Diagonalization of a Matrix:

Definition of diagonalization of a matrix:

D = P 1 AP .

Example:

Let

4 6
A =
5
.
3
Then,
1
2 0 1 2 4 6 1 2
D= = = P 1 AP,
0 1 1 1 3 5 1 1
where

2 0 1 2
D= , P =
1 1 1
.
0

Theorem:
An n n matrix A is diagonalizable if and only if it has n linearly
independent eigenvector.

Proof:

A is diagonalizable. Then, there exists a nonsingular matrix P and a

1
 1 0 L 0
0 2 L 0
D =
M M O M ,

0 0 L n
such that
D = P 1 AP AP = PD
1 0 L 0
0 L 0 .
A[col1 (P) col2 (P) L coln (P)] = [col1 (P) col2 (P) L coln (P)] 2

M M O M

0 0 L n

Then,
Acol j ( P) = j col j ( P), j = 1,2, K, n.
That is,
col 1 ( P ), col 2 ( P ), K , col n ( P )

are eigenvectors associated with the eigenvalues 1 , 2 ,K, n .

Let x1 , x 2 , K , x n be n linearly independent eigenvectors of A

associated with the eigenvalues 1 , 2 ,K, n . That is,

2
 1 0 L 0
0 2 L 0
D =
M M O M .

0 0 L n

Since Ax j = j x j ,

1 0 L 0
0 L 0
AP = A[x1 x2 L xn ] = [x1 x2 L xn ] 2
= PD
M M O M .

0 0 L n

Thus,

P 1 AP = P 1 PD = D ,

P 1 exists because x1 , x 2 , K , x n are linearly independent and thus P

Important result:
An n n matrix A is diagonalizable if all the roots of its characteristic
equation are real and distinct.

Example:

Let

4 6
A=
3 5 .
Find the nonsingular matrix P and the diagonal matrix D such that

D = P 1 AP
Solution:

3
We need to find the eigenvalues and eigenvectors of A first. The characteristic
equation of A is

+4 6
det(I A) = = ( + 1)( 2) = 0 .
3 5
= 1 or 2 .
By the above important result, A is diagonalizable. Then,

1. As = 2 ,

1
Ax = 2x (2I A)x = 0 x = r , r R.
1
2. As = 1 ,

2
Ax = x ( I A)x = 0 x = t , t R.
1
Thus,

1 2
1 and 1

are two linearly independent eigenvectors of A.

Let

1 2 2 0
P= D= .
0 1
and
1 1
Then, by the above theorem,

D = P 1 AP .

Note (very important):

D = P 1 AP ,

Note:
1
For any n n diagonalizable matrix A, D = P AP , then

4
 A k = PD k P 1 , k = 1, 2,K
where

1k 0 L 0

0 k2 L 0
D k
=
M M O M .

0 0 L kn

Example:

5 3
Is A = diagonalizable?
3 1

[solution:]

5 3
det( I A) = = ( 2 ) = 0 .
2

3 +1

Then, = 2, 2 .
As = 2,
1
(2 I A )x = 0 x = t , t R.
1
1
Therefore, all the eigenvectors are spanned by . There does not exist two linearly
1
independent eigenvectors. By the previous theorem, A is not diagonalizable.

Note:
An n n matrix may fail to be diagonalizable since

Not all roots of its characteristic equation are real numbers.

It does not have n linearly independent eigenvectors.

Note:
5
The set S j consisting of both all eigenvectors of an n n matrix A

associated with eigenvalue j and zero vector 0 is a subspace of R n . S j

is called the eigenspace associated with j .