MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

CHAPTER 6
EIGENVALUES AND EIGENVECTORS
6.1 Eigenvalues and Eigenvectors

Definition Let A be an n  n matrix. A nonzero vector x is an eigenvector of A if for

~

some scalar  , A x =  x . The scalar  is called an eigenvalue of A.

~ ~

Definition det (I − A) = 0  characteristic equation of A

Definition Nullspace of (I − A)  solution space of (I − A) x = 0

~ ~
 eigenspace of A corresponding to the eigenvalue

Computation of Eigenvalues, Eigenvectors and Finding a Basis for the Eigenspace of A

Step 1 Find eigenvalues of A by solving det (I − A) = 0

Step 2 For each eigenvalue, find the corresponding eigenvector
Step 3 Obtain a basis for the eigenspace of A

Example 6.1.1

Find the characteristics equation, eigenvalues, the corresponding eigenvectors, and bases
0 0 − 2
  3
for the eigenspace for the matrix A =  1 2 1  . Verify that tr( A) =  i and
1 0 3 
i =1

3
det( A) =  i .
i =1

Solution:

det ( I3 − A) = 0

  1 0 0   0 0 −2  
 
 det    0 1 0  −  1 2 1   = 0
 0 0 1 1 0 3 
    
 0 2 
 
 det  − 1  − 2 − 1  = 0
 −1  − 3 
 0

School of Mathematical Sciences (SMS) Page 6-1 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

 2
 ( − 2) =0 (cofactor expansion along the 2nd column)
−1  − 3

 ( − 2)( ( − 3) + 2) = 0
 ( − 2)(2 − 3 + 2) = 0
 ( − 2)( − 1)( − 2) = 0
 ( − 1)( − 2) 2 = 0   = 1,  = 2,  = 2 (repeated roots)

tr( A) = 0 + 2 + 3 = 5
3


i =1
i = 1 + 2 + 3 = 1 + 2 + 2 = 5
3
 tr( A) =  i = 5 (verified)
i =1

1 2
det( A) = (−2) = (−2)(0 − 2) = 4
1 0
3


i =1
i = 1  2  3 = 1 2  2 = 4
3
 det( A) =  i = 4 (verified)
i =1

 0 2  x1   0 
    
To find the eigenvectors, consider (I − A) x = 0 , ie  − 1  − 2 − 1  x2  =  0 
 −1  − 3  x3   0 
~ ~
 0

For the eigenvalue  = 1 , we have

1 0 2  x1   0   1 0 2 0 1 0 2 0 1 0 2 0
          
 − 1 − 1 − 1  x 2  =  0  →  − 1 − 1 − 1 0  →  0 − 1 1 0  →  0 1 − 1 0 
 − 1 0 − 2  x   0   − 1 0 − 2 0   0 0 0 0  0 0 0 0
  3         

Now, x1 + 2 x3 = 0
x2 − x3 = 0

Let x3 = s . Then, x2 = s and x1 = −2 s

 x1   −2 s   −2 
     
The solution space x =  x2  =  s  = s  1  , s 
x   s   1
~

 3    

School of Mathematical Sciences (SMS) Page 6-2 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

 x1   − 2 
   
 the solution space of (I − A) x = 0 is of the form x =  x2  = s 1 
x   1 
~ ~ ~

 3  
 − 2
 
  1  is a basis for the nullspace of (I − A) corresponding to  = 1
 1 
 

 − 2
 
  1  is a basis for the eigenspace of A corresponding to  = 1
 1 
 

For the eigenvalue  = 2 , we have

 2 0 2  x1   0   2 0 2 0   1 0 1 0   1 0 1 0 
          
 − 1 0 − 1 x2  =  0  →  − 1 0 − 1 0  →  − 1 0 − 1 0  →  0 0 0 0 
 − 1 0 − 1 x   0   − 1 0 − 1 0   − 1 0 − 1 0   0 0 0 0 
  3         

Now, x1 + x3 = 0

Let x3 = s and x2 = t . Then, x1 = − s

 x1   − s   −1  0 
       
The solution space x =  x2  =  t  = s  0  + t  1  , s, t 
x   s   1   0
~

 3      
 x1   −1  0 
 
 the solution space of (I − A) x = 0 is of the form x =  x2  = s  0  + t  1  , s, t 
x   1   0
~ ~ ~

 3    
 − 1  0 
   
  0 ,  1  is a basis for the nullspace of (I − A) corresponding to  = 2
 1   0 
   

 − 1  0 
   
  0 ,  1  is a basis for the eigenspace of A corresponding to  = 2
 1   0 
   

School of Mathematical Sciences (SMS) Page 6-3 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Example 6.1.2

Find the characteristics equation, eigenvalues, the corresponding eigenvectors, and bases

3 0 1
 
for the eigenspace for the matrix A =  0 2 4  .
0 0 1
 

Solution:

det (I − A) = 0

1 0 0  3 0 1
   
 det(  0 1 0  −  0 2 4 ) = 0
0 0 1   0 0 1 

 − 3 0 −1 
 
 det 0  −2 −4  = 0
 0  − 1
 0

 ( − 3)( − 2)( − 1) = 0 (upper triangular matrix)

  = 1,  = 2,  = 3 (distinct / different roots)

 − 3 0 − 1  x1   0 
    
To find the eigenvectors, consider (I − A) x = 0 , ie  0  − 2 − 4  x2  =  0 
 0  − 1 x3   0 
~ ~
 0

For the eigenvalue  = 1 , we have

 − 2 0 − 1  x1   0   − 2 0 − 1 0 
      
 0 − 1 − 4  x2  =  0  →  0 − 1 − 4 0 
 0 0  x3   0   0 0 0 
 0 0

Now, − 2 x1 − x3 = 0
− x2 − 4 x3 = 0

Let x3 = s . Then, x2 = −4s and x1 = − 12 s

School of Mathematical Sciences (SMS) Page 6-4 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

 x1   − 12 s   1 
     
The solution space x =  x2  =  − 4s  = − 12 s 8 
x   s   − 2
~

 3    
 x1   1 
   
 the solution space of (I − A) x = 0 is of the form x =  x2  = − 12 s 8 
x   − 2
~ ~ ~

 3  
 1 
 
  8  is a basis for the nullspace of (I − A) corresponding to  = 1
 − 2
 

 1 
 
  8  is a basis for the eigenspace of A corresponding to  = 1
 − 2
 

For the eigenvalue  = 2 , we have

 − 1 0 − 1  x1   0   − 1 0 − 1 0 
      
 0 0 − 4  x2  =  0  →  0 0 − 4 0 
 0 0 1  x   0   0 0 1 0 
  3     

Now, − x1 − x3 = 0
− 4 x3 = 0
x3 = 0

Therefore x1 = 0 , x3 = 0 and we let x2 = s

 x1   0   0 
     
The solution space x =  x2  =  s  = s 1 
 x  0 0
~

 3    
 x1   0   0 
     
 the solution space of (I − A) x = 0 is of the form x =  x2  =  s  = s 1 
 x  0 0
~ ~ ~

 3    
 0
 
  1  is a basis for the nullspace of (I − A) corresponding to  = 2
 0
 

School of Mathematical Sciences (SMS) Page 6-5 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

 0
 
  1  is a basis for the eigenspace of A corresponding to  = 2
 0
 

For the eigenvalue  = 3 , we have

 0 0 − 1  x1   0   0 0 −1 0
      
 0 1 − 4  x2  =  0  →  0 1 − 4 0 
 0 0 2  x   0  0 0 2 0
  3     

Now, − x3 = 0
x2 − 4 x3 = 0
2 x3 = 0

Therefore x2 = 0 , x3 = 0 and we let x1 = s

 x1   s   1 
     
The solution space x =  x2  =  0  = s 0 
 x   0 0
~

 3    
 x1   s   1 
     
 the solution space of (I − A) x = 0 is of the form x =  x2  =  0  = s 0 
 x   0 0
~ ~ ~

 3    
1
 
  0  is a basis for the nullspace of (I − A) corresponding to  = 3
0
 

1
 
  0  is a basis for the eigenspace of A corresponding to  = 3
0
 

School of Mathematical Sciences (SMS) Page 6-6 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Example 6.1.3

Find the characteristics equation, eigenvalues, the corresponding eigenvectors, and bases
 1 2 − 2
 
for the eigenspace for the matrix A =  − 2 5 − 2  .
 − 6 6 − 3
 

Solution:

det (I − A) = 0

 1 0 0   1 2 −2 
   
det(  0 1 0  −  −2 5 −2 ) = 0
 0 0 1   −6 6 −3 
   

  − 1 −2 2 
 
det  2  −5 2 =0
 6 −6  + 3 


 −5 2 2 2 2  −5
 −1 +2 +2 =0
−6  + 3 6  +3 6 −6

(  − 1) (  − 5)(  + 3) − (2)(−6)  + 2 2 (  + 3) − (2)(6)  + 2 (2)(−6) − 6 (  − 5) = 0

 3 − 3 2 − 9 + 27 = 0
(  − 3) (  2 − 9 ) = 0
(  − 3)(  − 3)(  + 3) = 0
(  − 3) (  + 3) = 0
2

 = −3,  = 3,  = 3 (same/repeated roots)

3
Checking: 
i =1
i = tr ( A ) ?  −3 + 3 + 3 = 3 = 1 + 5 − 3

 −4 −2 2  x1   0   1 −4 1 0 
      
When  = −3,  2 −8 2  x2  =  0  →  0 3 −1 0 
 6 −6 0  x   0  0 0 0 0
  3     

School of Mathematical Sciences (SMS) Page 6-7 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

x1 − 4 x2 + x3 = 0
3x2 − x3 = 0

Let x2 = s, then x3 = 3s
x1 = 4 x2 − x3 = 4(s) − 3s = s

 x1   s  1
     
The solution space x =  x2  =  s  = s  1  , s 
 x   3s   3
~

 3    

1
  1  is a basis for the nullspace of (I − A) corresponding to  = −3
 3
 

1
  1  is a basis for the eigenspace of A corresponding to  = −3
 3
 

 2 −2 2   x1   0   1 −1 1 0 
      
When  = 3,  2 −2 2   x2  =  0  →  0 0 0 0 
 6 −6 6   x   0   0 0 0 0 
  3     

x1 − x2 + x3 = 0

Let x2 = s and x3 = t , then x1 = x2 − x3 = s − t

 x1   s − t   1   −1
       
The solution space x =  x2  =  s  = s  1  + t  0  , s, t 
x   t   0  1 
~

 3      

 1   −1 
 
  1  ,  0   is a basis for the nullspace of (I − A) corresponding to  = 3
 0   1  
    

 1   −1 
 
  1  ,  0   is a basis for the eigenspace of A corresponding to  = 3
 0   1  
    

School of Mathematical Sciences (SMS) Page 6-8 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Theorem 6.1.1 If k is a positive integer,  is an eigenvalue of a matrix A, and x is

~

a corresponding eigenvector, then  is an eigenvalue of A and x is a corresponding

k k
~
eigenvector.

Example 6.1.4

Find the eigenvalues and corresponding eigenvectors of the matrix A7 where

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

Solution:

From Example 6.1.1, the eigenvalues of A are  = 2 and  = 1  the eigenvalues of A7

are  = 2 7 = 128 and  = 17 = 1 .

 − 2
 
An eigenvector of A correspond to  = 1 is  1   an eigenvector of A7 corresponds to
 1 
 
 − 2
 
 = 1 = 1 is also  1  .
7

 1 
 
 − 1
 
An eigenvector of A correspond to  = 2 is  1   an eigenvector of A7 corresponds to
 1
 
 −1
 = 2 = 128 is also  1  .
7

1
 

School of Mathematical Sciences (SMS) Page 6-9 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Example 6.1.5

Find the eigenvalues and the corresponding eigenvectors of the matrix A9 where

 1 2 − 2
 
A =  − 2 5 − 2 .
 − 6 6 − 3
 

Solution:

From Example 6.1.3, the eigenvalues of A are  = 3 and  = −3  the eigenvalues of A9

are  = 39 = 19683 and  = (−3)9 = −19683.

1  −1
   
Eigenvectors of A correspond to  = 3 are  1  and  0  .  Eigenvectors of A
9

0 1
   
1  −1
   
correspond to  = 3 = 19683 are also  1  and  0  .
9

0 1
   

1
 
An eigenvector of A correspond to  = −3 is  1   an eigenvector of A9 corresponds to
 3
 
1
 = (−3) = −19683 is also  1  .
9

 3
 

School of Mathematical Sciences (SMS) Page 6-10 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

6.2 Diagonalization

Definition A square matrix A is diagonalizable if  an invertible matrix P such that

−1
P AP is a diagonal matrix. The matrix P is said to diagonalize A.

Theorem 6.2.1 Let A be an n  n matrix. Then A is diagonalizable if and only if A

has n linearly independent eigenvectors.

Example 6.2.1

 0 0 −2 
 
Find a matrix P that diagonalizes A =  1 2 1  .
1 0 3 
 

Solution:

From Example 6.1.1, the characteristic equation of A is found to be ( − 1)( − 2)2 = 0.

 − 2
 
We also found that  1  is a basis for the eigenspace of A corresponding to  = 1 , and
 1 
 

 − 1  0 
   
 0 ,  1  forms a basis for the eigenspace of A corresponding to  = 2 .
 1   0 
   

Since A has 3 linearly independent eigenvectors, A is diagonalizable.

 − 2  − 1  0
     
We let p1 =  1  , p 2 =  0  , and p3 =  1  .
~
 1  ~ 1 ~  0
     

 − 2 − 1 0
 
The matrix P =  1 0 1 diagonalizes A. It can be checked that
 1 1 0 

1 0 0
−1  
P AP =  0 2 0  .
0 0 2
 
Evaluate P −1 using adjoint, or ERO ( P | I ) → ( I | P −1 ) , and verify using calculator.

School of Mathematical Sciences (SMS) Page 6-11 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Example 6.2.2
 1 2 − 2
 
Find a matrix P that diagonalizes A =  − 2 5 − 2  .
 − 6 6 − 3
 
Solution:

From Example 6.1.3,  = −3,  = 3,  = 3

1
 
In addition,  1  is a basis for the eigenspace of A corresponding to  = −3.
 3
 

 1   −1 
 
Also,  1  ,  0   is a basis for the eigenspace of A corresponding to  = 3.
 0   1  
    

 1 1 −1
 
The matrix P =  1 1 0  diagonalizes A. It can be checked that
3 0 1 
 
 1/ 3 −1/ 3 1/ 3  1 2 −2  1 1 −1  −3 0 0
−1      
D = P AP =  −1/ 3 4 / 3 −1/ 3  −2 5 −2  1 1 0  =  0 3 0 .
 −1 0     3 
 1  −6 6 −3  3 0 1   0 0

Alternatives:
 1 −1 1
 
The matrix P =  1 0 1  diagonalizes A. It can be checked that
0 1 3 

 −1/ 3 4 / 3 1/ 3  1 2 −2  1 −1 1  3 0 0 
−1      
D = P AP =  −1 1 0  −2 5 −2  1 0 1 =  0 3 0 .
 1/ 3 −1/ 3 1/ 3   3   0 0 −3 
  −6 6 −3  0 1

1 1 −1
 
The matrix P =  1 1 0  diagonalizes A. It can be checked that
0 3 1 

 −1/ 3 4 / 3 −1/ 3  1 2 −2  1 1 −1  3 0 0
−1      
D = P AP =  1/ 3 −1/ 3 1/ 3  −2 5 −2  1 1 0  =  0 −3 0 .
 −1 0     3 
 1  −6 6 −3  0 3 1   0 0

School of Mathematical Sciences (SMS) Page 6-12 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Theorem 6.2.2 If an n  n matrix A has n distinct eigenvalues, then A is

diagonalizable.

Example 6.2.3

 1 − 1 − 1
 
Find the eigenvalues of A =  1 3 1  and decide whether A is diagonalizable.
 − 3 1 − 1
 

Solution:

det (  I − A) = 0

 −1 1 1
−1  −3 −1 = 0
3 −1  +1

 − 3 −1 −1 −1 −1  − 3
(  − 1) −1 +1 = 0 (cofactor expansion along the 1st row)
−1  + 1 3  +1 3 −1

(  − 1) (  − 3)(  + 1) − (−1)(−1)  − 1 − (  + 1 − (−1)(3) ) + 1 (−1)(−1) − 3 (  − 3) = 0

 3 − 3 2 − 4 + 12 = 0

( − 2)( + 2)( − 3) = 0   = 2 ,  = −2 ,  = 3

Since A has 3 distinct eigenvalues, therefore A is diagonalizable.

Example 6.2.4

 4 3 − 2
 
Find the eigenvalues of A =  0 1 1  and decide whether A is diagonalizable.
 0 0 − 2
 

det (  I − A) = 0

 −4 −3 2
0  −1 −1 = 0
0 0 +2

School of Mathematical Sciences (SMS) Page 6-13 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

(  − 4 )(  − 1)(  + 2 ) = 0
 = −2,1, 4

Since A has 3 distinct eigenvalues, therefore A is diagonalizable.

Example 6.2.5

3 0 0
 
Determine whether the matrix A =  0 2 0  is diagonalizable. If so, find a matrix P that
0 1 2
 
−1
diagonalizes A and find P AP.

Solution:

det (I − A) = 0

 −3 0 0
 −2 = (  − 3)(  − 2 ) = 0
2
0 0
0 −1  −2

 = 3,  = 2,  = 2

Since A has 2 distinct eigenvalues, we cannot use Theorem 6.2.2 to decide whether A is
diagonalizable or not. We therefore find how many linearly independent eigenvectors A
has.

 − 3 0 0  x1   0 
    
To find the eigenvectors, consider (I − A) x = 0 , ie  0  −2 0  x2  =  0 
 0 − 1  − 2  x3   0 
~ ~

For the eigenvalue  = 3 , we have

 0 0 0  x1   0   0 0 0 0 
      
 0 1 0  x2  =  0  →  0 1 0 0 
 0 − 1 1  x   0   0 0 1 0 
  3     

Now, x2 = 0
x3 = 0

Let x1 = s .

School of Mathematical Sciences (SMS) Page 6-14 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

 x1   s   1 
     
The solution space x =  x2  =  0  = s 0 
 x   0 0
~

 3    
 x1   s   1 
     
 the solution space of (I − A) x = 0 is of the form x =  x2  =  0  = s 0 
 x   0 0
~ ~ ~

 3    
1
 
  0  is a basis for the nullspace of (I − A) corresponding to  = 3
 0
 
1
 
  0  is a basis for the eigenspace of A corresponding to  = 3
 0
 

For the eigenvalue  = 2 , we have

 − 1 0 0  x1   0   1 0 0 0 
      
 0 0 0  x2  =  0  →  0 0 0 0 
 0 − 1 0  x   0   0 1 0 0 
  3     

Now, x1 = 0
x2 = 0

Let x3 = s .
 x1   0   0 
     
The solution space x =  x2  =  0  = s 0 
 x   s  1
~

 3    
 x1   0   0 
     
 the solution space of (I − A) x = 0 is of the form x =  x2  =  0  = s 0 
 x   s  1
~ ~ ~

 3    
0
 
  0  is a basis for the nullspace of (I − A) corresponding to  = 2
1
 
0
 
  0  is a basis for the eigenspace of A corresponding to  = 2
1
 
Since A has only 2 linearly independent eigenvectors, therefore by Theorem 6.2.1, A is
NOT diagonalizable.

School of Mathematical Sciences (SMS) Page 6-15 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Example 6.2.6

2 0 0 
 
Determine whether the matrix A =  0 − 1 1  is diagonalizable. If so, find a matrix P
 0 0 − 1
 
−1
that diagonalizes A and find P AP.

Solution:

det (I − A) = 0

 −2 0 0
 +1 −1 = (  − 2 )(  + 1) = 0
2
0
0 0  +1

 = 2, −1

 0 0 0  x1   0   0 3 −1 0 
      
When  = 2,  0 3 −1 x2  =  0  →  0 0 3 0 
 0 0 3  x   0  0 0 0 0
  3     

x3 = 0
x2 = 0
x1 = s
 x1   s  1
     
The solution space x =  x2  =  0  = s  0  , s 
 x  0  0
~

 3    

1
  0  is a basis for the nullspace of (I − A) corresponding to  = 2
0
 

1
  0  is a basis for the eigenspace of A corresponding to  = 2
0
 

School of Mathematical Sciences (SMS) Page 6-16 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

 −3 0 0   x1   0   −3 0 0 0 
      
When  = −1,  0 0 −1  x2  =  0  →  0 0 −1 0 
 0 0 0  x  0  0 0 0 0
  3     

x3 = 0
x1 = 0
x2 = t
 x1   0   0 
     
The solution space x =  x2  =  t  = t  1  , t 
 x   0  0
~

 3    

0
  1  is a basis for the nullspace of (I − A) corresponding to  = −1
0
 

0
  1  is a basis for the eigenspace of A corresponding to  = −1
0
 

Since A has only 2 linearly independent eigenvectors, therefore by Theorem 6.2.1, A is

NOT diagonalizable.

Theorem 6.2.3 Let D = P −1 AP be a diagonal matrix for some invertible matrix P.

Then Ak = PD k P −1.

Example 6.2.7

 0 0 −2 
 
Compute A where A =  1 2 1  .
13

1 0 3 
 

Solution:
 − 2 − 1 0
 
From Example 6.1.1 and 6.2.1, it was shown that the matrix P =  1 0 1
 1 1 0 

1 0 0

−1

diagonalizes A and D = P AP =  0 2 0 .
0 0 2 


School of Mathematical Sciences (SMS) Page 6-17 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

A13
= PD13 P −1
 − 2 − 1 0  1 0   − 1 0 − 1
13
0
    
= 1 0 1  0 2 13
0  1 0 2
 1
 1 0   0 0 213  1 1 1
 
 −8190 0 −16382 
 
=  8191 8192 8191 
 8191 16383 
 0

Example 6.2.8

 1 2 − 2
 
Compute A where A =  − 2 5 − 2  .
10

 − 6 6 − 3
 

Solution:

 1 1 −1
 
From Example 6.1.3 and 6.2.2, it was shown that the matrix P =  1 1 0  diagonalizes
3 0 1 
 
 −3 0 0 
−1  
A and D = P AP =  0 3 0  .
 0 0 3
 
A10
= PD10 P −1
 1 1 −1  −3 0 0   1/ 3 −1/ 3 1/ 3 
10

    
=  1 1 0   0 3 0   −1/ 3 4 / 3 −1/ 3 
 3 0 1   0 0 3   −1 0 
    1
 310 0 0 
 
= 0 310 0 
 0 0 310 

 59049 0  0
 
= 0 59049 0 
 0 59049 
 0

School of Mathematical Sciences (SMS) Page 6-18 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

6.3 Cayley-Hamilton Theorem

Theorem 6.3.1 Every square matrix satisfies its own characteristic equation,
that is, if c0 + c1 + c2 2 + + cn−1 n−1 +  n = 0 is the characteristic equation of a
matrix A, then c0 I + c1 A + c2 A2 + + cn−1 An−1 + An = 0.

The theorem also provides an alternative approach to calculate powers and inverse of
a matrix.

Example 6.3.1

3 6
Verify that the matrix A =   satisfies its characteristic equation.
1 2

Solution:

det( I − A) = 0

 − 3 −6
=0
−1  − 2

(  − 3)(  − 2 ) − (−1)(−6) = 0
 2 − 5 + 6 − 6 = 0
 2 − 5 = 0

Replacing  with A,

A2 − 5 A
2
3 6 3 6
=  − 5 
1 2 1 2
15 30  15 30 
= − 
5 10   5 10 
0 0
= 
0 0

School of Mathematical Sciences (SMS) Page 6-19 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Example 6.3.2

1 4
Verify that the matrix A =  −1
 satisfies its characteristic equation. Hence, find A and
 2 3
4
A.

Solution:

det( I − A) = 0

 − 1 −4
=0
−2  − 3

(  − 1)(  − 3) − (−2)(−4) = 0
 2 − 4 + 3 − 8 = 0
 2 − 4 − 5 = 0

Replacing  with A,

A2 − 4 A − 5 I 2
2
1 4 1 4 1 0
=  − 4  − 5 
2 3  2 3 0 1
9 16   4 16   5 0 
= − − 
8 17   8 12   0 5 
0 0
= 
0 0

A2 − 4 A − 5 I 2 = 0
A−1 ( A2 − 4 A − 5 I 2 ) = 0
A − 4 I 2 − 5 A−1 = 0
5 A−1 = A − 4 I 2
1
A−1 = ( A − 4I 2 )
5

School of Mathematical Sciences (SMS) Page 6-20 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

1
A−1 = ( A − 4I 2 )
5
1 1 4 1 0
=   − 4 
5  2 3 0 1
1  −3 4 
=  
5  2 −1

A2 = 4 A + 5 I 2
A3 = A ( 4 A + 5 I 2 ) = 4 A2 + 5 A
= 4 ( 4 A + 5I 2 ) + 5 A
= 16 A + 20 I 2 + 5 A
= 21A + 20 I 2
A = A ( 21A + 20 I 2 ) = 21A2 + 20 A
4

= 21( 4 A + 5 I 2 ) + 20 A = 84 A + 105 I 2 + 20 A
= 104 A + 105 I 2
1 4 1 0
= 104   + 105  
2 3 0 1
 104 416  105 0 
= + 
 208 312   0 105 
 209 416 
= 
 208 417 

Alternative:

A2 − 4 A − 5 I 2 = 0
A2 = 4 A + 5 I 2
A4
= ( A2 )
2

= ( 4 A + 5I 2 )
2

2
 1 4 1 0
=  4  + 5 
  2 3 0 1
2
 9 16 
= 
 8 17 
 209 416 
= 
 208 417 

School of Mathematical Sciences (SMS) Page 6-21 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Example 6.3.3

2 1 
Verify that the matrix A =  6
 satisfies its characteristic equation. Hence, find A .
 5 −2 

Solution:

det( I − A) = 0

 − 2 −1
=0
−5  + 2

(  − 2 )(  + 2 ) − (−5)(−1) = 0
2 − 4 − 5 = 0
2 − 9 = 0

Replacing  with A,

A2 − 9 I 2 = 0
A2 − 9 I 2
2
2 1 1 0
=  − 9 
5 −2  0 1
9 0 9 0
= − 
0 9 0 9
0 0
= 
0 0

A2 − 9 I 2 = 0  A2 = 9 I 2
 729 0 
( A ) = (9I )
2 3
2
3
= 729 I 23 = 729 I 2 =  
 0 729 

School of Mathematical Sciences (SMS) Page 6-22 August 2022 Semester

MAT1024 Linear Algebra & Applications BAS, BINDS, BSDA

Example 6.3.4

 1 2 −2 
 
Find A−1 if A =  −2 5 −2  using the Cayley-Hamilton theorem.
 −6 6 −3 
 

Solution:

From Example 6.1.3, the characteristic equation is found to be  3 − 3 2 − 9 + 27 = 0.

Replacing  with A,

A3 − 3 A2 − 9 A + 27 I 3 = 0
A−1 ( A3 − 3 A2 − 9 A + 27 I 3 ) = 0
A2 − 3 A − 9 I 3 + 27 A−1 = 0
27 A−1 = − A2 + 3 A + 9 I 3

A−1 =
1
27
( − A2 + 3 A + 9 I 3 )

  1 2 −2  2  1 2 −2   1 0 0  
1       
=  −  −2 5 −2  + 3  −2 5 −2  + 9  0 1 0  
27    −6 6 −3   0 0 1  
−6 6 −3 
     
  9 0 0   1 2 −2   9 0 0  
1       
=  −  0 9 0  + 3  −2 5 −2  +  0 9 0  
27       
  0 0 9   −6 6 −3   0 0 9  
 1 2 −2 
1 
=  −2 5 −2 
9 
 −6 6 −3 

School of Mathematical Sciences (SMS) Page 6-23 August 2022 Semester