VIETNAM NATIONAL UNIVERSITY-HCMC

International University
Lecture Notes/Slides for

APPLIED LINEAR ALGEBRA

Chapter 3B. Subspaces, Span and Linear

independence

Chapter 3B. Subspaces, Span and Linear independence Page 1/31

Definition (Recall: Linear Combination)

Let u~1 , · · · , u~n , ~v be vectors. Then ~v is said to be a linear combination
of the vectors u~1 , · · · , u~n if there exist scalars, a1 , · · · , an such that

~v = a1 u~1 + · · · + an u~n

Definition (Span of a Set of Vectors)

The collection of all linear combinations of a set of vectors {~
u1 , · · · , u~k } in
n
R is known as the span of these vectors and is written as
u1 , · · · , u~k }.
span{~

u1 , u~2 , . . . , u~k }, then

Additional Terminology. If U = span{~
U is spanned by the vectors u~1 , u~2 , . . . , u~k .
the vectors u~1 , u~2 , . . . , u~k span U.
the set of vectors {~
u1 , u~2 , . . . , u~k } is a spanning set for U.

Chapter 3B. Subspaces, Span and Linear independence Spanning Page 2/31
 Problem
T T
∈ R3 . Show that
 
Let u~ = 1 1 0 and ~v = 3 2 0
 T
~ = 4 5 0
w is in span {~u , ~v }.

Solution
For a vector to be in span {~ u , ~v }, it must be a linear combination of these
~ ∈ span {~
vectors. If w u , ~v }, we must be able to find scalars a, b such that

~ = a~
w u + b~v

     
4 1 3
 5  = a 1 +b 2 
0 0 0
This is equivalent to the following system of equations

a + 3b = 4
a + 2b = 5
Chapter 3B. Subspaces, Span and Linear independence Spanning Page 3/31

Solution (continued)
We solving this system the usual way, constructing the augmented matrix
and row reducing to find the reduced row-echelon form .
   
1 3 4 1 0 7
→ ··· →
1 2 5 0 1 −1

The solution is a = 7, b = −1. This means that

u − ~v
~ = 7~
w

~ is in span {~
Therefore we can say that w u , ~v }.

Chapter 3B. Subspaces, Span and Linear independence Spanning Page 4/31
 Span of a Set of Vectors

Example
Let ~x ∈ R3 be a nonzero vector. Then span{~x } = {k~x | k ∈ R} is a line
through the origin having direction vector ~x .

Problem
 
 
0 0
Describe the span of the vectors u~ =  1  and ~v =  2 .
2 3

Chapter 3B. Subspaces, Span and Linear independence Spanning Page 5/31

Solution
Notice
  that any linear combination of the vectors u~ and ~v yields a vector
0
 y  in the YZ -plane.
z
 
0
Suppose we take an arbitrary vector  y  in the YZ -plane. It turns out
z
we can write any such vector as a linear combination of u~ and ~v .
     
0 0 0
 y  = (−3y + 2z)  1  + (2y − z)  2 
z 2 3

u , ~v } is the YZ -plane.
Hence, span{~

Chapter 3B. Subspaces, Span and Linear independence Spanning Page 6/31
 Span of a Set of Vectors

Consider the previous example where the span of u~ and ~v was the
YZ -plane. Suppose we add another vector w ~ , and consider the span of
u~, ~v , and w
~ . What would happen to the span?

Scenario
  1 Suppose w ~ is a vector in the YZ -plane. For example,
0
~ =  4 . Then w
w ~ is in the span of u~, ~v . Adding w
~ to the set doesn’t
1
change the span at all.

span {~ ~ } = span {~
u , ~v , w u , ~v }

Chapter 3B. Subspaces, Span and Linear independence Spanning Page 7/31

Span of a Set of Vectors

Scenario
  2 Suppose w ~ is not in the YZ -plane. For example, suppose
1
~ =  0 .
w
2
Notice that now, the three vectors span R3 . Any vector in R3 can be
written as a linear combination of u~, ~v , w
~ as follows:
       
x 0 0 1
 y  = (−4x + 5y + 2z)  1  + (2x + 2y − z)  2  + (x)  0 
z 2 3 2

You can see that the span of these three vectors depended on whether w ~
was in span{~u , ~v } or not. In the next section, we will examine the
distinction between these two scenarios using the concept of linear
independence.

Chapter 3B. Subspaces, Span and Linear independence Spanning Page 8/31
 Linearly Independent Set of Vectors

Definition
u1 , u~2 , ..., u~k } be a set of vectors in Rn . This set is linearly
Let {~
independent if no vector in the set is in the span of the other vectors of
that set.
If a set of vectors is not linearly independent, we call it linearly dependent.

Chapter 3B. Subspaces, Span and Linear independence Independence Page 9/31

A Linearly Dependent Set

Problem
     
0 0 0
Consider the vectors u~ =  1  , ~v =  2  , w
~ =  4 . Is the set
2 3 1
{~ ~ } linearly independent?
u , ~v , w

Solution
Notice that we can write w
~ as a linear combination of u~, ~v as follows:
     
0 0 0
 4  = (−10)  1  + (7)  2 
1 2 3

Hence, w
~ is in span{~ u , ~v }. By the definition, this set is not linearly
independent (it is linearly dependent).

Chapter 3B. Subspaces, Span and Linear independence Independence Page 10/31
 A Linearly Independent Set

Problem
Consider the vectors
     
0 0 1
u~ =  1  , ~v =  2  , w
~ = 0 
2 3 2

Is the set {~ ~ } linearly independent?

u , ~v , w

Solution
We cannot write any of the three vectors as a linear combination of the
other two. (We will see how to show this soon.) Therefore the set
{~ ~ } is linearly independent.
u , ~v , w

Chapter 3B. Subspaces, Span and Linear independence Independence Page 11/31

Linear Independence as a Linear Combination

The following theorem provides a familiar way to check if a set of vectors is

linearly independent.

Theorem
u1 , · · · , u~k } in Rn is linearly independent if and
The collection of vectors, {~
only if whenever
X n
ai u~i = ~0
i=1

it follows that each ai = 0.

Thus {~ u1 , · · · , u~k } in Rn is linearly independent exactly when the system of
linear equations AX = 0 has only the trivial solution, where A is the n × k
matrix having these vectors as columns.

Chapter 3B. Subspaces, Span and Linear independence Independence Page 12/31
 Linear Independence

We can state the conclusion of this theorem in another way: The set of
vectors {~
u1 , ..., u~k } is linearly independent if and only if there is no
nontrivial linear combination which equals zero. If a linear combination of
the vectors equals zero, then all the coefficients of the combination are
zero.
If the set is linearly independent, then

a1 u~1 + ... + ak u~k = 0

implies that
a1 = a2 = ... = ak = 0

Chapter 3B. Subspaces, Span and Linear independence Independence Page 13/31

Linear Independence

We can use the reduced row-echelon form of the matrix to determine if the
columns form a linearly independent set of vectors.

Problem
Determine whether the following set of vectors are linearly independent.
     
 1 2 0 
 2 , 1 , 1 
3 0 1
 

Chapter 3B. Subspaces, Span and Linear independence Independence Page 14/31
 Solution
Construct the 3 x 3 matrix A having these vectors as columns:
 
1 2 0
A= 2 1 1 
3 0 1

By the above theorem, the set of vectors is linearly independent if the

system AX = 0 has only the trivial solution. We can see this from the
reduced row-echelon form of the matrix A.
 
1 0 0
 0 1 0 
0 0 1
Since all columns are pivot columns (and the rank of A is 3), the vectors
are linearly independent.

Chapter 3B. Subspaces, Span and Linear independence Independence Page 15/31

Problem
Determine whether the following vectors are linearly independent. If they
are linearly dependent, write one of the vectors as a linear combination of
the others.        

 1 2 0 8 

2 7 1 5
        
 , , , 
  4   17   3   11 
 
1 2 0 11
 

Solution
Construct the matrix A using these vectors as columns.
 
1 2 0 8
 2 0 1 5 
A=  4

0 3 11 
1 3 0 11

Chapter 3B. Subspaces, Span and Linear independence Independence Page 16/31
 Solution (continued)
The reduced row-echelon form of this matrix is
 
1 0 0 2
 0 1 0 1 
 
 0 0 1 3 
0 0 0 0

Since the rank of A is 3 < 4, the vectors are linearly dependent.

Therefore, there are infinitely many solutions to AX = 0, one of which is
 
−2
 −1 
 
 −3 
1

Chapter 3B. Subspaces, Span and Linear independence Independence Page 17/31

Solution (continued)
Therefore we can write:
        
1 0 2 8 0
 2  1   0
     5   0 
−2   − 1 
  − 3  + 1
 11  =  0
  
4  3   0  
1 0 3 11 0

We can rewrite this as:

      

1 0 2 8
 2  1 
 + 3 0
    5 
2 4 
 + 1 = 
 3   0   11 
1 0 3 11

This shows that one of the vectors can be written as a linear combination
of the other three vectors. While here we chose the fourth vector, we could
have chosen any of the vectors to isolate.

Chapter 3B. Subspaces, Span and Linear independence Independence Page 18/31
 Subspaces

Definition
Let V be a nonempty collection of vectors in Rn . Then V is a subspace if
whenever a and b are scalars and u~ and ~v are vectors in V , a~
u + b~v is also
in V .

Subspaces are closely related to the span of a set of vectors which we

discussed earlier.

Theorem
Let V be a nonempty collection of vectors in Rn . Then V is a subspace of
Rn if and only if there exist vectors {~
u1 , ..., u~k } in V such that

V = span {~
u1 , ..., u~k }

Chapter 3B. Subspaces, Span and Linear independence Subspaces and Basis Page 19/31

Subspaces

Subspaces are also related to the property of linear independence.

Theorem
If V is a subspace of Rn , then there exist linearly independent vectors
{~u1 , ..., u~k } of V such that

V = span {~
u1 , ..., u~k }

In other words, subspaces of Rn consist of spans of finite, linearly

independent collections of vectors in Rn .

Chapter 3B. Subspaces, Span and Linear independence Subspaces and Basis Page 20/31
 Basis of a Subspace

Definition
Let V be a subspace of Rn . Then {~
u1 , ..., u~k } is called a basis for V if the
following conditions hold:
u1 , ..., u~k } = V
span{~
{~
u1 , ..., u~k } is linearly independent.

The following theorem claims that any two bases of a subspace must be of
the same size.
Theorem
Let V be a subspace of Rn and suppose {~
u1 , ..., u~k } and {~v1 , ..., ~vm } are
two bases for V . Then k = m.

Chapter 3B. Subspaces, Span and Linear independence Subspaces and Basis Page 21/31

Dimension

The previous theorem shows than all bases of a subspace will have the
same size. This size is called the dimension of the subspace.

Definition
Let V be a subspace of Rn . Then the dimension of V is the number of a
vectors in a basis of V .

Chapter 3B. Subspaces, Span and Linear independence Subspaces and Basis Page 22/31
 Properties of Rn

Note that the dimension of Rn is n.

There are some other important properties of vectors in Rn .
Theorem
u1 , ..., u~n } is a linearly independent set of a vectors in Rn , then
If {~
{~u1 , ..., u~n } is a basis for Rn .
u1 , ..., u~m } spans Rn . Then m ≥ n.
Suppose {~
u1 , ..., u~n } spans Rn , then {~
If {~ u1 , ..., u~n } is linearly independent.
u1 , u~2 , ..., u~k } is a set of vectors in Rn with k > n, then the set is
If {~
linearly dependent.

It follows then that a basis is a minimal spanning set. If a subspace has

dimension d, then any spanning set has size at least d, and any spanning
set of size d must be a basis (and is therefore independent).

Chapter 3B. Subspaces, Span and Linear independence Subspaces and Basis Page 23/31

Row and Column Space

Definition
Let A be an m × n matrix. The column space of A is the span of the
columns of A. The row space of A is the span of the rows of A.

Problem
Find the rank of the matrix A and describe the column and row spaces
efficiently.
 
1 2 1 3 2
A= 1  3 6 0 2 
3 7 8 6 6

Chapter 3B. Subspaces, Span and Linear independence Special Spaces Page 24/31
 Example: Column Space

Solution
To find the column space, we first find the reduced row-echelon form of A:
 
1 0 −9 9 2
 0 1 5 −3 0 
0 0 0 0 0

Therefore rank(A) = 2.
Note the first two columns are the pivot columns. All columns of the above
reduced row-echelon matrix are in
   
 1 0 
span  0  ,  1 
0 0
 

Chapter 3B. Subspaces, Span and Linear independence Special Spaces Page 25/31

Example: Column Space

Solution (continued)
To construct the column space, we use the pivot columns of the original
matrix - in this case, the first and second columns. Therefore the column
space of A is    
 1 2 
span  1  ,  3 
3 7
 

Chapter 3B. Subspaces, Span and Linear independence Special Spaces Page 26/31
 Example: Row Space
Solution (continued)
To find the row space of A we again look at the reduced row-echelon form
of the matrix.
 
1 0 −9 9 2
 0 1 5 −3 0 
0 0 0 0 0
The row space of A is the span of the non-zero rows of the above matrix:
   
span 1 0 −9 9 2 , 0 1 5 −3 0

Notice that the vectors used in the description of the column space are
from the original matrix, while those in the row space are from the reduced
row-echelon form of the original matrix.

Chapter 3B. Subspaces, Span and Linear independence Special Spaces Page 27/31

Null Space

Definition
The null space of A, or kernel of A is defined as:

ker(A) = {X : AX = 0}

We also speak of the image of A, Im(A), which is all vectors of the form
AX where X is in Rn .
To find ker(A), we solve the system of equations AX = 0.

Problem
Find ker(A) for the matrix A:
 
1 2 1
A =  0 −1 1 
2 3 3

Chapter 3B. Subspaces, Span and Linear independence Special Spaces Page 28/31
 Null Space

Solution
The first step is to set up the augmented matrix:
 
1 2 1 0
 0 −1 1 0 
2 3 3 0

Place this matrix in reduced row-echelon form:

 
1 0 3 0
 0 1 −1 0 
0 0 0 0

Chapter 3B. Subspaces, Span and Linear independence Special Spaces Page 29/31

Null Space

Solution (continued)
The solution to this system of equations is
  
 3t 
 t :t∈R
t
 

Therefore the null space of A is the span of this vector:

 
 3 
ker(A) = span  1 
1
 

Chapter 3B. Subspaces, Span and Linear independence Special Spaces Page 30/31
 Nullity

Definition
The dimension of the null space of a matrix is called the nullity, denoted
null(A).

Theorem
Let A be an m × n matrix. Then,

rank(A) + null(A) = n

For instance, in the last example, A was a 3 × 3 matrix. The rank was 2
and the nullity was 1 (since the null space had dimension 1).

rank(A) + null(A) = 2 + 1 = 3 = n

Chapter 3B. Subspaces, Span and Linear independence Special Spaces Page 31/31