A fractional boundary value problem is considered. By
means of Banach contraction principle, Leray-Schauder nonlinear alternative,
properties of the Green function, and Guo-Krasnosel'skii fixed point
theorem on cone, some results on the existence, uniqueness, and positivity
of solutions are obtained.
1. Introduction
Fractional differential equations are a natural generalization of ordinary differential equations. In the last few decades many authors pointed out that differential equations of fractional order are suitable for the metallization of various physical phenomena and that they have numerous applications in viscoelasticity, electrochemistry, control and electromagnetic, and so forth, see [1–4].
This work is devoted to the study of the following fractional boundary value problem (P1):cD0+qu(t)=f(t,u(t),cD0+σu(t)),0<t<1,u(0)=u′′(0)=0,u′(1)=αu′′(1),
where f:[0,1]×ℝ×ℝ→ℝ is a given function, 2<q<3, 1<σ<2 and cD0+q denotes the Caputo's fractional derivative. Our results allow the function f to depend on the fractional derivative cD0+σu(t) which leads to extra difficulties. No contributions exist, as far as we know, concerning the existence of positive solutions of the fractional differential equation (1.1) jointly with the nonlocal condition (1.2).
Our mean objective is to investigate the existence, uniqueness, and existence of positive solutions for the fractional boundary value problem (P1), by using Banach contraction principle, Leray-Schauder nonlinear alternative, properties of the Green function and Guo-Krasnosel'skii fixed point theorem on cone.
The research in this area has grown significantly and many papers appeared on this subject, using techniques of nonlinear analysis, see [5–14].
In [6], El-Shahed considered the following nonlinear fractional boundary value problemD0+qu(t)+λa(t)f(u(t))=0,0<t<1,u(0)=u′(0)=u′(1)=0,
where 2<q≤3 and D0+q is the Riemann-Liouville fractional derivative. Using the Krasnoselskii's fixed-point theorem on cone, he proved the existence and nonexistence of positive solutions for the above fractional boundary value problem.
Liang and Zhang in [9] studied the existence and uniqueness of positive solutions by the properties of the Green function, the lower and upper solution method and fixed point theorem for the fractional boundary value problem D0+qu(t)+f(t,u(t))=0,0<t<1,u(0)=u′(0)=0,u′(1)=∑i=1m-2βiu′(ζi),
where 2<q≤3 and D0+q is the Riemann-Liouville fractional derivative.
In [5] Bai and Lü investigated the existence and multiplicity of positive solutions for nonlinear fractional differential equation boundary value problem:D0+qu(t)+f(t,u(t))=0,0<t<1,u(0)=u(1)=0,
where 1<q≤1 and D0+q is the Riemann-Liouville fractional derivative. Applying fixed-point theorems on cone, they prove some existence and multiplicity results of positive solutions.
This paper is organized as follows, in the Section 2 we cite some definitions and lemmas needed in our proofs. Section 3 treats the existence and uniqueness of solution by using Banach contraction principle, Leray Schauder nonlinear alternative. Section 4 is devoted to prove the existence of positive solutions with the help of Guo-Krasnoselskii Theorem, then we give some examples illustrating the previous results.
2. Preliminaries and Lemmas
In this section, we present some lemmas and definitions from fractional calculus theory which will be needed later.
Definition 2.1.
If g∈C([a,b]) and α>0, then the Riemann-Liouville fractional integral is defined by
Ia+αg(t)=1Γ(α)∫atg(s)(t-s)1-αds.
Definition 2.2.
Let α≥0, n=[α]+1. If f∈ACn[a,b] then the Caputo fractional derivative of order α of f defined by
cDa+αg(t)=1Γ(n-α)∫atg(n)(s)(t-s)α-n+1ds,
exists almost everywhere on [a,b] ([α] is the entire part of α).
Lemma 2.3 (see [15]).
Let α,β>0 and n=[α]+1, then the following relations hold: cD0+αtβ-1=(Γ(β)/Γ(β-α))tβ-1, β>n and cD0+αtk=0, k=0,1,2,…,n-1.
Lemma 2.4 (see [15]).
For α>0, g(t)∈C(0,1), the homogenous fractional differential equation
cDa+αg(t)=0
has a solution
g(t)=c1+c2t+c3t2+⋯+cntn-1,
where, ci∈R, i=0,…,n, and n=[α]+1.
Denote by L1([0,1],ℝ) the Banach space of Lebesgue integrable functions from [0,1] into ℝ with the norm ||y||L1=∫01|y(t)|dt.
The following Lemmas gives some properties of Riemann-Liouville fractional integrals and Caputo fractional derivative.
Lemma 2.5 (see [16]).
Let p,q≥0, f∈L1[a,b]. Then I0+pI0+qf(t)=I0+p+qf(t)=I0+qI0+pf(t) and cD0+qI0+qf(t)=f(t), for all t∈a,b].
Lemma 2.6 (see [15]).
Let β>α>0. Then the formula cD0+αI0+βf(t)=I0+β-αf(t), holds almost everywhere on t∈a,b], for f∈L1[a,b] and it is valid at any point x∈[a,b] if f∈C[a,b].
Now we start by solving an auxiliary problem.
Lemma 2.7.
Let 2<q<3, 1<σ<2 and y∈C[0,1]. The unique solution of the fractional boundary value problem
Dc0+qu(t)=y(t),0<t<1,u(0)=u′′(0)=0,u′(1)=αu′′(1)
is given by
u(t)=1Γ(q-2)∫01G(t,s)y(s)ds,
where
G(t,s)={(t-s)q-1q(q-1)+αt(1-s)3-q-t(1-s)q-2q-1,s≤tαt(1-s)3-q-t(1-s)q-2q-1,t≤s.
Proof.
Applying Lemmas 2.4 and 2.5 to (2.5) we get
u(t)=I0+qy(t)+c1+c2t+c3t2.
Differentiating both sides of (2.8) and using Lemma 2.6 it yields
u′(t)=I0+q-1y(t)+c2+c3t,u′′(t)=I0+q-2y(t)+c3.
The first condition in (2.5) implies c1=c3=0, the second one gives c2=αI0+q-2y(1)-I0+q-1y(1). Substituting c2 by its value in (2.8), we obtain
u(t)=I0+qy(t)+t(αI0+q-2y(1)-I0+q-1y(1))
that can be written as
u(t)=1Γ(q-2)∫0t[(t-s)q-1q(q-1)+αt(1-s)3-q-t(1-s)q-2q-1]y(s)ds+1Γ(q-2)∫t1[αt(1-s)3-q-t(1-s)q-2q-1]y(s)ds
that is equivalent to
u(t)=1Γ(q-2)∫01G(t,s)y(s)ds,
where G is defined by (2.7). The proof is complete.
3. Existence and Uniqueness Results
In this section we prove the existence and uniqueness of solutions in the Banach space E of all functions u∈C2[0,1] into ℝ, with the norm ||u||=maxt∈[0,1]|u(t)|+maxt∈[0,1]|cD0+σu(t)|. We know that cD0+σu∈C[0,1], 1<σ<2, see [15]. Denote by E+={u∈E,u(t)≥0,t∈[0,1]}. Throughout this section, we suppose that f∈C([0,1]×ℝ×ℝ,ℝ). Define the integral operator T:E→E byTu(t)=1Γ(q-2)∫01G(t,s)f(s,u(s),cD0+σu(s))ds.
Lemma 3.1.
The function u∈E is solution of the fractional boundary value problem (P1) if and only if Tu(t)=u(t), for all t∈[0,1].
Proof.
Let u be solution of (P1) and v(t)=∫01G(t,s)f(s,u(s),cD0+σu(s))ds. In view of (2.10) we have
v(t)=I0+qf(t,u(t),cD0+σu(t))+αtI0+q-2f(1,u(1),cD0+σu(1))-tI0+q-1f(1,u(1),cD0+σu(1)).
With the help of Lemma 2.6 we obtain
cD0+qv(t)=D0+qcI0+qf(t,u(t),cD0+σu(t))+tαI0+q-2f(1,u(1),cD0+σu(1))-tI0+q-1f(1,u(1),cD0+σu(1))=f(t,u(t),cD0+σu(t)).
It is clear that v satisfies conditions (1.2), then it is a solution for the problem (P1). The proof is complete.
Theorem 3.2.
Assume that there exist nonnegative functions g,h∈L1([0,1],ℝ+) such that for all x,y∈ℝ and t∈[0,1], one has
|f(t,x,x¯)-f(t,y,y¯)|≤g(t)|x-y|+h(t)|x¯-y¯|,Cg+Ch<1,Ag+Ah<(2-σ)Γ(2-σ),
where
Cg=‖I0+q-1g‖L1+|α|I0+q-2g(1)+I0+q-1g(1),Ag=2I0+q-1g(1)+|α|I0+q-2g(1),Ch=‖I0+q-1h‖L1+|α|I0+q-2h(1)+I0+q-1h(1),Ah=2I0+q-1h(1)+|α|I0+q-2h(1).
Then the fractional boundary value problem (P1) has a unique solution u in E.
To prove Theorem 3.2, we use the following property of Riemann-Liouville fractional integrals.
Lemma 3.3.
Let q>0, f∈L1([a,b],ℝ+). Then, for all t∈a,b] we have
I0+q+1f(t)≤‖I0+qf‖L1.
Proof.
Let f∈L1([a,b],ℝ+), then
‖I0+qf‖L1=∫01I0+qf(r)dr≥1Γ(q)∫at∫arf(s)(r-s)1-qdsdr=1Γ(q)∫at(∫stf(s)(r-s)1-qdr)ds=I0+q+1f(t).
Now we prove Theorem 3.2.
Proof.
We transform the fractional boundary value problem to a fixed point problem. By Lemma 3.1, the fractional boundary value problem (P1) has a solution if and only if the operator T has a fixed point in E. Now we will prove that T is a contraction. Let u,v∈E, in view of (2.10) we get
Tu(t)-Tv(t)=1Γ(q-2)∫01G(t,s)(f(s,u(s),cD0+σu(s))-f(s,v(s),cD0+σv(s)))ds=I0+q(f(t,u(t),cD0+σu(t))-f(s,v(t),cD0+σv(t)))+tα[I0+q-2(f(1,u(1),cD0+σu(1))-f(1,v(1),cD0+σv(1)))-tI0+q-1(f(1,u(1),cD0+σu(1))-f(1,v(1),cD0+σv(1)))],
with the help of (3.4) we obtain
|Tu(t)-Tv(t)|≤max|u(t)-v(t)|(I0+qg(t)+|α|I0+q-2g(1)+I0+q-1g(1))+max|D0+σcu(t)-cD0+σv(t)|(I0+qh(t)+|α|I0+q-2h(1)+I0+q-1h(1)).
Lemma 3.3 implies
|Tu(t)-Tv(t)|≤‖u-v‖[‖I0+q-1g‖L1+|α|I0+q-2g(1)+I0+q-1g(1)+‖I0+q-1h‖L1+|α|I0+q-2h(1)+I0+q-1h(1)]=‖u-v‖(Cg+Ch).
In view of (3.5) it yields
|Tu-Tv|<‖u-v‖.
On the other hand we have
cD0+σTu-cD0+σTv=1Γ(2-σ)∫0t(Tu)′(s)-(Tv)′(s)(t-s)σ-1ds,
where
(Tu)′(t)=1Γ(q-2)∫01G1(t,s)f(s,u(s),cD0+σu(s))ds,G1(t,s)=∂G(t,s)∂t={(t-s)q-2q+α(1-s)3-q-(1-s)q-2q-1,0≤s≤t≤1α(1-s)3-q-(1-s)q-2q-1,0≤t≤s≤1.
Therefore
cD0+σTu-cD0+σTv=1Γ(q-2)Γ(2-σ)∫0t∫01(t-s)-σ+1G1(s,r)×(f(r,u(r),cD0+σu(r))-f(r,v(r),cD0+σv(r)))drds.
Applying hypothesis (3.4) we get
|cD0+σTu-cD0+σTv|≤max|u-v|Γ(q-2)Γ(2-σ)∫0t∫01(t-s)-σ+1|G1(s,r)|g(r)drds+max|cD0+σu-cD0+σv|Γ(q-2)Γ(2-σ)∫0t∫01(t-s)-σ+1|G1(s,r)|h(r)drds.
Let us estimate the term ∫01(∂G(s,r)/∂s)g(r)dr. We have
∫01|G1(s,r)|g(r)dr≤Γ(q-2)(2(q-2)I0+q-1g(1)q-1+|α|I0+q-2g(1))=Γ(q-2)(2I0+q-1g(1)+|α|I0+q-2g(1))=Γ(q-2)Ag.
Consequently (3.16) becomes
|cD0+σTu-cD0+σTv|≤‖u-v‖1(2-σ)Γ(2-σ)(Ag+Ah).
With the help of hypothesis (3.5) it yields
|cD0+σTu-cD0+σTv|≤‖u-v‖.
Taking into account (3.12)–(3.19) we obtain
‖Tu-Tv‖<‖u-v‖,
from here, the contraction principle ensures the uniqueness of solution for the fractional boundary value problem (P1). This finishes the proof.
Now we give an existence result for the fractional boundary value problem (P1).
Theorem 3.4.
Assume that f(t,0,0)≠0 and there exist nonnegative functions k,h,g∈L1([0,1],ℝ+), ϕ,ψ∈C(ℝ+,ℝ+*) nondecreasing on ℝ+ and r>0, such that
|f(t,x,x¯)|≤k(t)ψ(|x|)+h(t)ϕ(|x¯|)+g(t),a.e.(t,x)∈[0,1]×R.(ψ(r)+ϕ(r)+1)(C1Γ(q-2)+C2(2-σ)Γ(2-σ))<r,
where C1=max{Ck,Ch,Cg}, C2=max{Ak,Ah,Ag}, Ch and Cg are defined as in Theorem 3.2 and
Ck=‖I0+q-1k‖L1+|α|I0+q-2k(1)+I0+q-1k(1),Ak=2I0+q-1k(1)+|α|I0+q-2k(1).
Then the fractional boundary value problem (P1) has at least one nontrivial solution u*∈E.
To prove this Theorem, we apply Leray-Schauder nonlinear alternative.
Lemma 3.5 (see [17]).
Let F be a Banach space and Ω a bounded open subset of F, 0∈Ω. T:Ω¯→F be a completely continuous operator. Then, either there exists x∈∂Ω, λ>1 such that T(x)=λx, or there exists a fixed point x*∈Ω¯.
Proof.
First let us prove that T is completely continuous. It is clear that T is continuous since f and G are continuous. Let Br={u∈E,∥u∥≤r} be a bounded subset in E. We shall prove that T(Br) is relatively compact.
(i) For u∈Br and using (3.21) we get
|Tu(t)|≤1Γ(q-2)∫01|G(t,s)|[k(s)ψ(|u(s)|)+h(s)ϕ(|cD0+σu(s)|)+g(s)]ds.
Since ψ and ϕ are nondecreasing then (3.24) implies
|Tu(t)|≤1Γ(q-2)∫01|G(t,s)|[k(s)ψ(‖u‖)+h(s)ϕ(‖u‖)+g(s)]ds≤1Γ(q-2)∫01|G(t,s)|[k(s)ψ(r)+h(s)ϕ(r)+g(s)]ds,
using similar techniques as to get (3.12) it yields
|Tu(t)|≤1Γ(q-2)[ψ(r)(‖I0+q-1k‖L1+|α|I0+q-2k(1)+I0+q-1k(1))+ϕ(r)(‖I0+q-1h‖L1+|α|I0+q-2h(1)+I0+q-1h(1))+(‖I0+q-1g‖L1+|α|I0+q-2g(1)+I0+q-1g(1))]=1Γ(q-2)(Ckψ(r)+Chϕ(r)+Cg).
Hence
|Tu(t)|≤C1Γ(q-2)[ψ(r)+ϕ(r)+1].
Moreover, we have
|(Tu(t))′|≤1Γ(q-2)[ψ(r)∫01|∂G∂t(t,s)|k(s)ds+ϕ(r)∫01|∂G∂t(t,s)|h(s)+∫01|∂G∂t(t,s)|g(s)ds].
Using (3.17) we obtain
|cD0+σTu|≤C2(2-σ)Γ(2-σ)(ψ(r)+ϕ(r)+1).
From (3.27) and (3.29) we get
‖Tu‖=(ψ(r)+ϕ(r)+1)(C1Γ(q-2)+C2(2-σ)Γ(2-σ)),
then T(Br) is uniformly bounded.
(ii) T(Br) is equicontinuous. Indeed for all t1, t2∈[0,1], t1<t2, u∈Br, let C=max(|f(t,u(t),cD0+σu(t))|, 0≤t≤1,∥u∥<r), therefore
|Tu(t1)-Tu(t2)|≤CΓ(q-2)(∫0t1|G(t1,s)-G(t2,s)|ds+∫t1t2|G(t1,s)-G(t2,s)|ds+∫t21|G(t1,s)-G(t2,s)|ds),
that implies
|Tu(t1)-Tu(t2)|≤CΓ(q-2)∫0t1(t2-s)q-1-(t1-s)q-1q(q-1)+(t2-t1)(|α|(1-s)3-q-(1-s)q-2q-1)ds+∫t1t2(t2-s)q-1q(q-1)+(t2-t1)(|α|(1-s)3-q-(1-s)q-2q-1)ds+∫t21(t1-t2)(|α|(1-s)3-q-(1-s)q-2q-1)ds.
Let us consider the function Φ(x)=xq-1-(q-1)x, we see that Φ is decreasing on [0,1], consequently (t2-s)q-1-(t1-s)q-1≤(q-1)(t2-t1), from which we deduce
|Tu(t1)-Tu(t2)|≤CΓ(q-2)[(t2-t1)∫0t11q(q-1)+|α(1-s)3-q-(1-s)q-2q-1|ds+∫t1t2(t2-s)q-1q(q-1)+(t2-t1)||α|(1-s)3-q-(1-s)q-2q-1|ds+(t1-t2)∫t21||α|(1-s)3-q-(1-s)q-2q-1|ds].
Some computations give
|Tu(t1)-Tu(t2)|≤C(t2-t1)Γ(q-2)(1q(q-1)+3|α|(q-2)+3q-1)+CΓ(q-2)∫t1t2(t2-s)q-1q(q-1)ds.
On the other hand we have
|cD0+σTu(t1)-cD0+σTu(t2)|≤+1Γ(2-σ)∫0t1((t1-s)-σ+1-(t2-s)-σ+1)|(Tu(s))′|ds+1Γ(2-σ)∫t1t2(t2-s)-σ+1|(Tu(s))′|ds.
Using (3.17) and (3.28) it yields
|(Tu(t))′|≤[ψ(r)+ϕ(r)+1]C2,
then
|cD0+σTu(t1)-cD0+σTu(t2)|≤[ψ(r)+ϕ(r)+1]C2(2-σ)Γ(2-σ)[2(t2-t1)2-σ+t22-σ-t12-σ],
when t1→t2, in (3.34) and (3.37) then |Tu(t1)-Tu(t2)| and |cD0+σTu(t1)-cD0+σTu(t2)| tend to 0. Consequently T(Br) is equicontinuous. From Arzelá-Ascoli Theorem we deduce that T is completely continuous operator.
Now we apply Leray Schauder nonlinear alternative to prove that T has at least one nontrivial solution in E.
Letting Ω={u∈E:∥u∥<r}, for any u∈∂Ω, such that u=λTu, 0<λ<1, we get, with the help of (3.27),
|u(t)|=λ|Tu(t)|≤|Tu(t)|≤C1Γ(q-2)[ψ(r)+ϕ(r)+1].
Taking into account (3.29) we obtain
|cD0+σu(t)|≤C2(2-σ)Γ(2-σ)(ψ(r)+ϕ(r)+1).
From (3.38), (3.39), and (3.22) we deduce that
‖u‖≤(ψ(r)+ϕ(r)+1)(C1Γ(q-2)+C2(2-σ)Γ(2-σ))<r,
this contradicts the fact that u∈∂Ω. Lemma 3.5 allows us to conclude that the operator T has a fixed point u*∈Ω¯ and then the fractional boundary value problem (P1) has a nontrivial solution u*∈E. The proof is complete.
4. Existence of Positive Solutions
In this section we investigate the positivity of solution for the fractional boundary value problem (P1), for this we make the following hypotheses.
f(t,u,v)=a(t)f1(u,v) where a∈C((0,1),(0,∞)) and f1∈C(ℝ+×ℝ,ℝ+).
0<∫01G(s,s)a(s)ds<∞.
Now we give the properties of the Green function.
Lemma 4.1.
Let G(t,s) be the function defined by (2.7). If α≥1 then G(t,s) has the following properties:
G(t,s)∈C([0,1]×[0,1]), G(t,s)>0, for all t, s∈]0,1[.
If t,s∈(τ,1), τ>0, then
0<τG(s,s)≤G(t,s)≤2τG(s,s).
Proof.
(i) It is obvious that G(t,s)∈C([0,1]×[0,1]), moreover, we have
αt(1-s)3-q-t(1-s)q-2q-1=tq-1(1-s)3-q[(q-1)α-(1-s)]≥tq-1(1-s)3-q[α-1+s],
which is positive if α≥1. Hence G(t,s) is nonnegative for all t,s∈]0,1[.
(ii) Let t,s∈(τ,1), it is easy to see that G(s,s)≠0, then we have
G(t,s)G(s,s)=(t-s)q-1(1-s)3-qqs[(q-1)α-(1-s)]+ts≤1+(1-s)2s≤2τ,0<τ≤s≤t<1,G(t,s)G(s,s)=ts≤2τ,0<τ≤t≤s≤1.
Now we look for lower bounds of G(t,s)G(t,s)G(s,s)≥ts≥τs≥τ,0<τ≤s≤t<1,0<τ≤t≤s≤1.
Finally, since G(s,s) is nonnegative we obtain 0<τG(s,s)≤G(t,s)≤(2/τ)G(s,s).
We recall the definition of positive of solution.
Definition 4.2.
A function u(t) is called positive solution of the fractional boundary value problem (P1) if u(t)≥0, for all t∈[0,1].
Lemma 4.3.
If u∈E+ and α≥1, then the solution of the fractional boundary value problem (P1) is positive and satisfies
mint∈(τ,1)(u(t)+cD0+σu(t))≥τ22‖u‖.
Proof.
First let us remark that under the assumptions on u and f, the function cD0+σu is nonnegative. From Lemma 3.1 we have
u(t)=1Γ(q-2)∫01G(t,s)a(s)f1(u(s),cD0+σu(s))ds.
Applying the right-hand side of inequality (4.1) we get
u(t)≤2τΓ(q-2)∫01G(s,s)a(s)f1(u(s),cD0+σu(s))ds.
Moreover, (4.1) gives
cD0+σu(t)=1Γ(q-2)Γ(2-σ)∫0t∫01(t-s)-σ+1G1(s,r)×a(r)f1(u(r),cD0+σu(r))dsdr≤1τ(2-σ)Γ(q-2)Γ(2-σ)∫01G1(r,r)a(r)f1(u(r),cD0+σu(r))dr.
Combining (4.7) and (4.8) yields
‖u‖≤2τΓ(q-2)∫01[G(s,s)+G1(s,s)(2-σ)Γ(2-σ)]a(s)f1(u(s),cD0+σu(s))ds,
hence
∫01[G(s,s)+G1(s,s)(2-σ)Γ(2-σ)]a(s)f1(u(s),cD0+σu(s))ds≥τΓ(q-2)2‖u‖.
In view of the-left hand side of (4.1), we obtain for all t∈(τ,1)u(t)≥τΓ(q-2)∫01G(s,s)a(s)f1(u(s),cD0+σu(s))ds,
on the other hand we have
cD0+σu(t)≥τ2-σ(2-σ)Γ(q-2)Γ(2-σ)∫01G1(r,r)a(r)f1(u(r),cD0+σu(r))dr.
From (4.11) and (4.12) we get
mint∈(τ,1)(u(t)+cD0+σu(t))≥τΓ(q-2)∫01[G(s,s)+G1(s,s)(2-σ)Γ(2-σ)]×a(s)f1(u(s),cD0+σu(s))ds,
with the help of (4.10) we deduce
mint∈(τ,1)(u(t)+cD0+σu(t))≥τ22‖u‖.
The proof is complete.
Define the quantities A0 and A∞ by A0=lim(|u|+|v|)→0f1(u,v)|u|+|v|,A∞=lim(|u|+|v|)→∞f1(u,v)|u|+|v|.
The case A0=0 and A∞=∞ is called superlinear case and the case A0=∞ and A∞=0 is called sublinear case.
The main result of this section is as follows.
Theorem 4.4.
Under the assumption of Lemma 4.3, the fractional boundary value problem (P1) has at least one positive solution in the both cases superlinear as well as sublinear.
To prove Theorem 4.4 we apply the well-known Guo-Krasnosel'skii fixed point theorem on cone.
Theorem 4.5 (see [18]).
Let E be a Banach space, and let K⊂E, be a cone. Assume Ω1 and Ω2 are open subsets of E with 0∈Ω1, Ω1¯⊂Ω2 and let
A:K∩(Ω2¯∖Ω1)⟶K
be a completely continuous operator such that
||𝒜u||≤||u||, u∈K∩∂Ω1, and ||𝒜u||≥||u||, u∈K∩∂Ω2, or
||𝒜u||≥||u||, u∈K∩∂Ω1, and ||𝒜u||≤||u||, u∈K∩∂Ω2.
Then 𝒜 has a fixed point in K∩(Ω2¯∖Ω1).
Proof.
To prove Theorem 4.4 we define the cone K by
K={u∈E+,mint∈(τ,1)(u(t)+cD0+σu(t))≥τ22‖u‖}.
It is easy to check that K is a nonempty closed and convex subset of E, hence it is a cone. Using Lemma 4.3 we see that TK⊂K. From the prove of Theorem 3.4, we know that T is completely continuous in E.
Let us prove the superlinear case. First, since A0=0, for any ε>0, there exists R1>0, such that
f1(u,v)≤ε(|u|+|v|)
for 0<|u|+|v|≤R1. Letting Ω1={u∈E,∥u∥<R1}, for any u∈K∩∂Ω1, it yields
Tu(t)=1Γ(q-2)∫01G(t,s)a(s)f1(u(s),cD0+σu(s))ds≤2ε‖u‖τΓ(q-2)∫01G(s,s)a(s)ds.
Moreover, we have
cD0+σTu(t)=1Γ(q-2)Γ(2-σ)∫0t∫01(t-s)-σ+1G1(s,r)×a(r)f1(u(r),cD0+σu(r))dsdr≤1τ(2-σ)Γ(q-2)Γ(2-σ)∫01G1(r,r)a(r)(|u(r)|+|cD0+σu(r)|)dr≤ε‖u‖τ(2-σ)Γ(q-2)Γ(2-σ)∫01G1(s,s)a(s)ds.
From (4.19) and (4.20) we conclude
‖Tu‖≤2ε‖u‖τΓ(q-2)∫01[G(s,s)+G1(s,s)(2-σ)Γ(2-σ)]a(s)ds.
In view of hypothesis (H2), one can choose ε such that
ε≤τΓ(q-2)2∫01[G(s,s)+(G1(s,s)/(2-σ)Γ(2-σ))]a(s)ds.
The inequalities (4.21) and (4.22) imply that ||Tu||≤||u||, for all u∈K∩∂Ω1. Second, in view of A∞=∞, then for any M>0, there exists R2>0, such that f1(u,v)≥M(|u|+|v|) for |u|+|v|≥R2. Let R=max{2R1,(2R2/τ2)} and denote by Ω2 the open set {u∈E/||u||<R}. If u∈K∩∂Ω2 then
mint∈(τ,1)(u(t)+cD0+σu(t))≥τ22‖u‖=τ22R≥R2.
Using the left-hand side of (4.1) and Lemma 4.3, we obtain
Tu(t)≥τMΓ(q-2)∫01G(s,s)a(s)(|u(s)|+|cD0+σu(s)|)ds≥τ3M||u||2Γ(q-2)∫01G(s,s)a(s)ds.
Moreover, we get with the help of (4.12)
cD0+σTu(t)≥τ4-σM||u||2(2-σ)Γ(q-2)Γ(2-σ)∫01G1(s,s)a(s)ds.
In view of (4.26) and (4.24) we can write
Tu(t)+cD0+σTu(t)≥τ3M||u||2Γ(q-2)∫01[G(s,s)+τ1-σG1(s,s)(2-σ)Γ(2-σ)]a(s)ds≥τ3M||u||2Γ(q-2)∫01[G(s,s)+G1(s,s)(2-σ)Γ(2-σ)]a(s)ds.
Let us choose M such that
M≥2Γ(q-2)τ3∫01[G(s,s)+(G1(s,s)/(2-σ)Γ(2-σ))]a(s)ds,
then we get Tu(t)+cD0+σTu(t)≥||u||. Hence,
||Tu||≥||u||,∀u∈K∩∂Ω2.
The first part of Theorem 4.5 implies that T has a fixed point in K∩(Ω̅2∖Ω1) such that R2≤||u||≤R. To prove the sublinear case we apply similar techniques. The proof is complete.
In order to illustrate our results, we give the following examples.
Example 4.6.
The fractional boundary value problem
cD0+5/2u=(t-110)3u+t2D0+5/4u+lnt,0<t<1,u(0)=u′′(0)=0,u′(1)=-12u′′(1)
has a unique solution in E.
Proof.
In this case we have f(t,x,y)=((t-1)/10)3x+t2y+lnt, 2<q=5/2<3, σ=(5/4)<2, α=-1/2 and
|f(t,x,x¯)-f(t,y,y¯)|≤(1-t10)3|x-x¯|+t2|y-y¯|,
then g(t)=((1-t)/10)3 and h(t)=t2. Some calculus give
‖I0+q-1g‖L1=0.17846×10-3,I0+q-1g(1)=0.48001×10-3,I0+q-2g(1)=016120×10-3,Cg=7.3907×10-4,Ag=1.0406×10-3,‖I0+q-1h‖L1=0.038210,I0+q-1h(1)=0.17194,I0+q-2h(1)=0.6018,Ch=0.49747,Ah=0.42448,Cg+Ch=0.49821<1,Ag+Ah=0.42552<(2-σ)Γ(2-σ)=0.91906.
Thus Theorem 3.2 implies that fractional boundary value problem (4.29) has a unique in E.
Example 4.7.
The fractional boundary value problem
Dc0+7/3u=(1-t)2(u4100(1+u2)+ln(1+(cD0+6/5u)2)+19)=0,0<t<1,u(0)=u′′(0)=0,u′(1)=32u′′(1)
has at least one nontrivial solution in E.
Proof.
We apply Theorem 3.4 to prove that the fractional boundary value problem (4.32) has at least one nontrivial solution. We have q=7/3, σ=6/5, α=3/2, and
|f(t,x,x¯)|=(1-t)2x4100(1+x2)+(1-t)2ln(1+x2¯)9+(1-t)2≤|x|2100(1-t)2+(1-t)2ln(1+x2¯)9+(1-t)2≤k(t)ψ(|x|)+h(t)ϕ(|x¯|)+g(t),
where k(t)=h(t)=g(t)=(1-t)2, ψ(x)=(x/10)2, ϕ(x¯)=ln(1+x2¯)/9, f(t,0,0)≠0. Let us find r such that (3.22) holds, for this we have
‖I0+q-1g‖L1=0.19382,I0+q-2g(1)=0.15998,I0+q-1g(1)=0.33595,Cg=C1=0.76974,Ag=C2=0.91187.
We see that (3.22) is equivalent to 1.2664(((r2)/100)+(ln(1+r2)/9)+1)-r which is negative for r=6.
BagleyR. L.TorvikP. J.A theoretical basis for the application of fractional calculus to viscoelasticity198327320121010.1122/1.549724HilferR.2000River Edge, NJ, USAWorld Scientificviii+46310.1142/97898128177471890104OldhamK. B.KOldham@TrentU.caFractional differential equations in electrochemistry201041191210.1016/j.advengsoft.2008.12.012PodlubnyI.1999198San Diego, Calif, USAAcademic Pressxxiv+340Mathematics in Science and Engineering1658022BaiZ.LüH.Positive solutions for boundary value problem of nonlinear fractional differential equation20053112495505216841310.1016/j.jmaa.2005.02.052ZBL1079.34048El-ShahedM.Positive solutions for boundary value problem of nonlinear fractional differential equation2007200781036810.1155/2011/3905432353784ZBL1149.26012Guezane-LakoudA.KelaiaiaS.Solvability of a three-point nonlinear boundary-value problem20102010139192729460ZBL1206.34033JafariH.Daftardar-GejjiV.Positive solutions of nonlinear fractional boundary value problems using Adomian decomposition method20061802700706227004710.1016/j.amc.2006.01.007ZBL1102.65136LiangS.liangsihua@163.comZhangJ.jihuiz@jlonline.comExistence and uniqueness of positive solutions to m-point boundary value problem for nonlinear fractional differential equationJournal of Applied Mathematics and Computing. In press10.1007/s12190-011-0475-2QiuT.qiutingting19833@163.comBaiZ.zhanbingbai@163.comExistence of positive solutions for singular fractional differential equations20082008, article 146SuX.LiuL.Existence of solution for boundary value problem of nonlinear fractional differential equation2007223291298235106810.1007/s11766-007-0306-2ZBL1150.34005XuX.JiangD.YuanC.Multiple positive solutions for the boundary value problem of a nonlinear fractional differential equation200971104676468810.1016/j.na.2009.03.0302548701ZBL1178.34006ZhaoY.SunS.HanZ.ZhangM.Positive solutions for boundary value problems of nonlinear fractional differential equations20112171669506958277568510.1016/j.amc.2011.01.103ZhaoY.SunS.HanZ.LiQ.The existence of multiple positive solutions for boundary value problems of nonlinear fractional differential equations201116420862097273607410.1016/j.cnsns.2010.08.017KilbasA. A.SrivastavaH. M.TrujilloJ. J.2006204Amsterdam, The NetherlandsElsevierxvi+523North-Holland Mathematics Studies2218073AnastassiouG. A.On right fractional calculus2009421365376254305510.1016/j.chaos.2008.12.013ZBL1198.26006DeimlingK.1985Berlin, GermanySpringerxiv+450787404GuoD. J.LakshmikanthamV.19885Boston, Mass, USAAcademic Pressviii+275Notes and Reports in Mathematics in Science and Engineering959889