MATH1049 Linear Algebra II lecture notes

Armando Martino

Spring 2024

Contents

Preamble 2
Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1 Groups 3

2 Fields and Vector Spaces 8

2.1 Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Vector spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Subspaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3 Bases 16
3.1 Spanning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2 Linear independence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 Bases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.4 Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

4 Linear Transformations 24
4.1 Matrix representation I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2 Kernel and image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.3 Isomorphism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.4 Dimension Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.5 Matrix representation II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

5 Determinants 34

6 Diagonalisability 38
6.1 Eigen-things . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.2 Diagonalisability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
6.3 Cayley–Hamilton Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

7 Coursework Sheets 46

1
2 CONTENTS

7.1 Coursework Sheet 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

7.2 Coursework Sheet 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
7.3 Coursework Sheet 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
7.4 Coursework Sheet 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
7.5 Coursework Sheet 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
7.6 Coursework Sheet 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
7.7 Coursework Sheet 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
7.8 Coursework Sheet 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
7.9 Coursework Sheet 9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Preamble

These lecture notes contain more–less verbatim what appears on the (black– or white–)board during the
lectures.
The colour coding - in the pdf version - signifies the following: what is red is being defined. Blue are
named theorems and statements (these could be referred to by these names). Finally green is used to
reference lecture notes of other modules (mostly MATH1048 Linear Algebra I).

Acknowledgement

For the most part, these notes were designed and written by Dr Bernhard Koeck, and originally typed
into LaTeX by the undergraduate student Thomas Blundell–Hunter in summer 2011. They have been
regularly updated to reflect the changes in the syllabus. The current version is a complete overhaul of
the technical side of the notes by Ján Špakula, so that they are available also in html.

®
MIX
Paper from
responsible sources
FSC® C020438
 1 GROUPS 3

1 Groups

The cartesian product G × H of two sets G and H is defined as the set

G × H := {(g, h) | g ∈ G, h ∈ H},

consisting of all (ordered) pairs (g, h) where g ∈ G and h ∈ H. For example, R2 = R × R.

Definition 1.1. (a) A binary operation on a set G is a function G × G → G (from the cartesian product
G × G = {(a, b) | a ∈ G, b ∈ G} to G). We write a ∗ b (or ab, or a ◦ b, or a + b) for the image of the
pair (a, b) ∈ G × G in G under this function.

(b) A group is a set G together with a binary operation G × G → G, (a, b) 7→ a ∗ b (which we usually
refer to as “group operation”, or “group multiplication”, or just “multiplication”) such that the
following axioms are satisfied:

(i) Associativity: For all a, b, c ∈ G, we have (a ∗ b) ∗ c = a ∗ (b ∗ c) in G.

(ii) Existence of the identity (or neutral) element: There exists e ∈ G, such that for all a ∈ G we
have e ∗ a = a = a ∗ e.
(iii) Existence of the right inverse: For every a ∈ G there exists b ∈ G such that a ∗ b = e.

(c) A group is called abelian (or commutative), if for all a, b ∈ G we have a ∗ b = b ∗ a.

(d) A group is called finite if G has only finitely many elements; in this case its cardinality |G| is
called the order of G.
Example 1.2. (a) The usual addition N × N → N, (m, n) 7→ m + n, defines a binary operation on N
(and so does multiplication), but subtraction does not, because for instance 2 − 3 = −1 ̸∈ N.

(b) The sets Z, Q, R and C together with addition as the binary operation are abelian groups. The
sets Q∗ := Q \ {0}, R∗ := R \ {0} and C∗ := C \ {0} together with multiplication as the binary
operation are abelian groups.

(c) The set N with addition is not a group, because, for example, there doesn’t exist a (right) inverse
to 1 ∈ N. (In other words, the equation 1+? = 0 has no solution in N.)

(d) For any n ∈ N, the set Rn together with vector addition is a group. For any m, n ∈ N the set
Mm×n (R) of real m–by–n matrices together with matrix addition is a group. For every n ∈ N,
the set GLn (R) of invertible real (n × n)–matrices together with matrix multiplication is a group,
called the general linear group. If n > 1, GLn (R) is not abelian: e.g.
           
1 1 1 0 2 1 1 0 1 1 1 1
· = , but · = .
0 1 1 1 1 1 1 1 0 1 1 2

Proposition 1.3. Let G be a group. Then:

(a) There exists precisely one identity element in G.

(b) For every a ∈ G there exists precisely one right inverse (call it b) to a, and this b is also a left
inverse of a (meaning that we also have b ∗ a = e). We write a−1 for this inverse of a (or −a if the
group operation is “+”).

Proof. (a) Suppose e and e′ are two identity elements in G.

=⇒ e ∗ e′ = e (because e′ is an identity element)
and e ∗ e′ = e′ (because e is an identity element)
=⇒ e = e′ .
4 1 GROUPS

(b) Let b be a right inverse of a, and let c be a right inverse of b.

=⇒ a ∗ (b ∗ c) = a ∗ e (because c is a right inverse of b)
=a (because e is the identity element)
and a ∗ (b ∗ c) = (a ∗ b) ∗ c (by associativity)
= e∗c (because b is a right inverse of a)
=c (because e is the identity element)
=⇒ a = c
=⇒ b ∗ a = b ∗ c = e. (because c is a right inverse of b)
In other words, b is also a left inverse of a.
Suppose that b′ is another right inverse of a.
=⇒ b = b ∗ e = b ∗ (a ∗ b′ ) = (b ∗ a) ∗ b′ = e ∗ b′ = b′ .
(These equalities hold by: definition of e, definition of b′ , associativity, the fact we just proved,
and by definition of e, respectively.)

Proposition 1.4. Let G be a group.

(a) (Cancellation) If a, b, z ∈ G and a ∗ z = b ∗ z (or z ∗ a = z ∗ b), then a = b.

(b) For all a ∈ G both the equation a ∗ x = b and y ∗ a = b have a unique solution in G. (Another
way to say this: for every a ∈ G, both the map G → G, x 7→ a ∗ x (called left translation by a) and
G → G, y 7→ y ∗ a (called right translation by a) are bijective.)

(c) For every a ∈ G we have (a−1 )−1 = a.

(d) For all a, b ∈ G we have (a ∗ b)−1 = b−1 ∗ a−1 .

(e) (Exponential laws) For any m ∈ Z and a ∈ G, we define:



a ∗ a ∗{z· · · ∗ a} if m > 0;
|

m times
am :=

e if m = 0;
(a−1 )|m|

if m < 0.

(In additive notation, i.e. when the group operation is “+”, we write ma instead of am .)
Then for all m, n ∈ Z and a ∈ G we have am+n = am ∗ an and amn = (am )n .
If a, b ∈ G commute (i.e. a ∗ b = b ∗ a), then for all m ∈ Z we have (a ∗ b)m = am ∗ bm .

Proof. (a) Multiply both sides of the equation by z−1 .

(b) Proof of the “another way”: Injectivity: use (a).

Surjectivity: If b ∈ G, then b ∗ a−1 is mapped to b under the right translation by a.

(c) Both (a−1 )−1 and a are solutions of the equation a−1 ∗ x = e (see Proposition 1.3/(b)). Now apply
(b).

(d) We have (a∗b)∗(b−1 ∗a−1 ) = a∗(b∗(b−1 ∗a−1 )) = a∗((b∗b−1 )∗a−1 ) = a∗(e∗a−1 ) = a∗a−1 =
e. =⇒ b−1 ∗ a−1 is a right inverse to a ∗ b.

(e) Left as an exercise.

 1 GROUPS 5

Example 1.5. (a) The group table of a finite group G = {a1 , a2 , . . . , an } is a table like this:

∗ ··· aj ···
.. .. ..
. . .
ai ··· ai ∗ a j · · ·
.. ..
. .

The trivial group is the group with exactly one element, say e. Its group table must be

∗ e
e e

Any group with two elements, say e and a, is given by the group table:

∗ e a
e e a
a a e

Any group with three elements, say e, a and b, must have group table:

∗ e a b
e e a b
a a b e
b b e a

(Note that a ∗ b = a would imply b = e by 1.4/(a).) There are two “essentially different” groups of
order 4.
Note that Proposition 1.4/(b) implies that group tables must satisfy “sudoku rules”, i.e. that every
group element must appear in each row and each column exactly once. However, not every table
obeying this rule is a group table of a group; for example the table below does not. Why? (Hint:
what is a ∗ a ∗ b?)
e a b c d
a e c d b
b c d e a
c d a b e
d b e a c

(b) Let m ∈ N and define Cm := {0, 1, . . . , m − 1}. Define a binary operation on the set Cm by:
(
x+y if x + y < m;
x ⊕ y :=
x+y−m if x + y ≥ m.

Then Cm together with ⊕ is an abelian group called the cyclic group of order m. (Caveat: these
notes will use ⊕ for the group operation on Cm , to distinguish it from “+” between numbers.
However it is very common to just use “+” for the operation on Cm .)

Proof (that Cm is a group). (•) Associativity: Let x, y, z ∈ {0, 1, . . . , m − 1}. Want to show (x ⊕
y) ⊕ z = x ⊕ (y ⊕ z) in Cm .
- First case: Suppose that x + y + z < m. Then also x + y < m and y + z < m.
=⇒ LHS = x + y ⊕ z = (x + y) + z = x + (y + z) = x ⊕ y + z = RHS (using associativity
of addition of integers).
6 1 GROUPS

- Second case: Suppose m ≤ x + y + z < 2m.

Subcase (i): Suppose x + y < m.
=⇒ LHS = x + y ⊕ z = x + y + z − m.
Subcase (ii): Suppose x + y ≥ m.
=⇒ LHS = x + y − m ⊕ z = x + y + z − m.
Hence in both subcases we have LHS = x + y + z − m.
Similarly we obtain that also RHS = x + y + z − m = LHS.
- Third case: Suppose x + y + z ≥ 2m =⇒ x + y ≥ m and y + z ≥ m.
=⇒ LHS = x + y − m ⊕ z = x + y + z − 2m (since (x + y − m) + z ≥ m)
and RHS = x ⊕ y + z − m = x + y + z − 2m (since x + (y + z − m) ≥ m).
=⇒ LHS = RHS.
(•) Identity element: 0 is an identity element in Cm , because for any x ∈ Cm we have x ⊕0 = x + 0
= x = 0 + x = 0 ⊕ x.
(•) Existence of right inverses: Let x ∈ Cm . If x ̸= 0, then the inverse to x is m − x ∈ Cm . If x = 0,
then the inverse to x = 0 is 0.

(c) Let S be a set (such as S = {1, 2, . . . , n} for some n ∈ N) and let Sym(S) denote the set of all
bijective maps π : S → S, also called permutations of S.
For any π, σ ∈ Sym(S), we denote σ ◦ π their composition (as functions, so (σ ◦ π)(s) = σ (π(s))
for all s ∈ S). This defines a binary operation on Sym(S).
Then Sym(S) together with composition is a group, called the permutation group of S (or some-
times also the symmetric group of S).
The identity element in Sym(S) is the identity function (denoted idS or just id), and the inverse of
π ∈ Sym(S) is the inverse function π −1 (as in Calculus I).
If S = {1, . . . , n} for some
 n ∈ N, we write
 Sn for Sym(S) and use the “table notation” to describe
1 2 ··· n
permutations π ∈ Sn as π(1) π(2) ··· π(n) . For example:
 −1  
1 2 3 4 1 2 3 4
= in S4 ,
3 1 2 4 2 3 1 4
     
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
◦ = in S5 .
3 4 1 2 5 2 3 5 1 4 4 1 5 3 2

Definition 1.6. Let n ≥ 1 and s ≤ n. Let a1 , . . . , as ∈ {1, . . . , n} be pairwise distinct. The permutation
π ∈ Sn such that

π(a1 ) = a2 , π(a2 ) = a3 , ..., π(as−1 ) = as , π(as ) = a1 ,

and π(a) = a for a ∈ {1, . . . , n} \ {a1 , . . . , as },

is denoted by ⟨a1 , . . . , as ⟩. Any permutation of this form is called a cycle. If s = 2, it is called a transpo-
sition. The number s is called the length (or order) of the cycle.
   
1 2 3 4 5 6 1 2 3 4
For example, ⟨3, 1, 5, 2⟩ = in S6 ; and ⟨3, 2⟩ = in S4 .
5 3 1 4 2 6 1 3 2 4
Proposition 1.7. Every permutation σ ∈ Sn is a composition of cycles.
 
1 2 3 4 5 6 7 8 9 10 11
Example 1.8. (a) Let σ = ∈ S11 .
2 5 6 9 1 4 10 11 3 7 8
Then σ = ⟨1, 2, 5⟩ ◦ ⟨3, 6, 4, 9⟩ ◦ ⟨7, 10⟩ ◦ ⟨8, 11⟩.
 
1 2 3 4
(b) Let τ = ∈ S4 . Then τ = ⟨1, 4⟩.
4 2 3 1
 1 GROUPS 7

General Recipe (which, with a bit of effort, can be turned into a proof of 1.7).
Denote by σ ∈ Sn the permutation that we want to write as a composition of cycles.
Start with some a ∈ {1, . . . , n} such that σ (a) ̸= a (e.g. a := 1). (If there is no such a then σ = id and
we are done.)
Let m ∈ N, m > 1, be the smallest number such that σ m (a) ∈ {a, σ (a), σ 2 (a), . . . , σ m−1 (a)}. (Actually
then necessarily σ m (a) = a.)
Let σ1 be the cycle σ1 = ⟨a, σ (a), σ 2 (a), . . . , σ m−1 (a)⟩.
Now repeat the steps: take b ∈ {1, . . . , n} \ {a, σ (a), . . . , σ m−1 (a)} such that σ (b) ̸= b. Let l ∈ N, l > 1,
be the smallest number such that σ l (b) ∈ {b, σ (b), . . . , σ l−1 (b)}. Let σ2 = ⟨b, σ (b), . . . , σ l−1 (b)⟩.
Continuing in this way, we find a decomposition into cycles: σ = σ1 ◦ σ2 ◦ · · ·.

Definition 1.9. Let n ≥ 1 and σ ∈ Sn . We write σ = σ1 ◦ σ2 ◦ · · · as a composition of cycles of lengths

s1 , s2 , . . .. Then the number

sgn(σ ) := (−1)(s1 −1)+(s2 −1)+··· ∈ {±1}

is called the sign (or signum) of σ .

For example, with σ as in 1.8/(a), we have sgn(σ ) = (−1)2+3+1+1 = −1.

We have sgn(id) = 1, and if τ is any transposition, then sgn(τ) = −1.

Theorem 1.10. (a) The definition of sgn(σ ) does not depend on the chosen cycle decomposition of
σ.

(b) For all σ , τ ∈ Sn we have sgn(σ ◦ τ) = sgn(σ ) · sgn(τ).

In Group Theory in Year 2. (But for one possible proof for (a), see also an optional problem on one of
the Courseworks.)
8 2 FIELDS AND VECTOR SPACES

2 Fields and Vector Spaces

2.1 Fields

Definition 2.1. A field is a set F together with binary operations on F, which we will refer to as addition
and multiplication, such that:

• F together with addition is an abelian group (we use the notation a + b, 0 or 0F , −a), and
• F × := F \ {0} together with multiplication is an abelian group (we use the notation a · b or ab, 1,
a−1 ),

and such that the following axiom holds:

Distributivity: For all a, b, c ∈ F we have a(b + c) = ab + ac in F.
Example 2.2. (a) The sets Q, R and C with the usual addition and multiplication are fields (see also
1.2/(b)).
(b) The set Z with the usual addition and multiplication is not a field because for instance there is no
multiplicative inverse of 2 in Z.
(c) The set F2 := {0, 1} together with the following operations is a field.
+ 0 1 · 0 1
0 0 1 0 0 0
1 1 0 1 0 1
Note that 1 + 1 = 2 in Q but 1 + 1 = 0 in F2 . F2 is the smallest field.
Proof (that F2 is a field):
F2 with “+”, and F2 \ {0} = {1} with “·”, are abelian groups (see 1.5/(a)).
Distributivity: We need to check a(b + c) = ab + ac for all a, b, c ∈ F2 .
First case: a = 0 =⇒ LHS = 0, RHS = 0 + 0 = 0.
Second case: a = 1 =⇒ LHS = b + c = RHS. □
(d) (without proof) Let p be a prime. The set F p := {0, 1, . . . , p − 1} together with the addition defined
in Example 1.5/(b) and the following multiplication is a field:
x · y := remainder left when xy is divided by p.
(Why does this not work when p is not a prime, e.g. if p = 4?)
Proposition 2.3. Let F be a field. Then:

(a) For all a ∈ F we have 0a = 0.

(b) For all a, b ∈ F we have (−a)b = −(ab).

Proof.

(a) We have 0 + 0a = 0a = (0 + 0)a = 0a + 0a (0 is neutral for “+” and by distributivity)

=⇒ 0 = 0a. (cancel 0a on both sides using 1.4/(a))
(b) We have ab + (−a)b = (a + (−a))b (by distributivity)
= 0b (by definition of the additive inverse)
=0 (by part (a))
=⇒ (−a)b is the additive inverse of ab, i.e. (−a)b = −(ab).
 2 FIELDS AND VECTOR SPACES 9

2.2 Vector spaces

Definition 2.4. Let F be a field. A vector space over F is an abelian group V (we will use “+” for the
binary operation) together with a map F × V → V (called scalar multiplication and written as (a, x) 7→
ax), such that the following axioms are satisfied:

(i) 1st distributivity law: For all a, b ∈ F and x ∈ V we have (a + b)x = ax + bx in V .

(ii) 2nd distributivity law: For all a ∈ F and x, y ∈ V we have a(x + y) = ax + ay in V .

(iii) For all a, b ∈ F and for all x ∈ V we have (ab)x = a(bx) in V .

(iv) For all x ∈ V we have 1x = x in V .

The elements of V are called vectors. The elements of F will be referred to as scalars. We write 0F and
0V for the neutral elements of F and V , respectively, and often just 0 for both (when it is clear from the
context if it is a scalar or a vector). Furthermore we use the notation u − v for u + (−v) when u, v are
both vectors, or both scalars.
Example 2.5. (a) For every n ∈ N the set Rn together with the usual addition and scalar multiplication
(as seen in Linear Algebra I) is a vector space over R. Similarly, for any field F, the set

F n := {(a1 , . . . , an ) : a1 , . . . , an ∈ F}

together with component-wise addition and the obvious scalar multiplication is a vector space over
F. For example F22 = {(0, 0), (0, 1), (1, 0), (1, 1)} is a vector space over F2 ; F = F 1 is a vector
space over F, and finally F 0 := {0} is a vector space over F.

(b) Let V be the additive group of C. We view the usual multiplication R × V → V, (a, x) 7→ ax, as
scalar multiplication of R on V . Then V is a vector space over R. Similarly, we can think of C or
R as vector spaces over Q.

(c) Let V denote the abelian group R (with the usual addition). For a ∈ R and x ∈ V we put a ⊗ x :=
a2 x ∈ V ; this defines a scalar multiplication

R ×V → V, (a, x) 7→ a ⊗ x,

of the field R on V . Which of the vector space axioms (see 2.4) hold for V with this scalar multi-
plication?

Solution. (i) We need to check whether (a + b) ⊗ x = a ⊗ x + b ⊗ x for all a, b ∈ R and x ∈ V .

LHS = (a + b)2 x; RHS = a2 x + b2 x = (a2 + b2 )x
=⇒ For a = 1, b = 1 and x = 1 we have LHS ̸= RHS.
=⇒ First distributivity law does not hold.
(ii) We need to check whether a ⊗)(x + y) = a ⊗ x + a ⊗ y for all a ∈ R and x, y ∈ V .
LHS = a2 (x + y)
=⇒ LHS = RHS
RHS = a2 x + a2 y = a2 (x + y)
=⇒ Second distributivity law does hold.
(iii) We need to check whether a ⊗ (b ⊗ x) ) = (ab) ⊗ x for all a, b ∈ R and x ∈ V .
2 2 2
LHS = a ⊗ (b x) = a (b x)
=⇒ LHS = RHS
RHS = (ab)2 x = (a2 b2 )x = a2 (b2 x)
=⇒ Axiom (iii) does hold.
(iv) We have 1 ⊗ x = 12 x = x for all x ∈ V .
=⇒ Axiom (iv) does hold.
10 2 FIELDS AND VECTOR SPACES

Proposition 2.6. Let V be a vector space over a field F and let a, b ∈ F and x, y ∈ V . Then we have:

(a) (a − b)x = ax − bx

(b) a(x − y) = ax − ay

(c) ax = 0V ⇐⇒ a = 0F or x = 0V

(d) (−1)x = −x

Proof. (a) (a − b)x + bx = ((a − b) + b)x (by first distributivity law)

= (a + ((−b) + b)) x = (a + 0F )x = ax (using field axioms)
=⇒ (a − b)x = ax − bx. (add −bx to both sides)

(b) On Coursework.

(c) “=⇒”: On Coursework.

“⇐=”: Put a = b and x = y in (a) and (b), respectively.

(d) Put a = 0 and b = 1 in (a) and use (c).

The next example is the “mother” of almost all vector spaces. It vastly generalises the fourth of the
following five ways of representing vectors and vector addition in R3 .
 2 FIELDS AND VECTOR SPACES 11

0.5

0.0
z

a+b
−0.5
−1.0 1.0
0 0.8
1 0.6
x 0.4

y
2 a
0.2
3
(I) 0.0
(II) a = (2.5, 0, −1) b = (1, 1, 0.5) a + b = (3.5, 1, −0.5)
     
1 2 3 1 2 3 1 2 3
(III) a = b= a+b =
2.5 0 −1 1 1 0.5 3.5 1 −0.5

a : {1, 2, 3} → R b : {1, 2, 3} → R a + b : {1, 2, 3} → R

1 7→ 2.5 1 7→ 1 1 7→ 3.5
(IV)
2 7→ 0 2 7→ 1 2 7→ 1
3 7→ −1 3 7→ 0.5 3 7→ −0.5

x
3
x
2 function_name
x a
y

1 x xx x a+b
x x b
0 x
x
−1 x
1.0 1.5 2.0 2.5 3.0
(V) x
Example 2.7. Let S be any set and let F be a field. Let
F S := { f : S → F}
denote the set of all maps from S to F. We define an addition on F S and a scalar multiplication of F on
F S as follows: When f , g ∈ F S and a ∈ F we set:
( f + g)(s) := f (s) + g(s) for any s ∈ S
(a f )(s) := a f (s) for any s ∈ S.
Then F S is a vector space over F (see below for the proof).
Special Cases:
12 2 FIELDS AND VECTOR SPACES

(a) Let S = {1, . . . , n}. Identifying any map f : {1, . . . , n} → F with the corresponding tuple ( f (1), . . . , f (n)),
we see that F S can be identified with the set F n of all n-tuples (a1 , . . . , an ) considered in Example
2.5/(a).
(b) Let S = {1, . . . , n} × {1, . . . , m}. Identifying any map f : {1, . . . , n} × {1, . . . , m} → F with the
corresponding matrix:  
f ((1, 1)) . . . f ((1, m))
 .. .. 
 . . 
f ((n, 1)) . . . f ((n, m))
we see that F S can be identified with the set Mn×m (F) of (n × m)-matrices
 
a11 . . . a1m
 .. .. .. 
 . . . 
an1 . . . anm
with entries in F. In particular Mn×m (F) is a vector space over F.
(c) Let S = N. Identifying any map f : N → F with the sequence ( f (1), f (2), f (3), . . . ) we see that
F N can be identified with the set of all infinite sequences (a1 , a2 , a3 , . . . ) in F.
(d) Let F = R and let S be an interval I in R. Then F S = RI is the set of all functions f : I → R. (We
can visualise these functions via their graph, similarly as in (V) above.)

Proof (that F S is a vector space over F): First, F S with the above defined “+” is an abelian group:
Associativity: Let f , g, h ∈ F S .
We need to show: ( f + g) + h = f + (g + h) in F S
⇐⇒ (( f + g) + h)(s) = ( f + (g + h))(s) ) for all s ∈ S.
LHS = ( f + g)(s) + h(s) = ( f (s) + g(s)) + h(s)
(by definition of addition in F S )
RHS = f (s) + (g + h)(s) = f (s) + (g(s) + h(s)
=⇒ LHS = RHS (by associativity in F)
Identity element: Let 0 denote the constant function S → F, s 7→ 0F .
For any f ∈ F S and s ∈ S we have ( f + 0)(s) = f (s) + 0(s) = f (s) + 0F = f (s), hence f + 0 = f .
Similarly we have 0 + f = f . (using definitions of 0 and “+”, and field axioms)
=⇒ 0 is the identity element.
Inverses: Let f ∈ F S . Define (− f )(s) := − f (s).
For any s ∈ S we have ( f + (− f ))(s) = f (s) + (− f )(s) = f (s) + (− f (s)) = 0F = 0(s).
=⇒ f + (− f ) = 0 in F S , so − f is the inverse to f . (↑ defns of “+”, “− f ”, 0, and field axioms)
Commutativity: Let f , g ∈ F S .
For any s ∈ S we have ( f + g)(s) = f (s) + g(s) = g(s) + f (s) = (g + f )(s).
=⇒ f + g = g + f . (↑ by the definition of “+”, and commutativity of + in F)
Now the four axioms from Definition 2.4 (only (i) and (iii) spelled out here, the others are similar):
First distributivity law: Let a, b ∈ F and f ∈ F S . We want to check that (a + b) f = a f + b f :
For all s ∈ S we have
((a + b) f )(s) = (a + b)( f (s)) (by definition of the scalar multiplication)
= a( f (s)) + b( f (s)) (by distributivity in F)
= (a f )(s) + (b f )(s) (by definition of the scalar multiplication)
= (a f + b f )(s) (by definition of addition in F S )
=⇒ (a + b) f = a f + b f .
Axiom (iii): Let a, b ∈ F and f ∈ F S . We want to check that (ab) f = a(b f ).
For all s ∈ S we have
 2 FIELDS AND VECTOR SPACES 13

((ab) f )(s) = (ab)( f (s)) (by definition of scalar multiplication in F S )

= a(b( f (s))) (by associativity of multiplication in F)
= a((b f )(s)) (by definition of scalar multiplication in F S )
= (a(b f ))(s) (by definition of scalar multiplication in F S )
=⇒ (ab) f = a(b f ).

2.3 Subspaces

Definition 2.8. Let V be a vector space over a field F. A subset W of V is called a subspace of V if the
following conditions hold:

(a) 0V ∈ W .
(b) “W is closed under addition”: for all x, y ∈ W we also have x + y ∈ W .
(c) “W is closed under scalar multiplication”: for all a ∈ F and x ∈ W we have ax ∈ W .

Note that condition (b) states that the restriction of the addition in V to W gives a binary operation
W ×W → W on W (addition in W ).
Similarly, condition (c) states that the scalar multiplication of F on V yields a map F ×W → W which
we view as a scalar multiplication of F on W .

Proposition 2.9. Let V be a vector space over a field F and let W be a subspace of V . Then W together
with the above mentioned addition and scalar multiplication is a vector space over F.

Proof. The following axioms hold for W because they already hold for V :

• associativity of addition;
• commutativity of addition;
• all the four axioms in Definition 2.4.

There exists an additive identity element in W by condition 2.8/(a) (i.e. 0W := 0V ∈ W ).

It remains to show that additive inverses exist: Let x ∈ W . Then −x = (−1)x (see 2.6/(d)) is in W by
condition 2.8/(c); and −x satisfies x + (−x) = 0W = 0V because it does so in V .

Example 2.10. (a) Examples of subspaces of Rn as seen in Linear Algebra I, such as the nullspace
of any real (n × m)-matrix, or the column space of any real (m × n)-matrix.

(b) The set of convergent sequences is a subspace of the vector space RN of all sequences (a1 , a2 , a3 , . . . )
in R. A subspace of this subspace (and hence of RN ) is the set of all sequences in R that converge
to 0. (See Calculus I for proofs).

(c) Let A ∈ Ml×m (R). Then W := {B ∈ Mm×n (R) | AB = 0} is a subspace of Mm×n (R).
Proof :

(i) We have A · 0 = 0 =⇒ 0 ∈ W .
(ii) Let B1 , B2 ∈ W
=⇒ A(B1 + B2 ) = AB1 + AB2 = 0 + 0 = 0
=⇒ B1 + B2 ∈ W .
(iii) Let a ∈ R and B ∈ W
=⇒ A(aB) = a(AB) = a0 = 0
=⇒ aB ∈ W . □

(d) Let I be a non-empty interval in R. The following subsets of the vector space RI consisting of all
functions from I to R are subspaces:
14 2 FIELDS AND VECTOR SPACES

(i) For any s0 ∈ I the subset W := { f ∈ RI : f (s0 ) = 0} of RI .

Proof :
(1) The zero function 0 vanishes at s0 =⇒ 0 ∈ W .
(2) Let f , g ∈ W
=⇒ ( f + g)(s0 ) = f (s0 ) + g(s0 ) = 0 + 0 = 0
=⇒ f + g ∈ W .
(3) Let a ∈ R and f ∈ W
=⇒ (a f )(s0 ) = a · f (s0 ) = a · 0 = 0
=⇒ a f ∈ W . □
(ii) The set of all continuous functions f : I → R (see Calculus I).
(iii) The set of all differentiable functions f : I → R (see Calculus I).
(iv) For any n ∈ N, the set Pn of polynomial functions f : I → R of degree at most n, is a subspace
by 3.2/(c) and 3.3. A function f : I → R is a polynomial function of degree at most n if there
exists a0 , . . . , an ∈ R such that:

f (s) = a0 + a1 s + · · · + an sn for all s ∈ I.

Denoting the function I → R, s 7→ sm , by t m , this means that f = a0t 0 + a1t 1 + · · · + ant n as

elements of the vector space RI . (We will also use the more natural notation 1 for t 0 , and t
for t 1 .)
(v) The space of solutions of a homogeneous linear differential equation (without further expla-
nation); e.g.:

Pn = { f ∈ RI : f is differentiable (n + 1) times and f (n+1) = 0}

(e) The subset Zn of the vector space Rn over R is closed under addition but not closed under scalar
multiplication: For instance, (1, 0, . . . , 0) ∈ Zn and 21 ∈ R, but 12 (1, 0, . . . , 0) ∈
/ Zn .

(f) The subsets W1 := {(a, 0) : a ∈ R} and W2 := {(0, b) : b ∈ R} are subspaces of R2 . The subset
W := W1 ∪W2 of the vector space R2 is closed under scalar multiplication but not under addition
because, for instance, (1, 0) and (0, 1) are in W but (1, 0) + (0, 1) = (1, 1) ∈
/ W.

Proposition 2.11. Let W1 ,W2 be subspaces of a vector space V over a field F. Then the intersection
W1 ∩W2 and the sum of subspaces

W1 +W2 := {x1 + x2 ∈ V | x1 ∈ W1 , x2 ∈ W2 }

are subspaces of V as well.

Proof : For W1 ∩W2 :

(a) We have 0V ∈ W1 and 0V ∈ W2 (because W1 and W2 are subspaces)

=⇒ 0V ∈ W1 ∩W2 . (by definition of intersection)
(b) Let x, y ∈ W1 ∩W2
=⇒ x, y ∈ W1 and x, y ∈ W2 (by definition of intersection)
=⇒ x + y ∈ W1 and x + y ∈ W2 (because W1 and W2 are subspaces)
=⇒ x + y ∈ W1 ∩W2 . (by definition of intersection)
(c) Let a ∈ F and x ∈ W1 ∩W2
=⇒ x ∈ W1 and x ∈ W2 (by definition of intersection)
=⇒ ax ∈ W1 and ax ∈ W2 (because W1 and W2 are subspaces)
=⇒ ax ∈ W1 ∩W2 . (by definition of intersection)
 2 FIELDS AND VECTOR SPACES 15

For W1 +W2 :

(a) We have 0V = 0V + 0V ∈ W1 +W2 .

(b) Let x, y ∈ W1 +W2
=⇒ ∃x1 , y1 ∈ W1 and ∃x2 , y2 ∈ W2 with x = x1 + x2 , y = y1 + y2 (by definition of W1 +W2 )
=⇒ x + y = (x1 + x2 ) + (y1 + y2 ) =
= (x1 + y1 ) + (x2 + y2 ) ∈ W1 +W2 . (because W1 and W2 are subspaces)
(c) Let a ∈ F and x ∈ W1 +W2
=⇒ ∃x1 ∈ W1 , x2 ∈ W2 such that x = x1 + x2 (by definition of W1 +W2 )
=⇒ ax = a(x1 + x2 ) = ax1 + ax2 ∈ W1 +W2 . (because W1 and W2 are subspaces) □

Example 2.12. Let W1 and W2 be as in 2.10/(f). Then W1 +W2 = R2 .

16 3 BASES

3 Bases

3.1 Spanning

Definition 3.1. Let V be a vector space over a field F. Let x1 , . . . , xn ∈ V .

(a) An element x ∈ V is called a linear combination of x1 , . . . , xn if there are a1 , . . . , an ∈ F such that

x = a1 x1 + · · · + an xn .

(b) The subset of V consisting of all linear combinations of x1 , . . . , xn is called the span of x1 , . . . , xn
and is denoted by Span(x1 , . . . , xn ) (or SpanF (x1 , . . . , xn )); i.e.

Span(x1 , . . . , xn ) = {a1 x1 + · · · + an xn | a1 , . . . , an ∈ F}.

(c) We say that x1 , . . . , xn span V , or that x1 , . . . xn form a spanning set of V , if V = Span(x1 , . . . , xn ),

i.e. every x ∈ V is a linear combination of x1 , . . . , xn .

(See also Section 6.3 of L.A.I.)

Example 3.2. (a) Let V := Mn×m (F) be the vector space of (n × m)-matrices with entries in F. For
i ∈ {1, . . . , n} and j ∈ {1, . . . , m}, let Ei j denote the (n × m)-matrix with zeroes everywhere except
at (i j) where it has the entry 1. Then the matrices Ei j ; i = 1, . . . , n; j = 1, . . . , m form a spanning
set of V .
Proof : Let A = (ai j ) ∈ Mn×m (F) be an arbitrary matrix. Then A = ∑ni=1 ∑mj=1 ai j Ei j .
2 3 = 2 1 0 + 3 0 1 + (−1) 0 0 + 5 0 0 = 2E + 3E + (−1)E + 5E .






For example: −1 5 00 00 10 01 11 12 21 22
   
1 i
(b) Do the vectors , ∈ C2 span the vector space C2 over C?
i 2
 
w
Solution. Let ∈ C2 be an arbitrary vector.
z
     
1 i w
We want to know if we can find a1 , a2 ∈ C such that a1 + a2 = .
i 2 z
   
1 i w R27→R2−iR1 1 i w
Hence −−−−−−−→ .
i 2 z 0 3 z − iw
As in  Linear
 Algebra
 I we conclude that this system is solvable. (Theorem 3.15 of L.A.I.)
1 i
Thus , span C2 .
i 2

(c) Pn = Span(t 0 ,t 1 ,t 2 , . . . ,t n ) = Span(1,t,t 2 , . . . ,t n ) (c.f. 2.10/(d)/(iv)).

Proposition 3.3. Let V be a vector space over a field F. Let x1 , . . . , xn ∈ V . Then Span(x1 , . . . , xn ) is the
smallest subspace of V that contains x1 , . . . , xn . Furthermore:

(a) If x ∈ Span(x1 , . . . , xn ) then Span(x1 , . . . , xn , x) = Span(x1 , . . . , xn ).

(b) For any a2 , . . . , an ∈ F we have Span(x1 , . . . , xn ) = Span(x1 , x2 − a2 x1 , . . . , xn − an x1 ).

Notes. • The first statement means:

(i) Span(x1 , . . . , xn ) is a subspace of V that contains x1 , . . . , xn , and

(ii) among all the subspaces of V with this property it is the smallest;
in other words if W is a subspace of V that contains x1 , . . . , xn , then Span(x1 , . . . xn ) ⊆ W .
 3 BASES 17

• The statement (b) implies that the span of the column vectors of any matrix does not change when
performing (standard) column operations.

Proof. Span(x1 , . . . , xn ) is a subspace of V :

(i) We have 0V = 0F x1 + · · · + 0F xn ∈ Span(x1 , . . . , xn ).

(ii) Let x, y ∈ Span(x1 , . . . , xn ).

=⇒ ∃a1 , . . . , an ∈ F, ∃b1 , . . . bn ∈ F such that x = a1 x1 + · · · + an xn and y = b1 x1 + · · · + bn xn ;
=⇒ x + y = (a1 x1 + · · · + an xn ) + (b1 x1 + · · · + bn xn )
= (a1 x1 + b1 x1 ) + · · · + (an xn + bn xn ) (using commutativity and associativity)
= (a1 + b1 )x1 + · · · + (an + bn )xn (using first distributivity law)
∈ Span(x1 , . . . , xn ) (by definition of Span)

(iii) Let x ∈ Span(x1 , . . . , xn ) and a ∈ F.

Write x = a1 x1 + . . . an xn with a1 , . . . , an ∈ F as above.
=⇒ ax = a(a1 x1 + · · · + an xn ) = a(a1 x1 ) + · · · + a(an xn ) (using distributivity)
= (aa1 )x1 + · · · + (aan )xn (by axiom 2.4/(iii))
∈ Span(x1 , . . . , xn ) (by definition of Span)

Span(x1 , . . . , xn ) contains x1 , . . . , xn :
. . . because xi = 0F · x1 + . . . 0F · xi−1 + 1 · xi + 0F · xi+1 + · · · + 0F · xn ∈ Span(x1 , . . . , xn ).
Span(x1 , . . . , xn ) is the smallest:
Let W be a subspace of V such that x1 , . . . , xn ∈ W .
Let x ∈ Span(x1 , . . . , xn ). Write x = a1 x1 + · · · + an xn with a1 , . . . an ∈ F.
=⇒ a1 x1 , . . . , an xn ∈ W (by condition 2.8/(c))
=⇒ x = a1 x1 + · · · + an xn ∈ W . (by condition 2.8/(b))
=⇒ Span(x1 , . . . , xn ) ⊆ W .
Part (a): Span(x1 , . . . , xn ) ⊆ Span(x1 , . . . , xn , x) =: W
(because W is a subspace of V and x1 , . . . , xn ∈ W )
Span(x1 , . . . , xn , x) ⊆ Span(x1 , . . . , xn ) =: W
e
e is a subspace of V and x1 , . . . , xn , x ∈ W
(because W e)
Part (b): Span(x1 , . . . , xn ) ⊆ Span(x1 , x2 − a2 x1 , . . . , xn − an x1 ) =: W
(because W is a subspace of V and x1 , . . . , xn ∈ W )
Span(x1 , x2 − a2 x1 , . . . , xn − an x1 ) ⊆ Span(x1 , . . . , xn ) =: W
e
(because W e is a subspace of V and x1 ∈ W e and for i = 2, . . . , n also xi − ai x1 ∈ W
e)

3.2 Linear independence

Definition 3.4. Let V be a vector space over a field F. Let x1 , . . . , xn ∈ V . We say that x1 , . . . , xn are
linearly independent (over F) if the following condition holds:
if a1 , . . . , an ∈ F and a1 x1 + · · · + an xn = 0V then a1 = · · · = an = 0F .
Otherwise we say that x1 , . . . , xn are linearly dependent. A linear combination a1 x1 + · · · + an xn is called
trivial if a1 = · · · = an = 0, otherwise it is called non-trivial. (See also Section 6.5 of L.A.I.)

Note. x1 , . . . , xn are linearly dependent ⇐⇒ ∃a1 , . . . , an ∈ F, not all zero, such that a1 x1 + · · · + an xn =
0V . In other words, ⇐⇒ there exists a non-trivial linear combination of x1 , . . . , xn which equals 0V .

Example 3.5. (a) Examples as seen in Linear Algebra I. (Section 6.5 of L.A.I.)
18 3 BASES

     
1 1 0
(b) The three vectors x1 = 0, x2 = 1, x3 = −1 ∈ F 3 are not linearly independent because
1 0 1
x1 − x2 − x3 = 0.
       
c1 1 0 c1
  3
(c) Determine all vectors c2 ∈ C such that x1 := i , x2 := 1 , x3 := c2  ∈ C3 are lin-
    
c3 1 0 c3
early dependent.

Solution. We apply Gaussian elimination:

   
1 0 c1 1 0 c1
 i 1 c2  −R37 →R3−R1
−−−−−−→ 0 1 c2 − ic1 
R27→R2−iR1
1 0 c3 0 0 c3 − c1

=⇒ The equation a1 x1 + a2 x2 + a3 x3 = 0 has a non-trivial solution (a1 , a2 , a3 )

if and only if c3 − c1 = 0.
=⇒ x1 , x2 , x3 are linearly dependent if and only if c1 = c3 .
(d) The two functions sin, cos ∈ RR are linearly independent.

Proof. Let a, b ∈ R such that a sin +b cos = 0 in RR .

=⇒ ( For all s ∈ R we have a · sin(s) + b · cos(s) = 0(s) = 0.
a · sin(0) + b · cos(0) = 0 =⇒ a · 0 + b · 1 = 0 =⇒ b = 0
=⇒
a · sin( π2 ) + b · cos( π2 ) = 0 =⇒ a · 1 + b · 0 = 0 =⇒ a = 0

(e) Let I ⊆ R be a non-empty open interval. Recall from 2.10/(d)/(iv) that for any i ∈ N0 , we denote
t i the polynomial function I → R, s 7→ si .
The vectors t 0 ,t 1 ,t 2 , . . . ,t n are linearly independent in RI .
Proof : Let a0 , . . . , an ∈ R such that a0t 0 + a1t 1 + · · · + ant n = 0
=⇒ a0 + a1 s + · · · + an sn = 0 for all s ∈ I
=⇒ a0 = · · · = an = 0, because any non-zero real polynomial of degree n has at most n real
roots. (This follows from the Fundamental Theorem of Algebra. Alternatively, it can be proved
by induction and using long division.)
Proposition 3.6. Let V be a vector space over a field F.

(a) A single vector x ∈ V is linearly independent if and only if x ̸= 0V .

(b) Every subset of any set of linearly independent vectors is linearly independent again. (This is
equivalent to: If a subset of a set of vectors is linearly dependent, then the set itself is linearly
dependent.)

(c) Let x1 , . . . , xn ∈ V and suppose that xi = 0V for some i ∈ {1, . . . , n}, or that xi = x j for some i ̸= j.
Then x1 , . . . , xn are linearly dependent.

(d) If x1 , . . . , xn ∈ V are linearly dependent then at least one vector xi among x1 , . . . , xn is a linear
combination of the other ones.

(e) Let x1 , . . . , xn ∈ V and x ∈ Span(x1 , . . . , xn ). Then x1 , . . . , xn , x are linearly dependent.

Proof. (a) “⇒”: If x = 0V then 1 · x = 0V is a non-trivial linear combination of x.

“⇐”: Let x ̸= 0V and let a ∈ F such that ax = 0V . Then a = 0 by 2.6/(c).
 3 BASES 19

(b) Let x1 , . . . , xn ∈ V be linearly independent and let y1 , . . . , ym be a subset of x1 , . . . , xn for some

m ≤ n.
Proceed by contradiction: Suppose that ∃b1 , . . . , bm ∈ F, not all zero, such that b1 y1 +· · ·+bm ym =
0V .
Extending the list b1 , . . . , bm by zeroes we get a1 , . . . , an ∈ F, not all zero, such that a1 x1 + · · · +
an xn = 0V .
So x1 , . . . , xn are linearly dependent, a contradiction.
(c) If xi = 0V then x1 , . . . , xn are linearly dependent by (a) and (b).
If xi = x j for some i ̸= j, then xi , x j are linearly dependent, because 1xi + (−1)x j = 0V . Now apply
(b).
(d) ∃a1 , . . . , an ∈ F, not all zero, such that a1 x1 + · · · + an xn = 0V .
After reordering we may assume an ̸= 0.
=⇒ xn = −a−1 −1 −1
n (a1 x1 + · · · + an xn ) = (−an a1 )x1 + · · · + (−an an−1 )xn−1
=⇒ xn is a linear combination of x1 , . . . , xn−1 .
(e) Let x ∈ Span(x1 , . . . , xn ).
=⇒ ∃a1 , . . . , an ∈ F, such that x = a1 x1 + · · · + an xn .
=⇒ 0V = a1 x1 + · · · + an xn + (−1)x
=⇒ We have a a non-trivial linear combination of x1 , . . . , xn , x which equals 0V .
(Because −1 ̸= 0 in any field.)

3.3 Bases

Definition 3.7. Let V be a vector space over a field F. Let x1 , . . . , xn ∈ V . We say that x1 , . . . , xn form a
basis of V if x1 , . . . , xn both span V and are linearly independent. (Compare Definition 6.40 of L.A.I.)
Example 3.8. (a) Let F be a field. The vectors e1 := (1, 0, . . . , 0); . . . ; en := (0, . . . , 0, 1) form a basis
F n , called the standard basis of F n (as in Linear Algebra I (Ex 6.41(a)).
(b) The polynomials 1,t, . . . ,t n form a basis of Pn . (see 2.10/(d)/(iv), 3.2/(c)).
(c) 1, i form a basis of the vector space C over R (cf 2.5/(b)).
(d) Determine a basis of the nullspace N(A) ⊆ R4 of the matrix
 
1 −1 3 2
 2 −1 6 7 
A :=   ∈ M4×4 (R).
 3 −2 9 9 
−2 0 −6 −10

Solution. We perform Gaussian elimination until the reduced lower echelon form (row opera-
tions):
   
1 −1 3 2 1 −1 3 2
 2 −1 6 7  R27→R2−2R1 0
  1 0 3
A=  3 −2 9 −−−−−−−→ 
9  R37→R3−3R1 0 1 0 3
R47→R4+2R1
−2 0 −6 −10 0 −2 0 −6
 
1 0 3 5
R17→R1+R2  0 1 0 3
−−−−−−−→   =: A
e
R37→R3−R2  0 0 0 0
R47→R4+2R2
0 0 0 0
20 3 BASES

Because performing row operations does not change the nullspace of a matrix (see Note below),
we have:
e = {x ∈ R4 : Ax
N(A) = N(A) e = 0}
  

 x1 

x
  
2 4
 ∈ R : x1 = −3x3 − 5x4 ; x2 = −3x4
=  x 
 3

 

x4

  

 −3x3 − 5x4 

−3x
 
4

=    : x 3 , x 4 ∈ R

 x3  

x4
 
     

 −3 −5 

  
0 −3 
= x3   + x4   : x3 , x4 ∈ R
 1 0 
 
0 1
 
   
−3 −5
 0  −3
= Span 
 1  ,  0  .
   

0 1

(Denote u1 = (−3, 0, 1, 0)T , u2 = (−5, −3, 0, 1)T ).

Also u1 and u2 are also linearly independent (because they are not multiples of each other).
=⇒ u1 and u2 form a basis of N(A).

(e) Let A ∈ Mn×n (R). Then:

A invertible ⇐⇒ The columns of A form a basis of Rn (for a proof see Theorem 5.6).

Note. Row operations do not change the nullspace of a matrix (say A). This is because vectors x are in
N(A) exactly if they are solutions to Ax = 0, i.e. solutions to the homogeneous system of linear equations
described by the matrix A. Row operations on A correspond to the “allowed” operations on the linear
system of equations.

Proposition 3.9. Let V be a vector space over a field F. Let x1 , . . . , xn ∈ V . The following statements
are equivalent:

(a) x1 , . . . , xn form a basis of V .

(b) x1 , . . . , xn form a minimal spanning set of V (i.e. x1 , . . . , xn span V and after removing any vector
from x1 , . . . , xn the remaining ones don’t span V anymore). (Compare Def 6.23 of L.A.I.)

(c) x1 , . . . , xn form a maximal linearly independent subset of V (i.e. x1 , . . . , xn are linearly independent
and for any x ∈ V the n + 1 vectors x1 , . . . , xn , x are linearly dependent).

(d) Every vector x ∈ V can be written in the form

x = a1 x1 + · · · + an xn

with coefficients a1 , . . . , an ∈ F uniquely determined by x.

Proof.
“(a) =⇒ (b)”: 1/Spanning: x1 , . . . , xn span V by definition of a basis.
2/Minimality: Suppose that the spanning set x1 , . . . , xn is not minimal.
 3 BASES 21

=⇒ After reordering we may assume that x1 , . . . , xn−1 span V .

=⇒ xn ∈ V = Span(x1 , . . . , xn−1 ).
=⇒ x1 , . . . , xn−1 , xn are linearly dependent (by 3.6/(e)). Contradiction.
(b) =⇒ (c): 1/Independence: Suppose that x1 , . . . , xn are linearly dependent.
=⇒ After reordering we have xn ∈ Span(x1 , . . . , xn−1 ) (by 3.6/(d))
=⇒ V = Span(x1 , . . . , xn ) = Span(x1 , . . . , xn−1 ) (by 3.3/(a))
This contradicts the minimality assumed in (b).
2/Maximality: Let x ∈ V = Span(x1 , . . . , xn )
=⇒ x1 , . . . , xn , x are linearly dependent (by 3.6/(c)).
(c) =⇒ (d): 1/Existence: Let x ∈ V .
=⇒ ∃b1 , . . . , bn , b ∈ F, not all zero, with b1 x1 + · · · + bn xn + bx = 0.
(because x1 , . . . , xn , x are linearly dependent)
=⇒ b ̸= 0 (because x1 , . . . , xn are linearly independent)
−1
=⇒ x = a1 x1 + · · · + an xn , where ai := −b bi .
2/Uniqueness: Suppose x = a1 x1 + · · · + an xn = b1 x1 + · · · + bn xn for some a1 , . . . , an , b1 , . . . , bn ∈ F.
=⇒ 0V = x − x = (a1 − b1 )x1 + · · · + (an − bn )xn
=⇒ a1 = b1 , . . . , an = bn . (because x1 , . . . , xn are linearly independent)
(d) =⇒ (a): 1/Spanning: Directly from (d).
2/Independence: Let a1 , . . . , an ∈ F such that a1 x1 + . . . an xn = 0V .
=⇒ a1 = · · · = an = 0. (from uniqueness, because also 0x1 + · · · + 0xn = 0V )

Corollary 3.10. Let V be a vector space over a field F. Suppose V = Span(x1 , . . . , xn ) for some
x1 , . . . xn ∈ V . Then a subset of x1 , . . . , xn forms a basis of V . In particular V has a basis.

Proof. By successively removing vectors from x1 , . . . , xn we arrive at a minimal spanning set y1 , . . . , ym

for some m ≤ n. Then y1 , . . . , ym form a basis of V (by 3.9 (b) =⇒ (a)).

3.4 Dimension

Theorem 3.11 (This Theorem allows us to define the dimension of a vector space.). Let V be a vector
space over a field F. Suppose x1 , . . . , xn and y1 , . . . , ym both form a basis of V . Then m = n. (Compare
Thm 6.44 from L.A.I.)

Definition 3.12. Let V be a vector space over a field F. If the vectors x1 , . . . , xn ∈ V form a basis of V ,
we say that V is of finite dimension, and call n the dimension of V . We write dimF (V ) or just dim(V )
for n. Note that n does not depend on the chosen basis x1 , . . . , xn by 3.11.

Example 3.13. (a) dimF (F n ) = n (by Example 3.8/(a)).

(b) dimR (Pn ) = n + 1 (by Example 3.8/(b)).

(c) dimR (C) = 2 (by Example 3.8/(c)).

(d) dimC (C3 ) = 3, dimR (C3 ) = 6. In general dimC (Cn ) = n, dimR (Cn ) = 2n.

(e) R as a vector space over Q is not finite dimensional (see 2.5/(b)).

(f) dimR (Mn×m (R)) = nm.

√
(g) dimQ (Q( −3)) = 2 (see Coursework 3).

(h) About dimF (Span(x1 , . . . , xn )). We determine its dimension by finding a basis, i.e. subset of
x1 , . . . , xn which still spans Span(x1 , . . . , xn ), and it is linearly independent. For example:
22 3 BASES

Let x1 ∈ V, x1 ̸= 0 =⇒ dimF (Span(x1 )) = 1.

Let x2 be another vector in V(.
2 if x1 , x2 are linearly independent
=⇒ dimF (Span(x1 , x2 )) =
1 if x1 , x2 are linearly dependent

Proof of Theorem 3.11. : It follows from the following Proposition.

Proposition 3.14. Let V be a vector space over a field F. Let x1 , . . . , xn ∈ V and y1 , . . . , ym ∈ V .

Suppose x1 , . . . , xn span V . If y1 , . . . , ym are linearly independent then m ≤ n.

Proof. We will show the contrapositive:

• If m > n then there exist c1 , . . . , cm ∈ F, not all zero, such that c1 y1 + · · · + cm ym = 0.

For every i ∈ {1, . . . , m} we can write yi = ai1 x1 + · · · + ain xn for some ai1 , . . . ain ∈ F.
(because x1 , . . . , xn span V )
=⇒ For all c1 , . . . , cm ∈ F we have: (using the axioms of a vector space)

c1 y1 + · · · + cm ym = c1 (a11 x1 + · · · + a1n xn ) + · · · + cm (am1 x1 + · · · + amn xn )

= (a11 c1 + · · · + am1 cm )x1 + · · · + (a1n c1 + · · · + amn cm )xn

=⇒ It suffices to show that the system of linear equations

a11 c1 + a21 c2 + · · · + am1 cm = 0

..
.
a1n c1 + a2n c2 + · · · + amn cm = 0

has a solution (c1 , . . . , cm ) ∈ F m different from (0, . . . , 0).

This follows from Gaussian elimination (as seen in Linear Algebra I for F = R (Thm 3.15(b) from
L.A.I.)): since m > n, we have more unknowns than equations.

Corollary 3.15 (Two-out-of-three basis criterion). Let V be a vector space over a field F. Let x1 , . . . xn ∈
V . Suppose two of the following three statements hold. Then x1 , . . . , xn form a basis:

(a) x1 , . . . , xn are linearly independent.

(b) x1 , . . . , xn span V .
(c) n = dimF (V ).

Proof. If (a) and (b) hold, then x1 , . . . , xn form a basis by definition.

Suppose (a) and (c) hold. If x1 , . . . , xn would not form a basis we could find an x ∈ V such that x1 , . . . , xn , x
are still linearly independent (by 3.9 (a) ⇐⇒ (c)). =⇒ n + 1 ≤ n by Proposition 3.14. Contradiction.
Suppose (b) and (c) hold. After reordering we may assume that x1 , . . . , xm form a minimal spanning set,
for some m ≤ n.
=⇒ x1 , . . . , xm form a basis (by 3.9 (a) ⇐⇒ (b)).
=⇒ m = n (by 3.11); i.e. x1 , . . . , xn form a basis.
     
1 −2 1
Example 3.16. (a) The vectors x1 := 2 , x2 :=    1 , x3 := 0 form a basis of the vector
 
3 0 1
space Q over Q, of the vector space R over R and of the vector space C3 over C.
3 3
 3 BASES 23

Proof : We first show that x1 , x2 , x3 are linearly independent over C:

Let c1 , c2 , c3 ∈ C such that c1 x1 + c2 x2 + c3 x3 = 0.
    
1 −2 1 c1 0
=⇒ 2 1 0 c2  = 0 .
3 0 1 c3 0

Gaussian elimination yields:

     
1 −2 1 1 −2 1 6 1 −2 1
R27→R2−2R1 R37→R3− R2
A := 2 1 0 −−−−−−−→ 0 5 −2 −−−−−−5−→ 0 5 −2 
R37→R3−3R1
3 0 1 0 6 −2 0 0 2/5

=⇒ c = (c1 , c2 , c3 ) = (0, 0, 0) is the only solution of Ac = 0.

=⇒ x1 , x2 , x3 are linearly independent over C and then also over R and Q.
=⇒ x1 , x2 , x3 form a basis of C3 , R3 and Q3 over the respective fields.
(by 3.15 since 3 = dimC (C3 ) = dimR (R3 ) = dimQ (Q3 )) □

(b) We view C as a vector space√ over C, R and Q √(see 2.5/(b)).

Let x1 := 1, x2 := 2, x3 := 2, x4 := i, x5 := i 3 ∈ C.
Determine dimF (SpanF (x1 , x2 , x3 , x4 , x5 )) for F = C, R and Q.
Solution. For F = C:
We have C = SpanC (x1 ) ⊆ SpanC (x1 , . . . , x5 ) ⊆ C.
=⇒ dimC (SpanC (x1 , . . . , x5 )) = dimC (C) = 1.
For F = R:
x1 and x4 span C as a vector space over R.
=⇒ SpanR (x1 , . . . , x5 ) = C.
=⇒ dimR (Span(x1 , . . . , x5 )) = dimR (C) = 2.
For F = Q: Observations:
x1 , x2 are LD over Q =⇒ SpanQ (x1 , . . . , x5 ) = SpanQ (x1 , x3 , x4 , x5 ).
Also x1 , x3 are LI over Q; x1 , x4 are LI over R.
⇝ Let us try to prove that x1 , x3 , x4 , x5 are linearly independent over Q.
a5 ∈ Q be such √
Let a1 , a3 , a4 ,√ that a1 x1 + a3 x3 + a4 x4 + a5 x5 = 0.
=⇒ (a1 + a3√ 2) + i(a4 + a5 3) √= 0.
=⇒ a1 + a3 2 = 0 and a4 + a5 3 = 0 (because 1 and i are √ linearly independent over Q)
a1
=⇒ a1 = a3 = 0 (If a3 ̸= 0 =⇒ 2 = − a3 ∈ Q. Contradiction.)
and a4 = a5 = 0 (similarly).
=⇒ x1 , x3 , x4 , x5 are linearly independent, so form a basis of SpanQ (x1 , x3 , x4 , x5 ).
=⇒ dimQ (SpanQ (x1 , x2 , x3 , x4 , x5 )) = dimQ (SpanQ (x1 , x3 , x4 , x5 )) = 4.

(c) Let V be a vector space of finite dimension over a field F and let W be a subspace of V . Then
dimF (W ) ≤ dimF (V ).

Proof : If vectors are L.I. in W , they are also L.I. in V . (by def. of L.I.)
=⇒ Any L.I. subset of W has at most dim(V ) elements (use 3.14 and 3.9 (a) ⇐⇒ (c))
=⇒ dim(W ) ≤ dim(V ) (by 3.9 (a) ⇐⇒ (c))
24 4 LINEAR TRANSFORMATIONS

4 Linear Transformations

Let F be a field (e.g. F = R, Q, C or F2 ).

Definition 4.1. An m × n-matrix A over a field F is an array
 
a11 . . . a1n
A =  ... .. 

. 
am1 . . . amn

with entries ai j in F. We use the notation Mm×n (F) for the set of all (m × n)-matrices over F (see also
2.7/(b)). We define addition and multiplication of matrices (and other notions) in the same way as in the
case F = R (as seen in Linear Algebra I).

For example:
      
1 1+i 1−i 1 − i + 3 + 3i 4 + 2i
• = = (matrices over C)
2 1−i 3 2 − 2i + 3 − 3i 5 − 5i
    
0 1 1 1 0 1
• = (as matrices over F2 )
1 1 0 1 1 0
    
0 1 1 1 0 1
• but = (as matrices over R)
1 1 0 1 1 2

Definition 4.2. Let V,W be vector spaces over a field F. A map L : V → W is called a linear transfor-
mation if the following two conditions hold:

(a) For all x, y ∈ V we have L(x + y) = L(x) + L(y) in W .

(b) For all a ∈ F and x ∈ V we have L(ax) = a(L(x)) in W .
Note. Then we also have L(0V ) = 0W and L(x − y) = L(x) − L(y) for all x, y ∈ V .

Proof. L(0V ) = L(0V + 0V ) = L(0V ) + L(0V )

=⇒ L(0V ) = 0W . (by cancelling L(0V ))
L(x − y) = L(x + (−1)y) = L(x) + L((−1)y) = L(x) + (−1)L(y) = L(x) − L(y). (using 2.6/(d))

Example 4.3. (a) Let A ∈ Mm×n (F). Then the map

LA :F n → F m
   
x1 a11 x1 + · · · + a1n xn
x =  ...  7→ Ax =  ..
   
. 
xn am1 x1 + · · · + amn xn

is a linear transformation.
 (Compare
 with Lemma 5.3 in L.A.I.)
a 0
For example, if A = ∈ M2×2 (R) for some a ∈ R then LA : R2 → R2 is given by x 7→ ax,
0 a
i.e. it isa stretch of the plane
 by a factor of a.
cos(φ ) sin(φ )
If A = for some 0 ≤ φ < 2π then LA : R2 → R2 is the clockwise rotation by
− sin(φ ) cos(φ )
the angle φ .

Proof that LA is a linear transformation.

1/ Let x, y ∈ F n .
=⇒ LA (x + y) = A(x + y) = Ax + Ay = LA (x) + LA (y).
 4 LINEAR TRANSFORMATIONS 25

2/ Let a ∈ F and x ∈ F n .
=⇒ LA (ax) = A(ax) = a(Ax) = a(LA (x)).
(The middle equality of both chains of equalities has been proved in Linear Algebra I for F = R,
see Thm 2.13(i) and (ii), the same proof works for any field F.)

(b) Let V be a vector space over a field F. Then the following maps are linear transformations (cf. Ex-
ample 5.4(c),(d) in L.A.I.):

• id: V → V , x 7→ x (identity)
• 0: V → V , x 7→ 0V (zero map)
• the map V → V , given by x 7→ ax, for any given a ∈ F fixed (stretch)

(c) Let L : V → W and M : W → Z be linear transformation between vector spaces over a field F.
Then their composition M ◦ L : V → Z is again a linear transformation. (See also Section 5.3 of
L.A.I.)

Proof that M ◦ L is a linear transformation.

1/ Let x, y ∈ V .
=⇒ (M ◦ L)(x + y) = M(L(x + y)) = M(L(x) + L(y))
= M(L(x)) + M(L(y)) = (M ◦ L)(x) + (M ◦ L)(y).
2/ Let a ∈ F and x ∈ V .
=⇒ (M ◦ L)(ax) = M(L(ax)) = M(a(L(x))) = a(M(L(x))) = a(M ◦ L)(x).

(d) Let V be the subspace of RR consisting of all differentiable functions. Then differentiation D :
V → RR , f 7→ f ′ , is a linear transformation.

Proof.
1/ Let f , g ∈ V =⇒ D( f + g) = ( f + g)′ = f ′ + g′ = D( f ) + D(g).
2/ Let a ∈ R and f ∈ V =⇒ D(a f ) = (a f )′ = a f ′ = a(D( f )).
(The middle equality in both chains of equalities has been proved in Calculus.)
 
x
(e) The map L : R2 → R, 1 7→ x1 x2 , is not a linear transformation.
x2
 
1
Proof. Let a = 2 and x = ∈ R2 . Then:
1
 !
2
L(ax) = L = 4, but aL(x) = 2 · 1 = 2.
2

4.1 Matrix representation I

Proposition 4.4 (Matrix representation I). Let F be a field. Let L : F n → F m be a linear transformation.
Then there exists a unique matrix A ∈ Mm×n (F) such that L = LA (as defined in 4.3/(a)). In this case we
say that A represents L (with respect to the standard bases of F n and F m ).
(See also Theorem 5.6 of L.A.I.)
 
c1    
3 2 2c 1 + c3 − 4c 2 2 −4 1
For example, the map R → R , c2  7→ , is represented by A = ∈
c2 0 1 0
c3
M2×3 (R).
26 4 LINEAR TRANSFORMATIONS

   
1 0
0  .. 
Proof. Let e1 :=  .  , . . . , en :=  .  denote the standard basis of F n .
   
.
. 0
0 1
Uniqueness: Suppose A ∈ Mm×n (F) satisfies L = LA .
=⇒ The jth column of A is Ae j = LA (e j ) = L(e j ) (for j = 1, . . . , n)
=⇒ A is the (m × n)-matrix with the column vector L(e1 ), . . . , L(en ).

 Let
Existence:  A be defined this way. We want to show L = LA .
c1
 .. 
Let c =  .  ∈ F n =⇒ c = c1 e1 + · · · + cn en ;
cn
=⇒ L(c) = L(c1 e1 ) + · · · + L(cn en ) = c1 L(e1 ) + · · · + cn L(en )
and LA (c) = ... = c1 LA (e1 ) + · · · + cn LA (en )
(because L and LA are linear transformations)
=⇒ L(c) = LA (c) because L(e j ) = LA (e j ) for all j = 1, . . . , n).

4.2 Kernel and image

Definition 4.5. Let L : V → W be a linear transformation between vector spaces V,W over a field F.
Then
ker(L) := {x ∈ V : L(x) = 0W }
is called the kernel of L, and

im(L) := {y ∈ W : ∃x ∈ V : y = L(x)}

is called the image of L.

Remark 4.6. Let F be a field and A ∈ Mm×n (F). Then

ker(LA ) = N(A)

where N(A) = {c ∈ F n : Ac = 0} denotes the nullspace of A (see also Section 6.2 of L.A.I.) and

im(LA ) = Col(A)

where Col(A) denotes the column space of A; i.e. Col(A) = Span(a1 , . . . , an ), where a1 , . . . , an denote
the n columns of A. (See also Section 6.4 of L.A.I.)

Proof. First assertion: by definition.

Second assertion: follows from 4.9/(i) applied to the standard basis of F n .

Proposition 4.7. Let V and W be vector spaces over a field F and let L : V → W be a linear transfor-
mation. Then:

(a) ker(L) is a subspace of V .

(b) im(L) is a subspace of W .

Proof. (a) We verify the three subspace axioms.

(i) We have 0V ∈ ker(L) (see Note after Defn 4.2.)

 4 LINEAR TRANSFORMATIONS 27

(ii) Let x, y ∈ ker(L);

=⇒ L(x + y) = L(x) + L(y) = 0W + 0W = 0W ;
=⇒ x + y ∈ ker(L).
(iii) Let a ∈ F and x ∈ ker(L);
=⇒ L(ax) = a(L(x)) = a0W = 0W ;
=⇒ ax ∈ ker(L).

(b) We verify the three subspace axioms.

(i) We have 0W = L(0V ) ∈ im(L).

(ii) Let x, y ∈ im(L);
=⇒ ∃v, w ∈ V such that x = L(v) and y = L(w);
=⇒ x + y = L(v) + L(w) = L(v + w) ∈ im(L).
(iii) Let y ∈ im(L) and a ∈ F;
=⇒ ∃x ∈ V such that y = L(x);
=⇒ ay = a(L(x)) = L(ax) ∈ im(L).

Example 4.8. Let A ∈ M4×4 (R) be as in 3.8/(d). Find a basis of the image, im(LA ), of LA : R4 → R4 , c 7→
Ac.

Solution. We perform column operations:

   
1 −1 3 2 1 0 0 0
 2 −1 6 7  C47→C4+3C2  2
  1 0 3
A=  3 −2 9 −−−−−−−→ 
9  C37→C3−3C1  3 1 0 3
C47→C4−2C1
−2 0 −6 −10 −2 −2 0 −6
 
1 0 0 0
C27→C2+C1  2 1 0 0
−−−−−−−→  3
 =: A
e
1 0 0
−2 −2 0 0

  
1 0
2 1
=⇒ The two vectors 
 3  ,  1  span im(LA )
  

−2 −2
(because im(LA ) = Col(A) = Col(Ã) by 4.6 and 3.3/(b))
=⇒ They form a basis on im(LA ).
(because they are also L.I., as they are not multiples of each other) □

Proposition 4.9. Let V and W be vector spaces over a field F and let L : V → W be a linear transfor-
mation. Let x1 , . . . , xn ∈ V . Then:

(i) If x1 , . . . , xn ∈ V span V , then L(x1 ), . . . , L(xn ) span im(L).

(ii) If L(x1 ), . . . , L(xn ) are linearly independent, then x1 , . . . , xn are linearly independent.

Proof. (a) First, Span(L(x1 ), . . . , L(xn )) ⊆ im(L) (by 3.3 Note (i)).
For the other inclusion, let y ∈ im(L);
=⇒ ∃x ∈ V such that y = L(x)
and ∃a1 , . . . , an ∈ F such that x = a1 x1 + · · · + an xn (since V = Span(x1 , . . . , xn ))
=⇒ y = L(x) = L(a1 x1 + · · · + an xn ) =
28 4 LINEAR TRANSFORMATIONS

a1 L(x1 ) + · · · + an L(xn ) ∈ Span(L(x1 ), . . . , L(xn ));

=⇒ im(L) ⊆ Span(L(x1 ), . . . , L(xn ));
=⇒ im(L) = Span(L(x1 ), . . . , L(xn )). (i.e. L(x1 ), . . . , L(xn ) span im(L))

(b) Let a1 , . . . , an ∈ F such that a1 x1 + · · · + an xn = 0V ;

=⇒ 0W = L(0V ) = L(a1 x1 + · · · + an xn ) = a1 L(x1 ) + · · · + an L(xn );
=⇒ a1 = · · · = an = 0 (since L(x1 ), . . . , L(xn ) are linearly independent)
=⇒ x1 , . . . , xn are linearly independent.

Proposition 4.10 (Kernel Criterion). Let V and W be vector spaces over a field F, and let L : V → W
be a linear transformation. Then:

L is injective ⇐⇒ ker(L) = {0V }.

Proof. “=⇒”:
Let x ∈ ker(L) =⇒ L(x) = 0W .
We also have L(0V ) = 0W .
=⇒ x = 0V . (by injectivity)
“⇐=”:
Let x, y ∈ V such that L(x) = L(y);
=⇒ L(x − y) = L(x) − L(y) = 0W ;
=⇒ x − y = 0V (since ker(L) = {0V })
=⇒ x = y.

4.3 Isomorphism

Definition 4.11. Let V,W be vector spaces over a field F. A bijective linear transformation L : V → W is
called an isomorphism. The vector spaces V and W are called isomorphic if there exists an isomorphism
L : V → W ; we then write V ∼
= W.

Example 4.12. (a) For any vector space V over a field F, the identity id : V → V is an isomorphism.

(b) If L : V → W is an isomorphism then the inverse map L−1 : W → V is an isomorphism as well.

(See also Def 5.21 from L.A.I.)

(c) If A ∈ Mn×n (R) is invertible then LA : Rn → Rn is an isomorphism.

 
2 a
(d) The map L : R → C, 7→ a + bi, is an isomorphism between the vector spaces R2 and C over
b
R.

(e) For any n ∈ N, the map

 
a0
n+1  .. 
L:R → Pn ,  .  7→ a0 + a1t + · · · + ant n
an

is an isomorphism between the vector spaces Rn+1 and Pn over R.

(f) For any m, n ∈ N we have Rmn ∼

= Mm×n (R).
 4 LINEAR TRANSFORMATIONS 29

Proofs. (b) and (c) see Coursework.

(d) and (e) follow from the following proposition and 3.8/(c) and (d), respectively.
(f) (only in the case m = n = 2) The map
 
a1  
a2 
R4 → M2×2 (R),   7→ a1 a2
a3  a3 a4
a4

is clearly an isomorphism.

Proposition 4.13. Let V be a vector space over a field F with basis x1 , . . . xn . Then the map
 
a1
n  .. 
L : F → V,  .  7→ a1 x1 + · · · + an xn
an

is an isomorphism. (We will later use the notation Ix1 ,...,xn for the map L.)
   
a1 b1
 ..   .. 
Proof. (1) Let a =  .  and b =  .  ∈ F n ;
an b
  n
a1 + b1
 .. 
=⇒ L(a + b) = L  . 
an + bn
= (a1 + b1 )x1 + · · · + (an + bn )xn (by definition of L)
= (a1 x1 + · · · + an xn ) + (b1 x1 + · · · + bn xn )
(by distributivity, commutativity and associativity)
= L(a) + L(b). (by definition of L)
 
b1
 .. 
(2) Let a ∈ F and b =  .  ∈ F n ;
b
 n 
ab1
 .. 
=⇒ L(ab) = L  . 
abn
= (ab1 )x1 + · · · + (abn )xn (by definition of L)
= a(b1 x1 + · · · + bn xn ) (using the axioms of a vector space)
= a(L(b)). (by definition of L)
    
 a1
   0 
  
(3) ker(L) =  .  ∈ F : a1 x1 + · · · + an xn = 0V =  ... 
 ..  n  
 
  
an 0
  
(because x1 , . . . , xn are linearly independent)
=⇒ L is injective. (by 4.10)

(4) im(L) = Span(L(e1 ), . . . , L(en )) (by 4.9/(i))

= Span(x1 , . . . , xn ) = V (because x1 , . . . , xn span V )
=⇒ L is surjective.
30 4 LINEAR TRANSFORMATIONS

Theorem 4.14. Let V and W be vector spaces over a field F of dimension n and m, respectively. Then
V and W are isomorphic if and only if n = m.

Proof. “⇐=”:
We assume that n = m.
=⇒ We have isomorphisms LV : F n → V and LW : F n → W (by 4.13)
=⇒ LW ◦ LV−1 is an isomorphism between V and W . (by 4.3/(b) and 4.12/(b))
“=⇒”:
We assume that V and W are isomorphic.
Let L : V → W be an isomorphism and let x1 , . . . , xn be a basis of V .
=⇒ L(x1 ), . . . , L(xn ) span im(L) = W (by 4.9/(i))
and are linearly independent −1
(by 4.9/(ii) applied to L and L(x1 ), . . . , L(xn ))
=⇒ L(x1 ), . . . , L(xn ) form a basis of W
=⇒ n = dimF (W ) = m.

4.4 Dimension Theorem

Theorem 4.15 (Dimension Theorem). Let V be a vector space over a field F of finite dimension and let
L : V → W be a linear transformation from V to another vector space W over F. Then:

dimF (ker(L)) + dimF (im(L)) = dimF (V ).

(In the textbooks this is sometimes called the rank–nullity theorem.)

Example 4.16.

L:V→W dim(ker(L)) dim(im(L)) dim(V) Verification

LA for =2 =2 =4 2+2 = 4
A ∈ M4×4 (R) by 3.8/(d) by 4.8
as in 4.8
LA for =3 =2 =5 3+2 = 5
A ∈ M3×5 (R)
below
Isomorphism =0 = dim(W ) = dim(V ) 0 + dim(W )
by 4.10 = dim(V )
by 4.14
Zero map = dim(V ) =0 = dim(V ) dim(V ) + 0
= dim(V )

Let  
1 −2 2 3 −1
A = −3 6 −1 1 −7 .
2 −4 5 8 −4
We want to find dimR (ker(LA )) and dimR (im(LA )) – we do this by finding bases for both ker(LA ) and
im(LA ).
We find a basis of the nullspace N(A):
   
1 −2 2 3 −1 1 −2 0 −1 3
Gaussian elimination
A = −3 6 −1 1 −7 −−−−−−−−−−−→ 0 0 0 0 0  =: Ã.
2 −4 5 8 −4 0 0 1 2 −2
 4 LINEAR TRANSFORMATIONS 31

Then
N(A) = {x ∈ R5 : Ax = 0} = {x ∈ R5 : Ãx = 0} =
  

 x1 

x
 
 2 
   

5
= x3  ∈ R : x1 = 2x2 + x4 − 3x5 , x3 = −2x4 + 2x5 =
 
x
  
 4

 

 

x5
  

 2x2 + x4 − 3x5 

x
 

 2 



=  −2x4 + 2x5  : x2 , x4 , x5 ∈ R =
 
x4

  


 

 
x5
       

 2 1 −3 

1 0 0

       

       
= x2 0 + x4 −2 + x5  2  : x2 , x4 , x5 ∈ R =
     


 0 1 0 


 
0 0 1
 
     
2 1 −3
1  0   0 
     
= SpanR 0 , −2 ,  2  .
     
0  1   0 
0 0 1
The three vectors above are linearly independent, since if a1 , a2 , a3 ∈ R and
       
2 1 −3 0
1 0  0  0
       
a1  0 + a2 −2 + a3  2  = 0 ,
       
0 1  0  0
0 0 1 0
then a1 = a2 = a3 = 0 by looking at the second, fourth and fifth coordinates, respectively.
Thus these three vectors are a basis of N(A), so dimR (ker(LA )) = dimR (N(A)) = 3.
We now find a basis of the image im(LA ):
   
1 −2 2 3 −1 1 0 0 0 0
column operations
A = −3 6 −1 1 −7 −−−−−−−−−→ −3 0 5 0 0 .
2 −4 5 8 −4 2 0 1 0 0
   
1 0
So the vectors −3 , 5 span im(LA ). Since they are obviously not multiples of each other, they
  
2 1
are linearly independent, hence form a basis of im(LA ). Consequently, dimR (im(LA )) = 2.

Proof of the Dimension Theorem. Let x1 , . . . , xr be a basis of ker(L).

We extend x1 , . . . , xr to a basis x1 , . . . , xn of the whole V for some n ≥ r (by adding L.I. vectors in V until
we obtain a maximal L.I. subset) and show below that L(xr+1 ), . . . , L(xn ) form a basis of im(L). Then
we have
dimF (ker(L)) + dimF (im(L)) = r + (n − r) = n = dimF (V ),
as we wanted to prove.
Proof that L(xr+1 ), . . . , L(xn ) form a basis of im(L):
32 4 LINEAR TRANSFORMATIONS

• L(xr+1 ), . . . , L(xn ) span im(L):

Let y ∈ im(L).
=⇒ ∃x ∈ V such that y = L(x) (by definition of im(L))
and ∃a1 , . . . , an ∈ F such that x = a1 x1 + · · · + an xn (since x1 , . . . , xn span V )
=⇒ y = L(x) = L(a1 x1 + · · · + an xn )
= a1 L(x1 ) + · · · + an L(xn ) (because L is a linear transformation)
= ar+1 L(xr+1 ) + · · · + an L(xn ) (because x1 , . . . , xr ∈ ker(L))
∈ Span(L(xr+1 ), . . . , L(xn ))
=⇒ im(L) ⊆ Span(L(xr+1 ), . . . , L(xn )).
We also have Span(L(xr+1 ), . . . , L(xn )) ⊆ im(L) (by 3.3 Note (i))
=⇒ im(L) = Span(L(xr+1 ), . . . , L(xn )).
• L(xr+1 ), . . . , L(xn ) are linearly independent:
Let ar+1 , . . . , an ∈ F such that ar+1 L(xr+1 ) + · · · + an L(xn ) = 0W .
=⇒ L(ar+1 xr+1 + · · · + an xn ) = 0W (because L is a linear transformation)
=⇒ ar+1 xr+1 + · · · + an xn ∈ ker(L) (by definition of kernel)
=⇒ ∃a1 , . . . ar ∈ F such that ar+1 xr+1 + · · · + an xn = a1 x1 + · · · + ar xr
(because x1 , . . . , xr span ker(L))
=⇒ a1 x1 + · · · + ar xr − ar+1 xr+1 − · · · − an xn = 0V
=⇒ a1 = · · · = ar = −ar+1 = · · · = −an = 0 (because x1 , . . . , xn are linearly independent)
=⇒ ar+1 = · · · = an = 0.

4.5 Matrix representation II

Proposition 4.17 (Matrix representation II). Let V and W be vector spaces over a field F with bases
x1 , . . . , xn and y1 , . . . , ym , respectively. Let L : V → W be a linear transformation. Then there exists a
unique matrix A ∈ Mm×n (F) that represents L with respect to x1 , . . . , xn and y1 , . . . , ym . Here we say that
A = (ai j ) ∈ Mm×n (F) represents L with respect to x1 , . . . , xn and y1 , . . . , ym if for all c1 , . . . , cn , d1 , . . . , dm ∈
F we have    
d1 c1
 ..   .. 
L(c1 x1 + · · · + cn xn ) = d1 y1 + · · · + dm ym ⇐⇒  .  = A  .  .
dm cn

Proof. Let A ∈ Mm×n (F). Then:

A represents L with respect to x1 , . . . , xn and y1 , . . . , ym
(⋆)
⇐⇒ The diagram
L
V −−→ W
Ix1 ,...,xn ↑ ↑ Iy1 ,...,ym
LA
Fn −−→ Fm
commutes, i.e. L ◦ Ix1 ,...,xn = Iy1 ,...,ym ◦ LA (see proof below)
⇐⇒ LA = Iy−1 1 ,...,ym
◦ L ◦ Ix1 ,...,xn =: M
⇐⇒ A represents M : F n → F m (with respect to the standard bases of F n and F m ).
Hence 4.17 follows from 4.4.
   
c1 d1
 ..   .. 
Proof of (⋆). Let c1 , . . . , cn ∈ F and let d1 , . . . , dm ∈ F be given by A  .  =  . 
cn dm
 
c1
 .. 
=⇒ (L ◦ Ix1 ,...,xn )  .  = L(c1 x1 + · · · + cn xn )
cn
 4 LINEAR TRANSFORMATIONS 33

   
c1 d1
 ..   .. 
and (Iy1 ,...,ym ◦ LA )  .  = Iy1 ,...,ym  .  = d1 x1 + · · · + dm ym .
cn dm
Hence: L(c1 x1 + · · ·+ c
 nx )
n = d y
1 1 + · · · + d y
mm 
c1 c1
 ..   .. 
⇐⇒ (L ◦ Iy1 ,...,ym )  .  = (Iy1 ,...,ym ◦ LA )  . 
cn cn
Therefore: A represents L ⇐⇒ L ◦ Ix1 ,...,xn = Iy1 ,...,ym ◦ LA .

Note. Given L, x1 , . . . , xn and y1 , . . . , ym as in 4.17 we find the corresponding matrix A as follows: For
each i = 1, . . . , n we compute L(xi ), represent L(xi ) as a linear combination of y1 , . . . , ym and write the
coefficients of this linear combination into the ith column of A.

Example 4.18. Find the matrix A ∈ M3×4 (R) representing differentiation D : P3 → P2 , f 7→ f ′ , with
respect to the bases 1,t,t 2 ,t 3 and 1,t,t 2 of P3 and P2 , respectively.

Solution. We have

D(1) = 0 = 0 + 0t + 0t 2
D(t) = 1 = 1 + 0t + 0t 2
D(t 2 ) = 2t = 0 + 2t + 0t 2
D(t 3 ) = 3t 2 = 0 + 0t + 3t 2
 
0 1 0 0
=⇒ A = 0 0 2 0.
0 0 0 3
 
1 −1
Example 4.19. Let B := ∈ M2×2 (R). Find the matrix A ∈ M2×2 (R) representing the linear
2 4
   
2 2 1 1
transformation LB : R → R , x 7→ Bx, with respect to the basis , of R2 (used for both
−2 −1
source and target space).

Solution.
          
1 1 −1 1 3 1 1
LB = = =3 +0
2 2 4 −2 −6 −2 −1
          
1 1 −1 1 2 1 1
LB = = =0 +2
−1 2 4 −1 −2 −2 −1
 
3 0
=⇒ A = .
0 2
34 5 DETERMINANTS

5 Determinants

In Linear Algebra I the determinant of a square matrix has been defined axiomatically (cf. Theorem 5.3
here). Here we begin with the following closed formula.
Definition 5.1 (Leibniz’ definition of determinant). Let F be a field. Let n ≥ 1 and A = (ai j )i, j=1,...,n ∈
Mn×n (F). Then
n
det(A) := ∑ sgn(σ ) ∏ ai,σ (i) ∈ F
σ ∈Sn i=1

is called the determinant of A. We also write |A| for det(A).

Example 5.2. (a) Let n = 1 and A = (a11 ) ∈ M1×1 (F).
We have S1 = {id} and
det(A) = sgn(id)a11 = a11 .
 
a11 a12
(b) Let n = 2 and A = ∈ M2×2 (F).
a21 a22
We have S2 = {id, ⟨1, 2⟩} and

det(A) = sgn(id)a11 a22 + sgn(⟨1, 2⟩)a12 a21 = a11 a22 − a12 a21 .
 
1 + 2i 3 + 4i
For example: if A = ∈ M2×2 (C) then
1 − 2i 2 − i
det(A) = (1 + 2i)(2 − i) − (3 + 4i)(1 − 2i) = (2 + 2 + 4i − i) − (3 + 8 + 4i − 6i) = −7 + 5i.
 
a11 a12 a13
(c) Let n = 3 and A = a21 a22 a23  ∈ M3×3 (F).
a31 a32 a33
We have S3 = {id, ⟨1, 2, 3⟩, ⟨1, 3, 2⟩, ⟨1, 3⟩, ⟨2, 3⟩, ⟨1, 2⟩} and

det(A) = sgn(id) a11 a22 a33 + sgn(⟨1, 2, 3⟩) a12 a23 a31 + sgn(⟨1, 3, 2⟩) a13 a21 a32
| {z } | {z } | {z }
=+1 =+1 =+1
+ sgn(⟨1, 3⟩) a13 a22 a31 + sgn(⟨2, 3⟩) a11 a23 a32 + sgn(⟨1, 2⟩) a12 a21 a33 .
| {z } | {z } | {z }
=−1 =−1 =−1

a11 a12 a13 a11 a12

⧹ × × ⧸
Trick to memorise: a21 a22 a23 a21 a22
⧸ × × ⧹
a31 a32 a33 a31 a32
(Rule of Sarrus)  
1 0 1
For example, let A = 0 1 0 ∈ M3×3 (F2 ).
1 1 0
=⇒ det(A) = (0 + 0 + 0) − (1 + 0 + 0) = 1 (because −1 = 1 in F2 ).
 
a11 . . . . . . a1n
.. 
 0 ...

. 
(d) Let A = (ai j ) be an upper (or lower) triangular matrix. So A is of the form  . .
 .
 .. . . . . . ... 

0 . . . 0 ann
In other words, we have ai j = 0 if i > j.
Then det(A) = a11 a22 · · · ann , i.e. det(A) is the product of the entries on the main diagonal. For
example, det(In ) = 1.
 5 DETERMINANTS 35

Proof. Let σ ∈ Sn , σ ̸= id =⇒ ∃i0 ∈ {1, . . . , n} such that i0 > σ (i0 );

=⇒ ai0 ,σ (i0 ) = 0 =⇒ ∏ni=1 ai,σ (i) = 0.
=⇒ det(A) = sgn(id) ∏ni=1 ai,id(i) = a11 a22 · · · ann .

Theorem 5.3 (Weierstrass’ axiomatic description of the determinant map). See Defn 4.1 of L.A.I. Let F
be a field. Let n ≥ 1. The map
det : Mn×n (F) → F, A 7→ det(A),
has the following properties and is uniquely determined by these properties:

(a) det 
is linear in each column:
    
a1s + b1s a1s b1s
.. . .
det C D = det C .. D + det C .. D.
     
.
ans + bns ans bns

(b) Multiplying any column of a matrix A ∈ Mn×n (F) with a scalar λ ∈ F changes det(A) by the factor
λ:    
λ a1s a1s
.. .
det C D = λ · det C .. D.
   
.
λ ans ans

(c) If two columns of A ∈ Mn×n (F) are equal, then det(A) = 0.

(d) det(In ) = 1.

Proof. Omitted here (please see the extended notes).

Remark 5.4. Theorem 5.3 and the following Corollary 5.5 also hold when “columns” are replaced with
“rows” (similar proofs).

Corollary 5.5. Let F be a field. Let A ∈ Mn×n (F). Then:

(a) For all λ ∈ F we have det(λ A) = λ n det(A).

(b) If a column of A is the zero column then det(A) = 0.

(c) Let B be obtained from A by swapping two columns of A. Then det(B) = −det(A).

(d) Let λ ∈ F and let B be obtained from A by adding the λ -multiple of the jth column of A to the ith
column of A (i ̸= j). Then det(B) = det(A).

Proof. (a) Apply Theorem 5.3/(b) n times.

(b) Apply Theorem 5.3/(b) with λ = 0.

(c) Let a, b denote the two columns of A to be swapped.

=⇒ det(A) + det(B) = det(. . . a . . . b . . .) + det(. . . b . . . a . . .)
= det(. . . a . . . b . . .) + det(. . . b . . . a . . .)
+ det(. . . a . . . a . . .) + det(. . . b . . . b . . .) (using 5.3/(c))
| {z } | {z }
=0 =0
= det(. . . a . . . a + b . . .) + det(. . . b . . . a + b . . .) (by 5.3/(a))
= det(. . . a + b . . . a + b . . .) (by 5.3/(a))
=0 (by 5.3/(c))
36 5 DETERMINANTS

(d) det(B) = det(. . . a + λ b . . . b . . .)

= det(. . . a . . . b . . .) + λ det(. . . b . . . b . . .) (by 5.3/(a),(b))
| {z }
=0
= det(A). (by 5.3/(c))

Theorem 5.6. Let F be a field. Let A ∈ Mn×n (F). Then the following are equivalent:

(a) A is invertible.

(b) The columns of A span F n .

(c) N(A) = {0}.

(d) det(A) ̸= 0.

For (a) ⇐⇒ (d), compare Thm 4.14 of L.A.I.

   
1 0
0  .. 
Before the proof, recall that we denote the standard basis vectors of F n as e1 =  .  , . . . , en =  . .
   
.
. 0
0 1
So, in particular, In = (e1 , . . . , en ).

Proof.
“(a) =⇒ (b)”: We’ll prove the following more precise statement:
(⋆) ∃B ∈ Mn×n (F) such that AB = In ⇐⇒ The columns of A span F n .
Proof of (⋆): Let a1 . . . , an denote the columns of A. Then:
LHS ⇐⇒ ∃b1 , . . . , bn ∈ F n such that Abi = ei for all i = 1, . . . , n
(we can use the columns of B = (b1 , . . . , bn ))
n
⇐⇒ ∃b1 , . . . , bn ∈ F such that a1 bi1 + · · · + an bin = ei for all i = 1, . . . , n
⇐⇒ e1 , . . . , en ∈ Span(a1 , . . . , an )
⇐⇒ RHS.
“(b) ⇐⇒ (c)”: The columns of A span F n
⇐⇒ The columns of A form a basis of F n (by 3.15)
⇐⇒ The columns of A are linearly independent (by 3.15)
⇐⇒ N(A) = {0} (by definition of N(A))
“(b) =⇒ (a)”: ∃B ∈ Mn×n (F) such that AB = In (by (⋆))
=⇒ A(BA) = (AB)A = In A = A = AIn
=⇒ A(BA − In ) = 0
=⇒ Every column of BA − In belongs to N(A)
=⇒ BA − In = 0 (because N(A) = {0} by (b) ⇐⇒ (c))
=⇒ BA = In
=⇒ A is invertible (as both AB = In and BA = In )
“(b) ⇐⇒ (d)”: We apply column operations to the matrix A until we arrive at a lower triangular matrix
C. Then:
The columns of A span F n
⇐⇒ the columns of C span F n (by 3.3/(b))
⇐⇒ all the diagonal elements of C are non-zero (because C is triangular)
⇐⇒ det(C) ̸= 0 (by 5.2/(d))
⇐⇒ det(A) ̸= 0 (because det(C) = λ det(A) for some non-zero λ ∈ F by 5.5)
 5 DETERMINANTS 37

Theorem 5.7. (See also Thm 4.21 of L.A.I.) Let F be a field. Let A, B ∈ Mn×n (F). Then:

det(AB) = det(A) · det(B)

Proof. Omitted. (See the proof of Thm 4.21 in L.A.I., it works for any field F.)
 
m 1 + 4i 1
Example 5.8. For each m ∈ N compute det(A ) ∈ C, here A = ∈ M2×2 (C).
5+i 1−i

Solution. det(A) = (1 + 4i)(1 − i) −(5 + i) = (1 + 4 + 4i − i) − (5 + i) = 2i.



 2m if m is of the form 4k;
2m i

if m is of the form 4k + 1;
=⇒ det(Am ) = det(A)m = (2i)m = (by 5.7)


 −(2m ) if m is of the form 4k + 2;
−(2m )i if m is of the form 4k + 3.

38 6 DIAGONALISABILITY

6 Diagonalisability

6.1 Eigen-things

Definition 6.1. Let V be a vector space over a field F and let L : V → V be a linear transformation from
V to itself.

(a) For any λ ∈ F the set

Eλ (L) := {x ∈ V : L(x) = λ x}
is called the eigenspace of L corresponding to λ .

(b) An element λ ∈ F is called an eigenvalue of L if Eλ (L) is not the zero space. In this case any
vector x in Eλ (L) different from the zero vector is called an eigenvector of L with eigenvalue λ .

(c) Let A ∈ Mn×n (F). The eigenspaces, eigenvalues and eigenvectors of A are, by definition, those of
LA : F n → F n , x 7→ Ax.

(Compare Definition 7.1 of L.A.I.)

Proposition 6.2. Let F, V and L be as in Defn 6.1. Then Eλ (L) is a subspace of V for every λ ∈ F.

Proof. (a) We have 0V ∈ Eλ (L) because L(0V ) = 0V = λ · 0V .

(b) Let x, y ∈ Eλ (L)

=⇒ L(x + y) = L(x) + L(y) = λ x + λ y = λ (x + y)
=⇒ x + y ∈ Eλ (L).

(c) Let a ∈ F and x ∈ Eλ (L)

=⇒ L(ax) = aL(x) = a(λ x) = (aλ )x = (λ a)x = λ (ax)
=⇒ ax ∈ Eλ (L).

Proposition 6.3. Let F be a field. Let A ∈ Mn×n (F) and λ ∈ F. Then:

λ is an eigenvalue of A ⇐⇒ det(λ In − A) = 0.

(pA (λ ) := det(λ In − A) is called the characteristic polynomial of A) (See also Proposition 7.5 in L.A.I.)

Proof. λ is an eigenvalue of A
⇐⇒ ∃x ∈ F n , x ̸= 0, such that Ax = λ x
⇐⇒ ∃x ∈ F n , x ̸= 0, such that (λ In − A)x = 0
⇐⇒ N(λ In − A) ̸= {0}
⇐⇒ det(λ In − A) = 0. (by 5.6 (c) ⇐⇒ (d))

Example 6.4. Determine the (complex!) eigenvalues of the matrix

 
5i 3
A := ∈ M2x2 (C)
2 −2i

and a basis of the eigenspace of A for each eigenvalue of A.

Solution.
 
λ − 5i −3
pA (λ ) = det(λ I2 − A) = det
−2 λ + 2i
= (λ − 5i)(λ + 2i) − 6 = λ 2 − (3i)λ + 4
 6 DIAGONALISABILITY 39

√
3i± −9−16 3i±5i
The two roots of this polynomial are λ1,2 = 2 = 2 = 4i or −i;
=⇒ eigenvalues of A are 4i and −i.
Basis of E4i (A): Apply Gaussian elimination to
     
−i −3 R17→iR1 1 −3i R27→R2+2R1 1 −3i
4iI2 − A = −−−−→ −−−−−−−→
−2 6i −2 6i 0 0

  
x1 2
=⇒ E4i (A) = ∈ C : x1 = (3i)x2
x2
    
(3i)x2 3i
= : x2 ∈ C = Span
x2 1
 
3i
=⇒ a basis of E4i (A) is (as it is L.I.).
1
Basis of E−i (A): Apply Gaussian elimination to
     
−6i −3 R1↔R2 −2 i R27→R2−(3i)R1 −2 i
−iI2 − A = −−−−→ −−−−−−−−→
−2 i −6i −3 0 0

  
x1 2 i
=⇒ E−i (A) = ∈ C : x1 = x2
x2 2
 i    i 
2 x2 2
= : x2 ∈ C = Span
x2 1
 
i
=⇒ a basis of E−i (A) is (as it is L.I.).
2

Example 6.5. Let V be the real vector space of infinitely often differentiable functions from R to R and
let D : V → V, f 7→ f ′ , denote differentiation (cf. 4.3/(d)). Then for every λ ∈ R the eigenspace of D
with eigenvalue λ is of dimension 1 with basis given by the function expλ : R → R, t 7→ eλt .

Proof. (eλt )′ = λ (eλt ) (by the chain rule)

=⇒ expλ ∈ Eλ (D).
Conversely, suppose f ∈ Eλ (D)
=⇒ ( f (t)e−λt )′ = f ′ (t)e−λt + f (t)(e−λt )′ (by the product rule)
= λ f (t)e−λt − λ f (t)e−λt (because f ∈ Eλ (D) and by the chain rule)
=0
=⇒ f (t)e −λt is a constant, say a ∈ R (by Calculus)
=⇒ f (t) = aeλt , i.e. f = a expλ .
Hence Eλ (D) = Span(expλ ).

6.2 Diagonalisability

Definition 6.6. (a) Let F, V and L : V → V be as in Defn 6.1. We say that L is diagonalisable if there
exists a basis x1 , . . . , xn of V such that the matrix D representing L with respect to this basis is a
diagonal matrix.

(b) Let F be a field. We say that a square matrix A ∈ Mn×n (F) is diagonalisable if the linear transfor-
mation LA : F n → F n , x 7→ Ax, is diagonalisable.
40 6 DIAGONALISABILITY

6.2.1 Diagonalisability (version 1)

Proposition 6.7. Let F, V and L : V → V be as in Defn 6.1. Then L is diagonalisable if and only if V
has a basis x1 , . . . , xn consisting of eigenvectors of L.

Proof. “=⇒”:
Suppose ∃ a basis x1 , . . . , xn of V such that the matrix D representing L is diagonal, with some λ1 , . . . , λn ∈
F on the main diagonal.
=⇒ for any c1 , . . . , cn ∈ F we have

L(c1 x1 + · · · + cn xn ) = (λ1 c1 )x1 + · · · + (λn cn )xn ,

     
λ1 c1 λ1 0 c1
 ..   . .. .
because  .  =   ·  ..  .
 

λn cn 0 λn cn

=⇒ in particular when c1 = 0, . . . , ci−1 = 0, ci = 1, ci+1 = 0, . . . , cn = 0 for some i ∈ {1, . . . , n}, we get

L(xi ) = λi xi

=⇒ x1 , . . . , xn are eigenvectors of L with eigenvalues λ1 , . . . , λn , respectively.

“⇐=”:
Let x1 , . . . , xn be a basis of V consisting of eigenvectors of L and let λi ∈ F denote the eigenvalue corre-
sponding to xi .  
λ1 0
Define a diagonal matrix D by D = 
 .. .

.
0 λn
=⇒ D represents L with respect to x1 , . . . , xn because

L(c1 x1 + · · · + cn xn ) = c1 L(x1 ) + · · · + cn L(xn ) = λ1 c1 x1 + · · · + λn cn xn

   
λ1 c1 c1
 ..   .. 
and  .  = D  .  ∀c1 , . . . , cn ∈ F.
λn cn cn

6.2.2 Diagonalisability (version 2)

Proposition 6.8. Let F be a field. Let A ∈ Mn×n (F). Then A is diagonalisable if and only if there exists
an invertible matrix M ∈ Mn×n (F) such that M −1 AM is a diagonal matrix. (In this case we say that M
diagonalises A.)

Proof preparation. Let M ∈ Mn×n (F) with column vectors x1 , . . . , xn . Suppose M is invertible. Then:
xi is an eigenvector of A with eigenvalue λi
⇐⇒ Axi = λi xi
⇐⇒ AMei = λi (Mei ) (because xi = Mei is the ith column of M)
⇐⇒ AMei = M(λi ei )
 6 DIAGONALISABILITY 41

⇐⇒ M −1 AMei = λi ei   (multiply with M −1 )

0
 .. 
.
 
0
−1
th
  th
⇐⇒ i column of M AM is 
λi  ← i place.

0
 
 .. 
.
0

Proof. “=⇒”
A is diagonalisable
=⇒ ∃ a basis x1 , . . . , xn of F n consisting of eigenvectors of A (by 6.7)
=⇒ The matrix M ∈ Mn×n (F) with columns x1 , . . . , xn is invertible (by 5.6 (b) =⇒ (a))
and M −1 AM is diagonal. (by “preparation” above)
“⇐=”
There exists an invertible M ∈ Mn×n (F) such that M −1 AM is diagonal
=⇒ the columns of M are eigenvectors of A (by “preparation” above)
and they form a basis of F n . (by 5.6 (a) =⇒ (b) and 3.15)
=⇒ A is diagonalisable. (by 6.7)
Example 6.9. Show that the matrix
 
0 −1 1
A := −3 −2 3 ∈ M3×3 (R)
−2 −2 3
is diagonalisable and find an invertible matrix M ∈ M3×3 (R) that diagonalises it.
Solution. First compute the characteristic polynomial of A:
 
λ 1 −1
pA (λ ) = det(λ I3 − A) = det  3 λ + 2 −3 
2 2 λ −3
= λ (λ + 2)(λ − 3) + 1(−3)2 + (−1)3 · 2 − (−1)(λ + 2)2 − λ (−3)(2) − 1 · 3(λ − 3)
= λ (λ 2 − λ − 6) − 6 − 6 + 2λ + 4 + 6λ − 3λ + 9
= λ 3 − λ 2 − λ + 1 = λ 2 (λ − 1) − (λ − 1) = (λ 2 − 1)(λ − 1) = (λ − 1)2 (λ + 1).
=⇒ Eigenvalues of A are 1 and −1.
Basis of E1 (A): We apply Gaussian elimination to 1 · I3 − A:
   
1 1 −1 1 1 −1
R27→R2−3R1
1 · I3 − A = 3 3 −3 −−−−−−−→ 0 0 0  =: Ã
R37→R3−2R1
2 2 −2 0 0 0
  
 b1 
=⇒ E1 (A) = N(1 · I3 − A) = N(Ã) = b2  ∈ R3 : b1 + b2 − b3 = 0
b3
 
  
 −b2 + b3 
=  b2  : b2 , b3 ∈ R
b3
 
         
 −1 1  −1 1
= b2 1 + b3 0 : b2 , b3 ∈ R = Span
      1 , 0
 
0 1 0 1
 
42 6 DIAGONALISABILITY

   
−1 1
Also x1 :=  1  , x2 := 0 are L.I. (as they are not multiples of each other)
0 1
=⇒ A basis of E1 (A) is x1 , x2 .
Basis of E−1 (A): We apply Gaussian elimination to (−1)I3 − A:
   
−1 1 −1 −1 1 −1
R27→R2+3R1
−I3 − A =  3 1 −3 −−−−−−−→  0 4 −6
R37→R3+2R1
2 2 −4 0 4 −6
1
 
−1 0 2
R7→R1− 41 R2
−−−−−−−→ 0 4 −6 =: Â

R37→R3−R2
0 0 0

=⇒ E−1 (A) = N((−1) · I3 − A) = N(Â) =

  
 b1 1 
= b2  ∈ R3 : −b1 + b3 = 0 and 4b2 − 6b3 = 0
2
b3
 
 1    1 
 2 b3  2
=  23 b3  : b3 ∈ R = Span  32 
b3 1
 
1
2
Also x3 :=  32  is linearly independent (as it is not 0)
1
=⇒ A basis of E−1 (A) is x3 .
−1 1 12
 

For M := (x1 , x2 , x3 ) =  1 0 32  we have det(M) = 21 + 32 − 1 = 1 ̸= 0

0 1 1
=⇒ x1 , x2 , x3 form a basis of R3 consisting of eigenvectors of A (by 5.6 and 3.13)
=⇒ A is diagonalisable and M diagonalises A. (by the proof of 6.8)

6.2.3 Diagonalisability (version 3)

Definition 6.10. Let F be a field. Let A ∈ Mn×n (F) and λ ∈ F be an eigenvalue of A.

(a) The algebraic multiplicity aλ (A) of λ is its multiplicity as a root of the characteristic polynomial
of A.
(b) The geometric multiplicity gλ (A) of λ is the dimension of the eigenspace Eλ (A).
Example 6.11. (a) In Example 6.9 we had pA (λ ) = (λ − 1)2 (λ + 1), so a1 (A) = 2 and a−1 (A) = 1.
Looking at the eigenspaces, we had g1 (A) = 2 and g−1 (A) = 1.
 
1 1
(b) Let A = ∈ M2×2 (F) (for any field F)
0 1
 
2 1
=⇒ pA (λ ) = (λ − 1) and a basis of E1 (A) is
0
=⇒ a1 (A) = 2 but g1 (A) = 1.
Theorem 6.12. Let F be a field. Let A ∈ Mn×n (F). Then A is diagonalisable if and only if the char-
acteristic polynomial of A splits into linear factors and the algebraic multiplicity equals the geometric
multiplicity for each eigenvalue of A.
 6 DIAGONALISABILITY 43

Proof. Omitted.
 
0 1
Example 6.13. Determine whether the matrix A = is diagonalisable when viewed as an
−1 0
element of M2×2 (R), of M2×2 (C) and of M2×2 (F2 ). If A is diagonalisable then determine an invertible
matrix M that diagonalises A.
 
λ −1
Solution. pA (λ ) = det = λ 2 + 1.
1 λ
For R:
λ 2 + 1 does not split into linear factors
=⇒ as an element of M2×2 (R) the matrix A is not diagonalisable. (by 6.12)
(Actually A is a rotation by 90◦ about the origin.)
For C:
pA (λ ) = λ 2 + 1 = (λ + i)(λ − i)
=⇒ a+i (A) = 1 and a−i (A) = 1.
Basis of Ei (A): We apply Gaussian elimination to iI2 − A:
     
i −1 R17→(−i)R1 1 i R27→R2−R1 1 i
iI2 − A = −−−−−−→ −−−−−−→ =: Ã
1 i 1 i 0 0

  
b1 2
=⇒ Ei (A) = N(iI2 − A) = N(Ã) = ∈ C : b1 + ib2 = 0
b2
    
−ib2 −i
= : b2 ∈ C = Span
b2 1
 
−i
Also is linearly independent (as it is not 0)
1
 
−i
=⇒ is a basis of Ei (A)
1
=⇒ gi (A) = 1.
Basis of E−i (A): We apply Gaussian elimination to (−i)I2 − A:
     
−i −1 R17→iR1 1 −i R27→R2−R1 1 −i
−iI2 − A = −−−−→ −−−−−−→ =: Â
1 −i 1 −i 0 0

  
b1 2
=⇒ E−i (A) = N((−i)I2 − A) = N(Â) = ∈ C : b1 − ib2 = 0
b2
    
ib2 i
= : b2 ∈ C = Span
b2 1
 
i
Also is linearly independent (as it is not 0)
1
 
i
=⇒ is a basis of E−i (A)
1
=⇒ g−i (A) = 1.
=⇒ A is diagonalisable
 when viewed as an element of M2×2 (C) (by 6.12)
−i i
and M = diagonalises A.
1 1
44 6 DIAGONALISABILITY

For F2 : pA (λ ) = λ 2 + 1 = (λ + 1)2 (since 1 + 1 = 0 in F2 )

=⇒ A has a single eigenvalue 1 = −1 and a1 (A) = 2.
Basis of E1 (A): We apply Gaussian elimination to 1 · I2 − A:
   
1 1 R27→R2−R1 1 1
I2 − A = −−−−−−→ =A
b
1 1 0 0

  
b1 2
=⇒ E1 (A) = N(I2 − A) = N(A) =
b ∈ F2 : b1 + b2 = 0
b2
    
b2 1
= : b2 ∈ F2 = Span
b2 1
 
1
Also is linearly independent (as it is not 0)
1
 
1
=⇒ is a basis of E1 (A)
1
=⇒ g1 (A) = 1.
=⇒ A is not diagonalisable. (by 6.12, since a1 (A) = 2 ̸= 1 = g1 (A))

6.3 Cayley–Hamilton Theorem

Theorem 6.14 (Cayley–Hamilton Theorem). Let F be a field, let A ∈ Mn×n (F) and let pA be the char-
acteristic polynomial of A. Then pA (A) is the zero matrix.
 
0 1
Example 6.15. Let A = ∈ M2 (F)
−1 0
=⇒ pA (λ ) = λ 2 + 1       (see Example 6.13)
−1 0 1 0 0 0
=⇒ pA (A) = A2 + 1 · I2 = + = .
0 −1 0 1 0 0

Proof of the Cayley–Hamilton Theorem

! 6.14.
a1 0
First Case: When A = .. is a diagonal matrix.
.
0 an

=⇒ pA (λ ) = det(λ · In − A) = (λ − a1 ) . . . (λ − an )
=⇒ pA (A) = (A − a1 In ) · · · (A − an In )
  
0 0 a1 − a2 0
 a2 − a1  0 
= ...
  
..  ..
 .  . 
0 an − a1 0 an − a2
 
a1 − an 0
 .. 
...
 . 

 an−1 − an 
0 0
= 0,

because the product of any two diagonal matrices with diagonal entries b1 , . . . , bn and c1 , . . . , cn respec-
tively, is the diagonal matrix with diagonal entries b1 c1 , . . . , bn cn .
 6 DIAGONALISABILITY 45

Preparatory Step: If A, M, D ∈ Mn×n (F) are such that M is invertible and D = M −1 AM, then:

pD (λ ) = det(λ · In − D) = det(λ · In − M −1 AM)

= det(M −1 (λ In )M − M −1 AM) = det(M −1 (λ In − A)M)
= det(M)−1 det(λ In − A)det(M) = det(λ In − A)
= pA (λ ).

In other words, the characteric polynomials of A and D are the same.

Another Preparatory Computation: If M, D ∈ Mn×n (F), M invertible, and k ≥ 0, then:

(MDM −1 )k = (MDM −1 )(MDM −1 ) · · · (MDM −1 ))

= MD(M −1 M)D(M −1 M) · · · (M −1 M)DM −1
= MDk M −1 .

Second Case: When A is a diagonalisable matrix.

=⇒ ∃ an invertible M ∈ Mn×n (F) such that M −1 AM = D where D is a diagonal matrix (by 6.8)
Denote pA (λ ) = λ n + an−1 λ n−1 + · · · + a1 λ + a0 the characteristic polynomial of A
=⇒ pA (A) = An + an−1 An−1 + · · · + a1 A + a0 In
= (MDM −1 )n + an−1 (MDM −1 )n−1 + · · · + a1 (MDM −1 ) + a0 In
(by Preparatory Computation above)
= MD M + an−1 MD M + · · · + a1 MDM + a0 MM −1
n −1 n−1 −1 −1

= M(Dn + an−1 Dn−1 + · · · + a1 D + a0 In )M −1

= M pA (D)M −1
= M pD (D)M −1 (by Preparatory Step above)
= M0M −1 = 0. (by the First Case)
General Case: Omitted.
46 7 COURSEWORK SHEETS

7 Coursework Sheets

7.1 Coursework Sheet 1

Submit a single pdf with scans of your work to Blackboard by Monday, 5 February 2024, 17:00.

Exercise 1
 
1 1 1
Consider the matrix A = 3 1 0 .
2 0 −1
 
1
(a) Show that the equation Ax = 2 has no solution in R3 .

3

(b) Find two linearly independent vectors y in R3 such that the equation Ax = y has a solution in R3 .

(Give reasons for your answers.)

Exercise 2

(a) Let w, z be complex numbers. Solve the linear equation wx = z; in other words, find all x ∈ C such
that wx = z. (Hint: You need to distinguish three cases.)

(b) Solve the following system of linear equations:

(5i)x1 + 3x2 = 12 + i
x1 − (2i)x2 = 3 − i.

Exercise 3

Let A be a real m × n matrix. Recall that its nullspace is the following set of vectors: N(A) = {u ∈ Rn |
Au = 0 ∈ Rm } ⊆ Rn . Prove that N(A) is a subspace of Rn .

Extra question (not assessed)

Multiplication of complex numbers defines a binary operation on C× := C \ {0}. Show that C× together
with this operation is an abelian group. (Here we consider the multiplication of complex numbers defined
like so: if a + bi and c + di (a, b, c, d ∈ R) are complex numbers, their product is declared to be the
complex number (ac − bd) + (ad + bc)i. In your arguments, you may without further discussion use the
usual laws of algebra for R. such as associativity for addition and multiplication of real numbers.)

Challenge question (not assessed)

Let G be a group such that for all a ∈ G we have a ∗ a = e. Show that G is abelian.
 7 COURSEWORK SHEETS 47

7.2 Coursework Sheet 2

Submit a single pdf with scans of your work to Blackboard by Monday, 12 February 2024, 17:00.

Exercise 1

Let G and H be groups with binary operations ⊞ and ⊙, respectively. We define a binary operation ∗ on
the cartesian product G × H by

(a, b) ∗ (a′ , b′ ) := (a ⊞ a′ , b ⊙ b′ ) (for a, a′ ∈ G and b, b′ ∈ H).

Show that G × H together with this operation is a group.

Exercise 2

For a, b ∈ R we define a ∗ b := a + b + ab ∈ R. Furthermore let G := R\{−1}.

(a) Show that a ∗ b ∈ G for all a, b ∈ G.

(b) Show that G together with the binary operation G × G → G, (a, b) 7→ a ∗ b, is a group.

Exercise 3

Let G = {s,t, u, v} be a group with s ∗ u = u and t ∗ t = v. Determine the group table of G. (There is only
one way of completing the group table for G. Give a reason for each step.)

Exercise 4

Write down the group tables for the groups C4 and C2 ×C2 (cf. Exercise 1). For every element a in C4
and C2 ×C2 determine the smallest positive integer m such that ma equals the identity element.

Extra question (not assessed — no need to submit)

Let G be a group whose binary operation is written additively, i.e. G × G → G, (a, b) 7→ a + b. Show
that m(na) = (mn)a for all a ∈ G and m, n ∈ Z. (Hint: You need to distinguish up to 9 cases.) Write
down the other two exponential laws in additive notation as well. (Formulate these laws as complete
mathematical statements including all quantifiers. No proofs are required.)

7.3 Coursework Sheet 3

Submit a single pdf with scans of your work to Blackboard by Monday, 19 February 2024, 17:00.

Exercise 1

Write down the group table for the permutation group S3 and show that S3 is not abelian. (You may find
it more convenient to write all elements of S3 in cycle notation.)
48 7 COURSEWORK SHEETS

Exercise 2
   
1 2 3 4 5 6 7 8 9 1 2 3 4 5 6
Let σ := ∈ S9 , τ := ∈ S6
3 7 6 1 8 9 4 2 5 4 2 5 6 3 1
 
1 2 ... n−1 n
and η := ∈ Sn (for any even n ∈ N).
n n−1 ... 2 1

(a) Determine the sign of σ , τ and η.

(b) Write σ 2 , σ −1 , τ 2 , τ −1 , η 2 and η −1 as a composition of cycles.

(c) Determine the sign of σ 2 , τ 2 and η 2 in two ways, firstly using (b) and secondly using (a) and
Theorem 1.10 (b).

Exercise 3

Let n ≥ 1. Let ⟨a1 , . . . , as ⟩ ∈ Sn be a cycle and let σ ∈ Sn be arbitrary. Show that

σ ◦ ⟨a1 , . . . , as ⟩ ◦ σ −1 = ⟨σ (a1 ), . . . , σ (as )⟩ in Sn .

(Note this is an equality between maps. Hence, in order to show this equality you need to show that both
sides are equal after applying them to an arbitrary element b of {1, 2, . . . , n}. To do so you will need to
distinguish whether b belongs to {σ (a1 ), . . . , σ (as )} or not.)

Exercise 4
√ √ √
Let Q( 5) denote the set of real numbers z of the form z = a + b 5 where a, b ∈ Q. Show that Q( 5)
together with the usual√addition and multiplication of real numbers is a field.
√ (Hint: You need to show
that√for any w, z ∈ Q( 5) also w + z, wz, −z and z−1 (if z ̸= 0) are in Q( 5) and that 0 and 1 are
√ in
Q( 5). Distributivity, commutativity and associativity for addition and multiplication hold in Q( 5)
because they hold in R.)

Exercise 5

Let F be a field. For any a, b ∈ F, b ̸= 0, we write a

b for ab−1 . Prove the following statements for any
a, a′ ∈ F and b, b′ ∈ F\{0}:

a a′ ab′ + a′ b a a′ aa′
(i) + = ; (ii) = .
b b′ bb′ b b′ bb′

Extra items for Exercise 5 (not assessed, do not submit):

a a′
(iii) = if and only if ab′ = a′ b;
b b′
a
ab′
(iv) b
a′
= (if in addition a′ ̸= 0).
b′
a′ b
 7 COURSEWORK SHEETS 49

Extra problems to think about (do not submit)

The solutions for this will not be provided (but possible to find in a book or google). Not necessary for
the rest of the module at all. Feel free to ignore.
Task 1. The aim is to prove Thm 1.10 from the notes, about the sign function on the symmetric groups
Sn . Here’s one possible path to a proof.

• Every cycle of length k can be written as a product of k − 1 transpositions.

• Thus, every permutation can be written as a product of transpositions.

• Let σ be a permutation, and write it as a product of transpositions. Define the number nsgn(σ )
(for “new sign’ ’) to be equal to 1 if the number of transpositions is even, and −1 if the number of
transpositions is odd. Again, apriori nsgn depends on how do we write σ as a product of transpo-
sitions. However, by the first point above, nsgn(σ ) = sgn(σ ), since every cycle decomposition of
σ gives also a way to write σ as a product of transpositions. So the goal now is to prove that nsgn
is well defined, and that it’s multiplicative.

• A way to prove the above is to find a way to characterise nsgn to be something intrinsic to a
permutation. Here’s such a thing: Given a permutation σ ∈ Sn , we say that σ reverses the pair
(i, j), if i, j ∈ {1, . . . , n}, i < j and σ (i) > σ ( j). Let isgn(σ ) be 1 if σ reverses even number of
pairs, and −1 if σ reverses odd number of pairs.

• Prove that if σ is a permutation and τ is a transposition, then isgn(σ ◦τ) = −isgn(σ ) = isgn(τ ◦σ ).

• From the previous point, conclude that isgn(σ ) = nsgn(σ ) (thus the sign is well defined).

• From the definition of nsgn, show that nsgn(σ ◦ τ) = nsgn(σ )nsgn(τ) for any two permutations
σ and τ.

Task 2. Prove that the number of elements of Sn (i.e. the order of the symmetric group Sn ) is n!.

7.4 Coursework Sheet 4

Submit a single pdf with scans of your work to Blackboard by Monday, 26 February 2024, 17:00.

Exercise 1
 
x1
The set R2
together with the usual vector addition forms an abelian group. For a ∈ R and x = ∈
x2
 
2 ax1
R we put a ⊗ x := ∈ R2 ; this defines a scalar multiplication
0

R × R2 → R2 , (a, x) 7→ a ⊗ x,

of the field R on R2 . Determine which of the axioms defining a vector space hold for the abelian group
R2 with this scalar multiplication. (Proofs or counterexamples are required.)
50 7 COURSEWORK SHEETS

Exercise 2

The set R>0 of positive real numbers together with multiplication forms an abelian group. Let Rn>0
denote the n-fold cartesian product of R>0 with itself (cf. Exercise 1 on Sheet 2). (You may find it
convenient to use the symbol ⊕ for the binary operation in the abelian group Rn>0 , that is (b1 , . . . , bn ) ⊕
(c1 , . . . , cn ) = (b1 c1 , . . . , bn cn ) for b1 , . . . , bn , c1 , . . . , cn ∈ R>0 .) Furthermore, for a ∈ Q and b = (b1 , . . . , bn ) ∈
Rn>0 we put a⊗b := (ba1 , . . . , ban ). Show that the abelian group Rn>0 together with the scalar multiplication

Q × Rn>0 → Rn>0 , (a, b) 7→ a ⊗ b,

is a vector space over Q.

Exercise 3

Let V be a vector space over the field F and let a ∈ F and x, y ∈ V .

(a) Show that a(x − y) = ax − ay in V .

(b) If ax = 0V show that a = 0F or x = 0V .

(Remember to give a reason for each step.)

Exercise 4

Let S be a set and let V be a vector space over a field F. Let V S denote the set of all maps from S to
V . We define an addition on V S and a scalar multiplication of F on V S as follows: let f , g ∈ V S and let
a ∈ F; then
( f + g)(s) := f (s) + g(s) and (a f )(s) := a ( f (s)) (for any s ∈ S).
Show that V S is a vector space over F. (For a complete proof many axioms need to be checked. In order
to save you some writing, your solution will be considered complete, if you check that there exists an
additive identity element in V S , that every element in V S has an additive inverse and that the second
distributivity law holds.)

7.5 Coursework Sheet 5

Submit a single pdf with scans of your work to Blackboard by Monday, 11 March 2024, 17:00.

Exercise 1

Let n ≥ 2. Which of the conditions defining a subspace are satisfied for the following subsets of the
vector space Mn×n (R) of real (n × n)-matrices? (Proofs or counterexamples are required.)

U := {A ∈ Mn×n (R) | rank(A) ≤ 1}

V := {A ∈ Mn×n (R) | det(A) = 0}
W := {A ∈ Mn×n (R) | trace(A) = 0}

(Recall that rank(A) denotes the number of non-zero rows in a row-echelon form of A and trace(A)
denotes the sum ∑ni=1 aii of the diagonal elements of the matrix A = (ai j ).)
 7 COURSEWORK SHEETS 51

Exercise 2

Which of the following subsets of the vector space RR of all functions from R to R are subspaces?
(Proofs or counterexamples are required.)

U := { f ∈ RR | f is differentiable and f ′ (−5) = 0}

V := { f ∈ RR | f is polynomial of the form f = at 2 for some a ∈ [0, ∞)}
= { f ∈ RR | ∃a ∈ [0, ∞] : ∀s ∈ R : f (s) = as2 }
W := { f ∈ RR | f is polynomial of the form f = at 3 or f = at 5 for some a ∈ R}
= { f ∈ RR | ∃i ∈ {3, 5} ∃a ∈ R : ∀s ∈ R : f (s) = asi }
X := { f ∈ RR | f is even}

(Recall that a function f : R → R is called even if f (−s) = f (s) for all s ∈ R.)

Exercise 3

Let F2 = {0, 1} denote the field with 2 elements.

(a) Let V be a vector space over F2 . Show that every non-empty subset W of V which is closed under
addition is a subspace of V .
(b) Show that {(0, 0), (1, 0)} is a subspace of the vector space F22 over F2 .
(c) Write down all subsets of F22 and underline those subsets which are subspaces. (No explanations
are required.)

Exercise 4

Let V be a vector space over a field F and let X,Y and Z be subspaces of V , such that X ⊆ Y . Show that
Y ∩ (X + Z) = X + (Y ∩ Z). (Note: this is an equality of sets, so you need to show that every vector in
the LHS also belongs to RHS, and vice versa.)

Extra question (not marked, do not submit)

Let V be a vector space over a field F. Putting S = V in Example 2.7 we obtain the vector space F V
consisting of all functions from V to F. Consider the subset

V ∗ := {L : V → F | L is a linear transformation},

consisting of all linear transformations from the vector space V to the (one-dimensional) vector space
F. Show that V ∗ is a subspace of F V . (To get you started, at the end of this sheet you’ll find a detailed
proof of the first of the three conditions that need to be verified for a subspace.)

Verification of the first condition of being a subspace, for V ∗ from the above Exercise

(You don’t need to reproduce this in your solution, just say that the first condition is proved.)
The first condition for a subspace asserts that the zero vector of the “big” vector space F V belongs to set
V ∗ that we are showing to be a subspace.
The zero vector ( = the additive identity element for vector addition) of F V is the zero function 0 : V → F,
defined by 0(v) = 0F for all v ∈ V , that is, it maps every vector v from V to the additive identity element
0F in the field F.
52 7 COURSEWORK SHEETS

We need to show that this function 0 belongs to the set V ∗ , in other words, that it is a linear transformation
from V to F. This entails checking two conditions:

(a) 0 is compatible with addition: take arbitrary vectors x, y ∈ V . We need to check that 0(x + y) =
0(x) + 0(y) in F:
LHS = 0F (by definition of 0)
RHS = 0F + 0F = 0F (by definition of 0 and the field axioms)
So LHS = RHS.
(b) 0 is compatible with scalar multiplication: take a vector x ∈ V and a scalar a ∈ F. We need to
check that 0(ax) = a(0(x)) in F:
LHS = 0F (by definition of 0)
RHS = a0F = 0F (by definition of 0 and Prop. 2.3(a))
So again LHS = RHS.

7.6 Coursework Sheet 6

Submit a single pdf with scans of your work to Blackboard by Monday, 18 March 2024, 17:00.

Exercise 1

Which of the following are spanning sets for the vector space P2 of polynomial functions of degree at
most 2? (Give reasons for your answers.)
1 2 2
(a) 2 , t + t, t − 1

(b) 1, 2t, t 2 , 3t 2 + 5
(c) t + 1, t 2 + t

Exercise 2

Determine whether the following are linearly independent sets of vectors in the vector space RR of all
functions from R to R. (Give reasons for your answers.)

(a) 1 + t, 1 + t + t 2 , 1 + t + t 2 + t 3 , 1 + t + t 2 + t 4
(b) sin, cos2 , sin3
(c) 1, sin2 , cos2

(Here for example sin2 denotes the function R → R, s 7→ (sin(s))2 .)

Exercise 3

Find a basis of the null space N(A) ⊂ R5 of the matrix

 
1 −3 3 −1 −1
A =  −2 6 −1 −3 −8  ∈ M3×5 (R)
3 −9 10 −4 −5
and hence determine its dimension.
 7 COURSEWORK SHEETS 53

Exercise 4

(a) Determine whether the following (2 × 2)-matrices form a basis of the vector space M2×2 (R) of all
(2 × 2)-matrices over R:
       
1 0 2 1 3 2 4 3
A1 = , A2 = , A3 = , A4 = .
0 0 0 0 1 0 2 1

(b) Find a basis of the subspace W := {A ∈ M2×2 (R) | trace(A) = 0} of the vector space M2×2 (R)
and hence determine the dimension of W . (Recall that trace(B) of a square matrix B = (bi j ) ∈
Mn×n (F) denotes the sum of its diagonal entries, trace(B) = ∑ni=1 bii .)

Extra exercise (not marked, do not submit)

  
w
We view C2
= : w, z ∈ C as a vector space over C, R and Q (cf. Example 3.16 (b)). Let x1 :=
z
   √     √   
i √2 0 i√ 3 1
, x2 := , x3 := , x4 := , x5 := ∈ C2 . Determine dimF (SpanF (x1 , x2 , x3 , x4 , x5 ))
0 5 1 3 3
for F = C, R and Q.

7.7 Coursework Sheet 7

Submit a single pdf with scans of your work to Blackboard by Monday, 22 April 2024, 17:00.

Exercise 1

Determine whether the following maps are linear transformations. (For a matrix A, AT denotes its
transpose, see Section 2.3 in L.A.I.) (Proofs or counterexamples are required.)
 
  x2  
x1 x1
(a) L : R2 → R3 , 7→  x1 + 2x2  (b) L : R2 → R, 7→ x12 + x1 x2
x2 x2
x1 − x2

(c) L : Mn×n (R) → Mn×n (R), A 7→ AT − A (d) L : P3 → P2 , f 7→ 2 f ′ + ( f (3))t 2

Exercise 2

Consider the linear transformation R3 → R4 given by L(x) = Ax where A is the matrix

 
1 0 3
 2 1 2 
A=  ∈ M4×3 (R).
 1 2 −5 
2 1 1

Find a basis of the image of L. Using the Dimension Theorem show that L is injective.
54 7 COURSEWORK SHEETS

Exercise 3

Let F be a field.

(a) Let A ∈ Mn×n (F) be an invertible matrix. Show that the linear transformation

LA : F n → F n , x 7→ Ax,

(cf. Example 4.3(a)) is an isomorphism.

(b) Let L : V → W be an isomorphism between vector spaces over F. Show that the inverse map
L−1 : W → V is a linear transformation (and hence an isomorphism as well).

Exercise 4

For y ∈ Rn let Ly : Rn → R denote the map given by x 7→ Ly (x) = x · y where x · y denotes the dot product
of x and y introduced in Linear Algebra I.

(a) For each y ∈ Rn show that Ly is a linear transformation and compute dimR (ker(Ly )).

(b) Let (Rn )∗ denote the vector space introduced in Coursework 5/Extra Question, that is

(Rn )∗ = {T : Rn → R | T is a linear transformation}.

n
(The task in the Extra Question was to show that it is a subspace of RR , and thus that it is a vector
space over R. For this exercise, you can assume that it is so.)
Show that the map L : Rn → (Rn )∗ , y 7→ Ly , is an isomorphism. (Hint: For surjectivity use
Proposition 4.4.)

7.8 Coursework Sheet 8

Submit a single pdf with scans of your work to Blackboard by Monday, 6 May 2024, 17:00.

Exercise 1

From Calculus we Rknow that for any polynomial function f : R → R of degree at most n, the function
I( f ) : R → R, s 7→ 0s f (u) du, is a polynomial function of degree at most n + 1. Show that the map

I : Pn → Pn+1 , f 7→ I( f ),

is an injective linear transformation, determine a basis of the image of I and find the matrix M ∈
M(n+2)×(n+1) (R) that represents I with respect to the basis 1,t, . . . ,t n of Pn and the basis 1,t, . . . ,t n+1
of Pn+1 .

Exercise 2
 
1−i α i
(a) Let α ∈ C and A :=  i − α 1 − α α − i  ∈ M3×3 (C). Compute det(A) ∈ C.
1−α 1 2+α
 7 COURSEWORK SHEETS 55

(b) Let F be a field, n be even and let c1 , . . . , cn ∈ F. Follow the blueprint of the proof of Example
5.2(d) and use Exercise 2(a) on Coursework Sheet 3 to compute the determinant of the matrix
 
0 . . . 0 c1
 .. 
 . · · 0 
B :=  ..  ∈ Mn×n (F).
 
 0 · · . 
cn 0 . . . 0

Exercise 3

Let F be a field, let n ≥ 1 and let a, b ∈ F. Furthermore let

 
b a a ... a

 a b a ... a 

A := 
 a a b ... a  ∈ Mn×n (F).

 .. .. .. .. 
 . . . . 
a a a ... b

Show that det(A) = (b + (n − 1)a)(b − a)n−1 . (Hint: Begin with R1 7→ R1 + R2 + . . . + Rn , apply Theorem
5.3(b) and use further row operations to arrive at an upper triangular matrix.)

Exercise 4
   
1+i 1−i 2 2 3+i 2i
Let A =  3 i −i  ∈ M3×3 (C) and B =  1 2 − i 2 + 2i  ∈ M3×3 (C).
1 2−i 2+i 1−i i 3
Compute det(A), det(B), det(AB) and det(A3 ).

7.9 Coursework Sheet 9

Submit a single pdf with scans of your work to Blackboard by Monday, 13 May 2024, 17:00.

Exercise 1

Let F be a field and let A ∈ Mn×n (F).

(a) If n = 2 show that pA (λ ) = λ 2 − trace(A)λ + det(A). (Recall that trace(B) of a square matrix
B = (bi j ) ∈ Mn×n (F) denotes the sum of its diagonal entries, trace(B) = ∑ni=1 bii .)

(b) Let k ≥ 1. Show that if λ is an eigenvalue of A then λ k is an eigenvalue of Ak .

(c) Suppose that F = Q, R or C and that A2 = In . Show that if λ is an eigenvalue of A then λ = 1

or λ = −1. Show that ker(LIn +A ) = E−1 (A) and that im(LIn +A ) = E1 (A). (Note: The notation
“Lmatrix ” is from Example 4.3 (a).)
56 7 COURSEWORK SHEETS

Exercise 2

Let F be a field and let A ∈ Mn×n (F) be a diagonalizable matrix.

(a) Let k ≥ 1. Show that Ak is diagonalizable.

(b) Show that the transpose AT of A is diagonalizable.

(c) Show that if A is invertible then A−1 is diagonalizable.

Exercise 3

Find the eigenvalues of each of the following matrices and determine a basis of the eigenspace for each
eigenvalue. Determine which of these matrices are diagonalizable; if so, write down a diagonalizing
matrix.  
2 2 1  
3 −1
A =  1 3 1  ∈ M3×3 (R), B = ∈ M2×2 (R),
1 5
1 2 2
 
1 0 0
C =  1 −1 2  as element of M3×3 (R) and as element of M3×3 (C).
1 −1 1
Compute C2020 .

Exercise 4

Let V be a vector space over a field F and let L, M be two linear transformations from V to itself.

(a) Suppose that L ◦ M = M ◦ L. Show that L(Eλ (M)) ⊆ Eλ (M) for all λ ∈ F.

(b) Suppose that V is of finite dimension. Show that L is injective if and only if it is surjective.