2.1.

Eigen values and Eigenvectors

Definition: Let A be an 𝑛 × 𝑛 matrix. A scalar 𝜆 is called an
eigenavlue of A if there exists a nonzero vector 𝑥 in ℝ𝑛 such
that 𝐴𝑥 = 𝜆𝑥 . The vector x is called an eigenvector
corresponding to 𝜆.
Let us look at the geometrical significance of an
eigenvector that corresponds to a nonzero eigenvalue. The
vector Ax is in the same or opposite direction as x,
depending on the sign of 𝜆 . See the figure below. An
eigenvector of A is thus a vector whose direction is
unchanged or reversed when multiplied by A.

Computation of Eigenvalues and Eigenvectors

Let A be an 𝑛 × 𝑛 matrix with eigenvalue 𝜆 and
corresponding eigenvector x. Thus 𝑨𝒙 – 𝝀𝒙. This equation
may be rewritten

𝑨𝒙 – 𝝀𝒙 = 0
giving
𝐴 − 𝜆𝐼𝑛 𝑥 = 0
This matrix equation represents a system of homogeneous
linear equations having matrix of coefficients 𝐴 − 𝜆𝐼𝑛 .
𝑥 = 0 is a solution to this system. However, eigenvectors
have been defined to be nonzero vectors. Further, nonzero
solutions to this system of equations can only exist if the
matrix of coefficients is singular, 𝐴 − 𝜆𝐼𝑛 = 0 . Hence,
solving the equation 𝐴 − 𝜆𝐼𝑛 = 0 for 𝜆 leads to all the
eigenvalues of A.
On expanding the determinant 𝐴 − 𝜆𝐼𝑛 , we get a
polynomial in 𝜆. This polynomial is called the characteristic
polynomial of A. The equation 𝐴 − 𝜆𝐼𝑛 = 0 is called the
characteristic equation of A.
The eigenvalues are then substituted back into the
equation 𝐴 − 𝜆𝐼𝑛 𝑥 = 0 to find the corresponding
eigenvectors.
We are always interested in knowing whether sets of
vectors form subspaces!
Theorem: Let A be an 𝑛 × 𝑛 matrix and 𝜆 an eigenvalue of
A. The set of all eigenvectors corresponding to 𝜆, together
with the zero vectors, is a subspace of ℝ𝑛 . This subspace is
called the eigenspace of 𝜆.
Proof: In order to show that the eigenspace is a subspace,
we have to show that it is closed under vector addition and
scalar multiplication.
Let 𝑥1 and𝑥2 be two vectors in the eigenspace of 𝜆 and let c
be a scalar. Then 𝐴𝑥1 = 𝜆𝑥1 and 𝐴𝑥2 = 𝜆𝑥2 . Hence,
 𝐴𝑥1 + 𝐴𝑥2 = 𝜆𝑥1 + 𝜆𝑥2
𝐴( 𝑥1 + 𝑥2 ) = 𝜆(𝑥1 + 𝑥2 )
Thus 𝑥1 + 𝑥2 is a vector in the eigenspace of 𝜆 . The
eigenspace is closed under addition.
Further, since 𝐴𝑥1 = 𝜆𝑥1 ,
𝑐𝐴𝑥1 = 𝑐𝜆𝑥1
𝐴 𝑐𝑥1 = 𝜆 𝑐𝑥1
Therefore 𝑐𝑥1 is a vector in the eigenspace of 𝜆 . The
eigenspace is closed under scalar multiplication.
Thus the eigenspce is a subspace.
Properties of Eigen Values
Property 1: The sum of the eigen values of a square matrix
𝐴 is the sum of the diagonal elements (trace) of 𝐴.
Property 2: The product of the eigen values is 𝐴 .

 a11 a12  a1n 

a a22  a2 n 
Proof: Let A   
21

     
 
 n1 n 2
a a  ann 

The eigen values are got from the characteristic equation

 a11   a12  a1n
a21 a22    a2 n
A  I  0
   
an1 an 2  ann  

Let A   I  a0  a1

n 1
n
   an . . . ………………….(1)

Setting   0 we get A  an …………………………….(2)

Also explaining A   I by first row we get

A   I  (1)n  n  (1)n1 (a11  a22    ann ) n1   …….(3)

Comparing (1) and (3) we get

a0  (1)n ; a1  (1)n1 (a11  a22    ann ); ……………(4)

If 1 , 2 ,n are the eigen values ( (i.e) the roots of the
characteristic equation) then
𝑆𝑢𝑚 𝑜𝑓 𝑡𝑕𝑒 𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒𝑠 = 𝑠𝑢𝑚 𝑜𝑓 𝑡𝑕𝑒 𝑟𝑜𝑜𝑡𝑠

= 1  2    n
𝑎1
=− = 𝑎11 + 𝑎22 + ⋯ + 𝑎𝑛𝑛 (Using 4)
𝑎0

= 𝑡𝑟𝑎𝑐𝑒 𝑜𝑓 𝐴
𝑃𝑟𝑜𝑑𝑢𝑐𝑡 𝑜𝑓 𝑡𝑕𝑒 𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒𝑠 = 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 𝑜𝑓 𝑡𝑕𝑒 𝑟𝑜𝑜𝑡𝑠

= 1 , 2 ,n
 𝑎𝑛 (−1)𝑛 𝑎𝑛
𝑛
= (−1) =
𝑎0 (−1)𝑛
= 𝑎𝑛
= 𝐴 (𝑓𝑟𝑜𝑚 2 )
Property 3: The eigen values of 𝐴 and its transpose 𝐴𝑇 are
the same.

Proof: It is enough if we prove that A and 𝐴𝑇 have the same

characteristic polynomial. Since for any square matrix
𝑀, 𝑀 = 𝑀𝑇 we have,
𝐴 − 𝜆𝐼 = (𝐴 − 𝜆𝐼)𝑇 = 𝐴𝑇 − (𝜆𝐼)𝑇 = 𝐴𝑇 − 𝜆𝐼
Hence the result.
Property 4: If 𝜆 is an eigen value of a non singular matrix
1
𝐴 then is an eigen value of 𝐴−1 .
𝜆

Proof : Let 𝑋 be an eigen vector corresponding to 𝜆.

Then 𝐴𝑋 = 𝜆𝑋. Since 𝐴 is non singular 𝐴−1 exists.

∴ 𝐴−1 𝐴𝑋 = 𝐴−1 (𝜆𝑋)

𝐼𝑋 = 𝜆𝐴−1 𝑋
1
∴ 𝐴−1 𝑋 = 𝑋.
𝜆
1
∴ is an eigen value of 𝐴−1 .
𝜆
Corollary: If 𝜆1 , 𝜆2 , … , 𝜆𝑛 are the eigen values of a non
1 1 1
singular matrix 𝐴 then , ,…, are the eigen values of
𝜆1 𝜆2 𝜆𝑛
𝐴−1 .
Property 5: If 𝜆 is an eigen value of 𝐴 then 𝑘𝜆 is an eigen
value of 𝑘𝐴 where 𝑘 is a scalar.
Proof: Let 𝑋 be an eigen vector corresponding to 𝜆.
Then 𝐴𝑋 = 𝜆𝑋.
Now, 𝑘𝐴 𝑋 = 𝑘(𝐴𝑋)
= 𝑘(𝜆𝑋) (by (1))
= 𝑘𝜆 𝑋.
∴ 𝑘𝜆 is an eigen value of 𝑘𝐴.

Property 6: If 𝜆 is an eigen value of 𝐴 then 𝜆𝑘 is an eigen

value of 𝐴𝑘 where 𝑘 is any positive integer.
Proof: Let 𝑋 be an eigen vector corresponding to 𝜆.
Then 𝐴𝑋 = 𝜆𝑋.

Now, 𝐴2 𝑋 = 𝐴𝐴 𝑋 = 𝐴(𝐴𝑋)
= 𝐴(𝜆𝑋) (by (1))
= 𝜆(𝐴𝑋)
= 𝜆(𝜆𝑋) (by (1))
= 𝜆2 𝑋.
∴ 𝜆2 is an eigen value of 𝐴2 .
Proceeding like this we can prove that 𝜆𝑘 is an eigen value
of 𝐴𝑘 for any positive integer.
Corollary: If 𝜆1 , 𝜆2 , … , 𝜆𝑛 are eigen values of 𝐴 then
𝜆1𝑘 , 𝜆𝑘2 , … , 𝜆𝑘𝑛 are given values of 𝐴𝑘 for any positive integer.

Applications
Eigenvalues and eigenvectors are extremely important tools
in applying mathematics. They are used in many branches
of engineering and the natural and social sciences. Now we
see use of eigenvectors in long term behavior of the
population distribution.
It is estimated that the number of people living in cities in
the United States during 2000 is 58 million. The number of
people living in the surrounding suburbs is 142 million.
 58 
Let us represent this information by the matrix 𝑋0 =   .
142 
Consider the population flow from cities to suburbs.
During 2000, the probability of a person staying in the city
was 0.96. Thus the probability of moving to the suburbs
was 0.04 (assuming that all those who moved went to the
suburbs). Consider now the reverse population flow, from
suburbia to city. The probability of a person moving to the
city was 0.01; the probability of remaining in suburbia was
0.99. These probabilities can be written as the elements of
a stochastic matrix P:
 (from) (to)
City Suburb

0.96 0.01 city

P  Suburb
0.04 0.99 

The probability of moving from location 𝐴 to location 𝐵 is

given by the element in column 𝐴 and row 𝐵 . In this
context, the stochastic matrix is called a matrix of
transition probabilities.
Now consider the population distribution in 2001, one year
later:
City population in 2001 = People who remained from 2000
+ People who moved in the from
the suburbs
= 0.96 × 58 + (0.01 × 142)
= 57.1 million
Suburban population in 2001 = People who moved in from
the city +People who stayed
from 2000
= 0.04 × 58 + (0.99 × 142)
= 142.9 million
Note that we can arrive at these numbers using matrix
multiplication:
 0.96 0.01  58   57.1 
0.04 0.99 142  142.9
    
Using 2000 as the base year, let 𝑋1 be the population in
2002, one year later. We can write
𝑋1 = 𝑃𝑋0
Assume that the population flow represented by the matrix
𝑃 is unchanged over the years. The population distribution
𝑋2 after 2 years is given by
𝑋2 = 𝑃𝑋1
After 3 years the population distribution is given by
𝑋3 = 𝑃𝑋2
After n years we get
𝑋𝑛 = 𝑃𝑋𝑛−1
The predictions of this model (to four decimal places) are

 58  City  57.1   56.245 

X0    X1    X2   
142  Suburb
142.9 143.755
 55.4327   54.6611 
X3    X4   
144.5672  145.3389 
and so on.
If the sequence 𝑋0 , 𝑋1 , 𝑋2 , … converges to some fixed vector X,
where 𝑃𝑋 = 𝑋. The population movement would then be in
a steady-state with the total city population and total
suburban population remaining constant thereafter. We
then write
𝑋0 , 𝑋1 , 𝑋2 , … → 𝑋
Since such a vector X satisfies 𝑃𝑋 = 𝑋 , it would be an
eigenvector of P corresponding the eigenvalue 1. Knowledge
of the existence and value of such a vector would give us
information about the long term behavior of the population
distribution.
Problem 1: Find the eigenvalue and eigenvectors of the
matrix
−4 −6
𝐴=
3 5
Solution: Let us first derive the characteristic polynomial
of A. We get
−4 −6 1 0 −4 − 𝜆 −6
𝐴 − 𝜆𝐼2 = −𝜆 =
3 5 0 1 3 5−𝜆
Note that the matrix 𝐴 − 𝜆𝐼2 is obtained by subtracting 𝜆
from the diagonal elements of A. The characteristic
polynomial of A is 𝐴 − 𝜆𝐼2 = −4 − 𝜆 5 − 𝜆 + 18 = 𝜆2 + 𝜆 −
2.
We now solve the characteristic equation of A.

𝜆2 + 𝜆 − 2 = 0
𝜆−2 𝜆+1 =0
𝜆 = 2 or -1
The eigenvalues of A are 2 and -1.
The corresponding eigenvectors are found by using these
values of 𝜆 in the equation 𝐴 − 𝜆𝐼2 𝑥 = 0. There are many
eigenvectors corresponding to each eigenvalue.
𝝀 = 𝟐: We solve the equation 𝐴 − 2𝐼2 𝑥 = 0 for x. The matrix
𝐴 − 2𝐼2 is obtained by subtracting 2 from the diagonal
elements of A. We get
−6 −6 𝑥1
=0
3 3 𝑥2
This leads to the system of equations
−6𝑥1 − 6𝑥2 = 0
3𝑥1 + 3𝑥2 = 0
giving 𝑥1 = −𝑥2 . The solutions to this system of equations
are 𝑥1 = −𝑟, 𝑥2 = 𝑟, where r is a scalar. Thus the
eigenvectors of A corresponding to 𝜆 = 2 are nonzero
vectors of the form
−1
𝑟
1
𝝀 = −𝟏: We solve the equation 𝐴 + 1𝐼2 𝑥 = 0 for x. The
matrix 𝐴 + 1𝐼2 is obtained by adding 1 to the diagonal
elements of A. We get
−3 −6 𝑥1
=0
3 6 𝑥2
This leads to the system of equations
−3𝑥1 − 6𝑥2 = 0
3𝑥1 + 3𝑥2 = 0
Thus 𝑥1 = −2𝑥2 . The solutions to these equations are
𝑥1 = −2𝑠, 𝑥2 = 𝑠, where s is a scalar. Thus the eigenvectors
of A corresponding to 𝜆 = −1 are nonzero vectors of the
form
−2
𝑠
1
Problem 2: Find the eigenvalues and eigenvectors of the
5 4 2
matrix 4 5 2
2 2 2
Solution: The matrix 𝐴 − 𝜆𝐼3 is obtained by subtracting 𝜆
from the diagonal elements of A. Thus
5−𝜆 4 2
𝐴 − 𝜆𝐼3 = 4 5−𝜆 2
2 2 2−𝜆
The characteristic polynomial of A is 𝐴 − 𝜆𝐼3 . Using row
and column operations to simplify determinants, we get
5−𝜆 4 2
𝐴 − 𝜆𝐼3 = 4 5−𝜆 2
2 2 2−𝜆
1−𝜆 −1 + 𝜆 0
= 4 5−𝜆 2
2 2 2−𝜆
1−𝜆 0 0
= 4 9−𝜆 2
2 4 2−𝜆
= 1−𝜆 9 − 𝜆 2 − 𝜆 − 8 = 1 − 𝜆 𝜆2 − 11𝜆 + 10
2
= 1 − 𝜆 𝜆 − 10 𝜆 − 1 = − 𝜆 − 10 𝜆 − 1
We now solve the characteristic equation of A:
2
= − 𝜆 − 10 𝜆 − 1 =0
𝜆 = 10 𝑜𝑟 1
The eigenvalues of A are 10 and 1.
The corresponding eigenvectors are found by using these
values of 𝜆 in the equation 𝐴 − 𝜆𝐼3 𝑥 = 0.
𝜆 = 10: We get
𝐴 − 10𝐼3 𝑥 = 0
−5 4 2 𝑥1
4 −5 2 𝑥2 = 0
2 2 −8 𝑥3
The solutions to this system of equations are 𝑥1 = 2𝑟, 𝑥2 =
2𝑟and 𝑥3 = 𝑟, where r is a scalar. Thus the eigenspace of
𝜆 = 10 is the one-dimensional space of vectors of the form
2
𝑟 2
1
𝜆 = 1: Let 𝜆 = 1 in 𝐴 − 𝜆𝐼3 𝑥 = 0. We get
𝐴 − 1𝐼3 𝑥 = 0
4 4 2 𝑥1
4 4 2 𝑥2 = 0
2 2 1 𝑥3
The solutions to this system of equations can be shown to
be 𝑥1 = −𝑠 − 𝑡, 𝑥2 = 𝑠, and 𝑥3 = 2𝑡, where s and t are scalars.
Thus the eigenspace of 𝜆 = 1 is the space of vectors of the
form
−𝑠 − 𝑡
𝑠
2𝑡
Separating the parameters s and t, we can write
−𝑠 − 𝑡 −1 −1
𝑠 =𝑠 1 +𝑡 0
2𝑡 0 2
Thus the eigenspace of 𝜆 = 1 is a two-dimensional subspace
of 𝑅 2 with basis
−1 −1
1 , 0
0 2
If an eigenvalue occurs as a k times repeated root of the
characteristic equation, we say that it is of multiplicity k.
Thus 𝜆 = 10 has multiplicity 1, while 𝜆 = 1 has multiplicity
2 in this problem.
Problem 3: Let A be an 𝑛 × 𝑛 matrix A with eigenvalues
𝜆1 , … . . 𝜆𝑛 and corresponding eigenvectors 𝑋1 , … . . 𝑋𝑛 . Prove
that if 𝑐 ≠ 0, then the eigenvalues of cA are 𝑐𝜆1 , … . . 𝑐𝜆𝑛 with
corresponding eigenvectors 𝑋1 , … . . 𝑋𝑛 .
Solution: Let 𝜆𝑖 be one of the eigenvalues of A with
corresponding eigenvector 𝑋𝑖 . Then 𝐴𝑋𝑖 = 𝜆𝑖 𝑋𝑖 . Multiply both
sides of this equation by c to get
𝑐𝐴𝑋𝑖 = 𝑐𝜆𝑖 𝑋𝑖 .
Thus 𝑐𝜆𝑖 is an eigenvalue of cA with corresponding
eigenvector 𝑋𝑖 .
3 10 5
Problem 4: If the given values of 𝐴 = −2 −3 −4 are
3 5 7
2, 2, 3 find the given values of 𝐴 and 𝐴 .
−1 2

Solution: Since 0 is not an eigen value of 𝐴, 𝐴 is a non

singular matrix and hence 𝐴−1 exists.
1 1 1
Eigen values of 𝐴−1 are , , and eigen values of 𝐴2 are
2 2 3
2 2 2
2 ,2 3 .
3 0 0
Problem 5: Find the eigen values of 𝐴 when 𝐴 = 5
5
4 0 .
3 6 1
Solution: The characteristic equation of 𝐴 is obviously
3 − 𝜆 94 − 𝜆) 1 − 𝜆 = 0.
Hence the eigen values of 𝐴 are 3, 4, 1.

∴ The eigen values of 𝐴5 are 35 , 45 , 15 .

Problem 6: Find the sum and product of the eigen values
3 −4 4
of the matrix 1 −2 4 without actually finding the eigen
1 −1 3
values.
3 −4 4
Solution: Let 𝐴 = 1 −2 4
1 −1 3
Sum of the eigen values = trace of 𝐴 = 3 + −2 + 3 = 4.
Product of the eigen values = 𝐴 .
3 −4 4
Now, 𝐴 = 1 −2 4 = 3 −6 + 4 + 4 3 − 4 − 4(−1 + 2)
1 −1 3
= −6 − 4 − 4 = −14.
∴ Product of the eigen values= −14.

EXERCISE
1. Determine the characteristic polynomials, eigenvalues,
and corresponding eigenspaces of the given 3 × 3 matrices.
3 2 −2 1 −2 2
𝑎. −3 −1 3 𝑏. −2 1 2
1 2 0 −2 0 3
15 7 −7
𝑐. −1 1 1
13 7 −5
2. Determine the characteristic polynomials, eigenvalues,
and corresponding eigenspaces of the matrix.
4 2 −2 2
1 3 1 −1
0 0 2 0
1 1 −3 5
3. Prove that if A is a diagonal matrix, then its eigenvalues
are the diagonal elements.

4. Prove that A and 𝐴𝑇 have the same eigenvalues.

5. Prove that λ = 0 is an eigenvalue of a matrix A if and only
if A is singular.
6. Prove that if the eigenvalues of a matrix A are
𝜆1 , … . . 𝜆𝑛 with corresponding eigenvectors 𝑋1 , … . . 𝑋𝑛 , then
𝜆1𝑚 , … . . 𝜆𝑚
𝑛 are eigenvalues of Am with corresponding
eigenvectors 𝑋1 , … . . 𝑋𝑛 .
7. Show that the following matrices satisfy their
characteristic equations.
0 2 8 −10 6 −8 −1 5
𝑎 . 𝑏 . 𝑐 . 𝑑 .
−1 3 5 −7 4 −6 −10 14

ANSWERS
1)
a. Characteristic polynomial −𝜆3 + 2𝜆2 + 𝜆 − 2 ;
eigenvalues 1,-1,2; corresponding eigenspaces
1   1   0 
r 0  , s  1 , t 1 
     
1   1  1 
b. Characteristic polynomial 1 − 𝜆 2 (3 − 𝜆) ; eigen
1 
values 1,3; corresponding eigenspaces r  1  ,
 
 2 
0 
s 1 
 
1 
 c. Characteristic polynomial 1 − 𝜆 2 − 𝜆 8 − 𝜆 ;
eigenvalues 1,2,8; corresponding eigenspaces
 1   0  1 
r  1 , s 1  , t 0 
     
 1  1  1 
2) Characteristic polynomial 2 − 𝜆 2 − 𝜆 4 − 𝜆 (6 − 𝜆) ;
eigenvalues 2,4,6; corresponding eigenspaces
 1   0   0  1 
 1 0   1  0 
r    s  , t   , p  
 0  1   0   0 
       
 0  1   1 1 