1. Define
by
This is a group homomorphism by direct calculation. We have
if and only if
a = 1 =
c, so ker
α =
K.
3. Since
,
H G and
G/
H = {
H,
G
H} is a group with
H as unity. Thus
σ :
G/
H → {1, − 1} is an isomorphism if
σ(
H) = 1 and
σ(
G −
H) = − 1. Let
ϕ :
G →
G/
H be the coset map
ϕ(
g) =
Hg, and let
α =
σϕ :
G → {1, − 1}. Then
α is a homomorphism and: If
g H, then
α(
g) =
σϕ(
g) =
σ(
H) = 1; if
g ∉
H, then
α(
g) =
σϕ(
g) =
σ(
G −
H) = − 1.
4.
a. We have 1
α−1(
X) because
α(1) = 1
X. If
g,
h α−1(
X), then
α(
g)
X and
α(
h)
X, so
α(
g−1) = [
α(
g)]
−1 X and
α(
gh) =
α(
g) ·
α(
h)
X. Thus
g−1 α−1(
X) and
gh α−1(
X), so
α−1(
X) is a subgroup of
G. If
X α(
G), let
g α−1(
X),
a G. Then
Thus
a−1ga α−1(
X) for all
a G; that is
α−1(
X)
G.
c. Since
X ∩
Y ⊆
X and
X ∩
Y ⊆
Y,
α−1(
X ∩
Y) ⊆
α−1(
X) ∩
α−1(
Y) by (b). If
g α−1(
X) ∩
α−1(
Y), then
α(
g)
X and
α(
g)
Y, so
α(
g)
X ∩
Y. Hence
g α−1(
X ∩
Y).
5.
a. If
gd = 1, then since
d m, 1 =
gm =
ρmg, so
g ker
ρm. Conversely, if
ρmg = 1, then
gm = 1. We have also that
gn = 1 (since
. Since
d =
xm +
yn,
, this gives
gd = (
gm)
x(
gn)
y = 1
x1
y = 1.
c. Let
where
o(
a) =
n. If
σ :
G →
G is an automorphism then
o(
σ(
a)) =
o(
a) =
n. Hence Theorem 8, §2.4 gives
σ(
a) =
am where
gcd(
m,
n) = 1 . Now let
g G, say
g =
ak. Then
It follows that σ = ρm.
7. Let
α :
G →
G and let ker
α =
X and
α(
G) =
Y where
X and
Y are finite sets. Since
Y ⊆
α(
G), let
Z ⊆
G be a finite set such that, if
y Y,
y =
α(
z) for some
z Z. If
g G, then
α(
g)
Y , so
If we write
, then
gh−1 ker
α, so
xi X. Hence
Hence
G =
X ∪
Z , so since
X ∪
Z a finite set,
G is finitely generated.
8.
a. Let
C6 =
g ,
o(
g) = 6, and write
K4 = {1,
a,
b,
ab}, where
a2 =
b2 = 1 and
ab =
ba. If
α :
C6 →
K4 is a homomorphism, then
α is determined by the choice of
α(
g) in
K4. If
α(
g) = 1 then
α is trivial (
α(
x) = 1 for all
x G) . If
α(
g) =
a (say), then
α(
gk) =
ak. If we
defineα by
α(
gk) =
ak, it is well defined because
Thus α is well defined, it is clearly a homomorphism. In the same way, there is a homomorphism carrying g to 1, a, b and ab; so there are four in all.
c. Let
D3 = {1,
a,
a2,
b,
ba,
ba2} where
o(
a) = 3,
o(
b) = 2,
aba =
b; and let
C4 =
c ,
o(
c) = 4. If
α :
D3 →
C4, then
o(
α(
a)) divides
o(
a) = 3, so
α(
a) = 1 because
C4 has no element of order 3. Similarly
o(
α(
b)) divides
o(
b) = 2, so
α(
b) = 1,
c2. If
α(
b) = 1, then
α is trivial. If
α(
b) =
c2, then
α(
bkam) =
c2k1
m =
c2k. If we now
defineα by this formula, it is possible (but tedious) to check it is well defined and a homomorphism. There is another way. Write
H =
a in
D3. Then
H D3 being of index 2, so
D3/
H = {
H,
bH}. Then there is an isomorphism {
H,
bH} →
σ{1,
c2} ⊆
C4 where
σ(
H) = 1,
σ(
bH) =
c2. If
ϕ :
D3 →
D3/
H is the coset map, we get
Then: σϕ(bkam) = σ(bkamH) = σ(bkH) = σ[(bH)k] = [σ(bH)]k = c2k. Hence the map we want is α = σϕ. This is the only non-trivial homomorphism.
9. No. If α : S4 → A4 has ker α = K = {ε, (1 2)(3 4), (1 3)(2 4), (1 4)(2 3)}, then A4/K ≅ α(A4). Now S4/K ≅ D3 [if a = K(1 2 3) and b = K(1 2), then o(g) = 3, o(b) = 2 and aba = b]. Hence α(S4) would be a subgroup of A4 which is isomorphic to D3. But there is no such subgroup: If o(σ) = 3 and o(τ) = 2 and στσ = τ, then σ is a 3-cycle, say σ = (1 2 3), and τ is one of (1 2)(3 4), (1 3)(2 4) or (1 4)(2 3). It is easy to check that στσ ≠ τ in each case.
10.
a.No. If
α :
S3 →
K4 is onto, then
K4 ≅
S3/ker
α, so
would divide
, a contradiction.
c. Yes. If
S3 = {
ε,
σ,
σ2,
τ,
τσ,
τσ2}, let
K =
σ . Then
, so
S3/
K = {
K,
τ(
K)}. If
C2 =
c ,
o(
c) = 2, then
S3 →
S3/
K →
C3 is onto, where
σ(
K) = 1,
σ[
τ(
K)] =
c. Thus
α =
σϕ :
S3 →
C2 is given by
11.
a. We have
θ(
gh) = (
gh,
gh) = (
g,
g)(
h,
h) =
θ(
g) ·
θ(
h) for all
g,
h G, so
θ is a homomorphism. If
θ(
g) = (1, 1), then
g = 1, so
θ is one-to-one.
b. (1) ⇒ (3). If
G is abelian, define
ϕ :
G ×
G →
G by
ϕ(
g,
h) =
gh−1. Then
Thus
ϕ is a homomorphism, and
ϕθ(
g) =
ϕ(
g,
g) =
gg−1 = 1 for all
g. Hence
θ(
G) ⊆ ker
ϕ. If (
g,
h)
ker
ϕ, then
gh−1 =
ϕ(
g,
h) = 1, so
h =
g. Thus (
g,
h)
θ(
G), so
a(
G) = ker
ϕ.
13. Let
G be simple. If
α :
G →
G1 is a nontrivial homomorphism, then ker
α ≠
G. Since ker
α G, ker
α = {1} by simplicity, that is
α is one-to-one. Hence
G ≅
α(
G) ⊆
G1.
Conversely, if G1 has a subgroup G0 and σ : G → G0 is an isomorphism, then σ : G → G1 is a (one-to-one) homomorphism, which is nontrivial because G0 ≠ {1}, being simple.
15. We have
G/
Z(
G) ≅
inn G, and
inn G is cyclic by hypothesis (a subgroup of
aut G). Hence
G is abelian by Theorem 2 §2.9. Thus
τ :
G →
G is an automorphism where
τ(
g) =
g−1 for all
g G. Write
aut G =
σ . Then
σk =
τ for some
k, so
σ2k =
τ2 =
ε. Thus
o(
σ) is finite, that is
is finite. Since
o(
τ) = 2 divides
, we are done.
17. a. If
g G′,
g = [
a1,
b1][
a2,
b2]
[
an,
bn] where [
a,
b] =
a−1b−1ab is a commutator. Then
19. a.Define
by
α(
A) = det
A. Then
because det
A ≠ 0 if
A is invertible, and
α is a homomorphism because det
AB = det
A · det
B. Since ker
α =
K, we get
.
21. Define
by
for
[Note that
because
z ≠ 0.] Then
α is a homomorphism because
, and
. Thus
by the isomorphism theorem.
23. If
a ≠ 0, then
τa,b is a bijection, so
. We have
and
, and
, so
G is a subgroup. Now define
by (
τa,b) =
a. Then
α is well defined: If
then
ax +
b =
a′
x +
b′ for all
x, so
a =
a′. Now
so
α is a homomorphism. Since ker
α =
K, we have
K G and
.
25. Define
α :
G ×
G ×
G →
G ×
G by
α(
a,
b,
c) = (
ac−1,
bc−1). Then
α is a homomorphism because
G is abelian:
Now
Hence, since α is onto, we are done by the isomorphism theorem.
27. Define
β :
α(
G) →
G/
K by
β[
α(
g)] =
Kg. This is well defined because
α(
g) =
α(
g1) implies
, so
, that is
Kg =
Kg1. It is a homomorphism because
Finally
Thus α(K) α(G) and α(G)/α(K) ≅ α(G) = G/K. Note: This can also be solved by verifying α(K) α(G) directly, defining γ : G → α(G)/α(K) by γ(g) = [α(K)]α(g), and showing ker γ = K.
29. a. If
and
then
and
. Thus
G is a subgroup of
. The center consists of all
with
b′ +
ac′ +
b =
b +
a′
c +
b′ for all
a,
b,
c. Taking
a = 1,
c = 0 gives
c′ = 0; taking
a = 0,
c = 1 gives
a′ = 0. Thus
. The map
is an isomorphism
.
31. Let
and
. Define
by
. Then
so
α is a homomorphism. Now
and and .
Thus , so .
33.
a. has subgroups {0}, {0, 2} and
. The factors are isomorphic to
,
, {0} respectively, so these are the only possible images.
c. The normal subgroups of
A4 are
and
A4 (Exercise 19 §2.8). Thus the factor groups are isomorphic to
A4,
C3 (since
) and {1}.
35. If
ϕ :
G1 →
G1/
X is the coset map, we have
β =
ϕα :
G →
G1/
X. This is onto because
α and
ϕ are both onto, and since
The isomorphism theorem completes the proof.
36.
a. It is closed because
α +
β is a homomorphism:
We have
α +
β =
β +
α because
Similarly, α + (β + γ) = (α + β) + γ. The unity is θ : X → Y where θ(x) = 0 for all x. Finally, the negative of α is −α : X → Y defined by (− α)(x) = − [α(x)] for all x.
c. For convenience, write
and
. Let
d = gcd (
m,
n), and write
e =
nd. Given
, define
by
This is well defined:
This means
. So
αk is well defined. We have
Hence we have αk+l = αk + αl; that is α(k + l) = α(k) + α(l).
α is onto. Let
, and write
. Then
Thus
, so
n my. This gives
nd mdy. But
nd and
md are relatively prime, so
nd y, that is
e y, say
y =
ek. But then
Thus λ = αk and α is onto.
We have ker
α = {
k αk = 0}. But
Hence
.
37.
a. The unity of
Gω is [1) = (1, 1, 1, . . .). The inverse of [
gi) is
. Finally
Thus Gω is associative.
c. F → σGω is an isomorphism when σ(f) = [f(i)) = [f(0), f(1), f(2), . . .).