Basis and Dimension of a Vector Space

Definition 11.1.1. A set of vectors 𝑆𝑆 = {𝐯𝐯1 , 𝐯𝐯2 , … , 𝐯𝐯𝑛𝑛 } in a vector space 𝑉𝑉 is called a basis for
𝑉𝑉 if the following conditions hold true.

1. 𝑆𝑆 spans 𝑉𝑉. 2. 𝑆𝑆 is linearly independent.

Remark. This definition tells us that a basis has two features:

It

• must have enough vectors to span the space;

• should not have so many vectors that one of them could be written as a linear
combination of the other vectors in it.

1 0 0
Example 11.1.1. Show that 𝑆𝑆 = ��0� , �1� , �0�� is a basis for ℝ3 .
0 0 1

Solution. In Example 10.2.3 and Example 10.2.6, we proved that span(𝑆𝑆) = ℝ3 and 𝑆𝑆 is
linearly independent, respectively. Thus 𝑆𝑆 is a basis for ℝ3 . This basis is called the standard
basis for ℝ3 .

Similarly, the set 𝑆𝑆 = {𝐞𝐞1 , 𝐞𝐞2 , … , 𝐞𝐞𝑛𝑛 } where

1 0 0 0
⎡0⎤ ⎡1⎤ ⎡0⎤ ⎡0⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⋮ ⋮ ⋮ ⋮
𝐞𝐞1 = ⎢ ⎥ , 𝐞𝐞2 = ⎢ ⎥ , ⋯ 𝐞𝐞𝑘𝑘 = ⎢ ⎥ , ⋯ 𝐞𝐞𝑛𝑛 = ⎢ ⎥.
⎢⋮⎥ ⎢⋮⎥ 1
⎢ ⎥ ⎢⋮⎥
⎢⋮⎥ ⎢⋮⎥ ⎢⋮⎥ ⎢⋮⎥
⎣0⎦ ⎣0⎦ ⎣0⎦ ⎣1⎦

𝐞𝐞𝑖𝑖 has 1 in the 𝑖𝑖th position and 0’s elsewhere. The set 𝑆𝑆 is called the standard basis for ℝ𝑛𝑛 .

⎧ ⎫
⎪ 1 0 −2 ⎪
Example 11.1.2. Show that 𝑆𝑆 = �2� , �1� , � 0 � is a basis for ℝ3 .
⎨� � � 1 ⎬
⎪ 3 2 ⎪
⎩ 𝐯𝐯1 𝐯𝐯2 𝐯𝐯3 ⎭

Solution. In Example 10.2.4 and Example 10.2.7, we obtained that span(𝑆𝑆) = ℝ3 and 𝑆𝑆 is
linearly independent, respectively. Hence 𝑆𝑆 is a basis for ℝ3 .

Example 11.1.1 and Example 11.1.2 show that basis for a vector space is not unique.

Mohammad Hassan Murad P a g e | 26

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Basis and Dimension of a Vector Space
Example 11.1.3. Show that the set

1 0 0 1 0 0 0 0
𝑆𝑆 = �� �,� �,� �,� ��
0 0 0 0 1 0 0 1

is a basis for 𝑀𝑀2,2 . This is called the standard basis for 𝑀𝑀2,2 .

𝑎𝑎11 𝑎𝑎12
Solution. Take an arbitrary matrix �𝑎𝑎 𝑎𝑎22 � ∈ 𝑀𝑀2,2.
21

Clearly,
𝑎𝑎11 𝑎𝑎12 1 0 0 1 0 0 0 0
�𝑎𝑎 𝑎𝑎22 � = 𝑎𝑎11 �0 � + 𝑎𝑎12 � � + 𝑎𝑎21 � � + 𝑎𝑎22 � �
21 0 0 0 1 0 0 1

Therefore, span(𝑆𝑆) = 𝑀𝑀2,2.

To show the linear independence, we solve the vector equation for 𝑐𝑐1 , 𝑐𝑐2 , 𝑐𝑐3 , and 𝑐𝑐4

𝑐𝑐1 𝐯𝐯1 + 𝑐𝑐2 𝐯𝐯2 + 𝑐𝑐3 𝐯𝐯3 + 𝑐𝑐4 𝐯𝐯4 = 𝟎𝟎

where

1 0 0 1 0 0 0 0
𝐯𝐯1 = � �, 𝐯𝐯2 = � �, 𝐯𝐯3 = � �, 𝐯𝐯4 = � �
0 0 0 0 1 0 0 1

So,

1 0 0 1 0 0 0 0 0 0
𝑐𝑐1 � � + 𝑐𝑐2 � � + 𝑐𝑐3 � � + 𝑐𝑐4 � �=� �
0 0 0 0 1 0 0 1 0 0
𝑐𝑐1 𝑐𝑐2 0 0
�𝑐𝑐 𝑐𝑐4 � = �0 0�
3

This equation is true if and only if 𝑐𝑐1 = 𝑐𝑐2 = 𝑐𝑐3 = 𝑐𝑐4 = 0. This shows that 𝑆𝑆 is linearly
independent. Thus 𝑆𝑆 is basis for 𝑀𝑀2,2 .

Mohammad Hassan Murad P a g e | 27

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Basis and Dimension of a Vector Space
Example 11.1.4. Show that the set 𝑆𝑆 = {1, 𝑥𝑥, 𝑥𝑥 2 } is a basis for 𝐏𝐏2 (𝑥𝑥) where

𝐏𝐏2 (𝑥𝑥) = {𝑎𝑎2 𝑥𝑥 2 + 𝑎𝑎1 𝑥𝑥 + 𝑎𝑎0 | 𝑎𝑎2 , 𝑎𝑎1 , 𝑎𝑎0 ∈ ℝ}.

Solution. Let us denote

𝐯𝐯1 = 1, 𝐯𝐯2 = 𝑥𝑥, 𝐯𝐯3 = 𝑥𝑥 2

Take an arbitrary polynomial 𝑎𝑎2 𝑥𝑥 2 + 𝑎𝑎1 𝑥𝑥 + 𝑎𝑎0 from 𝐏𝐏2 (𝑥𝑥). Then obviously,

𝑎𝑎2 𝑥𝑥 2 + 𝑎𝑎1 𝑥𝑥 + 𝑎𝑎0 = 𝑎𝑎0 𝐯𝐯1 + 𝑎𝑎1 𝐯𝐯2 + 𝑎𝑎2 𝐯𝐯3

This shows that span(𝑆𝑆) = 𝐏𝐏2 (𝑥𝑥). Now we will show the linear independence of 𝑆𝑆.

𝑐𝑐1 𝐯𝐯1 + 𝑐𝑐2 𝐯𝐯2 + 𝑐𝑐3 𝐯𝐯3 = 𝟎𝟎

𝑐𝑐1 + 𝑐𝑐2 𝑥𝑥 + 𝑐𝑐3 𝑥𝑥 2 = 0

By differentiating both sides with respect to 𝑥𝑥, we obtain

𝑐𝑐2 + 2𝑐𝑐3 𝑥𝑥 = 0

Again differentiating, we get

2𝑐𝑐3 = 0

gives 𝑐𝑐1 = 𝑐𝑐2 = 𝑐𝑐3 = 0. ∎

1
In Example 10.2.1, we observed that 𝐯𝐯 = �1� ∈ ℝ3 can be written as a linear combination of
1
⎧ ⎫
⎪ 1 0 −1 ⎪
vectors in the set 𝑆𝑆 = �2� , �1� , � 0 � in various ways.
⎨� � � 1 ⎬
⎪ 3 2 ⎪
⎩ 𝐯𝐯1 𝐯𝐯2 𝐯𝐯3 ⎭

For example,

𝐯𝐯 = 𝐯𝐯1 − 𝐯𝐯2 or 𝐯𝐯 = 2𝐯𝐯1 − 3𝐯𝐯2 + 𝐯𝐯3 .

The next theorem tells us that under what condition this representation is unique.

Mohammad Hassan Murad P a g e | 28

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Basis and Dimension of a Vector Space
Theorem 11.1.1. (Uniqueness of Basis Representation)

If 𝑆𝑆 = {𝐯𝐯1 , 𝐯𝐯2 , … , 𝐯𝐯𝑛𝑛 } is a basis for a vector space 𝑉𝑉, then every vector in 𝑉𝑉 can be written in
one and only one way as a linear combination of vectors in S.

Proof. The existence portion of the proof is straightforward. Since 𝑆𝑆 is a basis for 𝑉𝑉, 𝑆𝑆 spans
𝑉𝑉, thus an arbitrary vector 𝐯𝐯 in 𝑉𝑉 can be expressed as

𝐯𝐯 = 𝑐𝑐1 𝐯𝐯1 + 𝑐𝑐2 𝐯𝐯2 + ⋯ + 𝑐𝑐𝑛𝑛 𝐯𝐯𝑛𝑛 (1)

for some scalars 𝑐𝑐1 , 𝑐𝑐2 , … , 𝑐𝑐𝑛𝑛 .

To prove uniqueness (that a vector can be represented in only one way), assume 𝐯𝐯 has
another representation

𝐯𝐯 = 𝑑𝑑1 𝐯𝐯1 + 𝑑𝑑2 𝐯𝐯2 + ⋯ + 𝑑𝑑𝑛𝑛 𝐯𝐯𝑛𝑛 (2)

Subtracting eq. (2) from eq. (1), we obtain

(𝑐𝑐1 − 𝑑𝑑1 )𝐯𝐯1 + (𝑐𝑐2 − 𝑑𝑑2 )𝐯𝐯2 + ⋯ + (𝑐𝑐𝑛𝑛 − 𝑑𝑑𝑛𝑛 )𝐯𝐯𝑛𝑛 = 𝟎𝟎.

𝑆𝑆 is linearly independent, however, so the only solution to this equation is the trivial solution

𝑐𝑐1 − 𝑑𝑑1 = 0, 𝑐𝑐2 − 𝑑𝑑2 = 0, ⋯, 𝑐𝑐𝑛𝑛 − 𝑑𝑑𝑛𝑛 = 0,

which means that 𝑐𝑐𝑖𝑖 = 𝑑𝑑𝑖𝑖 for all 𝑖𝑖 = 1, 2, … , 𝑛𝑛 and 𝐯𝐯 has only one representation for the basis
𝑆𝑆. ∎

Theorem 11.1.2. If 𝑆𝑆 = {𝐯𝐯1 , 𝐯𝐯2 , … , 𝐯𝐯𝑛𝑛 } is a basis for a vector space 𝑉𝑉, then every set
containing more than 𝑛𝑛 vectors in 𝑉𝑉 is linearly dependent, hence cannot be a basis.

Proof. Consult the textbook.

⎧ ⎫
⎪ 1 0 −2 3 ⎪
Question. Is 𝑆𝑆 = �2� , �1� , � 0 � , � 1 � a basis for ℝ3 ?
⎨� � � �−2 ⎬
⎪ 3 2 1 ⎪
⎩ 𝐯𝐯1 𝐯𝐯2 𝐯𝐯3 𝐯𝐯4 ⎭

Answer. NO. Because in Example 11.1.1 we found that one basis of ℝ3 contains three vectors,
since 𝑆𝑆 has four vectors, it must be linearly dependent, hence 𝑆𝑆 is not a basis for ℝ3 .

Mohammad Hassan Murad P a g e | 29

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Basis and Dimension of a Vector Space
Corollary (to Theorem 11.1.2) If a vector space 𝑉𝑉 has one basis with 𝑛𝑛 vectors, then every
basis for 𝑉𝑉 has 𝑛𝑛 vectors.

Proof. Let us take 𝑆𝑆1 = {𝐮𝐮1 , 𝐮𝐮2 , … , 𝐮𝐮𝑚𝑚 } and 𝑆𝑆2 = {𝐯𝐯1 , 𝐯𝐯2 , … , 𝐯𝐯𝑛𝑛 } are two bases for 𝑉𝑉. We want
to show that 𝑚𝑚 = 𝑛𝑛.

Since 𝑆𝑆1 is a basis, then by Theorem 11.1.2, 𝑆𝑆2 cannot have more than 𝑚𝑚 vectors. That is,
𝑛𝑛 ≤ 𝑚𝑚.

By a similar argument, since 𝑆𝑆2 is a basis, the set 𝑆𝑆1 cannot have more than 𝑛𝑛 vectors. That is,
𝑚𝑚 ≤ 𝑛𝑛. These two inequalities prove that 𝑚𝑚 = 𝑛𝑛. ∎

Dimension of a Vector Space

Definition 11.1.2 If a vector space 𝑉𝑉 has a basis consisting of 𝑛𝑛 vectors, then the number 𝑛𝑛
is called the dimension of 𝑉𝑉, denoted by dim(𝑉𝑉). If 𝑉𝑉 consists of the zero vector alone, the
dimension of 𝑉𝑉 is defined as zero. That is, if 𝑉𝑉 = {𝟎𝟎}, then dim(𝑉𝑉) = 0.

In other words, if a vector space is spanned by a set of finitely many vectors of 𝑉𝑉, then it is
called finite-dimensional, otherwise it is called an infinite-dimensional vector space.

By using Examples 11.1.1, 11.1.3, and 11.1.4, we can write

dim(ℝ3 ) = 3, dim�𝑀𝑀2,2 � = 4, dim(𝐏𝐏2 ) = 3.

In general, dim(ℝ𝑛𝑛 ) = 𝑛𝑛, dim�𝑀𝑀𝑚𝑚,𝑛𝑛 � = 𝑚𝑚𝑚𝑚, dim(𝐏𝐏𝑛𝑛 ) = 𝑛𝑛 + 1.

Some examples of finite-dimensional vector spaces are ℝ𝑛𝑛 , 𝐏𝐏𝑛𝑛 , 𝑀𝑀𝑚𝑚,𝑛𝑛 . On the other hand,
𝐶𝐶[𝑎𝑎, 𝑏𝑏], the vector space of all continuous functions defined over [𝑎𝑎, 𝑏𝑏] with 𝑎𝑎 < 𝑏𝑏, is an
example of an infinite-dimensional vector space.

Question. Does every vector space have a basis?

Answer. Yes. Unfortunately, it is beyond the scope of our discussion.

Mohammad Hassan Murad P a g e | 30

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Basis and Dimension of a Vector Space
Dimension of a Subspace
𝑥𝑥
Example 11.1.5. Find the dimension of 𝑊𝑊 = ��0� : 𝑥𝑥, 𝑧𝑧 ∈ ℝ�, a subspace of ℝ3 .
𝑧𝑧

Solution. In Example 10.1.1, we showed that 𝑊𝑊 is subspace of ℝ3 . To find its dimension, we

𝑥𝑥
take an arbitrary vector 𝐯𝐯 = �0� ∈ 𝑊𝑊. This vector can be written as
𝑧𝑧
𝑥𝑥 1 0
�0� = 𝑥𝑥 �0� + 𝑧𝑧 �0�.
𝑧𝑧 0 1

⎧ ⎫
⎪ 1 0 ⎪
This shows that span(𝑆𝑆) = 𝑊𝑊 where 𝑆𝑆 = �0� , �0� .
⎨� �1 ⎬
⎪ 0 ⎪
⎩ 𝐯𝐯1 𝐯𝐯2 ⎭

Furthermore, 𝑆𝑆 is linearly independent. To show this, consider the following equation

𝑐𝑐1 𝐯𝐯1 + 𝑐𝑐2 𝐯𝐯2 = 𝟎𝟎

1 0 0
𝑐𝑐1 �0� + 𝑐𝑐2 �0� = �0�
0 1 0
𝑐𝑐1 0
� 0 � = �0�.
𝑐𝑐2 0

The last equation is true if and only if 𝑐𝑐1 = 𝑐𝑐2 = 0. Therefore, 𝑆𝑆 is linearly independent. Hence
𝑆𝑆 is a basis for 𝑊𝑊. Thus, dim(𝑊𝑊) = 2.

Example 11.1.6. Find the dimension of the subspace 𝑆𝑆2 that is a set of all 2 × 2 real
symmetric matrices.

Solution. Let

𝑎𝑎 𝑏𝑏
𝑆𝑆2 = �� � ∈ 𝑀𝑀2,2 | 𝑎𝑎, 𝑏𝑏, 𝑐𝑐 ∈ ℝ�.
𝑏𝑏 𝑐𝑐

Mohammad Hassan Murad P a g e | 31

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Basis and Dimension of a Vector Space
In Example 10.1.2, we showed that 𝑆𝑆2 is subspace of 𝑀𝑀2,2 . To find its dimension, we take an
arbitrary matrix 𝐴𝐴 from 𝑆𝑆2 . Then

𝑎𝑎 𝑏𝑏 1 0 0 1 0 0
𝐴𝐴 = � � = 𝑎𝑎 � � + 𝑏𝑏 � � + 𝑐𝑐 � �.
𝑏𝑏 𝑐𝑐 0 0 1 0 0 1

1 0 0 1 0 0
This shows that 𝑆𝑆 = �� �,� �,� �� spans 𝑆𝑆2 .
���
0 0 ���
1 0 ���
0 1
𝐯𝐯1 𝐯𝐯2 𝐯𝐯3

To show that 𝑆𝑆 is linearly independent, consider the following equation

𝑐𝑐1 𝐯𝐯1 + 𝑐𝑐2 𝐯𝐯2 + 𝑐𝑐3 𝐯𝐯3 = 𝟎𝟎

1 0 0 1 0 0 0 0
𝑐𝑐1 � � + 𝑐𝑐2 � � + 𝑐𝑐3 � �=� �
0 0 1 0 0 1 0 0
𝑐𝑐1 𝑐𝑐2 0 0
�𝑐𝑐 𝑐𝑐3 � = �0 0�
2

The last equation is true if and only if 𝑐𝑐1 = 𝑐𝑐2 = 𝑐𝑐3 = 0. This proves that 𝑆𝑆 is linearly
independent. Therefore, 𝑆𝑆 is a basis for 𝑆𝑆2 . Thus, dim(𝑆𝑆2 ) = 3.

Theorem 11.1.3. (Basis Test in an 𝒏𝒏-Dimensional Space)

Let 𝑉𝑉 be a vector space of dimension 𝑛𝑛 and 𝑆𝑆 = {𝐯𝐯1 , 𝐯𝐯2 , … , 𝐯𝐯𝑛𝑛 } be a set of vectors in 𝑉𝑉.

1. If 𝑆𝑆 is linearly independent, then 𝑆𝑆 is a basis for 𝑉𝑉.

2. If 𝑆𝑆 spans 𝑉𝑉, that is span(𝑆𝑆) = 𝑉𝑉, then 𝑆𝑆 is a basis for 𝑉𝑉.

Proof. See Appendix.

Example 11.1.7. Show 𝑆𝑆 is a basis for ℝ3 .

0 −2 1
𝑆𝑆 = ��2�, � 1 �, �1��
1 0 1

Solution. Since dim(ℝ3 ) = 3, we just need to check if 𝑆𝑆 is either linearly independent or

spans ℝ3 . This is left for practice. Please try.
Mohammad Hassan Murad P a g e | 32
Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Row Space, Column Space, and Rank of a Matrix
Let 𝐴𝐴 be an 𝑚𝑚 × 𝑛𝑛 matrix over ℝ.

𝐜𝐜1 𝐜𝐜2 𝐜𝐜𝑛𝑛 𝐫𝐫1𝑇𝑇 𝐫𝐫2𝑇𝑇 𝑇𝑇

𝐫𝐫𝑚𝑚

𝑎𝑎11 𝑎𝑎12 ⋯ 𝑎𝑎1𝑛𝑛 𝐫𝐫1 𝑎𝑎11 𝑎𝑎21 ⋯ 𝑎𝑎𝑚𝑚1 𝐜𝐜1𝑇𝑇

𝑎𝑎21 𝑎𝑎22 ⋯ 𝑎𝑎2𝑛𝑛 𝐫𝐫2 𝑎𝑎12 𝑎𝑎22 ⋯ 𝑎𝑎𝑚𝑚2 𝐜𝐜2𝑇𝑇
𝐴𝐴 = � ⋮ ⋮ � 𝐴𝐴𝑇𝑇 = � ⋮
⋮ ⋮ ⋮ ⋮ ⋮ �
𝑎𝑎𝑚𝑚1 𝑎𝑎𝑚𝑚2 ⋯ 𝑎𝑎𝑚𝑚𝑚𝑚 𝐫𝐫𝑚𝑚 𝑎𝑎1𝑛𝑛 𝑎𝑎2𝑛𝑛 ⋯ 𝑎𝑎𝑚𝑚𝑚𝑚 𝐜𝐜𝑛𝑛𝑇𝑇

Definition 11.2.1.

The vectors 𝐫𝐫1 = (𝑎𝑎11 , 𝑎𝑎12 , … , 𝑎𝑎1𝑛𝑛 ), 𝐫𝐫2 = (𝑎𝑎21 , 𝑎𝑎22 , … , 𝑎𝑎2𝑛𝑛 ), … , 𝐫𝐫𝑚𝑚 = (𝑎𝑎𝑚𝑚1 , 𝑎𝑎𝑚𝑚2 , … , 𝑎𝑎𝑚𝑚𝑚𝑚 ) are
called the row vectors of 𝐴𝐴.
𝑎𝑎11 𝑎𝑎12 𝑎𝑎1𝑛𝑛
𝑎𝑎21 𝑎𝑎22 𝑎𝑎2𝑛𝑛
Similarly, the vectors 𝐜𝐜1 = � ⋮ � , 𝐜𝐜2 = � ⋮ � , … , 𝐜𝐜𝑛𝑛 = � ⋮ � are called the column
𝑎𝑎𝑚𝑚1 𝑎𝑎𝑚𝑚2 𝑎𝑎𝑚𝑚𝑚𝑚
vectors of 𝐴𝐴.

Note that 𝐫𝐫𝑖𝑖 ∈ ℝ𝑛𝑛 for 𝑖𝑖 = 1, 2, … , 𝑚𝑚 and 𝐜𝐜𝑖𝑖 ∈ ℝ𝑚𝑚 for 𝑖𝑖 = 1, 2, … , 𝑛𝑛.

Definition 11.2.2.

1. The row space of 𝐴𝐴, 𝑅𝑅(𝐴𝐴), is the subspace of ℝ𝑛𝑛 spanned by the row vectors of 𝐴𝐴.

2. The column space of 𝐴𝐴, 𝐶𝐶(𝐴𝐴), is the subspace of ℝ𝑚𝑚 spanned by the column vectors of 𝐴𝐴.

In other words, if 𝐫𝐫1 , 𝐫𝐫2 , … , 𝐫𝐫𝑚𝑚 and 𝐜𝐜1 , 𝐜𝐜2 , … , 𝐜𝐜𝑛𝑛 are the row and column vectors of 𝐴𝐴,
respectively, then

𝑅𝑅(𝐴𝐴) = rowspace(𝐴𝐴) = span{𝐫𝐫1 , 𝐫𝐫2 , … , 𝐫𝐫𝑚𝑚 },

𝐶𝐶(𝐴𝐴) = colspace(𝐴𝐴) = span{𝐜𝐜1 , 𝐜𝐜2 , … , 𝐜𝐜𝑛𝑛 }.

Observe that 𝐶𝐶(𝐴𝐴) = 𝑅𝑅(𝐴𝐴𝑇𝑇 ). In other words, the column space of 𝐴𝐴 is same as the row space
of 𝐴𝐴𝑇𝑇 ,

Mohammad Hassan Murad P a g e | 33

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Row Space, Column Space, and Rank of a Matrix
Recall that two matrices are called row equivalent if and only if one can be obtained from
another by a sequence of elementary row operations.

For example, the matrices

1 0 −1 1 0 −1
𝐴𝐴 = �2 1 0 � and 𝐵𝐵 = �6 1 −4�
3 2 1 1 2 3

are row equivalent, since

4𝑅𝑅1 +𝑅𝑅2 →𝑅𝑅2
1 0 −1 −2𝑅𝑅 1 0 −1
1 +𝑅𝑅3 →𝑅𝑅3
𝐴𝐴 = �2 1 0 � �⎯⎯⎯⎯⎯⎯⎯⎯� �6 1 −4� = 𝐵𝐵.
3 2 1 1 2 3

Theorem 11.2.1. If 𝐴𝐴 is row equivalent to 𝐵𝐵, then 𝑅𝑅(𝐴𝐴) = 𝑅𝑅(𝐵𝐵).

Proof. See Appendix.

Example 11.2.1. Finding a basis and dimension of a row space of

1 3 1 3
⎡ 0 1 1 0⎤
⎢ ⎥
𝐴𝐴 = ⎢−3 0 6 −1⎥.
⎢ 3 4 −2 1 ⎥
⎣ 2 0 −4 −2⎦

Solution. Reducing 𝐴𝐴 by applying a sequence of elementary row operations to its row

echelon form, we obtain

1 3 1 3 3𝑅𝑅1 +𝑅𝑅3 →𝑅𝑅3 1 3 1 3 −9𝑅𝑅2+𝑅𝑅3→𝑅𝑅3 1 3 1 3

⎡ 0 ⎤ −3𝑅𝑅1 +𝑅𝑅4 →𝑅𝑅4 ⎡
1 1 0 −2𝑅𝑅 +𝑅𝑅 →𝑅𝑅 0 1 1 0 ⎤ 5𝑅𝑅 2 +𝑅𝑅4 →𝑅𝑅4 ⎡0 1 1 0⎤
⎢ ⎥ 1 5 5 ⎢ ⎥ 6𝑅𝑅2+𝑅𝑅5→𝑅𝑅5 ⎢ ⎥
𝐴𝐴 = ⎢−3 0 6 −1⎥ �⎯⎯⎯⎯⎯⎯⎯⎯� ⎢0 9 9 8 ⎥ �⎯⎯⎯⎯⎯⎯⎯⎯� ⎢0 0 0 8⎥
⎢ 3 4 −2 1 ⎥ ⎢0 −5 −5 −8⎥ ⎢0 0 0 −8⎥
⎣ 2 0 −4 −2⎦ ⎣0 −6 −6 −8⎦ ⎣0 0 0 −8⎦
1 3 1 3 1 3 1 3
1 ⎡0 1 ⎤ 8𝑅𝑅3 +𝑅𝑅4 →𝑅𝑅4 ⎡
1 0 8𝑅𝑅 +𝑅𝑅 →𝑅𝑅 0 1 1 0⎤
𝑅𝑅 →𝑅𝑅3
8 3 ⎢ ⎥ 3 5 5 ⎢ ⎥
�⎯⎯⎯⎯� ⎢0 0 0 1 ⎥ �⎯⎯⎯⎯⎯⎯⎯� ⎢0 0 0 1⎥ = 𝐵𝐵
⎢0 0 0 −8⎥ ⎢0 0 0 0⎥
⎣0 0 0 −8⎦ ⎣0 0 0 0⎦

Mohammad Hassan Murad P a g e | 34

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Row Space, Column Space, and Rank of a Matrix
The non-zero row vectors of 𝐵𝐵 in row echelon form are:

𝐰𝐰1 = (1, 3, 1, 3), 𝐰𝐰2 = (0, 1, 1, 0), 𝐰𝐰3 = (0, 0, 0, 1)

Hence 𝑆𝑆 = {𝐰𝐰1 , 𝐰𝐰2 , 𝐰𝐰3 } forms a basis for 𝑅𝑅(𝐵𝐵). Thus, by the Theorem 11.2.1, 𝑆𝑆 is a basis for
𝑅𝑅(𝐴𝐴). And dim�𝑅𝑅(𝐴𝐴)� = 3.

Example 11.2.2. Find a basis and dimension of the column space of

1 3 1 3
⎡ 0 1 1 0⎤
⎢ ⎥
𝐴𝐴 = ⎢−3 0 6 −1⎥.
⎢ 3 4 −2 1 ⎥
⎣ 2 0 −4 −2⎦

Solution. Since the column space of 𝐴𝐴 is same as the row space of 𝐴𝐴𝑇𝑇 , by reducing 𝐴𝐴𝑇𝑇 to its
row echelon form, we obtain
−3𝑅𝑅 +𝑅𝑅 →𝑅𝑅
1 0 −3 3 2 −𝑅𝑅11+𝑅𝑅32→𝑅𝑅32 1 0 −3 3 2
3 1 0 4 0 −3𝑅𝑅1+𝑅𝑅3→𝑅𝑅3 0 1 9 −5 −6
𝐴𝐴𝑇𝑇 = � � �⎯⎯⎯⎯⎯⎯⎯⎯� � �
1 1 6 −2 −4 0 1 9 −5 −6
3 0 −1 1 2 0 0 8 −8 −4
1 0 −3 3 2 1 0 −3 3 2
0
−𝑅𝑅2 +𝑅𝑅3 →𝑅𝑅3 1 9 −5 −6 𝑅𝑅3↔𝑅𝑅4 0 1 9 −5 −6
�⎯⎯⎯⎯⎯⎯⎯� � � �⎯⎯⎯� � � = 𝐵𝐵
0 0 0 0 0 0 0 8 −8 −4
0 0 8 −8 −4 0 0 0 0 0

The non-zero row vectors of 𝐵𝐵 in row echelon form are:

𝐰𝐰1 = (1, 0, −3, 3, 2), 𝐰𝐰2 = (0, 1, 9, −5, −6), 𝐰𝐰3 = (0, 0, 8, −8, −4).

Hence 𝑆𝑆 = {𝐰𝐰1 , 𝐰𝐰2 , 𝐰𝐰3 } forms a basis for 𝑅𝑅(𝐵𝐵). Thus, by the Theorem 11.2.1, 𝑆𝑆 is a basis for
𝑅𝑅(𝐴𝐴𝑇𝑇 ). This is equivalent to saying that 𝑆𝑆 ′ = {𝐯𝐯1 , 𝐯𝐯2 , 𝐯𝐯3 } forms a basis for 𝐶𝐶(𝐴𝐴) = 𝑅𝑅(𝐴𝐴𝑇𝑇 )
where

1 0 0
⎡ 0⎤ ⎡ 1⎤ ⎡ 0⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
𝐯𝐯1 = 𝐰𝐰1𝑇𝑇 = ⎢−3⎥ , 𝐯𝐯2 = 𝐰𝐰2𝑇𝑇 = ⎢ 9 ⎥ , 𝐯𝐯3 = 𝐰𝐰3𝑇𝑇 = ⎢ 8 ⎥.
⎢ 3⎥ ⎢−5⎥ ⎢−8⎥
⎣ 2⎦ ⎣−6⎦ ⎣−4⎦

And dim�𝐶𝐶(𝐴𝐴)� = 3.

Mohammad Hassan Murad P a g e | 35

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University
 Row Space, Column Space, and Rank of a Matrix
Definition 11.2.3. The dimension of row space of a matrix 𝐴𝐴 is called the row rank of 𝐴𝐴,
and the dimension of column space of a matrix 𝐴𝐴 is called the column rank of 𝐴𝐴.

Theorem 11.2.2. For any 𝑚𝑚 × 𝑛𝑛 matrix 𝐴𝐴,

dim�𝑅𝑅(𝐴𝐴)� = dim�𝐶𝐶(𝐴𝐴)�.

In other words, row rank of a matrix is equal to its column rank.

Proof. Please consult the textbook.

Definition 11.2.4. The dimension of the row (or column) space of a matrix is called the rank
of 𝐴𝐴 and is denoted by rank(𝐴𝐴).

Example 11.2.3. Find the rank of the matrix of

1 3 1 3
⎡ 0 1 1 0⎤
⎢ ⎥
𝐴𝐴 = ⎢−3 0 6 −1⎥.
⎢ 3 4 −2 1 ⎥
⎣ 2 0 −4 −2⎦

Solution. From Example 11.2.1 and Example 11.2.2, we find that

dim�𝑅𝑅(𝐴𝐴)� = dim�𝐶𝐶(𝐴𝐴)� = 3.

Hence rank(𝐴𝐴) = 3.

Mohammad Hassan Murad P a g e | 36

Assistant Professor
Department of Mathematics and Natural Sciences
Brac University