Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
STUDY ON CLASS OF IMPROVED Q-BERNOULLI
MATRIX AND ITS PROPERTIES
A THESIS SUBMITTED TO THE GRADUATE
SCHOOL OF APPLIED SCIENCES
OF
NEAR EAST UNIVERSITY
By
IBRAHIM YUSUF KAKANGI
In Partial Fulfilment of the Requirements for
the Degree of Master of Science
in
Mathematics
NICOSIA, 2017
IBR
AH
IM Y
USU
F KA
KA
NG
I, A ST
UD
YO
N A
CL
ASSE
S OF Q
-BE
RN
OU
LL
I MA
TR
IX A
ND
ITS PR
OPE
RT
IES,
NE
U, 2017
STUDY ON CLASS OF IMPROVED Q-BERNOULLI
MATRIX AND ITS PROPERTIES
A THESIS SUBMITTED TO THE GRADUATE
SCHOOL OF APPLIED SCIENCES
OF
NEAR EAST UNIVERSITY
By
IBRAHIM YUSUF KAKANGI
In Partial Fulfilment of the Requirements for
the Degree of Master of Science
in
Mathematics
NICOSIA, 2017
Ibrahim Yusuf KAKANGI: STUDY ON CLASS OF IMPROVED Q-BERNOULLI
MATRIX AND ITS PROPERTIES
Approval of Director of Graduate School of
Applied Sciences
Prof. Dr. Nadire Çavuş
We certify that, this thesis is satisfactory for the award of the degree of Master of
Sciences in Mathematics.
Examining Committee in Charge:
Prof. Dr. Allaberen Ashyralyev Committee Chairman, Department of
Mathematics, Near East University.
Assoc.Prof. Dr. Suzan Cival Buranay External Examiner, Department of
Mathematics, Eastern Mediterranean
University.
Assis. Prof. Dr. Mohammad Momenzadeh Supervisor, Department of
Mathematics, Near East University.
I hereby declare that all information in this document has been obtained and presented in
accordance with academic rules and ethical conduct. I also declare that, as required by these
rules and conduct, I have fully cited and referenced all material and results that are not original
to this work.
Name, Last name: Yusuf Ibrahim Kakangi
Signature:
Date
ACKNOWLEDGMENTS
My deepest appreciation goes to Almighty God for being my strength, help and my reference
point towards the completion of my master degree course.
I wish to express my sincere gratitude to my supervisor Ass.Prof.Dr. Mohammad
MOMENZADEH invaluable assistance, guidance and thorough supervision. His keen eyes for
details and uncompromising insistence on high standard has ensured the success of this thesis.
I am most indebted to my sponsors: Kaduna State Government, Nigeria. And I am most
grateful to my tireless parent, brother, sisters, Nephew, cousins, relatives, friends and course
mate, for their support, useful advice, and encouragement towards the completion of my
master’s programme. May almighty Allah bless you and grant all your heart desires.
To Dr. Ramalan Yero ….
i
ABSTRACT
Since 19th century, a lot of q-Bernoulli numbers and polynomials has been introduced. Carlitz
was the first who made a generation of q-Bernoulli numbers, afterwards, a lot of researcher’s
works on a new form of q-Bernoulli numbers and matrices. In this thesis, we introduce
ordinary Bernoulli and q-Bernoulli matrices and their related Pascal matrices and their
relations. At the end by using generating function and improved q-exponential function we
work on a new class of q-Bernoulli matrix and related properties are given. Our definition is
more significant since it demonstrates a better definition of q-Bernoulli matrix and the
properties are convinced the ordinary case as well.
KEYWORDS: Bernoulli Number; Bernoulli Matrices; q-Bernoulli Number; q-Bernoulli
Matrices; Improved q-Bernoulli number; and Improved q-Bernoulli Matrices
ii
ÖZET
19. yüzyıldan beri, bir sürü q-Bernoulli sayısı ve polinomları tanıtıldı. Carlitz, daha sonra, q-
Bernoulli sayılarının ve matrislerinin yeni bir formuyla ilgili birçok araştırmacı tarafından q-
Bernoulli sayıları üreten ilk kişiydi. Bu tezde sıradan Bernoulli ve q-Bernoulli matrislerini ve
ilgili Pascal matrislerini ve bunların ilişkilerini tanıtmaktayız. Sonunda üretme fonksiyonu ve
geliştirilmiş q-üstel fonksiyonu kullanılarak q-Bernoulli matrisinin yeni bir sınıfında çalıştık
ve ilgili özellikler verildi. Tanımımız, q-Bernoulli matrisinin daha iyi tanımlanmasını
gösterdiği için daha belirgindir ve özellikler olağan durumu ikna eder.
ANAHTAR KELİMELER: Bernoulli sayısı; Bernoulli matrisleri; q-Bernoulli sayısı; q-
Bernoulli matrisleri; Geliştirilmiş q-Bernoulli sayısı; ve
Geliştirilmiş q-Bernoulli Matrisleri
iii
TABLE OF CONTENTS
ACKNOWLEDGMENTS………………………………………………………………... i
ABSTRACT……………………………………………………………………………….. ii
ÖZET………………………………………………………………………………………. iii
TABLE OF CONTENTS………………………………………………………………… iii
CHAPTER 1: INTRODUCTION…………………………………………………………… 1
1.1 Quantum Calculus.........................................................................................................1
1.1.1 Definition q And h-Differentiation............................................................................2
1.1.2 Definition q And h-Derivative..................................................................................2
1.1.3 Lemma Linearity Of q And h-Derivative..................................................................3
1.2 q-Taylor’s Formula For Polynomial.............................................................................4
1.2.1 q-Analogue Of Some q-Combinatory.........................................................................4
1.2.2 Some Properties of q-Calculus Functions..................................................................5
1.3 q-Exponential Function.................................................................................................6
1.3.1 Gauss binomial formula..............................................................................................6
1.3.2 Heines Binomial Formula...........................................................................................6
1.4.3 q-Euler Identities.........................................................................................................7
1.3.4 q-Exponential Functions..............................................................................................8
1.3.5 Relationship Between eqx and Eqx...........................................................................8
1.3.6 q-Derivative Of The q-Exponential Functions............................................................9
1.3.7 Convergence Of q-Exponentials Functions.................................................................9
1.4 q-Trigonometric Functions.........................................................................................10
iv
1.4.1 Properties Of q-Trigonometric Functions.................................................................10
1.4.2 q-Derivetive Of q-Trigonometric Functions.............................................................10
1.5 Improved q-Fxponential Function..............................................................................11
1.5.1 Definition ℇ qz.........................................................................................................11
1.5.2 Basic Definitions on Improved q-Exponential Function...........................................11
1.5.3 Unification Of q-Exponential Functions.................................................................12
1.5.4 Improved q-Trigonometric Functions......................................................................13
1.6 Bernoulli Number.......................................................................................................13
1.6.1 Recurrence Formula for Ordinary Bernoulli Number...............................................14
1.6.2 Kronecker Delta.......................................................................................................15
1.6.3 Lemma Explicit Definition of Bernoulli Number....................................................16
1.6.4 Proposition Bernoulli Numbers as Rational numbers............................................16
1.6.5 Bernoulli Polynomials.............................................................................................16
1.6.6 Some Properties Of Bernoulli Polynomials............................................................17
CHAPTER 2: BERNOULLI MATRIX AND SOME PROPERTIES…………………... 20
2.1 Bernoulli Matrix..........................................................................................................20
2.1.1 Definition Ordinary Matrix and some Properties...................................................20
2.1.2 Definition Bernoulli Matrix and Bernoulli Polynomials..........................................21
2.1.3 Theorem Bernoulli Polynomial Matrix of x and y..................................................21
2.1.4 Definition Inverse Of Bernoulli Matrix..................................................................24
2.2 Bernoulli Matrix and Generalized Pascal Matrix.......................................................25
2.2.1 Definition Pascal Matrix........................................................................................25
2.2.2 Theorem Relationship Between Barnoulli Polynomial Matrix And Pascal Matrix 25
2.2.3 Theorem Inverse Of Bernoulli Polynomial Matrix and Pascal Matrix....................28
v
CHAPTER 3: Q-BERNOULLI MATRIX AND ITS PROPERTIES…………………… 32
3.1 Q-Bernoulli Matrix.....................................................................................................32
3.1.1 Definition q-Bernoulli numbers...............................................................................32
3.1.2 Definition q-Bernoulli Polynomials.........................................................................33
3.1.3 Definition q-Bernoulli Matrix..................................................................................33
3.1.4 Definition q-Bernoulli Polynomials Matrix.............................................................34
3.1.5 Theorem Inverse of q-Bernoulli Matrix...................................................................34
3.2 q-Bernoulli Matrix and q-Pascal Matrices.......................................................................36
3.2.1 Definition Pascal Matrix and Inverse of Pascal Matrix...........................................36
3.2.2 Theorem Relationship Between q-Bernoulli Polynomial Matrix and Pascal Matrix
............................................................................................................................................37
3.2.3 Definition Inverse of q-Bernoulli Polynomials Matrix............................................41
3.2.4 Corollary......................................................................................................................42
CHAPTER 4: IMPROVED Q-BERNOULLI MATRIX AND ITS PROPERTIES……. 44
4.1 History of q-Bernoulli Numbers......................................................................................44
4.1.1 Definition Carlitz q-Bernoulli Number....................................................................44
4.2 Improved q-Bernoulli Numbers......................................................................................45
4.2.1 Lemma Recurrence Formula For Improved q-Bernoulli Number...........................45
4.2.3 Lemma Advantage Of Improved q-Exponential Function......................................46
4.2.4 Improved q-Bernoulli Polynomials...........................................................................48
4.2.5 Theorem Additive Theory........................................................................................48
4.3 Q-Improved Bernoulli Matrix And Its Properties..........................................................51
vi
4.3.1 Definition Improved q-Bernoulli Matrix and Improved q-Bernoulli Polynomials
Matrix.................................................................................................................................51
4.3.2 Theorem Improved q-Bernoulli Polynomial Matrix in terms of x and y................51
4.3.3 Theorem Inverse of Improved q-Bernoulli matrix………………………………. 52
CHAPTER 5: SUMMARY AND CONCLUSION………………………………………. 54
REFERENCE……………………………………………………………………………… 54
vii
CHAPTER 1
INTRODUCTION
Majority of the work in this chapter was presented from The book ‘ A comprehensive
treatment of q-Calculus’ (Ernst.T, 2012) , ‘Quantum Calculus’ (Kac.V, and Cheung.P,2002)
and any other information that is not from there was cited by means of reference.
Among the most important sequence in mathematics is the sequence of Bernoulli numbers Bn
it has a quiet good relationship to the number theories, for example you can express the value
of 𝜁(2𝑛) using the Bernoulli number, where 𝑛 is a positive integer and 𝜁(𝑛) is a Riemann
zeta function [1] you can also find the uses of Bernoulli number in analysis, for instance, they
also find it in the Euler-Maclaurins formula, the formula that is very useful in physics and
mathematics, in asymptotic of a q-special functions, the Bernoulli numbers is very essential.
Bernoulli matrix, Pascal matrix are some example of matrix with binomial coefficients as their
element which are very important in matrix theory and combinatory. So many researchers has
been showing interest in this related area, for that reason we also want to relate this kind of
matrix with quantum calculus (q-calculus), but just before then here are some terminologies
that one needs to know about q-calculus.
1.1 Quantum Calculus
If
limx→ x0
f ( x )− f (x0)x−x0
(1.1)
exists, that gives us the well known definition of the derivative dydx of a function f(x) at a point
x=x0. However if we assume that x=q x0 or x=x0+h, where q is a fixed constant not equal to
1 and h is a fixed constant not equal to 0, and do not take the limit, we enter into different
concept of mathematics called the quantum calculus.
1
Quantum calculus involves two types of derivative which are; q-derivative and h-derivative
that leads to the study of q-calculus and h-calculus respectively. In the course of studying
quantum calculus in relation to the ordinary calculus, so many important results and notions in
number theory, combinatory and different area of mathematics have been discovered.
For instance, a q-derivative ofxn=[ n ]q xn−1 , where
[ n ]q=qn−1q−1
(1.2)
and [ n ]q represent the ordinary 'n' in the ordinary derivative of xn.
1.1.1 Definition q and h-Differentiation
A q-differential and h-differential of an arbitrary function say f (x) on the set of real numbers
are defined as (Kac.V, and Cheung.P,2002)
dq f (x)=f (qx )−f ( x )(1.3)
And
dh f ( x )=f (x+h )− f ( x )(1.4)
Respectively.
Particularly keeping in mind that, dq x=(q−1 ) x also that dh x=h.
Considering the q-differential and the h-differential we defined their corresponding quantum
derivative as
1.1.2 Definition q and h-Derivative
Supposed f ( x ) is an arbitrary function on the set of real numbers R. Its q-derivative and h-
derivative are defined as
Dq f ( x )=dq f ( x )
dq x=
f (qx )−f ( x )(q−1 ) x
(1.5)
where, q ≠ 1 ,∧x ≠0
Dh f ( x )=dh f (x )
dh x=
f (x+h )−f (x)h
(1.6)
2
with h ≠ 0
referred to as the q-derivative and h-derivative respectively, of the arbitrary function f (x)
If we noticed that
limq→1
D q f (x )=limh→ 0
D h f (x)=df (x)dx
provided that the function f ( x ) is differentiable. Looking at notation of Leibniz df ( x )
dx, which
has to do with the ratio of two ‘infinitesimals’ is somewhat difficult to understand, because
there is need to give further detail of the notion of the differential df ( x ). But on the other hand,
one can easily see on the notion of q-calculus and h- calculus also the q-derivative and the h-
derivative are plain ratios.
1.1.3 Lemma Linearity of q and h-Derivative
Just like in the concept of ordinary derivative, the linear operator behaves in the same way
while finding the q-derivative or h-derivative of a function. In essence, if Dq and Dh are q-
derivative and h-derivative, then for any constants a and b, the following property hold:
Dq (af ( x )+bg ( x ) )=a D q f (x )+b D q g ( x )(1.7)
Dh (af ( x )+bgf (x ) )=a Dh f ( x )+b Dh g ( x )(1.8)
Example: if f ( x )=xn, and n is an integer greater than zero, then the q-derivative and h-
derivative can be find as
Dq xn=qn xn−xn
qx−x=
(qn−1)xn
(q−1)x=
(qn−1)(q−1)
xn−1(1.9)
Dh xnh=
(x+h )n+xn
h=nxn−1+
n (n−1 ) xn−2 h2
+…+hn−1(1. 10)
but there is a frequent appearance of (q¿¿n−1)
(q−1)¿ in the q-derivative so therefore we used the
notation
[ n ]q=(q¿¿ n−1)
(q−1)=qn−1+…+1 ¿
3
And it is referred to as the q-analogue of n, for any integer n greater than zero, and hence (1.9)
becomes
Dq xn=[n ]q xn−1(1.11)
1.2 q-Taylor’s Formula For Polynomial
Before going to q-Taylor formula, lets recall the generalized Taylor formula in the ordinary
calculus.
Taylor theorem says
f ( x )=∑n=0
∞
f (n)(a)(x−a)n
n !,(1.12)
is the power series of any function f ( x ) which has derivative of all kind of order is analytic at
x=a, provided we can write it as a power series about a point x=a.
We can increase the definition of a function to a more interesting domain by Taylor expansion
of an analytic function. For instance, if we defined the exponentials as a square matrices and a
complex number by using the Taylor expansion of ex, with which then we express the q-
analogue of the following expression where the q-Taylor formula follows
1.2.1 q-Analogue of Some q-Combinatory
Definition 1.2.1 (Kac.V, and Cheung.P,2002) If n is a positive integer. we defined the q-
analogue of n ! as:
[n]q !={ 1 , for n=0[n ] [ n−1 ] !q , for n ≥1
(1.13)
Definition 1.2.2 The q- binomial coefficients of any integer n ≥ k ≥1, is defined as
(nk)q=
[n ] !q
[ k ] !q [n−k ] !q,(1.14)
and (1.13) satisfies (n0)q=1 also
(nk)q=0 for n<k (Naim and KUS, 2015)
4
(nk)q=( n
n−k )q,(1.15)
and
(nk)q(k
j)q=(nj)q
(n− jk− j)q
(1.16)
the q-analogue of binomial function (x−a)n is defined as:
(x−a)qn={ 1 , for n=0
( x−a ) ( x−aq ) …( x−aqn−1) , for n=1,2 , …(1.17)
Definition 1.2.3 (Kac.V, and Cheung.P,2002)
For n<0, q-analogue of (x−a)−n is defined as:
( x−a )q−n= 1
( x−q−n a )qn , for n=1,2 , …(1.18)
Definition 1.2.4 (Kac.V, and Cheung.P,2002)
Letα∈Z , the q-analogue of α is defined as:
[ α ]q=1−qα
1−q(1.19)
Definition 1.2.5 (Kac.V, and Cheung.P,2002)
Generalize q-polynomial function is defined as
pn ( x )=(x−a)q
n
[n ] !q(1.20)
Where pn(x ) is a polynomial.
1.2.2 Some Properties of q-Calculus Functions
Proposition 1.2.1 (Kac.V, and Cheung.P,2002)
The following properties hold for any integerm , n∈Z .
a. Dq ( x−a )qn=[ n ]q ( x−a )q
n−1
b. ( x−a )qm+n ≠ ( x−a )q
m ( x−a )qn .
5
c. ( x−a )qm+n=( x−a )q
m ( x−qm a )qn .
d. Dq( 1( x−a )q
n )=[−n ]q ( x−qna )q−n−1
.
e. (−1 )n qn ( n−1)
2 ( x−q−n+1 a )qn=(a−x )q
n .
f. Dq (a−x )qn=[−n ]q (a−qx )q
n−1 .
g. Dq( 1(a−x)q
n )= [ n ]q(a−x)q
n+1 .
By using the above definitions and proposition we eventually come up with the q-Taylor
binomial formula for polynomial (Kac.V, and Cheung.P,2002) as
f ( x )=∑j=0
N
(D¿¿q j f )(c)( x−c)q
j
[ j ] !q(1.21)¿
1.3 q-Exponential Function
Before we study the Euler identity and the q-exponential function, there is need to understand
the concept of Gauss’s binomial formula and Hein’s binomial formula which were both
derived from the q-Taylor binomial formula, in this case we assumed that f ( x )=(x−a)qn.
With x as a variable and using (1.17) of definition 1.2.2 we obtain the Gauss binomial formula
1.3.1 Gauss binomial formula
(x+a)qn=∑
j=0
n
[nk ]qq
(n2)
ak xn−k (1.22)
Where,
[nk ]q=
[ n ]q [n−1 ]q … [n+k−1 ]q[ k ] !q
=[ n ] !q
[ k ] !q [ n−k ] !q
Is the q-binomial coefficient
And the Heine’s binomial formula
1.3.2 Heine’s Binomial Formula
6
1(1−x )q
n =1+∑k=1
∞ [ n ]q [n+1 ]q … [ n+k−1 ]q[ k ] !q
xk(1.23)
(Kac.V, and Cheung.P,2002)
Now considering (1.22) by replacing x and a by 1 and x respectively i.e
(1+x)qn=∑
k=0
n
q(n2 )[nk ]
qx
k
and (1.22)
1(1−x )q
n =∑j=1
∞ [ n ]q [n+1 ]q… [ n+k−1 ]q[k ] !q
xk ,
What will happen if we take the limit of n as n → ∞ in both the expression? Depending on the
value of x, the result is infinitely small or infinitely large so therefore producing not
interesting result in the ordinary calculus i.e when q=1. But in q-calculus it is entirely
different because, an example is, assuming |q|<1 , the infinite product
(1+x)q∞=(1+x ) (1+qx ) (1+q2 x ) …
will eventually converge to some finite limit. Therefore if we let |q|<1 , we have
limn→ ∞
[ n ]q= limn → ∞
1−qn
1−q= 1
1−q(1.24)
and
limn → ∞ (n
k)=limn →∞
(1−qn ) (1−qn−1 ) … (1−qn−k +1 )(1−q)¿¿
¿
¿ 1(1−q)(1−q2)…(1−qk)
(1.25)
So therefore there is difference in the behaviour between the q-analogues of integer and
binomial coefficients for a n larger integer to their ordinary counterparts.
Taking the limits as n → ∞ and substituting (1.24) and (1.25) in the Heine’s and
Gauss’s binomial formula we develop two identities of formal power series in x (with the
assumption that |q|<1∨¿). (Kac.V, and Cheung.P,2002)
1.4.3 q-Euler Identities
(1+x )q∞=∑
k=0
∞
qk ( k−1)
2 xk
(1−q ) (1−q2 ) …( 1−qk )(1.26)
7
1(1−x )q
∞ =∑k=0
∞ xk
(1−q)(1−q2)… (1−qk )(1.27)
and call (1.26) and (1.27) Euler’s first and second identities or E1 and E2 respectively (Kac.V,
and Cheung.P,2002) because he was the one that reveals them at the time of his live before
Gauss’s and Heine. Also the identities relate infinite product and infinite sums but they don’t
have classical analogue because each and every term in the sum don’t have meaning when
q=1.
1.3.4 q-Exponential Functions
Studying those identities helps us to define the q-analogue of the exponential function, but
before then, lets recall the Taylors’s exponential function expansion. i.e
ex=∑k=0
∞ xk
k !(1.28)
From (1.27) dividing both the numerator and the denominator of the R.H.S by 1−q we got
∑k=0
∞ xk
1 (1−q2)1−q
… (1−qk)1−q
=∑k=0
∞ ( x1−q )
k
[ k ] !q(1.29)
Definition: (Kac.V, and Cheung.P,2002) The classical exponential function ex has a q-
analogue as
eqx=∑
k=0
∞ xk
[k ] !q(1.30)
By using (1.29) and (1.30) we get
eqx /(1−q)= 1
(1−x)q∞ ,
or its equivalent
eqx= 1
(1−(1−q) x)q∞ (1.31)
That is the case of E2, we can also use E1 to defined another q-exponential function.
Definition (Kac.V, and Cheung.P,2002)
8
Eqx=∑
k=0
∞
qk (k−1 )
2 xk
[ k ] !q=(1+ (1−q ) x)q
∞ (1.32)
We can relate (1.31) and (1.32) as
1.3.5 Relationship Between eqx and Eq
x
eqx Eq
− x=1(1.33)
From the above property we can say
e1/ qx =∑
k=0
∞ (1−1/q )k xk
(1−1/q ) (1−1/q2 ) … (1−1/qk )
¿∑k=0
∞
qk (k−1)/2 (1−q ) k xk
(1−q ) (1−q2 ) … (1−qk )=Eq
x (1.34)
1.3.6 q-Derivative Of The q-Exponential Functions
And the q-derivative of the two q-exponential function is given as
Dq eqx=∑
k=0
∞ Dq xk
[ k ] !q=∑
k=0
∞ [ k ]q xk−1
[k ] !q=∑
k =0
∞ xk−1
[ k−1 ] !q=¿eq
x (1.35)¿
Dq Eqx=∑
k=0
∞
qk (k−1 )
2 Dq xk
[ k ] !q=∑
k=0
∞
qk ( k−1 )
2 [ k ]q xk−1
[k ] !q=∑
k=0
∞
qk ( k−1)
2 xk−1
[ k−1 ] !q=Eq
qx(1.36)
1.3.7 Convergence Of q-Exponential Functions
The series of non-negative terms in q-calculus converges if a bounded sequence is formed by
its partial sums, so for two classical q- exponential functions we can find interval of
convergence as follows
Let
eqx=∑
k=0
∞ xk
[k ] !q
Then by using De-Alembert theory
limk → ∞| xk +1
[ k+1 ] !q
xk
[ k ] !q|= lim
k → ∞| xk+1
[ k+1 ] !q∙
[ k ] !q
xk |,
9
¿ limk → ∞| x
[ k+1 ]q|,
using [ k+1 ]q=qk+1
q−1
¿ limk → ∞|x(q−1)
qk+1−1 |,¿|x||q−1|<1 ,
Hence converges and the interval of convergence is
|x|< 1¿1−q∨¿ .¿
Similarly we can prove the other q-exponential function as
Let
Eqx=∑
k=0
∞
qk (k−1 )
2 xk
[ k ] !q
By using De-Alembert we see that
limk → ∞|q( k+1 ) k xk+1
[k+1 ] !q
qk (k−1) xk
[ k ] !q|=|q( k+1) k xk+1
[ k+1 ] !q∙
[ k ] !q
qk (k −1 )xk|,
¿ limk → ∞| q (k +1) k x
qk(k−1) [ k+1 ]q|,¿ lim
k → ∞|qk x (q−1)qk+1−1 |=0 .
1.4 q-Trigonometric Functions
By using the well-known Euler formula in terms of exponential function, we can define the q-
analogues of the two trigonometric functions.
Proposition 1.3.1 (Kac.V, and Cheung.P,2002) The sine and cosine q-analogue function are
given by
10
sin q x=eq
ix−eq−ix
2 i, sinq x=
Eqix−Eq
−ix
2i(1.37)
cosq x=eq
ix+eq−ix
2,cosq x=
Eqix +Eq
−ix
2(1.38)
1.4.1 Properties Of q-Trigonometric Functions
We can see from (1.37) and (1.38) that
cosq x cosq x+sin q x sinq x=1(1.39)
1.4.2 q-Derivative Of q-Trigonometric Functions
The q-derivative of the q-trigonometric function is given by
Dq sinq x=cosq x , Dq sinq x=cosq qx (1.40)
Dq sinq x=−sin q x , Dq cosq x ¿−sin q qx(1.41)
And (1.39), (1.40) and (1.41) are being proved by proposition 1.3.1
1.5 Improved q-Exponential Function
There are two exponential functions that are define by Euler in the previous section, both there
are some properties that are lost, for example
eq (−x )= 1Eq ( x )
,Eq (−x )= 1eq ( x )
Which allows us to defined the improved q-exponential function as
1.5.1 Definition ℇ qz
Let ℇ qz be new q-exponential function, and defined as (Jan L. & Cieśliński, 2011)
ℇ qz=eq
z /2 Eqz /2=∏
k=0
∞ 1+(1−q) z2
qk
1−(1−q) z2
qk(1.42)
11
Where eqz∧Eq
z are the standard q-exponential functions. Classical Cayley transformation
motivated the above definition. (the infinite product representation is valid for |q|<1¿.
1.5.2 Basic Definitions on Improved q-Exponential Function
Definition 1.5.2: If a is any real or complex number, then we defined the following terms as
(a ;q )n=∏j=0
n−1
(1−q j a ) ,n∈N (1.43)
(a ;q )0=1(1.44)
{n }=1+q+…+qn−1
12(1+qn+1)
=[n]
12(1+qn+1)
=2(1−qn+1)
(1−q)(1+qn+1)(1.45 )
Therefore
{n }!={1 } {2 }… {n }=[ n ]q ! 2n
(−1 ;q )n
Definition 1.5.2: Bernoulli number can be demonstrated in term of improved q-Bernoulli
number by the following recurrence relation:
∑k=0
n
(nk )q
(−1, q )n−k
2n−k bk ,q−bn , q={1 ,n=10 , n ≠ 1
(1.46)
Where bk ,q is the Bernoulli number.
Definition 1.5.4: (Wikipedia) If x and y are real or complex parameter, then the
summation by Newton expansion in an ordinary case as
( x+ y )n=∑k=0
n
(nk ) xk yn−k .(1.47)
In the same manner, the following q-addition of the expression is define as (Zhang.Z and
JunWang, 2006)
( x⊕q y )n=∑k =0
n
(nk )q
(−1 ,q )k . (−1 , q )n−k
2n xk yn−k , n=0,1,2 ,… (1.48)
1.5.3 Unification Of q-Exponential Functions
The following statement holds true
ℇ qx .ℇ q
y=ℇq ( x⊕q y )
12
Proof
ℇ qx .ℇ q
y=(∑n=0
∞ xn
[ n ] !q
(−1; q)n
2n )(∑m=0
∞ ym
[ m ] !q
(−1 ;q)m
2m )¿∑
n=0
∞ ∞
(∑k=0
n xk
[ k ] !q
(−1 ;q)k
2kyn−k
[ n−k ] !q
(−1;q)n−k
2n−k )…¿∑
n=0
∞ ∞
(∑k=0
n (−1 ;q)k
2k
(−1;q )n−k
2n−kxk
[k ] !q
yn−k
[n−k ] !q )¿∑
n=0
∞ ∞
(∑k=0
n (−1 ;q)k
2k
(−1;q )n−k
2n−kxk
[k ] !q
yn−k
[n−k ] !q )¿∑
n=0
∞ ∞
(∑k=0
n
(nk )q
(−1; q)k .(−1; q)n−k
2n xk yn−k )¿ℇ q ( x⊕ y )
As required.
1.5.4 Improved q-Trigonometric Functions
We can use the natural way to define the new q-sine and q-cosine functions as
s∈¿q x=ℇq
ix−ℇq−ix
2i¿
C osq x=ℇq
ix+ℇq−ix
2(1.49)
1.6 Bernoulli Numbers
In this work Bernoulli numbers will be defined by the exponential generating function
tet−1
=∑n=0
∞
Bnt n
n!(1.50)
We see that the first Bernoulli number is easy to find, i.e.
B0= limt → 0
tet−1
13
¿ limt →0
1e t , L ' Hospital
¿ 1e0
¿1
B1= limt →0
ddt ( t
et−1 )¿ lim
t →0
e t−1−tet
(e t−1)2 , L' Hospital
¿ limt →0
−t2(et−1)
¿ limt →0
−12 e t
¿−12
1.6.1 Recurrence Formula for Ordinary Bernoulli Numbers
Continuing in this way we will use the tool that we have i.e the ordinary exponential function
in order to derive the recurrence formula for Bernoulli numbers
tet−1
=∑n=0
∞
Bnt n
n!
t=(∑n=0
∞ t n
n! )(∑n=0
∞
Bnt n
n! )−∑n=0
∞
Bnt n
n !,
By using the Cauchy product of two series (Rudin, 1964) i.e
Given the two series ∑ an and ∑ bn we write
Cn=∑ ak bn−k (n=0,1,2 , …) ,
Then ∑ cn is said to be the multiplication of the two series.
Going back to our work we see that
(∑n=0
∞
an)(∑n=0
∞
bk)=∑n=0
∞
(∑k=0
n
an bn−k ) ,
14
we obtain
t=∑n=0
∞
(∑k=0
n
Bkt k
k !t n−k
(n−k )! )−∑n=0
∞
Bnt n
n !,
t=∑n=0
∞
(∑k=0
n
Bk (nk) t n
n! )−∑n=0
∞
Bnt n
n !,
t=∑n=0
∞
(∑k=0
n
(nk )Bk ) tn
n !−∑
n=0
∞
Bnt n
n !.
By comparing the power of t we have
∑k=0
n
(nk )B k−Bn={ 1, for n=10 , for others
(1.51)
which is the recurrence formula for Bernoulli numbers.
1.6.2 Kronecker Delta
Proposition 1.6.1 (Riordan, 1968) If Bn is a Bernoulli number number then,
∑k=0
n 1k+1 (nk)Bn−k=δn , 0 ,
Where δ n ,0 is called Kronecker delta
Proof
Prove by Cauchy product on generating function.
Since Kronecker delta is defined as
δ nm={1 , for n=m0 , for n≠m
Then we can write (1.6.2) as
∑k=0
n
(nk )Bk−Bn=δn , 1 ,
Since
(nn)Bn=Bn
when we assume that n−1=m, i.e
∑k=0
n−1
(nk )Bk=δ n ,1 .
15
we will then have
∑k=0
m
(m+1k )Bk=δm , 0 ,
by opening the summation we have
(m+10 )B0+(m+1
1 )B1+…+(m+1m )Bm=δm ,0
∑k=0
∞
( m+1m−k) Bm−k=δm , 0
Why? Because
(m+1m−k )= (m+1 )!
(m−k ) ! (m+1−m+k )!= (m+1 ) m!
(k+1 )!k ! (m−k )!=(m
k ) m+1k+1
Therefore we have
∑k=0
m
(mk ) m+1k+1
Bm−k=δm, 0
¿∑k=0
m
(mk ) 1k+1
Bm−k=(m+1)−1δm , 0={ 1m+1
, for m=0
0 , for m≠ 0
Which is the same as proposition 1.6.1
∑k=0
n 1k+1 (nk)Bn−k=δn , 0 ,
1.6.3 Lemma Explicit Definition of Bernoulli Number (Arakaya.T & et.al, 2014)
Bernoulli number satisfy the recurrence
∑k=0
n
(n+1k )Bk=n+1 , for n≥ 0(1.52)
1.6.4 Proposition Bernoulli Numbers as Rational numbers (Arakaya.T & et.al, 2014)
The Bernoulli numbers are rational numbers.
Solving for the first seven of the Bernoulli numbers using the above recurrence
B0=1 , B1=−12
, B2=16
, B3=0 ,B4=−130
, B5=0 , B6=1
40, …
16
1.6.5 Bernoulli Polynomials
When we multiply the left hand side of (1.51) with ext and rise it to the power of some
arbitrary constant say α a real or complex parameter.
( te t−1 )
α
ext=∑n=0
∞
Bn(α )(x ) t n
n!(1.53)
It is called the generating function for Bernoulli polynomial.
The generalized Bernoulli polynomials are given as
Bn (x )=∑k=0
n
(nk )Bn−k xk (1.54)
Then by using the above expression we obtain the few Bernoulli polynomials:
B0 (x )=1
B1 ( x )=x−12
B2 ( x )=x2−x+ 16
B3 (x )=x3−32
x2
+ 12
x
B4 ( x )=x4−2 x3+x2− 130
B4 ( x )=x5−52
x4
+ 53
x3−16
x
1.6.6 Some Properties Of Bernoulli Polynomials
Proposition1.6.2 (Zhang.Z and JunWang, 2006) for all integers greater than or equal to one,
then the following holds
Bn (x )=∑k=0
n
(nk )Bn xn−k
Proof (Kac.V and Cheung.P, 2002)
Let
17
Fn ( x )=∑k =0
n
(nk)Bn xn−k
It is quiet simple to see that (a) Fn (0 )=Bn , for n≥ 0. (b) Fn' ( x )=nFn−1 ( x ) , for n≥ 1
Since these characters uniquely characterize Bn(x) you see that the (a) is so simple to find out
because
Fn (0 )=(nk )Bk xn+( n
n−1) Bn−1 xn−1+…+(nn)Bn x0
Fn (0 )=0+0+…+(nn)Bn=Bn
But for (b) and using the fact that if n>k≥ 0 ,
Differentiating both side and for n≥ 1, we see that
ddx
Fn
( x )=∑k=0
n−1
(nk )(n−k )Bn xn−k−1
But
(n−k )(nk)= n !
(n−k−1)
Multiplying and dividing by n we get
ddx
Fn ( x )=∑k=0
n−1
n(nk )B k xn−1−k=n∑k=0
n−1
(nk )B k xn−k−1 ,
as required.
Now from the Bernoulli polynomial generating function we deduced the following
proposition:
Proposition (1.6.3) (Zhang.Z and JunWang, 2006) If α and β are two real or complex
parameters then we say that
Bn(α +β ) (x+ y )=∑
k=0
n
(nk )Bn
( α ) ( x ) Bn−k( β ) ( y )(1.55)
Proof
Here we see that there are two polynomial with variables of x∧ y with index α +β multiplying
themselves.
It follows from (1.54)
18
( te t−1 )
α
ext=∑n=0
∞
Bn(α )(x ) t n
n !,
( te t−1 )
β
e yt=∑n=0
∞
Bn( β )( y ) t n
n!,
( te t−1 )
α+ β
e(x+ y)t=(∑n=0
∞
Bn( α )(x) t n
n! )(∑n=0
∞
Bn( β )( y ) t n
n !) ,
¿∑n=0
∞
(∑k=0
n
Bk(α )(x) tk
k ! )(∑n=0
∞
Bn−k( β ) ( y ) tn−k
(n−k )! ) ,
¿∑n=0
∞
(∑k=0
n
(nk )Bn( α )(x)Bn−k
( β ) ( y )) tn
n!,
Therefore we have
∑n=0
∞
Bn( α+β ) ( x+ y ) t n
n !=∑
n=0
∞
(∑k=0
n
(nk )Bn( α )(x )Bn−k
( β ) ( y )) t n
n!,
by comparing the coefficient of t n
n ! ,
we have
Bn(α +β ) (x+ y )=∑
k=0
n
(nk )Bn
( α ) ( x ) Bn−k( β ) ( y ) .
as required.
But
Bn(0 ) ( x )=xn(1.56)
when we interchange x and y in the addition (1.54) and put β=0 the equation yields
Bn(α ) ( x+ y )=∑
k=0
n
(nk )Bn(α ) ( y ) xn−k ,(1.57)
As a special case by putting α=1 we have
Bn (x+ y )=∑k=0
n
(nk )Bn( y )xn−k (1.58)
19
CHAPTER 2
BERNOULLI MATRIX AND SOME PROPERTIES
Majority of the work in this chapter was presented from a Journal Bernoulli matrix and its
algebraic properties (Zhizheng & JunWang, 2006) and any other information was cited by
means of reference.
2.1 Bernoulli Matrix
Before we talk about the Bernoulli matrix, let recall the definition of the algebra matrix that
says:
2.1.1 Definition Ordinary Matrix and some Properties
The rectangular arrangement of numbers is what we referred to as Matrix. A matrix that has m
number of rows and n number of columns is said to be of size m× n and can be displayed as:
A=[ a11 a12 ⋯ a1n
a21 a22 ⋯ a2n
⋮ ⋮ ⋱am1 am 2 ⋯ amn
]=[aij ]
,
where, the entries a ij are real numbers and they can also be complex in some other kind of
matrix.
Matrix has some properties as follows:
Supposes a matrix A=[aij ] and a matrix B=[b ij ] are m× n matrix and α is a scalar with another
matrix C=[c ij ], then,
Matrix addition is defined as: A+B=[aij+bij ]
Matrix subtraction is defined as: A−B=[a ij−bij ]
Scalar multiplication is defined as: αA=[α aij]
20
Multiplication of matrix is defined as: AC=[∑k=1
n
aik cik ],a.nd so many other properties. Now lets go back to the Bernoulli matrix.
2.1.2 Definition Bernoulli Matrix and Bernoulli Polynomials
Supposed Bn is an nth Bernoulli number and Bn (x ) is a Bernoulli polynomial and α∈ R, then
the generalized Bernoulli matrix Bα=[ Bijα ] and Bernoulli polynomial matrix
Bα ( x )= [B i , j(α ) ( x ) ] wherei , j=0,1,2 , …n. are defined respectively as (Zhang.Z and JunWang,
2006)
[B¿¿α ]ij={( ij)B
i− j
α
, if i≥ j
0 , otherwise ,(2.1)¿
[B¿¿i , j ( α ) ( x )]ij={( ij)B
i− j
(α )
( x ) , if i≥ j
0 ,otherwise ,(2.2 )¿
B(1 ) ( x )=B ( x ) is the Bernoulli polynomial matrix, and B (0 )=B is the Bernoulli matrix
Example: the 3×3 Bernoulli polynomial is given by
2.1.3 Theorem Bernoulli Polynomial Matrix of x and y
The following relation for generalized Bernoulli matrix holds true
B(α +β ) (x+ y )=B (α ) ( x ) B( β ) ( y )=B( α ) ( y ) B( β ) ( x )(2.3)
21
[ B (x ) ]3×3=[ 1 0 02 x−1 1 0
3 x2−3 x+ 12
3 x−32
1 ]
Proof
(Zhang.Z and JunWang, 2006)
In general case we need to show that the sum of the two Bernoulli matrices with the index
(α +β ) from the L.H.S is equal to their product at the R.H.S.
Now, the Bernoulli matrix of the L.H.S has the element of the form
[ B(α +β )(x+ y) ]ij={( ij)Bi− j
(α+β ) (x+ y ) , if i≥ j ,
0 , for others .
But on the R.H.S since it has to do with the multiplication of two matrices, therefore the
number of the row of the first Bernoulli matrix has to be the same as the number of column of
the other Bernoulli matrix. So that the elements are
[ B(α )( x)]ik={( ik )B i− j
(α) (x ) , if i≥ k ,
0 , for others .
[ B(β)( y)]kj={(kj)Bi− j
( β) ( y ) ,if k≥ j ,
0 , for others .
Now multiplying those two matrices we see that
( [B(α )(x )] [ B(β )( y) ])ij=∑k=1
n
[ B(α )(x )]ik [ B(β)( y)¿ ]kj .
Opening that summation we get
¿ [ B(α ) ( x ) ]i 1 [ B( β ) ( y ) ]1 j+ [ B( β ) ( y ) ]i2 [ B( β ) ( y ) ]2 j+…+ [B (α ) ( x ) ]ii [B ( β) ( y ) ]ij+[ B( α ) (x ) ]i (i+1) [B ( β) ( y ) ](i+ 1) j+…+ [B( α ) (x ) ]¿ [ B( β ) ( y ) ]nj
On the i index over there up to the point that we are going to reach ii index the result is not
equal to zero but after that where the k index become greater than i index it becomes zero.
These are the zeros of i index, what about the zeros of j index? When ever k< j the result will
be zero up to the point where k= j,
now assume that i≥ j, we have
¿ [ B(α ) ( x ) ]ij [ B( β ) ( y ) ] jj+[ B( β ) ( y ) ]ij+1 [ B( β ) ( y ) ] j+1 , j+…+ [ B( α ) ( x ) ]ii [ B( β ) ( y ) ]ij
¿( ij)Bi− j
α ( x )( jj)B j− j
β ( y )+( ij+1)B i− j−1
α ( x )( j+1j )B1
β ( y )+…+(ii) Bi−iα ( x )( i
j)B i− jβ ( y )
22
All the terms with the i< j becomes zero leaving first and the last term with
( jj)B j− j
β ( y )=(ii) Bi−iα ( x )=1
Now, it follows from (1.6.6), by simplifying the combinatorial
( ij)B i− j
α (x )( jj) B j− j
β ( y )=∑k= j
i
( ik) Bi−k
α (x )(kj) Bk− j
β ( y )
∑k= j
i
( ik )B i−k
α ( x )(kj)Bk− jβ ( y )=∑
k = j
i
( ij)( i− j
k− j)Bi−kα ( x ) Bk− j
β ( y )
¿( ij)∑k=0
i− j
(i− jk )B
i− j−k
α
( x ) Bkβ ( y )
¿( ij)B
i− j
( α+β )
( x+ y ) ,
Which implies (2.3)
2.1.4 Corollary
B( α1+α 2+…+α n ) ( x1+x2+…+ xn)=Bα 1 ( x1 ) Bα2 ( x2 ) …Bαn ( xn ) .(2.4)
Proof Using mathematical induction
And taking x1=x2=…=xn=x ,α 1=α 2=…=α n=α .
we see that for n=1
Bα 1 ( x1 )=Bα1 ( x1) ,
we assume that it is true for n=k,
B( α1+α 2+…+α k ) ( x1+x2+…+xk )=Bα 1 ( x1 ) Bα 2 ( x2 ) …Bαk ( xk ) .
For n=k+1 gives
B( α1 +α 2+…+α k+α k+1 ) ( x1+x2+…+xk+xk+1 )
¿ Bα 1 ( x1 ) Bα2 ( x2 ) …Bαk ( xk ) Bαk +1 ( xk+1 ) .
Therefore the right hand side becomes
B( α1+α 2+…+α k ) ( x1+x2+…+xk ) Bα k+1 ( xk+1 ) ,
which satisfies theorem (2.1).
We can further assume that x=0 or α=1 , then the Bernoulli matrix will have a simple
powers.
2.1.5 Corollary (Zhang.Z and JunWang, 2006)
23
If
(B¿¿α (x))k=B( kα ) (kx ) ,(2.5)¿
and specially,
( B ( x ))k=B(k ) ( kx )
(B¿¿(α ))k=B (kα )¿
Proof:
By applying mathematical induction on α in all the above expression we see that it is true.
2.1.4 Definition Inverse Of Bernoulli Matrix
Let D= [d i , j ] (0 ≤i , j ≤n ) be (n+1 ) × (n+1 ) matrix which is defined as (Zhang.Z and JunWang,
2006)
d i , j={ 1i− j+1 ( i
j)if i ≥ j ,
0 , otherwise .(2.6)
Theorem 2.1.4: Inverse of Bernoulli matrix can be defined by the previous definition of D.
That means
B−1=D .
Also,
(B¿¿(k ))−1=D k¿
Proof
We need to show that D is the inverse of B.
[B]ij={( ij)Bi− j , if i ≥ j ,
0 , otherwise .
Di , j={ 1i− j+1 ( i
j)if i ≥ j ,
0 , otherwise .
Now if we take their matrix we have
[ BD ]i , j={1 ,if i= j0 , if i≠ j
=δ i , j=[ B ] [ D ]=I(n+1 )×(n+1)
which is going to be a of delta kronecker matrix
24
[ BD ]ij=∑k=0
i−1
[B]ik[ D ]kj .
expanding the summation we have
[ [ B ] [ D ] ]i , j=[ B ]i , 0 [ D ]0 ,i+ [ B ]i, 1 [ D ]1 i+… [B ] ii [ D ] ij+…+ [ B ]i ,i+1 [ D ]i+1 j+…+ [ B ]¿−1 [ D ]n−1 j+ [ B ]i , n [ D ]n ,i
Let i= j
[ [ B ] [ D ] ]i , j=[ B ]i , 0 [ D ]0 ,i+ [ B ]i, 1 [ D ]1 i+…+ [ B ]i , n [ D ]n ,i
Treating the two matrices in terms of their component
∑k= j
i
( ik )B i−k
1k− j+1 (k
j )=( ij)∑k= j
i 1k− j+1 ( i− j
k− j )Bi−k ,
¿( ij)∑k=0
i− j 1k+1 (i− j
k )Bi− j−k=( ij)δi− j ,0 ,
Satisfying proposition (1.49)
The result becomes BD=I , i.e., B−1=D . Looking at this result and corollary 2.1.3 we noticed
that
(B¿¿(k ))−1=(B¿¿k )−1=(B ¿¿−1)k=Dk .¿¿¿
as required.(Zhang.Z and JunWang, 2006)
2.2 Bernoulli Matrix and Generalized Pascal Matrix
2.2.1 Definition Pascal Matrix
Supposed x is an unknown variable and n is an integer that is not equal to zero, then the
generalized (n+1)×(n+1) Pascal matrix is denoted as P [ x ]= [ pij ] ,(i , j=0 , 1 , …,n) and
defined as (Zhang.Z and JunWang, 2006)
Pij={ ( ij) x i− j , i≥ j
0 , otherwise .(2.7)
2.2.2 Theorem Relationship Between Bernoulli Polynomial Matrix And Pascal Matrix
For a Bernoulli matrix and Pascal matrix of a non zero real number x
B (x+ y )=P [ x ] B ( y )=P [ y ] B (x )(2.8)
25
Specially (Zhang.Z and JunWang, 2006)
B (x )=P [ x ] B .(2.9)
Proof.
The matrix on the R.H.S in i and j component can be written as
[ B(x+ y) ]ij={( ij) Bi− j ( x+ y ) , if i ≥ j
0 , if i< j
and L.H.S is
[ p(x ) ]ij={( ij)x i− j , if i≥ j
0 , for i< j
[ B( y )]ij={( ij)Bi− j ( y ) , if i≥ j
0 ,if i< j
multiplying the two matrices on the L.H.S
[ P(x )B( y )]ij=∑k=0
i−1
p[ x]ik[B ( y )]kj
expanding the summation
p [ x ]i 0 [B ( y ) ]0 j+ p [ x ]i 1 [B ( y ) ]1 j+…+ p [ x ]ii [B ( y ) ]ij+…
+ p [ x ]i ,i+1 [ B ( y ) ]i+1 j+…+ p [ x ]¿−1 [B ( y ) ]n−1 j+ p [x ]¿ [B ( y ) ]nj
assume i≤ j
p[ x ]ij [B ( y )]jj+ p[ x]i , j+1[B ( y )] j+1 , j+…+ p [ x ]ii [ B ( y ) ]ijBy putting in terms of their respective matrix
( ij) xi− j( i
j)B j− j ( y )+( ij+1) xi− j−1 B1 ( y )+…+( i
i) xi−i( ij)B i− j ( y )
All the terms with i< j component disappear leaving the ones with i≥ j
Since for i= j we have
(ii)x i−i=( jj)B j− j ( y )=1
∑k= j
i
( ij) xi− j( j
j )B j− j ( y )=∑k = j
i
( ik )x i−k (k
j) Bk− j ( y )
26
¿( ij)∑k= j
i
( i− jk− j)Bk− j ( y ) x i−k
¿( ij)∑k=0
i− j
(i− jk )Bk ( y) x i− j−k=( i
j)B i− j(x+ y) ,
Which give B (x+ y )=P(x)B ( y ) .
In the same manner we can obtain the other part of (2.8). i.e.
B (x+ y )=P(x)B ( x )
Proof.
The matrix on the R.H.S in i and j component can be written as
[ B(x+ y) ]ij={( ij) Bi− j ( x+ y ) , if i ≥ j
0 , if i< j
and L.H.S is
[ p[ y ]]ij={( ij) y i− j ,if i ≥ j
0 , for i< j
[ B( y ) ]ij={( ij)Bi− j ( x ) ,if i ≥ j
0 , if i< j
multiplying the two matrices on the L.H.S
[ p [ y ] B (x)]ij=∑k=0
i−1
p [ y ]ik [B ( x )]kj
expanding the summation
p [ y ]i 0 [B ( x ) ]0 j+ p [ y ]i 1 [B ( x ) ]1 j+…+ p [ y ] ii [B ( x ) ]ij+…
+ p [ y ]i , i+1 [B ( x ) ]i+1 j+…+ p [ y ]¿−1 [B ( x ) ]n−1 j+ p [ y ]¿ [ B ( x ) ]nj
assume i≤ j
p[ y ]ij [B ( x )]jj+ p[ y ]i , j+1[B ( x )] j+1 , j+…+ p [ y ]ii [B ( x ) ]ijby putting in terms of their respective matrix
( ij) y i− j( i
j)B j− j ( x )+( ij+1) y i− j−1 B1 ( x )+…+( i
i) y i−i( ij )Bi− j ( x )
all the terms with i< j component disappear leaving the ones with i≥ j
since for i= j we have
27
(ii) y i−i=( jj)B j− j ( x )=1
by considering the terms where i> j we see that by using (1.6.8),
∑k= j
i
( ij) y i− j( j
j)B j− j ( x )=∑k = j
i
( ik ) y i− k(k
j)Bk− j ( x )
¿( ij)∑k= j
i
( i− jk− j)Bk− j ( x ) y i−k
¿( ij)∑k=0
i− j
(i− jk )Bk (x) y i− j−k=( i
j)B i− j (x+ y )
which give B (x+ y )=P [ y ] B ( x ) .In the same manner, we can obtain B (x+ y )=P [ x ] B ( y ) .
Example
B( x )=(1 0 0 0 ⋯x 1 0 0 ⋯
x2−x+ 16
2 x−1 1 0 ⋯
x3−32
x2+12
x 3 x2−3 x+ 12
3 x−32
1 ⋯
⋯ ⋯ ⋯ ⋯ ⋯)(n+1)×(n+1)
=(1 0 0 0 ⋯x 1 0 0 ⋯x2 2 x 1 0 ⋯x3 3 x2 3 x 1 ⋯⋯ ⋯ ⋯ ⋯ ⋯
)(n+1)×(n+1)(
1 0 0 0 ⋯
−12
1 0 0 ⋯
16
−1 1 0 ⋯
0 12
−32
1 ⋯
⋯ ⋯ ⋯ ⋯ ⋯
)(n+1)×(n+1)
¿ P [ x ] B.
2.2.3 Theorem Inverse Of Bernoulli Polynomial Matrix and Pascal Matrix
B−1 ( x )=B−1 [−x ]=DP [−x ] (2.10)
(Zhang.Z and JunWang, 2006)
Proof:
Lets try and see how it operate before getting into the details of the proof
28
Pascal matrix is given as
[ P[ x] ]ij={( ij)x i− j , if i≥ j
0 , for i< ji.e
P[x]=
(1 0 0 0 ⋯x 1 0 0 ⋯x2 2 x 1 0 ⋯x3 3 x2 3 x 1 ⋯⋯ ⋯ ⋯ ⋯ ⋯
)and its inverse as
[ P[ x ]−1 ]ij={( ij)(−x )i− j ,if i ≥ j
0 , for i< jI.e
P [ x ]−1
=
(1 0 0 0 ⋯
−x 1 0 0 ⋯x2 −2 x 1 0 ⋯
−x3 3 x2 −3 x 1 ⋯⋯ ⋯ ⋯ ⋯ ⋯
)multiplying them we have
P[x]P[-x] =
(1 0 0 0 ⋯x 1 0 0 ⋯x2 2 x 1 0 ⋯x3 3 x2 3 x 1 ⋯⋯ ⋯ ⋯ ⋯ ⋯
)(1 0 0 0 ⋯
−x 1 0 0 ⋯x2 −2 x 1 0 ⋯
−x3 3 x2 −3 x 1 ⋯⋯ ⋯ ⋯ ⋯ ⋯
)
=(1 0 0 0 ⋯0 1 0 0 ⋯0 0 1 0 ⋯0 0 0 1 ⋯⋯ ⋯ ⋯ ⋯ ⋯
)=I
There are three cases involved
first
29
assume i< j
∑k=0
n
P [x ]ik P [−x ]kj=P[ x ]i 0 P[−x ]0 j+P [x ]i 1 P [−x ]1 j+…
+P [ x ]ii P [−x ]ij+P [ x ]i ,i+1 P [−x ]i+ 1, j+…+P [x ]¿ P[−x ]nj
¿0+0+0+…+0=0 ,
Second
assume i= j
¿ P[ x ]i 0 P[−x ]0 j+P [x ]i 1 P [−x ]1 j+…
+P [ x ]ii P [−x ]ij+P [x ] i ,i+1 P [−x ]i+ 1, j+…+P [ x ]¿P [−x ]nj
¿0+(ii)(x )i−i( jj)(−x) j− j+0=0+1+0+…=1 ,
The last case,
assume that i> j
∑k= j
i
( ik ) ( x )i−k (k
j)(−x )k− j=∑k= j
i
( ik )(kj) ( x )
i−k
(−1 )k− j ,
changing the boundary of summation we get
¿∑m=0
i− j
( ij)
( i− j ) (i− j−1 ) …(i− j−m)m!
x i− j(−1)m ,
¿( ij)x
i− j
∑m=0
i− j (i− j ) !(i− j−m )!m !
(−1)m,
¿( ij)x
i− j
∑m=0
l l !( l−m )! m!
(−1)m ,
¿( ij)x
i− j
∑m=0
l
( lm)(−1)m .
Now from Newton expansion formula of binomial that says
(a+b )n=∑k=0
n
(nk)ak bn−k=∑
k=0
n
(nk ) (−1 )n−k=0
If a=1 , b=−1 or a=−1 , b=1 , then
the final expression becomes
¿( ij)x
i− j
∑m=0
l
( lm) (1−1 )m=0
30
According to Theorem 2.2.1, we have
B (x+ y )=P [ x ] B ( y ) ,
if y=0, we have
B (x )=P [ x ] B ,
According to the discussion that we have above that
(P [ x ])−1=P [−x ] ,
multiplying from the left side by P[−x ]
P [− x ] B ( x )=B ,
now multiplying by B−1 ( x )
P [− x ]=B B−1 (x ) ,
B−1 ( x )=DP [−x ] .
Since B−1=D
Example
B−1( x )=(1 0 0 0 ⋯x 1 0 0 ⋯
x2−x+ 16
2 x−1 1 0 ⋯
x2− 32
x2+ 12
x 3 x2−3 x+ 12
3 x−32
1 ⋯
⋯ ⋯ ⋯ ⋯ ⋯)(n+1 )×(n+1)
−1
=(1 0 0 0 ⋯12
1 0 0 ⋯
13
1 1 0 ⋯
14
1 32
1 ⋯
⋯ ⋯ ⋯ ⋯ ⋯
)(n+1)×(n+1)
×(1 0 0 0 ⋯
−x 1 0 0 ⋯x2 −2 x 1 0 ⋯
− x3 3 x2 −3 x 1 ⋯⋯ ⋯ ⋯ ⋯ ⋯
)(n+1)×(n+1)
¿ DP [− x].
31
CHAPTER 3
Q-BERNOULLI MATRIX AND ITS PROPERTIES
Majority of the work in this Chapter was presented from a journal ‘Q-Bernoulli Matrices and
Their Some Properties’ (Naim and KUS, 2015) and any other information was cited by means
of reference.
3.1 Q-Bernoulli Matrix
Having studying the basic concept of q-calculus in Chapter 1 and Bernoulli matrix with some
of its properties in Chapter 2 of this theses, we now continue to see how the q-Bernoulli matrix
is being defined by using the q-Bernoulli polynomials and then some other properties of it,
3.1.1 Definition q-Bernoulli numbers
For integer n ≥ 1 and the Bernoulli numbers Bn. The q-Bernoulli numbers Bn (q ) are defined as
(Naim and KUS, 2015)
Bn (q )=Bn[ n ]q!n !
(3.1 )
By using the above definition we see that the first six q-Bernoulli numbers are:
B0 (q )=B0[ 0 ]q!0 !
=1 ,
B1 ( q )=B1[ 1 ]q !1!
=−12
,
B2 ( q )=B2[2 ]q !2!
=[ 2 ]q!12
,
B3 (q )=B3[3 ]q !3 !
=0 ,
B4 (q )=B4[ 4 ]q !4 !
=[ 4 ]q!720
,
B5 (q )=B5[5 ]q !5 !
=0 ,
32
B6 (q )=B6[ 6 ]q!6 !
=[6 ]q!
30240.
3.1.2 Definition q-Bernoulli Polynomials
The q-Bernoulli polynomials Bn (x , q ) as is defined as (Naim and KUS, 2015)
Bn (x , q )=∑k=0
n
(nk )
qB
n(q ) xn−k (3.2)
The first six q-Bernoulli polynomials also can be seen to be:
B0 (x ,q )=1 ,
B1 ( x , q )=x−12
,
B2 ( x , q )=x2−x+[ 2 ]q!12
,
B3 (x , q )=x3−3 x2
2+
[ 2 ]q! x4
,
B4 ( x ,q )=x4−2x3+[2 ]q x2
2+
[ 4 ]q!720
,
B5 (x , q )=x5−5 x4
2+
5 [ 2 ]q x3
6+
[4 ]q ! x144
.
Theorem 3.1.3 (Naim and KUS, 2015) When we apply the commutativity property on x and
y such that xy=qyx, then
Bn (x+ y , q )=∑k=0
n
(nk )
qyn−k Bk ( x , q ) ,(3.3)
with the same approached on the theorem we get
Bn (x+ y , q )=∑k=0
n
(nk )
qxn−k Bk ( y ,q )(3.4)
3.1.3 Definition q-Bernoulli Matrix
If Bn (q ) is an nth q-Bernoulli number, then the (n+1 ) × (n+1 ) q-Bernoulli matrix Bij (q )=[b ij (q ) ] is defined as (Naim and KUS, 2015)
33
Bij (q )={( ij)q
b i− j (q ) , if i≥ j
0 , otherwise(3.5)
where 0≤ i, j ≤ n .
For example
B(q )=(1 0 0 0 0
−12
1 0 0 0
[2 ]q2⋅3 !
[ 2 ]q2! 1 0 0
0[ 3 ]q [2 ]q
2⋅3 ![ 3 ]q2 ! 1 0
[ 4 ]q30⋅4 !
0[ 4 ]q [ 3 ]q
2⋅3 ![ 4 ]q2!
1)
3.1.4 Definition q-Bernoulli Polynomials Matrix
Suppose Bn (x , q ) is a q-Bernoulli polynomial. The q-Bernoulli polynomial matrix as (Naim
and KUS, 2015)
Bij ( x , q )={( ij)q
Bi− j ( x , q ) , if i≥ j
0 , otherwise(3.6)
3.1.5 Theorem Inverse of q-Bernoulli Matrix
Let D (q )=[d ij (q ) ] be (n+1 ) × (n+1 ) matrix that is defined as (Naim and KUS, 2015)
d ij ( q )={(nk )q
[ i− j ]q !(i− j+1 )!
, if i≥ j
0 , otherwise(3.7 )
Then D (q ) is called the inverse of the q-Bernoulli matrix.
Proof (Naim and KUS, 2015) If B (q ) is the q-Bernoulli matrix, and using the definition of
D (q ) above, then
34
to prove this there is need to show that the multiplication [ B (q ) D(q)] is equal to q-kronecker
delta.
Now
[ B (q ) ]ik={( ik)q
Bi−k (q ) , if i≥ k
0 , otherwise
[ D (q ) ]kj={(kj)q
[ k− j ]q!(k− j+1 )!
, if k ≥ j
0 , otherwise
Putting them in their matrix form [ B (q ) D(q)]ij=∑k=0
n
[B (q ) ]ik [ D (q ) ]kj
expanding the summation we have
[ B (q ) ]i 0 [D (q ) ]0 j+[ B (q ) ]i 1 [D (q ) ]1 j+…+[ B (q ) ]¿ [ D (q ) ]nj ,
assume that i≤ j
[ B (q ) ]ij [ D (q ) ] jj+ [B (q ) ]i , j +1 [ D (q ) ] j+1 , j+…+ [B (q ) ]ii [D (q ) ]ij ,
( ij)q
Bi− j (q )( jj)q
[ j− j ]q!( j− j+1 ) !
+( ij+1)
qB i− j−1 (q )( j+1
j )q
[1 ]q !(2 ) !
+…+(ii)qBi−i (q )( i
j)q
[i− j ]q !( i− j+1 ) !
,
All the component with i< j becomes zero leaving the terms with
(ii)q=( j
j)q=1
Then by considering the terms which are not equal to zero and one
[ B (q ) D(q)]ij=∑k=0
n
Bik D kj(q) ,
¿∑k= j
i
( ik )
qBi− k(q)(kj)q
[ k− j ]q !(k− j+1 )!
,
¿∑k= j
i
( ik )
q(kj)q
[ k− j ]q!(k− j+1 )!
Bi−k (q ) ,
¿( ij)q∑k= j
i
( i− jk− j)q
[k− j ]q !( k− j+1 )!
bi−k (q ) ,
Let t=k− j
35
¿( ij)q
∑k= j
i
(i− jt )
q
[t ]q !( t+1 ) !
bi−k (q ) ,
Shifting the summation to start from t=0 to i− j
¿( ij)q∑t=0
i− j
(i− jt )
q
[t ]q !( t+1 ) !
bi− j−t (q ) ,
and multiplying by [i− j−t ]q !(i− j−t )!
¿( ij)q∑t= j
i− j
(i− jt )
q
[t ]q !( t+1 ) !
Bi− j−t[ i− j−t ]q !( i− j−t ) !
,
¿( ij)q
[ i− j ]q !(i− j )! ∑t= j
i− j
(i− jt )
q
1(t+1 ) !
Bi− j−t ,
By using the orthogonality relation for Bernoulli numbers i.e Proposition 1.6.1
gives
[ B (q ) D (q 0 ) ]ij=( ij )q
[ i− j ]q !(i− j ) !
δi− j ,0=δi , j .(3.8)
And (3.8) is of Kronecker delta.
3.2 q-Bernoulli Matrix and q-Pascal Matrices
3.2.1 Definition Pascal Matrix and Inverse of Pascal Matrix
The generalized (n+1 ) × (n+1 ) q-Pascal matrix P ( x , q )=[ pij (q)] is defined as (Naim and KUS,
2015)
pij ( q )={( ij)q
x i− j , if i≥ j
0 , otherwise(3.9)
and the inverse of the generalized q-Pascal matrix P−1 ( x , q ) [ pij' (q ) ] as
Pij' (q )={( i
j)qq(i− j
2 )(−x )i− j ,if i ≥ j
0 , otherwise .
(3.10)
Now, the factorization of q-Pascal matrix can be generalized by the following theorem
36
3.2.2 Theorem Relationship Between q-Bernoulli Polynomial Matrix and Pascal Matrix
(Naim and KUS, 2015) Supposed the q-Bernoulli polynomial matrix B (x , q ) and the
generalized q-Pascal matrix P ( x , q ), then
B (x+ y , q )=P ( y , q ) B ( x , q )=P ( x , q ) B ( y , q )(3.11)
the interchanging occur as a result of commutative property
And specially
B (x , q )=P (x , q ) B (q )(3.12)
Proof (Naim and KUS, 2015) Consider B (x , q ) as the q-Bernoulli polynomial matrix and
P ( y ,q ) as generalized q-Pascal matrix. Then we see that
( P ( y , q ) B ( x ,q ) )ij=∑k=0
n
Pik (q ) Bkj ( x ,q )=∑k=0
n
[P ( y , q )]ik [B ( x , q )]kj ,
Expanding
[ P ( y , q ) ]i 0 [ B ( x , q ) ]0 j+[ P ( y , q ) ]i 1 [B ( x ,q ) ]1 j+…+ [P ( y , q ) ]¿ [ B ( x , q ) ]nj ,
Supposed i≤ j
[ P ( y , q ) ]ij [B ( x ,q ) ] jj+[ P ( y , q ) ]i , j+1 [B ( x ,q ) ] j+1 , j+…+[ P ( y ,q ) ]ii [B ( x ,q ) ]ij ,by expressing them in their matrix form:
( ij)q
y i− j( jj)q
B j− j (x ,q )+( ij+1)
qy i− j−1( j+1
j )qB1 ( x ,q )+…+(ii)q
y i−i( ij)q
B i− j (x ,q ) ,
(ii)q=( j
j)q=1 ,
and
i< j →0
now
( P ( y ,q ) B ( x ,q ) )ij=∑k=0
n
Pik (q ) B kj ( x , q ) ,
¿∑k= j
i
( ik )
qy
i−k
(kj)qB
k− j( x ,q ) ,
¿∑k= j
i
( ij)q
( i− jk− j)q
yi−k
Bk− j ( x ,q ) ,
let t=k− j
37
¿∑k= j
i
( ij)q
(i− jt )
qy
i−k
Bt (x , q ) ,
shifting the summation to start from t=0¿ i− j
¿( ij)q
∑t=0
i− j
(i− jt )
qy
i− j−t
Bt ( x , q ) ,
comparing it with (3.3 ), we have the equivalent
( P ( y ,q ) B ( x , q ) )ij=( ij)q
Bi− j
( x+ y ,q )= (B ( x+ y , q ) )ij ,
Similarly we can obtain the second part of (3.2.3). i.e.
B (x+ y , q )=P (x ,q ) B ( y ,q ) .
Proof (Naim and KUS, 2015)
( P ( x ,q ) B ( y , q ) )ij=∑k=0
n
Pik (x , q ) Bkj ( y ,q )=∑k=0
n
[ P (x ,q )]ik[B ( y , q )]kj
Opening the summation
[ P ( x ,q ) ]i 0 [B ( y , q ) ]0 j+[ P ( x ,q ) ]i 1 [B ( y , q ) ]1 j+…+ [P ( x , q ) ]¿ [B ( y ,q ) ]nj
Supposed i≤ j
[ P ( x , q ) ]ij [B ( y ,q ) ] jj+ [ P ( x , q ) ]i , j+1 [ B ( y ,q ) ] j+1 , j+…+ [ P ( x , q ) ]ii [ B ( y ,q ) ]ijExpressing them in their matrix form:
( ij)q
x i− j( jj)q
B j− j ( y , q )+( ij+1)
qxi− j−1( j+1
j )qB1 ( y ,q )+…+(ii)q
x i−i( ij)q
Bi− j ( y , q )
(ii)q=( j
j)q=1
and
i< j →0
Now
( P ( x , q ) B ( y ,q ) )ij=∑k=0
n
Pik (x , q ) Bkj ( y , q ) ,
¿∑k= j
i
( ik )
qx
i−k
(kj )q
Bk− j
( y ,q ) ,
¿∑k= j
i
( ij)q
( i− jk− j)q
xi−k
Bk− j ( y ,q ) ,
Let t=k− j
38
¿∑k= j
i
( ij)q
(i− jt )
qx
i−k
Bt ( y , q ) ,
Shifting the summation to start from t=0¿ i− j
¿( ij)q
∑t=0
i− j
(i− jt )
qx
i− j−t
Bt ( y , q ) ,
Comparing it with (3.3 ), we have the equivalent
( P ( x , q ) B ( y , q ) )ij=( ij)q
Bi− j
( x+ y ,q )= (B ( x+ y , q ) )ij .
and also poof (3.12) in similar way with (3.11). i.e.
B (x , q )=P (x , q ) B (q ) .
Proof. (Naim and KUS, 2015)
[ P ( x ,q ) B(q) ]ij=∑k=0
n
[P ( x , q ) ]ik [ B (q ) ]kj ,
expanding the summation we have
[ P ( x , q ) ]i 0 [B ( q ) ]0 j+[ P ( x , q ) ]i 1 [B ( q ) ]1 j+…+ [P ( x ,q ) ]¿ [B ( q ) ]nj ,
assume that i≤ j
[ P ( x ,q ) ]ij [B (q ) ] jj+[ P ( x , q ) ]i , j+1 [B (q ) ] j+1 , j+…+[ P ( x ,q ) ]ii [ B (q ) ]ij ,writing them in their matrix form
( ij)q
x i− j( jj)q
B j− j (q )+( ij+1)q
x i− j−1( j+1j )
qB1 (q )+…+(ii)q
x i−i( ij)q
Bi− j (q ) ,
all the terms with the combination where i< j will tends to zero leaving
(ii)q=( j
j)q=1 ,
then the other part will be
( P ( x , q ) B (q ) )ij=∑k=0
n
Pik (q ) Bkj (q ) ,
¿∑k= j
i
( ik )
qx
i−k
(kj )q
Bk− j
( q ) ,
¿∑k= j
i
( ik )
q(kj)q
xi−k
Bk− j ( q ) ,
¿( ij)q∑k= j
i
( i− jk− j)q
xi−k
Bk− j (q ) ,
39
¿( ij )q
∑k= j
i
(i− jk )
qx
i−k
Bk (q ) ,
Shifting the summation to start from k=0¿ i− j
¿( ij)q∑k=0
i− j
(i− jk )
qx
i−k
Bi− j−k ( q ) ,
¿( ij)q
Bi− j
( x ,q ) ,
¿ ( B ( x , q ) )ij .
Example
3×3 q-Bernoulli polynomial matrix
( P ( y , q ) B ( x , q ) )ij
=
( 1 0 0y 1 0y2 [ 2]q y 1 )×(
1 0 0
x−12 1 0
x2−[ 2 ]q
2x+
[2 ]q12
[2 ]q−[2 ]q
21 )
=(1 0 0
(x+ y )−12 1 0
x2+[2 ]q xy+ y2−[2 ]q
2( x+ y )+
[2 ]q12
[2 ]q ( x+ y )−[2 ]q
21 )
¿ B (x+ y ,q )
( P ( x ,q ) B (q ) )ij=¿
(1 0 0 0x 1 0 0x2 [2 ]q x 1 0x3 [3 ]q x2 [3 ]q x 1 )×(
1 0 0 0
−12
1 0 0
[2 ]q2⋅3 !
[2 ]q2 !
1 0
0[2 ]q [ 3 ]q
2⋅3 ![ 3 ]q2 !
1)
40
=(1 0 0 0
x−12
1 0 0
x2−[2 ]q
2x+
[2 ]q12
[ 2 ]q x−[2 ]q
21 0
x3−[3 ]q
2x2+
[2 ]q [ 3 ]q12
x [ 3 ]q x2−[2 ]q [ 3 ]q
2+[ 2 ]q [3 ]q
12[ 3 ]q x−
[3 ]q2
1)
¿ B (x , q )
3.2.3 Definition Inverse of q-Bernoulli Polynomials Matrix
(Naim and KUS, 2015) If B (x , q ) is a q-Bernoulli polynomial matrix, then B−1 ( x ,q )=[c ij (q ) ], where
c ij ( q )={ [ i ]q ![ j ]q !∑t=0
i− j q( t2)(− x)t
[ t ]q! (i− j−t +1 )!, if i≥ j
0 , otherwise .
,(3.13)
3.2.4 Corollary
(Naim and KUS, 2015) Let P ( x , q ) be the generalized q-Pascal matrix and B (q ) be q-Bernoulli matrix. Then we can use the factorization of B (x , q ) in (3.12)
B−1 ( x , q )=B−1 (q ) P−1 ( x ,q )=D ( q ) P−1 ( x , q ) .
With the inverse of generalized q-Pascal matrix (3.10), and considering the R.H.S
[ D (q ) ]ik={( ik )
q
[ i−k ]q !( i−k+1 )!
, if i≥ k ,
0 , otherwise .
Pkj' (q )={(k
j)qq(k− j
2 )(−x )k− j , if k ≥ j ,
0 , otherwise .
[ D (q ) P−1(x , q)]ij=∑k=0
n
[ D (q ) ]ik [ P−1(x ,q)]kj ,
¿ [ D (q ) ]i 0 [ P−1(x ,q) ]0 j+[ D (q ) ]i 1 [ P−1(x , q)]1 j+…+ [D (q ) ]¿ [ P−1 (x ,q ) ]nj ,
assume that i≤ j
41
¿ [ D (q ) ]ij [P−1(x ,q)] jj+[ D (q ) ]i , j+1 [ P−1(x , q)] j+1 , j+…+[ D (q ) ]ii [ P−1 ( x , q ) ]ij ,
¿( ij )q
[ i− j ]q !( i− j+1 )! ( j
j)qq
( j − j2 ) (−x ) j− j+( i
j+1)q
[ i− j−1 ]q !(i− j )! ( j+1
j )qq
(12 )(−x )1+¿
…+[ i−i ]q!
(1 )! ( ij)q
q(i− j
2 ) (−x )i − j ,
but for
(ii)q=( j
j)q=1 ,∧i< j=0
considering the other part
(D(q)P−1( x , q))ij=∑k =0
n
Dik(q)Pkj' (q) ,
¿∑k= j
i
( ik )
q
[ i−k ]q !(i−k+1 )! (k
j )q(−x)
k− j
q(k− j
2 ),
¿∑k= j
i
( ik )
q(kj)q
[ i−k ]q! q(k− j
2 )
( i−k+1 )!(−x )k− j ,
¿∑k= j
i
( ij)q
( i− jk− j)q
[i−k ]q !q(k− j
2 )
( i−k+1 ) !(−x)k− j ,
¿( ij)q∑k= j
i
( i− jk− j)q
[i−k ]q !q(k− j
2 )
( i−k+1 ) !(−x)k− j ,
Let t=k− j
¿( ij)q∑k= j
i
(i− jt )
q
[ i−k ]q! q(t2 )
( i−t +1 ) !(−x )t ,
Shifting the summation to start at t=0¿ i− j
¿( ij)q∑t=0
i− j
(i− jt )
q
[ i− j−t ]q! q(t2 )
( i− j−t+1 ) !(−x )t ,
¿[ i ]q ![ j ]q !∑t=0
i− j q( t2)(−x )t
[ t ]q! ( i− j−t +1 )!,
¿c ij (q ) .
42
CHAPTER 4
IMPROVED Q-BERNOULLI MATRIX AND ITS PROPERTIES
In this chapter, together with the knowledge that we obtained in the previous Books and
Journals, we used some properties in other related materials as they are been cited by means of
reference and at the end we developed another form of Bernoulli matrix called ‘ THE
IMPROVED BERNOULLI MATRIX’ where the Bernoulli polynomials are generated with
the improved q-exponential function. But before then lets us give a brief history of q-Bernoulli
numbers.
4.1 History of q-Bernoulli Numbers
Carlitz was the first person to studied the q-analogue of Bernoulli numbers together with
Bernoulli polynomials in the middle of last century where he introduced a new sequence as
{βn }n≥ 0 , and relationship between Bernoulli polynomials and Euler polynomials are been
proved in (H. M. Srivastava & Pint´er, 2004) . and they also presented the generalized
polynomials. Properties of Genocchi polynomials and Euler polynomials are been investigated
by kim et al. in (T., 2006)- (Kim, 2007), some recurrence relationship are also given there, the
q-extension of Genocchi numbers are presented in different manner in (Cenkci & et.al, q-
extensions of Genocchi numbers, 2006) by Cenkci et al. The new concept of the q-Genocchi
number and polynomials are presented by Kim in (Kim, 2007). In (Cenkci & et.al, q-
extensions of Genocchi numbers, 2006), The q-Genocchi zeta function and l−¿ function
through the use of generating functions and Mellin transformation are been discuss by Simsek
et al. in (Simsek & et.al, 2008), There so many recent interesting research on this related area
by so many authors as in: Kurt V. (Kurt V. , 2014), Gabuarry and Kurt B. (Gabaury & Kur,
2012) , Kurt in (Kurt & et.al, 2013), Srivastava in (Srivastava & et.al, 2004), (Srivastava &
Vignat, 2012), Choi in (Choi & et.al, 2008), Nalci and Pashaev in (Nalci & Pashaev, 2012),
Luo in (Luo, 2010), and Srivastava in (Srivastava & Luo, 2006), (Srivastava & Luo, 2011),
(Cenkci & et.al, 2008), and Cheon in (Cheon, 2003).
43
4.1.1 Definition Carlitz q-Bernoulli Number
We first present here the initial recurrence q-Bernoulli number by Carlitz as:
∑k=0
n
(nk ) βk qk+1−βn={1 , n=10 , n>1
From the Bernoulli generating function. i.e
tet−1
=∑n=0
∞
Bnt n
n!
There are a lot definition of the quantum form of Bernoulli number and the Bernoulli
polynomials, we find their differences according to their application, for example we can
defined it by generating function, so because we have a several types of quantum exponential
function, so we have the several types of the q-Bernoulli numbers as well, we can also defined
it arbitrary like its been done in the previous chapter.
4.2 Improved q-Bernoulli Numbers
In this chapter, the classical definition of quantum calculus concept will be used, by recalling
(1.41) of Definition 1.4.2 we can have the following lemma.
4.2.1 Lemma Recurrence Formula For Improved q-Bernoulli Number
We can equivalently define (1.41) by means of the generating function as:
xεq ( x )−1
=∑n=0
∞
bn ,qxn
[n]q !(4.1)
Proof
Let
xεq ( x )−1
=∑n=0
∞
bn ,qxn
[n]q !
Cross and multiply
x=[ εq ( x )−1 ]∑n=0
∞
bn , qxn
[ n ]q !
44
¿∑n=0
∞
bn ,qxn
[n]q !. εq ( x )−∑
n=0
∞
bn ,qxn
[n ]q !
¿∑n=0
∞
bn ,qxn
[n]q !.
(−1 , q )n2n
xn
[n]q!−∑
n=0
∞
bn ,qxn
[n ]q !
By using the Cauchy product of series we have
¿∑n=0
∞
(∑k=0
n
bk ,qxk
[k ]q !.
(−1 , q )n−k
2n−kxn−k
[n−k ]q! )−∑n=0
∞
bn , qxn
[n ]q !
¿∑n=0
∞
(∑k=0
n
[nk ]q bk ,q(−1 , q )n−k
2n−k ) xn
[n]q!−∑
n=0
∞
bn ,qxn
[n ]q !
¿∑n=0
∞
(∑k=0
n
[nk ]q(−1, q )n−k
2n−k . bk , q−bn ,q) xn
[n]q !
By comparing the power of x, we get
¿∑k=0
n
[nk ]q
(−1 ,q )n−k
2n−k . bk ,q−bn ,q={1, n=10 , n≠1
¿∑k=0
n−1
[nk ]q
(−1 ,q )n−k
2n−k bk ,q={1 , n=10 , n≠ 1
(4.2)
Which is the recurrence formula for the improved q-Bernoulli number as required.
By using the expression in (4.2) we can have the first few improved q-Bernoulli numbers as:
b0 ,q=1
b1, q=−12
=−1{2 }q
b2, q=14
q(q+1)q2+q+1
=q [2]q
4 [3 ]q
b3 ,q=0
4.2.3 Lemma Advantage Of Improved q-Exponential Function
All the coefficient of the improved quantum Bernoulli numbers are zero except the initial one.
i.e
bn ,q=0 , for n=2 p+1 , ( p∈ N )
Proof
Let
45
f ( x )=∑n=0
∞
bn , qxn
[ n ]q !
Subtracting the first term in the above expression
¿∑n=0
∞
bn ,qxn
[n ]q!−b1 , q x
But b1, q=−12
¿ xεq ( x )−1
+ x2
¿x2 ( 2
εq ( x )−1+1)
¿ x2 ( ε q ( x )+1
εq ( x )−1 )We assume that f ( x ) is an even function, i.e f ( x )=f (−x ) .
Now,
f (−x )=−x2 ( εq ( x )+1
εq ( x )−1 )From the definition of ε q (x )=Eq( x
2 )eq ( x2 ) implies that
ε q (−x )=Eq(−x2 )eq(−x
2 )=eq−1(−x
2 )Eq−1(−x
2 )=εq−1 (x )
Therefore,
f (−x )=−x2 (
1εq ( x )
+1
1ε q (x )
−1 )Multiplying by
εq ( x )εq ( x ) , we have
¿− x2 (
1εq ( x )
+1
1εq (x )
−1 ) εq ( x )εq ( x )
46
¿− x2 (
εq ( x )εq ( x )
+ε q ( x )
εq (x )εq (x )
−εq ( x ) )¿− x
2 ( 1+εq ( x )1−ε q ( x ) )
¿ x2 ( ε q ( x )+1
εq ( x )−1 )=f ( x )
As required.
And the previous lemma is one of the advantages of the improved q-exponential function over
the ordinary q-exponential function.
4.2.4 Improved q-Bernoulli Polynomials
We can also use the means of generating function to defined the improved q-Bernoulli
polynomials Bn ,qα ( x ) as
( tεq ( t )−1 )
α
ε q ( xt )=∑n=0
∞
bn , qα (x) t n
[ n ]q !(4.3)
Where α is a real or complex parameter.
We can observed that, Bn(1) ( x )=Bn (x ) and Bn
(1) (0 )=Bn are the classical improved Bernoulli
polynomials and classical Bernoulli numbers, respectively.
The Bernoulli polynomials can also be defined with respect to x and y as
tεq ( t )−1
∙ εq ( xt ) ∙ ε q ( yt )=∑n=0
∞
Bn ,q ( x , y ) t n
[n ]q !(4.4)
4.2.5 Theorem Additive Theory
supposed x , y∈C , then
Bn ,q ( x , y )=∑k=0
n
[nk ]qbk, q(x⊕q y)
n−k
=∑k=0
n
[nk ]q
(−1, q )n−k
2n−k bk ,q ( x ) yn−k (4.5)
Proof
By using (4.4)
47
tεq ( t )−1
∙ εq ( xt ) ∙ ε q ( yt )=∑n=0
∞
Bn ,q ( x , y ) t n
[n ]q !
(∑n=0
∞
Bn , qt n
[ n ]q ! )(∑n=0
∞
( x⊕q y )n t n
[ n ]q ! )=∑n=0
∞
Bn ,q ( x , y ) t n
[ n ]q!
∑n=0
∞
(∑k=0
n
Bk , qt k
[ k ]q ! ( x⊕q y )n−k t n−k
[ n−k ]q! )=∑n=0
∞
Bn ,q ( x , y ) t n
[ n ]q !
∑n=0
∞
(∑k=0
n
Bk , q ( x⊕q y )n−k tk
[ k ]q !t n−k
[ n−k ]q! )=∑n=0
∞
Bn ,q ( x , y ) t n
[ n ]q !
∑n=0
∞
(∑k=0
n
[nk ]q Bk , q
( x⊕q y )n−k) t n
[ n ]q !=∑
n=0
∞
Bn , q ( x , y ) t n
[ n ]q !
Comparing the power t we have
∑k=0
n
[nk ]qB
k , q( x⊕q y )
n−k
=∑n=0
∞
Bn,q ( x , y )
As required. For the second equation we must use 4.3.2, then it follows
tεq ( t )−1
∙ εq ( xt )∙ ε q ( yt )=(∑n=0
∞
Bn , q (x ) t n
[n ]q ! )(∑n=0
∞ t n yn
[ n ]q !(−1 , q )n
2n )By using Cauchy product we lead to
(∑n=0
∞
∑k=0
n
(nk)q
Bk , q (x ) yn−k (−1 ,q )n−k
2n−k ) tn
[ n ]q !=∑
n=0
∞
Bn , q (x , y ) tn
[n ]q !
Thus the last equation holds true.
4.2.6 Theorem for a real or complex parameter α , the following holds is true
Bn ,q(α+β )( x⊕q y )=∑
n=0
∞
[nk ]qBn ,q
(α ) ( x ) Bn−k , q( β ) ( y )(4.6)
Proof
Considering the right hand side
( tεq ( t )−1 )
α
ε q ( xt )=∑n=0
∞
bn , qα (x) t n
[ n ]q !
48
( tεq ( t )−1 )
β
ε q ( yt )=∑n=0
∞
bn , qβ ( y ) t n
[ n ]q!
Multiplying them we obtain
( tεq ( t )−1 )
( α+β )
ε q ( x⊕q y ) (t )=(∑n=0
∞
bn , qα (x ) t n
[n ]q ! )(∑n=0
∞
bn ,qβ ( y ) t n
[ n ]q ! )By using the Cauchy product of series formula
¿∑n=0
∞
(∑k=0
n
bk ,qα (x ) t k
[ k ]q !bn−k , q
β ( y) tn−k
[ n−k ]q ! )¿∑
n=0
∞
(∑k=0
n
bk ,qα (x )bn−k ,q
β ( y) t k
[ k ]q!tn−k
[ n−k ]q ! )¿∑
n=0
∞
(∑k=0
n
[nk ]q bk ,qα (x )bn−k ,q
β ( y) t n
[n ]q ! )¿∑
n=0
∞
(∑k=0
n
[nk ]q bk ,qα (x )bn−k ,q
β ( y)) tn
[ n ]q !
But the left hand side is
∑n=0
∞
Bn , q(α+β) ( x⊕q y ) tn
[ n ]q !
By comparing the power of t from the both side
Bn ,q(α+β )( x⊕q y )=∑
k=0
n
[nk ]qbk , q
α (x)bn−k ,qβ ( y )
As required base on the consequence of lemma 4.3 which is equivalent to (1.55)
lemma 4.2.7 The improved Bernoulli polynomials can also be demonstrated as
Bn ,q ( x )=∑k =0
n
[nk ]qbk ,q xn−k (−1 ;q )n−k
2n−k (4.7)
Proof. Put x=1 at additive theorem 4.3.1, then we lead to this equality.
By using the above expression we obtain the first few Bernoulli polynomials as
B0 ,q ( x )=1
B1 ,q ( x )=[2]q x
2−1
2
49
B2 ,q ( x )=[ 2 ]q x2
2 −x+q [2 ]q4 [ 3 ]q
B3 ,q ( x )=[ 2 ]q [ 3 ]q x3
4 −3 [ 2 ]q x2
4 +3 q [ 2 ]q x
4 [3 ]q
4.3 Improved q-Bernoulli Matrix And Its Properties
4.3.1 Definition Improved q-Bernoulli Matrix and Improved q-Bernoulli Polynomials Matrix
The generalized (n+1 ) × (n+1 ) q-improved Bernoulli polynomial matrix as
Bqα ( x )= [B i , j
α (q , x ) ] (i , j=0,1, …, n ) is defined by the following formular
Bi , jα (q , x )={( i
j)q
(−1; q)i− j
2i− j b i− j ,qα ( x ) ,if i ≥ j
0 , otherwise(4.8)
And B(1 ) ( x )=Bq ( x )∧B (0 )=Bq are called the improved q-Bernoulli polynomial matrix and
improved Bernoulli matrix respectively.
Example:
[ Bq( x )]=[ 1 0 02[2 ]q x−2 1 0
[ 2]q x2−2 x+q[ 2 ]q2[3 ]q
[2 ]q x−1 1 ] And
[ Bq ]=[1 0 0
−12 1 0
q[ 2]q
4[ 3 ]q−1 1 ]
Are the example of improved q-Bernoulli polynomial and Bernoulli matrix respectively. When
we tend q to 1 from the left side, we reach to the form of ordinary Bernoulli matrix.
4.3.2 Theorem Improved q-Bernoulli Polynomial Matrix in terms of x and y
The following relation for improved q-Bernoulli matrix with respect to x and y holds true:
¿
Proof
50
The case for i< j , i= j follows the same way as the previous proof in the generalized Bernoulli
matrix and for i> j we have:
∑k= j
i
( ik )
q
(−1 ;q)i−k
2i−k bi−k ,q
( x )(kj)q
(−1; q)k− j
2k− j b i−k ,q ( y )
¿( ij)q∑k=0
i− j
(i− jk )
q
(−1 ;q )i−k− j (−1 ;q )k2i− j b
i−k , q
(x ) bk , q ( y )
¿( ij)q
Bi− j
( x⊕q y )
And the last expression is additive theory as required.
Lemma 4.3.3 The following relation for improved q-Bernoulli numbers hold true
∑k=0
m
(mk )q
(−1;q )m−k
2m−k ∙bm−k
[k+1 ]=δ m,0(4.10)
Proof. At lemma (1.50) we proved this relation for Bernoulli numbers. The proof is exactly
similar to the ordinary case.
Definition 4.3.4 the inverse of improved q-Bernoulli matrix is defined by a matrix D, which is
d i , j(q)={ 1[ i− j+1 ]q ( i
j)qif i≥ j ,
0 , otherwise .(4.11)
4.3.3 Theorem Inverse of Improved q-Bernoulli matrix
Inverse of Improves q-Bernoulli matrix can be defined by the previous definition of D. That
means
B−1 (q )=D ( q ) .(4.12)
Proof
Since both of the matrices are lower triangle, their multiplications are also lower triangle. For
the another entries we may use the similar calculation as follow
¿∑k= j
i
( ik )
q(kj)q
1[ k− j+1 ]q
(−1 ;q )i−k
2i−k bi−k , q
¿( ij)q∑k= j
i
( i− jk− j)q
1[ k− j+1 ]q
(−1 ;q )i−k
2i−k bi−k , q
51
¿( ij )q
∑k=0
i− j
(i− jk )
q
1[ k+1 ]q
(−1; q )i− j−k
2i− j−k bi− j−k ,q=( ij)q
δ i− j , 0
As required.
Satisfying theorem 4.3.3.
52
CHAPTER 5
SUMMARY AND CONCLUSION
There are several definition for q-Bernoulli matrix and we works on a classes of improved q-
Bernoulli matrix which is more suitable to make a q-analogue of the same concept. Some
properties of improved q-Bernoulli numbers and polynomials allowed us to work this concept
easily. At the end of the day, we may defined these numbers by using different generating
functions. But the improved one works better and convinced the ordinary case better.
53
REFERENCE
Cenkci.M, Can.M. & Kurt.V. (2006). q-extensions of Genocchi numbers. J. Korean Maths. J.
Korean Math. Soc. 43(1), 183-198.
Cenkci.M, Kurt.V, Rim.S.H. & Y.Simsek. (2008). On (i; q) Bernoulli and Euler numbers.
Elsivier. 128(2), 280-312.
Cheon, G. (2003). A note on the Bernoulli and Euler polynomials. Elsevier. 16(3), 365-368
Choi. J, Anderson.P. & Srivastava.H. (2008). Some q-extensions of the Apostol-Bernoulli and
Apostol-Euler polynomials of order n and the multiple Hurwitz zeta function. Elsevier.
199(2), 723–737
Gabaury, S. & Kurt. B. (2012). Some relations involving Hermite-based Apostol-Genocchi
polynomials. Apply Mathematics Science. 6(82), 4091-4102.
Srivastava,H.M. & Pinter, A. (2004). Remarks on some relationships between the Bernoulli
and Euler polynomials. Elsevier. 17(4). 375-380.
Kim, T. (2007). A note on the q-Genocchi numbers and polynomials. Journal of Inequality
and application. 2007(71452)8.
Kim.D, Kurt.B, Kurt.V. (2013). Some Identities on the Generalized q-Bernoulli, q-Euler, and
q−Genocchi Polynomials. Abstract and Applied Analysis. 2013(293532), 6.
Luo Q.M. & Srivastava. M.H. (2006). Some relationships between the Apostol-Bernoulli and
Apostol-Euler-polynomials. Elsevier. 51(3–4), 631-642.
Luo, Q.M. (2010). Some results for the q-Bernoulli and q-Euler polynomials. Elsevier. 363(1),
7-18.
54
Naim.T. & KUS, S. (2015). Q-Bernoulli Matrices and Their Some Properties. Gazi University
Journal Of Science. 21(2).
Nalci.S. & Pashaev, O. K. (2012). q-Bernoulli Numbers and Zeros of q-Sine Function.
http://arxiv.org/abs/1202.2265v1.
Riordan, J. (1968). Combinatorial Identities. NewYork, London Sydney: Wiley.
Rudin, W. (1964). Principle Of Mathematical Induction. United State of America, U.S:
McGraw-Hill, Inc.
Srivastava, H.M. & pinter.A. (2004). Remarks on some relationships between the Bernoulli
and Euler polynomials. Applied Mathematics. 17(4), 375-380.
Srivastava.H.M. & Luo. (2006). Some relationships between the Apostol-Bernoulli and
Apostol-Euler-polynomials. Elsevier. 51(3–4), 631-642.
Srivastava.H.M. & Luo. (2011). q-Extensions of some relationships between the Bernoulli and
Euler polynomials. Taiwan J. Mathematics. 15(1)241-257.
Srivastava, H. M. & Vignat.C. (2012). Probabilistic proofs of some relationships between the
Bernoulli and Euler polynomials. European Journal and Applied Mathematics. 5(2).
Ernst. T. (2012). A comprehensive treatment of q-Calculus. United Kingdom: Springer Basel.
Kim.T. (2006). q-Generalized Euler numbers and polynomials. Springer. 13(3), 293–298.
Tsuneo, A., Tomoyoshi, I., Masanobu, K., and Don, Z. (2014). Bernoulli Numbers and Zeta
Function. Japan: Springer.
Victor. K. Cheung, & Pokman. (2002). Quantum Calculus. United state of America: Springer-
Verlag New York Berlin Heidelberg.
Zhizheng, Z. (1997). The Linear Algebra of the Generalized Pascal Matrix. Elsevier Science
Inc. 250(1), 51-60.
Zhizheng, Z., & JunWang. (2006). Bernoulli matrix and its algebraic properties. Elsevier B.V.
154(11), 1622–1632.
55
56