Abstract
In this paper, we are concerned with a critical Grushin-type problem. By applying Lyapunov–Schmidt reduction argument and attaching appropriate assumptions, we prove that this problem has infinitely many positive multi-bubbling solutions with arbitrarily large energy and cylindrical symmetry. Instead of estimating the corresponding derivatives of the reduced functional in locating the concentration points of the solutions, we employ the local Pohozaev identities to locate them.
Similar content being viewed by others
1 Introduction and main results
Consider the following semilinear elliptic equations involving Grushin operators:
where \(\alpha \ge 0\), \(\{m_1,m_2\}\subset \mathbb {N}^+\), K(x) is a function defined in \(\mathbb {R}^{m_1+m_2}\),
is the so-called Grushin operator, \(Q_\alpha :=m_1+(\alpha +1)m_2\) is the appropriate homogeneous dimension, and the power \(\frac{Q_\alpha +2}{Q_\alpha -2}\) is the corresponding critical exponent. For general case \(\alpha >0\), the entire positive solutions of (1.1) with \(K(x)=0\) were discussed by Monti and Morbidelli [21]. For \(\alpha =1\), the problem (1.1) becomes into
This case appeared very early in connection with the Cauchy–Riemann Yamabe problem solved by Jerison and Lee [14]. The Sharp Sobolev estimates for the Grushin operator \(G_1=-\Delta _y-4|y|^2\Delta _z\) were calculated by Beckner [1] in low dimension by using hyperbolic symmetry and conformal geometry.
However, as far as we know, there is very little literature on the existence of infinitely many solutions for (1.1) and even (1.2), which is one of our motivations to study this type of problem. In this paper, we shall study the critical Grushin-type problem (1.2) by using Lyapunov–Schmidt reduction argument and local Pohozaev identities. Under appropriate assumptions on K(x), we will prove that (1.2) possesses infinitely many multi-bubbling solutions with arbitrarily large energy and cylindrical symmetry. Our first technique is to transform (1.2) into a Hardy–Sobolev-type problem by a tricky change of variable.
If \(K(x)=K(|y|,z)\), and \(u=\psi (|y|,z)\) solves (1.2), then for \(|y|=\gamma \),
We define \(v(\gamma ,z)=\psi (\sqrt{\gamma },z),\) then
Hence v satisfies
If we denote \(V(x)=K(\sqrt{|y|},z)\), \(k=\frac{m_1+2}{2}\) when \(m_1\) is even, \(N=k+m_2\) and \(2^\star :=\frac{2(N-1)}{N-2}\), then \(u=v(|y|,z)\) solves
That is to say, the problem (1.3) is a special form of (1.2).
In the main part of this paper, we will construct infinitely many cylindrically symmetric multi-bubbling solutions for problem (1.3) by applying Lyapunov–Schmidt reduction argument and local Pohozaev identities. As derived and analyzed above, these solutions of (1.3) will produce infinitely many multi-bubbling solutions with cylindrical symmetry for the Grushin-type problem (1.2) when V(x) is cylindrically symmetric. This is also a motivation for us to study the problem (1.3).
Our main idea is motivated by the paper of Peng et al. [22], where the case of \(\alpha =0\) for problem (1.1) is concerned, which leads to study the following problem
The first existence result for (1.4) as far as our knowledge is due to Benci and Cerami [3], where they assumed that K was nonnegative and \(\Vert K\Vert _{L^{N/2}(\mathbb {R}^N)}\) was suitably small. Later in Chen et al. [8], it was proved that (1.4) had infinitely many nonradial solutions under the assumptions that nonnegative K was radially symmetric and \(r^2K(r)\) had a local maximum point or a local minimum point. Recently, Vétois and Wang [25] extended the result in [8] to the optimal dimension four.
In recent work by Peng et al. (see [22]), the authors relaxed the conditions on K and they assumed that \(K(x)=K(|x'|,x'')\) for \((x',x'')\in \mathbb {R}^2\times \mathbb {R}^{N-2}\) and \(r^2K(r,x'')\) has a stable critical point \((r_0,x_0'')\) with \(r_0>0,\ K(r_0,x_0'')>0\) and \(\deg \big (\nabla (r^2K(r,x'')),(r_0,x_0'')\big )\ne 0\), under which they constructed infinitely many positive bubbling solutions for (1.4) that all concentrate at \((r_0,x_0'')\). It is worthwhile to point out that in that paper for the first time, the authors used local Pohozaev identities in determining the location of the bubbles. Motivated by their idea, in the present paper, we will also construct infinitely many bubbling solutions which all concentrate at one point which can be a saddle point of some function related to the potential in our problem. We shall present the details at the end of this section.
For other results about the existence of multiple solutions to noncompact elliptic problems, please refer to [7, 12, 16, 18, 19, 24, 26,27,28,29,30,31,32].
Now we state our assumptions on V(x), which are as below.
(V): \(V(x)=V(|z'|,z'')\in C^1(\mathbb {R}^N)\) is nonnegative bounded for \(x=(y,z',z'') \in \mathbb {R}^k\times \mathbb {R}^2\times \mathbb {R}^{N-k-2}\). Setting \(r=|z'|\), the function \(r^2V(r,z'')\) has a stable critical point \((r_0,z_0'')\) satisfying \(r_0>0,\ V(r_0,z_0'')>0\), and \(\deg \big (\nabla (r^2V(r,z'')),B_\varrho (r_0,z_0'')\big )\ne 0\) for some small constant \(\varrho >0\).
The potential function V(x) is attached some cylindrical symmetry. In fact, the solutions we are going to construct are also of cylindrical symmetry. For similar problems, we refer readers to [2, 6]. In [2], Badiale and Tarantello analyzed the existence and non-existence of cylindrical solutions for a nonlinear elliptic equation in \(\mathbb {R}^3\). In [6], Cao, Peng and Yan constructed some multipeak solutions for a kind of Webster scalar curvature problems on the CR sphere with cylindrically symmetric curvature.
Our main results in this paper are as follows.
Theorem 1.1
Assume that \(N\ge 5,\ \frac{N+1}{2}\le k<N-1\) and (V) holds. Then the problem (1.3) has infinitely many cylindrically symmetric multi-bubbling solutions, whose energy can be made arbitrarily large.
Corollary 1.2
Under the assumptions of Theorem 1.1, if \(m_1=2k-2,\ m_2=N-k\) and \(K(x)=V(|z'|,z'')\), then the critical Grushin-type problem (1.2) has infinitely many multi-bubbling solutions with arbitrarily large energy and cylindrical symmetry.
Remark 1.3
\((\mathrm{i})\) The condition \(\frac{N+1}{2}\le k<N-1\) is equivalent to \(1<h:=N-k\le k-1\), which is needed for the proof of Lemma B.2. \((\mathrm{ii})\) For the case of \(N=4\), the estimate is totally different and the logarithm will appear (see [25]). Thus we only consider the case of \(N\ge 5\). \((\mathrm{iii})\) In order to estimate the local Pohozaev identity (3.1), we have to constrain the potential V independent of the first layer variables y (see Lemma 3.4).
Remark 1.4
In this paper, we only consider the case of \(\alpha =1\) for the Grushin operator \(G_\alpha \) since our methods to obtain the main results require the nondegeneracy of solution to the limit equation and that is true for \(\alpha =1\) (see [4]), while for \(\alpha >0\) and \(\alpha \ne 1\), the nondegeneracy is unknown and thus the corresponding infinitely many solutions problem for general Grushin operator \(G_\alpha \) cannot be obtained here and which is valuable to consider in the future.
There is no doubt that the reduction argument has become more and more useful in studying the existence and properties of solutions for noncompact elliptic problems. The studies [5, 6, 8,9,10,11,12,13, 15,16,17,18,19, 22,23,24, 26,27,31] and so on provide a powerful support. Particularly, Wei and Yan [30, 31] created a new idea, which is to use the number of the bumps or bubbles as the parameter in constructing solutions. That is also what we do in this paper. Next we shall sketch the proof of Theorem 1.1.
It is well known from [4, 20] that
are the only solutions to problem:
and \(U_{\xi ,\lambda }(x)\) is nondegenerate in \(D^{1,2}(\mathbb {R}^N)\), where
endowed with the norm \(\Vert u\Vert ^2=\int _{\mathbb {R}^N}|\nabla u|^2\text {d}x\), which is induced by the inner product \((u,v)=\int _{\mathbb {R}^N}\nabla u\nabla v\text {d}x\).
We shall use \(U_{\xi ,\lambda }(x)\) to build up the approximate solution for problem (1.3). In what follows, we always assume that \(m>0\) is a large integer, \(\lambda \in \big [T_0m^{\frac{N-2}{N-4}},T_1m^{\frac{N-2}{N-4}}\big ]\) for some constants \(T_1>T_0>0\), and \(|(\bar{r},\bar{z}'')-( r_0,z_0'')|\le \varrho \) with \(\bar{z}'':=(\bar{z}_3, \bar{z}_4,\ldots ,\bar{z}_{N-k})\in \mathbb {R}^{N-k-2}\). Define
and set
Let
where \(\xi _i':=\big (\bar{r}\cos \frac{2(i-1)\pi }{m},\bar{r}\sin \frac{2(i-1)\pi }{m}\big )\in \mathbb {R}^2,\ i=1,2,\ldots ,m\).
By the assumption (V), we can choose \(\delta >0\) as a small constant such that \(r^2V(r,z'')\ge C>0\) for \(|(r,z'')-(r_0,z_0'')|\le 10\delta \). To accelerate the decay of \(U_{\xi ,\lambda }\), we define a smooth cutoff function \(\eta (x)=\eta (|y|,|z'|,z'')\) satisfying \(\eta (x)=1\) if \(|(|y|,r,z'')-(0,r_0,z_0'')|\le \delta \) and \(\eta (x)=0\) if \(|(|y|,r,z'')-(0,r_0,z_0'')|\ge 2\delta \) with \(0\le \eta (x)\le 1\) in \(\mathbb {R}^N\). Denote
Theorem 1.1 is a direct consequence of the following result.
Theorem 1.5
Under the assumptions of Theorem 1.1, there exists an integer \(m_0>0\) such that for any integer \(m\ge m_0\), the problem (1.3) has a solution \(u_m\) of the form
where \(\varphi _m\in H_s\). Moreover, as \(m\rightarrow +\infty \), \(\lambda _m\in \big [T_0m^{\frac{N-2}{N-4}},T_1m^{\frac{N-2}{N-4}}\big ]\), \((\bar{r}_m,\bar{z}_m'')\rightarrow (r_0,z_0'')\) and \(\lambda _m^{-\frac{N-2}{2}}\Vert \varphi _m\Vert _{L^\infty (\mathbb {R}^N)}\rightarrow 0\).
We would like to remark that the concentration point \((r_0,z_0'')\) can be a saddle point of \(r^2V(r,z'')\). Thus, the procedure to determine the location of the bubbles in [8, 9, 12, 17, 27,28,29,30,31] cannot take effect any more. Because the solutions constructed there all concentrate at some local maximum points or local minimum points of some functions. Thereupon, a natural method is to estimate the derivatives of the reduced functional. However, the computations are very complicated at times (see [23] for example) and it doesn’t work at all in some cases (refer to [22]). So is our case. In fact, we only use one derivative of the reduced functional. The other two are useless for that the error terms destroy the dominant terms. Motivated and inspired by [22], we will employ some local Pohozaev identities to look for the algebraic equations determining the location of the bubbles. To be precise, we turn to prove that if \((\bar{r}, \bar{z}'')\) satisfies the local Pohozaev identities (3.1) and (3.2) (see Proposition 3.1) in a suitable neighborhood \(D_\rho \) of \((0,r_0, z_0'')\ (0\in \mathbb {R}^k)\), then \(\frac{\partial \mathcal F}{\partial \bar{r}}=0\) and \(\frac{\partial \mathcal F}{\partial \bar{z}_j}=0\ (j=3,4,\ldots ,N-k)\), where
is the reduced functional and
is the energy functional corresponding to problem (1.3). By using such identities (3.1) and (3.2), we only need to estimate the error term \(\varphi _{\bar{r},\bar{z}'',\lambda }\) away from the concentration points, which simplifies the calculations greatly. Moreover, to deal with the large number of bubbles in the solution, the reduction procedure will be performed in a weighted space instead of the standard Sobolev space. Since the equations we study in this paper include a singular term, the computations become much more delicate and complicated.
The paper in the sequel is organized as follows. In Sect. 2, we will carry out the finite-dimensional reduction procedure and obtain a good estimate for the error term. Precisely, we will use \(\bar{W}_{\bar{r},\bar{z}'',\lambda }\) as the approximation solution and consider the linearization of problem (1.3) around \(\bar{W}_{\bar{r},\bar{z}'',\lambda }\). The unique solvability of the linearized problem (2.1) will be assured by contradiction discussion and Fredholm’s alternative theorem. After that, we continue to study a perturbation problem for (1.3), namely (2.21). The Contraction Mapping Principle will be employed to prove that (2.21) is uniquely solvable, in which the estimates of \(\mathcal {R}(\varphi )\) and \(\mathcal P_m\) (see (2.23)) will play an important role. In Sect. 3, we will study the reduced finite-dimensional problem to obtain a true solution and complete the proof of Theorem 1.5. In other words, we will apply the local Pohozaev identities (3.1) and (3.2) together with one derivative (3.3) of the reduced functional to find the algebraic equations (3.21), (3.22), (3.23) and finally obtain suitable parameters \(\bar{r}, \bar{z}'',\lambda \), which correspond to the true solution. All the technical estimates will be left in “Appendices A, B and C”. For simplicity and without confusion, sometimes \(\int f(x)\) will denote \(\int _{\mathbb {R}^N}f(x)\text {d}x\). Throughout this paper, C signifies various positive constants independent of m and \(\lambda \).
2 Finite-dimensional reduction
In this section, we perform a finite-dimensional reduction. Let
and
where \(\tau :=\frac{N-4}{N-2}\). Denote
For some real numbers \(c_l\), consider
Lemma 2.1
Assume that \(\varphi _m\) solves problem (2.1) for \(f=f_m\). If \(\Vert f_m\Vert _{**}\rightarrow 0\) as \(m\rightarrow +\infty \), then \(\Vert \varphi _m\Vert _*\rightarrow 0\) as \(m\rightarrow +\infty \).
Proof
We argue by contradiction. Suppose that there exist \(m\rightarrow +\infty , \bar{r}_m\rightarrow r_0, \bar{z}_m''\rightarrow z_0''\), \(\lambda _m\in \big [T_0m^{\frac{N-2}{N-4}},T_1m^{\frac{N-2}{N-4}}\big ]\), and \(\varphi _m\) solves (2.1) for \(f=f_m, \lambda =\lambda _m, \bar{r}=\bar{r}_m, \bar{z}''=\bar{z}_m''\) with \(\Vert f_m\Vert _{**}\rightarrow 0\) and \(\Vert \varphi _m\Vert _*\ge C>0\). We may assume \(\Vert \varphi _m\Vert _*=1\) without loss of generality. For simplicity, we drop the subscript m.
By applying Green representation to \(|\varphi |\), we have
Due to Lemmas B.1, B.2 and B.3, we can prove
where \(n_l:=-1\) if \(l=1\) and \(n_l:=1\) if \(l=2,3,\ldots ,N-k\).
Next we estimate \(c_l\ (l=1,2,\ldots ,N-k)\). Multiplying (2.1) by \(\bar{U}_{1,l}\ (l=1,2,\ldots ,N-k)\) and integrating, we find that
Taking into account the facts that
and
we can deduce from Lemma B.1 that
and
Noting that
and by direct computation, we can prove
Combining (2.10), (2.11) and (2.13), we have
Additionally, direct calculation shows that for some \(\bar{c}>0\),
After (2.14) and (2.15) substituted in (2.6), we get
Now inserting (2.3), (2.4), (2.5), (2.16) into (2.2), we obtain
Since \(\Vert \varphi \Vert _*=1\), it follows from (2.17) that there exists \(R>0\) such that
The function \(\tilde{\varphi }(x):=\lambda ^{-\frac{N-2}{2}}\varphi \big (\lambda ^{-1}x+(0,\xi _1)\big )\) converges uniformly in any compact set to a solution u(x) of
Owing to that \(\varphi \in H_s\) solves (2.1), then u is even in \(z_2\) and u is perpendicular to the kernel of (2.19). Thus \(u=0\), which contradicts to (2.18). \(\square \)
By Lemma 2.1 and applying similar arguments to the proof of Proposition 4.1 in [10], we can prove the following result.
Lemma 2.2
There exist \(m_0>0\) and a constant \(C>0\) independent of m such that for any \(m\ge m_0\) and all \(f\in L^\infty (\mathbb {R}^N)\), problem (2.1) has a unique solution \(\varphi :=L_m(f)\). Moreover,
Now we consider
The main result in this section is
Proposition 2.3
There exists \(m_0>0\) such that for each \(m\ge m_0\), \(\lambda \in \big [T_0m^{\frac{N-2}{N-4}},T_1m^{\frac{N-2}{N-4}}\big ]\), \(\bar{r}\in [r_0-\varrho ,r_0+\varrho ]\) and \(\bar{z}''\in B_\varrho (z_0'')\), problem (2.21) has a unique solution \(\varphi =\varphi _{\bar{r},\bar{z}'',\lambda }\) satisfying
where \(\varepsilon >0\) is a small constant.
Rewrite (2.21) as
where
and
In order to apply the Contraction Mapping Principle to prove that (2.23) is uniquely solvable, we need to estimate \(\mathcal {R}(\varphi )\) and \(\mathcal P_m\), respectively.
Lemma 2.4
If \(N\ge 5\), then \(\Vert \mathcal {R}(\varphi )\Vert _{**}\le C\Vert \varphi \Vert _*^{2^{\star }-1}\).
Proof
For \(\theta \in (1,2]\),
Thus \(|\mathcal {R}(\varphi )|\le C\frac{|\varphi |^{2^{\star }-1}}{|y|}\), which together with Hölder inequality and (2.9) ensures that
where we use the symmetry and
Therefore, \(\Vert \mathcal {R}(\varphi )\Vert _{**}\le C\Vert \varphi \Vert _*^{2^{\star }-1}\). \(\square \)
Lemma 2.5
If \(N\ge 5\), then there exists a small constant \(\varepsilon >0\) such that \(\Vert \mathcal P_m\Vert _{**}\le C\left( \frac{1}{\lambda }\right) ^{1+\varepsilon }\).
Proof
Firstly, we estimate \(J_0\). By symmetry, we may assume \(x\in \Omega _1\). Then it holds
If \(|(|y|,r,z'')-(0,r_0,z_0'')|\ge 2\delta \), then \(J_0=0\).
If \(\delta<|(|y|,r,z'')-(0,r_0,z_0'')|<2\delta \), then
Thus we have
and
which imply that
If \(|(|y|,r,z'')-(0,r_0,z_0'')|\le \delta \), then
By (2.25), (2.12) and choosing \(\frac{N-2}{2}<\sigma \le \frac{N-2\tau }{2}\), we have
By Hölder inequality, (2.24) and (2.12), we get
Hence
Secondly, we estimate \(J_1\). Due to (2.7) and
we can obtain that
Thus
Thirdly, we estimate \(J_2\). By similar calculation to \(J_1\), we get
Fourthly, we estimate \(J_3\). Noting the fact that for \(x\in \text {supp}|\nabla \eta |\),
and (2.27), we can deduce that
Therefore,
Combining (2.26), (2.28), (2.29) and (2.31), we finish the proof. \(\square \)
Now we prove Proposition 2.3 by the Contraction Mapping Principle.
Proof of Proposition 2.3
Set
Problem (2.23) is equivalent to
where \(L_m\) is defined in Lemma 2.2. We will prove that \(\mathcal {A}\) is a contraction mapping from \(\mathcal {N}\) to \(\mathcal {N}\).
On the one hand, by (2.20), Lemmas 2.4 and 2.5, we have
Hence \(\mathcal {A}\) maps \(\mathcal {N}\) to \(\mathcal {N}\).
On the other hand,
Since \(|\mathcal {R}'(\varphi )|\le C\frac{|\varphi |^{2^\star -2}}{|y|}\) and
we get that there is \(0<\vartheta <1\) such that
which implies that
Thus \(\mathcal {A}\) is a contraction mapping on \(\mathcal {N}\).
In virtue of the Contraction Mapping Principle, there exists an unique \(\varphi =\varphi _{\bar{r},\bar{z}'',\lambda }\in \mathcal {N}\) satisfying problem (2.23). Moreover, by (2.20), Lemmas 2.4 and 2.5 again, we achieve (2.22). \(\square \)
3 Proof of the main result
In this section, we prove the existence of infinitely many positive multi-bubbling solutions for problem (1.3). Namely, we choose suitable \((\bar{r}, \bar{z}'',\lambda )\) satisfying \(c_l=0\ ( l=1,2,\ldots ,N-k)\), and thus \(\bar{W}_{\bar{r},\bar{z}'',\lambda }+\varphi _{\bar{r},\bar{z}'',\lambda }\) is a solution of problem (1.3).
For this purpose, the following result is needed. We point out in advance that the left-hand side of (3.1) is the local Pohozaev identity generating from scaling, while the left-hand side of (3.2) is the local Pohozaev identities generating from translations.
Proposition 3.1
Suppose that \((\bar{r},\bar{z}'',\lambda )\) satisfies
where \(D_\rho :=\big \{(y,z',z'')\in \mathbb {R}^N:|(|y|,|z'|,z'')-(0,r_0,z_0'')|\le \rho \big \}\) with \(\rho \in (2\delta ,5\delta )\) and \(u_m:=\bar{W}_{\bar{r},\bar{z}'',\lambda }+\varphi _{\bar{r},\bar{z}'',\lambda }\) gotten from Proposition 2.3. Then
Proof
Since \(\text {supp}\eta \subset D_\rho \), we find \(\bar{U}_{\xi _i,\lambda }=0\) in \(\mathbb {R}^N{\setminus } D_\rho \) for \(i=1,2,\ldots ,m\). It follows from (3.1), (3.2), (3.3) that
for \(v=\left\langle x,\nabla u_m\right\rangle , \ v=\frac{\partial u_m}{\partial z_j} \ (j=3,4,\ldots ,N-k)\) and \(v=\frac{\partial \bar{W}_{\bar{r},\bar{z}'',\lambda }}{\partial \lambda }\), respectively.
By direct computations, we can prove
for some constants \(a_1>0,\ a_2<0,\ a_3>0\).
By means of integration by parts and (2.22), we get
for \(v=\left\langle x,\nabla \varphi _{\bar{r},\bar{z}'',\lambda }\right\rangle \) and \(v=\frac{\partial \varphi _{\bar{r},\bar{z}'',\lambda }}{\partial z_j} \ (j=3,4,\ldots ,N-k)\), respectively, which together with (3.4) imply that
for \(v=\left\langle x,\nabla \bar{W}_{\bar{r},\bar{z}'',\lambda }\right\rangle \) and \(v=\frac{\partial \bar{W}_{\bar{r},\bar{z}'',\lambda }}{\partial z_j} \ (j=3,4,\ldots ,N-k)\), respectively.
Noting
we can obtain
In addition, for \(j=3,4,\ldots ,N-k\),
By (3.8), (3.9) and (3.5), we have
By (3.8), (3.10) and (3.6), we get for \(j=3,4,\ldots ,N-k\),
It follows from (3.11) and (3.12) that
In terms of (3.4), (3.7), (3.13) and
we obtain that
which implies \(c_1=0\). \(\square \)
Next we will prove there really exists \((\bar{r},\bar{z}'',\lambda )\) satisfying (3.1), (3.2) and (3.3). To achieve it, we have to find the equivalent forms of (3.1), (3.2) and (3.3), respectively.
Now we estimate (3.3).
Lemma 3.2
Equation (3.3) is equivalent to
where \(B_1>0,\ B_3>0\) are defined in Lemma A.1.
Proof
Since \(u_m=\bar{W}_{\bar{r},\bar{z}'',\lambda }+\varphi _{\bar{r},\bar{z}'',\lambda }\), we have
Taking into account (2.32), (2.9), (2.22) and for \(\theta \in (1,2]\),
we can obtain
By (2.11), (2.13) and (2.22), we get
Therefore,
By Lemma A.1 in “Appendix A”, the proof is completed. \(\square \)
Now we estimate (3.1) and (3.2).
Lemma 3.3
Equation (3.2) is equivalent to
Lemma 3.4
Equation (3.1) is equivalent to
Proof
Firstly, we have
Secondly, by (2.21), we get
Inserting (3.15) into (3.14), we obtain
It follows from (3.16), (2.22) and
that for some small \(\varepsilon >0\),
Noting
and by Lemma 3.3 and (3.17), we have
\(\square \)
To simplify the results in Lemmas 3.3 and 3.4 and obtain the final equivalent forms of (3.1) and (3.2), we need the following several results.
Lemma 3.5
There holds
Proof
By (2.21), we have
Similarly to the arguments of \(I_1\) in Lemma 3.2 and applying (2.22), we have
As similar as the arguments of (2.11) and by (2.22), we get
By using the estimates of \(J_0,J_2,J_3\) in Lemma 2.5, (2.9) and (2.22), we obtain
\(\square \)
Lemma 3.6
There holds
Proof
By Hardy–Sobolev inequality and Lemma 3.5, we have
Hence
\(\square \)
In view of Lemma 3.6, the following result is true.
Lemma 3.7
There exists \(\rho \in (3\delta ,4\delta )\) such that
Lemma 3.8
For any \(g(r,z'')\in C(\mathbb {R}^{N-k-1},\mathbb {R})\), there holds
Proof
Although the proof is similar to that of Lemma 3.4 in [22], we give the details for completeness. Since \(u_m=\bar{W}_{\bar{r},\bar{z}'',\lambda }+\varphi \), we have
It is similar to the estimate of \(\tilde{J}_2\) in Lemma 3.5 that
By Lemma 3.5, we have
Inserting (3.19) and (3.20) into (3.18), we obtain
where
and
\(\square \)
Choosing \(g(r,z'')=\frac{\partial V(r,z'')}{\partial z_j}\) and \(g(r,z'')=\frac{1}{2r}\frac{\partial (r^2 V(r,z''))}{\partial r}\) in Lemma 3.8, respectively, we have the following result.
Remark 3.9
For \(j=3,4,\ldots ,N-k\),
Now we are ready to prove Theorem 1.5.
Proof of Theorem 1.5
Combining Lemmas 3.2, 3.3, 3.4, 3.7 and Remark 3.9, we find that there exists \(\rho \in (3\delta ,4\delta )\) such that (3.1), (3.2) and (3.3) are, respectively, equivalent to
Set \(\lambda =tm^{\frac{1}{\tau }}\), then we get from (3.23) that
Let \(F(t,\bar{r},\bar{z}'')=\big (\nabla _{\bar{r},\bar{z}''}(\bar{r}^2V(\bar{r},\bar{z}'')),B_1V(\bar{r},\bar{z}'')-\frac{B_3}{t^{N-4}}\big )\), then
Hence there exist \(t_m\in [T_0,T_1]\) and \((\bar{r}_m,\bar{z}_m'')\in B_\varrho (r_0,z_0'')\) satisfying (3.21), (3.22) and (3.24). \(\square \)
References
Beckner, W.: On the Grushin operator and hyperbolic symmetry. Proc. Am. Math. Soc. 129, 1233–1246 (2001)
Badiale, M., Tarantello, G.: A Sobolev–Hardy inequality with applications to a nonlinear elliptic equation arising in astrophysics. Arch. Ration. Mech. Anal. 163, 259–293 (2002)
Benci, V., Cerami, G.: Existence of positive solutions of the equation\(-\Delta u+a(x)u=u^{\frac{N+2}{N-2}}\) in \(\mathbb{R}^N\). J. Funct. Anal. 88, 90–117 (1990)
Castorina, D., Fabbri, I., Mancini, G., Sandeep, K.: Hardy–Sobolev extremals, hyperbolic symmetry and scalar curvature equations. J. Differ. Equ. 246, 1187–1206 (2009)
Cao, D., Heinz, H.P.: Uniqueness of positive multi-lump bound states of nonlinear Schrödinger equations. Math. Z. 243, 599–642 (2003)
Cao, D., Peng, S., Yan, S.: On the Webster scalar curvature problem on the CR sphere with a cylindrical-type symmetry. J. Geom. Anal. 23, 1674–1702 (2013)
Cerami, G., Devillanova, G., Solimini, S.: Infinitely many bound states for some nonlinear scalar field equations. Calc. Var. Partial. Differ. Equ. 23, 139–168 (2005)
Chen, W., Wei, J., Yan, S.: Infinitely many solutions for the Schrödinger equations in \(\mathbb{R}^N\) with critical growth. J. Differ. Equ. 252, 2425–2447 (2012)
Deng, Y., Lin, C.S., Yan, S.: On the prescribed scalar curvature problem in \(\mathbb{R}^N\), local uniqueness and periodicity. J. Math. Pures Appl. 104, 1013–1044 (2015)
del Pino, M., Felmer, P., Musso, M.: Two-bubble solutions in the super-critical Bahri–Coron’s problem. Calc. Var. Partial. Differ. Equ. 16, 113–145 (2003)
Gheraibia, B., Wang, C., Yang, J.: Existence and local uniqueness of bubbling solutions for the Grushin critical problem. Differ. Integral Equ. 32, 49–90 (2019)
Guo, Y., Li, B.: Infinitely many solutions for the prescribed curvature problem of polyharmonic operator. Calc. Var. Partial. Differ. Equ. 46, 809–836 (2013)
Guo, Y., Nie, J., Niu, M., Tang, Z.: Local uniqueness and periodicity for the prescribed scalar curvature problem of fractional operator in \(\mathbb{R}^N\). Calc. Var. Partial. Differ. Equ. 56, 118 (2017)
Jerison, D., Lee, J.M.: Extremals for the Sobolev inequality on the Heisenberg group and the CR Yamabe problem. J. Am. Math. Soc. 1, 1–13 (1988)
Li, Y.Y.: On a singularly perturbed elliptic equation. Adv. Differ. Equ. 2, 955–980 (1997)
Li, Y.Y.: On \(-\Delta u=K(x)u^5\) in \(\mathbb{R}^3\). Commun. Pure Appl. Math. 46, 303–340 (1993)
Li, Y.Y., Wei, J., Xu, H.: Multi-bump solutions of \(-\Delta u=K(x)u^{\frac{n+2}{n-2}}\) on lattices in \(\mathbb{R}^n\). J. Reine Angew. Math. 743, 163–211 (2018)
Lin, F.H., Ni, W.M., Wei, J.: On the number of interior peak solutions for a singularly perturbed Neumann problem. Commun. Pure Appl. Math. 60, 252–281 (2007)
Lucia, M., Tang, Z.: Multi-bump bound states for a Schrödinger system via Lyapunov–Schmidt reduction. Nonlinear Differ. Equ. Appl. 24, 65 (2017)
Mancini, G., Fabbri, I., Sandeep, K.: Classification of solutions of a critical Hardy–Sobolev operator. J. Differ. Equ. 224, 258–276 (2006)
Monti, R., Morbidelli, D.: Kelvin transform for Grushin operators and critical semilinear equations. Duke Math. J. 131, 167–202 (2006)
Peng, S., Wang, C., Yan, S.: Construction of solutions via local Pohozaev identities. J. Funct. Anal. 274, 2606–2633 (2018)
Rey, O.: Boundary effect for an elliptic Neumann problem with critical nonlinearity. Commun. Partial Differ. Equ. 22, 1055–1139 (1997)
Rey, O., Wei, J.: Arbitrary number of positive solutions for an elliptic problem with critical nonlinearity. J. Eur. Math. Soc. 7, 449–476 (2005)
Vétois, J., Wang, S.: Infinitely many solutions for cubic nonlinear Schrödinger equations in dimension four. Adv. Nonlinear Anal. 8, 715–724 (2019)
Wang, Z.Q.: Construction of multi-peaked solutions for a nonlinear Neumann problem with critical exponent in symmetric domains. Nonlinear Anal. 27, 1281–1306 (1996)
Wang, C., Wang, Q., Yang, J.: On the Grushin critical problem with a cylindrical symmetry. Adv. Differ. Equ. 20, 77–116 (2015)
Wang, L., Wei, J., Yan, S.: A Neumann problem with critical exponent in nonconvex domains and Lin-Ni’s conjecture. Trans. Am. Math. Soc. 362, 4581–4615 (2010)
Wang, L., Wei, J., Yan, S.: On Lin-Ni’s conjecture in convex domains. Proc. Lond. Math. Soc. 102, 1099–1126 (2011)
Wei, J., Yan, S.: Infinitely many positive solutions for the nonlinear Schrödinger equations in \(\mathbb{R}^N\). Calc. Var. Partial. Differ. Equ. 37, 423–439 (2010)
Wei, J., Yan, S.: Infinitely many solutions for the prescribed scalar curvature prolem on \(S^N\). J. Funct. Anal. 258, 3048–3081 (2010)
Yan, S., Yang, J.: Infinitely many solutions for an elliptic problem involving critical Sobolev and Hardy–Sobolev exponents. Calc. Var. Partial. Differ. Equ. 48, 587–610 (2013)
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Zhongwei Tang: Supported by NSFC (11571040, 11671331). Chunhua Wang: Supported by NSFC (11671162) and CCNU18CXTD04.
Appendices
Appendix A: Energy expansion
In this section, we give the estimate of energy expansion for the approximate solution.
Lemma A.1
If \(N\ge 5\), then
where \(B_1,B_2,B_3\) are some positive constants.
Proof
It can be checked that
Moreover, we have
On the one hand, we get
for some constant \(B_0>0\) and \(B_1:=B_0\int U_{0,1}^2\). On the other hand, we have
for some constant \(B_2>0\). Hence we obtain
\(\square \)
Appendix B: Basic estimates
In this section, we give three basic estimates, especially Lemma B.1 will run through the whole paper.
Lemma B.1
(Lemma B.1, [27]) Let \(\alpha \ge 1,\beta \ge 1\) and \(i\ne j\). Then for any \(0<\sigma \le \min \{\alpha ,\beta \}\), there is a constant \(C>0\) such that
Lemma B.2
(Lemma B.2, [27]) For any constant \(0<\sigma <N-2\), there is a constant \(C>0\) such that
Lemma B.3
Let \(N\ge 5\). Then there is a constant \(C>0\) and a small constant \(\iota >0\) such that
Proof
The proof is similar to that of Lemma B.3 in [27], but we offer the details for completeness. Assume \(\tilde{x}=(\tilde{y},\tilde{z})\in \Omega _1\), it holds \(|\tilde{z}-\xi _i|\ge |\tilde{z}-\xi _1|\). By Lemma B.1 and (2.9), we get
And then
By Lemma B.2, we obtain
Therefore,
since \(1-\tau -\frac{2\tau }{N-2}>0\). \(\square \)
Appendix C: Proof of inequality (2.13)
Proof of (2.13)
Firstly, it is similar to the proof of (2.11) that
Secondly, by (2.30) and as similar as (2.11), we get
Thirdly,
By (2.30), (2.8) and (2.9), we have
Taking into account the facts that for \(\theta \in (0,1)\),
and for \(\zeta =N-2\) or \(\zeta =\frac{N-2}{2}+\tau \),
we can deduce that
By (2.12), we have
Owing to (2.24), (2.12) and (2.9), we get
In terms of (2.32) and (2.12), we obtain
In virtue of (2.12), we achieve
\(\square \)
Rights and permissions
About this article
Cite this article
Liu, M., Tang, Z. & Wang, C. Infinitely many solutions for a critical Grushin-type problem via local Pohozaev identities. Annali di Matematica 199, 1737–1762 (2020). https://doi.org/10.1007/s10231-019-00940-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10231-019-00940-y