Citation: Bashir Ahmad, Madeaha Alghanmi, Sotiris K. Ntouyas, Ahmed Alsaedi. A study of fractional differential equations and inclusions involving generalized Caputo-type derivative equipped with generalized fractional integral boundary conditions[J]. AIMS Mathematics, 2019, 4(1): 26-42. doi: 10.3934/Math.2019.1.26
[1] | M. Manigandan, Subramanian Muthaiah, T. Nandhagopal, R. Vadivel, B. Unyong, N. Gunasekaran . Existence results for coupled system of nonlinear differential equations and inclusions involving sequential derivatives of fractional order. AIMS Mathematics, 2022, 7(1): 723-755. doi: 10.3934/math.2022045 |
[2] | Saima Rashid, Abdulaziz Garba Ahmad, Fahd Jarad, Ateq Alsaadi . Nonlinear fractional differential equations and their existence via fixed point theory concerning to Hilfer generalized proportional fractional derivative. AIMS Mathematics, 2023, 8(1): 382-403. doi: 10.3934/math.2023018 |
[3] | Zaid Laadjal, Fahd Jarad . Existence, uniqueness and stability of solutions for generalized proportional fractional hybrid integro-differential equations with Dirichlet boundary conditions. AIMS Mathematics, 2023, 8(1): 1172-1194. doi: 10.3934/math.2023059 |
[4] | Muneerah AL Nuwairan, Ahmed Gamal Ibrahim . The weighted generalized Atangana-Baleanu fractional derivative in banach spaces- definition and applications. AIMS Mathematics, 2024, 9(12): 36293-36335. doi: 10.3934/math.20241722 |
[5] | Kishor D. Kucche, Sagar T. Sutar, Kottakkaran Sooppy Nisar . Analysis of nonlinear implicit fractional differential equations with the Atangana-Baleanu derivative via measure of non-compactness. AIMS Mathematics, 2024, 9(10): 27058-27079. doi: 10.3934/math.20241316 |
[6] | Iyad Suwan, Mohammed S. Abdo, Thabet Abdeljawad, Mohammed M. Matar, Abdellatif Boutiara, Mohammed A. Almalahi . Existence theorems for Ψ-fractional hybrid systems with periodic boundary conditions. AIMS Mathematics, 2022, 7(1): 171-186. doi: 10.3934/math.2022010 |
[7] | Mouffak Benchohra, John R. Graef, Nassim Guerraiche, Samira Hamani . Nonlinear boundary value problems for fractional differential inclusions with Caputo-Hadamard derivatives on the half line. AIMS Mathematics, 2021, 6(6): 6278-6292. doi: 10.3934/math.2021368 |
[8] | Bashir Ahmad, Ahmed Alsaedi, Ymnah Alruwaily, Sotiris K. Ntouyas . Nonlinear multi-term fractional differential equations with Riemann-Stieltjes integro-multipoint boundary conditions. AIMS Mathematics, 2020, 5(2): 1446-1461. doi: 10.3934/math.2020099 |
[9] | Kangqun Zhang . Existence and uniqueness of positive solution of a nonlinear differential equation with higher order Erdélyi-Kober operators. AIMS Mathematics, 2024, 9(1): 1358-1372. doi: 10.3934/math.2024067 |
[10] | Abdulwasea Alkhazzan, Wadhah Al-Sadi, Varaporn Wattanakejorn, Hasib Khan, Thanin Sitthiwirattham, Sina Etemad, Shahram Rezapour . A new study on the existence and stability to a system of coupled higher-order nonlinear BVP of hybrid FDEs under the p-Laplacian operator. AIMS Mathematics, 2022, 7(8): 14187-14207. doi: 10.3934/math.2022782 |
We investigate a new class of boundary value problems for generalized Caputo-type fractional differential equations and inclusions supplemented with Katugampola type generalized fractional integral boundary conditions. Precisely, we study the following problems:
{ρcDα0+y(t)=f(t,y(t)), t∈J:=[0,T],y(T)=m∑i=1σiρIβ0+y(ηi)+κ, δy(0)=0, ηi∈(0,T), | (1.1) |
and
{ρcDα0+y(t)∈F(t,y(t)), t∈J:=[0,T],y(T)=m∑i=1σiρIβ0+y(ηi)+κ, δy(0)=0, ηi∈(0,T), | (1.2) |
where ρcDα0+ denotes the generalized Caputo-type fractional derivative of order 1<α≤2,ρ>0, ρIβ0+ is the Katugampola type fractional integral of order β>0,ρ>0, f:J×R→R is a continuous function, σi∈R,i=1,2,…,m,κ∈R, δ=t1−ρddt, and F:J×R→P(R) is a multivalued function (P(R) is the family of all nonempty subjects of R).
Here we emphasize that the problem considered in the present paper is motivated by Laskin's work [1] on the generalization of the Feynman and Wiener path integrals in the context of fractional quantum mechanics and fractional statistical mechanics. One can find more details in the articles [2, 3]. It is expected that the results obtained in this paper will provide more leverage in dealing with Feynman and Wiener path type integrals involving an index like ρ>0 in (1.1), instead of a fixed choice ρ=1 (Caputo fractional derivative case). Moreover, chaos for a fractional order differential equation involving two parameters (α and ρ>0) becomes more complicated than the one containing Caputo fractional derivative of order α (generalized Caputo-type fractional derivative with ρ=1); one can find more details in [4, 5]. It is worthwhile to notice that Katugampola fractional integral unifies the Riemann-Liouville and Hadamard integrals into a single integral [6]. Thus, our results are more general in the context of integral boundary conditions.
The topic of fractional-order differential equations and inclusions attracted significant attention in recent years and several results on fractional differential equations involving Riemann-Liouville, Caputo, Hadamard type derivatives, supplemented with a variety of boundary conditions, can be found in the related literature [7, 8, 9]. The interest in the subject owes to its extensive applications in various disciplines of science and engineering, for instance, see the papers [10, 11, 12, 13, 14, 15, 16, 17, 18], and the references cited therein. In a recent paper [19], the authors studied fractional differential equations involving Caputo-Katugampola derivative. In a more recent work [20], the authors studied a fractional order boundary value problem involving Katugampola-type generalized fractional derivative and generalized fractional integral.
We organize the rest of the paper as follows. Section 2 contains preliminary material related to our work. The existence and uniqueness results for the problem (1.1), obtained with the aid of the standard fixed point theorem, are presented in Section 3. The existence results for the inclusions problem (1.2) are derived in Section 4. Examples are provided to demonstrate the application of the main theorems.
For c∈R,1≤p≤∞, let Xpc(a,b) denote the space of all complex-valued Lebesgue measurable functions ϕ on (a,b) endowed with the norm:
‖ϕ‖Xpc=(∫ba|xcϕ(x)|pdxx)1/p<∞. |
We denote by L1(a,b) the space of all Lebesgue measurable functions φ on (a,b) equipped with the norm:
‖φ‖L1=∫ba|φ(x)|dx<∞. |
Let G=C(J,R) denote the Banach space of all continuous functions from [0,T] to R endowed with the norm defined by ‖y‖=supt∈[0,T]|y(t)|.
Recall that
AC^n(J, \mathbb{R})=\{h:J\to \mathbb{R}: h, h', \dots, h^{(n-1)}\in C(J, \mathbb{R}) ~ \text{and} ~ h^{(n-1)} \text{is absolutely continuous} \}. |
For 0\leq \epsilon < 1, we define C_{\epsilon, \rho}(J, \mathbb{R})=\{ f:J\to \mathbb{R}:(t^\rho-a^\rho)^\epsilon f(t)\in C(J, \mathbb{R})\} equipped with the norm \|f\|_{C_{\epsilon, \rho}}=\|(t^\rho-a^\rho)^\epsilon f(t)\|_C. Moreover, let us introduce AC^n_\delta(J), which consists of the functions f that have absolutely continuous \delta^{n-1}-derivative, where \delta=t^{1-\rho}\frac{d}{dt}. Thus we define spaces AC^n_\delta(J, \mathbb{R})=\Big\{f:J\to \mathbb{R}:\delta^{n-1}f\in AC(J, \mathbb{R}), ~\delta=t^{1-\rho}\frac{d}{dt}\Big\}, and C^n_{\delta, \epsilon}(J, \mathbb{R})=\Big\{f:J\to \mathbb{R}:\delta^{n-1}f\in C(J, \mathbb{R}), \delta^{n}f\in C_{\epsilon, \rho}(J, \mathbb{R}), \delta=t^{1-\rho}\frac{d}{dt}\Big\} endowed with the norms \|f\|_{C^n_{\delta}}=\sum_{k=0}^{n-1}\|\delta^kf\|_C and \|f\|_{C^n_{\delta, \epsilon}}=\sum_{k=0}^{n-1}\|\delta^kf\|_C+\|\delta^nf\|_{C_{\epsilon, \rho}} respectively. Here we use the convention C^n_{\delta, 0}=C^n_{\delta}.
Definition 2.1. [6] The generalized fractional integral of order \alpha>0 and \rho>0 of f \in X_c^p(a, b) for -\infty < a < t < b < \infty, is defined by
(^{\rho}I^{\alpha}_{a^+}f)(t)=\frac{\rho^{1-\alpha}}{\Gamma(\alpha)}\int_a^{t}\frac{s^{\rho-1}}{(t^{\rho}-s^{\rho})^{1-\alpha}}f(s)ds. | (2.1) |
Note that the integral in (2.1) is called the left-sided fractional integral. Similarly we can define right-sided fractional integral ^{\rho}I^{\alpha}_{b^-}f as
(^{\rho}I^{\alpha}_{b^-}f)(t)=\frac{\rho^{1-\alpha}}{\Gamma(\alpha)}\int_t^{b}\frac{s^{\rho-1}}{(s^{\rho}-t^{\rho})^{1-\alpha}}f(s)ds. | (2.2) |
Definition 2.2. [21] The generalized fractional derivative, associated with the generalized fractional integrals (2.1) and (2.2) for 0 \le a < t < b < \infty, are defined by
\begin{eqnarray}\label{rfd} (^{\rho}D^{\alpha}_{a^+}f)(t)&=&\Big(t^{1-\rho}\frac{d}{dt}\Big)^n(^{\rho}I^{n-\alpha}_{a^+}f)(t)\nonumber\\ &=&\frac{\rho^{\alpha-n+1}}{\Gamma(n-\alpha)}\Big(t^{1-\rho}\frac{d}{dt}\Big)^n\int_a^t\frac{s^{\rho-1}}{(t^{\rho}-s^{\rho})^{\alpha-n+1}}f(s)ds \end{eqnarray} | (2.3) |
and
\begin{eqnarray} \label{lfd} (^{\rho}D^{\alpha}_{b^-}f)(t)&=&\Big(-t^{1-\rho}\frac{d}{dt}\Big)^n(^{\rho}I^{n-\alpha}_{b^-}f)(t)\nonumber\\ &=&\frac{\rho^{\alpha-n+1}}{\Gamma(n-\alpha)}\Big(-t^{1-\rho}\frac{d}{dt}\Big)^n\int_t^b\frac{s^{\rho-1}}{(s^{\rho}-t^{\rho})^{\alpha-n+1}}f(s)ds, \end{eqnarray} | (2.4) |
if the integrals in (2.3) and (2.4) exist.
Definition 2.3. [22] For \alpha\geq0 and f \in AC^n_\delta[a, b], the generalized Caputo-type fractional derivatives {}^{\rho}_{c}D_a^{\alpha} and {}^{\rho}_{c}D_b^{\alpha} are defined via the above generalized fractional derivatives as follows
{}^{\rho}_{c}D_{a^+}^{\alpha} f(x)={}^{\rho}D_{a^+}^{\alpha}\Bigg[f(t)-\sum\limits_{k=0}^{n-1}\frac{\delta^k f(a)}{k!}\Big(\frac{t^{\rho}-a^{\rho}}{\rho}\Big)^k\Bigg](x), ~~\delta=x^{1-\rho}\frac{d}{dx}, | (2.5) |
{}^{\rho}_{c}D_{b^-}^{\alpha} f(x)={}^{\rho}D_{b^{-}}^{\alpha}\Bigg[f(t)-\sum\limits_{k=0}^{n-1}\frac{(-1)^k\delta^k f(b)}{k!}\Big(\frac{b^{\rho}-t^{\rho}}{\rho}\Big)^k\Bigg](x), ~~\delta=x^{1-\rho}\frac{d}{dx}, | (2.6) |
where n=[\alpha]+1.
Lemma 2.1. [22] Let \alpha\geq0, n=[\alpha]+1 and f \in AC^n_\delta[a, b], where 0 < a < b < \infty. Then,
(1) for \alpha \notin \mathbb{N},
{}^{\rho}_{c}D_{a^+}^{\alpha} f(t)=\frac{1}{\Gamma(n-\alpha)}\int_{a}^{t}\Big(\frac{t^\rho-s^\rho}{\rho}\Big)^{n-\alpha-1}\frac{(\delta^n f)(s) ds}{s^{1-\rho}}={}^{\rho}I_{a^+}^{n-\alpha} (\delta^n f)(t), | (2.7) |
{}^{\rho}_{c}D_{b^-}^{\alpha} f(t)=\frac{1}{\Gamma(n-\alpha)}\int_{t}^{b}\Big(\frac{s^\rho-t^\rho}{\rho}\Big)^{n-\alpha-1}\frac{(-1)^n(\delta^n f)(s) ds}{s^{1-\rho}}={}^{\rho}I_{b^-}^{n-\alpha} (\delta^n f)(t); | (2.8) |
(2) for \alpha \in \mathbb{N},
{}^{\rho}_{c}D_{a^+}^{\alpha} f= \delta^n f, ~~~{}^{\rho}_{c}D_{b^-}^{\alpha} f= (-1)^n\delta^n f. | (2.9) |
Lemma 2.2. [22] Let f \in AC_{\delta}^{n}[a, b] or C_{\delta}^{n}[a, b] and \alpha\in \mathbb{R}. Then
{}^{\rho}I_{a^+}^{\alpha}{}^{\rho}_{c}D_{a^+}^{\alpha}f(x)=f(x)-\sum\limits_{k=0}^{n-1}\frac{(\delta^k f)(a)}{k!}\Big(\frac{x^{\rho}-a^{\rho}}{\rho}\Big)^k, |
{}^{\rho}I_{b^-}^{\alpha}{}^{\rho}_{c}D_{b^-}^{\alpha}f(x)=f(x)-\sum\limits_{k=0}^{n-1}\frac{(-1)^k(\delta^k f)(a)}{k!}\Big(\frac{b^{\rho}-x^{\rho}}{\rho}\Big)^k. |
In particular, for 0 < \alpha\leq 1, we have
{}^{\rho}I_{a^+}^{\alpha}{}^{\rho}_{c}D_{a^+}^{\alpha}f(x)=f(x)-f(a), |
{}^{\rho}I_{b^-}^{\alpha}{}^{\rho}_{c}D_{b^-}^{\alpha}f(x)=f(x)-f(b). |
Next we define a solution for the problem (1.1).
Definition 2.4. A function y\in AC_\delta^2([0, T], {\Bbb R}) is said to be a solution of (1.1) if y satisfies the equation {}^{\rho}_cD^{\alpha}y(t)=f(t, y(t)) on [0, T], and the conditions y(T)=\sum_{i=1}^{m}\sigma_i~{}^{\rho}I^{\beta}y(\eta_i)+\kappa, ~ \delta y(0)=0.
Relative to the problem (1.1), we consider the following lemma.
Lemma 2.3. Let h\in C (0, T)\cap L^1(0, T), y\in AC_\delta^2(J) and
\Omega=1-\sum\limits_{i=1}^{m}\sigma_i \frac{\eta_i^{\rho\beta}}{\rho^{\beta}\Gamma(\beta+1)}\ne 0. | (2.10) |
Then the integral solution of the linear boundary value problem (BVP):
\left\{\begin{array} {ll} ^{\rho}_cD^{\alpha}_{0^+}y(t)= h(t), ~~ t\in J:=[0, T], \\[0.3cm] \displaystyle y(T)= \sum\limits_{i=1}^{m}\sigma_i{}^{\rho}I_{0^+}^{\beta}y(\eta_i)+\kappa, ~~\delta y(0)=0, ~~ \eta_i\in (0, T), \end{array} \right. | (2.11) |
is given by
y(t)={}^{\rho}I_{0^+}^{\alpha}h(t)+\frac{1}{\Omega}\Big\{- {}^{\rho}I_{0^+}^{\alpha}h(T)+\sum\limits_{i=1}^{m}\sigma_i{}^{\rho}I_{0^+}^{\alpha+\beta}h(\eta_i)+\kappa \Big\}. | (2.12) |
Proof. Applying ^{\rho}I_{0^+}^{\alpha} on both sides of the fractional differential equation in (2.11) and using Lemma 2.2, we get
y(t)={}^{\rho}I_{0^+}^{\alpha}h(t)+c_{1}+c_{2}\frac{t^{\rho}}{\rho}=\frac{\rho^{1-\alpha}}{\Gamma(\alpha)}\int_{0}^{t}s^{\rho-1} (t^{\rho}-s^{\rho})^{\alpha-1}h(s)ds+c_{1}+c_{2}\frac{t^{\rho}}{\rho}, | (2.13) |
for some c_{1}, c_{2}\in \mathbb{R}. Taking \delta-derivative of (2.13), we get
\delta y(t)={}^{\rho}I_{0^+}^{\alpha-1}h(t)+c_{2}=\frac{\rho^{2-\alpha}}{\Gamma(\alpha-1)}\int_{0}^{t}s^{\rho-1} (t^{\rho}-s^{\rho})^{\alpha-2}h(s)ds+c_{2}. | (2.14) |
Using the boundary condition \delta y(0)=0 in (2.14), we get c_2=0. Applying the generalized integral {}^{\rho}I_{0^+}^{\beta} to (2.13) after inserting the value of c_2 in it, we get
{}^{\rho}I_{0^+}^{\beta}y(t)={}^{\rho}I_{0^+}^{\alpha+\beta}h(t)+c_1\frac{t^{\rho\beta}}{\rho^{\beta}\Gamma(\beta+1)}. | (2.15) |
Making use of the first boundary condition y(T)=\sum_{i=1}^{m}\sigma_i~{}^{\rho}I^{\beta}y(\eta_i)+\kappa in (2.15), we get
{}^{\rho}I_{0^+}^{\alpha}h(T)+c_1=\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}h(\eta_i)+\sum\limits_{i=1}^{m}\sigma_i c_1\frac{\eta_i^{\rho\beta}}{\rho^{\beta}\Gamma(\beta+1)}+\kappa, |
which, on solving for c_1 together with (2.10), yields
c_1=\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}h(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}h(\eta_i)+\kappa\Big\}. |
Substituting the values of c_1 and c_2 in (2.13), we obtain the solution (2.12). The converse follows by direct computation. The proof is completed.
Using Lemma 2.3, we define an operator \mathcal{N}:\mathcal{G}\rightarrow\mathcal{G} by
\mathcal{N} y(t) = {}^{\rho}I_{0^+}^{\alpha}f(t, y(t))+\frac{1}{\Omega}\Big\{- {}^{\rho}I_{0^+}^{\alpha}f(T, y(T))+\sum\limits_{i=1}^{m}\sigma_i{}^{\rho}I_{0^+}^{\alpha+\beta}f(\eta_i, y(\eta_i))+\kappa \Big\}. | (3.1) |
In the following, for brevity, we set the notation:
\Lambda=\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\frac{1}{|\Omega|}\Big\{\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\sum\limits_{i=1}^{m}|\sigma_i|\frac{\eta_i^{\rho(\alpha+\beta)}}{\rho^{\alpha+\beta}\Gamma(\alpha+\beta+1)}\Big\}. | (3.2) |
Our first existence result for the problem (1.1) relies on Leray-Schauder nonlinear alternative [23].
Theorem 3.1. Assume that
{(A_{1})} | f(t, y)| \leq p(t)\psi (\|y\|), \forall (t, y)\in [0, T]\times {\mathbb{R}}, where p\in L^{1}([0, T], {\mathbb{ R}}^{+}) and \psi:{\mathbb{R}}^{+}\rightarrow {\ \mathbb{R}}^{+} is a nondecreasing function;
{(A_{2})} we can find a positive constant W satisfying the inequality:
\frac{W}{\displaystyle \psi (W)\Big({}^{\rho}I_{0^+}^{\alpha}p(T) +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T) +\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i) +|\kappa|\Big\}\Big)} \gt 1. |
Then there exists at least one solution for the problem (1.1) on [0, T].
Proof. Consider the operator \mathcal{N}:\mathcal{G}\to\mathcal{G} defined by (3.1) and show that it is continuous and completely continuous. We establish in four steps.
(ⅰ) \mathcal{N} is continuous. Let \{y_{n}\} be a sequence such that y_{n}\to y in \mathcal{G}. Then
\begin{eqnarray*} |\mathcal{N}(y_n)(t)-\mathcal{N}(y)(t)| & \leq &{}^{\rho}I_{0^+}^{\alpha}| f(t, y_n(t))-f(t, y(t))| +\frac{1}{\Omega}\Bigg\{{}^{\rho}I_{0^+}^{\alpha}| f(T, y_n(T))-f(T, y(T))|\\ \displaystyle &&+ \sum\limits_{i=1}^{m} |\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}| f(\eta_i, y_n(\eta_i))-f(\eta_i, y(\eta_i))|\Bigg\} \leq \Lambda\|f(\cdot, y_n)-f(\cdot, y)\|. \end{eqnarray*} |
Since f is a continuous function, therefore, we have
\|\mathcal{N}(y_{n})-\mathcal{N}(y)\|\leq \Lambda\|f(\cdot, y_n)-f(\cdot, y)\| \to 0 , \, \, \mbox{as}\, \, n\to\infty. |
(ⅱ) The operator \mathcal{N} maps bounded sets into bounded sets in \mathcal{G}.
For any \bar{r}>0, it is indeed enough to show that there exists a positive constant \ell such that \|\mathcal{N}(y)\|\leq \ell for y\in B_{\bar{r}}=\{y\in \mathcal{G}: \|y\|\leq \bar{r}\}. By the assumption (A_{1}), for each t\in J, we have
\begin{eqnarray*} |\mathcal{N}(y)(t)|& \leq &{}^{\rho}I_{0^+}^{\alpha}|f(t, y(t))| +\frac{1}{|\Omega|}\Bigg\{{}^{\rho}I_{0^+}^{\alpha}| f(T, y(T))|+ \sum\limits_{i=1}^{m} |\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}| f(\eta_i, y(\eta_i))|+|\kappa|\Bigg\}\\ &\leq &{}^{\rho}I_{0^+}^{\alpha}p(T)\Omega (\|y\|) +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T)\psi (\|y\|)+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i)\psi (\|y\|)+|\kappa| \Big\}\\ &\leq &\psi (\|y\|)\Big({}^{\rho}I_{0^+}^{\alpha}p(T) +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T)+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i) +|\kappa|\Big\}\Big). \end{eqnarray*} |
Thus
\|\mathcal{N}(y)\| \leq \psi (\bar{r})\Big({}^{\rho}I_{0^+}^{\alpha}p(T) +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T)+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i) +|\kappa|\Big\}\Big):=\ell. |
(ⅲ) \mathcal{N} maps bounded sets into equicontinuous sets of \mathcal{G}.
Let t_{1}, t_{2}\in (0, T], \, t_{1} < t_{2}, B_{\bar{r}} be a bounded set of \mathcal{G} as in (ii) and let y\in B_{\bar{r}}. Then
\begin{eqnarray*} |\mathcal{N}(y)(t_{2})-\mathcal{N}(y)(t_{1})| &\le &\Big|{}^{\rho}I_{0^+}^{\alpha}f(t_2, y(t_2))-{}^{\rho}I_{0^+}^{\alpha}f(t_1, y(t_1))\Big|\\ &\le&\frac{\rho^{1-\alpha}\psi(\bar{r})}{\Gamma(\alpha)}\Big|\int_{0}^{t_1}\Big[\frac{s^{\rho-1}}{(t_2^{\rho}-s^{\rho})^{1-\alpha}}-\frac{s^{\rho-1}}{(t_1^{\rho}-s^{\rho})^{1-\alpha}}\Big] p(s)ds+ \int_{t_1}^{t_2}\frac{s^{\rho-1}}{(t_2^{\rho}-s^{\rho})^{1-\alpha}}p(s)ds\Bigg|\\ &\to& 0 ~\text{as}~ t_{1}\longrightarrow t_{2}, ~ \text{independent ~of} ~y. \end{eqnarray*} |
From the steps (i)-(iii), we deduce by the Arzelá-Ascoli theorem that \mathcal{N}:\mathcal{G}\longrightarrow \mathcal{G} is completely continuous.
(ⅳ) There exists an open set V \subseteq \mathcal{G} with y \ne \nu \mathcal{N}(y) for \nu \in (0, 1) and y\in \partial V.
Let y\in \mathcal{G} be a solution of y-\nu \mathcal{N} y=0 for \nu \in [ 0, 1]. Then, for t\in [0, T], we obtain
\begin{eqnarray*} |y(t)| &=& |\nu (\mathcal{N}y)(t)| \\ &\le&{}^{\rho}I_{0^+}^{\alpha}|f(t, y(t))| +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}|f(T, y(T))|+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}|f(\eta_i, y(\eta_i))| +|\kappa|\Big\}\\ &\le&{}^{\rho}I_{0^+}^{\alpha}p(T)\psi (\|y\|) +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T)\psi (\|y\|)+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i)\psi (\|y\|) +|\kappa|\Big\}\\ &\leq &\psi (\|y\|)\Big({}^{\rho}I_{0^+}^{\alpha}p(T) +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T) +\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i) +|\kappa|\Big\}\Big), \end{eqnarray*} |
which, on taking the norm for t \in [0, T], implies that
\displaystyle\frac{\|y\| }{\displaystyle \psi (\|y\|)\Big({}^{\rho}I_{0^+}^{\alpha}p(T) +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T) +\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i) +|\kappa|\Big\}\Big)}\leq 1. |
By the assumption (A_{2}), there exists a positive constant W such that \|y\| \ne W. Next we define V = \{y \in \mathcal{G} : \|y\| < W\} and note that the operator \mathcal{N} :\overline{V} \to \mathcal{G} is continuous and completely continuous. By the choice of V, there does not exist any y\in \partial V satisfying y=\nu \mathcal{N}(y) for some \nu \in (0, 1). In consequence, by the nonlinear alternative of Leray-Schauder type [23], we deduce that there exists a fixed point y\in \overline{V} for the operator \mathcal{N}, which is a solution of the problem (1.1).
In our next result, we make use of Banach contraction mapping principle to establish the uniqueness of solutions for the problem (1.1).
Theorem 3.2. Suppose that
(A_3) there exists a nonnegative constant L such that
|f(t, u)-f(t, v)|\le L\|u-v\| , \quad \mbox{for}\, \, t\in [0, T]\, \, \, \mbox{and every}\, \, u, v \in {\mathbb R}. |
Then the problem (1.1) has a unique solution on [0, T] if
L\Lambda \lt 1, | (3.3) |
where \Lambda is defined by (3.2).
Proof. Consider the operator \mathcal{N}: \mathcal{G}\to \mathcal{G} associated with the problem (1.1) defined by (3.1). With \Lambda given by (3.2), we fix
r\geq\frac{\Lambda f_0+|\kappa|/|\Omega|}{1-L\Lambda}, \, \, f_0=\sup\limits_{t\in [0, T]}|f(t, 0)|, |
and show that FB_{r}\subset B_{r} , where B_{r}= \{y\in \mathcal{G}:\|y\|\leq r\}. For y\in B_{r} , using (A_{3}), we get
\begin{eqnarray*} |\mathcal{N}(y)(t)| &\le& {}^{\rho}I_{0^+}^{\alpha}[|f(t, y(t))-f(t, 0)|+|f(t, 0)|] +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}[|f(T, y(T))-f(T, 0)|+|f(T, 0)|]\\ &&+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}[|f(\eta_i, y(\eta_i))-f(\eta_i, 0)|+|f(\eta_i, 0)|] +|\kappa|\Big\}\\ &\le& (L\|y\|+f_0)\Bigg[\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\frac{1}{|\Omega|}\Bigg\{\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)} +\sum\limits_{i=1}^{m}|\sigma_i|\frac{\eta_i^{\rho(\alpha+\beta)}}{\rho^{\alpha+\beta}\Gamma(\alpha+\beta+1)}\Bigg\}\Bigg]+\frac{|\kappa|}{|\Omega|}\\ &\le&(Lr+f_0)\Lambda+\frac{|\kappa|}{|\Omega|}\le r, \end{eqnarray*} |
which, on taking the norm for t\in [0, T], yields \|\mathcal{N}(y)\|\le r. This shows that \mathcal{N} maps B_r into itself. Now we show that the operator \mathcal{N} is a contraction. Let y, u\in \mathcal{G}. Then we get
\begin{eqnarray*} |\mathcal{N}(y)(t)-\mathcal{N}(u)(t)| &\le& {}^{\rho}I_{0^+}^{\alpha}|f(t, y(t))-f(t, u(t))| +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}|f(T, y(T))-f(T, u(T))|\\ &&+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}|f(\eta_i, y(\eta_i))-f(\eta_i, u(\eta_i))|\Big\}\\ &\le& L \Lambda \|y-u\|. \end{eqnarray*} |
Consequently we obtain \|\mathcal{N}(y)-\mathcal{N}(u)\| \le L\Lambda\|y-u\|, which shows that \mathcal{N} is a contraction by means of (3.3). Thus the contraction mapping principle applies and the operator \mathcal{N} has a unique fixed point. This shows that there exists a unique solution for the problem (1.1) on [0, T].
Now we prove the uniqueness of solutions for the problem (1.1) by applying a fixed point theorem for nonlinear contractions due to Boyd and Wong [24].
Definition 3.1. A mapping \mathcal{H}:E\to E is called a nonlinear contraction if we can find a continuous nondecreasing function \phi:\mathbb{R^+} \to \mathbb{R^+} such that \phi(0)=0, \phi(\xi) < \xi for all \xi>0 and \|\mathcal{H}y-\mathcal{H}u\|\leq \phi(\|y-u\|), \forall y, \, u\in E ( E is a Banach space).
Lemma 3.1. (Boyd and Wong) [24] Let E be a Banach space and let N:E\to E be a nonlinear contraction. Then N has a unique fixed point in E.
Theorem 3.3. Assume that
(A_4) \displaystyle |f(t, y)-f(t, u)|\leq g(t) \frac{|y-u|}{G^*+|y-u|}, \, t\in [0, T], \, y, u\geq 0, where g:[0, T]\to \mathbb{R^+} is continuous and
G^*={}^{\rho}I_{0^+}^{\alpha}g(T)+\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}g(T)+\sum\limits_{i=1}^{m}|\sigma_i|{}^{\rho}I_{0^+}^{\alpha+\beta}g(\eta_i)\Big\}. | (3.4) |
Then the problem (1.1) has a unique solution on [0, T].
Proof. Let \Psi:\mathbb{R}^+\to \mathbb{R}^+ be a continuous nondecreasing function such that \Psi(0)=0 and \Psi(\xi) < \xi for all \xi>0, defined by
\Psi(\xi)=\frac{G^* \xi}{G^*+\xi}, \, \, \forall \xi\geq 0. |
Let y, u\in \mathcal{G}. Then
|f(s, y(s))-f(s, u(s))|\leq \frac{g(s)}{G^*} \Psi(\|y-u\|), |
so that
\begin{eqnarray*} | \mathcal{N}(y)(t)-\mathcal{N}(u)(t)| &\le& {}^{\rho}I_{0^+}^{\alpha}\Big(g(t)\frac{|y(t)-u(t)|}{G^*+|y(t)-u(t)|}\Big)\\ &&+\frac{1}{|\Omega|}\Bigg \{{}^{\rho}I_{0^+}^{\alpha}\Big(g(T)\frac{|y(T)-u(T)|}{G^*+|y(T)-u(T)|}\Big)+\sum\limits_{i=1}^{m}|\sigma_i|{}^{\rho}I_{0^+}^{\alpha+\beta}\Big(g(\eta_i)\frac{|y(\eta_i)-u(\eta_i)|}{G^*+|y(\eta_i)-u(\eta_i)|}\Big)\Bigg\}\\ &\le& \frac{|y(t)-u(t)|}{G^*+|y(t)-u(t)|}\Bigg\{{}^{\rho}I_{0^+}^{\alpha}g(T)+\frac{1}{|\Omega|}\Bigg \{{}^{\rho}I_{0^+}^{\alpha}g(T)+\sum\limits_{i=1}^{m}|\sigma_i|{}^{\rho}I_{0^+}^{\alpha+\beta}g(\eta_i)\Bigg\}\Bigg\}, \end{eqnarray*} |
for t\in[0, T]. By the condition (3.4), we deduce that \|\mathcal{N}(y)-\mathcal{N}(u)\|\leq \Psi(\|y-u\|) and hence \mathcal{N} is a nonlinear contraction. Thus it follows from the fixed point theorem due to Boyd and Wong [24] that the operator \mathcal{N} has a unique fixed point in \mathcal{G}, which is indeed a unique solution of problem (1.1).
Example 3.1. Let us consider the following boundary value problem
\left\{\begin{array} {ll} \displaystyle {}^{1/3}_{~~c}D^{5/4}_{0^+}y=f(t, y), \ t\in [0, 2], \\[0.3cm] y(2)=2{}^{1/3}I^{3/4}y(1/2)+1/2~{}^{1/3}I^{3/4}y(3/2)+1/4, ~~~\delta y(0)=0, \end{array} \right. | (3.5) |
where \rho=1/3, \, \alpha=5/4, \, \sigma_1=2, \, \sigma_2=1/2, \, \beta=3/4, \, \eta_1=1/2, \, \eta_2=3/2, \, \kappa=1/4, T=2 and f(t, y(t)) will be fixed later.
Using the given data, we find that |\Omega|=4.543695998 and \Lambda=7.572001575, where \Omega and \Lambda are given by (2.10) and (3.2) respectively.
For illustrating Theorem 3.1, we take
f(t, y)= \frac{(1+t)}{30}\Bigg( \frac{|y|}{|y|+1}+y+\frac{1}{8}\Bigg), | (3.6) |
and find that p(t)=\frac{(1+t)}{30} and \psi(\|y\|)=||y||+\frac{9}{8}. By condition (A_2), we have W>0.7066246467 . Thus, the hypothesis of Theorem 3.1 holds true, which implies that the problem (3.5) has at least one solution.
Furthermore, for the uniqueness results, Theorem 3.2 can be illustrated by choosing
f(t, y)=\frac{\tan^{-1} y+ e^{-t} }{2\sqrt{81+\sin t}}. | (3.7) |
Clearly the condition (A_3) is satisfied with L=1/18. Also
L\Lambda\approx 0.4206667542 \lt 1. |
Obviously all the conditions of Theorem 3.2 hold and consequently the problem (3.5) with f(t, y) given by (3.7) has a unique solution on [0, 2] by the conclusion of Theorem 3.2.
Finally, for illustrating Theorem 3.3, we take
f(t, y)= t\Bigg( \frac{|y|}{|y|+11}+\frac{1}{8}\Bigg). | (3.8) |
Here we choose g(t)=(1+t) and find that
G^*={}^{\rho}I_{0^+}^{\alpha}g(T)+\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}g(T)+\sum\limits_{i=1}^{m}|\sigma_i|{}^{\rho}I_{0^+}^{\alpha+\beta}g(\eta_i)\Big\}\approx 9.923097014, |
and
|f(t, y)-f(t, u)|=t\Bigg( \frac{|y|-|u|}{11+|y|+|u|+\frac{|y||u|}{11}}\Bigg)\leq \frac{(1+t)|y-u|}{9.923097014+|y-u|}. |
So, the conclusion of Theorem 3.3 applies to the problem (3.5) with f(t, y) given by (3.8).
In this section, we present existence results for the problem (1.2).
Definition 4.1. A function y\in AC_\delta^2([0, T], {\mathbb R}) is called a solution of the problem (1.2) if y(T)=\sum_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\beta}y(\eta_i)+\kappa, \delta y(0)=0 and there exists function v\in L^1([0, T], {\mathbb R}) such that v(t)\in F(t, y(t)) a.e. on [0, T] and
\begin{array}{ll} y(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v(t) +\frac{1}{\Omega}\Big\{- {}^{\rho}I_{0^+}^{\alpha}v(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v(\eta_i) +\kappa \Big\}. \end{array} |
Here we prove an existence result for the problem (1.2) by applying nonlinear alternative for Kakutani maps [23] when F has convex values and is of Carathéodory type.
Theorem 4.1. Assume that
(B_1) F :[0, T] \times {\mathbb R} \to {\mathcal{P}_{cp, c}}({\mathbb R}) is L^1-Carathéodory, where {\mathcal P}_{cp, c}({\mathbb R}) =\{Y \in {\mathcal{P}}({\mathbb {\mathbb R}}): Y \mbox{ is compact and convex}\};
(B_2) there exist a continuous nondecreasing function \varphi : [0, \infty) \to (0, \infty) and a function p \in L^1([0, T], {\mathbb R}^+) such that
\|F(t, y)\|_\mathcal{P}:=\sup\{|x|: x \in F(t, y)\}\le p(t)\varphi(\|y\|) ~~ \mbox{for each}~~ (t, y) \in [0, T] \times {\mathbb R}; |
(B_3) there exists a constant \widehat{W}>0 satisfying
\frac{\widehat{W}}{\displaystyle \varphi (\widehat{W})\Big({}^{\rho}I_{0^+}^{\alpha}p(T)+\frac{1}{\Omega}\Big(~^{\rho}I_{0^+}^{\alpha}p(T)+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i)+\kappa \Big)\Big)} \gt 1. |
Then there exists at least one solution for the problem (1.2) on [0, T].
Proof. Define an operator {\mathcal M}: C([0, T], {\mathbb R})\longrightarrow {\mathcal P}(C([0, T], {\mathbb R})) by
{\mathcal M}(y)=\{h \in C([0, T], {\mathbb R}): h(t)=\mathcal{F}(y)(t)\}, | (4.1) |
where
\mathcal{F}(y)(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v(t) +\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v(\eta_i)+\kappa\Big\}, |
for v \in S_{F, y}. Here S_{F, y} denotes the set of selections of F and is defined by
S_{F, y} := \{ v \in L^1([0, T], {\mathbb R}) : v (t) \in F (t, y(t)) ~\mbox{a.e. on} ~ [0, T]\}, |
for each y \in C([0, T], {\mathbb R}). Notice that the fixed points of the operator {\mathcal M} are solutions of the problem (1.2).
To show that {\mathcal M} satisfies the assumptions of Leray-Schauder nonlinear alternative [23], we split the proof in several steps.
Step 1. {\mathcal M}(y) is convex for each y \in C([0, T], {\mathbb R}) as S_{F, y} is convex (F has convex values).
Step 2. Let B_r = \{y \in C([0, T], {\mathbb R}): \|y\| \le r \} be a bounded ball in C([0, T], {\mathbb R}), where r is a positive number. Then, for each h \in {\mathcal M} (y), \, y \in B_{r}, there exists v \in S_{F, y} such that
h(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v(t)+\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v(\eta_i) +\kappa\Big\} |
with
\|h\| \leq \varphi (r)\Big({}^{\rho}I_{0^+}^{\alpha}p(T) +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T)+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i) +|\kappa|\Big\}\Big):=\ell_1. |
This shows that {\mathcal M} maps bounded sets (balls) into bounded sets in C([0, T], {\mathbb R}).
Step 3. In order to show that {\mathcal M} maps bounded sets into equicontinuous sets of C([0, T], {\mathbb R}), we take t_{1}, t_{2}\in (0, T], \, t_{1} < t_{2}, and y\in B_{r}. Then we find that
|h(t_{2})-h(t_{1})| \le\frac{\rho^{1-\alpha}\varphi(r)}{\Gamma(\alpha)}\Big|\int_{0}^{t_1}\Big[\frac{s^{\rho-1}}{(t_2^{\rho}-s^{\rho})^{1-\alpha}}-\frac{s^{\rho-1}}{(t_1^{\rho}-s^{\rho})^{1-\alpha}}\Big] p(s)ds+ \int_{t_1}^{t_2}\frac{s^{\rho-1}}{(t^{\rho}-s^{\rho})^{1-\alpha}}p(s)ds\Bigg|, |
which tends to zero independently of y \in B_{r} as t_2-t_1 \to 0. In view of the foregoing steps, it follows by the Arzelá-Ascoli theorem that {\mathcal M}: C([0, T], {\mathbb R}) \to {\mathcal{P}}(C([0, T], {\mathbb R})) is completely continuous.
Step 4. In our next step, we show that {\mathcal M} is upper semi-continuous (u.s.c.). Since {\mathcal M} is completely continuous, it is enough to establish that it has a closed graph (see [25, Proposition 1.2]). For that, let y_n \to y_*, h_n \in {\mathcal M} (y_n) and h_n \to h_*. Then we have to show that h_* \in {\mathcal M} (y_*). Associated with h_n \in {\mathcal M} (y_n), we can find v_n \in S_{F, y_n} such that for each t \in [0, T],
h_n(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v_n(t)+\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v_n(s)v_n(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v_n(\eta_i)+\kappa \Big\}. |
Next, for each t \in [0, T], we establish that there exists v_* \in S_{F, y_*} satisfying
h_*(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v_*(t)+\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v_*(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v_*(\eta_i)+\kappa \Big\}. |
Consider the linear operator \Theta: L^1([0, T], {\mathbb R}) \to C([0, T], {\mathbb R}) given by
v\mapsto \Theta v(t) =\displaystyle {}^{\rho}I_{0^+}^{\alpha}v(t) +\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v(\eta_i)+\kappa\Big\}. |
Notice that \|h_n(t)-h_*(t)\| \to 0 as n\to \infty. Thus we deduce by the closed graph theorem [26] that \Theta \circ S_F is a closed graph operator. Furthermore, we have h_n(t) \in \Theta(S_{F, y_n}). As y_n \to y_*, we have
h_*(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v_*(t)+\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v_*(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v_*(\xi)+\kappa \Big\}, \, \text{for ~some} ~ v_* \in S_{F, y_*}. |
Step 5. Finally, we show the existence of an open set U\subseteq C([0, T], {\mathbb R}) such that y\notin \lambda {\mathcal M} (y) for any \lambda \in (0, 1) and all y\in \partial U. For that we take \lambda \in (0, 1) and y\in \lambda {\mathcal M} (y). Then there exists v \in L^1([0, T], {\mathbb R}) with v \in S_{F, y} such that, for t \in [0, T], we have
y(t)=\displaystyle \lambda {}^{\rho}I_{0^+}^{\alpha}v(t) +\frac{\lambda}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v(\eta_i)+\kappa\Big\}. |
As in the second step, one can obtain
\begin{eqnarray*} |y(t)| &\le&\displaystyle {}^{\rho}I_{0^+}^{\alpha}|v(T)| +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}|v(T)|+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}|v(\eta_i)| +|\kappa|\Big\}\\ &\leq& \varphi (\|y\|)\Big({}^{\rho}I_{0^+}^{\alpha}p(T)+\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T)+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i) +|\kappa|\Big\}\Big), \end{eqnarray*} |
which implies that
\displaystyle\frac{\|y\| }{\displaystyle \varphi (\|y\|)\Bigg({}^{\rho}I_{0^+}^{\alpha}p(T)+\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}p(T)+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}p(\eta_i) +|\kappa|\Big\}\Bigg)}\leq 1. |
By the hypothesis (B_{3}), we can find \widehat{W} such that \|y\| \ne \widehat{W}. Setting \mathcal{Y} = \{y \in C(J, {\mathbb R}) : \|y\| < \widehat{W}\}, we notice that the operator {\mathcal M} :\overline{\mathcal{Y}} \to \mathcal{P}(C(J, {\mathbb R})) is compact multi-valued, u.s.c. with convex closed values. From the choice of \mathcal{Y}, there does not exist any y \in \partial \mathcal{Y} satisfying y \in \lambda {\mathcal M} (y) for some \lambda \in (0, 1). In consequence, we deduce by the nonlinear alternative of Leray-Schauder type [23] that {\mathcal M} has a fixed point y \in \overline{\mathcal{Y}} which is a solution of the problem (1.2). This completes the proof.
Consider a mapping H_d : {\mathcal{P}}(X) \times {\mathcal{P}}(X) \to {\mathbb R} \cup \{\infty\} defined by
H_d(S, V ) = \max\{ \sup _{s \in S}d(a, V), \sup _{v \in V}d(S, v)\}, |
where d(S, v) = \inf_{s\in S}d(s; v), d(s, V) = \inf_{v\in V}d(s; v) and (X, d) is a metric space induced from the normed space (X; \|\cdot\|). Note that ({\mathcal P}_{cl, b}(X), H_d) is a metric space (see [27]), where {\mathcal P}_{cl, b}(X)=\{Y \in {\mathcal{P}}(X) : Y \mbox{ is closed and bounded}\}.
The following result, dealing with the existence of solutions for the problem (1.2) with nonconvex valued right hand side of the inclusion, relies on Covitz and Nadler's fixed point theorem for multivalued maps [28].
Theorem 4.2. Assume that
(C_1) F : [0, T] \times {\mathbb R} \to {\mathcal P}_{cp}({\mathbb R}) is such that F(\cdot, y) : [0, T] \to {\mathcal P}_{cp}({\mathbb R}) is measurable for each y \in {\mathbb R}, where {\mathcal P}_{cp}({\mathbb R})=\{Y \in {\mathcal{P}}({\mathbb R}) : Y \mbox{ is compact}\};
(C_2) H_d(F(t, y), F(t, \bar{y}))\le \mu(t)|y-\bar{y}| for almost all t \in [0, T] and y, \bar{y} \in {\mathbb R} with \mu \in C([0, T], {\mathbb R}^+) and d(0, F(t, 0))\le \mu(t) for almost all t \in [0, T].
Then the problem (1.2) has at least one solution on [0, T] provided that
\vartheta=\|\mu\|\Lambda \lt 1, | (4.2) |
where \Lambda is given by (3.2).
Proof. By the assumption (C_1), the set S_{F, y} is nonempty for each y \in C([0, T], {\mathbb R}) and F has a measurable selection by Theorem Ⅲ.6 in [29]. Now we proceed to show that the operator {\mathcal M} : (C [0, T], {\mathbb R}) \to {\mathcal P}_{cl}(C([0, T], {\mathbb R})) ({\mathcal P}_{cl}(C([0, T], {\mathbb R}))= \{Y \in P({C([0, T], {\mathbb R})}) : Y~\mbox{ is closed} \}) is a contraction so that Covitz and Nadler's Theorem [28] is applicable.
In the first step, we show that {\mathcal M}(y) \in {\mathcal P}_{cl}((C [0, T], {\mathbb R})) for each y \in C([0, T], {\mathbb R}). Let \{u_n\}_{n \ge 0} \in {\mathcal M}(y) with u_n \to u ~(n \to \infty) in C([0, T], {\mathbb R}). Then u \in C([0, T], {\mathbb R}) and there exists v_n \in S_{F, y_n} satisfying
u_n(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v_n(t) +\frac{1}{\Omega}\Big\{- {}^{\rho}I_{0^+}^{\alpha}v_n(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v_n(\eta_i)+\kappa\Big\} \, \, \text{for each} ~ t \in [0, T]. |
In view of the compact values of F, we pass onto a subsequence (if necessary) to find that v_n converges to v in L^1 ([0, T], {\mathbb R}). For v \in S_{F, y} and for each t \in [0, T], we have
u_n(t)\to u(t) =\displaystyle {}^{\rho}I_{0^+}^{\alpha}v(t) +\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v(\eta_i)+\kappa\Big\}. |
Thus u \in {\mathcal M}(y).
Now, for each y, \bar{y} \in C([0, T], R), we establish that there exists \vartheta < 1 (defined by (4.2)) satisfying
H_d({\mathcal M}(y), {\mathcal M}(\bar{y}))\le\vartheta \|y-\bar{y}\|. |
Let y, \bar{y} \in C([0, T], {\mathbb R}) and h_1 \in {\mathcal M}(y). Then there exists v_1(t) \in F(t, y(t)) satisfying
h_1(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v_1(t)+\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v_1(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v_1(\eta_i)+\kappa \Big\}, |
for each t \in [0, T]. By (C_2), we have
H_d(F(t, y), F(t, \bar{y}))\le \mu(t)|y(t)-\bar{y}(t)|. |
Therefore, we can find w \in F(t, \bar{y}(t)) satisfying
|v_1(t)-w|\le \mu(t)|y(t)-\bar{y}(t)|, ~~t \in [0, T]. |
Introduce \mathcal{U} : [0, T] \to \mathcal{P}({\mathbb R}) by
\mathcal{U}(t)=\{w \in {\mathbb R} : |v_1(t)-w|\le \mu(t)|y(t)-\bar{y}(t)|\}. |
As \mathcal{U}(t)\cap F(t, \bar{y}(t)) is measurable (Proposition Ⅲ.4 [29]), we can find a measurable selection v_2(t) for \mathcal{U} such that v_2(t) \in F(t, \bar{y}(t)) satisfying |v_1(t)-v_2(t)|\le \mu(t)|y(t)-\bar{y}(t)| for each t \in [0, T].
Define
h_2(t)=\displaystyle {}^{\rho}I_{0^+}^{\alpha}v_2(t)+\frac{1}{\Omega}\Big\{-{}^{\rho}I_{0^+}^{\alpha}v_2(T)+\sum\limits_{i=1}^{m}\sigma_i~{}^{\rho}I_{0^+}^{\alpha+\beta}v_2(\eta_i)+\kappa \Big\}, |
for each t \in [0, T]. Then
\begin{array}{ll} |h_1(t)- h_2(t)| &\le \displaystyle {}^{\rho}I_{0^+}^{\alpha}|v_1(t)-v_2(t)| +\frac{1}{|\Omega|}\Big\{{}^{\rho}I_{0^+}^{\alpha}|v_1(T)-v_2(T)|+\sum\limits_{i=1}^{m}|\sigma_i|~{}^{\rho}I_{0^+}^{\alpha+\beta}|v_1(\eta_i)-v_2(\eta_i)| \Big\}\\ &\displaystyle \leq \|\mu\|\Bigg[\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\frac{1}{|\Omega|}\Bigg\{\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\sum\limits_{i=1}^{m}|\sigma_i|\frac{\eta_i^{\rho(\alpha+\beta)}}{\rho^{\alpha+\beta}\Gamma(\alpha+\beta+1)}\Bigg\}\Bigg]\|y-\bar{y}\|. \end{array} |
Hence
\| h_1-h_2\| \le \|\mu\|\Bigg[\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\frac{1}{|\Omega|}\Bigg\{\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\sum\limits_{i=1}^{m}|\sigma_i|\frac{\eta_i^{\rho(\alpha+\beta)}}{\rho^{\alpha+\beta}\Gamma(\alpha+\beta+1)}\Bigg\}\Bigg]\|y-\bar{y}\|. |
Analogously, switching the roles of y and \overline{y}, we can obtain
H_d({\mathcal M}(y), {\mathcal M}(\bar{y})) \le \|\mu\|\Bigg[\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\frac{1}{|\Omega|}\Bigg\{\frac{T^{\rho \alpha}}{\rho^{\alpha}\Gamma(\alpha+1)}+\sum\limits_{i=1}^{m}|\sigma_i|\frac{\eta_i^{\rho(\alpha+\beta)}}{\rho^{\alpha+\beta}\Gamma(\alpha+\beta+1)}\Bigg\}\Bigg]\|y-\bar{y}\|. |
So {\mathcal M} is a contraction. Thus, by Covitz and Nadler's fixed point theorem [28], the operator {\mathcal M} has a fixed point y, which corresponds to a solution of (1.2).
Example 4.1. Consider the following boundary value problem
\left\{\begin{array} {ll} \displaystyle {}^{1/3}_{~~c}D^{5/4}_{0^+}y\in F(t, y), \ t\in [0, 2], \\[0.3cm] y(2)=2{}^{1/3}I^{3/4}y(1/2)+1/2~{}^{1/3}I^{3/4}y(3/2)+1/4, ~~~\delta y(0)=0, \end{array} \right. | (4.3) |
where F(t, y) will be fixed later.
For illustrating Theorem 4.1, we take
F(t, y)=\Bigg[\frac{e^{-t} }{\sqrt{900+t}}\Bigg(\sin y+\frac{1}{2}\Bigg)~, \, \frac{(1+t)}{30}\Bigg(\frac{|y|}{|y|+1}+y+\frac{1}{8}\Bigg)\Bigg]. | (4.4) |
Using the given data, we find \displaystyle p(t)=\frac{(1+t)}{30}, \varphi(\|y\|)=||y||+\frac{9}{8}, and by condition (B_3), we have \widehat{W}> 0.7066246467. Thus all conditions of Theorem 4.1 are satisfied and consequently, there exists at least one solution for the problem (4.3) with F(t, y) given by (4.4) on [0, 2].
In order to demonstrate the application of Theorem 4.2, let us choose
F(t, y)=\Bigg[\frac{e^{-t }}{\sqrt{900+t}}\Bigg(\tan^{-1} y+\frac{1}{2}\Bigg)~, \, \frac{(1+t)}{30}\Bigg(\frac{|y|}{|y|+1}+\frac{1}{8}\Bigg)\Bigg]. | (4.5) |
Clearly
H_d(F(t, y), F(t, \bar{y}))\leq \frac{(t+1)}{30} \|y-\bar{y}\|. |
Letting \displaystyle \mu(t)= \frac{(t+1)}{30}, it is easy to check that d(0, F(t, 0))\le \mu(t) holds for almost all t \in [0, 2] and \vartheta \approx 0.7572001575 < 1 (\vartheta is given by 4.2). As the hypotheses of Theorem 4.2 are satisfied, we conclude that the problem (4.3) with F(t, y) given by (4.5) has at least one solution on [0, 2].
We have developed the existence theory for fractional differential equations and inclusions involving Caputo-type generalized fractional derivative equipped with generalized fractional integral boundary conditions (in the sense of Katugampola). Standard fixed point theorems for single-valued and multi-valued maps are employed to obtain the desired results, which are well illustrated with the aid of examples. Our results are new in the given configuration and contribute significantly to the existing literature on the topic.
The authors gratefully acknowledge the referees for their useful comments on their paper.
The authors declare that they have no conflict of interests.
[1] | N. Laskin, Fractional quantum mechanics and Levy path integrals, Phys. Lett. A, 268 (2000), 298-305. |
[2] | N. Laskin, Fractals and quantum mechanics, Chaos, 10 (2000), 780-790. |
[3] | B. N. N. Achar, B. T. Yale, J. W. Hanneken, Time fractional Schr\"odinger equation revisited, Adv. Math. Phys., 2018 (2013), 1-11. |
[4] |
D. Baleanu, G. C. Wu, S. D. Zeng, Chaos analysis and asymptotic stability of generalized Caputo fractional differential equations, Chaos Soliton. Fract., 102 (2017), 99-105. doi: 10.1016/j.chaos.2017.02.007
![]() |
[5] |
G. C. Wu, D. Baleanu, Z. G. Deng, et al. Lattice fractional diffusion equation in terms of a Riesz-Caputo difference, Physica A-statistical Mechanics and Its Applications, 438 (2015), 335-339. doi: 10.1016/j.physa.2015.06.024
![]() |
[6] | U. N. Katugampola, New Approach to a generalized fractional integral, Appl. Math. Comput., 218 (2011), 860-865. |
[7] | I. Podlubny, Fractional Differential Equations, Academic Press, San Diego, 1999. |
[8] | A. A. Kilbas, H. M. Srivastava, J. J. Trujillo, Theory and Applications of Fractional Differential Equations, Vol. 204, North-Holland Mathematics Studies, Elsevier Science B.V., Amsterdam, 2006. |
[9] | B. Ahmad, A. Alsaedi, S. K. Ntouyas, et al. Hadamard-type fractional differential equations, inclusions and inequalities, Springer, Cham, 2017. |
[10] |
M. Benchohra, J. Henderson, S. K. Ntouyas, et al. Existence results for fractional order functional differential equations with infinite delay, J. Math. Anal. Appl., 338 (2008), 1340-1350. doi: 10.1016/j.jmaa.2007.06.021
![]() |
[11] |
J. R. Wang, Y. Zhang, Analysis of fractional order differential coupled systems, Math. Methods Appl. Sci., 38 (2015), 3322-3338. doi: 10.1002/mma.3298
![]() |
[12] |
B. Ahmad, S. K. Ntouyas, J. Tariboon, A study of mixed Hadamard and Riemann-Liouville fractional integro-differential inclusions via endpoint theory, Appl. Math. Lett., 52 (2016), 9-14. doi: 10.1016/j.aml.2015.08.002
![]() |
[13] |
B. Ahmad, R. Luca, Existence of solutions for a sequential fractional integro-differential system with coupled integral boundary conditions, Chaos Soliton. Fract., 104 (2017), 378-388. doi: 10.1016/j.chaos.2017.08.035
![]() |
[14] | Z. Y. Gao, J. R. Wang, Y. Zhou, Analysis of a class of fractional nonlinear multidelay differential systems, Discrete Dyn. Nat. Soc., 2017 (2017), 1-15. |
[15] | B. Ahmad, S. K. Ntouyas, Existence results for fractional differential inclusions with Erdelyi-Kober fractional integral conditions, An. Ştiinţ. Univ. "Ovidius" Constanţa Ser. Mat., 25 (2017), 5-24. |
[16] | M. Shabibi, Sh. Rezapour, S. M. Vaezpour, A singular fractional integro-differential equation, Politehn. Univ. Bucharest Sci. Bull. Ser. A Appl. Math. Phys., 79 (2017), 109-118. |
[17] | M. Xu, Z. Han, Positive solutions for integral boundary value problem of two-term fractional differential equations, Bound. Value Probl., 2018 (2018), 100. |
[18] |
J. Henderson, R. Luca, Positive solutions for a system of coupled fractional boundary value problems, Lith. Math. J., 58 (2018), 15-32. doi: 10.1007/s10986-018-9385-4
![]() |
[19] | R. Almeida, A. B. Malinowska, T. Odzijewicz, Fractional differential equations with dependence on the Caputo-Katugampola derivative, J. Comput. Nonlinear Dynam., 11 (2016), 61017. |
[20] |
B. Ahmad, M. Alghanmi, S. K. Ntouyas, et al. Fractional differential equations involving generalized derivative with Stieltjes and fractional integral boundary conditions, Appl. Math. Lett., 84 (2018), 111-117. doi: 10.1016/j.aml.2018.04.024
![]() |
[21] | U. N. Katugampola, A new approach to generalized fractional derivatives, Bull. Math. Anal. Appl., 6 (2014), 1-15. |
[22] |
F. Jarad, T. Abdeljawad, D. Baleanu, On the generalized fractional derivatives and their caputo modification, J. Nonlinear Sci. Appl., 10 (2017), 2607-2619. doi: 10.22436/jnsa.010.05.27
![]() |
[23] | A. Granas, J. Dugundji, Fixed Point Theory, Springer-Verlag, New York, 2003. |
[24] | D. W. Boyd and J. S. W. Wong, On nonlinear contractions, Proc. Am. Math. Soc., 20 (1969), 485-464. |
[25] | K. Deimling, Multivalued Differential Equations, Walter De Gruyter, Berlin-New York, 1992. |
[26] | A. Lasota, Z. Opial, An application of the Kakutani-Ky Fan theorem in the theory of ordinary differential equations, Bull. Acad. Polon. Sci. Ser. Sci. Math. Astronom. Phys., 13 (1965), 781-786. |
[27] | M. Kisielewicz, Stochastic Differential Inclusions and Applications. Springer Optimization and Its Applications, Vol. 80, Springer, New York, 2013. |
[28] |
H. Covitz and S. B. Nadler Jr., Multivalued contraction mappings in generalized metric spaces, Israel J. Math., 8 (1970), 5-11. doi: 10.1007/BF02771543
![]() |
[29] | C. Castaing and M. Valadier, Convex Analysis and Measurable Multifunctions, Vol. 580, Lecture Notes in Mathematics, Springer-Verlag, Berlin-Heidelberg-New York, 1977. |
1. | Subramanian Muthaiah, Dumitru Baleanu, Existence of Solutions for Nonlinear Fractional Differential Equations and Inclusions Depending on Lower-Order Fractional Derivatives, 2020, 9, 2075-1680, 44, 10.3390/axioms9020044 | |
2. | Mehmet Giyas Sakar, Onur Saldır, A New Reproducing Kernel Approach for Nonlinear Fractional Three-Point Boundary Value Problems, 2020, 4, 2504-3110, 53, 10.3390/fractalfract4040053 | |
3. | Bashir Ahmad, Badrah Alghamdi, Ahmed Alsaedi, Sotiris K. Ntouyas, Existence results for Riemann-Liouville fractional integro-differential inclusions with fractional nonlocal integral boundary conditions, 2021, 6, 2473-6988, 7093, 10.3934/math.2021416 | |
4. | Muthaiah Subramanian, Shorog Aljoudi, Existence and Ulam–Hyers Stability Analysis for Coupled Differential Equations of Fractional-Order with Nonlocal Generalized Conditions via Generalized Liouville–Caputo Derivative, 2022, 6, 2504-3110, 629, 10.3390/fractalfract6110629 | |
5. | Jinsheng Du, Cuizhi Lu, Yirong Jiang, Heng Xie, Uniqueness and stability for generalized Caputo fractional differential quasi-variational inequalities, 2021, 0, 2191-0294, 10.1515/ijnsns-2020-0294 | |
6. | Jeffrey W. Lyons, Differentiation of Solutions of Caputo Boundary Value Problems with Respect to Boundary Data, 2024, 12, 2227-7390, 1790, 10.3390/math12121790 | |
7. | Jeffrey W. Lyons, CONTINUOUS DEPENDENCE ON BOUNDARY CONDITIONS FOR CAPUTO FRACTIONAL DIFFERENTIAL EQUATIONS, 2024, 54, 0035-7596, 10.1216/rmj.2024.54.823 | |
8. | Keyu Zhang, Jiafa Xu, Solvability for a system of Hadamard-type hybrid fractional differential inclusions, 2023, 56, 2391-4661, 10.1515/dema-2022-0226 | |
9. | Madeaha Alghanmi, Shahad Alqurayqiri, Existence Results of Nonlocal Fractional Integro-Neutral Differential Inclusions with Infinite Delay, 2025, 9, 2504-3110, 46, 10.3390/fractalfract9010046 |