SUPERLINEAR PERTURBATIONS OF THE EIGENVALUE PROBLEM

FOR THE ROBIN LAPLACIAN PLUS AN INDEFINITE AND

UNBOUNDED POTENTIAL

NIKOLAOS S. PAPAGEORGIOU, VICENŢIU D. RĂDULESCU, AND DUSAN D. REPOVŠ

Abstract. We consider a superlinear perturbation of the eigenvalue problem for the

Robin Laplacian plus an indefinite and unbounded potential. Using variational tools and
critical groups, we show that when λ is close to a nonprincipal eigenvalue, the problem
has seven nontrivial solutions. We provide sign information for six of them.

1. Introduction
Let Ω ⊆ RN (N > 2) be a bounded domain with a C 2 -boundary ∂Ω. In this paper, we
study the following parametric semilinear Robin problem
−∆u(z) + ξ(z)u(z) = λu(z) + f (z, u(z)) in Ω,
(
(Pλ ) ∂u
+ β(z)u = 0 on ∂Ω, λ ∈ R.
∂n
In this problem ξ ∈ Ls (Ω) with s > N and is indefinite (that is, sign-changing). We
assume that ξ(·) is bounded from above (that is, ξ + ∈ L∞ (Ω)). So, the differential operator
(left hand side) of (Pλ ) is not coercive. In the reaction (right hand side) of (Pλ ), we have the
parametric term u 7→ λu and a perturbation f (z, x) which is measurable function such that
f (z, ·) is continuously differentiable. We assume that f (z, ·) exhibits superlinear growth near
±∞, but without satisfying the usual in such cases Ambrosetti-Rabinowitz condition (the
AR-condition for short). Instead we employ a less restrictive condition which incorporates
in our framework superlinear nonlinearities with slower growth near ±∞ which fail to
satisfy the AR-condition. So, problem (Pλ ) can be viewed as a perturbation of the classical
eigenvalue problem for the operator u 7→ −∆u + ξ(z)u with Robin boundary condition.
In the past such problems were studied primarily in the context of Dirichlet equations
with no potential. The first work is that of Mugnai [5], who used general linking theorem of
Marino-Saccon [4] to produce three nontrivial solutions. The work of Mugnai was extended
by Rabinowitz-Su-Wang [15] who based their method of proof on bifurcation theory, vari-
ational techniques and critical groups and produced three nontrivial solutions. Analogous
results for scalar periodic equations, were proved by Su-Zeng [16]. All the aforementioned
work use the AR-condition to express the superlinearity of the perturbation f (z, ·). A more
general superlinearity condition, was employed by Ou-Li [7] who also produced three non-
trivial solutions for λ > 0 near a nonprincipal eigenvalue. As we already mentioned earlier,
in all aforementioned works, there is no potential term and so the differential operator is
coercive. This facilitates the analysis of the problem. Papageorgiou-Rădulescu-Repovš [12]
went beyond Dirichlet problems and studied Robin problems with an indefinite potential.
In [12] the emphasis is on the existence and multiplicity of positive solutions. So, the condi-
tions on the perturbation f (z, ·) are different, leading to a bifurcation type result describing
◦
the change in the set of positive solutions as the parameter λ moves in R+ = (0, +∞).

Key words and phrases. Superlinear perturbation, regularity theory, maximum principle, constant sign
and nodal solutions, critical groups, indefinite potential
2010 Mathematics Subject Classification. Primary: 35J20. Secondary: 35J60, 58E05.
1
2 N.S. PAPAGEORGIOU, V.D. RĂDULESCU, AND D.D. REPOVŠ

We also mention the works of Castro-Cassio-Velez [1], Papageorgiou-Papalini [8] (Dirichlet

problems) and Hu-Papageorgiou [3] (Robin problems) who also produce seven nontrivial so-
lutions. In Castro-Cassio-Velez [1] there is no potential term, while Papageorgiou-Papalini
[8] and Hu-Papageorgiou [3] have an indefinite potential term and moreover, provide sign
information for all solution they produce. Finally we mention the work of Papageorgiou-
Rădulescu [11] who prove multiplicity results for nearly resonant Robin problems.
Here using variational tools from the critical point theory together with suitable trunca-
tion, perturbation and comparison techniques and using also critical groups (Morse theory),
we show that when the parameter λ > 0 is close to an eigenvalue of (−∆u + ξu, H 1 (Ω))
with Robin boundary condition, then the problem has seven nontrivial smooth solutions
providing sign information for six of them.

2. Mathematical Background-Hypotheses
The main space in the analysis of problem (Pλ ) are the Sobolev space H 1 (Ω), the Banach
space C 1 (Ω) and the ”boundary” Lebesgue spaces Lp (∂Ω), 1 6 p 6 ∞.
The Sobolev space H 1 (Ω) is a Hilbert space with inner product
Z Z
(u, h) = uhdz + (Du, Dh)RN dz for all u, h ∈ H 1 (Ω).
Ω Ω
By k · k we denote the norm corresponding to this inner product. So
1/2
kuk = kuk22 + kDuk22 for all u ∈ H 1 (Ω).


The Banach space C 1 (Ω) is ordered by the positive (order) cone C+ = {u ∈ C 1 (Ω) :
u(z) > 0 for all z ∈ Ω}. This cone has a nonempty interior given by
intC+ = {u ∈ C+ : u(z) > 0 for all z ∈ Ω}.
On ∂Ω we consider the (N − 1)-dimensional Hausdorff measure (surface measure) τ (·).
Using this measure, we can define in the usual way the boundary value spaces Lp (∂Ω)
1 6 p 6 ∞. From the theory of Sobolev spaces we know that there exists a unique
continuous linear map γ0 : H 1 (Ω) → L2 (∂Ω), known as the ”trace map”, such that

γ0 (u) = u for all u ∈ H 1 (Ω) ∩ C(Ω).

∂Ω
So, the trace map extends the notion of boundary values to all Sobolev functions. We
know that
imγ0 = H 1/2,2 (∂Ω) and ker γ0 = H01 (Ω).
 
1 p 2(N − 1)
The linear map γ0 (·) is compact from H (Ω) into L (∂Ω) for all p ∈ 1, if
N −2
N > 3 and into Lp (∂Ω) for all 1 6 p < ∞, if N = 2.
In the sequel for the sake of notational simplicity we drop the use of the map γ0 (·). All
restrictions of Sobolev functions on ∂Ω are understood in the sense of traces.
Let x ∈ R. We set x± = max{±x, 0} and the given u ∈ H 1 (Ω) we define u± (z) = u(z)±
for all z ∈ Ω. We know that
u± ∈ H 1 (Ω), u = u+ − u− , |u| = u+ + u− .
Given u, v ∈ H 1 (Ω) with u 6 v, we set
[u, v] = {h ∈ H 1 (Ω) : u(z) 6 h(z) 6 v(z) for a.a. z ∈ Ω}.
By intC 1 (Ω) [u, v] we denote the interior in the C 1 (Ω)-norm topology of [u, v] ∩ C 1 (Ω).
Let us introduce our hypotheses on the potential function ξ(·) and the boundary coeffi-
cient β(·).
 SUPERLINEAR PERTURBATIONS OF THE EIGENVALUE PROBLEM 3

H0 : ξ ∈ Ls (Ω) with s > N if N > 2 and s > 1 if N = 2, ξ + ∈ L∞ (Ω) and β ∈ W 1,∞ (∂Ω)
with β(z) > 0 for all z ∈ ∂Ω.
As we mentioned in the introduction, our analysis of problem (Pλ ) relies on the spectrum
of u 7→ −∆u + ξ(z)u with Robin boundary condition. So, we consider the following linear
eigenvalue problem

 −∆p u(z) + ξ(z)u(z) = λ̂u(z) in Ω,
(1) ∂u
 + β(z)u = 0 on ∂Ω.
∂n
We say that λ̂ ∈ R is an ”eigenvalue”, if problem (1) admits a nontrivial solution û ∈
H 1 (Ω) known as an ”eigenfunction” corresponding to the eigenvalue λ̂. From hypotheses
H0 and the regularity theory of Wang [17], we know that û ∈ C 1 (Ω).
Let γ : H 1 (Ω) → R be the C 2 -functional defined by
Z Z
2 2
γ(u) = kDuk2 + ξ(z)u dz + β(z)u2 dσ for all u ∈ H 1 (Ω).
Ω ∂Ω
From D’Agui-Marano-Papageorgiou [2] (see also Papageorgiou-Rădulescu [10]), we know
that there exists µ > 0 such that
(2) γ(u) + µkuk22 > Ĉkuk2 for some Ĉ > 0, all u ∈ H 1 (Ω).
Using (2) and the spectral theorem for compact, self adjoint operators on Hilbert space,
we show (see [2], [10]), that the spectrum of (1) consists of a sequence {λ̂k }k∈N of distinct
eigenvalues such that λ̂k → +∞ as k → ∞. There is also a corresponding sequence
{ûk }k∈N ⊆ H 1 (Ω) of eigenfunctions which form an orthogonal basis for H 1 (Ω) and an
orthogonal basis for L2 (Ω). As we already mentioned ûk ∈ C 1 (Ω) for all k ∈ N. By E(λ̂k )
we denote the eigenspace corresponding to the eigenvalue λ̂k . We have E(λ̂k ) ⊆ C 1 (Ω) for
all k ∈ N, it is finite dimensional and
H 1 (Ω) = ⊕ E(λ̂k ).
k>1

Moreover, each eigenspace E(λ̂k ) has the ”Unique Continuation Property” (the UCP for
short) which says
”if u ∈ E(λ̂k ) and u(·) vanishes on a set of positive measure,
then u ≡ 0”
The first (principal) eigenvalue λ̂1 is simple, that is dim E(λ̂1 ) = 1. All the eigenvalues
γ(u)
admit variational characterizations in the terms of the Rayleigh quotient , u ∈ H 1 (Ω),
kuk22
u 6= 0. We have
 
γ(u) 1
(3) λ̂ = inf : u ∈ H (Ω), u 6= 0 ,
kuk22
 
γ(u) k
λ̂k = sup : u ∈ H k = ⊕ E(λ̂m ), u 6= 0
kuk22 m=1
 
γ(u)
(4) = inf : u ∈ Ĥk = ⊕ E(λ̂m ), u 6= 0 , k > 2.
kuk22 m>k

In (3) the infimum is realized on E(λ̂1 ), while in (4) both the supremum and the infimum
are realized on E(λ̂k ).
From (3) it follows that the elements of E(λ̂1 ) have fixed sign, while from (4) and the
orthogonality of the eigenspaces, we see that the elements of E(λ̂k ), k > 2 are nodal (sign
4 N.S. PAPAGEORGIOU, V.D. RĂDULESCU, AND D.D. REPOVŠ

changing). By û1 we denote the positive, L2 -normalized (that is kûk2 = 1) eigenfunction

corresponding to λ̂1 . The regularity theory and the Hopf maximum principle imply that
û1 ∈ intC+ .
Let X be a Banach space, c ∈ R and ϕ ∈ C 1 (X, R). We introduce the following sets
Kϕ = {u ∈ X : ϕ0 (u) = 0} (the critical set of ϕ),
ϕc = {u ∈ X : ϕ 6 c}.
We say that ϕ(·) satisfies the ”C-condition”, if the following property holds:
”Every sequence {un }n>1 such that
{ϕ(un )}n>1 ⊆ R is bounded
and (1 + kun kX )ϕ0 (un ) → 0 in X ∗ as n → ∞,
admits a strongly convergent subsequence”.
This is a compactness-type condition on the functional ϕ(·). Since the ambient space
is not general locally compact (being infinite dimensional), the burden of compactness is
passed to the functional ϕ(·). Using the C-condition one can prove a deformation theorem
from which follows the minimax theory of the critical values of ϕ(·) (see, for example,
Papageorgiou-Rădulescu-Repovš [13], Chapter 5).
Let (Y1 , Y2 ) be a topological pair such that Y2 ⊆ Y1 ⊆ X. Given k ∈ N0 by Hk (Y1 , Y2 )
we denote the k th -relative singular homology group for the pair (Y1 , Y2 ) with Z-coefficients.
If ϕ ∈ C 1 (X, R), u ∈ Kϕ is isolated and c = ϕ(u), then the critical groups of ϕ at u, are
defined by
Ck (ϕ, u) = Hk (ϕc ∩ U, ϕc ∩ U \ {u}) for all k ∈ N0 ,
with U being a neighborhood of u such that Kϕ ∩ ϕc ∩ U = {u}. The excision property
of singular homology, implies that the above definition of critical groups, is independent of
the choice of the isolating neighborhood U . We say that a Banach X has the ”Kadec-Klee
property” if the following is true
w
”un → u in X and kun kX → kukX ⇒ un → u in X”.
A uniformly convex space has the Kadec-Klee property. In particular then Hilbert spaces
have the Kadec-Klee property.
By A ∈ L(H 1 (Ω), H 1 (Ω)∗ ), we denote the operator defined by
Z
hA(u), hi = (Du, Dh)RN dz for all u, h ∈ H 1 (Ω).
Ω
Also by δk,m we denote the Kronecker symbol defined by

1, if k = m
δk,m =
0, if k 6= m.

( Finally by 2∗ we denote the Sobolev critical exponent corresponding to 2, that is, 2∗ =

2N
, if N > 3
N −2
+∞, if N = 2.
Now we will introduce the hypotheses on the perturbation f (z, x).
H1 : f : Ω × R → R is a measurable function such that for a.a. z ∈ Ω, f (z, 0) = 0,
f (z, ·) ∈ C 1 (R) and
(i) |fx0 (z, x)| 6 a(z)[1 r−2 ∞ ∗
Z x + |x| ] for a.a. z ∈ Ω, all x ∈ R, with a ∈ L (Ω), 2 < r < 2 ;
F (z, x)
(ii) if F (z, x) = f (z, s)ds, then lim = +∞ uniformly for a.a. z ∈ Ω;
0 x→±∞ x2
 SUPERLINEAR PERTURBATIONS OF THE EIGENVALUE PROBLEM 5
   
N
(iii) there exists τ ∈ (r − 2) max 1, , 2∗ such that
2

f (z, x)x − 2F (z, x)

0 < β̂0 6 lim inf uniformly for a.a. z ∈ Ω;
x→±∞ |x|τ

f (z, x)
(iv) fx0 (z, 0) = lim = 0 uniformly for a.a. z ∈ Ω;
x→0 x
(v) there exist C , δ > 0 and q > 2 such that F (z, x) > −C ∗ |x|q for a.a. z ∈ Ω, all
∗

x ∈ R and 0 6 f (z, x)x for a.a. z ∈ Ω, all 0 6 |x| 6 δ0 ;

(vi) there exist constants C− < 0 < C+ and m ∈ N, m > 2 such that

[λ̂m+1 − ξ(z)]C+ + f (z, C+ ) 6 0 6 [λ̂m+1 − ξ(z)]C− + f (z, C− ) for a.a. z ∈ Ω;

(vii) for every ρ > 0 there exists ξˆρ > 0 such that for a.a. z ∈ Ω, the function x 7→
f (z, x) + ξˆρ x is nondecreasing on [−ρ, ρ].
Remarks: Hypotheses H1 (ii), (iii) imply that

f (z, x)
lim = ±∞ uniformly for a.a. z ∈ Ω.
x→±∞ x
Hence for a.a. z ∈ Ω, f (z, ·) is superlinear. However, thus superlinearity of the perturba-
tion term is not expressed using the AR-condition which is common in the literature when
dealing with superlinear problems. Recall that the AR-condition says that there exists
q > 2 and M > 0 such that

(5a) 0 < qF (z, x) 6 f (z, x)x for a.a. z ∈ Ω, all |x| > M

(5b) 0 < essinf F (·, ±M )

Ω

(see Mugnai [6]). Integrating (5a) and using (5b), we obtain the weaker condition

C0 |x|q 6 F (z, x) for a.a. z ∈ Ω, all |x| > M,

⇒ C0 |x|q 6 f (z, x)x for a.a. z ∈ Ω, all |x| > M (see (5a)).

So we see that the AR-condition implies that f (z, ·) has at least (q − 1)-polynomial
growth. In this paper, instead of the AR-condition, we employ the less restrictive condition
H1 (iii), which allows the consideration of superlinear nonlinearities with ”slower” growth
near ±∞ which fail to satisfy the AR-condition. The following example illustrates this. For
the sake of simplicity we drop the z-dependence of f and assume that ξ ∈ L∞ (Ω). Suppose
that for some m ∈ N, we have C > |λ̂m+2 | + kξk∞ , C > 0. Then the function

x − (C + 1)|x|q−2 x, if |x| 6 1

f (x) = 2 < q,
x ln |x| − Cx, if 1 < x

satisfies hypotheses H1 but fails to satisfy the AR-condition.

By ϕλ : H 1 (Ω) → R we denote the energy functional for problem (Pλ ) λ > 0, defined by
Z
1 λ
ϕλ (u) = γ(u) − kuk22 − F (z, u)dz for all u ∈ H 1 (Ω).
2 2 Ω

We have ϕλ ∈ C 2 (H 1 (Ω)).
6 N.S. PAPAGEORGIOU, V.D. RĂDULESCU, AND D.D. REPOVŠ

3. Constant Sign Solutions

In this section we prove the existence of four nontrivial smooth constant sign solutions
when λ ∈ (λ̂m , λ̂m+1 ).

Proposition 1. If hypotheses H0 , H1 hold and λ̂m < λ < λ̂m+1 (see H1 (vi)), then problem
(Pλ ) has at least four nontrivial solutions of constant sign

u0 , û ∈ intC+ , u0 6= û,
v0 , v̂ ∈ −intC+ , v0 6= v̂.

Proof. Let µ > 0 be as in (2) and consider the Carathéodory function gλ0 (z, x) defined by

(λ + µ)x+ + f (z, x+ ),

if x 6 C+
(5) gλ+ (z, x) =
(λ + µ)C+ + f (z, C+ ), if C+ < x.
Z x
We set G+
λ = gλ+ (z, s)ds and consider the C 1 -functional Ψ+ 1
λ : H (Ω) → R defined by
0
Z
1 µ
Ψ+ 2
λ (u) = 2 γ(u) + 2 kuk2 − G+ 1
λ (z, u)dz for all u ∈ H (Ω).
Ω

From (2) and (5), we see that Ψ+ λ (·) is coercive. Also, using the Sobolev embedding
theorem and the compactness of the trace map, we see that Ψ+ λ (·) is sequentially weakly
lower semicontinuous. So, by the Weierstrass-Tonelli theorem, we can find u0 ∈ H 1 (Ω) such
that

Ψ+
 + 1

(6) λ (u0 ) = inf Ψλ (u) : u ∈ H (Ω) .

Let t > 0 be small so that tû1 (z) 6 min{C+ , δ0 } for all z ∈ Ω (recall that û1 ∈ intC+ ).
Using (5) and hypothesis H1 (v) we have

t2
Ψ+ [λ̂1 − λ] < 0 (since λ > λ̂1 , kû1 k2 = 1),
λ (tû1 ) 6
2
⇒ Ψ+ +
λ (u0 ) < 0 = Ψλ (0) (see (6)),
⇒ u0 6= 0.

From (6) we have

0
(Ψ+
λ ) (u0 ) = 0,

Z Z Z
(7) ⇒ hA(u0 ), hi + [ξ(z) + µ]u0 hdz + β(z)u0 hdσ = gλ+ (z, u0 )hdz
Ω ∂Ω Ω
1
for all h ∈ H (Ω).

In (7) first we choose h = −u− 1

0 ∈ H (Ω). Then

γ(u− − 2
0 ) + µku0 k2 = 0 see (5)
⇒ Ĉku− 2
0 k 6 0 (see (2))
⇒ u0 > 0, u0 6= 0.
 SUPERLINEAR PERTURBATIONS OF THE EIGENVALUE PROBLEM 7

Next in (7) we choose h = (u0 − C+ )+ ∈ H 1 (Ω). We have

Z Z
+ +
hA(u0 ), (u0 − C+ ) i + [ξ(z) + µ]u0 (u0 − C+ ) dz + β(z)u0 (u0 − C+ )+ dσ
Ω ∂Ω
Z
+
= [(λ + µ)C+ + f (z, C+ )] (u0 − C+ ) dz (see (5))
Ω
Z h i
6 (λ̂m+1 + µ)C+ + f (z, C+ ) (u0 − C+ )+ dz (since λ < λ̂m+1 )
ZΩ
6 [ξ(z) + µ]C+ (u0 − C+ )+ dz (see hypotheses H1 (vi)),
Ω
⇒ u0 6 C+ .
So, we have proved that
(8) u0 ∈ [0, C+ ], u0 6= 0.
From (8), (5) and (7) it follows that u0 is a positive solution of (Pλ ) and we have
−∆u0 (z) + ξ(z)u0 (z) = λu0 (z) + f (z, u0 (z)) for a.a. z ∈ Ω,
(
(9) ∂u0
+ β(z)u0 = 0 on ∂Ω.
∂n
(see Papageorgiou-Rădulescu [9]).
We consider the following functions

 0, if 0 6 u0 (z) 6 1
ϑ̂λ (z) = f (z, u0 (z))
 λ − ξ(z) + , if 1 < u0 (z)
u0 (z)

(λ − ξ(z))u0 (z) + f (z, u0 (z)), if 0 6 u0 (z) 6 1
and γ̂λ =
0, if 1 < u0 (z).
On account of hypotheses H0 , we have
ϑ̂λ ∈ Ls (Ω) (s > N ) and |ϑ̂λ (z)| 6 |λ − ξ(z)| + C1 [1 + u0 (z)r−1 ]
for a.a. z ∈ Ω, some C1 > 0.
If N > 3 (the case N = 2 is clear since then 2∗ = +∞), then
 
N 2N N 2N
(r − 2) < −2 = = 2∗ .
2 N −2 2 N −2
Since u0 ∈ H 1 (Ω), by the Sobolev embedding theorem we have
(r−2) N
u0 2
∈ L1 (Ω)
N
⇒ ϑ̂λ ∈ L 2 (Ω).
From (9) we have

 −∆u0 (z) = ϑ̂λ (z)u0 (z) + γ̂λ (z) for a.a. z ∈ Ω,
∂u0
 + β(z)u0 = 0 on ∂Ω.
∂n
From Lemma 5.1 of Wang [17], we obtain that
u0 ∈ L∞ (Ω).
Then the Calderon-Zygmund estimates (see Lemma 5.2 of Wang [17]) imply u0 ∈ W 2,s (Ω).
N
By Sobolev embedding theorem we have W 2,s (Ω) ,→ C 1,α (Ω) with α = 1 − > 0. So,
s
8 N.S. PAPAGEORGIOU, V.D. RĂDULESCU, AND D.D. REPOVŠ

u0 ∈ C 1,α (Ω). Let ρ = kuk∞ and let ξˆρ > 0 be as postulated by hypothesis H1 (vii). From
(9) we have
h i
∆u0 (z) 6 kξ + k∞ + ξˆρ u0 (z) for a.a. z ∈ Ω
(see hypotheses H0 ),
⇒ u0 ∈ intC+ (by the maximum princinple).
Evidently choosing ξˆρ > 0 even bigger if necessary, we can have that for a.a. z ∈ Ω, the
function
x 7→ [λ + ξˆρ ]x + f (z, x)
is nondecreasing on [−ρ, ρ] (ρ = ku0 k∞ ). We have
−∆u0 (z) + [ξ(z) + ξˆρ ]u0 (z)
= [λ + ξˆρ ]u0 (z) + f (z, u0 (z))
6 [λ + ξˆρ ]C+ + f (z, C+ ) (see (8))
6 [ξ(z) + ξˆρ ]C+ for a.a. z ∈ Ω (see hypothesis H1 (vi)),
h i
⇒ ∆(C+ − u0 )(z) 6 kξ + k∞ + ξˆρ (C+ − u0 (z)) for a.a. z ∈ Ω,
⇒ C+ − u0 ∈ intC+ ,
(10) ⇒ u0 ∈ intC 1 (Ω) [0, C+ ].

Let ϕ+ 1 1
λ : H (Ω) → R be the C -functional defined by
Z
+ 1 µ − 2 λ + 2
ϕλ = γ(u) + ku k2 − ku k2 − F (z, u+ )dz for all u ∈ H 1 (Ω).
2 2 2 Ω
From (5) it is clear that

ϕ+
λ = Ψ+
λ
[0,C+ ] [0,C+ ]
⇒ u0 is a local C (Ω)-minimizer of ϕ+
1
λ (see (10)),
⇒ u0 is a local H 1 (Ω)-minimizer of ϕ+
λ
(see Papageorgiou-Rădulescu [9]).
It is easy to see that
Kϕ+ ⊆ C+ (regularity theory),
λ

⇒ Kϕ+ ⊆ intC+ ∪ {0} (maximum principle).

λ

So, we may assume Kϕ+ is finite. Otherwise we already have an infinity of positive smooth
λ
solutions and so we are done. Then on account of Theorem 5.7.6, p.449 of Papageorgiou-
Rădulescu-Repovš [13], we can find ρ0 ∈ (0, 1) small such that
ϕ+
 +  +
(11) λ (u0 ) < inf ϕλ (u) : ku − u0 k = ρ0 = mλ .
Hypothesis H1 (ii) implies that
(12) ϕ+
λ (tû1 ) → −∞ as t → +∞.

Claim: The functional ϕ+

λ satisfies the C-condition.
Consider a sequence {un }n>1 ⊆ H 1 (Ω) such that
(13) |ϕ+
λ (un )| 6 C2 for some C2 > 0, all n ∈ N,
0 ∗
(14) (1 + kun k)(ϕ+ 1
λ ) (un ) → 0 in H (Ω) as n → ∞.
 SUPERLINEAR PERTURBATIONS OF THE EIGENVALUE PROBLEM 9

From (14) we have

Z Z Z Z
hA(un ), hi + ξ(z)un hdz + β(z)un hdσ − µu−
n hdz − [λu∗n + f (z, u+
n )]hdz
Ω ∂Ω Ω Ω
εn khk
(15) 6
1 + kun k
for all h ∈ H 1 (Ω), with εn → 0+ .

In (15) we choose h = −u− 1

n ∈ H (Ω). Then

γ(u− − 2
n ) + µkun k2 6 εn for all n ∈ N,
⇒ Ĉku− 2
n k 6 εn for all n ∈ N (see (2)),
(16) ⇒ u− 1
n → 0 in H (Ω) as n → ∞.

Next in (15) we choose h = u+ 1

n ∈ H (Ω). We obtain
Z
(17) −γ(un ) + [λ(u+
+ 2 + +
n ) + f (z, un )un ]dz 6 εn for all n ∈ N.
Ω

On the other hand from (13) and (16), we have

Z
(18) γ(u+
n ) − [λ(u+ 2 +
n ) + 2F (z, un )]dz 6 C3 for some C3 > 0, all n ∈ N.
Ω

We add (17) and (18) and obtain

Z
(19) [f (z, u+ + +
n )un − 2F (z, un )]dz 6 C4 for some C4 > 0, all n ∈ N.
Ω

Hypotheses H1 (i), (iii) imply that we can find β̂1 (0, β̂0 ) and C5 > 0 such that

(20) β̂1 |x|τ − C5 6 f (z, x)x − 2F (z, x) for a.a. z ∈ Ω, all x ∈ R.

We use (20) in (19) and obtain that

(21) {u+ τ
n }n>1 ⊆ L (Ω) is bounded.

First assume that N > 3. From hypothesis H1 (iii) we see that without any loss of
generality, we may assume that τ < r < 2∗ . So, we can find t ∈ (0, 1) such that
1 1−t t
(22) = + ∗.
r τ 2
From the interpolation inequality (see Papageorgiou-Winkert [14], Proposition 2.3.17,
p.116), we have

ku+ + 1−t + t
n kr 6 kun kτ kun k2∗
(23) ⇒ ku+ r + tr
n kr 6 C6 kun k for some C6 > 0, all n ∈ N
(see (21) and use the Sobolev embedding theorem).

From hypothesis H1 (i) we have

(24) f (z, x)x 6 C7 [1 + xr ] for a.a. z ∈ Ω, all x > 0, some C7 > 0.

10 N.S. PAPAGEORGIOU, V.D. RĂDULESCU, AND D.D. REPOVŠ

In (15) we choose h = u+ 1
n ∈ H (Ω). Then
Z
γ(u+ + 2 + 2
n ) + µkun k2 6 [λ + µ]kun k2 + f (z, u+ +
n )un dz + εn
Ω
6 [|λ| + µ] ku+ 2 + tr
n k2 + C8 [1 + kun k ]
for some C8 > 0 (see (24) and (23))
6 C9 1 + ku+ tr
 
nk
for some C9 > 0 (recall 2 6 τ and see (21)),
⇒ Ĉku+ 2
6 C9 1 + ku+ tr
 
(25) nk nk for all n ∈ N.
N
Using (22) and the fact that τ > (r −2) (see hypothesis H1 (iii) and recall that N > 3),
2
we see that tr < 2. So, from (25) it follows that
{u+ 1
n }n>1 ⊆ H (Ω) is bounded,
(26) ⇒ {un }n>1 ⊆ H 1 (Ω) is bounded (see (16)).
We may assume that
w
(27) un → u in H 1 (Ω) as n → ∞.
In (15) we choose h = un − u ∈ H 1 (Ω), pass tot the limit as n → ∞ and use (23), the
Sobolev embedding theorem and the compactness of the trace map, we obtain
lim hA(un ), un − ui = 0,
n→∞
(28) ⇒ kDun k2 → kDuk2 .
From (27), (28) and the Kadec-Klee property of H 1 (Ω), we infer that
(29) un → u in H 1 (Ω) as n → ∞.
This proves that ϕ+λ satisfies the C-condition when N > 3.
∗
If N = 2, then 2 = +∞ and by the Sobolev embedding theorem, we have H 1 (Ω) ,→
η
L (Ω) compactly for all 1 6 η < ∞. Then for the previous argument to work, we replace
2∗ (= +∞) with η > r > τ . We choose t ∈ (0, 1) such that
1 1−t t
= +
r τ η
η(r − t)
⇒ tr = ,
η−τ
⇒ tr → r − τ as η → +∞ and r − τ < 2 (see H1 (iii)).
So, we choose η > r big enough so that tr < 2 and reasoning as above, we obtain (26)
and then from that and the Kadec-Klee property, we reach again (29). We conclude that
ϕ+λ satisfies the C-condition. This proves the Claim.
Then (11), (12) and the Claim, permit the use of the mountain pass theorem. So, we can
find û ∈ H 1 (Ω) such that
(30) û ∈ Kϕ+ ⊆ intC+ ∪ {0} and m+ +
λ 6 ϕλ (û) (see(11))
λ

From (11) and (30) it follows that û 6= u0 . If we show that û 6= 0, then this will be the
second positive solution of (Pλ ).
On account of hypotheses H1 (i), (iv), we have
(31) |f (z, x)| 6 C10 [|x| + |x|r−1 ] for a.a. z ∈ Ω, all x ∈ R, some C10 > 0.
 SUPERLINEAR PERTURBATIONS OF THE EIGENVALUE PROBLEM 11

We have
µ + |λ| − 2
Z
|ϕλ (u) − ϕ+
λ (u)| 6 kun k2 + |F (z, −u− )|dz
2
Ω
6 C11 kuk2 + kukr for some C11 > 0 (see (31)).

(32)
Also for h ∈ H 1 (Ω) we have
hϕ0λ (u) − (ϕ+ 0 r−1
 
λ ) (u), hi 6 C12 kuk + kuk khk for some C12 > 0,
⇒ kϕ0λ (u) − (ϕ+ 0 r−1
 
(33) λ ) (u)kH 1 (Ω)∗ 6 C12 kuk + kuk .
From (32), (33) and the C 1 -continuity of critical groups (see Papageorgiou-Rădulescu-
Repovš [13], Theorem 6.3.4, p.503), we have
(34) Ck (ϕλ , 0) = Ck (ϕ+
λ , 0) for all k ∈ N0 .

By hypothesis λ ∈ (λ̂m , λ̂m+1 ) and m > 2. So, u = 0 is a nondegenerate critical point of

ϕλ with Morse index dm = dim H m > 2 (since m > 2). Then by Proposition 6.2.6, p. 479,
we have
Ck (ϕλ , 0) = δk,dm Z for all k ∈ N0 ,
(35) ⇒ Ck (ϕ+
λ , 0) = δk,dm Z for all k ∈ N0 (see (34)).
On the other hand from the previous part of the proof we know that û ∈ Kϕ+ is of
λ
mountain pass type. Therefore Theorem 6.5.8, p. 527, of Papageorgiou-Rădulescu-Repovš
[13], implies that
(36) C1 (ϕ+
λ , û) 6= 0.
From (36), (35) and since dm > 2, we conclude that û 6= 0 and so û ∈ intC+ is the second
positive solution of (Pλ ) distinct from u0 .
For the negative solutions, we consider the Carathéodory function gλ− (z, x) defined by

− (λ + µ)C− + f (z, C− ), if x 6 C−
gλ (z, x) = − −
(λ + µ)(−x ) + f (z, −x ), if C− < x.
Z x
We set G− λ (z, x) = gλ− (z, s)ds and consider the C 1 -functionals Ψ− − 1
λ , ϕλ : H (Ω) → R
0
defined by
Z
1 µ
Ψ−
λ (u) = γ(u) + 2
kuk2 − G−
λ (z, u)dz
2 2 Ω
Z
1 µ + 2 λ − 2
and ϕ−
λ (u) = γ(u) + ku k2 − ku k2 − F (z, −u− )dz
2 2 2 Ω

for all u ∈ H 1 (Ω).

Working with these two functions as above, we produce two negative solutions v0 , v̂ ∈
−intC+ , v0 6= v̂.


4. Nodal Solutions
In this section we show that when λ is close to λ̂m+1 (near resonance) we can generate
two nodal (sign changing) solutions.
Proposition 2. If hypotheses H0 , H1 hold and λ ∈ (λ̂m , λ̂m+1 ) (see H1 (vi)), then we can
find δ̂ > 0 such that for all λ ∈ (λ̂m+1 − δ̂, λ̂m+1 ) problem (Pλ ) has at least two nodal
solutions y0 , ŷ ∈ C 1 (Ω).
12 N.S. PAPAGEORGIOU, V.D. RĂDULESCU, AND D.D. REPOVŠ

Proof. From Proposition 2.3 of Rabinowitz-Su-Wang [15], we know that there exists δ1 > 0
such that for all λ ∈ (λ̂m+1 − δ1 , λ̂m+1 ) problem (Pλ ) has at least two nontrivial solutions
y0 , ŷ ∈ H 1 (Ω). As before, using the regularity theory of Wang [17], we obtain that y0 , ŷ ∈
C 1 (Ω). Note that the result of Rabinowitz-Su-Wang [15] is for Dirichlet problems with
ξ ≡ 0. However, their proof is based on the abstract bifurcation theorem of Rabinowitz (see
Theorem 2.1 in [15]) and so it applies verbatim in our case too.
We will show that we can have these two solutions y0 , ŷ ∈ C 1 (Ω) to be nodal. From the
proof of Proposition 2.3 of! Rabinowitz-Su-Wang [15] and using hypothesis H1 (iv), we see
λ − λ̂1
that given ε ∈ 0, (recall λ > λ̂1 ), we can find 0 < δ̂ 6 δ1 such that
2

(37) λ ∈ (λ̂m+1 − δ̂, λ̂m+1 ) ⇒ |f (z, w(z))| 6 εw(z) for a.a. z ∈ Ω,

with w = y0 or w = ŷ. Suppose that w ∈ intC+ (the reasoning is similar if w ∈ −intC+ ).
We have
Z
λ̂1 wû1 dz
Ω
Z Z
= hA(û1 ), wi + ξ(z)û1 wdz + β(z)û1 wdσ
Ω ∂Ω
Z Z Z Z
∂w
= (−∆w)û1 dz + û1 dσ + ξ(z)û1 wdz + β(z)û1 wdσ
Ω ∂Ω ∂n Ω ∂Ω
(using Green’s identity)
Z
= [λw − f (z, w)]û1 dz (since w is a solution of (Pλ ))
Ω
Z " #
λ − λ̂1 λ − λ̂
> λw − w û1 dz (see (37) and recall 0 < ε 6 )
Ω 2 2
Z
λ + λ̂1
= wû1 dz
Ω 2
Z
> λ̂1 wû1 dz, a contradiction.
Ω

So, w = y0 or w = ŷ can not be constant sign and so y0 , ŷ ∈ C 1 (Ω) are nodal solutions
of (Pλ ) for λ ∈ (λ̂m+1 − δ̂, λ̂m+1 ).


5. Seventh Nontrivial Solution

In this case we prove the existence of a seventh nontrivial solution for problem (Pλ )
when λ ∈ (λ̂m , λ̂m+1 ). However, we are unable to provide sign information for this seventh
solution.
Proposition 3. If hypotheses H0 , H1 (i), (iv) hold and λ < λ̂m+2 , then there exists ρ > 0
such that
ϕλ > C̃0 > 0
Ĥm+2 ∩∂Bρ

with Ĥm+2 = ⊕ E(λ̂k ), Bρ = {u ∈ H 1 (Ω) : kuk < ρ}.

k>m+2

Proof. Hypotheses H1 (i), (iv) imply that given ε > 0, we can find Cε > 0 such that
ε
(38) |F (z, x)| 6 x2 + Cε |x|r for a.a. z ∈ Ω, all x ∈ R.
2
 SUPERLINEAR PERTURBATIONS OF THE EIGENVALUE PROBLEM 13

Let u ∈ Ĥm+2 . We have

1 λ ε
ϕλ (u) > γ(u) − kuk22 − kuk2 − Ĉε kukr
2 2 2
for some Ĉε > 0 (see (38))
C13 − ε
> kuk2 − Ĉε kukr for some C13 > 0 (recall λ < λ̂m+2 ).
2
Choose ε ∈ (0, C13 ). Then we obtain
ϕλ (u) > C14 kuk2 − Ĉε kukr for some C14 > 0, all u ∈ Ĥm+2 .
Since 2 < r, we can find ρ ∈ (0, 1) small such that
ϕλ (u) > C̃0 > 0 for all u ∈ Ĥm+2 ∩ ∂Bρ .

Let ûm+2 ∈ E(λ̂m+2 ) with kûm+2 k = 1 and let V = H m+1 ⊕ Rûm+2 , with H m+1 =
m+1
⊕ E(λ̂k ). For ρ1 > 0, we introduce the set
k=1

C = {u = u + ϑûm+2 : u ∈ H m+1 , ϑ > 0, kuk 6 ρ1 }.

Evidently we have



∂C = C0 = u = u + ϑûm+2 : u ∈ H m+1 , ϑ > 0, kuk = ρ1 or u ∈ H m+1 , kuk 6 ρ1 , ϑ = 0

Proposition 4. If hypotheses H0 , H1 (i), (ii), (iv), (v) hold and λ ∈ (λ̂m , λ̂m+1 ), then there
exist ρ1 > 0 and δ̃ > 0 such that
ϕλ 6 C̃1 < C̃0
C0

with C̃0 > 0 as in Proposition 3.

Proof. From hypotheses H1 (i), (ii), (v) given η > 0, we can find Ĉη∗ > 0 such that
η
(39) F (z, x) > x2 − Ĉη∗ |x|q for a.a. z ∈ Ω, all x ∈ R.
2
The space V is finite dimensional and so all norms are equivalent. Let u ∈ V . We have
1 λ 2 η 2
Ĉη∗
ϕλ (u) 6 γ(u) − kuk2 − kuk2 + kukqq (see (39))
2 2 2 q
h i
6 C15 λ̂m+2 − λ − η kuk2 + C16 kukq for some C15 , C16 = C16 (η) > 0.
Since η > 0 arbitrary, choosing η > 0 big, we have
ϕλ (u) 6 C16 kukq − C17 kuk2 for some C17 > 0.
Recall that q > 2. Then we can find ρ1 ∈ (0, 1) small such that

ϕλ 6 0 < C̃0 (see Proposition 3).

V ∩∂Bρ

If u ∈ H m+1 , kuk 6 ρ1 , then

1 λ
ϕλ (u) 6 γ(u) − kuk22 + C ∗ kukqq
2 2
1h i
6 λ̂m+1 − λ kuk22 + C ∗ kukqq (see H1 (v))
2
6 C18 ρ21 (since q > 2,λ < λ̂m+1 and ρ1 ∈ (0, 1)).
14 N.S. PAPAGEORGIOU, V.D. RĂDULESCU, AND D.D. REPOVŠ

Choosing ρ1 ∈ (0, 1) even smaller if necessary, we have

ϕλ 6 C̃1 < C̃0

H m+1 ∩∂Bρ1

for all λ ∈ (λ̂m , λ̂m+1 ) and with C̃0 > 0 (as in Proposition 3).
Therefore we conclude that
ϕλ 6 C̃1 < C̃0 for λ ∈ (λ̂m , λ̂m+1 ).
C0

Now we are ready to produce the seventh nontrivial smooth solution of (Pλ ).
Proposition 5. If hypotheses H0 , H1 hold and λ ∈ (λ̂m+1 − δ̂, λ̂m+1 ) (see Proposition 2),
then problem (Pλ ) has a seventh nontrivial solution ỹ ∈ C 1 (Ω).
Proof. Let D = H m+1 ∩ ∂Bρ1 . From Proposition 6.6.5, p. 532, of Papageorgiou-Rădulescu-
Repovš [13], we know that
{C, C0 } and D homologically link in dimension dm+1 + 1.
with dm+1 = dim H m+1 . Then Propositions 3 and 4 and Corollary 6.6.8 of [13], imply that
there exists ỹ ∈ Kϕλ ⊆ C 1 (Ω) (see Wang [17]) such that
(40) Cdm+1 +1 (ϕλ , ỹ) 6= 0.
From the proof of Proposition 1, we know that u0 ∈ intC+ and v0 ∈ −intC+ are local
−
minimizers of ϕ+
λ and of ϕλ respectively. Note that

(41) ϕλ = ϕ+
λ and ϕλ = ϕ+
λ .
C+ C+ −C+ −C+

So, it follows that u0 ∈ intC+ and v0 ∈ −intC+ are also local minimizers of ϕλ (see [9]).
Therefore we have
(42) Ck (ϕλ , u0 ) = Ck (ϕλ , v0 ) = δk,0 Z for all k ∈ N0 .
Also, again from the proof of Proposition 1, we know that the solutions û ∈ intC+
−
and v̂ ∈ −intC+ are critical points of mountain pass type of the functionals ϕ+
λ and ϕλ
respectively. Therefore we have
−
(43) C1 (ϕ+
λ , û) 6= 0 and C1 (ϕλ , v̂) 6= 0(see (36)).
From (41) and since û ∈ intC+ , v̂ ∈ −intC+ , we have
       
−
(44) Ck ϕ+ λ 1
, û = C k ϕλ 1 , û and C k ϕλ , v̂ = Ck ϕλ , v̂
C (Ω) C (Ω) C 1 (Ω) C 1 (Ω)

for all k ∈ N0 .
But on account of Theorem 6.6.26, p. 545, of Papageorgiou-Rădulescu-Repovš [13], we
have
−
(45) Ck (ϕ+
λ , û) = Ck (ϕλ , û) and Ck (ϕλ , v̂) = Ck (ϕλ , v̂)
for all k ∈ N0 .
Since ϕλ ∈ C 2 (H 1 (Ω)), from (42), (43), (45) and Proposition 6.5.9, p. 529, of Papageorgiou-
Rădulescu-Repovš [13], we infer that
(46) Ck (ϕλ , û) = Ck (ϕλ , v̂) = δk,1 Z for all k ∈ N0 .
Recall that
(47) Ck (ϕλ , 0) = δk,dm Z (see (35)).
 SUPERLINEAR PERTURBATIONS OF THE EIGENVALUE PROBLEM 15

Moreover from Corollary 6.2.40, p. 449, of Papageorgiou-Rădulescu-Repovš [13], we have

(48) Ck (ϕλ , y0 ) = Ck (ϕλ , ŷ) = 0 for k 6∈ [dm , dm+1 ] (recall that dm > 2).
From (40), (42), (46), (47), (48), we infer that
ỹ 6∈ {u0 , v0 , û, v̂, 0, y0 , ŷ},
⇒ ỹ ∈ C 1 (Ω) is the seventh nontrivial solution of (Pλ )
(λ ∈ (λ̂m+1 − δ̂, λ̂m+1 )).

So, summarizing our findings for problem (Pλ ), we can state the following multiplicity
theorem.
Theorem 6. If hypotheses H0 , H1 hold, then there exists δ̂ > 0 such that for all λ ∈
(λ̂m+1 − δ̂, λ̂m+1 ) problem (Pλ ) has at least seven distinct nontrivial smooth solutions
u0 , û ∈ intC+ , v0 , v̂ ∈ −intC+ , y0 , ŷ ∈ C 1 (Ω) nodal
ỹ ∈ C 1 (Ω).
Remarks: Is it possible to show that ỹ is nodal (see [3], [8]). Also, it seems that we can
not generate more that seven solutions without symmetry hypotheses (see [1]).

References
[1] A. Castro-J. Cassio-C. Velez: ”Existence of seven solutions for an asymptotically linear Dirichlet
problem without symmetries” Annali Mat. Pura Appl. 192 (2013), 607-619.
[2] G. D’Agui-S. Marano-N.S. Papageorgiou: ”Multiple solutions to a Robin problem with indefinite
weight and asymmetric reaction” J. Math. Anal. Appl 433 (2016), 1821-1845.
[3] S. Hu-N.S. Papageorgiou: ”Semilinear Robin problems with indefinite potential and competition
phenomena” Acta Appl. Math. DOI 10.1007/s10440-019-00284-y.
[4] A. Marino-C. Saccon: ”Some variational theorems of mixed type and elliptic problems with
jumping nonlinearities” Ann. Scuola Norm. Sup. Pisa, Cl. Sci 25 (1997), 631-665.
[5] D. Mugnai: ”Multiplicity of critical points in presence of linking: application to a superlinear
boundary value problem” Nonlin. Differ. Equ. Appl. (NoDEA) 11 (2004), 379-391.
[6] D. Mugnai: ”Addendum to: Multiplicity of critical points in presence of linking: application
to a superlinear boundary value problem” Nonlin. Differ. Equ. Appl. (NoDEA) 11 (2004), no.
3,379-391 and a comment on the generalized Ambroseti-Rabinowitz condition” Nonlin. Differ.
Equ. Appl. (NoDEA) 19 (2012), 299-311.
[7] Z.Q. Ou-C. Li: ”Existence of three nontrivial solutions for a class of superlinear elliptic equa-
tions” J. Math. Anal. Appl. 390 (2012), 418-426.
[8] N.S. Papageorgiou-F. Papalini: ”Seven solutions with sign information for sublinear equations
with unbounded and indefinite potential and no symmetries” Israel J. Math. 201 (2014), 761-
796.
[9] N.S. Papageorgiou-V.D. Rădulescu: ”Multiple solutions with precise sign for nonlinear para-
metric Robin problems”, J. Differential Equ. 256 (2014), 2449-2479.
[10] N.S. Papageorgiou-V.D. Rădulescu: ”Robin problems with indefinite unbounded potential and
reaction of arbitrary growth” Rev. Mat. Complut. 19 (2016), 91-126.
[11] N.S. Papageorgiou-V.D. Rădulescu: ”Robin problems near resonance at any nonprincipal eigen-
value” Results Math. 71 (2017), 1389-1412.
[12] N.S. Papageorgiou-V.D. Rădulescu-D.D. Repovš: ”Positive solutions for perturbations of the
Robin eigenvalue problem plus an indefinite potential” Discr. Cont. Dyn. Systems 37 (2017),
2589-2618.
[13] N.S. Papageorgiou-V.D. Rădulescu-D.D. Repovš: ”Nonlinear Analysis-Theory and Methods”
Springer Nature, Switerland, 2019.
[14] N.S Papageorgiou-P. Winkert: ”Applied Nonlinear Functional Analysis” W. De Gruyter, Berlin,
2018.
16 N.S. PAPAGEORGIOU, V.D. RĂDULESCU, AND D.D. REPOVŠ

[15] P. Rabinowitz-J. Su-Z.Q. Wang: ”Multiple solutions of superlinear elliptic equations” Rend.
Lincei Mat. Appl. 18 (2007), 97-108.
[16] J. Su-R. Zeng: ”Multiple periodic solutions of superlinear ordinary differential equations with
a parameter” Nonlinear Anal. 74 (2011), 6442-6450.
[17] S. Wang: ”Neumann problems of semilinear elliptic equations involving critical Sobolev expo-
nents” J. Differential Equ. 93 (1991), 283-310.

(N.S. Papageorgiou) National Technical University, Department of Mathematics, Zografou

Campus, Athens 15780, Greece
Email address: npapg@math.ntua.gr

(V.D. Rădulescu) Faculty of Applied Mathematics, AGH University of Science and Technol-
ogy, al. Mickiewicza 30, 30-059 Kraków, Poland & Department of Mathematics, University
of Craiova, 200585 Craiova, Romania
Email address: radulescu@inf.ucv.ro

(D.D. Repovš) University of Ljubljana, Faculty of Education and Faculty of Mathematics

and Physics, Kardeljeva ploscad 16, SI-1000 Ljubljana, SLOVENIA
Email address: dusan.repovs@guest.arnes.si