Abstract
Motivated by some recent developments in the existence theory of fractional difference equations, in this paper we consider boundary value problem
where \(t\in [0,N+1]_{\mathbb {N}_0} \) and N (\(N\ge 2\)) is an integer. The nonlinear function \(f:[\nu -1,\nu +N]_{\mathbb {N}_{\nu -1}}\times \mathbb {R}\rightarrow \mathbb {R^+}\) is assumed to be continuous. We establish some useful inequalities satisfied by the Green’s function associated with above boundary value problem. Sufficient conditions are developed to ensure the existence and nonexistence of positive solutions for the boundary value problem.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In recent years the continuous fractional calculus has seen tremendous growth due to the fact that many problems in science and engineering can be modeled by fractional differential equations. In many situations, the mathematical models based on fractional operators provide more suitable results than analogous models involving classical integer order depravities. The qualitative theory of fractional differential equations has been studied comprehensively in [1,2,3,4] and references therein. However quite recently the discrete fractional calculus has gained a great deal of interest by several researchers. Miller and Ross [5] initiated the study of discrete fractional calculus. Recently, important developments in this direction has been reported by Atici and Eloe [6,7,8], Holm [9, 10], Abdeljawad [11], Goodrich [12], Jia et al. [13], Goodrich and Peterson [14].
There are few papers dealing with the existence of positive solutions for discrete fractional boundary value problems. In [15], Atici and Eloe obtained sufficient conditions for the existence of positive solutions of the following two-point boundary value problem for a finite fractional difference equation
where \(1 < \nu \le 2\) is a real number and, \(b \ge 2\) an integer and \(f:[\nu -1,\nu +b-1]_{\mathbb {N}_{\nu -1}}\times \mathbb {R}\rightarrow \mathbb {R}\) is continuous.
Goodrich [16] considered a discrete fractional boundary value problem of the form
where \(f:[\nu -1,\nu +b-1]_{\mathbb {N}_{\nu -1}}\times \mathbb {R}\rightarrow \mathbb {R}\) is continuous, \(g : C([\nu -2,\nu +b]_{\mathbb {N}_{\nu -2}} , \mathbb {R})\) is a given functional. He proved the existence and uniqueness of solutions using tools from nonlinear functional analysis.
Goodrich [17] investigated a nonlinear discrete fractional boundary value problem given by
where \(t \in [0, b + 1]_{\mathbb {N}_0}\) , \(f : [\nu - 1, \nu + b]_{\mathbb {N}_{\nu -1}} \times \mathbb {R} \rightarrow \mathbb {R}\), and \(b \in \mathbb {N}_0\). The Green’s function for this problem is studied and some new results are obtained for \(1< \nu < 2\).
Inspired by the above cited works, we consider the following class of two-point boundary value problem for fractional difference equations
where \(1 < \nu \le 2\) is a real number and, \(N \ge 2\) an integer and \(f:[\nu -1,\nu +N]_{\mathbb {N}_{\nu -1}}\times \mathbb {R}\rightarrow \mathbb {R}\) is continuous. This boundary value problem is similar to the problem discussed in [17]. The difference here is that the boundary condition at \(\nu +N\) involves a difference of order \(\nu -1\) rather than order one. We obtain a different Green’s function as compared to one obtained in [17]. We shall establish an equivalent summation representation of above problem and obtain various existence and non-existence results for positive solutions. Furthermore, for eigenvalue problem, intervals for parameter (eigenvalues) are obtained for which there exist positive solutions or no positive solution.
Rest of the paper is organized as follows: In Sect. 2, we shall list some basic definitions and properties of discrete fractional operators. In Sect. 3 we shall obtain an equivalent summation representation of the boundary value problem (1). Some useful inequality will be established for the Green’s function. Finally in Sect. 4 we shall use tools from functional analysis to establish several existence results for positive solutions.
2 Preliminaries
In this section we review some basic definitions and properties of discrete fractional operators. For details, we refer the reader to [14].
Definition 1
For \(\nu >0\), the \(\nu \)-th fractional sum of a function \(u:\mathbb {N}_a\rightarrow \mathbb {R}\) is defined as
for \(t\in \mathbb {N}_{a+\nu }:=\{a+\nu ,a+\nu +1,\ldots \}\). Also the fractional difference of order \(\nu >0\) is defined by \(\Delta _a^{\nu }u(t):=\Delta ^{n}\Delta _a^{\nu -n}u(t)\) where \( n-1<\nu \le n\) with \(n\in \mathbb {N}\) and \(t\in N_{a+n-\nu }\). The fractional difference operator \(\Delta _a^{\nu }\) maps functions defined on \(\mathbb {N}_a\) to functions defined on \(\mathbb {N}_{a+n-\nu }\).
Lemma 1
Assume \(\mu ,\nu >0\). Then following properties hold for fractional sum and difference:
-
(i)
\(\Delta t^{(\nu )}=\nu t^{(\nu -1)}\),
-
(ii)
\(\nu ^{(\nu )}=\Gamma (\nu +1)\),
-
(iii)
\(\Delta _{a+\mu }^{-\nu }(t-a)^{(\mu )}=\frac{\Gamma (\mu +1)}{\Gamma (\mu +\nu +1)}(t-a)^{(\mu +\nu )}\) and
-
(iv)
\(\Delta _{a+\mu }^{\nu }(t-a)^{(\mu )}=\frac{\Gamma (\mu +1)}{\Gamma (\mu -\nu +1)}(t-a)^{(\mu -\nu )}\),
whenever expressions in (i)–(iv) are well defined.
Lemma 2
[14] Assume u be a real valued function and \(\mu ,\nu >0\). Then
for all \(t\in \mathbb {N}_{\nu +\mu }.\)
Lemma 3
[9] Let \(u : \mathbb {N}_a \rightarrow \mathbb {R}\) and \(n - 1 < \nu \le n\). Consider the \(\nu \)-th-order discrete fractional equation
and the corresponding discrete fractional initial value problem
Then the general solution of (2) is given by
for some \(c_i\in \mathbb {R}\), \(i=1,2,\ldots ,n\), \(t\in \mathbb {N}_{a+\nu -n}\), and the unique solution to (3) is given by
3 Green’s function and its properties
To establish existence theorems, the boundary value problem for the fractional difference equation is reduced to fractional summation equation. This is standard practice in existence theory of fractional difference equations. In the following lemma we reduce the boundary value problem (1) to an equivalent summation equation.
Lemma 4
Let \(1<\nu \le 2\) and \(h\in C ([\nu -1,\nu +N]_{\mathbb {N}_{\nu -1}})\). Then a function u is solution of discrete fractional boundary value problem
if and only if u(t), for \(t\in [\nu -2,\nu +N]_{\mathbb {N}_{\nu -2}}\) is solution of
where
Proof
In view of Lemma 3, the general solution of the fractional difference equation \(\Delta _{\nu -2}^{\nu }u(t)+h(t+\nu -1)=0\) is given by
where \(t\in [\nu -1,\nu +N]_{\mathbb {N}_{\nu -1}}\) and \(c_1, c_2\in \mathbb {R}\). Now applying the boundary condition \(u(\nu -2)=0,\) we immediately get \(c_2=0\). Applying \(\Delta _{\nu -1}^{\nu -1}\) on both sides of equation (5) and by using Lemmas 1 and 2, we arrive at
Using the boundary condition \(\Delta _{\nu -1}^{\nu -1}u(\nu +N)=0\) in Eq. (6) we have
Since \(\nu -1\le 1\), therefore
Substituting (7) in Eq. (5) we get
\(\square \)
Lemma 5
The Green’s function G(t, s) in (4) satisfies the following properties:
-
(i)
\(G(t,s)>0\) for \(t\in [\nu -1,\nu +N]_{\mathbb {N}_{\nu -1}}\) and \(s\in [0,N+1]_{\mathbb {N}_0},\)
-
(ii)
\(\max _{t\in [\nu -1,\nu +N]_{\mathbb {N}_{\nu -1}}}G(t,s)=G(s+\nu ,s)\) and
-
(iii)
\(\min _{t\in \left[ \frac{N+\nu }{4}, \frac{3(N+\nu )}{4}\right] _{\mathbb {N}_{\nu -1}}}G(t,s) \ge \gamma \max _{t\in [\nu -1,\nu +N]_{\mathbb {N}_{\nu -1}}}G(t,s)\) for some \(\gamma \in (0,1), s\in [0,N+1]_{\mathbb {N}_0}\).
Proof
-
(i)
For \(t-\nu <s\le N+1\), clearly \(G(t,s)>0.\) Since \(t^{(q)}\) is increasing function for \(0<q<1,\) therefore \((t-s-1)^{(\nu -1)}< t^{(\nu -1)}\). Hence \(G(t,s)>0\) for \(0<s< t-\nu \le N+1\).
-
(ii)
For \(t-\nu <s\le N+1\), obviously \(\Gamma (\nu )\Delta _tG(t,s)=(\nu -1)t^{(\nu -2)}>0\) which implies that G is increasing for \(0\le t\le s+\nu \). Furthermore \(t^{(q)}\) is a decreasing function for \(q\in (-1,0)\), which implies \(\Delta _tG(t,s)=(\nu -1)[t^{(\nu -2)}-(t-s-1)^{(\nu -2)}]<0\) for \(s< t-\nu \le N\). Thus G is decreasing for \(s+\nu <t\le \nu +N\). Hence \(\max _{t\in [\nu -1,\nu +N]_{\mathbb {N}_{\nu -2}}}G(t,s)=G(s+\nu ,s).\)
Now
Observe that, for \(\frac{(N+\nu )}{4}\le t\le \frac{3(N+\nu )}{4} \) and \(t-\nu <s\le N,\) we have \(t^{(\nu -1)}\ge \left( \frac{N+\nu }{4}\right) ^{(\nu -1)} \) and \((s+\nu )^{(\nu -1)}\le (N+\nu )^{(\nu -1)}\). Therefore
Now, for \(\frac{(N+\nu )}{4}\le t\le \frac{3(N+\nu )}{4} \) and \(s\le t-\nu ,\) we have \((t-s-1)^{\nu -1}\le (\frac{3}{4}(N+\nu )-s-1)^{\nu -1} <(\frac{3}{4}(N+\nu )-1)^{\nu -1}\) and \(t^{(\nu -1)}\ge (s+\nu )^{(\nu -1)}.\) Consequently
Therefore
where \(\gamma :=\min \{ \frac{(\frac{1}{4}(N+\nu ))^{(\nu -1)}}{(N+\nu )^{(\nu -1)}},~1- \frac{(\frac{3}{4}(N+\nu )-1)^{(\nu -1)}}{(\frac{3}{4}(N+\nu ))^{(\nu -1)}}\}\). Note that \(\frac{(\frac{1}{4}(N+\nu ))^{(\nu -1)}}{(N+\nu )^{(\nu -1)}}<1\) and also \(\frac{(\frac{3}{4}(N+\nu )-1)^{(\nu -1)}}{(\frac{3}{4}(N+\nu ))^{(\nu -1)}} <1\). Therefore \(0<\gamma <1\). \(\square \)
4 Existence of positive solutions
In recent decades, the Krasnoselskii fixed point theorem and its generalizations have been frequently applied to establish the existence of solutions in the study of boundary value problems. We shall use following well-known Guo–Krasnoselskii fixed point theorem to establish sufficient conditions for the existence of at least one positive solution for boundary value problem (1).
Definition 2
We call a function u(t) a positive solution of problem (1), if \(u(t)\in C[\nu -2,\nu +N]_{{\mathbb {N}}_{\nu -2}}\) and \(u(t)\ge 0 \) for \(t\in [\nu -2,\nu +N]_{{\mathbb {N}}_{\nu -2}}\) and satisfies (1).
Theorem 1
[18] Let B be a Banach space and \(K\subseteq B\) be a cone. Assume \(\Omega _1\) and \(\Omega _2\) are open discs contained in B with \(0\in \Omega _1\), and \(\bar{\Omega }_{1}\subset \Omega _2.\) Furthermore assume that \(T:K\cap (\bar{\Omega }_2\backslash {\Omega }_{1})\rightarrow K\) be a completely continuous operator such that, either
-
(i)
\(\Vert Tu\Vert \le \Vert u\Vert \) for \(u\in K\cap \partial \Omega _1\) and \(\Vert Tu\Vert \ge \Vert u\Vert \) for \(u\in K\cap \partial \Omega _2\), or
-
(ii)
\(\Vert Tu\Vert \ge \Vert u\Vert \) for \(u\in K\cap \partial \Omega _1\) and \(\Vert Tu\Vert \le \Vert u\Vert \) for \(u\in K\cap \partial \Omega _2\).
Then T has at least one fixed point in \(K\cap (\bar{\Omega }_2\backslash {\Omega }_{1}).\)
We define Banach space B by
equipped with the norm \(\Vert u\Vert =\max |u(t)|,~t\in [\nu -2,~\nu +N]_{\mathbb {N}_{\nu -2}}.\) In addition, for some \({\gamma }\in (0,1)\) we define
Let \(T:B\rightarrow B\) be an operator defined as
then \(u\in B\) is a solution of (10) if and only if \(u\in B\) is a solution of (1).
Lemma 6
Let T be defined as in (10) and K in (9). Then \(T:K\rightarrow K\) and T is completely continuous.
Proof
Let \(u\in K.\) Then by Lemma 5, it follows that
Therefore \(T:K\rightarrow K\). One can easily prove that T is completely continuous. \(\square \)
Following conditions on the growth of function f will be used in the sequel to establish some existence results.
-
(\(\mathbf{H_1}\)) There exists \(\rho >0\) such that \(f(t,u)\le \frac{1}{2}\eta \rho \) whenever \(0\le u\le \rho \),
-
(\(\mathbf{H_2}\)) There exists \(\rho >0\) such that \(f(t,u)\ge \mu \rho \) whenever \({\gamma } \rho \le u\le \rho .\)
For convenience, we introduce following notations:
We now can prove the following existence result.
Theorem 2
Suppose that there are distinct \(\rho _{1},\rho _{2}>0\) such that condition (H\(_{1}\)) holds at \(\rho =\rho _{1}\) and condition (H\(_{2}\)) holds at \(\rho =\rho _{2}.\) Suppose also that \(f(t,u)\ge 0\) and continuous. Then the problem (1) has at least one positive solution, say \(u_{0},\) such that \(|u_{0}|\) lies between \(\rho _{1}\) and \(\rho _{2}.\)
Proof
We shall assume without loss of generality that \(0<\rho _{1}<\rho _{2}.\) For \(u\in K\), by Lemma 6 \(Tu\in K\) and T is a completely continuous operator. Now put \(\Omega _{1}=\{u\in K:||u||<\rho _{1}\}\). Note that for \(u\in \partial \Omega _{1},\) we have that \(||u||=\rho _{1},\) so that condition \((H_{1})\) holds for all \(u\in \partial \Omega _{1}.\) So for \(u\in K\cap \partial \Omega _{1},\) we find that
Whence we find that \(||Tu||\le ||u||,\) whenever \(u\in K\cap \partial \Omega _{1}.\) Thus we get that the operator T is a cone compression on \(K\cap \Omega _{1}.\) On the other hand, put \(\Omega _{2}=\{u\in K:||u||<\rho _{2}\}\). Note that for \(u\in \partial \Omega _{2},\) we have that \(||u||=\rho _{2},\) so that condition \((H_{2})\) holds for all \(u\in \partial \Omega _{2}.\) Also note that \(\left\{ \lfloor \frac{N+1}{2}\rfloor +\nu \right\} \subset \left[ \frac{N+\nu }{4},\frac{3(N+\nu )}{4}\right] .\) So, for \(u\in K\cap \partial \Omega _{2},\) we find that
Whence \(||Tu||\ge ||u||,\) whenever \(u\in K\cap \partial \Omega _{2}.\) Thus we get that the operator T is a cone expansion on \(K\cap \partial \Omega _{2}.\) So, it follows from Theorem 1 that the operator T has a fixed point. This means that (1) has a positive solution, say \(u_{0}\) with \(\rho _{1}\le ||u_0||\le \rho _{2},\) as claimed. \(\square \)
We now establish some results that yield the existence or non existence of positive solutions under the assumption that f(t, u) has the special form \(f(t,u)\equiv \lambda F_{1}(t)F_{2}(u).\)
-
(\(\mathbf{H_3}\)) We assume that \(\lambda >0,\) and that \(F_{1},F_{2}\) are nonnegative. Let
$$\begin{aligned} F_{0}=\lim _{u\rightarrow 0}\frac{F_{2}(u)}{u}, \qquad F_{\infty }=\lim _{u\rightarrow \infty }\frac{F_{2}(u)}{u}. \end{aligned}$$
Theorem 3
Assume that condition rm (H\(_{3})\) holds.
-
(\(\mathbf{H_4}\)) If \(F_{0}=0\) and \(F_{\infty }=\infty ,\) then for all \(\lambda >0\) problem (1) has a positive solution.
-
(\(\mathbf{H_5}\)) If \(F_{0}=\infty \) and \(F_{\infty }=0,\) then for all \(\lambda >0\) problem (1) has a positive solution.
Proof
In order to prove that boundary value problem (1) has a positive solution, it is sufficient to show that operator \(T:K\rightarrow K\) defined by Eq. (10) has a fixed point. The operator \(T:K\rightarrow K\) is completely continuous. Define \(K_{\rho }=\left\{ u\in K:\Vert u\Vert <\rho \right\} \) and \(\partial K_{\rho }=\left\{ u\in K:\Vert u\Vert =\rho \right\} ,\) then the operator \(T:K_{\rho }\rightarrow K\) defined as \(Tu(t):=\lambda \sum _{s=0}^{N+1}G(t,s)F_{1}(s)F_{2}(u(s))\) is completely continuous. Now, \(F_{0}=0\) implies, for \(\epsilon \in (\lambda \sum _{s=0}^{N+1}G(s+\nu ,s)F_{1}(s))^{-1},\) there exists positive \(\rho _{1}>0\) such that \(F_{2}(u)<\epsilon |u|\), whenever \(0<|u|<\rho _{1}.\) Let \(\Omega _{\rho _{1}}=\{u\in K:\Vert u\Vert <\rho _{1}\}.\) For \(u\in \partial \Omega _{\rho _{1}}\cap K,\) we have
which imply that \(\Vert Tu(t)\Vert \le \rho _{1}=\Vert u\Vert ,\) for \(u\in \partial \Omega _{\rho _{1}}\cap K.\) The assumption \(F_{\infty }=\infty \) implies, for \(M>(\lambda \gamma ^{2}\sum _{s=\lceil \frac{N+\nu }{4}\rceil } ^{\lfloor \frac{3(N+\nu )}{4}\rfloor } G(s+\nu ,s)F_{1}(s))^{-1},\) there exists \(N_0>0\) such that \(F_{2}(u)>M|u|,\) whenever \(|u|>N_0.\) Take \(\rho _{2}>\{\rho _{1},\frac{N_0}{\gamma }\},\) and let \(\Omega _{\rho _{2}}=\{u\in K:\Vert u\Vert <\rho _{2}\}.\) Then for \(u\in \partial \Omega _{\rho _{2}}\cap K\), we have \( u(t)\ge \gamma \Vert u\Vert =\gamma \rho _{2}>N_0\), for \(\frac{N+\nu }{4}\le t\le \frac{3(N+\nu )}{4}.\) Hence
which implies that \(||Tu||\ge ||u||,\) for \(u\in \partial \Omega _{\rho _{2}}\cap K.\) From Theorem 1(i), it follows that T has at least one fixed point in \(K\cap (\bar{\Omega }_{\rho _{2}}/{\Omega }_{\rho _{1}}).\) From Lemma 4, the fixed point \(u\in K\cap (\bar{\Omega }_{\rho _{2}}/{\Omega }_{\rho _{1}})\) is the positive solution of (1).
The hypothesis \(F_{0}\,{=}\,\infty \) implies that, for \(M\,{>}\,(\lambda \gamma ^{2}\sum _{s=\lceil \frac{N+\nu }{4}\rceil }^{\lfloor \frac{3(N+\nu )}{4}\rfloor } G(s\,{+}\,\nu ,s)F_{1}(s))^{-1},\) there exists a positive \(\rho _{1}>0\) such that \(F_{2}(u)>M|u|,\) whenever \(0<|u|<\rho _{1}.\) Let \(\Omega _{\rho _{1}}=\{u\in K:\Vert u\Vert <\rho _{1}\}.\) For \(u\in \partial \Omega _{\rho _{1}}\cap K,\) we have
which implies that \(||Tu||\ge ||u||\), for \(u\in \partial \Omega _{\rho _{1}}\cap K.\)
\(F_{\infty }=0\) implies, for \(\epsilon \in \Big (2\lambda \sum _{s=0}^{N+1} G(s+\nu ,s)F_{1}(s)\Big )^{-1},\) there exists positive \(N_0>0\) such that \(F_{2}(u)<\epsilon |u|,\) for \(|u|>N_0.\) Therefore, \(F_{2}(u)\le \epsilon |u|+\Lambda ,\) for \(u\in [0,+\infty ),\) where \(\Lambda ={\max }_{0\le u\le N_0}F_{2}(u)+1.\) Let \(\Omega _{\rho _{2}}=\{u\in K: ||u||<\rho _{2}\},\) where \(\rho _{2}>\{\rho _{1},2\Lambda \lambda \sum _{s=0}^{N+1}G(s+\nu ,s)F_{1}(s)\},\) then, for \(u\in \partial \Omega _{\rho _{2}}\cap K,\) we have
which implies \(||Tu||\le ||u||,\) for \(u\in \partial \Omega _{\rho _{2}}\cap K.\) It follows from Theorem 1 (ii) that T has at least one fixed point u in \(K\cap (\bar{\Omega }_{\rho _{2}}/{\Omega }_{\rho _{1}}).\) From Lemma 4, the fixed point \(u\in K\cap (\bar{\Omega }_{\rho _{2}}/{\Omega }_{\rho _{1}})\) is the positive solution of (1). \(\square \)
Theorem 4
Assume \(F_{0}=0\) or \(F_{\infty }=0\) and condition (H\(_{3})\) holds. Then there exists \(\lambda _{0}>0\) such that for all \(\lambda >\lambda _{0}\) problem (1) has a positive solution.
Proof
Choose \(\rho >0\) and define \(K_{\rho }=\{u\in K:||u||<\rho \},\) \(\partial K_{\rho }=\{u\in K:||u||=\rho \}.\) Then, the operator \(T:K_{\rho }\rightarrow K\) is completely continuous. For fixed \(\rho _{1}>0,\) we define \(\lambda _{0}:=\rho _{1}\Big (\gamma m_{\rho _{1}}\sum _{s=\lceil \frac{N+\nu }{4}\rceil }^{\lfloor \frac{3(N+\nu )}{4}\rfloor } G(s+\nu ,s)F_{1}(s)\Big )^{-1}\) and \(\Omega _{\rho _{1}}:=\{u\in K:||u||<\rho _{1}\},\) where \(m_{\rho _{1}}:={\min _{\gamma \rho _{1}\le u\le \rho _{1}} F_{2}(u)}\). By \((H_{3}),m_{\rho _{1}>0}\). Then for \(u\in K\cap \partial \Omega _{\rho _{1}},\) by the Lemma 5, we have
which implies that
Now, \(F_{0}=0\) implies, for \(\epsilon \in \Big (0, (\lambda \sum _{s=0}^{N+1}G(s+\nu ,s)F_{1}(s))^{-1}\Big ),\) there exists positive \(\overline{\rho }_{2}>0\) such that \(F_{2}(u)<\epsilon |u|,\) whenever \(0<|u|<\overline{\rho }_{2}.\) Let \(\Omega _{\rho _{2}}=\{u\in K: ||u||<\rho _{2}\},0<\rho _{2}<\left\{ \rho _{1},\overline{\rho }_{2}\right\} .\) For \(u\in K\cap \partial \Omega _{\rho _{2}},\) we have
which imply that \(||Tu||\le \rho _{2}=||u||\), for \(u\in K\cap \partial \Omega \rho _{2}.\) \(F_{\infty }=0\) implies, for \(\epsilon \in \Big (0,~ (2\lambda \sum _{s=0}^{N+1}G(s+\nu ,s)F_{1}(s))^{-1}\Big ),\) there exists positive \(N_0>0\) such that \(F_{2}(u)<\epsilon |u|,\) for \(|u|>N_0.\) Thus, \(F_{2}(u)\le \epsilon |u|+\Lambda ,\) for \(u\in [0,+\infty ),\) where \(\Lambda ={\max _{0\le u\le N_0}}F_{2}(u)+1.\) Let \(\Omega _{\rho _{3}}=\{u\in K: ||u||<\rho _{3}\},\) where \(\rho _{3}>\{\rho _{1},2\lambda \Lambda \sum _{s=0}^{N+1}G(s+\nu ,s)F_{1}(s)\},\) then, for \(u\in K\cap \partial \Omega _{\rho _{3}},\) it follows that
which imply that \(||Tu||\le \rho _{3}=||u||,\) for \(u\in K\cap \partial \Omega _{\rho _{3}}.\) It follows from Theorem 5 that T has a fixed point in \(u\in K\cap \bar{\Omega }_{\rho _{2}}/{\Omega }{\rho _{1}}\) or \(u\in K\cap \bar{\Omega }_{\rho _{3}}/{\Omega }{\rho _{1}}\) according to \(F_{0}=0\) or \(F_{\infty }=0,\) respectively. Consequently, problem (1) has a positive solution for \(\lambda >\lambda _{0}.\) \(\square \)
Theorem 5
Assume \(F_{0}=F_{\infty }=0,\) and (H\(_{3})\) holds. Then there exists \(\lambda _{0}>0\) such that for all \(\lambda >\lambda _{0}\) problem (1) has two positive solutions.
Proof
Let \(\rho _{3},\rho _{4}>0\) such that \(\rho _{3}<\rho _{4}.\) By the same argument as for inequality (11), there exists \(\lambda _{0}>0\) such that
Since \(F_{0}=F_{\infty }=0,\) then, it follows from the proof of Theorem 4 that we can choose \(0<\rho _{1}<\rho _{3}\) and \(\rho _{2}>\rho _{4}\) such that \(||Tu||<||u||\) for \(u\in K\cap \partial \Omega _{\rho _{i}}(i=1,2).\)
Hence, by Theorem 1 the operator T has two fixed points \(u_{1}\in K\cap \bar{\Omega }_{\rho _{3}}/{\Omega }\rho _{1}\) and \( u_{2}\in K\cap \bar{\Omega }_{\rho _{2}}/{\Omega }\rho _{4},\) which are the distinct positive solutions of problem (1). \(\square \)
Theorem 6
Assume that condition (H\(_{3})\) holds. If \(F_{0}<\infty \) and \(F_{\infty }<\infty ,\) then there exists \(\lambda _{0}>0\) such that for all \(0<\lambda <\lambda _{0}\) problem (1) has no positive solution.
Proof
Since \(F_0<\infty \) and \(F_\infty <\infty \), then for arbitrary \(\epsilon _{1},\epsilon _{2}>0\) there exist \(0<\rho _1<\rho _2\) such that,
Let
Then we have
On contrary, assume that w(t) is a positive solution of problem (1), then, for \(0<\lambda <\lambda _{0}\), where \(\lambda _{0}=({\sigma \epsilon \sum _{s=0}^{N+1}G(s+\nu ,s)F_{1}(s)})^{-1}\),for some \(0<\sigma <1\). Then, we have
That is, \(\Vert w\Vert \le \frac{\Vert w\Vert }{\sigma }<\Vert w\Vert \), which is a contradiction. Hence, problem (1) has no positive solution. \(\square \)
Proofs of the following theorems are similar to Theorems 4–6.
Theorem 7
Assume that condition (H\(_{3})\) holds and \(F_{0}=\infty \) or \(F_{\infty }=\infty .\) Then there exists \(\lambda _{0}>0\) such that for all \(0<\lambda <\lambda _{0}\) problem (1) has a positive solution.
Theorem 8
Assume that condition (H\(_{3})\) holds and \(F_{0}=F_{\infty }=\infty \). Then there exists \(\lambda _{0}>0\) such that for all \(0<\lambda <\lambda _{0}\) problem (1) has two positive solutions.
Theorem 9
Assume that condition (H\(_{3})\) holds and \(F_{0}>0\) or \(F_{\infty }>0.\) Then there exists \(\lambda _{0}>0\) such that for all \(\lambda >\lambda _{0}\) problem (1) has no positive solution.
Example 1
Consider the discrete boundary value problem
where \(\nu =\frac{3}{2}\) and \(N=5\). Let \(f(t,u(t)):=\frac{1+t^{2}e^{-1/2t}}{1+t^{2}}+\frac{t^{2}[1+\sin ^{2}u(t)]}{1+u^{2}(t)}\), \(t\in [\frac{1}{2},\frac{13}{2}]_{\mathbb {N}_{\frac{1}{2}}}.\) By computations, we have the following estimate:
Also, note that
Now we compute the value of \(\gamma \) as
The Green’s function for boundary value problem (12) is given by
The value of \(\eta \) and \(\mu \) are computed as:
Since
Then, \(\mu =1/\varsigma \approx 3.241557911\). Choose \(\rho _{1}=1140\). Then we have \(f(t,u)\le 111.623< \frac{1}{2}\eta \rho _1=\frac{1}{2}(0.20397)(1095)\approx 116.263\). Now, choose \(\rho _{2}=0.001\). Then \(f(t,u(t))\ge 0.0233455>\mu \rho _2=0.00324156.\)
All conditions of the Theorem 2 are satisfied. Therefore, boundary value problem (12) has a positive solution satisfying \(0.001\le |u|\le 1140.\)
Example 2
Consider the following discrete boundary value problem
where \(1<\nu \le 2\) and \(N>3\) is an integer. Let \(\nu =3/2\) and \(N=4\). Taking \(f(t,u(t))=\lambda e^{-t}\sqrt{t^2+1}[\frac{u^2(t)}{\pi +u^3(t)}],\) for \(t\in [\nu -2, \nu +N]_{\mathbb {N}_{\nu -2}}.\) Obviously \(f(t,u(t))=\lambda F_{1}({t})F_{2}({u(t)}),\) where \(F_{1}(t):=\sqrt{t^2+1}e^{-t}\) and \(F_{2}(u(t)):=\frac{u^2(t)}{\pi +u^3(t)}\) satisfy assumption \((H_{3}).\) By computations \(m_{\rho _{1}}={\min _{\gamma \rho _{1}\le u\le \rho _{1}}}F_{2}(u)={\min _{0.9696\le u\le 8}}(\frac{u^2(t)}{\pi +u^3(t)})\approx 0.23195,\) for \(\rho _{1}=8.\) Furthermore
Moreover, \(F_{0}=0\) and \(F_{\infty }=0.\) All assumptions of the Theorem 5 are satisfied. Consequently, the boundary value problem (14) has two positive solution.
References
Agarwal, R.P., Benchohra, M., Hamani, S.: A survey on existence results for boundary value problems of nonlinear fractional differential equations and inclusions. Acta Appl. Math. 109, 973–1033 (2010)
Henderson, J., Luca, R.: Boundary Value Problems for Systems of Differential, Difference and Fractional Equations. Elsevier Ltd, Amsterdam (2015)
Tisdell, C.C.: Basic existence and a priori bound results for solutions to systems of boundary value problems for fractional differential equations. Electron. J. Differ. Equ. 2016, 1–9 (2016)
Zhang, S.: Existence results of positive solutions to boundary value problem for fractional differential equation. Positivity 13, 583–599 (2009)
Miller, K.S., Ross, B.: Fractional difference calculus. In: Proceedings of the International Symposium on Univalent Functions, Fractional Calculus and their Applications, Nihon University, Koriyama, Japan, pp 139–152 (1988)
Atici, F.M., Eloe, P.W.: A transform method in discrete fractional calculus. Int. J.f Differ. Equ. 2, 165–176 (2007)
Atici, F.M., Eloe, P.W.: Fractional q-calculus on a time scale. J. Nonlinear Math. Phys. 14, 341–352 (2007)
Atici, F.M., Eloe, P.W.: Initial value problems in discrete fractional calculus. Proc. Am. Math. Soc. 137, 981–989 (2009)
Holm, M.: Sum and difference compositions in discrete fractional calculus. CUBO A Math. J. 13, 153–184 (2011)
Holm, M.: Solutions to a discrete, nonlinear, (N\(-1\), 1) fractional boundary value problem. Int. J. Dyn. Syst. Differ. Equ. 3, 267–287 (2011)
Abdeljawad, T.: On Riemann and Caputo fractional differences. Comput. Math. Appl. 62, 1602–1611 (2011)
Goodrich, C.S.: On a discrete fractional three-point boundary value problem. J. Differ. Equ. Appl. 18, 397–415 (2012)
Jia, B., Erbe, L., Peterson, A.: Some relations between the Caputo fractional difference operators and integer-order differences. Electron. J. Differ. Equ. 2015, 1–7 (2015)
Goodrich, C.S., Peterson, A.C.: Discrete Fractional Calculus. Springer International Publishing, Switzerland (2015)
Atici, F.M., Eloe, P.W.: Two-point boundary value problems for finite fractional difference equations. J. Differ. Equ. Appl. 17, 445–456 (2011)
Goodrich, C.S.: Existence and uniqueness of solutions to a fractional difference equation with nonlocal conditions. Comput. Math. Appl. 61, 191–202 (2011)
Goodrich, C.S.: Solutions to a discrete right-focal fractional boundary value problem. Int. J. Differ. Equ. 5, 195–216 (2010)
Guo, D., Lakshmikantham, V.: Nonlinear Problems in Abstract Cones. Academic Press, Boston (1988)
Acknowledgements
We are very grateful to the reviewer for his useful comments that led to improvement of this manuscript.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
ur Rehman, M., Iqbal, F. & Seemab, A. On existence of positive solutions for a class of discrete fractional boundary value problems. Positivity 21, 1173–1187 (2017). https://doi.org/10.1007/s11117-016-0459-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11117-016-0459-4