Research ArticlePositive Solutions for a Class of Discrete Mixed Boundary ValueProblems with the (p q)-Laplacian Operator
Cuiping Li1 and Zhan Zhou 12
1School of Mathematics and Information Science Guangzhou University Guangzhou 510006 China2Center for Applied Mathematics Guangzhou University Guangzhou 510006 China
Correspondence should be addressed to Zhan Zhou zzhou0321hotmailcom
Received 31 December 2019 Accepted 15 February 2020 Published 26 March 2020
Academic Editor Beatrice Di Bella
Copyright copy 2020 Cuiping Li and Zhan Zhou)is is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited
In this paper we consider the existence of solutions for the discrete mixed boundary value problems involving (p q)-Laplacianoperator By using critical points theory we obtain the existence of at least two positive solutions for the boundary value problemunder appropriate assumptions on the nonlinearity
1 Introduction
In recent years with the development of mechanical engi-neering control system computer science and economicsthe existence of solutions of difference equations hasattracted wide attention (see [1ndash6]) For example applyingRicceri variational principle to obtain the existence ofmultiple solutions [7ndash9] taking the invariant sets ofdescending flow to prove the existence of sign-changingsolutions [10] making the linking theorem to get the ex-istence and multiplicity of periodic solutions [11] and usingcritical point theory to obtain the existence of homoclinicsolutions [12ndash15] and heteroclinic solutions [16]
As we know the fixed-point method and upper andlower solution techniques are important tools to solve theexistence of solutions for boundary value problems (see
[17 18]) But recently it is more common to use criticalpoint theory to study Dirichlet boundary value problems(see [19ndash23]) More result on difference equations by usingcritical point theories can be referred to [24ndash27]
In [28] DrsquoAguı et al established the existence of at leasttwo positive solutions for the following discrete Dirichletboundary value problem
minusΔ ϕp(Δu(k minus 1))1113872 1113873 + q(k)ϕp(u(k)) λf(k u(k)) k isin Z(1 N)
u(0) u(N + 1) 0
⎧⎨
⎩
(1)
where q(k)ge 0 for all k isin 1 2 N Unlike this DrsquoAguı et al in [29] proved that there are at
least two nonzero weak solutions for the following mixedboundary value problem
minus q(x) uprime(k)1113868111386811138681113868
1113868111386811138681113868pminus2
uprime(x)1113872 1113873prime + s(x)|u(k)|pminus2u(x) λf(x u(x)) x isin [a b]
u(a) uprime(b) 0
⎧⎨
⎩ (2)
where pgt 1 q s isin Linfin([a b]) with q0 ess inf [ab]qgt 0 s0
ess inf [ab]sge 0 f [a b] times R⟶ R is an L1-Caratheodoryfunction and λ is a real positive parameter
As a discrete analogy of the abovementioned problemwe consider the existence of positive solutions for the fol-lowing discrete mixed boundary value problem
HindawiDiscrete Dynamics in Nature and SocietyVolume 2020 Article ID 5414783 9 pageshttpsdoiorg10115520205414783
minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) λf(k u(k)) k isin Z(1 N)
u(0) Δu(N) 0
⎧⎨
⎩ Df
λ1113872 1113873 (3)
where Z(a b) denote the discrete interval a a + 1 b
for any integers a and b with alt b N be a positive integerf(k u) is continuous in u for each k isin Z(1 N) Δu(k)
u(k + 1) minus u(k) is the forward difference operatorϕr R⟶ R is the r-Laplacian given by ϕr(u) |u|rminus 2u
with u isin R 1lt qleplt +infin s(k)ge 0 for all k isin Z(1 N) andλ is a positive parameter
In this paper under suitable assumptions on the non-linearity f we use the theory of two nonzero critical points(see [30]) to ensure that there are at least two nonzerosolutions for problem (D
f
λ ) )e two nonzero critical pointstheorem is an appropriate combination of local minimumtheorem (see [31]) and classical AmbrosettindashRabinowitztheorem (see [32]) An important hypothesis of mountainpass theorem is PalaisndashSmale condition It satisfies the ap-plication of infinite dimensional space by requiring thecondition that the nonlinear term is stronger than p-superlinearity at infinity In order to obtain the existence oftwo nonzero solutions we can assume the classicalAmbrosettindashRabinowitz condition and nonlinear algebraiccondition (see (40) in )eorem 2) hold that is morewidespread than the p-sublinearity at zero Moreover whenwe require that f(k 0)ge 0 for all k isin Z(1 N) we can usestrongmaximum principle to obtain the existence of positivesolutions which has been proved in Lemma 2
Let slowast min s(k) k isin Z(1 N) a special case of ourmain result is stated as follows
Theorem 1 Let f R⟶ R be a continuous function suchthat
limt⟶0+
f(t)
tpminus14(a)
limt⟶+infin
f(t)
tpminus14(b)
then for each λ isin (0 (1 + slowastNpminus1pNp)min 1113864supcgt0(cq
max|ξ|lec 1113938ξ0 f(t)dt) supcgt0(cpmax|ξ|lec 1113938
ξ0 f(t)dt)1113865) the
problemminusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) λf(u(k)) k isin Z(1 N)
u(0) Δu(N) 0
⎧⎨
⎩
(5)
admits at least two positive solutions
)e structure of the article is as follows In Section 2some basic definitions and properties are given In Section 3we give the main results Under suitable hypothesis Lemma1 is used to obtain that the problem (D
f
λ ) possesses at leasttwo positive solutions Finally some examples are given toillustrate our main results
2 Preliminaries
In this section we recall some definitions notations andproperties Consider the N-dimensional Banach space
S u Z(0 N + 1)⟶ R u(0) Δu(N) 0 (6)
and define the norm
u 1113944N+1
k1|Δu(k minus 1)|
p⎛⎝ ⎞⎠
1p
(7)
and uinfin max |u(k)| k isin Z(1 N) is another norm in S
Proposition 1 -e following inequality holds
uinfin lemaxpNpminus1
1 + slowastNpminus11113888 1113889
1qup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1q
pNpminus1
1 + slowastNpminus11113888 1113889
1pup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1p⎧⎨
⎩
⎫⎬
⎭ (8)
Proof Let u isin S then there exist klowast isin Z(1 N) such that|u(klowast)| max |u(k)| k isin Z(1 N)
Since
u klowast
( 11138571113868111386811138681113868
1113868111386811138681113868 1113944
klowast
k1Δu(k minus 1)|le 1113944
klowast
k11)
1minus1p1113944
klowast
k1|Δu(k minus 1)|
p)1p leN
1minus1pu⎛⎝⎛⎝
1113868111386811138681113868111386811138681113868111386811138681113868(9)
2 Discrete Dynamics in Nature and Society
then
upinfin leN
pminus1u
p (10)
If ||u||infin gt 1 then
1 + slowastNpminus1( 1113857uq
infinp
leupinfin + Npminus1 1113936
Nk1 s(k)|u(k)|q
p
leNpminus 1
pu
p+ 1113944
N
k1s(k)|u(k)|
q⎛⎝ ⎞⎠leNpminus1
pu
p+
Npminus1
q1113944
N
k1s(k)|u(k)|
q
(11)
that is
uinfin lepNpminus 1
1 + slowastNpminus11113888 1113889
1qup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1q
(12)
If uinfin le 1 then
1 + slowastNpminus 1( 1113857u
pinfin
ple
upinfin + Npminus 1 1113936
Nk1 s(k)|u(k)|q
p
leNpminus 1
pu
p+
Npminus 1
q1113944
N
k1s(k)|u(k)|
q
(13)
that is
uinfin lepNpminus 1
1 + slowastNpminus11113888 1113889
1p||u||p
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1p
(14)
In summary we have
uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1qup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1pup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1p⎧⎨
⎩
⎫⎬
⎭ (15)
Put
F(k t) ≔ 1113946t
0f(k ξ)dξ forall(k t) isin Z(1 N) times R (16)
and consider the function Jλ S⟶ R for all λgt 0 by
Jλ Φ minus λΨ (17)
where
Φ ≔ Φ1 +Φ2Φ1(u) up
p
Φ2(u) 1113936
Nk1 s(k)|u(k)|q
q
Ψ(u) ≔ 1113944N
k1F(k u(k))
(18)
It is clear that Φ1Φ2Ψ isin C1(SR) and their Gateauxderivatives at the point u isin S are given by
Φ1prime(u)(v) 1113944N+1
k1ϕp(Δu(k minus 1))v(k)
minus 1113944N+1
k1ϕp(Δu(k minus 1))v(k minus 1)
1113944N
k1ϕp(Δu(k minus 1))v(k) minus 1113944
N
k0ϕp(Δu(k))v(k)
minus 1113944N
k1Δϕp(Δu(k minus 1))v(k)
Φ2prime(u)(v) 1113944N
k1s(k)ϕq(u(k))v(k)
Ψprime(u)(v) 1113944N
k1f(k u(k))v(k)
(19)
Discrete Dynamics in Nature and Society 3
for all u v isin S So we have
Jλprime(u)(v) 1113944N
k1minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) minus λf(k u(k))1113960 1113961v(k) (20)
Hence a critical point u of Jλ is a solution of problem(D
f
λ )Now we recall a definition and a two nonzero critical
points theorem for the readerrsquos convenience
Definition 1 Let X be a real Banach space we say that aGateaux differentiable function Jλ X⟶ R satisfies the(PS)-condition if any sequence un1113864 1113865nisinNsubeX such that
(i) Jλ(un)⟶ c isin R as n⟶ +infin(ii) Jλprime(un)⟶ 0 as n⟶ +infin has a convergent
subsequence
Lemma 1 Let X be a real Banach space and ΦΨ isin C1(SR)
such that infX(Φ) Φ(0) Ψ(0) 0 Assume that there arer isin R and ω isin X with 0ltΦ(ω)lt r such that
supuisinΦminus1(minusinfinr]Ψ(u)
rltΨ(ω)
Φ(ω) (21)
and for each
λ isin Λ Φ(ω)
Ψ(ω)
r
supuisinΦminus1(minusinfinr]Ψ(u)1113888 1113889 (22)
the functional Jλ Φ minus λΨ satisfies the (PS)-condition and itis unbounded from below
)en for each λ isin Λ the functional Jλ admits at least twononzero critical points uλ1 uλ2 such that Jλ(uλ1)lt 0lt Jλ(uλ2)
In order to obtain the positive solution of problem (Df
λ )we establish the following strong maximum principle
Lemma 2 Fix u isin S such that either
u(k)gt 0 or minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k))ge 0
(23)
for all k isin Z(1 N) -en either ugt 0 in Z(1 N) or u equiv 0
Proof Let j isin Z(1 N) such that
u(j) min u(k) k isin Z(1 N) (24)
If u(j)gt 0 then it is easy know that ugt 0 in Z(1 N)If u(j)le 0 then by (23) we have
minusΔ ϕp(Δu(j minus 1))1113872 1113873ge minuss(j)ϕq(u(j))ge 0 (25)
that is
ϕp(Δu(j))leϕp(Δu(j minus 1)) (26)
Since ϕp(u) is increasing in u we have
Δu(j)leΔu(j minus 1) (27)
By the definition of u(j) we know that
Δu(j)ge 0
Δu(j minus 1)le 0(28)
By combining (27) with (28) we get u(j + 1) u(j)
u(j minus 1) If j minus 1 0 we have u(j) u(j minus 1) 0 Other-wise j minus 1 isin Z(1 N) replacing j minus 1 by j we knowu(j minus 2) u(j minus 1) Continuing in this way we haveu(j) u(j minus 1) middot middot middot u(0) 0 Similarly we haveu(j) u(j + 1) middot middot middot u(N + 1) )us u(k) u(0) 0and forallk isin Z(1 N)
Now put
F+(k t) 1113946
t
0f k ξ+
( 1113857dξ forall(k t) isin Z(1 N) times R (29)
where ξ+ max ξ 0
Define J+λ Φ1 +Φ2 minus λΨ+ and
Ψ+(u) 1113936Nk1 F+(k u(k)) Standard arguments show that
J+λ isin C1(S R) and the critical points of J+
λ are precisely thesolutions of the following problem
minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) λf k u+(k)( ) k isin Z(1 N)
u(0) Δu(N) 0
⎧⎨
⎩ Df+
λ1113872 1113873 (30)
Lemma 3 If f(k 0)ge 0 for each k isin Z(1 N) any nonzerocritical point of the functional J+
λ is a positive solution ofproblem (D
f
λ )
Proof Since a critical point of J+λ is a solution of problem
Df+
λ the conclusion follows by the discrete maximumprinciple ([33] Proposition 1)
4 Discrete Dynamics in Nature and Society
Next we suppose that f(k 0)ge 0 and f(k x) f(k 0)
for all xle 0 and for all k isin Z(1 N) Put
Linfin ≔ minkisinZ(1N)
liminft⟶+infin
F(k t)
tp
1113957s 1113944
N
k1s(k)
(31)
we have the following result
Lemma 4 If Linfin gt 0 then Jλ satisfies (PS)-condition and it isunbounded from below for all λ isin (2pN + 1113957s minus 2pminus1qLinfin +infin)
Proof Let λgt 2pN + 1113957s minus 2pminus 1qLinfin We consider a sequenceun1113864 1113865nisinNsube S such that Jλ(un)⟶ c isin R and Jλprime(un)⟶ 0 as
n⟶ +infin Let u+n max un 01113864 1113865 and uminus
n max minusun 01113864 1113865 forall n isin N We first prove that uminus
n1113864 1113865 is bounded On one handwe have
Δuminusn (k minus 1)
11138681113868111386811138681113868111386811138681113868p le minus ϕp Δun(k minus 1)( 1113857Δuminus
n (k minus 1)
s(k)∣uminusn (k)∣q minuss(k)∣un(k)∣qminus2
un(k)uminusn (k)
(32)
for all k isin Z(1 N) So
uminusn
p 1113944
N+1
k1Δuminus
n (k minus 1)1113868111386811138681113868
1113868111386811138681113868p
le minus 1113944N+1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k) + 1113944
N1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k minus 1)
minus 1113944N
k1ϕp Δun(k minus 1)( 1113857u
minusn(k) + 1113944
N
k1ϕp Δun(k)( 1113857u
minusn(k)
1113944N
k1Δϕp Δun(k minus 1)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
1113944
N
k1s(k) u
minusn(k)
11138681113868111386811138681113868111386811138681113868q
minus 1113944N
k1s(k)ϕq un(k)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
(33)
On the other hand we assume that
f(k u) f(k u) if ugt 0
f(k 0) if ule 01113896 (34)
for each k isin Z(1 N) then
Ψprime un( 1113857 uminusn( 1113857 1113944
N
k1f k un(k)( 1113857u
minusn (k)ge 0 (35)
)erefore
uminusn
p le uminusn
p+ 1113944
N
k1s(k)|u(k)|
q
le minusΦ1prime un( 1113857 uminusn( 1113857 minusΦ2prime un( 1113857 u
minusn( 1113857 + λΨprime un( 1113857 u
minusn( 1113857
minusJλprime un( 1113857 uminusn( 1113857
(36)
for all n isin N which leads to uminusn pminus 1⟶ 0 as n⟶ +infin
So we have uminusn ⟶ 0 as n⟶ +infin It means that there
exists an Mgt 0 such that uminusn leM From (10) we know that
uminusn infin leN1minus 1pM c for all k isin Z(1 N)
Next we suppose that the sequence un1113864 1113865 is unboundedthat is u+
n1113864 1113865 is unboundedAs Linfin gt 0 we know that there exists an l isin R such that
Linfin gt lgt 2pN + 1113957s minus 2pminus 1λq From the definition of Linfin thereis δk gt 0 such that F(k t)gt l|t|p for all tgt δk Furthermoresince F(k t) is a continuous function there exists a constantC(k)ge 0 such that F(k t)ge l|t|p minus C(k) with t isin [minusc δk]
)us F(k t)ge l|t|p minus C(k) for all sge minus c and k isin Z(1 N)
We can obtain that
1113944
N
k1F k un(k)( 1113857ge 1113944
N
k1l un(k)1113868111386811138681113868
1113868111386811138681113868p
minus Cge l un
p
infin minus C (37)
for all k isin Z(1 N) where C 1113936Nk1 C(k) that is
Ψ un( 1113857ge l un
p
infin minus C (38)
Hence for all un such that uninfinge 1 we conclude that
Discrete Dynamics in Nature and Society 5
Jλ un( 1113857 1113936
N+1k1 Δun(k minus 1)
11138681113868111386811138681113868111386811138681113868p
p+
1113936Nk1s(k) un(k)
11138681113868111386811138681113868111386811138681113868q
qminus λΨ un( 1113857
le2pminus 1
p1113944
N
k1un(k)
11138681113868111386811138681113868111386811138681113868p
+ 1113944N
k1un(k minus 1)
11138681113868111386811138681113868111386811138681113868p⎛⎝ ⎞⎠
+1q
1113944
N
k1s(k)∣un(k) ∣q minus λΨ un( 1113857
le2pminus 1(2N minus 1)
pun
p
infin +1113957s
qun
q
infin minus λΨ un( 1113857
le2pN + 1113957s minus 2pminus 1
qminus λl1113888 1113889 un
p
infin + λC
(39)
Since 2pN + 1113957s minus 2pminus 1q minus λllt 0 we can getlimn⟶+infin Jλ(un) minusinfin and this is absurd Hence Jλ sat-isfies (PS)-condition
Let un1113864 1113865 be such that uminusn1113864 1113865 is bounded and u+
n1113864 1113865 is un-bounded From the proof above we can see that Jλ is un-bounded from below
3 Main Results
)e main results of this paper are as follows
Theorem 2 Let f Z(1 N) times R⟶ R be a continuousfunction satisfying f(k 0)ge 0 for all k isin Z(1 N) If there aretwo constants c and d with dlt c such that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ∣ lec
F(k ξ)max1cq
1cp
1113882 1113883
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(40)
)en for each λ isin Λ1 with
Λ1 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1pNpminus 1( 1113857min cq cp
1113936Nk1 max |ξ∣lecF(k ξ)
⎛⎝ ⎞⎠ (41)
the problem (Df
λ ) admits at least two positive solutions
Proof Put ΦΨ as in (18) It is clear that infX(Φ) Φ(0)
Ψ(0) 0 According to Lemma 3 we know that a nonzerocritical point in S of the functional J+
λ is precisely a positivesolution of problem (D
f
λ ) Next we just need to provecondition (21) of Lemma 1
We observe that Linfin gt 0 from (40) and Λ1 is nonde-generate Fix λ isin Λ1 Lemma 4 ensures that Jλ satisfies
(PS)-condition for all λgt 2pN + 1113957s minus 2pminus 1qLinfin and it isunbounded from below We let u isin Φminus 1(minusinfin r] that is(upp) + (1113936
Nk1 s(k)|u(k)|qq)le r Put
r 1 + slowastN
pminus 1
pNpminus1 min cq c
p1113864 1113865 (42)
If r (1 + slowastNpminus 1pNpminus 1)cq it means that cge 1
According to (8) we obtain
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max c cqp
1113966 1113967 c (43)
If r (1 + slowastNpminus 1pNpminus 1)cp we know 0lt clt 1 then
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max cpq
c1113966 1113967 c (44)
To sum up we know that |u(k)|le c for all k isin Z(1 N)
Furthermore we have
Ψ(u) 1113944N
k1F(k u(k))le 1113944
N
k1max|ξ|lec
F(k ξ) (45)
for all u isin S with Φ(u)le r Hence
supuisinΦminus1(minusinfinr]Ψ(u)
rle
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ|lec
F(k ξ)max1cq
1cp
1113882 1113883
(46)
6 Discrete Dynamics in Nature and Society
Now let ω(k) d for all k isin Z(1 N) and ω(0)
Δω(N) 0 Clearly ω isin S It is easy to account that Φ(ω)
dppminus 1 + dqqminus 11113957s then
Ψ(ω)
Φ(ω)
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s (47)
Consequently from (46) (47) and assumption (40) wecan obtain
supuisinΦminus 1(minusinfinr]
Ψ(u)
rltΨ(ω)
Φ(ω)
(48)
Moreover because 0lt dlt c and from (40) we obtain
0ltdpp
minus 1+ d
minus 11113957slt
1 + slowastNpminus 1
pNpminus1 min cq c
p1113864 1113865 (49)
that is mean that 0ltΦ(ω)lt r
Hence the problem (Df
λ ) admits at least two positivesolutions by Lemma 1 and Lemma 3 for all λ isin Λ1
Remark 1 If f(k t) is a nonnegative function and there aretwo positive constants c d with dlt c such that
pNpminus 1
1 + slowastNpminus1 max
1113936Nk1 F(k c)
cq1113936
Nk1 F(k c)
cp1113896 1113897
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(50)
then the result of)eorem 2 is also valid for each λ isin Λ2 with
Λ2 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1 mincq
1113936Nk1 F(k c)
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889 (51)
)ere are some consequences of )eorem 2 as follows
Corollary 1 Let g R⟶ [0 +infin) be a continuous functionsuch that f(k t) α(k)g(t) where α(k)gt 0 for allk isin Z(1 N) Put A 1113936
Nk1 α(k) G(t) 1113938
t
0 g(ξ)dξ for allt isin R and Llowastinfin min
kisin[1N]α(k) liminf
t⟶+infin(G(t)tp)gt 0
If there exists cgt dgt 0 such that
pNpminus 1
1 + slowastNpminus1 AG(c)max
1cq
1cp
1113882 1113883
ltminAG(d)
dppminus1 + dqqminus11113957s
qLlowastinfin2pN + 1113957s minus 2pminus11113896 1113897
(52)
then the problem Df
λ has at least two positive solutions foreach λ isin Λ3 with
Λ3 ≔ maxdppminus 1 + dqqminus 11113957s
AG(d)2pN + 1113957s minus 2pminus 1
qLlowastinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1min cq cp
AG(c)1113888 1113889 (53)
Proof Consider the function f Z(1 N) times R⟶ R isgiven as
f(k ξ) α(k)g(ξ) forallk isin Z(1 N) ξ isin R (54)
so that
1113944
N
k1max|ξ|lec
F(k ξ) AG(c) 1113944N
k1F(k d) AG(d) (55)
)en the conclusion can be obtained by)eorem 2
Corollary 2 Assume f be a continuous function withf(k 0)ge 0 and
limsupt⟶0+
F(k t)
tp +infin (56)
limt⟶+infin
F(k t)
tp +infin (57)
for all k isin Z(0 N) Put λlowast (1 + slowastNpminus 1pNpminus 1)min supcgt01113864
(cq1113936Nk1max |ξ∣lecF(k ξ)) supcgt0(cp 1113936
Nk1 max |ξ∣lecF(k ξ))
-en for each λ isin (0 λlowast) the problem (Df
λ ) admits at least twopositive solutions
Proof We know that Linfin +infin from (57) Fix λ isin (0 λlowast)and then there exists cgt 0 such that
λlt1 + slowastN
pminus 1
pNpminus1
middot min supcgt0
cq
1113936Nk1 max|ξ|lecF(k ξ)
supcgt0
cp
1113936Nk1 max|ξ|lecF(k ξ)
⎧⎨
⎩
⎫⎬
⎭
(58)
From (56) we can also obtain
limsupt⟶0+
1113936Nk1 F(k t)
tp +infin (59)
and then there exists d isin (0 c) such that (1113936Nk1 F(k d)
dppminus 1 + dqqminus 11113957s)gt (1λ) )erefore )eorem 2 ensures theconclusion
Discrete Dynamics in Nature and Society 7
Remark 2 If f(k t) is a nonnegative function for all(k t) isin Z(1 N) times [0 +infin) As long as condition (56) holdsfor at least one k isin Z(1 N) then Corollary 2 ensures thatthe solutions are obtained for each
λ isin 01 + slowastN
pminus 1
pNpminus1 min supcgt0
cq
1113936Nk1 F(k c)
supcgt0
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889
(60)
Remark 3 When f(k t) f(t) for all k isin Z(1 N) )e-orem 1 can be ensured by Corollary 2 Obviously condition(4(a)) implies f(0)ge 0 Specially if f is nonnegative weonly need condition (4(a)) to get the corresponding so-lution for each
λ isin 01 + slowastN
pminus 1
pNpmin sup
cgt0
cq
F(c) sup
cgt0
cp
F(c)1113896 11138971113888 1113889 (61)
Example 1 Let p 4 q 2 N 3 s(k) 12 andf(k t)
et
Put
x(c) c2
ec minus 1 (62)
)en
xprime(c) c (2 minus c)ec minus 2( )
ec minus 1( )2 (63)
Let z(c) (2 minus c)ec minus 2 then zprime(c) (1 minus c)ec So z(c)
is increasing in c isin (0 1) and decreasing in c isin (1 +infin)Since z(0) 0 and z(+infin) minusinfin there exists an uniquec1 isin (1 +infin) such that z(c1) 0 )us x(c) in increasing inc isin (0 c1) and decreasing in c isin (c1 +infin) )is means thatsupcgt0 x(c) x(c1) In fact c1 asymp 15936
Similarly put y(c) c4ec minus 1 we can show that thereexists a unique c2 isin (3 +infin) such that supcgt0y(c) y(c2)
In fact c2 asymp 39207
Since
y c2( 1113857 c42
ec2 minus 1gt
c41ec1 minus 1
c21x c1( 1113857gt x c1( 1113857 (64)
then
1 + slowastNpminus 1
pNpmin sup
cgt0x(c) sup
cgt0y(c)1113896 1113897
1 + slowastN
pminus 1
pNpx c1( 1113857 asymp 06496
(65)
)erefore for each λ isin (0 06496) the problem
minusΔ ϕ4(Δu(k minus 1))( 1113857 + 12ϕ2(u(k)) λeu(k) k 1 2 3
u(0) Δu(3) 0
⎧⎨
⎩
(66)
admits at least two positive solutions
Example 2 Let N 3 p 3 and q 2 and f be a functionas follows
f(k t)
0 if tlt 0kt
3radic
if 0le tle 1kt
3radic
+ 225t2 minus 225 if tgt 1
⎧⎪⎪⎨
⎪⎪⎩(67)
From Remark 1 we can choose c 1 and d 002 Easycalculation shows that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1F(k c)max
1cq
1cp
1113882 1113883 pNpminus 1
1 + slowastNpminus1 1113944
3
k111139461
0
kt
3radic
dt
asymp 13693
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s
11139363k1 1113946
002
0
kt
3radic
dt
dppminus1 + dqqminus11113957sasymp 31386
qLinfin2pN + 1113957s minus 2pminus1
166
limt⟶+infin
kt
3radic
+ 225t2 minus 225t2
asymp 34091
(68)
which satisfy condition (50) )us for each λ isin(03186 07303) the problem
minusΔ ϕ3(Δu(k minus 1))( 1113857 + 8ϕ2(u(k)) λf(k u(k)) k 1 2 3
u(0) Δu(3) 01113896
(69)
admits at least two positive solutions
Data Availability
No data were used to support the study
Conflicts of Interest
)e authors declare that there are no conflicts of interestregarding the publication of this paper
Authorsrsquo Contributions
All authors contributed equally to the writing of this paperAll authors read and approved the final manuscript
Acknowledgments
)is work was supported by the National Natural ScienceFoundation of China (Grant no 11971126) and Program forChangjiang Scholars and Innovative Research Team inUniversity (Grant no IRTminus16R16)
References
[1] R P Agarwal Difference Equations and Inequalities -eoryMethods and Applications Marcel Dekker Inc New YorkNY USA 2000
[2] A Kristaly M Mihailescu V Radulescu and S TersianldquoSpectral estimates for a nonhomogeneous difference
8 Discrete Dynamics in Nature and Society
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9
minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) λf(k u(k)) k isin Z(1 N)
u(0) Δu(N) 0
⎧⎨
⎩ Df
λ1113872 1113873 (3)
where Z(a b) denote the discrete interval a a + 1 b
for any integers a and b with alt b N be a positive integerf(k u) is continuous in u for each k isin Z(1 N) Δu(k)
u(k + 1) minus u(k) is the forward difference operatorϕr R⟶ R is the r-Laplacian given by ϕr(u) |u|rminus 2u
with u isin R 1lt qleplt +infin s(k)ge 0 for all k isin Z(1 N) andλ is a positive parameter
In this paper under suitable assumptions on the non-linearity f we use the theory of two nonzero critical points(see [30]) to ensure that there are at least two nonzerosolutions for problem (D
f
λ ) )e two nonzero critical pointstheorem is an appropriate combination of local minimumtheorem (see [31]) and classical AmbrosettindashRabinowitztheorem (see [32]) An important hypothesis of mountainpass theorem is PalaisndashSmale condition It satisfies the ap-plication of infinite dimensional space by requiring thecondition that the nonlinear term is stronger than p-superlinearity at infinity In order to obtain the existence oftwo nonzero solutions we can assume the classicalAmbrosettindashRabinowitz condition and nonlinear algebraiccondition (see (40) in )eorem 2) hold that is morewidespread than the p-sublinearity at zero Moreover whenwe require that f(k 0)ge 0 for all k isin Z(1 N) we can usestrongmaximum principle to obtain the existence of positivesolutions which has been proved in Lemma 2
Let slowast min s(k) k isin Z(1 N) a special case of ourmain result is stated as follows
Theorem 1 Let f R⟶ R be a continuous function suchthat
limt⟶0+
f(t)
tpminus14(a)
limt⟶+infin
f(t)
tpminus14(b)
then for each λ isin (0 (1 + slowastNpminus1pNp)min 1113864supcgt0(cq
max|ξ|lec 1113938ξ0 f(t)dt) supcgt0(cpmax|ξ|lec 1113938
ξ0 f(t)dt)1113865) the
problemminusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) λf(u(k)) k isin Z(1 N)
u(0) Δu(N) 0
⎧⎨
⎩
(5)
admits at least two positive solutions
)e structure of the article is as follows In Section 2some basic definitions and properties are given In Section 3we give the main results Under suitable hypothesis Lemma1 is used to obtain that the problem (D
f
λ ) possesses at leasttwo positive solutions Finally some examples are given toillustrate our main results
2 Preliminaries
In this section we recall some definitions notations andproperties Consider the N-dimensional Banach space
S u Z(0 N + 1)⟶ R u(0) Δu(N) 0 (6)
and define the norm
u 1113944N+1
k1|Δu(k minus 1)|
p⎛⎝ ⎞⎠
1p
(7)
and uinfin max |u(k)| k isin Z(1 N) is another norm in S
Proposition 1 -e following inequality holds
uinfin lemaxpNpminus1
1 + slowastNpminus11113888 1113889
1qup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1q
pNpminus1
1 + slowastNpminus11113888 1113889
1pup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1p⎧⎨
⎩
⎫⎬
⎭ (8)
Proof Let u isin S then there exist klowast isin Z(1 N) such that|u(klowast)| max |u(k)| k isin Z(1 N)
Since
u klowast
( 11138571113868111386811138681113868
1113868111386811138681113868 1113944
klowast
k1Δu(k minus 1)|le 1113944
klowast
k11)
1minus1p1113944
klowast
k1|Δu(k minus 1)|
p)1p leN
1minus1pu⎛⎝⎛⎝
1113868111386811138681113868111386811138681113868111386811138681113868(9)
2 Discrete Dynamics in Nature and Society
then
upinfin leN
pminus1u
p (10)
If ||u||infin gt 1 then
1 + slowastNpminus1( 1113857uq
infinp
leupinfin + Npminus1 1113936
Nk1 s(k)|u(k)|q
p
leNpminus 1
pu
p+ 1113944
N
k1s(k)|u(k)|
q⎛⎝ ⎞⎠leNpminus1
pu
p+
Npminus1
q1113944
N
k1s(k)|u(k)|
q
(11)
that is
uinfin lepNpminus 1
1 + slowastNpminus11113888 1113889
1qup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1q
(12)
If uinfin le 1 then
1 + slowastNpminus 1( 1113857u
pinfin
ple
upinfin + Npminus 1 1113936
Nk1 s(k)|u(k)|q
p
leNpminus 1
pu
p+
Npminus 1
q1113944
N
k1s(k)|u(k)|
q
(13)
that is
uinfin lepNpminus 1
1 + slowastNpminus11113888 1113889
1p||u||p
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1p
(14)
In summary we have
uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1qup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1pup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1p⎧⎨
⎩
⎫⎬
⎭ (15)
Put
F(k t) ≔ 1113946t
0f(k ξ)dξ forall(k t) isin Z(1 N) times R (16)
and consider the function Jλ S⟶ R for all λgt 0 by
Jλ Φ minus λΨ (17)
where
Φ ≔ Φ1 +Φ2Φ1(u) up
p
Φ2(u) 1113936
Nk1 s(k)|u(k)|q
q
Ψ(u) ≔ 1113944N
k1F(k u(k))
(18)
It is clear that Φ1Φ2Ψ isin C1(SR) and their Gateauxderivatives at the point u isin S are given by
Φ1prime(u)(v) 1113944N+1
k1ϕp(Δu(k minus 1))v(k)
minus 1113944N+1
k1ϕp(Δu(k minus 1))v(k minus 1)
1113944N
k1ϕp(Δu(k minus 1))v(k) minus 1113944
N
k0ϕp(Δu(k))v(k)
minus 1113944N
k1Δϕp(Δu(k minus 1))v(k)
Φ2prime(u)(v) 1113944N
k1s(k)ϕq(u(k))v(k)
Ψprime(u)(v) 1113944N
k1f(k u(k))v(k)
(19)
Discrete Dynamics in Nature and Society 3
for all u v isin S So we have
Jλprime(u)(v) 1113944N
k1minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) minus λf(k u(k))1113960 1113961v(k) (20)
Hence a critical point u of Jλ is a solution of problem(D
f
λ )Now we recall a definition and a two nonzero critical
points theorem for the readerrsquos convenience
Definition 1 Let X be a real Banach space we say that aGateaux differentiable function Jλ X⟶ R satisfies the(PS)-condition if any sequence un1113864 1113865nisinNsubeX such that
(i) Jλ(un)⟶ c isin R as n⟶ +infin(ii) Jλprime(un)⟶ 0 as n⟶ +infin has a convergent
subsequence
Lemma 1 Let X be a real Banach space and ΦΨ isin C1(SR)
such that infX(Φ) Φ(0) Ψ(0) 0 Assume that there arer isin R and ω isin X with 0ltΦ(ω)lt r such that
supuisinΦminus1(minusinfinr]Ψ(u)
rltΨ(ω)
Φ(ω) (21)
and for each
λ isin Λ Φ(ω)
Ψ(ω)
r
supuisinΦminus1(minusinfinr]Ψ(u)1113888 1113889 (22)
the functional Jλ Φ minus λΨ satisfies the (PS)-condition and itis unbounded from below
)en for each λ isin Λ the functional Jλ admits at least twononzero critical points uλ1 uλ2 such that Jλ(uλ1)lt 0lt Jλ(uλ2)
In order to obtain the positive solution of problem (Df
λ )we establish the following strong maximum principle
Lemma 2 Fix u isin S such that either
u(k)gt 0 or minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k))ge 0
(23)
for all k isin Z(1 N) -en either ugt 0 in Z(1 N) or u equiv 0
Proof Let j isin Z(1 N) such that
u(j) min u(k) k isin Z(1 N) (24)
If u(j)gt 0 then it is easy know that ugt 0 in Z(1 N)If u(j)le 0 then by (23) we have
minusΔ ϕp(Δu(j minus 1))1113872 1113873ge minuss(j)ϕq(u(j))ge 0 (25)
that is
ϕp(Δu(j))leϕp(Δu(j minus 1)) (26)
Since ϕp(u) is increasing in u we have
Δu(j)leΔu(j minus 1) (27)
By the definition of u(j) we know that
Δu(j)ge 0
Δu(j minus 1)le 0(28)
By combining (27) with (28) we get u(j + 1) u(j)
u(j minus 1) If j minus 1 0 we have u(j) u(j minus 1) 0 Other-wise j minus 1 isin Z(1 N) replacing j minus 1 by j we knowu(j minus 2) u(j minus 1) Continuing in this way we haveu(j) u(j minus 1) middot middot middot u(0) 0 Similarly we haveu(j) u(j + 1) middot middot middot u(N + 1) )us u(k) u(0) 0and forallk isin Z(1 N)
Now put
F+(k t) 1113946
t
0f k ξ+
( 1113857dξ forall(k t) isin Z(1 N) times R (29)
where ξ+ max ξ 0
Define J+λ Φ1 +Φ2 minus λΨ+ and
Ψ+(u) 1113936Nk1 F+(k u(k)) Standard arguments show that
J+λ isin C1(S R) and the critical points of J+
λ are precisely thesolutions of the following problem
minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) λf k u+(k)( ) k isin Z(1 N)
u(0) Δu(N) 0
⎧⎨
⎩ Df+
λ1113872 1113873 (30)
Lemma 3 If f(k 0)ge 0 for each k isin Z(1 N) any nonzerocritical point of the functional J+
λ is a positive solution ofproblem (D
f
λ )
Proof Since a critical point of J+λ is a solution of problem
Df+
λ the conclusion follows by the discrete maximumprinciple ([33] Proposition 1)
4 Discrete Dynamics in Nature and Society
Next we suppose that f(k 0)ge 0 and f(k x) f(k 0)
for all xle 0 and for all k isin Z(1 N) Put
Linfin ≔ minkisinZ(1N)
liminft⟶+infin
F(k t)
tp
1113957s 1113944
N
k1s(k)
(31)
we have the following result
Lemma 4 If Linfin gt 0 then Jλ satisfies (PS)-condition and it isunbounded from below for all λ isin (2pN + 1113957s minus 2pminus1qLinfin +infin)
Proof Let λgt 2pN + 1113957s minus 2pminus 1qLinfin We consider a sequenceun1113864 1113865nisinNsube S such that Jλ(un)⟶ c isin R and Jλprime(un)⟶ 0 as
n⟶ +infin Let u+n max un 01113864 1113865 and uminus
n max minusun 01113864 1113865 forall n isin N We first prove that uminus
n1113864 1113865 is bounded On one handwe have
Δuminusn (k minus 1)
11138681113868111386811138681113868111386811138681113868p le minus ϕp Δun(k minus 1)( 1113857Δuminus
n (k minus 1)
s(k)∣uminusn (k)∣q minuss(k)∣un(k)∣qminus2
un(k)uminusn (k)
(32)
for all k isin Z(1 N) So
uminusn
p 1113944
N+1
k1Δuminus
n (k minus 1)1113868111386811138681113868
1113868111386811138681113868p
le minus 1113944N+1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k) + 1113944
N1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k minus 1)
minus 1113944N
k1ϕp Δun(k minus 1)( 1113857u
minusn(k) + 1113944
N
k1ϕp Δun(k)( 1113857u
minusn(k)
1113944N
k1Δϕp Δun(k minus 1)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
1113944
N
k1s(k) u
minusn(k)
11138681113868111386811138681113868111386811138681113868q
minus 1113944N
k1s(k)ϕq un(k)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
(33)
On the other hand we assume that
f(k u) f(k u) if ugt 0
f(k 0) if ule 01113896 (34)
for each k isin Z(1 N) then
Ψprime un( 1113857 uminusn( 1113857 1113944
N
k1f k un(k)( 1113857u
minusn (k)ge 0 (35)
)erefore
uminusn
p le uminusn
p+ 1113944
N
k1s(k)|u(k)|
q
le minusΦ1prime un( 1113857 uminusn( 1113857 minusΦ2prime un( 1113857 u
minusn( 1113857 + λΨprime un( 1113857 u
minusn( 1113857
minusJλprime un( 1113857 uminusn( 1113857
(36)
for all n isin N which leads to uminusn pminus 1⟶ 0 as n⟶ +infin
So we have uminusn ⟶ 0 as n⟶ +infin It means that there
exists an Mgt 0 such that uminusn leM From (10) we know that
uminusn infin leN1minus 1pM c for all k isin Z(1 N)
Next we suppose that the sequence un1113864 1113865 is unboundedthat is u+
n1113864 1113865 is unboundedAs Linfin gt 0 we know that there exists an l isin R such that
Linfin gt lgt 2pN + 1113957s minus 2pminus 1λq From the definition of Linfin thereis δk gt 0 such that F(k t)gt l|t|p for all tgt δk Furthermoresince F(k t) is a continuous function there exists a constantC(k)ge 0 such that F(k t)ge l|t|p minus C(k) with t isin [minusc δk]
)us F(k t)ge l|t|p minus C(k) for all sge minus c and k isin Z(1 N)
We can obtain that
1113944
N
k1F k un(k)( 1113857ge 1113944
N
k1l un(k)1113868111386811138681113868
1113868111386811138681113868p
minus Cge l un
p
infin minus C (37)
for all k isin Z(1 N) where C 1113936Nk1 C(k) that is
Ψ un( 1113857ge l un
p
infin minus C (38)
Hence for all un such that uninfinge 1 we conclude that
Discrete Dynamics in Nature and Society 5
Jλ un( 1113857 1113936
N+1k1 Δun(k minus 1)
11138681113868111386811138681113868111386811138681113868p
p+
1113936Nk1s(k) un(k)
11138681113868111386811138681113868111386811138681113868q
qminus λΨ un( 1113857
le2pminus 1
p1113944
N
k1un(k)
11138681113868111386811138681113868111386811138681113868p
+ 1113944N
k1un(k minus 1)
11138681113868111386811138681113868111386811138681113868p⎛⎝ ⎞⎠
+1q
1113944
N
k1s(k)∣un(k) ∣q minus λΨ un( 1113857
le2pminus 1(2N minus 1)
pun
p
infin +1113957s
qun
q
infin minus λΨ un( 1113857
le2pN + 1113957s minus 2pminus 1
qminus λl1113888 1113889 un
p
infin + λC
(39)
Since 2pN + 1113957s minus 2pminus 1q minus λllt 0 we can getlimn⟶+infin Jλ(un) minusinfin and this is absurd Hence Jλ sat-isfies (PS)-condition
Let un1113864 1113865 be such that uminusn1113864 1113865 is bounded and u+
n1113864 1113865 is un-bounded From the proof above we can see that Jλ is un-bounded from below
3 Main Results
)e main results of this paper are as follows
Theorem 2 Let f Z(1 N) times R⟶ R be a continuousfunction satisfying f(k 0)ge 0 for all k isin Z(1 N) If there aretwo constants c and d with dlt c such that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ∣ lec
F(k ξ)max1cq
1cp
1113882 1113883
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(40)
)en for each λ isin Λ1 with
Λ1 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1pNpminus 1( 1113857min cq cp
1113936Nk1 max |ξ∣lecF(k ξ)
⎛⎝ ⎞⎠ (41)
the problem (Df
λ ) admits at least two positive solutions
Proof Put ΦΨ as in (18) It is clear that infX(Φ) Φ(0)
Ψ(0) 0 According to Lemma 3 we know that a nonzerocritical point in S of the functional J+
λ is precisely a positivesolution of problem (D
f
λ ) Next we just need to provecondition (21) of Lemma 1
We observe that Linfin gt 0 from (40) and Λ1 is nonde-generate Fix λ isin Λ1 Lemma 4 ensures that Jλ satisfies
(PS)-condition for all λgt 2pN + 1113957s minus 2pminus 1qLinfin and it isunbounded from below We let u isin Φminus 1(minusinfin r] that is(upp) + (1113936
Nk1 s(k)|u(k)|qq)le r Put
r 1 + slowastN
pminus 1
pNpminus1 min cq c
p1113864 1113865 (42)
If r (1 + slowastNpminus 1pNpminus 1)cq it means that cge 1
According to (8) we obtain
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max c cqp
1113966 1113967 c (43)
If r (1 + slowastNpminus 1pNpminus 1)cp we know 0lt clt 1 then
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max cpq
c1113966 1113967 c (44)
To sum up we know that |u(k)|le c for all k isin Z(1 N)
Furthermore we have
Ψ(u) 1113944N
k1F(k u(k))le 1113944
N
k1max|ξ|lec
F(k ξ) (45)
for all u isin S with Φ(u)le r Hence
supuisinΦminus1(minusinfinr]Ψ(u)
rle
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ|lec
F(k ξ)max1cq
1cp
1113882 1113883
(46)
6 Discrete Dynamics in Nature and Society
Now let ω(k) d for all k isin Z(1 N) and ω(0)
Δω(N) 0 Clearly ω isin S It is easy to account that Φ(ω)
dppminus 1 + dqqminus 11113957s then
Ψ(ω)
Φ(ω)
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s (47)
Consequently from (46) (47) and assumption (40) wecan obtain
supuisinΦminus 1(minusinfinr]
Ψ(u)
rltΨ(ω)
Φ(ω)
(48)
Moreover because 0lt dlt c and from (40) we obtain
0ltdpp
minus 1+ d
minus 11113957slt
1 + slowastNpminus 1
pNpminus1 min cq c
p1113864 1113865 (49)
that is mean that 0ltΦ(ω)lt r
Hence the problem (Df
λ ) admits at least two positivesolutions by Lemma 1 and Lemma 3 for all λ isin Λ1
Remark 1 If f(k t) is a nonnegative function and there aretwo positive constants c d with dlt c such that
pNpminus 1
1 + slowastNpminus1 max
1113936Nk1 F(k c)
cq1113936
Nk1 F(k c)
cp1113896 1113897
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(50)
then the result of)eorem 2 is also valid for each λ isin Λ2 with
Λ2 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1 mincq
1113936Nk1 F(k c)
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889 (51)
)ere are some consequences of )eorem 2 as follows
Corollary 1 Let g R⟶ [0 +infin) be a continuous functionsuch that f(k t) α(k)g(t) where α(k)gt 0 for allk isin Z(1 N) Put A 1113936
Nk1 α(k) G(t) 1113938
t
0 g(ξ)dξ for allt isin R and Llowastinfin min
kisin[1N]α(k) liminf
t⟶+infin(G(t)tp)gt 0
If there exists cgt dgt 0 such that
pNpminus 1
1 + slowastNpminus1 AG(c)max
1cq
1cp
1113882 1113883
ltminAG(d)
dppminus1 + dqqminus11113957s
qLlowastinfin2pN + 1113957s minus 2pminus11113896 1113897
(52)
then the problem Df
λ has at least two positive solutions foreach λ isin Λ3 with
Λ3 ≔ maxdppminus 1 + dqqminus 11113957s
AG(d)2pN + 1113957s minus 2pminus 1
qLlowastinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1min cq cp
AG(c)1113888 1113889 (53)
Proof Consider the function f Z(1 N) times R⟶ R isgiven as
f(k ξ) α(k)g(ξ) forallk isin Z(1 N) ξ isin R (54)
so that
1113944
N
k1max|ξ|lec
F(k ξ) AG(c) 1113944N
k1F(k d) AG(d) (55)
)en the conclusion can be obtained by)eorem 2
Corollary 2 Assume f be a continuous function withf(k 0)ge 0 and
limsupt⟶0+
F(k t)
tp +infin (56)
limt⟶+infin
F(k t)
tp +infin (57)
for all k isin Z(0 N) Put λlowast (1 + slowastNpminus 1pNpminus 1)min supcgt01113864
(cq1113936Nk1max |ξ∣lecF(k ξ)) supcgt0(cp 1113936
Nk1 max |ξ∣lecF(k ξ))
-en for each λ isin (0 λlowast) the problem (Df
λ ) admits at least twopositive solutions
Proof We know that Linfin +infin from (57) Fix λ isin (0 λlowast)and then there exists cgt 0 such that
λlt1 + slowastN
pminus 1
pNpminus1
middot min supcgt0
cq
1113936Nk1 max|ξ|lecF(k ξ)
supcgt0
cp
1113936Nk1 max|ξ|lecF(k ξ)
⎧⎨
⎩
⎫⎬
⎭
(58)
From (56) we can also obtain
limsupt⟶0+
1113936Nk1 F(k t)
tp +infin (59)
and then there exists d isin (0 c) such that (1113936Nk1 F(k d)
dppminus 1 + dqqminus 11113957s)gt (1λ) )erefore )eorem 2 ensures theconclusion
Discrete Dynamics in Nature and Society 7
Remark 2 If f(k t) is a nonnegative function for all(k t) isin Z(1 N) times [0 +infin) As long as condition (56) holdsfor at least one k isin Z(1 N) then Corollary 2 ensures thatthe solutions are obtained for each
λ isin 01 + slowastN
pminus 1
pNpminus1 min supcgt0
cq
1113936Nk1 F(k c)
supcgt0
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889
(60)
Remark 3 When f(k t) f(t) for all k isin Z(1 N) )e-orem 1 can be ensured by Corollary 2 Obviously condition(4(a)) implies f(0)ge 0 Specially if f is nonnegative weonly need condition (4(a)) to get the corresponding so-lution for each
λ isin 01 + slowastN
pminus 1
pNpmin sup
cgt0
cq
F(c) sup
cgt0
cp
F(c)1113896 11138971113888 1113889 (61)
Example 1 Let p 4 q 2 N 3 s(k) 12 andf(k t)
et
Put
x(c) c2
ec minus 1 (62)
)en
xprime(c) c (2 minus c)ec minus 2( )
ec minus 1( )2 (63)
Let z(c) (2 minus c)ec minus 2 then zprime(c) (1 minus c)ec So z(c)
is increasing in c isin (0 1) and decreasing in c isin (1 +infin)Since z(0) 0 and z(+infin) minusinfin there exists an uniquec1 isin (1 +infin) such that z(c1) 0 )us x(c) in increasing inc isin (0 c1) and decreasing in c isin (c1 +infin) )is means thatsupcgt0 x(c) x(c1) In fact c1 asymp 15936
Similarly put y(c) c4ec minus 1 we can show that thereexists a unique c2 isin (3 +infin) such that supcgt0y(c) y(c2)
In fact c2 asymp 39207
Since
y c2( 1113857 c42
ec2 minus 1gt
c41ec1 minus 1
c21x c1( 1113857gt x c1( 1113857 (64)
then
1 + slowastNpminus 1
pNpmin sup
cgt0x(c) sup
cgt0y(c)1113896 1113897
1 + slowastN
pminus 1
pNpx c1( 1113857 asymp 06496
(65)
)erefore for each λ isin (0 06496) the problem
minusΔ ϕ4(Δu(k minus 1))( 1113857 + 12ϕ2(u(k)) λeu(k) k 1 2 3
u(0) Δu(3) 0
⎧⎨
⎩
(66)
admits at least two positive solutions
Example 2 Let N 3 p 3 and q 2 and f be a functionas follows
f(k t)
0 if tlt 0kt
3radic
if 0le tle 1kt
3radic
+ 225t2 minus 225 if tgt 1
⎧⎪⎪⎨
⎪⎪⎩(67)
From Remark 1 we can choose c 1 and d 002 Easycalculation shows that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1F(k c)max
1cq
1cp
1113882 1113883 pNpminus 1
1 + slowastNpminus1 1113944
3
k111139461
0
kt
3radic
dt
asymp 13693
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s
11139363k1 1113946
002
0
kt
3radic
dt
dppminus1 + dqqminus11113957sasymp 31386
qLinfin2pN + 1113957s minus 2pminus1
166
limt⟶+infin
kt
3radic
+ 225t2 minus 225t2
asymp 34091
(68)
which satisfy condition (50) )us for each λ isin(03186 07303) the problem
minusΔ ϕ3(Δu(k minus 1))( 1113857 + 8ϕ2(u(k)) λf(k u(k)) k 1 2 3
u(0) Δu(3) 01113896
(69)
admits at least two positive solutions
Data Availability
No data were used to support the study
Conflicts of Interest
)e authors declare that there are no conflicts of interestregarding the publication of this paper
Authorsrsquo Contributions
All authors contributed equally to the writing of this paperAll authors read and approved the final manuscript
Acknowledgments
)is work was supported by the National Natural ScienceFoundation of China (Grant no 11971126) and Program forChangjiang Scholars and Innovative Research Team inUniversity (Grant no IRTminus16R16)
References
[1] R P Agarwal Difference Equations and Inequalities -eoryMethods and Applications Marcel Dekker Inc New YorkNY USA 2000
[2] A Kristaly M Mihailescu V Radulescu and S TersianldquoSpectral estimates for a nonhomogeneous difference
8 Discrete Dynamics in Nature and Society
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9
then
upinfin leN
pminus1u
p (10)
If ||u||infin gt 1 then
1 + slowastNpminus1( 1113857uq
infinp
leupinfin + Npminus1 1113936
Nk1 s(k)|u(k)|q
p
leNpminus 1
pu
p+ 1113944
N
k1s(k)|u(k)|
q⎛⎝ ⎞⎠leNpminus1
pu
p+
Npminus1
q1113944
N
k1s(k)|u(k)|
q
(11)
that is
uinfin lepNpminus 1
1 + slowastNpminus11113888 1113889
1qup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1q
(12)
If uinfin le 1 then
1 + slowastNpminus 1( 1113857u
pinfin
ple
upinfin + Npminus 1 1113936
Nk1 s(k)|u(k)|q
p
leNpminus 1
pu
p+
Npminus 1
q1113944
N
k1s(k)|u(k)|
q
(13)
that is
uinfin lepNpminus 1
1 + slowastNpminus11113888 1113889
1p||u||p
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1p
(14)
In summary we have
uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1qup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1pup
p+
1113936Nk1 s(k)|u(k)|q
q1113888 1113889
1p⎧⎨
⎩
⎫⎬
⎭ (15)
Put
F(k t) ≔ 1113946t
0f(k ξ)dξ forall(k t) isin Z(1 N) times R (16)
and consider the function Jλ S⟶ R for all λgt 0 by
Jλ Φ minus λΨ (17)
where
Φ ≔ Φ1 +Φ2Φ1(u) up
p
Φ2(u) 1113936
Nk1 s(k)|u(k)|q
q
Ψ(u) ≔ 1113944N
k1F(k u(k))
(18)
It is clear that Φ1Φ2Ψ isin C1(SR) and their Gateauxderivatives at the point u isin S are given by
Φ1prime(u)(v) 1113944N+1
k1ϕp(Δu(k minus 1))v(k)
minus 1113944N+1
k1ϕp(Δu(k minus 1))v(k minus 1)
1113944N
k1ϕp(Δu(k minus 1))v(k) minus 1113944
N
k0ϕp(Δu(k))v(k)
minus 1113944N
k1Δϕp(Δu(k minus 1))v(k)
Φ2prime(u)(v) 1113944N
k1s(k)ϕq(u(k))v(k)
Ψprime(u)(v) 1113944N
k1f(k u(k))v(k)
(19)
Discrete Dynamics in Nature and Society 3
for all u v isin S So we have
Jλprime(u)(v) 1113944N
k1minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) minus λf(k u(k))1113960 1113961v(k) (20)
Hence a critical point u of Jλ is a solution of problem(D
f
λ )Now we recall a definition and a two nonzero critical
points theorem for the readerrsquos convenience
Definition 1 Let X be a real Banach space we say that aGateaux differentiable function Jλ X⟶ R satisfies the(PS)-condition if any sequence un1113864 1113865nisinNsubeX such that
(i) Jλ(un)⟶ c isin R as n⟶ +infin(ii) Jλprime(un)⟶ 0 as n⟶ +infin has a convergent
subsequence
Lemma 1 Let X be a real Banach space and ΦΨ isin C1(SR)
such that infX(Φ) Φ(0) Ψ(0) 0 Assume that there arer isin R and ω isin X with 0ltΦ(ω)lt r such that
supuisinΦminus1(minusinfinr]Ψ(u)
rltΨ(ω)
Φ(ω) (21)
and for each
λ isin Λ Φ(ω)
Ψ(ω)
r
supuisinΦminus1(minusinfinr]Ψ(u)1113888 1113889 (22)
the functional Jλ Φ minus λΨ satisfies the (PS)-condition and itis unbounded from below
)en for each λ isin Λ the functional Jλ admits at least twononzero critical points uλ1 uλ2 such that Jλ(uλ1)lt 0lt Jλ(uλ2)
In order to obtain the positive solution of problem (Df
λ )we establish the following strong maximum principle
Lemma 2 Fix u isin S such that either
u(k)gt 0 or minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k))ge 0
(23)
for all k isin Z(1 N) -en either ugt 0 in Z(1 N) or u equiv 0
Proof Let j isin Z(1 N) such that
u(j) min u(k) k isin Z(1 N) (24)
If u(j)gt 0 then it is easy know that ugt 0 in Z(1 N)If u(j)le 0 then by (23) we have
minusΔ ϕp(Δu(j minus 1))1113872 1113873ge minuss(j)ϕq(u(j))ge 0 (25)
that is
ϕp(Δu(j))leϕp(Δu(j minus 1)) (26)
Since ϕp(u) is increasing in u we have
Δu(j)leΔu(j minus 1) (27)
By the definition of u(j) we know that
Δu(j)ge 0
Δu(j minus 1)le 0(28)
By combining (27) with (28) we get u(j + 1) u(j)
u(j minus 1) If j minus 1 0 we have u(j) u(j minus 1) 0 Other-wise j minus 1 isin Z(1 N) replacing j minus 1 by j we knowu(j minus 2) u(j minus 1) Continuing in this way we haveu(j) u(j minus 1) middot middot middot u(0) 0 Similarly we haveu(j) u(j + 1) middot middot middot u(N + 1) )us u(k) u(0) 0and forallk isin Z(1 N)
Now put
F+(k t) 1113946
t
0f k ξ+
( 1113857dξ forall(k t) isin Z(1 N) times R (29)
where ξ+ max ξ 0
Define J+λ Φ1 +Φ2 minus λΨ+ and
Ψ+(u) 1113936Nk1 F+(k u(k)) Standard arguments show that
J+λ isin C1(S R) and the critical points of J+
λ are precisely thesolutions of the following problem
minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) λf k u+(k)( ) k isin Z(1 N)
u(0) Δu(N) 0
⎧⎨
⎩ Df+
λ1113872 1113873 (30)
Lemma 3 If f(k 0)ge 0 for each k isin Z(1 N) any nonzerocritical point of the functional J+
λ is a positive solution ofproblem (D
f
λ )
Proof Since a critical point of J+λ is a solution of problem
Df+
λ the conclusion follows by the discrete maximumprinciple ([33] Proposition 1)
4 Discrete Dynamics in Nature and Society
Next we suppose that f(k 0)ge 0 and f(k x) f(k 0)
for all xle 0 and for all k isin Z(1 N) Put
Linfin ≔ minkisinZ(1N)
liminft⟶+infin
F(k t)
tp
1113957s 1113944
N
k1s(k)
(31)
we have the following result
Lemma 4 If Linfin gt 0 then Jλ satisfies (PS)-condition and it isunbounded from below for all λ isin (2pN + 1113957s minus 2pminus1qLinfin +infin)
Proof Let λgt 2pN + 1113957s minus 2pminus 1qLinfin We consider a sequenceun1113864 1113865nisinNsube S such that Jλ(un)⟶ c isin R and Jλprime(un)⟶ 0 as
n⟶ +infin Let u+n max un 01113864 1113865 and uminus
n max minusun 01113864 1113865 forall n isin N We first prove that uminus
n1113864 1113865 is bounded On one handwe have
Δuminusn (k minus 1)
11138681113868111386811138681113868111386811138681113868p le minus ϕp Δun(k minus 1)( 1113857Δuminus
n (k minus 1)
s(k)∣uminusn (k)∣q minuss(k)∣un(k)∣qminus2
un(k)uminusn (k)
(32)
for all k isin Z(1 N) So
uminusn
p 1113944
N+1
k1Δuminus
n (k minus 1)1113868111386811138681113868
1113868111386811138681113868p
le minus 1113944N+1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k) + 1113944
N1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k minus 1)
minus 1113944N
k1ϕp Δun(k minus 1)( 1113857u
minusn(k) + 1113944
N
k1ϕp Δun(k)( 1113857u
minusn(k)
1113944N
k1Δϕp Δun(k minus 1)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
1113944
N
k1s(k) u
minusn(k)
11138681113868111386811138681113868111386811138681113868q
minus 1113944N
k1s(k)ϕq un(k)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
(33)
On the other hand we assume that
f(k u) f(k u) if ugt 0
f(k 0) if ule 01113896 (34)
for each k isin Z(1 N) then
Ψprime un( 1113857 uminusn( 1113857 1113944
N
k1f k un(k)( 1113857u
minusn (k)ge 0 (35)
)erefore
uminusn
p le uminusn
p+ 1113944
N
k1s(k)|u(k)|
q
le minusΦ1prime un( 1113857 uminusn( 1113857 minusΦ2prime un( 1113857 u
minusn( 1113857 + λΨprime un( 1113857 u
minusn( 1113857
minusJλprime un( 1113857 uminusn( 1113857
(36)
for all n isin N which leads to uminusn pminus 1⟶ 0 as n⟶ +infin
So we have uminusn ⟶ 0 as n⟶ +infin It means that there
exists an Mgt 0 such that uminusn leM From (10) we know that
uminusn infin leN1minus 1pM c for all k isin Z(1 N)
Next we suppose that the sequence un1113864 1113865 is unboundedthat is u+
n1113864 1113865 is unboundedAs Linfin gt 0 we know that there exists an l isin R such that
Linfin gt lgt 2pN + 1113957s minus 2pminus 1λq From the definition of Linfin thereis δk gt 0 such that F(k t)gt l|t|p for all tgt δk Furthermoresince F(k t) is a continuous function there exists a constantC(k)ge 0 such that F(k t)ge l|t|p minus C(k) with t isin [minusc δk]
)us F(k t)ge l|t|p minus C(k) for all sge minus c and k isin Z(1 N)
We can obtain that
1113944
N
k1F k un(k)( 1113857ge 1113944
N
k1l un(k)1113868111386811138681113868
1113868111386811138681113868p
minus Cge l un
p
infin minus C (37)
for all k isin Z(1 N) where C 1113936Nk1 C(k) that is
Ψ un( 1113857ge l un
p
infin minus C (38)
Hence for all un such that uninfinge 1 we conclude that
Discrete Dynamics in Nature and Society 5
Jλ un( 1113857 1113936
N+1k1 Δun(k minus 1)
11138681113868111386811138681113868111386811138681113868p
p+
1113936Nk1s(k) un(k)
11138681113868111386811138681113868111386811138681113868q
qminus λΨ un( 1113857
le2pminus 1
p1113944
N
k1un(k)
11138681113868111386811138681113868111386811138681113868p
+ 1113944N
k1un(k minus 1)
11138681113868111386811138681113868111386811138681113868p⎛⎝ ⎞⎠
+1q
1113944
N
k1s(k)∣un(k) ∣q minus λΨ un( 1113857
le2pminus 1(2N minus 1)
pun
p
infin +1113957s
qun
q
infin minus λΨ un( 1113857
le2pN + 1113957s minus 2pminus 1
qminus λl1113888 1113889 un
p
infin + λC
(39)
Since 2pN + 1113957s minus 2pminus 1q minus λllt 0 we can getlimn⟶+infin Jλ(un) minusinfin and this is absurd Hence Jλ sat-isfies (PS)-condition
Let un1113864 1113865 be such that uminusn1113864 1113865 is bounded and u+
n1113864 1113865 is un-bounded From the proof above we can see that Jλ is un-bounded from below
3 Main Results
)e main results of this paper are as follows
Theorem 2 Let f Z(1 N) times R⟶ R be a continuousfunction satisfying f(k 0)ge 0 for all k isin Z(1 N) If there aretwo constants c and d with dlt c such that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ∣ lec
F(k ξ)max1cq
1cp
1113882 1113883
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(40)
)en for each λ isin Λ1 with
Λ1 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1pNpminus 1( 1113857min cq cp
1113936Nk1 max |ξ∣lecF(k ξ)
⎛⎝ ⎞⎠ (41)
the problem (Df
λ ) admits at least two positive solutions
Proof Put ΦΨ as in (18) It is clear that infX(Φ) Φ(0)
Ψ(0) 0 According to Lemma 3 we know that a nonzerocritical point in S of the functional J+
λ is precisely a positivesolution of problem (D
f
λ ) Next we just need to provecondition (21) of Lemma 1
We observe that Linfin gt 0 from (40) and Λ1 is nonde-generate Fix λ isin Λ1 Lemma 4 ensures that Jλ satisfies
(PS)-condition for all λgt 2pN + 1113957s minus 2pminus 1qLinfin and it isunbounded from below We let u isin Φminus 1(minusinfin r] that is(upp) + (1113936
Nk1 s(k)|u(k)|qq)le r Put
r 1 + slowastN
pminus 1
pNpminus1 min cq c
p1113864 1113865 (42)
If r (1 + slowastNpminus 1pNpminus 1)cq it means that cge 1
According to (8) we obtain
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max c cqp
1113966 1113967 c (43)
If r (1 + slowastNpminus 1pNpminus 1)cp we know 0lt clt 1 then
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max cpq
c1113966 1113967 c (44)
To sum up we know that |u(k)|le c for all k isin Z(1 N)
Furthermore we have
Ψ(u) 1113944N
k1F(k u(k))le 1113944
N
k1max|ξ|lec
F(k ξ) (45)
for all u isin S with Φ(u)le r Hence
supuisinΦminus1(minusinfinr]Ψ(u)
rle
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ|lec
F(k ξ)max1cq
1cp
1113882 1113883
(46)
6 Discrete Dynamics in Nature and Society
Now let ω(k) d for all k isin Z(1 N) and ω(0)
Δω(N) 0 Clearly ω isin S It is easy to account that Φ(ω)
dppminus 1 + dqqminus 11113957s then
Ψ(ω)
Φ(ω)
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s (47)
Consequently from (46) (47) and assumption (40) wecan obtain
supuisinΦminus 1(minusinfinr]
Ψ(u)
rltΨ(ω)
Φ(ω)
(48)
Moreover because 0lt dlt c and from (40) we obtain
0ltdpp
minus 1+ d
minus 11113957slt
1 + slowastNpminus 1
pNpminus1 min cq c
p1113864 1113865 (49)
that is mean that 0ltΦ(ω)lt r
Hence the problem (Df
λ ) admits at least two positivesolutions by Lemma 1 and Lemma 3 for all λ isin Λ1
Remark 1 If f(k t) is a nonnegative function and there aretwo positive constants c d with dlt c such that
pNpminus 1
1 + slowastNpminus1 max
1113936Nk1 F(k c)
cq1113936
Nk1 F(k c)
cp1113896 1113897
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(50)
then the result of)eorem 2 is also valid for each λ isin Λ2 with
Λ2 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1 mincq
1113936Nk1 F(k c)
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889 (51)
)ere are some consequences of )eorem 2 as follows
Corollary 1 Let g R⟶ [0 +infin) be a continuous functionsuch that f(k t) α(k)g(t) where α(k)gt 0 for allk isin Z(1 N) Put A 1113936
Nk1 α(k) G(t) 1113938
t
0 g(ξ)dξ for allt isin R and Llowastinfin min
kisin[1N]α(k) liminf
t⟶+infin(G(t)tp)gt 0
If there exists cgt dgt 0 such that
pNpminus 1
1 + slowastNpminus1 AG(c)max
1cq
1cp
1113882 1113883
ltminAG(d)
dppminus1 + dqqminus11113957s
qLlowastinfin2pN + 1113957s minus 2pminus11113896 1113897
(52)
then the problem Df
λ has at least two positive solutions foreach λ isin Λ3 with
Λ3 ≔ maxdppminus 1 + dqqminus 11113957s
AG(d)2pN + 1113957s minus 2pminus 1
qLlowastinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1min cq cp
AG(c)1113888 1113889 (53)
Proof Consider the function f Z(1 N) times R⟶ R isgiven as
f(k ξ) α(k)g(ξ) forallk isin Z(1 N) ξ isin R (54)
so that
1113944
N
k1max|ξ|lec
F(k ξ) AG(c) 1113944N
k1F(k d) AG(d) (55)
)en the conclusion can be obtained by)eorem 2
Corollary 2 Assume f be a continuous function withf(k 0)ge 0 and
limsupt⟶0+
F(k t)
tp +infin (56)
limt⟶+infin
F(k t)
tp +infin (57)
for all k isin Z(0 N) Put λlowast (1 + slowastNpminus 1pNpminus 1)min supcgt01113864
(cq1113936Nk1max |ξ∣lecF(k ξ)) supcgt0(cp 1113936
Nk1 max |ξ∣lecF(k ξ))
-en for each λ isin (0 λlowast) the problem (Df
λ ) admits at least twopositive solutions
Proof We know that Linfin +infin from (57) Fix λ isin (0 λlowast)and then there exists cgt 0 such that
λlt1 + slowastN
pminus 1
pNpminus1
middot min supcgt0
cq
1113936Nk1 max|ξ|lecF(k ξ)
supcgt0
cp
1113936Nk1 max|ξ|lecF(k ξ)
⎧⎨
⎩
⎫⎬
⎭
(58)
From (56) we can also obtain
limsupt⟶0+
1113936Nk1 F(k t)
tp +infin (59)
and then there exists d isin (0 c) such that (1113936Nk1 F(k d)
dppminus 1 + dqqminus 11113957s)gt (1λ) )erefore )eorem 2 ensures theconclusion
Discrete Dynamics in Nature and Society 7
Remark 2 If f(k t) is a nonnegative function for all(k t) isin Z(1 N) times [0 +infin) As long as condition (56) holdsfor at least one k isin Z(1 N) then Corollary 2 ensures thatthe solutions are obtained for each
λ isin 01 + slowastN
pminus 1
pNpminus1 min supcgt0
cq
1113936Nk1 F(k c)
supcgt0
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889
(60)
Remark 3 When f(k t) f(t) for all k isin Z(1 N) )e-orem 1 can be ensured by Corollary 2 Obviously condition(4(a)) implies f(0)ge 0 Specially if f is nonnegative weonly need condition (4(a)) to get the corresponding so-lution for each
λ isin 01 + slowastN
pminus 1
pNpmin sup
cgt0
cq
F(c) sup
cgt0
cp
F(c)1113896 11138971113888 1113889 (61)
Example 1 Let p 4 q 2 N 3 s(k) 12 andf(k t)
et
Put
x(c) c2
ec minus 1 (62)
)en
xprime(c) c (2 minus c)ec minus 2( )
ec minus 1( )2 (63)
Let z(c) (2 minus c)ec minus 2 then zprime(c) (1 minus c)ec So z(c)
is increasing in c isin (0 1) and decreasing in c isin (1 +infin)Since z(0) 0 and z(+infin) minusinfin there exists an uniquec1 isin (1 +infin) such that z(c1) 0 )us x(c) in increasing inc isin (0 c1) and decreasing in c isin (c1 +infin) )is means thatsupcgt0 x(c) x(c1) In fact c1 asymp 15936
Similarly put y(c) c4ec minus 1 we can show that thereexists a unique c2 isin (3 +infin) such that supcgt0y(c) y(c2)
In fact c2 asymp 39207
Since
y c2( 1113857 c42
ec2 minus 1gt
c41ec1 minus 1
c21x c1( 1113857gt x c1( 1113857 (64)
then
1 + slowastNpminus 1
pNpmin sup
cgt0x(c) sup
cgt0y(c)1113896 1113897
1 + slowastN
pminus 1
pNpx c1( 1113857 asymp 06496
(65)
)erefore for each λ isin (0 06496) the problem
minusΔ ϕ4(Δu(k minus 1))( 1113857 + 12ϕ2(u(k)) λeu(k) k 1 2 3
u(0) Δu(3) 0
⎧⎨
⎩
(66)
admits at least two positive solutions
Example 2 Let N 3 p 3 and q 2 and f be a functionas follows
f(k t)
0 if tlt 0kt
3radic
if 0le tle 1kt
3radic
+ 225t2 minus 225 if tgt 1
⎧⎪⎪⎨
⎪⎪⎩(67)
From Remark 1 we can choose c 1 and d 002 Easycalculation shows that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1F(k c)max
1cq
1cp
1113882 1113883 pNpminus 1
1 + slowastNpminus1 1113944
3
k111139461
0
kt
3radic
dt
asymp 13693
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s
11139363k1 1113946
002
0
kt
3radic
dt
dppminus1 + dqqminus11113957sasymp 31386
qLinfin2pN + 1113957s minus 2pminus1
166
limt⟶+infin
kt
3radic
+ 225t2 minus 225t2
asymp 34091
(68)
which satisfy condition (50) )us for each λ isin(03186 07303) the problem
minusΔ ϕ3(Δu(k minus 1))( 1113857 + 8ϕ2(u(k)) λf(k u(k)) k 1 2 3
u(0) Δu(3) 01113896
(69)
admits at least two positive solutions
Data Availability
No data were used to support the study
Conflicts of Interest
)e authors declare that there are no conflicts of interestregarding the publication of this paper
Authorsrsquo Contributions
All authors contributed equally to the writing of this paperAll authors read and approved the final manuscript
Acknowledgments
)is work was supported by the National Natural ScienceFoundation of China (Grant no 11971126) and Program forChangjiang Scholars and Innovative Research Team inUniversity (Grant no IRTminus16R16)
References
[1] R P Agarwal Difference Equations and Inequalities -eoryMethods and Applications Marcel Dekker Inc New YorkNY USA 2000
[2] A Kristaly M Mihailescu V Radulescu and S TersianldquoSpectral estimates for a nonhomogeneous difference
8 Discrete Dynamics in Nature and Society
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9
for all u v isin S So we have
Jλprime(u)(v) 1113944N
k1minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) minus λf(k u(k))1113960 1113961v(k) (20)
Hence a critical point u of Jλ is a solution of problem(D
f
λ )Now we recall a definition and a two nonzero critical
points theorem for the readerrsquos convenience
Definition 1 Let X be a real Banach space we say that aGateaux differentiable function Jλ X⟶ R satisfies the(PS)-condition if any sequence un1113864 1113865nisinNsubeX such that
(i) Jλ(un)⟶ c isin R as n⟶ +infin(ii) Jλprime(un)⟶ 0 as n⟶ +infin has a convergent
subsequence
Lemma 1 Let X be a real Banach space and ΦΨ isin C1(SR)
such that infX(Φ) Φ(0) Ψ(0) 0 Assume that there arer isin R and ω isin X with 0ltΦ(ω)lt r such that
supuisinΦminus1(minusinfinr]Ψ(u)
rltΨ(ω)
Φ(ω) (21)
and for each
λ isin Λ Φ(ω)
Ψ(ω)
r
supuisinΦminus1(minusinfinr]Ψ(u)1113888 1113889 (22)
the functional Jλ Φ minus λΨ satisfies the (PS)-condition and itis unbounded from below
)en for each λ isin Λ the functional Jλ admits at least twononzero critical points uλ1 uλ2 such that Jλ(uλ1)lt 0lt Jλ(uλ2)
In order to obtain the positive solution of problem (Df
λ )we establish the following strong maximum principle
Lemma 2 Fix u isin S such that either
u(k)gt 0 or minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k))ge 0
(23)
for all k isin Z(1 N) -en either ugt 0 in Z(1 N) or u equiv 0
Proof Let j isin Z(1 N) such that
u(j) min u(k) k isin Z(1 N) (24)
If u(j)gt 0 then it is easy know that ugt 0 in Z(1 N)If u(j)le 0 then by (23) we have
minusΔ ϕp(Δu(j minus 1))1113872 1113873ge minuss(j)ϕq(u(j))ge 0 (25)
that is
ϕp(Δu(j))leϕp(Δu(j minus 1)) (26)
Since ϕp(u) is increasing in u we have
Δu(j)leΔu(j minus 1) (27)
By the definition of u(j) we know that
Δu(j)ge 0
Δu(j minus 1)le 0(28)
By combining (27) with (28) we get u(j + 1) u(j)
u(j minus 1) If j minus 1 0 we have u(j) u(j minus 1) 0 Other-wise j minus 1 isin Z(1 N) replacing j minus 1 by j we knowu(j minus 2) u(j minus 1) Continuing in this way we haveu(j) u(j minus 1) middot middot middot u(0) 0 Similarly we haveu(j) u(j + 1) middot middot middot u(N + 1) )us u(k) u(0) 0and forallk isin Z(1 N)
Now put
F+(k t) 1113946
t
0f k ξ+
( 1113857dξ forall(k t) isin Z(1 N) times R (29)
where ξ+ max ξ 0
Define J+λ Φ1 +Φ2 minus λΨ+ and
Ψ+(u) 1113936Nk1 F+(k u(k)) Standard arguments show that
J+λ isin C1(S R) and the critical points of J+
λ are precisely thesolutions of the following problem
minusΔ ϕp(Δu(k minus 1))1113872 1113873 + s(k)ϕq(u(k)) λf k u+(k)( ) k isin Z(1 N)
u(0) Δu(N) 0
⎧⎨
⎩ Df+
λ1113872 1113873 (30)
Lemma 3 If f(k 0)ge 0 for each k isin Z(1 N) any nonzerocritical point of the functional J+
λ is a positive solution ofproblem (D
f
λ )
Proof Since a critical point of J+λ is a solution of problem
Df+
λ the conclusion follows by the discrete maximumprinciple ([33] Proposition 1)
4 Discrete Dynamics in Nature and Society
Next we suppose that f(k 0)ge 0 and f(k x) f(k 0)
for all xle 0 and for all k isin Z(1 N) Put
Linfin ≔ minkisinZ(1N)
liminft⟶+infin
F(k t)
tp
1113957s 1113944
N
k1s(k)
(31)
we have the following result
Lemma 4 If Linfin gt 0 then Jλ satisfies (PS)-condition and it isunbounded from below for all λ isin (2pN + 1113957s minus 2pminus1qLinfin +infin)
Proof Let λgt 2pN + 1113957s minus 2pminus 1qLinfin We consider a sequenceun1113864 1113865nisinNsube S such that Jλ(un)⟶ c isin R and Jλprime(un)⟶ 0 as
n⟶ +infin Let u+n max un 01113864 1113865 and uminus
n max minusun 01113864 1113865 forall n isin N We first prove that uminus
n1113864 1113865 is bounded On one handwe have
Δuminusn (k minus 1)
11138681113868111386811138681113868111386811138681113868p le minus ϕp Δun(k minus 1)( 1113857Δuminus
n (k minus 1)
s(k)∣uminusn (k)∣q minuss(k)∣un(k)∣qminus2
un(k)uminusn (k)
(32)
for all k isin Z(1 N) So
uminusn
p 1113944
N+1
k1Δuminus
n (k minus 1)1113868111386811138681113868
1113868111386811138681113868p
le minus 1113944N+1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k) + 1113944
N1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k minus 1)
minus 1113944N
k1ϕp Δun(k minus 1)( 1113857u
minusn(k) + 1113944
N
k1ϕp Δun(k)( 1113857u
minusn(k)
1113944N
k1Δϕp Δun(k minus 1)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
1113944
N
k1s(k) u
minusn(k)
11138681113868111386811138681113868111386811138681113868q
minus 1113944N
k1s(k)ϕq un(k)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
(33)
On the other hand we assume that
f(k u) f(k u) if ugt 0
f(k 0) if ule 01113896 (34)
for each k isin Z(1 N) then
Ψprime un( 1113857 uminusn( 1113857 1113944
N
k1f k un(k)( 1113857u
minusn (k)ge 0 (35)
)erefore
uminusn
p le uminusn
p+ 1113944
N
k1s(k)|u(k)|
q
le minusΦ1prime un( 1113857 uminusn( 1113857 minusΦ2prime un( 1113857 u
minusn( 1113857 + λΨprime un( 1113857 u
minusn( 1113857
minusJλprime un( 1113857 uminusn( 1113857
(36)
for all n isin N which leads to uminusn pminus 1⟶ 0 as n⟶ +infin
So we have uminusn ⟶ 0 as n⟶ +infin It means that there
exists an Mgt 0 such that uminusn leM From (10) we know that
uminusn infin leN1minus 1pM c for all k isin Z(1 N)
Next we suppose that the sequence un1113864 1113865 is unboundedthat is u+
n1113864 1113865 is unboundedAs Linfin gt 0 we know that there exists an l isin R such that
Linfin gt lgt 2pN + 1113957s minus 2pminus 1λq From the definition of Linfin thereis δk gt 0 such that F(k t)gt l|t|p for all tgt δk Furthermoresince F(k t) is a continuous function there exists a constantC(k)ge 0 such that F(k t)ge l|t|p minus C(k) with t isin [minusc δk]
)us F(k t)ge l|t|p minus C(k) for all sge minus c and k isin Z(1 N)
We can obtain that
1113944
N
k1F k un(k)( 1113857ge 1113944
N
k1l un(k)1113868111386811138681113868
1113868111386811138681113868p
minus Cge l un
p
infin minus C (37)
for all k isin Z(1 N) where C 1113936Nk1 C(k) that is
Ψ un( 1113857ge l un
p
infin minus C (38)
Hence for all un such that uninfinge 1 we conclude that
Discrete Dynamics in Nature and Society 5
Jλ un( 1113857 1113936
N+1k1 Δun(k minus 1)
11138681113868111386811138681113868111386811138681113868p
p+
1113936Nk1s(k) un(k)
11138681113868111386811138681113868111386811138681113868q
qminus λΨ un( 1113857
le2pminus 1
p1113944
N
k1un(k)
11138681113868111386811138681113868111386811138681113868p
+ 1113944N
k1un(k minus 1)
11138681113868111386811138681113868111386811138681113868p⎛⎝ ⎞⎠
+1q
1113944
N
k1s(k)∣un(k) ∣q minus λΨ un( 1113857
le2pminus 1(2N minus 1)
pun
p
infin +1113957s
qun
q
infin minus λΨ un( 1113857
le2pN + 1113957s minus 2pminus 1
qminus λl1113888 1113889 un
p
infin + λC
(39)
Since 2pN + 1113957s minus 2pminus 1q minus λllt 0 we can getlimn⟶+infin Jλ(un) minusinfin and this is absurd Hence Jλ sat-isfies (PS)-condition
Let un1113864 1113865 be such that uminusn1113864 1113865 is bounded and u+
n1113864 1113865 is un-bounded From the proof above we can see that Jλ is un-bounded from below
3 Main Results
)e main results of this paper are as follows
Theorem 2 Let f Z(1 N) times R⟶ R be a continuousfunction satisfying f(k 0)ge 0 for all k isin Z(1 N) If there aretwo constants c and d with dlt c such that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ∣ lec
F(k ξ)max1cq
1cp
1113882 1113883
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(40)
)en for each λ isin Λ1 with
Λ1 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1pNpminus 1( 1113857min cq cp
1113936Nk1 max |ξ∣lecF(k ξ)
⎛⎝ ⎞⎠ (41)
the problem (Df
λ ) admits at least two positive solutions
Proof Put ΦΨ as in (18) It is clear that infX(Φ) Φ(0)
Ψ(0) 0 According to Lemma 3 we know that a nonzerocritical point in S of the functional J+
λ is precisely a positivesolution of problem (D
f
λ ) Next we just need to provecondition (21) of Lemma 1
We observe that Linfin gt 0 from (40) and Λ1 is nonde-generate Fix λ isin Λ1 Lemma 4 ensures that Jλ satisfies
(PS)-condition for all λgt 2pN + 1113957s minus 2pminus 1qLinfin and it isunbounded from below We let u isin Φminus 1(minusinfin r] that is(upp) + (1113936
Nk1 s(k)|u(k)|qq)le r Put
r 1 + slowastN
pminus 1
pNpminus1 min cq c
p1113864 1113865 (42)
If r (1 + slowastNpminus 1pNpminus 1)cq it means that cge 1
According to (8) we obtain
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max c cqp
1113966 1113967 c (43)
If r (1 + slowastNpminus 1pNpminus 1)cp we know 0lt clt 1 then
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max cpq
c1113966 1113967 c (44)
To sum up we know that |u(k)|le c for all k isin Z(1 N)
Furthermore we have
Ψ(u) 1113944N
k1F(k u(k))le 1113944
N
k1max|ξ|lec
F(k ξ) (45)
for all u isin S with Φ(u)le r Hence
supuisinΦminus1(minusinfinr]Ψ(u)
rle
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ|lec
F(k ξ)max1cq
1cp
1113882 1113883
(46)
6 Discrete Dynamics in Nature and Society
Now let ω(k) d for all k isin Z(1 N) and ω(0)
Δω(N) 0 Clearly ω isin S It is easy to account that Φ(ω)
dppminus 1 + dqqminus 11113957s then
Ψ(ω)
Φ(ω)
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s (47)
Consequently from (46) (47) and assumption (40) wecan obtain
supuisinΦminus 1(minusinfinr]
Ψ(u)
rltΨ(ω)
Φ(ω)
(48)
Moreover because 0lt dlt c and from (40) we obtain
0ltdpp
minus 1+ d
minus 11113957slt
1 + slowastNpminus 1
pNpminus1 min cq c
p1113864 1113865 (49)
that is mean that 0ltΦ(ω)lt r
Hence the problem (Df
λ ) admits at least two positivesolutions by Lemma 1 and Lemma 3 for all λ isin Λ1
Remark 1 If f(k t) is a nonnegative function and there aretwo positive constants c d with dlt c such that
pNpminus 1
1 + slowastNpminus1 max
1113936Nk1 F(k c)
cq1113936
Nk1 F(k c)
cp1113896 1113897
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(50)
then the result of)eorem 2 is also valid for each λ isin Λ2 with
Λ2 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1 mincq
1113936Nk1 F(k c)
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889 (51)
)ere are some consequences of )eorem 2 as follows
Corollary 1 Let g R⟶ [0 +infin) be a continuous functionsuch that f(k t) α(k)g(t) where α(k)gt 0 for allk isin Z(1 N) Put A 1113936
Nk1 α(k) G(t) 1113938
t
0 g(ξ)dξ for allt isin R and Llowastinfin min
kisin[1N]α(k) liminf
t⟶+infin(G(t)tp)gt 0
If there exists cgt dgt 0 such that
pNpminus 1
1 + slowastNpminus1 AG(c)max
1cq
1cp
1113882 1113883
ltminAG(d)
dppminus1 + dqqminus11113957s
qLlowastinfin2pN + 1113957s minus 2pminus11113896 1113897
(52)
then the problem Df
λ has at least two positive solutions foreach λ isin Λ3 with
Λ3 ≔ maxdppminus 1 + dqqminus 11113957s
AG(d)2pN + 1113957s minus 2pminus 1
qLlowastinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1min cq cp
AG(c)1113888 1113889 (53)
Proof Consider the function f Z(1 N) times R⟶ R isgiven as
f(k ξ) α(k)g(ξ) forallk isin Z(1 N) ξ isin R (54)
so that
1113944
N
k1max|ξ|lec
F(k ξ) AG(c) 1113944N
k1F(k d) AG(d) (55)
)en the conclusion can be obtained by)eorem 2
Corollary 2 Assume f be a continuous function withf(k 0)ge 0 and
limsupt⟶0+
F(k t)
tp +infin (56)
limt⟶+infin
F(k t)
tp +infin (57)
for all k isin Z(0 N) Put λlowast (1 + slowastNpminus 1pNpminus 1)min supcgt01113864
(cq1113936Nk1max |ξ∣lecF(k ξ)) supcgt0(cp 1113936
Nk1 max |ξ∣lecF(k ξ))
-en for each λ isin (0 λlowast) the problem (Df
λ ) admits at least twopositive solutions
Proof We know that Linfin +infin from (57) Fix λ isin (0 λlowast)and then there exists cgt 0 such that
λlt1 + slowastN
pminus 1
pNpminus1
middot min supcgt0
cq
1113936Nk1 max|ξ|lecF(k ξ)
supcgt0
cp
1113936Nk1 max|ξ|lecF(k ξ)
⎧⎨
⎩
⎫⎬
⎭
(58)
From (56) we can also obtain
limsupt⟶0+
1113936Nk1 F(k t)
tp +infin (59)
and then there exists d isin (0 c) such that (1113936Nk1 F(k d)
dppminus 1 + dqqminus 11113957s)gt (1λ) )erefore )eorem 2 ensures theconclusion
Discrete Dynamics in Nature and Society 7
Remark 2 If f(k t) is a nonnegative function for all(k t) isin Z(1 N) times [0 +infin) As long as condition (56) holdsfor at least one k isin Z(1 N) then Corollary 2 ensures thatthe solutions are obtained for each
λ isin 01 + slowastN
pminus 1
pNpminus1 min supcgt0
cq
1113936Nk1 F(k c)
supcgt0
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889
(60)
Remark 3 When f(k t) f(t) for all k isin Z(1 N) )e-orem 1 can be ensured by Corollary 2 Obviously condition(4(a)) implies f(0)ge 0 Specially if f is nonnegative weonly need condition (4(a)) to get the corresponding so-lution for each
λ isin 01 + slowastN
pminus 1
pNpmin sup
cgt0
cq
F(c) sup
cgt0
cp
F(c)1113896 11138971113888 1113889 (61)
Example 1 Let p 4 q 2 N 3 s(k) 12 andf(k t)
et
Put
x(c) c2
ec minus 1 (62)
)en
xprime(c) c (2 minus c)ec minus 2( )
ec minus 1( )2 (63)
Let z(c) (2 minus c)ec minus 2 then zprime(c) (1 minus c)ec So z(c)
is increasing in c isin (0 1) and decreasing in c isin (1 +infin)Since z(0) 0 and z(+infin) minusinfin there exists an uniquec1 isin (1 +infin) such that z(c1) 0 )us x(c) in increasing inc isin (0 c1) and decreasing in c isin (c1 +infin) )is means thatsupcgt0 x(c) x(c1) In fact c1 asymp 15936
Similarly put y(c) c4ec minus 1 we can show that thereexists a unique c2 isin (3 +infin) such that supcgt0y(c) y(c2)
In fact c2 asymp 39207
Since
y c2( 1113857 c42
ec2 minus 1gt
c41ec1 minus 1
c21x c1( 1113857gt x c1( 1113857 (64)
then
1 + slowastNpminus 1
pNpmin sup
cgt0x(c) sup
cgt0y(c)1113896 1113897
1 + slowastN
pminus 1
pNpx c1( 1113857 asymp 06496
(65)
)erefore for each λ isin (0 06496) the problem
minusΔ ϕ4(Δu(k minus 1))( 1113857 + 12ϕ2(u(k)) λeu(k) k 1 2 3
u(0) Δu(3) 0
⎧⎨
⎩
(66)
admits at least two positive solutions
Example 2 Let N 3 p 3 and q 2 and f be a functionas follows
f(k t)
0 if tlt 0kt
3radic
if 0le tle 1kt
3radic
+ 225t2 minus 225 if tgt 1
⎧⎪⎪⎨
⎪⎪⎩(67)
From Remark 1 we can choose c 1 and d 002 Easycalculation shows that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1F(k c)max
1cq
1cp
1113882 1113883 pNpminus 1
1 + slowastNpminus1 1113944
3
k111139461
0
kt
3radic
dt
asymp 13693
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s
11139363k1 1113946
002
0
kt
3radic
dt
dppminus1 + dqqminus11113957sasymp 31386
qLinfin2pN + 1113957s minus 2pminus1
166
limt⟶+infin
kt
3radic
+ 225t2 minus 225t2
asymp 34091
(68)
which satisfy condition (50) )us for each λ isin(03186 07303) the problem
minusΔ ϕ3(Δu(k minus 1))( 1113857 + 8ϕ2(u(k)) λf(k u(k)) k 1 2 3
u(0) Δu(3) 01113896
(69)
admits at least two positive solutions
Data Availability
No data were used to support the study
Conflicts of Interest
)e authors declare that there are no conflicts of interestregarding the publication of this paper
Authorsrsquo Contributions
All authors contributed equally to the writing of this paperAll authors read and approved the final manuscript
Acknowledgments
)is work was supported by the National Natural ScienceFoundation of China (Grant no 11971126) and Program forChangjiang Scholars and Innovative Research Team inUniversity (Grant no IRTminus16R16)
References
[1] R P Agarwal Difference Equations and Inequalities -eoryMethods and Applications Marcel Dekker Inc New YorkNY USA 2000
[2] A Kristaly M Mihailescu V Radulescu and S TersianldquoSpectral estimates for a nonhomogeneous difference
8 Discrete Dynamics in Nature and Society
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9
Next we suppose that f(k 0)ge 0 and f(k x) f(k 0)
for all xle 0 and for all k isin Z(1 N) Put
Linfin ≔ minkisinZ(1N)
liminft⟶+infin
F(k t)
tp
1113957s 1113944
N
k1s(k)
(31)
we have the following result
Lemma 4 If Linfin gt 0 then Jλ satisfies (PS)-condition and it isunbounded from below for all λ isin (2pN + 1113957s minus 2pminus1qLinfin +infin)
Proof Let λgt 2pN + 1113957s minus 2pminus 1qLinfin We consider a sequenceun1113864 1113865nisinNsube S such that Jλ(un)⟶ c isin R and Jλprime(un)⟶ 0 as
n⟶ +infin Let u+n max un 01113864 1113865 and uminus
n max minusun 01113864 1113865 forall n isin N We first prove that uminus
n1113864 1113865 is bounded On one handwe have
Δuminusn (k minus 1)
11138681113868111386811138681113868111386811138681113868p le minus ϕp Δun(k minus 1)( 1113857Δuminus
n (k minus 1)
s(k)∣uminusn (k)∣q minuss(k)∣un(k)∣qminus2
un(k)uminusn (k)
(32)
for all k isin Z(1 N) So
uminusn
p 1113944
N+1
k1Δuminus
n (k minus 1)1113868111386811138681113868
1113868111386811138681113868p
le minus 1113944N+1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k) + 1113944
N1
k1ϕp Δun(k minus 1)( 1113857u
minusn (k minus 1)
minus 1113944N
k1ϕp Δun(k minus 1)( 1113857u
minusn(k) + 1113944
N
k1ϕp Δun(k)( 1113857u
minusn(k)
1113944N
k1Δϕp Δun(k minus 1)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
1113944
N
k1s(k) u
minusn(k)
11138681113868111386811138681113868111386811138681113868q
minus 1113944N
k1s(k)ϕq un(k)( 1113857u
minusn (k) minusΦ1prime un( 1113857 u
minusn( 1113857
(33)
On the other hand we assume that
f(k u) f(k u) if ugt 0
f(k 0) if ule 01113896 (34)
for each k isin Z(1 N) then
Ψprime un( 1113857 uminusn( 1113857 1113944
N
k1f k un(k)( 1113857u
minusn (k)ge 0 (35)
)erefore
uminusn
p le uminusn
p+ 1113944
N
k1s(k)|u(k)|
q
le minusΦ1prime un( 1113857 uminusn( 1113857 minusΦ2prime un( 1113857 u
minusn( 1113857 + λΨprime un( 1113857 u
minusn( 1113857
minusJλprime un( 1113857 uminusn( 1113857
(36)
for all n isin N which leads to uminusn pminus 1⟶ 0 as n⟶ +infin
So we have uminusn ⟶ 0 as n⟶ +infin It means that there
exists an Mgt 0 such that uminusn leM From (10) we know that
uminusn infin leN1minus 1pM c for all k isin Z(1 N)
Next we suppose that the sequence un1113864 1113865 is unboundedthat is u+
n1113864 1113865 is unboundedAs Linfin gt 0 we know that there exists an l isin R such that
Linfin gt lgt 2pN + 1113957s minus 2pminus 1λq From the definition of Linfin thereis δk gt 0 such that F(k t)gt l|t|p for all tgt δk Furthermoresince F(k t) is a continuous function there exists a constantC(k)ge 0 such that F(k t)ge l|t|p minus C(k) with t isin [minusc δk]
)us F(k t)ge l|t|p minus C(k) for all sge minus c and k isin Z(1 N)
We can obtain that
1113944
N
k1F k un(k)( 1113857ge 1113944
N
k1l un(k)1113868111386811138681113868
1113868111386811138681113868p
minus Cge l un
p
infin minus C (37)
for all k isin Z(1 N) where C 1113936Nk1 C(k) that is
Ψ un( 1113857ge l un
p
infin minus C (38)
Hence for all un such that uninfinge 1 we conclude that
Discrete Dynamics in Nature and Society 5
Jλ un( 1113857 1113936
N+1k1 Δun(k minus 1)
11138681113868111386811138681113868111386811138681113868p
p+
1113936Nk1s(k) un(k)
11138681113868111386811138681113868111386811138681113868q
qminus λΨ un( 1113857
le2pminus 1
p1113944
N
k1un(k)
11138681113868111386811138681113868111386811138681113868p
+ 1113944N
k1un(k minus 1)
11138681113868111386811138681113868111386811138681113868p⎛⎝ ⎞⎠
+1q
1113944
N
k1s(k)∣un(k) ∣q minus λΨ un( 1113857
le2pminus 1(2N minus 1)
pun
p
infin +1113957s
qun
q
infin minus λΨ un( 1113857
le2pN + 1113957s minus 2pminus 1
qminus λl1113888 1113889 un
p
infin + λC
(39)
Since 2pN + 1113957s minus 2pminus 1q minus λllt 0 we can getlimn⟶+infin Jλ(un) minusinfin and this is absurd Hence Jλ sat-isfies (PS)-condition
Let un1113864 1113865 be such that uminusn1113864 1113865 is bounded and u+
n1113864 1113865 is un-bounded From the proof above we can see that Jλ is un-bounded from below
3 Main Results
)e main results of this paper are as follows
Theorem 2 Let f Z(1 N) times R⟶ R be a continuousfunction satisfying f(k 0)ge 0 for all k isin Z(1 N) If there aretwo constants c and d with dlt c such that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ∣ lec
F(k ξ)max1cq
1cp
1113882 1113883
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(40)
)en for each λ isin Λ1 with
Λ1 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1pNpminus 1( 1113857min cq cp
1113936Nk1 max |ξ∣lecF(k ξ)
⎛⎝ ⎞⎠ (41)
the problem (Df
λ ) admits at least two positive solutions
Proof Put ΦΨ as in (18) It is clear that infX(Φ) Φ(0)
Ψ(0) 0 According to Lemma 3 we know that a nonzerocritical point in S of the functional J+
λ is precisely a positivesolution of problem (D
f
λ ) Next we just need to provecondition (21) of Lemma 1
We observe that Linfin gt 0 from (40) and Λ1 is nonde-generate Fix λ isin Λ1 Lemma 4 ensures that Jλ satisfies
(PS)-condition for all λgt 2pN + 1113957s minus 2pminus 1qLinfin and it isunbounded from below We let u isin Φminus 1(minusinfin r] that is(upp) + (1113936
Nk1 s(k)|u(k)|qq)le r Put
r 1 + slowastN
pminus 1
pNpminus1 min cq c
p1113864 1113865 (42)
If r (1 + slowastNpminus 1pNpminus 1)cq it means that cge 1
According to (8) we obtain
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max c cqp
1113966 1113967 c (43)
If r (1 + slowastNpminus 1pNpminus 1)cp we know 0lt clt 1 then
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max cpq
c1113966 1113967 c (44)
To sum up we know that |u(k)|le c for all k isin Z(1 N)
Furthermore we have
Ψ(u) 1113944N
k1F(k u(k))le 1113944
N
k1max|ξ|lec
F(k ξ) (45)
for all u isin S with Φ(u)le r Hence
supuisinΦminus1(minusinfinr]Ψ(u)
rle
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ|lec
F(k ξ)max1cq
1cp
1113882 1113883
(46)
6 Discrete Dynamics in Nature and Society
Now let ω(k) d for all k isin Z(1 N) and ω(0)
Δω(N) 0 Clearly ω isin S It is easy to account that Φ(ω)
dppminus 1 + dqqminus 11113957s then
Ψ(ω)
Φ(ω)
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s (47)
Consequently from (46) (47) and assumption (40) wecan obtain
supuisinΦminus 1(minusinfinr]
Ψ(u)
rltΨ(ω)
Φ(ω)
(48)
Moreover because 0lt dlt c and from (40) we obtain
0ltdpp
minus 1+ d
minus 11113957slt
1 + slowastNpminus 1
pNpminus1 min cq c
p1113864 1113865 (49)
that is mean that 0ltΦ(ω)lt r
Hence the problem (Df
λ ) admits at least two positivesolutions by Lemma 1 and Lemma 3 for all λ isin Λ1
Remark 1 If f(k t) is a nonnegative function and there aretwo positive constants c d with dlt c such that
pNpminus 1
1 + slowastNpminus1 max
1113936Nk1 F(k c)
cq1113936
Nk1 F(k c)
cp1113896 1113897
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(50)
then the result of)eorem 2 is also valid for each λ isin Λ2 with
Λ2 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1 mincq
1113936Nk1 F(k c)
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889 (51)
)ere are some consequences of )eorem 2 as follows
Corollary 1 Let g R⟶ [0 +infin) be a continuous functionsuch that f(k t) α(k)g(t) where α(k)gt 0 for allk isin Z(1 N) Put A 1113936
Nk1 α(k) G(t) 1113938
t
0 g(ξ)dξ for allt isin R and Llowastinfin min
kisin[1N]α(k) liminf
t⟶+infin(G(t)tp)gt 0
If there exists cgt dgt 0 such that
pNpminus 1
1 + slowastNpminus1 AG(c)max
1cq
1cp
1113882 1113883
ltminAG(d)
dppminus1 + dqqminus11113957s
qLlowastinfin2pN + 1113957s minus 2pminus11113896 1113897
(52)
then the problem Df
λ has at least two positive solutions foreach λ isin Λ3 with
Λ3 ≔ maxdppminus 1 + dqqminus 11113957s
AG(d)2pN + 1113957s minus 2pminus 1
qLlowastinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1min cq cp
AG(c)1113888 1113889 (53)
Proof Consider the function f Z(1 N) times R⟶ R isgiven as
f(k ξ) α(k)g(ξ) forallk isin Z(1 N) ξ isin R (54)
so that
1113944
N
k1max|ξ|lec
F(k ξ) AG(c) 1113944N
k1F(k d) AG(d) (55)
)en the conclusion can be obtained by)eorem 2
Corollary 2 Assume f be a continuous function withf(k 0)ge 0 and
limsupt⟶0+
F(k t)
tp +infin (56)
limt⟶+infin
F(k t)
tp +infin (57)
for all k isin Z(0 N) Put λlowast (1 + slowastNpminus 1pNpminus 1)min supcgt01113864
(cq1113936Nk1max |ξ∣lecF(k ξ)) supcgt0(cp 1113936
Nk1 max |ξ∣lecF(k ξ))
-en for each λ isin (0 λlowast) the problem (Df
λ ) admits at least twopositive solutions
Proof We know that Linfin +infin from (57) Fix λ isin (0 λlowast)and then there exists cgt 0 such that
λlt1 + slowastN
pminus 1
pNpminus1
middot min supcgt0
cq
1113936Nk1 max|ξ|lecF(k ξ)
supcgt0
cp
1113936Nk1 max|ξ|lecF(k ξ)
⎧⎨
⎩
⎫⎬
⎭
(58)
From (56) we can also obtain
limsupt⟶0+
1113936Nk1 F(k t)
tp +infin (59)
and then there exists d isin (0 c) such that (1113936Nk1 F(k d)
dppminus 1 + dqqminus 11113957s)gt (1λ) )erefore )eorem 2 ensures theconclusion
Discrete Dynamics in Nature and Society 7
Remark 2 If f(k t) is a nonnegative function for all(k t) isin Z(1 N) times [0 +infin) As long as condition (56) holdsfor at least one k isin Z(1 N) then Corollary 2 ensures thatthe solutions are obtained for each
λ isin 01 + slowastN
pminus 1
pNpminus1 min supcgt0
cq
1113936Nk1 F(k c)
supcgt0
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889
(60)
Remark 3 When f(k t) f(t) for all k isin Z(1 N) )e-orem 1 can be ensured by Corollary 2 Obviously condition(4(a)) implies f(0)ge 0 Specially if f is nonnegative weonly need condition (4(a)) to get the corresponding so-lution for each
λ isin 01 + slowastN
pminus 1
pNpmin sup
cgt0
cq
F(c) sup
cgt0
cp
F(c)1113896 11138971113888 1113889 (61)
Example 1 Let p 4 q 2 N 3 s(k) 12 andf(k t)
et
Put
x(c) c2
ec minus 1 (62)
)en
xprime(c) c (2 minus c)ec minus 2( )
ec minus 1( )2 (63)
Let z(c) (2 minus c)ec minus 2 then zprime(c) (1 minus c)ec So z(c)
is increasing in c isin (0 1) and decreasing in c isin (1 +infin)Since z(0) 0 and z(+infin) minusinfin there exists an uniquec1 isin (1 +infin) such that z(c1) 0 )us x(c) in increasing inc isin (0 c1) and decreasing in c isin (c1 +infin) )is means thatsupcgt0 x(c) x(c1) In fact c1 asymp 15936
Similarly put y(c) c4ec minus 1 we can show that thereexists a unique c2 isin (3 +infin) such that supcgt0y(c) y(c2)
In fact c2 asymp 39207
Since
y c2( 1113857 c42
ec2 minus 1gt
c41ec1 minus 1
c21x c1( 1113857gt x c1( 1113857 (64)
then
1 + slowastNpminus 1
pNpmin sup
cgt0x(c) sup
cgt0y(c)1113896 1113897
1 + slowastN
pminus 1
pNpx c1( 1113857 asymp 06496
(65)
)erefore for each λ isin (0 06496) the problem
minusΔ ϕ4(Δu(k minus 1))( 1113857 + 12ϕ2(u(k)) λeu(k) k 1 2 3
u(0) Δu(3) 0
⎧⎨
⎩
(66)
admits at least two positive solutions
Example 2 Let N 3 p 3 and q 2 and f be a functionas follows
f(k t)
0 if tlt 0kt
3radic
if 0le tle 1kt
3radic
+ 225t2 minus 225 if tgt 1
⎧⎪⎪⎨
⎪⎪⎩(67)
From Remark 1 we can choose c 1 and d 002 Easycalculation shows that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1F(k c)max
1cq
1cp
1113882 1113883 pNpminus 1
1 + slowastNpminus1 1113944
3
k111139461
0
kt
3radic
dt
asymp 13693
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s
11139363k1 1113946
002
0
kt
3radic
dt
dppminus1 + dqqminus11113957sasymp 31386
qLinfin2pN + 1113957s minus 2pminus1
166
limt⟶+infin
kt
3radic
+ 225t2 minus 225t2
asymp 34091
(68)
which satisfy condition (50) )us for each λ isin(03186 07303) the problem
minusΔ ϕ3(Δu(k minus 1))( 1113857 + 8ϕ2(u(k)) λf(k u(k)) k 1 2 3
u(0) Δu(3) 01113896
(69)
admits at least two positive solutions
Data Availability
No data were used to support the study
Conflicts of Interest
)e authors declare that there are no conflicts of interestregarding the publication of this paper
Authorsrsquo Contributions
All authors contributed equally to the writing of this paperAll authors read and approved the final manuscript
Acknowledgments
)is work was supported by the National Natural ScienceFoundation of China (Grant no 11971126) and Program forChangjiang Scholars and Innovative Research Team inUniversity (Grant no IRTminus16R16)
References
[1] R P Agarwal Difference Equations and Inequalities -eoryMethods and Applications Marcel Dekker Inc New YorkNY USA 2000
[2] A Kristaly M Mihailescu V Radulescu and S TersianldquoSpectral estimates for a nonhomogeneous difference
8 Discrete Dynamics in Nature and Society
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9
Jλ un( 1113857 1113936
N+1k1 Δun(k minus 1)
11138681113868111386811138681113868111386811138681113868p
p+
1113936Nk1s(k) un(k)
11138681113868111386811138681113868111386811138681113868q
qminus λΨ un( 1113857
le2pminus 1
p1113944
N
k1un(k)
11138681113868111386811138681113868111386811138681113868p
+ 1113944N
k1un(k minus 1)
11138681113868111386811138681113868111386811138681113868p⎛⎝ ⎞⎠
+1q
1113944
N
k1s(k)∣un(k) ∣q minus λΨ un( 1113857
le2pminus 1(2N minus 1)
pun
p
infin +1113957s
qun
q
infin minus λΨ un( 1113857
le2pN + 1113957s minus 2pminus 1
qminus λl1113888 1113889 un
p
infin + λC
(39)
Since 2pN + 1113957s minus 2pminus 1q minus λllt 0 we can getlimn⟶+infin Jλ(un) minusinfin and this is absurd Hence Jλ sat-isfies (PS)-condition
Let un1113864 1113865 be such that uminusn1113864 1113865 is bounded and u+
n1113864 1113865 is un-bounded From the proof above we can see that Jλ is un-bounded from below
3 Main Results
)e main results of this paper are as follows
Theorem 2 Let f Z(1 N) times R⟶ R be a continuousfunction satisfying f(k 0)ge 0 for all k isin Z(1 N) If there aretwo constants c and d with dlt c such that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ∣ lec
F(k ξ)max1cq
1cp
1113882 1113883
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(40)
)en for each λ isin Λ1 with
Λ1 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1pNpminus 1( 1113857min cq cp
1113936Nk1 max |ξ∣lecF(k ξ)
⎛⎝ ⎞⎠ (41)
the problem (Df
λ ) admits at least two positive solutions
Proof Put ΦΨ as in (18) It is clear that infX(Φ) Φ(0)
Ψ(0) 0 According to Lemma 3 we know that a nonzerocritical point in S of the functional J+
λ is precisely a positivesolution of problem (D
f
λ ) Next we just need to provecondition (21) of Lemma 1
We observe that Linfin gt 0 from (40) and Λ1 is nonde-generate Fix λ isin Λ1 Lemma 4 ensures that Jλ satisfies
(PS)-condition for all λgt 2pN + 1113957s minus 2pminus 1qLinfin and it isunbounded from below We let u isin Φminus 1(minusinfin r] that is(upp) + (1113936
Nk1 s(k)|u(k)|qq)le r Put
r 1 + slowastN
pminus 1
pNpminus1 min cq c
p1113864 1113865 (42)
If r (1 + slowastNpminus 1pNpminus 1)cq it means that cge 1
According to (8) we obtain
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max c cqp
1113966 1113967 c (43)
If r (1 + slowastNpminus 1pNpminus 1)cp we know 0lt clt 1 then
|u(k)|le uinfin lemaxpNpminus 1
1 + slowastNpminus11113888 1113889
1q
r1q
pNpminus 1
1 + slowastNpminus11113888 1113889
1p
r1p⎧⎨
⎩
⎫⎬
⎭ max cpq
c1113966 1113967 c (44)
To sum up we know that |u(k)|le c for all k isin Z(1 N)
Furthermore we have
Ψ(u) 1113944N
k1F(k u(k))le 1113944
N
k1max|ξ|lec
F(k ξ) (45)
for all u isin S with Φ(u)le r Hence
supuisinΦminus1(minusinfinr]Ψ(u)
rle
pNpminus 1
1 + slowastNpminus1 1113944
N
k1max|ξ|lec
F(k ξ)max1cq
1cp
1113882 1113883
(46)
6 Discrete Dynamics in Nature and Society
Now let ω(k) d for all k isin Z(1 N) and ω(0)
Δω(N) 0 Clearly ω isin S It is easy to account that Φ(ω)
dppminus 1 + dqqminus 11113957s then
Ψ(ω)
Φ(ω)
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s (47)
Consequently from (46) (47) and assumption (40) wecan obtain
supuisinΦminus 1(minusinfinr]
Ψ(u)
rltΨ(ω)
Φ(ω)
(48)
Moreover because 0lt dlt c and from (40) we obtain
0ltdpp
minus 1+ d
minus 11113957slt
1 + slowastNpminus 1
pNpminus1 min cq c
p1113864 1113865 (49)
that is mean that 0ltΦ(ω)lt r
Hence the problem (Df
λ ) admits at least two positivesolutions by Lemma 1 and Lemma 3 for all λ isin Λ1
Remark 1 If f(k t) is a nonnegative function and there aretwo positive constants c d with dlt c such that
pNpminus 1
1 + slowastNpminus1 max
1113936Nk1 F(k c)
cq1113936
Nk1 F(k c)
cp1113896 1113897
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(50)
then the result of)eorem 2 is also valid for each λ isin Λ2 with
Λ2 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1 mincq
1113936Nk1 F(k c)
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889 (51)
)ere are some consequences of )eorem 2 as follows
Corollary 1 Let g R⟶ [0 +infin) be a continuous functionsuch that f(k t) α(k)g(t) where α(k)gt 0 for allk isin Z(1 N) Put A 1113936
Nk1 α(k) G(t) 1113938
t
0 g(ξ)dξ for allt isin R and Llowastinfin min
kisin[1N]α(k) liminf
t⟶+infin(G(t)tp)gt 0
If there exists cgt dgt 0 such that
pNpminus 1
1 + slowastNpminus1 AG(c)max
1cq
1cp
1113882 1113883
ltminAG(d)
dppminus1 + dqqminus11113957s
qLlowastinfin2pN + 1113957s minus 2pminus11113896 1113897
(52)
then the problem Df
λ has at least two positive solutions foreach λ isin Λ3 with
Λ3 ≔ maxdppminus 1 + dqqminus 11113957s
AG(d)2pN + 1113957s minus 2pminus 1
qLlowastinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1min cq cp
AG(c)1113888 1113889 (53)
Proof Consider the function f Z(1 N) times R⟶ R isgiven as
f(k ξ) α(k)g(ξ) forallk isin Z(1 N) ξ isin R (54)
so that
1113944
N
k1max|ξ|lec
F(k ξ) AG(c) 1113944N
k1F(k d) AG(d) (55)
)en the conclusion can be obtained by)eorem 2
Corollary 2 Assume f be a continuous function withf(k 0)ge 0 and
limsupt⟶0+
F(k t)
tp +infin (56)
limt⟶+infin
F(k t)
tp +infin (57)
for all k isin Z(0 N) Put λlowast (1 + slowastNpminus 1pNpminus 1)min supcgt01113864
(cq1113936Nk1max |ξ∣lecF(k ξ)) supcgt0(cp 1113936
Nk1 max |ξ∣lecF(k ξ))
-en for each λ isin (0 λlowast) the problem (Df
λ ) admits at least twopositive solutions
Proof We know that Linfin +infin from (57) Fix λ isin (0 λlowast)and then there exists cgt 0 such that
λlt1 + slowastN
pminus 1
pNpminus1
middot min supcgt0
cq
1113936Nk1 max|ξ|lecF(k ξ)
supcgt0
cp
1113936Nk1 max|ξ|lecF(k ξ)
⎧⎨
⎩
⎫⎬
⎭
(58)
From (56) we can also obtain
limsupt⟶0+
1113936Nk1 F(k t)
tp +infin (59)
and then there exists d isin (0 c) such that (1113936Nk1 F(k d)
dppminus 1 + dqqminus 11113957s)gt (1λ) )erefore )eorem 2 ensures theconclusion
Discrete Dynamics in Nature and Society 7
Remark 2 If f(k t) is a nonnegative function for all(k t) isin Z(1 N) times [0 +infin) As long as condition (56) holdsfor at least one k isin Z(1 N) then Corollary 2 ensures thatthe solutions are obtained for each
λ isin 01 + slowastN
pminus 1
pNpminus1 min supcgt0
cq
1113936Nk1 F(k c)
supcgt0
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889
(60)
Remark 3 When f(k t) f(t) for all k isin Z(1 N) )e-orem 1 can be ensured by Corollary 2 Obviously condition(4(a)) implies f(0)ge 0 Specially if f is nonnegative weonly need condition (4(a)) to get the corresponding so-lution for each
λ isin 01 + slowastN
pminus 1
pNpmin sup
cgt0
cq
F(c) sup
cgt0
cp
F(c)1113896 11138971113888 1113889 (61)
Example 1 Let p 4 q 2 N 3 s(k) 12 andf(k t)
et
Put
x(c) c2
ec minus 1 (62)
)en
xprime(c) c (2 minus c)ec minus 2( )
ec minus 1( )2 (63)
Let z(c) (2 minus c)ec minus 2 then zprime(c) (1 minus c)ec So z(c)
is increasing in c isin (0 1) and decreasing in c isin (1 +infin)Since z(0) 0 and z(+infin) minusinfin there exists an uniquec1 isin (1 +infin) such that z(c1) 0 )us x(c) in increasing inc isin (0 c1) and decreasing in c isin (c1 +infin) )is means thatsupcgt0 x(c) x(c1) In fact c1 asymp 15936
Similarly put y(c) c4ec minus 1 we can show that thereexists a unique c2 isin (3 +infin) such that supcgt0y(c) y(c2)
In fact c2 asymp 39207
Since
y c2( 1113857 c42
ec2 minus 1gt
c41ec1 minus 1
c21x c1( 1113857gt x c1( 1113857 (64)
then
1 + slowastNpminus 1
pNpmin sup
cgt0x(c) sup
cgt0y(c)1113896 1113897
1 + slowastN
pminus 1
pNpx c1( 1113857 asymp 06496
(65)
)erefore for each λ isin (0 06496) the problem
minusΔ ϕ4(Δu(k minus 1))( 1113857 + 12ϕ2(u(k)) λeu(k) k 1 2 3
u(0) Δu(3) 0
⎧⎨
⎩
(66)
admits at least two positive solutions
Example 2 Let N 3 p 3 and q 2 and f be a functionas follows
f(k t)
0 if tlt 0kt
3radic
if 0le tle 1kt
3radic
+ 225t2 minus 225 if tgt 1
⎧⎪⎪⎨
⎪⎪⎩(67)
From Remark 1 we can choose c 1 and d 002 Easycalculation shows that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1F(k c)max
1cq
1cp
1113882 1113883 pNpminus 1
1 + slowastNpminus1 1113944
3
k111139461
0
kt
3radic
dt
asymp 13693
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s
11139363k1 1113946
002
0
kt
3radic
dt
dppminus1 + dqqminus11113957sasymp 31386
qLinfin2pN + 1113957s minus 2pminus1
166
limt⟶+infin
kt
3radic
+ 225t2 minus 225t2
asymp 34091
(68)
which satisfy condition (50) )us for each λ isin(03186 07303) the problem
minusΔ ϕ3(Δu(k minus 1))( 1113857 + 8ϕ2(u(k)) λf(k u(k)) k 1 2 3
u(0) Δu(3) 01113896
(69)
admits at least two positive solutions
Data Availability
No data were used to support the study
Conflicts of Interest
)e authors declare that there are no conflicts of interestregarding the publication of this paper
Authorsrsquo Contributions
All authors contributed equally to the writing of this paperAll authors read and approved the final manuscript
Acknowledgments
)is work was supported by the National Natural ScienceFoundation of China (Grant no 11971126) and Program forChangjiang Scholars and Innovative Research Team inUniversity (Grant no IRTminus16R16)
References
[1] R P Agarwal Difference Equations and Inequalities -eoryMethods and Applications Marcel Dekker Inc New YorkNY USA 2000
[2] A Kristaly M Mihailescu V Radulescu and S TersianldquoSpectral estimates for a nonhomogeneous difference
8 Discrete Dynamics in Nature and Society
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9
Now let ω(k) d for all k isin Z(1 N) and ω(0)
Δω(N) 0 Clearly ω isin S It is easy to account that Φ(ω)
dppminus 1 + dqqminus 11113957s then
Ψ(ω)
Φ(ω)
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s (47)
Consequently from (46) (47) and assumption (40) wecan obtain
supuisinΦminus 1(minusinfinr]
Ψ(u)
rltΨ(ω)
Φ(ω)
(48)
Moreover because 0lt dlt c and from (40) we obtain
0ltdpp
minus 1+ d
minus 11113957slt
1 + slowastNpminus 1
pNpminus1 min cq c
p1113864 1113865 (49)
that is mean that 0ltΦ(ω)lt r
Hence the problem (Df
λ ) admits at least two positivesolutions by Lemma 1 and Lemma 3 for all λ isin Λ1
Remark 1 If f(k t) is a nonnegative function and there aretwo positive constants c d with dlt c such that
pNpminus 1
1 + slowastNpminus1 max
1113936Nk1 F(k c)
cq1113936
Nk1 F(k c)
cp1113896 1113897
ltmin1113936
Nk1 F(k d)
dppminus1 + dqqminus11113957s
qLinfin2pN + 1113957s minus 2pminus11113896 1113897
(50)
then the result of)eorem 2 is also valid for each λ isin Λ2 with
Λ2 ≔ maxdppminus 1 + dqqminus 11113957s
1113936Nk1 F(k d)
2pN + 1113957s minus 2pminus 1
qLinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1 mincq
1113936Nk1 F(k c)
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889 (51)
)ere are some consequences of )eorem 2 as follows
Corollary 1 Let g R⟶ [0 +infin) be a continuous functionsuch that f(k t) α(k)g(t) where α(k)gt 0 for allk isin Z(1 N) Put A 1113936
Nk1 α(k) G(t) 1113938
t
0 g(ξ)dξ for allt isin R and Llowastinfin min
kisin[1N]α(k) liminf
t⟶+infin(G(t)tp)gt 0
If there exists cgt dgt 0 such that
pNpminus 1
1 + slowastNpminus1 AG(c)max
1cq
1cp
1113882 1113883
ltminAG(d)
dppminus1 + dqqminus11113957s
qLlowastinfin2pN + 1113957s minus 2pminus11113896 1113897
(52)
then the problem Df
λ has at least two positive solutions foreach λ isin Λ3 with
Λ3 ≔ maxdppminus 1 + dqqminus 11113957s
AG(d)2pN + 1113957s minus 2pminus 1
qLlowastinfin1113896 1113897
1 + slowastNpminus 1
pNpminus1min cq cp
AG(c)1113888 1113889 (53)
Proof Consider the function f Z(1 N) times R⟶ R isgiven as
f(k ξ) α(k)g(ξ) forallk isin Z(1 N) ξ isin R (54)
so that
1113944
N
k1max|ξ|lec
F(k ξ) AG(c) 1113944N
k1F(k d) AG(d) (55)
)en the conclusion can be obtained by)eorem 2
Corollary 2 Assume f be a continuous function withf(k 0)ge 0 and
limsupt⟶0+
F(k t)
tp +infin (56)
limt⟶+infin
F(k t)
tp +infin (57)
for all k isin Z(0 N) Put λlowast (1 + slowastNpminus 1pNpminus 1)min supcgt01113864
(cq1113936Nk1max |ξ∣lecF(k ξ)) supcgt0(cp 1113936
Nk1 max |ξ∣lecF(k ξ))
-en for each λ isin (0 λlowast) the problem (Df
λ ) admits at least twopositive solutions
Proof We know that Linfin +infin from (57) Fix λ isin (0 λlowast)and then there exists cgt 0 such that
λlt1 + slowastN
pminus 1
pNpminus1
middot min supcgt0
cq
1113936Nk1 max|ξ|lecF(k ξ)
supcgt0
cp
1113936Nk1 max|ξ|lecF(k ξ)
⎧⎨
⎩
⎫⎬
⎭
(58)
From (56) we can also obtain
limsupt⟶0+
1113936Nk1 F(k t)
tp +infin (59)
and then there exists d isin (0 c) such that (1113936Nk1 F(k d)
dppminus 1 + dqqminus 11113957s)gt (1λ) )erefore )eorem 2 ensures theconclusion
Discrete Dynamics in Nature and Society 7
Remark 2 If f(k t) is a nonnegative function for all(k t) isin Z(1 N) times [0 +infin) As long as condition (56) holdsfor at least one k isin Z(1 N) then Corollary 2 ensures thatthe solutions are obtained for each
λ isin 01 + slowastN
pminus 1
pNpminus1 min supcgt0
cq
1113936Nk1 F(k c)
supcgt0
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889
(60)
Remark 3 When f(k t) f(t) for all k isin Z(1 N) )e-orem 1 can be ensured by Corollary 2 Obviously condition(4(a)) implies f(0)ge 0 Specially if f is nonnegative weonly need condition (4(a)) to get the corresponding so-lution for each
λ isin 01 + slowastN
pminus 1
pNpmin sup
cgt0
cq
F(c) sup
cgt0
cp
F(c)1113896 11138971113888 1113889 (61)
Example 1 Let p 4 q 2 N 3 s(k) 12 andf(k t)
et
Put
x(c) c2
ec minus 1 (62)
)en
xprime(c) c (2 minus c)ec minus 2( )
ec minus 1( )2 (63)
Let z(c) (2 minus c)ec minus 2 then zprime(c) (1 minus c)ec So z(c)
is increasing in c isin (0 1) and decreasing in c isin (1 +infin)Since z(0) 0 and z(+infin) minusinfin there exists an uniquec1 isin (1 +infin) such that z(c1) 0 )us x(c) in increasing inc isin (0 c1) and decreasing in c isin (c1 +infin) )is means thatsupcgt0 x(c) x(c1) In fact c1 asymp 15936
Similarly put y(c) c4ec minus 1 we can show that thereexists a unique c2 isin (3 +infin) such that supcgt0y(c) y(c2)
In fact c2 asymp 39207
Since
y c2( 1113857 c42
ec2 minus 1gt
c41ec1 minus 1
c21x c1( 1113857gt x c1( 1113857 (64)
then
1 + slowastNpminus 1
pNpmin sup
cgt0x(c) sup
cgt0y(c)1113896 1113897
1 + slowastN
pminus 1
pNpx c1( 1113857 asymp 06496
(65)
)erefore for each λ isin (0 06496) the problem
minusΔ ϕ4(Δu(k minus 1))( 1113857 + 12ϕ2(u(k)) λeu(k) k 1 2 3
u(0) Δu(3) 0
⎧⎨
⎩
(66)
admits at least two positive solutions
Example 2 Let N 3 p 3 and q 2 and f be a functionas follows
f(k t)
0 if tlt 0kt
3radic
if 0le tle 1kt
3radic
+ 225t2 minus 225 if tgt 1
⎧⎪⎪⎨
⎪⎪⎩(67)
From Remark 1 we can choose c 1 and d 002 Easycalculation shows that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1F(k c)max
1cq
1cp
1113882 1113883 pNpminus 1
1 + slowastNpminus1 1113944
3
k111139461
0
kt
3radic
dt
asymp 13693
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s
11139363k1 1113946
002
0
kt
3radic
dt
dppminus1 + dqqminus11113957sasymp 31386
qLinfin2pN + 1113957s minus 2pminus1
166
limt⟶+infin
kt
3radic
+ 225t2 minus 225t2
asymp 34091
(68)
which satisfy condition (50) )us for each λ isin(03186 07303) the problem
minusΔ ϕ3(Δu(k minus 1))( 1113857 + 8ϕ2(u(k)) λf(k u(k)) k 1 2 3
u(0) Δu(3) 01113896
(69)
admits at least two positive solutions
Data Availability
No data were used to support the study
Conflicts of Interest
)e authors declare that there are no conflicts of interestregarding the publication of this paper
Authorsrsquo Contributions
All authors contributed equally to the writing of this paperAll authors read and approved the final manuscript
Acknowledgments
)is work was supported by the National Natural ScienceFoundation of China (Grant no 11971126) and Program forChangjiang Scholars and Innovative Research Team inUniversity (Grant no IRTminus16R16)
References
[1] R P Agarwal Difference Equations and Inequalities -eoryMethods and Applications Marcel Dekker Inc New YorkNY USA 2000
[2] A Kristaly M Mihailescu V Radulescu and S TersianldquoSpectral estimates for a nonhomogeneous difference
8 Discrete Dynamics in Nature and Society
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9
Remark 2 If f(k t) is a nonnegative function for all(k t) isin Z(1 N) times [0 +infin) As long as condition (56) holdsfor at least one k isin Z(1 N) then Corollary 2 ensures thatthe solutions are obtained for each
λ isin 01 + slowastN
pminus 1
pNpminus1 min supcgt0
cq
1113936Nk1 F(k c)
supcgt0
cp
1113936Nk1 F(k c)
1113896 11138971113888 1113889
(60)
Remark 3 When f(k t) f(t) for all k isin Z(1 N) )e-orem 1 can be ensured by Corollary 2 Obviously condition(4(a)) implies f(0)ge 0 Specially if f is nonnegative weonly need condition (4(a)) to get the corresponding so-lution for each
λ isin 01 + slowastN
pminus 1
pNpmin sup
cgt0
cq
F(c) sup
cgt0
cp
F(c)1113896 11138971113888 1113889 (61)
Example 1 Let p 4 q 2 N 3 s(k) 12 andf(k t)
et
Put
x(c) c2
ec minus 1 (62)
)en
xprime(c) c (2 minus c)ec minus 2( )
ec minus 1( )2 (63)
Let z(c) (2 minus c)ec minus 2 then zprime(c) (1 minus c)ec So z(c)
is increasing in c isin (0 1) and decreasing in c isin (1 +infin)Since z(0) 0 and z(+infin) minusinfin there exists an uniquec1 isin (1 +infin) such that z(c1) 0 )us x(c) in increasing inc isin (0 c1) and decreasing in c isin (c1 +infin) )is means thatsupcgt0 x(c) x(c1) In fact c1 asymp 15936
Similarly put y(c) c4ec minus 1 we can show that thereexists a unique c2 isin (3 +infin) such that supcgt0y(c) y(c2)
In fact c2 asymp 39207
Since
y c2( 1113857 c42
ec2 minus 1gt
c41ec1 minus 1
c21x c1( 1113857gt x c1( 1113857 (64)
then
1 + slowastNpminus 1
pNpmin sup
cgt0x(c) sup
cgt0y(c)1113896 1113897
1 + slowastN
pminus 1
pNpx c1( 1113857 asymp 06496
(65)
)erefore for each λ isin (0 06496) the problem
minusΔ ϕ4(Δu(k minus 1))( 1113857 + 12ϕ2(u(k)) λeu(k) k 1 2 3
u(0) Δu(3) 0
⎧⎨
⎩
(66)
admits at least two positive solutions
Example 2 Let N 3 p 3 and q 2 and f be a functionas follows
f(k t)
0 if tlt 0kt
3radic
if 0le tle 1kt
3radic
+ 225t2 minus 225 if tgt 1
⎧⎪⎪⎨
⎪⎪⎩(67)
From Remark 1 we can choose c 1 and d 002 Easycalculation shows that
pNpminus 1
1 + slowastNpminus1 1113944
N
k1F(k c)max
1cq
1cp
1113882 1113883 pNpminus 1
1 + slowastNpminus1 1113944
3
k111139461
0
kt
3radic
dt
asymp 13693
1113936Nk1 F(k d)
dppminus1 + dqqminus11113957s
11139363k1 1113946
002
0
kt
3radic
dt
dppminus1 + dqqminus11113957sasymp 31386
qLinfin2pN + 1113957s minus 2pminus1
166
limt⟶+infin
kt
3radic
+ 225t2 minus 225t2
asymp 34091
(68)
which satisfy condition (50) )us for each λ isin(03186 07303) the problem
minusΔ ϕ3(Δu(k minus 1))( 1113857 + 8ϕ2(u(k)) λf(k u(k)) k 1 2 3
u(0) Δu(3) 01113896
(69)
admits at least two positive solutions
Data Availability
No data were used to support the study
Conflicts of Interest
)e authors declare that there are no conflicts of interestregarding the publication of this paper
Authorsrsquo Contributions
All authors contributed equally to the writing of this paperAll authors read and approved the final manuscript
Acknowledgments
)is work was supported by the National Natural ScienceFoundation of China (Grant no 11971126) and Program forChangjiang Scholars and Innovative Research Team inUniversity (Grant no IRTminus16R16)
References
[1] R P Agarwal Difference Equations and Inequalities -eoryMethods and Applications Marcel Dekker Inc New YorkNY USA 2000
[2] A Kristaly M Mihailescu V Radulescu and S TersianldquoSpectral estimates for a nonhomogeneous difference
8 Discrete Dynamics in Nature and Society
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9
problemrdquo Communications in Contemporary Mathematicsvol 12 no 6 pp 1015ndash1029 2010
[3] L M Li and Z Zhou ldquoInfinitely many positive solutions for acoupled discrete boundary value problemrdquoDiscrete Dynamicsin Nature and Society vol 2019 Article ID 8052497 7 pages2019
[4] J X Ling and Z Zhou ldquoPositive solutions of the discreteDirichlet problem involving the mean curvature operatorrdquoOpen Mathematics vol 17 no 1 pp 1055ndash1064 2019
[5] Q Q Zhang ldquoHomoclinic orbits for discrete Hamiltoniansystems with local super-quadratic conditionsrdquo Communi-cations on Pure amp Applied Analysis vol 18 no 1 pp 425ndash4342019
[6] Z Zhou and J X Ling ldquoInfinitely many positive solutions fora discrete two point nonlinear boundary value problem withϕc-Laplacianrdquo Applied Mathematics Letters vol 91 pp 28ndash34 2019
[7] G Bonanno and P Candito ldquoInfinitely many solutions for aclass of discrete non-linear boundary value problemsrdquo Ap-plicable Analysis vol 88 no 4 pp 605ndash616 2009
[8] G H Lin and Z Zhou ldquoHomoclinic solutions of discreteφ-Laplacian equations with mixed nonlinearitiesrdquo Commu-nications on Pure amp Applied Analysis vol 17 no 5pp 1723ndash1747 2018
[9] J S Yu and B Zheng ldquoModeling Wolbachia infection inmosquito population via discrete dynamical modelrdquo Journalof Difference Equations and Applications vol 25 no 11pp 1549ndash1567 2019
[10] P Mei Z Zhou and G H Lin ldquoPeriodic and subharmonicsolutions for a 2nth-order ϕc-Laplacian difference equationcontaining both advances and retardationsrdquo Discrete ContDyn-S vol 12 pp 2085ndash2095 2019
[11] M K Moghadam L Li and S Tersian ldquoExistence of threesolutions for a discrete anisotropic boundary value problemrdquoBulletin of the Iranian Mathematical Society vol 44 no 4pp 1091ndash1107 2018
[12] L Erbe B G Jia and Q Q Zhang ldquoHomoclinic solutions ofdiscrete nonlinear systems via variational methodrdquo Journal ofApplied Analysis and Computation vol 9 pp 271ndash294 2019
[13] S Heidarkhani and M Imbesi ldquoMultiple solutions for partialdiscrete Dirichlet problems depending on a real parameterrdquoJournal of Difference Equations and Applications vol 21 no 2pp 96ndash110 2015
[14] G H Lin Z Zhou and J S Yu ldquoGround state solutions ofdiscrete asymptotically linear Schrodinge equations withbounded and non-periodic potentialsrdquo Journal of Dynamicsand Differential Equations 2019
[15] M Mihǎilescu V Rǎdulescu and S Tersian ldquoHomoclinicsolutions of difference equations with variable exponentsrdquoTopological Methods in Nonlinear Analysis vol 38 pp 277ndash289 2011
[16] J Kuang and Z Guo ldquoHeteroclinic solutions for a class of p-Laplacian difference equations with a parameterrdquo AppliedMathematics Letters vol 100 Article ID 106034 2020
[17] C Bereanu and J Mawhin ldquoBoundary value problems forsecond-order nonlinear difference equations with discreteφ-Laplacian and singular φrdquo Journal of Difference Equationsand Applications vol 14 no 10-11 pp 1099ndash1118 2008
[18] W G Kelly and A C Peterson Difference Equations AnIntroduction with Applications Academic Press San DiegoCA USA 1991
[19] G Bonanno P Jebelean and C Serban ldquoSuperlinear discreteproblemsrdquo Applied Mathematics Letters vol 52 pp 162ndash1682016
[20] J Henderson and H B )ompson ldquoExistence of multiplesolutions for second-order discrete boundary value prob-lemsrdquo Computers amp Mathematics with Applications vol 43no 10-11 pp 1239ndash1248 2002
[21] Y H Long and J L Chen ldquoExistence of multiple solutions tosecond-order discrete Neumann boundary problemsrdquo Ap-plied Mathematics Letters vol 83 pp 7ndash14 2018
[22] B Ricceri ldquoA general variational principle and some of itsapplicationsrdquo Journal of Computational and Applied Math-ematics vol 133 no 1-2 pp 401ndash410 2000
[23] Z Zhou and D F Ma ldquoMultiplicity results of breathers for thediscrete nonlinear Schrodinger equations with unboundedpotentialsrdquo Science China Mathematics vol 58 no 4pp 781ndash790 2015
[24] R P Agarwal K Perera and D OrsquoRegan ldquoMultiple positivesolutions of singular discrete p-Laplacian problems via var-iational methodsrdquo Advances in Difference Equationsvol 2015 no 2 Article ID 690272 2005
[25] A Nastasi C Vetro and F Vetro ldquoPositive solutions ofdiscrete boundary value problems with the (p q)-Laplacianoperatorrdquo Electronic Journal of Differential Equationsvol 225 pp 1ndash12 2017
[26] Z Zhou and M T Su ldquoBoundary value problems for 2nth-order ϕc-Laplacian difference equations containing bothadvance and retardationrdquo Applied Mathematics Lettersvol 41 pp 7ndash11 2015
[27] Z Zhou J S Yu and Y M Chen ldquoHomoclinic solutions inperiodic diffrence equations with saturable nonlinearityrdquoScience China Mathematics vol 54 no 1 pp 83ndash93 2011
[28] G DrsquoAguı J Mawhin and A Sciammetta ldquoPositive solutionsfor a discrete two point nonlinear boundary value problemwith p-Laplacianrdquo Journal of Mathematical Analysis andApplications vol 447 no 1 pp 383ndash397 2017
[29] G DrsquoAguı A Sciammetta and E Tornatore ldquoTwo non-zerosolutions for Sturm-Liouville equations with mixed boundaryconditionrdquo Nonlinear Anal Real World Appl vol 47pp 324ndash331 2019
[30] G Bonanno and G DrsquoAguı ldquoTwo non-zero solutions forelliptic Dirichlet problemsrdquo Zeitschrift fur Analysis und ihreAnwendungen vol 35 no 4 pp 449ndash464 2016
[31] G Bonanno ldquoA critical point theorem via the Ekeland var-iational principlerdquo Nonlinear Analysis -eory Methods ampApplications vol 75 no 5 pp 2992ndash3007 2012
[32] A Ambrosetti and P H Rabinowitz ldquoDual variationalmethods in critical point theory and applicationsrdquo Journal ofFunctional Analysis vol 14 no 4 pp 349ndash381 1973
[33] G Bonanno P Candito and G DrsquoAguı ldquoVariational methodson finite dimensional Banach space and discrete problemsrdquoAdvanced Nonlinear Studies vol 14 no 4 pp 915ndash939 2014
Discrete Dynamics in Nature and Society 9