Linear Algebra - MA2034

4 Linear Transformations

Definition 4.1: If 𝑇 ∶ 𝑉 → 𝑊 is a mapping from a vector space 𝑉 to a vector space 𝑊, then 𝑇 is

called a linear transformation from 𝑉 to 𝑊 if the following two properties hold for all vectors 𝒖
and 𝒗 in 𝑉 and for all scalars 𝑘:

(i) 𝑇(𝑘𝒖) = 𝑘𝑇(𝒖) [Homogeneity property]

(ii) 𝑇(𝒖 + 𝒗) = 𝑇(𝒖) + 𝑇(𝒗) [Additivity property] .

In the special case where 𝑉 = 𝑊, the linear transformation 𝑇 is called a linear operator on the
vector space 𝑉.

The homogeneity and additivity properties of a linear transformation 𝑇 ∶ 𝑉 → 𝑊 can be used in

combination to show that if 𝒗𝟏 and 𝒗𝟐 are vectors in 𝑉 and 𝑘1 and 𝑘2 are any scalars, then

𝑇(𝑘1 𝒗1 + 𝑘2 𝒗2 ) = 𝑘1 𝑇(𝒗1 ) + 𝑘2 𝑇(𝒗2 ).

More generally, if 𝒗1 , 𝒗2 , . . . , 𝒗𝑟 are vectors in 𝑉 and 𝑘1 , 𝑘2 , . . . , 𝑘𝑟 are any scalars, then

𝑇(𝑘1 𝒗1 + 𝑘2 𝒗2 +··· +𝑘𝑟 𝒗𝑟 ) = 𝑘1 𝑇(𝒗1 ) + 𝑘2 𝑇(𝒗2 ) +··· +𝑘𝑟 𝑇(𝒗𝑟 ).

Theorem 4.1: Let 𝑇 ∶ 𝑉 → 𝑊 be a linear transformation; where 𝑉 and 𝑊 are vector spaces.
Then:
(a) 𝑇(𝟎) = 𝟎.
(b) 𝑇(𝒖 − 𝒗) = 𝑇(𝒖) − 𝑇(𝒗) for all 𝒖 and 𝒗 in 𝑉.

Proof.
Let 𝒖 be any vector in 𝑉. Since 0𝒖 = 𝟎, it follows from the homogeneity property in Definition 1
that

𝑇(𝟎) = 𝑇(0𝒖) = 0𝑇(𝒖) = 𝟎

which proves (a).

We can prove part (b) by rewriting 𝑇(𝑢 − 𝑣) as

𝑇(𝒖 − 𝒗) = 𝑇(𝒖 + (−1)𝒗) = 𝑇(𝒖) + (−1)𝑇(𝒗) = 𝑇(𝒖) − 𝑇(𝒗).

Example 1: If 𝐴 is an 𝑚 × 𝑛 matrix and 𝑇: ℝ𝑛 ⟶ ℝ𝑚 is defined by 𝑇(𝑋) = 𝐴𝑋 for every 𝑋 ∈ ℝ𝑛 ,

then 𝑇 is a linear transformation from ℝ𝑛 into ℝ𝑚 .

1
Dr. Supem Samarasiri, Department of Mathematics, University of Moratuwa
 Linear Algebra - MA2034

Since 𝑇(𝑋 + 𝑌) = 𝐴(𝑋 + 𝑌) = 𝐴𝑋 + 𝐴𝑌 and 𝑇(𝑎𝑋) = 𝐴(𝑎𝑋) = 𝑎𝐴𝑋.

Example 2: The Zero Transformation

Let 𝑉 and 𝑊 be any two vector spaces. The mapping 𝑇: 𝑉 → 𝑊 such that 𝑇(𝒗) = 𝟎 for every 𝒗
in 𝑉 is a linear transformation called the zero transformation. To see that 𝑇 is linear, observe that

𝑇(𝒖 + 𝒗) = 𝟎, 𝑇(𝒖) = 𝟎, 𝑇(𝒗) = 𝟎, and 𝑇(𝑘𝒖) = 𝟎 ; ∀ 𝒖, 𝒗 ∈ 𝑽 .

Therefore,
𝑇(𝒖 + 𝒗) = 𝑇(𝒖) + 𝑇(𝒗) and 𝑇(𝑘𝒖) = 𝑘𝑇(𝒖) ; ∀ 𝒖, 𝒗 ∈ 𝑽

Example 3: The Identity Operator

Let 𝑉 be any vector space. The mapping 𝐼: 𝑉 → 𝑉 defined by 𝐼(𝑣) = 𝑣 is called the identity
operator on 𝑉. We will leave it for you to verify that 𝐼 is linear.

Example 4: The Evaluation Transformation

Let 𝑉 be a subspace of 𝐹(−∞, ∞), let
𝑥1 , 𝑥2 , . . . , 𝑥𝑛

be a sequence of distinct real numbers, and let 𝑇 ∶ 𝑉 → ℝ𝑛 be the transformation

𝑇(𝑓) = (𝑓(𝑥1 ), 𝑓(𝑥2 ), . . . , 𝑓(𝑥𝑛 )) ⟶ (1)

that associates with 𝑓 the 𝑛-tuple of function values at 𝑥1 , 𝑥2 , . . . , 𝑥𝑛 . We call this the evaluation
transformation on 𝑉 at 𝑥1 , 𝑥2 , . . . , 𝑥𝑛 . Thus, for example, if

𝑥1 = − 1, 𝑥2 = 2, 𝑥3 = 4

and if 𝑓(𝑥) = 𝑥 2 − 1, then

𝑇(𝑓) = (𝑓(𝑥1 ), 𝑓(𝑥2 ), 𝑓(𝑥3 )) = (0,3,15)

The evaluation transformation in (1) is linear, for if 𝑘 is any scalar, and if 𝑓 and 𝑔 are any functions
in 𝑉, then
𝑇(𝑘𝑓) = ((𝑘𝑓)(𝑥1 ), (𝑘𝑓)(𝑥2 ), … , (𝑘𝑓)(𝑥𝑛 ))

= (𝑘𝑓(𝑥1 ), 𝑘𝑓(𝑥2 ), … , 𝑘𝑓(𝑥𝑛 ))

= 𝑘(𝑓(𝑥1 ), 𝑓(𝑥2 ), … , 𝑓(𝑥𝑛 )) = 𝑘𝑇(𝑓)

and

2
Dr. Supem Samarasiri, Department of Mathematics, University of Moratuwa
 Linear Algebra - MA2034

𝑇(𝑓 + 𝑔) = ((𝑓 + 𝑔)(𝑥1 ), (𝑓 + 𝑔)(𝑥2 ), . . . , (𝑓 + 𝑔)(𝑥𝑛 )) = (𝑓(𝑥1 ) + 𝑔(𝑥1 ), 𝑓(𝑥2 ) +

𝑔(𝑥2), . . . , 𝑓(𝑥𝑛 ) + 𝑔(𝑥𝑛 )) = (𝑓(𝑥1 ), 𝑓(𝑥2 ), . . . , 𝑓(𝑥𝑛 )) + (𝑔(𝑥1 ), 𝑔(𝑥2 ), . . . , 𝑔(𝑥𝑛 )) = 𝑇(𝑓) +
𝑇(𝑔).

Theorem 4.2 Let 𝑇 ∶ 𝑉 → 𝑊 be a linear transformation, where 𝑉 is finite-dimensional.

If 𝑆 = {𝒗𝟏 , 𝒗𝟐 , … , 𝒗𝒏 } is a basis for 𝑉, then the image of any vector 𝒗 in 𝑉 can be expressed as

𝑇(𝒗) = 𝑐1 𝑇(𝒗𝟏 ) + 𝑐2 𝑇(𝒗𝟐 ) +··· +𝑐𝑛 𝑇(𝒗𝒏 )

where 𝑐1 , 𝑐2 , . . . , 𝑐𝑛 are the coefficients required to express 𝒗 as a linear combination of the

vectors in the basis 𝑆.

Proof. Express 𝒗 as 𝒗 = 𝑐1 𝒗𝟏 + 𝑐2 𝒗𝟐 +··· +𝑐𝑛 𝒗𝒏 and use the linearity of 𝑇.

Example 5:
Computing with Images of Basis Vectors
Consider the basis 𝑆 = {𝑣1 , 𝑣2 , 𝑣3 } for ℝ3 , where

𝑣1 = (1,1,1), 𝑣2 = (1,1,0), 𝑣3 = (1,0,0)

Let 𝑇 ∶ ℝ3 → ℝ2 be the linear transformation for which

𝑇(𝑣1 ) = (1,0), 𝑇(𝑣2 ) = (2, −1), 𝑇(𝑣3 ) = (4,3)

Find a formula for 𝑇(𝑥1 , 𝑥2 , 𝑥3 ), and then use that formula to compute 𝑇(2, −3,5).

Solution.
We first need to express 𝑥 = (𝑥1 , 𝑥2 , 𝑥3 ) as a linear combination of 𝑣1 , 𝑣2 , and 𝑣3 . If we write

(𝑥1 , 𝑥2 , 𝑥3 ) = 𝑐1 (1,1,1) + 𝑐2 (1,1,0) + 𝑐3 (1,0,0)

then on equating corresponding components, we obtain

𝑐1 + 𝑐2 + 𝑐3 = 𝑥1

𝑐1 + 𝑐2 = 𝑥2

𝑐1 = 𝑥3

which yields 𝑐1 = 𝑥3 , 𝑐2 = 𝑥2 − 𝑥3 , 𝑐3 = 𝑥1 − 𝑥2 , so

(𝑥1 , 𝑥2 , 𝑥3 ) = 𝑥3 (1,1,1) + (𝑥2 − 𝑥3 )(1,1,0) + (𝑥1 − 𝑥2 )(1,0,0)

= 𝑥3 𝑣1 + (𝑥2 − 𝑥3 )𝑣2 + (𝑥1 − 𝑥2 )𝑣3

Thus
𝑇(𝑥1 , 𝑥2 , 𝑥3 ) = 𝑥3 𝑇(𝑣1 ) + (𝑥2 − 𝑥3 )𝑇(𝑣2 ) + (𝑥1 − 𝑥2 )𝑇(𝑣3 )

3
Dr. Supem Samarasiri, Department of Mathematics, University of Moratuwa
 Linear Algebra - MA2034

= 𝑥3 (1,0) + (𝑥2 − 𝑥3 )(2, −1) + (𝑥1 − 𝑥2 )(4,3)

= (4𝑥1 − 2𝑥2 − 𝑥3 , 3𝑥1 − 4𝑥2 + 𝑥3 )

From this formula we obtain 𝑇(2, −3,5) = (9,23) .

Theorem 4.3: Let 𝑈, 𝑉, 𝑊 be vector spaces. If 𝑇1 : 𝑈 → 𝑉 and 𝑇2 : 𝑉 → 𝑊 are linear

transformations, then (𝑇2 ∘ 𝑇1 ): 𝑈 → 𝑊 is also a linear transformation.
Proof: Will be done in class.

Definition 4.2: Let 𝑉 and 𝑊 be vector spaces,

Let 𝑇 ∶ 𝑉 → 𝑊 be a linear transformation, then the set of vectors in 𝑉 that 𝑇 maps into 0 ∈ 𝑊
is called the kernel of 𝑇 and is denoted by ker(𝑇). The set of all vectors in 𝑊 that are images
under 𝑇 of at least one vector in 𝑉 is called the range of 𝑇 and is denoted by 𝑅(𝑇). More
precisely:
ker(𝑇) ∶= {𝒗 ∈ 𝑉 | 𝑇(𝒗) = 𝟎 ∈ 𝑊}

𝑅(𝑇) ∶= {𝒘 ∈ 𝑊 | ∃ 𝒗 ∈ 𝑉 𝑠𝑢𝑐ℎ 𝑡ℎ𝑎𝑡 𝒘 = 𝑇(𝒗) } .

Theorem 4.4: Let 𝑉 and 𝑊 be vector spaces. If 𝑇 ∶ 𝑉 → 𝑊 is a linear transformation, then:

(a) The kernel of 𝑇 is a subspace of 𝑉, and

(b) The range of 𝑇 is a subspace of 𝑊.

Proof: Will be explained in class.

Theorem 4.5: Dimension Theorem for Linear Transformations

If 𝑇 ∶ 𝑉 → 𝑊 is a linear transformation from a finite-dimensional vector space 𝑉 to a vector space
𝑊, then the range of 𝑇 is finite-dimensional, and

dim(𝑅(𝑇)) + dim(ker(𝑇)) = dim(𝑉).

4
Dr. Supem Samarasiri, Department of Mathematics, University of Moratuwa
 Linear Algebra - MA2034

Definition 4.3: (Rank and Nullity of Linear Transformations)

Let 𝑇 ∶ 𝑉 → 𝑊 be a linear transformation. If the range of 𝑇 is finite-dimensional, then its
dimension is called the rank of 𝑻; and if the kernel of 𝑇 is finite-dimensional, then its dimension
is called the nullity of 𝑻. The rank of 𝑇 is denoted by rank(𝑇) and the nullity of 𝑇 by nullity(𝑇).

Example 6: Kernel and Range of the Zero Transformation

Let 𝑇 ∶ 𝑉 → 𝑊 be the zero transformation. Since 𝑇 maps every vector in 𝑉 into 0, it follows that
ker(𝑇) = 𝑉.
Moreover, since 0 is the only image under 𝑇 of vectors in 𝑉, it follows that 𝑅(𝑇) = {0}.

Example 7: Kernel and Range of the Identity Operator

Let 𝐼: 𝑉 → 𝑉 be the identity operator. Since 𝐼(𝑣) = 𝑣 for all vectors in 𝑉, every vector in 𝑉 is the image
of some vector (namely, itself); thus 𝑅(𝐼) = 𝑉. Since the only vector that 𝐼 maps into 0 is 0, it follows that
𝑘𝑒𝑟(𝐼) = {0}.

Example 8: Kernel and Range of a Rotation

Let 𝑇 ∶ ℝ2 → ℝ2 be the linear operator that rotates each vector in the 𝑥𝑦-plane through the angle 𝜃. Since
every vector in the 𝑥𝑦-plane can be obtained by rotating some vector through the angle 𝜃, it follows that
𝑅(𝑇) = ℝ2 . Moreover, the only vector that rotates into 0 is 0, so 𝑘𝑒𝑟(𝑇) = {0}.

5
Dr. Supem Samarasiri, Department of Mathematics, University of Moratuwa