A positive solution of
asymptotically periodic Schrödinger equations with local superlinear nonlinearities
Gui-Dong Li, Yong-Yong Li and Chun-Lei Tang
BSchool of Mathematics and Statistics, Southwest University, Chongqing, 400715, China Received 31 October 2019, appeared 11 May 2020
Communicated by Dimitri Mugnai
Abstract. In this paper, we investigate the following Schrödinger equation
−∆u+V(x)u=λf(u) inRN,
whereN≥3,λ>0,Vis an asymptotically periodic potential and the nonlinearity term f(u)is only locally defined for|u|small and satisfies some mild conditions. By using Nehari manifold and Moser iteration, we obtain the existence of positive solutions for the equation with sufficiently largeλ.
Keywords: Schrödinger equation, positive solution, locally defined nonlinearity, asymptotically periodic potential.
2020 Mathematics Subject Classification: 35J50, 35A01, 35B09, 35D30.
1 Introduction
In recent years, many researchers consider the following Schrödinger equation
−∆u+V(x)u= f(x,u), inRN, (1.1) where N ≥ 3, V is a given potential and f ∈ C(RN×R,R). Knowledge of the solutions of Eq. (1.1) has a great importance for studying standing wave solutions for
ih∂Ψ
∂t =−h2∆Ψ+W(x)Ψ− f(x,Ψ), for all x∈Ω, (NLS) where h > 0, W is the real-valued potential and Ω is a domain in RN. Eq. (NLS) is one of the main objects of the quantum physics, because it appears in problems involving nonlinear optics, plasma physics and condensed matter physics.
Eq. (1.1) has been researched intensively, see [1,3,5,7,10,11,13,14,19,21,22,28] and references therein. In the above works, we observe that many interesting conditions on f have been studied. Notice that, it seems necessary that the condition can be assumed on f at infinity, that is, f is assumed to be subcritical (or critical) at infinity, i.e.,
BCorresponding author. Email: tangcl@swu.edu.cn
(f0) 0≤lim|s|→+∞ f(x,s)s
|s|2∗ < +∞ uniformly forx ∈RN,
where the number 2∗ is denoted by N2N−2 and called the critical Sobolev exponent for the embedding ofH1(RN)intoLp(RN). This aim is to ensure that the associated energy functional would be well defined and of class C1 on H1(RN), and then its critical points are precisely the solutions of Eq. (1.1) by using variational methods. Certainly, many researchers tried to seek some suitable conditions to replace (f0). If there does not exist an assumption on f at infinity, can it be proved that there exists a nontrivial solution for Eq. (1.1)? Mathematically this problem is interesting. Accordingly, Costa and Wang [9] have considered the following equation
−∆u= λf(u), inΩ,
whereλ>0 is a parameter,Ωis a bounded smooth domain inRN(N≥3), and f :R→Ris a function of classC1 satisfying the following conditions:
(f1) f(−u) =−f(u)for any|u| ≤δ (for someδ >0);
(f2) there existsγ∈(2, 2∗)such that lim sup|s|→0 f|(ss|γ)s =0;
(f3) there existsβ∈(2, 2∗)such that lim inf|s|→0 f|(ss|)βs >0;
(f4) there exists µ ∈ (2, 2∗) such that s f(s) ≥ µF(s) > 0 for all |s| small, where F(s) = Rs
0 f(t)dt.
Motivated by Costa and Wang [9], do Ó et al. [12] have studied the following equation
−∆u+V(x)u=λf(u) inRN, (P) whereVsatisfies(V1)–[(V2)or(V3)],
(V1) V∈C(RN,R)and infx∈RNV(x)>0,
(V2) V(x) → ∞ as |x| → ∞, or more generally, for every M > 0, meas{x ∈ RN : V(x) ≤ M}<+∞,
(V3) the function [V(x)]−1 belongs to L1(RN),
and f :R→Ris a function of classC1 satisfying(f10)−(f20)and(f4), (f10) there exists p∈(2, 2∗)such that lim sup|s|→0 f|(ss|)ps <+∞,
(f20) there existsq∈ (2, 2∗)such that lim inf|s|→0F|s(|sq) >0, where F(s) =Rs
0 f(t)dt.
Further results for related problems can be found in [8,15,23,24] and references therein.
Inspired by the above works, we are concerned with the existence of positive solutions for asymptotically periodic Eq. (P) with a locally defined nonlinearity term, namely V satisfies (V4),
(V4) there exists a 1-periodic function V∞(x) ∈ L∞(RN) such that 0 ≤ V(x) ≤ V∞(x), infx∈RNV∞(x)>0 andV(x)−V∞(x)∈ F1, where
F1:={h(x): for any ε>0, meas{x ∈B1(y):|h(x)| ≥ε} →0 as|y| →∞},
and f satisfies(f5)–(f6),
(f5) f ∈ C(R,R)and there exist p > 2,δ ∈ (0, 1)such that the function s 7→ f(s)
sp−1 is nonde- creasing and f(s)>0 on (0,δ],
(f6) there existsq∈(2, 2∗)such that lim infs→0+ Fs(qs) >0, where F(s) =Rs
0 f(t)dt.
As is well known, if f were assumed to be superlinear and subcritical (or critical) at infinity, then the associated energy functional
I(u) = 1 2
Z
RN(|∇u|2+V(x)u2)dx−λ Z
RNF(u)dx
would be of class C1 on H1(RN) and has the mountain pass geometry. Classically, it is a minimax principle that shows the mountain pass level is a critical level of the functional (see [4,5,26]). Here, the assumptions(f5)–(f6)we make on the nonlinearity f(u)refer solely to its behavior in a neighborhood ofu=0, and we will show that they suffice for the existence of a positive solution of Eq. (P) whenλis large enough. Exactly we give our main result.
Theorem 1.1. Assume that N ≥ 3, (V4) and (f5)–(f6) hold. Then there exists λ1 > 0 such that Eq.(P)has a positive solution forλ≥λ1.
Remark 1.2. In this paper, we study the existence of positive solutions for Schrödinger equa- tions with the assumptions of Theorem 1.1 that has never been investigated. For the case where the nonlinear term is only locally defined for |u| small, we should point out that we refer [8,9,12,15] for references in this direction. Costa and Wang [9] considered Eq. (P) in bound domain. do Ó et al. [12] considered Eq. (P) when V was coercive potential or sat- isfied that [V(x)]−1 belongs to L1(RN). Li and Zhong [15] studied the Kirchhoff equation when the nonlinearity term was sub-linear growth. Chu and Liu [8] investigated quasi-linear Schrödinger equations in the radial space. In these papers, they have the compactness and get certain solutions easily. However, in our cases we do not have compact embedding, which is the main difficulty in this paper. Due to this difficult, the methods in [8,9,12,15] fail in our case, so we will use a different way to overcome the lack of compactness.
We now make some comments on the key ingredients of the analysis in this paper. Fol- lowing the idea of [8,9,12,15], we first extend the nonlinear term f and introduce a modified nonlinear Schrödinger equation. Next, we show by variational methods that the modified nonlinear Schrödinger equation possesses a positive ground state solution. Finally, our ap- proach is inspired by the results of [2,6,9,12,26] and is based on the fact that we can show a priori bound of the form
|u|∞< Cλ−β, β>0,
for a class of solutions for the modified nonlinear Schrödinger equation.
The organization of this paper is as follows. In the next section we reserve for setting the framework and establishing some preliminary results. Theorem1.1 is proved in Section 3.
2 Preliminaries
From now on, we will use the following notations.
• H1(RN)is the usual Sobolev space endowed with the usual norm kuk2H =
Z
RN(|∇u|2+u2)dx.
• Lp(RN)is the usual Lebesgue space endowed with the norm
|u|pp =
Z
RN|u|pdx and |u|∞ =ess sup
x∈RN
|u(x)| for all p ∈[1, +∞).
• E:= u∈L2(RN):|∇u| ∈L2(RN)and R
RNV(x)u2dx< +∞ has the norm kuk2=
Z
RN(|∇u|2+V(x)u2)dx.
• measΩdenotes the Lebesgue measure of the setΩ.
• u± :=max{±u, 0}andK:={u∈ E:u+ 6=0}.
• h·,·idenotes action of dual.
• Br(y):= {x∈RN :|x−y| ≤r}andBr :={x∈RN :|x| ≤r}.
• Cdenotes a positive constant and is possibly various in different places.
We work in the spaceE and recall some facts that the normsk · k and k · kH are equivalent andE,→ Ls(RN)for anys ∈ [2, 2∗]is continuous. The proof can be done similarly to that in [19] and details are omitted here. We start by observing that(f5)−(f6)imply that p≤qand
|f(s)s| ≤C|s|p, for any|s| ≤δ.
In order to prove our main result via variational methods, we need to modify and extend f(u) for outside a neighborhood ofu=0 to get fe(u). We set
ef(s):=
0, s ≤0,
f(s), 0<s≤ δ, C1sp−1, δ <s,
and fixC1 > 0 such that ef ∈ C(R,R+). Combining with the definition of fe, one can easily obtain the following lemma.
Lemma 2.1. Suppose that(f5)hold. Then (a) lims→+∞ Fe(s)
s2 = +∞, where Fe(s) =Rs
0 ef(t)dt,
(b) there exists C>0such that|ef(s)s| ≤C|s|pand|Fe(s)| ≤C|s|p for all s∈R, (c) there existsµ∈ (2,p)such that the function s7→ ef(s)
sµ−1 is strictly increasing on(0,+∞),
Now let us consider the modified equation of Eq. (P) given by (−∆u+V(x)u=λef(u),
u∈E. (Pe)
The corresponding energy functional Ie(u) = 1
2 Z
RN(|∇u|2+V(x)u2)dx−λ Z
RNFe(u)dx is of classC1 by a standard argument and whose derivative is given by
hIe0(u),vi=
Z
RN(∇u· ∇v+V(x)uv)dx−λ Z
RN ef(u)vdx, v∈ E.
Formally, critical points of Ie are solutions of Eq. (Pe). We note that critical points of Ie with L∞-norm less than or equal toδare also solutions of the original Eq. (P). We recall the Nehari manifold
N :=nu∈E\ {0}:hIe0(u),ui=0o
=nu∈ K:hIe0(u),ui=0o , and set
c:= inf
u∈NIe(u). Lemma 2.2. Suppose that(V4)and(f5)hold. Then
(a) for any u∈K , there exists a unique tu >0such that tuu∈ N. Moreover, the maximum ofIe(tu) for t>0is achieved at tu,
(b) there existsρ>0such thatkuk ≥ρfor all u∈ N,
(c) the functionalI is bounded from below onN by a positive constant.
Proof. (a)For anyu ∈K, we define Ψ(t):=Ie(tu) = t
2
2 Z
RN(|∇u|2+V(x)u2)dx−λ Z
RN Fe(tu)dx, t∈ (0,+∞). It follows from (b) of Lemma2.1and the Sobolev inequality that
Z
RNFe(tu)dx≤ C Z
RN|tu|pdx ≤Ctpkukp. Thus one has
Ψ(t)≥ t
2
2kuk2−λCtpkukp.
Then there exists t0 >0 such thatΨ(t0)>0. We setΩ={x∈RN :u(x)>0}. Combining (a) in Lemma2.1with Fatou’s lemma, we have
lim inf
t→∞ Z
Ω
Fe(tu) (tu)2u
2dx= +∞.
Hence lim sup
t→∞
Ψ(t) t2 = 1
2kuk2−λlim inf
t→∞ Z
Ω
Fe(tu)
t2 dx= 1
2kuk2−λlim inf
t→∞ Z
Ω
Fe(tu) (tu)2u
2dx=−∞.
One could deduce Ψ(t) → −∞ as t → +∞. So there exists tu > 0 such that Ψ(tu) = maxt>0Ψ(t) and Ψ0(tu) = 0, i.e., Ie(tuu) = maxt>0Ie(tu) and tuu ∈ N. Suppose that there existst1 >t2>0 such thattiu∈ N,i=1, 2, one has
Z
Ω
ef(t1u)u2 t1u dx=
Z
Ω
ef(t2u)u2 t2u dx,
which contradicts (c) of Lemma2.1. Thus we can conclude thattuis unique.
(b)For anyu∈ N, combining the Sobolev embedding and (b) of Lemma2.1, one obtains kuk2= λ
Z
RN ef(u)udx≤ Cλ Z
RN|u|pdx≤Cλkukp. (2.1) It follows from (2.1) that there existsρ>0 independent ofusuch that
ρ≤ kuk.
(c)Also from (b) of Lemma2.1 and the Sobolev inequality, we have Ie(u)≥ 1
2kuk2−Cλkukp.
Sincep>2, there existsσ >0 such thatIe(u)≥ σ42 >0 forkuk=σ >0. For anyv∈ N, there existst0 >0 such thatt0kvk=σ. Combining with (a)-(b) of Lemma2.2, one obtains
Ie(v)≥Ie(t0v)≥ σ
2
4 . This completes the proof.
From Lemmas2.1–2.2, one can easily know (see also [19,26]) c= inf
u∈NI(e u) = inf
u∈Ksup
t>0
Ie(tu) =min
γ∈Γ max
t∈[0,1]
Ie(γ(t)),
whereΓ={γ∈C([0, 1],E):γ(0) =0,Ie(γ(t))<0}. Notice that,c>0 from (c) of Lemma2.2.
In order to prove our results, we introduce the following equation
−∆u+V∞(x)u=λfe(u), (P∞) and it follows from [16,19,26,27] that Eq. (P∞) has a positive ground state solutionω. From Lemma2.2 and(V4), there exists a uniquetω >0 such thattωω∈ N and
c≤Ie(tωω)≤Ie∞(tωω)≤Ie∞(ω):=c∞, (2.2) whereIe∞ is the energy functional associated with Eq. (P∞).
Lemma 2.3. Suppose that(V4)and(f5)hold. If u∈ N andIe(u) =c, then u is a nontrivial solution of Eq.(Pe).
Proof. Inspired by the method in [18], one supposes by contradiction thatuis not a nontrivial solution of Eq. (Pe). Then there existsφ∈ Esuch that
hIe0(u),φi=
Z
RN(∇u· ∇φ+V(x)uφ)dx−λ Z
RN ef(u)φdx<−1.
Letε ∈(0, 1)be small enough. Then hIe0(tu+sφ),φi ≤ −1
2, for any|t−1| ≤ε, |s| ≤ε. (2.3) We set a curve
γ(t) =tu+sτ(t)φ, t >0,
where τ∈C(R,[0, 1])is a smooth cut-off function such thatτ(t) =1 for|t−1| ≤ 2ε,τ(t) =0 for|t−1| ≥ε. Obviously,γis a continuous. We can claim thatIe(γ(t))<cfor anyt∈ (0,+∞). Indeed, it follows from Lemma2.2that Ie(γ(t)) =Ie(tu)<Ie(u) =cfor|t−1| ≥ε. When
|t−1|< ε, owing toΦ(s):= Ie(tu+sτ(t)φ)is ofC1on [0,ε], there exists ¯s∈ (0,ε)such that Ie(tu+sτ(t)φ) =Ie(tu) +hIe0(tu+sτ¯ (t)φ),ετ(t)φi ≤Ie(tu)− 1
2ετ(t)<c, where the inequality holds from (2.3). HenceIe(γ(t))<cfor anyt∈ (0,+∞).
We denote J(u) = hIe0(u),ui. According to Lemma2.2 and the definition of γ, we have J(γ(1−ε)) = J((1−ε)u) > 0 and J(γ(1+ε)) = J((1+ε)u) < 0. By the continuity of t 7→ J(γ(t)) there exists t0 ∈ (1−ε, 1+ε) such that J(γ(t0)) = 0. Thus γ(t0) ∈ N and I(e γ(t0))<c, which is a contradiction. This completes the proof.
Lemma 2.4. Suppose that(V4)and(f5)hold. Then the Cerami sequence forIeat level m>0(shortly:
(Ce)msequence) is bounded in E.
Proof. We recall the(Ce)m sequence{un}, that is,
Ie(un)→m, kIe0(un)k(1+kunk)→0.
Then
o(1) =hIe0(un),u−ni=−ku−nk2.
Consequently we could deduce that {u+n} is also a (Ce)m sequence. For the sake of con- venience, we denote u+n by un. By a contradiction, we assume that kunk → +∞ and set vn = kuun
nk. Obviously up to a subsequence, there exists a nonnegative function v ∈ E such that vn → v ∈ E,vn → v ∈ L2loc(RN) and vn(x) → v(x) a.e. in RN. We denote Ω1 ={x ∈RN :v(x)>0}. If measΩ1>0, Fatou’s lemma and (a) of Lemma2.1imply
lim inf
n→∞ Z
RN
Fe(un)
u2n v2ndx≥lim inf
n→∞ Z
Ω1
Fe(un)
u2n v2ndx= +∞.
Then
0=lim sup
n→∞
Ie(un) kunk2 = 1
2−λlim inf
n→∞ Z
RN
Fe(un)
u2n v2ndx=−∞, which is a contradiction. Thusv=0. We denote
α:= lim
n→∞ sup
z∈RN Z
B1(z)v2ndx. (2.4)
If α = 0, we have vn → 0 in Lp(RN) from the Lions lemma [17,26]. Combining with (b) of Lemma2.1, we obtainR
RNFe(2√
mvn)dx= o(1). By the continuity ofIe, there existstn ∈[0, 1]
such thatIe(tnun) =maxt∈[0,1]Ie(tun). Sincekunk →+∞, one has 2
√m
kunk ≤1 asnlarge enough.
We observe that
Ie(tnun) +o(1)≥Ie 2√
m kunkun
+o(1) =2mkvnk2−λ Z
RNFe(2√
mvn)dx+o(1)
=2m+o(1).
In view ofIe(un)→mand (a) of Lemma2.2, we can see thattn∈(0, 1)andhIe0(tnun),tnuni= 0 asnlarge enough. Hence by Lemma 2.3 in [20], one has
m=Ie(un) +o(1)
=Ie(un)− 1
µhIe0(un),uni+o(1)
= 1
2 − 1 µ
Z
RN |∇un|2+V(x)u2n
dx+λ Z
RN
1 µ
ef(un)un−Fe(un)
dx+o(1)
≥ µ−2 2µ t2n
Z
RN |∇un|2+V(x)u2n
dx+λ Z
RN
1 µ
fe(tnun)tnun−Fe(tnun)dx+o(1)
=Ie(tnun)− 1
µhIe0(tnun),tnuni+o(1)
=Ie(tnun) +o(1)
≥2m+o(1), which is a contradiction.
Ifα>0, there exists{zn} ⊂RN such that α 2 ≤
Z
B1(zn)v2ndx.
If{zn}is bounded, there exists R>0 such that α 2 ≤
Z
BR
v2ndx.
which is a contradiction with vn → 0 in L2loc(RN). Then {zn} is unbounded, up to a subse- quence,|zn| →∞. We setwn(x):=vn(x+zn), wherewnsatisfies
α 2 ≤
Z
B1
w2ndx,
up to a subsequence, there exists w ∈ E such that wn * w in E, wn → w in L2loc(RN) and wn(x)→ w(x)a.e. inRN. Evidently, measΩ2 > 0 whereΩ2 = {x ∈ RN : w(x) >0}. In fact wn(x) = un(kx+zn)
unk . Also from Fatou’s lemma and (a) of Lemma2.1, one obtains lim inf
n→∞
1 kunk2
Z
RNFe(un)dx
=lim inf
n→∞
1 kunk2
Z
RNFe(un(x+zn))dx
≥lim inf Z
Ω2
Fe(un(x+zn)) [un(x+zn)]2 w
2ndx
= +∞.
Hence
0=lim sup
n→∞
Ie(un) kunk = 1
2 −λlim inf
n→∞
1 kunk2
Z
RNFe(un)dx
=−∞, which is a contradiction. In a word, the(Ce)m sequence {un}is bounded inE.
Proposition 2.5. Suppose that(V4)and(f5)hold. Then Eq.(Pe)has a positive ground state solution.
Proof. Notice that 0<c≤c∞. Therefore, one of the two cases occurs:
Case 1. c=c∞. It follows from (2.2) that
c∞ ≤Ie(tωω)≤ Ie∞(tωω)≤Ie∞(ω) =c∞.
Thenω is also a positive ground state solution of Eq. (Pe) from Lemma2.3.
Case 2. 0 < c < c∞. We see easily Ie satisfies the mountain pass geometry. From the mountain pass theorem [25,26] and Lemma 2.4, there exists a nonnegative and bounded sequence {un} ∈Esuch that
Ie(un)→c, kIe0(un)k(1+kunk)→0.
Then there exists a nonnegative function u ∈ E such that up to a subsequence, un * u in E, un → u in L2loc(RN) and un(x) → u(x) a.e. in RN. For any ϕ ∈ C0∞(RN), one has 0 = hIe0(un),ϕi+o(1) = hIe0(u),ϕi, i.e., uis a nonnegative solution of Eq. (Pe). Ifu 6= 0 inE, combining Lemma 2.3 in [20] with Fatou’s lemma one obtains
c=Ie(un) +o(1)
=Ie(un)− 1
µhIe0(un),uni+o(1)
= 1
2 − 1 µ
Z
RN |∇un|2+V(x)u2ndx+λ Z
RN
1 µ
ef(un)un−Fe(un)
dx+o(1)
≥ 1
2 − 1 µ
Z
RN |∇u|2+V(x)u2
dx+λ Z
RN
1 µ
fe(u)u−Fe(u)
dx+o(1)
=Ie(u)− 1
µhIe0(u),ui+o(1)
=Ie(u) +o(1). (2.5)
At the same time, one knows c ≤ Ie(u) from the definition of c and u ∈ N. Applying the strongly maximum principle, we could deduce that u is a positive ground state solution of Eq. (Pe).
We assume that u = 0 (otherwise we complete the proof). Then there exists α ≥ 0 such that
nlim→∞ sup
z∈RN Z
B1(z)
|un|2dx =α.
Indeed, ifα=0, applying the Lions lemma [17,26] we obtain
un →0 in Lp(RN). (2.6)
HenceIe(un)→0 asn→∞from (b) in Lemma2.1, which contradictsc>0. Then there exists {zn} ⊂RN such thatR
B1(zn)|un|2dx≥ α2 >0.
If{zn}is bounded, there existsR >0 such that R
BR(0)|un|2dx ≥ α2 >0, which is a contra- diction with un →0 in L2loc(RN). Then{zn}is unbounded. After extracting a subsequence if
necessary, we have (i) |zn| →+∞,
(ii) un(·+zn)*v6=0 inE.
From Lemma 2.4 in [19], we have 0=hIe0(un),ϕ(· −zn)i+o(1)
=
Z
RN[∇un· ∇ϕ(· −zn) +V(x)unϕ(x−zn)]dx−λ Z
RN fe(un)ϕ(x−zn)dx+o(1)
=
Z
RN[∇un· ∇ϕ(· −zn) +V∞(x)unϕ(x−zn)]dx−λ Z
RN ef(un)ϕ(x−zn)dx+o(1)
=hIe∞0 (v),ϕi+o(1).
Thenv is a nontrivial solution of Eq. (P∞). Notice that, also from [19], we obtain c=Ie(un)− 1
µhIe0(un),uni+o(1)
= 1
2− 1 µ
Z
RN |∇un|2+V(x)u2n
dx+λ Z
RN
1 µ
ef(un)un−Fe(un)
dx+o(1)
= 1
2− 1 µ
Z
RN |∇un|2+V∞(x)u2n dx +λ
Z
RN
1 µ
fe(un(·+zn))un(·+zn)−Fe(un(·+zn))
dx+o(1)
=Ie∞(un(·+zn))− 1
µhIe∞0 (un(·+zn)),un(·+zn)i+o(1)
=Ie∞(v) +o(1)
≥c∞+o(1), which is a contradiction.
In conclusion, whether Case 1 occurs or Case 2 occurs, we can prove Proposition2.5.
3 Proof of Theorem 1.1
Lemma 3.1. Suppose that (V4) and (f5) hold. If u is a critical point of Ie, then u ∈ L∞(RN). Furthermore, there exists a positive constant C independent ofλsuch that
|u|∞ ≤Cλ2∗ −1p Z
RN|∇u|2dx 2(22∗ −∗ −2p)
.
Proof. We prove the result by using the Moser iteration. For eachk>0, we define uk(x) =
(u(x), if|u(x)| ≤k,
±k, if ±u(x)>k.
Forβ>1, we use ϕk = |uk|2(β−1)uas a test function inhIe0(u),ϕkito obtain Z
RN|uk|2(β−1)|∇u|2dx+2(β−1)
Z
RN|uk|2(β−2)uuk∇u· ∇ukdx +
Z
RNV(x)|uk|2(β−1)u2dx=λ Z
RN ef(u)|uk|2(β−1)udx. (3.1)
Then we use the Sobolev inequality to yield β2
Z
RN
|uk|2(β−1)|∇u|2dx+2(β−1)|uk|2(β−2)uuk∇u· ∇uk dx
≥
Z
RN|uk|2(β−1)|∇u|2+ (β−1)2|uk|2(β−2)u2|∇uk|2+2(β−1)|uk|2(β−2)uuk∇u· ∇ukdx
≥
Z
RN
∇|uk|β−1u
2dx
≥C Z
RN
|uk|β−1u
2∗
dx 22∗
, (3.2)
where we also have used the facts thatu2|∇uk|2 ≤u2k|∇u|2andβ>1. From (b) in Lemma2.1, we deduce
Z
RN ef(u)|uk|2(β−1)udx≤C Z
RN|u|p|uk|2(β−1)dx. (3.3) Combining (3.1), (3.2) and (3.3), we obtain
Z
RN
|uk|β−1u
2∗
dx 22∗
≤Cβ2λ Z
RN|u|p−2|uk|2(β−1)u2dx
≤Cβ2λ Z
RN|u|2∗dx
p2−∗2 Z
RN
|uk|2(β−1)u2
2∗ 2∗ −p+2
dx
!2∗ −2∗p+2 . Lettingk→∞, we have
|u|β·2∗ ≤ Cβ2λ2β1 Z
RN|∇u|2dx p4β−2
|u| 2·2∗β 2∗ −p+2
. (3.4)
To carry out an iteration process, we set βm =
2∗−p+2 2
m+1
, m=0, 1, . . . Then we have
2·2∗βm
2∗−p+2 =2∗βm−1. By (3.4), one obtains
|u|βm·2∗ ≤ Cβ2mλ2βm1 Z
RN|∇u|2dx 4βmp−2
|u|2·2∗βm 2∗ −p+2
= (Cλ)2βm1 β
1
mβm
Z
RN|∇u|2dx 4βmp−2
|u|βm−1·2∗. By the Moser iteration, we have
|u|βm·2∗ ≤ (Cλ)∑mi=02β1i
∏
m i=0β
1 βi
i
Z
RN|∇u|2dx
p−42∑mi=0 1
βi
|u|2∗. (3.5)
Sinceβ0 =2∗−2p+2>1 andβi =βi0+1, we observe that
∑
m i=01 βi
=
∑
m i=01 βi0+1
,
∏
m i=0β
1 βi
i =
∏
m i=0βi0+1
1
βi+1
0 = (β0)∑
m i=0 i+1
βi+1
0 .
One can easily see
∑
∞ i=0i+1 βi0+1
= β∗ <+∞,
∑
∞ i=01 βi0+1
= 2
2∗−p. Lettingm→∞in (3.5), we conclude thatu ∈L∞(RN)and
|u|∞ ≤Cλ2∗ −1pββ
∗ 0
Z
RN|∇u|2dx 2(2p∗ −−2p)
|u|2∗ ≤Cλ2∗ −1p Z
RN|∇u|2dx 2(22∗ −∗ −2p)
. (3.6) This completes the proof.
Proof of Theorem1.1. By proposition2.5, Eq. (Pe) has a positive ground solutionu. Combining the Sobolev embedding and (b) of Lemma2.1, one obtains
c=Ie(u)− 1
µhIe0(u),ui ≥ 1
2 − 1 µ
kuk2. (3.7)
We can see that there existsv ∈K∩L∞(RN)such that|v|∞<1. Since (f6), there existsC>0 independent ofλsuch that
Fe(tv)≥C|tv|q, t∈[0, 1].
At the same time there existsλ0 >0 such that Ie(v)<0 for λ≥λ0. Then from the definition ofc, we have
c≤ max
t∈[0,1]
Ie(tv)
= max
t∈[0,1]
t2 2
Z
RN(|∇v|2+V(x)v2)dx−λ Z
RNFe(tv)dx
≤ max
t∈[0,1]
t2 2
Z
RN(|∇v|2+V(x)v2)dx−Ctqλ Z
RN|v|qdx
≤Cλ−q−22. (3.8)
Combining (3.6), (3.7) and (3.8), we have
|u|∞ ≤Cλ2∗ −1pkuk22∗ −∗ −2p ≤Cλ2∗ −1pλ
1 2−q·22∗ −∗ −2p
. Sincep,q∈ (2, 2∗), there existsλ1≥ λ0such that
|u|∞ ≤Cλ
2∗ −q (2∗ −p)(2−q)
1 ≤ δ.
Therefore, from the definition of ef, we can conclude thatuis also a positive solution of Eq. (P) forλ≥λ1. This completes the proof of Theorem1.1.
Acknowledgment
This work was partially supported by Postgraduate research and innovation project of Chongqing (No. CYB19082), Fundamental Research Funds for the Central Universities (No.
XDJK2020D032) and National Natural Science Foundation of China (No.11971393). The au- thors express their gratitude to the handling editor and anonymous reviewers for careful reading and helpful suggestions for the improvement of the original manuscript.
References
[1] C. O. Alves, M. A. S. Souto, M. Montenegro, Existence of a ground state solution for a nonlinear scalar field equation with critical growth,Calc. Var. Partial Differential Equations 43(2012), No. 3–4, 537–554.https://doi.org/10.1007/s00526-011-0422-y;MR2875652;
Zbl 1237.35037.
[2] C. O. Alves, M. A. Souto, Existence of solutions for a class of elliptic equations in RN with vanishing potentials, J. Differential Equations 252(2012), No. 10, 5555–5568. https:
//doi.org/10.1016/j.jde.2012.01.025;MR2902126;Zbl 1250.35103.
[3] T. Bartsch, Z. Q. Wang, Existence and multiplicity results for some superlinear ellip- tic problems on RN, Comm. Partial Differential Equations 20(1995), No. 9–10, 1725–1741.
https://doi.org/10.1080/03605309508821149;MR1349229;Zbl 0837.35043.
[4] T. Bartsch, Z.-Q. Wang, M. Willem, The Dirichlet problem for superlinear ellip- tic equations, in: Stationary partial differential equations. Vol. II, Handb. Differ. Equ., Elsevier/North-Holland, Amsterdam, 2005, pp. 1–55.https://doi.org/10.1016/S1874- 5733(05)80009-9;MR2181481;Zbl 1071.35002.
[5] H. Berestycki, P.-L. Lions, Nonlinear scalar field equations. I. Existence of a ground state, Arch. Rational Mech. Anal. 82(1983), No. 4, 313–345. https://doi.org/10.1007/
BF00250556;MR695536;Zbl 0533.35029.
[6] H. Brézis, T. Kato, Remarks on the Schrödinger operator with singular complex poten- tials,J. Math. Pures Appl. (9)58(1979), No. 2, 137–151.MR539217;Zbl 0408.35025.
[7] H. Brézis, L. Nirenberg, Positive solutions of nonlinear elliptic equations involving critical Sobolev exponents, Comm. Pure Appl. Math. 36(1983), No. 4, 437–477. https:
//doi.org/10.1002/cpa.3160360405;MR709644;Zbl 0541.35029.
[8] C. Chu, H. Liu, Existence of positive solutions for a quasilinear Schrödinger equation, Nonlinear Anal. Real World Appl.44(2018), 118–127.https://doi.org/10.1016/j.nonrwa.
2018.04.007;MR3825156;Zbl 1406.35347.
[9] D. G. Costa, Z.-Q. Wang, Multiplicity results for a class of superlinear elliptic problems, Proc. Amer. Math. Soc.133(2005), No. 3, 787–794.https://doi.org/10.1090/S0002-9939- 04-07635-X;MR2113928;Zbl 1210.35048.
[10] V. Coti Zelati, P. H. Rabinowitz, Homoclinic type solutions for a semilinear elliptic PDE on RN, Comm. Pure Appl. Math.45(1992), No. 10, 1217–1269. https://doi.org/10.
1002/cpa.3160451002;MR1181725;Zbl 0785.35029.
[11] R. de Marchi, Schrödinger equations with asymptotically periodic terms, Proc.
Roy. Soc. Edinburgh Sect. A 145(2015), No. 4, 745–757. https://doi.org/10.1017/
S0308210515000104;MR3371574;Zbl 1331.35136.
[12] J. M.doÓ, E. Medeiros, U. Severo, On the existence of signed and sign-changing solu- tions for a class of superlinear Schrödinger equations,J. Math. Anal. Appl.342(2008), No. 1, 432–445.https://doi.org/10.1016/j.jmaa.2007.11.058;MR2440810;Zbl 1139.35094.
[13] Q. Guo, J. Mederski, Ground states of nonlinear Schrödinger equations with sum of periodic and inverse-square potentials, J. Differential Equations 260(2016), No. 5, 4180–
4202.https://doi.org/10.1016/j.jde.2015.11.006;MR3437584;Zbl 1335.35232.
[14] W. Kryszewski, A. Szulkin, Generalized linking theorem with an application to a semilinear Schrödinger equation, Adv. Differential Equations 3(1998), No. 3, 441–472.
MR1751952;Zbl 0947.35061.
[15] L. Li, X. Zhong, Infinitely many small solutions for the Kirchhoff equation with local sublinear nonlinearities,J. Math. Anal. Appl.435(2016), No. 1, 955–967.https://doi.org/
10.1016/j.jmaa.2015.10.075;MR3423438;Zbl 1328.35026.
[16] Y. Li, Z.-Q. Wang, J. Zeng, Ground states of nonlinear Schrödinger equations with potentials, Ann. Inst. H. Poincaré Anal. Non Linéaire 23(2006), No. 6, 829–837. https:
//doi.org/10.1016/j.anihpc.2006.01.003;MR2271695;Zbl 1111.35079.
[17] P.-L. Lions, The concentration-compactness principle in the calculus of variations. The locally compact case. I, Ann. Inst. H. Poincaré Anal. Non Linéaire1(1984), No. 2, 109–145.
http://www.numdam.org/item?id=AIHPC_1984__1_2_109_0;MR778970.Zbl 0541.49009.
[18] J.-Q. Liu, Y.-Q. Wang, Z.-Q. Wang, Solutions for quasilinear Schrödinger equations via the Nehari method,Comm. Partial Differential Equations29(2004), No. 5–6, 879–901.https:
//doi.org/10.1081/PDE-120037335;MR2059151;Zbl 1140.35399.
[19] J. Liu, J.-F. Liao, C.-L. Tang, A positive ground state solution for a class of asymptotically periodic Schrödinger equations, Comput. Math. Appl. 71(2016), No. 7, 965–976. https:
//doi.org/10.1016/j.camwa.2016.08.010;MR3547688;Zbl 1362.35114.
[20] S. Liu, On superlinear problems without the Ambrosetti and Rabinowitz condition, Nonlinear Anal.73(2010), No. 3, 788–795. https://doi.org/10.1016/j.na.2010.04.016;
MR2653749;Zbl 1192.35074.
[21] J. Mederski, Ground states of a system of nonlinear Schrödinger equations with periodic potentials, Comm. Partial Differential Equations41(2016), No. 9, 1426–1440.https://doi.
org/10.1080/03605302.2016.1209520;MR3551463;Zbl 1353.35267.
[22] S. Peng, C. Wang, S. Yan, Construction of solutions via local Pohozaev identities, J. Funct. Anal. 274(2018), No. 9, 2606–2633. https://doi.org/10.1016/j.jfa.2017.12.
008;MR3771837;Zbl 1392.35148.
[23] P. Pucci, J. Serrin, A note on the strong maximum principle for elliptic differential inequalities, J. Math. Pures Appl. (9) 79(2000), No. 1, 57–71. https://doi.org/10.1016/
S0021-7824(99)00146-4;MR1742565;Zbl 0952.35163.
[24] P. Pucci, J. Serrin, The strong maximum principle revisited, J. Differential Equa- tions196(2004), No. 1, 1–66.https://doi.org/10.1016/j.jde.2003.05.001;MR2025185;
Zbl 1109.35022.
[25] M. Schechter, A variation of the mountain pass lemma and applications, J. London Math. Soc. (2) 44(1991), No. 3, 491–502. https://doi.org/10.1112/jlms/s2-44.3.491;
MR1149010;Zbl 0756.35032.
[26] M. Willem, Minimax theorems, Progress in Nonlinear Differential Equations and their Applications, Vol. 24, Birkhäuser Boston, Inc., Boston, MA, 1996. https://doi.org/10.
1007/978-1-4612-4146-1;MR1400007;Zbl 0856.49001.
[27] M. Yang, Ground state solutions for a periodic Schrödinger equation with superlinear nonlinearities,Nonlinear Anal. 72(2010), No. 5, 2620–2627. https://doi.org/10.1016/j.
na.2009.11.009;MR2577823;Zbl 1184.35139.
[28] J. Zhang, W. Zou, Solutions concentrating around the saddle points of the potential for critical Schrödinger equations, Calc. Var. Partial Differential Equations54(2015), No. 4, 4119–4142.https://doi.org/10.1007/s00526-015-0933-z;MR3426106;Zbl 1339.35109.