Linear algebra I

Remi Abgrall
September 23, 2019

Contents
1 Construction of Z 1
1.1 Equivalence relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Construction of Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Groups 4
2.1 Definitions and first properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Morphisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Product of groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 Subgroups, caracterisation of sub groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5 Other properties of groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.6 Distinguished groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.7 Quotient groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.7.1 Equivalence relations that are compatible with the law of a group . . . . . . . . . . . 10
2.7.2 Quotient groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.7.3 A fundemental example Z/nZ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.8 The symetric group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1 Construction of Z
1.1 Equivalence relations
Let us recall what is an equivalence relation. If E is a set, R is a relation if, for any (x, y) 2 E ⇥ E, xRy is
either true or false: this is a predicat. We say that R is
• Reflexive if and only if 8x 2 E, xRx,
• Symetric if and only if 8x, y 2 E,
xRy ) yRx.

• antisymetric if and only if for any x, y 2 E,

xRy and yRx ) x = y

• transitive if and only if 8(x, y, z) 2 E 3

xRy and yRz ) xRz.

Definition 1.1 (Equivalence relation). R, a relation on E is an equivalence relation if and only if it is

reflexive, symetric and transitive.

1
 Linear algebra I
Remi Abgrall
September 23, 2019

Contents
1 Construction of Z 1
1.1 Equivalence relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Construction of Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Groups 4
2.1 Definitions and first properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Morphisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Product of groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 Subgroups, caracterisation of sub groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5 Other properties of groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.6 Distinguished groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.7 Quotient groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.7.1 Equivalence relations that are compatible with the law of a group . . . . . . . . . . . 10
2.7.2 Quotient groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.7.3 A fundemental example Z/nZ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.8 The symetric group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1 Construction of Z
1.1 Equivalence relations
Let us recall what is an equivalence relation. If E is a set, R is a relation if, for any (x, y) 2 E ⇥ E, xRy is
either true or false: this is a predicat. We say that R is
• Reflexive if and only if 8x 2 E, xRx,
• Symetric if and only if 8x, y 2 E,
xRy ) yRx.

• antisymetric if and only if for any x, y 2 E,

xRy and yRx ) x = y

• transitive if and only if 8(x, y, z) 2 E 3

xRy and yRz ) xRz.

Definition 1.1 (Equivalence relation). R, a relation on E is an equivalence relation if and only if it is

reflexive, symetric and transitive.

1
Notation: xRy or x ⌘ y mod R.
Definition 1.2 (Equivalence class, quotient set). Let R be an equivalence relation. For any x 2 E, we call
equivalence class of x modulo R the set
x = {y 2 E, xRy}.
The set of equivalence classes is called the quotient set by R, denoted as E/R.
we easily see that

• none of the equivalence class is empty: xRx: x 2 x.

• Any y 2 E belongs to an equivalence class: again y 2 y.
• For any x, y 2 E, x \ y = ; or x = y.
An example of relation and equivalent class is given through the construction of Z from N and Q from Z.

1.2 N
Let us recall what is admitted for N
Definition 1.3 (Peano’s Axiom). There exists a set denoted by N containing an element denoted by 0 and
on which there exists an injective application ⌫ : N ! N such that
1. 0 62 range ⌫
2. If A is a subset of N, 0 2 A and such that

n 2 A ! ⌫(n) 2 A

then A = N
We will denote ⌫(n) = n + 1, and ⌫(0) = 1, ⌫(1) = 2, etc
The set N? is N {0}

One can construct an order on N: we say that p  q if:

• either p = q
• or q can be written as q = ⌫(⌫(nu(. . . ⌫(k)))) = ⌫ ⌫ . . . ⌫(k)
On can also define an operation denoted by +: if p,q 2 N:

p = ⌫| ⌫ {z. . . ⌫}(0), q = ⌫| ⌫ {z. . . ⌫}(0), then p + q = ⌫| ⌫ {z. . . ⌫} ⌫| ⌫ {z. . . ⌫}(0).

p times q times p times q times

Clearly, p + q = q + p, p + (q + r) = (p + q) + r and 0 + p = p + 0 = p from the properties of the composition

of functions.
Similarly, p ⇥ q is defined by
p ⇥ q = p + p + ... + p
| {z }
q times

and we see that p ⇥ q = q ⇥ p: this is because we compose ⌫ with itself and evaluate on 0. Then 1 ⇥ p = p.

2
1.3 Construction of Z
On N ⇥ N we define the relation:

(p, q) ⇠ (m, n) ! p + n = m + q.

We define Z = N ⇥ N/ ⇠. The mapping

proj : N!Z
m 7! [(m, 0)]

is an injection
Proof. Let m, n 2 Z such that proj(m) = proj(n) This means that [(m, 0)] = [(n, 0)]: (m, 0) 2 [(n, 0)] :
m + 0 = n + 0 so that m = n.
Thus we can identify N and proj(N), using this we seen that N ⇢ Z. We also define Z? = Z {0}.
We can extend the operation + and the order of N:
• Operation +: On N ⇥ N we define

(m, n) + (p, q) = (m + p, n + q).

We see that (m, n) ⇠ (m0 , n0 ) and (p, q) ⇠ (p0 , q 0 ) implies (m + p, n + q) ⇠ (m0 + p0 , n0 + q 0 )

Proof. We have
m + n0 = n + m0 and p + q 0 = q + p0
so that
(m + n0 ) + (p + q 0 ) = (n + m0 ) + (q + p0 )
i.e., because of the associativity in N,

(m + p) + (n0 + q 0 ) = (n + q) + (m0 + p0 )

that is (m + p, n + q) ⇠ (m0 + p0 , n0 + q 0 ).

This operation is called addition and has the following properties:

– Commutativity: [(m, n)] + [(p, q)] = [(p, q)] + [(m, n)]
– Associativity: [(a, b)] + [(m, n)] + [(p, q)] = [(a, b)] + [(m, n)] + [(p, q)].
– for any [(m, n)], [(m, n)] + [(0, 0)] = [(m, n)]: [(0, 0)] is a neutral element
– For any [(m, n)], [(m, n)] + [(n, m)] = [(0, 0)]: any element of Z has an inverse. If x 2 Z the inverse
of x is denoted by x.
It is said that (Z, +) is a commutative group

• Order : we say that a, b 2 Z are such that a  b if b a 2 N. This is a total order. For this we need
to show that if x 2 Z then either x 2 N of x 2 N.

Proof. Le x 2 Z, x = [(m, n)] Since the order on N is total, we have either m  n or n  m. If m  n,

then n m 2 N and x = ([m, n]) = [(0, n m)]: x 2 N.
If n  m, then m n 2 N and x = [(m, n)] = [(m n, 0)]: x 2 N
Hence the order is total.

3
 One can define a multiplication:s

(n1 , n2 ) ⇥ (m1 , m2 ) = (n1 m1 + n2 m2 , n1 m2 + m1 n2 )

It is compatible with the relation ⇠: if (n1 , n2 ) ⇠ (n01 , n02 ) and m1 , m2 ) ⇠ (m01 , m02 ) then (n1 , n2 )⇥(m1 , m2 ) ⇠
(n01 , n02 ) ⇥ (m01 , m02 ).
This enable to define a multiplication on Z:

[(n1 , n2 )] ⇥ ([m1 , m2 )] = [(n1 m1 + n2 m2 , n1 m2 + m1 n2 )]

Note that

[(n, 0)] ⇥ [(m, 0)]] = [(mn, 0)], [(0, n)] ⇥ [(m, 0)] = ([0, mn)], [(0, n)] ⇥ [(0, m)] = [(mn, 0)]

i.e. for n, m 2 N
n ⇥ m = nm, n ⇥ ( m) = (nm), ( n) ⇥ ( m) = nm.
One can also show that: for any n, m, p 2 Z, we have
• 1 is the neutral element: 1 ⇥ m = m,
• Commutativity: n ⇥ m = m ⇥ n,

• Associativity:
n ⇥ (m ⇥ p) = (n ⇥ m) ⇥ p

• the distributivity with respect to +:

n ⇥ (m + p) = n ⇥ m + n ⇥ p.

(Z, +, ⇥) is an integer ring: if x ⇥ y = 0 then either x = 0 or y = 0.

In what follows, n ⇥ m is denoted as nm. We denote by Z? = Z {0}.

2 Groups
2.1 Definitions and first properties
Definition 2.1. Let G be a set and ? an ”operation”, i.e. an application defined from G ⇥ D onto G:

8(x, y) 2 G ⇥ G, x ? y 2 G

that has the following properties:

1. For any x, y, z 2 G, (x ? y) ? z = x ? (y ? z): ? is associative
2. There exists e 2 G such that for any x 2 G, x ? e = e ? x = x: e is a neutral element

3. for any x 2 G, there exists y 2 G such that x ? y = y ? x = e: y is an inverse of x

is called a group.
If in addition, for any x, y 2 G, x ? y = y ? x, the group G is said to be commutative (or Abelian).
When there is no ambiguity, we talk about the group G instead of the group (G, ?).
We have the following properties:

1. The neutral element is unique:

4
 Proof. Assume that we have two neutral elements e and e0 . Then

e0 = e ? e0 = e0 ? e = e.

2. Let x 2 G with two inverses, y and y 0 . Then y = y 0 .

Proof. We have e = x ? y = x ? y 0 . Then

y = y ? e = y ? (x ? y 0 ) = (y ? x) ? y 0 = e ? y 0 = y 0 .

1
Since the inverse of x is unique, it is denoted by x
1
3. Let a, b 2 G. There exists a unique x 2 G such that a ? x = b, this is x = a ? b.
Indeed, if there is a solution,
1 1 1
a ?b=a ? (a ? x) = (a ? a) ? x = x

and we see that

1 1
a ? (a ? b) = (a ? a ) ? b = e ? b = b.

Example 2.2 (Examples of group, and other sets that are not group).
• (Z, +) is a group, The neutral is 0

• (N, +) is not a group: 1 has no inverse for +

• (Q, +) is a group
• (Q 0, ⇥) is a group, the neutral is 11
• (Q, ⇥) is not a group: there is no x 2 Q such that x ⇥ 0 = 1

• Let G = {e} with the operation e ? e = e: this is an Abelian group

• Let G = {e, a} with the operation e ? e = e e ? a = a ? e = a is an Abelian group
We have define xn .

• If n > 0, x1 = x and xn+1 = xn ? x

1
• if n < 0, xn = x n

• If n = 0, x0 = e
and we see that for any n, m 2 Z, xn+m = xn ? xm .
When a group is Abelian, we will often write the law ? as +. The neutral element will be denoted by 0,
and denote the inverse of x 2 G as x. Similarly xn will be denoted by nx
1I will often denote by Q? , R?, C? , the set Q, R, C without 0

5
2.2 Morphisms
Definition 2.3. Let (G, ?) and (F, •) two groups. An application f : G ! F is called a group morphism if
for any x, y 2 G,
f (x ? y) = f (x) • f (y).

Example 2.4 (Example of morphisms).

• For (G, ?) = (Z, +), and a 2 Z, let us define

f: Z!Z
x 7! na(= n ⇥ a)

Then for any x, y 2 Z, f (x + y) = f (x) + f (y): this is a group morphism.

• For (G, ?) = (Q? , ⇥), and a 2 Z let us define f (x) = xa we clearly have f (xy) = f (x)f (y)
We have the following properties:

• If e is the neutral in G and e0 that of F , f (e) = e0

Proof. We have e ? e = e, so f (e) • f (e) = f (e). Let us consider (f (e)) 1 the inverse of f (e) in F .
✓ ◆ ✓ ◆
e0 = (f (e)) 1 • f (e) = (f (e)) 1 • f (e) • f (e) = (f (e)) 1 • f (e) • f (e) = f (e).

Note. Taking for any x 2 G the fact that x ? e = x implies f (x) • f (e) = f (x) and only this (i.e. no
inversion) is not enough since we do not assume that for any y 2 F , there exists a x 2 E with f (x) = y.
1 1
• for any x 2 E, f (x ) = f (x) .

Proof. for any x, e0 = f (e) = f (x ? x 1

) = f (x) • f (x 1
). Since f (x) has only one inverse, f (x 1
)=
1
f (x) .
n
• For any x 2 G, any n 2 Z, f (xn ) = f (x) .

n
Proof. The proof is by induction for n 0, and then we use the fact that x = (xn ) 1

In the case of Abelian groups, f (0E ) = 0F , and since in most cases, there is no ambiguity, 0 will be 0E or
0G . We also have f ( x) = f (x), f (nx) = nf (x).

2.3 Product of groups

Definition 2.5. Let {Gi }i2I a family of groups ((Gi , •). We consider G = ⇥i2I Gi the cartesian product,
and we define the law: For any x, y 2 G, x = (xi )i2I , y = (yi )i2I ,

x ? y = z, z = (zi )i2I , zi = xi • yi

Theorem 2.6. (G, •) is a group with neutral element 0 = (0Gi )i2I

The proof is left as an exercise.

6
2.4 Subgroups, caracterisation of sub groups
Definition 2.7. Let (G, ·) be a group. We call sub-group of (G, ·) any group (H, ·) where H ⇢ G and is
stable by the law ·:
8(x, y) 2 H, x · y 2 H.

We can consider the canonical injection i : H ! G defined by i(x) = x. This is a group morphism. The
neutral of H is the neutral of G. The inverse of any x 2 H, when H is seen as a group, is the inverse of x
when x 2 H ⇢ G is seen as an element of G: x has the same inverse in H or G. . . Fortunately.
Theorem 2.8 (Characterisation of a subgroup). Le G be a group, e 2 G is neutral, and H ⇢ G. The
following properties are equivalent:

1. H is a subgroup of G,
1
2. H is stable by ·, and e 2 H and 8x 2 H, x 2 H,
1
3. H is stable by ·, H 6= ;, and 8x 2 H, x 2 H,

4. H 6= ; and 8(x, y) 2 H 2 , xy 1
2 H.
Proof. • (1) implies (2): this is what has been written above
• (2) implies (3) and (3) implies (4) are clear
1
• (4) implies (2) Since H =
6 ;, there exists a 2 H, and so e = aa 2 H. Then taking x = e and any x,
we have that x 1 = ex 1 2 H.
1 1 1
Last H is stable: for any x, y 2 H, we have first y 2 H and hence xy = x y 2H
• (2) implies (1): For any x, y 2 H, we can define a law (denoted by • for 2 seconds such that:

x • y = xy

This law is an internal law: we need to check that x • y 2 H, and this is true because if x, y 2 H,
x · y 2 H: H is stable. In addition the element e of H (the neutral of · on G is such that for any x 2 H
e • x = e · x = x = x · e = x • e: this is the neutral for •. Last x • (y • z) = (x • y) • z: (H•) is a group.

2.5 Other properties of groups

Theorem 2.9 (Intersection of groups). Let G be a group and (Hi )i2I a family of subgroups of G. Then the
intersection H = \i2I Hi is a subgroup of G
Proof. Since e 2 Hi for any i 2 I, H = \i2I Hi 6= ;. In addition, for any x, y 2 H, x, y 2 Hi , hence
xy 1 2 Hi , hence, xy 1 \i2I Hi : H is a subgroup of G.
Definition 2.10. We call subgroup of G generated by A ⇢ G, and denote it by gr(A), the intersection of
all the groups of G that contain A.
If A is such that gr(A) = G, we say that A generates G. By convention gr(;) = {e}
Theorem 2.11. Let G be agroup with neutral e, and A ⇢ G. Let us consider H ⇢ G which elements are
either e or can be written as:
a"11 a"22 . . . a"nn
with "i 2 {±1}. Then H is the subgroup of G generated by A.

7
Proof. H is not empty.
"0 "0 "0
If x 2 H and y 2 H, then we have x = a"11 a"22 . . . a"nn and b = b11 b22 . . . bmm with ai , bj 2 A and "j = ±1
as well as "0j . We have

"0 "0 "0 1

xy 1
= a"11 a"22 . . . a"nn b11 b22 . . . bmm
"0 "0 "01
= a"11 a"22 . . . a"nn bm 1 bm 21 . . . b1
"0 "0 "01
= a"11 a"22 . . . a"nn bm 1 bm 21 . . . b1 2H

The converse is true: if gr(A) contains e and all the elements of the type a"11 a"22 . . . a"nn , with "i 2 {±1}.
This subset is a group as we have shown previously, so H = gr(A).

Remark 2.12. If H1 and H2 are subgroups of G, H1 [ H2 is not a subgroup og G in general.

For example, in G = Z with +. Let

H1 = {x 2 Z, 9p 2 Z, x = 2p} and H2 = {x 2 Z, 9p 2 Z, x = 3p}.

H1 and H2 are subgroups of Z, but H1 [ K2 is not a subgroup of Z because + is not stable: 2 + 3 = 5 is

neither a multiple of 2 or 3, 5 62 H1 \ H2 .
Theorem 2.13 (Image of a morphism, reciprocical image of a subgroup by a morphism). Let (G, ·) and
(F, ·) be groups, f : G ! F a morphism. Then
• If H is a subgroup of G, f (H) is a subgroup of F

• if H ⇢ F is a subgroup of F then
1
f (H) = {x 2 G, 9y 2 H, y = f (x)}

is a subgroup of G
1 1 1
Proof. • f (H) contains f (e) = e, so it is not empty. If x, y 2 H, f (x)f (y) = f (x)f (y ) = f (xy )2
f (H): this is a subgroup of F .
1
• Since eF = f (eE ), if H is a subgroup of F , eF 2 H hence eE 2 f (H): This is a non empty set.
Let x, y 2 f (H). There exists x and y such that x = f (x ) and y = f (y 0 ). Then xy
1 0 0 0 1
=
f (x0 (y 0 ) 1 ) 2 f 1 (H).

There are two important examples: The image of f , Imf = {z 2 F, 9x 2 G, z = f (x)} and the kernel of
f, Kerf = f 1 ({0F }) = {x 2 G, f (x) = 0F }. We have the foloowing important result
Theorem 2.14 (Necessary and suffisant condition of injectivity of a morphism). Let f : G ! F a group
morphism. f is injective if and only if Kerf = {0E }.

Proof. • Assume that f is injective. Since f (eE ) = eF , we know that {eE } 2 textKerf . It there exists
x 6= eE in Ker f , ther would be two di↵erent solutions to f (z) = eF , whhich is in contradiction with
the injectivity property.
1 1
• Assume that Kerf = {eE }. Take x, y 2 E such that f (x) = f (y). Then f (x) f (y) = f (x)f y =
f (xy 1 ) = e, this means that xy 1 2 Ker f = {eE }, i..e x = y.

Theorem 2.15 (Necessary and suffisant condition of surjectivity of a morphism). Let f : G ! F a group
morphism. f is surjective if and only if Im f = F .

8
Proof. • If f is surjective, for any y 2 F there exists x 2 E wth f (x) = y: F ⇢ Im f ⇢ F , hence
F = Im f
• If F = Im f , for any y 2 F there exists x 2 E such that f (x) = y: f is surjective.

We see that a morphism is bijective if and only if Kerf = {eE } and Im f = F . We say that f is an
isomorphism. When E = F an isomorphism is also called an automorphism.

2.6 Distinguished groups

Let G be a group. For any a 2 G, we can define an automorphism 'a by
1
'a (x) = axa .

This is also called an interior automorphism. We have:

1. 'e = IdE ,
2. for any a, b 2 G, 'a 'b = 'ab

Proof.
1 1 1 1 1 1 1
'a 'b (x) = 'a 'b (x) = 'a bxb = a bxb a = a bxb a = a bx(ab) = ab x(ab) = 'ab (x).

3. From this we see that 'a 'a 1 = IdG : 'a is a bijection

1 1 1 1 1
4. For any x, y, and a 2 G, 'a (x)'a (y) = axa aya = axa aya = axya = 'a (xy): 'a is an
automorphism of G
If we denote by Aut(G) the set of automorphisms defined on G with the law , it is easy to see it is a
group (not necessarily commutative, enev if G is commutative), and

: G ! Aut(G)
a 7! 'a

is a morphism between G and Aut(G). We can look for its kernel

Ker = {a 2 G, 8x 2 G, ax = xa}

This is called the center of G, denoted by Z(G): the set of elements that commute with all elements of G.
When a group is commutative, Z(G) = G.
If H is a subgroup of G, and a 2 G, 'a (H) is a subgroup of G, denoted by aHa 1 .
Definition 2.16 (Distinguished groups). A subgroup H of G is said to be distinguished if it is stable by any
automorphism 'a :
8a 2 G, aHa 1 = H, i.e. , 8x 2 H, axa 1 2 H.
Proposition 2.17. If H is a distinguished subgroup of G, it is stable and invariant by any interior auto-
morphism.
Proof. Let H be a distinguished subgroup of H, for any a 2 G we already have aHa 1 ⇢ H. Let us show
that H ⇢ aHa 1 .
Let be any (a, h) 2 G ⇥ H. Since 'a 'a 1 Id G , we see that h is the image by 'a of 'a 1 (h) 2 H because
H is stable by 'a 1 . This shows that H ⇢ aHa 1 , and hence aHa 1 = H.

9
Remark 2.18.
• {e} is a distinguished subgroup
• Any subgroup of an abelian group is distiguished.
Theorem 2.19. Let f : G ! F be a group morphism. If H is a distinguished subgroup of G, f (H) is a
distinguished subgroup of f (G). If H 0 is a distinguisged subgroup of F , f 1 (H 0 ) is a distinguished subgroup
of G
Proof. We only prove the second claim, the first one is obvious.
Let x 2 f 1 (H 0 ) anda 2 G. since f (axa 1 ) = f (a)f (x)f (a 1
) 2 H 0 because f (x) 2 H 0 and H 0 is
distinguished, axa 1 ) 2 f 1 (H 0 ) which is distinguished.
As a corrollary, we get that Kerf is distinguished. If one wants to show that H ⇢ G is distinguished,
one way of doing so is to identify f : G ! F for a suitable F such that H = Ker f .

2.7 Quotient groups

2.7.1 Equivalence relations that are compatible with the law of a group
If G is a group and R an equivalence relation, we say it is compatible on the left with the composition law
(x · y) if for any z 2 G and any x, y such that xRy then zxRzy. It is compatible on the right if xzRyz.
Theorem 2.20. Let G be a group. Any equivalence relation that is compatible on the left (resp. on the
right) with the composition law in G is of the form x 1 y 2 H (resp yx 1 2 H) where H is a subgroup of G.
Conversely, Any relation of this type is an equivalence relation compatible with the composition in G.
Proof. We will assume that the equivalence relation is compatible on the left. Let H = e. Note that if xRy
then x 1 xRx 1 y, i.e. eRx 1 y. Since the relation is symmetric, we see that xRy implies x 1 y 2 H: Since
e 2 H it is non empty, and also stable.
Conversely, define H the relation xHy if and only if x 1 y 2 H, and we assume that H is a group. This
is an equivalence relation:
1
• it is reflexive: x x = e 2 H so xHx
• it is symetric:
1 1 1 1
xHy ) x y2H)y x= x y 2 H ) yHx.
• It is transitive: xHy and yHz is equivalent to
1 1 1 1 1
x y 2 H and y z 2 H ) (x y)(y z) = x z2H
because H is stable
This shows it is an equivalence relation
Definition 2.21 (Left and right classes). Let G be a group, H a subgroup, a 2 G. The sets
aH = {ah|h 2 H} and Ha = {ha|h 2 H}
are called left and right classes of a, according to H. These are equivalent classes for the relations Rl and
Rr defined by:
8(x, y) 2 G2 , xRl y () x 1 y 2 H,
8(x, y) 2 G2 , xRr y () yx 1
2 H.
Exercise 1. Let G be a finite group (a group with a finite number of elements, and H a subgroup of G.
1. Show that H is a finite group,
2. Show that |H| divides |G|

10
2.7.2 Quotient groups
The question here is the following: having an equivalence relation, we have the set of equivalence classes
G/R. We have a natural injection:
i G ! G/R
x 7! x
A priori, i has no special property. However we can ask the question: under which condition(s), can we
define a group structure on G/R so that i becomes a group morphism? The answer is given in the following
theorem
Theorem 2.22. Let G be a group. The equivalence relations that are compatible with the group structure of
G are those of the form x 1 y 2 H where H is a distinguished subgroup.
Before proving the result, let us first explain in more detail the statement. Saying that an equivalence is
compatible with the group structure of G means that

xRy and x0 Ry 0 ) (xx0 )R(yy 0 ). (1)

Proof. • Let R be an equivalence relation satisfying (1), and let H be the class of e. We show that H is
a distiguished group: First it is a subgroup of G. Second, Let x, x0 , y, y 0 as in (1):

x 1
y 2 H and (x0 ) 1 0
y 2 H ) (xx0 ) 1
(yy 0 ) 2 H.

Since x0 Rx0 , we see that xx0 Ryx0 , i.e. (xx0 ) 1 yx0 = (x0 ) 1 x1 y x0 2 H. When we particularise this
to x and y = e, we see that first x 2 H, and that for any x0 2 G, (x0 ) 1 xx0 2 H, which means that H
is a distinguished group.

• Assume now that H is a distinguished group, and consider xRy and aRb. Let us show that (xa)R(yb0 ).
We have
(xa) 1 (yb) = a 1 x 1 y a a 1 b).
Since aRb, a 1 b 2 H. Since xRy, x 1 y 2 H. Since H is distinguished, a 1
x 1
y a 2 H. Since H is
a group, it follows that (xa) 1 (yb) 2 H.

Theorem 2.23 (Theorem and definition). Let G be a group, H a distinguished subgroup, and let us consider
G/H the set of equivalence classes defined by the relation x 1 y 2 H. Then the law

x · y = xy

defines a group structure on G/H. The canonical injection G ! G/H is a group morphism. if in addition
G is an Abelian group, G/H is also an Abelian group.
Proof. Since H is distinguished, the composition on G is compatible with the equivalence relation: this just
means that if x 2 a, y 2 b then xy 2 ab, so x · y = xy is well defined because independant of the choices of
x and y. The rest is clear.
Theorem 2.24. Let G and F be two groups, f : G ! F a group morphism, let ' the canonical surjection
G ! G/Ker f , j the canonical injection from G/Ker f to F . Then there exists a unique group morphism f
from G/Ker f to Imf such that f = j f ' and it is an isomorphism.
A few words of explanation. ' : G ! G/Ker f is just defined by '(x) = x̄. The canonical injection
j : G/Ker f ! F is defined as follow: For any x, y 2 G, f (x) = f (y) if and only if x 1 y 2 textKerf so we
can define j(x̄) = f (x), this is independant of the choice of z 2 x̄.

11
Proof. We set
f¯ : G/Ker f ! textImf
x̄ 7! f (x)
It is well defined because if y 2 x̄, x 1 y 2 Ker f , so f x 1 y = eF , i..e f (x) = f (y): the value is independant
of the element we have chosen to compute. By definition, for any x 2 G, f (x) = j f '(x). We easily see
it is a group morphism.
It is injective: if x̄ 2 Kerf¯, we have e = f¯(x̄) = f (x), so x 2 Kerf , and hence x̄ = e. It is surjective: if
y 2 Im f, there exists x 2 G such that y = f (x), if x0 2 x̄, we also have y = f (x0 ), so y = f¯(x̄). Hence it is
an isomorphism.

2.7.3 A fundemental example Z/nZ

Theorem 2.25. If H is a subgroup of (Z, +), then there exists p 2 Z such that H = pZ
Proof. Let us define J = {x 2 H, x > 0}. If J = ;, it means that any x 2 H is  0. But if there exists a
x < 0, then x 2 H and x > 0 which is contradiction, do H = {0} = 0Z in that case.
Hence we can assume that J 6= ;, and we can consider a = min{x 2 J}. Clearly a > 0, and a 2 H by
definition. Let x 2 J. There exis p, q 2 N such that x = pa + r and 0  r < a. We see that r = x pa 2 H
and since r 0, either r = 0 of r 2 J. Since a is the smallest element of J, the only possibility is r = 0.
If x 2 H and x  0, x = 0 or x 2 J, so we see that any element of H is a multiple of a.
Conversely, for any a 2 Z, aZ is a subgroup of Z: 0 2 aZ and for any x, y 2 aZ, there exists p, q 2 Z such
that x = ap and y = aq, so x y = (p q)a 2 aZ.
Definition 2.26. For any n 2 N, the group Z/nZ is called the group of integer number modulo n. If n = 0
it is isomorphic to Z. If n 6= 0, it is a cyclic group, i.e. there exists m 2 N such that for any x 2 Z/nZ,
mx = 0: indeed nx = 0.

2.8 The symetric group

Another very important example of group is the symmetric group.
Definition 2.27. Let E be a set, we denote by SE the set of bijection from E onto E, they are called
permutations.
Theorem 2.28. (SE , ) is a group, called the symmetric group.
Proof. Since the composition of two bijections is a bijection, is an internal law. The neutral element is
IdE
If E = {0, 1, . . . n}, the symetric group SE is more often denoted as Sn . This is finite group which cardinal
is n! (prove it by induction on n)
It is clear that if f : E ! F is a bijection, then SE and SF are isomorphic, and an isomorphism is
': S E ! SF
1
s 7! f s f
Thanks to this result, SE is isomorphic to Sn when |E| = n, and we will only get interested in that case.
The symmetric group is important because of the following result:
Theorem 2.29 (Cayley). Any finite group is isomorphic to a subgroup of S|E| .
Proof. We construct a injective morphism from G to S|E| by:
a 2 G 7! sa
with sa defined by: 8x 2 G, sa (x) = ax. This is a bijection of G. If a is such sa = Id, then a = e: the
morphism is injective, and its image, with which G is isomorphic to, is a subgroup of S|E| .

12
 If s 2 Sn , we represent it by a table
✓ ◆
1 2 ... p ...n
.
s(1) s(2) ... s(p) . . . s(n)

For example S3 is composed of

✓ ◆ ✓ ◆ ✓ ◆ ✓ ◆ ✓ ◆ ✓ ◆
1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
e= , ⌧1 = , ⌧2 = , ⌧3 = , 1 = , 2 = .
1 2 3 2 1 3 1 3 2 3 2 1 2 3 1 3 1 2
2
⌧1,2,3 are called transposition (they only permute two indices), we see that ⌧1,2,3 = e. The 1,2 are called
2 3 2 3
cycle, and we see that 1 = 2 , while 1 = e, 3 = 2 , 3 = e.
More generaly, we have the following definitions:

Definition 2.30 (Orbits, cycle, transposition). Let s 2 Sn , and let us define the relation xRs y if and only
if there exists p 2 N such that y = sp (x). The equivalence classes are called orbits.
A cycle is a permutation such that there exists only one orbit that is not reduced to a point. This orbit is
called the support of the cycle, its cardinal is the lenght of the permutation. We call transposition a cycle of
lenght 2.

Not all permutations are cycle (but for this we need to work in Sn with n 4. An example is
✓ ◆
1 2 3 4
.
2 1 4 3

which has two orbits of lenght 2.

13