Basic Group Theory (Week 5) PDF
Basic Group Theory (Week 5) PDF
A group is a set G together with a function ∗ : G × G → G, assigning to each pair (a, b) of elements
of G another element a ∗ b ∈ G, satisfying the following three axioms:
There is no requirement that a ∗ b = b ∗ a for all a, b ∈ G. If this property does hold, we say that G is
abelian.
by applying successively axioms G2, G3 (for a0 ), G1, G3 (for a00 ) and finally G2 again.
If G is abelian, and only in this case, we sometimes use additive notation, writing a∗b = a+b, denoting
the identity element by 0, and the inverse element of a by −a.
5
The order of G is the cardinality |G|, either a positive integer or ∞. Any group G of order one consists
of the identity element only and is called the trivial group.
We can extend the product of elements in G to subsets: If S and T are subsets of a group G, we define
ST = {st : s ∈ S, t ∈ T }.
The structure of a group G is revealed by its subgroups and homomorphisms into other groups.
This implies that f (1G ) = 1G0 and that f (g −1 ) = f (g)−1 for all g ∈ G.
For all a, b ∈ G we have f (a) = f (b) if and only if ab−1 ∈ ker f . Hence f is injective iff ker f = {1G }.
The image of f is the set theoretic image, defined as above by
im f := f (G) = {g 0 ∈ G : f −1 (g 0 ) 6= ∅}.
Hom(G, G0 ) = {homomorphisms f : G → G0 }.
We sometimes abuse terminology and say that G is or is a copy of G0 , when we really mean only that
G ' G0 . For example, any two trivial groups are isomorphic, so we allow ourselves to say that the
trivial group is the unique group with one element.
6
The set Aut(G) forms a group under composition, whose identity element is the identity automor-
phism, which sends g 7→ g for all g ∈ G.
Many automorphisms arise from within the group itself as follows. For each element g in a group G,
the map
cg : G → G given by cg (x) = gxg −1 ∀x ∈ G
is an automorphism of G, called conjugation by g. Automorphisms of this form are called inner
automorphisms. The function c : G → Aut(G) sending g 7→ cg is a homomorphism.
2.3 Subgroups
The subsets {1} and G are subgroups of G. All other subgroups of G, if any, are called proper
subgroups. We write H ≤ G to indicate that H is a subgroup of G which is possibly equal to G itself.
We write H < G for a subgroup which is not equal to G.
Lemma 2.1 Let G be a group and let H be a nonempty finite subset of G. Then H ≤ G if and only if
SG1 holds.
Proof: Let h be an element of the nonempty set H. Since H is finite, the powers h, h2 , h3 , . . . must
eventually repeat, so we have hi = hj for some positive integers i < j. It follows that hj−i = 1, so
SG2 holds, and h · hj−i−1 , so SG3 holds. Hence H is a subgroup of G.
This proof moved from H to a particular kind of subgroup of G. A group C is cyclic if there exists an
element c ∈ C such that C = {cn : n ∈ Z}. In this case we write C = hci.
More generally, if S is any subset of G, the subgroup generated by S is the smallest subgroup hSi of
G containing S. That is, \
hSi = H
S⊆H≤G
is the intersection of all subgroups H of G which contain the subset S.
7
2.4 Cosets and quotient spaces
A left coset of a subgroup H < G is a subset of G of the form gH = {gh : h ∈ H}. Two left cosets
are either equal or disjoint; we have
gH = g 0 H ⇔ g −1 g 0 ∈ H.
In particular, we have gH = H if and only if g ∈ H. The set of left cosets of H in G is denoted G/H,
and is called the quotient of G by H.
A right coset of H in G is a subset of the form Hg = {hg : h ∈ H}. Two right cosets are either equal
or disjoint; we have
Hg = Hg 0 ⇔ g 0 g −1 ∈ H.
In particular, we have Hg = H if and only if g ∈ H. The set of right cosets of H in G is denoted
H\G.
A coset is a left or right coset. Any element of a coset is called a representative of that coset. We have
canonical bijections H → gH and H → Hg, sending h 7→ gh and h 7→ hg, respectively. Hence if H
is finite, all cosets have cardinality |H|.
There are an equal number (including infinity) of left and right cosets in G. We denote this number by
[G : H] = |G/H| = |H\G|,
and call it the index of H in G. If G is finite, then G is partitioned into [G : H] cosets, each of
cardinality |H|. It follows that [G : H] = |G|/|H|. In particular we have
Example 1: If |G| = p a prime, then G has no proper subgroups. Hence for any nonidentity element
g ∈ G we have G = hgi, so G is cyclic.
Example 2: The order of any element in a finite group G divides |G|. in particular, we have g |G| = 1
for all g ∈ G. The smallest positive integer e such that g e = 1 for all g ∈ G is called the exponent of
G. If g ∈ G has order m, then m divides e, which in turn divides |G|.
The converse of Lagrange’s theorem is false. The smallest counterexample is the group A4 of order 12,
which has no subgroup of order 6. However, the converse of Lagrange’s theorem is true for subgroups
H of prime power order. This is part of the Sylow theorems, which we will prove later. However, one
special case is easy:
Proposition 2.2 Any group of even order contains an element of order two.
Proof: Suppose G has even order |G| = 2m. Pair the nonidentity elements with their inverses. Since
there are 2m − 1 such elements, at least one of them is paired with itself. This is a nonidentity element
g ∈ G such that g = g −1 . Thus, g is an element of G of order two.
8
2.5 Normal subgroups and quotient groups
Let G be a group and let H ≤ G be a subgroup. One attempts to define a group structure on the set
G/H by the rule:
gH ∗ g 0 H = gg 0 H, ∀g, g 0 ∈ G. (1)
However, this rule is only well-defined when every left coset of H in G is also a right coset of H in G.
The subgroup H is said to be normal in G if gH = Hg for all g ∈ H. On the level of elements, this
means that ghg −1 ∈ H for all g ∈ G and h ∈ H. If G is abelian, then ghg −1 = h, so every subgroup
is normal in G. Thus, being normal in G is a weakening, with respect to H, of the abelian condition.
We write H / G or H E G to indicate that H is a normal subgroup of G.
When, and only when H E G, the set G/H = H\G becomes a group under the operation given by
(1). We call G/H the quotient of G by H. It is a group of order equal to the index of H in G:
|G/H| = [G : H].
Proof: Exercise.
Theorem 2.4 (First isomorphism theorem) Let f : G → G0 be a group homomorphism with kernel
K = ker f . Then K is a normal subgroup of G, and there is an isomorphism
∼
f¯ : G/K −→ im f, given by f¯(gK) = f (g). (2)
Theorem 2.5 Let f : G → G0 be a group homomorphism with kernel K = ker f . Let H be a normal
subgroup of G contained in K. Then there is a surjective homomorphism
f¯ : G/H im f, given by f¯(gH) = f (g), (3)
with ker f¯ = K/H.
9
Proof: Again, this is a good exercise.
Conversely, every normal subgroup H E G is the kernel of a surjective homomorphism from G into
another group. Namely, the canonical homomorphism
[x, y] = xyx−1 y −1
[G, G] = h[x, y] : x, y ∈ Gi
is the subgroup generated by all commutators. For g ∈ G we have g[x, y]g −1 = [gxg −1 , gxg −1 ]. It
follows that [G, G] E G. The quotient G/[G, G] is called the abelianization of G, and is often denoted
Gab , because of the following result.
Proposition 2.6 The quotient G/[G, G] of a group G by its commutator subgroup is abelian. More-
over, G/[G, G] is the largest abelian quotient of G, in the following sense: If f : G → A is a homo-
morphism from G to an abelian group A, then [G, G] < ker f , so f factors through a homomorphism
f¯ : G/[G, G] −→ A.
Proof: Exercise.
with
ker f1 = {1}, im f1 = ker f2 ' G1 , im f2 = G3 ' G2 /G1 .
10
2.7 The second isomorphism theorem (correspondence theorem)
Theorem 2.7 Let G be a group with normal subgroup H E G, and let πH : G → G/H be the
canonical homomorphism.
There is also a correspondence theorem for homomorphisms, the first part of which is just Thm. 2.5
above.
11
2.8 The third isomorphism theorem
The final isomorphism theorem concerns products of subgroups. If H and K are subgroups of a group
G, the product HK = {hk : h ∈ H, k ∈ K} contains the identity, but it need not be closed under
the operation in G, hence HK need not be a subgroup of G. However, it will be so under an additional
condition.
Theorem 2.9 (Third Isomorphism Theorem) Let H and K be subgroups of a group G and assume
that K ≤ NG (H). Then
Proof: We have already proved the first part, and the second part is easy. As for the third part, it is
clear that f is surjective, so it remains to determine ker f . Let k ∈ K. Then f (k) = 1 in HK/H iff
ι(k) ∈ H, which means that k ∈ H. But k ∈ K, so we have f (k) = 1 iff k ∈ H ∩ K, as claimed.
Let H and K be groups with identity elements 1H and 1K . Then the direct product
H × K = {(h, k) : h ∈ H, k ∈ K}
12
is a group under the operation
(h, k)(h0 , k 0 ) = (hh0 , kk 0 )
and identity element (1H , 1K ). The direct product of finitely many groups G1 , . . . , Gn is defined simi-
larly; we confine our discussion to the case n = 2.
H 0 ∩ K 0 = {1} and H 0 K 0 = G.
2. H ∩ K = {1};
3. G = HK.
Proof: Let h ∈ H, k ∈ K and parse the commutator [h, k] = hkh−1 k −1 in two ways. On the one
hand, [h, k] = (hkh−1 )k −1 ∈ K since K E G. On the other hand, [h, k] = h(kh−1 k −1 ) ∈ H, since
H E G. But H ∩ K is trivial, so [h, k] = 1. Hence h and k commute for all h ∈ H and k ∈ K. It is
now immediate that f is a homomorphism, which is surjective by assumption 3. Finally, if f (h, k) = 1,
we have h = k −1 ∈ H ∩ K = {1}, so h = k = 1. Therefore f is an isomorphism, as claimed.
If H and K are abelian groups then we often write H ⊕ K instead of H × K, in accordance with our
use of additive notation for abelian groups.
Recall that a group G with two normal subgroups H, K E G is the direct product G = H × K iff
HK = G and H ∩ K = {1}. This situation often occurs with the variation that only one of the
subgroups is normal.
13
2. H ∩ K = {1}.
3. HK = G.
Suppose G is the semidirect product of H and K and (say) H is normal in G. On the set H × K, define
a group law as follows:
(h, k)(h0 , k 0 ) = (hkh0 k −1 , kk 0 ).
Let H o K denote the set H × K with this group law.
Proposition 2.12 If G is a semidirect product of two subgroups H and K with H E G, then the map
(h, k) 7→ hk is a group isomorphism
∼
H o K −→ G.
Proof: exercise.
2.11 Conjugacy
Some of our earlier notions can be expressed in terms of conjugacy. For example, we have x ∈ Z(G)
if and only if G x = {x}. And H E G if and only if H is a union of conjugacy classes in G.
|G|
|G x| = [G : CG (x)] = .
|CG (x)|
14
Proof: Exercise
The last formula in Prop. 2.13 is very useful for computing the sizes of conjugacy classes and central-
izers in finite groups. It implies for example that |G x| divides |G|. Since G is the disjoint union of its
conjugacy classes, we have
Corollary 2.14 (The class equation) Let G be a finite group, let X1 , . . . , Xk be its conjugacy classes,
choose xi ∈ Xi for 1 ≤ i ≤ k, and let Gi = CG (xi ). Then we have
k
X
[G : Gi ] = |G|.
i=1
Alternatively,
1 1 1
+ + ··· + = 1.
|G1 | |G2 | |Gk |
Proof: We have G partitioned as G = Z(G) ∪ {noncentral elements}. Every conjugacy class has
size a power of p. This power is zero precisely for those classes consisting of a single element
in Z(G) and every conjugacy class of noncentral elements has size a positive power of p. Hence
|{noncentral elements}| is divisible by p. As |G| is also divisible by p, it follows that p divides |Z(G)|.
Since |Z(G)| ≥ 1, it follows that a positive power of p divides |Z(G)|.
Groups usually arise as symmetries of mathematical structure. The most basic structure is just a set.
For any set X, the symmetric group on X is the group SX of bijections σ : X → X from X to itself,
where the group operation is composition of functions: (στ )(x) = σ(τ (x)). The identity element is
the identity function e(x) ≡ x. The elements of SX are usually called permutations.
∼
If β : X → Y is a bijection between two sets X and Y , then we get a group isomorphism Sβ : SX −→
SY , defined by
Sβ (σ) = β ◦ σ ◦ β −1 .
If X = Y then β ∈ SX and Sβ is conjugation by β.
Assume now that X is finite, say |X| = n. Then SX is finite and one checks that
|SX | = n! = n(n − 1) · · · 2 · 1.
15
Given σ ∈ SX , define an equivalence relation ∼ on X by the rule:
σ
The equivalence classes are called σ-orbits. If O is a σ-orbit and x ∈ O, the distinct elements of O are
Number the σ-orbits in X as O1 , . . . Oq , where |O1 | ≥ |O2 | ≥ · · · ≥ |Oq |, and set λi = |Oi | for each
i. Thus, we have a set partition
q
G
X= Oi
i=1
Let us choose xi ∈ Oi , for each σ-orbit Oi , and set xji = σ j−1 (xi ), so that
Oi = {xji : 1 ≤ λi }.
Proposition 3.1 Two elements σ, σ 0 ∈ SX are conjugate if and only if λ(σ) = λ(σ 0 ).
Conversely, suppose λ(σ 0 ) = λ(σ). Let O1 , . . . , Oq be the σ-orbits and let O10 , . . . , Oq0 0 be the σ 0 -orbits.
Since λ(σ) = λ(σ 0 ) we have q = q 0 and |Oi | = |Oi0 | for each i. Since the orbits are disjoint there exists
16
a permutation τ ∈ SX such that τ (Oi ) = Oi0 for each i. Replacing σ by τ στ −1 , we may assume that
Oi0 = Oi for each i.
Choose xi ∈ Oi , and set xji = σ j−1 (xi ), yij = (σ 0 )j−1 (xi ), so that x1i = yi1 = xi and
Oi = {x1i , x2i , . . . , xλi i } = {yi1 , yi2 , . . . , yiλi }.
Every element of X may be uniquely expressed in the form xji or in the form yij , where 1 ≤ i ≤ q and
1 ≤ j ≤ λi . Let π ∈ SX be the permutation sending xji 7→ yij for each such i, j. For x = xji , we have
πσπ −1 (x) = πσσ j (xi ) = πσ j+1 (xi ) = (σ 0 )j+1 (xi ) = σ 0 (x).
It follows that πσπ −1 = σ 0 , so σ and σ 0 are conjugate.
When X = {1, 2, . . . , n} we write Sn = SX . If X is any set with |X| = n then by labelling the
∼
elements of X we obtain a bijection f : X → {1, 2, . . . , n}, whence a group isomorphism Sf : SX −→
Sn . This isomorphism is noncanonical, because it depends on the chosen labelling f . Moreover, the set
X need not be ordered, as {1, 2, . . . , n} is. The ordering gives additional structure to Sn that is missing
in SX when X is unordered
As in (4), an element in Sn is a product of cycles, but now the symbols in the cycle are numbers in
{1, 2, . . . , n}. For example, the element σ ∈ S6 sending
1 7→ 4, 2 7→ 6, 3 7→ 1, 4 7→ 3, 5 7→ 5, 6 7→ 2
may be written as
σ = (1 4 3)(2 6),
and has cycle type λ(σ) = [3, 2, 1]. Note that 5, which is fixed, is omitted in the cycle decomposition
of σ, but is counted in the cycle type of σ.
We multiply using the cycle decomposition by following the path of each number, starting with the
right-most cycle. For example, we have
(1 4 3)(2 6)(4 6 5) = (1 4 2 6 5 3).
Note that the cycle type of τ = (1 4 3)(2 6)(4 6 5) is not [3, 3, 2], because the cycles are not disjoint.
We first have to make them disjoint, as we have done above. We obtained a single 6-cycle (1 4 2 6 5 3),
so the cycle type of τ in S6 is λ(τ ) = [6].
17