Home Limit profiles and the existence of bound-states in exterior domains for fractional Choquard equations with critical exponent
Article Open Access

Limit profiles and the existence of bound-states in exterior domains for fractional Choquard equations with critical exponent

  • Fumei Ye , Shubin Yu and Chun-Lei Tang EMAIL logo
Published/Copyright: June 8, 2024

Abstract

This article is devoted to studying the existence of positive solutions to the following fractional Choquard equation:

( Δ ) s u + u = Ω u ( y ) p x y N α d y u p 2 u + ε Ω u ( y ) 2 α , s * x y N α d y u 2 α , s * 2 u , in Ω , u = 0 , on R N \ Ω ,

where Ω is an exterior domain with smooth boundary Ω such that R N \ Ω is bounded, N > 2 s , 2 < p < 2 α , s * , 2 α , s * N + α N 2 s is the fractional Hardy-Littlewood-Sobolev critical exponent, and ε > 0 is a parameter. We establish the limit profiles and uniqueness of positive radial ground-states for the limit equation without the critical exponent as α sufficiently close to N . Then, combining variational method, barycentric functions, and Brouwer degree theory, we determine the existence of positive bound-state solutions provided that ε > 0 is sufficiently small.

MSC 2010: 35J65; 35Q55; 35B33

1 Introduction

In this article, we focus our attention on the following nonlinear fractional Choquard equation:

(1.1) i t Φ + ( Δ ) s Φ = ( I α Φ p ) Φ p 2 Φ ,

where N > 2 s , s ( 0 , 1 ) , α ( 0 , N ) , p ( 2 , 2 α , s * ) , and 2 α , s * N + α N 2 s is the fractional Hardy-Littlewood-Sobolev critical exponent, and I α is the Riesz potential given by

I α ( x ) = C N , α x N α , with C N , α = Γ N α 2 Γ α 2 π N 2 2 α .

In particular, the fractional Laplace operator ( Δ ) s appears as an infinitesimal generator for Lévy stable diffusion processes [3], which is defined by

( Δ ) s u ( x ) C ( N , s ) P.V. R N u ( x ) u ( y ) x y N + 2 s d y , u S ( R N ) ,

where C ( N , s ) is a suitable normalization constant, P.V. stands for the principle value of the integral, and S ( R N ) denotes the Schwartz space of rapidly decreasing smooth functions (see [24] for more details).

An important issue concerning nonlinear evolution equations such as (1.1) is to study their standing wave solutions. A stationary wave solution is a solution of the form Φ ( t , x ) = e i λ t u ( x ) , where λ R and u : R N R is a time-independent that must solve the following fractional Choquard equation:

(1.2) ( Δ ) s u + λ u = ( I α u p ) u p 2 u , in R N .

In the case of s = 1 2 , N = 3 , α = 2 , and p = 2 , the aforementioned equation is reduced to the following form:

Δ u + λ u = ( I 2 u 2 ) u , in R 3 ,

which has been used to model the dynamics of pseudo-relativistic boson stars in [10,12]. Moreover, the regularity, existence, nonexistence, symmetry, and decay properties of weak solutions for Problem (1.2) have been extensively studied. We refer the reader to [9,14,15,31] and references therein. On the other hand, it is worth noting that, if s 1 , Problem (1.2) reduces to a well-known Choquard equation, which can be applied to describe an electron trapped in its own hole and also used to describe the quantum mechanics of stationary polarons (see [19] and [26]), respectively. In this regime, the existence of solutions and the dynamic behaviors have been studied in [2123,27,28] and references therein.

Compared to the previous results on R N , we mainly focus on the existence of positive solutions for Problem (1.2) in exterior domains, i.e.,

(1.3) ( Δ ) s u + u = ( I α u p ) u p 2 u , in Ω , u = 0 , on R N \ Ω ,

where Ω is an exterior domain with smooth boundary Ω such that R N \ Ω is bounded. Noting that the research in exterior domain problems originated from [4], the authors considered the following Schrödinger equation:

(1.4) Δ u + λ u = u p 2 u , in Ω

and proved that (1.4) has no ground-state solutions. Based on the aforementioned facts, they analyzed the behavior of Palais-Smale sequences and showed a precise estimate of the higher levels of energy. Then, the existence of positive solutions for (1.4) can be obtained. Then, a great attention has been paid to Schrödinger equations or fractional Schrödinger equations in exterior domains. We refer the reader to [1,2,6,7,33] and the related results mentioned there. It is worth noting that there are few works concerning the existence of positive solutions for Choquard equations in exterior domains. Indeed, the corresponding results are presented in [5,8,17,34]. For example, the authors in [8] considered the following nonlinear Choquard equation in exterior domains:

(1.5) Δ u + u = ( I α u p + 1 ) u p + ε ( I α u 2 α * ) u 2 α * 1 , in Ω , u > 0 , in Ω , u = 0 , on Ω ,

where 2 α * N + α N 2 is the Hardy-Littlewood-Sobolev critical exponent and p = 1 if N = 3 , 4 , 5 or 2 < p + 1 < 7 3 if N = 3 . Combining the variational method, Brouwer degree theory, and deformation lemma, they proved the existence of positive solutions for Problem (1.5). In particular, the restriction of exponent p is strictly dependent on the uniqueness of radial solutions to the limit equation of Problem (1.5), which is the key point in studying the existence of positive solutions of elliptic problems in exterior domains. However, the uniqueness is missing for Choquard equations in the fractional setting. To this end, inspired by Seok [29], the authors in [17] proved the uniqueness of the radial solution as α 0 for the limit equation:

(1.6) ( Δ ) s u + u = ( I α u p ) u p 2 u , in R N ,

where p ( 2 , 2 s * ) and 2 s * 2 N N 2 s is the fractional Sobolev critical exponent. Then, the existence of positive solutions for Problem (1.3) can be obtained when α > 0 small enough.

Mainly motivated by [17,29], we first intend to establish the limit profiles and uniqueness of the radial ground-states for equation (1.6) as α sufficiently close to N . When α 0 , the limit profiles of Problem (1.6) is strictly depend on some estimates of C N , α (see [17, Proposition 2.12]). However, as far as α N is considered, it is necessary to utilize a suitable scaling and transform (1.6) into the following form:

(1.7) ( Δ ) s u + u = R N u ( y ) p x y N α u p 2 u in R N .

Given this reality, in this article, we consider the relevant issues of Problem (1.7) instead of Problem (1.6). By establishing the convergence properties of Choquard term in (1.7), the following result can be obtained.

Theorem 1.1

Let 2 < p < 2 α , s * and α sufficiently close to N, then Problem (1.7) has a unique positive radial ground-state solution.

According to the aforementioned result, the following Choquard equation in exterior domains can be considered

(1.8) ( Δ ) s u + u = Ω u ( y ) p x y N α d y u p 2 u , in Ω , u = 0 , on Ω ,

where N > 2 s , 2 < p < 2 α , s * , and Ω is also an exterior domain. Then, by repeating the same arguments [17, Theorem 1.1], the existence of positive solutions for Problem (1.8) as α sufficiently close to N can be directly obtained when the complement of Ω in R N is small in the sense that R N \ Ω B ( 0 , η ) for some small constant η > 0 . Here, B ( 0 , η ) is the Euclidean ball of radius η > 0 .

Corollary 1.2

There exist α N > 0 close to N and η > 0 sufficiently small such that if R N \ Ω B ( 0 , η ) , then Problem (1.8) has a positive solution for every α ( α N , N ) .

In what follows, we are particularly interested in the existence of positive solutions for Problem (1.8) with the fractional Hardy-Littlewood-Sobolev critical exponent. In this perspective, it is natural to consider the following problem:

(1.9) ( Δ ) s u + u = Ω u ( y ) p x y N α d y u p 2 u + ε Ω u ( y ) 2 α , s * x y N α d y u 2 α , s * 2 u , in Ω , u = 0 , on Ω ,

where N > 2 s , 2 < p < 2 α , s * . To this end, it is necessary to consider the information regarding the solutions to the limit problem:

(1.10) ( Δ ) s u + u = R N u ( y ) p x y N α d y u p 2 u + ε R N u ( y ) 2 α , s * x y N α d y u 2 α , s * 2 u , in R N .

To our knowledge, the uniqueness of solution for limit Problem (1.10) is not ensured. Meanwhile, due to the appearance of the fractional Hardy-Littlewood-Sobolev critical exponent, it gives rise to another difficulty: the lack of compactness of Sobolev embedding. Inspired by [2,8], the difficulties can be overcome by restricting the parameter ε > 0 sufficiently small. Then, we shall prove that Problem (1.10) has a ground-state solution, whereas Problem (1.9) does not have a ground-state solution, which is the essential feature of elliptic problems in exterior domains.

Theorem 1.3

Problem (1.9) has no ground-state solution.

Once Theorem 1.3 is established, in order to obtain the existence of positive solutions for Problem (1.9), it is necessary to observe higher levels of energy and study the behaviors of the Palais-Smale sequences as ε 0 . Then, we verify the compactness of Palais-Smale sequences in an interval of energy levels higher than the ground-state level (Proposition 3.7). Based on the aforementioned descriptions, we can investigate the existence of positive bound-states to Problem (1.8) with a small critical perturbation, i.e., Problem (1.9).

Theorem 1.4

There exist α N > 0 close to N , η > 0 and ε > 0 such that if R N \ Ω B ( 0 , η ) and ε ( 0 , ε ) ; then, Problem (1.9) has a positive solution for every α ( α N , N ) .

An outline of this article is as follows. The limit profiles and uniqueness for Problem (1.7) as α N are developed in Section 2. In Section 3, we prove that Problem (1.9) does not have a ground-state solution, and we provide some technical lemmas, which are crucial to prove our main results. Finally, in Section 4, we will give the proof of Theorem 1.4 by combining the Brouwer degree theory, barycentric functions, and deformation lemma.

Throughout this article, we make use of the following notations:

  • The fractional Sobolev space H s ( R N ) is defined for any s ( 0 , 1 ) by

    H s ( R N ) = u L 2 ( R N ) : ( Δ ) s 2 u L 2 ( R N ) ,

    which is a Hilbert space endowed with the norm:

    u H s ( R N ) 2 = R N ( Δ ) s 2 u 2 d x + R N u 2 d x , u H s ( R N ) ,

    where

    R N ( Δ ) s 2 u 2 d x = R N R N u ( x ) u ( y ) 2 x y N + 2 s d y d x .

  • H 0 s ( Ω ) is denoted by one sub-space of H s ( R N ) ,

    H 0 s ( Ω ) = { u H s ( R N ) : u = 0 , a.e. in R N \ Ω } ,

    and endowed with the norm

    u H 0 s ( Ω ) 2 = Q u ( x ) u ( y ) 2 x y N + 2 s d y d x + Ω u 2 d x ,

    where Q R 2 N \ ( Ω c × Ω c ) . Obviously,

    u H s ( R N ) = u H 0 s ( Ω ) , u H 0 s ( Ω ) .

  • For p 1 , the (standard) L p -norm of u L p ( R N ) is denoted by u L p ( R N ) . Moreover, H r s ( R N ) denotes the radial subspace of H s ( R N ) .

  • For any x R N and r > 0 , B ( x , r ) { y R N : x y < r } .

  • C , C 1 , C 2 , denote the positive constants possibly different in different places.

2 Limit profiles and uniqueness of limit equation

In this section, for N > 2 s , s ( 0 , 1 ) , α ( 0 , N ) , and p ( 2 , 2 α , s * ) , we are intended to establish the uniqueness of positive radial ground-state solutions of (1.7) as α N , i.e., Theorem 1.1 is proved. We recall the limit equation (1.6), i.e.,

(2.1) ( Δ ) s v + v = C N , α R N v ( y ) p x y N α v p 2 v , in R N , v H s ( R N ) .

In particular, the existence of a ground-state solution to (2.1) when p ( 2 , 2 α , s * ) has been studied in [9]. Making the change of variables u = s ( N , α , p ) v , we are led to study the following equation:

( Δ ) s u + u = R N u ( y ) p x y N α u p 2 u , in R N ,

i.e., Problem (1.7). Here,

s ( N , α , p ) ( C N , α ) 1 2 p 2 1 N α 1 2 p 2 , as α N .

Obviously, 2 α , s * 2 s * as α N .

In what follows, we insert the fractional concentration compactness principle established in [11, Lemma 2.2].

Lemma 2.1

Let N > 2 s and q [ 2 , 2 s * ) . Assume that { u n } is bounded in H s ( R N ) and it satisfies

lim n sup y R N B ( y , R ) u n ( x ) q d x = 0 ,

where R > 0 . Then, u n 0 in L p ( R N ) for 2 < p < 2 s * .

Next, we reveal the Hardy-Littlewood-Sobolev inequality and the estimate of the Riesz potential as in [20, Section 4.3].

Lemma 2.2

Let t , r > 1 and α ( 0 , N ) be such that

1 t + 1 r = 1 + α N .

Then, for any u L t ( R N ) and v L r ( R N ) , there holds

R N R N u ( x ) v ( y ) x y N α d y d x C ( N , α , t ) u L t ( R N ) v L r ( R N ) .

Lemma 2.3

Let 1 r < s < and α ( 0 , N ) satisfy

1 r 1 s = α N .

Then, for u L r ( R N ) , there holds

R N u ( y ) y N α d y L s ( R N ) C ( N , α , r ) u L r ( R N ) ,

where C ( N , α , r ) is a positive constant depending only on N , α , and r .

The following lemma gives an upper bound for the Riesz potential energy (see [28, Lemma 3.3]).

Lemma 2.4

Let r ( 1 , + ) . For every α N r , N , if f L 1 ( R N ) L r ( R N ) vis nonnegative and x R N ,

1 N α f ( x ) R N f d y + C N α ( r α N ) 1 1 r R N f r d y 1 r .

In particular, if the function g L 1 ( R N ) is nonnegative, then

R N 1 N α f g d y R N f d y R N g d y + C N α ( r α N ) 1 1 r R N f r d y 1 r R N g d y .

As a consequence of the aforementioned lemma, similar to [28, Lemma 3.4], we can prove the convergence results as follows.

Lemma 2.5

Let r ( 1 , + ) , { α n } be a sequence in N r , N converging to N , and let { f n } be a bounded sequence of functions in L r ( R N ) . If { f n } converges strongly to f in L 1 ( R N ) , then

(2.2) lim n + 1 N α n f n ( x ) = R N f d y , for any x R N .

If, moreover, { g n } is a sequence of functions that converges to g strongly in L 1 ( R N ) , then

(2.3) lim n + R N 1 N α n f n g n d y = R N f d y R N g d y .

Proof

Given ε ( 0 , 1 ) , for every n N , in order to prove (2.2), we see that

1 N α n f n ( x ) R N f n d y = R N 1 x y N α n 1 f n ( y ) d y = B ( x , ε ) + B c ( x , ε ) 1 x y N α n 1 f n ( y ) d y .

For each x R N , combining with the Hölder inequality and Lemma 2.4, it can be seen that

B ( x , ε ) 1 x y N α n 1 f n ( y ) d y = B ( 0 , ε ) 1 y N α n 1 f n ( x + y ) d y B ( 0 , ε ) 1 y N α n + 1 f n ( x + y ) d y 2 B ( 0 , ε ) 1 1 r f n L r ( R N ) + C ( N α n ) ( r α n N ) 1 1 r f n L r ( R N ) .

Now, we prove that for each y R N ,

(2.4) lim n 1 x y N α n 1 χ B c ( 0 , ε ) = 0 ,

where the symbol χ represents the characteristic function. In fact, by means of the triangle inequality, one has

x y x + y ( 1 + x ) ( 1 + y ) .

Moreover, if y R N \ B ( x , ε ) , there holds

x y ( 1 λ ) ρ + λ ( y x ) = λ ( 1 + y ) ,

where λ = ρ x + ρ + 1 . Hence, we conclude that

1 x + 1 N α n 1 ( 1 + y ) N α n 1 x y N α n x + ρ + 1 ρ N α n 1 ( 1 + y ) N α n ,

for y R N \ B ( x , ε ) , which ensures that (2.4) holds. Consequently, together this with Lebesgue’s dominated convergence theorem, one has

lim n B c ( x , ε ) 1 x y N α n 1 f n ( y ) d y = 0 .

Since ε > 0 is arbitrary, this implies that (2.2) holds.

Next, we prove (2.3). We assume that, up to a subsequence, g n g almost everywhere in R N as n and for every n N , there exists g ˜ L 1 ( R N ) such that g n g ˜ in R N . In view of (2.2), we infer that

lim n + 1 N α n f n ( x ) R N f d y ,

for every x R N . By means of Lemma 2.4, the sequence 1 N α n f n n N is uniformly bounded over R N ; combining these facts and Lebesgue’s dominated convergence theorem, then we obtain the desired conclusions.□

According to Lemma 2.5, the following result can be directly obtained.

Corollary 2.6

Let 2 < p < 2 α , s , { α n } > 0 be a sequence converging to N and { u n } H r s ( R N ) be a sequence converging weakly to some u 0 H r s ( R N ) in H s ( R N ) . Then,

R N 1 N α n u n p u n p d x R N u 0 p d x 2 .

In addition, for any ϕ H s ( R N ) ,

R N 1 N α n u n p u n p 2 u n ϕ d x R N u 0 p d x R N u 0 p 2 u 0 ϕ d x .

Now, we introduce the variational structure of Problem (1.7). Naturally, associated with Problem (1.7), the energy functional J : H s ( R N ) R is of the form

J ( u ) = 1 2 R N ( Δ ) s 2 u 2 + u 2 d x 1 2 p R N 1 N α u p u p d x .

It is standard to verify that J C 1 ( H s ( R N ) , R ) . Meanwhile, it is natural to consider the Nehari manifold related to J defined by

N = { u H s ( R N ) \ { 0 } : J ( u ) u = 0 } .

It is obvious that the ground-state energy level c of (1.7) satisfies the mountain pass characterization, where

c inf u H s ( R N ) \ { 0 } max t 0 J ( t u ) .

Moreover,

c = inf u N J ( u ) = J ( ψ ) ,

where the function ψ is the ground-state solution for (1.7). Furthermore, letting α N , Corollary 2.6 indicates that the functional J approaches a limit functional

J N ( u ) = 1 2 R N ( Δ ) s 2 u 2 + u 2 d x 1 2 p R N u p d x 2 .

It can be seen that for p ( 2 , 2 s * ) , the limit functional J N C 1 ( H s ( R N ) , R ) and its Euler-Lagrange equation is

(2.5) ( Δ ) s u + u = R N u p d x u p 2 u , in R N .

The existence and properties of the ground-state to (2.5) can be deduced. Indeed, the solutions of Problem (2.5) and

(2.6) ( Δ ) s v + v = v p 2 v , in R N ,

are related to each other. It is well known that Problem (2.6) has a unique (up to symmetries) positive ground-state solution v , which is radially symmetric and represents a polynomial decay behavior (see [13], for instance). Define

u = v R N v p d x 1 2 p 2 ,

then we infer that u is a positive ground-state solution of Problem (2.5). Furthermore, the ground-state u of Problem (2.5) inherits the sign, uniqueness, symmetry, and decay properties of the ground-state v corresponding to Problem (2.6).

Based on the previous arguments, the following theorem can be established, which is crucial for the proof of Theorem 1.1.

Theorem 2.7

Fix 2 < p < 2 α , s * . Let { u α } be a family of positive radial ground-states to (1.7) for α N , and let u 0 H s ( R N ) be a unique positive radial ground-state of (2.5). Then, there holds

lim α N u α u 0 H s ( R N ) = 0 .

The proof is based on the following lemmas.

Lemma 2.8

There holds

lim α N c = c N ,

where c N inf v H s ( R N ) max t 0 J N ( t v ) .

Proof

For u H s ( R N ) \ { 0 } , in view of Corollary 2.6, we deduce, as α N ,

c max t > 0 J ( t u ) = max t > 0 t 2 2 R N ( Δ ) s 2 u 2 + u 2 d x t 2 p 2 p R N 1 N α u p u p d x max t > 0 t 2 2 R N ( Δ ) s 2 u 2 + u 2 d x t 2 p 2 p R N u p d x 2 = max t > 0 J N ( t u ) .

Taking the infimum with respect to u H s ( R N ) \ { 0 } , we observe

(2.7) limsup α N c c N .

Moreover, we suppose u α H s ( R N ) \ { 0 } satisfying

c = J ( u α ) and J ( u α ) = 0 .

Since

c = J ( u α ) 1 2 p J ( u α ) u α = 1 2 1 2 p R N ( Δ ) s 2 u α 2 + u α 2 d x ,

then Inequality (2.7) ensures that the ground-state solution u α is bounded in H s ( R N ) as α N . Therefore, it can be inferred from Lemma 2.4 that

liminf α N c = liminf α N max t > 0 J ( t u α ) liminf α N max t > 0 J N ( t u α ) c N .

The proof is complete.□

Next, let { α n } > 0 be a sequence such that α n N as n + and choose a sequence { u α n } of positive radial ground-state solutions for (1.7).

Lemma 2.9

There exists a positive radial solution u 0 H s ( R N ) of (2.5) such that, up to a subsequence,

lim n + u α n u 0 H s ( R N ) = 0 .

Proof

Since J ( u α n ) = 0 , we have

J ( u α n ) = 1 2 1 2 p u α n H s ( R N ) 2 .

Then, we see from Lemma 2.8 that u α n H s ( R N ) is uniformly bounded. Hence, up to a subsequence, there exists u 0 H s ( R N ) satisfying

u α n u 0 weakly in H s ( R N ) .

From this and Corollary 2.6, it can be inferred that J N ( u 0 ) = 0 , i.e., u 0 is a weak solution of (2.5). Therefore,

u α n H s ( R N ) 2 = R N 1 N α n u α n p u α n p d x R N u 0 p d x 2 = u 0 H s ( R N ) 2 , as n + ,

which implies u α n u 0 in H s ( R N ) . Then, it remains to prove that u 0 is positive. By means of Lemma 2.3 and Sobolev embedding, we have

u α n H s ( R N ) 2 = R N 1 N α n u α n p u α n p d x C u α n L 2 N p N + α n ( R N ) 2 p C ˜ u α n H s ( R N ) 2 p .

Hence, we obtain a uniform lower bound for u α n H s ( R N ) , which, passing to a limit, ensures that u 0 is nontrivial. Obviously, u 0 is nonnegative; then, the maximum principle implies that u 0 is positive.□

Lemma 2.10

It results that

J N ( u 0 ) = c N .

Namely, u 0 is a unique positive radial ground-state solution of (2.5).

Proof

According to Corollary 2.6 and Lemmas 2.8 and 2.9, one sees immediately that

c N = lim n + c = lim n + J ( u α n ) = lim n + 1 2 u α n H s ( R N ) 2 1 2 p R N ( 1 N α n u α n p ) u α n p d x = 1 2 u 0 H s ( R N ) 2 1 2 p R N u 0 p d x 2 = J N ( u 0 ) .

Proof of Theorem 2.7

As a conclusion of Lemmas 2.9 and 2.10, the proof is complete.□

The following lemma is a fractional version of [29, Lemma 5.1], and its proof is presented in [17, Lemma 3.7]

Lemma 2.11

Let 2 N N + 2 s q 2 . Then, the operator ( ( Δ ) s + I ) 1 is bounded from L q ( R N ) into H s ( R N ) .

For p ( 2 , 2 α , s * ) , choose and fix α N ( N 2 ) p 2 , N , then we define an operator T ( α , u ) given by

T ( α , u ) u ( ( Δ ) s + I ) 1 1 N α u p u p 2 u , α ( α N , N ) , u ( ( Δ ) s + I ) 1 R N u p d x u p 2 u , α = N .

Lemma 2.12

The operator T is a continuous map from ( α N , N ] × H r s ( R N ) into H r s ( R N ) , and T is continuously differentiable with respect to u on ( α N , N ] × H r s ( R N ) .

Proof

Let { ( α j , u j ) } be a sequence in ( α N , N ] × H r s ( R N ) converging to some ( α , u ) ( α N , N ] × H r s ( R N ) . By compactness of H r s ( R N ) L w ( R N ) for every w ( 2 , 2 s * ) , we may assume u j u in L w ( R N ) . We only prove the case: α = N and α j N . To this end, in view of Lemma 2.11, it is sufficient to show that 1 N α j u j p u j p 2 u j converges to R N u p d x u p 2 u in L q ( R N ) , where q = p p 1 . Since p ( 2 , 2 α , s * ) , we observe that q [ 2 N N + 2 s , 2 ] .

Note that

1 N α j u j p u j p 2 u j R N u p d x u p 2 u L q ( R N ) 1 N α j u j p R N u j p d x u j p 2 u j L q ( R N ) + R N u j p d x u j p 2 u j R N u p d x u p 2 u L q ( R N ) I 1 + I 2 .

Now, we claim that

I 1 , I 2 = o ( 1 ) , as j .

Indeed, in view of Lemma 2.5, we can conclude that

I 1 1 N α j u j p R N u p d x + R N u p d x R N u j p d x u j p 2 u j L p p 1 ( R N ) 1 N α j u j p R N u p d x u j p 2 u j L p p 1 ( R N ) + R N u j p d x R N u p d x u j p 2 u j L p p 1 ( R N ) = o ( 1 ) , as j .

On the other hand, it is easy to verify that

I 2 = R N u j p d x R N u p d x u j p 2 u j + R N u p d x ( u j p 2 u j u p 2 u ) L p p 1 ( R N ) R N u j p d x R N u p d x u j p 2 u j L p p 1 ( R N ) + R N u p d x ( u j p 2 u j u p 2 u ) L p p 1 ( R N ) = o ( 1 ) , as j .

Denoting L = ( ( Δ ) s + I ) 1 and differentiating T with respect to u , we have

T u ( α , u ) [ ψ ] = ψ L p 1 N α u p 2 u ψ u p 2 u + ( p 1 ) 1 N α u p u p 2 u ψ , α ( α N , N ) , ψ L p R N u p 1 ψ d x u p 2 u + ( p 1 ) R N u p d x u p 2 ψ , α = N .

Then, one can repeat the same arguments earlier to see that T u is continuous on ( α N , N ] × H r s ( R N ) .□

Lemma 2.13

Let u 0 be the unique positive radial ground-state solution of (2.5). Then, there exists a neighborhood U N ( α N , N ] × H r s ( R N ) of a point ( N , u 0 ) ( α N , N ] × H r s ( R N ) such that (1.7) admits a unique solution in U N .

Proof

Note that

T u ( N , u 0 ) [ ψ ] = ψ L p R N u 0 p 1 ψ d x u 0 p 2 u 0 + ( p 1 ) R N u 0 p d x u 0 p 2 ψ .

Since u 0 satisfies polynomial decay, we deduce that the map

ψ p R N u 0 p 1 ψ d x u 0 p 2 u 0 + ( p 1 ) R N u 0 p d x u 0 p 2 ψ

is compact from H r s ( R N ) into L 2 ( R N ) . Therefore, the composite map

ψ L p R N u 0 p 1 ψ d x u 0 p 2 u 0 + ( p 1 ) R N u 0 p d x u 0 p 2 ψ

is compact from H r s ( R N ) into H r s ( R N ) ; then, it follows that T u ( N , u 0 ) is bounded and the kernel of T u ( N , u 0 ) is trivial. Applying the Fredholm alternative, we know that T u ( N , u 0 ) is an onto map. This concludes that T u ( N , u 0 ) : H r s ( R N ) H r s ( R N ) is a linear isomorphism. Moreover, by means of the implicit function theorem, the desired result is obtained.□

Proof of Theorem 1.1

Assume, by contradiction, that there exist sequences { α i } > 0 , { u α i 1 } H r s ( R N ) and { u α i 2 } H r s ( R N ) such that α i N as i + , { u α i 1 } and { u α i 2 } are the sequences of positive radial ground-state solutions of (1.7), and u α i 1 u α i 2 for any i N . Furthermore, Theorem 2.7 implies that { u α i 1 } and { u α i 2 } converge to a unique positive radial solution u 0 of (2.5) in H r s ( R N ) . This contradicts with Lemma 2.13, which ensures the uniqueness of positive radial ground-states of (1.7) for p ( 2 , 2 α , s * ) and α sufficiently close to N .□

3 Nonexistence result and technical lemmas

In this section, we will give some results that are crucial to prove main theorems. To this end, we choose α N sufficiently close to N and fix throughout this section such that Problem (1.7) has a unique positive radial ground-state solution for every α ( α N , N ) .

Some necessary notations are as follows: let us denote by J ε , Ω : H 0 s ( Ω ) R the energy functional associated with Problem (1.9), defined by

J ε , Ω ( u ) = 1 2 Q u ( x ) u ( y ) 2 x y N + 2 s d y d x + Ω u 2 d x 1 2 p Ω Ω u ( y ) p u ( x ) p x y N α d y d x ε 2 2 α , s * Ω Ω u ( y ) 2 α , s * u ( x ) 2 α , s * x y N α d y d x .

Meanwhile, we need to consider the following functionals:

J ε , ( u ) 1 2 R N ( Δ ) s 2 u 2 + u 2 d x 1 2 p R N R N u ( y ) p u ( x ) p x y N α d y d x ε 2 2 α , s * R N R N u ( y ) 2 α , s * u ( x ) 2 α , s * x y N α d y d x

and

J Ω ( u ) 1 2 Q u ( x ) u ( y ) 2 x y N + 2 s d y d x + Ω u 2 d x 1 2 p Ω 1 N α u p u p d x ,

which correspond to the limit Problems (1.10) and (1.8), respectively. Using standard arguments, it is easy to prove that the functionals J ε , Ω , J Ω C 1 ( H 0 s ( Ω ) , R ) and J ε , C 1 ( H s ( R N ) , R ) . Moreover, the Nehari manifolds associated with J ε , Ω , J ε , , and J Ω are given by

N ε , Ω { u H 0 s ( Ω ) \ { 0 } : J ε , Ω ( u ) u = 0 } , N ε , { u H s ( R N ) \ { 0 } : J ε , ( u ) u = 0 } ,

and

N Ω { u H 0 s ( Ω ) \ { 0 } : J Ω ( u ) u = 0 } .

Furthermore, we define the following minimization problems:

c ε , Ω inf u N ε , Ω J ε , Ω ( u )

and

c Ω inf u N Ω J Ω ( u ) .

Obviously, the functional J ε , possesses a mountain pass geometry. This means that there exist r > 0 and θ > 0 , independent of ε , such that if u H s ( R N ) = r , then J ε , ( u ) θ > 0 , and there exists e H s ( R N ) such that e H s ( R N ) > r and J ε , ( e ) < 0 . Therefore, we can consider the mountain pass value

c ε , inf γ Γ max t [ 0 , 1 ] J ε , ( γ ( t ) ) ,

where

Γ { γ C ( [ 0 , 1 ] , H s ( R N ) ) : γ ( 0 ) = 0 and γ ( 1 ) = e } .

On the other hand, it is easy to check that the functionals J ε , Ω and J Ω also have the mountain pass geometry and c ε , Ω , c Ω θ > 0 for all ε > 0 .

Lemma 3.1

There holds

c ε , = inf u N ε , J ε , ( u ) .

Proof

Letting u N ε , , it is obvious that

(3.1) max t 0 J ε , ( t u ) = J ε , ( u ) .

We can take t ¯ R + and v = t ¯ u such that J ε , ( v ) < 0 , which implies γ ( t ) = t v Γ and then c ε , J ε , ( u ) . Thus, we can infer that

c ε , inf u N ε , J ε , ( u ) .

On the other hand, set { u n } H s ( R N ) is a ( P S ) c ε , sequence for J ε , , i.e.,

J ε , ( u n ) c ε , and J ε , ( u n ) u n 0 , as n + .

In view of (3.1), for every n N , there exists t n R + such that

J ε , ( t n u n ) t n u n = 0 .

Hence, t n u n N ε , , and it follows that

(3.2) u n H s ( R N ) 2 = t n 2 p 2 R N R N u n ( y ) p u n ( x ) p x y N α d y d x + ε t n 2 2 α , s * 2 R N R N u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x .

Due to J ε , ( u n ) u n = o n ( 1 ) and the fact that c ε , > 0 , we can directly conclude that { t n } is bounded and t n 0 as n + . Then, we can assume that there exists t ( 0 , ) such that, up to a subsequence,

t n t , as n + .

Combining (3.2) and J ε , ( u n ) u n = o n ( 1 ) , we have

(3.3) o n ( 1 ) = ( 1 t n 2 p 2 ) R N R N u n ( y ) p u n ( x ) p x y N α d y d x + ε ( 1 t n 2 2 α , s * 2 ) R N R N u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x .

Taking the limit, one has

( 1 ( t ) 2 p 2 ) L + ε ( 1 ( t ) 2 2 α , s * 2 ) M = 0 ,

where

R N R N u n ( y ) p u n ( x ) p x y N α d y d x L

and

R N R N u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x M .

Hence, t = 1 , which means that t n 1 as n + . By virtue of t n u n N ε , , the following holds:

inf u N ε , J ε , ( u n ) J ε , ( t n u n ) = t n 2 J ε , ( u n ) + ( 1 t n 2 p 2 ) 2 p R N R N u n ( y ) p u n ( x ) p x y N α d y d x + ε ( 1 t n 2 2 α , s * 2 ) 2 2 α , s * R N R N u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x = t n 2 J ε , ( u n ) + o n ( 1 ) = ( t n 2 1 ) J ε , ( u n ) + J ε , ( u n ) + o n ( 1 ) ,

which indicates that

c ε , inf u N ε , J ε , ( u ) .

The proof follows.□

Now, define

(3.4) S h , l inf u D s , 2 ( R N ) \ { 0 } R N ( Δ ) s 2 u 2 d x R N R N u ( y ) 2 α , s * u ( x ) 2 α , s * x y N α d y d x 1 2 α , s * ,

which is well-defined (see [15, Section 2] for instance). Then, the following result holds.

Lemma 3.2

There exists ε ˜ > 0 such that if ε ( 0 , ε ˜ ) , then

c ε , 0 , α + 2 s 2 N + 2 α ε 2 s N α + 2 s S h , l N + α α + 2 s .

Proof

Let u be a ground-state solution of Problem (1.7), then there exists t > 0 such that t u N ε , . Thus, we can derive that

c ε , J ε , ( t u ) J ( t u ) max t 0 J ( t u ) = J ( u ) = c .

Note that

c < α + 2 s 2 N + 2 α ε 2 s N α + 2 s S h , l N + α α + 2 s ,

for ε > 0 small enough. Hence, the proof is complete.□

Lemma 3.3

Let { u n } H s ( R N ) be a ( P S ) c ε , sequence for J ε , . Then, there exist τ , ρ > 0 and { z n } R N such that

limsup n B ( z n , τ ) u n ( x ) 2 d x ρ > 0 .

Proof

Assume, by contradiction, that Lemmas 2.12.2 state that

R N R N u n ( y ) p u n ( x ) p x y N α d y d x C R N u n ( x ) 2 N p N + α d x N + α N = o n ( 1 ) ,

which guarantees that

c ε , = J ε , ( u n ) + o n ( 1 ) = 1 2 u n H s ( R N ) 2 ε 2 2 α , s * R N R N u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x + o n ( 1 ) = 1 2 1 2 2 α , s * u n H s ( R N ) 2 + o n ( 1 ) = α + 2 s 2 N + 2 α u n H s ( R N ) 2 + o n ( 1 ) .

According to the definition of S h , l and the fact that J ε , ( u n ) u n = o n ( 1 ) , we observe

c ε , = α + 2 s 2 N + 2 α u n H s ( R N ) 2 + o n ( 1 ) α + 2 s 2 N + 2 α S h , l N + α α + 2 s ε 2 s N α + 2 s + o n ( 1 ) , for any ε > 0 ,

which contradicts Lemma 3.2.□

Now, we present the existence result for Problem (1.10).

Proposition 3.4

There exists ε ˜ > 0 such that for each ε ( 0 , ε ˜ ) , there exists a positive function ψ ε N ε , , radially symmetric in the origin satisfying J ε , ( ψ ε ) = 0 and

c ε , = J ε , ( ψ ε ) = inf u N ε , J ε , ( u ) ,

where c ε , denotes the mountain pass level of J ε , . Then, ψ ε is a ground-state solution of Problem (1.10).

Proof

According to the mountain pass theorem [32], we can deduce that J ε , satisfies the mountain pass geometry. Therefore, there exists a sequence { u n } H s ( R N ) satisfying

(3.5) J ε , ( u n ) c ε , and J ε , ( u n ) 0 , as n + .

Combining with (3.5) and p < 2 α , s * , we observe that

J ε , ( u n ) + u n H s ( R N ) o n ( 1 ) + o n ( 1 ) J ε , ( u n ) 1 2 p J ε , ( u n ) u n = 1 2 1 2 p u n H s ( R N ) 2 + ε 1 2 p 1 2 2 α , s * R N R N u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x 1 2 1 2 p u n H s ( R N ) 2 ,

which implies that { u n } is bounded in H s ( R N ) . Then, there exists u H s ( R N ) such that, up to a subsequence, u n u in H s ( R N ) , u n u in L loc r ( R N ) for 1 r < 2 s * , and u n u a.e. in R N .

On the one hand, if u = 0 , then the fact that c ε , > 0 indicates that u n 0 in H s ( R N ) . By Lemma 3.3, we can conclude that there exist τ , ρ > 0 and { z n } R N such that

(3.6) B ( z n , τ ) u n 2 d x ρ > 0 .

Let us consider the sequence w n ( ) u n ( + z n ) . Note that

w n H s ( R N ) = u n H s ( R N ) , J ε , ( w n ) = J ε , ( u n ) , and J ε , ( w n ) = o n ( 1 ) .

Then, { w n } is a ( P S ) c ε , sequence for J ε , and { w n } is bounded in H s ( R N ) . Furthermore, there exists w H s ( R N ) such that w n w in H s ( R N ) and J ε , ( w ) = 0 . In addition, combining (3.6), Brezis-Lieb lemma, and the compactness of H s ( R N ) L 2 ( B ( 0 , τ ) ) , we observe that

ρ u n L 2 ( B ( z n , τ ) ) = w n L 2 ( B ( 0 , τ ) ) = w n w L 2 ( B ( 0 , τ ) ) + w L 2 ( B ( 0 , τ ) ) = o n ( 1 ) + w L 2 ( B ( 0 , τ ) ) ,

which ensures that w is nontrivial. Moreover, by virtue of Fatou’s lemma, we obtain that

c ε , J ε , ( w ) 1 2 J ε , ( w ) w = 1 2 1 2 p R N R N w ( y ) p w ( x ) p x y N α d y d x + ε 1 2 1 2 2 α , s * R N R N w ( y ) 2 α , s * w ( x ) 2 α , s * x y N α d y d x 1 2 1 2 p liminf n + R N R N w n ( y ) p w n ( x ) p x y N α d y d x + ε 1 2 1 2 2 α , s * liminf n + R N R N w n ( y ) 2 α , s * w n ( x ) 2 α , s * x y N α d y d x liminf n + 1 2 1 2 p R N R N w n ( y ) p w n ( x ) p x y N α d y d x + ε 1 2 1 2 2 α , s * R N R N w n ( y ) 2 α , s * w n ( x ) 2 α , s * x y N α d y d x = liminf n + J ε , ( w n ) 1 2 J ε , ( w n ) w n = c ε , .

Hence, we have J ε , ( w ) = c ε , . Consequently, w is a ground-state solution to limit Problem (1.10).

On the other hand, if u 0 , we claim that u is a weak solution of (1.10). In view of the weak version of Brezis-Lieb-type result for the Choquard term (see [18, Lemma 2.4] for the subcritical case and [16, formula (2.15)] for the critical case), it is clearly seen that J ε , ( u ) = 0 , i.e., the claim holds. Similar to the previous arguments, we can deduce that u is a ground-state solution to limit Problem (1.10).

Without loss of generality, the ground-state solution of limit Problem (1.10) is denoted by w . Let w * be the symmetric decreasing rearrangement of w . It is clear that

w L 2 ( R N ) = w * L 2 ( R N ) .

Furthermore, from the fractional Polya-Szegö inequality in [25] and formula (A.11) in [30], we have

( Δ ) s 2 w * L 2 ( R N ) ( Δ ) s 2 w L 2 ( R N ) ( Δ ) s 2 w L 2 ( R N ) .

By Riesz’s rearrangement inequality (see [20, Theorem 3.4]), there hold

R N R N w * ( y ) p w * ( x ) p x y N α d y d x R N R N w ( y ) p w ( x ) p x y N α d y d x

and

R N R N w * ( y ) 2 α , s * w * ( x ) 2 α , s * x y N α d y d x R N R N w ( y ) 2 α , s * w ( x ) 2 α , s * x y N α d y d x .

Thus,

c ε , J ε , ( t * w * ) J ε , ( t * w ) max t 0 J ε , ( t w ) = J ε , ( w ) = c ε ,

for some t * > 0 satisfying t * w * N ε , , which states that J ε , ( t * w * ) = c ε , and t * w * is a critical point of J ε , in N ε , . Let us denote ψ ε t * w * , so ψ ε H r s ( R N ) and

J ε , ( ψ ε ) = c ε , = inf u N ε , J ε , ( u ) ,

as desired.□

In what follows, let 0 < ε < ε ˜ and ψ ε be the ground-state solution of Problem (1.10), where ε ˜ and ψ ε are given in Proposition 3.4.

Lemma 3.5

There holds

(3.7) c ε , Ω = c ε , .

Proof

Define

w n ( x ) ξ ( x ) ψ ε ( x z n ) ,

where { z n } Ω is a sequence of points satisfying z n + as n + . In particular, ξ C ( R N , [ 0 , 1 ] ) is a cut-off function defined by

ξ ( x ) = 0 , if x B ( 0 , η ) , 1 , if x B c ( 0 , 2 η ) ,

where η is the smallest positive number such that R N \ Ω B ( 0 , η ) . Choose t n > 0 such that t n w n N ε , Ω , and then,

(3.8) c ε , Ω = inf u N ε , Ω J ε , Ω ( u ) J ε , Ω ( t n w n ) = J ε , ( t n w n ) , n N .

Now, we prove that

(3.9) ξ ( + z n ) ψ ε ψ ε in H s ( R N ) , as n + ,

i.e.,

R N ( Δ ) s 2 [ ξ ( x + z n ) ψ ε ( x ) ψ ε ( x ) ] 2 + ξ ( x + z n ) ψ ε ( x ) ψ ε ( x ) 2 d x = o n ( 1 ) .

Note that

R N ( Δ ) s 2 [ ξ ( x + z n ) ψ ε ( x ) ψ ε ( x ) ] 2 d x = R N R N [ ξ ( x + z n ) ψ ε ( x ) ψ ε ( x ) ] [ ξ ( y + z n ) ψ ε ( y ) ψ ε ( y ) ] 2 x y N + 2 s d y d x = R N R N [ ξ ( x + z n ) 1 ] ψ ε ( x ) [ ξ ( y + z n ) 1 ] ψ ε ( y ) 2 x y N + 2 s d y d x .

Then, we define

Φ n ( x , y ) [ ξ ( x + z n ) 1 ] ψ ε ( x ) [ ξ ( y + z n ) 1 ] ψ ε ( y ) 2 x y N + 2 s .

Since lim n + z n = + , we observe that

Φ n ( x , y ) 0 , a.e. in R N × R N .

In addition, a direct application of the mean value theorem gives

Φ n ( x , y ) ψ ε ( x ) ψ ε ( y ) 2 x y N + 2 s + C 1 ψ ε ( y ) 2 x y N + 2 s 2 χ B ( y , 1 ) ( x ) + C 2 ψ ε ( y ) 2 x y N + 2 s χ B ( y , 1 ) c ( x ) L 1 ( R N × R N ) .

Consequently, the Lebesgue’s dominated convergence theorem leads to

R N R N [ ξ ( x + z n ) 1 ] ψ ε ( x ) [ ξ ( y + z n ) 1 ] ψ ε ( y ) 2 x y N + 2 s d y d x 0 , as n + .

On the other hand, let Φ n ( x ) = ξ ( x + z n ) ψ ε ( x ) ψ ε ( x ) 2 . Since

Φ n ( x ) 0 , a.e. in R N ,

and

Φ n ( x ) 4 ψ ε ( x ) 2 L 1 ( R N ) .

Using the Lebesgue’s dominated convergence theorem again, we have

R N ξ ( x + z n ) ψ ε ( x ) ψ ε ( x ) 2 d x 0 , as n + .

Hence, we can conclude that (3.9) holds.

Finally, we claim that

(3.10) lim n + t n 1 .

Once the claim holds, then

J ε , ( t n w n ) = J ε , ( ψ ε ) + o n ( 1 ) = c ε , + o n ( 1 ) ,

which, together with (3.8), signifies that

c ε , Ω c ε , .

Thus, the fact that N ε , Ω N ε , immediately implies that the desired result is reached. Now, we prove (3.10). Since t n w n N ε , Ω , we observe that

t n w n H 0 s ( Ω ) 2 = Ω Ω t n w n ( y ) p t n w n ( x ) p x y N α d y d x + ε Ω Ω t n w n ( y ) 2 α , s * t n w n ( x ) 2 α , s * x y N α d y d x .

By virtue of the definition of the w n , we infer that

w n H s ( R N ) 2 = t n 2 p 2 R N R N ξ ( x + z n ) ψ ε ( y ) p ξ ( x + z n ) ψ ε ( x ) p x y N α d y d x + ε t n 2 2 α , s * 2 R N R N ξ ( x + z n ) ψ ε ( y ) 2 α , s * ξ ( x + z n ) ψ ε ( x ) 2 α , s * x y N α d y d x .

Obviously, { t n } is bounded and t n 0 as n + . Then, it is straightforward to see from (3.9) and ψ ε N ε , that (3.10) holds.□

Proof of Theorem 1.3

If there exists u such that u H 0 s ( Ω ) ,

J ε , Ω ( u ) = 0 and J ε , Ω ( u ) = inf u N ε , Ω J ε , Ω ( u ) = c ε , Ω = c ε , .

Without loss of generality, it can be supposed u * 0 in Ω . Letting u = 0 in R N \ Ω , we observe that u H s ( R N ) ,

J ε , ( u ) u = 0 and J ε , ( u ) = c ε , ,

which imply that u is a ground-state solution of Problem (1.10). Then, the maximum principle ensures that u is strictly positive in R N , which is impossible.□

Arguing as [6, Lemma 3.1], one can have the following compactness lemma.

Lemma 3.6

Let { u n } be a ( P S ) c sequence for J Ω . Then, up to a subsequence, there exist k N and k sequences of points { z n j } such that z n j + as n + and k + 1 sequences of functions { u n j } H s ( R N ) , 0 j k , such that

u n ( x ) = u n 0 ( x ) + j = 1 k u n j ( x z n j ) ,

u n 0 = u n u 0 , in H 0 s ( Ω ) ,

u n j u j in H s ( R N ) , for 1 j k ,

where u 0 is a weak solution of (1.8) and u j are nontrivial solutions of equation (1.7) for every 1 j k . Furthermore,

u n H 0 s ( Ω ) 2 u 0 H 0 s ( Ω ) 2 + j = 1 k u j H s ( R N ) 2

and

J Ω ( u n ) J Ω ( u 0 ) + j = 1 k J ( u j ) .

Based on the fact of Theorem 1.3, we can construct some energy-level interval close to c and verify the compactness of nonnegative Palais-Smale sequences of J ε , Ω at these energy levels.

Proposition 3.7

For every δ 0 , c 2 , there exists ε ¯ > 0 , such that, if { u n } is a nonnegative ( P S ) c sequence of J ε , Ω with c ( c + δ , 2 c δ ) and ε ( 0 , ε ¯ ) , then u n u 0 0 in H 0 s ( Ω ) .

Proof

Assume, by contradiction, that there exist δ 0 0 , c 2 , a sequence { ε i } with ε i 0 and { u n i } H 0 s ( Ω ) such that

(3.11) J ε i , Ω ( u n i ) d i and J ε i , Ω ( u n i ) 0 , as n + ,

with

(3.12) d i ( c + δ 0 , 2 c δ 0 ) ,

and we suppose that for every i , u n i 0 in H 0 s ( Ω ) as n + .

Firstly, we claim that { u n i } is bounded in H 0 s ( Ω ) . Note that, for every i ,

(3.13) d i + o n ( 1 ) u n i H 0 s ( Ω ) + o n ( 1 ) J ε i , Ω ( u n i ) 1 2 p J ε i , Ω ( u n i ) u n i = 1 2 1 2 p u n i H 0 s ( Ω ) 2 + ε 1 2 p 1 2 2 α , s * R N R N u n i ( y ) 2 α , s * u n i ( x ) 2 α , s * x y N α d y d x 1 2 1 2 p u n i H 0 s ( Ω ) 2 ,

which ensures the claim holds. Moreover, we have the following estimate:

(3.14) limsup n + u n i H 0 s ( Ω ) 2 1 2 1 2 p 1 d i 1 2 1 2 p 1 ( 2 c δ 0 ) .

Next, we claim that there exist ρ > 0 , r > 0 and i 0 N such that

liminf n + sup z R N B ( z , r ) u n i 2 d x ρ , for every i i 0 .

Indeed, if the case does not occur, then for every r > 0 and j N , there is { i j } N with i j + as j + satisfying

liminf n + sup z R N B ( z , r ) u n i j 2 d x < 1 j ,

which, together with Lemma 2.1, yields that

limsup n + Ω u n i j q d x = o j ( 1 ) , q ( 2 , 2 s * ) ,

and then,

limsup n + Ω Ω u n i j ( y ) p u n i j ( x ) p x y N α d y d x = o j ( 1 ) .

Thus, using J ε n j , Ω ( u n i j ) = o j ( 1 ) , we infer that

(3.15) limsup n + u n i j H 0 s ( Ω ) 2 = ε i j limsup n + Ω Ω u n i j ( y ) 2 α , s * u n i j ( x ) 2 α , s * x y N α d y d x + o j ( 1 ) .

It follows from (3.4) that

limsup n + Ω Ω u n i j ( y ) 2 α , s * u n i j ( x ) 2 α , s * x y N α d y d x S h , l 2 α , s * [ limsup n + u n i j H 0 s ( Ω ) 2 ] 2 α , s * .

Combining (3.14), one has

limsup n + Ω Ω u n i j ( y ) 2 α , s * u n i j ( x ) 2 α , s * x y N α d y d x S h , l 2 α , s * 1 2 1 2 p 1 ( 2 c δ 0 ) 2 α , s * ,

which implies that

(3.16) ε i j limsup n + Ω Ω u n i j ( y ) 2 α , s * u n i j ( x ) 2 α , s * x y N α d y d x = o j ( 1 ) .

In view of (3.15) and (3.16), we deduce that

limsup n + u n i j H 0 s ( Ω ) 2 = o j ( 1 ) .

Furthermore, there holds

d i j = limsup n + J ε i j , Ω ( u n i j ) limsup n + 1 2 u n i j H 0 s ( Ω ) 2 = o j ( 1 ) ,

which contradicts with (3.12). Thus, the claim is true. Moreover, up to a subsequence of { u n i } , there exists { z n i } R N with z n i + as n + such that

(3.17) B ( z n i , r ) u n i 2 d x ρ 2 , for any n , i N .

Therefore, combining with (3.11) and (3.17), for every n N , we can choose n i N satisfying

(3.18) z n i i > i , B ( z n i i , r ) u n i i 2 d x ρ 2 ,

(3.19) J ε i , Ω ( u n i i ) d i < 1 i and J ε i , Ω ( u n i i ) < 1 i .

Let us denote the sequences { z n i i } and { u n i i } by { z i } and { u i } , respectively. Namely, (3.18)–(3.19) hold for { z i } and { u i } . Arguing as (3.13) and (3.16), we deduce that { u i } is bounded in H 0 s ( Ω ) . Furthermore,

J ε i , Ω ( u i ) J Ω ( u i ) = ε i 2 2 α , s * Ω Ω u i ( y ) 2 α , s * u i ( x ) 2 α , s * x y N α d y d x = o i ( 1 )

and

sup φ H 0 s ( Ω ) 1 J ε i , Ω ( u i ) φ J Ω ( u i ) φ sup φ H 0 s ( Ω ) 1 ε n Ω Ω u i ( y ) 2 α , s * u i ( x ) 2 α , s * 2 u i ( x ) φ ( x ) x y N α d y d x = o i ( 1 ) ,

for φ H 0 s ( Ω ) . Then, it holds that

J Ω ( u i ) d [ c + δ 0 , 2 c δ 0 ] and J Ω ( u i ) 0 , as i + .

Namely, { u i } is a ( P S ) d sequence for J Ω . Since { u i } is bounded in H 0 s ( Ω ) , there exists u 0 H 0 s ( Ω ) such that, up to a subsequence, u i u 0 in H 0 s ( Ω ) and J Ω ( u 0 ) = 0 .

Now, we show that u i u 0 in H 0 s ( Ω ) . If this is false, according to the aforementioned facts and z i + as i + , there holds

B ( z i , r ) u 0 2 d x < ρ 4 and B ( z i , r ) u i 2 d x B ( z i , r ) u 0 2 d x < ρ 4 .

Therefore,

ρ 2 B ( z i , r ) u i 2 d x < B ( z i , r ) u 0 2 d x + ρ 4 < ρ 2 ,

which is absurd. Then, by means of Lemma 3.6, there exist k N , k 1 and u 1 , u 2 , , u k solutions of Problem (1.7) such that

(3.20) J Ω ( u i ) = J Ω ( u 0 ) + j = 1 k J ( u j ) + o i ( 1 ) .

Finally, we claim that u 0 0 . Assume, by contradiction, that u 0 = 0 , and then, (3.20) leads to

J Ω ( u i ) = J Ω ( u 0 ) + j = 1 k J ( u j ) + o i ( 1 ) k c + o i ( 1 ) ,

which ensures that d k c . Then, we must have k = 1 and u 1 > 0 . Hence,

J Ω ( u i ) = J ( u 1 ) + o i ( 1 ) = c + o i ( 1 ) ,

which is a contradiction.□

As a consequence of Proposition 3.7, we can prove the following result.

Proposition 3.8

For every δ 0 , c 2 , there exists ε ¯ > 0 , such that, if { u n } is a nonnegative ( P S ) c sequence for J ε , Ω restricted to N ε , Ω with c ( c + δ , 2 c δ ) and ε ( 0 , ε ¯ ) , then u n u 0 0 in H 0 s ( Ω ) . Moreover, u 0 is a critical point of J ε , Ω in N ε , Ω .

Proof

Let { u n } N ε , Ω be such that

J ε , Ω ( u n ) c ( c + δ 0 , 2 c δ 0 ) and J ε , Ω ( u n ) * = o n ( 1 ) , as n + ,

where

J ε , Ω ( u ) * sup J ε , Ω ( u ) φ : φ T u N ε , Ω and φ H 0 s ( Ω ) 1 , u N ε , Ω .

Hence, there is a sequence { μ n } R so that

(3.21) J ε , Ω ( u n ) = μ n I ε , Ω ( u n ) + o n ( 1 ) ,

where

I ε , Ω ( u ) = u H 0 s ( Ω ) 2 Ω Ω u ( y ) p u ( x ) p x y N α d y d x ε Ω Ω u ( y ) 2 α , s * u ( x ) 2 α , s * x y N α d y d x .

Note that

I ε , Ω ( u n ) u n = I ε , Ω ( u n ) u n 2 p I ε , Ω ( u n ) = 2 u n H 0 s ( Ω ) 2 2 p Ω Ω u n ( y ) p u n ( x ) p x y N α d y d x 2 2 α , s * ε Ω Ω u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x 2 p u n H 0 s ( Ω ) 2 + 2 p Ω Ω u n ( y ) p u n ( x ) p x y N α d y d x + 2 p ε Ω Ω u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x = ( 2 2 p ) u n H 0 s ( Ω ) 2 + ε ( 2 p 2 2 α , s * ) Ω Ω u n ( y ) 2 α , s * u n ( x ) 2 α , s * x y N α d y d x < 0 .

Thus, we can assume that I ε , Ω ( u n ) u n l 0 . If l = 0 , it can be inferred from

I ε , Ω ( u n ) u n ( 2 2 p ) u n H 0 s ( Ω ) 2

that u n H 0 s ( Ω ) 0 as n + , which contradicts with u n H 0 s ( Ω ) > 0 . Therefore, l 0 and μ n = o n ( 1 ) . In addition, we derive from (3.21) that J ε , Ω ( u n ) = o n ( 1 ) and { u n } is a ( P S ) c sequence for J ε , Ω in H 0 s ( Ω ) . It follows from Proposition 3.7 that the proof is complete.□

4 Proof of Theorem 1.4

In this section, we focus our main attention to establish the proof of Theorem 1.4. First, we give some properties of the levels c ε , and c .

Lemma 4.1

There holds

lim ε 0 c ε , = c .

Proof

For every ε ( 0 , ε ˜ ) , where ε ˜ is given in Proposition 3.4, let ψ ε N ε , be a ground-state solution of Problem (1.10) and let t ε > 0 satisfying t ε ψ ε N . Let us consider a sequence { ε n } with ε n 0 ; thus,

1 2 1 2 p ψ ε n H s ( R N ) c ε , c ,

which yields that { ψ ε n } is bounded in H r s ( R N ) . Hence, there exists ψ 0 H r s ( R N ) such that, up to a subsequence, ψ ε n ψ 0 in H r s ( R N ) . Note that H r s ( R N ) is continuously embedding in L q ( R N ) for all q [ 2 , 2 s * ] and compactly embedding if q ( 2 , 2 s * ) . By means of Lemma 2.2 and ψ ε n N ε n , , we observe that

ψ ε n H s ( R N ) 2 = R N R N ψ 0 ( y ) p ψ 0 ( x ) p x y N α d y d x + o n ( 1 )

and

ψ 0 H s ( R N ) 2 = R N R N ψ 0 ( y ) p ψ 0 ( x ) p x y N α d y d x .

It follows that

(4.1) ψ ε n H s ( R N ) 2 = ψ 0 H s ( R N ) 2 + o n ( 1 ) .

We claim that ψ 0 0 . Otherwise, if ψ 0 = 0 , we infer from (4.1) that ψ ε n 0 in H r s ( R N ) and

0 < θ c ε n , = J ε n , ( ψ ε n ) = o n ( 1 ) .

This contradiction validates the claim. Hence,

(4.2) ψ ε n ψ 0 0 in H r s ( R N ) , as n + .

In addition, since t ε n ψ ε n N , one has

ψ ε n H s ( R N ) 2 = t ε n 2 p 2 R N R N ψ ε n ( y ) p ψ ε n ( x ) p x y N α d y d x ,

which, together with (4.2), yields that t ε n 1 . Furthermore, it can be directly seen from Lemma 2.2 and Sobolev embedding that

ε n 2 2 α , s * R N R N t ε n ψ ε n ( y ) 2 α , s * t ε n ψ ε n ( x ) 2 α , s * x y N α d y d x ε n t n 2 α , s * 2 2 α , s * C ψ ε n H s ( R N ) 2 α , s * = o n ( 1 ) .

Therefore,

c J ( t ε n ψ ε n ) = J ε n , ( t ε n ψ ε n ) + ε n 2 2 α , s * R N R N t ε n ψ ε n ( y ) 2 α , s * t ε n ψ ε n ( x ) 2 α , s * x y N α d y d x J ε n , ( t ε n ψ ε n ) + o n ( 1 ) = c ε n , + o n ( 1 ) c + o n ( 1 ) .

The proof follows.□

Let us define the barycentric function β : H s ( R N ) R N by

β ( u ) = R N u ( x ) 2 ϱ ( x ) x d x ,

where ϱ C ( R + , R ) is a non-increasing real function defined by

ϱ ( t ) = 1 , if 0 t r , r t , if t > r ,

for some r > 0 such that R N \ Ω B ( 0 , r ) . We define the sets

P { u H 0 s ( Ω ) : u 0 } , C 0 { u N Ω P : β ( u ) = 0 } , C 0 , ε { u N ε , Ω P : β ( u ) = 0 } ,

and c 0 , c 0 , ε given by

c 0 = inf u C 0 J Ω ( u ) and c 0 , ε = inf u C 0 , ε J ε , Ω ( u ) .

It is clearly seen that

c 0 c Ω θ and c 0 , ε c ε , Ω θ .

Lemma 4.2

There holds

lim ε 0 c 0 , ε = c 0 .

Proof

For every ε > 0 , we fix u ε C 0 satisfying

J Ω ( u ε ) c 0 + ε .

Moreover, we consider τ ε u ε > 0 such that τ ε u ε N ε , Ω . Then,

c 0 , ε J ε , Ω ( τ ε u ε ) J Ω ( τ ε u ε ) max τ > 0 J Ω ( τ u ε ) = J Ω ( u ε ) c 0 + ε .

Thus, it suffices to prove that the reverse inequality. Let w ε C 0 , ε and t ε > 0 be such that J ε , Ω ( w ε ) c 0 , ε + ε and t ε w ε N Ω . Hence, t ε w ε C 0 and

(4.3) c 0 = inf u C 0 J Ω ( u ) J Ω ( t ε w ε ) = J ε , Ω ( t ε w ε ) + ε 2 2 α , s * Ω Ω t ε w ε ( y ) 2 α , s * t ε w ε ( x ) 2 α , s * x y N α d y d x J ε , Ω ( w ε ) + ε 2 2 α , s * Ω Ω t ε w ε ( y ) 2 α , s * t ε w ε ( x ) 2 α , s * x y N α d y d x c 0 , ε + ε + ε t ε 2 2 α , s * 2 2 α , s * Ω Ω w ε ( y ) 2 α , s * w ε ( x ) 2 α , s * x y N α d y d x .

Since w ε N ε , Ω , we deduce that

J ε , Ω ( w ε ) = 1 2 1 2 p w ε H 0 s ( Ω ) 2 + ε 1 2 p 1 2 2 α , s * Ω Ω w ε ( y ) 2 α , s * w ε ( x ) 2 α , s * x y N α d y d x c 0 , ε + ε ,

which ensures that Ω Ω w ε ( y ) 2 α , s * w ε ( x ) 2 α , s * x y N α d y d x is bounded. By the mountain pass geometry of J ε , Ω , we infer that w ε H 0 s ( Ω ) 0 and

Ω Ω w ε ( y ) p w ε ( x ) p x y N α d y d x 0 .

Combining the aforementioned arguments and t ε w ε N Ω , we find that { t ε } is bounded. Hence,

t ε 2 2 α , s * 2 2 α , s * Ω Ω w ε ( y ) 2 α , s * w ε ( x ) 2 α , s * x y N α d y d x C , for some C > 0 ,

which, together with (4.3), states that c 0 c 0 , ε + o ε ( 1 ) . The thesis follows.□

For further discussions, we shall define D by

D { u V P : β ( u ) = 0 } ,

where

V = u H 0 s ( Ω ) : R N R N u ( y ) p u ( x ) p x y N α d y d x = 1 .

Then, it is necessary to consider the following minimization problems:

b = inf u D u H 0 s ( Ω ) 2 and K = inf u V u H 0 s ( Ω ) 2 .

Lemma 4.3

There holds b > K .

Proof

It is obvious that b K . We claim that b K . Otherwise, there exists a minimizing sequence { w n } H 0 s ( Ω ) verifying

Ω Ω w n ( y ) p w n ( x ) p x y N α d y d x = 1 , β ( w n ) = o n ( 1 ) , and w n H 0 s ( Ω ) 2 = K + o n ( 1 ) .

According to the Ekeland variational principle, we assume that

h ( w n ) * 0 , where h ( u ) = Q u ( x ) u ( y ) 2 x y N + 2 s d y d x + Ω u 2 d x = u H 0 s ( Ω ) 2 .

Letting u n = K 1 2 ( p 2 ) w n , we can see that

J Ω ( u n ) = 1 2 1 2 p K p p 1 + o n ( 1 ) and J Ω ( u n ) = o n ( 1 ) , in H 0 s ( Ω ) .

Combining Lemma 3.6 and the uniqueness of ground-state solutions of Problem (1.7), we infer that there exists a sequence { z n } R N with z n + and v n 0 in H s ( R N ) , such that

u n ( x ) = ψ ( x z n ) + v n ( x ) , for x R N .

We consider the sets ( R N ) n + = { x R N x , z n > 0 } and ( R N ) n = R N \ ( R N ) n + . Therefore, there exists B ( z n , r 0 ) ( R N ) n + such that

(4.4) ψ ( x z n ) 1 2 ψ ( 0 ) > 0 , for all x B ( z n , r 0 ) and n 1 .

Here, we have used the radial lemma and the mean value theorem. Furthermore, similar to [33, Lemma 3.2], we can deduce that

( R N ) n ψ ( x z n ) 2 ϱ ( x ) x d x = o n ( 1 ) ,

which, together with (4.4) and the Cauchy-Schwarz inequality, states that

β ( ψ ( x z n ) ) , z n z n R N = ( R N ) n + ψ ( x z n ) 2 ϱ ( x ) x , z n z n R N d x + ( R N ) n ψ ( x z n ) 2 ϱ ( x ) x , z n z n R N d x B ( z n , r 0 ) 1 4 ψ ( 0 ) 2 ϱ ( x ) x , z n z n d x ( R N ) n ψ ( x z n ) 2 ϱ ( x ) x d x C o n ( 1 ) > 0 ,

where C is a positive constant. However, observing that v n 0 in H s ( R N ) and β ( u n ) = 0 , we infer that β ( ψ ( x z n ) ) = o n ( 1 ) , which is absurd. Hence, we must have b > K .□

Lemma 4.4

One has c 0 > c .

Proof

Let u C 0 , then

u Ω Ω u ( y ) p u ( x ) p x y N α d y d x 1 2 p D ,

which indicates that

b p Ω Ω u ( y ) p u ( x ) p x y N α d y d x u H 0 s ( Ω ) 2 p .

Since u N Ω ,

b p u H 0 s ( Ω ) 2 = b p Ω Ω u ( y ) p u ( x ) p x y N α d y d x u H 0 s ( Ω ) 2 p ,

i.e.,

u H 0 s ( Ω ) 2 b p p 1 .

Therefore, we can infer that

J Ω ( u ) = J Ω ( u ) 1 2 p J Ω ( u ) u = 1 2 1 2 p u H 0 s ( Ω ) 2 1 2 1 2 p b p p 1 , for u C 0 .

Hence,

1 2 1 2 p b p p 1 c 0 .

Moreover, we easily see from the Lemma 4.3 that

c = J ( ψ ) = 1 2 1 2 p K p p 1 < 1 2 1 2 p b p p 1 c 0 ,

as desired.□

Let ζ : R N [ 0 , 1 ] be a function C such that ζ ( x ) = ξ ( x η ) , where η is the smallest positive number such that R N \ Ω B ( 0 , η ) and ξ : R + { 0 } [ 0 , 1 ] is a non-decreasing function C such that

ξ ( t ) = 0 , if t 1 , 1 , if t 2 .

For every z R N , let us consider a function ψ z , η H s ( R N ) defined by

ψ z , η ( x ) = ζ ( x ) ψ ( x z ) = ξ x η ψ ( x z ) ,

where ψ N is a ground-state solution of Problem (1.7) with radially symmetric and monotonically decreasing. Let t z , η ε > 0 be such that ϕ η , ε ( z ) = t z , η ε ψ z , η ( x ) N ε , .

Lemma 4.5

The following statements hold:

  1. lim η 0 ψ z , η H s ( R N ) 2 ψ H s ( R N ) 2 uniformly in z ;

  2. For all p [ 2 , 2 α , s * ] , lim η 0 R N R N ψ z , η ( y ) p ψ z , η ( x ) p x y N α d y d x = R N R N ψ ( y ) p ψ ( x ) p x y N α d y d x uniformly in z ;

  3. lim η , ε 0 J ε , ( ϕ η , ε ( z ) ) = c uniformly in z ;

  4. For each η fixed,

    lim z + , ε 0 J ε , ( ϕ η , ε ( z ) ) = c .

Proof

(1) Similar to [1, Lemma 4.1], it is easy to check that (1) holds, so we omit the details.

(2) In view of Lemma 2.2, we have

R N R N ψ z , η ( y ) p ψ z , η ( x ) p x y N α d y d x R N R N ψ ( y ) p ψ ( x ) p x y N α d y d x R N R N ζ ( x + z ) p 1 ζ ( y + z ) p ψ ( y ) p ψ ( x ) p x y N α d y d x + R N R N ζ ( y + z ) p 1 ψ ( y ) p ψ ( x ) p x y N α d y d x R N R N ζ ( x + z ) p 1 ψ ( y ) p ψ ( x ) p x y N α d y d x + R N R N ζ ( y + z ) p 1 ψ ( y ) p ψ ( x ) p x y N α d y d x C R N R N ζ ( y + z ) p 1 ψ ( y ) p ψ ( x ) p x y N α d y d x C R N ψ ( x ) 2 N p N + α d x N + α 2 N B ( 0 , 2 η ) ψ ( 0 ) 2 N p N + α d x N + α 2 N ,

which ensures that

R N R N ψ z , η ( y ) p ψ z , η ( x ) p x y N α d y d x R N R N ψ ( y ) p ψ ( x ) p x y N α d y d x , as η 0

uniformly in z .

(3) We claim that

lim η , ε 0 t z , η ε = 1 , uniformly in z .

Indeed, letting η n 0 , ε n 0 and { z n } R N , we show that { t z n , η n ε n } is bounded. Otherwise, suppose

t z n , η n ε n + .

We see from t z n , η n ε n ψ z n , η n N ε n , that

J ε n , ( t z n , η n ε n ψ z n , η n ) t z n , η n ε n ψ z n , η n = 0 ,

and hence,

ψ z n , η n H s ( R N ) 2 = ( t z n , η n ε n ) 2 p 2 R N R N ψ z n , η n ( y ) p ψ z n , η n ( x ) p x y N α d y d x + ε n ( t z n , η n ε n ) 2 2 α , s * 2 R N R N ψ z n , η n ( y ) 2 α , s * ψ z n , η n ( x ) 2 α , s * x y N α d y d x .

Furthermore,

ψ z n , η n H s ( R N ) 2 = ( t z n , η n ε n ) 2 p 2 R N R N ξ y + z n η n ψ ( y ) p ξ x + z n η n ψ ( x ) p x y N α d y d x + ε n ( t z n , η n ε n ) 2 2 α , s * 2 R N R N ξ y + z n η n ψ ( y ) 2 α , s * ξ x + z n η n ψ ( x ) 2 α , s * x y N α d y d x

and

ψ z n , η n H s ( R N ) 2 ( t z n , η n ε n ) 2 p 2 R N R N ψ ( y ) p ψ ( x ) p x y N α d y d x , as n + .

Then, lim n + ψ z n , η n H s ( R N ) = + , which contradicts with Item- ( 1 ) . Hence, the sequence { t z n , η n ε n } is bounded, then there exists t 0 0 such that

lim n + t z n , η n ε n = t 0 < + .

Note that t 0 0 . Otherwise, there holds ψ z n , η n 0 H s ( R N ) , which is absurd. Hence, t 0 ( 0 , + ) . Furthermore, it can be deduced from Items- ( 1 ) and ( 2 ) that

ψ H s ( R N ) 2 = t 0 2 p 2 R N R N ψ ( y ) p ψ ( x ) p x y N α d y d x .

Since ψ N , we observe that t 0 = 1 . Thus, the aforementioned arguments assure that

lim η , ε 0 ϕ η , ε H s ( R N ) 2 = lim η , ε 0 t z , η ε ψ z , η H s ( R N ) 2 = ψ H s ( R N ) 2 , uniformly in z .

Therefore,

lim η , ε 0 J ε , ( ϕ η , ε ( z ) ) = J ( ψ ) = c ,

uniformly in z .

(4) For each η fixed, let us consider the sequence { z n } R N with z n + . The same arguments used in the proof of Theorem 1.3 yield that

lim z n + , ε 0 ϕ η , ε ( z n ) H s ( R N ) 2 = ψ H s ( R N ) 2

and

lim z n + , ε 0 R N R N ψ z n , η ( y ) p ψ z n , η ( x ) p x y N α d y d x = R N R N ψ ( y ) p ψ ( x ) p x y N α d y d x .

Consequently,

lim z n + , ε 0 J ε , ( ϕ η , ε ( z n ) ) = J ( ψ ) = c .

The following lemma is a direct conclusion of Lemma 4.5(3).

Lemma 4.6

Fixed δ 0 , c 2 , there exist η 0 = η 0 ( δ ) and ε 2 = ε 2 ( δ ) such that

sup z R N J ε , ( ϕ η , ε ( z ) ) < 2 c δ ,

for all η < η 0 and ε < ε 2 .

In what follows, we fix δ 0 , c 2 with c + δ < c 0 . Furthermore, we assume Ω is fixed such that R N \ Ω B ( 0 , η ) with η < η 0 ( δ ) .

Lemma 4.7

There exist ε 3 > 0 and R 0 > η satisfying

  1. J ε , ( ϕ η , ε ( z ) ) c ε , , c 0 , ε + c ε , 2 , for z R 0 and ε ( 0 , ε 3 ) ;

  2. β ( ϕ η , ε ( z ) ) , z > 0 , for z = R 0 .

Proof

(1) By means of ϕ η , ε ( z ) N ε , Ω , we have

c ε , Ω = inf u N ε , Ω J ε , Ω ( u ) J ε , Ω ( ϕ η , ε ( z ) ) .

Combining the aforementioned inequality, Lemma 3.5, and Theorem 1.3, one has

c ε , = c ε , Ω < J ε , Ω ( ϕ η , ε ( z ) ) , ε ( 0 , ε ˜ ) .

Moreover, in view of Lemmas 4.1, 4.2, and 4.4, for ς small enough, there exists ε 3 > 0 such that

c < c 0 + c 2 ς < c 0 , ε + c ε , 2 < c 0 + c 2 + ς , ε ( 0 , ε 3 ) .

Furthermore, by virtue of Item-(4) of Lemma 4.5, there exists R 0 > η such that

J ε , Ω ( ϕ η , ε ( z ) ) < c 0 + c 2 ς , ε ( 0 , ε 3 ) , and z R 0 .

The previous estimates lead that (1) holds.

(2) The proof is similar to [1, Lemma 4.3], so we omit it.□

Now, we define the following sets:

G { ϕ η , ε ( z ) : z R 0 } , h C ( P N ε , Ω , P N ε , Ω ) : h ( u ) = u if J ε , Ω ( u ) < c 0 , ε + c ε , 2 ,

and

Γ { h ( G ) : h } .

Then, the next result can be found in [1, Lemma 5.1], whose arguments is standard, so it will be omitted.

Lemma 4.8

If A Γ , then A C 0 , ε for all ε ( 0 , ε 3 ) , where ε 3 is given in Lemma 4.7.

Finally, we establish the existence of positive solutions for Problem (1.9).

Proof of Theorem 1.4

Let us define the level

c ( ε ) inf A Γ sup u A J ε , Ω ( u ) .

By means of Lemma 4.8, for every A Γ , there is a u ˜ A C 0 , ε . Hence,

c 0 , ε = inf u C 0 , ε J ε , Ω ( u ) J ε , Ω ( u ˜ ) sup u A J ε , Ω ( u ) ,

for all A Γ . Furthermore,

c 0 , ε inf A Γ sup u A J ε , Ω ( u ) = c ( ε ) , ε ( 0 , ε 3 ) .

On the other hand, choose δ 0 , c 2 such that c + δ < c 0 . In view of Lemmas 4.24.4, there exists ε 4 > 0 such that

c + δ < c 0 , ε , ε ( 0 , ε 4 ) .

Noting that for every A Γ , c ( ε ) sup u A J ε , Ω ( u ) , there holds

c ( ε ) sup ϕ η , ε ( z ) G J ε , Ω ( h ( ϕ η , ε ( z ) ) ) , h .

Letting h = I R N , it is easy to see from Lemma 4.6 that

c ( ε ) sup ϕ η , ε ( z ) G J ε , Ω ( ϕ η , ε ( z ) ) sup z R 0 J ε , Ω ( ϕ η , ε ( z ) ) < 2 c δ , ε ( 0 , ε 2 ) .

Thus, setting ε 0 min { ε 2 , ε 3 , ε 4 } , we infer that

c + δ < c ( ε ) < 2 c δ , for ε ( 0 , ε 0 ) ,

then there exists μ > 0 such that

0 < μ < c ( ε ) c 0 , ε + c ε , 2 .

In what follows, we claim that for each μ 0 , c ( ε ) c 0 , ε + c ε , 2 , there exists u μ n ( P N ε , Ω ) J ε , Ω 1 ( c ( ε ) μ n , c ( ε ) + μ n ) satisfying

J ε , Ω ( u μ n ) * < μ n .

Assume, by contradiction, that there exists μ 0 > 0 with μ 0 < c ( ε ) c 0 , ε + c ε , 2 such that

J ε , Ω ( u ) * μ 0 2 , u ( P N ε , Ω ) J ε , Ω 1 ( c ( ε ) μ 0 , c ( ε ) + μ 0 ) .

Applying the deformation lemma [32, Lemma 5.15], there exists ρ : [ 0 , 1 ] × P N ε , Ω P N ε , Ω is continuous such that

ρ ( t , u ) = u , if u J ε , Ω 1 ( c ( ε ) μ 0 , c ( ε ) + μ 0 ) , t [ 0 , 1 ] ,

and

ρ 1 , J ε , Ω c ( ε ) + μ 0 2 J ε , Ω c ( ε ) μ 0 2 ,

where

J ε , Ω c { u P N ε , Ω : J ε , Ω ( u ) c , c R } .

According to the definition of c ( ε ) , we have

sup u A ^ J ε , Ω ( u ) < c ( ε ) + μ 0 2 , for some A ^ Γ ,

which states that A ^ J ε , Ω c ( ε ) + μ 0 2 . Then, the deformation lemma yields

ρ ( 1 , A ^ ) ρ 1 , J ε , Ω c ( ε ) + μ 0 2 J ε , Ω c ( ε ) μ 0 2 ,

i.e.,

J ε , Ω ( u ) c ( ε ) μ 0 2 , u ρ ( 1 , A ^ ) .

Therefore,

(4.5) sup u ρ ( 1 , A ^ ) J ε , Ω ( u ) c ( ε ) μ 0 2 .

In addition, since A ^ Γ , there exists h such that h ( G ) = A ^ . Define h ^ : P N ε , Ω P N ε , Ω given by h ^ ( u ) = ρ ( 1 , h ( u ) ) . Due to h C ( P N ε , Ω , P N ε , Ω ) , h ^ C ( P N ε , Ω , P N ε , Ω ) , which indicates that, if

J ε , Ω ( u ) < c 0 , ε + c ε , 2 < c ( ε ) μ 0 ,

then

h ^ ( u ) = ρ ( 1 , h ( u ) ) = ρ ( 1 , u ) = u ,

i.e., h ^ . Noting that μ 0 < c ( ε ) c 0 , ε + c ε , 2 , we have

h ^ ( G ) = ρ ( 1 , h ( G ) ) = ρ ( 1 , A ^ ) Γ .

Combining the definition of c ( ε ) , we can show that

c ( ε ) sup u ρ ( 1 , A ^ ) J ε , Ω ( u ) ,

which, together with (4.5), signifies that

c ( ε ) c ( ε ) μ 0 2 .

This leads to an obvious contradiction. Hence, there exists

u μ n ( P N ε , Ω ) J ε , Ω 1 ( c ( ε ) μ n , c ( ε ) + μ n ) ,

with μ n 0 such that

J ε , Ω ( u μ n ) * < μ n .

To sum up, there exists ε ¯ min { ε ˜ , ε 0 } such that for any ε ( 0 , ε ¯ ) , { u μ n } is a nonnegative ( P S ) c ( ε ) sequence for J ε , Ω restricted to N ε , Ω with c ( ε ) ( c + δ , 2 c δ ) . As an immediate consequence, Proposition 3.8 ensures that u μ n u ε 0 and J ε , Ω ( u ε ) = 0 . Consequently, there exists ε > 0 such that for all ε ( 0 , ε ) , Problem (1.9) has at least one nonnegative solution u H 0 s ( Ω ) for every α ( α N , N ) with J ε , Ω ( u ) [ c + δ , 2 c δ ] . By virtue of the maximum principle, u is strictly positive.□

Acknowledgements

The authors would like to thank the anonymous referee for several valuable suggestions and comments that helped to improve this article.

  1. Funding information: This project was supported by the National Natural Science Foundation of China (No.12371120), Natural Science Foundation of Chongqing, China (No. CSTB2023NSCQ-BHX0226), and the special subsidy from Chongqing human resources and Social Security Bureau.

  2. Author contributions: F. Ye wrote the manuscript; S. Yu contributed the analyses; C-L. Tang contributed to the conception of the study.

  3. Conflict of interest: The authors do not have conflict of interest.

  4. Data availability statement: Data sharing is not applicable to this article as no new data were created or analyzed in this study.

References

[1] C. O. Alves, G. M. Bisci, and C. T. Ledesma, Existence of solutions for a class of fractional elliptic problems on exterior domains, J. Differential Equations 268 (2020), 7183–7219. 10.1016/j.jde.2019.11.068Search in Google Scholar

[2] C. O. Alves and L. Freitas, Existence of a positive solution for a class of elliptic problems in exterior domains involving critical growth, Milan J. Math. 85 (2017), 309–330. 10.1007/s00032-017-0274-9Search in Google Scholar

[3] D. Applebaum, Lévy processes-from probability to finance and quantum groups, Not. Am. Math. Soc. 51 (2004), 1336–1347. Search in Google Scholar

[4] V. Benci and G. Cerami, Positive solutions of some nonlinear elliptic problems in exterior domains, Arch. Rational Mech. Anal. 99 (1987), 283–300. 10.1007/BF00282048Search in Google Scholar

[5] P. Chen and X. Liu, Positive solutions for a Choquard equation in exterior domain, Comm. Pure Appl. Anal. 20 (2021), 2237–2256. 10.3934/cpaa.2021065Search in Google Scholar

[6] J. N. Correia and G. M. Figueiredo, Existence of positive solution for a fractional elliptic equation in exterior domain, J. Differential Equations 268 (2020), 1946–1973. 10.1016/j.jde.2019.09.024Search in Google Scholar

[7] J. N. Correia and C. P. Oliveira, Existence of a positive solution for a class of fractional elliptic problems in exterior domains involving critical growth, J. Math. Anal. Appl. 506 (2022), 125543. 10.1016/j.jmaa.2021.125543Search in Google Scholar

[8] J. N. Correia and C. P. Oliveira, Positive solution for a class of Choquard equations with Hardy-Littlewood-Sobolev critical exponent in exterior domains, Complex Var. Elliptic Equ. 68 (2023), 1485–1520, https://doi.org/10.1080/17476933.2022.2056888. https://doi.org/10.1080/17476933.2022.2056888Search in Google Scholar

[9] P. d’Avenia, G. Siciliano, and M. Squassina, On fractional Choquard equations, Math. Models Methods Appl. Sci. 25 (2015), 1447–1476. 10.1142/S0218202515500384Search in Google Scholar

[10] A. Elgart and B. Schlein, Mean field dynamics of boson stars, Commun. Pure Appl. Math. 60 (2007), 500–545. 10.1002/cpa.20134Search in Google Scholar

[11] P. Felmer, A. Quaas, and J. Tan, Positive solutions of nonlinear Schrödinger equation with the fractional Laplacian, Proc. Roy. Soc. Edinburgh Sect. A 142 (2012), 1237–1262. 10.1017/S0308210511000746Search in Google Scholar

[12] R. L. Frank and E. Lenzmann, On ground-states for the L2-critical boson star equation. 2009. arXiv:0910.2721. Search in Google Scholar

[13] R. L. Frank, E. Lenzmann, and L. Silvestre, Uniqueness of radial solutions for the fractional Laplacian, Appl. Math. 69 (2016), 1671–1725. 10.1002/cpa.21591Search in Google Scholar

[14] W. Guan, V. D. Rădulescu, and D. Wang, Bound states of fractional Choquard equations with Hardy-Littlewood-Sobolev critical exponent, J. Differential Equations 355 (2023), 219–247. 10.1016/j.jde.2023.01.023Search in Google Scholar

[15] X. He and V. D. Rădulescu, Small linear perturbations of fractional Choquard equations with critical exponent, J. Differential Equations 282 (2021), 481–540. 10.1016/j.jde.2021.02.017Search in Google Scholar

[16] X. He, V. D. Rădulescu, and W. Zou, Normalized ground-states for the critical fractional Choquard equation with a local perturbation, J. Geom. Anal. 32 (2022), 51. 10.1007/s12220-022-00980-6Search in Google Scholar

[17] C. T. Ledesma and O. H. Miyagaki, Positive solutions for a class of fractional Choquard equation in exterior domain, Milan J. Math. 90 (2022), 519–554. 10.1007/s00032-022-00361-2Search in Google Scholar

[18] C. T. Ledesma, Existence of positive solutions for a class of fractional Choquard equation in exterior domain, Discrete Contin. Dyn. Syst. 42 (2022), 3301–3328. 10.3934/dcds.2022016Search in Google Scholar

[19] E. H. Lieb, Existence and uniqueness of the minimizing solution of Choquard’s nonlinear equation, Stud. Appl. Math. 57 (1976/77), 93–105. 10.1002/sapm197757293Search in Google Scholar

[20] E. H. Lieb and M. P. Loss, Analysis, American Mathematical Society, Providence RI, 2001. Search in Google Scholar

[21] G. Li and C. Tang, Existence of positive ground-state solutions for Choquard equation with variable exponent growth, Discrete Contin. Dyn. Syst. Ser. S, 12 (2019), 2035–2050. 10.3934/dcdss.2019131Search in Google Scholar

[22] Y. Li, G. Li, and C. Tang, Multiplicity and concentration of positive solutions for critical Choquard equations with concave perturbation, J. Math. Anal. Appl. 524 (2023), 24 pp. 10.1016/j.jmaa.2023.127112Search in Google Scholar

[23] V. Moroz and J. Van Schaftingen, Groundstates of nonlinear Choquard equations: existence, qualitative properties and decay asymptotics, J. Funct. Anal. 265 (2013), 153–184. 10.1016/j.jfa.2013.04.007Search in Google Scholar

[24] E. Nezza, G. Palatucci, and E. Valdinoci, Hitchhiker’s guide to the fractional Sobolev spaces, Bull. Sci. Math. 136 (2012), 521–573. 10.1016/j.bulsci.2011.12.004Search in Google Scholar

[25] Y. Park, Fractional Polya-Szegö inequality, J. Chungcheong Math. Soc. 42 (2011), 267–271. Search in Google Scholar

[26] S. Pekar, Untersuchung über die Elektronentheorie der Kristalle, Akademie Verlag, Berlin, 1954. 10.1515/9783112649305Search in Google Scholar

[27] D. Qin and X. Tang, On the planar Choquard equation with indefinite potential and critical exponential growth, J. Differential Equations 285 (2021), 40–98. 10.1016/j.jde.2021.03.011Search in Google Scholar

[28] D. Ruiz and J. Van Schaftingen, Odd symmetry of least energy nodal solutions for the Choquard equation, J. Differential Equations 264 (2018), 1231–1262. 10.1016/j.jde.2017.09.034Search in Google Scholar

[29] J. Seok, Limit profiles and uniqueness of ground-states to the nonlinear Choquard equations, Adv. Nonlinear Anal. 8 (2019), 1083–1098. 10.1515/anona-2017-0182Search in Google Scholar

[30] R. Servadei and E. Raffaella, Variational methods for non-local operators of elliptic type, Discrete Contin. Dyn. Syst. 33 (2013), 2105–2137. 10.3934/dcds.2013.33.2105Search in Google Scholar

[31] X. Wang and Z. Yang, Symmetry and monotonicity of positive solutions for a Choquard equation with the fractional Laplacian, Complex Var. Elliptic Equ. 67 (2022), 1211–1228. 10.1080/17476933.2020.1863385Search in Google Scholar

[32] M. Willem, Minimax Theorems, Birkhäuser, Boston, 1996. 10.1007/978-1-4612-4146-1Search in Google Scholar

[33] S. Yu, C. Tang, and Z. Zhang, Normalized solutions of mass subcritical fractional Schrödinger equations in exterior domains, J. Geom. Anal. 33 (2023), 30 pp. 10.1007/s12220-023-01215-ySearch in Google Scholar

[34] X. Yang, Existence of positive solution for the Choquard equation in exterior domain, Complex Var. Elliptic Equ. 67 (2022), 2043–2059. 10.1080/17476933.2021.1913133Search in Google Scholar

Received: 2023-06-08
Revised: 2023-08-03
Accepted: 2024-05-06
Published Online: 2024-06-08

© 2024 the author(s), published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Regular Articles
  2. Existence of normalized peak solutions for a coupled nonlinear Schrödinger system
  3. Orbital stability of the trains of peaked solitary waves for the modified Camassa-Holm-Novikov equation
  4. Global boundedness in a two-dimensional chemotaxis system with nonlinear diffusion and singular sensitivity
  5. Existence and uniqueness of solution for a singular elliptic differential equation
  6. The bounded variation capacity and Sobolev-type inequalities on Dirichlet spaces
  7. Vanishing and blow-up solutions to a class of nonlinear complex differential equations near the singular point
  8. Existence, uniqueness, localization and minimization property of positive solutions for non-local problems involving discontinuous Kirchhoff functions
  9. Concentration phenomena for a fractional relativistic Schrödinger equation with critical growth
  10. Beyond the classical strong maximum principle: Sign-changing forcing term and flat solutions
  11. Monotonicity of solutions for parabolic equations involving nonlocal Monge-Ampère operator
  12. On a nonlinear Robin problem with an absorption term on the boundary and L1 data
  13. Multiple solutions for the quasilinear Choquard equation with Berestycki-Lions-type nonlinearities
  14. Gradient estimates for a class of higher-order elliptic equations of p-growth over a nonsmooth domain
  15. Global existence and decay estimates of the classical solution to the compressible Navier-Stokes-Smoluchowski equations in ℝ3
  16. Global existence and finite-time blowup for a mixed pseudo-parabolic r(x)-Laplacian equation
  17. Multiple positive solutions for a class of concave-convex Schrödinger-Poisson-Slater equations with critical exponent
  18. A minimization problem with free boundary for p-Laplacian weakly coupled system
  19. Multiplicity of semiclassical solutions for a class of nonlinear Hamiltonian elliptic system
  20. k-convex solutions for multiparameter Dirichlet systems with k-Hessian operator and Lane-Emden type nonlinearities
  21. Infinitely many solutions for Hamiltonian system with critical growth
  22. On optimal control in a nonlinear interface problem described by hemivariational inequalities
  23. Variational–hemivariational system for contaminant convection–reaction–diffusion model of recovered fracturing fluid
  24. Potential and monotone homeomorphisms in Banach spaces
  25. Critical fractional Schrödinger-Poisson systems with lower perturbations: the existence and concentration behavior of ground state solutions
  26. Supercritical Hénon-type equation with a forcing term
  27. Concentration of blow-up solutions for the Gross-Pitaveskii equation
  28. Mountain-pass-type solutions for Schrödinger equations in R2 with unbounded or vanishing potentials and critical exponential growth nonlinearities
  29. Regularity for critical fractional Choquard equation with singular potential and its applications
  30. Double phase anisotropic variational problems involving critical growth
  31. Normalized solutions of NLS equations with mixed nonlocal nonlinearities
  32. Nontrivial solutions for resonance quasilinear elliptic systems
  33. Standing waves for Choquard equation with noncritical rotation
  34. Low regularity conservation laws for Fokas-Lenells equation and Camassa-Holm equation
  35. Modified quasilinear equations with strongly singular and critical exponential nonlinearity
  36. Limit profiles and the existence of bound-states in exterior domains for fractional Choquard equations with critical exponent
  37. Nests of limit cycles in quadratic systems
  38. Regularity of minimizers for double phase functionals of borderline case with variable exponents
  39. Boundedness, existence and uniqueness results for coupled gradient dependent elliptic systems with nonlinear boundary condition
  40. Ground state solutions for magnetic Schrödinger equations with polynomial growth
  41. Nonoccurrence of Lavrentiev gap for a class of functionals with nonstandard growth
  42. Dynamics for wave equations connected in parallel with nonlinear localized damping
  43. Boussinesq's equation for water waves: Asymptotics in Sector I
  44. Optimal global second-order regularity and improved integrability for parabolic equations with variable growth
  45. Normalized solutions of Schrödinger equations involving Moser-Trudinger critical growth
  46. Normalized solutions for the double-phase problem with nonlocal reaction
  47. Normalized solutions for Sobolev critical fractional Schrödinger equation
  48. Well-posedness of Cauchy problem of fractional drift diffusion system in non-critical spaces with power-law nonlinearity
  49. Improved results on planar Klein-Gordon-Maxwell system with critical exponential growth
  50. Existence and multiplicity of solutions for a new p(x)-Kirchhoff equation
  51. Quasiconvex bulk and surface energies: C1,α regularity
  52. Time decay estimates of solutions to a two-phase flow model in the whole space
  53. Bounds for the sum of the first k-eigenvalues of Dirichlet problem with logarithmic order of Klein-Gordon operators
  54. The properties of a new fractional g-Laplacian Monge-Ampère operator and its applications
  55. A fractional profile decomposition and its application to Kirchhoff-type fractional problems with prescribed mass
  56. Cauchy problem for a non-Newtonian filtration equation with slowly decaying volumetric moisture content
  57. Multiplicity of normalized semi-classical states for a class of nonlinear Choquard equations
  58. The second Yamabe invariant with boundary
  59. Peaked solitary waves and shock waves of the Degasperis-Procesi-Kadomtsev-Petviashvili equation
  60. Normalized solutions for the Choquard equations with critical nonlinearities
  61. Boundedness and long-time behavior in a parabolic-elliptic system arising from biological transport networks
  62. Initial boundary value problem and exponential stability for the planar magnetohydrodynamics equations with temperature-dependent viscosity
  63. Nonexistence of mean curvature flow solitons with polynomial volume growth immersed in certain semi-Riemannian warped products
  64. Analysis of a vector-borne disease model with vector-bias mechanism in advective heterogeneous environment
  65. Normalized solutions for the Kirchhoff equation with combined nonlinearities in ℝ4
  66. Nonexistence of the compressible Euler equations with space-dependent damping in high dimensions
  67. Multiplicity of solutions for a nonhomogeneous quasilinear elliptic equation with concave-convex nonlinearities
  68. On semilinear inequalities involving the Dunkl Laplacian and an inverse-square potential outside a ball
  69. Besov regularity for the elliptic p-harmonic equations in the non-quadratic case
  70. Uniqueness and nondegeneracy of ground states for Δ u + u = ( I α u 2 ) u in R 3 when α is close to 2
  71. Existence of solutions for a class of quasilinear Schrödinger equations with Choquard-type nonlinearity
  72. Weighted Hardy-Adams inequality on unit ball of any even dimension
  73. Existence and regularity for a p-Laplacian problem in ℝN with singular, convective, and critical reaction
  74. Temporal periodic solutions of non-isentropic compressible Euler equations with geometric effects
  75. Nodal solutions for a zero-mass Chern-Simons-Schrödinger equation
  76. The modified quasi-boundary-value method for an ill-posed generalized elliptic problem
  77. Nonlocal heat equations with generalized fractional Laplacian
  78. Choquard equations with recurrent potentials
  79. Local and global well-posedness of the Maxwell-Bloch system of equations with inhomogeneous broadening
  80. The Riemann problem for two-layer shallow water equations with bottom topography
  81. Global stability and asymptotic profiles of a partially degenerate reaction diffusion Cholera model with asymptomatic individuals
  82. On Kirchhoff-Schrödinger-Poisson-type systems with singular and critical nonlinearity
  83. Energy-variational solutions for viscoelastic fluid models
  84. Stability on 3D Boussinesq system with mixed partial dissipation
  85. Existence and multiplicity results for non-autonomous second-order Hamiltonian systems
  86. Erratum
  87. Erratum to “Normalized solutions for the Kirchhoff equation with combined nonlinearities in ℝ4
Downloaded on 7.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/anona-2024-0020/html
Scroll to top button