Further Notions on Matrices

Math 203
Lecture 22

Lecturers
Dr. Hasan Ibrahim
Dr. Mohamed Salman
Dr. Rami Younes
 Linear Transformations
Eigenvectors and Eigenvalues of a Matrix

Lecture Outline

1 Linear Transformations

2 Eigenvectors and Eigenvalues of a Matrix

Computing Eigenvalues
Computing Eigenvectors

Further Notions on Matrices

 Linear Transformations
Eigenvectors and Eigenvalues of a Matrix

Elementary Linear Algebra, Howard Anton and Chris Rorres, 10th

ed.:
section 4.9 on linear transformations
section 5.1 on eigenvalues and eigenvectors
(not to study in detail but to be surveyed for a solid background).

Further Notions on Matrices

 Linear Transformations
Eigenvectors and Eigenvalues of a Matrix

Linear Transformations
Recall that a function f is a rule that to each element a in a
domain A assigns an element b, denoted f (a), in a codomain B.

Example 2 3
2 3 1 5
The matrix A = 44 1 2 15 defines the function
5 1 4 0
Remark
Any n ⇥ m matrix A can be used to define a function T from Rm T : R4 ! R3 , with w = T (x1 , x2 , x3 , x4 ) = Ax:
into Rn as follows:

To each x in Rm , T assigns w = Ax in Rn

()

(ii)
(i)

w = T (x) For any x = (x1 , x2 , x3 , x4 ), w = (w1 , w2 , w3 ) given in figures (i)

and (ii). For example the image of the zero vector 0R4 is the zero
That is, if A = [aij ], the vector that T assigns to vector 0R3 and the image of the vector x = ( 1, 0, 2, 3) is
x = (x1 , x2 , . . . , xm ) is w = Ax = (w1 , w2 , . . . , wn ) given by: obtained as the product w = Ax = (15, 5, 3).

Further Notions on Matrices

 Linear Transformations
Eigenvectors and Eigenvalues of a Matrix

Linear Transformations
A Function T : Rm ! Rn that admits the following properties
(i) and (ii) is said to be a linear transformation:
Examples
(i) T (x + y) = T (x) + T (y), for all x and y
In what follows we survey few typical examples of linear
(ii) T (kx) = kT (x), for all scalar k and for all x. transformations. In each case, the matrix of the transformation in
the standard basis is given.
Remarks
1 Projection along the direction of v.
Linear transformations may be defined between any
For example, when v = e1 :
two vector spaces V and W , not only V = Rm and  
V = Rn . Not in this course, however. 1 0 x
T (x, y ) = (x, 0) =
A Function T : Rm ! Rn is a linear transformation if 0 0 y
and only if T (x) = Ax, where A is the n ⇥ m matrix
A = [T (e1 ) T (e2 ) . . . T (em )]
with e1 , e2 , . . . , em is the standard basis of Rm .
A is said to be the matrix of the linear transformation
T in the standard basis.
Proof 2 Reflection with respect to a line.
If x = (x1 , x2 , . . . , xm ) is in Rm : For example, when the line is the subspace y = x:
 
0 1 x
T (x, y ) = (y , x) =
T (x) = T (x1 e1 + x2 e2 + · · · + xm em ) 1 0 y

= T (x1 e1 ) + T (x2 e2 ) + · · · + T (xm em ) (by (i))

= x1 T (e1 ) + x2 T (e2 ) + · · · + xm T (em ) (by (ii))
2 3
x1
6 . 7
= [T (e1 ) T (e2 ) . . . T (em )] 6 7
4 .. 5 = Ax
xm
Further Notions on Matrices
 Linear Transformations
Eigenvectors and Eigenvalues of a Matrix

Linear Transformations
5 Rotation through an angle ✓.
3 Projection onto a plane.
For example, when the plane is the xz plane: T (x, y ) = (x cos ✓ y sin ✓, x sin ✓ + y cos ✓)
2 32 3  
1 0 0 x cos ✓ sin ✓ x
=
4
T (x, y , z) = (x, 0, z) = 0 0 0 5 4 y5 sin ✓ cos ✓ y
0 0 1 z

6 Rotation about an axis through an angle ✓.

For example, when the axis is the z axis:

4 Dilations and Contractions. T (x, y , z) = (x cos ✓ y sin ✓, x sin ✓ + y cos ✓, z)

  2 32 3
k 0 x cos ✓ sin ✓ 0 x
T (x, y ) = (kx, ky ) =
0 k y 4
= sin ✓ cos ✓ 0 5 4 y5
0 0 1 z

Contraction: 0  k < 1 Dilation: k > 1

Further Notions on Matrices

 Linear Transformations Computing Eigenvalues
Eigenvectors and Eigenvalues of a Matrix Computing Eigenvectors

Eigenvectors and Eigenvalues

Let A be a square matrix of order n. Eigenvalues
Eigenvectors An eigenvalue of A is any scalar such that there exists some
An eigenvector of A is a vector in Rn such that Ax = x for x 6= 0 in Rn such that Ax = x.
some scalar . Notice that at least one nonzero vector is required to satisfy the
Geometrically, an eigenvector of A is a vector x such that the equation Ax = x for to qualify as eigenvalue. This is to avoid
image of x by the linear transformation T (x) = Ax is parallel any scalar becoming eigenvalue, since A0 = 0 for any .
to x. Examples
(i) = 1 and = 0 are both eigenvalues of the projection
matrix A in the preceding example.

(ii) 3 0
= 3 is an eigenvector of A = since
8 1
 
1 1
A =3 (but how was = 3 obtained?).
2 2

Another perspective is that, an eigenvector defines a Remarks

direction along which multiplication by A amounts to (i) If is an eigenvalue of A then any solution of Ax = x is
multiplication by a scalar. called an eigenvector that corresponds to .
Example 2 3 (ii) is an eigenvalue of A if:
1 0 0
Ax = x for x 6= 0 () (A In )x = 0 for x 6= 0
The matrix A = 40 0 05 defines projection on the
0 0 1 (iii) (ii)=) is an eigenvalue of A if and only if:
xz plane (see example 3 on the preceding slide).
A In is singular () det(A In ) = 0
It follows that any vector in the xz plane is an eigenvector
of A (since the projection of any such vector is itself, that is, (iv) (ii)=) the eigenvectors that correspond to an eigenvalue
Ax = x). Furthermore, any vector orthogonal to the plane is actually constitute the subspace: null(A In ) 6= {0},
an eigenvector of A, since Ax = 0 = 0x. which is called the eigenspace of and is denoted E ,
(dim(E ) 1).

Further Notions on Matrices

 Linear Transformations Computing Eigenvalues
Eigenvectors and Eigenvalues of a Matrix Computing Eigenvectors

Computing Eigenvalues
Theorem 2 3
If A is an n ⇥ n matrix, then: 0 1 0
(i) PA ( ) = det(A In ) is a polynomial of degree n, called
2 The eigenvalues of A = 40 0 15 are the
the characteristic polynomial of A. 4 17 8
solutions of the characteristic equation
In PA ( ), the coefficient of n is ( 1)n , that of n 1
is
( 1)n 1 tr(A) and the constant term is det(A).
1 0
(ii) is an eigenvalue of A if and only if is a root of the PA ( ) = 0 () 0 1 =0
characteristic polynomial PA ( ), that is, a solution of the 4 17 8
equation
3 2
() +8 17 +4=0
PA ( ) = det(A In ) = 0,
known as the characteristic equation of A.
Using a calculator, it is found
p that the only p
eigenvalues of
Examples A are: = 4, = 2 + 3 or = 2 3.
 Notice that,
3 0
1 The characteristic polynomial of A = is
8 1 ( 1)3 1 tr(A) = tr(A) = 8 = coef. of 2 in PA ( )
det(A) = 4 = constant term in PA ( )

3 0 Remarks
PA ( ) = det(A I2 ) =
8 1 The diagonal terms of a triangular matrix are the
2 eigenvalues of the matrix (be the matrix upper
=( + 1)( 3) = 2 3 triangular, lower triangular or diagonal).

Therefore , the only eigenvalues of A are the roots of To find all eigenvectors of A:
PA ( ) = det(A I2 ): = 1 and = 3. (i) Find all eigenvalues
Notice that, (ii) For each eigenvalue , solve the linear system
2 1
( 1) tr(A) = tr(A) = 2 = coef. of in PA ( )
det(A) = 3 = constant term in PA ( ) (A In )x = 0 to find E = null(A In ).

Further Notions on Matrices

 Linear Transformations Computing Eigenvalues
Eigenvectors and Eigenvalues of a Matrix Computing Eigenvectors

Computing Eigenvectors
Example   
 0 0 x1 0
3 0 The system (2) is = , whose general
2 Find all eigenvectors of the matrix A = . 8 4 x2 0
8 1 
0.5
Solution solution is x = t .
1
The eigenvalues of A have been found to be = 1 and
= 3, (see the preceding slide).
E3 = {(0.5t, t)|t in R}
Eigenvectors corresponding to 1: E 1 .
=
= span{(0.5, 1)}
These are the vectors x = (x1 , x2 ) such that

Remarks
Ax = x () (A + I2 )x = 0 (1)
(i) {(0, 1)} is a basis of E 1 .
   (ii) {(0.5, 1)} is a basis of E3 .
4 0 x1 0
The system (1) is = , whose general (iii) The vectors (0, 1) and (0.5, 1) are eigenvectors that
8 0 x2 0
 correspond to distinct eigenvalues and they are
0
solution is x = . linearly independent.
t
Therefore, the eigenspace of = 1 is: (iv) {(0, 1), (0.5, 1)} is a basis of R2 that consists of
eigen vectors of A. In the next lecture, we will
explore the consequences of having such a basis.
E 1 = {(0, t)|t in R}
= span{(0, 1)}

Eigenvectors corresponding to = 3: E3 .
These are the vectors x = (x1 , x2 ) such that

Ax = 3x () (A 3I2 )x = 0 (2)

Further Notions on Matrices

 Linear Transformations Computing Eigenvalues
Eigenvectors and Eigenvalues of a Matrix Computing Eigenvectors

Computing Eigenvectors
Examples
2 32 3 2 3
3 Find bases for the eigenspaces of 1 0 2 x1 0
2 3 The system (3) is 4 1 1 15 4x2 5 = 405 whose
0 0 2 1 0 2 x3 0
A = 41 2 15 . general solution can be found to be
1 0 3
Solution x = ( 2s, s, s) = s( 2, 1, 1).
(i) Eigenvalues of A.
The eigenvalues of A are the roots of the It follows that E1 = span{( 2, 1, 1)} and a basis of the
characteristic polynomial eigenspace E1 is {( 2, 1, 1)}.

Eigensvectors corresponding to = 2: E2 .
PA ( ) = det(A I3 )
These are the vectors x = (x1 , x2 , x3 ) such that
0 2
= 1 2 1
Ax = 2x () (A 2I3 )x = 0 (4)
1 0 3

=
3
+5
2
8 +4 2 32 3 2 3
2 0 2 x1 0
2 The system (4) is 4 1 0 1 5 4 5 4
x2 = 05 whose
= (1 )(2 )
1 0 1 x3 0
general solution can be found to be
Use a calculator if factoring proves difficult. It
follows that the eigenvalues are = 1 and = 2. x = ( s, t, s) = s( 1, 0, 1) + t(0, 1, 0).
(ii) Eigenvectors of A.
Eigensvectors corresponding to = 1: E1 It follows that E2 = span{( 1, 0, 1), (0, 1, 0)} and a basis
These are the vectors x = (x1 , x2 , x3 ) such that of the eigenspace E2 is {( 1, 0, 1), (0, 1, 0)}.

Ax = x () (A I3 )x = 0 (3)

Further Notions on Matrices