See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/307518397

Positive solutions of kirchhoff type problem with singular and critical

nonlinearities in dimension four

Article · September 2016

DOI: 10.3934/cpaa.2016.15.1841

CITATIONS READS

4 285

4 authors, including:

Chun-Lei Tang Jia-Feng Liao

Southwest University in Chongqing China West Normal University
188 PUBLICATIONS   2,585 CITATIONS    26 PUBLICATIONS   155 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Critical point theory -- Application to elliptic partial differential equations View project

Pure Mathematics; Analysis; Functional Analysis; partial differential equation View project

All content following this page was uploaded by Chun-Lei Tang on 12 October 2016.

The user has requested enhancement of the downloaded file.

 Positive solutions of Kirchhoff type problem with
singular and critical nonlinearities in dimension four∗
Rui-Qi Liu, Chun-Lei Tang, Jia-Feng Liao, Xing-Ping Wu†
School of Mathematics and Statistics, Southwest University, Chongqing 400715,
People’s Republic of China

Abstract: In this article, we study the existence and multiplicity of positive solutions
for the Kirchhoff type problem with singular and critical nonlinearities
  Z 
2 3 λ
− a + b |∇u| dx ∆u = µu + |x|β uγ , in Ω,



Ω
(0.1)

 u > 0, in Ω,

u = 0, on ∂Ω,

where Ω ⊂ R4 is a bounded smooth domain and a, b > 0, λ > 0. For all µ > 0 and
γ ∈ (0, 1), 0 ≤ β < 3, we obtain one positive solution for problem (0.1). Particularly, we
prove that this problem has at least two positive solutions for all µ > bS 2 and γ ∈ (0, 21 ),
2 + 2γ < β < 3.
Key words: Kirchhoff type problem; Critical exponent; Singularity; Positive solution;
Perturbation method

1 Introduction and main result

In this paper, we investigate the following Kirchhoff type problem with singularity
and critical exponent
  Z 
2 3 λ
− a + b |∇u| dx ∆u = µu + |x|β uγ , in Ω,


Ω
(1.1)

 u > 0, in Ω,

u = 0, on ∂Ω,

where Ω is a bounded smooth domain in R4 and a, b > 0, µ, λ > 0, γ ∈ (0, 1),

0 ≤ β < 3 is a constant.
In problem (1.1), if we replace µu3 + |x|βλuγ by f (x, u), it reduces to the following
Kirchhoff type of equations with Dirichlet boundary value conditions
  Z 
− a + b |∇u|2 dx ∆u = f (x, u), in Ω,
Ω (1.2)
u = 0, on ∂Ω,


∗
Supported by Nationa Science Foundation of China(No. 11471267)
†
Corresponding author. E-mail address: wuxp@swu.edu.cn (X.-P. Wu).

1
which is related to the stationary analogue of the equation
Z L
∂ 2u
  2
P0 E ∂u 2 ∂ u
ρ 2 − + | | dx = f (x, t, u)
∂t h 2L 0 ∂x ∂x2

proposed by Kirchhoff in [10]. Many authors have paid much attention to the solv-
ability of the Kirchhoff problem (1.2). However, the singular Kirchhoff type problems
have few considered except [11], [13] and [17]. In [11], Lei et al. have studied the
existence and multiplicity of positive solutions for equations
  Z 
2 5 λ
− a + b |∇u| dx ∆u = u + uγ , in Ω,


Ω

 u > 0, in Ω,

u = 0, on ∂Ω,

where Ω is a smooth bounded domain in R3 . They have obtained two positive so-
lutions via the variational methods and perturbation methods. And recently, the
solvability or multiplicity of the Kirchhoff type equation with critical exponent has
been paid much attention by many authours; we refer readers to [7], [8], [11], [12],
[16], [17], [22], [27]. Before [22], many papers have studied the Kirchhoff type prob-
lems only in dimension three or less except for [7], [12], [15]. Daisuke Naimen in [22],
first concerned the following Kirchhoff type problem in R4 ,
  Z 
2 3 q
− a + b |∇u| dx ∆u = µu + λu , in Ω,


Ω

 u > 0, in Ω,

u = 0, on ∂Ω,

where Ω ⊂ R4 is a smooth bounded domain, and a, λ, µ > 0, b ≥ 0, 1 ≤ q < 3.

The author used the variational methods and concentration compactness argument
to deal with the problem and obtained the existence results.
Inspired by [11] and [22], we consider the 4-dimensional critical and singular
problem (1.1). In a critical case, the typical difficulty is due to the lack of the
compactness of the Sobolev embedding H01 (Ω) ,→ L4 (Ω).R 4Particularly in our problem,
not only the nonlocal term but also the nonlinearity Ω u dx is crucial. Moreover, the
non-differentiability of the singular term on H01 (Ω) makes it harder for us to solve the
case. In this paper, we use the variational and perturbation methods to deal with
the difficulties. To our best knowledge, this article is the first one that attacks this
kind of critical and singular kirchhoff type equation in dimension four.
We need the following preliminaries. The H01 (Ω) space is equipped with the inner
product and the norm
Z
1
hu, vi = (∇u, ∇v)dx, ||u|| = hu, ui 2 .
Ω
R 1
| · |p denotes the usual Lp − norm, i.e. |u|p = ( Ω |u|p dx) p . Let S be the best Sobolev
constant, namely R
Ω
|∇u|2 dx
S := inf
1 . (1.3)
u∈H01 (Ω)\{0}
R
|u|4 dx 2
Ω

2
We define the functional
(u+ )1−γ
Z Z
a b µ λ
I(u) = kuk2 + kuk4 − + 4
(u ) dx − dx, ∀u ∈ H01 (Ω),
2 4 4 Ω 1−γ Ω |x|β

where u± = max{±u(x), 0}. Generally, if u ∈ H01 (Ω) with u > 0 and for all ϕ ∈ H01 (Ω)
it satisfies
(u+ )−γ
Z Z Z
2 + 3


a + bkuk (∇u, ∇ϕ)dx − µ(u ) ϕdx − λ β
ϕdx = 0,
Ω Ω Ω |x|

we call that u is a solution of problem (1.1).

Now we state our main result.

Theorem 1.1. Assume a, b > 0, µ > 0, 0 < γ < 1 and 0 ≤ β < 3. Then there exists
λ∗ > 0 such that problem (1.1) has at least one positive solution for all 0 < λ < λ∗ .

Theorem 1.2. Assume a, b > 0, µ > bS 2 , 0 < γ < 21 and 2 + 2γ < β < 3. Then
there exists λ∗∗ > 0 such that problem (1.1) has at least two positive solutions for all
0 < λ < λ∗∗ .

Remark 1. To the best of our knowledge, problem (1.1) has not been investigated
before. On one hand, our results extend a resent result of Daisuke Naimen [22] to
negative index. On the other hand, comparing with [11], our results extend the so-
lutions of the singular and critical Kirchhoff type problem from dimension three to
dimension four.

This paper is organized as follows. In Section 2, we give the proof of Theorem

1.1. In Section 3, we show the proof of Theorem 1.2.

2 Proof of Theorem 1.1

In this Section, we show that problem (1.1) has a positive solution which is a local
minimizer of the functional I.

Lemma 2.1. Assume a, b, µ > 0, 0 < γ < 1 and 0 ≤ β < 3. Then there exist λ∗ > 0
and R, ρ > 0 such that, for every λ ∈ (0, λ∗ ), we have

I(u)|u∈SR ≥ ρ, and inf I(u) < 0, (2.1)

u∈BR

where SR = {u ∈ H01 (Ω) : kuk = R}, BR = {u ∈ H01 (Ω) : kuk ≤ R}.

Proof. Let R0 be a constant such that Ω ⊂ B(0, R0 ) = {x ∈ R4 : |x| < R0 }. By

Hölder’s inequality and (1.3), for all u ∈ H01 (Ω), exists a constant C0 > 0 such that

(u+ )1−γ |u|1−γ

Z Z
dx ≤ dx
Ω |x|β Ω |x|
β
Z  1−γ Z  3+γ
4
4 1 4
≤ |u| dx 4β dx
Ω Ω |x| 3+γ

3
 Z  3+γ
− 1−γ 1 4
≤S 2 ||u||1−γ 4β dx
B(0,R0 ) |x| 3+γ
1−γ
≤C0 R03+γ−β S − 2 ||u||1−γ . (2.2)

Combining (1.3) and (2.2), we have

(u+ )1−γ
Z Z
a 2 b 4 µ + 4 λ
I(u) = kuk + kuk − (u ) dx − dx
2 4 4 Ω 1 − γ Ω |x|β
a µ − bS 2 λ 3+γ−β − 1−γ
≥ kuk2 − kuk 4
− C 0 R 0 S 2 kuk1−γ
2 4S 2 1−γ
µ − bS 2
 
1−γ a 1+γ 3+γ λ 3+γ−β − 1−γ
≥kuk kuk − kuk − C0 R0 S 2 . (2.3)
2 4S 2 1−γ
µ−bS 2 3+γ
We set g(t) = a2 t1+γ − 4S 2
t for t > 0.
  12
2 2a(1+γ)S 2
When µ > bS , it’s easy to obtain a constant R1 = (µ−bS 2 )(3+γ)
> 0, such
1−γ
(1−γ)S 2
that maxt>0 g(t) = g(R1 ) > 0. Letting λ1 = 2C0 R03+γ−β
g(R1 ), it follows that there
exists a constant ρ > 0 such that I(u)|SR1 ≥ ρ for every λ ∈ (0, λ1 ).
When 0 < µ ≤ bS 2 , from (2.3) we see that I(u) → +∞ as ||u|| → +∞.
Therefore, I is coercive. Of course, we can find a R2 > 0 and a constant ρ > 0 such
1−γ
(1−γ)S 2
that I(u)|SR2 ≥ ρ for every λ ∈ (0, λ2 ), where λ2 = 2C0 R03+γ−β
g(R2 ) and g(R2 ) > 0.
Thus, for all µ > 0, there exist R, λ∗ , ρ > 0 such that I(u)|SR ≥ ρ for every
λ ∈ (0, λ∗ ). Choosing u ∈ BR with u+ 6= 0, one gets

(u+ )1−γ
Z
I(tu) λ
lim+ 1−γ = − dx < 0.
t→0 t 1 − γ Ω |x|β

So we have I(tu) < 0 for all u+ 6= 0 and t small enough. Hence, we obtain

m1 = inf I(u) < 0. (2.4)

u∈BR

for kuk small enough. So we complete the proof of Lemma 2.1. 

Theorem 2.2. Assume a, b, µ > 0, 0 < γ < 1, 0 ≤ β < 3 and 0 < λ < λ∗ (λ∗
defined in Lemma 2.1). Then problem (1.1) has a positive solution u∗ ∈ H01 (Ω) with
I(u∗ ) < 0.

Proof. First, we prove that there exists u∗ ∈ BR such that I(u∗ ) = m1 < 0. From
(2.1), we deduce
Z
a 2 b 4 µ
kuk + kuk − (u+ )4 dx ≥ ρ, for u ∈ SR ,
2 4 4 Ω
and Z
a b µ
kuk2 + kuk4 − (u+ )4 dx ≥ 0, for u ∈ BR . (2.5)
2 4 4 Ω

4
Let {un } ⊂ BR be a minimizing sequence satisfying limn→∞ I(un ) = m1 < 0. Since
{un } is bounded and BR is closed and convex, there exists u∗ ∈ BR ⊂ H01 (Ω) and a
subsequence still denoted as {un } such that

1
un * u∗ , weakly in H0 (Ω),

un → u∗ , strongly in Lp (Ω) (1 ≤ p < 4), (2.6)

u (x) → u (x), a.e. in Ω.
n ∗

By the Lebesgue dominated convergence theorem, we obtain

(u+ 1−γ
(u+ 1−γ
Z Z
n) ∗)
dx = dx + o(1). (2.7)
Ω |x|β Ω |x|β
Set wn = un − u∗ , we have
||un ||2 = ||wn ||2 + ||u∗ ||2 + o(1), (2.8)
||un ||4 = ||wn ||4 + ||u∗ ||4 + 2||wn |2 ||u∗ ||2 + o(1). (2.9)
By Brézis-Lieb’s Lemma (see [4]), it holds that
Z Z Z
(un ) dx = (wn ) dx + (u+
+ 4 + 4 4
∗ ) dx + o(1). (2.10)
Ω Ω Ω

If u∗ = 0, we get wn = un , which implies that wn ∈ BR . If u∗ 6= 0, by (2.8), we have

wn ∈ BR for n large sufficiently. Thus from (2.5), it holds that
Z
a 2 b 4 µ
kwn k + kwn k − (w+ )4 dx ≥ 0. (2.11)
2 4 4 Ω n
Therefore, from (2.7) to (2.11) we have
m1 = I(un ) + o(1) Z
a 2 b 4 b 2 2 µ
= I(u∗ ) + kwn k + kwn k + kwn k ku∗ k − (w+ )4 dx + o(1)
2 4 2 4 Ω n
b
≥ I(u∗ ) + kwn k2 ku∗ k2 + o(1)
2
≥ I(u∗ ) + o(1)
≥ m1 + o(1).
This implies I(u∗ ) = m1 < 0 and u∗ ≡ 6 0. It follows that u∗ is a local minimizer of
I. Therefore there exists δ1 ∈ (0, 1) such that (1 + t)u∗ ∈ BR for |t| ≤ δ1 . Define
h : [−δ1 , δ1 ] by h(t) = I((1 + t)u∗ ). Clearly, h(t) achieves its minimum at t = 0,
namely
(u+ 1−γ
Z Z
0 2 4 + 4 ∗)
h (t)|t=0 = aku∗ k + bku∗ k − µ (u∗ ) dx − λ dx = 0. (2.12)
Ω Ω |x|β
Next, we prove that u∗ is a solution of problem (1.1) and u∗ > 0. For any
ψ ∈ H01 (Ω), ψ ≥ 0, let t > 0 small enough, such that u∗ + tψ ∈ BR , we have
I(u∗ + tφ) − I(u∗ )
0 ≤ lim inf
t→0+
Z t Z
= (a + bku∗ k ) (∇u∗ , ∇φ)dx − µ (u+
2 3
∗ ) φdx (2.13)
ΩZ Ω
λ ((u∗ + tφ)+ )1−γ − (u+∗)
1−γ
− lim sup dx.
1 − γ t→0+ Ω |x|β t

5
 Notice that
((u∗ + tφ)+ )1−γ − (u+ 1−γ
Z
λ ∗)
lim sup dx
1 − γ t→0+ Ω |x|β t
((u∗ + tψ)+ )1−γ − (u+ 1−γ
Z
λ ∗)
≥ lim inf dx
1 − γ t→0+ {x∈Ω:u∗ (x)≥0} |x|β t
((u∗ + tψ)+ )1−γ − (u+ 1−γ
Z
λ ∗)
+ lim inf dx
1 − γ t→0+ {x∈Ω:u∗ (x)<0} |x|β t
((u∗ + tψ)+ )1−γ − (u+ 1−γ
Z
λ ∗)
≥ lim inf dx
1 − γ t→0+ {x∈Ω:u∗ (x)≥0} |x|β t
−γ
(u+
Z
∗) ψ
≥λ dx. (2.14)
{x∈Ω:u∗ (x)≥0} |x|β

Therefore, from the estimate above and (2.13) we deduce that

−γ
(u+
Z Z Z
2 + 3 ∗) ψ
(a + bku∗ k ) (∇u∗ , ∇ψ)dx − µ (u∗ ) ψdx − λ dx ≥ 0, ψ ≥ 0. (2.15)
Ω Ω Ω |x|β
We define Ψ ∈ H01 (Ω) by
Ψ = (u+ +
∗ + εφ) ,

for any φ ∈ H01 (Ω) and ε > 0. Then replacing ψ with Ψ in (2.15), from (2.12), it
holds that
−γ
(u+
Z  
2 + 3 ∗)
0≤ (a + bku∗ k )(∇u∗ , ∇Ψ) − µ(u∗ ) Ψ − λ Ψ dx
Ω |x|β
Z
= (a + bku∗ k2 )(∇u∗ , ∇(u+ ∗ + εφ))dx
{x|u+
∗ +εφ>0}
−γ
(u+
Z  
+ 3 + ∗) +
− µ(u∗ ) (u∗ + εφ) + λ (u∗ + εφ) dx
{x|u+
∗ +εφ>0}
|x|β
(u+ 1−γ
Z Z
2 4 + 4 ∗)
≤aku∗ k + bku∗ k − µ (u∗ ) dx − λ dx
Ω Ω |x|β
−γ
(u+
Z  
2 + 3 ∗)
+ε (a + bku∗ k )(∇u∗ , ∇φ) − µ(u∗ ) φ − λ φ dx
Ω |x|β
Z
− (a + bku∗ k2 )(∇u∗ , ∇(u+ ∗ + εφ))dx
+
{x|u∗ +εφ≤0}
−γ
(u+
Z  
+ 3 + ∗) +
+ µ(u∗ ) (u∗ + εφ) + λ (u∗ + εφ) dx
{x|u+
∗ +εφ≤0}
|x|β
−γ
(u+
Z  
2 + 3 ∗)
≤ε (a + bku∗ k )(∇u∗ , ∇φ) − µ(u∗ ) φ − λ φ dx
Ω |x|β
Z
−ε (a + bku∗ k2 )(∇u∗ , ∇φ)dx. (2.16)
{x|u+
∗ +εφ≤0}

Since the measure of the domain of integration {x|u+

∗ + εφ ≤ 0} → 0 as ε → 0, we
have Z
lim (∇u∗ , ∇φ)dx = 0.
ε→0 {x|u+
∗ +εφ≤0}

6
Thus, dividing by ε and letting ε → 0 in (2.16), we get that
−γ
(u+
Z Z Z
2 + 3 ∗)
(a + bku∗ k ) (∇u∗ , ∇φ)dx − µ(u∗ ) φdx − λ β
φdx ≥ 0. (2.17)
Ω Ω Ω |x|

Since the arbitrariness of φ, this inequality also holds for −φ, i.e.
−γ
(u+
Z Z Z
2 + 3 ∗)
(a + bku∗ k ) (∇u∗ , ∇φ)dx − µ(u∗ ) φdx − λ β
φdx = 0. (2.18)
Ω Ω Ω |x|

Replacing the test function φ with u− −

∗ in (2.18), one gets ku∗ k = 0, which implies
that u∗ ≥ 0. Then by (2.15), we have
Z
(∇u∗ , ∇ψ)dx ≥ 0, ψ ∈ H01 (Ω),
Ω

which means that u∗ ∈ H01 (Ω) and satisfies

−∆u∗ ≥ 0, in Ω.

Therefore, by the strong maximum principle we have u∗ > 0 in Ω. Then we can

conclude that u∗ is a positive solution of problem (1.1) with I(u∗ ) = m1 < 0. This
completes the proof of Theorem 2.2. 

3 Proof of Theorem 1.2

In this Section, we prove the multiplicity of the positive solutions of problem (1.1).
From Section 2, we have obtained that problem (1.1) has a positive solution for all
µ > 0, 0 < γ < 1 and 0 ≤ β < 3. Hence, in the following, we only need to obtain the
second positive solution of problem (1.1) with µ > bS 2 , 0 < γ < 21 and 2+2γ < β < 3.
Then we can complete the proof of Theorem 1.2.
In order to overcome the difficulty caused by the singular term and get the second
positive solution of problem (1.1), we consider the following perturbation problem:
  Z 
2 3 λ
− a + b |∇u| dx ∆u = µu + |x|β (u+α)γ , in Ω,


Ω
(3.1)

 u > 0, in Ω,

u = 0, on ∂Ω.

We look for solutions of problem (3.1) by finding critical points of C 1 -functional on

H01 (Ω) given by
(u+ + α)1−γ − α1−γ
Z Z
a 2 b 4 µ + 4 λ
Iα (u) = kuk + kuk − (u ) dx − dx.
2 4 4 Ω 1−γ Ω |x|β
If u is a solution of problem (3.1), it satisfies
Z Z Z
2

+ 3 ϕ
a + bkuk (∇u, ∇ϕ)dx − µ (u ) ϕdx − λ dx = 0, (3.2)
Ω Ω Ω |x|β (u+ + α)γ
for any ϕ ∈ H01 (Ω).

7
 First, we show that the perturbation problem has at least one mountain-pass
solution.

Lemma 3.1 Let a, b > 0, µ > bS 2 , 0 < α < 1, 0 < γ < 1 and 0 ≤ β < 3. Then
2
a2 s2
if {vn } ⊂ H01 (Ω) is a (P S)c sequence for Iα with c < 4(µ−bs 2 ) − Dλ
1+γ , then {v }
n
2
2C0 R03+γ−β 1+γ
 
strongly converges in H01 (Ω) up to subsequences, where D = a(1+γ)
4(1−γ) 1−γ .
aS 2

Proof. Let {vn } ⊂ H01 (Ω) be such that

Iα (vn ) → c, Iα0 (vn ) → 0 as n → ∞. (3.3)
We first claim that {vn } is bounded in H01 (Ω). From the subadditivity of t1−γ , we
have
(v + + α)1−γ − α1−γ ≤ (v + )1−γ ∀v ∈ H01 (Ω). (3.4)
From (2.2), (3.3) and (3.4), we have
1
1 + c + o(1)kvn k ≥ Iα (vn ) − hIα0 (vn ), vn i
4 Z
(vn+ + α)1−γ − α1−γ
Z
a 2 λ vn λ
≥ kvn k + dx − dx
4 4 ZΩ |x|β (vn+ + α)γ 1−γ Ω |x|β
vn− (vn+ )1−γ
Z
a λ λ
≥ kvn k2 − dx − dx
4 4 Ω Z|x|β αγ 1Z− γ Ω |x|β
(vn+ )1−γ
Z
a 2 λ 4 1
− 4β 3 λ
≥ kvn k − γ ( |vn | dx) ( |x| dx) −
4 3 4 dx
4 4α Ω Ω 1 − γ Ω |x|β
0
a Cλ 1 λ 1−γ
≥ kvn k2 − γ R03−β S − 2 ||vn || − C0 R03+γ−β S − 2 kvn k1−γ ,
4 4α 1−γ
0
where C > 0 is a constant. Since 0 < 1 − γ < 2, this proves our claim. Therefore
{vn } is bounded in H01 (Ω). Up to a subsequence, there exists v∗ ∈ H01 (Ω) such that

v * v∗ , weakly in H01 (Ω),
 n


 vn → v∗ , strongly in Lp (Ω) (1 ≤ p < 4),
(3.5)

 vn (x) → v∗ (x), a.e. in Ω,

there exists k ∈ L1 (Ω) such that for all n, |vn (x)| ≤ k(x) a.e. in Ω.


|v∗ | |v∗ (x)|

For given α > 0 and |x|β (vn+
+α)γ
≤ αγ |x|β
, by the dominated convergence theorem and
(3.5) one gets
(vn+ + α)−γ (v∗+ + α)−γ
Z Z
lim v∗ dx = v∗ dx. (3.6)
n→∞ Ω |x|β Ω |x|β
Moreover, from (3.5), we have
 
 vn  |vn | k(x)
 |x|β (v + + α)γ  ≤ αγ |x|β ≤ αγ |x|β .
 
n

Thus, applying the dominated convergence theorem it holds that

(vn+ + α)−γ (v∗+ + α)−γ
Z Z
lim vn dx = v∗ dx. (3.7)
n→∞ Ω |x|β Ω |x|β

8
 Set wn = vn − v∗ , we claim that kwn k → 0 as n → ∞. Otherwise, there is a
subsequence of {wn } such that limn→∞ kwn k = l > 0. Since limn→∞ hIα0 (vn ), v∗ i = 0
and from (3.6) we obtain
(v∗+ + α)−γ
Z Z
2 4 2 2 + 4
akv∗ k + bkv∗ k + bl kv∗ k − µ (v∗ ) dx − λ v∗ dx = 0. (3.8)
Ω Ω |x|β
By Brézis-Lieb’s Lemma and (3.3), we have
hIα0 (vn ), vn i = akv∗Zk2 + akwn k2 + Zbkv∗ k4 + bkwn kZ4 + 2bkwn k2 kv∗ k2
(vn+ + α)−γ
−µ (wn+ )4 dx − µ (v∗+ )4 dx − λ vn dx + o(1)
Ω Ω Ω |x|β
= o(1).
Thus, from the equality above, (3.7) and (3.8), one has
Z
2 4 2
akwn k + bkwn k + bkwn k kv∗ k − µ (wn+ )4 dx = o(1).
2
(3.9)
Ω

Then let n → ∞ in (3.9), from (1.3) we have

al2 + bl4 + bl2 kv∗ k2 ≤ µS −2 l4 ,
which implies that
(a + b||v∗ ||2 )S 2
l2 ≥ . (3.10)
µ − bS 2
Moreover, from (3.3) and (3.9), applying Brézis-Lieb’s Lemma and letting n → ∞,
we infer that
a b
Iα (v∗ ) = c − l2 − l2 kv∗ k2 . (3.11)
4 4
Since vn is bounded in H01 (Ω), we can set d = ||vn ||2 + o(1). From (3.3), we have
limn→∞ hIα0 (vn ), ϕi = 0, for all ϕ ∈ H01 (Ω). Then we get
(v∗+ + α)−γ
Z Z Z
+ 3
(a + bd) (∇v∗ , ∇ϕ)dx − µ (v∗ ) ϕdx − λ ϕdx = 0. (3.12)
Ω Ω Ω |x|β
Replacing ϕ with v∗− in (3.12), one gets ||v∗− || = 0, which implies that v∗ ≥ 0.

On one hand, combining (2.2), (3.4), (3.8) and using the Young inequality, we
obtain
(v∗+ + α)1−γ − α1−γ (v∗+ + α)−γ
Z Z
a 2 b 2 2 λ λ
Iα (v∗ ) ≥ kv∗ k − l kv∗ k − dx + v∗ dx
4 4 1−γ Ω |x|β 4 Ω |x|β
|v∗ |1−γ
Z
a λ b 2
≥ kv∗ k2 − β
dx − l kv∗ k2
4 1 − γ Ω |x| 4
a 2 λ 3+γ−β − 1−γ b
≥ kv∗ k − C0 R0 S 2 kv∗ k1−γ − l2 kv∗ k2
4 1−γ 4
2
! 1+γ
a(1 + γ) 2C0 R03+γ−β 2 b
≥ − 1−γ λ 1+γ − l2 kv∗ k2
4(1 − γ) aS 2 4
2 b
= −Dλ 1+γ − l2 kv∗ k2 .
4
(3.13)

9
 2
a2 S 2
On the other hand, by (3.11), (3.10) and c < 4(µ−bS 2 )
− Dλ 1+γ , we have

a b
Iα (v∗ ) = c − l2 − l2 kv∗ k2
4 4
a (a + b||v∗ ||2 )S 2 b 2
≤ c− 2
− l kv∗ k2
4 µ − bS 4
a2 S 2 b 2
≤ c− 2
− l kv∗ k2
4(µ − bS ) 4
2 b
< −Dλ 1+γ − l2 kv∗ k2 ,
4
which contradicts the inequality (3.13). Therefore vn → v∗ in H01 (Ω) as n → ∞. It
follows that Iα satisfies the (P S)c condition. This completes the proof of Lemma 3.1.

According to [4] and [25], we introduce the cut-off Talenti function,

ετ (x)
uε (x) = ∈ H01 (Ω)
ε2+ |x| 2

where ε > 0 and τ ∈ C0∞ (Ω) is a cut-off function with 0 ≤ τ (x) ≤ 1 and τ (x) = 1 for
x in some neighborhood of 0 ∈ Ω. Set vε (x) = u|uεε(x)
|4
, we can obtain
R
2 2
 Ω |∇v | dx = S + O(ε ),

(3.14)
R v 4 dx = 1.

Ω 

Lemma 3.2 Suppose a, b > 0, µ > bS 2 , 0 < γ < 1, 0 ≤ β < 3 and R, ρ > 0,
0 < λ < λ∗ (where λ∗ , R and ρ are as in Lemma 2.1). Then the functional Iα satisfies
the following conditions
(i) Iα (u) ≥ ρ > 0 if u ∈ SR ;
(ii) there exists u∗∗ ∈ H01 (Ω) such that ku∗∗ k > R and Iα (u∗∗ ) < ρ.

Proof. (i) From (3.4) it holds that

Iα (u) ≥ I(u), ∀u ∈ H01 (Ω).
Therefore, the conclusion follows from (2.1).

(ii) For every vε ∈ H01 (Ω), vε 6= 0, and t > 0. Using (3.14), we have
µt4
Z
a 2 2 b 4 4
Iα (tvε ) = t kvε k + t kvε k − (vε )4 dx
2 4 4 Ω
(tvε + α)1−γ − α1−γ
Z
λ
− dx.
1−γ Ω |x|β
µt4
Z
a 2 b
≤ t kvε k2 + t4 kvε k4 − (vε )4 dx
2 4 4 Ω
a(S + O(ε2 )) 2 (µ − bS 2 + O(ε2 )) 4
= t − t
2 4
(µ − bS 2 ) 4
≤ aSt2 − t.
8

10
Since µ > bS 2 , we can see that Iα (tvε ) → −∞ as t → ∞. Therefore we can choose
t0 > 0 such that ||u∗∗ || > R and Iα (u∗∗ ) < ρ, where u∗∗ = t0 vε . This completes the
proof of Lemma 3.2. 

Lemma 3.3. Assume a, b > 0, µ > bS 2 , 0 < γ < 21 , 2 + 2γ < β < 3, 0 < α < 1 and
0 < λ < λ̄ small enough. There exists a vε ∈ H01 (Ω) such that
a2 S 2 2
sup Iα (tvε ) < − Dλ 1+γ , (3.15)
t≥0 4(µ − bS 2 )
where D is as in Lemma 3.1.
2 2 1+γ 2 2 2
a S a S
Proof. Letting λ < ( 4(µ−bS 2 )D )
2 , we have
4(µ−bS 2 )
− Dλ 1+γ > 0 for any λ ∈
 1+γ

a2 S 2
0, ( 4(µ−bS 2 )D )
2 . Since Iα (tvε ) → −∞ as t → ∞, there exists tε > 0 such that
Iα (tε vε ) = maxt>0 Iα (tvε ). By Lemma 3.2, we obtain that Iα (tε vε ) ≥ ρ > 0. Therefore
by the continuity of Iα , there exist positive constants t1 and t2 such that 0 < t1 ≤
tε ≤ t2 . Set Iα (tε vε ) = A(tε vε ) − λB(tε vε ), where
a b µ
A(tε vε ) = t2ε kvε k2 + t4ε kvε k4 − t4ε ,
2 4 4
and
(tε vε + α)1−γ − α1−γ
Z
1
B(tε vε ) = dx.
1−γ Ω |x|β

First, we prove that there exists a constant C1 > 0( independent of λ, ε), such
a2 S 2 2
that A(tε vε ) ≤ 4(µ−bS 2 ) + C1 ε . Indeed, by (3.14), it holds that

a b µ
A(tε vε ) = t2ε kvε k2 + t4ε kvε k4 − t4ε
2 4 4
a(S + O(ε )) 2 (µ − bS 2 + O(ε2 )) 4
2
≤ tε − tε
2 4
a2 S 2 + O(ε2 )
≤
4(µ − bS 2 + O(ε2 ))
a2 S 2
≤ 2
+ O(ε2 )
4(µ − bS )
a2 S 2
≤ 2
+ C 1 ε2 . (3.16)
4(µ − bS )
Next, we prove that there exists a constant C2 > 0 (independent of λ, ε) such
10+2γ−3β
that B(tε vε ) ≥ C2 ε 4 . In fact, we have
(tε vε + α)1−γ − α1−γ
Z
1
B(tε vε ) ≥ dx
(1 − γ) |x|≤ε 43 |x|β
(tε vε )1−γ − α1−γ
Z
1
≥ dx
(1 − γ) |x|≤ε 43 |x|β
ε1−γ
Z Z
1
≥C 2 2 1−γ β
dx − C β
dx
|x|≤ε 4 (ε + |x| ) |x| |x|≤ε 4 |x|
3 3

11
 3
ε4
ε1−γ
Z Z
≥C 3 3(1−γ) 3β
dx − C r3−β dx
|x|≤ε 4 ε 2 ε 4 0
1−γ− 23 (1−γ)− 34 β−3 3− 34 β
≥Cε − Cε
10+2γ−3β 12−3β
≥Cε 4 − Cε 4

10+2γ−3β
≥C2 ε 4 , (3.17)
where C and C2 are positive constants and independent of λ, ε. Therefore, combining
(3.16) and (3.17), we have
Iα (tε vε ) = A(tε vε ) − λB(tε vε )
a2 S 2 10+2γ−3β
≤ 2
+ C1 ε2 − C2 λε 4 .
4(µ − bS )
 4(1+γ)
 3β−6(1+γ)
1
C2
Since 2 + 2γ < β < 3, let ε = λ 1+γ ,λ < C1 +D
, it holds that
10+2γ−3β 2 14+6γ−3β
C1 ε2 − C2 λε 4
 − C2 λ
= C1 λ 1+γ 4(1+γ)
6(1+γ)−3β
2

= λ 1+γ C1 − C2 λ 4(1+γ)

2
< −Dλ 1+γ .
Thus, we conclude that
Iα (tε vε ) = A(tε vε ) − λB(tε vε )
a2 S 2 2
6(1+γ)−3β
≤ + C 1 ε − C 2 λε 4(1+γ)
4(µ − bS 2 )
a2 S 2 2
≤ 2
− Dλ 1+γ ,
4(µ − bS )
( 4(1+γ)
)
  3β−6(1+γ) 2 2 1+γ
C2 a S
which implies that (3.15) holds when λ < λ̄ = min C1 +D
, ( 4(µ−bS 2 )D )
2 .

This completes the proof of Lemma 3.3. 

Theorem 3.4. Let a, b > 0 and 0 < γ < 12 . For any 0 < α < 1, 2 + 2γ < β < 3 and
all µ > bS 2 , there exists λ∗∗ > 0 such that 0 < λ < λ∗∗ , problem (3.1) has at least a
positive solution vα ∈ H01 (Ω) with Iα (vα ) > ρ (here ρ is as in Lemma 2.1).

Proof. Let λ∗∗ = min{λ∗ , λ̄}, Lemmas 3.1-3.3 hold for 0 < λ < λ∗∗ . Now we define
Γ := {γ ∈ C([0, 1], H01 (Ω)) : γ(0) = 0, γ(1) = u∗∗ } ,
cα := inf γ∈Γ maxt∈[0,1] Iα (γ(t)).
By Lemma 3.2 and applying the mountain-pass lema, there exists a sequence {vn } ⊂
H01 (Ω), such that
Iα (vn ) → cα > ρ, and Iα0 (vn ) → 0, (3.18)
Moreover, from Lemma 3.2 and Lemma 3.3, we obtain
a2 S 2 2
0 < ρ < cα ≤ max Iα (tu1 ) ≤ sup Iα (tu1 ) < 2
− Dλ 1+γ . (3.19)
t∈[0,1] t≥0 4(µ − bS )

12
 According to Lemma 3.1, we know that {vn } ⊂ H01 (Ω) has a convergent subsequence,
still denoted by {vn }, we may assume that vn → vα in H01 (Ω) as n → ∞. Conse-
quently, we have
Iα (vα ) = lim Iα (vn ) = cα > ρ > 0. (3.20)
n→∞
This implies that vα 6≡ 0. Moreover, from the continuity of Iα , we obtain that vα is a
solution of problem (3.1). Then vα satisfies (3.2). Therefore, taking the test function
ϕ = vα− in (3.2), it follows that kvα− k = 0. Thus, we have vα ≥ 0 and vα 6≡ 0. Hence,
vα is a positive solution of problem (3.1), by the strong maximum principle. The
proof is completed. 

Next, we shall prove that the limit point of a family of solutions {vα } of prob-
lem (3.1) is the second positive solution of the case with µ > bS 2 , 0 < γ < 21 and
2 + 2γ < β < 3.

Theorem 3.5. Assume a, b > 0, µ > bS 2 , 0 < λ < λ∗∗ (λ∗∗ is as in Theorem 3.4)
and 0 < γ < 21 , 2 + 2γ < β < 3. Then problem (1.1) has a positive solution v0
satisfying I(v0 ) > 0.

Proof. Noting that {vα } is a family of positive solutions of problem (3.1), we have
 
2 3 λ λ
−(a + bkvα k )∆vα = µvα + β ≥ min {1, µ} min 1, β .
|x| (vα + α)γ R0 2γ
Clearly, it follows that
 
1 λ
−∆vα ≥ min {1, µ} min 1, β .
a + bkvα k2 R0 2γ
Denoting by e the positive solution of the following problem
(
−∆u = 1, in Ω,
u = 0, on ∂Ω,
then e(x) > 0 in Ω. Therefore, by comparison principle we have
 
1 λ
vα ≥ min {1, µ} min 1, β e > 0. (3.21)
a + bkvα k2 R0 2γ
Furthermore, from (3.19), (3.4) and (2.2) we obtain
a2 S 2 2 1
− Dλ 1+γ > Iα (vα ) − hIα0 (vα ), vα i
4(µ − bS 2 ) 4 Z
(vα + α)1−γ − α1−γ
Z
a 2 λ vα λ
= kvα k + dx − dx
4 4 Ω |x|β (vα + α)γ 1−γ Ω |x|β
a λ 1−γ
≥ kvα k2 − C0 R03+γ−β S − 2 kvα k1−γ ,
4 1−γ
which implies that {vα } is bounded in H01 (Ω). Up to a subsequence, there exists
v0 ∈ H01 (Ω) such that

1
vα * v0 , weakly in H0 (Ω),

vα → v0 , strongly inLp (Ω)(1 ≤ p < 4), (3.22)

v (x) → v (x), a.e. in Ω.
α 0

13
Then, we prove that vα → v0 in H01 (Ω) as α → 0.
As usually, letting wα = vα − v0 , we claim that kwα k → 0 as α → 0. Otherwise,
there exists a subsequence of wα such that limα→0 kwα k = l > 0. Since |x|β (vvαα+α)γ ≤
1−γ
vα
|x|β
, by the dominated convergence theorem and (3.22), we obtain

v01−γ
Z Z
vα
lim dx = dx. (3.23)
α→0 Ω |x|β (vα + α)γ Ω |x|β
Therefore, by hIα0 (vα ), vα i = 0 and using Brézis-Lieb’s Lemma, it holds that

akwα k2 + akv0 k2 + bkwα k4 + bkv0 k4 + 2bkwα k2 kv0 k2

Z Z Z 1−γ
4 4 v0 (3.24)
−µ wα dx − µ v0 dx − λ β
dx = o(1).
Ω Ω Ω |x|

From (3.21), passing to the limit as α → 0, we have v0 (x) > 0. Thus letting α → 0,
vα * v0 in (3.2) and taking the test function ϕ = φ ∈ H01 (Ω) ∩ C0 (Ω) (C0 (Ω) is the
subset of C(Ω) consisting of functions with compact support in Ω) we have
Z Z Z −γ
2 2 3 v0
(a + bl + bkv0 k ) (∇v0 , ∇φ)dx = µ v0 φdx + λ β
φdx. (3.25)
Ω Ω Ω |x|

We show that (3.25) holds for any φ ∈ H01 (Ω). Indeed, since H01 (Ω) ∩ C0 (Ω) is dense
in H01 (Ω), then for any φ ∈ H01 (Ω), there exists a sequence {φn } ⊂ H01 (Ω) ∩ C0 (Ω)
such that φn → φ, as n → ∞. For n, m ∈ N + large enough, replacing φ with φn − φm
in (3.25), we obtain
Z Z
(a + bl + bkv0 k ) (∇v0 , ∇|φn − φm |)dx = µ v03 |φn − φm |dx
2 2
Ω ΩZ
(3.26)
v0−γ
+λ β
|φn − φm |dx.
Ω |x|

On one hand, since φn → φ, from (3.26) we can infer that { |x|φβnvγ } is a Cauchy
0
φn
sequence in L1 (Ω), hence there exists ν ∈ L1 (Ω) satisfying |x|β v0γ
→ ν(x), which means
φn
that |x|β v0γ
→ ν(x) in measure. By Riesz’s Theorem, { |x|φβnvγ } has a subsequence, still
0
denoted by { |x|φβnvγ }, such that
0

φn
→ ν(x), a.e. x ∈ Ω. (3.27)
|x|β v0γ

On the other hand, with |x|φβnvγ → |x|φβ vγ a.e. in Ω, thus from (3.27) we get |x|φβ vγ = ν(x).
0 0 0
So Ω |x|φβnvγ dx → Ω |x|φβ vγ dx as n → ∞. Then, taking the test function φ = φn in
R R
0 0
(3.25) and passing to the limit as n → ∞, then we deduce that (3.25) holds for
φ ∈ H01 (Ω).
In particular, take φ = v0 as the test function in (3.25), we have
Z Z 1−γ
2 4 2 2 4 v0
akv0 k + bkv0 k + bl kv0 k − µ v0 dx − λ β
dx = 0. (3.28)
Ω Ω |x|

14
From (3.24) and (3.28), one gets
Z
2 4 2 2
akwα k + bkwα k + bkwα k kv0 k − µ wα4 dx = o(1). (3.29)
Ω

Passing to the limit as α → 0 in (3.29) and by (1.3), we obtain

al2 + bl4 + bl2 kv0 k2 ≤ µS −2 l4 .
This implies that
(a + b||v0 ||2 )S 2
l2 ≥ . (3.30)
µ − bS 2
Furthermore, applying Brézis-Lieb’s Lemma and by (3.29), we infer
a b
I(v0 ) = Iα (vα ) − l2 − l2 kv0 |2 + o(1). (3.31)
4 4
On one hand, combining (3.28) and (2.2), using the Young inequality, we deduce
  Z 1−γ
a 2 1 1 v0 b
I(v0 ) ≥ kv0 k − − λ β
dx − l2 kv0 k2
4 1−γ 4 Ω |x| 4
Z 1−γ
a λ v0 b
≥ kv0 k2 − β
dx − l2 kv0 k2
4 1 − γ Ω |x| 4
a λ 1−γ b
≥ kv0 k2 − C0 R03+γ−β S − 2 kv0 k1−γ − l2 kv0 k2
4 1−γ 4
2
! 1+γ
a(1 + γ) 2C0 R03+γ−β 2 b
≥− 1−γ λ 1+γ − l2 kv0 k2
4(1 − γ) aS 2 4
2 b
= − Dλ 1+γ − l2 kv0 k2 . (3.32)
4
2 2 2
a S
On the other hand, since Iα (vα ) < 4(µ−bS 2 ) − Dλ
1+γ , thus, by (3.31) and (3.30), we

obtain
a2 S 2 2 a b
I(v0 ) < 2
− Dλ 1+γ − l2 − l2 kv0 ||2
4(µ − bS ) 4 4
a2 S 2 2 a (a + b||v0 ||2 )S 2 b
≤ 2
− Dλ 1+γ −
2
− l2 kv0 k2
4(µ − bS ) 4 µ − bS 4
a2 S 2 2 a 2 2
S b
≤ − Dλ 1+γ − − l2 kv0 k2
4(µ − bS 2 ) 4(µ − bS 2 ) 4
2 b
= −Dλ 1+γ − l2 kv0 k2 ,
4
which contradicts the inequality (3.32). Consequently we know that l = 0, this
implies vα → v0 in H01 (Ω) as α → 0, and combining (3.25), we obtain that v0 is
a positive solution of problem (1.1). Moreover, from Theorem 3.4 it follows that
I(v0 ) = limα→0 Iα (vα ) > ρ > 0. This completes the proof of Theorem 3.5. 

Acknowledgements

The authors express their gratitude to the reviewers for careful reading and
helpful suggestions which led to an improvement of the original manuscript.

15
 References
[1] C.O. Alves, F.J.S.A. Corrêa, G.M. Figueiredo, On a class of nonlocal elliptic problems
with critical growth, Differential Equation and Applications 23 (2010) 409–417.
[2] C.O. Alves, F.J.S.A. Corrêa, T.F. Ma, Positive solutions for a quasilinear elliptic
equation of Kirchhoff type, Comput. Math. Appl. 49 (2005) 85–93.
[3] A. Ambrosetti, P.H. Rabinowitz, Dual variational methods in critical point theory
and applications, J. Funct. Anal. 14 (1973) 349–381.
[4] H. Brézis, L. Nirenberg, Positive solutions of nonlinear elliptic equations involving
critical Sobolev exponents, Comm. Pure Appl. Math. 36 (1983) 437–477.
[5] B. Cheng, New existence and multiplicity of nontrivial solutions for nonlocal elliptic
Kirchhoff type problems, J. Math. Annal. Appl. 394 (2012) 488–495.
[6] B. Cheng, X. Wu, Existence results of positive solutions of Kirchhoff type problems,
Nonlinear Anal. 71 (2009) 4883–4892.
[7] G.M. Figueiredo, Existence of a positive for a Kirchhoff problem type with critical
growth via truncation argument, J. Math. Anal. Appl. 401 (2013) 706–713.
[8] A. Hamydy, M. Massar, N. Tsouli, Existence of solution for p-Kirchhoff type problems
with critical exponents, Electronic J. Differential Equations 105 (2011) 1–8.
[9] J. Jin, X. Wu, Infinitely many radial solutions for Kirchhoff-type problems in RN , J.
Math. Annal. Appl. 369 (2010) 564–574.
[10] G. Kirchhoff, Mechanik, Teubner, Leipzig, 1883.
[11] C.-Y. Lei, J.-F. Liao, C.-L. Tang, Multiple positive solutions for Kirchhoff type of
problems with singularity and critical exponents, J. Math. Anal. Appl. 421 (2015)
521–538.
[12] J. Liu, J.-F. Liao, C.-L. Tang, Positive solutions for Kirchhoff-type equations with
critical exponent in RN , J. Math. Anal. Appl. 429 (2015) 1153–1172.
[13] J.-F. Liao, P. Zhang, J. Liu, C.-L. Tang, Existence and multiplicity of positive so-
lutions for a class of Kirchhoff type problems with singularity, J. Math. Anal. Appl.
430 (2015) 1124–1148.
[14] J.L. Lions, On some questions in boundary value problems of mathematical physics,
in: Contemporary Developments in Continuum Mechanics and Partial Differential
Equations, Proc. Internat. Sympos. Inst. Mat, Univ. Fed. Rio de Janeiro, 1997, in:
North-Holland Math. Stud. vol. 30, North-Holland, Amsterdam, 1978, pp. 284–346.
[15] S. Liang, S. Shi, Soliton solutions to Kirchhoff type problems involving the critical
growth in RN , Nonlinear Anal. 81 (2013) 31–41.
[16] S. Liang, J. Zhang, Existence of solutions for Kirchhoff type problems with critical
growth in R3 , Nonlinear Anal. Real World Appl. 17 (2014) 126–136.
[17] X. Liu, Y. Sun, Multiple positive solutions for Kirchhoff type problems with singu-
larity, Commun. Pure Appl. Anal. 12 (2013) 721–733.

16
 [18] Y. Li, F. Li, J. Shi, Existence of a positive solution to Kirchhoff type problems
without compactness conditions, J. Differential Equations 253 (2012) 2285–2294.
[19] A. Mao, S. Luan, Sign-changing solutions of a class of nonlocal quasilinear elliptic
boundary value problems, J. Math. Anal. Appl. 383 (2011) 239–243.
[20] A. Mao, Z. Zhang, Sign-changing and multiple solutions of Kirchhoff type problems
without the P.S. condition, Nonlinear Anal. 70 (2009) 1275–1287.
[21] T.F. Ma, J.E. M. Rivera, Positive solutions for a nonlinear nonlocal elliptic trans-
mission problem, Appl. Math. Lett. 16 (2003) 243–248.
[22] D. Naimen, The critical problem of Kirchhoff type elliptic equations in dimension
four, J. Differential Equations 257 (2014) 1168–1193.
[23] J. Nie, X. Wu, Existence and multiplicity of non-trivial solutions for Schrödinger-
Kirchhoff-type equations with radial potential, Nonlinear Anal. 75 (2012) 3470–3479.
[24] J. Sun, C. Tang, Existence and multipicity of solutions for Kirchhoff type equations,
Nonlinear Anal. 74 (2011) 1212–1222.
[25] G. Talenti, Best constant in Sobolev inequality, Ann. Mat. Pura Appl. 110(1976)
353–372.
[26] L. Wei, X. He, Multiplicity of high energy solutions for superlinear Kirchhoff equa-
tions, J. Appl. Math. Comput. 39 (2012) 473–487.
[27] Q. Xie, X. Wu, C. Tang, Existence and multiplicity of solutions for Kirchhoff type
problem with critical exponent, Commun. Pure Appl. Anal. 12 (2013) 2773–2786.
[28] Y. Yang, J. Zhang, Positive and negative solutions of a class of nonlocal problems,
Nonlinear Anal. 73 (2010) 25–30.
[29] Z. Zhang, K. Perera, Sign changing solutions of Kirchhoff type problems via invariant
sets of descent flow, J. Math. Anal. Appl. 317 (2006) 456–463.

17

View publication stats