Lecture 7. Maps Between Spaces: change of basis.

Composition of Linear Mapping

Definition 1. Suppose 𝑉, 𝑈 and 𝑊 be vector spaces over 𝕂 and suppose that 𝐹 ∶
𝑉 → 𝑈 and 𝐺 ∶ 𝑈 → 𝑊 are linear mappings. The composition of linear
mapping 𝐺 ∘ 𝐹 is the mapping from 𝑉 to 𝑊 defined by (𝐺 ∘ 𝐹)(𝑣) = 𝐺(𝐹(𝑣)).
Proposition 1. Let 𝑉, 𝑈 and 𝑊 be vector spaces over 𝕂. If 𝐹 ∶ 𝑉 → 𝑈 and 𝐺 ∶
𝑈 → 𝑊 are linear mappings, then 𝐺 ∘ 𝐹 is a linear mapping.
Example 1. Let 𝐹, 𝐺 ∶ ℝ! → ℝ! be linear mappings defined by

𝑥 𝑥+𝑦 𝑥 2𝑥
𝐹 1𝑦4 = 1 4 and 𝐺 1𝑦4 = : =.
𝑥−𝑦 𝑥 − 3𝑦

Then
𝑥 𝑥+𝑦
(𝐺 ∘ 𝐹 )(𝑣 ) = 𝐺?𝐹 (𝑣 )@ = 𝐺 A𝐹 1𝑦4B = 𝐺 1 4=
𝑥−𝑦
2(𝑥 + 𝑦) 2𝑥 + 3𝑦
: ==: =.
(𝑥 + 𝑦) − 3(𝑥 − 𝑦) −2𝑥 + 4𝑦

𝑥 2𝑥
(𝐹 ∘ 𝐺 )(𝑣 ) = 𝐹?𝐺 (𝑣)@ = 𝐹 A𝐺 1𝑦4B = 𝐹 : ==
𝑥 − 3𝑦
2𝑥 + (𝑥 − 3𝑦) 3𝑥 − 3𝑦
: ==: =.
2𝑥 − (𝑥 − 3𝑦) 𝑥 + 3𝑦

We note that 𝐹 ∘ 𝐺 ≠ 𝐺 ∘ 𝐹.
Proposition 2. The composition operation for any linear mappings 𝐹, 𝐺, 𝐻 is
associative, that is,
𝐻 ∘ (𝐺 ∘ 𝐹 ) = (𝐻 ∘ 𝐺 ) ∘ 𝐹.

Matrix representation of a linear mapping

Let 𝑉 be a vector space over 𝕂 and 𝑆 = {𝑢" , . . . , 𝑢# } be a basis of 𝑉. Then for any
𝑣 = 𝜆" 𝑢" + . . . + 𝜆# 𝑢# we can find its coordinates with respect to 𝑆 and have

𝜆"
[𝑣 ]$ = M ⋮ O.
𝜆#

Recall that by the correspondence 𝑣 ⟼ [𝑣]$ we can construct isomorphism

between the spaces 𝑉 and ℝ# . Below we see that if 𝑑𝑖𝑚𝑉 = 𝑛, then there is an
isomorphism between 𝐻𝑜𝑚(𝑉, 𝑉 ) and 𝑀#,# with respect to the basis 𝑆.
Definition 2. Let 𝑇 ∈ 𝐻𝑜𝑚(𝑉, 𝑉) over 𝕂. Let 𝑆 = {𝑢" , . . . , 𝑢# } is a basis of 𝑉.
Suppose

𝑇(𝑢" ) = 𝑎"" 𝑢" + 𝑎!" 𝑢! +. . . +𝑎#" 𝑢# ,

𝑇(𝑢! ) = 𝑎"! 𝑢" + 𝑎!! 𝑢! +. . . +𝑎#! 𝑢# ,
…
𝑇(𝑢# ) = 𝑎"# 𝑢" + 𝑎!# 𝑢! +. . . +𝑎## 𝑢# ,

Then
𝑎"" 𝑎"! … 𝑎"#
[𝑇]$ = [ 𝑎!"
𝑎!! … 𝑎!#
\
… … … …
𝑎#" 𝑎#! … 𝑎##

is said to be the matrix representation of 𝑇 relative to the basis 𝑆.

𝑥 𝑥+𝑦
Example 2. a) Let 𝑇 ∶ ℝ! → ℝ! be linear mappings defined by 𝑇 1𝑦4 = 1𝑥 − 𝑦4
and 𝑆 = {𝑒" , 𝑒! } be the standard basis. Then

1 1 1 0
𝑇(𝑒") = 𝑇 1 4 = 1 4 = 1 4 + 1 4 = 𝑒" + 𝑒! = 1 ⋅ 𝑒" + 1 ⋅ 𝑒! ,
0 1 0 1
0 1 1 0
𝑇(𝑒! ) = 𝑇 1 4 = 1 4 = 1 4 − 1 4 = 𝑒" − 𝑒! = 1 ⋅ 𝑒" + (−1) ⋅ 𝑒! .
1 −1 0 1

The matrix representation of 𝑇 in the basis 𝑆 is

[𝑇]$ = 11 1
4.
1 −1
𝑥
b) Again, consider the linear mapping 𝑇 in the preceding example 𝑇 1𝑦4 =
𝑥+𝑦 1 1
1 4 and with new basis 𝑆 & = {𝑓" , 𝑓! } where 𝑓" = 1 4 and 𝑓! = 1 4. Then
𝑥−𝑦 1 0

1 2 1 1 𝜆 + 𝜆!
𝑇(𝑓" ) = 𝑇 1 4 = 1 4 = 𝜆" 1 4 + 𝜆! 1 4 = : " =.
1 0 1 0 𝜆"

We obtain 𝜆" + 𝜆! = 2 and 𝜆" = 0. It implies 𝜆" = 0 and 𝜆! = 2. Thus,

1 1
𝑇 (𝑓" ) = 0 1 4 + 2 1 4 = 0𝑓" + 2𝑓! = 0 ⋅ 𝑓" + 2 ⋅ 𝑓! .
1 0

1 1 1 1 𝜇 +𝜇
𝑇(𝑓! ) = 𝑇 1 4 = 1 4 = 𝜇" 1 4 + 𝜇! 1 4 = 1 " 𝜇 ! 4.
0 1 1 0 "
We obtain 𝜇" + 𝜇! = 1 and 𝜇" = 1. It implies 𝜇" = 1 and 𝜇! = 0. Thus,
 1 1
𝑇(𝑓! ) = 1 1 4 + 0 1 4 = 1𝑓" + 0𝑓! = 1 ⋅ 𝑓" + 0 ⋅ 𝑓! .
1 0

The matrix representation of 𝑇 in the basis 𝑆′ is

[𝑇]$& = 10 1
4.
2 0

Properties of Matrix Representations

Theorem 1. Let 𝑇 ∶ 𝑉 → 𝑉 be a linear mapping and 𝑆 be a basis of 𝑉. Then for
any 𝑣 ∈ 𝑉, we have [𝑇]$ [𝑣]$ = [𝑇(𝑣 )]$ .

𝑥 2𝑥 + 𝑦
Let 𝑇 1𝑦4 = : = and 𝑆 = {𝑔" , 𝑔! } be a basis, where
3𝑥 − 4𝑦

1 1
𝑔" = 1 4 and 𝑔! = 1 4.
−2 −1

2
Let 𝑣 = 1 4. Then
3

2 1 1
𝑣 = 1 4 = −5 1 4 + 7 1 4 = −5𝑔" + 7𝑔! .
3 −2 −1

So
[𝑣 ]' = 1−54.
7
1 0 1 1
𝑇 (𝑔" ) = 𝑇 1 4 = 1 4 = −11 1 4 + 11 1 4 = −11𝑔" + 11𝑔! .
−2 11 −2 −1
1 1 1 1
𝑇 (𝑔! ) = 𝑇 1 4 = 1 4 = −8 1 4 + 9 1 4 = −8𝑔" + 9𝑔! .
−1 7 −2 −1

Then
[𝑇]$ = 1−11 −8
4.
11 9

2 7 1 1
𝑇(𝑣 ) = 𝑇 1 4 = 1 4 = − 1 4 + 8 1 4 = −𝑔" + 8𝑔!
3 −6 −2 −1
−1
and so [𝑇 (𝑣)]$ = 1 4.
8
[𝑇 ]$ [𝑣]$ = 1−11 −8 −5 −1
4 1 4 = 1 4 = [𝑇 (𝑣)]$ .
11 9 7 8
Theorem 2. Let 𝑉 be an 𝑛 −dimensional vector space over 𝕂, let 𝑆 be a basis of
𝑉. Let 𝑀#,# be a vector space of 𝑛 × 𝑛 matrices. Then the mapping

𝜑 ∶ 𝐻𝑜𝑚(𝑉, 𝑉 ) → 𝑀#,#

defined by 𝜑(𝑇 ) = [𝑇 ]$ is a vector space isomorphism. That is, for any 𝐹, 𝐺 ∈

𝐻𝑜𝑚(𝑉, 𝑉 ) and any 𝜆 ∈ 𝕂,
1. 𝜑(𝐹 + 𝐺) = [𝐹 ]$ + [𝐺 ]$
2. 𝜑(𝜆𝐹) = 𝜆[𝐹 ]$
3. 𝜑 is one-to-one and onto.

Let 𝑉 be a vector space over 𝕂. Then we select two different bases, say 𝑆 and 𝑆′
of 𝑉 . Then there arises a question: According to these bases how do our
representations change?

Change of basis matrix

Definition 3. Let 𝑉 be a vector space over 𝕂 and 𝑆 = {𝑢" , 𝑢! , . . . , 𝑢# } and 𝑆′ =
{𝑣" , 𝑣! , . . . , 𝑣# } be bases of 𝑉. Write basis vectors in 𝑆′ as a linear combination of
basis vectors in 𝑆. Suppose

𝑣" = 𝑎"" 𝑢" + 𝑎!" 𝑢! +. . . +𝑎#" 𝑢# ,

𝑣! = 𝑎"! 𝑢" + 𝑎!! 𝑢! +. . . +𝑎#! 𝑢# ,
…
𝑣# = 𝑎"# 𝑢" + 𝑎!# 𝑢! +. . . +𝑎## 𝑢# ,
and
𝑎"" 𝑎"! … 𝑎"#
𝑎 𝑎!! … 𝑎!#
𝑃 = [ !" \
… … … …
𝑎#" 𝑎#! … 𝑎##

Then 𝑃 is called the change-of-basis matrix or transformation matrix from the

”old” basis 𝑆 to the ”new” basis 𝑆′.
1 1
Example 3. Consider 𝑉 = ℝ! . Let 𝑆 = {𝑢" , 𝑢! } = {1 4 , 1 4} and 𝑆′ =
1 0
1 2
{𝑣" , 𝑣! } = {1 4 , 1 4}. We write the vectors 𝑣" and 𝑣! as linear combinations of
2 5
𝑢" and 𝑢! . Then

1 1 1 𝜆 + 𝜆!
1 4 = 𝜆" 1 4 + 𝜆! 1 4 = : " = . Then 𝜆" = 2, 𝜆! = −1.
2 1 0 𝜆"

2 1 1 𝜇 + 𝜇!
1 4 = 𝜇" 1 4 + 𝜇! 1 4 = 1 " 4 . Then 𝜇" = 5, 𝜇! = −3.
5 1 0 𝜇"
Thus,
 𝑣" = 2𝑢" − 𝑢!
𝑣! = 5𝑢" − 3𝑢!
and hence,

2 5
𝑃=1 4.
−1 −3

Proposition 3. Let 𝑃 be a change-of-basis matrix from a basis 𝑆 to a basis 𝑆′ and

𝑄 be a change-of-basis matrix from a basis 𝑆′ to a basis 𝑆. Then 𝑄 = 𝑃(" .

1 1
Consider 𝑉 = ℝ! . Let 𝑆 = {𝑢" , 𝑢! } = {1 4 , 1 4} and 𝑆′ = {𝑣" , 𝑣! } =
1 0
1 2
{1 4 , 1 4}. We have just seen that the change-of-basis matrix from 𝑆 to 𝑆′ is
2 5
2 5
𝑃=1 4.
−1 −3
We will see below that the change-of-basis matrix 𝑄 from 𝑆′ to 𝑆 is 𝑃(" .

1 1 2 𝜆 + 2𝜆!
1 4 = 𝜆" 1 4 + 𝜆! 1 4 = : " = . Then 𝜆" = 3, 𝜆! = −1.
1 2 5 2𝜆" + 5𝜆!

1 1 2 𝜇 + 2𝜇!
1 4 = 𝜇" 1 4 + 𝜇! 1 4 = : " = . Then 𝜇" = 5, 𝜇! = −2.
0 2 5 2𝜇" + 5𝜇!
Thus,
𝑢" = 3𝑣" − 𝑣!
𝑢! = 5𝑣" − 2𝑣!
and hence,

3 5 4.
𝑄=1
−1 −2

Let us find the inverse of 𝑃 by the formula given in Lecture 6 and have

1
1−3
−54 −3 −54 1 3 54
𝑃(" = = −1 = = 𝑄.
2 ⋅ (−3) − 5 ⋅ (−1) 1 2 1 2 −1 −2

Change of coordinates
Theorem 3. Let 𝑃 be the change-of-basis matrix from a basis 𝑆 to a basis 𝑆′ in a
vector space 𝑉. Then

𝑃[𝑣 ]$& = [𝑣]$ or 𝑃(" [𝑣 ]$ = [𝑣]$& .

 1 0 0
Let 𝑆 = {𝑒" , 𝑒! , 𝑒) } = sM0O , M1O , M0Ot and 𝑆 & = {𝑓" , 𝑓! , 𝑓) } =
0 0 1
1 1 1 1
)
sM1O , M1O , M0Ot. We know that they are bases of ℝ . Let 𝑣 = M−2O.
1 0 0 3

Since 𝑓" = 𝑒" + 𝑒! + 𝑒) , 𝑓! = 𝑒" + 𝑒! , 𝑓) = 𝑒" we have

1 1 1
𝑃 = M1 1 0O.
1 0 0
1
Then it is clear that 𝑣 = 𝑒" − 2𝑒! + 3𝑒) and have [𝑣]$ = M−2O. One can write
3
3
𝑣 = 3𝑓" − 5𝑓! + 3𝑓) . Consequently, [𝑣 ]$& = M−5O. Then
3

1 1 1 3 1 ⋅ 3 + 1 ⋅ (−5) + 1 ⋅ 3 1
𝑃 [𝑣]$& = M1 1 0O M−5O = [1 ⋅ 3 + 1 ⋅ (−5) + 0 ⋅ 3\ = M−2O = [𝑣 ]$ .
1 0 0 3 1 ⋅ 3 + 0 ⋅ (−5) + 0 ⋅ 3 3

Two matrix representations

Theorem 4. Let 𝑃 be he change-of-basis matrix from a basis 𝑆 to a basis 𝑆′ vector
space 𝑉. Then for any linear mapping 𝑇 on 𝑉 (𝑇 ∶ 𝑉 → 𝑉)

[𝑇 ]$& = 𝑃(" [𝑇]$ 𝑃.

𝑥 𝑥+𝑦
Example 4. Let 𝑇 ∶ ℝ! → ℝ! be linear mappings defined by 𝑇 1𝑦4 = 1𝑥 − 𝑦4
and 𝑆 = {𝑒" , 𝑒! } be the standard basis. The matrix representation of T in the
basis 𝑆 is

[𝑇]$ = 11 1
4.
1 −1

1 1
If 𝑆 = {𝑓" , 𝑓! } where 𝑓" = 1 4 and 𝑓! = 1 4,then
1 0

[𝑇]$& = 10 1
4.
2 0
One can easily obtain that

1 1 0 1
𝑃=1 4 and 𝑃(" = 1 4.
1 0 1 −1

Then
0 1 1 1 1 1
𝑃(" [𝑇]$ 𝑃 = 1 41 41 4=
1 −1 1 −1 1 0

1 −1 1 1 0 1
=1 41 4=1 4 = [𝑇]$&
0 2 1 0 2 0

Similarity
Definition 4. Suppose matrices 𝐴 and 𝐵 be matrices for which there exists an
invertible matrix 𝑃 such that 𝐵 = 𝑃(" 𝐴𝑃. Then 𝐵 is said to be similar to 𝐴.

1 1 0 1
Example 5. If 𝐴 = 1 4 and 𝐵 = 1 4, then 𝐵 is similar to 𝐴, because
1 −1 2 0
1 1
there exists matrix 𝑃 = 1 4 such that
1 0
0 1 1 1 1 1 0 1
1 41 41 4=1 4.
1 −1 1 −1 1 0 2 0

Theorem 5. Two matrices represent the same linear mapping if and only if the
matrices are similar.

Glossary
matrix representation матричное представление
similarity подобие

Exercises for lecture 7

Algebra of Linear Operators
1. Let 𝐹 and 𝐺 be the linear operators on ℝ! defined by 𝐹(𝑥, 𝑦) = (𝑥 + 𝑦, 0) and
𝐺(𝑥, 𝑦) = (−𝑦, 𝑥). Find formulas defining the linear operators:
(a) 𝐹 + 𝐺, (b) 5𝐹 − 3𝐺, (c) 𝐹𝐺, (d) 𝐺𝐹, (e) 𝐹 ! , ( f ) 𝐺 ! .

Matrices and Linear Operators

2. Let 𝐹: ℝ! →ℝ! be defined by 𝐹(𝑥, 𝑦) = (4𝑥 + 5𝑦, 2𝑥 − 𝑦).
(a) Find the matrix 𝐴 representing 𝐹 in the usual basis 𝐸.
(b) Find the matrix 𝐵 representing 𝐹 in the basis S={𝑢" , 𝑢! } =
{(1, 4), (2, 9)}.
 (c) Find 𝑃 such that 𝐵 = 𝑃(" 𝐴𝑃.
(d) For 𝑣 = (𝑎, 𝑏), find [𝑣 ]$ and [𝐹(𝑣)]$ . Verify that[𝐹 ]$ [𝑣]$ = [𝐹(𝑣)]$ .

3. Find the matrix representing each linear transformation 𝑇 on ℝ) relative to the

usual basis of ℝ) :
(a) 𝑇(𝑥, 𝑦, 𝑧) = (𝑥, 𝑦, 0). (b) 𝑇 (𝑥, 𝑦, 𝑧) = (𝑧, 𝑦 + 𝑧, 𝑥 + 𝑦 + 𝑧).
(c) 𝑇(𝑥, 𝑦, 𝑧) = (2𝑥 − 7𝑦 − 4𝑧, 3𝑥 + 𝑦 + 4𝑧, 6𝑥 − 8𝑦 + 𝑧).

Change of Basis
4. Find the change-of-basis matrix P from the usual basis 𝐸 of ℝ! to a basis 𝑆,
the change-of-basis matrix 𝑄 from 𝑆 back to 𝐸, and the coordinates of 𝑣 = (𝑎, 𝑏)
relative to 𝑆, for the following bases 𝑆:
(a) 𝑆 = {(1,2), (3,5)}. (c) 𝑆 = {(2,5), (3,7)}.
(b) 𝑆 = {(1, −3), (3, −8)}. (d) 𝑆 = {(2,3), (4,5)}.

Linear Operators and Change of Basis

5. Consider the linear operator 𝐹 on ℝ! defined by 𝐹 (𝑥, 𝑦) = (5𝑥 + 𝑦, 3𝑥 − 2𝑦)
and the following bases of ℝ! :
𝑆 = {(1,2), (2,3)} and 𝑆 & = {(1,3), (1,4)}
(a) Find the matrix 𝐴 representing 𝐹 relative to the basis 𝑆.
(b) Find the matrix 𝐵 representing 𝐹 relative to the basis 𝑆′.
(c) Find the change-of-basis matrix 𝑃 from 𝑆 to 𝑆′.
(d) How are 𝐴 and 𝐵 related?

Homework 7
Algebra of Linear Operators
1. Let 𝐹 and 𝐺 be the linear operators on ℝ! defined by 𝐹(𝑥, 𝑦) = (𝑥 + 𝑦, 0) and
𝐺(𝑥, 𝑦) = (−𝑦, 𝑥). Find formulas defining the linear operators:
(a) 𝐺𝐹, (b) 𝐹 ! , ( c ) 𝐺 ! .

Matrices and Linear Operators

2. Let𝐹: ℝ! → ℝ! be defined by 𝐹(𝑥, 𝑦) = (5𝑥 + 4𝑦, −𝑥 + 2𝑦).
(a) Find the matrix 𝐴 representing 𝐹 in the usual basis 𝐸.
(b) Find the matrix 𝐵 representing 𝐹 in the basis S={𝑢" , 𝑢! } =
{(4, 1), (3, 4)}.
(c) Find 𝑃 such that 𝐵 = 𝑃(" 𝐴𝑃.
(d) For 𝑣 = (𝑎, 𝑏), find [𝑣 ]$ and [𝐹(𝑣)]$ . Verify that[𝐹 ]$ [𝑣]$ = [𝐹(𝑣)]$ .

3. Find the matrix representing each linear transformation 𝑇 on ℝ) relative to the

usual basis of ℝ) :
(a) 𝑇(𝑥, 𝑦, 𝑧) = (0, 𝑦, 𝑧). (b) 𝑇(𝑥, 𝑦, 𝑧) = (𝑦, 𝑥 + 𝑧, 𝑥 + 𝑦 + 𝑧).
 (c) 𝑇(𝑥, 𝑦, 𝑧) = (2𝑥 − 5𝑦 − 2𝑧, 3𝑥 + 𝑦 + 6𝑧, 4𝑥 − 2𝑦 + 𝑧).

Change of Basis
4. Find the change-of-basis matrix P from the usual basis 𝐸 of ℝ! to a basis 𝑆,
the change-of-basis matrix 𝑄 from 𝑆 back to 𝐸, and the coordinates of 𝑣 = (𝑎, 𝑏)
relative to 𝑆, for the following bases 𝑆:
(a) 𝑆 = {(1,3), (2,5)}. (c) 𝑆 = {(3,5), (2,7)}.
(b) 𝑆 = {(1, −2), (2, −8)}. (d) 𝑆 = {(3,2), (5,4)}.

Linear Operators and Change of Basis

5. Consider the linear operator 𝐹 on ℝ! defined by 𝐹 (𝑥, 𝑦) = (3𝑥 + 𝑦, 5𝑥 − 2𝑦)
and the following bases of ℝ! :
𝑆 = {(1,3), (3,4)} and 𝑆 & = {(1,2), (1,4)}
(a) Find the matrix 𝐴 representing 𝐹 relative to the basis 𝑆.
(b) Find the matrix 𝐵 representing 𝐹 relative to the basis 𝑆′.
(c) Find the change-of-basis matrix 𝑃 from 𝑆 to 𝑆′.
(d) How are 𝐴 and 𝐵 related?