Lecture 8 - Vector Spaces

Part 2
COM1033 - Foundations of Computing II

Prof. Ferrante Neri

Learning outcomes

By the end of this lecture we will have learned:

⊚ Bases of Vector Spaces
⊚ Steinitz’ Lemma
⊚ Grassmann Formula

Learning Resources
Chapter 8, Sections 8.5: F. Neri, Linear Algebra for
Computational Sciences and Engineering, 2019, pages 304-319

1
Basis of a Vector Space
Definition of Basis
Definition
Let (𝐸, +, ·) be a finite-dimensional vector space. A basis
𝐵 = {v1 , v2 , . . . , vn } of (𝐸, +, ·) is a set of vectors ∈ 𝐸 that verify
the following properties.

⊚ v1 , v2 , . . . , vn are linearly independent

⊚ v1 , v2 , . . . , vn span 𝐸, i.e. 𝐸 = 𝐿 (v1 , v2 , . . . , vn ).

Example
Let us consider the vector space ℝ3 . A basis 𝐵 of ℝ3

© 0 ª © 0 ª © 1 ª
­ 0 ®, ­ 1 ®, ­ 0 ®
­ ® ­ ® ­ ®
« 1 ¬ « 0 ¬ « 0 ¬
as they are linearly independent and all the numbers in ℝ3
can be derived by their linear combination. 3
Example: Basis of a Vector Space

Example
The following vectors span ℝ3

© 0 ª © 0 ª © 1 ª © 1 ª
­ 0 ®, ­ 1 ®, ­ 0 ®, ­ 2 ®
­ ® ­ ® ­ ® ­ ®
« 1 ¬ « 0 ¬ « 0 ¬ « 3 ¬
as they still allow to generate all the numbers in ℝ3 but are not
linearly independent, thus not a basis.

4
Steinitz’s Lemma

Lemma

Steinitz’s Lemma. Let (𝐸, +, ·) be a finite-dimensional vector

space and 𝐿 (v1 , v2 , . . . , vn ) = 𝐸 its span.
Let w1 , w2 , . . . , ws be 𝑠 linearly independent vectors ∈ 𝐸.
It follows that 𝑠 ≤ 𝑛, i.e. the number of a set of linearly independent
vectors cannot be higher than the number of vectors spanning the
vector space.

Corollary
Let (𝐸, +, ·) be a finite-dimensional vector space and
𝐿 (v1 , v2 , . . . , vn ) = 𝐸 its span and 𝐵 = {w1 , w2 , . . . , ws } be its
basis. It follows that 𝑠 ≤ 𝑛.

5
Steinitz’ Lemma

Example
The following four vectors span ℝ3

© 0 ª © 0 ª © 1 ª © 1 ª
v1 = ­ 0 ®, v2 = ­ 1 ®, v3 = ­ 0 ®, v4 = ­­ 2 ®®
­ ® ­ ® ­ ®

« 1 ¬ « 0 ¬ « 0 ¬ « 3 ¬
while the following vectors are linearly independent ∈ ℝ3
© 1 ª © 0 ª © 0 ª
w1 = ­ 4 ®, w2 = ­ 5 ®, w3 = ­­ 2 ®®.
­ ® ­ ®

« 0 ¬ « 0 ¬ « 1 ¬
Steinitz’s lemma says that any set of linearly independent
vectors belonging to a set 𝐸 cannot be more numerous than a
any set of vectors spanning 𝐸.

6
Dimension of a Vector Space
Order of a Basis

Definition
The number of vectors composing a basis is said order of a basis.

Theorem

Let (𝐸, +, ·) be a finite-dimensional vector space. All the bases of a

vector space have the same order.

Example
If we consider two bases of ℝ3 , they will have the same order.
We know that in ℝ3 at most three vectors can be linearly
independent, thus a basis of ℝ3 can be composed of at most
three vectors. We have seen that at least three vectors are
needed to span ℝ3 . Thus each basis must have three vectors.

8
Dimension of a Vector Space

Definition
Let (𝐸, +, ·) be a finite-dimensional vector space. The order of
a basis of (𝐸, +, ·) is said dimension of (𝐸, +, ·) and is indicated
with dim (𝐸, +, ·) or simply with dim (𝐸).

Theorem

Let (𝐸, +, ·) be a finite-dimensional vector space.

The dimension dim (𝐸, +, ·) = 𝑛 of a vector space (or simply
dim (𝐸)) is

⊚ the maximum number of linearly independent vectors of 𝐸;

⊚ the minimum number of vectors spanning 𝐸

9
Dimension of a Vector Space

Proof.
If dim (𝐸, +, ·) = 𝑛, then let 𝐵 = {v1 , v2 , . . . , vn } be it basis. Let
us assume, by contradiction, that there exist 𝑛 + 1 linearly
independent vectors in 𝐸 (those 𝑛 vectors in 𝐵 plus one). Since
𝐵 is a basis, its 𝑛 vectors span 𝐸. For Steinitz Lemma,
𝑛 + 1 ≤ 𝑛, that is a contradiction.
Let us assume, by contradiction, that 𝑛 − 1 vectors are enough
to span 𝐸. Since 𝐵 is a basis then its 𝑛 vectors are linearly
independent. For Steinitz’s Lemma, 𝑛 ≤ 𝑛 − 1, that is a
contradiction.
□

10
Dimension of a Vector space
Example
Let us consider the vector space (𝐸, +, ·) where

𝐸 = {(𝑥, 𝑦, 𝑧) ∈ ℝ3 |𝑥 − 3𝑦 − 7𝑧 = 0}.

In order to determine span and basis of this vector space, let

us assume that 𝑦 = 𝛼, 𝑧 = 𝛽, and let us solve the equation with
respect to 𝑥 : 𝑥 = 3𝛼 + 7𝛽. Hence, there are ∞2 solutions of the
kind (3𝛼 + 7𝛽, 𝛼, 𝛽). This expression can be written as

(3𝛼 + 7𝛽, 𝛼, 𝛽) = (3𝛼, 𝛼, 0) + (7𝛽, 0, 𝛽) = 𝛼 (3, 1, 0) + 𝛽 (7, 0, 1) .

Hence, 𝐸 = 𝐿 ((3, 1, 0) , (7, 0, 1)). The two vectors are linearly

independent and compose a basis 𝐵 = {(3, 1, 0) , (7, 0, 1)}.
Thus, dim (𝐸, +, ·) = 2.
11
Dimension of a Vector Space

Example
Let us consider now the vector space (𝐸, +, ·) where

𝑥−𝑦=0

 
 


 
 


 
 

𝐸 = (𝑥, 𝑦, 𝑧) ∈ ℝ3 𝑦+𝑧=0 .
  
 3𝑥 + 𝑧 = 0

 
 

 
  
Since the incomplete matrix is non-singular, the only solution
of the linear system is (0, 0, 0). Hence, 𝐿 ((0, 0, 0)) = 𝐸. In this
special case, the vector space is composed only of the null
vector. Since there are no linearly independent vectors,
dim (𝐸, +, ·) = 0.

12
Dimension of a Vector Space
Example
More generally, if a vector subspace of ℝ3 is identified by 3
linear equations in 3 variables

⊚ if the rank of the system is 3, then dim (𝐸, +, ·) = 0. The

geometrical interpretation of this vector subspace is the
origin of a system of coordinates in three-dimensional
space (only one point).
⊚ if the rank of the system is 2, the system has ∞1 solutions
and dim (𝐸, +, ·) = 1. The geometrical interpretation of this
vector subspace is a line passing through the origin.
⊚ if the rank of the system is 1, the system has ∞2 solutions
and dim (𝐸, +, ·) = 2. The geometrical interpretation of this
vector subspace is a plane passing through the origin.
13
Grassmann’s Formula
Unique representation

Lemma

Let (𝐸, +, ·) be a vector space. Let the vectors v1 , v2 , . . . , vn ∈ 𝐸. If

the 𝑛 vectors are linearly dependent while 𝑛 − 1 are linearly
independent, there is a unique way to express one vector as the linear
combination of the others:

∀vk ∈ 𝐸, ∃!𝜆1 , 𝜆2 , . . . , 𝜆 𝑘−1 , 𝜆 𝑘+1 , . . . , 𝜆𝑛 ≠ 0, 0, . . . , 0 such that

vk = 𝜆1 v1 + 𝜆2 v2 + . . . + 𝜆 𝑘−1 vk−1 + 𝜆 𝑘+1 vk+1 + . . . + 𝜆𝑛 vn

15
Example: Unique representation

Example
Let us consider the following linearly dependent vectors ∈ ℝ3 :
© 1 ª © 1 ª © 2 ª
v1 = ­ 0 ®, v2 = ­ 1 ®, v3 = ­­ 1 ®®.
­ ® ­ ®

« 1 ¬ « 1 ¬ « 2 ¬
Any pair of them is linearly independent. Let us express v3 as
a linear combination of the other two vectors:

© 2 ª © 1 ª © 1 ª
­ 1 ® = 𝜆1 ­ 0 ® + 𝜆2 ­ 1 ®
­ ® ­ ® ­ ®
2
« ¬ 1
« ¬ « 1 ¬
which results in the system

16
Example: Unique representation

Example
𝜆1 + 𝜆2 = 2






𝜆 =1
2

 𝜆1 + 𝜆2 = 2



that is determined and its only solution is 𝜆1 , 𝜆2 = 1, 1.

Lemma
Basis Extension theorem. Any list of linearly independent
vectors ∈ 𝑈 but not spanning 𝑈 can be extended to a basis of 𝑈.

Example
The vectors of ℝ3 , (0, 0, 1) and (1, 0, 0) are linearly
independent. We can add (1, 4, 1) to the list and obtain a basis.
17
Grassmann’s Formula
Theorem
Grassmann’s Formula. Let (𝐸, +, ·) be a finite-dimensional vector
space. Let (𝑈 , +, ·) and (𝑉 , +, ·) be vector subspaces of (𝐸, +, ·).
Then, dim (𝑈 + 𝑉) + dim (𝑈 ∩ 𝑉) = dim (𝑈) + dim (𝑉).

Proof.
Let us suppose that dim (𝑈) = 𝑟 and dim (𝑉) = 𝑠. (𝑈 ∩ 𝑉 , +, ·)
is a vector subspace and let 𝐵𝑈∩𝑉 = {t1 , t2 , . . . , tl } be its basis.
Since all the vectors contained in 𝐵𝑈∩𝑉 are also vectors in 𝑈,
for the basis extension theorem, we can obtain 𝐵𝑈 and 𝐵𝑉
from 𝐵𝑈∩𝑉 :

𝐵𝑈 = {t1 , t2 , . . . , tl , ul+1 , ul+2 , . . . , ur }

𝐵𝑉 = {t1 , t2 , . . . , tl , vl+1 , vl+2 , . . . , vs } .

18
Grassmann’s Formula
Proof.
From the definition of the sum set

𝑆 = 𝑈 + 𝑉 = {w ∈ 𝐸|∃u ∈ 𝑈 , v ∈ 𝑉 |w = u + v}

w=u+v=
𝜆1 t1 + 𝜆2 t2 + . . . + 𝜆 𝑙 tl + 𝑎 𝑙+1 ul+1 + 𝑎 𝑙+2 ul+2 + . . . + 𝑎 𝑟 ur
+𝜇1 t1 + 𝜇2 t2 + . . . + 𝜇𝑙 tl + 𝑏 𝑙+1 vl+1 + 𝑏 𝑙+2 vl+2 + . . . + 𝑏 𝑠 vs =
= (𝜆1 + 𝜇1 ) t1 + (𝜆2 + 𝜇2 ) t2 + . . . (𝜆 𝑙 + 𝜇𝑙 ) tl +
+𝑎 𝑙+1 ul+1 + 𝑎 𝑙+2 ul+2 + . . . + 𝑎 𝑟 ur + 𝑏 𝑙+1 vl+1 + . . . + 𝑏 𝑠 vs
This means that 𝑟 + 𝑠 − 𝑙 vectors span 𝑈 + 𝑉

𝐿 (t1 , t2 , . . . , tl , ul+1 , ul+2 , . . . , ur , vl+1 , vl+2 , . . . , vs ) = 𝑈 + 𝑉.

19
Grassmann’s Formula
Proof.
Let us now check the linear independence of these 𝑟 + 𝑠 − 𝑙
vectors. Let us impose that

𝛼 1 t1 + 𝛼 2 t2 + . . . + 𝛼 𝑙 tl +
+𝛽 𝑙+1 ul+1 + 𝛽 𝑙+2 ul+2 + . . . + 𝛽 𝑟 ur +
+𝛾𝑙+1 vl+1 + 𝛾𝑙+2 vl+2 + . . . + 𝛾𝑠 vs = o.

Hence,

𝛼1 t1 + 𝛼 2 t2 + . . . + 𝛼 𝑙 tl +
+𝛽 𝑙+1 ul+1 + 𝛽 𝑙+2 ul+2 + . . . + 𝛽 𝑟 ur = d =
= − (𝛾𝑙+1 vl+1 + 𝛾𝑙+2 vl+2 + . . . + 𝛾𝑠 vs )

It follows that d ∈ 𝑉 and d ∈ 𝑈. Hence, d ∈ 𝑈 ∩ 𝑉.

20
Grassmann’s Formula
Proof.
This means that d can be expressed as

d = 𝛼′1 t1 + 𝛼′2 t2 + . . . + 𝛼′𝑙 tl

Since t1 , t2 , . . . , tl are linearly independent, there is only one

way (Lemma of unique representation), to represent d as a
linear combination of them. Thus, 𝛽 𝑙+1 = 𝛽 𝑙+2 = . . . = 𝛽 𝑟 = 0
and 𝛼1 = 𝛼′1 , 𝛼2 = 𝛼′2 , . . . 𝛼 𝑙 = 𝛼′𝑙 . Hence, we can write the
expression above as

𝛼 1 t1 + 𝛼 2 t2 + . . . + 𝛼 𝑙 tl +
+𝛾𝑙+1 vl+1 + 𝛾𝑙+2 vl+2 + . . . + 𝛾𝑠 vs = o.

The vectors are linearly independent (basis of 𝑉), thus

𝛼 1 = 𝛼2 = . . . = 𝛼 𝑙 = 𝛾𝑙+1 = 𝛾𝑙+2 = . . . = 𝛾𝑠 = 0. 21
Grassmann’s Formula

Proof.
Thus these 𝑟 + 𝑠 − 𝑙 vectors compose a basis 𝐵𝑈+𝑉 :

𝐵𝑈+𝑉 = {t1 , t2 , . . . , tl , ul+1 , ul+2 , . . . , ur , vl+1 , vl+2 , . . . , vs }

that is dim (𝑈 + 𝑉) = 𝑟 + 𝑠 − 𝑙, i.e.

dim (𝑈 + 𝑉) = dim (𝑈) + dim (𝑉) − dim (𝑈 ∩ 𝑉) .

22
Example: Grassmann’s Formula

Example
Let us consider the vector space ℝ3 , +, · and two vector

subspaces: (𝑈 , +, ·) and (𝑉 , +, ·), respectively.

Let us verify/interpret Grassmann’s formula in the following
cases:

⊚ if 𝑈 = {(0, 0, 0)} and 𝑉 = ℝ3 then dim (𝑈) = 0 dim (𝑉) = 3.

In this case 𝑈 ∩ 𝑉 = (0, 0, 0) = 𝑈 and 𝑈 + 𝑉 = ℝ3 = 𝑉. It
follows that dim (𝑈 + 𝑉) + dim (𝑈 ∩ 𝑉) = 3 + 0 =
dim (𝑈) + dim (𝑉) = 0 + 3.

23
Example: Grassmann’s Formula

Example
⊚ If the dimension of both 𝑈 and 𝑉 is 1, i.e. only one linearly
independent vector and thus one line, we can distinguish
two subcases
◦ the two vectors in 𝑈 and 𝑉, respectively, represent two lines
passing through the origin. The intersection
𝑈 ∩ 𝑉 = (0, 0, 0) is the origin, while the sum 𝑈 + 𝑉 is the
plane that contains the two vectors. It follows that
dim (𝑈 + 𝑉)+dim (𝑈 ∩ 𝑉) = 2+0 = dim (𝑈)+dim (𝑉) = 1+1
◦ the two vectors in 𝑈 and 𝑉, respectively, represent two
coinciding lines. Both intersection and sum coincide with
the vector, i.e. 𝑈 ∩ 𝑉 = 𝑈 + 𝑉 = 𝑈 = 𝑉. It follows that
dim (𝑈 + 𝑉)+dim (𝑈 ∩ 𝑉) = 1+1 = dim (𝑈)+dim (𝑉) = 1+1

24
Example: Grassmann’s Formula

Example
⊚ if the dimension of 𝑈 is 1 while that of 𝑉 is 2, i.e. one line
passing through the origin and one plane passing through
the origin, we can distinguish two subcases:
◦ the line does not lay in the plane. It follows that
𝑈 ∩ 𝑉 = (0, 0, 0) and 𝑈 + 𝑉 = ℝ3 . Hence,
dim (𝑈 + 𝑉)+dim (𝑈 ∩ 𝑉) = 3+0 = dim (𝑈)+dim (𝑉) = 1+2
◦ the line lays in the plane. It follows that 𝑈 ∩ 𝑉 = 𝑈 and
𝑈 + 𝑉 = 𝑉. Hence,
dim (𝑈 + 𝑉)+dim (𝑈 ∩ 𝑉) = 2+1 = dim (𝑈)+dim (𝑉) = 1+2

25
Example: Grassmann’s Formula

Example
⊚ If the dimension of both 𝑈 and 𝑉 is 2, i.e. two linearly
independent vectors and thus two planes passing through
the origin, we can distinguish two subcases
◦ the planes do not coincide. It follows that 𝑈 ∩ 𝑉 is a line
while 𝑈 + 𝑉 = ℝ3 . Hence,
dim (𝑈 + 𝑉)+dim (𝑈 ∩ 𝑉) = 3+1 = dim (𝑈)+dim (𝑉) = 2+2
◦ the planes coincide. It follows that 𝑈 ∩ 𝑉 = 𝑈 + 𝑉 and
𝑈 + 𝑉 = 𝑈 = 𝑉, i.e. intersection and sum are the same
coinciding plane. Hence,
dim (𝑈 + 𝑉)+dim (𝑈 ∩ 𝑉) = 2+2 = dim (𝑈)+dim (𝑉) = 2+2

26
Summary and next lecture

Summary
⊚ A basis of a vector space is a set of linearly independent
vectors spanning the set of the vector space
⊚ The number of vectors in a basis is a special number that
characterises the dimension of a problem and is the
maximum number of linearly independent vectors and the
minimum number of vectors spanning the set
⊚ Grassmann’s formula states that if we sum two sets, its
resulting dimension depends on each set separately and
their intersection

The next lecture

We will introduce linear mappings.
27
Tutorial
Exercise

Let us consider the vector subspaces (𝑈 , +, ·) and (𝑉 , +, ·) where

𝑈 = {(𝑥, 𝑦, 𝑧) ∈ ℝ3 |𝑥 − 𝑦 + 𝑧 = 0}
𝑉 = {(𝑥, 𝑦, 𝑧) ∈ ℝ3 |𝑥 + 2𝑦 + 2𝑧 = 0}

1. Determine a basis of 𝑈
2. Determine the dimension of 𝑈
3. Determine a basis of 𝑉
4. Determine the dimension of 𝑉
5. Determine a basis of 𝑈 ∩ 𝑉
6. Determine the dimension of 𝑈 ∩ 𝑉
7. Determine the dimension of 𝑈 + 𝑉
8. Provide a geometrical interpretation of 𝑈, 𝑉, 𝑈 ∩ 𝑉, and
𝑈 +𝑉 29
Exercise

Let us consider the vector subspaces (𝑈 , +, ·) and (𝑉 , +, ·) where

( ( )
𝑥−𝑦+𝑧 =0
𝑈 = (𝑥, 𝑦, 𝑧) ∈ ℝ3
𝑦 − 2𝑧 = 0
𝑉 = (𝑥, 𝑦, 𝑧) ∈ ℝ3 𝑥 − 𝑦 − 𝑧 = 0


1. Determine a basis of 𝑈
2. Determine the dimension of 𝑈
3. Determine a basis of 𝑉
4. Determine the dimension of 𝑉
5. Determine a basis of 𝑈 ∩ 𝑉
6. Determine the dimension of 𝑈 ∩ 𝑉
7. Determine the dimension of 𝑈 + 𝑉
8. Provide a geometrical interpretation of 𝑈, 𝑉, 𝑈 ∩ 𝑉, 𝑈 + 𝑉 30
Exercise

Let us consider the vector subspaces (𝑈 , +, ·) and (𝑉 , +, ·) where

( ( )
𝑥−𝑦+𝑧 =0
𝑈 = (𝑥, 𝑦, 𝑧) ∈ ℝ3
𝑦−𝑧=0
( ( )
𝑥 + 2𝑦 + 𝑧 = 0
𝑉 = (𝑥, 𝑦, 𝑧) ∈ ℝ3
𝑥=0

1. Determine a basis of 𝑈
2. Determine the dimension of 𝑈
3. Determine a basis of 𝑉
4. Determine the dimension of 𝑉
5. Determine a basis of 𝑈 ∩ 𝑉
6. Determine the dimension of 𝑈 ∩ 𝑉
7. Determine the dimension of 𝑈 + 𝑉
8. Provide a geometrical interpretation of 𝑈, 𝑉, 𝑈 ∩ 𝑉, 𝑈 + 𝑉
31
Exercise

Let v1 ,v2 ,. . .,vn be 𝑛 vectors ∈ 𝐸. We know that

𝐿 (v1 , v2 , . . . , vn ) = 𝑈 ⊂ 𝐸

1. What can we conclude about the linear dependence of the

vectors?
2. What can we conclude about the dimension of 𝐸?
3. If the vectors are linearly independent what can we
conclude about the dimension of 𝐸?

32
Exercise

Let (𝑈 , +, ·) and (𝑉 , +, ·) be two vector spaces.

1. If 𝑈 ∩ 𝑉 = 𝑈 = 𝑉 what can we say about 𝑈 + 𝑉 and its

dimension?
2. If 𝑈 ∩ 𝑉 = {o} what can we say about 𝑈 + 𝑉 and its
dimension?
3. If dim (𝑈) = dim (𝑉) = 0 what is 𝑈 + 𝑉?
4. Could ever happen that 𝑈 ∩ 𝑉 = ∅?

33
THANK
YOU