J J. Edu. & Sci., Vol. (21), No.

(4) 2008 K

Some basic properties of idempotent matrices

Alaa A. Hammodat Ali A. Bilal Akram S. Mohammed

Dept. of Math. Dept. of Math. Dept. of Math.
College of Education College of Education College of Science
University of Mosul University of Mosul University of Tikrit

Received Accepted
25 / 12 / 2006 03 / 10 / 2007

‫ﺍﻟﺨﻼﺼﺔ‬

‫ﻓﻲ ﻫﺫﺍ ﺍﻟﺒﺤﺙ ﻗﺩﻤﻨﺎ ﺒﻌﺽ ﺍﻟﺨﻭﺍﺹ ﻟﻠﻤﺼﻔﻭﻓﺎﺕ ﺍﻟﻤﺘﺤﺎﻴﺩﺓ ﺍﻟﻤﺘﺒﺎﺩﻟﺔ ﺼﻔﺭﻴﹰﺎ ﻭﺃﻋﻁﻴﻨﺎ‬
.‫ﺒﻌﺽ ﺍﻟﺨﻭﺍﺹ ﻟﻠﻤﺼﻔﻭﻓﺎﺕ ﻏﻴﺭ ﺍﻟﻤﻨﻔﺭﺩﺓ‬

Abstract
In this paper we give some properties of the zero commut
idempotent matrices ,and give some properties of non-singular matrices .

124
Some basic properties of idempotent matrices.

1. Introduction
Throughout, this paper all matrices considered are square and
commutative unless other wise stated.
In [2], Koliha, Racocevic and Straskraba present new results on the
invertibility of the sum of projectors, new relations between the non-
singularity of the difference and the sum of projectors, and present a
simple proof of the invertibility of n×n matrix A exists by showing that
N(A)={0}. In this paper we present some basic properties of an
idempotent matrices and relations between the range and the null spaces
and give some results on the non-singularity of the difference and sum of
idempotent matrices .We recall that;
1) A matrix A is said to be idempotent if A 2=A .
2) A two matrices A and B are said to be zero commut if AB = BA =0 .
3) A null space of a matix A is the set of all solutions to the equation
v v v
A x =0 , Null(A)={ x ∈ R n : A x =0} , and we denote that N(A) .
4) A range space of a matrix A is the set of all solutions to the equation
v v
A x = x , and we denote that R(A) .

2. Idempotent matrices
In this section we give some basic properties of the idempotent
matrices.

Lemma 2.1 :
If A is idempotent matrix, then (I-A) is idempotent .
Proof:
Trivial.

Proposition 2.2 [2] :

If A is a matrix, then N(A)=R(I- A),
also N(I-A)=R(A) .

Proposition 2.3 :
If A is a matrix, then R(A) ∩ N(A) ={0} .
Proof:
v v v
Let x ∈ R(A) ∩ N(A) , then x ∈ R(A) and x ∈ N(A) .
v v v v
Then x =A x and A x =0, so, x =0 .
Hence R(A) ∩ N(A)={0} . ■

Proposition 2.4 :
If A and B be are idempotent matrices, then R(A) ∩ R[B(I-A)]={0}.
Proof:
v v v v v
Let y ∈ R(A) ∩ R[B(I-A)] , then y =A y and y =B(I-A) y .

125
Alaa A. Hammodat & Ali A. Bilal & Akram S. Mohammed

v v v v
So A y =AB(I-A) y =AB y -ABA y =0 .
v v v
But A y = y ,so y =0 .
Hence R(A) ∩ R[B(I-A)]={0} . ■

Proposition 2.5 :
If A and B be are matrices, with AB=A and BA=B . Then N(A)=N(B) .
Proof:
v v v v
Let x ∈ N(A), then A x =0, so BA x =0, but BA=B. Therefore B x =0
v
and x ∈ N(B) . Hence N(A) ⊆ N(B) .
Similarly N(B) ⊆ N(A) .
Therefore N(A)=N(B) . ■

Proposition 2.6 :
If A and B be are idempotent matrices, then R(AB) = R(A) ∩ R(B) .
Proof:
v v v
Let x ∈ R(AB), then x =AB x .
v v
So (I-A) x = (I-A)AB x =0 .
v
Hence x ∈ N(I-A)=R(A) [by Proposition2.2] .
v v
Similarly x ∈ R(B) ,so x ∈ R(A) ∩ R(B) .
Therefore R(AB) ⊆ R(A) ∩ R(B) .
v v v v v
Now, let y ∈ R(A) ∩ R(B) ,then y =A y and y =B y .
v v v v
So A y =B y . That is A y =AB y ,
v v v v
but y =A y ,therefore y =AB y ∈ R(AB) .
So R(A) ∩ R(B) ⊆ R(AB) .
Hence R(AB) = R(A) ∩ R(B) . ■

Proposition 2.7 :
If A and B be are idempotent matrices, and R(A) ∩ R(B)={0}, then
N(A-B)=N(A) ∩ N(B) .
Proof:
v v v v
Let x ∈ N(A-B), then (A-B) x =0 and A x =B x .
v v v v
So A x =A 2 x =AB x =BA x .
v v v v
Now, from A x =BA x we get (B-I)A x =0 , so A x ∈ N(B-I)=R(B)
v v v
[by Proposition2.2], but A x =B x ,so B x ∈ R(B) .
v v v v
Similarly from B x =AB x we get A x ,B x ∈ R(A) .
v v v
So A x ,B x ∈ R(A) ∩ R(B)={0}, therefore x ∈ N(A) .
v v
Also , x ∈ N(B) , so x ∈ N(A) ∩ N(B) .
Hence N(A-B) ⊆ N(A) ∩ N(B) .
v v v
Now, let y ∈ N(A) ∩ N(B) , then A y =0 and B y =0 .
v v
So (A-B) y =0, therefore y ∈ N(A-B) .
So N(A) ∩ N(B) ⊆ N(A-B) .
Hence N(A-B)=N(A) ∩ N(B) . ■

126
Some basic properties of idempotent matrices.

Proposition 2.8 :
If A and B be are idempotent matrices, then R[B(I-A)]=N(A) ∩ R(B).
Proof:
v v v
Let x ∈ R[B(I-A)] , then x =B(I-A) x .
v v v v
So A x =AB(I-A) x =AB x -ABA x =0 .
v
So x ∈ N(A) .
v v v v v
Also, (I-B) x =(I-B)B(I-A) x =0 , so x =B x , therefore x ∈ R(B) .
v
So x ∈ N(A) ∩ R(B) .
Hence R[B(I-A)] ⊆ N(A) ∩ R(B) .
v v v v
Now, let y ∈ N(A) ∩ R(B) then A y =0 and y =B y .
v v
So A y =AB y =0 .
v v v v
Now, B(I-A) y = B(I-A)B y = B y -BAB y .
v v v v
Therefore B(I-A) y = B y = y , so y ∈ R[B(I-A)] .
Hence N(A) ∩ R(B) ⊆ R[B(I-A)] .
Therefore R[B(I-A)]=N(A) ∩ R(B) . ■

Proposition 2.9 :
If A and B be are idempotent matrices, then R(A)=R(B) if and only if
N(A)=N(B) .
Proof:
v v v v
Let R(A)=R(B), then A x =B x and let x ∈ N(A), then A x =0, so
v
B x =0.
v
Then x ∈ N(B) , therefore N(A) ⊆ N(B) .
Similarly N(B) ⊆ N(A) , so N(A)=N(B) .
v v v
Now, let N(A)=N(B) and let y ∈ R(A) , then y =A y .
v v v
So B y =BA y =AB y .
v v v
From B(I-B) y =0 , we get (I-B) y ∈ N(B)=N(A) and A(I-B) y =0 .
v v v v v v
So A y =AB y , but y =A y and B y =AB y .
v v v
Therefore y =B y and y ∈ R(B) , so R(A) ⊆ R(B) .
Similarly R(B) ⊆ R(A) .
Hence R(A)=R(B) . ■

Proposition 2.10 :
If A and B be are idempotent matrices, then N(A) ∩ N(B)=N(A+B-AB).
Proof:
v v v v
Let x ∈ N(A) ∩ N(B) , then A x =0 and B x =0 , so AB x =0 .
v v
Therefore (A+B-AB) x =0 and x ∈ N(A+B-AB) .
So N(A) ∩ N(B) ⊆ N(A+B-AB) .
v v v v v
Now, let y ∈ N(A+B-AB) , then (A+B-AB) y =0 , so A y +B y =AB y .
v v v v v
Hence A 2 y +AB y =A 2 B y . So A y =0 and y ∈ N(A) .

127
Alaa A. Hammodat & Ali A. Bilal & Akram S. Mohammed

v v
Similarly B y =0 and y ∈ N(B) .
v
So y ∈ N(A) ∩ N(B) .
Hence N(A+B-AB) ⊆ N(A) ∩ N(B) .
Therefore N(A) ∩ N(B)=N(A+B-AB) . ■

3. Basic properties of zero commut idempotent matrices.

In this section we present some basic properties of the zero
commut idempotent matrices.

Proposition 3.1 :
If A and B be are zero commut idempotent matrices, then
N(A-B) =N(A) ∩ N(B) .
Proof:
v v v v
Let x ∈ N(A-B) , then (A-B) x =0 and A x =B x .
v v
So A x =AB x =0 .
v
Hence x ∈ N(A) .
v v
Similarly x ∈ N(B), so x ∈ N(A) ∩ N(B) .
Hence N(A-B) ⊆ N(A) ∩ N(B) .
v v v v
Now, let y ∈ N(A) ∩ N(B) , then A y =0 and B y =0 , so (A-B) y =0 .
v
Therefore y ∈ N(A-B) , so N(A) ∩ N(B) ⊆ N(A-B) .
Hence N(A-B) =N(A) ∩ N(B) . ■

Proposition 3.2 :
If A and B be are zero commut idempotent matrices, then
N(A+B) =N(A) ∩ N(B) .
Proof:
Similar to the proof of proposition 3.1

Corollary 3.3 :
If A and B be are zero commut idempotent matrices, then
N(A+B)=N(A-B) .

Proposition 3.4 :
If A and B be are zero commut idempotent matrices, then
R(A+B) = R(A)+R(B) .
Proof:
v v v
Let y = y 1 + y 2 ∈ R(A) +R(B),
v v
where y 1 ∈ R(A) and y 2 ∈ R(B) .
v v v v v v
So (A+B)( y 1 + y 2 )=A y 1 +A y 2 +B y 1 +B y 2
v v v v
= y 1 + y 2 +A y 2 +B y 1 -0
v v v v v v
= y 1 + y 2 +A y 2 +B y 1 -AB( y 1 + y 2 )

128
Some basic properties of idempotent matrices.

v v v v v v
= y 1 + y 2 +A( y 2 -B y 2 )+B( y 1 -A y 1 )
v v v
=y 1+y 2 =y .
v v v
Therefore y = y 1 + y 2 ∈ R(A+B) , so R(A)+R(B) ⊆ R(A+B) .
v v v v v
Now, let x ∈ R(A+B) , then x =(A+B) x =A x +B x .
v v v v v v
Let x 1 =A x , then A x 1 =A x = x 1 , so x 1 ∈ R(A) .
v v v v v v
Also, let x 2 =B x , then B x 2 =B x = x 2 , so x 2 ∈ R(B) .
v v v
Therefore x = x 1 + x 2 ∈ R(A)+R(B) , so R(A+B) ⊆ R(A)+R(B) .
Hence R(A+B) = R(A)+R(B) . ■
Corollary 3.5 :
If A is an idempotent matrix, then R n =R(A)+N(A) .
Proof:
Since A(I-A)=(I-A)A=0 .
Then A and I-A are zero commut idempotent matrices .
Now, from [Proposition3.4] we get R(A+I-A)=R(A)+R(I-A) .
Hence R(I)=R(A)+N(A) .
Therefore R n = R(A)+N(A) . ■
Proposition 3.6 :
If A is an idempotent matrix, then R n =R(A) ⊕ N(A) .
Proof:
By [Corollary3.5] we get R n =R(A)+N(A), and by
[Proposition2.3] we get R(A) ∩ N(A) ={0} .
Hence R n =R(A) ⊕ N(A) . ■

4. Basic properties of non-singular matrices.

In this section we give some basic properties of a non-singular
matrices.
Remark:
The only nonsingular idempotent matrix is identity matrix (I n ).
Every idempotent matrix (except I n ) is singular but a singular matrix may
not be idempotent.
Theorem 4.1 [1]:
An n×n matrix A over a number field F has rank n if and only if
−1
A exists, that is, if and only if A is non-singular.
Theorem 4.2 [2]:
If A and B be are idempotent matrices, then the following
conditions are equivalent:
1) A-B is non-singular.
2) A+B and I-AB are non-singular.

129
Alaa A. Hammodat & Ali A. Bilal & Akram S. Mohammed

Proposition4.3:
Let A and B be are idempotent matrices and AB is non-singular,
then N(A) ∩ R(B)={0} .
Proof:
v v v v
Let x ∈ N(A) ∩ R(B), then A x =0 and x =B x .
v v v
So A x =AB x . Hence AB x =0 .
v v
Therefore x ∈ N(AB)={0} (since AB is non-singular), so x =0 .
Hence N(A) ∩ R(B)={0} . ■

Theorem 4.4 :
If A and B be are zero commut idempotent matrices, and A-B is
non-singular, then R n = R(A) ⊕ R(B) .
Proof:
v v v
Let x ∈ R(A) ∩ R(B) , then x ∈ R(A) and x ∈ R(B) .
v v v v v v v
Therefore x =A x and x =B x , so A x =B x , then (A-B) x =0 .
v
Hence x ∈ N(A-B)=0 (since A-B is non-singular) .
v
So x =0 ,hence R(A) ∩ R(B)={0} .
v v
Since (A-B)(I-A-B) x =0 , then (I-A-B) x ∈ N(A-B)=0 ,
v v v v
and (I-A-B) x =0 .Hence x =A x +B x ∈ R(A)+R(B) by [Proposition3.4].
So R n =R(A)+R(B) .
Hence R n = R(A) ⊕ R(B) . ■

Proposition 4.5 :
If A and B be are zero commut idempotent matrices, then the
following conditions are equivalent :
1) A-B is non-singular .
2) N(A) ∩ N(B)={0} .
Proof:
v v v v
(1→2) let x ∈ N(A) ∩ N(B), then A x =0 and B x =0 . So (A-B) x =0,
v v
therefore x ∈ N(A-B)=0 (since A-B is non-singular) . So x =0 .
Hence N(A) ∩ N(B)={0} .
(2→1) from [Proposition3.1] we get N(A) ∩ N(B)=N(A-B) ,
but N(A) ∩ N(B)={0} , then N(A-B)={0} .
Hence A-B is non-singular . ■

References:-
1) F. E. Hohn, "Introduction to linear algebra", The Macmillan
Company, New York, (1972).
2) J. J. Koliha, V. Rakocevic and I. Straskraba, "The difference and
sum of projectors", Linear Algebra Appl. 388(2004) ,279-288 .
3) J. J. Koliha, "Power bounded and exponentially bounded matrices",
Applications of Math., 44(1999), 289-308 .

130