Introduction to Group Structures

Hong-Jian Lai

October 2002

1. Group Actions
(1.1) A group action of a group G acting on a set S is a map from G × S to S (written
as g · s or g(s) for g ∈ G and s ∈ S) such that g1 · (g2 · s) = g1 g2 · s and eG · s = s.
If a group G is acting on a set S, we call S a G-set. Suppose S is a G-set. Define a
relation s1 ∼ s2 ⇐⇒ s1 = g(s2 ) for some g ∈ G, then ∼ is an equivalence relation. (The
equivalence classes are called orbits of S under G. If an orbit contains s ∈ S, then the
orbit is called the orbit of s. The orbit of s := sG = {g(s)|g ∈ G}.)
Proof: Check definition of equivalence relation.

(1.1a) Let H ≤ G and S be the left cosets of H. Then g · aH = (ga)H is an action of G on

S, called the left multiplication action.
(1.1b) Let H be a subgroup of G, and a, b ∈ G. Define a ∼ b ⇐⇒ for some g ∈ H,
a = hbh−1 . Verify that ∼ is an equivalence relation. Then h · a = hah−1 is an action of
H on G, (called the action of H on G by conjugation). The orbits of G acting on G by
conjugation are called conjugacy classes.
(1.1c) Let G ≤ SΩ . Then G acts on Ω by g · s = g(s). If G = hgi, then the cycle decompo-
sition of g gives the orbits of this action.

(1.2) Suppose that G is acting on a set Ω. For s ∈ Ω, the stabilizer of s in G is

Gs = {g ∈ G|gs = s}.

The kernel of the action is

ker(G/Ω) = {g ∈ G|gs = s, ∀s ∈ Ω}.

Then
(i) Gs ≤ G.

1
 (ii) There is a group homomorphism φ : G 7→ SΩ such that ker(G/Ω) = ker(φ)  G.

Proof: (i) For g1 , g2 ∈ Gs , g1 g2 (s) = g1 (g2 (s)) = g1 (s) = s, and g1−1 (s) = g1−1 (g1 (s)) =
e(s) = s.
(ii) ∀g ∈ G, let φ(g) : Ω 7→ Ω be given by φ(g)(s) = g(s). Since g −1 (g(s)) == e(s) = s, φ(g)
is onto. Since g(s1 ) = g(s2 ) implies that s1 = g −1 (g(s1 )) = g −1 (g(s2 )) = s2 , φ(g) is 1-1.
Therefore, φ(g) ∈ SΩ . That g1 (g2 (s)) = (g1 g2 )(s) implies that φ is a group homomorphism.
Finally,

ker(φ) = {g ∈ G : φ(g) = identity in SΩ } = {g ∈ G : φ(g)(s) = s, ∀s ∈ Ω} = ker(G/Ω).

(1.2a) Let A ⊆ G and let G acts on S = G by conjugation, i. e. g(x) = gxg −1 , ∀x ∈

S, g ∈ G. Then ker(G/A) = CG (A). In particular, ker(G/G) = Z(G).
Proof Checking definition.
(1.2b) Conjugation in Sn : If σ = (a1 a2 · · · ak1 )(b1 b2 · · · bk2 ) · · · and τ ∈ Sn , then

τ στ −1 = (τ (a1 )τ (a2 ) · · · τ (ak1 ))(τ (b1 )τ (b2 ) · · · τ (bk2 )) · · ·

(1.2c) Definition: for a positive integer n, a partition of n is a nondecreasing sequence of

Pr
integers 0 < n1 ≤ n2 · · · nr such that i=1 ni = n. If a permutation σ ∈ Sn is the product of
disjoint cycles of length n1 , n2 , · · · , nr , respectively, where 0 < n1 < · · · < nr is a partition
of n, then the cycle type of σ is n1 , n2 , · · · , nr .
Two elements of Sn are conjugate if and only if they have the same cycle type. The
number of conjugacy classes of Sn is the same as the number of partitions of n.
(1.3) A group G acts transitively on a set S if for any pair s, s0 ∈ S, there if a g ∈ G such
that g(s) = s0 . (In other words, G has only one orbit: G = sG for any s ∈ S).
Let H ≤ G, and let G acts on the set X of left cosets of H by left multiplication
(g(f H) = (gf )H). Then
(i) G acts transitively.
(ii) GH = H.
(iii) ker(G/X) = ∩g∈G gHg −1 , which is the largest normal subgroup of G contained in
H.
Proof (i) and (ii) follows by definition.

ker(G/X) = {g ∈ G|g(xH) = xH, ∀x ∈ G}

= {g ∈ G|x−1 gx ∈ H, ∀x ∈ G}
= {g ∈ G|g ∈ xHx−1 , ∀x ∈ G}.

2
Thus ker(G/X)  G. When x = 1, we have ker(G/X) ≤ H. Use definition to show that if
N  G and N ≤ H, then N ≤ ker(G/X).

(1.3a) (Cayley, Thm 4.6) Every group is isomorphic to a subgroup of a symmetric group
S|G| .
Proof Let G acts on G by left (or right) multiplication.
(1.3b) If |G| = n < ∞ and if p is the smallest prime dividing n, then any subgroup H of
index p is normal.
Proof: Assume |G : H| = p. Let L be the collection of left coset of H in G and let G
act on L by left multiplication: g(xH) = (gx)H. Let K = ker(G/L) be the kernel of the
action. (want to show H = K!) By (1.3)(iii), K ≤ H, and so by the index formula:

|G : K| = |G : H||H : K| = p|H : K|.

By the 1st Iso Thm, G/K is iso to a subgroup of Sp (since |L| = |G : H| = p), and so by
Lagrange’s Thm,

p|H : K| = |G/K| is a factor of |Sp | = p! =⇒ |H : K| is a factor of (p − 1)!.

Since p is the smallest prime factor of |G|, |H : K| = 1.

(1.4) (Thm 4.3) If X is an G-set and if x ∈ X, then |xG | = |G : Gx |.

Proof Define φ : xG → { left cosets of Gx } by: φ(g(x)) = gGx . Verify that φ is well de-
fined: If h(x) = g(x), then g −1 h ∈ Gx and so gGx = hGx .
φ is an injection: φ(g(x)) = φ(h(x)) ⇐⇒ gGx = hGx ⇐⇒ g = hk for some k ∈ Gx . Hence
g(x) = hk(x) = h(x).
φ is onto: Definition.

(1.5) (Burnside) Let G be a finite group and X be a finite G-set. For each g ∈ G, de-
fine Xg = {x ∈ X|g(x) = x}. If r denotes the number of orbits in X under G, then
X
r|G| = |Xg |.
g∈G

Proof: Let (x1 )G , (x2 )G , · · · , (xr )G be the orbits of X under G. Let P denote all the pairs
(x, g) where g(x) = x. (P is number of 1’s in the |X| × |G| (0,1)-matrix where the (x, g)
entry is 1 iff (g(x) = x. Thus |Xg | = number of 1’s in the gth column and |Gx | = number
of 1’s in the xth row.) Hence
X X
|P | = |Xg | = |Gx |.
g∈G x∈X

3
By (1.4), |Gx | = |G|/|xG |, and so
r r
X X 1 X X 1 X
|Gx | = |G| = |G| = |G| 1 = |G|r.
x∈X x∈X
|xG | i=1 x∈(x )
|xG | i=1
i G

(1.6) Let X be a G-set. The fixed points of the action is

XG = {x ∈ X|g(x) = x∀g ∈ G}.

XG is the union of all single element orbits. If (x1 )G , (x2 )G , · · · , (xs )G are the orbits of X
under G that are non-single-element orbits, then
s
X
|X| = |XG | + |(xi )G |.
i=1

Proof By (1.1).

(1.6a) (The Class Equation, Cor 4.5) Let |G| < ∞ and let g1 , g2 , · · · , gr be representatives
of the distinct conjugacy classes of G − Z(G). Then
r
X
|G| = |Z(G)| + |G : CG (gi )|.
i=1

Proof: Let G acts on X = G by conjugation. Then XG = Z(G) and CG (gi ) = Ggi . (1.6a)
follows by (1.4).

(1.6b) Let G be a group of order pn for some prime p, and let X be a finite G-set. Then
|X| ≡ |XG | (mod p).
Proof: By (1.4), each |(xi )G | is a factor of |G| = pn . (1.6b) follows then by (1.6).
(1.6c) Consider the examples when G = Q8 and when G = D8 .

(1.7) Let G be a group with |G| = pn (p a prime) and let X be a G-set. Then |X| ≡ |XG |
(mod p).
Proof: By (1.6) and (1.4).

(1.7a) If |G| = pm for some prime p, then |Z(G)| > 0.

Proof: G acts on X = G by conjugation. By (1.7), p divides |XG | and 1 ∈ XG .

(1.7b) If |G| = p2 , then G ∈ {Zp2 , Zp × Zp }.

Proof G/Z(G) is cyclic. By (I.4.9a), G is abelian. If G has an element or order p2 , then

G = Zp2 . Otherwise choose x ∈ G with |x| = p and y ∈ G−hxh. Show G = hx, yh∼= hxh×hyh.

4
(1.8) (Cauthy, Thm 5.2) If |G| < ∞ and if p is a prime dividing |G|, then there is some
g ∈ G such that |g| = p.
Proof Let X = {(gp , · · · , g2 , g1 )|gi ∈ G and gp · · · g2 g1 = 1}.
(Step 1) |X| = |G|p−1 (gp = (gp−1 · · · g2 g1 )−1 ). Since p divides |G|, p divides |X|.
(Step 2) Let σ = (1, 2, · · · , p) and H = hσh. Let σ act on X by σ(gp , · · · , g2 , g1 ) =
(gσ(p) , · · · , gσ(2) , gσ(1) ) = (g1 , gp , · · · , g3 , g2 ) ∈ X. By (1.7), |X| ≡ |XH | (mod p). By (Step
1) and since (1, 1, 1, · · · , 1) ∈ XH , XH has at least p elements. Note that σ(gp , · · · , g2 , g1 ) =
(gp , · · · , g2 , g1 ) implies a = g1 = g2 = · · · = gp and so |a| = p.

5
 2. Sylow’s Theorems

Groups in this section are all finite groups.

(2.1) Definition: Let G be a group and p be a prime. A (sub)group H with H| = pa

is called a p-(sub)group. A maximal p-subgroup of G is a Sylow p-subgroup of G. The
set of all Sylow p-subgroups of G is Sylp (G), and np = |Sylp (G)|.
(2.2) (Lemma 5.5) Let H be a p-group of G. Then

[NG (H) : H] = [G : H](mod p).

Proof: Let L be the left cosets of H in G, and let H act on L by h(xH) = (hx)H. Then
L is an H-set. Note that |L| = [G : H], and note that ∀h ∈ H, x ∈ G,

hxH = xH ←→ x−1 hxH = H ⇐⇒ x−1 hx ∈ H ⇐⇒ x ∈ NG (H).

Hence the number of fixed points of this action is |LH | = [NG (H) : H]. Since |H| = pn , by
(1.6b), [NG (H) : H] = |LH | ≡ [G : H](mod p).
(2.2a) (Cor 5.6) Let H be a p-subgroup of a finite group G. If p divides [G : H], then
NG (H) 6= H.

(2.3) The First Sylow Theorem (Thm 5.7) Let G be a group of order pa m, where
a > 0 and where p is a prime not dividing m. Then
(i) G contains a subgroup of order pi for each i where 0 ≤ i ≤ a,
(ii) every subgroup H of G of order pi is a normal subgroup of a subgroup of order pi+1
for 1 ≤ i < a.
Proof: (i) By induction on a. Clearly G has a subgroup of order 1. Suppose that H is a
subgroup of order pi for some 0 ≤ i < a. Since i < a, p divides pa−i m = |G : H|.
By (2.2a), |NG (H) : H| > 0. Since H  NG (H), the group NG (H)/H has p as a
factor, and so by Cauchy’s Theorem (1.8), NG (H)/H has a subgroup K 0 of order p. Let
K = {k ∈ NG (H)|kH ∈ K 0 }. Then K is the inverse image of a homomorphism, and so
K ≤ N( G) ≤ G. Since |K : H| = |K 0 | = p, |K| = pi+1 .

(ii) Note that in (i), H ≤ K ≤ NG (H). Since H  NG (H), H  K.

(2.4) The Second Sylow Theorem (Thm 5.9) Let G be a finite group and let P1 , P2 ∈

6
Syl(G). Then there is some g ∈ G such that P1 = gP2 g −1 .

Proof: (The trick here is to let one (P2 , say) act on the left cosets of the other (P1 ).)
Let L be the collection of left cosets of P1 , and let P2 act on L by

∀y ∈ P2 , y(gP1 ) = (yg)P1 .

The fixed points: By (1.6b), |LP2 | ≡ |L| (mod p). Since P1 ∈ Syl(G), |L| = |G : P1 | is not
divisible by p. Hence |LP2 | > 0. Let zP1 ∈ LP2 . Then

∀y ∈ P2 , y(zP1 ) = zP1 =⇒ z −1 P2 z ≤ P1 =⇒ z −1 P2 z = P1 ,

where the last equality follows by counting. Finally, set g = z −1 .

(2.5) The Third Sylow Theorem (Thm 5.10) If |G| = pa m < ∞ (where p is a prime)
and if np = |Sylp (G)|, then

both np ≡ 1 (modp) and np |pa m.

Proof: Action 1: Fix Q ∈ Sylp (G). Let Q act on Sylp (G) by conjugation:

∀x ∈ Q and ∀P ∈ Sylp (G), x(P ) = xP x−1 .

The fixed points: By (1.6b), |(Sylp (G))Q | ≡ |Sylp (G)| (mod p).

(Sylp (G))Q = {P ∈ Sylp (G)|xP x−1 = P ∀x ∈ Q} = {P ∈ Sylp (G)|Q ≤ NG (P )},

and so P, Q ∈ Sylp (N (P )). Apply 2nd Sylow Theorem to NG (P ), there is a g ∈ NG (P ) such

that Q = gP g −1 = P , (the last equality follows from g ∈ NG (P ). Thus (Sylp (G))Q = {Q}.
By (1.6b), np = |Sylp (G)| ≡ |(Sylp (G))Q | = 1 (mod p).
Action 2: Let G act on Sylp (G) by conjugation. By (2.4), the action is transitive,
and so GP = Sylp (G) is the only orbit. By (1.4), np = |Sylp (G)| = |G : GP | and so by
Lagrange’s theorem, np divides |G|.

(2.6) Sylow’s Theorem (A combined statement) Let G be a group of order pa m, where p

is a prime not dividing m.
(i) Sylp (G) 6= ∅, and if H is a p-subgroup of G such that [G : H] ≡ 0 (mod p), then
there exists a p subgroup H 0 of G such that [H 0 : H] = p.
(ii) If P is a Sylow p-subgroup of G and Q is a p-subgroup of G, then there exists g ∈ G
such that Q ≤ gP g −1 . In particular, any two Sylow p-subgroups of G are conjugate in G.
(iii) The number of Sylow p-subgroups np ≡ 1 (mod p) and np |m.

7
 3. Theorems on Sylow p-groups

In this section, we make the following assumptions: |G| = pa m for some prime p, where
integers a > 0 and m satisfies (m, p) = 1.

(3.1) Suppose that P ∈ Sylp (G), and N ≤ G such that NG (P ) ≤ N . Then NG (N ) = N .

Proof: It suffices to show that NG (N ) ⊆ N . ∀x ∈ NG (N ), Q = xP x−1 ∈ Sylp (G).
Since x ∈ NG (N ) and P ≤ N , then Q = xP x−1 ≤ xN x−1 = N , and so P, Q ∈ Sylp (N ).
By 2nd Sylow Theorem, ∃y ∈ N , such that P = yQy −1 = (yx)P (yx)−1 . It follows that
yx ∈ NG (P ) ⊆ N . But y ∈ N , and so x ∈ N .

(3.2) Suppose that H = H1 ∩ H2 is maximal in the set of {Hi ∩ Hj : Hi , Hj are distinct

members in Sylp (G)}. Let N = NG (H). Then
(i) If K ∈ Sylp (N ), then there exists exactly one P ∈ Sylp (G) such that K = P ∩ N .
(ii) If P ∈ Sylp (G) such that H ≤ P , then P ∩ N ∈ Sylp (N ).
(iii) If K1 , K2 ∈ Sylp (N ) are two distinct members, then K1 ∩ K2 = H.
(iv) If K ∈ Sylp (N ), then NN (K) = N ∩ NG (P ) for some P ∈ Sylp (G) such that H ≤ P .
(v) [N : H] > 1, and every Sylow p-subgroup of N contains H properly.
(vi) |Sylp (N )| > 1.

Proof: Since H 6= H1 , then N 6= NH1 (H) ⊆ H1 ∩ N . Therefore, there exists a K ∈

Sylp (N ) such that H 6= H1 ∩ N ≤ K. By 1st Sylow Theorem, there exists a P ∈ Sylp (G)
such that K ≤ P . Since
N 6= H1 ∩ N ≤ P ∩ H1 ,

We must have H1 = P , and so K = P ∩ N = H1 ∩ N (this proves (i) and (ii)).

Let K1 , K2 ∈ Sylp (N ) be two distinct members, and let K ∈ Sylp (N ) be a member

such that H ≤ K. By 2nd Sylow Theorem, there exists z ∈ N such that K1 = zKz −1 .
Since z ∈ N = NG (H) and since H ≤ K, H = zHz −1 ≤ zKz −1 = K1 . Similarly, H ≤ K2 .
Therefore, H ⊆ K1 ∩K2 . By (iv) (just shown above), there exist P1 , P2 ∈ Sylp (G) such that
Ki = Pi ∩ N . Therefore, H ⊆ P1 ∩ P2 . By the maximality of H, H = P1 ∩ P2 = K1 ∩ K2 .
This proves (iii).

Let K = H1 ∩N . Then by checking the definition for normalizer, we have K
NG (H1 )∩

8
N . If for any x ∈ N , xKx−1 = K, then K ⊆ xH1 x−1 , ∀x ∈ N , and so by (iv), H1 = xH1 x−1 .
It follows that x ∈ NG (H1 ) ∩ N = K. This would imply that

H1 ∩ N = K = NG (H1 ) ∩ N = NN (K).

This proves (iv) with P = H1 .

Note that H1 ∈ Sylp (G) and that H ⊆ H1 but H 6= H1 . Thus [N : H] ≡ [G : H] ≡ 0

mod p) (See (2.2) in the preceding section), and so [N : H] > 1. This proves the first half
of (v).

Recall that H = H1 ∩H2 . Let K1 = H1 ∩N and K2 = H2 ∩N . By (ii), K1 , K2 ∈ Sylp (N ).

By (i), K1 6= K2 . Therefore, |Sylp (N )| > 1. This proves (vi).

Let K ∈ Sylp (N ). By (vi), we can find another K 0 ∈ Sylp (N ) − {K}. By (iii),

H = K ∩ K 0 . This proves the later half of (v).

9
 4. Nilpotent and Solvable Groups

(4.1) (The Central Series) Let G be a group, and let C(G) denote the center of G; and
define

C0 (G) = {e}
C1 (G) = C(G)
C2 (G) = inverse image of C(G/C1 (G)) under the projection G 7→ G/C1 (G)
··· = ···
Ci (G) = inverse image of C(G/Ci−1 (G)) under the projection G 7→ G/Ci−1 (G)
··· = ···

Then, each Ci (G)  G, and the series

{e} < C1 (G) < C2 (G) · · ·

is called the ascending central series of G. A group G is nilpotent if for some integer
n, Gn = G.

(4.1a) If Gn = G, then Gn+1 = G; and for each i, C(G/Ci (G)) = Ci+1 (G)/Ci (G).
(4.1b) Any abelian group is nilpotent.
(4.1c) (Thm 7.2) Every finite p-group is nilpotent. (By the class equation).
(4.1d) (Thm 7.3) The direct product of a finite number of nilpotent groups is nilpotent.

Proof: It suffices to show the case when the factor number is 2.

(i) If H1  H and K1  K, then H/H1 × K/K1 ∼ = (H × K)/(H1 × K1 ).
(ii) Ci (H) × Ci (K) = Ci (H × K)

As (i) is a special case of (1.8) in Products of Groups, we only need to prove (ii). Note first
that
C1 (H × K) = C(H × K) = C(H) × C(K) = C1 (H) × C1 (K).

In particular, for any i,

C(H/Ci (H) × K/Ci (K)) = [C(H/Ci (H)) × C(K/Ci (K)).

10
 Fixed an i ≥ 1. Assume inductively that

Ci (H × K) = Ci (H) × Ci (K).

Then Ci (H)
H and Ci (K)
K. Let πH : H 7→ H/Ci (H) and πK : K 7→ K/Ci (K) be
the canonical projections, and π = (πH , πK ) : H/Ci (H) × K/Ci (K) be the homomorphism
defined componentwise. By (i), there is an isomorphism

ψ : H/Ci (H) × K/Ci (K) 7→ (H × K)/(Ci (H) × Ci (K)) = (H × K)/Ci (H × K).

Let φ = ψ · π : H × K 7→ (H × K)/Ci (H × K). Then

Ci+1 (H × K) = φ−1 [C((H × K)/Ci (H × K))] = π −1 ψ −1 [C((H × K)/Ci (H × K))]

= π −1 [C(H/Ci (H) × K/Ci (K))]
= π −1 [C(H/Ci (H)) × C(K/Ci (K))]
−1 −1
= πH [C(H/Ci (H))] × πK [C(K/Ci (K))]
= Ci+1 (H) × Ci+1 (K).

Thus by induction, (ii) always holds.

Suppose that both H and K are nilpotent. Then

{eH } < C1 (H) < C2 (H) < · · · < Cm (H) = H,

and
{eK } < C1 (K) < C2 (K) < · · · < Cn (K) = K.

We may assume that m ≥ n. Then Cn+1 (K) = · · · Cm (K) = K and so,

{eH } × {eK } < C1 (H) × C1 (K) < · · · < Cm (H) × Cm (K).

By (ii), (only the last equality is needed),

{eH×K } < C1 (H × K) < · · · < Cm (H × K) = H × K.

(4.2) (Lemma 7.4) Let G be nilpotent, and let H < G be proper. Then H < NG (H) is also
proper.

11
Proof: G is nilpotent =⇒ Gm = G. Let n be the largest such that Gn < H. Since
H 6= G, n < m. Pick a ∈ Cn+1 (H) − H. Then a ∈ NG (H) − H.

(4.3) If |G| < ∞, and if P ∈ Sylp (P ), then NG (NG (P )) = NG (P ).

Proof: This is a special case of (3.1). A direct proof is as follows. P  NG (P ). Since

∀x ∈ NG (NG (P )) =⇒ x(NG (P ))x−1 = NG (P ), xP x−1 = P =⇒ x ∈ NG (P ).

(4.4) (Prop. 7.5) A finite group is nilpotent if and only if G is the direct product of
its Sylow subgroups.

Proof: (Sufficiency) If G is the direct product of its Sylow subgroups, then by (4.1c)
and (4.1d), G is nilpotent.

(Necessity) Let G be nilpotent. If G itself is a p-group, then done. Assume that G is

not a p-group.

Step 1 Every Sylow p-subgroup is normal in G.

Let P ∈ Sylp (G). Then P 6= G. By (4.2), P 6= NG (P ). If NG (H) 6= G, then by
(4.2) (with H replaced by NG (H)), NG (P ) 6= NG (NG (H)). On the other hand, by (4.3),
NG (P ) = NG (NG (H)). This implies NG (H) = G and so Sylp (G) = {P }.

Step 2 If Pi is a Sylow pi -subgroup, and Pj is a Sylow pj -subgroup, where pi and pj are

two distinct primes, then Pi ∩ Pj = {e} and Pi Pj = Pj Pi .
Consider the order of an element in the intersection to see {e} = Pi ∩ Pj . For x ∈ Pi , y ∈ Pj ,
xyx−1 y −1 ∈ Pi P| , as Pi and Pj are normal in G.

Step 3 Suppose |G| = pn1 1 pn2 2 · · · pnmm , where p1 , p2 , · · · , pm are the distinct primes dividing
|G|, and P1 , P2 , · · · , Pm are the corresponding Sylow pi -subgroups, then G = P1 P2 · · · Pm .
By Lagrange and Step 2, every element in P1 · · · Pi−1 Pi+1 · · · Pm has order dividing

n n
pn1 1 pn2 2 · · · pi−1
i−1 i+1
pi+1 · · · pnmm ,

and so
Pi ∩ P1 · · · Pi−1 Pi+1 · · · Pm = {e}.

12
As P1 P2 · · · Pm ≤ G, and as

pn1 1 pn2 2 · · · pnmm = |P1 P2 · · · Pm | ≤ |G| = pn1 1 pn2 2 · · · pnmm ,

we must have G = P1 P2 · · · Pm .
Combine Step 1 and Step 3 to conclude that

G∼
= P1 × P2 × · · · × Pm .

(4.5) (Thm 7.8) Let G be a group and let a, b ∈ G. The element aba−1 b−1 is called a
commutator of G. The subgroup

G0 = haba−1 b−1 : a, b ∈ Gi

is called the commutator subgroup of G. Moreover,

(i) G0 is an invariant subgroup of G, and so G0  G.
(ii) G/G0 is abelian.
(iii) Suppose N  G. Then G/N is abelian if and only if G0 < N .

Proof: (i) ∀g ∈ G, and ∀a, b, ∈ G,

g(aba−1 b−1 )g −1 = g(ab)g −1 (ab)−1 a(bg)a−1 (bg)−1 bgb−1 g −1 .

(ii) ∀a, b ∈ G, ab = aba−1 b−1 ba, and so (ab)G0 = (ba)G0 .

(iii) Suppose that (ab)N = (ba)N ⇐⇒ aba−1 b−1 ∈ N .

(4.6) Let G(0) = G. For each i ≥ 1, define G(i) = (G(i−1) )0 . Then

(i) · · ·  G(i)  G(i−1) · · ·  G(1)  G(0) = G.
(ii) ∀i, G(i)  G.

Proof: (4.5) =⇒ (ii) holds when i = 1. Then argue by induction and use the corre-
spondence in the isomorphism theorems. (Another way to see this is that each of the G(i)
is an invariant subgroup.)

(4.7) A group G is solvable if G(n) = {e} for some integer n.

13
(4.8) Every nilpotent group is solvable.

Proof: G is nilpotent =⇒ Cn (G) = G.

C(G/Gn−1 (G)) = Cn (G)/Cn−1 (G) is abelian =⇒ G(1) < Cn−1 (G).
Similarly,

G(2) = (G(1) )0 < Cn−1 (G)0 < Cn−2 (G)

··· = ···
G(n) = (G(1) )0 < C1 (G)0 = C(G) = {e}

(4.9) (Thm 7.11)

(i) Every subgroup and every homomorphic image of a solvable group is solvable.
(ii) Suppose N  G. Then G is solvable if and only if both N and G/N are solvable.

Proof: (i). Suppose f : G 7→ H is an epimorphism. Then

f (G(1) ) = h{f (a)f (b)f (a)−1 f (b)−1 : a, b ∈ G}i = H (1) ,

and

f (G(i) ) = h{f (a)f (b)f (a)−1 f (b)−1 : a, b ∈ G(i−1) }i = h{f (a)f (b)f (a)−1 f (b)−1 : a, b ∈ H (i−1) }i = H (i) .

If H < G, then G(i) ∩ H = H (i) .

(ii). Let π : G :7→ G/N be the canonical map. Note that π(G(n) ) = (G/N )(n) = {eG/N }. It
follows that G(n) ≤ Kerπ = N . By (i), as a subgroup of a solvable group N , G(n) is also
solvable.

(4.10) If n ≥ 5, then Sn is not solvable.

Proof Suppose that Sn is solvable. Then by (4.9). An , as a subgroup of Sn , must be

solvable. Consider the commutator group A0n . Since An is not abelian, A0n 6= {e}. Since
(i)
An is simple, and since A0n
An , we must have A0n = An . It follows that An = An for any
i ≥ 1, and so An cannot be solvable.

14