1 Lineartransformations: Linear Transformation

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Lecture Notes by M. G. P.

Prasad, IITG

Class Notes for the Lectures on the Topic of Linear Transformations (MA102)

1 LinearTransformations
Definition 1.1. Let V and W be vector spaces over the field F . A linear transformation
from V into W is a function T : V → W such that

T (u + v) = T u + T v for all vectors u and v in V ,

T (α v) = α T u for all scalars α ∈ F and for all vectors v ∈ V .


The two conditions in the above definition can be written as a single condition as
follows:

T (αu + βv) = α T u + β T v for all vectors u, v ∈ V and for all scalars α, β ∈ F .

Examples:
• Let V and W be vector spaces over the field F . The Zero Tranformation 0 : V → W
defined by 0v = 0 for all v ∈ V , is a linear transformaton from V into W .
• Let V be a vector space over the field F . The Identity Transformation I : V → V
defined I(v) = v for all v ∈ V is a linear transformtion from V into V .
• Let P denote the vector space of all real polynomials over the real field. Let P (x) =
a0 + a1 x + a2 x2 + · · · + an xn ∈ P. Define a map T : P → P by
d
(T P )(x) = (P (x)) = a1 +2a2 x+· · ·+nan xn−1 for all x ∈ R, for each P ∈ P .
dx
Then, T is a linear transformation from P into P and it called differential transfor-
mation / operator.
• Let A be a fixed m × n matrix with entries in the field R. The function T : Rn → Rm
defined by T x = Ax for x = (x1 , . . . , xn )T ∈ Rn is a linear transformation from Rn
into Rm . The function S : Rm → Rn defined by Sx = xA for x = (x1 , . . . , xm ) ∈ Rm
is a linear transformation from Rm into Rn .
• Let C(R) = {f : R → R : f is continuous on R} denote the vector space of all con-
tinuous functions from R into R over the real field. Define T : C(R) → C(R) by
Z x
(T f ) (x) = f (t) dt for all x ∈ R, for each f ∈ C(R) .
0

Then T is a linear transformation from C(R) into C(R). The function T f is not only
continuous but it has a continuous derivative.

Page No. 1
Lecture Notes by M. G. P. Prasad, IITG

• Let P denote the vector space of all real polynomials over the real field. Let P (x) =
a0 + a1 x + a2 x2 + · · · + an xn ∈ P. Define a map T : P → R by
(T P )(x) = P (3) for all x ∈ R, for each P ∈ P .
Then, T is a linear transformation from P into R.

Theorem 1.1. Let V and W be vector spaces over the field F . Let T be a linear transfro-
mation from V into W . Then
1. T (0) = 0.
2. T (−v) = −v for all v ∈ V .
3. T (u − v) = T u − T v for all u ∈ V and v ∈ V .
Proof.
Proof of (1):
Observe that
T (0) = T (0 + 0) = T (0) + T (0) (by Linearity property)
This gives that,
T (0) − T (0) = T (0) =⇒ 0 = T (0) .

Proof of (2):
Observe that for any vector v ∈ V ,
T (v) + T (−v) = T (v + (−v)) = T (0) = 0 .

Proof of (3):
Observe that for all u ∈ V and v ∈ V ,
T (u − v) = T (u + (−v)) = T (u) + T (−v) = T (u) + (−1)T (v) = T u − T v .

Question: Is T : R → R defined by T (x) = x + 1 for all x ∈ R linear transformation?


Answer: T (x) = x + 1 is not linear because T (0) = 1 6= 0.

Question: Can there be a linear transformation T : R2 → R2 satisfying T (1, 0) = (3, 2),


T (0, 1) = (2, 3) and T (1, 1) = (3, 3)?
Answer: Observe that T ((1, 0) + (0, 1)) = T (1, 0) + T (0, 1) = (3, 2) + (2, 3) = (5, 5). Also
observe that T (1, 1) = T ((1, 0) + (0, 1)) = (5, 5) 6= (3, 3) and hence no such linear tranfor-
mation exists.

Page No. 2
Lecture Notes by M. G. P. Prasad, IITG

Theorem 1.2. Let V be finite dimensional vector space over the field F with dim (V ) = n.
Let B = {u1 , u2 , . . . , un } be an (ordered) basis for V . Let W be a vector space over the
same field F and let u1 , . . ., un be arbitrary vectors in W . There there is a unique linear
transformation T : V → W such that T (ui ) = wi for all i = 1, 2, . . ., n.

Proof. Let v ∈ V . Since B is an ordered basis for V , there exists scalars α1 , α2 , . . ., αn


such that n
X
v= αi ui .
i=1

Define a map T : V → W by
n n
!
X X
T (v) = αi wi for each v = αi ui ∈V .
i=1 i=1

Observe that T is well defined, because for each vector v, its coordinates [v]B = (α1 , . . . , αn )
with respect to the ordered basis B is unique.
Further T satisfies T (ui ) = wi for i = 1, 2, . . ., n.

Claim: T is linear
Let u = α1 u1 + α2 u2 + · · · + αn un and v = β1 u1 + β2 u2 + · · · + βn un be any two vectors in
V and let a and b any two scalars in F . Then

au + bv = (aα1 + bβ1 ) u1 + · · · + (aαn + bβn ) un .

Now,
T (au + bv) = T ((aα1 + bβ1 ) u1 + · · · + (aαn + bβn ) un )
= (aα1 + bβ1 ) w1 + · · · + (aαn + bβn ) wn
= aα1 w1 + · · · + aαn wn + bβ1 w1 + · · · + bβn wn
= a (α1 w1 + · · · + αn wn ) + b (β1 w1 + · · · + βn wn )
= aT (u) + bT (v) .

Claim: T is unique
Let S be a linear transformation from V to W such that S(ui ) = wi for i = 1, 2, . . ., n.
For the vector v = ni=1 αi ui , we have
P

n
!
X
S(v) = S αi ui
i=1

n
X
= αi S(ui )
i=1

Page No. 3
Lecture Notes by M. G. P. Prasad, IITG

n
X
= αi w i
i=1

so that S is exactly the rule T which we defined above. This shows that the linear trans-
formation T with T (ui ) = wi for each i is unique.
This completes the proof.

Note: The above theorem shows that a linear transformation is completely determined by
its action on any basis of domain. One can see it in the following example.

Example: Find a unique linear transformation T : R2 → R3 such that T (1, 2) = (3, 2, 1)


and T (3, 4) = (6, 5, 4).
Observe that B = {u1 = (1, 2), u2 = (3, 4)} is an ordered basis for R2 . Then, by previous
theorem, there is a unique linear transformation from R2 into R3 such that T (ui ) = wi for
i = 1, 2. Here w1 = (3, 2, 1) and w2 = (6, 5, 4).
Step 1: Expressing Standard Ordered Basis Vectors in terms of given ordered basis.
Consider the standard ordered basis of R2 , namely, Bs = {e1 = (1, 0), e2 = (0, 1)}. Express
e1 and e2 as linear combination of vectors of the given ordered basis B as follows.

e1 = (1, 0) = −2u1 + u2 = −2(1, 2) + (3, 4) ,


       
3 −1 3 −1
e2 = (0, 1) = u1 + u2 = (1, 2) + (3, 4) .
2 2 2 2

Step 2: Finding T (ei ) for each i

T (e1 ) = T (−2u1 + u2 ) = −2T (u1 ) + u2 ) = −2(3, 2, 1) + (6, 5, 4) = (0, 1, 2) ,


T (e2 ) = T ((3/2)u1 + (−1/2)u2 ) = (3/2)T (u1 ) + (−1/2)T (u2 )
= (3/2)(3, 2, 1) + (−1/2)(6, 5, 4) = ((3/2), (1/2), (−1/2)) .

Step 3: Finding T (v)


Observe that any vector v = (x, y) ∈ R2 can be written as

v = (x, y) = xe1 + ye2 = x(1, 0) + y(0, 1) .

Therefore

T (v) = T (x, y) = T (xe1 + ye2 ) = xT (e1 ) + yT (e2 ) = x(0, 1, 2) + y ((3/2), (1/2), (−1/2))

T (x, y) = ((3/2)y, x + (1/2)y, 2x + (−1/2)y)

Example: Let P2 (R) denote the vector space of all real polynomials of degree atmost 2
over the real field. Let T be a linear transformation T : R2 → P2 (R) such that T (1, 0) =

Page No. 4
Lecture Notes by M. G. P. Prasad, IITG

2 − 3x + x2 and T (0, 1) = 1 − x2 . Then, find T (2, 3) and T (a, b)


Observe that
T (2, 3) = T (2(1, 0) + 3(0, 1)) = 2T (1, 0) + 3T (0, 1)
= 2(2 − 3x + x2 ) + 3(1 − x2 ) = 7 − 6x − x2 .
Similarly,
T (a, b) = T (a(1, 0) + b(0, 1)) = aT (1, 0) + bT (0, 1)
= a(2 − 3x + x2 ) + b(1 − x2 ) = (2a + b) + (−3a)x + (a − b)x2 .

Definition 1.2. Let V and W be vector spaces over the field F and let T be a linear
transformation from V into W . Then, the range space of T , denoted by range (T ) and the
kernel of T , denoted by ker (T ) are defined as

Range of T = range (T ) = {T (v) ∈ W : v ∈ V } ,

Kernel of T = ker (T ) = {v ∈ V : T v = 0} .

The kernel of T is also called the null space of T .


Theorem 1.3. Let V and W be vector spaces over the field F and let T be a linear
transformation from V into W . Then, the kernel of T is a subspace of V . That is,
ker (T ) 4 V .

Proof.
Since T (0) = 0, we have 0 ∈ ker(T ) and hence the kernel of T is non-empty.
Let v1 and v2 be any two vectors in the kernel of T and let c be any scalar in F .
To show: cv1 + v2 is in the kernel of T
Since v1 ∈ ker (T ), we have T (v1 ) = 0. Similarly, since v2 ∈ ker (T ), we have T (v2 ) = 0.
Now,
T (cv1 + v2 ) = c T (v1 ) + T (v2 ), since T is linear
=c0 + 0 = 0.
Therefore, cv1 + v2 ∈ ker (T ). Thus, the kernel of T is a subspace of V .
This completes the proof.

Theorem 1.4. Let V and W be vector spaces over the field F and let T be a linear
transformation from V into W . Then, the range space of T is a subspace of W . That is,
range (T ) 4 W .

Proof.
Since T (0) = 0, we have 0 ∈ range(T ) and hence the range of T is non-empty.
Let w1 and w2 be any two vectors in the range of T and let c be any scalar in F .
To show: cw1 + w2 is in the range of T

Page No. 5
Lecture Notes by M. G. P. Prasad, IITG

Since w1 ∈ range (T ), there exists v1 ∈ V such that T (v1 ) = w1 . Similarly, since w2 ∈


range (T ), there exists v2 ∈ V such that T (v2 ) = w2 .
Since v1 ∈ V and v2 ∈ V and c ∈ F , we have
cv1 + v2 ∈ V .
Since T is linear, we have
T (cv1 + v2 ) = c T (v1 ) + T (v2 ) = c w1 + w2 .
Since w1 ∈ W , w2 ∈ W and c ∈ F , we have cw1 + w2 ∈ W because W is a vector space.
Since T (cv1 + v2 ) = cw1 + w2 ∈ W , we conclude that (cv1 + v2 ) ∈ range (T ). Therefore,
the range space of T is a subspace of W . This completes the proof.

Example: Let A be an m × n real matrix. Let T : Rn → Rm be the linear transformation


defined by T x = Ax for all x ∈ Rn . Find range (T ) and ker (T ).
Observe that
range (T ) = {T x ∈ Rm : x ∈ Rn } = {Ax : x ∈ Rn } = Col (A) 4 Rm .
ker (T ) = {x ∈ Rn : T x = 0} = {x ∈ Rn : Ax = 0} = Null (A) 4 Rn .

Definition 1.3. Let V and W be vector spaces over the field F and let T be a linear
transformation from V into W . Assume that V is finite-dimensional. Then the rank of T
is the dimension of the range space of T and the nullity of T is the dimension of the kernel
(or null space) of T .
Example: Let P3 denote the vector space of all real polynomials of degree atmost 3 over
the real field. Let T : P3 → P3 be the linear transformation defined by (T P )(x) = P 0 (x).
Fine the rank and nullity of T .
Oberve that any polynomial P (x) = a0 + a1 x + a2 x2 ∈ P2 can be written as
d  a1 a2 
P (x) = a0 + a1 x + a2 x2 = a0 x + x 2 + x 3 .
dx 2 3
The range of T is all of P2 and hence the rank of T is the dimension of P2 which is equal
to 3.
The kernel of T is given by
ker (T ) = {P ∈ P3 : T (P ) = P 0 is a zero polynomial } = {All constant polynomials } .
Therefore, the nullity of T is 1.

Exercise: Read and understand Examples 6.61, 6.62, 6.65, 6.66 in David Poole’s Book
(4th Edition).

The following theorem is one of most important results in linear algebra.

Page No. 6
Lecture Notes by M. G. P. Prasad, IITG

Theorem 1.5. Rank-Nullity Theorem (or Rank Theorem):


Let V and W be vector spaces over the field F and let T : V → W be a linear transformation
from V into W . Suppose that V is finite-dimensional. Then
rank (T ) + nullity (T ) = dim (V ) .
Proof.
Given that V is a finite-dimensional vector space and hence dim (V ) = n (say).
Since ker (T ) 4 V and V is finite-dimensional, ker (T ) is also finite-dimensional.
Let {v1 , v2 , . . . , vk } be a basis for ker (T ) and hence nullity (T ) = dim (ker (T )) = k.

Since {v1 , v2 , . . . , vk } is a linearly independent set in V , it can be extended to form a


basis for V . Let B = {v1 , v2 , . . . , vk , vk+1 , . . . , vn } be such a basis for V .

We shall now prove that C = {T vk+1 , . . . , T vn } is a basis for the range of T .


Claim 1: C spans range (T )
Let w ∈ range (T ). Then, there exists v ∈ V such that T (v) = w.
Since v ∈ V and B is a basis for V , we can find scalars c1 , . . ., cn such that
v = c1 v1 + . . . + ck vk + ck+1 vk+1 + . . . + cn vn .
Now,
T v = T (c1 v1 + . . . + ck vk + ck+1 vk+1 + . . . + cn vn )
= c1 T (v1 ) + . . . + ck T (vk ) + ck+1 T (vk+1 ) + . . . + cn T (vn )
Since v1 , . . ., vk are in ker (T ), we have
T (vi ) = 0 for i = 1, 2, . . . , k .
Therefore
T v = ck+1 T (vk+1 ) + · · · + cn T (vn ) .
This shows that the range of T is spanned by C .

Claim 2: C is linearly independent


Suppose that there are scalars ck+1 , . . ., cn such that
ck+1 T (vk+1 ) + · · · + cn T (vn ) = 0 .
Since T is linear, we have
T (ck+1 vk+1 + · · · + cn vn ) = 0 .
This means that ck+1 vk+1 + · · · + cn vn is in the kernel of T and hence it is expressible as a
linear combination of the basis vectors v1 , . . ., vk of ker (T ) as follows:
ck+1 vk+1 + · · · + cn vn = c1 v1 + · · · + ck vk

Page No. 7
Lecture Notes by M. G. P. Prasad, IITG

It gives that
c1 v1 + · · · + ck vk − ck+1 vk+1 − · · · − cn vn = 0 .
Since B is a basis for V , it follows that c1 = · · · = cn = 0. In particular, ck+1 = · · · = cn =
0. Therefore C is linearly independent.

Since C is a basis for the range of T , it follows that the rank of T is equal to (n − k).

rank (T ) + nullity (T ) = (n − k) + k = n = dim (V ) .

This completes the proof.

Example: Let V be the vector space of all symmetric 2 × 2 real matrices over the real
field. Let P2 denote the vector space of all real polynomials of degree atmost 2 over the
real field. Define a linear transformation T : V → P2 by
 
a b
T = (a − b) + (b − c)x + (c − a)x2 .
b c

Find the rank and nullity of T .


The nullity of T is easier to compute directly than the rank. So we proceed to compute
ker (T ).

    
a b a b
ker (T ) = : T =0
b c b c
  
a b 2
= : (a − b) + (b − c)x + (c − a)x = 0 for all x ∈ R
b c
  
a b
= : (a − b) = (b − c) = (c − a) = 0
b c
  
a b
= : a=b=c
bc
  
c c
= : c∈R
c c
 
1 1
= span
1 1
 
1 1
Therefore, is a basis for the kernel of T and hence the nullity of T is 1.
1 1
Observe that the dimension of V is 3.
By the Rank-Nullity theorem, it follows that

rank (T ) = dim (V ) − nullity (T ) = 3 − 1 = 2 .

Page No. 8
Lecture Notes by M. G. P. Prasad, IITG

Exercise: Read and understand Example 6.67 in David Poole’s Book (4th Edition).

Composition of Linear Transformations:


Let U , V , W be vector spaces over the same field F . If T : U → V and S : V → W are lin-
ear transformations, then the composition ST is the function ST : U → W defined by the
equation (ST )(x) = S(T (x)) for every x ∈ U is also a linear transformation from U into W .

Algebraic Operations on Linear Transformations and Forming a Vector Space:


Let V and W be vector spaces over the same field F .
Let T , T1 and T2 be linear transformations from V into W . Let c be any scalar in F . We
define the addition and scalar multiplication as follows:

(T1 + T2 )(x) = T1 (x) + T2 (x) for all x ∈ V ,

(c T )(x) = c T (x) for all x ∈ V .


Then, the set of all linear transformations of V into W together with these operations
(addition and scalar multiplication) will form a vector space over the field F .
We shall denote it by

L(V, W ) = {T : V → W : T is a linear transformation } .

We remind that L(V, W ) is defined only when V and W are vector spaces over the same
field.
In case of L(V, V ), we simply write it as L(V ).
If V is a vector space over the field F then a linear tranformation T : V → V from V into
V itself is called a linear operator on V .
If V is a vector space over the field F then a linear tranformation f : V → F from the
vector space V into the field F is called a linear functional on V .

Theorem 1.6. Let V be an n-dimensional vector space over the field F and let W be an
m-dimensional vector space over the (same) field F . Then, the vector space L(V, W ) is
finite-dimensional and has dimension mn.

One-to-One (Injective), Onto (Surjective) and Bijective:

Definition 1.4. Let V and W be vector spaces over the field F and let T : V → W be a
function from V into W .
The function T : V → W is said to be one-to-one (or one-one or injective) if T maps
distinct vectors in V to distinct vectors in W .
The function T : V → W is said to be onto (or surjective) if the range of T is equal to W .
The function T : V → W is said to be bijective if T is both one-to-one and onto.

Page No. 9
Lecture Notes by M. G. P. Prasad, IITG

That is, T : V → W is one-to-one if, for all u and v in V ,

u 6= v implies that T (u) 6= T (v) .

The above statement is equivalent to the following:


T : V → W is one-to-one if, for all u and v in V ,

T (u) = T (v) implies that u=v.

Another way to write the definition of onto is as follows:


T : V → W is onto if for each w ∈ W , there is at least one v in V such that T (v) = w.

Example 1: Let T : R2 → R3 be defined by T ([x, y]T ) = [2x, (x − y), 0]T for each
[x, y]T ∈ R2 .
First observe that if T ([x1 , y1 ]T ) = T ([x2 , y2 ]T ) then [x1 , y1 ]T = [x2 , y2 ]T .
Suppose that
   
2x1 2x2
T ([x1 , y1 ]T ) = T ([x2 , y2 ]T ) =⇒ (x1 − y1 ) = (x2 − y2 ) .
0 0

This gives that 2x1 = 2x2 and (x1 − y1 ) = (x2 − y2 ). Solving these equations, we see that
x1 = x2 and y1 = y2 . Therefore, [x1 , y1 ]T = [x2 , y2 ]T .
Thus, T is one-to-one.

Observe that [0, 0, 1]T ∈ R3 , but there is no vector [x, y]T ∈ R2 such that T ([x, y]T ) =
[0, 0, 1]T . Therefore, T is NOT onto.

Example 2: Let T : P3 → P2 be defined by T (P ) = P 0 .


Observe that the range of T is P2 and hence T is onto.

Observe that P (x) = x3 and Q(x) = x3 + 1 are mapped by T to the same polynomial
R(x) = 3x2 . Therefore T is NOT one-to-one.

Example 3: Let T : M2 (R) → M2 (R) be defined by T (A) = AT for each A ∈ M2 (R).


Suppose that A and B in M2 (R) with T (A) = T (B). Then,

T (A) = T (B) =⇒ (A)T = B T =⇒ A = B .

Therefore T is one-to-one.

Observe that for each A ∈ M2 , there exists B = AT such that T (B) = (AT )T = A.

Page No. 10
Lecture Notes by M. G. P. Prasad, IITG

Therefore, T is onto.

Theorem 1.7. A linear transformation T : V → W is one-to-one if and only if ker (T ) =


{0}.

Proof.
Proof of ⇒:
Given that T is one-to-one. To show: ker (T ) = {0}.
If v ∈ ker (T ) then T (v) = 0.
We also know that T (0) = 0. Therefore T (v) = T (0).
Since T is one-to-one, T (v) = T (0) implies that v = 0.
Therefore, ker (T ) = {0}.

Proof of ⇐:
Given that ker (T ) = {0}. To show: T is one-to-one.
Let u and v be in V such that T (u) = T (v). To show: u = v.
Now,
T (u − v) = T (u) − T (v) = 0 =⇒ u − v ∈ ker (T ) .
Since ker (T ) = {0}, we get

(u − v) = 0 =⇒ u=v.

Theorem 1.8. Let V and W be finite-dimensional vector spaces over the field F . Suppose
that dim (V ) = dim (W ). Then T is one-to-one if and only if T is onto.

Proof.
Given that dim (V ) = dim (W ) = n (say).
Proof of ⇒:
Also given that T : V → W is one-to-one. To show: T is onto.
Since T is one-to-one, by previous theorem, it follows that ker (T ) = {0} and hence the
nullity of T is 0.
Then, by Rank-Nullity Theorem, it follows that

rank (T ) = dim (V ) − nullity (T ) = n − 0 = n .

Since rank (T ) = n, range (T ) 4 W and dim (W ) = n, it follows that range (T ) = W .


Therefore T is onto.

Proof of ⇐:

Page No. 11
Lecture Notes by M. G. P. Prasad, IITG

Given that T is onto. To show: T is one-to-one. Sufficient To show: ker (T ) = {0}


Since T is onto, it follows that range (T ) = W and hence rank (T ) = dim (W ) = n.
Then, by Rank-Nullity Theorem, it follows that
nullity (T ) = dim (V ) − rank (T ) = n − n = 0 .
Since nullity (T ) = 0, it follows that ker (T ) = {0}.
This completes the proof.

Theorem 1.9. Let V and W be finite-dimensional vector spaces over the field F . If
dim (V ) > dim (W ), then any linear transformation T : V → W is NOT one-to-one
(injective).
Proof.
Given that dim (V ) > dim (W ).
To show: Any linear map T : V → W is not one-to-one.
We know that range (T ) 4 W and hence rank (T ) ≤ dim (W ). It gives that −rank (T ) ≥
−dim (W ).
We know that
nullity (T ) = dim (V ) − rank (T ) ≥ dim (V ) − dim (W ) > 0 .
Since nullity (T ) > 0, the kernel of T must contain vectors other than 0 and hece T can
not be one-to-one.

Theorem 1.10. Let V and W be finite-dimensional vector spaces over the field F . If
dim (V ) < dim (W ), then any linear transformation T : V → W is NOT onto (surjective).
Proof. Given that dim (V ) < dim (W ).
To show: Any linear map T : V → W is not onto.
We know that
rank (T ) = dim (V ) − nullity (T )
≤ dim (V )
< dim (W )
Since rank (T ) < dim (W ), the subspace range (T ) is a proper subset of W and hence T is
not onto.

Theorem 1.11. Let V and W be vector spaces over the field F and let T : V → W be a
one-to-one linear transformation. If S = {v1 , v2 , . . . , vk } is a linearly independent set in V
then T (S) = {T (v1 ), T (v2 ), . . . , T (vk )} is a linearly independent set in W .

Page No. 12
Lecture Notes by M. G. P. Prasad, IITG

Proof.
Let c1 , . . ., ck be scalars such that

c1 T (v1 ) + · · · + ck T (vk ) = 0 .

Since T is linear, it follows that c1 T (v1 ) + · · · + ck T (vk ) = T (c1 v1 + · · · + ck vk ) and hence

T (c1 v1 + · · · + ck vk ) = 0 =⇒ (c1 v1 + · · · + ck vk ) ∈ ker (T ) .

Since T is one-to-one, ker (T ) = {0}. Therefore,

(c1 v1 + · · · + ck vk ) = 0 .

Since S is linearly independent in V , it gives that

c1 = · · · = ck = 0 .

Therefore, T (S) is linearly independent in W .

Corollary 1.1. Let V and W be finite-dimensional vector spaces over the field F with
dim (V ) = dim (W ). Let T : V → W be a one-to-one linear transformation. Then T maps
a basis for V to a basis for W .

Proof.
Given that dim (V ) = dim (W ) = n (say).
Let B = {v1 , . . . , vn } be a basis for V .
To show: T (B) = {T (v1 ), . . . , T (vn )} is a basis for W .
Since B is a basis for V , B = {v1 , . . . , vn } is a linearly independent set in V .
Then, by previous theorem, it follows that T (B) = {T (v1 ), . . . , T (vn )} is a linearly inde-
pendent set in W .
Since dim (W ) = n and T (B) is a linearly independent set in W having n elements, we
get span (T (B) = W .
Therefore, T (B) is a basis for W .

Example: Let T : R2 → P1 (R) be the linear transformation defined by


 
a
T = a + (a + b)x .
b

Find a basis for P1 (R).


Let B = {e1 = (1, 0)T , e2 = (0, 1)T } be the standard basis for R2 .
Then by above corollary, it follows that T (B) = {T (e1 ), T (e2 )} is a basis for the range
space of T .
That is, T (B) = {T (e1 ) = 1 + x, T (e2 ) = x} is a basis for the range space of T .

Page No. 13
Lecture Notes by M. G. P. Prasad, IITG

Since dim (P1 (R)) = 2, range (T ) 4 P1 (R) and rank (T ) = 2, it follows that P1 (R) =
range (T ).
Therefore, T (B) = {T (e1 ) = 1 + x, T (e2 ) = x} is a basis for P1 (R).

Definition 1.5. Let f : V → W be a function from a vector space V into a vector space
W . If there exists a function g : W → V such that f (g(w)) = w for every w ∈ W and
g(f (v)) = v for every v ∈ V , then we say that f is invertible and the function g is called
the inverse of f . The inverse of f is unique and it is denoted by f −1 .

Theorem 1.12. Let V and W be vector spaces over the field F and let T : V → W be a
linear transformation. Then, T is invertible if and only if T is both one-to-one and onto.

Definition 1.6. Let V and W be vector spaces over the field F . A linear transformation
T : V → W is an isomorphism if T is both one-to-one and onto (that is, T is bijective).
Definition 1.7. Let V and W be vector spaces over the (same) field F . Then we say that
V is isomorphic to W if there exists a linear transformation T : V → W which is both
one-to-one and onto (that is, T is an isomorphism).
If V is isomorphic to W , then we write it as V ∼
= W.

Theorem 1.13. Let V and W be vector spaces over the (same) field F . Then, V is
isomorphic to W if and only if dim (V ) = dim (W ).

Theorem 1.14. Every n-dimensional vector space over the field F is isomorphic to the
vector space F n over the field F .

Theorem 1.15. Every n-dimensional vector space over the real field R is isomorphic to
the vector space Rn over the real field R.
Theorem 1.16. Let V be a finite-dimensional vector space over the field F . Let T : V → V
be a linear operator on V . Then the following statements are equivalent:
1. T is an isomorphism.
2. T is one-to-one.
3. T is onto.
4. ker (T ) = {0}.
5. range (T ) = V .

Page No. 14
Lecture Notes by M. G. P. Prasad, IITG

2 Representation of Linear Transformations by Ma-


trices
Let V be an n-dimensional vector space over the field F and let W be an m-dimensional
vector space over the (same) field F . Let B = {v1 , . . . , vn } be an ordered basis for V and
C = {w1 , . . . , wm } be an ordered basis for W . If T : V → W is any linear transfor-
mation from V into W , then T is completely determined by its action on the
basis vectors vj .

First observe that each of the vector T vj , j = 1, 2, . . ., n is uniquely expressible as a linear


combination of basis vectors of C as follows:
m
X
T vj = aij ui ,
i=1

where the scalars a1j , a2j , . . ., amj being the coordinates of T vj in the ordered basis C .
That is,  
a1j
   a2j 
T vj C =  . .  j = 1, 2, . . . , n .
 
 .
amj
Define an m × n matrix A by
      
A = T v1 C T v2 C · · · T vn C .
 
That is, A is formed by putting j-th column as T vj C . The above matrix A is called the
matrix of T with respect to the ordered bases B and C and it is denoted by
 
T C ←B .

Then for every vector v in V , it satisfies


     
T (v) C = T C ←B v B .

The relationship is illustrated below

T
v −→ T (v)
 
y y
 
  T C ←B  
v B −→ T (v) C

Page No. 15
Lecture Notes by M. G. P. Prasad, IITG

Theorem 2.1. Let V be an n-dimensional vector space over the field F and let W be an
m-dimensional vector space over the (same) field F . Let B = {v1 , . . . , vn } be an ordered
basis for V and C = {w1 , . . . , wm } be an ordered basis for W . If T : V → W is a linear
transformation from V into W , then the m × n matrix defined by
        
T C ←B = T v1 C T v2 C · · · T vn C (1)

satisfies      
T (v) C = T C ←B v B (2)
for every vector v in V .
Moreover the matrix T C ←B given in (1) is the only matrix with the property (2).

Remark: In the above theorem, if V = W and B = C , then the matrix T B←B is simply
 
 
written as T B .

Example: Let T : R3 → R2 be the linear transformation defined by

T [x, y, z]T = [(x − 2y), (x + y − 3x)]T for every [x, y, z]T ∈ R3 .




Let B = {v1 = (1, 0, 0), v2 = (0, 1, 0), v3 = (0, 0, 1)} be the ordered basis for R3 and let
C = {w1 = (0, 1), w2 = (1, 0)} be the ordered basis for R 2 . Find the matrix of T with
respected to the ordered bases B and C . Also verify previous theorem for v = [1, 3, −2]T .
Step 1: Computing T vj for j = 1, . . ., n

T v1 = [1, 1]T , T v2 = [−2, 1]T , T v3 = [0, −3]T .

Step 2: Finding the coordinates of T vj , ∀j with respect to the ordered basis C


       
1 0 1   1
T v1 = =1 +1 =⇒ T v1 C = .
1 1 0 1
       
−2 0 1   1
T v2 = =1 + (−2) =⇒ T v1 C = .
1 1 0 −2
       
0 0 1   −3
T v3 = = (−3) + (0) =⇒ T v1 C = .
−3 1 0 0
 
Step 3: Forming the matrix T C ←B
By (1),  
         1 1 −3
T C ←B = T v1 C T v2 C T v3 C = .
1 −2 0
     
Step 4: Writing in the form of T (v) C = T C ←B v B

Page No. 16
Lecture Notes by M. G. P. Prasad, IITG

Let v = [x, y, z]T be any vector in R3 . Then,


 
  x
v B = y  .
z

Therefore
 
  x  
      1 1 −3   x + y − 3z
T (v) C = T C ←B v B = y = .
1 −2 0 x − 2y C
z B

Step 5: Verifying previous theorem for the vector v = [1, 3, −2]

T v = T [1, 3, −2]T = [−5, 10]T .




Observe that 
1
[v]B =  3  ,
−2
 
  10
T (v) C = .
−5
Using all these facts, we confirm that
 
  1  
    1 1 −3   10  
T C ←B v B = 3 = = T (v) C .
1 −2 0 −5
−2

Example: Let D : P3 → P3 be the differential operatorD(p(x)) = p0 (x). Let B =


{1, x, x2 , x3 } be the standard ordered basis for P3 . Find the matrix of D with respect to
the ordered basis B.
Step 1: Computing Dvj for j = 1, . . ., n

D(1) = 0, D(x) = 1, D(x2 ) = 2x, D(x3 ) = 3x2 .

Step 2: Finding the coordinates of Dvj , ∀j with respect to the ordered basis B
       
0 1 0 0
0 0
2 , 0 .
     2
    3
  
D(1) B = 0
 , D(x) B
=   ,
0 D(x ) B
= 0 D(x ) B
= 3
0 0 0 0
 
Step 3: Forming the matrix D B

Page No. 17
Lecture Notes by M. G. P. Prasad, IITG

By (1),  
0 1 0 0
  0 0 2 0
D B = 
0
 .
0 0 3
0 0 0 0

Example: Let T : V = R2 → W = R2 be the linear transformation defined by T (x1 , x2 ) =


((x1 + x2 ), (x1 − x2 )). Let B = {e1 = (1, 0), e2 = (0, 1)} be the standard ordered basis
for V = R2 and let C = {w1 = (1, 1), w2 = (−1, 1)} be the ordered basis for
 W =R.
2

Find the matrix of T with respect to the ordered bases B and C . Find T (v) C for every


v ∈ V = R2 .
Step 1: Computing T ej for each j

T (e1 ) = [1, 1]T , T (e2 ) = [1, −1]T .

Step 2: Finding the coordinates of T ej for each j with respect to the ordered basis C
   
  1   0
T (e1 ) C = , T (e2 ) C = .
0 −1
 
Step 3: Forming the matrix T C ←B
By (1),  
       1 0
T C ←B = T e1 C T e2 C = .
0 −1
 
Step 4: Finding T (v) C for every v ∈ V = R2
 
x
Let v = (x1 , x2 ) ∈ V = R . Then v B = 1 .
2
 
x2
By (2),     
      1 0 x1 x1
T (v) C = T C ←B v B = = .
0 −1 x2 −x2
Cross Verfication:
       
x1 x1 + x2 1 −1
T = = x1 + (−x2 ) = x1 w1 + (−x2 )w2 .
x2 x1 − x2 1 1

Exercise: Let T : R2 → R2 be the linear transformation defined by T (x1 , x2 ) = ((x1 +


x2 ), (x1 − x2 )). Let B = {u1 = (1, 1), u2 = (−1,1)} and let C = {v1 = (2, 2), v2 = (3, 2)}
be two ordered bases for R2 . Find T C ←B and T B←C .

Exercise: Let C denote the set of all complex numbers forming a vector space over the real
field R. Let T : C → C be a linear transformation defined by T (z) = i z for every z ∈ C.

Page No. 18
Lecture Notes by M. G. P. Prasad, IITG

Let B = {v1 = 1, v2 = i} be an ordered basis for C. Find T B .


 

What happens to representing matrices when transformations are added?

Theorem 2.2. Let V and W be two vector spaces over the field F with ordered bases B
and C respectively.  T : V → W and S : V → W be two linear transformations.
 Let
Let T C ←B and S C ←B denote the matrices of T and S respectively with respect to the
ordered bases B and C .

If c is any scalar in the field F , then (c T + S) is a linear transformaton from V into


W.  
Let c T + S C ←B denote the matrix of the linear transformation (c T + S) with respect to
the ordered bases B and C .
Then,      
c T + S C ←B = c T C ←B + S C ←B .

Example: Let V = R2 be the vector space with the ordered basis B = {v1 = (1, 0), v2 =
(0, 1)}. Let W = R2 be the vector space with the ordered basis C = {w1 = (1, 1), w2 =
(−1, 1)}.
Let T : V → W be the linear transformation defined by T (x1 , x2 ) = ((x1 + x2 ), (x1 − x2 )).
Let S : V → W be the linear transformation defined by S(x1 , x2 ) = (x2 , x1 ).
Let c be a fixed real number.
Then cT + S : V → W is a linear transformation given by

(cT + S)(x1 , x2 ) = ((cx1 + (1 + c)x2 ), ((1 + c)x1 − cx2 )) .


     
Verify c T + S C ←B = c T C ←B + S C ←B .
Answer:  
  1 0
T C ←B = .
0 −1
 
  (1/2) (1/2)
S C ←B = .
(1/2) (−1/2)
 
  c + (1/2) (1/2)
c T + S C ←B = .
(1/2) −c − (1/2)

What happens to representing matrices when we compose transformations?

Theorem 2.3. Let U , V and W be three vector spaces over the field F with ordered bases
B, C and D respectively. Let T : U → V and S : V → W be two linear transformations.
Then the composition (S ◦ T ) = S(T (u)) for every u ∈ U is a linear transformation from
U to W .

Page No. 19
Lecture Notes by M. G. P. Prasad, IITG

Let T C ←B denote the matrix of T with respect to the ordered bases B and C .
 

Let S D←C denote the matrix of S with respect to the ordered bases C and D.
 
 
Let S ◦ T D←B denote the matrix of the linear transformation S ◦ T with respect to the
ordered bases B and D.
Then,      
S ◦ T D←B = S D←C T C ←B .

Proof.
Let the ordered basis B for U be given by B = {u1 , . . . , up }.
We have       
S ◦ T D←B = (S ◦ T )u1 D · · · (S ◦ T )up D .
 
Now, the i-th column of S ◦ T D←B is
   
(S ◦ T )ui D = S(T (ui )) D
   
= S D←C T (ui ) C
     
= S D←C T C ←B ui B
   
= S D←C T C ←B ei
   
which is the i-th column of S D←C T C ←B .
This completes the proof.

Example:
Let U = R2 be the vector space with the ordered basis B = {u1 = (0, 1), u2 = (1, 0)}.
Let V = R2 be the vector space with the ordered basis B = {v1 = (1, 0), v2 = (0, 1)}.
Let W = R2 be the vector space with the ordered basis C = {w1 = (1, 1), w2 = (−1, 1)}.
Let T : U → V be the linear transformation defined by T (x1 , x2 ) = ((x1 + x2 ), (x1 − x2 )).
Let S : V → W be the linear transformation defined by S(x1 , x2 ) = (x2 , x1 ).
Then S ◦ T : U → W is a linear transformation given by

(S ◦ T )(x1 , x2 ) = ((x1 − x2 ), (x1 + x2 )) .


     
Verify S ◦ T D←B = S D←C T C ←B .
Answer:  
  1 1
T C ←B = .
−1 1
 
  (1/2) (1/2)
S D←C = .
(1/2) (−1/2)
 
  0 1
S ◦ T D←B = .
1 0

Exercise: Read and understand Example 6.81 in David Poole’s Book (4th Edition).

Page No. 20
Lecture Notes by M. G. P. Prasad, IITG

What happens to representing matrices of an Linear Operator when we do change of ordere

For the sake of simplicity, we shall consider this question only for linear operators T :
V → V on a finite dimensional vecto space V , so that we can use a single ordered basis B
for domain space and co-domain/ range space.
Let V be a finite dimensional vector space over the field F . Let B ={u1 , . . . ,un} and
C = {v1 , . . . , vn } be two ordered bases for V . How are the matrices T B and T C are
related?

Theorem 2.4. Let V be a finite dimensional vector space over the field F . Let B =
{u1 , . . . , un } and C = {v1 , . . . , vn } be two ordered bases for V . Let T : V → V be a linear
operator on V . If  
P = P 1 P2 · · · Pn
is an n × n matrix with columns
 
Pj = vj B for j = 1, 2, . . . , n ,

then
T C = P −1 T B P .
   

Alternatively, if S is the invertible operator on V defined by S(uj ) = vj for j = 1, 2, . . .,


n, then    −1    
T C = S B T B S B.

Example:
Let T : R2 → R2 be the linear operator defined by T (x1 , x2 ) = (x1 , 0). Let B = {u1 =
(1, 0), u2 = (0, 1)} be the standard ordered basis for R 2 and let C =
 {v1 = (1, 1), v2 =
2 −1
(2, 1)} be another ordered basis for R . Then verify T C = P T B P.
Answer:    
1 2 −1 −1 2
P = and P = .
1 1 1 −1
 
  1 0
T B= .
0 0
 
  −1 −2
T C = .
1 2

Definition 2.1. Let A and B be n × n matrices. We say that B is similar to A if there is


an invertible n × n matrix P such that B = P −1 AP .

Page No. 21
Lecture Notes by M. G. P. Prasad, IITG

Remark: Let A and B be n × n matrices. The statement B is similar to A means that on


each n-dimensional vector space, the similar matrices A and B represent the same linear
transformation in two (possibly) different ordered bases.

Exercises: In every lecture slides (PDF file) uploaded in the moodle course webpage of
MA102, several exercises are given towrds end (last few slides) and solve all those exer-
cises/ problems.

Coordinates of a Vector in the Ordered Basis:


Let V be an n-dimensional vector space over the field F . Let B = {v1 , . . . , vn } be an
ordered basis for V . For any vector v ∈ V , there exist scalars x1 , . . ., xn in F such that

v = x1 v1 + x2 v2 + · · · + xn vn .

Then  T
x = x1 x2 · · · xn
is called the coordinates of v with respect to the ordered basis B and it is denoted by [v]B .

Properties:

• [u + v]B = [u]B + [v]B for all u and v in V .

• [α v]B = α [v]B for all v ∈ V and for all α ∈ F .

Define a map B : F n → V by

B(x1 , . . . , xn ) = x1 v1 + · · · + xn vn .

That is, B maps [v]B to v.


Then,

• B is a linear transformation.

• B is bijective.

Define the coordinate map S : V → F n by

S(v) = [v]B .

Then,

Page No. 22
Lecture Notes by M. G. P. Prasad, IITG

• S is a linear transformation.

• S is bijective.

• S is the inverse of B.

• S is an isomorphism.

Page No. 23

You might also like