Eigen values and eigen vectors

Let A is a n × n matrix.
Suppose the linear transformation Y = AX transforms X into a scalar multiple of itself.
i.e. AX = Y = λ X (1)
Then the scalar λ is known as an eigen value of the matrix A and the corresponding non-zero
vector X is known as an eigen vector of the matrix A corresponding to the eigen value λ .
Thus the eigen values or characteristic values or proper values or latent roots are scalars λ which
satisfies the following equation
AX = λ X for X ≠ 0 (2)
⇒ AX − λ X = 0 ⇒ AX − λ IX = 0 ⇒ ( A − λI ) X = 0 (3)
which represents a system of n homogeneous equation in n variables x1 , x2 , x3 , .... , xn . This

sum of equations has non-trivial solutions if the coefficient matrix ( A − λ I ) is singular.

i.e. A − λ I = 0
a11 − λ a12 a13 … a1n
a21 a22 − λ a23 … a2 n
i.e. a31 a32 a33 − λ … a3n =0 (4)

an1 an 2 an3 … ann − λ

Expansion of this determent gives a nth degree of polynomial Pn ( λ ) which is known as

characteristic polynomial of A .
On expanding the determinants, the characteristics equation takes the form
( −1)n λ n + k1λ n−1 + k2λ n−2 + .... + kn−1λ + kn = 0 (5)
where k’s are expressible in terms of the elements aij , i, j = 1, 2, 3, ...., n .

The roots of the characteristic equation (5) are called the eigen values.

Notes:
1. If all the eigen values of a matrix A are different, then the corresponding eigen vectors are
linearly independent vectors.
 2. If the eigen values of a matrix is repeating, when two, or more eigen values are equal,
then may or may not possible to get the linearly independent eigen vectors corresponding
to the repeated roots. Hence the set of eigen vectors may be linearly independent or
linearly dependent.

3. Algebraic multiplicity of an eigen value λ is the order of the eigen value as a root of the
characteristic polynomial (i.e. if λ is a double root then algebraic multiplicity is two).

4. Geometric multiplicity of an eigen value λ is the number of linearly independent eigen

vectors corresponding to λ .

5. If X 1 and X 2 are two eigen vectors of a matrix A, corresponding to the same eigen value
λ of A, then any linear combination of the form k1 X1 + k2 X 2 is also gives eigen vector
of A corresponding to the same eigen value λ . Thus the eigen vector corresponding to
the eigen value is not unique but may be any one of the vectors cX i (c is a constant).

6. If λ1 , λ2 , λ3 , ...., λn are roots of the characteristic equation then

Trace( A) = λ1 + λ2 + λ3 + .... + λn , A = λ1 i λ2 i λ3 i .... i λn .

{
Spectrum A = λ1 , λ2 , λ3 , ...., λn , }
Spectral radius = λi , i = 1, 2, 3, ...., n , { }
λi = max λ j , j = 1, 2, 3, ...., n .

Remarks:

a) If matrix A is upper triangular or lower triangular or diagonal matrix then the diagonal
entries are the eigen values of the matrix A.

b) λ is an eigen value of a matrix A if and only if there exists a non-zero vector X such that
AX = λ X .

c) An eigen vector X of the equation AX = λ X , can’t correspond to more than one eigen
value λ of the matrix.

d) If X is an eigen vector of a matrix A, corresponding to an eigen value λ of A, and c is

any non-zero scalar, then cX is an eigen vector of A for the same eigen value λ .

e) If λ is an eigen value of A, then cX is an eigen vector of cA (c is an arbitrary constant).

A
f) If λ is an eigen value of a non-singular matrix A, then is an eigen value of adjoint of
λ
A.
 Properties of eigen values

1. Any square matrix A and its transpose A′ have the same eigen values.

We have ( A − λ I )′ = A′ − λ I ′ = A′ − λ I

( A − λ I )′ = A′ − λ I (∵ A′ = A)
A − λ I = A′ − λ I

∴ A − λ I = 0 ⇔ A′ − λ I = 0
i.e. λ is an eigen value of A if and only if it is an eigen value of A′ .
Hence the result.

Corollaries:
¾ The square matrix A and P −1 AP have the same characteristic roots, where P is a non-
singular square matrix of same order as A.
¾ If The A and B are two non-singular matrices, then the matrices AB and BA have the same
eigen values.
¾ If The A and B are square matrices and A is non-singular, then the matrices AB −1 and
BA−1 have the same eigen values.

2. The eigen values of a triangular matrix are just the diagonal elements of the matrix.

a11 a12 a13 … a1n

0 a22 a23 … a2 n
Let 0 0 a33 … a3n = 0 is a triangular matrix of order n.

0 0 0 … ann

( )
Then A − λ I = ( a11 − λ )( a22 − λ ) .... ann − λ .

Roots of A − λ I = 0 are λ = a11 , a22 , ...., ann .

Hence the eigen values of A are the diagonal elements of A. i.e. a11 , a22 , ...., ann .

Corollary: The eigen values of a diagonal matrix are just the diagonal elements of the matrix.
 3. The eigen values of an idempotent matrix are either zero or unity.

Let A be an idempotent matrix so that A2 = A .

If λ be an eigen value of A, then there exists a non-zero vector X such that AX = λ X .
We have AX = λ X (1)
Pre-multiply by A, A ( AX ) = A ( λ X )

i.e. A2 X = λ ( AX )
AX = λ ( λ X ) ⎡∵ A2 = A and AX = ( λ X ) ⎤
⎣ ⎦
AX = λ 2 X (2)

( )
From (1) and (2), we get λ 2 X = λ X or λ 2 − λ X = 0 or λ 2 − λ = 0 (∵ X ≠ 0)

∴ λ = 0 or λ = 1 .
Hence the result.

4. The sum of the eigen values of a matrix is the sum of the elements of the principal
diagonal.

Consider a square matrix A,

a11 a12 a13
A = a21 a22 a23
a31 a32 a33

a11 − λ a12 a13

So that A − λ I = a21 a22 − λ a23
a31 a32 a33 − λ

( )
On expanding, A − λ I = −λ 3 + λ 2 a11 + a22 + a33 − k1λ + k2 (3)
where k1 and k2 are expressible in terms of the elements aij , i, j = 1, 2, 3 .

If λ1 , λ2 , λ3 be the eigen values of A, then

(
A − λ I = ( −1) ( λ − λ1 )( λ − λ2 ) λ − λ3
3
) (4)
= −λ 3 + λ 2 ( λ1 + λ2 + λ3 ) − k3λ + k4
where k3 and k4 are expressible in terms of the elements λk , k = 1, 2, 3 .

Equating the RHS of (3) and (5), and comparing the coefficients of λ 2 , we get
λ1 + λ2 + λ3 = a11 + a22 + a33 . Hence the result.
 5. The product of the eigen values of a matrix A is equal to its determinant.

If λ1 , λ2 , λ3 be the eigen values of A, then

A − λ I = ( −1) ( λ − λ1 )( λ − λ2 ) λ − λ3
3
( ) (5)
= −λ 3 + λ 2 ( λ1 + λ2 + λ3 ) − k3λ + k4

Put If λ = 0 in equation (5), we get A = k4

where k4 is expressible in terms of the elements λk , k = 1, 2, 3 .
Hence the result.

Corollaries:
¾ If A is a singular matrix, then at least one of its eigen value is zero.
¾ If one of the eigen values of A is zero, then A = 0 . i.e. A is a singular matrix.
¾ If all the eigen values of A is non-zero, then A ≠ 0 . i.e. A is a non-singular matrix.

1
6. If λ is an eigen value of a matrix A, then is the eigen value of A−1 .
λ

If X be the eigen vector corresponding to λ , then AX = λ X (5)

Pre-multiplying both sides by A−1 , we get A−1 AX = A−1λ X
i.e. IX = λ A−1 X or (
X = λ A−1 X )
⎛1⎞
i.e. A−1 X = ⎜ ⎟ X
⎝λ⎠
1
This being of the same form as (5), shows that is an eigen value of the inverse matrix A−1 .
λ
Hence the result.

Corollary: If λ1, λ2 , λ3 , ...., λn are the eigen values of a matrix A, then A−1 has the eigen values
1 1 1 1
, , , ...., .
λ1 λ2 λ3 λ3
 1
7. If λ has an eigen value of an orthogonal matrix, then is also its eigen value.
λ

1
We know that if λ is an eigen value of an matrix A, then is an eigen value of A−1 [Property
λ
5].
Since A is an orthogonal matrix, A−1 is same as its transpose A′ .
1
∴ is an eigen value of A′ .
λ
But the matrices A and A′ have the same eigen values, since the determinants A − λ I and
A′ − λ I are the same.
1
Hence is an eigen value of A.
λ

Corollary: The matrices A and A′ have the same eigen values where A′ is the transpose of A.

8. If λ1 , λ2 , λ3 , ...., λn are the eigen values of a matrix A, then Am has the eigen values

λ1m , λ2m , λ3m , ...., λnm (m is a positive integer).

Let λi be the eigen value of A and X i the corresponding eigen vector. Then
AX i = λi X i (6)

(
Pre-multiply by A, we get A AX i = A λi X i) ( )
( ) ( )
⇒ A2 X i = λi AX i = λi λi X i = λi2 X i

Similarly, A3 X i = λi3 X i , . . .

In general, Am X i = λim X i which is of the same form as (6).

Hence λim is an value of Am .

The corresponding eigen vector is the same X i .