Next Article in Journal
From Annular to Toroidal Pseudo Knots
Next Article in Special Issue
Symmetric Functions and Rings of Multinumbers Associated with Finite Groups
Previous Article in Journal
Modeling Pollutant Diffusion in the Ground Using Conformable Fractional Derivative in Spherical Coordinates with Complete Symmetry
Previous Article in Special Issue
Spectral Properties of Dual Unit Gain Graphs
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Solving the Dual Generalized Commutative Quaternion Matrix Equation AXB = C

by
Lei Shi
1,
Qing-Wen Wang
1,2,*,
Lv-Ming Xie
1 and
Xiao-Feng Zhang
3
1
Department of Mathematics and Newtouch Center for Mathematics, Shanghai University, Shanghai 200444, China
2
Collaborative Innovation Center for the Marine Artificial Intelligence, Shanghai 200444, China
3
Shanghai Newtouch Software Co., Ltd., Shanghai 200127, China
*
Author to whom correspondence should be addressed.
Symmetry 2024, 16(10), 1359; https://doi.org/10.3390/sym16101359
Submission received: 11 September 2024 / Revised: 6 October 2024 / Accepted: 11 October 2024 / Published: 13 October 2024
(This article belongs to the Special Issue Exploring Symmetry in Dual Quaternion Matrices and Matrix Equations)

Abstract

:
Dual generalized commutative quaternions have broad application prospects in many fields. Additionally, the matrix equation A X B = C has important applications in mathematics and engineering, especially in control systems, economics, computer science, and other disciplines. However, research on the matrix equation A X B = C over the dual generalized commutative quaternions remains relatively insufficient. In this paper, we derive the necessary and sufficient conditions for the solvability of the dual generalized commutative quaternion matrix equation A X B = C . Furthermore, we provide the general solution expression for this matrix equation, when it is solvable. Finally, a numerical algorithm and an example are provided to confirm the reliability of the main conclusions.

1. Introduction

In 1843, the renowned mathematician William Rowan Hamilton [1] introduced the concept of quaternions, making a groundbreaking discovery. The set of quaternions is typically denoted as
H = { q = q 0 + q 1 i + q 2 j + q 3 k : i 2 = j 2 = k 2 = 1 , i j k = 1 , q 0 , q 1 , q 2 , q 3 R } .
It is easy to verify that H is a four-dimensional non-commutative division ring over the real number field R . Quaternions exhibit a diverse array of applications, spanning not only the realm of mathematics but also playing a pivotal role in mechanics, quantum physics, signal processing, and numerous other disciplines [2,3,4]. Nevertheless, the non-commutativity of quaternion multiplication poses significant challenges in academic research. Corrado Segre [5] expanded on the definition of quaternions, overcoming the difficulty that quaternions are not commutative and proposing the concept of commutative quaternions of the following form.
S = { a = a 0 + a 1 i + a 2 j + a 3 k : a 0 , a 1 , a 2 , a 3 R } ,
here i , j , k satisfying the following conditions.
i 2 = k 2 = 1 , j 2 = 1 , i j = j i = k , j k = k j = i , k i = i k = j .
Presently, commutative quaternions have garnered widespread application in the domains of digital signal and image processing [6,7,8,9]. Tian et al. [10] expanded upon the notion of commutative quaternions, introducing the concept of generalized commutative quaternions, which is defined as
S g = { a = a 0 + a 1 i + a 2 j + a 3 k : a 0 , a 1 , a 2 , a 3 R } ,
where i , j and k satisfy the following fundamental properties.
i 2 = α , j 2 = β , k 2 = i j k = α β , i j = j i = k , j k = k j = β i , k i = i k = α j , α , β R { 0 } .
Specifically, when α = 1 , β = 1 , the generalized commutative quaternion algebra S g reduces to the commutative quaternion algebra S . Similarly, the set of generalized commutative quaternion matrices can be written as
S g m × n = { X = X 0 + X 1 i + X 2 j + X 3 k : X 0 , X 1 , X 2 , X 3 R m × n } ,
here i , j and k satisfy the same conditions as above.
In 1873, Clifford [11] introduced the concept of dual numbers as an extension of real numbers. This is a development that has garnered significant attention due to their pivotal role in kinematic synthesis, robotics, and numerous other disciplines [12,13]. As an extension of quaternions, dual quaternions have found extensive applications in theoretical kinematics, 3D computer graphics, robotics, and numerous other domains [14,15,16]. Similarly, we can further extend the generalized commutative quaternion concept by incorporating dual numbers, thus arriving at the concept of the dual generalized commutative quaternion. For the definition of dual numbers and dual generalized commutative quaternions, please refer to Section 2.
As everyone knows, the classical complex matrix equation
A X B = C
has a wide range of applications in the field of control systems, which has attracted the attention of many scholars. Roger Penrose [17] provided the general solution and solvable conditions for the matrix Equation ( 1 ) . In 2003, Liao and Bai [18] conducted in-depth research on the least squares solution of matrix Equation ( 1 ) based on their work, especially with regard to symmetric positive semidefinite matrices. Subsequently, the centrosymmetric solution to the matrix Equation ( 1 ) was formulated by Peng [19]. Building upon this foundation, Deng et al. [20] delved into the generalized formulation of the Hermitian solution pertaining to the same matrix equation. Moreover, many scholars have studied various particular solutions of matrix equation (1), as detailed in references [21,22,23,24,25]. Additionally, corresponding numerical algorithms for these specific solutions have been proposed, as discussed in references [26,27,28,29,30]. Later, the research on matrix Equation ( 1 ) was gradually extended to the quaternion algebras. Ivan Kyrchei [31] first derived explicit determinantal representation formulas for the solution of the quaternionic matrix Equation (1). Subsequently, Ivan Kyrchei [32] formulated determinantal representation formulas for the quaternion two-sided generalized Sylvester matrix equation A 1 X 1 B 1 + A 2 X 2 B 2 = C . Recently, Chen et al. [33] shifted the perspective to the dual quaternions, providing the necessary and sufficient conditions for the equation to be solvable. In addition, Si and Wang [34] studied the solvability and general solution over dual split quaternions. At present, the research results are still being further expanded.
Up until now, there is a paucity of research information pertaining to the Equation ( 1 ) in the context of dual generalized commutative quaternions. Therefore, this article aims to establish the necessary and sufficient conditions for the solvability of the Equation ( 1 ) over dual generalized commutative quaternions; furthermore, we present the expression of the general solution of the Equation ( 1 ) when it is consistent.
For clarity, hereby introduce some symbols. Let R m × n , S g m × n denote the set of all m × n matrices with real-valued entries and the set of all m × n matrices whose elements belong to the set S g , respectively. For A R m × n , its transpose, conjugate transpose, and rank are represented by A T , A * and r a n k ( A ) , respectively. Additionally, I and O refer to the identity matrix and zero matrix, respectively. The Kronecker product of two matrices A = ( a i j ) R m × n and B = ( b i j ) R s × t is defined as A B = ( a i j B ) R m s × n t . The vectorization operator is denoted by Vec ( A ) , which is expressed as Vec ( A ) = ( x 1 T , x 2 T , . . . , x n T ) , where x i ( i = 1 , 2 , , n ) represents the i-th column vector of A. The Moore-Penrose inverse is denoted as A and satisfies the following equations.
A A A = A , A A A = A , ( A A ) * = A A , ( A A ) * = A A .
Lastly, we denote L A = I A A , R A = I A A .
The structure of this article is outlined as follows. We devote Section 2 to revisiting several definitions, fundamental properties and lemmas that serve as the foundation for our subsequent analysis. In Section 3, we derive the necessary and sufficient conditions for the solvability of the matrix Equation (1) over the dual generalized commutative quaternions. Furthermore, we present the general expression for the solution when it is consistent. Finally, accompanying algorithms and numerical examples are offered in Section 4 to demonstrate the application and validity of our theoretical results.

2. Preliminaries

This section delves into a review of definitions about dual numbers, dual generalized commutative quaternions, and their attendant propositions. Furthermore, we also introduce some lemmas and properties that will be used in subsequent proofs.
Definition 1
([35]). Suppose that x 0 , x 1 R , we say that x is a dual number if x is in the form
x = x 0 + x 1 ϵ ,
where ϵ is an infinitesimal unit, satisfying ϵ 2 = 0 . Denote that x 0 and x 1 as the real part of x and the dual part of x, respectively.
The multiplication of infinitesimal unit ϵ with real numbers, complex numbers, and quaternions exhibits commutativity. The collection of dual numbers is denoted as
D = { x = x 0 + x 1 ϵ | x 0 , x 1 R , ϵ 2 = 0 } .
For x = x 0 + x 1 ϵ , y = y 0 + y 1 ϵ D , if x 0 = y 0 and x 1 = y 1 , then x = y . The addition and multiplication operations of dual numbers satisfy the following properties.
x + y = x 0 + y 0 + ( x 1 + y 1 ) ϵ , x y = x 0 y 0 + ( x 0 y 1 + x 1 y 0 ) ϵ .
Definition 2.
If z 0 , z 1 S g , then z is a dual generalized commutative quaternion if z is in the form
z = z 0 + z 1 ϵ
where ϵ is an infinitesimal unit, satisfying ϵ 2 = 0 .
The set of dual generalized commutative quaternions is formally defined as
DS g = { z = z 0 + z 1 ϵ | ϵ 2 = 0 , z 0 , z 1 S g } .
The addition and multiplication rules for the dual generalized commutative quaternions are similar to the dual number.
Let X 0 , X 1 S g m × n .  X is called a dual generalized commutative quaternion matrix if it has the form X = X 0 + X 1 ϵ . We use DS g m × n to represent all dual generalized commutative quaternion matrices. For X , Y DS g m × n , we have X = Y , if X 0 = Y 0 , X 1 = Y 1 . Furthermore,
X + Y = X 0 + Y 0 + ( X 1 + Y 1 ) ϵ , X Y = X 0 Y 0 + ( X 0 Y 1 + X 1 Y 0 ) ϵ .
Definition 3
([36]). For X = X 0 + X 1 i + X 2 j + X 3 k S g m × n , where X 0 , X 1 , X 2 , X 3 R m × n , α , β R { 0 } , there are three kinds of real representations of X denoted by X σ η , η { i , j , k } , where
X σ i = X 0 α X 1 β X 2 α β X 3 X 1 X 0 β X 3 β X 2 X 2 α X 3 X 0 α X 1 X 3 X 2 X 1 X 0 ,
X σ j = V m X σ i = X 0 α X 1 β X 2 α β X 3 X 1 X 0 β X 3 β X 2 X 2 α X 3 X 0 α X 1 X 3 X 2 X 1 X 0 ,
X σ k = W m X σ i = X 0 α X 1 β X 2 α β X 3 X 1 X 0 β X 3 β X 2 X 2 α X 3 X 0 α X 1 X 3 X 2 X 1 X 0 .
Here,
V m = I m 0 0 0 0 I m 0 0 0 0 I m 0 0 0 0 I m ,
W m = I m 0 0 0 0 I m 0 0 0 0 I m 0 0 0 0 I m .
.
The following lemmas are useful in the subsequent proof process.
Lemma 1
([36,37]). For X , Y S g m × n , Z S g n × s , λ R , α R { 0 } and β R { 0 } , set
G n = I n 0 0 0 0 α I n 0 0 0 0 β I n 0 0 0 0 α β I n ,
R n = 0 α I n 0 0 I n 0 0 0 0 0 0 α I n 0 0 I n 0 ,
S n = 0 0 β I n 0 0 0 0 β I n I n 0 0 0 0 I n 0 0 ,
T n = 0 0 0 α β I n 0 0 β I n 0 0 α I n 0 0 I n 0 0 0 .
Then the following statements are satisfied.
( 1 ) X = Y X σ η = Y σ η , η { i , j , k } ,
( 2 ) ( X + Y ) σ η = X σ η + Y σ η , η { i , j , k } ,
( 3 ) ( λ X ) σ η = ( X λ ) σ η = λ X σ η , η { i , j , k } ,
( 4 ) ( X Z ) σ i = X σ i Z σ i , ( X Z ) σ j = X σ j V n Z σ j , ( X Z ) σ k = X σ k W n Z σ k , V n and W n are in the form of ( 2 ) and ( 3 ) , respectively,
( 5 ) ( a ) R m 1 X σ i R n = X σ i , S m 1 X σ i S n = X σ i , T m 1 X σ i T n = X σ i , ( b ) R m 1 X σ j R n = X σ j , S m 1 X σ j S n = X σ j , T m 1 X σ j T n = X σ j , ( c ) R m 1 X σ k R n = X σ k , S m 1 X σ k S n = X σ k , T m 1 X σ k T n = X σ k ,
( 6 ) X i * σ i = G n 1 X σ i T G m , X j * σ j = G n 1 X σ j T G m , X k * σ k = G n 1 X σ k T G m ,
( 7 ) ( a ) X = 1 4 I m I m i I m j I m k X σ i I n 1 α I n i 1 β I n j 1 α β I n k , ( b ) X = 1 4 I m I m i I m j I m k X σ j I n 1 α I n i 1 β I n j 1 α β I n k , ( c ) X = 1 4 I m I m i I m j I m k X σ k I n 1 α I n i 1 β I n j 1 α β I n k ,
( 8 ) ( a ) R n V n = V n R n , R n G n 1 = G n 1 R n T , R n W n = W n R n , R n 1 = 1 α R n , ( b ) S n V n = V n S n , S n G n 1 = G n 1 S n T , S n W n = W n S n , S n 1 = 1 β S n , ( c ) T n V n = V n T n , T n G n 1 = G n 1 T n T , T n W n = W n T n , T n 1 = 1 α β T n .
Lemma 2
([38]). Let A R m × s , B R r × n , C R m × n . Then the following statements are equivalent:
( 1 ) The matrix equation A X B = C is consistent.
( 2 )   A A C B B = C .
( 3 )   A A C = C , C B B = C .
In that case, the general solution of the matrix equation A X B = C can be expressed as
X = A C B + L A V + U R B ,
where U , V are arbitrary real matrices with appropriate order.
Lemma 3
([17]). Suppose that A R m × n , b R m , then the matrix equation
A x = b
is consistent if and only if
A A b = b .
In that case, the general solution to the matrix Equation ( 4 ) can be expressed as x = A b + L A u , where u is an arbitrary real vector with appropriate size. Moreover, the matrix Equation ( 4 ) has a unique solution x = A b if r a n k ( A ) = n .
Lemma 4
([39]). For A C m × n , B C n × s , and C C s × t , then we obtain
Vec A B C = C T A Vec ( B ) .

3. Solving the Matrix Equation ( 1 ) over DS g

In this section, we consider solving matrix Equation ( 1 ) in combination with the previous lemmas.
Theorem 1.
Let A = A 0 + A 1 ϵ DS g m × n , B = B 0 + B 1 ϵ DS g n × s , C = C 0 + C 1 ϵ DS g m × s . Set
A 0 η = A 0 σ i , η = i A 0 σ j V n , η = j A 0 σ k W n , η = k , A 1 η = A 1 σ i , η = i A 1 σ j V n , η = j A 1 σ k W n , η = k , B 0 η = B 0 σ i , η = i B 0 σ j V s , η = j B 0 σ k W s , η = k , B 1 η = B 1 σ i , η = i B 1 σ j V s , η = j B 1 σ k W s , η = k ,
C 0 η = C 0 σ η , C 1 η = C 1 σ η ,
C 00 = A 0 η A 0 η C 0 η B 0 η B 1 η + A 1 η A 0 η C 0 η B 0 η B 0 η ,
A 00 = A 1 η L A 0 η , B 00 = R B 0 η B 1 η ,
L = B 0 η T A 00 , M = B 00 T A 0 η , N = B 0 η T A 0 η ,
a = Vec ( V ) , b = Vec ( U ) , c = Vec ( X 1 η ) , σ = a b c ,
Q = L M N , d = Vec C 1 η C 00 ,
V n and V s are in the form of ( 2 ) , while W n and W s are in the form of ( 3 ) . Then, the following statements are equivalent:
( 1 ) The dual generalized commutative quaternion matrix Equation ( 1 ) is consistent.
( 2 ) The system of matrix equations
A 0 η X 0 η B 0 η = C 0 η , A 0 η X 0 η B 1 η + A 0 η X 1 η B 0 η + A 1 η X 0 η B 0 η = C 1 η .
is consistent.
( 3 ) The following equations
A 0 η A 0 η C 0 η B 0 η B 0 η = C 0 η , A 0 η A 0 η C 0 η = C 0 η , C 0 η B 0 η B 0 η = C 0 η , Q Q d = d ,
hold. In this case, the general solution to the dual generalized commutative quaternion matrix Equation ( 1 ) can be formulated as
X = X 0 + X 1 ϵ ,
( a ) when η = i ,
X 0 = 1 16 I n I n i I n j I n k X 0 i + R n 1 X 0 i R n + S n 1 X 0 i S n + T n 1 X 0 i T n I n 1 α I n i 1 β I n j 1 α β I n k ,
X 1 = 1 16 I n I n i I n j I n k X 1 i + R n 1 X 1 i R n + S n 1 X 1 i S n + T n 1 X 1 i T n I n 1 α I n i 1 β I n j 1 α β I n k ,
( b ) when η = j ,
X 0 = 1 16 I n I n i I n j I n k X 0 j R n 1 X 0 j R n S n 1 X 0 j S n + T n 1 X 0 j T n I n 1 α I n i 1 β I n j 1 α β I n k ,
X 1 = 1 16 I n I n i I n j I n k X 1 j R n 1 X 1 j R n S n 1 X 1 j S n + T n 1 X 1 j T n I n 1 α I n i 1 β I n j 1 α β I n k ,
( c ) when η = k ,
X 0 = 1 16 I n I n i I n j I n k X 0 k + R n 1 X 0 k R n S n 1 X 0 k S n T n 1 X 0 k T n I n 1 α I n i 1 β I n j 1 α β I n k ,
X 1 = 1 16 I n I n i I n j I n k X 1 k + R n 1 X 1 k R n S n 1 X 1 k S n T n 1 X 1 k T n I n 1 α I n i 1 β I n j 1 α β I n k .
In above ( a ) , ( b ) and ( c ) ,
X 0 η = A 0 η C 0 η B 0 η + L A 0 η V + U R B 0 η ,
σ = a b c = Vec ( V ) Vec ( U ) Vec ( X 1 η ) = Q d + L Q u , u R 48 n 2 ,
where U , V , X 1 η , u are arbitrary matrices over R with appropriate sizes.
Proof. 
Firstly, we demonstrate ( 1 ) ( 2 ) . We prove the case of η = i in detail, and the other two cases can be proved similarly.
Assuming X DS g n × n is the solution of the dual generalized commutative quaternion matrix Equation ( 1 ) , then X can be expressed as
X = X 0 + X 1 ϵ .
Substituting ( 5 ) into the matrix Equation ( 1 ) , we obtain
( 1 ) A 0 + A 1 ϵ X 0 + X 1 ϵ B 0 + B 1 ϵ = C 0 + C 1 ϵ , A 0 X 0 B 0 + A 0 X 0 B 1 + A 0 X 1 B 0 + A 1 X 0 B 0 ϵ = C 0 + C 1 ϵ ,
i.e.,
A 0 X 0 B 0 = C 0 , A 0 X 0 B 1 + A 0 X 1 B 0 + A 1 X 0 B 0 = C 1 .
By applying the formula ( 2 ) and ( 4 ) of Lemma 1 to the system ( 6 ) , we obtain system
A 0 i X 0 i B 0 i = C 0 i , A 0 i X 0 i B 1 i + A 0 i X 1 i B 0 i + A 1 i X 0 i B 0 i = C 1 i .
At this time, ( X 0 i , X 1 i ) is the solution of the system ( 7 ) .
Conversely, if the system ( 7 ) has a solution ( X 0 i , X 1 i ) , which can be expressed as
X 0 i = a 11 a 12 a 13 a 14 a 21 a 22 a 23 a 24 a 31 a 32 a 33 a 34 a 41 a 42 a 43 a 44 , a u v R n × n , u , v = 1 , 2 , 3 , 4 ,
X 1 i = b 11 b 12 b 13 b 14 b 21 b 22 b 23 b 24 b 31 b 32 b 33 b 34 b 41 b 42 b 43 b 44 , b u v R n × n , u , v = 1 , 2 , 3 , 4 .
Using the formula ( 5 ) of Lemma 1, we derive
A 0 i R n 1 X 0 i R n B 0 i = C 0 i , A 0 i R n 1 X 0 i R n B 1 i + A 0 i R n 1 X 1 i R n B 0 i + A 1 i R n 1 X 0 i R n B 0 i = C 1 i .
Therefore, ( R n 1 X 0 i R n , R n 1 X 1 i R n ) is also a solution to the system ( 7 ) . Similarly, ( S n 1 X 0 i S n , S n 1 X 1 i S n ) and ( T n 1 X 0 i T n , T n 1 X 1 i T n ) are also solutions to the system ( 7 ) .
Let
Z 1 = 1 4 ( X 0 i + R n 1 X 0 i R n + S n 1 X 0 i S n + T n 1 X 0 i T n ) , Z 2 = 1 4 ( X 1 i + R n 1 X 1 i R n + S n 1 X 1 i S n + T n 1 X 1 i T n ) .
Then ( Z 1 , Z 2 ) is also a solution to the system ( 7 ) . That is
A 0 i Z 1 B 0 i = C 0 i , A 0 i Z 1 B 1 i + A 0 i Z 2 B 0 i + A 1 i Z 1 B 0 i = C 1 i .
By direct computation, we obtain
Z 1 = c 0 α c 1 β c 2 α β c 3 c 1 c 0 β c 3 β c 2 c 2 α c 3 c 0 α c 1 c 3 c 2 c 1 c 0 ,
where
c 0 = 1 4 ( a 11 + a 22 + a 33 + a 44 ) , c 1 = 1 4 1 α a 12 + 1 α a 34 + a 21 + a 43 , c 2 = 1 4 1 β a 13 + 1 β a 24 + a 31 + a 42 , c 3 = 1 4 1 α β a 14 + 1 β a 23 + 1 α a 32 + a 41 ,
and
Z 2 = d 0 α d 1 β d 2 α β d 3 d 1 d 0 β d 3 β d 2 d 2 α d 3 d 0 α d 1 d 3 d 2 d 1 d 0 .
Here
d 0 = 1 4 ( b 11 + b 22 + b 33 + b 44 ) , d 1 = 1 4 1 α b 12 + 1 α b 34 + b 21 + b 43 , d 2 = 1 4 1 β b 13 + 1 β b 24 + b 31 + b 42 , d 3 = 1 4 1 α β b 14 + 1 β b 23 + 1 α b 32 + b 41 .
Construct X 0 , X 1 S g n × n , and they satisfy
X 0 σ i = X 0 i = Z 1 , X 1 σ i = X 1 i = Z 2 ,
i.e., ( X 0 i , X 1 i ) is the solution of the system ( 7 ) . In addition, applying the formula ( 2 ) and ( 4 ) of Lemma 1 to the system ( 7 ) , we get the system ( 6 ) . Therefore,
X = X 0 + X 1 ϵ ,
is the solution of the matrix Equation ( 1 ) . In this case, we employ the item ( 7 ) in Lemma 1 to express the specific forms of X. It can be written as
X 0 = c 0 + c 1 i + c 2 j + c 3 k = 1 4 I n I n i I n j I n k Z 1 I n 1 α I n i 1 β I n j 1 α β I n k = 1 16 I n I n i I n j I n k ( X 0 i + R n 1 X 0 i R n + S n 1 X 0 i S n + T n 1 X 0 i T n ) I n 1 α I n i 1 β I n j 1 α β I n k ,
X 1 = d 0 + d 1 i + d 2 j + d 3 k = 1 4 I n I n i I n j I n k Z 2 I n 1 α I n i 1 β I n j 1 α β I n k = 1 16 I n I n i I n j I n k ( X 1 i + R n 1 X 1 i R n + S n 1 X 1 i S n + T n 1 X 1 i T n ) I n 1 α I n i 1 β I n j 1 α β I n k .
Next, we divide the system ( 7 ) into two parts,
A 0 i X 0 i B 0 i = C 0 i ,
and
A 0 i X 0 i B 1 i + A 0 i X 1 i B 0 i + A 1 i X 0 i B 0 i = C 1 i .
Applying the Lemma 2 to the matrix Equation ( 8 ) yields
X 0 i = A 0 i C 0 i B 0 i + L A 0 i V + U R B 0 i ,
where U , V R 4 n × 4 n are arbitrary matrices. Substitute (10) into the ( 9 ) to obtain
A 00 V B 0 i + A 0 i U B 00 + A 0 i X 1 i B 0 i = C 1 i C 00 .
Using the Vec operator on both sides of Equation (11) and Lemma 4, we derive
B 0 i T A 00 Vec ( V ) + B 00 T A 0 i Vec ( U ) + B 0 i T A 0 i Vec X 1 i = Vec C 1 i C 00 .
Therefore, ( 12 ) can be expressed as
L a + M b + N c = d .
That is
Q σ = d .
Thus, ( 13 ) is solvable if and only if
Q Q d = d .
At this time, the general solution of matrix Equation (13) is given by
σ = Q d + L Q u ,
where u R 48 n 2 is an arbitrary vector.
Finally, utilizing Lemmas 2 and 3, we establish the equivalence between ( 2 ) ( 3 ) easily. □

4. Numerical Example

To further elaborate on the key findings of this article, we now present a numerical algorithm and an example.
Example 1.
Let α = 1 , β = 1 . Take η = i as an example, the other two cases are similar.
A = A 0 + A 1 ϵ = 1 i 2 j 2 2 k + i + k j + k ϵ , B = B 0 + B 1 ϵ = 1 + 2 j + k i + 2 k i + 2 k 2 + i + 2 j + k + 1 + 2 i + j + k 1 + j + 2 k 2 i + j 1 + i + j + 3 k ϵ , C = C 0 + C 1 ϵ = 1 + 2 i + j + k 1 + 4 j + 3 k + 1 + i + j + 2 k i + 2 j + 2 k ϵ .
Utilizing MATLAB8.6 and Algorithm, we obtain
A 0 i A 0 i C 0 i B 0 i B 0 i C 0 i = 8.8668 × 10 15 , A 0 i A 0 i C 0 i C 0 i = 2.9286 × 10 15 , C 0 i B 0 i B 0 i C 0 i = 8.5821 × 10 15 , Q Q d d = 1.4093 × 10 14 .
In this case, the general solution of the matrix Equation ( 1 ) can be formulated as
X = X 0 + X 1 ϵ ,
where
X 0 = 1 16 I 2 I 2 i I 2 j I 2 k ( X 0 i + R 2 1 X 0 i R 2 + S 2 1 X 0 i S 2 + T 2 1 X 0 i T 2 ) I 2 1 α I 2 i 1 β I 2 j 1 α β I 2 k , X 1 = 1 16 I 2 I 2 i I 2 j I 2 k ( X 1 i + R 2 1 X 1 i R 2 + S 2 1 X 1 i S 2 + T 2 1 X 1 i T 2 ) I 2 1 α I 2 i 1 β I 2 j 1 α β I 2 k ,
X 0 i = 0.12573 0.043649 0.17884 0.022075 0.17501 0.13434 0.17171 0.0257 0.049808 0.18161 0.042146 0.042912 0.042146 0.042912 0.049808 0.18161 0.17884 0.022075 0.12573 0.043649 0.17171 0.0257 0.17501 0.13434 0.042146 0.042912 0.049808 0.18161 0.049808 0.18161 0.042146 0.042912 0.17501 0.13434 0.17171 0.0257 0.12573 0.043649 0.17884 0.022075 0.042146 0.042912 0.049808 0.18161 0.049808 0.18161 0.042146 0.042912 0.17171 0.0257 0.17501 0.13434 0.17884 0.022075 0.12573 0.043649 0.049808 0.18161 0.042146 0.042912 0.042146 0.042912 0.049808 0.18161 + L A 0 i V + U R B 0 i ,
Vec ( X 1 i ) = X 1 i 1 X 1 i 2 X 1 i 3 X 1 i 4 ,
X 1 i 1 = 0.13404 0.029119 0.090283 0.11942 0.13544 0.11942 0.20831 0.029119 0.14232 0.021957 0.058871 0.099797 0.097791 0.099797 0.081445 0.021957 , X 1 i 2 = 0.090283 0.11942 0.13404 0.029119 0.20831 0.029119 0.13544 0.11942 0.058871 0.099797 0.14232 0.021957 0.081445 0.021957 0.097791 0.099797 , X 1 i 3 = 0.13544 0.11942 0.20831 0.029119 0.13404 0.029119 0.090283 0.11942 0.097791 0.099797 0.081445 0.021957 0.14232 0.021957 0.058871 0.099797 , X 1 i 4 = 0.20831 0.029119 0.13544 0.11942 0.090283 0.11942 0.13404 0.029119 0.081445 0.021957 0.097791 0.099797 0.058871 0.099797 0.14232 0.021957 .
Here U , V are arbitrary matrices over R with appropriate sizes.

5. Conclusions

In this article, we conducted an in-depth study of the solution to the dual generalized quaternion matrix Equation (1), and derived a general solution expression when it is consistent. This research not only provides a new perspective and method for understanding and solving this special type of matrix equation, but also contributes new knowledge to the field of dual generalized commutative quaternion algebra. To show our research results more intuitively, we provide a specific numerical example. Through this example, we can clearly see how to apply the existence conditions and general solution expressions to solve the dual generalized quaternion matrix equation.
We will study more complex matrix equations and tensor equations over the dual generalized commutative quaternions in the future, which may have broader applications in physics and engineering fields.

Author Contributions

Conceptualization, L.S. and Q.-W.W.; methodology, L.S. and Q.-W.W.; software, L.S., X.-F.Z. and L.-M.X.; validation, L.S.; formal analysis, L.S.; investigation, L.S. and L.-M.X.; resources, Q.-W.W.; data curation, L.S.; writing—original draft preparation, Q.-W.W. and L.S.; writing—review and editing, Q.-W.W., L.S. and L.-M.X.; supervision, Q.-W.W.; project administration, Q.-W.W.; funding acquisition, Q.-W.W. All authors have read and agreed to the published version of the manuscript.

Funding

This research is supported by the National Natural Science Foundation of China (No. 12371023).

Data Availability Statement

Data are contained within the article.

Conflicts of Interest

Author Xiao-Feng Zhang was employed by the company Shanghai Newtouch Software Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. Hamilton, W.R. Lectures on Quaternions. In Landmark Writings in Western Mathematics 1640–1940; Hodges and Smith: Dublin, Ireland, 1853. [Google Scholar] [CrossRef]
  2. Assefa, D.; Mansinha, L.; Tiampo, K.F.; Rasmussen, H.; Abdella, K. Local quaternion Fourier transform and color image texture analysis. Signal Process. 2010, 90, 1825–1835. [Google Scholar] [CrossRef]
  3. Bihan, N.L.E.; Mars, J. Singular value decomposition of matrices of quaternions: A new tool for vector-sensor signal processing. IEEE Trans. Signal Process. 2004, 84, 1177–1199. [Google Scholar] [CrossRef]
  4. Cyrus, J.; Clive, C.T.; Danilo, P.M. A class of quaternion valued affine projection algorithms. Signal Process. 2013, 93, 1712–1723. [Google Scholar] [CrossRef]
  5. Segre, C. The real representations of complex elements and extension to bicomplex systems. Math. Ann. 1892, 40, 413–467. [Google Scholar] [CrossRef]
  6. Catoni, F. Commutative (Segre’s) quaternion fields and relation with Maxwell equations. Adv. Appl. Clifford Algebras 2008, 18, 9–28. [Google Scholar] [CrossRef]
  7. Guo, L.; Zhu, M.; Ge, X. Reduced biquaternion canonical transform, convolution and correlation. Signal Process. 2011, 91, 2147–2153. [Google Scholar] [CrossRef]
  8. Isokawa, T.; Nishimura, H.; Matsui, N. Commutative quaternion and multistate Hopfield neural networks. In Proceedings of the The 2010 International Joint Conference on Neural Networks (IJCNN), Barcelona, Spain, 18–23 July 2010; pp. 1–6. [Google Scholar]
  9. Pei, S.C.; Chang, J.H.; Ding, J.J. Commutative reduced biquaternions and their Fourier transform for signal and image processing applications. IEEE Trans. Signal Process. 2004, 52, 2012–2031. [Google Scholar] [CrossRef]
  10. Tian, Y.; Liu, X.; Zhang, Y. Least-squares solutions of the generalized reduced biquaternion matrix equations. Filomat 2023, 37, 863–870. [Google Scholar] [CrossRef]
  11. Clifford, W.K. Preliminary sketch of bi-quaternions. Proc. Lond. Math. Soc. 1873, 4, 381–395. [Google Scholar] [CrossRef]
  12. Angeles, J. The dual generalized inverses and their applications in kinematic synthesis. In Latest Advances in Robot Kinematics; Springer: Dordrecht, The Netherlands, 2022. [Google Scholar]
  13. Gu, Y.L.; Luh, J. Dual-number transformation and its applications to robotics. IEEE J. Robot. Autom. 1987, 3, 615–623. [Google Scholar] [CrossRef]
  14. Brambley, G.; Kim, J. Unit dual quaternion-based pose optimization for visual runway observations. IET Cyber Syst. Robot. 2020, 2, 181–189. [Google Scholar] [CrossRef]
  15. Cheng, J.; Kim, J.; Jiang, Z.; Che, W. Dual quaternion-based graph SLAM. Robot. Auton. Syst. 2016, 77, 15–24. [Google Scholar] [CrossRef]
  16. Wang, X.; Yu, C.; Lin, Z. A dual quaternion solution to attitude and position control for rigid body coordination. IEEE Trans. Rob. 2012, 28, 1162–1170. [Google Scholar] [CrossRef]
  17. Penrose, R. A generalized inverse for matrices. Math. Proc. Camb. Philos. Soc. 1955, 51, 406–413. [Google Scholar] [CrossRef]
  18. Liao, A.P.; Bai, Z.Z. The constrained solutions of two matrix equations. Acta Math. Appl. Sin. Engl. Ser. 2002, 18, 671–678. [Google Scholar] [CrossRef]
  19. Peng, Z.Y. The centro-symmetric solutions of linear matrix equation AXB = C and its optimal approximation. Chin. J. Eng. Math. 2003, 20, 60–64. [Google Scholar] [CrossRef]
  20. Deng, Y.B.; Bai, Z.Z.; Gao, Y.H. Iterative orthogonal direction methods for Hermitian minimum norm solutions of two consistent matrix equations. Numer. Linear Algebra Appl. 2006, 13, 801–823. [Google Scholar] [CrossRef]
  21. Cvetković-ilić, D.S. The reflexive solutions of the matrix equation AXB = C. Comput. Math. Appl. 2006, 51, 897–902. [Google Scholar] [CrossRef]
  22. Cvetković-ilić, D.S. Re-nnd solutions of the matrix equation AXB = C. J. Aust. Math. Soc. 2008, 84, 63–72. [Google Scholar] [CrossRef]
  23. Hu, S.; Yuan, Y. The symmetric solution of the matrix equation AXB = D on subspace. Comput. Appl. Math. 2022, 41, 373. [Google Scholar] [CrossRef]
  24. Ji, Z.D.; Li, J.F.; Zhou, X.L.; Duan, F.J.; Li, T. On solutions of matrix equation AXB = C under semi-tensor product. Linear Multilinear Algebra 2021, 69, 1935–1963. [Google Scholar] [CrossRef]
  25. Yang, J.; Deng, Y.B. On the solutions of the equation AXB = C under Toeplitz-like and Hankel matrices constraint. Math. Meth. Appl. Sci. 2018, 41, 2074–2094. [Google Scholar] [CrossRef]
  26. Chen, F.; Li, T.Y. Two-step AOR iteration method for the linear matrix equation AXB = C. Comp. Appl. Math. 2021, 40, 89. [Google Scholar] [CrossRef]
  27. Hou, J.J.; Peng, Z.Y.; Zhang, Z.L. An iterative method for the least squares symmetric solution of matrix equation AXB = C. Numer. Algorithms 2006, 42, 181–192. [Google Scholar] [CrossRef]
  28. Li, J.F.; Hu, X.Y.; Zhang, L. Numerical solutions of AXB = C for centrosymmetric matrix X under a specified submatrix constraint. Numer. Linear Algebra Appl. 2011, 18(5), 857–873. [Google Scholar] [CrossRef]
  29. Liu, Z.Y.; Li, Z.; Ferreira, C.; Zhang, Y.L. Stationary splitting iterative methods for the matrix equation AXB = C. Appl. Math. Comput. 2020, 378, 125195. [Google Scholar] [CrossRef]
  30. Peng, Z.Y. New matrix iterative methods for constraint solutions of the matrix equation AXB = C. J. Comput. Appl. Math. 2010, 235, 726–735. [Google Scholar] [CrossRef]
  31. Kyrchei, I.I. Explicit determinantal representation formulas for the solution of the two-sided restricted quaternionic matrix equation. J. Appl. Math. Comput. 2018, 58, 335–365. [Google Scholar] [CrossRef]
  32. Kyrchei, I. Cramer’s Rules for Sylvester Quaternion Matrix Equation and Its Special Cases. Adv. Appl. Clifford Algebras 2018, 28, 90. [Google Scholar] [CrossRef]
  33. Chen, Y.; Wang, Q.W.; Xie, L.M. Dual quaternion matrix equation AXB = C with applications. Symmetry 2024, 16, 287. [Google Scholar] [CrossRef]
  34. Si, K.W.; Wang, Q.W. The general solution to a classical matrix equation AXB = C over the dual split quaternion algebra. Symmetry 2024, 16, 491. [Google Scholar] [CrossRef]
  35. Qi, L.Q.; Ling, C.; Yan, H. Dual quaternions and dual quaternion vectors. Commun. Appl. Math. Comput. 2022, 4, 1494–1508. [Google Scholar] [CrossRef]
  36. Ren, B.Y.; Wang, Q.W.; Chen, X.Y. The η-anti-Hermitian solution to a constrained matrix equation over the generalized Segre quaternion algebra. Symmetry 2023, 15, 592. [Google Scholar] [CrossRef]
  37. Kösal, H.H.; Tosun, M. Universal similarity factorization equalities for commutative quaternions and their matrices. Linear Multilinear A 2019, 67, 926–938. [Google Scholar] [CrossRef]
  38. Wang, Q.W. A system of matrix equations and a linear matrix equation over arbitrary regular rings with identity. Linear Algebra Appl. 2004, 384, 43–54. [Google Scholar] [CrossRef]
  39. Yuan, S.F.; Wang, Q.W.; Yu, Y.B.; Tian, Y. On hermitian solutions of the split quaternion matrix equation AXB + CXD = E. Adv. Appl. Clifford Algebras 2017, 27, 3235–3252. [Google Scholar] [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Shi, L.; Wang, Q.-W.; Xie, L.-M.; Zhang, X.-F. Solving the Dual Generalized Commutative Quaternion Matrix Equation AXB = C. Symmetry 2024, 16, 1359. https://doi.org/10.3390/sym16101359

AMA Style

Shi L, Wang Q-W, Xie L-M, Zhang X-F. Solving the Dual Generalized Commutative Quaternion Matrix Equation AXB = C. Symmetry. 2024; 16(10):1359. https://doi.org/10.3390/sym16101359

Chicago/Turabian Style

Shi, Lei, Qing-Wen Wang, Lv-Ming Xie, and Xiao-Feng Zhang. 2024. "Solving the Dual Generalized Commutative Quaternion Matrix Equation AXB = C" Symmetry 16, no. 10: 1359. https://doi.org/10.3390/sym16101359

APA Style

Shi, L., Wang, Q.-W., Xie, L.-M., & Zhang, X.-F. (2024). Solving the Dual Generalized Commutative Quaternion Matrix Equation AXB = C. Symmetry, 16(10), 1359. https://doi.org/10.3390/sym16101359

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop