Indonesian Journal of Mathematics and Applications (2023), 1-2, 1–12

RESEARCH ARTICLE

Neutral Caputo-Fabrizio Fractional Differential

Equation with Hyers-Ulam Stability
M. Rohman,*† İ. Eryılmaz,‡ and M.N. Huda¶
†Department of Madrasah Ibtidaiyah Teacher Education, School of Islamic Studies Ma’had Aly Al-Hikam Malang, Indonesia
‡Department of Mathematics, Faculty of Science, Ondokuz Mayıs Üniversitesi, Türkiye
¶Department of Mathematics, Mulawarman University, Samarinda Ulu, Samarinda, 75123, Indonesia
*Corresponding author. Email: minanurrohmanali@gmail.com

Abstract
In this paper, we will discuss the existence and uniqueness of a neutral Caputo-Fabrizio differential equation
which fulfills several required conditions. Furthermore, using the Picard operator, we show that the
problem has Hyers-Ulam type stability.

Keywords: fixed point theorem, neutral fractional differential equation, Caputo-Fabrizio derivative, Hyers-Ulam stability

1. INTRODUCTION
It is well known that differential equations have applications in many fields. This is why a differential
equation offers a mathematical model of a real phenomenon with respect to changes in time. In the last
few decades, there has emerged a branch of differential equations with orders in the form of fractional
numbers called fractional differential equations. Although the term attracted many mathematicians
recently, the actual embryo of the fractional derivative first appeared in 1965 when Leibniz revealed
to L’Hospital that it was possible to generalize the differential operation for non-integer numbers.
The reason of using a fractional derivative is its own properties. If we can know what happened in a
short span of time, then we will have better information about the phenomenon as a whole. This
property is found in fractional derivatives which are not owned by classical derivatives. This can be
easily seen from the definition of the general derivative (Grünwald-Letnikov algorithm) for order α
under gamma function Γ (α + 1) = α!, i.e.,

j k
x–a
h
dα 1
α
k
h i X
α f (x) = lim α (–1) f (x – kh)
dx h↓0 h k
k=0
j k
x–a
h
1 X k Γ (α + 1)
= lim α (–1) f (x – kh)
h↓0 h k!Γ (α – k + 1)
k=0

where a is the starting point of differentiation (see [24]). If α ∈ N, by the definition of gamma
function (see Definition 1), then the series on the right hand side tends to zero for all k ≥ α + 1.
Besides, for a finite α ∈ N, f (x – kh) → f (x) whenever h → 0. Thus, it can be said that the classical
2 M. Rohman et al.

derivative has a local property. On the contrary, this series does not tend to zero for α < N. Therefore,
fractional derivative has non-local property.
Recently, there has been a lot of research related to fractional derivatives, in both pure analysis
and applications. There are two main types of fractional derivatives that are widely used, which are
the left-hand definition (Riemann-Liouville) and the right-hand definition (Caputo). These two
terms arise because fractional derivatives are obtained by applying fractional integrals (see Definition
2). The main idea of Riemann-Liouville (RL) fractional derivative algorithm is to integrate with
order n – α where n – 1 < α < n, n ∈ N, and then taking the whole derivative of order n. On the other
hand, the idea of Caputo derivative algorithm is to differentiate the order n first before taking the
integration of order n – α. This process causes both fractional derivative to require a starting point
when carrying out the differentiation process. Several applications of the Caputo derivative can be
found in [4, 15, 16, 30]. Unfortunately, the same reason also causes fractional derivatives to have a
weakness, namely there is a point of singularity in their kernels. By modifying the kernel using an
exponential function, Caputo and Fabrizio introduce a new definition of fractional derivative which
is known as Caputo-Fabrizio (CF) derivative (see Definition 4).
For any Banach spaces X and Y, Ulam said that the surjective isometry mapping U : X → Y is
affine [20], i.e., there is a linear mapping L : X → Y such that U (x) = L (x) + U(0). This means that
if U (0) = 0, then all surjective isometries are linear mappings. This finding offers two new problems,
related to the terms ε-linear and ε-isometry, which later got affirmative answers from Hyers.
The first answer from Hyers is as follows. For any ε-linear mapping f : X → Y, i.e.

f (x + y) – f (x) – f (y) ≤ ε, x, y ∈ X, ε > 0,

there exists a unique linear mapping L : X → Y such that

f (x) – L(x) ≤ kε, x ∈ X (1)

for some k ∈ R+
[8]. Not long after, Hyers gave an affirmative answer to the second question posed
by Ulam. For any surjective ε-isometry mapping f : X → Y with f (0) = 0, i.e.

f (x) – f (y) – x – y ≤ ε, x, y ∈ X, ε > 0,

there exists a unique linear isometry mapping U : X → Y such that

f (x) – U(x) ≤ kε, x ∈ X (2)

for some k ∈ R+ [9]. The value of k = 2 is the smallest number that satisfies the inequality (2)
above [22]. The surjective property can be removed if X and Y are equipped with weak topologies
[6, 25]. Note that the condition f (0) = 0 does not lose generality, because if f (x) is ε-isometry, then
f (x) – f (0) is also ε-isometry.
Combining inequalities (1) and (2), Ulam [28] said that this type of "stability" can be applied to
differential equations (see Definition 6). This type of stability can be used in ordinary differential
equations (e.g. [7, 10]) and fractional differential equations (e.g. [1, 6, 13, 14, 15, 17, 29, 31]). Liu et
al. gave the existence and uniqueness of a CF fractional differential equation which is Hyers-Ulam
stable [18]. Recently, Wang [30] showed that the following CF fractional differential equation with
boundary value condition,
 ′
 x (t) + CF D0β [x (t)] = f (t, x (t)) , t ∈ [0, 1]

 x (1) = I α [x (ξ)] , (3)
ξ ∈ [0, 1]

0
has a unique solution and Hyers-Ulam stable, where x(t) ∈ C1 ([0, 1] ,R), f : [0, 1] × R → R is a
continuous function, and I0α [·] is the Riemann-Liouville fractional integral of order α > 0 with
starting point 0.
 Indonesian Journal of Mathematics and Applications 3

Motivated by [13, 14], this paper will show the existence and uniqueness of solution together
with Hyers-Ulam stability of the following CF neutral fractional differential equation:
β
 
CF Dα x
[ (t)] = f t, x (t) , CF D0 [x (t)] , t ∈ [0, 1]


0



x (t) = η (t) , t ∈ [–r, 0) (4)



 x (0) = x , x′ (0) = x


0 1
β
where CF D0α [x (t)] and CF D0 [x (t)], respectively, are Caputo-Fabrizio fractional derivatives of
order α and β, 1 < α < 2 and 0 < β < 1; x0 and x1 are real scalars; f : [0, 1] × Cr × Cr → R; and
η : [–r, 0) → R are continuous functions. Here, Cr will denote the Banach space of all real-valued
continuous functions on [–r, 1] under the maximum norm.

2. Preliminaries
This section will contain basic information that will be used in the following discussion.
Definition 1. ([24], Gamma function) The gamma function Γ:R+ → R is defined as
Z ∞
Γ (α) = e–t tα–1 dt.
0

It is easy to see that n! = Γ (n + 1) for any n ∈ N and Γ (α + 1) = αΓ (α). Note that the gamma
function can be extended to negative non-integer arguments using analytic continuation and so
α! = Γ (α + 1) for all α < Z– . Deploying Gamma function, the Riemann-Liouville fractional integral
is defined as follows.
Definition 2. ([24], RL fractional integral) Let –∞ ≤ a < x ≤ b ≤ ∞ and f : [a, b] → R be a locally
integrable function on [a, b]. The RL fractional integral of order α > 0 with starting point a of function f (x)
is defined as
Z x
α
h i 1 α–1
a Ix f (x) = (x – y) f (y) dy
Γ (α) a
or Z b
α
h i 1 α–1
x Ib f (x) = (x – y) f (y) dy.
Γ (α) x
The first definition is called forward (left) definition while the other is backward (right) definition.
As their names, the forward definition uses the previous value to determine the current state of f (x)
under R-L fractional integral while the backward definition uses the future value. Therefore, the
forward definition is to find out the effect of something that previously happened, while the backward
definition is to know the future effect on the present. For this reason, the forward definition and
backward definition are also known as causal and anti-causal, respectively. Using this fractional
integral, we get the following definition.
Definition 3. ([24], R-L and Caputo fractional derivative) Assume that all assumptions in Definition 2
are satisfied. For n – 1 < α < n, the Riemann-Liouville and Caputo fractional derivative of order α with
starting point a = 0 for a function f (x), respectively, are defined as
Z x
α
h i
α
h i 1 dn n–α–1
0 Dx f (x) = D0 f (x) = (x – y) f (y) dy
Γ (n – α) dxn 0
and x
1
Z
C α n–α–1 (n)
f (x) = C D0α f (x) =
h i h i
0 Dx (x – y) f (y) dy.
Γ (n – α) 0
4 M. Rohman et al.

Note that the function f (x) must be n-differentiable continuous function, i.e., f ∈ Cn ([a, b] ,R ).
There are some calculations that give the relationship between RL and Caputo fractional derivatives,
e.g.

C
D0α f (x) = D0α f (x) – D0α f (0) + D0α–1 f (1) (0) + · · · + D0α–n+1 f (n–1) (0) .
h i h i  h i h i h i

Hence, if 0 < α < 1, then

C
h i h i
D0α f (x) = D0α f (x) – f (0) .
From Definition 2, it can be seen that both RL and Caputo fractional derivatives are undefined
when x = y, since the kernel is a power function and n – 1 < α. The following definition is another
fractional derivative which is obtained by changing the kernel.
Definition 4. ([3, 9], Caputo-Fabrizio) The new Caputo-Fabrizio (CF) fractional derivative of order α
(0 < α < 1) for a function f ∈ C1 ([a, b] ,R ) is defined as
i (2 – α) M(α) Z x  –α 
CF α
f (x) = CF D0α f (x) = (x – y) f (1) (y) dy,
h i h
0 Dx 2 (1 – α)
exp
1–α
0
2
where M(α) is a normalization function with M (α) = 2–α . For n ≥ 1 and 0 < α < 1, new Caputo-Fabrizio
(CF) fractional derivative of order α + n is defined as
i (2 – α) M(α) Z x  –α 
CF
D0α+n (x – y) f (n+1) (y) dy.
h
f (x) = exp
2 (1 – α) 0 1–α
The following results can be found in [23, 26].
Definition 5. (Picard operator) Let X be metric space and P : X → X be an operator. If there exists u0 ∈ X
such that
(i) the set of fixed points FP := {x ∈ X : P (x) = x} of operator P contains a unique u0 , i.e., FP = {u0 },
(ii) the sequence (Pn (x))n∈N converges to u0 for all x ∈ X,
then P is called Picard operator.
Lemma 1. (Gronwall lemma) Let φ and ψ be positive real-valued continuous functions on interval [a, b], i.e.
φ, ψ ∈ C ([a, b] ,R), and φ be an increasing function. If x ∈ C ([a, b] , R+ ) is the solution of the foollowing
inequality
Z b
x (t) ≤ φ (t) + ψ (s) x (s) ds, t ∈ [a, b] ,
a
then
b
Z 
 
x (t) ≤ φ (t) exp  ψ (s) ds , t ∈ [a, b] .
a

Lemma 2. (Abstract Gronwall lemma) Let X be an ordered metric space. If there exists an increasing Picard
operator P : X → X with FP = {xP }, then i x ≤ P(x) ⇒ x ≤ xP and ii x ≥ P(x) ⇒ x ≥ xP for all x ∈ X.
Let Y and Z be Banach spaces and (xn ) ⊂ Y ∩ Z be a sequence convergence to x. Since
∥xn – x∥Y + ∥xn – x∥Z → 0 whenever n → ∞, we must have x ∈ Y ∩ Z. Thus, we have the following
proposition.
Proposition 1. Let Y and Z be a Banach space and X = Y ∩ Z. Then X is a Banach space under the norm
∥x∥X = ∥x∥Y + ∥x∥Z .
 Indonesian Journal of Mathematics and Applications 5

Motivated by inequalities (1) and (2), following Jung’s method [11], we are ready to give the
definition of the stability of fractional differential equation (4).

Definition 6. (Hyers-Ulam stability) Let ε > 0. The CF fractional differential equation (4) is Hyers-Ulam
stable if and only if for any approximation solution y(t) of

CF CF β
 
D0α [x (t)] – f t, xt , D0 [x (t)] ≤ ε, t ∈ [–r, 1] ,

there is a constant k and exact solution x(t) of CF fractional differential equation (4) such that

x(t) – y(t) ≤ kε.

Note that Hyers-Ulam stability implies Lyapunov stability [21].

3. Main Result
In this section we will give a theorem that is the main objective of this paper by showing the
existence and uniqueness of the solution of CF fractional differential equation (4) and its stability in
Hyers-Ulam sense.

Theorem 1. (Main theorem) Suppose that CF fractional differential equation (4) satisfies the following
conditions:
β 1
(i) f ∈ C ([–r, 1] ×R×R,R) and CF D0 [x (t)] ≤ Γ(2– β)
x(t) ,

(ii) there exists a real number N > 0 such that

f (t, δ1 , δ2 ) – f (t, γ1 , γ2 ) ≤ N ( δ1 – γ1 + δ2 – γ2 )
for t ∈ [0, 1] , δi , γi ∈ R where i = 1, 2, and

N N(2–β)M(β)(α–1)
(iii) Γ(α+1)
+ 2(1–β)Γ(α)
= c < 1, where M (β) is a normalization function.
Then, the Caputo Fabrizio fractional differential equation (4) has a unique solution in C ([–r, 1] ,R) ∩
C1 ([–r, 1] ,R) and stable in the sense of Hyers-Ulam.
β
Proof. Let X := x ∈ C ([–r, 1] ,R) : CF D0 [x (t)] ∈ C1 ([–r, 1] ,R) be a vector space and define a
n o

function F : X → R+ ∪ {0} by

CF β
F (x) = max x (t) + max D0 [x (t)] .
t∈[–r,1] t∈[–r,1]

Obviously, F is well defined and F (x) = 0 if and only if x = 0. It is easy to see that F is a positive
homogenous function. Indeed x (t) , CF Dβ x∈ R, the triangle inequality of R shows that
0 [ (t)]
F (x + y) ≤ F (x) + F(y). Hence, F is a norm function on X. Indeed, by Proposition 1, F is a Banach
norm on X. For convenience, let F (x) = ∥x∥. Since x ∈ X and f is a continuous function, the value of
Z t
CF β 1 α–1 β
f s, xs , CF D0 [x (s)] ds
h  i  
α
I0 f t, xt , D0 [x (t)] = (t – s)
Γ (α) 0
exists for 0 ≤ s < t ≤ 1. Thus, our problem is equivalence to
 Rt α–1 β
 x0 + x1 + Γ(1α) 0 (t – s) f s, xs , CF D0 [x (s)] ds, t ∈ [0, 1] ,
 


x (t) = 
 η (t) ,

t ∈ [–r, 0) .
6 M. Rohman et al.

Let T : X → X be an operator which maps x ∈ X to its solution. Then, for all t ∈ [–r, 0) we have
Tx (t) – Ty (t) = 0. On the other hand, for t ∈ [0, 1] and l ∈ [–r, 0) we get

max xs – ys = max x (s) – y (s)

0≤s<t 0≤s<t
= max x(s + l) – y(s + l)
l≤s+l<t+l
≤ max x(s + l) – y(s + l)
–r ≤s+l<1
= x–y .

Using this result and the fact that x0 = y0 and x1 = y1 , the condition (ii) will show that

Z t
1 α–1 β
f s, xs , CF D0 [x (s)] ds – y0
 
Tx (t) – Ty (t) = x0 + x1 + (t – s)
Γ (α) 0
Z t
1 α–1 β
f s, ys , CF D0 y (s) ds
 h i
– y1 – (t – s)
Γ (α) 0
Z t " 
1 α–1 β
f s, xs , CF D0 [x (s)]

≤ (t – s)
Γ (α) 0
i #
CF β
 h
– f s, ys , D0 y (s) ds

N
Z t "
α–1
≤ (t – s) xs – ys
Γ (α) 0

β β
i#
CF
D0 [x (s)] – CF D0
h
+ y (s) ds

N
≤ max x (s) – y (s)
Γ (α) 0≤s<t
β β
i !Z t α–1
CF
D0 [x (s)] – CF D0 y (s)
h
+ max (t – s) ds
0≤s<t 0

Ntα
≤ max x (s) – y (s)
αΓ (α) 0≤s<t
β β
i !
CF
D0 [x (s)] – CF D0
h
+ max y (s)
0≤s<t

N
≤ x–y .
Γ (α + 1)

for α > 0. On the other side, if we take the CF fractional derivative of Tx (t) and Ty (t) for 0 < β < 1,
then
 Indonesian Journal of Mathematics and Applications 7

CF β β
D0 [Tx (t)] – CF D0 Ty (t)
h i

(2 – β) M(β)
Z t –β
!
= exp (t – s) D1 (Tx (t) – Ty (t)) ds
2 (1 – β) 0 1–β
Z t
(2 – β) M(β)
!
–β
= exp (t – s)
2 (1 – β) 0 1–β
α–1 t
Z " 
α–2 β
f s, xs , CF D0 [x (s)]

(t – s)
Γ (α) 0
# !
CF β

– f s, xs , D0 [x (s)] ds ds

(2 – β) M(β) (α – 1)
Z t –β
!
≤ exp (t – s)
2 (1 – β) Γ (α) 0 1–β
Z t
α–2 CF β
 
(t – s) f s, xs , D0 [x (s)]
0
! !
CF β
 
– f s, xs , D0 [x (s)] ds ds

(2 – β) M(β) (α – 1)
≤ max x (s) – y (s)
2 (1 – β) Γ (α) 0≤s<t

β β
i !
CF
D0 [x (s)] – CF D0 y (s)
h
+ max
0≤s<t
Z t !
–β
exp (t – s) ds
0 1–β
N (2 – β) M (β) (α – 1)
≤ x–y .
2 (1 – β) Γ (α)

Therefore,

β CF β
D0 [Tx (t)] – CF D0 Ty (t)
h i
Tx – Ty = max Tx (t) – Ty (t) + max
0≤t<1 0≤t<1
N (2 – ) M (β) (α – 1)
!
N β
≤ + x–y .
Γ (α + 1) 2 (1 – β) Γ (α)

N N(2–β)M(β)(α–1)
Since Γ(α +1)
+ 2((1–β))Γ(α) = c < 1, T is a contraction mapping. Hence, the problem has a
unique solution in X (see [24], Theorem 3.4 and [12], Theorem 2.1).
To get the Hyers-Ulam stability, let x ∈ X be the solution of the problem and y ∈ X be its
β
h i  h i
approximation solution as in Definition 6. Let g (t) = CF D0α y (t) – f t, yt , CF D0 y (t) . Then
8 M. Rohman et al.

Z t
1 α–1 β
f s, ys , CF D0 y (s) ds
 h i
y (t) – y0 – y1 – (t – s)
Γ (α) 0
β
= I0α CF D0α y (t) – f t, yt , CF D0 y (t)
h h i  h ii

h i
= I0α g(t)
Z t
1 α–1
= (t – s) g(s)ds
Γ (α) 0
εtα εtα
≤ = .
αΓ (α) Γ (α + 1)

Recall that y0 = x0 and y1 = x1 . Hence,

1
Z t
α–1 CF β
 h i
y (t) = y0 + y1 t + (t – s) f s, ys , D0 y (s) ds
Γ (α) 0
1
Z t
α–1 CF β
 h i
= x0 + x1 t + (t – s) f s, ys , D0 y (s) ds.
Γ (α) 0

Since y (t) = x(t) for any t ∈ [–r, 0), we get y (t) – x(t) = 0 and there is nothing to be proved. For
t ∈ [0, 1] we have

y (t) – x(t)

1
Z t
α–1 CF β
 
= y (t) – x0 – x1 t – (t – s) f s, xs , D0 [x (s)] ds
Γ (α) 0
1
Z t
α–1 CF β
 h i
≤ y (t) – y0 – y1 t – (t – s) f s, ys , D0 y (s) ds
Γ (α) 0
Z t
1 α–1 β β
f s, ys , CF D0 y (s) – f s, xs , CF D0 [x (s)] ds
h  h i  i
+ (t – s)
Γ (α) 0
εtα
≤
Γ (α + 1)
Z t
1 α–1 β β
f s, ys , CF D0 y (s) – f s, xs , CF D0 [x (s)] ds
 h i  
+ (t – s)
Γ (α) 0
Z t
εtα N α–1

β β

ys – xs + CF D0 y (s) – CF D0 [x (s)] ds.
h i
≤ + (t – s)
Γ (α + 1) Γ (α) 0

Now, let ys – xs = µs ∈ C ([–r, 1] , R+ ) and an operator P : C ([–r, 1] , R+ ) → C ([–r, 1] , R+ ) be

defined as
(
A1 + A2 + A3 , t ∈ [0, 1]
Pµ (t) =
0, t ∈ [–r, 0) .
α t α–1
Rt α–1 CF β
where A1 = Γ(εαt +1) , A2 = ΓN µs ds, and A3 = ΓN
R
(α) 0
(t – s) (α) 0
(t – s) D0 [µs ]ds.
 Indonesian Journal of Mathematics and Applications 9

By this definition, we get

Pµ (t) – Pω (t)

N
Z t Z t
α–1 α–1 CF β β
D0 [µs ] – CF D0 [ωs ] ds
 
= (t – s) (µs – ωs )ds + (t – s)
Γ (α) 0 0
N
Z t
α–1
≤ (t – s) µs – ωs ds
Γ (α) 0
Z t !
α–1 CF β β
+ (t – s) D0 [µs ] – CF D0 [ωs ] ds
0
N
!Z t
β β α–1
≤ max µs – ωs + max CF D0 [µs ] – CF D0 [ωs ] (t – s) ds
Γ (α) 0≤s<t 0≤s<t 0
tα N N
≤ ∥µ – ω∥ ≤ ∥µ – ω∥,
αΓ (α) αΓ (α)

and

CF β β
D0 [Pµ (t)] – CF D0 [Pω (t)]

(2 – β) M(β) –β
Z t !
= (t – s) D1 (Pµ (t) – Pω (t)) ds
exp
2 (1 – β) 0 1–β
Z t
(2 – β) M(β) (α – 1) N t
!"
–β
Z
α–2
= exp (t – s) (t – s) (µ (s) – ω (s)) ds
2 (1 – β) 0 1 – β Γ (α ) 0
(α – 1) N t  #
Z
α–2 CF β CF β

+ (t – s) D0 [µs ] – D0 [ωs ] ds ds
Γ (α) 0
Z t
(2 – β) M(β) (α – 1) N
!
–β
≤ ∥µ – ω∥ exp (t – s) ds
2 (1 – β) Γ (α) 0 1–β
N (2 – β) M (β) α – 1
≤ ∥µ – ω∥.
2 (1 – β) Γ (α)

Therefore,

N (2 – β) M (β) (α – 1)
!
N
∥Pµ – Pω∥ ≤ + x–y ,
Γ (α + 1) 2 (1 – β) Γ (α)
and again the condition (iii) shows that P is a contraction. Hence, P is a Picard operator and FP = {u0 }
for some u0 ∈ X. Thus,

εtα N
Z t
α–1
u0 (t) ≤ + (t – s) u0 (s)ds
Γ (α + 1) Γ (α) 0
N
Z t
α–1 CF β
h i
+ (t – s) D0 u0 (s) ds,
Γ (α) 0

and the condition (i) shows that

10 M. Rohman et al.

εtα N
Z t
α–1
u0 (t) ≤ + (t – s) u0 (s)ds
Γ (α + 1) Γ (α) 0
N
Z t
α–1
+ (t – s) u0 (s)ds.
Γ (α) Γ (2 – β) 0

Applying Gronwall lemma, we get

!Z t
εtα
" #
N 1 α–1
u0 (t) ≤ exp 1+ (t – s) ds
Γ (α + 1) Γ (α) Γ (2 – β) 0
εt α " α
t N 1
!#
≤ exp 1+
Γ (α + 1) Γ (α) Γ (2 – β)
" !#
ε N 1
≤ exp 1+ .
Γ (α + 1) Γ (α) Γ (2 – β)

By the definition of operator P, we see that P is an increasing Picard operator. If we set

u (t) = y (t) – x(t) for some t ∈ [–r, 1], then u(t) ≤ Pu(t). Thus, the abstract Gronwall lemma implies
u(t) ≤ u0 (t) and so

y(t) – x(t) ≤ kε, t ∈ [–r, 1]

  
where k = Γ(α1+1) exp ΓN
(α)
1 + 1
Γ(2–β)
. This result completes the proof.
□

Example. Consider the following problem

CF D5/4 1 1 1 CF 3/4
 h i h i



 0
x(t) = 10 x (t) + 10 x (t – 0, 1) + 10 D0 x(t – 0.1) , t ∈ [0, 1]
x (t) = 0, 3 , t ∈ [–0.1, 0)



 ′
x (0) = x (0) = 0



and ε-perturbed problem

h i 1 1 1 CF 3/4 h
CF
D05/4 y(t) –
i
y (t) + y (t – 0, 1) + D0 y(t – 0.1) ≤ ε
10 10 10

for ε > 0. The problem gives α = 54 , β = 34 and M 43 = 85 . If we set N = 17 , then these problems
 

satisfy condition (i) and (ii) of Theorem 1. Since Γ 54 ≈ 0.90640 and Γ 94 ≈ 1.13300, we have
   

N N (2 – β) M (β) (α – 1)
c= + ≈ 0.75653 < 1
Γ (α + 1) 2 (1 – β) Γ (α)

and hence satisfies the condition (iii). Therefore, the problem has a unique solution and Hyers-Ulam
stable with
" !#
1 N 1
k= exp 1+ ≈ 1.22952 > 0.
Γ (α + 1) Γ (α) Γ (2 – β)
 Indonesian Journal of Mathematics and Applications 11

Acknowledgement
The authors are very grateful to the referee(s) for their constructive suggestions.

Funding This research received no specific grant from any funding agency in the public, commercial,
or not-for-profit sectors.

References
[1] R. Agarwal, S. Hristova and D. O’Regan, Ulam type stability for non-instaneous impulsive
Caputo fractional differential equations with finite state dependent delay, Georgian Math. J., 40
(2021), 499–517.

[2] D. Baleanu, A. Mousalou and S. Rezapour, On the existence of solutions for some infinite
coefficient-symmetric Caputo-Fabrizio fractional integro-differential equations, Bound. Value
Probl. 2017 (2017), article no. 145. DOI: 10.1186/s13661-017-0867-9.

[3] M. Caputo and M. Fabrizio, A new definition of fractional derivative without singular kernel,
Progr. Fract. Differ. Appl. 1 (2015), 73–85.

[4] L. Chen, R. Wu, J. Cao and J.-B. Liu, Stability and synchronization of memristor-based
fractional-order delayed neural networks, Neural Netw. 71 (2015), 37–44.

[5] L. Cheng and Y. Dong, A note on the stability of nonsurjective ε-isometries of Banach spaces,
Proc. Amer. Math. Soc. 148 (2020), 4837–4844.

[6] Q. Dai, R. Gao, Z. Li and C. Wang, Stability of Ulam-Hyers and Ulam-Hyers-Rassias for a
class of fractional differential equations, Adv. Differ. Equ. 2020 (2020), article no. 103.

[7] J. Huang and Y. Li, Hyers-Ulam stability of delay differential equations of first order, Math.
Nachr. 289 (2016), 60–66.

[8] D.H. Hyers, On the stability of the linear functional equation, Proceedings of the National
Academy of Sciences of the United States of America, 27 (1941), 222–224.

[9] D.H. Hyers and S.M. Ulam, On approximate isometries, Bull. Amer. Math. Soc. 51 (1945),
288–292.

[10] S.-M. Jung, Hyers-ulam stability of linear differential equations of first order, Appl. Math. Lett.
17 (2004), 1135–1140.

[11] S.-M. Jung, Hyers-Ulam-Rassias Stability of Functional Equation in Nonlinear Analysis, (Springer,
New York, 2011).

[12] S.-M. Jung and H. Rezaei, A fixed point approach to the stability of linear differential equations,
Bull. Malays. Math. Sci. Soc. 38 (2015), 855–865.

[13] H.E. Khochemane, A. Ardjouni and S. Zitouni, Existence and Ulam stability for two orders
delay Fractional differential equations, Rend. Mat. Appl. 40 (2019), 141–158.

[14] H.E. Khochemane, A. Ardjouni and S. Zitouni, Existence and Ulam stability results for two
orders neutral Fractional differential equations, Afr. Mat. 33 (2022), 141–158.

[15] S. Krim, S. Abbas and M. Benchohra, Caputo-Hadamard implicit fractional differential equa-
tions with delay, Sao Paolo J. Math. Sci. 15 (2021), 463–484.
12 M. Rohman et al.

[16] P. Kumar, D. Baleanu, V.S. Erturk and V. Govindaraj, A delayed plant disease model with
Caputo fractional derivatives, Adv. Cont. Discr. Mod. 2022 (2022), article no. 11.

[17] H. Liu, and Y. Li, Hyers-Ulam stability of linear fractional differential equations with variable
coefficients, Adv. Differ. Equ. 2020 (2020), article no. 404.

[18] K. Liu, M. Fečkan, D. O’Regan and J.R. Wang, Hyers-Ulam stability and existence of solutions
for differential equations with Caputo-Fabrizio Fractional derivative, Mathematics, 7 (2019),
article no. 333. doi.org/10.3390/math7040333.

[19] J. Losada and J.J. Nieto, Properties of a new fractional derivative without singular kernel, Prog.
Fract. Differ. Appl., 1 (2015), 87–92.

[20] S. Mazur and S. Ulam, Sur les transformations isométriques d’espaces vectoriels normés, C R
Acad. Sci. Paris. 194 (1932), 946–948.

[21] M. Obłoza, Connections between Hyers and Lyapunov stability of the ordinary differential
equations, Rocznik Nauk.-Dydakt. Prace Mat. 14 (1997), 141–146.

[22] M. Omladič and P. Šemrl, P., On Nonlinear Perturbation of Isometries, , Math. Ann. 303 (1995),
617–628.

[23] D. Otrocol and V. Ilea, Ulam stability for a delay differential equation, Cent. Eur. J. Math. 11
(2013), 1296–1303.

[24] I. Podlubny, Fractional Differential Equations, (Academic Press, California, 1999).

[25] M. Rohman and I. Eryßlmaz, Weak stability of ε-isometry Mapping on Real Banach Spaces,
Euoro. J. Sci. Tech. 34 (2022), 110–114.

[26] H.A. Salem and M. Vath, An abstract Gronwall lemma and applications to global existence
results for functional differential and integral equations of fractional order, J. Integral Equ. Appl.
16 (2004), 411–439.

[27] H. Smith, An Introduction to Delay Differential Equations with Applications to the Life Sciences,
(Springer, New York, 2011).

[28] S.M. Ulam, A Collection of the Mathematical Problems, (Interscience, New York, 1960).

[29] J.R. Wang, L. Lv and Y. Zhou, Ulam stability and data dependence for fractional differential
equations with Caputo derivative, Electron. J. Qual. Theory Differ. Equ. 2011 (2011), article no.
63. DOI: 10.14232/ejqtde.2011.1.63.

[30] S. Wang, The Ulam stability of fractional differential equation with the Caputo-Fabrizio
derivative, J. Funct. Spaces 2022 (2022), article ID 7268518. DOI: 10.1155/2022/7268518

[31] Z. Yang, J. Zhang, J. Hu and J. Mei, Finite-time stability criteria for a class of high-order
fractional Cohen-Grossberg neural networks with delay, Complexity, 2020 (2020), DOI:
10.1155/2020/3604738.