Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
0% found this document useful (0 votes)
12 views

Module 4 Diagonalisation

The document discusses diagonalization of matrices. It provides definitions and properties of similar matrices and diagonalizable matrices. It presents a theorem on conditions for a matrix to be diagonalizable. Examples are given to demonstrate finding the eigenvalues and eigenvectors of a matrix and using them to diagonalize the matrix by obtaining a diagonal matrix that is similar to the original matrix. Procedures and steps involved in diagonalizing a matrix are also outlined.
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
12 views

Module 4 Diagonalisation

The document discusses diagonalization of matrices. It provides definitions and properties of similar matrices and diagonalizable matrices. It presents a theorem on conditions for a matrix to be diagonalizable. Examples are given to demonstrate finding the eigenvalues and eigenvectors of a matrix and using them to diagonalize the matrix by obtaining a diagonal matrix that is similar to the original matrix. Procedures and steps involved in diagonalizing a matrix are also outlined.
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 35

MODULE - 4

Matrix Theory

Section : 3 - Diagonalisation of a Matrix

Nobin Thomas Amal Jyothi College of Engineering


Similar Matrices

If A and B are square matrices, then we say that B is


similar to A if there is an invertible matrix P such that
B = P −1 AP .

Note that if B is similar to A, then it is also true that A is similar


to B since we can express A as A = Q−1 BQ by taking Q = P −1 .
This being the case, we will usually say that A and B are similar
matrices if either is similar to the other.

Nobin Thomas Amal Jyothi College of Engineering


Similar Matrices : Properties

Property Description
Determinant A and P −1 AP have the same determinant.
Invertibility A is invertible if and only if P −1 AP is invertible.
Rank A and P −1 AP have the same rank.
Nullity A and P −1 AP have the same nullity.
Trace A and P −1 AP have the same trace.
Characteristic A and P −1 AP have the same characteristic poly-
polynomial nomial.
Eigenvalues A and P −1 AP have the same eigenvalues.
Eigenspace If λ is an eigenvalue of A (and hence of P −1 AP )
dimension then the eigenspace of A corresponding to λ and
the eigenspace of P −1 AP corresponding to λ have
the same dimension.

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix

A square matrix A is called diagonalizable if it is similar to


a diagonal matrix, i.e., if there exists an invertible matrix
P and a diagonal matrix D such that P −1 AP = D

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation : Different Interpretations

⋄ A square matrix A is called diagonalizable if it is similar to


a diagonal matrix D.

⋄ A square matrix A of order n is diagonalizable if and only if


there exists a basis of eigenvectors for Rn .

⋄ A linear operator T : V → V is said to be diagonalizable if


and only if there exists a basis S of V such that the matrix
of T relative to this basis S is a diagonal matrix D.

⋄ If T is a linear operator on a finite dimensional vector space


V . Then T is is said to be diagonalizable if and only if there
exists a basis S of V such that each element of the basis S
is an eigenvector.

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix

Theorem
Let A be a square matrix with order n.

1. A is diagonalizable if and only if A has n linearly indepen-


dent eigenvectors.

2. A is diagonalizable if and only if the geometric multiplicity


of every eigenvalue is equal to the algebraic multiplicity.

3. If A has n distinct eigenvalues then A is diagonalizable.

Note:
Converse of 3 is not true: A is diagonalizable even though A has
same eigenvalues.

Nobin Thomas Amal Jyothi College of Engineering


A Procedure for Diagonalizing an n × n Matrix
Suppose A is the given matrix. We can diagonalize the matrix A
using the following steps:
1. Find out the eigenvalues and eigenvectors of the given ma-
trix.
2. Determine whether the matrix is actually diagonalizable or
not.
3. If the matrix is diagonalizable, then form the matrix P whose
column vectors are the n basis vectors you obtained in Step
1.
4. Find P −1 .
5. Then P −1 AP = D is the required diagonalized matrix. Here
D is a diagonal matrix whose successive diagonal entries are
the eigenvalues λ1 , λ2 , · · · , λn that correspond to the succes-
sive columns of P .
Nobin Thomas Amal Jyothi College of Engineering
Diagonalisation of a Matrix - Example 1

 
1 0
Diagonalize the matrix A =
1 2

The eigenvalues and eigenvectors are found by solving the eigen-


value probelm
AX = λX
Characteristic equation is given by

λ2 − trace(A)λ + det(A) = 0
trace(A) = sum of the main diagonal elements of A = 1 + 2 = 3
1 0
det(A) = = 1(2) − 1(0) = 2
1 2

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 1
λ2 − 3λ + 2 = 0
=⇒ λ = 1, 2

Here A has 2 distinct eigenvalues, so A is diagonalizable


When λ = 1

(A − λI)X = 0 =⇒ (A − I)X = 0
    
0 0 x 0
=
1 1 y 0
For finding the corresponding eigenvector we have to find out a
linearly independent equation.The row which contains zero only
is linearly dependant. Taking the second row:
 
x y 1
x = 1 y = −1 X1 =
1 1 −1

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 1
When λ = 2

(A − λI)X = 0 =⇒ (A − 2I)X = 0
    
−1 0 x 0
=
1 0 y 0
For finding the corresponding eigenvector we have to find out a
linearly independent equation. Taking the first row:
 
x y 0
x=0 y = −(−1) = 1 X2 =
−1 0 1

Form a matrix P by writing the eigenvectors as column vectors


which is known as the modal matrix
 
  1 0
P = X1 X2 =
−1 1

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 1

 
1 0
P −1 =
1 1
D = P −1 AP is the required diagonalized matrix. Here D is a
diagonal matrix whose main diagonal entries are the eigenvalues
of A.

D = P −1 AP

     
1 0 1 0 1 0 1 0
=
0 2 1 1 1 2 −1 1

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 2

 
−2 2 −3
Diagonalize the matrix A =  2 1 −6
−1 −2 0

The eigenvalues and eigenvectors are found by solving the eigen-


value probelm
AX = λX
Characteristic equation is given by

λ3 − trace(A)λ2 + (M11 + M22 + M33 )λ − det(A) = 0


trace(A) = sum of the main diagonal elements of A
= −2 + 1 + 0 = −1

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 2

1 −6 −2 −3 −2 2
M11 = = −12, M22 = = −3, M33 = = −6
−2 0 −1 0 2 1

M11 + M22 + M33 = −12 + −3 − 6 = −21


−2 2 −3
1 −6 2 −6 2 1
|A| = 2 1 −6 = −2 −2 −3 = 45
−2 0 −1 0 −1 −2
−1 −2 0
∴ Characteristic equation is

λ3 + λ2 − 21λ − 45= 0
=⇒ λ = 5, −3, −3

∴ the eigenvalues are 5,-3,-3

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 2
When λ = 5
1 ≤ GM(λ) ≤ AM(λ) =⇒ GM(5) = AM(5) = 1

(A − λI)X = 0 =⇒ (A − 5I)X = 0
    
−7 2 −3 x 0
 2 −4 −6 y  = 0
−1 −2 −5 z 0
For finding the corresponding eigenvector we have to find out two
linearly independent equations. Taking the first two rows.

x y z
−7 2 −3
2 −4 −6
2 −3 −7 −3 −7 2
x= = −24, y = − = −48, z = = 24
−4 −6 2 −6 2 −4
Nobin Thomas Amal Jyothi College of Engineering
Diagonalisation of a Matrix - Example 2
∴ the eigenvector corresponding to λ = 5 is given by
     
x −24 1
X1 = y  = −48 =  2 
z 24 −1

When λ = −3

(A − λI)X = 0 =⇒ (A + 2I)X = 0
    
1 2 −3 x 0
2 4 −6 y  = 0
1 2 −3 z 0
   
1 2 −3 1 2 −3
R → R2 − 2R1
E =  2 4 −6  2 ∼ 0 0 0 
R3 → R3 − R1
1 2 −3 0 0 0

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 2
GM(−3) = n − rank(A − 3I) = n − r = 3 − 1 = 2 = AM(−3).

GM(λ) = AM(λ) for every eigenvalue λ of A, so A is diagonal-


izable
Take y = s, z = t

x + 2y − 3z = 0 =⇒ x = 3z − 2y = 3t − 2s
       
x 3t − 2s 3 −2
X= y =
   s  =t 0 +s 1 
  
z t 1 0
∴ the eigenvectors corresponding to λ = −3 are given by
   
3 −2
X2 = 0 , X3 =  1 
1 0

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 2
Form a matrix P by writing the eigenvectors as column vectors
which is known as the modal matrix
 
  1 3 −2
P = X1 X2 X3 =  2 0 1 
−1 1 0
 
1 2 −3
1
P −1 =  1 2 5 
8
−2 4 6
D = P −1 AP is the required diagonalized matrix. Here D is a
diagonal matrix whose main diagonal entries are the eigenvalues
of A.
D = P −1 AP
     
5 0 0 1 2 −3 −2 2 −3 1 3 −2
0 −3 0  = 1  1 2 5   2 1 −6  2 0 1 
8
0 0 −3 −2 4 6 −1 −2 0 −1 1 0
Nobin Thomas Amal Jyothi College of Engineering
Diagonalisation of a Matrix - Example 3

Diagonalize the following matrix, if possible.


 
−3 −7 −5
2 4 3
1 2 2

The eigenvalues and eigenvectors are found by solving the eigen-


value problem
Ax = λx
Characteristic equation is given by
det(A − λI) = 0
that is

λ3 − trace(A)λ2 + (M11 + M22 + M33 )λ − det(A) = 0


Nobin Thomas Amal Jyothi College of Engineering
Diagonalisation of a Matrix - Example 3

trace(A) = sum of the main diagonal elements of A


= −3 + 4 + 2 = 3

4 3 −3 −5 −3 −7
M11 = = 2, M22 = = −1, M33 = =2
2 2 1 2 2 4
M11 + M22 + M33 = 2 − 1 + 2 = 3
−3 −7 −5
|A| = 2 4 3 =1
1 2 2
∴ Characteristic equation is
λ3 − 3λ2 + 3λ − 1= 0
=⇒ λ = 1, 1, 1
∴ the eigenvalues are 1,1,1
Nobin Thomas Amal Jyothi College of Engineering
Diagonalisation of a Matrix - Example 3
When λ = 1
1 ≤ GM(λ) ≤ AM(λ)
Here
AM(−1) = 3
Hence GM(1) is either 1 or 2 or 3

(A − λI)x = 0 =⇒ (A − I)x = 0
    
−4 −7 −5 x 0
2 3 3   y = 0
 
1 2 1 z 0
 
−4 −7 −5
E=  2 3 3  R1 ↔ R3
1 2 1

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Matrix - Example 3
 
1 2 1
R → R2 − 2R1
∼ 2 3 3  2
R3 → R3 + 4R1
−4 −7 −5
 
1 2 1
∼  0 −1 1  R3 → R3 + R2
0 1 −1
 
1 2 1
∼  0 −1 1 
0 0 0

GM(−3) = n − rank(A − 3I) = n − r = 3 − 2 = 1.

Here
GM(1) = 1 ̸= AM(1) = 3
So the given matrix is not diagonalizable.
Nobin Thomas Amal Jyothi College of Engineering
Computing Powers of a Matrix
The problem of computing powers of a matrix is greatly simplified when the
matrix is diagonalizable.

Suppose that A is a diagonalizable n × n matrix, such that P diagonalizes A


 k
λ1 0 · · · 0 λ1 0 ··· 0
  
 0 λ2 · · · 0  k
 0 λ2 · · · 0 
k
D= .  =⇒ D =  .
   
. . . .. .. .. 
 .. .. .. ..   .. . . . 
0 0 · · · λn 0 0 · · · λkn

D =P −1 AP =⇒ D2 = (P −1 AP )(P −1 AP ) = P −1 A(P P −1 )AP


= P −1 A2 P

λk1 0 ··· 0
 
0 λk2 ··· 0
Ak = P Dk P −1 ; where Dk =  .
 
.. .. .. 
 .. . . . 
0 0 ··· λkn

Nobin Thomas Amal Jyothi College of Engineering


Orthogonal Matrices

A square matrix A is said to be orthogonal if its transpose


is the same as its inverse, that is, if

A−1 = AT

or, equivalently, if

AAT = AT A = I

Nobin Thomas Amal Jyothi College of Engineering


Orthogonal Matrices : Examples

 
cos x − sin x
A=
sin x cos x
 
1/3 2/3 −2/3
B =  −2/3 2/3 1/3 
2/3 1/3 2/3

Nobin Thomas Amal Jyothi College of Engineering


Orthogonal Matrices : Properties

Theorem
The following are equivalent for an n × n matrix A.
(a) A is orthogonal.
(b) The row vectors of A form an orthonormal set in Rn with
the Euclidean inner product.
(c) The column vectors of A form an orthonormal set in Rn with
the Euclidean inner product.

Nobin Thomas Amal Jyothi College of Engineering


Symmetric Matrices

A square matrix A is symmetric if it is equal to its trans-


pose
A = AT

 
0 −7 −5
A = −7 4 3 symmetric
−5 3 2

 
3 −7 5
B = 4 4 3 nonsymmetric
5 3 2

Nobin Thomas Amal Jyothi College of Engineering


Eigenvalues of Symmetric Matrices

If A is an n × n symmetric matrix, then the following properties


are true
(1) A is diagonalizable
(2) All eigenvalues of A are real numbers
(3) Eigenvectors from different eigenspaces are orthogonal.

Nobin Thomas Amal Jyothi College of Engineering


Orthogonal Diagonalization

A matrix A is orthogonally diagonalizable if there exists an or-


thogonal matrix P such that P −1 AP = D is diagonal

Nobin Thomas Amal Jyothi College of Engineering


Fundamental Theorem of Symmetric Matrices

Theorem
An n × n matrix A is orthogonally diagonalizable if and only if
A is symmetric.

Nobin Thomas Amal Jyothi College of Engineering


A Procedure for Orthogonal Diagonalization of a Symmetric Matrix

Let A be a a symmetric matrix.


1. Find all eigenvalues of A and determine the algebraic mul-
tiplicity of each.
✽ eigenvectors corresponding to distinct eigenvalues are orthognoal

2. For each eigenvalue of algebraic multiplicity 1, choose the


unit eigenvector.
3. For each eigenvalue of the multiplicity to be k ≥ 2, find
a set of k linearly independent eigenvectors. If this set
{v1 , v2 , · · · , vk } is not orthonormal, apply the Gram-Schmidt
orthonormalization process.
4. The composite of steps (2) and (3) produces an orthonor-
mal set of n eigenvectors. Use these orthonormal and thus
linearly independent eigenvectors as column vectors to form
the matrix P .
5. Then P T AP = D is the required diagonalized matrix.
Nobin Thomas Amal Jyothi College of Engineering
Diagonalization of a Symmetric Matrix - Example 1

Find an orthogonal matrix P that diagonalizes


 
3 −2 4
A = −2 6 2
4 2 3

Characteristic equation is given by

λ3 − trace(A)λ2 + (M11 + M22 + M33 )λ − det(A) = 0


λ3 − 12λ2 + 21λ + 98= 0
=⇒ λ = −2, 7, 7
∴ the eigenvalues are -2,7,7

Nobin Thomas Amal Jyothi College of Engineering


Diagonalization of a Symmetric Matrix - Example 1
The usual calculations produce bases for the eigenspaces:
   1  
1 −2 −1
λ = 7 : v1 = 0 v2 = 1 , λ = −2 : v3 = − 21 
    
1 0 1

⟨v1 , v2 ⟩ = v1 · v2 = (1, 0, 1) · (−1/2, 1, 0) = −1/2


Although v1 andv2 are linearly independent, they are not or-
thogonal.
Using Gram–Schmidt process we can find a vector

⟨v2 , v1 ⟩
z = v2 − v1
∥v1 ∥2

such that {v1 , z} form an orthogonal basis for the eigenspace


corresponding to λ = 7.
Nobin Thomas Amal Jyothi College of Engineering
Diagonalization of a Symmetric Matrix - Example 1

⟨v2 , v1 ⟩
z = v2 − v1
∥v1 ∥2
 1  
−2 1
−1/2
= 1 − 2
 0
1 + 02 + 12
0 1
 1  
−2 1
1
=  1  + 0
4
0 1
 1
−4
= 1 
1
4

Normalize v1 and z to obtain the following orthonormal basis for


the eigenspace for λ = 7.
Nobin Thomas Amal Jyothi College of Engineering
Diagonalization of a Symmetric Matrix - Example 1

 
− √12
v1  
u1 = = 1 
∥v1 ∥ 
1


2
 
− √118
z  4 
u2 = = √ 
∥z∥  18 
√1
18

An orthonormal basis for the eigenspace for λ = −2 is


 2
−3
v3  1
u3 = = − 3 
∥v3 ∥ 2
3

Nobin Thomas Amal Jyothi College of Engineering


Diagonalisation of a Symmetric Matrix - Example 1

 
− √12 − √118 − 32
h i  
P = u1 u2 u3 =  1 √4 − 31 
 
 18 
√1 √1 2
2 18 3

P −1 = P T
D = P T AP is the required diagonalized matrix.
D = P T AP
  √1 T    √1 
− 2 − √118 − 32 − 2 − √118 − 23

7 0 0 3 −2 4
0 7 0 = 1 √4 − 31 
 −2 6 2  1 √4 − 13 
 
18 18 
1 √1 2 √1 √1 2
0 0 −2 √
3
4 2 3 3
2 18 2 18

Nobin Thomas Amal Jyothi College of Engineering

You might also like