SEBHA UNIVERSITY JOURNAL OF PURE & APPLIED SCIENCES VOL.20 NO. 1 2021
مجلة جامعة سبها للعلوم البحتة والتطبيقية
Sebha University Journal of Pure & Applied Sciences
Journal homepage: www.sebhau.edu.ly/journal/index.php/jopas
On A Matrix Hypergeometric Differential Equation
Salah Hamd1, Faisal Saleh Abdalla2, Ahmed Shletiet3
of Mathematics, College of Education, Omar Al-Mukhtar University, Al-Bayda,Libya.
of general science, College of mechanical engineering technology – Benghazi, Benghazi,Libya
3Department of Mathematics, faculty of science, University of Ajdabiya, Ajdabiya, Libya.
1Department
2Department
Keywords:
Special functions.
A matrix Hypergeometric
differential Equation.
Matrix.
Gamma function.
Jordan Canonical Form.
ABSTRACT
In this paper we consider a matrix Hypergeometric differential equation, which are special matrix
functions and solution of a specific second order linear differential equation. The aim of this work is
to extend a well known theorem on Hypergeometric function in the complex plane to a matrix
version, and we show that the asymptotic expansions of Hypergeometric function in the complex
plane ” that are given in the literature are special members of our main result. Background and
motivation are discussed.
حل املعادلة التفاضلية لدالة فوق هندسية في الصورة املصفوفية
و أحمد شليتيت2و فيصل صالح عبدهللا1صالح حمد
3
ليبيا، جامعة عمر املختار، كلية التربية،قسم الرياضيات1
ليبيا، جامعة بتغازى، كلية تقنية الهندسة امليكاتيكية،قسم العلوم العامة2
ليبيا، جامعة اجدابيا، كلية العلوم،قسم الرياضيات3
:الكلمات املفتاحية
امللخص
.دوال خاصة
.دالة جاما
في هذا البحث تمت دراسة املعادلة التفاضلية لدالة فوق هندسية في الصورة املصفوفية والتي تمثل دالة
والهدف من هذه الدراسة هو تعميم نظرية.مصفوفية خاصة لحل معادلة تفاضلية خطية من الرتبة الثانية
مشهورة على الدوال فوق الهندسية لتكون في الصيغة املصفوفية بدال من الصورة التي كانت عليها في املستوى
باألضافة إلى ذلك تم أثبات تقريب املقرب للدوال فوق الهندسية في املستوى املركب والتي تم عرضها في،املركب
خالل هذا البحث تم مناقشة االساسيات والدوافع.بحوث سابقة كحالة خاصة من النتيجة التي توصلنا إليها
.التي استندت عليها هذه الدراسة
.صيغة جوردان القانونية
.مصفوفة
.معادلة تفاضلية فوق هندسية
Introduction
Generalization and extension of scalar special functions to matrix
special functions have been developed in the past two decades The
Gamma matrix function, whose eigenvalues are all in the right open
half-plane is presented and investigated by L. J´odar, J. Cort´es [1]
for matrices in 𝐶 𝑟𝑥𝑟 . Hermite matrix polynomials are introduced and
discussed by L. J´odar et al [2] and some of their properties are
provided in E. Defez, L. J´odar [3]. Other classical orthogonal
polynomials as Laguerre and Chebyshev have been extended to
orthogonal matrix polynomials, and some results have been studied
in L. J´odar, J. Sastre [4] and E. Defez, L. J´odar [5]. Relations
between the Beta, Gamma and the Hypergeometric matrix function
are given in L. J´odar, J. G. Cort´es [6] and R. S. Batahan [7]. These
special functions of matrices have developed an important tool in
both theory and applications. The main our goal is that , some cases
of the asymptotic expansions of 2F1(a, b; c; z) have been provided
in the literature, they are all limited by a narrow domain of validity
in the complex plane of the variable. Overcoming this restriction, we
provide new asymptotic expansion for the matrix hypergeometric
function .The order of presentation in this article is as follows. In
section 2 we provide basic necessary notation, definitions and
auxiliary theorems that need to be cited in the sequel. In section 3 we
provide our main results.
Preliminaries
In this part we elaborate on some necessary language that is adopted
from L. J´odar, J. Sastre [4] and N. J. Higham [8]. Denote by
*Corresponding author:
E-mail addresses: salahh92@gmail.com, (F. S. Abdalla2) faisal@ceb.edu.ly, (A. Shletiet) Ahmed.shletiet@uoa.edu.ly
Article History : Received 27 March 2021- Received in revised form 25 May 2021 - Accepted 01 june 2021
On A Matrix Hypergeometric Differential Equation
Hamd et al.
λ1, λ2 · · · , λn the distinct eigenvalues of a matrix P ∈ C r×r . The
spectrum σ(P) of P ∈ Cr×r , denotes the set of all the eigenvalues of P.
We put γ(P). and 𝝔(P) the real numbers
𝜸(𝑷) = 𝒎𝒂𝒙{𝑹𝒆(𝝀): 𝝀 ∈ 𝝈(𝑷)},
(𝟐. 𝟏)
𝝔(𝑷) = 𝐦𝐢𝐧{𝑹𝒆(𝝀): 𝝀 ∈ 𝝈(𝑷)}
holomorphic function f(λ) at a point was defined as a regular analytic
function in a neighborhood of the point, see e.g. W. Wasow [9]. It is
called holomorphic in a set if it is holomorphic at every point of the
set. A matrix is called holomorphic if every entry of it is a
holomorphic function. If 𝒇(𝝀) 𝒂𝒏𝒅 𝒈(𝝀) are homomorphic function
of the complex variable λ, which are defined in an open set Ω of the
complex plane, and P is matrix in Cr×r with σ(P) ⊂ Ω, than from the
properties of the matrix functional calculus, see N. Dunford, J.
Schwartz [10], it follows that
𝒇(𝑷)𝒈(𝑷) = 𝒈(𝑷)𝒇(𝑷)
(2.2)
A set of complex numbers is called positive stable if all the elements
of the set have positive real part and a square matrix P is called
positive stable if σ(P) is positive stable.
If P is a positive stable matrix in Cr×r , than Γ(P) is well defined, see
L. J´odar, J. G. Cort´es [1]
∞
𝜞(𝑷) = ∫ 𝒆−𝒕 𝒕𝒑−𝟏 𝒅𝒕
(𝟐. 𝟑)
𝟎
If 𝒇(𝑷) is well defined and T is an invertible matrix in
𝒇〖(TP𝑻−𝟏 )〗= 𝑻 𝒇(𝑷)𝑻−𝟏
(2.4)
It is a standard result that for any matrix P ∈Cr×r
there exist a nonsingular matrix T ∈Cr×r such that
𝑻−𝟏 𝑷 𝑻 = 𝑱 = 𝒅𝒊𝒂𝒈(𝑱𝟏 , 𝑱𝟐 , … … , 𝑱𝑵 )
(𝟐. 𝟓)
Where
𝝀𝒌 𝟏 𝟎 … 𝟎
⋮
𝟎 𝝀𝒌 𝟏 ⋱
Jk=Jk(λk)= ⋮
(2.6)
⋱ ⋱ ⋱ 𝟎 ∈𝑪𝒎𝒌 ×𝒎𝒌
⋱ 𝝀𝒌 𝟏
⋮
[ 𝟎 … … 𝟎 𝝀𝒌 ]
Cr×r ,
then
Lemma2.1 (matrix function via Jordan canonical form). Let f be
defined on σ(P), P ∈C r×r and let p have the Jordan canonical form
(2.5) subject to (2.6).Then
𝐟(𝐏) = 𝐓 𝐟(𝐉)𝐓 −𝟏 = 𝐓 𝐝𝐢𝐚𝐠(𝐟(𝐉𝟏 ), 𝐟(𝐉𝟐 ),· …
(𝟐. 𝟕)
· , 𝐟(𝐉𝐬 ))𝐓 −𝟏
Where
(𝝀𝒌 )
𝒇(𝝀𝒌 ) 𝒇(𝟏) (𝝀𝒌 ) … (𝒎 −𝟏)!
𝒌
𝒇(𝝀𝒌 )
⋱
⋮
𝒇(𝑱𝒌 )= 𝟎
∈𝑪𝒎𝒌 ×𝒎𝒌
(2.8)
⋱
⋱ 𝒇(𝟏) (𝝀𝒌 )
⋮
…
𝟎 𝒇(𝝀 ) ]
[ 𝟎
𝒌
Proof: The proof of this lemma is already proved in [8]
A hypergeometric function is the sum of a hypergeometric series,
which is defined as follows see e.g. F. W. J. Olver [11] . The
hypergeometric function 𝒑𝑭𝒒 (a1, a2, · · · ap; b1, b2, · · · bq; z) is
defined by means of a hypergeometric series as
𝒑𝑭𝒒 (𝒂1, 𝒂2,· · · 𝒂𝒑 ; 𝒃1, 𝒃2,· · · 𝒃q; 𝒛)
∞
=∑
𝑛=0
(𝑎1 )𝑛(𝑎2 )𝑛 · · · (𝑎𝑃 )𝑛 𝑧 𝑛
.
(𝑏1 )𝑛(𝑏2 )𝑛 · · · (𝑏𝑞 )𝑛 𝑛!
Recall that the shifted factorial (a)n is defined b
(𝒂)𝒏 = 𝒂(𝒂 + 𝟏)(𝒂 + 𝟐) · · · (𝒂 + 𝒏 − 𝟏) , 𝒏 ∈
𝑵 𝒂𝒏𝒅 (𝒂)𝟎 = 𝟏.
Gauss’s hypergeometric equation is a second order differential
equation with three regular singular points {0, 1,∞}, that is
𝒛(𝟏 − 𝒛)𝒇 ′′ + [𝒄 − (𝟏 + 𝒂 + 𝒃)𝒛]𝒇′ − 𝒂𝒃𝒇 = 𝟎.
JOPAS Vol.20 No. 1 2021
(𝒄)𝒏
𝒏!
Theorem 2.2 for Re 𝒄 > 𝑹𝒆 𝒃 > 𝟎 we have
2F1 (𝒂, 𝒃; 𝒄; 𝒛) =
𝜞(𝒄)
𝜞(𝒃)𝜞(𝒄 − 𝒃)
𝟏
∫𝟎 𝒕 𝒃−𝟏 (𝟏 − 𝒕) 𝒄−𝒃−𝟏 (𝟏 − 𝒛𝒕)−𝒂 𝒅𝒕,
(2.9)
for all 𝒛 ∈ 𝑪 and | 𝒛 | < 𝟏.
The hypergeometric function with matrix arguments F(A, B; C; z),
see L. J´odar and J. G. Sastre [6], is a solution of the differential
equation
𝒛(𝟏 − 𝒛)𝑾 (𝟐) − 𝒛𝑨𝑾(𝟏) + 𝑾(𝟏) (𝑪 + 𝒛(𝒏 − 𝟏)𝑰) + 𝒏𝑨𝑾
= 𝟎
which is defined by
𝑭(𝑨, 𝑩; 𝑪; 𝒛) =
𝟏
(∫ (1 − 𝒕𝒛)−𝑨 𝒕 𝑩−𝑰 (1 − 𝒕)𝑪−𝑩−𝑰 𝒅𝒕) 𝜞 −𝟏 (𝑩) 𝜞 −𝟏
𝟎
(𝑪 − 𝑩)𝜞(𝑪)
(𝟐. 𝟏𝟎)
Proposition 2.3 Let f(t) be a complex valued function of a real
variable t such that (𝑖) 𝑓(𝑡)𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑛 (0, ∞),
∞
(𝑖𝑖) 𝑓(𝑡) ∼ ∑ 𝑎𝑛𝑡 𝜉𝑛−1 𝑎𝑠 𝑡 ⟶ 0 𝑤𝑖𝑡ℎ 0 < 𝜉 0 < 𝜉1 < 𝜉 2
𝑛=0
The symbols 𝓞, o and ∼, due to Bachmann and Landau (1927), which
are also used by e.g. F. W. J. Olver [11] and A. Erd´elyi [12].
Concerning the definition and elementary properties of asymptotic
series we refer to W. Wa-sow [9] and A. Erd´elyi [12].
𝒇(𝒎𝒌 −𝟏)
and has a solutions
𝒏
∞ (𝒂)𝒏 (𝒃)𝒏 𝒛
2F1 (𝒂, 𝒃; 𝒄; 𝒛) = ∑𝒏=𝟎
< ……
𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑙𝑜𝑠𝑠 𝑜𝑓 𝑔𝑒𝑛𝑒𝑟𝑎𝑙𝑖𝑡𝑦 𝑎𝑠𝑠𝑢𝑚𝑒 𝑎0 ≠ 0,
(𝑖𝑖𝑖) 𝑓𝑜𝑟 𝑠𝑜𝑚𝑒 𝑓𝑖𝑥𝑒𝑑 𝑐 > 0 𝑓(𝑡)
= 𝑂{𝑒 𝑐𝑡 } 𝑎𝑠 𝑡 ⟶ ∞.
Then we have
∞ −𝜆 𝑡
𝑑𝑣
𝑑𝑛 𝛤(𝜉𝑛 + 𝑣)
𝑘
𝑎𝑠 𝑅𝑒 𝜆𝑘 ⟶
𝑓(𝑡)𝑑𝑡) ∼ ∑∞
𝜉𝑛+𝑣
𝑣 (∫0 𝑒
𝑛=0
𝑑𝜆𝑘
𝜆𝑘
∞
(2.11)
where 𝑑𝑛 = (−1)𝑣 𝑎𝑛 and
𝜋
𝜋
| 𝑎𝑟𝑔(𝜆𝑘 )| ≤
− 𝛿 <
𝑓𝑜𝑟 𝑠𝑜𝑚𝑒 𝛿 𝑠𝑢𝑐ℎ 𝑡ℎ𝑎𝑡 0 < 𝛿
2
2
𝜋
.
<
2
Proof: The proof of this proposition is already proved in [13] .
Lemma 2.4 Suppose 𝑄 is a positive stable matrix in Cr×r and suppose
also that 𝑓(𝑡)𝐼 ∈Cr×r , where f(t) is a function of a real variable t such
that
(𝑖) 𝑓(𝑡) 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑛 (0, ∞).
∞
(𝑖𝑖) 𝑓(𝑡) ∼ ∑ 𝑎𝑛 𝑡 𝜉𝑛−1 𝑎𝑠 𝑡 ⟶ 0
With
𝑛=0
(2.12)
0 < 𝜉0 < 𝜉1 < 𝜉2 < · · · · · ·
(𝑖𝑖𝑖) 𝑓𝑜𝑟 𝑠𝑜𝑚𝑒 𝑓𝑖𝑥𝑒𝑑 𝑐 > 0, 𝑓(𝑡) = 𝑂{𝑒 𝑐𝑡 } 𝑎𝑠
𝑡⟶ ∞
(iv) 𝑄 = 𝑃 has the Jordan canonical form subject to (2.5) and (2.6)
with
𝜋
𝜋
| 𝑎𝑟𝑔(𝜆𝑘 ) | ≤
− 𝛿 <
for some δ such that
0 <
𝜋
2
2
.
𝛿 <
2
Then we have
∞
𝑻−𝟏 [∫𝟎 𝒆−𝑸𝒕 𝒇(𝒕) dt] T ∼ 𝜱𝑸 as Re λk → ∞
(2.13)
Where 𝜱𝑸 = diag(𝜱𝟏 , 𝜱𝟐 , · · · , 𝜱𝑺 ) is a square block diagonal
matrix in Cr×r with blocks
Φk ∈ ℂ𝒎𝒌 ×𝒎𝒌 .
𝜱𝒌
=
𝒅𝒏 𝜞(𝝃𝒏 +𝒎𝒌 −𝟏)
∞
∑
)
𝒅𝒏 𝜞(𝝃𝒏
𝒅𝒏 𝜞(𝝃𝒏 + 𝟏)
𝒌 −𝟏
∑∞
∑∞
… 𝒏=𝟎 𝝀 𝝃𝒏+𝒎
(𝒎𝒌 −𝟏)!
𝝃𝒏+𝟏
𝒏=𝟎 𝝀 𝝃𝒏
𝒏=𝟎
𝒌
𝝀𝒌
𝒌
⋮
𝒅𝒏 𝜞(𝝃𝒏 )
∑∞
⋱ ∑∞ 𝒅𝒏𝜞(𝝃𝒏 + 𝟏)
𝟎
𝒏=𝟎 𝝀 𝝃𝒏
𝝃𝒏+𝟏
𝒏=𝟎
𝒌
𝝀𝒌
⋱
⋱
⋮
𝒅𝒏 𝜞(𝝃𝒏 )
∑∞
…
𝟎
𝒏=𝟎 𝝀 𝝃𝒏
𝟎
[
]
(2.14)
or we can write 𝜱𝒌 𝒂𝒔
𝒌
183
On A Matrix Hypergeometric Differential Equation
where
∑∞
𝒏=𝟎
𝜱𝒌 = 𝑫𝒌 + 𝑬𝒌 = 𝑫𝒌 [ 𝑰𝒎𝒌 + 𝑫−𝟏
𝒌 𝑬𝒌 ]
𝑫𝒌 =
𝒅𝒏 𝜞(𝝃𝒏 )
𝟎
𝝃𝒏
𝝀𝒌
And
⋮
𝟎
𝟎
∑∞
𝒏=𝟎
[
⋮
⋮
𝟎
[
∑∞
𝒏=𝟎
𝑬𝒌 =
𝒅𝒏 𝜞(𝝃𝒏 + 𝟏)
𝝃𝒏+𝟏
𝝀𝒌
⋱
𝟎
…
𝒅𝒏 𝜞(𝝃𝒏 )
⋱
…
⋱
𝝃𝒏
𝝀𝒌
…
⋱
⋱
…
⋱
𝟎
∑∞
𝒏=𝟎
𝟎
⋮
𝟎
𝒅𝒏 𝜞(𝝃𝒏 )
∑∞
𝒏=𝟎 𝝀 𝝃𝒏
𝒌
∈ ℂ𝒎𝒌 ×𝒎𝒌
]
𝝃𝒏+𝒓𝒌 −𝟏
∑∞
𝒏=𝟎
⋮
(𝒓𝒌 −𝟏)!
𝒅𝒏 𝜞(𝝃𝒏 + 𝟏)
𝝃𝒏+𝟏
𝝀𝒌
∈ ℂ𝒎𝒌 ×𝒎𝒌
…
]
𝟎
Proof: The proof of this lemma is already proved in [13] .
3 On a matrix Hypergeometric Differential Equation.
In this section we apply the machinery of the previous
sections to show that in a certain sense
𝐹(𝑎𝐼, 𝑏𝐼, 𝐶, 𝑧) ∼ 𝐼 𝑎𝑠 𝑅𝑒 𝜆 ⟶ ∞
𝑎𝑛𝑑
𝐹(𝑎𝐼, 𝑏𝐼, 𝜃𝑄, 𝑧) ∼ 𝐼 𝑎𝑠 𝜃 ⟶ ∞ ,
with 𝑄 a fixed matrix.
Theorem 3.1 Suppose C - bI ∈ ℂ r×r has a Jordan canonical form
subject to (2.5) and (2.6) then for any fixed matrices aI, bI in ℂ r×r, b >
0 we have
𝑇 −1 𝐹(𝑎𝐼, 𝑏𝐼; 𝐶; 𝑧)𝑇 𝐼 𝑎𝑠 𝑅𝑒 𝜆𝑘 ⟶ ∞
(3.1)
uniformly for | z |≤ δ < 1 with δ a fixed number.
Proof Consider the following integral as a function of a matrix C
1
𝜑(C) = ∫0 (1 − 𝑧𝑡)−𝑎𝐼 t (b−1)I (1 − t) C−(b+1)I dt.
(3.2)
Setting
1 − t = e −u , dt = e −u du
when t = 1 ⇒ u → ∞,
also when t = 0 ⇒ u = 0 ,
so we have
∞
φ(C) = ∫0 (1 + z(e−u − 1))−aI (1- e−u ) (b−1)I e−u(C- (b+1)I) e−u dt .
∞
= ∫0 (1 + z(e−u − 1))−aI (1- e−u ) (b−1)I e−u(C-bI) dt .
Note that
𝑢2
−𝑢
𝑢3
(1 + 𝑧(𝑒 −𝑢 − 1))−𝑎𝐼 =(1 + 𝑧( + − +
1!
2!
3!
Therefore
(1 + 𝑧(𝑒 −𝑢 − 1))−𝑎𝐼 ∼ I as u→ 0
Also
(1 + 𝑒 −𝑢 )−(𝑏−1)𝐼 =(1 − 1 +
−𝑢
−𝑢
𝑢2
1!
+
𝑢3
𝑢2
2!
−
𝑢4
𝑢3
3!
+
𝑢4
4!
𝑢4
4!
… ))−𝑎𝐼
… )(𝑏−1)𝐼
𝑢(𝑏−1) (1 − + − + … … )(𝑏−1) 𝐼
2!
3!
4!
1!
When u⟶ 0 we have
(1 + 𝑒 −𝑢 )(𝑏−1)𝐼 ∼ ub-1 I
thus
(1 + 𝑧(𝑒 −𝑢 − 1))−𝑎𝐼 (1- 𝑒 −𝑢 ) (b−1)I ∼ub-1 I dt
as u⟶ 0.
Now lemma 2.4 implies that
1
T −1 [∫0 (1 − 𝑧𝑡)−𝑎𝐼 t (b−1)I (1 − t) C−(b−1)I ] T
1
∫0 (1
𝑧𝑡)−𝑢𝐼𝑘
(𝑏−1)𝐼𝑘
𝑡)𝐽𝑘 −𝐼
1
= ∫0 t (b−1)I (1 − t)Jk −I dt
∼ ΦC−bI as Re λk ⟶ ∞ k = 1,2 …..
It is readily observed that
F(aI, bI; C; z)
1
𝒅𝒏 𝜞(𝝃𝒏 +𝒓𝒌 −𝟏)
𝝀𝒌
Hamd et al.
and obtain
1
T−1[β(Bi ,C−bI)]T=T−1[∫0 t (b−1)I (1 − t) C−(b+1)I dt] T
(1 −
dt ∼
ΦC−bI as
=
−
𝑡
Re λk ⟶ ∞
(3.3)
Note that by letting a = 0 in equation (3.3), we obtain the following
representation for the Beta matrix. Namely
1
β(C −bI, bI) = ∫0 𝑡 (𝑏−1)𝐼 (1 − 𝑡)𝐶−(𝑏+1)𝐼 dt
= Γ−1 (C)Γ(C − bI)Γ(bI)
(3.4)
= ( ∫0 (1 − 𝑡𝑧) −𝑎𝐼 t (b−1)I (1 − t) C−(b+1)I dt ) Γ −1 (bI)Γ−1 (C − bI)Γ(C)
(3.5)
in the sense that T −1 F(aI, bI; C; z)T ∼ I .
Example Let C = c ∈ (0, ∞) in the equation (3.5), then for any fixed
a and
b > 0 we have
1
𝜑(C) = ∫0 𝑡 𝑏−1 (1 − t) c−b−1 (1 − zt) −a dt.
Setting 1 − t = e −u , dt = e −u du when
t = 1 ⇒ u ⟶ ∞ , also when t = 0 ⇒ u = 0
so we have
1
𝜑(C)=∫0 (1 + 𝑧(𝑒 −𝑢 − 1))−𝑎 (1-1 − 𝑒 −𝑢 )b-1 e-u(c-b) du .
Note that
(1 + 𝑧(𝑒 −𝑢 − 1))−𝑎 (1-1 − 𝑒 −𝑢 )b-1 ~ ub-1
as u→ 0
Now Watson’s lemma implies that
1
𝛤(𝑏)
∫0 (1 + 𝑧(𝑒 −𝑢 − 1))−𝑎 (1-1 − 𝑒 −𝑢 )b-1 e-u(c-b) du ~
𝑏 as
c→ ∞
or
1
∫0 t b−1 (1 − t) c−b−1 (1 − zt) −a dt . ~
(𝑏−𝑐)
Γ(b)
(b−c)b
as c→ ∞
(3.6)
By assuming a=0 in (3.6) we obtain
1
β(b, c − b) = ∫0 t b−1 (1 − t) c−b−1 dt ~
Γ(b)
(3.7)
(b−c)b
By (3.6) and (3.7) we derive asymptotic expansion for 2F1(a,b;c; z)
when c
Approaches infinity,
(3.8)
2F1(a,b;c; z) ~ 1 as c→ ∞
Corollary 3.2 Suppose 𝐶 − 𝑏𝐼 = 𝜃𝑄 where θ ∈ (0, ∞) and 𝑄 is
a constant matrix and has a Jordan canonical form subject to (2.5)
and (2.6) then
1
T −1 [∫0 (1 − zt)−aI t (b−1)I (1 − t)c−(b+1)I dt] T ∼ Ψ θQ as θ →
∞
(3.9)
Where 𝛹𝜃𝑄 = 𝑑𝑖𝑎𝑔(𝛹1, 𝛹2,· · · , 𝛹𝑠) is a square block diagonal
matrix in ℂ r×r, with blocks 𝛹𝑘 ∈ ℂ𝑚𝑘×𝑚𝑘 ,
Ψk=
∑∞
𝑛=0
𝑎𝑛 𝛤(𝜉𝑛 )
𝜃 𝜉𝑛
∑∞
𝑛=0
𝑑𝑛 𝜆𝑘 𝛤(𝜉𝑛 + 1)
(𝜃𝜆𝑘 )𝜉𝑛 + 1
𝑎 𝛤(𝜉 )
∞
∑𝑛=0 𝑛 𝜉𝑛 𝑛
𝜃
⋯
∑∞
𝑛=0
𝑑𝑛 𝜆𝑘 𝛤(𝜉𝑛 +𝑚𝑘 −1)
(𝜃𝜆𝑘 )𝜉𝑛+𝑚𝑘 −1 (𝑚𝑘 −1)!
⋮
𝑑𝑛 𝜆𝑘 𝛤(𝜉𝑛 + 1)
⋱
0
∑∞
𝑛=0 (𝜃𝜆 )𝜉𝑛 + 1
𝑘
⋱
⋱
⋮
𝑎𝑛 𝛤(𝜉𝑛 )
∞
∑𝑛=0 𝜉𝑛
0
…
0
[
]
𝜃
Moreover,
T −1 F(aI, bI; θC; z)T ∼ I as θ ⟶ ∞
(3.10)
Uniformly for | z | ≤ δ < 1 with 𝛿 a fixed number.
Proof: By Watson’s lemma we have
1
𝑎𝑛 𝛤(𝜉𝑛 )
∫0 (1 − 𝑧𝑡) −𝑎 𝑡 (𝑏−1) (1 − 𝑡)(𝜃𝜆𝑘)−1 dt ∼ ∑∞
𝑛=0
𝜉𝑛
θ→∞.
and by proposition 2.3 as 𝜃 → ∞ we get
1
dv
( ∫0 1 − zt) −a t
dθv
v
dn λk Γ(ξn + v)
∑∞
n=0 (θλ )ξn+v .
k
t (b−1)
as
(𝜃𝜆𝑘 )
(1 − t) (θλk )−1
dt
)
∼
Thus by the lemma 2.1 we have
JOPAS Vol.20 No. 1 2021
184
On A Matrix Hypergeometric Differential Equation
1
∫0 (1
zt)−aIk
t (b−1)Ik
t)θJk −1
−
(1 −
dt ∼ Ψk
and the results (3.9) and (3.10) follow.
Conclusion
We show that
and
Hamd et al.
as θ ⟶ ∞ ,
𝐹(𝑎𝐼, 𝑏𝐼, 𝐶, 𝑧) ∼ 𝐼 𝑎𝑠 𝑅𝑒 𝜆 ⟶ ∞
𝐹(𝑎𝐼, 𝑏𝐼, 𝜃𝑄, 𝑧) ∼ 𝐼 𝑎𝑠 𝜃 → ∞
with 𝑄 a fixed matrix.
References:
[1]- J´odar L. and Cort´e J. G. , Some Properties Of Gamma And
Beta Matrix Function, Appl . Math. Lett, 11, (1998), 89-93
[2]- J´odar L . and Company R. Hermite Matrix Polynomials And
Second Order Matrix Differential Equation , Approximation
Theory and its Application 12(2) , 20-30, (1996).
[3]- Defez E. and J´odar, L. Some Applications of the Hermite Matrix
Polynomials Series
Expansions, Journal Of Computational
Applied Mathematics 99, 105-117, (1998).
[4]- J`odar L. and Sastre J. The growth Of Laguerre Matrix
Polynomials On Bounded
Intervals, appl. math. lett, 13,
(2000),21-26.
[5]- Defez E. , and J´odar L . Chebyshev Matrix
Polynomials
And Second Order Matrix Differential Equation. Utilitas
Mathematica , (2002). 61. 107-123.
[6]- J´odar L. and Cort´es J. G. On the Hypergeometric Matrix
Function, J. Comp. Appl . Math.99 , (1998),205-217.
[7]- Batahan R. S. Generalized Form Of Hermite Matrix Polynomials
Via the Hypergeometric Matrix Function, Advances In Linear
Algebra And Matrix Theory, (2014) ,4,134-141.
JOPAS Vol.20 No. 1 2021
185