Journal of Applied Analysis and Computation Website:http://www.jaac-online.

com
Volume 15, Number 1, February 2025, 547–563 DOI:10.11948/20240147

EXISTENCE RESULTS FOR A NONLINEAR

ELLIPTIC PROBLEM DRIVEN BY A
NON-HOMOGENEOUS OPERATOR

John R. Graef1,† , Shapour Heidarkhani2 , Lingju Kong1 and

Shahin Moradi2

Abstract In this paper, the authors discuss the existence of at least one weak
solution and infinitely many weak solutions to a parametric nonlinear Dirich-
let problem involving a nonhomogeneous differential operator of p-Laplacian
type. Their approach is based on variational methods. Some recent results
are extended and improved, and an example is presented to demonstrate the
application of the main results.
Keywords Nonlinear elliptic problem, one weak solution, infinitely many
solutions, variational methods.
MSC(2010) 34B15, 35J15.

1. Introduction
Let Ω ⊂ RN (N ≥ 2) be a bounded domain with a smooth boundary ∂Ω and consider
the problem

 −divA(x, ∇u) = λf (x, u), in Ω,
(Pλf )
 u = 0, on ∂Ω,

where A : Ω × RN → RN is a function admitting a potential and satisfying some

natural conditions such that the differential operator divA(x, ∇u) includes the usual
p-Laplacian (p > 1). Here, λ is a positive parameter and f : Ω × R → R is a suitable
Carathéodory function.
Recently, many authors have applied variational methods to study the existence
of multiple solutions of nonlinear Dirichlet problems involving a nonhomogeneous
differential operator of p-Laplacian type and containing a parameter; we refer the
reader to [5, 7–11, 18, 19] and references cited therein as examples of such results.
Based on a recent abstract critical point theorem proved by Bonanno [3], Kristály
† The corresponding author.
1 Department of Mathematics, University of Tennessee at Chattanooga, Chat-
tanooga, TN 37403, USA
2 Department of Mathematics, Faculty of Sciences, Razi University, Kerman-

shah 67149, Iran

Email: John-Graef@utc.edu(J. R. Graef),
s.heidarkhani@razi.ac.ir(S. Heidarkhani), Lingju-Kong@utc.edu(L. Kong),
shahin.moradi86@yahoo.com(S. Moradi)
548 J. R. Graef, S. Heidarkhani. L. Kong & S. Moradi

et al. [18] established the existence of three weak solutions to the problem

 −div(a(x, ∇u)) = λf (x, u), in Ω,
 u = 0, on ∂Ω,

where N ≥ 2 and the nonlinearities a : Ω × RN → RN and f : Ω × R → R satisfy

certain structural conditions. Bonanno et al. [5] used variational methods to obtain
some new contributions on the problem (Pλf ).
In the present paper, motivated by the papers of Bonanno, D’Aguı̀, and Livrea
[5], Colasuonno, Pucci, and Varga [9], and Kristály, Lisei, and Varga [18], we study
the problem (Pλf ) in the case where A admits a potential A : Ω × RN → R such
that
(A) A = A(x, ξ) is a continuous function on Ω × RN with a continuous derivative
with respect to ξ, A = ∂ξ A, and:
(i) A(x, 0) = 0 and A(x, ξ) = A(x, −ξ) for every x ∈ Ω and ξ ∈ RN .
(ii) A(x, ·) is strictly convex in RN for all x ∈ Ω.
(iii) There exist constants a1 , a2 , with 0 < a1 ≤ a2 such that

A(x, ξ)ξ ≥ a1 |ξ|p and |A(x, ξ)| ≤ a2 |ξ|p−1 (1.1)

for every (x, ξ) ∈ Ω × RN .

Motivated by the above facts, in the present paper we use a smooth version of
Theorem 2.1 of [6] (that is sometimes viewed as a more precise version of Ricceri’s
Variational Principle [21]) to investigate the existence of at least one weak solution
and the existence of infinitely many weak solutions to the problem (Pλf ). In fact,
we shall study the existence of at least one non-trivial weak solution to (Pλf ) under
an assumption on the asymptotic behavior of the nonlinear function f at zero (see
Theorem 3.1 below). In addition, under suitable conditions on the oscillatory be-
havior of f at infinity, we discuss the existence of infinitely many weak solutions to
(Pλf ). We prove the existence of an interval on λ in which the problem (Pλf ) admits
a sequence of solutions that are unbounded in the space W 1,p (Ω) (see Theorem
4.1). Additionally, some consequences of Theorem 4.1 are presented. By replacing
the conditions at infinity on the nonlinear term, by a similar one at zero, we obtain
a sequence of pairwise distinct solutions strongly converging at zero (see Theorem
4.3).
We have organized the remainder of the paper as follows. In Section 2, we recall
some basic definitions and the tools to be used in our proofs. In Sections 3 and 4,
we state and prove the main results of the paper.

2. Preliminaries
The key argument used in proving our results is the following version of Ricceri’s
variational principle [21, Theorem 2.1] as given by Bonanno and Molica Bisci in [6].
Theorem 2.1. Let X be a reflexive real Banach space, let Φ, Ψ : X −→ R be two
Gâteaux differentiable functionals such that Φ is sequentially weakly lower semi-
continuous, strongly continuous, and coercive, and Ψ is sequentially weakly upper
Nonlinear elliptic problem 549

semicontinuous. For every r > inf X Φ, let

supu∈Φ−1 (−∞,r) Ψ(u) − Ψ(u)

φ(r) := inf ,
u∈Φ−1 (−∞,r) r − Φ(u)
θ := lim inf φ(r), and δ := lim inf φ(r).
r→+∞ r→(inf X Φ)+

Then:
 
1
(a) for every r > inf X Φ and every λ ∈ 0, , the restriction of the functional
φ(r)
Iλ = Φ − λΨ to Φ ((−∞, r)) admits a global minimum, which is a critical
−1

point (local minimum) of Iλ in X.

 
1
(b) If θ < +∞, then for each λ ∈ 0, , either
θ
(b1 ) Iλ possesses a global minimum,
or
(b2 ) there is a sequence {un } of critical points (local minima) of Iλ such that

lim Φ(un ) = +∞.

n→+∞
 
1
(c) If δ < +∞, then for each λ ∈ 0, , either
δ
(c1 ) there is a global minimum of Φ which is a local minimum of Iλ ,
or
(c2 ) there is a sequence of pairwise distinct critical points (local minima) of
Iλ that weakly converges to a global minimum of Φ.

We refer the interested reader to the papers [1, 2, 12, 14, 16] in which Theorem
2.1 has been successfully employed to prove the existence of at least one non-trivial
solution to boundary value problems, and to the papers [4, 13, 15, 17] in which
Theorem 2.1 was used to show the existence of infinitely many solutions.
Next, we give some pertinent definitions and notations. Throughout the paper,
Ω is a bounded domain in RN with 1 < p < N , W01,p (Ω) is the usual Sobolev space
endowed with the norm
∥u∥ = ∥∇u∥p ,
′
and W −1,p (Ω) is its dual space, where p1 + p1′ = 1 . It is well known that if
Np
1 < p < N and p∗ = , then there is a constant T = T (N, p) such that
N −p

∥u∥p∗ ≤ T ∥u∥ (2.1)

for every u ∈ W01,p (Ω). Such a constant has been sharply determined by Talenti
in [22] as given by the formula

1− p1 ( ) N1
Γ(1 + N2 )Γ(N )

−1 −1 p−1
T =π 2 N p , (2.2)
N −p Γ( Np )Γ(1 + N − Np )
550 J. R. Graef, S. Heidarkhani. L. Kong & S. Moradi

where Γ is the gamma function. Clearly, inequality (2.1), in conjunction with

Hölder’s inequality, implies that for every s ∈ [1, p∗ ],
(p∗ −s)
∥u∥s ≤ T |Ω| (p∗ s) ∥u∥ (2.3)

for all u ∈ W01,p (Ω), where |Ω| is the Lebesgue measure of Ω. Notice that the
embedding W01,p (Ω) ,→ Ls (Ω) is compact provided s ∈ [1, p∗ ).
Following [9], we will assume that A : Ω × RN → RN is a function admitting a
smooth potential A : Ω × RN → R as given in condition (A) above.
In [9], it has been explicitly observed that (A)(i) and (A)(iii) imply that

a1 |ξ|p ≤ pA(x, ξ) ≤ a2 |ξ|p (2.4)

for every (x, ξ) ∈ Ω × RN . Moreover, it is possible to obtain the following lemma.

Lemma 2.1. ( [9, Lemma 2.5]) Let A satisfy condition (A). Then the functional
Φ : W01,p (Ω) → R defined by
Z
Φ(u) = A(x, ∇u(x))dx (2.5)
Ω

is convex, weakly lower semicontinuous, and belongs to the class C 1 in W01,p (Ω)
with Z
Φ′ (u)(v) = A(x, ∇u(x))∇v(x)dx
Ω
′
for every u, v ∈ W01,p (Ω). Moreover, Φ′ : W01,p (Ω) → W −1,p (Ω) satisfies the
condition:

(S+ ) For every sequence {un } in W01,p (Ω) such that un ⇀ u weakly in W01,p (Ω) and
Z
lim sup A(x, ∇un ) · (∇un − ∇u)dx ≤ 0,
n→∞ Ω

then un → u strongly in W01,p (Ω).

Given a Carathéodory function f : Ω×R → R and a positive function a ∈ Lα (Ω)

with α > N/p and 1 < q ≤ p, we say that f is of type (Gf,a,q ) if it satisfies the
growth condition:

(Gf,a,q ) there exist positive constants M1 and M2 such that

|f (x, t)| ≤ a(x) M1 + M2 |t|q−1

(2.6)

for a.e. x ∈ Ω and all t ∈ R.

We will also need the following lemma to prove our main results.
Lemma 2.2. ( [9, Lemma 3.2], [5, Lemma 2.2]) Assume that f is of type (Gf,a,q )
Rt
and set F (x, t) = 0 f (x, s)ds. Then, the functional Ψ : W01,p (Ω) → R defined by
Z
Ψ(u) = F (x, u(x))dx (2.7)
Ω
Nonlinear elliptic problem 551

is in the class C 1 with

Z
′
Ψ (u)(v) = f (x, u(x))v(x)dx.
Ω

′
Moreover, the operator Ψ′ : W01,p (Ω) → W −1,p (Ω) is compact and sequentially
weakly continuous in W01,p (Ω).
Remark 2.1. As was pointed out in [5, Remark 2.1], Colasuonno, Pucci, and
Varga [9, Lemma 3.2], proved the compactness of Ψ′ for 1 < q < p. But the same
argument can be adopted in order to show that it is also true if q = p.
Next, we define what is meant by a weak solution of (Pλf ).

Definition 2.1. A function u ∈ W01,p (Ω) is a (weak) solution of the BVP (Pλf ) if
Z Z
A(x, ∇u(x))∇v(x)dx − λ f (x, u(x))v(x)dx = 0
Ω Ω

for every v ∈ W01,p (Ω).

Hence, in view of Lemmas 2.1 and 2.2, we consider the functional Iλ : W01,p (Ω) →
R defined by Iλ = Φ(u) − λΨ(u) for λ > 0 and u ∈ W01,p (Ω).
Lemma 2.3. A function u ∈ W01,p (Ω) is a critical point of Iλ in W01,p (Ω) if and
only if u is a solution of (Pλf ).

3. Existence of one solution

Here is our main result on the existence of one solution to the problem (Pλf ). In
1 1
what follows α′ is the conjugate of α, that is, + ′ = 1.
α α
Theorem 3.1. Assume that f satisfies (Gf,a,q ) and there are sets D ⊆ Ω and
G ⊂ D such that

ess inf x∈G F (x, ξ)

lim sup = +∞ (3.1)
ξ→ 0+ |ξ|p

and
ess inf x∈D F (x, ξ)
lim inf > −∞. (3.2)
ξ→0+ |ξ|p

Then, for each

 
 a1 
λ∈Λ=
0, sup (p∗ −α′ p)
  ,
θ>0 M2 q−p 
∥a∥α pT p |Ω| (p∗ α′ ) M1 θ1−p + θ
q

the problem (Pλf ) admits at least one non-trivial weak solution u ∈ W01,p (Ω).
552 J. R. Graef, S. Heidarkhani. L. Kong & S. Moradi

Proof. Our aim is to apply Theorem 2.1(a) to the problem (Pλf ). We introduce
the functionals Φ and Ψ as given in (2.5) and (2.7), respectively. Lemmas 2.1 and
2.2 establish that Φ and Ψ are of class C 1 , while condition (2.4) assures that
a1 a2
∥u∥p ≤ Φ(u) ≤ ∥u∥p (3.3)
p p

for every u ∈ W01,p (Ω). By using the first inequality in (3.3), it follows that

lim Φ(u) = +∞,

∥u∥→+∞

so Φ is coercive.
Choose λ such that
a1
0 < λ < sup (p∗ −α′ p)
 .
θ>0 M2 q−p
∥a∥α pT p |Ω| (p∗ α′ ) M1 θ1−p + θ
q

Then, there exists θ̄ > 0 such that

a1
λ< (p∗ −α′ p)
 .
M2 q−p
∥a∥α pT p |Ω| (p∗ α′ ) M1 θ̄1−p + θ̄
q

Set p
a1 |Ω| p∗ p
r= θ̄ .
pT p
Now since α > N/p implies that 1 < α′ < α′ q ≤ α′ p < p∗ , condition (Gf,a,q ),
Hölder’s inequality, and (2.3) imply
Z Z
M2
Ψ(u) ≤ M1 a(x)|u(x)|dx + a(x)|u(x)|q dx
Ω q Ω
M2
≤ M1 ∥a∥α ∥u∥α′ + ∥a∥α ∥u∥qα′ q
q
(p∗ −α′ ) M2 (p∗ −α′ q)
≤ M1 ∥a∥α T |Ω| (p∗ α′ ) ∥u∥ + ∥a∥α T q |Ω| (p∗ α′ ) ∥u∥q (3.4)
q

for every u ∈ W01,p (Ω). Hence, in view of (3.3),

(   p1 )
−1
n
1,p
o
1,p pr
Φ (−∞, r) = u ∈ W0 | Φ(u) < r ⊆ u ∈ W0 | ∥u∥ ≤ ,
a1

so (3.4) implies that

 p1  pq
(p∗ −α′ ) (p∗ −α′ q)
 
pr M2 pr
sup Ψ(u) ≤ M1 ∥a∥α T |Ω| (p∗ α′ ) + ∥a∥α T q |Ω| (p∗ α′ ) .
Φ(u)≤r a1 q a1

Since 0 ∈ Φ−1 (−∞, r) and Φ(0) = Ψ(0) = 0, we have

(supu∈Φ−1 (−∞,r) Ψ(u)) − Ψ(u)

φ(r) = inf
u∈Φ−1 (−∞,r) r − Φ(u)
Nonlinear elliptic problem 553

supu∈Φ−1 (−∞,r) Ψ(u)

≤
r
  p1   pq
(p∗ −α′ ) p 1−p M2 (p∗ −α′ q) p q−p
∗ α′ ) q ∗ α′ )
≤ M1 ∥a∥α T |Ω| (p r p + ∥a∥α T |Ω| (p r p
a1 q a1
 ! 1−p
p
! q−p
p

∗ −α′ p) p p
p (p T pr M2 T pr
= ∥a∥α T p |Ω| (p∗ α′ ) M1 p + p

a1 a1 |Ω| p∗ q a1 |Ω| p∗
p p (p(p∗ −α ′ p)
 
∗ α′ ) 1−p M2 q−p
= ∥a∥α T |Ω| M1 θ̄ + θ̄
a1 q
1
< .
λ
 
Hence, by Theorem 2.1, for every λ ∈ Λ ⊂ 0, φ(r) 1
the functional Iλ admits at
least one critical point (local minima) uλ ∈ Φ−1 (−∞, r).
To complete the proof we need to show that uλ is nontrivial. First, we show
that
Ψ(u)
lim sup = +∞. (3.5)
∥u∥→0 Φ(u)
+

By our assumptions at zero, namely, (3.1) and (3.2), we can choose a sequence
{ξn } ⊂ R+ converging to zero and two constants ζ and κ with ζ > 0 such that

ess inf x∈G F (x, ξn )

lim = +∞
n→+∞ |ξn |p
and
ess inf F (x, ξ) ≥ κ|ξ|p ,
x∈D

for ξ ∈ [0, ζ]. Now for the given G and D, take the function v ∈ W01,p (Ω) such that:
(j) v(x) ∈ [0, 1] for every x ∈ Ω,
(jj) v(x) = 1 for every x ∈ G,
(jjj) v(x) = 0 for every x ∈ Ω \ D.
Hence, fix Y > 0 and consider a positive real number η with

η meas (G) + κ D\G |v(x)|p dx

R
Y < a2 .
∥v∥p
p
Then, there is n0 ∈ N such that ξn < ζ and

ess inf F (x, ξn ) ≥ η|ξn |p

x∈G

for all n > n0 . Now, for every n > n0 , using the properties of the function v (that
is 0 ≤ ξn v(x) < ζ for large n), by (3.3),
Z Z
F (x, ξn )dx + F (x, ξn v(x)dx
Ψ(ξn v) G D\G
=
Φ(ξn v) Φ(ξn v)
554 J. R. Graef, S. Heidarkhani. L. Kong & S. Moradi
Z
η meas (G) + κ |v(x)|p dx
D\G
> a2
∥v∥p
p
>Y.

Since Y can be arbitrarily large,

Ψ(ξn v)
lim = +∞,
n→∞ Φ(ξn v)
from which (3.5) clearly follows.
Hence, there exists a sequence {wn } ⊂ W01,p (Ω) strongly converging to zero with
wn ∈ Φ−1 (−∞, r) and

Iλ (wn ) = Φ(wn ) − λΨ(wn ) < 0.

Since uλ is a global minimum of the restriction of Iλ to Φ−1 (−∞, r), we conclude

that
Iλ (uλ ) < 0, (3.6)
so uλ is not trivial. This completes the proof of the theorem.
Remark 3.1. If in addition to the hypotheses of Theorem 3.1, we ask that
a1
sup (p∗ −α′ p)
  > 1, (3.7)
θ>0 M2 q−p
∥a∥α pT p |Ω| (p∗ α′ ) M1 θ 1−p + θ
q

then the conclusion of the theorem holds with λ = 1, that is, for problem (Pλf )
without a parameter.
Next we present an example to illustrate Theorem 3.1.
Example 3.1. Consider the problem

 −divA(x, ∇u) = λf (u), in Ω,
(3.8)
 u = 0, on ∂Ω,

where Ω ⊂ R4 with meas (Ω) = 1 and f (ξ) = 2ξ + 1 for every ξ ∈ R. Condition

(2.6) is satisfied with a(t) ≡ 1, M1 = 1, M2 = 2, and q = 2. Note that p = 2 and
F (ξ) = ξ 2 + ξ for ξ ∈ R. All the conditions of Theorem 3.1 are satisfied, so there
exists ρ > 0 such that for any λ ∈ (0, ρ), problem (3.8) has at least one non-trivial
weak solution u ∈ W01,2 (Ω).
We now give some remarks on our results.
Remark 3.2. In Theorem 3.1 we searched for the critical points of the functional
Iλ naturally associated with the problem (Pλf ). We note that, in general, Iλ can
be unbounded from below in W01,p (Ω). For example, in the case where f (ξ) =
1 + |ξ|γ−p ξ for ξ ∈ R with γ > p, for any fixed u ∈ W01,p (Ω) \ {0} and ι ∈ R, we
obtain
ιγ
Z
a2
Iλ (ιu) =Φ(ιu) − λ F (ιu(x))dx ≤ ιp ∥u∥p − λι∥u∥p − λ ∥u∥γ → −∞
Ω p γ
Nonlinear elliptic problem 555

as ι → +∞. Therefore, condition (I2 ) in [20, Theorem 2.2] is not satisfied. Hence,
we can not use direct minimization to find the critical points of the functional Iλ .
Remark 3.3. We wish to point out that the energy functional Iλ associated with
the problem (Pλf ) may not be coercive. For example, if F (ξ) = |ξ|s with s ∈ (p, +∞)
for ξ ∈ R, for any fixed u ∈ W01,p (Ω) \ {0} and ι ∈ R we have
Z
a2
Iλ (ιu) =Φ(ιu) − λ F (ιu(x))dx ≤ ιp ∥u∥p − λιs ∥u∥s → −∞
Ω p
as ι → −∞.
Remark 3.4. From (3.6) we can easily see that the map

λ 7→ Iλ (uλ ) for λ ∈ (0, λ∗ ) (3.9)

is negative. Furthermore, we have

lim ∥uλ ∥ = 0.
λ→0+

To see this, by considering that Φ is coercive and that for λ ∈ (0, λ∗ ), the solution
uλ ∈ Φ−1 (−∞, r), we have that there exists a positive constant L such that ∥uλ ∥ ≤
L for every λ ∈ (0, λ∗ ). It is easy to see that, since f is bounded, an application of
Hölder’s inequality implies there exists a positive constant M such that
Z
f (x, uλ (x))uλ (x)dx ≤ M (3.10)
Ω

for λ ∈ (0, λ∗ ). Since uλ is a critical point of Iλ , we have Iλ′ (uλ )(v) = 0, for
v ∈ W01,p (Ω) and λ ∈ (0, λ∗ ). In particular Iλ′ (uλ )(uλ ) = 0, that is,
Z
′
Φ (uλ )(uλ ) = λ f (x, uλ (x))uλ (x)dx (3.11)
Ω

for λ ∈ (0, λ∗ ). Then, since

0 ≤ a1 ∥uλ ∥p ≤ Φ′ (uλ )(uλ ),

from (3.11),
Z
0 ≤ a1 ∥uλ ∥p ≤ Φ′ (uλ )(uλ ) ≤ λ f (x, uλ (x))uλ (x)dx
Ω

for λ ∈ (0, λ∗ ). Letting λ → 0+ , by (3.10), we have limλ→0+ ∥uλ ∥ = 0.

Finally, we show that the map

λ 7→ Iλ (uλ )

is strictly decreasing in (0, λ∗ ). To see this, observe that for any u ∈ W01,p (Ω),
 
Φ(u)
Iλ (u) = λ − Ψ(u) . (3.12)
λ
556 J. R. Graef, S. Heidarkhani. L. Kong & S. Moradi

Now, fix 0 < λ1 < λ2 < λ∗ and let uλ1 and uλ2 be global minimums of the functional
Iλi restricted to Φ(−∞, r) for i = 1, 2. Also, let
   
Φ(uλi ) Φ(v)
mλi = − Ψ(uλi ) = inf − Ψ(v) ,
λi v∈Φ−1 (−∞,r) λi

for i = 1, 2.
Since λ > 0, it is clear that (3.9) together with (3.12) imply

mλi < 0, for i = 1, 2. (3.13)

Moreover,
mλ2 ≤ mλ1 (3.14)
since 0 < λ1 < λ2 . Then, (3.12)–(3.14) and the fact that 0 < λ1 < λ2 , implies

Iλ2 (ūλ2 ) = λ2 mλ2 ≤ λ2 mλ1 < λ1 mλ1 = Iλ1 (ūλ1 ),

so that the map λ 7→ Iλ (uλ ) is strictly decreasing in λ ∈ (0, λ∗ ).

Remark 3.5. Observe that Theorem 3.1 above is a bifurcation result in the sense
that the pair (0, 0) belongs to the closure of the set
n o
(uλ , λ) ∈ W01,p (Ω) × (0, +∞) : uλ is a non-trivial weak solution of (Pλf )

in W01,p (Ω) × R. Indeed, by Remark 3.4, we have that

∥uλ ∥ → 0 as λ → 0.

Hence, there exist two sequences {ui } in W01,p (Ω) and {λi } in R+ (here ui = uλi )
such that
λi → 0+ and ∥ui ∥ → 0
as i → +∞. Moreover, we want to emphasis that due to the fact that the map

λ 7→ Iλ (uλ ), λ ∈ (0, λ∗ )

is strictly decreasing, for every λ1 , λ2 ∈ (0, λ∗ ), with λ1 ̸= λ2 , the solutions ūλ1 and
ūλ2 guaranteed by Remark 3.4 are different.
Remark 3.6. If f is non-negative, then the weak solution obtained by Theorem
3.1 is non-negative. To see this, let u0 be a non-trivial weak solution of the problem
(Pλf ). For the sake of a contradiction, assume that the set D = {x ∈ Ω : u0 (x) < 0}
is non-empty and of positive measure. Set v̄(x) = min{0, u0 (x)} for all x ∈ Ω.
Clearly, v̄ ∈ W01,p (Ω) and
Z Z
A(x, ∇u0 (x))∇v̄(x)dx − λ f (x, u0 (x))v̄(x)dx = 0.
Ω Ω

Thus, from our sign assumptions on f , we have

Z Z
p
0 ≤ a1 ∥u0 ∥D ≤ A(x, ∇u0 (x))∇u0 (x)dx = λ f (x, u0 (x))u0 (x)dx ≤ 0.
D D

Hence, u0 = 0 in D, which is a contradiction.

Nonlinear elliptic problem 557

4. Existence of infinitely many solutions

In this section, we formulate our main result on the existence of infinitely many
weak solutions to the problem (Pλf ). First we introduce some additional notation.
Let R : Ω → [0, +∞) be the distance function defined by R(x) = d(x, ∂Ω) for each
x ∈ Ω. Thus, for every fixed x0 ∈ Ω, B(x0 , R(x0 )) = {x ∈ Ω : |x−x0 | < R(x0 )} ⊆ Ω,
and for a ∈ Lα (Ω) with α > N/p, we let
N
R(x0 )

|B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, ξ)
2
B ∞ = lim sup N −p .
a2 R(x0 )

ξ→∞
N
(2 − 1)|B(0, 1)|ξ p
p 2

Theorem 4.1. Assume that f satisfies (Gf,a,q ) and there exist x0 ∈ Ω and two real
sequences {dn } and {bn } with
lim bn = ∞
n→∞

such that:

(A1 ) f (x, t) ≥ 0 for (x, t) ∈ B(x0 , R(x0 )) × R;

N −p p
a2 R(x0 )

a1 |Ω| p∗ p
(A2 ) (2N − 1)|B(0, 1)|dpn < b for n ∈ N;
p 2 pT p n
N
R(x0 )

Kbn − |B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, dn )
2
(A3 ) A = lim
∞
p N −p ! < B∞ ,
n→∞
a2 R(x0 )

a1 |Ω| p∗ p
bn − (2N − 1)|B(0, 1)|dpn
pT p p 2
where
p ! p1 p ! pq
(p∗ −α′ ) |Ω| p∗ p M2 (p∗ −α′ q) |Ω| p∗ p
Kbn = M1 ∥a∥α T |Ω| (p∗ α′ ) b + ∥a∥α T q |Ω| (p∗ α′ ) b .
Tp n q Tp n
 
1 1
Then, for each λ ∈ , , the problem (Pλf ) admits an unbounded sequence
B ∞ A∞
of solutions.
Proof. Our wish here is to again apply Theorem 2.1. Consider the functionals Φ
and Ψ as given in (2.5) and (2.7). The regularity assumptions on Φ and Ψ required
in Theorem 2.1 are satisfied. Take
p
a1 |Ω| p∗ p
rn = b
pT p n

for all n ∈ N. We see that rn > 0 for all n ∈ N and rn → ∞ as n → ∞. From the
definition of Φ and in view of (3.3), for every rn > 0,
(   p1 )
n o prn
Φ −1
(−∞, rn ) = u ∈ W01,p : Φ(u) < rn ⊆ u∈ W01,p : ∥u∥ ≤ ,
a1
558 J. R. Graef, S. Heidarkhani. L. Kong & S. Moradi

and inequality (3.4) assures that

 p1  pq
(p∗ −α′ ) (p∗ −α′ q)
 
prn M2 prn
sup Ψ(u) ≤ M1 ∥a∥α T |Ω| (p∗ α′ ) + ∥a∥α T q |Ω| (p∗ α′ ) .
Φ(u)≤rn a1 q a1

For each n ∈ N, define

x ∈ Ω\B̄(x0 , R(x0 )),



 0,
 2d
n
wn (x) = (R(x0 ) − |x − x0 |), x ∈ B(x0 , R(x0 ))\B̄(x0 , R(x0 )/2), (4.1)

 R(x0 )
x ∈ B(x0 , R(x0 )/2).

dn ,

Clearly, wn ∈ W01,p (Ω). A direct computation based on (3.3) shows that

a1 2p
Φ(wn ) ≥ |B(x0 , R(x0 )) \ B̄(x0 , R(x0 )/2)|dpn
p [R(x0 )]p
a1 2p
|B(0, 1)| [R(x0 )]N − [R(x0 )/2]N dpn


= p
p [R(x0 )]
N −p
a1 R(x0 )

= (2N − 1)|B(0, 1)|dpn .
p 2
Similarly,
N −p
R(x0 )

a2
Φ(wn ) ≤ (2N − 1)|B(0, 1)|dpn . (4.2)
p 2
In view of condition (A1 ),
Z
Ψ(wn ) = F (x, wn (x))dx
ZΩ
≥ F (x, dn )dx
B(x0 ,R(x0 )/2)

≥ |B(x0 , R(x0 )/2)|essinf B(x0 ,R(x0 )/2) F (x, dn )

N
R(x0 )

= |B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, dn ).
2

From (A2 ), we have Φ(wn ) < rn . Also, Φ(0) = Ψ(0) = 0. Therefore, for large n,
from (A3 ) and (4.2),

(supu∈Φ−1 (−∞,rn ) Ψ(u)) − Ψ(u)

φ(rn ) = inf
u∈Φ−1 (−∞,r n) rn − Φ(u)
N
R(x0 )

supu∈Φ−1 (−∞,rn ) Ψ(u) − |B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, dn )
2
≤
rn − Φ(wn )
N
R(x0 )

Kbn − |B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, dn )
2
≤ p N −p !.
a2 R(x0 )

a1 |Ω| p∗ p N p
bn − (2 − 1)|B(0, 1)|dn
pT p p 2
Nonlinear elliptic problem 559

Hence, from (A2 ), λ ≤ limn→∞ φ(rn ) ≤ A∞ < ∞ follows.

Next, we show that Iλ is unbounded from below. Let {cn } → ∞ be a sequence
of positive numbers to be determined. Let {yn } in W01,p (Ω) be given by

x ∈ Ω\B̄(x0 , R(x0 )),



 0,
 2c
n
yn (x) = (R(x0 ) − |x − x0 |), x ∈ B(x0 , R(x0 ))\B̄(x0 , R(x0 )/2), (4.3)

 R(x 0)
cn , x ∈ B(x0 , R(x0 )/2).


Then N −p
R(x0 )

a2
Φ(yn ) ≤ (2N − 1)|B(0, 1)|cpn . (4.4)
p 2
Hence,
N −p
R(x0 )

a2
Iλ (yn ) ≤ (2N − 1)|B(0, 1)|cpn
p 2
N
R(x0 )

−λ |B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, cn ).
2

Now, we consider the two cases.

1 1
If B∞ < ∞, then since λ > , we can fix ϵ > 0 such that ϵ < B∞ − . Hence,
B∞ λ
there exists νϵ ∈ N and a positive sequence {hn } → ∞ such that
N
R(x0 )

|B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, hn )
2
N −p !
R(x0 )

∞ a2 N
>(B − ϵ) (2 − 1)|B(0, 1)|hpn
p 2

for all n > νϵ . Hence,

N −p
R(x0 )

a2
Iλ (yn ) ≤ (2N − 1)|B(0, 1)|hpn
p 2
N
R(x0 )

−λ |B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, hn )
2
N −p
a2 R(x0 )

< (2N − 1)|B(0, 1)|hpn (1 − λ(B∞ − ϵ)).
p 2

Since 1 − λ(B∞ − ϵ) < 0, letting {hn } be the sequence {cn } in (4.3) and taking
(4.4) into account, we have
lim Iλ (yn ) = −∞.
n→∞

1
Now consider the situation where B∞ = ∞ and fix D > . There exists νD and
λ
a positive sequence {kn } → ∞ such that
N
R(x0 )

|B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, kn )
2
560 J. R. Graef, S. Heidarkhani. L. Kong & S. Moradi

N −p !
R(x0 )

a2 N
>D (2 − 1)|B(0, 1)|knp
p 2

for all n > νD , and moreover

N −p
a2 R(x0 )

Iλ (yn ) < (2N − 1)|B(0, 1)|knp (1 − λD).
p 2
Since 1 − λD < 0, letting {kn } be the sequence {cn } in (4.3), and arguing as before,
we see that
lim Iλ (yn ) = −∞.
n→∞
Hence, in both cases Iλ is unbounded from below. The conclusion of the theorem
then follows from Theorem 2.1(b).
Remark 4.1. If {dn } and {bn } are two real sequences with limn→∞ bn = ∞,
such that the condition (A2 ) in Theorem 4.1 is satisfied, then under the conditions
A∞ = 0 and B ∞ = ∞, Theorem 4.1 assures that for every λ > 0 the problem (Pλf )
admits infinitely many weak solutions.
Theorem 4.2. Assume that f satisfies (Gf,a,q ), condition (A1 ) holds, and
N
R(x0 )

|B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, ξ)
Kξ 2
(A4 ) lim inf p < lim sup N −p .
a2 R(x0 )
ξ→∞ a1 |Ω| p∗

p ξ→∞
pT p ξ N
(2 − 1)|B(0, 1)|ξ p
p 2
Then, for each
N −p p
R(x0 )
  
a2 N p a1 |Ω| p∗ p
(2 − 1)|B(0, 1)|ξ ξ 
p 2 pT p

λ ∈  , ,
 
Kξ
N
 R(x0 ) 
|B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, ξ)
2

the problem (Pλf ) has an unbounded sequence of weak solutions.

Proof. Let {dn } ≡ 0 and choose a sequence {bn } of positive numbers tending to
∞ such that
Kbn Kξ
lim p = lim inf p .
n→∞ a |Ω| p∗ ξ→∞ a |Ω| p∗
1 p 1 p
bn ξ
pT p pT p
From Theorem 4.1 the conclusion follows.
Next, we point out two simple consequences of our main results. First, from
Theorem 4.1 we have the following corollary.
Corollary 4.1. Assume that f satisfies (Gf,a,q ), there exist two real sequences {dn }
and {bn }, with limn→∞ bn = ∞, such that (A1 ) and (A2 ) hold, A∞ < 1, and
B∞ > 1. Then, the problem
(
−divA(x, ∇u) = f (x, u), in Ω,
(P f )
u = 0, on ∂Ω,

has an unbounded sequence of weak solutions.

Nonlinear elliptic problem 561

The following corollary is a consequence of Theorem 4.2.

Corollary 4.2. Assume that B∞ > 1 and
Kξ
lim inf p < 1.
ξ→∞ a1 |Ω| p∗ p
ξ
pT p
Then, the problem (P f ) has an unbounded sequence of weak solutions.
Now put
N
R(x0 )

|B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, ξ)
2
B 0 = lim sup N −p .
a2 R(x0 )

ξ→0
N
(2 − 1)|B(0, 1)|ξ p
p 2
Arguing as in the proof of Theorem 4.1, but using conclusion (c) of Theorem 2.1
instead of (b), we can obtain the following result.
Theorem 4.3. Assume that f satisfies (Gf,a,q ), (A1 ) holds, and there exist two real
sequences {sn } and {en } with limn→∞ en = 0 such that
N −p p
a2 R(x0 )

N p a1 |Ω| p∗ p
(A5 ) (2 − 1)|B(0, 1)|sn < e for n ∈ N,
p 2 pT p n
N
R(x0 )

K en − |B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, sn )
2
(A6 ) A0 = lim p N −p ! < B0 .
n→∞
a2 R(x0 )

a1 |Ω| p∗ p
en − (2N − 1)|B(0, 1)|spn
pT p p 2
1 1
Then, for each λ ∈ (λ3 , λ4 ) with λ3 := and λ4 := , the problem (Pλf ) has a
B0 A0
sequence of pairwise distinct weak solutions that strongly converges to 0 in W01,p (Ω).
Our final existence result is contained in the following theorem.
Theorem 4.4. Assume that f satisfies (Gf,a,q ) and assume that
N
R(x0 )

|B(0, 1)|essinf B(x0 ,R(x0 )/2) F (x, ξ)
Kξ 2
(A7 ) lim inf p < lim sup N −p .
a2 R(x0 )

ξ→0+ a1 |Ω| p∗ ξ→0
ξ p N
(2 − 1)|B(0, 1)|ξ p
pT p p 2
Then, for each
 
 
 1 1 
λ∈ 
R(x0 )
N , Kξ
,
|B(0,1)|essinf B(x0 ,R(x0 )/2) F (x,ξ) lim inf ξ→0+
 
2 p
lim supξ→0 ∗
 
a1 |Ω| p
R(x0 ) N −p ξp
 
a2
p 2 (2N −1)|B(0,1)|ξ p pT p

the problem (Pλf ) has a sequence of pairwise distinct weak solutions that strongly
converges to 0 in W01,p (Ω).
562 J. R. Graef, S. Heidarkhani. L. Kong & S. Moradi

Proof. Let {sn } ≡ 0 and choose a sequence {en } of positive numbers tending to
0 such that
K en Kξ
lim p = lim inf p .
n→∞ a |Ω| p∗ ξ→∞ a |Ω| p∗
1 p 1 p
en ξ
pT p pT p
The conclusion follows from Theorem 4.3.
Remark 4.2. By applying Theorem 4.3 instead of Theorem 4.1, results similar to
Remark 4.1 and Corollaries 4.1 and 4.2 can be obtained.

Declarations
The authors have no competing interests to disclose.

References
[1] G. A. Afrouzi, M. Bohner, G. Caristi, S. Heidarkhani and S. Moradi, An ex-
istence result for impulsive multi-point boundary value systems using a local
minimization principle, J. Optim. Theo. Appl., 2018, 177, 1–20.
[2] D. Barilla, M. Bohner, S. Heidarkhani and S. Moradi, Existence results for dy-
namic Sturm-Liouville boundary value problems via variational methods, Appl.
Math. Comput., 2021, 409, 125614, 10 pp.
[3] G. Bonanno, Some remarks on a three critical points theorem, Nonlinear Anal.,
2021, 54, 651–665.
[4] G. Bonanno and P. Candito, Infinitely many solutions for a class of discrete
non-linear boundary value problems, Appl. Anal., 2009, 88, 605–616.
[5] G. Bonanno, G. D’Aguı̀ and R. Livrea, Triple solutions for nonlinear elliptic
problems driven by a non-homogeneous operator, Nonlinear Anal., 2020, 197,
111862, 17 pp.
[6] G. Bonanno and G. Molica Bisci, Infinitely many solutions for a boundary
value problem with discontinuous nonlinearities, Bound. Value Probl., 2009,
2009, 670675, 20 pp.
[7] C. Chen and H. Wang, Ground state solutions for singular p-Laplacian equation
in RN , J. Math. Anal. Appl., 2009, 351, 773–780.
[8] F. Colasuonno and P. Pucci, Multiplicity of solutions for p(x)-polyharmonic
elliptic Kirchhoff equations, Nonlinear Anal., 2011, 74, 5962–5974.
[9] F. Colasuonno, P. Pucci and R. Varga, Multiple solutions for an eigenvalue
problem involving p-Laplacian type operators, Nonlinear Anal., 2012, 75, 4496–
4512.
[10] F. Demengel and E. Hebey, On some nonlinear equations involving the p-
Laplacian with critical Sobolev growth, Adv. Differ. Equ., 1998, 3, 533–574.
[11] Y. Deng and H. Pi, Multiple solutions for p-harmonic type equations, Nonlinear
Anal., 2009, 71, 4952–4959.
[12] M. Galewski and G. Molica Bisci, Existence results for one-dimensional frac-
tional equations, Math. Meth. Appl. Sci., 2016, 39, 1480–1492.
Nonlinear elliptic problem 563

[13] J. R. Graef, S. Heidarkhani and L. Kong, Infinitely many solutions for systems
of Sturm-Liouville boundary value problems, Results Math., 2014, 66, 327–341.
[14] J. R. Graef, S. Heidarkhani, L. Kong and S. Moradi, On an anisotropic discrete
boundary value problem of Kirchhoff type, J. Differ. Equ. Appl., 2021, 27, 1103–
1119.
[15] S. Heidarkhani, Infinitely many solutions for systems of n two-point boundary
value Kirchhoff-type problems, Ann. Polon. Math., 2013, 107, 133–152.
[16] S. Heidarkhani, G. A. Afrouzi, S. Moradi, G. Caristi and B. Ge, Existence of
one weak solution for p(x)-biharmonic equations with Navier boundary condi-
tions, Z. Angew. Math. Phys., 2016, 67, 1–13.
[17] S. Heidarkhani, Y. Zhao, G. Caristi, G. A. Afrouzi and S. Moradi, Infinitely
many solutions for perturbed impulsive fractional differential systems, Appl.
Anal., 2017, 96, 1401–1424.
[18] A. Kristály, H. Lisei and C. Varga, Multiple solutions for p-Laplacian type
equations, Nonlinear Anal., 2008, 68, 1375–1381.
[19] P. D. Nápoli and M. C. Mariani, Mountain pass solutions to equations of p-
Laplacian type, Nonlinear Anal., 2003, 54, 1205–1219.
[20] P. H. Rabinowitz, Minimax Methods in Critical Point Theory with Applications
to Differential Equations, CBMS Reg. Conf. Ser. Math., 1986, 65, Amer. Math.
Soc., Providence, RI.
[21] B. Ricceri, A general variational principle and some of its applications, J.
Comput. Appl. Math., 2000, 113, 401–410.
[22] G. Talenti, Best constants in Sobolev inequality, Ann. Mat. Pura Appl., 1976,
110, 353–372.