Definition of matrices, entries, size of a matrix.

Special types of matrices.

Matrix operations and some laws.

Matrix multiplication and differences with real
number multiplication.
Different ways to represent matrix multiplication.
Representing a linear system by a matrix equation.
Matrix transpose and some laws.
Inverses of square matrices
Elementary matrices
If x is a real number, it is easy to solve

2 x  5,

since corresponding to '2', there is another

number ' 12 ' such that 2  12  1, allowing us to have

1
2  2 x  12  5
 1 x  52
 x  52
If X is a matrix, how can we solve

 2 1 3   2 1 1
 0 1 3  X   3 5 0  ?
   
 3 1 2 0 1 4 
   
 
Is there a matrix   such that
 
 
 
  2 1 3    2 1 1
  0 1 3  X    3 5 0 
     
  3 1 2    0 1 4 
     
  2 1 3    2 1 1
  0 1 3  X    3 5 0 
     
  3 1 2    0 1 4 
     

    2 1 1
 X   3 5 0 
    
    0 1 4 
    
 I (identity matrix)
  2 1 1
 X   3 5 0 
  
  0 1 4 
  
Let A be a square matrix of order n.

A is said to be an invertible matrix if there exists

another square matrix B of the same order such
that
AB  BA  I n

If such a B exists, it is called an inverse of A.

A is said to be singular if it has no inverse.

The definition of an invertible matrix is an
existential one.

"I tried very hard to find B but I could not..."

...does not mean B does not exists!

Could there be more than one such B?

  2 0 1  2 1 1
Is  2 1 2  the inverse of  6 3 2  ?
   
 3 1 0  5 2 2 
   

 2 0 1 2 1 1   
Check  2 1 2  6 3 2    
    
 3 1 0  5 2 2   
    

 2 1 1  2 0 1  
Check  6 3 2  2 1 2    
    
 5 2 2  3 1 0   
    
  2 1 1  3 1 1 
Find X if  6 3 2  X   0 0 2 .
   
 5 2 2  1 0 1
   

 2 0 1 2 1 1   2 0 1 3 1 1 
 2 1 2  6 3 2  X   2 1 2  0 0 2 
     
 3 1 0  5 2 2   3 1 0  1 0 1 
     

 7 2 3   7 2 3 
IX   8 2 2   X   8 2 2 .
   
 9 3 5   9 3 5 
   
 1 0 
Show that   is singular.
1 0 

1 0 
Suppose   is invertible, then there must be a
1 0 
a b 
square matrix of order 2, say   such that
c d

(By definition  a b 1 0   1 0 

of inverse)  c d 1 0    0 1 
    
 a b 1 0   1 0    a  b 0    1 0 
 c d 1 0    0 1  c  d 0 0 1
   
    

 a contradiction, by looking at the (2, 2)-entry

1 0 
Thus it is impossible for   to have an inverse,
1 0 
and so the matrix has to be singular.
  2 1 1
We knew A   6 3 2  is invertible because
 
 5 2 2 
 
 2 0 1
we were 'given' B   2 1 2  to 'test' whether
 
 3 1 0
 
AB and BA are both equal to I .
1 0 
We had to use contradiction to show that   has
1 0 
no inverse. What if the matrix was bigger and more
complicated?
Is there a more
effective way to
check whether a
matrix is invertible?

Yes...but before that...

If A is an invertible square matrix and

AB1  AB2
then B1  B2 .

If A is an invertible square matrix and

C1 A  C 2 A
then C1  C2 .
If A is not an invertible matrix, the cancellation
law may not hold:

1 0 
A  (known to be singular by previous example)
1 0 

 2 0 Check that AB1  AB2 , but B1  B2 .

B1   
 0 1 

 2 0
B2   
 1 3 
 If B and C are inverses of a square matrix A, then

B  C.

That is, if A is an invertible square matrix, then

it has one and only one inverse.
Inverses are unique!

Since inverses are unique, we will write A1 as the

inverse of A if A is invertible.
It turns out that to check whether a given square
matrix B is the inverse of A, we only need to check
either
AB  I OR BA  I

The reason will be explained in Section 2.4.

 a b
Let A    . Show that if ad  bc  0, then
c d
b
1  d
ad bc ad bc 
A is invertible and A    c a 
.
 ad bc ad bc 

Solution:
b
 a b   addbc ad bc 
Check    c a 
 c d  ad bc ad bc 
 a b
Let A    . Show that if ad  bc  0, then
c d
b
1  d
ad bc ad bc 
A is invertible and A    c a 
.
 ad bc ad bc 

Solution:
b
 addbc ad bc  a b 
Check   c a  
 ad bc ad bc   c d 
1) Let A be an invertible matrix (so A1 exists) and
c a non zero scalar. Then cA is invertible and

(cA)1  1c A1.
2) Let A be an invertible matrix (so A1 exists) then
AT is invertible and
Inverse of transpose
( AT )1  ( A1)T . equal transpose of
inverse
3) Let A be an invertible matrix (so A1 exists) then
A1 is invertible and

( A1)1  A.
4) Let A and B be two invertible matrices of the same
size (so A1 and B 1 exists) then AB is invertible and

( AB)1  B 1 A1. Inverse of product

equal product of
inverse*
4) Let A and B be two invertible matrices of the same
size (so A1 and B 1 exists) then AB is invertible and

( AB)1  B 1 A1.

As an extension to the above, if A1 , A2 ,..., Ak are

invertible matrices of the same size, then ( A1 A2 ... Ak )
is invertible and

( A1 A2 ... Ak )1  Ak 1... A2 1 A11.

Let A be a square matrix and n be a non negative
integer, then
 I if n  0;
An   AA... A if n  1.
 n times

If A is an invertible square matrix, then An is

invertible and we define

A n  ( A1)n  A1 A1... A1 and ( An )1  A n

n times
 1 2   3 2 
1
A  A  
 1 3   1 1 

1 2 1 2   
A 
2
   
1 3 1 3   

2 3 2  3 2   
A     
 1 1  1 1   
Let B be a m  n matrix.  
 
a1  1 0 0 ... 0  B  
 
 
 
 
 
a1B  1 0 0 ... 0   
 
 
 
 ... 
 first row of B
Let B be a m  n matrix.  
 
a2   0 1 0 ... 0  B  
 
 
 
 
 
a2 B   0 1 0 ... 0   
 
 
 
 ... 
 second row of B
Let B be a m  n matrix.  
 
ai   0 ... c ... 0  B  
 
 
ith position  
 
 
ai B   0 ... c ... 0    ith row
 
 
 
 c ... 
 c times ith row of B
Let B be a m  n matrix. What is the product equal to?
m columns

1 0 ... ... ... 0 

ith row 0 ... ... 0  
  
 c  
m   B 

1
 
rows
  

 
0 0 ... ... ... 1 
 
 
 c  row i 
 
 
 
Note that the product can be obtained by
performing elementary row operation cRi to B.
1 0 ... ... ... 0 
ith row 0 ... ... 0  
  
 c  
  B 

1
 
  

 
0 0 ... ... ... 1 
 
 
cRi  c  row i 
B  
 
 
Note that the matrix can also be obtained by
performing cRi on I m . cRi Im
1 0 ... ... ... 0 
ith row 0 ... ... 0  
  
 c  
  B 

1
 
  

 
0 0 ... ... ... 1 
 
 
cRi  c  row i 
B  
 
 
It seems like we can 'represent' performing the
elementary row operation cRi on B by
pre-multiplying a suitable matrix to B.

cRi
B C
cRi
Im E

Then we have EB  C.
Let's investigate the other two types of elementary
row operations!
 Let B be a m  n matrix. What is the product equal to?
ith col jth col
1 
 
  
ith row  0 1  
  B 
  
jth row  1 0  
  
 
 
 1   
 row j of B 
 
 row i of B 
 
 
Note that the product can be obtained by
performing elementary row operation Ri  R j to B.
1 
 
  
 0 1  
  B 
  
 1 0  
  
 
  
 1  
row j of B
Ri  R j  
B  row i of B 
 
 
Note that the matrix can also be obtained by
performing Ri  R j on I m .
Im
1  Ri  R j
 
  
 0 1  
  B 
  
 1 0  
  
 
  
 1  
row j of B
Ri  R j  
B  row i of B 
 
 
It seems like we can 'represent' performing the
elementary row operation Ri  R j on B by
pre-multiplying a suitable matrix to B.

Ri  R j
B C
Ri  R j
Im E

Then we have EB  C.
Let's investigate the last type of elementary
row operation!
 Let B be a m  n matrix. What is the product equal to?
ith col jth col
1 
 
  
ith row  1  
  B 
  
jth row  k 1  
  
 
 
 1   
(i  j)  
(row j of B)  k (row i of B)  

  jth row
 
 
 Let B be a m  n matrix. What is the product equal to?
jth col ith col
1 
 
  
jth row 1 k  
  B 
  
ith row  1  
  
 
 
 1   
(i  j)   jth row
 
(row j of B)  k (row i of B)  
 
 
 Note that the product can be obtained by
performing elementary row operation R j  kRi to B.
1 
 
  
 1 k  
  B 
  
 1  
  
 
  
(i  j)  
1
  jth row
R j  kRi  
B  
 
 
 Note that the product can be obtained by
performing elementary row operation R j  kRi to B.
1 
 
  
 1  
  B 
  
 k 1  
  
 
  
(i  j)  
1
 
R j  kRi  
B   jth row
 
 
Note that the matrix can also be obtained by
performing R j  kRi on I m .
Im
1 
  R j  kRi
  
 1 k  
  B 
  
 1  
  
 
  
 1 
  jth row
R j  kRi  
B  
 
 
Note that the matrix can also be obtained by
performing R j  kRi on I m .
Im
1 
  R j  kRi
  
 1  
  B 
  
 k 1  
  
 
  
 1 
 
R j  kRi  
B   jth row
 
 
It seems like we can 'represent' performing the
elementary row operation R j  kRi on B by
pre-multiplying a suitable matrix to B.

R j  kRi
B C
R j  kRi
Im E

Then we have EB  C.
 1 0 ... ... ... 0 
0 ... ... 0  E is invertible and E 1
 
 c  is shown below
E  
1 (check it!)
 
 
 
0 0 ... ... ... 1  1 0 ... ... ... 0 
0 ... ... 0 
 
1  1

E  c

1
 
 
 
0 0 ... ... ... 1 
 1 
 
  E is invertible and E 1
 0 1 
E   is shown below
  (check it!)
 1 0 
  1 
   
 
 1  
 0 1 
1
E   
 
 1 0 
 
 
 
 1 
 1 
 
  E is invertible and E 1
 1 k 
is shown below
E  
  (check it!)
 1 
  1
   
  
 1   
 1 k 
E 1   
 
 1 
 
 
 
 1 
 1 
 
  E is invertible and E 1
 1 
is shown below
E  
  (check it!)
 k 1 
  1
   
  
 1   
 1 
E 1   
 
 k 1 
 
 
 
 1 
Let A be a m  n matrix. For each of the three types
of elementary row operations we can perform on A :

B1 E1
cRi cRi
Ri  R j Ri  R j
A B2 Im E2
R j  kRi R j  kRi
B3 E3
there are matrices Ei (i  1, 2, 3) such that
Ei A  Bi for i  1, 2,3.
Furthermore, each Ei is an invertible matrix.
A square matrix is an elementary matrix if it can be
obtained from an identity matrix by performing
a single elementary row operation.
Note that all elementary matrices are invertible and
their inverses are themselves elementary matrices.
E1 1
Ri E11
cRi c

Ri  R j Ri  R j
Im E2 Im E2 1
R j  kRi R j  kRi
E3 E31
Note that if an elementary matrix E represents a
single elementary row operation X , E 1 represents
the elementary row operation that 'un-do' X .
 1 0 1 1 
Let A   0 2 3 1 . Find a sequence of elementary
 
 1 0 3 4 
 
matrices E1 , E2 ,..., Ek such that Ek Ek1... E2 E1 A is the
reduced row-echelon form of A.

Write down the inverses of each of the elementary

matrices and describe which elementary row
operation these inverses represent.
 Lecture 06:
Elementary matrices (cont’d)
Determinants (till Example 2.5.13)