An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative

An Inverse Problem for Diffusion Equation with Two parameters

Fractional Derivative

Presented by: Aniqa Iqbal

Supervised by: Dr. Salman Amin Malik

COMSATS University Islamabad

June 26, 2024

1 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative

Outline

• What is a Fractional calculus?

• Definitions of the some Fractional Operators

• Mittag Leffler Function

• Solution of an Inverse problem

• Uniqueness of the solution

• Conclusion

• Bibliography

2 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative

A Bit of History

What is the Fractional calculus?

The Fractional Calculus (FC) is a generalization of classical calculus concerned with operations
of integration and differentiation of arbitrary(non-integer) order.

• Beginning of Fractional calculus (1695).

• Contributions of different mathematician in Fractional Calculus.

• First conference was held in 1974.

• First Mathematical journal "Fractional Calculus and Applied Analysis" created in 1998.

3 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Definitions of the Fractional Operators

Definitions of the Fractional Operators

Riemann-Liouville Fractional Integral

Let f ∈ L1 (a, b) and order α > 0, then Riemann-Liouville fractional integral is defined as

1 Zt
Iaα+ ,t f (t) = (t − τ )α−1 f (τ )d(τ ), t ∈ (a, b]. (1)
Γ(α) a

Riemann-Liouville Fractional Derivative

RL
Let f (t) ∈ L1 (a, b), Riemann-Liouville fractional derivative Daα+ ,t of order 0 < α < 1 is
defined as

RL d 1−α 1 d Z t f (τ )
Daα+ ,t (f (t)) = Ia+ ,t f (t) = dτ, t ∈ (a, b]. (2)
dt Γ(1 − α) dt a (t − τ )α

4 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Definitions of the Fractional Operators

Definitions of the some Fractional Operators

Caputo fractional derivative

Let f (t) ∈ AC [a, b], then Caputo fractional derivative c Daα+ ,t of order 0 < α < 1 is defined as.

c α d 1 Z t
f ′ (τ )
Da+,t (f (t)) = Ia1−α
+ ,t
f (t) = dτ, t ∈ (a, b]. (3)
dt Γ(1 − α) a (t − τ )α

Relation between Riemann-Liouville and Caputo Fractional derivatives

Let n − 1 < α < n, n ∈ N. Then the relation between Riemann-Liouville fractional derivative
and Caputo fractional derivative is given by the equation

n−1
C f k (a)
Daα+ ,t f RL
Daα+ ,t f (t − a)k−α .
X
(t) = (t) − (4)
k=0 Γ(k − α + 1)
C
provided that Daα+ ,t and RL
Daα+ ,t exists.

5 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Definitions of the Fractional Operators

Definitions of the some Fractional Operators

Hilfer Fractional Derivative

Hilfer Fractional derivative is a generalization of a Riemann-Liouville and Caputo fractional

derivatives. The left-sided Hilfer fractional derivative of order 0 < α < 1 , β ∈ [0, 1], and
(1−β)(1−α)
Ia+ ,t f (t) ∈ AC [a, b] is defined as

!
H β(1−α) (1−β)(1−α)
Daα,β
+ ,t
f (t) = Ia+ ,t DIa+ ,t f (t), (5)
!
= Iaγ−α
+ ,t
DIa1−γ
+ ,t
f (t), t ∈ (a, b]. (6)

d
where D := dt and γ = α − αβ + β.

6 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Definitions of the Fractional Operators

Hilfer Fractional Derivative

Remark 1

• When β = 0, (5) reduces to the Riemann–Liouville fractional derivative 0 < α < 1

H d 1−α
Daα+ ,t f (t) = dt Ia+ ,t f (t) = RL
Daα+ ,t f (t).
• When β = 1 , (5) reduces to the Caputo fractional derivative of order 0 < α < 1
H
Daα,1
+ ,t
f (t) = Ia1−α d
+ ,t dt
f (t) = c
Daα+ ,t f (t).

Laplace Transform of Hilfer fractional Derivative

The Laplace transform of Hilfer fractional Derivative of order n − 1 < α < n and 0 ≤ β ≤ 1 is
defined as
n−1
d n−k−1 (1−β)(n−α)
L( H D0α,β ) = s α F (s) − s k−β(n−α)
X
+ ,t
I f (0). (7)
k=0 dt n−k−1 0+ ,t

7 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Mittag Leffler Function

Mittag Leffler Function (MLF)

Definition of MLF

• The Mittag-Leffler function of two parameters is defined as for Re(β) > 0, Re(α) > 0.
∞
X zk
Eα,β (z) = α, β, z ∈ C.
k=0 Γ(αk + β)

• For β = 1, Eα,β (z) reduces to the classical Mittag-Leffler function, i.e.,

∞
X zk
Eα,1 (z) = Eα (z) = ,
k=0 Γ(αk + 1)

Laplace transform of the Mittag Leffler function

The Laplace transform of the Mittag Leffler function is given as

s α−β
L{t β−1 Eα,β (λt α )} = , |λs −α | < 1. (8)
sα − λ
8 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Statement of an Inverse Problem

Statement of an Inverse Problem

• We use the Hilfer fractional derivative in the below inverse problem of determining the
distribution u(x,t) and source term f(x) of a generalized time- fractional diffusion equation
[1].

D0α,γ
+ ,t
u(x , t) − uxx (x , t) = f (x ), 0 < x < 1, 0 < t < T , 0 < α ≤ γ ≤ 1, (9)

I01−γ
+ ,t
u(x , t)|t=0 = g(x ), u(x , T ) = h(x ), 0<x <1 (10)

u(1, t) = 0, ux (0, t) = ux (1, t), 0<t≤T (11)

where g(x) and h(x) are the initial and the final conditions, respectively.

9 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Formal solution and source term

Spectral Problems

• The spectral problem for the boundary value problem (9)-(11) given by

X ′′ (x ) = −λX (x ), x ∈ (0, 1). (12)

X (1) = 0, X ′ (0) = X ′ (1).

is non-self-adjoint and the adjoint problem of the spectral problem is

Y ′′ (x ) = −λY (x ), x ∈ (0, 1). (13)

Y ′ (0) = 0, Y (1) = Y (0).

10 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Formal solution and source term

Bi-orthogonal system of functions

The set of eigenfunctions corresponding to eigenvalues λ0 = 0 and λn = 2πn form a bi-orthogonal

system of functions [2] under the following one-to-one correspondences between

ϕ0 (x ) = 2(1 − x ), ϕ1n (x ) = 4(1 − x ) cos λn x , ϕ2n (x ) = 4 sin λn x . (14)

and
ψ0 (x ) = 1 ψ1n (x ) = cos λn x , ψ2n (x ) = x sin λn x . (15)

We seek a solution and source function of (9)-(11) in the form

∞
X
u(x , t) = u0 (t)ϕ0 (x ) + ukn (t)ϕkn (x ), (16)
n=1
k=1,2
∞
X
f (x ) = f0 ϕ0 (x ) + fkn ϕkn (x ). (17)
n=1
k=1,2
where u0 (t), ukn (t), f0 and fkn for n ∈ N are unknowns to be determined.
11 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Formal solution and source term

continued...

Taking the fractional derivative under the integral,we have the following linear fractional differ-
ential equations
D0α,γ u (t) = f0 ,
+ ,t 0
(18)

D0α,γ u (t) + λ2n u1n (t) = f1n ,

+ ,t 1n
(19)

D0α,γ u (t) + λ2n u2n (t) − 2λn u1n (t) = f2n .

+ ,t 2n
(20)

By using the Laplace transform techniques of Hilfer fractional derivative, we get

f0 g0 γ−1
u0 (t) = tα + t . (21)
Γ(1 + α) Γ(γ)
u1n (t) = f1n t α Eα,α+1 (−λ2n t α ) + g1n t γ−1 Eα,γ (−λ2n t α ). (22)

12 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Formal solution and source term

continued...

"
α
u2n (t) = f2n t Eα,α+1 (−λ2n t α ) + g2n t γ−1
Eα,γ (−λ2n t α ) + 2λn f1n t 2α Eα,α+1
2
(−λ2n t α )
#
+ g1n t α+γ−1 Eα,γ
2
(−λ2n t α ) . (23)

we obtain the unknowns f0 and fkn k=1,2, n=1,2,. . . as taking (21)-(23). This implies
" #
Γ(1 + α) g0 γ−1
f0 = h0 − T (24)
Tα Γ(γ)
h1n − g1n E2n (T )
f1n = , (25)
E1n (T )
h2n − g2n E2n (T ) − 2λn [f1n E3n (t) + g1n E4n (t)]
f2n = . (26)
E1n (T )

13 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Formal solution and source term

Continued...

The solution of the IP is given by the series (16) and (17), where the unknowns u0 (t), u1n (t), u2n (t)
are calculated from (21)-(23), while f0 , f1n , f2n are obtained from (24)-(26).
where
E1n (T ) = T α Eα,α+1 (−λ2n T α ), E2n (T ) = T γ−1 Eα,γ (−λ2n T 2α ),

E3n (T ) = T 2α Eα,α+1
2
(−λ2n T α+γ−1 ), E4n (T ) = T α+γ−1 Eα,γ
2
(−λ2n T α ).

14 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Formal solution and source term
Uniqueness of solution

Uniqueness of solution

Theorem 1:

Let {u1 (x , t), f1 (x )} and {u2 (x , t), f2 (x )} be two regular solution sets of the inverse source
problem (10)-(11). Then u1 (x , t) = u2 (x , t) implies f1 (x ) = f2 (x ).

Proof: Consider the functions

ukn(i) (t) = ⟨ui (x , t), ψkn (x )⟩, fkn(i) = ⟨fi (x ), ψkn (x )⟩. (27)

for i=1,2 . Now we take the Hilfer fractional derivative of equation (27) when k=1;

D0α,γ u
+ ,t 1n(i)
(t) = ⟨D0α,γ u (x , t), ψ1n (x )⟩.
+ ,t i
(28)

By virtue of (9), the following system of fractional differential equations is obtained

D0α,γ u
+ ,t 1n(1)
(t) + λ2n u1n(1) (t) = f1n(1) ,

D0α,γ u
+ ,t 1n(2)
(t) + λ2n u1n(2) (t) = f1n(2) .
15 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Formal solution and source term
Uniqueness of solution

continued...

By using the Laplace transform technique the solutions are given by

u1n(1) (t) = f1n(1) E1n (t) + g1n(1) E2n (t), (29)

u1n(2) (t) = f1n(2) E1n (t) + g1n(2) E2n (t). (30)

Using the assumption u1 (x , t) = u2 (x , t) =⇒ u1n(1) (t) = u1n(2) (t), we get

(f1n(1) − f1n(2) )E1n (t) = 0. (31)

By taking Laplace transform, we get

f1n(1) − f1n(2) f1n(1) − f1n(2)
= 0, Re(s) > 0 =⇒ = 0,
s(s α + λ2n ) (η + λ2n )

By using Caushy Integral theorem, integrating along the different disks, we have f¯1n (x ) = 0
Similarly f¯0 (x ) = 0, f¯2n (x ) = 0 for ∀n ∈ N. Hence, we have f1 (x ) = f2 (x )
16 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Conclusion

Conclusion

• We introduced the fractional order operators of order α.

• We determined the solution and source term of a diffusion equation involved Hilfer
Fractional derivative by using Eigenfunctions expansion method.
• We established a bi-orthogonal basis of L2 space corresponding to a nonself-adjoint
boundary value problem.
• We proved the Uniqueness of the inverse probelm by using the concept of Caushy Integral
Theorem.

17 / 18
An Inverse Problem for Diffusion Equation with Two parameters Fractional Derivative
Bibliography

Bibliography

Khaled M Furati, Olaniyi S Iyiola, and Mokhtar Kirane.

An inverse problem for a generalized fractional diffusion.
Applied Mathematics and Computation, 249:24–31, 2014.
VA Il’in.
How to express basis conditions and conditions for the equiconvergence with trigonometric
series of expansions related to non-self-adjoint differential operators.
Computers & Mathematics with Applications, 34(5-6):641–647, 1997.

18 / 18