University of Cape Town

Department of Mathematics and Applied Mathematics

MAM3000W

3AL Notes 2024

Morgan Vandeyar

Last updated 14th March 2024 at 9:16am

Contents

1 Group Theory 2
1.1 Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Homomorphisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Cosets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Lagrange’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 Normal subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.6 Factor groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2 The Isomorphism Theorems 19

2.1 The First Isomorphism Theorem . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 The Second Isomorphism Theorem . . . . . . . . . . . . . . . . . . . . . . 23
2.3 The Third Isomorphism Theorem . . . . . . . . . . . . . . . . . . . . . . . 26
2.4 The Correspondence Theorem . . . . . . . . . . . . . . . . . . . . . . . . . 27

3 p-Groups 34
3.1 Group actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2 Conjugation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3 Cauchy’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4 p-Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

1
 Group Theory 1

Mon 12 Feb Let’s start the course by reviewing some basic group theory. Hopefully this will all be
very familiar to you from 2IA. The content of this chapter is just for review purposes,
you won’t be expected to memorise any of the definitions, results or proofs. Everything
in this chapter will form part of a toolbox for proving more complex results in the rest of
this course. So it is very important you’re comfortable with all of these ideas! 😊

1.1 Groups
Definition 1.1. A group is a pair (G, ·) consisting of a set G and a binary operation ·
on G satisfying the following properties:

(i) Identity: There exists an element e ∈ G such that a · e = e · a = a for all a ∈ G,

(ii) Inverses: For each a ∈ G, there exists some b ∈ G such that a · b = b · a = e,

(iii) Associativity: For each a, b, c ∈ G, we have a · (b · c) = (a · b) · c.

A group is called abelian if it satisfies the commutativity property, that is, a · b = b · a

for all a, b ∈ G. #

Before we cover basic examples of groups, let’s discuss some notation, basic properties
and related concepts.
We call G the underlying set of the group (G, ·). Often, we will just use G to denote
the group, when it is clear what the binary operation · is from context.

2
 1.1. Groups

Often times, we will write groups multiplicatively. This means we use the shorthand
ab for a · b. Moreover, we denote the identity element by 1G or 1 and the inverse of an
element a by a−1 . For a positive integer n, we write

a0 = 1,

| · a ·{z· · · · a},
an = a
n times
−n −1
a =a
| · a−1{z· · · · · a−1} .
n times

For example, a3 = a · a · a and a−2 = a−1 · a−1 .

The alternative to writing a group multiplicatively is writing a group additively: a + b
instead of a · b. The identity is written as 0G or 0 and inverse of a is written as −a. We
typically use the additive notation when talking about abelian groups.
Now for some basic properties of groups. Since this chapter is just a review, the proofs
are not included. However, it’s worth trying to prove these properties yourself so that
you get back into the algebra mindset.

Remark 1.2. Let G be a group and a, b ∈ G. Then the following properties hold:

1. The identity element of G is unique,

2. The inverse of a is unique,

3. (ab)−1 = b−1 a−1 ,

4. For all n, m ∈ Z, we have an am = an+m and (an )m = anm . ✠

Finally, before we get to examples of groups, here are some more algebra terms that
will allow us discuss groups more robustly.

Definition 1.3. Let G be a group. The centre of G, denoted Z(G), is a subset of G

consisting of all elements a ∈ G such that ab = ba for all b ∈ G.
The order of an element a ∈ G is the smallest positive integer n such that an = 1.
We denote it by |a|. If such an integer does not exist, we say a has infinite order. If the
set G is finite, the order of a group G is the cardinality of G (that is, the number of
elements in G), denoted |G|. Otherwise, we say that the group has infinite order.
A subset H ⊆ G is called a subgroup of G if H forms a group under the same
operation as G. We write H ≤ G. #

3
 1.1. Groups

Example 1.4.

1. The sets Z, Q, R and C (the integers, rational, real and complex numbers, respec-
tively) form abelian groups under addition. In each group, the identity element is
0 and the inverse of an element a is −a. Moreover, Z ≤ Q ≤ R ≤ C.

2. The sets Q× = Q \ {0}, R× = R \ {0} and C× = C \ {0} form abelian groups

under multiplication. In each group, the identity element is 1 and the inverse of a
is a−1 = 1/a. Furthermore, Q× ≤ R× ≤ C× .

3. A cyclic group is a group where all elements can be expressed as a power of a

certain element g. That is, each element can be written as g n for some n ∈ Z.
We say that g is the generator. We write hgi = {g k : k ∈ Z} for the cyclic group
generated by g. The order of a cyclic group equals the order of the generator. Every
cyclic group is abelian.

Wed 14 Feb 4. For each positive integer n, the integers modulo n, denoted Zn = {[0], [1], . . . , [n−1]},
forms a group under addition (where [m] is the equivalence class of m under modulo
n). Namely, [a]+[b] = [a+b]. The identity is [0] and the inverse of [a] is −[a] = [−a].
Such groups are both abelian and cyclic.

5. The Klein group is K4 = {1, a, b, c} where a2 = b2 = c2 = 1, ab = ba = c,

bc = cb = a and ac = ca = b. That is, each non-identity element has order 2
and the product of any two distinct non-identity elements is the third non-identity
element. The Klein group is abelian.

6. Let m, n be positive integers (m ≥ 2 preferably). If R = Zm , Z, Q, R or C, then the

general linear group of degree n over R is the set of n×n invertible matrices with
entries from R equipped with the operation of matrix multiplication. It is denoted
by GLn (R). In general, it is not abelian. The special linear group of degree n
over R, denoted SLn (R), is a subgroup of GLn (R) consisting of all matrices with
determinant equal to 1.

7. For a positive integer n, the symmetric group of degree n is the set of permutations
of n objects equipped with the operation of composition. The group is denoted by
Sn and it has order n!. In the case where n = 3, we use the notation

S3 = {ε, σ, σ 2 , τ, τ σ, τ σ 2 },

4
 1.1. Groups

where σ 3 = τ 2 = ε and στ σ = τ . ♦

In the previous example, we mentioned that the special linear group is a subgroup of
the general linear group. In order to prove such a statement, we can use the following:

Remark 1.5 (Subgroup Test). If G is a group and H ⊆ G, to prove that H is a

subgroup of G it suffices to show the following:

1. 1 ∈ H (where 1 is the identity element of G),

2. If a, b ∈ H, then ab ∈ H,

3. If a ∈ H, then a−1 ∈ H.

An alternate (often faster) method is to just show the following:

1. 1 ∈ H,

2. If a, b ∈ H, then ab−1 ∈ H (alternatively, a−1 b ∈ H). ✠

In the next example, we use the Subgroup Test to show that SLn (R) ≤ GLn (R).

Example 1.6. First, note that the identity element of GLn (R) is the n × n identity
matrix, In . We know det(In ) = 1, thus In ∈ SLn (R). Now let A, B ∈ SLn (R). We need
to show that AB −1 ∈ SLn (R). That is, we need to show det(AB −1 ) = 1. From properties
of the determinant, we have

det(AB −1 ) = det(A) det(B −1 ) = det(A) det(B)−1 .

Since A, B ∈ SLn (R) by assumption, we have det(A) = det(B) = 1. Thus,

det(AB −1 ) = det(A) det(B)−1 = 1 · 1−1 = 1.

So, AB −1 ∈ SLn (R) and we can now conclude that SLn (R) is a subgroup of GLn (R). ♦

Now we see a new example of a group which we will be using in this course.

Example 1.7. Consider a regular n-gon (n ≥ 3). The symmetries of a regular n-gon are
a collection of rotations and reflections that preserve the shape of our n-gon. For example,
consider a square (n = 4). There are 4 rotations and 4 reflections we can do that keep
the shape of the square the same:

5
 1.1. Groups

R0 R90 R180 R270

H V D D0

We can rotate the square (around the centre) by 0 degrees, 90 degrees, 180 degrees or 270
degrees. We can also flip the square along the horizontal axis, vertical axis and the two
diagonals. These rotations and reflections obviously keep the square in the same place
but permute the vertices.
We can also compose these symmetries (i.e. rotations and reflections). For example,
we could rotate by 90 degrees then flip along the horizontal. The set of symmetries under
the operation of composition forms a group called the dihedral group, Dn . A dihedral
group always consists of n rotational symmetries and n reflectional symmetries. Thus
|Dn | = 2n.
We can express the dihedral group Dn in a more abstract way which allow us to do
computations more easily. Namely,

Dn = {1, r, r2 , . . . , r n−1 , s, sr, sr 2 , . . . , sr n−1 },

where r and s satisfy |r| = n, |s| = 2 and rsr = s. We call r and s the generators of Dn .
In the case of the square, let r represent the rotation by 90 degrees (R90 in the
diagram) and s represent the horizontal reflection (H). Any other reflection or rotation
can be expressed as a composition of r and s. For example, rotating by 180 degrees (R180 )
is r2 . On the other hand, reflection about the off diagonal (D′ ) is the same as a rotation
of 90 degrees (R90 ) followed by a horizontal reflection (H). That is, the reflection D′ can
be represented by sr. ♦

Exercise 1.8. Show that Z(D4 ) = {1, r2 }.

6
 1.2. Homomorphisms

1.2 Homomorphisms
Homomorphisms are a very important concept in algebra. These are maps (functions) that
preserve group structure. Homomorphisms give us a way to find interesting connections
between seemingly unrelated groups.

Definition 1.9. Let G and G′ be groups. A map φ : G → G′ is called a (group)

homomorphism if φ(ab) = φ(a)φ(b) for all a, b ∈ G. An injective homomorphism is
called a monomorphism and a surjective homomorphism is called an epimorphism. A
homomorphism that is both injective and surjective is called an isomorphism. In this
case, we say G and G′ are isomorphic and write G ∼
= G′ . An isomorphism from G to itself
is called an automorphism. #

There are also several important subsets that are related to homomorphisms, as out-
lined in the following definition:

Thu 15 Feb Definition 1.10. Let φ : G → G′ be a homomorphism. The kernel of φ is the following
subset of G:

ker φ = {g ∈ G : φ(g) = 1G′ }.

The image of φ is the following subset of G′ :

φ(G) = im φ = {g ′ ∈ G′ : there exists g ∈ G such that φ(g) = g ′ }.

For any subset H ⊆ G, the image of H under φ is similarly defined:

φ(H) = {g ′ ∈ G′ : there exists h ∈ H such that φ(h) = g ′ }. #

Example 1.11. Consider the groups Z and Z3 (under addition and addition modulo
3, respectively). Let φ : Z → Z3 be the map defined by φ(n) = [n] for all n ∈ Z,
where [n] represents the equivalence class of n under modulo 3. We claim that this is a
homomorphism. To prove this, we take two elements n, m ∈ Z and show φ(n + m) =
φ(n) + φ(m). Using the definition of our map φ, we have φ(n) + φ(m) = [n] + [m]. From
properties of modular arithmetic, we have [n] + [m] = [n + m]. Thus,

φ(n) + φ(m) = [n] + [m] = [n + m] = φ(n + m).

Thus φ is a homomorphism. In fact, φ is an epimorphism. Note that Z3 = {[0], [1], [2]}

and φ(0) = [0], φ(1) = [1] and φ(2) = [2]. ♦

7
 1.2. Homomorphisms

Note: In the above example, n+m is addition of integers in the group Z and φ(n)+φ(m)
is addition of equivalence classes modulo 3 in Z3 .

Question: Is φ a monomorphism?

Now we look at some useful properties of homomorphisms.

Remark 1.12 (Homomorphism properties). Let φ : G → G′ be a homomorphism.

Then the following properties hold:

1. φ(1G ) = 1G′ (homomorphisms preserve the identity),

2. φ(an ) = φ(a)n for all a ∈ G and n ∈ Z (for the case where n = −1, homomorphisms
preserve inverses),

3. a, b ∈ G commute if and only if φ(a), φ(b) ∈ G′ commute (homomorphisms preserve

commutativity),

4. ker φ ≤ G and im φ ≤ G′ ,

5. φ is a monomorphism if and only if ker φ = {1G },

6. φ is an epimorphism if and only if im φ = G′ ,

7. φ is an isomorphism if and only if φ−1 is a homomorphism,

8. If φ is an isomorphism, then |a| = |φ(a)| for all a ∈ G (isomorphisms preserve

order). ✠

Note: Property 7 gives us a new way to prove a homomorphism φ is an isomorphism.

Namely, we just need to provide an inverse homomorphism φ−1 : G′ → G.

Exercise 1.13. Prove the above properties.

In the following example, we look at some examples of isomorphic groups.

Example 1.14.

1. All finite cyclic groups of order n are isomorphic Zn ,

2. All infinite cyclic groups are isomorphic to Z,

8
 1.3. Cosets

3. K4 is isomorphic to Z2 ×Z2 (the group whose underlying set is the Cartesian product
of Z2 with itself and who operation is component-wise addition modulo 2),

4. Up to isomorphism, there are two groups of order 4. Namely, Z4 and K4 (∼

= Z2 ×Z2 ).
If a group of order 4 is cyclic, it must be isomorphic to Z4 . Otherwise, it is isomorphic
to K4 . ♦

1.3 Cosets
Definition 1.15. Let H be a subgroup of a group G and a ∈ G. The left coset of H
containing a is the set aH = {ah : h ∈ H}. We call a the representative of aH and
use |aH| to denote the number of elements in the set aH. We similarly define the right
coset of H containing a by Ha = {ha : h ∈ H}. #

Note: Cosets are not necessarily subgroups of G. They are merely subsets. However, H
is always a subgroup. For additive groups, we write left cosets as a + H = {a + h : h ∈ H}.
We write right cosets in a similar manner.

Remark 1.16 (Coset properties). Let G be a group with subgroup H and elements
a, b. Then,

1. H = 1H = H1,

2. |aH| = |H|,

3. a ∈ aH,

4. aH = H if and only if a ∈ H,

5. aH = bH if and only if a ∈ bH (or, equivalently, a−1 b ∈ H),

6. aH = Ha if and only if H = aHa−1 = {aha−1 : h ∈ H},

7. If H ⊆ Z(G), then aH = Ha for all a ∈ G,

8. aH is a subgroup of G if and only if a ∈ H,

9. Exactly one of the following holds: Either aH = bH or aH ∩ bH = ∅.

9
 1.3. Cosets

These properties also hold for right cosets (though the final part of property 5 becomes
ab−1 ∈ H). ✠

Exercise 1.17. Prove the above properties.

Coset property 6 suggests that it the left and right cosets of H containing a are not
necessarily equal, that is, aH 6= Ha. The following example shows us that it is indeed
the case.

Mon 19 Feb Example 1.18. Consider S3 , the group symmetric group of order 3, and its subgroup
H = {ε, τ }. We have

σH = {σ, στ } = {σ, τ σ 2 }
Hσ = {σ, τ σ}.

These two sets are not equal, so we have σH 6= Hσ. Now let’s consider the subgroup
K = {ε, σ, σ 2 }. The left and right cosets of K with representative τ are

τ K = {τ, τ σ, τ σ 2 }
Kτ = {τ, στ, σ 2 τ } = {τ, τ σ 2 , τ σ}.

So, in this case, τ K = Kτ . Coset property 6 says that this can happen only when
K = τ Kτ −1 . Let’s confirm this:

τ Kτ −1 = τ Kτ = {τ 2 , τ στ, τ σ 2 τ } = {ε, σ 2 , σ} = K. ♦

Question: Why are H and K subgroups of S3 ? Why do the following identities hold:
στ = τ σ 2 , τ σ = σ 2 τ , τ στ = σ 2 and τ σ 2 τ = σ?

Exercise 1.19. Compare the left and right cosets of K with representative τ σ. Are the
left and right cosets equal? Do the same for τ σ 2 .

In the next example, we will look at cosets in an additive group.

Example 1.20. Consider the group Z under addition. For each n ∈ Z, we have the
subgroup hni generated by n. Namely, hni = nZ = {. . . , −2n, −n, 0, n, 2n, . . .}. Let’s

10
 1.3. Cosets

look at the subgroup H = 3Z. Since Z is an additive group, the cosets of 3Z are written
additively, that is, m + 3Z for m ∈ Z. The left cosets of H are

0 + 3Z = {. . . , −6, −3, 0, 3, 6, . . .} = 3 + 3Z = . . . ,
1 + 3Z = {. . . , −5, −2, 1, 4, 7, . . .} = 4 + 3Z = . . . ,
2 + 3Z = {. . . , −4, −1, 2, 5, 8, . . .} = 5 + 3Z = . . . .

Thus the left cosets of 3Z are exactly the equivalence classes [0], [1] and [2] in Z3 . Since
Z is abelian, coset property 7 tells us that the left and right cosets of 3Z are equal, that
is 0 + 3Z = 3Z + 0 = [0], etc.
In fact, the same is true for any nZ where n is some integer. Thus, the left and right
cosets of nZ are precisely the equivalence classes [0], [1], . . . , [n − 1] in Zn . ♦

In the previous example, we saw how the cosets of a subgroup H with different rep-
resentatives can actually be the same sets. For example, 0 + 3Z = 3 + 3Z. This means
the number of cosets of H does not necessarily equal the number of elements in G. In
particular, we found that H = 3Z had 3 left cosets even though Z has infinitely many
elements.

Definition 1.21. Let G be a group with subgroup H. The index of H in G is the

number of (distinct) left (equivalently, right) cosets of H in G. It is denoted |G : H|. #

Note: The number of left cosets of H always equals the number of right cosets of H.
This is true regardless of whether aH = Ha for all a ∈ G. Note that it is possible that
the number cosets (i.e. index) is infinite.

Thus, following Example 1.20, we can write |Z : 3Z| = 3. Furthermore, for any n ∈ Z,
we have |Z : nZ| = n.

Example 1.22. Consider the dihedral group D4 and its subgroup K = Z(D4 ) = {1, r2 }.
We can find the index of K in D4 by calculating all the left cosets then counting them.
The left cosets of K are

K = {1, r2 }
rK = {r, r3 } = r3 K
sK = {s, sr2 } = sr2 K
srK = {sr, sr3 } = sr3 K.

Thus, |D4 : K| = 4. ♦

11
 1.4. Lagrange’s Theorem

Exercise 1.23. Compute the right cosets of K in D4 . How do they compare to the left
cosets?

Remark 1.24. Recall that a partition of a set S is a collection of nonempty disjoint

subsets of S whose union is S. For example, the sets {0}, Z+ and Z− for a partition of
the integers.
Coset properties 3 and 9 mean that the family of left cosets of H forms a partition of
G. Similarly for the family of right cosets of H. Diagrammatically, this looks like:

aH G
a
bH
ah b

H d
1 h
cH dH
c

1.4 Lagrange’s Theorem

The following theorem shows how powerful the concept of cosets is. Namely, it allows us
to relate orders of subgroups to the order of the group.

Theorem 1.25 (Lagrange’s Theorem). Let H be a subgroup of a finite group G. Then

the order of H divides the order of G. Moreover, the number of cosets of H in G is
|G|
|G : H| = .
|H|

Note: While the index of a subgroup is defined for infinite groups G, the quotient |G|/|H|
does not make sense when G is infinite. Though notice that the theorem requires G to
be finite, so we run into no problems here.

Lagrange’s Theorem is quite helpful since if we know the order of a subgroup, we have
an idea what the order of the overall group could be. But what would the converse of this
theorem look like? Given the order of G, is there necessarily a subgroup for every divisor
of that order? For example, does a group of order 12 have to have a subgroup of order 6?

12
 1.5. Normal subgroups

In general, no. The most simple example is the alternating group A4 which has order 12
but no subgroup of order 6.
One of the goals for this course is to find a partial converse of Lagrange’s Theorem.
Basically, we are going to look for some conditions we can add to the theorem in order to
make its converse true.
For now, let’s look at some important corollaries of Lagrange’s Theorem.

Corollary 1.26. If G is a finite group, the order of each element in G divides the order
of G.

In a way, the above corollary is even more powerful than Lagrange’s Theorem since
now we only need to know the order of a single element to get an idea of the order of G.
The next corollary shows us how knowing the order of a group can tell us about its
structure.

Corollary 1.27. Every group of prime order is cyclic.

Question: If G is a group with prime order p, what group is G isomorphic to?

1.5 Normal subgroups

In Example 1.18, we saw that it is possible (but not guaranteed) for the left and right
cosets of a subgroup to be equal (by which we mean aH = Ha for all a ∈ G). The
subgroups H whose left and right cosets are equal play quite an important role in group
theory. These subgroups allow us to get a better understanding of the structure of the
whole group. As such, these subgroups have a special name.

Definition 1.28. Let G be a group. A subgroup H of G is called normal in G if

gH = Hg for all g ∈ G. We write H ◁ G. #

While looking at cosets, we already came across some normal subgroups:

Example 1.29.

1. Following Example 1.20, we have that 3Z ◁ Z. In fact, nZ ◁ Z for any n ∈ Z.

2. Following Example 1.18 and the subsequent exercise, the subgroup K = {ε, σ, σ 2 }
of S3 is normal. On the other hand, H = {ε, τ } is not normal in S3 .

13
 1.5. Normal subgroups

3. Following Example 1.22 and the subsequent exercise, Z(D4 ) ◁ D4 . ♦

For the above examples, we explicitly calculated all of the left and right cosets of a
given subgroup. However, the following theorem gives us a way to simplify the process.

Theorem 1.30 (Normality Test). Let H be a subgroup of a group G. Then the fol-
lowing are equivalent:

(i) H ◁ G,

(ii) gHg −1 ⊆ H for all g ∈ G,

(iii) gHg −1 = H for all g ∈ G.

This theorem tells us that in order to prove a subgroup H is normal, all we need to
do is show that one of the equivalent conditions is true.

Mon 22 Feb Example 1.31. Let R = Q, R or C. We know that SLn (R) is a subgroup of GLn (R). We
claim that SLn (R) is in fact normal in GLn (R). To prove this, we will use the Normality
Test. We will show that for all A ∈ GLn (R), we have A SLn (R)A−1 ⊆ SLn (R). To show
this subset inclusion, let B ∈ SLn (R). We will show ABA−1 ∈ SLn (R). The condition for
this to hold is that det(ABA−1 ) = 1. By properties of the determinant, we have

det(ABA−1 ) = det(A) det(B) det(A)−1 = det(A)1 det(A)−1 = det(A) det(A)−1 = 1.

Thus ABA−1 ∈ SLn (R) for all A ∈ GLn (R), B ∈ SLn (R), and so SLn (R) ◁ GLn (R). ♦

Remark 1.32 (Normal subgroup properties). Let G be a group with subgroups H

and K. Then the following hold:

1. If G is abelian, then H ◁ G (every subgroup of an abelian group is normal),

2. If H ⊆ Z(G), then H ◁ G,

3. If H ◁ G and H ⊆ K, then H ◁ K. ✠

Exercise 1.33. Prove the above properties.

The following theorem also gives us a way to prove a subgroup is normal (in the very
special case where the subgroup has index 2).

Theorem 1.34. Let H be a subgroup of a group G. If |G : H| = 2, then H ◁ G.

14
 1.6. Factor groups

Exercise 1.35. Prove the above theorem.

Property 3 says that if H is a normal subgroup of G then H is normal in any other

subgroup K it is contained it. However, is it true if H ◁ K and K ◁ G, then H ◁ G? In
general, this is not the case. We can use the above theorem to provide a counterexample.

Example 1.36. Consider the dihedral group D4 and the subgroups H = hsi = {1, s} and
K = hs, r2 i = {1, s, r2 , sr2 }. Clearly H ⊆ K. By Lagrange’s Theorem, |K : H| = 4/2 = 2,
so H ◁ K by Theorem 1.34. Moreover, |D4 : K| = 8/4 = 2, so K ◁ D4 . However, H is not
normal in D4 (consider rsr−1 = sr2 ∈
/ hsi). ♦

Question: Why is H not normal in D4 ? Provide a full explanation.

This example shows us that we need to be careful when considering nested subgroups
and normality. It’s very important that when we talk about a ‘normal subgroup’, we are
clear which group it is normal in. We’ll be returning to this idea in later chapters!

1.6 Factor groups

Definition 1.37. Let G be a group with normal subgroup H. The set of left cosets of
H is denoted by

G/H = {gH : g ∈ G}.

The set G/H equipped with the operation

(aH) · (bH) = (ab)H,

for a, b ∈ G, the set G/H is a group called the factor group (or quotient group) of G
by H. The identity of G/H is H and the inverse of gH ∈ G/H is g −1 H. #

Question: Does it matter whether we define the quotient group using left cosets instead
of right cosets?

Note: Factor groups are only defined for normal subgroups. So, whenever we see a group
of the form G/H, we immediately know that H must be normal in G.

Remark 1.38 (Factor group properties). Let G be a group with H ◁ G. Then the
following properties hold:

15
 1.6. Factor groups

1. The order of G/H is the index of H in G. If G is finite, by Lagrange’s Theorem,

|G|
|G/H| = |G : H| = ,
|H|

2. If G is abelian, then G/H is abelian,

3. If G is cyclic with G = hgi, then G/H is cyclic with G/H = hgHi, ✠

Exercise 1.39. Prove the above properties.

Remark 1.40. For each normal subgroup H of G, there is a canonical (or natural)
projection map π between G and the factor group of H. It is defined by

π : G → G/H
g 7→ gH.

The canonical projection is an epimorphism. ✠

Question: Why is the canonical projection an epimorphism?

Theorem 1.41. A subgroup H of a group G is normal in G if and only if it is the kernel

of a homomorphism.

Exercise 1.42. Prove the above theorem.

Example 1.43. Recall from Example 1.20 that, for each n ∈ Z+ , the cosets of nZ in
Z correspond to the equivalence classes [0], [1], . . . , [n − 1] from Zn . We also saw that
nZ ◁ Z. Thus, we can consider the factor group Z/nZ. In fact, through the isomorphism
φ : a + nZ 7→ [a], we obtain Z/nZ ∼
= Zn . ♦

Question: Why is φ an isomorphism?

Recall that the group Zn was obtained by taking the integers modulo n. Factor groups
generalise the “modulo” concept to arbitrary groups.

Example 1.44. We saw previously that K = Z(D4 ) is normal in D4 . Furthermore, we

calculated the left cosets of K in Example 1.22. The set of left cosets of K is then
{K, rK, sK, srK}. (Note that, for example, r3 K = rK, so only one of these appears in
the set). Thus, we have the factor group D4 /K = {K, rK, sK, srK}.

16
 1.6. Factor groups

Since D4 /K is a group with order 4, it is isomorphic to either Z4 or K4 . We know

that Z4 is cyclic but K4 is not. Thus, to determine which group D4 /K is isomorphic to,
we just need to determine if it is cyclic or not. If D4 /K was cyclic, it would contain some
element that generates the group. That is, an element of order 4. However, all elements
have order at most 2 since (rK)2 = (sK)2 = (srK)2 = K. Hence, D4 /K is not cyclic and
so we have D4 /K ∼= K4 . ♦

Finally, we see a theorem that shows us how the structure of a factor group can tell
us about the structure of the original group. It might seem a bit strange that we would
know more about the factor group than the original group, but we will see this in action
later in the course!

Theorem 1.45. Let G be a group. If G/Z(G) is cyclic then G is abelian.

Exercise 1.46. Prove the above theorem.

Hint: Let the generator of G/Z(G) be denoted by gH (where H = Z(G)). Show that
every element of G can be written as g n z for n ∈ Z and z ∈ Z(G).

To finish off the chapter, here are some mathematical memes:

Michael Kinyon
@ProfKinyon
I

Ah, group theory, where we can write H/H = {H} and have it
make perfect sense

7:18 pm · 02 Feb 22

retweet 4 🗩9 🧡 117

After going through this chapter, is it clear to you why H/H = {H}? 😊

17
 1.6. Factor groups

physicist
@FinitePhysicist
I

i am the anti-galois: survived into my 20s but no notable

mathematical achievements

6:20 pm · 06 Feb 22

retweet 45 🗩7 🧡 364

Évariste Galois was a French mathematician who lived in the 19th century. He made
many contributions to algebra and helped develop the concept of normal groups. He
notably died at age 20 in a duel. If you continue with algebra in honours, you’ll learn
more about his work!

18
 The Isomorphism Theorems 2

Wed 28 Feb The Isomorphism Theorems, also called Noether’s Theorems after mathematician Emmy
Noether who discovered them, are very important and useful tools in algebra. They show
us how homomorphic groups relate to each other and the connection between groups and
their normal subgroups. We will use these theorems a lot during the course and they will
help us build a partial converse to Lagrange’s Theorem.

2.1 The First Isomorphism Theorem

The First Isomorphism Theorem will give us a way to connect homomorphisms to factor
groups. This is powerful since it allows us to study factor groups without dealing with
cosets, or, conversely study homomorphisms by looking at factor groups.

Theorem 2.1 (First Isomorphism Theorem). Let φ : G → G′ be a group homomor-

phism. Then ker φ ◁ G and

G/ ker φ ∼
= im φ.

Proof. We start by showing ker φ ◁ G. Take k ∈ ker φ, that is, φ(k) = 1G′ . Then, for all
g ∈ G,

φ(gkg −1 ) = φ(g)φ(k)φ(g −1 ) = φ(g)φ(g −1 ) = φ(gg −1 ) = φ(1G ) = 1G′ .

Thus, gkg −1 ∈ ker φ for all g ∈ G, k ∈ ker φ. That is, g(ker φ)g −1 ⊆ ker φ for all g ∈ G.
By the Normality Test, this means that ker φ ◁ G.

19
 2.1. The First Isomorphism Theorem

Now let us construct an isomorphism Φ : G/ ker φ → im φ. Notice that G/ ker φ is

well-defined since ker φ is normal in G. Moreover, the elements of G/ ker φ have the form
g ker φ, as they are the left cosets of ker φ in G. As we are mapping into the image of
φ, it would make sense to apply φ to the representative element g in our mapping. We
define

Φ(g ker φ) = φ(g).

To check that Φ is well-defined, we need to make sure that it gives the same output
regardless of the coset representative. Suppose a ker φ = b ker φ. By the coset properties,
ab−1 ∈ ker φ. That is,

1G′ = φ(ab−1 ) = φ(a)φ(b)−1 .

Multiplying both sides by φ(b) on the right, we obtain φ(b) = φ(a). Thus Φ(a ker φ) =
Φ(b ker φ) and so Φ is well-defined.
The mapping preserves operations since, for a, b ∈ G, we have

Φ((a ker φ)(b ker φ)) = Φ((ab) ker φ) = φ(ab) = φ(a)φ(b) = Φ(a ker φ)Φ(b ker φ).

Moreover, Φ is a surjection since for every g ′ ∈ im φ there is some g ∈ G such that φ(g) =
g ′ (by definition of the image) and so Φ(g ker φ) = g ′ . Now to prove injectivity, we take
a ker φ, b ker φ ∈ G/ ker φ and prove that Φ(a ker φ) = Φ(b ker φ) implies a ker φ = b ker φ.
Note that

Φ(a ker φ) = Φ(b ker φ)

=⇒ φ(a) = φ(b)
=⇒ 1G′ = φ(a)−1 φ(b)
=⇒ 1G′ = φ(a−1 b)
=⇒ a−1 b ∈ ker φ
=⇒ a ker φ = b ker φ.

Thus, Φ is an isomorphism and so we have G/ ker φ ∼

= im φ.

The First Isomorphism Theorem tells us that each homomorphism from a group G
leads to a factor group with the same structure as the image of the homomorphism. For
example, if we have a group G missing a desirable property but we are able to construct
a homomorphism to a group with that desirable property, we can “give” our group that

20
 2.1. The First Isomorphism Theorem

desirable property by taking the quotient of that group with respect to the kernel. Another
use is that it allows us to look at factor groups without delving into cosets.
The following corollary provides a special case of the First Isomorphism Theorem
where φ is not only a homomorphism but an epimorphism (that is, a surjective homomor-
phism).

Corollary 2.2. Let φ : G → G′ be an epimorphism. Then im φ = G′ and so

G/ ker φ ∼
= G′ .

Note: Often this corollary is referred to as the First Isomorphism Theorem as well.

Question: What if the homomorphism in the First Isomorphism Theorem is both sur-
jective and injective? What does the theorem tell us then?

Note: If a question asks to prove G/H ∼

= G′ for groups G and G′ with H ◁ G, often we
will use the First Isomorphism Theorem. To do so, construct an epimorphisms φ : G → G′
with ker φ = H.

This is what the First Isomorphism Theorem looks like diagrammatically:

G G0

1
ker ϕ ϕ(1)

x
x ker ϕ ϕ(x)

a
ϕ(a) = ϕ(b)
a ker ϕ b

..
.

The theorem tells us that a homomorphism φ organises elements of G together according

to which coset of ker φ the elements belong. But this is the same as collecting elements

21
 2.1. The First Isomorphism Theorem

which map to the same element of G. From this, we get a correspondence between the
cosets of a normal subgroup of G and the elements of G′ (or, in the case of φ not being
an epimorphism, the elements of im φ = φ(G)).

Example 2.3. Let us show that GLn (R)/ SLn (R) ∼

= R× .To do so, we will construct an
epimorphism φ : GLn (R) → R× with ker φ = SLn (R).
A very natural way to go from a group of matrices to the real numbers is by taking
the determinant. Let us try the epimorphism φ : A 7→ det(A) for A ∈ GLn (R). Now, the
kernel of φ consists of all n × n matrices with determinant 1. However, this is precisely
the group SLn (R). Thus, by the First Isomorphism Theorem, GLn (R)/ SLn (R) ∼ = R× . ♦

Question: Why is φ : A 7→ det(A) an epimorphism? (I believe this was covered last

year, so check your IA notes 😊)

Thu 29 Feb Example 2.4. Let G be a group. Take x ∈ G, then the map σx : G → G defined
by σx (g) = xgx−1 is called an inner automorphism of G. The group of all inner
automorphisms of G under composition is denoted by Inn(G) = {σx : x ∈ G}.
Let’s construct a homomorphism between G and Inn(G). The most natural one would
be the map that takes an element x to its corresponding inner automorphism, σx , that is,
φ : x 7→ σx . However, we need to check that it is a homomorphism. That is, φ(x)φ(y) =
φ(xy) for all x, y ∈ G. More specifically, σx ◦ σy = σxy . We need to show the equality of
these two maps σx ◦ σy and σxy . To do so, we show that σx (σy (g)) = σxy (g) for all g ∈ G.
Notice that

σx (σy (g)) = x(ygy −1 )x−1 = (xy)g(xy)−1 = σxy (g),

for all x, y, g ∈ G. That is, we have the equality of homomorphisms.

The kernel consists of φ elements of G whose associated inner automorphism is the
identity mapping (which is the identity in Inn(G)). That is, the elements x such that σx
satisfies σx (g) = g for all g ∈ G. That is to say, xgx−1 = g. However, this is exactly
xg = gx, meaning the kernel of φ consists of all elements of all x that commute with
every element g ∈ G. Thus, ker φ = Z(G). Moreover, im φ = Inn(G) since all inner
automorphisms can be represented as σx for some x ∈ G and so φ(x) = σx .
We can conclude that G/Z(G) ∼= Inn(G) for all groups G. ♦

Question: Why is Inn(G) a group?

Exercise 2.5. Prove that S3 ∼

= Inn(S3 ).

22
 2.2. The Second Isomorphism Theorem

The above exercise shows how using normal subgroups (that is, groups whose left and
right cosets are equal) can lead to us better understanding the group itself.

Exercise 2.6. Prove that G/{1} ∼

= G.

2.2 The Second Isomorphism Theorem

Mon 4 Mar Definition 2.7. Let G be a group with subgroups H and K. The product of H and K
is the set

HK = {hk : h ∈ H, k ∈ K}. #

Note that the product HK is a set, not necessarily a subgroup itself. For example,
consider G = S3 with H = hτ i and K = hτ σ 2 i. Then

HK = {ε, τ, τ σ 2 , σ 2 }.

We see that |HK| = 4 and |S3 | = 6. Since 4 does not divide 6, it is impossible for HK
to be a subgroup of S3 , via Lagrange’s Theorem. What a nice application of Lagrange’s
Theorem!
Luckily, the following lemma tells us the conditions for HK to be a subgroup of G.

Lemma 2.8. Let H and K be subgroups of G. Then HK is a subgroup of G if and only

if KH ⊆ HK.

Proof. First assume that HK is a subgroup of G. We take kh ∈ KH and prove kh ∈ HK.

The inverse of kh exists since it belongs to the group G. So, (kh)−1 = h−1 k −1 ∈ HK
(recall that H and K are groups and so h−1 ∈ H and k −1 ∈ K). Since HK is a group,
the inverse (h−1 k −1 )−1 must be in HK. That is, kh ∈ HK and so KH ⊆ HK.
Now assume that KH ⊆ HK. First, note that 1 ∈ HK since 1 ∈ H and 1 ∈ K. Now
take a = h1 k1 , b = h2 k2 ∈ HK. We will prove ab−1 ∈ HK which shows that HK is closed
under products and inverses. We have

ab−1 = h1 k1 (h2 k2 )−1 = h1 k1 k2−1 h−1

2 .

Note that the inverse of b indeed exists since b is in the group G. Let k3 = k1 k2−1 ∈ K and
h3 = h−1
2 ∈ H. Then we have ab
−1
= h1 k3 h3 . Since KH ⊆ HK, we can write k3 h3 = h4 k4
and so ab−1 = h1 h4 k4 ∈ HK. Since HK is closed under products and inverses, it must
be a subgroup.

23
 2.2. The Second Isomorphism Theorem

Theorem 2.9 (Second Isomorphism Theorem). Let H and K be subgroups of G

with K ◁ G. Then the following hold:

(i) H ∩ K ◁ H,

(ii) HK is a subgroup of G,

(iii) K ◁ HK.

Moreover, we have the following isomorphism of factor groups:

H/(H ∩ K) ∼
= (HK)/K.

Proof.

(i) We must first prove that H ∩ K is a subgroup of H. It is clear that H ∩ K is a

group as it is the intersection of two groups. Moreover, by the definition of intersection,
H ∩ K must be a subset of H and so H ∩ K is a subgroup of H.

By the Normality Test, H ∩ K ◁ H is equivalent to saying h(H ∩ K)h−1 ⊆ H ∩ K for all

h ∈ H. Take arbitrary h ∈ H and a ∈ H ∩ K. We want to prove that x = hah−1 ∈ H ∩ K.
Since a, h ∈ H, it is clear that x ∈ H so it remains to prove that x ∈ K, and then x will
belong to the intersection H ∩ K. Since K is normal in G, we have hkh−1 ∈ K for all
k ∈ K. Taking k = a, we see that x = hah−1 ∈ K. Thus, x ∈ H ∩ K. Thus H ∩ K ◁ H.

(ii) Using Lemma 2.8, it suffices to show that KH ⊆ HK. Take kh ∈ KH. Observe
that kh = (hh−1 )kh = h(h−1 kh). Note that since K is normal in G, we have that
gKg −1 ⊆ K for all g ∈ G, thus gkg −1 = k ′ for some k ′ ∈ K. Therefore, we can write
h−1 kh = k ′ ∈ K and so kh = hk ′ ∈ HK. Thus KH ⊆ KH and HK is a subgroup of G.

(iii) First, it is clear that K is a subgroup of HK since for all k ∈ K, we have

k = 1k ∈ HK. Since K is normal in G, we have that gKg −1 ⊆ K for all g ∈ HK ⊆ G.
Thus, by the Normality Test, K ◁ HK.

Finally, we show that H/(H ∩ K) ∼

= (HK)/K. We start by constructing an epimor-
phism φ : H → HK/K with ker φ = H ∩ K, then we can use the First Isomorphism
Theorem to show the desired isomorphism of factor groups.
We define φ(h) = hK. Note that h ∈ HK, so the coset hK is indeed in the factor
group HK/K, and so our map φ is well-defined. Taking h, h′ ∈ H, we have

φ(hh′ ) = (hh′ )K = (hK)(h′ K) = φ(h)φ(h′ ),

24
 2.2. The Second Isomorphism Theorem

using the factor group operation. Thus, φ preserves the operation and so φ is a homo-
morphism. Notice that φ is surjective since for all cosets (hk)K in HK/K we can write
(hk)K = (hK)(kK) = (hK)(K) = hK = φ(h).
Now we look at the kernel of φ. The identity element in the factor group HK/K is
the coset K. So ker φ = {h ∈ H : hK = K}. By the properties of cosets, hK = K
is equivalent to saying h ∈ K. Thus, ker φ = H ∩ K. Thus, by the First Isomorphism
Theorem, H/(H ∩ K) ∼ = (HK)/K.

The Second Isomorphism is sometimes referred to as the Diamond Isomorphism The-

orem. The reason why comes from the lattice of the groups used in the theorem:

HK

K H

H ∩K

{1}

The Second Isomorphism Theorem says that each circled pair forms a factor group iso-
morphic to the other circled pair’s factor group.

Example 2.10. A group G is called metabelian if there exists an abelian normal sub-
group K such that the factor group G/K is abelian. All abelian groups are metabelian.
An example of a non-abelian metabelian group is S3 .
We can use the Second Isomorphism Theorem to prove that every subgroup of a
metabelian group is metabelian. Let H be a subgroup of G and K be the abelian normal
subgroup such that G/K is abelian. Applying the Second Isomorphism Theorem, we
obtain

H/(H ∩ K) ∼
= HK/K.

Note that the factor group HK/K is a subgroup of G/K since HK is a subgroup of G. By
assumption, G/K is abelian, thus the subgroup HK/K is also abelian. Thus, H/(H ∩ K)

25
 2.3. The Third Isomorphism Theorem

is abelian. As mentioned in the Second Isomorphism Theorem, H ∩ K is normal in H,

moreover, since H ∩ K is a subgroup of K, it must be abelian. Thus we have found a
normal subgroup of H such that its factor group is abelian. That is, H is metabelian. ♦

Question: To prove that all abelian groups are metabelian what can we take the normal
subgroup K to be? Why is S3 metabelian?

2.3 The Third Isomorphism Theorem

Thu 7 Mar Everyone’s favourite things about fractions or quotients is cancellation, right? The next
isomorphism theorem shows us that we have a similar cancellation when considering
quotients of groups (i.e. factor groups).

Theorem 2.11 (Third Isomorphism Theorem). Let H and K be normal subgroups

of G with H ⊆ K. Then K/H ◁ G/H and

(G/H)/(K/H) ∼
= G/K.

Proof. We will use the First Isomorphism Theorem to show this isomorphism of groups.
Let us construct an epimorphism φ : G/H → G/K with ker φ = K/H. A good place to
start is by looking at the form of the elements on either side of our mapping. The elements
of G/H have the form gH and the elements of G/K have the form gK. This suggests we
should try the map φ : gH 7→ gK. Note that φ is well-defined since if gH = g ′ H then
g −1 g ′ ∈ H ⊆ K and so φ(gH) = gK = g ′ K = φ(g ′ H). Taking gH, g ′ H ∈ G/H, we have

φ((gH)(g ′ H)) = φ(gg ′ H) = gg ′ K = (gK)(g ′ K) = φ(gH)φ(g ′ H).

Moreover, for all gK ∈ G/K, we have φ(gH) = gK. Thus φ is a well-defined epimorphism
and so we can apply the First Isomorphism Theorem. But let’s first check the kernel.
Remember that the identity of a factor group G/K is K itself. Thus, gH ∈ ker φ if and
only if φ(gH) = gK = K. That is, g ∈ K. If g ∈ K then gH belongs to the factor group
K/H. Thus, ker φ = K/H. Finally, applying the First Isomorphism Theorem, we obtain
(G/H)/(K/H) ∼= G/K as desired.
Note that it follows from the fact that ker φ = K/H that K/H ◁ G/H.

Note: (G/H)/(K/H) is a group of cosets of K/H with representatives from G/H, i.e.
representatives are cosets of H in G. That is,

(G/H)/(K/H) = {gH(K/H) : gH ∈ G/H}.

26
 2.4. The Correspondence Theorem

Thankfully, applying the First Isomorphism Theorem gave us a way to avoid dealing with
this jumble of cosets.

A nice thing about the Third Isomorphism Theorem is it goes along with our intuition
of fraction cancellation (note the “cancellation” of the Hs on the left-hand side).

Example 2.12. Consider G = Z with normal subgroups H = 6Z ⊆ K = 2Z. The Third

Isomorphism Theorem says

(Z/6Z)/(2Z/6Z) ∼
= Z/2Z.

The cosets of 6Z in 2Z are

0 + 6Z = {. . . , −12, −6, 0, 6, 12, . . .} = 6 + 6Z = . . . ,

2 + 6Z = {. . . , −10, −4, 2, 8, 14, . . .} = 8 + 6Z = . . . ,
4 + 6Z = {. . . , −8, −2, 4, 10, 16, . . .} = 10 + 6Z = . . . .

Thus 2Z/6Z is a group consisting of 3 elements. Recall from Example 1.43 that Z/6Z ∼
= Z6
under the isomorphism φ : a + 6Z 7→ [a]. Under this same isomorphism, 2Z/6Z =
{6Z, 2 + 6Z, 4 + 6Z} is isomorphic to the cyclic subgroup h[2]i = {[0], [2], [4]} of Z6 . Then,
Z6 /h[2]i ∼
= Z2 . ♦

Note: It is true that 2Z/6Z ∼= Z3 since it is a cyclic group containing 3 elements. How-
ever, we cannot write Z6 /Z3 ∼
= Z2 . The reason for this is Z3 is not a subgroup of Z6 and
we cannot factor a group by something that is not a normal subgroup. The only reason
we can go from (Z/6Z)/(2Z/6Z) to Z6 /h[2]i is that Z/6Z ∼
= Z6 and 2Z/6Z ∼
= h[2]i under
the same isomorphism, meaning that the subgroup inclusion remains.

2.4 The Correspondence Theorem

The Correspondence Theorem, sometimes known as the Lattice Theorem or Fourth Iso-
morphism Theorem, shows that every subgroup containing some normal group K corre-
sponds to a factor group that is a subgroup of G/K. Basically, it allows us to uncover
the subgroup structure of G by looking at the subgroup structure of a factor group.

Lemma 2.13. Let K and H be subgroups of G with K ⊆ H and K ◁ G. Then,

(i) H/K is a subgroup of G/K,

27
 2.4. The Correspondence Theorem

(ii) If H1 is a subgroup of G with K ⊆ H1 , then H ⊆ H1 if and only if H/K ⊆ H1 /K.

Proof.

(i) Under the natural projection π : G → G/K, the image of H is H/K. Since π is a
homomorphism, H/K must be a subgroup of G/K.

(ii) Suppose H ⊆ H1 . Then π(H) ⊆ π(H1 ), that is, H/K ⊆ H1 /K. Now suppose
that H/K ⊆ H1 /K. Take h ∈ H, then hK ∈ H/K ⊆ H1 /K, so hK = h1 K for some
h1 ∈ H1 . Since K ⊆ H1 , we have h ∈ h1 K ⊆ h1 H1 = H1 . Thus H ⊆ H1 .

Mon 11 Mar Theorem 2.14 (Correspondence Theorem). Let G be a group and K ◁ G. Define a
map

Θ : {H : H is a subgroup of G with K ⊆ H} → {H : H is a subgroup of G/K}

by Θ(H) = H/K. Then,

(i) If H is a subgroup of G/K then H = H/K where H = {g ∈ G : gK ∈ H},

(ii) Θ is a bijection.

For subgroups H, H1 , H2 of G containing K, we have

(iii) Θ preserves containment: H ⊆ H1 if and only if H/K ⊆ H1 /K,

(iv) Θ preserves normality: H ◁ H1 if and only if H/K ◁ H1 /K,

(v) Θ preserves intersections: (H1 /K) ∩ (H2 /K) = (H1 ∩ H2 )/K,

(vi) Θ preserves products: if H1 H2 is a subgroup of G then (H1 /K)(H2 /K) = (H1 H2 )/K.

Proof.

(i) Let H be a subgroup of G/K and H = {g ∈ G : gK ∈ H}. We first need to prove

that H/K is well-defined, that is, H is a subgroup of G and K ◁ H. To show that H is
a subgroup of G, take h, h′ ∈ H. Then hK, h′ K ∈ H. Since H is a subgroup of G/K, we
must have (hK)−1 (h′ K) = h−1 h′ K ∈ H. Thus h−1 h′ ∈ H, and so H is a subgroup of G.

Notice that for all k ∈ K we have k ∈ H since the identity kK = K of G/K must be
contained in the subgroup H, and then by the definition of H, k ∈ H. Therefore, K ⊆ H.
It follows that K ◁ H since H ⊆ G and K ◁ G.

28
 2.4. The Correspondence Theorem

Now we show that H = H/K. We start by showing H ⊆ H/K. Let gK ∈ H. Then

g ∈ H, by definition of H. Thus, gK ∈ H/K and so H ⊆ H/K. Now we show that
H/K ⊆ H. Take hK ∈ H/K. We know h ∈ H. That means hK ∈ H, by the definition
of H. Thus H/K ⊆ H and so H = H/K.

(ii) By (i), Θ is surjective, since for all subgroups H of G/K, we can define a subgroup
H of G with K ⊆ H such that Θ(H) = H/K = H. By Lemma 2.13, Θ is injective. If it
is not clear why, take Θ(H) = Θ(H1 ) and plug H and H1 into (ii) of the lemma, then by
symmetry we have injectivity. Thus Θ is a bijection.

(iii) This follows from Lemma 2.13.

(iv) Suppose H ◁ H1 . By (iii), we already have H/K ⊆ H1 /K, so it remains to prove

normality. Take x = (h1 K)(hK)(h1 K)−1 ∈ (h1 K)(H/K)(h1 K)−1 . Then x = h1 hh−1
1 K

Since H is normal in H1 , there is some h′ ∈ H such that h1 hh−1 ′ ′

1 = h . Thus, x = h K ∈

H/K. Therefore, (h1 K)(H/K)(h1 K)−1 ⊆ H/K and so H/K ◁ H1 /K.

Now suppose that H/K ◁ H1 /K. Take h ∈ H and h1 ∈ H1 . Then by the normality
of the factor groups, (h1 K)(hK)(h1 K)−1 ∈ H/K. That is h1 hh−1 ′
1 K = h K for some

h′ ∈ H. That is, h1 hh−1 ′ ′ ′

1 ∈ h K. Recall that K ⊆ H, so we have h K ⊆ h H = H. Thus,

h1 Hh−1
1 ⊆ H and by the Normality Test, H ◁ H1 .

(v) Left as an exercise.

(vi) Left as an exercise.

As previously mentioned, the Correspondence Theorem allows us to see the subgroup

structure of a factor group G/K. One tool we’ve used before when examining subgroups
is the lattice diagram. The Correspondence Theorem gives us a way to go from the lattice
diagram of subgroups containing K to the lattice diagram of subgroups of G/K.

Wed 13 Mar Example 2.15. Let G = hai be a cyclic group of order 12. Consider the subgroups
K1 = ha6 i and K2 = ha4 i.
The Fundamental Theorem of Finite Cyclic Groups says that the subgroups of a cyclic
group of order n are in one-to-one correspondence with the divisors of n. Moreover, each
subgroup is the cyclic group of order d, for some unique divisor of n. Thus the subgroups
of G are the cyclic subgroups generated by elements of order 1, 2, 3, 4, 6 or 12. That is,
ha12 i = {1}, ha6 i, ha4 i, ha2 i and hai = G, respectively. From this, we can construct the
subgroup lattice diagram for G. Note that had i ⊆ hak i if and only if k | d.

29
 2.4. The Correspondence Theorem

G = hai

ha2 i ha3 i

K1 = ha4 i K2 = ha6 i

{1}

Now, if we are interested in the subgroup structure of G/K1 , the Correspondence

Theorem says that we need to look at the subgroup structure for subgroups containing
K1 . This means restriction our attention to the section of the sublattice diagram that
involves K1 . That is, the section of the diagram that falls between K1 and G. So we
consider H/K1 for H = K1 , ha2 i, or G. Since Θ from the Correspondence Theorem
preserves inclusion, we are left with the following subgroup structure for G/K1 :

G/K1

ha2 i/K1

K1 /K1 = {K1 }

Similarly, we can inspect the subgroup structure of G/K2 . This means restriction our
attention to the upper square in the subgroup lattice diagram. By taking the quotient of
each subgroup by K2 , we obtain the subgroup lattice diagram for G/K2 :

G/K2

ha2 i/K2 ha3 i/K2

K2 /K2 = {K2 }
♦

30
 2.4. The Correspondence Theorem

In fact, since the map Θ in the Correspondence Theorem is a bijection, it allows us

to go in the other direction too, from subgroups of G/K to subgroups of G containing K.
We see this in the following example:

Example 2.16. Recall that K = Z(D4 ) = {1, r2 } is a normal subgroup of the dihedral
group D4 . The subgroups of D4 /K are D4 /K, {K} and

H1 = hrKi = {K, rK},

H2 = hsKi = {K, sK},
H3 = hsrKi = {K, srK}.

The subgroup lattice diagram of D4 /K is therefore:

D4 /K

H1 H2 H3

{K}

By the Correspondence Theorem, each of these subgroups Hi of D4 /K corresponds to

some subgroup of D4 , say Hi containing K1 . Using the map defined in the theorem, we
obtain the following subgroups of D4 :

H1 = {g ∈ D4 : gK ∈ H1 } = {1, r, r2 , r3 },
H2 = {g ∈ D4 : gK ∈ H2 } = {1, r2 , s, sr 2 },
H3 = {g ∈ D4 : gK ∈ H3 } = {1, r2 , sr, sr 3 }.

Putting this into a lattice diagram, we obtain the following subgroup structure for groups
of D4 containing K:

D4

H1 H2 H3

K
♦

31
 2.4. The Correspondence Theorem

Before we return to our new favourite kind of groups (metabelian groups), we need a
lemma.

Thu 14 Mar Lemma 2.17. If H and K are normal subgroups of a group G, then HK = KH and
HK is a normal subgroup of G.

Proof. By the Second Isomorphism Theorem, since K ◁G, we have that HK is a subgroup
of G. By Lemma 2.8, KH ⊆ HK so now we need to show that HK ⊆ KH. Take
hk ∈ HK. Since K ◁ G, we have hkh−1 ∈ K. Thus hk = (hkh−1 )h ∈ KH. So
HK = KH.
To show that HK ◁ G, take hk ∈ HK and g ∈ G. We will show that ghkg −1 ∈ HK.
We have

ghkg −1 = ghg −1 gkg −1 = h′ k ′ ,

for some h′ ∈ H and k ′ ∈ K, following from H and K being normal in G. Therefore we

have shown that gHKg −1 ⊆ HK, and so HK ◁ G.

Example 2.18. Using the Correspondence Theorem, we can return to the concept of
metabelian groups (Example 2.10) and look at whether the factor groups of a metabelian
group are metabelian.
Let G be a metabelian group where K ◁ G such that K and G/K are abelian. For
an arbitrary H ◁ G, we will prove that G/H is metabelian. To do so, we will find some
S ◁ G/H such that S and (G/H)/S are abelian. By the Correspondence Theorem, S
necessarily has the form N /H for some subgroup N of G with H ◁ N . So, let’s look for
this subgroup of G that contains H as a normal subgroup.
From the Second Isomorphism Theorem, H ◁ KH (here we have swapped the tradi-
tional roles of H and K, but that’s fine because H is normal now too). Moreover,

K/(K ∩ H) ∼
= KH/H.

By Lemma 2.17, KH ◁G. Then Correspondence Theorem (iv) tells us that KH/H ◁G/H.
We claim that KH/H is our special normal abelian subgroup S.
So, we need to show that KH/H and (G/H)/(KH/H) are abelian. By assumption,
K is a abelian, thus the factor group K/(K ∩ H) is also abelian and so is KH/H, by
the Second Isomorphism Theorem. Using two applications of the Third Isomorphism
Theorem, we have

(G/H)/(KH/H) ∼
= G/KH ∼
= (G/K)/(KH/K).

32
 2.4. The Correspondence Theorem

Again, by assumption G/K is abelian and thus so are any of its factor groups. Therefore
(G/H)/(KH/H) is also abelian. Finally, we conclude that any factor group G/H of a
metabelian group G is also metabelian. ♦

We end this chapter with an illustration depicting the First Isomorphism Theorem:

The poster depicts a factorisation. At the top

we find a mapping f , from A to B, represented
by the work of picking vegetables from the field
and depositing them in the warehouse. It de-
composes into three mappings, a surjective one
(left), the projection from A to A/ ker(f ), given
by the partition of the vegetables according to
their storage place; a bijective one (bottom),
from A/ ker(f ) to im(f ), given by the deposit
of the piles in the corresponding boxes; and an
injective one (right), the inclusion from im(f )
to B, given by the movement of the boxes to its
destination. The farmer is the mathematician
Emmy Noether, author of this theorem and a
fundamental figure of modern day algebra.
—Enric Cosme, Núria Tamarit

SOURCE: International Day of Mathematics

33
p-Groups 3

Recall that one of our goals for this course is to find a partial converse to Lagrange’s
Theorem. p-Groups will help us do so. These are groups whose orders are prime powers.
However, before we look at p-groups, we need to look at a new kind of algebraic structure
which will allow us to examine properties of p-groups.
Throughout this chapter, we will be considering X to be a nonempty set.

3.1 Group actions

Definition 3.1. Let G be a group and X a set. A group action by G on X is a map
∗ : G × X → X satisfying the following properties:

(i) Compatibility: g1 ∗ (g2 ∗ x) = (g1 g2 ) ∗ x for all g1 , g2 ∈ G and x ∈ X,

(ii) Identity: 1 ∗ x = x for all x ∈ X.

We say that G acts on X and call X a G-set #

Note: The set X is not necessarily a group itself. Also note the convention of using infix
notation g ∗ x instead of ∗(g, x) when apply a group action.

In some texts, ∗ : G × X → X is specified as a left group action on X. A right

group action is then a map ∗ : X × G → X, where x ∗ g ∈ X. We are going to focus on
left group actions, so we omit writing “left”.

Example 3.2. The following are some simple examples of group actions:

34
 3.1. Group actions

1. For any group G and set X, there is the trivial group action defined by g ∗ x = x
for all g ∈ G and x ∈ X.

2. If G is a group then G acts on G via the group action ∗ : G × G → G defined by

g ∗ x = gx for all g, x ∈ G. Notice that on the right-hand side we are using the
group operation from G. We call this G acting on itself by left multiplication.

3. If H is a subgroup of a group G and L represents the set of left cosets of H in G,

then G acts on L by the action g ∗ (g ′ H) = g(g ′ H) = gg ′ H.

4. If V is a vector space over a field F . Then the multiplicative group F × of nonzero

scalars acts on V by scalar multiplication. Namely, ∗ : F × × V → V defined by
λ ∗ v = λv satisfies the group action axioms. ♦

Example 3.3. A group G can act on itself by the group action defined by g ∗ x = gxg −1 .
This action is called conjugation. Moreover, if H is a normal subgroup of G, then G
acts on H by conjugation. ♦

Exercise 3.4. Prove that the maps defined in Examples 3.2 and 3.3 are group actions.

Definition 3.5. Let G be a group acting on a set X. A subset S of X is said to be

stable under the action of G if g ∗ x ∈ S for all g ∈ G and x ∈ S. The stabiliser of an
element x ∈ X, is the subgroup of G given by

S(x) = {g ∈ G : g ∗ x = x}. #

Note: You can think of a stable subset as a set that is closed under the group action.
Note that under this action, S itself becomes a G-set. Also note that subsets of X can be
called stable, whereas a stabiliser is a subgroup of G.

Question: Why is S(x) a subgroup of G?

Recall that an equivalence relation ∼ on a set X is a binary relation on X that

satisfies the following:

1. Reflexivity: x ∼ x for all x ∈ X,

2. Symmetry: x ∼ y implies y ∼ x,

35
 3.1. Group actions

3. Transitivity: if x ∼ y and y ∼ z, then x ∼ z.

The set [x] = {y ∈ X : x ∼ y} is called the equivalence class of x. All of the equivalence
classes of a set X under a given relation ∼ form a partition of X. An example of an
equivalence relation is the relation of congruency modulo some integer n on the set of
integers, Z. That is, the relation ∼ defined by a ∼ b if and only if a ≡ b mod n. Recall
that from this equivalence relation, we obtain the group Zn of integers modulo n, whose
elements are the equivalence classes under that relation.

Theorem 3.6. Let G be a group acting on a set X. The relation ∼ on X, defined by

x ∼ y if and only if y = g ∗ x for some g ∈ G, is an equivalence relation on X.

Proof. We must show that ∼ satisfies the three equivalence relation conditions: reflexivity,
symmetry and transitivity. Take x, y, z ∈ X. We have 1 ∗ x = x, therefore x ∼ x and
so reflexivity holds. Suppose that x ∼ y, then y = g ∗ x for some g ∈ G. Consider
g −1 ∗ (g ∗ x) = g −1 ∗ y. However, we also have g −1 ∗ (g ∗ x) = (g −1 g) ∗ x = x. Thus,
x = g −1 ∗ y and so symmetry holds. Now suppose that x ∼ y and y ∼ z. Thus, y = g ∗ x
and z = g ′ ∗ y for some g, g ′ ∈ G. Thus z = g ′ ∗ (g ∗ x) = (g ′ g) ∗ x and so x ∼ z. Hence ∼
is an equivalence relation on X.

Definition 3.7. Let G be a group acting on a set X. The equivalence class of x ∈ X

under the relation ∼ (from above theorem) is called the orbit of x under G. It is denoted
G ∗ x, or Gx, and we write
orbits partitons the set X
G ∗ x = {g ∗ x : g ∈ G}.

We say that G acts transitively on X if X has one unique orbit. Meaning for all
x, x′ ∈ X, we can find some g ∈ G such that g ∗ x = x′ . We call the action transitive.#

Note: The orbit of x is a subset of X, it consists of elements from X. Moreover, since

orbits are defined as equivalence classes, the orbits of X form a partition of X.

Graphically, orbits look something like this:

36
 3.1. Group actions

g5 ∗ x5

x5
·
g5 ∗ x4 ·
·
g∗x
x4 x1 ∀g ∈ G

g3 ∗ x3 g1 ∗ x1 x
x3 x2

g2 ∗ x2

On the left-hand side, we have the orbit of x1 , which is G ∗ x1 = {x1 , x2 , x3 , . . .}, where
each xi is obtained by acting upon some other xj by some g ∈ G. On the right-hand side,
we have the orbit of an element x. This orbit consists of only one element, G ∗ x = {x}.
A singleton orbit arises when g ∗ x = x for all g ∈ G. Meaning no matter which g acts on
x, we still get x after doing the group action. We will later see that these singleton orbits
can be quite useful.

Example 3.8. Consider the subgroup G = {1, r2 } of the dihedral group D4 , which rep-
resents the symmetries of a square. Recall that r2 represents a rotation by 180 degrees.
We can label the vertices of the square as a, b, c and d, as shown in the following diagram:

1 = R0 r2 = R180

a b c d

d c b a

Then G acts on the set X = {a, b, c, d} by the action defined by g ∗ x = the vertex where
x has moved to under the rotation g. For example, when we apply the rotation r2 = R180
to the vertex a, it moves into the original position of vertex c. Thus r2 ∗ a = c. (As an
exercise, prove that this is indeed a group action).
Let’s consider the orbit of each vertex. To do so, let’s calculate g ∗x for each g ∈ G and
x ∈ X. By the definition of group action, we do not need to perform these calculations
for g = 1, since 1 ∗ x = x for all x ∈ X. We have r2 ∗ a = c, r2 ∗ b = d, r2 ∗ c = a and

37
 3.1. Group actions

r2 ∗ d = b. So we have the following orbits:

G ∗ a = G ∗ c = {a, c},
G ∗ b = G ∗ d = {b, d}.

Notice how the orbits indeed form a partition of X = {a, b, c, d}. ♦

Theorem 3.9 (Orbit-Stabiliser Theorem). Let G be a group acting on a set X and

take x ∈ X. The size of the orbit G ∗ x is |G : S(x)|.

Proof. To prove this, let us construct a bijection between elements in the orbit G ∗ x and
the left cosets of S(x). Take y ∈ G ∗ x, then there is some g ∈ G such that y = g ∗ x. We
define a map φ from G ∗ x to the set of left cosets of S(x) by φ(y) = gS(x). To show that
this is well-defined, suppose y = g1 ∗ x = g2 ∗ x for some g1 , g2 ∈ G. We need to show that
g1 S(x) = g2 S(x), that is, g1−1 g2 ∈ S(x). We have

g1−1 g2 ∗ x = g1−1 ∗ (g2 ∗ x) = g1−1 ∗ (g1 ∗ x) = 1 ∗ x = x.

Thus g1−1 g2 ∈ S(x) and so g1 S(x) = g2 S(x). Therefore the image of y under φ is in fact
unique and independent of the choice of representative g.
Now we will show that φ is injective. Let y, y ′ ∈ G ∗ x. Then y = g ∗ x and y ′ = g ′ ∗ x
for some g, g ′ ∈ G. Suppose that φ(y) = φ(y ′ ). Then gS(x) = g ′ S(x), meaning that
g −1 g ′ ∈ S(x) and so g −1 g ′ ∗ x = x. This simplifies to g −1 ∗ y ′ = x. We have y = g ∗ x and
so y = g ∗ (g −1 ∗ y ′ ) = y ′ . Thus, φ is injective.
To show that φ is surjective, let gS(x) be a left coset of S(x). Let y = g ∗ x ∈ G ∗ x.
Then φ(y) = gS(x). So, φ is surjective. Therefore, we have shown that there is a bijection
between the orbit G ∗ x and the set of left cosets of S(x). Thus, |G ∗ x| = |G : S(x)|.

The Orbit-Stabiliser Theorem actually gives us a way to relate the size of the set X
to the indices of its stabilisers.

Corollary 3.10 (Orbit Equation). Let G be a group acting on a finite set X. Let T
denote a set of representatives of the distinct equivalence classes of ∼ (equivalently, the
representatives of all distinct orbits). Then,
X X
|X| = |G ∗ x| = |G : S(x)|.
x∈T x∈T

Proof. The first equality holds since the equivalence classes of ∼ (or, orbits) partition X.
The second equality holds by the Orbit-Stabiliser Theorem.

38
 3.1. Group actions

Definition 3.11. Let G be a group acting on a set X. We say that g ∈ G fixes an

element x ∈ X if g ∗ x = x. For g ∈ G, the fixed subset of g is the set of elements
x ∈ X that are fixed by g. We write this as

X g = {x ∈ X : g ∗ x = x}.

On the other hand, the fixed subset of G (or just fixed set) is the set of all x ∈ X that
are fixed by all elements of G. We write this as

Xf = {x ∈ X : g ∗ x = x for all g ∈ G}.

We call the elements of Xf the fixed points of the group action. #

Note: Fixed sets X g play a similar role to stabilisers S(x). One looks at the elements
of X fixed by a given g, the other looks at the elements of G which fix a given x.

Returning to the concept of orbits, we can divide the set of orbits of X into orbits
which are singletons and orbit which are not singletons. If an orbit of some x ∈ X is a
singleton, it means that the only element y ∈ X that satisfies y = g ∗ x, for some g ∈ G,
is the element y = x. That is, g ∗ x = x for all g ∈ G. This means that x belongs to the
fixed set Xf .

Remark 3.12. For an element x ∈ X, we have x ∈ Xf if and only if G ∗ x = {x}.

Moreover, Xf is the union of singleton orbits. ✠

Because the orbits of a set X partition that set, we have another version of the Orbit
Equation, where the sum is split between orbits which are singletons and orbits which are
not singletons.

Corollary 3.13 (Orbit Decomposition Theorem). Let G be a group acting on a fi-

nite set X. Let G ∗ x1 , G ∗ x2 , . . ., G ∗ xn denote the distinct nonsingleton orbits of X.
Then,
X
n
|X| = |Xf | + |G : S(xi )|.
i=1

Note: Sometimes this corollary is referred to as the Orbit Equation too.

39
 3.2. Conjugation

3.2 Conjugation
In Example 3.3, we saw that a group G can act on itself by conjugation, defined by
g ∗ x = gxg −1 . This is an important example of a group action. In later sections, we
will apply our results about group actions to the group action of conjugation to obtain
a partial converse to Lagrange’s Theorem. For now, let’s look at how our results about
group actions work for conjugation in particular.

Definition 3.14. Let G be a group and a, b ∈ G. We say that a and b are conjugate in
G if b = gag −1 for some g ∈ G. That is, a and b belong to the same orbit.
The conjugates of an element a are all elements of the form gag −1 for some g ∈ G.
That is, all elements in G ∗ a, the orbit of a. We call the orbit of a under conjugation the
conjugacy class of a. We write

class a = G ∗ a = {g ∗ a : g ∈ G} = {gag −1 : g ∈ G}. #

Note: Since the conjugacy classes of G are precisely the orbits of G, they also form a
partition of G.

Example 3.15.

1. If G is an abelian group then the action of G on itself by conjugation is the trivial

group action g ∗ x = gxg −1 = gg −1 x = x. The conjugacy class of each x ∈ G is the
singleton {x}.

2. In any group G, class 1 = {1}. Moreover, class a = {a} if and only if a ∈ Z(G),
since gag −1 = a for all g ∈ G if and only if a ∈ Z(G). ♦

Remark 3.16. Under conjugation, the stabiliser of x ∈ G is

S(x) = {g ∈ G : gxg −1 = x} = {g ∈ G : gx = xg}.

So, the stabiliser consists of all elements that commute with x. This has a special name,
the centraliser of x. It is a subgroup of G denoted by

CG (x) = {g ∈ G : gx = xg}.

Moreover, the centre of G can be written as

\
Z(G) = CG (x) = {g ∈ G : gx = xg for all x ∈ X}. ✠
x∈G

40
 3.2. Conjugation

Question: Why is the centraliser of x a subgroup of G? Why is the centre equal to the
intersection of the centralisers?

Theorem 3.17. Let G be a group and take a, b ∈ G. If a and b are conjugate, then their
orders are equal, that is, |a| = |b|.

Proof. Since a and b are conjugate, b = gag −1 for some g ∈ G. Recall that for each g ∈ G,
we have an inner automorphism σg : G → G defined by σg (x) = gxg −1 (Example 2.4).
Since a and b are conjugate, b = σg (a). Since σg is an automorphism, orders are preserved
and |a| = |b|.

This theorem means that we can determine conjugacy classes by looking at the orders
of the elements. Namely, the conjugacy class of a will be a subset of the set of elements
of G which have order |a|.

Example 3.18. Consider the dihedral group D3 = {1, r, r2 , s, sr, sr 2 } where |r| = 3,
|s| = 2 and rsr = s. The only elements of order 3 are r and r2 . By Theorem 3.17,
the only possible conjugate of r is r2 , that is class r ⊆ {r, r2 }. We confirm that r2 is
a conjugate of r by looking at srs−1 = r2 . Thus class r = {r, r2 }. Through similar
reasoning (now considering elements of order 2), we obtain class s ⊆ {s, sr, sr 2 }. Notice
that sr = rsr−1 and sr2 = r2 sr−2 . Thus class s = {s, sr, sr 2 }. Therefore, the distinct
conjugacy classes of D3 are class 1, class r and class s. ♦

Note: Using inner automorphisms, we have another way to express the conjugacy class
of an element a, namely,

class a = {σg (a) : g ∈ G},

where σg is the inner automorphism associated with g.

The definition of conjugation (g ∗ x = gxg −1 ) is reminiscent of normal subgroups (H

is normal in G if and only if gHg −1 = H for all g ∈ G). Let’s examine this connection to
see if we can learn anything interesting about conjugation or normal subgroups.
First, we can extend the conjugation group action to act on the set of all subsets of
G, that is the power set P(G). Then the conjugation group action ∗ : G × P(G) → P(G)
is defined by

g ∗ H = gHg −1 = {ghg −1 : h ∈ H},

for each g ∈ G and H ∈ P(G).

41
 3.2. Conjugation

Exercise 3.19. Prove that conjugation is a group action on P(G).

Definition 3.20. Let H and K be subsets of G. We call K a conjugate of H if there

is some g ∈ G such that K = gHg −1 . #

Remark 3.21. Let H be a subgroup of G. If H ◁ G, then gHg −1 = H for all g ∈ G.

Thus the only possible conjugate of H is H itself. We call H self-conjugate. ✠

In Remark 3.16, we saw that the stabiliser of x under conjugation is also known as
the centraliser CG (x) of x. We similarly have a special name for the stabiliser of a set H
under conjugation.

Remark 3.22. The stabiliser of a set H under conjugation is called the normaliser of
H in G, written as

NG (H) = S(H) = {g ∈ G : gHg −1 = H}.

Now take H to be a subgroup of G, not just a subset. Notice how NG (H) consists of all
elements g for which gHg −1 = H, one of the Normality Test conditions. It follows that
H ◁ NG (H). In fact, NG (H) is the largest subgroup of G that contains H as a normal
subgroup. Therefore, H ◁ G if and only if NG (H) = G. ✠

Question: Why is NG (H) the largest subgroup of G which contains H as a normal

subgroup?

By the above remark, we can see how the conjugation group action on a subgroup
gives us a way to determine whether a group is normal by examining the stabiliser.
Now let’s return to conjugation by G on G and look at the fixed set of the action. We
have

Gf = {a ∈ G : gag −1 = a for all g ∈ G}

= {a ∈ G : ga = ag for all g ∈ G}
= Z(G).

So, the fixed set of G is the centre of G. Using this, we obtain a special version of the
Orbit Equation for the group action of conjugation called the Class Equation.

Theorem 3.23 (Class Equation). Let G be a finite group and class a1 , class a2 , . . .,
class an be the distinct nonsingleton conjugacy classes. Then
X
n
|G| = |Z(G)| + |G : CG (ai )|.
i=1

42
 3.3. Cauchy’s Theorem

Notice how each of the terms on the right-hand side are divisors of |G|. The centre is a
subgroup of G thus its order must divide the order of G by Lagrange’s Theorem. Moreover,
each centraliser CG (ai ) is a subgroup of G, thus its index |G : CG (ai )| = |G|/|CG (ai )| also
divides |G|.

Example 3.24. In Example 3.18, we saw that the conjugacy classes of D3 were class 1,
class r and class s. The only singleton conjugacy class is class 1 = {1}. Therefore,
|Z(D3 )| = 1. On the other hand, |D3 : CD3 (r)| = |class r| = 2, by the Orbit-Stabiliser
Theorem. Similarly, |D3 : CD3 (s)| = |class s| = 3. Thus,

|D3 | = |Z(D3 )| + |D3 : CD3 (r)| + |D3 : C(D3 )(s)|

= 1 + 2 + 3 = 6. ♦

3.3 Cauchy’s Theorem

Recall that our goal is to find a partial converse to Lagrange’s Theorem. This next result
shows us why we have been looking at group actions. In particular, the Orbit Equation
will give us a way to find an element of a certain order, given the order of the group. This
is one of the partial converses to Lagrange’s Theorem we will see in this course.

Theorem 3.25 (Cauchy’s Theorem). Let G be a finite group. If p is a prime that

divides the order of G, then G has an element of order p.

Proof. Let X = {(g1 , g2 , . . . , gp ) : gi ∈ G, g1 g2 · · · gp = 1}. Take (g1 , g2 , . . . , gp ) ∈ X.

Then

g1 g2 · · · gp−1 gp = 1 =⇒ gp = (g1 g2 · · · gp−1 )−1 .

Therefore, the entry gp is determined by the p − 1 elements p1 , p2 , . . . , pn−1 . Thus |X| =

|G|p−1 . Consider the cyclic group Cp = hci where |c| = p. We define a group action
∗ : Cp × X → X by

c ∗ (g1 , g2 , . . . , gp ) = (g2 , g3 , . . . , gp , g1 ).

Notice how we have only defined the action for the element c ∈ Cp . However, all other
elements of Cp can be written as cn where 1 ≤ n ≤ p. Thus, cn ∗ x = c ∗ (c ∗ (· · · ∗ x)) for
all x ∈ X, where we are applying the group action n times. Basically, this group action
shifts all entries in a p-tuple by the order of the group element.

43
 3.3. Cauchy’s Theorem

Our goal is to use the Orbit Equation. For this, we need to understand the fixed set
and the nonsingleton orbits of the group action.
Note that Xf = {(g, g, . . . , g) : g ∈ G}, since the p-tuples where all entries are equal
will not change under this group action. Furthermore, we have X 6= Xf . This is because,
for a nonidentity element g ∈ G, the p-tuple (g, g −1 , 1, . . . , 1) belongs to X but not Xf .
Take x ∈ X with x ∈
/ Xf , that is, some x has a nonsingleton orbit. Then there is some
cm ∈ Cp such that cm ∗ x 6= x. Thus S(x) is a proper subgroup of Cp since cm ∈
/ S(x). By
Lagrange’s Theorem, |S(x)| must divide |Cp | = p. Since S(x) is a proper subgroup, we
must have |S(x)| = 1. Thus, |Cp : S(x)| = p/1 = p whenever x ∈ X has a nonsingleton
orbit.
Suppose that there are s nonsingleton orbits in X. Substituting |X| = |G|p−1 , n = s
and |Cp : S(xi )| = p into the Orbit Equation, we obtain

|G|p−1 = |Xf | + sp.

Notice that p divides both |G|p−1 and sp. Thus p must divide |Xf |. We know that |Xf | ≥ 1
since (1, 1, . . . , 1) ∈ X. Since p divides |Xf | and |Xf | ≥ 1, there are at least p elements in
|Xf |, including at least p − 1 tuples consisting of nonidentity elements. Thus, there are
at least p − 1 nonidentity elements g ∈ G such that g p = 1. Therefore, there are at least
p − 1 elements of G of order p. For any prime p, we have p − 1 ≥ 1. Thus G has at least
one element of order p.

Note: In the final step of the proof, the order of g must be p since g p = 1 implies |g| | p
so either |g| = 1 or p. However, g is not the identity, so we must have |g| = p.

The original proof of Cauchy’s Theorem used the Class Equation and induction on
the order of G. Give it a try if you’re interested!

Corollary 3.26. Let G be a finite group. If p is a prime that divides the order of G, then
G has a subgroup of order p.

Proof. By Cauchy’s Theorem, there exists some g ∈ G with order p. Then the cyclic
subgroup hgi of G has order p.

Sometimes the above corollary is also referred to as Cauchy’s Theorem.

Corollary 3.27. If G is a group of order 6, then G ∼

= Z6 or G ∼
= D3 .

44
 3.4. p-Groups

Proof. Since 2 and 3 are primes which divide |G|, Cauchy’s Theorem says that there
exist a, b ∈ G such that |a| = 2 and |b| = 3. Consider the cosets hbi and ahbi. We have
hbi = {1, b, b2 } and ahbi = {a, ab, ab2 }. This makes for a total of 6 distinct elements. Thus
G = {1, b, b2 , a, ab, ab2 }, since G has order 6.
Consider the product ba, which doesn’t appear in the above list of elements of G. So
ba must equal one of the previously listed elements. The only possibilities are that ba = ab
or ba = ab2 .
If ba = ab then (ab)2 = a2 b2 = b2 and (ab)3 = a3 b3 = a. So the order of ab must
instead be 6. This means that habi = G and so G is isomorphic to the cyclic group Z6 .
On the other hand, if ba = ab2 , then bab = a. From this, we see that the map
φ : G → D3 , defined by φ(a) = s and φ(b) = r, is an isomorphism. Thus G ∼
= D3 .

Question: Why are the elements of ahbi distinct? Why are the only two possibilities
ba = ab or ba = ab2 ? Why is φ an isomorphism?

The following corollary generalises the previous corollary to groups of order 2p for odd
prime p. The proof is not included, but it is very similar to the above proof. You won’t
be expected to know this result though.

Corollary 3.28. If p is an odd prime, then every group of order 2p is either cyclic or
dihedral.

3.4 p-Groups
Following Cauchy’s Theorem, we go a step further to look at groups whose orders are
only divisible by a single prime p.

Definition 3.29. Let p be a prime. A group G is called a p-group if |G| = pn for some
n ≥ 1. A subgroup H of a group G is called a p-subgroup if H is a p-group. #

Example 3.30.

1. For each prime p and positive integer k, the groups Zpk and Zp ×Zp ×· · ·×Zp = (Zp )k
are p-groups.

2. The dihedral group D4 is a 2-group since it has order 8 = 23 . On the other hand,
D3 is not a p-group since it has order 6, which is not a power of a prime.

45
 3.4. p-Groups

3. While Z6 is not a p-group, h[3]i is a 2-subgroup of Z6 . ♦

Theorem 3.31. A finite nontrivial group G is a p-group if and only if the order of every
element of G is a power of p.

Proof. By Lagrange’s Theorem, if |G| is a power of p, the only possible orders of elements
are powers of p. On the other hand, suppose the order of every element of G is a power
6 p that also divides |G|. Then q divides |G|. By
of p. Suppose there is a prime q =
Cauchy’s Theorem, there is an element of order q. This contradicts the assumption that
every element’s order is a power of p. Thus |G| must be a power of p since p is the only
prime divisor of |G|.

Theorem 3.32. Let G be a p-group. Then Z(G) is nontrivial.

Exercise 3.33. Prove the above theorem using the Class Equation.

Using this theorem, we obtain another partial converse to Lagrange’s Theorem.

Corollary 3.34. If G is a p-group with order pn , then for each k with 1 ≤ k ≤ n, G has
a normal subgroup of order pk .

Proof. We use proof by induction on n. If n = 1, then we can take G to be the normal

subgroup. Assume that the corollary is true for n − 1. Let G be a group of order pn .
By Theorem 3.32, the centre of G is a nontrivial subgroup. Then p divides the order of
Z(G). By Cauchy’s Theorem, there is some g ∈ Z(G) with order p. Let H = hgi. Since
H ⊆ Z(G), we have that H is a normal subgroup of G, moreover it has order p. So we’re
done with the case of k = 1.
Now consider the factor group G/H, which has order pn /p = pn−1 . By the inductive
hypothesis, for each k ′ with 1 ≤ k ′ ≤ n−1, there is some normal subgroup of G/H of order
′
pk , say K. By the Correspondence Theorem, we can write K = K/H for some normal
′ ′
subgroup K of G. By Lagrange’s Theorem, |K| = |K||H| = pk p = pk +1 . Therefore,
(upon labelling k = k ′ + 1) we have found a normal subgroup K of G with order pk where
1 ≤ k ≤ n.

Recall that every group of order 4 is isomorphic to either K4 (∼

= Z2 × Z2 ) or Z4 . The
following theorem generalises this to groups of order equal to any squared prime.

46
 3.4. p-Groups

Theorem 3.35. If G is a group of order p2 , then G is abelian. Moreover, G is isomorphic

to either Zp × Zp or Zp2 .

Proof. Note that G is a p-group. Therefore, the centre of G is nontrivial and so has order
p or p2 . If |Z(G)| = p then Z(G) 6= G. However, the factor group G/Z(G) has prime
order p and so it is cyclic. By Theorem 1.45, G must be abelian and so Z(G) = G. This
is a contradiction so we must have |Z(G)| = p2 . Therefore G = Z(G) and G is abelian.
Now, if G has an element of order p2 , then G is cyclic and is thus isomorphic to Zp2 .
On the other hand, if there are no elements of order p2 , all nonidentity elements must have
order p. Take some nonidentity elements a, b ∈ G such that b ∈ / hai. Since |a| = p, we
have hai ∼
= Zp . Moreover, hai is a subgroup of ha, bi. Since b ∈
/ hai, this must be a strict
subgroup and so the only possibility for the order of ha, bi is p2 . That is, G = ha, bi. The
map φ(an , bm ) = an bm is an isomorphism from hai × hbi to G. However, hai ∼ = Zp ∼ = hbi
and so G ∼
= Zp × Zp .

Question: Why is it possible to find b ∈ G with b ∈

/ hai? Why is φ an isomorphism?

Note: The theorem does not extend to groups of order p3 . For example, the group D4
has order 23 = 8 and is not abelian.

Wed 26 Apr Theorem 3.36. Let G be a p-group. Suppose G acts on a set X, then p divides |X|−|Xf |.
Equivalently, |X| ≡ |Xf | mod p.

Exercise 3.37. Prove the above theorem. Hint: Use the Orbit Equation.

Theorem 3.38. Let G be a p-group. If H is a proper subgroup of G, then NG (H) 6= H.

Proof. Let X denote the set of left cosets of H in G. We define a group action on X by H
using left multiplication: h∗gH = hgH. Note that |Xf | ≥ 1 since h∗H = H for all h ∈ H
and so H ∈ Xf . By Theorem 3.36, p divides |X|−|Xf |. However, |X| = |G : H| = |G|/|H|.
So p also divides |X|. Thus |Xf | > 1. Notice that gH ∈ Xf if and only if hgH = gH for
all h ∈ H. This is equivalent to saying g −1 hg ∈ H. That is, g −1 Hg = H. However, this
is precisely the condition for g to be in NG (H). Thus gH ∈ Xf if and only if g ∈ NG (H).
Since |Xf | > 1, there is some element gH ∈ Xf , with g ∈ G, such that gH 6= H, that is,
g∈
/ H. Meaning that there exists some g ∈ G such that g ∈ NG (H) and g ∈
/ H. Thus we
conclude that NG (H) 6= H.

47
 3.4. p-Groups

Corollary 3.39. Let G be a p-group. If H is a proper subgroup of G, then we have that

|G : H| ≡ |NG (H) : H| mod p.

Proof. In the proof of the previous theorem, we see that g ∈ NG (H) if and only if
gH ∈ Xf . So Xf consists of left cosets with representatives from NG (H). This means
|Xf | = |NG (H) : H|. Recall that |X| = |G : H|. Furthermore, from Theorem 3.36, we
have |G : H| ≡ |NG (H) : H| mod p.

48
 3.4. p-Groups

Coming soon...

Sylow Theorems
Finitely Generated Abelian Groups

49