Abstract
In this paper, we study the following weighted nonlocal system with critical exponents related to the Stein–Weiss inequality
By using moving plane arguments in integral form, we obtain symmetry, regularity and asymptotic properties, as well as sufficient conditions for the nonexistence of solutions to the nonlocal Stein–Weiss system.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Under suitable symmetry hypotheses, notably radial symmetry, classical estimates and embedding properties of function spaces admit substantial improvements. For instance, the following radial estimate of Strauss [32] establishes that all radial functions \(u\in H^1({\mathbb {R}}^N)\) (\(N\ge 2\)) satisfy
This inequality shows that a control on the \(H^1\) norm of u gives a pointwise bound and decay of u, which are false in the general case. This phenomenon is quite natural, in the sense that symmetric functions can be regarded as functions defined on lower dimensional manifolds, hence satisfying stronger estimates, extended by the action of some group of symmetries. Radial functions are essentially functions on \({\mathbb {R}}^+\), while the norms on \({\mathbb {R}}^N\) introduce suitable dimensional weights connected to the volume form. Related weighted interpolation inequalities are due to Caffarelli et al. [4].
A central role in the analysis developed in this paper is played by the fractional integral
Weighted \(L^p\) estimates for \(T_\mu \) is a fundamental problem of harmonic analysis, with a wide range of applications. Starting from the classical one-dimensional case studied by Hardy and Littlewood, an exhaustive analysis has been made on the admissible classes of weights and ranges of indices (see [30] and the references therein). In the special case of power weights the optimal result is due to Stein and Weiss [31], which established the following weighted Hardy–Littlewood–Sobolev inequality, which is now called the Stein–Weiss inequality.
Proposition 1.1
(Weighted HLS inequality [27]) Let \(1<t,r<\infty \), \(0<\mu <N\), \(\alpha +\beta \ge 0\) and \(0<\alpha +\beta +\mu \le N\), \(f\in L^{t}({\mathbb {R}}^N)\), and \(h\in L^{r}({\mathbb {R}}^N)\). There exists a sharp constant \(C_{ t,r,\alpha , \beta ,\mu , N}\) such that
where
and
where C is independent of f and h. Moreover, for any \(h\in L^{r}({\mathbb {R}}^N)\), we have
where s satisfies \(1+\frac{1}{s}=\frac{1}{r}+\frac{\alpha +\beta +\mu }{N}\) and \(\frac{\alpha }{N}<\frac{1}{s}<\frac{\alpha +\mu }{N}\).
This inequality reduces to the classical Hardy–Littlewood–Sobolev inequality if \(\alpha =\beta =0\). In this case, Lieb [28] applied the Riesz rearrangement inequalities to prove that the best constant for the classical Hardy–Littlewood–Sobolev inequality can be achieved by some extremals. Lieb also classified the solutions of the integral equation
as an open problem. Dou and Zhu [12] classified the extremal functions of the reversed HLS inequality and they computed the best constant. In fact, Eq. (1.1) arises as an Euler–Lagrange equation for a functional under a constraint in the context of the Hardy–Littlewood–Sobolev inequality and is closely related to the well-known fractional equation
When \(N\ge 3\), \(\tau =2\), Eq. (1.2) goes back to
which is a special case of the Lane–Emden equation
The classification of the solutions of Eq. (1.3) and the related best Sobolev constant play an important role in the Yamabe problem, which is the prescribed scalar curvature problem on Riemannian manifolds. It is well known that for \(0<p<\frac{N+2}{N-2}\), Gidas and Spruck [16] proved that Eq. (1.4) has no positive solutions. This result is optimal in the sense that for any \(p\ge \frac{N+2}{N-2}\), there are infinitely many positive solutions to (1.4). Gidas et al. [15], Caffarelli et al. [3] proved the symmetry and uniqueness of the positive solutions respectively. Chen and Li [6], Li [21] simplified the results above as an application of the moving plane method. Wei and Xu [33] generalized the classification of the solutions of the more general Eq. (1.2) with \(\tau \) being any even number between 0 and N. Later on, Chen et al. [8] developed the method of moving planes in integral form in order to prove that any critical points of the functional is radially symmetric and they gave a positive answer to the Lieb open problem involving Eq. (1.1). Li [24] also studied the regularity of the locally integrable solution for problem (1.1) and used the moving sphere method to establish the classification of solutions.
For the doubly weighted case, Lieb [28] proved the existence of a sharp constant, provided that either one of r and t equals 2 or \(r=t\). For \(1<r,t<\infty \) with \(\frac{1}{r}+\frac{1}{t}=1\), the sharp constant is given by Beckner in [1, 2]. It is well known that the corresponding Euler-Lagrange equations for the Stein–Weiss inequality are the system of integral equations
where \(0<p,q<+\infty \), \(0<\mu <N\), \(\frac{\alpha }{N}<\frac{1}{p+1}<\frac{\mu +\alpha }{N}\) and \(\frac{1}{p+1}+\frac{1}{q+1}=\frac{\mu +\alpha +\beta }{N}\). In [10] and [17], the authors obtained the symmetry, monotonicity and the optimal integrability of solutions to problem (1.5). In the special case when \(\mu =N-2\), since the integral system (1.5) is equivalent to the nonlinear singular PDE system
Chen and Li [11] proved the uniqueness of the solutions and classified solutions of problem (1.6) if \(\alpha =\beta \) and \(p=q\). Next, Lei et al. [20] studied the asymptotic radial symmetry and growth estimates of positive solutions for (1.5). Liu and Lei [26] discussed the nonexistence results for (1.5), and they also considered the existence of positive solutions for the following weighted system with double bounded coefficients
where \(0<p,q<+\infty \), \(0<\mu +\alpha +\beta <N\), \(\frac{\alpha }{N}<\frac{1}{p+1}<\frac{\mu +\alpha }{N}\) and \(\frac{\beta }{N}<\frac{1}{q+1}<\frac{\mu +\beta }{N}\).
More generally, Chen et al. [5] established the symmetry and regularity results related to the weighted Hardy–Sobolev type system
where
and nonnegative constants \(\lambda _{i}\), \(\mu _{i}\), \(\gamma _{i}\) \((i=1,2)\) are not equal to zero simultaneously. For the special case of (1.8) corresponding to \(\alpha =\beta =0\), there are some contributions on the system of integral equations
where \(0<\gamma <N\), \(1\le p,q\le \frac{N+\gamma }{N-\gamma }\) with \(p+q\le \frac{N+\gamma }{N-\gamma }\). When \(\gamma =2\), Li and Ma [22] proved the symmetry and uniqueness of the positive solutions for (1.9) with critical exponents \(p+q=\frac{N+2}{N-2}\). Furthermore, Yu [34] studied the more general integral system
where f, g satisfy the following monotonicity conditions: \(f(s_{1},s_{2})\) and \(g(s_{1},s_{2})\) are nondecreasing in \(s_{i}\) for fixed \(s_{j}\) and, additionally, \(\frac{f(s_{1},s_{2})}{s_{1}^{p_{1}}s_{2}^{q_{1}}}\) and \(\frac{g(s_{1},s_{2})}{s_{1}^{p_{2}}s_{2}^{q_{2}}}\) are nondecreasing in \(s_{i}\) for fixed \(s_{j}\) with \(p_{i},q_{i}\ge 0\) and \(p_{i}+q_{i}=\frac{N+\gamma }{N-\gamma }\).
From the weighted Hardy–Littlewood–Sobolev inequality with \(\alpha =\beta \) and \(t=r\), assuming that \(|u|^{p}\in L^{t}({\mathbb {R}}^N)\), we easily get the following inequality
where t satisfies
Furthermore, if \(u\in H^{1}({\mathbb {R}}^N)\), from the Sobolev embedding theorems, we have
and hence
Accordingly, the critical exponent \(2_{\alpha ,\mu }^{*}:=\frac{2N-2\alpha -\mu }{N-2}\) (\(2_{*\alpha ,\mu }:=2-\frac{2\alpha +\mu }{N}\)) is called the upper (lower) critical exponent in the sense of the weighted Hardy–Littlewood–Sobolev inequality. It is obvious that
Therefore, we see that the best constant problem of (1.11) is related to the following critical nonlocal Hartree equation
which is a special case of the weighted Choquard equation
The classification of solutions to problem (1.13) has attracted a lot of interest recently. If \(\alpha =0\), Eq. (1.13) reduces to
Miao et al. [29] established the existence of solutions of (1.14) if \(p=2\), \(\mu =4\) and \(N\ge 5\). For the symmetry and uniqueness of solutions for the nonlocal Hartree equation, by using the moving plane method introduced in [8, 9], Liu [25], Lei [19] and Du and Yang [14] classified the positive solutions of problem (1.14) with the critical exponent \(\frac{2N-\mu }{N-2}\). Moreover, Du and Yang [14] also proved the nondegeneracy of the unique solutions for the equation when \(\mu \) is close to N. As applications, Ding et al. [35] investigated the existence of semiclassical solutions of the critical Choquard equation with critical frequency.
The readers may turn to [18, 19] and the references therein for more backgrounds about the Hartree type equations. For (1.13) with \(\alpha \ne 0\), the authors in [13] proved the existence of positive ground state solutions the critical equation by a nonlocal version of the concentration-compactness principle. They also established the regularity of positive solutions and proved the symmetry of these solutions by the moving plane method in integral form [7]. Finally, we recall that Li et al. [23] studied the equation without variational structure and classified the nonpositive solutions.
2 Main results
This paper is devoted to the study of some qualitative properties to the positive solutions of three nonlocal elliptic systems with weighted Stein–Weiss type convolution part. We first consider the following nonlocal system without a variational structure
where \(N\ge 3\), \(\alpha \ge 0\), \(0<\mu <N\), p, \(q>1\) and \(0<2\alpha +\mu \le N\).
In Sect. 3, by investigating an equivalent integral system with Riesz potential, we are able to prove some qualitative properties of the positive solutions for problem (2.1). In fact, we obtain the symmetry result for the positive solutions of (2.1) via the moving plane arguments of integral form, which can be easily applied to more complicated equations without maximum principles.
Theorem 2.1
Suppose that \(N\ge 3\), \(\alpha \ge 0\), \(0<\mu <N\), p, \(q>1\) and \(0<2\alpha +\mu \le N\). If \((u,v)\in L^{s_{0}}({\mathbb {R}}^N)\times L^{s_{0}}({\mathbb {R}}^N) \) is a pair of positive solutions of system (2.1) with \(s_{0}=\frac{N(p+q-1)}{N+2-2\alpha -\mu }\), then u and v are radially symmetric and decreasing about the origin.
If \(p+q=2\cdot 2_{\alpha ,\mu }^{*}-1\), then \(s_{0}=\frac{2N}{N-2}\), and so \((u,v) \in L^{2^{*}}({\mathbb {R}}^N)\times L^{2^{*}}({\mathbb {R}}^N)\). Assuming that p, q lie in some suitable intervals depending on the parameters \(\alpha \), \(\mu \), then we can apply the regularity lifting lemma [9] to prove that the positive integral solutions possess better integral properties.
Theorem 2.2
Suppose that \(N=3\), 4, 5, 6, \(\alpha \ge 0\), \(0<\mu <N\) and \(N-2\le 2\alpha +\mu \le N \). Let \((u,v) \in L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\) be a pair of positive solutions of system (2.1), where p, q satisfy \(\frac{2(N-2\alpha -\mu )}{N-2}\le p,q\le \min \left\{ \frac{4}{N-2},\frac{N+6-2(2\alpha +\mu )}{N-2}\right\} \) and \(p+q=2\cdot 2_{\alpha ,\mu }^{*}-1\). Then \((u,v)\in L^{s}({\mathbb {R}}^{N})\times L^{s}({\mathbb {R}}^{N})\) with
By using the symmetry and regularity results obtained above, we establish the asymptotic behaviour of solutions at infinity.
Theorem 2.3
Suppose that \(N=3\), 4, 5, 6, \(\alpha \ge 0\), \(0<\mu <N\), p, \(q>1\) and \(N-2\le 2\alpha +\mu \le N \). Let \((u,v) \in L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\) be a pair of positive solutions of system (2.1). If p, q satisfy \(p+q=2\cdot 2_{\alpha ,\mu }^{*}-1\) and \(\frac{2(N-2\alpha -\mu )}{N-2}\le p,q\le \min \left\{ \frac{4}{N-2}, \frac{N+2+2(N+2-2\alpha -\mu )}{N-2}\right\} \), then the following properties hold.
-
(1)
If \(0\le \alpha <2\), then both u(x) and v(x) are bounded and, moreover, we have u(x), \(v(x)\in C^{\infty }({\mathbb {R}}^N-\{0\})\).
-
(2)
For large |x|, we have \(u(x)\backsimeq \frac{C}{|x|^{N-2}}\) and \(v(x)\backsimeq \frac{C}{|x|^{N-2}}\).
Next, we are interested in the following nonlocal system with variational structure
where \(N\ge 3\), \(\alpha \ge 0\), \(0<\mu <N\), p, \(q>1\) and \(0<2\alpha +\mu \le N\). Notice that system (2.2) becomes (1.13) if \(p=q\) and \(u=v\), but problem (2.2) has not been well studied if \(p\ne q\).
In Sect. 4, we are concerned with the nonexistence of positive solutions to system (2.2), provided that \(p+q=2\cdot 2_{\alpha ,\mu }^{*}\), which is called a critical condition. We first prove that system (2.2) has no positive solutions in the subcritical case.
Theorem 2.4
Assume that \(N\ge 3\), \(\alpha \ge 0\), \(0<\mu <N\), p, \(q>1\) and \(0<2\alpha +\mu \le N\). Let (u, v) \(\in W_{loc}^{2,2}({\mathbb {R}}^N)\times W_{loc}^{2,2}({\mathbb {R}}^N) \) be a pair of solutions of (2.2). If \(p+q<2\cdot 2_{\alpha ,\mu }^{*}\), then \(u\equiv v\equiv 0\).
Analogously to the arguments for problem (2.1), we can also draw the conclusions for the system (2.2), such as symmetry, regularity and asymptotic behavior. Here we shall assume that u, v are integrable solutions belonging to \(L^{s_{0}}({\mathbb {R}}^N)\) with \(s_{0}=\frac{N(p+q-2)}{N+2-2\alpha -\mu }\).
We establish the following symmetry result.
Theorem 2.5
Suppose that \(N\ge 3\), \(\alpha \ge 0\), \(0<\mu <N\), p, \(q\ge 2\) and \(0<2\alpha +\mu \le N \). Let \((u,v)\in L^{s_{0}}({\mathbb {R}}^N)\times L^{s_{0}}({\mathbb {R}}^N) \) be a pair of positive solutions of system (2.2) with \(s_{0}=\frac{N(p+q-2)}{N+2-2\alpha -\mu }\). Then u and v are radially symmetric and decreasing about the origin.
As we can see, for the critical case \(p+q=2\cdot 2_{\alpha ,\mu }^{*}\), we get \((u,v) \in L^{2^{*}}({\mathbb {R}}^N)\times L^{2^{*}}({\mathbb {R}}^N)\). Hence, arguing in the same way as Theorems 2.2 and 2.3, the regularity and decay properties are stated as follows.
Theorem 2.6
Suppose that \(N=3\), 4, 5, 6, \(\alpha \ge 0\), \(0<\mu <N\), p, \(q\ge 2\) and \(N-2\le 2\alpha +\mu \le N \). Let \((u,v) \in L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\) be a pair of positive solutions of system (2.2). If p, q satisfy \(p+q=2\cdot 2_{\alpha ,\mu }^{*}\) and \(\frac{2(N-2\alpha -\mu )}{N-2}\le p-1,q-1\le \min \left\{ \frac{4}{N-2}, \frac{N+2+2(N+2-2\alpha -\mu )}{N-2}\right\} \), then \((u,v)\in L^{s}({\mathbb {R}}^N)\times L^{s}({\mathbb {R}}^N)\) with
Theorem 2.7
Suppose that \(N=3\), 4, 5, 6, \(\alpha \ge 0\), \(0<\mu <N\), p, \(q\ge 2\) and \(N-2\le 2\alpha +\mu \le N \). Let \((u,v) \in L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\) be a pair of positive solutions of system (2.2). If p, q satisfy \(p+q=2\cdot 2_{\alpha ,\mu }^{*}\) and \(\frac{2(N-2\alpha -\mu )}{N-2}\le p-1,q-1\le \min \left\{ \frac{4}{N-2}, \frac{N+2+2(N+2-2\alpha -\mu )}{N-2}\right\} \), then the following properties hold true.
-
(1)
If \(0\le \alpha <2\), then both u(x) and v(x) are bounded and, moreover, we have u, \(v\in C^{\infty }({\mathbb {R}}^N-\{0\})\).
-
(2)
For large |x|, we have \(u(x)\backsimeq \frac{C}{|x|^{N-2}}\) and \(v(x)\backsimeq \frac{C}{|x|^{N-2}}\).
Finally, we study the following Hamiltonian-type system
where \(N\ge 3\), \(0<\mu _{1},\mu _{2}<N\), \(\alpha _{1}\), \(\alpha _{2}\ge 0\), \(0<2\alpha _{1}+\mu _{1}\le N\), \(0<2\alpha _{2}+\mu _{2}\le N\) and p, \(q>1\).
In the last section, we verify the symmetry of positive solutions of the Hamiltonian system (2.3) with convolution part.
Theorem 2.8
Suppose that \(N\ge 3\), \(\alpha _{i}\ge 0\), \(0<\mu _{i}<N\) and \(0<2\alpha _{i}+\mu _{i}\le min\left\{ 4, N\right\} \), \(i=1\), 2. If \((u,v)\in D^{1,2}({\mathbb {R}}^N)\times D^{1,2}({\mathbb {R}}^N)\) is a pair of positive solutions of (2.3) and \((p, q)=\left( \frac{2N-2\alpha _{1}-\mu _{1}}{N-2}, \frac{2N-2\alpha _{2}-\mu _{2}}{N-2}\right) \), then u and v are radially symmetric and decreasing about the origin.
An outline of the paper is as follows. In Sect. 3 we mainly focus on the nonlocal Hartree system (2.1). By translating the equation into an equivalent integral system, we apply a regularity lifting lemma to obtain the regularity of the solutions and the moving plane methods in integral form to study the symmetry of the positive solutions. Besides these, the decay at infinity is also shown by careful estimates. In Sect. 4 we will study system (2.2). Firstly, by establishing a Pohožaev identity, we prove a non-existence result. In this part we will also prove the regularity of the solutions by some iterative arguments and singular integral analysis. Finally, we prove the symmetry of solutions for the Hamiltonian system (2.3). This is done by using the moving plane method in integral form.
3 Qualitative properties for the nonlocal system (2.1)
In this section, we discuss the qualitative properties of system (2.1) with the critical condition \(p+q=2\cdot 2_{\alpha ,\mu }^{*}-1\), including symmetry, regularity and asymptotic behavior at infinity. It is worth noting that system (2.1) is equivalent to the following integral system in \({\mathbb {R}}^{N}\),
where \(R_{N}=\frac{\varGamma (\frac{N-2}{2})}{4\pi ^{\frac{N}{2}}}\).
3.1 Symmetry
In this subsection, we establish the symmetry of the positive solutions for (3.1) by means of the moving plane method in integral forms developed by Chen et al. [8]. We start this part with some basic definitions. For \(\lambda \in {{\mathbb {R}}}\), define
and
By straightforward computation we obtain
and
Since \(|x^{\lambda }-y^{\lambda }|=|x-y|\) and \(|x^{\lambda }-y|=|x-y^{\lambda }|\), it follows that
and
For p, \(q>1\), we have the following estimates.
Lemma 3.1
Under the assumption of Theorem 2.1, for any \(\lambda <0\), there exists a constant \(C>0\) such that
and
Proof
For any \(x\in \Sigma _{\lambda }\), notice \(|x^{\lambda }-y|\ge |x-y|\), by using the mean value theorem, from (3.2) and (3.3), we easily deduce
and
In virtue of \((u,v)\in L^{s_{0}}({\mathbb {R}}^N)\times L^{s_{0}}({\mathbb {R}}^N) \) with \(s_{0}=\frac{N(p+q-1)}{N+2-2\alpha -\mu }\), we suppose that \(z\in L^{k}({\mathbb {R}}^N)\), \(h\in L^{t}({\mathbb {R}}^N)\), where k and t satisfy
By applying the HLS inequality and the Hölder inequality, we have
From (3.6), we obtain
Analogously, we have
and
from which we can deduce that
and
Additionally, using the weighted HLS inequality again, we have
and
We deduce that \(z\in L^{k}(\Sigma _{\lambda }^{u})\) and \(h\in L^{t}(\Sigma _{\lambda }^{v})\).
Combining (3.11), (3.12) with (3.7), we see that (3.4) holds. Similarly, we obtain
Thus, inserting (3.10) and (3.13) into (3.14), we complete the proof. \(\square \)
Lemma 3.2
Under the assumption of Theorem 2.1, there exists \(M>0\) such that for any \(\lambda <-M\), we have
Proof
Since u, v are integrable, letting \(\lambda \rightarrow -\infty \), we have
and
where the constant C is the same as in Lemma 3.1. Hence, inserting those inequalities into (3.4) and (3.5), as \(\lambda \rightarrow -\infty \), it follows that
which shows that \(\Sigma _{\lambda }^{u}=\Sigma _{\lambda }^{v}=\emptyset .\) Therefore, there exists \(M>0\) such that for any \(\lambda <-M\), relation (3.15) holds. \(\square \)
We now can move the plane \(T_{\lambda }=\left\{ x\in {\mathbb {R}}^N|x_{1}=\lambda \right\} \) to the right as long as (3.15) is satisfied. Naturally, denote
We observe that \(\lambda _{0}<+\infty \).
Next, we deduce the following auxiliary property.
Lemma 3.3
Under the assumption of Theorem 2.1, then for any \(\lambda _{0}<0\), we have
Proof
Suppose that at \(\lambda _{0}<0\), there holds \(u(x)\le u_{\lambda _{0}}(x)\) and \(v(x)\le v_{\lambda _{0}}(x)\), but \(u(x)\not \equiv u_{\lambda _{0}}(x)\) or \(v(x)\not \equiv v_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}} \).
We claim that there exists \(\varepsilon >0\) such that \(u(x)\le u_{\lambda }(x)\) and \(v(x)\le v_{\lambda }(x)\) on \(\Sigma _{\lambda } \) for any \(\lambda \in \left[ \lambda _{0},\lambda _{0}+\varepsilon \right) \).
Indeed, for any \(\eta >0\), we can choose \(R>0\) large enough such that
and
For such \(R>0\) and \(\lambda >\lambda _{0}\), we show that the measures of the sets \(\Sigma _{\lambda }^{u}\cap B_{R}(0)\) and \(\Sigma _{\lambda }^{v}\cap B_{R}(0)\) go to 0 as \(\lambda \rightarrow \lambda _{0}\).
Assume that
From (3.8), we obtain
Thus, by (3.3), we yield
Combining with (3.9), it follows that
From (3.2), we have
Naturally, for any \(\delta >0\), we define
and
Obviously, we get
and
where \({\mathcal {L}}\) is the Lebesgue measure. For any \( x \in \Sigma _{\lambda }^{u}\cap D_{\delta }\), since
we have
Hence, by the Chebyshev inequality, for fixed \(\delta >0\), we obtain that
if \(\lambda \rightarrow \lambda _{0}\). Notice that
From (3.19)–(3.21), as \(\lambda \rightarrow \lambda _{0}\) and \(\delta \rightarrow 0\), we can easily get
Analogously, we obtain
Combining (3.22), (3.23), (3.17) with (3.18), there exists \(\varepsilon >0\) such that for any \(\lambda \in [ \lambda _{0},\lambda _{0}+\varepsilon )\),
and
where the constant C is the same as in Lemma 3.1. By the same arguments as above, we can conclude that \(\Sigma _{\lambda }^{u}=\Sigma _{\lambda }^{v}=\emptyset \). Therefore, there exists \(\varepsilon >0\) such that \(u(x)\le u_{\lambda }(x)\) and \(v(x)\le v_{\lambda }(x)\) on \(\Sigma _{\lambda } \) for any \(\lambda \in \left[ \lambda _{0},\lambda _{0}+\varepsilon \right) \). This contradicts the definition of \(\lambda _{0}\), hence we obtain \(u(x)\equiv u_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}}\).
Similarly, if \(v(x)\not \equiv v_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}}\), which is also a contradiction. The proof is completed. \(\square \)
3.2 Proof of Theorem 2.1
Clearly, we can also move the plane from \(+\infty \) to left, and define
where \(\Sigma _{\rho }^{'}=\left\{ x\in {\mathbb {R}}^N|x_{1}>\rho \right\} \).
If \(\lambda _{0}=\lambda _{1}\ne 0\), then both u and v are radially symmetric and decreasing about the plane \(x_{1}=\lambda _{0}\), which implies \(u(x)\equiv u_{\lambda _{0}}(x)\) and \(v(x)\equiv v_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}}\). Since \(|x-y|<|x^{\lambda _{0}}-y|\) and \(|y|>|y^{\lambda _{0}}|\), we deduce from (3.9) that
Therefore, we obtain that
which is impossible. Hence, we get \(\lambda _{0}=\lambda _{1}=0\). Notice that the direction of \(x_{1}\) is arbitrary, hence u, v are radially symmetric and decreasing about origin. \(\square \)
3.3 Regularity
Since the integrability and the regularity play an essential role in estimating the decay rates of u(x) if \(|x|\rightarrow \infty \), it is necessary for us to discuss the integrability of solutions to system (2.1) by applying the regularity lifting theorem (see in [9, Theorem 3.3.1]) to (2.1).
Lemma 3.4
Let X and Y be Banach spaces with norms \(\Vert \cdot \Vert _{X}\) and \(\Vert \cdot \Vert _{Y}\), respectively. The subspace \(Z=X\cap Y\) of X and Y, is endowed with a new norm by
Suppose that \({\mathscr {T}}\) is a contraction map from Banach space X into itself and from Banach space Y into itself. If \(f\in X\) and there exists a function \(g\in Z=X\cap Y\) such that \(f={\mathscr {T}}f+g\), then f also belongs to Z.
For some constant \( A>0 \), we define
and \(u_{B}(x)=u(x)-u_{A}(x)\). Similarly, we can define \( v_{A}(x) \) and \(v_{B}(x)\). Then we define the functions
Suppose a, \(b \in L^{s}({\mathbb {R}}^{N})\), \(m\in L^{k}({\mathbb {R}}^{N})\), and \(h \in L^{t}({\mathbb {R}}^{N}) \). We define the operator
with the norm
Hence, we deduce that (u, v, z, h) satisfies the operator equation
Next, we will obtain the main result of Theorem 2.2 by proving the following two lemmas.
Lemma 3.5
Assume that \(p+q=2\cdot 2_{\alpha ,\mu }^{*}-1\) and s, k, t satisfy
Then for A sufficiently large, \({\mathscr {T}}_{A}\) is a contraction map from \(L^{s}({\mathbb {R}}^{N})\times L^{s}({\mathbb {R}}^{N})\times L^{k}({\mathbb {R}}^{N})\times L^{t}({\mathbb {R}}^{N})\) to itself.
Proof
Since \(s>\frac{N}{N-2}\), \(2N>[q(N-2)-4]s\) and \(\frac{1}{s}-\frac{1}{k}=\frac{q(N-2)-4}{2N}\), by the Hardy–Littlewood–Sobolev inequality and the Hölder inequality, we have
Similarly, from \(2N>[p(N-2)-4]s\) and \(\frac{1}{s}-\frac{1}{t}=\frac{p(N-2)-4}{2N}\), we get
In addition, notice that \(s>\frac{2N}{2N+p(N-2)-2(N+2-2\alpha -\mu )}\) and \( s>\frac{2N}{2N+q(N-2)-2(N+2-2\alpha -\mu )}\). Using the weighted HLS inequality and the Hölder inequality, we obtain
and
By virtue of \(u,v\in L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\), we can choose A large enough such that
Thus, \({\mathscr {T}}_{A}\) is a contraction map from \(L^{s}({\mathbb {R}}^{N})\times L^{s}({\mathbb {R}}^{N})\times L^{k}({\mathbb {R}}^{N})\times L^{t}({\mathbb {R}}^{N})\) to itself. \(\square \)
Lemma 3.6
Suppose that \(3\le N\le 6\), \(\alpha \ge 0\), \(0<\mu <N\) and \(N-2\le 2\alpha +\mu \le N \). Let \((u,v,z,h) \in L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{k_{0}}({\mathbb {R}}^N)\times L^{t_{0}}({\mathbb {R}}^N)\) be a set of positive solutions of system (3.1) with \(k_{0}=\frac{2N}{4-(q-1)(N-2)}\) and \(t_{0}=\frac{2N}{4-(p-1)(N-2)}\), where p, q satisfying \(\frac{2(N-2\alpha -\mu )}{N-2}\le p,q\le \min \left\{ \frac{4}{N-2},\frac{N+6-2(2\alpha +\mu )}{N-2}\right\} \) and \(p+q=2\cdot 2_{\alpha ,\mu }^{*}-1\). Then \((u,v,z,h)\in L^{s}({\mathbb {R}}^{N})\times L^{s}({\mathbb {R}}^{N})\times L^{k}({\mathbb {R}}^{N})\times L^{t}({\mathbb {R}}^{N})\) with
Proof
Firstly, under the assumption of Lemma 3.5, we claim that
In fact, we know that \(|u_{B}|\le A\) and \(u_{B}=0\) for \(|x|>A\). Following the same estimates as in the proof of the above Lemma, we easily have
and
where \(A_{1}=\frac{2Nsq}{2N+[2(N+2-2\alpha -\mu )-p(N-2)]s}\), \(A_{2}=\frac{2Nsp}{2N+[2(N+2-2\alpha -\mu )-q(N-2)]s}\), \(k_{0}=\frac{2N}{4-(q-1)(N-2)}\) and \(t_{0}=\frac{2N}{4-(p-1)(N-2)}\) satisfy \(2N+[2(N+2-2\alpha -\mu )-p(N-2)]s>0\), \(2N+[2(N+2-2\alpha -\mu )-q(N-2)]s>0\), \(4-(q-1)(N-2)>0\) and \(4-(p-1)(N-2)>0\). Moreover, we can get
and
where \(A_{3}=\frac{2Ns(q-1)}{2N+[2(N+2-2\alpha -\mu )-(p+1)(N-2)]s}\), \(A_{4}=\frac{2Ns(p-1)}{2N+[2(N+2-2\alpha -\mu )-(q+1)(N-2)]s}\), and we require \(2N+[2(N+2-2\alpha -\mu )-(p+1)(N-2)]s>0\) and \(2N+[2(N+2-2\alpha -\mu )-(q+1)(N-2)]s>0\). Hence, we have
which implies \({\mathscr {T}}_{A}\) is also a contraction map from \(L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{k_{0}}({\mathbb {R}}^N)\times L^{t_{0}}({\mathbb {R}}^N)\) to itself. Write
and
Evidently, if s, k and t satisfy
we deduce that \( (u,v,z,h)\in L^{s}({\mathbb {R}}^{N})\times L^{s}({\mathbb {R}}^{N})\times L^{k}({\mathbb {R}}^{N})\times L^{t}({\mathbb {R}}^{N})\) by the regularity lifting theorem. From (3.24), if \(\frac{1}{s}-\frac{1}{k}=\frac{q(N-2)-4}{2N}\) , \(\frac{1}{s}-\frac{1}{t}=\frac{p(N-2)-4}{2N}\), then s, k, t should satisfy
More accurately, we deduce that if p, q satisfy \(\frac{2(N-2\alpha -\mu )}{N-2}<p,q\le min\left\{ \frac{4}{N-2},\right. \left. \frac{N+6-2(2\alpha +\mu )}{N-2}\right\} \), then \((u,v,z,h)\in L^{s}({\mathbb {R}}^{N})\times L^{s}({\mathbb {R}}^{N})\times L^{k}({\mathbb {R}}^{N})\times L^{t}({\mathbb {R}}^{N})\) with
The proof is now complete. \(\square \)
3.4 Decay
In this part, we will show the decay rate of the solutions of the critical weighted Hartree system (2.1).
3.5 Proof of Theorem 2.3
We first prove that \(|x|^{\alpha }h(x)\in L^{\infty }({\mathbb {R}}^{N})\). It is obvious that
Thus, for any \(r>0\), we obtain
On the one hand, for \(x\in {\mathbb {R}}^{N}-B_{2r}(0)\), we have \(|x-y|>|y|\), we have
where \(1<k<min\left\{ \frac{N}{N-q(N-2)}, \frac{N}{\mu +\alpha }\right\} \) if \(N-q(N-2)>0\), while \(1<k< \frac{N}{\mu +\alpha }\) if \(N-q(N-2)\le 0\). For \(x\in B_{2r}(0)\), we have
Thus, we can obtain that
On the other hand,
As in the preceding estimates, we have
where \(1<k<\min \left\{ \frac{N}{N-q(N-2)}, \frac{N}{\mu }\right\} \) if \(N-q(N-2)>0\), while \(1<k< \frac{N}{\mu }\) if \(N-q(N-2)\le 0\).
We observe that
where \(\frac{N}{\alpha }\le k< \frac{N}{N-q(N-2)}\) if \(N-q(N-2)>0\), while \(k\ge \frac{N}{\alpha }\) if \(N-q(N-2)\le 0\). Therefore, we get
By (3.25), (3.26), (3.27), we can conclude that
Secondly, we claim that \(v(x)\in L^{\infty }({\mathbb {R}}^{N})\). From (3.1) and (3.28), we have
For any \(r>0\), we decompose as follows
On the one hand, for \(x\in {\mathbb {R}}^{N}-B_{2r}(0)\),
where \(1<k<\min \left\{ \frac{N}{N-p(N-2)}, \frac{N}{N-2+\alpha }\right\} \) if \(N-p(N-2)>0\), while \(1<k< \frac{N}{N-2+\alpha }\) if \(N-p(N-2)\le 0\). However, for \(x\in B_{2r}(0)\),
Consequently, we have
On the other hand,
Clearly,
In addition, we also have
Since
where \(\frac{N}{N-2+\alpha }<k<\frac{N}{N-p(N-2)}\) if \(N-p(N-2)>0\), while \(k>\frac{N}{N-2+\alpha }\) if \(N-p(N-2)\le 0\), from which we conclude that
Combining with (3.32)–(3.34), we obtain
Through (3.29), (3.30), (3.31), (3.35), we deduce that
Finally, we prove that u(x), \(v(x)\in C^{\infty }({\mathbb {R}}^N-\{0\})\). For any \(x\in {\mathbb {R}}^{N}-\{0\}\), we decompose v(x) as follows
where \(r<\frac{|x|}{2}\). It has been established in [27, Chapter 10] that for any \(\delta <2\),
If we can obtain
then together with (3.36) we can conclude that \(v(x)\in C^{\delta }({\mathbb {R}}^{N}-\{0\})\). Thus, combining the classical bootstrap technique [27, Chapter 10], we prove that \(v(x)\in C^{\infty }({\mathbb {R}}^{N}-\{0\})\).
In the following we will show that
Define
We claim that
Indeed, for any small \(t<r\), \(0<\theta <1\) and if \(e_{i}\) is the unit ith vector, then
Since \(|x|^{\alpha }h(x)\), \(v(x)\in L^{\infty }({\mathbb {R}}^N)\), it follows that
Clearly, we also see that
Then by the regularity result we have
where the parameters \(\alpha \), \(\mu \), k satisfy one of following four cases:
-
(1)
if \(N-p(N-2)>0\), \(\alpha \le 1\), then \(\frac{N}{N-1+\alpha }<k<\frac{N}{N-p(N-2)}\);
-
(2)
if \(N-p(N-2)>0\), \(1<\alpha <2\), then \(1<k<\frac{N}{N-p(N-2)}\);
-
(3)
if \(N-p(N-2)\le 0\), \(\alpha \le 1\), then \(k>\frac{N}{N-1+\alpha }\);
-
(4)
if \(N-p(N-2)\le 0\), \(1<\alpha <2\), then \(k>1\).
Therefore we can deduce that
Thus, from (3.37), (3.38) and the Lebesgue dominated convergence theorem, we can conclude that \(\psi (x)\in C^{1}({\mathbb {R}}^{N}-\{0\})\). Repeating the above process, we can deduce \(\psi (x)\in C^{\infty }({\mathbb {R}}^{N}-\{0\})\), which implies \( K_{2}\in C^{\infty }({\mathbb {R}}^{N}-\{0\}) \), so that \(v(x)\in C^{\infty }({\mathbb {R}}^N-\{0\})\). Similarly, we have \(|x|^{\alpha }z(x)\in L^{\infty }({\mathbb {R}}^N)\). Therefore, we can also obtain \(u(x)\in L^{\infty }({\mathbb {R}}^N)\) and \(u(x)\in C^{\infty }({\mathbb {R}}^N-\{0\})\).
Write \(A=R_{N}\int _{{\mathbb {R}}^{N}} z(y)u^{q}(y)dy\). From Lemma 3.6, we have \((u,v,z,h)\in L^{s}({\mathbb {R}}^N)\times L^{s}({\mathbb {R}}^N)\times L^{k}({\mathbb {R}}^N)\times L^{t}({\mathbb {R}}^N)\), where s, k and t satisfy
We can take
where \(\varepsilon >0\) sufficiently small. Applying the Hölder inequality, we have
For fixed \(R>0\),
For large |x|, by the Lebesgue dominated convergence theorem and
we can see that \(\lim \limits _{|x|\rightarrow \infty }|M_{1}|=0\).
Decompose \(M_{2}\) into two parts by
and
Since \(|x-y|\ge \frac{|x|}{2}\) when \(y\in ({\mathbb {R}}^N - B_{R}(0))- B_{\frac{|x|}{2}}(x)\), we have
which implies \(M_{21}\rightarrow 0\) as \(R\rightarrow +\infty \).
In the following, we estimate \(M_{22}\) as \(|x|\rightarrow +\infty \). Clearly, from Theorem 2.1 we know u, v, z, h are radially symmetric and decreasing about \(x_{0}=0\). Then we can write
and
Notice that \(\frac{|x|}{2}<|y|<\frac{3|x|}{2}\) for \(y\in B_{\frac{|x|}{2}}(x)\), we deduce that
and
Therefore,
By choosing \(\frac{1}{s}=\frac{N-2}{2N}+\frac{2\varepsilon }{2Nq}\) and \(\frac{1}{k}=\frac{2N-q(N-2)-\varepsilon }{2N}\) with sufficiently small \(\varepsilon >0\) such that \(\frac{q}{s}+\frac{1}{k}>1\), together with the integrability results, we get \((u,z)\in L^{s}({\mathbb {R}}^N)\times L^{k}({\mathbb {R}}^N)\).
Since u, z are decreasing about \(x_{0}=0\), we have
and
from which we conclude that
Inserting (3.40) into (3.39), as \(|x|\rightarrow +\infty \), we have
Therefore, we conclude that
which implies \(u(x)\backsimeq \frac{C}{|x|^{N-2}}\) as \(|x|\rightarrow +\infty \). Similarly, we have \(v(x)\backsimeq \frac{C}{|x|^{N-2}}\) as \(|x|\rightarrow +\infty \). The proof is completed. \(\square \)
4 Conclusions for the variational system (2.2)
In this section, we are going to study the nonlocal variational system (2.2). By using similar arguments as for the system (2.1), we can also prove symmetry and regularity properties, as well as the decay of the positive solutions to system (2.2). Furthermore, we establish the nonexistence results under the subcritical condition.
4.1 Nonexistence results for the subcritical case
We first obtain the corresponding Poho\({\check{z}}\)aev type identity for the subcritical case of system (2.2).
Lemma 4.1
Assume that \(N\ge 3\), \(0<\mu <N\), \(\alpha \ge 0\) and \(0<2\alpha +\mu \le N\). Let (u, v) \(\in W_{loc}^{2,2}({\mathbb {R}}^N)\times W_{loc}^{2,2}({\mathbb {R}}^N) \) be a pair of solutions of (2.2), then there holds
Proof
We define a cut-off function \(\varphi \in C_{0}^{\infty }({\mathbb {R}}^N)\) with \(0\le \varphi \le 1\), satisfying \(\varphi =1 \) in \(B_{1}(0)\) and \(\varphi =0 \) outside \(B_{1}(0)\). For \(0<\lambda <\infty \) and \(x\in {\mathbb {R}}^N\), we denote
Multiplying the first and the second equation of (2.2) by \(\psi _{u,\lambda }(x)\) and \(\psi _{v,\lambda }(x)\) respectively, and integrating by part, we get
and
But
and
Next, we claim that
Indeed, letting
we have
Then we have
and
A direct calculation shows that
We can also deduce that
Combining (4.4) with (4.5) and adding (4.6) to (4.7), we conclude from the dominated convergence Theorem that
Therefore, taking (4.1)–(4.3) into account, we infer that
The proof is now complete. \(\square \)
4.2 Proof of Theorem 2.4
We first multiply the first equation of (2.2) by u and multiply the second equation of (2.2) by v, then it follows that
and
Together with the identity in Lemma 4.1, we deduce
If \( p+q<\frac{2(2N-2\alpha -\mu )}{N-2}\), we get
The proof is complete. \({} \Box \)
4.3 Qualitative properties for the critical case
Next we focus on the qualitative results of system (2.2) with critical condition \(p+q=\frac{2(2N-2\alpha -\mu )}{N-2}=2\cdot 2_{\alpha ,\mu }^{*}\), including symmetry, regularity and asymptotic behavior at infinity.
4.3.1 Symmetry
Analogously, we consider the following equivalent integral system in \({\mathbb {R}}^{N}\)
For \(\lambda \in {{\mathbb {R}}}\), define
and
We rewrite u(x) and \(u_{\lambda }(x)\) as
and
Since \(|x^{\lambda }-y^{\lambda }|=|x-y|\) and \(|x^{\lambda }-y|=|x-y^{\lambda }|\), it follows that
and
For the case p, \(q>2\), we have the following property.
Lemma 4.2
Under the assumption of Theorem 2.5, for any \(\lambda <0\), there exists a constant \(C>0\) such that
and
Proof
For any \(x\in \Sigma _{\lambda }\), notice \(|x^{\lambda }-y|\ge |x-y|\), using the mean value theorem, from (4.10) and (4.11), we easily deduce
and
In virtue of \((u,v)\in L^{s_{0}}({\mathbb {R}}^N)\times L^{s_{0}}({\mathbb {R}}^N) \) with \(s_{0}=\frac{N(p+q-2)}{N+2-2\alpha -\mu }\), assume that \(z\in L^{k}({\mathbb {R}}^N)\), \(h\in L^{t}({\mathbb {R}}^N)\), where \(k,\ t\) such that
Then we have
By applying the HLS inequality and the Hölder inequality, we lead
Similarly, we obtain
Combining with relations (3.10)–(3.13), we easily obtain (4.12) and (4.13). We complete the proof. \(\square \)
For the case \(p=q=2\) or \(p>2\) and \(q=2\), Lemma 4.2 can be replaced by the following lemmas.
Lemma 4.3
Under the assumption of Theorem 2.5, if \(p=q=2\), for any \(\lambda <0\), there exists a constant \(C>0\) such that
and
Lemma 4.4
Under the assumption of Theorem 2.5, if \(p>2\) and \(q=2\), for any \(\lambda <0\), there exists a constant \(C>0\) such that
and
In the following, we provide the starting of the moving plane methods by using the \(L^{s_{0}}\) estimates proved above.
Lemma 4.5
Under the assumption of Theorem 2.5, there exists \(M>0\) such that for any \(\lambda <-M\), we have
Proof
Since u, v are integrable, if \(\lambda \rightarrow -\infty \), we have
and
where the constant C is the same as in Lemma 4.2. Hence, inserting those inequalities into (4.12) and (4.13), as \(\lambda \rightarrow -\infty \), it follows
which shows that \(\Sigma _{\lambda }^{u}=\Sigma _{\lambda }^{v}=\emptyset .\) Therefore, there exists \(M>0\) such that for any \(\lambda <-M\), relation (4.19) holds. \(\square \)
We now can move the plane \(T_{\lambda }=\left\{ x\in {\mathbb {R}}^N|x_{1}=\lambda \right\} \) to the right as long as (4.19) is satisfied. Setting
we observe that \(\lambda _{0}<+\infty \).
Next, we deduce the following property.
Lemma 4.6
Under the assumption of Theorem 2.5, then for any \(\lambda _{0}<0\), we have
Proof
Suppose that at \(\lambda _{0}<0\), there holds \(u(x)\le u_{\lambda _{0}}(x)\) and \(v(x)\le v_{\lambda _{0}}(x)\), but \(u(x)\not \equiv u_{\lambda _{0}}(x)\) or \(v(x)\not \equiv v_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}} \).
We claim that there exists \(\varepsilon >0\) such that \(u(x)\le u_{\lambda }(x)\) and \(v(x)\le v_{\lambda }(x)\) on \(\Sigma _{\lambda } \) for any \(\lambda \in \left[ \lambda _{0},\lambda _{0}+\varepsilon \right) \).
Actually, for any \(\eta >0\), there exists a suitable \(R>0\) large enough such that
and
For such \(R>0\) and \(\lambda >\lambda _{0}\), we can verify that the measure of the set \(\Sigma _{\lambda }\cap B_{R}(0)\) goes to 0 as \(\lambda \rightarrow \lambda _{0}\).
By contradiction, we assume that
From (3.8), we obtain
Thus, by (4.11), we yield
Combining with (3.9), it follows that
together with (4.10), we have
Therefore, we can apply a similar argument as in Lemma 3.3 to conclude that \(u(x)\equiv u_{\lambda _{0}}(x)\) and \( v(x)\equiv v_{\lambda _{0}}(x) \) for any \( x\in \Sigma _{\lambda _{0}}.\) \(\square \)
4.4 Proof of Theorem 2.5
Similarly, we can move the plane from \(+\infty \) to left, and define
where \(\Sigma _{\rho }^{'}=\left\{ x\in {\mathbb {R}}^N|x_{1}>\rho \right\} \). If \(\lambda _{0}=\lambda _{1}\ne 0\), then both u and v are radially symmetric and decreasing about the plane \(x_{1}=\lambda _{0}\), which implies \(u(x)\equiv u_{\lambda _{0}}(x)\) and \(v(x)\equiv v_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}}\). Since \(|x-y|<|x^{\lambda _{0}}-y|\) and \(|y|>|y^{\lambda _{0}}|\), it is easy to deduce from (3.9) that
Therefore, we obtain that
which is impossible. Hence, we get \(\lambda _{0}=\lambda _{1}=0\). Notice that the direction of \(x_{1}\) is arbitrary, hence u, v are radially symmetric and decreasing about the origin. \({} \Box \)
4.4.1 Regularity
Taking similar derivations as Theorem 2.2, we need to define the functions
Suppose a, \(b \in L^{s}({\mathbb {R}}^{N})\), \(m\in L^{k}({\mathbb {R}}^{N})\), and \(h \in L^{t}({\mathbb {R}}^{N}) \), we also define the operator
Then (u, v, z, h) satisfies the operator equation
In order to prove the main result of Theorem 2.6, we shall establish the following two lemmas.
Lemma 4.7
Assume that \(p+q=2\cdot 2_{\alpha ,\mu }^{*}\) and s, k, t satisfy
Then for A sufficiently large, \({\mathscr {T}}_{A}\) is a contraction map from \(L^{s}({\mathbb {R}}^{N})\times L^{s}({\mathbb {R}}^{N})\times L^{k}({\mathbb {R}}^{N})\times L^{t}({\mathbb {R}}^{N})\) to itself.
Lemma 4.8
Suppose that \(3\le N\le 6\), \(\alpha \ge 0\), \(0<\mu <N\) and \(N-2\le 2\alpha +\mu \le N \). Let \((u,v,z,h) \in L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{k_{0}}({\mathbb {R}}^N)\times L^{t_{0}}({\mathbb {R}}^N)\) be a set of positive solutions of system (4.9) with \(k_{0}=\frac{2N}{4-(q-2)(N-2)}\) and \(t_{0}=\frac{2N}{4-(p-2)(N-2)}\), where p, q satisfying \(p+q=2\cdot 2_{\alpha ,\mu }^{*}\) and \(\frac{2(N-2\alpha -\mu )}{N-2}\le p-1,q-1\le min\left\{ \frac{4}{N-2}, \frac{N+2+2(N+2-2\alpha -\mu )}{N-2}\right\} \). Then \((u,v,z,h)\in L^{s}({\mathbb {R}}^{N})\times L^{s}({\mathbb {R}}^{N})\times L^{k}({\mathbb {R}}^{N})\times L^{t}({\mathbb {R}}^{N})\) with
and
4.4.2 Decay
The proof of this part is the same as the Theorem 2.3, here we omit for convenience.
5 Symmetry for the Hamiltonian system (2.3)
In this section, we shall prove the symmetry of (2.3) by discussing the following equivalent integral system in \({\mathbb {R}}^{N}\)
where \(p=2_{\alpha _{1},\mu _{1}}^{*}=\frac{2N-2\alpha _{1}-\mu _{1}}{N-2}\) and \(q=2_{\alpha _{2},\mu _{2}}^{*}=\frac{2N-2\alpha _{2}-\mu _{2}}{N-2}\).
For \(\lambda \in {{\mathbb {R}}}\), define
and
We easily get
and
Since \(|x^{\lambda }-y^{\lambda }|=|x-y|\) and \(|x^{\lambda }-y|=|x-y^{\lambda }|\), then it follows that
Similarly, we obtain
First, we consider the case where \(2\alpha _{1}+\mu _{1}\ne 4\) and \(2\alpha _{2}+\mu _{2}\ne 4\). We have the following property.
Lemma 5.1
Suppose that \(\alpha _{i}\ge 0\), \(0<\mu _{i}<N\), \(2\alpha _{i}+\mu _{i}\le 3\) if \(N=3\) while \(2\alpha _{i}+\mu _{i}<4\) if \(N\ge 4\) \((i=1,2)\). For any \(\lambda <0\), there exists a constant \(C>0\) such that
and
Proof
For any \(x\in \Sigma _{\lambda }\), notice \(|x^{\lambda }-y|\ge |x-y|\), using the mean value theorem, from (5.2) we know that
By applying the HLS inequality and the Hölder inequality, we obtain
Analogously, we also have
From (3.8) and (3.9), we can deduce that
and
Additionally, using the weighted HLS inequality again, we have
and
Inserting (5.8)–(5.11) to (5.6) and (5.7), we can obtain (5.4) and (5.5). The proof is completed. \(\square \)
For the case \(2\alpha _{i}+\mu _{i}=4\) \((i=1,2)\) or \(2\alpha _{1}+\mu _{1}=4\) and \(2\alpha _{2}+\mu _{2} \ne 4\), by the same derivation of the above, it is not difficult to find that the Lemma 5.1 would be replaced by the following lemmas.
Lemma 5.2
Suppose that \(N\ge 4\), \(\alpha _{i}\ge 0\), \(0<\mu _{i}<N\) and \(2\alpha _{i}+\mu _{i}=4(i=1,2)\). For any \(\lambda <0\), there exists a constant \(C>0\) such that
and
Lemma 5.3
Suppose that \(N\ge 4\), \(\alpha _{i}\ge 0\), \(0<\mu _{i}<N\), \(2\alpha _{1}+\mu _{1}=4\) and \(2\alpha _{2}+\mu _{2}<4\). For any \(\lambda <0\), there exists a constant \(C>0\) such that
and
The integral inequalities in Lemmas 5.1–5.3 can provide a beginning of the procedure of moving plane methods in integral forms. Thus, we are going to prove that for sufficiently small \(\lambda \), there holds \(u(x)\le u_{\lambda }(x)\) and \( v(x)\le v_{\lambda }(x) \) for any \(x\in \Sigma _{\lambda }\), which implies that we can start to move the plane from \(-\infty \) to the right.
Lemma 5.4
Suppose that \(N\ge 3\), \(\alpha _{i}\ge 0\), \(0<\mu _{i}<N\) and \(0<2\alpha _{i}+\mu _{i}\le min\left\{ 4, N\right\} \). Let \((u,v) \in L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\times L^{\frac{2N}{N-2}}({\mathbb {R}}^N)\) be a pair of positive solutions of system (5.1), then there exists \(M>0\) such that for any \(\lambda <-M\), we have
Proof
Since u, v are integrable, letting \(\lambda \rightarrow -\infty \), we have
and
where the constant C is the same as in Lemma 5.1. Hence, as \(\lambda \rightarrow -\infty \) in (5.4) and (5.5), we easily get
which shows that \(\Sigma _{\lambda }^{u}=\Sigma _{\lambda }^{v}=\emptyset .\) Therefore, there exists \(M>0\) such that for any \(\lambda <-M\), relation (5.12) holds. \(\square \)
Consequently, we now move the plane \(T_{\lambda }=\left\{ x\in {\mathbb {R}}^N|x_{1}=\lambda \right\} \) to the right as long as (5.12) is satisfied. We can certainly define
hence \(\lambda _{0}<+\infty \). This can be seen by applying a similar argument as in the above lemmas from \(\lambda \) near \(+\infty \).
Next, we deduce that u, v are symmetric about the critical plane \(x_{1}=\lambda _{0}\) in the \(x_{1}\) direction.
Lemma 5.5
Under the assumption of Lemma 5.4, for any \(\lambda _{0}<0\), we have
Proof
Suppose on the contrary that at \(\lambda _{0}<0\), there hold \(u(x)\le u_{\lambda _{0}}(x)\) and \(v(x)\le v_{\lambda _{0}}(x)\), but \(u(x)\not \equiv u_{\lambda _{0}}(x)\) or \(v(x)\not \equiv v_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}} \). It is sufficient to claim that there exists an \(\varepsilon >0\) such that \(u(x)\le u_{\lambda }(x)\) and \(v(x)\le v_{\lambda }(x)\) on \(\Sigma _{\lambda } \) for any \(\lambda \in \left[ \lambda _{0},\lambda _{0}+\varepsilon \right) \).
Indeed, for any \(\eta >0\), we can choose suitable \(R>0\) large enough such that
and
For such \(R>0\) and \(\lambda >\lambda _{0}\), we can also show that the measures of the sets \(\Sigma _{\lambda }^{u}\cap B_{R}(0)\) and \(\Sigma _{\lambda }^{v}\cap B_{R}(0)\) go to 0 as \(\lambda \rightarrow \lambda _{0}\).
By contradiction, we assume that
From (3.8) and (3.9), we obtain
Thus, by (5.2) and (5.3), we yield
Naturally, take the same derivations as in Lemma 3.3, we obtain that
and
where \({\mathcal {L}}\) is the Lebesgue measure.
Combining (5.16)–(5.19), there exists an \(\varepsilon >0\) such that for any \(\lambda \in [ \lambda _{0},\lambda _{0}+\varepsilon )\),
and
where the constant C is the same as in Lemma 5.1. By the same arguments as above, we can conclude that \(\Sigma _{\lambda }^{u}=\Sigma _{\lambda }^{v}=\emptyset \). Therefore, there exists an \(\varepsilon >0\) such that \(u(x)\le u_{\lambda }(x)\) and \(v(x)\le v_{\lambda }(x)\) on \(\Sigma _{\lambda } \) for any \(\lambda \in \left[ \lambda _{0},\lambda _{0}+\varepsilon \right) \). This contradicts the definition of \(\lambda _{0}\), then we have \(u(x)\equiv u_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}}\).
Similarly, if \(v(x)\not \equiv v_{\lambda _{0}}\) on \(\Sigma _{\lambda _{0}}\), we obtain a contradiction. The proof is completed. \(\square \)
5.1 Proof of Theorem 2.8
Similarly, we can also move the plane from \(+\infty \) to left, and define
where \(\Sigma _{\rho }^{'}=\left\{ x\in {\mathbb {R}}^N|x_{1}>\rho \right\} \).
If \(\lambda _{0}=\lambda _{1}\ne 0\), then both u and v are radially symmetric and decreasing about the plane \(x_{1}=\lambda _{0}\), which implies \(u(x)\equiv u_{\lambda _{0}}(x)\) and \(v(x)\equiv v_{\lambda _{0}}(x)\) on \(\Sigma _{\lambda _{0}}\). Thus, we also get
from which we can deduce that
This contradiction shows that \(\lambda _{0}=\lambda _{1}=0\). Since the direction of \(x_{1}\) is arbitrary, we conclude that u and v are radially symmetric and decreasing about the origin. \({} \Box \)
References
Beckner, W.: Pitt’s inequality with sharp convolution estimates. Proc. Am. Math. Soc. 136, 1871–1885 (2008)
Beckner, W.: Weighted inequalities and Stein–Weiss potentials. Forum Math. 20, 587–606 (2008)
Caffarelli, L., Gidas, B., Spruck, J.: Asymptotic symmetry and local behavior of semilinear elliptic equations with critical Sobolev growth. Commun. Pure Appl. Math. 42, 271–297 (1989)
Caffarelli, L., Kohn, R., Nirenberg, L.: First order interpolation inequalities with weights. Compos. Math. 53, 259–275 (1984)
Chen, L., Liu, Z., Lu, G.: Symmetry and regularity of solutions to the weighted Hardy–Sobolev type system. Adv. Nonlinear Stud. 16, 1–13 (2016)
Chen, W., Li, C.: Classification of solutions of some nonlinear elliptic equations. Duke Math. J. 63, 615–622 (1991)
Chen, W., Li, C., Ou, B.: Classification of solutions for a system of integral equations. Commun. Partial Differ. Equ. 30, 59–65 (2005)
Chen, W., Li, C., Ou, B.: Classification of solutions for an integral equation. Commun. Pure Appl. Math. 59, 330–343 (2006)
Chen, W., Li, C.: Methods on Nonlinear Elliptic Equations. AIMS Book Series on Differential Equations and Dynamical Systems, vol. 4 (2010)
Chen, W., Jin, C., Li, C., Lim, J.: Weighted Hardy–Littlewood–Sobolev inequalities and systems of integral equations. Discrete Contin. Dyn. Syst. S, 164–172 (2005)
Chen, W., Li, C.: The best constant in a weighted Hardy–Littlewood–Sobolev inequality. Proc. Am. Math. Soc. 136, 955–962 (2008)
Dou, J., Zhu, M.: Reversed Hardy–Littewood–Sobolev inequality. Int. Math. Res. Not. 19, 9696–9726 (2015)
Du, L., Gao, F., Yang, M. On elliptic equations with Stein–Weiss type convolution parts. Math. Z. (2022). https://doi.org/10.1007/s00209-022-02973-1
Du, L., Yang, M.: Uniqueness and nondegeneracy of solutions for a critical nonlocal equation. Discrete Contin. Dyn. Syst. 39, 5847–5866 (2019)
Gidas, B., Ni, W., Nirenberg, L.: Symmetry of positive solutions of nonlinear elliptic equations in \({\mathbb{R}}^{N}\), mathematical analysis and applications. Adv. Math. Suppl. Stud., 7a, Academic Press, New York. part A, pp. 369–402 (1981)
Gidas, B., Spruck, J.: Global and local behavior of positive solutions of nonlinear elliptic equations. Commun. Pure Appl. Math. 34, 525–598 (1981)
Jin, C., Li, C.: Quantitative analysis of some system of integral equations. Calc. Var. Partial Differ. Equ. 26, 447–457 (2006)
Lei, Y.: Qualitative analysis for the static Hartree-type equations. SIAM J. Math. Anal. 45, 388–406 (2013)
Lei, Y.: Liouville theorems and classification results for a nonlocal Schrödinger equation. Discrete Contin. Dyn. Syst. 38, 5351–5377 (2018)
Lei, Y., Li, C., Ma, C.: Asymptotic radial symmetry and growth estimates of positive solutions to weighted Hardy–Littlewood–Sobolev system of integral equations. Calc. Var. Partial Diffe. Equ. 16, 1–13 (2016)
Li, C.: Local asymptotic symmetry of singular solutions to nonlinear elliptic equations. Invent. Math. 123, 221–231 (1996)
Li, C., Ma, L.: Uniqueness of positive bound states to Schrödinger systems with critical exponents. SIAM J. Math. Anal. 40, 1049–1057 (2008)
Li, X., Yang, M., Zhou, X.: Qualitative properties and classification of solutions for elliptic equations with Stein–Weiss type convolution part. Sci. China Math. https://doi.org/10.1007/s11425-021-1918-1
Li, Y.: Remark on some conformally invariant integral equations: the method of moving spheres. J. Eur. Math. Soc. 6, 153–180 (2004)
Liu, S.: Regularity, symmetry, and uniqueness of some integral type quasilinear equations. Nonlinear Anal. 71, 1796–1806 (2009)
Liu, X., Lei, Y.: Existence of positive solutions for integral systems of the weighted Hardy–Littlewood–Sobolev type. Discrete Contin. Dyn. Syst. 40, 467–489 (2020)
Lieb, E., Loss, M.: Analysis, Graduate Studies in Mathematics. American Mathematical Society, Providence (2001)
Lieb, E.: Sharp constants in the Hardy–Littlewood–Sobolev and related inequalities. Ann. Math. 118, 349–374 (1983)
Miao, C., Xu, G., Zhao, L.: Global well-posedness, scattering and blow-up for the energy-critical, focusing Hartree equation in the radial case. Colloq. Math. 114, 213–236 (2009)
Stein, E.: Harmonic Analysis: Real-Variable Methods, Orthogonality, and Oscillatory Integrals, Princeton Mathematical Series, vol. 43. Princeton University Press, Princeton (1993)
Stein, E., Weiss, G.: Fractional integrals on \(n\)-dimensional Euclidean space. J. Math. Mech. 7, 503–514 (1958)
Strauss, W.: Existence of solitary waves in higher dimensions. Commun. Math. Phys. 55, 149–162 (1977)
Wei, J., Xu, X.: Classification of solutions of higher order conformally invariant equations. Math. Ann. 313, 207–228 (1999)
Yu, X.: Liouville type theorems for integral equations and integral systems. Calc. Var. Partial Differ. Equ. 46, 75–95 (2013)
Ding Y., Gao F., Yang M., Semiclassical states for Choquard type equations with critical growth: critical frequency case. Nonlinearity 33, 6695–6728 (2020)
Acknowledgements
Minbo Yang was partially supported by NSFC (11971436, 12011530199) and ZJNSF (LZ22A010001, LD19A010001). The research of Vicențiu D. Rădulescu was supported by a grant of the Romanian Ministry of Research, Innovation and Digitization, CNCS/CCCDI-UEFISCDI, project number PCE 137/2021, within PNCDI III.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by P. H. Rabinowitz.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Yang, M., Rădulescu, V.D. & Zhou, X. Critical Stein–Weiss elliptic systems: symmetry, regularity and asymptotic properties of solutions. Calc. Var. 61, 109 (2022). https://doi.org/10.1007/s00526-022-02221-8
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00526-022-02221-8