Next Article in Journal
Applications of Riccati–Bernoulli and Bäcklund Methods to the Kuralay-II System in Nonlinear Sciences
Previous Article in Journal
New Insights on Keller–Osserman Conditions for Semilinear Systems
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Quasilinear Fractional Neumann Problems

by
Dimitri Mugnai
1,*,† and
Edoardo Proietti Lippi
2,†
1
Department of Ecological and Biological Sciences, University of Tuscia, Largo dell’Università, 01100 Viterbo, Italy
2
Department of Mathematics and Statistics, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia
*
Author to whom correspondence should be addressed.
These authors contributed equally to this work.
Mathematics 2025, 13(1), 85; https://doi.org/10.3390/math13010085
Submission received: 8 November 2024 / Revised: 24 December 2024 / Accepted: 27 December 2024 / Published: 29 December 2024
(This article belongs to the Section C1: Difference and Differential Equations)

Abstract

:
We study an elliptic quasilinear fractional problem with fractional Neumann boundary conditions, proving an existence and multiplicity result without assuming the classical Ambrosetti–Rabinowitz condition. Improving previous results, we also provide the weak formulation of solutions without regularity assumptions and we provide an example, even in the linear case, for which no regularity can indeed be assumed.

1. Introduction

The aim of this paper is to study quasilinear problems driven by mixed operators with fractional Neumann boundary conditions. More precisely, we consider the operator
L : = α Δ p + β ( Δ ) p s .
Here, α , β > 0 , s ( 0 , 1 ) , p > 1 , Δ p is the classical p Laplacian, acting as Δ p u : = div ( | D u | p 2 D u ) , and ( Δ ) p s is the fractional p-Laplacian, namely
( Δ ) p s u ( x ) = C N , s , p P V R N | u ( x ) u ( y ) | p 2 u ( x ) u ( y ) | x y | N + p s d y .
Here, P V stands for the Cauchy principal value and the constant C N , s , p , defined as
C N , s , p : = s 2 2 s 1 Γ p s + p + N 2 2 π N / 2 Γ ( 1 s ) ,
is the usual normalization constant for ( Δ ) p s (see [1] for more details), but its value will not play a role in our analysis.
The operator in (1) acts on a bounded open subset Ω R N with a smooth boundary of class C 1 .
Mixed operators of the form (1) have raised increasing interest in recent years, mainly in the Hilbert setting (namely, when p = 2 ), but the quasilinear case has also been the object of several results about existence, multiplicity, and qualitative properties of solutions, both in the elliptic and in the parabolic case. The list of references being huge, we just quote some very recent ones and their related bibliography, [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16]. We also refer to [17], where a more general superposition of local and nonlocal operators is considered.
As for the boundary conditions, we combine the operator in (1) with the so-called ( α , β ) -Neumann conditions, introduced in [18] for p = 2 and in [19] for general p. These conditions are made of two contributions. The first one corresponds to the local part and is defined on Ω , while the second corresponds to the nonlocal part and is defined on R N Ω ¯ , namely
( α , β ) - Neumann   conditions: : N s , p u = 0 i n R N Ω ¯ , | D u | p 2 u ν = 0 o n Ω .
Here,
N s , p u ( x ) : = C N , s , p Ω | u ( x ) u ( y ) | p 2 u ( x ) u ( y ) | x y | N + p s d y , x R N Ω ¯ ,
is the nonlocal normal p derivative, or fractional p Neumann boundary condition and describes the natural Neumann boundary condition in the presence of the fractional p Laplacian. It was introduced in [20,21] as an extension of the notion of nonlocal normal derivative introduced in [22] for the fractional Laplacian, i.e., for p = 2 . Nonlocal Neumann conditions have been treated recently, for instance in [23,24,25,26,27,28,29,30,31], as they imply a number of mathematical advantages that different conditions do not permit, see [22].
From (3) it is clear that the name ( α , β ) -Neumann conditions is related to the fact that they consider both the local part and the nonlocal part of the operator in (1).
From now on, for the sake of simplicity, we will set C N , s , p = 1 .
In this framework, we are interested in dealing with problems of type
α Δ p u + β ( Δ ) p s u = f ( x , u ) in   Ω , with   ( α , β ) - homogeneous   Neumann   conditions .
More precisely, we start in Section 2, giving the definition of the suitable functional space to study problems such as (5). We also recall the nonlocal counterpart of the divergence theorem and the integration by parts formula stated in [20,21]. However, we give these results in a more general setting, and we also give an example to better explain why such a generalization is needed. In particular, we consider the linear case in dimension 1, showing that, in some cases, solutions cannot be differentiable in the whole space, see Example 1. Moreover, we give the definition of weak solutions and some properties that they satisfy.
We give our main results in Section 3 and Section 4. In the former, under a suitable hypothesis, we give an L estimate for weak solutions; namely, we prove that they are bounded in the whole of R N . In this case, the proof mainly relies on a suitable choice of test functions and an iteration argument.
In Section 4, we deal with a superlinear problem in the presence of a source term, which does not satisfy the so-called Ambrosetti–Rabinowitz condition. In particular, we prove the existence of two nontrivial weak solutions, which do not change signs. Here, the strategy is to apply a Mountain Pass argument to suitable truncated functionals. Moreover, the absence of the Ambrosetti–Rabinowitz condition makes it harder to prove a compactness property for such functionals, as they do not satisfy the Palais–Smale condition. To overcome this difficulty, we prove that these functionals satisfy the Cerami condition. After that, by using the Mountain Pass Theorem with the Cerami condition, we prove the existence of one nontrivial solution for problem (5).
As far as we know, these are the first results in the quasilinear case for operators of this type.
Finally, in Appendix A we give the details on the results stated in Example 1 of Section 2.

2. Functional Setting

In this section, we give the correct framework in order to study the operator in (1) with ( α , β ) -Neumann conditions. First, we introduce the norm
u p : = α Ω | D u | p d x + β 2 Q | u ( x ) u ( y ) | p | x y | N + p s d x d y + Ω | u ( x ) | p d x ,
where Q = ( Ω × Ω ) ( Ω × ( R N Ω ) ) ( ( R N Ω ) × Ω ) . Thus, we can define the space
X α , β : = { u : R N R : u < + } .
We observe that, setting
X = u : R N R : u L p ( Ω ) and Q | u ( x ) u ( y ) | p | x y | N + p s d x d y < + ,
we can write X α , β = W 1 , p ( Ω ) X . Then, it is not hard to see that X α , β is an uniformly convex Banach space. Moreover, if
p = p N N p if p < N , + if p N ,
we have that the embedding of X α , β in L r ( Ω ) is compact for every r [ 1 , p ) .
Now, we recall the analogous of the divergence theorem and of the integration by parts formula for the nonlocal case:
Proposition 1.
Let u : R N R be such that u C 2 ( Ω ) and
( Δ ) p s u L 1 ( Ω ) .
Assume that the function
Ω × ( R N Ω ) ( x , y ) | u ( x ) u ( y ) | p 1 | x y | N + p s b e l o n g s t o L 1 ( Ω × ( R N Ω ) ) .
Then,
Ω ( Δ ) p s u d x = R N Ω N s , p u d x .
Proposition 2.
Let u , v : R N R be such that u , v C 2 ( Ω ) and
Q ( x , y ) | u ( x ) u ( y ) | p 1 ( v ( x ) v ( y ) ) | x y | N + p s b e l o n g s t o L 1 ( Q ) .
Assume, also, that
( Δ ) p s u v L 1 ( Ω )
and that the function
( R N Ω ) × Ω ( x , y ) | u ( x ) u ( y ) | p 1 v ( x ) | x y | N + p s b e l o n g s t o L 1 ( ( R N Ω ) × Ω ) .
Then,
1 2 R 2 N ( C Ω ) 2 | u ( x ) u ( y ) | p 2 ( u ( x ) u ( y ) ) ( v ( x ) v ( y ) ) | x y | N + p s d x d y = Ω v ( Δ ) p s u d x + R N Ω v N s , p u d x .
A few comments on Propositions 1 and 2 are mandatory. These results were first given in [20] (Theorem 6.3), where u and v are assumed to be in the Schwartz space S ( R N ) . They are also stated in [21] (Propositions 2.5 and 2.6) for any u and v bounded and of class C 2 ( R N ) .
However, these regularity assumptions on u and v may be too much as solutions of problem (5) may not even be of class C 1 ( R N ) as we will show in the forthcoming Example 1. For this reason, in a similar fashion to [32] (Lemmas 5.1 and 5.2), which cover the case p = 2 , we give Propositions 1 and 2 in a more general setting. We stress that our assumptions are enough to guarantee that the integrals in (6) and (7) are finite. Moreover, the proofs remain the same.
Next, we give an example of a solution for problem (5), which is not of class C 1 ( R N ) when s ( 0 , 1 / 2 ] . For more details, see Appendix A.
Example 1.
Let N = 1 , p = 2 and Ω = ( 1 , 1 ) . We start defining u : ( 1 , 1 ) R as
u ( x ) = x ( x 1 ) 2 ( x + 1 ) 2 .
Clearly,
lim x 1 + u ( x ) = lim x 1 u ( x ) = 0 .
Then, defining for every x > 1
g s ( x ) : = 1 1 y ( y 1 ) 2 ( y + 1 ) 2 ( x y ) 1 + 2 s d y 1 1 d y ( x y ) 1 + 2 s ,
we can extend u in R as
u ( x ) = g s ( x ) x < 1 x ( x 1 ) 2 ( x + 1 ) 2 1 x 1 g s ( x ) x > 1 .
We note that by extending u in this way, we have u C ( R ) and N s u ( x ) = 0 for any x R [ 1 , 1 ] . In addition,
lim x 1 + u ( x ) = lim x 1 u ( x ) = + s ( 0 , 1 2 ) , 4 3 s = 1 2 , 0 s ( 1 2 , 1 ) ,
so u C 1 ( R ) if s ( 0 , 1 / 2 ] .
On the other hand, computing u ( x ) and ( Δ ) s u ( x ) in ( 1 , 1 ) (see also [33] for more details on the computation of ( Δ ) s u ( x ) ), we can define
f s ( x ) : = u ( x ) + ( Δ ) s u ( x ) .
In this way, we have that u is a solution of
u ( x ) + ( Δ ) s u ( x ) = f s ( x ) x ( 1 , 1 ) , u ( x ) = 0 x { 1 , 1 } , N s u = 0 x R [ 1 , 1 ] ,
where f s L q for any q 1 (see Appendix A).
We also stress that, even if s ( 0 , 1 / 2 ] , u satisfies the hypotheses of Proposition 1. This is in agreement with the computation
1 1 ( Δ ) s u ( x ) d x = 0 .
The integration by parts formula in Proposition 2 leads to the definition of weak solutions, which shows that X α , β is the natural space for problems ruled by the operator in (1). We remark that we give the definition of weak solutions and the next two results in the case of non homogeneous boundary conditions. In this case, the functional space that we consider is
X α , β g , h : = u : R N R : u + Ω h u d σ + R N Ω ¯ g u d x R
Definition 1.
Let f L p ( Ω ) , g L 1 ( R N Ω ¯ ) and h L 1 ( Ω ) . We say that u X α , β g , h is a weak solution of
α Δ p u + β ( Δ ) p s u = f ( x ) i n   Ω , β N s , p u = g i n   R N Ω ¯ , | D u | p 2 u ν = h o n   Ω ,
whenever
α Ω D u · D v d x + β 2 R 2 N ( C Ω ) 2 J p ( u ( x ) u ( y ) ) ( v ( x ) v ( y ) ) | x y | N + p s d x d y = Ω f v d x + Ω h v d σ + R N Ω ¯ g v d x
for every v X α , β g , h , where
J p ( u ( x ) u ( y ) ) : = | u ( x ) u ( y ) | p 2 ( u ( x ) u ( y ) ) .
Now, we give a sort of maximum principle for weak solutions of (8).
Proposition 3.
Let f L p ( Ω ) , g L 1 ( R N Ω ) , and h L 1 ( Ω ) . Let u X α , β g , h be a weak solution of (8) with f 0 , g 0 and h 0 . Then, u is constant.
Proof. 
First, we notice that v 1 belongs to X α , β g , h . So, using it as a test function in (9), we obtain
0 Ω f d x = Ω h d σ R N Ω ¯ g d x 0 .
Hence, f = 0 a.e. in Ω and g = 0 a.e. in R N Ω . Now, taking v = u as a test function in (9), we obtain
R 2 N ( C Ω ) 2 | u ( x ) u ( y ) | p | x y | N + p s d x d y = 0 ,
so u must be constant.  □
The next result states that, if u is a weak solution of (8), then the nonlocal boundary condition in R N Ω ¯ is satisfied almost everywhere, as in related classical results for local operators.
Theorem 1.
Let u X α , β g , h be a weak solution of (8). Then, β N s , p u = g a.e. in R N Ω ¯ .
Proof. 
First, we take v X α , β g , h such that v 0 in Ω ¯ as a test function in (9), obtaining
R N Ω ¯ g v d x = β 2 Ω R N Ω ¯ J p ( u ( x ) u ( y ) ) v ( y ) | x y | N + p s d y d x + β 2 R N Ω ¯ Ω J p ( u ( x ) u ( y ) ) v ( x ) | x y | N + p s d y d x = β Ω R N Ω ¯ J p ( u ( x ) u ( y ) ) v ( y ) | x y | N + p s d y d x = β R N Ω ¯ v ( y ) Ω J p ( u ( x ) u ( y ) ) | x y | N + p s d x d y = β R N Ω ¯ v ( y ) N s , p u ( y ) d y .
Therefore,
R N Ω ¯ ( β N s , p u ( x ) g ( x ) ) v ( x ) d x = 0
for every v X α , β g , h , which is 0 in Ω . In particular, this is true for every v C c ( R N Ω ¯ ) , and, so, β N s , p u ( x ) = g ( x ) a.e. in R N Ω ¯ .  □

3. L Estimate

In this section, we give a boundedness result for weak solutions. First, it is useful to introduce the following notation for the norm in W 1 , p ( Ω ) , namely
u W 1 , p ( Ω ) : = α Ω | D u | p d x + Ω | u | p d x 1 / p .
Clearly, this norm is equivalent to the usual norm in W 1 , p ( Ω ) .
The main result of this section is stated as follows.
Theorem 2.
Let N 3 , f L q ( Ω ) with q > N / p and α β 0 . If u is a weak solution of
α Δ p u + β ( Δ ) p s u + | u | p 2 u = f ( x ) i n   Ω , w i t h   ( α , β ) N e u m a n n   c o n d i t i o n s ,
then, u L ( R N ) and
u L ( R N ) C ( f L q ( Ω ) + f L q ( Ω ) 1 p 1 )
for some constant C > 0 .
Proof. 
If the weak solution u is identically zero, we have nothing to prove. Otherwise, we take δ ( 0 , 1 ) and set
u ˜ = δ u u L p ( Ω ) + f L q ( Ω ) and f ˜ = δ u L p ( Ω ) + f L q ( Ω ) p 1 f ,
(if N p , we replace p with any number larger than p by the usual Sobolev embedding theorems, see e.g., [34]). With this definition, we find that u ˜ is a weak solution of
α Δ p u ˜ + β ( Δ ) p s u ˜ + | u ˜ | p 2 u ˜ = f ˜ in   Ω , with   ( α , β ) Neumann   conditions .
For every k N , we define c k : = 1 2 k ,
v k : = u ˜ c k , w k : = ( v k ) + : = max { v k , 0 } , U k : = w k L p ( Ω ) p .
From the Dominated Convergence Theorem, we have
lim k U k = lim k w k L p ( Ω ) p = ( u ˜ 1 ) + L p ( Ω ) p .
Moreover, for k = 0 , we have w 0 = ( v 0 ) + = u + ˜ , and, so,
U 0 = Ω w 0 p ( x ) d x p / p = Ω u ˜ + p ( x ) d x p / p u ˜ L p ( Ω ) p δ p .
Now, we can use w k + 1 as a test function in (13), obtaining
α Ω | D u ˜ | p 2 D u ˜ · D w k + 1 d x + β 2 R 2 N ( C Ω ) 2 J p ( u ˜ ( x ) u ˜ ( y ) ) ( w k + 1 ( x ) w k + 1 ( y ) ) | x y | N + p s d x d y + Ω | u ˜ | p 2 u ˜ w k + 1 d x = Ω f ˜ w k + 1 d x .
We note that, for a.e. x , y R N , simple algebraic reasoning shows that
| w k + 1 ( x ) w k + 1 ( y ) | p = | ( v k + 1 ) + ( x ) ( v k + 1 ) + ( y ) | p | ( v k + 1 ) + ( x ) ( v k + 1 ) + ( y ) | p 2 ( ( v k + 1 ) + ( x ) ( v k + 1 ) + ( y ) ) ( v k + 1 ( x ) v k + 1 ( y ) ) = | w k + 1 ( x ) w k + 1 ( y ) | p 2 ( w k + 1 ( x ) w k + 1 ( y ) ) ( u ˜ ( x ) u ˜ ( y ) ) ,
so that, being that w k + 1 = u ˜ c k + 1 where u ˜ > c k + 1 , we have
Q J p ( u ˜ ( x ) u ˜ ( y ) ) ( w k + 1 ( x ) w k + 1 ( y ) ) | x y | N + p s d x d y Q | w k + 1 ( x ) w k + 1 ( y ) | p | x y | N + p s d x d y 0 .
Moreover,
Ω | D u ˜ | p 2 D u ˜ · D w k + 1 d x = Ω { u ˜ > c k + 1 } | D u ˜ | p 2 D u ˜ · D v k + 1 d x = Ω | D w k + 1 | p d x
and
Ω | u ˜ | p 2 u ˜ w k + 1 d x = { u ˜ > c k + 1 } | u ˜ | p 2 u ˜ ( u ˜ c k + 1 ) d x Ω w k + 1 p d x .
By using (18)–(20) in (16), we obtain
α Ω | D w k + 1 | p d x + Ω w k + 1 p d x Ω f ˜ w k + 1 d x .
Then, by the Sobolev inequality,
U k + 1 = Ω w k + 1 p d x p / p C w k + 1 W 1 , p ( Ω ) p Ω w k + 1 f ˜ d x ,
for some C > 0 . Moreover, by definition, v k + 1 v k , and, so,
w k + 1 w k .
We note that
w k + 1 = ( u ˜ c k + 1 ) + = u ˜ 1 + 1 2 k + 1 + 1 2 k 1 2 k + = v k + 1 2 k + 1 1 2 k + ,
and, consequently,
{ w k + 1 > 0 } = { v k + 1 > 0 } = w k > 1 2 k + 1 .
We also note that
p p q 1 > p p N / p 1 = p 1 .
As a consequence,
τ : = p p p q 1 1 < p p 1 ,
and we also observe that
τ > p p 1 > 1 .
A simple computation shows that
1 p + 1 q + 1 τ = 1 .
With (25) in mind, we can use the generalized Hölder Inequality with exponents p , q, and τ , and, together with (23) and the definition of U k + 1 , we have
Ω w k + 1 | f ˜ | d x = Ω { w k + 1 > 0 } w k + 1 | f ˜ | d x f ˜ L q ( Ω ) w k + 1 L p ( Ω ) | Ω { w k + 1 > 0 } | 1 / τ f L q ( Ω ) 2 p w k L p ( Ω ) Ω w k > 1 2 k + 1 1 / τ f L q ( Ω ) 2 p U k 1 / p 2 p ( k + 1 ) Ω w k > 1 2 k + 1 w k p 1 / τ C 1 k U k 1 / p U k p / p τ ,
for some C 1 > 0 . We define the exponent
γ : = 1 p + p p τ
and observe that, from (24),
γ > 1 .
Now, from (21) and (26), we have
U k + 1 C 1 k U k γ .
By iterating, from (27), we find
U k + 1 C 1 ( k + 1 ) k 2 γ k U 0 γ k + 1 .
Then, if we take δ sufficiently small in (15), we can conclude that
lim k U k = 0 ,
and, recalling (14), we obtain
( u ˜ 1 ) + L p ( Ω ) p = 0 ,
hence, u ˜ 1 . So, from (12),
u ( x ) u L p ( Ω ) + f L q ( Ω ) δ
for every x Ω .
A similar argument can be made for u . Clearly, setting
u ¯ = δ u u L p ( Ω ) + f L q ( Ω ) and f ¯ = δ u L p ( Ω ) + f L q ( Ω ) p 1 f ,
it follows that u ¯ is a weak solution of
α Δ p u ¯ + β ( Δ ) p s u ¯ + | u ¯ | p 2 u ¯ = f ¯ in   Ω , with   ( α , β ) Neumann   conditions .
Reasoning as above, we can conclude that u ¯ 1 , hence
u ( x ) u L p ( Ω ) + f L q ( Ω ) δ
for every x Ω . This, together with (28), implies that
| u ( x ) | u L p ( Ω ) + f L q ( Ω ) δ
for every x Ω .
On the other hand, using u as a test function in (10) and recalling (25), we obtain
u W 1 , p ( Ω ) p = α Ω | D u | p d x + Ω | u | p d x α Ω | D u | p d x + β 2 Ω | u ( x ) u ( y ) | p | x y | N + p s d x d y + Ω | u | p d x = Ω u f d x u L p ( Ω ) f L q ( Ω ) | Ω | 1 / τ .
From this last inequality and the Sobolev inequality, we can deduce
u L p ( Ω ) p 1 C 2 f L q ( Ω ) ,
for some C 2 > 0 . Using this in (29), we obtain
u L ( Ω ) C 3 ( f L q ( Ω ) + f L q ( Ω ) 1 p 1 )
for some C 3 > 0 . From Theorem 1, we easily obtain N s , p u = 0 a.e. in R N Ω ¯ , which implies
u ( x ) = Ω | u ( x ) u ( y ) | p 2 u ( y ) | x y | N + p s d y Ω | u ( x ) u ( y ) | p 2 | x y | N + p s d y
so that
| u ( x ) | u L ( Ω ) , x R N Ω ¯ ,
see also [21]. This concludes the proof.  □

4. A Superlinear Problem Without the Ambrosetti–Rabinowitz Condition

In this section, we consider the problem
α Δ p u + β ( Δ ) p s u + | u | p 2 u = f ( x , u ) in   Ω , with   ( α , β ) Neumann   conditions ,
where f : Ω × R R is a Carathéodory function such that f ( x , 0 ) = 0 for almost every x Ω . In addition, we assume the following hypotheses taken from [21] as improvements of those in [35,36]:
( f 1 )
there exist a L q ( Ω ) , a 0 , with q ( ( p ) , p ) , c > 0 and r ( p , p ) , such that
| f ( x , t ) | a ( x ) + c | t | r 1
for a.e. x Ω and for all t R ;
( f 2 )
denoting F ( x , t ) = 0 t f ( x , σ ) d σ , we have
lim t ± F ( x , t ) | t | p = +
uniformly for a.e. x Ω ;
( f 3 )
if σ ( x , t ) = f ( x , t ) t p F ( x , t ) , then there exist θ > 1 and β L 1 ( Ω ) , β 0 , such that
σ ( x , t 1 ) θ σ ( x , t 2 ) + β
for a.e. x Ω and all 0 t 1 t 2 or t 2 t 1 0 ;
( f 4 )
lim t 0 f ( x , t ) | t | p 2 t = 0
uniformly for a.e. x Ω .
Definition 2.
Let u X α , β . With the same assumptions on f as above, we say that u is a weak solution of (32) if
α Ω | D u | p 2 D u · D v d x + β 2 Q J p ( u ( x ) u ( y ) ) ( v ( x ) v ( y ) ) | x y | N + p s d x d y + Ω | u | p 2 u v d x = Ω f ( x , u ) v d x
for every v X α , β .
Following this definition, we have that any critical point of the C 1 functional I : X α , β R , defined as
I ( u ) = 1 p u p Ω F ( x , u ) d x
is a weak solution of (32).
Proposition 4.
Setting A ( u ) = u p , the functional A : X α , β X α , β satisfies the ( S ) property, that is, for every sequence ( u n ) n such that u n u in X α , β as n and
lim n A ( u n ) , u n u = 0 ,
there holds
lim n u n = u .
Proof. 
Clearly, A is weakly lower semicontinuous in X α , β , so that
A ( u ) lim inf n A ( u n ) .
On the other hand, A is convex and, so,
A ( u ) A ( u n ) + A ( u n ) , u n u ,
which, passing to the limit, gives
A ( u ) lim sup n A ( u n ) .
Thus, combining (33) and (34), we obtain
A ( u ) = lim n A ( u n )
as desired.  □
We have the following result.
Theorem 3.
If hypotheses ( f 1 )–( f 4 ) hold, then problem (32) has two nontrivial constant sign solutions.
In order to prove Theorem 3, we introduce the functionals
I ± ( u ) = 1 p u p Ω F ( x , u ± ) d x ,
where u + : = max { u , 0 } and u : = max { u , 0 } denote the classical positive and negative parts of u, respectively.
Now, we want to prove that both I ± satisfy the Cerami condition, (C) for short, which states that any sequence ( u n ) n in X α , β such that ( I ± ( u n ) ) n is bounded and ( 1 + u n ) I ± ( u n ) 0 as n , admits a convergent subsequence.
Proposition 5.
Under the assumptions of Theorem 3, the functionals I ± satisfy the (C) condition.
Proof. 
We give the proof for I + , the proof for I being analogous.
Let ( u n ) n in X α , β be such that
| I + ( u n ) | M 1
for some M 1 > 0 and all n 1 , and
( 1 + u n ) I + ( u n ) 0   in   X α , β   as   n .
From (36), we have
| I + ( u n ) h | ε n h 1 + u n
for every h X α , β and with ε n 0 as n , that is
| α Ω | D u n | p 1 D u n · D h d x + β 2 Q J p ( u n ( x ) u n ( y ) ) ( h ( x ) h ( y ) ) | x y | N + p s d x d y + Ω | u n | p 2 u n h d x Ω f ( x , ( u n ) + ) h d x | ε n h 1 + u n .
Taking h = ( u n ) in (37), we obtain
| α Ω | D ( u n ) | p d x + β 2 Q J p ( u n ( x ) u n ( y ) ) ( ( u n ) ( x ) ( u n ) ( y ) ) | x y | N + p s d x d y + Ω | ( u n ) | p d x | ε n
Observing that, similarly to (17), we have
| ( u n ) ( x ) ( u n ) ( y ) | p J p ( u n ( x ) u n ( y ) ) ( ( u n ) ( x ) ( u n ) ( y ) ) ,
thus, from (38), we obtain
( u n ) p ε n ,
and, so,
( u n ) 0   in   X α , β   as   n .
Now, taking h = ( u n ) + in (37), we obtain
α Ω | D ( u n ) + | p d x β 2 Q J p ( u n ( x ) u n ( y ) ) ( ( u n ) + ( x ) ( u n ) + ( y ) ) | x y | N + p s d x d y Ω | ( u n ) + | p d x + Ω f ( x , ( u n ) + ) ( u n ) + d x ε n .
From (35), we have
α Ω | D u n | p d x + β 2 Q | u n ( x ) u n ( y ) | p | x y | N + p s d x d y + Ω | u n | p d x p Ω F ( x , ( u n ) + ) d x p M 1
for M 1 > 0 and n 1 , which, together with (38), gives
α Ω | D ( u n ) + | p d x + β 2 Q J p ( u n ( x ) u n ( y ) ) ( ( u n ) + ( x ) ( u n ) + ( y ) ) | x y | N + p s d x d y + Ω | ( u n ) + | p d x p Ω F ( x , ( u n ) + ) d x M 2
for some M 2 > 0 and all n 1 . Adding (41) to (42), we obtain
Ω f ( x , ( u n ) + ) ( u n ) + d x p Ω F ( x , ( u n ) + ) d x M 3
for some M 3 > 0 and all n 1 , that is
Ω σ ( x , ( u n ) + ) d x M 3 .
Now, we want to prove that ( ( u n ) + ) n is bounded in X α , β , and, for this, we argue by contradiction. Passing to a subsequence if necessary, we assume that ( u n ) + as n . Defining y n = ( u n ) + / ( u n ) + , we can assume that
y n y   in   X α , β   and   y n y   in   L q ( Ω )
for every q [ p , p ) and for some y 0 .
First, we treat the case y 0 . We define the set
Z ( y ) = { x Ω : y ( x ) = 0 } ,
so that | Ω Z ( y ) | > 0 and ( u n ) + for a.e. x Ω Z ( y ) as n . By hypothesis ( f 2 ), we have
lim n F ( x , ( u n ) + ( x ) ) ( u n ) + p = lim n F ( x , ( u n ) + ( x ) ) ( u n ) + ( x ) p y n ( x ) p = +
for almost every x Ω Z ( y ) . On the other hand, by Fatou’s Lemma
Ω lim inf n F ( x , ( u n ) + ( x ) ) ( u n ) + p d x lim inf n Ω F ( x , ( u n ) + ( x ) ) ( u n ) + p d x ,
which leads to
lim n Ω F ( x , ( u n ) + ( x ) ) ( u n ) + p d x = + .
From (35), we have
1 p u n p + Ω F ( x , ( u n ) + ( x ) ) d x M 1
for all n 1 .
Recalling that u n p 2 p 1 ( ( u n ) + p + ( u n ) p ) , from (40), we obtain
2 p 1 p ( u n ) + p + Ω F ( x , ( u n ) + ( x ) ) d x M 4
for some M 4 > 0 . Dividing by ( u n ) + p ,
2 p 1 p + Ω F ( x , ( u n ) + ( x ) ) ( u n ) + p d x M 4 ( u n ) + p .
Passing to the limit, we have
lim sup n Ω F ( x , ( u n ) + ( x ) ) ( u n ) + p d x 2 p 1 p ,
which is in contradiction with (45), and this concludes the case y 0 .
Now, we deal with the case y 0 . We consider the continuous functions ϕ n : [ 0 , 1 ] R , defined as ϕ n ( t ) = I + ( t ( u n ) + ) with t [ 0 , 1 ] and n 1 . So, we can define t n such that
ϕ n ( t n ) = max t [ 0 , 1 ] ϕ n ( t ) .
Now, if λ > 0 , we set v n : = ( p λ ) 1 p y n X α , β . From (44), v n 0 in L q ( Ω ) for all q [ 1 , p ) . Performing some integration, from ( f 1 ), we have
Ω F ( x , v n ( x ) ) d x Ω a ( x ) | v n ( x ) | d x + C Ω | v n ( x ) | r d x
for some C > 0 , which implies that
lim n Ω F ( x , v n ( x ) ) d x = 0 .
Since ( u n ) + , there exists n 0 1 such that ( p λ ) 1 p / ( u n ) + ( 0 , 1 ) for all n n 0 . Then, from (46),
ϕ n ( t n ) ϕ n ( p λ ) 1 p ( u n ) +
for all n n 0 . So,
I + ( t n ( u n ) + ) I + ( ( p λ ) 1 p y n ) = λ y n p Ω F ( x , v n ( x ) ) d x .
Then, (47) implies that
I + ( t n ( u n ) + ) λ y n p + o ( 1 ) ,
and, since λ is arbitrary, we have
lim n I + ( t n ( u n ) + ) = + .
We observe that 0 t n ( u n ) + ( u n ) + for all n 1 ; so, from ( f 3 ), we know that
Ω σ ( x , t n ( u n ) + ) d x θ σ ( x , ( u n ) + ) d x + β 1
for all n 1 . Clearly, I + ( 0 ) = 0 . In addition, by (38), we have
I + ( ( u n ) + ) = 1 p Ω | D u n | p d x 1 p Ω | D ( u n ) | p d x + 1 p Ω | u n | p d x 1 p Ω | ( u n ) | p d x Ω F ( x , ( ( u n ) + ) d x + 1 p { x , y Q : u ( x ) > 0 , u ( y ) > 0 } | u n ( x ) u n ( y ) | p | x y | N + p s d x d y 1 p { x , y Q : u ( x ) 0 , u ( y ) 0 } | ( u n ) ( x ) ( u n ) ( y ) | p | x y | N + p s d x d y 2 p { x , y Q : u ( x ) > 0 , u ( y ) 0 } | ( u n ) + ( x ) + ( u n ) ( y ) | p | x y | N + p s d x d y = I + ( u n ) 2 p { x , y Q : u ( x ) > 0 , u ( y ) 0 } | ( u n ) + ( x ) + ( u n ) ( y ) | p | x y | N + p s d x d y + o ( 1 ) I + ( u n ) + o ( 1 )
where o ( 1 ) 0 as n . By (35), we obtain that I + ( ( u n ) + ) M 7 for some M 7 > 0 and all n N . Together with (48), this implies that t n ( 0 , 1 ) for all n n 1 n 0 . Since t n is a maximum point, we have
0 = t n ϕ n ( t n ) = + β 2 Q J p ( t n u n ( x ) t n u n ( y ) ) ( t n ( u n ) + ( x ) t n ( u n ) + ( y ) ) | x y | N + p s d x d y + α Ω | D t n ( u n ) + | p d x + Ω | t n ( u n ) + | p d x Ω f ( x , t n ( u n ) + ) t n ( u n ) + d x .
As in (17), we have
| t n ( u n ) + ( x ) t n ( u n ) + ( y ) ) | p J p ( t n u n ( x ) t n u n ( y ) ) ( t n ( u n ) + ( x ) t n ( u n ) + ( y ) ) ,
so that
t n ( u n ) + p Ω f ( x , t n ( u n ) + ) t n ( u n ) + d x 0 .
Adding (50) to (49), we obtain
t n ( u n ) + p p Ω F ( x , t n ( u n ) + ) d x θ Ω σ ( x , ( u n ) + ) d x + β 1 ,
that is
p I + ( t n ( u n ) + ) θ Ω σ ( x , ( u n ) + ) d x + β 1 .
Hence, from (48), we obtain
lim n σ ( x , ( u n ) + ) d x = + .
By combining (43) and (51), we obtain a contradiction, which concludes the case y 0 .
In conclusion, we have proved that ( ( u n ) + ) n is bounded in X α , β , and from (40), we know that ( u n ) n is bounded in X α , β . Then, we can assume
u n u   in   X α , β   and   u n u   in   L q ( Ω )
with q [ 1 , p ) and for some u X α , β . Taking h = u n u in (37), we have
u n p α Ω | D u n | p 2 D u n · D u d x β 2 Q J p ( u n ( x ) u n ( y ) ) ( u ( x ) u ( y ) ) | x y | N + p s d x d y Ω | u n | p 2 u n u d x Ω f ( x , ( u n ) + ) ( u n u ) d x = ε n .
From ( f 1 ) and (52), we know that
lim n Ω | f ( x , ( u n ) + ) ( u n u ) | d x = 0 .
Passing to the limit in (53), we obtain
lim n ( u n p β 2 Q J p ( u n ( x ) u n ( y ) ) ( u ( x ) u ( y ) ) | x y | N + p s d x d y α Ω | D u n | p 2 D u n · D u d x Ω | u n | p 2 u n u d x ) = 0 .
The ( S ) property, see Proposition 4, implies that u n u ; so, from the uniform convexity of X α , β (as 1 < p < ), we know that u n u in X α , β as n . Then, I + satisfies the (C) condition, which concludes the proof.  □
We can now give the proof of Theorem 3.
Proof of Theorem 3.
We want to apply the Mountain Pass Theorem to I + . From Proposition 5, we know that I + satisfies the (C) condition, so we only have to verify the geometric conditions.
From ( f 1 ) and ( f 4 ), for every ε > 0 , there exists C ε such that
F ( x , t ) ε p | t | p + C ε | t | r
for a.e. x Ω and all t R . Then,
I + ( u ) = 1 p u p Ω F ( x , u + ) d x 1 p u p ε p u p p C ε u r r 1 ε C 1 p u p C 2 u r
for some C 1 , C 2 > 0 . From this, we obtain that, if u = ρ is small enough, then
inf u = ρ I + ( u ) > 0 .
Now, we take u X α , β with u > 0 and t > 0 ; then,
I + ( u ) = t p p u p Ω F ( x , t u ) d x = t p p u p t p Ω F ( x , t u ) | t u | p u p d x .
By Fatou’s Lemma,
Ω lim inf t F ( x , t u ) | t u | p u p d x lim inf t Ω F ( x , t u ) | t u | p u p d x ,
so, from ( f 2 ), we know that
lim t Ω F ( x , t u ) | t u | p u p d x = + .
Then,
lim t I + ( t u ) = ,
therefore, there exists e X α , β such that e > ρ and I + ( e ) < 0 .
Now, we can apply the Mountain Pass Theorem (see, e.g., [37]) to I + and obtain a nontrivial critical point u. In particular, we have
0 = α Ω | D u | p d x + β 2 Q J p ( u ( x ) u ( y ) ) ( u ( x ) u ( y ) ) | x y | N + p s d x d y + Ω | u | p d x Ω f ( x , u + ) u d x = α Ω | D u | p d x + β 2 Q J p ( u ( x ) u ( y ) ) ( u ( x ) u ( y ) ) | x y | N + p s d x d y + Ω | u | p d x .
By (39), we have
| u ( x ) u ( y ) | p J p ( u ( x ) u ( y ) ) ( u ( x ) u ( y ) ) ,
so that
0 u p ,
and, thus, u 0 . Then, I + ( u ) = I ( u ) , and, so, u 0 is a solution of (32). Arguing in the same way for I , we can find a nontrivial negative solution of (32).  □

Author Contributions

Conceptualization, D.M. and E.P.L.; Methodology, D.M. and E.P.L.; Validation, D.M. and E.P.L.; Investigation, D.M. and E.P.L.; Writing—original draft, D.M. and E.P.L.; Writing—review & editing, D.M. and E.P.L. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding authors.

Acknowledgments

D.M. is a member of GNAMPA (Gruppo Nazionale per l’Analisi Matematica, la Probabilità e le loro Applicazioni) of INdAM (Istituto Nazionale di Alta Matematica ’Francesco Severi’) and is supported by the FFABR “Fondo per il finanziamento delle attività base di ricerca” 2017, by the INdAM-GNAMPA Project 2023 “Variational and non-variational problems with lack of compactness”, and by the INdAM-GNAMPA Project 2024 “Nonlinear problems in local and nonlocal settings with applications”. E.P.L. is a member of GNAMPA (Gruppo Nazionale per l’Analisi Matematica, la Probabilità e le loro Applicazioni) of INdAM (Istituto Nazionale di Alta Matematica ‘Francesco Severi’) and is supported by the Australian Laureate Fellowship FL190100081.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A. Details of Example 1

In this appendix, we give more details on the computations of Example 1. Thus, as in Example 1, we let N = 1 , p = 2 , and Ω = ( 1 , 1 ) and define u : ( 1 , 1 ) R as
u ( x ) = x ( x 1 ) 2 ( x + 1 ) 2 .
Clearly,
lim x 1 + u ( x ) = lim x 1 u ( x ) = 0 .
Then, in agreement with (30), for every x > 1 , we define
g s ( x ) : = 1 1 y ( y 1 ) 2 ( y + 1 ) 2 ( x y ) 1 + 2 s d y 1 1 d y ( x y ) 1 + 2 s .
Let us compute the integrals above. First,
1 1 d y ( x y ) γ = 1 γ 1 ( x 1 ) 1 γ ( x + 1 ) 1 γ if   γ 1 log x + 1 x 1 if   γ = 1 .
In order to compute
1 1 y 5 2 y 3 + y ( x y ) 1 + 2 s d y
we use the identity
1 1 y n ( x y ) γ d y = 1 1 y n x n ( x y ) γ d y + x n 1 1 d y ( x y ) γ = k = 0 n 1 x k 1 1 y n k 1 ( x y ) γ 1 d y + x n 1 1 d y ( x y ) γ .
When s 1 / 2 , by using (A3) (multiple times if needed), we obtain
1 1 y ( x y ) 1 + 2 s d y = 1 1 d y ( x y ) 2 s + x 1 1 d y ( x y ) 1 + 2 s ,
1 1 y 3 ( x y ) 1 + 2 s d y = 1 1 d y ( x y ) 2 s 2 + 3 x 1 1 d y ( x y ) 2 s 1 3 x 2 1 1 d y ( x y ) 2 s + x 3 1 1 d y ( x y ) 1 + 2 s
and
1 1 y 5 ( x y ) 1 + 2 s d y = 1 1 d y ( x y ) 2 s 4 + 5 x 1 1 d y ( x y ) 2 s 3 10 x 2 1 1 d y ( x y ) 2 s 2 + 10 x 3 1 1 d y ( x y ) 2 s 1 5 x 4 1 1 d y ( x y ) 2 s + x 5 1 1 d y ( x y ) 1 + 2 s
Accordingly, by (A2), we can compute both the integrals in (A1) to obtain
g s ( x ) = 2 s ( x 1 ) 2 s ( x + 1 ) 2 s ( x + 1 ) 2 s ( x 1 ) 2 s [ 1 5 2 s ( x + 1 ) 5 2 s ( x 1 ) 5 2 s + 5 x 4 2 s ( x + 1 ) 4 2 s ( x 1 ) 4 2 s 10 x 2 2 3 2 s ( x + 1 ) 3 2 s ( x 1 ) 3 2 s + 10 x 3 6 x 2 2 s ( x + 1 ) 2 2 s ( x 1 ) 2 2 s 5 x 4 6 x 2 + 1 1 2 s ( x + 1 ) 1 2 s ( x 1 ) 1 2 s x 5 2 x 3 + x 2 s ( x + 1 ) 2 s ( x 1 ) 2 s ] ,
When s = 1 / 2 , by (A2) and (A3), we obtain
g 1 2 ( x ) : = 5 x 5 28 3 x 3 + 13 3 x 1 2 ( 5 x 6 11 x 4 + 7 x 2 1 ) log x + 1 x 1 .
Thus, we can extend u in R as
u ( x ) = g s ( x ) x < 1 x ( x 1 ) 2 ( x + 1 ) 2 1 x 1 g s ( x ) x > 1
so that, by construction, N s u ( x ) = 0 for any x R [ 1 , 1 ] .
We observe that, extending u in this way, we have u C ( R ) , which is in agreement with [22] (Prosition 5.2). Moreover, one can compute
lim | x | + g s ( x ) = 0 ,
which is in agreement with [22] (Proposition 3.13), that is, u ( x ) at infinity tends to its integral mean in ( 1 , 1 ) , namely
lim | x | + u ( x ) = 1 2 1 1 u ( x ) d x = 0 .
However, computing the first derivative of g s ( x ) , we see that
lim x 1 + u ( x ) = lim x 1 u ( x ) = + s ( 0 , 1 2 ) , 4 3 s = 1 2 , 0 s ( 1 2 , 1 ) ,
so u C 1 ( R ) if s ( 0 , 1 2 ] .
Now, we compute u ( x ) and ( Δ ) s u ( x ) for any x ( 1 , 1 ) . Indeed,
u ( x ) = 20 x 3 12 x
for any x ( 1 , 1 ) . Moreover,
( Δ ) s u ( x ) = P V + u ( x ) u ( y ) | x y | 1 + 2 s d y = 1 x ( x 1 ) 2 ( x + 1 ) 2 + g s ( y ) | x y | 1 + 2 s d y + P V 1 1 x ( x 1 ) 2 ( x + 1 ) 2 y ( y 1 ) 2 ( y + 1 ) 2 | x y | 1 + 2 s d y + 1 + x ( x 1 ) 2 ( x + 1 ) 2 g s ( y ) | x y | 1 + 2 s d y .
To compute
P V 1 1 x ( x 1 ) 2 ( x + 1 ) 2 y ( y 1 ) 2 ( y + 1 ) 2 | x y | 1 + 2 s d y
we take ε > 0 small enough and consider, separately, the integrals
1 x ε x ( x 1 ) 2 ( x + 1 ) 2 y ( y 1 ) 2 ( y + 1 ) 2 ( y x ) 1 + 2 s d y
and
x + ε 1 x ( x 1 ) 2 ( x + 1 ) 2 y ( y 1 ) 2 ( y + 1 ) 2 ( x y ) 1 + 2 s d y .
Exploiting (A2) and (A3), we can compute the integrals in (A6) and (A7). Then, adding them up and taking the limit for ε 0 , (A5) becomes
( Δ ) s u ( x ) = 1 5 2 s ( 1 + x ) 5 2 s ( 1 x ) 5 2 s 5 x 4 2 s ( 1 + x ) 4 2 s + ( 1 x ) 4 2 s + 10 x 2 2 3 2 s ( 1 + x ) 3 2 s ( 1 x ) 3 2 s 10 x 3 6 x 2 2 s ( 1 + x ) 2 2 s + ( 1 x ) 2 2 s + 5 x 4 6 x 2 + 1 1 2 s ( 1 + x ) 1 2 s ( 1 x ) 1 2 s + 1 + x ( x 1 ) 2 ( x + 1 ) 2 g s ( y ) | x y | 1 + 2 s d y + 1 x ( x 1 ) 2 ( x + 1 ) 2 + g s ( y ) | x y | 1 + 2 s d y
for s 1 / 2 , while
( Δ ) 1 2 u ( x ) = 8 x 3 + 20 3 x + ( 5 x 4 6 x 2 + 1 ) log 1 + x 1 x + 1 + x ( x 1 ) 2 ( x + 1 ) 2 g 1 2 ( y ) | x y | 2 d y + 1 x ( x 1 ) 2 ( x + 1 ) 2 + g 1 2 ( y ) | x y | 2 d y .
With (A8) and (A9) in mind, we can define, for any s ( 0 , 1 ) and x ( 1 , 1 ) ,
f s ( x ) : = 20 x 3 + 12 x + ( Δ ) s u ( x ) .
As a consequence, we have that u is a solution of
u ( x ) + ( Δ ) s u ( x ) = f s ( x ) x ( 1 , 1 ) , u ( x ) = 0 x { 1 , 1 } , N s u = 0 x R [ 1 , 1 ] .
Notice that, in this setting, such a problem corresponds to (5) with α = β = 1 .
Now, we want to check that f s ( x ) L q ( ( 1 , 1 ) ) for any q 1 . First, the non-integral part of f s ( x ) is bounded in ( 1 , 1 ) . Thus, we claim that
1 + x ( x 1 ) 2 ( x + 1 ) 2 g s ( y ) | x y | 1 + 2 s d y belongs to L q ( ( 1 , 1 ) ) for any q 1 .
To this aim, we fix ε > 0 . We recall that, since N s u = 0 in R [ 1 , 1 ] , we have
u L ( R ) u L ( ( 0 , 1 ) ) < + .
Thus, recalling (31) and (A1), we obtain
1 + ε + x ( x 1 ) 2 ( x + 1 ) 2 g s ( y ) | x y | 1 + 2 s d y 2 u L ( R ) 1 + ε + 1 ( y x ) 1 + 2 s d y = u L ( R ) s ( 1 x + ε ) 2 s u L ( R ) s ε 2 s .
Therefore,
1 1 1 + ε + x ( x 1 ) 2 ( x + 1 ) 2 g s ( y ) | x y | 1 + 2 s d y q d x 2 u L ( R ) s ε 2 s q < + .
On the other hand, if y ( 1 , 1 + ε ) , we have
| g s ( y ) | c ε ( y 1 ) 2 s
for some c ε > 0 . Indeed, for this estimate, it is enough to note that 5 y 4 6 y 2 + 1 = ( y 1 ) ( y + 1 ) ( 5 y 2 1 ) and y 5 2 y 3 + y = y ( y 1 ) 2 ( y + 1 ) 2 , so that all the functions in square brackets in (A4) are bounded when y ( 1 , 1 + ε ) . As a consequence, for every x ( 1 , 1 ) , we have
1 1 + ε x ( x 1 ) 2 ( x + 1 ) 2 g s ( y ) | x y | 1 + 2 s d y | x | ( x 1 ) 2 ( x + 1 ) 2 1 1 + ε 1 ( y x ) 1 + 2 s d y + c ε 1 1 + ε ( y 1 ) 2 s ( y x ) 1 + 2 s d y = : I 1 ( x ) + I 2 ( x ) .
First, we clearly have
I 1 ( x ) = | x | ( x 1 ) 2 ( x + 1 ) 2 2 s [ ( 1 x ) 2 s ( 1 x + ε ) 2 s ]
which is bounded in ( 1 , 1 ) , and, so, I 1 L q ( ( 1 , 1 ) ) for any q 1 . Moreover, since x ( 1 , 1 ) , we have
I 2 ( x ) c ε 1 1 + ε ( y x ) 2 s ( y x ) 1 + 2 s d y = c ε ( log ( 1 x + ε ) log ( 1 x ) ) .
Recalling the inequality
| a + b | q 2 q 1 ( | a | q + | b | q ) for any a , b R ,
we obtain that, for any q 1 ,
1 1 | I 2 ( x ) | q d x c 2 1 1 ( | log ( 1 x + ε ) | q + | log ( 1 x ) | q ) d x < + ,
for some c 2 = c 2 ( s , q , ε ) > 0 .
Combining the integrability of I 1 and I 2 with (A12), we obtain
1 1 1 1 + ε x ( x 1 ) 2 ( x + 1 ) 2 g s ( y ) | x y | 1 + 2 s d y q d x c 3 1 1 ( | I 1 ( x ) | q + | I 2 ( x ) | q ) d x < +
for every q 1 and for some c 3 = c 3 ( s , q , ε ) . Finally, (A11) and (A13), together with (A12), imply the claim in (A10).
A similar argument shows that
1 x ( x 1 ) 2 ( x + 1 ) 2 + g s ( y ) | x y | 1 + 2 s d y belongs to L q ( ( 1 , 1 ) ) for any q 1 .
This concludes the proof that f s ( x ) L q ( ( 1 , 1 ) ) for any q 1 .

References

  1. Warma, M. Local Lipschitz continuity of the inverse of the fractional p-Laplacian, Hölder type continuity and continuous dependence of solutions to associated parabolic equations on bounded domains. Nonlinear Anal. 2016, 135, 129–157. [Google Scholar] [CrossRef]
  2. Biagi, S.; Dipierro, S.; Valdinoci, E.; Vecchi, E. Semilinear elliptic equations involving mixed local and nonlocal operators. Proc. R. Soc. Edinb. A Math. 2021, 151, 1611–1641. [Google Scholar] [CrossRef]
  3. Biagi, S.; Dipierro, S.; Valdinoci, E.; Vecchi, E. Mixed local and nonlocal elliptic operators: Regularity and maximum principles. Commun. Partial. Differ. Equ. 2022, 47, 585–629. [Google Scholar] [CrossRef]
  4. Biagi, S.; Dipierro, S.; Valdinoci, E.; Vecchi, E. A Brezis-Nirenberg type result for mixed local and nonlocal operators. arXiv 2022, arXiv:2209.07502. [Google Scholar]
  5. Biagi, S.; Dipierro, S.; Valdinoci, E.; Vecchi, E. A Faber-Krahn inequality for mixed local and nonlocal operators. J. Anal. Math. 2023, 150, 405–448. [Google Scholar] [CrossRef]
  6. Biagi, S.; Dipierro, S.; Valdinoci, E.; Vecchi, E. A Hong-Krahn-Szegö inequality for mixed local and nonlocal operators. Math. Eng. 2023, 5, 1–25. [Google Scholar] [CrossRef]
  7. Biagi, S.; Mugnai, D.; Vecchi, E. A Brezis-Oswald approach for mixed local and nonlocal operators. Commun. Contemp. Math. 2024, 26, 2250057. [Google Scholar] [CrossRef]
  8. Biagi, S.; Punzo, F.; Vecchi, E. Global solutions to semilinear parabolic equations driven by mixed local-nonlocal operators. arXiv 2024, arXiv:2406.17731. [Google Scholar] [CrossRef]
  9. Biagi, S.; Vecchi, E. Multiplicity of positive solutions for mixed local-nonlocal singular critical problems. Calc. Var. Partial Differ. 2024, 63, 221. [Google Scholar] [CrossRef]
  10. De Filippis, C.; Mingione, G. Gradient regularity in mixed local and nonlocal problems. Math. Ann. 2024, 388, 261–328. [Google Scholar] [CrossRef]
  11. Dipierro, S.; Perera, K.; Sportelli, C.; Valdinoci, E. An existence theory for superposition operators of mixed order subject to jumping nonlinearities. Nonlinearity 2024, 37, 055018. [Google Scholar] [CrossRef]
  12. Garain, P.; Kinnunen, J. On the regularity theory for mixed local and nonlocal quasilinear elliptic equations. Trans. Am. Math. Soc. 2022, 375, 5393–5423. [Google Scholar] [CrossRef]
  13. Giacomoni, J.; Mukherjee, T.; Sharma, L. On an eigenvalue problem associated with mixed operators under mixed boundary conditions. arXiv 2024, arXiv:2411.16499v1. [Google Scholar]
  14. Perera, K.; Sportelli, C. A multiplicity result for critical elliptic problems involving differences of local and nonlocal operators. Topol. Methods Nonlinear Anal. 2024, 63, 717–731. [Google Scholar] [CrossRef]
  15. Sharma, L.; Mukherjee, T. On critical Ambrosetti–Prodi type problems involving mixed operator. J. Elliptic Parabol. Equ. 2024, 10, 1187–1216. [Google Scholar] [CrossRef]
  16. Su, X.; Valdinoci, E.; Wei, Y.; Zhang, J. Regularity results for solutions of mixed local and nonlocal elliptic equations. Math. Z. 2022, 302, 1855–1878. [Google Scholar] [CrossRef]
  17. Dipierro, S.; Perera, K.; Sportelli, C.; Valdinoci, E. An existence theory for nonlinear superposition operators of mixed fractional order. Commun. Contemp. Math. 2024. [Google Scholar] [CrossRef]
  18. Dipierro, S.; Proietti Lippi, E.; Valdinoci, E. (Non)local logistic equations with Neumann 406 conditions. Ann. Inst. Henri Poincaré C Anal. Non Linéaire 2023, 40, 1093–1166. [Google Scholar] [CrossRef] [PubMed]
  19. Mugnai, D.; Proietti Lippi, E. On mixed local-nonlocal operators with (α, β)-Neumann conditions. Rend. Circ. Mat. Palermo 2022, 71, 1035–1048. [Google Scholar] [CrossRef]
  20. Barrios, B.; Montoro, L.; Peral, I.; Soria, F. Neumann conditions for the higher order s-fractional Laplacian (−Δ)su with s > 1. Nonlinear Anal. 2020, 193, 111368. [Google Scholar] [CrossRef]
  21. Mugnai, D.; Proietti Lippi, E. Neumann fractional p-Laplacian: Eigenvalues and existence results. Nonlinear Anal. 2019, 188, 455–474. [Google Scholar] [CrossRef]
  22. Dipierro, S.; Ros-Oton, X.; Valdinoci, E. Nonlocal problems with Neumann boundary conditions. Rev. Mat. Iberoam. 2017, 33, 377–416. [Google Scholar] [CrossRef]
  23. Dipierro, S.; Valdinoci, E. Description of an ecological niche for a mixed local/nonlocal dispersal: An evolution equation and a new Neumann condition arising from the superposition of Brownian and Lévy processes. Phys. A 2021, 575, 126052. [Google Scholar] [CrossRef]
  24. Foghem, G.; Kassmann, M. A general framework for nonlocal Neumann problems. Commun. Math. Sci. 2024, 22, 15–66. [Google Scholar] [CrossRef]
  25. Grube, F.; Hensiek, T. Robust nonlocal trace spaces and Neumann problems. Nonlinear Anal. 2024, 241, 113481. [Google Scholar] [CrossRef]
  26. Kreisbeck, C.; Schönberger, H. Non-constant functions with zero nonlocal gradient and their role in nonlocal Neumann-type problems. Nonlinear Anal. 2024, 249, 113642. [Google Scholar] [CrossRef]
  27. Mugnai, D.; Proietti Lippi, E. Linking over cones for the Neumann fractional p-Laplacian. J. Differ. Equ. 2021, 271, 797–820. [Google Scholar] [CrossRef]
  28. Abatangelo, N. A remark on nonlocal Neumann conditions for the fractional Laplacian. Arch. Math. 2020, 114, 699–708. [Google Scholar] [CrossRef]
  29. Alves, C.O.; Ledesma, C.T. Multiplicity of solutions for a class of fractional elliptic problems with critical exponential growth and nonlocal Neumann condition. Commun. Pure Appl. Anal. 2021, 20, 2065–2100. [Google Scholar] [CrossRef]
  30. Audrito, A.; Felipe-Navarro, J.; Ros-Oton, X. The Neumann problem for the fractional Laplacian: Regularity up to the boundary. Ann. Sc. Norm. Super. Cl. Sci. 2023, 24, 155–1222. [Google Scholar] [CrossRef] [PubMed]
  31. Cinti, E.; Colasuonno, F. Existence and non-existence results for a semilinear fractional Neumann problem. NoDEA Nonlinear Differ. Equ. Appl. 2023, 30, 79. [Google Scholar] [CrossRef]
  32. Dipierro, S.; Proietti Lippi, E.; Valdinoci, E. The role of Allee effects for Gaussian and Lévy dispersals in an environmental niche. J. Math. Biol. 2024, 89, 19. [Google Scholar] [CrossRef] [PubMed]
  33. Dyda, B. Fractional calculus for power functions and eigenvalues of the fractional Laplacian. Fract. Calc. Appl. Anal. 2012, 15, 536–555. [Google Scholar] [CrossRef]
  34. Di Nezza, E.; Palatucci, G.; Valdinoci, E. Hitchhiker’s guide to the fractional Sobolev spaces. Bull. Sci. Math. 2012, 136, 521–573. [Google Scholar] [CrossRef]
  35. Jeanjean, L. On the existence of bounded Palais-Smale sequences and application to a Landesman-Lazer-type problem set on RN. Proc. Roy. Soc. Edinb. Sect. A 1999, 129, 787–809. [Google Scholar] [CrossRef]
  36. Mugnai, D.; Papageorgiou, N.S. Wang’s multiplicity result for superlinear (p, q)-equations without the Ambrosetti-Rabinowitz condition. Trans. Am. Math. Soc. 2014, 366, 4919–4937. [Google Scholar] [CrossRef]
  37. Motreanu, D.; Motreanu, V.V.; Papageorgiou, N.S. Topological and Variational Methods with Applications to Nonlinear Boundary Value Problems; Springer: New York, NY, USA, 2014. [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

Mugnai, D.; Proietti Lippi, E. Quasilinear Fractional Neumann Problems. Mathematics 2025, 13, 85. https://doi.org/10.3390/math13010085

AMA Style

Mugnai D, Proietti Lippi E. Quasilinear Fractional Neumann Problems. Mathematics. 2025; 13(1):85. https://doi.org/10.3390/math13010085

Chicago/Turabian Style

Mugnai, Dimitri, and Edoardo Proietti Lippi. 2025. "Quasilinear Fractional Neumann Problems" Mathematics 13, no. 1: 85. https://doi.org/10.3390/math13010085

APA Style

Mugnai, D., & Proietti Lippi, E. (2025). Quasilinear Fractional Neumann Problems. Mathematics, 13(1), 85. https://doi.org/10.3390/math13010085

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop