2.

7 EIGENVALUES AND EIGENVECTORS

Introduction

We know that many physical systems arising in engineering applications can be

represented as discrete models involving matrices. Some key parameters
describing physical systems (e.g., the resonance frequency) are closely related
to eigenvalues of the matrix representing the system. For example, the natural
frequency of the bridge is the eigenvalue of smallest magnitude of a system that
models the bridge. The engineers exploit this knowledge to ensure the stability
of their constructions. Eigenvalue analysis is also used in the design of car
stereo systems, where it helps to reduce the vibration of the car due to the
music. In electrical engineering, the application of eigenvalues and eigenvectors
is useful for decoupling three-phase systems through symmetrical component
transformation.

An eigenvector of an 𝑛 × 𝑛 matrix A is nonzero vector X such that 𝐴𝑋 = 𝜆𝑋 for

some scalar 𝝀. A scalar 𝝀 is called an eigenvalue of A if there is a nontrivial
solution X of 𝐴𝑋 = 𝜆𝑋 ; such an X is called an eigenvector corresponding to 𝝀.
 AX =  X  ( A −  I ) X = O

where I is the nth order identity matrix. To find all 𝝀, the determinant of this
matrix ( A −  I ) equated to zero.

a11 −  a12 .... a1n

a21 a22 −  .... a2 n
i.e, A − I = =0 (1)
.... .... .... ....
an1 an 2 .... ann − 

is called as the characteristic equation of the matrix A . On expanding the

determinant, the characteristic equation takes the form
n n n-1 n-2
(-1) λ +k1λ +k 2λ +.....+k n =0 (2)

where the ki ’s are expressible in terms of the elements aij of the matrix A .

The roots of the characteristic equation of A are the eigenvalues, latent roots or
the characteristic roots of the matrix A .
The matrix equation (1) represents n homogenous linear equations
(a11 −  ) x1 + a12 x2 + ..... + a1n xn = 0
a21 x1 + (a22 −  ) x2 + ..... + a2 n xn = 0
(3)
.... .... .... ....
an1 x1 + an 2 x2 + ..... + (ann −  ) xn = 0

which will have a non-trivial solution if and only if the coefficient matrix is
singular, i.e, if A −  I = 0 .

The characteristic equation of A has n roots (need not be distinct) and

corresponding to each root, the equation (3) will have a non-zero solution,
which is the eigenvector or latent vector.

Examples:

5 4 
1. Find the eigenvalues and eigenvectors of the matrix A =  .
1 2 

Solution: The characteristic equation is A −  I = 0

5− 4
i.e, = 0 or  2 − 7 + 6 = 0
1 2−

or ( − 6)( − 1) = 0

  = 6,1

If X = [ x1 , x2 ]T is an eigenvector corresponding to the eigenvalue  ,then

5 −  4   x1 
(A − I )X =  =0.
 1 2 −    x2 

 −1 4   x1 
Corresponding to  = 6 , we have  = 0 which gives only one
1 −4   x2 
equation − x1 + 4 x2 = 0 .

x1 x2
 = giving the eigenvector (4,1)T .
4 1

4 4  x 
Corresponding to  = 1 , we have    x  = 0 which gives only one equation
1

 1 1  2
x1 + x2 = 0 .
 x1 x2
 = giving the eigenvector (1, −1)T .
1 −1

1 1 3
2. Find the eigenvalues and eigenvectors of the matrix A = 1 5 1 .
3 1 1

1−  1 3
Solution: The characteristic equation is given by A −  I = 1 5 −  1 = 0
3 1 1− 

i.e  3 − 7 2 + 36 = 0 or ( + 2)( − 3)( − 6) = 0 .

  = −2,3,6

If X =  x1 , x2 , x3  is an eigenvector corresponding to an eigenvalue  , then

T

we have

1 −  1 3   x1 

 A −  I  X =  1 5 −  1   x2  = O
 3 1 1 −    x3 

For  = 2 , we have the equations 3x1 + x2 + 3x3 = 0

x1 + 7 x2 + x3 = 0
3x1 + x2 + 3x3 = 0

x1 x2 x3
Solving, we obtain = = , an eigenvector is (−1, 0,1)T .
−1 0 1

Also, every non-zero multiple of this vector is an eigenvector corresponding to

 = −2 .

Similarly, the eigenvectors corresponding to  = 3 and  = 6 are the arbitrary

non-zero multiples of the vectors (1, −1,1)T and (1, 2,1)T .

Exercises:

1.Find the eigenvalues and eigenvectors of the matrices:

 8 −6 2   2 1 −1 2 0 1
a)  −6 7 −4 b)  1 1 −2 c) 0 2 0 
 
 2 −4 3   −1 −2 1  1 0 2 
Properties of Eigenvalues:

1. Any square matrix A and its transpose AT have the same eigenvalues.

2. The eigenvalues of a triangular matrix are the diagonal elements of the

matrix.

3. The eigenvalues of an idempotent matrix are either zero or unity.

4. The sum of the eigenvalues of a matrix is the sum of the elements of the
principal diagonal(trace of a matrix).

5. The product of the eigenvalues of a matrix is equal to its determinant.

1
6. If  is an eigenvalue of a matrix A , then is the eigenvalue of A −1 .


1
7. If  is an eigenvalue of an orthogonal matrix, then is also its eigenvalue.


8. If 1, 2 ,....n are the eigenvalues of a square matrix A of order n then Am has
the eigenvalues 1m , 2m ,.....n m where m is a positive integer.

Cayley-Hamilton Theorem

Statement: Every square matrix satisfies its own characteristic equation.

i.e if the characteristic equation for the nth order square matrix A is

A −  I = (−1) n  n + k1 n −1 + k2 n − 2 + ..... + kn = 0 , then

(−1)n An + k1 An−1 + k2 An−2 + ..... + kn I = 0 …………………. (4)

Multiplying (4) by A −1 , we get (−1)n An−1 + k1 An−2 + ..... + kn−1I + kn A−1 = 0 .

1
 A−1 = − (−1)n An −1 + k1 An −2 + .... + kn −1I  .
kn

This result gives the inverse of A in terms of n −1 powers of A and is

considered as a practical method for the computation of the inverse of the large
matrices.
Examples:

1 4
1. Verify Cayley-Hamilton theorem for the matrix A =   and hence find its
2 3
inverse.
1−  4
Solution: The characteristic equation of A is given by = 0 or
2 3−
 2 − 4 − 5 = 0 .

By Cayley-Hamilton theorem, A satisfies its characteristic equation, i.e.

A2 − 4 A − 5I = O .

1 4  1 4  1 4  1 0 
Now, A2 − 4 A − 5I =     −4  −5 
2 3 2 3 2 3 0 1 
9 16   4 16  5 0 
= − − 
8 17  8 12  0 5 
0 0 
= =O
0 0 

This verifies the theorem.

Multiplying by A −1 , we get A − 4I − 5 A−1 = O or

1 1  1 4  1 0   1  −3 4 
A−1 = ( A − 4 I ) =    −4  =  .
5 5  2 3 0 1   5  2 −1

1 2
2. Using Cayley-Hamilton theorem, find A8 , if A =  .
 2 −1

1−  2
Solution: The characteristic equation of A is given by = 0 or
2 −1 − 
2 − 5 = 0 .

A8 = ( A2 ) 4 = ( A2 − 5 I + 5 I ) 4
= ( A2 − 5I ) 4 + 4( A2 − 5I )3 5I + 6( A2 − 5I ) 2 (5I ) 2 + 4( A2 − 5I )(5I )3 + (5I ) 4

By Cayley-Hamilton theorem, A2 − 5I = O .

 A8 = (5I ) 4 = 625I .
 1 2 0 
3. Using Cayley-Hamilton theorem, find A , where A =  2 −1 0  .
−2

0 0 −1

1−  2 0
Solution: The characteristic equation of A is given by 2 −1 −  0 = 0 or
0 0 −1 − 
− 3 −  2 + 5 + 5 = 0 (5)
1
Multiplying by A −1 , we get − A2 − A + 5I + 5 A−1 = O or A−1 =  A2 + A − 5I  .
5

Multiplying by A −1 , we get − A − I + 5 A−1 + 5 A−2 = O or

−2 −1
5A = A + I − 5A
1
= A + I − 5    A2 + A − 5I 
5
= 6 I − A2

 6 0 0  5 0 0   1 0 0 
1   1
  
 A = 6 I − A  =  0 6 0 − 0 5 0  = 0 1 0 
1
−2 2

5 5  5 0 0 5 
 0 0 6 0 0 1    

2 2 1
4. If A = 1 3 1  , then find the eigenvalues of A −1 and hence, find its
1 2 2
determinant.

Solution: The eigenvalues of A are given by the roots of its characteristic

equation, i.e A −  I =  3 − 7 2 + 11 − 5 = 0 .

  = 5,1,1 .

1
If  is an eigenvalue of A , then is an eigenvalue of A −1 .


1
Therefore, the eigenvalues of A −1 are ,1,1 .
5

1
 A−1 = .
5

Exercises:
1. Verify Cayley-Hamilton theorem for the following matrices and find its
inverse:

 2 −1 1   7 2 −2 3 2 4
a)  −1 2 −1 b)  −6 −1 2  c)  4 3 2 
 
 1 −1 2   6 2 −1  2 4 3 

 2 −1 2 
2. If A = −1 2 −1 , find A4 .
 1 −1 2 

2 3 4
3. Find the eigenvalues of A − 2 A + I where A = 0 4 2 .
2

0 0 3 

Iterative method to obtain largest eigenvalue of a matrix (Rayleigh Power

method)

Consider the matrix equation AX =  X which gives the following homogenous

system of equations
(a11 −  ) x1 + a12 x2 + ..... + a1n xn = 0
a21 x1 + (a22 −  ) x2 + ..... + a2 n xn = 0
.... .... .... ....
an1 x1 + an 2 x2 + ..... + (ann −  ) xn = 0

In many applications, it is required to compute numerically the largest

eigenvalue and the corresponding eigenvector.

We start with a column vector X which is as near the solution as possible and
evaluate AX , which is written as  (1) X (1) after normalisation, that is, we obtain
 (1) from AX which is largest in magnitude. Hence, the resulting vector X (1)
will have largest component unity. This gives the first approximation  (1) to the
largest eigenvalue and X (1) to the eigenvector. Similarly, we evaluate
AX (1) =  (2) X (2) which gives the second approximation. We repeat this process till
X (r) − X ( r −1) becomes negligible. Then,  ( r ) will be the largest eigenvalue of A
and X ( r ) , the corresponding eigenvector.

This iterative method to find the dominant eigenvalue of a matrix is known as

Rayleigh’s power method.
For the initial approximation, we can take the vector X = [1, 0, 0]T , if it is not
explicitly provided.

Note: The largest eigenvalue of A −1 is the smallest eigenvalue of A .

Examples:

1. Determine the largest eigenvalue and the corresponding eigenvector of the

 2 −1 0 
matrix A = −1 2 −1 up to 4 iterations.
 0 −1 2 

Solution: Let the initial approximation to the required eigenvector be

X = 1, 0, 0 .
T

 2 −1 0  1   2   1 
AX =  −1 2 −1 0 =  −1 = 2  −0.5 =  (1) X (1)
 0 −1 2  0  0   0 

 2 −1 0   1   2.5  1 
AX (1)
=  −1 2 −1  −0.5 =  −2  = 2.5  −0.8 =  (2) X (2)
     
 0 −1 2   0  0.5  0.2 

 2 −1 0   1   2.8   1 
AX (2)      
=  −1 2 −1  −0.8 =  −2.8  = 2.8  −1  =  (3) X (3)
 0 −1 2   0.2  1.204 0.43

 2 −1 0   1   2.9841 0.87 
AX (3)      
=  −1 2 −1  −1  =  −3.43  = 3.43  −1  =  (4) X (4)
 0 −1 2  0.43 1.8522  0.54 

Hence, the largest eigenvalue is 3.43 and the corresponding eigenvector is

0.87, −1, 0.54  .
T

2. Find the largest eigenvalue and the corresponding eigenvector of the matrix
 1 3 −1
A =  3 2 4  with the initial approximation 1, 1, 0 . Carry out 4 iterations.
T

 −1 4 10 

Solution: Given X = 1, 1, 0 we have

T
  1 3 −1 1   4 0.8
AX =  3 2 4  1  = 5  = 5  1  =  (1) X (1)
     
 −1 4 10  0  3 0.6

 1 3 −1 0.8 3.2 0.3478

AX (1)      
=  3 2 4   1  = 6.8 = 9.2  0.7391 =  (2) X (2)
 −1 4 10  0.6  9.2   1 

 1 3 −1 0.3478  1.5652   0.1241

AX (2)
=  3 2 4   0.7391 =  6.5217  = 12.6087 0.5172 =  (3) X (3)
     
 −1 4 10   1  12.6087   1 

 1 3 −1  0.1241  0.6759   0.0566

AX (2)      
=  3 2 4  0.5172 =  5.4069  = 11.9448 0.4527 =  (4) X (4)
 −1 4 10   1  11.9448  1 

Hence, the largest eigenvalue is 11.9448 and the corresponding eigenvector is

0.0566, 0.4527, 1
T

Exercises:

1. Find the largest eigenvalue and the corresponding eigenvector of the matrix
1 −3 2 
 4 4 −1 using the initial approximation 1, 0, 0 T . Carry out 4 iterations.
   
6 3 5 

2. Find the dominant eigenvalue and the corresponding eigenvector of the

 6 −2 2 
matrix  −2 3 −1 . Take an appropriate initial approximation and carry out 4
 2 −1 3 

iterations.