Landesman–Lazer condition revisited: the influence of vanishing and oscillating nonlinearities
Pavel Drábek
B1, 2and Martina Langerová
21Department of Mathematics, University of West Bohemia, Univerzitní 8, 306 14 Plze ˇn, Czech Republic
2NTIS, University of West Bohemia, Technická 8, 306 14 Plze ˇn, Czech Republic
Received 7 September 2015, appeared 20 October 2015 Communicated by László Simon
Abstract. In this paper we deal with semilinear problems at resonance. We present a sufficient condition for the existence of a weak solution in terms of asymptotic proper- ties of nonlinearity. Our condition generalizes the classical Landesman–Lazer condition and it also covers the cases of vanishing and oscillating nonlinearities.
Keywords: resonance problem, semilinear equation, Landesman–Lazer condition, sad- dle point theorem, critical points.
2010 Mathematics Subject Classification: 35J20, 35J25, 35B34, 35B38.
1 Introduction
Let Ω ⊆ Rn be a bounded domain, g: R → R be a bounded continuous function and f ∈ L2(Ω). We consider the boundary value problem
−∆u−λku+g(u) = f inΩ,
u=0 on ∂Ω. (1.1)
Hereλk, k≥1, is thek-th eigenvalue of the eigenvalue problem
−∆u−λu=0 inΩ,
u=0 on∂Ω. (1.2)
By a solutionof (1.1) we understand a functionu∈ H:=W01,2(Ω)satisfying (1.1) in the weak sense,i.e.,
Z
Ω
∇u∇vdx−λk Z
Ω
uvdx+
Z
Ω
g(u)vdx=
Z
Ω
f vdx (1.3)
holds for any test functionv∈ H.
BCorresponding author. Email: pdrabek@kma.zcu.cz
Let m≥ 1 be a multiplicity of λk. We arrange the eigenvalues of (1.2) into the increasing sequence:
0<λ1 <λ2≤ · · · ≤λk−1 <λk =· · · =λk+m−1 <λk+m ≤λk+m+1 ≤ · · · →∞.
The corresponding eigenfunctions,(φn), form an orthogonal basis for both L2(Ω)andH. We assume that everyφn is normalized with respect to theL2norm,i.e., kφnk2 =1, n= 1, 2, . . . . We use the scalar product(u,v) =R
Ω∇u∇vdxand the induced normkuk= R
Ω|∇u|2dx1/2
on H. We split the spaceHinto the following three subspaces spanned by the eigenfunctions of (1.2) as follows:
Hˆ := [φ1, . . . ,φk−1], H¯ := [φk, . . . ,φk+m−1], H˜ := [φk+m,φk+m+1, . . .].
ThenH = Hˆ ⊕H¯ ⊕H˜ with dim ˆH= k−1, dim ¯H= m, dim ˜H= ∞. Of course, ifk =1 then m= 1 (λ1 is a simple eigenvalue) and ˆH = ∅. We split an elementu ∈ H as u = uˆ+u¯+u,˜
ˆ
u∈ H, ¯ˆ u∈ H¯ and ˜u∈ H. We split a function˜ f ∈ L2(Ω)as f = f¯+ f⊥, where R
Ω f⊥vdx= 0 for anyv∈ H.¯
The purpose of this paper is to introduce a rather general sufficient condition of the Landesman–Lazer type for the existence of a solution of (1.1).
If(un)⊂ H is a sequence such thatkunk2 →∞and there existsφ0 ∈ H,¯ kuun
nk2 → φ0 in L2(Ω), then
nlim→∞
Z
Ω
G(un)dx−
Z
Ω
f u¯ ndx
= ±∞. (SC)±
Here,G(s) =Rs
0 g(τ)dτis the antiderivative ofg.
Theorem 1.1. Assume that either(SC)+or else(SC)−holds. Then the problem(1.1)has at least one solution.
Remark 1.2. Note that the sufficient condition which is similar to(SC)+ but more restrictive than(SC)+ was introduced recently in [8] where the resonance problem with respect to the Fuˇcík spectrum of the Laplacian was studied. In this paper, we benefit from the fact that the resonance occurs at the eigenvalue which allows us to split the underlying function space H into the sum of orthogonal subspaces. In contrast with [8], where such splitting is impossible, we can get rid of the f⊥-part of the right-hand side f in (SC)±. This makes our conditions more general and geometrically more transparent.
In the following we interpret our conditions(SC)±in historical context. We first consider a bounded continuous nonlinear functiong: R→Rwith finite limitsg(±∞):=lims→±∞g(s). Example 1.3. Let us assume that
g(∓∞)
Z
Ω
φ+dx−g(±∞)
Z
Ω
φ−dx<
Z
Ω
f¯φdx
< g(±∞)
Z
Ω
φ+dx−g(∓∞)
Z
Ω
φ−dx
(LL)±
holds for all eigenfunctions φ associated with λk. This is the classical Landesman–Lazer condition (see [14]). Assumekunk2 → ∞and kun
unk2 →φ0 for some eigenfunctionφ0. Then by
l’Hospital’s rule we have
nlim→∞
1 kunk2
Z
Ω
G(un)dx−
Z
Ω
f u¯ ndx
= lim
n→∞ Z
Ω
G(un) un − f¯
un kunk2 dx
=
Z
Ω
g(+∞) + f¯
φ+0 dx−
Z
Ω
g(−∞) + f¯ φ0−dx.
The last expression is either positive or negative due to (LL)± and hence conditions (SC)± hold. In other words, we proved that(LL)±imply(SC)±.
Assume, moreover,g(−∞)<0<g(+∞)(think, for example, aboutg(s) =arctans). Then problem (1.1) has a solution for all f which belong to the “strip” (given by inequalities(LL)+) around the linear subspace
L2(Ω)⊥:=
f ∈ L2(Ω): Z
Ω
fφdx=0 for allφ∈ H¯
of L2(Ω).
We note that the conditions (LL)± are empty if g(−∞) = g(+∞). However, we prove existence results even in this case.
Example 1.4. It follows from Theorem 1.1 that the problem (1.1) with g(s) = (e+|s|)sgnln(se+|s|) (eis Euler’s number) has at least one solution for f ∈L2(Ω)⊥. Indeed,
|slim|→∞G(s) = lim
|s|→∞ln(ln(e+|s|)) =∞
implies that (SC)+ holds true. Hence, our conditions(SC)± cover the case ofvanishing non- linearitiesg(±∞) =0 (see [7]). It should be emphasized, that in contrast with previous works on vanishing nonlinearities our approach does not require any kind of symmetry or sign condition about g(cf. [2–4,9,11,13]). At the same time, it generalizes the results from [10,12].
The verification of (SC)± does not require the existence of limits g(±∞) at all. See the following example.
Example 1.5. Consider g(s) = arctans+c·coss with an arbitrary constant c ∈ R. An easy calculation yields that (SC)+ is satisfied, and hence, according to Theorem1.1, problem (1.1) has at least one solution for any f ∈ L2(Ω)satisfying
Z
Ω
fφdx
< π 2
Z
Ω
|φ|dx (1.4)
for anyφ∈ H. On the other hand, the conditions¯ (LL)± and various generalizations of(LL)± (see, e.g. [5,6]) do not apply if |c| ≥ π2, due to the fact that these conditions are vacuous in this case.
Remark 1.6. In fact, the above mentioned case g(s) = arctans+c·coss is covered by the so called potential Landesman–Lazer condition:
G∓ Z
Ω
φ+dx−G± Z
Ω
φ−dx <
Z
Ω
f¯φdx< G± Z
Ω
φ+dx−G∓ Z
Ω
φ−dx (PLL)±
where G± := lims→±∞ G(s)
s . Indeed, l’Hospital’s rule implies G− = −π2, G+ = π2 and the condition (PLL)+ reduces to (1.4). For the use of (PLL)± see, e.g. the papers [1,18–22].
The conditions(PLL)± eliminate the influence of the boundedoscillating termc·coss which disappears “in an average” as|s| →∞.
However, the conditions (PLL)± do not cover the case like g(s) = s
1+s2 +c·coss, where c∈Ris an arbitrary constant. Indeed, both conditions are empty due to the factG± =0. On the other hand, it follows from our Theorem1.1 that (1.1) with g given above has a solution for any f ∈ L2(Ω)⊥. This fact illustrates that our conditions(SC)±refine also the conditions (PLL)±and complement the results from [15] and [16].
In the following example we treat rather general nonlinearity.
Example 1.7. It follows from our Theorem1.1that the boundary value problem
−∆u−λku+ u
(e+u2)ln(e+u2)1/2 +c·cosu= f inΩ, u=0 on∂Ω,
(1.5)
has a solution for arbitraryc∈Rand for any f ∈ L2(Ω)satisfying Z
Ω
fφdx =0
for allφ∈ H. Indeed, since¯
|slim|→∞G(s) = lim
|s|→∞
h ln
ln(e+s2)1/2+c·sinsi
= ∞
the condition(SC)+is satisfied. The existence result for problems of type (1.5) does not follow from any work published in the literature so far.
Examples and remarks presented above justify the novelty of our work and show that our conditions(SC)±are new and recover previously published results.
2 Preliminaries
In this section we stress some helpful facts used in the proof of Theorem1.1.
Lemma 2.1. There exist c1>0, c2 >0such that for any u∈ H we have Z
Ω
|∇uˆ|2dx−λk Z
Ω
(uˆ)2dx≤ −c1kuˆk2 (2.1)
and
Z
Ω
g(u)uˆdx−
Z
Ω
fuˆdx
≤c2kuˆk. (2.2)
Proof. The inequality (2.1) follows from the variational characterization of λk, (2.2) follows from the Hölder inequality, the boundedness ofgand the fact f ∈ L2(Ω).
Lemma 2.2. There exist c3>0, c4>0such that for any u∈ H we have Z
Ω
|∇u˜|2dx−λk Z
Ω
(u˜)2dx≥c3ku˜k2 (2.3)
and
Z
Ω
g(u)u˜dx−
Z
Ω
fu˜dx
≤ c4ku˜k. (2.4)
Proof. The inequality (2.3) is also a consequence of the variational characterization of λk, and (2.4) follows similarly as (2.2).
Lemma 2.3. There exist c5>0such that for any u∈ H we have
Z
Ω
G(u)dx−
Z
Ω
f udx
≤c5kuk2. (2.5)
Proof. The inequality (2.5) follows from the Hölder inequality, the boundedness of g and the fact f ∈ L2(Ω).
3 Proof of Theorem 1.1
We define theenergy functionalassociated with (1.1),E: H→R, by E(u):= 1
2 Z
Ω
|∇u|2dx− λk 2
Z
Ω
(u)2dx+
Z
Ω
G(u)dx−
Z
Ω
f udx, u∈H. Obviously, all critical points ofE satisfy (1.3) and vice versa.
We apply the Saddle Point Theorem due to P. Rabinowitz [17] to prove the existence of a critical point ofE.
Theorem 3.1. LetE ∈C1(H,R)and H=H−⊕H+,dimH− <∞,dimH+= ∞. Assume that (a) there exist a bounded neighborhood D of o in H−and a constantα∈Rsuch thatE
∂D ≤α;
(b) there exists a constant β> αsuch thatE
H+ ≥β;
(c) E satisfies (PS)condition, i.e., if(E(un))⊂ Ris a bounded sequence and∇E(un)→ o in H, then there exist a subsequence(unk)⊂(un)and an element u∈ H such that unk →u in H.
Then the functionalE has a critical point in H.
At first we verify the Palais–Smale condition.
Lemma 3.2. Let us assume(SC)±. ThenE satisfies(PS)condition.
Proof. In the first step we prove that (un) is bounded in L2(Ω). Assume the contrary, i.e., kunk2 →∞. Setvn:= kuun
nk2. Then E(un)
kunk22 := 1 2
Z
Ω
|∇vn|2dx− λk 2
Z
Ω
(vn)2dx+
Z
Ω
G(un)
kunk22 dx− 1 kunk2
Z
Ω
f vndx →0. (3.1)
The second term is equal to −λ2k since kvnk2 = 1, the last two terms go to zero due to Lemma2.3.
Then it follows from (3.1) that(vn)is a bounded sequence inH. Passing to a subsequence, if necessary, we may assume that there exists v ∈ H such that vn * v (weakly) in H and vn →vin L2(Ω).
For arbitraryw∈ H,
0← (∇E0(un),w) kunk2 =
Z
Ω
∇vn∇wdx−λk Z
Ω
vnwdx
+ 1
kunk2
Z
Ω
g(un)wdx− 1 kunk2
Z
Ω
f wdx.
(3.2)
We haveR
Ω∇vn∇wdx →R
Ω∇v∇wdx byvn *v in H,R
Ωvnwdx →R
Ωvwdx byvn →v in L2(Ω), ku1
nk2
R
Ω f wdx → 0 and ku1
nk2
R
Ωg(un)wdx → 0 by f ∈ L2(Ω), the boundedness of g and by our assumptionkunk2 →∞. Then it follows from (3.2) that
Z
Ω
∇v∇wdx−λk Z
Ω
vwdx=0
holds for arbitrary w ∈ H, i.e., v = φ0 ∈ H¯ is an eigenfunction associated with λk. That is,
un
kunk2 →φ0in L2(Ω).
Now, by the assumption∇E(un)→oand the orthogonal decomposition of H, we have o(kuˆnk) = (∇E(un), ˆun) =
Z
Ω
|∇uˆn|2dx−λk Z
Ω
(uˆn)2dx+
Z
Ω
g(un)uˆndx−
Z
Ω
fuˆndx. (3.3) By Lemma2.1, it follows from (3.3) that
o(1)≤ −c1kuˆnk+c2
withc1,c2 >0 independent ofn. Hencekuˆnkis a bounded sequence.
Similarly, we also have o(ku˜nk) = (∇E(un), ˜un) =
Z
Ω
|∇u˜n|2dx−λk Z
Ω
(u˜n)2dx+
Z
Ω
g(un)u˜ndx−
Z
Ω
fu˜ndx. (3.4) By Lemma2.2, it follows from (3.4) that
o(1)≥c3ku˜nk −c4
withc3,c4> 0 independent ofn. Henceku˜nkis a bounded sequence. Let us split nowE(un) as follows
E(un) = 1 2
Z
Ω
|∇uˆn|2dx−λk 2
Z
Ω
(uˆn)2dx
| {z }
A
+1 2
Z
Ω
|∇u˜n|2dx−λk 2
Z
Ω
(u˜n)2dx
| {z }
B
+
Z
Ω
G(un)dx−
Z
Ω
f u¯ ndx
| {z }
C
−
Z
Ω
f⊥uˆndx−
Z
Ω
f⊥u˜ndx
| {z }
D
.
The boundedness of kuˆnk and ku˜nk implies that A,B and D are bounded terms. On the other hand, (SC)+ forces C → +∞ and (SC)− forces C → −∞. In particular, we conclude E(un) → ±∞ which contradicts the assumption of the boundedness of (E(un)). We thus proved that(un)is a bounded sequence inL2(Ω).
In the second step we select a strongly convergent subsequence (in H) from (un). Let us examine again the terms in
E(un):= 1 2
Z
Ω
|∇un|2dx− λk 2
Z
Ω
(un)2dx+
Z
Ω
G(un)dx−
Z
Ω
f undx.
By the assumption (E(un)) is bounded. The boundedness of the sequence (un) in L2(Ω) implies thatR
Ω(un)2dx,R
ΩG(un)dx andR
Ω f undxare bounded independently of n, as well.
Therefore, kunk2 =R
Ω|∇un|2dxmust be also bounded. Hence, we may assume, without lost of generality, that un*u inH for someu∈ H, andun→uin L2(Ω). Then
0←(∇E(un),un−u) =
Z
Ω
∇un∇(un−u)dx−λk Z
Ω
un(un−u)dx +
Z
Ω
g(un)(un−u)dx−
Z
Ω
f(un−u)dx.
Since
−λk Z
Ω
un(un−u)dx+
Z
Ω
g(un)(un−u)dx−
Z
Ω
f(un−u)dx→0,
we conclude that Z
Ω
∇un∇(un−u)dx→0, as well. So,
Z
Ω
|∇un|2dx−
Z
Ω
∇un∇udx→0 which together with
Z
Ω
∇un∇udx− kunk2 →0 (due to the weak convergenceun*u) yields
kunk → kuk.
The uniform convexity of H then implies that un → u in H. Hence E satisfies the condition (c)in Theorem3.1.
Now we prove that also the hypotheses(a)and(b)hold. To this end we have to consider separately the case(SC)+ and(SC)−.
1. Let us assume that(SC)+ holds. We set
H−:= H,ˆ H+ := H¯ ⊕H.˜ It follows from Lemmas2.1and2.3that
kulimˆk→∞E(uˆ):= lim
kuˆk→∞
1 2 Z
Ω
|∇uˆ|2dx− λk 2
Z
Ω
(uˆ)2dx+
Z
Ω
G(uˆ)dx−
Z
Ω
fuˆdx
= −∞ (3.5)
for ˆu∈ H.ˆ
On the other hand, we prove that there existsβ∈Rsuch that
uinf∈H+E(u)≥ β.
Assume the contrary,i.e., there exists a sequence(un)⊂ H+such that
nlim→∞E(un) =−∞. (3.6)
Thenkunk2 →∞, and forvn:= kuun
nk2 (vn∈ H+) we have 0≥lim sup
n→∞
E(un)
kunk22 :=lim sup
n→∞
1 2
Z
Ω
|∇vn|2dx− λk 2
Z
Ω
(vn)2dx
+
Z
Ω
G(un) kunk22 dx−
Z
Ω
f vn kunk2dx
.
(3.7)
Clearly, by Lemma2.3, we have Z
Ω
G(un) kunk22 dx−
Z
Ω
f vn
kunk2dx→0. (3.8)
It follows from (3.7) and (3.8) that kvnkis a bounded sequence. Passing to a subsequence, if necessary, we may assume that there exists v ∈ H+ such that vn * v in H and vn → v in L2(Ω). Moreover,
lim inf
n→∞ Z
Ω
|∇vn|2dx≥
Z
Ω
|∇v|2dx (3.9)
by the weak lower semicontinuity of the norm in H. We deduce from (3.7)–(3.9) that Z
Ω
|∇v|2dx−λk Z
Ω
(v)2dx ≤0,
and hence, from Lemma 2.2, it follows that v = φ0 ∈ H¯ is an eigenfunction associated with λk. That is,
un
kunk2 →φ0 in L2(Ω).
By Lemma 2.2, by the properties of the orthogonal decomposition of H+ and f and by the condition(SC)+, we have forun ∈ H+
nlim→∞E(un):= lim
n→∞
1 2
Z
Ω
|∇un|2dx− λk 2
Z
Ω
(un)2dx +
Z
Ω
G(un)dx−
Z
Ω
f undx
= lim
n→∞
1 2
Z
Ω
|∇u˜n|2dx−λk 2
Z
Ω
(u˜n)2dx+
Z
Ω
G(un)dx−
Z
Ω
f u¯ ndx−
Z
Ω
f⊥u˜ndx
≥ lim
n→∞
h
c3ku˜nk2− kf⊥k2ku˜nk2i+ lim
n→∞
Z
Ω
G(un)dx−
Z
Ω
f u¯ ndx
= +∞.
This contradicts (3.6). By (3.5) there exists R> 0 such that for D:= {u ∈ H− :kuk ≤ R}the following inequality holds
sup
u∈∂D
E(u)<α:= β−1.
Hence, we proved that the hypotheses (a) and (b) in Theorem3.1hold.
2. Let us assume that(SC)− holds. In this case we set H−:= Hˆ ⊕H,¯ H+ := H.˜ Letu∈ H+. Then by Lemmas2.2and2.3, we have
E(u):= 1 2
Z
Ω
|∇u|2dx− λk 2
Z
Ω
(u)2dx+
Z
Ω
G(u)dx−
Z
Ω
f udx
≥c3kuk2−c5kuk2≥ c3kuk2−c6kuk. Hence, there exists β∈Rsuch thatE(u)≥ βfor allu∈ H+.
On the other hand, we prove that
kuk→lim∞,u∈H−
E(u) =−∞. (3.10)
Notice, that dimH−<∞implies that the normsk · kandk · k2are equivalent onH−. Assume, by contradiction, that (3.10) does not hold,i.e., there exist a sequence(un)⊂ H−and a constant c∈Rsuch thatkunk2→∞and
E(un)≥ c. (3.11)
Set vn := kuun
nk2. Due to dimH− < ∞ we may assume that there exists v ∈ H− such that vn→vboth in HandL2(Ω). Then
0≤ lim inf
n→∞
E(un) kunk22
= lim inf
n→∞
1 2
Z
Ω
|∇vn|2dx−λk 2
Z
Ω
(vn)2dx+
Z
Ω
G(un) kunk22 dx−
Z
Ω
f vn kunk2 dx
= 1 2
Z
Ω
|∇v|2dx−λk 2
Z
Ω
(v)2dx,
(3.12)
by Lemma2.3. According to Lemma2.1, (3.12) impliesv = φ0 ∈ H, an eigenfunction associ-¯ ated with λk. Hence kuun
nk2 → φ0 in L2(Ω). It follows from the orthogonal decomposition of H−and f, Lemma2.1 and(SC)− that forun∈ H−
nlim→∞E(un):= lim
n→∞
1 2
Z
Ω
|∇un|2dx −λk 2
Z
Ω
(un)2dx+
Z
Ω
G(un)dx−
Z
Ω
f undx
= lim
n→∞
1 2
Z
Ω
|∇uˆn|2dx−λk 2
Z
Ω
(uˆn)2dx+
Z
Ω
G(un)dx−
Z
Ω
f u¯ ndx−
Z
Ω
f⊥uˆndx
≤ lim
n→∞
−c1kuˆnk2+c2kuˆnk+ lim
n→∞
Z
Ω
G(un)dx−
Z
Ω
f u¯ ndx
=−∞.
This contradicts (3.11),i.e., (3.10) holds true. Let us choose again D:= {u ∈ H− : kuk ≤ R}. Then, forR>0 large enough, we have
sup
u∈∂D
E(u)< α:=β−1.
Thus, the hypotheses (a) and (b) in Theorem3.1are satisfied.
Recall that the hypothesis (c) in Theorem3.1is proved in Lemma3.2for both cases(SC)±. It then follows from Theorem3.1that under the assumptions(SC)±there exists a critical point ofE. Since this is also a solution of (1.1), the proof of Theorem1.1 is finished.
Acknowledgements
This research was supported by the Grant 13-00863S of the Grant Agency of Czech Republic and by the project LO1506 of the Czech Ministry of Education, Youth and Sports.
References
[1] D. A. Bliss, J. Buerger, A. J. Rumbos, Periodic boundary-value problems and Dancer–
Fuˇcík spectrum under conditions of resonance, Electron. J. Differential Equations 2011, No. 112, 1–34.MR2836793
[2] A. Cañada, k-set contractions and nonlinear vector boundary value problems, J. Math.
Anal. Appl.117(1986), No. 1, 1–22.MR0843001
[3] P. Drábek, Bounded nonlinear perturbations of second order linear elliptic problems, Comment. Math. Univ. Carolin.22(1981), No. 2, 215–221.MR0620358
[4] P. Drábek, Existence and multiplicity results for some weakly nonlinear elliptic problems at resonance,Casopis Pˇest. Mat. (Math. Bohemica)ˇ 108(1983), No. 3, 272–284.MR0711264 [5] P. Drábek, On the resonance problem with nonlinearity which has arbitrary linear
growth,J. Math. Anal. Appl.127(1987), 435–442.MR0915069
[6] P. Drábek, Landesman–Lazer condition for nonlinear problems with jumping nonlinear- ities,J. Differential Equations85(1990), 186–199.MR1052334
[7] P. Drábek,Solvability and bifurcations of nonlinear equations, Longman Scientific & Techni- cal, Pitman Research Notes in Mathematics Series, Vol. 264, Harlow, 1992.MR1175397 [8] P. Drábek, S. B. Robinson, On the solvability of resonance problems with respect to the
Fuˇcík Spectrum,J. Math. Anal. Appl.418(2014), 884–905.MR3206687
[9] D. G. de Figueiredo, W. M. Ni, Perturbations of second order linear elliptic problems by nonlinearities without Landesman–Lazer condition, Nonlinear Anal.3(1979), 629–634.
MR0541873
[10] S. Fu ˇcík, M. Krbec, Boundary value problems with bounded nonlinearity and general nullspace of linear part,Math. Z.155(1977), 129–138.MR0473513
[11] C. P. Gupta, Solvability of a boundary value problem with the nonlinearity satisfying a sign condition,J. Math. Anal. Appl.129(1988), 482–492.MR0924305
[12] P. Hess, A remark on the preceding paper of Fuˇcík and Krbec, Math. Z. 155(1977), 139–141.MR0473514
[13] R. Iannacci, M. N. Nkashama, J. R. Ward, Nonlinear second order elliptic partial dif- ferential equations at resonance,Trans. Amer. Math. Soc.311(1989), 711–726.MR0951886 [14] E. M. Landesman, A. C. Lazer, Nonlinear perturbations of linear elliptic boundary value
problems at resonance,J. Math. Mech.19(1970), 609–623.MR0267269
[15] M. N. Nkashama, S. B. Robinson, Resonance and nonresonance in terms of average values,J. Differential Equations132(1996), 46–65.MR1418499
[16] M. N. Nkashama, S. B. Robinson, Resonance and non-resonance in terms of average values. II,Proc. Roy. Soc. Edinburgh Sect. A131(2001), No. 5, 1217–1235.MR1862451 [17] P. H. Rabinowitz, Minimax methods in critical point theory with applications to differential
equations, Amer. Math. Soc., Providence, RI, 1986.MR0845785
[18] C. L. Tang, Solvability for two-point boundary value problems, J. Math. Anal. Appl.
216(1997), 368–374.MR1487269
[19] P. Tomiczek, A generalization of the Landesman–Lazer condition,Electron. J. Differential Equations2001, No. 4, 1–11.MR1811777
[20] P. Tomiczek, Potential Landesman–Lazer type conditions and the Fucik spectrum, Elec- tron. J. Differential Equations2005, No. 94, 1–12.MR2162255
[21] P. Tomiczek, The Duffing equation with the potential Landesman–Lazer condition,Non- linear Anal.70(2009), 735–740.MR2468415
[22] P. Tomiczek, Periodic problem with a potential Landesman Lazer condition,Bound. Value Probl.2010, Art. ID 586971, 8 pp.MR2728256;url