8.

Linear Transformation
Learning Objectives

After finishing this module, students should be able to:

1. Identify linear transformations,

2. Determine the standard matrix for a linear transformation,
3. Explain the properties of linear transformation in planes and
spaces (R2 and R3),
4. Explain eigenspace as a Kernel for a linear transformation T
5. Explain the meaning behind an eigenvalue of [T] geometrically,
6. Explain the similarity invariance property.
Scope
Linear Transformation
(Definition)

Linear Transformation in
Euclidean Space

Composition of Linear
Transformations

Kernel and range

Inverse of a linear
transformation

Similarity
Pre-test module
Pre-test
• Which one of these diagrams describes a function?
a. b. c.

• Which one of these diagrams describes an injective function(one to one)?

a. b. c.

• Which one of these diagrams describes a surjective function(onto)?

a. b. c.

• Which one of these diagrams describes a bijective function?

a. b. c.
 Linear Transformation (Definition)

Linear Transformation in Euclidean

Space

Composition of Linear Transformations

Kernel and range

Inverse of a Linear Transformation

Similarity

8.1 Definition of Linear

Transformation
Function

Definition 8.1: Function

Function is a rule f that associates with each element in a set D one and
only one element in a set C.

Domain Codomain

Notation: f: D  K
f: x f(x)
Function


Function

Not a function


Not a function
Linear Transformation

Definition 8.2. : Linear Transformation

If T: V  W is a function from vector space V into a vector space W, then T is
called a linear transformation from V to W if, for all vectors u, v in V and all
scalars k satisfy:
1. T u  v   T (u )  T ( v )
2. T  k u   kT (u )
If V = W, then T is called a linear operator on V.

u T(u)
V W
v T(v)
u+v T(u+v) = T(u) + T(v)
kv T(kv) = k.T(v)
Exercise

 2a  b   ab   b2 
a   a   a  
f1 :   0 f2 :   0 f3 :   b
     
b  9a  b  9a  b  9 a  2b 
     

 2a  b   ab 
a   a  
f4 :   5 f5 :   b
   
b  9a  b  3 a  0.9 b 
   
Example 1: Linear Transformation
y
T(a)
1.

a T maps vector a to T(a)

x

T(b) b

y y
T(a) T(a)
2. 3.

a a
-2a x T(a+b) = T(a) + T(b)
b

T(b)
T(-2a)=-2T(a)
Example 2: the zero transformation and
the identity operator
a. Let V and W be any two vector spaces; the linear transformation is defined as
T0: V  W where T0(v) = 0 for every v element of V

T0  u   v   0
0  0
  T0 u    T0  v 

T0: V  W is a linear transformation and is called as the zero transformation

b. The following transformation is called the identity operator in V.

T1 : V  V
T1  v   v For every v element of V
Example 3: Linear Transformation

Linear transformation from P2 to P3

for a vector v  a 0  a1 x  a 2 x  P
2 2
T: P2  P3

a transform ation is defined as T  v   a1 x  a 2 x

2 3

T  k v   ka 1 x  ka 2 x
2 3

 k a1 x  a2 x 
2 3

 kT  v 

T  v  w   a1  b1  x  a2  b 2  x
2 3

2 2 3 3
 a1 x  b1 x  a2 x  b 2 x
 a1 x  a2 x   b 
2 3 2 3
1 x  b2 x
 T u   T  v 
Example 4: a nonlinear transformation

A function is defined from R2 to R3 as follows

T: R2  R3

T : v1 , v 2   v1 , v 2 ,5 

T  v  w   v 1  w 1 , v 2  w 2 ,5 
T  v   T  w   v 1  w 1 , v 2  w 2 ,10 
T v  w   T v   T w 

T is not a linear transformation.

Properties of Linear Transformation

Theorem 8.1. : Properties of Linear Transformation

1. T(0) = 0
2. T(-a) = - T(a)
3. T(a –b) = T(a) – T(b)

V W
0 0
-a -T(a)

a-b T(a-b) = T(a) - T(b)

 Linear Transformation (Definition)

Linear Transformation in Euclidean

Space

Composition in Linear Transformation

Kernel and range

Inverse of a Linear Transformation

Similarity

8.2 Linear Transformation in

Euclidean Space
Linear Transformation in Euclidean Space
Rn Rm

T(x) =Ax
x Ax
Linear Transformation
R R

Every linear transformation from a Euclidean space to another Euclidean

space can always be expressed as a matrix multiplication, this matrix is
called the standard matrix
If m = n, then T is called linear operator.
Notation: T:Rn  Rm
TA(x) = Ax
[TA] = [T] = A standard matrix of transformation T
Projection onto x-axis
y

 a1 
 
 a2 
a  a1 
 
T(a)  0 
x

 a1 
a    Standard
 a2  Matrix

 a1   a1   1 .a1  0 .a 2   1 0   a1 
T             
 a 2   0   0 .a1  0 .a 2   0 0  a 2 
Example 1: projection onto y-axis
1. Find the projection of vector (x, y) onto the y-axis.
2. Write down each component as the linear combination of x and y.
3. Find the standard matrix of the linear operator that projects vectors
onto the y-axis.
4. Find all eigenvalues of the standard matrix.
Answers:
(1) (x, y) is projected to (0, y)

 x   0   0. x  0. y 
(2) T        
y
    y 0 . x  1 . y 
0 0
(3) standard matrix :  
0 1
(4) Eigenvalues of [T] are 1 and 0
Eigenvalues and Eigenvectors of Linear
Operators

Definition 8.3.: Eigenvalues and Eigenvectors of Linear Operators

If T: Rm  Rn is a linear operator, then λ is the eigenvalue of T if there is a
nonzero vector x such that T(x) = λx. This nonzero vector is called the
eigenvector of T that corresponds to λ.

Rn Rn

T(a) =Aa
x Ax=λx
Linear operator

R R

x eigenvector
λ eigenvalue
Example 5: linear transformation
5 6
Let f:R2  R2 where f(x) = Ax and A   
3  4 

Find the linear transformation of the vectors x, y, and z.

 4  1   2
x    y    z   
2 3 2
 5  6  4   8   5  6  1    13   5  6  2    2 
f  x         f  y         f  z        
3  4  2   4  3  4  3    9  3  4  2    2 

y  y
y y

z
x

x

Ax Az

x 
Ay
x
x eigenvector of A y not an eigenvector of A z eigenvector of A
Eigenvalue of the standard matrix that
project vectors on the x-axis

v
1 0
standard matrix  T    
0 0
u

Eigenvalues of [T]
b = T(b) a = T(a) x
are 1 and 0

For every horizontal vector a, then T(a) = 1.a.

For every vertical vector u, then T(u) = 0.u = 0
The eigenvalue of T is the same as the eigenvalue of [T], i.e. 1 and 0. Eigenvectors
of T that corresponds to 1 are nonzero horizontal vectors. Nonzero vertical vectors
are eigenvectors of T that corresponds to 0.
Expansion and Compression
In the x-direction In the y-direction
y y
(x,ky)

T(a)
(x,y) (kx,y) (x,y)
a a
T(a)
x x

k 0 1 0
standard matrix Standard matrix 0
0 1   k 


Expansion: k >1; Compression 0 <k<1

 Dilation and Contraction

If k is a nonzero scalar, then the operator T(x) = kx in R2 or R3 is called a

contraction with factor k (if 0 ≤ k ≤1) and dilation with factor k if k ≥ 1.

y y

T(a)
a
a T(a)

x x
k≥1 0≤k≤1
The eigenvalue of T is k, every nonzero vector in R2 are eigenvectors.
The determinant of a Linear Operator

Definition 8.4.: The determinant of a linear operator

If T: Rm  Rn is a linear operator with the standard matrix [T] then the
determinant of the linear operator T is
det(T) = det([T])

Example 6:
a. T is the orthogonal projection on the x - axis where the
1 0
standard matrix is ; T     , then det( T )  0
0 0

k 0
b. the determinan t of dilation w ith factor k in R is k  det 
2 2

0 k
Reflection about the x-axis in R2

 x
a 
y Standard
Matrix

  x   x  x  1 . x  0 . y  1 0 x
T a     T     
  y 
 y 
 
 y 
  0 . x  1 . y  0  1   y 

Eigenvalue of T are 1 and -1.

Determinant of T is -1
Reflection about the y-axis in R2
y

T(ā) ā

x
 x  x   1.x  0. y   1 0  x 
T    y  
 y 
   
 0.x  1. y 
 0 1   y 
 

 1 0
Standard matrix = [T ]   det (T) = -1
 0 1 

Eigenvalues of T: -1 and 1. Eigenvectors that corresponds to -1 are

every nonzero horizontal vectors. Eigenvectors that corresponds to 1
are every nonzero vertical vectors.
Reflection about y=x in R2

y
y=x
T(a)
 x y 0 . x  1 . y  0 1x
T    1 . x  0 . y   1
 y 
 
x
     0   y 
a

Eigenvalues of T are 1 and -1. Vectors (a, a) where a ≠0 are eigenvectors

that corresponds to 1.
a. Find eigenvectors that corresponds to -1.
b. Which vectors are mapped to the zero vector?
c. Find det(T)
Reflection about y = -x in R2

y = -x y

x
T(a)

 x   y   0.x  1. y  0  1 x 

T      

 y    1 0   y 
    x   1.x  0. y 
 0  1
standard matrix : T    
Det (T) = -1
  1 0  Eigenvalues of T are 1 and -1
 Reflection about the xy-plane in R3
z
 x  x   1.x  0. y  0.z 
     
T   y     y    0.x  1. y  0.z 
a   
 z

 z 
 0 . x  0 . y   1 . z 


1 0 0 x
y   
 0 1 0 y

0 0  1
z

T(a)
1 0 0 
x  
standard matrix  T   0 1 0
 
 0 0  1
a. Find eigenvectors that corresponds to eigenvalues of T.
b. Which vectors are mapped to the zero vector?
c. Find det(T)
Rotation in R2

• Rotation through an angle α, counterclockwise

 x  x*  y
• Equation T    * T(a) =(x*, y*)
 y 
  y 

x  r cos  , y  r sin 
α a=(x, y)
x  r cos     , y  r sin     
* *

β
x  r cos  cos   r sin  sin 
*
x
y  r cos  sin   r sin  cos 
*

 x  x *   r cos  cos   r sin  sin    x cos   y sin    cos   sin    x 

T    *      x sin   y cos   
 y  y r cos  sin   r sin  cos     sin  cos    y 
     
 cos   sin  
standard matrix  T     [T] is an orthogonal matrix
 sin  cos  
Rotation about the positive x-axis in R3

• Counterclockwise rotation through an angle α

z

(x*,y*,z*) y
α Standard
Matrix
(x,y,z)
x

 x  x  1 0 0 x
      
T   y     y cos   z sin     0 cos   sin    y 
     0 sin  cos    z 
  z    y sin   z cos  
 Finding the Standard Matrix
Theorem 8.2: If T : Rn  Rm then T(x) = [T]x (T can be done by multiplying
with matrix [T]). The standard matrix [T] is obtained through
[T] = [ T(e1) T(e2) ……T(en) ]

i.e. the matrix where the columns are the result of mapping the vectors of
the normal standard basis.
u1 
   2u 1  3 u 2 
Example 7: T : R3  R2 where T : u 2    
 4u3 
u 3 

2 .1  3 .0   2  3  0 
T (i)     T ( j)    T (k )   
 0  4 
 4 .0  0 
2 3 0
Hence, [T ]  
0 0 4 
Shear in the x and y-direction

in the x-direction with factor k

y

(x,y) (x + ky,y)
a
T(a)
x

1 k
standard matrix  
0 1

How do you obtain the standard matrix of shear in the y-direction?

Earth’s Rotation

• The earth rotates on its axis, which is an axis that goes through the north and
south pole. Because the sun rises from the east and sets to the west, then the
earth rotates from west to east. From the south pole perspective, the earth has
a clockwise rotation, but seeing this from the north pole, it seems that the earth
rotates counterclockwise.
 Linear Transformation (Definition)

Linear Transformation in Euclidean

Space

Composition of Linear Transformations

Kernel and range

Inverse of a Linear Transformation

Similarity

8.3 Composition of Linear

Transformations
Composition of Linear Transformations

Theorem 8.3 : Composition of Linear Transformations

If T1: V  U and T2: U  W are linear transformations, then the
composition of T2 with T1 is denoted by T2 o T1 (“T2 circle T1”) and it is a
linear transformation defined by the formula (T2oT1)(v) = T2(T1(v)), where v
is a vector in V

V U W

v T1(v) T2(T1(v))
Example 8: Composition of Two Linear
Transformations

2 3  a 2
T1 : R  P ; T1 :    a  bx  ( a  2 b ) x
b
c 0 c1 
T2 : P  M ; T2 : c 0  c1 x  c 2 x
3 2x2 2
 
0 c1  2 c 2 
2 2x2
T2  T1 : R  M ;
 a 2 a b 
T2  T1 :    a  bx  ( a  2 b ) x  
b 0 a  2( a  2 b ) 

2 2 1  2 1 
T2  T1    0 
1   2  2(2  2 .1 )  0 10 

Example 9: Composition of Two Rotations
Rotation followed by another rotation
y
T2(T1(a))

α T1(a)
β
x
a

Is T2(T1(a)) = T1(T2(a)) ?
Example 10: Composition of Rotation and
Projection
a. Rotation followed by a projection b. Projection followed by a rotation

y y

T2(T1(a))
T1(a) a
β β T1(a)
T2(T1(a)) a x x

Conclusion:
Composition of linear transformations is not commutative
Composition of 3 Linear Transformations

Given three linear transformations T1, T2, and T3.

Domain of T3 is the range of T2.
Domain of T2 is the range of T1.
V U W X

v T1(v) T2(T1(v)) T3(T2(T1(v)))

Standard Matrix of Composition of Linear
Transformations in Euclidean Space
y  a1 
a   
 a2 
T2(T1(a)) T1 orthogonal projection on x-axis
T2 reflection about the line y = x
a
β T1(a) 1 0 0 1
T1     T2    
x
0 0  1 0

The standard matrix of T2oT1 is obtained by:

  a1     a1   0   a1    0 1  1 0   a1   0 0   a1 
T2  T1    T2     
  a  T2  T1          
 a   0 
 2    1  a 
  2   1 0  0 0   a2   1 0   a2 

0 1  1 0 0 0
 T2  T1        
1 0  0 0  1 0
Standard Matrix of Composition of Linear
Transformations in Euclidean Space

Theorem 8.4 : Composition of Linear Transformations

If T1: Rn  Rm and T2: Rm  Rk, then [T2oT1] = [T2][T1].

V U W

v T1(v) =[T1] v T2(T1(v))= [T2][T1]v

Standard Matrix Projection-Rotation

Projection followed by a rotation

y

a
β T1(a)
x

1 0  cos   sin  
T1     T2   
0 0  sin  cos  

 cos   sin   1 0   cos  0

[T2 o T1]=    
 sin  cos   0  0 0 
 0
 Linear Transformation (Definition)

Linear Transformation in Euclidean

Space

Composition of Linear Transformations

Kernel and range

Inverse of a Linear Transformation

Similarity

8.4 Kernel and range

Kernel and range

Definition 8.6: Kernel and range

If T: V  W is a linear transformation, then the set of vectors in V that is
mapped to 0 is called as the kernel of T (denoted by ker(T)).
The set of vectors in W that are images of vectors in V is called as the
range of T (denoted by R(T))

Ker(T) 0 R(T)
Rank and Nullity

Definition 8.7: rank and nullity

If T: V  W is a linear transformation, then the dimension of range T is
rank(T) and the dimension of kernel T is nullity(T).

Ker(T) 0 R(T)
Example 11: kernel, range, nullity, and
rank
1. Zero transformation T1: V  W, where T(v) = 0 for every v in V

V T1 W

Ker(T1) 0 R(T1)

Ker(T1) = V, nullity(T1) = dim(V)

R(T1) = {0}, rank(T1) = 0
Example 12: kernel, range, nullity, and rank
2. The identity operator T2 (v) = v

V W
T2
a T(a) = a
Ker(T2) 0 0

Ker(T2) = {0}, nullity(T2 ) = 0

R(T2) = V, rank(T2) = dim (V)
Example 13: kernel, range, nullity, and rank

3. Transformation T3:R2  R2 projection on the x-axis

y
L1

L2 x
R(T3)
Ker(T3)

Ker(T3) = L1, nullity(T3) = 1

R(T3) = L2, rank(T3) = 1
Exercise 2: Kernel and range shear in the x-
direction

(x,y) (x +ky, y)
a
T(a)
x

(x +ky, y) = (0, 0) if and only if x = 0 and y = 0, then Ker (T) = {0}

Exercise:
1. Find range and rank T
2. Find eigenvalues and eigenvectors of T
Answers to Exercise 2

1. R(T) = R2

y
b = T(b)

Standard matrix

T(a) = ka k 0
a 0 1
x

2. Eigenvalues of T are k and 1. Eigenvectors that corresponds to k are

nonzero horizontal vectors. Nonzero vertical vectors are eigenvectors
that corresponds to 1.
 Linear Transformation (Definition)

Linear Transformation in Euclidean

Space

Composition of Linear Transformations

Kernel and range

Inverse of a Linear Transformation

Similarity

8.5 Inverse of a Linear

Transformation
One-to-One Linear Transformation

Definition 8.6.: one-to-one linear transformation

T: V W is a one-to-one linear transformation if distinct vectors in V are
mapped to distinct vectors in W

Example 14: one-to-one linear transformation

• rotation
• reflection
• Identity operator

Orthogonal projection on the x-axis is not a one-to-one linear operator

Inverse of Linear Transformations
V T W

T-1

T: V  W is a bijective linear transformation (one-to-one and onto)

T-1: W  V is the inverse of T and is also a linear transformation

T-1(T(v)) = T-1(w) = v
T(T-1(w)) = T(v) = w

Note: If it is a one-to-one function but not onto (surjective) then we can define
the inverse where its domain is the range of the original linear
transformation:
T-1: R(T)  V
Example 15: Inverse of Linear
Transformations
y
T(a) a = T-1(T(a))
  x    y 
T : R  R where T       x 
2 2
 y  
     2 

 0  1
standard matrix T :  1 
 2 0

Are inverse to one another
1 
0 2
standard matrix T :  
 1 0

  x  0 2  y   x 
 x 
1    0 2  x   2 y  T
1 T     x  
T     1    y  1 0  
 y 
   0   y    x       y
 2   
Properties of One-to-one Linear
Transformations
• If T: Rn  Rm has a standard matrix [T] that has an inverse, then [T] x =
b is consistent with exactly one solution for every b in Rn.
• [T] x = b is consistent, then for every b in Rn there is an x in Rn such that
[T]x = T(x) = b. R(T) = Rn.
• Solution of [T] x = b is exactly one, then T is one-to-one.
Properties of One-to-one Linear
Transformations

Theorem 8.5. :
If A is a nxn matrix that has an inverse and is a standard matrix for the
linear operator TA : Rn  Rn then the following are equivalent.
1. A has inverse
2. R(TA) = Rn
3. Ker(TA) = {0}
4. TA is one-to-one

• Every one-to-one linear operator (injective) has an invertible standard

matrix . E.g. rotation, dilation.
• A one-to-one function can be not onto (surjective), but a one-to-one
linear operator is also onto (surjective and injective).
 Isomorphism
Definition 8.7.:
Isomorphism is a linear transformation that is both one-to-one (injective) and onto
(surjective). If there is an isomorphism from vector space V to W (or vice versa),
then it is said that V and W are isomorphic.

 a 
 
 b 
Example 16: T: R4 P3 where T      ax  bx  cx  d
3 2

c 
 
 d 
 
R4 P3
a
b 
 
c 
 
d  ≈
−
3 2
ax  bx  cx  d

Isomorphism has an inverse and that inverse is also an isomorphism

Example 12: Isomorphism

  a1 a2 a3  
T : M 2 X 3  R , where T      a1 b3 
6
a2 a3 b1 b2
 b
 1 b2 b3  

M2x3 R6

 a1
b
 1
a2
b2
a3 
b3  ≈
− a1 a2 a3 b1 b2 b3 

R R

M2x3 and R6 are isomorphic

Exercise 3

Answer with TRUE or FALSE:

1. Every one-to-one linear transformation is an isomorphism.
2. T : R  R w here T  x   2 x is a linear tra nsformatio n.
3. Projection on the x-axis is an isomorphism.
4. T : R 2  R 3 w here T a , b   0 , a , b  is an isomorphis m.

a b
5. T : P  M 2 X 2
3

w here T ax  bx  cx  d  
3 2
  is an isomorphis m.
c d

Answers:
1. False, 2. True 3. False, 4. False, 5. True
 Linear Transformation (Definition)

Linear Transformation in Euclidean

Space

Composition of Linear Transformations

Kernel and range

Inverse of a Linear Transformation

Similarity

8.6 Similarity
Two Similar Matrices

Definition 8.8.:
If A and B are two nxn matrices, then B is similar to A if there is an invertible matrix
P such that B = P-1AP

If B is similar to A, then A is similar to B because A = PBP-1 and we simply say that A

and B are similar.

Example 17:

1 1 1 2  14 0 1 4 0
A B P 1
P 
3  3  0 0 2 
2 1 2

A = PBP-1
A and B are similar
Example 18: two similar matrices

The following matrices A and D are similar.

5 6 2 1 1  1 1
A    P    P   
3  4  2 
 1 1  1
Diagonal matrix

 1  1  5  6  2 1 2 0 
    
1 2  3  4  1 1   0  1 
P-1 A P D

A is diagonalizable to D, then A is similar to D

64
Example 19: two similar matrices

The following matrices A and B are similar

1 2
A   
3 4
1 1 1 0  1 2  1 0
B          
1 
6 4 0 2  3 4  0 2 

Note that det(A) = det(B) = -2; trace(A) =trace(B) = 5

Similarity Invariant
Two similar matrices will have several things in common, one of these
things are their determinants.

e.g A and B are similar, then there is matrix P such that B = P-1AP

1
det( B )  det( P AP )
1
 det( P ) det( A) det( P )
1
 det( P ) det( P ) det( A)
 det( A)

Furthermore, two similar matrices have other things in common..

 Similarity Invariants
Property Description

Determinant Det(A) = det(P-1AP)

Invertibility A is invertible if and only if P-1AP is invertible

Rank dan nullity Rank(A) = rank(P-1AP); Nullity(A) = nullity(P-1AP);

Trace Trace(A) = trace(P-1AP)

Characteristic A and P-1AP have the same characteristic polynomial

Polynomial
Eigenvalues A and P-1AP have the same eigenvalues
(may have different eigenvectors)
 Key Concepts
Summarize the material you just learned.

Check your summary, does it contain all of the following key concepts?

 Definition of linear transformation (LT)  Nullity and rank of LT

 Zero transformation  Inverse of LT
 Eigenvalues and eigenvectors of
 The identity transformation
LTs
 The standard matrix  General LT
 The method to find the standard matrix  Standard matrix of general LT
 LT in the Euclidean space  Determinant of an LT
 Properties of LT  Isomorphism
 Composition of LTs  similarity
 kernel and range
Post-test module

Copyright © 2004-2005 NameOfTheOrganization. All rights reserved.

 Post-test

A Answer whether it is TRUE or FALSE

1. Projection on the x-axis is a one-to-one linear operator.
2. T is one-to-one if and only if nullity(T) = 0.
3. Reflection about the y = x line is a one-to-one linear transformation
4. Rotasi through an angle α is a one-to-one linear transformation.
5. If two matrices are similar, then they have the same eigenvalues.
6. If two matrices have the same determinant and trace, then they are similar.

B. What is the relation between Ker(T), Null(T), Null([T]); and the relation between
Rank([T]), R (T) and Coll([T]), if T is a linear operator from a Euclidean space to
another Euclidean space.
C. What is the relation between a kernel of a linear transformation and the
eigenspace?
Congratulations, you have finished Module 8.