1.8-1.

9: Linear Transformations
Goal: think of the equation Ax = b in terms of the multiplication by A
function: its input is x and its output is b.

Primary One: Primary Four:

22 = 4 Think of this as: 2 4
squaring
32 = 9 3 9

Last week: Today:

  1   1 multiply by A  
8 3 4 10 10
2 = Think of this as: 2
5 1 2 9 9
1 1

  1   1 multiply by A  
8 3 4 4 4
0 = 0
5 1 2 7 7
1 1
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 1 of 24
 Goal: think of the equation Ax = b in terms of the multiplication by A
function: its input is x and its output is b.

In this class, we are interested in functions that are linear (see p6 for the definition).
Key skills:
i Determine whether a function is linear (p7-9);
ii Find the standard matrix of a linear function (p12-13);
iii Describe existence and uniqueness of solutions in terms of linear functions
(p17-23).

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 2 of 24

 Definition: A function f from Rn to Rm is a rule that assigns to each vector x
in Rn a vector f (x) in Rm . We write f : Rn Rm .

Rn is the domain of f .
Rm is the codomain of f .
f (x) is the image of x under f .
The range is the set of all
images. It is a subset of the
codomain.

Example: f : R R given by f (x) = x2 .

Its domain = codomain = R, its range = {zero and positive numbers}.
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 3 of 24
 Examples:
3
  x1 x2
2 3 x1
f : R R , defined by f = 2x2 .
x2
0
The range of f is the plane z = 0 (it is obvious that the range must be a
subset of the plane z = 0, and with a bit of work (see p18), we can show
that all points in R3 with z = 0 is the image of some point in R2 under f ).

 
3 2 8 3 4
h : R R , given by the matrix transformation h(x) = x.
5 1 2

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 4 of 24

 Geometric Examples:

g : R2 R2 , given by reflection through the x2 -axis.

  
x1 x1
 x2 x2
g = . g
x2 x2 g(x)
x

x1 x1

S : R2 R2 , given by dilation by a factor of 3.

x2 S(x)
S(x) = 3x. x2
S
x

HKBU Math 2207 Linear Algebra

x1 x
Semester 1 2018, Week 3, Page 5 of 124
 In this class, we will concentrate on functions that are linear. (For historical
reasons, people like to say linear transformation instead of linear function.)
Definition: A function T : Rn Rm is a linear transformation if:
1. T (u + v) = T (u) + T (v) for all u, v in the domain of T ;
2. T (cu) = cT (u) for all scalars c and for all u in the domain of T .
For your intuition: the name linear is because these functions preserve lines:
A line through the point p in the direction v is the set p + sv, where s is any number.
If T is linear, then the image of this set is
1 2
T (p + sv) = T (p) + T (sv) = T (p) + sT (v),

the line through the point T (p) in the direction T (v).

(If T (v) = 0, then the image is just the point T (p).)
Fact: A linear transformation T must satisfy T (0) = 0.
Proof: Put c = 0 in condition 2.
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 6 of 24
 Definition: A function T : Rn Rm is a linear transformation if:
1. T (u + v) = T (u) + T (v) for all u, v in the domain of T ;
2. T (cu) = cT (u) for all scalars c and all u, in the domain of T .
3
  x1 x2
x1
Example: Is f = 2x2 is not linear:
x2
0
 
1
Take u = and c = 2:
1

     16
1 2
f 2 =f = 4 .
1 2
0

  1 2 16
1 So condition 2 is false for f .
2f = 2 2 = 4 6= 4 .

1
0 0 0
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 7 of 24
 Definition: A function T : Rn Rm is a linear transformation if:
1. T (u + v) = T (u) + T (v) for all u, v in the domain of T ;
2. T (cu) = cT (u) for all scalars c and all u, in the domain of T .
   
x1 x1
Example: g = (reflection through the x2 -axis) is linear:
x2 x2
           
u1 + v1 u1 v1 u1 v1 u1 v1
1. g = = + =g +g .
u2 + v2 u2 + v2 u2 v2 u2 v2
       
cu1 cu1 u1 u1
2. g = =c = cg .
cu2 cu2 u2 u2
Notice from the previous two examples:
To show that a function is linear, check both conditions for general u, v, c (i.e.
use variables).
To show that a function is not linear, show that one of the conditions is not
satisfied for a particular numerical values of u and v (for 1) or of c and u (for 2).
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 8 of 24
 1 0
EXAMPLE: Let e 1 = , e2 = . Suppose T : R2 R3 is a linear transformation with
0 1

1 0
Te1 = 0 and Te2 = 1 .
2 1
3
Find the image of .
2

Solution:

x2 2 x2
1
2 0
8

1 4x
3

0
3
2
x1 1 x
1 2 3 0
1

T3e 1 + 2e 2 = 3Te 1 + 2Te 2

 Definition: A function T : Rn Rm is a linear transformation if:
1. T (u + v) = T (u) + T (v) for all u, v in the domain of T ;
2. T (cu) = cT (u) for all scalars c and all u, in the domain of T .

For simple functions, we can combine the two conditions at the same time, and
check just one statement: T (cu + dv) = cT (u) + dT (v), for all scalars c, d and
all vectors u, v.

Example: S(x) = 3x (dilation by a factor of 3) is linear:

S(cu + dv) = 3(cu + dv) = 3cu + 3dv = S(cu) + S(dv).

Important Example: All matrix transformations T (x) = Ax are linear:

T (cu + dv) = A(cu + dv) = A(cu) + A(dv) = cAu + dAv = cT (u) + dT (v).

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 9 of 24

 In general:
Write ei for the vector with 1 in row i and 0 in all other rows.

1 0 0
For example, in R3 , we have e1 = 0, e2 = 1, e3 = 0.
0 0 1

x1
n ..
{e1 , . . . , en } span R , and x = . = x1 e1 + + xn en .
xn
So, if T : Rn Rm is a linear transformation, then
x
| | | 1
..
T (x) = T (x1 e1 + +xn en ) = x1 T (e1 )+ +xn T (en ) = T (e1 ) . . . T (en ) . .

| | | xn
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 11 of 24
 Theorem 10: The matrix of a linear transformation: Every linear
transformation T : Rn Rm is a matrix transformation: T (x) = Ax where A
is the standard matrix for T , the m n matrix given by

| | |
A = T (e1 ) . . . T (en ) .
| | |

Example: S : R2 R2 , given by dilation by a factor of 3, S(x) = 3x.

           
1 1 3 0 0 0
S(e1 ) = S =3 = , S(e2 ) = S =3 = .
0 0 0 1 1 3
   
3 0 3 0
So the standard matrix of S is , i.e. S(x) = x.
0 3 0 3

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 12 of 24

    
x1 x1
Example: g = (reflection through the x2 -axis):
x2 x2

g(e1 ) =

| |  
1 0
The standard matrix of g is g(e1 ) g(e2 ) =
.
0 1
| |
    
1 0 x1 x1
Indeed, = .
0 1 x2 x2
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 13 of 24
 Vertical Contraction and Expansion

Image of the Standard

Unit Square Matrix
x2 x2

0


k


0
k
x1 x1
1 1
0 0
0<k<1 k>1

Chapter 1 Lay, Linear Algebra and Its Applications, Second EditionUpdate A1.8.T2
Copyright c 2000 by Addison Wesley Longman. All rights reserved.
 Projection onto the x1-axis

Image of the Standard

Unit Square Matrix

x2

x1
0 1
0 0

Chapter 1 Lay, Linear Algebra and Its Applications, Second EditionUpdate A1.8.T4
Copyright c 2000 by Addison Wesley Longman. All rights reserved.
 EXAMPLE: T : R2 R2 given by rotation counterclockwise about the
origin through an angle :

x2
( sin , cos ) (0, 1)

(cos , sin )
x1
(1, 0)

Chapter 1 Lay, Linear Algebra and Its Applications, Second EditionUpdate A1.8.01
Copyright c 2000 by Addison Wesley Longman. All rights reserved.
 Now we rephrase our existence and uniqueness questions in terms of functions.
Definition: A function f : Rn Rm is onto (surjective) if each y in Rm is the
image of at least one x in Rn .
Other ways of saying this:
The range is all of the codomain Rm ,
The equation f (x) = y always has a solution.
Definition: A function f : Rn Rm is one-to-one (injective) if each y in Rm is
the image of at most one x in Rn .
Other ways of saying this:
??? (something that only works for linear transformations, see p20),
The equation f (x) = y has no solutions or a unique solution.

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 17 of 24

 Definition: A function f : Rn Rm is onto (surjective) if each y in Rm is the
image of at least one x in Rn .
Definition: A function f : Rn Rm is one-to-one (injective) if each y in Rm is
the image of at most one x in Rn .
3
  x1 x2
2 3 x1
Example: f : R R , defined by f = 2x2 .
x2
0
0
f is not onto, because f (x) = 0 does not have a solution.
1

0
f is not one-to-one: f (x) = 0 whenever x2 = 0 (regardless of what x1 is) .
0

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 18 of 24

 There is an easier way to check if a linear transformation is one-to-one:
Definition: The kernel of a linear transformation T : Rn Rm is the set of
solutions to T (x) = 0.
Fact: If T (v1 ) = T (v2 ), then v1 v2 is in the kernel of T .
 
Example: Let T be projection onto the x1 -axis. 2
The kernel of T is the x2 -axis. 2
     
2 2 2
T =T = .
 
2 1 0 2
      1
2 2 0
= , which is in the kernel.
2 1 1

Proof of Fact: If T (v1 ) = T (v2 ) = y, then T (v1 v2 ) = T (v1 ) T (v2 ) = y y = 0.

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 19 of 24
 There is an easier way to check if a linear transformation is one-to-one:
Definition: The kernel of a linear transformation T : Rn Rm is the set of
solutions to T (x) = 0.
Fact: If T (v1 ) = T (v2 ), then v1 v2 is in the kernel of T .

Theorem: A linear transformation is one-to-one if and only if its kernel is {0}.

Warning: this only works for linear transformations. For other functions, the
solution sets of f (x) = y and f (x) = 0 are not related.

Proof:
Suppose T is one-to-one. So T (x) = 0 has at most one solution. Since 0 is a
solution, it must be the only one. So its kernel is {0}.
Suppose the kernel of T is {0}. Then, from the Fact, if there are vectors v1 , v2
with T (v1 ) = T (v2 ) = y, then v1 v2 = 0, i.e. v1 = v2 .
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 20 of 24
 Definition: The kernel of a linear transformation T : Rn Rm is the set of
solutions to T (x) = 0.
Theorem: A linear transformation is one-to-one if and only if its kernel is {0}.
So a matrix transformation x 7 Ax is one-to-one if and only if the set of
solutions to Ax = 0 is {0}. This is equivalent to many other things:

Theorem: Uniqueness of solutions to linear systems: For a matrix A, the

following are equivalent:
a. Ax = 0 has no non-trivial solution.
b. If Ax = b is consistent, then it has a unique solution.
c. The columns of A are linearly independent.
d. rref(A) has a pivot in every column (i.e. all variables are basic).
e. The linear transformation x 7 Ax is one-to-one.

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 21 of 24

 Now lets think about onto and existence of solutions.
Recall that the range of a linear transformation T : Rn Rm is the set of
images, i.e. the set of y in Rm with T (x) = y for some x in Rn .
So, the range of the linear transformation x 7 Ax is the set of b for which
Ax = b has a solution.
So the range of T is the span of the columns of A (see week 2 p17).

Example: The standard

 matrix of projection onto
1 0
the x1 -axis is .
0 0
   
1 0
Its range is the x1 -axis, which is also Span , .
0 0

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 22 of 24

 The range of the linear transformation x 7 Ax is the set of b for which Ax = b
has a solution.
And a linear transformation Rn Rm is onto if and only if its range is all of Rm .
Putting these together: x 7 Ax is onto if and only if Ax = b is always
consistent, and this is equivalent to many things:

Theorem 4: Existence of solutions to linear systems: For an m n matrix

A, the following statements are logically equivalent (i.e. for any particular matrix
A, they are all true or all false):
a. For each b in Rm , the equation Ax = b has a solution.
b. Each b in Rm is a linear combination of the columns of A.
c. The columns of A span Rm .
d. rref(A) has a pivot in every row.
e. The linear transformation x 7 Ax is onto.

HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 23 of 24

 Conceptual problems regarding linear independence and linear transformations:
In problems without specific numbers, its often better not to use row-reduction.
The all-important equation: T (c1 v1 + + cp vp ) = c1 T (v1 ) + + cp T (vp ).
Example: Prove that, if {u, v, w} is linearly dependent and T is a linear
transformation, then {T (u), T (v), T (w)} is linearly dependent.
Step 1 Rewrite the mathematical terms in the question as formulas.
What we know: there are scalars c1 , c2 , c3 not all zero with c1 u + c2 v + c3 w = 0.
What we want to show: there are scalars d1 , d2 , d3 not all zero such that
d1 T (u) + d2 T (v) + d3 T (w) = 0.
Step 2 Fill in the missing steps by rearranging vector equations.
Answer: We know c1 u + c2 v + c3 w = 0 for some scalars c1 , c2 , c3 not all zero.
Apply T to both sides: T (c1 u + c2 v + c3 w) = T (0).
Because T is a linear transformation: c1 T (u) + c2 T (v) + c3 T (w) = 0.
Because c1 , c2 , c3 are not all zero, this is a linear dependence relation among
T (u), T (v), T (w).
HKBU Math 2207 Linear Algebra Semester 1 2018, Week 3, Page 24 of 24