Next Article in Journal
LPI Radar Waveform Recognition Based on Deep Convolutional Neural Network Transfer Learning
Next Article in Special Issue
An Efficient Algorithm for Nontrivial Eigenvectors in Max-Plus Algebra
Previous Article in Journal
A Note on the Truncated-Exponential Based Apostol-Type Polynomials
Previous Article in Special Issue
Some Results on the Symmetric Representation of the Generalized Drazin Inverse in a Banach Algebra
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Rank Equalities Related to the Generalized Inverses A‖(B1,C1), D‖(B2,C2) of Two Matrices A and D

1
Faculty of Mathematics and Physics, Huaiyin Institute of Technology, Huaian 223003, China
2
Instituto de Matemática Multidisciplinar, Universitat Politècnica de València, 46022 Valencia, Spain
*
Authors to whom correspondence should be addressed.
Symmetry 2019, 11(4), 539; https://doi.org/10.3390/sym11040539
Submission received: 27 March 2019 / Revised: 9 April 2019 / Accepted: 10 April 2019 / Published: 15 April 2019
(This article belongs to the Special Issue Matrices and Symmetry)

Abstract

:
Let A be an n × n complex matrix. The ( B , C ) -inverse A ( B , C ) of A was introduced by Drazin in 2012. For given matrices A and B, several rank equalities related to A ( B 1 , C 1 ) and B ( B 2 , C 2 ) of A and B are presented. As applications, several rank equalities related to the inverse along an element, the Moore-Penrose inverse, the Drazin inverse, the group inverse and the core inverse are obtained.
MSC:
15A09; 15A03

1. Introduction

The set of all m × n matrices over the complex field C will be denoted by C m × n . Let A * , R ( A ) , N ( A ) and rank ( A ) denote the conjugate transpose, column space, null space and rank of A C m × n , respectively.
For A C m × n , if X C n × m satisfies A X A = A , X A X = X , ( A X ) * = A X and ( X A ) * = X A , then X is called a Moore-Penrose inverse of A [1,2]. This matrix X always exists, is unique and will be denoted by A .
Let A C n × n . It can be easily proved that exists a non-negative integer k for which rank ( A k ) = rank ( A k + 1 ) holds. The Drazin index of A is the smallest non-negative k such that rank ( A k ) = rank ( A k + 1 ) , and denoted by ind ( A ) . A matrix X C n × n such that X A k + 1 = A k , X A X = X and A X = X A hold, where k = ind ( A ) , is called a Drazin inverse of A. It can be proved, (see, e.g., [3] Chapter 4), that the Drazin inverse of any square matrix A exists, is unique, and will be denoted by A D . If ind ( A ) 1 , then the Drazin inverse of A is called the group inverse and denoted by A # .
The core inverse of a complex matrix was introduced by Baksalary and Trenkler in [4]. Let A C n × n , a matrix X C n × n is called a core inverse of A, if it satisfies A X = P R ( A ) and R ( X ) R ( A ) , where P R ( A ) denotes the orthogonal projector onto R ( A ) . If such a matrix X exists, then it is unique and denoted by A # . In [4] it was proved that a square matrix A is core invertible if and only if ind ( A ) 1 .
In [5], Mary introduced a new type of generalized inverse, namely, the inverse along an element. Let A , D C n × n . We say that A is invertible along D if there exists Y C n × n such that
Y A D = D = D A Y , R ( Y ) R ( D ) and N ( D ) N ( Y ) .
If such Y exists, then it is unique and denoted by A D . The inverse along an element extends some known generalized inverses, for example, the group inverse, the Drazin inverse and the Moore-Penrose inverse.
In ([6] Definition 4.1), Benítez et al. gave the following definition extending simultaneously the notion of the ( B , C ) -inverse from elements in rings [7] to rectangular matrices and the invertibility along an element. Let A C m × n and B , C C n × m , the matrix A is said to be ( B , C ) -invertible, if there exists a matrix Y C n × m such that
Y A B = B , C A Y = C , R ( Y ) R ( B ) and N ( C ) N ( Y ) .
If such a matrix Y exists, then it is unique and denoted by A ( B , C ) . Many existence criteria and properties of the ( B , C ) -inverse can be found in, for example, [6,8,9,10,11,12,13,14]. From the definition of the ( B , C ) -inverse, it is evident that R ( Y ) = R ( B ) and N ( C ) = N ( Y ) . The ( B , C ) -inverse of A is a generalization of some well-known generalized inverses. By ([7] p. 1910), the Moore-Penrose inverse of A coincides with the ( A * , A * ) -inverse of A. A D is the ( D , D ) -inverse of A, A D is the ( A k , A k ) -inverse of A, where k = ind ( A ) and A # is the ( A , A ) -inverse of A. By ([15] Theorem 4.4), we have that the ( A , A * ) -inverse is the core inverse of A.
Let A C m × n be a matrix of rank r, let T be a subspace of C n of dimension s r and let S be a subspace of C m of dimension m s . The matrix A has a { 2 } -inverse X such that R ( X ) = T and N ( X ) = S if and only if A T S = C n (see, e.g., [3] Section 2.6). In this case, X is unique and is denoted by A T , S ( 2 ) . Many properties of A T , S ( 2 ) can be found in, for example, [3,16,17,18,19].
The theory of the generalized inverses has many applications, as one can see in [3]. Another important application is the study of singular systems of differential equations (see, e.g., [20,21]).
The main purpose of the manuscript is twofold: to research the rank of the difference A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) and to apply this study to characterize when A A ( B 1 , C 1 ) = D D ( B 2 , C 2 ) . These results are contained in Section 3 and Section 4. The paper finishes by particularizing the previous results to some standard generalized inverses.

2. Preliminaries

The following lemmas about the partitioned matrices [ A , B ] and A B will be useful in the sequel.
Lemma 1.
Let A C m × n and B C m × k .
(1)
([22] Theorem 5) For any A A { 1 } and B B { 1 } , we have
rank ( [ A , B ] ) = rank ( A ) + rank ( [ I m A A ] B ) = rank ( [ I m B B ] A ) + rank ( B ) ;
(2)
rank ( [ A , B ] ) = rank ( B ) if and only if R ( A ) R ( B ) .
Proof. 
Since (1) was proved in ([22] Theorem 5), we only give the proof of (2). Observe that rank ( [ A , B ] ) = dim [ R ( A ) + R ( B ) ] = dim R ( A ) + dim R ( B ) dim [ R ( A ) R ( B ) ] , which leads to rank ( [ A , B ] ) = rank ( B ) dim R ( A ) = dim [ R ( A ) R ( B ) ] R ( A ) R ( B ) . ☐
Lemma 2.
Let A C m × n and B C k × n .
(1)
([22] Theorem 5) For any A A { 1 } and B B { 1 } , we have
rank A B = rank ( A ) + rank ( B [ I n A A ] ) = rank ( A [ I n B B ] ) + rank ( B ) ;
(2)
rank A B = rank ( B ) if and only if N ( B ) N ( A ) .
Proof. 
Again, only the the proof of (2) will be given. Since rank A B = rank ( [ A * , B * ] ) , by employing item (2) of Lemma 1 we get
rank A B = rank ( B ) rank ( [ A * , B * ] ) = rank ( B * ) R ( A * ) R ( B * ) .
The proof finishes by recalling the equality R ( X * ) = N ( X ) valid for any matrix X, where the superscript denotes the orthogonal complement. ☐
Lemma 3.
([22], Theorem 5) Let A C m × n and B C m × k .
(1)
If there is a matrix Q C n × t such that R ( A ) R ( A Q ) , then
rank ( [ A Q , B ] ) = rank ( [ A , B ] ) ;
(2)
If there is a matrix P C k × s such that R ( B ) R ( B P ) , then
rank ( [ A , B P ] ) = rank ( [ A , B ] ) .
Proof. 
We only prove (1) since (2) is analogous. Observe that R ( A ) R ( A Q ) and R ( A ) = R ( A Q ) are equivalent. Now, (1) is evident from the expression rank ( [ X , Y ] ) = dim R ( X ) + dim R ( Y ) dim [ R ( X ) R ( Y ) ] valid for any pair of matrices X and Y with the same number of rows. ☐
Lemma 4.
([22], Theorem 5) Let A C m × n and B C k × n .
(1)
If there is a matrix Q C t × m such that N ( Q A ) N ( A ) , then
rank Q A B = rank A B ;
(2)
If there is a matrix P C s × k such that N ( P B ) N ( B ) , then
rank A P B = rank A B .
Proof. 
The proof is similar than the proof of Lemma 2, item (2). ☐
From Lemma 3 and Lemma 4, we have the following two lemmas.
Lemma 5.
Let A C m × n and B C m × k . If there are matrices Q C n × t and P C k × s such that R ( A ) R ( A Q ) and R ( B ) R ( B P ) , then
rank ( [ A Q , B P ] ) = rank ( [ A , B ] ) .
Lemma 6.
Let A C m × n and B C k × n . If there are matrices Q C t × m and P C s × k such that N ( Q A ) N ( A ) and N ( P B ) N ( B ) , then
rank Q A P B = rank A B .
Lemma 7.
([23] Theorem 2.1) Let P , Q C n × n be any two idempotent matrices. The difference P Q satisfies the following rank equality:
rank ( P Q ) = rank P Q + rank ( [ P , Q ] ) rank ( P ) rank ( Q ) .
The following lemma gives the calculation and the characterization of the existence of the ( B , C ) -inverse of a matrix A.
Lemma 8
([6] Theorem 4.4). Let A C m × n and B , C C n × m . The following statements are equivalent:
(1)
the ( B , C ) -inverse of A exists;
(2)
rank ( B ) = rank ( C ) = rank ( C A B ) .
In this case, A ( B , C ) = B ( C A B ) C .
Lemma 9
([6] Corollary 7.2). Let A C n × m and G C m × n . The following statements are equivalent:
(1)
the matrix A is invertible along G;
(2)
the outer inverse A R ( G ) , N ( G ) ( 2 ) exists.

3. Main Theorem

In this section, several rank equalities related to the generalized inverse A ( B 1 , C 1 ) and D ( B 2 , C 2 ) of A and D are derived, where A , D C n × m and B 1 , B 2 , C 1 , C 2 C m × n .
Theorem 1.
Let A , D C n × m , B 1 , B 2 , C 1 , C 2 C m × n . If A is ( B 1 , C 1 ) invertible and D is ( B 2 , C 2 ) invertible with A ( B 1 , C 1 ) is the ( B 1 , C 1 ) -inverse of A and D ( B 2 , C 2 ) is the ( B 2 , C 2 ) -inverse of D, then
rank ( A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) ) = rank C 1 C 2 + rank ( [ A B 1 , D B 2 ] ) rank ( B 1 ) rank ( B 2 ) = rank C 1 C 2 + rank ( [ A B 1 , D B 2 ] ) rank ( C 1 ) rank ( C 2 ) .
Proof. 
Since A ( B 1 , C 1 ) and D ( B 2 , C 2 ) are outer inverse of A and D we have have that A A ( B 1 , C 1 ) and D D ( B 2 , C 2 ) are idempotents. By Lemma 7, we have
rank ( A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) ) = rank A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) + rank ( [ A A ( B 1 , C 1 ) , D D ( B 2 , C 2 ) ] ) rank ( A A ( B 1 , C 1 ) ) rank ( D D ( B 2 , C 2 ) ) .
Since A ( B 1 , C 1 ) A B 1 = B 1 and R ( A ( B 1 , C 1 ) ) = R ( B 1 ) , thus
rank ( B 1 ) rank ( A ( B 1 , C 1 ) A ) rank ( A ( B 1 , C 1 ) ) = rank ( B 1 ) .
Similarly, the expressions C 1 A A ( B 1 , C 1 ) = C 1 and N ( A ( B 1 , C 1 ) ) = N ( C 1 ) imply
rank ( C 1 ) rank ( A A ( B 1 , C 1 ) ) rank ( A ( B 1 , C 1 ) ) = rank ( C 1 ) .
From (4) and (5), we have
rank ( A A ( B 1 , C 1 ) ) = rank ( A ( B 1 , C 1 ) A ) = rank ( B 1 ) = rank ( C 1 ) .
In an analogous manner, for the ( B 2 , C 2 ) -invertible matrix D, we have
rank ( D D ( B 2 , C 2 ) ) = rank ( D ( B 2 , C 2 ) D ) = rank ( B 2 ) = rank ( C 2 ) .
By Lemma 8, we have A ( B 1 , C 1 ) = B 1 ( C 1 A B 1 ) C 1 and D ( B 2 , C 2 ) = B 2 ( C 2 D B 2 ) C 2 . Thus
rank A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) = rank A B 1 ( C 1 A B 1 ) C 1 D B 2 ( C 2 D B 2 ) C 2
rank ( [ A A ( B 1 , C 1 ) , D D ( B 2 , C 2 ) ] ) = rank ( [ A B 1 ( C 1 A B 1 ) C 1 , D B 2 ( C 2 D B 2 ) C 2 ] ) .
By ([6] Theorem 3.4), we have R ( C 1 ) = R ( C 1 A B 1 ) and N ( B 1 ) = N ( C 1 A B 1 ) . Then we have
C 1 = C 1 A B 1 ( C 1 A B 1 ) C 1 and B 1 = B 1 ( C 1 A B 1 ) C 1 A B 1 .
In a analogous manner, for the ( B 2 , C 2 ) -invertible matrix D, we have
C 2 = C 2 D B 2 ( C 2 D B 2 ) C 2 and B 2 = B 2 ( C 2 D B 2 ) C 2 D B 2 .
Since the conditions in (10) and (11) imply that N ( A B 1 ( C 1 A B 1 ) C 1 ) N ( C 1 ) and N ( D B 2 ( C 2 D B 2 ) C 2 ) N ( C 2 ) , respectively. Thus, Lemma 6 and the condition in (8) imply
rank A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) = rank C 1 C 2 .
By (10) and (11), we have
A B 1 = A B 1 ( C 1 A B 1 ) C 1 A B 1 and D B 2 = D B 2 ( C 2 D B 2 ) C 2 D B 2 .
The expressions in (13) imply that R ( A B 1 ) R ( A B 1 ( C 1 A B 1 ) C 1 ) and R ( D B 2 ) R ( D B 2 ( C 2 D B 2 ) C 2 ) . Thus, Lemma 5 and the expression in (9) imply
rank ( [ A A ( B 1 , C 1 ) , D D ( B 2 , C 2 ) ] ) = rank ( [ A B 1 , D B 2 ] ) .
The proof is completed by using (3), (6), (7), (13) and (14). ☐
Example 1.
Let
A = 1 0 0 0 , B 1 = 1 1 1 1 , C 1 = 0 0 1 1
and
D = 0 0 0 1 , B 2 = 1 2 2 4 , C 2 = 1 1 1 1 .
By using Lemma 8 we get that A is ( B 1 , C 1 ) -invertible, D is ( B 2 , C 2 ) -invertible, and
A ( B 1 , C 1 ) = 1 1 1 1 , D ( B 2 , C 2 ) = 1 / 2 1 / 2 1 1 .
These latter computations can be easily performed by a numerical software, e.g., Octave. Now,
rank A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) = rank 1 1 1 1 = 1 .
Evidently, rank C 1 C 2 = 1 , rank ( B 1 ) = rank ( B 2 ) = 1 , and
rank [ A B 1 , D B 2 ] = rank 1 1 0 0 0 0 2 4 = 2 ,
which exemplifies Theorem 1.

4. Applications

In this section, as an application of the preceding section, we will characterize when A A ( B 1 , C 1 ) = D D ( B 2 , C 2 ) holds. Also, several rank equalities for the Moore-Penrose inverse, the Drazin inverse, the group inverse, the core inverse and the inverse along an element will be presented in view of the rank equalities for the ( B , C ) -inverse presented in Theorem 1.
Theorem 2.
Let A , D C n × m , B 1 , B 2 , C 1 , C 2 C m × n . If A is ( B 1 , C 1 ) invertible and D is ( B 2 , C 2 ) invertible, then the following statements are equivalent:
(1)
A A ( B 1 , C 1 ) = D D ( B 2 , C 2 ) ;
(2)
R ( A B 1 ) R ( D B 2 ) and N ( C 1 ) N ( C 2 ) ;
(3)
R ( D B 2 ) R ( A B 1 ) and N ( C 2 ) N ( C 1 ) ;
(4)
N ( A ( B 1 , C 1 ) ) N ( D D ( B 2 , C 2 ) ) and R ( A A ( B 1 , C 1 ) ) R ( D B 2 ) ;
(5)
N ( D ( B 2 , C 2 ) ) N ( A A ( B 2 , C 2 ) ) and R ( D D ( B 2 , C 2 ) ) R ( A B 1 ) .
Proof. 
By B 1 = A ( B 1 , C 1 ) A B 1 , we have rank ( A B 1 ) rank ( B 1 ) = rank ( A ( B 1 , C 1 ) A B 1 ) rank ( A B 1 ) , that is rank ( A B 1 ) = rank ( B 1 ) . Similarly, we have rank ( D B 2 ) = rank ( B 2 ) . By Lemma 2, we have
rank C 1 C 2 = rank ( C 1 ) + rank ( C 2 [ I n C 1 C 1 ] ) .
By Lemma 1, we have
rank ( [ A B 1 , D B 2 ] ) = rank ( D B 2 ) + rank ( [ I n D B 2 ( D B 2 ) ] A B 1 ) ,
where ( D B 2 ) is any inner inverse of D B 2 .
Assume that (1) holds. By Theorem 1 we have
rank C 1 C 2 + rank ( [ A B 1 , D B 2 ] ) rank ( B 1 ) rank ( B 2 ) = 0 .
By employing rank ( D B 2 ) = rank ( B 2 ) and Lemma 8, the equality in (17) can be written as
rank C 1 C 2 + rank ( [ A B 1 , D B 2 ] ) rank ( C 1 ) rank ( D B 2 ) = 0 .
Having in mind (18), (15), and (16), we have that rank ( C 2 [ I n C 1 C 1 ] ) = 0 and rank ( [ I n D B 2 ( D B 2 ) ] A B 1 ) = 0 . That is R ( A B 1 ) R ( D B 2 ) and N ( C 1 ) N ( C 2 ) , we have just obtained (2).
By using rank ( A B 1 ) = rank ( B 1 ) , Lemma 8 and (17) we get
rank C 1 C 2 + rank ( [ A B 1 , D B 2 ] ) rank ( A B 1 ) rank ( C 2 ) = 0 .
The proof of (1) ⇒ (3) finishes as the proof of (1) ⇒ (2).
(2) ⇒ (1). The hypotheses are clearly equivalent to A B 1 = D B 2 ( D B 2 ) A B 1 and C 2 = C 2 C 1 C 1 , which in view of (15), (16), and Theorem 1 lead to (1).
(3) ⇒ (1). The proof is similar than the proof of (2) ⇒ (1).
(1) ⇔ (4). From the proof of Theorem 1 and rank ( D B 2 ) = rank ( B 2 ) , we have
rank ( A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) ) = rank A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) + rank ( [ A A ( B 1 , C 1 ) , D B 2 ] ) rank ( A A ( B 1 , C 1 ) ) rank ( D B 2 ) .
By A ( B 1 , C 1 ) A A ( B 1 , C 1 ) = A ( B 1 , C 1 ) , we have N ( A ( B 1 , C 1 ) A ) N ( A ( B 1 , C 1 ) ) . Thus, by Lemma 2, the expression (19) can be written as
rank ( A A ( B 1 , C 1 ) D D ( B 2 , C 2 ) ) = rank A ( B 1 , C 1 ) D D ( B 2 , C 2 ) + rank ( [ A A ( B 1 , C 1 ) , D B 2 ] ) rank ( A A ( B 1 , C 1 ) ) rank ( D B 2 ) .
By Lemma 1 and Lemma 2, we have
rank A ( B 1 , C 1 ) D D ( B 2 , C 2 ) = rank ( A ( B 1 , C 1 ) ) + rank ( D D ( B 2 , C 2 ) [ I n ( A ( B 1 , C 1 ) ) A ( B 1 , C 1 ) ] )
and
rank ( [ A A ( B 1 , C 1 ) , D B 2 ] ) = rank ( D B 2 ) + rank ( [ I n D B 2 ( D B 2 ) ] A A ( B 1 , C 1 ) ) .
By (6) and R ( B 1 ) = R ( A ( B 1 , C 1 ) ) , we have
rank ( A A ( B 1 , C 1 ) ) = rank ( A ( B 1 , C 1 ) ) .
Thus by (20) and (21), we have that A A ( B 1 , C 1 ) = D D ( B 2 , C 2 ) if and only if both
rank ( D D ( B 2 , C 2 ) [ I n ( A ( B 1 , C 1 ) ) A ( B 1 , C 1 ) ] ) = 0
and
rank ( [ I n D B 2 ( D B 2 ) ] A A ( B 1 , C 1 ) ) = 0
hold. It is easy to see that rank ( D D ( B 2 , C 2 ) [ I n ( A ( B 1 , C 1 ) ) A ( B 1 , C 1 ) ] ) = 0 is equivalent to N ( A ( B 1 , C 1 ) ) N ( D D ( B 2 , C 2 ) ) and rank ( [ I n D B 2 ( D B 2 ) ] A A ( B 1 , C 1 ) ) = 0 is equivalent to R ( A A ( B 1 , C 1 ) ) R ( D B 2 ) .
The proof of ( 1 ) ( 5 ) is similar to the proof of ( 1 ) ( 4 ) . ☐
Since the inverse of a matrix A along D coincides with the ( D , D ) -inverse of A, Theorem 1 and Theorem 2 lead to the following corollaries.
Corollary 1.
Let A , B , D 1 , D 2 C n × n . If A is invertible along D 1 and B is invertible along D 2 , then
rank ( A A D 1 B B D 2 ) = rank D 1 D 2 + rank ( [ A D 1 , B D 2 ] ) rank ( D 1 ) rank ( D 2 ) = rank D 1 D 2 + rank ( [ A D 1 , B D 2 ] ) rank ( D 1 ) rank ( D 2 ) .
Corollary 2.
Let A , B , D 1 , D 2 C n × n . If A is invertible along D 1 and B is invertible along D 2 , then the following statements are equivalent:
(1)
A A D 1 = B B D 2 ;
(2)
R ( A D 1 ) R ( B D 2 ) and N ( D 1 ) N ( D 2 ) ;
(3)
R ( B D 2 ) R ( A D 1 ) and N ( D 2 ) N ( D 1 ) ;
(4)
N ( A D 1 ) N ( B B D 2 ) and R ( A A D 1 ) R ( B D 2 ) ;
(5)
N ( B D 2 ) N ( A A D 1 ) and R ( B B D 2 ) R ( A D 1 ) .
Let A C n × n . By ([7] p. 1910), we have that the Moore-Penrose inverse of A coincides with the ( A * , A * ) -inverse of A, the Drazin inverse of A coincides with the ( A k , A k ) -inverse of A for some integer k and A is group invertible if and only if A is ( A , A ) -invertible. By ([15] Theorem 4.4), we have that the ( A , A * ) -inverse coincides with the core inverse of A. We have that the ( A * , A ) -inverse coincides with the dual core inverse of A. Thus, by Theorem 1 and Theorem 2, more results of the inverse along an element, the Moore-Penrose inverse, Drazin inverse, core inverse and dual core inverse can be obtained. We give some characterizations of these results as follows, and leaving the remaining parts to the reader to research. Also, some rank characterizations of the EP elements can be got by the following Corollary 3.
Corollary 3.
Let A , B C n × n . Then
(1)
Let A and B be the Moore-Penrose inverse of A and B, respectively. We have
rank ( A A B B ) = rank A * B * + rank ( [ A , B ] ) rank ( A ) rank ( B ) ;
(2)
Let ind ( A ) = k , ind ( B ) = l and A D , B D be the Drazin inverse of A and B, respectively. We have
rank ( A A D B B D ) = rank A k B l + rank ( [ A k + 1 , B l + 1 ] ) rank ( A k ) rank ( B l ) ;
(3)
Let ind ( A ) = ind ( B ) = 1 and A # , B # be the group inverse of A and B, respectively. We have
rank ( A A # B B # ) = rank A B + rank ( [ A 2 , B 2 ] ) rank ( A ) rank ( B ) ;
(4)
Let ind ( A ) = ind ( B ) = 1 and A # , B # be the core inverse of A and B, respectively. We have
rank ( A A # B B # ) = rank A * B * + rank ( [ A 2 , B 2 ] ) rank ( A ) rank ( B ) .
Proof. 
(1). Since A coincides with the ( A * , A * ) -inverse of A and B coincides with the ( B * , B * ) -inverse of B, by Theorem 1, we have
rank ( A A B B ) = rank A * B * + rank ( [ A A * , B B * ] ) rank ( A * ) rank ( B * ) .
That is
rank ( A A B B ) = rank A * B * + rank ( [ A , B ] ) rank ( A ) rank ( B ) ,
by the following obvious facts rank ( [ A A * , B B * ] ) = rank ( [ A , B ] ) , rank ( A * ) = rank ( A ) , rank ( B * ) = rank ( B ) .
( 2 ) , ( 3 ) , ( 4 ) are obvious. ☐
Some particular cases can be obtained from the previous corollary by setting the matrix B to some concrete generalized inverses. For example, when B = A we have rank ( A A A A ) = 2 [ rank ( A , A ) rank ( A ) ] , which can be used to get a characterization of the EP matrices ( A A = A A ) or the co-EP matrices ( A A A A is nonsingular, see [24]).

Author Contributions

Writing-original draft preparation, W.W.; writing-review and editing, W.W. and S.X.; methodology, S.X. and J.B.; supervision, S.X. and J.B.

Funding

This research received no external funding.

Acknowledgments

The second author is grateful to China Scholarship Council for giving him a scholarship for his further study in Universitat Politècnica de València, Spain.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Moore, E.H. On the reciprocal of the general algebraic matrix. Bull. Am. Math. Soc. 1920, 26, 394–395. [Google Scholar]
  2. Penrose, R. A generalized inverse for matrices. Math. Proc. Camb. Philos. Soc. 1955, 51, 406–413. [Google Scholar] [CrossRef]
  3. Ben-Israel, A.; Greville, T.N.E. Generalized Inverse: Theory and Applications, 2nd ed.; Springer: New York, NY, USA, 2003. [Google Scholar]
  4. Baksalary, O.M.; Trenkler, G. Core inverse of matrices. Linear Multilinear Algebra 2010, 58, 681–697. [Google Scholar] [CrossRef]
  5. Mary, X. On generalized inverse and Green’s relations. Linear Algebra Appl. 2011, 434, 1836–1844. [Google Scholar] [CrossRef]
  6. Benítez, J.; Boasso, E.; Jin, H.W. On one-sided (B,C)-inverses of arbitrary matrices. Electron. J. Linear Algebra 2017, 32, 391–422. [Google Scholar] [CrossRef]
  7. Drazin, M.P. A class of outer generalized inverses. Linear Algebra Appl. 2012, 436, 1909–1923. [Google Scholar] [CrossRef] [Green Version]
  8. Boasso, E.; Kantún-Montiel, G. The (b,c)-inverses in rings and in the Banach context. Mediterr. J. Math. 2017, 14, 112. [Google Scholar] [CrossRef]
  9. Drazin, M.P. Left and right generalized inverses. Linear Algebra Appl. 2016, 510, 64–78. [Google Scholar] [CrossRef]
  10. Ke, Y.Y.; Cvetković-Ilić, D.S.; Chen, J.L.; Višnjić, J. New results on (b,c)-inverses. Linear Multilinear Algebra 2018, 66, 447–458. [Google Scholar] [CrossRef]
  11. Ke, Y.Y.; Višnjić, J.; Chen, J.L. One sided (b,c)-inverse in rings. arXiv, 2016; arXiv:1607.06230. [Google Scholar]
  12. Rakić, D.S. A note on Rao and Mitra’s constrained inverse and Drazin’s (b,c) inverse. Linear Algebra Appl. 2017, 523, 102–108. [Google Scholar] [CrossRef]
  13. Wang, L.; Castro-González, N.; Chen, J.L. Characterizations of the outer generalized inverses. Can. Math. Bull. 2017, 60, 861–871. [Google Scholar] [CrossRef]
  14. Xu, S.Z.; Benítez, J. Existence criteria and expressions of the (b,c)-inverse in rings and its applications. Mediterr. J. Math. 2018, 15, 14. [Google Scholar] [CrossRef]
  15. Rakić, D.S.; Dinčić, N.Č.; Djordjević, D.S. Group, Moore-Penrose, core and dual core inverse in rings with involution. Linear Algebra Appl. 2014, 463, 115–133. [Google Scholar] [CrossRef]
  16. Liu, Y.H. How to characterize equalities for the generalized inverse A T , S ( 2 ) of a matrix. Kyungpook Math. J. 2003, 43, 605–616. [Google Scholar]
  17. Liu, Y.H.; Wei, M.S. Rank equalities for submatrices in generalized inverses M T , S ( 2 ) of M. Appl. Math. Comput. 2004, 152, 499–504. [Google Scholar] [CrossRef]
  18. Liu, Y.H.; Wei, M.S. Rank equalities related to the generalized inverses A T , S ( 2 ) , B T 1 , S 1 ( 2 ) of two matrices A and B. Appl. Math. Comput. 2004, 159, 19–28. [Google Scholar] [CrossRef]
  19. Wei, Y.M. A characterization and representation of the generalized inverse A T , S ( 2 ) and its applications. Linear Algebra Appl. 1998, 280, 87–96. [Google Scholar] [CrossRef]
  20. Boyarintsev, Y. Methods of Solving Singular Systems of Ordinary Differential Equations, Pure and Applied Mathematics; A Wiley-Interscience Publication; John Wiley and Sons, Ltd.: Chichester, UK, 1992. [Google Scholar]
  21. Campbell, S. Singular Systems of Differential Equations; Pitman Advanced Pub. Program: London, UK, 1980. [Google Scholar]
  22. Marsaglia, G.; Styan, G.P.H. Equalities and inequalities for ranks of matrices. Linear Multilinear Algebra 1974, 2, 269–292. [Google Scholar] [CrossRef]
  23. Tian, Y.G.; Styan, G.P.H. Rank equalities for idempotent and involutory matrices. Linear Algebra Appl. 2001, 335, 101–117. [Google Scholar] [CrossRef]
  24. Benítez, J.; Rakočević, V. Matrices A such that AAAA are nonsingular. Appl. Math. Comput. 2010, 217, 3493–3503. [Google Scholar] [CrossRef]

Share and Cite

MDPI and ACS Style

Wang, W.; Xu, S.; Benítez, J. Rank Equalities Related to the Generalized Inverses A‖(B1,C1), D‖(B2,C2) of Two Matrices A and D. Symmetry 2019, 11, 539. https://doi.org/10.3390/sym11040539

AMA Style

Wang W, Xu S, Benítez J. Rank Equalities Related to the Generalized Inverses A‖(B1,C1), D‖(B2,C2) of Two Matrices A and D. Symmetry. 2019; 11(4):539. https://doi.org/10.3390/sym11040539

Chicago/Turabian Style

Wang, Wenjie, Sanzhang Xu, and Julio Benítez. 2019. "Rank Equalities Related to the Generalized Inverses A‖(B1,C1), D‖(B2,C2) of Two Matrices A and D" Symmetry 11, no. 4: 539. https://doi.org/10.3390/sym11040539

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