Inverse of a Matrix

If A and B are n x n matrices, and

if AB = BA = In, we say that B is
the inverse of A, and we write B = A–1.

• The concept of the inverse of a matrix

is analogous to that of the reciprocal
of a real number.
 Inverse of a Matrix—Definition

Let A be a square n x n matrix.

If there exists an n x n matrix A–1 with

the property that
AA–1 = A–1A = In

then we say that A–1 is the inverse of A.

E.g. 2—Verifying that a Matrix Is an Inverse

Verify that B is the inverse of A,

where:

2 1  3 −1
A=  and B= 
5 3   −5 2 

• We perform the matrix multiplications

to show that AB = I and BA = I.
E.g. 2—Verifying that a Matrix Is an Inverse

2 1  3 −1
5 3   −5 2 
  
 2  3 + 1( −5) 2( −1) + 1 2 
= 
5  3 + 3( −5) 5( −1) + 3  2 
1 0
= 
 0 1
 E.g. 2—Verifying that a Matrix Is an Inverse

 3 −1 2 1
 −5 2  5 3 
  
 3  2 + ( −1)5 3  1 + ( −1)3 
= 
( −5)2 + 2  5 ( −5)1 + 2  3 
1 0
= 
 0 1
Finding the Inverse
of a 2 x 2 Matrix
Finding the Inverse of a 2 x 2 Matrix

The following rule provides a simple way

for finding the inverse of a 2 x 2 matrix,
when it exists.

• For larger matrices, there’s a more general

procedure for finding inverses—which we
consider later in this section.
 Inverse of a 2 x 2 Matrix

If
a b 
A= 
c d 
then
−1 1  d −b 
A =  
ad − bc  −c a

• If ad – bc = 0, then A has no inverse.

 E.g. 3—Finding the Inverse of a 2 x 2 Matrix

Let A be the matrix

4 5
A= 
2 3

Find A–1 and verify that

AA–1 = A–1A = I2
 E.g. 3—Finding the Inverse of a 2 x 2 Matrix

Using the rule for the inverse of a 2 x 2

matrix, we get:
−1 1  3 −5 
A =  
4  3 − 5  2  −2 4 
1  3 −5   32 − 52 
=   = 
2  −2 4   −1 2 
• To verify that this is indeed the inverse of A,
we calculate AA–1 and A–1A.
 E.g. 3—Finding the Inverse of a 2 x 2 Matrix

−1  4 5   3
− 2
5
AA =   
2

 2 3   −1 2 
 4  32 + 5( −1) 4( − 52 ) + 5  2   1 0 
= 3 = 
 2  2 + 3( −1) 2( − 2 ) + 3  2  0 1
5

−1  3
− 2 
5
4 5
A A= 2
 2 3
 −1 2   
 32  4 + ( − 52 )2 32  5 + ( − 52 )3   1 0 
=  =  0 1
 ( −1)4 + 2  2 ( −1)5 + 2  3   
Determinant of a Matrix

The quantity ad – bc that appears in

the rule for calculating the inverse of
a 2 x 2 matrix is called the determinant
of the matrix.

• If the determinant is 0, the matrix does not

have an inverse (since we cannot divide by 0).
Finding the Inverse
of an n x n Matrix
 Finding the Inverse of an n x n Matrix

For 3 x 3 and larger square matrices,

the following technique provides
the most efficient way to calculate their
inverses.
 Finding the Inverse of an n x n Matrix

If A is an n x n matrix, we first construct the n

x 2n matrix that has the entries of A on the left
and of the identity matrix In on the right:

 a11 a12 a1n 1 0 0

a a a2 n 
 21 22 0 1 0
 
 
an1 an 2 ann 0 0 1
Finding the Inverse of an n x n Matrix

We then use the elementary row operations

on this new large matrix to change the left
side into the identity matrix.
• This means that we are changing the large matrix
to reduced row-echelon form.

 a11 a12 a1n 1 0 0

a a a2 n 
 21 22 0 1 0
 
 
an1 an 2 ann 0 0 1
Finding the Inverse of an n x n Matrix

The right side is transformed

automatically into A–1.

• We omit the proof of this fact.

 a11 a12 a1n 1 0 0

a a a2 n 
 21 22 0 1 0
 
 
an1 an 2 ann 0 0 1
 E.g. 4—Finding the Inverse of a 3 x 3 Matrix

Let A be the matrix

 1 −2 −4 

A =  2 −3 −6 
 −3 6 15 

(a) Find A–1.

(b) Verify that AA–1 = A–1A = I3.
E.g. 4—Inverse of a 3 x 3 Matrix Example (a)

We begin with the 3 x 6 matrix whose left half

is A and whose right half is the identity matrix.

 1 −2 −4 1 0 0 
 2 −3 −6 0 1 0 
 
 −3 6 15 0 0 1

• We then transform the left half of this new matrix

into the identity matrix—by performing the following
sequence of elementary row operations on the
entire new matrix.
E.g. 4—Inverse of a 3 x 3 Matrix Example (a)

 1 −2 −4 1 0 0 
R2 − 2R1 →R2
⎯⎯⎯⎯⎯ 
→ 0 1 
2 −2 1 0 
R3 + 3R1 →R3
0 0 3 3 0 1

 1 −2 −4 1 0 0 
1R
⎯⎯⎯
3 3 
→ 0 1 2 −2 1 0 
 
0 0 1 1 0 31 
E.g. 4—Inverse of a 3 x 3 Matrix Example (a)

 1 0 0 −3 2 0 
R1 + 2R2 →R1
⎯⎯⎯⎯⎯ 
→ 0 1 2 −2 1 0 
 
0 0 1 1 0 31 

 1 0 0 −3 2 0
R2 − 2R3 →R2
⎯⎯⎯⎯ ⎯ → 0 1 0 −4 1 − 32 
 
0 0 1 1 0 3
1

E.g. 4—Inverse of a 3 x 3 Matrix Example (a)

We have now transformed the left half

of this matrix into an identity matrix.

• This means we’ve put the entire matrix

in reduced row-echelon form.
 E.g. 4—Inverse of a 3 x 3 Matrix Example (a)

Note that, to do this in as systematic a fashion

as possible, we first changed the elements
below the main diagonal to zeros—just as we
would if we were using Gaussian elimination.

 1 −2 −4 1 0 0   1 −2 −4 1 0 0 
 2 −3 −6 0 1 0  0 1 2 −2 1 0 
   
 −3 6 15 0 0 1 0 0 3 3 0 1
E.g. 4—Inverse of a 3 x 3 Matrix Example (a)

Then, we changed each main diagonal

element to a 1 by multiplying by
the appropriate constant(s).

 1 −2 −4 1 0 0   1 −2 −4 1 0 0 
0 1 
2 −2 1 0  0 1 2 −2 1 0 
  
0 0 3 3 0 1 0 0 1 1 0 31 
E.g. 4—Inverse of a 3 x 3 Matrix Example (a)

Finally, we completed the process

by changing the remaining entries
on the left side to zeros.

 1 0 0 −3 2 0   1 0 0 −3 2 0
0 1 2 −2 1 0  0 1 0 −4 1 − 2 
   3

0 0 1 1 0 31  0 0 1 1 0 3
1

E.g. 4—Inverse of a 3 x 3 Matrix Example (a)

The right half is now A–1.

 −3 2 0
−1  2
A =  −4 1 − 3 
 1 0 1
3
E.g. 4—Inverse of a 3 x 3 Matrix Example (b)

We calculate AA–1 and A–1A, and verify that

both products give the identity matrix I3.

 1 −2 −4   −3 2 0   1 0 0 
 
AA−1 =  2 −3 −6   −4 1 − 32  = 0 1 0 
 −3 6 15   1 0 1
3
0 0 1

 −3 2 0   1 −2 −4   1 0 0 
 
A−1A =  −4 1 − 32   2 −3 −6  = 0 1 0 
 1 0 1
3
−3 6 15  0 0 1
E.g. 5—Matrix that Does Not Have an Inverse

Find the inverse of the matrix.

2 −3 −7 
1 2 7 
 
 1 1 4 
E.g. 5—Matrix that Does Not Have an Inverse

2 −3 −7 1 0 0  1 2 7 0 1 0
1 2 7 0 1 0 R  R2
2 −3 −7 1 0 0 
  1
 
 1 1 4 0 0 1  1 1 4 0 0 1
E.g. 5—Matrix that Does Not Have an Inverse

1 2 7 0 0
1
R2 − 2R1 →R2
⎯⎯⎯⎯⎯ 
→ 0 −7 −21 1 −2 0 
R3 −R1 →R3
0 −1 − 3 0 −1 1 

1 2 7 0 1 0
− 71 R2
⎯⎯⎯→ 0 1 3 − 71 2 
0
 7
0 −1 −3 0 −1 1
E.g. 5—Matrix that Does Not Have an Inverse

 0
2 3
1 0 1 7 7
R3 +R2 →R3
⎯⎯⎯⎯⎯ →  0 1 3 − 1
1 0 
R1 − 2R2 →R1  7 
0 0 0 − 7
1
− 7 1
5

• At this point, we would like to change the 0

in the (3, 3) position of this matrix to a 1,
without changing the zeros in the (3, 1) and (3, 2)
positions.
E.g. 5—Matrix that Does Not Have an Inverse

 0
2 3
1 0 1 7 7
R3 +R2 →R3
⎯⎯⎯⎯⎯ →  0 1 3 − 1
1 0 
R1 − 2R2 →R1  7 
0 0 0 − 7
1
− 7 1
5

• However, there is no way to accomplish this.

• No matter what multiple of rows 1 and/or 2 we add

to row 3, we can’t change the third zero in row 3
without changing the first or second zero as well.
E.g. 5—Matrix that Does Not Have an Inverse

 0
2 3
1 0 1 7 7
R3 +R2 →R3
⎯⎯⎯⎯⎯ →  0 1 3 − 1 2
0 
R1 − 2R2 →R1  7 7

0 0 0 − 71 − 5
7 1

• Thus, we cannot change the left half

to the identity matrix.

• So, the original matrix doesn’t have an inverse.

 Matrix that Does Not Have an Inverse

If we encounter a row of zeros on

the left when trying to find an inverse—as
in Example 5—then the original matrix
does not have an inverse.