \frac{t^{p^*(t)}}{p^*(t)} (\alpha^\frac{\alpha}{p} \beta^\frac{\beta}{p}) \int_\Omega Q (y_0) \frac{|u_\epsilon|^{p^*(t)}}{|x - x_0|^t } dx \Big ]\,. \end{align*} Note that \begin{equation} \label{ina} \sup_{t \geq 0} \big(\frac{t^p}{p} A - \frac{t^{p^*(t)}}{p^*(t)} B \big) =\big(\frac{1}{p} - \frac{1}{p^*(t)} \big) \big(\frac{A}{B^\frac{p}{p^*(t)}}\big)^{\frac{p^*(t)}{p^*(t) - p}}, \quad A, B > 0. \end{equation} From (H2), \eqref{e8}, \eqref{e9} and \eqref{ina} it follows that \begin{align} &\sup_{t \geq 0} y(t u_1, t v_1) \nonumber\\ & = y (t_\epsilon u_1, t_\epsilon v_1) \nonumber\\ &\leq \frac{p - t}{p(N - t)} \Big( \frac{(\alpha + \beta) \int_\Omega \Big(|\nabla u_\epsilon|^p - \mu \frac{|u_\epsilon|^p}{|x|^p } \Big) dx }{((\alpha^\frac{\alpha}{p} \beta^\frac{\beta}{p}) \|Q\|_\infty \int_\Omega \frac{|u_\epsilon|^{p^*(t)}}{|x - x_0|^t } dx)^\frac{N - p}{N -t}} \Big)^\frac{N - t}{p - t} \nonumber\\ &\leq \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \Big[ \Big(\Big (\frac{\alpha}{\beta} \Big)^\frac{\beta}{\alpha + \beta} + \Big (\frac{\beta}{\alpha} \Big)^\frac{\alpha}{\alpha + \beta} \Big) \frac{(A_{\mu, t})^\frac{N - t}{p - t} + O (\epsilon ^{b(\mu) p + p - N})}{(A_{\mu,t})^\frac{N - t}{p - t} + O (\epsilon ^{b(\mu) p^*(t) - N + t})} \Big]^{\frac{N - t}{p - t}} \nonumber \\ &\leq \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \Big[ \Big(\Big (\frac{\alpha}{\beta} \Big)^\frac{\beta}{\alpha + \beta} + \Big (\frac{\beta}{\alpha} \Big)^\frac{\alpha}{\alpha + \beta} \Big) (A_{\mu, t})^\frac{N - t}{p - t} \Big]^{\frac{N - t}{p - t}} + O (\epsilon ^{b(\mu) p + p - N}) \nonumber\\ &\leq \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \Big(S_{t, \alpha, \beta} \Big)^\frac{N - t}{p - t} + O (\epsilon ^{b(\mu) p + p - N}). \label{sa4} \end{align} On the other hand, (H2) implies that there exists $r_1 < r$, such that for $x \in B_{r_1} (y_0)$, $|Q (x) - Q (x_0)| \leq C |x - x_0|^{\vartheta}$. Thus \begin{equation} \label{sa5} \begin{split} \big| \int_\Omega (Q (x) - Q (x_0)) \frac{|u_\epsilon|^{p^*(t)}}{|x - x_0|^t } dx \big| &\leq C \int_\Omega |Q (x) - Q (x_0)| \frac{|u_\epsilon|^{p^*(t)}}{|x - x_0|^t } dx \\ & = C \int_{B_{2 r} (x_0)} \frac{|x - x_0|^\vartheta |u_\epsilon |^{p^*(t)}}{|x - x_0|^t } dx = O ( \epsilon^{\vartheta - t}) \end{split} \end{equation} From \eqref{sa3}, \eqref{sa4} and \eqref{sa5}, we conclude that \begin{equation} \label{sa6} \begin{split} \sup_{t \geq 0} I(t u_1, t v_1) &= I (t_\epsilon u_1, t_\epsilon v_1) \\ &\leq \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \Big(S_{t, \alpha, \beta} \Big)^\frac{N - t}{p - t} + O (\epsilon ^{b(\mu) p + p - N}). \end{split} \end{equation} Observe that there exists $\Lambda_1^* > 0$, such that for $0 < \lambda < \Lambda_1^*$ and \[ \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \Big(S_{t, \alpha, \beta} \Big)^\frac{N - t}{p - t} - d \lambda^\frac{p}{p - q} >0. \] Then for $0 < \lambda < \Lambda_1^*$, there exists $t_1 \in (0, 1)$, such that \begin{equation} \label{sa1} \begin{split} \sup_{0 \leq t \leq t_1} J (t u_1, t v_1) &\leq \sup_{0 \leq t \leq t_1} \frac{1}{p} t^p \int_\Omega \Big(|\nabla u_1|^p + |\nabla v_1|^p \Big) dx \\ & < \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \Big(S_{t, \alpha, \beta} \Big)^\frac{N - t}{p - t} - d \lambda^\frac{p}{p - q}. \end{split} \end{equation} On the other hand, \begin{equation} \label{sa2} \begin{split} &\sup_{t \geq t_1} J (t u_1, t v_1)\\ &\leq \sup_{t \geq t_1} \Big[I (t u_1, t v_1) - \frac{\lambda}{q} t^q \int_\Omega h (x)\frac{|u_1|^q}{|x|^s} dx - \frac{1}{p^*(s_1)} t^{p^*(s_1)} \int_\Omega \frac{|u_1|^{p^*(s_1)} }{|x|^{s_1}} dx \Big ] \\ &\leq \sup_{t \geq t_1} I (t u_1, t v_1) - \frac{\lambda}{q} t_1^q \int_\Omega h (x) |u_1|^q dx - \frac{1}{p^*(s_1)} t_1^{p^*(s_1)} \int_\Omega \frac{|u_1|^{p^*(s_1)} }{|x|^{s_1}} dx\Big ] \\ &\leq \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \Big(S_{t, \alpha, \beta} \Big)^\frac{N - t}{p - t} + O (\epsilon ^{b(\mu) p + p - N}) \\ &\quad - C \int_\Omega \frac{|u_\epsilon|^{p^*(s_1)} }{|x|^{s_1}} dx - \lambda C \int_\Omega h (x) \frac{|u_\epsilon|^q}{|x|^s} dx \end{split} \end{equation} From \eqref{e9}, \begin{equation} \label{san1} \int_\Omega \frac{|u_\epsilon|^{p^*(s_1)}}{|x|^{s_1} } dx \geq O (\epsilon ^{b(\mu) p^*(s_1) - N + s_1}). \end{equation} Also, from \eqref{e10}, it follows that \begin{equation} \label{san2} \int_\Omega h (x) \frac{|u_\epsilon|^q}{|x|^s} dx \geq \beta_0 \int_\Omega \frac{|u_\epsilon|^q}{|x|^s} dx \geq \begin{cases} C \epsilon^{N - s + (1 - \frac{N}{p})q}, & q > \frac{N - s}{b (\mu)},\\ C \epsilon^{N - s + (1 - \frac{N}{p})q} |\ln \epsilon|, & q = \frac{N - s}{b (\mu)},\\ C \epsilon^{q (b(\mu) + 1 - \frac{N}{p})q}, & q < \frac{N - s}{b (\mu)}. \end{cases} \end{equation} Since $q \geq \frac{N - s}{b (\mu)}$, by \eqref{sa2}-\eqref{san2} we have \begin{align*} \sup_{t \geq t_1} J (t u_1, t v_1) &\leq \frac{p - t}{p(N - t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \big(S_{t, \alpha,\beta}\big)^\frac{N - t}{p - t} + O (\epsilon ^{b(\mu) p + p - N}) \\ &\quad + O (\epsilon ^{b(\mu) p^*(s_1) - N + s_1}) - \lambda \begin{cases} C \epsilon^{N - s + (1 - \frac{N}{p})q}, & q > \frac{N - s}{b (\mu)},\\ C \epsilon^{N - s + (1 - \frac{N}{p})q} |\ln \epsilon|, & q = \frac{N - s}{b (\mu)}. \end{cases} \end{align*} Note that $b(\mu) p + p - N < b(\mu) p^*(s_1) - N + s_1$, then we have \begin{equation} \label{san3} \begin{split} \sup_{t \geq t_1} J (t u_1, t v_1) &\leq \frac{p - t}{p(N -t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \big(S_{t, \alpha, \beta} \big)^\frac{N - t}{p - t} \\ &\quad + O (\epsilon ^{b(\mu) p^*(s_1) - N + s_1}) - \lambda \begin{cases} C \epsilon^{N - s + (1 - \frac{N}{p})q}, & q > \frac{N - s}{b (\mu)},\\ C \epsilon^{N - s + (1 - \frac{N}{p})q} |\ln \epsilon|, & q = \frac{N - s}{b (\mu)}. \end{cases} \end{split} \end{equation} Note that $N > p^2$, $b (\mu) \geq \frac{N - s}{q}$. Thus \[ \big[N -s + (1 - \frac{N}{p})q \big] \frac{p - q}{q} < b(\mu) p + p - N - \big[N -s + (1 - \frac{N}{p})q \big]. \] Choose $\lambda = \epsilon^\tau$, where $\big[N - s + (1 - \frac{N}{p})q \big] \frac{p - q}{q} < \tau < b(\mu) p + p - N - \big[N - s + (1 - \frac{N}{p})q \big]$. Then \[ \lambda O (\epsilon^{N - s + (1 - \frac{N}{p})q}) = O (\epsilon^{\tau + N - s + (1 - \frac{N}{p})q}), \quad d \lambda^\frac{p}{p - q} = O (\epsilon^\frac{p \tau}{p - q}). \] Since $\tau + N - s + (1 - \frac{N}{p})q < \frac{p \tau}{p - q}$, $\tau + N - s + (1 - \frac{N}{p})q < b(\mu) p + p - N$, taking $\epsilon$ small enough we deduce that there exists $\delta > 0$, such that \begin{equation} \label{san4} O (\epsilon ^{b(\mu) p^*(s_1) - N + s_1}) - \lambda O (\epsilon^{N - s + (1 - \frac{N}{p})q}) < - d \lambda^\frac{p}{p - q}, \quad \forall \lambda : 0 < \lambda^\frac{p}{p - q} < \delta. \end{equation} Choose $\Lambda_1 = \min \{\Lambda_1^*, \frac{p - q}{p} \delta\}$. Then for all $\lambda \in (0, \Lambda_1)$ we have \[ \sup_{t \geq t_1} J (t u_1, t v_1) \leq \frac{p - t}{p(N -t)} \frac{1}{\|Q\|_\infty^\frac{N - p}{p - t}} \big(S_{t, \alpha,\beta} \big)^\frac{N - t}{p - t} - d \lambda^\frac{p}{p - q}. \] Together with \eqref{sa1}, we get the conclusion of Lemma \ref{lem11}. \end{proof} \section{Proof of the main results} \begin{proof}[Proof of Theorem \ref{thmA}] Let \begin{gather*} r: = \|(u,v)\|\,, \\ f(r): = \frac{1}{p}r^p-\frac{1}{p^*(s_1)} A_{\mu,s_1}^{-\frac{p^*(s_1)}{p}}r^{p^*(s_1)} -\frac{1}{p^*(s_2)}A_{\mu,s_2}^{-\frac{p^*(s_2)}{p}}r^{p^*(s_2)} - \frac{1}{p^* (t)} S_{t, \alpha, \beta}^{-\frac{p^* (t)}{p}} \|Q\|_\infty\,, \\ h(r): = \frac{\lambda}{q} \Big(\frac{N \omega_N R_0^{N - s}}{N- s} \Big)^\frac{p^* (s) - q}{p^* (s)} A_{\mu, s}^{-\frac{q}{p}} r^q\,. \end{gather*} From \eqref{san}, \eqref{hardy2} and \eqref{hold}, \[ J (u,v) \geq f(r)-h(r). \] Note that $p < p^*(s_1)$, $p^*(s_2)$, $p^*(t)$, it is easy to see that there exists $\varrho > 0$ such that $f(r)$ achieves its maximum at $\varrho$ and $f(\varrho)>0$. Therefore, there exists $\Lambda_{11}>0$, such that for $0 < \lambda <\Lambda_{11}$, \begin{equation}\label{e3.1} \inf_{\|(u,v)\|= \varrho} I(u,v) \geq f(\varrho) - h(\varrho) > 0. \end{equation} On the other hand, set $B_\varrho=\{(u,v);\;\|(u,v)\|\leq \varrho \}$. For $(u,v) \neq (0,0)$, we can choose $d > 0$ small enough, such that $(d u,d v)\in B_\varrho$ and \begin{equation}\label{e3.2} I(d u,d v) < 0. \end{equation} Thus, \begin{equation}\label{e3.3} -\infty < \inf_{(u,v)\in B_\varrho} I(u,v) < 0. \end{equation} Now we can apply the Ekeland variational principle in \cite{MW} and obtain $\{(u_n,v_n)\}\subset B_\varrho$, such that \begin{gather}\label{e3.4} I(u_n,v_n)\leq\inf_{(u,v)\in B_\varrho}I(u,v)+\frac{1}{n},\\ \label{e3.5} I(u_n,v_n)\leq I(u,v)+\frac{1}{n}\|(u_n-u,v_n-v)\|, \end{gather} for all $(u,v)\in B_R$. Define \begin{equation}\label{e3.6} J_1(u,v):= J (u,v)+\frac{1}{n}\|(u_n-u,v_n-v)\|. \end{equation} By \eqref{e3.5}, we have $(u_n,v_n)$ is the minimizer of $J_1(u,v)$ on $B_\varrho$. \eqref{e3.1}, \eqref{e3.3} and \eqref{e3.4} imply that there exists $\epsilon>0$ and $k\in N$, such that for $n\geq k$, $\{(u,v),\:\|(u,v)\|\leq \varrho -\epsilon\}$. Therefore, for $n\geq k$ and $(\phi,\varphi)\in W$, we can choose $t>0$ small enough, such that $(u_n+t\phi,v_n+t\varphi)\in B_\varrho$ and \[ \frac{J_1 (u_n+t\phi,v_n+t\varphi)- J_1 (u_n,v_n)}{t}\geq 0. \] That is, \begin{equation}\label{e3.7} \frac{J (u_n+t\phi,v_n+t\varphi)- J (u_n,v_n)}{t} + \frac{1}{n}\|(\phi,\varphi)\|\geq 0. \end{equation} Passing to the limit in \eqref{e3.7} as $n\to0$, one can obtain \[ \langle J'(u_n,v_n),(\phi,\varphi)\rangle\geq-\frac{1}{n}\|(\phi,\varphi)\|, \] which implies \begin{equation}\label{e3.8} \|J'(u_n,v_n)\|\leq\frac{1}{n}. \end{equation} Combining \eqref{e3.4} and \eqref{e3.8}, there holds \begin{gather}\label{e3.9} \lim_{n\to\infty} J (u_n,v_n)= \inf_{(u,v)\in B_\varrho}J(u,v) < 0,\\ \label{e3.10} \lim_{n\to\infty}J'(u_n,v_n)=0. \end{gather} We note that there exists $\Lambda_{11}^* \in (0, \Lambda_{11})$, such that for $0<\lambda < \Lambda_{11}^*$, and $c_0>\inf_{(u,v)\in B_\varrho } I(u,v)$, where $c_0$ is defined in Lemma \ref{ps3}. Thus, \eqref{e3.9} and \eqref{e3.10} and Lemma Lemma \ref{ps3} imply that for $0<\lambda < \Lambda_{11}^*,\;(u_n,v_n)\to(u,v)$ strongly in $W$. Therefore, $(u,v)$ is a nontrivial solution of problem \eqref{1} satisfying $J(u,v)=\inf_{(u,v)\in B_\varrho} J (u,v) < 0$. Note that $J (u,v)=J (|u|,|v|)$ and $$ (|u|,|v|)\in \{(u,v),\:\|(u,v)\|\leq \varrho - \epsilon \}, $$ we have $I(|u|,|v|)=\inf_{(u,v)\in B_\varrho} J(u,v)$ and $J'(|u|,|v|)=0$. Then problem \eqref{1} has a nontrivial nonnegative solution. By the strongly maximum principle, we get the conclusion of Theorem \ref{thmA}. \end{proof} \begin{proof}[Proof of Theorem \ref{the1}] In view of the proof of Theorem \ref{thmA}, we know that for $0<\lambda< \Lambda_{11}$, there exists $\varrho > 0$, such that $\inf_{\|(u,v)\|=\varrho} I(u,v) \geq \vartheta^* > 0$. Moreover, \eqref{e3.9} and \eqref{e3.10} hold. We note that there exists $\Lambda_{12} \in (0,\Lambda_{11})$, such that for $0< \lambda < \Lambda_{12},\;c_*> \inf_{(u,v) \in B_\varrho} J(u,v)$, where $c_*$ is defined in Lemma \ref{ps2}. Thus \eqref{e3.9} and \eqref{e3.10} and Lemma \ref{ps2} imply that $(u_n,v_n) \to (u,v)$ strongly in $W$. Standard argument shows that for $0 < \lambda< \Lambda_{12}$, problem \eqref{1} has at least one positive solution satisfying $J(u,v)<0$ and $J'(u,v)=0$. Now we prove a second positive solution. It is easy to see $J (0,0)=0$. Set $\Lambda^{**}= \min \{\Lambda_{12}, \Lambda_1\}$, where $\Lambda_1$ is given in Lemma \ref{lem11}. Then it follows from Lemma \ref{lem11} there exists $(u',v')\in W \setminus \{0 \}$, such that for $0< \lambda < \Lambda^{**}$, $$ \sup_{t \geq 0} J (tu',tv')< c_*. $$ On the other hand we obtain that $\lim_{l\to\infty} J(lu',lv')=-\infty$. Thus there exists $l'>0$ such that $\|(l'u',l'v')\|> \varrho$ and $J(l'u',l'v')<0$. Let $$ c:=\inf_{\gamma\in\Gamma}\sup_{t\in[0,1]}J(\gamma(t)), $$ where $$ \Gamma:=\{\gamma\in C^0([0,1],W):\gamma(0)=(0,0),\,\gamma(1)=(l'u',l'v')\}. $$ Thus, it follows from the mountain pass lemma in \cite{A. Ambrosetti} that there exists a sequence $(u_n,v_n) \in W$ such that \begin{gather*} \lim_{n\to\infty}J(u_n,v_n)=c,\\ \lim_{n\to\infty} J'(u_n,v_n)=0. \end{gather*} Moreover, $c\in(0,c_*)$. From Lemma \ref{ps2}, $(u_n,v_n)\to(\overline{u},\overline{v})$ strongly in $W$, which implies that $J (\overline{u},\overline{v})=c$ and $J'(\overline{u},\overline{v})=0$, Therefore, $(\overline{u},\overline{v})$ is a second nontrivial solution of \eqref{1}. Set $u^+=\max\{u,0\},v^+=\max\{v,0\}$. Replacing \[ \int_{\Omega} \frac{|u|^q}{|x|^s} dx,\quad \int_{\Omega} \frac{|v|^q}{|x|^s} dx,\quad \int_{\Omega}\frac{|u|^{p^*(s_1)}}{|x|^{s_1}}dx,\quad \int_{\Omega}\frac{|v|^{p^*(s_2)}}{|x|^{s_2}}dx, \quad \int_{\Omega} Q(x) \frac{|u|^\alpha |v|^\beta}{|x - x_0|^t}dx \] by \begin{gather*} \int_{\Omega} \frac{(u^+)^q}{|x|^s} dx,\quad \int_{\Omega} \frac{(v^+)^q}{|x|^s} dx,\quad \int_{\Omega}\frac{(u^+)^{p^*(s_1)}}{|x|^{s_1}}dx,\\ \int_{\Omega}\frac{(v^+)^{p^*(s_2)}}{|x|^{s_2}}dx, \quad \int_{\Omega} Q(x) \frac{(u^+)^\alpha (v^+)^\beta}{|x - x_0|^t}dx \end{gather*} and repeating the above process, we have a nonnegative solution $(\widetilde{u},\widetilde{v})$ of problem \eqref{1} satisfying $J (\widetilde{u},\widetilde{v}) > 0$. Then by the strongly maximum principle, we have a second positive solution. \end{proof} \subsection*{Acknowledgments} The authors would like to thank the anonymous referees for his/her valuable suggestions and comments. \begin{thebibliography}{99} \bibitem{C. Alves} C. Alves, D. Filho, M.\ Souto; \emph{On systems of elliptic equations involving subcritical or critical Sobolev exponents}, Nonlinear Anal. 42 (2000) 771-787. \bibitem{A. Ambrosetti} A. Ambrosetti, P. H. Rabinowitz; \emph{Dual variational methods in critical point theory and applications}, J. Funct. Anal. 14 (1973) 349-381. \bibitem{L. Caffarelli} L. Caffarelli, R. Kohn, L. Nirenberg; \emph{First order interpolation inequality with weights}, Compos. Math. 53 (1984) 259-275. \bibitem{D. Cao} D. Cao, P. Han; \emph{Solutions for semilinear elliptic equations with critical exponents and Hardy potential}, J. Differential Equations 205 (2004) 521-537. \bibitem{D. Cao1} D. Cao, D. Kang; \emph{Solutions of quasilinear elliptic problems involving a Sobolev exponent and multiple Hardy-type terms}, J. Math. Anal. Appl. 333 (2007) 889-903. \bibitem{Degiovanni} M. Degiovanni, S. Lancelotti; \emph{Linking solutions for p-Laplace equations with nonlinearity at critical growth}, J. Funct. Anal. 256 (2009) 3643-3659. \bibitem{F. Demengel} F. Demengel, E. Hebey; \emph{On some nonlinear equations involving the p-Laplacian with critical Sobolev growth}, Adv. Differential Equations 3 (1998) 533-574. \bibitem{R. Filippucci} R. Filippucci, P. Pucci, F. Robert; \emph{On a p-Laplace equation with multiple critical nonlinearities}, J. Math. Pures Appl. 91 (2009) 156-177. \bibitem{J. Garcia Azorero} J. Garcia Azorero, I. Peral; \emph{Hardy inequalities and some critical elliptic and parabolic problems}, J. Differential Equations 144 (1998) 441-476. \bibitem{N. Ghoussoub} N. Ghoussoub, F. Robert; \emph{The effect of curvature on the best constant in the Hardy–Sobolev inequality}, Geom. Funct. Anal. 16 (2006) 897-908. \bibitem{N. Ghoussoub1} N. Ghoussoub, C. Yuan; \emph{Multiple solutions for quasi-linear PDEs involving the critical Sobolev and Hardy exponents}, Amer. Math. Soc. 352 (2000) 5703-5743. \bibitem{N. Ghoussoub2} N. Ghoussoub, C. Yuan; \emph{Multiple solutions for quasi-linear PDEs involving the critical Sobolev and Hardy exponents}, Trans. Amer. Math. Soc. 352 (2000) 5703-5743. \bibitem{Han1} P. Han; \emph{Multiple positive solutions for a critical growth problem with Hardy potential}, Proc. Edinb. Math. Soc. 49 (2006) 53-69. \bibitem{Han2} P. Han; \emph{Multiple solutions to singular critical elliptic equations}, Israel J. Math. 156 (2006) 359-380. \bibitem{Han4} P. Han; \emph{Many solutions for elliptic equations with critical exponents}, Israel J. Math. 164 (2008) 125-152. \bibitem{P. Han} P. Han; \emph{Quasilinear elliptic problems with critical exponents and Hardy terms}, Nonlinear Anal. 61 (2005) 735-758. \bibitem{Han3} P. Han, L. Zhaoxia; \emph{Solutions for a singular critical growth problem with a weight}, J. Math. Anal. Appl. 327 (2007) 1075-1085. \bibitem{Kang} D. Kang; \emph{On the quasilinear elliptic problem with a critical Hardy–Sobolev exponent and a Hardy term}, Nonlinear Anal. 69 (2008) 2432-2444. \bibitem{S. Peng} D. Kang, S. Peng; \emph{Solutions for semilinear elliptic problems with critical Sobolev–Hardy exponents and Hardy potential}, Appl. Math. Lett. 18 (2005) 1094-1100. \bibitem{Y. Li} Y. Li, Q. Guo, P. Niu; \emph{The existence of solutions for quasilinear elliptic problems with combined critical Sobolev-Hardy terms}, J. Math. Anal. Appl. 388 (2012) 525-538. \bibitem{P.L. Lions} P. L. Lions; \emph{The concentration-compactness principle in the calculus of variations. the limit case, part 1}, Rev. Mat. Iberoam. 1 (1985) 145-201. \bibitem{MW} J. Mawhin, M. Willem; \emph{Critical Point Theory and Hamiltonian Systems}, Appl. Math. Sci. 74, Springer, New York (1989). \bibitem{Nia1} N. Nyamoradi; \emph{Existence and multiplicity of solutions to a singular elliptic system with critical Sobolev-Hardy exponents and concave-convex nonlinearities}, J. Math. Anal. Appl. 396 (2012) 280-293. \bibitem{P. Rabinowitz} P. Rabinowitz; \emph{Minimax methods in critical points theory with applications to differential equations}, in: CBMS Regional Conference Series in Mathematics, vol. 65, Providence, RI, 1986. \bibitem{Shang} Y. Y. Shang; \emph{Existence and multiplicity of positive solutions for some Hardy-Sobolev critical elliptic equation with boundary singularities}, Nonlinear Anal. 75 (2012) 2724-2734. \bibitem{Xuan1} B. Xuan, J. Wang' \emph{Existence of a nontrivial weak solution to quasilinear elliptic equations with singular weights and multiple critical exponents}, Nonlinear Analysis 72 (2010) 3649-3658. \bibitem{J. Zhang} J. Zhang, Z. Wei; \emph{Existence of multiple positive solutions to singular elliptic systems involving critical exponents}, Nonlinear Anal. 75 (2012) 559-573. \end{thebibliography} \end{document}