MODULE-II

LINEAR TRANSFORMATION

Introduction
In this chapter we will discuss linear transformations, which are the most natural kind of a
map from one vector space to another, and show how they are intimately related with matrices. In
our discussions we will give concrete examples as often as possible, and use the general properties
we have shown about vector spaces to motivate results relevant to solving systems of linear
equations and solving differential equations.

Topic Learning Objectives:

Upon Completion of this module, students will be able to:

 Transform vector space from one dimension to other using linear transformation.
 Apply rank-Nullity theorem.
 Estimate inner product of vectors, orthogonal and orthonormal basis of a vector space

Linear Transformation/Linear Mapping:

Let U and V be two vector spaces over the same field F. The mapping 𝑇: 𝑈 → 𝑉 is said to be a
linear transformation if
i. 𝑇(𝛼 + 𝛽) = 𝑇(𝛼) + 𝑇(𝛽), ∀𝛼, 𝛽 ∈ 𝑈 ii. 𝑇(𝑐.
𝛼) = 𝑐. 𝑇(𝛼), ∀𝑐 ∈ 𝐹, 𝛼 ∈ 𝑈
Note:
1) Linear transformation is also called as linear mapping or linear operator.
2) In the definition the sign + in 𝛼 + 𝛽 denotes the addition in the vector space U, where
as the sign + in 𝑇(𝛼) + 𝑇(𝛽) denotes the addition in the space V.
Problems:

1. If 𝑓: 𝑉3(𝑅) → 𝑉2(𝑅) is defined by f(x, y, z) = (x + y, y + z), show that f is a linear

transformation.
Solution: Let 𝑎 = (𝒙𝟏, 𝒚𝟏, 𝒛𝟏) & 𝖰 = (𝒙𝟐, 𝒚𝟐, 𝒛𝟐) be any two elements of 𝑽𝟑(𝑹) Consider
𝑓(𝛼 + 𝛽) = 𝑓(𝑥1 + 𝑥2, 𝑦1 + 𝑦2, 𝑧1 + 𝑧2)
= (𝑥1 + 𝑥2 + 𝑦1 + 𝑦2, 𝑦1 + 𝑦2 + 𝑧1 + 𝑧2)
= [(𝑥1 + 𝑦1) + ( 𝑥2 + 𝑦2), (𝑦1 + 𝑧1) + (𝑦2 + 𝑧2)]
 = [(𝑥1 + 𝑦1), (𝑦1 + 𝑧1)] + [(𝑥2 + 𝑦2) + (𝑦2 + 𝑧2)]
= 𝑓(𝑥1, 𝑦1, 𝑧1) + 𝑓(𝑥2, 𝑦2, 𝑧2)
⇒ 𝑓(𝛼 + 𝛽) = 𝑓(𝛼) + 𝑓(𝛽)

Let 𝑐 ∈ 𝑅 be arbitrary.
Consider 𝑓(𝑐. 𝛼) = 𝑓(𝑐. 𝑥1, 𝑐. 𝑦1, 𝑐. 𝑧1) = (𝑐. 𝑥1 + 𝑐. 𝑦1, 𝑐. 𝑦1 + 𝑐. 𝑧1)
= 𝑐(𝑥1 + 𝑦1, 𝑦1 + 𝑧1) = 𝑐. 𝑓(𝑥1, 𝑦1, 𝑧1) = 𝑐. 𝑓(𝛼) ⇒
𝑓(𝑐. 𝛼) = 𝑐. 𝑓(𝛼)
Therefore, f is a linear transformation.

2. If 𝑇: 𝑉3(𝑅) → 𝑉3(𝑅) is defined by 𝑇 (𝑥1, 𝑥2, 𝑥3) = (0, 𝑥2, 𝑥3), show that T is a linear
transformation.
Solution: Let 𝛼 = (𝑥1, 𝑥2, 𝑥3) & 𝛽 = (𝑦1, 𝑦2, y3) be any two elements of 𝑉3(𝑅) Consider
𝑇(𝛼 + 𝛽) = 𝑇((𝑥1, 𝑥2, 𝑥3) + (𝑦1, 𝑦2, 𝑦3))
= 𝑇(𝑥1 + 𝑦1, 𝑥2 + 𝑦2, 𝑥3 + 𝑦3)
= (0, 𝑥2 + 𝑦2, 𝑥3 + 𝑦3)
= (0, 𝑥2, 𝑥3) + (0, 𝑦2, 𝑦3)
= 𝑇(𝑥1, 𝑥2, 𝑥3) + 𝑇(𝑦1, 𝑦2, 𝑦3)
⟹ 𝑇(𝛼 + 𝛽) = 𝑇(𝛼) + 𝑇(𝛽) Let
𝑐 ∈ 𝑅 be arbitrary.
Consider,𝑇(𝑐. 𝛼) = 𝑇[𝑐. (𝑥1, 𝑥2, 𝑥3)] = 𝑇[(𝑐𝑥1, c𝑥2, c𝑥3)] = (0, c𝑥2, c𝑥3) = c.(0, 𝑥2, 𝑥3)
𝑇(𝑐. 𝛼) = 𝑐. 𝑇(0, 𝑥2, 𝑥3) = 𝑐. 𝑇(𝛼)
⇒ 𝑇(𝑐. 𝛼) = 𝑐. 𝑇(𝛼)
Therefore, T is a linear transformation.
3. If T is a mapping from 𝑉2(𝑅) into 𝑉2(𝑅) defined by
𝑇(𝑥, 𝑦) = (𝑥 𝑐𝑜𝑠𝜃 − 𝑦𝑠𝑖𝑛𝜃, 𝑥 𝑠𝑖𝑛𝜃 + 𝑦 𝑐𝑜𝑠𝜃), show that T is a linear transformation.
4. Find the linear transformation 𝑓: 𝑅2 → 𝑅2 such that 𝑓(1, 1) = (0, 1) and
𝑓(−1, 1) = (3, 2).
5. Find the linear transformation 𝑇: 𝑅2 → 𝑅3 such that 𝑇(1, 1) = (0, 1, 2) and 𝑇(−1, 1) = (2,
1, 0).

Ordered basis of V:
Definition: Let V be a vector space over a field F and dim[V] = n > 0. Choose a basis
𝒂𝟏, 𝒂𝟐, 𝒂𝟑, … , 𝒂𝒏 of V written in order. This basis 𝑩 = {𝒂𝟏, 𝒂𝟐, 𝒂𝟑, … , 𝒂𝒏}is called an ordered basis
of V. In an ordered basis, the order in which the elements are arranged is taken in to account.
Example:
[{(1, 0, 0), (0, 1, 0), (0, 0, 1)}, {(0, 1, 0), (0, 0, 1), (1, 0, 0)}, {(0, 0, 1), (1, 0, 0), (0, 1,0)}]
are three different basis of R3.

Note:

1) If 𝐵 = {𝛼1, 𝛼2, 𝛼3, … , 𝛼𝑛} be an ordered basis of V, then for 𝛼 ∈ 𝑉

𝛼= c1𝛼1+ c2𝛼2+ c3𝛼3+…+c𝑛𝛼𝑛, 𝑐𝑖 ∈ F are called coordinates of the vector 𝛼

2) The coordinates depend on the choice of the ordered basis.

Problems:

1. Find the coordinate vector of (3, -2, 1) relative to (i) ordered standard basis

(ii) the ordered basis {(1, 1, 1), (1, 0, 0), (1, 1, 0)} of 𝑅3.

Solution:

(i) Standard basis is (1, 0, 0), (0, 1, 0), (0, 0, 1)

∴ (3, −2, 1) = 𝑐1(1, 0, 0) + 𝑐2(0, 1, 0) + 𝑐3(0, 0, 1) ⇒ 𝑐1 = 3, 𝑐2 = −2, 𝑐3 = 1
⇒ (3, −2, 1) = (𝑐1, 𝑐2, 𝑐3).
Therefore, coordinate vector of (3, -2, 1) relative to standard basis is (3, -2, 1) itself.

(ii) Let (3, −2, 1) = 𝑐1(1, 1, 1) + 𝑐2(1, 0, 0) + 𝑐3(1, 1, 0) = (𝑐1 + 𝑐2 + 𝑐3, 𝑐1 + 𝑐3, 𝑐1)

⇒ 𝑐1 + 𝑐2 + 𝑐3 = 3, 𝑐1 + 𝑐3 = −2, 𝑐1 = 1
Solving we get c1 = 1, c1 = 5, c3 = -3 which are coordinates of (3, -2, 1) with respect to the
ordered basis (1, 1, 1), (1, 0, 0), (1, 1, 0).

Kernel and image of a linear transformations:

To every linear transformation we associate two important subspaces.
Let 𝑇: 𝑉 → 𝑊 be linear, the kernel of T, Written kerT, is {𝑣𝜖𝑉: 𝑇(𝑣) = 0𝑤} and the image of T,
written imT, is {𝑣𝜖𝑉: 𝑇(𝑣)}. In other words, the image is what we normally mean by the image
of a function.
Property of all linear maps
Lemma: Let 𝑇: 𝑉 → 𝑊 be a linear map. Then 𝑇(0𝑣) = 0𝑤.
Proof. 𝑇(0𝑣) = 𝑇(0𝑣 + 0𝑣) = 𝑇(0𝑣) + 𝑇(0𝑣)
By the definition of linearity now add −𝑇(0𝑣) to both sides
𝑇(0𝑣) − 𝑇(0𝑣) = 𝑇(0𝑣) + 𝑇(0𝑣) − 𝑇(0𝑣)
0𝑤 = 𝑇(0𝑣)

Problem 1. Let 𝐴 = [0 1] so that we have a linear map 𝑇𝐴: 𝑅2 − 𝑅2 given by 𝑇𝐴(𝑥) = 𝐴𝑥. We
0 0
will find 𝑖𝑚 𝑇𝐴 and 𝑘𝑒𝑟𝑇𝐴.
𝑥
Solution: 𝑖𝑚 𝑇𝐴 = {𝑇𝐴 (𝑦) : 𝑥, 𝑦 ∈ 𝑅}

0 1 𝑥
= {( 0 0 ) (𝑦) : 𝑥, 𝑦 ∈ 𝑅}

𝑦
= {(0) : 𝑥, 𝑦 ∈ 𝑅}

Another way to write this is that 𝑖𝑚 𝑇𝐴 = 𝑠𝑝𝑎𝑛 and so dim 𝑖𝑚 𝑇𝐴 = 1.

Now we’ll do the kernel

𝑥 0 𝑥 2
𝑘𝑒𝑟 𝑇𝐴 = {𝑇𝐴 (𝑦) = (0 ) : (𝑦) ∈ 𝑅 }

0 1 𝑥
0 𝑥 2

𝑦 0 𝑥 2
𝑅 }

𝑥
= {( ) : 𝑥 ∈ 𝑅}
0

Again, we could write this as 𝑘𝑒𝑟 𝑇𝐴 = 𝑠𝑝𝑎𝑛 .

The kernel and image are equal in this case.

Range and Kernel of a linear transformation

 Definition: Let 𝑇: 𝑈 → 𝑉 be a linear transformation. The range of T is the set, 𝑅(𝑇) = {𝑇(𝛼)/𝛼 ∈
𝑈}, i.e., 𝑅(𝑇) is the set of all T-images of the elements of U. Clearly, 𝑅(𝑇) ⊆ 𝑉.

Definition: Let 𝑇: 𝑈 → 𝑉 be a linear transformation. The Kernel (or null space) of T is the set
𝑁(𝑇) = {𝛼 ∈ 𝑈/𝑇(𝛼) = 0} where 0 is the zero vector of V.
Rank and Nullity:

Definition: Let 𝑇: 𝑈 → 𝑉 be a linear transformation. Then,

1) If 𝑅(𝑇) is finite dimensional, the dimension of 𝑅(𝑇)is called the Rank of linear transformation
and is denoted by 𝑟(𝑇).

2) If 𝑁(𝑇) is finite dimensional, the dimension of 𝑁(𝑇)is called the Nullity of linear
transformation and is denoted by 𝑛(𝑇).

Rank-Nullity theorem

Let 𝑇: 𝑈 → 𝑉 be a linear transformation and U be a finite dimensional vector space. Then 𝑑𝑖𝑚
𝑅 (𝑇) + 𝑑𝑖𝑚 𝑁 (𝑇) = 𝑑𝑖𝑚 𝑈 𝑖. 𝑒. 𝑟(𝑇) + 𝑛(𝑇) = 𝑑𝑖𝑚 𝑈. [or Rank+Nullity=dim(domain)]

The null space of the matrix 𝐴, denoted by 𝑁(𝐴), is the set of all n-dimensional column vectors X
such that 𝐴𝑋 = 0.

Problems:

1. Let 𝑇: 𝑉2(𝑅) → 𝑉2(𝑅) defined by 𝑇(𝑥, 𝑦) = (𝑥 + 𝑦, 𝑥). Find the range space, kernel, rank, and
nullity. Also, verify the rank-nullity theorem.
Solution:
To find 𝑹(𝑻): We shall write the images of the elements w.r.t standard basis
𝑇(𝑒1) = 𝑇(1,0) = (1,1)&𝑇(𝑒2) = 𝑇(0,1) = (1,0)

Therefore, the matrix A of T is (by echelon form)

There are 2 non-zero rows. ∴ Rank of A is 2
Hence, 𝑅(𝑇) is subspace generated by(1,1)&(0, −1).
∴ 𝑅(𝑇) = {𝑥(1,1) + 𝑦(0, −1)} = {(𝑥, 𝑥) + (0, −𝑦)}
𝑅(𝑇) = {𝑥, 𝑥 − 𝑦} = 𝑉2(𝑅) To
find 𝑵(𝑻):
Let𝑇(𝑥, 𝑦) = 0 ⇒ (𝑥 + 𝑦, 𝑥) = 0 ⇒ 𝑥 + 𝑦 = 0, 𝑥 = 0 ⇒ 𝑥 = 0, 𝑦 = 0
∴ 𝑁(𝑇) contains only zero elements of 𝑉2(𝑅)
∴ 𝑁(𝑇) = {(0,0)} ⇒ Nullspace = {(0,0)}
𝑑𝑖𝑚[𝑁(𝑇)] = 0 ⇒ Nullity = 0
∴ Rank+Nullity = 2 + 0 = 2 = dim(𝑉2(𝑅))
Hence rank-nullity theorem is verified.
2. Let 𝑇: 𝑉 → 𝑊 be a linear transformation defined by
𝑇(𝑥, 𝑦, 𝑧) = (𝑥 + 𝑦, 𝑥 − 𝑦, 2𝑥 + 𝑧). Find the range, null space, rank, nullity and hence
verify the rank-nullity theorem Solution:
We shall write the images of the elements of standard basis
𝑇(1, 0, 0) = (1, 1, 2) = 𝑎1
𝑇(0, 1, 0) = (1, −1, 0) = 𝑎2
𝑇(0, 0, 1) = (0, 0, 1) = 𝑎3
Clearly the set {𝑎1, 𝑎2, 𝑎3}generates 𝑅(𝑇).
1 1 2
Consider the matrix 𝐴 = [1 −1 0]
0 0 1

1 1 2 1 1 2 1 1 2
𝐴 = [1 −1 0] ∼ [0 −2 −2] ∼ [0 1 1] (By echelon form)
0 0 1 0 0 1 0 0 1

∴ dim 𝑅(𝑇) = 3 ⇒ 𝑟𝑎𝑛𝑘 = 3 . 𝑅(𝑇) is subspace generated by (1,1,2), (0,1,1)&(0,0,1).

∴ 𝑅(𝑇) = {𝑥(1,1,2) + 𝑦(0,1,1) + 𝑧(0,0,1)} = {(𝑥, 𝑥, 2𝑥) + (0, 𝑦, 𝑦) + (0,0, 𝑧)}
𝑅(𝑇) = {𝑥, 𝑥 + 𝑦, 2𝑥 + 𝑦 + 𝑧/𝑥, 𝑦, 𝑧 ∈ 𝑅} = 𝑅3 To
find 𝑁(𝑇):
Consider 𝑇(𝑥, 𝑦, 𝑧) = (0, 0, 0)
(𝑥 ,𝑥 + 𝑦, 2𝑥 +𝑦 + 𝑧) = (0, 0, 0)
⇒ 𝑥 = 0, 𝑥 + 𝑦 = 0, 2𝑥 + 𝑦 + 𝑧 = 0
⇒ 𝑥 = 0, 𝑦 = 0, 𝑧 = 0
𝑁(𝑇) contains only zero vector
𝑁(𝑇) = (0, 0, 0) ⇒ 𝑑𝑖𝑚 𝑁 (𝑇) = 0 ⇒ Nullity = 0
∴ Rank+Nullity = 3 + 0 = 3 = 𝑑𝑖𝑚(𝑅3) Hence
rank-nullity theorem is verified.
3. Let 𝑇: 𝑅3 → 𝑅3 be defined by 𝑇(1, 0, 0) = (1, 1, 2), 𝑇(0, 1, 0) = (1, −1, 0), 𝑇(0, 0, 1) =
(0, 0, 1). Find the range space, null space, rank and nullity of 𝑇 and verify the rank-nullity theorem.
4. Let 𝑇: 𝑅3 → 𝑅3 be defined by 𝑇(𝑥, 𝑦, 𝑧) = (𝑥 + 𝑦, 𝑥 + 𝑧, 𝑦 + 𝑧). Find the range space, null
space, rank and nullity of 𝑇 and hence verify the rank-nullity theorem.
Matrix of the linear transformation:
Let V and W be vector spaces of dimension m and n respectively over the same field F. Let 𝐵1 =
{𝑎1, 𝑎2, 𝑎3, … , 𝑎𝑚} & 𝐵2 = {𝑏1, 𝑏2, 𝑏3, … , 𝑏𝑛}be the ordered basis of V and W respectively. Let
𝑇: 𝑉 → 𝑊 be a linear transformation defined as
𝑇(𝑎1) = 𝑎11𝑏1 + 𝑎21𝑏2 + 𝑎31𝑏3 + … … . + 𝑎𝑛1𝑏𝑛
𝑇(𝑎2) = 𝑎12𝑏1 + 𝑎22𝑏2 + 𝑎32𝑏3 + … … . + 𝑎𝑛2𝑏𝑛 … …
…………………………………….
𝑇(𝑎𝑚) = 𝑎1𝑚𝑏1 + 𝑎2𝑚𝑏2 + 𝑎3𝑚𝑏3 + … … . + 𝑎𝑛𝑚𝑏𝑛
The transpose of the coefficient matrix is called the “matrix of the linear transformation” relative to
the basis 𝐵1 & 𝐵2. This matrix is denoted by 𝐴𝑇 and is given by
𝑎11 𝑎12 . 𝑎1𝑚
𝑎21 𝐴𝑇 = [ . 𝑎22 . . 𝑎2𝑚
. .]
𝑎𝑛1 𝑎𝑛2 . 𝑎𝑛𝑚
Problems:
1. Find the matrix of the linear transformation 𝑇: 𝑉2(𝑅) → 𝑉3(𝑅) defined by 𝑇(𝑥, 𝑦) = (𝑥 + 𝑦, 𝑥, 3𝑥
− 𝑦) with respect to standard bases.
Solution: By data 𝑇(𝑥, 𝑦) = (𝑥 + 𝑦, 𝑥, 3𝑥 − 𝑦)
B1 = {(1, 0), (0, 1)} & B2 = {(1, 0, 0), (0, 1, 0), (0, 0, 1)} Now
T (1, 0) = (1, 1, 3) & T (0, 1) = (1, 0, -1)
𝑇 (1, 0) = (1, 1, 3) = 𝑐1(1, 0, 0) + 𝑐2(0, 1, 0) + 𝑐3(0, 0, 1)
⇒ (1, 1, 3) = (𝑐1, 𝑐2, 𝑐3) ⇒ 𝑐1 = 1, 𝑐2 = 1, 𝑐3 = 3
𝑇 (0, 1) = (1, 0, −1) = 𝑐1(1, 0, 0) + 𝑐2(0, 1, 0) + 𝑐3(0, 0, 1)
⇒ (1, 0, −1) = (𝑐1, 𝑐2, 𝑐3) ⇒ 𝑐1 = 1, 𝑐2 = 0, 𝑐3 = −1
11
∴ 𝐴𝑇 = [1 0 ] is matrix of the linear transformation.
3 −1
2. Find the matrix of the linear transformation 𝑇: 𝑉3(𝑅) → 𝑉2(𝑅) defined by 𝑇(𝑥, 𝑦, 𝑧) =
(𝑥 + 𝑦, 𝑦 + 𝑧) with respect to 𝐵1 = {(1, 1, 1), (1, 0, 0), (1, 1, 0)}, 𝐵2 = {(1, 0), (0, 1)}.
Solution: By data 𝑻(𝒙, 𝒚, 𝒛) = (𝒙 + 𝒚, 𝒚 + 𝒛)
𝐵1 = {(1, 1, 1), (1, 0, 0), (1, 1, 0)}, 𝐵2 = {(1, 0), (0, 1)}.

𝑇(1, 1, 1) = (2,2), 𝑇(1, 0, 0) = (1,0) , 𝑇(1, 1, 0) = (2,1)

𝑇(1, 1, 1) = (2, 2) = 𝑐1(1, 0) + 𝑐2(0, 1) = (𝑐1, 𝑐2) ⇒ 𝑐1 = 2, 𝑐2 = 2

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

 𝑇(1, 1, 0) = (2, 1) = 𝑐1(1, 0) + 𝑐2(0, 1) = (𝑐1, 𝑐2) ⇒ 𝑐1 = 2, 𝑐2 = 1
2 1 2
∴ 𝐴 𝑇= [ ] is matrix of the linear transformation.
2 0 1
1 2
3. Given the matrix 𝐴 = [ 0 1], find the linear transformation 𝑇: 𝑉2(𝑅) → 𝑉3(𝑅) relative to
−1 3
the basis 𝐵1 = {(1, 1), (−1, 1)}, 𝐵2 = {(1, 1, 1), (1, −1, 1), (0, 0, 1)}.
Solution: Define 𝑇: 𝑉2(𝑅) → 𝑉3(𝑅) by
𝑇(1, 1) = 1(1, 1, 1) + 0(1, −1, 1) + (−1)(0, 0, 1) = (1, 1, 0)
𝑇(−1, 1) = 2(1, 1, 1) + 1(1, −1, 1) + 3(0, 0, 1) = (3, 1, 6)
Let (𝑥, 𝑦) ∈ 𝑉2(𝑅)be arbitrary, then
(𝑥, 𝑦) = 𝑐1(1, 1) + 𝑐2(−1, 1) = (𝑐1 − 𝑐2, 𝑐1+ 𝑐2) ⇒ 𝑐1 − 𝑐2= 𝑥 & 𝑐1+ 𝑐2 = 𝑦

Solving we get 𝑐 &𝑐

Consider,
Applying linear transformation on both sides
𝑇(𝑥, 𝑦) = 𝑥+𝑦 𝑇(1, 1) + 𝑦−𝑥 𝑇(−1, 1)
2 2

= 𝑥+𝑦 (1,1,0) + 𝑦−𝑥 (3,1,6) = [𝑥+𝑦 + 3𝑦−3𝑥 , 𝑥+𝑦 + 𝑦−𝑥 , 6𝑦−6𝑥]

2 2 2 2 2 2 2

⇒ 𝑇(𝑥, 𝑦) = (−𝑥 + 2𝑦, 𝑦, −3𝑥 + 3𝑦) is the required transformation relative to the basis 𝐵1 & 𝐵2.
4. Find the matrix of the linear transformation 𝑇: 𝑉3(𝑅) → 𝑉2(𝑅) defined by 𝑇(𝑥, 𝑦, 𝑧) = (𝑥 + 𝑦,
𝑦 + 𝑧) with respect to standard bases of 𝑉3(𝑅) and 𝑉2(𝑅).
5. Find the matrix of the linear transformation 𝑇: 𝑉2(𝑅) → 𝑉3(𝑅)
defined by 𝑇(𝑥, 𝑦) = (−𝑥 + 2𝑦, 𝑦,-3x+3𝑦) relative to bases 𝐵1 = {(1, 1), (−1, 1)} and 𝐵2 = {(1, 1,
1), (1, −1, 1), (0, 0, 1)}.

Change of Basis

Definition: Let 𝑇: 𝑉 → 𝑉 be linear transformation defined over Field F,

Let 𝐵1 = {𝑎1, 𝑎2, 𝑎3, … , 𝑎𝑚} be basis of V and 𝐵2 = {𝑏1, 𝑏2, 𝑏3, … , 𝑏𝑛} be another basis of V.
Then we obtain change of basis from old basis 𝐵1 to new basis 𝐵2 by expressing 𝑏𝑖 as a linear
combination of 𝑎1, 𝑎2, 𝑎3, … , 𝑎𝑚
𝑏1 = 𝑐11𝑎1 + 𝑐21𝑎2 + 𝑐31𝑎3 + … … . + 𝑐𝑛1𝑎𝑛 𝑏2
= 𝑐12𝑎1 + 𝑐22𝑎2 + 𝑐32𝑎3 + … … . + 𝑐𝑛2𝑎𝑛 … …
…………………………………….
𝑇(𝑎𝑚) = 𝑐1𝑚𝑎1 + 𝑐2𝑚𝑎2 + 𝑐3𝑚𝑎3 + … … . + 𝑐𝑛𝑚𝑎𝑛
The transpose of the coefficient of above system of equation gives matrix of change of basis from
𝐵1𝑡𝑜 𝐵2. This matrix is given by
𝑐11 𝑐12 . 𝑐1𝑚
𝑃 = [𝑐21. 𝑐22. .. 𝑐2. 𝑚]
𝑐𝑛1 𝑐𝑛2 . 𝑐𝑛𝑚
Problems:
1. Consider the following basis of 𝑅2, 𝑆 = {𝑢1, 𝑢2} = {(1,2), (3,5)}, 𝑆′ = {𝑣1, 𝑣2} = {(1, −1),
(1, −2)}. Find the change of basis matrix P from 𝑆 to 𝑆′. Also find the change of basis
matrix Q from 𝑆′ to 𝑆.
Solution: i) Let 𝑆 = {𝑢1, 𝑢2} = {(1,2), (3,5)}. Express 𝑆′ vectors {𝑣1, 𝑣2} as a linear combination
of vectors {𝑢1, 𝑢2} of 𝑆.
𝑣1 = 𝑐1𝑢1 + 𝑐2𝑢2 ⇒ (1, −1) = 𝑐1(1, 2) + 𝑐2 (3, 5)
1 = 𝑐1 + 3𝑐2 ⇒ 𝑐1 = −8, 𝑐2 = 3
−1 = 2𝑐1 + 5𝑐2
𝑣2 = 𝑐3𝑢1 + 𝑐4𝑢2 ⇒ (1, −2) = 𝑐3(1, 2) + 𝑐4 (3, 5)
1 = 𝑐3 + 3𝑐4 ⇒ 𝑐3 = −11, 𝑐4 = 4
−2 = 2𝑐3 + 5𝑐4
Change of basis from 𝑆 to 𝑆′ is −8 − 11
𝑃=[ ]3 4
ii) To obtain change of basis from 𝑆′𝑡𝑜 𝑆, Express 𝑆 vectors {𝑢1, 𝑢2} as a linear combination of 𝑆′
vectors {𝑣1, 𝑣2}.
𝑢1 = 𝑐1𝑣1 + 𝑐2𝑣2 ⇒ (1, 2) = 𝑐1(1, −1) + 𝑐2 (1, −2)
1 = 𝑐1 + 𝑐2 ⇒ 𝑐1 = 4, 𝑐2 = −3
2 = −𝑐1 − 2𝑐2
𝑢2 = 𝑐3𝑣1 + 𝑐4𝑣2 ⇒ (3, 5) = 𝑐3(1, −1) + 𝑐4 (1, −2)
3 = 𝑐1 + 𝑐2 ⇒ 𝑐1 = 11, 𝑐2 = −8
5 = −𝑐1 − 2𝑐2
Change of basis from 𝑆 to 𝑆′ is 𝑄 = [ 4 11 ]
−3 −8
2. Consider the following basis of 𝑅3, 𝐸 = {𝑒1, 𝑒2, 𝑒3} and 𝑆 = {𝑢1, 𝑢2, 𝑢3} =
{(1,0, 1), (2, 1, 2), (1, 2, 2)}. Find the change of basis matrix P from E to S. Also find the
change of basis matrix Q from S to E.
Solution: i) Let 𝐸 = {𝑒1, 𝑒2, 𝑒3} = {(1,0,0), (0,1,0), (0,0,1)}. To find change of basis from E
to S, express S vectors {𝑢1, 𝑢2, 𝑢3} as a linear combination of vectors {𝑒1, 𝑒2, 𝑒3} of E.
𝑢1 = 𝑐1𝑒1 + 𝑐2𝑒2 + 𝑐3𝑒3 ⇒ (1, 0,1) = 𝑐1(1, 0,0) + 𝑐2 (0, 1,0) + 𝑐3(0, 0, 1) ⇒
𝑐1 = 1, 𝑐2 = 0, 𝑐3 = 1
𝑢2 = 𝑐4𝑒1 + 𝑐5𝑒2 + 𝑐6𝑒3 ⇒ (2, 1,2) = 𝑐3(1, 0,0) + 𝑐4 (0, 1,0) + 𝑐5(0, 0, 1) ⇒
𝑐4 = 2, 𝑐5 = 1, 𝑐6 = 2
𝑢3 = 𝑐7𝑒1 + 𝑐8𝑒2 + 𝑐9𝑒3 ⇒ (1, 2,2) = 𝑐1(1, 0,0) + 𝑐2 (0, 1,0) + 𝑐3(0, 0, 1)
⇒ 𝑐7 = 1, 𝑐8 = 2, 𝑐9 = 2

1 2 1
Change of basis from E to S is 𝑃 = [0 1 2]
1 2 2
ii) To obtain change of basis from S to E, Express E vectors {𝑒1, 𝑒2, 𝑒3} as a linear combination
of S vectors {𝑢1, 𝑢2, 𝑢3}.
𝑒1 = 𝑐1𝑢1 + 𝑐2𝑢2 + 𝑐3𝑢3 ⇒ (1, 0, 0) = 𝑐1(1, 0,1) + 𝑐2 (2,1,2) + 𝑐3(1,2,2) 1
= 𝑐1 + 2𝑐2 + 𝑐3
0 = 0𝑐1 + 2𝑐2 + 2𝑐3
0 = 𝑐1 + 2𝑐2 + 2𝑐3 ⇒ 𝑐1 = −2, 𝑐2 = 2, 𝑐3 = −1
𝑒2 = 𝑐4𝑢1 + 𝑐5𝑢2 + 𝑐6𝑢3 ⇒ (0, 1, 0) = 𝑐4(1, 0,1) + 𝑐5 (2,1,2) + 𝑐6(1,2,2) 0 =
𝑐1 + 2𝑐2 + 𝑐3
1 = 0𝑐1 + 2𝑐2 + 2𝑐3
0 = 𝑐1 + 2𝑐2 + 2𝑐3 ⇒ 𝑐4 = −2, 𝑐5 = 1, 𝑐6 = 0
𝑒3 = 𝑐7𝑢1 + 𝑐8𝑢2 + 𝑐9𝑢3 ⇒ (0, 0, 1) = 𝑐7(1, 0,1) + 𝑐8 (2,1,2) + 𝑐9(1,2,2) 0 =
𝑐1 + 2𝑐2 + 𝑐3
0 = 0𝑐1 + 2𝑐2 + 2𝑐3
1 = 𝑐1 + 2𝑐2 + 2𝑐3 ⇒ 𝑐1 = 3, 𝑐2 = −2, 𝑐3 = 1

−2 −2 3
Change of basis from S to E is 𝑄 = [ 2 1 −2]
−1 0 1
 3. Consider the following basis of 𝑅2, 𝐸 = {𝑒1, 𝑒2} = {(1,0), (0,1)}, 𝑆 = {𝑣1, 𝑣2} = {(1,3), (1,
4)}. Find the change of basis matrix P from E to S and change of basis matrix Q from S to
E.
4. Consider the linear transformation 𝑇 on 𝑅2 defined by 𝑇(𝑥, 𝑦) = (5𝑥 − 𝑦, 2𝑥 + 𝑦) and the
following basis of 𝑅2 𝐸 = {𝑒1, 𝑒2} = {(1,0), (0,1)}, 𝑆 = {𝑣1, 𝑣2} = {(1, 4), (2, 7)}.
Find the change of basis matrix P from E to S and change of basis matrix Q from S to E.

Singular and Non-singular linear transformations:

The n × n matrix F is called non-singular if 𝑑𝑒𝑡(𝐹) ≠ 0, in other words, if 𝐾𝑒𝑟 𝐹 = 0. and singular
if 𝑑𝑒𝑡(𝐹) = 0. So a non-singular matrix is invertible, and a singular matrix is not. We call the linear
transformation 𝐹 ∶ 𝑅𝑛 → 𝑅𝑛 non-singular if F is a non-singular matrix, and we call it singular if F
is a singular matrix.

Problems
1. Determine whether or not each of the following linear maps is nonsingular. If not, find a nonzero
vector v whose image is 0. (i) 𝐹: 𝑅2 → 𝑅2 defined by 𝐹(𝑥, 𝑦) = (𝑥 − 𝑦, 𝑥 − 2𝑦). (ii) 𝐺: 𝑅2 → 𝑅2
defined by 𝐺(𝑥, 𝑦) = (2𝑥 − 4𝑦, 3𝑥 − 6𝑦).
Solution: (i) Find Ker F by setting 𝐹(𝑣) = 0, 𝑤ℎ𝑒𝑟𝑒 𝑣 = (𝑥, 𝑦) (𝑥
− 𝑦, 𝑥 − 2𝑦) = (0,0).
𝑥 − 𝑦 = 0, 𝑥 − 2𝑦 = 0
The only solution is 𝑥 = 0, 𝑦 = 0. Hence F is nonsingular
(ii) Find Ker G by setting 𝐺(𝑣) = 0, 𝑤ℎ𝑒𝑟𝑒 𝑣 = (𝑥, 𝑦) (2𝑥
− 4𝑦, 3𝑥 − 6𝑦) = (0,0).
2𝑥 − 4𝑦 = 0,3𝑥 − 6𝑦 = 0
The system has nonzero solutions, because y is a free variable. Hence, G is singular. Let 𝑦 = 1 to
obtain the solution𝑉 = (2,1); 1Þ, which is a nonzero vector, such that 𝐺(𝑣) = 0.
2. The linear map 𝐹: 𝑅2 → 𝑅2 defined by 𝐹(𝑥, 𝑦) = (𝑥 − 𝑦, 𝑥 − 2𝑦) is nonsingular. Find a
formula for 𝐹−1.
Solution: set 𝐹(𝑥, 𝑦) = (𝑎, 𝑏), so that 𝐹−1(𝑎, 𝑏) = (𝑥, 𝑦)
We have (𝑥 − 𝑦, 𝑥 − 2𝑦) = (𝑎, 𝑏)
Solve for x and y in terms of a and b to get 𝑥 = 2𝑎 − 𝑑, 𝑦 = 𝑎 − 𝑏. Thus
 𝐹−1(𝑎, 𝑏) = (2𝑎 − 𝑏, 𝑎 − 𝑏)

3. Let 𝐹: 𝑅2 → 𝑅3 defined by 𝐹(𝑥, 𝑦) = (𝑥 + 𝑦, 𝑥 − 2𝑦, 3𝑥 + 𝑦). (i) Show that F is nonsingular

(ii) find a formula for 𝐹−1.

Solution: (i) Find Ker F by setting 𝐹(𝑣) = 0, 𝑤ℎ𝑒𝑟𝑒 𝑣 = (𝑥, 𝑦) (𝑥

+ 𝑦, 𝑥 − 2𝑦, 3𝑥 + 𝑦) = (0,0,0).
𝑥 + 𝑦 = 0, 𝑥 − 2𝑦 = 0, 3𝑥 + 𝑦 = 0
The only solution is 𝑥 = 0, 𝑦 = 0. Hence F is nonsingular
(ii) Although G is nonsingular, it is not invertible, because 𝑅2and 𝑅3 have different dimensions.
Accordingly, 𝐹−1 does not exist.

Invertible linear transformations:

Let 𝐹 𝑉 → 𝑉 be a linear operator. F is said to be invertible if it has an inverse—that is, if there

exists 𝐹−1 in A(V) such that 𝐹𝐹−1 = 𝐹−1𝐹 = 𝐼. On the other hand, F is invertible as a mapping if F
is both one-to-one and onto. In such a case, 𝐹−1 is also linear and 𝐹−1is the inverse of F as a linear
operator.
Note: Let F be a linear operator on a finite-dimensional vector space V. Then the following four
conditions are equivalent.
(i) F is nonsingular: Ker F=0. (iii) F is an onto mapping.
(ii) F is one-to-one. (iv) F is invertible.

Problems:
1. Let 𝐹 be the linear operator on 𝑅2 defined by 𝑓(𝑥, 𝑦) = (2𝑥 + 𝑦, 3𝑥 + 2𝑦). (a) Show that F is
nonsingular. (b) Find a formula for 𝐹−1.
Solution:
a) to show that 𝐹 is invertible, we need only show that F is nonsingular. Set 𝑓(𝑥, 𝑦) = (0,0) to
obtain the homogeneous system
2𝑥 + 𝑦 = 0; 3𝑥 + 2𝑦 = 0
Solve for x and y to get 𝑥 = 0; 𝑦 = 0. Hence, 𝐹 is non singular and so invertible.
b) To find a formula for 𝐹−1 we set 𝐹(𝑥, 𝑦) = (𝑠, 𝑡) and so 𝐹−1(𝑠, 𝑡) = (𝑥, 𝑦)
 2𝑥 + 𝑦 = 𝑠; 3𝑥 + 2𝑦 = 𝑡
Solve for x and y in terms of s and t to obtain 𝑥 = 2𝑠 − 𝑡; 𝑦 = −3𝑠 + 2𝑡. Thus
𝐹−1(𝑠, 𝑡) = (2𝑠 − 𝑡, −3𝑠 + 2𝑡)
where we rewrite the formula for 𝐹−1 using x and y instead of s and t.
2. Let 𝐺: 𝑅2 → 𝑅3be defined by 𝐺(𝑥, 𝑦) = (𝑥 + 𝑦, 𝑥 − 2𝑦, 3𝑥 + 𝑦), (a) Show that G is nonsingular.
(b) Find a formula for 𝐺−1.
Solution:
(a) Set 𝐺(𝑥, 𝑦) = (0,0,0) to find Ker G. We have
𝑥 + 𝑦 = 0, 𝑥 − 2𝑦 = 0, 3𝑥 + 𝑦 = 0
The only solution is 𝑥 = 0, 𝑦 = 0; hence, G is nonsingular.
(b) Although G is nonsingular, it is not invertible, because 𝑅2 and 𝑅3 have different dimensions.
(Thus, Theorem 5.9 does not apply.) Accordingly, 𝐺−1 does not exist.
3. Consider the linear operator T on 𝑅3 defined by 𝑇(𝑥, 𝑦) = (2𝑥, 4𝑥 − 𝑦, 2𝑥 + 3𝑦 − 𝑧).
(a) Show that T is invertible (b) Find formulas for 𝑇−1.
Solution:
(a) Set 𝑇(𝑥, 𝑦) = (0,0,0) to find Ker G. We have
2𝑥 = 0, 4𝑥 − 𝑦 = 0, 2𝑥 + 3𝑦 − 𝑧 = 0
The only solution is 𝑥 = 0, 𝑦 = 0, z = 0; Hence, T is non singular and so invertible.
b) To find a formula for 𝑇−1 we set 𝑇(𝑥, 𝑦) = (𝑟, 𝑠, 𝑡) and so 𝑇−1(𝑟, 𝑠, 𝑡) = (𝑥, 𝑦, 𝑧)
2𝑥 = 𝑟; 4𝑥 − 𝑦 = 𝑠; 2𝑥 + 3𝑦 − 𝑧 = 𝑡 Solve for x and y in
terms of s and t to obtain 𝑥 = 𝑟 ; 𝑦 = 2𝑟 − 𝑠; 𝑧 = 7𝑟 − 3𝑠 − 𝑡 . Thus
2
𝑟
𝑇−1( 𝑟, 𝑠, 𝑡) = ( , 2𝑟 − 𝑠, 7𝑟 − 3𝑠 − 𝑡) 2
where we rewrite the formula for 𝑇−1 using x and y instead of s and t.

Video Links:

1. Linear transformation
https://www.youtube.com/watch?v=is1cg5yhdds
2. Rank-Nullity theorem https://www.youtube.com/watch?v=-
LmAFzcKYPA