P ERMANENT OF C IRCULANT M ATRICES

T O FIND A GENERAL METHOD FOR COMPUTING THE

PERMANENT OF CIRCULANT MATRICES
 P ERMANENT OF C IRCULANT M ATRICES
T O FIND A GENERAL METHOD FOR COMPUTING THE
PERMANENT OF CIRCULANT MATRICES

By

Adeel B HATTI
Department of Mathematics & Social Science
Sukkur IBA University

Bachelor Thesis
A thesis submitted in partial fulfillment of the requirements for the
degree of Bachelor of Science in Mathematics

Defended publicly on 20 January 2020

This thesis has been submitted to the Department of Mathematics & Social Science as
partial fulfillment of the requirement for the award of Bachelor’s Degree in Mathematics
to

Name : Adeel B HATTI

Reg. No. : 143-21-0006

Thesis has been approved by the supervisor

Dr. Syed Hyder Ali Muttaqi Shah
Department of Mathematics & Social Science
Sukkur IBA University

and the defense/promotion committee:

Dr. Javed Hussain Brohi
HoD Mathematics
Sukkur IBA Univeristy

External member:
Dr. Israr Memon
Department of Mathematics
Shah Abdul Latif University, Khairpur

Internal member:
Dr. Irshad Ahmed
Department of Mathematics & Social Science
Sukkur IBA Univeristy

Project Coordinator:
Dr. Fazeel Anwar
Department of Mathematics & Social Science
Sukkur IBA University

Copyright © 2020 by Adeel Bhatti

ISBN 000-00-0000-000-0
vi

An electronic version of this dissertation is available at

http://repository.suk-iba.edu.pk/.
 D ECLARATION

I Adeel B HATTI hereby declare that this thesis has been composed solely by myself, dur-
ing the period of my BS study. I am aware of and understand the university’s & HEC
policy on plagiarism and I certify that this thesis is my own work, except where indicated
by reference, and the work presented in it has not been submitted in support of another
degree or qualification from this or any other university or institute of learning. If a vio-
lation of the university & HEC policy on research has occurred in this thesis then I shall
be liable for punishable action under the rules of the university & HEC.

, 20 January 2020

Adeel B HATTI
Reg. No. 143-21-0006
 C ERTIFICATE

It is certified that Adeel B HATTI (Reg. No. 143-21-0006) has carried out all the work
related to this thesis under my supervision at the Department of Mathematics & Social
Science, Sukkur IBA University. The work fulfills the requirement for the award of a BS
degree in Mathematics.

, 20 January 2020

Dr. Syed Hyder Ali Muttaqi Shah

Department of Mathematics & Social Science
Sukkur IBA University
Sukkur, Pakistan
 " Pure mathematics is,
in its way, the poetry of logical ideas."
Albert Einstein
 C ONTENTS

Summary xv
Preface xvii
1 Preliminaries 1
1.1 Fibonacci Sequence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Lucas Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Multiplicity Function m t (ω): . . . . . . . . . . . . . . . . . . . . . . . . 2
1.5 Function µ(ω): . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.6 Polynomials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.7 Matrix. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.7.1 Tridiagonal Matrices . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.7.2 Circulant Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.8 Submatrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.9 Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.9.1 Symmetric Groups . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.10 Character χ:. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.11 Generalized Matrix Function . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Permanent and Its Elementary Properties 7
2.1 Permanent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 3×3 Example for calculating the permanent . . . . . . . . . . . . . 7
2.2 Elementary Theorems for Permanents . . . . . . . . . . . . . . . . . . . 8
3 Some Special Tridiagonal Matrices and its Permanents 13
3.1 Tridiagonal Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Matrix Contraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4 Conclusion 19
A Appendix 21
A.1 Ordinary Differential Equation . . . . . . . . . . . . . . . . . . . . . . . 21
A.1.1 Classification of ODEs . . . . . . . . . . . . . . . . . . . . . . . . 21
A.2 Autonomous and Non-Autonomous ODEs . . . . . . . . . . . . . . . . . 22
A.3 Dynamical System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
B Appendix 25

xiii
 S UMMARY
The Thesis mainly presents the idea of the permanent and how the world has progressed
through permanenents throughout the thesis our main objective was to find the perma-
nent of circulant matrices which hasn’t been possible but we’ll try and get to it as close
as possible

xv
 P REFACE
the purpose of writing this thesis was to get the idea that how a permanent works all the
motivation for writing it comes from my supervisor Dr. Fazeel Anwar he motivated me
to work on an open problem that I worked on .

Adeel Bhatti
20 January 2020

xvii
 1
P RELIMINARIES
" Matrices act.
They don’t just sit there."
- Gilbert Strang

" Science is a way of thinking much more than

it is a body of knowledge."
- Carl Sagan

This chapter provides the essential definitions and theorems needed to follow the main
ideas discussed later. It begins with the basics, such as Fibonacci and Lucas numbers, and
then introduces matrices, polynomials, and some special types of matrices. We also define
a particular set of increasing sequences that will play an important role in the next chap-
ters. If you’re already comfortable with these topics, you may choose to skim this chapter
and move ahead.

Parts of this chapter have been published in Annalen der Physik 324, 289 (1906) [? ].

1
 2 1. P RELIMINARIES

1 1.1. F IBONACCI S EQUENCE

The Fibonacci sequence, a n is defined as:
a n+1 = a n + a n−1 f or n ≥ 1 (1.1)
where a 0 = 0, a 1 = 1.

1.2. LUCAS S EQUENCE

The Lucas sequence, b n , is defined as:
b n+1 = b n + b n−1 f or n ≥ 1 (1.2)
where b 0 = 2, b 1 = 1.

1.3. S EQUENCES
A sequence is an arrangement of numbers in a particular order

We begin by introducing key sets and functions that will be used throughout this work.
• Set Γr,n (All Integer Sequences): The set Γr,n consists of all sequences ω = (ω1 , . . . , ωr )
of integers such that each element lies between 1 and n, inclusive. Formally,
Γr,n = (ω1 , . . . , ωr ) ∈ Zr | 1 ≤ ωi ≤ n for all i = 1, . . . , r .
© ª

• Set G r,n (Nondecreasing Sequences): The subset G r,n of Γr,n contains all nonde-
creasing sequences:
G r,n = (ω1 , . . . , ωr ) ∈ Γr,n | ω1 ≤ ω2 ≤ · · · ≤ ωr .
© ª

• Set Q r,n (Strictly Increasing Sequences): The subset Q r,n of Γr,n contains all strictly
increasing sequences:
Q r,n = (ω1 , . . . , ωr ) ∈ Γr,n | ω1 < ω2 < · · · < ωr .
© ª

1.4. M ULTIPLICITY F UNCTION m t (ω):

For any sequence ω ∈ Γr,n and any integer t ∈ {1, 2, . . . , n}, the function m t (ω) denotes the
number of times the integer t appears in the sequence ω. That is,
m t (ω) = |{i ∈ {1, . . . , r } | ωi = t }| .
Examples: Let ω = (2, 4, 4, 4, 5, 5) ∈ Γ6,5 . Then:
m 1 (ω) = 0,
m 2 (ω) = 1,
m 3 (ω) = 0,
m 4 (ω) = 3,
m 5 (ω) = 2.
1.5. F UNCTION µ(ω): 3

1.5. F UNCTION µ(ω): 1

Define the function µ(ω) as the product of the factorials of the multiplicities of each
integer t in the sequence ω:
n
µ(ω) =
Y
m t (ω)!.
t =1

Continuing the example: For ω = (2, 4, 4, 4, 5, 5), we have:

µ(ω) = m 1 (ω)! · m 2 (ω)! · m 3 (ω)! · m 4 (ω)! · m 5 (ω)! = 0! · 1! · 0! · 3! · 2! = 1 · 1 · 1 · 6 · 2 = 12.

1.6. P OLYNOMIALS
Let a 0 , a 1 , ..., a k be complex numbers. A complex polynomial of degree n ≥ 0 is a function
of the form:

p(x) = c n · x n + c n−1 · x n−1 + .... + c 1 · x + c 0 c n ̸= 0

If all the coefficients of p are real, then p is a real polynomial.

1.7. M ATRIX
an m × n matrix is a rectangular array:
 
a 11 a 12 ... a 1n
a a 22 ... a 2n 
 21 
A = [a i j ] = 
 .. .. .. ..  (1.3)
 . . . . 

a n1 a n2 ... a nn

where a i j ∈ C
The set of all m × n matrices with complex entries is denoted by M m×n (C)
Following are the special types of matrices that we are interested in:

1.7.1. T RIDIAGONAL M ATRICES

A matrix A = [a i j ] is tridiagonal if a i j = 0 whenever |i − j | ≥ 2
 
b1 c1 0 ··· 0
 .. .. 
a 2

b2 c2 . . 

..
 
.
 
0
 a3 b3 0  
 ...
A = [a i j ] =  .. .. .. (1.4)

 . . . c n−1 
0 ··· 0 an bn 
 
 . .. .. .. .. 
 . 
 . . . . . 
0 ··· 0 0 0 m×n
 4 1. P RELIMINARIES

1.7.2. C IRCULANT M ATRICES

1 Let c = [c 1 , c 2 , c 3 , ..., c n ] of length n, the circulant matrix C is defined as:
 
c1 c2 ... cn
c c1 ... c n−1 
 n 
C = [a i j ] = 
 .. .. .. ..  (1.5)
. . . . 

c2 ... cn c1

1.8. S UBMATRIX
A submatrix of a matrix A is a matrix obtained from A by deleting some rows and/or
columns.
Let A = (a i j ) ∈ M m,n , and let α ∈ G h,m , β ∈ G k,n , where G r,s denotes the set of strictly
increasing sequences of length r from {1, 2, . . . , s}.
Then:

• A[α | β] denotes the h × k submatrix of A. That is, select rows indexed by α and
columns indexed by β.

• If α ∈ Q h,m and β ∈ Q k,n , where Q r,s is the set of all r -element subsets of {1, 2, . . . , s},
then A[α | β] is called a submatrix of A.

• If α = β, we simplify the notation to A[α].

• A(α | β) denotes the (m − h) × (n − k) submatrix obtained by deleting rows indexed

by α and columns indexed by β.

• A(α | −): submatrix obtained by deleting only rows α.

• A(− | β): submatrix obtained by deleting only columns β.

Example:
Let  
1 2 3
A = 4 5 6 , α = {1, 3}, β = {2, 3}.
7 8 9
Then: µ ¶
2 3
A[α | β] = A(α | β) = 1 .
¡ ¢
,
8 9

1.9. G ROUPS
Def: Let G be a set and ∗ : G × G →
− G be a binary operation on G. The pair (G,∗) is called
a group if following properties hold:

• Associativity: ∗ is an associative binary operation, that is,

(a ∗ b) ∗ c = a ∗ (b ∗ c) , ∀a, b, c ∈ G
1.10. C HARACTER χ: 5

• Existence of Identity: There exists an element e ∈ G such that:

1
e ∗ a = a ∗ e = e , ∀a ∈ G

• Existence of Inverse: For all a ∈ G, there exists b ∈ G such that:

a ∗ b = b ∗ a = e , ∀a ∈ G

1.9.1. S YMMETRIC G ROUPS

A symmetric group, denoted as S n , consists of all possible permutations (rearrange-
ments) of a set of n distinct elements.

Note: We usually denote the members of this set as "σ"

1.10. C HARACTER χ:
Def: A character (χ) in this setting is a group homomorphism from a subgroup H of the
symmetric group S m to the multiplicative group of complex numbers C× that is:

χ : H → C× (1.6)

1.11. G ENERALIZED M ATRIX F UNCTION

Let A be a n ×n matrix and Let S n denote the symmetric group of degree n and let H be a
subgroup of S n of order h. Let χ be a character of degree 1 on H , i.e., a nontrivial homo-
morphism of H into the complex numbers. we define the generalized matrix function
d χ by
n
χ(σ)
X Y
d χ (A) = a i ,σ(i ) . (1.7)
σ∈H i =1

Special Cases:
• Determinant: if H = S m and χ(σ) = sg n(σ) (the sign of the permutation), then:
X n
Y
d et (A) = sgn(σ) a i ,σ(i ) (1.8)
σ∈S n i =1
 6 1. P RELIMINARIES

• Permanent: if H = S m and χ(σ) = 1, then:

1
n
X Y
Per (A) = a i ,σ(i ) (1.9)
σ∈S n i =1

We’ll explain the permanent in a very detailed manner that what it is and how it works
 2
P ERMANENT AND I TS E LEMENTARY
P ROPERTIES
This chapter is about what permanent is, how it works, its application, and its most ba-
sic properties. It starts with the most basic definition of permanent. After defining the
permanent, it further explores the properties of the permanent, what things we can do
with the permanent, and what things we cannot do with the permanent. the properties
involving permanents all this chapter covers all the basis that we need for permanent.

2.1. P ERMANENT
The permanent of a matrix is denoted by per m(A) and is calculated by the formula:

n
X Y
Per m(A) = a i ,σ(i )
σ∈S n i =1

Remark: Computing the permanent of an arbitrary n × n matrix is a #p-complete prob-

lem. It is computationally extremely expensive to compute the permanent, as we can see
with each n we have n! terms, and for each, the number of terms will increase drastically,
so as the number of operations and computational cost.

2.1.1. 3×3 E XAMPLE FOR CALCULATING THE PERMANENT

Consider the matrix:
 
a 11 a 12 a 13
A = a 21 a 22 a 23  (2.1)
a 31 a 32 a 33

7
 8 2. P ERMANENT AND I TS E LEMENTARY P ROPERTIES

The symmetric group S 3 consists of the six permutations:

µ ¶ µ ¶ µ ¶
1 2 3 1 2 3 1 2 3
σ1 = σ2 = σ3 =
1 2 3 1 3 2 2 1 3
µ ¶ µ ¶ µ ¶
1 2 3 1 2 3 1 2 3
2 σ4 =
2 3 1
σ5 =
3 1 2
σ6 =
3 2 1

Using the formula:

perm(A) = a 11 a 22 a 33 + a 11 a 23 a 32 + a 12 a 21 a 33 + a 12 a 23 a 31 + a 13 a 21 a 32 + a 13 a 22 a 31
Thus, the permanent of A is:
perm(A)=a 11 a 22 a 33 + a 11 a 23 a 32 + a 12 a 21 a 33 + a 12 a 23 a 31 + a 13 a 21 a 32 + a 13 a 22 a 31

2.2. E LEMENTARY T HEOREMS FOR P ERMANENTS

Theorem 1 (a) The permanent function on m × n matrices, m ≤ n, is a multilinear
function of the rows of each matrix. If m = n, it is also a multilinear function of the
columns.

(b) If A is an m × n matrix, m ≤ n, and P and Q are permutation matrices of orders m

and n, respectively, then
per(P AQ) = per(A).

(c) If A is an n-square matrix, then

per(A T ) = per(A).

Theorem 2 If A is an m × n matrix, 2 ≤ m ≤ n, and α ∈ Q r,m , then

per(A[α | ω]) per(A(α | ω)).

X
per(A) =
ω∈Q r,n

in particular, for any i , 1 ≤ i ≤ m,

n
X
per(A) = a t t · per(A(i | t )).
t =1

if m = n and β ∈ Q r,n , then also

per(A[ω | β]) per(A(ω | β)).

X
per(A) = (1.3)
ω∈Q r,n

and for any j , 1 ≤ j ≤ n.

n
X
per(A) = a t j per(A(t | j )), for any j , 1 ≤ j ≤ n.
t =1

Theorem 3 If B and C are m × n and n × m matrices, respectively, m ≤ n, then

X 1 ¡ ¢ ¡ ¢
per(BC ) = per B [1, . . . , m|ω] per C [ω|1, . . . , m] .
ω∈G m,n µ(ω)
2.2. E LEMENTARY T HEOREMS FOR P ERMANENTS 9

Example (Theorem 3 for Permanent)

Let  
· ¸ 1 0
1 2 3
B= , C = 0 1 .
4 5 6
1 1
2
Here, B is 2 × 3, C is 3 × 2, so m = 2, n = 3.
The subsets ω ∈ G 2,3 are all 2-element subsets of {1, 2, 3}:

ω1 = (1, 2), ω2 = (1, 3), ω3 = (2, 3).

For each ωi , compute:

per(B [:, ωi ]) and per(C [ωi , :]).

1. For ω1 = (1, 2):

· ¸ · ¸
1 2 1 0
B [:, ω1 ] = , C [ω1 , :] =
4 5 0 1

per(B [:, ω1 ]) = 1 · 5 + 2 · 4 = 13, per(C [ω1 , :]) = 1 · 1 + 0 · 0 = 1.

2. For ω2 = (1, 3):

· ¸ · ¸
1 3 1 0
B [:, ω2 ] = , C [ω2 , :] =
4 6 1 1

per(B [:, ω2 ]) = 1 · 6 + 3 · 4 = 18, per(C [ω2 , :]) = 1 · 1 + 0 · 1 = 1.

3. For ω3 = (2, 3):

· ¸ · ¸
2 3 0 1
B [:, ω3 ] = , C [ω3 , :] =
5 6 1 1

per(B [:, ω3 ]) = 2 · 6 + 3 · 5 = 27, per(C [ω3 , :]) = 0 · 1 + 1 · 1 = 1.

Now apply the theorem:

1
per(B [:, ω]) per(C [ω, :])
X
per(BC ) =
ω∈G 2,3 µ(ω)

Since all µ(ω) = 1 (distinct indices), we get:

per(BC ) = 13 · 1 + 18 · 1 + 27 · 1 = 58.

Theorem 4 Let A = (a i j ) and B = (b i j ) be n × n matrices. Then

n
per(A[α | β]) per(B (α | β)),
X X
per(A + B ) =
r =0 α,β∈Q r,n

where per(A[α | β]) = 1 and det(B (α | β)) = det(B ) when r = 0, and det(B (α | β)) = 1 when
r = n.
 10 2. P ERMANENT AND I TS E LEMENTARY P ROPERTIES

Detailed Example: Verification of Theorem 4 for 3 × 3 Matrices

Let:    
1 0 2 0 2 1
A = 0 1 1  , B = 3 1 0 
1 1 0 1 0 2
2 Then:  
1 2 3
A + B = 3 2 1
2 1 2
Step 1: Compute per(A + B ), using the formula:

3
X Y
per(A + B ) = (A + B )i ,σ(i )
σ∈S 3 i =1

Permutations of {1, 2, 3} and corresponding products:

σ1 = (1, 2, 3) : 1 · 2 · 2 = 4
σ2 = (1, 3, 2) : 1 · 1 · 1 = 1
σ3 = (2, 1, 3) : 2 · 3 · 2 = 12
σ4 = (2, 3, 1) : 2 · 1 · 2 = 4
σ5 = (3, 1, 2) : 3 · 3 · 1 = 9
σ6 = (3, 2, 1) : 3 · 2 · 2 = 12

per(A + B ) = 4 + 1 + 12 + 4 + 9 + 12 = 42
Step 2: Use Theorem 4

3
per(A[α|β]) · per(B [α′ |β′ ])
X X
per(A + B ) =
r =0 α,β∈Q r,3

We consider each r from 0 to 3:

r = 0:
per(;) · per(;) = 1 · 1 = 1
r = 1: ¡ ¢2
Choose all 1 × 1 minors: There are 31 = 9 combinations.
Compute all A[i | j ] · B [i ′ | j ′ ] for i , j = 1, 2, 3:
Example: · ¸
1 0
A[1|1] = 1, B [2, 3|2, 3] = , per = 1 · 2 + 0 · 0 = 2
0 2
- Similar for the other 8 combinations.
r = 2: ¡ ¢2
Choose all 2 × 2 minors: There are 32 = 9 combinations.
2.2. E LEMENTARY T HEOREMS FOR P ERMANENTS 11

Example: ·¸
1 0
- A[1, 2|1, 2] = , per = 1 - B [3|3] = [2], per = 2 - Product = 1 · 2 = 2
0 1
Repeat for all 9 combinations.
r = 3:
2
per(A) = 1 · 1 · 0 + 0 · 1 · 1 + 2 · 1 · 1 + · · · = 12 , per(B [;]) = 1 ⇒ 12 · 1 = 12

Final Sum:
per(A + B ) = 1 + 13 + 16 + 12 = 42
 3
S OME S PECIAL T RIDIAGONAL
M ATRICES AND ITS P ERMANENTS
This Chapter gives some permanents of some special types of triadiagonal matrices,
such that the calculation of their permanents is quite easy by a single formula. First, we’ll
define those matrices and then provide a comprehensive proof of their permanents.

3.1. T RIDIAGONAL M ATRIX

Def: A tridiagonal matrix is a matrix in which only the main diagonal, the above and
below it diagonal have non-zero elements. All other elements have zero value. Because
of this, tridiagonal matrices of order less than or equal to 3 appear to be meaningless.
e.g: Let A be a 6 × 6 Tridiagonal Matrix then:
 
a 11 a 22 0 0 0 0
a a 22 a 23 0 0 0 
 21 
 0 a 32 a 33 a 34 0 0 
 
A=
 0 0 a 43 a 44 a 45 0 

 
 0 0 0 a 54 a 55 a 56 
0 0 0 0 a 65 a 66

Def: A matrix is said to be a (-1,0,1)-matrix if each of its entries are -1 , 0 , 1.

Laplace Expansion: The Laplace expansion equation is written as:
n
a i , j (−1)i + j M i , j
X
det(A) = for i ∈ {1, 2, · · · , n}
j =1

3.2. M ATRIX C ONTRACTION

Let A = [a i j ] be an m × n real matrix with row vectors α1 , α2 , . . . , αm .
Contraction on a column

13
 14 3. S OME S PECIAL T RIDIAGONAL M ATRICES AND ITS P ERMANENTS

A matrix A is said to be contractible on a column k if: - Column k contains exactly two

nonzero entries. - Suppose these two entries are a i k ̸= 0 and a j k ̸= 0, where i ̸= j .
Procedure: 1. Replace row i in matrix A with:

a j k αi + a i k α j

2. Delete rows j and column k. 3. The resulting (m − 1) × (n − 1) matrix is called the

contraction of A on column k relative to rows i and j . 4. Denoted as:

3 A i j :k

Contraction on a row
A matrix A is contractible on a row k if: - Row k contains exactly two nonzero entries. -
Suppose these two entries are a ki ̸= 0 and a k j ̸= 0, where i ̸= j .
Procedure
1. Replace column i with:
a k j αTi + a ki αTj
(Transpose vectors are used here since we’re dealing with columns.) 2. Delete columns
j and row k. 3. The resulting matrix is called the contraction of A on row k relative to
columns i and j . 4. Denoted as:
h iT
A k:i j = A Tij :k

Matrix Contraction Chain

We say that a matrix A can be contracted to a matrix B if: - B = A, or - There exists a
sequence of matrices A 0 , A 1 , . . . , A t , such that: - A 0 = A - A t = B - A r is a contraction of
A r −1 for r = 1, 2, . . . , t

Theorem 5 Let C be a matrix given below and let B be a contraction of C then:

per (C ) = per (B ) (3.1)

 
c 11 −c 12 0
 .. 
c
 21 c 22 −c 23 . 

. .. .. .. 
Let C =  . . . . .
 

 

 c n−2,n−2 −c n−2,n−1 0 

 −c n−1,n−2 −c n−1,n−1 −c n=1,n 
0 −c n,n−1 c n,n

Proof: It suffices to consider the case where B is the contraction of C on column 1 relative
to rows 1 and 2. Thus, C and B have the form
 
c 11 −c 12 hc c +c c i
11 22 12 21
C = −c 21 c 22  , B =
D
0 D
3.2. M ATRIX C ONTRACTION 15

where c i j ̸= 0.
Using the Laplace expansion of the permanent with respect to column 1, we obtain
µ· ¸¶ µ· ¸¶
c 22 c 12
per(C ) = c 11 · per + c 21 · per
D D

Hence, by linearity of the permanent, perC = per B

Let A n denote an n × n (−1, 0, 1)-tridiagonal Toeplitz matrix as follows: for n ≥ 3 3

 
1 −1 0
−1 1 −1 
 
 .. 

−1 1 . 
An =  (3.2)
 
.. ..

. . −1
 
 
 
 −1 1 −1
0 −1 1

Corollary 1. Let A n be the n ×n (−1, 0, 1)-tridiagonal Toeplitz matrix as in (3.2). Then,

for n ≥ 3
per A n = F n+1
where F n is the nth Fibonacci number.
Proof:
If n = 3, then we have  
1 −1 0
A 3 = −1 1 −1
0 −1 1
and hence per A 3 = 3 = F 4 .
p
Let A n be the pth contraction of A n , 1 ≤ p ≤ n − 2. From the definition of A n , the
matrix A n can be contracted on column 1 so that
 
2 −1
−1 1 −1 
 
1
 .. 
An = 

−1 1 . 

.. ..
 
. . −1
 

−1 1

Since the matrix A 1n can be contracted on column 1 and F 4 = 3, F 3 = 2,

   
3 −2 F 4 −F 3
−1 1 −1  −1 1 −1 
   
2
 . .
  . .

An = 

−1 1 . =
 
−1 1 . 

.. .. .. ..
   
. . −1  . . −1
   

−1 1 −1 1
 16 3. S OME S PECIAL T RIDIAGONAL M ATRICES AND ITS P ERMANENTS

Furthermore, the matrix A 2n can be contracted on column 1 so that

 
5 −3
−1
 1 −1 

3
 .. 
An = 

−1 1 . 

.. ..
 
. .
 
 −1
−1 1

3 where F 5 = 5, F 4 = 3. Continuing this process, we obtain

 
F r +2 −F r +1
 −1
 1 −1 

 .. 
A rn =  −1 1 . for 3 ≤ r ≤ n − 4.
 

.. ..
 
. .
 
 −1
−1 1

Hence,  
F n−1 −F n−2 0
A n−3
n =  −1 1 −1
0 −1 1

which, by contraction of A n−4

n on column 1, gives
· ¸ · ¸
F n−2 + F n−1 −F n−1 Fn −F n−1
A n−2
n = =
−1 1 −1 1

By applying Theorem 5, we obtain

per A n = per A n−2

n = F n+1 .

Lemma 1: Let Vn be an n × n matrix of the form:

−1 2 0
 
1 −1 1 0 
 
0 1 −1 1 0 
 
 .. 

0 1 −1 1 . 
Vn =  ,
 
.. .. .. ..
. . . .
 

 0 


 0 1 −1 1 0 
 0 1 −1 1
0 1 −1
and let Vnm denote the mth contraction of the matrix Vn . Then

perVnn−2 = L −n ,
3.2. M ATRIX C ONTRACTION 17

where L −n is the nth negatively subscripted Lucas number.

Proof: From the definition of the matrix Vn , the matrix Vn can be contracted on col-
umn 1, yielding:

3 −1 0
 
 .. 
1
 −1 1 . 

 .. 
1

Vn =  0 1 −1 . 0

.
.. .. ..
 


 . . . 1 0

 3
 0 1 −1 1
0 1 −1

Since Vn1 can be contracted on column 1 and using the values L −3 = −4, L −2 = 3, we
obtain:
   
−4 3 0 L −3 L −2 0
1 −1 1   1 −1 1 
   
 ..   .. 
2 0

1 −1 .   0
 
1 −1 . 
Vn =  = . .

 .. .. .. ..   . .. .. ..
. . . . . .

 .   . 
   
0 1 −1 1  0 1 −1 1
0 0 1 −1 0 0 1 −1

Further contracting Vn2 on column 1 and using L −4 = 7, we get:

   
7 −4 0 L −4 L −3 0
1 −1 1   1 −1 1 
   
 ..   .. 
3 0

1 −1 .   0
 
1 −1 . 
Vn =  = . .

 .. .. .. ..   . .. .. ..
. . . . . .

.   . 
   
0 1 −1 1  0 1 −1 1
0 0 1 −1 0 0 1 −1

Continuing this process, we reach:

 
L −(n−2) L −(n−3) 0
Vnn−3 =  1 −1 1 ,
0 1 −1

which, by contraction of Vnn−4 , gives:

· ¸
L −(n−1) L −(n−2)
Vnn−2 = .
1 −1

Calculating the permanent of Vnn−2 :

perVnn−2 = L −(n−2) − L −(n−1) ,

 18 3. S OME S PECIAL T RIDIAGONAL M ATRICES AND ITS P ERMANENTS

and using the identity for Lucas numbers, we obtain:

perVnn−2 = L −n .

3
 4
C ONCLUSION
HIV dynamical models (with treatment and without treatment) has been presented in
this thesis. It is concluded that

19
 A
A PPENDIX
This appendix present the basic mathematical concepts and definitions [1–3] that are
used in this thesis to simulate the dynamical system for HIV.

A.1. O RDINARY D IFFERENTIAL E QUATION

If an equation contains ordinary derivatives of one or more unknown functions with
respect to single independent variable is called ordinary differential equation. Th n th
order ordinary differential equation in the normal form is,

d n y(t ) ′ ′′
= f (t , y(t ), y (t ), y (t ), · · · , y n−1 (t )) (A.1)
dtn

for a ≤ t ≤ b and t ∈ ℜ where ℜ is set of Real Values. For example, d x(t )/d t = 5x(t ) − 3.
Any function g (t ), defined on interval I possesses atleast n derivatives that are contin-
uous on I , which when substituted into an n th order ordinary differential equation re-
duces the equation to an identity, is said to be a solution of the equation on the interval
I . If g (t ) is the solution of the (A.1) then it must satisfies the given ODE and initial con-
ditions.
d n g (t ) ′ ′′
− f (t , g (t ), g (t ), g (t ), ..., g n−1 (t )) = 0
dtn

A.1.1. C LASSIFICATION OF ODE S

The ordinary differential equations (ODEs) are classified by order and linearity as fol-
lows.

O RDER
The order of an ordinary differential equation is the order of the highest derivative in
the equation. For example, d 2 x(t )/d t 2 = 5x(t ) − 3 is a 2nd order ordinary differential
equation.

21
 22 A. A PPENDIX

L INEARITY
A An n th order ordinary differential equation is said to be linear if it is of the following form.

d n x(t ) d n−1 x(t ) d 2 x(t ) d x(t )

a n (t ) + a n−1 (t ) + · · · + a 2 (t ) + a 1 (t ) + a 0 (t )x(t ) = f (t ) (A.2)
dtn d t n−1 dt2 dt
where x(t ) has atleast n derivative with respect to t , hence is the solution of (A.2) for
t ∈ I and also a i (t ) for i = 0, 1, 2, · · · , n are continuous for every t ∈ I . For example,
d 3 x(t )/d t 3 = 5d 2 x(t )/d t 2 − 3 is 3rd order linear ordinary differential equation.

N ON - LINEAR O RDINARY D IFFERENTIAL E QUATION

Equations of form other than (A.2) are known as Non-linear Ordinary Differential Equa-
tions. For example, d y/d t = y 2 for some t belongs to some interval I is a non-linear
ODE.

A.2. AUTONOMOUS AND N ON -AUTONOMOUS ODE S

An ordinary differential equation in which the independent variable does not appear
explicitly is said to be autonomous ODE.
d n y(t ) ′ ′′ ′′′
n
= f (y, y , y , y , · · · , y n−1 ) (A.3)
dt
where t belongs to some interval I and y(t ) be the solution of (A.3) For example, d x(t )/d t =
x(t ) + sin(x(t )) is a autonomous equation. If the any term of the an ordinary differential
equation contain independent variable explicitly then it is said to be non-autonomous
ODE.
d n y(t ) ′ ′′ ′′′
n
= f (t , y, y , y , y , · · · , y n−1 ) (A.4)
dt
where t belongs to some interval I and y(t ) be the solution of (A.4), for example, d x(t )/d t =
x(t ) + sin(x(t )) + t 2 + 7t is a non-autonomous equation.

A.3. DYNAMICAL S YSTEM

A dynamical system is any system that allows one to determine (at least theoretically)
the future states of the system given its present or past state. For example, the recursive
formula (difference equation) x n+1 = (1.05)x n , for n = 0, 1, 2, · · · is a dynamical system,
since the future state x n+1 is determined given the previous state, x n . The differential
equations are also know as a dynamical system such as initial value problem (the in-
formation about the system is known at either the present or past state). For example,
d x/d t = −2x, where the solution x(t ) shows the state of the system at any time t . If the
initial condition is known i.e. x(t 0 ) = x 0 , then system state can be determined at any
past, present or future instant of time by finding the exact solution i.e. x(t ) = x 0 e −2(t −t0 )
for t ≥ t 0 . Following are the the types of behaviour of dynamical systems.
1. Equilibrium solutions: A constant solution of the autonomous differential equa-
tion d x/d t = f (x) is called an equilibrium of the equation. In other words, it is a
solution which satisfies f (x) = 0. The solutions either converge to the equilibrium
or diverge away from it.
R EFERENCES 23

2. Periodic orbits: A periodic solution x = x(t ), y = y(t ). A periodic solution is called

a cycle. If p is the period of the solution, then X (t + p) = X (t ) and a particle placed A
on the curve at X 0 will cycle around the curve and return to X 0 in p units of time
(see [1], page no. 366).

3. Chaotic orbit: An orbit that exhibits an unstable behavior that is not itself fixed
or periodic is called a chaotic orbit. At any point in such an orbit, there are points
arbitrarily near that will move away from the point during further iteration. In
terms of solutions, it means they are very sensitive to small perturbations in the
initial conditions and almost all of them do not appear to be either periodic or
converge to equilibrium solutions

R EFERENCES
[1] D. G. Zill, A First Course in Differential Equations with Modeling Applications, 10th
ed. (Brooks Cole, 2012).

[2] R. L. Burden and J. D. Faires, Numerical Analysis, 9th ed. (Brooks Cole, 2010).

[3] R. K. Nagle, E. B. Saff, and A. D. Snider, Fundamentals of Differential Equations, 8th

ed. (Addison Wesley, 2011).
 B
A PPENDIX

1 %% main file
2 %% Constants
3 d = 0.02;
4 k = 100;
5 s = 10;
6 beta = 2.4e−5;
7 m1 = 2.4;
8 m2 = 0.24;
9 %% Initial Conditions.
10 ini_x = [500 800 1000 1200];
11 ini_y = [1e−6 10 50 0];
12 ini_z = [60 70 30 100];
13 name1 = {'T(0) = 500','T(0) = 800','T(0) = 1000','T(0) = 1200'};
14 name2 = {'T*(0) = 1e−6','T*(0) = 10','T*(0) = 50','T*(0) = 0'};
15 name3 = {'V(0) = 60','V(0) = 70','V(0) = 30','V(0) = 100'};
16 %% Interval of Computation
17 a = 0;
18 b = 400;
19 h = 0.1; %Step size in the interval.
20 %% Plotting Style
21 CC = {'y','r','g','b','g','y',[.5 .6 .7],[.8 .2 .6]};
22 Markers = {'+','o','*','x','v','d','^','s','>','<'};
23 linespec = {'−',':','−−','−.'};
24 %% System of Equations 2.1
25 f1 = @(t,x,y,z)s−d*x − beta*x*z;
26 f2 = @(t,x,y,z)beta*x*z − m2*y;
27 f3 = @(t,x,y,z)k*y − m1*z;
28 for i=1:length(ini_x)
29 ini = [ini_x(i) ini_y(i) ini_z(i)];
30 RK4_Method = RK4(a,b,h,ini,f1,f2,f3);
31 h1 = figure(1)
32 plot(RK4_Method(:,1),RK4_Method(:,2),linespec{i},'color',...
33 CC{i},'LineWidth',2,'DisplayName',name1{i})
34 axis tight
35 xlabel('Time (Days)'), ylabel('T(t)')

25
 26 B. A PPENDIX

36 legend('Location','northeast')
37 hold on
38 %% fig file writing
39 set(findall(gcf,'−property','FontSize'),'FontSize',12)
40 set(gca,'FontName','Times New Roman')
set(gcf,'Units','Inches','Position', [0,5.8, 5.5, 2])
B 41

42 h2 = figure(2)
43 plot(RK4_Method(:,1),RK4_Method(:,3),linespec{i},'color',...
44 CC{i},'LineWidth',2,'DisplayName',name2{i})
45 axis tight
46 xlabel('Time (Days)'), ylabel('T*(t)')
47 legend
48 hold on
49 %% fig file writing
50 set(findall(gcf,'−property','FontSize'),'FontSize',12)
51 set(gca,'FontName','Times New Roman')
52 set(gcf,'Units','Inches','Position', [0,3, 5.5, 2])
53 h3 = figure(3)
54 plot(RK4_Method(:,1),RK4_Method(:,4),linespec{i},'color',...
55 CC{i},'LineWidth',2,'DisplayName',name3{i})
56 axis tight
57 xlabel('Time (Days)'), ylabel('V(t)')
58 legend('Location','northeast')
59 hold on
60 %% fig file writing
61 set(findall(gcf,'−property','FontSize'),'FontSize',12)
62 set(gca,'FontName','Times New Roman')
63 set(gcf,'Units','Inches','Position', [0,0.3, 5.5, 2])
64 h4 = figure(4)
65 plot3(RK4_Method(:,2),RK4_Method(:,3),RK4_Method(:,4),...
66 linespec{i},'color',CC{i},'LineWidth',2)
67 % title('Plotting the system')
68 axis tight
69 xlabel('T(t)'), ylabel('T*(t)'), zlabel('V(t)')
70 grid on
71 hold on
72 %% fig file writing
73 set(findall(gcf,'−property','FontSize'),'FontSize',12)
74 set(gca,'FontName','Times New Roman')
75 set(gcf,'Units','Inches','Position', [8, 4, 5, 3])
76 end
77 %% Save the file as PNG
78 print(h1,'RK4_HealthyCells_System_1','−dpng','−r600');
79 print(h2,'RK4_InfectedCells_System_1','−dpng','−r600');
80 print(h3,'RK4_VirusLoad_System_1','−dpng','−r600');
81 print(h4,'RK4_PhasePortrait_System_1','−dpng','−r600');

mfile: main.m