Elementary Linear Algebra (2024-25 @IITH)

Amit Tripathi

Indian Institute of Technology, Hyderabad

(IIT) Elementary Linear Algebra (2024-25 @IITH) 1 / 33

Isomorphism of vector spaces

Definition
An isomorphism of the vector spaces V and W is a bijective linear
transformation T : V → W . When the map defining the isomorphism is
clear from the context, we write

V ∼
= W.
(a) Id : R2 → R2 is an isomorphism.
(b) T : R2 → R2 , T (x1 , x2 ) = (x1 , x1 + x2 ) is an isomorphism.
(c) Is every non-zero linear transformation R → R an isomorphism?
(d) Is every non-zero linear transformation R2 → R2 an isomorphism?

(IIT) Elementary Linear Algebra (2024-25 @IITH) 2 / 33

Isomorphism of vector spaces

Question: Suppose T : V → W is a bijective linear transformation and

let S : W → V be the inverse map. Is S also a linear transformation?

In particular, we conclude that isomorphism is an equivalence relation on

the set of all (finite dimensional) vector spaces.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 3 / 33

Isomorphisms

Remark
It is clear by rank-nullity that if T is an isomorphism, then

dim(V ) = dim(W ).

The converse also holds:

Theorem
Let V be finite dimensional vector space. Then V is isomorphic to a
vector space W if and only if they have the same dimension.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 4 / 33

Isomorphisms
Proof.
One side implication is clear. Conversely, suppose dim(V ) = dim(W ) = n.
Sufficient to show that
V ∼= F n.
Let {α1 , · · · , αn } be an ordered basis. Then there is a unique linear
transformation T : V → F n such that

T (α1 ) =e1 , · · · · · · , T (αn ) = en .

It is clear that T is a bijective map. Thus, T defines an isomorphism

which completes the proof.

In particular, any finite dimensional vector space V is isomorphic to F n

(upto a choice of basis of V ) where n = dim(V ).

(IIT) Elementary Linear Algebra (2024-25 @IITH) 5 / 33

Direct sum

Definition
Let W and W ′ be two subspaces of V such that W ∩ W ′ = {0}. Then
W + W ′ is also written as W ⊕ W ′ .

(a) Let V = R2 and let W be the subspace spanned by the vector (1, 0)
and W ′ be the subspace spanned by the vector (0, 1). Then
V = W ⊕ W ′.
(b) If V = W ⊕ W ′ , then any vector α can be written uniquely as

α = β + β′

where β ∈ W and β ′ ∈ W ′ . For example,

(x, y ) = (x, 0) + (0, y ).

(IIT) Elementary Linear Algebra (2024-25 @IITH) 6 / 33

Representation of linear transformations by matrices

Let T : V → W be a linear transformation between finite dimensional

vector spaces.
Let B = {α1 , · · · , αn } be a basis for V and C = {β1 , · · · , βm } be a basis
for W .
Each of the vector T (αi ) ∈ W can be expressed in terms of the basis
{β1 , · · · , βm }. Set

T (α1 ) = a11 β1 + a21 β2 + · · · + am1 βm

T (α2 ) = a12 β1 + a22 β2 + · · · + am2 βm (1)
······
T (αn ) = a1n β1 + a2n β2 + · · · + amn βm

where aij ∈ R.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 7 / 33

Representation of linear transformations by matrices

Thus, if we know the mn scalars {aij }, and the bases {αi }, {βj }, then we
have determined T . We define a m × n matrix A as following:

The (i, j)’th entry of A is aij .

In other words  
a11 a12 ··· a1n
 a21 a22 ··· a2n 
A=
· · · · · ·

··· 
am1 am2 ··· amn

(IIT) Elementary Linear Algebra (2024-25 @IITH) 8 / 33

Representation of linear transformations by matrices
The matrix A determines T completely (in terms of bases B and C).
We show how: Let v = c1 α1 + · · · + cn αn be any vector in V . Then the
coordinate vector of v with respect to the basis {α1 , · · · , αn } is the
column vector  
c1
 c2 
V = 
· · ·
cn
Let us see the image of V when it is multiplied with A above. We have
    
a11 a12 · · · a1n c1 a11 c1 + a12 c2 + · · · + a1n cn
 a21 a22 · · · a2n 
  c2  =  a21 c1 + a22 c2 + · · · + a2n cn 
   
AV =  · · · ··· ···  · · ·  ··· 
am1 am2 · · · amn cn am1 c1 + am2 c2 + · · · + amn cn

(IIT) Elementary Linear Algebra (2024-25 @IITH) 9 / 33

Representation of linear transformations by matrices

The matrix AV is an m × 1 column vector. We set

w = (a11 c1 +a12 c2 +· · ·+a1n cn )β1 +· · ·+(am1 c1 +am2 c2 +· · ·+amn cn )βm .

We started with an arbitrary vector v ∈ V , expressed it in terms of the

basis {αi } of V and then multiplied the corresponding coordinate vector
by A and ended up with coordinates of a vector w ∈ W .
Question: What is the relation between v and w .

(IIT) Elementary Linear Algebra (2024-25 @IITH) 10 / 33

Representation of linear transformations by matrices

We have

T (v ) = T (c1 α1 + · · · + cn αn )
= c1 T (α1 ) + · · · + cn T (αn )
= c1 (a11 β1 + a21 β2 + · · · + am1 βm ) + · · ·
+ cn (a1n β1 + a2n β2 + · · · + amn βm )
= (a11 c1 + a12 c2 + · · · + a1n cn ) β1 + · · ·
+ (am1 c1 + am2 c2 + · · · + amn cn ) βm
= w.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 11 / 33

Representation of linear transformations by matrices

Theorem
Let V be an n-dimensional vector space and let W be an m-dimensional
vector space. Let B be an ordered basis of V and C be an ordered basis of
W . For each linear transformation T : V → W , there is an m × n matrix
A such that
[T (α)]C = A[α]B , for each vector α ∈ V .
Here the (i, j)th entry of A is obtained as

T (αj ) = a1j β1 + · · · + aij βi + · · · + amj βm .

(IIT) Elementary Linear Algebra (2024-25 @IITH) 12 / 33

Representation of linear transformations by matrices

Note that the matrix corresponding to T will depend upon the choses
basis of both the vector spaces V and W .
An important case is when V = W i.e., when T is a linear operator.
In this case, we can take C = B. We shall then denote the matrix of T
with respect to the basis B by [T ]B , and the linear transformation can
be described in terms of matrices as

[T (α)]B = [T ]B [α]B .

Remark
It is important to keep in mind that the matrix representation of a linear
transformation is always with respect to chosen basis for the domain and
the codomain vector spaces.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 13 / 33

Example

Example
Let V = R2 and let T : V → V be the linear transformation defined as

T (x1 , x2 ) = (x1 , 0).

Let B = {e1 , e2 } be the standard ordered basis for V . Then

T (e1 ) = (1, 0) = 1 · e1 + 0 · e2 = a11 e1 + a21 e2

T (e2 ) = (0, 0) = 0 · e1 + 0 · e2 = a12 e1 + a22 e2

Thus, we get  
1 0
[T ]B =
0 0

(IIT) Elementary Linear Algebra (2024-25 @IITH) 14 / 33

Example

Example
Now, let C = {e1′ , e2′ } be another ordered basis where e1′ = (1, 1) and
e2′ = (1, 0).

T (e1 ) = e1 = 0 · e1′ + 1 · e2′ = a11 e1′ + a21 e2′

T (e2 ) = (0, 0) = 0 · e1′ + 0 · e2′ = a12 e1′ + a22 e2′

Thus, we get  
0 0
[T ]B,C =
1 0
where the use of subscripts B and C is to emphasis that the matrix
representation of T depends upon the chosen basis.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 15 / 33

Example

Example
Let V be a finite dimensional vector space and let B = {α1 , · · · , αn } be a
basis of V . Let T : V → V be the identity operator i.e. T (α) = α for all
α ∈ V . Then

T (α1 ) = α1 = 1 · α1 + 0 · α2 + · · · + 0 · αn
T (α2 ) = α2 = 0 · α1 + 1 · α2 + · · · + 0 · αn
············
T (αn ) = αn = 0 · α1 + 0 · α2 + · · · + 1 · αn

It follows that the matrix of T with respect to any arbitrary basis B is the
identity matrix.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 16 / 33

Rotation revisited

Example
    
x cos(θ) − sin(θ) x
Rθ : R2 → R2 , 7→ .
y sin(θ) cos(θ) y
In other words,

(x, y ) 7→ (x cos(θ) − y sin(θ), x sin(θ) + y cos(θ)).

As R vector spaces
R2 ∼
= C.
How does Rθ looks over C ?

(IIT) Elementary Linear Algebra (2024-25 @IITH) 17 / 33

Rotation revisited

Example
Rθ
R2 / R2

 
C
? /C

Question: How does reflection over x-axis looks like?

(IIT) Elementary Linear Algebra (2024-25 @IITH) 18 / 33

Representation of linear transformations by matrices

Let V , W , and Z be finite dimensional vector spaces and let B, C, D be

ordered bases for V , W , and Z , respectively. Let T : V → W and
S : W → Z be linear transformations. So the situation is as below:

T S
V,B W,C Z, D

(IIT) Elementary Linear Algebra (2024-25 @IITH) 19 / 33

Representation of linear transformations by matrices

Let A be the matrix of T with respect to the basis B and C. Let B be the
matrix of S with respect to the basis C and D.

[T ]B,C = A, and [S]C,D = B.

Composing the two linear transformation S and T , we get

S◦T
V,B Z, D

(IIT) Elementary Linear Algebra (2024-25 @IITH) 20 / 33

Representation of linear transformations by matrices

Question: What is the matrix of the composed map S ◦ T with respect

to the basis B on V and D on W ?
It is the product of the corresponding matrices.

[S ◦ T ]B,D = BA.

Remark
Let V be a finite dimensional vector space and let B be a basis of V . A
linear operator
T :V →V
is an isomorphism if and only if the associated matrix [T ]B is invertible. In
fact, we have
[T −1 ]B = ([T ]B )−1 .

(IIT) Elementary Linear Algebra (2024-25 @IITH) 21 / 33

The set of all linear transformations between two vector
spaces

Let A be an m × n matrix and let

TA : Rn → Rm

be the linear transformation induced by A.

Question: What is the matrix of TA with respect to the standard basis
on Rn and Rm ?

(IIT) Elementary Linear Algebra (2024-25 @IITH) 22 / 33

The set of all linear transformations between two vector
spaces

Fix standard ordered basis on Rn and Rm . Consider two sets:

(a) The set of linear transformations T : Rm → Rn .
(b) The set of m × n matrices.
These two sets are bijective. The second set is a vector space.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 23 / 33

The set of all linear transformations between two vector
spaces

Let Hom(V , W ) denote the set of all the linear transformations from V to
W.
Let T , T ′ ∈ Hom(V , W ), and let c ∈ R, we define T + T ′ and cT as
follows: For any v ∈ V , set

(T + T ′ )(v ) = T (v ) + T ′ (v )
(cT )(v ) = c(T (v )).

With these vector addition and scalar multiplication, Hom(V , W ) is a

vector space.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 24 / 33

The set of all linear transformations between two vector
spaces

Question: Suppose dim(V ) = n and dim(W ) = m. What is the dimension

of the linear space
Hom(V , W )
mn

(IIT) Elementary Linear Algebra (2024-25 @IITH) 25 / 33

Linear Operators

Definition
A linear operator is a linear transformation from a vector space to itself.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 26 / 33

Eigenvectors and eigenvalues

Definition
Let T : V → V be a linear operator. Suppose there is a scalar c and a
nonzero vector v ∈ V such that

T (v ) = cv .

Then we say that c is an eigenvalue of T and v is an eigenvector.

Observe that if v is an eigenvector, then λv is also an eigenvector for any

scalar λ ̸= 0.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 27 / 33

Eigenvectors and eigenvalues

Example
Define T : R3 → R3 as

T (x, y , z) = (2z, x + z, 2x).

Then v = (v1 , v2 , v3 ) ∈ R3 is an eigenvector if for some scalar c,

T (v ) = cv , i.e.
(2v3 , v1 + v3 , 2v1 ) = c(v1 , v2 , v3 ).
Equating coordinates, we get the following equations

2v3 = cv1
v1 + v3 = cv2
2v1 = cv3 .

(IIT) Elementary Linear Algebra (2024-25 @IITH) 28 / 33

Eigenvectors and eigenvalues

Example
Solving them we get following possibilities:
(a) v1 = v2 = v3 = 0.
(b) c = 2 and v1 = v2 = v3 .
(c) c = −2, v1 = −v3 and v2 = 0.
The first solution gives v = 0 which is not allowed. With eigenvalue
c = 2, one possible eigenvector is (1, 1, 1) and with eigenvalue c = −2,
one possible eigenvector is (1, 0, −1).

(IIT) Elementary Linear Algebra (2024-25 @IITH) 29 / 33

Eigenvectors and eigenvalues

Theorem
A scalar λ is an eigenvalue if and only if T − λI has no inverse.
Equivalently, if A is a matrix representation of T with respect to some
basis B, then

λ is an eigenvalue ⇐⇒ |A − λI | = 0.

Proof.
Any eigenvalue λ corresponds to some eigenvector v ̸= 0. So we have
Av = λv . This is equivalent to saying that

(A − λI )X = 0,

has a non-trivial solution which is equivalent to the assertion.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 30 / 33

Eigenvectors corresponding to distinct eigenvalues

Theorem
Let T : V → V be a linear operator and let λ1 , · · · , λn be distinct
eigenvalues. Let vi ∈ V be an eigenvector corresponding λi for
i = 1, · · · , n. Then the set {v1 , · · · , vn } is linearly independent.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 31 / 33

Eigenvectors corresponding to distinct eigenvalues

Proof.
We argue by contradiction. Suppose v1 , · · · , vn are dependent. Since any
eigenvector is by definition non-zero, we get that the set {v1 } is linearly
independent. Let m be the first integer ≥ 1 such that the set {v1 , · · · , vm }
is linearly independent but the set {v1 , · · · , vm+1 } is linearly dependent.
Let
c1 v1 + · · · + cm vm + cm+1 vm+1 = 0,
be a linear relation. Applying T on both sides, we get

c1 λ1 v1 + · · · + cm+1 λm+1 vm+1 = 0.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 32 / 33

Eigenvectors corresponding to distinct eigenvalues

Proof.
Multiplying the first equation by λm+1 and subtracting from the second,
we get
c1 (λ1 − λm+1 )v1 + · · · + cm (λm − λm+1 )vm = 0.
Since the set {v1 , · · · , vm } is assumed to be independent, we conclude that

c1 (λ1 − λm+1 ) = · · · = cm (λm − λm+1 ) = 0.

Since λi are all distinct eigenvalues, we get

c1 = · · · = cm = 0,

which completes the proof.

(IIT) Elementary Linear Algebra (2024-25 @IITH) 33 / 33