Introduction to Fractional Calculus

A Project
Submitted to the University of Nizwa in Partial Fulfillment of the
Requirements for the Degree of Bachelor of Education in Mathematics

By

AMNA ABDULSALAM YOUNIS AL-AMRY

Supervised by
Professor Dr. Ahmed S. El-Karamany

2010
 Supervisor Certification

We certify that this thesis entitled" Introduction of Fractional

Calculus" have been made by the Bachelor student

AMNA ABDULSALAM YOUNIS AL-AMRY under my supervision at the

University of Nizwa as Partial Fulfillment of the Requirements for the

Degree of Bachelor of Education in Mathematics.

Signature

Professor Dr. Ahmed Sadek El-Karamany

In view of the available of the recommendations I forwarded this

thesis for debate the examining committee.

Signature
 Committee Certification

We certify that, we have read this project entitled" Introduction of Fractional

Calculus " and examining committee examined the student AMNA

ABDULSALAM YOUNIS AL-AMRY in its contents and what it connected with it,

and that our opinion it meets the standard of project for degree bachelor of Education

in Mathematics.

1) Signature

Name: Professor Dr. Ahmed S. El-Karamany

Chairman and Supervisor. Date: 27/5/2010

2) Signature

Name: Professor Dr. Mohammad Elatrash

Member Date: 27/5/2010

3) Signature
Name: Dr. Mahmood Khalid Jasim
Associate Professor
Member Date: 27/5/2010

Approved by:
Professor Dr. Ahmed S. El-Karamany
Head of Department of Mathematical and Physical Sciences.
Date: 27/5/2010
 Abstract

Abstract

This work is devoted to explore Fractional Calculus which has been used
successfully to modify many existing models of Physical processes.
The Work consists of 5 chapters. In chapter 1; the introduction, some historical notes
and an approach to fractional calculus using generalization of Cauchy formula for
n –folde integral. Laplace transform of the convolution and its properties are given.
Integrals and derivatives of fractional order are defined in chapter 2. In this Chapter
integrals and derivatives of fractional order of power and exponential functions are
calculated. Semi-integrals and semi-derivatives of power function, exponential function,
sine and cosine functions are given in chapter 3. In chapter 4 some illustrating solved
problems are given. The Tautochrone problem is discussed in chapter 5 as an application
of semi-derivatives in solving Abel’s integral equation. A list of references is given at the
end of this project.

i
 ACKNOWLEDGEMENT

Thanks firstly and finally to God who deserves all thanks.

I would like to express my deep appreciation to my supervisor

Professor Dr. Ahmed Sadek El-Karamany the Head of department of

Mathematical and Physical Sciences for his encouragement, Support,

guidance and valuable suggestions. His enthusiasm and interest for

science are contagious.

I was also lucky to be able to associate myself with the talented

and hard working members of the DMPS.

My special thanks are due to the University of Nizwa, College

of Arts & Sciences and to all staff members for their assistance and

support during my study.

My thanks also go to my friends and colleagues for their support

and help.

AMNA ABDULSALAM YOUNIS AL-AMRY

 Contents

Contents

Chapter 1
1.1 Introduction……………………………………………………………………. …..1
1.2 Birth of Fractional Calculus ………………………………………………………..2
1.3 Fractional Calculus as Generalization of Integer Order Calculus…………....... 3
Chapter 2
Fractional integrals and derivatives……………………………………………………6

Chapter 3
Semi- integrals and Semi-derivatives………………………………………………….12
Chapter 4
Semi- Integrals and Semi-Derivatives of Some Algebraic Functions ……………….16
Chapter 5: Application
The Tautochrone Curve ……………………………………………………………… 19
References …………………………………………………………………………… 22

i
 1

Chapter 1
Introduction
1.1 Introduction
Fractional calculus is three centuries old as the conventional calculus, but not very
popular among science and/or engineering community. The beauty of this subject
is that fractional derivatives (and integrals) are not a local (or point) property (or
quantity). Thereby this considers the history and non-local distributed effects. In
other words, perhaps this subject translates the reality of nature better! Therefore
to make this subject available as popular subject to science and engineering community,
it adds another dimension to understand or describe basic nature in a better
way. Perhaps fractional calculus is what nature understands, and to talk with nature
in this language is therefore efficient. For the past three centuries, this subject was with
mathematicians, and only in last few years, this was pulled to several (applied) fields
of engineering , science and economics. However, recent attempt is on to have
the definition of fractional derivative as local operator specifically to fractal science
theory. Next decade will see several applications based on this 300 years (old) new
subject, which can be thought of as superset of fractional differintegral calculus, the
conventional integer order calculus being a part of it. Differintegration is an operator
doing differentiation and sometimes integrations, in a general sense. In this project,
fractional order is limited to only real numbers; the complex order differentigrations
are not touched. Also the applications and discussions are limited to fixed fractional
order differintegrals, and the variable order of differintegation is kept as a future
research subject. Perhaps the fractional calculus will be the calculus of twenty-first
century. In this project, attempt is made to make this topic application oriented for
regular science and engineering applications. Therefore, rigorous mathematics is
kept minimal. In this introductory chapter.
1.2 Birth of Fractional Calculus
In a letter dated 30th September 1695, L’Hopital wrote to Leibniz asking him a
particular notation that he had used in his publication for the nth derivative of a
function
 2

D nf
Dx n
i.e., what would the result be if n = 1/2. Leibniz’s response “ an apparent paradox

from which one day useful consequences will be drawn.” (Debnath 2004) In these
words, fractional calculus was born. Studies over the intervening 300 years have proved
at least half right. It is clear that within the twentieth century especially numerous
applications have been found. However, these applications and mathematical background
surrounding fractional calculus are far from paradoxical. While the physical meaning is
difficult to grasp, the definitions are no more rigorous than integer order counterpart.

1.3 Fractional Calculus as Generalization of Integer Order Calculus

Let us consider n an integer and when we say xn we quickly visualize x multiply n
times will give the result. Now we still get a result if n is not an integer but fail to

visualize how. Like to visualize 2π is hard to visualize, but it exists.

An approach to fractional calculus may be given using the generalization of Cauchy

formula for n –folder integral , the Laplace Transform of the convolution and its
properties (Gorenflo and Mainardi 1997).
The Laplace Transform of the function f (t ) is defined by
∞

∫
L{ f (t )} = e − st f (t )dt = F ( s ) (1.1)
0

The convolution of two functions f (t ) and g (t ) is denoted by (f ∗ g )(t ) and is

defined by the relation
t t

∫ ∫
(f ∗ g )(t ) = f (t − x ) g (x )dx = f (t − y ) g ( y )dy (1.2)
0 0

It is easy to prove that

t t

∫ ∫
(f ∗ g )(t ) = f (t − x ) g (x )dx = ( g ∗ f )(t ) = g (t − x )f (x )dx (1.3)
0 0
 3

We prove the following theorem

Theorem 1.1:
L{( f ∗ g )} = L{ f (t )}L{g (t )} = F ( s )G ( s ) (1.4)
where
F ( s ) = L{ f (t )}, G ( s ) = L{g (t )} (1.5)

Proof: From the definitions of Laplace transform and the convolution we get
∞ t ∞ t

∫ ∫ ∫ ∫e
− st − st
L{ f ∗ g} = e ( f ( y ) g (t − y )dy ) dt = f ( y ) g (t − y ) dydt (1.6)
0 0 0 0

The region of integration is a simple vertically region in the tOy plane where Oy is the
vertical axis and Ot is the horizontal axis.
It is the triangle in the first coordinate quarter whose base in the infinity and whose sides

are the axis O t and the line y = t : RV = {(t , y ) : 0 ≤ t < ∞ 0 ≤ y ≤ t} .

Reversing the order of integration we convert the region to simple horizontally:

R h = {(t , y ) : y ≤ t < ∞ 0 ≤ y < ∞} and the integral (1.6) takes the form
∞ t ∞ ∞

∫ ∫e ∫ ∫e
− st − st
L{ f ∗ g} = f ( y ) g (t − y )dydt = f ( y ) g (t − y ) dtdy
0 0 0 y
∞ ∞

∫ f ( y)(∫ e
− st
= g (t − y )dt )dy
0 y

Substituting t − y = x into the internal integral, where y is kept fixed, we get

∞ ∞ ∞ ∞

∫ ∫e ∫ ∫e
− st − s ( x+ y )
L{ f ∗ g} == f ( y )( g (t − y )dt )dy = f ( y )( g ( x)dx) dy
y =0 t= y y =0 x =0
∞ ∞

∫ ∫
− sy
=( e f ( y )dy )( e − sx g ( x)dx) = F ( s )G ( s )
0 0

This proves the theorem. As a consequence we have

t

∫
−1
L {F (s )G (s )} = (f ∗ g ) = f (t − x ) g (x )dx (1.7)
0
 4

Below is the Laplace transforms and inverse Laplace transforms of some functions that
play an essential role in the generalization we are going to present
The Laplace transform and the inverse Laplace transform of some functions:
Laplace Transform The Inverse Laplace Transform
1 (n − 1)! 1 t n −1
L {t n −1} = −1
L { n}=
sn s Γ (n )
2 t t
F (s ) F (s )
∫ ∫
−1
L { f (x )dx } = L { } = f (x )dx
s s
0 0

3 t t t t
F (s ) F (s )
∫∫ ∫∫ f (x )dxdx
−1
L { f (x )dxdx } = 2 L { 2 }=
00
s s 00

4 t t t t t t
F (s ) F (s )
∫∫ … ∫ ∫∫… ∫
−1
L{ f (x )dxdx …dx } = n L { n }= f (x ) dxdx …dx
0o 0
s s 00 0 n
n n
−n
=D f (t )

Where
∞ ∞

∫ ∫
2
p −1 − x
Γ( p ) = x e dx = 2 x 2 p −1e − x dx (1.8)
0 0

1 t n −1
Using the convolution theorem with G (s ) = n , g (t ) = we obtain
s Γ(n )

t n −1
t t
F (s ) 1 1
∫ ∫
−1
L { n } = (f ∗ )(t ) = f (x )(t − x ) n −1dx = x n −1f (t − x )dx
s Γ ( n ) Γ ( n )
0
Γ ( n )
0

Therefore, we get from the last row

t t
1 1
∫ ∫
−n
D f (t ) = (t − x ) n −1 f (x )dx = x n −1f (t − x )dx (1.9)
Γ (n ) 0
Γ (n ) 0

Here we have used that if two continuous functions have the same Laplace transform they
are identical (Learch Theorem). This means that we restrict our derivation to include only
continuous functions
 5

Equation (1.8) is called the Cauchy formula for n-folder integral where n ∈ .
Riemann and Liouville generalized Cauchy formula to any fractional number ν and got
the Riemann-Liouville fractional integral
t
1
∫
−ν
D f (t ) = (t − x )ν −1 f (x )dx ν∈ +
,t > 0 (1.9)
Γ (ν ) 0

When 0 < ν < 1 we have an improper integral of second type which is convergent by
the p- test since 1 −ν < 1 . The integral is also convergent for ν = 1 . Therefore the
conditions for the existence of integral (1.9) is that f (t ) should be Riemann integrable
on (0, ∞) .
The Riemann – Liouville fractional derivative is defined by

D ν f (t ) = D m ⎡⎣ D − ( m −ν )f (t ) ⎤⎦
t
dm dm 1
dt Γ(m −ν ) ∫0
= m D − ( m −ν ) f (t ) = m [ (t − x )m −ν −1 f (x )dx ]
dt
m −1 <ν ≤ m , m ∈
We shall not discuss here other definitions of fractional derivatives where the function
f (t ) is required to be Lebesgue integrable and absolutely continuous (Samko et all
1993).

AMNA ABDULSALAM YOUNIS AL-AMRY

 6

Chapter 2
Fractional integrals and derivatives

Definition (1): Let ν > 0 and let f be continuous on (0, ∞) and integrable on any

finite subinterval [a , b ] . Then we shall say

that f is an element in the class C, and for t > 0 , we call
t
D −ν f =
1
Γ(ν ) ∫0
(t − x )ν −1 f (x ) dx (2.1a)

the Riemann-Liouville fractional integral of f (t ) of order ν > 0 .The symbol

J ν (f ) is also used for the expression (2.1) . We can add to the definition (1) that
t

∫ f (t ) dt
−1
J f (t ) = D f (t ) = f (t ) , J f (t ) = D f (t ) =
0 0 1
(2.1b)
0

Using the linearity property of the definite integral we obtain

t
1
∫
−ν
D [C 1f (t ) ± C 2 g (t )] = (t − x )ν −1[C 1f (x ) ± C 2 g ( x )]dx (2.2)
Γ (ν ) 0

t t
C C
∫ ∫
= 1 (t − x )ν −1 f (x )dx ± 2 (t − x )ν −1 g (x )dx = C 1D −ν f (t ) ± C 2 D −ν g (t )
Γ (ν ) Γ (ν )
0 0

Where C 1 and C 2 are constants.

Because we now have a well-defined class of functions to which our definition of

the fractional integral applies, it is worthwhile to see the fractional integrals of
the power function . The fractional integration of even the simplest functions can lead
to complicated higher transcendental functions.
Example (1): The fractional integral of the power function.

Let f (t ) = t λ , λ > −1 . Then,

t
1
∫
−ν λ
D (t ) = (t − x )ν −1 x λ dx (2.3)
Γ (ν ) 0

Substituting x = t u into Eq. (2.3) we get

(t ν +λ ) t ν +λ t ν +λ Γ (λ + 1)
1

∫
−ν λ
D (t ) = (1 − u )ν −1u λ du = B (ν , λ + 1) = (2.4)
Γ (ν ) 0
Γ (ν ) Γ (ν + λ + 1)
 7

Example (2): The fractional integral of the constant function

Let f (t ) = C . Then, using the preceding expression (2.4) and the linearity

property (2.2) one obtains

Ct ν +λ Γ (λ + 1) C tν
D −ν (C ) = lim[D −ν (Ct λ )] = lim = (2.5)
λ →0 λ →0 Γ (ν + λ + 1) Γ (ν + 1)
As special case we get the fraction integral of f (t ) = 1

tν
D −ν (1) = ν >0 (2.6)
Γ (ν + 1)
Example (3): The fractional integral of the exponential function

f (t ) = e αt , α ≠ 0.

e αt
t t
1
∫ ∫
−ν αt
D (e ) = x ν −1e α (t −x )dx = x ν −1e −α x dx
Γ (ν ) 0
Γ (ν ) 0

Substitution α x = u leads to:

αt
e αt α −ν e αt α −ν
∫u
−ν αt ν −1 −u
D (e ) = e du = γ (ν ,αt ) = E t (ν ,α ) (2.7)
Γ (ν ) 0
Γ (ν )

Where: the incomplete gamma function γ (ν , x ) is defined by the relation:

x

∫
γ (ν , x ) = t ν −1e −t dt (2.8)
0

and the Et- function E t (ν ,α ) is defined by the relation:

α −ν e αt γ (ν ,αt )
E t (ν , α ) = (2.9)
Γ (ν )
Definition (2):: The fractional derivative of f (t ) of order µ > 0 (if exists) can be

defined in terms of fractional integral D −ν f (t ) as:

D µ f (t ) = D m ⎡⎣ D − ( m − µ ) f (t ) ⎤⎦ (2.10)

Where m is an integer ≥ [ µ ] , and [ x ] = n , x ∈ [n , n + 1) is the ceiling function

(the greatest integer that equal or less than x ).

Example (4): The fractional derivative of the power function

f (t ) = t λ , λ > −1
 8

Applying the definition (2.10) Using Eq. (2.4) with ν = m − µ we get

⎡ Γ (λ + 1) t λ − µ + m ⎤
D µ (t λ ) = D m [D − ( m − µ ) (t λ )] = D m ⎢ ⎥ (2.11)
⎣ Γ (1 + m + λ − µ ) ⎦
Since m is integer, Eq. (2.11) yields

[Γ (λ + 1)][(λ − µ + m )(λ − µ + m − 1)… (λ − µ + 1)] t λ − µ

D µ (t λ ) = (2.12)
Γ (1 + m + λ − µ )
From the properties of the Gamma function we have
Γ (λ − µ + 1 + m ) = (λ − µ + m )(λ − µ + m − 1)… (λ − µ + 1)Γ (λ − µ + 1) (2.13)
Substituting from (2.13) into Eq. (2.12) we get
Γ (λ + 1)
D µ (t λ ) = (t λ − µ ) (2.14)
Γ (λ + 1 − µ )
For integer m and n we have
Γ (m + 1)
D n (t m ) = (t m−n ), m>n (2.15)
Γ (m − n + 1)
Example (5): The fractional derivative of a constant function
Eq. (2.14) when λ → 0 yields

t −µ
D µ (1) = , ( µ is a noninteger ) (2.16)
Γ (1 − µ )
Before proving some properties of the fractional derivatives we introduce the Leibniz
rule for the fractional derivative of product of two functions.
Leibniz rule: For integer n the Leibniz rule for the product of two n time’s
differentiable functions can be written in the form
k =n
Γ(n + 1)
D n [ f (t ) g (t )] = ∑ Γ(k + 1)Γ(n + 1 − k ) Dn−k f (t ) Dk g (t )
k =0
n(n − 1) n−2
( ) ( )
= D n f (t ) g (t ) + n D n−1 f (t ) Dg (t ) +
2
( )
D f (t ) D 2 g (t ) + (
+ f (t ) D n g (t ) )
k =n
Γ( n + 1)
= ∑ Γ(k + 1)Γ(n + 1 − k ) Dk f (t ) Dn−k g (t ) (2.17)
k =0

The analogy between the two expressions (2.14) and (2.15) suggests to generalize
(2.17) for the fractional derivative of order ν > 0 as follows
 9

∞
Γ(ν + 1)
Dν [ f (t ) g (t )] = ∑ Γ(k + 1)Γ(ν + 1 − k ) Dν −k f (t ) D k g (t )
k =0

( ) (
= D f (t ) g (t ) + ν Dν −1 f (t ) Dg (t ) +
ν
) (
+ f (t ) Dν g (t ) +)
∞
Γ(ν + 1)
= ∑ Γ(k + 1)Γ(ν + 1 − k ) Dk f (t ) Dν −k g (t ) (2.18)
k =0

Since ν is not integer the upper limit of the sum in (2.18) is infinite.
Now, we are able to prove the following properties of the fractional derivative.
Property (1): Let g (t ) = C is the constant function, then

Dν [Cf (t )] = CDν [ f (t )] (2.19)

Proof: since D 0C = C , and D mC = 0, m = 1, 2, we get upon substituting

g (t ) = C into (2.18):
∞
Γ(ν + 1)
ν
D [ f (t )C ] = ∑ Γ(k + 1)Γ(ν + 1 − k ) Dν −k f (t ) Dk C = ( Dν f (t ) ) C = C ( Dν f (t ) )
k =0

Property (2):

(
Dν [ f (t ) + h(t )] = Dν f (t ) + Dν h(t ) ) ( ) (2.20)

Proof: Let g (t ) = 1 = t 0 then, Dν [ f (t ) + h(t )] = Dν [t 0 ( f (t ) + h(t ) )] and

∞
Γ(ν + 1)
Dν [t 0 ( f (t ) + h(t ) )] = ∑ Γ(k + 1)Γ(ν + 1 − k ) Dν −k f (t ) Dk t 0
k =0
∞
Γ(ν + 1)
+ ∑ Γ(k + 1)Γ(ν + 1 − k ) Dν −k h(t ) Dk t 0 = ( Dν f (t ) ) + ( Dν g (t ) )
k =0

Combining properties (1) and (2) we obtain the following linearity property of the
fractional derivative

( )
D ν [C 1f (t ) + C 2 h (t )] = C 1 D ν f (t ) + C 2 D ν h (t ) ( ) (2.21)

Where C 1 and C 2 are constants.

Theorem (1): The Index law for fractional integrals

(i) D −ν ( D − µ f (t )) = D − (ν + µ ) f (t ) (2.25a)

(ii) D − µ ( D −ν f (t )) = D −ν ( D − µ f (t )) (2.25b)
For f ∈C , 0 < ν < 1 and 0 < µ < 1
 10

Proof: Let g (t ) = D − µ f (t ) and using Eq. (2.1) we have

t
1
∫
−µ
g (t ) = D f (t ) = (t − ξ ) µ −1 f (ξ )d ξ (2.26)
Γ( µ )
0

Applying D −ν to Eq. (2.26) we obtain

t x
1
∫ ∫
−ν −µ
D (D f (t )) = (t − x )ν −1 ( (x − ξ ) µ −1 f (ξ )d ξ )dx
Γ(ν )Γ( µ )
0 0
(2.27)
t x
1
∫∫ (t − x )
ν −1
= (x − ξ ) µ −1 f (ξ )d ξ dx
Γ(ν )Γ( µ )
00

If we consider the vertical axis is O ξ then the region of integration in the preceding
double integral is the triangle OPM in Fig. (1) and it is a simple vertical region

R v = {( x, ξ ) : 0 ≤ x ≤ t , 0 ≤ ξ ≤ x}

Changing the region of integration from simple vertical to simple horizontal we get

R h = {(x , ξ ) : x ≤ ξ ≤ t , 0 ≤ x ≤ t } .Reversing the order of integration we get

t t
1
∫∫
−ν −µ
D (D f (t )) = ( (x − ξ ) µ −1 (t − x )ν −1dx )f (ξ )d ξ (2.28)
Γ(ν )Γ( µ )
0 ξ

In the inner integral we substitute

x = ξ + u (t − ξ ) , dx = (t − ξ )du , ( ξ and t are fixed) . (2.29)
As x = ξ + u (t − ξ ) = ξ → u = 0, x = t = ξ + u (t − ξ ) → u = 1 , and
x − ξ = u (t − ξ ), (t − x ) = (t − ξ )(1 − u ) , then Eq. (2.28) leads to
t 1
1
∫∫
−ν −µ
D (D f (t )) = ( u µ −1 (1 − u )ν −1 (t − ξ )ν + µ −1du )f (ξ )d ξ
Γ(ν )Γ( µ )
0 0
 11

t 1 t
1 B ( µ ,ν )
=
Γ(ν )Γ( µ ) ∫∫
( u µ −1 (1 − u )ν −1du )(t − ξ )ν + µ −1 f (ξ )d ξ =
Γ(ν )Γ( µ ) ∫
(t − ξ )ν + µ −1 f (ξ )d ξ
0 0 0

t
1
=
Γ(ν + µ ) ∫
(t − ξ )ν + µ −1 f (ξ )d ξ = D − (ν + µ ) f (t ) (2.30)
0

This proves the part (i) of the theorem.

t
1
∫
−ν
To prove part (ii) we let h (t ) = D f (t ) = (t − ξ )ν −1 f (ξ )d ξ .
Γ(ν )
0

Applying D − µ we get
t
1
∫
−µ
D (h(t )) = (t − x) µ −1 h( x)dx
Γ( µ )
0
t x
1
∫ ∫
−µ −ν
D (D f (t )) = (t − x) µ −1 ( ( x − ξ )ν −1 f (ξ ) dξ )dx
Γ(ν )Γ( µ )
0 0

t x
1
∫∫ (t − x)
µ −1
= ( x − ξ )ν −1 f (ξ )dξ dx (2.31)
Γ(ν )Γ( µ )
00

Since the region of integration is the same region shown in Fig. (1) we get upon
reversing the order of integration
t t
1
∫∫ (t − x)
−µ −ν µ −1
D (D f (t )) = ( x − ξ )ν −1 f (ξ )dxd ξ
Γ(ν )Γ( µ )
0ξ

The substitution (2.29) yields

t 1
1
∫∫
−µ −ν
D (D f (t )) = ( uν −1 (1 − u ) µ −1 (t − ξ )ν + µ −1 du ) f (ξ )d ξ (2.32)
Γ(ν )Γ( µ )
0 0

Since B ( µ ,ν ) = B (ν , µ ) we obtain

D − µ ( D −ν f (t )) = D − ( µ +ν ) f (t ) = D − (ν + µ ) f (t ) (2.33)
Equations (2.30) and (2.33) leads to

D − µ ( D −ν f (t )) = D −ν ( D − µ f (t )) = D − ( µ +ν ) f (t ) = D − (ν + µ ) f (t ) (2.34)
This proves the Theorem.
Theorem (1) can be also written in the form

J ν J µ f (t ) = J µ J ν f (t ) = J ν + µ f (t ) (2.35)
 12

Chapter 3
Semi- integrals and Semi-derivatives
The most useful application of derivatives and integrals of noninteger order involve
the semiderivatives and semi-integrals (Oldham and Spanier 2002).
1
Substituting ν = in the definition of fractional integral given by Eq. (2.1a)
2
leads to the following expression of semi integral:
t t
−1 1
f (x )dx 1 f (t − x )dx
D 2 f (t ) =
π 0 t −x ∫ =
π ∫ x
(3.1)
0

Rewriting Eq. (2.11)

t
d d 1
dt Γ(1 −ν ) ∫0
D f (t ) = D − (1−ν ) f (t ) = [
ν
(t − x )−ν f (x )dx ], 0 < ν ≤ 1 , and
dt

1
substituting ν = we obtain the semi- derivative in the form:
2
t t
1
⎡ −1 ⎤ 1 d f (x )dx 1 d f (t − x )dx
D f (t ) = D ⎢ D 2 f (t ) ⎥ =
2
⎣ ⎦ π dt ∫ t −x
=
π dt ∫ x
(3.2)
0 0

Now, the semi-integrals and the semi-derivatives of some elementary functions are
deduced
1) The constant function : Using Equations (3.1) and (3.2) one obtains:
t
−1 1 Cdx t
D 2C =
π 0 x∫ = 2C
π
(3.3a)

1 t C
D 2C = D (2C )= (3.3b)
π πt
The following example illustrate the property

Example (3.1) : Verify that D 1/ 2 (D −1/ 2C ) = D 0C = C ,

Solution: Substituting x = t sin 2 u , we obtain:

t
2C d x dx 2C d t π
∫
−1/ 2
D 1/ 2
(D C)= = ( ) =C
π dt 0
t −x π dt 2
 13

Now, using the definition (3.1) and Eq. (3.3a) and the substitution x = t sin 2 u , we
get
π
t 2
−1 −1 −1 t 1 t dx 4Ct
D 2 (D C ) = D
2 2 (2C
π
)=
π ∫ (2C )
π t −x
=
π 0 ∫
sin 2 udu = Ct
0

t
2C d x dx 2C d t π
∫
−1/ 2
D 1/ 2
(D C)= = ( ) =C
π dt 0
t −x π dt 2
t
−1 −1
∫
Note that: Cdx = Ct = D −1C = D 2 (D 2C ) .
0

−1
Example (3.2) : Evaluate D 2 t
Solution: Using the same substitution and the definition we get

t 3 π 3
−1 1 xdx 2t 2 2 4t 2
(t ) = ∫ = ∫ sin u du =
2 3
D
π 0
t −x π 0
3 π
1 −1
Let us verify the relation: D 2 (D 2 t ) = t ,(we use the substitution x = t sin 2 u )
3 3
1 −14t 2 − 1 4t 2 1
D ( D 2t ) = D (
2 ) = D [D 2 ( )] 2
3 π 3 π
3 π
t 2
1 d 4x dx 8 d 2 2
= ( ) = ∫
π dt 0 3 π t − x 3π dt
(t sin 4 udu ) = t ∫
0

We have used the Wallis formulas

π 2 π 2 ⎧1 m ∈ ` is odd
(m − 1)!! ⎪
∫ sin xdx = ∫ cos xdx = p where p = ⎨ π
m m

0 0
m !! ⎪⎩ 2 m ∈ ` is even

2) : The power function f (t ) = t n , n ∈ `

1
Substituting ν = , λ = n in Eq. (2.4) ) we get:
2
n+1
−1 t 2 Γ (n + 1) 2 (2n )!! n + 12
D 2 (t ) =
n
= t
Γ (n + 32 ) π (2n + 1)!!
Using the following properties of double factorial:
 14

(2n + 1)! (2n + 1)!

(2n)!! = 2n n ! and (2n + 1)!! = = we get
(2n )!! 2n n !
n+1
−1 (n !) 2 (4t ) 2
D 2 (t n ) = (3.4a)
(2n + 1)! π
1
Substituting µ = , m = 1, λ = n in Eq. (2.15) we get:
2
n −1
1 t 2 Γ (n + 1) (2n )!! (2n + 1) n − 12
D (t n ) =
2 = t which yields
Γ ( n + 12 ) π (2n + 1)!!
1 (n !) 2 (4t ) n
D (t ) =
2 n
(3.4b)
(2n )! π t
It should be noted that Equations (3.4) are valid for n = 0

3): the exponential function f (t ) = e αt , α > 0

e α (t −x ) e αt e −α x dx
t t
−1 1
∫ ∫
αt
D 2 (e ) = dx = ,
π 0 x π 0
x

For α > 0, the substitution α x = u 2 leads to:

αt
−1 2(e αt ) e αt
∫e
αt −u 2
D 2 (e ) = du = erf ( αt ) (3.5)
α π 0
α
t
2
∫
2
Where erf (t ) = e − x dx is the error function.
π 0

Setting α = ln a where a > 0, a ≠ 1 in Eq. (3.5) we get

−1 at
D 2 (at ) = erf ( t ln a ) (3.6)
π ln a
The semi- derivative of the exponential function can be obtained from (3.5) using the
definition (3.2):
1 1
D 2 (eα t ) = α eα t erf ( α t ) + (3.7)
πt
Where used the Leibniz’s formula
v (t )
d dv du
dt ∫ f (x )dx =f (v (t )).
dt
− f (u (t )).
dt
(3.8)
u (t )
 15

−1 1 1
From Eqs. (3.3b) , (3.5) and (3.7) we get: D 2 (e αt ) = D 2 (e αt − 1) ,
α
4): The functions sint and cost :
t
−1 1 sin(t − x )dx 2
π ∫
D 2 (sin t ) = = [sin t C ( t ) − cos t S ( t )] (3.9)
0
x π
−1 2
D 2 (cos t ) = [cos t C ( t ) + sin t S ( t )] (3.10)
π
Where C (t ) and S (t ) are the Fresnel integrals defined by the relations
t t

∫
C (t ) = cos( y )dy , ∫
S (t ) = sin( y 2 )dy , and therefore, we have
2

0 0

t t
C( t)= ∫ cos( y S( t)= ∫ sin( y
2 2
)dy , )dy , (3.11)
0 0

The semi -derivatives can be obtained from the definition (3.2) using Leibniz’ formula
(3.8) as:
1 −1 d 2
D 2 (sin t ) = D {D 2 (sin t )} = { [sin t C ( t ) − cos t S ( t )]}
dt π
2 1
= [cos t C ( t ) + sin t S ( t )] + [sin t cos t − cos t sin t ] . Therefore,
π πt
1 2
D 2 (sin t ) = [cos t C ( t ) + sin t S ( t )] (3.12)
π
Taking into consideration Eq. (3.10) we get
1 −1
D 2 (sin t ) = D 2 (cos t ) (3.13)
Similarly, we get
1 1 −1
D 2 (cos t ) = −D 2 (sin t ) (3.14)
πt
Taking into consideration Eq. (3.3a) the preceding equation can be written in the
form:
−1 1
D 2 (sin t ) = D 2 (1 − cos t ) (3.15)

AMNA ABDULSALAM YOUNIS AL-AMRY

 16

Chapter 4
Examples
(Semi- Integrals and Semi-Derivatives of Some Algebraic Functions)
f (x ) f (x )dx 11
t t
f (t − x )dx D
1/2
f ( x)
∫ ∫
−1/ 2
D f (x ) = =
π 0 t −x π 0
x = D[ D
−1/2
f ( x )]

1 x zero
1 1 dy
∫
( ) y = xz
x ⎯⎯⎯⎯
dy = xdz
→
π 0
y x −y
1
1
π ∫
−1/ 2 −1/ 2
= (1 − z ) z dz
0
1 1 1
= B( , )= π
π 2 2

π
2 x
( x) 1 x − y dy
∫
y = xz
⎯⎯⎯⎯
dy = xdz
→
π 0
y 2

x 3 1 x π x π
= B( , )= ( )=
π 2 2 π 2 2
3
1 1
x
dy y =(1+ x )sin θ ,
2 1
1+ x π
∫ 1+ x − y y
⎯⎯⎯⎯⎯⎯⎯⎯
dy = 2(1+ x )sin θ cos θ d θ
→
(1 + x ) π x
0
−1
tan x
2 (1 + x ) sin θ cos θ d θ 2
∫
−1
= = tan ( x )
π (1 + x ) sin θ cos θ π
0

4
1+ x x
1 + x − y dy 1 tan −1 ( x )
1 y =(1+ x )sin 2 θ , +
π
∫ y
⎯⎯⎯⎯⎯⎯⎯⎯
dy = 2(1+ x )sin θ cosθ d θ
→ πx π
0

x (1 + x ) −1
= + tan ( x )
π π
5 1
1
x
dy y =(1+ x )sin θ ,2 1− x
(1 + x )3/ 2
π
∫ (1 + x − y )
3
y
⎯⎯⎯⎯⎯⎯⎯⎯
dy = 2(1+ x )sin θ cosθ d θ
→
(1 + x ) 2 π x
0

2 x
=
(1 + x ) π
 17

6
1 1
x
dy y =(1− x )sinh θ 2 1
1− x π
∫ 1− x + y y
⎯⎯⎯⎯⎯⎯⎯⎯⎯
dy = 2(1− x )sinh θ cosh θ d θ
→
(1 − x ) π x
0

2
= tanh −1 ( x )
π
7
1− x x 1
1 1 − x + y dy [
∫
y =(1− x )sinh 2 θ πx
⎯⎯⎯⎯⎯⎯⎯⎯⎯
dy = 2(1− x )sinh θ cosh θ d θ
→
π y tanh −1( x )
0 − ]
π
x (1 − x )
+ tanh −1 ( x )
π π

8 1
1
x
dy 1+ x
∫
y =(1− x )sinh 2 θ
(1 − x )3/ 2 ⎯⎯⎯⎯⎯⎯⎯⎯⎯
dy = 2(1− x )sinh θ cosh θ d θ
→ (1 − x ) 2 π x
π 0 (1 − x + y )3 y

2 x
=
(1 − x ) π
9
1 x 1− x
1 dy y =(1− x ) tan 2 θ
(
1− x
)
π
∫ (1 − x + y ) y
⎯⎯⎯⎯⎯⎯⎯⎯→
dy = 2(1− x ) tan θ sec θ d θ
2
(1 − x )3/ 2 π x
x sin −1( x )
0 +
(1 − x )3/ 2 π x
2sin −1 ( x )
=
π (1 − x )
10
1 x
1+ x
1 dy y =(1+ x ) tanh θ 2
(
1+ x
)
∫
π (1 + x − y ) y
⎯⎯⎯⎯⎯⎯⎯⎯⎯ →
dy = 2(1+ x ) tanh θ sech θ d θ
2
(1 + x )
3/ 2
πx
0
−1
−1 x sinh ( x)
2sinh ( x ) −
= (1 + x )
3/ 2
πx
π (1 + x )

π
11 x
x 1 x − y dy y = x sin t 2

(1 + x ) π
∫ (1 + x − y ) y
⎯⎯⎯⎯⎯⎯→
dy = 2 x sin t cos tdt 2(1 + x )3/ 2
0

π
= π −
1+ x

π
12 1 x
1 dy 2
y = x sin t
x (1 + x )
π
∫ (1 + x − y ) x −y y
⎯⎯⎯⎯⎯⎯→
dy = 2 x sin t cos tdt
−
2(1 + x )3/ 2
0
 18

π
=
1+ x

π
13 x
x 1 x − y dy
∫ (1 − x + y )
2
y = x sin t
⎯⎯⎯⎯⎯→
(1 − x ) π y dy = 2 x sin t cos tdt 2(1 − x )3/ 2
0

π
=− π +
1− x

π
14 1 x
1 dy
∫ (1 − x + y )
2
y = x sin t
x (1 − x )
⎯⎯⎯⎯⎯→
π x −y y dy = 2 x sin t cos tdt 2(1 − x )3/ 2
0

π
=
1− x

AMNA ABDULSALAM YOUNIS AL-AMRY

 19

Chapter 5
Applications
The Tautochrone Curve
A tautochrone curve or isochrone curve is the curve for which the time taken by an object
sliding without friction in uniform gravity to its lowest point independent of its starting
point. Abel was interested in the tautochrone problem; that is, determining a curve in the
( x , y ) plane such that the time required for a particle to slide down the curve to its
lowest point is independent of its initial placement on the curve.
Firstly, we shall consider the time required for descent as a function of initial height.
Let us fix the lowest point of the curve at the origin and position the curve in the positive
quadrant of the plane, denoting by M ( x 0 , y 0 ) the initial point and P ( x , y ) any point

on the curve between O (0,0) and ( x 0 , y 0 ) . We denote the curve by S = S ( y ) ( at

P )and then, S (0) = 0 . we also have the following conditions: at point M we have
t = 0, y = y 0 , v = 0 ( the initial velocity) and at point O we have
t = T ( y 0 ), y = 0, S = 0 . Assuming no frictional losses we may apply the
conservation of energy law (the sum of kinetic energy and potential energy is constant):

1 2 1
EM = mvM + mgy0 = E p = mvP2 + mgy . Since vM = 0 we get:
2 2
1 dS 2
m ( ) = mg ( y 0 − y ) (5.1)
2 dt
 20

dS
Since the distance S is decreasing as the time increases we have < 0.
dt
dS dS dy
Therefore, Eq. (5.1) leads to =( ) = − 2 g ( y 0 − y ) . Thus, separating the
dt dy dt
variables one obtains
dS
()dy
dy
2 g dt = − (5.2)
(y 0 − y )
Integrating from t = 0, to t = T which corresponds from y = y 0 to y = 0 we get
T 0
(dS / dy )
∫
2 g dt = − ∫ (y 0 − y )
dy . Therefore,
0 y0

y0
(dS / dy )
2g T ( y 0 ) = ∫ (y 0 − y )
dy (5.3)
0

Equation (5.3) can be written in the form

y
(dS / dz )
2g T ( y ) = ∫ (y − z )
dz (5.4)
0

Equation (5.4) is the Abel’s integral equation. We shall use the fractional calculus to
obtain the solution of this equation (Samko et all 1993). Recalling the definition of the
y
1 f (z )
π ∫
−1/2
semi – integral: D f (y ) = dz we get
0
y −z

2 gT ( y ) = π D −1/2 ((dS / dy )) = π D −1/2 (DS ( y )) = π D 1/2 (S ( y )) . (5.5)

Therefore, we get

2g
S (y ) = D −1/2{T ( y )} (5.6)
π
Thus, we have obtained the solution of Abel’s integral equation in terms of the semi-
integral of T (y ) .
 21

Now, to get the required curve we must set T ( y ) = T = Const . and use

y
D −1/2{C } = 2C and Eq. (5.6) becomes
π
y 2g
S ( y ) = 2T = 2 ay , a = 2 gT 2 / π 2 (5.7)
π π
Differentiating both sides with respect to y one obtains

dS ( y ) dx a
= 1 + ( )2 = , from which
dy dy y
dx 2 a dx a
1+ ( ) = → ( )2 = − 1 (5.8)
dy y dy y
Let

y = a sin 2 (θ / 2) . (5.9)
Then from (5.8) we get
dx dy
= cot(θ / 2) → = tan(θ / 2) (5.10)
dy dx
dy dθ
From (5.9) and (5.10) we get = a sin(θ / 2) cos(θ / 2) = tan(θ / 2) which leads to
dx dx
dx dx a
= a cos 2 (θ / 2) , and we get = (1 + cos θ )
dθ dθ 2
a
From the preceding equation we obtain x = (θ + sin θ ) + C . The curve is passing
2
through the origin so, x = 0 at y = 0 and therefore, C = 0 . Thus we obtain the
parametric equations of the tautochrone curve (a cycloid)
a a
x = (θ + sin θ ) y = (1 − cos θ ) (5.11)
2 2
AMNA ABDULSALAM YOUNIS AL-AMRY
 22

References

[1] L. Debnath, (2004), A brief historical introduction to fractional calculus, INT. J.

MATH. EDUC. SCI., TECHNOL., vol. 35, No 4, 487-501.

[2] Gorenflo, R. and F. Mainardi (1997) , Fractional calculus : integral and differential

equations of fractional order, in Fractals and Fractional Calculus in Continuum

Mechanics (Ed. A. Carpinteri and F. Mainardi), Springer Verlag , Wien.

[3] Oldham Keith B. and Spanier Jerome.( 2002), The fractional Calculus, 2nd Ed,

Dover Publications , INC, Mineola, New York.

[4] I. Padlubny (1999), Fractional Differential Equations, Academic Press, N.Y.

[5] Samko S. G., Kilbas, A. A. and O. I. Marichev (1993), Fractional Integrals and

Derivatives, Theory and Applications, Gordon and Breach, Amsterdam.

Learning Outcomes

1. I knew about modern branches in Mathematics and their

applications.
2. I reviewed many subjects such as Laplace Transforms, Gamma
and Beta functions, double integrals and techniques of
integration. I learned how to use many mathematical skills.
3. I learned new subjects like the error function, Fresnel integrals
Dawson’s function, incomplete Gamma function .and Et
function which were not taught in the undergraduate program.
4. This Project helped me to gain much knowledge in physics and
geometry.
5. Helped me in gaining critical thinking.

AMNA ABDULSALAM YOUNIS AL-AMRY