Several existence theorems of twin positive solutions are established for a
nonlinear m-point boundary value problem of third-order p-Laplacian dynamic equations on time
scales by using a fixed point theorem. We present two theorems and four corollaries which generalize
the results of related literature. As an application, an example to demonstrate our results is given.
The obtained conditions are different from some known results.
1. Introduction
A time scale T is a nonempty closed subset of ℝ.
We make the blanket assumption that 0 and T are points in T.
By an interval (0,T),
we always mean the intersection of the real interval (0,T) with the given time scale, that is, (0,T)∩T.
In this paper, we will be concerned with the existence of
positive solutions of the p-Laplacian dynamic equations on time scales:(ϕp(uΔ∇))∇+a(t)f(t,u(t))=0,t∈(0,T),ϕp(uΔ∇(0))=∑i=1m−2aiϕp(uΔ∇(ξi)),uΔ(0)=0,u(T)=∑i=1m−2biu(ξi),where ϕp(s) is p-Laplacian operator; that is, ϕp(s)=|s|p−2s,p>1,ϕp−1=ϕq,1/p+1/q=1,0<ξ1<⋯<ξm−2<ρ(T),
and
ai,bi∈[0,+∞),i=1,2,…,
satisfy 0<∑i=1m−2ai<1 and ∑i=1m−2bi<1;
a(t)∈Cld([0,T],[0,+∞)) and there exists t0∈(ξm−2,T) such that a(t0)>0;
f∈C([0,T]×[0,+∞),[0,+∞)).
We point out that the Δ-derivative and the ∇-derivative in (1.2) and the Cld space in (H2) are defined in Section 2.
Recently, there has been much attention paid to the
existence of positive solutions for third-order nonlinear boundary value
problems of differential equations. For example, see [1–10] and the
listed references. Anderson [2] considered the following third-order nonlinear
problem:x′′′(t)=f(t,x(t)),t1≤t≤t3,x(t1)=x′(t2)=0,γx(t3)+δx′′(t3)=0.He used the Krasnoselskii and
the Leggett and Williams fixed-point theorems to prove the existence of
solutions to the nonlinear problem (1.3). Li [6] considered the existence of
single and multiple positive solutions to the nonlinear singular third-order
two-point boundary value problem:u′′′(t)+λa(t)f(u(t))=0,0<t<1,u(0)=u′(0)=u′′(1)=0.Under various assumptions on a and f,
they established intervals of the parameter λ which yield the existence of at least two and
infinitely many positive solutions of the boundary value problem by using
Krasnoselski's fixed-point theorem of cone expansion-compression type. Liu et
al. [7] discussed the existence of at least one or two nondecreasing positive solutions
for the following singular nonlinear
third-order differential equations:x′′′(t)+λα(t)f(t,x(t))=0,a<t<b,x(a)=x′′(a)=x′(b)=0.Green's function and the
fixed-point theorem of cone expansion-compression type are utilized in their
paper. In [8], Sun considered the following nonlinear singular third-order
three-point boundary value problem:u′′′(t)−λa(t)F(t,u(t))=0,0<t<1,u(0)=u′(η)=u′′(1)=0.He obtained various results on
the existence of single and multiple positive solutions to the boundary value
problem (1.6) by using a fixed-point theorem of cone expansion-compression type
due to Krasnosel'skii. In [10], Zhou and Ma studied the existence and iteration
of positive solutions for the following third-order generalized right-focal
boundary value problem with p-Laplacian operator:(ϕp(u′′))′(t)=q(t)f(t,u(t)),0≤t≤1,u(0)=∑i=1mαiu(ξi),u′(η)=0,u′′(1)=∑i=1nβiu′′(θi).They established a corresponding
iterative scheme for (1.7) by using the monotone iterative technique.
On the other hand, the existence of positive solutions
for third-order nonlinear boundary value problems of difference equations is
also extensively studied by a number of authors (see [1, 3, 5, 9] and the
listed references). The present work is motivated by a recent paper [4]. In
[4], Henderson and Yin considered the existence of solutions for a third-order
boundary value problem on a time-scale equation of the formuΔ3=f(t,u,uΔ,uΔΔ),t∈T,which is uniform for the
third-order difference equation and the third-order differential equation.
2. Preliminaries and Lemmas
For
convenience, we list the following definitions which can be found in [4, 11–15].
Definition 2.1.
Let T be a time scale. For t<supT and r>infT,
define the forward jump operator σ and the backward jump operator ρ,
respectively, byσ(t)=inf{τ∈T∣τ>t}∈T,ρ(r)=sup{τ∈T∣τ<r}∈Tfor all t,r∈T.
If σ(t)>t, t is said to be right-scattered, and if ρ(r)<r,r is said to be left-scattered; if σ(t)=t, t is said to be right-dense, and if ρ(r)=r,r is said to be left-dense. If T has a right-scattered minimum m,
define Tk=T−{m};
otherwise set Tk=T.
If T has a left-scattered maximum M,
define Tk=T−{M};
otherwise set Tk=T.
Definition 2.2.
For f:T→R and t∈Tk,
the delta derivative of f at the point t is defined to be the number fΔ(t) (provided that it exists), with the property
that for each ϵ>0 there is a neighborhood U of t such that|f(σ(t))−f(s)−fΔ(t)(σ(t)−s)|≤ϵ|σ(t)−s|for all s∈U.
For f:T→R and
t∈Tk,
the nabla derivative of f at t
is denoted by f∇(t) (provided that it exists), with the property
that for each ϵ>0 there is a neighborhood U of t such that|f(ρ(t))−f(s)−f∇(t)(ρ(t)−s)|≤ϵ|ρ(t)−s|for all s∈U.
Definition 2.3.
A function f is left-dense continuous (i.e., ld-continuous) if f is continuous at each left-dense point in T,
and its right-sided limit exists at each right-dense point in T.
Definition 2.4.
If ϕΔ(t)=f(t),
then one defines the delta integral by∫abf(t)Δt=ϕ(b)−ϕ(a).
If F∇(t)=f(t),
then one defines the nabla integral by∫abf(t)∇t=F(b)−F(a).
To prove the main results in this paper, we will
employ several lemmas. These lemmas are based on the linear BVP(ϕp(uΔ∇))∇+h(t)=0,t∈(0,T),ϕp(uΔ∇(0))=∑i=1m−2aiϕp(uΔ∇(ξi)),uΔ(0)=0,u(T)=∑i=1m−2biu(ξi).
Lemma 2.5.
If ∑i=1m−2ai≠1 and ∑i=1m−2bi≠1,
then for h∈Cld[0,T] the BVP (2.6)-(2.7) has the unique solutionu(t)=−∫0t(t−s)ϕq(∫0sh(τ)∇τ−A)∇s+C,whereA=−∑i=1m−2ai∫0ξih(τ)∇τ1−∑i=1m−2ai,C=∫0T(T−s)ϕq(∫0sh(τ)∇τ−A)∇s−∑i=1m−2bi∫0ξi(ξi−s)ϕq(∫0sh(τ)∇τ−A)∇s1−∑i=1m−2bi.
Proof.
(i) Let
u be a solution, then we will show that (2.8)
holds. By taking the nabla integral of problem (2.6) on (0,t),
we haveϕp(uΔ∇(t))=−∫0th(τ)∇τ+AthenuΔ∇(t)=ϕq(−∫0th(τ)∇τ+A)=−ϕq(∫0th(τ)∇τ−A).By taking the nabla integral of
(2.11) on (0,t),
we can getuΔ(t)=−∫0tϕq(∫0sh(τ)∇τ−A)∇s+B.By taking the delta integral of
(2.12) on (0,t),
we can getu(t)=−∫0t(t−s)ϕq(∫0sh(τ)∇τ−A)∇s+Bt+C.Similarly, let t=0 on (2.10), then we have ϕp(uΔ∇(0))=A;
let t=ξi on (2.10), then we haveϕp(uΔ∇(ξi))=−∫0ξih(τ)∇τ+A.
Let t=0 on (2.12), then we haveuΔ(0)=B.Let t=T on (2.13), then we haveu(T)=−∫0T(T−s)ϕq(∫0sh(τ)∇τ−A)∇s+BT+C.Similarly, let t=ξi on (2.13), then we haveu(ξi)=−∫0ξi(ξi−s)ϕq(∫0sh(τ)∇τ−A)∇s+Bξi+C.By the boundary condition (2.7),
we can getB=0,A=∑i=1m−2ai(−∫0ξih(τ)∇τ+A).Solving
(2.19), we getA=−∑i=1m−2ai∫0ξih(τ)∇τ1−∑i=1m−2ai.By the boundary condition (2.7),
we can obtain−∫0T(T−s)ϕq(∫0sh(τ)∇τ−A)∇s+C=∑i=1m−2bi[−∫0ξi(ξi−s)ϕq(∫0sh(τ)∇τ−A)∇s+C].Substituting (2.20) in the above
expression, one hasC=∫0T(T−s)ϕq(∫0sh(τ)∇τ−A)∇s−∑i=1m−2bi∫0ξi(ξi−s)ϕq(∫0sh(τ)∇τ−A)∇s1−∑i=1m−2bi.
(ii) We show that the
function u given in (2.8) is a solution.
Let u be as in (2.8). By [12, Theorem 2.10(iii)] and
taking the delta derivative of (2.8), we haveuΔ(t)=−∫0tϕq(∫0sh(τ)∇τ−A)∇s;moreover, we get
uΔ∇(t)=−ϕq(∫0th(τ)∇τ−A),ϕp(uΔ∇)=−(∫0th(τ)∇τ−A).Taking the nabla derivative of
this expression yields (ϕp(uΔ∇))∇=−h(t).
Also, routine calculation verifies that u satisfies the boundary value conditions in
(2.7) so that u given in (2.8) is a solution of (2.6) and
(2.7). The proof is complete.
Lemma 2.6.
Assume(H1)holds. For h∈Cld[0,T] and h≥0,
the unique solution u of (2.6) and (2.7) satisfiesu(t)≥0fort∈[0,T].
According to Lemma 2.5, we getu(0)=C=∫0T(T−s)φ0(s)∇s−∑i=1m−2bi∫0ξi(ξi−s)φ0(s)∇s1−∑i=1m−2bi≥∫0T(T−s)φ0(s)∇s−∑i=1m−2bi∫0ξi(T−s)φ0(s)∇s1−∑i=1m−2bi=∫0T(T−s)φ0(s)∇s−∑i=1m−2bi(∫0T(T−s)φ0(s)∇s−∫ξiT(T−s)φ0(s)∇s)1−∑i=1m−2bi=∫0T(T−s)φ0(s)∇s+∑i=1m−2bi∫ξiT(T−s)φ0(s)∇s1−∑i=1m−2bi≥0,u(T)=−∫0T(T−s)φ0(s)∇s+C=−∫0T(T−s)φ0(s)∇s+∫0T(T−s)φ0(s)∇s−∑i=1m−2bi∫0ξi(ξi−s)φ0(s)∇s1−∑i=1m−2bi≥−∫0T(T−s)φ0(s)∇s+∫0T(T−s)φ0(s)∇s−∑i=1m−2bi∫0ξi(T−s)φ0(s)∇s1−∑i=1m−2bi=∑i=1m−2bi∫ξiT(T−s)φ0(s)∇s1−∑i=1m−2bi≥0.If t∈(0,T),
we haveu(t)=−∫0t(t−s)φ0(s)∇s+11−∑i=1m−2bi[∫0T(T−s)φ0(s)∇s−∑i=1m−2bi∫0ξi(ξi−s)φ0(s)∇s]≥−∫0T(T−s)φ0(s)∇s+11−∑i=1m−2bi[∫0T(T−s)φ0(s)∇s−∑i=1m−2bi∫0ξi(T−s)φ0(s)∇s]=11−∑i=1m−2bi[−(1−∑i=1m−2bi)∫0T(T−s)φ0(s)∇s+∫0T(T−s)φ0(s)∇s−∑i=1m−2bi∫0ξi(T−s)φ0(s)∇s]=11−∑i=1m−2bi∑i=1m−2bi∫ξiT(T−s)φ0(s)∇s≥0.
So u(t)≥0,t∈[0,T].
Lemma 2.7.
Assume(H1)holds. If h∈Cld[0,T] and h≥0,
then the unique solution u of (2.6) and (2.7) satisfiesinft∈[0,T]u(t)≥γ∥u∥,whereγ=∑i=1m−2bi(T−ξi)T−∑i=1m−2biξi,∥u∥=maxt∈[0,T]|u(t)|.
Proof.
It is easy to
check that uΔ(t)=−∫0tφ0(s)∇s≤0;
this implies that∥u∥=u(0),mint∈[0,T]u(t)=u(T).It is easy to see that uΔ(t2)≤uΔ(t1) for any t1,t2∈[0,T] with t1≤t2.
Hence, uΔ(t) is a decreasing function on [0,T].
This means that the graph of u(t) is concave down on (0,T).
For each i∈{1,2,…,m−2},
we haveu(T)−u(0)T−0≥u(T)−u(ξi)T−ξi,that is,Tu(ξi)−ξiu(T)≥(T−ξi)u(0),so thatT∑i=1m−2biu(ξi)−∑i=1m−2biξiu(T)≥∑i=1m−2bi(T−ξi)u(0).With the boundary condition u(T)=∑i=1m−2biu(ξi),
we haveu(T)≥∑i=1m−2bi(T−ξi)T−∑i=1m−2biξiu(0).This completes the proof.
Let the norm on Cld[0,T] be the maximum norm. Then, the Cld[0,T] is a Banach space. It is easy to see that BVP
(1.1)-(1.2) has a solution u=u(t) if and only if u is a fixed point of the operator(Au)(t)=−∫0t(t−s)ϕq(∫0sa(τ)f(τ,u(τ))∇τ−A)∇s+C,whereA=−∑i=1m−2ai∫0ξia(τ)f(τ,u(τ))∇τ1−∑i=1m−2ai,C=∫0T(T−s)ϕq(∫0sa(τ)f(τ,u(τ))∇τ−A)∇s−∑i=1m−2bi∫0ξi(ξi−s)ϕq(∫0sa(τ)f(τ,u(τ))∇τ−A)∇s1−∑i=1m−2bi.DenoteK={u∣u∈Cld[0,T],u(t)≥0,inft∈[0,T]u(t)≥γ∥u∥},where γ is the same as in Lemma 2.7. It is obvious
that K is a cone in Cld[0,T].
By Lemma 2.7, A(K)⊂K.
So by applying Arzela-Ascoli theorem on time scales [16], we can obtain that A(K) is relatively compact. In view of Lebesgue's
dominated convergence theorem on time scales [13], it is easy to prove that A is continuous. Hence, A:K→K is completely continuous.
Lemma 2.8.
A:K→K is completely continuous.
Proof.
First, we
show that A maps bounded set into bounded
set.
Assume c>0 is a constant and u∈Kc¯={u∈K:∥u∥≤c}.
Note that the continuity of f guarantees that there is c′>0 such that f(t,u(t))≤ϕp(c′) for t∈[0,T].
So∥Au∥=maxt∈[0,T]|Au(t)|≤C≤∫0T(T−s)ϕq(∫0sa(τ)f(τ,u(τ))∇τ−A)∇s1−∑i=1m−2bi≤c′∫0T(T−s)ϕq(∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ/(1−∑i=1m−2ai))∇s1−∑i=1m−2bi.That is, AKc¯ is uniformly bounded.
In addition, notice that for any t1,t2∈[0,T],
we have|Au(t1)−Au(t2)|=|∫0t1(t2−t1)ϕq(∫0sa(τ)f(τ,u(τ))∇τ−A)∇s+∫t1t2(t2−s)ϕq(∫0sa(τ)f(τ,u(τ))∇τ−A)∇s|≤c′|t1−t2|[∫0Tϕq(∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ1−∑i=1m−2ai)∇s+Tmaxs∈[0,T]ϕq(∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ1−∑i=1m−2ai)].So, by applying Arzela-Ascoli
theorem on time scales [16], we obtain that AKc¯ is relatively compact.
Finally, we prove that A:Kc¯→K is continuous. Suppose that {un}n=1∞⊂Kc¯ and un(t) converges to u*(t) uniformly on [0,T]. Hence, {Aun(t)}n=1∞ is uniformly bounded and equicontinuous on [0,T]. The Arzela-Ascoli theorem on time scales [16]
tells us that there exists uniformly convergent subsequence in {Aun(t)}n=1∞. Let {Aun(m)(t)}m=1∞ be a subsequence which converges to v(t) uniformly on [0,T]. In addition,0≤Aun(t)≤c′∫0T(T−s)ϕq(∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ/(1−∑i=1m−2ai))∇s1−∑i=1m−2bi.Observe the expression of {Aun(m)(t)}, and then letting m→∞, we obtainv(t)=−∫0t(t−s)ϕq(∫0sa(τ)f(τ,u*(τ))∇τ−A*)∇s+C*,whereA*=−∑i=1m−2ai∫0ξia(τ)f(τ,u*(τ))∇τ1−∑i=1m−2ai,C*=11−∑i=1m−2bi[∫0T(T−s)ϕq(∫0sa(τ)f(τ,u*(τ))∇τ−A*)∇s−∑i=1m−2bi∫0ξi(ξi−s)ϕq(∫0sa(τ)f(τ,u*(τ))∇τ−A*)∇s]. Here, we have used the Lebesgue
dominated convergence theorem on time scales [13]. From the
definition of A,
we know that v(t)=Au*(t) on [0,T]. This shows that each subsequence of {Aun(t)}n=1∞ uniformly converges to Au*(t). Therefore, the sequence {Aun(t)}n=1∞ uniformly converges to Au*(t). This means that A is continuous at u*∈Kc¯. So, A is continuous on Kc¯ since u* is arbitrary. Thus, A is completely continuous. This proof is
complete.
Since∫0sa(τ)f(τ,u(τ))∇τ−A=∫0sa(τ)f(τ,u(τ))∇τ+∑i=1m−2ai∫0ξia(τ)f(τ,u(τ))∇τ1−∑i=1m−2ai≥0,then φ(s)≥0.
For all t∈(0,T],
we have(∫0t(t−s)φ(s)∇st)∇=t∫0tφ(s)∇s−∫0t(t−s)φ(s)∇stρ(t)≥0.In fact, let ψ(t)=t∫0tφ(s)∇s−∫0t(t−s)φ(s)∇s;
taking the nabla derivative of this expression, we haveψ∇(t)=∫0tφ(s)∇s+tφ(t)−∫0tφ(s)∇s=tφ(t)≥0.Hence, ψ(t) is a nondecreasing function on [0,T].
That is,ψ(t)≥0.For all t∈(0,T],∫0t(t−s)φ(s)∇st≤∫0T(T−s)φ(s)∇sT.By (2.51), for ξi(i=1,…,m−2),
we have∫0ξi(ξi−s)φ(s)∇s≤ξiT∫0T(T−s)φ(s)∇s.
Lemma 2.10 (See [17]).
Let E be a Banach space, and let K⊂E be a cone. Assume Ω1,Ω2 are open bounded
subsets of E with 0∈Ω1,Ω1¯⊂Ω2, and let
F:K∩(Ω2¯∖Ω1)→K be a completely continuous operator such that
∥Fu∥≤∥u∥,u∈K∩∂Ω1, and ∥Fu∥≥∥u∥,u∈K∩∂Ω2,
or
∥Fu∥≥∥u∥,u∈K∩∂Ω1,
and ∥Fu∥≤∥u∥,u∈K∩∂Ω2.
Then, F has a fixed point in K∩(Ω2¯∖Ω1).
Now, we introduce the following notations. LetA0={11−∑i=1m−2bi∫0T(T−s)ϕq[∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ1−∑i=1m−2ai]∇s}−1,B0={T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)∫0T(T−s)ϕq[∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ1−∑i=1m−2ai]∇s}−1.For l>0,Ωl={u∈K:∥u∥<l},
and ∂Ωl={u∈K:∥u∥=l},α(l)=sup{∥Au∥:u∈∂Ωl},β(l)=inf{∥Au∥:u∈∂Ωl},by Lemma 2.6, where
α and β are well defined.
3. Main ResultsTheorem 3.1.
Assume(H1), (H2), and (H3)hold, and assume that the following conditions
hold:
pi∈C([0,+∞),[0,+∞)),i=1,2, and liml→0+¯p1(l)lp−1<A0p−1,liml→∞¯p2(l)lp−1<A0p−1;
ki∈L1([0,T],[0,+∞)),i=1,2;
there exist 0<c1≤c2 and 0≤λ2<p−1<λ1 such thatf(t,l)≤p1(l)+k1(t)lλ1,(t,l)∈[0,T]×[0,c1],f(t,l)≤p2(l)+k2(t)lλ2,(t,l)∈[0,T]×[c2,+∞);
there exists
b>0 such thatmin{f(t,l):(t,l)∈[0,T]×[γb,b]}≥(bB0)p−1.
Then, problem (1.1)-(1.2) has at least two positive solutions u1*,u2* satisfying 0<∥u1*∥<b<∥u2*∥.
Theorem 3.2.
Assume(H1), (H2),
and (H3)hold, and assume that the following conditions
hold:
pi∈C([0,+∞),[0,+∞)),i=3,4, and liml→0+¯p3(l)lp−1>(B0γ)p−1,liml→∞¯p4(l)lp−1>(B0γ)p−1;
ki∈L1([0,T],[0,+∞)),i=3,4;
there exist 0<c3≤c4 and 0≤λ4<p−1<λ3 such thatf(t,l)≥p3(l)−k3(t)lλ3,(t,l)∈[0,T]×[0,c3],f(t,l)≥p4(l)−k4(t)lλ4,(t,l)∈[0,T]×[c4,+∞);
there exists
a>0 such thatmax{f(t,l):(t,l)∈[0,T]×[0,a]}≤(aA0)p−1.
Then, problem (1.1)-(1.2) has at least two positive solutions u3*,u4* satisfying 0<∥u3*∥<a<∥u4*∥.
Proof of Theorem 3.1.
Letϵ=12min[A0p−1−liml→0+¯p1(l)lp−1,A0p−1−liml→∞¯p2(l)lp−1],then there exist 0<a¯1≤c1 and c2≤a¯2<+∞ such thatp1(l)≤(A0p−1−ϵ)lp−1,0≤l≤a¯1,p2(l)≤(A0p−1−ϵ)lp−1,a¯2≤l≤+∞.If 0≤l≤a¯1,u∈∂Ωl,
then 0≤u(t)≤l,0≤t≤T.
By condition (A3),
we havef(t,u(t))≤p1(u(t))+k1(t)uλ1(t)≤(A0p−1−ϵ)up−1(t)+k1(t)uλ1(t)≤(A0p−1−ϵ)∥u∥p−1+k1(t)∥u∥λ1=(A0p−1−ϵ)lp−1+k1(t)lλ1so that∫0sa(τ)f(τ,u(τ))∇τ−A=∫0sa(τ)f(τ,u(τ))∇τ+∑i=1m−2ai∫0ξia(τ)f(τ,u(τ))∇τ1−∑i=1m−2ai≤∫0sa(τ)[(A0p−1−ϵ)lp−1+k1(τ)lλ1]∇τ+∑i=1m−2ai∫0ξia(τ)[(A0p−1−ϵ)lp−1+k1(τ)lλ1]∇τ1−∑i=1m−2ai.Therefore,∥Au∥≤C=11−∑i=1m−2bi(∫0T(T−s)φ(s)∇s−∑i=1m−2bi∫0ξi(ξi−s)φ(s)∇s)≤11−∑i=1m−2bi∫0T(T−s)φ(s)∇s≤11−∑i=1m−2bi×∫0T(T−s)ϕq{∫0sa(τ)[(A0p−1−ϵ)lp−1+k1(τ)lλ1]∇τ+∑i=1m−2ai∫0ξia(τ)[(A0p−1−ϵ)lp−1+k1(τ)lλ1]∇τ1−∑i=1m−2ai}∇s.It follows thatα(l)l≤11−∑i=1m−2bi×∫0T(T−s)ϕq{∫0sa(τ)[A0p−1−ϵ+k1(τ)lλ1−p+1]∇τ+∑i=1m−2ai∫0ξia(τ)[A0p−1−ϵ+k1(τ)lλ1−p+1]∇τ1−∑i=1m−2ai}∇s.Noticing λ1−p+1>0,
we haveliml→0+¯α(l)l≤11−∑i=1m−2bi∫0T(T−s)ϕq[∫0sa(τ)(A0p−1−ϵ)∇τ+∑i=1m−2ai∫0ξia(τ)(A0p−1−ϵ)∇τ1−∑i=1m−2ai]∇s=(A0p−1−ϵ)1/(p−1)1−∑i=1m−2bi∫0T(T−s)ϕq[∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ1−∑i=1m−2ai]∇s=(A0p−1−ϵ)1/(p−1)A0−1=(1−A0−(p−1)ϵ)1/(p−1)<1.Therefore, there exist 0<a1<a¯1 such that α(a1)<a1.
It implies that ∥Au∥<∥u∥,u∈∂Ωa1.
If a¯2≤l<+∞ and u∈∂Ωl,
then 0≤u(t)≤l.
Similar to the above argument, noticing that λ2−p+1<0,
we can get lim¯l→∞(α(l)/l)<1.
Therefore, there exist 0<a¯2<a2 such that α(a2)<a2.
It implies that ∥Au∥<∥u∥,u∈∂Ωa2.
On the other hand, since f:[0,T]×[0,+∞)→[0,+∞) is continuous, by condition (A4),
there exist a1<b1<b<b2<a2 such thatmin{f(t,l):(t,l)∈[0,T]×[γbi,bi]}≥(biB0)p−1,i=1,2.If u∈∂Ωb1,
then γb1≤u(t)≤b1,0≤t≤T.
Applying Lemma 2.9, it follows that∥Au∥=max0≤t≤T|(Au)(t)|≥−∫0T(T−s)φ(s)∇s+11−∑i=1m−2bi(∫0T(T−s)φ(s)∇s−∑i=1m−2bi∫0ξi(ξi−s)φ(s)∇s)=∑i=1m−2bi1−∑i=1m−2bi∫0T(T−s)φ(s)∇s−∑i=1m−2bi∫0ξi(ξi−s)φ(s)∇s1−∑i=1m−2bi≥∑i=1m−2bi1−∑i=1m−2bi∫0T(T−s)φ(s)∇s−∑i=1m−2biξiT(1−∑i=1m−2bi)∫0T(T−s)φ(s)∇s=T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)∫0T(T−s)φ(s)∇s≥T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)∫0T(T−s)ϕq[∫0sa(τ)(b1B0)p−1∇τ+∑i=1m−2ai∫0ξia(τ)(b1B0)p−1∇τ1−∑i=1m−2ai]∇s=T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)b1B0∫0T(T−s)ϕq[∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ1−∑i=1m−2ai]∇s=b1B0B0−1=b1=∥u∥.
In the same way, we
can prove
that if u∈∂Ωb2,
then ∥Au∥≥∥u∥.
Now, we consider the operator A on Ωb1¯∖Ωa1 and Ωa2¯∖Ωb2,
respectively. By Lemma 2.10, we assert that the operator A has two fixed points u1*,u2*∈K such that a1≤∥u1*∥≤b1 and b2≤∥u2*∥≤a2.
Therefore, ui*,i=1,2,
are positive solutions of problem (1.1)-(1.2).
Proof of Theorem 3.2.
Letϵ=12min[liml→0+¯p3(l)lp−1−(B0γ)p−1,liml→∞¯p4(l)lp−1−(B0γ)p−1],then there exist 0<b¯3≤c3 and c4≤b¯4<+∞ such thatp3(l)≥[(B0γ)p−1+ϵ]lp−1,0≤l≤b¯3,p4(l)≥[(B0γ)p−1+ϵ]lp−1,b¯4≤l≤+∞. If 0≤l≤b¯3,u∈∂Ωl,
then γl≤u(t)≤l,0≤t≤T.
By Lemma 2.9 and condition (B3),
we have∥Au∥=max0≤t≤T|(Au)(t)|≥T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)∫0T(T−s)φ(s)∇s≥T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)×∫0T(T−s)ϕq{∫0sa(τ)[p3(u(τ))−k3(τ)uλ3]∇τ+∑i=1m−2ai∫0ξia(τ)[p3(u(τ))−k3(τ)uλ3]∇τ1−∑i=1m−2ai}∇s≥T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)×∫0T(T−s)ϕq{∫0sa(τ)[((B0γ)p−1+ϵ)(γl)p−1−k3(τ)lλ3]∇τ+∑i=1m−2ai∫0ξia(τ)[((B0/γ)p−1+ϵ)(γl)p−1−k3(τ)lλ3]∇τ1−∑i=1m−2ai}∇s.It follows thatβ(l)l≥T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)×∫0T(T−s)ϕq{∫0sa(τ)[B0p−1+γp−1ϵ−k3(τ)lλ3−p+1]∇τ+∑i=1m−2ai∫0ξia(τ)[B0p−1+γp−1ϵ−k3(τ)lλ3−p+1]∇τ1−∑i=1m−2ai}∇s.Noticing λ3−p+1>0,
we getliml→0+¯β(l)l≥T∑i=1m−2bi−∑i=1m−2biξiT(1−∑i=1m−2bi)×∫0T(T−s)ϕq[∫0sa(τ)(B0p−1+γp−1ϵ)∇τ+∑i=1m−2ai∫0ξia(τ)(B0p−1+γp−1ϵ)∇τ1−∑i=1m−2ai]∇s=(B0p−1+γp−1ϵ)1/(p−1)B0−1=(1+γp−1B0−(p−1)ϵ)1/(p−1)>1.Therefore, there
exists
b3 with 0<b3<a such that β(b3)>b3.
It implies that ∥Au∥>∥u∥ for u∈∂Ωb3.
If b¯4≤γl<+∞ and u∈∂Ωl,
then b¯4≤γl≤u(t)≤l,0≤t≤T.
Similar to the above argument, noticing that λ4−p+1<0,
we can get lim̲l→+∞(β(l)/l)>1.
Therefore, there exist
b4 with 0<b4<+∞ such that β(b4)>b4.
It implies that ∥Au∥>∥u∥ for u∈∂Ωb4.
By condition (B4),
we can see that there exist b3<a3<a<a4<b4 such thatmax{f(t,l):(t,l)∈[0,T]×[0,ai]}≤(aiA0)p−1,i=3,4.If u∈∂Ωa3,
then 0≤u(t)≤a3, 0≤t≤T,
and f(t,u(t))≤(a3A0)p−1.
It follows that∥Au∥≤11−∑i=1m−2bi∫0T(T−s)φ(s)∇s≤11−∑i=1m−2bia3A0∫0T(T−s)ϕq[∫0sa(τ)∇τ+∑i=1m−2ai∫0ξia(τ)∇τ1−∑i=1m−2ai]∇s=a3=∥u∥.Similarly, if u∈∂Ωa4,
then ∥Au∥≤∥u∥.
Now, we study the operator A on Ωa3¯∖Ωb3 and Ωb4¯∖Ωa4,
respectively. By Lemma 2.10, we assert that the operator A has two fixed points u3*,u4*∈K such that b3≤∥u3*∥≤a3 and a4≤∥u4*∥≤b4.
Therefore, ui*,i=3,4,
are positive solutions of problem (1.1)-(1.2).
4. Further Discussion
If the
conditions of Theorems 3.1 and
3.2 are weakened, we will get the existence of
single positive solution of problem (1.1)-(1.2).
Corollary 4.1.
Assume(H1), (H2),
and (H3)hold, and assume that the following conditions
hold:
p1∈C([0,+∞),[0,+∞)),
and lim¯l→0+(p1(l)/lp−1)<A0p−1;
k1∈L1([0,T],[0,+∞));
there exist c1>0 and λ1>p−1 such thatf(t,l)≤p1(l)+k1(t)lλ1,(t,l)∈[0,T]×[0,c1];
there exists
b>0 such thatmin{f(t,l):(t,l)∈[0,T]×[γb,b]}≥(bB0)p−1.
Then, problem (1.1)-(1.2) has at least one positive solution.
Corollary 4.2.
Assume(H1), (H2),
and (H3)hold, and assume that the following conditions
hold:
p2∈C([0,+∞),[0,+∞)), and lim¯l→∞(p2(l)/lp−1)<A0p−1;
k2∈L1([0,T],[0,+∞));
there exist c2>0 and 0≤λ2<p−1 such thatf(t,l)≤p2(l)+k2(t)lλ2,(t,l)∈[0,T]×[c2,+∞);
there exists
b>0 such thatmin{f(t,l):(t,l)∈[0,T]×[γb,b]}≥(bB0)p−1.
Then, problem (1.1)-(1.2) has at least one positive solution.
Corollary 4.3.
Assume(H1), (H2),
and (H3)hold, and assume that the following conditions
hold:
p3∈C([0,+∞),[0,+∞)), and lim̲l→0+(p3(l)/lp−1)>(B0/γ)p−1;
k3∈L1([0,T],[0,+∞));
there exist c3>0 and λ3>p−1 such thatf(t,l)≥p3(l)−k3(t)lλ3,(t,l)∈[0,T]×[0,c3];
there exists
a>0 such thatmax{f(t,l):(t,l)∈[0,T]×[0,a]}≤(aA0)p−1.
Then, problem (1.1)-(1.2) has at least one positive solution.
Corollary 4.4.
Assume(H1), (H2),
and (H3)hold, and assume that the following conditions
hold:
p4∈C([0,+∞),[0,+∞)), and lim̲l→∞(p4(l)/lp−1)>(B0/γ)p−1;
k4∈L1([0,T],[0,+∞));
there exist c4>0 and 0≤λ4<p−1 such thatf(t,l)≥p4(l)−k4(t)lλ4,(t,l)∈[0,T]×[c4,+∞);
there exists
a>0 such thatmax{f(t,l):(t,l)∈[0,T]×[0,a]}≤(aA0)p−1.
Then, problem (1.1)-(1.2) has at least one positive solution.
The proof of the above results is similar to those of
Theorems 3.1 and
3.2; thus we omit it.
5. Some Examples
In this
section, we present a simple example to explain our results. We only study the
case T=R,(0,T)=(0,1).
Let f(t,0)≡0.
Consider the following BVP:(ϕ3(u′′))′+f(t,u(t))=0,t∈(0,1),ϕ3(u′′(0))=12ϕ3(u′′(12)),u′(0)=0,u(1)=12u(12), wheref(t,u)={223u3+min{1t(1−t),2u}u5,(t,u)∈[0,1]×[0,1],225,(t,u)∈[0,1]×[1,3],74u+36min{1t(1−t),2u3}u3,(t,u)∈[0,1]×[3,+∞).It is easy to check that f:[0,1]×[0,+∞)→[0,+∞) is continuous. In this case, p=3,a(t)≡1,m=3,a1=b1=1/2,
and ξ1=1/2,
and it follows from a direct calculation thatA0=[11−1/2∫01(1−s)ϕq(s+(1/2)⋅(1/2)1−1/2)ds]−1=1.1062,γ=b1(1−ξ1)1−b1ξ1=(1/2)(1−1/2)1−(1/2)⋅(1/2)=13.We haveB0=[b1−b1ξ11−b1∫0T(1−s)ϕq(s+a1ξ11−a1)ds]−1=[1/2−(1/2)⋅(1/2)1−1/2∫01(1−s)(s+(1/2)⋅(1/2)1−1/2)1/2ds]−1=4.4248<5.Choosing c1=1,c2=3,b=3,λ1=5/2,λ2=3/2,p1(u)=223u3,p2(u)=74u,
and k1(t)=k2(t)=1/t(1−t),
it is easy to check thatf(t,u)≤p1(u)+k1(t)u5/2,(t,u)∈[0,1]×[0,1],f(t,u)≤p2(u)+k2(t)u3/2,(t,u)∈[0,1]×[3,+∞),limu→0¯p1(u)u2=liml→0¯223u3u2=0<A02=(1.1062)2,limu→∞¯p2(u)u2=liml→∞¯74uu2=0<A02=(1.1062)2,min{f(t,u):(t,u)∈[0,1]×[1,3]}=225>(13.274)2=(bB0)2.It follows that f satisfies the conditions (A1)–(A4) of Theorem 3.1; then problem (5.1) has
at least two positive solutions.
Acknowledgment
Project supported by the National Natural Science Foundation of China (Grant
NO. 10471040).
AgarwalR. P.HendersonJ.Positive solutions and nonlinear eigenvalue problems for third-order difference equations19983610–12347355MR1666152ZBL0933.3900310.1016/S0898-1221(98)80035-7AndersonD. R.Green's function for a third-order generalized right focal problem20032881114MR201974010.1016/S0022-247X(03)00132-XZBL1045.34008HaoZ. C.Nonnegative solutions for semilinear third-order difference equation boundary value
problems2001212225229MR1840991ZBL0993.39004HendersonJ.YinW. K. C.Existence of solutions for third-order boundary value problems on a time scale2003456–911011111MR2000581ZBL1057.3901110.1016/S0898-1221(03)00091-9KongL.KongQ.ZhangB.Positive solutions of boundary value problems for third-order functional difference equations2002443-4481489MR1912843ZBL1057.3901410.1016/S0898-1221(02)00170-0LiS.Positive solutions of nonlinear singular third-order two-point boundary value problem20063231413425MR226221510.1016/j.jmaa.2005.10.037ZBL1107.34019LiuZ.UmeJ. S.AndersonD. R.KangS. M.Twin monotone positive solutions to a singular nonlinear third-order differential equation20073341299313MR233255710.1016/j.jmaa.2006.12.067ZBL1123.34016SunY.Positive solutions of singular third-order three-point boundary value problem20053062589603MR213633510.1016/j.jmaa.2004.10.029ZBL1074.34028YangC.WengP.Green functions and positive solutions for boundary value problems of third-order difference equations2007544567578MR2340847ZBL1130.3901210.1016/j.camwa.2007.01.032ZhouC.MaD.Existence and iteration of positive solutions for a generalized right-focal boundary value problem with p-Laplacian operator20063241409424MR2262480ZBL1115.3402910.1016/j.jmaa.2005.10.086AgarwalR. P.O'ReganD.Nonlinear boundary value problems on time scales2001444527535MR1822100ZBL0995.3401610.1016/S0362-546X(99)00290-4AticiF. M.GuseinovG. Sh.On Green's functions and positive solutions for boundary value problems on time scales20021411-27599MR191154810.1016/S0377-0427(01)00437-XZBL1007.34025AulbachB.NeidhartL.Integration on measure chainsProceedings of the 6th International Conference on Difference EquationsJuly-August 2004Augsburg, GermanyCRC239252MR2092560ZBL1083.26005BohnerM.PetersonA.2001Boston, Mass, USABirkhäuserx+358MR1843232ZBL0978.39001BohnerM.PetersonA.2003Boston, Mass, USABirkhäuserxii+348MR1962542ZBL1025.34001AgarwalR. P.BohnerM.ŘehákP.Half-linear dynamic equations2003Dordrecht, The NetherlandsKluwer Academic Publishers157MR2060210ZBL1056.34049GuoD.LakshmikanthamV.19885Boston, Mass, USAAcademic Pressviii+275Notes and Reports in Mathematics in Science and EngineeringMR959889ZBL0661.47045