Week 7

4.4 Matrix Transformations II

Projection of a Vector ~v on a Line Through the Origin with Direction Vector d~ = [a b c]T .
We will let ~v = [x y z]T .

a
~
~v d ~ ax + by + cz
b .
d= 2
Recall that projd~~v =
~ 2
a + b2 + c 2
kdk
c
ax + by + cz
a2 x + aby + acz
w1 = 2
a=
a + b2 + c 2
a2 + b 2 + c 2
acx + bcy + c2 z
ax + by + cz
c
=
w2 = 2
a + b2 + c 2
a2 + b 2 + c 2
ax + by + cz
acx + bcy + c2 z
w3 = 2
c
=
a + b2 + c 2
a2 + b 2 + c 2
Then the standard matrix for this transformation is

2
a ab ac
1
ab b2 bc
2
a + b2 + c2
ac bc c2
Projection of a Vector ~v onto a Plane Through the Origin with Normal Vector ~n = [a b c]T .
Here we want the vector ~v proj~n~v or if we let ~v = [x y z]T , then we have

2

x
a ab ac
x
1
2

y 2
ab b bc
y
~v proj~n~v =
a + b2 + c 2
z
ac bc c2
z

2

x
a ab ac
1 0 0
1
2

y
ab b bc
0 1 0 2
=
a + b2 + c 2
2
z
ac bc c
0 0 1
2

b + c2 ab
ac
x
1
2
2

ab a + c
bc
y
= 2
a + b2 + c 2
2
2
ac
bc a + b
z
Reflection of a Vector ~v in a Line Through the Origin with Direction Vector d~ = [a b c]T
2

a b2 c2
2ab
2ac
1

.
2ab
b2 a2 c2
2bc
Standard Matrix is 2
a + b2 + c 2
2
2
2
2ac
2bc
c a b
Reflection of a Vector ~v in a Plane Through the Origin with Normal Vector ~n = [a b c]T

2
b + c 2 a2
2ab
2ac
1
2ab
a2 + c2 b2
2bc .
Standard Matrix is 2
a + b2 + c 2
2
2ac
2bc
a + b2 c2
1

Projection of a vector ~v onto a Line Through the Origin with Direction Vector d~ = [a b]T
If we let ~v = [x y]T , then we have
T (~v ) = projd~~v
 
 
[x y]T [a b]T a
x
T
=
y
b
k[a b]T k2
 
1
a
= 2
(ax + by)
2
b
a +b
 2

1
a x + bay
= 2
a + b2 abx + b2 y
 2
 
1
a ab x
= 2
a + b2 ab b2 y
Reflection of a vector ~v about a Line Through the Origin with Direction Vector d~ = [a b]T
If we let ~v = [x y]T , then we have
T (~v ) = ~v + 2(projd~~v ~v )
= 2projd~~v ~v
 

 2
 
  
1
x
a ab
1 0
x
T
=
2 2
2
2
y
0 1
y
a + b ab b
 2
 
2
1
a b
2ab
x
= 2
2
2
2
2ab
b a
y
a +b
Example:

1
Find the matrix for the reflection through the plane with normal vector n = 2 followed by a
3
projection onto the line x = 2t, y = 3t, z = 5t.
Answer:
[P R]

22
(2)(3) (2)(5)
2 + 32 12
2(1)(2)
2(1)(3)
1
1
(2)(3) (3)2 (5)(5)
2(1)(2) 12 + 32 22
2(2)(3)
= 2
2
2
2
2
2
2 + (3) + 5
1 +2 +3
2
2
(2)(5) (3)(5)
5
2(1)(3)
2(2)(3) 1 + 22 32

12 172 2
1
48 258
3
=
494
30 430 5

5.1 Subspaces
We will be looking at the set of all vectors with n real components or Rn . This is an example of a
vector space. We will study other vector spaces in the next chapter.
Definition:
We say a subset of vectors W in Rn is a subspace of Rn if
1. ~0 W
2. If ~x W and ~y W , then ~x + ~y W .
3. If ~x W and a is a scalar, then a~x W .
Examples
1. Rn is a subspace of Rn
2. {~0} is a subspace of Rn
3. Every plane P through the origin in R3 is a subspace of R3 .
Proof: Let ~n be the normal vector of the plane P .
Then P = {~v R3 : ~v ~n = 0}.
1. ~0 P since ~0 ~n = 0.
2. If ~u, ~v P , then ~u ~n = 0 and ~v ~n = 0.
Also,
~u + ~v ~n = ~u ~n + ~v ~n
= 0+0
= 0
3. If ~u P and a R, then a~u ~n = a(~u ~n) = a(0) = 0.
Thus, P is a subspace of R3 .
4. Lines L through the origin are subspaces of R3 .
Proof: Let d~ be the direction vector of L.
Then L = {td~ : t R}.
1. ~0 L since when t = 0 0d~ = ~0.
2. If ~u, ~v L, then ~u = td~ and ~v = sd~ for some scalars s and t.
Then ~u + ~v = td~ + sd~ = (t + s)d~ which belongs to L.
3. If a R and ~u L, then ~u = k d~ for some k R.
~ = (ak)d~ which belongs to L.
Then a~u = a(k d)

Example 5
 

x
Is W =
: x 0 a subspace of R2 ?
y

Theorem 1:
R has no subspaces other than R and {~0}.
Proof:
Suppose R does have a subspace H, where H 6= {~0} and H 6= R.
Since H is non-empty, there exists some non-zero real number H.
Since H is a subspace, any scalar multiple of is also in H.
1
In particular = 1 H.

Since 1 H, any scalar multiple of 1 is also in H and so every real number is in H so H = R.

This is a contradiction since we assumed H 6= R.
Theorem 2:
Let H1 and H2 be subspaces of Rn . Then H1 H2 is a subspace of Rn .
Proof:
1. ~0 H1 H2 since ~0 H1 and ~0 H2 .
2. Let ~u, ~v H1 H2 .
Then ~u H1 and ~u H2
and ~v H1 and ~v H2 .
Thus, ~u + ~v H1 and ~u + ~v H2 since H1 and H2 are subspaces (closed under addition).
Therefore, ~u + ~v H1 H2 .
3. Let ~u H1 H2 and a R.
Then ~u H1 and ~u H2 .
Thus, a~u H1 and a~u H2 .
Therefore, a~u H1 H2 .

Example to Illustrate Theorem 2

y : 2x y z = 0 . and H2 =
y : x + 2y + 3z = 0 .
Let H1 =

z
z
Then H1 H2 is the intersection of two planes through the origin.
In this case, by solving the above system of equations
2x y z = 0
x + 2y + 3z = 0

x
1/5
we obtain the solution y = 7/5 t, which is a line through the origin.
z
1
We showed in Example 4 that this is also a subspace of R3 .

Span
Definition
If ~v1 , ~v2 , . . . , ~vk Rn , then S = span{~v1 , ~v2 , . . . , ~vk } is the set of all linear combinations of ~v1 , ~v2 , . . . , ~vk },
and we say {~v1 , ~v2 , . . . , ~vk } span S.
S = {t1~v1 + t2~v2 + . . . + tk~vk : ti R}
Examples
~ or d~ spans L.
1. A line L in R3 = span{d}
2. A plane P in R3 = span{~u, ~v }, where ~u, ~v are non-parallel, non-zero vectors in R3 . Equivalently,
we could say ~u and ~v span the plane P .
Row Space, Column Space & Null Space
There are 3 particular subspaces associated with an m n matrix A.
The Row
Space
of A, or Row(A), is the set of all linear combinations of the row vectors of A.
~r1
~r2

ie. If A = .. , then
.
~rm
Row(A) =
The Column Space of A, or Col(A), is the set of all linear combinations of the column vectors
of A.


ie. If A = ~c1 ~c2 ~cn , then
Col(A)=

 The Null Space of A, Null(A), is the set of all solutions to the matrix equation A~x = ~0.
ie. Null(A) =
Exercise
Verify that Row(A), Col(A) and Null(A) are indeed subspaces. What are they subspaces of?

Theorem 1:
Let U = span{~v1 , ~v2 , . . . , ~vk } Rn .
(i) U is a subspace of Rn containing each ~vi .
(ii) If W is a subspace of Rn and each ~vi W , then U W .
Proof:
(i)

1. ~0 U since ~0 = 0~v1 + 0~v2 + . . . + 0~vk

2. If ~x, ~y U , then we can write ~x = t1~v1 + t2~v2 + . . . + tk~vk and ~y = s1~v1 + s2~v2 + . . . + sk~vk .
Then ~x + ~y = (t1 + s1 )~v1 + (t2 + s2 )~v2 + . . . + (tk + sk )~vk and thus lies in U .
3. If ~x U , then ~x = t1~v1 + t2~v2 + . . . + tk~vk .
Then a~x = at1~v1 + at2~v2 + . . . + atk~vk and thus lies in U .
Finally, each ~vi lies in U since ~v1 = 1~v1 + 0~v2 + . . . + 0~vk etc.

(ii) Let ~u U . We want to show ~u W .

Since ~u U , we can write ~u = t1~v1 + t2~v2 + . . . + tk~vk .
6

Now, since ~v1 , ~v2 , . . . , ~vk W , t1~v1 , t2~v2 , . . . tk~vk W also, since a subspace is closed under
scalar multiplication.
Since W is also closed under addition, we have t1~v1 + t2~v2 + . . . + tk~vk W and so ~u W .
Example 3:
If ~x, ~y Rn , show span{~x + ~y , ~x ~y } =span{~x, ~y }.
Solution:
To show two sets are equal we need to show they are each contained inside the other. In other words,
to show A = B, we need to show A B and B A.
We will use the result of Theorem 3(ii) above.
Is ~x + ~y , ~x ~y span{~x, ~y }?
Yes, since ~x + ~y = 1(~x) + 1(~y ) and ~x ~y = 1~x + (1)~y .
Thus, by Theorem 3(ii), span{~x + ~y , ~x ~y } span{~x, ~y }.
Is ~x, ~y span{~x + ~y , ~x ~y }?
Yes, since
~x =
~y =

(~x + ~y ) +
(~x + ~y ) +

Thus, span{~x, ~y } span{~x + ~y , ~x ~y }.

Example 4:

7
1
2
Is ~b span{~u, ~v }, where ~b = 4 ~u = 2 ~v = 5.
3
5
6

(~x ~y )
(~x ~y )

Solution

Example 5:

1
3
4
Do ~v1 , ~v2 , ~v3 span R3 where ~v1 = 4 ~v2 = 2 ~v3 = 6?
3
2
7
Solution

Note: The vector equation x1~v1 + x2~v2 + x3~v3 = ~b is equivalent to the matrix equation A~x = ~b. By
placing the vectors ~v1 , ~v2 , ~v3 as columns of a matrix A and reducing the augmented matrix [A|~b], we
can see that this system is only consistent for some vectors ~b R3 . This tells us that there are some
vectors in Rn can not be written as a linear combination of the vectors ~v1 , ~v2 , ~v3 and so these vectors
do not span R3 .
What is significant about the coefficient matrix A in this case?
Example 5:

1
2
4
2
3

Does the set of vectors {~v1 , ~v2 , ~v3 , ~v4 } span R , where ~v1 = 2 ~v2 = 3 ~v3 = 5 ~v4 = 8?
1
3
7
2
Solution