Vector Spaces

Dr. Jagannath Bhanja

IIITDM Kancheepuram, Chennai
Mathematical structures in linear algebra

(1) Field (See Chapter 1)

(2) Vector Space
(3) ...........

1
A vector space V over a field F

A vector space ⟨ V , F , +, . ⟩ consists of the following:

(1) a field F of scalars;

(2) a set V of objects, called vectors;
(3) an operation + : V × V −→ V , called vector addition, which
satisfies the following axioms:
(a) addition is commutative

α + β = β + α, for all α, β ∈ V ;

(b) addition is associative

α + (β + γ) = (α + β) + γ, for allα, β, γ ∈ V ;

2
contd.

(c) there is a unique vector 0 ∈ V called the zero vector such that

α + 0 = α, for all α ∈ V ;

(d) for each vector α ∈ V , there is a unique vector −α ∈ V

such that α + (−α) = 0;

(4) an operation . : F × V −→ V , called scalar multiplication,

which satisfies the following axioms.
(e) 1.α = α, for all α ∈ V ;
(f) (c1 c2 )α = c1 (c2 α), for all c1 , c2 ∈ F , α ∈ V ;
(g) c(α + β) = cα + cβ, for all α, β ∈ V , c ∈ F ;
(h) (c1 + c2 )α = c1 α + c2 α, for all c1 , c2 ∈ F , α ∈ V .

3
Example 1 : The Euclidean space ⟨Rn , R, +, .⟩

V = Rn = {(x1 , x2 , . . . , xn ) : xi ∈ R} and F = R.

Let α = (x1 , x2 , . . . , xn ), β = (y1 , y2 , . . . , yn ) ∈ V

Let us define vector addition and scalar multiplication as follows:
Define + : V × V −→ V as (vector addition)

α + β = (x1 + y1 , x2 + y2 , . . . , xn + yn )

and . : F × V −→ V as

cα = (cx1 , cx2 , . . . , cxn )

Then ⟨Rn , R, +, .⟩ = Rn is a vector space.

4
Verification

(a)

α + β = (x1 + y1 , x2 + y2 , . . . , xn + yn )
= (y1 + x1 , y2 + x2 , . . . , yn + xn )
= β + α.

(b) Let γ = (z1 , z2 , . . . , zn ). Then

α + (β + γ) = (x1 , x2 , . . . , xn ) + (y1 + z1 , y2 + z2 , . . . , yn + zn )
= (x1 + y1 + z1 , x2 + y2 + z2 , . . . , xn + yn + zn )
= (x1 + y1 , x2 + y2 , . . . , xn + yn ) + (z1 , z2 , . . . , zn )
= (α + β) + γ.

5
contd.

(c) Let 0 = (0, 0, . . . , 0) ∈ V = Rn such that

α + 0 = (x1 , . . . , xn ) + (0, . . . , 0) = (x1 , . . . , xn ) = α.
(d) For every α = (x1 , x2 , . . . , xn ), there exists
−α = (−x1 , −x2 , . . . , −xn ) ∈ V such that α + (−α) = 0.
(e) 1 · α = (1 · x1 , 1 · x2 , . . . , 1 · xn ) == (x1 , x2 , . . . , xn )α.
(f) For scalars c1 , c2 and a vector α = (x1 , x2 , . . . , xn ), we have

(c1 c2 ) · α = (c1 c2 x1 , c1 c2 x2 , . . . , c1 c2 xn )
= c1 · (c2 x1 , c2 x2 , . . . , c2 xn )
= c1 · (c2 · α).

6
contd.

(g) Let α = (x1 , x2 , . . . , xn ) and β = (y1 , y2 , . . . , yn ), we have

c(α + β) = c(x1 + y1 , x2 + y2 , . . . , xn + yn )
= (c · (x1 + y1 ), c · (x2 + y2 ), . . . , c · (xn + yn ))
= (cx1 + cy1 , cx2 + cy2 , . . . , cxn + cyn ))
= cα + cβ.

(h) For scalars c1 , c2 and a vector α = (x1 , x2 , . . . , xn ), we have

(c1 + c2 ) · α = ((c1 + c2 )x1 , (c1 + c2 )x2 , . . . , (c1 + c2 )xn )

= (c1 x1 + c2 x1 , c1 x2 + c2 x2 , . . . , c1 xn + c2 xn )
= (c1 x1 , c1 x2 , . . . , c1 xn ) + (c2 x1 , c2 x2 , . . . , c2 xn )
= c1 · α + c2 · α.
7
Example 2: The n-tuple space ⟨F n , F , +, .⟩

Let F be a field and let

V = F n = {(x1 , x2 , . . . , xn ) : xi ∈ F } .

Let α = (x1 , x2 , . . . , xn ), β = (y1 , y2 , . . . , yn ) ∈ V = F n .

Define + : V × V −→ V as (vector addition)

α + β = (x1 + y1 , x2 + y2 , . . . , xn + yn )

and . : F × V −→ V as

cα = (cx1 , cx2 , . . . , cxn )

Show that ⟨F n , F , +, .⟩ is a vector space.

8
Example 3 : The space of m × n matrices ⟨F m×n , F , +, .⟩

Let F be a field and

V = F m×n = {A = [aij ]m×n : aij ∈ F } .

We define vector addition and scalar mulatiplication as follows,

where A = [aij ], B = [bij ] ∈ V and c ∈ F

[A + B]ij = [aij + bij ]

and
[cA]ij = [caij ].

Show that F m×n is a vector space over F

Note that F n×n is not a field

9
Example 4: The set of all real valued continuous functions
defined on [0, 1]

Let V = {f : f : [0, 1] −→ R and f is continuous on [0,1]}

We define
+ : V × V −→ V

as (f + g )(s) = f (s) + g (s) for s ∈ [0, 1].

· : R × V −→ V

as (cf )(s) = cf (s) for s ∈ [0, 1]

Show that ⟨V , R, +, .⟩ is a a vector space.

10
Example 5: The space of polynomial functions over a field

Assignment

11
Problem 1

Let V = {(x, y ) : x, y ∈ R}. We define

(x1 , y1 ) + (x2 , y2 ) = (x1 + x2 , y1 + y2 )

and
c(x, y ) = (cx, y )

Prove or disprove that ⟨V , R, +, .⟩ is a vector space.

Solution: Verify (c1 + c2 )α ̸= c1 α + c2 α.

12
Note 1

Let V be a vector space over a field F . We have

0 = 0 + 0, (additive identity)

c0 = c(0 + 0), c ∈F

c0 = c0 + c0, (c(α + β) = cα + cβ)

Add −(c0) ∈ V on both sides

c0 + (−(c0)) = (c0 + c0) + (−(c0)),

0 = c0 + (c0 + −(c0)) , (Associative)

13
Note 1 contd.

0 = c0 + 0, (Existence of inverse)

0 = c0 (additive identity )

c0 = 0 for all c ∈ F .

Qn. Show that 0α = 0 for all α ∈ V , where 0 is the additive

identity in the field F and 0 is the zero vector in the vector
space V .

14
Note 2

0 = 0α, (see last question)

= (1 + (−1)) α
= 1.α + (−1)α ( Reason: (c1 + c2 )α = c1 α + c2 α)
= α + (−1)α (Reason : 1.α = α)
=⇒ additive inverse of α, −α = (−1)α.

15
Note 3

Prove that if cα = 0, then c = 0 or α = 0.

Proof : Suppose that c ̸= 0 (else 0α = 0).
Since 0 ̸= c ∈ F and F is a field, c −1 ∈ F .

cα = 0

=⇒ c −1 (cα) = c −1 0 = 0

=⇒ (c −1 c)α = 0 Reason: (c1 c2 )α = c1 (c2 α)

=⇒ 1.α = 0 Reason : c −1 c = 1

=⇒ α=0 Reason : 1.α = α

16
Linear Combination

Definition
Let V be a vector space over a field F . A vector β ∈ V is said to
be a linear combination of vectors α1 , α2 , . . . , αn in V if there exist
scalars c1 , c2 , . . . , cn in F such that
n
X
β = c1 α1 + c2 α2 + . . . + cn αn = ci αi .
i=1

17
Problem 2

Show that (x, y , z) ∈ R3 is a linear combination of vectors

α = (1, 1, 1), β = (0, 1, 1) and γ = (0, 0, 1).
Solution : Find scalars (if exist) a, b, c ∈ R such that

(x, y , z) = aα + bβ + cγ

(x, y , z) = a(1, 1, 1) + b(0, 1, 1) + c(0, 0, 1)

(x, y , z) = (a, a + b, a + b + c)

(x, y , z) = x(1, 1, 1) + (y − x)(0, 1, 1) + (z − y )(0, 0, 1).

18
Problem 3

Prove or disprove that (1, 2, 3) is a linear combination of

α = (1, 1, 1) and β = (0, 1, 1).
Ans. No.

(1, 2, 3) = a(1, 1, 1) + b(0, 1, 1)

=⇒ a + b = 2 and a + b = 3, lead us to a contradiction.

19
Problem 4

Let R be the real field. Find all vectors in R3 that are linear
combination of (1, 0, −1), (0, 1, 1) and (1, 1, 1).
Solution: Objective is to find all linear combinations of vectors
(1, 0, −1), (0, 1, 1) and (1, 1, 1).
That is, find all (x, y , z) such that there exist a, b, c ∈ R such that

a(1, 0, −1) + b(0, 1, 1) + c(1, 1, 1) = (x, y , z).

That is, find a, b, c (if exist) such that

    
1 0 1 a x
 0 1 1  b  =  y 
    

−1 1 1 c z

=⇒ AX = Y
20
Problem 4 contd.

Find a row-reduced echelon matrix which is row-equivalent to A.

       
1 0 1 1 0 1 1 0 1 1 0 0
A= 0 1 1 ∼ 0 1 1 ∼ 0 1 1 ∼ 0 1 0 
       

−1 1 1 0 1 2 0 0 1 0 0 1

By Theorem 12, A is invertible (A ∼ I ). By Theorem 13, the system

AX = Y has a solution X for all Y . Hence for every
Y t = (x, y , z) ∈ R3 , there exists X t = (a, b, c) such that

a(1, 0, −1) + b(0, 1, 1) + c(1, 1, 1) = (x, y , z).

21
Subspace

Definition
Let V be a vector space over a field F . A subspace of V is a subset
W of V which itself a vector space over F with the operations of
vector addition and scalar multiplication defined on V .

22
Examples of subspaces

Example-1. Let V be a vector space over the field F . Then the

subset {0} of V is a subspace of V and it is called the zero
subspace.

23
Theorem 1

Let V be a vector space over the field F . A non-empty subset W of

V is a subspace of V if and only if

∀ α, β ∈ W , c ∈ F =⇒ cα + β ∈ W .

Proof:
Case 1: Suppose that W is a subspace of V . So, W is a vector
space over the field F . Thus, if c ∈ F and α ∈ W , then cα ∈ W
(closed under scalar multiplication). Again, as W is closed under
vector addition, for any vector β ∈ W one has cα + β ∈ W . Hence

∀ α, β ∈ W , c ∈ F =⇒ cα + β ∈ W .

24
Theorem 1 contd.

Case 2: Suppose that W is a non-empty subset of V with the

property that

∀ α, β ∈ W , c ∈ F =⇒ cα + β ∈ W . − − − −(1)

Since W ̸= ϕ, there exists γ ∈ W and hence 0 = (−1)γ + γ ∈ W ,

by equation (1).
For all α, β ∈ W , α + β = 1 · α + β ∈ W by equation (1).
For all α ∈ W , cα = cα + 0 ∈ W by equation (1).
Furthermore, −α = (−1)α + 0 ∈ W for all α ∈ W by (1).
Since W ⊆ V , ⟨W , F , +, .⟩ satisfies the rest of the axioms (verify!)
of a vector space and thus W is a subspace of V .

25
Examples of subspaces

Example-2. Note that W = {(0, x2 . . . , xn ) : xi ∈ F } is a subspace

of F n .
Proof: Clearly 0 = (0, 0, . . . , 0) ∈ W . So ϕ ̸= W ⊆ F n . Let
α = (0, x2 , . . . , xn ), β = (0, y2 , . . . , yn ) ∈ W and c ∈ F .

cα + β = (0, cx2 + y2 , . . . , cxn + yn ) ∈ W

By Theorem 1, W is a subspace of F n .

Example-3. Prove that W = {(1 + x2 , x2 , x3 , . . . , xn ) : xi ∈ F } is

not a subspace of F n .
Reason: 0 = (0, 0, . . . , 0) ∈
/ W.

26
Examples contd.

Example-4. Prove that the solution set of the homogeneous system

AX = 0 is subspace of F n×1 where A ∈ F m×n .

Let S = {X ∈ F n×1 : AX = 0}. Clearly 0 ∈ S ̸= ϕ.

Let X1 , X2 ∈ S. Then AX1 = AX2 = 0.

=⇒ A(cX1 + X2 ) = cAX1 + AX2 = 0,

This implies

∀ X1 , X2 ∈ S, c ∈ F =⇒ cX1 + X2 ∈ S.

Hence, by Theorem 1, S is a subspace of F n×1 .

27
Theorem 2

Let V be a vector space over the field F . Let W1 , W2 be two

subspaces of V . Then W1 ∩ W2 is a subspace of V .
Proof: Since W1 and W2 are subspace of V . Since W1 and W2 are
themselves vector spaces, 0 ∈ W1 and 0 ∈ W2 . Thus,
0 ∈ W1 ∩ W2 , that implies W1 ∩ W2 ̸= ∅.
Now, let α, β ∈ W1 ∩ W2 and c ∈ F . This implies α, β ∈ W1 and
α, β ∈ W2 . Since W1 and W2 are subspaces of V , by Theorem 1,
we have cα + β ∈ W1 and cα + β ∈ W2 . This implies
cα + β ∈ W1 ∩ W2 . Hence, by Theorem 1, W1 ∩ W2 is a subspace
of V .
Corollary: Intersection of any collection of subspaces of a vector
space V is a subspace of V .

28
The subspace spanned by S

Definition: Let S be a subset of a vector space V . The subspace

spanned by S is defined as the intersection all subspaces of V which
contains S.
\
Subspace spanned by S = {W : S ⊆ W , W is a subspace of V }.

Note-1. Subspace spanned by S is the smallest subspace which

contains S.
Note-2. If S = {α1 , α2 , . . . , αn }, we call the subspace spanned by
S as the subspace spanned by the vectors α1 , α2 , . . . , αn .

29
L(S)= the set of all linear combinations of vectors in S

Let S be a non-empty subset of a vector space V . The set of all

linear combinations of vectors in S is denoted by L(S). In other
words
( n )
X
L(S) = ci αi : ci ∈ F , αi ∈ S, n ∈ N .
i=1

Example. Let S = {(1, 0, 0), (0, 0, 1)}. Then

L(S) = {a(1, 0, 0) + b(0, 0, 1) : a, b ∈ R}

= {(a, 0, b) : a, b ∈ R} .

30
Lemma

S ⊆ L(S) and L(S) is a subspace of V .

Proof: Since S ̸= ϕ, let S = {α1 , α2 , . . . , αn }. Observe that

α1 = 1 · α1 + 0 · α2 + · · · + 0 · αn . So, α1 ∈ L(S). Similarly, it can
be shown that αi ∈ L(S). Therefore, S ⊆ L(S) and hence, L(S) is
nonempty.
To show L(S) is a subspace of V , by Theorem 1, it is equivalent to
check that for all α, β ∈ L(S) and c ∈ F , cα + β ∈ L(S).
Since α, β ∈ L(S), we have α = ni=1 ci αi and β = ni=1 di αi .
P P

Thus, cα + β = ni=1 (cci )αi + ni=1 di αi = ni=1 (cci + di )αi is a

P P P

linear combination of vectors in S. Thus cα + β ∈ L(S). Hence,

L(S) is a subspace of V .

31
Theorem 3

Let S be a non-empty subset of a vector space V over the field F .

Then the subspace spanned by the set S is the set of all linear
combinations of vectors in S.
Proof.
Let \
W ∗ = {W : S ⊆ W , W is a subspace of V }

We want to show that

W ∗ = L(S) − − − (a)

By the previous lemma, S ⊆ L(S) and L(S) is a subspace of V , and

thus
W ∗ ⊆ L(S) − − − −(i).

32
Theorem 3 contd.

It remains to show that L(S) ⊂ W ∗ . It is enough to show that, if

W is a subspace containing S, then L(S) ⊆ W .
Let x ∈ L(S). Then x is a linear combination of vectors in S. Since
W is a subspace of V and S ⊆ W , every linear combination of
vectors in S is also a member of W and thus x ∈ W .

x ∈ L(S) =⇒ x ∈ W . Thus L(S) ⊆ W .

Hence we proved that

\
L(S) ⊆ {W : S ⊆ W , W is a subspace of V } = W ∗ − −(ii).

By (i) and (ii),

\
W∗ = {W : S ⊆ W , W is a subspace of V } = L(S)−−−(a).
33
Note 1: ( Visit previous lecture notes)

Find the solution space of the system RX = 0

 
0 1 −3 0 21 )
x2 − 3x3 + 21 x5 = 0
R= 0 0 0 1 2 
 
x4 + 2x5 = 0
0 0 0 0 0

No. of non-zero rows of R, r = 2, No. of variables, n = 5.

k1 = 2, k2 = 4 =⇒ Pivot variables = {xk1 , xk2 } = {x2 , x4 }.
No. of free variables = n − r = 5 − 2 = 3.
Free variables = {x1 , x3 , x5 }.
Set the free variables as: x1 = a, x3 = b, x5 = c
=⇒ x2 = 3b − 12 c, x4 = −2c.

34
Note 1 contd. (back to chapter one !)

Solution set S = (a, 3b − 12 c, b, −2c, c) : a, b, c ∈ R



 
1
S = a(1, 0, 0, 0, 0) + b(0, 3, 1, 0, 0) + c(0, − , 0, −2, 1) : a, b, c ∈ R
2

 
1
= Span of (1, 0, 0, 0, 0), (0, 3, 1, 0, 0), (0, − , 0, −2, 1) .
2

35
Problem

Let W be set of all (x1 , x2 , x3 , x4 , x5 ) ∈ R5 which satisfies

2x1 − x2 + 34 x3 − x4 =0
2
x1 + 3 x3 − x5 = 0
9x1 − 3x2 + 6x3 − 3x4 − 3x5 = 0

Find a finite set of vectors which spans W .

36
Row space and Column space of a matrix

Let A ∈ F m×n with rows {R1 , R2 , . . . , Rm } and columns

{C1 , C2 , . . . , Cn }. Then

Row space of A = The subspace spanned by R1 , R2 , . . . , Rm .

Column space of A = The subspace spanned by C1 , C2 , . . . , Cn .

Note .: Row space of A ⊆ F 1×n and Column space of

A ⊆ F m×1 .

37
Example

" #
1 0 0
Let A =
0 1 0
where
! ! !
1 0 0
R1 = (1, 0, 0), R2 = (0, 1, 0), C1 = , C2 = , C3 =
0 1 0

Row space of A

= {a(1, 0, 0) + b(0, 1, 0) : a, b ∈ F } = {(a, b, 0) : a, b ∈ F } .

Column Space of A
( ! ! ! ) ( !)
1 0 0 a
= a +b +c : a, b, c ∈ F = . 38
0 1 0 b
Observations

(i) Column space of AB is same as column space of A.

(ii) Row space of AB is same as row space of B.

39
Observations cont...


" # x
a11 a12 a13  
AX =  y 
a21 a22 a23
z
" #
a11 x + a12 y + a13 z
=
a21 x + a22 y + a23 z
! ! !
a11 a12 a13
=x +y +z
a21 a22 a23

= xC1 + yC2 + zC3 (Ci is the ith column of A).

(1) AX is a linear combination of columns of the matrix A.

(2) Every column of AB is a linear combination of columns of A.
(3) Every row of AB is a linear combination of rows of B.
40
Linearly Dependent (L.D.) and Linearly Independent (L.I.)

Definition:
Let V be a vector space over the field F . A subset
S = {α1 , α2 , . . . , αn } of V is said to be linearly dependent if there
exist scalars c1 , c2 , . . . , cn ∈ F , ci ̸= 0 for at least one i, such that

c1 α1 + c2 α2 + . . . + cn αn = 0.

Definition:
A set which is not linearly dependent is called a linearly independent
set.
In other words, if c1 α1 + c2 α2 + . . . + cn αn = 0, then
c1 = c2 = · · · = cn = 0.

41
Note

1. Any set which contains a linearly dependent set is linearly

dependent.
2. Any subset of a linearly independent set is linearly
independent.
3. Any set which contains the 0 vector is linearly dependent.
Reason 1 · 0 = 0.

42
Problem 1

Show that α1 = (3, 0, −3), α2 = (−1, 1, 2), α3 = (4, 2, −2) and

α4 = (2, 1, 1) are linearly dependent (L.D.) on R 3 .

Solution: Find scalars c1 , c2 , c3 , c4 (at least one ci ̸= 0) such that

c1 α1 + c2 α2 + c3 α3 + c4 α4 = 0.

2α1 + 2α2 − α3 + 0α4 = 0.

43
Problem 2

Show that e1 = (1, 0, 0), e2 = (0, 1, 0) and e3 = (0, 0, 1) is a

linearly independent subset of F 3 .
Solution: Consider c1 e1 + c2 e2 + c3 e3 = 0

=⇒ c1 (1, 0, 0) + c2 (0, 1, 0) + c3 (0, 0, 1) = (0, 0, 0)

=⇒ (c1 , c2 , c3 ) = (0, 0, 0)

=⇒ c1 = c2 = c3 = 0.

Hence {e1 , e2 , e3 } is a L.I . subset of F 3 .

44
Note

Note:
{e1 = (1, 0, 0, . . . , 0, 0), e2 = (0, 1, 0, . . . , 0, 0) . . . , en = (0, 0, 0, . . . , 0, 1)}
is a linearly independent subset of F n .

45
Applications

(i) The columns of A forms a linearly independent set if and

only if AX = 0 has only trivial solution.

(ii) If A is an invertible matrix if and only if the columns of A

forms a linearly independent set (By note (i) and Theorem 13,
chapter 1).

46
Basis

Definition:
Let V be a vector space over the field F . A non-empty set B ⊆ V
is a basis for V if

1. B is a linearly independent subset of V and

2. V = span B (= L(B) ).

Definition:
A vector space V is called finite dimensional if it has a finite basis.

47
Example-1

The set B =
{e1 = (1, 0, 0, . . . , 0, 0), e2 = (0, 1, 0, . . . , 0, 0) . . . , en = (0, 0, 0, . . . , 0, 1)}
is a basis of Rn .
Verification:
Claim 1: B is a linearly independent set in Rn .
Consider

c1 (1, 0, . . . , 0)+c2 (0, 1, 0, . . . , 0)+· · ·+cn (0, 0, . . . , 0, 1) = (0, 0, . . . , 0)

=⇒ (c1 , c2 , . . . , cn ) = (0, 0, . . . , 0)

=⇒ c1 = c2 = · · · = cn = 0

=⇒ B is a L.I . set.

48
Example-1 contd.

Claim 2 : Rn = span B. That is every vector of Rn can be written

as a linear combination of vectors of B.
Let x = (x1 , x2 , . . . , xn ) be any arbitrary vector in Rn . As

(x1 , x2 , . . . , xn ) = x1 (1, 0, . . . , 0)+x2 (0, 1, . . . , 0)+· · ·+xn (0, . . . , 0, 1)

=⇒ (x1 , x2 , . . . , xn ) = x1 e1 + x2 e2 + . . . + xn en

we get that
span B = Rn .

Hence, by claims 1 and 2, B is a basis of Rn .

Note: B = {e1 , e2 , . . . , en } is called the standard basis of Rn .

49
Example-2

The set B =
{e1 = (1, 0, 0, . . . , 0, 0), e2 = (0, 1, 0, . . . , 0, 0) . . . , en = (0, 0, 0, . . . , 0, 1)}
is a basis of the vector space F n , where F is a field.

50
Example-3

The set B = {(0, 1, 1), (1, 0, 1), (1, 1, 0)} is a basis for R3 .
Verification:
Claim 1: B is a linearly independent set in R3 .
Consider

c1 (0, 1, 1) + c2 (1, 0, 1) + c3 (1, 1, 0) = (0, 0, 0)

=⇒ (0, c1 , c1 ) + (c2 , 0, c2 ) + (c3 , c3 , 0) = (0, 0, 0)

=⇒ (c2 + c3 , c1 + c3 , c1 + c2 ) = (0, 0, 0)

=⇒ c2 + c3 = 0, c1 + c3 = 0, c1 + c2 = 0

On solving these three equations we obtain

c1 = 0, c2 = 0 and c3 = 0. Thus, B is a linearly independent set
of R3 .
51
Example-3 contd.

Claim 2: R3 = span B. That is, we have to show that every vector

of R3 is a linear combination of vectors of B.
Let (x, y , z) be an arbitrary vector of R3 . We want to find scalars
c1 , c2 , c3 such that

c1 (0, 1, 1) + c2 (1, 0, 1) + c3 (1, 1, 0) = (x, y , z)

=⇒ (0, c1 , c1 ) + (c2 , 0, c2 ) + (c3 , c3 , 0) = (x, y , z)

=⇒ (c2 + c3 , c1 + c3 , c1 + c2 ) = (x, y , z)

=⇒ c2 + c3 = x, c1 + c3 = y , c1 + c2 = z

52
Example-3 contd.

Upon solving these three equations we obtain

c1 = −x+y
2
+z
, c2 = x−y2 +z and c3 = x+y2 −z . Thus,

(x, y , z)
−x + y + z x −y +z x +y −z
= (0, 1, 1) + (1, 0, 1) + (1, 1, 0).
2 2 2
This implies
span B = R3 .

Hence, B is a basis of R3 .

53
Problem-3:

Find a basis of the solution space of the system RX = 0, where

 
0 1 −3 0 12
R =  0 0 0 1 2 .
 

0 0 0 0 0

Solution.
No. of non-zero rows of R, r = 2, No. of variables, n = 5.
k1 = 2, k2 = 4 =⇒ Pivot variables = {xk1 , xk2 } = {x2 , x4 }.
No. of free variables = n − r = 5 − 2 = 3.
Free variables = {x1 , x3 , x5 }.
Set the free variables as: x1 = a, x3 = b, x5 = c
=⇒ x2 = 3b − 12 c, x4 = −2c.

54
Problem-3 contd...

Solution set S = (a, 3b − 12 c, b, −2c, c) : a, b, c ∈ R



 
1
S = a(1, 0, 0, 0, 0) + b(0, 3, 1, 0, 0) + c(0, − , 0, −2, 1) : a, b, c ∈ R
2

 
1
= Span of (1, 0, 0, 0, 0), (0, 3, 1, 0, 0), (0, − , 0, −2, 1) .
2

Let
1
E1 = (1, 0, 0, 0, 0), E3 = (0, 3, 1, 0, 0) and E5 = (0, − , 0, −2, 1).
2

55
Problem-3 contd...

We prove that {E1 , E3 , E5 } is a linearly independet set.

Let
1
c1 (1, 0, 0, 0, 0) + c2 (0, 3, 1, 0, 0) + c3 (0, − , 0, −2, 1) = (0, 0, 0, 0, 0)
2
1
⇒ (c1 , 3c2 − c3 , c2 , −2c3 , c3 ) = (0, 0, 0, 0, 0)
2
This implies c1 = 0, c2 = 0 and c3 = 0. Thus, {E1 , E3 , E5 } is a
linearly independet set.
Hence {E1 , E3 , E5 } is a basis of the solution space S.

56
Problem 4 (assignment)

Find a basis for the solution set S of the system of equations

2x1 − x2 + 34 x3 − x4 =0
2
x1 + 3 x3 − x5 = 0 .
9x1 − 3x2 + 6x3 − 3x4 − 3x5 = 0

57
Problem-5

Find a basis for the vector space F m×n of all m × n matrices over
the field F .
Solution: Define the matrices Ei,j in the following way:
(i). The ij th entry of matrix Ei,j is one
(ii). All other entries are zero.
Verify that the set B = {Eij : i = 1, 2, . . . , m; j = 1, 2, . . . , n} is a
basis for F m×n .

This is an extension of the idea of the standard basis of F n .

58
The column vectors of an invertible matrix form a basis

Let P ∈ F n×n be an invertible matrix. Let P1 , P2 , . . . , Pn be the

columns of P. Show that B = {P1 , P2 , . . . , Pn } is a basis of F n×1 .
Claim 1: B is a L.I. set.
Consider x1 P1 + x2 P2 + . . . + xn Pn = 0.
 
x1
h i x2 
=⇒ P1 P2 . . . Pn =0
 

 ... 
xn

=⇒ PX = 0

=⇒ X = 0, ( P is an invertible matrix)

59
Contd.

=⇒ x1 = x2 = . . . = xn = 0 =⇒ B is an L.I. set

Claim 2: F n×1 = Span B. That is, every column vector can be

expressed as a linear combination of column vectors of matrix P.
Let Y ∈ F n×1 be any column vector. Consider the matrix
multiplication P −1 Y , denote this by X . So, Y = PX . As the
matrix PX can be written as a linear combination of column vectors
of P, we have

Y = PX = x1 P1 + x2 P2 + . . . + xn Pn ∈ L({P1 , P2 , . . . , Pn })

Thus, F n×1 = Span B.

Hence, by claims 1 and 2, B (the set of all columns of P) is a basis
of F n×1 . 60
A simple way of checking a set of vectors in F n is a basis or not

Idea: Given any set B = {α1 , α2 , . . . , αn } of n vectors in F n ,

consider the matrix A where the column vectors are precisely
α1 , α2 , . . . , αn . Show that A is invertible. Then by above
observation the set B = {α1 , α2 , . . . , αn } is a basis for F n .

61
Theorem 4

Let V be a vector space which is spanned by a finite set of vectors

β1 , β2 , . . . , βm . Then any linearly independent set of vectors in V
contains no more than m elements.
Proof: It suffices to show that every subset of V which contains
more than m elements is linearly dependent (L.D.). Let S be such a
set. Let S = {α1 , α2 , . . . , αn }, where n > m.
Since span B = V , we have
α1 = A11 β1 + A21 β2 + · · · + Am1 βm
α2 = A12 β1 + A22 β2 + · · · + Am2 βm
..
.
αn = A1n β1 + A2n β2 + · · · + Amn βm .

62
Theorem 4 contd.

For each j = 1, 2, . . . , n, we have

m
X
αj = A1j β1 + A2j β2 + · · · + Amj βm = Aij βi .
i=1
We want to show that S is L.D. That means we are searching for
some x1 , x2 , . . . , xn , at least one of them is not zero, such that

x1 α1 + x2 α2 + . . . + xn αn = 0 − − − − − −(i)

63
Theorem 4 contd.

Consider
x1 α1 + x2 α2 + . . . + xn αn
n
X
= xj αj
j=1

n m
!
X X
= xj Aij βi
j=1 i=1
 
m
X n
X
=  Aij xj  βi (This is just an rearrangement of terms)
i=1 j=1
     
n
X n
X Xn
= A1j xj  β1 +  A2j xj  β2 + · · · +  Amj xj  βm .
j=1 j=1 j=1

64
Theorem 4 contd.

Consider the system of equations

n
X
A1j xj = 0
j=1

n
X
A2j xj = 0
j=1

..
.
n
X
Amj xj = 0
j=1

This is a homogeneous linear system with m equations and n

variables.
65
Theorem 4 contd.

Since m < n, this system has a non-trivial solution say

x1∗ , x2∗ , . . . , xn∗ (at least one xj∗ ̸= 0), that is
X n
Aij xj∗ = 0, for each i = 1, 2, . . . , m − − − (ii).
j=1

Therefore we got x1∗ , x2∗ , . . . , xn∗ (at least one xj∗ ̸= 0) such that
 
m
X Xn
x1∗ α1 + x2∗ α2 + . . . + xn∗ αn =  Aij xj∗  βi = 0 (by (ii))
i=1 j=1

Hence the set S = {α1 , α2 , . . . , αn } is a L.D. set.

This completes the proof.

66
Corollaries to Theorem 4

Corollary 1. If V is a finite dimensional vector space, then any two

bases of V have the same number of elements.
Proof: B1 = {β1 , β2 , . . . , βm } and B2 = {α1 , α2 , . . . , αn } be two
bases of V . Then by definition
(i) B1 is L.I. and span B1 = V .
(ii) B2 is L.I. and span B2 = V .
Since span B1 = V and B2 is an L.I. set, n ≤ m − − − (a) (by
Theorem 4).
Similarly, since span B2 = V and B1 is an L.I. set, m ≤ n − − − (b)
(by Theorem 4).
Hence, by (a) and (b), m = n.
This completes the proof of the corollary.
67
Dimension

Definition:
The dimension of a finite-dimensional vector space V is the number
of elements in a basis for V .
The dimension of the zero vector space is zero.
If V is not a finite dimensional vector space we say it is an infinite
dimensional vector space.

68
Examples

Example-1.
As B = {e1 = (1, 0, 0, . . . , 0, 0), e2 = (0, 1, 0, . . . , 0, 0), . . . ,
en = (0, 0, 0, . . . , 0, 1)} is a basis of F n over F ,

dimension of F n = dim(F n ) = n.

Example-2. The dimension of F m×n = mn,

as the set {Eij : i = 1, 2, . . . , m; j = 1, 2, . . . , n} (see previous
lecture) is a basis of F m×n .

69
Examples

Example-3. Let AX = 0 be a system of homogeneous equations,

where A is an m × n matrix, X is an n × 1 column vector, and 0 is
the m × 1 zero vector. Let R be the row-reduced-echelon form of
matrix A and r be the number of non-zero rows of R. Then the
dimension of the solution space of the homogeneous system of linear
equations AX = 0 is of dimension n − r (the number of free
variables).

70
Examples

For example consider the system of equations RX = 0, where

 
0 1 −3 0 12
R =  0 0 0 1 2 .
 

0 0 0 0 0

Solution space
 
1
S = Span of (1, 0, 0, 0, 0), (0, 3, 1, 0, 0), (0, − , 0, −2, 1) .
2

We have also verified that the set

(1, 0, 0, 0, 0), (0, 3, 1, 0, 0), (0, − 21 , 0, −2, 1) is a basis of S.


Hence, the dimension of the solution space S is 3 = 5 − 2.

71
Examples of infinite dimensional vector spaces

Example-4. Let P be the set of all polynomials with real coefficients.

Verify that the set {1, x, x 2 , x 3 , . . . , x n , . . .} is a basis for P. Since
{1, x, x 2 , x 3 , . . . , x n , . . .} is an infinite set, the vector space P is an
infinite dimensional vector space.
Example-5. Let C [a, b] be the set of all continuous functions from
[a, b] to R. Then C [a, b] is an infinite dimensional vector space.

72
Corollaries to Theorem 4

Corollary 2. Let V be a finite dimensional vector space and let n =

dim V . Then
1. any subset of V which contains more than n vectors is a L.D.;
2. no subset of V which contains fewer than n vectors can span
V.
Proof. Let B = {β1 , β2 , . . . , βn } be a basis of V . Then
(i) B is L.I. and
(ii) span B = V .
Proof of (1). Obvious from Theorem 4.
Proof of (2). Suppose that V = span {γ1 , γ2 , . . . , γp }. Since B is a
linearly independent set and span B = V , by Theorem 4 we get
n ≤ p. That means, any set of vectors which spans V contains at
least n vectors. This proves (2).
73
Lemma

Let S be a linearly independent subset of a vector space V .

Suppose that there is a vector β ∈ V − L(S). Then S ∪ {β} is
an L.I. subset of V .
Proof. Let S = {α1 , α2 , . . . , αn }.
Claim. The set {α1 , α2 , . . . , αn , β} is an L.I. set.
Consider

c1 α1 + c2 α2 + · · · + cn αn + bβ = 0 − − − − − (i)

Then b = 0; otherwise β = − cb1 α1 − c2

b α2 − ··· − cn
b αn ∈ L(S), a
contradiction.

74
Proof continued

From Equation (1) we have

c1 α1 + c2 α2 + · · · + cn αn = 0.

Since S is an L.I. set, we must have c1 = c2 = · · · = cn = 0. As we

already have cn+1 = 0, the set {α1 , α2 , . . . , αn , β} is an L.I. subset
of V .

75
Theorem 5

In a finite-dimensional vector space V every non-empty linearly

independent set of vectors is part of a basis.
Proof: Let S0 be a linearly independent subset of V .
Claim: S0 is a part of a basis for W .
We extend S0 to a basis for V in the following way.
If span S0 = V , then we are done, as S0 is L.I. and span S0 = V ,
hence a basis for V .
If not, that is span S0 ̸= V , then there is an element (say)
β1 ∈ V − span S0 . Let S1 = S0 ∪ {β1 }. By the previous lemma S1
is an L.I. subset of V .

76
Theorem 5 contd

If span S1 = V , then we are done, as S1 is L.I. and span S1 = V ,

hence a basis for V .
If not, apply the previous lemma to obtain a β2 ∈ V − span (S1 )
such that S2 = S1 ∪ {β2 } is an L.I. set.
If we continue in this way, then (in not more than dim V steps) we
reach an L.I. set

Sm = S0 ∪ {β1 , β2 , . . . , βm }

which is a basis for V .

77
Example

Let S0 = {(1, 1, 1)}. Find a basis for R 3 which contains S0 .

Solution :

L(S0 ) = {a(1, 1, 1) : a ∈ R} = {(a, a, a) : a ∈ R}

Clearly, β1 = (1, 1, 0) ∈
/ L(S0 ). By Theorem 5,
S1 = S0 ∪ {β1 } = {(1, 1, 1), (1, 1, 0)} is a L.I. subset of R 3 .

L(S1 ) = {a(1, 1, 1) + b(1, 1, 0) = (a + b, a + b, a) : a, b ∈ R}

Clearly β2 = (1, 0, 0) ∈
/ L(S1 ). By Theorem 5,
S2 = S1 ∪ {β2 } = {(1, 1, 1), (1, 1, 0), (1, 0, 0)} is a L.I. set. Verify
that L(S2 ) = R 3 . Hence, S2 is a basis for R 3 .
78
Ordered Basis

Definition:
Let V be a finite-dimensional vector space. An ordered basis for V
is a finite sequence of vectors which is linearly independent and
spans V .

Note: Let B = {α1 , α2 , . . . , αn } be an ordered basis for V . Let

α ∈ V = span {α1 , α2 , . . . , αn }. Then there exist some scalars
x1 , x2 , . . . , xn such that

α = x1 α1 + x2 α2 + · · · + xn αn − − − −(1)

This representation is unique.

79
If not, then there exist scalars y1 , y2 , . . . , yn ∈ F such that

α = y1 α1 + y2 α2 + · · · + yn αn − − − −(2)

From (1) and (2), we get

x1 α1 + x2 α2 + · · · + xn αn = α = y1 α1 + y2 α2 + · · · + yn αn .

This implies

(x1 − y1 )α1 + (x2 − y2 )α2 + · · · + (xn − yn )αn = 0.

Since B is linearly independent, we have

x1 − y1 = x2 − y2 = · · · = xn − yn = 0.

Thus, x1 = y1 , x2 = y2 , . . . , xn = yn . Hence the representation is

unique. 80
Coordinate Matrix

Definition:
The coordinate matrix of the vector α relative to the ordered basis
B is  
x1
 x2 
 
[α]B =  . 

.
 .. 
xn

81
Example

Example: Find the coordinate matrices of the vector α = (1, 2, 3)

with respect to the bases
B1 = {ϵ1 = (1, 0, 0), ϵ2 = (0, 1, 0), ϵ3 = (0, 0, 1)} and
B2 = {α1 = (1, 1, 1), α2 = (0, 1, 1), α3 = (0, 0, 1)}.
Solution: Note that

α = (1, 2, 3) = 1 · ϵ1 + 2 · ϵ2 + 3 · ϵ3

and
α = (1, 2, 3) = 1 · α1 + 1 · α2 + 1 · α3 .

Therefore    
1 1
[α]B1 =  2  and [α]B2 =  1  .
   

3 1
82
Relation between [α]B1 and [α]B2 ?

Note that
α1 = ϵ1 + ϵ2 + ϵ3 ,

α2 = 0ϵ1 + ϵ2 + ϵ3

and
α3 = 0ϵ1 + 0ϵ2 + ϵ3 .

So, we have
     
1 0 0
P1 = [α1 ]B1 =  1  , P2 = [α2 ]B1 =  1  and P3 = [α3 ]B1 =  0  .
     

1 1 1

83
Relation between [α]B1 and [α]B2 ?

Let  
1 0 0
P = [P1 , P2 , P3 ] =  1 1 0 
 

1 1 1
Then
    
1 0 0 1 1
P · [α]B2 =  1 1 0   1  =  2  = [α]B1 .
    

1 1 1 1 3

Verify that the matrix P is invertible and [α]B2 = P −1 [α]B1 .

84
Relation between [α]B1 and [α]B2 ?

This phenomenon also holds in an arbitrary finite dimensional

vector space V .

Proof: Let B1 = {α1 , α2 , . . . , αn } and B2 = {β1 , β2 , . . . , βn } be

two ordered bases of a finite-dimensional vector space V . Let
α ∈ V . Let
Xn Xn
α= xi αi and α = yj βj
i=1 j=1

be the unique representations of α with respect to the bases B1 and

B2 , respectively. Thus,
   
x1 y1
x2 y2
   
   
[α]B1 = ..  and [α]B = 
2 .. .
. .
   
   
xn yn 85
contd..

Since βj ∈ V = span {α1 , α2 , . . . , αn } for each j = 1, 2, . . . , n,

there exist unique scalars P1j , P2j , . . . , Pnj such that

βj = P1j α1 + P2j α2 + · · · + Pnj αn .

That is
β1 = P11 α1 + P21 α2 + · · · + Pn1 αn

β2 = P12 α1 + P22 α2 + · · · + Pn2 αn

..
.

βn = P1n α1 + P2n α2 + · · · + Pnn αn .

86
contd.

Now,
α = y1 β1 + y2 β2 + · · · + yn βn
= y1 (P11 α1 + P21 α2 + · · · + Pn1 αn )
+ y2 (P12 α1 + P22 α2 + · · · + Pn2 αn )
..
.
+ yn (P1n α1 + P2n α2 + · · · + Pnn αn )
= (P11 y1 + P12 y2 + · · · + P1n yn )α1
+ (P21 y1 + P22 y2 + · · · + P2n yn )α2
..
.
+ (Pn1 y1 + Pn2 y2 + · · · + Pnn yn )αn .

87
contd.

But, we already have α = x1 α1 + x2 α2 + · · · + xn αn , and since the

representation of α relative to the basis B1 is unique, we have

x1 = P11 y1 + P12 y2 + · · · + P1n yn

x2 = P21 y1 + P22 y2 + · · · + P2n yn

..
.

xn = Pn1 y1 + Pn2 y2 + · · · + Pnn yn .

88
contd.

Writing these equations in matrix form we get

    
x1 P11 P12 · · · P1n y1
x2 P21 P22 · · · P2n y2
    
    
 .. = .. .. .. ..  .. .
. . . . . .
    
    
xn Pn1 Pn2 · · · Pnn yn

=⇒ X = PY , − − − − −(1)

where X = [α]B1 and Y = [α]B2 .

89
contd.

Next, we claim that the matrix P is invertible, that we show in the

following two steps:
Claim (1): X = 0 ⇐⇒ Y = 0.
Proof: Let X = 0. This implies x1 = x2 = · · · = xn = 0, that
further implies α = 0.
As α = y1 β1 + y2 β2 + · · · + yn βn = 0 and B2 = {β1 , β2 , . . . , βn } is
linearly independent, we get y1 = y2 = · · · = yn = 0. Thus, Y = 0.

By the same argument we can prove that if Y = 0 then X = 0.

Claim (2): P is an invertible matrix.
Proof: Consider the system of equations PY = 0. As X = PY , we
get X = 0. Then from Claim 1 it follows that Y = 0. This means
the sytem PY = 0 has only the trivial solution. Hence, by Theorem
7 (of chapter 1) we get that P is invertible. 90
Theorem 7

We just proved the following theorem:

Theorem 7.
Let V be a n-dimensional vector space over the field F and let
B1 = {α1 , α2 , . . . , αn } and B2 = {β1 , β2 , . . . , βn } be two ordered
bases of V . Then there is a unique, necessarily invertible, n × n
matrix P with entries in F such that (i) [α]B1 = P[α]B2 and (ii)
[α]B2 = P −1 [α]B1 for every α ∈ V . The columns of P are given by
Pj = [βj ]B1 for j = 1, 2, . . . , n.

91
An application

Note: For a given ordered basis B1 of a finite-dimensional vector

space V and a given invertible matrix P, it is possible to construct
another ordered basis B2 of V .

Example: Let B1 =
{α1 = (0, 1, 1, 1), α2 = (1, 0, 1, 1), α3 = (1, 1, 0, 1), α4 = (1, 1, 1, 0)}
be an ordered basis for R 4 and let
 
1 0 1 0
 1 0 0 0 
P=
 

 0 1 0 0 
0 1 4 2

be an invertible matrix. Find an ordered basis for R 4 different from

the basis B1 .
92
Solution

We want to find β1 , β2 , β3 , and β4 such that

   
1 0
 1   0 
[β1 ]B1 = P1 =  , [β2 ]B1 = P2 =  ,
   
 0   1 
0 1
   
1 0
 0   0 
[β3 ]B1 = P3 =  , [β4 ]B1 = P4 =  .
   
 0   0 
4 2

93
solution contd.

From the definition of coordinate matrices, it follows that

β1 = 1α1 + 1α2 + 0α3 + 0α4 = (1, 1, 2, 2)

β2 = 0α1 + 0α2 + 1α3 + 1α4 = (2, 2, 1, 1)

β3 = 1α1 + 0α2 + 0α3 + 4α4 = (4, 5, 5, 1)

β4 = 0α1 + 0α2 + 0α3 + 2α4 = (2, 2, 2, 0).

Hence, we got another ordered basis

B2 = {(1, 1, 2, 2), (2, 2, 1, 1), (4, 5, 5, 1), (2, 2, 2, 0)}

different from the given ordered basis B1 .

94
Row rank / Column rank

Let A be an m × n matrix over a field F . Let {R1 , R2 , . . . , Rm } be

the rows of A and {C1 , C2 , . . . , Cn } be the columns of A.
Row space of A = span {R1 , R2 , . . . , Rm }.
Column space of A = span {C1 , C2 , . . . , Cn }.
Then the row space of A is a subspace of F 1×n and the column
space of A is a subspace of F m×1 .

Definitions:
Row rank of A = dimension of row space of A.
Column rank of A = dimension of column space of A.

95
Basis of a row-reduced echelon matrix

Let A be an m × n matrix over the field F . Let R be the

row-reduced echelon form of A. The non-zero rows of R forms a
basis for the row space of R. Since two row-equivalent matrices
have same row space, thus a basis for the row space of R is also a
basis for the row space of A.

Hence

Row rank of A = Row rank of R = No. of non-zero rows of R.

96