Chapter 04 Matrices

Introduction

A matrix is a rectangular array of numbers. In a rectangular array, the numbers are

written in rows and columns. If the array contains m rows and n columns we write the

array of mn numbers using double suffixes as follows:

 a11 a12 a13 . . a1n 

 
 a 21 a 22 a 23 . . a2n 
a a32 a33 . . a3n 
A=  31 
 . . . . . . 
 . . . . . . 

a a mn 
 m1 am2 a m3 . .

The array is enclosed by a pair of parenthesis to indicate that all these numbers belong to

the same array. The numbers in the array are called the elements of the matrix. When

referring to a matrix in general (for example when proving general theorems), for

simplicity we denote such a matrix by a ij mn . This symbol indicates that the ij
th
element

(i.e. the element at the intersection of the i th row and the j th column) of the matrix is aij ,

and mn, is referred to as the type of the matrix .A matrix having the same number n of

rows and columns is called a square matrix of order n. Sometimes, a matrix is

conveniently named just by means of an upper case letter if it is refereed several times.

Example: -

 1 2 3
A =   is a matrix of type 23.
  1 0 2

Definition

Two matrices A and B are said to be equal, written A  B, if they are of the same type and

the corresponding elements of A and B are equal.

Basic Operations on Matrices

1. Matrix addition

Let A and B be two matrices of the same type. The sum of A and B, written A  B ,

is defined to be the matrix obtained by adding the corresponding elements of A and

B.

If A = ( a ij ) mn and B = (bij ) mn then A + B = (cij ) mn ,where cij  aij  bij .

Example: -

 1 3  3  3 3 2  4 6 1
If A    , B    then A  B    .
 0 2  5 9 2 1  9 4  4

2. Scalar multiplication

The product of a scalar k and a matrix A, written kA, is defined to be the matrix

obtained by multiplying each element of A by k.

If A = ( a ij ) mn then kA = (cij ) mn where cij  ka ij .

Note:-

i. A and kA are of the same type

ii. -1A is written as –A. Thus A – B stands for A + (-1B).

iii. 0A, which has all the elements equal to 0 is denoted by 0.

iv. Thus for a matrices A and O of the same type,

A + 0 = 0 + A = A.

A + (-A) = 0 ;The matrix 0 is called a zero matrix.

3. Matrix Multiplication

Let A = ( a ij ) m p and B = (bij ) pn ( the number of columns of A is equal to the

number of rows of B). Then the product of A and B written AB, is defined by

2
 p
AB = (cij ) mn , where cij   aik bkj .
k 1

If the number of columns of A is equal to the number of rows of B, we say that A

and B are conformable for the product AB.

1 2  1 2 0
For example, if A = .   and B =   , then A and B are of type 22 and
3 1   1 4 3

23 respectively. Therefore, they are conformable for the product AB and

1  1  2  (1) 1  2  2  4 1  0  2  3    1 10 6 
AB =   =   .
 3  1  1  (1) 3  2  1  4 3  0  1  3   2 10 3 

However, A and B are not conformable for the product BA.

Note: -

i. If the number of columns of A is not equal to the number of rows of B, then the

product AB is not defined. If the matrices A and B both are same order square

matrix, then they are conformable for both products AB and BA.

1 2  2 2 14 12 
For example, if A =   and B =   then AB =   and
3 0 6 5 6 6

 8 12 
BA =   .
18 12 

ii. In general AB  BA.

iii. A.0 = 0.A = 0.

iv. If A is a square matrix of order n then Ap for any positive integer p is defined by A 

A  …A (p factors)

Basic Laws of Algebra of Matrices

If 𝐴, 𝐵, 𝐶 ∈ 𝑀𝑚,𝑛 (𝐹) and 𝛼, 𝛽 ∈ 𝐹 then

1. A + B = B + A Commutivity

2. (A + B) + C = A + (B + C) Associativity

3
 3. 𝛼 (𝐴 + 𝐵) = 𝛼𝐴 + 𝛼𝐵 distributivity of a Scalar

4. If B = 0 (a matrix of all zeros) then

A+B = A +0 = A

5. (𝛼 + 𝛽)𝐴 = 𝛼𝐴 + 𝛽𝐴

6. 𝛼(𝛽𝐴) = 𝛽(𝛼𝐴) = (𝛼𝛽)𝐴

7. 𝑂𝐴 = 0

More Definitions:

 a11 0 . . . 0 
 
 0 a 22 0 . . 0 
1. A square matrix of the form  0 0 a33 0 . 0  is called a diagonal
 
 . . . . . . 
 0 . 0 a nn 
 . .

matrix. The above diagonal matrix may also be written as diag(a11 , a22 ,, ann ) .

2. A diagonal matrix whose diagonal elements are equal is called a scalar matrix.

3. A scalar matrix of order n whose diagonal elements are equal to 1, denoted by I n

is called the unit matrix of order n or the identity matrix of order n. If the order

of the identity matrix is not important in a discussion, the identity matrix is

denoted by I . Thus a scalar matrix is of the form kI for some scalar k. If I and A

are matrices of same order, then AI  IA  A.

 a11 0 . . . 0 
 
 a 21 a 22 0 . . 0 
1. A square matrix of the form  a31 a32 a33 0 . 0  is called a lower
 
 . . . . . . 
a . a nn 
 n1 an2 . .

triangular matrix.

If all the diagonal elements of a lower triangular matrix are equal to 0, then it is

4
 called a strictly lower triangular matrix.

A = ( a ij ) mn .is lower triangular if and only if aij = 0, for j > i, and

A is strictly lower triangular if and only if aij = 0, for j  i.

Similarly upper triangular matrix is defined

2. The transpose of a given matrix A, written AT, is the matrix obtained by inter

changing each row with the corresponding column.

If A = ( a ij ) mn then AT = ( a ji ) nm .

Note: -

i. (AT)T = A and (kA)T = kAT.

ii. If A is a lower triangular matrix then AT is an upper triangular matrix, and vice

versa.

1 2 3  1 0 2
iii. If A =  0 1 1 then A =  2 1 3  .
  T

2 3 4   3 1 4 
   

iv. If A = AT (This can happen only if A is a square matrix), then A is called a

symmetric matrix. Thus if A = ( a ij ) nn is a symmetric matrix, then a ij  a ji for i,

j = 1, 2, …,n.

Example:-

1 1 −1 1 1 −1
𝐴=( 1 2 0) 𝐴𝑇 = ( 1 2 0)
−1 0 5 −1 0 5

v. If A = –AT, then A is called a skew symmetric matrix. In a skew symmetric matrix

the diagonal elements are equal to 0.

Example: -

0 1 −2 0 −1 2
𝐴 = (−1 0 3) 𝐴𝑇 = ( 1 0 −3)
2 −3 0 −2 3 0

5
 vi. If A is a square matrix and if a matrix B exists such that AB = BA = I, then A is

called a non-singular (or invertible) matrix, and B is called the inverse of A (and

A is the inverse of B). The inverse of A is denoted by A–1.

Thus B = A–1, and A = B–1. From these two equations we get (A-1)-1 = A.

vii. A square matrix, which is not non-singular is called a singular matrix, then

it is not possible to find a matrix B such that AB = I or BA = I.

1 –1
viii. If A is non-singular and if k  0 then (kA) –1 = A .
k

ix. The inverse of a matrix is unique.

For: If AB = BA = I and AC = CA = I for the non-singular matrix A, then

C = B.

If a square matrix has a row of zeros or a column of zeros, then A is singular.

Proposition

The product of two lower triangular matrices is a lower triangular matrix.

Home work

Prove the above theorem

Proposition

1. If A and B are matrices of the same type then (A + B)T = AT + BT

2. If A and B are matrices that are conformable for the product AB, then (AB)T =

BT.AT.

3. If A and B are non-singular matrices of the same order, then AB and BA are

non-singular and (AB) –1 = B–1A –1.

4. If A is a non-singular matrix then AT is non-singular, and (AT) –1 = (A–1) T.

5. If A is a non-singular lower triangular matrix then the diagonal elements of A are

non-zero and A –1 is also a lower triangular matrix.

6
 6. Any non-singular lower triangular matrix can be written as a product of a uni-

diagonal lower triangular matrix and a diagonal matrix.

Note:-

i. AB = O does not necessarily mean that A = O or B = O.

 2 1
  3 1
For example, if A =  8  4  and B =   , then AB = O.
 2 1  6 2
 

Similarly, AX = AY does not imply always that X = Y. However, if A is non-singular

then AX = AY implies X = Y.

Example:-

 1 1 1
 
i. Let A =  0 1 1 .Determine A2 and A3, and show that A3 = 3A2 - 3A + I, where I
 0 0 1
 

is the identity matrix of order 3. Deduce that (A - I)(A2 - 2A + I) = O.

Does this imply A2 - 2A + I = O? Justify your answer.

1 n f ( n) 
n  
ii. Prove that for any positive integer n, A has the form  0 1 n ,
0 0 1 


where f(n) is a function of n to be determined.

iii. Show that Ar – Ar -1 = (r - 1)A2 - (2r - 3)A + (r - 2)I for any rℤ+, and deduce that

1
An = [ n(n - 1)A2 - 2n(n - 2)A + (n - 1)(n - 2)I].
2

Solution:

1.

 1 1 1  1 1 1  1 2 3 
2     
A =  0 1 1  0 1 1 =  0 1 2  and
 0 0 1  0 0 1  0 0 1 
    

7
  1 1 1  1 2 3   1 1  2 1  2  3  1 3 6
      
3
A =  0 1 1  0 1 2  =  0 1 1 2 =  0 1 3
 0 0 1  0 0 1   0 1  0 0 1
    0  

 1 2 3   1 1 1  1 0 0   1 3 6 
       
Now 3A - 3A + I = 3 0 1 2   3 0 1 1   0 1 0    0 1 3 
2

 0 0 1   0 0 1  0 0 1   0 0 1 
       

A3 =3A2 - 3A + I. and hence, A3 - I3 = 3A(A - I)

i.e. (A - I)(A2 + A + I - 3A) = O, and hence (A - I) (A2 - 2A + I) = O .

0 0 1
 
Since (A - 2A + I) =  0 0 0  ------ (1), we have A2 - 2A + I  O.
2

 0 0 0
 

(A – I)(A2 –2A + I) = 0 does not imply that A2 –2A + I = 0

 1 3 1  2  3   1 1 1  1 4 1  2  3  4 
 4    
2. Now, A = 6  0 1 3   0 1 1 =  0 1 4 .
0 0
 1   0 0 1  0 0 1 


1 n f ( n) 
  n
n(n  1)
n
Let us assume that A is of the form  0 1 n  , where f(n) = r  2
,
0 0 1 
r 1


and prove by Mathematical Induction that actually An has this form.

When n = 1, the assertion is true.

Assume that the assertion is true for n = r > 1.

1 r f ( r )   1 1 1  1 r  1 1  r  f ( r ) 
= A .A =  0 1    
Now, A r+1 r
r   0 1 1 =  0 1 r 1  =
0 0
 1   0 0 1  0 0 1 


 1 r 1 f (r  1) 
 
0 1 r 1  ,
0 1 
 0

as 1 + r + f(r) = 1 + r + r(r + 1)/2 = (r + 1)(r + 2)/2 = f(r + 1).

8
i.e. The assertion is true for n = r +1.

Hence, by Mathematical Induction our assertion is true for all n 1.

0 1 f (r )  f (r  1)   0 1 r 
r r-1    
3. Now, A - A = 0 0 1  = 0 0 1
0 0   0 0 0
 0   

= A - I + (r - 1)(A2 -2A + I) (by (1))

= (r - 1)A2 - (2r -3)A + (r - 2)I.

Summing up for r = 1, 2, …, n, we have

n n n
 An - A0 =  (r  1) A 2  (2r  3) A   (r  2) I
r 1 r 1 r 1

1
An = [{n(n + 1) -2n}A2 - 2{n(n + 1) - 3n}A + {n(n + 1) - 4n +2}I] (as A0 = I)
2

1
= [n(n - 1)A2 - 2n(n - 2)A + (n - 1)(n - 2)I ].
2

Inverse of a Matrix and System of Linear Equations

Introduction

There are standard methods to find the inverse of a non-singular matrix of any order. In this

lesson we consider only special types matrices and find their inverses.

Matrix of order 2

a b  w x a b
Consider a non-singular matrix A =   . Suppose that A–1 =   . Then  
 c d   y z c d

 w x   1 0
  =   implies
 y z   0 1

aw + by = 1 ---------- (1)

ax + bz = 0 ------------(2)

cw + dy = 0 -----------(3)

9
 cx + dz = 1 ----------- (4)

From (1) and (3) we have (ad – bc)w = d -------- (5)

From (2) and (4) we have (ad – bc)x = -b --------- (6)

0
If ad = bc, then by (5) and (6), b = d = 0 and then the first column of A becomes   #
0

as A is nonsingular.

 ad  bc. By (5) and (6) w = d/(ad – bc) and x = – b/(ad – bc)

By (2 ) and (3) y = – c/(ad – bc) and z = a/(ad – bc).

1  d  b
A–1 =  .
ad  bc   c a 

Diagonal Matrix

Let D  diag(d1 , d 2 , , d n ) be non-singular. Clearly, di  0 for i = 1, 2, …, n.

Otherwise, D has a row of 0s #

It can be verified that DE = I, where E  diag(1 d1 ,1 d 2 , , 1 d n ) .

Since the inverse of a matrix is unique, D 1  diag (1 d1 ,1 d 2 ,  , 1 d n ) .

Lower Triangular Matrix

Let A = (aij) be a non-singular lower triangular matrix. Then by A-1 = (bij) is also a non-

singular lower triangular matrix. Hence aii and bii are nonzero for all i.

 a11 0 . . . 0   b11 0 . . . 0 
   
 a 21 a 22 0 . . 0   b21 b22 0 . . 0 
Let A =  a31 a32 a33 0 . 0  and A–1 = b b32 b33 0 . 0 .
   31 
 . . . . . .   . . . . . . 
a . a nn  b . bnn 
 n1 an2 . .  n1 bn 2 . .

Then AA–1 = I. ---------- (0)

10
Equating the first column of LHS of (0) to the first column of RHS, we have

a11 b11 = 1 -------- (1)

a21b11 + a22b21 = 0 ------- (2)

a31b11 + a32b21 +a33b31 = 0 ------- (3)

a41b11 + a42b21 + a43b31 + a44b41 = 0 ---------- (4)

…………………………………

a
k 1
b  0 --------- (n)
nk k 1

By (1) we get b11 = 1/a11.

Substituting the known value b11 in (2), we get b21. Substituting the known values a31, a32,

a33, b11and b21 in (3) we get b31. Similarly, from (4) we get b41, and so on.

ie. bi1, i = 1, 2, …., n are known.

Next, equate the second column of LHS to the second column of RHS of (0) to determine

all the values of bi2 and so on.

Finally all bij are determined and A–1 is determined.

Example

 2 0 0
 
Find the inverse of  1 3 0  .
  1 1 4
 

Solution

 2 0 0 a 0 0
   
Let A =  1 3 0  and A-1 = b d 0  . Then AA-1 = I implies
  1 1 4 c e f 
  

2a = 1 ------- (1) 3d = 1 --------- (4)

a + 3b = 0 ------- (2) d + 4e = 0 --------- (5)

11
 -a + b + 4c = 0 -------- (3) 4f = 1 --------- (6)

By (1), (2) and (3), we get a = 1/2, b = -1/6, c = 1/6, and by (4), (5) and (6) we get d = 1/3,

e = -1/12 and f = 1/4.

 6 0 0
1 
Hence, A =.   2 4 0  .
-1
12  
 2 1 1

Example

1
 1 0 0  1 0 0
   
Verify that  a 1 0   a 1 0
 c b 1  ab  c  b 1 
   

Solution

 1 0 0  1 0 0 1 0 0
    
 a 1 0   a 1 0    0 1 0 .
 c b 1   ab  c  b 1   0 0 1 
    

Example

 2 0 0
 
Express A =  1 3 0  as a product of a diagonal matrix and a uni-diagonal lower
  1 1 4
 

triangular matrix. Hence, find the inverse of A.

Solution

 2 0 0   2 0 0  1 0 0
    
 1 3 0    0 3 0  1 / 3 1 0 .
  1 1 4   0 0 4   1 / 4 1 / 4 1 
    

1 1 1
 2 0 0  1 0 0  2 0 0  1 0 0 1 / 2 0 0 
        
  1 3 0   1/ 3 1 0  0 3 0    1/ 3 1 0  0 1 / 3 0 
 1 1 4   1/ 4 1/ 4 1   0 0 4  1 / 6  1 / 4 1  0 0 1 / 4 
       

12
  1/ 2 0 0 
 
=   1/ 6 1/ 3 0 .
 1 / 6  1 / 12 1 / 4 
 

Using the fact that the transpose of an upper triangular matrix is a lower triangular matrix,

and the above result on lower triangular matrices we can find the inverse of any non-

singular upper triangular matrix.

Note : -

1. A matrix which can be written as a product of a lower triangular matrix and an

upper triangular matrix.

Let A be a non-singular matrix such that A = LU, where L is a lower triangular

matrix and U is an upper triangular matrix. Then A–1 = U–1L–1. Since U–1 and L–1

can be found by the method indicated in (3), we can find A–1.

2. Not every non-singular matrix can be expressed as a product of a lower triangular

matrix and an upper triangular matrix.

Example

2 1 1  1 0 0   2 1  1
    
If A =  6 5  2  and A = -1
 3 1 0   0 b c  , find a, b, c, d and hence find A .
4 8 4   2 a 1 0 0 d 
    

Solution

Equating the elements of A to the elements of the given product of matrices we have a = 3,

b = 3, c = 1 and d = 4.

1 1
 2 1  1  1 0 0  6  2 2   1 0 0   2 1  1
    1    
A =  0 3 1   3 1 0  =  0 4  1   3 1 0  =  0 b c  .
-1

0 0 4   2 3 1 12 
    0 0 1    4  3 1   0 0 d 

13
Example

1 p q
 
Let A =  0 t r  , where p, q, r, t are real numbers such that t2 = 1.
0 0 t 
 

 1  pt pr  qt 
 
-1
Show that A =  0 t  r .
0 t 
 0

1 6 0 1 0 0 1 6 0 
     
Let M =  0  1 1  , L =  0 1 0  and U =  0 b c  , where a, b, c, d are real numbers.
1 4 1 1 a 1 0 0 d 
     

Find a, b, c, d such that M = LU. Hence find M-1.

Solution

 1 p q   1  pt pr  qt  1 0 0
    
Since  0 t r   0 t  r =  0 1 0 ,
0 0 t  0 t  0 0 1
  0  

 1  pt pr  qt 
 
we see that  0 t  r  is the inverse of A.
0 t 
 0

1 6 0 1 0 0  1 6 0 
    
Since  0  1 1  =  0 1 0   0 b c  , we get b = -1, c = 1, 6 + ab = 4 and
1 4 1 1 a 1 0 0 d 
    

ac + d = 1. Thus, a = 2, b = -1, c = 1 and d = -1.

1 0 0 1 6 0 
   
 L =  0 1 0  and U =  0  1 1  .
1 2 1  0 0  1
   

1 0 0 1 6 6 
   
 0  1  1 .
-1
By the first part L =  0 1 0  and U-1 =
 1  2 1  0 0  1
   

Now M = LU implies that M-1 =U-1L-1, and hence

14
 1 6 6   1 0 0   5  6 6 

1
   
M =  0  1  1  0 1 0 =  1 1  1 .
 0 0  1   1  2 1   1 2  1
   

Solution of a system of n linear equations in n variables

A general system of n linear equation in n variables x1 , x2 , , xn can be written as

a11x1 + a12x2 + … + a1nxn = b1

a21x1 + a22x2 + … + a2nxn = b2

……………………………………

an1x1 + an2x2 + … + annxn = bn

In the matrix notation this system can be written as AX = B, where

 x1   b1 
   
 x2  b 
A  aij mn X=   and B =  2  .
. .
   
x  b 
 n  n

The matrix A is called the coefficient matrix of the given system of equations.

If A is a non-singular matrix then X = A–1B is the unique solution of the system.

If A is singular, then either the system has no solutions or has an infinite number of

solutions.

Consider the following examples.

Example: -

Solve x + 2y = 4

2x + 3y = 7.

Solution

15
  1 2  x  4
This can be written as AX = B, where A   , X   and B    .
 2 3  y 7

 3  2  3 2   4  2
A-1 = -   , and hence, X =     =   .
 2 1   2 1 7 1

Hence, x = 2, y = 1 is the unique solution of the given system of equations.

Example

Solve x + 2y = 2

2x + 4y = 3.

Solution

 1 2  x   2  1 2
      . The matrix   is singular as 1.4 - 2.2 = 0, and the second equation
 2 4  y   3   2 4

of the system reads as 2(x + 2y) = 3, and by the first equation 4 = 3.#

Hence the system has no solutions.

Example

Solve x + 2y = 2

2x + 4y = 4

Solution

x = a, y = (2 – a)/2 is a solution of the system for any real a.

 There are infinite number of solutions for the system.

Example

Solve the system of linear equations

x + 4y + 4z = 7

3x + 2y + 2z = 6

9x + 6y + 2z = 14

Solution

16
 1 4 4  x 7
     
This can be written as AX = B, where A =  3 2 2  , X =  y  and B =  6 .
9 6 2 z 14 
     

Now, if

1 0 0 𝑑 𝑒 𝑓
𝐴 = (𝑎 1 0) (0 𝑔 ℎ)
𝑏 𝑐 1 0 0 𝑖

1. 𝑑 = 1 ⟹ 𝑑 = 1

1. 𝑒 = 4 ⟹ 𝑒 = 4

1. 𝑓 = 4 ⟹ 𝑓 = 4

𝑎. 𝑑 = 3 ⟹ 𝑎 = 3

𝑎. 𝑒 + 𝑔 = 2 ⟹ 12 + 𝑔 = 2 ⟹ 𝑔 = −10

𝑎. 𝑓 + ℎ = 2 ⟹ 12 + ℎ = 2 ⟹ ℎ = −10

𝑏. 𝑑 = 9 ⟹ 𝑏 = 9

𝑏. 𝑒 + 𝑐. 𝑔 = 1 ⟹ 36 − 10𝑐 = 6 ⟹ 𝑐 = 3

𝑏. 𝑓 + 𝑐. ℎ + 𝑖 = 1 ⟹ 36 − 30 + 𝑖 = 2 ⟹ 𝑖 = −4

1 0 0 1 4 4 
   
Let L =  3 1 0  and U =  0  10  10  .
9 3 1 0 0  4 
  

Then 𝐴 = 𝐿𝑈 ⟹ 𝐴−1 = 𝑈 −1 𝐿−1

2
1 0
5
−1 1
𝑈 −1 = 0
10 4
−1
(0 0
4)

And

1 0 0
−1
𝐿 = (−3 1 0)
0 −3 1

17
 −1 2
0
5 5
3 −17 1
𝐴−1 =
10 20 4
3 −1
(0 4 4)

𝑋 = 𝐴−1 𝐵

−1 2
0 7
5 5
3 −17 1  
𝑋= 10 20 4  6 .
3 −1 14 
0  
( 4 4 )

 x 1
 
 y  = (1/2).
z 1
 

Thus, x = 1, y = ½, z = 1 is the unique solution of the given system.

Example

Express the following system of equations in the matrix form AX = B, where A is the

coefficient matrix, XT = (x y z) and BT = (3 2 1).

6x  3y  2z = 3

3x + 2y + z = 2

2x + y + z = 1

Find an upper triangular matrix U whose diagonal elements are equal to 1 and satisfies

UUT = A, and hence, solve the given system of equations.

Solution

1 a b  6  3  2
   
Let U =  0 1 c  . Then UU = A =   3 2
T
1  implies a = 1, b = 2 and c = 1.
0 0 1  2 1 1 
  

Now, AX = B  UUTX = B  X = (UUT )-1B =(UT)-1U-1B = (U-1)TU-1B

18
 1 0 0  1 1 1  3   6 
     
X = 1 1 0  0 1  1 2    7 .
1  1 1  0 0 1  1   6 
     

 x = 6, y = 7, z = 6 is the unique solution of the system

Note: -

In general, for a given matrix A, it is not possible to find an upper triangular matrix U such

that UUT = A, but it is possible in this example.

19
20