UNIT – 1 Linear Algebra

LECTURE-8
EIGEN VALUES
AND
EIGEN VECTOR
Eigen values and Eigen vectors:
a) Characteristic Matrix:
For a given square matrix A, A   I matrix is called the characteristic
matrix, where  is scalar and I is the unit matrix.
2 2 1
Let A  1 3 1 
 
1 2 2 
2 2 1 1 0 0   2   2 1 
A   I  1 3 1    0 1 0    1 3 1 
     
1 2 2  0 0 1   1 2 2   
Characteristic matrix
b) Characteristic Polynomial:
The Determinant A   I when expanded will give a polynomial,
which we call as characteristic polynomial of matrix A .
 2 2 1
For example 1 3 1
1 2 2
 (2   )(6  5   2  2)  2(2    1)  1(2  3   )
  3  7 2  11  5
c) Characteristic Equation: The equation A   I  0 is called the
characteristic equation of matrix A .
 3  7 2  11  5  0
d) Characteristic Roots or Eigen Values: The roots of the
characteristic equation A   I  0 are called characteristic roots or eigen
values of matrix A .

 3  7 2  11  5  0
i.e.
(  1)(  1)(  5)  0    1,1,5
Characteristic roots are 1, 1, 5.
 Some Important properties of Eigen Values:

'
1.Any square matrix A and its transpose matrix A have the same
Eigen values.
2.The sum of the Eigen values of a matrix is equal to the trace of
the matrix.
3.The product of the Eigen values of a matrix A is equal to the
determinant of the A .
4.If 1 , 2 , 3 ,........, n are the Eigen values of the matrix A , then the
Eigen values of
(i) kA are k 1 , k 2 , k 3 ,........, k n .

1 1 1 1 1
(ii) A are , , ,........., .
1 2 3 n
(iii) Am are 1m , 2m , 3m ,........, n m
Theorem. Let A be a square matrix of order n.
If λ is an eigenvalue of A, then:
1
1. If A is invertible, then is an eigenvalue of A-1.

2. A is not invertible if and only if   0 is an eigenvalue of A.
3. If  is any number, then    is an eigenvalue of A   I n .
4. If A and B are similar, then they have the same characteristic
polynomial (which implies they also have the same eigenvalues).
 Characteristic Vectors or Eigen Vectors:

 is an eigenvalue of an n X n matrix A if and only if the equation

has a nontrivial solution

Geometrically, a transformation matrix rotates, stretches, or shears the vectors it acts

upon. The eigenvectors for a linear transformation matrix are the set of vectors that are
only stretched, with no rotation or shear. The eigenvalue is the factor by which an
eigenvector is stretched. If the eigenvalue is negative, the direction is reversed.

Eigenvalues and eigenvectors are used to study principal axes of the rotational motion
of rigid bodies, eigenvalues and eigenvectors have a wide range of applications, for
example in stability analysis, vibration analysis, atomic orbitals, facial recognition, and
matrix diagonalization.
Properties of Eigen Vectors;
1. The eigenvector X of a matrix is not unique.
2. If 1 , 2 , 3 ,..............., n be distinct eigenvalues of
an n  n matrix then corresponding eigenvectors
X1 , X 2 , X 3 ,..............., X n from a linearly independent set.
3. If two or more eigenvalues are equal it may or may not
be possible to get linearly independent eigenvectors
corresponding to the equal roots.
4. Two eigenvectors X 1 and X 2 are called orthogonal vectors
1X2  0 .
'
if X
5. Eigenvectors of a symmetric matrix corresponding to
different eigenvalues are orthogonal.
EXAMPLE:
Find the Eigen values and eigen vectors. Also find a basis of the eigenspace E2 corresponding to
the eigenvalue 2
Let

𝜆1 = 1
Eigen values are 𝜆2 = 2
𝜆3 = 2
So we have a homogeneous system of linear equations, we solve it by Gaussian Elimination:
So we have a homogeneous system of linear equations, we solve it by Gaussian Elimination:
The product of a matrix and a vector depends in interesting ways on the entries of each. Eigenvectors represent a
coincidence of direction.
 Non symmetric matrices with non repeated eigenvalues:
3 1 4
Example: Find the eigenvalues and eigenvectors of a matrix A  0 2 6  .
 
0 0 5 
3 1 4
A  I  0 2 6  (3   )(2   )(5   )
0 0 5

Hence the characteristic equation of matrix A is given by

A   I  0    2,   3,   5

Thus eigen values of the matrix A are 2, 3, 5.

The eigenvectors of matrix A corresponding to the eigen value  is given by the zero
solution of the equation ( A   I ) X  0
 3   1 4   x1  0 
OR  0 2 6   x2   0  … (1)
    
 0 0 5     x3  0 
(I) When  =2 ,the corresponding eigenvector is given by

3  2 1 4   x1  0 
 0 22 6   x2   0 
    
 0 0 5  2   x3  0 
1 1 4   x1  0 
0 0 6   x   0
  2  
0 0 3   x3  0 
 x1  x2  x3  0 , 6 x3  0  x3  0 … (2)
From (2)  x1  x2  0  0  x1   x2 ,
Hence x1  k , x2  k , x3  0.
 k 
Eigenvector X 1   k 
 
 0 
When   3 , substituting in (1), the corresponding eigenvector is given by
3  3 1 4   x1  0 
 0 23 6   x2   0 
    
 0 0 5  3  x3  0 

0 1 4   x1  0 
 0 1 6   x2   0 
    
0 0 2   x3  0 

 x2  4 x3  0 ,  x2  6 x3  0, 2 x3  0

From which we obtain x3  0, x2  0, let x1  k , and let k  1

1 
Hence, X 2  0  can be taken as eigenvector of A corresponding to the
 
0 
eigenvalue   3 .
Again, when   5 , substituting in (1), the corresponding eigenvector is given by
3  5 1 4   x1  0 
 0 25 6   x2   0 
    
 0 0 5  5  x3  0 

 2 1 4   x1  0 
  0 3 6   x2   0 
    
 0 0 0   x3  0 
 2 x1  x2  4 x3  0 , 3 x2  6 x3  0
By cross-multiplication method, we have
x1 x x x x x
 2  3  1  2  3 k
6  12 0  12 6  0 3 2 1
 3k   3 
Hence, X 3   2k    2  can be taken as eigenvector of A corresponding to the
   
 k  1 
eigenvalue   5 , where k is arbitrary constant.
Non symmetric matrix with repeated eigen values:
Example: Find all eigen values and all eigenvectors of the matrix
 2 2 3
A   2 1 6 
 
 1 2 0 
Solution: Characteristic equation of A is

2   2 3
2 1   6  0
1 2 0  
 (2   )  2    12  2(2  6)  3(4  1   )  0
  3   2  21  45  0 … (1)
 (  3)(  3)(  5)  0
   3, 3,5
 To find the eigenvectors for corresponding eigen values, we will
consider the matrix equation
(A  I )X  0
 2   2 3   x  0 
 2 1   6   y   0 
    
 1 2    z  0 
On putting  =5 in above, it becomes
 7 2 3  x  0 
 2 4 6   y   0 
    
 1 2 5  z  0
We have -7x+2y-3z=0, 2x-4y-6z=0, -x-2y-5z=0
Solving first and second equations by cross multiplication method, we
have
x y z
  k
1 2 1
k  1
Hence the eigenvector is X 1   2k   k  2 
   
  k   1
Putting  =-3 in above equation , it becomes

1 2 3  x   0 
2 4 6   y    0 
    
 1 2 3   z   0 

We have x+2y-3z=0, 2x+4y-6z=0, -x-2y+3z=0

First, second and third equation is same.
1
Let x  k 1 , y  k 2 , then z   k1  k 2 
3
 
 k1 
 
Hence the eigen vector is  k2 
1 
  k1  k2  
3 
 0
Let k1  0, k2  3 , hence X 2   3 
 
 2 
Since the matrix is non-symmetric , the corresponding eigenvectors X2 and X3 must
be linearly independent. This can be done by choosing,
3
k1  1, k2  0, Hence X 3  0  .
1 
SYMMETRIC MATRICES WITH NONREPEATED EIGENVALUES
Example: Find the eigenvalues and the corresponding eigenvectors of the matrix
 2 5 4 
A   5 7 5 .
 
 4 5 2 
Solution:
2   5 4
A  I  5 7 5 =0 or   3  3 2  90  216  0
4 5 2  
On solving above equation.   3, 6,12
Matrix equation for eigenvectors  A   I  X  0
 2   5 4   x  0
 5 7 5   y   0 … (1)
    
 4 5 2     z   0 

Eigenvector corresponding to eigen value   3

Equation (1) becomes,

 2  3 5 4   x  0
 5 73 5   y   0
    
 4 5 2  3  z   0 
1 5 4   x  0
      x  5 y  4z  0
 5 10 5 y  0 
     5 x  10 y  5 z  0
 4 5 1   z   0 
By cross-multiplication method
x y z.
x

y

z or  
25  40 20  5 10  25 1 1 1
 1 
Eigenvector
X 1   1
 

 1 

Eigenvector corresponding to eigenvalue   6
Equation (1) becomes,
 2  6 5 4   x 0
 5 76 5   y   0
    

 4 5 2  6 
z 
0

4 5 4  x  0
      4 x  5 y  4 z  0
 5 13 5 y  0 
     5 x  13 y  5 z  0

4 5 4
z  
0
By cross-multiplication method
x y z x y z .
  or  
25  52 20  20 52  25 1 0 1
1 
 
Eigenvector X 2   0 

 1
Eigenvector corresponding to eigenvalue   12
Equation (1) becomes,
 2  12 5 4   x  0
 5 7  12 5   y   0 
    
 4 5 2  12   z  0 
 14 5 4   x  0
 14 x  5 y  4 z  0
 5 5 5   y   0 
      5x  5 y  5z  0
 4 5 14   z  0 
By cross-multiplication method
x y z x y z
  or   .
25  20 20  70 70  25 1 2 1
1 
Eigenvector X 3   2 
 
1 
Applications of a Square Matrix’s Eigenvalues and Eigenvectors

1. System of Communication: Claude Shannon utilized eigenvalues to calculate the theoretical limit of how much information can be
carried via a communication channel such as a telephone line or the air. The eigenvectors and eigenvalues of the communication
channel (represented as a matrix) are calculated, and then the eigenvalues are water filled. The eigenvalues are then essentially the
gains of the channel’s fundamental modes, which are recorded by the eigenvectors.

2. Bridge Construction: The smallest magnitude eigenvalue of a system that models the bridge is the natural frequency of the bridge.
Engineers use this knowledge to guarantee that their structures are stable.

3. Automobile Stereo System Design: Eigenvalue analysis is also employed in the design of car stereo systems, where it aids in the
reproduction of car vibration caused by music.

4. Electrical Engineering: The use of eigenvalues and eigenvectors to decouple three-phase systems via symmetrical component
transformation is advantageous. Also eigenvalues and eigenvectors aid in analyzing circuits and networks. By diagonalizing the
impedance or admittance matrix, engineers can simplify complex circuit calculations and predict the response to different inputs.

5. Mechanical Engineering: Eigenvalues and eigenvectors enable us to “decompose” a linear process into smaller, more manageable
tasks. When stress is applied to a “plastic” solid, for example, the deformation can be divided into “principle directions,” or the
directions where the deformation is greatest. The eigenvectors in the principle directions are the eigenvectors, and the associated
eigenvalue is the percentage deformation in each principle direction. Also, Mechanical engineers utilize eigenvalues and eigenvectors
to analyze dynamic systems such as rotating machinery and vehicle dynamics. Eigenvalues determine the stability of rotational
speeds or vehicle maneuvers, while eigenvectors represent the modes of motion

Eigenvalue analysis is commonly used by oil firms to explore land for oil. Because oil, dirt, and other substances all produce linear
systems with varying eigenvalues, eigenvalue analysis can help pinpoint where oil reserves lie. Oil companies set up probes all-
around a site to pick up the waves created by a massive truck vibrating the ground. The waves are modified when they move through
the different substances in the earth. The oil corporations are directed to possible drilling sites based on the study of these waves.
Example: Application in Mechanical Engineering
In mechanical engineering, eigenvalues can be used to determine natural frequencies of a
2 0
system. Consider a 2-DOF mass-spring system with mass matrix M= and stiffness
0 1
5 −1
matrix K= . Find the natural frequencies.
−1 5
Solution:
1.Form the generalized eigenvalue problem: det (K−λM)=0
2.Find Eigenvalues:
5−2λ −1
2. K−λM= =0
−1 5−λ
3. Determinant: (5−2λ)(5−λ)−1⋅(−1) = 0
4. λ^2−15λ+24 = 0
5. Solving for λ: λ=3 or λ=8
3.Natural Frequencies:
3. ω1=3​, ω2=8.
Eigenvalues and eigenvectors in data analysis are applied primarily in techniques like
Principal Component Analysis (PCA) for dimensionality reduction, where they identify the
directions of maximum variance in datasets. They are also crucial for feature extraction,
aiding in simplifying data representation while retaining essential information.
 Practice Problems
6 2
1. Given matrix A=
2 3

Find the eigenvalues.

2 1
Find the corresponding eigenvectors. Ans : , ,
1 −2

2. An electrical circuit has an impedance matrix Z= 4 1 . Determine the eigenvalues and

1 3
eigenvectors to simplify the circuit analysis.

3. Consider a mass-spring system with mass matrix M = 3 0

and stiffness matrix K = 7
1
.
0 2 1 4
Find the natural frequencies of the system.

2 0.8
4. Given the covariance matrix Σ =
0.8 1.5
Find the eigenvalues.
Find the eigenvectors.
Use the eigenvectors to determine the principal components.

10 4 .
5. A communication channel is represented by the matrix matrix A=
4 5
Calculate the eigenvalues and eigenvectors to determine the fundamental modes of the channel.
Diagonalisation of Matrix
An n x n matrix A is said to be diagonalizable if there is a non-singular
matrix P such that A = PDP-1 where D is a diagonal matrix.
The eigen values of a diagonal matrix are the numbers down its diagonal and the
eigenvector corresponding to the ith diagonal element is the ith standard basis
element ei.
If the matrix A has n distinct real eigen values then A can be diagonalised using a real
non-singular matrix P to produce a real diagonal matrix D. The columns of P can be
taken to be the eigenvectors of A.
More generally, A can be diagonalised precisely when n linearly independent real
eigenvectors of A can be found. In which case P is formed by taking these
eigenvectors as its columns. The ith diagonal element of D is the eigenvalue
corresponding to the ithcolumn vector of P.
METHOD TO FIND D

That is, find an invertible matrix P and a diagonal matrix D such that A = D A D-1

Step 1. Find the eigenvalues of A

Step 2. Find three linearly independent eigenvectors of A

Step 3. Construct P from the vectors in step 2

Step 4. Construct D by finding P-1 . Then

D= P-1 A P
(These statements remain true if some of the eigenvalues of A are complex, provided
we accept complex matrices for P and D.)
1 6 1 
Examples: Find a matrix P which diagonalizes the matrix A  1 2 0  .
 
0 0 3 
Solution: The characteristic equation of A is A  I  0

1  6 1
 1 2 0 0
0 0 3
 (  3)(  1)(  4)  0
   3, 1, 4.

 x
Let X   y  be eigen vector corresponding to  is given by [A-  I]X=0
 
 z 
1   6 1   x  0 
 1 2 0   y   0 
    
 0 0 3     z  0 
(i) When  =-1, then above matrix equation becomes:

 2 6 1   x  0 
1 3 0   y   0
    
 0 0 4   z  0 

Since the rank of the coefficient matrix is 2, so selecting first two rows, we get
x y z  3
  , So X 1   1  is eigen vector for  =-1.
3 1 0  0 

(I) When  =3, the matrix equation becomes

 2 6 1   x  0 
 1 1 0   y   0 
    
 0 0 4   z  0 

Since rank of coefficient matrix is 2 , so that

1
, So eigen vector X 2   1  is eigen vector for
x y z
   =3.
1 1 4
 4 
(I) When  =4, then matrix equation becomes

 3 6 1   x  0 
 1 2 0   y    0  . Since the rank of coefficient matrix is 2 , so that
    
 0 0 1  z   0 
x y z
 
2 1 0
 3 1 2 
The modal matrix P=[X1,X2,X3] P=  1 1 1  .
 0 4 0 

 4 8 1  4 8 1 1 6 1   3 1 2 
1 
0 0 5 and D  P 1 AP 
1 
Find P-1 P 1  0 0 5 1 2 0   1 1 1 
20   20    
 4 12 4   4 12 4  0 0 3  0 4 0 

 1 0 0 
Hence the diagonal matrix D=  0 3 0  .
 
 0 0 4 
Complex Matrix:
Conjugate of a matrix: the matrix formed by replacing the elements of a matrix by
their respective conjugate numbers is called the conjugate of A and is denoted by A .

A  (aij )mn , then A  (a ij )mn

For example

3  4i 2  i 4  3  4i 2  i 4 
If A    then A   
 i 2 3i    i 2 3i 
Theorem

If A and B be two matrices and their conjugate matrices A and B , then

(i) ( A)  A (ii) ( A  B)  A  B , (iii) (k A)  k A (iv) ( AB)  AB .

Transpose of a conjugate matrix:

The conjugate of a matrix A is denoted by A or A* .

( A)'  A or A*

The (i, j)th element of A =(j, i)th element of A =complex conjugate of (j, i)th
element of A.

 2  3i 1  2i 2  4i 
Example: if A  3  4i 4  3i 2  6i  , find A
 
 5 5  6i 3 

 2  3i 1  2i 2  4i   2  3i 1  2i 2  4i 
Solution: we have, A  3  4i 4  3i 2  6i  , A  3  4i 4  3i 2  6i 
   
 5 5  6i 3   5 5  6i 3 

 2  3i 3  4i 5 
A  ( A)'  1  2i 4  3i 5  6i  ANS.
 
 2  4i 2  6i 3 
Hermitian Matrix: A square matrix A  (aij ) is said to be Hermitian matrix if the

(i, j)th of A, is equal to the conjugate complex of the (j, i)th element of A.
A = Aθ
i.e., aij  a ji for all i and j

 2 3  4i   a b  id 
For example,   , 
 3  4i 1   b  id c 
Hence the all elements of the principal diagonal are real.
A necessary and sufficient condition for a matrix A to be Hermitian is that A  A .

 1 1  i 2
Example: Prove that matrix A  1  i 3 i  is Hermitian Matrix.
 
 2 i 0 
Skew Hermitian Matrix: A square matrix A  (aij ) is said to be Skew-Hermitian
matrix if the (i, j)th of A, is equal to the negative of conjugate complex of the
(j, i)th element of A.
Aθ = -A
i.e., aij  a ji for all i and j

If A is a skew-Hermitian matrix, then

aii  aii
aii  aii  0

Obiously, a is either apure imaginary number or must be zero.

ii

 0 3  4i 
3  4i 
 0 
A necessary and sufficient condition for a matrix A to be skew-Hermitian is that
A   A .
  i 3  2i 2  i 
Example: show that A   3  2i 0 3  4i  si skew-Hermitian matrx.
 
 2  i 3  4i 2i 

 i 3  2i 2  i 
Solution: A   3  2i 0 3  4i 
 
 2  i 3  4i 2i 

 i 3  2i 2  i   i 3  2i 2  i 
( A)'   3  2i 0 3  4i  A   3  2i 0 3  4i 
   
 2  i 3  4i 2i   2  i 3  4i 2i 

 i 3  2i 2  i 
   3  2i 0 3  4i    A
 
 2  i 3  4i 2i 

A   A  is a skew  Hermitian matrix .

Unitary Matrx: A suare matrx is said to be unitary marix if A A  AA  I ,
where I is the identity matrix.

1  1 1 i
Example: Show that the matrx A  1  i 1  is unitary.
3 

1  1 1 i
Solution: We have A   
3 1  i 1 

1  1 1 i  1  1 1 i
A'    A  ( A '
)   
3 1  i 1  3 1  i 1 

1  1 1  i  1  1 1  i  1  3 0  1 0 
A A  1  i 1  1  i 1   3 0 3   0 1   I
3  3     
Similarly AA  I
 A A  AA  I

Hence matrix A is Unitary.