Some facts of the product of two matrices:

In real number system i) if 𝑎𝑏 = 𝑎𝑐 and 𝑎 ≠ 0 then 𝑏 = 𝑐

ii) if 𝑎𝑏 = 0 then either 𝑎 = 0 or 𝑏 = 0
The above results are not in general true in matrix arithmetic.

a) Failure of the cancellation law:

0 1 1 1 2 5
Consider 𝐴 = ( ), 𝐵=( ) and 𝐶 = ( )
0 2 3 4 3 4
3 4
We see that 𝐴𝐵 = 𝐴𝐶 = ( ) but 𝐵 ≠ 𝐶.
6 8
That means we cannot cancel 𝐴 in the equality 𝐴𝐵 = 𝐴𝐶. So the normal cancellation law
fails in case of matrix.

b) A zero product with non-zero factors:

0 1 3 7
Consider 𝐴=( ) and 𝐷 = ( ) none of which, is a zero matrix but their product
0 2 0 0
0 1 3 7 0 0
is a zero matrix because we see that 𝐴𝐷 = ( )( )=( )
0 2 0 0 0 0

c) Property of Unit matrix/ Identity matrix

1 0
We know that the Identity matrix, 𝐼2 = ( ) is a unit matrix of order 2.
0 1
1 0 0
𝐼3 = (0 1 0) is a unit matrix of order 3
0 0 1
1 0 0 0
𝐼4 = ( 0 1 0 0 ) is a unit matrix of order 4 and so on.
0 0 1 0
0 0 0 1

If 𝐴 is an 𝑚 × 𝑛 matrix, it can be shown that 𝑨𝒎×𝒏 𝑰𝒏 = 𝑨𝒎×𝒏 and 𝑰𝒎 𝑨𝒎×𝒏 = 𝑨𝒎×𝒏

So the identity matrix plays the same role in matrix algebra as the number ‘1’ plays in real
number system.
 Inverse of a square matrix

Definition: If 𝐴 is a square matrix and if a matrix 𝐵 of the same size can be found such that
𝐴𝐵 = 𝐵𝐴 = 𝐼 then 𝐴 is said to invertible (or non-singular) and 𝐵 is called the inverse of 𝐴 .
If no such matrix 𝐵 can be found then 𝐴 is not invertible and is said to be a singular matrix.
The relationship 𝐴𝐵 = 𝐵𝐴 = 𝐼 is not changed by interchanging 𝐴 and 𝐵, so if 𝐴 is invertible
and 𝐵 is the inverse of 𝐴, then it is also true that 𝐵 is invertible and 𝐴 is the inverse of 𝐵. Thus
when 𝐴𝐵 = 𝐵𝐴 = 𝐼, we say that 𝐴 and 𝐵 are inverses of each other.

Theorem: If 𝐵 and 𝐶 are both inverses of the matrix 𝐴, then 𝐵 = 𝐶, that is the inverse of a
square matrix is unique if it exists.

𝑎 𝑏
Theorem: The matrix 𝐴 = ( ) is invertible if and only if, 𝑎𝑑 − 𝑏𝑐 ≠ 0, in which case, the
𝑐 𝑑
𝟏 𝒅 −𝒃 𝟏 𝒅 −𝒃
inverse of 𝐴 is given by the formula 𝑨−𝟏 = ( ) = ( ). The
|𝑨| −𝒄 𝒂 𝒂𝒅−𝒃𝒄 −𝒄 𝒂
quantity 𝑎𝑑 − 𝑏𝑐 is called the determinant of the matrix 𝐴.
1
Now we confirm the validity of the formula by showing that 𝐴𝐴−1 = 𝐴−1 𝐴 = 𝐼. [𝐴−1 ≠ ]
𝐴

1 𝑎 𝑏 𝑑 −𝑏 1 𝑎𝑑 − 𝑏𝑐 −𝑎𝑏 + 𝑎𝑏
𝐴𝐴−1 = ( )( ) = 𝑎𝑑−𝑏𝑐 ( )
𝑎𝑑−𝑏𝑐 𝑐 𝑑 −𝑐 𝑎 𝑐𝑑 − 𝑐𝑑 −𝑏𝑐 + 𝑎𝑑
1 𝑎𝑑 − 𝑏𝑐 0 𝑎𝑑−𝑏𝑐 1 0 1 0
= ( )= ( )=( )=𝐼
𝑎𝑑−𝑏𝑐 0 𝑎𝑑 − 𝑏𝑐 𝑎𝑑−𝑏𝑐 0 1 0 1
1 𝑑 −𝑏 𝑎 𝑏 1 𝑎𝑑 − 𝑏𝑐 𝑏𝑑 − 𝑏𝑑
Similarly, 𝐴−1 𝐴 = ( )( ) = 𝑎𝑑−𝑏𝑐 ( )
𝑎𝑑−𝑏𝑐 −𝑐 𝑎 𝑐 𝑑 −𝑎𝑐 + 𝑎𝑐 −𝑏𝑐 + 𝑎𝑑
1 𝑎𝑑 − 𝑏𝑐 0 𝑎𝑑−𝑏𝑐 1 0 1 0
= ( )= ( )=( )=𝐼
𝑎𝑑−𝑏𝑐 0 𝑎𝑑 − 𝑏𝑐 𝑎𝑑−𝑏𝑐 0 1 0 1

So 𝐴𝐴−1 = 𝐴−1 𝐴 = 𝐼 is verified.

𝑎 𝑏 𝑘𝑎 𝑘𝑏 𝑘𝑎 𝑘𝑏
[ Note: We know that 𝑘. 𝐴 = 𝑘 ( )=( ) . So if we have a matrix 𝐵 = ( ),
𝑐 𝑑 𝑘𝑐 𝑘𝑑 𝑘𝑐 𝑘𝑑
𝑎 𝑏
we can take the common factor ‘𝑘’ outside the matrix and have 𝐵 = 𝑘 ( ).
𝑐 𝑑
4 0 −8 2 0 −4
5 0 1 0
Example: 𝐴 = ( ) = 5( ) 𝐵 = (2 −6 0 ) = 2 (1 −3 0 )
10 15 2 3
0 −16 12 0 −8 6

−6 3 0 1 − 1⁄2 0
And 𝐶 = ( ) = −6 ( ) etc. ]
2 1 −9 − 1⁄3 − 1⁄6 3⁄2
Theorem: If 𝐴 and 𝐵 are invertible matrices of same sizes then 𝐴𝐵 is also invertible and
(𝑨𝑩)−𝟏 = 𝑩−𝟏 𝑨−𝟏 .

Proof: Since 𝐴 and 𝐵 are invertible 𝐴−1 and 𝐵 −1 exist and 𝐴𝐴−1 = 𝐴−1 𝐴 = 𝐼 and
𝐵𝐵 −1 = 𝐵 −1 𝐵 = 𝐼 are satisfied.

So (𝐴𝐵)(𝐵 −1 𝐴−1 ) = 𝐴(𝐵𝐵 −1 )𝐴−1 = 𝐴𝐼𝐴−1 = 𝐴𝐴−1 = 𝐼 and

(𝐵 −1 𝐴−1 )(𝐴𝐵) = 𝐵 −1 (𝐴−1 𝐴)𝐵 = 𝐵 −1 𝐼𝐵 = 𝐵 −1 𝐵 = 𝐼

So (𝐴𝐵)(𝐵 −1 𝐴−1 ) = (𝐵 −1 𝐴−1 )(𝐴𝐵) = 𝐼, showing that 𝐵 −1 𝐴−1 is the inverse of 𝐴𝐵

That is (𝐴𝐵)−1 = 𝐵 −1 𝐴−1 . Similarly it can be shown that

(𝐴𝐵𝐶)−1 = 𝐶 −1 𝐵 −1𝐴−1 and (𝐴𝐵𝐶𝐷)−1 = 𝐷 −1 𝐶 −1 𝐵−1 𝐴−1 and so on.

A product of any number of invertible matrices is invertible and the inverse of the product is the
product of the inverses in the reverse order.

Powers of a square matrix: If 𝐴 is a square matrix, we define the non-negative integer powers
of 𝐴 to be 𝐴0 = 𝐼, 𝐴𝑛 = 𝐴. 𝐴. 𝐴 … … … … . . 𝐴 (𝑛 factors) and if 𝐴 is invertible, we define the
negative integer powers of 𝐴 to be 𝐴−𝑛 = 𝐴−1 . 𝐴−1 … … … 𝐴−1(𝑛 factors). These definitions are
parallel to those for real numbers. So the usual laws of non-negative exponents hold; for
example: 𝐴𝑟 𝐴𝑠 = 𝐴𝑟+𝑠 and (𝐴𝑟 )𝑠 = 𝐴𝑟𝑠 .

Theorem: If 𝐴 is invertible and 𝑛 is a non-negative integer, then

i) 𝐴−1 is invertible and (𝐴−1 )−1 = 𝐴
ii) 𝐴𝑛 is invertible and (𝐴𝑛 )−1 = 𝐴−𝑛 = (𝐴−1 )𝑛
1
iii) 𝑘𝐴 is invertible for any non-zero scalar 𝑘 and (𝑘𝐴)−1 = 𝐴−1 𝑘 −1 = 𝑘 𝐴−1.

Theorem: If 𝐴 is an invertible matrix, then 𝐴𝑇 is also invertible and (𝑨𝑻 )−𝟏 = (𝑨−𝟏 )𝑻

Proof: Since 𝐴 is an invertible matrix, we have 𝐴𝐴−1 = 𝐴−1 𝐴 = 𝐼 .

Now using the property: (𝐴𝐵)𝑇 = 𝐵 𝑇 𝐴𝑇 , that is using 𝐵 𝑇 𝐴𝑇 = (𝐴𝐵)𝑇 , we may write

𝐴𝑇 (𝐴−1 )𝑇 = (𝐴−1 𝐴)𝑇 = (𝐼)𝑇 = 𝐼 and

(𝐴−1 )𝑇 𝐴𝑇 = (𝐴𝐴−1 )𝑇 = (𝐼)𝑇 = 𝐼

So 𝐴𝑇 (𝐴−1 )𝑇 = (𝐴−1 )𝑇 𝐴𝑇 = 𝐼, which, shows that (𝐴−1 )𝑇 is the inverse of 𝐴𝑇 ,

That is (𝐴𝑇 )−1 = (𝐴−1 )𝑇

Matrix Polynomials:
Suppose that 𝐴 is a square matrix of size 𝑛 × 𝑛. Then if
𝑃(𝑥) = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 + … … … + 𝑎𝑚 𝑥 𝑚 (1)
is any polynomial of degree 𝑚, then we can define the 𝑛 × 𝑛 matrix 𝑃(𝐴) to be
𝑃(𝐴) = 𝑎0 𝐼 + 𝑎1 𝐴 + 𝑎2 𝐴2 + … … … + 𝑎𝑚 𝐴𝑚 (2)
where, 𝐼 is the 𝑛 × 𝑛 identity matrix. The expression of the form in equation (2) is called
a matrix polynomial in 𝐴.

Q: Compute the inverses of the following matrices:

6 4 𝑐𝑜𝑠𝜃 𝑠𝑖𝑛𝜃
i) 𝐴 = ( ) ii) 𝐵 = ( )
−2 −1 −𝑠𝑖𝑛𝜃 𝑐𝑜𝑠𝜃

6 4 −1 −4 ⁄
) = (− 1 2 −2)
1
i) |𝐴| = | | = −6 + 8 = 2, So 𝐴−1 = 2 (
−2 −1 2 6 1 3
𝑐𝑜𝑠𝜃 𝑠𝑖𝑛𝜃 𝑐𝑜𝑠𝜃 −𝑠𝑖𝑛𝜃
ii) |𝐵| = | | = 𝑐𝑜𝑠 2 𝜃 + 𝑠𝑖𝑛 2 𝜃 = 1, So 𝐵 −1 = ( )
−𝑠𝑖𝑛𝜃 𝑐𝑜𝑠𝜃 𝑠𝑖𝑛𝜃 𝑐𝑜𝑠𝜃

3 1
Q: Use the matrix 𝐴 = ( ) to verify that (𝐴𝑇 )−1 = (𝐴−1 )𝑇
5 2

3 5 3 5 1 2 −5 2 −5
𝐴𝑇 = ( ) |𝐴𝑇 | = | | = 1 So (𝐴𝑇 )−1 = 𝑇 ( )=( )
1 2 1 2 |𝐴 | −1 3 −1 3

|𝐴| = |3 1 1 2 −1 2 −1
|=6−5= 1 So 𝐴−1 = |𝐴| ( )=( )
5 2 −5 3 −5 3

2 −5
and so (𝐴−1 )𝑇 = ( ), which, shows that (𝐴𝑇 )−1 = (𝐴−1 )𝑇 .
−1 3

−3 −1 −1 2
Q: Find the matrix 𝐴 if i) (5𝐴𝑇 )−1 = ( ) ii) (𝐼 + 2𝐴)−1 = ( )
5 2 4 5

i) Taking inverses on both sides we get [Since (𝐴−1 )−1 = 𝐴]

−1
[(5𝐴𝑇 )−1 ]−1 = (−3 −1) −3 −1
; Here | | = −6 + 5 = −1
5 2 5 2

1 2 1 −2 −1 1 −2 −1 −2⁄5 −1⁄5
⇒ 5𝐴𝑇 = ( )=( ) ⇒ 𝐴𝑇 = ( )=( )
−1 − 5 −3 5 3 5 5 3 1 3⁄5

−2⁄5 1
So 𝐴 = ( ) [because (𝐴𝑇 )𝑇 = 𝐴]
−1⁄5 3⁄5
ii) Taking inverses on both sides we get

2 −1
[(𝐼 + 2𝐴)−1 ]−1 = (−1 −1 2
) ; Here | | = −5 − 8 = −13
4 5 4 5

1 5 −2 − 5⁄13 2⁄13
⇒ 𝐼 + 2𝐴 = −13 ( )=( )
−4 −1 4⁄13 1⁄13

− 5⁄13 2⁄13 1 0 −18⁄13 2⁄13

⇒ 2𝐴 = ( )−( )=( )
4⁄13 ⁄
1 13 0 1 4⁄13 − 12⁄13

−9⁄13 1⁄13
So 𝐴 = ( )
2⁄13 − 6⁄13

2 0
Q: If 𝐴 is the matrix ( ) Compute i) 𝐴3 ii) 𝐴−3 iii) 𝐴2 − 2𝐴 + 𝐼
4 1
2 0 2 0 2 0 4 0 2 0 8 0
i) 𝐴3 = 𝐴. 𝐴. 𝐴 = ( )( )( )=( )( )=( )
4 1 4 1 4 1 12 1 4 1 28 1

8 0 −1 1 1 0 1 1 0 1⁄8 0
ii) 𝐴−3 = (𝐴3 )−1 = ( ) = 3 ( )=8( )=( )
28 1 |𝐴 | −28 8 −28 8 − 7⁄2 1

2 0 2 0 2 0 1 0
iii) 𝐴2 − 2𝐴 + 𝐼 = ( )( )− 2( )+( )
4 1 4 1 4 1 0 1

4 0 4 0 1 0 4−4+1 0−0+0
=( )−( )+( )=( )
12 1 8 2 0 1 12 − 8 + 0 1−2+1
1 0
=( )
4 0

Q: Simplify the following expressions:

i) (𝐴𝐵)−1 (𝐴𝐶 −1)(𝐷−1 𝐶 −1 )−1𝐷−1 ii) (𝐴𝐶 −1 )−1 (𝐴𝐶 −1 )(𝐴𝐶 −1)−1 𝐴𝐷−1

i) (𝐴𝐵)−1 (𝐴𝐶 −1)(𝐷−1 𝐶 −1 )−1𝐷−1

= 𝐵 −1 𝐴−1 (𝐴𝐶 −1)(𝐶 −1 )−1(𝐷−1 )−1 𝐷−1 = 𝐵 −1 (𝐴−1 𝐴)(𝐶 −1 𝐶)(𝐷𝐷 −1 )
= 𝐵 −1 (𝐼)(𝐼)(𝐼) = 𝐵 −1

ii) (𝐴𝐶 −1 )−1 (𝐴𝐶 −1)(𝐴𝐶 −1 )−1 𝐴𝐷 −1

= 𝐼(𝐶 −1 )−1 𝐴−1 𝐴𝐷 −1 = 𝐼𝐶𝐼𝐷 −1 = 𝐶𝐷−1

Q: i) Show that if 𝐴, 𝐵 and 𝐴 + 𝐵 are invertible matrices of the same size, then
𝐴(𝐴−1 + 𝐵 −1 )𝐵(𝐴 + 𝐵)−1 = 𝐼

ii) Using the above result show that 𝐴−1 + 𝐵 −1 is invertible and find its inverse.