Vector space

Example for vector space:

1. Check whether the set 𝑉 = 𝑥, 𝑦 𝑥, 𝑦 ∈ ℝ} with the vector addition defined by
𝑥1 , 𝑦1 + 𝑥2 , 𝑦2 = (𝑥1 𝑦2 , 𝑥2 𝑦1 ) and the scalar multiplication defined by
𝑐 𝑥, 𝑦 = (𝑐𝑥, 𝑐𝑦) is a vector space.
Answer: NO
You have already verified that V is closed with respect to addition and scalar multiplication.
It contains additive identity and additive inverse.
Check for commutativity..
For 1, 2 , 3, 4 ∈ 𝑉, 1, 2 + 3, 4 = 1 × 4, 2 × 3 = (4, 6)
3, 4 + 1, 2 = 3 × 2, 4 × 1 = (6, 4)
But, 4, 6 ≠ 6, 4 . So commutativity fails in V.
Therefore, V is not a vector space.
2. Let 𝑉 = 0 , be a set of only one element “0”. This set is also a vector space over ℝ.

Note: Elements of the vector space are called vectors.

Problems:

𝑎 𝑏
1. Prove that the set 𝑉 = 𝑎, 𝑏, 𝑐, 𝑑 ∈ ℝ is a vector space over ℝ with respect usual
𝑐 𝑑
matrix addition and scalar multiplication.

2. Prove that the set of polynomials with real coefficients of degree less than or equal to n is a
vector space.
 Subspaces:
A subset 𝐻 of 𝑉 is called a subspace of 𝑉 if 𝐻 is also a vector space (using the same
addition and scalar multiplication as on 𝑉) .
For example,

𝑥 0
The set U = 𝑥, 𝑦 ∈ ℝ is a vector space over ℝ with respect usual matrix
0 𝑦
addition and scalar multiplication, which is a subset of the vector space

𝑎 𝑏
𝑉= 𝑎, 𝑏, 𝑐, 𝑑 ∈ ℝ .
𝑐 𝑑
So, 𝑈 is a subspace of 𝑉.
If 𝐻 is a subset of V, then to check that 𝐻 is a subspace of V we need only check that 𝐻 satisfies
the following:
additive identity 𝟎 ∈𝐻
closed under addition 𝑢, 𝑣 ∈ 𝐻 𝑖𝑚𝑝𝑙𝑖𝑒𝑠 𝑢 + 𝑣 ∈ 𝐻;
closed under scalar multiplication 𝑐 ∈ ℝ 𝑎𝑛𝑑 𝑢 ∈ 𝐻 𝑖𝑚𝑝𝑙𝑖𝑒𝑠 𝑐𝑢 ∈ 𝐻;

So, More formally we can define subspace as:

Definition: A nonempty set 𝐻 of a vector space 𝑉 is said to be a subspace if,

(i) 𝟎 ∈ 𝐻
(ii) For each 𝒖 and 𝒗 in 𝐻, the sum 𝒖 + 𝒗 is in 𝐻.
(iii) For each 𝒖 in 𝐻 and each scalar 𝑐, the vector 𝑐𝒖 is in 𝐻.
Examples:
1. For any vector space 𝑉, 0 and 𝑉 are always subspaces.

2. Let 𝑊 = 𝑥, 0 𝑥 ∈ ℝ } be a subspace of ℝ2 .
Because,
0, 0 ∈ 𝑊
For 𝑥1 , 0 , 𝑥2 , 0 ∈ 𝑊,
𝑥1 , 0 + 𝑥2 , 0 = 𝑥1 + 𝑥2 , 0 ∈ 𝑊.
For a scalar 𝑐 and 𝑥, 0 ∈ 𝑊,
𝑐 𝑥, 0 = 𝑐𝑥, 0 ∈ 𝑊.
There fore W is a subspace.
3. Let 𝑊 = 0, 𝑦 𝑦 ∈ ℝ } be also a subspace of ℝ2 (verify).
4. Check whether the subset 𝑈 = 𝑥, 𝑦 ∈ ℝ2 𝑥 2 + 𝑦 2 = 1} is a subspace?
Answer: No (why??)…
Because,
U is set of all points in circumference of the unit circle 𝑥 2 + 𝑦 2 = 1,

It is not passing through the origin.

So, 0, 0 ∉ 𝑈. Thus U is not a subspace.
 Thus the Subspaces of ℝ2 are
(i) {0}
(ii) ℝ2
(iii) All the lines passing through the origin.

Subspaces of ℝ3 are:
(i) {0}
(ii) ℝ3
(iii) All the lines passing through the origin
(iv) All the planes passing through the origin.
Problems:
𝑥 −𝑥
1. Let 𝑉 be a vector space of all 2 × 2 matrices then show that 𝑊 = 𝑦 𝑧 𝑥, 𝑦, 𝑧 ∈ ℝ
is a subspace of 𝑉.

0 0
Solution: Clearly, ∈𝑊
0 0
𝑥1 −𝑥1 𝑥2 −𝑥2
Now, for 𝑦1 𝑧1 , 𝑦2 𝑧2 ∈ 𝑊,

𝑥1 −𝑥1 𝑥2 −𝑥2 𝑥1 + 𝑥2 −(𝑥1 + 𝑥2 )

𝑦1 𝑧1 + 𝑦2 𝑧2 = 𝑦1 + 𝑦2 𝑧1 + 𝑧2
∈𝑊

Thus 𝑊 is closed under addition.

𝑥 −𝑥 𝑥 −𝑥 𝑐𝑥 −𝑐𝑥
Let 𝑐 ∈ ℝ and 𝑦 𝑧 ∈ 𝑊, then 𝑐 𝑦 𝑧 = 𝑐𝑦 𝑐𝑧 ∈ 𝑊
𝑊 is also closed under scalar multiplication. Hence 𝑊 is a subspace of 𝑉.
 2. Let 𝑋 = 1, 2, −3 , 𝑌 = (−2, 3, 0) be two vectors in 𝑅3 and 𝑊 be the set of vectors of the
form 𝑎𝑋 + 𝑏𝑌 where 𝑎 and 𝑏 are real numbers. Show that 𝑊 is a subspace of 𝑅3 .
Solution:
Here 𝑊 = 𝑎𝑋 + 𝑏𝑌 𝑎, 𝑏 ∈ ℝ}
Since 0 = 0 ∙ 𝑋 + 0 ∙ 𝑌 , hence 0 ∈ 𝑊.
Now let 𝑢, 𝑣 ∈ 𝑊,
this implies 𝑢 = 𝑎1 𝑋 + 𝑏1 𝑌 = 𝑎1 1, 2, −3 + 𝑏1 (−2, 3, 0) for some scalars 𝑎1 , 𝑏1
and 𝑣 = 𝑎2 𝑋 + 𝑏2 𝑌 = 𝑎2 1, 2, −3 + 𝑏2 (−2, 3, 0) for some scalars 𝑎2 , 𝑏2 .
Then 𝑢 + 𝑣 = 𝑎1 1, 2, −3 + 𝑏1 −2, 3, 0 + 𝑎2 1, 2, −3 + 𝑏2 (−2, 3, 0)
= (𝑎1 +𝑎2 ) ∙ 1, 2, −3 + (𝑏1 +𝑏2 ) ∙ (−2, 3, 0)
= (𝑎1 +𝑎2 )𝑋 + (𝑏1 +𝑏2 )𝑌 ∈ 𝑊.
Similarly, we can prove 𝑊 is closed under scalar multiplication.
Theorem 1: A nonempty subset 𝑊 of a vector space 𝑉 over 𝑅 is a subspace if and only if for
any vectors 𝑢, 𝑣 ∈ 𝑊 and any scalar 𝑐, 𝑐𝑢 + 𝑣 ∈ 𝑊.
Proof:
Suppose that 𝑊 is a nonempty subset a vector space 𝑉 such that 𝑐𝑢 + 𝑣 ∈ 𝑊 for all vectors
𝑢, 𝑣 ∈ 𝑊 and all scalar 𝑐 in 𝑅.
Since 𝑊 is non-empty, there is a vector 𝑤 in 𝑊, and hence
−1 𝑤 + 𝑤 = −𝑤 + 𝑤 = 𝟎 is in 𝑾.
Then, for any vector 𝑤 ∈ 𝑊 and any scalar 𝑐, 𝑐𝑤 + 0 = 𝒄𝒘 ∈ 𝑾.
Let 𝑢, 𝑣 ∈ 𝑊, then 1𝑢 + 𝑣 = 𝒖 + 𝒗 ∈ 𝑾.
Therefore 𝑾 is a subspace of 𝑽.
Conversely,
If 𝑊 is a subspace of 𝑉, then for any scalar 𝑐 and any vectors 𝑢, 𝑣 ∈ 𝑊,
 𝑐𝑢 ∈ 𝑊 and hence 𝒄𝒖 + 𝒗 ∈ 𝑾.
This prove the above theorem.

Note: We can also use property 𝑐𝑢 + 𝑣 in the theorem 1 as a definition of a subspace.

Theorem 2: If 𝐻 and 𝐾 are subspaces of 𝑉, then 𝐻 ∩ 𝐾 is also a subspace of 𝑉.
Proof: Since 𝐻 and 𝐾 are subspaces, 0 ∈ 𝐻 and 0 ∈ 𝐾.
Hence 𝟎 ∈ 𝑯 ∩ 𝑲.
Suppose that 𝑢, 𝑣 ∈ 𝐻 ∩ 𝐾.
This implies that, 𝑢, 𝑣 ∈ 𝐻 and 𝑢, 𝑣 ∈ 𝐾
Since 𝐻 and 𝐾 are subspaces, 𝑢 + 𝑣 ∈ 𝐻 and 𝑢 + 𝑣 ∈ 𝐾, hence 𝑢 + 𝑣 ∈ 𝐻 ∩ 𝐾.
Therefore, 𝑯 ∩ 𝑲 is closed under addition.
Let 𝑘 ∈ ℝ and 𝑢 ∈ 𝐻 ∩ 𝐾, this implies that 𝑢 ∈ 𝐻 and 𝑢 ∈ 𝐾.
Since 𝐻 and 𝐾 are subspaces, 𝑘𝑢 ∈ 𝐻 and 𝑘𝑢 ∈ 𝐾, hence 𝑘𝑢 ∈ 𝐻 ∩ 𝐾.
Thus 𝑯 ∩ 𝑲 is closed under scalar multiplication.
Therefore 𝑯 ∩ 𝑲 is a subspace of 𝑽.
 Note: Is 𝐻 ∪ 𝐾 also a subspace of 𝑉 ?
Consider two subspaces 𝐻 = 𝑥, 0 𝑥 ∈ ℝ } and
𝐾= 0, 𝑦 𝑦 ∈ ℝ } of ℝ2 .
Now, (2, 0) and (0, 3) are in 𝐻 ∪ 𝐾,
But, 𝟎, 𝟐 + 𝟎, 𝟑 = (𝟐, 𝟑) ∉ 𝑯 ∪ 𝑲.
Therefore, 𝑯 ∪ 𝑲 need not be a subspace.

Theorem 3: If 𝑊1 and 𝑊2 are subspaces of a vector space 𝑉, then the set

𝑊1 + 𝑊2 = 𝑤1 + 𝑤2 𝑤1 ∈ 𝑊1 𝑎𝑛𝑑 𝑤2 ∈ 𝑊2 } is also a subspace.
(Prove this)
Theorem 3: If 𝑊1 and 𝑊2 are subspaces of a vector space 𝑉, then the set
𝑊1 + 𝑊2 = 𝑤1 + 𝑤2 𝑤1 ∈ 𝑊1 𝑎𝑛𝑑 𝑤2 ∈ 𝑊2 } is also a subspace.
Proof: Since 𝑊1 and 𝑊2 are subspaces, 0 ∈ 𝑊1 and 0 ∈ 𝑊2 .
Hence 𝟎 = 𝟎 + 𝟎 ∈ 𝑾𝟏 + 𝑾𝟐 .
Suppose that 𝑢, 𝑣 ∈ 𝑊1 + 𝑊2 .
Then, 𝑢 = 𝑤1 + 𝑤2 , 𝑣 = 𝑤1′ + 𝑤2′ for some 𝑤1 , 𝑤1′ ∈ 𝑊1 and 𝑤2 , 𝑤2′ ∈ 𝑊2 .
Hence 𝑢 + 𝑣 = (𝑤1 +𝑤2 ) + 𝑤1′ + 𝑤2′ = 𝑤1 + 𝑤1′ + 𝑤2 + 𝑤2′ ∈ 𝑊1 + 𝑊2
Because 𝑊1 and 𝑊2 are subspaces, 𝑤1 + 𝑤1′ ∈ 𝑊1 and 𝑤2 + 𝑤2′ ∈ 𝑊2 .
Let 𝑘 ∈ ℝ and 𝑢 ∈ 𝑊1 + 𝑊2 , then 𝑢 = 𝑤1 + 𝑤2 , for some 𝑤1 ∈ 𝑊1 and 𝑤2 ∈ 𝑊2 .
Since 𝑊1 and 𝑊2 are subspaces, 𝑘𝑤1 ∈ 𝑊1 and 𝑘𝑤2 ∈ 𝑊2 ,
hence 𝑘𝑢 = 𝑘 𝑤1 + 𝑤2 = 𝑘𝑤1 + 𝑘𝑤2 ∈ 𝑊1 + 𝑊2 .
Therefore 𝑾𝟏 + 𝑾𝟐 is a subspace of 𝑽.
Linear combination:
A linear combination of a set 𝑣1 , . . . , 𝑣𝑛 of vectors in a vector space V is a vector of the
form
𝑐1 𝑣1 + 𝑐2 𝑣2 + ⋯ + 𝑐𝑛 𝑣𝑛 for some scalars 𝑐1 , 𝑐2 , … , 𝑐𝑛 .

For example, Let 𝑣1 = (1, 2, 3) and 𝑣2 = (1, 0, 1) in 𝑅3 , then the vector

3, 6, 11 = 3 1, 2, 3 + 2(1, 0, 1) is a linear combination of the vectors 𝑣1 and 𝑣2 .

Suppose we want to express the vector 𝑢 = 1, −2, 5 as a linear combination of the vectors
𝑢1 = 1, 1, 1 , 𝑢2 = 1, 2, 3 , 𝑢3 = 2, −1, 1 .
We have to find the scalars 𝑐1 , 𝑐2 and 𝑐3 such that 𝑢 = 𝑐1 𝑢1 + 𝑐2 𝑢2 + 𝑐3 𝑢3 ;
that is, 1, −2, 5 = 𝑐1 (1, 1, 1) + 𝑐2 (1, 2, 3) + 𝑐3 2, −1, 1 .
 1 1 1 2 𝑐1 𝑐2 2𝑐3
or −2 = 𝑐1 1 + 𝑐2 2 + 𝑐3 −1 = 𝑐1 + 2𝑐2 + −𝑐3
5 1 3 1 𝑐1 3𝑐2 𝑐3

1 𝑐1 + 𝑐2 + 2𝑐3
−2 = 𝑐1 +2𝑐2 − 𝑐3
5 𝑐1 +3𝑐2 + 𝑐3
The vectors on the left hand and right hand sides of (2) are equal if and only if their
corresponding entries are both equal. That is,

1 = 𝑐1 + 𝑐2 + 2𝑐3
−2 = 𝑐1 +2𝑐2 − 𝑐3
5 = 𝑐1 + 3𝑐2 + 𝑐3
We solve this system by row reducing the augmented matrix of the system as follows:
 1 1 2 1 1 1 2 1 1 1 2 1
1 2 −1 −2 ~ 0 1 −3 −3 ~ 0 1 −3 −3
1 3 1 5 0 2 −1 4 0 0 5 10
So equivalent system is:

𝑐1 + 𝑐2 + 2𝑐3 = 1
𝑐2 − 3𝑐3 = −3
5𝑐3 = 10
Back-substitution yields the solution: 𝑐3 = 2, 𝑐2 = −3 + 3𝑐3 = −3 + 6 = 3 and
𝑐1 = 1 − 𝑐2 − 2𝑐3 = 1 − 3 − 2 2 = −6
Thus, 𝑢 = −6𝑢1 + 3𝑢2 + 2𝑢3 .
Note: In the above example we can observe that, the vector equation
𝒗 = 𝑐1 𝑢1 + 𝑐2 𝑢2 + ⋯ + 𝑐𝑛 𝑢𝑛
has the same solution as the linear system whose augmented matrix is
[𝑢1 𝑢2 … 𝑢𝑛 𝒗] −−− −(1)
In particular, 𝒗 can be expressed as a linear combination of 𝑢1 , 𝑢2 , … , 𝑢𝑛 if and only if there
exists a solution to the linear system corresponding to 1 (that is, the the linear system
corresponding to 1 is consistent).

Solved Problems:
1. Can 𝑣 = (2, −5, 3) in 𝑅3 be represented as a linear combination of the vectors
𝑢1 = 1, −3, 2 , 𝑢2 = 2, −4, −1 , 𝑢3 = 1, −5, 7 .
Solution: The vector 𝑣 can be written as a linear combination of 𝑢1 , 𝑢2 , 𝑢3 if and only if
there exist scalars scalars 𝑐1 , 𝑐2 and 𝑐3 such that 𝑣 = 𝑐1 𝑢1 + 𝑐2 𝑢2 + 𝑐3 𝑢3 ;
that is, 2, −5, 3 = 𝑐1 1, −3, 2 + 𝑐2 2, −4, −1 + 𝑐3 1, −5, 7 .
This vector equation is equivalent to a system of three linear equations in three unknowns.
That is,

2 = 𝑐1 + 2𝑐2 + 𝑐3
−5 = −3𝑐1 −4𝑐2 − 5𝑐3
3 = 2𝑐1 −𝑐2 + 7 𝑐3
If we reduce the augmented matrix for the above system, we get,

1 2 1 2 1 2 1 2 1 2 1 2
−3 −4 −5 −5 ~ 0 2 −2 1 ~ 0 2 −2 1
3
2 −1 7 3 0 −5 5 −1 0 0 0 2
Here, 𝒓𝒂𝒏𝒌 𝒄𝒐𝒆𝒇𝒇𝒊𝒄𝒊𝒆𝒏𝒕 𝒎𝒂𝒕𝒓𝒊𝒙 = 𝟐 ≠ 𝒓𝒂𝒏𝒌 𝒂𝒖𝒈𝒎𝒆𝒏𝒕𝒆𝒅 𝒎𝒂𝒕𝒓𝒊𝒙 = 𝟑.
 Thus the system is inconsistent and so has no solution. Thus, 𝑣 cannot be written as a linear
combination of 𝑢1 , 𝑢2 𝑎𝑛𝑑 𝑢3 .
2. For which value of 𝑘 the vectors 𝑢 = (1, −2 , 𝑘) in 𝑅3 be a linear combination of the vectors
𝑢1 = 3, 0, −2 and 𝑢2 = (2, −1, −5).
Solution: The vector 𝑢 can be written as a linear combination of 𝑢1 , 𝑢2 , 𝑢3 if and only if there
exist scalars scalars 𝑐1 , 𝑐2 such that 𝑢 = 𝑐1 𝑢1 + 𝑐2 𝑢2 ;
that is, 1, −2, 𝑘 = 𝑐1 3, 0, −2 + 𝑐2 (2, −1, −5)
This vector equation is equivalent to a system of three linear equations in three unknowns.
If we reduce the augmented matrix for this system, we obtain,

3 2 1 3 2 1 3 2 1 3 2 1
0 −1 −2 ~ 0 −1 −2 ~ 0 −1 −2 ~ 0 −1 −2
11 2
−2 −5 𝑘 0 − 𝑘+ 0 −11 3𝑘 + 2 0 0 3𝑘 + 24
3 3
 The above system is consistent if and only if 3𝑘 − 24 = 0, that is 𝑘 = 8. Hence, for 𝑘 = 8 the
vectors 𝑢 = (1, −2 , 𝑘) in 𝑅3 can be written as a linear combination of the vectors
𝑢1 = 3, 0, −2 and 𝑢2 = (2, −1, −5).

3 1 1 1
3. Write the matrix 𝐸 = as a linear combination of the matrices 𝐴 = ,
1 −1 1 0
0 0 0 2
𝐵= 𝑎𝑛𝑑 𝐶 = .
1 1 0 −1
Solution: To write the matrix E as a linear combination of the matrices A, 𝐵 𝑎𝑛𝑑 𝐶 , we have
to find values of 𝑐1 , 𝑐2 𝑎𝑛𝑑 𝑐3 such that 𝐸 = 𝑐1 𝐴 + 𝑐2 𝐵 + 𝑐3 𝐶.

3 1 1 1 0 0 0 2 𝑐1 𝑐1 + 2𝑐3
That is, = 𝑐1 + 𝑐2 + 𝑐3 =
1 −1 1 0 1 1 0 −1 𝑐1 + 𝑐2 𝑐2 − 𝑐3
 3 1 𝑐1 𝑐1 + 2𝑐3
=
1 −1 𝑐1 + 𝑐2 𝑐2 − 𝑐3
this implies, 3 = 𝑐1 , 1 = 𝑐1 + 2𝑐3 , 1 = 𝑐1 + 𝑐2 and −1 = 𝑐2 − 𝑐3
Solving above equations, we get, 𝑐1 = 3, 𝑐3 = −1 and 𝑐2 = −2.
Therefore, 𝐸 = 3𝐴 − 2𝐵 − 1𝐶 .

1 2 7
4. Let 𝑎1 = −2 , 𝑎2 = 5 , 𝑎𝑛𝑑 𝑏 = 4 . Determine whether 𝑏 can be written as a linear
−5 6 −3
combination 𝑎1 𝑎𝑛𝑑 𝑎2 . (Try this).
 Span of vectors:
If 𝑣1 , 𝑣2 , … , 𝑣𝑘 are vectors in a vector space 𝑉, then the set of all linear combinations of
𝑣1 , 𝑣2 , … , 𝑣𝑘 is called span (or linear span) of the vectors 𝑣1 , 𝑣2 , … , 𝑣𝑘 . It is denoted by
Span{𝒗𝟏 , 𝒗𝟐 , … , 𝒗𝒌 }.
That is, Span{𝑣1 , 𝑣2 , … , 𝑣𝑘 } is the collection of all vectors that can be written in the form
𝑐1 𝑣1 + 𝑐2 𝑣2 + ⋯ + 𝑐𝑘 𝑣𝑘 with 𝑐1 , 𝑐2 , … , 𝑐𝑘 scalars.
Therefore,
Span{𝑣1 , 𝑣2 , … , 𝑣𝑘 } = {𝑐1 𝑣1 + 𝑐2 𝑣2 + ⋯ + 𝑐𝑘 𝑣𝑘 𝑐1 , 𝑐2 , … , 𝑐𝑘 ∈ 𝑅 .

Note: Asking whether a vector 𝑏 is in Span{𝑣1 , 𝑣2 , … , 𝑣𝑘 } means to asking whether the vector
equation 𝑐1 𝑣1 + 𝑐2 𝑣2 + ⋯ + 𝑐𝑘 𝑣𝑘 = 𝑏 has a solution, or, equivalently, asking whether the linear
system with augmented matrix 𝑣1 , 𝑣2 , … , 𝑣𝑘 𝑏 has a solution.
Theorem:
If 𝑣1 , 𝑣2 are vectors in a vector space 𝑉, then the set 𝐻 = 𝑆𝑝𝑎𝑛{𝑣1 , 𝑣2 } is a subspace of 𝑉.
Proof:
The zero vector is in 𝐻, because 0 = 0𝑣1 + 0𝑣2 .
To prove that 𝐻 is closed under the vector addition, take two vectors in 𝐻,
that is, let 𝑢, 𝑤 ∈ 𝐻, then 𝑢 = 𝑐1 𝑣1 + 𝑐2 𝑣2 for some scalars 𝑐1 𝑎𝑛𝑑 𝑐2
and 𝑤 = 𝑑1 𝑣1 + 𝑑2 𝑣2 for some scalars 𝑑1 𝑎𝑛𝑑 𝑑2
Now, 𝑢 + 𝑤 = 𝑐1 𝑣1 + 𝑐2 𝑣2 + 𝑑1 𝑣1 + 𝑑2 𝑣2
= 𝑐1 + 𝑑1 𝑣1 + (𝑐2 + 𝑑2 ) 𝑣2 . So, 𝑢 + 𝑤 ∈ 𝐻.
Let 𝑘 be a scalar and 𝑢 ∈ 𝐻, then 𝑢 = 𝑐1 𝑣1 + 𝑐2 𝑣2 for some scalars 𝑐1 𝑎𝑛𝑑 𝑐2
Now, 𝑘𝑢 = 𝑘(𝑐1 𝑣1 + 𝑐2 𝑣2 ) = 𝑘𝑐1 𝑣1 + 𝑘𝑐2 𝑣2 which shows that 𝑐𝑢 is in 𝐻.
Thus 𝐻 is a subspace of V.
Note : We can generalize the above theorem for 𝑛 vectors, as:
If 𝑣1 , 𝑣2 , … , 𝑣𝑛 are vectors in a vector space 𝑉, then the set 𝐻 = 𝑆𝑝𝑎𝑛{𝑣1 , 𝑣2 , … , 𝑣𝑛 } is a
subspace of 𝑉. (Prove this !!).
A Geometric Description of 𝑺𝒑𝒂𝒏{𝒗} and 𝑺𝒑𝒂𝒏{𝒖, 𝒗}:
Let 𝒗 be a nonzero vector in 𝑅3 . Then 𝑆𝑝𝑎𝑛{𝒗} is the set of all scalar multiples of 𝒗, which
is the set of points on the line in 𝑅3 through 𝒗 and 𝟎. See figure 1.
If 𝒖 and 𝒗 are nonzero vectors in 𝑅3 , with 𝒗 not a multiple of 𝒖, then Span{u, v} is the plane
in 𝑅3 that contains 𝒖, 𝒗, and 𝟎. In particular, Span{u, v } contains the line in 𝑅3 through 𝒖 and
𝟎 and the line through 𝒗 and 𝟎. See Figure 2.
Linearly independent and linearly dependent vectors:
Let 𝑉 be a vector space over 𝑅, the set {𝑣1 , 𝑣2 , … , 𝑣𝑛 } of vectors is said to be linearly
dependent there exist scalars 𝑐1 , 𝑐2 , … , 𝑐𝑛 in 𝑅, not all of them zero, such that
𝑐1 𝑣1 + 𝑐2 𝑣2 + ⋯ + 𝑐𝑛 𝑣𝑛 = 0

For example, { 2, 3, 1 , 1, −1, 2 , 7, 3, 8 } is linearly dependent set in 𝑅3 because

2(2, 3, 1) + 3(1, −1, 2) + (−1)(7, 3, 8) = (0, 0, 0).

The set {𝑣1 , 𝑣2 , … , 𝑣𝑛 } of vectors in 𝑉 is said to be linearly independent the vector

equation 𝑐1 𝑣1 + 𝑐2 𝑣2 + ⋯ + 𝑐𝑛 𝑣𝑛 = 0 has only the trivial solution, that is 𝑐1 = 0,
𝑐2 = 0, … , 𝑐𝑛 = 0.
 -or-
The set {𝑣1 , 𝑣2 , … , 𝑣𝑛 } of vectors in V is called linearly independent if the only choice of
𝑐1 , 𝑐2 , . . . , 𝑐𝑛 ∈ 𝑹 that makes c1 𝑣1 + 𝑐2 𝑣2 + · · · +𝑐𝑛 𝑣𝑛 equal 0 is 𝑐1 = 𝑐2 = · · · = 𝑐𝑛 = 0 .

For example,
{ 1, 0, 0 , 0, 1, 0 , 0,0, 1 } is linearly independent set in 𝑅3 because
𝑐1 1, 0, 0 , +𝑐2 0, 1, 0 + 𝑐3 0,0, 1 = 0, 0, 0 ,
this implies, 𝑐1 , 𝑐2 , 𝑐3 = 0, 0, 0 ,
hence 𝑐1 = 0, 𝑐2 = 0 and 𝑐3 = 0.

Note: Any set of vectors containing the 0 vector is linearly dependent (why?).
Pivots are non zero leading entry of a row in an echelon form of the matrix.
Pivot columns are the columns of A (original matrix) that contains a pivot.

1 2 1 2 1 2 1 2
Example: Let A = −3 −4 −5 −5 and 𝐵 = 0 2 −2 1 be an echelon form of 𝐴, then
3
2 −1 7 3 0 0 0 2
 How to check the given set of vectors are linearly independent ?
Answer: A practical way of checking whether vectors 𝑣1 , 𝑣2 , … , 𝑣𝑛 in a vector space 𝑉 are linearly
independent is to use Gaussian elimination: Write all vectors as columns of a matrix A and perform
Gaussian elimination until the matrix is in row echelon form:
– The pivot columns indicate the vectors, which are linearly independent of the vectors on the
left.
– The non-pivot columns can be expressed as linear combinations of the pivot columns on
their left. For instance, the row-echelon form

1 3 0
0 0 2
tells us that the first and third columns are pivot columns. The second column is a non-pivot column
because it is three times the first column.
All column vectors are linearly independent if and only if all columns are pivot columns. If there is at
least one non-pivot column, the columns (and, therefore, the corresponding vectors) are linearly dependent.
Check whether the following set of vectors is linearly dependent or linearly independent.
i) { 1, 4, 5 , 4, 4, 8 , 3, −3, 0 }
Solution: Let 𝑢1 = 1, 4, 5 , 𝑢2 = 4, 4, 8 𝑎𝑛𝑑 𝑢3 = 3, −3, 0
Method 1:
Set a linear combination of the vectors 𝑢1 , 𝑢2 , 𝑢3 equal to the zero vector using unknowns 𝑐1 ,
𝑐2 , 𝑐3 to obtain the equivalent system of linear equations and then reduce the system to
echelon form. This yields
𝑐1 𝑢1 + 𝑐2 𝑢2 + 𝑐3 𝑢3 = 0,
that is, 𝑐1 1, 4, 5 + 𝑐2 4, 4, 8 + 𝑐3 (3, −3, 0) = (0, 0, 0)
or 𝑐1 , 4𝑐1 , 5𝑐1 + 4𝑐2 , 4𝑐2 , 8𝑐2 + 3𝑐3 , −3𝑐3 , 0 = 0, 0, 0
𝑐1 + 4𝑐2 + 3𝑐3 , 4𝑐1 + 4𝑐2 − 3𝑐3 , 5𝑐1 + 8𝑐2 = (0, 0, 0)
Thus 𝑐1 + 4𝑐2 + 3𝑐3 = 0
4𝑐1 + 4𝑐2 − 3𝑐3 = 0
5𝑐1 + 8𝑐2 = 0

1 4 3 𝑐1 0
Matrix of the above system is: 4 4 −3 𝑐2 = 0
5 8 0 𝑐3 0
𝑅2 ↔ 𝑅2 − 4𝑅1
𝑅3 ↔ 𝑅3 − 5𝑅1
1 4 3 𝑐1 0
~ 0 −12 −15 𝑐2 = 0
0 −12 −15 𝑐3 0
𝑅3 ↔ 𝑅3 − 𝑅2
1 4 3 𝑐1 0
~ 0 −12 −15 𝑐2 = 0
0 0 0 𝑐3 0
which is the echelon form of the original system.
 or 𝑐1 + 4𝑐2 + 3𝑐3 = 0
−12𝑐2 − 15𝑐3 = 0
The echelon system has only two nonzero equations in three unknowns; hence, we can take
5
arbitrary value for 𝑐3 . If we take 𝑐3 = 1, then we get, 𝑐2 = − and 𝑐1 = 2, which is the non
4

5
zero solution of the above system. Therefore, 2𝑢1 − 𝑢2 + 𝑢3 = 0. Thus, 𝑢1 , 𝑢2 , 𝑢3 are
4

linearly dependent vectors.

Method 2: Form the matrix A whose columns are 𝑢1 , 𝑢2 , 𝑢3 and reduce to echelon form

1 4 3
A= 4 4 −3
5 8 0
𝑅2 ↔ 𝑅2 − 4𝑅1
𝑅3 ↔ 𝑅3 − 5𝑅1
1 4 3
~ 0 −12 −15
0 −12 −15
𝑅3 ↔ 𝑅3 − 𝑅2
1 4 3
~ 0 −12 −15
0 0 0

The third column does not have a pivot; hence, the third vector 𝑢3 is a linear combination
of the first two vectors 𝑢1 and 𝑢2 . Thus, the vectors are linearly dependent. (Observe that the
matrix A is also the coefficient matrix in Method 1. In other words, this method is essentially
the same as the first method.)
ii) { 1, 0, −1, 2 , 4, 2, 0, −1 , 6, 4, −2, 3 }
Solution: Let 𝑢1 = 1, 0, −1, 2 , 𝑢2 = 4, 2, 0, −1 𝑎𝑛𝑑 𝑢3 = 6, 4, −2, 3 .
Form the matrix A whose columns are 𝑢1 , 𝑢2 , 𝑢3 and reduce to echelon form:

1 4 6
A= 0 2 4
−1 0 −2
2 −1 3
9
𝑅3 ↔ 𝑅3 + 𝑅1 𝑅3 ↔ 𝑅3 − 2𝑅2 𝑅4 ↔ 𝑅4 + 4 𝑅2
𝑅4 ↔ 𝑅4 − 2𝑅1 1 4 6 9
𝑅4 ↔ 𝑅4 + 2 𝑅2 1 4 6 1 4 6
~ 0 2 4 ~ 0 2 4 ~ 0 2 4
0 4 4 0 0 −4 0 0 −4
0 −9 −9 0 0 9 0 0 0
Every column of 𝐴 has a pivot entry; hence, no vector is a linear combination of the previous
vectors. Thus, the vectors are linearly independent.
iii) { 1, 2, −1 , 1, −2, 1 , −3, 2, −1 }
Solution: Let 𝑢1 = 1, 2, −1 , 𝑢2 = 1, −2, 1 𝑎𝑛𝑑 𝑢3 = −3, 2, −1 .
Form the matrix A whose columns are 𝑢1 , 𝑢2 , 𝑢3 and reduce to echelon form:
𝑅2 ↔ 𝑅2 − 2𝑅1
1 1 −3 𝑅3 ↔ 𝑅3 + 𝑅1 1 1 −3
A = 2 −2 2 ~ 0 −4 8
−1 1 −1 0 2 −4

1
𝑅3 ↔ 𝑅3 + 2 𝑅2
1 1 −3
~ 0 −4 8
0 0 0

The third column does not have a pivot; hence, the third vector 𝑢3 is a linear combination of
the first two vectors 𝑢1 and 𝑢2 . Thus, the vectors are linearly dependent.
 1 4 2
iv) Check whether the following vectors 𝑣1 = 2 , 𝑣2 = 5 , 𝑣3 = 1 are. linearly
3 6 0
independent or not.
Solution: Form the matrix A whose columns are 𝑣1 , 𝑣2 , 𝑣3 and reduce to echelon form:
𝑅2 ↔ 𝑅2 − 2𝑅1
1 4 2 1
𝑅3 ↔ 𝑅3 − 3𝑅1 4 2 1 4
𝑅3 ↔ 𝑅3 − 2𝑅2 2
A= 2 5 1 ~ 0 −3 −3 ~ 0 −3 −3
3 6 0 0 −6 −6 0 0 0

The third column does not have a pivot; hence, the third vector 𝑢3 is a linear combination of
the first two vectors 𝑢1 and 𝑢2 . Thus, the vectors are linearly dependent.
Characterization of linear dependence:
Theorem: Let 𝑉 be a vector space over 𝑅, the set {𝑣1 , 𝑣2 , … , 𝑣𝑛 } of vectors in 𝑉 is linearly
dependent if and only if at least one them is linear combination of the others.
Proof: Suppose that the vector 𝑣𝑗 can be expressed as the linear combination of other
vectors, that is
𝑣𝑗 = 𝑐1 𝑣1 + 𝑐2 𝑣2 + ⋯ + 𝑐𝑗−1 𝑣𝑗−1 + 𝑐𝑗+1 𝑣𝑗+1 + ⋯ + 𝑐𝑛 𝑣𝑛 , for some scalars 𝑐1 , 𝑐2 , … , 𝑐𝑛 .
Then, 0 = 𝑐1 𝑣1 + 𝑐2 𝑣2 + ⋯ + 𝑐𝑗−1 𝑣𝑗−1 + −1 𝑣𝑗 + 𝑐𝑗+1 𝑣𝑗+1 + ⋯ + 𝑐𝑛 𝑣𝑛
where the coefficient of 𝑣𝑗 is −1 which is not equal to zero.
Hence the set {𝑣1 , 𝑣2 , … , 𝑣𝑛 } of vectors is linearly dependent.
Conversely,
Suppose the set {𝑣1 , 𝑣2 , … , 𝑣𝑛 } is linearly dependent, then there exist
scalars 𝑑1 , 𝑑2 , … , 𝑑𝑛 not all zero, say 𝑑𝑖 ≠ 0, such that
 𝑑1 𝑣1 + 𝑑2 𝑣2 + ⋯ + 𝑑𝑖−1 𝑣𝑖−1 + 𝑑𝑖 𝑣𝑖 + 𝑑𝑖+1 𝑣𝑖+1 + ⋯ + 𝑑𝑛 𝑣𝑛 = 0
or 𝑑𝑖 𝑣𝑖 = − 𝑑1 𝑣1 + 𝑑2 𝑣2 + ⋯ + 𝑑𝑖−1 𝑣𝑖−1 + 𝑑𝑖+1 𝑣𝑖+1 + ⋯ + 𝑑𝑛 𝑣𝑛
Since 𝑑𝑖 ≠ 0, divide the above equation by 𝑏𝑖 , then we get
𝑑1 𝑑2 𝑑𝑖−1 𝑑𝑖+1 𝑑1
𝑣𝑖 = − 𝑣 − 𝑣 … − 𝑣𝑖−1 − 𝑣𝑖+1 … − 𝑣
𝑑𝑖 1 𝑑𝑖 2 𝑑𝑖 𝑑𝑖 𝑑𝑛 𝑛

and hence 𝑣𝑖 is a linear combination of other vectors.

 Basis and Dimension for a vector space:
A set ℬ = 𝑢1 , 𝑢2 , … , 𝑢𝑛 of vector space 𝑉 is said to be a basis of 𝑉 if it has following
properties:
(i) ℬ is linearly independent.
(ii) ℬ spans 𝑉, that is 𝑉 = 𝑠𝑝𝑎𝑛 ℬ .
Thus, the basis of the vector space 𝑉 is a linearly independent set that spans 𝑉.

1 0 0
Example:1. The vectors 𝑒1 = 0 , 𝑒2 = 1 , 𝑒3 = 0 forms the basis of 𝑅3 . Because
0 0 1
𝑒1 , 𝑒2 𝑎𝑛𝑑 𝑒3 are linearly independent and 𝑆𝑝𝑎𝑛 𝑒1 , 𝑒2 , 𝑒3 = 𝑅3 . This basis is called standard
basis of 𝑅3 .
2. The set ℬ = { 1,0, … , 0 , 0, 1, … , 0 , … , 0, 0, … , 1 } of vectors in 𝑅𝑛 is the standard basis of
𝑅𝑛 .
Theorem 1:
If ℬ is a basis of a vector space 𝑉, then
(i) ℬ is a minimal spanning set, that is, there is no smaller set of vectors in ℬ that spans 𝑉.
(ii) ℬ is a maximal linearly independent set of vectors in V, that is, adding any other vector to
this set will make it linearly dependent.
Dimension of a vector space:
A vector space in general has many different bases, for example set ℬ = { 1,0 , (0, 1)} is the
standard basis of 𝑅2 also we can verify that the linearly independent sets 1, 1 , 1, −1 ,
1, 0 , 2, 1 . . so on, are the bases of 𝑅2 . So here we observe that each basis of 𝑅2 has only two
vectors. That means number vectors in a basis is fixed, this number is called dimension of the
vector space.
or The number of vectors in a basis is called the dimension. It is denoted by 𝐝𝐢𝐦 𝑽 .
Examples:
1. Since the standard basis of the vector space 𝑅3 has 3 vectors, dim 𝑅3 = 3.
2. 𝑑𝑖𝑚 𝑅𝑛 = 𝑛.
3. Let 𝑀2 (𝑅) be the vector space of 2 × 2 real matrices and basis of 𝑀2 (𝑅) is

1 0 0 1 0 0 0 0
, , , . Hence the dim 𝑀2 𝑅 = 4.
0 0 0 0 1 0 0 1
Problems:
1.Check whether the set { 𝟐, 𝟑 , (𝟎, 𝟒)} is a basis of 𝑹𝟐 .
Solution: The set ℬ = { 2, 3 , (0, 4)} is a basis of 𝑅2 if and only if ℬ is linearly independent
spanning set of 𝑅2 .
Let 𝑢1 = (2, 3) and 𝑢2 = 0, 4 . Since 𝑢1 is not a scalar multiple of 𝑢2 and 𝑢2 is
not a scalar multiple of 𝑢1 , 𝑢1 and 𝑢2 are linearly independent.
 Next, we need to show that an arbitrary vector 𝑎, 𝑏 𝑖𝑛 𝑅2 is a linear combination of 𝑢1
and 𝑢2 , that is, 𝑎, 𝑏 = 𝑐1 2, 3 + 𝑐2 0, 4 = (2𝑐1 , 3𝑐1 + 4𝑐2 )
This implies, 𝑎 = 2𝑐1 , 𝑏 = 3𝑐1 + 4𝑐2
𝑎 3𝑎 2𝑏−3𝑎
or 𝑐1 = and hence 4𝑐2 = 𝑏 − 3𝑐1 = 𝑏 − =
2 2 2

2𝑏−3𝑎
∴ 𝑐2 = .
8

Thus we can express any vector (𝑎, 𝑏) in 𝑅2 as linear combination of 𝑢1 and 𝑢2 . Hence ℬ
is a basis of 𝑅2 .
2. Check whether the set { 𝟏, 𝟑, 𝟏 , 𝟎, 𝟐, 𝟎 , 𝟎, 𝟎, 𝟕 } is a basis of 𝑹𝟑 .
Solution: Let 𝑢1 = 1, 3,1 , 𝑢2 = 0, 2, 0 𝑎𝑛𝑑 𝑢3 = 0, 0, 7 .
To check whether the set is independent, form the matrix whose rows are given
vectors and reduce it in to echelon form:
 1 0 0 𝑅2 ↔ 𝑅2 − 3𝑅1 1 0 0
That is: A = 3 2 0 ~ 0 2 0
0 0 7 0 0 7
Every column of 𝐴 has a pivot entry; Thus, the vectors are linearly independent.
Next, we have to show that given set spans vector space 𝑅3 .
Let 𝑎, 𝑏, 𝑐 be an arbitrary 𝑖𝑛 𝑅3 ,
𝑎, 𝑏, 𝑐 = 𝑐1 1, 3, 1 + 𝑐2 0, 2, 0 + 𝑐3 (0, 0, 7) = (𝑐1 , 3𝑐1 + 2𝑐2 , 𝑐1 + 7𝑐3 )
this implies, 𝑎 = 𝑐1 , 𝑏 = 3𝑐1 + 2𝑐2 𝑐 = 𝑐1 + 7𝑐3
1 1
or 𝑐1 = 𝑎, 𝑐2 = (𝑏 − 3𝑎) and 𝑐3 = 𝑐−𝑎
2 7

Thus we can express any vector (𝑎, 𝑏, 𝑐) in 𝑅3 as linear combination of 𝑢1 , 𝑢2 and 𝑢3 .

Hence 𝑢1 , 𝑢2 , 𝑢3 is a basis of 𝑅3 .
Theorem: Let 𝑉 be a vector space with the basis 𝐵 = {𝑢1 , 𝑢2 , … , 𝑢𝑛 }, then every vector 𝑣 in 𝑉
can be uniquely written as a linear combination of vectors in 𝐵.
Proof:
Let 𝑣 ∈ 𝑉, since B is a basis and hence span 𝐵 = V,
so, there exist scalars 𝑐1 , 𝑐2 , … , 𝑐𝑛 such that 𝑣 = 𝑐1 𝑢1 + 𝑐2 𝑢2 + ⋯ + 𝑐𝑛 𝑢𝑛 -----(1)
Suppose 𝑣 = 𝑑1 𝑢1 + 𝑑2 𝑢2 + ⋯ + 𝑑𝑛 𝑢𝑛 ------(2)
for some scalars 𝑑1 , 𝑑2 , … , 𝑑𝑛 ,
Then, subtracting equations (1) and (2) we get,
0 = 𝑣 − 𝑣 = (𝑐1 −𝑑1 ) 𝑢1 + (𝑐2 −𝑑2 )𝑢2 + ⋯ + (𝑐𝑛 −𝑑𝑛 )𝑢𝑛 .
Since the vectors 𝑢1 , 𝑢2 , … , 𝑢𝑛 are linearly independent,
𝑐1 − 𝑑1 = 0, 𝑐2 −𝑑2 = 0, … , 𝑐𝑛 − 𝑑𝑛 = 0
⟹ 𝑐1 = 𝑑1 , 𝑐2 = 𝑑2 , … , 𝑐𝑛 = 𝑑𝑛 .
Row space and column space of a matrix:
The important subspaces associated with a matrix are row space and column space. These
subspaces are explicitly defined by linear combinations.
Definitions:
The row space of an 𝑚 × 𝑛 matrix 𝐴, denoted by Row𝐴, is the set of all linear combinations of
the rows of 𝐴. If 𝑅1 , 𝑅2 , … , 𝑅𝑚 the rows of 𝐴, then
Row𝐴 = 𝑎1 𝑅1 + 𝑎2 𝑅2 + ⋯ + 𝑎𝑚 𝑅𝑚 𝑎1 , 𝑎2 , … , 𝑎𝑚 𝑎𝑟𝑒 𝑠𝑐𝑎𝑙𝑎𝑟𝑠}
or Row𝐴 = 𝑆𝑝𝑎𝑛 𝑅1 , 𝑅2 , … , 𝑅𝑚
The column space of an 𝑚 × 𝑛 matrix 𝐴, denoted by Col𝐴, is the set of all linear combinations
of the columns of 𝐴. If 𝐶1 , 𝐶2 , … , 𝐶𝑚 the columns of 𝐴, then
Col 𝐴 = 𝑏1 𝐶1 + 𝑏2 𝐶2 + ⋯ + 𝑏𝑛 𝐶𝑛 𝑏1 , 𝑏2 , … , 𝑏𝑛 𝑎𝑟𝑒 𝑠𝑐𝑎𝑙𝑎𝑟𝑠}
or Col𝐴 = 𝑆𝑝𝑎𝑛 𝐶1 , 𝐶2 , … , 𝐶𝑛
Note:
1.Since the spanning set is a subspace, The 𝑅𝑜𝑤 𝐴 and 𝐶𝑜𝑙 𝐴 are subspaces of 𝑅𝑛 and 𝑅𝑚
respectively.
2. A vector in 𝐶𝑜𝑙 𝐴 can be written as 𝐴x for some x in Rn , thus
𝐶𝑜𝑙 𝐴 = 𝑏 ∈ 𝑅𝑚 𝑏 = 𝐴x, for some x in 𝑅𝑛 }
Hence we conclude that, any vector 𝑣 in 𝑅𝑚 is in Col 𝐴 if and only if the linear system
𝐴𝑥 = 𝑣 is consistent(has a solution).
3. Row 𝐴 = 𝐶𝑜𝑙 𝐴𝑇 and 𝐶𝑜𝑙 𝐴 = 𝑟𝑜𝑤 𝐴𝑇 .
Problem: Check whether the vector 𝑢 = (3, −1, 3) is in the column space of the matrix

2 4 −2 1
A = −2 −5 7 3 .
3 7 −8 6
Solution: We know that vector 𝑢 is in the column space of matrix 𝐴 if and only if the system
𝐴x = 𝑢 is consistent(from the above given note). Now to check the consistency of the
system, reduce the augmented matrix into the echelon form.

𝑅2 ↔ 𝑅2 + 𝑅1
2 4 −2 1 3 3 2 4 −2 1 3
𝑅3 ↔ 𝑅3 − 2 𝑅1
[A u] = −2 −5 7 3 −1 ~ 0 −1 5 4 2
9 3
3 7 −8 6 3 0 1 −5 −
2 2

𝑅3 ↔ 𝑅3 + 𝑅2 2 4 −2 1 3
~ 0 −1 5 4 2
17 1
0 0 0 2
2
It is clear that the equation 𝐴x = 𝑢 is consistent, so 𝑣 is in 𝐶𝑜𝑙 𝐴.
Basis of row space :
If 𝑀 is obtained from 𝐴 by row operations(𝑀 is row equivalent to 𝐴), then each row of 𝑀 is a
row of 𝐴 or a linear combination of row of 𝐴. So, 𝐴 and 𝑀 have the same row space, that is
𝑅𝑜𝑤 𝐴 = 𝑅𝑜𝑤 𝐵.
If 𝐵 is an echelon form of 𝐴, then its non zero rows are linearly independent(explained in the
below problem). Thus the nonzero rows of 𝐵 form a basis of the row space of 𝐵 and hence 𝐴,
because 𝐴 is row equivalent to 𝐵.
Example: Consider the following echelon matrix 𝐴, whose pivots have been circled:
 Observe that the rows 𝑅2 , 𝑅3 , 𝑅4 have 0′ 𝑠 in the second column below the pivot in 𝑅1 , and
hence any linear combination of 𝑅2 , 𝑅3 , 𝑅4 must have 0 as its second entry. Thus 𝑅1 cannot be a
linear combination of the rows below it. Similar any nonzero row cannot be a linear combination
other rows in the matrix. Thus the nonzero rows of matrix 𝐴 are linearly independent.

Basis of column space:

When matrix 𝐴 is row reduced to a matrix 𝐵, the columns of 𝐵 are totally different from
columns of 𝐴. But, the equation 𝐴𝑥 = 0 and 𝐵𝑥 = 0 have exactly same set of solutions(Gauss
elimination technique). That is, the columns of 𝐴 have exactly the same linear dependence
relationships as the columns of 𝐵.
If 𝐵 is the echelon form of 𝐴, then the of pivot columns of 𝐵 are linearly independent. Since
𝐴 is row equivalent to 𝐵, the pivot columns of 𝐴 are also linearly independent. Thus the pivot
 forms the basis of column space of 𝐴.
Warning: Be careful to use the pivot columns of 𝐴 itself for the basis of 𝐶𝑜𝑙 𝐴. Row operations
can change the column space of a matrix. The columns of an echelon form 𝐵 of 𝐴 are often not in
the column space of 𝐴.

1. Find the basis for column space and row space of the matrix 𝐴 =

Solution: Row reduce 𝐴 to echelon form:

𝑅2 ↔ 𝑅2 − 𝑅1
𝑅3 ↔ 𝑅3 + 3𝑅2
𝑅3 ↔ 𝑅3 − 2𝑅1 𝑅4 ↔ 𝑅4 + 𝑅2
𝑅4 ↔ 𝑅4 − 3𝑅1
𝐴 ~
 The two nonzero rows of the echelon form of A form a basis 𝑅𝑜𝑤 𝐴. Thus the basis of 𝑅𝑜𝑤 𝐴
is 1, 3, 1, −2, −3 , 0, 1, 2,1, −1
The pivot columns of 𝐴 (not from echelon form) form the basis of 𝐶𝑜𝑙 𝐴, thus the basis of Col 𝐴

1 3
is 1 , 4 .
2 3
3 8

Note: For a given set S = {𝑢1 , 𝑢2 , … , 𝑢𝑘 } of vectors 𝑅𝑛 , we mean a subspace spanned by S as a

column space (row space) of the matrix 𝐴, whose columns (rows) are given vectors.
Problems:
1. Find the dimension of the subspace 𝑊 of 𝑅3 spanned by the vectors
3, 1, 0 , 2, 1, 0 , 1, 1, −2 .
Solution: Form the matrix 𝐴 whose columns are the vectors in the given set. Then, reduce 𝐴 to
echelon form:

3 2 1 1 1 1 1 1 1
𝐴= 1 1 1 ~ 3 2 1 ~ 0 −1 −2
0 0 −2 0 0 −2 0 0 −2
Every column has a pivot entry; hence, no vector is a linear combination of the previous
vectors. Thus, the vectors are linearly independent and Hence they form the basis for the
subspace 𝑊 spanned by the vectors 3, 1, 0 , 2, 1, 0 , 1, 1, −2 .
Therefore the dim 𝑊 = 3.
2. Find the dimension of the subspace 𝑊 of 𝑅3 spanned by the vectors
2, 4, 2 , 1, −1, 0 , 1, 2, 1 , (0, 3, 1) .
Solution: Form the matrix 𝐴 whose columns are the vectors in the given set. Then, reduce 𝐴 to
echelon form:

2 1 1 0 2 1 1 0 2 1 1 0
𝐴 = 4 −1 2 3 ~ 0 −3 0 3 ~ 0 −3 0 3
2 0 1 1 0 −1 0 1 0 0 0 0
The echelon matrix has two non-pivot columns. Hence, the four vectors are linearly
dependent and do not form a basis of 𝑊. Since 1st and 2nd columns of are pivot columns, they
are linearly independent and the form the basis of 𝑊. Thus the dimension of 𝑊 is 2.
3. Let 𝑊 be the subspace spanned by vectors 𝑢1 = 1, −2, 5, −3 , 𝑢2 = 3, 8, −3 − 5 ,
𝑢3 = 2, 3, 1, −4 . (i) Find a basis and dimension of 𝑊.
(ii) Extend the basis of 𝑊 to a basis of 𝑅4 .
 (i) Form the matrix 𝐴 whose columns are the vectors in the given set. Then, reduce 𝐴 to
echelon form:

1 3 2 1 3 2 1 3 2
𝐴 = −2 8 3 ~ 0 14 7 ~ 0 14 7
5 −3 1 0 −18 −9 0 0 0
−3 −5 −4 0 4 2 0 0 0
The echelon matrix has two non-pivot columns. Hence, the given vectors are linearly
dependent and do not form a basis of 𝑊. Since 1st and 2nd columns of are pivot columns, they
are linearly independent and the form the basis of 𝑊. Thus the dimension of 𝑊 is 2.
(ii) We seek four linearly independent vectors, which include the above two linearly
independent vectors. The four vectors 1, −2, 5, 8 , 3, 8, −3, −5 , 0, 0, 1, 0 𝑎𝑛𝑑
(0, 0, 0, 1) are linearly and so they form a basis of 𝑅4 , which is an extension of the basis of 𝑊.
 4. Determine whether 1, 1, 1, 1 , 1, 2, 3, 2 , 2, 5, 6, 4 (2, 6, 8, 5) form a basis of 𝑅4 . If not
find the dimension of subspace spanned by the these vectors.
(Ans: No, dimension of subspace spaaned by given vectors is 3).

3 1 −2
5. Check whether [5 17 − 20] is in the row space of 𝐴 = 4 0 1 .
−2 4 −3
( Check whether the vector 5 17 − 20 𝑇 is in the column space of 𝐴𝑇 ).

Null space of matrix:

The null space of an 𝑚 × 𝑛 matrix 𝐴, written as 𝑁𝑢𝑙 𝐴, is the set of all solutions of the
homogeneous equation 𝐴x = 𝟎. In set notation,
𝑁𝑢𝑙 𝐴 = 𝐱 𝐱 𝑖𝑠 𝑖𝑛 𝑅𝑛 𝑎𝑛𝑑 𝐴𝐱 = 𝟎}
 5
1 −3 −2
Example 1: Let 𝐴 = , and let 𝑢 = 3 . Determine if 𝒖 belongs to the null
−5 9 1
−2
space of 𝐴.
Solution: To test if 𝒖 satisfies 𝐴𝒖 = 𝟎, simply compute

5
1 −3 −2 5−9+4 0
𝐴𝑢 = 3 = = .
−5 9 1 −25 + 27 − 2 0
−2
Thus 𝒖 is in 𝑁𝑢𝑙 𝐴.
Theorem: Let 𝐴 an 𝑚 × 𝑛 matrix, then the 𝑁𝑢𝑙 𝐴 is a subspace of 𝑅𝑛 .
Proof: Since 𝐴 ∙ 0 = 0, 0 ∈ 𝑁𝑢𝑙𝐴.
Let 𝑢, 𝑣 ∈ 𝑁𝑢𝑙 𝐴, then 𝐴𝑢 = 0 and 𝐴𝑣 = 0.
Hence 𝐴 𝑢 + 𝑣 = 𝐴𝑢 + 𝐴𝑣 = 0 + 0 = 0, there fore 𝑢 + 𝑣 ∈ 𝑁𝑢𝑙 𝐴.
 Let 𝑐 be a scalar, 𝑢 ∈ 𝑁𝑢𝑙 𝐴, then 𝐴𝑢 = 0.
Hence 𝐴 𝑐𝑢 = 𝑐 𝐴𝑢 = 𝑐 ∙ 0 = 0, there fore 𝑐𝑢 ∈ 𝑁𝑢𝑙 𝐴.
Therefore, 𝑁𝑢𝑙 𝐴 is a subspace of 𝑅𝑛 .

Note: There is no obvious relation between vectors in 𝑁𝑢𝑙 𝐴 and the entries in 𝐴. We say that
𝑁𝑢𝑙 𝐴 is defined implicitly, because it is defined by a condition that must be checked. No
explicit list or description of the elements in 𝑁𝑢𝑙 𝐴 is given. However, solving the equation
𝐴𝒙 = 𝟎 amounts to producing an explicit description of 𝑁𝑢𝑙 𝐴.

Example: Find the Null space of the matrix

 Solution: To find the 𝑁𝑢𝑙 𝐴 , we have to find the solution of the system 𝐴𝑥 = 0.
Now reduce the augmented matrix 𝐴 0 into echelon form, we get

−3 6 −1 1 −7 0 1 −2 2 3 −1 0
1 −2 2 3 −1 0 ~ −3 6 −1 1 −7 0
2 −4 5 8 −4 0 2 −4 5 8 −4 0

1 −2 2 3 −1 0 1 −2 2 3 −1 0
~ 0 0 5 10 −10 0 ~ 0 0 5 10 −10 0
0 0 1 2 −2 0 0 0 0 0 0 0
So, equivalent system of the echelon form is:

𝑥1
1 −2 2 3 −1 𝑥2 0
0 0 5 10 −10 𝑥3 = 0
0 0 0 0 0 𝑥4 0
𝑥5
 that is , 𝑥1 − 2𝑥2 + 2𝑥3 + 3𝑥4 − 𝑥5 = 0
5𝑥3 + 10𝑥4 − 10𝑥5 = 0 or 𝑥3 + 2𝑥4 − 2𝑥5 = 0
Since 2nd , 4th and 5th columns of matrix 𝐴 are non-pivot columns, 𝑥2 , 𝑥4 and 𝑥5 are free
variables, so we can put arbitrary values for these variables.
Let 𝑥2 = 𝑘1 , 𝑥4 = 𝑘2 and 𝑥5 = 𝑘3
Thus, 𝑥3 = −2𝑥4 + 2𝑥5 = −2𝑘2 + 2𝑘3
𝑥1 = 2𝑥2 − 2𝑥3 − 3𝑥4 + 𝑥5 = 2𝑘1 − 2 −2𝑘2 + 2𝑘3 − 3𝑘2 + 𝑘3
= 2𝑘1 + 𝑘2 − 3𝑘3

𝑥1 2𝑘1 + 𝑘2 − 3𝑘3 2𝑘1 𝑘2 −3𝑘3

𝑥2 𝑘1 𝑘1 0 0
Hence, 𝑥3 = −2𝑘2 + 2𝑘3 = 0 + −2𝑘2 + 2𝑘3
𝑥4 𝑘2 0 𝑘2 0
𝑥5 𝑘
 𝑥1 2 1 −3
𝑥2 1 0 0
Hence, 𝑥3 = 𝑘1 0 + 𝑘2 −2 + 𝑘3 2 , 𝑘1 , 𝑘2 𝑎𝑛𝑑 𝑘3 𝑎𝑟𝑒 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦 𝑟𝑒𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟𝑠.
𝑥4 0 1 0
𝑥5 0 0 1

𝒖 𝒗 𝒘

Every linear combination of 𝒖, 𝒗, 𝑎𝑛𝑑 𝒘 is an element of 𝑁𝑢𝑙 𝐴 and vice versa. Thus
𝑁𝑢𝑙 𝐴 = 𝑆𝑝𝑎𝑛 {𝒖, 𝒗, 𝒘}.
Since 𝒖, 𝒗 𝑎𝑛𝑑 𝒘 are linearly independent, the set {𝒖, 𝒗, 𝒘} is the basis of 𝑁𝑢𝑙 𝐴.
Problem: Find bases for the row space, the column space, and the null space of the matrix
 Coordinates:
Let 𝑉 be an 𝑛-dimensional vector space with basis 𝐵 = {𝑢1 , 𝑢2 , … , 𝑢𝑛 }. Then any vector
𝑣 ∈ 𝑉 can be expressed uniquely as a linear combination of the basis vectors in 𝐵, say
𝑣 = 𝑐1 𝑢1 + 𝑐2 𝑢2 + ⋯ + 𝑐𝑛 𝑢𝑛
These 𝑛 scalars 𝑐1 , 𝑐2 , … , 𝑐𝑛 are called the coordinates of 𝑣 relative to the basis 𝐵, and they
form a vector [𝑐1 , 𝑐2 , … , 𝑐𝑛 ] in 𝑅𝑛 called the coordinate vector of 𝑣 relative to 𝐵. We denote this
vector by 𝑣 𝐵 .
Thus,
𝑣 𝐵 = [𝑐1 , 𝑐2 , … , 𝑐𝑛 ].

Note: The above 𝑛 scalars 𝑐1 , 𝑐2 , … , 𝑐𝑛 also form the coordinate column vector [𝑐1 , 𝑐2 , … , 𝑐𝑛 ]𝑇 of
𝑣 relative to 𝐵.
 5 1
Example 1: Let 𝑣 = 3 and 𝐵 = {𝑢1 , 𝑢2 , 𝑢3 } be the basis of 𝑅3 where 𝑢1 = −1 ,
4 0

1 0
𝑢2 = 1 and 𝑢3 = 1 . Determine the coordinate vector of 𝑣 relative to 𝐵.
0 1
Solution: Set 𝑣 = 𝑐1 𝑢1 + 𝑐2 𝑢2 + 𝑐3 𝑢3 that is, set 𝑣 as a linear combination of the basis
vectors using unknown scalars 𝑐1 , 𝑐2 , 𝑐3 .
This yields,

5 1 1 0
3 = 𝑐1 −1 + 𝑐2 1 + 𝑐3 1
4 0 0 1
 5 𝑐1 𝑐2 0 𝑐1 + 𝑐2
That is , 3 = −𝑐1 + 𝑐2 + 𝑐3 = − 𝑐1 + 𝑐2 + 𝑐3 .
4 0 0 𝑐3 𝑐3
The equivalent system of linear equations is as follows:
𝑐1 + 𝑐2 = 5, − 𝑐1 +𝑐2 + 𝑐3 = 3, 𝑐3 = 4
this implies, 𝑐1 + 𝑐2 = 5, − 𝑐1 +𝑐2 = −1
Thus the solution of the system is 𝑐1 = 3, 𝑐2 = 2 and 𝑐3 = 4.
Therefore, 𝑣 = 3𝑢1 + 2𝑢2 + 4𝑢3
and so, 𝑣 𝐵 = [3, 2, 4].
2. Find the coordinate vector of 𝑣 = (𝑎, 𝑏, 𝑐) in 𝑅3 relative to the basis
𝐵 = 𝑢1 , 𝑢2 , 𝑢3 = 1, 1,1 , 1, 1,0 , 1, 0, 0 .
Answer: 𝑣 𝐵 = 𝑐, 𝑏 − 𝑐, 𝑎 − 𝑏 .
Here equations are, 𝑐1 +𝑐2 + 𝑐3 = 𝑎, 𝑐1 + 𝑐2 = 𝑏 & 𝑐1 = 𝑐.
 Problems:

1 2 3
1. Consider the basis 𝐵 = , for 𝑅2 . If x 𝐵 = , what is x ?
2 1 2
2. Let 𝑣1 = 3, 6, 2 , 𝑣2 = −1, 0,1 , 𝑥 = (3, 12, 7) and 𝐵 = 𝑣1 , 𝑣2 is a basis of
subspace 𝐻 of 𝑅3 . Determine 𝑥 𝐵 in 𝐻.

Trace of the matrix:

The sum of the diagonal entries of a square matrix 𝐴 is called trace of the trace of the matrix
𝐴 and it is denoted by 𝑡𝑟 𝐴 .

12 3 4
Example: Let 𝐴 = 5 6 9 , then 𝑡𝑟 𝐴 = 12 + 6 + 4 = 22.
3 4 4
Properties of trace: Let A and B be 𝑛 × 𝑛 matrices and c be a scalar, then
(i) 𝑡𝑟 𝑐𝐴 = 𝑐 𝑡𝑟(𝐴)
(ii) 𝑡𝑟 𝐴 ± 𝐵 = 𝑡𝑟 𝐴 ± 𝑡𝑟(𝐵)
(iii) 𝑡𝑟 𝐴𝐵 = 𝑡𝑟(𝐵𝐴)
(iv) 𝑡𝑟 𝐵−1 𝐴𝐵 = 𝑡𝑟 𝐴
(v) 𝑡𝑟 𝐴𝐴𝑇 = σ𝑛𝑖=1 σ𝑛𝑗=1 𝑎𝑖𝑗
2
.

1 2 5 2
Example: Let 𝐴 = and 𝐵 = . Verify that (i) 𝑡𝑟 𝐵−1 𝐴𝐵 = 𝑡𝑟 𝐴 and
3 4 −1 3
(ii) 𝑡𝑟 𝐴𝐴𝑇 = σ2𝑖=1 σ2𝑗=1 𝑎𝑖𝑗
2
.

1 3 −2 1 3 −2 1 2 5 2
Answer: Here, 𝐵−1 = −1
. So 𝐵 𝐴𝐵 =
17 1 5 17 1 5 3 4 −1 3
1 3 −2 3 8 1 −13 −12
= = .
17 1 5 11 18 17 58 98
 1 1
Thus, 𝑡𝑟 𝐵−1 𝐴𝐵 = −13 + 98 = 85 = 5 = 1 + 4 = 𝑡𝑟 𝐴 .
17 17

1 2 1 3 1 2 + 22 11
(ii) 𝐴𝐴𝑇 = =
3 4 2 4 11 32 + 42
Thus, 𝑡𝑟 𝐴𝐴𝑇 = 12 + 22 + 32 + 42 = σ2𝑖=1 σ2𝑗=1 𝑎𝑖𝑗
2
.

Eigenvalues and Eigenvectors:

A scalar 𝜆 is said to be eigenvalue of a square matrix 𝐴, if there exist a nonzero vector 𝑋
such that 𝐴𝑋 = 𝜆𝑋,
Any vector satisfying this relation is called an eigenvector of 𝐴 corresponding to the
eigenvalue 𝜆 .
We note that each scalar multiple 𝑘𝑋 of an eigenvector 𝑋 corresponding to the eigenvalue 𝜆 is
also such an eigenvector,
 because,
𝐴 𝑘𝑋 = 𝑘 𝐴𝑋 = 𝑘 𝜆𝑋 = 𝜆(𝑘𝑋).
The set 𝐸𝜆 of all such eigenvectors (including zero vector) is a subspace of 𝑉, called the
eigen space of 𝜆.
Note: Let 𝐴 be square matrix and 𝜆 be the eigen value of 𝐴, then there exist a nonzero vector
𝑋 such that 𝐴𝑋 = 𝜆𝑋
that is, 𝐴𝑋 − 𝜆𝑋 = 0
or 𝐴 − 𝜆𝐼 𝑋 = 0
The linear system 𝐴 − 𝜆𝐼 𝑋 = 0 has nonzero solution if and only if the matrix
𝐴 − 𝜆𝐼 is singular( that is det 𝐴 − 𝜆𝐼 = 0). Therefore the eigen value 𝜆 is the root
of the characteristic equation det 𝐴 − 𝜆𝐼 = 0. Thus, the terms characteristic value
and characteristic vector (or proper value and proper vector) are sometimes used
Relation between trace and eigenvalues of the matrix:
Let 𝜆1 , 𝜆2 , … , 𝜆𝑛 be the eigenvalues of an 𝑛 × 𝑛 matrix 𝐴, then
(i) 𝑡𝑟 𝐴 = 𝜆1 , +𝜆2 , + ⋯ + 𝜆𝑛
(ii) det 𝐴 = 𝜆1 𝜆2 … 𝜆𝑛 .

Computing Eigenvalues, Eigenvectors and Eigenspaces:

1. Find the characteristic equation det 𝐴 − 𝜆𝐼 = 0 of 𝐴.
2. Solve the characteristic equation for the eigenvalues(roots of the characteristic equation).
3. For each eigenvalue solve the system 𝐴 − 𝜆𝐼 𝑋 = 0, which gives the corresponding
eigenvector.
Problems: Check whether 𝜆1 +𝜆2 + 𝜆3 equals the trace for the following matrices

3 4 2 0 0 2
(i) A = 0 1 2 (ii) B = 0 2 0
0 0 0 2 0 0
Solution: (i) For the eigenvalues, solve the characteristic equation det 𝐴 − 𝜆𝐼 = 0 ,
Since 𝐴 is upper triangular matrix its characteristic equation is:
3−𝜆 1−𝜆 0−𝜆 =0
this implies, 3 − 𝜆 = 0, 1 − 𝜆 = 0 𝑎𝑛𝑑 0−𝜆 =0
Thus the eigenvalues are 𝜆1 = 3, 𝜆2 = 1, 𝜆3 = 0.
Hence, 𝜆1 +𝜆2 + 𝜆3 = 3 + 1 + 0 = 𝑡𝑟 𝐴 .

−𝜆 0 2
(ii) Characteristic equation of 𝐵 is, det 0 2−𝜆 0 =0
2 0 −𝜆
 that is, −𝜆 2 − 𝜆 −𝜆 − 0 + 2 0 − 2 2 − 𝜆 =0
𝜆2 2 − 𝜆 − 4 2 − 𝜆 = 0
2 − 𝜆 𝜆2 − 4 = 0 or 2−𝜆 𝜆−2 𝜆+2 =0
Hence the eigenvalues are: 2, 2, −2.
Thus, 𝜆1 +𝜆2 + 𝜆3 = 2 + 2 − 2 = 2 = 𝑡𝑟 𝐵 .

5 4 6 3
2. Are 𝑢 and 𝑣 Eigenvectors of 𝐴, if 𝐴 = , 𝑢= and 𝑣 = ?
1 2 −5 −2
Answer: Characteristic equation of 𝐴 is : 𝜆2 − 𝑡𝑟𝐴 𝜆 + det 𝐴 = 0
that is, 𝜆2 − 7𝜆 + 6 = 0 or 𝜆−6 𝜆−1 =0
Thus the eigenvalues are: 6, 1.

5 4 6 10
Now consider 𝐴𝑢 = = ≠ 𝜆𝑢, for 𝜆 = 6 𝑜𝑟 1. Therefore 𝑢 is not an
1 2 −5 −4
eigenvector of 𝐴.
 5 4 3 7
Next, 𝐴𝑣 = = ≠ 𝜆𝑣, for 𝜆 = 6 𝑜𝑟 1. Therefore 𝑣 is also not an
1 2 −2 −1
eigenvector of 𝐴.
3. Determine the Eigenvalues and corresponding eigenvectors of the matrix

3 −2 4
𝐴 = −2 6 2 .
4 2 3
Solution: Characteristic equation of the matrix 𝐴 is:
𝜆3 − 𝑆1 𝜆2 + 𝑆2 𝜆 − 𝑆3 = 0
Where 𝑆1 = 𝑡𝑟 𝐴 = 3 + 6 + 3 = 12
𝑆2 = 𝑠𝑢𝑚 𝑜𝑓 𝑚𝑖𝑛𝑜𝑟𝑠 𝑜𝑓 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 𝑒𝑛𝑡𝑟𝑖𝑒𝑠 𝑜𝑓 𝐴 = 𝐴11 + 𝐴22 + 𝐴33

6 2 3 4 3 −2
= + + = 14 − 7 + 14 = 21.
2 3 4 3 −2 6
 𝑆3 = 𝐷𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑎𝑛𝑡 𝑜𝑓 𝐴

6 2 −2 2 −2 6
=3 − −2 +4 = 3 14 + 2 −14 + 4 −28 = −98
2 3 4 3 4 2
Thus the characteristic equation is:
𝜆3 − 12𝜆2 + 21𝜆 − (−98) = 0 or 𝜆3 − 12𝜆2 + 21𝜆 + 98 = 0.
Eigen values are: −2, 7, 7.
Next, we have to find the corresponding eigenvectors:
When 𝜆 = −2, 𝐴 − 𝜆𝐼 𝑋 = 𝐴 − (−2)𝐼 𝑋 = 𝐴 + 2𝐼 𝑋 = 0

3 + 2 −2 4 𝑥 5 −2 4 𝑥
that is, −2 6 + 2 2 𝑦 = −2 8 2 𝑦 = 0
4 2 3+2 𝑧 4 2 5 𝑧
Since two rows of coefficient matrix are linearly independent, we can use cross ratio
method to solve the above system.
 𝑥 −𝑦 𝑧 𝑥 −𝑦 𝑧
that is, −2 4 = 5 4 = 5 −2 or = =
−36 18 36
8 2 −2 2 −2 8
𝑥 𝑦 𝑧
or = = = 𝑘1 , 𝑘1 𝑖𝑠 𝑎 𝑛𝑜𝑛𝑧𝑒𝑟𝑜 𝑠𝑐𝑎𝑙𝑎𝑟.
2 1 −2
𝑥
this implies, = 𝑘1 ℎ𝑒𝑛𝑐𝑒 𝑥 = 2𝑘1 , 𝑦 = 𝑘1 and 𝑧 = −2𝑘1 .
2

𝑥 2𝑘1 2
therefore, X𝜆=−2 = 𝑦 = 𝑘1 = 𝑘1 1 .
𝑧 −2𝑘1 −2

2
Thus the corresponding eigenspace 𝐸𝜆=−2 = 𝑘1 1 | 𝑘1 𝑖𝑠 𝑎𝑛 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦 𝑟𝑒𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 .
−2
or 𝐸𝜆=−2 = 𝑆𝑝𝑎𝑛 2, 1, −2 .
When 𝜆 = 7, 𝐴 − 𝜆𝐼 𝑋 = 𝐴 − 7𝐼 𝑋 = 0

3 − 7 −2 4 𝑥 −4 −2 4 𝑥
that is, −2 6 − 7 2 𝑦 = −2 −1 2 𝑦 = 0
4 2 3−7 𝑧 4 2 −4 𝑧
In the above system, no two rows are linearly dependent. So we can’t use cross ratio
method to compute the eigenvectors. But we reduce the above system into,

−4 −2 4 𝑥
0 0 0 𝑦 =0
0 0 0 𝑧
Equivalent linear equation is: −4𝑥 − 2𝑦 + 4𝑧 = 0 or 2𝑥 + 𝑦 − 2𝑧 = 0
Since we have only one equation with 3 unknowns, two of them are free variables. Here 𝑦
1 𝑘2
and 𝑧 are free variables. Put 𝑦 = 𝑘2 and 𝑧 = 𝑘3 . Hence, 𝑥 = 2𝑧 − 𝑦 = 𝑘3 − .
2 2
 𝑘2 1
𝑥 𝑘3 − − 1
2 2
therefore, the corresponding eigenvector: X𝜆=7 = 𝑦 = 𝑘2 = 𝑘2 1 + 𝑘3 0 .
𝑧 𝑘3 0 1
Thus the corresponding eigenspace:

1
− 1
2
𝐸𝜆=7 = 𝑘2 1 + 𝑘3 0 | 𝑘2 , 𝑘3 𝑎𝑟𝑒 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦 𝑟𝑒𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟𝑠 .
0 1
1
or 𝐸𝜆=7 = 𝑆𝑝𝑎𝑛 − , 1, 0 , (1, 0, 1) .
2
4. Determine the Eigenvalues and corresponding eigenvectors of the matrix

2 −1 1
𝐴 = −1 2 −1 .
1 −1 2
Solution: Characteristic equation of the matrix 𝐴 is:
𝜆3 − 𝑆1 𝜆2 + 𝑆2 𝜆 − 𝑆3 = 0
Where 𝑆1 = 𝑡𝑟 𝐴 = 2 + 2 + 2 = 6
𝑆2 = 𝑠𝑢𝑚 𝑜𝑓 𝑚𝑖𝑛𝑜𝑟𝑠 𝑜𝑓 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 𝑒𝑛𝑡𝑟𝑖𝑒𝑠 𝑜𝑓 𝐴 = 𝐴11 + 𝐴22 + 𝐴33

2 −1 2 1 2 −1
= + + =3+3+3=9
−1 2 1 2 −1 2
2 −1 −1 −1 −1 2
𝑆3 = 𝐷𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑎𝑛𝑡 𝑜𝑓 𝐴 = 2 − −1 +1
−1 2 1 2 1 −1
= 2 3 + −1 + −1 = 4.
 Thus the characteristic equation is:
𝜆3 − 6𝜆2 + 9𝜆 − 4 = 0.
So, Eigen values are: 4, 1,1.
Next, we have to find the corresponding eigenvectors:
When 𝜆 = 4, 𝐴 − 𝜆𝐼 𝑋 = 𝐴 − 4𝐼 𝑋 = 0

2 − 4 −1 1 𝑥 −2 −1 1 𝑥
that is, −1 2 − 4 −1 𝑦 = −1 −2 −1 𝑦 = 0
1 −1 2−4 𝑧 1 −1 −2 𝑧
Since two rows of coefficient matrix are linearly independent, we can use cross ratio
method to solve the above system.

𝑥 −𝑦 𝑧 𝑥 −𝑦 𝑧
that is, −1 1 = −2 1 = −2 −1 or = =
3 3 3
−2 −1 −1 −1 −1 −2
 𝑥 −𝑦 𝑧
or = = = 𝑘1 , 𝑘1 𝑖𝑠 𝑎 𝑛𝑜𝑛𝑧𝑒𝑟𝑜 𝑠𝑐𝑎𝑙𝑎𝑟.
1 1 1
𝑥
this implies, = 𝑘1 ℎ𝑒𝑛𝑐𝑒 𝑥 = 𝑘1 , 𝑦 = −𝑘1 and 𝑧 = 𝑘1 .
1

𝑥 𝑘1 1
therefore, the corresponding eigenvector is: X𝜆=4 = 𝑦 = −𝑘1 = 𝑘1 −1 .
𝑧 𝑘1 1

1
Thus the corresponding eigenspace 𝐸𝜆=4 = 𝑘1 −1 | 𝑘1 𝑖𝑠 𝑎𝑛 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦 𝑟𝑒𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 .
1
or 𝐸𝜆=4 = 𝑆𝑝𝑎𝑛 1, −1, 1 .
When 𝜆 = 1, 𝐴 − 𝜆𝐼 𝑋 = 𝐴 − 𝐼 𝑋 = 0

2 − 1 −1 1 𝑥 1 −1 1 𝑥
that is, −1 2 − 1 −1 𝑦 = −1 1 −1 𝑦 = 0
1 −1 2 − 1 𝑧 1 −1 1 𝑧
In the above system, no two rows are linearly dependent. So we can’t use cross ratio
method to compute the eigenvectors. But we reduce the above system into,

1 −1 1 𝑥
0 0 0 𝑦 =0
0 0 0 𝑧
Equivalent linear equation is: 𝑥 − 𝑦 + 𝑧 = 0. Since we have only one equation with 3
unknowns, two of them are free variables. Here 𝑦 and 𝑧 are free variables. Put 𝑦 = 𝑘2 and 𝑧 =
𝑘3 . Hence, 𝑥 = 𝑦 − 𝑧 = 𝑘2 − 𝑘3 .
 𝑥 𝑘2 − 𝑘3 1 −1
therefore, the corresponding eigenvector: X𝜆=1 = 𝑦 = 𝑘2 = 𝑘2 1 + 𝑘3 0 .
𝑧 𝑘3 0 1
Thus the corresponding eigenspace:

1 −1
𝐸𝜆=1 = 𝑘2 1 + 𝑘3 0 | 𝑘2 , 𝑘3 𝑎𝑟𝑒 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦 𝑟𝑒𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟𝑠 .
0 1
or 𝐸𝜆=1 = 𝑆𝑝𝑎𝑛 1, 1, 0 , (1, 0, −1) .
5. Determine the Eigenvalues and corresponding eigenvectors of the matrix

13 −4 2
𝐴 = −4 13 −2 .
2 −2 10