Solvability of a fractional integral equation with

the concept of measure of noncompactness

Kishin Sadarangani Bessem Samet

Abstract
We study the existence of solutions for a new class of integral equations
involving a fractional integral with respect to another function. Our tech-
niques are based on the measure of non-compactness concept combined with
a generalized version of Darbo’s theorem. Some examples are presented to
illustrate the obtained results.

1 Introduction
Many problems in several branches of science such as engineering, physics,
biology and other disciplines can be modeled as an integral equation (see, for
example [10, 23, 32, 34, 35]). The most used techniques to study the existence of
solutions to an integral equation are based on fixed point arguments. For exam-
ple, Banach contraction principle was used by several authors in order to estab-
lish existence results for different kinds of integral equations (see, for example
[27, 29, 30, 33]). Similarly, many solvability results were derived via Schauder’s
fixed point theorem (see, for example [14, 25, 26]). On the other hand, in the
absence of compacity and the Lipschitz condition, the above mentioned theorems
cannot be applied. In such situation, the measure of non-compactness concept
can be useful in order to avoid the mentioned problems.
Several existence results for integral equations were established using fixed
point theorems involving measures of non-compactness. For more details, we
Received by the editors in May 2016 - In revised form in August 2016.
Communicated by E. Colebunders.
2010 Mathematics Subject Classification : 45G05, 26A33, 74H20.
Key words and phrases : Integral equation; fractional integral with respect to another function;
Riemann-Liouville fractional integral; Erdélyi-Kober fractional integral; Hadamard fractional in-
tegral; measure of non-compactness; Darbo’s theorem.

Bull. Belg. Math. Soc. Simon Stevin 24 (2017), 135–151

136 K. Sadarangani – B. Samet

refer the reader to [1, 3, 4, 5, 7, 8, 9, 12, 13, 15, 20, 21, 28] and the references therein.
On the other hand, due to the importance of fractional calculus in modelling
various problems from many fields of science and engineering, a great attention
has been focused recently on the study of fractional integral equations (see, for
example [2, 6, 11, 16, 17, 18, 24]).
In this paper, we deal with a new class of fractional integral equations involv-
ing a fractional integral with respect to another function. Using a fixed point the-
orem involving measures of non-compactness, an existence result is established.
Some numerical examples are also provided.

2 Problem formulation and preliminaries

The main purpose of this paper is to study the following integral equation
Z t ′
f 2 (t, x (t)) g (τ )k(τ, x (τ ))
x (t) = f 1 (t, x (t)) + dτ, t ∈ [ a, T ], (2.1)
Γ(α ) a ( g(t) − g(τ ))1−α

where α ∈ (0, 1), 0 ≤ a < T, k : [ a, T ] × R → R, g : [ a, T ] → R and

f i : [ a, T ] × R → R (i = 1, 2), from the point of view of the theory of existence of
solutions. Eq.(2.1) can be written in the form
α
x (t) = f 1 (t, x (t)) + f 2 (t, x (t)) Ia,g k(·, x (·))(t), t ∈ [ a, T ],
α is the fractional integral of order α with respect to the function g
where Ia,g
defined by
Z t
α 1 g′ (τ )ψ (τ )
Ia,g ψ (t) = dτ, t ∈ [ a, T ].
Γ(α ) a ( g(t) − g(τ ))1−α

In the particular case g(τ ) = τ, Eq.(2.1) models some problems in the queuing
α is the Riemann-Liouville fractional
theory and biology (see [19]). In such case, Ia,g
integral defined by
Z t
1 ψ (τ )
Iaα ψ(t) = dτ, t ∈ [ a, T ]. (2.2)
Γ(α ) a ( t − τ )1− α

In the case g(τ ) = τ β (β > 0), Ia,g

α is the Erdélyi-Kober fractional integral defined

by
Z t
α β τ β−1 ψ (τ )
Ia,β ψ(t) = dτ, t ∈ [ a, T ]. (2.3)
Γ ( α ) a ( t β − τ β )1− α
α is the Hadamard fractional
However, in the case g(τ ) = ln τ and a > 0, Ia,g
integral defined by
Z t α−1
1 t ψ (τ )
Jaα ψ(t) = ln dτ, t ∈ [ a, T ]. (2.4)
Γ(α ) a τ τ

For more details on fractional calculus, we refer the reader to [31].

Solvability of a fractional integral equation... 137

Using a measure of non-compactness in the set C([ a, T ]; R ) of real continuous

functions in [ a, T ] introduced by Banaś and Goebel [8], we give sufficient con-
ditions for the existence of at least one solution to Eq.(2.1). To the best of our
knowledge, an integral equation of the type (2.1) has not been considered earlier.
Before presenting and proving the main results of this paper, we need to recall
some basic concepts and present some preliminaries results that will be useful
later.
Let ( H, k · k) be a given Banach space. We denote by B H the family of all
nonempty bounded subsets of H. A mapping η : B H → [0, ∞) is said to be a
measure of non-compactness (see [8]) if it satisfies the following axioms:
(A1) for all B ∈ B H , we have

η ( B) = 0 =⇒ B is precompact;

(A2) for every pair ( B1 , B2 ) ∈ B H × B H , we have

B1 ⊆ B2 =⇒ η ( B1 ) ≤ η ( B2 );

(A3) for every B ∈ B H ,

η ( B ) = η ( B) = η (coB),
where coB denotes the closed convex hull of B;
(A4) for every pair ( B1 , B2 ) ∈ B H × B H and λ ∈ (0, 1), we have

η (λB1 + (1 − λ) B2 ) ≤ λη ( B1 ) + (1 − λ)η ( B2 );

(A5) if { Bn } is a sequence of closed and decreasing (w.r.t ⊆) sets in B H such that

∞
\
η ( Bn ) → 0 as n → ∞, then B∞ := Bn is nonempty and compact.
n =1
A function ϕ : [0, ∞) → [0, ∞) is said to be a comparison function if it satisfies
the following properties:
(i) ϕ is non-decreasing;
(ii) lim ϕ(n) (t) = 0, for all t > 0, where ϕ(n) denotes the n-iteration of ϕ.
n→∞
Examples of comparison functions are
t
ϕ(t) = λ t, λ ∈ [0, 1); ϕ(t) = arctan t; ϕ(t) = ln(1 + t); ϕ(t) = .
1+t
Through this paper, we denote by Φ the set of comparison functions.
In [5], Aghajani et al. proved the following useful result in order to prove that
a function is a comparison function.

Lemma 2.1. Let ϕ : [0, ∞) → [0, ∞) be a non-decreasing and upper semi-continuous

function. Then the following conditions are equivalent:
(i) lim ϕ(n) (t) = 0 for all t > 0.
n→∞
(ii) ϕ(t) < t for all t > 0.

In the same paper, the authors proved the following generalization of Darbo’s
theorem (see [1]).
138 K. Sadarangani – B. Samet

Lemma 2.2. Let C be a nonempty, bounded, closed and convex subset of a Banach space
H. Let U : C → C be a continuous mapping such that

η (UW ) ≤ ϕ(η (W )), W ⊆ C,

where ϕ ∈ Φ and η is a measure of non-compactness in H. Then U has at least one fixed

point.
For given functions ϕ, ψ : [0, ∞) → R and a real number λ, define the opera-
tors Tmax and T by

Tmax ( ϕ, ψ)(t) = max{ ϕ(t), ψ(t)}, t≥0

and
T (λ, ϕ)(t) = λϕ(t), t ≥ 0.
e the
Denote by C([0, ∞); R ) the set of real continuous functions in [0, ∞) and by Φ
set Φ ∩ C([0, ∞); R ).
The following lemma is interesting for our purpose.
Lemma 2.3. The following properties hold:
(i) Tmax (Φ e ×Φ e ) ⊆ Φ.
e
e ) ⊆ Φ.
(ii) T ([0, 1] × Φ e
e × Φ.
Proof. Let ( ϕ, ψ) ∈ Φ e We have to prove that Tmax ( ϕ, ψ) : [0, ∞) → [0, ∞)
e Since the maximum of a finite number of continuous functions is
belongs to Φ.
e×
continuous, we have Tmax ( ϕ, ψ) ∈ C([0, ∞); R ). Moreover, since ( ϕ, ψ) ∈ Φ
e
Φ ⊆ Φ × Φ, then ϕ and ψ are non-decreasing functions, which yields Tmax ( ϕ, ψ)
is also a non-decreasing function. From Lemma 2.1, we have

ϕ(t) < t, ψ(t) < t, t > 0.

Then
Tmax ( ϕ, ψ)(t) < t, t > 0.
Again, by Lemma 2.1, we get

lim [ Tmax ( ϕ, ψ)](n) (t) = 0, t > 0.

n→∞

e
Therefore, Tmax ( ϕ, ψ) ∈ Φ. As consequence, Tmax ( ϕ, ψ) ∈ Φ ∩ C([0, ∞); R ) = Φ,
which proves (i). A similar argument can be used to prove (ii).

Let H = C([ a, T ]; R ) be the set of real continuous functions in [ a, T ]. Such a

set is a Banach space with respect to the norm

kvk = max{|v(t)| : t ∈ [ a, T ]}, v ∈ C([ a, T ]; R ).

We will use the following notations through this paper. Let B ∈ B H . For v ∈ B
and ρ ≥ 0, set

ω (v, ρ) = sup{|v(t) − v(s)| : t, s ∈ [ a, T ], |t − s| ≤ ρ}.

Solvability of a fractional integral equation... 139

We define the mapping Ω : B H × [0, ∞) → [0, ∞) by

Ω( B, ρ) = sup{ω (v, ρ) : v ∈ B}, ( B, ρ) ∈ B H × [0, ∞).

It was proved in [8] that the mapping η : B H → [0, ∞) defined by

η ( B) = lim Ω( B, ρ), B ∈ BH
ρ → 0+

is a measure of non-compactness in the Banach space H.

Now, we are ready to present our main results. This is the aim of the next
section.

3 Main results
Our first existence result is obtained under the following hypotheses:
(H1) The functions f i : [ a, T ] × R → R (i = 1, 2) are continuous.
(H2) There exist functions ϕ1 , ϕ2 : [0, ∞) → [0, ∞) such that
(i) ϕi (0) = 0, i = 1, 2.
(ii) ϕ1 is non-decreasing and continuous.
e
(iii) ϕ2 ∈ Φ.
(iv) There exist δ ≥ 0 and θ ∈ Φ e such that

ϕ1 (t) ≤ δθ (t), t ≥ 0.

(v) For all (t, u, v) ∈ [ a, T ] × R × R, we have

| f i (t, u) − f i (t, v)| ≤ ϕi (|u − v|), i = 1, 2.

(H3) The function k : [ a, T ] × R → R is continuous and it satisfies

|k(t, u)| ≤ ϕ(|u|), (t, u) ∈ [ a, T ] × R,

where ϕ : [0, ∞) → [0, ∞) is a non-decreasing function.

(H4) g ∈ C([ a, T ]; R ) ∩ C1 ((a, T ]; R ) with g′ (t) > 0, for all t ∈ (a, T ].
(H5) There exists r0 > 0 such that

( ϕ1 (r0 ) + f 1∗ )Γ(α + 1) + ( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 )( g(T ) − g(a))α ≤ r0 Γ(α + 1)

and, moreover
ϕ (r 0 )
δ+ ( g(T ) − g(a))α ≤ 1,
Γ ( α + 1)
where
f i∗ = max{| f i (t, 0)| : t ∈ [ a, T ]}, i = 1, 2.
Our main result is the following.
140 K. Sadarangani – B. Samet

Theorem 3.1. Under the assumptions (H1)-(H5), Eq.(2.1) has at least one solution
y∗ ∈ C([ a, T ]; R ). Moreover, we have ky∗ k ≤ r0 .
Proof. Let us consider the operator U defined on H = C([ a, T ]; R ) by
Z t ′
f (t, x (t)) g (τ )k(τ, x (τ ))
(Ux )(t) = f 1 (t, x (t)) + 2 dτ, ( x, t) ∈ H × [ a, T ].
Γ(α ) a ( g(t) − g(τ ))1−α
(3.1)
At first, we show that the operator U maps H into itself. Set
Z t ′
g (τ )k(τ, x (τ ))
(Gx )(t) = dτ, ( x, t) ∈ H × [ a, T ]. (3.2)
a ( g(t) − g(τ ))1−α
From assumption (H1), we have just to prove that G maps H into itself, that is,
Gx : [ a, T ] → R is continuous for every x ∈ H. Observe that Gx is well-defined.
In fact, for all t ∈ [ a, T ], from assumption (H3), we have
Z t
g′ (τ ) ϕ(k x k)
|(Gx )(t)| ≤ ϕ(k x k) 1 − α
dτ = ( g(t) − g(a))α < ∞. (3.3)
a ( g(t) − g(τ )) α
Let us check the continuity of Gx at the point a. Let {tn } be a sequence in [ a, T ]
such that tn → a+ as n → ∞. Using (3.3), for all n ∈ N, we have
Z tn
g′ (τ ) ϕ(k x k)
|(Gx )(tn )| ≤ ϕ(k x k) 1 − α
dτ = ( g(tn ) − g(a))α .
a ( g(tn ) − g(τ )) α
Letting n → ∞ in the above inequality and using the continuity of g, we get
lim (Gx )(tn ) = 0 = (Gx )(a),
n→∞
which proves the continuity of Gx at the point a. Now, let t ∈ (a, T ] be fixed and
{tn } be a sequence in (a, T ] such that tn → t as n → ∞. Without restriction of the
generality, we may assume that tn ≥ t for n large enough. We have
|(Gx )(tn ) − (Gx )(t)| =
Z t Z t ′ 
 n g′ (τ )k(τ, x (τ )) g ( τ ) k ( τ, x ( τ )) 
 
 a ( g(t ) − g(τ ))1−α dτ − a ( g(t) − g(τ ))1−α dτ  , n ∈ N.
n
For n large enough, we can write
|(Gx )(tn ) − (Gx )(t)|
Z t  ′  

 g (τ )k(τ, x (τ )) g′ (τ )k(τ, x (τ )) 
≤  1 − α
− 1 − α
dτ 
a ( g(tn ) − g(τ )) ( g(t) − g(τ ))
Z t ′

 n g (τ )k(τ, x (τ )) 
+  dτ 

t ( g(tn ) − g(τ ))1− α
Z t 
g′ (τ ) g′ (τ )
≤ ϕ(k x k) − dτ
a ( g(t) − g(τ ))1−α ( g(tn ) − g(τ ))1−α
Z tn
g′ (τ )
+ ϕ(k x k) dτ
t ( g(tn ) − g(τ ))1− α
ϕ(k x k)
= (( g(t) − g(a))α + ( g(tn ) − g(t))α − ( g(tn ) − g(a))α )
α
ϕ(k x k)
+ ( g(tn ) − g(t))α .
α
Solvability of a fractional integral equation... 141

Since g is continuous in [ a, T ], we have

ϕ(k x k)
lim (( g(t) − g(a))α + ( g(tn ) − g(t))α − ( g(tn ) − g(a))α ) +
n→∞ α
ϕ(k x k)
( g(tn ) − g(t))α = 0,
α
which yields lim |(Gx )(tn ) − (Gx )(t)| = 0. Then Gx is continuous at t. As
n→∞
consequence, Gx ∈ H, for all x ∈ H, and U : H → H is well-defined.
On the other hand, for an arbitrarily fixed x ∈ H and t ∈ [ a, T ], we have
Z
| f (t, x (t))| t g′ (τ )|k(τ, x (τ ))|
|(Ux )(t)| ≤ | f 1 (t, x (t))| + 2 dτ
Γ(α ) a ( g(t) − g(τ ))1− α
≤ | f 1 (t, x (t)) − f 1 (t, 0)| + | f 1 (t, 0)|
Z
| f 2 (t, x (t)) − f 2 (t, 0)| + | f 2 (t, 0)| t g′ (τ ) ϕ(| x (τ )|)
+ 1− α
dτ
Γ(α ) a ( g(t) − g(τ ))
( ϕ2 (| x (t)|) + f 2∗ ) ϕ(k x k)
≤ ϕ1 (| x (t)|) + f 1∗ + ( g(t) − g(a))α
Γ ( α + 1)
( ϕ2 (k x k) + f 2∗ ) ϕ(k x k)
≤ ϕ1 (k x k) + f 1∗ + ( g(T ) − g(a))α .
Γ ( α + 1)
Then
( ϕ2 (k x k) + f 2∗ ) ϕ(k x k)
kUx k ≤ ϕ1 (k x k) + f 1∗ + ( g(T ) − g(a))α , x ∈ H.
Γ ( α + 1)
From the above inequality, the fact that the functions ϕ1 , ϕ2 , ϕ : [0, ∞) → [0, ∞)
are non-decreasing, and assumption (H5), we infer that the operator U maps
B(0, r0 ) into itself, where

B(0, r0 ) = { x ∈ H : k x k ≤ r0 }.

Now, we claim that the operator U : B(0, r0 ) → B(0, r0 ) is continuous. We

write U in the form
1
Ux = F1 x + F2 x + GX, x ∈ B(0, r0 ),
Γ(α )
where
( Fi x )(t) = f i (t, x (t)), ( x, t) ∈ B(0, r0 ) × [ a, T ], i = 1, 2
and G is defined by (3.2). In order to prove our claim, it is sufficient to show that
the operators Fi (i = 1, 2) and G are continuous on B(0, r0 ). Firstly, we show that
Fi are continuous operators on B(0, r0 ). To do this, we take a sequence { xn } ⊂
B(0, r0 ) and x ∈ B(0, r0 ) such that k xn − x k → 0 as n → ∞, and we have to prove
that k Fi xn − Fi x k → 0 as n → ∞. In fact, for all t ∈ [ a, T ], we have
|( Fi xn )(t) − ( Fi x )(t)| = | f i (t, xn (t)) − f i (t, x (t))|
≤ ϕi (| xn (t) − x (t)|)
≤ ϕi (k xn − x k).
142 K. Sadarangani – B. Samet

Then

k Fi xn − Fi x k ≤ ϕi (k xn − x k), n ∈ N.

Letting n → ∞ in the above inequality, using the continuity of the functions ϕi

(i = 1, 2) and the fact that ϕi (0) = 0, we get

 
lim k Fi xn − Fi x k ≤ ϕi lim k xn − x k = ϕi (0) = 0.
n→∞ n→∞

This proves that Fi is a continuous operator on B(0, r0 ), for all i = 1, 2. Next, we

show that G is a continuous operator on B(0, r0 ). To do this, we fix a real number
ε > 0 and we take arbitrary functions x, y ∈ B(0, r0 ) such that k x − yk < ε. For
all t ∈ [ a, T ], we have

Z t ′
g (τ )|k(τ, x (τ )) − k(τ, y(τ ))|
|(Gx )(t) − (Gy)(t)| ≤ dτ
a ( g(t) − g(τ ))1−α
Z t
g′ (τ )
≤ u (r 0 , ε ) dτ
a ( g(t) − g(τ ))1− α
u (r 0 , ε )
= ( g(t) − g(a))α
α
u (r 0 , ε )
≤ ( g(T ) − g(a))α ,
α

where

u(r0 , ε) = sup{|k(τ, v) − k(τ, w)| : τ ∈ [ a, T ], v, w ∈ [−r0 , r0 ], |v − w| < ε}.

Therefore,
u (r 0 , ε )
kGx − Gyk ≤ ( g(T ) − g(a))α .
α

Since k is uniform continuous on the compact [ a, T ] × [−r0 , r0 ], we have u(r0 , ε) →

0 as ε → 0+ and, therefore, the last inequality gives us

lim k Gx − Gyk = 0.
ε → 0+

Then G is continuous on B(0, r0 ) and U maps continuously the set B(0, r0 ) into
itself.
Further, let W be a nonempty subset of B(0, r0 ). Let ρ > 0 be fixed, x ∈ W and
t1 , t2 ∈ [ a, T ] be such that |t1 − t2 | ≤ ρ. Without restriction of the generality, we
Solvability of a fractional integral equation... 143

may assume t1 ≥ t2 . We have

|(Ux )(t1) − (Ux )(t2)|

≤ | f 1 (t1 , x (t1 )) − f 1 (t2 , x (t2 ))|
 
 f 2 (t1 , x (t1 )) Z t1 g′ (τ )k(τ, x (τ )) Z
f 2 (t2 , x (t2 )) t2 g′ (τ )k(τ, x (τ )) 
+ dτ − dτ 
Γ( α) a ( g ( t1 ) − g ( τ )) 1 − α Γ( α) a ( g ( t2 ) − g ( τ )) 1 − α

≤ | f 1 (t1 , x (t1 )) − f 1 (t2 , x (t2 ))|

 
 f 2 (t1 , x (t1 )) Z t1 g′ (τ )k(τ, x (τ )) Z
f 2 (t2 , x (t2 )) t1 g′ (τ )k(τ, x (τ )) 
+ dτ − dτ 
Γ( α) a ( g ( t1 ) − g ( τ )) 1 − α Γ( α) a ( g ( t1 ) − g ( τ )) 1 − α
Z Z t2 
| f 2 (t2 , x (t2 ))|  t1 g′ (τ )k(τ, x (τ )) g′ (τ )k(τ, x (τ )) 
+  1 − α
dτ − 1 − α
dτ 
Γ( α) a ( g ( t1 ) − g ( τ )) a ( g ( t2 ) − g ( τ ))
≤ | f 1 (t1 , x (t1 )) − f 1 (t1 , x (t2 ))| + | f 1 (t1 , x (t2 )) − f 1 (t2 , x (t2 ))|
Z
| f (t , x (t1 )) − f 2 (t1 , x (t2 ))| + | f 2 (t1 , x (t2 )) − f 2 (t2 , x (t2 ))| t1 g′ (τ )|k(τ, x (τ ))|
+ 2 1 dτ
Γ( α) a ( g ( t1 ) − g ( τ ))1− α
Z t Z t2 
| f 2 (t2 , x (t2 )) − f 2 (t2 , 0)| + | f 2 (t2 , 0)|  1 g′ (τ )k(τ, x (τ )) g′ (τ )k(τ, x (τ )) 
+  1 − α
dτ − 1 − α
dτ 
Γ( α) a ( g ( t1 ) − g ( τ )) a ( g ( t1 ) − g ( τ ))
Z Z t2 
| f 2 (t2 , x (t2 )) − f 2 (t2 , 0)| + | f 2 (t2 , 0)|  t2 g′ (τ )k(τ, x (τ )) g′ (τ )k(τ, x (τ )) 
+  1 − α
dτ − 1 − α
dτ 
Γ( α) a ( g ( t1 ) − g ( τ )) a ( g ( t2 ) − g ( τ ))
ϕ2 (| x (t1 ) − x (t2 )|) + ω f 2 (r0 , ρ)
≤ ϕ1 (| x (t1 ) − x (t2 )|) + ω f 1 (r0 , ρ) + ϕ(r0 )( g(t1 ) − g( a))α
Γ( α + 1)
Z
( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 ) t1 g′ (τ )
+ dτ
Γ( α) t2 ( g ( t1 ) − g ( τ ))1− α
( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 )
+ (( g(t2 ) − g( a))α + ( g(t1 ) − g(t2 ))α − ( g(t1 ) − g( a))α )
Γ( α + 1)
ϕ2 (ω ( x, ρ)) + ω f 2 (r0 , ρ)
≤ ϕ1 (ω ( x, ρ)) + ω f 1 (r0 , ρ) + ϕ(r0 )( g( T ) − g( a))α
Γ( α + 1)
2( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 )
+ ( g(t1 ) − g(t2 ))α
Γ( α + 1)
ϕ2 (Ω(W, ρ)) + ω f 2 (r0 , ρ)
≤ ϕ1 (Ω(W, ρ)) + ω f 1 (r0 , ρ) + ϕ(r0 )( g( T ) − g( a))α
Γ( α + 1)
2( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 )
+ [ω ( g, ρ)]α ,
Γ( α + 1)

where

ω f i (r0 , ρ) = sup{| f i (t, y) − f i (s, u)| : u ∈ [−r0 , r0 ], t, s ∈ [ a, T ], |t − s| ≤ ρ}, i = 1, 2.

Therefore,

ω (Ux, ρ) ≤ ϕ1 (Ω(W, ρ)) + ω f 1 (r0 , ρ)+

ϕ2 (Ω(W, ρ)) + ω f 2 (r0 , ρ) α 2( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 )
ϕ(r0 )( g(T ) − g(a)) + [ω ( g, ρ)]α .
Γ ( α + 1) Γ ( α + 1)

Passing to the limit as ρ → 0+ , we obtain

ϕ2 (η (W ))
η (UW ) ≤ ϕ1 (η (W )) + ϕ(r0 )( g(T ) − g(a))α .
Γ ( α + 1)
144 K. Sadarangani – B. Samet

Taking in consideration assumption (H2)-(iv), we get

 
ϕ(r0 )( g(T ) − g(a))α
η (UW ) ≤ δ+ Tmax (θ, ϕ2 )(η (W ))
Γ ( α + 1)
= T (λ, Tmax (θ, ϕ2 ))(η (W )),

where
ϕ(r0 )( g(T ) − g(a))α
λ = δ+ .
Γ ( α + 1)
Since λ ∈ [0, 1] (see (H5)), from Lemma 2.3, T (λ, Tmax (θ, ϕ2 )) ∈ Φ. Finally, by
Lemma 2.2, the operator U has at least one fixed point y∗ ∈ B(0, r0 ), which is a
solution to Eq.(2.1) satisfying ky∗ k ≤ r0 . This completes the proof.

If in Theorem 3.1 we add the following assumptions:

(H6) For i = 1, 2, f i ([ a, T ] × [0, ∞)) ⊆ [0, ∞) and f i is non-decreasing with respect
to each variable on [ a, T ] × [0, ∞).
(H7) k([ a, T ] × [0, ∞)) ⊆ [0, ∞) and k is non-decreasing with respect to each vari-
able on [ a, T ] × [0, ∞).
Then we obtain the following result.
Theorem 3.2. Under the assumptions (H1)-(H7), Eq.(2.1) has at least one solution
y∗ ∈ C([ a, T ]; R ) which is non-negative and non-decreasing on the interval [ a, T ].
Proof. Let us consider the operator U defined on H = C([ a, T ]; R ) by (3.1). Fol-
lowing the proof of Theorem 3.1, we know that the operator U maps H into itself,
and that, for x ∈ H, we have the estimate
( ϕ2 (k x k) + f 2∗ ) ϕ(k x k)
kUx k ≤ ϕ1 (k x k) + f 1∗ + ( g(T ) − g(a))α , x ∈ H,
Γ ( α + 1)

which yields U ( B(0, r0 )) ⊆ B(0, r0 ). Now, we consider the subset Br+0 of B(0, r0 )
defined by
Br+0 = { x ∈ B(0, r0 ) : x (t) ≥ 0, t ∈ [ a, T ]}.
In virtue (H6) and (H7), we may infer that the operator U maps Br+0 into itself.
Following the same argument used in the proof of Theorem 3.1, we obtain that U
is a continuous operator on Br+0 . Next, we take the subset Q of Br+0 defined by

Q = { x ∈ Br+0 : x is non-decreasing}.

Obviously, the set Q is nonempty, bounded, closed and convex. Let us prove that
the operator U maps Q into itself. To do this, let x ∈ Q be fixed and (t, s) be a
pair of elements in [ a, T ] × [ a, T ] such that t ≥ s. By assumption (H6), we have

f i (t, x (t)) ≥ f i (s, x (s)), i = 1, 2.

So we have just to prove that

Z t ′ Z s ′
g (τ )k(τ, x (τ )) g (τ )k(τ, x (τ ))
dτ ≥ dτ. (3.4)
a ( g(t) − g(τ ))1−α a ( g(s) − g(τ ))1−α
Solvability of a fractional integral equation... 145

We have
Z t ′ Z s ′
g (τ )k(τ, x (τ )) g (τ )k(τ, x (τ ))
dτ − dτ
a ( g(t) − g(τ ))1−α a ( g(s ) − g(τ ))1− α
Z s 
g′ (τ ) g′ (τ )
= − k(τ, x (τ )) dτ
a ( g(t) − g(τ ))1−α ( g(s) − g(τ ))1−α
Z t ′
g (τ )k(τ, x (τ ))
+ dτ.
s ( g(t) − g(τ ))1− α

Using assumption (H7), we obtain

Z t ′ Z s ′
g (τ )k(τ, x (τ )) g (τ )k(τ, x (τ ))
dτ − dτ
a ( g(t) − g(τ ))1−α a ( g(s ) − g(τ ))1− α
Z s 
g′ (τ ) g′ (τ )
≥ k(s, x (s)) − dτ
a ( g(t) − g(τ ))1−α ( g(s) − g(τ ))1−α
Z t
g′ (τ )
+k(s, x (s)) dτ
s ( g(t) − g(τ ))1− α
Z t Z s 
g′ (τ ) g′ (τ )
= k(s, x (s)) dτ − dτ
a ( g(t) − g(τ ))1− α a ( g(s ) − g(τ ))1− α
k(s, x (s))
= (( g(t) − g(a))α − ( g(s) − g(a))α ) ≥ 0.
α
Then (3.4) holds and Ux is a non-decreasing function. Therefore, u maps Q into
itself. Finally, using the same argument of the proof of Theorem 3.1, we obtain

η (UW ) ≤ T (λ, Tmax (θ, ϕ2 ))(η (W )), W ⊆ Q,

where
ϕ(r0 )( g(T ) − g(a))α
λ = δ+ ∈ [0, 1].
Γ ( α + 1)
An application of Lemma 2.2 gives us the desired result.

4 Particular cases and examples

In this section, using Theorem 3.1, we give some existence results for some func-
tional equations involving various types of fractional integrals. We present also
some illustrative examples.

4.1 A functional equation involving Riemann-Liouville fractional integral

Let us consider the integral equation

Z t
f 2 (t, x (t)) k(τ, x (τ ))
x (t) = f 1 (t, x (t)) + dτ, t ∈ [ a, T ], (4.1)
Γ(α ) a ( t − τ )1− α
146 K. Sadarangani – B. Samet

where α ∈ (0, 1), 0 ≤ a < T, k : [ a, T ] × R → R and f i : [ a, T ] × R → R (i = 1, 2).

Eq.(4.1) can be written as
x (t) = f 1 (t, x (t)) + f 2 (t, x (t)) Iaα k(·, x (·))(t), t ∈ [ a, T ],
where Iaα is the Riemann-Liouville fractional integral of order α defined by (2.2).
Take
g(t) = t, t ∈ [ a, T ]
in Theorem 3.1, we obtain the following existence result.
Corollary 4.1. Suppose that all the assumptions (H1)-(H3) are satisfied. Moreover,
suppose that there exists r0 > 0 such that
( ϕ1 (r0 ) + f 1∗ )Γ(α + 1) + ( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 )(T − a)α ≤ r0 Γ(α + 1)
and
ϕ (r 0 )
δ+ (T − a)α ≤ 1.
Γ ( α + 1)
Then (4.1) has at least one solution y∗ ∈ C([ a, T ]; R ). Moreover, we have ky∗ k ≤
r0 .
We present the following example to illustrate the above result.
Example 4.2. Consider the integral equation
√  Z t
t π t x (τ )
x (t) = + + | x (t)| p dτ, t ∈ [0, 1]. (4.2)
2 8 3 2
0 ( τ + 1) t − τ )

The above equation can be written in the form (4.1) with a = 0, T = 1, α = 21 ,

t
f 1 (t, u) = (t, u) ∈ [0, 1] × R,
,
2√
 
π t
f 2 (t, u) = + |u| , (t, u) ∈ [0, 1] × R,
8 3
√
π u
k(t, u) = 2
, (t, u) ∈ [0, 1] × R.
2 ( t + 1)
We can check easily that √
all the assumptions
√
(H1)-(H3) are satisfied with ϕ1 (t) =
π π
θ (t) = δ = 0, ϕ2 (t) = 8 t and ϕ(r ) = 2 r. Moreover, in this case, we have
√
π
f 1∗ = 1
2 and f 2∗ = 24 . Observe that for r0 = 1, we have
ϕ (r 0 )
δ+ (T − a)α = 1.
Γ ( α + 1)
Moreover,

( ϕ1 (r0 ) + f 1∗ )Γ(α + 1) + ( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 )(T − a)α =

√ √
π π π
+ ≤ = r 0 Γ ( α + 1 ).
4 12 2
Then all the required assumptions of Corollary 4.1 are satisfied. As consequence,
Eq.(4.2) has at least one solution y∗ ∈ C([0, 1]; R ) satisfying ky∗ k ≤ 1.
Solvability of a fractional integral equation... 147

4.2 A functional equation involving Erdélyi-Kober fractional integral

Let us consider the integral equation

Z t β−1
β f 2 (t, x (t)) τ k(τ, x (τ ))
x (t) = f 1 (t, x (t)) + dτ, t ∈ [ a, T ], (4.3)
Γ(α ) a ( t β − τ β )1− α

where α ∈ (0, 1), β > 0, 0 ≤ a < T, k : [ a, T ] × R → R and f i : [ a, T ] × R → R

(i = 1, 2). Eq.(4.3) can be written as
α
x (t) = f 1 (t, x (t)) + f 2 (t, x (t)) Ia,β k(·, x (·))(t), t ∈ [ a, T ],

α is the Erdélyi-Kober fractional integral defined by (2.3).

where Ia,β
Take
g(t) = t β , t ∈ [ a, T ]
in Theorem 3.1, we obtain the following existence result.

Corollary 4.3. Suppose that all the assumptions (H1)-(H3) are satisfied. Moreover,
suppose that there exists r0 > 0 such that
 α
( ϕ1 (r0 ) + f 1∗ )Γ(α + 1) + ( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 ) T β − a β ≤ r0 Γ(α + 1)

and
ϕ (r 0 )  β α
δ+ T − a β ≤ 1.
Γ ( α + 1)
Then (4.3) has at least one solution y∗ ∈ C([ a, T ]; R ). Moreover, we have
k y ∗ k ≤ r0 .

We present the following example to illustrate the above result.

Example 4.4. We consider the integral equation

37 Z t
e −1 t βe 19 τ β−1 x (τ )
x (t) = + (t + 1 + arctan | x (t)|)
1− α √ dτ, t ∈ [0, 1],
10 76Γ(α) 0 tβ − τ β τ+1
(4.4)
where α ∈ (0, 1) and β > 0. The above equation can be written in the form (4.3)
with a = 0, T = 1,

e −1 t
f 1 (t, u) = , (t, u) ∈ [0, 1] × R,
10
t + 1 + arctan |u|
f 2 (t, u) = , (t, u) ∈ [0, 1] × R,
2
37
e 19 u
k(t, u) = √ , (t, u) ∈ [0, 1] × R.
38 t + 1
148 K. Sadarangani – B. Samet

We can check easily that all the assumptions (H1)-(H3) are satisfied with ϕ1 (t) =
37
e −1
θ (t) = δ = 0, ϕ2 (t) = 2t and ϕ(r ) = e38
19
r. Moreover, in this case, we have f 1∗ = 10
∗
and f 2 = 1. On the other hand, we have [22]
37
e 19
= 0.0922... ≤ 0.8856 ≤ Γ(α + 1), for α ∈ (0, 1).
76
Then for r0 = 12 , we have
37
ϕ (r 0 )  β β
α e 19
76
δ+ T −a = ≤ 1.
Γ ( α + 1) Γ ( α + 1)
Next, we have

α
∗ ( ϕ2 (r0 ) + f 2∗ ) ϕ(r0 ) T β − a β
( ϕ 1 (r 0 ) + f 1 ) +
Γ ( α + 1)
37
e −1 5e 19 1
= + ≤ 0.16698... ≤ = r0 .
10 304Γ(α + 1) 2
Then all the required assumptions of Corollary 4.3 are satisfied. As consequence,
Eq.(4.4) has at least one solution y∗ ∈ C([0, 1]; R ) satisfying ky∗ k ≤ 12 .

4.3 A functional equation involving Hadamard fractional integral

We consider the integral equation

Z t  α−1
f 2 (t, x (t)) t k(τ, x (τ ))
x (t) = f 1 (t, x (t)) + ln dτ, t ∈ [ a, T ], (4.5)
Γ(α ) a τ τ
where 0 < a < T, α ∈ (0, 1), k : [ a, T ] × R → R and f i : [ a, T ] × R → R (i = 1, 2).
Eq.(4.5) can be written as
x (t) = f 1 (t, x (t)) + f 2 (t, x (t)) Jaα k(·, x (·))(t), t ∈ [ a, T ],
where Jaα is the Hadamard fractional integral of order α defined by (2.4).
Take
g(t) = ln t, t ∈ [ a, T ]
in Theorem 3.1, we obtain the following existence result.
Corollary 4.5. Suppose that all the assumptions (H1)-(H3) are satisfied. Moreover,
suppose that there exists r0 > 0 such that
 
∗ ∗ T α
( ϕ1 (r0 ) + f 1 )Γ(α + 1) + ( ϕ2 (r0 ) + f 2 ) ϕ(r0 ) ln ≤ r0 Γ ( α + 1) (4.6)
a
and  α
ϕ (r 0 ) T
δ+ ln ≤ 1. (4.7)
Γ ( α + 1) a
Then (4.5) has at least one solution y∗ ∈ C([ a, T ]; R ). Moreover, we have ky∗ k ≤
r0 .
Solvability of a fractional integral equation... 149

We end the paper with the following example that illustrates the above result.

Example 4.6. Let us consider the integral equation

Z t  α−1
−2 −3 t
x (t) = e t + 2αe x (t) ln x (τ ) dτ, t ∈ [1, e1 ], (4.8)
1 τ

where α ∈ (0, 1). Eq.(4.8) can be written in the form (4.5) with a = 1, T = e1 ,

f 1 (t, u) = e−2 t, (t, u) ∈ [1, e1 ] × R,

u
f 2 (t, u) = , (t, u) ∈ [1, e1 ] × R,
2
k(t, u) = 4e−3 Γ(α + 1)tu, (t, u) ∈ [1, e1 ] × R.

We can check easily that all the assumptions (H1)-(H3) are satisfied with ϕ1 (t) =
θ (t) = δ = 0, ϕ2 (t) = 2t and ϕ(r ) = 4e−2 Γ(α + 1)r. In this case, we have f 1∗ = e−1
2
and f 2∗ = 0. Moreover, for r0 = e4 , the inequalities (4.6) and (4.7) are satisfied.
Therefore, by Corollary 4.5, Eq.(4.8) has at least one solution y∗ ∈ C([1, e1 ]; R )
2
such that ky∗ k ≤ e4 .

Acknowledgements. The first author was partially supported by the project

MTM-2013-44357-P. The second author would like to extend his sincere apprecia-
tion to the Deanship of Scientific Research at King Saud University for its funding
of this research through the International Research Group Project No. IRG14-04.

References
[1] R. Agarwal, M. Meehan, D. O’Regan, Fixed Point Theory and Applications,
Cambridge University Press, 2004.
[2] R.P. Agarwal, B. Samet, An existence result for a class of nonlinear integral
equations of fractional orders, Nonlinear Analysis: Modelling and Control.
21 (5) (2016) 616–629.
[3] A. Aghajani, R. Allahyari, M. Mursaleen, A generalization of Darbo’s the-
orem with application to the solvability of systems of integral equations,
J. Comput. Appl. Math. 260 (2014) 68–77.
[4] A. Aghajani, J. Banaś, Y. Jalilian, Existence of solutions for a class of non-
linear Volterra singular integral equations, Comput. Math. Appl. 62 (2011)
1215–1227.
[5] A. Aghajani, J. Banaś, N. Sabzali, Some generalizations of Darbo fixed point
theorem and applications, Bull. Belg. Math. Soc. Simon Stevin. 20 (2) (2013)
345–358.
[6] A. Aghajani, E. Pourhadi, J.J. Trujillo, Application of measure of noncom-
pactness to a cauchy problem for fractional differential equation in Banach
spaces, Fract. Calc. Appl. Anal. 16 (2013) 962–977.
150 K. Sadarangani – B. Samet

[7] J. Banaś, Measures of noncompactness in the study of solutions of nonlinear

differential and integral equations, Cent. Eur. J. Math. 10 (6) (2012) 2003–
2011.
[8] J. Banaś, K. Goebel, Measures of Noncompactness in Banach Space. Lecture
Notes in Pure and Applied Mathematics, Vol. 60. Dekker, New York, 1980.
[9] J. Banaś, R. Rzepka, An application of a measure of noncompactness in the
study of asymptotic stability, Appl. Math. Lett. 16 (2003) 1–6.
[10] D. Beglov, B. Roux, An integral equation to describe the solvation of polar
molecules in liquid water, J. Phys. Chem. B. 101 (1997) 7821–7826.
[11] M. Benchohra, D. Seba, Integral equations of fractional order with multiple
time delays in banach spaces, Electron. J. Differential Equations. 56 (2012)
1–8.
[12] U. Çakan, I. Ozdemir, An application of Darbo fixed-point theorem to a class
of functional integral equations, Numer. Funct. Anal. Optim. 36 (2015) 29–40.
[13] M. Cichon, M.A. Metwali, On quadratic integral equations in Orlicz spaces,
J. Math. Anal. Appl. 387 (2012) 419–432.
[14] B.C. Dhage, D. O’Regan, A fixed point theorem in Banach algebras with ap-
plications to nonlinear integral equation, Funct. Differ. Equ. 7 (3-4) (2000)
259–267.
[15] M.A. Darwish, On monotonic solutions of a singular quadratic integral
equation with supremum, Dynam. Systems Appl. 17 (2008) 539–550.
[16] M.A. Darwish, On quadratic integral equation of fractional orders, J. Math.
Anal. Appl. 311 (2005) 112–119.
[17] M.A. Darwish, On Erdélyi-Kober fractional Urysohn-Volterra quadratic in-
tegral equation, Appl. Math. Comput. 273 (2016) 562–569.
[18] M.A. Darwish, K. Sadarangani, On a quadratic integral equation with supre-
mum involving Erdélyi-Kober fractional order, Mathematische Nachrichten.
288 (5-6) (2015) 566–576.
[19] K. Deimling, Nonlinear Functional Analysis, Springer-Verlag, Berlin, 1985.
[20] B.C. Dhage, M. Imdad, Asymptotic behaviour of nonlinear quadratic func-
tional integral equations involving Carathodory, Nonlinear Anal. 71 (2009)
1285–1291.
[21] B.C. Dhage, S.S. Bellale, Local asymptotic stability for nonlinear quadratic
functional integral equations, Electron. J. Qual. Theory Differ. Equ. 10 (2008)
1–13.
[22] G.M. Fichtenholz, Differential and Integral Calculus II, PWN, Warsaw, 1980
(in Polish).
[23] H.W. Hethcote, D.W. Tudor, Integral equation models for endemic infectious
diseases, J. Math. Biology. 9 (1980) 37–47.
Solvability of a fractional integral equation... 151

[24] M. Jleli, M. Mursaleen, B. Samet, On a class of q-integral equations of frac-

tional orders, Electron. J. Differ. Equ. 17 (2016) 1–14.
[25] S. Kablin, L. Nibenbebg, On a theorem of P. Nowosad, J. Math. Anal. Appl.
17 (1967) 61–67.
[26] J. Klamka, Schauder’s fixed point theorem in nonlinear controllability, Con-
trol and Cybernetics, 29 (1) (2000) 153-165. problems
[27] V. Muresan, Volterra integral equations with iterations of linear modification
of the argument, Novi Sad J. Math. 33(2003) 1–10.
[28] M. Mursaleen, S.A. Mohiuddine, Applications of measures of noncompact-
ness to the infinite system of differential equations in l p spaces, Nonlinear
Anal. 75 (2012) 2111–2115.
[29] I.M. Olaru, An integral equation via weakly Picard operators, Fixed Point
Theory. 11(2010) 97–106.
[30] I.M. Olaru, Generalization of an integral equation related to some epidemic
models, Carpathian J. Math. 26 (1) (2010) 92–96.
[31] S.G. Samko, A.A. Kilbas, O.I. Marichev, Fractional Integrals and Derivatives.
Theory and Applications, Gordon and Breach, Yverdon et alibi, 1993.
[32] HA. Schenck, Improved integral formulation for acoustic radiation prob-
lems, J Acoust Soc Am. 44 (1968) 41–58.
[33] M.A. Serban, I.A. Rus, A. Petrusel, A class of abstract Volterra equations via
weakly Picard operators technique, Math. Ineq. Appl. 13(2010) 255–269.
[34] J. Sladek, V. Sladek, S.N. Atluri, Local boundary integral equation (LBIE)
method for solving problems of elasticity with nonhomogeneous material
properties, Computational Mechanics. 24 (2000) 456–462.
[35] T. Terai, On calculation of sound fields around three dimensional objects by
integral equation methods, J Sound Vib. 69 (1) (1980) 71–100.

Departamento de Matemàticas, Universidad de Las Palmas de Gran Canaria,

Campus de Tafira Baja,
35017 Las Palmas de Gran Canaria, Spain.
E-mail: ksadaran@dma.ulpgc.es

Department of Mathematics, College of Science,

King Saud University,
P.O. Box 2455, Riyadh, 11451, Saudi Arabia.
E-mail: bsamet@ksu.edu.sa
Copyright of Bulletin of the Belgian Mathematical Society - Simon Stevin is the property of
Department of Pure Mathematics & Computer Algebra, Ghent University and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.