Honam Mathematical J. 43 (2021), No. 1, pp.

141–151
https://doi.org/10.5831/HMJ.2021.43.1.141

GENERALIZATION OF LAGUERRE MATRIX

POLYNOMIALS FOR TWO VARIABLES

Asad Ali∗ and Muhammad Zafar Iqbal

Abstract. The main object of the present paper is to introduce the gen-
eralized Laguerre matrix polynomials for two variables. We prove that
these matrix polynomials are characterized by the generalized hypergeo-
metric matrix function. An explicit representation, generating functions
and some recurrence relations are obtained here. Moreover, these matrix
polynomials appear as solution of a differential equation.

1. Introduction and preliminaries

Orthogonal matrix polynomials comprise an emerging fields of study, with

important results in both Lie group theory and number theory its applications
being still is contained to appear in the literature. Theory of classical orthogo-
nal polynomials are extended to the orthogonal matrix polynomials ([9]). The
study of functions of matrices is a very popular topic in the Matrix Analysis
literature. Matrix generalization of special functions has become important
in the last two decades. The reason of importance have many motivations.
For instance, using special matrix functions provides solutions for some phys-
ical problems. Also, special matrix functions are in connection with different
matrix functions ([4], [8], [14]). Jodar et al introduced Laguerre matrix poly-
nomials in ([11]). Some important properties of Laguerre matrix polynomials
such as asymptotic expressions relations between different matrix functions and
generating matrix functions are studied ( [8], [9], [11]). Indeed, in recent pa-
pers, matrix polynomials have significant emergent. Some results in the theory
of classical orthogonal polynomials have been extended to orthogonal matrix
polynomials ( [1], [2], [4], [5], [7], [11], [13]) Throughout this paper, for a ma-
trix A in C r×r , its spectrum σ(A) denotes the set of all eigenvalues of A. The

Received November 4, 2020. Accepted February 9, 2021.

2020 Mathematics Subject Classification. 33C45, 42C05.
Key words and phrases. Laguerre matrix polynomials, generating functions, recurrence
relations.
*Corresponding author
142 A. Ali and M. Z. Iqbal

two-norm will be denoted by ||A||2 and it is defined by ([10])

||Ax||2
||A||2 = sup
x̸=0 ||x||2
1
where for a vector x in C r×r , ||x||2 = (xT x) 2 is the Euclidean norm of x. Let
us denote the real numbers M (A) and m(A) as in the following
(1) M (A) = max {Re(z) : z ∈ σ(A)} ; m(A) = min {Re(z) : z ∈ σ(A)} .
If f (z) and g(z) are holomorphic functions of the complex variable z, which are
defined in an open set of the complex plane, and A, B are matrices in C r×r ,
with σ(A) ⊂ Ω and σ(B) ⊂ Ω, such that AB = BA, then from the properties
of the matrix functional calculus in ([6]) , it follows that
(2) f (A)g(B) = g(B)f (A).
Throughout this study, a matrix polynomial of degree n in x means an expres-
sion of the form n
(3) Pn (x) = An xn + An−1 xn−1 + ... + A1 x1 + A0 .
where x is a real variable or complex variable, Aj , for 0 ≤ j ≤ n and An ̸= 0,
where 0 is the null matrix or zero matrix in C r×r . We recall reciprocal gamma
function denoted by Γ−1 (z) = Γ(z)
1
is an entire function of the complex variable
z. Then, for any matrix A in C r×r
, the image of Γ−1 (A) acting on A, denoted
by Γ−1 (A) is a well defined matrix. Furthermore, if A is a matrix such that
(4) A + nI is invertible f or every integer n ≥ 0,
where I is the identity matrix in C r×r , then from ([11]) it follows that
(5) (A)n = A(A + I)...(A + (n − 1)I) = Γ(A + nI)Γ−1 (A); n ≥ 1; (A)0 = 1.
If A is a positive stable matrix in C r×r , then the gamma matrix function, Γ(A),
is defined ([12]) by
∫ ∞
(6) Γ(A) = e−t tA−1 dt, Re(A) > 0.
0

And if A, B is a positive stable matrices in C r×r , then the beta matrix function,
β(A, B), are defined ([12]) by
∫ 1
(7) β(A, B) = tA−1 (1 − t)B−1 dt, Re(A) >, Re(B) > 0.
o

The generalized hypergeometric matrix function p Fq (p, q ∈ N) given in ([13])

[ ] ∑∞
A1 , . . . , A p ; (A1 )n · · · (Ap )n xn
F
p q x =
(8) B1 , . . . , B q ;
n=0
(B1 )n · · · (Bq )n n!
= p Fq (A1 , . . . , Ap ; B1 , . . . , Bq ; x).
 Generalization of Laguerre Matrix polynomials for two variables 143

Where Ai and Bj are matrices in C r×r such that Bj ; 1 ≤ j ≤ q satisfy

condition. With p = 1 and q = 0, one gets the following relation by ([13])
∑∞
(A)n xn
(9) 1 F0 (A; −; x) = (1 − x)−A = ,
n=0
n!

In ([12], [14], [15]) if λ is a complex number with Re(λ) > 0 and A is a positive
stable matrix in Cr×r , with A + nI invertible for every integer n ≥ 1, then the
(λ,A)
nth Laguerre matrix polynomials Ln (x, y) for two variables are defined by
the following generating matrix relation:
∞
∑ −λxt
(10) L(λ,A)
n (x, y)tn = (1 − yt)−I−A e( 1−yt ) .
n=0

From the above equation we have

∑
n
(I + A)n (−xλ)k (y)n−k
(11) L(λ,A) (x, y) = , λ ≥ 0.
n
(n − k)!(I + A)k k!
k=o

The second-order matrix differential equations of the form

( ( ) )
d2 x d nI (λ,A)
(12) xI 2 + (1 + A) − λ I + L1,n (x, y) = 0.
dx y dx y
where n ∈ N, and A ∈ Cr×r . And Cr×r denotes the vector space containing all
square matrices with r rows and r columns with entries in the complex number
C.

2. Generalized Laguerre Matrix polynomials for two variables

(λ,A)
We begin by defining generalized Laguerre matrix polynomials Lp,n (x, y)
for two variables by the following generating function:
( ) ∑ ∞
1 −λxp tp
(13) exp = L(λ,A)
n,p (x, y) t
n
(1 − yt)I+A (1 − yt)p n=0

( )
p ∈ N; x, y ∈ C, A ∈ Cr×r .
Here and elsewhere, let N, R and C be the sets of positive integers, real num-
ber and complex numbers, respectively, and let N0 := N ∪ {0}. Obviously
(λ,A) (λ,A)
Ln,1 (x, y) = Ln (x, y).
Hereafter we explore certain formulas and properties involving the general-
ized Laguerre Matrix polynomials in (13). Throughout, let F (p; λ, x, y, t) be
the left-handed generating function in (13).
144 A. Ali and M. Z. Iqbal

Explicit representation.
We give an explicit expression of the generalized Laguerre matrix polyno-
(λ,A)
mials Ln,p (x, y) for two variables in the following theorem.

Theorem 2.1. Let x, ∈ C, p ∈ N, n ∈ N0 and A is positive stable matrix

in Cr×r . Then
∑
[n/p]
(−λ)k (y)n−pk
(14) L(λ,A)
n,p (x, y) = (I + A)n xpk
k! (I + A)pk (n − pk)!
k=0

(I + A)n ∑ (λ)k (−1)(p+1)k (−nI)pk (y)n−pk pk

[n/p]
(15) = x .
n! k! (I + A)pk
k=0

Here and throughout, [m] denotes the greatest integer less than or equal to
m ∈ R. Or, equivalently,
(16)  
−n −n + 1 −n − 1 + p
I, I, . . . , I ; ( )
(I + A)n y n  p p p x 
p
L(λ,A)
n,p (x, y) =

p Fp  (−1)p+1 λ .
n! A + I A + 2I A + pI y 
, , ..., ;
p p p
Proof. Expanding the exponential in the left-hand side of (13), we find
∞
∑
1 λk (−1)k xpk tpk
F (p; λ, x, y, t) = .
(1 − yt)I+A+pk k!
k=0

Employing the binomial theorem

∑∞
(A)n xn
(17) (1 − x)−A = 1 F0 (A; −; x) = , (A ∈ Cr×r ; |x| < 1),
n=0
n!

we obtain the following double series

∑∞ ∑∞
(−1)k λk (I + A + pk)n xpk y n n+pk
(18) F (p; λ, x, y, t) = t .
n=0
k! n!
k=0

Recall ([11]) if A(n, k) and B(n, k) are matrices in Cr×r and satisfying the
spectral condition (4) for n ≥ 0, k ≥ 0, then it follows that
∞ ∑
∑ ∞ ∞ ∑
∑ n
(19) A(n, k) = A(n, n − k),
n=0 k=0 n=0 k=0

∞ ∑
∞ ∞ ∑ [n]
∑ ∑ p

(20) A(n, k) = A(n, n − pk), (p ∈ N),

n=0 k=0 n=0 k=0
 Generalization of Laguerre Matrix polynomials for two variables 145

∞ ∑ [n] ∞ ∑
∞
∑ p
∑
(21) A(n, k) = A(n, n + pk), (p ∈ N),
n=0 k=0 n=0 k=0
where Ax,y denotes a function of two variables x and y and the involved
double series is assumed to be absolutely convergent.
Applying (20) in (18), we get
∑∞ [n/p]
∑ (−1)k λk (I + A + pk)n−pk xpk y n−pk
(22) F (p; λ, x, y, t) = tn .
n=0
k! (n − pk)!
k=0
Equating the coefficients of tn in the right members of (13) and (22) yields
∑ (−1)k λk (I + A + pk)n−pk
[n/p]
(23) L(λ,A)
n,p (x, y) = xpk y n−pk .
k! (n − pk)!
k=0
Using ([12]) a known identity
1 (−1)k (−nI)k
(24) I= (k, n ∈ N0 ; 0 ≤ k ≤ n) ,
(n − k)! n!
we derive
(25)
(I + A)n 1 (−1)pk (−nI)pk
(I+A+pk)n−pk = and I= ; 0 ≤ pk ≤ n.
(I + A)pk (n − pk)! n!
Hence, using (25) in (23) leads to the desired identity (15).

Finally, applying the multiplication formula

p ( )
(−1)pk (−nI)pk (−1)pk (p)pk ∏ j − n − 1
(26) = I ; 0 ≤ pk ≤ n.
n! n! j=1
p n

( p ∈ N; n ∈ N0 )
to (15) gives the equivalent expression (16).

Generating function.
We establish two generating functions for the generalized Laguerre matrix
(λ,A)
polynomials Ln,p (x, y) for two variables in the following theorem.

Theorem 2.2. Let λ, t, x, y c ∈ C, p ∈ N and A is positive stable matrix

in Cr×r and satisfying the spectral condition (4). Then
( ( )p )
A + I A + 2I A + pI xzt
eyt 0 Fp ; , , ..., ; −λ
p p p p
(27) ∞
∑ Ln,p (x, y) tn
(λ,A)
=
n=0
(I + A)n
146 A. Ali and M. Z. Iqbal

and
( ( )p )
c c+1 c+p−1
1 p, p , . . . , ; xt
p Fp
p
A+pI −λ
(1 − yt)c A+I A+2I
p , p , ..., p ; 1 − yt
(28) ∞
∑ (λ,A)
(c)n Ln,p (x, y) tn
= (|t| < 1).
n=0
(I + A)n

Proof. Using (14), (21), and (26), we have

∑∞ ∑∞ ∞
(yt)n ∑ (−λxp z p tp )k
(λ,A)
Ln,p (xz, y) tn
=
n=0
(I + A)n n=0
n! k! (I + A)pk
k=0
(29) ∑∞ ( )pk
(−λ)k xzt
= eyt p ( ) .
∏ A+jI p
k=0 k!
p
j=1 k

In view of (8), the rightmost term of (29) can be expressed as the left-hand
side of (27).

Employing (14), (17), and (21), we find

∑∞ (λ,A) ∑∞ ∑ ∞ k
(c)n Ln,p (x, y) tn (c + pk)n (yt)n (c)pk {−λ(xt)p }
= ·
n=0
(I + A)n n! k! (I + A)pk
k=0 n=0
∑∞ { ( )p }k
1 (c)pk xt
= −λ ,
(1 − yt)c k! (I + A)pk 1 − yt
k=0

which, upon using (26) and (8), leads to the left-hand member of (28).

It is noted that the case c = I + A of (28) yields the generating function

(13).

Recurrence relation.
We give some recurrence relations involving the generalized Laguerre matrix
(λ,A)
polynomials Ln,p (x, y) for two variables and their derivative in the following
theorem.

Theorem 2.3. Let λ, t, x, c ∈ C, p ∈ N and A is positive stable matrix in

Cr×r and satisfying the spectral condition (4). Also let D = dx
d
. Then
(λ,A)
n,p (x, y) − n Ln,p (x, y) + y(A + (n + 1)I) Ln−1,p (x, y) = 0;
x λDL(λ,A) (λ,A)
(30)
{
0 (0 ≤ n ≤ p − 1)
(31) DL(λ,A)
n,p (x, y) = (λ,A+p)
−p λxp−1 Ln−p,p (x, y) (n ≥ p);
 Generalization of Laguerre Matrix polynomials for two variables 147

(32)
(λ,A) 2 p (λ,A+p)
y(A + (n + 1)I) Ln−1,p (x, y) − n L(λ,A)
n,p (x, y) = p λ x Ln−p,p (x, y) (n ≥ p).
Proof. From (29), we can set
∑∞ (λ,A) ( )
Ln,p (x) tn −λxp tp
(33) G(p; λ, x, y, t) := = eyt Φ ,
n=0
(I + A)n pp
where the function
( ) ∑∞ ( )pk
−λ xp tp (−λ)k xt
∏ ( A+jI )
Φ = p .
pp p
k=0 k! p
j=1 k

Differentiating G(p; λ, x, y, t) with respect to x and t, respectively, gives

( )
yt ′ −λ xp tp −λ
Gx (p; λ, x, y, t) = e Φ p
· p−1 xp−1 tp
p p
and
( ) ( )
−λ xp tp ′ −λ xp tp −λ p p−1
yt
Gt (p; λ, x, y, t) = ye Φ +e Φ yt
· x t .
pp pp pp−1
Combining Gx (p; λ, x, y, t) and Gt (p; λ, x, y, t) yields
(34) λx Gx (p; λ, x, y, t) − t Gt (p; λ, x, y, t) + yt G(p; λ, x, y, t) = 0.
Applying the series in (33) to (34), we obtain
(35)
∑∞ ∞ ∞
λx DLn,p (x, y) tn ∑ n Ln,p (x, y) tn ∑
(λ,A) (λ,A) (λ,A)
Ln−1,p (x, y) tn
− +y = 0.
n=1
(I + A)n n=1
(I + A)n n=1
(I + A)n−1
We find from (35) that each coefficient of tn should be zero, which gives (30).

Differentiating both sides of (13) provides

∞
∑ ( )
1 −λxp tp ( )
DL(λ,A)
n,p (x, y) t
n
= exp · −λp xp−1 tp
n=1
(1 − yt)I+A+p (1 − yt)p
∞
∑
= −pλ xp−1 (λ,A+p)
Ln,p (x, y) tn+p
n=0
∑∞
(λ,A+p)
= −pλ xp−1 Ln−p,p (x) tn ,
n=p

which, upon equating the coefficients of tn (n ≥ p) in the leftmost and rightmost

members, produces (31).

Setting (31) in (30) provides (32).

148 A. Ali and M. Z. Iqbal

Differential equation.
We provide a differential equation which is satisfied by the generalized La-
(λ,A)
guerre polynomials Ln,p (x, y) for two variables in the following theorem (for
differential equation whose solution is p Fq , see, e.g., [15, Section 47]).
Theorem 2.4. Let λ, t, x, c ∈ C, p ∈ N and A is positive stable matrix in
d
C r×r and satisfying the spectral condition (4). Also let θ = x dx . Then

∏p ( )
1θ 1
(θI − I + A + jI)
p j=0 p
(36) 
( )p ∏ p
x 1
+λ(−1)p−1 (θ + j − n − 1)I  L(λ,A)
p,n (x, y) = 0.
y j=0
p

Proof. We find from (15) that

( ( )p )
(I + A)n (y)n −n −n + 1 −n + p − 1 I + A 2I + A pI + A p−1 x
ϕ= p Fp I, I... I; , ... ; λ(−1) .
n! p p p p p p y
p (
∏ )
j−n−1
[n
p] I (−1)(p−1)k λk xpk (y)n−pk
(I + A)n ∑ j=1 p
k
= p (
∏ ) ,
n! jI+A
k=o
p k!
j=1 k

1 pk pk
Since p θx = kx , it follows that
p (
∏ ) ( )
  jn−1 j+A+k−1
[n
p] I I (−1)(p−1)k λk xpk (y)n−pk
∏
p
(I + A)n ∑ j=1 p p
1θ I k
(θ − 1 + A + j) = p ( ) ,
p j=0 p n! ∏
k=o
jI+A
p (k − 1)!
i=1 k
( ) ( )
jI+A jI+A+k−1
But the last factor in p is p so that
k
p (
∏ )
j−n−1
[n
p] I (−1)(p−1)k λk xpk (y)n−pk
(I + A)n ∑ j=1 p
k
= p (
∏ ) ,
n!
k=o
jI+A
p (k − 1)!
j=1 k−1

Now we replace k by (k+1) and have

p (
∏ )
  j−n−1
[n
p] I (−1)(p−1)(k+1) λk+1 xp(k+1) (y)n−p(k+1)
∏
p
(I + A)n ∑ j=1 p
1θ 1 k+1
(θ − 1 + A + j) I  = p ( ) ,
p j=0 p n! ∏ jI+A
k=o
p k!
j=1 k
 Generalization of Laguerre Matrix polynomials for two variables 149

p (
∏ ) ∏ p ( )
j−n−1 k+j−n−1
( )p [n
p] I I (−1)(p−1)k λk xpk (y)n−pk
p−I x (I + A)n ∑ j=1 p
k j=1
p
= λ(−1) p (
∏ ) ,
y n! jI+A
k=o
p k!
j=1 k
 
( )p ∏p
x  1
= λ(−1)p−1 (θ + j − n − 1)I  ϕ.
y j=0
p

Some other properties.

We provide some other identities involving the generalized Laguerre poly-
(λ,A)
nomials Ln,p (x, y) for two variables in the following theorem.

Theorem 2.5. Let λ, t, x, y c ∈ C, p, n ∈ N and A, B is positive stable

matrix in Cr×r and satisfying the spectral condition (4). Then
∑
n (λ,B)
(A − B)k Ln−k,p (x, y)
(37) L(λ,A)
n,p (x, y) = ;
k!
k=0

∑
n
(λ,A) (λ,B)
(λ,A+B+I)
(38) Ln,p (z, y) = Lk,p (x, y) Ln−k,p (z, y),
k=0
1
where xp + z p ∈ C \ {0} and w := (xp + z p ) p whose principal branch can be
chosen;
∑
n (λ,A)
(I + A)n (1 − z)n−k z k Lk,p (x, y)
(39) L(λ,A)
n,p (xz, y) = .
(n − k)! (I + A)k
k=0

Proof. From (13), we have

∞
∑ ( )
−I−A −λxp tp
L(λ,A)
n,p (x, y) t
n
= (1 − yt) exp
n=0
(1 − yt)p
( )
−λxp tp
= (1 − t)−(A−B) · (1 − yt)−I−B exp
(1 − yt)p
∑∞ ∑∞
(A − B)k (λ,B)
= Ln,p (x, y)tn+k
n=0
k!
k=0
∑∞ ∑n
(A − B)k (λ,B)
= Ln−k,p (x, y)tn ,
n=0
k!
k=0

which, upon equating the coefficients of tn , yields (37).

150 A. Ali and M. Z. Iqbal

We find from (13) that

∑∞ ∑ n
(λ,A) (λ,B)
Lk,p (x, y) Ln−k,p (z, y) tn
n=0 k=0
( ) ( )
−I−A −λxp tp −I−B −λz p tp
= (1 − yt) exp (1 − t) exp
(1 − yt)p (1 − yt)p
( )
−λw t
p p
= (1 − yt)−I−(A+B+I) exp
(1 − yt)p
∑∞
= L(A+B+I)
n,p (w, y) tn ,
n=0
which, upon equating the coefficients of tn , gives (38).

We consider
( ( )p )
A + I A + 2I A + pI xzt
yt
e 0 Fp ; , , ..., ; −λ
p p p p
( ( )p )
A + I A + 2I A + pI x(zt)
=e (1−z)yt yzt
e 0 Fp ; , , ..., ; −λ ,
p p p p
which, in view of (27), produces
∞
(∞ ) ( ∞ (λ,A) )
∑ (λ,A)
Ln,p (xz, y) tn ∑ (1 − zt)n y n tn ∑ Lk,p (x, y) z k tk
= .
n=0
(I + A)n n=0
n! (I + A)k
k=0
Then, from the last equality, we obtain (39).

3. Conclusion remarks

The generalized Laguerre matrix polynomials for two variables are intro-
duced here and their properties and formulas presented are hoped to be poten-
(λ,A) (λ,A)
tially useful. Since Ln,1 (x, y) = Ln (x, y), the results in Section 2 reduce
to yield certain properties and formulas for the classical Laguerre matrix poly-
(λ,A)
nomials Ln (x, y) for two variables.

References
[1] H. Abd-Elmageed, M. Abdalla, M. Abul-Ez and N. Saad. Some results on the first
Appell matrix function, Linear and Multilinear Algebra. 68(2) (2020), 278-292.
[2] A. Ali, M. Z. Iqbal, B. Anwar and A. Mehmood. Generalization of Bateman polynomial,
Int. J. of Anal. and Appl. 17(5) (2019), 803-808.
[3] A. Ali, M. Z. Iqbal, T. Iqbal and M. Haider. Study of Generalized k-hypergeometric
Function, Int. J. of Math. and Comp. Sci. 16(1) (2021), 379-388.
[4] R. S. Batahan. A new extension of Hermite matrix polynomials and its applications,
Linear Algebra and its Application. 419 (2006), 82–92.
 Generalization of Laguerre Matrix polynomials for two variables 151

[5] A. Shehata. Connections between Legendre with Hermite and Laguerre matrix polyno-
mials, Gazi Uni. J. of Sci. 28(2) (2015), 221-230.
[6] A. Shehata. Some relations on Laguerre matrix polynomials Malaysian J. of Math. Sci.
9(3) (2015), 443-462.
[7] A. Shehata. On modified Laguerre matrix polynomials, J. of Natural Sci. and Math.
8(2) (2015), 153-166.
[8] A. Shehata. Lie algebra and Laguerre matrix polynomials of one variable, Gen. Lett. in
Math. 4(1) (2018), 1-5.
[9] A. J. Duran and F. A. Grnbaum. A survey on orthogonal matrix polynomials satisfying
second order differential equations, J. of Comp. and Appl. Math. 178(1-2) (2005), 169-
190.
[10] G.Golub and C. F. Van Loan. Matrix Computations, The Johns Hopkins University
Press, Baltimore (1989).
[11] L. Jodar, R. Company and Navarro E. Navarro. Laguerre matrix polynomials and sys-
tems of second order differential equations, App. Numer. Math. 15(1) (1994), 53-63.
[12] G. Dattoli, and A. Torre. Operational methods and two variable laguerre polynomials.
Atti Rendiconti Acc. Torino. 32 (1998), 1-7.
[13] L. Jodar and J. C. Cortes. On the hypergeometric matrix function, J. of Comp. and App
Math. 199 (1-2) (1998), 205-217.
[14] R. S. Batahan and A. A. Bathanya, On generalized Laguerre matrix polynomials, Acta
Univ. Sapientiae, Mathematica, 6(2) (2014), 121-134.
[15] K. Subuhi; and N. A. M. Hassan. 2-variable Laguerre matrix polynomials and Lie-
algebraic techniques, J. of Phy. A: Math. and Theo. 43(23) (2010), 235204.

Asad Ali
Department of Mathematics and Statistics,
University of Agriculture Faisalabad 38000, Pakistan
E-mail:mrasadali5100@gmail.com

Muhammad Zafar Iqbal

Department of Mathematics and Statistics,
University of Agriculture Faisalabad 38000, Pakistan
E-mail:mzts2004@hotmail.com