1. First
ε gal(
E :
F) because
ε(
a) =
a for all
a F. If
σ,
τ gal(
E :
F) then
σ(
a) =
a for all
a F, so
a =
σ−1(
a) for all
a; hence (since
σ−1 is an automorphism)
σ−1 gal(
E :
F). Finally
στ(
a) =
σ[
τ(
a)] =
σ(
a) =
a, so
στ gal(
E :
F).
3. Let
σ(
ui) =
τ(
ui) for all
i where
σ,
τ gal(
E :
F). If
, write
,
ai F. Then
As
was arbitrary, this shows that
σ =
τ.
5. Let
σ :
E →
E be an automorphism. Then if
,
σ(
n) =
n ·
σ(1) =
n. Thus
for all
131 Hence
σ gal
7. Put
u =
e2πi/6. Then
u satisfies
x6 − 1 = (
x2 − 1)(
x2 +
x + 1)(
x2 −
x + 1), and the only roots in
are the sixth roots of unity 1,
u,
u2,
u3,
u4,
u5. But
u3 = − 1 so
u satisfies
x3 + 1 = (
x + 1)(
x2 −
x + 1). Hence
u is a root of
m =
x2 −
x + 1 . Since
m is irreducible over
, it is the minimal polynomial of
u. The other root of
m is
u5 (because (
u5)
3 = (
u3)
5 = (− 1)
5 = − 1), so Theorem 1 gives
where
σ(
u) =
u5. Note that
σ2 =
ε because
σ2(
u) =
σ[
σ(
u)] =
σ(
u5) = [
σ(
u)]
5 =
u25 =
u. Hence
9. The minimal polynomial of
i is
x2 + 1, and that of
is
x2 − 3. By Lemma 2, there exists a
-isomorphism
with
σ0(
i) = −
i. This extends to an automorphism
σ of
satisfying
. Thus
,
σ(
i) = −
i and
. Similarly, there exists
with
τ(
i) =
i and
. Note that
σ(
i)
{
i, −
i} and
so
by Theorem 2. Now observe that
στ(
i) = −
i =
τσ(
i) and
Hence
τσ =
στ by Theorem 2. Since
o(
σ) = 2 =
o(
τ), it follows that {
ε,
σ,
τ,
στ} ≅
C2 ×
C2. But
and
so
10. a. Write
. Then
u satisfies
x4 − 2, irreducible by the Eisenstein criterion. The roots of
x4 − 2 in
are
u, −
u,
iu and −
iu, and the only ones in
are
u and −
u. Thus Theorem 1 gives
so
where
σ(
u) = −
u.
11. Choose
u E,
u ∉
F. Thus [
F(
u) :
F] ≠ 1 so, since [
E :
F] = 2 is prime,
E =
F(
u) by Theorem 5 §6.2 (the multiplication theorem). If
m is the minimal polynomial of
u over
F, then °
m = 2. Thus since one root
u lies in
E, so does the other (their sum is the negative of the coefficient of
x in
m). Now the result is clear by Theorem 1.
13. Write
. Now
x4 − 2 is irreducible (Eisenstein) so
x4 − 2 and
x2 + 1 are the minimal polynomials of
u and
i. The roots are {
u, −
u,
iu, −
iu} and {
i, −
i}, and all are in
E so |
G| ≤ 8 by Theorem 1 where
. By Lemma 1 there exists
satisfying
σ0(
u) =
iu. Extend it to an automorphism
σ of
where
σ(
i) =
i. Thus
σ(
u) =
iu and
σ(
i) =
i, and so
σ2(
u) = −
u,
σ3(
u) = −
iu,
σ4(
u) =
u. Thus
o(
σ) = 4. Next let
have
τ0(
i) = −
i, extend
τ0 to an automorphism
τ of
where
τ(
u) =
u. Then
τ2 =
ε so
o(
τ) = 2. Finally
Thus
στσ =
τ, whence
σ,
τ ≅
D4. Thus |
σ,
τ |= 8 so, since
σ,
τ ⊆
G and |
G| = 8,
G =
σ,
τ ≅
D4.
15. a. If
and
, then
is 0 or
. So
a = 1 is none of these, whence
by the proof of Theorem 6.
16. a. If
and
, then
is 0 or
. So
a = 1 is neither of these, whence
by the proof of Theorem 6.
17. If
then
for some
. If
σ ≠
ε in gal(
then
and
. Now let
. If
u ≠
σ(
u) let
u <
σ(
u). If
u <
a <
σ(
u),
, then
σ(
u) <
σ(
a) =
a, a contradiction. A similar argument diminates
σ(
u) <
u.
19. We proceed by induction on
n. If
n = 1 then
E =
F(
u1) = {
f(
u1)
f F[
x]} . Hence
σ(
f(
u1)) =
f(
σ(
u1)) =
f(
τ(
u1)) =
τ(
f(
u1)) for all
f, as required. In general, write
K =
F(
u1,
u2, . . .,
un−1) so that
E =
K(
un) . By induction,
σ =
τ on
K, so
σ,
τ gal(
K :
F) . Since
σ(
un) =
τ(
un) the result follows from the case
n = 1.
21. Let
σ gal (
F(
t) :
F). If
,
f,
g relatively prime in
F[
t], then
, so if
,
σ(
λ(
t)) =
λ(
σ(
t)). Thus
σ is determined completely by
f and
g. Clearly
f ≠ 0,
g ≠ 0. Now suppose
. Then
. Suppose
Then
Suppose
n >
m. Then
Since
anfn is the only term not involving
g, it follows that
g|
anfn. Hence
g|
an since
f,
g are relatively prime, so °
g = 0. Similarly
n <
m implies that
g|
bmtfm so, again, °
g ≤ 1 . Thus either °
g ≤ 1 or
n =
m. But if
n =
m we have
t(
b0gn +
b1fgn−1 +
+
bnfn) =
a0gn +
+
anfn. This yields
g|(
an −
bnt)
fn and °
g ≤ 1 in all cases. Now observe that
so
. Hence
and the same type of argument implies °
f ≤ 1. Hence we have proved that
so that
σ =
σM where
. Now observe that
σM[
σN(
λ(
x))] =
σNMλ(
x) holds for all
λ(
t)
F(
t). In particular if
σ gal(
F(
t) :
F) and
σ =
σM, and if
σ−1 =
σN, we have
where
. It follows that
c′′x2 +
d′′x =
a′′x +
b′′ so
c′′ =
b′′ = 0,
a′′ =
d′′ ≠ 0 so
and so
. Thus the map
M σM from
GL2(
F) →
gal (
F(
t) :
F) is an onto group homomorphism. Moreover the above shows that the kernel is
Z.
22. a. (3) ⇒ (1). Let
E ⊇
F be a field and suppose
f has a repeated root
u in
E. Then
f(
u) = 0 =
f′(
u) by Theorem 3 §6.4. But (3) implies 1 =
fg +
f′
h in
F[
x] with
g,
h F[
x], and this is valid in
E[
x]. But then, taking
x =
u gives 1 = 0 in
E, a contradiction.
23. Here f′ = nxn−1 − 1. Write d = gcd (f, f′); we must show d = 1 by the preceding exercise. Let E ⊇ F be a splitting field for f. If d ≠ 1 then °d > 1 so d has a root u in E. Thus f(u) = 0 = f′(u) so un = u and nun−1 = 1. Hence (n − 1)u = 0, a contradiction if char F = 0 . But if char F = p it implies p|(n − 1), contrary to hypothesis.
25. If
E ⊇
F and
f F[
x] is separable over
F, let
q be any irreducible factor of
f in
E[
x]. We must show that
q has no repeated root in
E. Now
f =
p1p2 pr in
F[
x] where the
pi are irreducible and (by hypothesis) separable. But
f =
p1p2 pr in
E[
x] so
q|
pi for some
i. Hence any repeated root of
q in some splitting field of
f would be a repeated root of
pi, contrary to assumption. So
q is separable.
27. Suppose
f =
xp −
a is not a power of a linear polynomial in
F[
x]; we must show it is irreducible. If
u is a root of
f in an extension
E ⊇
F, then
up =
a so
f =
xp −
up = (
x −
u)
p in
E[
x] because the characteristic is
p. Then
u ∉
F and
F(
u) is a splitting field of
f over
F. Let
q be an irreducible factor of
f in
F[
x]. Then
q|(
x −
u)
p in
E[
x] so
q = (
x −
u)
t. Then
t > 1 because
u ∉
F so
q′ = 0 by Lemma 5. Hence
q =
g(
xp) by Theorem 4. Thus, the factorization of
f into irreducibles in
F[
x] takes the form
Since def f = p, it follows that f = gi(xp) = qi for some i, so f is irreducible.
28.
a. (3) ⇒ (1). If
E ⊇
F is algebraic and
u E, the minimal polynomial of
u is separable by (3), and so
E ⊇
F is a separable extension.
c. Let E ⊇ F be algebraic, F perfect. If K ⊇ E is algebraic, then K ⊇ F is algebraic by Corollary 1, Theorem 6 §6.2, hence separable by hypothesis.
29.
a. If
F is perfect, and
a F, let
E be the splitting field of
f =
xp −
a. If
u E is a root of
f then
up =
a, so
f =
xp −
a =
xp −
up = (
x −
u)
p. Let
q be an irreducible factor of
f in
F[
x]. Then
q = (
x −
u)
m. But
E ⊇
F is finite, hence separable by hypothesis. Thus
q has distinct roots, whence
m = 1 and
q =
x −
u. But then
u F and
up =
a, as required.
Conversely, let
q be irreducible in
F[
x]. If
q is not separable then
q′ = 0 so
q =
f(
xp) for some
f F[
x] by Theorem 4, say
. By hypothesis, let
,
bi F. Then
This contradicts the irreducibility of q.
30.
a. Let
q be the minimal polynomial of
u over
F. If
K =
F(
up) let
m K[
x] be the minimal polynomial of
u over
K. Then
q K[
x] and
q(
u) = 0, so
m|
q. But
q has distinct roots by hypothesis, so
m has distinct roots. On the other hand,
xp −
up K[
x] and
xp −
up = (
x −
u)
p in
E[
x]. Hence
m|(
x −
u)
p so
m = (
x −
u)
r. Since
m has distinct roots,
r = 1 and so
u K.
c. Extend
to a basis
of
E over
F. Suppose
, say
. Write
, so
, whence
. Thus
spans
E and so is
F -independent because dim
E =
n (Theorem 7 §6.1). But then
is
F-independent.
31. If
E ⊇
F is separable, the result is Exercise 26. If
E ⊇
K and
K ⊇
F are separable, we may assume char
F =
p by the Corollary to Theorem 4. By the preceding exercise,
E =
K(
Ep) and
K =
F(
Kp). If
u E then
where
and
ui E. But
where
aij F and
for all
i,
j. Thus
. Thus
E ⊇
F is separable by the preceding exercise.
32.
a. Let
p and
q be the minimal polynomials of
u over
F and
K respectively. Then
p K[
x] and
p(
u) = 0, so
q|
p. Since
p has distinct roots in some splitting field
L ⊇
K,
q is separable over
K.
c. Let
u,
. Then
F(
u) ⊇
F is separable and, since
is separable over
F(
u) by (a),
is also separable. Hence
is separable by the preceding exercise. Since
and
uv are in
, they are separable over
F, so
S is a field. Clearly
S ⊇
F since the minimal polynomial of
a F is
x −
a, and
S ⊇
F is separable by the definition of
S. If
E ⊇
K ⊇
F and
K ⊇
F is separable, then each
u K is separable over
F; that is
u K. Hence
K ⊆
S.