1. a. D4 = {1, a, a2, a3, b, ba, ba2, ba3} where o(a) = 4, o(b) = 2 and aba = b. The classes are {1}, {a, a3}, {a2}, {b, ba2}, {ba, ba3}. Since normal subgroups are of orders 1, 2, 4, 8, they are {1}, {1, a2}, {1, a, a2, a3}, {1, b, a2, ba2}, {1, a2, ba, ba3} and D4.
3. If
am =
g−1ag with
g G, then
By induction we get
for all
k ≥ 0. Since
G is finite, let
gk = 1,
k ≥ 1. Then
whence 1 −
mk =
qn. This gives 1 =
mk +
qn, so gcd (
m,
n) = 1.
5. Let
H =
class a1 ∪ . . . ∪
classan. Given
g G and
h H, let
h classai. Then
g−1hg classai ⊆
H, so
g−1hg H. Then
g−1Hg ⊆
H, as required.
7. Let
K =
g−1Hg, so
H =
gKg−1. We claim
N(
K) =
g−1N(
H)
g. Let
a N(
K) so
a−1Ka =
K. To show
a g−1N(
H)
g it suffices to show
gag−1 N(
H). But
as required. Hence N(K) ⊆ g−1N(H)g. Similarly N(H) ⊆ gN(K)g−1 so g−1N(H)g ⊆ N(K).
9. If |class a| = 2 then |G : N(a)| = 2 by Theorem 2. Hence N(a) is normal in G. Since N(a) ≠ G we are done if N(a) ≠ {1}. But N(a) = {1} implies |G| = |G : N(a)| = 2 so G is abelian and every conjugacy class is a singleton, a contradiction.
11. We have
H ⊆
N(
H) ⊆
G and
H has finite index
m in
G. So
by Exercise 31 §2.6. Thus |G : N(H)| is finite, and H has |G : N(H)| conjugates by Theorem 2.
13. Write
X = {
g−1Hg g G} and
Y = {
N(
H)
g g G}. Define
ϕ :
X →
Y by
ϕ(
g−1Hg) =
N(
H)
g. Then:
Thus ϕ is well defined and one-to-one; it is clearly onto.
15. Let
α =
γ1γ2. . .
γr in
Sn where the
γi are disjoint cycles. Given
σ Sn,
and this is the factorization of σ−1ασ into disjoint cycles by Lemma 3 §2.8. Since γi and σ−1γiσ have the same length, σ−1ασ has the same cycle structure as σ. Conversely let α and β have the same cycle structure, say
where the
γi (the
δi) are disjoint cycles, and
γi and
δi have equal length for each
i. If
γi = (
ki1,
ki2, . . .,
kis) and
δi = (
li1li2. . .
lis) for all
i define
σ Sn by
Then σ is a permutation and σ−1γiσ = δi by Lemma 3 §2.8. Hence σ−1ασ = β, that is α and β are conjugate.
17. a. S4 consists of
ε, three permutations of type (
a b)(
c d), eight 3-cycles, six 2-cycles and six 4-cycles. These are the conjugacy classes by Exercise 15. So, by Theorem 1, each normal subgroup
H S4 is a union of these classes. Since |
H| divides 24 = |
S4|, and since
ε H, the only possibilities are
H =
ε,
H =
K,
H =
A4 and
H =
S4.
19. It suffices to show that G is abelian (then there are |G| conjugacy classes). So assume G is not abelian. Then there must be a nonsingleton conjugacy class, say class a. Thus there is only one singleton class, {1}. This means |G| = 1 + |class a| = 1 + |G : N(a)|. Write m = |G : N(a)| so that |G| = md, d ≥ 2. But then md = 1 + m gives 1 = m(d − 1) so m = 1, that is |class a| = 1, contrary to assumption.
21. If
G/
K and
K are
p-groups, let
g G. Then
in
G/
K for some
n, so
. Since
K is itself a
p-group,
for some
m; that is
. Thus
o(
g) is a power of
p, so
G is a
p-group. Conversely, subgroups of
p-groups are clearly
p-groups, as are images because if
then
23. We have |
G| =
pn for some
n ≥ 1 so
for all
g G. It follows that
in
for all [
gi) in
, so
is a
p-group. If 1 ≠
g and 1 ≠
h then [
g, 1, 1, . . .), [1,
g, 1, 1, . . .), . . . are all distinct in
. So
.
25. By Theorem 1,
H is a union of
G-conjugacy classes; suppose there are
m of these which are singletons. The remaining classes in
H have order a multiple of
p by Theorem 3 (since
G is a
p-group) and
p divides |
H| (since
H ≠ {1}). Hence
p|
m. Since
m > 0 (because {1} ⊆
H) this means
m > 1 and there exists
a ≠ 1, such that class
a = {
a} ⊆
H. But class
a = {
a} means
a Z(
G), so
a H ∩
Z(
G).
26. a. Let |
G| =
p3,
G nonabelian. Write
Z =
Z(
G). Then
Z ≠ {1} by Theorem 6 and
Z ≠
G because
Gisnonabelian so |
Z| =
p or
p2. If |
Z| =
p2 then |
G/
Z| =
p so
G/
Z is cyclic and
G is abelian by Theorem 2 §2.9, contrary to assumption. So |
Z| =
p. Then |
G/
Z| =
p2 so
G/
Z is abelian by Theorem 8. Hence
G′ ⊆
Z so
G′ = {1} or
G′ =
Z. But
G′ = {1} implies
G is abelian. So
G′ =
Z. Finally, if
K G, |
K| =
p, then
G/
K is abelian so
G′ ⊆
K. Thus |
K| =
p = |
Z| = |
G′| implies
K =
G′.
27. We proceed by induction on
n where |
G| =
pn. If
n = 1 it is clear. In general |
H| =
pm. If
H = {1} it is clear. If
H ≠ {1} let
K =
H ∩
Z(
G). Then
K G and
K ≠ {1} by Exercise 25. Let |
K| =
pb. We have
H/
K G/
K so, by induction, let
where
and
for all
i. Thus
Gi G and |
Gi/
Gi+1| =
p for all
i. Similarly
where
Gi G for each
i and |
Gi/
Gi+1| =
p. These
Gi do it.
29. Since
C G, let
Z[
G/
C] =
K/
C. Since |
G/
C| > 1, Theorem 6 shows
C ⊂
K. But
K G so
K H by Exercise 26 §2.8. If
k K then
kC is in the center of
G/
C, so
h−1k−1hk C H. Hence
k−1Hk ⊆
H, and similarly
kHk−1 ⊆
H. Thus
k N(
H) and we have shown
K ⊆
N(
H).
31. The associativity is verified as follows:
These are equal. The unity is (0, 0, 0) and (
x,
y,
z)
−1 = (−
x, −
y, −
z −
xy) for all (
x,
y,
z). The group is nonabelian because (1, 1, 0) · (1, 0, 0) = (2, 1, − 1) while (1, 0, 0) · (1, 1, 0) = (2, 1, 0). Finally
for all
k ≥ 2 by induction on
k. If
k =
p,
. Since
and
p is odd, this is (0, 0, 0) so
o((
x,
y,
z)) =
p.
33.
a. Given
a G, |class
a| = |
G :
N(
a)| by Theorem 2. But
so the fact that |G : Z(G)| is finite implies |G : N(a)| is finite.
c. If
G =
X and
a G, write
,
xi X,
. We claim that
N(
xi) ⊆
N(
a). For if
g N(
xi) for all
i then
gxi =
xig for all
i, so
ga =
ag; that is
g N(
a). But each
N(
xi) has finite index by Theorem 2, so
N(
xi) has finite index by Poincaré's Theorem [Exercise 33 §2.6], so
N(
a) has finite index. Thus |class
a| is finite.
e. If
a,
b G∗ we have
N(
a) ∩
N(
b) ⊆
N(
ab). Since
N(
a) ∩
N(
b) has finite index (Poincaré's theorem), so does
N(
ab) by the following Lemma. Thus
ab G∗.
i. Lemma. If
K ⊆
H ⊆
G are groups and |
G :
K| is finite, then |
G :
H| is finite.
Proof. We have
H = ∪ {
hK h H}, a finite union because {
gK g G} is finite. If
H =
h1K ∪
∪
hnK, a disjoint union, then
Hence there are at most a finite number of cosets gH.
Next,
N(
a−1) = {
g ga−1 =
a−1g} = {
g ga =
ag} =
N(
a), so
a−1 G∗ too. Clearly 1
G∗, so
G∗ is a subgroup. Finally,
G∗ is itself an FC-group. Indeed, if
a G∗
Now let
σ :
G →
G be an automorphism. If
a G∗ then
Since class
a is finite, this shows class
σ(
a) is finite, that is
σ(
a)
G∗.