:

CHAPTER ONE

MATRIX ALGEBRA

1.1 Introduction

Definition: A matrix is a rectangular array (or arrangement) of numbers or variables

enclosed in parentheses or brackets. The number (or variable) in a matrix are called entry
or element of the matrix. A capital letter is generally used to denote a matrix and lower
case letters with double subscripts generally denote its elements.

The number of rows (horizontal arrays) and the number of columns (vertical arrays)
determine the dimension of the matrix, which is also known as the order or size of the
matrix. For instance, a matrix having 3 rows and 4 columns is said to be a matrix of
dimension order 3x4 (read as “3 by 4). Notice that in expressing the dimension of a
matrix, the number of rows always written first and the number of columns afterwards.

Example 1.1 Consider the matrix

1 3 4  1st Row
A   
0
 5  2  2nd Row

1st column 2nd column 3rd column

Since A has two rows and three columns, A is said to be a 2x3 matrix or A is a matrix of
dimension (order) 2x3.

Note that every element in a matrix has definite location (position) allotted to it. For
instance, in a matrix A given above, -3 is located in the first row and the second column
and it can be denoted as a12 =-3. Similarly, a11 =1, a13 =4, a12 =0, a22 =5 and a23 =-2.
In general, a matrix A of dimension or order mxn (a matrix with m rows and n columns)
can be written as

1
 :

 a 11 a 12 a 13 . . . a 1n 
 
 a 21 a 22 a 23 . . . a 2n 
. . . . 
A 
. . . . 
. . . . 
 
a 
 m1 a m2 a m3 a mn 
Notice that each element of the above matrix has a subscript that indicates the number of
the row and column respectively which fully describes the position of the element. For
example, a43 represents the element found at the fourth row and the third column of the
matrix.

Remarks
1. The above general matrix A can be also denoted as a ij mn where i represent the

number of row and j represent the number of column to which the element belongs.
2. A matrix, by definition, is simply an arrangement of numbers and has no numerical
value
Exercise - 1
 2 5 0 7
 
Given A   3 1  3 1
 4 1 5 
 2

a) What is the dimension or size of A?

b) Find a14, a23, a33 and a34?
c) Write the second row and the third column?
Types of Matrices
1. Raw matrix; a matrix which consists of only one row. An n-component row matrix
such as (a1 a2 a3 . . . an) is said to be a matrix of order 1xn. Thus (2 0 -5) is a 1x3
row matrix.

2
 :

2. Column matrix: is a matrix consists of only one column. An n-component column

 a1 
 
a2   3
   
 5
matrix like   is said to be a matrix of order nx1. Thus
.
 is a column matrix
.   1
   
 0 
.  
a 
 n
of order 4  1 .
Definition A matrix that consists only one row (column) is known as a row (column)
vector. Vectors are special cases of matrices and obey all the rules that characterize
the operations of matrices.

3. Null or Zero Matrixes: is a matrix in which all of its entries (elements) are zero.
 0 0
 
For example, the matrix A =  0 0  is a 3x2 null matrix
 0 0
 
 0  0 0
Moreover, (0),  , 0 0,   are examples of null or zero matrices
 0  0 0 
4. Square matrix: It is a matrix which has equal number of rows and columns.
A matrix having an order of n x n (n rows and n columns) is called a square matrix of
order n.
2 1
For example, the matrix, A=   is a matrix of order (2) or 2x2.
3 0
 3 0 1
 
Similarly, (5) and  2 1 4  are square matrices of order 1 and 3 respectively.
  1 3 5
 
0 0
Notice that the matrix A=   can be termed as a square null matrix since it is
0 0 
both square and null matrix.

Definition The diagonal (or main diagonal) of a square matrix consists of elements
of the form aii or consist all elements found in the same number of rows and
columns.

3
 :

For example, in the matrix

3 4 0 
 
A  1 5  2  , the diagonal elements are a11 =3, a22 =5 and a33 = 6
3  3 6 
 
5. Diagonal – Matrix: is a square matrix which has zero entries (or elements) every
where other than the main diagonal.
 2 0 0
 2 0  
For example,   and  0  5 0  are diagonal matrices.
 0 3 0 0 4 
 
Note that the diagonal elements in a diagonal matrix may also be zero. For instance,
3 0 0 0
  is a diagonal matrix and every square null matrix like   can be a
0 0 0 0 

diagonal matrix.
6. Scalar Matrix: is a diagonal matrix whose diagonal elements are equal.
2 0 0
3 0  
For example,   and  0 2 0  are scalar matrices.
 0 3 0 2 
 0

7. Identify or Unit Matrix: it is a diagonal matrix, particularly a scalar matrix, whose

every diagonal element is one. An identity matrix of order n is denoted by In.
1 0 0
1 0   
For example I 2    and I 3   0 1 0  are identity matrices.
 0 1 0 0 1

8. Triangular Matrix
Square matrix whose every element above the diagonal is zero is said to be a lower -
triangular matrix
 5 0 0
 2 0  
For example,   and   2 6 0  are lower - triangular matrices.
 3 1   0 4 1
 
A square matrix whose every element below the diagonal is zero is called an
upper-triangular matrix

4
 :

3 1 5 
5 1   
For example,   and  0  4  2  are upper - triangular matrices.
 0  2 0 0 4 

Note that any or all of the diagonal elements in both types of triangular matrices can
be zero.
 4 0 5
3 0   
For example,   and  0 0 0  are lower and upper-triangular matrices
 4 0  0 0 0
 
respectively, and every square-null matrix and identity matrix can be both lower and
upper triangular matrices.

Example 1.2
a) Which of the following matrices are diagonal matrices?
1 0 0 
1 2  5 0  6 0 0  
A    , B   , C   , D   0 1 0 
 0 0  0 0 0 3 0  0 0 1
 
b) Which of the following matrices are lower or upper - triangular matrices?
 0 0 0
 3 1    0 0 5 0 0
A    , B   0 0 0 , C    , D   
 0 0 4 0 0   0 0  2 3 0
 
Answers
a) A is not a diagonal matrix because a12 =2 is an off-diagonal element and non-zero.
B is a diagonal matrix because it is a square matrix and every of its non-diagonal
elements are zero.
C is not a diagonal matrix because it is not a square matrix.
D is a diagonal because of the same reason as B.
b)
A is an upper - triangular matrix
B is a lower - triangular matrix
C can be both lower and upper-triangular matrix
D can not be both types because it is not a square matrix.

5
 :

Exercise - 2
1. Say "True or false" for each of the following
a) Every null or zero matrixes is a diagonal.
b) Every identity matrix is a scalar matrix.
c) Every matrix has a diagonal.
d) Every diagonal matrix is a scalar matrix

1.2 Basic Matrix Operations

The rules for matrix operations help you to know how matrices are combined. Not all
operations are well defined for all matrices, in particular, the dimensions (or order) of
matrices will need to satisfy some basic conditions before they can be manipulated.

A. Equality of Matrices
Two matrices A and B are said to be equal and written as A=B if
i. They have the same order (or dimension), and
ii. Each element of one matrix is the same as the corresponding element of the other.
Example 1.3 Determine whether the following pair of matrices are equal or not.
 5 1
  5  3 4
a) A    3 2 B   
 4 0  1 2 0 

3 5 3 5
b) A    B   
1  4  1 4 

2  3 4 2  3 4
c) A=   B   
1 0 5 1 0 5 

Answers
a) A ≠ B because they have different orders (A3x2 and B2x3)
b) A ≠ B because the element a22 in A is -4 and the corresponding element in B, b22 =4.
c) A = B because both are of the same order and all the corresponding elements are
equal.

6
 :

Exercise - 3
1. State the two conditions that should be satisfied by two matrices in order to equal.
2. Given an example of two equal lower - triangular matrices.
3. Find the value of each variable if
5 x  2 5 3 
    
3 0  3 8  2y) 

B. Addition and Subtraction of matrices

Addition and subtraction on matrices can be well defined (applied) if and only if the
matrices involved have the same dimension or order. Matrices that have the same
dimension are said to be conformable for additions or subtraction. Those matrices that
have different dimensions are said to be non-conformable and thus can not be added or
subtracted.
1 2  2  5
For example,   and   are conformable for addition or subtracted but
3 1  4 0 
3 4 1  0 2
  and   are non-conformable and their sum or difference do not
6 6  3  5 8 
exist.

Definition The sum of two conformable matrices A and B, denoted by A+B, is a matrix
whose elements obtained by adding the corresponding elements of A and B.
More specifically, if
 a 11 a 12 a 13   b11 b12 b13 
   
A=  a 21 a 22 a 23  and B   b 21 b 22 b 23 
a  b b 
 31 a 32 a 33   31 32 b 33 

Then
 a 11  b11 a 12  b12 a 13  b13 
 
A  B   a 21  b 21 a 22  b 22 a 23  b 23 
a  b a 32  b 32 a 33  b 33 
 31 31

Similarly,

7
 :

 a 11  b11 a 12  b12 a 13  b13 

 
A  B   a 21  b 21 a 22  b 22 a 23  b 23 
a  b a 32  b 32 a 33  b 33 
 31 31

2 3 4   1 0 2
Example 3.4 If A    and B   , then find
 0 1  5  4 3 4
a) A+B
b) A-B
Answers
 2  (1) 3  0 4  2  1 3 6
a) A  B    
 04 1  3  5  4   4 4  1

 2  (1) 3  0 4  2   3 3 2
b) A  B    
0  4 1  3  5  4    4 2  9 
Remarks
1. The dimension of A+B or A=B is the same as that of A and B.
2. It is possible to add or subtract three or more matrices if they have the same dimension
(i.e. if they are confirmable).

C. Scalar Multiplication
In matrix algebra, real numbers are called Scalars. Multiplying a matrix by a number (or
scalar) is known as a scalar multiplication.
Scalar multiplication is carried out by multiplying each element of the matrix by scalar.
2 1
Example 1.5 Given A    , then find
3  2 
a) 3A
b) -4A
Answers
 3x2 3x1   6 3
a) 3A      
 3x3 3x(2)   9  6 

  4x2  4x1    8  4
b)  4A      
  4x3  4x(4)    12 8 

8
 :

Note that the negative of a matrix A, denoted as -A, is a matrix obtained by multiplying
all of its elements by -1and - A is called an additive inverse of A since it is always true
that -A+A = 0 = A+(-A) , where 0 is a mull matrix having the same order as A.

2 1   2 1 
For Example, if A   , and  A    ,
 3  2    3 2 
 2  (2) 1  (1)   0 0 
Then A  A  A  (A)    
 3  (3)  2  2   0 0 
D. Multiplication of Matrices
At the first place all matrices can not be multiplied by each other. Matrices should
satisfy certain criteria to be multiplied with each other and this criterion can be
summarized as follows:
Conformability condition for multiplication
Two matrices can be multiplied with each other if and only if the number of columns of
the first matrix is equal to the number of rows of the second matrix and they are said to
be conformable for multiplication.
Given a matrix A of order mxp and matrix B of order qxn, then the matrix product of A
and B, denoted by AB(not BA), is defined only if p=q. In this case, A and B are said to
be conformable for multiplication. If p ≠ q, A and B are non- conformable for
multiplication and thus the product AB is not defined.
It is important to check the dimensions before starting multiplication process.
If A is a matrix of dimension mxp and B a matrix of dimension pxn, then the product AB
is defined since the number of columns of A is “P” which is the same as the number of
rows of B. More over, AB is matrix with a dimension mxn. This is can be shown
schematically as follows.

Matrices: A B = AB
Orders: mxp pxq mxq

Example 1.6
Determine whether the product AB is defined or not for each of the following pairs of
matrices. If AB is defined, find its dimension or order.
a) A with order of 2x3 and B with order of 3 x 5

9
 :

b) A with order of 1 x 2 and B with order of 2 x3

c) A with order of 3 x 2 and B with order of 3 x 4
Answers
a) Since the number of columns of A is 3 which is equal to the number of rows B, AB is
defined and has a dimension of 2x5
Matrices: A B = AB
Dimension: 2x3 3x5 2x5

b) Since the number of columns A is equal to the number of rows of B,

Which equal to 2, AB exists and has a dimension of 1x3.
Matrices: A B = AB
Dimensions: 1x2 2x3 1x3
 Notice here that A is a row matrix and also the product AB is a row matrix.
c) AB is not defined because the number of columns of A is 2, which is different from
the number of rows of B i.e. 3.
Now, let us come to the question “How the multiplication process is take place for
matrices which satisfy the conformability condition?” To make the process easy and
clear, the multiplication process will be described through the following two matrices.

Consider two matrices A and B given as

 b11 b12 
 a11 a12 a13   
A=   and B=  b 21 b 22 
 a 21 a 22 a 23  b 
 31 b32 
Then our interest is to find the product AB.
Notice that before starting the multiplication processes, it is important to check that A
and B are conformable for multiplication or not. To do this, observe that A is 2x3 matrix
and B is 3x2 matrix. So, A and B are conformable for multiplication which means that
the product AB is defined because the number of columns of A, which is 3, is the same
as the number of rows of B. More over, AB is 2x2 matrix.

Now let us denote the product matrix AB as

 d11 d11 
AB=D=  
 d 21 d 22 

10
 :

The multiplication process will be easy and convenient, if the matrices are arranged as
shown below.
b b 
 a11 a12 a13   11 12   d11 d12 
  b 21 b 22    
 a 21 a 22 a 23     d 21 d 22 
 b31 b32 
2x3 3x2 = 2x2
 Then to get each element in the product matrix AB = D, we do the followings:
To get d 11 (1st row and 1st column in D), multiply each element in the first row of
matrix A by the corresponding element in the first column of matrix B and add the
products to obtain d11 =a11b11 +a12 b 21 +a13b31

To get d12 (1st row and 2nd column in D), multiply each element in the first row of
matrix A by the corresponding element in the second column of matrix B and add the
products. The result will be d12 =a11b12 +a12 b 22 +a13b32

To get d21 (2nd row and 1st column in D), multiply each element in the second row of
matrix A by the corresponding element in the first column of matrix B and add the
products. The result will be d 21 =a 21b11 +a 22 b 21 +a 23b31

To get d22 (2nd row and 2nd column in D), multiply each element in the second row of
matrix A by the corresponding element in the second column of matrix B and add the
products. The result will be d 22 =a 21b12 +a 22 b 22 +a 23b33
Remember that you should not bother for the above rigorous multiplication process since
you can tackle it automatically with out such long process after you gone though some
exercises on it.

Example 1.7
Find the product AB, given
 1 3
 2 3 -1   
a) A=   and B=  2 0 
 -2 1 2   -1 2 
 

11
 :

1 2  1 0
b) A=   and B =  
 2 1 1 0
Answers
 2 0  1 2 3   2(1)+0(2) 2(2)+0(0) 2(3)+0(-1)   2 4 6 
a) AB=    = = 
1 3   2 0 -1 1(1)+3(2) 1(2+3(0) 1(3)+3(-1)   7 2 0 
But BA is not defined and thus AB ≠ BA

1 2  1 0  1(1) +2(1) 1(0) +2(0)   3 0 

b) AB=    = = 
 2 1  1 0   2(1)+1(1) 2(0) +2(0)   3 0 
1 0  1 2  1(1) +0(2) 1(2) +0(1)   1 2 
BA=    =  .
 1 0   2 1   2(1)+1(1) 1(2) +0(1)   1 2 
Thus AB ≠ BA even if both AB and B are defined
Remark: Both of the product matrices AB and BA are well defined iff A & B are
square matrices of same order and A has a dimension of mxn and B dimension of nxm.
Exercise-4
1. Write the condition that should be fulfilled by two matrices A and B to get the product
a) AB
b) BA
2. For what types of matrices A and B that both of the product matrices AB and BA are
well defined.
3. for the following pairs of matrices compute the product AB if it is possible.
 1 3
 2 -1 1  
a) A=   and B =  0 -1 
1 3 -2   -2 2 
 
 1
 
b) A=  3 -2 -4  and B =  2 
 -3 
 

13
 :

1 0   2 -5 
c) A=   and B=  
 0 1 4 3
 3 -1 1  1 0 
d) A=   and B=  
 4 2 -2   2 1
 2 -3 -2 
 2 -1 3 0   
  1 0 1
e) A=  -3 4 2 1  and B= 
 0 -2 1 4   -1 2 0
   
 2 -2 -3 

 1 1 1 2 
4) Given A=   find a matrix B such that AB =  
 0 1  0 3
Properties of Matrix Multiplication
Property 1 Matrix multiplication is not commutative. That is AB  BA.
Example 1.9
 1 2  3 -1 
A=   and B=   ,then find AB and BA?
-1 0   4 5
Answers
Let  1 2   3 -1   11 9
AB=    = 
 -1 0   4 5   -3 1
 3 -1   1 2   4 6
BA=    = 
 4 5   -1 0   -1 8

Thus AB ≠ BA even if both AB and BA exist.

Example 1.10
 1 
Given A = (3 2) and B =   , find AB and BA?
 4
Answers
  1
AB = (3 2)   = 5
 4
 1    3  1
AB =   (3 2) =  
4  12 8 
Thus AB  BA
There may be some cases where AB = BA

Example 3.11

14
 :

2 0   3 -1 
Let A =   and B =   , then find AB and BA?
 0 2  4 -2 
Answers
 2 0   3 -1  6 -2 
AB =     = 
 0 2   4 -2   8 -4 
3 1  2 0  6 2
BA =     =  
 4 2   0 2 8 4 
Thus AB = BA
Notice that in this example, A is a scalar matrix i.e, A is a diagonal matrix with equal
diagonal elements.
But this is true not only in this example but also for any two conformable matrices A and
B which either of them or both of them are scalar matrices.

Property 2 Matrix multiplication is associative. That is, for matrices A, B and C, which
are conformable for multiplication, it is always true that (AB) C = A (BC).
If A is mxq matrix, B is pxq of matrix, and C is qxn matrix then
(AB) is mxq matrix and since C is qxn matrix,
(AB) C will be mxn matrix. Amxp Bpxq = (AB)mxq Cqxn = (AB) Cmxn
(BC) is pxn matrix and since A is mxp matrix, A (BC) will be mxn matrix.
Amxp (Bpxq Cqxn) = Amxp (BC)pxn = A(BC) mxn

Example 1.12
2 3 0 1
1
Given: A = 
2 3
 , B = 1  and C =   , then find (AB) C & A (BC).
0 0  0
2 4 1  
 -1  2
 1 2   
Answer
  2 3 0  1 
 1 2 3    
(AB) C =   1 0 0  0
1 

2 3
 -1 1 2    2
 

1
1 6 6     13 
=    0 =  
6 7 2     10 
 2

15
 :

 2 3 0 1  
1 2 3     
A (BC) =   1 0 0  0 
2 3 1   
  -1 1 2   2 
 

 2
1 2 3    13 
=   1  =   Thus, (AB) C = A (BC)
2 3 1     10 
3
Property 3. Matrix multiplication is distributive over addition. That is, for matrices A, B
and C which are conformable for addition and multiplication, it is always true that
A (B + C) = AB + AC
(B + C) A = BA + CA
Notice also that A (B + C) is not always equal to (B + C) A since matrix multiplication is
not commutative.

Property 4 An identity matrix I is a multiplicative identity for matrix multiplication.

That is, for any non-zero (null) matrix A there is a corresponding identity matrix I, which
is conformable with A such that pre multiplying A by I gives the matrix A (i.e, IA = A).
Example 1.13 Find IA and AI, given
1 0 2 1 1
a) I =   and A =  
0 1  3 0  2 

1 0 3  2
b) I =   and A =  
0 1  1 4 

Answers
1 0 2 1 1 2 1 1
a) IA =     =   = A. But AI is not defined
0 1  3 0  2 3 0  2 

1 0 3 2 3 2
b) IA =     =   =A
0 1  1 4 1 4 

3 2 1 0 3 2
AI =     =   =A
1 4  0 1 1 4 
Notice from this example that, if A is any non-zero square matrix then there is a
corresponding identity matrix I such that IA = A = AI

16
 :

If A is a square matrix, then it is possible to multiply it to any power as A.A = A 2, A.A.A

= A3, --- and A2, A3, A4,… are said to be powers of a matrix A.

Example 1.14 Find A2 given

1 0
a) A =  
3 4 

1 0 0
 
b) A =  0 0 0
0 1 
 0

Answers
1 0 1 0 1 0
a) A2 = A.A =     =  
3 4  3 4 15 16 
1 0 0 1 0 0 1 0 0
2      
b) A = A.A =  0 0 0 0 0 0 =  0 0 0  =A
0 1  0 1  0 1 
 0  0  0

Definition A square matrix A is said to be an idempotent if it remains itself when it is

multiplied to any power. That a square matrix A is idempotent if A2 = A3 = - - - = A.
1 0 0
 
For instance, the matrix A =  0 0 0  in the above example is an idempotent matrix.
0 1
 0

Similarly, any identity matrix I is an idempotent since I2 = I3 . . . = I.

Exercise -5
1. Write the major properties of matrix multiplication.
2. Is matrix multiplication always commutative?
3. Is any null or zero matrix is an idempotent matrix?
1  2 4 2 0
4. Given A =   , B =   and C =   , then show that
3  1 -3   -1 1
a) A (B + C)  (B + C) A
b) (B + C) A = BA + CA
 3 4 2
5. Given A =   , than show that A is an identity matrix.
 2 3

17
 :

D. Matrix Transposition
A very useful operation in matrix algebra is that of transposition. The transpose of a
matrix A, is the matrix in which the rows of the original matrix A becomes columns and
the columns of A becomes rows. This transpose matrix is denoted by AT.

Definition The transpose matrix, AT, is the original matrix A with its rows and columns
interchanged. AT will have its dimensions reversed when compared with A. In other
words, if A is mxn matrix, then AT will be nxm matrix.

Example 1.15 Find the transpose of the matrix A, given by

1 2 3 
A 
 2 5 7
Answer
Since A is 2 x 3, its transpose, AT will be 3x 2. Placing the first row of A, (1,2,3), as the
first column of AT and the second row of A, ( 2 5 7 ) , as the second column of AT
yields
1 2 
 
AT=  2 5 
3 7 
 

2 5 3 2 5 3
  T T  
Example .16 Given A=  5 4 -1  , find A . A =  5 4 -1  =A
 3 -1 6   3 -1 6 
   
Notice here that the elements in A are arranged in such a way that every element above
the diagonal is the same as the corresponding element situated symmetrically to it below
the diagonal, i,e, a12 = a 21 = 5, a13 =a 31 =3 and a 23 =a 32 = -1 .Such type of matrix is
said to be a symmetric matrix. Formally the symmetric matrix can be defined as follows.
Definition A symmetric matrix is a square matrix that is equal to its transpose.
A square matrix A is a symmetric matrix if and only if A = AT
From the definition, it is clear that
 A matrix has to be square matrix in order to be symmetric. A non - square matrix A
can not be a symmetric matrix since its dimension and that of its transpose AT are
different.
 Every diagonal matrix is a symmetric matrix.

18
 :

Remember an identity matrix I is an idempotent matrix. It also satisfies the definition of

a symmetric matrix. Thus, an identity matrix I is both idempotent and symmetric matrix,
i.e, it is idempotent-symmetric matrix. Such types of matrices will have useful
applications in some of your further economic courses, particularly in Econometrics.

Properties of Transposes

A 
T
T
1. The transpose of the transpose matrix is is the original matrix A. That is,

A 
T
T
=A

 2 1 3 T
Example 3.17 Find  A T  , given A= 
T
 =A
3 1 4
2 3 
 T 
Answer A =  1 1 
3 
 4 

Now, since  A T  means the transpose of AT , bring the first column of AT as first row
T

of  A T  and the second column of AT as the second row of  A T  .This will gives
T T

A 
T
T
=A

2. The transpose of a sum of matrices is the sum of the transposes of matrices. That is,
for two matrices A and B conformable for addition, it always true that

 A  B
T
 AT  BT

This property is also true for three or more matrices which are conformable for addition.
Example 3.18
1 2   3 1
A=   and B =   ,then find
3 0  -1 1
a)  A+B 
T
Given
b)A T +BT
Answers
1 2   3 1   4 3 
i) A + B =  + = ,
 3 0   -1 1  2 1 

19
 :

4 3 4 2
Then (A + B )T is the transpose of   which is  ,
2 1 3 1
4 2
Thus,  A  B  T=  
3 1
 3 -1
BT =   , then
 1 3   1 1 
ii) AT    and
 2 0 1 3   3 -1   4 2 
A T +BT =  + = 
 2 0   1 1  3 1 
 4 2
There fore, (A+B)T =  T
 = A +B
T

 2 1
3. The transpose of the product matrix AB is the product of the transposes of individual
matrix, taken in the reverse order. That is, for, two matrices A and B which are

conformable for multiplication,  AB =BT .AT

T

This rule can be extended to any product matrix made up of three or more conformable
matrices. For instances, for three matrices A,B and C conformable for multiplication),
the transpose of the product can be defined as

 ABC
T
=CT (AB)T =CT BT AT

2 1
  1 3 
Example 3.19 Given A=  4 3  and B=  ,
1 0   2 2


Show that  AB =BT AT

T

Answers

To get  AB  , compute AB first and then its transpose.

T

AB is 3 x 2 matrix as A is 3 x 2 and B is 2 x 2 specifically,

 2 1 4 8 
  1 3   
AB=  4 3    = 10 18 
1 0   2 2  1 3 
   
Then transposing AB gives,
4 10 1
 AB
T
= 
8 18 3
More over to get BTAT, compute BT and AT first and then their product.

20
 :

1 2  T 2 4 1
BT =   and A =  
3 2 1 3 0 
1 2   2 4 1  4 10 1
BT A T =    = 
 3 2  1 3 0   8 18 3 
Therefore,  AB  =BT A T
T

Exercise-6
1) State the rules of matrix transposition
2) Write what a symmetric matrix means.
3) Write true or false for each of the followings
a) Any scalar matrix is a symmetric matrix
b) A square-null matrix is both idempotent and symmetric.
c) For any two conformable matrix A and B, (AB) T = STBT,
4) Find ABT and AT B if they are defined, for the following pairs of matrices,
 1
 
a) A=(2 4 3 ) and B =  0 
 2
 

1 0 0  1 
   
b) A   2 1 0  and B=  2 
0 2 2 3
   
5) Find ITI, if I is an identify matrix of order 2.
1 0 
6) Find A2 +2AT , for A =  
 2 3

1 0   2 4 
  + 
8 9  0 6 
A2+2AT =
3 4 
= 
 8 15 
1.4 Determinants
To each square matrix there is a corresponding real number assigned to it. This number
is said to be the determinant of the matrix.
Definition A determinant of a square matrix A is a uniquely defined scalar or number
which is the characteristic of the particular matrix A, which denoted by A or det (A).

21
 :

The determinant of an n x n matrix A is called determinant of order n or the nth-order

determinant. For instance, the third-order determinant is the determinant of a 3 x 3
matrix.
From the above definition, observe that
 the determinant is defined only for square matrix
 although a matrix is an array or arrangement of numbers, its determinant is only
a definite number or value,
To get a clear understanding on the method of finding the determinant of a square
matrix, it is better and convenient to start with a 2 x 2 matrix and proceeds further for the
higher order square matrices.
 a11 a12 
Definition The determinant of the 2 x 2 matrix A ( A ) =   is a11 a22 – a21 a12.
 a21 a22 
It is termed as the 2nd– order determinant.
Example 3.20 Find the determinant of A, A , given

1 2 1 2
a)A =    A = =1(5) -2(4) = -3
4 5 4 5

2  3 2 3
b)A =    A = =2 (-6) – (-3)4 = -12 + 12 = 0
4  6 4 6

Notice that for a 1 x 1 matrix A = (a11), the determinant of A is defined to be the number
a11, i.e, A = a11 for A = (a11). For examples, if A = (-6), then A = -6.

Definition A square matrix A, whose determinant is zero, is said to be a singular matrix.

1 2
If A  0, then A is called non-singular. For example, A =   is a singular matrix
1 2 
since A = 1 (2) – 2(1) = 0. Such property of the determinant of a matrix will be useful

for finding the inverse of a matrix.

Before defining the determinant of a 3 x 3 matrix it is better to discuss on the new

concepts called minors and cofactors since these concepts simplify the required
computation.

22
 :

Minors and Cofactors

Consider a matrix A with typical element aij. That is,
 a 11 a 12 a 13 . . . a 1n 
 
 a 21 a 22 a 23 . . . a 2n 
A= 
    
 
 a n1 a n2 a n3 . . . a nn 
 

Definition The minor associated with the element aij, which denoted as Mij, is the
determinant of the sub-matrix formed by deleting the ith row and jth column of the
matrix A.
2 3
For example, if A =   , then M11, the minor of a11 = 2, is the determinant of the sub
4 5 

matrix formed by deleting the first row and first column of A. That is, M11 = 5 = 5.

Notice that the number of minors for a square matrix A is equal to the number of
elements in A. For instance, there are 9 possible minors for a 3 x 3 matrix A since there
are 9 elements in A.
3 4 7
 
Example 3.21 Given A =  2 1 3  , find the minors of the elements a13 and a23.
7 1 
 2

Answers
To get the minor of a13, M13, delete the first row and third column of A to form a sub
2 1
matrix   and then evaluate the determinant of this sub matrix. This will result in
7 2 

2 1
M13 = = 2(2) – 1(7) = -3.
7 2
Similarly, to get the minor of a23, M23, delete the 2nd row and 3rd column of A to obtain a
3 4
sub matrix   and then evaluate its determinant. This will result in
7 2 

3 4
M23 = = 3(2) – 4(7) = -22
7 2

23
 :

Definition The cofactor of the element aij, denoted by Cij, is the minor of that element
coupled with the appropriate sign determined by (-1)i + j
. In short, the cofactor is the
signed minor which obtained as
Cij = (-1) i + j Mij, i = 1,2, - - n j = 1,2, - - n.
Notice that
If i + j is an even number, then the cofactor and the minor associated with aij are equal,
i.e, Cij = Mij.
If i + j is an odd number, then the cofactor is the negative of the minor, i.e., Cij = -Mij.
3 4 7
 
Example 3.22 Find C11 for A =  2 1 3
7 1 
 2

Answers
1 3
First delete the first row and first column of A to obtain the sub matrix   and
2 1 
1 3
evaluate its determinant to get the minor M11 = = -5. Then after compute C11, the
2 1

cofactor of a11 = 3, as
C11 = (-1) i + j M11
= (-1) 1 + 1 – 5
= (-1)2 – 5
=-5
Definition The cofactor matrix of A, denoted by C, is a square matrix formed by
replacing each element aij of A with its associated cofactor Cij.
Notice that the following sign matrix obtained from (-1)i+j will be useful to covert Mij
directly in to Cij.
   . . .
 
   . . .
   . . .
 
      

3 4 7
 
Example 3.23 Find the cofactor matrix C for A =  2 1 3
7 1
 2

Answers

24
 :

  
 
Remember that the signed matrix from (-1) i+i
for 3x3 matrix is      .
  
 
Use the formula Cij = (-1)i+j Mij by taking the sign of (-1)i+j from the above sign matrix
to get
1 3 2 3 2 1
C11 = M11 = = -5 C12 = M12 = -   (19)  19 C13 = M13 =  3
2 1 7 1 7 2

4 7 3 7
C21 = -M21 = -   (10)  10 C22 = M22 =  1 (46)   46
2 1 7 1

3 4 4 7
C23 = - M23 = -   (22)  22 C31 = M31 = 5
7 2 1 4

3 7 3 4
C32 = M32 = -  () (5)  5 C33 = M33 =  5
2 3 2 1
Therefore,
 C11 C12 C13   5 19 3 
   
C =  C 21 C 22 C 23  =  10  46 22
C C33   5  5 
 31 C32  5

Obtaining the Determinant of a 3x3 matrix

 a11 a12 a13 
 
Consider a 3x3 matrix A given by A =  a 21 a 22 a 23 
a a33 
 31 a32

The determinant of A, A , can be obtained through the following steps.

Step 1: Take any row or column of A

Step 2: Multiply the elements of the row or column taken in step 1by the corresponding
cofactors,
Step 3. Add all products in step 2 to reach the determinant, A .

This method of finding the determinant is known as cofactor Expansion.

The other equivalent way of obtaining A is through the following steps.

Step 1: Take any row or column of A.

Step 2: Attach each element in the given row or column with its corresponding sign
obtained from (-1)i+j or the sign matrix.

25
 :

Step 3: Multiply the signed elements by their respective minors.

Step 4: Add all products in step 3 to get the determinate A

This method of finding the determinant is known as minor expansion.

For example, using the first row of the above 3x3 matrix A and the corresponding
cofactors, the determinant can be obtained as
A = a11 C11 + a12 C12 + a13 C13

Alternatively, using the corresponding minors and the respective signs of the elements
  
 
obtained from (-1) i+j
or the sign matrix      , the determinant can be obtained as
  
 
A = a11M11 - a12M12 + a13 M13

Similarly it is possible to calculate A by expanding a long the first column of A using

cofactors as A = a11 C11 + a21 C21 + a31C31

Or using minors as A = a11 M11 - a21 M21 + a31 M31

3 4 7
 
Example 3.24 Find A for A   2 1 3
7 1
 2

Answers
Using the first row of A for expansion and through minor expansion method A can be

obtained as follows.
Attach the corresponding sign to each element in the first row by recalling the sign
 3 4 7 
 
matrix as  2 1 3  and find the respective minors as
7 2 1 

1 3 2 3 2 1
M11 = , M12 = and M13 =
2 1 7 1 7 2

Then calculate A as

A = a11 M11 - a12 M12 + a13 M13

26
 :

1 3 2 3 2 1
=3 -4 +7
2 1 7 1 7 2

= 3 (1-6) - 4 (2-21) + 7 (4-7)

= -15 + 76 - 21
= 40
Notice that expanding a long any row or column of the matrix results the same an answer
for A . Thus the choice of which particular row or column to be taken may be guided by

computational ease. One rule of thumb is to choose the row or column of the matrix with
the largest number of zero elements, if it exists.
12 98 15 
 
Example 1.25 Find A for A =  0 25 0 
 21 84 19
 
Answers
It is better to expand a long the second row since it has the largest number of zeros.
Attach the corresponding sign to each element of this row from the sign matrix as
12 98 15 
  
 0 25 0   and multiply these signed elements by their respective minors to get the
 21 84 19 
 
determinant as A = - a21 M21 + a22 M22 - a23 M23

98 15 12 15 12 98
=0 + 25 0
84 19 21 19 21 19
= 0 + 25 (228 - 315) - 0
= 25 (-87)
= - 2175
The methods of finding the determinant outlined above for the case of a 3x3 matrix A
can be extended in a similar way for any higher orders square matrix. But from a
computational point of view, obtaining the cofactors or minors becomes a fairly
complicated task. For example, in the case of a 4x4 matrix A given below
 a 11 a 12 a 13 a 14 
 
 a 21 a 22 a 23 a 24 
A= 
a a 32 a 33 a 34 
 31 
 a 41 a 42 a 43 a 44 


27
 :

Obtaining any minor, say M14, would involve solving the determinant of a 3x3 matrix,
a 21 a 22 a 23
since, M14  a 31 a 32 a 33 and the determinant of the above 3x3 matrix, M14 obtained
a 41 a 42 a 43

through finding the corresponding three 2x2 minors or cofactors a long a given row or
column. So to get the determinant of this 4x4 matrix A, four 3x3 minors or cofactors
each of them in turn involves three 2x2 minors or cofactors for a total of twelve 2x2
minors or cofactors should be calculated. Thus calculating the determinant of a large
square matrix using the methods out lined above is quite long and cumbersome.
Consequently to get determinant of square matrices higher than order 3, it is advisable to
use other alternatives like methods which depend on the properties of the determinant
that are stated in the next subsection, or using computer aids.

Exercise -7
1. Define each of the following terms or concepts
a) Determinant b) Minor c) Cofactor d) Cofactor matrix e) Singular matrix
 2 1
2) Given A =   , find
3 4 
a) A b) Cofactor matrix, C c) AT and compare it with A

 2 3 5
 
3) Give A  1 9 8 , then find A
4 6 7 
 
a) By expanding a long the first row. b) By expanding a long the second column
c) What do you conclude from the results in (a) and (b)
2 2 4 9
 
4 1 0 2
4) Find A for A  
4 1 0 0
 
3 1 
 2 1

Properties of Determinants
This subsection focuses on the major properties of determinants that are useful in
reducing the computational burden of calculating the determinant of large square matrix.

28
 :

Each property will be illustrated with example using 2x2 matrices. This is due to only to
make the computations easy. The properties also true for higher order square matrices.
Property 1 The determinant of the transpose matrix of A, A T , is the same as the

determinant A for any square matrix A. That is, for any square matrix A, A  A T .

1 2 
Example 1.26 For A    , find A and A T
3 4 
Answers
1 3 T
A  4  6   2 and since A T   , A  4  6  2 . Thus A  A T
 2 4
Property 2 If B is obtained from A by interchanging the rows or columns of A, then
B  A

1 2 
Example 1.27 Given A   , , find B if B is formed
2 4
a) by interchanging the rows of A b) by interchanging the columns of A

Answers
3 4 
a) B    and B  6  4  2.Since A   2, B   A
1 2 
 2 1
b) B    and B  6  4  2.Since A   2, B   A
4 3
Property 3 If a matrix has two or more identical rows or columns its determinant will be
zero.
1 2 1 1 
Example 1.28 Find A and B , for A    and B   
1 2  2 2
Answers
A  2  2  0 ( identical rows) and B  2  2  0 ( identical Clumns )

Property 4 If one of the rows of A is a multiple of any other row or if one of the
columns of A is a multiple of any other column, then A  0

 5 3 2 4 
Example 1.29 Find A and B for A    and B   
10 6  3 6
Answers

29
 :

In A the second row is twice of the first row and A  30  30  0 .

In B the second column is twice of the first column and B  12 12  0 .

Property 5 If B is formed from A by adding a multiple of one row of A to any other

row, or a multiple of one column of A to any other column, then, the value of the
determinant remains unchanged i . e, B  A .

1 2 
Example 1.30 Given A   , find the determinant of B formed by
3 4 
a) adding twice of the second row to the first row
b) subtracting the second column from the first column
Answers
 7 10
a) B    and B  28  30   2. Since A  4  6   2, B  A
3 4 
 1 2
ii) B    and B   4  2   2. Thus B  A
 1 4 

Property 6 Multiplying any row or any column of a matrix A by a scalar (number) k,

multiplies the determinant of A by k. That is, if B is formed from A by multiplying a row
or column of A with K, then B  K A

1 2 
Example 1.31 Given A   , find the determinant of B formed by multiplying the
3 4 
first row of A with k=3.
Answers
 3x1 3x2   3 6 
B     and B  12  18   6 .
3 4   3 4 

Since A   2 and  6  3 2, B  3 A  K A .

Notice that multiplying every element of nxn matrix A by a scalar K, multiplies the
determinant by Kn . In other words if B is formed from nxn matrix A by multiplying
every element of A by K, then B  K n A .

Example 1.32 Find the determinant of B formed by multiplying every element of

1 2 
A    by 3 and compare it with A .
3 4 

30
 :

Answer
 3x1 3x2   3 6 
B       and B  36  54   18
 3x3 3x4   9 12
Since A   2 and  18  32  2  B  32 A  K 2 A .

Property 7 The determinant of the product of two square matrices A and B of the same
order is equal to the product of each determent. That is, AB  A B .

1 2   2 3
Example 1.33 Given A    and B    , then show that AB  A B .
3 4   4 5
Answer
1 2  2 3  10 13 
AB        and AB  10(29) 13(22)  4 .
 3 4  4 5   22 29
Since A  2 and B  2, A B  4 , therefore, AB  A B .

Property 8 The determinant a triangular matrix is equal to the product of the diagonal
elements.
 a 11 0 0
 
For example If A   a 21 a 22 0  then A  a 11 a 22 a 33
a 
 31 a 32 a 33 
Exercise -8
1. Write the major properties of Determinants
2. Find A if

a) A is an identity matrix of order n

b) A is a 3 x 3 diagonal matrix having 2,3, and -1 on its main diagonal.
3. Illustrate using example that for two square matrices A and B, A  B  A  B

4. Explain why each of the following is true

1 5 4 
 
a) If A   2 3  1, then A  0
1 5 4 
 
 2 3 3  2
b) If A    and B   , then B   A
 5 4 4 5 

31
 :

2 3 4
 
c) If A   4 6 8 , then A  0
1 5  2 


 4 1 5   4 3 1
   
d) If A    3 2 3 , and B   1 2 0 , then A  B
 1 0 6   5 3 6 
 

1.5 Matrix Inversion

In the previous sections, you learned the operations of addition, subtraction and
multiplication of matrices. What do you think about division? Is it possible to divide one
matrix by another matrix?
You also remember that in the set of real numbers, for each number a (except zero) there
exists a real number a-1 = 1/a, the multiplicative inverse of a, such that aa-1 = a-1a =1. Do
you think that there exists a multiplicative inverse for any matrix other than the null
matrix? This section attempts to show the methods of finding the multiplicative inverse
of a matrix. Finding the inverse of a matrix has important applications in solving large
systems of linear equations and in determining the solutions of many practical problems
which can be expressed in matrix form.

If A and B are square matrices, with equal dimension n, which satisfy the equations
AB = BA = I, Where I is the identity matrix having the same order as A or B, then B is
called the inverse of A. B is said to be the inverse because, if it exits with this property,
then it is uniquely determined by A and can be written as B  A 1 .
2 3  2  3
For example, if A =   and B =   , then
1 2  1 2 

2 3  2  3   2(2)  3(1) 2(3)  3(2)   1 0

AB =     =  = = I
1 2  1 2  1(2)  2(1) 1(3)  2(2)   0 1

 2  3  2 3   2(2)  (3)1 2(3)  (3)2   1 0

BA =    = = = I
1 2  1 2    1(2)  2(1) 1(3)  2(2)   0 1
Thus, since AB = BA = I, B is said to be the inverse of A. That is B = A-1.
Definition The inverse of a square matrix A (if it exists), denoted by A -1 (read as “A
inverse”), is a matrix that satisfied the condition AA-1 = A-1 A = I, Where I is the identity
matrix having the same order as A.

32
 :

Notes
 Inverse of a matrix is defined only for square matrices. Non-square matrices do not
have multiplicative inverses.
 Not every square matrix has an inverse. A square matrix A has an inverse only if it is
non-singular (i.e, A  0). If A is a singular matrix (i.e, A = 0), then it does not

have an inverse.
 Inverse of a matrix, if it exists, is unique. That is, a matrix can not have more than
one inverse.
There are two methods of finding the inverse of a given square matrix: Co-factor
Method and Gauss – Jordan elimination Method, and each of them will be discussed
in the next two sub section in details.
A. THE CO-FACTOR METHOD
Consider a 3 x 3 matrix A given by
 a 11 a 12 a 13 
 
A=  a 21 a 22 a 23 
a a 33 
 31 a 32

Then the inverse of A can be obtained through the following steps.

Step 1 Find the determinant of A, A .

If A = 0 (i.e, A is singular), then A has not an inverse.

If A  0 (i.e, A is non-singular), then A has an inverse and proceed to the next

step.

Step 2: Determine the cofactor matrix of A, C.

 c11 c12 c13 
 
C =  c21 c22 c23  , Where Cij is the cofactor of the element aij in A.
c c33 
 31 c32

Step 3 Find the adjoint matrix of A, denoted as adj (A).

Definition The adjoint matrix of A, denoted as adj (A), is the transpose of the
cofactor matrix of A (or C). That is,

33
 :

 c11 c 21 c31 
T

adj (A) = C =  c12 c 22 c32 
c c33 
 13 c 23

Step 4 Divide each elements of adj (A) by A to get A-1. That is, compute A-1 as

 c11 c 21 c 31 
-11 1  
A = adj(A) =  c12 c 22 c 32  .
A A 
 c13 c 23 c 33 

Notice that the above steps can be used to find the inverse of any square matrix and thus
it is possible to generalize the method of finding inverse matrix as follows.
Definition The inverse of an n x n matrix A is the adjoint matrix of A divided by the
determinant of A. That is,
1
A-1 = adj(A) .
A

1 2 3
 
Example 1.34 Find A -1
for A =  0 1  1
1 1 
 2

Answer
First compute A . To do this, use the first row of A for expansion, attach each elements

of this row with the corresponding sign from the sign matrix and calculate A as

1 1 0 1 0 1
A = M11 -2M12 + 3M13 = 2 3 = 3 – 2(1) + 3(-1) = -2
2 1 1 1 1 2

Since A = -2  0, A has an inverse.

Then determine the cofactor matrix of A, (C).

 M11  M12 M13 
 
C=  - M 21 M 22  M 23 
 M  M 32 M 33 
 31

 1 1 0 1 0 1 
  
 2 1 1 1 1 2 
   3 1  1
2 3 1 3 1 2  
=    4 2 0
 2 1 1 1 1 2 
   5 1 1
 2 3 1 3 1 2
 1 1 
 0 1 0 1 

34
 :

Transpose C to get the adjoint matrix as

 3 4  5
 
adj (A) = C =   1
T
2 1
 1 1
 0

Finally Calculate A-1 as

 3 4  5    3/2 2 5/2 
1 1    
  1  2 1    1/2  1/2 
-1
A = adj(A) = 1
A 2
 1 0 1   1/2 0  1/2 

Notice that it is possible to check the correctness of the result through verifying that A-1
satisfies AA-1 = I.

Exercise -9
1. Does an inverse matrix exist for any square matrix?
2. Find the inverse of the following matrices if it is possible using the Co-factor method.
 1 2 1 3 0 0 1 2 3
5 7      
a) A =   b) B =  0 1 0  c) C = 0 1 0  d) D = 0 1 2
2 3  5
 2 3  0
 0 3  0 0
 1 

B. Gauss – Jordan elimination Method

This method is particularly very useful in computing the inverse of high – order matrix in
which evaluating high order determinants becomes so complicated to apply the Co-factor
method.
The Gauss – Jordan elimination Method involves augmenting the matrix to be inverted
by including the identity matrix next to it. Then applying the necessary elementary row
operations to both matrices until the original matrix is transformed into the identity
matrix. By doing this, the former identity matrix will have been transformed into a new
matrix which will be the inverse of the original matrix.
Elementary operations mean the following three types of row or column operations:
i. Interchanging any two rows (or columns)
ii. Multiplying any row (or column) by any non-zero constant number; and
iii. Adding a multiple of one row (or column) to a multiple of another row (column).
Notes
1 An elementary operation should be either row or column operation, but not both at
the same time.

35
 :

2 An elementary operation should involve every element of the row or column. It is

prohibited to operate in some elements of the row (or column) and ignore some
elements of the row (or column).
3 Although elementary operation can be conducted either on rows or columns, only
row operations will be applied in this sub section.
Notations
 Denotes interchanging, example R1  R2 means row 1 and row are interchanged.

 Denotes replacement, for instance  3R1  R2  R2 means minus 3 times row one
plus row two in the pervious augmented matrix replaces row two in the new augmented
matrix.

 2 1
Example 1.35 Find the inverse of A   
 1 1
Answer
First augment matrix A with corresponding identity matrix. That is, write it as
2 1 1 0 
A I or  
 1 1 0 1 

Notice that the Gauss – Jordan elimination Method starts from A I  and apply the

necessary row operations to end up with I A -1  . In other words, the matrix A has to
be transformed into the identity matrix so that the first column becomes 1,0 and the
second column becomes 0,1.
To do this, it will be convenient to start working systematically on the first column, and
then on the second column.
Now, to change a11 element (or 2) to 1, multiply the first row of the above augmented
1
matrix by :
2

 1 -1 2 1 2 0 

1 0 1 
 1
R R 
 1
2 1 1

Then change a 21 element (or -1) to 0 by adding row one and row two of the above
augmented matrix:

36
 :

1 -1 2 1 2 0 
  R1  R2 R2 
0 1 2 1 2 1 

Next multiply the second row of the above augmented matrix by 2 to change a 22 element
(or 1/2) to 1.
 1 -1 2 1 2 0 
  2 R2 R2 
0 1 1 2 
Finally, in the above augmented matrix, add 1/2 times row two and row one to change
the a12 element (or -1/2) to 0:

1 0 1 1
  1
R  R R
2 2 1 1

 0 1 1 2
1 1 
Therefore, the inverse of is A will be A 1    .
1 2 
Remember that you can easily check your answer by showing
 2  1 1 1  1 0 
AA 1     I.
 1 11 2   0 1
Notice also that, to save time and space, the necessary row operations can be performed
on two or more rows simultaneously as it will be illustrated in the next example.
 3 1 1 
 
Example1.36 Find the inverse of B    1 1 0 
 2 2  1
 
Answer
Write the desired augmented matrix as
 3  1 1 1 0 0
 
1 1 0 0 1 0
2 2  1 0 0 1 

Now just like the previous example the best procedure will be first transform the first
column into 1,0,0,then the second into 0,1,0 and finally the third into 0,0,1.
1
To do this, first multiply the first row of the above augmented matrix by to get 1 in the
3
a11 position:

 1 1 3 1 3 1 3 0 0 
 
 -1 1 0 0 1 0 1
R R1
3 1

2  1 0 0 1 
 2

37
 :

Then add the first and the second rows of the above augmented matrix to get 0 in
1
the a 21 position and also multiply the third row by and add it on the first row to get 0
2
in the a 31 position:

 1 1 3 1 3 1 3 0 0 
   R-11  R 2 R 2 
0 2 3 1 3 1 3 1 0   R 3  R1R 3 
 0 - 4 3 5 6 1 3 0 -1 2  2 
 
To change the a 22 element (or 2/3) to 1, multiply the second row of the above
3
augmented matrix by :
2

 1 1 3 1 3 1 3 0 0 
 
0 1 12 12 32 0  3
R R 2
2 2

 0 - 4 3 5 6 1 3 0 -1 2 
 
1
Next add times of the second row on the first row of the above augmented matrix to
3
4
get 0 in the a12 position and also add times of the second row on the third row to get 0
3
in the a 32 position:

1 0 1 2 1 2 1 2 0 
   13 R 2  R1R1 
0 1 1 2 1 2 3 2 0   4 
0 0 3 2 1 2 - 1 2   3 R 2  R 3 R 3 

3 2
To change the a 33 element (or ) to 1, multiply the third row by :
2 3

1 0 1 2 1 2 1 2 0 
 
0 1 1 2 1 2 3 2 0   2
R R 3
3 3

 0 0 1 2 3 4 3 -1 3
 
1
Finally, in the above augmented matrix, add times of row three on both first and
2
second row to get 0 in the position of a13 and a 23 respectively:

 1 0 0 1 6 -1 6 1 6 
   -21R 3  R1R1 
0 1 0 1 6 5 6 1 6   -1 
 0 0 1 2 3 4 3 -1 3  2 R 3  R 2 R 2 
 
Therefore, the inverse of B will be

38
 :

 1 -1 1 
 6 6 6
 1 
B  1
-1 5
6 6 6
 
 2 4 -1 
 3 3 3
Notes
1. It is possible to know a matrix has an inverse or not through the Gauss – Jordan
Elimination Method. A square matrix has not an inverse (or it is singular) if it can not be
transformed into the identity matrix through any type of row operations.
 3 1
Example 3.37 Find the inverse of A   
 3 1
Answer
3 1 1 0
 
3 1 0 1
 1 1 3 1 3 0
 
1 0 1 
 1
R R
3 1 1

 3
 1 1 3 1 3 0
  3 R1  R2 R2 
 0 0 -1 1 
Since every element in the second row is 0, no further row operations can provide the
identity matrix and hence the matrix A has no inverse.

2. Sometimes interchanging two rows will be necessary to perform the row operations
through the procedure explained in the pervious examples.
 0 1 3
 
Example 1.38 Find the inverse of B   1 2 1
 1 0 
 -2

Answer
0 1 3 1 0 0
 
1 2 1 0 1 0
1  2 1 
 0 0 0

1  2 0 1 0 0
 
1 2 1 0 1 0 R1R3 
0 3 0 0 1 
 1

39
 :

1  2 0 1 0 0
 
0 4 1 -1 1 0  -R1R 2 R 2 
0 3 0 0 1 
 1

1  2 0 1 0 0
 
0 1 1 4 -1 4 14 0 1
R R 2
4 2

0 1 
 1 3 0 0

1 0 12 12 12 0

0 1 1 4 -1 4 14 0


2R 2  R1 R1
- R 2  R 3 R 3 
0 1 
 0 11 4 1 4 -1 4

1 0 12 12 12 0 
 
0 1 1 4 -1 4 14 0   4
R R 3
11 3

0 4 11
 0 1 1 11 - 1 11

1 0 0 5 11 6 11 - 2 11
   -21R 3  R1R1 
0 1 0 - 3 11 3 11 - 1 11  -1 
0 4 11   4 R 3  R 2 R 2 
 0 1 1 11 - 1 11

 5 11 6 11 - 2 11
 
Therefore, the inverse of B will be B-1   - 3 11 3 11 - 1 11 
 1 11 4 11 
 - 1 11

Properties of the Inverse Matrix

Property 1 The inverse of an inverse matrix gives the original matrix.

That is, A 1   A
1

Property 2 The inverse of the product matrix AB is equal to the product of the inverse
matrices in reverse order. That is,
AB1  B1A 1 Provided that A and B are of the same order and A-1 and B-1 exist.
Property 3 The inverse of the transpose is equals to the transpose of the inverse. That is,
(AT) -1 = (A-1) T
Property 4 The determinant of the inverse matrix is 1 over the determinant of the
1
original matrix. That is, A 1 
A

Exercise -10
1. Write the advantages of the Gauss – Jordan Elimination Method over the Co-factor
method in computing an inverse of a matrix.

40
 :

0 1  5 -3 
2. Given A =   and B =   , then show that
2 8 2 -1 
a) (A-1)-1 =A
b) (AB)-1 = B-1A-1

3. Find the inverse of the following matrices if it is possible using the Gauss – Jordan
Elimination Method
 3 2 3 2 2 3
2  3    
a) A =   b) B =   1 0 3 c) C = 1 0  3
1 1   2
 1  1  3
 4 0 

 -3 0.5 2
 
d) D =   6 1 4
 6 0 
 3

1.7 Linear Dependence and Rank of a Matrix

The concept of linear dependence is based on the idea of linear combination. So, before
presenting the former idea it will be better to have some idea on the later concept.
Consider the matrix A given by
 a 11 a 12 a 13 
 
A=  a 21 a 22 a 23 
a a 33 
 31 a 32

A linear combination of row1, 2, and 3 is obtained by summing up the multiple of the

first, second and third rows by a row vector (or a constant K1, K2, and K3 where at least
one of them is non-zero). That is, the linear combination of the three rows can be
expressed as
 a 11 a 12 a 13 
 
K 1 K2 K 3   a 21 a 22 a 23 
a a 33 
 31 a 32

 K1 (a11 a12 a13 )  K 2 (a 21 a 22 a 23 )  K 3 (a31 a32 a33 )

K1a11  K 2a 21  K3a 31 K1a12  K 2a 22  K 3a 32 K1a13  K 2a 23  K 3a 33 

=
 C1 C2 C3 

41
 :

1 2 1
 
Example 1.39 Write the linear combination of all rows of A =  3 2 4
 -1 - 3 
 0

Answer
K1 1 -2 1  K 2 3 2 4  K 3  - 1 0 - 3

K1  3K 2  K3 - 2K1  2K 2 K1  4K 2  3K 3 
  C1 C2 C3 
Notice that a linear combination of columns of a matrix can be obtained similarly by
summing up the constant multiple of each column.

Definition: The rows (columns) of a matrix are said to be linearly dependent if some
linear combination of them is a null or zero row (column). Otherwise they are said to be
linearly independent.

For instance, in example 1.39 the rows will be linearly dependent if the three elements
C1, C2 and C3 are all equal to zero for some values of K1, K2, and K3 with at least one of
them is non-zero.

Example 1.40 Test for linear dependence of a matrix

1 2 4 
 
A=  2 4 8 
3 12
 6

Answer
Row1 (R1), Row2 (R2) and Row3 (R3) are linearly dependent since
-1R1 -1R2+ R3= -1(1 2 4)-1(2 4 8)+(3 6 12)
=(-1 -2 -4)+(-2 -4 -8)+(3 6 12)=(0 0 0)
Furthermore, R1 and R2 are linearly dependent since -2R1 + R2 = (0 0 0). Similarly R1
and R3 are linearly dependent since -3R1 + R3 = (0 0 0).

Definition: The rank of a matrix is the maximum number of linearly independent rows
(columns) in the matrix.

There are two possible ways of determining the rank of a matrix:

42
 :

I. Determinant Method
II. Gauss Jordan Elimination Method
I. Determinant Method
This method involves evaluating the determinants of the matrix (if it is a square matrix)
and possible square sub-matrices contained in the matrix. Then the dimension of the
highest with non-zero determinant is taken as the rank of the matrix.
The major steps involved in this method for (n  n) square matrix, A, are as follows:

Step 1 Compute the determinant of A, A

 If A  0 , then the rank of A is n.

 If A  0 , go to the next step.

Step 2 Find any square sub matrix of order n-1 and evaluate its determinant.
 If the determinant is non-zero, then the rank of A is n-1.
 If not, evaluate the determinant of any next lower sub matrix and proceed
it until a non-zero determinant is obtained.
Examples 3.41 Find the rank of the following matrices.
1 2 4 
 5 2  
a) A =   b) B=  2 4 8 
 2  3  3
 6 12

Answers
5 2
a) Since A   11  0 , the rank of A is 2 and thus the two rows (columns) are
2 3
linearly independent.
b) The rank of B cannot be 3 as B  0 . The rank of B cannot also be 2 since the

determinants of all possible (2 2) sub-matrices in B are zero. For instance,

1 2 2 4
 0,  0 and so on.
2 4 3 6
Since the next square sub matrix would be of order 1 which is non-zero, rank of B is 1.

N.B:
 If the rank of a matrix is 1, then all the rows (columns) of the matrix are said
to be linearly dependent. Hence, all the three rows (columns) of the above
matrices, B, are linearly dependent.

43
 :

 The first step in determining the rank of non-square matrix is evaluating the
determinant of the highest square sub matrix contained in the matrix as the
determinant is defined only for square matrices.
 3 3 2
Examples 1.42 Find the rank of A   
 2 0 5 
Answer
The rank of A can not be 3 since it is impossible to find any (3 3) matrix in it. Thus, the
rank of A can be at most 2.
To see to see whether it is actually 2, find any (2 2) non-zero determinant sub matrix.

3 3
Since  6  0 , the rank of A is 2.
2 0

Notes
1. The rank of a matrix can not exceed the number of rows or columns which ever
smaller. For instance, the rank of a (3 4) matrix can not be more than 3 (the
number of rows) which is less than the number of columns.
2. The rank of a matrix is at least one, unless the matrix has all zero elements,
i,e, when it is a null matrix. The rank of a null matrix is zero.
3. Rank of the transpose of a matrix A is the same as the rank of A.

II. Gauss Jordan Elimination Method

In this method, the row operations used in the Gauss Jordan Elimination Method are
applied to a matrix and the rank of the matrix is taken as the dimension of the highest
order identity matrix that remains in the final reduced matrix.
The method is particularly useful in finding the rank of high order matrix in which
evaluating the determinant of possible square sub matrices would be tedious and
cumbersome.
1 2 1
 
Examples 1.43 Find the rank of A=  2 1 1 using Gauss Jordan Elimination
4 3 
 5

Answer

44
 :

1 2 1
 
2 1 1
4 3 
 5

1 2 1
   - 2R 1  R 2  R 2 
0 -3 - 1  
0  - 3R 1  R 3  R 3 
 3 - 1

1 2 1 
   - 1 R 
0 1 1 3  R2

0 3 2 
 3 - 1 

1 0 1 3
   - 2R 2  R 1  R 1 
0 1 1 3  
0  3R 2  R 3  R 3 
 0 0 

Since no further reductions are possible and the highest order identity matrix in the final
reduced matrix is (2 2) , the rank of A is 2.

Exercise -11
1. Write the steps involves for finding the rank of non-square matrix.
2. Determine the rank of the following matrices using the determinant method and test
the rows for linear independence.
 1 1 3
2  1  
a) A =   b) B =  2 1 1
1 1   0
 -2 1 

1 0 0 1 0 0  1
   
c) C = 1 1 1 d) D =  0 2 2 1
0 1 2 1
 1  1 3

3. Determine the rank of the following matrices using Gauss Jordan Elimination Method.
 1 2 3 1 3 4  2
   
a) A=  2 3 1 b) B=  2 6 8 -4
- 2 - 1 3 3 
 -3  0 3

45
 :

1 -2 0 3 5
 
2 1 0 2 4
c) C= 
0 -6 2 0 0
 
- 2 4 
 2 1 -1

1.8. Trace of a Square Matrix

Definition
Let A = (ai;j) be a square matrix of order n. The trace of the matrix is the sum of the
main diagonal:

Trace (A) = i 1 ai, j

n

Notation: The trace of a matrix A is also commonly denoted as Tr(A) or TrA.

Properties:
1. The trace is a linear transformation from the space of square matrices to the real
numbers. In other words, if A and B are square matrices with real (or complex)
entries, of same order and c is a scalar, then
Trace (A + B) = trace (A) + trace (B);
Trace (cA) = c*trace (A):
2. For the transpose and conjugate transpose, we have for any square matrix A with real
(or complex) entries,
Trace (At) = trace (A);
Trace (A*) = trace (A):
3. If A and B are matrices such that AB is a square matrix, then
Trace (AB) = trace (BA):
For this reason it is possible to define the trace of a linear transformation, as the
choice of basis does not affect the trace. Thus, if A; B; C are matrices such that ABC
is a square matrix, then
Trace (ABC) = trace (CAB) = trace (BCA):
4. If B is in invertible square matrix of same order as A, then
Trace (A) = trace (B-1 AB): In other words, the trace of similar matrices are equal.
In particular, trace A*A 0 with equality if and only if A = 0.

46