Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Linear Algebra through Matrices; Lecture -3A, 31 March

Department of Mathematics and Statistics Indian Institute of Technology - Kanpur

Inverse of a Matrix

1 Inverse of a Matrix
Having understood the matrix multiplication, we now define the inverse of a square matrix. At
this stage, we will just look at the definition and basic properties. We will come across the study
of invertible matrices again and again.

Definition 1.1. Let A ∈ Mm,n (C). Then

1. B ∈ Mn,m (C) is said to be a left inverse of A if BA = In .

2. C ∈ Mn,m (C) is called a right inverse of A if AC = Im .


" # " #
1 2
Example 1.2. Let A = [1 2]. Define B = or B = then AB = 1 but you cannot
0 −1/2
get BA = I2 . For if, there exists a B such that BA = I2 then
" # " # " # " #!
−2 −2 −2 −2
= I2 = (BA) = B [1 2] = B · 0 = 02×1 ,
1 1 1 1

a contradiction.

3. A is invertible (has an inverse) if there exists B ∈ Mn (C) such that AB = BA = In . As


we need the same matrix B, A must be a square matrix.

Lemma 1.3. Let A ∈ Mn (C). If there exist B, C ∈ Mn (C) such that AB = In and CA = In then
B = C, i.e., If A has a left inverse and a right inverse then they are equal.

Proof. Note that C = CIn = C(AB) = (CA)B = In B = B.

Remark 1.4. Lemma 1.3 implies that whenever A is invertible, the inverse is unique. Thus, we
denote the inverse of A by A−1 . That is, AA−1 = A−1 A = I.

Taking transpose/conjugate transpose on both sides gives

Theorem 1.5. Let A and B be two invertible matrices. Then,

1. (A−1 )−1 = A.

2. (AB)−1 = B −1 A−1 .

3. (A∗ )−1 = (A−1 )∗ and (AT )−1 = (A−1 )T .


2

" # " #
1 2 2 3
Example 1.6. 1. The matrices A = and D = do not have inverse!!
0 0 4 6
Ans: Later, we will see that each of them have ‘determinant’ 0, so they are not invertible. But,
let us us look at important alternative methods.
Method 1: Note that A[2, :] = 0. Hence, for each x, A[2, :]x = [0, 0]T 6= [0, 1]T = the second
row of the identity matrix. A similar argument holds for B[:, 2].
Thus, an invertible matrix can not have a zero row or a zero column.
" #
−2
Method 2: Verify that A = 02×1 . Now, get a contradiction by showing that if A has an
1
inverse, say B then BA = I2 and
" # " # " # " #
−2 −2 −2 −2
= I2 = (BA) = B(A ) = B · 0 = 0.
1 1 1 1
" #
−3
Similarly, D = 02×1 . Then, we can use similar ideas as above to show that D can not be
2
invertible. Thus, an invertible matrix can not have two columns which are multiple
of each other.
 
1 2 3
 
2. C = 2 −1 1 does not have inverse.

3 2 5
Ans: Note that C[:, 3] = C[:, 1] + C[:, 2]. So, if there exists a B such that CB = BC = I3
then
 
0
 
0 = e3 = (BC)[:, 3] = BC[:, 3] = B(C[:, 1] + C[:, 2]) = BC[:, 1] + BC[:, 2]
 
1
 
1
 
= (BC)[:, 1] + (BC)[:, 2] = e1 + e2 = 1

,
0

a contradiction. Thus, an invertible matrix cannot have a row which is sum of other
rows. A similar statement holds for columns.
We summarize them in the next example.
" #
a b
3. Let A = .
c d

(a) If ad − bc = 0 then prove that


i. either A[1, :] = 0T ;
3

ii. or A[:, 1] = 0;
iii. or A[2, :] = 0T ;
iv. or A[:, 2] = 0;
v. or αA[2, :] + βA[1, :] = 0T for some α, β ∈ C.
# 2] = 0 for some α, β ∈ C, or equivalently, there exists a non-zero
vi. or αA[:, 1] +" βA[:,
α
vector x = such that Ax = 02×1 .
β
Hence, using matrix product, prove that A is not invertible. The above ideas can be
generalized for higher order square matrices.
" #
d −b
(b) If ad − bc 6= 0. Then, verify that A−1 = 1
ad−bc .
−c a
" # " #
2 3 1 7 −3
(c) For example, the inverse of equals 2 .
4 7 −4 2

You might also like