Next Article in Journal
Evaluating the Fractal Pattern of the Von Koch Island Using Richardson’s Method
Previous Article in Journal
A Mixed Finite Volume Element Method for Nonlinear Time Fractional Fourth-Order Reaction–Diffusion Models
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Normalized Ground States for the Sobolev Critical Fractional Kirchhoff Equation with at Least Mass Critical Growth

School of Mathematics, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
*
Author to whom correspondence should be addressed.
Fractal Fract. 2025, 9(8), 482; https://doi.org/10.3390/fractalfract9080482
Submission received: 10 June 2025 / Revised: 6 July 2025 / Accepted: 8 July 2025 / Published: 24 July 2025

Abstract

In this paper, we delve into the following nonlinear fractional Kirchhoff-type problem ( a + b | | ( Δ ) s 2 u | | 2 2 ) ( Δ ) s u + λ u = g ( u ) + | u | 2 s * 2 u in R 3 with prescribed mass R 3 | u | 2 d x = ρ > 0 , where s ( 3 4 , 1 ) , λ R , 2 s * = 6 3 2 s . Under some general growth assumptions imposed on g, we employ minimization of the energy functional on the linear combination of Nehari and Poho z ˘ aev constraints intersected with the closed ball in the L 2 ( R 3 ) of radius ρ to prove the existence of normalized ground state solutions to the equation. Moreover, we provide a detailed description for the asymptotic behavior of the ground state energy map.

1. Introduction and Main Results

In this paper, we consider the existence of normalized solutions for the following fractional Kirchhoff equation
( a + b | | ( Δ ) s 2 u | | 2 2 ) ( Δ ) s u + λ u = g ( u ) + | u | 2 s * 2 u in R 3
with prescribed mass
R 3 | u | 2 d x = ρ > 0 ,
where s ( 3 4 , 1 ) , λ R , 2 ¯ s = 2 + 8 s 3 and 2 s * = 6 3 2 s . The operator ( Δ ) s is the fractional Laplacian defined by
( Δ ) s u ( x ) : = C s P . V . R 3 u ( x ) u ( y ) | x y | 3 + 2 s d y , x R 3 ,
and
| | ( Δ ) s 2 u | | 2 2 = R 6 | u ( x ) u ( y ) | 2 | x y | 3 + 2 s d x d y ,
where P . V . is the Cauchy principal value on the integral and C s is some positive normalization constant [1].
Equation (1) has its origin in the theory of nonlinear vibration. In the case of s = 1 and h ( x , u ) = g ( u ) + | u | 2 s * 2 u , problem (1) reduces to the following Kirchhoff equation
( a + b | | ( Δ ) s 2 u | | 2 2 ) ( Δ ) s u + λ u = g ( u ) + | u | 2 s * 2 u in R 3 .
which is related to the stationary case of the following equation:
u t t ( a + b Ω | u | 2 d x ) Δ u = g ( u ) + | u | 2 s * 2 u , in Ω × ( 0 , + ) , u ( x , t ) = 0 , on Ω × [ 0 , + ) .
where a is the initial tension, b is related to the intrinsic properties of the string, Ω R 3 is a smooth domain, u stands for displacement, and f C ( Ω × R , R ) is the external force. Equation (3) comes from the following model for the modified d’Alembert wave equation:
ρ 2 u t 2 ρ 0 h + E 2 L 0 L | u x | 2 d x 2 u x 2 = f ( x , u ) ,
for free vibrations of elastic strings, which was proposed by G.Kirchhoff [2] in 1883 and considered theoretically or experimentally by several physicists [3,4]. Arosio and Panizzi [5] Introduced the research background and significance of this issue, and listed existing research results on this issue. Moreover, they considered the Cauchy problem for the quasilinear hyperbolic integro-differential equation.
In 2013, Fiscella and Valdinoci [6] proved the existence of non-negative solutions for a Kirchhoff type problem driven by a nonlocal integrodifferential operator:
M ( | | u | | Z 2 ) L K u = λ f ( x , u ) + | u | 2 s * 2 u in Ω , u = 0 in R N Ω ,
where L K is a nonlocal operator, defined as follows:
L K = 1 2 R N ( u ( x + y ) + u ( x y ) 2 u ( x ) ) K ( y ) d y ,
for all x R N and K : R N { 0 } ( 0 , + ) is a measurable function with some properties. When K ( x ) = | x | N + 2 s , the problem (5) becomes
M R 2 N | u ( x ) u ( y ) | 2 | x y | N + 2 s d x d y L K u = λ f ( x , u ) + | u | 2 s * 2 u in Ω , u = 0 in R N Ω ,
which is the fractional Kirchhoff-type problem. In 2014, Autuori, Fiscella, and Pucci [7] dealt with the existence and asymptotic behavior of non-negative solutions of (7) when M is zero at zero; that is, the problem is degenerate. In view of variational methods, Pucci and Saldi [8] studied the existence and multiplicity of nontrivial non-negative weak solutions of (7). Figueiredo, Ikoma and J u ´ nior [9] studied the existence and concentration results of a class of Kirchhoff-type equations with general nonlinear terms. Pucci and R a ˇ dulescu [10] mainly discussed the progress of nonlinear Kirchhoff problems and introduced the research background and application fields of the Kirchhoff problem, especially for the study of closed Riemann manifolds. In addition, one can refer to [11,12] to know more about the mathematical and physical background of (3).
In recent years, stationary Kirchhoff problems have been widely studied. However, there are few papers on the fractional Kirchhoff-type problem. Under the condition g : R R is an odd function satisfying Berestycki–Lions-type assumptions, with the aid of minimax arguments, Ambrosio and Isernia [13] established a multiplicity result for
p + q ( 1 + s ) R N | ( Δ ) s 2 u | 2 d x ( Δ ) s u = g ( u ) in R N ,
provided that q is sufficiently small. Without the ( A R ) -condition, Mingqi Xiang, Binlin Zhang and Miaomiao Yang [14] used variational methods combined with a cut-off function technique to investigate the existence of radial solutions for a fractional Kirchhoff-type problem:
a + b R N | ( Δ ) s 2 u | 2 d x + R N | u | 2 d x θ 1 ( ( Δ ) s u + u ) = f ( u ) in R N .
Jin and Liu [15] considered the following fractional Kirchhoff equation
a + b R N | ( Δ ) s 2 u | 2 d x ( Δ ) s u + u = f ( u ) in R N .
and the existence of solutions to (10) without the A R -condition is proven when the parameter b is small. Moreover, they studied the asymptotic behavior of solutions as b 0 . Li [16] considered the fractional Kirchhoff equation with critical growth:
a + b R N | ( Δ ) s 2 u | 2 d x ( Δ ) s u + u = f ( x , u ) + | u | 2 s * 2 u in R 3 .
and proved the existence of ground state solutions. Gu, Ming, and Yang [17] made use of variational methods and compactness analysis to investigate the existence of ground state solutions for a non-autonomous fractional-order Kirchhoff equation.
Motivated by the works above, we are concerned with (1). Let G ( u ) : = 0 u g ( s ) d s , H ( u ) : = g ( u ) u 2 G ( u ) and h = H . Now, we make the following conditions:
( A 0 ) g , h C ( R ) and there exists a constant C > 0 such that
| h ( t ) | C | t | + | t | 2 s * 1 for t R .
( A 1 ) η : = lim sup | t | 0 G ( t ) | t | 2 ¯ s < .
( A 2 ) lim | t | G ( t ) | t | 2 ¯ s = .
( A 3 ) lim | t | G ( t ) | t | 2 s * = 0 .
( A 4 ) 2 + 8 s 3 H ( t ) h ( t ) t for t R .
( A 5 ) 2 + 8 s 3 G ( t ) g ( t ) t 2 s * G ( t ) for any t R .
( A 6 ) H ( ζ 0 ) > 0 for some ζ 0 0 .
Before we present our main results, for any s ( 0 , 1 ) , we recall the definition of the fractional Sobolev space
H s ( R 3 ) = u L 2 ( R 3 ) : | | ( Δ ) s 2 u | | 2 2 < +
endowed with the norm
| | u | | H s ( R 3 ) 2 = R 3 ( | ( Δ ) s 2 u | 2 + u 2 ) d x .
Now, we set
S : = { u H s ( R 3 ) | | | u | | 2 2 = ρ } . ,
and
D : = { u H s ( R 3 ) : | | u | | 2 2 ρ } .
The solutions to (1) are critical points of the energy functional
E ( u ) = a 2 | | ( Δ ) s 2 u | | 2 2 + b 4 | | ( Δ ) s 2 u | | 2 4 R N G ( u ) d x 1 2 s * | | u | | 2 s * 2 s * .
where G ( u ) = 0 u g ( s ) d s . From [18], we obtain that the following Poho z ˘ aev identity
3 2 s 2 a + b | | ( Δ ) s 2 u | | 2 2 | | ( Δ ) s 2 u | | 2 2 = 3 R 3 G ( u ) d x + 3 2 s * R 3 | u | 2 s * d x .
Furthermore, let H ( u ) = g ( u ) u 2 G ( u ) and h = H , we define the Poho z ˘ aev manifold
P = { u H s ( R 3 ) { 0 } : P ( u ) = 0 } ,
where
P ( u ) = s a + b | | ( Δ ) s 2 u | | 2 2 | | ( Δ ) s 2 u | | 2 2 3 2 R 3 H ( u ) d x s | | u | | 2 s * 2 s * .
The constraint manifold P constitutes the complete solution space of nontrivial solutions to (1) and exhibits parameter invariance with respect to λ . All nontrivial solutions to (1) belongs to S P D P . Therefore, any solution u that simultaneously satisfies (1) and achieves the energy minimization condition E ( u ) = inf S P E can be characterized as a normalized ground state solution to the system (1).
Remark 1.
Bieganowski and Mederski [19] showed the existence of ground state solutions to the problem Δ u + λ u = g ( u ) with the L 2 -norm constrained manifold. Moreover, Li, R a ˇ dulescu and Zhang [20] extended the results in [19] from the Sobolev subcritical setting to the Sobolev critical framework. Taking inspiration from the above paper, can we extend some of the results in the article to the fractional Kirchhoff Equation (1)? However, the emergence of a fractional operator greatly limits the range of dimensions studied and increases the difficulty of calculation. Furthermore, nonlocal terms greatly increase the difficulty of energy estimation. In our opinion, the range estimation of the minimization of energy functional cannot be directly applied to the analysis in the above article. Of course, we believe that with the continuous deepening of research, better methods will be found in the future to solve this problem.
Theorem 1.
Let ( A 0 )–( A 6 ) hold and
2 s * η C s , 2 ¯ s , p 2 ¯ s ρ 4 s 3 3 < b ,
a 4 s 3 S 3 < b 2 s b 3 2 s
hold. Then, there is u D P , such that
E ( u ) = inf D P E > 0 ,
and if, in addition, g is odd, then inf D P E = inf S P E and u D P is a positive and radially symmetric normalized ground state solution to (1).
For any ρ > 0 , we define
D ρ : = { u H s ( R 3 ) : | | u | | 2 2 ρ }
and
S ρ : = { u H s ( R 3 ) : | | u | | 2 2 = ρ } .
Then, we obtain the asymptotic behavior of the ground state energy map ρ inf S ρ P E as follows:
Theorem 2.
Under the assumptions of Theorem 1, then
  • (a) the ground state energy map ρ inf S ρ P E is strictly decreasing;
  • (b) if η 0 ,
    •   (1) the map ρ inf S ρ P E is continuous;
    •   (2) inf S ρ P E 0 + as ρ .

2. Preliminaries

First, we establish some estimates. Under assumptions ( A 0 ), ( A 1 ), ( A 3 ), and ( A 5 ), for each ε > 0 , a constant C ε > 0 can be found, which satisfying
H ( t ) ( 2 s * 2 ) G ( u ) ( 2 s * 2 ) ε | t | 2 s * + ( ε + η ) | t | 2 ¯ s + C ε | t | q
for any t R , where 2 q 2 s * . Alternatively, in view of ( A 5 ), one obtains that
G ( t ) , H ( t ) 0 , t R .
According to [21], we know
S = inf D s , 2 ( R 3 ) { 0 } | | ( Δ ) s 2 u | | 2 2 | | u | | 2 s * 2
where S is the best Sobolev constant and D s , 2 ( R 3 ) : = { u L 2 s * ( R 3 ) : | | ( Δ ) s 2 u | | 2 2 < } .
It is known that [22] S is attained by the functions u ~ ( x ) = κ ( μ 2 + | x x 0 | 2 ) 3 2 s 2 , x R 3 , where κ R { 0 } , κ > 0 and x 0 R 3 are fixed constants. By [23], it is clear that u * is a solution to the problem
( Δ ) s u = | u | 2 s * 2 u in R 3
and | | u * | | 2 s * 2 s * = S 3 2 s . For any ε > 0 , set U ε ( x ) = ε 3 2 s 2 u * x ε . Fix φ C 0 ( R 3 , [ 0 , 1 ] ) , such that ssup φ B 2 r ( 0 ) and φ ( x ) = 1 for x B r ( 0 ) . Let v ε ( x ) = φ ( x ) U ε ( x ) . From [24] and [25], it follows that
| | ( Δ ) s 2 v ε ( x ) | | 2 2 S 3 2 s + O ( ε 3 2 s ) ,
| | v ε ( x ) | | 2 s * 2 s * = S 3 2 s + O ( ε 3 ) ,
| | v ε ( x ) | | p p = O ( ε ( 3 2 s ) p 2 ) , p < 3 3 2 s , O ( ε 3 ( 2 p ) + 2 s p 2 | log ε | ) , p = 3 3 2 s , O ( ε 3 ( 2 p ) + 2 s p 2 ) , 3 3 2 s < p < 6 3 2 s .
Proposition 1
(Fractional Gagliardo–Nirenberg inequality [26]). Let s ( 3 4 , 1 ) , q ( 2 , 2 s * ) , then, for all u H s ( R 3 ) , there exists a constant B q , s > 0 depending on p , s such that
| | u | | q q B s , q q | | ( Δ ) s 2 u | | 2 N ( q 2 ) 2 s | | u | | 2 q N ( q 2 ) 2 s .
Proposition 2
([27]). Let s ( 0 , 1 ) . For any u H s ( R 3 ) , the following equation holds
| | ( Δ ) s 2 u | | 2 2 | | ( Δ ) s 2 u * | | 2 2 .

3. Proof of Theorem 1

Lemma 1.
Assume that ( A 0 ), ( A 1 ), ( A 3 ), ( A 5 ), ( A 6 ) and (15) hold. For any u D P , there exists δ > 0 , such that | | ( Δ ) s 2 u | | 2 2 δ .
Proof. 
Taking any p ¯ s < q < p s * . For any u D P D , from (14) and (18), we can obtain that
s a + b | | ( Δ ) s 2 u | | 2 2 | | ( Δ ) s 2 u | | 2 2 = 3 2 R 3 [ g ( u ) u 2 G ( u ) ] d x + s | | u | | 2 s * 2 s * = 3 2 R 3 H ( u ) d x + s | | u | | 2 s * 2 s * 3 2 ( 2 s * 2 ) R 3 ε | u | 2 s * + ( ε + η ) | u | 2 ¯ s + C ε | u | q d x + s | | u | | 2 s * 2 s * = s ( 2 s * ε + 1 ) | | u | | 2 s * 2 s * + 2 s * s ( ε + η ) | | u | | 2 ¯ s 2 ¯ s + s 2 s * C ε | | u | | q q .
In view of (20) and (25), we have
a | | ( Δ ) s 2 u | | 2 2 + b | | ( Δ ) s 2 u | | 2 4 ( 2 s * ε + 1 ) | | u | | 2 s * 2 s * + 2 s * ( ε + η ) | | u | | 2 ¯ s 2 ¯ s + 2 s * C ε | | u | | q q ( 2 s * ε + 1 ) S 2 s * 2 | | ( Δ ) s 2 u | | 2 2 s * + 2 s * ( ε + η ) B s , 2 ¯ s 2 ¯ s ρ 4 s 3 3 | | ( Δ ) s 2 u | | 2 4 + 2 s * C ε B s , q q ρ q ( 1 γ q , s ) | | ( Δ ) s 2 u | | 2 q γ q , s ,
that is,
a | | ( Δ ) s 2 u | | 2 2 + ( b 2 s * ( ε + η ) B s , 2 ¯ s 2 ¯ s ρ 4 s 3 3 ) | | ( Δ ) s 2 u | | 2 4 ( 2 s * ε + 1 ) S 2 s * 2 | | ( Δ ) s 2 u | | 2 2 s * + 2 s * C ε B s , q q ρ q ( 1 γ q , s ) | | ( Δ ) s 2 u | | 2 q γ q , s ,
As q γ q , s > 4 and 2 s * η B s , 2 ¯ s 2 ¯ s ρ 4 s 3 3 < b , taking ε < b 2 s * η B s , 2 ¯ s 2 ¯ s ρ 4 s 3 N p s * η B s , 2 ¯ s 2 ¯ s ρ 4 s 3 3 , we complete the proof. □
Let u H s ( R 3 ) { 0 } satisfy
4 η B s , p ¯ s p ¯ s R 3 u 2 d x 4 s 3 3 < b
For each λ > 0 , we define the scaling functional ϕ ( λ ) : = E ( λ 3 2 u ( λ · ) ) . The following Lemma holds:
Lemma 2.
Suppose that u H s ( R N ) { 0 } satisfies (27) and ( A 0 ), ( A 1 ), ( A 3 )–( A 6 ) hold. Then, there is an interval λ 0 ( 0 , + ) , such that
  • (i) ϕ ( λ 0 ) ϕ ( λ ) for any λ ( 0 , + ) and the strict inequality holds for λ ( 0 , + ) { λ 0 } .
  • (ii)
    P ( λ 3 2 u ( λ · ) ) > 0 i f λ ( 0 , λ 0 ) , = 0 i f λ = λ 0 , < 0 i f λ ( λ 0 , + ) .
Proof. 
Let u H s ( R N ) { 0 } satisfy (27). From the ( A 1 ) and (25), one obtain
ϕ ( λ ) = E ( λ 3 2 u ( λ x ) ) = a λ 2 s 2 | | ( Δ ) s 2 u | | 2 2 + b λ 4 s 4 | | ( Δ ) s 2 u | | 2 4 R 3 G ( λ 3 2 u ( x ) ) λ 3 d x λ 2 s * s 2 s * | | u ( x ) | | 2 s * 2 s * .
Obviously, ϕ ( λ ) 0 as λ 0 + . Moreover, ϕ ( λ ) as λ because of (19). Set R : = | | u | | 2 2 = | | λ 3 2 u ( λ · ) | | 2 2 > 0 . From (18), (20), (25), it is not difficult to find that for any ϵ > 0 , there exists C ϵ > 0 , such that
R N G ( u ) d x ( ε + η ) | | u | | p ¯ s p ¯ s + C ε | | u | | 2 s * 2 s * ( ε + η ) B s , 2 ¯ s 2 ¯ s R 4 s 3 3 | | ( Δ ) s 2 u | | 2 4 + C ε S 2 s * 2 | | ( Δ ) s 2 u | | 2 2 s * .
Therefore,
ϕ ( λ ) λ 4 s = a 2 λ 2 s | | ( Δ ) s 2 u | | 2 2 + b 4 | | ( Δ ) s 2 u | | 2 4 1 λ 4 s R 3 G ( λ 3 2 u ( λ x ) ) d x λ ( 2 s * 4 ) s 2 s * | | u ( x ) | | 2 s * 2 s * a 2 λ 2 s | | ( Δ ) s 2 u | | 2 2 + b 4 ( ε + η ) B s , 2 ¯ s 2 ¯ s R 4 s 3 3 | | ( Δ ) s 2 u | | 2 4 λ ( 2 s * 4 ) s C ε S 2 s * 2 | | ( Δ ) s 2 u | | 2 2 s * λ ( 2 s * 4 ) s 2 s * | | u ( x ) | | 2 s * 2 s * .
By (27), one can obtain ϕ ( λ ) > 0 for small enough ε > 0 and λ > 0 . Hence, there exists some λ 0 > 0 , such that ϕ max = ϕ ( λ 0 ) . In particular, ϕ ( λ 0 ) = 0 . Since
ϕ ( λ ) λ 4 s 1 = a s λ 2 s | | ( Δ ) s 2 u | | 2 2 + b s | | ( Δ ) s 2 u | | 2 4 3 2 λ 3 4 s R 3 H ( λ 3 2 u ) d x s λ 2 s * s 4 s | | u | | 2 s * 2 s *
and combined with the (14), we can easily see that P ( λ 0 N p u ( λ 0 x ) ) = λ 0 ϕ ( λ 0 ) = 0 . Set
m 1 ( λ ) : = a s λ 2 s | | ( Δ ) s 2 u | | 2 2 + b s | | ( Δ ) s 2 u | | 2 4 , m 2 ( λ ) : = 3 2 λ 3 4 s R 3 H ( λ N p u ) d x + s λ 2 s * s 4 s | | u | | 2 s * 2 s * .
It is easy to know that m 1 ( λ ) is strictly decreasing on λ ( 0 , + ) . By virtue of ( A 4 ), we have the function m 2 ( λ ) , which is strictly increasing on λ ( 0 , + ) and tends to + as n . This means that ϕ ( λ ) > 0 if λ ( 0 , λ 0 ) and ϕ ( λ ) < 0 if λ ( λ 0 , + ) . □
Lemma 3.
Assume that ( A 0 ), ( A 1 ), ( A 3 )–( A 6 ), and (15) hold. Then, E is coercive on D P .
Proof. 
For any u D P , according to Lemma 1, (1) and ( A 5 ), one has
E ( u ) = E ( u ) 1 4 s P ( u ) = a 4 | | ( Δ ) s 2 u | | 2 2 + 3 8 s R 3 g ( u ) u 2 + 8 s 3 G ( u ) d x + ( 1 4 1 2 s * ) | | u | | 2 s * 2 s * > 0
Then, E is bounded from below on D P . Now, we suppose that there exits a sequence { u n } D P , such that | | u n | | and E ( u n ) is bounded from above. Since u n D , we see that | | ( Δ ) s 2 u n | | 2 2 as n . Set λ n s : = 1 | | ( Δ ) s 2 u n | | 2 > 0 and v n : = λ n 3 2 u n ( λ n · ) . Obviously, λ n 0 + as n . Hence, we obtain
R N | v n | 2 d x = λ n 3 R 3 | u n ( λ n x ) | 2 d x = R N | u n | 2 d x ρ
and
| | ( Δ ) s 2 v n | | 2 2 = λ n 2 s | | ( Δ ) s 2 u n | | 2 2 = 1 | | ( Δ ) s 2 u n | | 2 2 | | ( Δ ) s 2 u n | | 2 2 = 1 ,
which implies that { v n } D is bounded in H s ( R N ) .
Set l ( z ) : = B 1 ( z ) | v n | 2 s * d x . From the integral absolute continuity and v n 0 , we have that l ( z ) is continuous on R N and B 1 ( z 0 ) | v n | 2 s * d x : = ξ > 0 for some z 0 R 3 . Then, we can find a large R > 0 , such that R 3 B R ( 0 ) | v n | 2 s * d x < ξ , which means l ( z ) = B 1 ( z ) | v n | 2 s * d x < ξ for any z R 3 B R + 1 ( 0 ) and sup z R 3 l ( z ) = sup z B ¯ R + 1 ( 0 ) l ( z ) .
In view of the compactness of B ¯ R + 1 ( 0 ) , we deduce that there exits y n B ¯ R + 1 ( 0 ) , such that l ( y n ) = sup z B ¯ R + 1 ( 0 ) l ( z ) , that is,
B 1 ( y n ) | v n | 2 s * d x = sup z R 3 B 1 ( z ) | v n | 2 s * d x .
In the following, we claim lim n B 1 ( y n ) | v n | 2 s * d x = 0 , which we prove through contradiction with lim sup n B 1 ( y n ) | v n | 2 s * d x > 0 . Then, v 0 , by virtue of (29) and (19), we have
0 < E ( u n ) | | ( Δ ) s 2 u n | | 2 p s * a 2 λ n ( 2 s * 2 ) s + b 4 λ n ( 2 · 2 s * 2 ) s 1 2 s * | | v n | | 2 s * 2 s * a 2 λ n ( 2 s * 2 ) s + b 4 λ n ( 2 · 2 s * 2 ) s 1 2 s * B 1 ( y n ) | v n | 2 s * d x < 0
if n is large enough. Obviously, it is impossible. Similar to [28], we can obtain lim n B 1 ( y n ) | v n | 2 s * d x = 0 . It is not difficult to obtain v n 0 in L 2 + 8 s 3 . Combining (18) and (19), for any λ > 0 , we can deduce that
R N G λ 3 2 v n ( λ x ) d x 0
as n . It is obvious to know that u n D P satisfies (27) when (15) holds. By Lemma 2, we have
E ( u n ) E λ 3 2 v n ( λ x ) = a λ 2 s 2 | | ( Δ ) s 2 v n | | 2 2 + b λ 4 s 4 | | ( Δ ) s 2 v n | | 2 4 + o ( 1 ) ,
which is a contradiction if λ is large enough. □
Lemma 4.
Assume that u H s ( R 3 ) { 0 } satisfies (27) and ( A 0 )–( A 6 ) hold. Then,
c : = inf D P E ( u ) < c * : = a S 3 2 s 2 T 2 s + b S 3 s 4 T 4 s S 3 2 s 2 s * T 2 s * s ,
where T > 0 is the unique maximum point of the function A ( t ) = a S 3 2 s 2 t 2 s + b S 3 s 4 t 4 s S 3 2 s 2 s * t 2 s * s , t > 0 .
Proof. 
From (22), (23), and (24), we set
u ε ( x ) = ( ρ 1 2 s | | v ε | | 2 1 s ) 3 2 s 2 v ε ( ρ 1 2 s | | v ε | | 2 1 s x ) .
Hence, it is not difficult to find that | | u ε | | 2 2 = ρ , that is, u ε S . Moreover, one obtains
| | ( Δ ) s 2 u ε | | 2 2 = | | ( Δ ) s 2 v ε | | 2 2 S 3 2 s + O ( ε 3 2 s ) , | | u ε | | 2 s * 2 s * = | | v ε | | 2 s * 2 s * = S 3 2 s + O ( ε 3 )
and
R 3 | u ε | 2 ¯ s d x = ( ρ 1 2 s | | v ε | | 2 1 s ) 2 s ( 3 4 s ) 3 R 3 | v ε | 2 ¯ s d x .
From Lemma 2, it is clear that there exists λ ε > 0 , such that λ ε 3 2 u ε ( λ ε · ) P . Due to u ε S , we have λ ε 3 2 u ε ( λ ε · ) S P D P . In view of (29) and (19), we have
0 < c = inf D P E E λ ε 3 2 u ε ( λ ε · ) = a λ ε 2 s 2 | | ( Δ ) s 2 u ε | | 2 2 + b λ ε 4 s 4 | | ( Δ ) s 2 u ε | | 2 4 R 3 G ( λ ε 3 2 u ( x ) ) λ ε 3 d x λ ε 2 s * s 2 s * | | u ε | | 2 s * 2 s * a λ ε 2 s 2 | | ( Δ ) s 2 u ε | | 2 2 + b λ ε 4 s 4 | | ( Δ ) s 2 u ε | | 2 4 λ ε 2 s * s 2 s * | | u ε | | 2 s * 2 s *
Let
l ( λ ε ) = a λ ε 2 s 2 | | ( Δ ) s 2 u ε | | 2 2 + b λ ε 4 s 4 | | ( Δ ) s 2 u ε | | 2 4 λ ε 2 s * s 2 s * | | u ε | | 2 s * 2 s * .
Obviously, l ( λ ε ) as λ ε + and l ( λ ε ) > 0 for λ ε is small enough. Moreover, we can easily find that there exits two constants λ ε 1 , λ ε 2 R + , independent of ε , such that sup λ ε > 0 l ( λ ε ) = sup λ ε [ λ ε 1 , λ ε 2 ] l ( λ ε ) .
Making use of ( A 2 ) and (19), for any M > 0 , there exists C M > , such that G ( t ) M | t | 2 ¯ s C M | t | 2 , t R . Therefore,
R 3 G λ ε 3 2 u ε ( λ ε x ) M R 3 | λ ε 3 2 u ε ( λ ε x ) | 2 ¯ s d x C M R 3 | λ ε 3 2 u ε ( λ ε x ) | 2 d x = M λ ε 4 s R 3 | u ε | 2 ¯ s d x C M R 3 | u ε | 2 d x = M λ ε 4 s ( ρ 1 2 s | | v ε | | 2 1 s ) 2 s ( 3 4 s ) 3 R 3 | v ε | 2 ¯ s d x C M ρ M λ ε 4 s ρ 3 2 s 3 K 1 , s ε ( 4 s 3 ) 2 3 C M ρ ,
where K 1 , s > 0 is a constant. Combining (22) and (23), one obtains
l ( λ ε ) a λ ε 2 s 2 S 3 2 s + O ( ε 3 2 s ) + b λ ε 4 s 4 S 3 s + O ( ε 3 2 s ) λ ε 2 s * s 2 s * S 3 2 s + O ( ε 3 )
Since M can be arbitrarily large, we can obtain c sup t 0 A ( t ) , where
A ( t ) = a S 3 2 s 2 t 2 s + b S 3 s 4 t 4 s S 3 2 s 2 s * t 2 s * s .
Then
A ( t ) = s S 3 2 s t 2 s 1 a + b S 3 2 s t 2 s t ( 2 s * 2 ) s .
Let A ( t ) = s S 3 2 s t 2 s 1 A ~ ( t ) , where
A ~ ( t ) = a + b S 3 2 s t 2 s t ( 2 s * 2 ) s , A ~ ( t ) = s t 2 s 1 2 b S 3 2 s ( 2 s * 2 ) t ( 2 s * 4 ) s .
Since s ( 3 4 , 1 ) , we can find a unique T, such that A ~ ( t ) > 0 for t ( 0 , T ) and A ~ ( t ) < 0 for t > T . Then, A ( T ) = 0 and T is the unique maximum point of A ( t ) . □
Due to
lim t 0 H ( t ) | t | 2 = lim | t | H ( t ) | t | 2 s * = 0 ,
we have the following profile decomposition.
Theorem 3
(Profile decomposition). Assume that { u n } is bounded in H s ( R 3 ) . Then, there exists sequence { u ~ i } i = 0 H s ( R 3 ) , { y n i } i = 0 R 3 for any n 1 , such that y 0 n = 0 , | y n i y n j | + as n for i j , and up to a sequence, the following conditions hold for any i 0 :
u n ( · + y n i ) u ~ i in H s ( R 3 ) as n ,
A 2 : = lim n | | ( Δ ) s 2 u n | | 2 2 = j = 0 i | | ( Δ ) s 2 u ~ i | | 2 2 + lim n | | ( Δ ) s 2 v n i | | 2 2 ,
where v n i ( · ) : = u n j = 0 i u ~ j ( · y n j ) and
lim sup n R N H ( u n ) d x = j = 0 H ( u ~ j ) d x ,
For more details, please refer to [19,20,29,30,31,32,33].
Lemma 5.
Suppose that ( A 0 )–( A 6 ) and (15) hold. Then, c : = inf D P E is attained. If, in addition, g is odd, then ρ is attained by a nonnegative and radially symmetric function in D P .
Proof. 
From Lemma 3, for any sequence { u n } D P , such that E ( u n ) c , we know that { u n } is bounded in H s ( R 3 ) . Set M : = { i 0 : u ~ i 0 } . Then, we claim that M 0 . We prove this through contradiction with M = 0 .
In view of (19) and (32), we have lim n R N H ( u n ) d x = 0 . Therefore, it is easy to know
a + b | | ( Δ ) s 2 u n | | 2 2 | | ( Δ ) s 2 u n | | 2 2 | | u n | | 2 s * 2 s * = o ( 1 ) .
By Lemma 1, up to a subsequence, denote by l 0 , such that
l : = lim n | | u n | | 2 s * 2 s * = lim n a + b | | ( Δ ) s 2 u n | | 2 2 | | ( Δ ) s 2 u n | | 2 2
and
A 2 : = lim n | | ( Δ ) s 2 u n | | 2 2 > 0 .
From (20), we have
lim n S 2 s * 2 | | ( Δ ) s 2 u n | | 2 2 s * lim n a + b | | ( Δ ) s 2 u n | | 2 2 | | ( Δ ) s 2 u n | | 2 2 .
Then
S 2 s * 2 A 2 s * a A 2 + b A 4 > b A 4 .
that is,
A 2 s * 4 > b S 2 s * 2
Making use of Lemma 4, one obtains
a + b S 3 2 s T 2 s T ( 2 s * 2 ) s = 0 , c * = a S 3 2 s 2 T 2 s + b S 3 s 4 T 4 s S 3 2 s 2 s * T 2 s * s .
Then
c * = a S 3 2 s 2 T 2 s + b S 3 s 4 T 4 s S 3 2 s 2 s * T 2 s * s = S 3 2 s T 2 s a 2 + b 4 S 3 2 s T 2 s 1 2 s * T ( 2 s * 2 ) s = S 3 2 s T 2 s 1 2 1 2 s * a + 1 4 1 2 s * b S 3 2 s T 2 s .
Consequently,
E ( u n ) = a 2 + b 4 | | ( Δ ) s 2 u n | | 2 2 | | ( Δ ) s 2 u n | | 2 2 1 2 s * a + b | | ( Δ ) s 2 u n | | 2 2 | | ( Δ ) s 2 u n | | 2 2 + o ( 1 ) = 1 2 1 2 s * a | | ( Δ ) s 2 u n | | 2 2 + 1 4 1 2 s * b | | ( Δ ) s 2 u n | | 2 4 + o ( 1 ) . < 1 2 1 2 s * a S 3 2 s T 2 s + 1 4 1 2 s * b S 3 s T 4 s .
Therefore,
A 2 = lim n | | ( Δ ) s 2 u n | | 2 2 S 3 2 s T 2 s .
Combining (35), we have
S 3 2 s T 2 s > ( b S 2 s * 2 ) 2 2 s * 4 ,
that is,
T > b 3 2 s 2 s ( 4 s 3 ) S 3 ( 3 2 s ) 4 s 2 ( 4 s 3 ) .
It follows from (16) and (38) that
A ~ ( b 3 2 s 2 s ( 4 s 3 ) S 3 ( 3 2 s ) 4 s 2 ( 4 s 3 ) ) = a + b 3 2 s 4 s 3 S 3 4 s 3 b 2 s 4 s 3 S 3 4 s 3 < 0 ,
which is a contradiction to Lemma 4. Thus, M 0 .
In the following, we claim that for every i 0 , it holds that
u n ( · + y n i ) u ~ i ,
in D s , 2 ( R 3 ) . Assume that there exists i 0 , such that
B : = lim n | | ( Δ ) s 2 v n | | 2 2 = lim n | | ( Δ ) s 2 ( u n ( · + y n i ) u ~ i ) | | 2 2 > 0
and
( a + b | | ( Δ ) s 2 u ~ i | | 2 2 ) | | ( Δ ) s 2 u ~ i | | 2 2 > 3 2 s R 3 H ( u ~ i ) d x + | | u ~ i | | 2 s * 2 s *
where v n ( · ) : = u n ( · + y n i ) u ~ i ( · ) . Set K ( · ) = 3 2 s H ( · ) + | · | 2 s * , in view of Vitali’s convergence theorem, one obtains
R 3 ( K ( u n ) K ( v n ) ) d x = R 3 0 1 d d s K ( u n t u ~ i ) d t d x = R 3 0 1 k ( u n t u ~ i ) u ~ i d t d x 0 1 R 3 k ( u ~ i t u ~ i ) u ~ i d t d x = 0 1 R 3 d d s K ( u n t u ~ i ) d t d x = R 3 K ( u ~ i ) d x
as n . There holds that
( a + b ( | | ( Δ ) s 2 v n | | 2 2 + | | ( Δ ) s 2 u ~ i | | 2 2 ) ) ( | | ( Δ ) s 2 v n | | 2 2 + | | ( Δ ) s 2 u ~ i | | 2 2 ) = 3 2 s R 3 H ( v n ) d x + | | v n | | 2 s * 2 s * + 3 2 s R 3 H ( u ~ i ) d x + | | u ~ i | | 2 s * 2 s * + o ( 1 ) .
Combining with (41), it is clear to know that
( a + b | | ( Δ ) s 2 v n | | 2 2 ) | | ( Δ ) s 2 v n | | 2 2 3 2 s R 3 H ( v n ) d x + | | v n | | 2 s * 2 s * + o ( 1 )
In view of ( A 6 ) and arguing as in [32], for any R > 0 , one can find a radial function u H 0 s ( B ( 0 , R ) ) L ( B ( 0 , R ) ) , such that R N H ( u ) d x > 0 . We define r ( · ) as
r ( u ) : = a + a 2 + 4 b ( 3 2 s R 3 H ( u ) d x + | | u | | 2 s * 2 s * ) 2 b | | ( Δ ) s 2 u | | 2 2 1 2 s
and R n = r ( v n ) . Obviously, we can see that u ( r ( u ) · ) P , so P is nonempty.
Now, we claim that R n 1 as n . If
( a + b | | ( Δ ) s 2 v n | | 2 2 ) | | ( Δ ) s 2 v n | | 2 2 > 3 2 s R 3 H ( v n ) d x + | | v n | | 2 s * 2 s *
holds for a.e. on n, by (42), it is easy to know R n 1 as n .
On the other hand,
( a + b | | ( Δ ) s 2 v n | | 2 2 ) | | ( Δ ) s 2 v n | | 2 2 3 2 s R 3 H ( v n ) d x + | | v n | | 2 s * 2 s *
implies that R n 1 . Since lim n ( | | u n | | 2 2 | | v n | | 2 2 ) = | | u ~ i | | 2 2 > 0 , we have v n D and v n ( R n · ) D P for a.e. on n.
Making use of the Br e ´ zis-Lieb Lemma, one obtains
c = inf D P E E ( v n ( R n · ) ) = E ( v n ( R n · ) ) 1 4 s P ( v n ( R n · ) ) = a 4 | | ( Δ ) s 2 v n ( R n · ) | | 2 2 + ( 1 4 1 2 s * ) | | v n ( R n · ) | | 2 s * 2 s * + 3 8 s R 3 g ( v n ( R n · ) ) v n ( R n · ) 2 + 8 s 3 G ( v n ( R n · ) ) d x = a 4 R n 3 + 2 s | | ( Δ ) s 2 v n | | 2 2 + ( 1 4 1 2 s * ) R n 3 | | v n | | 2 s * 2 s * + 3 8 s R n 3 R 3 g ( v n ) v n 2 + 8 s 3 G ( v n ) d x a 4 | | ( Δ ) s 2 v n | | 2 2 + ( 1 4 1 2 s * ) | | v n | | 2 s * 2 s * + 3 8 s R 3 g ( v n ) v n 2 + 8 s 3 G ( v n ) d x = a 4 | | ( Δ ) s 2 u n | | 2 2 a 4 | | ( Δ ) s 2 u ~ i | | 2 2 + ( 1 4 1 2 s * ) | | u n | | 2 s * 2 s * ( 1 4 1 2 s * ) | | u ~ i | | 2 s * 2 s * + 3 8 s R 3 g ( u n ) u n 2 + 8 s 3 G ( u n ) d x 3 8 s R 3 g ( u ~ i ) u ~ i 2 + 8 s 3 G ( u ~ i ) d x + o ( 1 ) a 4 | | ( Δ ) s 2 u n | | 2 2 + ( 1 4 1 2 s * ) | | u n | | 2 s * 2 s * + 3 8 s R 3 g ( u n ) u n 2 + 8 s 3 G ( u n ) d x + o ( 1 ) = E ( u n ) 1 4 s P ( u n ) + o ( 1 ) = E ( u n ) + o ( 1 ) = c + o ( 1 )
which implies that R n 1 as n .
From Theorem 3, it is clear that
( a + b | | ( Δ ) s 2 v n | | 2 2 ) | | ( Δ ) s 2 v n | | 2 2 = 3 2 s R 3 H ( v n ) d x + | | v n | | 2 s * 2 s * + o ( 1 ) = | | v n | | 2 s * 2 s * + o ( 1 ) .
Similar to the above analysis, we can easily know that lim n E ( v n ) c * . Moreover,
E ( u ~ i ) = E ( u ~ i ) 1 4 s P ( u ~ i ) = a 4 | | ( Δ ) s 2 u ~ i | | 2 2 + ( 1 4 1 2 s * ) | | u ~ i | | 2 s * 2 s * + 3 8 s R 3 g ( u ~ i ) u ~ i 2 + 8 s 3 G ( u ~ i ) d x 0 ,
since E ( u n ) = E ( v n ) + E ( u ~ i ) + o ( 1 ) and ( A 0 ) hold. Hence, we have
c = lim n E ( u n ) = lim n E ( v n ) + E ( u ~ i ) = lim n E ( v n ) + E ( u ~ i ) c * ,
which is a contradiction to Lemma 4. Hence, we obtain
0 < ( a + b | | ( Δ ) s 2 u ~ i | | 2 2 ) | | ( Δ ) s 2 u ~ i | | 2 2 3 2 s R 3 H ( u ~ i ) d x + | | u ~ i | | 2 s * 2 s * ,
which yields that r ( u ~ i ) 1 and u ~ i ( r ( u ~ i ) · ) P D .
If r ( u ~ i ) > 1 , then passing to a subsequence u n ( · + y n i ) u ~ i ( x ) for a.e. x R 3 . Using the Fatou Lemma, it is not difficult to deduce that
c = inf D P E E ( u ~ i ( r ( u ~ i ) x ) ) = E ( u ~ i ( r ( u ~ i ) x ) ) 1 4 s P ( u ~ i ( r ( u ~ i ) x ) ) = a 4 | | ( Δ ) s 2 u ~ i ( r ( u ~ i ) x ) | | 2 2 + ( 1 4 1 2 s * ) | | u ~ i ( r ( u ~ i ) x ) | | 2 s * 2 s * + 3 8 s R 3 g ( u ~ i ( r ( u ~ i ) x ) ) u ~ i ( r ( u ~ i ) x ) 2 + 8 s 3 G ( u ~ i ( r ( u ~ i ) x ) ) d x = a 4 r ( u ~ i ) 3 + 2 s | | ( Δ ) s 2 u ~ i ( x ) | | 2 2 + ( 1 4 1 2 s * ) r ( u ~ i ) 3 | | u ~ i ( x ) | | 2 s * 2 s * + 3 8 s r ( u ~ i ) 3 R 3 g ( u ~ i ( x ) ) u ~ i ( x ) 2 + 8 s 3 G ( u ~ i ( x ) ) d x < a 4 | | ( Δ ) s 2 u ~ i ( x ) | | 2 2 + ( 1 4 1 2 s * ) | | u ~ i ( x ) | | 2 s * 2 s * + 3 8 s R 3 g ( u ~ i ( x ) ) u ~ i ( x ) 2 + 8 s 3 G ( u ~ i ( x ) ) d x lim inf n a 4 | | ( Δ ) s 2 u n ( · + y n i ) | | 2 2 + ( 1 4 1 2 s * ) | | u n ( · + y n i ) | | 2 s * 2 s * + 3 8 s R 3 g ( u n ( · + y n i ) ) u n ( · + y n i ) 2 + 8 s 3 G ( u n ( · + y n i ) ) d x lim n E ( u n ) = c ,
which is a contradiction. Thus, r ( u ~ i ) = 1 , u ~ i D P , and u n ( · + y n i ) u ~ i ( x ) in D s , 2 ( R 3 ) . Moreover, by (19), we have
lim n R 3 H ( u n ( · + y n i ) ) d x = R 3 H ( u ~ i ) d x
and
lim n E ( u n ) = E ( u ~ i ) = c .
In addition, we suppose that g is odd. Then, G and H are even, so that G ( | u | ) = G ( u ) and H ( | u | ) = H ( u ) for all u H s ( R 3 ) . Define v ~ i : = | u ~ i | * as the Schwarz symmetrization of | u ~ i | . Then | | v ~ i | | 2 = | | u ~ i | | 2 , that is, v ~ i D . Furthermore, due to
a 2 | | ( Δ ) s 2 v ~ i | | 2 2 + b 4 | | ( Δ ) s 2 v ~ i | | 2 4 a 2 | | ( Δ ) s 2 u ~ i | | 2 2 + b 4 | | ( Δ ) s 2 u ~ i | | 2 4 = R N G ( u ~ i ) d x + 1 2 s * | | u ~ i | | 2 s * 2 s * = R N G ( v ~ i ) d x + 1 2 s * | | v ~ i | | 2 s * 2 s * ,
we obtain that r ( v ~ i ) 1 and v ~ i ( r ( v ~ i ) ) P , where r ( · ) is defined as (43). Using a similar argument as before, we can obtain that r ( v ~ i ) = 1 , so v ~ i P . It is clear that E ( v ~ i ) = c and v ~ i is radially symmetric. □
Lemma 6.
Suppose that ( A 0 )–( A 6 ), and (15) hold. If g is odd, inf S P E < E ( u ) for any u ( D S ) P .
Proof. 
Suppose, by contradiction, that there exits u ~ ( D S ) P , such that
inf D P E = c = E ( u ~ ) inf S P E .
which implies that u ~ is a local minimizer for E on D P . Since D P is an open set in P , we know that u ~ is a local minimizer for E on P . This resules in a Lagrange multiplier μ R , such that
E ( u ~ ) v + μ P ( u ~ ) v = 0 ,
for any v H s ( R 3 ) . Namely, u ~ is a weak solution to
( a ( 1 + μ ) + b ( 1 + 2 μ ) | | ( Δ ) s 2 u ~ | | 2 2 ) ( Δ ) s u ~ = g ( u ~ ) + 3 4 s μ h ( u ~ ) + 1 + 2 s * 2 μ | u ~ | 2 s * 2 u ~ .
In particular, u ~ satisfies the following Nehari-type identity
( a ( 1 + μ ) + b ( 1 + 2 μ ) | | ( Δ ) s 2 u ~ | | 2 2 ) | | ( Δ ) s 2 u ~ | | 2 2 = R 3 g ( u ~ ) u ~ + 3 4 s μ h ( u ~ ) u ~ + 1 + 2 s * 2 μ | u ~ | 2 s * d x .
Moreover, since u ~ P , then P ( u ~ ) = 0 , that is,
a + b | | ( Δ ) s 2 u ~ | | 2 2 | | ( Δ ) s 2 u ~ | | 2 2 = 3 2 s R 3 H ( u ~ ) d x + | | u ~ | | 2 s * 2 s * .
On the other hand, u ~ satisfies Pohozaev and Nehari identities. Therefore,
( a ( 1 + μ ) + b ( 1 + 2 μ ) | | ( Δ ) s 2 u ~ | | 2 2 ) | | ( Δ ) s 2 u ~ | | 2 2 = 3 2 s R 3 H ( u ~ ) + 3 4 s μ ( h ( u ~ ) u ~ 2 H ( u ~ ) ) + 1 2 2 s * 1 + 2 s * 2 μ | u ~ | 2 s * d x
Combining these two identities, it is clear to know
μ a | | ( Δ ) s 2 u ~ | | 2 2 + 2 b | | ( Δ ) s 2 u ~ | | 2 4 9 8 s 2 R 3 h ( u ~ ) u ~ 2 H ( u ~ ) d x 3 3 2 s | | u ~ | | 2 s * 2 s * = 0
By ( A 6 ) and (45), we have
a | | ( Δ ) s 2 u ~ | | 2 2 + 2 b | | ( Δ ) s 2 u ~ | | 2 4 9 8 s 2 R 3 h ( u ~ ) u ~ 2 H ( u ~ ) d x 3 3 2 s | | u ~ | | 2 s * 2 s * a | | ( Δ ) s 2 u ~ | | 2 2 + 3 4 s 3 2 s | | u ~ | | 2 s * 2 s * < 0 .
which indicates μ = 0 . Hence, u ~ is a weak solution to
( a + b | | ( Δ ) s 2 u ~ | | 2 2 ) ( Δ ) s u ~ = g ( u ~ ) + | u ~ | 2 s * 2 u ~ .
and
( a + b | | ( Δ ) s 2 u ~ | | 2 2 ) | | ( Δ ) s 2 u ~ | | 2 2 = R 3 g ( u ~ ) u ~ d x + | | u ~ | | 2 s * 2 s * .
By (45), we have
R 3 g ( u ~ ) u ~ d x = 3 2 s R 3 H ( u ~ ) d x ,
i.e.,
R 3 g ( u ~ ) u ~ 2 s * G ( u ~ ) d x = 0 .
From the elliptic regularity theory and ( A 6 ), u ~ is continuous in R 3 and g ( u ~ ) u ~ = 2 s * G ( u ~ ) for x R 3 . Due to u ~ H s ( R 3 ) , there is an open interval Ω R such that 0 Ω ¯ and 2 G ( u ) = g ( u ) u for u Ω ¯ . If g is odd, in the light of Lemma 5, we may assume that u ~ is nonnegative and radially symmetric. However, it is impossible to show that u ~ is a solution to
( a + b A 2 ) ( Δ ) s u ~ = ( 2 s * C + 1 ) | u ~ | 2 s * 2 u ~ ,
so that u ~ is not a solution to (46), where A 2 = | | ( Δ ) s 2 u ~ | | 2 2 . Because the nonnegative and radial solution of (47) is a Aubin–Talenti instanton, up to a scaling and a translation, which is not L 2 -integrable. □
Proof of Theorem 1.
Considering Lemmas 5 and 6, we know that c = inf S P E is attained. Furthermore, we can infer that there exists Lagrange multipliers λ , μ R , such that u ~ S P is a solution to
( a ( 1 + μ ) + b ( 1 + 2 μ ) | | ( Δ ) s 2 u ~ | | 2 2 ) ( Δ ) s u ~ + λ u ~ = g ( u ~ ) + 3 4 s μ h ( u ~ ) + 1 + 2 s * 2 μ | u ~ | 2 s * 2 u ~ .
In view of the Nehari and Pohozaev identities for (48), one deduces that
( a ( 1 + μ ) + b ( 1 + 2 μ ) | | ( Δ ) s 2 u ~ | | 2 2 ) | | ( Δ ) s 2 u ~ | | 2 2 = 3 2 s R 3 H ( u ~ ) + 3 4 s μ ( h ( u ~ ) u ~ 2 H ( u ~ ) ) + 1 2 2 s * 1 + 2 s * 2 μ | u ~ | 2 s * d x
Due to u ~ S P P , we have P ( u ~ ) = 0 , so
μ a | | ( Δ ) s 2 u ~ | | 2 2 + 2 b | | ( Δ ) s 2 u ~ | | 2 4 9 8 s 2 R 3 h ( u ~ ) u ~ 2 H ( u ~ ) d x 3 3 2 s | | u ~ | | 2 s * 2 s * = 0 .
Similar to the proof of Lemma 6, it is clear that μ = 0 . Combining Lemma 5, under the condition g is odd, we can deduce that u ~ S P is a positive and radially symmetric normalized ground state solution to (1). □

4. Proof of Theorem 2

Proof of Theorem 2 (a).
Taking into account Lemmas 5 and 6, there exists u ρ 1 S ρ 1 P , u ρ 2 S ρ 2 P such that ϑ 1 = E ( u ρ 1 ) , ϑ 2 = E ( u ρ 2 ) for ρ 1 > ρ 2 > 0 . Thus, it is easy to obtain that u ρ 1 ( D ρ 1 S ρ 1 ) P , so that ϑ ρ 1 = inf S ρ 1 P E < E ( u ρ 2 ) = ϑ ρ 2 . □
Proof of Theorem 2 (b).
Assume that η 0 .
(1) Now, we suppose that ρ n ρ + as n . It follows from Lemma 5 that there exists u n D ρ n P , such that E ( u n ) = inf D ρ n P E . Similar to the proof of Lemma 5, up to a translation and up to a subsequence, it is not difficult to know u n u ~ in H s ( R 3 ) and r ( u ~ ) 1 . Furthermore, u ~ ( r ( u ~ ) · ) D ρ P D ρ n P .If r ( u ~ ) > 0 , from Fatou lemma and ( A 5 ), we have
c = inf D ρ P E E ( u ~ ( r ( u ~ ) x ) ) = E ( u ~ ( r ( u ~ ) x ) ) 1 4 s P ( u ~ ( r ( u ~ ) x ) ) = a 4 | | ( Δ ) s 2 u ~ ( r ( u ~ ) x ) | | 2 2 + ( 1 4 1 2 s * ) | | u ~ ( r ( u ~ ) x ) | | 2 s * 2 s * + 3 8 s R 3 g ( u ~ ( r ( u ~ ) x ) ) u ~ ( r ( u ~ ) x ) 2 + 8 s 3 G ( u ~ i ( r ( u ~ i ) x ) ) d x = a 4 r ( u ~ ) 3 + 2 s | | ( Δ ) s 2 u ~ ( x ) | | 2 2 + ( 1 4 1 2 s * ) r ( u ~ ) 3 | | u ~ ( x ) | | 2 s * 2 s * + 3 8 s r ( u ~ ) 3 R 3 g ( u ~ ( x ) ) u ~ ( x ) 2 + 8 s 3 G ( u ~ ( x ) ) d x < a 4 | | ( Δ ) s 2 u ~ ( x ) | | 2 2 + ( 1 4 1 2 s * ) | | u ~ ( x ) | | 2 s * 2 s * + 3 8 s R 3 g ( u ~ ( x ) ) u ~ ( x ) 2 + 8 s 3 G ( u ~ ( x ) ) d x lim inf n a 4 | | ( Δ ) s 2 u n | | 2 2 + ( 1 4 1 2 s * ) | | u n | | 2 s * 2 s * + 3 8 s R 3 g ( u n ) u n 2 + 8 s 3 G ( u n ) d x lim n E ( u n ) lim n inf D ρ n P E lim n inf D ρ P E = inf D ρ P E ,
which implies that r ( u ~ ) = 1 . Similar to the proof of Lemma 5, we can easily deduce that
E ( u ~ ) = inf D ρ P E = lim n inf D ρ n P E .
On the other hand, we suppose that ρ n ρ as n . Choosing u D ρ P so that E ( u ) = inf D ρ P E . Similar to [33], we set θ n : = ρ n ρ 1 as n , v n = θ n u . Taking into account Lemma 2, it is easy to find λ n > 0 , such that λ n 3 2 v n ( λ n · ) P .
Now, we claim that there exists λ > 0 , such that λ n λ as n . We prove this through indirection. Suppose λ n as n passing to a subsequence. By ( A 5 ), we have that
a λ n 2 s θ n 2 | | ( Δ ) s 2 u | | 2 2 + b λ n 4 s θ n 4 | | ( Δ ) s 2 u | | 2 4 λ n 2 s * s θ n 2 s * | | u | | 2 s * 2 s * ,
that is,
b | | ( Δ ) s 2 u | | 2 4 λ n 2 s * s 4 s θ n 2 s * 4 | | u | | 2 s * 2 s * a | | ( Δ ) s 2 u | | 2 2 λ n 2 s θ n 2 ,
as n , which is a contradiction. In addition, we need to exclude λ n 0 as n passing to a subsequence. If lim n λ n 0 , it is clear to know that
b | | ( Δ ) s 2 u | | 2 4 = 3 2 s θ 8 s 3 R 3 H ( λ n 3 2 θ n u ( x ) ) | λ n 3 2 θ n u ( x ) ) | 2 + 8 s 3 | u | 2 + 8 s 3 d x + λ n 2 s * s 4 s θ n 2 s * 4 | | u | | 2 s * 2 s * a | | ( Δ ) s 2 u | | 2 2 λ n 2 s θ n 2 ,
which is impossible. Thus, up to a subsequence, λ n λ as n and λ 3 2 u ( λ · ) P . In view of Lemma 2, we have
lim sup n inf D ρ n P E lim n E ( λ n 3 2 v n ( λ · ) ) = E ( λ 3 2 u ( λ · ) ) = E ( u ) = inf D ρ P E
Furthermore, we can infer that
inf D ρ P E lim sup n inf D ρ n P E lim sup n inf D ρ n P E inf D ρ P E ,
that is,
lim n inf D ρ n P E = inf D ρ P E ,
since D ρ n P D ρ P . Hence, the continuity of the ground state energy map is proven.
(2) Set lim n ρ n = + and u H s ( R 3 ) is a ground state solution to the Equation (1) with ρ = 1 . Moreover, E ( u ) = inf D 1 P E = inf S 1 P E , Without a loss of generality, we may suppose that ρ n > 1 , and define u n = ρ n u . Making use of Lemma 2, due to u n S ρ n D ρ n , there exists λ n > 0 such that v n : = λ n 3 2 u n ( λ n · ) P . Since | | v n | | 2 2 = | | u n | | 2 2 = ρ n , we have v n D ρ n P . It results that
0 < inf D n P E E ( v n ) a 2 λ n 2 s ρ n 2 | | ( Δ ) s 2 u | | 2 2 + b 4 λ n 4 s ρ n 4 | | ( Δ ) s 2 u | | 2 4
As v n P , we have
a 2 λ n 2 s ρ n 2 | | ( Δ ) s 2 u | | 2 2 + b 4 λ n 4 s ρ n 4 | | ( Δ ) s 2 u | | 2 4 = a 2 | | ( Δ ) s 2 v n | | 2 2 + b 4 | | ( Δ ) s 2 v n | | 2 4 = 3 8 s R 3 H ( v n ) d x + 1 4 | | v n | | 2 s * 2 s * + a 4 | | ( Δ ) s 2 v n | | 2 2 = 3 2 s λ n 3 R 3 H ( λ n 3 2 ρ n u ) d x + λ n 2 s * s ρ 2 s * | | u | | 2 s * 2 s * + a 4 λ n 2 s ρ n 2 | | ( Δ ) s 2 u | | 2 2 ,
that is,
b 4 | | ( Δ ) s 2 u | | 2 4 = 3 2 s ρ n 4 λ n 3 4 s R 3 H ( λ n 3 2 ρ n u ) d x + λ n 2 s * s 4 s ρ n 2 s * 4 | | u | | 2 s * 2 s * a 4 | | ( Δ ) s 2 u | | 2 2 λ n 2 s ρ n 2 = 3 2 s ρ n 8 s 3 R 3 H ( λ n 3 2 ρ n u ( x ) ) | λ n 3 2 ρ n u ( x ) ) | 2 + 8 s 3 | u | 2 + 8 s 3 d x + λ n 2 s * s 4 s ρ n 2 s * 4 | | u | | 2 s * 2 s * a 4 | | ( Δ ) s 2 u | | 2 2 λ n 2 s ρ n 2 .
Considering lim n ρ n = , if λ n s ρ n l > 0 as n , we have λ n 3 2 ρ n 0 as n . As η 0 , we can obtain
3 2 s ρ n 8 s 3 R 3 H ( λ n 3 2 ρ n u ( x ) ) | λ n 3 2 ρ n u ( x ) ) | 2 + 8 s 3 | u | 2 + 8 s 3 d x + λ n 2 s * s 4 s ρ n 2 s * 4 | | u | | 2 s * 2 s * a 4 | | ( Δ ) s 2 u | | 2 2 λ n 2 s ρ n 2 ,
as n , which is a contradiction. Obviously, it is impossible that (50) holds when λ n s ρ n as n . Thus, we can obtain λ n s ρ n 0 and λ n 3 2 ρ n 0 as n . It follows from (49) that lim n inf D n P E = 0 , that is, inf D n P E 0 + as ρ . □

5. Conclusions

This article mainly studies a class of nonlinear fractional Kirchhoff equation with at least critical mass growth. We focus on the existence and properties of solutions to Equation (1) and establish a series of assumptions ( A 0 )–( A 6 ) to ensure that the solutions in question satisfy specific conditions. By establishing a novel variational framework that combines Nehari and Pohozaev constraint methods within an L 2 -ball of prescribed radius ρ , we show the existence of solutions. Under the condition where g is odd, the solution is a positive and radially symmetric normalized ground state solution to Equation (1). In addition, we give a detailed description for the asymptotic behavior of the ground state energy map. This study has significant physical implications and provides important theoretical basis for in-depth research on nonlinear optics and Bose Einstein condensates

Author Contributions

Conceptualization, P.J.; writing—original draft preparation, P.J.; writing—review and editing, F.C. All authors have read and agreed to the published version of the manuscript.

Funding

This research was supported by the National Natural Science Foundation of China grant number No. 12172166. The APC was funded by the National Natural Science Foundation of China grant number No. 12172166.

Data Availability Statement

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.

Acknowledgments

We sincerely thank the anonymous reviewers for their valuable comments and suggestions that helped improve this manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Nezza, E.D.; Palatucci, G.; Valdinoci, E. Hitchhiker’s guide to the fractional sobolev spaces. Bull. Sci. Math. 2012, 136, 521–573. [Google Scholar] [CrossRef]
  2. Kirchhoff, G. Mechanik; Teubner: Leipzig, Germany, 1883. [Google Scholar]
  3. Narashima, R. Nonlinear vibration of an elastic string. J. Sound Vib. 1968, 8, 134–146. [Google Scholar] [CrossRef]
  4. Oplinger, D.W. Frequency response of a nonlinear stretched string. J. Acoust. Soc. Am. 1960, 32, 1529–1538. [Google Scholar] [CrossRef]
  5. Arosio, A.; Panizzi, S. On the Well-Posedness of the Kirchhoff String. Trans. Am. Math. Soc. 2002, 305–330. [Google Scholar] [CrossRef]
  6. Fiscella, A.; Valdinoci, E. A critical Kirchhoff type problem involving a nonlocal operator. Nonlinear Anal. 2014, 94, 156–170. [Google Scholar] [CrossRef]
  7. Autuori, G.; Fiscella, A.; Pucci, P. Stationary Kirchhoff problems involving a fractional elliptic operator and a critical nonlinearity. Nonlinear Anal. 2015, 125, 699–714. [Google Scholar] [CrossRef]
  8. Pucci, P.; Saldi, S. Critical stationary Kirchhoff equations in R N involving nonlocal operators. Rev. Mat. Iberoam. 2016, 32, 1–22. [Google Scholar] [CrossRef]
  9. Figueiredo, G.; Ikoma, N.; Júnior, J. Existence and cocentration result for the Kirchhoff type equations with general nonlinearities. Arch. Ration. Mech. Anal. 2014, 213, 931–979. [Google Scholar] [CrossRef]
  10. Pucci, P.; Rădulescu, V. Progress in nonlinear Kirchhoff problems. Nonlinear Anal. 2019, 186, 1–5. [Google Scholar] [CrossRef]
  11. Lu, S.Y.; Mao, A.M. Existence and concentration behavior of normalized solutions for critical Kirchhoff type equations with general nonlinearities. Z. Angew. Math. Phys. 2024, 75, 153. [Google Scholar] [CrossRef]
  12. Cavalcanti, M.M.; Domingos Cavalcanti, V.N.; Soriano, J.A. Global existence and uniform decay rates for the Kirchhoff-Carrier equation with nonlinear dissipation. Adv. Differ. Equ. 2001, 6, 701–730. [Google Scholar] [CrossRef]
  13. Ambrosio, V.; Isernia, T. A multiplicity result for a fractional Kirchhoff equation in R N with a general nonlinearity. Commun. Contemp. Math. 2018, 20, 1750054. [Google Scholar] [CrossRef]
  14. Xiang, M.Q.; Zhang, B.L.; Yang, M.M. A fractional Kirchhoff-type problem in R N without the (AR) condition. Complex Var. Elliptic Equ. 2016, 61, 1481–1493. [Google Scholar] [CrossRef]
  15. Jin, H.; Liu, W. Fractional Kirchhoff equation with a general critical nonlinearity. Appl. Math. Lett. 2017, 74, 140–146. [Google Scholar] [CrossRef]
  16. Li, K.X. Ground state solutions for fractional Kirchhoff type equations with critical growth. Electron. J. Differ. Equ. 2024, 10, 1–14. [Google Scholar] [CrossRef]
  17. Gu, G.Z.; Mu, C.Y.; Yang, Z.P. Existence of Ground State Solutions for a Class of Non-Autonomous Fractional Kirchhoff Equations. Fractal Fract. 2024, 8, 113. [Google Scholar] [CrossRef]
  18. Wang, Y.; Liu, Y. Positive and negative solutions for the nonlinear fractional kirchhoff equation in R N. SN Part. Differ. Equ. Appl. 2020, 1, 25. [Google Scholar] [CrossRef]
  19. Bieganowski, B.; Mederski, J. Normalized ground states of the nonlinear Schrödinger equation with at least mass critical growth. J. Funct. Anal. 2021, 280, 108989. [Google Scholar] [CrossRef]
  20. Li, Q.Q.; Rădulescu, V.; Zhang, W. Normalized ground states for the Sobolev critical Schrödinger equation with at least mass critical growth. Nonlinearity 2024, 37, 025018. [Google Scholar] [CrossRef]
  21. Lu, G.Z.; Shen, Y.S. Existence of Solutions to Fractional p-Laplacian Systems with Homogeneous Nonlinearities of Critical Sobolev Growth. Adv. Nonlinear Stud. 2020, 20, 579–597. [Google Scholar] [CrossRef]
  22. Cotsiolis, A.; Tavoularis, N.K. Best constants for Sobolev inequalities for higher order fractional derivatives. J. Math. Anal. Appl. 2004, 295, 225–236. [Google Scholar] [CrossRef]
  23. Li, G.; Ye, H. Existence of positive ground state solutions for the nonlinear Kirchhoff type equations in R 3. J. Differ. Equ. 2014, 257, 566–600. [Google Scholar] [CrossRef]
  24. Servadei, R.; Valdinoci, E. The Brezis-Nirenberg result for the fractional Laplacian. Trans. Am. Math. Soc. 2014, 367, 67–102. [Google Scholar] [CrossRef]
  25. Teng, K. Existence of ground state solutions for the nonlinear fractional Schrödinger-Poisson system with critical Sobolev exponent. J. Differ. Equ. 2016, 261, 3061–3106. [Google Scholar] [CrossRef]
  26. Frank, R.L.; Lenzmann, E.; Silvestre, L. Uniqueness of Radial Solutions for the Fractional Laplacian. Commun. Pure Appl. Math. 2016, 69, 1671–1726. [Google Scholar] [CrossRef]
  27. Luo, H.J.; Zhang, Z.T. Normalized solutions to the fractional Schrödinger equations with combined nonlinearities. Calc. Var. Part. Differ. Equ. 2020, 59, 1–35. [Google Scholar] [CrossRef]
  28. Bartsch, T.; Wang, Z.; Willem, M. The Dirichlet problem for superlinear elliptic equations. In Handbook of Differential Equations: Stationary Partial Differential Equations; Elsevier: Amsterdam, The Netherlands, 2015; Volume 2, pp. 1–55. [Google Scholar]
  29. Devillanova, G.; Solimini, S. Some Remarks on Profile Decomposition Theorems. Adv. Nonlinear Stud. 2016, 16, 795–805. [Google Scholar] [CrossRef]
  30. Mederski, J. Nonradial Solutions of Nonlinear Scalar Field Equations. Nonlinearity 2017, 33, 6349–6380. [Google Scholar] [CrossRef]
  31. Liu, Z.; Squassina, M.; Zhang, J. Ground states for fractional Kirchhoff equations with critical nonlinearity in low dimension. Nonlinear Differ. Equ. Appl. Nodea 2017, 24, 50. [Google Scholar] [CrossRef]
  32. Berestycki, H.; Lions, P.L. Nonlinear scalar field equations, I existence of a ground state. Arch. Ration. Mech. Anal. 1983, 82, 313–346. [Google Scholar] [CrossRef]
  33. Jeanjean, L.; Lu, S.S. A mass supercritical problem revisited. Calc. Var. Part. Differ. Equ. 2020, 59, 174–217. [Google Scholar] [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Ji, P.; Chen, F. Normalized Ground States for the Sobolev Critical Fractional Kirchhoff Equation with at Least Mass Critical Growth. Fractal Fract. 2025, 9, 482. https://doi.org/10.3390/fractalfract9080482

AMA Style

Ji P, Chen F. Normalized Ground States for the Sobolev Critical Fractional Kirchhoff Equation with at Least Mass Critical Growth. Fractal and Fractional. 2025; 9(8):482. https://doi.org/10.3390/fractalfract9080482

Chicago/Turabian Style

Ji, Peng, and Fangqi Chen. 2025. "Normalized Ground States for the Sobolev Critical Fractional Kirchhoff Equation with at Least Mass Critical Growth" Fractal and Fractional 9, no. 8: 482. https://doi.org/10.3390/fractalfract9080482

APA Style

Ji, P., & Chen, F. (2025). Normalized Ground States for the Sobolev Critical Fractional Kirchhoff Equation with at Least Mass Critical Growth. Fractal and Fractional, 9(8), 482. https://doi.org/10.3390/fractalfract9080482

Article Metrics

Back to TopTop