This action might not be possible to undo. Are you sure you want to continue?
An Introduction to Galois Theory
Andrew Baker
[07/04/2010]
Department of Mathematics, University of Glasgow.
Email address: a.baker@maths.gla.ac.uk
URL: http://www.maths.gla.ac.uk/∼ajb
Q(
3
√
2, ζ
3
)
Q(
3
√
2)
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
Q(
3
√
2 ζ
3
)
2
.
.
.
.
.
.
.
.
.
.
.
.
3
Q(
3
√
2 ζ
2
3
)
2
3
Q(ζ
3
)
3
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Q
Gal(E/Q)
∼
= S
3
¦id, (1 2 3), (1 3 2)¦
3
2
¦id, (2 3)¦
2
3
¦id, (1 3)¦
2
3
¦id, (1 2)¦
2
3
¦id¦
The Galois Correspondence for Q(
3
√
2, ζ
3
)/Q
ii
Introduction: What is Galois Theory?
Much of early algebra centred around the search for explicit formulae for roots of polynomial
equations in one or more unknowns. The solution of linear and quadratic equations in a single
unknown was well understood in antiquity, while formulae for the roots of general real cubics and
quartics was solved by the 16th century. These solutions involved complex numbers rather than
just real numbers. By the early 19th century no general solution of a general polynomial equation
‘by radicals’ (i.e., by repeatedly taking nth roots for various n) was found despite considerable
eﬀort by many outstanding mathematicians. Eventually, the work of Abel and Galois led to a
satisfactory framework for fully understanding this problem and the realization that the general
polynomial equation of degree at least 5 could not always be solved by radicals. At a more
profound level, the algebraic structure of Galois extensions is mirrored in the subgroups of their
Galois groups, which allows the application of group theoretic ideas to the study of ﬁelds. This
Galois Correspondence is a powerful idea which can be generalized to apply to such diverse
topics as ring theory, algebraic number theory, algebraic geometry, diﬀerential equations and
algebraic topology. Because of this, Galois theory in its many manifestations is a central topic
in modern mathematics.
In this course we will focus on the following topics.
• The solution of polynomial equations over a ﬁeld, including relationships between roots,
methods of solutions and location of roots.
• The structure of ﬁnite and algebraic extensions of ﬁelds and their automorphisms.
We will study these in detail, building up a theory of algebraic extensions of ﬁelds and their
automorphism groups and applying it to solve questions about roots of polynomial equations.
The techniques we will meet can also be applied to study the following some of which may be
met by people studying more advanced courses.
• Classic topics such as squaring the circle, duplication of the cube, constructible numbers
and constructible polygons.
• Applications of Galois theoretic ideas in Number Theory, the study of diﬀerential
equations and Algebraic Geometry.
There are many good introductory books on Galois Theory, some of which are listed in the
Bibliography. In particular, [2, 3, 8] are all excellent sources and have many similarities to the
present approach to the material.
c (A. J. Baker (2009)
Contents
Introduction: What is Galois Theory? ii
Chapter 1. Integral domains, ﬁelds and polynomial rings 1
Basic notions, convention, etc 1
1.1. Recollections on integral domains and ﬁelds 1
1.2. Polynomial rings 6
1.3. Identifying irreducible polynomials 11
1.4. Finding roots of complex polynomials of small degree 15
1.5. Automorphisms of rings and ﬁelds 17
Exercises on Chapter 1 21
Chapter 2. Fields and their extensions 25
2.1. Fields and subﬁelds 25
2.2. Simple and ﬁnitely generated extensions 27
Exercises on Chapter 2 30
Chapter 3. Algebraic extensions of ﬁelds 33
3.1. Algebraic extensions 33
3.2. Splitting ﬁelds and Kronecker’s Theorem 36
3.3. Monomorphisms between extensions 39
3.4. Algebraic closures 42
3.5. Multiplicity of roots and separability 44
3.6. The Primitive Element Theorem 48
3.7. Normal extensions and splitting ﬁelds 50
Exercises on Chapter 3 50
Chapter 4. Galois extensions and the Galois Correspondence 53
4.1. Galois extensions 53
4.2. Working with Galois groups 54
4.3. Subgroups of Galois groups and their ﬁxed ﬁelds 56
4.4. Subﬁelds of Galois extensions and relative Galois groups 57
4.5. The Galois Correspondence and the Main Theorem of Galois Theory 58
4.6. Galois extensions inside the complex numbers and complex conjugation 60
4.7. Galois groups of even and odd permutations 60
4.8. Kaplansky’s Theorem 63
Exercises on Chapter 4 66
Chapter 5. Galois extensions for ﬁelds of positive characteristic 69
5.1. Finite ﬁelds 69
5.2. Galois groups of ﬁnite ﬁelds and Frobenius mappings 72
5.3. The trace and norm mappings 74
Exercises on Chapter 5 75
Chapter 6. A Galois Miscellany 77
iii
iv CONTENTS
6.1. A proof of the Fundamental Theorem of Algebra 77
6.2. Cyclotomic extensions 77
6.3. Artin’s Theorem on linear independence of characters 82
6.4. Simple radical extensions 84
6.5. Solvability and radical extensions 85
6.6. Symmetric functions 89
Exercises on Chapter 6 90
Bibliography 93
Solutions 95
Chapter 1 95
Chapter 2 101
Chapter 3 104
Chapter 4 105
Chapter 5 108
Chapter 6 110
CHAPTER 1
Integral domains, ﬁelds and polynomial rings
Basic notions, convention, etc
In these notes, a ring will always be a ring with unity 1 = 0. Most of the rings encountered
will also be commutative. An ideal I R will always mean a twosided ideal. An ideal I R in
a ring R is proper if I = R, or equivalently if I R. Under a ring homomorphism ϕ: R −→ S,
1 ∈ R is sent to 1 ∈ S, i.e., ϕ(1) = 1.
1.1. Definition. Let ϕ: R −→ S be a ring homomorphism.
• ϕ is a monomorphism if it is injective, i.e., if for r
1
, r
2
∈ R,
ϕ(r
1
) = ϕ(r
2
) =⇒ r
1
= r
2
,
or equivalently if ker ϕ = ¦0¦.
• ϕ is an epimorphism if it is surjective, i.e., if for every s ∈ S there is an r ∈ R with
ϕ(r) = s.
• ϕ is an isomorphism if it is both a monomorphism and an epimorphism, i.e., if it is
invertible (in which case its inverse is also an isomorphism).
1.1. Recollections on integral domains and ﬁelds
The material in this section is standard and most of it should be familiar. Details may be
found in [3, 5] or other books containing introductory ring theory.
1.2. Definition. A commutative ring R in which there are no zerodivisors is called an
integral domain or an entire ring. This means that for u, v ∈ R,
uv = 0 =⇒ u = 0 or v = 0.
1.3. Example. The following rings are integral domains.
(i) The ring of integers, Z.
(ii) If p is a prime, the ring of integers modulo p, F
p
= Z/p = Z/(p).
(iii) The rings of rational numbers, Q, real numbers, R, and complex numbers, C.
(iv) The polynomial ring R[X], where R is an integral domain; in particular, the polynomial
rings Z[X], Q[X], R[X] and C[X] are all integral domains.
1.4. Definition. Let I R be a proper ideal in a commutative ring R.
• I is a prime ideal if for u, v ∈ R,
uv ∈ I =⇒ u ∈ I or v ∈ I.
• I is a maximal ideal R if whenever J R is a proper ideal and I ⊆ J then J = I.
• I R is principal if
I = (p) = ¦rp : r ∈ R¦
for some p ∈ R. Notice that if p, q ∈ R, then (q) = (p) if and only if q = up for some
unit u ∈ R. We also write p [ x if x ∈ (p).
• p ∈ R is prime if (p) R is a prime ideal; this is equivalent to the requirement that
whenever p [ xy with x, y ∈ R then p [ x or p [ y.
1
2 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
• R is a principal ideal domain if it is an integral domain and every ideal I R is principal.
1.5. Example. Every ideal I Z is principal, so I = (n) for some n ∈ Z which we can always
take to be nonnegative, i.e., n 0. Hence Z is a principal ideal domain.
1.6. Proposition. Let R be a commutative ring and I R an ideal.
(i) The quotient ring R/I is an integral domain if and only if I is a prime ideal.
(ii) The quotient ring R/I is a ﬁeld if and only if I is a maximal ideal.
1.7. Example. If n 0, the quotient ring Z/n = Z/(n) is an integral domain if and only if
n is a prime.
For any (not necessarily commutative) ring with unity there is an important ring homomor
phism η: Z −→ R called the unit or characteristic homomorphism which is deﬁned by
η(n) = n1 =
1 + + 1
. .. .
n
if n > 0,
−(1 + + 1
. .. .
−n
) if n < 0,
0 if n = 0.
Since 1 ∈ R is nonzero, ker η Z is a proper ideal and using the Isomorphism Theorems we
see that there is a quotient monomorphism η: Z/ ker η −→ R which allows us to identify the
quotient ring Z/ ker η with the image ηZ ⊆ R as a subring of R. By Example 1.5, there is a
unique nonnegative integer p 0 such that ker η = (p); this p is called the characteristic of R
and denoted char R.
1.8. Lemma. If R is an integral domain, its characteristic char R is a prime.
Proof. Consider p = char R. If p = 0 we are done. So suppose that p > 0. The quotient
monomorphism η: Z/ ker η −→ R identiﬁes Z/ ker η with the subring imη = imη of the integral
domain R. But every subring of an integral domain is itself an integral domain, hence Z/ ker η is
an integral domain. Now by Proposition 1.6(i), ker η = (p) is prime ideal and so by Example 1.7,
p is a prime.
1.9. Remark. When discussing a ring with unit R, we can consider it as containing as a
subring of the form Z/(char R) since the quotient homomorphism η: Z/(char R) −→ R gives
an isomorphism Z/(char R) −→ imη, allowing us to identify these rings. In particular, every
integral domain contains as a subring either Z = Z/(0) (if char R = 0) or Z/(p) if p = char R > 0
is a nonzero prime. This subring is sometimes called the characteristic subring of R. The rings
Z and Z/n = Z/(n) for n > 0 are often called core rings. When considering integral domains,
the rings Z and F
p
= Z/p = Z/(p) for p > 0 a prime are called prime rings.
Here is a useful and important fact about rings which contain a ﬁnite prime ring F
p
.
1.10. Theorem (Idiot’s Binomial Theorem). Let R be a commutative ring with unit con
taining F
p
for some prime p > 0. If u, v ∈ R, then
(u +v)
p
= u
p
+v
p
.
Proof. We have p1 = 0 in R, hence pt = 0 for any t ∈ R. The Binomial Expansion yields
(1.1) (u +v)
p
= u
p
+
p
1
u
p−1
v +
p
2
u
p−2
v
2
+ +
p
p −1
uv
p−1
+v
p
.
Now suppose that 1 j p −1. Then we have
p
j
=
p (p −1)!
j! (p −j)!
= p
(p −1)!
j! (p −j)!
.
1.1. RECOLLECTIONS ON INTEGRAL DOMAINS AND FIELDS 3
There are no factors of p appearing in (p −1)!, j! or (p −j)!, so since this number is an integer
it must be divisible by p, i.e.,
(1.2a) p [
p
j
,
or equivalently
(1.2b)
p
j
≡ 0 (mod p).
Hence in R we have
p
j
1 = 0.
Combining the divisibility conditions of (1.2) with the expansion of (1.1), we obtain the
required equation in R,
(u +v)
p
= u
p
+v
p
.
1.11. Definition. Let R be a ring. An element u ∈ R is a unit if it is invertible, i.e., there
is and element v ∈ R for which
uv = 1 = vu.
We usually write u
−1
for this element v, which is necessarily unique and is called the (multi
plicative) inverse of u in R. We will denote the set of all invertible elements of R by R
and
note that it always forms a group under multiplication.
1.12. Definition. A commutative ring k is a ﬁeld if every nonzero element u ∈ k is a unit.
This is equivalent to requiring that k
= k −¦0¦.
The familiar rings Q, R and C are all ﬁelds.
1.13. Example. If n 1, the quotient ring Z/n is a ﬁeld if and only if n is a prime.
1.14. Proposition. Every ﬁeld is an integral domain.
Proof. Let k be a ﬁeld. Suppose that u, v ∈ k and uv = 0. If u = 0, we can multiply by
u
−1
to obtain
v = u
−1
uv = 0,
hence v = 0. So at least one of u, v must be 0.
A nonzero element p ∈ R is irreducible if for u, v ∈ R,
p = uv =⇒ u or v is a unit.
1.15. Lemma. Let R be an integral domain. If p ∈ R is a nonzero prime then it is an
irreducible.
Proof. Suppose that p = uv for some u, v ∈ R. Then p [ u or p [ v, and we might as well
assume that u = tp for some t ∈ R. Then (1 −tv)p = 0 and so tv = 1, showing that v is a unit
with inverse t.
Now let D be an integral domain. A natural question to ask is whether D is isomorphic to
a subring of a ﬁeld. This is certainly true for the integers Z which are contained in the ﬁeld of
rational numbers Q, and for a prime p > 0, the prime ring F
p
is itself a ﬁeld.
1.16. Definition. The ﬁelds Q and F
p
where p > 0 is a prime are the prime ﬁelds.
Of course, we can view Z as a subring of any subﬁeld of the complex numbers so an answer
to this question may not be unique! However, there is always a ‘smallest’ such ﬁeld which is
unique up to an isomorphism.
4 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
1.17. Theorem. Let D be an integral domain.
(i) There is a ﬁeld of fractions of D, Fr(D), which contains D as a subring.
(ii) If ϕ: D −→ F is a ring monomorphism into a ﬁeld F, there is a unique homomorphism
¯ ϕ: Fr(D) −→ F such that ¯ ϕ(t) = ϕ(t) for all t ∈ D ⊆ Fr(D).
D
inc
ϕ
F
Fr(D)
∃! ¯ ϕ
Proof. (i) Consider the set
P(D) = ¦(a, b) : a, b ∈ D, b = 0¦.
Now introduce an equivalence relation ∼ on P(D), namely
(a
t
, b
t
) ∼ (a, b) ⇐⇒ ab
t
= a
t
b.
Of course, it is necessary to check that this relation is an equivalence relation; this is left as an
exercise. We denote the equivalence class of (a, b) by [a, b] and the set of equivalence classes by
Fr(D).
We deﬁne addition and multiplication on Fr(D) by
[a, b] + [c, d] = [ad +bc, bd], [a, b][c, d] = [ac, bd].
We need to verify that these operations are well deﬁned. For example, if [a
t
, b
t
] = [a, b] and
[c
t
, d
t
] = [c, d], then
(a
t
d
t
+b
t
c
t
)bd = a
t
d
t
bd +b
t
c
t
bd = ab
t
d
t
d +b
t
bcd
t
= (ad +bc)b
t
d
t
,
and so (a
t
d
t
+ b
t
c
t
, b
t
d
t
) ∼ (ad + bc, bd); hence addition is well deﬁned. A similar calculation
shows that (a
t
c
t
, b
t
d
t
) ∼ (ac, bd), so multiplication is also well deﬁned. It is now straightforward
to show that Fr(D) is a commutative ring with zero 0 = [0, 1] and unit 1 = [1, 1]. In fact, as we
will soon see, Fr(D) is a ﬁeld.
Let [a, b] ∈ Fr(D). Then [a, b] = [0, 1] if and only if (0, 1) ∼ (a, b) which is equivalent to
requiring that a = 0; notice that for any b = 0, [0, b] = [0, 1]. We also have [a, b] = [1, 1] if and
only if a = b.
Now let [a, b] ∈ Fr(D) be nonzero, i.e., a = 0. Then b = 0 and [a, b], [b, a] ∈ Fr(D) satisfy
[a, b][b, a] = [ab, ba] = [1, 1] = 1,
so [a, b] has [b, a] as an inverse. This shows that Fr(D) is a ﬁeld.
We can view D as a subring of Fr(D) using the map
j : D −→ Fr(D); j(t) = [t, 1]
which is a ring homomorphism; it is easy to check that it is a monomorphism. Therefore we
may identify t ∈ D with j(t) = [t, 1] ∈ Fr(D) and D with the subring imj ⊆ Fr(D).
(ii) Consider the function
Φ: P(D) −→ F; Φ(a, b) = ϕ(a)ϕ(b)
−1
.
1.1. RECOLLECTIONS ON INTEGRAL DOMAINS AND FIELDS 5
If (a
t
, b
t
) ∼ (a, b), then
Φ(a
t
, b
t
) = ϕ(a
t
)ϕ(b
t
)
−1
= ϕ(a
t
)ϕ(b)ϕ(b)
−1
ϕ(b
t
)
−1
= ϕ(a
t
b)ϕ(b)
−1
ϕ(b
t
)
−1
= ϕ(ab
t
)ϕ(b
t
)
−1
ϕ(b)
−1
= ϕ(a)ϕ(b
t
)ϕ(b
t
)
−1
ϕ(b)
−1
= ϕ(a)ϕ(b)
−1
= Φ(a, b),
so Φ is constant on each equivalence class of ∼. Hence we may deﬁne the function
¯ ϕ: Fr(D) −→ F; ¯ ϕ([a, b]) = Φ(a, b).
It is now easy to verify that ¯ ϕ is a ring homomorphism which agrees with ϕ on D ⊆ Fr(D).
The next three corollaries are left as an exercise.
1.18. Corollary. If F is a ﬁeld then F = Fr(F).
1.19. Corollary. If D is a subring of a ﬁeld F, then Fr(D) ⊆ Fr(F) = F and Fr(D) is
the smallest subﬁeld of F containing D.
1.20. Corollary. Let D
1
and D
2
be integral domains and let ϕ: D
1
−→ D
2
be a ring
monomorphism. Then there is a unique induced ring homomorphism ϕ
∗
: Fr(D
1
) −→ Fr(D
2
)
which satisﬁes ϕ
∗
(t) = ϕ(t) whenever t ∈ D
1
⊆ Fr(D
1
).
D
1
ϕ
inc
D
2
inc
Fr(D
1
)
ϕ
∗
Fr(D
2
)
Moreover, this construction has the following properties.
• If ϕ: D
1
−→ D
2
and θ: D
2
−→ D
3
are monomorphisms between integral domains then
θ
∗
◦ ϕ
∗
= (θ ◦ ϕ)
∗
as homomorphisms Fr(D
1
) −→ Fr(D
3
).
• For any integral domain D, the identity homomorphism id: D −→ D induces the
identity homomorphism (id)
∗
= id: Fr(D) −→ Fr(D).
D
1
ϕ
inc
D
2
θ
inc
D
3
inc
Fr(D
1
)
ϕ
∗
Fr(D
2
)
θ
∗
Fr(D
3
)
D
id
inc
D
inc
Fr(D)
id
∗
= id
Fr(D)
1.21. Remarks. (a) When working with a ﬁeld of fractions we usually adopt the familiar
notation
a
b
= a/b = [a, b]
for the equivalence class of (a, b). The rules for algebraic manipulation of such symbols are the
usual ones for working with fractions, i.e.,
a
1
b
1
+
a
2
b
2
=
a
1
b
2
+a
2
b
1
b
1
b
2
,
a
1
b
1
a
2
b
2
=
a
1
b
1
a
2
b
2
=
a
1
a
2
b
1
b
2
.
The ﬁeld of fractions of an integral domain is sometimes called its ﬁeld of quotients, however as
the word quotient is also associated with quotient rings we prefer to avoid using that terminology.
(b) Corollary 1.20 is sometimes said to imply that the construction of Fr(D) is functorial in the
integral domain D.
6 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
1.2. Polynomial rings
Let R be a commutative ring. We will make frequent use of the ring R[X] of polynomials
over R in an indeterminate X. This consists of elements of form
p(X) = p
0
+p
1
X + +p
m
X
m
where m 0 and p
0
, p
1
, . . . , p
m
∈ R; such p(X) are called polynomials. Addition and multipli
cation in R[X] are deﬁned by
(p
0
+p
1
X + +p
m
X
m
) + (q
0
+q
1
X + +q
m
X
m
) =
(p
0
+q
0
) + (p
1
+q
1
)X + + (p
m
+q
m
)X
m
,
and
(p
0
+p
1
X + +p
m
X
m
)(q
0
+q
1
X + +q
m
X
m
) =
(p
0
q
0
) + (p
0
q
1
+p
1
q
0
)X + + (p
0
q
m
+p
1
q
m−1
+ +p
m−1
q
1
+p
m
q
0
)X
2m
.
Then R[X] is a commutative ring with the constant polynomials 0 and 1 as its zero and unit.
We identify r ∈ R with the obvious constant polynomial; this allows us to view R as a subring
of R[X] and the inclusion function inc: R −→ R[X] is a monomorphism.
More generally, we inductively can deﬁne the ring of polynomials in n indeterminates
X
1
, . . . , X
n
over R,
R[X
1
, . . . , X
n
] = R[X
1
, . . . , X
n−1
][X
n
]
for n 1. Again there is an inclusion monomorphism inc: R −→ R[X
1
, . . . , X
n
] which sends
each element of R to itself considered as a constant polynomial.
These polynomial rings have an important universal property.
1.22. Theorem (Homomorphism Extension Property). Let ϕ: R −→ S be a ring homomor
phism.
(i) For each s ∈ S there is a unique ring homomorphism ϕ
s
: R[X] −→ S for which
• ϕ
s
(r) = ϕ(r) for all r ∈ R,
• ϕ
s
(X) = s.
R
inc
ϕ
S
R[X]
∃! ϕ
s
(ii) For n 1 and s
1
, . . . , s
n
∈ S, there is a unique ring homomorphism
ϕ
s
1
,...,s
n
: R[X
1
, . . . , X
n
] −→ S
for which
• ϕ
s
1
,...,s
n
(r) = ϕ(r) for all r ∈ R,
• ϕ
s
1
,...,s
n
(X
i
) = s
i
for i = 1, . . . , n.
R
inc
ϕ
S
R[X
1
, . . . , X
n
]
∃! ϕ
s
1
,...,s
n
1.2. POLYNOMIAL RINGS 7
Proof. (Sketch)
(i) For a polynomial p(X) = p
0
+p
1
X + +p
m
X
m
∈ R[X], we deﬁne
(1.3) ϕ
s
(p(X)) = p
0
+p
1
s + +p
m
s
m
∈ S.
It is then straightforward to check that ϕ
s
is a ring homomorphism with the stated properties
and moreover is the unique such homomorphism.
(ii) is proved by induction on n using (i).
We will refer to ϕ
s
1
,...,s
n
as the extension of ϕ by evaluation at s
1
, . . . , s
n
. It is standard to
write
p(s
1
, . . . , s
n
) = ϕ
s
1
,...,s
n
(p(X
1
, . . . , X
n
)).
An extremely important special case occurs when we start with the identity homomorphism
id: R −→ R and r
1
, . . . , r
n
∈ R; then we have the homomorphism
ε
r
1
,...,r
n
= id
r
1
,...,r
n
: R[X
1
, . . . , X
n
] −→ R.
Slightly more generally we may take the inclusion of a subring inc: R −→ S and s
1
, . . . , s
n
∈ S;
then
ε
s
1
,...,s
n
= inc
s
1
,...,s
n
: R[X
1
, . . . , X
n
] −→ S
is called evaluation at s
1
, . . . , s
n
and we denote its image by
R[s
1
, . . . , s
n
] = ε
s
1
,...,s
n
R[X
1
, . . . , X
n
] ⊆ S.
Then R[s
1
, . . . , s
n
] is a subring of S, called the subring generated by s
1
, . . . , s
n
over R.
Here is an example illustrating how we will use such evaluation homomorphisms.
1.23. Example. Consider the inclusion homomorphism inc: Q −→ C. We have the evalu
ation at i homomorphism ε
i
, for which ε
i
(X) = i. We easily see that ε
i
Q[X] ⊆ C is a subring
Q[i] ⊆ C consisting of the complex numbers of form a +bi with a, b ∈ Q.
Notice that if we had used −i instead of i, evaluation at −i, ε
−i
, we would also have
ε
−i
Q[X] = Q[i]. These evaluation homomorphisms are related by complex conjugation since
ε
−i
(p(X)) = ε
i
(p(X)),
which is equivalent to the functional equation
ε
−i
= ( ) ◦ ε
i
.
Notice also that in these examples we have
ker ε
−i
= ker ε
i
= (X
2
+ 1) Q[X],
hence we also have
Q[i]
∼
= Q[X]/(X
2
+ 1).
In fact (X
2
+ 1) is actually a maximal ideal and so Q[i] ⊆ C is a subﬁeld; later we will write
Q(i) for this subﬁeld.
1.24. Proposition. Let R be an integral domain.
(i) The ring R[X] of polynomials in an indeterminate X over R is an integral domain.
(ii) The ring R[X
1
, . . . , X
n
] of polynomials in the indeterminates X
1
, . . . , X
n
over R is an
integral domain.
1.25. Corollary. Let k be a ﬁeld and n 1. Then the polynomial ring k[X
1
, . . . , X
n
] in
the indeterminates X
1
, . . . , X
n
is an integral domain.
As we will make considerable use of such rings we describe in detail some of their important
properties. First we recall long division in a polynomial ring k[X] over a ﬁeld k; full details can
be found in a basic course on commutative rings or any introductory book on this subject.
8 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
1.26. Theorem (Long Division). Let k be a ﬁeld. Let f(X), d(X) ∈ k[X] and assume that
d(X) = 0 so that deg d(X) > 0. Then there are unique polynomials q(X), r(X) ∈ k[X] for
which
f(X) = q(X)d(X) +r(X)
and either deg r(X) < deg d(X) or r(X) = 0.
In the situation discussed in this result, the following names are often used. We refer to the
process of ﬁnding q(X) and r(X) as long division of f(X) by d(X). Also,
f(X) = the dividend, d(X) = the divisor, q(X) = the quotient, r(X) = the remainder.
1.27. Example. For k = Q, ﬁnd the quotient and remainder when f(X) = 6X
4
− 6X
3
+
3X
2
−3X + 1 is divided by d(X) = 2X
2
+ 1.
Solution. In the usual notation we have the following calculation.
3X
2
−3X
2X
2
+ 1 [ 6X
4
−6X
3
+ 3X
2
−3X + 1
6X
4
+ 0X
3
+ 3X
2
+ 0X + 0
−6X
3
+ 0X
2
−3X + 1
−6X
3
+ 0X
2
−3X + 0
1
Hence
6X
4
−6X
3
+ 3X
2
−3X + 1 = (3X
2
−3X)(2X
2
+ 1) + 1,
giving q(X) = 3X
2
−3X and r(X) = 1.
1.28. Example. For k = F
5
, ﬁnd the quotient and remainder when f(X) = 10X
5
+ 6X
4
−
6X
3
+ 3X
2
−3X + 1 is divided by d(X) = 2X
2
+ 1.
Solution. First notice that working modulo 5 we have
f(X) = 10X
5
+ 6X
4
−6X
3
+ 3X
2
−3X + 1 ≡ X
4
+ 4X
3
+ 3X
2
+ 2X + 1 (mod 5).
Notice also following multiplicative inverses in F
5
:
2
−1
≡ 3 (mod 5), 3
−1
≡ 2 (mod 5), 4
−1
≡ 4 (mod 5).
We have the following calculation.
3X
2
+ 2X
2X
2
+ 1[ 6X
4
+ 4X
3
+ 3X
2
+ 2X + 1
6X
4
+ 0X
3
+ 3X
2
+ 0X + 0
4X
3
+ 0X
2
+ 2X + 1
4X
3
+ 0X
2
+ 2X + 0
1
Hence
6X
4
−6X
3
+ 3X
2
−3X + 1 ≡ (3X
2
+ 2X)(2X
2
+ 1) + 1 (mod 5),
giving q(X) = 3X
2
+ 2X and r(X) = 1.
An important consequence of Theorem 1.26 is the following which makes use of the Euclidean
Algorithm.
1.2. POLYNOMIAL RINGS 9
1.29. Corollary. Let k be a ﬁeld and X an indeterminate. Let f(X), g(X) ∈ k[X] be
nonzero. Then there are a(X), b(X) ∈ k[X] such that
a(X)f(X) +b(X)g(X) = gcd(f(X), g(X)).
Here the greatest common divisor gcd(f(X), g(X)) of f(X), g(X) is the monic polynomial
of greatest degree which divides both of f(X), g(X).
1.30. Proposition. Let k be a ﬁeld and X an indeterminate. Then a nonconstant polyno
mial p(X) ∈ k[X] is an irreducible if and only if it is a prime.
Proof. By Lemma 1.15 we already know that p(X) is irreducible if it is prime. So sup
pose that p(X) is irreducible and that p(X) [ u(X)v(X) for u(X), v(X) ∈ k[X]. Then by
Corollary 1.29, there are a(X), b(X) ∈ k[X] such that
a(X)p(X) +b(X)u(X) = gcd(p(X), u(X)).
But since p(X) is irreducible, gcd(p(X), u(X)) = p(X) or gcd(p(X), u(X)) = 1. In the latter
case,
a(X)p(X) +b(X)u(X) = 1,
and multiplying through by v(X) gives
a(X)p(X)v(X) +b(X)u(X)v(X) = v(X)
and so p(X) [ v(X). This shows that p(X) [ u(X) or p(X) [ v(X), and so p(X) is prime.
1.31. Theorem. Let k be a ﬁeld and X an indeterminate.
(i) Every ideal I k[X] is principal, i.e., I = (h(X)) for some h(X) ∈ k[X].
(ii) The ideal (p(X)) k[X] is prime if and only if p(X) = 0 or p(X) is irreducible in k[X].
(iii) The quotient ring k[X]/(p(X)) is an integral domain if and only if p(X) = 0 or p(X)
is irreducible in k[X].
(iv) The quotient ring k[X]/(p(X)) is a ﬁeld if and only if p(X) is an irreducible in k[X].
Proof. (i) Let I k[X] and assume that I = (0). Then there must be at least one element
of I with positive degree and so we can choose h(X) ∈ I of minimal degree, say d = deg h(X).
Now let p(X) ∈ I. By Long Division, there are q(X), r(X) ∈ k[X] such that
p(X) = q(X)h(X) +r(X) and deg r(X) < d or r(X) = 0.
Since p(X) and h(X) are in the ideal I, we also have
r(X) = p(X) −q(X)h(X) ∈ I.
If r(X) = 0, this would contradict the minimality of d, so we must have r(X) = 0, showing that
p(X) = q(X)h(X). Thus I ⊆ (p(X)) ⊆ I and therefore I = (p(X)).
(ii) This follows from Proposition 1.30.
(iii) This follows from Proposition 1.6(i).
(iv) Since k[X] is an integral domain and not a ﬁeld, it follows that if k[X]/(p(X)) is a ﬁeld
then because it is an integral domain, p(X) is an irreducible by (iii).
Suppose that p(X) is irreducible (and hence is nonzero). Then for any q(X) ∈ k[X] with
q(X) / ∈ (p(X)), by Corollary 1.29 we can ﬁnd suitable a(X), b(X) ∈ k[X] for which
a(X)p(X) +b(X)q(X) = gcd(p(X), q(X)).
But gcd(p(X), q(X)) = 1 since p(X) is irreducible, so
a(X)p(X) +b(X)q(X) = 1.
This shows that in the quotient ring k[X]/(p(X)) the residue class of q(X) has the residue class
of b(X) as its inverse.
10 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
1.32. Remark. In connection with Theorem 1.31(i), notice that if p(X) ∈ k[X], then pro
vided d = deg p(X) > 0, we have for some p
d
= 0,
p(X) = p
0
+p
1
X + +p
d
X
d
= p
d
q(X),
where
q(X) = p
−1
d
p
0
+p
−1
d
p
1
X + +p
−1
d
p
d−1
X
d−1
+X
d
.
This easily implies that as ideals of k[X], (p(X)) = (q(X)). So we can always ﬁnd a monic
polynomial as the generator of a given ideal, and this monic polynomial is unique.
1.33. Proposition (Unique Factorization Property). Every nonconstant polynomial f(x) ∈
k[X] has a factorization
f(x) = cp
1
(X) p
k
(X),
where c ∈ k, and p
1
(X), . . . , p
k
(X) ∈ k[X] are irreducible monic polynomials. Moreover, c is
unique and the sequence of polynomials p
1
(X), . . . , p
k
(X) is unique apart from the order of the
terms.
Proof. (Sketch)
Existence is proved by induction on the degree of f(X) and begins with the obvious case
deg f(X) = 1. If deg f(X) > 1, then either f(X) is already irreducible, or f(X) = f
1
(X)f
2
(X)
with both factors of positive degree, and therefore deg f
j
(X) < deg f(X). This gives the
inductive step.
To prove uniqueness, suppose that
p
1
(X) p
k
(X) = q
1
(X) q
(X)
where p
i
(X), q
j
(X) ∈ k[X] are irreducible monic polynomials. Then by Proposition 1.30, each
p
i
(X) is prime, hence divides one of the q
j
(X), hence must equal it. By reordering we can
assume that p
i
(X) = q
i
(X) and k . After cancelling common factors we obtain
q
k+1
(X) q
(X) = 1,
and so we see that k = .
1.34. Corollary. Suppose that f(X) ∈ k[X] factors into linear factors
f(X) = c(X −u
1
) (X −u
d
),
where u
1
, . . . , u
d
∈ k. Then the sequence of roots u
1
, . . . , u
d
is unique apart from the order. In
particular, if v
1
, . . . , v
r
are the distinct roots, then
f(X) = c(X −v
1
)
m
1
(X −v
r
)
m
r
,
where m
i
> 0 and this factorization is unique apart from the order of the pairs (v
i
, m
i
).
1.35. Corollary. The number of distinct roots of a nonconstant polynomial f(X) ∈ k[X]
is at most deg f(X).
1.36. Definition. If k is a ﬁeld and X an indeterminate, then the ﬁeld of fractions of k[X]
is the ﬁeld of rational functions, k(X). The elements of k(X) are fractions of the form
a
0
+a
1
X + +a
m
X
m
b
0
+b
1
X + +b
n
X
n
with a
i
, b
j
∈ k and b
0
+b
1
X + +b
n
X
n
= 0.
1.3. IDENTIFYING IRREDUCIBLE POLYNOMIALS 11
1.3. Identifying irreducible polynomials
When k is a ﬁeld, we will need some eﬀective methods for deciding when a polynomial in
k[X] is irreducible.
Let us consider factorisation of polynomials over Q. If f(X) ∈ Z[X] then we can also consider
f(X) as an element of Q[X]. If R = Z or Q, we say that f(X) has a proper factorisation over
R if f(X) = g(X)h(X) for some g(X), h(X) ∈ R[X] with deg g(X) > 0 and deg h(X) > 0.
1.37. Proposition (Gauss’s Lemma). Let f(X) ∈ Z[X]. Then f(X) has a proper factori
sation over Z if and only it has a proper factorisation over Q.
So to ﬁnd factors of f(X) it is suﬃcient to look for factors in Z[X]. Our next result is a
special case of the Eisenstein Irreducibility Test. The version here is slightly more general than
the more usual one which corresponds to taking s = 0.
1.38. Proposition (Eisenstein Test). Let f(X) ∈ Z[X] and s ∈ Z. Choose a
i
∈ Z so that
f(X) = a
0
+a
1
(X −s) + +a
d−1
(X −s)
d−1
+a
d
(X −s)
d
,
where d = deg f(X). Suppose that p > 0 is a prime for which the following three conditions
hold:
• a
k
≡ 0 (mod p) for k = 0, . . . , d −1;
• a
0
≡ 0 (mod p
2
);
• a
d
≡ 0 (mod p).
Then f(X) is irreducible in Q[X] and hence also in Z[X].
1.39. Example. Let p 2 be a prime. Then the polynomial
Φ
p
(X) = 1 +X + +X
p−1
∈ Z[X]
is irreducible in Q[X] and hence also in Z[X].
Proof. Working in Z[X],
Φ
p
(X)(X −1) = (1 +X + +X
p−1
)(X −1)
= X
p
−1
= (1 + (X −1))
p
−1
=
p
¸
k=1
p
k
(X −1)
k
≡ (X −1)
p
(mod p),
since by (1.2a), p divides
p
k
=
p!
k! (p −k)!
when k = 1, . . . , p −1. Hence
Φ
p
(X) ≡ (X −1)
p−1
(mod p)
Also,
p
1
= p ≡ 0 (mod p
2
),
giving
(1.4) Φ
p
(X) = (X −1)
p−1
+c
p−2
(X −1)
p−2
+ +c
1
(X −1) +c
0
with c
r
≡ 0 (mod p) and c
0
= p. So the Eisenstein Test can be applied here with s = 1 to show
that Φ
p
(X) is irreducible in Z[X].
12 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
1.40. Example. As examples we have the irreducible polynomials
Φ
2
(X) = 1 +X,
Φ
3
(X) = 1 +X +X
2
,
Φ
5
(X) = 1 +X +X
2
+X
3
+X
4
,
Φ
7
(X) = 1 +X +X
2
+X
3
+X
4
+X
5
+X
6
,
Φ
11
(X) = 1 +X +X
2
+X
3
+X
4
+X
5
+X
6
+X
7
+X
8
+X
9
+X
10
.
These are examples of the cyclotomic polynomials Φ
n
(X) ∈ Z[X] which are deﬁned for all
n 1 by
(1.5a) X
n
−1 =
¸
d[n
Φ
d
(X),
where the product is taken over all the positive divisors of n. For example,
X
2
−1 = (X −1)(X + 1) = Φ
1
(X)Φ
2
(X),
X
3
−1 = (X −1)(X
2
+X + 1) = Φ
1
(X)Φ
3
(X),
X
4
−1 = (X −1)(X + 1)(X
2
+ 1) = Φ
1
(X)Φ
2
(X)Φ
4
(X),
X
5
−1 = (X −1)(X
4
+X
3
+X + 1) = Φ
1
(X)Φ
5
(X),
X
6
−1 = (X −1)(X + 1)(X
2
+X + 1)(X
2
−X + 1) = Φ
1
(X)Φ
2
(X)Φ
3
(X)Φ
6
(X),
X
12
−1 = (X −1)(X + 1)(X
2
+X + 1)(X
2
+ 1)(X
2
−X + 1)(X
4
−X
2
+ 1)
= Φ
1
(X)Φ
2
(X)Φ
3
(X)Φ
4
(X)Φ
6
(X)Φ
12
(X).
Cyclotomic polynomials can be computed recursively using Equation (1.5a). If we know Φ
k
(X)
for k < n, then
(1.5b) Φ
n
(X) =
X
n
−1
¸
d[n
d<n
Φ
d
(X)
.
The degree of Φ
n
(X) involves a function of n probably familiar from elementary Number Theory.
1.41. Definition. The Euler function ϕ: N −→N is deﬁned by
ϕ(n) = number of k = 1, . . . , n for which gcd(n, k) = 1
= [(Z/n)
[ = number of units in Z/n
= number of generators of the cyclic group Z/n.
In particular, if p 2 is a prime then ϕ(p) = p −1. Of course, ϕ(1) = 1.
It can be shown that for each natural number n,
(1.6)
¸
d[n
ϕ(d) = n.
Notice that we can inductively determine ϕ(n) using this equation. For example, if p and q are
distinct primes, then
ϕ(pq) = pq −(ϕ(p) +ϕ(q) +ϕ(1)) = pq −(p −1) −(q −1) −1 = (p −1)(q −1).
It is also true that whenever m, n are coprime, i.e., when gcd(m, n) = 1,
(1.7) ϕ(mn) = ϕ(m)ϕ(n).
1.3. IDENTIFYING IRREDUCIBLE POLYNOMIALS 13
Thus if n = p
r
1
1
p
r
s
s
where p
1
< p
2
< < p
s
are the prime factors of n and r
j
> 0, then
(1.8) ϕ(n) = ϕ(p
r
1
1
) ϕ(p
r
s
s
).
Furthermore, if p is a prime and r > 0, then
(1.9) ϕ(p
r
) = (p −1)p
r−1
.
Notice that as a result, ϕ(n) is even when n > 2.
1.42. Remark. For those who know about the M¨obius function µ (which takes values 0, ±1)
and M¨obius inversion, the latter can be used to solve Equation (1.6) for ϕ, giving
(1.10) ϕ(n) =
¸
d[n
µ(d)
n
d
.
Similarly, the formulae of (1.5) lead to
(1.11) Φ
n
(X) =
¸
d[n
(X
n/d
−1)
µ(d)
.
So for example, if p, q are distinct primes, then using standard properties of µ,
Φ
pq
(X) = (X
pq
−1)
µ(1)
(X
pq/p
−1)
µ(p)
(X
pq/q
−1)
µ(q)
(X
pq/pq
−1)
µ(pq)
= (X
pq
−1)(X
q
−1)
−1
(X
p
−1)
−1
(X −1) =
(X
pq
−1)(X −1)
(X
q
−1)(X
p
−1)
.
Recall that an element ζ of a ﬁeld K is a primitive nth root of unity if
min¦k : 1 k and ζ
k
= 1¦ = n.
We think of ζ
n
= e
2πi/n
as the standard complex primitive nth root of unity. Then every
complex nth root of unity has the form ζ
k
n
= e
2πik/n
for k = 0, 1, . . . , n −1.
1.43. Theorem. For each n 1, the cyclotomic polynomial Φ
n
(X) is irreducible in Q[X]
and hence in Z[X]. The complex roots of Φ
n
(X) are the primitive nth roots of unity,
ζ
k
n
= e
2πik/n
(0 k n −1, gcd(k, n) = 1).
and the number of these is deg Φ
n
(X) = ϕ(n). Hence,
Φ
n
(X) =
¸
t=1,...,n−1
gcd(t,n)=1
(X −ζ
t
n
).
Proof. We will give a reformulation and proof of this in Theorem 6.2.
1.44. Example. For n = 6 we have
ζ
6
= e
2πi/6
= e
πi/3
=
1
2
+
√
3
2
i.
Then ϕ(6) = 2 and
Φ
6
(X) = X
2
−X + 1 = (X −ζ
6
)(X −ζ
5
6
).
It is also worth recording a related general result on cyclic groups.
1.45. Proposition. Let n 1 and C = 'g` be a cyclic group of order n and a generator g.
Then an element g
r
∈ C is a generator if and only if gcd(r, n) = 1; the number of such elements
of C is ϕ(n).
This leads to a useful group theoretic result.
1.46. Lemma. Let G be a ﬁnite group satisfying the following condition:
• For each n 1, there are at most n solutions of x
n
= ι in G.
14 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
Then G is cyclic and in particular is abelian.
Proof. Let θ
G
(d) denote the number of elements in G of order d. By Lagrange’s Theorem,
θ
G
(d) = 0 unless d divides [G[. Since
G =
¸
d[[G[
¦g ∈ G : [g[ = d¦,
we have
[G[ =
¸
d[[G[
θ
G
(d).
Recall the Euler ϕfunction satisﬁes Equation (1.6), hence
[G[ =
¸
d[[G[
ϕ(d).
Combining these we obtain
(1.12)
¸
d[[G[
θ
G
(d) =
¸
d[[G[
ϕ(d).
Let d be a divisor of [G[. By Proposition 1.45, for each element g ∈ G of order d, the cyclic
subgroup 'g` G has ϕ(d) generators, each of order d. As there are at most d such elements g
in G, this gives θ
G
(d) ϕ(d). So
¸
d[[G[
θ
G
(d)
¸
d[[G[
ϕ(d).
Now if θ
G
(d) < ϕ(d) for some d, we would have a strict inequality in place of Equation (1.12).
Hence θ
G
(d) = ϕ(d) for all d. In particular, there are ϕ([G[) elements of order [G[, hence there
must be an element of order [G[, so G is cyclic.
The above results for polynomials over Q and Z have analogues over the ﬁeld of fractions
k(T) and polynomial ring k[T], where k is a ﬁeld.
A polynomial f(X) ∈ k[T][X] is an element of k(T)[X]. If R = k[T] or k(T), we say that
f(X) has a proper factorisation over R if f(X) = g(X)h(X) for some g(X), h(X) ∈ R[X] with
deg g(X) > 0 and deg h(X) > 0.
1.47. Proposition (Gauss’s Lemma). Let f(X) ∈ k[T][X]. Then f(X) has a proper fac
torisation over k[T] if and only it has a proper factorisation over k(T).
Here is another version of the Eisenstein Test; again we state a version which is slightly
more general than the usual one which corresponds to the case where s = 0.
1.48. Proposition (Eisenstein Test). Let f(X) ∈ k[T][X] and s ∈ k[T]. Choose a
i
∈ k[T]
so that
f(X) = a
0
+a
1
(X −s) + +a
d−1
(X −s)
d−1
+a
d
(X −s)
d
,
where d = deg f(X). Suppose that p(T) ∈ k[T] is an irreducible for which the following three
conditions hold:
• a
k
≡ 0 (mod p(T)) for k = 0, . . . , d −1;
• a
0
≡ 0 (mod p(T)
2
);
• a
d
≡ 0 (mod p(T)).
Then f(X) is irreducible in k(T)[X] and hence also in k[T][X].
1.49. Example. Let k be a ﬁeld. Then the polynomial X
n
−T is irreducible in k(T)[X].
1.4. FINDING ROOTS OF COMPLEX POLYNOMIALS OF SMALL DEGREE 15
1.4. Finding roots of complex polynomials of small degree
♠
♥ ♦
♣
In this section we work within the complex numbers and take k ⊆ C. In practice we
will usually have k = R or k = C.
For monic linear (degree 1) or quadratic (degree 2) polynomials, methods of ﬁnding roots are
very familiar. Let us consider the cases of cubic (degree 3) and quartic (degree 4) polynomials.
Cubic polynomials: Cardan’s method. The following 16th century method of ﬁnding
roots of cubics is due to Jerˆome Cardan who seems to have obtained some preliminary versions
from Niccol`o Tartaglia by somewhat disreputable means! For historical details see [2, 3].
A monic cubic
f(X) = X
3
+a
2
X
2
+a
1
X +a
0
∈ C[X]
can be transformed into one with no quadratic term by a change of variables X −→ X −a
2
/3
giving
g(X) = f(X −a
2
/3) = X
3
−
a
1
−
1
3
a
2
2
X +
a
0
−
a
1
a
2
3
+
2a
3
2
27
∈ C[X].
Clearly ﬁnding the roots of f(X) is equivalent to ﬁnding those of g(X), so we may as well
assume that we want to ﬁnd the complex roots of
f(X) = X
3
+pX +q ∈ C[X].
Suppose that x ∈ C is a root of f(X), i.e.,
(1.13) x
3
+px +q = 0.
If we introduce u ∈ C for which
x = u −
p
3u
,
then
u −
p
3u
3
+p
u −
p
3u
+q = 0
and so
u
3
−
p
3
27u
3
+q = 0,
hence
u
6
+qu
3
−
p
3
27
= 0.
Solving for u
3
we obtain
u
3
= −
q
2
±
1
2
q
2
+
4p
3
27
,
where
q
2
+
4p
3
27
denotes one of the complex square roots of the discriminant of the quadratic
equation
U
2
+qU −
p
3
27
= 0.
Now if we take u to be a cube root of one of the complex numbers
−
q
2
±
1
2
q
2
+
4p
3
27
16 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
we obtain the desired root of f(X) as x = u − p/3u. Notice that we have a choice of 2 values
for u
3
and for each of these a choice of 3 values for u, diﬀering by factors of the form ω
r
for
r = 0, 1, 2 where ω = e
2πi/3
is a primitive cube root of 1. However, since
1
−q +
q
2
+
4p
3
27
=
−q −
q
2
+
4p
3
27
q
2
−(q
2
+ 4p
3
/27)
= −27
−q −
q
2
+
4p
3
27
4p
3
,
it is easy to verify that there are in fact only 3 choices of the root x which we can write
symbolically as
x =
3
−
q
2
+
1
2
q
2
+
4p
3
27
+
3
−
q
2
−
1
2
q
2
+
4p
3
27
(1.14)
or more precisely as
x =
3
−
q
2
+
1
2
q
2
+
4p
3
27
−
p
3
3
−
q
2
+
1
2
q
2
+
4p
3
27
. (1.15)
1.50. Example. Find the complex roots of the polynomial
f(X) = X
3
+ 3X −10 ∈ R[X].
Solution. Applying the method above, we reduce to the quadratic equation
U
2
−10U −1 = 0
whose roots are 5 ±
√
26 ∈ R. Notice that 5 +
√
26 > 0 and 5 −
√
26 < 0; we also have
5 −
√
26 =
−1
5 +
√
26
.
Now 5 +
√
26 has the complex cube roots
3
5 +
√
26,
3
5 +
√
26 ω,
3
5 +
√
26 ω
2
.
Here we have x = u −1/u, so the 3 complex roots of f(X) are
3
5 +
√
26 −
1
3
5 +
√
26
ω
r
(r = 0, 1, 2).
Notice that one of these is real, namely
3
5 +
√
26 −
1
3
5 +
√
26
=
3
5 +
√
26
2
−1
3
5 +
√
26
.
Quartic polynomials: Ferrari’s method. The following method of ﬁnding roots of quar
tics was publicised by Cardan who attributed it to his student Lodovicio Ferrari.
A general monic quartic polynomial
f(X) = X
4
+a
3
X
3
+a
2
X
2
+a
1
X +a
0
∈ C[X]
can be transformed into one with no cubic term by a change of variables X −→ X−a
2
/3 giving
g(X) = f(X −a
3
/4) =
Y
4
+
a
2
−
3
8
a
2
3
Y
2
+
1
8
a
3
3
−
1
2
a
2
a
3
+a
1
Y −
1
16
a
2
a
2
3
−
3
256
a
4
3
+
1
4
a
1
a
3
+a
0
.
1.5. AUTOMORPHISMS OF RINGS AND FIELDS 17
Clearly ﬁnding the roots of f(X) is equivalent to ﬁnding those of g(X), so we may as well
assume that we want to ﬁnd the complex roots of
f(X) = X
4
+pX
2
+qX +r ∈ C[X].
Suppose that x is a root and introduce numbers y, z such that z = x
2
+ y (we will ﬁx the
values of these later). Then
z
2
= x
4
+ 2x
2
y +y
2
= −px
2
−qx −r + 2x
2
y +y
2
= (2y −p)x
2
−qx +y
2
−r.
Now choose y to make the last quadratic expression in x a square,
(1.16) (2y −p)x
2
−qx + (y
2
−r) = (Ax +B)
2
.
This can be done by requiring the vanishing of the discriminant
(1.17) q
2
−4(2y −p)(y
2
−r) = 0.
Notice that if y = p/2 then we would require q = 0 and then
f(X) = X
4
+pX
2
+r = (X
2
)
2
+p(X
2
) +r = 0
can be solved by solving
Z
2
+pZ +r = 0.
Since Equation (1.17) is a cubic in y, we can use the method of solution of cubics to ﬁnd a root
y = t say. Then for Equation (1.16) we have
(x
2
+t)
2
= (Ax +B)
2
,
whence
x
2
= −t ±(Ax +B).
Thus taking the two square roots of the right hand side we obtain 4 values for x, which we write
symbolically as
x = ±
−t ±(Ax +B).
1.51. Remark. In the case of cubic and quartic polynomials over C we can obtain all the
roots by repeatedly taking square or cube roots (or radicals). Consequently such polynomials are
said to be solvable by radicals. Later we will see that this is not true in general for polynomials
of degree at least 5; this is one of the great early successes of this theory.
1.5. Automorphisms of rings and ﬁelds
1.52. Definition. Let R be a ring and R
0
⊆ R a subring.
• An automorphism of R is a ring isomorphism α: R −→ R. The set of all such auto
morphisms is denoted Aut(R).
• An automorphism of R over R
0
is a ring isomorphism α: R −→ R for which α(r) = r
whenever r ∈ R
0
. The set of all automorphisms of R over R
0
is denoted Aut
R
0
(R).
1.53. Proposition. For a ring R with a subring R
0
⊆ R, Aut(R) and Aut
R
0
(R) form
groups under composition of functions.
Proof. The composition α◦β of two automorphisms α, β: R −→ R is also an automorphism
of R as is the inverse of α. The identity function id = id
R
: R −→ R is an automorphism. Hence
Aut(R) forms a group under composition. The argument for Aut
R
0
(R) is similar.
1.54. Proposition. Let R be one of the core rings Z or Z/n with n > 1. Then
18 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
(i) The only automorphism of R is the identity, i.e., Aut(R) = ¦id¦.
(ii) If S is a ring containing a core ring R and α ∈ Aut(S), then α restricts to the identity
on R, i.e., α(r) = r for all r ∈ R. Hence, Aut(S) = Aut
R
(S).
Proof. (i) For such a core ring R, every element has the form k1 for some k ∈ Z. For an
automorphism α of R,
α(k1) =
α(1) + +α(1)
. .. .
k
if k > 0,
−(α(1) + +α(1)
. .. .
−k
) if k < 0,
α(0) if k = 0
=
1 + + 1
. .. .
k
if k > 0,
−(1 + + 1
. .. .
−k
) if k < 0,
0 if k = 0
=k1.
Thus α = id.
(ii) For α ∈ Aut(S), α(1) = 1 and a similar argument to that for (i) shows that α(r) = r for all
r ∈ R.
1.55. Proposition. Let D be an integral domain and α: D −→ D be an automorphism.
Then the induced homomorphism gives an automorphism α
∗
: Fr(D) −→ Fr(D).
Proof. Given α, the induced homomorphism α
∗
: Fr(D) −→ Fr(D) exists and we need
to show it has an inverse. The inverse automorphism α
−1
: D −→ D also gives rise to an
induced homomorphism (α
−1
)
∗
: Fr(D) −→ Fr(D). Since α
−1
◦ α = id = α ◦ α
−1
, we can apply
Corollary 1.20 to show that
(α
−1
)
∗
◦ (α)
∗
= id = (α)
∗
◦ (α
−1
)
∗
.
Hence (α)
∗
is invertible with inverse (α
−1
)
∗
.
1.56. Corollary. There is a monomorphism of groups
( )
∗
: Aut(D) −→ Aut(Fr(D)); α −→ α
∗
.
1.57. Example. The ﬁeld of fractions of the ring of integers Z is the ﬁeld of rationals Q.
The homomorphism
( )
∗
: Aut(Z) −→ Aut(Q); α −→ α
∗
is an isomorphism and hence Aut(Q) = ¦id¦.
Combining this example with Proposition 1.54(ii) we obtain another useful result.
1.58. Proposition. Let k be one of the prime ﬁelds Q or F
p
with p > 0 prime. If R is
a ring containing k as a subring, then every automorphism of R restricts to the identity on k,
i.e., Aut(R) = Aut
k
(R).
Recalling Deﬁnition 1.36, we have an example which shows that the monomorphism of
Corollary 1.56 need not be an epimorphism. Here we take D = Q[X] and Fr(Q[X]) = Q(X).
1.5. AUTOMORPHISMS OF RINGS AND FIELDS 19
1.59. Example. The homomorphism
( )
∗
: Aut(Q[X]) −→ Aut(Q(X)); α −→ α
∗
is a monomorphism but it is not an epimorphism since there is an automorphism
γ : Q(X) −→Q(X); γ(f(X)) = f(1/X)
which sends X ∈ Q[X] ⊆ Q(X) to 1/X / ∈ Q[X] and so does not restrict to an automorphism of
Q[X].
Let k be a ﬁeld. The group of invertible 2 2 matrices over k is the 2 2 general linear
group over k,
GL
2
(k) =
¸
a
11
a
12
a
21
a
22
: a
ij
∈ k, a
11
a
22
−a
12
a
21
= 0
The scalar matrices form a normal subgroup
Scal
2
(k) = ¦diag(t, t) : t ∈ k, t = 0¦ GL
2
(k).
The quotient group is called the 2 2 projective general linear group over k,
PGL
2
(k) = GL
2
(k)/ Scal
2
(k).
Notice that GL
2
(k) has another interesting subgroup called the aﬃne subgroup,
Aﬀ
1
(k) =
¸
a b
0 1
: a, b ∈ k, a = 0
GL
2
(k).
1.60. Example. Let k be a ﬁeld and X an indeterminate. Then Aut
k
(k[X]) and hence
Aut
k
(k(X)), contains a subgroup isomorphic to Aﬀ
1
(k). In fact, Aut
k
(k[X])
∼
= Aﬀ
1
(k).
Proof. We begin by showing that to each aﬃne matrix
A =
¸
a b
0 1
∈ Aﬀ
1
(k)
there is an associated automorphism α
A
: k[X] −→k[X].
For this we use the element aX + b ∈ k[X] together with the extension result of Theo
rem 1.22(i) to obtain a homomorphism α
A
: k[X] −→ k[X] with α
A
(X) = aX + b. Using the
inverse matrix
A
−1
=
¸
a
−1
−a
−1
b
0 1
we similarly obtain a homomorphism α
A
−1 : k[X] −→k[X] for which
α
A
−1(X) = a
−1
X −a
−1
b.
Using the same line of argument as in the proof of Proposition 1.55 (or doing a direct calculation)
we see that α
A
−1 is the inverse of α
A
an so α
A
∈ Aut
k
(k[X]). It is straightforward to check
that for A
1
, A
2
∈ Aﬀ
1
(k),
α
A
2
A
1
= α
A
1
◦ α
A
2
,
(note the order!) hence there is a homomorphism of groups
Aﬀ
1
(k) −→ Aut
k
(k[X]); A −→ α
A
−1,
which is easily seen to be a monomorphism. Composing with ( )
∗
we see that there is a
monomorphism Aﬀ
1
(k) −→ Aut
k
(k(X)). In fact, this is also an epimorphism and we leave the
proof of this as an exercise.
1.61. Example. Let k be a ﬁeld and X an indeterminate. Then
(i) Aut
k
(k(X)) contains a subgroup isomorphic to PGL
2
(k).
(ii) In fact, Aut
k
(k(X))
∼
= PGL
2
(k).
20 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
Proof. (i) We begin by showing that to each invertible matrix
A =
¸
a
11
a
12
a
21
a
22
∈ GL
2
(k)
there is an associated automorphism α
A
: k(X) −→k(X).
We begin by choosing the element (a
11
X +a
12
)/(a
21
X +a
22
) ∈ k(X) and then using Theo
rem 1.22(i) to obtain a homomorphismk[X] −→k(X) that sends X to (a
11
X+a
12
)/(a
21
X+a
22
).
By applying ( )
∗
to this we obtain a homomorphism (known as a fractional linear transforma
tion) α
A
: k(X) −→k(X) for which
α
A
(X) =
a
11
X +a
12
a
21
X +a
22
.
Again we ﬁnd that
α
A
2
A
1
= α
A
1
◦ α
A
2
.
There is an associated homomorphism of groups GL
2
(k) −→ Aut
k
(k(X)) sending A to α
A
−1
.
However, this is not an injection in general since for each scalar matrix diag(t, t),
α
diag(t,t)
(X) =
tX
t
= X,
showing that α
diag(t,t)
is the identity function.
In fact it is easy to see that Scal
2
(k)GL
2
(k) is the kernel of this homomorphism. Therefore
passing to the quotient PGL
2
(k) = GL
2
(k)/ Scal
2
(k) we obtain a monomorphism PGL
2
(k) −→
Aut
k
(k(X)). There is one case where Scal
2
(k) is the trivial group, namely k = F
2
.
(ii) To show that every automorphism of k(X) is a fractional linear transformation is less
elementary. We give a sketch proof for the case of k = C; actually this argument can be modiﬁed
to work for any algebraically closed ﬁeld, but an easy argument then shows the general case.
Let α ∈ Aut
C
(C(X)). There is an associated rational (hence meromorphic) function f given
by z −→ f(z), where α(X) = f(X), deﬁned on C with the poles of f deleted. If we write
f(X) =
p(X)
q(X)
where p(X), q(X) ∈ C[X] have no common factors of positive degree, then the order of f(X) is
ord f = max¦deg p(X), deg q(X)¦.
Now let c ∈ C. Then the number of solutions counted with algebraic multiplicity of the equation
f(z) = c turns out to be ordf. Also, if deg p(X) deg q(X) then the number of poles of f
counted with algebraic multiplicity is also ordf. Finally, if deg p(X) > deg q(X) then we can
write
f(X) = p
1
(X) +
p
0
(X)
q(X)
,
where p
0
(X), p
1
(X) ∈ C[X] and deg p
0
(X) < deg q(X). Then the number of poles of f counted
with algebraic multiplicity is
deg p
1
(X) + ord
p
0
q
.
Now it is easy to see that since α is invertible so is the function f. But this can only happen
if the function f is injective which means that all of these numbers must be 1, hence ordf = 1.
Thus
f(X) =
aX +b
cX +d
= constant
and the matrix
¸
a b
c d
must be invertible.
21
Clearly not every fractional linear transformation α
A
: k(X) −→k(X) maps polynomials to
polynomials so ( )
∗
: Aut
k
(k[X]) −→ Aut
k
(k(X)) is not an epimorphism.
Now we turn to a more familiar ﬁeld R, the real numbers.
1.62. Proposition. The only automorphism of the ﬁeld R is the identity function, hence
Aut(R) = ¦id¦.
Proof. First we note that Q ⊆ R is a subring and if α ∈ Aut(R) then α(q) = q for q ∈ Q
by Example 1.57.
We recall from Analysis that the rational numbers are dense in the real numbers in the
sense that each r ∈ R can be expressed as a limit r = lim
n→∞
q
n
, where q
n
∈ Q. Then for a
continuous function f : R −→R, its value at r depends on its values on Q since
f(r) = f( lim
n→∞
q
n
) = lim
n→∞
f(q
n
).
We will show that an automorphism α ∈ Aut(R) is continuous.
First recall that for x, y ∈ R,
x < y ⇐⇒ 0 < y −x ⇐⇒ y −x = t
2
for some nonzero t ∈ R.
Now for α ∈ Aut(R) and s ∈ R, we have α(s
2
) = α(s)
2
. Hence,
x < y =⇒ α(y) −α(x) = α(t)
2
for some nonzero t ∈ R =⇒ α(x) < α(y).
So α preserves order and ﬁxes rational numbers.
Now let x ∈ R and ε > 0. Then we can choose a rational number q such that 0 < q ε.
Taking δ = q we ﬁnd that for y ∈ R with [y −x[ < δ (i.e., −δ < y −x < δ) we have
−δ = α(−δ) < α(y) −α(x) < α(δ) = δ,
hence
[α(y) −α(x)[ < δ ε.
This shows that α is continuous at x.
Thus every automorphism of R is continuous function which ﬁxes all the rational numbers,
hence it must be the identity function.
1.63. Remark. If we try to determine Aut(C) the answer turns out to be much more
complicated. It is easy to see that complex conjugation ( ): C −→C is an automorphism of C
and ﬁxes every real number, i.e., ( ) ∈ Aut
R
(C); in fact, Aut
R
(C) = ¦id, ( )¦. However, it is
not true that every α ∈ Aut(C) ﬁxes every real number! The automorphism group Aut(C) is
actually enormous but it is hard to ﬁnd an explicit element other than id and ( ). Note that
given an automorphism α ∈ Aut(C), the composition α ◦ ( ) ◦ α
−1
is also self inverse, so there
are many elements of order 2 in the group Aut(C).
Exercises on Chapter 1
1.1. Let R be a ring. Show that
¦n ∈ Z : n > 0 and n1 = 0¦ = ¦n ∈ Z : n > 0 and nr = 0 for all r ∈ R¦.
Deduce that if char R > 0 then these sets are nonempty and
char R = min¦n ∈ Z : n > 0 and nr = 0 for all r ∈ R¦.
1.2. Let R be an integral domain.
(a) Show that every subring S ⊆ R is also an integral domain. What is the relationship
between char S and char R?
22 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
(b) If R is a ﬁeld, give an example to show that a subring of R need not be a ﬁeld.
1.3. For each of the following rings R, ﬁnd the characteristic char R and the characteristic
subring of R. Determine which of these rings is an integral domain. In (b) and (c), A is an
arbitrary commutative ring.
(a) Any subring R ⊆ C.
(b) The polynomial ring R = A[X].
(c) The ring of n n matrices over A,
R = Mat
n
(A) =
a
11
. . . a
1n
.
.
.
.
.
.
.
.
.
a
n1
. . . a
nn
¸
¸
¸
: a
ij
∈ A
.
1.4. If R is a commutative ring with unit containing the prime ﬁeld F
p
for some prime p > 0,
show that the function ϕ: R −→ R given by ϕ(t) = t
p
, deﬁnes a ring homomorphism. Give
examples to show that ϕ need not be surjective or injective.
1.5. Let R and S be rings with unity and Q S a prime ideal.
(a) If ϕ: R −→ S is a ring homomorphism, show that
ϕ
−1
Q = ¦r ∈ R : ϕ(r) ∈ Q¦ ⊆ R
is a prime ideal of R.
(b) If R ⊆ S is a subring, show that Q∩ R is a prime ideal of R.
(c) If the word ‘prime’ is replaced by ‘maximal’ throughout, are the results in parts (a)
and (b) still true? [Hint: look for a counterexample.]
(d) If R ⊆ S is a subring and P R is a maximal ideal, suppose that Q S is a prime ideal
for which P ⊆ Q. Show that Q∩ R = P.
1.6. Let k be a ﬁeld, R be a ring with unit and let ϕ: k −→ R be a ring homomorphism. Show
that ϕ is a monomorphism.
1.7. Consider the sets
Z(i) = ¦u +vi : u, v ∈ Z¦ ⊆ C, Q(i) = ¦u +vi : u, v ∈ Q¦ ⊆ C.
(a) Show that Z(i) and Q(i) are subrings of C. Also show that Z(i) is an integral domain,
Q(i) is a ﬁeld and Z(i) is a subring of Q(i).
(b) Show that the inclusion homomorphism inc: Z(i) −→Q(i) extends to a monomorphism
inc
∗
: Fr(Z(i)) −→Q(i).
(c) Show that inc
∗
is an isomorphism, so Fr(Z(i)) = Q(i).
1.8. Let R be a commutative ring.
(a) If a, b ∈ R, show that there is a unique ring homomorphism ψ
a,b
: R[X] −→ R[X] for
which ψ
a,b
(r) = r if r ∈ R and ψ
a,b
(X) = aX +b. If c, d ∈ R, determine ψ
a,b
◦ ψ
c,d
. If
a is a unit, show that ψ
a,b
is an isomorphism and ﬁnd its inverse.
(b) Now suppose that R = k is a ﬁeld and a, b ∈ k with a = 0. Prove the following.
(i) If f(X) ∈ k[X], the deg ψ
a,b
(f(X)) = deg f(X).
(ii) If p(X) ∈ k[X] is a prime then so is ψ
a,b
(p(X)).
(iii) If p(X) ∈ k[X] is an irreducible then so is ψ
a,b
(p(X)).
1.9. Let k be a ﬁeld and k[[X]] be the set consisting of all power series
∞
¸
k=0
a
k
X
k
= a
0
+a
1
X + +a
k
X
k
+ ,
23
with a
k
∈ k.
(a) Show that this can be made into an integral domain containing k[X] as a subring by
deﬁning addition and multiplication in the obvious way.
(b) Show that
¸
∞
k=0
a
k
X
k
∈ k[[X]] is a unit if and only if a
0
= 0.
(c) Show that Fr(k[[X]]) consists of all ﬁnitetailed Laurent series
∞
¸
k=
a
k
X
k
= a
X
+a
+1
X
+1
+ +a
k
X
k
+
for some ∈ Z and a
k
∈ k.
1.10. Taking k = Q, ﬁnd the quotient and remainder when performing long division of f(X) =
6X
4
−6X
3
+ 3X
2
−3X −2 by d(X) = 2X
3
+X + 3.
1.11. Taking k = F
3
, ﬁnd the quotient and remainder when performing long division of
f(X) = 2X
3
+ 2X
2
+X + 1 by d(X) = 2X
3
+ 2X.
1.12. Let p > 0 be a prime. Suppose that f(X) = a
0
+ a
1
X + + a
n
X
n
∈ Z[X] with
p a
n
and that f(X) ∈ F
p
[X] denotes the polynomial obtained by reducing the coeﬃcients of
f(X) modulo p. If f(X) is irreducible, show that f(X) is irreducible. Which of the following
polynomials in Z[X] is irreducible?
X
3
−X + 1, X
3
+ 2X + 1, X
3
+X −1, X
5
−X + 1, X
5
+X −1, 5X
3
−10X +X
2
−2.
1.13. Find generators for each of the following ideals:
I
1
= ¦f(X) ∈ Q[X] : f(i) = 0¦ Q[X], I
2
= ¦f(X) ∈ Q[X] : f(
√
2 i) = 0¦ Q[X],
I
3
= ¦f(X) ∈ Q[X] : f(
√
2) = 0¦ Q[X], I
4
= ¦f(X) ∈ R[X] : f(
√
2) = 0¦ R[X],
I
5
= ¦f(X) ∈ R[X] : f(
√
2 i) = 0¦ R[X], I
6
= ¦f(X) ∈ R[X] : f(ζ
3
) = 0¦ R[X].
1.14. Consider the inclusion inc: Q −→C and its extension to ε
√
2
: Q[X] −→C.
Determine the image ε
√
2
Q[X] ⊆ C. What is ε
−
√
2
Q[X] ⊆ C? Find ker ε
√
2
Q[X] and
ker ε
−
√
2
Q[X]; are these maximal ideals?
1.15. Let ω = (−1 +
√
3i)/2 ∈ C. Consider the inclusion inc: Q −→ C and its extension
to ε
ω
: Q[X] −→ C. Determine the image ε
ω
Q[X] ⊆ C. Determine ker ε
ω
Q[X] and decide
whether it is maximal. Find another evaluation homomorphism with the same kernel and image.
1.16. Consider the inclusion inc: Q −→ C and its extension to ε
α
: Q[X] −→ C where α is
one of the 4 complex roots of the polynomial f(X) = X
4
− 2 ∈ Q[X]. Determine the image
ε
α
Q[X] ⊆ C and the ideal ker ε
α
Q[X]; is the latter ideal maximal? What happens if α is
replaced by one of the other roots of f(X)?
Repeat this problem starting with the inclusion of the real numbers into the complex num
bers inc: R −→C and ε
α
: R[X] −→C.
1.17. Use Cardan’s method to ﬁnd the complex roots of the polynomial
f(X) = X
3
−9X
2
+ 21X −5.
1.18. Consider the real numbers
α =
3
10 +
√
108 +
3
10 −
√
108, β =
3
1 +
2
3
7
3
+
3
1 −
2
3
7
3
.
Find rational cubic polynomials f(X) and g(X) for which f(α) = 0 = g(β). Hence determine
these real numbers.
24 1. INTEGRAL DOMAINS, FIELDS AND POLYNOMIAL RINGS
1.19. Prove the ﬁnal part of Example 1.60 by showing that there is an isomorphism of groups
Aﬀ
1
(k)
∼
= Aut
k
(k[X]).
1.20. Let k be any ﬁeld. Consider the 6 automorphisms α
j
: k(X) −→ k(X) (j = 1, . . . , 6)
deﬁned by
α
1
(f(X)) = f(X), α
2
(f(X)) = f(1 −X), α
3
(f(X)) = f(1/X),
α
4
(f(X)) = f((X −1)/X), α
5
(f(X)) = f(1/(1 −X)), α
6
(f(X)) = f(X/(X −1)).
Show that the set consisting of these elements is a subgroup Γ
k
Aut
k
(k(X)) isomorphic to
the symmetric group S
3
. When k = F
2
, show that Γ
k
∼
= GL
2
(k).
1.21. Determine the cyclotomic polynomial Φ
20
(X).
1.22. Let p > 0 be a prime.
(a) Show that for k 1, the cyclotomic polynomial Φ
p
k(X) satisﬁes
Φ
p
k(X) = Φ
p
(X
p
k−1
)
and has as its complex roots the primitive p
k
th roots of 1.
(b) Show that Φ
p
k(X) ∈ Q[X] is irreducible.
(c) Generalize part (a) to show that if n = p
r
1
1
p
r
k
k
is the prime power factorization of n
with the p
i
being distinct primes and r
i
> 0, then
Φ
n
(X) = Φ
p
1
p
k
(X
p
r
1
−1
1
p
r
k
−1
k
).
1.23. For n 2, show that
X
ϕ(n)
Φ
n
(X
−1
) = Φ
n
(X).
1.24. Show that for n 1, ζ
n
+ζ
−1
n
= 2 cos(2π/n).
Find expressions for ζ
5
+ζ
−1
5
and ζ
2
5
+ζ
−2
5
in terms of cos(2π/5). Hence ﬁnd a rational polynomial
which has cos(2π/5) as a root.
1.25. Let p > 0 be a prime and K be a ﬁeld with char K = p.
(a) Show that if ζ ∈ K is a pth root of 1 then ζ = 1. Deduce that if m, n > 0 and p n,
then every np
m
th root of 1 in K is an nth root of 1.
(b) If a ∈ K, show that the polynomial X
p
−a ∈ K[X] has either no roots or exactly one
root in K.
CHAPTER 2
Fields and their extensions
2.1. Fields and subﬁelds
2.1. Definition. Let K and L be ﬁelds and suppose that K ⊆ L is a subring. Then we
say that K is a subﬁeld of L; L is also said to be an extension (ﬁeld) of K. We write K L or
L/K to indicate this, and write K < L if K is a proper subﬁeld of L, i.e., if K = L.
An important fact about an extension of ﬁelds L/K is that L is a Kvector space whose
addition is the addition in the ﬁeld L while scalar multiplication is deﬁned by
u x = ux (u ∈ K, x ∈ L).
2.2. Definition. We will call dim
K
L the degree or index of the extension L/K and use the
notation [L : K] = dim
K
L. An extension of ﬁelds L/K is ﬁnite (dimensional ) if [L : K] < ∞,
otherwise it is inﬁnite (dimensional ).
2.3. Example. Show that the extension C/R is ﬁnite, while R/Q and C/Q are both inﬁnite.
Solution. We have
C = ¦x +yi : x, y ∈ R¦,
so 1, i span C as a vector space over R. Since i / ∈ R, these elements are also linearly independent
over R and therefore they form a basis, whence [C : R] = 2. The inﬁniteness of R/Q and C/Q are
consequences of the fact that any ﬁnite dimensional vector space over Q is countable, however
R and C are uncountable. A basis for the Qvector space R is known as a Hamel basis.
2.4. Example. Consider the extension Q(
√
2)/Q where
Q(
√
2) = ¦x +y
√
2 : x, y ∈ Q¦.
Show that [Q(
√
2) : Q] = 2.
Solution. The elements 1,
√
2 clearly span the Qvector space Q(
√
2). Now recall that
√
2 / ∈ Q. If the elements 1,
√
2 were linearly dependent we would have u + v
√
2 = 0 for some
u, v ∈ Q not both zero; in fact it is easy to see that we would then also have u, v both nonzero.
Thus we would have
√
2 = −
u
v
∈ Q,
which we know to be false. Hence 1,
√
2 are linearly independent and so form a basis for Q(
√
2)
over Q and [Q(
√
2) : Q] = 2.
If we have two extensions L/K and M/L then it is a straightforward to verify that K M
and so we have another extension M/K.
2.5. Definition. Given two extensions L/K and M/L, we say that L/K is a subextension
of M/K and sometimes write L/K M/K.
2.6. Theorem. Let L/K be a subextension of M/K.
(i) If one or both of the dimensions [L : K] or [M : L] is inﬁnite then so is [M : K].
25
26 2. FIELDS AND THEIR EXTENSIONS
(ii) If the dimensions [L : K] and [M : L] are both ﬁnite then so is [M : K] and
[M : K] = [M : L] [L : K].
Proof. (i) If [M : K] is ﬁnite, choose a basis m
1
, . . . , m
r
of M over K. Now any element
u ∈ M can be expressed as
u = t
1
m
1
+ +t
r
m
r
,
where t
1
, . . . , t
r
∈ K; but since K ⊆ L, this means that m
1
, . . . , m
r
spans M over L and so
[M : L] < ∞. Also L is a Kvector subspace of the ﬁnite dimensional Kvector space M, hence
[L : K] < ∞.
(ii) Setting r = [L : K] and s = [M : L], choose a basis
1
, . . . ,
r
of L over K and a basis
m
1
, . . . , m
s
of M over L.
Now let v ∈ M. Then there are elements y
1
, . . . , y
s
∈ L for which
v = y
1
m
1
+ +y
s
m
s
.
But each y
j
can be expressed in the form
y
j
= x
1j
1
+ +x
rj
r
for suitable x
ij
∈ K. Hence,
v =
s
¸
j=1
r
¸
i=1
x
ij
i
m
j
=
s
¸
j=1
r
¸
i=1
x
ij
(
i
m
j
),
where each coeﬃcient x
ij
is in K. Thus the elements
i
m
j
(i = 1, . . . , r, j = 1, . . . , s) span the
Kvector space M.
Now suppose that for some t
ij
∈ K we have
s
¸
j=1
r
¸
i=1
t
ij
(
i
m
j
) = 0.
On collecting terms we obtain
s
¸
j=1
r
¸
i=1
t
ij
i
m
j
= 0,
where each coeﬃcient
¸
r
i=1
t
ij
i
is in L. By the linear independence of the m
j
over L, this
means that for each j,
r
¸
i=1
t
ij
i
= 0.
By the linear independence of the
i
over K, each t
ij
= 0.
Hence the
i
m
j
form a basis of M over K and so
[M : K] = rs = [M : L] [L : K].
We will often indicate subextensions in diagrammatic form where larger ﬁelds always go
above smaller ones and the information on the lines indicates dimensions
M
[M:L]
[M:K]=[M:L] [L:K]
L
[L:K]
K
We often suppress ‘composite’ lines such as the dashed one. Such towers of extensions are our
main objects of study. We can build up sequences of extensions and form towers of arbitrary
2.2. SIMPLE AND FINITELY GENERATED EXTENSIONS 27
length. Thus, if L
1
/K, L
2
/L
1
, . . . , L
k
/L
k−1
is a such a sequence of extensions, there is a
diagram
L
k
L
k−1
L
1
K
2.2. Simple and ﬁnitely generated extensions
2.7. Definition. Let F be a ﬁeld and K F. Given elements u
1
, . . . , u
r
∈ F we set
K(u
1
, . . . , u
r
) =
¸
KLF
u
1
,...,u
r
∈L
L
which is the smallest subﬁeld in F that contains K and the elements u
1
, . . . , u
r
. The ex
tension K(u
1
, . . . , u
r
)/K is said to be generated by the elements u
1
, . . . , u
r
; we also say that
K(u
1
, . . . , u
r
)/K is a ﬁnitely generated extension of K. An extension of the form K(u)/K is
called a simple extension of K with generator u.
We can extend this to the case of an inﬁnite sequence u
1
, . . . , u
r
, . . . in F and denote by
K(u
1
, . . . , u
r
, . . .) F the smallest extension ﬁeld of K containing all the elements u
r
.
It can be shown that
(2.1) K(u
1
, . . . , u
r
) =
f(u
1
, . . . , u
r
)
g(u
1
, . . . , u
r
)
∈ F : f(X
1
, . . . , X
r
), g(X
1
, . . . , X
r
) ∈ K[X
1
, . . . , X
r
], g(u
1
, . . . , u
r
) = 0
.
Reordering the u
i
does not change K(u
1
, . . . , u
n
).
2.8. Proposition. Let K(u)/K and K(u, v)/K(u) be simple extensions. Then
K(u, v) = K(u)(v) = K(v)(u).
More generally,
K(u
1
, . . . , u
n
) = K(u
1
, . . . , u
n−1
)(u
n
)
and this is independent of the order of the sequence u
1
, . . . , u
n
.
2.9. Theorem. For a simple extension K(u)/K, exactly one of the following conditions
holds.
(i) The evaluation at u homomorphism ε
u
: K[X] −→ K(u) is a monomorphism and on
passing to the fraction ﬁeld gives an isomorphism (ε
u
)
∗
: K(X) −→ K(u). In this case,
K(u)/K is inﬁnite and u is said to be transcendental over K.
(ii) The evaluation at u homomorphism ε
u
: K[X] −→ K(u) has a nontrivial kernel
ker ε
u
= (p(X)) where p(X) ∈ K[X] is an irreducible monic polynomial of positive de
gree and the quotient homomorphism ¯ ε
u
: K[X]/(p(X)) −→ K(u) is an isomorphism.
In this case K(u)/K is ﬁnite with [K(u) : K] = deg p(X) and u is said to be algebraic
over K.
28 2. FIELDS AND THEIR EXTENSIONS
Proof. (i) If ker ε
u
= (0), all that needs checking is that (ε
u
)
∗
is an epimorphism; but as
u is in the image of (ε
u
)
∗
this is obvious.
(ii) When ker ε
u
= (0), Theorem 1.31(iv) implies that the image of ε
u
is a subﬁeld of K(u) and
since it contains u it must equal K(u). Hence ¯ ε
u
is an isomorphism. Using Long Division, we
ﬁnd that every element of K[X]/(p(X)) can be uniquely expressed as a coset of the form
f(X) + (p(X)),
where deg f(X) < deg p(X). Hence every element of K[X]/(p(X)) can be uniquely expressed
as a linear combination over K of the d cosets
1 + (p(X)), X + (p(X)), X
2
+ (p(X)), . . . , X
d−1
+ (p(X)),
where d = deg p(X). Via the isomorphism ¯ ε
u
under which ¯ ε
u
(X
k
+ (p(X))) = u
k
, we see that
the elements 1, u, . . . , u
d−1
form a basis for K(u) over K.
2.10. Example. For the extension Q(
√
2,
√
3)/Q we have [Q(
√
2,
√
3) : Q] = 4.
Proof. By Example 2.4 we know that [Q(
√
2) : Q] = 2. We have the following tower of
extensions.
Q(
√
2,
√
3)
[Q(
√
2,
√
3):Q(
√
2)]
[Q(
√
2,
√
3):Q]=2[Q(
√
2,
√
3):Q(
√
2)]
Q(
√
2)
2
Q
We will show that [Q(
√
2,
√
3) : Q(
√
2)] = 2.
Notice that if u ∈ Q(
√
2,
√
3) = Q(
√
2)(
√
3) then u = a + b
√
3 for some a, b ∈ Q(
√
2),
so 1,
√
3 span Q(
√
2,
√
3) over Q(
√
2). But if these are linearly dependent then
√
3 ∈ Q(
√
2).
Writing
√
3 = v +w
√
2
with v, w ∈ Q, we ﬁnd that
v
2
+ 2w
2
+ 2vw
√
2 = 3 ∈ Q,
and hence 2vw
√
2 ∈ Q. The possibilities v = 0 or w = 0 are easily ruled out, while v, w = 0
would implies that
√
2 ∈ Q which is false. So 1,
√
3 are linearly independent over Q(
√
2)
and therefore form a basis of Q(
√
2,
√
3). This shows that [Q(
√
2,
√
3) : Q(
√
2)] = 2 and so
[Q(
√
2,
√
3) : Q] = 4.
2.11. Remark. There are some other subﬁelds of Q(
√
2,
√
3) which are conveniently dis
played in the following diagram.
Q(
√
2,
√
3)
2
2
2
Q(
√
2)
2
Q(
√
3)
2
Q(
√
6)
2
Q
One idea in the veriﬁcation of Example 2.10 can be extended to provide a useful general
result whose proof is left as an exercise.
2.2. SIMPLE AND FINITELY GENERATED EXTENSIONS 29
2.12. Proposition. Let p
1
, . . . , p
n
be a sequence of distinct primes p
i
> 0. Then
√
p
n
/ ∈ Q(
√
p
1
, . . . ,
√
p
n−1
).
Hence [Q(
√
p
1
, . . . ,
√
p
n
) : Q(
√
p
1
, . . . ,
√
p
n−1
)] = 2 and [Q(
√
p
1
, . . . ,
√
p
n
) : Q] = 2
n
.
2.13. Example. For the extension Q(
√
2, i)/Q we have [Q(
√
2, i) : Q] = 4.
Proof. We know that [Q(
√
2) : Q] = 2. Also, i / ∈ Q(
√
2) since i is not real and Q(
√
2) R.
Since i
2
+ 1 = 0, we have Q(
√
2, i) = Q(
√
2)(i) and [Q(
√
2, i) : Q(
√
2)] = 2. Using the formula
[Q(
√
2, i) : Q] = [Q(
√
2, i) : Q(
√
2)] [Q(
√
2) : Q],
we obtain [Q(
√
2, i) : Q] = 4.
This example also has several other subﬁelds, with only Q(
√
2) = Q(
√
2, i) ∩ R being a
subﬁeld of R.
C
2
∞ R
∞ Q(
√
2, i)
2
2
2
Q(
√
2)
2
Q(i)
2
Q(
√
2 i)
2
Q
2.14. Example. For n 1, let E
n
= Q(2
1/n
) R, where 2
1/n
∈ R denotes the positive real
nth root of 2.
(i) Show that [E
n
: Q] = n.
(ii) If m 1 with m [ n, show that E
m
E
n
and determine [E
n
: E
m
].
(iii) If m, n are coprime, show that E
mn
= Q(2
1/m
, 2
1/n
).
Solution. (i) Consider the evaluation homomorphism ε
2
1/n : Q[X] −→ E
n
. Applying the
Eisenstein Test 1.38 using the prime 2 to the polynomial X
n
−2 ∈ Z[X], we ﬁnd that
ker ε
2
1/n = (X
n
−2) Q[X],
and the induced homomorphism ¯ ε
2
1/n : Q[X]/(X
n
− 2) −→ E
n
is an isomorphism. Hence
[E
n
: Q] = n.
(ii) Since n/m is an integer,
2
1/m
= (2
1/n
)
n/m
∈ E
n
,
so
E
m
= Q(2
1/m
) ⊆ E
n
.
By Theorem 2.6 we have
n = [E
n
: Q] = [E
n
: E
m
] [E
m
: Q] = m[E
n
: E
m
],
whence [E
n
: E
m
] = n/m.
(iii) By (ii) we have E
m
E
mn
and E
n
E
mn
, hence Q(2
1/m
, 2
1/n
) E
mn
. As gcd(m, n) = 1,
there are integers r, s for which rm+sn = 1 and so
1
mn
=
rm+sn
mn
=
r
n
+
s
m
.
30 2. FIELDS AND THEIR EXTENSIONS
This shows that
2
1/mn
= (2
1/n
)
r
(2
1/m
)
s
∈ Q(2
1/m
, 2
1/n
),
whence E
mn
Q(2
1/m
, 2
1/n
). Combining these inclusions we obtain E
mn
= Q(2
1/m
, 2
1/n
).
Exercises on Chapter 2
2.1. Let p ∈ N be an prime. Show that the extension Q(
√
p)/Q has [Q(
√
p) : Q] = 2.
2.2. Let p, q > 0 be distinct primes. Show that [Q(
√
p,
√
q) : Q(
√
p)] = 2.
2.3. Prove Proposition 2.12 by induction on n.
2.4. Let K a ﬁeld with char K = 2 and suppose that L/K is an extension. If a, b ∈ K are
distinct, suppose that u, v ∈ L satisfy u
2
= a and v
2
= b. Show that K(u, v) = K(u +v).
[Hint: ﬁrst show that u±v = 0 and deduce that u−v ∈ K(u+v); then show that u, v ∈ K(u+v).]
2.5. Show that [Q(i) : Q] = 2.
2.6. Show that [Q(
√
3, i) : Q] = 4. Find the three subﬁelds L Q(
√
3, i) with [L : Q] = 2 and
display their relationship in a diagram, indicating which ones are subﬁelds of R.
2.7. Let ζ
5
= e
2πi/5
∈ C.
(a) Explain why [Q(ζ
5
) : Q] = 4.
(b) Show that cos(2π/5), sin(2π/5) i ∈ Q(ζ
5
).
(c) Show that for t ∈ R,
cos 5t = 16 cos
5
t −20 cos
3
t + 5 cos t.
(d) Show that the numbers cos(2kπ/5) with k = 0, 1, 2, 3, 4 are roots of the polynomial
f(X) = 16X
5
−20X
3
+ 5X −1 = (X −1)(4X
2
+ 2X −1)
2
and deduce that [Q(cos(2π/5)) : Q] = 2.
(e) Display the relationship between the ﬁelds Q, Q(cos(2π/5)), and Q(ζ
5
) in a suitable
diagram.
2.8. This question is for those who like lots of calculation or using Maple. Let ζ
7
= e
2πi/7
∈ C.
(a) Explain why [Q(ζ
7
) : Q] = 6.
(b) Show that cos(2π/7), sin(2π/7) i ∈ Q(ζ
7
).
(c) Show
cos 7t = 64 cos
7
t −112 cos
5
t + 56 cos
3
t −7 cos t.
Show that the numbers cos(2kπ/7) with k = 0, 1, . . . , 6 are roots of the polynomial
f(X) = 64X
7
−112X
5
+ 56X
3
−7X −1 = (X −1)(8X
3
+ 4X
2
−4X −1)
2
and deduce that [Q(cos(2π/7)) : Q] = 3.
(d) Show that sin(2π/7) i is a root of
g(X) = 64X
7
+ 112X
5
+ 56X
3
+ 7X = X(64X
6
+ 112X
4
+ 56X
2
+ 7)
and that 64X
6
+112X
4
+56X
2
+7 ∈ Q[X] is irreducible. What is [Q(sin(2π/7) i) : Q]?
(e) Display the relationship between the ﬁelds Q, Q(cos(2π/7)), Q(sin(2π/7) i) and Q(ζ
7
)
in a diagram.
(f) Is i ∈ Q(ζ
7
)?
2.9. In this question we continue to consider the situation described in Example 2.14.
31
(a) Show that
Aut
Q
(E
n
) =
¦id¦ if n is odd,
¦id, τ
n
¦
∼
= Z/2 if n is even,
where τ
n
has composition order 2.
(b) Let E =
¸
n1
E
n
R. Show that Aut
Q
(E) = ¦id¦.
(c) Display the 6 subﬁelds of E
12
in a diagram.
(d) Which of the subﬁelds in part (c) contain the element 2
1/2
+ 2
1/3
?
CHAPTER 3
Algebraic extensions of ﬁelds
3.1. Algebraic extensions
Let L/K be an extension of ﬁelds. From Theorem 2.9(ii), recall the following notion.
3.1. Definition. An element t ∈ L is algebraic over K if there is a nonzero polynomial
p(X) ∈ K[X] for which p(t) = 0.
Notice in particular that for an element t ∈ K, the polynomial p(X) = X − t ∈ K[X]
satisﬁes p(t) = 0, so t is algebraic over K.
Theorem 2.9 allows us to characterize algebraic elements in other ways.
3.2. Proposition. Let t ∈ L. Then the following conditions are equivalent.
(i) t is algebraic over K.
(ii) The evaluation homomorphism ε
t
: K[X] −→ L has nontrivial kernel.
(iii) The extension K(t)/K is ﬁnite dimensional.
3.3. Definition. If t ∈ L is algebraic over K then by Proposition 3.2,
ker ε
t
= (minpoly
K,t
(X)) = (0),
where minpoly
K,t
(X) ∈ K[X] is an irreducible monic polynomial called the minimal polynomial
of t over K. The degree of minpoly
K,t
(X) is called the degree of t over K and is denoted
deg
K
t.
3.4. Proposition. If t ∈ L is algebraic over K then
[K(t) : K] = deg minpoly
K,t
(X) = deg
K
t.
Proof. This follows from Theorem 2.9(ii).
3.5. Remark. Suppose that t ∈ L is algebraic over K and that p(X) ∈ ker ε
t
with
deg p(X) = deg minpoly
K,t
(X). Then minpoly
K,t
(X) [ p(X) and so
p(X) = uminpoly
K,t
(X)
for some u ∈ K. In particular, when p(X) is monic,
p(X) = minpoly
K,t
(X).
We will often use this without further comment.
3.6. Example. Consider C/Q. The minimal polynomial of
√
2 ∈ C over Q is
minpoly
Q,
√
2
(X) = X
2
−2.
Proof. Clearly X
2
−2 ∈ ker ε
√
2
since (
√
2)
2
−2 = 0. By Example 2.4,
deg minpoly
Q,
√
2
(X) = [Q(
√
2) : Q] = 2,
hence
minpoly
Q,
√
2
(X) = X
2
−2.
3.7. Example. Consider C/Q. The minimal polynomial of i ∈ C over Q is X
2
+ 1.
33
34 3. ALGEBRAIC EXTENSIONS OF FIELDS
Proof. Clearly X
2
+ 1 ∈ ker ε
i
since i
2
+ 1 = 0. As [Q(i) : Q] = 2, we have
minpoly
Q,i
(X) = X
2
+ 1.
3.8. Example. Consider C/Q. Find the minimal polynomial of the primitive 6th root of
unity, ζ
6
∈ C over Q.
Solution. Recall from Example 1.44 that ζ
6
is a root of the irreducible cyclotomic poly
nomial
Φ
6
(X) = X
2
−X + 1.
Then Φ
6
(X) ∈ ker ε
ζ
6
so minpoly
Q,ζ
6
(X) [ Φ
6
(X). Since Φ
6
(X) is irreducible and monic, we
must have
minpoly
Q,ζ
6
(X) = Φ
6
(X)
and so deg
Q
ζ
6
= 2.
3.9. Example. Consider C/Q. Find the minimal polynomial of
√
2 +
√
3 over Q.
Solution. Notice that
√
3 −
√
2 =
(
√
3 −
√
2)(
√
3 +
√
2)
(
√
3 +
√
2)
=
1
√
2 +
√
3
∈ Q(
√
2 +
√
3).
So we have
√
2 =
1
2
(
√
2 +
√
3) −(
√
3 −
√
2)
∈ Q(
√
2 +
√
3),
√
3 =
1
2
(
√
2 +
√
3) + (
√
3 −
√
2)
∈ Q(
√
2 +
√
3),
hence Q(
√
2,
√
3) Q(
√
2 +
√
3). Since Q(
√
2 +
√
3) Q(
√
2,
√
3) we must have
Q(
√
2 +
√
3) = Q(
√
2,
√
3).
Referring to Example 2.10 we see that
deg
Q
(
√
2 +
√
3) = 4.
Let us ﬁnd a nonzero polynomial in ker ε
√
2+
√
3
Q[X].
Referring to Example 2.10 or Proposition 2.12 we see that
√
2 +
√
3 / ∈ Q(
√
2), hence
deg
Q(
√
2)
(
√
2 +
√
3) = 2.
One polynomial in ker ε
√
2+
√
3
Q(
√
2)[X] is
(X −(
√
2 +
√
3))(X −(
√
2 −
√
3)) = X
2
−2
√
2X −1.
Since this is monic and of degree 2,
minpoly
Q(
√
2),
√
2+
√
3
(X) = X
2
−2
√
2X −1.
Similarly,
minpoly
Q(
√
2),−
√
2+
√
3
(X) = X
2
+ 2
√
2X −1.
Consider
p(X) = minpoly
Q(
√
2),
√
2+
√
3
(X) minpoly
Q(
√
2),−
√
2+
√
3
(X)
= (X
2
−2
√
2X −1)(X
2
+ 2
√
2X −1)
= X
4
−10X
2
+ 1.
Then p(
√
2 +
√
3) = 0 so p(X) ∈ ker ε
t
. Since deg p(X) = 4 and p(X) is monic, we have
minpoly
Q,
√
2+
√
3
(X) = X
4
−10X
2
+ 1.
3.1. ALGEBRAIC EXTENSIONS 35
3.10. Definition. Let L/K be a ﬁnite extension. An element u ∈ L for which L = K(u) is
called a primitive element for the extension L/K.
Later we will see that when char K = 0 every ﬁnite extension L/K has a primitive element.
3.11. Lemma. Let L/K be a ﬁnite extension and u ∈ L. Then u is a primitive element for
L/K if and only if deg
K
u = [L : K].
Proof. K(u) ⊆ L is a ﬁnite dimensional Kvector subspace. Then K(u) = L if and only
dim
K
K(u) = dim
K
L. Since deg
K
u = dim
K
K(u) and [L : K] = dim
K
L the result follows.
Sometimes the minimal polynomial of an element in an extension is introduced in a diﬀerent
but equivalent way.
3.12. Proposition. Let t ∈ L be algebraic over K. Then
I(t) = ¦f(X) ∈ K[X] : f(t) = 0¦ ⊆ K[X]
is an ideal which is principal and has an irreducible monic generator q(X) ∈ K[X]. In fact,
q(X) = minpoly
K,t
(X).
Proof. It is easy to see that I(t) K[X] and therefore I(t) = (q(X)) for some monic
generator q(X). To see that q(X) is irreducible, suppose that q(X) = q
1
(X)q
2
(X) with
deg q
i
(X) < deg q(X). Now as q
1
(t)q
2
(t) = 0, we must have q
1
(t) = 0 or q
2
(t) = 0, hence
q
1
(X) ∈ I(t) or q
2
(X) ∈ I(t). These possibilities give q(X) [ q
1
(X) or q(X) [ q
2
(X) and
so deg q(X) deg q
1
(X) or deg q(X) deg q
2
(X), contradicting the above assumption that
deg q
i
(X) < deg q(X).
The irreducible monic polynomial minpoly
K,t
(X) is in I(t) so q(X) [ minpoly
K,t
(X) and
therefore q(X) = minpoly
K,t
(X).
The next Lemma will often be useful.
3.13. Lemma. Let L/K be an extension and suppose that u
1
, . . . , u
n
∈ L are algebraic. Then
K(u
1
, . . . , u
n
)/K is a ﬁnite extension.
Proof. Use induction on n together with Proposition 2.8 and Theorem 2.6(ii).
We now come to an important notion for extensions.
3.14. Definition. The extension L/K is algebraic or L is algebraic over K if every element
t ∈ L is algebraic over K.
3.15. Proposition. Let L/K be a ﬁnite extension. Then L/K is algebraic.
Proof. Let t ∈ L. Since the Kvector space L is ﬁnite dimensional, when viewed as
elements of this vector space, the powers 1, t, . . . , t
n
, . . . must be linearly dependent over K.
Hence for suitable coeﬃcients c
j
∈ K not all zero and some m 1 we have
c
0
+c
1
t + +c
m
t
m
= 0.
But this means that t is algebraic over K.
3.16. Proposition. Let M/L and L/K be algebraic extensions. Then the extension M/K
is algebraic.
Proof. Let u ∈ M. Then u is algebraic over L, so there is a polynomial
p(X) = p
0
+p
1
X + +p
m
X
m
∈ L[X]
of positive degree with p(u) = 0. By Lemma 3.13, the extension K(p
0
, . . . , p
m
)/K is ﬁnite and
so is K(p
0
, . . . , p
m
, u)/K(p
0
, . . . , p
m
). By Theorem 2.6(ii), K(p
0
, . . . , p
m
, u)/K is ﬁnite, so by
Proposition 3.15, u is algebraic over K.
36 3. ALGEBRAIC EXTENSIONS OF FIELDS
3.17. Definition. For an extension L/K, let
L
alg
= ¦t ∈ L : t is algebraic over K¦ ⊆ L.
3.18. Proposition. For an extension L/K, L
alg
is a subﬁeld containing K and L
alg
/K is
algebraic.
Proof. Clearly K ⊆ L
alg
. We must show that L
alg
L.
Let u, v ∈ L
alg
. Then by Lemma 3.13, K(u, v)/K is a ﬁnite dimensional extension, hence
every element of K(u, v) is algebraic over K. In particular, u + v and uv are in K(u, v) and if
u = 0, u
−1
is also in K(u, v). Therefore u +v, uv and u
−1
are all algebraic over K.
3.19. Example. In the extension C/Q we can consider C
alg
C which is called the subﬁeld
of algebraic numbers. Similarly, in the extension R/Q the subﬁeld
R
alg
= C
alg
∩ R C
consists of all the real algebraic numbers. Elements of C−C
alg
are called transcendental complex
numbers; examples are e and π. The sets C
alg
and R
alg
are both countable, whereas C and R
are uncountable, so there are in fact many more transcendental numbers but it can be hard to
determine whether a given number is transcendental or not. A more usual notation for C
alg
is Q since this is the algebraic closure of Q which will be discussed later. When dealing with
algebraic extensions of Q we will usually work with subﬁelds of Q = C
alg
.
We end this section with a technical result.
3.20. Proposition. Let K(u)/K be a ﬁnite simple extension. Then there are only ﬁnitely
many subextensions F/K K(u)/K.
Proof. Consider the minimal polynomial minpoly
K,u
(X) ∈ K[X]. Now for any subexten
sion F/K K(u)/K we can also consider
minpoly
F,u
(X) = c
0
+c
1
X + +c
k−1
X
k−1
+X
k
∈ F[X],
which divides minpoly
K,u
(X) in F[X]. The Unique Factorization Property 1.33 implies that
minpoly
K,u
(X) has only ﬁnitely many monic divisors in K(u)[X], so there are only a ﬁnite
number of possibilities for minpoly
F,u
(X). Now consider F
0
= K(c
0
, c
1
, . . . , c
k−1
), the extension
ﬁeld of K generated by the coeﬃcients of minpoly
F,u
(X). Then F
0
F and so minpoly
F,u
(X) ∈
F
0
[X] is irreducible since it is irreducible in F[X]; hence minpoly
F,u
(X) = minpoly
F
0
,u
(X). We
have
[K(u) : F] = deg minpoly
F,u
(X) = deg minpoly
F
0
,u
(X) = [K(u) : F
0
],
hence F = F
0
.
This shows that there are only ﬁnitely many subextensions F/K K(u)/K, each of which
has the form K(a
0
, a
1
, . . . , a
−1
), where
a
0
+a
1
X + +a
−1
X
−1
+X
∈ K(u)[X]
is a factor of minpoly
K,u
(X) in K(u)[X].
3.2. Splitting ﬁelds and Kronecker’s Theorem
We can now answer a basic question. Let K be a ﬁeld and p(X) ∈ K[X] be a polynomial
of positive degree.
3.21. Question. Is there an extension ﬁeld L/K for which p(X) has a root in L?
A stronger version of this question is the following.
3.2. SPLITTING FIELDS AND KRONECKER’S THEOREM 37
3.22. Question. Is there an extension ﬁeld E/K for which p(X) factorizes into linear factors
in E[X]?
3.23. Definition. p(X) ∈ K[X] splits in E/K or over E if it factorizes into linear factors
in E[X].
Of course, if we have such a ﬁeld E then the distinct roots u
1
, . . . , u
k
of p(X) in E generate
a subﬁeld K(u
1
, . . . , u
k
) E which is the smallest subﬁeld of E that answers Question 3.22.
3.24. Definition. Such a minimal extension of K is called a splitting ﬁeld of p(X) over K
and we will sometimes denote it by K(p(X)) or K
p
.
We already know how to answer Question 3.21.
3.25. Theorem (Kronecker’s Theorem: ﬁrst version). Let K be a ﬁeld and p(X) ∈ K[X]
be a polynomial of positive degree. Then there is a ﬁnite extension L/K for which p(X) has a
root in L.
Proof. We begin by factorizing p(X) ∈ K[X] into irreducible monic factors q
j
(X) together
with a constant factor c:
p(X) = cq
1
(X) q
r
(X).
Now for any j we can form the quotient ﬁeld K[x]/(q
j
(X)) which is a ﬁnite dimensional (simple)
extension of K and in which the coset X + (q
j
(X)) satisﬁes the equation
q
j
(X + (q
j
(X))) = 0 + (q
j
(X)).
Hence p(X) has a root in K[x]/(q
j
(X)).
Of course, this construction is only interesting if q
j
(X) to has degree bigger than 1 since a
linear polynomial already has a root in K.
To answer Question 3.22 we iterate this construction. Namely, having found one root u
1
in
an extension L
1
/K we discard the linear factor X −u
1
and consider the polynomial
p
1
(X) =
p(X)
X −u
1
∈ L
1
[X].
We can repeat the argument to form a ﬁnite extension of L
1
(and hence of K) containing a
root of p
1
(X) and so on. At each stage we either already have another root in L
1
or we need to
enlarge the ﬁeld to obtain one.
3.26. Theorem (Kronecker’s Theorem: second version). Let K be a ﬁeld and p(X) ∈ K[X]
be a polynomial of positive degree. Then there is a ﬁnite extension E/K which is a splitting
ﬁeld of p(X) over K.
In practise we often have extension ﬁelds ‘lying around in nature’ containing roots and we
can work inside of these. When working over Q (or any other subﬁeld of C) we can always ﬁnd
roots in C by the Fundamental Theorem of Algebra. We then refer to a subﬁeld of C which is
a splitting ﬁeld as the splitting subﬁeld.
3.27. Example. Find a splitting ﬁeld E/Q for p(X) = X
4
−4 over Q and determine [E : Q].
Solution. Notice that
p(X) = (X
2
−2)(X
2
+ 2),
so ﬁrst we adjoin the roots ±
√
2 of (X
2
− 2) to form Q(
√
2, −
√
2) = Q(
√
2) which gives an
extension Q(
√
2)/Q of degree 2.
38 3. ALGEBRAIC EXTENSIONS OF FIELDS
Next consider the polynomial X
2
+ 2 ∈ Q(
√
2)[X]. The complex roots of X
2
+ 2 are ±
√
2i
and these are not real, so this polynomial is irreducible in Q(
√
2)[X]. Hence we need to consider
Q(
√
2,
√
2i) = Q(
√
2, i) and the extension Q(
√
2, i)/Q(
√
2) which has degree 2.
C
∞
Q(
√
2, i)
adjoin roots of X
2
+ 2 2
Q(
√
2)
adjoin roots of X
2
− 2 2
Q
Thus the splitting subﬁeld of p(X) over Q in C is Q(
√
2, i) and [Q(
√
2, i) : Q] = 4.
Of course we could have started by ﬁrst adjoining roots of X
2
+2 and then adjoining roots
of X
2
−2, thus giving the tower
C
∞
Q(
√
2, i)
adjoin roots of X
2
− 2 2
Q(
√
2i)
adjoin roots of X
2
+ 2 2
Q
An important point is that if a splitting ﬁeld exists inside of a given extension ﬁeld F/K, it is
unique as a subﬁeld of F.
3.28. Proposition. Let F/K be an extension ﬁeld and p(X) ∈ K[X]. If E
1
, E
2
F are
splitting subﬁelds for p(X) over K then E
1
= E
2
.
Proof. Let u
1
, . . . , u
k
∈ F be the distinct roots of p(X) in F. By deﬁnition, K(u
1
, . . . , u
k
)
is the smallest subﬁeld containing K and all the u
j
. But K(u
1
, . . . , u
k
) must be contained in
any splitting subﬁeld, so E
1
= K(u
1
, . . . , u
k
) = E
2
.
Since we will frequently encounter quadratic polynomials we record a useful result on roots
of such polynomials. Recall that p(X) = aX
2
+ bX + c ∈ K[X] is quadratic if a = 0 and its
discriminant is
∆ = b
2
−4ac ∈ K.
The proof of the next result is the standard one which works provided 2 has an inverse in K,
i.e., when char K = 2.
3.29. Proposition. Let K be a ﬁeld of characteristic diﬀerent from 2. Then the quadratic
polynomial p(X) = aX
2
+bX +c ∈ K[X] has
• no roots in K if ∆ is not a square in K;
• one root −b/(2a) = −(2a)
−1
b if ∆ = 0;
3.3. MONOMORPHISMS BETWEEN EXTENSIONS 39
• two distinct roots
−b +δ
2a
= (2a)
−1
(−b +δ),
−b −δ
2a
= (2a)
−1
(−b −δ),
if ∆ = δ
2
for some nonzero δ ∈ K.
In particular, the splitting ﬁeld of p(X) over K is K if ∆ is a square in K and K(δ) otherwise,
where δ is one of the two square roots of ∆ in some extension of K such as the algebraic closure
K which we will introduce in Section 3.4.
3.30. Example. Find a splitting ﬁeld E/Q for p(X) = X
3
−2 over Q and determine [E : Q].
Solution. By the Eisenstein Test 1.38, p(X) is irreducible over Q. One root of p(X) is
3
√
2 ∈ R so we adjoin this to Q to form an extension Q(
3
√
2)/Q of degree 3. Now
p(X) = (X −
3
√
2)(X
2
+
3
√
2X + (
3
√
2)
2
)
and the second factor has the nonreal complex roots
3
√
2 ζ
3
,
3
√
2 ζ
2
3
lying in the extension
Q(
3
√
2, ζ
3
)/Q(
3
√
2) of degree 2. So the splitting subﬁeld of X
3
− 2 in C over Q is Q(
3
√
2, ζ
3
)
with [Q(
3
√
2, ζ
3
) : Q] = 6.
An alternative strategy would have been to adjoin one of the other roots
3
√
2 ζ
3
or
3
√
2 ζ
2
3
ﬁrst. We could also have begun by adjoining ζ
3
to form the extension Q(ζ
3
)/Q, but none of
the roots of p(X) lie in this ﬁeld so the extension Q(
3
√
2, ζ
3
)/Q(ζ
3
) of degree 3 is obtained by
adjoining one and hence all of the roots.
Figure 3.1 shows all the subﬁelds of the extension Q(
3
√
2, ζ
3
)/Q.
C
∞
2
R
∞ Q(
3
√
2, ζ
3
)
Q(
3
√
2)
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
Q(
3
√
2 ζ
3
)
2
3
Q(
3
√
2 ζ
2
3
)
2
3
Q(ζ
3
)
3
2
Q
Figure 3.1. The subﬁelds of Q(
3
√
2, ζ
3
)/Q
3.3. Monomorphisms between extensions
3.31. Definition. For extensions F/K and L/K, let Mono
K
(L, F) denote the set of all
monomorphisms L −→ F which ﬁx the elements of K.
3.32. Remark. We always have Aut
K
(F) ⊆ Mono
K
(F, F) and Mono
K
(F, F) is closed under
composition but is not always a group since elements are not necessarily invertible. If F/K is
ﬁnite, then we do have Mono
K
(F, F) = Aut
K
(F) since every injective Klinear transformation
is surjective and so invertible.
40 3. ALGEBRAIC EXTENSIONS OF FIELDS
We will also use the following notation.
3.33. Definition. Let F/K be an extension and p(X) ∈ K[X]. Set
Roots(p, F) = ¦u ∈ F : p(u) = 0¦,
the set of roots of p(X) in F. This is always a ﬁnite set which may of course be empty, which
happens precisely when p(X) has no root in F.
Suppose that p(X) ∈ K[X] is an irreducible polynomial which we might as well assume is
monic, and let F/K be an extension. Then if t ∈ F is a root of p(X), the evaluation homomor
phism ε
t
: K[X] −→ F factors through the quotient monomorphism ¯ε
t
: K[X]/(p(X)) −→ F
whose image is K(t) F. Of course, there is one such monomorphism for each root of p(X) in
F. If we ﬁx one such root t
0
and identify K[X]/(p(X)) with K(t
0
) via ¯ε
t
0
, then each root of
p(X) in F gives rise to a monomorphism ϕ
t
=¯ε
t
◦ ¯ε
−1
t
0
: K(t
0
) −→ F for which ϕ
t
(t
0
) = t.
K(t
0
)
ϕ
t
=ε
t
◦ε
−1
t
0
K[X]/(p(X))
ε
t
0
∼
=
ε
t
F
Notice that if ϕ: K[X]/(p(X)) −→ F is any homomorphism extending the identity function
on K, then the coset X + (p(X)) must be sent by ϕ to a root of p(X) in F, hence every such
homomorphism arises this way. This discussion is summarized in the following result.
3.34. Proposition. Let F/K be a ﬁeld extension. Let p(X) ∈ K[X] be an irreducible
polynomial with t
0
∈ F be a root of p(X). Then there is a bijection
Roots(p, F) ←→ Mono
K
(K(t
0
), F)
given by t ←→ ϕ
t
, where ϕ
t
: K(t
0
) −→ F has the eﬀect ϕ
t
(t
0
) = t.
3.35. Example. Show that Mono
Q
(Q(
√
2), C) has two elements.
Solution. We have Q(
√
2)
∼
= Q[X]/(X
2
− 2) where X
2
− 2 is irreducible over Q. Hence
the Qmonomorphisms we want send
√
2 to ±
√
2 which are the complex roots of X
2
− 2. In
fact both possibilities occur, giving monomorphisms id, α: Q(
√
2) −→C, where
α(a +b
√
2) = a −b
√
2.
We can replace C by Q(
√
2) to obtain
Mono
Q
(Q(
√
2), C) = Mono
Q
(Q(
√
2), Q(
√
2)) = Aut
Q
(Q(
√
2)).
We will see that this is not always true.
3.36. Example. Show that Mono
Q
(Q(
3
√
2), C) has 3 elements but Mono
Q
(Q(
3
√
2), Q(
3
√
2))
contains only the identity function.
Solution. Here minpoly
Q,
3
√
2
(X) = X
3
− 2 and there are 3 complex roots
3
√
2,
3
√
2 ζ
3
,
3
√
2 ζ
2
3
. As two of these roots are not real, Mono
Q
(Q(
3
√
2), Q(
3
√
2)) contains only the identity
since Q(
3
√
2) R.
Each of the above roots corresponds to one of the subﬁelds Q(
3
√
2), Q(
3
√
2 ζ
3
) or Q(
3
√
2 ζ
2
3
)
of C and there are 3 monomorphisms α
0
, α
1
, α
2
: Q(
3
√
2) −→C given by
α
0
(a +b
3
√
2 +c(
3
√
2)
2
) = a +b
3
√
2 +c(
3
√
2)
2
,
α
1
(a +b
3
√
2 +c(
3
√
2)
2
) = a +b
3
√
2 ζ
3
+c(
3
√
2)
2
ζ
2
3
,
α
2
(a +b
3
√
2 +c(
3
√
2)
2
) = a +b
3
√
2 ζ
2
3
+c(
3
√
2)
2
ζ
3
.
3.3. MONOMORPHISMS BETWEEN EXTENSIONS 41
These mappings have images
α
0
Q(
3
√
2) = Q(
3
√
2), α
1
Q(
3
√
2) = Q(
3
√
2 ζ
3
), α
2
Q(
3
√
2) = Q(
3
√
2 ζ
2
3
).
3.37. Proposition. Let F/K and L/K be extensions.
(i) For p(X) ∈ K[X], each monomorphism α ∈ Mono
K
(L, F) restricts to a function
α
p
: Roots(p, L) −→ Roots(p, F) which is an injection.
(ii) If α ∈ Mono
K
(L, L), then α
p
: Roots(p, L) −→ Roots(p, L) is a bijection.
Proof. (i) For u ∈ Roots(p, L) we have
p(α(u)) = α(p(u)) = α(0) = 0,
so α maps Roots(p, L) into Roots(p, F). Since α is an injection its restriction to Roots(p, L) ⊆ L
is also an injection.
(ii) From (i), α
p
: Roots(p, L) −→ Roots(p, L) is an injective function from a ﬁnite set to itself,
hence it is also surjective by the Pigeon Hole Principle. Thus α
p
: Roots(p, L) −→ Roots(p, L)
is a bijection.
Part (ii) says that any automorphism of L/K permutes the set of roots in L of a polynomial
p(X) ∈ K[X]. This gives us a strong hold on the possible automorphisms. In the case of ﬁnite,
or more generally algebraic, extensions it is the key to understanding the automorphism group
and this is a fundamental insight of Galois Theory.
3.38. Example. Determine Mono
Q
(Q(
3
√
2, ζ
3
), C).
Solution. We have already met the extension Q(
3
√
2, ζ
3
)/Q in Example 3.30 and we will
make use of information from there. We build up the list of monomorphisms in stages.
First consider monomorphisms that ﬁx
3
√
2 and hence ﬁx the subﬁeld Q(
3
√
2). These form
the subset
Mono
Q(
3
√
2)
(Q(
3
√
2, ζ
3
), C) ⊆ Mono
Q
(Q(
3
√
2, ζ
3
), C).
We know that Q(
3
√
2, ζ
3
) = Q(
3
√
2)(ζ
3
) and that ζ
3
is a root of the irreducible cyclotomic
polynomial Φ
3
(X) = X
2
+ X + 1 ∈ Q(
3
√
2)[X]. So there are two monomorphisms id, α
0
ﬁxing
Q(
3
√
2), where α
0
has the eﬀect
α
0
:
3
√
2 −→
3
√
2
ζ
3
−→ ζ
2
3
.
Next we consider monomorphisms that send
3
√
2 to
3
√
2 ζ
3
. This time we have 2 distinct ways to
extend to elements of Mono
Q
(Q(
3
√
2, ζ
3
), Q(
3
√
2, ζ
3
)) since again we can send ζ
3
to either ζ
3
or
ζ
2
3
. The possibilities are
α
1
:
3
√
2 −→
3
√
2 ζ
3
ζ
3
−→ ζ
3
, α
t
1
:
3
√
2 −→
3
√
2 ζ
3
ζ
3
−→ ζ
2
3
.
Finally we consider monomorphisms that send
3
√
2 to
3
√
2 ζ
2
3
. There are again two possibilities
α
2
:
3
√
2 −→
3
√
2 ζ
2
3
ζ
3
−→ ζ
3
, α
t
2
:
3
√
2 −→
3
√
2 ζ
2
3
ζ
3
−→ ζ
2
3
.
These are all 6 of the required monomorphisms. It is also the case here that
Mono
Q
(Q(
3
√
2, ζ
3
), C) = Mono
Q
(Q(
3
√
2, ζ
3
), Q(
3
√
2, ζ
3
)) = Aut
Q
(Q(
3
√
2, ζ
3
)),
so these form a group. It is a nice exercise to show that Aut
Q
(Q(
3
√
2, ζ
3
))
∼
= S
3
, the symmetric
group on 3 objects. It is also worth remarking that [ Aut
Q
(Q(
3
√
2, ζ
3
))[ = [Q(
3
√
2, ζ
3
) : Q].
We end this section with another useful result.
42 3. ALGEBRAIC EXTENSIONS OF FIELDS
3.39. Proposition. Let L/K be an extension and α ∈ Mono
K
(L, L). Then α restricts to
an automorphism α
alg
: L
alg
−→ L
alg
.
Proof. Suppose that u ∈ L
alg
, say p(u) = 0 for some p(X) ∈ K[X] of positive degree.
Then
p(α(u)) = α(p(u)) = α(0) = 0,
so α maps L
alg
⊆ L into itself and therefore gives rise to a restriction α
alg
: L
alg
−→ L
alg
which
is also a monomorphism. We must show that α
alg
is a bijection by showing it is surjective.
Let v ∈ L
alg
and suppose that q(v) = 0 for some q(X) ∈ K[X] of positive degree. Now
Roots(q, L) = ∅ since it contains v, and it is also ﬁnite. Then α
q
: Roots(q, L) −→ Roots(q, L)
is a bijection by Proposition 3.37(ii), hence v = α
q
(w) = α(w) for some w ∈ Roots(q, L) ⊆ L
alg
.
This shows that v ∈ imα and so α
alg
is surjective.
3.4. Algebraic closures
An important property of the complex numbers is that C is algebraically closed.
3.40. Theorem (Fundamental Theorem of Algebra for C). Every nonconstant polynomial
p(X) ∈ C[X] has a root in C.
3.41. Corollary. Every nonconstant polynomial p(X) ∈ C[X] has a factorization
p(X) = c(X −u
1
) (X −u
d
),
where c, u
1
, . . . , u
d
∈ C and this is unique apart from the order of the roots u
j
.
It is natural to pose the following question.
3.42. Question. Let K be a ﬁeld. Is there an algebraically closed ﬁeld F containing K?
By taking F
alg
we might as well ask that such a ﬁeld be algebraic over K.
3.43. Definition. Let K be a ﬁeld. An extension F/K is called an algebraic closure of K
if F is algebraic over K and algebraically closed.
3.44. Theorem. Let K be a ﬁeld.
(i) There is an algebraic closure of K.
(ii) Let F
1
and F
2
be algebraic closures of K. Then there is an isomorphism ϕ: F
1
−→ F
2
which ﬁxes the elements of K.
K
.·
F
1
ϕ
F
2
Hence algebraic closures are essentially unique.
Proof. See [3] for a proof using Zorn’s Lemma (see Axiom 3.48) which is logically equiva
lent to the Axiom of Choice.
Because of the uniqueness we usually ﬁx some choice of algebraic closure of K and write K
or K
alg cl
, referring to it as the algebraic closure of K. We are already familiar with the example
C = C. There are some immediate consequences of Theorem 3.44. We will temporarily write
E
1
.
= E
2
to indicate that for extensions E
1
/K and E
2
/K there is an isomorphism E
1
−→ E
2
ﬁxing the elements of K.
3.45. Proposition. Let K be a ﬁeld.
(i) If L/K is an algebraic extension, then L
.
= K.
3.4. ALGEBRAIC CLOSURES 43
(ii) If L/K is an extension, then so is L/K and (L)
alg
.
= K.
Proof. (i) By Proposition 3.16, every element of L is algebraic over K. Since L is alge
braically closed it is an algebraic closure of K.
(ii) Every nonconstant polynomial in (L)
alg
[X] has a root in L; indeed, by Proposition 3.16,
all of its roots are in fact algebraic over K since (L)
alg
is algebraic over K. Hence these roots
lie in (L)
alg
, which shows that it is algebraically closed.
For example, we have Q = C
alg
and R = C.
There is a stronger result than Theorem 3.44(ii), the Monomorphism Extension Theorem,
which we will ﬁnd useful. Again the proof uses Zorn’s Lemma which we state below. First we
need some deﬁnitions.
3.46. Definition. A partially ordered set (X, ) consists of a set X and a binary relation
such that whenever x, y, z ∈ X,
• x x;
• if x y and y z then x z;
• if x y and y x then x = y.
(X, ) is totally ordered if for every pair x, y ∈ X, at least one of x y or y x is true.
3.47. Definition. Let (X, ) be a partially ordered set and Y ⊆ X.
• y ∈ X is an upper bound for Y if for every y ∈ Y , y y.
• An element x ∈ X is a maximal element of X if
x y =⇒ y = x.
3.48. Axiom (Zorn’s Lemma). Let (X, ) be a partially ordered set in which every totally
ordered subset has an upper bound. Then X has a maximal element.
3.49. Theorem (Monomorphism Extension Theorem). Let M/K be an algebraic extension
and L/K M/K. Suppose that ϕ
0
: L −→ K is a monomorphism ﬁxing the elements of K.
Then there is an extension of ϕ
0
to a monomorphism ϕ: M −→ K.
K
M
ϕ
L
ϕ
0
K
=
K
Proof. We consider the set X consisting of all pairs (F, θ), where F/L M/L and θ: F −→
K extends ϕ
0
. We order X using the relation for which (F
1
, θ
1
) (F
2
, θ
2
) whenever F
1
F
2
and θ
2
extends θ
1
. Then (X, ) is a partially ordered set.
Suppose that Y ⊆ X is a totally ordered subset. Let
¯
F =
¸
(F,θ)∈Y
F.
Then
¯
F/L M/L. Also there is a function
¯
θ:
¯
F −→ K deﬁned by
¯
θ(u) = θ(u)
44 3. ALGEBRAIC EXTENSIONS OF FIELDS
whenever u ∈ F for (F, θ) ∈ Y . It is straightforward to check that if u ∈ F
t
for (F
t
, θ
t
) ∈ Y
then
θ
t
(u) = θ(u),
so
¯
θ is welldeﬁned. Then for every (F, θ) ∈ Y we have (F, θ) (
¯
F,
¯
θ), so (
¯
F,
¯
θ) is an upper
bound for Y . By Zorn’s Lemma there must be a maximal element of X, (M
0
, θ
0
).
Suppose that M
0
= M, so there is an element u ∈ M for which u / ∈ M
0
. Since M is algebraic
over K it is also algebraic over M
0
, hence u is algebraic over M
0
. If
minpoly
M
0
,u
(X) = a
0
+ +a
n−1
X
n−1
+X
n
,
then the polynomial
f(X) = θ
0
(a
0
) + +θ
0
(a
n−1
)X
n−1
+X
n
∈ (θ
0
M
0
)[X]
is also irreducible and so it has a root v in K (which is also an algebraic closure of θ
0
M
0
K). The Homomorphism Extension Property 1.22 of the polynomial ring M
0
[X] applied to
the monomorphism θ
0
: M
0
−→ K yields a homomorphism θ
t
0
: M
0
[X] −→ K extending θ
0
and for which θ
t
0
(u) = v. This factors through the quotient ring M
0
[X]/(minpoly
M
0
,u
(X)) to
give a monomorphism θ
tt
0
: M
0
(u) −→ K extending θ
0
. But then (M
0
, θ
0
) (M
0
(u), θ
tt
0
) and
(M
0
, θ
0
) = (M
0
(u), θ
tt
0
), contradicting the maximality of (M
0
, θ
0
). Hence M
0
= M and so we
can take ϕ = θ
0
.
3.50. Example. Let u ∈ K and suppose that p(X) = minpoly
K,u
(X) ∈ K[X]. Then for
any other root of p(X), v ∈ K say, there is a monomorphism ϕ
v
: K(u) −→ K with ϕ
v
(u) = v.
This extends to a monomorphism ϕ: K −→ K.
3.51. Definition. Let u, v ∈ K. Then v is conjugate to u over K or is a conjugate of u
over K if there is a monomorphism ϕ: K −→ K ﬁxing K for which v = ϕ(u).
3.52. Lemma. If u, v ∈ K, then v is conjugate to u over K if and only if minpoly
K,u
(v) = 0.
Proof. Suppose that v = ϕ(u) for some ϕ ∈ Mono
K
(K, K). If
minpoly
K,u
(X) = a
0
+a
1
X + +a
d−1
X
d−1
+X
d
,
then
a
0
+a
1
u + +a
d−1
u
d−1
+u
d
= 0
and so
a
0
+a
1
v + +a
d−1
v
d−1
+v
d
= ϕ(a
0
+a
1
u + +a
d−1
u
d−1
+u
d
) = 0.
The converse follows from Example 3.50.
3.5. Multiplicity of roots and separability
Let K be a ﬁeld. Suppose that f(X) ∈ K[X] and u ∈ K is a root of f(X), i.e., f(u) = 0.
Then we can factor f(X) as f(X) = (X −u)f
1
(X) for some f
1
(X) ∈ K[X].
3.53. Definition. If f
1
(u) = 0 then u is a multiple or repeated root of f(X). If f
1
(u) = 0
then u is a simple root of f(X).
We need to understand more clearly when an irreducible polynomial has a multiple root
since this turns out to be important in what follows. Consider the formal derivative on K[X],
i.e., the function ∂ : K[X] −→ K[X] given by
∂(f(X)) = f
t
(X) = a
1
+ 2a
2
X + +da
d
X
d−1
,
where f(X) = a
0
+a
1
X +a
2
X
2
+ +a
d
X
d
with a
j
∈ K.
3.54. Proposition. The formal derivative ∂ : K[X] −→ K[X] has the following properties.
3.5. MULTIPLICITY OF ROOTS AND SEPARABILITY 45
(i) ∂ is Klinear.
(ii) ∂ is a derivation, i.e., for f(X), g(X) ∈ K[X],
∂(f(X)g(X)) = ∂(f(X))g(X) +f(X)∂(g(X)).
(iii) If char K = 0, then ker ∂ = K and ∂ is surjective.
(iv) If char K = p > 0, then
ker ∂ = ¦h(X
p
) : h(X) ∈ K[X]¦
and im∂ is spanned by the monomials X
k
with p (k + 1).
Proof. (i) This is routine.
(ii) By Klinearity, it suﬃces to verify this for the case where f(X) = X
r
and g(X) = X
s
with
r, s 0. But then
∂(X
r+s
) = (r +s)X
r+s−1
= rX
r−1
X
s
+sX
r
X
s−1
= ∂(X
r
)X
s
+X
r
∂(X
s
).
(iii) If f(X) = a
0
+a
1
X +a
2
X
2
+ +a
d
X
d
then
∂(f(X)) = 0 ⇐⇒ a
1
= 2a
2
= = da
d
= 0.
So ∂(f(X)) = 0 if and only if f(X) = a
0
∈ K. It is also clear that every polynomial g(X) ∈ K[X]
has the form g(X) = ∂(f(X) where f(X) is an antiderivative of g(X).
(iv) For a monomial X
m
, ∂(X
m
) = mX
m−1
and this is zero if and only if p [ m. Using this we
see that
∂(a
0
+a
1
X +a
2
X
2
+ +a
d
X
d
) = 0 ⇐⇒ a
m
= 0 whenever p m.
Also, im∂ is spanned by the monomials X
k
for which ∂(X
k+1
) = 0, which are the ones with
p (k + 1).
We now apply the formal derivative to detect multiple roots.
3.55. Proposition. Let f(X) ∈ K[X] have a root u ∈ L for some extension L/K. Then u
is a multiple root of f(X) if and only if f(X) and f
t
(X) have a common factor of positive degree
in K[X] which vanishes at u.
Proof. Working in L[X], let f(X) = (X −u)f
1
(X). Then
f
t
(X) = f
1
(X) + (X −u)f
t
1
(X),
so f
t
(u) = f
1
(u). Hence u is a multiple root if and only if f(X) and f
t
(X) have a common
factor in L[X] (and hence in K[X] by Proposition 3.12) and which vanishes at u.
3.56. Corollary. If f(X) is irreducible in K[X] then a root u is a multiple root if and
only if f
t
(X) = 0. In particular, this can only happen if char K > 0.
3.57. Corollary. If char K = 0 and f(X) is irreducible in K[X], then every root of f(X)
is simple.
3.58. Example. For n 1, show that each of the roots of f(X) = X
n
−1 in C is simple.
Solution. We have f
t
(X) = ∂(X
n
−1) = nX
n−1
, so for any root ζ of f(X),
f
t
(ζ) = nζ
n−1
= 0.
3.59. Example. Show that 2i is a multiple root of f(X) = X
4
+ 8X
2
+ 16.
Solution. We have f
t
(X) = 4X
3
+16X. Using Long Division and the Euclidean Algorithm
we ﬁnd that gcd(f(X), f
t
(X)) = X
2
+4, where 2i is also a root of X
2
+4. Hence 2i is a multiple
root of f(X). In fact, X
4
+ 8X
2
+ 16 = (X
2
+ 4)
2
, so this is obvious.
46 3. ALGEBRAIC EXTENSIONS OF FIELDS
3.60. Example. Let p > 0 be a prime and suppose that L/F
p
is an extension. Show that
each of the roots of f(X) = X
p
−1 in L is multiple.
Solution. We have f
t
(X) = ∂(X
p
− 1) = pX
p−1
= 0, so if ζ is any root of f(X) then
f
t
(ζ) = 0. Later we will see that 1 is the only root of X
p
−1.
3.61. Definition. An irreducible polynomial p(X) ∈ K[X] is separable over K if every
root of p(X) in an extension L/K is simple. By Corollary 3.56, this is equivalent to requiring
that p
t
(X) = 0. If u ∈ L is a multiple root of p(X), then the multiplicity of u in p(X) is the
maximum m such that p(X) = (X −u)
m
q(X) for some q(X) ∈ L[X].
3.62. Proposition. Let K be a ﬁeld and let K be an algebraic closure. If the irreducible
polynomial p(X) ∈ K[X] has distinct roots u
1
, . . . , u
k
∈ K, then the multiplicities of the u
j
are
equal. Hence in K[X],
p(X) = c(X −u
1
)
m
(X −u
k
)
m
,
where c ∈ K and m 1.
Proof. Let u ∈ K be a root of p(X) and suppose that it has multiplicity m, so we can
write p(X) = (X −u)
m
p
1
(X) where p
1
(X) ∈ K(u)[X] and p
1
(u) = 0.
Now let v ∈ K be any other root of p(X). By Proposition 3.34, there is a monomorphism
ϕ
v
: K(u) −→ K for which ϕ
v
(u) = v. When p(X) is viewed as an element of K(u)[X], the
coeﬃcients of p(X) are ﬁxed by ϕ
v
. Then
ϕ
v
((X −u)
m
p
1
(X)) = (X −u)
m
p
1
(X),
and so
(X −v)
m
¯ p
1
(X) = (X −u)
m
p
1
(X),
where ¯ p
1
(X) ∈ K[X] is obtained applying ϕ
v
to the coeﬃcients of p
1
(X). Now by Corollary 1.34,
(X − v)
m
must divide p
1
(X) in K[X], and therefore the multiplicity of v must be at least m.
Interchanging the rˆoles of u and v we ﬁnd that the multiplicities of u and v are in fact equal.
3.63. Corollary. Let K be a ﬁeld and let K be an algebraic closure. If the irreducible
polynomial p(X) ∈ K[X] has distinct roots u
1
, . . . , u
k
∈ K which are all simple then in K[X],
p(X) = c(X −u
1
) (X −u
k
),
where c ∈ K and k = deg p(X).
3.64. Corollary. Let K be a ﬁeld and let u ∈ K. Then the number of distinct conjugates
of u is
deg minpoly
K,u
(X)
m
,
where m is the multiplicity of u in minpoly
K,u
(X).
3.65. Definition. An algebraic element u ∈ L in an extension L/K is separable if its
minimal polynomial minpoly
K,u
(X) ∈ K[X] is separable.
3.66. Definition. An algebraic extension L/K is called separable if every element of L is
separable over K.
3.67. Example. An algebraic extension L/K of a ﬁeld of characteristic 0 is separable by
Corollary 3.57.
3.68. Definition. Let L/K be a ﬁnite extension. The separable degree of L over K is
(L : K) = [ Mono
K
(L, K)[.
3.5. MULTIPLICITY OF ROOTS AND SEPARABILITY 47
3.69. Lemma. For a ﬁnite simple extension K(u)/K,
(K(u) : K) = [ Roots(minpoly
K,u
, K)[.
If K(u)/K is separable, then [K(u) : K] = (K(u) : K).
Proof. This follows from Proposition 3.34 applied to the case L = K.
Any ﬁnite extension L/K can be built up from a succession of simple extensions
(3.1) K(u
1
)/K, K(u
1
, u
2
)/K(u
1
), , L = K(u
1
, . . . , u
k
)/K(u
1
, . . . , u
k−1
).
So we can use the following to compute (L : K) = (K(u
1
, . . . , u
k
) : K).
3.70. Proposition. Let L/K and M/L be ﬁnite extensions. Then
(M : K) = (M : L)(L : K).
Proof. For α ∈ Mono
K
(M, K) let α
L
∈ Mono
K
(L, K) be its restriction to L. By the
Monomorphism Extension Theorem 3.49, each element of Mono
K
(L, K) extends to a monomor
phism M −→ K, so every element β ∈ Mono
K
(L, K) has the form β = α
L
for some α ∈
Mono
K
(M, K). Since (L : K) = [ Mono
K
(L, K)[, we need to show that the number of such α is
always (M : L) = [ Mono
L
(M, K)[.
So given β ∈ Mono
K
(L, K), choose any extension to a monomorphism
¯
β: K −→ K; by
Proposition 3.39,
¯
β is an automorphism. Of course, restricting to M K we obtain a monomor
phism M −→ K. Now for any extension β
t
: M −→ K of β we can form the composition
¯
β
−1
◦ β
t
: M −→ K; notice that if u ∈ L, then
¯
β
−1
◦ β
t
(u) =
¯
β
−1
(β(u)) = u,
hence
¯
β
−1
◦ β
t
∈ Mono
L
(M, K). Conversely, each γ ∈ Mono
L
(M, K) gives rise to a monomor
phism
¯
β ◦ γ : M −→ K which extends β. In eﬀect, this shows that there is a bijection
¸
extensions of β to monomorphism a M −→ K
¸
←→ Mono
L
(M, K),
so (M : L) = [ Mono
L
(M, K)[ agrees with the number of extensions of β to a monomorphism
M −→ K. Therefore we have the desired formula (M : K) = (M : L)(L : K).
3.71. Corollary. Let L/K be a ﬁnite extension. Then (L : K) [ [L : K].
Proof. If L/K is a simple extension then by Propositions 3.62 and 3.34 we know that this
is true. The general result follows by building up L/K as a sequence of simple extensions as
in (3.1) and then using Theorem 2.6(ii) which gives
[L : K] = [K(u
1
) : K] [K(u
1
, u
2
) : K(u
1
)] [K(u
1
, . . . , u
k
) : K(u
1
, . . . , u
k−1
)].
For each k, (K(u
1
, . . . , u
k
) : K(u
1
, . . . , u
k−1
)) divides [K(u
1
, . . . , u
k
) : K(u
1
, . . . , u
k−1
)], so the
desired result follows.
3.72. Proposition. Let L/K be a ﬁnite extension. Then L/K is separable if and only if
(L : K) = [L : K].
Proof. Suppose that L/K is separable. If K E L, then for any u ∈ L, u is alge
braic over E, and in the polynomial ring E[X] we have minpoly
E,u
(X) [ minpoly
K,u
(X). As
minpoly
K,u
(X) is separable, so is minpoly
E,u
(X), and therefore L/E is separable. Clearly E/K
is also separable. We have (L : K) = (L : E) (E : K) and [L : K] = [L : E] [E : K], so to
verify that (L : K) = [L : K] it suﬃces to show that (L : E) = [L : E] and (E : K) = [E : K].
Expressing L/K in terms of a sequence of simple extensions as in (3.1), we have
(L : K) = (K(u
1
) : K) (L : K(u
1
, . . . , u
k−1
)),
[L : K] = [K(u
1
) : K] [L : K(u
1
, . . . , u
k−1
)].
48 3. ALGEBRAIC EXTENSIONS OF FIELDS
Now we can apply Lemma 3.69 to each of these intermediate separable simple extensions to
obtain (L : K) = [L : K].
For the converse, suppose that (L : K) = [L : K]. We must show that for each u ∈ L, u is
separable. For the extensions K(u)/K and L/K(u) we have (L : K) = (L : K(u)) (K(u) : K)
and [L : K] = [L : K(u)] [K(u) : K]. By Corollary 3.71, there are some positive integers r, s for
which [L : K(u)] = r(L : K(u)) and [K(u) : K] = s(K(u) : K). Hence
(L : K(u))(K(u) : K) = rs(L : K(u))(K(u) : K),
which can only happen if r = s = 1. Thus (K(u) : K) = [K(u) : K] and so u is separable.
3.73. Proposition. Let L/K and M/L be ﬁnite extensions. Then M/K is separable if and
only if L/K and M/L are separable.
Proof. If M/K is separable then [M : K] = (M : K) and so by Proposition 3.70,
[M : L][L : K] = (M : L)(L : K).
This can only happen if [M : L] = (M : L) and [L : K] = (L : K), since (M : L) [M : L] and
(L : K) [L : K]. By Proposition 3.72 this implies that L/K and M/L are separable.
Conversely, if L/K and M/L are separable then [M : L] = (M : L) and [L : K] = (L : K),
hence
[M : K] = [M : L][L : K] = (M : L)(L : K) = (M : K).
Therefore M/K is separable.
3.6. The Primitive Element Theorem
3.74. Definition. For a ﬁnite simple extension L/K, an element u ∈ L is called a primitive
element for the extension if L = K(u).
3.75. Theorem (Primitive Element Theorem). Let L/K be a ﬁnite separable extension.
Then L has a primitive element.
Proof. The case where K is a ﬁnite ﬁeld will be dealt with in Proposition 5.16. So we will
assume that K is inﬁnite.
Since L is built up from a sequence of simple extensions it suﬃces to consider the case
L = K(u, v). Let p(X), q(X) ∈ K[X] be the minimal polynomials of u and v over K. Suppose
that the distinct roots of p(X) in K are u = u
1
, . . . , u
r
, while the distinct roots of q(X) are
v = v
1
, . . . , v
s
. By the separability assumption, r = deg p(X) and s = deg q(X).
Since K is inﬁnite, we can choose an element t ∈ K for which
t =
u −u
i
v
j
−v
whenever j = 1. Then taking w = u+tv ∈ L, we ﬁnd that w = u
i
+tv
j
whenever j = 1. Deﬁne
the polynomial (of degree r)
h(X) = p(w −tX) ∈ K(w)[X] ⊆ L[X].
Then h(v) = p(u) = 0, but h(v
j
) = p(u
i
) = 0 for any j = 1 by construction of t, so none of the
other v
j
is a zero of h(X).
Now since the polynomials h(X), q(X) ∈ K(w)[X] have exactly one common root in K,
namely v, by separability their greatest common divisor in K(w)[X] is a linear polynomial which
must be X −v, hence v ∈ K(w) and so u = w −tv ∈ K(w). This shows that K(u, v) K(w)
and therefore K(w) = K(u, v).
3.76. Corollary. Let L/K be a ﬁnite extension of a ﬁeld of characteristic 0. Then L has
a primitive element.
3.6. THE PRIMITIVE ELEMENT THEOREM 49
Proof. Since Q K, K is inﬁnite and by Example 3.67 L/K is separable.
To ﬁnd a primitive element we can always use the method suggested by the proof of Theo
rem 3.75, however a ‘try it and see’ approach will often be suﬃcient.
3.77. Example. Find a primitive element for the extension Q(
√
3, i)/Q.
Solution. Consider
√
3+i. Then working over the subﬁeld Q(
√
3) Q(
√
3, i) we ﬁnd that
i / ∈ Q(
√
3) R and
(X −(
√
3 +i))(X −(
√
3 −i)) = X
2
−2
√
3X + 4 ∈ Q(
√
3)[X],
hence
X
2
−2
√
3X + 4 = minpoly
Q(
√
3),
√
3+i
(X).
Now taking
(X
2
−2
√
3X + 4)(X
2
+ 2
√
3X + 4) = X
4
−4X
2
+ 16 ∈ Q[X],
we see that minpoly
Q,
√
3+i
(X) [ (X
4
−4X
2
+ 16) in Q[X]. Notice that
(
√
3 +i)
−1
=
(
√
3 −i)
(
√
3 +i)(
√
3 −i)
=
(
√
3 −i)
3 + 1
=
1
4
(
√
3 −i) ∈ Q(
√
3 +i),
since (
√
3 +i)
−1
∈ Q(
√
3 +i). Hence
√
3 =
1
2
((
√
3 +i) + (
√
3 −i)), i =
1
2
((
√
3 +i) −(
√
3 −i)),
are both in Q(
√
3 +i), showing that Q(
√
3, i) Q(
√
3 +i) and so Q(
√
3, i) = Q(
√
3 +i). Thus
we must have deg minpoly
Q,
√
3+i
(X) = 4, and so minpoly
Q,
√
3+i
(X) = X
4
−4X
2
+ 16.
There is a general phenomenon illustrated by Example 3.77.
3.78. Proposition. Let u ∈ K be separable over K. Then
minpoly
K,u
(X) = (X −α
1
(u)) (X −α
d
(u)),
where α
1
, . . . , α
d
are the elements of Mono
K
(K(u), K). In particular, the polynomial
(X −α
1
(u)) (X −α
d
(u)) ∈ K[X]
is in K[X] and is irreducible therein.
Proof. Since K(u) is separable then by Lemma 3.52,
d = deg minpoly
K,u
(X) = [K(u) : K] = (K(u) : K).
In Example 3.77 we have
[Q(
√
3, i) : Q] = [Q(
√
3, i) : Q(
√
3)][Q(
√
3) : Q] = 2 2 = 4.
There are four monomorphisms α
k
: Q(
√
3, i) −→Q(
√
3, i) given by
α
1
= id, α
2
=
√
3 −→
√
3
i −→ −i
, α
3
=
√
3 −→ −
√
3
i −→ i
, α
4
=
√
3 −→ −
√
3
i −→ −i
.
Then
α
2
(
√
3 +i) = (
√
3 −i), α
3
(
√
3 +i) = (−
√
3 +i), α
4
(
√
3 +i) = (−
√
3 −i),
so
(X −
√
3 −i)(X −
√
3 +i)(X +
√
3 −i)(X +
√
3 +i) = X
4
−4X
2
+ 16 ∈ Q[X].
Hence this polynomial is irreducible. So we have [Q(
√
3+i) : Q] = 4 and Q(
√
3+i) = Q(
√
3, i).
50 3. ALGEBRAIC EXTENSIONS OF FIELDS
3.7. Normal extensions and splitting ﬁelds
Let K be an algebraic closure for the ﬁeld K and let E/K K/K be a ﬁnite extension. If
ϕ ∈ Mono
K
(E, K), then by Remark 3.32, ϕE = E if and only if ϕE E.
3.79. Definition. E/K is normal if ϕE = E for every ϕ ∈ Mono
K
(E, K).
3.80. Remark. If E/K is a normal extension then whenever an irreducible polynomial
p(X) ∈ K[X] has a root in E, it splits in E since by Lemma 3.52 each pair of roots of p(X) is
conjugate over K and one can be mapped to the other by a monomorphism K −→ K which
must map E into itself.
3.81. Theorem. A ﬁnite extension E/K is normal if and only if it is a splitting ﬁeld over
K for some polynomial f(X) ∈ K[X].
Proof. Suppose that E/K is normal. Then there is a sequence of extensions
K K(u
1
) K(u
1
, u
2
) K(u
1
, . . . , u
n
) = E
Construct a polynomial by taking
f(X) = minpoly
K,u
1
(X) minpoly
K,u
2
(X) minpoly
K,u
n
(X).
Then by Remark 3.80, f(X) splits in E. Also, E is generated by some of the roots of f(X).
Hence E is a splitting ﬁeld for f(X) over K.
Now suppose that E is a splitting ﬁeld for g(X) ∈ K[X], so that E = K(v
1
, . . . , v
k
), where
v
1
, . . . , v
k
are the distinct roots of g(X) in E. Now any monomorphism θ ∈ Mono
K
(E, K) must
map these roots to θ(v
1
), . . . , θ(v
k
) which are also roots of g(X) and therefore lie in E (see
Proposition 3.34). Since θ permutes the roots v
j
, we have
θE = θK(v
1
, . . . , v
k
) = K(θ(v
1
), . . . , θ(v
k
)) = K(v
1
, . . . , v
k
) = E.
3.82. Corollary. Let E/L and L/K be ﬁnite extensions. If E/K is normal then E/L is
normal.
Proof. If E is the splitting ﬁeld of a polynomial f(X) ∈ K[X] over K, then E is the
splitting ﬁeld of f(X) over L.
These result makes it easy to recognize a normal extension since it is suﬃcient to describe it
as a splitting ﬁeld for some polynomial over K. In Chapter 4 we will see that separable normal
extensions play a central rˆole in Galois Theory, indeed these are known as Galois extensions.
Exercises on Chapter 3
3.1. Prove Proposition 3.2.
3.2. Finding splitting subﬁelds E C over Q and determine [E : Q] for each of the following
polynomials.
p
1
(X) = X
4
−X
2
+1, p
2
(X) = X
6
−2, p
3
(X) = X
4
+2, p
4
(X) = X
4
+5X
3
+10X
2
+10X+5.
[Hint: for p
4
(X), consider p
4
(Y −1) ∈ Q[Y ].]
3.3. Prove that Aut
Q
(Q(
3
√
2, ζ
3
))
∼
= S
3
, the symmetric group on 3 elements, as claimed in the
solution of Example 3.38. [Hint: work out the eﬀect of each automorphism on the three roots of
the polynomial X
3
−2.]
3.4. Let k be a ﬁeld of characteristic char k = p > 0 and k(T) be the ﬁeld of rational functions
in T over k. Show that the polynomial g(X) = X
p
− T ∈ k(T)[X] is irreducible and has a
multiple root in k(T). How does g(X) factor in k(T)[X]?
51
3.5. Find primitive elements for the extensions Q(
√
5,
√
10)/Q, Q(
√
2, i)/Q, Q(
√
3, i)/Q,
Q(
4
√
3, i)/Q, in each case ﬁnding it minimal polynomial over Q. [Hint: look for elements of
high degree over Q, or use the method of proof of Theorem 3.75.]
3.6. Prove the following converse of Proposition 3.20:
Let L/K be a ﬁnite extension. If there are only ﬁnitely many subextensions F/K L/K,
then L/K is simple, i.e., L = K(w) for some w ∈ L.
[Hint: First deal with the case where L = K(u, v), then use induction on n to prove the general
case L = K(u
1
, . . . , u
n
).]
3.7. Let K be a ﬁeld. Show that every quadratic (i.e., of degree 2) extension E/K is normal.
Is such an extension always separable?
3.8. Let f(X) ∈ Q[X] be an irreducible polynomial of odd degree greater than 1 and having
only one real root u ∈ R. Show that Q(u)/Q is not a normal extension.
CHAPTER 4
Galois extensions and the Galois Correspondence
In this Chapter we will study the structure of Galois extensions and their associated Galois
groups, in particular we will explain how these are related through the Galois Correspondence.
Throughout the chapter, let K be a ﬁeld.
4.1. Galois extensions
4.1. Definition. A ﬁnite extension E/K is a (ﬁnite) Galois extension if it is both normal
and separable.
From Section 3.5 we know that for such a Galois extension E/K, [E : K] = (E : K) and also
every monomorphism ϕ ∈ Mono
K
(E, K) maps E into itself, hence restricts to an automorphism
of E which will be denoted ϕ
[
E
.
K
E
ϕ
ϕ

E
∼
=
E
K
=
K
Also, by the Monomorphism Extension Theorem 3.49, every automorphism α ∈ Aut
K
(E) ex
tends to a monomorphism E −→ K ﬁxing elements of K. So there is a bijection
Mono
K
(E, K) ←→ Aut
K
(E)
and we have
(4.1) [ Aut
K
(E)[ = (E : K) = [E : K].
4.2. Definition. For a ﬁnite Galois extension E/K, the group
Gal(E/K) = Aut
K
(E)
is called the Galois group of the extension or the Galois group of E over K. The elements of
Gal(E/K) are called (Galois) automorphisms of E/K.
Notice that Equation (4.1) implies
(4.2) [ Gal(E/K)[ = (E : K) = [E : K].
We can also reformulate the notion of conjugacy introduced in Deﬁnition 3.51.
4.3. Definition. Let E/K a ﬁnite Galois extension and u, v ∈ E. Then v is conjugate to u
if there is a ϕ ∈ Gal(E/K) for which v = ϕ(u); we also say that v is a conjugate of u.
It is easy to see that for u, v ∈ K, there is a ﬁnite Galois extension E/K in which v is
a conjugate of u if and only v is a conjugate of u over K in the old sense. Here is a slightly
diﬀerent way to understand this. First notice that every element ϕ ∈ Aut
K
(K, K) restricts to a
monomorphism E −→ K whose image is contained in E, hence gives rise to an automorphism
ϕ
E
: E −→ E. Similarly, if F/K is any ﬁnite normal extension with E F, every automorphism
53
54 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
θ: F −→ F restricts to an automorphism θ
F
E
: E −→ E. The proof of the next result is left as
an exercise.
4.4. Proposition. If E/K is a ﬁnite Galois extension, then the function
Aut
K
(K, K) −→ Aut
K
(E, E); ϕ −→ ϕ
E
is a surjective group homomorphism. If F/K K/K is any ﬁnite normal extension with E F
then there is a surjective group homomorphism
Aut
K
(F, F) −→ Aut
K
(E, E); θ −→ θ
F
E
.
Furthermore, for ϕ ∈ Aut
K
(K, K) we have
(ϕ
F
)
F
E
= ϕ
E
.
4.2. Working with Galois groups
Let E/K be a ﬁnite Galois extension. Then we know that E is a splitting ﬁeld for some
polynomial over K since E/K is normal. We also know that E is a simple extension of K
since E/K is separable. Hence E is a splitting ﬁeld for the minimal polynomial of any primitive
element for E/K; this minimal polynomial has degree [E : K]. It is often convenient to use these
facts to interpret elements of the Galois group as permutations of the roots of some polynomial
which splits over E.
4.5. Example. Describe the Galois group Gal(Q(
√
2,
√
3)/Q) as a subgroup of the group of
permutations of the roots of (X
2
−2)(X
2
−3) ∈ Q[X].
Solution. We have
[Q(
√
2,
√
3) : Q] = [Q(
√
2,
√
3) : Q(
√
2)] [Q(
√
2) : Q] = 4,
and the following nontrivial elements of the Galois group together with the element identity
α
1
= id:
α
2
=
¸
¸
¸
√
2 −→ −
√
2
−
√
2 −→
√
2
√
3 −→
√
3
−
√
3 −→ −
√
3
¸
, α
3
=
¸
¸
¸
√
2 −→
√
2
−
√
2 −→ −
√
2
√
3 −→ −
√
3
−
√
3 −→
√
3
¸
, α
4
=
¸
¸
¸
√
2 −→ −
√
2
−
√
2 −→
√
2
√
3 −→ −
√
3
−
√
3 −→
√
3
¸
.
Writing the roots in the list
√
2, −
√
2,
√
3, −
√
3 and numbering them from 1 to 4, these auto
morphisms correspond to the following permutations in S
4
expressed in cycle notation:
α
2
←→ (1 2), α
3
←→ (3 4), α
4
←→ (1 2)(3 4).
4.6. Example. Using a primitive element u for the extension, describe the Galois group
Gal(Q(
√
2,
√
3)/Q) as a subgroup of the group of permutations of the roots of minpoly
Q,u
(X) ∈
Q[X].
Solution. We have Q(
√
2,
√
3) = Q(
√
2 +
√
3) and the conjugates of u =
√
2 +
√
3 are
±
√
2 ±
√
3. Listing these as
√
2 +
√
3,
√
2 −
√
3, −
√
2 +
√
3, −
√
2 −
√
3,
and after numbering them accordingly, we ﬁnd the correspondences
α
2
←→ (1 3)(2 4), α
3
←→ (1 2)(3 4), α
4
←→ (1 4)(2 3).
Next we summarize the properties of Galois groups that can be deduced from what we
have established so far. Recall that for an extension F/K and a polynomial f(X) ∈ K[X],
Roots(f, F) denotes the set of roots of f(X) in F.
4.2. WORKING WITH GALOIS GROUPS 55
4.7. Recollection. Recall that an action of a group G on a set X is transitive if for every
pair of elements x, y ∈ X, there is an element g ∈ G such that y = gx (so there is only one
orbit); the action is faithful or eﬀective if for every nonidentity element h ∈ G, there is an
element z ∈ X such that hz = z.
4.8. Theorem. Let E/K be a ﬁnite Galois extension. Suppose that E is the splitting ﬁeld
of a separable irreducible polynomial f(X) ∈ K[X] of degree n. Then the following are true.
(i) Gal(E/K) acts transitively and faithfully on Roots(f, E).
(ii) Gal(E/K) can be identiﬁed with a subgroup of the group of permutations of Roots(f, E).
If we order the roots u
1
, . . . , u
n
then Gal(E/K) can be identiﬁed with a subgroup of S
n
.
(iii) [ Gal(E/K)[ divides n! and is divisible by n.
As we have seen in Examples 4.5 and 4.6, in practise it is often easier to use a not necessarily
irreducible polynomial to determine and work with a Galois group.
4.9. Example. The Galois extension Q(ζ
8
)/Q has degree [Q(ζ
8
) : Q] = 4 and it has the
following automorphisms apart from the identity:
α: ζ
8
−→ ζ
3
8
, β: ζ
8
−→ ζ
5
8
, γ : ζ
8
−→ ζ
7
8
.
If we list the roots of the minimal polynomial
minpoly
Q,ζ
(X) = Φ
8
(X) = X
4
+ 1
in the order ζ
8
, ζ
3
8
, ζ
5
8
, ζ
7
8
, we ﬁnd that these automorphisms correspond to the following permu
tations in S
4
:
α ←→ (1 2)(3 4), β ←→ (1 3)(2 4), γ ←→ (1 4)(2 3).
So the Galois group Gal(Q(ζ
8
)/Q) corresponds to
¦id, (1 2)(3 4), (1 3)(2 4), (1 4)(2 3)¦ S
4
.
Noticing that
ζ
8
=
1
√
2
+
1
√
2
i,
we easily ﬁnd that
√
2, i ∈ Q(ζ
8
); hence Q(
√
2, i) Q(ζ
8
). Since [Q(
√
2, i) : Q] = 4, we have
Q(
√
2, i) = Q(ζ
8
). Notice that Q(
√
2, i) is the splitting ﬁeld of f(X) = (X
2
− 2)(X
2
+ 1) over
Q. Now list the roots of f(X) in the order
√
2, −
√
2, i, −i, and observe that
α:
¸
¸
¸
√
2 −→ −
√
2
−
√
2 −→
√
2
i −→ −i
−i −→ i
¸
←→ (1 2)(3 4), β:
¸
¸
¸
√
2 −→ −
√
2
−
√
2 −→
√
2
i −→ i
−i −→ −i
¸
←→ (1 2),
γ :
¸
¸
¸
√
2 −→
√
2
−
√
2 −→ −
√
2
i −→ −i
−i −→ i
¸
←→ (3 4).
In this description, the Galois group Gal(Q(ζ
8
)/Q) = Gal(Q(
√
2, i)/Q) corresponds to the
subgroup
¦id, (1 2), (3 4), (1 2)(3 4)¦ S
4
.
While it can be hard to determine Galois groups in general, special arguments can sometimes
be exploited.
4.10. Example. Suppose that f(X) = X
3
+ aX
2
+ bX + c ∈ Q[X] is an irreducible cubic
and that f(X) has only one real root. Then Gal(Q(f(X))/Q)
∼
= S
3
.
56 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
Proof. Let u
1
∈ R be the real root of f(X) and let u
2
, u
3
be the remaining complex
roots. Then Q(f(X)) = Q(u
1
, u
2
, u
3
) and in fact [Q(f(X)) : Q] = 6 since [Q(f(X)) : Q] [ 6
and u
2
/ ∈ Q(u
1
) R. Hence Gal(Q(f(X))/Q) is isomorphic to a subgroup of S
3
and so
Gal(Q(f(X))/Q)
∼
= S
3
since the orders agree. We also have Q(f(X)) ∩ R = Q(u
1
).
The Galois group Gal(Q(f(X))/Q) contains an element of order 3 which corresponds to a
3cycle when viewed as a permutation of the roots u
1
, u
2
, u
3
; we can assume that this is (1 2 3).
It also contains an element of order 2 obtained by restricting complex conjugation to Q(f(X));
this ﬁxes u
1
and interchanges u
2
, u
3
, so it corresponds to the transposition (2 3).
4.11. Remark. Such examples occur when the cubic polynomial f(X) has local maximum
and minimum at real values c
+
and c
−
with f(c
+
), f(c
−
) > 0 or f(c
+
), f(c
−
) < 0. This happens
for example with f(X) = X
3
−3X +3 which has local extrema at ±1 and f(1) = 1, f(−1) = 5.
Given a Galois extension E/K, we will next study subextensions L/K E/K and sub
groups Γ Gal(E/K), focusing on the relationship between objects of these types.
4.3. Subgroups of Galois groups and their ﬁxed ﬁelds
Let E/K a Galois extension and suppose that Γ Gal(E/K). Consider the subset of
elements of E ﬁxed by Γ,
E
Γ
= ¦u ∈ E : ∀γ ∈ Γ, γ(u) = u¦.
4.12. Lemma. E
Γ
E is a subﬁeld of E containing K.
Proof. For u, v ∈ E
Γ
and γ ∈ Γ,
γ(u +v) = γ(u) +γ(v) = u +v, γ(uv) = γ(u)γ(v) = uv.
Also, if u = 0,
γ(u
−1
) = γ(u)
−1
= u
−1
.
Finally, if t ∈ K then γ(t) = t, so K E
Γ
.
4.13. Definition. E
Γ
E is the ﬁxed subﬁeld of Γ.
By Proposition 3.73, the extensions E/E
Γ
and E
Γ
/K are separable. E/E
Γ
is also normal,
so this is a Galois extension; we will identify its Galois group. Notice that
[E : E
Γ
] = (E : E
Γ
) = [ Gal(E/E
Γ
)[.
Now each element of Gal(E/E
Γ
) is also an element of Gal(E/K) and Gal(E/E
Γ
) Gal(E/K).
Notice that by deﬁnition Γ Gal(E/E
Γ
), so Lagrange’s Theorem implies that [Γ[ divides
[ Gal(E/E
Γ
)[. In fact we have
4.14. Proposition. For Γ Gal(E/K), we have Gal(E/E
Γ
) = Γ and the equations
[E : E
Γ
] = [ Gal(E/E
Γ
)[ = [Γ[, [E
Γ
: K] =
[ Gal(E/K)[
[Γ[
.
Proof. We know that E/E
Γ
is separable, so by the Primitive Element Theorem 3.75 it is
simple, say E = E
Γ
(u). Now let the distinct elements of Γ be γ
1
= id, γ
2
, . . . , γ
h
, where h = [Γ[.
Consider the polynomial of degree h
f(X) = (X −u)(X −γ
2
(u)) (X −γ
h
(u)) ∈ E[X].
Notice that f(X) is unchanged by applying any γ
k
to its coeﬃcients since the roots γ
j
(u) are
permuted by γ
k
. Hence, f(X) ∈ E
Γ
[X]. This shows that
[E : E
Γ
] = [E
Γ
(u) : E
Γ
] h = [Γ[.
4.4. SUBFIELDS OF GALOIS EXTENSIONS AND RELATIVE GALOIS GROUPS 57
Since Γ Gal(E/E
Γ
), we also have
h = [Γ[ [ Gal(E/E
Γ
)[ = [E : E
Γ
].
Combining these two inequalities we obtain
[E : E
Γ
] = [ Gal(E/E
Γ
)[ = [Γ[ = h
and therefore Γ = Gal(E/E
Γ
).
4.4. Subﬁelds of Galois extensions and relative Galois groups
Let E/K a Galois extension and suppose that L/K E/K (i.e., K L E). Then E/L
is also a Galois extension whose Galois group Gal(E/L) is sometimes called the relative Galois
group of the pair of extensions E/K and L/K. The following is immediate.
4.15. Lemma. The relative Galois group of the pair of extensions L/K E/K is a subgroup
of Gal(E/K), i.e., Gal(E/L) Gal(E/K), and its order is [ Gal(E/L)[ = [E : L].
4.16. Proposition. Let L/K E/K. Then L = E
Gal(E/L)
.
Proof. Clearly L E
Gal(E/L)
. Suppose that u ∈ E − L. By Theorem 4.8(i), there is
an automorphism θ ∈ Gal(E/L) such that θ(u) = u, hence u / ∈ E
Gal(E/L)
. This shows that
E
Gal(E/L)
L and therefore E
Gal(E/L)
= L.
We need to understand when Gal(E/L) Gal(E/K) is actually a normal subgroup. The
next result explains the connection between the two uses of the word normal which both ulti
mately derive from their use in Galois theory.
4.17. Proposition. Let E/K be a ﬁnite Galois extension and L/K E/K.
(i) The relative Galois group Gal(E/L) of the pair of extensions L/K E/K is a normal
subgroup of Gal(E/K) if and only if L/K is a normal extension.
(ii) If L/K is normal and hence a Galois extension, then there is a group isomorphism
Gal(E/K)/ Gal(E/L)
∼
=
−→ Gal(L/K); αGal(E/L) −→ α
[
L
.
Proof. (i) Suppose that Gal(E/L) Gal(E/K), i.e., for all α ∈ Gal(E/L) and β ∈
Gal(E/K), we have βαβ
−1
∈ Gal(E/L). Now if u ∈ L, then for any γ ∈ Gal(E/K) and
α ∈ Gal(E/L), γ(u) ∈ E satisﬁes
αγ(u) = γ(γ
−1
αγ(u)) = γ(u),
since γ
−1
αγ ∈ Gal(E/L); hence γ(u) ∈ E
Gal(E/L)
= L. By the Monomorphism Extension
Theorem 3.49, every monomorphism L −→ K ﬁxing K extends to a monomorphism E −→ K
which must have image E, so the above argument shows that L/K is normal.
Conversely, if L/K is normal, then for every ϕ ∈ Gal(E/K) and v ∈ L, ϕ(v) ∈ L, so for
every θ ∈ Gal(E/L), θ(ϕ(v)) = ϕ(v) and therefore
ϕ
−1
θϕ(v) = v.
This shows that ϕ
−1
θϕ ∈ Gal(E/L). Hence for every ϕ ∈ Gal(E/K),
ϕGal(E/L)ϕ
−1
= Gal(E/L),
which shows that Gal(E/L) Gal(E/K).
(ii) If α ∈ Gal(E/K), then αL = L since L/K is normal. Hence we can restrict α to an
automorphism of L,
α
[
L
: L −→ L; α
[
L
(u) = α(u).
58 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
Then α
[
L
is the identity function on L if and only if α ∈ Gal(E/L). It is easy to see that the
function
Gal(E/K) −→ Gal(L/K); α −→ α
[
L
is a group homomorphism whose kernel is Gal(E/L). Thus we obtain an injective homomor
phism
Gal(E/K)/ Gal(E/L) −→ Gal(L/K)
for which
[ Gal(E/K)/ Gal(E/L)[ =
[E : K]
[E : L]
= [L : K] = [ Gal(L/K)[.
Hence this homomorphism is an isomorphism.
4.5. The Galois Correspondence and the Main Theorem of Galois Theory
We are now almost ready to state our central result which describes the Galois Correspon
dence associated with a ﬁnite Galois extension. We will use the following notation. For a ﬁnite
Galois extension E/K, let
S(E/K) = the set of all subgroups of Gal(E/K);
F(E/K) = the set of all subextensions L/K of E/K.
Each of these sets is ordered by inclusion. Since every subgroup of a ﬁnite group is a ﬁnite
subset of a ﬁnite set, S(E/K) is also a ﬁnite set. Deﬁne two functions by
Φ
E/K
: F(E/K) −→ S(E/K); Φ
E/K
(L) = Gal(E/L),
Θ
E/K
: S(E/K) −→ F(E/K); Θ
E/K
(Γ) = E
Γ
.
4.18. Theorem (Main Theorem of Galois Theory). Let E/K be a ﬁnite Galois extension.
Then the functions Φ
E/K
and Θ
E/K
are mutually inverse bijections which are orderreversing.
F(E/K)
Φ
E/K
S(E/K)
Θ
E/K
Under this correspondence, normal subextensions of E/K correspond to normal subgroups of
Gal(E/K) and vice versa.
Proof. We know from Proposition 4.16 that for an extension L/K in F(E/K),
Θ
E/K
(Φ
E/K
(L)) = Θ
E/K
(Gal(E/L)) = E
Gal(E/L)
= L.
Also, by Proposition 4.14 for H ∈ S(E/K) we have
Φ
E/K
(Θ
E/K
(Γ)) = Φ
E/K
(E
Γ
) = Gal(E/E
Γ
) = Γ.
This shows that Φ
E/K
and Θ
E/K
are mutually inverse and so are inverse bijections.
Let L
1
/K, L
2
/K ∈ F(E/K) satisfy L
1
/K L
2
/K. Then Gal(E/L
2
) Gal(E/L
1
) since
L
1
⊆ L
2
and so if α ∈ Gal(E/L
2
) then α ﬁxes every element of L
1
. Hence Φ
E/K
(L
2
) Φ
E/K
(L
1
)
and so Φ
E/K
reverses order.
Similarly, if Γ
1
, Γ
2
∈ S(E/K) and Γ
1
Γ
2
, then E
Γ
2
E
Γ
1
since if w ∈ E
Γ
2
then it is ﬁxed
by every element of Γ
1
(as Γ
1
is a subset of Γ
2
). Hence Θ
E/K
reverses order.
There is an immediate consequence of the Main Theorem 4.18 which is closely related to
Proposition 3.20.
4.19. Corollary. Let E/K be a ﬁnite Galois extension. Then there are only ﬁnitely many
subextensions L/K E/K.
Proof. Since the set S(E/K) is ﬁnite, so is F(E/K).
4.5. THE GALOIS CORRESPONDENCE AND THE MAIN THEOREM OF GALOIS THEORY 59
When dealing with a ﬁnite Galois extension E/K, we indicate the subextensions in a diagram
with a line going upwards indicating an inclusion. We can also do this with the subgroups of
the Galois group Gal(E/K) with labels indicating the index of the subgroups. In eﬀect, the
Galois Correspondence inverts these diagrams.
4.20. Example. Figure 4.1 shows the Galois Correspondence for the extension of Exam
ple 3.30.
E = Q(
3
√
2, ζ
3
)
Q(
3
√
2)
2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Q(
3
√
2 ζ
3
)
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
Q(
3
√
2 ζ
2
3
)
2
3
Q(ζ
3
)
3
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Q
Φ
E/K
Gal(E/Q)
Gal(E/Q(ζ
3
))
3
2
Gal(E/Q(
3
√
2))
2
3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Gal(E/Q(
3
√
2 ζ
3
))
2
3
Gal(E/Q(
3
√
2 ζ
2
3
))
2
3
¦id¦
Figure 4.1. The Galois Correspondence for E = Q(
3
√
2, ζ
3
)/Q
As noted at the end of Example 3.38, the Galois group here is Gal(Q(
3
√
2, ζ
3
)/Q)
∼
= S
3
. It
is useful to make this isomorphism explicit. First take the 3 roots of the polynomial X
3
−2 for
which E is the splitting ﬁeld over Q; these are
3
√
2,
3
√
2 ζ
3
,
3
√
2 ζ
2
3
which we number in the order
they are listed. Then the monomorphisms id, α
0
, α
1
, α
t
1
, α
2
, α
t
2
extend to automorphisms of E,
each of which permutes these 3 roots in the following ways given by cycle notation:
α
0
= (2 3), α
1
= (1 2 3), α
t
1
= (1 2), α
2
= (1 3 2), α
t
2
= (1 3).
We ﬁnd that
Gal(E/Q(ζ
3
)) = ¦id, α
1
, α
2
¦
∼
=¦id, (1 2 3), (1 3 2)¦, Gal(E/Q(
3
√
2)) = ¦id, α
0
¦
∼
=¦id, (2 3)¦,
Gal(E/Q(
3
√
2 ζ
3
)) = ¦id, α
t
2
¦
∼
=¦id, (1 3)¦, Gal(E/Q(
3
√
2 ζ
2
3
)) = ¦id, α
t
1
¦
∼
=¦id, (1 2)¦.
Notice that ¦id, (1 2 3), (1 3 2)¦ S
3
and so Q(ζ
3
)/Q is a normal extension. Of course Q(ζ
3
)
is the splitting ﬁeld of X
3
−1 over Q.
60 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
4.6. Galois extensions inside the complex numbers and complex conjugation
When working with Galois extensions contained in the complex numbers it is often useful
to make use of complex conjugation as an element of a Galois group. Let E/Q be a ﬁnite Galois
extension with E/Q C/Q. Setting E
R
= R ∩ E, we have Q E
R
E.
4.21. Proposition. Complex conjugation ( ): C −→C restricts to an automorphism of E
over Q, ( )
E/Q
: E −→ E. Furthermore,
(i) ( )
E/Q
agrees with the identity function if and only if E
R
= E.
(ii) If E
R
= E, then
( )
E/Q
= ¦id, ( )
E/Q
¦
∼
= Z/2,
hence, E
R
= E
'( )
E/Q
`
and [E : E
R
] = 2.
Proof. Let u ∈ E. As E/Q is normal, minpoly
Q,u
(X) ∈ Q[X] splits over E, so all of its
complex roots lie in E. But ( ) permutes the roots of this minimal polynomial. Therefore ( )
maps E into itself.
(i) For z ∈ C, z = z if and only if z ∈ R.
(ii) Here [
( )
E/Q
[ = 2, and
E
'( )
E/Q
`
= ¦u ∈ E : u = u¦ = E
R
.
C
2
∞
R
∞
E
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
E
R
Q
We will usually write ( ) rather than ( )
E/Q
when no confusion seems likely to result.
4.22. Example. Consider the cyclotomic extension Q(ζ
8
)/Q where
ζ
8
= e
πi/4
=
1
√
2
+
1
√
2
i.
From Example 4.9 we know that
Q(ζ
8
) = Q(
√
2, i), [Q(ζ
8
) : Q] = 4,
and we easily see that
Q(ζ
8
)
R
= Q(
√
2).
4.7. Galois groups of even and odd permutations
We have seen that for a monic separable polynomial f(X) ∈ K[X] of degree n, the Galois
group of its splitting ﬁeld E over K can naturally be thought of as a subgroup of the symmetric
group S
n
, where we view the latter as permuting the roots of f(X). It is reasonable to ask when
Gal(E/K) A
n
rather than just Gal(E/K) S
n
.
We ﬁrst recall an interpretation of the sign of a permutation σ ∈ S
n
, sgn σ = ±1. For each
pair i, j with 1 i < j n, exactly one of the inequalities σ(i) < σ(j) or σ(j) < σ(i) must
4.7. GALOIS GROUPS OF EVEN AND ODD PERMUTATIONS 61
hold and the ratio (σ(j) − σ(i))/(j − i) is either positive or negative. It is easily veriﬁed that
the righthand side of the following equation must have value ±1 and so
(4.3) sgnσ =
¸
1i<jn
σ(j) −σ(i)
j −i
.
Note that this is sometimes used as the deﬁnition of sgnσ.
Suppose that f(X) factorizes over E as
f(X) = (X −u
1
) (X −u
n
) =
n
¸
i=1
(X −u
i
).
Here u
1
, . . . , u
n
∈ E are the roots of f(X); as we have assumed that f(X) is separable, the u
i
are distinct.
4.23. Definition. The discriminant of f(X) is
Discr(f(X)) =
¸
1i<jn
(u
j
−u
i
)
2
∈ E.
Notice that Discr(f(X)) = 0 since u
i
= u
j
if i = j.
4.24. Remark. There is an explicit formula for computing Discr(f(X)) is terms of its
coeﬃcients. For polynomials
p(X) = a
0
+a
1
X + +a
m
X
m
, q(X) = b
0
+b
1
X + +b
n
X
n
,
their resultant is the (m+n) (m+n) determinant (with n rows of a
i
’s and m rows of b
i
’s)
(4.4) Res(p(X), q(X)) =
a
0
a
1
. . . . . . . a
m
0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
0 a
0
a
1
. . . . . . . a
m
0 . . . . . . . . . . . . . . . . . . . . . . . . . . 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 a
0
a
1
. . . . . . . a
m
b
0
b
1
. . . . . . . b
n
0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
0 b
0
b
1
. . . . . . . b
n
0 . . . . . . . . . . . . . . . . . . . . . . . . . . 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 . . . . . . . . . . . . . . . . . . . . . . . . . . 0 b
0
b
1
. . . . . . . . . . . . b
n
.
Then if f(X) is monic with d = deg f(X),
(4.5) Discr(f(X)) = (−1)
d(d−1)/2
Res(f(X), f
t
(X)).
So for example,
Discr(X
3
+pX +q) = (−1)
3
Res(X
3
+pX +q, 3X
2
+p)
= (−1)
q p 0 1 0
0 q p 0 1
p 0 3 0 0
0 p 0 3 0
0 0 p 0 3
= −4p
3
−27q
2
.
Here are some low degree examples of discriminants obtained with the aid of Maple.
Discr(a
0
+a
1
X +X
2
) = −4a
0
+a
2
1
. n = 2:
Discr(a
0
+a
1
X +a
2
X
2
+X
3
) = −27a
2
0
+ 18a
0
a
1
a
2
+a
2
1
a
2
2
−4a
3
2
a
0
−4a
3
1
. n = 3:
62 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
n = 4: Discr(a
0
+a
1
X +a
2
X
2
+a
3
X
3
+X
4
) = 18a
3
a
3
1
a
2
−6a
2
3
a
2
1
a
0
−192a
3
a
1
a
2
0
−27a
4
1
+ 144a
2
a
2
3
a
2
0
+ 144a
0
a
2
1
a
2
+ 256a
3
0
−4a
3
3
a
3
1
−128a
2
2
a
2
0
+ 16a
4
2
a
0
−4a
3
2
a
2
1
+ 18a
3
3
a
1
a
2
a
0
−80a
3
a
1
a
2
2
a
0
−27a
4
3
a
2
0
+a
2
2
a
2
3
a
2
1
−4a
3
2
a
2
3
a
0
.
n = 5: Discr(a
0
+a
1
X +a
2
X
2
+a
3
X
3
+a
4
X
4
+X
5
) = 2250a
4
a
2
3
a
3
0
−36a
0
a
3
4
a
3
1
−128a
2
3
a
4
1
+ 2000a
2
0
a
3
a
2
1
−900a
1
a
3
3
a
2
0
−2500a
3
0
a
4
a
1
−50a
2
0
a
2
4
a
2
1
−900a
4
a
3
2
a
2
0
−27a
4
4
a
4
1
−3750a
3
a
2
a
3
0
+ 356a
3
2
a
2
2
a
4
a
1
a
0
+ 560a
3
a
2
2
a
2
4
a
2
0
−2050a
3
a
2
a
2
0
a
4
a
1
−80a
2
3
a
2
a
4
a
1
3
−630a
3
3
a
2
a
4
a
0
2
+ 825a
2
3
a
2
2
a
2
0
+ 16a
3
3
a
3
2
a
0
+ 2000a
2
a
2
4
a
3
0
−6a
2
2
a
2
4
a
3
1
−128a
2
2
a
4
4
a
2
0
+ 16a
4
2
a
3
4
a
0
−4a
3
2
a
3
4
a
2
1
+ 108a
5
3
a
2
0
+ 108a
5
2
a
0
−746a
3
a
2
a
0
a
4
2
a
1
2
−27a
4
2
a
2
1
+ 256a
5
4
a
3
0
−4a
3
3
a
2
2
a
2
1
+ 144a
3
a
2
2
a
3
1
+ 144a
2
4
a
4
1
a
3
+ 3125a
4
0
+ 256a
5
1
−72a
4
3
a
2
a
1
a
0
+ 18a
3
a
2
a
3
4
a
3
1
+ 560a
2
3
a
2
a
0
a
2
1
+ 16a
4
3
a
3
1
+ 18a
3
a
2
3
a
4
a
1
2
−72a
3
a
2
4
a
4
a
0
+ 144a
3
2
a
2
a
4
3
a
0
2
−192a
4
4
a
1
a
3
a
0
2
−630a
3
a
2
3
a
1
a
0
+ 24a
2
3
a
4
2
a
1
a
0
+a
3
2
a
2
2
a
4
2
a
1
2
−6a
4
3
a
1
2
a
3
2
a
0
−80a
3
a
2
2
a
4
3
a
1
a
0
−4a
3
2
a
2
3
a
4
2
a
0
+ 2250a
1
a
2
2
a
2
0
−1600a
3
a
3
4
a
3
0
−192a
4
a
4
1
a
2
−1600a
0
a
3
1
a
2
−4a
3
3
a
3
1
a
2
4
−27a
4
3
a
2
4
a
2
0
+ 1020a
4
2
a
3
2
a
0
2
a
1
+ 18a
3
3
a
2
a
4
2
a
0
a
1
+ 160a
2
a
4
3
a
0
2
a
1
+ 144a
2
a
4
4
a
0
a
1
2
+ 24a
4
a
1
2
a
3
3
a
0
+ 1020a
0
a
4
a
2
2
a
1
2
+ 160a
0
a
4
a
1
3
a
3
.
So for example,
Discr(X
5
+a
4
X
4
+a
0
) = a
3
0
(3125a
0
+ 256a
5
4
), Discr(X
5
+a
1
X +a
0
) = 256a
5
1
+ 3125a
4
0
.
4.25. Proposition. For every σ ∈ Gal(E/K),
σ(Discr(f(X))) = Discr(f(X)).
Hence Discr(f(X)) ∈ E
Gal(E/K)
= K.
Proof. For σ ∈ Gal(E/K) S
n
, we have
σ(Discr(f(X))) =
¸
1i<jn
(u
σ(j)
−u
σ(i)
)
2
=
¸
¸
1i<jn
(u
σ(j)
−u
σ(i)
)
¸
2
.
Now for each pair i, j with i < j,
σ(u
j
−u
i
) = u
σ(j)
−u
σ(i)
,
and by Equation (4.3)
(4.6)
¸
1i<jn
(u
σ(j)
−u
σ(i)
) = sgn σ
¸
1i<jn
(u
j
−u
i
) = (±1)
¸
1i<jn
(u
j
−u
i
).
Hence σ(Discr(f(X))) = Discr(f(X)). Since E
Gal(E/K)
= K, we have Discr(f(X)) ∈ K.
Now let
δ(f(X)) =
¸
1i<jn
(u
j
−u
i
) ∈ E.
Then δ(f(X))
2
= Discr(f(X)), so the square roots of Discr(f(X)) are ±δ(f(X)). Now consider
the eﬀect of σ ∈ Gal(E/K) on δ(f(X)) ∈ E. By Equation (4.6),
σ(δ(f(X))) = sgn σ δ(f(X)) = ±δ(f(X)).
If δ(f(X)) ∈ K, this means that sgnσ = 1. On the other hand, if δ(f(X)) / ∈ K then
K(δ(f(X))) = E
Gal(E/K)∩A
n
.
Of course [ Gal(E/K)/ Gal(E/K) ∩ A
n
[ = 2.
4.8. KAPLANSKY’S THEOREM 63
4.26. Proposition. The Galois group Gal(E/K) S
n
is contained in A
n
if and only if
Discr(f(X)) is a square in K.
4.27. Example. For the polynomials of Examples 6.40 and 6.42 we obtain
Discr(X
5
−35X
4
+ 7) = −4611833296875 = −3
3
5
6
7
4
29 157,
δ(X
5
−35X
4
+ 7) = ±5
3
3 7
2
√
3 29 157 i = ±18375
√
13659 i / ∈ Q;
Discr(X
5
+ 20X + 16) = 1024000000 = 2
16
5
6
,
δ(X
5
+ 20X + 16) = ±2
8
5
3
∈ Q.
4.8. Kaplansky’s Theorem
In this section we give a detailed account of the Galois theory of irreducible rational poly
nomials f(X) = X
4
+ aX
2
+ b ∈ Q[X]. The following result describes the Galois groups that
occur and the proof introduces some useful computational techniques.
4.28. Theorem (Kaplansky’s Theorem). Let f(X) = X
4
+aX
2
+b ∈ Q[X] be irreducible.
(i) If b is a square in Q then Gal(Q(f(X))/Q)
∼
= Z/2 Z/2.
(ii) If b(a
2
−4b) is a square in Q then Gal(Q(f(X))/Q)
∼
= Z/4.
(iii) If neither b nor b(a
2
−4b) is a square in Q then Gal(Q(f(X))/Q)
∼
= D
8
.
Proof. Let g(X) = X
2
+ aX + b ∈ Q[X]. Notice that g(X) must be irreducible since
otherwise f(X) would factorize, hence (a
2
−4b) is not a square in Q. Setting d = (a
2
−4b) ∈ Q
and taking δ to be a square root of d (so δ / ∈ Q), we ﬁnd that the roots of g(X) are (−a±δ)/2 / ∈ Q.
Then the roots of f(X) are ±u, ±v, where
u
2
=
(−a +δ)
2
, v
2
=
(−a −δ)
2
,
so the splitting ﬁeld of f(X) over Q is E = Q(u, v) which contains the quadratic extension
Q(δ)/Q. Since deg f(X) = 4, we also have 4 [ [E : Q]. In fact, since E is obtained by at most 3
successive quadratic extensions we also have [E : Q] [ 8.
(i) We have
(uv)
2
= u
2
v
2
=
a
2
−d
4
=
4b
4
= b,
hence uv is a square root of b which is in Q. Setting c = uv ∈ Q, we ﬁnd that v = c/u ∈ Q(u).
This shows that E = Q(u) and we have the following Galois tower.
E = Q(u)
2
Q(δ)
2
Q
In particular [E : Q] = 4 = [ Gal(E/Q)[. Notice that for the Galois extension Q(δ)/Q there
must be a normal subgroup N Gal(E/Q) with
Q(δ) = E
N
, Gal(Q(δ)/Q) = Gal(E/Q)/N.
Hence there is an element σ ∈ Gal(E/Q) for which σ(δ) = −δ. This element must also have
the eﬀects σ(u) = ±v and σ(v) = ±u. Given u we might as well choose v so that σ(u) = v.
64 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
There is also an element τ ∈ N for which τ(u) = −u and we also have τ(v) = −v. Notice that
if σ(v) = −u then easy calculation shows that
τσ(v) = στ(v) = u, τσ(δ) = στ(δ) = −δ,
hence we might as assume that σ(v) = u since if necessary we can replace our original choice
by τσ.
We now have
σ(u) =
c
u
, τ(u) = −u, τσ(u) = στ(u) = −
c
u
.
These satisfy
σ
2
= τ
2
= (στ)
2
= id = the identity, στ = τσ.
This shows that
Gal(Q(f(X))/Q) = Gal(E/Q) = ¦id, σ, τ, στ¦
∼
= Z/2 Z/2 = the Klein 4group.
(ii) If bd is a square in Q, then
(uvδ)
2
= u
2
v
2
d = bd,
which is a square in Q, so we can write uvδ = c ∈ Q or equivalently v = c/(uδ) ∈ Q(u) since
Q(δ) Q(u). This shows that E = Q(u, v) = Q(u) and again we have a Galois tower
E = Q(u)
2
Q(δ)
2
Q
with [E : Q] = 4 = [ Gal(E/Q)[.
Since Q(δ)/Q is Galois there is an element σ ∈ Gal(E/Q) with σ(δ) = −δ and this has the
eﬀect σ(u) = ±v; given u we might as well choose v so that σ(u) = v. Notice that
σ(v) =
c
σ(uδ)
= −
c
vδ
= −u,
so σ
2
(u) = −u. This shows that
Gal(Q(f(X))/Q) = Gal(E/Q) = ¦id, σ, σ
2
σ
3
¦
∼
= Z/4 = a cyclic group of order 4.
(iii) Suppose that d, b and bd are not squares in Q. By an easy calculation we ﬁnd that (uv)
2
= b,
so uv ∈ E is a square root of b in E. Suppose that uv ∈ Q(δ); then uv = p+qδ for some p, q ∈ Q.
By squaring we obtain
b = (p
2
+q
2
d) + 2pqδ,
and so pq = 0. We cannot have q = 0 since this would imply that b was a square in Q; if
p = 0 then b = q
2
d and so bd = (qd)
2
, implying that bd was a square in Q. Thus we have
Q(uv) ∩ Q(δ) = Q. A similar discussion shows that
Q(uvδ) ∩ Q(δ) = Q = Q(uvδ) ∩ Q(uv).
4.8. KAPLANSKY’S THEOREM 65
So we have a Galois tower which includes the following subﬁelds.
E = Q(u, v)
Q(uv, δ)
Q(δ)
2
Q(uv)
2
Q(uvδ)
2
Q
Choose
α ∈ Gal(E/Q(uv)) Gal(E/Q)
so that α(δ) = −δ. By renaming −v to v if necessary, we may assume that v = α(u) and so
u = α(v). Notice that α
2
= id.
Choose
β ∈ Gal(E/Q(δ)) Gal(E/Q)
with β(uv) = −uv. We must have either β(u) = −u or β(v) = −v, so by interchanging ±δ if
necessary we can assume that β(u) = −u and β(v) = v. Notice that β
2
= id.
Choose
γ ∈ Gal(E/Q(δ, uv)) Gal(E/Q)
so that γ(u) = −u. Then we must have γ(v) = −v since γ(uv) = uv. Notice that γ
2
= id.
Setting σ = αβ we ﬁnd σ(u) = −v and σ(v) = u. Then σ
2
= γ and σ has order 4. Also,
ασα = βσβ = σ
−1
.
The eight elements
id, σ, γ, σ
−1
, α, ασ, αγ, ασ
−1
form a group isomorphic to the dihedral group of order 8, D
8
. Therefore we have
Gal(Q(f(X))/Q) = Gal(E/Q)
∼
= D
8
,
and [E : Q] = 8. The corresponding Galois tower is
E = Q(u, v)
2
Q(uv, δ)
2
2
2
Q(δ)
2
Q(uv)
2
Q(uvδ)
2
Q
4.29. Example. We have the following Galois groups:
Gal(Q(X
4
+ 1)/Q)
∼
= Z/2 Z/2; Gal(Q(X
4
+ 4X
2
+ 2)/Q)
∼
= Z/4;
Gal(Q(X
4
+ 2X
2
+ 2)/Q)
∼
= D
8
.
66 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE
Exercises on Chapter 4
4.1. If f(X) ∈ K[X] is a separable polynomial, prove that the splitting ﬁeld of f(X) over K is
a ﬁnite Galois extension of K.
4.2. Let K be a ﬁeld for which char K = 2, 3 and suppose that f(X) ∈ K[x] is a cubic
polynomial.
(a) Show that there u, v ∈ K with u = 0 such that f(uX + v) = X
3
+ aX + b for some
a, b ∈ K. If f(X) is monic, deduce that a, b ∈ K; under what conditions is this always
true?
(b) If g(X) = X
3
+ aX + b ∈ K[x] is irreducible and E = K(g(X)) is its splitting ﬁeld
over K, explain why Gal(E/K) is isomorphic to one of the groups S
3
or A
3
.
(c) Continuing with the notation and assumptions of (b), suppose that w
1
, w
2
, w
3
are the
distinct roots of g(X) in E and let
∆ = (w
1
−w
2
)
2
(w
2
−w
3
)
2
(w
1
−w
3
)
2
∈ E.
Show that
∆ = −4b
3
−27a
2
,
and hence ∆ ∈ K. If δ = (w
1
−w
2
)(w
3
−w
3
)(w
1
−w
3
), show that
Gal(E/K)
∼
=
A
3
if δ ∈ K,
S
3
if δ / ∈ K.
[Hint: Consider K(δ) E and the eﬀect on the element δ of even and odd permutations
in Gal(E/K) S
3
.]
4.3. Show that f(X) = X
3
−3X+1 ∈ Q[X] is irreducible over Q, and show that its discriminant
is a square in Q. Prove that the Galois group of f(X) over Q is cyclic.
4.4. This is a revision exercise on ﬁnite groups of small order.
(a) Show that every nonabelian ﬁnite group has order at least 6.
(b) Let D
8
be the dihedral group with the eight elements
ι, α, α
2
, α
3
, β, βα, βα
2
, βα
3
satisfying
α
4
= ι, β
2
= ι, βαβ = α
−1
= α
3
.
Find all the normal subgroups of D
8
.
4.5. Use Kaplansky’s Theorem 4.28 to ﬁnd the Galois group of the splitting ﬁeld E of the
polynomial X
4
+ 3 ∈ Q[X] over Q. Determine all the subextensions F E for which F/Q is
Galois.
4.6. Find the Galois groups for each of the following extensions:
Q(X
3
−10)/Q; Q(
√
2)(X
3
−10)/Q(
√
2); Q(
√
3 i)(X
3
−10)/Q(
√
3 i);
Q(
√
23 i)(X
3
−X −1)/Q(
√
23 i); K(X
3
−X −1)/K for K = Q, Q(
√
5), Q(
√
5 i), Q(i).
4.7. Let p > 0 be a prime. Let K be a ﬁeld with char K = p. Suppose that 0 = a ∈ K and
f(X) = X
p
−a ∈ K[X]. Let L/K where L is a splitting ﬁeld for f(X) over K.
(a) Show that f(X) has p distinct roots in L. If u ∈ L is one such root, describe the
remaining roots and show that L contains p distinct pth roots of 1.
(b) Suppose that K contains p distinct pth roots of 1. Show that either f(X) is irreducible
over K or it factors into p distinct linear factors over K.
67
(c) Suppose that the only pth root of 1 in K is 1. Show that either f(X) is irreducible
over K or it has a root in K.
4.8. Let K be a ﬁeld of characteristic char K = p where p > 0 is a prime. Suppose that
0 = a ∈ K and f(X) = X
p
− a ∈ K[X]. Show that if f(X) has no root in K then it is
irreducible over K.
CHAPTER 5
Galois extensions for ﬁelds of positive characteristic
In this chapter we will investigate extensions of ﬁelds of positive characteristic, especially
ﬁnite ﬁelds. A thorough account of ﬁnite ﬁelds and their applications can be found in [6].
Throughout this chapter we will assume that K is a ﬁeld of prime characteristic p = char K >
0, containing the prime subﬁeld F
p
.
5.1. Finite ﬁelds
If K is a ﬁnite ﬁeld, then K is an F
p
vector space. Our ﬁrst goal is to count the elements of
K. Here is a more general result.
5.1. Lemma. Let F be a ﬁnite ﬁeld with q elements and let V be an Fvector space. Then
dim
F
V < ∞ if and only if V is ﬁnite in which case [V [ = q
dim
F
V
.
Proof. If d = dim
F
V < ∞, then for a basis v
1
, . . . , v
d
we can express each element v ∈ V
uniquely in the form v = t
1
v
1
+ + t
d
v
d
, where t
1
, . . . , t
d
∈ F. Clearly there are exactly q
d
such expressions, so [V [ = q
d
.
Conversely, if V is ﬁnite then any basis has ﬁnitely many elements and so dim
F
V < ∞.
5.2. Corollary. Let F be a ﬁnite ﬁeld and E/F an extension. Then E is ﬁnite if and only
if E/F is ﬁnite and then [E[ = [F[
[E:F]
.
5.3. Corollary. Let K be a ﬁnite ﬁeld. Then K/F
p
is ﬁnite and [K[ = p
[K:F
p
]
.
Our next task is to show that for each power p
d
there is a ﬁnite ﬁeld with p
d
elements. We
start with the algebraic closure F
p
of F
p
and consider the polynomial
Θ
p
d(X) = X
p
d
−X ∈ F
p
[X].
Notice that Θ
t
p
d
(X) = −1, hence by Proposition 3.55 every root of Θ
p
d(X) in F
p
is simple. There
fore by Corollary 1.35 Θ
p
d(X) must have exactly p
d
distinct roots in F
p
, say 0, u
1
, . . . , u
p
d
−1
.
Then in F
p
[X] we have
X
p
d
−X = X(X −u
1
) (X −u
p
d
−1
),
and each root is separable over F
p
. Let
F
p
d = ¦u ∈ F
p
: Θ
p
d(u) = 0¦ ⊆ F
p
, F
0
p
d
= ¦u ∈ F
p
d : u = 0¦.
Notice that u ∈ F
0
p
d
if and only if u
p
d
−1
= 1.
5.4. Proposition. For each d 1, F
p
d is a ﬁnite subﬁeld of F
p
with p
d
elements and
F
0
p
d
= F
p
d
. Furthermore, the extension F
p
d/F
p
is a separable splitting ﬁeld.
Proof. If u, v ∈ F
p
d then by the Idiot’s Binomial Theorem 1.10,
(u +v)
p
d
−(u +v) = (u
p
d
+v
p
d
) −(u +v) = (u
p
d
−u) + (v
p
d
−v) = 0,
(uv)
p
d
−uv = u
p
d
v
p
d
−uv = uv −uv = 0.
69
70 5. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC
Furthermore, if u = 0 then u
p
d
−1
= 1 and so u has multiplicative inverse u
p
d
−2
. Hence F
p
d F
p
.
Notice that F
p
F
p
d, so F
p
d/F
p
is a ﬁnite extension. In any ﬁeld the nonzero elements are
always invertible, hence F
0
p
d
= F
p
d
.
5.5. Definition. The ﬁnite subﬁeld F
p
d F
p
is called the Galois ﬁeld of order p
d
.
The notation GF(p
d
) is often used in place of F
p
d. Of course, F
p
1 = GF(p
1
) = GF(p) = F
p
and [F
p
d : F
p
] = d.
5.6. Proposition. Let d 1.
(i) F
p
d F
p
is the splitting subﬁeld for each of the polynomials X
p
d
− X and X
p
d
−1
− 1
over F
p
.
(ii) F
p
d F
p
is the unique subﬁeld with p
d
elements.
(iii) If F is any ﬁeld with p
d
elements then there is a monomorphism F −→F
p
with image
F
p
d, hence F
∼
= F
p
d.
Proof. (i) As F
p
d consists of exactly the roots of Θ
p
d(X) in F
p
, it is the splitting subﬁeld.
The nonzero elements of F
p
d are the roots of X
p
d
−1
−1, so F
p
d is also the splitting subﬁeld for
this polynomial.
(ii) Let F F
p
have p
d
elements. Notice that the nonzero elements of F form a group F
under multiplication. This group is abelian and has p
d
−1 elements, so by Lagrange’s Theorem,
each element u ∈ F
has order dividing p
d
− 1, therefore u
p
d
−1
= 1 and so u
p
d
= u. But this
means every element of F is a root of Θ
p
d(X) and so F F
p
d; equality follows since these
subﬁelds both have p
d
elements.
(iii) Apply the Monomorphism Extension Theorem 3.49 for K = L = F
p
and M = F. By (ii),
the image of the resulting monomorphism must be F
p
d, therefore F
∼
= F
p
d.
It is worth noting the following consequence of this result and the construction of F
p
d.
5.7. Corollary. Let K be a ﬁnite ﬁeld of characteristic p. Then K/F
p
is a ﬁnite Galois
extension.
5.8. Example. Consider the polynomial X
4
−X ∈ F
2
[X]. By inspection, in the ring F
2
[X]
we ﬁnd that
X
4
−X = X
4
+X = X(X
3
+ 1) = X(X + 1)(X
2
+X + 1).
Now X
2
+ X + 1 has no root in F
2
so it must be irreducible in F
2
[X]. Its splitting ﬁeld is a
quadratic extension F
2
(w)/F
2
where w is one of the roots of X
2
+X +1, the other being w+1
since the sum of the roots is the coeﬃcient of X. This tells us that every element of F
4
= F
2
(w)
can be uniquely expressed in the form a +bw with a, b ∈ F
2
. To calculate products we use the
fact that w
2
= w + 1, so for a, b, c, d ∈ F
2
we have
(a +bw)(c +dw) = ac + (ad +bc)w +bdw
2
= (ac +bd) + (ad +bc +bd)w.
5.9. Example. Consider the polynomial X
9
− X ∈ F
3
[X]. Let us ﬁnd an irreducible
polynomial of degree 2 in F
3
[X]. Notice that X
2
+1 has no root in F
3
, hence X
2
+1 ∈ F
3
[X] is
irreducible; so if u ∈ F
3
is a root of X
2
+ 1 then F
3
(u)/F
3
has degree 2 and F
3
(u) = F
9
. Every
element of F
9
can be uniquely expressed in the form a + bu with a, b ∈ F
3
. Multiplication is
carried out using the relation u
2
= −1 = 2.
By inspection, in the ring F
3
[X] we ﬁnd that
X
9
−X = X(X
8
−1) = (X
3
−X)(X
2
+ 1)(X
2
+X −1)(X
2
−X −1).
So X
2
+X−1 and X
2
−X−1 are also quadratic irreducibles in F
3
[X]. We can ﬁnd their roots
in F
9
using the quadratic formula since in F
3
we have 2
−1
= (−1)
−1
= −1. The discriminant of
5.1. FINITE FIELDS 71
X
2
+X −1 is
1 −4(−1) = 5 = 2 = u
2
,
so its roots are (−1)(−1 ±u) = 1 ±u. Similarly, the discriminant of X
2
−X −1 is
1 −4(−1) = 5 = 2 = u
2
and its roots are (−1)(1 ±u) = −1 ±u. Then we have
F
9
= F
3
(u) = F
3
(1 ±u) = F
3
(−1 ±u).
There are two issues we can now clarify.
5.10. Proposition. Let F
p
m and F
p
n be two Galois ﬁelds of characteristic p. Then F
p
m
F
p
n if and only if m [ n.
Proof. If F
p
m F
p
n, then by Corollary 5.2,
p
n
= (p
m
)
[F
p
n:F
p
m]
= p
m[F
p
n:F
p
m]
,
so m [ n.
If m [ n, write n = km with k 1. Then for u ∈ F
p
m we have u
p
m
= u, so
u
p
n
= u
p
mk
= (u
p
m
)
p
m(k−1)
= u
p
m(k−1)
= = u
p
m
= u.
Hence u ∈ F
p
n and therefore F
p
m F
p
n.
This means that we can think of the Galois ﬁelds F
p
n as ordered by divisibility of n. The dia
gram of subﬁelds for F
p
24 can be seen in Figure 5.1 which shows extensions with no intermediate
subextensions.
F
p
24
F
p
8 F
p
12
F
p
4 F
p
6
F
p
2
F
p
3
F
p
Figure 5.1. The subﬁelds of F
p
24
5.11. Theorem. The algebraic closure of F
p
is the union of all the Galois ﬁelds of charac
teristic p,
F
p
=
¸
n1
F
p
n.
Furthermore, each element u ∈ F
p
is separable over F
p
.
Proof. Let u ∈ F
p
. Then u is algebraic over F
p
and the extension F
p
(u)/F
p
is ﬁnite. Hence
by Corollary 5.2, F
p
(u) F
p
is a ﬁnite subﬁeld. Proposition 5.10 now implies that F
p
(u) = F
p
n
for some n. The separability statement follows from Corollary 5.7.
We will require a useful fact about Galois ﬁelds.
72 5. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC
5.12. Proposition. The group of units F
p
d
in F
p
d is cyclic.
This is a special case of a more general result about arbitrary ﬁelds.
5.13. Proposition. Let K be a ﬁeld. Then every ﬁnite subgroup U K
is cyclic.
Proof. Use Corollary 1.35 and Lemma 1.46.
5.14. Definition. w ∈ F
p
d
is called a primitive root if it is a primitive (p
d
− 1)th root of
unity, i.e., its order in the group F
p
d
is (p
d
−1), hence 'w` = F
p
d
.
5.15. Remark. Unfortunately the word primitive has two confusingly similar uses in the
context of ﬁnite ﬁelds. Indeed, some authors use the term primitive element for what we have
called a primitive root, but that conﬂicts with our usage, although as we will in the next result,
every primitive root is indeed a primitive element in our sense!
5.16. Proposition. The extension of Galois ﬁelds F
p
nd/F
p
d is simple, i.e., F
p
nd = F
p
d(u)
for some u ∈ F
p
nd.
Proof. By Proposition 5.12, F
p
nd has a primitive root w say. Then every element of
F
p
nd can be expressed as a polynomial in w, so F
p
nd F
p
d(w) F
p
nd. This implies that
F
p
nd = F
p
d(w).
5.17. Remark. This completes the proof of the Primitive Element Theorem 3.75 which we
had previously only established for inﬁnite ﬁelds.
5.18. Example. In Example 5.8 we ﬁnd that F
4
= F
2
(w) has the two primitive roots w
and w + 1.
5.19. Example. In Example 5.9 we have F
9
= F
3
(u) and F
9
is cyclic of order 8. Since
ϕ(8) = 4, there are four primitive roots and these are the roots of the polynomials X
2
+X −1
and X
2
−X −1 which we found to be ±1 ±u.
We record a fact that is very important in Number Theory.
5.20. Proposition. Let p > 0 be an odd prime.
(i) If p ≡ 1 (mod 4), the polynomial X
2
+ 1 ∈ F
p
[X] has two roots in F
p
.
(ii) If p ≡ 3 (mod 4) the polynomial X
2
+1 ∈ F
p
[X] is irreducible, so F
p
2
∼
= F
p
[X]/(X
2
+1).
Proof. (i) We have 4 [ (p −1) = [F
p
[, so if u ∈ F
p
is a generator of this cyclic group, the
order of u
[F
×
p
[/4
is 4, hence this is a root of X
2
+ 1 (the other root is −u
[F
×
p
[/4
).
(ii) If v ∈ F
p
is a root of X
2
+1 then v has order 4 in F
p
. But then 4 [ (p −1) = [F
p
[, which is
impossible since p −1 ≡ 2 (mod 4).
Here is a generalization of Proposition 5.20.
5.21. Proposition. F
p
d contains a primitive nth root of unity if and only if p
d
≡ 1 (mod n)
and p n.
5.2. Galois groups of ﬁnite ﬁelds and Frobenius mappings
Consider an extension of Galois ﬁelds F
p
nd/F
p
d. By Proposition 5.6(i), Corollary 5.7 and
Proposition 3.73, this extension is Galois and
[ Gal(F
p
nd/F
p
d)[ = [F
p
nd : F
p
d] = n.
We next introduce an important element of the Galois group Gal(F
p
nd/F
p
d).
5.2. GALOIS GROUPS OF FINITE FIELDS AND FROBENIUS MAPPINGS 73
5.22. Definition. The (relative) Frobenius mapping for the extension F
p
nd/F
p
d is the func
tion F
d
: F
p
nd −→F
p
nd given by F
d
(t) = t
p
d
.
5.23. Proposition. The relative Frobenius mapping F
d
: F
p
nd −→ F
p
nd is an automor
phism of F
p
nd that ﬁxes the elements of F
p
d, so F
d
∈ Gal(F
p
nd/F
p
d). The order of F
d
is n, so
Gal(F
p
nd/F
p
d) = 'F
d
`, the cyclic group generated by F
d
.
Proof. For u, v ∈ F
p
nd, we have the identities
F
d
(u +v) = (u +v)
p
d
= u
p
d
+v
p
d
, F
d
(uv) = (uv)
p
d
= u
p
d
v
p
d
,
so F
d
is a ring homomorphism. Also, for u ∈ F
p
d we have
F
d
(u) = u
p
d
= u,
so F
d
ﬁxes the elements of F
p
d. To see that F
d
is an automorphism, notice that the composition
power F
n
d
= F
d
◦ ◦ F
d
(with n factors) satisﬁes
F
n
d
(t) = t
p
nd
= t
for all t ∈ F
p
nd, hence F
n
d
= id. Then F
d
is invertible with inverse F
−1
d
= F
n−1
d
. This also shows
that the order of F
d
in the group Aut
F
p
d
(F
p
nd) is at most n. Suppose the order is k with k n;
then every element u ∈ F
p
nd satisﬁes the equation F
k
d
(u) = u which expands to u
p
kd
= u, hence
u ∈ F
p
kd. But this can only be true if k = n.
Frobenius mappings exist on the algebraic closure F
p
. For d 1, consider the function
F
d
: F
p
−→F
p
; F
d
(t) = t
p
d
.
5.24. Proposition. Let d 1.
(i) F
d
: F
p
−→ F
p
is an automorphism of F
p
which ﬁxes the elements of F
p
d. In fact for
u ∈ F
p
, F
d
(u) = u if and only if u ∈ F
p
d.
(ii) The restriction of F
d
to the Galois subﬁeld F
p
dn agrees with the relative Frobenius
mapping F
d
: F
p
nd −→F
p
nd.
(ii) If k 1, then F
k
d
= F
kd
. Hence in the automorphism group Aut
F
p
d
(F
p
), F
d
has inﬁnite
order, so Aut
F
p
d
(F
p
) is inﬁnite.
Proof. This is left as an exercise.
The Frobenius mapping F = F
1
is often called the absolute Frobenius mapping since it exists
as an element of each of the groups Aut
F
p
(F
p
) and Aut
F
p
(F
p
n) = Gal(F
p
n/F
p
) for every n 1.
In Gal(F
p
nd/F
p
d) = 'F
d
`, for each k with k [ n there is the cyclic subgroup
F
k
d
of order
[
F
k
d
[ = n/k.
5.25. Proposition. For k [ n, the ﬁxed subﬁeld of
F
k
d
in F
p
nd is F
'F
k
d
`
p
nd
= F
p
dk.
F
p
nd
n/k
F
'F
k
d
`
p
nd
= F
p
dk
k
F
p
d
Proof. For u ∈ F
p
nd we have F
k
d
(u) = u
p
dk
, hence F
k
d
(u) = u if and only if u ∈ F
p
dk.
74 5. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC
Figure 5.2 shows the subgroup diagram corresponding to the lattice of subﬁelds of F
p
24
shown in Figure 5.1.
Gal(F
p
24/F
p
) = 'F`
∼
= Z/24
F
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
F
3
F
4
F
6
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
F
8
F
12
F
24
= ¦id¦
Figure 5.2. The subgroups of the Galois groups of F
p
24/F
p
5.3. The trace and norm mappings
For an extension of Galois ﬁelds F
p
nd/F
p
d, consider the function T
F
p
nd
/F
p
d
: F
p
nd −→ F
p
nd
deﬁned by
T
F
p
nd
/F
p
d
(u) = u +u
p
d
+u
p
2d
+ +u
p
(n−1)d
= u + F
d
(u) + F
2d
(u) + + F
(n−1)d
(u).
Notice that
F
d
(T
F
p
nd
/F
p
d
(u)) = u
p
d
+u
p
2d
+u
p
3d
+ +u
p
nd
= u
p
d
+u
p
2d
+u
p
3d
+ +u
p
(n−1)d
+u = T
F
p
nd
/F
p
d
(u).
So by Proposition 5.24(i), T
F
p
nd
/F
p
d
(u) ∈ F
p
d. If we modify T
F
p
nd
/F
p
d
to have codomain F
p
d, we
obtain the relative trace
Tr
F
p
nd
/F
p
d
: F
p
nd −→F
p
d; Tr
F
p
nd
/F
p
d
(u) = u +u
p
d
+u
p
2d
+ +u
p
(n−1)d
.
5.26. Proposition. The relative trace Tr
F
p
nd
/F
p
d
is a surjective F
p
dlinear mapping and
whose kernel is an F
p
dvector subspace of dimension n −1.
Proof. Clearly Tr
F
p
nd
/F
p
d
is additive. For t ∈ F
p
d we have t
p
d
= t, so F
p
dlinearity follows
from the formula
tu + (tu)
p
d
+ (tu)
p
2d
+ + (tu)
p
(n−1)d
= tu +tu
p
d
+tu
p
2d
+ +tu
p
(n−1)d
.
To see that Tr
F
p
nd
/F
p
d
is surjective, notice that Tr
F
p
nd
/F
p
d
(u) = 0 if and only if u is a root of
the polynomial
X +X
p
d
+X
p
2d
+ +X
p
(n−1)d
∈ F
p
d[X]
which has degree p
(n−1)d
and so has at most p
(n−1)d
< p
nd
roots in F
p
nd. This means that
ker Tr
F
p
nd
/F
p
d
cannot be the whole of F
p
nd. Tr
F
p
nd
/F
p
d
is surjective since its codomain has
dimension 1.
75
There is a multiplicative version of this construction. Consider the function
N: F
p
nd
−→F
p
nd
for which
N(u) = uu
p
d
u
p
2d
u
p
(n−1)d
= uF
d
(u) F
2d
(u) F
(n−1)d
(u).
Then we have
F
d
(N(u)) = u
p
d
u
p
2d
u
p
3d
u
p
nd
= u
p
d
u
p
2d
u
p
3d
u
p
(n−1)d
u
= uu
p
d
u
p
2d
u
p
3d
u
p
(n−1)d
= N(u).
So by Proposition 5.24(i), N(u) ∈ F
p
d. By redeﬁning the codomain we obtain the relative norm
Norm
F
p
nd
/F
p
d
: F
p
nd
−→F
p
d
; Norm
F
p
nd
/F
p
d
(u) = uu
p
d
u
p
2d
u
p
(n−1)d
.
5.27. Proposition. The relative norm Norm
F
p
nd
/F
p
d
is a surjective group homomorphism.
Proof. Multiplicativity is obvious. The kernel of Norm
F
p
nd
/F
p
d
consists of the roots in F
p
nd
of the polynomial
X
1+p
d
++p
(n−1)d
−1 ∈ F
p
d[X],
so
[ ker Norm
F
p
nd
/F
p
d
[ 1 +p
d
+ +p
(n−1)d
=
p
nd
−1
p
d
−1
.
Hence
[ imNorm
F
p
nd
/F
p
d
[ =
p
nd
−1
[ ker Norm
F
p
nd
/F
p
d
[
p
d
−1.
Since imNorm
F
p
nd
/F
p
d
F
p
d
, we also have
[ imNorm
F
p
nd
/F
p
d
[ p
d
−1,
therefore
imNorm
F
p
nd
/F
p
d
= F
p
d
.
Exercises on Chapter 5
5.1. Show that Proposition 5.13 also applies to an integral domain in place of a ﬁeld.
5.2. What happens to Theorem 5.20 if we try to take p = 2.
5.3. Let f(X) ∈ F
p
d[X] be an irreducible polynomial with deg f(X) = n. Find the splitting ﬁeld
of f(X). Deduce that for any other irreducible polynomial g(X) ∈ F
p
d[X] with deg g(X) = n,
the splitting ﬁelds of f(X) and g(X) over F
p
d agree.
5.4. Find the smallest Galois ﬁelds containing all the roots of the following polynomials, in
each case ﬁnd a primitive root of this Galois ﬁeld:
(a) X
8
−1 ∈ F
41
[X]; (b) X
8
−1 ∈ F
5
[X]; (c) X
8
−1 ∈ F
11
[X]; (d) X
8
−1 ∈ F
2
[X].
5.5. Let w ∈ F
p
d
be a primitive root. If < d, show that w / ∈ F
p
. Deduce that deg
F
p
w = d
and d [ ϕ(p
d
−1).
5.6. Let p > 0 be a prime. Suppose that d 1, and K/F
p
d is an extension. For a ∈ K, let
g
a
(X) = X
p
d
−X −a ∈ K[X].
76 5. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC
(a) If the polynomial g
a
(X) is irreducible over K, show that the splitting ﬁeld E of g
a
(X)
over K is separable and Gal(E/K)
∼
= F
p
d. [Hint: show that if u ∈ E is a root of g
a
(X)
in an extension E/K, then so is u +t for every t ∈ F
p
.]
(b) If d = 1, show that g
a
(X) is irreducible over K if and only if it has no root in K.
(c) If K is a ﬁnite ﬁeld and d > 1, explain why g
a
(X) can never be irreducible over K.
5.7. Let p be an odd prime, d 1 and write q = p
d
.
(a) Consider ¦±1¦ = ¦1, −1¦ as a group under multiplication. Show that there is a unique
group homomorphism λ
q
: F
q
−→ ¦±1¦ which is characterized by the requirement
that for every u ∈ F
q
, λ
q
(u) = 1 if and only if u = v
2
for some v ∈ F
q
. Is λ
q
always
surjective?
(b) Consider the set of all squares in F
q
,
Σ
q
= ¦u
2
∈ F
q
: u ∈ F
q
¦ ⊆ F
q
.
Show that the number of elements of Σ
q
is [Σ
q
[ = (q +1)/2. Deduce that if t ∈ F
q
then
the set
t −Σ
q
= ¦t −u
2
∈ F
q
: u ∈ F
q
¦
has [t −Σ
q
[ = (q + 1)/2 elements.
(c) If t ∈ F
q
, show that
[Σ
q
∩ (t −Σ
q
)[ 1.
Deduce that every element of F
q
is either a square or can be written as the sum of two
squares.
(d) Deduce that the equation x
2
+y
2
+z
2
= 0 has at least one nontrivial solution in F
q
.
(e) What can you say about the case p = 2?
CHAPTER 6
A Galois Miscellany
In this chapter we will explore some miscellaneous topics in Galois Theory. Historically,
Galois Theory has always been an important tool in Number Theory and Algebra, stimulating
the development of subjects such as Group Theory, Ring Theory and such diverse areas as Dif
ferential Equations, Complex Analysis and Algebraic Geometry. Many of the ideas introduced
in this chapter are of great importance in these and other mathematical areas.
6.1. A proof of the Fundamental Theorem of Algebra
We will prove the Fundamental Theorem of Algebra for the complex numbers C. This proof
is essentially due to Gauss but he did not use the historically more recent Sylow theory. It is
interesting to compare the proof below with others which use the topology of the plane and circle
or Complex Analysis; our proof only uses the connectivity of the real line (via the Intermediate
Value Theorem) together with explicit calculations in C involving square roots.
6.1. Theorem (The Fundamental Theorem of Algebra). The ﬁeld of complex numbers C is
algebraically closed and R = C.
Proof. We know that [C : R] = 2, so C/R is algebraic. Let p(X) ∈ C[X] be irreducible.
Then any root u of p(X) in the algebraic closure C is algebraic over R, so in C[X] we have
p(X) [ minpoly
R,u
(X). The splitting ﬁeld of p(X) over C is contained in the splitting ﬁeld E of
minpoly
R,u
(X)(X
2
+ 1) over R. Since C E, we have 2 [ [E : R] and so 2 [ [ Gal(E/R)[.
Now consider a 2Sylow subgroup P Gal(E/R) and recall that [ Gal(E/R)[/[P[ is odd.
For the ﬁxed subﬁeld of P, we have
[E
P
: R] =
[ Gal(E/R)[
[P[
,
which shows that E
P
/R has odd degree. The Primitive Element Theorem 3.75 allows us to
write E
P
= R(v) for some v whose minimal polynomial over R must also have odd degree.
But by the Intermediate Value Theorem, every real polynomial of odd degree has a real root,
so irreducibility implies that v has degree 1 over R and therefore E
P
= R. This shows that
Gal(E/R) = P, hence Gal(E/R) is a 2group.
As C/R is a Galois extension, we can consider the normal subgroup Gal(E/C)Gal(E/R) for
which [ Gal(E/R)[ = 2 [ Gal(E/C)[. We must show that [ Gal(E/C)[ = 1, so suppose not. From
the theory of 2groups, there is a normal subgroup N Gal(E/C) of index 2, so we can consider
the Galois extension E
N
/C of degree 2. But from known properties of C (see Proposition 3.29),
every quadratic aX
2
+ bX + c ∈ C[X] has complex roots (because we can ﬁnd square roots
of every complex number). So we cannot have an irreducible quadratic polynomial in C[X].
Therefore [ Gal(E/C)[ = 1 and E = C.
6.2. Cyclotomic extensions
We begin by discussing the situation for cyclotomic extensions over Q using material dis
cussed in Section 1.3. Let ζ
n
= e
2πi/n
, the standard primitive nth root of 1 in C. In Theo
rem 1.43, it was claimed that the irreducible polynomial over Q which has ζ
n
as a root was the
77
78 6. A GALOIS MISCELLANY
nth cyclotomic polynomial
Φ
n
(X) =
¸
t=1,...,n−1
gcd(t,n)=1
(X −ζ
t
n
).
6.2. Theorem. Let n 2. Then
• Q(ζ
n
) = Q[X]/(Φ
n
(X));
• [Q(ζ
n
) : Q] = ϕ(n);
• Gal(Q(ζ
n
)/Q)
∼
= (Z/n)
, where the element t
n
∈ (Z/n)
acts on Q(ζ
n
) by t
n
ζ
n
= ζ
t
n
.
Proof. Since the complex roots of Φ
n
(X) are the powers ζ
t
n
with t = 1, . . . , n − 1 and
gcd(t, n) = 1, Q(ζ
n
) is the splitting ﬁeld of Φ
n
(X) over Q and indeed Q(ζ
n
) = Q(ζ
t
n
) whenever t
has the above properties and so ζ
t
n
is a primitive nth root of unity. The main step in the proof
is to show that Φ
n
(X) ∈ Z[X] is irreducible. To do this we will show that every power ζ
t
n
as
above is actually a Galois conjugate of ζ
n
over Q, therefore
Φ
n
(X) = minpoly
Q,ζ
n
(X) = minpoly
Q,ζ
t
n
(X)
and hence Φ
n
(X) is irreducible.
Consider
Z(ζ
n
) = ¦a
0
+a
1
ζ
n
+ +a
r
ζ
r
n
: r 0, a
j
∈ Z¦ ⊆ Q(ζ
n
).
Then Z(ζ
n
) is a subring of Q(ζ
n
) and so is an integral domain. Its group of units contains the
cyclic subgroup 'ζ
n
` of order n.
Let p > 0 be a prime which does not divide n. Let P Z(ζ
n
) be a maximal ideal which
contains p; then the quotient ring Z(ζ
n
)/P is a ﬁeld of characteristic p. In fact, it is a ﬁnite
ﬁeld, say F
p
d for some d. Let π: Z(ζ
n
) −→F
p
d be the quotient homomorphism.
Inside the group of units of Z(ζ
n
) is the subgroup of powers of ζ
n
, 'ζ
n
` Z(ζ
n
)
; this
is a cyclic subgroup of order n. We claim that when restricted to 'ζ
n
`, π gives an injective
group homomorphism, π
t
: 'ζ
n
` −→ F
p
d
. To see this, suppose that π
t
(ζ
r
n
) = 1 for some r =
1, 2, . . . , n −1; then ζ
r
n
−1 ∈ P. By elementary Group Theory we can assume that r [ n and so
p r. On factoring we have
(ζ
n
−1)(ζ
r−1
n
+ +ζ
n
+ 1) ≡ (ζ
n
−1)r (mod P),
so ζ
n
− 1 ∈ P or r ∈ P since maximal ideals are prime. But Z ∩ P = (p) and so r / ∈ P, hence
ζ
n
−1 ∈ P. Recalling that
ζ
n−1
n
+ +ζ
n
+ 1 = 0,
we see that n ∈ P and hence p [ n, thus contradicting our original assumption on n. So π
t
is
injective.
Writing u = π
t
(u), we can consider the eﬀect of the absolute Frobenius map F: F
p
d −→F
p
d
on ζ
t
n
= ζ
t
n
,
F(ζ
t
n
) = (ζ
t
n
)
p
= ζ
tp
n
.
This shows that in the Galois extension F
p
d/F
p
, ζ
t
n
is conjugate to ζ
tp
n
; by iterating this we ﬁnd
that ζ
t
n
is conjugate to every power of the form ζ
tp
k
n
.
Now let t = 1, . . . , n −1 and gcd(t, n) = 1. Suppose there is a factorization
Φ
n
(X) = f(X) minpoly
Q,ζ
n
(X)
for some monic polynomial f(X) ∈ Z[X] and f(ζ
t
n
) = 0. Consider the prime power factorization
t = p
r
1
1
p
r
m
m
, where the p
j
are primes with 2 p
1
< < p
m
and r
j
1 with. Since
gcd(t, n) = 1 we also have p
j
n s.
Now consider a maximal ideal P
1
Z[ζ
n
] containing p
1
. Reducing modulo P
1
and working
in the resulting extension F
p
d
1
1
/F
p
1
, we ﬁnd that ζ
n
is conjugate to ζ
p
r
1
1
n
. By separability and
6.2. CYCLOTOMIC EXTENSIONS 79
the fact that the reduction map π
1
: Z[ζ
n
] −→F
p
d
1
1
is injective on the powers of ζ
n
, we ﬁnd that
f(ζ
p
r
1
1
n
) = 0 and so f(ζ
p
r
1
1
n
) = 0 in Z[ζ
n
]. This shows that minpoly
Q,ζ
n
(ζ
p
r
1
1
n
) = 0 and so ζ
p
r
1
1
n
is
conjugate to ζ
n
.
Repeating this argument starting with ζ
p
r
1
1
n
and using the prime p
2
we ﬁnd that
minpoly
Q,ζ
n
(ζ
p
r
1
1
p
r
2
2
n
) = 0
and so ζ
p
r
1
1
p
r
2
2
n
is conjugate to ζ
n
. Continuing in this fashion, for each j = 1, . . . , m we have
minpoly
Q,ζ
n
(ζ
p
r
1
1
p
r
2
2
p
r
j
j
n
) = 0
and so ζ
p
r
1
1
p
r
j
j
n
is conjugate to ζ
n
. When j = m, this shows that minpoly
Q,ζ
n
(ζ
t
n
) = 0. Hence
ζ
t
n
is conjugate to ζ
n
in the extension Q(ζ
n
)/Q.
6.3. Theorem. For n > 2, consider the cyclotomic extension Q(ζ
n
)/Q where ζ
n
= e
2πi/n
.
Then Q(ζ
n
)
R
= Q(ζ
n
). Furthermore,
Q(ζ
n
)
R
= Q(ζ
n
)
/( ))
= Q(ζ
n
+ζ
n
) = Q(cos(2π/n)),
and
[Q(cos(2π/n)) : Q] =
ϕ(n)
2
.
Proof. Recall that
Gal(Q(ζ
n
)/Q)
∼
= Z/n
,
where the residue class of r acts by sending ζ
n
to ζ
r
n
. Complex conjugation corresponds to the
residue class of −1 ≡ n −1 (mod n). Making use of the identities
e
θi
= cos θ + sin θ i, cos θ =
1
2
(e
θi
+e
−θi
),
we obtain
cos(2π/n) =
1
2
(ζ
n
+ζ
n
) =
1
2
(ζ
n
+ζ
−1
n
).
Complex conjugation ﬁxes each of the real numbers cos(2πk/n) for k = 1, 2, . . . , n − 1. The
residue class of r acts by sending cos(2π/n) to cos(2πr/n); it is elementary to show that
cos(2πr/n) = cos(2π/n) unless r ≡ 1 (mod n). Hence
( )
= ¦id, ( )¦ = Gal(Q(cos(2π/n))/Q).
Thus we have
Q(ζ
n
)
'( )`
= Q(cos(2π/n)),
and so [Q(cos(2π/n)) : Q] = ϕ(n)/2. Notice that ζ
n
is a root of the polynomial
X
2
−2 cos(2π/n)X + 1 ∈ Q(cos(2π/n))[X],
so we also have
(6.1) minpoly
Q(cos(2π/n)),ζ
n
(X) = X
2
−2 cos(2π/n)X + 1.
6.4. Example. We have
[Q(ζ
24
) : Q] = ϕ(24) = 8
and
Gal(Q(ζ
24
)/Q)
∼
= Z/2 Z/2 Z/2.
80 6. A GALOIS MISCELLANY
Proof. By Theorem 1.43 we have [Q(ζ
24
) : Q] = 8. Also,
ζ
6
24
= i, ζ
3
24
=
√
2
2
+
√
2
2
i, ζ
8
24
= −
1
2
+
√
3
2
i,
and all of these numbers are in Q(ζ
24
), hence Q(
√
2,
√
3, i) Q(ζ
24
). It is easy to check that
[Q(
√
2,
√
3, i) : Q] = 8,
which implies that
Q(ζ
24
) = Q(
√
2,
√
3, i).
Using this we ﬁnd that
Gal(Q(ζ
24
)/Q)
∼
= Z/2 Z/2 Z/2.
We also have cos(2π/24) = cos(π/12) ∈ Q(ζ
24
). Since
cos(2π/12) = cos(π/6) =
√
3
2
,
we have
2 cos
2
(π/12) −1 =
√
3
2
and so
4 cos
4
(π/12) −4 cos
2
(π/12) + 1 =
3
4
,
giving
16 cos
4
(π/12) −16 cos
2
(π/12) + 1 = 0.
Then
16X
4
−16X
2
+ 1 = 16 minpoly
Q,cos(π/12)
(X).
Note that case (i) of Kaplansky’s Theorem 4.28 applies to the polynomial minpoly
Q,cos(π/12)
(X).
For this example, Gal(Q(ζ
24
)/Q) has 2
3
−1 = 7 subgroups of each of the orders 2 and 4; it
is an interesting exercise to ﬁnd them all together with their ﬁxed subﬁelds.
6.5. Remark. The minimal polynomial for cos(π/12) can also be found as follows. We have
Φ
24
(ζ
24
) = 0, hence since
Φ
24
(X) = X
8
−X
4
+ 1,
we obtain
ζ
8
24
−ζ
4
24
+ 1 = 0.
Then after multiplying by ζ
−4
24
we have
ζ
4
24
−1 +ζ
−4
24
= 0,
giving
(ζ
4
24
+ζ
−4
24
) −1 = 0.
Now
(ζ
24
+ζ
−1
24
)
4
= (ζ
4
24
+ζ
−4
24
) + 4(ζ
2
24
+ζ
−2
24
) + 6,
hence
ζ
4
24
+ζ
−4
24
= (ζ
24
+ζ
−1
24
)
4
−4(ζ
2
24
+ζ
−2
24
) −6.
Similarly,
(ζ
24
+ζ
−1
24
)
2
= ζ
2
24
+ζ
−2
24
+ 2,
so
ζ
2
24
+ζ
−2
24
= (ζ
24
+ζ
−1
24
)
2
−2.
Combining these we have
(ζ
24
+ζ
−1
24
)
4
−4(ζ
24
+ζ
−1
24
)
2
+ 1 = 0,
6.2. CYCLOTOMIC EXTENSIONS 81
and so
16 cos
4
(π/12) −16 cos
2
(π/12) + 1 = 0.
This method will work for any n where ϕ(n) is even, i.e., when n > 2.
6.6. Remark. The polynomial that expresses cos nθ as a polynomial in cos θ is the n
th Chebsyhev polynomial of the ﬁrst kind T
n
(X) ∈ Z[X]. Here are the ﬁrst few of these
polynomials:
T
2
(X) = 2X
2
−1, T
3
(X) = 4X
3
−3X,
T
4
(X) = 8X
4
−8X
2
+ 1, T
5
(X) = 16X
5
−20X
3
+ 5X,
T
6
(X) = 32X
6
−48X
4
+ 18X
2
−1, T
7
(X) = 64X
7
−112X
5
+ 56X
3
−7X.
These form a system of orthogonal polynomials which can be computed in Maple using the
command orthopoly[T](n,X).
Now let K be a ﬁeld with characteristic char K n. The polynomial Φ
n
(X) has integer
coeﬃcients, so we can view it as an element of K[X] since either Q K or F
p
K and we
can reduce the coeﬃcients modulo p. In either case it can happen that Φ
n
(X) factors in K[X].
However, we can still describe the splitting ﬁeld of X
n
−1 over K and its Galois group.
6.7. Theorem. If char K n, then the splitting ﬁeld of X
n
−1 over K is K(ζ), where ζ ∈ K
is a primitive nth root of unity. The Galois group Gal(K(ζ)/K) is isomorphic to a subgroup
of (Z/n)
, hence it is abelian with order dividing ϕ(n).
Proof. Working in K, we know that Φ
n
(ζ) = 0, hence the roots of minpoly
K,ζ
(X) ∈ K[X]
are primitive roots of 1. So X
n
− 1 splits over K(ζ) and each element α ∈ Gal(K(ζ)/K) has
the action α(ζ) = ζ
r
α
, where gcd(r
α
, n) = 1. Hence Gal(K(ζ)/K) is isomorphic to a subgroup
of Gal(Q(ζ
n
)/Q)
∼
= (Z/n)
which implies that it is abelian and its order divides ϕ(n).
6.8. Remark. When p = char K > 0, this Galois group only depends on the largest subﬁeld
of K which is algebraic over F
p
. For example, if K = F
p
d(T) then the value of d is the crucial
factor. The precise outcome can be determined with the aid of Proposition 5.21.
6.9. Example. We have the following splitting ﬁelds and Galois groups.
(i) The splitting ﬁeld of X
4
−1 over F
3
(T) is F
9
(T) and
Gal(F
9
(T)/F
3
(T))
∼
= (Z/4)
∼
= Z/2.
(ii) By Proposition 5.20, X
4
−1 splits over F
5
(T) and the Galois group Gal(F
5
(T)/F
5
(T))
is trivial.
Proof. (i) By Proposition 5.20, X
4
−1 is separable over F
3
(T) and has irreducible factors
(X − 1), (X + 1) and (X
2
+ 1). The splitting ﬁeld of (X
2
+ 1) over F
3
is F
9
= F
3
(ζ), where
ζ
2
+ 1 = 0, so (X
2
+ 1) splits over F
9
(T). Also,
Gal(F
9
/F
3
)
∼
= (Z/4)
∼
= Z/2,
with generator σ satisfying σ(ζ) = ζ
−1
= −ζ. This generator clearly extends to an automor
phism of F
9
(T) which ﬁxes T.
(ii) By Proposition 5.20, X
4
−1 splits over F
5
.
82 6. A GALOIS MISCELLANY
6.3. Artin’s Theorem on linear independence of characters
Let G be a group and K a ﬁeld.
6.10. Definition. A group homomorphism χ: G −→ K
is called a character of G with
values in K.
6.11. Example. Given any ring homomorphism ϕ: R −→ K we obtain a character of R
in K by restricting ϕ to a map χ
ϕ
: R
−→ K
.
6.12. Example. Given an automorphism α: K −→ K, χ
α
: K
−→ K
is a character of
K
in K.
6.13. Example. Let E/K be a Galois extension and σ ∈ Gal(E/K). Then χ
σ
: E
−→ E
is a character.
6.14. Definition. Let χ
1
, . . . , χ
n
be characters of a group G in a ﬁeld K. Then χ
1
, . . . , χ
n
are linearly independent if for t
1
, . . . , t
n
∈ K,
t
1
χ
1
+ +t
n
χ
n
= 0 =⇒ t
1
= = t
n
= 0.
If χ
1
, . . . , χ
n
are not linearly independent then they are linearly dependent.
In this deﬁnition, the functional equation means that for all g ∈ G,
t
1
χ
1
(g) + +t
n
χ
n
(g) = 0.
6.15. Theorem (Artin’s Theorem). Let χ
1
, . . . , χ
n
be distinct characters of a group G in a
ﬁeld K. Then χ
1
, . . . , χ
n
are linearly independent.
Proof. We proceed by induction on n. For n = 1 the result is easily veriﬁed. For the
inductive assumption, suppose that it holds for any n k.
Let χ
1
, . . . , χ
k+1
be a set of k + 1 distinct characters for which there are t
1
, . . . , t
k+1
∈ K
not all zero and such that
(6.2) t
1
χ
1
+ +t
k+1
χ
k+1
= 0.
If one of the t
i
is zero, say t
r
= 0, then χ
1
, . . . , χ
r−1
, χ
r+1
, . . . , χ
k+1
is linearly dependent,
contradicting the inductive assumption. Hence all of the t
i
must be nonzero. As χ
1
= χ
2
, there
must be an element g
0
∈ G for which χ
1
(g
0
) = χ
2
(g
0
). So for all g ∈ G, Equation (6.2) applied
to g
0
g yields
t
1
χ
1
(g
0
g) + +t
k+1
χ
k+1
(g
0
g) = 0,
and therefore since χ
j
(g
0
g) = χ
j
(g
0
)χ
j
(g), we see that
t
1
χ
1
(g
0
)χ
1
+ +t
k+1
χ
k+1
(g
0
)χ
k+1
= 0.
Multiplying Equation (6.2) by χ
1
(g
0
) and subtracting gives
t
2
(χ
2
(g
0
) −χ
1
(g
0
))χ
2
+t
3
(χ
3
(g
0
) −χ
1
(g
0
))χ
3
+ +t
k+1
(χ
k+1
(g
0
) −χ
1
(g
0
))χ
k+1
= 0,
in which the coeﬃcient t
2
(χ
2
(g
0
)−χ
1
(g
0
)) is not zero. Hence χ
2
, . . . , χ
k+1
is linearly dependent,
again contradicting the inductive assumption. So χ
1
, . . . , χ
k+1
is linearly independent, which
demonstrates the inductive step.
6.16. Corollary. Suppose that α
1
, . . . , α
n
are distinct automorphisms of the ﬁeld K. Let
t
1
, . . . , t
n
∈ K be a sequence of elements, not all of which are 0. Then there is a z ∈ K for
which
t
1
α
1
(z) + +t
n
α
n
(z) = 0.
Hence the Klinear transformation t
1
α
1
+ +t
n
α
n
: K −→ K is nontrivial.
6.3. ARTIN’S THEOREM ON LINEAR INDEPENDENCE OF CHARACTERS 83
6.17. Corollary. Let E/K be a ﬁnite Galois extension of degree n and let α
1
, . . . , α
n
be the distinct elements of Gal(E/K). Then the function α
1
+ + α
n
: E −→ E is a non
trivial Klinear transformation whose image is contained in K. Hence the associated Klinear
transformation
Tr
E/K
: E −→ K; Tr
E/K
(x) = α
1
(x) + +α
n
(x)
is surjective.
The function Tr
E/K
: E −→ K is called the trace mapping of E/K.
Proof. First note that for x ∈ E and γ ∈ Gal(E/K),
γ(α
1
(x) + +α
n
(x)) = γα
1
(x) + +γα
n
(x) = α
1
(x) + +α
n
(x),
since the list γα
1
, . . . , γα
n
is the same as α
1
, . . . , α
n
apart from its order. Hence,
α
1
(x) + +α
n
(x) ∈ E
Gal(E/K)
= K.
The rest of the statement follows directly from Corollary 6.16.
Suppose that E/K is a ﬁnite Galois extension with cyclic Galois group Gal(E/K) = 'σ` of
order n. For each u ∈ E
, the element uσ(u) σ
n−1
(u) ∈ E satisﬁes
σ(uσ(u) σ
n−1
(u)) = σ(u) σ
n−1
(u)σ
n
(u) = σ(u) σ
n−1
(u)u,
hence in uσ(u) σ
n−1
(u) ∈ E
/σ)
= K. Now using this we deﬁne a group homomorphism
N
E/K
: E
−→ K
; N
E/K
(u) = uσ(u) σ
n−1
(u).
N
E/K
is called the norm mapping for E/K and generalizes the norm mapping for ﬁnite ﬁelds
of Section 5.3.
There is another homomorphism
δ
E/K
: E
−→ E
; δ
E/K
(u) = uσ(u)
−1
.
Notice that for u ∈ E
,
N
E/K
(δ
E/K
(u)) = (uσ(u)
−1
)(σ(u)σ
2
(u)
−1
σ
n−1
(u)σ
n
(u)
−1
) = 1,
since σ
n
(u) = u. So imδ
E/K
ker N
E/K
. Our next result is an important generalization of
Proposition 5.27.
6.18. Theorem (Hilbert’s Theorem 90). Let E/K be a ﬁnite Galois extension with cyclic
Galois group Gal(E/K) = 'σ` of order n. Then imδ
E/K
= ker N
E/K
. Explicitly, if u ∈ E
and
uσ(u) σ
n−1
(u) = 1, then there is a v ∈ E
such that u = vσ(v)
−1
.
Proof. Let u ∈ ker N
E/K
.
The characters σ
k
: E
−→ E
with k = 0, 1, . . . , n−1 are distinct and linearly independent
by Artin’s Theorem 6.15. Consider the function
id +uσ +uσ(u)σ
2
+ +uσ(u) σ
n−2
(u)σ
n−1
: E
−→ E.
This cannot be identically zero, so for some w ∈ E, the element
v = w +uσ(w) +uσ(u)σ
2
(w) + +uσ(u) σ
n−2
(u)σ
n−1
(w)
is nonzero. Notice that
uσ(v) = uσ(w) +uσ(u)σ
2
(w) +uσ(u)σ
2
(u)σ
3
(w) + +uσ(u)σ
2
(u) σ
n−1
(u)σ
n
(w) = v,
since
uσ(u)σ
2
(u) σ
n−1
(u)σ
n
(w) = w.
Thus we have u = vσ(v)
−1
as required.
84 6. A GALOIS MISCELLANY
6.4. Simple radical extensions
In this section we will investigate splitting ﬁelds of polynomials of the form X
n
−a, where
char K n. We call these simple radical extensions and later in Deﬁnition 6.33 we introduce a
more general notion of radical extension.
6.19. Proposition. Let f(X) = X
n
−a ∈ K[X] be irreducible and separable over K. Then
the splitting ﬁeld of f(X) over K has the form K(u, ζ), where u is a root of f(X) and ζ is a
primitive nth root of 1.
6.20. Corollary. If K contains a primitive nth root of 1, ζ, then the splitting ﬁeld of
f(X) = X
n
− a over K has the form K(u), where u is a root of f(X). The Galois group
Gal(K(u)/K) is cyclic of order n with a generator σ for which σ(u) = ζu.
In the more general situation of Proposition 6.19,
¦id¦ Gal(K(ζ, u)/K(ζ)) Gal(K(ζ, u)/K),
where Gal(K(ζ, u)/K(ζ)) is cyclic and
Gal(K(ζ)/K))
∼
= Gal(K(ζ, u)/K))/ Gal(K(ζ, u)/K(ζ))
is abelian. The Galois Correspondence identiﬁes the following towers of subﬁelds and subgroups.
K(ζ, u)
Gal(K(ζ, u)/K)
K(ζ)
Gal(K(ζ, u)/K(ζ))
K
¦id¦
6.21. Definition. Let K be a ﬁeld with char K n and which contains a primitive nth
root of 1, ζ say. Then L/K is a simple nKummer extension if L = K(u) where u
n
= a for
some a ∈ K. L/K is an (iterated) nKummer extension if L = K(u
1
, . . . , u
k
) where u
n
1
=
a
1
, . . . , u
n
k
= a
k
for some elements a
1
, . . . , a
k
∈ K.
Note that in this deﬁnition we do not require the polynomials X
n
− a
j
∈ K[X] to be
irreducible.
6.22. Proposition. Let K(u)/K be a simple nKummer extension. Then K(u)/K is a
Galois extension and Gal(K(u)/K) is cyclic with order dividing n.
Proof. Suppose that u
n
= a ∈ K. Then in K[X] we have
X
n
−a = (X −u)(X −ζu) (X −ζ
n−1
u).
Clearly the roots of X
n
−a are distinct and so K(u)/K is separable over K; in fact, K(u) is a
splitting ﬁeld of X
n
−a over K. This means that K(u)/K is Galois.
For each α ∈ Gal(K(u)/K) we have α(u) = ζ
r
α
u for some r
α
= 0, 1 . . . , n −1. Notice that
for β ∈ Gal(K(u)/K),
βα(u) = β(ζ
r
α
u) = ζ
r
α
β(u) = ζ
r
α
ζ
r
β
u = ζ
r
α
+r
β
u,
and so r
βα
= r
α
+r
β
. Hence the function
ρ: Gal(K(u)/K) −→ 'ζ` ; ρ(α) = ζ
r
α
,
is a group homomorphism. As 'ζ` is cyclic of order n, Lagrange’s Theorem implies that the
image of ρ has order dividing n. Since every element of Gal(K(u)/K) is determined by its eﬀect
6.5. SOLVABILITY AND RADICAL EXTENSIONS 85
on u, ρ is injective, hence [ Gal(K(u)/K)[ divides n. In fact, Gal(K(u)/K) is cyclic since every
subgroup of a cyclic group is cyclic.
6.23. Example. Let n 1 and q ∈ Q. Then Q(ζ
n
,
n
√
q)/Q(ζ
n
) is a simple nKummer
extension.
6.24. Example. Q(i,
√
2)/Q(i) is a simple 4Kummer extension with Gal(Q(i,
√
2)/Q(i))
cyclic of order 2.
Proof. We have (
√
2)
4
−4 = 0, but
X
4
−4 = (X
2
−2)(X
2
+ 2),
and
X
2
−2 = minpoly
Q(i),
√
2
(X).
The corresponding group homomorphism ρ: Gal(Q(i)(
√
2)/Q(i)) −→ 'i` has image
imρ = ¦1, −1¦ 'i` .
Here is a converse to Proposition 6.22.
6.25. Proposition. Suppose that char K n and there is an element ζ ∈ K which is a
primitive nth root of unity. If E/K is a ﬁnite Galois extension with cyclic Galois group of
order n, then there is an element a ∈ E such that E = K(a) and a is a root of a polynomial of
the form X
n
−b with b ∈ K. Hence E/K is a simple nKummer extension.
Proof. We have
N
E/K
(ζ
−1
) = ζ
−n
= 1,
so by Hilbert’s Theorem 6.18, there is an element a ∈ E for which ζ
−1
= aσ(a)
−1
. Then
σ(a) = ζa and the elements σ
k
(a) = ζ
k
a for k = 0, 1, . . . , n − 1 are distinct, so they must be
the n conjugates of a. Also note that
X
n
−a
n
= (X −a)(X −ζa) (X −ζ
n−1
a) = (X −a)(X −σ(a)) (X −σ
n−1
(a)),
hence a
n
∈ K since it is ﬁxed by σ. Since K(a) E, this shows that
n = [K(a) : K] [E : K] = n
and therefore
[K(a) : K] = [E : K] = n,
whence K(a) = E.
6.5. Solvability and radical extensions
We begin by recalling some ideas about groups, see [3, 5] for further details.
6.26. Definition. A group G is solvable, soluble or soluable if there is a chain of subgroups
(called a subnormal series)
¦1¦ = G
G
−1
G
1
G
0
= G
in which G
k+1
G
k
and each composition factor G
k
/G
k+1
is abelian; we usually write
¦1¦ = G
G
−1
G
1
G
0
= G.
If each composition factor is a cyclic group of prime order the subnormal series is called a
composition series. A group which is not solvable is called insolvable.
86 6. A GALOIS MISCELLANY
6.27. Remark. It is a standard result that we can always reﬁne (i.e., add extra terms) a
subnormal series of a solvable group to obtain a composition series. The primes appearing as
well as the number of times each occurs are all determined by [G[, only their order varying for
diﬀerent composition series.
6.28. Example. Let G be a ﬁnite abelian group. Then G is solvable.
6.29. Example. Let G be a ﬁnite pgroup, where p is a prime. Then G is solvable.
In fact, for a ﬁnite pgroup G, there is always a normal subgroup of a pgroup with index p,
so in this case we can assume each quotient G
k
/G
k+1
is cyclic of order p.
6.30. Proposition. Let G be a group.
(i) If G is solvable then every subgroup H G and every quotient group G/N is solvable.
(ii) If N G and G/N are solvable then so is G.
In the opposite direction we can sometimes see that a group is insolvable. Recall that a
group is simple if it has no nontrivial proper normal subgroups.
6.31. Proposition. Let G be a ﬁnite group. Then G is insolvable if any of the following
conditions holds:
(i) G contains a subgroup which is a nonabelian simple group.
(ii) G has a quotient group which is a nonabelian simple group.
(iii) G has a composition series in which one of the terms is a nonabelian simple group.
6.32. Example. For n 5, the alternating and symmetric groups A
n
and S
n
are insolvable.
Proof. This follows from the fact that if n 5, A
n
is a simple group and A
n
S
n
with
quotient group S
n
/A
n
∼
= Z/2.
Now we explain how this relates to ﬁelds and their extensions. Let K be a ﬁeld and L/K a
ﬁnite extension. For simplicity, we assume also that char K = 0.
6.33. Definition. L/K is a radical extension of K if it has the form L = K(a
1
, a
2
, . . . , a
n
)
with
a
d
k
k
∈ K(a
1
, a
2
, . . . , a
k−1
)
for some d
k
1. Thus every element of L is expressible in terms of iterated roots of elements
of K.
We will need the following Lemma and its Corollary. According to [4], several text books
make subtle errors or omissions related to this result, so beware when reading other sources!
6.34. Lemma. Let L/K be a ﬁnite Galois extensions and let L(u)/L be a radical extension.
Let E/L be an extension where E is a splitting ﬁeld for the polynomial minpoly
K,u
(X) ∈ L[X].
Then E/L is a radical Galois extension. In particular, if L/K is a radical Galois extension
then so is E/K.
Proof. Suppose that u
d
= a ∈ L with a = 0. Then X
d
− a has a d distinct roots in E,
and if v is any other root then (v/u)
d
= 1, so there are d distinct dth roots of unity in E.
Hence there is a primitive dth root of unity ζ ∈ E and the subﬁeld L(ζ, u) E is normal over
L, so L(ζ, u)/L is a radical Galois extension. But L(ζ, u)/K need not be Galois. However, if
u = u
1
, . . . , u
t
∈ E are the distinct roots of minpoly
K,u
(X) in E, then
E = L(ζ, u, u
1
, . . . , u
t
).
But this is clearly a radical extension of L.
6.5. SOLVABILITY AND RADICAL EXTENSIONS 87
If L/K is a radical Galois extension, say L = K(a
1
, . . . , a
n
), then
E = L(a
1
, . . . , a
n
, ζ, u, u
1
, . . . , u
t
),
which is a radical Galois extension of K.
6.35. Corollary. If L/K is a radical extension then it is contained in a radical Galois
extension L
t
/K.
Proof. Writing L = K(a
1
, a
2
, . . . , a
n
) as in Deﬁnition 6.33, this is proved by induction
on n using Lemma 6.34.
In the next deﬁnition, the word Galois is superﬂuous because of the preceding results.
6.36. Definition. If L is the splitting ﬁeld of a polynomial f(X) ∈ K[X], then f(X) is
solvable by radicals over K if L is contained in a radical (Galois) extension of K.
6.37. Definition. L/K is solvable if L L
t
where L
t
/K is a ﬁnite radical Galois extension
of K.
6.38. Theorem. Let E/K be a ﬁnite Galois extension. Then E/K is solvable if and only
if the group Gal(E/K) is solvable.
Proof. Suppose that E E
t
where E
t
/K is a ﬁnite radical Galois extension, so
E
t
= K(ζ, u
1
, . . . , u
m
),
where ζ
d
= 1, u
d
1
1
∈ K(ζ) and u
d
r
r
∈ K(ζ, u
1
, . . . , u
r−1
) for r = 2, . . . , m with d
1
d
m
[ d. If
G
r
Gal(E
t
/K) and
(E
t
)
G
r
= K(ζ, u
1
, . . . , u
r
),
with
(E
t
)
G
0
= K(ζ),
then
¦1¦ = G
m
G
m−1
G
0
Gal(E
t
/K)
and
G
r−1
/G
r
∼
= Gal(K(ζ, u
1
, . . . , u
r
)/K(ζ, u
1
, . . . , u
r−1
)),
which is abelian by Proposition 6.22. Hence Gal(E
t
/K) is solvable, and since Gal(E/K) is a
quotient group of Gal(E
t
/K), is also solvable by Proposition 6.30.
Now suppose that Gal(E/K) is solvable and let n = [ Gal(E/K)[. Let E
t
be the splitting
ﬁeld of X
n
−1 over E, so E
t
contains a primitive nth root of unity ζ and therefore it contains
a primitive dth root of unity for every divisor d of n. Now Gal(E
t
/E) Gal(E
t
/K) and
by Theorem 6.7, Gal(E
t
/E) is abelian. Also, Gal(E
t
/K)/ Gal(E
t
/E)
∼
= Gal(E/K) which is
solvable, so Gal(E
t
/K) is solvable by Proposition 6.30. We will now show that E
t
/K is a
radical extension.
Clearly K(ζ)/K is radical. Then Gal(E
t
/K(ζ)) Gal(E
t
/K) is solvable. Let
¦1¦ = G
G
−1
G
1
G
0
= Gal(E
t
/K(ζ))
be a composition series. The extension (E
t
)
G
1
/K(ζ) is radical by Proposition 6.25. Similarly,
each extension (E
t
)
G
k+1
/(E
t
)
G
k
is radical. Hence E
t
/K(ζ) is radical, as is E
t
/K.
6.39. Example. The Galois group of the extension Q(ζ
3
,
3
√
2)/Q is solvable.
88 6. A GALOIS MISCELLANY
Proof. We have already studied this extension in Example 3.30 and 4.20. Clearly Q(ζ
3
,
3
√
2)
is a radical extension of Q and
Q(ζ
3
,
3
√
2) = Q(ζ
3
)(
3
√
2).
We know that Gal(Q(ζ
3
,
3
√
2)/Q)
∼
= S
3
, where we identify each element of the Galois group with
a permutation of the three roots of X
3
−2 in Q(ζ
3
,
3
√
2) which we list in the order
3
√
2,
3
√
2 ζ
3
,
3
√
2 ζ
2
3
.
We have the following towers of subﬁelds and subgroups related under the Galois Correspon
dence.
Q(ζ
3
,
3
√
2)
_
S
3
.
Q(ζ
3
) = Q(ζ
3
,
3
√
2)
A
3
3
A
3
= Gal(Q(ζ
3
,
3
√
2)/Q(ζ
3
))
2
Q
2
¦id¦
3
Here Q(ζ
3
)/Q is itself a Galois extension and A
3
S
3
. Notice that A
3
∼
= Z/3 and S
3
/A
3
∼
= Z/2,
so we have the following composition series for S
3
:
¦id¦ A
3
S
3
.
It is also interesting to reverse the question and ask whether there are extensions which are
not solvable. This was a famous problem pursued for several hundred years. To ﬁnd examples,
we ﬁrst recall that the smallest nonabelian simple group is A
5
which has order 60. We should
therefore expect to look for a polynomial of degree at least 5 to ﬁnd a Galois group for a splitting
ﬁeld to be simple or occur as a composition factor of such a Galois group. Here is an explicit
example over Q.
6.40. Example. The splitting ﬁeld of the polynomial f(X) = X
5
−35X
4
+7 ∈ Q[X] is not
solvable.
Proof. Let E C be the splitting ﬁeld of f(X) over Q. Using the Eisenstein Test 1.38
with p = 7, we ﬁnd that f(X) is irreducible over Q. By Theorem 4.8(iii), 5 divides the order of
Gal(E/Q), so by Cauchy’s Lemma this group contains an element of order 5.
Now observe that
f
t
(X) = 5X
4
−140X
3
= 5X
3
(X −28), f
tt
(X) = 20X
4
−420X
2
= 20X
2
(X −21).
There are two turning points, namely a maximum at x = 0 and a minimum at x = 28. Then
f(0) = 7 > 0 > f(28) = −4302585,
hence there are three real roots of f(X) and two nonreal complex ones. Then complex conju
gation restricts to an element of order 2 in Gal(E/Q) which interchanges the nonreal roots and
ﬁxes the others. If we list the roots of f(X) as u
1
, u
2
, u
3
, u
4
, u
5
with u
1
, u
2
being the nonreal
roots, then the transposition (1 2) ∈ S
5
corresponds to this element. Furthermore, the only
elements of S
5
of order 5 are 5cycles; by taking an appropriate power we can assume that there
is a 5cycle of the form (1 2 3 4 5) corresponding to an element of Gal(E/Q) which we can view
as a subgroup of S
5
. The next lemma shows that Gal(E/Q)
∼
= S
5
.
6.41. Lemma. Let n 1. Suppose that H S
n
and H contains the elements (1 2) and
(1 2 n). Then H = S
n
.
The proof is left as an exercise. This completes the veriﬁcation of Example 6.40.
6.6. SYMMETRIC FUNCTIONS 89
It is worth remarking that the most extreme version of this occurs when we ask for a Galois
group which is simple. There has been a great deal of research activity on this question in the
past few decades, but apparently not all simple groups are known to occur as Galois groups of
extensions of Q or other ﬁnite subextensions of C/Q. Here is an example whose Galois group
is A
5
; this is veriﬁed using Proposition 4.26.
6.42. Example. The Galois group of f(X) = X
5
+20X+16 over Qis Gal(Q(f(X))/Q)
∼
= A
5
,
hence it is not solvable.
6.6. Symmetric functions
Let k be a ﬁeld. Consider the polynomial ring on n indeterminates k[X
1
, . . . , X
n
] and its
ﬁeld of fractions K = k(X
1
, . . . , X
n
). Each permutation σ ∈ S
n
acts on k[X
1
, . . . , X
n
] by
σ f(X
1
, . . . , X
n
) = f
σ
(X
1
, . . . , X
n
) = f(X
σ(1)
, . . . , X
σ(n)
).
Viewed as a function σ : k[X
1
, . . . , X
n
] −→k[X
1
, . . . , X
n
] is a ring isomorphism; this extends to
a ring isomorphism σ : k(X
1
, . . . , X
n
) −→ k(X
1
, . . . , X
n
). Varying σ we obtain actions of the
group S
n
on k[X
1
, . . . , X
n
] and k(X
1
, . . . , X
n
) by ring isomorphisms ﬁxing k and in the latter
case it is by ﬁeld automorphisms ﬁxing k.
6.43. Definition. The ﬁeld of symmetric functions on n indeterminates is
Sym
n
(k) = k(X
1
, . . . , X
n
)
S
n
k(X
1
, . . . , X
n
).
So if f(X
1
, . . . , X
n
) ∈ k(X
1
, . . . , X
n
), then
f(X
1
, . . . , X
n
) ∈ Sym
n
(k) ⇐⇒ ∀σ ∈ S
n
f(X
1
, . . . , X
n
) = f(X
σ(1)
, . . . , X
σ(n)
).
6.44. Theorem. The extension k(X
1
, . . . , X
n
)/ Sym
n
(k) is a ﬁnite Galois extension for
which Gal(k(X
1
, . . . , X
n
)/ Sym
n
(k))
∼
= S
n
.
Proof. There are elements of k[X
1
, . . . , X
n
] ⊆ k(X
1
, . . . , X
n
) called elementary symmetric
functions,
e
k
=
¸
i
1
<i
2
<<i
k
X
i
1
X
i
2
X
i
k
,
where 1 k n. It is easy to see that for every σ ∈ S
n
, e
σ
k
= e
k
, so e
k
∈ Sym
n
(k). Working in
the ring k(X
1
, . . . , X
n
)[Y ] we have
f
n
(Y ) = Y
n
−e
1
Y
n−1
+ + (−1)
n−1
e
n−1
Y + (−1)
n
e
n
= 0,
hence the roots of this polynomial are the X
i
. So k(X
1
, . . . , X
n
) is the splitting ﬁeld of f
n
(Y )
over Sym
n
(k). Now S
n
Gal(k(X
1
, . . . , X
n
)/ Sym
n
(k)), hence
[k(X
1
, . . . , X
n
) : Sym
n
(k)] = [ Gal(k(X
1
, . . . , X
n
)/ Sym
n
(k))[ [S
n
[ = n!.
But as every element of Gal(k(X
1
, . . . , X
n
)/ Sym
n
(k)) permutes the roots of f
n
(Y ) and is de
termined by this permutation, we also have
n! [ Gal(k(X
1
, . . . , X
n
)/ Sym
n
(k))[.
Combining these inequalities we obtain [ Gal(k(X
1
, . . . , X
n
)/ Sym
n
(k))[ = n! and therefore
Gal(k(X
1
, . . . , X
n
)/ Sym
n
(k)) = S
n
.
6.45. Remark. In fact, this proof shows that the extension k(X
1
, . . . , X
n
)/k(e
1
, . . . , e
n
) is
Galois of degree n!. Since k(e
1
, . . . , e
n
) Sym
n
(k) we can also deduce that k(e
1
, . . . , e
n
) =
Sym
n
(k). Hence every element of Sym
n
(k) is a rational function in the e
i
. Analogous results
are true for polynomials, i.e.,
k[X
1
, . . . , X
n
]
S
n
= k[e
1
, . . . , e
n
].
90 6. A GALOIS MISCELLANY
6.46. Corollary. If n 5, the extension k(X
1
, . . . , X
n
)/ Sym
n
(k) is not solvable.
Exercises on Chapter 6
6.1. Let p > 0 be a prime and G a group of order [G[ = p
n
for some n 1. Show by induction
on n that there is a normal subgroup N G with [N[ = p
n−1
. [Hint: what do you know about
the centre of G? Use this information to produce a quotient group of smaller order than G.]
6.2. Let K be a ﬁeld for which char K = 2 and n 1 be odd. If K contains a primitive nth
root of unity, show that then K contains a primitive 2nth root of unity.
6.3. Find all values of n 1 for which ϕ(n) [ 4. Using this, determine which roots of unity lie
in the following ﬁelds:
Q(i), Q(
√
2 i), Q(
√
3 i), Q(
√
5 i).
6.4. (a) Describe the elements of (Z/24)
explicitly and verify that this group is isomorphic
to Z/2 Z/2 Z/2. Describe the eﬀect of each element on Q(ζ
24
) and Q(cos(π/12)) under the
action described in Theorem 6.2.
(b) Determine the group (Z/20)
and describe the eﬀect of each of its elements on Q(ζ
20
) and
Q(cos(π/10)) under the action described in Theorem 6.2.
6.5. Let n 1.
(a) What can you say about sin(2π/n) and Gal(Q(sin(2π/n))/Q))?
(b) Determine sin(π/12) and Gal(Q(sin(π/12))/Q)).
6.6. In this question, work in the cyclotomic ﬁeld Q(ζ
5
) where ζ
5
= e
2πi/5
.
(a) Describe the Galois group Gal(Q(ζ
5
)/Q) and its action on Q(ζ
5
).
(b) Determine the minimal polynomial of cos(2π/5) over Q. Hence show that
cos(2π/5) =
−1 +
√
5
4
.
For which other angles θ is cos θ a root of this minimal polynomial? What is the value
of sin(2π/5) ?
(c) Find the tower of subﬁelds of Q(ζ
5
) and express them as ﬁxed ﬁelds of subgroups of
Gal(Q(ζ
5
)/Q).
6.7. In this question, let p be an odd prime and let ζ
p
= e
2πi/p
∈ Q(ζ
p
) C.
(a) Consider the product
ξ =
(p−1)/2
¸
r=1
(ζ
r
p
−ζ
−r
p
) ∈ Q(ζ
p
).
Show that
ξ
2
= (−1)
(p−1)/2
p−1
¸
r=1
(1 −ζ
r
p
).
(b) Deduce that
ξ
2
=
p if p ≡ 1 (mod 4),
−p if p ≡ 3 (mod 4).
91
(c) Conclude that
ξ =
±
√
p if p ≡ 1 (mod 4),
±
√
p i if p ≡ 3 (mod 4).
and also
√
p ∈ Q(ζ
p
) if p ≡ 1 (mod 4) and
√
p i ∈ Q(ζ
p
) if p ≡ 3 (mod 4).
6.8. Prove Lemma 6.41. [Hint: show that every 2cycle of the form (i i + 1) is in H by
considering elements of the form (1 2 n)
r
(1 2)(1 2 n)
n−r
.]
6.9. This question is about an additive version of Hilbert’s Theorem 90, see Theorem 6.18.
Let E/K be a Galois extension with cyclic Galois group Gal(E/K) = 'σ` of order n.
(a) Show that the function
T : E −→ E; T(u) = u +σ(u) +σ
2
(u) + +σ
n−1
(u),
takes values in K and use this to deﬁne a Klinear mapping Tr
E/K
: E −→ K.
(b) If v ∈ E has Tr
E/K
(v) = 0, show that there is a w ∈ E such that v = w −σ(w).
[Hint: Show that there is an element t ∈ E for which Tr
E/K
t = 0, then consider
w =
1
(Tr
E/K
t)
vσ(t) + (v +σ(v))σ
2
(t) + + (v +σ(v)σ
2
(t) + +σ
n−2
(v))σ
n−1
(t)
and adapt the proof of Hilbert’s Theorem 90 in Theorem 6.18, using Tr
E/K
in place of N
E/K
.]
6.10. (a) For n 1 and 1 k n, the kth power sum s
k
∈ k[X
1
, . . . , X
n
]
S
n
is deﬁned by
s
k
=
¸
1in
X
k
i
.
Prove the formula
s
k
= e
1
s
k−1
−e
2
s
k−2
+ + (−1)
k−1
e
k−1
s
1
+ (−1)
k
ke
k
.
(b) For n 1 and 1 k n, the total symmetric function is deﬁned by
h
k
=
¸
j
1
j
2
j
k
X
j
1
X
j
2
X
j
k
,
i.e., the sum of all the monomials in the X
i
of degree k.
(i) For large values of n, express h
1
, h
2
, h
3
in terms of the elementary symmetric functions
e
1
, e
2
, e
3
.
(ii) Show that the power sum functions s
k
of the previous question satisfy
s
k
= −(h
1
s
k−1
+h
2
s
k−2
+ +h
k−1
s
1
) +kh
k
.
Bibliography
[1] E. Artin, Galois Theory, Dover Publications (1998); ISBN 0 486 62342 4.
[2] JP. Escoﬁer, Galois theory, SpringerVerlag, New York (2001); ISBN 0387987657. [Highly recommended,
especially for its historical notes]
[3] J. B. Fraleigh, A First Course in Abstract Algebra, Addison Wesley (1999); ISBN 0 201 33596 4. [Highly
recommended]
[4] T. W. Hungerford, A counterexample in Galois theory, American Mathematical Monthly 97 (1997), 54–57.
[5] S. Lang, Algebra, Addison Wesley (1993); ISBN 0 201 55540 9.
[6] R. Lidl & H. Niederreiter, Finite Fields, Cambridge University Press (1997); ISBN 0 521 39231 4.
[7] J. Rotman, Galois Theory, SpringerVerlag (1998); ISBN 0 387 98541 7.
[8] I. Stewart, Galois Theory, Chapman and Hall (1989); ISBN 0 412 345501. [Very highly recommended.]
93
Solutions
Chapter 1
1.1. Clearly ¦n ∈ Z : n > 0 and nr = 0 for all r ∈ R¦ ⊆ ¦n ∈ Z : n > 0 and n1 = 0¦. If
0 < n ∈ Z and n1 = 0, then for every r ∈ R,
nr = r + +r
. .. .
n
= (1 + + 1
. .. .
n
)r = (n1)r = 0r = 0,
so ¦n ∈ Z : n > 0 and n1 = 0¦ ⊆ ¦n ∈ Z : n > 0 and nr = 0 for all r ∈ R¦. Hence these sets
are in fact equal. When char R = p > 0 they must both be nonempty. Now by deﬁnition of
characteristic,
char R = min¦n ∈ Z : n > 0 and n1 = 0¦ = min¦n ∈ Z : n > 0 and nr = 0 for all r ∈ R¦.
1.2. (a) Let u, v ∈ S and suppose that uv = 0; then u = 0 or v = 0 since u, v ∈ R and R is
an integral domain. Consider the unit homomorphisms η: Z −→ R and η
t
: Z −→ S. Then for
n ∈ Z, η
t
(n) = η(n), so ker η
t
= ker η and therefore char S = char R.
(b) Q is a ﬁeld and Z ⊆ Q is a subring which is not a ﬁeld.
1.3. (a) For any subring R ⊆ C, R is an integral domain with characteristic subring Z and
char R = 0.
(b) The characteristic subring of A[X] is the same as that of A and char A[X] = char A. A[X]
is an integral domain if and only if A is an integral domain.
(c) If we identify A with the subring of scalar matrices in Mat
n
(A), then the characteristic
subring of Mat
n
(A) is the same as that of A and char Mat
n
(A) = char A. If n > 1 then
Mat
n
(A) is not commutative, in any case it always has zerodivisors since any singular matrix
is a zerodivisor.
1.4. The main thing to check is that ϕ(u + v) = ϕ(u) + ϕ(v) which is a consequence of the
Idiot’s Binomial Theorem. For R = F
p
[X], ϕ is not surjective, while for R = F
p
[X]/(X
2
), ϕ is
not injective.
1.5. (a) Recall from the Isomorphism Theorems of basic Ring Theory that ϕ
−1
Q R; we need
to show it is a prime ideal. Suppose that u, v ∈ R with uv ∈ ϕ
−1
Q; then ϕ(u)ϕ(v) = ϕ(uv) ∈ Q
and so ϕ(u) ∈ Q or ϕ(v) ∈ Q, hence u ∈ ϕ
−1
Q or v ∈ ϕ
−1
Q.
(b) Consider the inclusion function inc: R −→ S; then inc
−1
Q = Q ∩ R, so this result follows
from (a).
(c) Consider Z ⊆ Q; then the zeroideal (0)
Q
Q has (0)
Q
∩Z = (0)
Z
Z but this is not maximal
in Z since for any prime p > 0, (p)
Z
Z is a (maximal) ideal that properly contains (0)
Z
.
(d) We have P ⊆ Q∩ R R with P R maximal; so P ⊆ Q∩ R. In fact Q only needs to be a
proper ideal of S for this argument to work.
1.6. The only proper ideal of k is the zero ideal (0), so ker ϕ = (0).
1.7. (a) Addition and multiplication follow from the obvious formulae
(u
1
+v
1
i)+(u
2
+v
2
i) = (u
1
+u
2
)+(v
1
+v
2
)i, (u
1
+v
1
i)(u
2
+v
2
i) = (u
1
u
2
−v
1
v
2
)+(u
1
v
2
+u
2
v
1
)i,
96 SOLUTIONS
with Z[i] and Q[i] both closed under these operations and containing 1 = 1 + 0i as a unity, so
they are subrings of the ﬁeld C; by Qu. 1.1, they are both integral domains. To see that Q[i] is
a ﬁeld, notice that if u +vi = 0 with u, v ∈ Q,
(u −vi)(u +vi) = (u +vi)(u −vi) = u
2
+v
2
= 0,
so
u
u
2
+v
2
+
v
u
2
+v
2
i ∈ Q(i)
is the inverse of u +vi. Hence every nonzero element of Q[i] has an inverse, therefore Q[i] is a
ﬁeld.
(b) & (c) The crucial point is that every element of Q[i] can be written as
1
n
(u + vi) with
n, u, v ∈ Z and n = 0. Then
inc
∗
(u +vi)
n
= inc
∗
(u +vi)
n + 0i
=
1
n
(u +vi),
so the latter element is in the image of inc
∗
which must therefore be a surjection.
1.8. (a) Existence and uniqueness of such an ψ
a,b
follow from the Homomorphism Extension
Property 1.22 and its eﬀect on f(X) =
¸
n
i=0
r
i
X
i
∈ R[X] where r
i
∈ R is
ψ
a,b
(f(X)) = f(aX +b) =
n
¸
i=0
r
i
(aX +b)
i
.
We have
ψ
a,b
◦ ψ
c,d
(X) = ψ
a,b
(cX +d) = c(aX +b) +d = caX + (cb +d) = ψ
ca,cb+d
(X).
By the uniqueness part of the Homomorphism Extension Property, we have ψ
a,b
◦ψ
c,d
= ψ
ca,cb+d
.
If a is a unit then ψ
a
−1
,−ba
−1 : R[X] −→ R[X] has the property that ψ
a
−1
,−ba
−1(aX+b) = X and
ψ
a,b
(a
−1
X −ba
−1
) = X, so by the uniqueness part of the Homomorphism Extension Property,
ψ
a,b
◦ ψ
a
−1
,−ba
−1 = id = ψ
a
−1
,−ba
−1 ◦ ψ
a,b
.
Therefore these are inverse isomorphisms, ψ
a
−1
,−ba
−1 = ψ
−1
a,b
.
(b) (i) If f(X) =
¸
n
i=0
c
i
X
i
∈ k[X] with c
i
∈ k and c
n
= 0, then deg f(X) = n. Now
ψ
a,b
(f(X)) =
n
¸
i=0
c
i
(aX +b)
i
= c
n
a
n
X
n
+ terms of lower degrees in X.
Since c
n
a
n
= 0, this shows that deg ψ
a,b
(f(X)) = deg f(X).
(ii) Suppose that ψ
a,b
(p(X)) [ g(X)h(X) for g(X), h(X) ∈ k[X]. Choose k(X) ∈ k[X] so that
g(X)h(X) = k(X)ψ
a,b
(p(X)). Since ψ
a,b
is an isomorphism, we have
ψ
−1
a,b
(g(X))ψ
−1
a,b
(h(X)) = ψ
−1
a,b
(k(X))p(X)
and as p(X) is prime, p(X) [ ψ
−1
a,b
(g(X)) or p(X) [ ψ
−1
a,b
(h(X)). Hence ψ
a,b
(p(X)) [ g(X) or
ψ
a,b
(p(X)) [ h(X) and so ψ
a,b
(p(X)) is prime.
(iii) This follows from (ii) and Proposition 1.30.
1.9. (a) Addition and multiplication are given by the usual formulae
(
∞
¸
k=0
a
k
X
k
) + (
∞
¸
k=0
b
k
X
k
) =
∞
¸
k=0
(a
k
+b
k
)X
k
, (
∞
¸
k=0
a
k
X
k
)(
∞
¸
k=0
b
k
X
k
) =
∞
¸
k=0
(
k
¸
=0
a
b
k−
)X
k
.
97
Clearly k[X] ⊆ k[[X]] is a subring. Given two nonzero elements a, b ∈ k[[X]] we may write
a =
∞
¸
k=k
0
a
k
X
k
, b =
∞
¸
0
b
X
with a
k
0
= 0 = b
0
. Then the lowest degree term in ab is a
k
0
b
0
X
k
0
+
0
with a
k
0
b
0
= 0. Hence
ab = 0. So k[[X]] is an integral domain.
(b) Let a =
¸
∞
k=0
a
k
X
k
∈ k[[X]]. Then a has an inverse in k[[X]] only if there is a b =
¸
∞
k=0
b
X
∈ k[[X]] with ab = 1, in particular this forces a
0
= 0 since otherwise the lowest term
in X in ab would be of degree greater than 0. Conversely, if a
0
= 0, then we can inductively
solve the system of equations
a
0
b
0
= 1,
n
¸
=0
a
b
n−
= a
0
b
n
+a
1
b
n−1
+ +a
n
b
n
= 0 (n 1),
to ensure that ab = 1.
(c) We can deﬁne make the set k((X)) of all such ﬁnite tailed Laurent series into a ring with
addition and multiplication deﬁned by
(
∞
¸
k=k
1
a
k
X
k
) + (
∞
¸
k=k
2
b
k
X
k
) =
∞
¸
k=mink
1
,k
2
¦
(a
k
+b
k
)X
k
,
(
∞
¸
k=k
0
a
k
X
k
)(
∞
¸
=
0
b
X
) =
∞
¸
k=mink
0
,
0
¦
(
k
¸
j=0
a
b
k−j
)X
k
.
Clearly k[[X]] ⊆ k((X)) is a subring. Notice that every element
¸
∞
k=k
0
a
k
X
k
∈ k((X)) with
k
0
< 0 can be written as
(
¸
r=0
a
r+k
0
X
r
)X
k
0
.
The inclusion inc: k[[X]] −→k((X)) extends to the monomorphism inc
∗
: Fr(k[[X]]) −→k((X))
for which
inc
∗
¸
∞
r=0
a
r+k
0
X
r
X
−k
0
= (
∞
¸
r=0
a
r+k
0
X
r
)X
k
0
,
so inc
∗
is surjective.
1.10. Here f(X) = (3X −3)d(X) + (−9X + 7).
1.11. Here f(X) = −X
3
−X
2
+X + 1 and d(X) = −X
3
−X with
f(X) = d(X) + (−X −X
2
+ 1) = d(X) + (2X
2
+ 2X + 1).
1.12. The reduction modulo p function
ρ: Z[X] −→F
p
[X]; ρ(f(X)) = f(X),
is a ring homomorphism. If f(X) = g(X)h(X) with g(X), h(X) ∈ Z[X], deg g(X) < deg f(X)
and deg h(X) < deg f(X), then
f(X) = ρ(g(X)h(X)) = ρ(g(X))ρ(h(X)) = g(X)h(X),
where deg g(X) < deg f(X) = deg f(X) and deg h(X) < deg f(X) = deg f(X). But this is
impossible since f(X) is irreducible. So f(X) must be irreducible.
X
3
−X +1 reduces modulo 3 to an irreducible since it has no roots modulo 3. So X
3
−X +1
is irreducible.
X
3
+2X +1 ≡ X
3
−X +1 (mod 3) so this polynomial reduces modulo 3 to an irreducible and
so is irreducible.
X
3
+X −1 reduces modulo 2 to an irreducible since it has no roots modulo 2. So X
3
+X −1
98 SOLUTIONS
is irreducible.
X
5
−X + 1 is irreducible modulo 3 and 5 so is itself irreducible.
X
5
+ X − 1 = (X
3
+ X
2
− 1)(X
2
− X + 1) and 5X
3
− 10X + X
2
− 2 = (5X + 1)(X
2
− 2) so
neither of these is irreducible.
1.13. I
1
= (X
2
+1), I
2
= (X
2
+2), I
3
= (X
2
−2), I
4
= (X−
√
2), I
5
= (X
2
+2), I
6
= X
2
+X+1.
1.14. The image is
ε
√
2
Q[X] = Q[
√
2] = ¦a +b
√
2 : a, b ∈ Q¦.
The image of ε
−
√
2
is ε
−
√
2
Q[X] = Q[
√
2] = ε
√
2
Q[X]. We have
ker ε
√
2
= ker ε
−
√
2
= (X
2
−2) Q[X]
which is a maximal ideal.
1.15. Notice that ω = (−1 +
√
3i)/2 = ζ
3
is a primitive 3rd root of unity and is a root of the
irreducible polynomial X
2
+X + 1 ∈ Q[X]. Then
ε
ω
Q[X] = Q[ω] = ¦a +bω : a, b ∈ Q¦, ker ε
ω
= (X
2
+X + 1) Q[X],
where (X
2
+ X + 1) Q[X] is a maximal ideal. The other complex root of X
2
+ X + 1 is ω
2
,
so the evaluation homomorphism ε
ω
2 has ε
ω
2 Q[X] = ε
ω
Q[X] and ker ε
ω
2 = ker ε
ω
.
1.16. We have
ε
α
Q[X] = Q[α] = ¦a +bα +cα
2
+dα
3
: a, b, c, d ∈ Q¦, ker ε
α
= (X
4
−2) Q[X],
and the latter ideal is maximal. The other complex roots of X
4
− 2 are −α, αi, −αi (notice
that two of these are real while the other two are not). Then
ker ε
−α
= ker ε
αi
= ker ε
−αi
= (X
4
−2) Q[X]
but although ε
−α
Q[X] = Q[α], we have
ε
αi
Q[X] = ε
−αi
Q[X] = Q[αi] = ¦a +bαi +cα
2
+dα
3
i : a, b, c, d ∈ Q¦ = Q[α],
so ε
αi
Q[X] = ε
α
Q[X] since one of these is a subset of R but the other is not.
If we replace Q by R, then in R[X],
X
4
−2 = (X
2
−
√
2)(X
2
+
√
2) = (X −
4
√
2)(X +
4
√
2)(X
2
+
√
2).
Let α be a root of X
4
−2. If α =
4
√
2, then
ε
α
R[X] = R[α] = ¦a +bα : a, b ∈ R¦ = R, ker ε
α
= (X −
4
√
2) R[X].
Similarly, if α = −
4
√
2, then
ε
−α
R[X] = R[−α] = ¦a −bα : a, b ∈ R¦ = R, ker ε
−α
= (X +
4
√
2) R[X].
If α
2
+ 2 = 0, then α / ∈ R and
ε
α
R[X] = R[α] = ¦a +bα : a, b ∈ R¦ = C, ker ε
α
= (X
2
+ 2) R[X].
1.17. First change variable to obtain
g(X) = f(X + 3) = X
3
−6X + 4.
Using Cardan’s method we have to solve the quadratic equation
U
2
+ 4U + 8 = 0,
which has roots
−2 ±2i = (
√
2)
3
e
3πi/4
.
99
Thus we can take
u =
√
2e
πi/4
ω
r
=
√
2
√
2
(1 +i)ω
r
= (1 +i)ω
r
(r = 0, 1, 3).
For the roots of g(X) we obtain 2,
√
3−1, −
√
3−1, while for f(X) we have 5,
√
3+2, −
√
3+2.
1.18. Work backwards with Cardan’s method. For α, take
−
q
2
= 10,
27q
2
+ 4p
3
108
= 108,
so q = −20 and p = 6. Thus α is a real root of f(X) = X
3
+ 6X − 20. Notice that 2 is a real
root of this polynomial and
f(X) = (X −2)(X
2
+ 2X + 10),
where X
2
+ 2X + 10 has no real roots. Therefore α = 2.
For β, take
−
q
2
= 1,
27q
2
+ 4p
3
108
=
28
27
,
so q = −2 and p = 1. Thus β is a real root of g(X) = X
3
+X −2 for which 1 is also a root and
g(X) = (X −1)(X
2
+X + 2),
where X
2
+X + 2 has no real roots. Therefore β = 1.
1.19. To see that the homomorphism
Aﬀ
1
(k) −→ Aut
k
(k[X]); A −→ α
A
−1,
described in the Proof of Example 1.60 is surjective, suppose that ϕ ∈ Aut
k
(k[X]) is any
automorphism. Let
ϕ(X) = a
0
+a
1
X + +a
n
X
n
with a
i
∈ k and a
n
= 0. If n = 0 then ϕk[X] = k ⊆ k[X] so ϕ would not be surjective, hence
we must have n 1. Suppose that show that n > 1. Then
ϕk[X] = ¦c + 0 +c
1
ϕ(X) + +c
k
ϕ(X)
k
: c
0
, c
1
, . . . , c
k
∈ k¦ = k[X].
But if k > 0 and c
k
= 0 then deg(c + 0 + c
1
ϕ(X) + + c
k
ϕ(X)
k
) = kn > 1, so X / ∈ ϕk[X],
which gives a contradiction. So we must have n = 1. Therefore ϕ(X) = a
0
+a
1
X and so ϕ = α
A
for some A ∈ Aﬀ
1
(k).
1.20. Calculation.
1.21. We have
deg Φ
20
(X) = ϕ(20) = ϕ(4)ϕ(5) = 2 4 = 8
and
X
20
−1 = (X
10
−1)(X
10
+ 1) = (X
10
−1)(X
2
+ 1)(X
8
−X
6
+X
4
−X
2
+ 1).
Since the roots of X
10
−1 are the 10th roots of unity, we ﬁnd that
Φ
20
(X) [ (X
2
+ 1)(X
8
−X
6
+X
4
−X
2
+ 1);
since cyclotomic polynomials are irreducible, we must have Φ
20
(X) = X
8
−X
6
+X
4
−X
2
+1.
1.22. (a) We have
X
p
k
−1 = (X
p
k−1
)
p
−1 = (X
p
k−1
−1)Φ
p
(X
p
k−1
),
so by (1.5),
¸
0jk
Φ
p
j (X) = Φ
p
(X
p
k−1
)
¸
0jk−1
Φ
p
j (X),
100 SOLUTIONS
and therefore Φ
p
k(X) = Φ
p
(X
p
k−1
). The complex roots of Φ
p
(X) are the primitive pth roots
of 1, so the roots of Φ
p
k(X) are their p
k−1
st roots which are the primitive p
k
th roots of 1.
(b) Using the formula of Equation 1.4, we have
Φ
p
k(X) = Φ
p
(X
p
k−1
) = (X
p
k−1
−1)
p−1
+c
p−2
(X
p
k−1
−1)
p−2
+ +c
1
(X
p
k−1
−1) +c
0
,
where c
r
≡ 0 (mod p) and c
0
= p. The Idiot’s Binomial Theorem gives
X
p
k−1
−1 ≡ (X −1)
p
k−1
(mod p)
so
Φ
p
k(X) = (X −1)
(p−1)p
k−1
+c
t
p−2
(X −1)
(p−2)p
k−1
+ +c
t
1
(X −1)
p
k−1
+c
t
0
,
where c
t
r
≡ 0 (mod p). In fact,
c
t
0
= Φ
p
k(1) = Φ
p
(1) = c
0
= p,
so the Eisenstein Test can be applied to show that Φ
p
k(X) is irreducible over Q.
(c) First notice that
deg Φ
n
(X) = ϕ(n) = (p
1
−1) (p
k
−1)p
r
1
−1
1
p
r
k
−1
k
,
and
deg Φ
p
1
p
k
(X
p
r
1
−1
1
p
r
k
−1
k
) = ϕ(p
1
p
k
)p
r
1
−1
1
p
r
k
−1
k
= (p
1
−1) (p
k
−1)p
r
1
−1
1
p
r
k
−1
k
,
so deg Φ
n
(X) = deg Φ
p
1
p
k
(X
p
r
1
−1
1
p
r
k
−1
k
). Also, each root ξ of Φ
n
(X),
(ξ
p
r
1
−1
1
p
r
k
−1
k
)
p
1
p
k
= ξ
n
= 1,
and no smaller power of (ξ
p
r
1
−1
1
p
r
k
−1
k
) has this property, hence (ξ
p
r
1
−1
1
p
r
k
−1
k
) is a root of
Φ
p
1
p
k
(X). This shows that Φ
n
(X) [ Φ
p
1
p
k
(X
p
r
1
−1
1
p
r
k
−1
k
). As these are monic polynomials of
the same degree they are equal.
1.23. By Theorem 1.43, Φ
n
(X) =
¸
t=1,...,n−1
gcd(t,n)=1
(X −ζ
t
n
), so
Φ
n
(X
−1
) =
¸
t=1,...,n−1
gcd(t,n)=1
(X
−1
−ζ
t
n
)
= X
−ϕ(n)
¸
t=1,...,n−1
gcd(t,n)=1
(1 −Xζ
t
n
)
= X
−ϕ(n)
¸
t=1,...,n−1
gcd(t,n)=1
(1 −Xζ
n−t
n
)
= X
−ϕ(n)
¸
t=1,...,n−1
gcd(t,n)=1
(1 −Xζ
−t
n
)
= X
−ϕ(n)
¸
t=1,...,n−1
gcd(t,n)=1
(ζ
t
n
−X)
= (−1)
ϕ(n)
X
−ϕ(n)
¸
t=1,...,n−1
gcd(t,n)=1
(X −ζ
t
n
)
= (−1)
ϕ(n)
X
−ϕ(n)
Φ
n
(X).
Since 2 [ ϕ(n) when n > 2 and the result is immediate when n = 2, we see that desired equation
always holds.
101
1.24. We have
ζ
n
+ζ
−1
n
= e
2πi/n
+e
−2πi/n
= (cos(2πi/n) + sin(2πi/n) i) + (cos(2πi/n) −sin(2πi/n) i) = 2 cos(2πi/n).
Now we have
ζ
5
+ζ
−1
5
= 2 cos(2π/5), ζ
2
5
+ζ
−2
5
= (ζ
5
+ζ
−1
5
)
2
−2 = 4 cos
2
(2π/5) −2.
We also have Φ
5
(X) = X
4
+X
3
+X
2
+X + 1, so
ζ
4
5
+ζ
3
5
+ζ
2
5
+ζ
5
+ 1 = 0.
Rearranging and using the formulae ζ
4
5
= ζ
−1
5
, ζ
3
5
= ζ
−2
5
, we have
(ζ
2
5
+ζ
−2
5
) + (ζ
5
+ζ
−1
5
) + 1 = 0,
hence
4 cos
2
(2π/5) + 2 cos(2π/5) −1 = 0.
Thus a suitable polynomial is 4X
2
+ 2X −1 ∈ Q[X].
1.25. (a) In K[X], by the Idiot’s Binomial Theorem 1.10,
X
p
−1 = X
p
+ (−1)
p
= (X + (−1))
p
= (X −1)
p
.
By the Unique Factorization Property 1.33, the only root of this polynomial in K must be 1.
Similarly,
X
np
m
−1 = (X
n
−1)
p
m
and the only roots of this must be nth roots of 1.
(b) If u ∈ K is a root of this polynomial then u
p
= a. As in (a) we have
X
p
−a = X
p
−u
p
= (X −u)
p
,
so u is the only root in K.
Chapter 2
2.1. This is similar to Example 2.4.
2.2. It is obvious that [Q(
√
p,
√
q) : Q(
√
p)] 2; if [Q(
√
p,
√
q) : Q(
√
p)] = 1 then
√
q ∈ Q(
√
p),
say
√
q = a +b
√
p for some a, b ∈ Q. Then
q = (a +b
√
p)
2
= (a
2
+b
2
p) + 2ab
√
p,
giving the simultaneous pair of equations
a
2
+b
2
p = q, 2ab = 0.
If b = 0 then
√
q ∈ Q which contradicts the result of Qu. 2.1. If a = 0 then
√
q = b
√
p. Writing
b = b
1
/b
2
with b
1
, b
2
∈ Z and gcd(b
1
, b
2
) = 1, we obtain
b
2
2
q = b
2
1
p
and so p [ b
2
and q [ b
1
. Writing b
1
= b
t
1
q and b
2
= b
t
2
q for suitable b
t
1
, b
t
2
∈ Z, we obtain
(b
t
2
)
2
p
2
q = (b
t
1
)
2
q
2
p,
hence
(b
t
2
)
2
p = (b
t
1
)
2
q.
From this we obtain p [ b
t
1
and q [ b
t
2
; but then p [ b
1
as well as p [ b
2
, contradicting the fact that
gcd(b
1
, b
2
) = 1. So
√
q / ∈ Q(
√
p).
2.3. Arrange the induction carefully.
102 SOLUTIONS
2.4. Notice that if v = ±u then b = v
2
= u
2
= a which is impossible; so v = ±u. Then
u −v =
(u −v)(u +v)
u +v
=
u
2
−v
2
u +v
=
a −b
u +v
∈ K(u +v).
Hence
u =
1
2
((u +v) + (u −v)) ∈ K(u +v), v =
1
2
((u +v) −(u −v)) ∈ K(u +v).
So K(u, v) K(u +v) K(u, v) and therefore K(u +v) = K(u, v).
2.5. Since 1, i span the Qvector space Q(i), we have [Q(i) : Q] 2. But also if x, y ∈ R, then
x +yi = 0 ⇐⇒ x = y = 0, so 1, i is a basis for Q(i) over Q. Hence [Q(i) : Q] = 2.
2.6. First notice that [Q(
√
3) : Q] = 2 (with Qbasis 1,
√
3) and Q(
√
3) R. Also, i / ∈ Q(
√
3)
and since i
2
+ 1 = 0, Q(
√
3, i) = Q(
√
3)(i) has [Q(
√
3, i) : Q(
√
3)] = 2. By Theorem 2.6(ii),
[Q(
√
3, i) : Q] = [Q(
√
3, i) : Q(
√
3)] [Q(
√
3) : Q] = 2 2 = 4.
The following three subﬁelds of Q(
√
3, i) are distinct and are extensions of Q having degree 2:
L
1
= Q(
√
3), L
2
= Q(i), L
3
= Q(
√
3 i). Then [L
r
∩ L
s
: Q] > 1 ⇐⇒ L
r
∩ L
s
= L
r
= L
s
, so
L
r
∩ L
s
= Q whenever r = s. The only real subﬁeld amongst these is L
1
.
C
2
∞ R
∞ Q(
√
3, i)
2
2
2
Q(
√
3)
2
Q(i)
2
Q(
√
3 i)
2
Q
2.7. (a) Since 5 is a prime,
[Q(ζ
5
) : Q] = [Q[X]/(Φ
5
(X)) : Q] = ϕ(5) = 5 −1 = 4.
(b) We have ζ
5
= cos(2π/5) +sin(2π/5) i ∈ Q(ζ
5
). But also ζ
−1
5
∈ Q(ζ
5
) and ζ
−1
5
= cos(2π/5) −
sin(2π/5) i ∈ Q(ζ
5
). Hence we have
cos(2π/5) =
1
2
ζ
5
+ζ
−1
5
∈ Q(ζ
5
), sin(2π/5) i =
1
2
ζ
5
−ζ
−1
5
∈ Q(ζ
5
).
(c) This can be found by repeated use of the double angle formula
cos(A+B) = cos Acos B −sin Asin B.
The polynomial T
n
(X) expressing cos nθ in terms of cos θ is called the nth Chebyshev polyno
mial, see Remark 6.6.
(d) For k = 0, 1, 2, 3, 4, cos(5(2kπ/5)) = cos(2kπ) = 1, so T
5
(cos 2kπ/5) − 1 = 0. So each of
the numbers cos(2kπ/5) is a root of the polynomial T
5
(X) −1 = (X −1)(4X
2
+2X −1)
2
. For
k = 1, 2, 3, 4, cos(2kπ/5) is a root of 4X
2
+ 2X −1, therefore
Q(cos(2π/5))
∼
= Q[X]/(4X
2
+ 2X −1), [Q(cos(2kπ/5)) : Q] = 2.
103
(e) Q(ζ
5
)
2
Q(cos(2π/5))
2
Q
2.8. This is similar to the previous question.
2.9. (a) If α ∈ Aut
Q
(E
n
) then α(2
1/n
)
n
= α(2) = 2, so α(2
1/n
) ∈ E
n
is also a real n
th root of 1. If n is odd, the only possibility is α(2
1/n
) = 2
1/n
, so α = id. If n is even, the
possibilities are α(2
1/n
) = ±2
1/n
. We can realize this automorphism starting with the evaluation
homomorphism ε
2
1/n : Q[X] −→ E
n
and precomposing with the isomorphism ψ: Q[X] −→Q[X]
for which ψ(X) = −X to form ε
t
2
1/n
= ε
2
1/n ◦ ψ. On passing to the quotient homomorphism of
ε
t
2
1/n
we obtain an automorphism τ
n
of E
n
under which τ
n
(2
1/n
) = −2
1/n
.
(b) Since E R, an automorphism α ∈ Aut
Q
(E) has the eﬀect
α(2
1/n
) =
2
1/n
if n is odd,
±2
1/n
if n is even.
If for some n we have α(2
1/n
) = −2
1/n
then
−2
1/n
= α(2
1/n
) = α(2
1/2n
)
2
> 0
since α(2
1/2n
) ∈ R. This contradiction shows that α(2
1/n
) = 2
1/n
for every n, so α = id.
(c) Assuming there are only 6 such subﬁelds, they form the following tower.
E
12
3
2
E
4
2
E
6
3
2
E
2
2
E
3
3
Q
(d) This element is a root of the polynomial
(X −(2
1/2
+ 2
1/3
))(X −(−2
1/2
+ 2
1/3
)) = X
2
−2(2
1/3
)X + 2
2/3
−2 ∈ E
3
[X],
so it is certainly an element of E
6
which is the only degree 2 extension of E
3
. If 2
1/2
+2
1/3
∈ E
3
then 2
1/2
∈ E
3
, which would imply 2 = [E
2
: Q] [ [E
3
: Q] = 3 which is false, so 2
1/2
+2
1/3
/ ∈ E
3
;
a similar argument shows that 2
1/2
+ 2
1/3
/ ∈ E
2
. Writing ω = e
2πi/3
, 2
1/2
+ 2
1/3
is a root of
(X −(2
1/2
+ 2
1/3
))(X −(2
1/2
+ 2
1/3
ω))(X −(2
1/2
+ 2
1/3
ω
2
))
= X
3
−3(2
1/2
)X
2
+ 6X −(2 + 2(2
1/2
)) ∈ E
2
[X],
so it cannot lie in E
4
since 2
1/2
+ 2
1/3
/ ∈ E
2
and 3 [E
4
: E
2
] = 2. So 2
1/2
+ 2
1/3
is in E
6
and
E
12
and none of the others.
104 SOLUTIONS
Chapter 3
3.1. Clearly, t is algebraic over K if and only if ker ε
t
= (0), i.e., (i) ⇐⇒ (ii). By Theorem 2.9,
(ii) ⇐⇒ (iii). Hence these three conditions are indeed equivalent.
3.2. The diagrams at the bottom indicate useful subﬁelds of the splitting ﬁelds occurring in
each of these examples.
p
1
(X) = X
4
−X
2
+1: The polynomial X
2
−X +1 has the complex roots e
±πi/3
=
1 ±
√
3 i
2
, so
the four roots of p
1
(X) are the complex square roots of these numbers, i.e., ±e
±πi/6
. Explicitly
these are
√
3
2
+
1
2
i, −
√
3
2
−
1
2
i,
√
3
2
−
1
2
i, −
√
3
2
+
1
2
i.
The splitting ﬁeld is E = Q(
√
3, i) and [E : Q] = 4.
p
2
(X) = X
6
− 2: The roots are the six complex 6th roots of 2, i.e.,
6
√
2e
2kπi/6
=
6
√
2e
kπi/3
for
k = 0, 1, 2, 3, 4, 5. Explicitly, these are
6
√
2,
6
√
2
2
+
6
√
2
√
3
2
i, −
6
√
2
2
+
6
√
2
√
3
2
i, −
6
√
2, −
6
√
2
2
−
6
√
2
√
3
2
i,
6
√
2
2
−
6
√
2
√
3
2
i.
The splitting ﬁeld is E = Q(
6
√
2,
√
3i) = Q(
6
√
2)(
√
3i) which has degree [E : Q] = 12.
p
3
(X) = X
4
+ 2: The roots are the four 4th roots of −2, i.e.,
4
√
2e
(2k+1)πi/4
for k = 0, 1, 2, 3.
Explicitly these are
1
4
√
2
+
1
4
√
2
i, −
1
4
√
2
+
1
4
√
2
i, −
1
4
√
2
−
1
4
√
2
i,
1
4
√
2
−
1
4
√
2
i.
The splitting ﬁeld is E = Q(
4
√
2, i) and [E : Q] = 8.
p
4
(X) = X
4
+ 5X
3
+ 10X
2
+ 10X + 5: Notice that
p
4
(Y −1) = Y
4
+Y
3
+Y
2
+Y + 1 = Φ
5
(Y ),
so the splitting ﬁeld of p
4
(X) over Q is the same as that of Φ
5
(Y ) over Q and this is the
cyclotomic ﬁeld Q(ζ
5
) where ζ
5
= cos(2π/5) + sin(2π/5)i with [Q(ζ
5
) : Q] = 4; in fact we have
Q(ζ
5
) = Q(cos(2π/5), sin(2π/5)i).
Q(
√
3, i)
Q(
√
3)
2
Q
2
Q(
6
√
2,
√
3, i)
Q(
6
√
2,
√
3)
2
Q(
√
3)
6
Q
2
Q(
4
√
2, i)
Q(
4
√
2)
2
Q
4
Q(cos(2π/5), sin(2π/5)i)
Q(cos(2π/5))
2
Q
2
3.3. List the three roots of X
3
−2 as u
1
=
3
√
2, u
2
=
3
√
2ζ
3
, u
3
=
3
√
2ζ
2
3
. Then each automorphism
α ∈ Aut
Q
(Q(
3
√
2, ζ
3
)) permutes these roots, so can be identiﬁed with the unique permutation
σ
α
∈ S
3
for which
α(u
i
) = u
σ
α
(i)
(i = 1, 2, 3).
We ﬁnd that (using cycle notation for permutations)
σ
id
= id, σ
α
0
= (2 3), σ
α
1
= (1 2 3), σ
α
1
= (1 2), σ
α
2
= (1 3 2), σ
α
2
= (1 3).
These are the six elements of S
3
, therefore Aut
Q
(Q(
3
√
2, ζ
3
))
∼
= S
3
.
105
3.4. Irreducibility is a consequence of the polynomial version of the Eisenstein Test 1.48.
Suppose that t ∈ k(T) is a root of g(X); then using the Idiot’s Binomial Theorem we have
(X −t)
p
= X
p
−t
p
= X
p
−T,
so t is in fact a root of multiplicity p, hence it is the only root of g(X) in k(T). This also gives
the factorization of g(X) into linear factors over k(T).
3.5. Q(
√
5,
√
10)/Q: Here [Q(
√
5,
√
10) : Q] = 4 and the element
√
5 +
√
10 has degree 4 with
minimal polynomial X
4
−30X
2
+ 25 which has roots ±
√
5 ±
√
10.
Q(
√
2, i)/Q: Here [Q(
√
2, i) : Q] = 4 and the element
√
2 + i has degree 4 with minimal
polynomial X
4
−2X
2
+ 9 which has roots ±
√
2 ±i.
Q(
√
3, i)/Q: Here [Q(
√
3, i) : Q] = 4 and the element
√
3 + i has degree 4 with minimal
polynomial X
4
−4X
2
+ 16 which has roots ±
√
3 ±i.
Q(
4
√
3, i)/Q: Here [Q(
4
√
3, i) : Q] = 8 and the element
4
√
3 + i has degree 8 with minimal
polynomial X
8
+ 4X
6
+ 40X
2
+ 4 which has roots ±
4
√
3 ±i and ±
4
√
3i ±i.
3.6. The induction is straightforward. Here is the argument that K(u, v)/K is simple. We
assume that K is inﬁnite since otherwise the result will be proved in Proposition 5.16.
Consider the subﬁelds K(u + tv) K(u, v) with t ∈ K. Then there are only ﬁnitely many
of these, so there must be s, t ∈ K such that s = t and K(u +sv) = K(u +tv). Then
(s −t)v = (u +sv) −(u +tv) ∈ K(u +tv),
hence v ∈ K(u +tv). This implies that
u = (u +tv) −tv ∈ K(u +tv),
hence K(u, v) K(u +tv) K(u, v) and so K(u, v) = K(u +tv).
3.7. If E/K is a quadratic extension then for any u ∈ E − K we have 1 < [K(u) : K] 2,
so [K(u) : K] = 2 = [E : K] and therefore K(u) = E. Then minpoly
K,u
(X) must factor into
linear factors over E, so both its roots in K lie in E. This shows that E is normal over K.
The example F
2
(Z)/F
2
(Z
2
) is not separable since X
2
− Z
2
∈ F
2
(Z
2
)[X] is irreducible but
not separable (see Qu. 3.4). If char K = 2 then all quadratic polynomials over K are separable.
3.8. Let E C be a splitting subﬁeld for f(X) over Q. Then if v ∈ C is a nonreal root of
f(X) we have v / ∈ Q(u), so f(X) does not split over Q(u) even though it has a root in this ﬁeld.
This means that there is a monomorphism ϕ ∈ Mono
Q
(Q(u), C) = Mono
Q
(Q(u), Q) for which
ϕ(u) = v, hence ϕQ(u) = Q(u) and so Q(u)/Q is not normal.
Chapter 4
4.1. By Theorem 3.81 we know that splitting ﬁelds are always normal, so it is only necessary
to show that the splitting ﬁeld E of p(X) over K is separable over K. Since E is obtained by
repeatedly adjoining roots of p(X), the result follows from Proposition 3.73 together with the
fact that if L/K E/K is separable and v ∈ E is a root of p(X), then L(v)/K is separable.
4.2. (a) Suppose that f(X) = c
3
X
3
+c
2
X
2
+c
1
X +c
4
with c
3
= 0. Then
f(uX +v) =
c
3
u
3
X
3
+ (3c
3
vu
2
+c
2
u
2
)X
2
+ (3c
3
uv
2
+c
1
u + 2c
2
uv)X + (c
3
v
3
+c
4
+c
1
v +c
2
v
2
),
so if we take u to be any cube root of c
3
and u = −c
2
/3c
3
then f(uX +v) has the desired form.
Notice that v ∈ K(u) and then f(uX +v) ∈ K(u), so provided that we can ﬁnd a cube root of
1/c
3
in K, we have f(uX +v) ∈ K.
(b) Viewing Gal(E/K) as a subgroup of S
3
, by Theorem 4.8 we know that 3 divides [ Gal(E/K)[;
106 SOLUTIONS
but the only subgroups of S
3
with this property are S
3
and A
3
.
(c) This is a tedious calculation! See Section 4.7 for the rest of this question.
4.3. If a/b is a rational root of f(X), we may assume that gcd(a, b) = 1. Now a
3
−3ab
2
+b
3
=
0, which easily implies that a, b = ±1; but 1 is certainly not a root. Hence there are no
rational roots and so no proper rational factors. By the formula following Proposition 4.25, the
discriminant of f(X) is
∆ = −27 −4(−3)
3
= 81 = 9
2
.
If the distinct roots of f(X) in C are u, v, w, the splitting subﬁeld K(v, w) = Q(u, v, w) C
satisﬁes 3 [ [Q(u, v, w) : Q] and [Q(u, v, w) : Q] [ 3! = 6. The Galois group Gal(Q(u, v, w)/Q) is a
subgroup of S
3
(viewed as the permutation group of ¦u, v, w¦). Since the discriminant is a square
in Q, Proposition 4.26 implies that Gal(Q(u, v, w)/Q) A
3
∼
= Z/3. So [ Gal(Q(u, v, w))/Q)[ = 3
and Gal(Q(u, v, w)) is cyclic of order 3 whose generator is a 3cycle which cyclically permutes
u, v, w.
4.4. (a) This should be a familiar result.
(b) The centre of D
8
is
α
2
which has order 2, and there are three normal subgroups of order
4, namely
'α` = ¦ι, α, α
2
, α
3
¦,
α
2
, β
= ¦ι, α
2
, β, βα
2
¦,
α
2
, βα
= ¦ι, α
2
, βα, βα
3
¦.
Notice that there are also four nonnormal subgroups of order 2,
'β` = ¦ι, β¦, 'βα` = ¦ι, βα¦,
βα
2
= ¦ι, βα
2
¦,
βα
3
= ¦ι, βα
3
¦.
4.5. This is an example of case (iii) of Kaplansky’s Theorem and we use the notation of the
proof. The discriminant here is δ
2
= −12, so we can take δ = 2
√
3i. The roots of X
2
+ 3 are
±
√
3i, so we may assume
u =
4
√
3 ζ
8
=
√
2
4
√
3
2
(1 +i), v =
4
√
3 ζ
−1
8
=
√
2
4
√
3
2
(1 −i),
where as usual ζ
8
= e
2πi/8
= (1 +i)/
√
2. Hence we have uv =
√
3 and uvδ = 6i. This gives the
diagram of subﬁelds of E
E = Q(
4
√
3 ζ
8
,
4
√
3 ζ
−1
8
) = Q(
4
√
3, ζ
8
)
Q(
√
3, i)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Q(
√
3i)
2
Q(
√
3)
2
Q(i)
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Q
Then α is the restriction of complex conjugation to E, while β(
√
3i) =
√
3i and β(
√
3) = −
√
3,
hence also β(i) = −i. Using the choices of the proof, we have
β(
4
√
3 ζ
8
) = −
4
√
3 ζ
8
, β(
4
√
3 ζ
−1
8
) = β(−
4
√
3 ζ
8
i) = −
4
√
3 ζ
8
i.
The eﬀects of σ and γ on the four roots
4
√
3 ζ
8
,
4
√
3 ζ
−1
8
, −
4
√
3 ζ
8
, −
4
√
3 ζ
−1
8
of f(X) are given in
permutation notation by σ = (1 4 3 2) and α = (1 2)(3 4), and these generate a dihedral
subgroup of S
4
. Using the previous question (but beware that the notation there is inconsistent
with that of the present situation!) we have the normal subgroups
σ
2
, 'σ` ,
σ
2
, α
,
σ
2
, ασ
,
107
and these have ﬁxed ﬁelds
E
'σ
2
`
= Q(
√
3, i), E
/σ)
= Q(i), E
'σ
2
,α`
= Q(
√
3), E
'σ
2
,ασ`
= Q(
√
3 i),
each of which is a normal extension of Q.
4.6. Q(X
3
− 10)/Q: This is similar to Example 4.20, with splitting ﬁeld Q(
3
√
10, ζ
3
) and
Gal(Q(
3
√
10, ζ
3
)/Q)
∼
= S
3
.
Q(
√
2)(X
3
−10)/Q(
√
2): The splitting ﬁeld is Q(
√
2,
3
√
10, ζ
3
), [Q(
√
2,
3
√
10) : Q(
√
2)] = 3 and
Q(
√
2,
3
√
10) Q(
√
2,
3
√
10, ζ
3
).
Since ζ
3
is not real, [Q(
√
2,
3
√
10, ζ
3
) : Q(
√
2)] = 6. The Galois group is isomorphic to S
3
.
Q(
√
3 i)(X
3
− 10)/Q(
√
3 i): Here Q(
√
3 i) = Q(ζ
3
), with [Q(ζ
3
) : Q] = 2. The splitting ﬁeld is
Q(
3
√
10, ζ
3
) and [Q(
3
√
10, ζ
3
) : Q(ζ
3
)] = 3, hence Gal(Q(
3
√
10, ζ
3
)/Q(ζ
3
))
∼
= Z/3 with generator σ
for which σ(
3
√
10) =
3
√
10 ζ
3
.
Q(
√
23 i)(X
3
−X −1)/Q(
√
23 i): First note that X
3
−X −1 ∈ Z[X] must be irreducible since
its reduction modulo 2, X
3
+ X + 1 ∈ F
2
[X], has no root in F
2
and hence has no linear factor
(see Qu. 1.10). To proceed further we can use the ideas of Qu. 4.2 above (see also Section 4.7).
The discriminant of the polynomial X
3
− X − 1 is ∆ = −23 and so δ =
√
23 i. Then if
E = Q(
√
23 i)(X
3
− X − 1) is the splitting ﬁeld of X
3
− X − 1 over Q, Gal(E/Q)
∼
= S
3
and
Gal(E/Q(
√
23 i))
∼
= A
3
.
K(X
3
− X − 1)/K for K = Q, Q(
√
5), Q(
√
5 i), Q(i): Continuing the preceding discussion,
notice that [E ∩ R : Q] = 3, so
√
5 / ∈ E, hence
Q(
√
5)(X
3
−X −1) = Q(X
3
−X −1)(
√
5)
and
[Q(
√
5)(X
3
−X −1) : Q(
√
5)] = [Q(X
3
−X −1) : Q] = 6,
hence Gal(Q(
√
5)(X
3
−X −1)/Q(
√
5))
∼
= S
3
. Similarly,
√
5 i / ∈ E and i / ∈ E, hence
Gal(Q(
√
5 i)(X
3
−X −1)/Q(
√
5 i))
∼
= S
3
∼
= Gal(Q(i)(X
3
−X −1)/Q(i)).
4.7. (a) Since char K = 0, f
t
(X) = pX
p−1
= 0, so if u ∈ L is any root of f(X) then
f
t
(u) = pu
p−1
= 0. By Proposition 3.55, there are no multiple roots, hence p distinct roots. If
u, v ∈ L are distinct roots, then (vu
−1
)
p
= 1, so v = uξ for ξ ∈ K a pth root of 1 with ξ = 1.
(b) If there is a root u / ∈ K, the Galois group Gal(L/K) acts in the following way. By The
orem 4.8, there must be an element γ ∈ Gal(L/K) with γ(u) = u. We can write γ(u) = uξ
γ
where ξ
γ
= 1 is a pth root of 1. Since γ(ξ
γ
) = ξ
γ
, for r 1 we have γ
r
(u) = uξ
r
γ
, which can
only equal u if p [ r. So u must have at least p conjugates which are all roots of f(X). Since
deg f(X)
p
, every root of f(X) is conjugate to u, so f(X) must be irreducible over K.
(c) Suppose that f(X) = g(X)h(X) with g(X) ∈ K[X] monic irreducible and 0 < d =
deg g(X) < p. Let L/K with L a splitting ﬁeld for f(X) over K and let w ∈ L be a root
of g(X). Arguing as in (a), we know that each root of g(X) has the form wξ where ξ is some
pth root of 1; moreover, L must contain p distinct pth roots of 1. Now the constant coeﬃcient
of g(X) is g(0) = (−1)
d
ξ
0
w
d
∈ K where ξ
0
is a pth root of 1. So
g(0)
p
= (−1)
dp
ξ
p
0
(w
p
)
d
= (−1)
dp
a
d
,
from which it follows that a
d
is a pth power in K. As gcd(p, d) = 1, there are integers r, s such
that rp +sd = 1, so we have
a = (a
r
)
p
(a
d
)
s
= a pth power in K.
Hence if f(X) is not irreducible in K[X] it has a root in K.
108 SOLUTIONS
4.8. If u ∈ L is a root of f(X) in an extension L/K then by the Idiot’s Binomial Theorem 1.10
X
p
−a = X
p
+ (−u)
p
= (X −u)
p
,
so u is the only such root in L and f(X) splits over L. If (X − u)
d
∈ K[X] for some d with
1 < d < p then u
d
∈ K. Since gcd(d, p) = 1, there are integers r, s such that rd +sp = 1. Hence
(u
d
)
s
(u
p
)
r
= u, where the left hand side is in K. This shows that u ∈ K. Hence either f(X)
has a root in K or it must be irreducible over K.
Chapter 5
5.1. By Theorem 1.17, an integral domain D always admits a monomorphism into a ﬁeld
j : D −→ F (e.g., F can be taken to be the ﬁeld of fractions of D), so any subgroup U D
becomes isomorphic to a subgroup jU F
, and if U is ﬁnite so is jU. Therefore jU and U
are cyclic.
5.2. The only root of X
2
+ 1 in F
2
is the multiple root 1.
5.3. The ﬁeld F
p
d[X]/(f(X)) is an extension of F
p
d which has degree n, hence it is a ﬁnite
ﬁeld with p
dn
elements, hence Proposition 5.6 implies that it is isomorphic to F
p
dn. Since the
extension F
p
dn/F
p
d is normal, F
p
dn is a splitting ﬁeld for f(X) over F
p
d.
5.4. (a) Here 41 is prime. Since 8 [ (41 −1), there is a primitive 8th root of unity in F
41
. 6 is
a primitive root for F
41
and 6
5
≡ 27 (mod 4)1 has order 8.
(b) Here 5 is prime 4 [ (5 −1), so there is a primitive 4th root of unity in F
5
, but no primitive
8th root of unity. In fact, 2 and 3 have order 4, so these are primitive roots for F
5
. Notice that
in F
5
[X],
X
8
−1 = (X
4
−1)(X
4
+ 1) = (X
4
−1)(X
2
−2)(X
2
−3),
where the polynomials X
2
−2 and X
2
−3 are irreducible. Therefore F
25
is the splitting ﬁeld for
X
8
−1 over F
5
and we have F
25
∼
= F
5
(u) = F
5
(v), where u
2
= 2 and v
2
= 3, so ±u and ±v are
primitive 8th roots of unity. To ﬁnd an element of order 24 in F
25
, we ﬁrst ﬁnd one of order 3.
Consider the polynomial X
2
+ X + 1 ∈ F
5
[X]; in F
5
, this has roots which have order 3. These
roots are given by (−1 ±w)/2, where w
2
= (1 −4) = −3 = 2, hence they are
(−1 ±u)
2
= −3 ±3u.
Now the elements ±(2 ±2u)u = ±(±4 + 2u) = ±4 ±2u all have order 8 3 = 24.
(c) Here 11 is prime and 8 [ (121−1) = 120, so F
121
is the splitting ﬁeld of X
8
−1 over F
11
. The
polynomial X
2
+1 is irreducible over F
11
so F
121
= F
11
(z) where z
2
= −1. Since 120 = 835,
it is suﬃcient to ﬁnd elements of order 8, 3 and 5 whose product will have order 120.
Suppose that a+bz ∈ F
121
with a, b ∈ F
11
. If this element has order 8, then (a+bz)
2
= ±z.
So let us solve
(a
2
−b
2
) + 2abz = z.
Then 2ab = 1 and b
2
= a
2
, hence b = ±a. Now we have 2a
2
= ±1 and so a
2
= ±1/2 = ±6.
Now 6 is not a square in F
11
but
7
2
≡ −6 ≡ 4
2
(mod 11),
so we have a = 4, b = ±4 and a = 7, b = ±7. Therefore the elements of order 8 in F
121
are
4 ±4z and 7 ±7z.
By the same approach as in (b), the elements of order 3 in F
121
are (−1 ±5z)/2 = 5 ±8z.
2 is a primitive root for F
11
so 4 = 2
2
has order 5.
Combining these we obtain the following primitive roots for F
121
: 7 ±z, 10 ±4z.
(d) In F
2
[X] we have X
8
−1 = (X −1)
8
, whose only root in F
2
is 1. So the splitting ﬁeld is F
2
.
109
5.5. Notice that F
p
(w) is a splitting ﬁeld of the separable polynomial X
p
d
−1
− 1 over F
p
, so
if w ∈ F
p
then F
p
(w) F
p
. Since F
p
(w) = F
p
d we have d ; we also have deg
F
p
w = d.
The number of conjugates of w is d, hence each primitive root of F
p
d has d conjugates and the
total number of these is the number of generators of the cyclic group F
p
d
∼
= Z/(p
d
− 1), i.e.,
ϕ(p
d
− 1). Hence d [ ϕ(p
d
− 1). This can also be interpreted in terms of the evident action of
Gal(F
p
d/F
p
)
∼
= Z/d on the set of all primitive roots of F
p
d; each orbit has exactly d elements,
so the number of orbits is ϕ(p
d
−1)/d which is an integer.
5.6. (a) First note that g
t
a
(X) = −1, so g
a
(X) is separable, hence E/K is separable. If u ∈ E
is a root of g
a
(X), then for t ∈ F
p
d,
g
a
(u +t) = (u +t)
p
d
−(u +t) −a = (u
p
d
−u −a) + (t
p
d
−t) = (u
p
d
−u −a) = 0,
hence u+t is also a root of g
a
(X). This means that E = K(u) since all the other roots of g
a
(X)
lie in K(u). As g
a
(X) is irreducible over K, [E : K] = p
d
= [ Gal(E/K)[ and so the following
p
d
automorphisms are the elements of Gal(E/K):
σ
t
: E −→ E; σ
t
(u) = u +t (t ∈ F
p
d).
It is easy to check that for s, t ∈ F
p
d, σ
s
◦σ
t
= σ
s+t
. Hence there is an isomorphism Gal(E/K)
∼
=
F
p
d with σ
t
corresponding to t ∈ F
p
d.
(b) If g
a
(X) is irreducible over K then it cannot have a root in K since its degree is greater
than 1.
Conversely, suppose that g
a
(X) has no root in K. Then if u ∈ E is any root of g
a
(X) in
a splitting ﬁeld over K, the other roots are the p elements u + t ∈ E (t ∈ F
p
). If u + t
0
= u
is a conjugate of u with 0 = t
0
∈ F
p
, there must be an element τ
t
0
∈ Gal(E/K) for which
τ
t
0
(u) = u + t
0
. Then 'τ
t
0
` must be isomorphic to a nontrivial subgroup of F
p
, but this must
be F
p
since this group is simple. Hence, u must have p conjugates and so g
a
(X) is irreducible
over K.
(c) If K is a ﬁnite ﬁeld and d > 1 then if g
a
(X) were irreducible over K, then by (a), E would
be ﬁnite and Gal(E/K)
∼
= F
p
d. But F
p
d is not cyclic, yet we know from Proposition 5.23 that
Gal(F
p
d/F
p
)
∼
= Z/d is cyclic.
5.7. (a) By Proposition 5.12, F
q
is a cyclic group. If p = 2 then [F
2
d
[ = 2
d
−1, which is odd,
so every element of F
2
d
is a square; we may therefore take λ
2
d(u) = 1 for all u ∈ F
2
d
. So now
suppose that p is odd. Then [F
p
d
[ = p
d
− 1, which is even. The set of squares in F
p
d
is the
normal subgroup
(F
p
d
)
2
= ¦u
2
: u ∈ F
p
d
¦ F
p
d
and it is easily seen that its quotient group has order 2, hence
F
p
d
/(F
p
d
)
2
∼
= ¦±1¦.
We may use this group isomorphism to deﬁne λ
q
. Clearly we have
ker λ
q
= (F
p
d
)
2
.
λ
q
is surjective if and only if p is odd.
Remark: when d = 1, λ
p
(u) =
u
p
, the Legendre symbol of u from Number Theory.
(b) If u ∈ Σ
q
, then either u = 0 or u = 0 and u = (±v)
2
for some v ∈ F
q
. Thus we have
[Σ
q
[ = 1 +
(q −1)
2
=
(q + 1)
2
.
Then
[t −Σ
q
[ = [Σ
q
[ =
(q + 1)
2
.
110 SOLUTIONS
(c) Since Σ
q
∪ (t −Σ
q
) ⊆ F
q
, we have
q [Σ
q
∪ (t −Σ
q
)[ = [Σ
q
[ +[t −Σ
q
[ −[Σ
q
∩ (t −Σ
q
)[.
This implies that
q (q + 1) −[Σ
q
∩ (t −Σ
q
)[
and so
[Σ
q
∩ (t −Σ
q
)[ 1.
Thus for every t ∈ F
q
, there are u, v ∈ F
q
(possibly 0) for which u
2
= t −v
2
, whence t = u
2
+v
2
.
(d) By (c), we may write −1 = a
2
+b
2
for some a, b ∈ F
q
, i.e.,
1
2
+a
2
+b
2
= 0.
Chapter 6
6.1. Now when n = 1, G
∼
= Z/p, which is abelian. Suppose that the result holds whenever
[G[ = p
k
with k < n. Now if [G[ = p
n
, recall that by Cauchy’s Lemma, the centre Z of G
is nontrivial. Hence G/Z has order [G/Z[ = p
k
with k < n. By the inductive hypothesis,
there is a normal subgroup M G/Z with [M[ = p
k−1
. By one of the Isomorphism Theorems,
there is a normal subgroup N G containing Z and satisfying N/Z = M ⊆ G/Z. Clearly
[N[ = [Z[ [M[ = p
n−1
. This establishes the inductive step and hence the desired result.
6.2. In this situation, for any nonzero t ∈ K, −t = t (since otherwise 2t = 0 and so t = 0).
If ζ ∈ K is a primitive nth root of unity, then (−ζ)
n
= (−1)
n
ζ
n
= −1, while (−ζ)
2n
=
(−1)
2n
ζ
2n
= 1. Hence −ζ ∈ K is a primitive 2nth root of unity.
6.3. Write n = 2
k
p
r
1
1
p
r
s
s
, where each p
j
is an odd prime, p
1
< p
2
< < p
s
, r
j
1 and
k 0. Then
ϕ(n) = ϕ(2
k
)ϕ(p
r
1
1
) ϕ(p
r
s
s
) = ϕ(2
k
)(p
1
−1)p
r
1
−1
1
(p
s
−1)p
r
s
−1
s
.
If s > 0 then ϕ(n) [ 4 happens precisely when r
1
= = r
s
= 1 and one of the following
possibilities occurs:
• p
1
= 5, s = 1 and k = 0 (hence n = 5);
• p
1
= 3, s = 1 and k = 0, 1, 2 (hence n = 3, 6, 12);
• s = 0 and k = 0, 1, 3 (hence n = 1, 2, 4, 8).
Q(i): Here degree [Q(i) : Q] = 2 and clearly the four 4th roots of unity ±1, ±i lie in this ﬁeld.
As ϕ(5) = 4, it has no 5th roots of unity except 1. If it contained a 3rd root of unity then it
would contain
√
3 and so Q(
√
3, i) Q(i) which is impossible since [Q(
√
3, i) : Q] = 4. From
this we see that the only roots of unity in Q(i) are ±1, ±i.
Q(
√
2 i): This ﬁeld contains only the square roots of unity ±1.
Q(
√
3 i): This contains the six 6th roots of unity ±1, ±
1
2
±
√
3
2
i.
Q(
√
5 i): This ﬁeld contains only the square roots of unity ±1.
6.4. (a) We have ϕ(24) = ϕ(8)ϕ(3) = 4 2 = 8. The elements of Z/24 which are invertible
are the residue classes modulo 24 of the numbers 1, 5, 7, 11, 13, 17, 19, 23. For each of these
numbers r, the residue class modulo 24, r, satisﬁes r
2
= 1, hence these all have order 2 except 1
which has order 1. Since (Z/24)
is abelian, it is isomorphic to Z/2 Z/2 Z/2. The eﬀect of
these elements on Q(ζ
24
) is given by r ζ
r
24
. Notice that 23 acts like complex conjugation. The
eﬀect on Q(cos(π/12)) is given by
r cos(π/12) = cos(πr/12),
111
so in particular,
−r cos(π/12) = cos(−πr/12) = cos(πr/12) = r cos(π/12).
(b) This is similar to (a). We have ϕ(20) = ϕ(4)ϕ(5) = 2 4 = 8 and the elements of (Z/20)
are the residue classes modulo 20 of the numbers 1, 3, 7, 9, 11, 13, 17, 19. This time there are
elements of order 4, for instance 7 and 13. Then we have (Z/20)
∼
= Z/2 Z/4.
6.5. For any n 1, let ζ
n
= e
2πi/n
= cos(2π/n) + sin(2π/n) i. Notice that if n is odd, then
Q(ζ
n
) = Q(−ζ
n
) where −ζ
n
is a primitive 2nth root of unity, so we might as well assume that
n is even from now on. We also have
ζ
n
−ζ
−1
n
= 2 sin(2π/n) i ∈ Q(ζ
n
).
(a) If 4 n then writing n = 2k with k odd, we have
[Q(ζ
n
) : Q] = ϕ(2k) = ϕ(2)ϕ(k) = ϕ(k),
while
[Q(ζ
2n
) : Q] = ϕ(4k) = ϕ(4)ϕ(k) = 2ϕ(k).
Hence, Q(ζ
n
) cannot contain ζ
2n
and by another simple argument it cannot contain i = ζ
k
2n
. So
we see that sin(2π/n) / ∈ Q(ζ
n
) in this situation. Notice that since i = ζ
k
2n
,
sin(2π/n) =
ζ
2
2n
−ζ
−2
2n
2i
∈ Q(ζ
2n
),
and by Theorem 6.3,
sin(2π/n) ∈ Q(ζ
2n
) ∩ R = Q(cos(π/n)).
Also, we have
[Q(cos(π/n)) : Q] = 2[Q(cos(2π/n)) : Q],
hence
Q(cos(π/n)) = Q(cos(2π/n))(sin(2π/n))
and
[Q(cos(2π/n))(sin(2π/n)) : Q(cos(2π/n))] = 2,
with
minpoly
Q(cos(2π/n)),sin(2π/n)
(X) = X
2
+ cos
2
(2π/n) −1.
If 4 [ n, we can write n = 4. Then i = ζ
n
, so i ∈ Q(ζ
n
), whence
sin(π/2) = sin(2π/n) =
ζ
n
−ζ
−1
n
i
∈ Q(ζ
n
).
Clearly
sin(π/2) ∈ Q(ζ
n
) ∩ R = Q(cos(2π/n)).
Consider the automorphism σ ∈ Gal(Q(ζ
n
)/Q)) for which σ(ζ
n
) = ζ
2+1
n
= −ζ
n
; it is easy to
see that σ has order 2. Then
σ(cos(2π/n)) = σ(cos(π/2)) = −cos(π/2),
σ(cos(π/)) = cos(π/),
σ(sin(2π/n)) = σ(sin(π/2)) =
−ζ
n
+ζ
−1
n
2(−ζ
n
)
=
sin(π/2) if is odd,
−sin(π/2) if is even.
From this we ﬁnd that when is odd,
Q(cos(2π/n)) = Q(cos(π/2)) = Q(cos(π/))(sin(π/2)) = Q(sin(π/2)),
since cos(π/) = 1 −2 sin
2
(π/2) ∈ Q(sin(π/2)). Thus we have [Q(sin(π/2)) : Q] = 2ϕ() and
Gal(Q(sin(π/2))/Q) = Gal(Q(cos(π/2))/Q) = (Z/4)
/¦1, −1¦.
112 SOLUTIONS
Similarly, if is even,
[Q(cos(π/))(sin(π/2)) : Q(cos(π/))] = 2
and we must have
Q(cos(2π/n)) = Q(cos(π/2)) = Q(sin(π/2))
with
Gal(Q(sin(π/2))/Q) = Gal(Q(cos(π/2))/Q) = (Z/4)
/¦1, −1¦
(b) We have
sin
2
(π/12) =
1 −cos(π/6)
2
=
2 −
√
3
4
,
and so
sin(π/12) =
2 −
√
3
2
=
√
6 −
√
2
4
.
Then
Q(sin(π/12)) = Q(
√
6 −
√
2) = Q(
√
2,
√
3).
and
Gal(Q(sin(π/12))/Q))
∼
= (Z/4)
/¦1, −1¦
∼
= Z/2 Z/2.
Here the eﬀect of the coset of the residue class of r ∈ (Z/4)
is given by
r sin(π/12) =
ζ
r
24
−ζ
r
24
i
r
= sin(rπ/12) i
1−r
.
Explicitly we have
1 sin(π/12) = −1 sin(π/12) = sin(π/12) =
√
6 −
√
2
4
,
5 sin(π/12) = −5 sin(π/12) = sin(5π/12) =
√
6 +
√
2
4
,
7 sin(π/12) = −7 sin(π/12) = −sin(7π/12) =
−
√
6 −
√
2
4
,
11 sin(π/12) = −11 sin(π/12) = −sin(11π/12) =
−
√
6 +
√
2
4
.
In terms of the generators
√
2 and
√
3 these act by
1
√
2 =
√
2, 1
√
3 =
√
3, 5
√
2 = −
√
2, 5
√
3 =
√
3,
7
√
2 =
√
2, 5
√
3 = −
√
3, 11
√
2 = −
√
2, 11
√
3 = −
√
3.
6.6. (a) We have
[ Gal(Q(ζ
5
)/Q)[ = [Q(ζ
5
) : Q] = deg Φ
5
(X) = ϕ(5) = 4,
and (Z/5)
is cyclic generated by the residue class 2. The action is given by
2 ζ
5
= ζ
2
5
, 2
2
ζ
5
= ζ
4
5
, 2
3
ζ
5
= ζ
3
5
, 2
4
ζ
5
= ζ
5
.
(b) We have ζ
5
+ζ
−1
5
= 2 cos(2π/5) and Φ
5
(ζ
5
) = 0, so since ζ
3
5
= ζ
−2
5
and ζ
4
5
= ζ
−1
5
,
(ζ
2
5
+ζ
−2
5
) + (ζ
5
+ζ
−1
5
) + 1 = 0
and therefore
(ζ
5
+ζ
−1
5
)
2
+ (ζ
5
+ζ
−1
5
) −1 = 0.
Hence
4 cos
2
(2π/5) + 2 cos(2π/5) −1 = 0.
113
The quadratic polynomial 4X
2
+ 2X − 1 ∈ Z[X] has discriminant 20 which is not a square in
Q, so this is this polynomial is irreducible over Q, therefore
minpoly
Q,cos(2π/5)
(X) = X
2
+
1
2
X −
1
4
.
The roots of this are
−1 ±
√
5
4
. As cos(2π/5) > 0 we must have cos(2π/5) =
−1 +
√
5
4
. We
also have cos(4π/5) =
−1 −
√
5
4
. As sin(2π/5) > 0,
sin
2
(2π/5) = 1 −cos
2
(2π/5) = 1 −
1 + 5 −2
√
5
16
=
5 +
√
5
8
,
hence sin(2π/5) =
5 +
√
5
8
.
(c) Gal(Q(ζ
5
)/Q)
∼
= Z/4 and has 3 subgroups ¦1¦ ¦1, 4¦ Gal(Q(ζ
5
)/Q), giving the following
tower of subﬁelds.
Q(ζ
5
)
2
Q(ζ
5
)
'4`
= Q(cos(2π/5)) = Q(
√
5)
2
Q
6.7. (a) We have
ξ
2
=
(p−1)/2
¸
r=1
(ζ
r
p
−ζ
−r
p
)
2
= (−1)
(p−1)/2
(p−1)/2
¸
r=1
(ζ
r
p
−ζ
−r
p
)(ζ
−r
p
−ζ
r
p
)
= (−1)
(p−1)/2
(p−1)/2
¸
r=1
(1 −ζ
−2r
p
)(1 −ζ
2r
p
)
= (−1)
(p−1)/2
p−1
¸
r=1
(1 −ζ
2r
p
)
= (−1)
(p−1)/2
(p−1)
¸
s=1
(1 −ζ
s
p
)
since each congruence 2x ≡ t (mod p) has exactly one solution modulo p for each t.
(b) Since
(−1)
(p−1)/2
=
−1 if p ≡ 1 (mod 4),
1 if p ≡ 3 (mod 4),
and
p−1
¸
s=1
(1 −ζ
s
p
) = Φ
p
(1) = p,
the result follows.
(c) Taking square roots we ﬁnd that
ξ =
±
√
p if p ≡ 1 (mod 4),
±
√
p i if p ≡ 3 (mod 4).
As ξ ∈ Q(ζ
p
), we see that
√
p ∈ Q(ζ
p
) if p ≡ 1 (mod 4) and
√
p i ∈ Q(ζ
p
) if p ≡ 3 (mod 4).
114 SOLUTIONS
6.8. Recall the wellknown formula
σ(i
1
i
r
)σ
−1
= (σ(i
1
) σ(i
r
)).
Then for 1 r n −2 we have
(1 2 n)
r
(1 2)(1 2 n)
n−r
= (1 2 n)
r
(1 2)((1 2 n)
r
)
−1
= (r + 1 r + 2),
while
(1 2 n)
n−1
(1 2)((1 2 n)
n−1
)
−1
= (1 2 n)
−1
(1 2)((1 2 n)
−1
)
−1
= (n 1) = (1 n).
This means that every such 2cycle (r + 1 r + 2) is in H. Also recall that every permutation
ρ ∈ S
n
is a product of 2cycles, so it suﬃces to show that every 2cycle (a b) ∈ S
n
is a product
of 2cycles of the form (r + 1 r + 2). Assuming that a < b, we also have
(a b) = (b −1 b) (a + 2 a + 3)(a + 1 a + 2)(a a + 1)(a + 1 a + 2)(a + 2 a + 3) (b −1 b),
and this is in H. Hence H = S
n
.
6.9. (a) For each u ∈ E,
σ(T(u)) = σ(u +σ(u) +σ
2
(u) + +σ
n−1
(u))
= σ(u) +σ
2
(u) + +σ
n
(u)
= σ(u) +σ
2
(u) + +σ
n−1
(u) +u = T(u),
so T(u) is ﬁxed by σ and all its powers, hence by Gal(E/K). Therefore T(u) is in E
Gal(E/K)
= K.
It is straightforward to verify that the resulting function Tr
E/K
: E −→ K is Klinear.
(b) Let v ∈ E and suppose that Tr
E/K
(v) = 0. By Artin’s Theorem 6.15, the linear combination
of characters id +σ + +σ
n−1
must be linearly independent, so there is an element t ∈ E for
which
Tr
E/K
t = t +σ(t) + +σ
n−1
(t) = 0.
Then
u = vσ(t) + (v +σ(v))σ
2
(t) + + (v +σ(v)σ
2
(t) + +σ
n−2
(v))σ
n−1
(t)
satisﬁes
u −σ(u) = v
σ(t) +σ
2
(t) + +σ
n−1
(t)
−
σ(v) + +σ
n−1
(v)
t
= v
t +σ(t) +σ
2
(t) + +σ
n−1
(t)
−
v +σ(v) + +σ
n−1
(v)
t
= (Tr
E/K
t)v −(Tr
E/K
v)t = (Tr
E/K
t)v.
So we obtain
v =
1
Tr
E/K
t
u −σ
1
Tr
E/K
t
u
.
6.10. (a) This can be proved by induction on n. Write
e
[m]
r
=
¸
i
1
<i
2
<<i
r
m
X
i
1
X
i
r
, s
[m]
r
=
¸
1im
X
r
i
.
Then we easily ﬁnd that
e
[m]
r
= e
[m−1]
r
+e
[m−1]
r−1
X
m
, s
[m]
r
= s
[m−1]
r
+X
r
m
.
Notice also that e
[m]
r
= 0 whenever r > m. The desired result is that for all n 1 and k 1,
s
[n]
k
= e
[n]
1
s
[n]
k−1
−e
[n]
2
s
[n]
k−2
+ + (−1)
k−1
e
[n]
k−1
s
[n]
1
+ (−1)
k
ke
[n]
k
.
115
When n = 1 we have s
[1]
r
= X
r
1
and e
[1]
1
= X
1
from which the result follows. Now suppose that
the result is true for some n 1. Then s
[n+1]
k
= s
[n]
k
+X
k
n+1
, while
e
[n+1]
1
s
[n+1]
k−1
−e
[n+1]
2
s
[n+1]
k−2
+ + (−1)
k−1
e
[n+1]
k−1
s
[n+1]
1
+ (−1)
k
ke
[n+1]
k
=
(e
[n]
1
+X
n+1
)(s
[n]
k−1
+X
k−1
n+1
) −(e
[n]
2
+e
[n]
1
X
n+1
)(s
[n]
k−2
+X
k−2
n+1
) +
+ (−1)
k−1
(e
[n]
k−1
+e
[n]
k−2
X
n+1
)(s
[n]
1
+X
n+1
) + (−1)
k
k(e
[n]
k
+e
[n]
k−1
X
n+1
)
= s
[n]
k
+ (e
[n]
1
X
k−1
n+1
−e
[n]
2
X
k−2
n+1
+ + (−1)
k−1
e
[n]
k−1
X
n+1
)
+ (s
[n]
k−1
−e
[n]
1
s
[n]
k−2
+ + (−1)
k−1
e
[n]
k−2
s
[n]
1
+ (−1)
k
ke
[n]
k−1
)X
n+1
+ (X
k
n+1
−e
[n]
1
X
k−1
n+1
+ + (−1)
k−1
e
[n]
k−2
X
2
n+1
)
= s
[n]
k
+X
k
n+1
= s
[n+1]
k
,
which demonstrates the inductive step.
(b)(i) We have h
1
= e
1
, h
2
= e
2
1
−e
2
and h
3
= e
3
−2e
1
e
2
+e
3
1
.
(ii) This can be done by induction on n in a similar way to part (a).
Q( 2, ζ3 ) H qq ggg qq gggggnnnnn gg qq 2 ggggg nnnnn qq ggg 2 nn 2 qq gggg √ √ 2 √ gggg qq3 3 3 3 qq Q( 2 ζ3 ) Q( 2) Q( 2 ζ3 ) qq yyy N N ff I qq yyy ff qq yyy ff qq yyy ff q yyy3 ff 3 yyy ff 3 Q(ζ3 ) yyy f k Q yyy fff kkkk k 2 yyy ff kkkk kk yyy ff kkkk yy kkkk QS
√ 3
ww ww ww ww 3 www w {id, (1 3)} {id, (1 2)} {id, (2 3)} w ww ww ww 2 ww 2 2 ww
o  ooo  ooo  o ooo  3  3 oooo 3  oo  ooo  oo  o  ooo  ooo  o
Gal(E/Q) ∼ S3 =
2
{id, (1 2 3), (1 3 2)}
{id}
√ The Galois Correspondence for Q( 3 2, ζ3 )/Q
ii
Introduction: What is Galois Theory? Much of early algebra centred around the search for explicit formulae for roots of polynomial equations in one or more unknowns. The solution of linear and quadratic equations in a single unknown was well understood in antiquity, while formulae for the roots of general real cubics and quartics was solved by the 16th century. These solutions involved complex numbers rather than just real numbers. By the early 19th century no general solution of a general polynomial equation ‘by radicals’ (i.e., by repeatedly taking nth roots for various n) was found despite considerable eﬀort by many outstanding mathematicians. Eventually, the work of Abel and Galois led to a satisfactory framework for fully understanding this problem and the realization that the general polynomial equation of degree at least 5 could not always be solved by radicals. At a more profound level, the algebraic structure of Galois extensions is mirrored in the subgroups of their Galois groups, which allows the application of group theoretic ideas to the study of ﬁelds. This Galois Correspondence is a powerful idea which can be generalized to apply to such diverse topics as ring theory, algebraic number theory, algebraic geometry, diﬀerential equations and algebraic topology. Because of this, Galois theory in its many manifestations is a central topic in modern mathematics. In this course we will focus on the following topics. • The solution of polynomial equations over a ﬁeld, including relationships between roots, methods of solutions and location of roots. • The structure of ﬁnite and algebraic extensions of ﬁelds and their automorphisms. We will study these in detail, building up a theory of algebraic extensions of ﬁelds and their automorphism groups and applying it to solve questions about roots of polynomial equations. The techniques we will meet can also be applied to study the following some of which may be met by people studying more advanced courses. • Classic topics such as squaring the circle, duplication of the cube, constructible numbers and constructible polygons. • Applications of Galois theoretic ideas in Number Theory, the study of diﬀerential equations and Algebraic Geometry. There are many good introductory books on Galois Theory, some of which are listed in the Bibliography. In particular, [2, 3, 8] are all excellent sources and have many similarities to the present approach to the material.
c A. J. Baker (2009)
Contents
Introduction: What is Galois Theory? Chapter 1. Integral domains, ﬁelds and polynomial rings Basic notions, convention, etc 1.1. Recollections on integral domains and ﬁelds 1.2. Polynomial rings 1.3. Identifying irreducible polynomials 1.4. Finding roots of complex polynomials of small degree 1.5. Automorphisms of rings and ﬁelds Exercises on Chapter 1 Chapter 2. Fields and their extensions 2.1. Fields and subﬁelds 2.2. Simple and ﬁnitely generated extensions Exercises on Chapter 2 Chapter 3. Algebraic extensions of ﬁelds 3.1. Algebraic extensions 3.2. Splitting ﬁelds and Kronecker’s Theorem 3.3. Monomorphisms between extensions 3.4. Algebraic closures 3.5. Multiplicity of roots and separability 3.6. The Primitive Element Theorem 3.7. Normal extensions and splitting ﬁelds Exercises on Chapter 3 Chapter 4. Galois extensions and the Galois Correspondence 4.1. Galois extensions 4.2. Working with Galois groups 4.3. Subgroups of Galois groups and their ﬁxed ﬁelds 4.4. Subﬁelds of Galois extensions and relative Galois groups 4.5. The Galois Correspondence and the Main Theorem of Galois Theory 4.6. Galois extensions inside the complex numbers and complex conjugation 4.7. Galois groups of even and odd permutations 4.8. Kaplansky’s Theorem Exercises on Chapter 4 Chapter 5. Galois extensions for ﬁelds of positive characteristic 5.1. Finite ﬁelds 5.2. Galois groups of ﬁnite ﬁelds and Frobenius mappings 5.3. The trace and norm mappings Exercises on Chapter 5 Chapter 6. A Galois Miscellany
iii
ii 1 1 1 6 11 15 17 21 25 25 27 30 33 33 36 39 42 44 48 50 50 53 53 54 56 57 58 60 60 63 66 69 69 72 74 75 77
iv
CONTENTS
6.1. A proof of the Fundamental Theorem of Algebra 6.2. Cyclotomic extensions 6.3. Artin’s Theorem on linear independence of characters 6.4. Simple radical extensions 6.5. Solvability and radical extensions 6.6. Symmetric functions Exercises on Chapter 6 Bibliography Solutions Chapter Chapter Chapter Chapter Chapter Chapter 1 2 3 4 5 6
77 77 82 84 85 89 90 93 95 95 101 104 105 108 110
CHAPTER 1
Integral domains, ﬁelds and polynomial rings
Basic notions, convention, etc In these notes, a ring will always be a ring with unity 1 = 0. Most of the rings encountered will also be commutative. An ideal I R will always mean a twosided ideal. An ideal I R in a ring R is proper if I = R, or equivalently if I R. Under a ring homomorphism ϕ : R −→ S, 1 ∈ R is sent to 1 ∈ S, i.e., ϕ(1) = 1. 1.1. Definition. Let ϕ : R −→ S be a ring homomorphism. • ϕ is a monomorphism if it is injective, i.e., if for r1 , r2 ∈ R, ϕ(r1 ) = ϕ(r2 ) =⇒ r1 = r2 , or equivalently if ker ϕ = {0}. • ϕ is an epimorphism if it is surjective, i.e., if for every s ∈ S there is an r ∈ R with ϕ(r) = s. • ϕ is an isomorphism if it is both a monomorphism and an epimorphism, i.e., if it is invertible (in which case its inverse is also an isomorphism). 1.1. Recollections on integral domains and ﬁelds The material in this section is standard and most of it should be familiar. Details may be found in [3, 5] or other books containing introductory ring theory. 1.2. Definition. A commutative ring R in which there are no zerodivisors is called an integral domain or an entire ring. This means that for u, v ∈ R, uv = 0 (i) (ii) (iii) (iv) =⇒ u = 0 or v = 0.
1.3. Example. The following rings are integral domains. The ring of integers, Z. If p is a prime, the ring of integers modulo p, Fp = Z/p = Z/(p). The rings of rational numbers, Q, real numbers, R, and complex numbers, C. The polynomial ring R[X], where R is an integral domain; in particular, the polynomial rings Z[X], Q[X], R[X] and C[X] are all integral domains. R be a proper ideal in a commutative ring R. uv ∈ I =⇒ u ∈ I or v ∈ I.
1.4. Definition. Let I
• I is a prime ideal if for u, v ∈ R,
• I is a maximal ideal R if whenever J R is a proper ideal and I ⊆ J then J = I. • I R is principal if I = (p) = {rp : r ∈ R} for some p ∈ R. Notice that if p, q ∈ R, then (q) = (p) if and only if q = up for some unit u ∈ R. We also write p  x if x ∈ (p). • p ∈ R is prime if (p) R is a prime ideal; this is equivalent to the requirement that whenever p  xy with x, y ∈ R then p  x or p  y.
1
If n n is a prime. FIELDS AND POLYNOMIAL RINGS • R is a principal ideal domain if it is an integral domain and every ideal I R is principal. Proof. hence pt = 0 for any t ∈ R. this p is called the characteristic of R and denoted char R. Remark.10. Lemma. Proof. So suppose that p > 0.5. hence Z/ ker η is an integral domain. 0. ker η Z is a proper ideal and using the Isomorphism Theorems we see that there is a quotient monomorphism η : Z/ ker η −→ R which allows us to identify the quotient ring Z/ ker η with the image ηZ ⊆ R as a subring of R.8. Consider p = char R. Example. n 0. its characteristic char R is a prime. 1. 1. Here is a useful and important fact about rings which contain a ﬁnite prime ring Fp . 1 2 p−1 p j p (p − 1)! (p − 1)! =p× . Let R be a commutative ring and I R an ideal. Since 1 ∈ R is nonzero.2 1. so I = (n) for some n ∈ Z which we can always take to be nonnegative. 1. The rings Z and Z/n = Z/(n) for n > 0 are often called core rings. 1. This subring is sometimes called the characteristic subring of R. j! (p − j)! j! (p − j)! Now suppose that 1 p − 1. If p = 0 we are done. then (u + v)p = up + v p . The Binomial Expansion yields (1. ker η = (p) is prime ideal and so by Example 1.1) (u + v)p = up + j p p−1 p p−2 2 p u v+ u v + ··· + uv p−1 + v p . −n 0 if n = 0. Proposition. Then we have = . the rings Z and Fp = Z/p = Z/(p) for p > 0 a prime are called prime rings. (ii) The quotient ring R/I is a ﬁeld if and only if I is a maximal ideal. We have p1 = 0 in R. allowing us to identify these rings. When considering integral domains. v ∈ R.5. every integral domain contains as a subring either Z = Z/(0) (if char R = 0) or Z/(p) if p = char R > 0 is a nonzero prime. If R is an integral domain. But every subring of an integral domain is itself an integral domain. 1. When discussing a ring with unit R.6. In particular. INTEGRAL DOMAINS. we can consider it as containing as a subring of the form Z/(char R) since the quotient homomorphism η : Z/(char R) −→ R gives an isomorphism Z/(char R) −→ im η. Hence Z is a principal ideal domain. 1. Every ideal I Z is principal. (i) The quotient ring R/I is an integral domain if and only if I is a prime ideal. i.. Theorem (Idiot’s Binomial Theorem). The quotient monomorphism η : Z/ ker η −→ R identiﬁes Z/ ker η with the subring im η = im η of the integral domain R. n η(n) = n1 = −(1 + · · · + 1) if n < 0.6(i). p is a prime. If u.e. the quotient ring Z/n = Z/(n) is an integral domain if and only if For any (not necessarily commutative) ring with unity there is an important ring homomorphism η : Z −→ R called the unit or characteristic homomorphism which is deﬁned by 1 + · · · + 1 if n > 0. Let R be a commutative ring with unit containing Fp for some prime p > 0. By Example 1.7. there is a unique nonnegative integer p 0 such that ker η = (p). Now by Proposition 1.9. Example.7.
This is certainly true for the integers Z which are contained in the ﬁeld of rational numbers Q. Let k be a ﬁeld. (1. (u + v)p = up + v p .13.14. Proof. j p 1 = 0. the prime ring Fp is itself a ﬁeld.2b) Hence in R we have p j ≡ 0 (mod p). 1. v ∈ k and uv = 0. hence v = 0. We will denote the set of all invertible elements of R by R× and note that it always forms a group under multiplication. Suppose that u. the quotient ring Z/n is a ﬁeld if and only if n is a prime. v must be 0. A natural question to ask is whether D is isomorphic to a subring of a ﬁeld. we can multiply by to obtain v = u−1 uv = 0. Of course. Let R be an integral domain. Proposition. Now let D be an integral domain. u−1 Proof. Then (1 − tv)p = 0 and so tv = 1. p = uv =⇒ u or v is a unit. and we might as well assume that u = tp for some t ∈ R. i. R and C are all ﬁelds. so since this number is an integer it must be divisible by p.e. 1. 1. Definition. We usually write u−1 for this element v.12. Definition. Definition. So at least one of u. i. p p . If p ∈ R is a nonzero prime then it is an irreducible. v ∈ R. there is always a ‘smallest’ such ﬁeld which is unique up to an isomorphism. Let R be a ring. If u = 0. which is necessarily unique and is called the (multiplicative) inverse of u in R.2a) or equivalently (1..e.15. v ∈ R.. This is equivalent to requiring that k× = k − {0}. we obtain the required equation in R. A commutative ring k is a ﬁeld if every nonzero element u ∈ k is a unit. The familiar rings Q. j Combining the divisibility conditions of (1. Lemma. Example. 1. 1. Every ﬁeld is an integral domain.11. An element u ∈ R is a unit if it is invertible. and for a prime p > 0.1. A nonzero element p ∈ R is irreducible if for u. showing that v is a unit with inverse t.1.2) with the expansion of (1.16. there is and element v ∈ R for which uv = 1 = vu. Then p  u or p  v. The ﬁelds Q and Fp where p > 0 is a prime are the prime ﬁelds. j! or (p − j)!. If n 1.1). Suppose that p = uv for some u. . we can view Z as a subring of any subﬁeld of the complex numbers so an answer to this question may not be unique! However. RECOLLECTIONS ON INTEGRAL DOMAINS AND FIELDS 3 There are no factors of p appearing in (p − 1)!. 1.
4 1. 1] ∈ Fr(D) and D with the subring im j ⊆ Fr(D). it is necessary to check that this relation is an equivalence relation.. there is a unique homomorphism ϕ : Fr(D) −→ F such that ϕ(t) = ϕ(t) for all t ∈ D ⊆ Fr(D). For example. namely (a . [0. (i) There is a ﬁeld of fractions of D. hence addition is well deﬁned. bd]. b] has [b. 1] which is a ring homomorphism. (i) Consider the set ∃! ϕ P(D) = {(a. This shows that Fr(D) is a ﬁeld. b] and [c . b] ∈ Fr(D) be nonzero. b d ) ∼ (ad + bc. b) = ϕ(a)ϕ(b)−1 . [a. Then b = 0 and [a. bd). b = 0}. a = 0. b ∈ D. 1]. notice that for any b = 0. b) by [a. Now introduce an equivalence relation ∼ on P(D). which contains D as a subring. it is easy to check that it is a monomorphism. (ii) Consider the function Φ : P(D) −→ F .17. b ] = [a. and so (a d + b c . FIELDS AND POLYNOMIAL RINGS 1. b] = [0. We also have [a. Fr(D) is a ﬁeld. d ] = [c. this is left as an exercise. b] + [c. b] = [1. ϕ D inc /F < Fr(D) Proof. d] = [ac. INTEGRAL DOMAINS. ba] = [1. b) : a. 1] and unit 1 = [1. so multiplication is also well deﬁned. Then [a. a] = [ab. We can view D as a subring of Fr(D) using the map j : D −→ Fr(D). Φ(a. Let [a. Now let [a. Theorem. b d ) ∼ (ac. It is now straightforward to show that Fr(D) is a commutative ring with zero 0 = [0. Of course. b) ⇐⇒ ab = a b. b] ∈ Fr(D). Therefore we may identify t ∈ D with j(t) = [t. [b. d] = [ad + bc. 1]. b) which is equivalent to requiring that a = 0. We need to verify that these operations are well deﬁned. We deﬁne addition and multiplication on Fr(D) by [a. b][b. . b]. d]. Fr(D). b][c. 1) ∼ (a. 1] if and only if (0. b ) ∼ (a. In fact. as we will soon see. j(t) = [t. then (a d + b c )bd = a d bd + b c bd = ab d d + b bcd = (ad + bc)b d .e. A similar calculation shows that (a c . Let D be an integral domain. (ii) If ϕ : D −→ F is a ring monomorphism into a ﬁeld F . i. b] = [0. b] and the set of equivalence classes by Fr(D). so [a. 1] = 1. bd]. a] ∈ Fr(D) satisfy [a. if [a . 1] if and only if a = b. a] as an inverse. bd). We denote the equivalence class of (a.
Then there is a unique induced ring homomorphism ϕ∗ : Fr(D1 ) −→ Fr(D2 ) which satisﬁes ϕ∗ (t) = ϕ(t) whenever t ∈ D1 ⊆ Fr(D1 ). If F is a ﬁeld then F = Fr(F ). The next three corollaries are left as an exercise. the identity homomorphism id : D −→ D induces the identity homomorphism (id)∗ = id : Fr(D) −→ Fr(D).18.e.20. so Φ is constant on each equivalence class of ∼. however as the word quotient is also associated with quotient rings we prefer to avoid using that terminology. (a) When working with a ﬁeld of fractions we usually adopt the familiar notation a = a/b = [a. × = = .20 is sometimes said to imply that the construction of Fr(D) is functorial in the integral domain D. Remarks. 1. b). then Φ(a . Corollary. b] b for the equivalence class of (a. D1 ϕ / D2 inc ϕ∗ inc Fr(D1 ) / Fr(D2 ) Moreover. Corollary.19. Let D1 and D2 be integral domains and let ϕ : D1 −→ D2 be a ring monomorphism. i. If D is a subring of a ﬁeld F . (b) Corollary 1. b ) = ϕ(a )ϕ(b )−1 = ϕ(a )ϕ(b)ϕ(b)−1 ϕ(b )−1 = ϕ(a b)ϕ(b)−1 ϕ(b )−1 = ϕ(ab )ϕ(b )−1 ϕ(b)−1 = ϕ(a)ϕ(b )ϕ(b )−1 ϕ(b)−1 = ϕ(a)ϕ(b)−1 = Φ(a. • For any integral domain D. b).1.21. 1. RECOLLECTIONS ON INTEGRAL DOMAINS AND FIELDS 5 If (a . ϕ([a. It is now easy to verify that ϕ is a ring homomorphism which agrees with ϕ on D ⊆ Fr(D). The rules for algebraic manipulation of such symbols are the usual ones for working with fractions.. Corollary. . D1 ϕ / D2 θ / D3 D id /D inc ϕ∗ inc θ∗ inc inc inc Fr(D1 ) / Fr(D2 ) / Fr(D3 ) / Fr(D) Fr(D) id∗ = id 1. b1 b2 b1 b2 b1 b2 b1 b2 b1 b2 The ﬁeld of fractions of an integral domain is sometimes called its ﬁeld of quotients. b]) = Φ(a. Hence we may deﬁne the function ϕ : Fr(D) −→ F .1. b). b ) ∼ (a. a1 a2 a1 b2 + a2 b1 a1 a2 a1 a2 a1 a2 + = . b). this construction has the following properties. • If ϕ : D1 −→ D2 and θ : D2 −→ D3 are monomorphisms between integral domains then θ∗ ◦ ϕ∗ = (θ ◦ ϕ)∗ as homomorphisms Fr(D1 ) −→ Fr(D3 ). 1. then Fr(D) ⊆ Fr(F ) = F and Fr(D) is the smallest subﬁeld of F containing D.
6 1.. . . . Xn ] = R[X1 . .sn (r) = ϕ(r) for all r ∈ R... . These polynomial rings have an important universal property. Xn over R. INTEGRAL DOMAINS. Again there is an inclusion monomorphism inc : R −→ R[X1 .... . . . this allows us to view R as a subring of R[X] and the inclusion function inc : R −→ R[X] is a monomorphism.. . Polynomial rings Let R be a commutative ring.sn : R[X1 . • ϕs (X) = s. . . . .. there is a unique ring homomorphism ϕs1 . . .sn . 1. Theorem (Homomorphism Extension Property). . Xn ] −→ S for which • ϕs1 . R[X1 . Addition and multiplication in R[X] are deﬁned by (p0 + p1 X + · · · + pm X m ) + (q0 + q1 X + · · · + qm X m ) = (p0 + q0 ) + (p1 + q1 )X + · · · + (pm + qm )X m .. . p1 . we inductively can deﬁne the ring of polynomials in n indeterminates X1 . R inc ϕ 8/ S R[X1 . We will make frequent use of the ring R[X] of polynomials over R in an indeterminate X. R inc ϕ /S = R[X] (ii) For n ∃! ϕs 1 and s1 . . Xn ] ∃! ϕs1 . . This consists of elements of form p(X) = p0 + p1 X + · · · + pm X m where m 0 and p0 . . and (p0 + p1 X + · · · + pm X m )(q0 + q1 X + · · · + qm X m ) = (p0 q0 ) + (p0 q1 + p1 q0 )X + · · · + (p0 qm + p1 qm−1 + · · · + pm−1 q1 + pm q0 )X 2m . . sn ∈ S. . Then R[X] is a commutative ring with the constant polynomials 0 and 1 as its zero and unit. .22.2.. . . • ϕs1 .. Let ϕ : R −→ S be a ring homomorphism. such p(X) are called polynomials.. . FIELDS AND POLYNOMIAL RINGS 1. . . . . . . We identify r ∈ R with the obvious constant polynomial. . .sn (Xi ) = si for i = 1. (i) For each s ∈ S there is a unique ring homomorphism ϕs : R[X] −→ S for which • ϕs (r) = ϕ(r) for all r ∈ R. n.. Xn ] which sends each element of R to itself considered as a constant polynomial. More generally.. . . pm ∈ R.. Xn−1 ][Xn ] for n 1..
. . . = In fact (X 2 + 1) is actually a maximal ideal and so Q[i] ⊆ C is a subﬁeld.. Corollary. sn ] is a subring of S. Consider the inclusion homomorphism inc : Q −→ C. . .. POLYNOMIAL RINGS 7 Proof. .sn as the extension of ϕ by evaluation at s1 . Notice that if we had used −i instead of i. .. . sn ∈ S. .. . Xn ] in the indeterminates X1 .rn : R[X1 . . Xn is an integral domain..sn R[X1 . . .24. Xn ] ⊆ S. .. (Sketch) (i) For a polynomial p(X) = p0 + p1 X + · · · + pm X m ∈ R[X]. . . .sn : R[X1 . ..sn (p(X1 .. (i) The ring R[X] of polynomials in an indeterminate X over R is an integral domain. .2. then εs1 . for which εi (X) = i. . (ii) is proved by induction on n using (i). It is then straightforward to check that ϕs is a ring homomorphism with the stated properties and moreover is the unique such homomorphism.rn = idr1 . . . 1. .. . As we will make considerable use of such rings we describe in detail some of their important properties. Then the polynomial ring k[X1 . Proposition. Here is an example illustrating how we will use such evaluation homomorphisms. . we would also have ε−i Q[X] = Q[i].. . rn ∈ R. .. Xn ] −→ S is called evaluation at s1 .. .. sn over R. . Xn ] −→ R. First we recall long division in a polynomial ring k[X] over a ﬁeld k. It is standard to write p(s1 . Notice also that in these examples we have ker ε−i = ker εi = (X 2 + 1) Q[X]. . Let k be a ﬁeld and n 1. These evaluation homomorphisms are related by complex conjugation since ε−i (p(X)) = εi (p(X)).1. .23. Example. . hence we also have Q[i] ∼ Q[X]/(X 2 + 1). sn and we denote its image by R[s1 . b ∈ Q. We easily see that εi Q[X] ⊆ C is a subring Q[i] ⊆ C consisting of the complex numbers of form a + bi with a.. ... ..25. (ii) The ring R[X1 . .. called the subring generated by s1 .. . . . We have the evaluation at i homomorphism εi . sn . .3) ϕs (p(X)) = p0 + p1 s + · · · + pm sm ∈ S. . full details can be found in a basic course on commutative rings or any introductory book on this subject. Then R[s1 . . 1. .. . sn ) = ϕs1 . . Let R be an integral domain. . . . ... . we deﬁne (1. Slightly more generally we may take the inclusion of a subring inc : R −→ S and s1 . sn ] = εs1 . evaluation at −i. . ... then we have the homomorphism εr1 . . . . Xn )). Xn ] of polynomials in the indeterminates X1 . . . . . . . . . ε−i . . which is equivalent to the functional equation ε−i = ( ) ◦ εi .. . later we will write Q(i) for this subﬁeld. An extremely important special case occurs when we start with the identity homomorphism id : R −→ R and r1 . 1. .sn = incs1 . ... We will refer to ϕs1 . Xn over R is an integral domain.
FIELDS AND POLYNOMIAL RINGS 1. In the situation discussed in this result. ﬁnd the quotient and remainder when f (X) = 6X 4 − 6X 3 + − 3X + 1 is divided by d(X) = 2X 2 + 1.26 is the following which makes use of the Euclidean Algorithm. giving q(X) = 3X 2 − 3X and r(X) = 1.26. In the usual notation we have the following calculation. 4−1 ≡ 4 (mod 5). For k = F5 . Let f (X). 3−1 ≡ 2 (mod 5). We have the following calculation. First notice that working modulo 5 we have f (X) = 10X 5 + 6X 4 − 6X 3 + 3X 2 − 3X + 1 ≡ X 4 + 4X 3 + 3X 2 + 2X + 1 (mod 5). For k = Q. 3X 2 − 3X 2X 2 + 1  6X 4 − 6X 3 + 3X 2 − 3X + 1 6X 4 + 0X 3 + 3X 2 + 0X + 0 − 6X 3 + 0X 2 − 3X + 1 − 6X 3 + 0X 2 − 3X + 0 1 Hence 6X 4 − 6X 3 + 3X 2 − 3X + 1 = (3X 2 − 3X)(2X 2 + 1) + 1. Notice also following multiplicative inverses in F5 : 2−1 ≡ 3 (mod 5). Solution. 1.8 1. ﬁnd the quotient and remainder when f (X) = 10X 5 + 6X 4 − + 3X 2 − 3X + 1 is divided by d(X) = 2X 2 + 1. Then there are unique polynomials q(X). giving q(X) = 3X 2 + 2X and r(X) = 1. r(X) = the remainder . . r(X) ∈ k[X] for which f (X) = q(X)d(X) + r(X) and either deg r(X) < deg d(X) or r(X) = 0. INTEGRAL DOMAINS.27. Example. Theorem (Long Division). the following names are often used. 6X 3 1. Let k be a ﬁeld. f (X) = the dividend . Example. We refer to the process of ﬁnding q(X) and r(X) as long division of f (X) by d(X). 3X 2 + 2X 2X 2 + 1 6X 4 + 4X 3 + 3X 2 + 2X + 1 6X 4 + 0X 3 + 3X 2 + 0X + 0 4X 3 + 0X 2 + 2X + 1 4X 3 + 0X 2 + 2X + 0 1 Hence 6X 4 − 6X 3 + 3X 2 − 3X + 1 ≡ (3X 2 + 2X)(2X 2 + 1) + 1 (mod 5). Solution. Also. q(X) = the quotient.28. d(X) ∈ k[X] and assume that d(X) = 0 so that deg d(X) > 0. An important consequence of Theorem 1. 3X 2 d(X) = the divisor .
Theorem. u(X)). p(X) is an irreducible by (iii). it follows that if k[X]/(p(X)) is a ﬁeld then because it is an integral domain. 1. g(X). So suppose that p(X) is irreducible and that p(X)  u(X)v(X) for u(X). gcd(p(X).29. so we must have r(X) = 0.e. If r(X) = 0. Let f (X). g(X)) of f (X). Then by Corollary 1. q(X)) = 1 since p(X) is irreducible. 1. This shows that p(X)  u(X) or p(X)  v(X). Proof. Here the greatest common divisor gcd(f (X). . POLYNOMIAL RINGS 9 1. Let k be a ﬁeld and X an indeterminate. b(X) ∈ k[X] such that a(X)f (X) + b(X)g(X) = gcd(f (X). there are a(X). r(X) ∈ k[X] such that p(X) = q(X)h(X) + r(X) and deg r(X) < d or r(X) = 0. Proof. q(X)). Let k be a ﬁeld and X an indeterminate. Then for any q(X) ∈ k[X] with q(X) ∈ (p(X)). Let k be a ﬁeld and X an indeterminate. (iii) This follows from Proposition 1. b(X) ∈ k[X] such that a(X)p(X) + b(X)u(X) = gcd(p(X).29 we can ﬁnd suitable a(X). (iv) Since k[X] is an integral domain and not a ﬁeld. Then a nonconstant polynomial p(X) ∈ k[X] is an irreducible if and only if it is a prime.2. Thus I ⊆ (p(X)) ⊆ I and therefore I = (p(X)). Proposition. b(X) ∈ k[X] for which / a(X)p(X) + b(X)q(X) = gcd(p(X). we also have r(X) = p(X) − q(X)h(X) ∈ I. By Lemma 1. this would contradict the minimality of d. But since p(X) is irreducible. This shows that in the quotient ring k[X]/(p(X)) the residue class of q(X) has the residue class of b(X) as its inverse. there are q(X). g(X) ∈ k[X] be nonzero. Now let p(X) ∈ I. Corollary. (iii) The quotient ring k[X]/(p(X)) is an integral domain if and only if p(X) = 0 or p(X) is irreducible in k[X]. (iv) The quotient ring k[X]/(p(X)) is a ﬁeld if and only if p(X) is an irreducible in k[X]. g(X)).. say d = deg h(X).1.6(i). (ii) The ideal (p(X)) k[X] is prime if and only if p(X) = 0 or p(X) is irreducible in k[X].15 we already know that p(X) is irreducible if it is prime. I = (h(X)) for some h(X) ∈ k[X]. (ii) This follows from Proposition 1. by Corollary 1. v(X) ∈ k[X]. By Long Division. u(X)) = p(X) or gcd(p(X). Since p(X) and h(X) are in the ideal I. But gcd(p(X).29. and so p(X) is prime. Then there must be at least one element of I with positive degree and so we can choose h(X) ∈ I of minimal degree.30. a(X)p(X) + b(X)u(X) = 1. u(X)) = 1. g(X) is the monic polynomial of greatest degree which divides both of f (X). and multiplying through by v(X) gives a(X)p(X)v(X) + b(X)u(X)v(X) = v(X) and so p(X)  v(X). so a(X)p(X) + b(X)q(X) = 1. (i) Every ideal I k[X] is principal. In the latter case. showing that p(X) = q(X)h(X).30. i.31. Then there are a(X). (i) Let I k[X] and assume that I = (0). Suppose that p(X) is irreducible (and hence is nonzero).
and p1 (X). . . ud is unique apart from the order. we have for some pd = 0. ud ∈ k.33. p(X) = p0 + p1 X + · · · + pd X d = pd q(X). Then by Proposition 1. . d d d This easily implies that as ideals of k[X]. . . bj ∈ k and b0 + b1 X + · · · + bn X n = 0. mi ). or f (X) = f1 (X)f2 (X) with both factors of positive degree. . If deg f (X) > 1. then the ﬁeld of fractions of k[X] is the ﬁeld of rational functions. . . . 1. qj (X) ∈ k[X] are irreducible monic polynomials. INTEGRAL DOMAINS. 1. then f (X) = c(X − v1 )m1 · · · (X − vr )mr . vr are the distinct roots. c is unique and the sequence of polynomials p1 (X). Proposition (Unique Factorization Property). where mi > 0 and this factorization is unique apart from the order of the pairs (vi . FIELDS AND POLYNOMIAL RINGS 1. and therefore deg fj (X) < deg f (X). where c ∈ k. . 1. The number of distinct roots of a nonconstant polynomial f (X) ∈ k[X] is at most deg f (X). By reordering we can assume that pi (X) = qi (X) and k . where u1 . Remark. . .32. The elements of k(X) are fractions of the form a0 + a1 X + · · · + am X m b0 + b1 X + · · · + bn X n with ai . then provided d = deg p(X) > 0. . To prove uniqueness. Corollary. (Sketch) Existence is proved by induction on the degree of f (X) and begins with the obvious case deg f (X) = 1. Then the sequence of roots u1 . and so we see that k = . . then either f (X) is already irreducible.31(i). Moreover. . suppose that p1 (X) · · · pk (X) = q1 (X) · · · q (X) where pi (X). k(X). each pi (X) is prime. . In connection with Theorem 1. This gives the inductive step. If k is a ﬁeld and X an indeterminate. pk (X) ∈ k[X] are irreducible monic polynomials. and this monic polynomial is unique. (p(X)) = (q(X)). pk (X) is unique apart from the order of the terms.34. In particular. Suppose that f (X) ∈ k[X] factors into linear factors f (X) = c(X − u1 ) · · · (X − ud ).35. Definition. . if v1 .10 1. hence must equal it. . notice that if p(X) ∈ k[X]. where q(X) = p−1 p0 + p−1 p1 X + · · · + p−1 pd−1 X d−1 + X d . . After cancelling common factors we obtain qk+1 (X) · · · q (X) = 1.36. Corollary. hence divides one of the qj (X). Proof. . Every nonconstant polynomial f (x) ∈ k[X] has a factorization f (x) = cp1 (X) · · · pk (X). 1. . So we can always ﬁnd a monic polynomial as the generator of a given ideal.30.
Φp (X)(X − 1) = (1 + X + · · · + X p−1 )(X − 1) = Xp − 1 = (1 + (X − 1))p − 1 p = k=1 p (X − 1)k k (mod p). .3. . Let f (X) ∈ Z[X] and s ∈ Z. .4) = p! k! (p − k)! (mod p) = p ≡ 0 (mod p2 ). If f (X) ∈ Z[X] then we can also consider f (X) as an element of Q[X]. Φp (X) = (X − 1)p−1 + cp−2 (X − 1)p−2 + · · · + c1 (X − 1) + c0 with cr ≡ 0 (mod p) and c0 = p. If R = Z or Q. 1. Choose ai ∈ Z so that f (X) = a0 + a1 (X − s) + · · · + ad−1 (X − s)d−1 + ad (X − s)d . Then f (X) is irreducible in Q[X] and hence also in Z[X]. Our next result is a special case of the Eisenstein Irreducibility Test. Let us consider factorisation of polynomials over Q. d − 1. h(X) ∈ R[X] with deg g(X) > 0 and deg h(X) > 0. Proposition (Eisenstein Test). So the Eisenstein Test can be applied here with s = 1 to show that Φp (X) is irreducible in Z[X]. Example. 1. we will need some eﬀective methods for deciding when a polynomial in k[X] is irreducible. . . Let p 2 be a prime.3. IDENTIFYING IRREDUCIBLE POLYNOMIALS 11 1.39.2a). . So to ﬁnd factors of f (X) it is suﬃcient to look for factors in Z[X]. • a0 ≡ 0 (mod p2 ). Working in Z[X].37. p divides p k when k = 1. Proof. ≡ (X − 1)p since by (1. Then f (X) has a proper factorisation over Z if and only it has a proper factorisation over Q. Let f (X) ∈ Z[X]. where d = deg f (X). Proposition (Gauss’s Lemma). . p − 1.1. . p 1 giving (1. we say that f (X) has a proper factorisation over R if f (X) = g(X)h(X) for some g(X). 1. The version here is slightly more general than the more usual one which corresponds to taking s = 0. • ad ≡ 0 (mod p). Suppose that p > 0 is a prime for which the following three conditions hold: • ak ≡ 0 (mod p) for k = 0. Then the polynomial Φp (X) = 1 + X + · · · + X p−1 ∈ Z[X] is irreducible in Q[X] and hence also in Z[X]. Hence Φp (X) ≡ (X − 1)p−1 Also. Identifying irreducible polynomials When k is a ﬁeld. .38.
If we know Φk (X) for k < n. X 5 − 1 = (X − 1)(X 4 + X 3 + X + 1) = Φ1 (X)Φ5 (X). In particular.41. The degree of Φn (X) involves a function of n probably familiar from elementary Number Theory. . Φ7 (X) = 1 + X + X 2 + X 3 + X 4 + X 5 + X 6 . when gcd(m. X 2 − 1 = (X − 1)(X + 1) = Φ1 (X)Φ2 (X). Cyclotomic polynomials can be computed recursively using Equation (1. Notice that we can inductively determine ϕ(n) using this equation. n These are examples of the cyclotomic polynomials Φn (X) ∈ Z[X] which are deﬁned for all 1 by Xn − 1 = dn (1. Φ11 (X) = 1 + X + X 2 + X 3 + X 4 + X 5 + X 6 + X 7 + X 8 + X 9 + X 10 .5a) Φd (X). then ϕ(pq) = pq − (ϕ(p) + ϕ(q) + ϕ(1)) = pq − (p − 1) − (q − 1) − 1 = (p − 1)(q − 1). For example.6) dn ϕ(d) = n. where the product is taken over all the positive divisors of n. . X 6 − 1 = (X − 1)(X + 1)(X 2 + X + 1)(X 2 − X + 1) = Φ1 (X)Φ2 (X)Φ3 (X)Φ6 (X)..e. It can be shown that for each natural number n. X 3 − 1 = (X − 1)(X 2 + X + 1) = Φ1 (X)Φ3 (X). Definition. X 4 − 1 = (X − 1)(X + 1)(X 2 + 1) = Φ1 (X)Φ2 (X)Φ4 (X). . Example. (1. n are coprime. Φ3 (X) = 1 + X + X 2 . n) = 1. if p 2 is a prime then ϕ(p) = p − 1. . INTEGRAL DOMAINS. As examples we have the irreducible polynomials Φ2 (X) = 1 + X.40.12 1. FIELDS AND POLYNOMIAL RINGS 1.7) ϕ(mn) = ϕ(m)ϕ(n).5a). k) = 1 = (Z/n)×  = number of units in Z/n = number of generators of the cyclic group Z/n. Φ5 (X) = 1 + X + X 2 + X 3 + X 4 . . Of course. The Euler function ϕ : N −→ N is deﬁned by ϕ(n) = number of k = 1. 1. then (1. i. It is also true that whenever m. if p and q are distinct primes. X 12 − 1 = (X − 1)(X + 1)(X 2 + X + 1)(X 2 + 1)(X 2 − X + 1)(X 4 − X 2 + 1) = Φ1 (X)Φ2 (X)Φ3 (X)Φ4 (X)Φ6 (X)Φ12 (X).5b) Φn (X) = Xn − 1 Φd (X) dn d<n . (1. For example. n for which gcd(n. ϕ(1) = 1.
then (1. Then every k complex nth root of unity has the form ζn = e2πik/n for k = 0.6) for ϕ. We think of ζn = e2πi/n as the standard complex primitive nth root of unity.3.43. Furthermore. Proposition. Lemma. We will give a reformulation and proof of this in Theorem 6. t (X − ζn ). ±1) o and M¨bius inversion.. if p. Let G be a ﬁnite group satisfying the following condition: • For each n 1. the number of such elements of C is ϕ(n).8) ϕ(n) = ϕ(pr1 ) · · · ϕ(prs ). Φn (X) = t=1. 2 2 It is also worth recording a related general result on cyclic groups.46. the latter can be used to solve Equation (1. Let n 1 and C = g be a cyclic group of order n and a generator g. the cyclotomic polynomial Φn (X) is irreducible in Q[X] and hence in Z[X].2.45. .9) Notice that as a result. Φpq (X) = (X pq − 1)µ(1) (X pq/p − 1)µ(p) (X pq/q − 1)µ(q) (X pq/pq − 1)µ(pq) = (X pq − 1)(X q − 1)−1 (X p − 1)−1 (X − 1) = (X pq − 1)(X − 1) . 1. Example. giving o n (1. the formulae of (1.. 2πi/6 =e πi/3 √ 1 3 = + i. n) = 1. 1. n − 1. The complex roots of Φn (X) are the primitive nth roots of unity. then using standard properties of µ. then s 1 (1. and the number of these is deg Φn (X) = ϕ(n).n−1 gcd(t.1. 1.5) lead to (1. 1.. For those who know about the M¨bius function µ (which takes values 0. IDENTIFYING IRREDUCIBLE POLYNOMIALS 13 Thus if n = pr1 · · · prs where p1 < p2 < · · · < ps are the prime factors of n and rj > 0. ϕ(n) is even when n > 2. 1. (X q − 1)(X p − 1) Recall that an element ζ of a ﬁeld K is a primitive nth root of unity if min{k : 1 k and ζ k = 1} = n. Remark. 1. s 1 ϕ(pr ) = (p − 1)pr−1 . . d dn Similarly. Hence.10) ϕ(n) = µ(d) . ..11) Φn (X) = dn (X n/d − 1)µ(d) . For n = 6 we have ζ6 = e Then ϕ(6) = 2 and 5 Φ6 (X) = X 2 − X + 1 = (X − ζ6 )(X − ζ6 ). if p is a prime and r > 0. n) = 1). For each n 1. k ζn = e2πik/n (0 k n − 1. Theorem.n)=1 Proof. . Then an element g r ∈ C is a generator if and only if gcd(r.42.44. . q are distinct primes. So for example. This leads to a useful group theoretic result. there are at most n solutions of xn = ι in G. gcd(k.
Then f (X) has a proper factorisation over k[T ] if and only it has a proper factorisation over k(T ). Suppose that p(T ) ∈ k[T ] is an irreducible for which the following three conditions hold: • ak ≡ 0 (mod p(T )) for k = 0. for each element g ∈ G of order d. INTEGRAL DOMAINS. Let θG (d) denote the number of elements in G of order d. Choose ai ∈ k[T ] so that f (X) = a0 + a1 (X − s) + · · · + ad−1 (X − s)d−1 + ad (X − s)d . we would have a strict inequality in place of Equation (1. Let k be a ﬁeld.12). dG θG (d). Proposition (Eisenstein Test). As there are at most d such elements g in G. • ad ≡ 0 (mod p(T )). . 1. .49. A polynomial f (X) ∈ k[T ][X] is an element of k(T )[X]. By Lagrange’s Theorem. the cyclic subgroup g G has ϕ(d) generators. Now if θG (d) < ϕ(d) for some d. where k is a ﬁeld. Let f (X) ∈ k[T ][X]. hence G = dG ϕ(d).45. h(X) ∈ R[X] with deg g(X) > 0 and deg h(X) > 0. each of order d. Combining these we obtain (1.47. 1. hence there must be an element of order G. In particular. this gives θG (d) ϕ(d). we say that f (X) has a proper factorisation over R if f (X) = g(X)h(X) for some g(X). where d = deg f (X).12) dG θG (d) = dG ϕ(d). . . Since G= we have G = dG {g ∈ G : g = d}. Then the polynomial X n − T is irreducible in k(T )[X]. again we state a version which is slightly more general than the usual one which corresponds to the case where s = 0. By Proposition 1. Here is another version of the Eisenstein Test.6). Let d be a divisor of G. So θG (d) dG dG ϕ(d). Proof. FIELDS AND POLYNOMIAL RINGS Then G is cyclic and in particular is abelian. θG (d) = 0 unless d divides G. Let f (X) ∈ k[T ][X] and s ∈ k[T ].48. d − 1. Hence θG (d) = ϕ(d) for all d. Then f (X) is irreducible in k(T )[X] and hence also in k[T ][X]. so G is cyclic.14 1. 1. The above results for polynomials over Q and Z have analogues over the ﬁeld of fractions k(T ) and polynomial ring k[T ]. there are ϕ(G) elements of order G. Recall the Euler ϕfunction satisﬁes Equation (1. Proposition (Gauss’s Lemma). . Example. If R = k[T ] or k(T ). • a0 ≡ 0 (mod p(T )2 ).
i.13) If we introduce u ∈ C for which x=u− then u− and so u3 − hence u6 + qu3 − Solving for u3 we obtain q 1 u3 = − ± 2 2 where equation q2 + q2 + 4p3 .4.1.. 27u3 p 3u 3 x3 + px + q = 0. (1. o A monic cubic f (X) = X 3 + a2 X 2 + a1 X + a0 ∈ C[X] can be transformed into one with no quadratic term by a change of variables X −→ X − a2 /3 giving 1 a1 a2 2a3 g(X) = f (X − a2 /3) = X 3 − a1 − a2 X + a0 − + 2 2 3 3 27 ∈ C[X]. For monic linear (degree 1) or quadratic (degree 2) polynomials.e. so we may as well assume that we want to ﬁnd the complex roots of f (X) = X 3 + pX + q ∈ C[X]. 3u p +q =0 3u +p u− 4p3 denotes one of the complex square roots of the discriminant of the quadratic 27 U 2 + qU − p3 = 0. Clearly ﬁnding the roots of f (X) is equivalent to ﬁnding those of g(X). 27 p3 + q = 0. Finding roots of complex polynomials of small degree ♠ ♥ ♦ ♣ In this section we work within the complex numbers and take k ⊆ C. FINDING ROOTS OF COMPLEX POLYNOMIALS OF SMALL DEGREE 15 1.4. 27 p3 = 0. 3]. The following 16th century method of ﬁnding roots of cubics is due to Jerˆme Cardan who seems to have obtained some preliminary versions o from Niccol` Tartaglia by somewhat disreputable means! For historical details see [2. methods of ﬁnding roots are very familiar. 27 Now if we take u to be a cube root of one of the complex numbers q 1 − ± 2 2 q2 + 4p3 27 . Suppose that x ∈ C is a root of f (X). Let us consider the cases of cubic (degree 3) and quartic (degree 4) polynomials. In practice we will usually have k = R or k = C. Cubic polynomials: Cardan’s method. p .
Here we have x = u − 1/u. we also have √ −1 √ . 4p3 q2 + 27 q 1 − + 2 2 1. 2 where ω = e2πi/3 is a primitive cube root of 1. Notice that we have a choice of 2 values for u3 and for each of these a choice of 3 values for u.14) or more precisely as (1. √ 5 + 26 Quartic polynomials: Ferrari’s method. 3 16 256 3 4 . However.15) x= 3 x= 3 q 1 − + 2 2 q2 + 4p3 + 27 3 q 1 − − 2 2 q2 + 4p3 27 q 1 − + 2 2 q2 + 4p3 − 27 3 p 3 . namely 3 √ 5 + 26 − 1 3 3 5+ √ = 26 5+ 3 2 √ 26 − 1 . q − (q 2 + 4p3 /27) 4p3 4p3 2+ −q + q 27 it is easy to verify that there are in fact only 3 choices of the root x which we can write symbolically as (1. diﬀering by factors of the form ω r for r = 0. INTEGRAL DOMAINS. A general monic quartic polynomial f (X) = X 4 + a3 X 3 + a2 X 2 + a1 X + a0 ∈ C[X] can be transformed into one with no cubic term by a change of variables X −→ X − a2 /3 giving g(X) = f (X − a3 /4) = 3 Y 4 + a2 − a2 Y 2 + 8 3 1 3 1 a − a2 a3 + a1 Y − 8 3 2 1 3 4 1 a2 a2 − a + a1 a3 + a0 . Find the complex roots of the polynomial f (X) = X 3 + 3X − 10 ∈ R[X].50. we reduce to the quadratic equation U 2 − 10U − 1 = 0 √ √ √ whose roots are 5 ± 26 ∈ R. 1. 2). Example. Solution. Applying the method above. 1.16 1. 5 − 26 = 5 + 26 √ Now 5 + 26 has the complex cube roots √ √ √ 3 3 3 5 + 26. Notice that one of these is real. 5 + 26 ω 2 . since 4p3 4p3 −q − q 2 + −q − q 2 + 27 1 27 = −27 = 2 . 5 + 26 ω. Notice that 5 + 26 > 0 and 5 − 26 < 0. FIELDS AND POLYNOMIAL RINGS we obtain the desired root of f (X) as x = u − p/3u. so the 3 complex roots of f (X) are 3 5+ √ 26 − 3 √ 5 + 26 1 ωr (r = 0. The following method of ﬁnding roots of quartics was publicised by Cardan who attributed it to his student Lodovicio Ferrari.
Since Equation (1. Hence Aut(R) forms a group under composition. Proposition. The identity function id = idR : R −→ R is an automorphism. Let R be one of the core rings Z or Z/n with n > 1. Proposition.17) Notice that if y = p/2 then we would require q = 0 and then . 1. f (X) = X 4 + pX 2 + r = (X 2 )2 + p(X 2 ) + r = 0 can be solved by solving Z 2 + pZ + r = 0.5.16) we have (x2 + t)2 = (Ax + B)2 . Then This can be done by requiring the vanishing of the discriminant (1. 1. • An automorphism of R is a ring isomorphism α : R −→ R. Suppose that x is a root and introduce numbers y. Later we will see that this is not true in general for polynomials of degree at least 5.16) (2y − p)x2 − qx + (y 2 − r) = (Ax + B)2 . 1. Aut(R) and AutR0 (R) form groups under composition of functions. z such that z = x2 + y (we will ﬁx the values of these later).54.51. Now choose y to make the last quadratic expression in x a square. q 2 − 4(2y − p)(y 2 − r) = 0. The composition α◦β of two automorphisms α.17) is a cubic in y.53. Consequently such polynomials are said to be solvable by radicals. so we may as well assume that we want to ﬁnd the complex roots of f (X) = X 4 + pX 2 + qX + r ∈ C[X]. Definition. we can use the method of solution of cubics to ﬁnd a root y = t say. Then for Equation (1. For a ring R with a subring R0 ⊆ R.5. this is one of the great early successes of this theory. Automorphisms of rings and ﬁelds 1. Proof. The argument for AutR0 (R) is similar. (1. Thus taking the two square roots of the right hand side we obtain 4 values for x. AUTOMORPHISMS OF RINGS AND FIELDS 17 Clearly ﬁnding the roots of f (X) is equivalent to ﬁnding those of g(X). The set of all automorphisms of R over R0 is denoted AutR0 (R). In the case of cubic and quartic polynomials over C we can obtain all the roots by repeatedly taking square or cube roots (or radicals). Remark.1. which we write symbolically as x = ± −t ± (Ax + B). Let R be a ring and R0 ⊆ R a subring. whence x2 = −t ± (Ax + B). 1. β : R −→ R is also an automorphism of R as is the inverse of α. • An automorphism of R over R0 is a ring isomorphism α : R −→ R for which α(r) = r whenever r ∈ R0 . Then z 2 = x4 + 2x2 y + y 2 = −px2 − qx − r + 2x2 y + y 2 = (2y − p)x2 − qx + y 2 − r. The set of all such automorphisms is denoted Aut(R).52.
(ii) If S is a ring containing a core ring R and α ∈ Aut(S). The inverse automorphism α−1 : D −→ D also gives rise to an induced homomorphism (α−1 )∗ : Fr(D) −→ Fr(D).56 need not be an epimorphism. Recalling Deﬁnition 1. α −→ α∗ . then every automorphism of R restricts to the identity on k. (i) For such a core ring R. (ii) For α ∈ Aut(S).. FIELDS AND POLYNOMIAL RINGS (i) The only automorphism of R is the identity. The homomorphism ( )∗ : Aut(Z) −→ Aut(Q). i. −k 0 if k = 0 =k1. There is a monomorphism of groups ( )∗ : Aut(D) −→ Aut(Fr(D)). 1. Let D be an integral domain and α : D −→ D be an automorphism. then α restricts to the identity on R. Let k be one of the prime ﬁelds Q or Fp with p > 0 prime. Hence (α)∗ is invertible with inverse (α−1 )∗ . −k α(0) if k = 0 1 + · · · + 1 if k > 0. k α(k1) = −(α(1) + · · · + α(1)) if k < 0..58. we have an example which shows that the monomorphism of Corollary 1. 1.36. α(1) + · · · + α(1) if k > 0.e.57. Proof. INTEGRAL DOMAINS. Since α−1 ◦ α = id = α ◦ α−1 .20 to show that (α−1 )∗ ◦ (α)∗ = id = (α)∗ ◦ (α−1 )∗ . For an automorphism α of R. α(r) = r for all r ∈ R. Aut(R) = Autk (R). Proof. Aut(R) = {id}. k = −(1 + · · · + 1) if k < 0. Proposition. we can apply Corollary 1. i.54(ii) we obtain another useful result. Then the induced homomorphism gives an automorphism α∗ : Fr(D) −→ Fr(D). Hence. Example.e. every element has the form k1 for some k ∈ Z.18 1. 1. Proposition.56.55. α −→ α∗ . If R is a ring containing k as a subring. the induced homomorphism α∗ : Fr(D) −→ Fr(D) exists and we need to show it has an inverse. 1. Corollary. The ﬁeld of fractions of the ring of integers Z is the ﬁeld of rationals Q. Thus α = id. Combining this example with Proposition 1. α(1) = 1 and a similar argument to that for (i) shows that α(r) = r for all r ∈ R. i. Given α. Aut(S) = AutR (S).e.. Here we take D = Q[X] and Fr(Q[X]) = Q(X). is an isomorphism and hence Aut(Q) = {id}.
Using the inverse matrix a−1 −a−1 b A−1 = 0 1 we similarly obtain a homomorphism αA−1 : k[X] −→ k[X] for which αA−1 (X) = a−1 X − a−1 b. contains a subgroup isomorphic to Aﬀ 1 (k). Let k be a ﬁeld and X an indeterminate. (note the order!) hence there is a homomorphism of groups Aﬀ 1 (k) −→ Autk (k[X]).59. Using the same line of argument as in the proof of Proposition 1. 1. Then (i) Autk (k(X)) contains a subgroup isomorphic to PGL2 (k). In fact. this is also an epimorphism and we leave the proof of this as an exercise. Let k be a ﬁeld and X an indeterminate. A −→ αA−1 . For this we use the element aX + b ∈ k[X] together with the extension result of Theorem 1. AUTOMORPHISMS OF RINGS AND FIELDS 19 1. 1.61. A2 ∈ Aﬀ 1 (k).22(i) to obtain a homomorphism αA : k[X] −→ k[X] with αA (X) = aX + b. Composing with ( )∗ we see that there is a monomorphism Aﬀ 1 (k) −→ Autk (k(X)). α −→ α∗ is a monomorphism but it is not an epimorphism since there is an automorphism γ(f (X)) = f (1/X) which sends X ∈ Q[X] ⊆ Q(X) to 1/X ∈ Q[X] and so does not restrict to an automorphism of / Q[X].5. PGL2 (k) = GL2 (k)/ Scal2 (k).60. which is easily seen to be a monomorphism. Then Autk (k[X]) and hence Autk (k(X)). Notice that GL2 (k) has another interesting subgroup called the aﬃne subgroup. a11 a22 − a12 a21 = 0 GL2 (k) = a21 a22 The scalar matrices form a normal subgroup Scal2 (k) = {diag(t. Autk (k(X)) ∼ PGL2 (k). Example. t) : t ∈ k. The group of invertible 2 × 2 matrices over k is the 2 × 2 general linear group over k. It is straightforward to check that for A1 . Example. γ : Q(X) −→ Q(X). = Proof. The homomorphism ( )∗ : Aut(Q[X]) −→ Aut(Q(X)). (ii) In fact. Let k be a ﬁeld. Autk (k[X]) ∼ Aﬀ 1 (k). a = 0 0 1 GL2 (k). αA2 A1 = αA1 ◦ αA2 . We begin by showing that to each aﬃne matrix A= a b ∈ Aﬀ 1 (k) 0 1 there is an associated automorphism αA : k[X] −→ k[X]. In fact. = .1. Aﬀ 1 (k) = a b : a. Example. t = 0} GL2 (k). b ∈ k. The quotient group is called the 2 × 2 projective general linear group over k. a11 a12 : aij ∈ k.55 (or doing a direct calculation) we see that αA−1 is the inverse of αA an so αA ∈ Autk (k[X]).
There is one case where Scal2 (k) is the trivial group. t). Therefore passing to the quotient PGL2 (k) = GL2 (k)/ Scal2 (k) we obtain a monomorphism PGL2 (k) −→ Autk (k(X)). deg q(X)}. q(X) where p0 (X). But this can only happen if the function f is injective which means that all of these numbers must be 1. Finally. this is not an injection in general since for each scalar matrix diag(t.t) (X) = tX = X. Let α ∈ AutC (C(X)). (i) We begin by showing that to each invertible matrix A= a11 a12 ∈ GL2 (k) a21 a22 there is an associated automorphism αA : k(X) −→ k(X). We begin by choosing the element (a11 X + a12 )/(a21 X + a22 ) ∈ k(X) and then using Theorem 1. if deg p(X) > deg q(X) then we can write p0 (X) f (X) = p1 (X) + . but an easy argument then shows the general case. Then the number of poles of f counted with algebraic multiplicity is p0 deg p1 (X) + ord . INTEGRAL DOMAINS. t −1 a11 X + a12 . FIELDS AND POLYNOMIAL RINGS Proof. Also. In fact it is easy to see that Scal2 (k) GL2 (k) is the kernel of this homomorphism. If we write f (X) = p(X) q(X) where p(X). namely k = F2 .20 1. Now let c ∈ C. actually this argument can be modiﬁed to work for any algebraically closed ﬁeld. if deg p(X) deg q(X) then the number of poles of f counted with algebraic multiplicity is also ord f . We give a sketch proof for the case of k = C. where α(X) = f (X). αdiag(t. There is an associated rational (hence meromorphic) function f given by z −→ f (z). p1 (X) ∈ C[X] and deg p0 (X) < deg q(X). q(X) ∈ C[X] have no common factors of positive degree. then the order of f (X) is ord f = max{deg p(X). c d . Then the number of solutions counted with algebraic multiplicity of the equation f (z) = c turns out to be ord f . Thus aX + b = constant f (X) = cX + d a b and the matrix must be invertible. deﬁned on C with the poles of f deleted. (ii) To show that every automorphism of k(X) is a fractional linear transformation is less elementary.22(i) to obtain a homomorphism k[X] −→ k(X) that sends X to (a11 X+a12 )/(a21 X+a22 ). q Now it is easy to see that since α is invertible so is the function f . However. There is an associated homomorphism of groups GL2 (k) −→ Autk (k(X)) sending A to αA .t) is the identity function. hence ord f = 1. a21 X + a22 showing that αdiag(t. By applying ( )∗ to this we obtain a homomorphism (known as a fractional linear transformation) αA : k(X) −→ k(X) for which αA (X) = Again we ﬁnd that αA2 A1 = αA1 ◦ αA2 .
Now we turn to a more familiar ﬁeld R. AutR (C) = {id. 1. 1. Proposition.62. Show that {n ∈ Z : n > 0 and n1 = 0} = {n ∈ Z : n > 0 and nr = 0 for all r ∈ R}. Proof. First recall that for x. Deduce that if char R > 0 then these sets are nonempty and char R = min{n ∈ Z : n > 0 and nr = 0 for all r ∈ R}. we have α(s2 ) = α(s)2 . so there are many elements of order 2 in the group Aut(C). Now for α ∈ Aut(R) and s ∈ R. This shows that α is continuous at x. i. We recall from Analysis that the rational numbers are dense in the real numbers in the sense that each r ∈ R can be expressed as a limit r = limn→∞ qn . in fact.63. So α preserves order and ﬁxes rational numbers. Then we can choose a rational number q such that 0 < q Taking δ = q we ﬁnd that for y ∈ R with y − x < δ (i. where qn ∈ Q. Note that given an automorphism α ∈ Aut(C). n→∞ n→∞ We will show that an automorphism α ∈ Aut(R) is continuous.57. Hence. y ∈ R.. −δ < y − x < δ) we have −δ = α(−δ) < α(y) − α(x) < α(δ) = δ. What is the relationship between char S and char R ? . (a) Show that every subring S ⊆ R is also an integral domain. the composition α ◦ ( ) ◦ α−1 is also self inverse. hence Aut(R) = {id}.2.e. hence α(y) − α(x) < δ ε. x < y =⇒ α(y) − α(x) = α(t)2 for some nonzero t ∈ R =⇒ α(x) < α(y).21 Clearly not every fractional linear transformation αA : k(X) −→ k(X) maps polynomials to polynomials so ( )∗ : Autk (k[X]) −→ Autk (k(X)) is not an epimorphism.. If we try to determine Aut(C) the answer turns out to be much more complicated. It is easy to see that complex conjugation ( ) : C −→ C is an automorphism of C and ﬁxes every real number. ( ) ∈ AutR (C). However. The only automorphism of the ﬁeld R is the identity function. 1. Now let x ∈ R and ε > 0. hence it must be the identity function. Let R be a ring. ( )}. Remark. it is not true that every α ∈ Aut(C) ﬁxes every real number! The automorphism group Aut(C) is actually enormous but it is hard to ﬁnd an explicit element other than id and ( ).1. Let R be an integral domain. Then for a continuous function f : R −→ R. its value at r depends on its values on Q since f (r) = f ( lim qn ) = lim f (qn ). First we note that Q ⊆ R is a subring and if α ∈ Aut(R) then α(q) = q for q ∈ Q by Example 1.e. Exercises on Chapter 1 1. Thus every automorphism of R is continuous function which ﬁxes all the rational numbers. ε. the real numbers. x<y ⇐⇒ 0<y−x ⇐⇒ y − x = t2 for some nonzero t ∈ R.
show that there is a unique ring homomorphism ψa. are the results in parts (a) and (b) still true? [Hint: look for a counterexample. a11 . Let k be a ﬁeld. ij R = Matn (A) = . ann 1.7. If R is a commutative ring with unit containing the prime ﬁeld Fp for some prime p > 0. Also show that Z(i) is an integral domain. . (c) The ring of n × n matrices over A. 1. suppose that Q S is a prime ideal for which P ⊆ Q. (c) Show that inc∗ is an isomorphism. 1. Consider the sets Z(i) = {u + vi : u.8. Q(i) = {u + vi : u. d ∈ R.3. an1 . (a) If a. determine ψa.. 1. deﬁnes a ring homomorphism. (a) Show that Z(i) and Q(i) are subrings of C. Let k be a ﬁeld and k[[X]] be the set consisting of all power series ∞ ak X k = a0 + a1 X + · · · + ak X k + · · · . Show that ϕ is a monomorphism.b (X) = aX + b. so Fr(Z(i)) = Q(i).5. Prove the following. . v ∈ Z} ⊆ C. Let R and S be rings with unity and Q S a prime ideal. show that ϕ−1 Q = {r ∈ R : ϕ(r) ∈ Q} ⊆ R is a prime ideal of R. . the deg ψa. ﬁnd the characteristic char R and the characteristic subring of R.] (d) If R ⊆ S is a subring and P R is a maximal ideal.b (f (X)) = deg f (X). . :a ∈A . Give examples to show that ϕ need not be surjective or injective. . (b) The polynomial ring R = A[X]. Show that Q ∩ R = P . show that Q ∩ R is a prime ideal of R. (i) If f (X) ∈ k[X].d . (b) Show that the inclusion homomorphism inc : Z(i) −→ Q(i) extends to a monomorphism inc∗ : Fr(Z(i)) −→ Q(i). b ∈ k with a = 0. (a) If ϕ : R −→ S is a ring homomorphism. b ∈ R. For each of the following rings R.22 1.6. (b) If R ⊆ S is a subring. 1.b (r) = r if r ∈ R and ψa. INTEGRAL DOMAINS. (c) If the word ‘prime’ is replaced by ‘maximal’ throughout.b is an isomorphism and ﬁnd its inverse.9. . 1. . (ii) If p(X) ∈ k[X] is a prime then so is ψa. (iii) If p(X) ∈ k[X] is an irreducible then so is ψa. 1. If c.b : R[X] −→ R[X] for which ψa. . Let R be a commutative ring. (a) Any subring R ⊆ C. (b) Now suppose that R = k is a ﬁeld and a. v ∈ Q} ⊆ C. k=0 . give an example to show that a subring of R need not be a ﬁeld. If a is a unit. A is an arbitrary commutative ring.b ◦ ψc. Q(i) is a ﬁeld and Z(i) is a subring of Q(i). show that the function ϕ : R −→ R given by ϕ(t) = tp . . a1n . . FIELDS AND POLYNOMIAL RINGS (b) If R is a ﬁeld.b (p(X)).b (p(X)). Determine which of these rings is an integral domain. In (b) and (c). show that ψa.4. R be a ring with unit and let ϕ : k −→ R be a ring homomorphism.
18.16. 1. Determine the image εα Q[X] ⊆ C and the ideal ker εα Q[X]. 3 Find rational cubic polynomials f (X) and g(X) for which f (α) = 0 = g(β). 1. Hence determine these real numbers. ﬁnd the quotient and remainder when performing long division of f (X) = 2X 3 + 2X 2 + X + 1 by d(X) = 2X 3 + 2X. Taking k = Q. (b) Show that ∞ ak X k ∈ k[[X]] is a unit if and only if a0 = 0. k=0 (c) Show that Fr(k[[X]]) consists of all ﬁnitetailed Laurent series ∞ ak X k = a X + a k= +1 X +1 + · · · + ak X k + · · · for some ∈ Z and ak ∈ k. What is ε−√2 Q[X] ⊆ C? Find ker ε√2 Q[X] and ker ε−√2 Q[X].23 with ak ∈ k. X 3 + X − 1. 1. Determine the image εω Q[X] ⊆ C. is the latter ideal maximal? What happens if α is replaced by one of the other roots of f (X)? Repeat this problem starting with the inclusion of the real numbers into the complex numbers inc : R −→ C and εα : R[X] −→ C.17. Determine ker εω Q[X] and decide whether it is maximal.15.14. Determine the image ε√2 Q[X] ⊆ C. . Suppose that f (X) = a0 + a1 X + · · · + an X n ∈ Z[X] with p an and that f (X) ∈ Fp [X] denotes the polynomial obtained by reducing the coeﬃcients of f (X) modulo p. 1. 1.13. X 5 − X + 1. ﬁnd the quotient and remainder when performing long division of f (X) = 6X 4 − 6X 3 + 3X 2 − 3X − 2 by d(X) = 2X 3 + X + 3. Let ω = (−1 + 3i)/2 ∈ C. √ I5 = {f (X) ∈ R[X] : f ( 2 i) = 0} R[X].10. Use Cardan’s method to ﬁnd the complex roots of the polynomial f (X) = X 3 − 9X 2 + 21X − 5. Which of the following polynomials in Z[X] is irreducible? X 3 − X + 1. Find generators for each of the following ideals: I1 = {f (X) ∈ Q[X] : f (i) = 0} Q[X]. 1. are these maximal ideals? √ 1. X 3 + 2X + 1. Taking k = F3 . Find another evaluation homomorphism with the same kernel and image. 1. Consider the inclusion inc : Q −→ C and its extension to εω : Q[X] −→ C. (a) Show that this can be made into an integral domain containing k[X] as a subring by deﬁning addition and multiplication in the obvious way. Consider the inclusion inc : Q −→ C and its extension to ε√2 : Q[X] −→ C. show that f (X) is irreducible.12.11. √ I2 = {f (X) ∈ Q[X] : f ( 2 i) = 0} Q[X]. If f (X) is irreducible. √ I4 = {f (X) ∈ R[X] : f ( 2) = 0} R[X]. Consider the real numbers α= 3 10 + √ 108 + 3 10 − √ 108. β= 3 1+ 2 3 7 + 3 3 1− 2 3 7 . X 5 + X − 1. 5X 3 − 10X + X 2 − 2. Consider the inclusion inc : Q −→ C and its extension to εα : Q[X] −→ C where α is one of the 4 complex roots of the polynomial f (X) = X 4 − 2 ∈ Q[X]. I6 = {f (X) ∈ R[X] : f (ζ3 ) = 0} R[X]. √ I3 = {f (X) ∈ Q[X] : f ( 2) = 0} Q[X]. 1. Let p > 0 be a prime.
For n 2. n > 0 and p n. show that the polynomial X p − a ∈ K[X] has either no roots or exactly one root in K. . (b) Show that Φpk (X) ∈ Q[X] is irreducible. r −1 r1 −1 ···pkk ). Prove the ﬁnal part of Example 1. α3 (f (X)) = f (1/X). Autk (k(X)) isomorphic to Show that the set consisting of these elements is a subgroup Γk the symmetric group S3 . . −1 1. (a) Show that for k 1. Let k be any ﬁeld. FIELDS AND POLYNOMIAL RINGS 1. (b) If a ∈ K. 1. Let p > 0 be a prime and K be a ﬁeld with char K = p.24.24 1. Consider the 6 automorphisms αj : k(X) −→ k(X) (j = 1. Let p > 0 be a prime.60 by showing that there is an isomorphism of groups Aﬀ 1 (k) ∼ Autk (k[X]). . 6) deﬁned by α1 (f (X)) = f (X).20. α4 (f (X)) = f ((X − 1)/X). α5 (f (X)) = f (1/(1 − X)). When k = F2 . Deduce that if m.25. then every npm th root of 1 in K is an nth root of 1. .23. show that Γk ∼ GL2 (k). = 1. −1 2 +ζ −2 in terms of cos(2π/5). ζn + ζn = 2 cos(2π/n). α6 (f (X)) = f (X/(X − 1)).21. = 1. INTEGRAL DOMAINS. Show that for n 1. (c) Generalize part (a) to show that if n = pr1 · · · prk is the prime power factorization of n 1 k with the pi being distinct primes and ri > 0.19. show that X ϕ(n) Φn (X −1 ) = Φn (X). Determine the cyclotomic polynomial Φ20 (X). (a) Show that if ζ ∈ K is a pth root of 1 then ζ = 1. Hence ﬁnd a rational polynomial Find expressions for ζ5 +ζ5 and ζ5 5 which has cos(2π/5) as a root. then Φn (X) = Φp1 ···pk (X p1 1. . α2 (f (X)) = f (1 − X). the cyclotomic polynomial Φpk (X) satisﬁes Φpk (X) = Φp (X p k−1 ) and has as its complex roots the primitive pk th roots of 1. 1.22.
The inﬁniteness of R/Q and C/Q are consequences of the fact that any ﬁnite dimensional vector space over Q is countable. Definition.5. Since i ∈ R. M 2. √ Show that [Q( 2) : Q] = 2. Consider the extension Q( 2)/Q where √ √ Q( 2) = {x + y 2 : x.2. We write K L or L/K to indicate this. L is also said to be an extension (ﬁeld ) of K. A basis for the Qvector space R is known as a Hamel basis. if K = L.e. v both nonzero. An important fact about an extension of ﬁelds L/K is that L is a Kvector space whose addition is the addition in the ﬁeld L while scalar multiplication is deﬁned by u · x = ux (u ∈ K. v ∈ Q not both zero. however R and C are uncountable. these elements are also linearly independent / over R and therefore they form a basis. Definition. Given two extensions L/K and M/L. (i) If one or both of the dimensions [L : K] or [M : L] is inﬁnite then so is [M : K]. y ∈ R}. Solution. Let L/K be a subextension of M/K. 2 are linearly independent and so form a basis for Q( 2) to over Q and [Q( 2) : Q] = 2. Let K and L be ﬁelds and suppose that K ⊆ L is a subring. We will call dimK L the degree or index of the extension L/K and use the notation [L : K] = dimK L.1. y ∈ Q}.CHAPTER 2 Fields and their extensions 2. Example. x ∈ L). Hence 1. If we have two extensions L/K and M/L then it is a straightforward to verify that K and so we have another extension M/K. i.. whence [C : R] = 2. Show that the extension C/R is ﬁnite. If the elements 1. Then we say that K is a subﬁeld of L. √ 2. 2. while R/Q and C/Q are both inﬁnite. 25 . Fields and subﬁelds 2. 2 were linearly dependent we would have u + v 2 = 0 for some / u. The elements 1. An extension of ﬁelds L/K is ﬁnite (dimensional ) if [L : K] < ∞. so 1. Example.6. Definition. 2 clearly span the Qvector space Q( 2). 2. and write K < L if K is a proper subﬁeld of L.3. We have C = {x + yi : x. in fact it is easy to see that we would then also have u. √ √ 2 ∈ Q. √ √ Now recall that √ Solution. i span C as a vector space over R.1. 2. v √ √ which we know √ be false. otherwise it is inﬁnite (dimensional ). Theorem.4. Thus we would have √ u 2 = − ∈ Q. we say that L/K is a subextension of M/K and sometimes write L/K M/K.
Hence. (ii) Setting r = [L : K] and s = [M : L]. (i) If [M : K] is ﬁnite. . r i=1 tij i is in L. . Also L is a Kvector subspace of the ﬁnite dimensional Kvector space M . this r tij i=1 i = 0. j = 1. each tij = 0. tr ∈ K. ys ∈ L for which v = y1 m1 + · · · + ys ms . hence [L : K] < ∞. where t1 . Thus the elements Kvector space M . Now suppose that for some tij ∈ K we have s r tij ( i mj ) = 0. . . . . . ms of M over L. . where each coeﬃcient means that for each j. . choose a basis 1 . . . . . . By the linear independence of the mj over L. s) span the where each coeﬃcient xij is in K. Such towers of extensions are our main objects of study. . Now any element u ∈ M can be expressed as u = t1 m 1 + · · · + tr m r . mr spans M over L and so [M : L] < ∞. Proof. . r of L over K and a basis m1 . . (i = 1. . . mr of M over K. . . . We can build up sequences of extensions and form towers of arbitrary . FIELDS AND THEIR EXTENSIONS (ii) If the dimensions [L : K] and [M : L] are both ﬁnite then so is [M : K] and [M : K] = [M : L] [L : K]. Then there are elements y1 . . . Now let v ∈ M . . . By the linear independence of the i over K.26 2. s r s r 1 + · · · + xrj r v= j=1 i=1 xij i mj = j=1 i=1 i mj xij ( i mj ). choose a basis m1 . but since K ⊆ L. . r. . We will often indicate subextensions in diagrammatic form where larger ﬁelds always go above smaller ones and the information on the lines indicates dimensions MF [M :L] A 6 & L [L:K] 1 [M :K]=[M :L] [L:K] K ! We often suppress ‘composite’ lines such as the dashed one. . this means that m1 . . But each yj can be expressed in the form yj = x1j for suitable xij ∈ K. . . Hence the i mj form a basis of M over K and so [M : K] = rs = [M : L] [L : K]. j=1 i=1 On collecting terms we obtain s j=1 r tij i=1 i mj = 0.
. . ur . . ur ) = which is the smallest subﬁeld in F that contains K and the elements u1 . . . g(X1 . . . exactly one of the following conditions holds. ur ) Reordering the ui does not change K(u1 . . g(u1 . The extension K(u1 . ur .. K(u)/K is inﬁnite and u is said to be transcendental over K. . Lk /Lk−1 is a such a sequence of extensions. . . . . We can extend this to the case of an inﬁnite sequence u1 . It can be shown that (2.9. . un ) = K(u1 . . . Theorem.. . ur )/K is said to be generated by the elements u1 . More generally. . .. . Xr ]. . . . . . Proposition. .. . . . Let F be a ﬁeld and K F . v)/K(u) be simple extensions. . ur ∈ F we set L K L F u1 . ur )/K is a ﬁnitely generated extension of K. . Then K(u. .2. . ur . . . .) F the smallest extension ﬁeld of K containing all the elements ur . . . un ). . . ur . ur ) ∈ F : f (X1 . . . . if L1 /K. . . in F and denote by K(u1 . un−1 )(un ) and this is independent of the order of the sequence u1 . . ur ) = 0 .1) K(u1 . Thus. . . . ur ) = f (u1 . . . For a simple extension K(u)/K. . . . we also say that K(u1 . . Simple and ﬁnitely generated extensions 2. .7. . (ii) The evaluation at u homomorphism εu : K[X] −→ K(u) has a nontrivial kernel ker εu = (p(X)) where p(X) ∈ K[X] is an irreducible monic polynomial of positive degree and the quotient homomorphism εu : K[X]/(p(X)) −→ K(u) is an isomorphism. . Given elements u1 .8. . there is a diagram Lk Lk−1 L1 K 2.2. . . . . . In this case. .2. 2. . L2 /L1 . . . Let K(u)/K and K(u. 2. . Definition. Xr ). . . g(u1 . . . un . v) = K(u)(v) = K(v)(u). . . . An extension of the form K(u)/K is called a simple extension of K with generator u. . . . . .ur ∈L K(u1 . K(u1 . . . . . . . In this case K(u)/K is ﬁnite with [K(u) : K] = deg p(X) and u is said to be algebraic over K. (i) The evaluation at u homomorphism εu : K[X] −→ K(u) is a monomorphism and on passing to the fraction ﬁeld gives an isomorphism (εu )∗ : K(X) −→ K(u). Xr ) ∈ K[X1 . SIMPLE AND FINITELY GENERATED EXTENSIONS 27 length.
Theorem 1. we see that the elements 1. 3):Q( 2)] Q √ √ √ We will show that [Q( 2. 3)/Q we have [Q( 2. This shows that [Q( 2. while v. √3) √Q( 2)] = 2.11. But if these are linearly dependent then 3 ∈ Q( 2).4 we know that [Q( 2) : Q] = 2. √ √ √ √ √ so 1. all that needs checking is that (εu )∗ is an epimorphism. 3) √ √ √ [Q( 2. 3). We have the following tower of extensions. b ∈ Q(√2). 3) = Q( 2)( 3) then u = a + b 3 for some a. Using Long Division. 3) which are conveniently displayed in the following diagram. Example.31(iv) implies that the image of εu is a subﬁeld of K(u) and since it contains u it must equal K(u). w √ 0 Q. . . we ﬁnd that every element of K[X]/(p(X)) can be uniquely expressed as a coset of the form f (X) + (p(X)). .28 2. √ √ Q( 2. X 2 + (p(X)). (i) If ker εu = (0). . By Example 2. Writing √ √ 3=v+w 2 with v. 3 span Q( 2. FIELDS AND THEIR EXTENSIONS Proof. √ : √ √ √ Notice that if u ∈ Q( 2. Via the isomorphism εu under which εu (X k + (p(X))) = uk . Remark. 3) over Q( 2). 3) : Q] = 4. √ √ Q( 2. . So 1. 3):Q( 2)] √ Q( 2) 2 √ √ √ √ √ [Q( 2. where deg f (X) < deg p(X). 3):Q]=2[Q( 2. . Hence εu is an isomorphism. 3 are linearly√ √ independent over Q( 2) √ √ √ and√ √ therefore form a basis of Q( 2. = √ would implies that 2 ∈ Q which is false. (ii) When ker εu = (0). we ﬁnd that √ v 2 + 2w2 + 2vw 2 = 3 ∈ Q. . where d = deg p(X). √ Proof.10 can be extended to provide a useful general result whose proof is left as an exercise. There are some other subﬁelds of Q( 2. but as u is in the image of (εu )∗ this is obvious. 3) √ Q( 2) rr rrr 2 rrrr r vvv vvv v 2 vvvv v √ Q( 3) 2 vvv vvv 2 vvv 2 v r rrr rrr rr 2 rrr √ Q( 6) Q One idea in the veriﬁcation of Example 2. X + (p(X)). . For the extension Q( 2. 3) : Q( 2)] = 2 and so [Q( 2. √ √ √ √ 2. ud−1 form a basis for K(u) over K. √ and hence 2vw 2 ∈√ The possibilities v = 0 or w = 0 are easily ruled out. X d−1 + (p(X)). u. . Hence every element of K[X]/(p(X)) can be uniquely expressed as a linear combination over K of the d cosets 1 + (p(X)). √ √ 2. 3) : Q] = 4.10. w ∈ Q.
2.2. SIMPLE AND FINITELY GENERATED EXTENSIONS
29
2.12. Proposition. Let p1 , . . . , pn be a sequence of distinct primes pi > 0. Then √ √ √ pn ∈ Q( p1 , . . . , pn−1 ). / √ √ √ √ √ √ Hence [Q( p1 , . . . , pn ) : Q( p1 , . . . , pn−1 )] = 2 and [Q( p1 , . . . , pn ) : Q] = 2n . √ √ 2.13. Example. For the extension Q( 2, i)/Q we have [Q( 2, i) : Q] = 4. √ √ √ Proof. We know that [Q( 2) : Q] = 2. Also, i ∈ Q( 2) since i is not real and Q( 2) R. / √ √ √ √ Since i2 + 1 = 0, we have Q( 2, i) = Q( 2)(i) and [Q( 2, i) : Q( 2)] = 2. Using the formula √ √ √ √ [Q( 2, i) : Q] = [Q( 2, i) : Q( 2)] [Q( 2) : Q], √ we obtain [Q( 2, i) : Q] = 4. √ √ This example also has several other subﬁelds, with only Q( 2) = Q( 2, i) ∩ R being a subﬁeld of R. C
2
R
∞
Q( 2)
t tt tt t tt √ t
2
∞
√ Q( 2, i)
2
uuu uuu 2 uuu u
uuu uuu u 2 uuu u
Q(i)
2
Q
r rrr rr2 rr rrr
√ Q( 2 i)
2.14. Example. For n nth root of 2.
1, let En = Q(21/n )
R, where 21/n ∈ R denotes the positive real
(i) Show that [En : Q] = n. (ii) If m 1 with m  n, show that Em En and determine [En : Em ]. (iii) If m, n are coprime, show that Emn = Q(21/m , 21/n ). Solution. (i) Consider the evaluation homomorphism ε21/n : Q[X] −→ En . Applying the Eisenstein Test 1.38 using the prime 2 to the polynomial X n − 2 ∈ Z[X], we ﬁnd that ker ε21/n = (X n − 2) Q[X], and the induced homomorphism ε21/n : Q[X]/(X n − 2) −→ En is an isomorphism. Hence [En : Q] = n. (ii) Since n/m is an integer, 21/m = (21/n )n/m ∈ En , so Em = Q(21/m ) ⊆ En . By Theorem 2.6 we have n = [En : Q] = [En : Em ] [Em : Q] = m[En : Em ], whence [En : Em ] = n/m. (iii) By (ii) we have Em Emn and En Emn , hence Q(21/m , 21/n ) there are integers r, s for which rm + sn = 1 and so rm + sn r s 1 = = + . mn mn n m Emn . As gcd(m, n) = 1,
30
2. FIELDS AND THEIR EXTENSIONS
This shows that 21/mn = (21/n )r (21/m )s ∈ Q(21/m , 21/n ), whence Emn Q(21/m , 21/n ). Combining these inclusions we obtain Emn = Q(21/m , 21/n ).
Exercises on Chapter 2 √ √ 2.1. Let p ∈ N be an prime. Show that the extension Q( p)/Q has [Q( p) : Q] = 2. √ √ √ 2.2. Let p, q > 0 be distinct primes. Show that [Q( p, q) : Q( p)] = 2. 2.3. Prove Proposition 2.12 by induction on n. 2.4. Let K a ﬁeld with char K = 2 and suppose that L/K is an extension. If a, b ∈ K are distinct, suppose that u, v ∈ L satisfy u2 = a and v 2 = b. Show that K(u, v) = K(u + v). [Hint: ﬁrst show that u±v = 0 and deduce that u−v ∈ K(u+v); then show that u, v ∈ K(u+v).] 2.5. Show that [Q(i) : Q] = 2. √ √ 2.6. Show that [Q( 3, i) : Q] = 4. Find the three subﬁelds L Q( 3, i) with [L : Q] = 2 and display their relationship in a diagram, indicating which ones are subﬁelds of R. 2.7. Let ζ5 = e2πi/5 ∈ C. (a) Explain why [Q(ζ5 ) : Q] = 4. (b) Show that cos(2π/5), sin(2π/5) i ∈ Q(ζ5 ). (c) Show that for t ∈ R, cos 5t = 16 cos5 t − 20 cos3 t + 5 cos t. (d) Show that the numbers cos(2kπ/5) with k = 0, 1, 2, 3, 4 are roots of the polynomial f (X) = 16X 5 − 20X 3 + 5X − 1 = (X − 1)(4X 2 + 2X − 1)2 and deduce that [Q(cos(2π/5)) : Q] = 2. (e) Display the relationship between the ﬁelds Q, Q(cos(2π/5)), and Q(ζ5 ) in a suitable diagram. 2.8. This question is for those who like lots of calculation or using Maple. Let ζ7 = e2πi/7 ∈ C. (a) Explain why [Q(ζ7 ) : Q] = 6. (b) Show that cos(2π/7), sin(2π/7) i ∈ Q(ζ7 ). (c) Show cos 7t = 64 cos7 t − 112 cos5 t + 56 cos3 t − 7 cos t. Show that the numbers cos(2kπ/7) with k = 0, 1, . . . , 6 are roots of the polynomial f (X) = 64X 7 − 112X 5 + 56X 3 − 7X − 1 = (X − 1)(8X 3 + 4X 2 − 4X − 1)2 and deduce that [Q(cos(2π/7)) : Q] = 3. (d) Show that sin(2π/7) i is a root of g(X) = 64X 7 + 112X 5 + 56X 3 + 7X = X(64X 6 + 112X 4 + 56X 2 + 7) and that 64X 6 + 112X 4 + 56X 2 + 7 ∈ Q[X] is irreducible. What is [Q(sin(2π/7) i) : Q]? (e) Display the relationship between the ﬁelds Q, Q(cos(2π/7)), Q(sin(2π/7) i) and Q(ζ7 ) in a diagram. (f) Is i ∈ Q(ζ7 )? 2.9. In this question we continue to consider the situation described in Example 2.14.
31
(a) Show that if n is odd, ∼ Z/2 if n is even, {id, τn } = where τn has composition order 2. (b) Let E = En R. Show that AutQ (E) = {id}. AutQ (En ) =
n 1
{id}
(c) Display the 6 subﬁelds of E12 in a diagram. (d) Which of the subﬁelds in part (c) contain the element 21/2 + 21/3 ?
.
t (X) ∈ K[X] is an irreducible monic polynomial called the minimal polynomial of t over K. 33 .√2 (X) = [Q( 2) : Q] = 2. so t is algebraic over K.2. 3.2.6. In particular. (iii) The extension K(t)/K is ﬁnite dimensional.9(ii). 3. We will often use this without further comment. 3.1. Definition.t (X) = degK t.1.t (X). 3. where minpolyK. Proposition. Algebraic extensions Let L/K be an extension of ﬁelds.t (X)) = (0). p(X) = minpolyK. Proof.√2 (X) = X 2 − 2.4. Theorem 2. 3. √ Proof. If t ∈ L is algebraic over K then by Proposition 3. This follows from Theorem 2. when p(X) is monic. If t ∈ L is algebraic over K then [K(t) : K] = deg minpolyK. The degree of minpolyK.CHAPTER 3 Algebraic extensions of ﬁelds 3. The minimal polynomial of √ 2 ∈ C over Q is minpolyQ. From Theorem 2.3. (ii) The evaluation homomorphism εt : K[X] −→ L has nontrivial kernel.7. An element t ∈ L is algebraic over K if there is a nonzero polynomial p(X) ∈ K[X] for which p(t) = 0.t (X) is called the degree of t over K and is denoted degK t. Then minpolyK. Definition. Consider C/Q. hence minpolyQ. 3. Remark.9 allows us to characterize algebraic elements in other ways. the polynomial p(X) = X − t ∈ K[X] satisﬁes p(t) = 0. (i) t is algebraic over K.√2 (X) = X 2 − 2. Example.t (X) for some u ∈ K. Let t ∈ L. Example. Notice in particular that for an element t ∈ K. Then the following conditions are equivalent. Consider C/Q.t (X). ker εt = (minpolyK.9(ii). √ deg minpolyQ.t (X)  p(X) and so p(X) = u minpolyK. 3. recall the following notion.5. Clearly X 2 − 2 ∈ ker ε√2 since ( 2)2 − 2 = 0. Proposition. By Example 2. Suppose that t ∈ L is algebraic over K and that p(X) ∈ ker εt with deg p(X) = deg minpolyK. The minimal polynomial of i ∈ C over Q is X 2 + 1.4.
Solution. 3. Since Φ6 (X) is irreducible and monic. Example. 2 √ √ √ √ √ √ 1 √ 3= ( 2 + 3) + ( 3 − 2) ∈ Q( 2 + 3). Since Q( 2 + 3) Q( 2. ALGEBRAIC EXTENSIONS OF FIELDS Proof. Solution. we must have minpolyQ. Since deg p(X) = 4 and p(X) is monic. 3) we must have √ √ √ √ Q( 2 + 3) = Q( 2. 2 √ √ √ √ √ √ Q( 2 + 3).9. √ √ hence Q( 2.ζ6 (X)  Φ6 (X). hence / √ √ degQ(√2) ( 2 + 3) = 2. we have minpolyQ. √ √ √ Referring to Example 2. Consider √ minpolyQ(√2).i (X) = X 2 + 1. Similarly.12 we see that 2 + 3 ∈ Q( 2). ζ6 ∈ C over Q. Then Φ6 (X) ∈ ker εζ6 so minpolyQ.−√2+√3 (X) √ √ = (X 2 − 2 2X − 1)(X 2 + 2 2X − 1) = X 4 − 10X 2 + 1. 3) Referring to Example 2. . √ One polynomial in ker ε√2+√3 Q( 2)[X] is √ √ √ √ √ (X − ( 2 + 3))(X − ( 2 − 3)) = X 2 − 2 2X − 1. Example. 3− 2= =√ ( 3 + 2) 2+ 3 So we have √ √ √ √ √ √ 1 √ 2= ( 2 + 3) − ( 3 − 2) ∈ Q( 2 + 3). Notice that √ √ √ √ √ √ √ √ ( 3 − 2)( 3 + 2) 1 √ √ √ ∈ Q( 2 + 3).√2+√3 (X) = X 4 − 10X 2 + 1. Consider C/Q. Recall from Example 1. Clearly X 2 + 1 ∈ ker εi since i2 + 1 = 0. Find the minimal polynomial of √ √ 2 + 3 over Q. Consider C/Q.−√2+√3 (X) = X 2 + 2 2X − 1. 3.√2+√3 (X) = X 2 − 2 2X − 1. √ minpolyQ(√2). we have minpolyQ.8. Find the minimal polynomial of the primitive 6th root of unity.ζ6 (X) = Φ6 (X) and so degQ ζ6 = 2.√2+√3 (X) minpolyQ(√2).34 3.10 or Proposition 2.10 we see that Let us ﬁnd a nonzero polynomial in ker ε√2+√3 Q[X]. 3). √ √ Then p( 2 + 3) = 0 so p(X) ∈ ker εt . Since this is monic and of degree 2.44 that ζ6 is a root of the irreducible cyclotomic polynomial Φ6 (X) = X 2 − X + 1. p(X) = minpolyQ(√2). As [Q(i) : Q] = 2. √ √ degQ ( 2 + 3) = 4.
t (X) and therefore q(X) = minpolyK.1. Let M/L and L/K be algebraic extensions. Let L/K be a ﬁnite extension. pm . . 3. Use induction on n together with Proposition 2. Proof. Then I(t) = {f (X) ∈ K[X] : f (t) = 0} ⊆ K[X] is an ideal which is principal and has an irreducible monic generator q(X) ∈ K[X]. ALGEBRAIC EXTENSIONS 35 3. Definition. Later we will see that when char K = 0 every ﬁnite extension L/K has a primitive element. tn . Let L/K be an extension and suppose that u1 . . Let L/K be a ﬁnite extension. 3. u)/K is ﬁnite.13. By Theorem 2. Then L/K is algebraic.14. 3. un )/K is a ﬁnite extension.12.10. The next Lemma will often be useful. so by Proposition 3. pm )/K is ﬁnite and so is K(p0 . Then u is algebraic over L. Then K(u) = L if and only dimK K(u) = dimK L. .16. . Let L/K be a ﬁnite extension and u ∈ L. suppose that q(X) = q1 (X)q2 (X) with deg qi (X) < deg q(X). Proposition. . . Then the extension M/K is algebraic. . . Let t ∈ L be algebraic over K. 3.t (X). Proposition. must be linearly dependent over K. . By Lemma 3. . . . Since degK u = dimK K(u) and [L : K] = dimK L the result follows. hence q1 (X) ∈ I(t) or q2 (X) ∈ I(t). The irreducible monic polynomial minpolyK.6(ii). K(u) ⊆ L is a ﬁnite dimensional Kvector subspace. Let t ∈ L. It is easy to see that I(t) K[X] and therefore I(t) = (q(X)) for some monic generator q(X). pm ).t (X).8 and Theorem 2. In fact.13. 3.t (X) is in I(t) so q(X)  minpolyK. The extension L/K is algebraic or L is algebraic over K if every element t ∈ L is algebraic over K. Lemma. u)/K(p0 . Sometimes the minimal polynomial of an element in an extension is introduced in a diﬀerent but equivalent way. K(p0 . Then K(u1 . Then u is a primitive element for L/K if and only if degK u = [L : K]. Lemma. . Now as q1 (t)q2 (t) = 0. we must have q1 (t) = 0 or q2 (t) = 0. . .15. Proposition. contradicting the above assumption that deg qi (X) < deg q(X). pm . These possibilities give q(X)  q1 (X) or q(X)  q2 (X) and so deg q(X) deg q1 (X) or deg q(X) deg q2 (X). We now come to an important notion for extensions. un ∈ L are algebraic. . 3. . . q(X) = minpolyK. To see that q(X) is irreducible. .11. .15. . . the extension K(p0 . . . . Hence for suitable coeﬃcients cj ∈ K not all zero and some m 1 we have c0 + c1 t + · · · + cm tm = 0. . . . the powers 1.6(ii). Proof. . Proof. But this means that t is algebraic over K. Since the Kvector space L is ﬁnite dimensional. . Proof. so there is a polynomial p(X) = p0 + p1 X + · · · + pm X m ∈ L[X] of positive degree with p(u) = 0. Proof. when viewed as elements of this vector space. Definition. . Let u ∈ M .3. . An element u ∈ L for which L = K(u) is called a primitive element for the extension L/K. t. u is algebraic over K.
3.u (X) ∈ K[X]. The sets Calg and Ralg are both countable. v) and if u = 0. When dealing with algebraic extensions of Q we will usually work with subﬁelds of Q = Calg .36 3. Proposition.u (X) = deg minpolyF0 . −1 X −1 K(u)/K. v) is algebraic over K. Now consider F0 = K(c0 . Now for any subextension F/K K(u)/K we can also consider minpolyF. a −1 ). Similarly. v).u (X).u (X) in K(u)[X]. . v ∈ Lalg . We must show that Lalg L.2. so there are only a ﬁnite number of possibilities for minpolyF. . each of which + X ∈ K(u)[X] . 3. whereas C and R are uncountable.33 implies that minpolyK. Splitting ﬁelds and Kronecker’s Theorem We can now answer a basic question.u (X) = [K(u) : F0 ]. Question. Let K(u)/K be a ﬁnite simple extension.u (X) = minpolyF0 . The Unique Factorization Property 1. Consider the minimal polynomial minpolyK. in the extension R/Q the subﬁeld Ralg = Calg ∩ R C consists of all the real algebraic numbers. Lalg is a subﬁeld containing K and Lalg /K is algebraic. This shows that there are only ﬁnitely many subextensions F/K has the form K(a0 . For an extension L/K. which divides minpolyK. Is there an extension ﬁeld L/K for which p(X) has a root in L? A stronger version of this question is the following. Then by Lemma 3. c1 . 3. 3. A more usual notation for Calg is Q since this is the algebraic closure of Q which will be discussed later.20.17. . Let u. Then there are only ﬁnitely many subextensions F/K K(u)/K.13.21. where a0 + a1 X + · · · + a is a factor of minpolyK. let Lalg = {t ∈ L : t is algebraic over K} ⊆ L.u (X) in F [X]. u + v and uv are in K(u. so there are in fact many more transcendental numbers but it can be hard to determine whether a given number is transcendental or not. We have [K(u) : F ] = deg minpolyF. v)/K is a ﬁnite dimensional extension.u (X) = c0 + c1 X + · · · + ck−1 X k−1 + X k ∈ F [X]. . Proposition. Proof. hence F = F0 . . In particular. ALGEBRAIC EXTENSIONS OF FIELDS 3. examples are e and π. . In the extension C/Q we can consider Calg C which is called the subﬁeld of algebraic numbers. We end this section with a technical result. Then F0 F and so minpolyF. uv and u−1 are all algebraic over K. ck−1 ). u−1 is also in K(u. 3. Example.u (X) ∈ F0 [X] is irreducible since it is irreducible in F [X]. Clearly K ⊆ Lalg .18. a1 . . Proof. Let K be a ﬁeld and p(X) ∈ K[X] be a polynomial of positive degree. the extension ﬁeld of K generated by the coeﬃcients of minpolyF. For an extension L/K.19. Therefore u + v. hence minpolyF. Definition. . hence every element of K(u. Elements of C−Calg are called transcendental complex numbers.u (X).u (X).u (X) has only ﬁnitely many monic divisors in K(u)[X]. K(u.
uk ) E which is the smallest subﬁeld of E that answers Question 3.24. Solution. .22. Notice that p(X) = (X 2 − 2)(X 2 + 2). Of course. .22. We then refer to a subﬁeld of C which is a splitting ﬁeld as the splitting subﬁeld.21. . In practise we often have extension ﬁelds ‘lying around in nature’ containing roots and we can work inside of these. . . . . SPLITTING FIELDS AND KRONECKER’S THEOREM 37 3. Such a minimal extension of K is called a splitting ﬁeld of p(X) over K and we will sometimes denote it by K(p(X)) or Kp . 3. . having found one root u1 in an extension L1 /K we discard the linear factor X − u1 and consider the polynomial p1 (X) = p(X) ∈ L1 [X]. Then there is a ﬁnite extension E/K which is a splitting ﬁeld of p(X) over K.26. To answer Question 3. Let K be a ﬁeld and p(X) ∈ K[X] be a polynomial of positive degree. this construction is only interesting if qj (X) to has degree bigger than 1 since a linear polynomial already has a root in K. Definition.25. We begin by factorizing p(X) ∈ K[X] into irreducible monic factors qj (X) together with a constant factor c: p(X) = cq1 (X) · · · qr (X). At each stage we either already have another root in L1 or we need to enlarge the ﬁeld to obtain one. Question. Proof.3. uk of p(X) in E generate a subﬁeld K(u1 . When working over Q (or any other subﬁeld of C) we can always ﬁnd roots in C by the Fundamental Theorem of Algebra. Namely. 3. Hence p(X) has a root in K[x]/(qj (X)). p(X) ∈ K[X] splits in E/K or over E if it factorizes into linear factors in E[X]. 3. We already know how to answer Question 3. . if we have such a ﬁeld E then the distinct roots u1 . Let K be a ﬁeld and p(X) ∈ K[X] be a polynomial of positive degree. Theorem (Kronecker’s Theorem: ﬁrst version). Theorem (Kronecker’s Theorem: second version). Now for any j we can form the quotient ﬁeld K[x]/(qj (X)) which is a ﬁnite dimensional (simple) extension of K and in which the coset X + (qj (X)) satisﬁes the equation qj (X + (qj (X))) = 0 + (qj (X)).27. √ √ √ √ so ﬁrst we adjoin the roots ± 2 of (X 2 − 2) to form Q( 2. Is there an extension ﬁeld E/K for which p(X) factorizes into linear factors in E[X]? 3. X − u1 We can repeat the argument to form a ﬁnite extension of L1 (and hence of K) containing a root of p1 (X) and so on. 3. Of course.2. Definition. Example.22 we iterate this construction. Find a splitting ﬁeld E/Q for p(X) = X 4 − 4 over Q and determine [E : Q].23. − 2) = Q( 2) which gives an √ extension Q( 2)/Q of degree 2. Then there is a ﬁnite extension L/K for which p(X) has a root in L.
38
3. ALGEBRAIC EXTENSIONS OF FIELDS
√ √ Next consider the polynomial X 2 + 2 ∈ Q( 2)[X]. The √ complex roots of X 2 + 2 are ± 2i and these are not √ real, so this polynomial is irreducible √ Q( 2)[X]. Hence we need to consider in √ √ √ Q( 2, 2i) = Q( 2, i) and the extension Q( 2, i)/Q( 2) which has degree 2. C
∞
√ Q( 2, i)
adjoin roots of X 2 + 2 2
√ Q( 2)
adjoin roots of X 2 − 2 2
Q √ √ Thus the splitting subﬁeld of p(X) over Q in C is Q( 2, i) and [Q( 2, i) : Q] = 4. Of course we could have started by ﬁrst adjoining roots of X 2 + 2 and then adjoining roots of X 2 − 2, thus giving the tower C
∞
√ Q( 2, i)
adjoin roots of X 2 − 2 2
√ Q( 2i)
adjoin roots of X 2 + 2 2
Q An important point is that if a splitting ﬁeld exists inside of a given extension ﬁeld F/K, it is unique as a subﬁeld of F . 3.28. Proposition. Let F/K be an extension ﬁeld and p(X) ∈ K[X]. If E1 , E2 splitting subﬁelds for p(X) over K then E1 = E2 . F are
Proof. Let u1 , . . . , uk ∈ F be the distinct roots of p(X) in F . By deﬁnition, K(u1 , . . . , uk ) is the smallest subﬁeld containing K and all the uj . But K(u1 , . . . , uk ) must be contained in any splitting subﬁeld, so E1 = K(u1 , . . . , uk ) = E2 . Since we will frequently encounter quadratic polynomials we record a useful result on roots of such polynomials. Recall that p(X) = aX 2 + bX + c ∈ K[X] is quadratic if a = 0 and its discriminant is ∆ = b2 − 4ac ∈ K. The proof of the next result is the standard one which works provided 2 has an inverse in K, i.e., when char K = 2. 3.29. Proposition. Let K be a ﬁeld of characteristic diﬀerent from 2. Then the quadratic polynomial p(X) = aX 2 + bX + c ∈ K[X] has • no roots in K if ∆ is not a square in K; • one root −b/(2a) = −(2a)−1 b if ∆ = 0;
3.3. MONOMORPHISMS BETWEEN EXTENSIONS
39
• two distinct roots −b + δ = (2a)−1 (−b + δ), 2a if ∆ = δ 2 for some nonzero δ ∈ K.
−b − δ = (2a)−1 (−b − δ), 2a
In particular, the splitting ﬁeld of p(X) over K is K if ∆ is a square in K and K(δ) otherwise, where δ is one of the two square roots of ∆ in some extension of K such as the algebraic closure K which we will introduce in Section 3.4. 3.30. Example. Find a splitting ﬁeld E/Q for p(X) = X 3 − 2 over Q and determine [E : Q]. √ Solution. By the Eisenstein Test 1.38, p(X) is irreducible over Q. One root of p(X) is √ 3 2 ∈ R so we adjoin this to Q to form an extension Q( 3 2)/Q of degree 3. Now √ √ √ 3 3 3 p(X) = (X − 2)(X 2 + 2X + ( 2)2 ) √ √ 2 and the second factor has the nonreal complex roots 3 2 ζ3 , 3 2 ζ3 lying in the extension √ √ √ Q( 3 2, ζ3√ )/Q( 3 2) of degree 2. So the splitting subﬁeld of X 3 − 2 in C over Q is Q( 3 2, ζ3 ) with [Q( 3 2, ζ3 ) : Q] = 6. √ √ 2 An alternative strategy would have been to adjoin one of the other roots 3 2 ζ3 or 3 2 ζ3 ﬁrst. We could also have begun by adjoining ζ3 to form the extension Q(ζ3 )/Q, but none of √ the roots of p(X) lie in this ﬁeld so the extension Q( 3 2, ζ3 )/Q(ζ3 ) of degree 3 is obtained by adjoining one and hence all of the roots. √ Figure 3.1 shows all the subﬁelds of the extension Q( 3 2, ζ3 )/Q. C
2
R
∞
VV3 iii q V iiii qqqq 2 2 iiii qq 2 VV VV q iiii ii qq √ 2 VVV3 √ iiii √ Q( 3 2 ζ3 ) VV Q( 3 2) Q( 3 2 ζ3 ) `` vvv VV vvv `` VV vvv `` VV vvv `` 3 vvv ``3 3 Q(ζ3 ) vvv ` vvv ``` ss 2 ss vvv `` s vvv `` sss vv sss
∞
√ Q( 3 2, ζ )
Q
√ Figure 3.1. The subﬁelds of Q( 3 2, ζ3 )/Q 3.3. Monomorphisms between extensions 3.31. Definition. For extensions F/K and L/K, let MonoK (L, F ) denote the set of all monomorphisms L −→ F which ﬁx the elements of K. 3.32. Remark. We always have AutK (F ) ⊆ MonoK (F, F ) and MonoK (F, F ) is closed under composition but is not always a group since elements are not necessarily invertible. If F/K is ﬁnite, then we do have MonoK (F, F ) = AutK (F ) since every injective Klinear transformation is surjective and so invertible.
40
3. ALGEBRAIC EXTENSIONS OF FIELDS
We will also use the following notation. 3.33. Definition. Let F/K be an extension and p(X) ∈ K[X]. Set Roots(p, F ) = {u ∈ F : p(u) = 0}, the set of roots of p(X) in F . This is always a ﬁnite set which may of course be empty, which happens precisely when p(X) has no root in F . Suppose that p(X) ∈ K[X] is an irreducible polynomial which we might as well assume is monic, and let F/K be an extension. Then if t ∈ F is a root of p(X), the evaluation homomorphism εt : K[X] −→ F factors through the quotient monomorphism εt : K[X]/(p(X)) −→ F whose image is K(t) F . Of course, there is one such monomorphism for each root of p(X) in F . If we ﬁx one such root t0 and identify K[X]/(p(X)) with K(t0 ) via εt0 , then each root of p(X) in F gives rise to a monomorphism ϕt = εt ◦ ε−1 : K(t0 ) −→ F for which ϕt (t0 ) = t. t0
ϕt =εt ◦ε−1 t
0
K(t0 ) o
εt0 ∼ =
K[X]/(p(X))
εt
*/
F
Notice that if ϕ : K[X]/(p(X)) −→ F is any homomorphism extending the identity function on K, then the coset X + (p(X)) must be sent by ϕ to a root of p(X) in F , hence every such homomorphism arises this way. This discussion is summarized in the following result. 3.34. Proposition. Let F/K be a ﬁeld extension. Let p(X) ∈ K[X] be an irreducible polynomial with t0 ∈ F be a root of p(X). Then there is a bijection Roots(p, F ) ←→ MonoK (K(t0 ), F ) given by t ←→ ϕt , where ϕt : K(t0 ) −→ F has the eﬀect ϕt (t0 ) = t. √ 3.35. Example. Show that MonoQ (Q( 2), C) has two elements. √ Solution. We have Q( 2) ∼ Q[X]/(X 2 − 2) where X 2 − 2 is irreducible over Q. Hence = √ √ the Qmonomorphisms we want send 2 to ± 2 which are √ complex roots of X 2 − 2. In the fact both possibilities occur, giving monomorphisms id, α : Q( 2) −→ C, where √ √ α(a + b 2) = a − b 2. √ We can replace C by Q( 2) to obtain √ √ √ √ MonoQ (Q( 2), C) = MonoQ (Q( 2), Q( 2)) = AutQ (Q( 2)). We will see that this is not always true. √ √ √ 3.36. Example. Show that MonoQ (Q( 3 2), C) has 3 elements but MonoQ (Q( 3 2), Q( 3 2)) contains only the identity function. √ √ Solution. Here minpolyQ, √2 (X) = X 3 − 2 and there are 3 complex roots 3 2, 3 2 ζ3 , 3 √ √ √ 2 3 2 ζ3 . As two of these roots are not real, MonoQ (Q( 3 2), Q( 3 2)) contains only the identity √ since Q( 3 2) R. √ √ √ 2 Each of the above roots corresponds to one of the subﬁelds Q( 3 2), Q( 3 2 ζ3 ) or Q( 3 2 ζ3 ) √ of C and there are 3 monomorphisms α0 , α1 , α2 : Q( 3 2) −→ C given by √ √ √ √ 3 3 3 3 α0 (a + b 2 + c( 2)2 ) = a + b 2 + c( 2)2 , √ √ √ √ 3 3 3 3 2 α1 (a + b 2 + c( 2)2 ) = a + b 2 ζ3 + c( 2)2 ζ3 , √ √ √ 2 √ 3 3 3 3 α2 (a + b 2 + c( 2)2 ) = a + b 2 ζ3 + c( 2)2 ζ3 .
ζ3 ). α1 : . ζ3 )) = AutQ (Q( 2. ζ3 )/Q in Example 3.3.37. √ Solution. α1 : 2 ζ3 −→ ζ3 ζ3 −→ ζ3 √ √ 2 Finally we consider monomorphisms that send 3 2 to 3 2 ζ3 . C) = MonoQ (Q( 2. then αp : Roots(p. in √ 3 First consider monomorphisms that ﬁx 2 and hence ﬁx the subﬁeld Q( 3 2). It is a nice exercise to show that AutQ (Q( 3 2. . ζ3 ). It is also worth remarking that  AutQ (Q( 2. Proposition. Q( 2. ζ3 )) = [Q( 3 2. (ii) From (i). L) into Roots(p. We have already met the extension Q( 3 2. These form the subset √ √ 3 3 MonoQ( √2) (Q( 2. Let F/K and L/K be extensions. (i) For u ∈ Roots(p. L) −→ Roots(p. or more generally algebraic. √ √ 3 3 α1 Q( 2) = Q( 2 ζ3 ). Determine MonoQ (Q( 3 2. ζ3 ) : Q]. Part (ii) says that any automorphism of L/K permutes the set of roots in L of a polynomial p(X) ∈ K[X].3. This time we have 2 distinct ways to √ extend to elements of MonoQ (Q( 3 2. The possibilities are √ √ √ √ 3 3 2 −→ 3 2 ζ3 2 −→ 3 2 ζ3 . 3. √ 3. hence it is also surjective by the Pigeon Hole Principle. so α maps Roots(p. ζ3 ). Q( 3 2. the symmetric = √ √ 3 group on 3 objects. Example. L) −→ Roots(p. L) is an injective function from a ﬁnite set to itself. L) is a bijection. ζ3 )) ∼ S3 . extensions it is the key to understanding the automorphism group and this is a fundamental insight of Galois Theory. 3 √ √ We know that Q( 3 2. ζ3 ). F ) which is an injection. α ﬁxing polynomial Φ3 (X) = X 0 √ Q( 3 2). There are again two possibilities √ √ √ 2 √ 2 3 3 2 −→ 3 2 ζ3 2 −→ 3 2 ζ3 . where α0 has the eﬀect √ √ 3 2 −→ 3 2 α0 : . each monomorphism α ∈ MonoK (L. So there are two monomorphisms id. This gives us a strong hold on the possible automorphisms. ζ3 ). L) −→ Roots(p. In the case of ﬁnite. F ). Thus αp : Roots(p. L) ⊆ L is also an injection. (i) For p(X) ∈ K[X]. MONOMORPHISMS BETWEEN EXTENSIONS 41 These mappings have images √ √ 3 3 α0 Q( 2) = Q( 2). ζ3 ). L). L) −→ Roots(p. √ so these form a group. C) ⊆ MonoQ (Q( 2. L) we have p(α(u)) = α(p(u)) = α(0) = 0. L) is a bijection. Since α is an injection its restriction to Roots(p. 2 ζ3 −→ ζ3 √ √ 3 Next we consider monomorphisms that send √ 2 to 3 2 ζ3 . ζ3 ) = Q( 3 2)(ζ3√ and that ζ3 is a root of the irreducible cyclotomic ) 2 + X + 1 ∈ Q( 3 2)[X]. αp : Roots(p. (ii) If α ∈ MonoK (L. We end this section with another useful result. α2 : . α2 : 2 ζ3 −→ ζ3 ζ3 −→ ζ3 These are all 6 of the required monomorphisms. We build up the list of monomorphisms √ stages. ζ3 )). Proof. C). √ √ 2 3 3 α2 Q( 2) = Q( 2 ζ3 ). C). It is also the case here that √ √ √ √ 3 3 3 3 MonoQ (Q( 2. ζ3 )) since again we can send ζ3 to either ζ3 or 2 ζ3 .30 and we will make use of information from there.38. F ) restricts to a function αp : Roots(p.
Let v ∈ Lalg and suppose that q(v) = 0 for some q(X) ∈ K[X] of positive degree. then L = K.42 3. . ALGEBRAIC EXTENSIONS OF FIELDS 3. 3. Let K be a ﬁeld. Now Roots(q. ud ∈ C and this is unique apart from the order of the roots uj . 3. There are some immediate consequences of Theorem 3.37(ii). Proposition. We will temporarily write . Then α restricts to an automorphism αalg : Lalg −→ Lalg . Theorem.41. L) −→ Roots(q. L). .44. We are already familiar with the example C = C. Theorem (Fundamental Theorem of Algebra for C). Question. . Then there is an isomorphism ϕ : F1 −→ F2 which ﬁxes the elements of K. We must show that αalg is a bijection by showing it is surjective. 3. so α maps Lalg ⊆ L into itself and therefore gives rise to a restriction αalg : Lalg −→ Lalg which is also a monomorphism. Is there an algebraically closed ﬁeld F containing K? By taking F alg we might as well ask that such a ﬁeld be algebraic over K. (i) If L/K is an algebraic extension. This shows that v ∈ im α and so αalg is surjective. and it is also ﬁnite. See [3] for a proof using Zorn’s Lemma (see Axiom 3. Algebraic closures An important property of the complex numbers is that C is algebraically closed. L) = ∅ since it contains v. Corollary. where c. Suppose that u ∈ Lalg . (i) There is an algebraic closure of K. Let K be a ﬁeld. L) ⊆ Lalg . . u1 . It is natural to pose the following question.40.45. K } eee ee }} ee }} e }} ~} ϕ / F2 F1 Hence algebraic closures are essentially unique. Then αq : Roots(q. Let L/K be an extension and α ∈ MonoK (L. . Proposition.42. . Proof.39. hence v = αq (w) = α(w) for some w ∈ Roots(q.48) which is logically equivalent to the Axiom of Choice.4. Proof. referring to it as the algebraic closure of K. 3. 3.43. Let K be a ﬁeld. Let K be a ﬁeld. Every nonconstant polynomial p(X) ∈ C[X] has a root in C. (ii) Let F1 and F2 be algebraic closures of K. E1 = E2 to indicate that for extensions E1 /K and E2 /K there is an isomorphism E1 −→ E2 ﬁxing the elements of K. 3. 3. An extension F/K is called an algebraic closure of K if F is algebraic over K and algebraically closed. Every nonconstant polynomial p(X) ∈ C[X] has a factorization p(X) = c(X − u1 ) · · · (X − ud ). L) is a bijection by Proposition 3. Then p(α(u)) = α(p(u)) = α(0) = 0. Definition.44. say p(u) = 0 for some p(X) ∈ K[X] of positive degree. Because of the uniqueness we usually ﬁx some choice of algebraic closure of K and write K or K alg cl .
(ii) If L/K is an extension. (X. the Monomorphism Extension Theorem. y • An element x ∈ X is a maximal element of X if x y =⇒ y = x. indeed. ) is a partially ordered set. θ1 ) (F2 . y. Let (X. z ∈ X. x then x = y.48. For example.49. we have Q = Calg and R = C. every element of L is algebraic over K. ) is totally ordered if for every pair x. Theorem (Monomorphism Extension Theorem). y or y x is true. by Proposition 3. • x x. We consider the set X consisting of all pairs (F. which shows that it is algebraically closed. Also there is a function θ : F −→ K deﬁned by θ(u) = θ(u) .46. Then F /L M/L.θ)∈Y F. θ). There is a stronger result than Theorem 3. 3.16. all of its roots are in fact algebraic over K since (L)alg is algebraic over K. Hence these roots lie in (L)alg . A partially ordered set (X. Since L is algebraically closed it is an algebraic closure of K.44(ii). y. 3. Let (X. 8K Ñ@ ÑÑ ÑÑ ÑÑ Ñ ÑÑ M ÑÑ ϕ0 Ñ ÑÑ ÑÑ Ñ ϕ L K = /K Proof.4.3. Suppose that ϕ0 : L −→ K is a monomorphism ﬁxing the elements of K. Then there is an extension of ϕ0 to a monomorphism ϕ : M −→ K. Let M/K be an algebraic extension and L/K M/K. Then X has a maximal element. • y ∈ X is an upper bound for Y if for every y ∈ Y . Definition. ) consists of a set X and a binary relation such that whenever x. then so is L/K and (L)alg = K. which we will ﬁnd useful.47. Then (X. Let F = (F. Suppose that Y ⊆ X is a totally ordered subset. ALGEBRAIC CLOSURES 43 . We order X using the relation for which (F1 . (i) By Proposition 3. at least one of x 3. 3. ) be a partially ordered set in which every totally ordered subset has an upper bound. Definition. ) be a partially ordered set and Y ⊆ X. where F/L M/L and θ : F −→ K extends ϕ0 . Axiom (Zorn’s Lemma). First we need some deﬁnitions. y ∈ X. Proof. (ii) Every nonconstant polynomial in (L)alg [X] has a root in L.16. Again the proof uses Zorn’s Lemma which we state below. • if x y and y • if x y and y z then x z. θ2 ) whenever F1 F2 and θ2 extends θ1 .
. v ∈ K.. f (u) = 0.u (X) = a0 + a1 X + · · · + ad−1 X d−1 + X d . there is a monomorphism ϕv : K(u) −→ K with ϕv (u) = v. We need to understand more clearly when an irreducible polynomial has a multiple root since this turns out to be important in what follows. Definition. Then v is conjugate to u over K or is a conjugate of u over K if there is a monomorphism ϕ : K −→ K ﬁxing K for which v = ϕ(u). 3. If minpolyM0 .u (X)) to give a monomorphism θ0 : M0 (u) −→ K extending θ0 .50.u (v) = 0. v ∈ K say.54. Suppose that M0 = M . the function ∂ : K[X] −→ K[X] given by ∂(f (X)) = f (X) = a1 + 2a2 X + · · · + dad X d−1 . Since M is algebraic / over K it is also algebraic over M0 . θ) is an upper bound for Y . Consider the formal derivative on K[X]. Multiplicity of roots and separability Let K be a ﬁeld. Suppose that v = ϕ(u) for some ϕ ∈ MonoK (K. This factors through the quotient ring M0 [X]/(minpolyM0 . then v is conjugate to u over K if and only if minpolyK. 3. 3. Hence M0 = M and so we can take ϕ = θ0 . The formal derivative ∂ : K[X] −→ K[X] has the following properties. But then (M0 . Then we can factor f (X) as f (X) = (X − u)f1 (X) for some f1 (X) ∈ K[X]. θ0 ). where f (X) = a0 + a1 X + a2 X 2 + · · · + ad X d with aj ∈ K. ALGEBRAIC EXTENSIONS OF FIELDS whenever u ∈ F for (F. contradicting the maximality of (M0 .44 3. Let u ∈ K and suppose that p(X) = minpolyK. If u. θ0 ) = (M0 (u).51. so (F . so θ is welldeﬁned. θ0 ).50. K). i. i.53. θ) ∈ Y . θ0 ) and (M0 . The converse follows from Example 3. The Homomorphism Extension Property 1. Then for any other root of p(X). If f1 (u) = 0 then u is a multiple or repeated root of f (X).u (X) ∈ K[X].u (X) = a0 + · · · + an−1 X n−1 + X n . 3. θ0 ) (M0 (u).e. Let u. θ). (M0 . Lemma.22 of the polynomial ring M0 [X] applied to the monomorphism θ0 : M0 −→ K yields a homomorphism θ0 : M0 [X] −→ K extending θ0 and for which θ0 (u) = v. then the polynomial f (X) = θ0 (a0 ) + · · · + θ0 (an−1 )X n−1 + X n ∈ (θ0 M0 )[X] is also irreducible and so it has a root v in K (which is also an algebraic closure of θ0 M0 K). Proof.e. Proposition. v ∈ K. θ0 ). It is straightforward to check that if u ∈ F for (F .5. If minpolyK. Suppose that f (X) ∈ K[X] and u ∈ K is a root of f (X). Definition. By Zorn’s Lemma there must be a maximal element of X. This extends to a monomorphism ϕ : K −→ K. θ) ∈ Y we have (F. then a0 + a1 u + · · · + ad−1 ud−1 + ud = 0 and so a0 + a1 v + · · · + ad−1 v d−1 + v d = ϕ(a0 + a1 u + · · · + ad−1 ud−1 + ud ) = 0. θ ) ∈ Y then θ (u) = θ(u). hence u is algebraic over M0 . 3. Example. θ) (F .52. 3. Then for every (F. If f1 (u) = 0 then u is a simple root of f (X). so there is an element u ∈ M for which u ∈ M0 ..
g(X) ∈ K[X]. im ∂ is spanned by the monomials X k for which ∂(X k+1 ) = 0. Proof.59. Solution. For n 1. 3. (iii) If char K = 0. f (ζ) = nζ n−1 = 0. But then ∂(X r+s ) = (r + s)X r+s−1 = rX r−1 X s + sX r X s−1 = ∂(X r )X s + X r ∂(X s ). Using this we see that ∂(a0 + a1 X + a2 X 2 + · · · + ad X d ) = 0 ⇐⇒ am = 0 whenever p m. so f (u) = f1 (u). We have f (X) = 4X 3 +16X. Proof. Show that 2i is a multiple root of f (X) = X 4 + 8X 2 + 16..5.57. Using Long Division and the Euclidean Algorithm we ﬁnd that gcd(f (X). Hence u is a multiple root if and only if f (X) and f (X) have a common factor in L[X] (and hence in K[X] by Proposition 3. 3. (i) This is routine. where 2i is also a root of X 2 + 4. 3. this can only happen if char K > 0. ∂(X m ) = mX m−1 and this is zero if and only if p  m. s 0. In particular. Then f (X) = f1 (X) + (X − u)f1 (X). let f (X) = (X − u)f1 (X). Solution. If f (X) is irreducible in K[X] then a root u is a multiple root if and only if f (X) = 0. In fact. Hence 2i is a multiple root of f (X). If char K = 0 and f (X) is irreducible in K[X]. So ∂(f (X)) = 0 if and only if f (X) = a0 ∈ K. Example. 3. . for f (X). Then u is a multiple root of f (X) if and only if f (X) and f (X) have a common factor of positive degree in K[X] which vanishes at u.56. then ker ∂ = {h(X p ) : h(X) ∈ K[X]} and im ∂ is spanned by the monomials X k with p (k + 1). f (X)) = X 2 + 4. (ii) By Klinearity. X 4 + 8X 2 + 16 = (X 2 + 4)2 . Proposition. so this is obvious. It is also clear that every polynomial g(X) ∈ K[X] has the form g(X) = ∂(f (X) where f (X) is an antiderivative of g(X).58. show that each of the roots of f (X) = X n − 1 in C is simple. We now apply the formal derivative to detect multiple roots. (iv) For a monomial X m . Corollary.3. 3. (iv) If char K = p > 0. then ker ∂ = K and ∂ is surjective. Let f (X) ∈ K[X] have a root u ∈ L for some extension L/K. Also. (ii) ∂ is a derivation.55.e. We have f (X) = ∂(X n − 1) = nX n−1 . Corollary. Example. it suﬃces to verify this for the case where f (X) = X r and g(X) = X s with r. then every root of f (X) is simple. Working in L[X]. MULTIPLICITY OF ROOTS AND SEPARABILITY 45 (i) ∂ is Klinear. i. (iii) If f (X) = a0 + a1 X + a2 X 2 + · · · + ad X d then ∂(f (X)) = 0 ⇐⇒ a1 = 2a2 = · · · = dad = 0. so for any root ζ of f (X). ∂(f (X)g(X)) = ∂(f (X))g(X) + f (X)∂(g(X)). which are the ones with p (k + 1).12) and which vanishes at u.
3. Corollary. this is equivalent to requiring that p (X) = 0.64. The separable degree of L over K is (L : K) =  MonoK (L.66. Definition. Example. Let u ∈ K be a root of p(X) and suppose that it has multiplicity m. then the multiplicities of the uj are equal. 3. Let K be a ﬁeld and let K be an algebraic closure. so we can write p(X) = (X − u)m p1 (X) where p1 (X) ∈ K(u)[X] and p1 (u) = 0. p(X) = c(X − u1 )m · · · (X − uk )m . . uk ∈ K. Then ϕv ((X − u)m p1 (X)) = (X − u)m p1 (X). (X − v)m must divide p1 (X) in K[X]. Definition.34.68.46 3. and therefore the multiplicity of v must be at least m. 3. If the irreducible polynomial p(X) ∈ K[X] has distinct roots u1 . Corollary.65. so if ζ is any root of f (X) then f (ζ) = 0. Interchanging the rˆles of u and v we ﬁnd that the multiplicities of u and v are in fact equal. Let K be a ﬁeld and let u ∈ K.60. Definition. . there is a monomorphism ϕv : K(u) −→ K for which ϕv (u) = v. 3. Proof. 3. o 3. . where c ∈ K and k = deg p(X). Show that each of the roots of f (X) = X p − 1 in L is multiple. K).67. Hence in K[X].63.56. 3. An irreducible polynomial p(X) ∈ K[X] is separable over K if every root of p(X) in an extension L/K is simple. m where m is the multiplicity of u in minpolyK.57. An algebraic extension L/K is called separable if every element of L is separable over K. Now by Corollary 1. Example. . Proposition. where c ∈ K and m 1. Solution. and so (X − v)m p1 (X) = (X − u)m p1 (X). then the multiplicity of u in p(X) is the maximum m such that p(X) = (X − u)m q(X) for some q(X) ∈ L[X]. 3. Let K be a ﬁeld and let K be an algebraic closure. Now let v ∈ K be any other root of p(X). uk ∈ K which are all simple then in K[X]. ALGEBRAIC EXTENSIONS OF FIELDS 3. If the irreducible polynomial p(X) ∈ K[X] has distinct roots u1 .u (X) ∈ K[X] is separable. . When p(X) is viewed as an element of K(u)[X]. An algebraic element u ∈ L in an extension L/K is separable if its minimal polynomial minpolyK. Let p > 0 be a prime and suppose that L/Fp is an extension. . Then the number of distinct conjugates of u is deg minpolyK. Let L/K be a ﬁnite extension.u (X) . We have f (X) = ∂(X p − 1) = pX p−1 = 0. An algebraic extension L/K of a ﬁeld of characteristic 0 is separable by Corollary 3.62. .34. Definition. p(X) = c(X − u1 ) · · · (X − uk ). If u ∈ L is a multiple root of p(X).u (X). Later we will see that 1 is the only root of X p − 1. .61. By Corollary 3. the coeﬃcients of p(X) are ﬁxed by ϕv . . where p1 (X) ∈ K[X] is obtained applying ϕv to the coeﬃcients of p1 (X). By Proposition 3.
. . . K) extends to a monomorphism M −→ K. uk ) : K(u1 . Suppose that L/K is separable. restricting to M K we obtain a monomorphism M −→ K. · · · . uk−1 )]. . . . Any ﬁnite extension L/K can be built up from a succession of simple extensions (3. For each k.1) and then using Theorem 2. K). . u is algebraic over E. K). If K E L. . and in the polynomial ring E[X] we have minpolyE. uk−1 )). so (M : L) =  MonoL (M.5.6(ii) which gives [L : K] = [K(u1 ) : K] [K(u1 . . K). Proof. K) be its restriction to L. (K(u1 . K) gives rise to a monomorphism β ◦ γ : M −→ K which extends β. . . Corollary. uk ) : K(u1 . 3. uk−1 )]. . . we have (L : K) = (K(u1 ) : K) · · · (L : K(u1 . K(u1 . . 3. . K). This follows from Proposition 3. . uk−1 )]. Then L/K is separable if and only if (L : K) = [L : K]. .u (X). . Therefore we have the desired formula (M : K) = (M : L)(L : K). each γ ∈ MonoL (M. K) has the form β = αL for some α ∈ MonoK (M. . then β −1 ◦ β (u) = β −1 (β(u)) = u. If L/K is a simple extension then by Propositions 3. .34 applied to the case L = K. Proof. The general result follows by building up L/K as a sequence of simple extensions as in (3. So we can use the following to compute (L : K) = (K(u1 . . Proof. choose any extension to a monomorphism β : K −→ K. Expressing L/K in terms of a sequence of simple extensions as in (3. .62 and 3.49. K) agrees with the number of extensions of β to a monomorphism M −→ K. hence β −1 ◦ β ∈ MonoL (M. uk ) : K(u1 . 3.1) K(u1 )/K.3. and therefore L/E is separable. . Let L/K be a ﬁnite extension. We have (L : K) = (L : E) (E : K) and [L : K] = [L : E] [E : K]. If K(u)/K is separable. u2 ) : K(u1 )] · · · [K(u1 . .u (X) is separable. . . . so is minpolyE.1). Then (M : K) = (M : L)(L : K). For α ∈ MonoK (M.39. then [K(u) : K] = (K(u) : K). β is an automorphism.72. then for any u ∈ L. . . (K(u) : K) =  Roots(minpolyK. by Proposition 3. so to verify that (L : K) = [L : K] it suﬃces to show that (L : E) = [L : E] and (E : K) = [E : K]. . . . . Since (L : K) =  MonoK (L. .70. . K) let αL ∈ MonoK (L. Let L/K be a ﬁnite extension. . MULTIPLICITY OF ROOTS AND SEPARABILITY 47 3.69. we need to show that the number of such α is always (M : L) =  MonoL (M.u . uk−1 )) divides [K(u1 . so the desired result follows. . so every element β ∈ MonoK (L.34 we know that this is true. . Clearly E/K is also separable. K). In eﬀect. each element of MonoK (L. uk−1 ). u2 )/K(u1 ). K). .71. L = K(u1 . Lemma. . As minpolyK. [L : K] = [K(u1 ) : K] · · · [L : K(u1 . . . Now for any extension β : M −→ K of β we can form the composition β −1 ◦ β : M −→ K.u (X)  minpolyK. So given β ∈ MonoK (L. . Proposition. By the Monomorphism Extension Theorem 3. Let L/K and M/L be ﬁnite extensions.u (X). Of course. notice that if u ∈ L. this shows that there is a bijection extensions of β to monomorphism a M −→ K ←→ MonoL (M. Conversely. For a ﬁnite simple extension K(u)/K. uk )/K(u1 . . K). . Proposition. Proof. uk ) : K). Then (L : K)  [L : K].
3. 3. Deﬁne the polynomial (of degree r) h(X) = p(w − tX) ∈ K(w)[X] ⊆ L[X]. Let L/K and M/L be ﬁnite extensions. We must show that for each u ∈ L. 3. an element u ∈ L is called a primitive element for the extension if L = K(u). hence v ∈ K(w) and so u = w − tv ∈ K(w). The Primitive Element Theorem 3. v) K(w) and therefore K(w) = K(u. which can only happen if r = s = 1. so none of the other vj is a zero of h(X). v). Then M/K is separable if and only if L/K and M/L are separable. Proof. . Let L/K be a ﬁnite extension of a ﬁeld of characteristic 0. vs . Definition.76. Suppose that the distinct roots of p(X) in K are u = u1 .16. but h(vj ) = p(ui ) = 0 for any j = 1 by construction of t. we ﬁnd that w = ui + tvj whenever j = 1. Corollary.73. By Corollary 3. For the converse. Then L has a primitive element. For the extensions K(u)/K and L/K(u) we have (L : K) = (L : K(u)) (K(u) : K) and [L : K] = [L : K(u)] [K(u) : K]. . By the separability assumption. by separability their greatest common divisor in K(w)[X] is a linear polynomial which must be X − v. q(X) ∈ K(w)[X] have exactly one common root in K. Proof. Thus (K(u) : K) = [K(u) : K] and so u is separable. r = deg p(X) and s = deg q(X).48 3. . ur . Let L/K be a ﬁnite separable extension. Since K is inﬁnite. If M/K is separable then [M : K] = (M : K) and so by Proposition 3. namely v. Then h(v) = p(u) = 0. u is separable. Since L is built up from a sequence of simple extensions it suﬃces to consider the case L = K(u. ALGEBRAIC EXTENSIONS OF FIELDS Now we can apply Lemma 3. since (M : L) [M : L] and (L : K) [L : K]. .75. . For a ﬁnite simple extension L/K.72 this implies that L/K and M/L are separable. [M : L][L : K] = (M : L)(L : K). This shows that K(u. Conversely.74. . This can only happen if [M : L] = (M : L) and [L : K] = (L : K). So we will assume that K is inﬁnite. q(X) ∈ K[X] be the minimal polynomials of u and v over K. 3. while the distinct roots of q(X) are v = v1 .6. v). Hence (L : K(u))(K(u) : K) = rs(L : K(u))(K(u) : K). we can choose an element t ∈ K for which u − ui t= vj − v whenever j = 1. Therefore M/K is separable. s for which [L : K(u)] = r(L : K(u)) and [K(u) : K] = s(K(u) : K). Now since the polynomials h(X). Proposition. there are some positive integers r. The case where K is a ﬁnite ﬁeld will be dealt with in Proposition 5. .70. By Proposition 3. hence [M : K] = [M : L][L : K] = (M : L)(L : K) = (M : K). Let p(X).69 to each of these intermediate separable simple extensions to obtain (L : K) = [L : K].71. if L/K and M/L are separable then [M : L] = (M : L) and [L : K] = (L : K). suppose that (L : K) = [L : K]. Theorem (Primitive Element Theorem). . Then L has a primitive element. . Then taking w = u + tv ∈ L.
.77. Example. however a ‘try it and see’ approach will often be suﬃcient. K). Proof. Consider 3 + i.√3+i (X). α3 = . √ 3. hence Now taking √ X 2 − 2 3X + 4 = minpolyQ(√3). √ √ There are four monomorphisms αk : Q( 3. Notice that √ √ √ √ ( 3 − i) ( 3 − i) 1 √ −1 √ ( 3 + i) = √ = = ( 3 − i) ∈ Q( 3 + i). There is a general phenomenon illustrated by Example 3. i) we ﬁnd that √ i ∈ Q( 3) R and / √ √ √ √ (X − ( 3 + i))(X − ( 3 − i)) = X 2 − 2 3X + 4 ∈ Q( 3)[X]. Let u ∈ K be separable over K.√3+i (X)  (X 4 − 4X 2 + 16) in Q[X]. αd are the elements of MonoK (K(u). 2 2 √ √ √ √ √ are both in Q( 3 + i).6.67 L/K is separable. so √ √ √ √ 3 − i)(X − 3 + i)(X + 3 − i)(X + 3 + i) = X 4 − 4X 2 + 16 ∈ Q[X]. and so minpolyQ. Then minpolyK. . i) = Q( 3 + i). . Then working over the subﬁeld Q( 3) Q( 3. THE PRIMITIVE ELEMENT THEOREM 49 Proof. α2 = i −→ −i i −→ i i −→ −i Then √ √ α2 ( 3 + i) = ( 3 − i). i) : Q] = [Q( 3. the polynomial (X − α1 (u)) · · · (X − αd (u)) ∈ K[X] is in K[X] and is irreducible therein. Since K(u) is separable then by Lemma 3.√3+i (X) = X 4 − 4X 2 + 16.77. i)/Q.52. α1 = id.77 we have √ √ √ √ [Q( 3. . d = deg minpolyK.78. . α4 = .3. √ √ (X 2 − 2 3X + 4)(X 2 + 2 3X + 4) = X 4 − 4X 2 + 16 ∈ Q[X]. 3. √ √ α4 ( 3 + i) = (− 3 − i). 3+1 4 ( 3 + i)( 3 − i) √ √ since ( 3 + i)−1 ∈ Q( 3 + i). √ √ √ Solution. i) −→ Q( 3. Find a primitive element for the extension Q( 3. To ﬁnd a primitive element we can always use the method suggested by the proof of Theorem 3. i) given by √ √ √ √ √ √ 3 −→ 3 3 −→ − 3 3 −→ − 3 . In Example 3. where α1 . √ √ √ Hence this polynomial is irreducible.√3+i (X) = 4. showing that Q( 3.75. Proposition. i) : Q( 3)][Q( 3) : Q] = 2 · 2 = 4. So we have [Q( 3 + i) : Q] = 4 and Q( 3 + i) = Q( 3. Thus we must have deg minpolyQ. In particular. i = (( 3 + i) − ( 3 − i)). we see that minpolyQ. K is inﬁnite and by Example 3. i). i) Q( 3 + i) and so Q( 3.u (X) = (X − α1 (u)) · · · (X − αd (u)).u (X) = [K(u) : K] = (K(u) : K). Since Q K. Hence √ √ √ 1 √ 1 √ 3 = (( 3 + i) + ( 3 − i)). (X − √ √ α3 ( 3 + i) = (− 3 + i).
Proof. Prove that AutQ (Q( 3 2. p4 (X) = X 4 +5X 3 +10X 2 +10X+5. Show that the polynomial g(X) = X p − T ∈ k(T )[X] is irreducible and has a multiple root in k(T ). θ(vk )) = K(v1 . . where v1 . . then by Remark 3. ALGEBRAIC EXTENSIONS OF FIELDS 3. 3. . vk ) = E. f (X) splits in E. Finding splitting subﬁelds E polynomials. . How does g(X) factor in k(T )[X]? . Since θ permutes the roots vj . ϕE = E if and only if ϕE E. . Suppose that E/K is normal. un ) = E Construct a polynomial by taking f (X) = minpolyK. u2 ) ··· K(u1 . Hence E is a splitting ﬁeld for f (X) over K.80. . so that E = K(v1 .u2 (X) · · · minpolyK. Now any monomorphism θ ∈ MonoK (E. If E is the splitting ﬁeld of a polynomial f (X) ∈ K[X] over K. o Exercises on Chapter 3 3. .1. . the symmetric group on 3 elements.81. .4.79.2. . 3. Let E/L and L/K be ﬁnite extensions. [Hint: for p4 (X). Definition. . as claimed in the = solution of Example 3. then E is the splitting ﬁeld of f (X) over L. . . .82. . Now suppose that E is a splitting ﬁeld for g(X) ∈ K[X]. .] 3. we have θE = θK(v1 . ζ3 )) ∼ S3 .80. . . it splits in E since by Lemma 3. . indeed these are known as Galois extensions. . . 3. K). Corollary. vk ) = K(θ(v1 ). E is generated by some of the roots of f (X). If E/K is a normal extension then whenever an irreducible polynomial p(X) ∈ K[X] has a root in E. .32. p2 (X) = X 6 −2. If ϕ ∈ MonoK (E.7.u1 (X) minpolyK. θ(vk ) which are also roots of g(X) and therefore lie in E (see Proposition 3. Normal extensions and splitting ﬁelds Let K be an algebraic closure for the ﬁeld K and let E/K K/K be a ﬁnite extension. K) must map these roots to θ(v1 ). . Then there is a sequence of extensions K K(u1 ) K(u1 .52 each pair of roots of p(X) is conjugate over K and one can be mapped to the other by a monomorphism K −→ K which must map E into itself. vk are the distinct roots of g(X) in E. These result makes it easy to recognize a normal extension since it is suﬃcient to describe it as a splitting ﬁeld for some polynomial over K.un (X). If E/K is normal then E/L is normal. 3. K).] √ 3. Also. A ﬁnite extension E/K is normal if and only if it is a splitting ﬁeld over K for some polynomial f (X) ∈ K[X]. E/K is normal if ϕE = E for every ϕ ∈ MonoK (E. . Prove Proposition 3. Remark. In Chapter 4 we will see that separable normal extensions play a central rˆle in Galois Theory. Proof. C over Q and determine [E : Q] for each of the following p3 (X) = X 4 +2. [Hint: work out the eﬀect of each automorphism on the three roots of the polynomial X 3 − 2. Theorem.38.34). consider p4 (Y − 1) ∈ Q[Y ]. p1 (X) = X 4 −X 2 +1.3.50 3. 3. . . . .2. Let k be a ﬁeld of characteristic char k = p > 0 and k(T ) be the ﬁeld of rational functions in T over k. vk ). Then by Remark 3.
then use induction on n to prove the general case L = K(u1 .] 3. Q( 3. [Hint: look for elements of high degree over Q.75. i)/Q. v).6. Let K be a ﬁeld.20: Let L/K be a ﬁnite extension. Prove the following converse of Proposition 3.] 3. Is such an extension always separable? 3. then L/K is simple.8.e. .. . i)/Q. of degree 2) extension E/K is normal.e.51 √ √ √ √ 3.7.5. Q( 4 3. Let f (X) ∈ Q[X] be an irreducible polynomial of odd degree greater than 1 and having only one real root u ∈ R. or use the method of proof of Theorem 3. L = K(w) for some w ∈ L. . . Show that every quadratic (i. Show that Q(u)/Q is not a normal extension. i.. 10)/Q. √ Find primitive elements for the extensions Q( 5. Q( 2. in each case ﬁnding it minimal polynomial over Q. [Hint: First deal with the case where L = K(u. . un ). i)/Q. If there are only ﬁnitely many subextensions F/K L/K.
.
Definition. 4.51. by the Monomorphism Extension Theorem 3. Definition.2)  Gal(E/K) = (E : K) = [E : K]. 4. in particular we will explain how these are related through the Galois Correspondence. A ﬁnite extension E/K is a (ﬁnite) Galois extension if it is both normal and separable. Similarly.1)  AutK (E) = (E : K) = [E : K].49. The elements of Gal(E/K) are called (Galois) automorphisms of E/K.1) implies (4. let K be a ﬁeld. there is a ﬁnite Galois extension E/K in which v is a conjugate of u if and only v is a conjugate of u over K in the old sense. v ∈ K. every automorphism 53 . Here is a slightly diﬀerent way to understand this. v ∈ E.CHAPTER 4 Galois extensions and the Galois Correspondence In this Chapter we will study the structure of Galois extensions and their associated Galois groups.3. hence gives rise to an automorphism ϕE : E −→ E. First notice that every element ϕ ∈ AutK (K.5 we know that for such a Galois extension E/K. Throughout the chapter. Notice that Equation (4. the group Gal(E/K) = AutK (E) is called the Galois group of the extension or the Galois group of E over K. 4.1. Definition. So there is a bijection = MonoK (E. We can also reformulate the notion of conjugacy introduced in Deﬁnition 3. Let E/K a ﬁnite Galois extension and u. K) ←→ AutK (E) and we have (4.2. if F/K is any ﬁnite normal extension with E F .1. every automorphism α ∈ AutK (E) extends to a monomorphism E −→ K ﬁxing elements of K. For a ﬁnite Galois extension E/K. K) restricts to a monomorphism E −→ K whose image is contained in E. K ~> ~~ ~~ ~~ ∼ = E ϕ / E ϕ E /K K Also. From Section 3. K) maps E into itself. [E : K] = (E : K) and also every monomorphism ϕ ∈ MonoK (E. we also say that v is a conjugate of u. It is easy to see that for u. hence restricts to an automorphism of E which will be denoted ϕE . Then v is conjugate to u if there is a ϕ ∈ Gal(E/K) for which v = ϕ(u). Galois extensions 4.
Roots(f. We have √ √ √ √ √ √ [Q( 2. α4 ←→ (1 4)(2 3).2. Solution. describe the Galois group Gal(Q( 2. and the following nontrivial elements of the Galois group together with the element identity α1 = id: √ √ √ √ √ √ √2 −→ −√2 √2 −→ √2 −→ −√2 √2 − 2 −→ √2 .4. . Recall that for an extension F/K and a polynomial f (X) ∈ K[X]. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE F θ : F −→ F restricts to an automorphism θE : E −→ E.54 4. − 2. 4. If F/K K/K is any ﬁnite normal extension with E then there is a surjective group homomorphism AutK (F. E). E 4. Using a primitive element u for the extension. − 2 − 3. 3)/Q) as a subgroup of the group of permutations of the roots of (X 2 − 2)(X 2 − 3) ∈ Q[X]. √ √ √ √ √ √ Solution. α3 ←→ (3 4). Proposition. F ) denotes the set of roots of f (X) in F . E). and after numbering them accordingly. It is often convenient to use these facts to interpret elements of the Galois group as permutations of the roots of some polynomial which splits over E. α3 = −√2 −→ −√2 . these automorphisms correspond to the following permutations in S4 expressed in cycle notation: α2 ←→ (1 2). K) −→ AutK (E. α2 = √ √3 −→ √3 −→ −√3 √3 −→ −√3 √3 − 3 −→ − 3 − 3 −→ − 3 −→ 3 3 √ √ √ √ Writing the roots in the list 2. K) we have (ϕF )F = ϕE . α3 ←→ (1 2)(3 4). 2 − 3. − 2 + 3. We have Q( 2. 3) = Q( 2 + 3) and the conjugates of u = 2 + 3 are √ √ ± 2 ± 3.5. Example. Hence E is a splitting ﬁeld for the minimal polynomial of any primitive element for E/K. Working with Galois groups F θ −→ θE . 3) : Q( 2)] [Q( 2) : Q] = 4. If E/K is a ﬁnite Galois extension. √ √ 4. 3. Then we know that E is a splitting ﬁeld for some polynomial over K since E/K is normal. F ) −→ AutK (E. Furthermore. We also know that E is a simple extension of K since E/K is separable.√ √ Example. Next we summarize the properties of Galois groups that can be deduced from what we have established so far. 4. Let E/K be a ﬁnite Galois extension. 3)/Q) as a subgroup of the group of permutations of the roots of minpolyQ. Describe the Galois group Gal(Q( 2. ϕ −→ ϕE F is a surjective group homomorphism. for ϕ ∈ AutK (K. − 3 and numbering them from 1 to 4. α4 ←→ (1 2)(3 4). 3) : Q] = [Q( 2. α4 = −√2 −→ √2 .6. The proof of the next result is left as an exercise. we ﬁnd the correspondences α2 ←→ (1 3)(2 4). Listing these as √ √ √ √ √ √ √ √ 2 + 3.u (X) ∈ Q[X]. this minimal polynomial has degree [E : K]. then the function AutK (K.
ζ8 . Example. and observe that √ √ √ √ 2 −→ −√2 2 −→ −√2 √ √ − 2 −→ − 2 −→ 2 2 ←→ (1 2)(3 4). Suppose that E is the splitting ﬁeld of a separable irreducible polynomial f (X) ∈ K[X] of degree n. Then Gal(Q(f (X))/Q) ∼ S3 .ζ (X) = Φ8 (X) = X 4 + 1 3 5 7 in the order ζ8 .4. i) Q(ζ8 ).6. . we ﬁnd that these automorphisms correspond to the following permutations in S4 : α ←→ (1 2)(3 4). special arguments can sometimes be exploited. i ∈ Q(ζ8 ). the Galois group Gal(Q(ζ8 )/Q) = Gal(Q( 2. If we list the roots of the minimal polynomial minpolyQ. While it can be hard to determine Galois groups in general. i) is the splitting ﬁeld of f (X) = (X 2 − 2)(X 2 + 1) over √ √ Q. 2 2 √ √ √ we√ easily ﬁnd that 2. Recall that an action of a group G on a set X is transitive if for every pair of elements x.8. Then the following are true. γ ←→ (1 4)(2 3). If we order the roots u1 . (1 2)(3 4). ζ8 . −i. Let E/K be a ﬁnite Galois extension. (ii) Gal(E/K) can be identiﬁed with a subgroup of the group of permutations of Roots(f. (i) Gal(E/K) acts transitively and faithfully on Roots(f.9. i) : Q] = 4.7. Notice that Q( 2. there is an element z ∈ X such that hz = z. . Theorem. the action is faithful or eﬀective if for every nonidentity element h ∈ G. E). Recollection. there is an element g ∈ G such that y = gx (so there is only one orbit).√ hence Q( 2. − 2. So the Galois group Gal(Q(ζ8 )/Q) corresponds to {id. (1 2)(3 4)} S4 . Now list the roots of f (X) in the order 2. E). Suppose that f (X) = X 3 + aX 2 + bX + c ∈ Q[X] is an irreducible cubic and that f (X) has only one real root. ←→ (1 2). i)/Q) corresponds to the subgroup {id. we have Q( 2. un then Gal(E/K) can be identiﬁed with a subgroup of Sn . 7 γ : ζ8 −→ ζ8 . ζ8 . Since [Q( 2. y ∈ X. in practise it is often easier to use a not necessarily irreducible polynomial to determine and work with a Galois group. (1 2). . α: β: i −→ −i i −→ i −i −→ i −i −→ −i √ √ √2 −→ √2 − 2 −→ − 2 γ: i −→ −i ←→ (3 4). The Galois extension Q(ζ8 )/Q has degree [Q(ζ8 ) : Q] = 4 and it has the following automorphisms apart from the identity: 3 α : ζ8 −→ ζ8 . i. 1 1 ζ8 = √ + √ i. −i −→ i √ In this description. (iii)  Gal(E/K) divides n! and is divisible by n. (1 4)(2 3)} Noticing that S4 .10. As we have seen in Examples 4. 4. β ←→ (1 3)(2 4). = . WORKING WITH GALOIS GROUPS 55 4. . i) = Q(ζ8 ).2.5 and 4. 4. 5 β : ζ8 −→ ζ8 . (3 4). 4. (1 3)(2 4). Example.
Hence Gal(Q(f (X))/Q) is isomorphic to a subgroup of S3 and so ∼ S3 since the orders agree. γ(u + v) = γ(u) + γ(v) = u + v. E is the ﬁxed subﬁeld of Γ. Notice that f (X) is unchanged by applying any γk to its coeﬃcients since the roots γj (u) are permuted by γk . By Proposition 3. Remark. so by the Primitive Element Theorem 3. Hence. 4. f (c− ) > 0 or f (c+ ). γ(u) = u}. where h = Γ. 4. E Γ E is a subﬁeld of E containing K. if u = 0. Γ [E : E Γ ] =  Gal(E/E Γ ) = Γ. Given a Galois extension E/K.73. 4. γ(u−1 ) = γ(u)−1 = u−1 . . Such examples occur when the cubic polynomial f (X) has local maximum and minimum at real values c+ and c− with f (c+ ). It also contains an element of order 2 obtained by restricting complex conjugation to Q(f (X)). say E = E Γ (u). so it corresponds to the transposition (2 3). f (X) ∈ E Γ [X]. This happens for example with f (X) = X 3 − 3X + 3 which has local extrema at ±1 and f (1) = 1. we can assume that this is (1 2 3).3. γh . Consider the polynomial of degree h f (X) = (X − u)(X − γ2 (u)) · · · (X − γh (u)) ∈ E[X]. For Γ Gal(E/K).11. Proof. Proof. u3 be the remaining complex roots. f (−1) = 5. γ2 . so Lagrange’s Theorem implies that Γ divides Γ ).14.75 it is simple. Also. u3 ) and in fact [Q(f (X)) : Q] = 6 since [Q(f (X)) : Q]  6 and u2 ∈ Q(u1 ) / R. Gal(Q(f (X))/Q) = The Galois group Gal(Q(f (X))/Q) contains an element of order 3 which corresponds to a 3cycle when viewed as a permutation of the roots u1 . E Γ EΓ. u3 . we will next study subextensions L/K E/K and subgroups Γ Gal(E/K). Consider the subset of elements of E ﬁxed by Γ. the extensions E/E Γ and E Γ /K are separable. so K 4. We know that E/E Γ is separable. this ﬁxes u1 and interchanges u2 . if t ∈ K then γ(t) = t. γ(uv) = γ(u)γ(v) = uv.12. For u. we will identify its Galois group. we have Gal(E/E Γ ) = Γ and the equations [E Γ : K] =  Gal(E/K) . f (c− ) < 0. This shows that [E : E Γ ] = [E Γ (u) : E Γ ] h = Γ. We also have Q(f (X)) ∩ R = Q(u1 ). Now let the distinct elements of Γ be γ1 = id. focusing on the relationship between objects of these types. v ∈ E Γ and γ ∈ Γ. E Γ = {u ∈ E : ∀γ ∈ Γ. Definition. Then Q(f (X)) = Q(u1 . Lemma. In fact we have  Gal(E/E 4. Let u1 ∈ R be the real root of f (X) and let u2 . Notice that [E : E Γ ] = (E : E Γ ) =  Gal(E/E Γ ). u3 . .56 4. . . Proposition.13. so this is a Galois extension. u2 . GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE Proof. E/E Γ is also normal. Subgroups of Galois groups and their ﬁxed ﬁelds Let E/K a Galois extension and suppose that Γ Gal(E/K). . Now each element of Gal(E/E Γ ) is also an element of Gal(E/K) and Gal(E/E Γ ) Gal(E/K). Notice that by deﬁnition Γ Gal(E/E Γ ). Finally. u2 .
then for any γ ∈ Gal(E/K) and α ∈ Gal(E/L).8(i). Then E/L is also a Galois extension whose Galois group Gal(E/L) is sometimes called the relative Galois group of the pair of extensions E/K and L/K. 4.e. then there is a group isomorphism Gal(E/K)/ Gal(E/L) − Gal(L/K). for all α ∈ Gal(E/L) and β ∈ Gal(E/K). hence γ(u) ∈ E Gal(E/L) = L. so the above argument shows that L/K is normal. SUBFIELDS OF GALOIS EXTENSIONS AND RELATIVE GALOIS GROUPS 57 Since Γ Gal(E/E Γ ). This shows that ϕ−1 θϕ ∈ Gal(E/L). Hence for every ϕ ∈ Gal(E/K). we have βαβ −1 ∈ Gal(E/L). Then L = E Gal(E/L) . hence u ∈ E Gal(E/L) . By Theorem 4. Combining these two inequalities we obtain [E : E Γ ] =  Gal(E/E Γ ) = Γ = h and therefore Γ = Gal(E/E Γ ). E L and therefore E We need to understand when Gal(E/L) Gal(E/K) is actually a normal subgroup. → ∼ = α Gal(E/L) −→ αL .17.. K L E). so for every θ ∈ Gal(E/L). Hence we can restrict α to an automorphism of L. The next result explains the connection between the two uses of the word normal which both ultimately derive from their use in Galois theory. ϕ(v) ∈ L. The following is immediate. Conversely. (i) Suppose that Gal(E/L) Gal(E/K). Let L/K E/K. i. every monomorphism L −→ K ﬁxing K extends to a monomorphism E −→ K which must have image E.15. i. (ii) If α ∈ Gal(E/K).. we also have h = Γ  Gal(E/E Γ ) = [E : E Γ ]. αL (u) = α(u). Proof. 4. Subﬁelds of Galois extensions and relative Galois groups Let E/K a Galois extension and suppose that L/K E/K (i.4. there is an automorphism θ ∈ Gal(E/L) such that θ(u) = u. since γ −1 αγ ∈ Gal(E/L). Clearly L E Gal(E/L) . Lemma. and its order is  Gal(E/L) = [E : L].e. Now if u ∈ L. The relative Galois group of the pair of extensions L/K E/K is a subgroup of Gal(E/K).4. Suppose that u ∈ E − L. (ii) If L/K is normal and hence a Galois extension. Proposition. γ(u) ∈ E satisﬁes αγ(u) = γ(γ −1 αγ(u)) = γ(u). Proof.16. if L/K is normal.e.. . Gal(E/L) Gal(E/K). By the Monomorphism Extension Theorem 3. αL : L −→ L. Let E/K be a ﬁnite Galois extension and L/K E/K. ϕ Gal(E/L)ϕ−1 = Gal(E/L). 4. This shows that / Gal(E/L) Gal(E/L) = L. which shows that Gal(E/L) Gal(E/K). then αL = L since L/K is normal. θ(ϕ(v)) = ϕ(v) and therefore ϕ−1 θϕ(v) = v.49. then for every ϕ ∈ Gal(E/K) and v ∈ L.4. (i) The relative Galois group Gal(E/L) of the pair of extensions L/K E/K is a normal subgroup of Gal(E/K) if and only if L/K is a normal extension. 4. Proposition.
Corollary. Proof. F(E/K) = the set of all subextensions L/K of E/K. if Γ1 .18. ΘE/K : S(E/K) −→ F(E/K). We will use the following notation. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE Then αL is the identity function on L if and only if α ∈ Gal(E/L). by Proposition 4. let S(E/K) = the set of all subgroups of Gal(E/K). ΦE/K F(E/K) o ΘE/K / S(E/K) Under this correspondence. ΘE/K (ΦE/K (L)) = ΘE/K (Gal(E/L)) = E Gal(E/L) = L. Hence ΦE/K (L2 ) ΦE/K (L1 ) and so ΦE/K reverses order.16 that for an extension L/K in F(E/K). Let L1 /K.19.  Gal(E/K)/ Gal(E/L) = 4. ΦE/K (L) = Gal(E/L). Let E/K be a ﬁnite Galois extension.14 for H ∈ S(E/K) we have ΦE/K (ΘE/K (Γ)) = ΦE/K (E Γ ) = Gal(E/E Γ ) = Γ. 4. The Galois Correspondence and the Main Theorem of Galois Theory We are now almost ready to state our central result which describes the Galois Correspondence associated with a ﬁnite Galois extension. Similarly. normal subextensions of E/K correspond to normal subgroups of Gal(E/K) and vice versa. Then Gal(E/L2 ) Gal(E/L1 ) since L1 ⊆ L2 and so if α ∈ Gal(E/L2 ) then α ﬁxes every element of L1 . [E : L] Hence this homomorphism is an isomorphism. Γ2 ∈ S(E/K) and Γ1 Γ2 . Proof. 4. so is F(E/K). Since the set S(E/K) is ﬁnite.5. ΘE/K (Γ) = E Γ . There is an immediate consequence of the Main Theorem 4. α −→ αL is a group homomorphism whose kernel is Gal(E/L). Then the functions ΦE/K and ΘE/K are mutually inverse bijections which are orderreversing. This shows that ΦE/K and ΘE/K are mutually inverse and so are inverse bijections. Hence ΘE/K reverses order.18 which is closely related to Proposition 3. Each of these sets is ordered by inclusion. Since every subgroup of a ﬁnite group is a ﬁnite subset of a ﬁnite set. S(E/K) is also a ﬁnite set. L2 /K ∈ F(E/K) satisfy L1 /K L2 /K. We know from Proposition 4. Then there are only ﬁnitely many subextensions L/K E/K. Theorem (Main Theorem of Galois Theory).20. Let E/K be a ﬁnite Galois extension. then E Γ2 E Γ1 since if w ∈ E Γ2 then it is ﬁxed by every element of Γ1 (as Γ1 is a subset of Γ2 ). Also. . Thus we obtain an injective homomorphism Gal(E/K)/ Gal(E/L) −→ Gal(L/K) for which [E : K] = [L : K] =  Gal(L/K). For a ﬁnite Galois extension E/K. Deﬁne two functions by ΦE/K : F(E/K) −→ S(E/K).58 4. It is easy to see that the function Gal(E/K) −→ Gal(L/K).
α2 .1.4. we indicate the subextensions in a diagram with a line going upwards indicating an inclusion.30. (1 3 2)}. It = is useful to make this isomorphism explicit. α1 . In eﬀect. Example. α1 . = = √ √ 2 3 3 Gal(E/Q( 2 ζ3 )) = {id. α1 = (1 2).20.5. (1 2)}. each of which permutes these 3 roots in the following ways given by cycle notation: α0 = (2 3). E = Q( 2. ζ3 )/Q) ∼ S3 . . α2 extend to automorphisms of E. √ 3 Gal(E/Q(ζ3 )) = {id. α0 } ∼{id. ζ3 ) H ii eeee k ii eeeeeekkkkkk e eee ii 2 kk eeee2 e ii kk 2 eeee ee kkk ii eeeeee √ 2 √ √ eee 3 ii ii Q( 3 2 ζ3 ) Q( 3 2 ζ3 ) Q( 3 2) ii N N rr I ii rr rr ii rr ii rr 3 rr3 rr 3 Q(ζ3 ) rr k Q rr kkk 2 kk rr kkk rr kkk rr kkk kkk QS ΦE/K √ 3 k 2 kkk ww kkk www k kkk ww w kkk kk ww 3 kkk 3 w w Gal(E/Q(ζ3 )) 3 kk ww kkk ww kkk yy ww kkk yy w kk ww yy kkk y √ √ kkk √w w yy 3 3 3 2 )) 3yyy Gal(E/Q( 2 ζ3 y Gal(E/Q( 2)) Gal(E/Q( 2 ζ3 )) yy yy 2 yy y 2 2 yy Gal(E/Q) {id} √ Figure 4. (2 3)}. (1 3 2)} S3 and so Q(ζ3 )/Q is a normal extension. 3 2 ζ3 which we number in the order they are listed.38. α0 . α2 = (1 3). α2 = (1 3 2). We ﬁnd that α1 = (1 2 3). (1 2 3). 3 2 ζ3 . = = Notice that {id. We can also do this with the subgroups of the Galois group Gal(E/K) with labels indicating the index of the subgroups. α2 } ∼{id. α1 } ∼{id. Gal(E/Q( 2 ζ3 )) = {id. (1 2 3). (1 3)}. these are 3 2. Gal(E/Q( 2)) = {id. α1 . the Galois Correspondence inverts these diagrams. The Galois Correspondence for E = Q( 3 2. THE GALOIS CORRESPONDENCE AND THE MAIN THEOREM OF GALOIS THEORY 59 When dealing with a ﬁnite Galois extension E/K. the Galois group here is Gal(Q( 3 2.1 shows the Galois Correspondence for the extension of Example 3. α2 } ∼{id. First take the √ roots of the polynomial X 3 − 2 for 3 √ √ 2 which E is the splitting ﬁeld over Q. Figure 4. 4. Then the monomorphisms id. Of course Q(ζ3 ) is the splitting ﬁeld of X 3 − 1 over Q. ζ3 )/Q √ As noted at the end of Example 3.
where we view the latter as permuting the roots of f (X). (ii) Here  ( )E/Q  = 2. 2 C R ∞ nE nnn nnn nn nnn 2 nnn ER e ee ee ee e ∞ Q We will usually write ( ) rather than ( )E/Q when no confusion seems likely to result. = hence. But ( ) permutes the roots of this minimal polynomial. Galois extensions inside the complex numbers and complex conjugation When working with Galois extensions contained in the complex numbers it is often useful to make use of complex conjugation as an element of a Galois group. z = z if and only if z ∈ R. For each pair i. then ( )E/Q = {id. i). Setting ER = R ∩ E. exactly one of the inequalities σ(i) < σ(j) or σ(j) < σ(i) must .22. Complex conjugation ( ) : C −→ C restricts to an automorphism of E over Q. Proof. Let u ∈ E. ( )E/Q } ∼ Z/2. Furthermore. so all of its complex roots lie in E. the Galois group of its splitting ﬁeld E over K can naturally be thought of as a subgroup of the symmetric group Sn .u (X) ∈ Q[X] splits over E. we have Q ER E.60 4. Example. (i) ( )E/Q agrees with the identity function if and only if ER = E. Proposition. We ﬁrst recall an interpretation of the sign of a permutation σ ∈ Sn . As E/Q is normal. and we easily see that √ Q(ζ8 )R = Q( 2). Galois groups of even and odd permutations We have seen that for a monic separable polynomial f (X) ∈ K[X] of degree n. Let E/Q be a ﬁnite Galois extension with E/Q C/Q. 4. Therefore ( ) maps E into itself. j with 1 i<j n. (i) For z ∈ C. and E ( )E/Q = {u ∈ E : u = u} = ER .7. Consider the cyclotomic extension Q(ζ8 )/Q where 1 1 ζ8 = eπi/4 = √ + √ i. 4. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE 4. [Q(ζ8 ) : Q] = 4. ER = E ( )E/Q and [E : ER ] = 2. ( )E/Q : E −→ E.6. It is reasonable to ask when Gal(E/K) An rather than just Gal(E/K) Sn . 2 2 From Example 4. 4.21.9 we know that √ Q(ζ8 ) = Q( 2. sgn σ = ±1. (ii) If ER = E. minpolyQ.
. . Remark. . . f (X)). . . . . . . . . . 0 . . . . . . . as we have assumed that f (X) is separable. . . .. . . . . . . . . .. . . . . . . . 0 . . . . . n = 2: n = 3: Discr(a0 + a1 X + X 2 ) = −4a0 + a2 . . . . . . . . . . . .. . .4. 3X 2 + p) q 0 = (−1) p 0 0 p q 0 p 0 0 p 3 0 p 1 0 0 3 0 0 1 0 = −4p3 − 27q 2 . . . 0 . . . j−i Note that this is sometimes used as the deﬁnition of sgn σ. . . . bn 0 . .. . Definition. . . . For polynomials p(X) = a0 + a1 X + · · · + am X m . . . am (4. . . .. . . . . . their resultant is the (m + n) × (m + n) determinant (with n rows of ai ’s and m rows of bi ’s) . . (4. . . There is an explicit formula for computing Discr(f (X)) is terms of its coeﬃcients. . a1 a0 . . . . . . It is easily veriﬁed that the righthand side of the following equation must have value ±1 and so (4. . . . . .. . . . . . . . . . . . am 0 . . . . .. .. . 4. . . 0 a0 a1 . . . b0 b1 . a0 0 . 0 1 2 2 1 . . .. . . 0 a1 . .. . . . . . . . . . .. . . The discriminant of f (X) is Discr(f (X)) = (uj − ui )2 ∈ E. . . . . . . q(X)) = . . . . un ∈ E are the roots of f (X). . . . . 0 b0 b1 . . . . . . . . . .... . .. . . . . . . . . . Here u1 . . .. .24. . . . . . 1 Discr(a0 + a1 X + a2 X 2 + X 3 ) = −27a2 + 18a0 a1 a2 + a2 a2 − 4a3 a0 − 4a3 . 0 0 b0 b1 . .4) Res(p(X). . . . . . . . . . . . .23. . .7. . .5) So for example. . . . . .. . bn Then if f (X) is monic with d = deg f (X). .. . . . . . . . . . .. . Here are some low degree examples of discriminants obtained with the aid of Maple. am 0 .. . . . Discr(X 3 + pX + q) = (−1)3 Res(X 3 + pX + q. . . .. . . . . . . 1 i<j n Notice that Discr(f (X)) = 0 since ui = uj if i = j. . . 0 . . .. . . . . . Suppose that f (X) factorizes over E as n f (X) = (X − u1 ) · · · (X − un ) = i=1 (X − ui ). .. . . . . . . . q(X) = b0 + b1 X + · · · + bn X n . . . . . .. . . the ui are distinct. . . . . bn 0 . . . . . . .. . . 4. . GALOIS GROUPS OF EVEN AND ODD PERMUTATIONS 61 hold and the ratio (σ(j) − σ(i))/(j − i) is either positive or negative. .. .. . . 0 3 Discr(f (X)) = (−1)d(d−1)/2 Res(f (X).3) sgn σ = 1 i<j n σ(j) − σ(i) . . . .. .
σ(Discr(f (X))) = Discr(f (X)).6). σ(δ(f (X))) = sgn σ δ(f (X)) = ±δ(f (X)). + 2000a2 a3 a2 − 900a1 a3 a2 − 2500a3 a4 a1 − 50a2 a2 a2 − 900a4 a3 a2 − 27a4 a4 − 3750a3 a2 a3 0 1 3 0 0 0 4 1 2 0 4 1 0 So for example. Of course  Gal(E/K)/ Gal(E/K) ∩ An  = 2. Hence Discr(f (X)) ∈ E Gal(E/K) = K. Discr(X 5 + a4 X 4 + a0 ) = a3 (3125a0 + 256a5 ).6) 1 i<j n (uσ(j) − uσ(i) ) = sgn σ (uj − ui ) = (±1) 1 i<j n (uj − ui ). Since Now let δ(f (X)) = E Gal(E/K) = K.25. For every σ ∈ Gal(E/K). (uσ(j) − uσ(i) )2 = Now for each pair i. σ(uj − ui ) = uσ(j) − uσ(i) . 1 0 Sn . On the other hand. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE n = 4: Discr(a0 + a1 X + a2 X 2 + a3 X 3 + X 4 ) = 18a3 a3 a2 − 6a2 a2 a0 − 192a3 a1 a2 − 27a4 1 3 1 0 1 + 144a2 a2 a2 + 144a0 a2 a2 + 256a3 − 4a3 a3 − 128a2 a2 + 16a4 a0 − 4a3 a2 3 0 1 0 3 1 2 0 2 2 1 + 18a3 a1 a2 a0 − 80a3 a1 a2 a0 − 27a4 a2 + a2 a2 a2 − 4a3 a2 a0 . For σ ∈ Gal(E/K) σ(Discr(f (X))) = 1 i<j n Discr(X 5 + a1 X + a0 ) = 256a5 + 3125a4 .3) (4. . we have Discr(f (X)) ∈ K. By Equation (4. and by Equation (4. 3 2 3 0 2 3 1 2 3 n = 5: Discr(a0 + a1 X + a2 X 2 + a3 X 3 + a4 X 4 + X 5 ) = 2250a4 a2 a3 − 36a0 a3 a3 − 128a2 a4 3 0 4 1 3 1 + 356a3 2 a2 2 a4 a1 a0 + 560a3 a2 a2 a2 − 2050a3 a2 a2 a4 a1 − 80a2 a2 a4 a1 3 − 630a3 3 a2 a4 a0 2 2 4 0 0 3 + 825a2 a2 a2 + 16a3 a3 a0 + 2000a2 a2 a3 − 6a2 a2 a3 − 128a2 a4 a2 + 16a4 a3 a0 − 4a3 a3 a2 3 2 0 3 2 4 0 2 4 1 2 4 0 2 4 2 4 1 + 108a5 a2 + 108a5 a0 − 746a3 a2 a0 a4 2 a1 2 − 27a4 a2 + 256a5 a3 − 4a3 a2 a2 + 144a3 a2 a3 3 0 2 2 1 4 0 3 2 1 2 1 + 144a2 a4 a3 + 3125a4 + 256a5 − 72a4 a2 a1 a0 + 18a3 a2 a3 a3 + 560a2 a2 a0 a2 + 16a4 a3 4 1 0 1 3 4 1 3 1 3 1 + 18a3 a2 3 a4 a1 2 − 72a3 a2 4 a4 a0 + 144a3 2 a2 a4 3 a0 2 − 192a4 4 a1 a3 a0 2 − 630a3 a2 3 a1 a0 + 24a2 3 a4 2 a1 a0 + a3 2 a2 2 a4 2 a1 2 − 6a4 3 a1 2 a3 2 a0 − 80a3 a2 2 a4 3 a1 a0 − 4a3 2 a2 3 a4 2 a0 + 2250a1 a2 a2 − 1600a3 a3 a3 − 192a4 a4 a2 − 1600a0 a3 a2 − 4a3 a3 a2 − 27a4 a2 a2 2 0 4 0 1 1 3 1 4 3 4 0 + 1020a4 2 a3 2 a0 2 a1 + 18a3 3 a2 a4 2 a0 a1 + 160a2 a4 3 a0 2 a1 + 144a2 a4 4 a0 a1 2 + 24a4 a1 2 a3 3 a0 + 1020a0 a4 a2 2 a1 2 + 160a0 a4 a1 3 a3 . we have 1 i<j n 2 (uσ(j) − uσ(i) ) . (uj − ui ) ∈ E. Now consider the eﬀect of σ ∈ Gal(E/K) on δ(f (X)) ∈ E. so the square roots of Discr(f (X)) are ±δ(f (X)). Proof. 1 i<j n Hence σ(Discr(f (X))) = Discr(f (X)). this means that sgn σ = 1. j with i < j.62 4. If δ(f (X)) ∈ K. 0 4 4. Proposition. 1 i<j n Then δ(f (X))2 = Discr(f (X)). if δ(f (X)) ∈ K then / K(δ(f (X))) = E Gal(E/K)∩An .
Sn is contained in An if and only if 4. v) which contains the quadratic extension Q(δ)/Q. Since deg f (X) = 4.27.8.40 and 6. 4.42 we obtain Discr(X 5 − 35X 4 + 7) = −4611833296875 = −33 · 56 · 74 · 29 · 157. This element must also have the eﬀects σ(u) = ±v and σ(v) = ±u.28. Example. = 2 − 4b) is a square in Q then Gal(Q(f (X))/Q) ∼ D . Kaplansky’s Theorem In this section we give a detailed account of the Galois theory of irreducible rational polynomials f (X) = X 4 + aX 2 + b ∈ Q[X]. √ √ δ(X 5 − 35X 4 + 7) = ±53 · 3 · 72 · 3 · 29 · 157 i = ±18375 13659 i ∈ Q. 2 v2 = (−a − δ) . This shows that E = Q(u) and we have the following Galois tower. The Galois group Gal(E/K) Discr(f (X)) is a square in K. we ﬁnd that the roots of g(X) are (−a±δ)/2 ∈ Q. Given u we might as well choose v so that σ(u) = v. 4. Let f (X) = X 4 + aX 2 + b ∈ Q[X] be irreducible. Theorem (Kaplansky’s Theorem). (uv)2 = u2 v 2 = 4 4 hence uv is a square root of b which is in Q. where u2 = (−a + δ) .8. Let g(X) = X 2 + aX + b ∈ Q[X]. KAPLANSKY’S THEOREM 63 4. Setting c = uv ∈ Q.4. E = Q(u) 2 Q(δ) 2 Q In particular [E : Q] = 4 =  Gal(E/Q). since E is obtained by at most 3 successive quadratic extensions we also have [E : Q]  8. (iii) If neither b nor b(a = 8 Proof. For the polynomials of Examples 6.26. / / Then the roots of f (X) are ±u. ±v. / Discr(X 5 + 20X + 16) = 1024000000 = 216 · 56 . = (ii) If b(a2 − 4b) is a square in Q then Gal(Q(f (X))/Q) ∼ Z/4. (i) We have 4b a2 − d = = b. (i) If b is a square in Q then Gal(Q(f (X))/Q) ∼ Z/2 × Z/2. Notice that g(X) must be irreducible since otherwise f (X) would factorize. Hence there is an element σ ∈ Gal(E/Q) for which σ(δ) = −δ. The following result describes the Galois groups that occur and the proof introduces some useful computational techniques. we ﬁnd that v = c/u ∈ Q(u). . Setting d = (a2 − 4b) ∈ Q and taking δ to be a square root of d (so δ ∈ Q). Gal(Q(δ)/Q) = Gal(E/Q)/N. Notice that for the Galois extension Q(δ)/Q there must be a normal subgroup N Gal(E/Q) with Q(δ) = E N . Proposition. we also have 4  [E : Q]. 2 so the splitting ﬁeld of f (X) over Q is E = Q(u. δ(X 5 + 20X + 16) = ±28 53 ∈ Q. In fact. hence (a2 − 4b) is not a square in Q.
This shows that Gal(Q(f (X))/Q) = Gal(E/Q) = {id. and so pq = 0. στ } ∼ Z/2 × Z/2 = the Klein 4group. We now have c c σ(u) = . σ 2 σ 3 } ∼ Z/4 = a cyclic group of order 4. Suppose that uv ∈ Q(δ). Q(δ) 2 Q with [E : Q] = 4 =  Gal(E/Q). Thus we have Q(uv) ∩ Q(δ) = Q. then (uvδ)2 = u2 v 2 d = bd. so uv ∈ E is a square root of b in E. implying that bd was a square in Q. By an easy calculation we ﬁnd that (uv)2 = b. c c = − = −u. given u we might as well choose v so that σ(u) = v. u u These satisfy σ 2 = τ 2 = (στ )2 = id = the identity. A similar discussion shows that Q(uvδ) ∩ Q(δ) = Q = Q(uvδ) ∩ Q(uv). GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE There is also an element τ ∈ N for which τ (u) = −u and we also have τ (v) = −v. σ(uδ) vδ .64 4. τ σ(δ) = στ (δ) = −δ. σ. This shows that E = Q(u. τ σ(u) = στ (u) = − . We cannot have q = 0 since this would imply that b was a square in Q. hence we might as assume that σ(v) = u since if necessary we can replace our original choice by τ σ. = (ii) If bd is a square in Q. v) = Q(u) and again we have a Galois tower E = Q(u) 2 στ = τ σ. σ. τ. which is a square in Q. then uv = p+qδ for some p. Notice that σ(v) = so σ 2 (u) = −u. By squaring we obtain b = (p2 + q 2 d) + 2pqδ. Notice that if σ(v) = −u then easy calculation shows that τ σ(v) = στ (v) = u. so we can write uvδ = c ∈ Q or equivalently v = c/(uδ) ∈ Q(u) since Q(δ) Q(u). b and bd are not squares in Q. This shows that Gal(Q(f (X))/Q) = Gal(E/Q) = {id. if p = 0 then b = q 2 d and so bd = (qd)2 . q ∈ Q. τ (u) = −u. = (iii) Suppose that d. Since Q(δ)/Q is Galois there is an element σ ∈ Gal(E/Q) with σ(δ) = −δ and this has the eﬀect σ(u) = ±v.
ασ. Setting σ = αβ we ﬁnd σ(u) = −v and σ(v) = u. E = Q(u. Then σ 2 = γ and σ has order 4. Therefore we have Gal(Q(f (X))/Q) = Gal(E/Q) ∼ D8 . = . = and [E : Q] = 8. = Gal(Q(X 4 + 4X 2 + 2)/Q) ∼ Z/4. D8 . v) 2 r 2 rrr 2 rrr rrr Q(uv. We have the following Galois groups: Gal(Q(X 4 + 1)/Q) ∼ Z/2 × Z/2. We must have either β(u) = −u or β(v) = −v. σ. σ −1 . By renaming −v to v if necessary. ασα = βσβ = σ −1 . δ) www www2 www ww Q(δ) vvv vvv vvv 2 vvv Q(uv) Q ppp ppp 2 ppp 2 ppp Q(uvδ) 4. ασ −1 form a group isomorphic to the dihedral group of order 8. δ) www www www ww Q(δ) vvv vvv vvv 2 vvv Q(uv) Q ppp ppp 2 ppp 2 ppp Q(uvδ) Choose α ∈ Gal(E/Q(uv)) Gal(E/Q) so that α(δ) = −δ. Choose γ ∈ Gal(E/Q(δ. KAPLANSKY’S THEOREM 65 So we have a Galois tower which includes the following subﬁelds. Also. Notice that γ 2 = id.4. Example. v) r rrr rrr rrr Q(uv.8. Notice that α2 = id. Notice that β 2 = id. γ.29. α. The eight elements id. so by interchanging ±δ if necessary we can assume that β(u) = −u and β(v) = v. Then we must have γ(v) = −v since γ(uv) = uv. = Gal(Q(X 4 + 2X 2 + 2)/Q) ∼ D8 . we may assume that v = α(u) and so u = α(v). uv)) Gal(E/Q) so that γ(u) = −u. The corresponding Galois tower is E = Q(u. αγ. Choose β ∈ Gal(E/Q(δ)) Gal(E/Q) with β(uv) = −uv.
β 2 = ι.1. (b) Suppose that K contains p distinct pth roots of 1. (a) Show that f (X) has p distinct roots in L. If δ = (w1 − w2 )(w3 − w3 )(w1 − w3 ). 3 − X − 1)/K for K = Q. Let L/K where L is a splitting ﬁeld for f (X) over K. βα2 .6. GALOIS EXTENSIONS AND THE GALOIS CORRESPONDENCE Exercises on Chapter 4 4. (b) Let D8 be the dihedral group with the eight elements ι. w3 are the distinct roots of g(X) in E and let ∆ = (w1 − w2 )2 (w2 − w3 )2 (w1 − w3 )2 ∈ E. and hence ∆ ∈ K. Use Kaplansky’s Theorem 4. Show that either f (X) is irreducible over K or it factors into p distinct linear factors over K. Let K be a ﬁeld for which char K = 2. under what conditions is this always true? (b) If g(X) = X 3 + aX + b ∈ K[x] is irreducible and E = K(g(X)) is its splitting ﬁeld over K. 4. Find the Galois groups for each of the following extensions: √ √ √ √ Q(X 3 − 10)/Q. explain why Gal(E/K) is isomorphic to one of the groups S3 or A3 . deduce that a. Let K be a ﬁeld with char K = p.] A3 S3 if δ ∈ K. w2 . (a) Show that every nonabelian ﬁnite group has order at least 6. If f (X) is monic. Q(i). √ √ √ 3 − X − 1)/Q( 23 i). βα3 satisfying α4 = ι. βα. Find all the normal subgroups of D8 . Suppose that 0 = a ∈ K and f (X) = X p − a ∈ K[X]. Q( 5).3. Let p > 0 be a prime. If u ∈ L is one such root. β. describe the remaining roots and show that L contains p distinct pth roots of 1. prove that the splitting ﬁeld of f (X) over K is a ﬁnite Galois extension of K. 4.4. show that Gal(E/K) ∼ = [Hint: Consider K(δ) in Gal(E/K) S3 . b ∈ K. If f (X) ∈ K[X] is a separable polynomial. and show that its discriminant is a square in Q. Q( 5 i). v ∈ K with u = 0 such that f (uX + v) = X 3 + aX + b for some a. b ∈ K.7. Show that ∆ = −4b3 − 27a2 . 4. if δ ∈ K.2. βαβ = α−1 = α3 . suppose that w1 .28 to ﬁnd the Galois group of the splitting ﬁeld E of the polynomial X 4 + 3 ∈ Q[X] over Q. Prove that the Galois group of f (X) over Q is cyclic. 3 and suppose that f (X) ∈ K[x] is a cubic polynomial. Q( 23 i)(X K(X 4.66 4. . (c) Continuing with the notation and assumptions of (b). α3 . α2 .5. Q( 3 i)(X 3 − 10)/Q( 3√ i). (a) Show that there u. α. This is a revision exercise on ﬁnite groups of small order. / E and the eﬀect on the element δ of even and odd permutations 4. Q( 2)(X 3 − 10)/Q( 2). 4. Determine all the subextensions F E for which F/Q is Galois. Show that f (X) = X 3 −3X +1 ∈ Q[X] is irreducible over Q.
. Suppose that 0 = a ∈ K and f (X) = X p − a ∈ K[X]. Show that either f (X) is irreducible over K or it has a root in K. 4.8. Show that if f (X) has no root in K then it is irreducible over K. Let K be a ﬁeld of characteristic char K = p where p > 0 is a prime.67 (c) Suppose that the only pth root of 1 in K is 1.
.
1. Lemma. . Let Fpd = {u ∈ Fp : Θpd (u) = 0} ⊆ Fp . A thorough account of ﬁnite ﬁelds and their applications can be found in [6]. especially ﬁnite ﬁelds.1.10. then for a basis v1 . . Notice that u ∈ F0d if and only if up p F0d p d −1 d d F0d = {u ∈ Fpd : u = 0}. p = 1. Corollary. where t1 . td ∈ F . and each root is separable over Fp . Then E is ﬁnite if and only if E/F is ﬁnite and then E = F [E:F ] . . 5. u1 . Proof. containing the prime subﬁeld Fp . (uv)p − uv = up v p − uv = uv − uv = 0. Proof. Here is a more general result. Furthermore. . Our next task is to show that for each power pd there is a ﬁnite ﬁeld with pd elements. For each d 1. Corollary. Therefore by Corollary 1. (u + v)p − (u + v) = (up + v p ) − (u + v) = (up − u) + (v p − v) = 0. 5. hence by Proposition 3. Let F be a ﬁnite ﬁeld with q elements and let V be an F vector space.4. We start with the algebraic closure Fp of Fp and consider the polynomial Θpd (X) = X p − X ∈ Fp [X]. 5. . Let K be a ﬁnite ﬁeld. . Clearly there are exactly q d such expressions. Notice that Θpd (X) = −1. . . Throughout this chapter we will assume that K is a ﬁeld of prime characteristic p = char K > 0. Our ﬁrst goal is to count the elements of K. Let F be a ﬁnite ﬁeld and E/F an extension. so V  = q d . vd we can express each element v ∈ V uniquely in the form v = t1 v1 + · · · + td vd . .2. . 5. 5. Then K/Fp is ﬁnite and K = p[K:Fp ] .CHAPTER 5 Galois extensions for ﬁelds of positive characteristic In this chapter we will investigate extensions of ﬁelds of positive characteristic. if V is ﬁnite then any basis has ﬁnitely many elements and so dimF V < ∞.3. Proposition. Then dimF V < ∞ if and only if V is ﬁnite in which case V  = q dimF V . Then in Fp [X] we have X p − X = X(X − u1 ) · · · (X − upd −1 ). 69 d d d d d d d d .35 Θpd (X) must have exactly pd distinct roots in Fp . Fpd is a ﬁnite subﬁeld of Fp with pd elements and × = Fpd .55 every root of Θpd (X) in Fp is simple. upd −1 . then K is an Fp vector space. the extension Fpd /Fp is a separable splitting ﬁeld. Finite ﬁelds If K is a ﬁnite ﬁeld. Conversely. . . say 0. v ∈ Fpd then by the Idiot’s Binomial Theorem 1. If u. If d = dimF V < ∞.
d The nonzero elements of Fpd are the roots of X p −1 − 1. hence F ∼ Fpd . = Proof. b ∈ F3 . so if u ∈ F3 is a root of X 2 + 1 then F3 (u)/F3 has degree 2 and F3 (u) = F9 .49 for K = L = Fp and M = F . (ii) Fpd Fp is the unique subﬁeld with pd elements. Definition. This group is abelian and has pd − 1 elements. In any ﬁeld the nonzero elements are always invertible. p pd 5.5. so for a. so Fpd is also the splitting subﬁeld for this polynomial. Then K/Fp is a ﬁnite Galois extension. The discriminant of . (iii) If F is any ﬁeld with pd elements then there is a monomorphism F −→ Fp with image Fpd . d d each element u ∈ F × has order dividing pd − 1. Example. therefore up −1 = 1 and so up = u. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC d d Furthermore. hence X 2 + 1 ∈ F3 [X] is irreducible. The ﬁnite subﬁeld Fpd Fp is called the Galois ﬁeld of order pd .8. 5. Let d 1. c. so by Lagrange’s Theorem. Fp1 = GF(p1 ) = GF(p) = Fp and [Fpd : Fp ] = d. subﬁelds both have p (iii) Apply the Monomorphism Extension Theorem 3. But this means every element of F is a root of Θpd (X) and so F Fpd . Let us ﬁnd an irreducible polynomial of degree 2 in F3 [X]. Notice that Fp Fpd . By inspection. Of course. d d (i) Fpd Fp is the splitting subﬁeld for each of the polynomials X p − X and X p −1 − 1 over Fp . (ii) Let F Fp have pd elements. the image of the resulting monomorphism must be Fpd .7. So X 2 + X − 1 and X 2 − X − 1 are also quadratic irreducibles in F3 [X]. Hence Fpd Fp . = It is worth noting the following consequence of this result and the construction of Fpd . d ∈ F2 we have (a + bw)(c + dw) = ac + (ad + bc)w + bdw2 = (ac + bd) + (ad + bc + bd)w. equality follows since these d elements. Every element of F9 can be uniquely expressed in the form a + bu with a. Example. b ∈ F2 . the other being w + 1 since the sum of the roots is the coeﬃcient of X. Consider the polynomial X 9 − X ∈ F3 [X]. (i) As Fpd consists of exactly the roots of Θpd (X) in Fp . Corollary. This tells us that every element of F4 = F2 (w) can be uniquely expressed in the form a + bw with a. Notice that the nonzero elements of F form a group F × under multiplication. in the ring F3 [X] we ﬁnd that X 9 − X = X(X 8 − 1) = (X 3 − X)(X 2 + 1)(X 2 + X − 1)(X 2 − X − 1). Its splitting ﬁeld is a quadratic extension F2 (w)/F2 where w is one of the roots of X 2 + X + 1. 5. Multiplication is carried out using the relation u2 = −1 = 2. hence F0d = F× . By inspection.70 5. Consider the polynomial X 4 − X ∈ F2 [X].9. if u = 0 then up −1 = 1 and so u has multiplicative inverse up −2 . To calculate products we use the fact that w2 = w + 1. therefore F ∼ Fpd . By (ii). Let K be a ﬁnite ﬁeld of characteristic p. it is the splitting subﬁeld. Now X 2 + X + 1 has no root in F2 so it must be irreducible in F2 [X].6. Notice that X 2 + 1 has no root in F3 . so Fpd /Fp is a ﬁnite extension. We can ﬁnd their roots in F9 using the quadratic formula since in F3 we have 2−1 = (−1)−1 = −1. Proposition. 5. The notation GF(pd ) is often used in place of Fpd . in the ring F2 [X] we ﬁnd that X 4 − X = X 4 + X = X(X 3 + 1) = X(X + 1)(X 2 + X + 1). b. 5.
2. { {{ {{ { {{ {{ {{ {{ {{ { {{ {{ { {{ gg gg gg gg Fp24 Fp8 Fp12 Fp4 Fp6 gg gg gg gg { {{ {{ { {{ Fp2 Fp3 Fp Figure 5. Proposition 5. Fp (u) Fp is a ﬁnite subﬁeld. pn = (pm )[Fpn :Fpm ] = pm[Fpn :Fpm ] . Then Fpm Fpn if and only if m  n.7.10 now implies that Fp (u) = Fpn for some n.1. so m  n. Let u ∈ Fp . write n = km with k up = up n mk 1. so its roots are (−1)(−1 ± u) = 1 ± u. Then u is algebraic over Fp and the extension Fp (u)/Fp is ﬁnite. The algebraic closure of Fp is the union of all the Galois ﬁelds of characteristic p. Fp = Fpn . There are two issues we can now clarify. each element u ∈ Fp is separable over Fp . The subﬁelds of Fp24 5. = up m(k−1) = · · · = up = u. If Fpm Fpn .5. Proof. Hence by Corollary 5. n 1 Furthermore. Proof. then by Corollary 5. so m m(k−1) m = (up )p Fpn . The separability statement follows from Corollary 5.1 which shows extensions with no intermediate subextensions. The diagram of subﬁelds for Fp24 can be seen in Figure 5. Theorem. Then we have F9 = F3 (u) = F3 (1 ± u) = F3 (−1 ± u). We will require a useful fact about Galois ﬁelds. 5. If m  n.1.10. . Similarly.11. m Hence u ∈ Fpn and therefore Fpm This means that we can think of the Galois ﬁelds Fpn as ordered by divisibility of n. Let Fpm and Fpn be two Galois ﬁelds of characteristic p. Proposition. Then for u ∈ Fpm we have up = u.2. FINITE FIELDS 71 X 2 + X − 1 is 1 − 4(−1) = 5 = 2 = u2 . the discriminant of X 2 − X − 1 is 1 − 4(−1) = 5 = 2 = u2 and its roots are (−1)(1 ± u) = −1 ± u.
19.e. its order in the group F× is (pd − 1). × × . 5. hence w = F× . 5. so if u ∈ F× is a generator of this cyclic group. this extension is Galois and  Gal(Fpnd /Fpd ) = [Fpnd : Fpd ] = n. (i) We have 4  (p − 1) = F× .2. 5. although as we will in the next result. Example. Definition. unity. This implies that Fpnd = Fpd (w). some authors use the term primitive element for what we have called a primitive root..46.16. Remark. there are four primitive roots and these are the roots of the polynomials X 2 + X − 1 and X 2 − X − 1 which we found to be ±1 ± u. Then every element of Fpnd can be expressed as a polynomial in w. Galois groups of ﬁnite ﬁelds and Frobenius mappings Consider an extension of Galois ﬁelds Fpnd /Fpd . Since 9 ϕ(8) = 4. the polynomial X 2 + 1 ∈ Fp [X] has two roots in Fp .9 we have F9 = F3 (u) and F× is cyclic of order 8. Let K be a ﬁeld.73. Fpnd has a primitive root w say.18. Indeed. 5. Proposition. every primitive root is indeed a primitive element in our sense! 5.e. Then every ﬁnite subgroup U Proof. (ii) If p ≡ 3 (mod 4) the polynomial X 2 +1 ∈ Fp [X] is irreducible. Use Corollary 1. But then 4  (p − 1) = F× . (ii) If v ∈ Fp is a root of X 2 + 1 then v has order 4 in F× .12. but that conﬂicts with our usage. hence this is a root of X 2 + 1 (the other root is −uFp /4 ). 5. which is p p impossible since p − 1 ≡ 2 (mod 4).15. 5. Proposition. pd pd 5. = Proof. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC 5. i. so Fpnd Fpd (w) Fpnd . Example. In Example 5.12. so Fp2 ∼ Fp [X]/(X 2 +1).7 and Proposition 3. i.21.17. By Proposition 5. Fpd contains a primitive nth root of unity if and only if pd ≡ 1 (mod n) and p n. We record a fact that is very important in Number Theory.20.20.75 which we had previously only established for inﬁnite ﬁelds. Proposition.14.35 and Lemma 1. In Example 5. w ∈ F× is called a primitive root if it is a primitive (pd − 1)th root of pd K × is cyclic.6(i). Here is a generalization of Proposition 5. Proposition. Proposition. We next introduce an important element of the Galois group Gal(Fpnd /Fpd ). Unfortunately the word primitive has two confusingly similar uses in the context of ﬁnite ﬁelds. Proof. 5. Corollary 5. By Proposition 5.8 we ﬁnd that F4 = F2 (w) has the two primitive roots w and w + 1.72 5. The extension of Galois ﬁelds Fpnd /Fpd is simple. 5. The group of units F× in Fpd is cyclic.. Let p > 0 be an odd prime. the p p order of uFp /4 is 4. (i) If p ≡ 1 (mod 4). pd This is a special case of a more general result about arbitrary ﬁelds. Remark.13. Fpnd = Fpd (u) for some u ∈ Fpnd . This completes the proof of the Primitive Element Theorem 3.
so Fd is a ring homomorphism. Proof.5. Proof.22. The (relative) Frobenius mapping for the extension Fpnd /Fpd is the funcd tion Fd : Fpnd −→ Fpnd given by Fd (t) = tp . For k  n. Let d 1. But this can only be true if k = n. Proposition. GALOIS GROUPS OF FINITE FIELDS AND FROBENIUS MAPPINGS 73 5. the cyclic group generated by Fd . so Fd ∈ Gal(Fpnd /Fpd ). In Gal(Fpnd /Fpd ) = Fd . 5. hence 1. then every element u ∈ Fpnd satisﬁes the equation Fk (u) = u which expands to up d u ∈ Fpkd . Proposition.25. In fact for u ∈ Fp . d d d d d d d Fd (uv) = (uv)p = up v p . Definition. (ii) The restriction of Fd to the Galois subﬁeld Fpdn agrees with the relative Frobenius mapping Fd : Fpnd −→ Fpnd . This also shows d d d that the order of Fd in the group AutFpd (Fpnd ) is at most n. notice that the composition power Fn = Fd ◦ · · · ◦ Fd (with n factors) satisﬁes d Fn (t) = tp d nd =t for all t ∈ Fpnd . hence Fn = id. 5. The relative Frobenius mapping Fd : Fpnd −→ Fpnd is an automorphism of Fpnd that ﬁxes the elements of Fpd .23. hence Fk (u) = u if and only if u ∈ Fpdk . Suppose the order is k with k n. To see that Fd is an automorphism. The order of Fd is n. d Fk Fpnd n/k d Fpnd = Fpdk Fk k Fpd Proof. Then Fd is invertible with inverse F−1 = Fn−1 . so Fd ﬁxes the elements of Fpd . Proposition. then Fk = Fkd . so Gal(Fpnd /Fpd ) = Fd . Hence in the automorphism group AutFpd (Fp ). Also. we have the identities Fd (u + v) = (u + v)p = up + v p .2. This is left as an exercise. For u. Fd (t) = t . consider the function (i) Fd : Fp −→ Fp is an automorphism of Fp which ﬁxes the elements of Fpd . Fd (u) = u if and only if u ∈ Fpd . d d 5. For d Fd : Fp −→ Fp . pd kd = u. (ii) If k 1. Fd has inﬁnite d order. so AutFpd (Fp ) is inﬁnite. Frobenius mappings exist on the algebraic closure Fp .24. v ∈ Fpnd . d d dk . For u ∈ Fpnd we have Fk (u) = up . the ﬁxed subﬁeld of Fk in Fpnd is Fpnd = Fpdk . The Frobenius mapping F = F1 is often called the absolute Frobenius mapping since it exists as an element of each of the groups AutFp (Fp ) and AutFp (Fpn ) = Gal(Fpn /Fp ) for every n 1. for u ∈ Fpd we have Fd (u) = up = u. for each k with k  n there is the cyclic subgroup Fk of order d  Fk  = n/k.
Clearly TrFpnd /Fpd is additive. So by Proposition 5. so Fpd linearity follows from the formula tu + (tu)p + (tu)p + · · · + (tu)p d 2d (n−1)d d = tu + tup + tup d 2d + · · · + tup (n−1)d . This means that ker TrFpnd /Fpd cannot be the whole of Fpnd . Proposition. If we modify TFpnd /Fpd to have codomain Fpd . notice that TrFpnd /Fpd (u) = 0 if and only if u is a root of the polynomial X + Xp + Xp + · · · + Xp d 2d (n−1)d ∈ Fpd [X] which has degree p(n−1)d and so has at most p(n−1)d < pnd roots in Fpnd . To see that TrFpnd /Fpd is surjective.1.26. (n−1)d + u = TFpnd /Fpd (u). TrFpnd /Fpd is surjective since its codomain has dimension 1.2. For t ∈ Fpd we have tp = t. .74 5.3. The relative trace TrFpnd /Fpd is a surjective Fpd linear mapping and whose kernel is an Fpd vector subspace of dimension n − 1. we obtain the relative trace TrFpnd /Fpd : Fpnd −→ Fpd . Gal(Fp24 /Fp ) = F ∼ Z/24 = F4 8 F F12 lll lll lll lll lll 2 F ll lll lll lll lll lll F6 F3 F24 = {id} Figure 5.24(i).2 shows the subgroup diagram corresponding to the lattice of subﬁelds of Fp24 shown in Figure 5. consider the function TFpnd /Fpd : Fpnd −→ Fpnd deﬁned by TFpnd /Fpd (u) = u + up + up + · · · + up Notice that Fd (TFpnd /Fpd (u)) = up + up + up + · · · + up = up + up + up + · · · + up d 2d 3d d 2d 3d nd d 2d (n−1)d = u + Fd (u) + F2d (u) + · · · + F(n−1)d (u). GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC Figure 5. The subgroups of the Galois groups of Fp24 /Fp 5. Proof. The trace and norm mappings For an extension of Galois ﬁelds Fpnd /Fpd . TrFpnd /Fpd (u) = u + up + up + · · · + up d 2d (n−1)d . TFpnd /Fpd (u) ∈ Fpd . 5.
5. (b) X 8 − 1 ∈ F5 [X]. pd p2d u u p3d ···u p(n−1)d So by Proposition 5. (d) X 8 − 1 ∈ F2 [X]. Let w ∈ F× be a primitive root.13 also applies to an integral domain in place of a ﬁeld. Exercises on Chapter 5 5. = up up up · · · up = uu u = N(u). For a ∈ K.24(i).4.6. let 5. . (c) X 8 − 1 ∈ F11 [X]. Let p > 0 be a prime. 5. we also have pd  im NormFpnd /Fpd  pd − 1. By redeﬁning the codomain we obtain the relative norm NormFpnd /Fpd : F× −→ F× . Find the splitting ﬁeld of f (X). 1 + pd + · · · + p(n−1)d = pnd − 1  ker NormFpnd /Fpd  pnd − 1 .5.20 if we try to take p = 2. Proposition. 5.1.3. the splitting ﬁelds of f (X) and g(X) over Fpd agree.2.27. in each case ﬁnd a primitive root of this Galois ﬁeld: (a) X 8 − 1 ∈ F41 [X]. show that w ∈ F× . What happens to Theorem 5. The kernel of NormFpnd /Fpd consists of the roots in Fpnd of the polynomial d (n−1)d X 1+p +···+p − 1 ∈ Fpd [X]. and d  ϕ(pd − 1). N(u) ∈ Fpd . Find the smallest Galois ﬁelds containing all the roots of the following polynomials. 5. The relative norm NormFpnd /Fpd is a surjective group homomorphism. Multiplicativity is obvious. If pd < d. Deduce that for any other irreducible polynomial g(X) ∈ Fpd [X] with deg g(X) = n. Consider the function N : F× −→ F× pnd pnd for which Then we have Fd (N(u)) = up up up · · · up d 2d 3d d 2d 3d nd (n−1)d N(u) = uup up · · · up d 2d (n−1)d = u Fd (u) F2d (u) · · · F(n−1)d (u). 5. Proof. Suppose that d d ga (X) = X p − X − a ∈ K[X]. pd F× . Deduce that degFp w = d / p 1. so  ker NormFpnd /Fpd  Hence  im NormFpnd /Fpd  = Since im NormFpnd /Fpd therefore im NormFpnd /Fpd = F× .75 There is a multiplicative version of this construction. pnd pd NormFpnd /Fpd (u) = uup up · · · up d 2d (n−1)d . pd − 1 pd − 1. Show that Proposition 5. Let f (X) ∈ Fpd [X] be an irreducible polynomial with deg f (X) = n. and K/Fpd is an extension.
(c) If K is a ﬁnite ﬁeld and d > 1. d 1 and write q = pd . Show that there is a unique group homomorphism λq : F× −→ {±1} which is characterized by the requirement q that for every u ∈ F× . Σq = {u2 ∈ Fq : u ∈ Fq } ⊆ Fq . (a) Consider {±1} = {1. Show that the number of elements of Σq is Σq  = (q + 1)/2. λq (u) = 1 if and only if u = v 2 for some v ∈ F× .7. −1} as a group under multiplication. (c) If t ∈ Fq . show that ga (X) is irreducible over K if and only if it has no root in K.] (b) If d = 1. (e) What can you say about the case p = 2? . 5. then so is u + t for every t ∈ Fp . (d) Deduce that the equation x2 + y 2 + z 2 = 0 has at least one nontrivial solution in Fq . Let p be an odd prime.76 5. explain why ga (X) can never be irreducible over K. Is λq always q q surjective? (b) Consider the set of all squares in Fq . show that the splitting ﬁeld E of ga (X) over K is separable and Gal(E/K) ∼ Fpd . [Hint: show that if u ∈ E is a root of ga (X) = in an extension E/K. GALOIS EXTENSIONS FOR FIELDS OF POSITIVE CHARACTERISTIC (a) If the polynomial ga (X) is irreducible over K. Deduce that if t ∈ Fq then the set t − Σq = {t − u2 ∈ Fq : u ∈ Fq } has t − Σq  = (q + 1)/2 elements. show that Σq ∩ (t − Σq ) 1. Deduce that every element of Fq is either a square or can be written as the sum of two squares.
hence Gal(E/R) is a 2group. The ﬁeld of complex numbers C is algebraically closed and R = C. P  which shows that E P /R has odd degree. there is a normal subgroup N Gal(E/C) of index 2. the standard primitive nth root of 1 in C. 6. The splitting ﬁeld of p(X) over C is contained in the splitting ﬁeld E of minpolyR. Let ζn = e2πi/n . 6.2. Many of the ideas introduced in this chapter are of great importance in these and other mathematical areas. Galois Theory has always been an important tool in Number Theory and Algebra.1. Since C E. so we can consider the Galois extension E N /C of degree 2. Theorem (The Fundamental Theorem of Algebra).43. As C/R is a Galois extension. This proof is essentially due to Gauss but he did not use the historically more recent Sylow theory.CHAPTER 6 A Galois Miscellany In this chapter we will explore some miscellaneous topics in Galois Theory.1. every quadratic aX 2 + bX + c ∈ C[X] has complex roots (because we can ﬁnd square roots of every complex number).3.u (X). The Primitive Element Theorem 3.29). For the ﬁxed subﬁeld of P . every real polynomial of odd degree has a real root. it was claimed that the irreducible polynomial over Q which has ζn as a root was the 77 . Ring Theory and such diverse areas as Differential Equations. From the theory of 2groups. Historically. so C/R is algebraic. we have [E P : R] =  Gal(E/R) . so suppose not. 6.u (X)(X 2 + 1) over R. Then any root u of p(X) in the algebraic closure C is algebraic over R. Now consider a 2Sylow subgroup P Gal(E/R) and recall that  Gal(E/R)/P  is odd. We know that [C : R] = 2. we can consider the normal subgroup Gal(E/C) Gal(E/R) for which  Gal(E/R) = 2  Gal(E/C). So we cannot have an irreducible quadratic polynomial in C[X]. Complex Analysis and Algebraic Geometry. our proof only uses the connectivity of the real line (via the Intermediate Value Theorem) together with explicit calculations in C involving square roots. Proof. In Theorem 1. We must show that  Gal(E/C) = 1. we have 2  [E : R] and so 2   Gal(E/R). Let p(X) ∈ C[X] be irreducible. Therefore  Gal(E/C) = 1 and E = C. A proof of the Fundamental Theorem of Algebra We will prove the Fundamental Theorem of Algebra for the complex numbers C. so irreducibility implies that v has degree 1 over R and therefore E P = R. This shows that Gal(E/R) = P . It is interesting to compare the proof below with others which use the topology of the plane and circle or Complex Analysis. stimulating the development of subjects such as Group Theory.75 allows us to write E P = R(v) for some v whose minimal polynomial over R must also have odd degree. But by the Intermediate Value Theorem. Cyclotomic extensions We begin by discussing the situation for cyclotomic extensions over Q using material discussed in Section 1. so in C[X] we have p(X)  minpolyR. But from known properties of C (see Proposition 3.
. Let π : Z(ζn ) −→ Fpd be the quotient homomorphism. t • Gal(Q(ζn )/Q) ∼ (Z/n)× . by iterating this we ﬁnd tp that ζ n is conjugate to every power of the form ζn . therefore Φn (X) = minpolyQ. aj ∈ Z} ⊆ Q(ζn ). . Now consider a maximal ideal P1 Z[ζn ] containing p1 . π gives an injective r group homomorphism. n) = 1. Let p > 0 be a prime which does not divide n. So π is injective. In fact. = t Proof. n − 1 and gcd(t. we ﬁnd that ζ n is conjugate to ζ n1 . . we see that n ∈ P and hence p  n. ζ n is conjugate to ζn .ζn (X) = minpolyQ. Let n 2. To see this. . Inside the group of units of Z(ζn ) is the subgroup of powers of ζn . .ζn (X) t and hence Φn (X) is irreducible.. . Its group of units contains the cyclic subgroup ζn of order n. But Z ∩ P = (p) and so r ∈ P . t t t tp This shows that in the Galois extension Fpd /Fp . Since the complex roots of Φn (X) are the powers ζn with t = 1. π : ζn −→ F× . By separability and 1 p r1 . . To do this we will show that every power ζn as above is actually a Galois conjugate of ζn over Q. Now let t = 1. this is a cyclic subgroup of order n.n−1 gcd(t. it is a ﬁnite ﬁeld. ζn Z(ζn )× . where the element tn ∈ (Z/n)× acts on Q(ζn ) by tn · ζn = ζn . 2. n) = 1 we also have pj n s. Then Z(ζn ) is a subring of Q(ζn ) and so is an integral domain. . Recalling that n−1 ζn + · · · + ζn + 1 = 0. t tp F(ζ n ) = (ζ n )p = ζn . Consider r Z(ζn ) = {a0 + a1 ζn + · · · + ar ζn : r 0. Let P Z(ζn ) be a maximal ideal which contains p. We claim that when restricted to ζn . On factoring we have r−1 (ζn − 1)(ζn + · · · + ζn + 1) ≡ (ζn − 1)r (mod P ). then the quotient ring Z(ζn )/P is a ﬁeld of characteristic p. Then t (X − ζn ). Reducing modulo P1 and working in the resulting extension Fpd1 /Fp1 . . Writing u = π (u). By elementary Group Theory we can assume that r  n and so p r.. . n − 1. then ζn − 1 ∈ P . Q(ζn ) is the splitting ﬁeld of Φn (X) over Q and indeed Q(ζn ) = Q(ζn ) whenever t t is a primitive nth root of unity. n − 1 and t gcd(t.. say Fpd for some d.n)=1 • Q(ζn ) = Q[X]/(Φn (X)). . Suppose there is a factorization t k Φn (X) = f (X) minpolyQ.. The main step in the proof has the above properties and so ζn t is to show that Φn (X) ∈ Z[X] is irreducible. where the pj are primes with 2 m 1 gcd(t. thus contradicting our original assumption on n.ζn (X) t for some monic polynomial f (X) ∈ Z[X] and f (ζn ) = 0. t=1.78 6. Theorem. • [Q(ζn ) : Q] = ϕ(n). suppose that π (ζn ) = 1 for some r = pd r 1. .2. we can consider the eﬀect of the absolute Frobenius map F : Fpd −→ Fpd t on ζ n = ζn . Since t = pr1 · · · prm . Consider the prime power factorization p1 < · · · < pm and rj 1 with. hence / ζn − 1 ∈ P . so ζn − 1 ∈ P or r ∈ P since maximal ideals are prime. A GALOIS MISCELLANY nth cyclotomic polynomial Φn (X) = 6. n) = 1.
this shows that minpolyQ. Continuing in this fashion. Example. . 2. . Then Q(ζn )R = Q(ζn ). Hence ( ) = {id. m we have p 1 p 2 ···p j minpolyQ. .ζn (ζn1 and so ζn r r p11 p22 p r1 r2 p2 )=0 is conjugate to ζn .ζn (ζn ) = 0. so we also have (6. ζn n n r r p11 ···pj j 6. consider the cyclotomic extension Q(ζn )/Q where ζn = e2πi/n .ζn (ζn1 ) = 0 and so ζn1 is p r1 p r1 p r1 Repeating this argument starting with ζn1 and using the prime p2 we ﬁnd that minpolyQ. Furthermore. Hence t is conjugate to ζ in the extension Q(ζ )/Q.3. we ﬁnd that p 1 f (ζn1 ) r = 0 and so conjugate to ζn . .6. For n > 2. for each j = 1. Complex conjugation corresponds to the residue class of −1 ≡ n − 1 (mod n). Notice that ζn is a root of the polynomial X 2 − 2 cos(2π/n)X + 1 ∈ Q(cos(2π/n))[X]. Recall that Gal(Q(ζn )/Q) ∼ Z/n× . Thus we have Q(ζn ) ( ) = Q(cos(2π/n)). . ( )} = Gal(Q(cos(2π/n))/Q). ϕ(n) . Making use of the identities ( ) = Q(ζn + ζ n ) = Q(cos(2π/n)). .1) minpolyQ(cos(2π/n)). CYCLOTOMIC EXTENSIONS 79 the fact that the reduction map π1 : Z[ζn ] −→ Fpd1 is injective on the powers of ζn . it is elementary to show that cos(2πr/n) = cos(2π/n) unless r ≡ 1 (mod n). p 1 f (ζn1 ) r 1 = 0 in Z[ζn ]. The residue class of r acts by sending cos(2π/n) to cos(2πr/n). and so [Q(cos(2π/n)) : Q] = ϕ(n)/2.2. [Q(ζ24 ) : Q] = ϕ(24) = 8 and Gal(Q(ζ24 )/Q) ∼ Z/2 × Z/2 × Z/2.ζn (X) = X 2 − 2 cos(2π/n)X + 1. Q(ζn )R = Q(ζn ) and [Q(cos(2π/n)) : Q] = Proof. . n − 1. 2 eθi = cos θ + sin θ i. This shows that minpolyQ. = 6.ζn (ζn1 2 j ) r r r =0 t and so ζn is conjugate to ζn . . 2 2 Complex conjugation ﬁxes each of the real numbers cos(2πk/n) for k = 1. = r where the residue class of r acts by sending ζn to ζn . 2 1 1 −1 cos(2π/n) = (ζn + ζ n ) = (ζn + ζn ). Theorem.4. We have . When j = m. we obtain 1 cos θ = (eθi + e−θi ).
2 2 2 2 √ √ and all of these numbers are in Q(ζ24 ). ζ24 = − + i. it is an interesting exercise to ﬁnd them all together with their ﬁxed subﬁelds. hence since Φ24 (X) = X 8 − X 4 + 1.cos(π/12) (X). 2 √ 3 2 cos (π/12) − 1 = 2 2 √ √ Q(ζ24 ) = Q( 2. 4 16 cos4 (π/12) − 16 cos2 (π/12) + 1 = 0. so −2 −1 2 ζ24 + ζ24 = (ζ24 + ζ24 )2 − 2.80 6. −4 Then after multiplying by ζ24 we have −4 4 ζ24 − 1 + ζ24 = 0. 16X 4 − 16X 2 + 1 = 16 minpolyQ. Gal(Q(ζ24 )/Q) has 23 − 1 = 7 subgroups of each of the orders 2 and 4. −1 −2 2 (ζ24 + ζ24 )2 = ζ24 + ζ24 + 2. . The minimal polynomial for cos(π/12) can also be found as follows. = We also have cos(2π/24) = cos(π/12) ∈ Q(ζ24 ). 6.5. √ √ √ 2 2 3 1 6 3 8 ζ24 = i. ζ24 = + i.43 we have [Q(ζ24 ) : Q] = 8. we obtain 8 4 ζ24 − ζ24 + 1 = 0. Remark. giving −4 4 (ζ24 + ζ24 ) − 1 = 0. hence −4 −1 −2 4 2 ζ24 + ζ24 = (ζ24 + ζ24 )4 − 4(ζ24 + ζ24 ) − 6. Similarly. By Theorem 1. A GALOIS MISCELLANY Proof. Combining these we have −1 −1 (ζ24 + ζ24 )4 − 4(ζ24 + ζ24 )2 + 1 = 0. It is easy to check that √ √ [Q( 2. For this example. hence Q( 2. 3. which implies that Using this we ﬁnd that Gal(Q(ζ24 )/Q) ∼ Z/2 × Z/2 × Z/2. 3. i) Q(ζ24 ).28 applies to the polynomial minpolyQ. 3. Now −1 −4 −2 4 2 (ζ24 + ζ24 )4 = (ζ24 + ζ24 ) + 4(ζ24 + ζ24 ) + 6. we have and so giving Then 3 4 cos4 (π/12) − 4 cos2 (π/12) + 1 = . Since √ 3 cos(2π/12) = cos(π/6) = .cos(π/12) (X). We have Φ24 (ζ24 ) = 0. Also. i) : Q] = 8. i). Note that case (i) of Kaplansky’s Theorem 4.
so (X 2 + 1) splits over F9 (T ). (i) By Proposition 5.. Hence Gal(K(ζ)/K) is isomorphic to a subgroup of Gal(Q(ζn )/Q) ∼ (Z/n)× which implies that it is abelian and its order divides ϕ(n). X 4 − 1 splits over F5 . For example. The splitting ﬁeld of (X 2 + 1) over F3 is F9 = F3 (ζ). where ζ ∈ K is a primitive nth root of unity. (i) The splitting ﬁeld of X 4 − 1 over F3 (T ) is F9 (T ) and Gal(F9 (T )/F3 (T )) ∼ (Z/4)× ∼ Z/2. hence it is abelian with order dividing ϕ(n). The Galois group Gal(K(ζ)/K) is isomorphic to a subgroup of (Z/n)× . 6. then the splitting ﬁeld of X n − 1 over K is K(ζ). X 4 − 1 splits over F5 (T ) and the Galois group Gal(F5 (T )/F5 (T )) is trivial.7. So X n − 1 splits over K(ζ) and each element α ∈ Gal(K(ζ)/K) has the action α(ζ) = ζ rα . n) = 1. i. This generator clearly extends to an automorphism of F9 (T ) which ﬁxes T . When p = char K > 0. . Here are the ﬁrst few of these polynomials: T2 (X) = 2X 2 − 1. Now let K be a ﬁeld with characteristic char K n. so we can view it as an element of K[X] since either Q K or Fp K and we can reduce the coeﬃcients modulo p. T4 (X) = 8X 4 − 8X 2 + 1. X 4 − 1 is separable over F3 (T ) and has irreducible factors (X − 1). 6.X). However. Working in K.6. (ii) By Proposition 5. T6 (X) = 32X 6 − 48X 4 + 18X 2 − 1. where gcd(rα . T5 (X) = 16X 5 − 20X 3 + 5X.e. Gal(F9 /F3 ) ∼ (Z/4)× ∼ Z/2. = 6. These form a system of orthogonal polynomials which can be computed in Maple using the command orthopoly[T](n.6. The polynomial that expresses cos nθ as a polynomial in cos θ is the nth Chebsyhev polynomial of the ﬁrst kind Tn (X) ∈ Z[X]. In either case it can happen that Φn (X) factors in K[X]. T7 (X) = 64X 7 − 112X 5 + 56X 3 − 7X. = = (ii) By Proposition 5. if K = Fpd (T ) then the value of d is the crucial factor. (X + 1) and (X 2 + 1).2. The precise outcome can be determined with the aid of Proposition 5. Proof. We have the following splitting ﬁelds and Galois groups.21. Theorem. we know that Φn (ζ) = 0.20. If char K n.20.20. where ζ 2 + 1 = 0. Remark. 6. Proof. when n > 2. this Galois group only depends on the largest subﬁeld of K which is algebraic over Fp . hence the roots of minpolyK. CYCLOTOMIC EXTENSIONS 81 and so 16 cos4 (π/12) − 16 cos2 (π/12) + 1 = 0. The polynomial Φn (X) has integer coeﬃcients. Example.9.ζ (X) ∈ K[X] are primitive roots of 1. we can still describe the splitting ﬁeld of X n − 1 over K and its Galois group. T3 (X) = 4X 3 − 3X. = = with generator σ satisfying σ(ζ) = ζ −1 = −ζ.8. Also. This method will work for any n where ϕ(n) is even. Remark.
. . . χk+1 is linearly dependent. . χα : K × −→ K × is a character of in K. .12. . 6. Hence χ2 .2) applied to g0 g yields t1 χ1 (g0 g) + · · · + tk+1 χk+1 (g0 g) = 0. So for all g ∈ G. Example. Suppose that α1 . χn be distinct characters of a group G in a ﬁeld K. 6. t1 χ1 (g) + · · · + tn χn (g) = 0. . . So χ1 . . In this deﬁnition. . . suppose that it holds for any n k. .2) t1 χ1 + · · · + tk+1 χk+1 = 0.16. . Corollary. αn are distinct automorphisms of the ﬁeld K. . 6. . . Then χ1 .82 6. For n = 1 the result is easily veriﬁed. there must be an element g0 ∈ G for which χ1 (g0 ) = χ2 (g0 ). Then there is a z ∈ K for which t1 α1 (z) + · · · + tn αn (z) = 0. If χ1 . . .10. Let χ1 . . Hence the Klinear transformation t1 α1 + · · · + tn αn : K −→ K is nontrivial. . . Given any ring homomorphism ϕ : R −→ K we obtain a character of R× in K by restricting ϕ to a map χϕ : R× −→ K × . . tn ∈ K be a sequence of elements. . 6. Definition. in which the coeﬃcient t2 (χ2 (g0 )−χ1 (g0 )) is not zero. . Artin’s Theorem on linear independence of characters Let G be a group and K a ﬁeld. . again contradicting the inductive assumption. Example. . χk+1 be a set of k + 1 distinct characters for which there are t1 . . Let E/K be a Galois extension and σ ∈ Gal(E/K). say tr = 0. t1 χ1 + · · · + tn χn = 0 =⇒ t1 = · · · = tn = 0. . we see that t1 χ1 (g0 )χ1 + · · · + tk+1 χk+1 (g0 )χk+1 = 0. We proceed by induction on n. tn ∈ K. . Proof. . χr+1 . χn are linearly independent if for t1 . . then χ1 . χn are linearly independent. . Given an automorphism α : K −→ K.2) by χ1 (g0 ) and subtracting gives t2 (χ2 (g0 ) − χ1 (g0 ))χ2 + t3 (χ3 (g0 ) − χ1 (g0 ))χ3 + · · · + tk+1 (χk+1 (g0 ) − χ1 (g0 ))χk+1 = 0. . Equation (6. Let t1 . the functional equation means that for all g ∈ G. . . Multiplying Equation (6.11. . . . If one of the ti is zero. . . 6. . Let χ1 .13. . χr−1 . As χ1 = χ2 . 6. . not all of which are 0. . contradicting the inductive assumption. .14. . . . . Theorem (Artin’s Theorem). χk+1 is linearly independent. For the inductive assumption.15. Then χσ : E × −→ E × is a character. .3. χk+1 is linearly dependent. Example. Hence all of the ti must be nonzero. K× 6. A group homomorphism χ : G −→ K × is called a character of G with values in K. . tk+1 ∈ K not all zero and such that (6. . Let χ1 . Definition. and therefore since χj (g0 g) = χj (g0 )χj (g). χn are not linearly independent then they are linearly dependent. . A GALOIS MISCELLANY 6. . . Then χ1 . χn be characters of a group G in a ﬁeld K. which demonstrates the inductive step. . .
Proof. There is another homomorphism δE/K : E × −→ E × . So im δE/K Proposition 5.18. Corollary. Explicitly. Hence the associated Klinear transformation TrE/K : E −→ K. then there is a v ∈ E × such that u = vσ(v)−1 . . Proof. Our next result is an important generalization of δE/K (u) = uσ(u)−1 . 1. Let E/K be a ﬁnite Galois extension with cyclic Galois group Gal(E/K) = σ of order n. the element uσ(u) · · · σ n−1 (u) ∈ E satisﬁes σ(uσ(u) · · · σ n−1 (u)) = σ(u) · · · σ n−1 (u)σ n (u) = σ(u) · · · σ n−1 (u)u.27. so for some w ∈ E. Then im δE/K = ker NE/K . . αn be the distinct elements of Gal(E/K). Consider the function id +uσ + uσ(u)σ 2 + · · · + uσ(u) · · · σ n−2 (u)σ n−1 : E × −→ E. . .6. TrE/K (x) = α1 (x) + · · · + αn (x) is surjective. . Notice that for u ∈ E × . Thus we have u = vσ(v)−1 as required. . Suppose that E/K is a ﬁnite Galois extension with cyclic Galois group Gal(E/K) = σ of order n. The rest of the statement follows directly from Corollary 6. Hence.3. . since σ n (u) = u. The function TrE/K : E −→ K is called the trace mapping of E/K.15. Let E/K be a ﬁnite Galois extension of degree n and let α1 . . . For each u ∈ E × .3.16. ker NE/K . αn apart from its order. . Notice that uσ(v) = uσ(w) + uσ(u)σ 2 (w) + uσ(u)σ 2 (u)σ 3 (w) + · · · + uσ(u)σ 2 (u) · · · σ n−1 (u)σ n (w) = v. . NE/K (δE/K (u)) = (uσ(u)−1 )(σ(u)σ 2 (u)−1 · · · σ n−1 (u)σ n (u)−1 ) = 1. . since uσ(u)σ 2 (u) · · · σ n−1 (u)σ n (w) = w. . . n−1 are distinct and linearly independent by Artin’s Theorem 6. . α1 (x) + · · · + αn (x) ∈ E Gal(E/K) = K. NE/K : E × −→ K × . This cannot be identically zero. γαn is the same as α1 . Now using this we deﬁne a group homomorphism NE/K (u) = uσ(u) · · · σ n−1 (u). NE/K is called the norm mapping for E/K and generalizes the norm mapping for ﬁnite ﬁelds of Section 5. Theorem (Hilbert’s Theorem 90). since the list γα1 . if u ∈ E × and u σ(u) · · · σ n−1 (u) = 1.17. ARTIN’S THEOREM ON LINEAR INDEPENDENCE OF CHARACTERS 83 6. 6. First note that for x ∈ E and γ ∈ Gal(E/K). the element v = w + uσ(w) + uσ(u)σ 2 (w) + · · · + uσ(u) · · · σ n−2 (u)σ n−1 (w) is nonzero. . Then the function α1 + · · · + αn : E −→ E is a nontrivial Klinear transformation whose image is contained in K. The characters σ k : E × −→ E × with k = 0. . hence in uσ(u) · · · σ n−1 (u) ∈ E σ = K. γ(α1 (x) + · · · + αn (x)) = γα1 (x) + · · · + γαn (x) = α1 (x) + · · · + αn (x). Let u ∈ ker NE/K .
u)/K))/ Gal(K(ζ. Proposition. then the splitting ﬁeld of f (X) = X n − a over K has the form K(u). u)/K(ζ)) = is abelian. . where u is a root of f (X). This means that K(u)/K is Galois. .22. . . is a group homomorphism. Then the splitting ﬁeld of f (X) over K has the form K(u.84 6. ak ∈ K. . A GALOIS MISCELLANY 6.19. uk ) where un = 1 a1 . For each α ∈ Gal(K(u)/K) we have α(u) = ζ rα u for some rα = 0. Proposition. Then K(u)/K is a Galois extension and Gal(K(u)/K) is cyclic with order dividing n. Corollary. where u is a root of f (X) and ζ is a primitive nth root of 1. ρ(α) = ζ rα . 6. ζ say. ζ). Let f (X) = X n − a ∈ K[X] be irreducible and separable over K. in fact. . u)/K). . . and so rβα = rα + rβ . Since every element of Gal(K(u)/K) is determined by its eﬀect . Let K be a ﬁeld with char K n and which contains a primitive nth root of 1. We call these simple radical extensions and later in Deﬁnition 6. The Galois Correspondence identiﬁes the following towers of subﬁelds and subgroups. In the more general situation of Proposition 6. . Hence the function ρ : Gal(K(u)/K) −→ ζ . Definition. {id} Gal(K(ζ.19. u)/K(ζ)) Gal(K(ζ. . Simple radical extensions In this section we will investigate splitting ﬁelds of polynomials of the form X n − a. k Note that in this deﬁnition we do not require the polynomials X n − aj ∈ K[X] to be irreducible. K(ζ. K(u) is a splitting ﬁeld of X n − a over K. Then L/K is a simple nKummer extension if L = K(u) where un = a for some a ∈ K. u) a Gal(K(ζ. 6. u)/K) 7 K(ζ) f } Gal(K(ζ. βα(u) = β(ζ rα u) = ζ rα β(u) = ζ rα ζ rβ u = ζ rα +rβ u. Clearly the roots of X n − a are distinct and so K(u)/K is separable over K. . . un = ak for some elements a1 . where char K n. u)/K(ζ)) 8 ' K {id} 6. . where Gal(K(ζ. Lagrange’s Theorem implies that the image of ρ has order dividing n.20. The Galois group Gal(K(u)/K) is cyclic of order n with a generator σ for which σ(u) = ζu. L/K is an (iterated ) nKummer extension if L = K(u1 . Proof. Then in K[X] we have X n − a = (X − u)(X − ζu) · · · (X − ζ n−1 u).33 we introduce a more general notion of radical extension. . 1 . As ζ is cyclic of order n. u)/K(ζ)) is cyclic and Gal(K(ζ)/K)) ∼ Gal(K(ζ. n − 1. Suppose that un = a ∈ K.4.21. ζ. Let K(u)/K be a simple nKummer extension. If K contains a primitive nth root of 1. . Notice that for β ∈ Gal(K(u)/K). 6.
soluble or soluable if there is a chain of subgroups (called a subnormal series) {1} = G G −1 i .5.18. so by Hilbert’s Theorem 6. Also note that X n − an = (X − a)(X − ζa) · · · (X − ζ n−1 a) = (X − a)(X − σ(a)) · · · (X − σ n−1 (a)). Since K(a) n = [K(a) : K] and therefore [K(a) : K] = [E : K] = n. We have ( 2)4 − 4 = 0. 6.23. We have NE/K (ζ −1 ) = ζ −n = 1.6. −1} Here is a converse to Proposition 6. 1. √ √ 6. 2)/Q(i) is a simple 4Kummer extension with Gal(Q(i. this shows that [E : K] = n ··· G1 G0 = G in which Gk+1 Gk and each composition factor Gk /Gk+1 is abelian.22. Gal(K(u)/K) is cyclic since every subgroup of a cyclic group is cyclic. n − 1 are distinct. so they must be the n conjugates of a. but X 4 − 4 = (X 2 − 2)(X 2 + 2). If each composition factor is a cyclic group of prime order the subnormal series is called a composition series.26. . E. there is an element a ∈ E for which ζ −1 = aσ(a)−1 . √ 6. SOLVABILITY AND RADICAL EXTENSIONS 85 on u. 6. hence  Gal(K(u)/K) divides n.25. Let n 1 and q ∈ Q. . Then σ(a) = ζa and the elements σ k (a) = ζ k a for k = 0. A group which is not solvable is called insolvable. Proposition. Q(i.√2 (X).24. In fact. . . 6. Example. Hence E/K is a simple nKummer extension. see [3. whence K(a) = E. Example. Suppose that char K n and there is an element ζ ∈ K which is a primitive nth root of unity. 2)/Q(i)) cyclic of order 2. A group G is solvable.5. hence an ∈ K since it is ﬁxed by σ. Then Q(ζn . If E/K is a ﬁnite Galois extension with cyclic Galois group of order n. . Definition. 5] for further details. and X 2 − 2 = minpolyQ(i). we usually write {1} = G G −1 · · · G1 G0 = G. then there is an element a ∈ E such that E = K(a) and a is a root of a polynomial of the form X n − b with b ∈ K. n q)/Q(ζn ) is a simple nKummer extension. Solvability and radical extensions We begin by recalling some ideas about groups. √ Proof. √ The corresponding group homomorphism ρ : Gal(Q(i)( 2)/Q(i)) −→ i has image im ρ = {1. ρ is injective. Proof.
Let G be a ﬁnite pgroup. For simplicity. Remark. (ii) If N G and G/N are solvable then so is G. ak−1 ) k for some dk of K. Then E/L is a radical Galois extension. where p is a prime. . so L(ζ.30. so beware when reading other sources! 6. Let G be a ﬁnite abelian group.27. a2 . But L(ζ. 6. Then X d − a has a d distinct roots in E.86 6. Then G is solvable. . u1 . u. Proposition. 6. Thus every element of L is expressible in terms of iterated roots of elements We will need the following Lemma and its Corollary. The primes appearing as well as the number of times each occurs are all determined by G. In the opposite direction we can sometimes see that a group is insolvable. u)/L is a radical Galois extension. Let G be a group.u (X) ∈ L[X]. 6. .u (X) in E. ut ).32. Example. (iii) G has a composition series in which one of the terms is a nonabelian simple group. Proposition.34. ut ∈ E are the distinct roots of minpolyK. It is a standard result that we can always reﬁne (i. . Let L/K be a ﬁnite Galois extensions and let L(u)/L be a radical extension. only their order varying for diﬀerent composition series. Let E/L be an extension where E is a splitting ﬁeld for the polynomial minpolyK. so there are d distinct dth roots of unity in E.33. 6.31. In particular. . . L/K is a radical extension of K if it has the form L = K(a1 . Then G is solvable. a2 . we assume also that char K = 0.28. then E = L(ζ. Hence there is a primitive dth root of unity ζ ∈ E and the subﬁeld L(ζ. . Definition.. Proof. if u = u1 .e. . 5. In fact. . A GALOIS MISCELLANY 6. 6. Let K be a ﬁeld and L/K a ﬁnite extension. add extra terms) a subnormal series of a solvable group to obtain a composition series. However. = Now we explain how this relates to ﬁelds and their extensions. . For n 5. for a ﬁnite pgroup G. u) E is normal over L. 1. several text books make subtle errors or omissions related to this result. if L/K is a radical Galois extension then so is E/K. . (i) If G is solvable then every subgroup H G and every quotient group G/N is solvable. . .29. (ii) G has a quotient group which is a nonabelian simple group. Lemma. This follows from the fact that if n quotient group Sn /An ∼ Z/2. . . Then G is insolvable if any of the following conditions holds: (i) G contains a subgroup which is a nonabelian simple group. Recall that a group is simple if it has no nontrivial proper normal subgroups. there is always a normal subgroup of a pgroup with index p. . An is a simple group and An Sn with Proof. Suppose that ud = a ∈ L with a = 0. But this is clearly a radical extension of L. Example. Example. . an ) with adk ∈ K(a1 . u)/K need not be Galois. the alternating and symmetric groups An and Sn are insolvable. 6. so in this case we can assume each quotient Gk /Gk+1 is cyclic of order p. According to [4]. and if v is any other root then (v/u)d = 1. Let G be a ﬁnite group.
If r 1 Gr Gal(E /K) and (E )Gr = K(ζ. each extension (E )Gk+1 /(E )Gk is radical. 6. If L/K is a radical extension then it is contained in a radical Galois extension L /K. In the next deﬁnition. u1 . ζ. this is proved by induction on n using Lemma 6.22.37. Theorem. Gal(E /K)/ Gal(E /E) ∼ Gal(E/K) which is = solvable. . . L where L /K is a ﬁnite radical Galois extension 6. ur )/K(ζ. . Let {1} = G G −1 Gm−1 ··· G0 Gal(E /K) · · · G1 G0 = Gal(E /K(ζ)) be a composition series. . . 3 2)/Q is solvable.30. where ζ d = 1. . . Example. . is also solvable by Proposition 6. Hence Gal(E /K) is solvable. and since Gal(E/K) is a quotient group of Gal(E /K). m with d1 · · · dm  d. . √ 6. . ud1 ∈ K(ζ) and udr ∈ K(ζ. . Similarly. Let E/K be a ﬁnite Galois extension. ur−1 )). Writing L = K(a1 . .36. so Gal(E /K) is solvable by Proposition 6. Clearly K(ζ)/K is radical.6. The extension (E )G1 /K(ζ) is radical by Proposition 6. . Suppose that E E where E /K is a ﬁnite radical Galois extension. . ur ). Corollary. u1 . Now Gal(E /E) Gal(E /K) and by Theorem 6. u1 . SOLVABILITY AND RADICAL EXTENSIONS 87 If L/K is a radical Galois extension. then {1} = Gm and Gr−1 /Gr ∼ Gal(K(ζ.30. . u1 . . then E = L(a1 . a2 . The Galois group of the extension Q(ζ3 .38. . . as is E /K.35. . . so E = K(ζ. . . . . . with (E )G0 = K(ζ). . Hence E /K(ζ) is radical. .7. which is a radical Galois extension of K. . ur−1 ) for r = 2. Proof.25. an ) as in Deﬁnition 6. ut ). . Then Gal(E /K(ζ)) Gal(E /K) is solvable. Definition. . . 6. Proof. . Gal(E /E) is abelian. 6. L/K is solvable if L of K. Definition.39. . Also. . . Now suppose that Gal(E/K) is solvable and let n =  Gal(E/K). say L = K(a1 . We will now show that E /K is a radical extension. so E contains a primitive nth root of unity ζ and therefore it contains a primitive dth root of unity for every divisor d of n. Then E/K is solvable if and only if the group Gal(E/K) is solvable. . um ). = which is abelian by Proposition 6. u1 .34. . . then f (X) is solvable by radicals over K if L is contained in a radical (Galois) extension of K. . an . Let E be the splitting ﬁeld of X n − 1 over E. If L is the splitting ﬁeld of a polynomial f (X) ∈ K[X]. the word Galois is superﬂuous because of the preceding results. .5. an ). . u1 . u.33.
Then complex conjugation restricts to an element of order 2 in Gal(E/Q) which interchanges the nonreal roots and ﬁxes the others. Example. so by Cauchy’s Lemma this group contains an element of order 5. Here is an explicit example over Q. Sn and H contains the elements (1 2) and The proof is left as an exercise. Let E C be the splitting ﬁeld of f (X) over Q. Lemma. hence there are three real roots of f (X) and two nonreal complex ones. We should therefore expect to look for a polynomial of degree at least 5 to ﬁnd a Galois group for a splitting ﬁeld to be simple or occur as a composition factor of such a Galois group. Using the Eisenstein Test 1. To ﬁnd examples. u4 . It is also interesting to reverse the question and ask whether there are extensions which are not solvable. 3 2) which we list in the order √ 2 √ √ 3 3 3 2. Clearly Q(ζ3 . then the transposition (1 2) ∈ S5 corresponds to this element.30 and 4. √ 5 S3 Q(ζ3 .41. 5 divides the order of Gal(E/Q). 3 2) is a radical extension of Q and √ √ 3 3 Q(ζ3 . If we list the roots of f (X) as u1 . There are two turning points. This was a famous problem pursued for several hundred years. we ﬁrst recall that the smallest nonabelian simple group is A5 which has order 60. Let n 1. We have already studied this extension in Example 3. 5 3 √ 3 2)/Q(ζ3 )) Q y ) {id} Here Q(ζ3 )/Q is itself a Galois extension and A3 S3 . 2) = Q(ζ3 )( 2). namely a maximum at x = 0 and a minimum at x = 28. This completes the veriﬁcation of Example 6. Suppose that H (1 2 · · · n). Notice that A3 ∼ Z/3 and S3 /A3 ∼ Z/2. We have the following towers of subﬁelds and subgroups related under the Galois Correspondence. √ We know that Gal(Q(ζ3 . 3 2)/Q) ∼ S3 . 2 ζ3 . Then f (0) = 7 > 0 > f (28) = −4302585. By Theorem 4. we ﬁnd that f (X) is irreducible over Q. A GALOIS MISCELLANY √ Proof. f (X) = 20X 4 − 420X 2 = 20X 2 (X − 21).40. the only elements of S5 of order 5 are 5cycles. by taking an appropriate power we can assume that there is a 5cycle of the form (1 2 3 4 5) corresponding to an element of Gal(E/Q) which we can view as a subgroup of S5 . u3 . = = so we have the following composition series for S3 : {id} A3 S3 . Furthermore. The next lemma shows that Gal(E/Q) ∼ S5 . 3 2) e 3 2 Q(ζ3 ) = Q(ζ3 . 6.40. = 6. u5 with u1 . 2 √ A 3 2) 3 i A3 = Gal(Q(ζ3 .8(iii). u2 being the nonreal roots. where we identify each element of the Galois group with = √ a permutation of the three roots of X 3 − 2 in Q(ζ3 . Now observe that f (X) = 5X 4 − 140X 3 = 5X 3 (X − 28). 2 ζ3 . Proof. The splitting ﬁeld of the polynomial f (X) = X 5 − 35X 4 + 7 ∈ Q[X] is not solvable. Then H = Sn .20.38 with p = 7.88 6. . u2 .
Xn )/ Symn (k)) = Sn . . . . The ﬁeld of symmetric functions on n indeterminates is Symn (k) = k(X1 . SYMMETRIC FUNCTIONS 89 It is worth remarking that the most extreme version of this occurs when we ask for a Galois group which is simple. . . . . . . Xn ). then f (X1 . we also have n!  Gal(k(X1 . . Example. . . . Symmetric functions Let k be a ﬁeld. . . en ) Symn (k) we can also deduce that k(e1 . . = hence it is not solvable. Xn ) is the splitting ﬁeld of fn (Y ) over Symn (k). Xn )/ Symn (k)) = n! and therefore Gal(k(X1 . .e. . . . . Xn )Sn So if f (X1 . Xn )/ Symn (k)) Sn  = n!. . Xn ) : Symn (k)] =  Gal(k(X1 . . . . . eσ = ek . . Xn )/ Symn (k) is a ﬁnite Galois extension for which Gal(k(X1 . i1 <i2 <···<ik where 1 k n. . . en ]. . . . . . Xn )/k(e1 . . In fact. Here is an example whose Galois group is A5 . . . . . Since k(e1 . . . i. this proof shows that the extension k(X1 . . . . Hence every element of Symn (k) is a rational function in the ei . Xn ). . Xn ) by ring isomorphisms ﬁxing k and in the latter case it is by ﬁeld automorphisms ﬁxing k. . . . . . . .. . . . . . .42. . Working in k the ring k(X1 . It is easy to see that for every σ ∈ Sn . ek = Xi1 Xi2 · · · Xik . . en ) is Galois of degree n!. . 6. . . Remark. . . Varying σ we obtain actions of the group Sn on k[X1 . Xn ) ∈ Symn (k) ⇐⇒ ∀σ ∈ Sn f (X1 . Xn ) = f (Xσ(1) . . . . . . But as every element of Gal(k(X1 . . 6. Each permutation σ ∈ Sn acts on k[X1 . . . Definition. . So k(X1 . . . . . . . . Analogous results are true for polynomials. . . . Xn ] is a ring isomorphism. . Xn )/ Symn (k)) ∼ Sn . en ) = Symn (k). . The Galois group of f (X) = X 5 +20X+16 over Q is Gal(Q(f (X))/Q) ∼ A5 . . . . . . Xn ) = f σ (X1 . . so ek ∈ Symn (k). . . . . . . . .6. .26. . hence [k(X1 . . . . . The extension k(X1 . . . Xn )/ Symn (k)). . 6. There has been a great deal of research activity on this question in the past few decades. k[X1 . . Xn ] ⊆ k(X1 . . . Xn ) ∈ k(X1 . . . Viewed as a function σ· : k[X1 . . this extends to a ring isomorphism σ· : k(X1 . . .6. Xσ(n) ). . . Xn ] and its ﬁeld of fractions K = k(X1 . Xn ). . . Xn )[Y ] we have fn (Y ) = Y n − e1 Y n−1 + · · · + (−1)n−1 en−1 Y + (−1)n en = 0. Xn ]Sn = k[e1 . . .44. . Xσ(n) ). . . . . . . this is veriﬁed using Proposition 4. . . 6. . . . Xn ) = f (Xσ(1) . . . = Proof.6.43. 6. Xn ). . . Consider the polynomial ring on n indeterminates k[X1 . Combining these inequalities we obtain  Gal(k(X1 . but apparently not all simple groups are known to occur as Galois groups of extensions of Q or other ﬁnite subextensions of C/Q. Theorem. . hence the roots of this polynomial are the Xi . k(X1 . . Xn ) −→ k(X1 . . . There are elements of k[X1 . Xn ] and k(X1 . Now Sn Gal(k(X1 . . Xn ] −→ k[X1 . . Xn ] by σ · f (X1 . . Xn )/ Symn (k)). . .45. . . Xn ) called elementary symmetric functions. . . . Xn )/ Symn (k)) permutes the roots of fn (Y ) and is determined by this permutation.
6. Q( 5 i). Let K be a ﬁeld for which char K = 2 and n 1 be odd. . Corollary. Show that p−1 ξ = (−1) (b) Deduce that ξ2 = p −p 2 (p−1)/2 r=1 r (1 − ζp ). . .7. (b) Determine the group (Z/20)× and describe the eﬀect of each of its elements on Q(ζ20 ) and Q(cos(π/10)) under the action described in Theorem 6. Using this. if p ≡ 3 (mod 4). Q( 3 i).2. A GALOIS MISCELLANY 6.5. if p ≡ 1 (mod 4). (b) Determine the minimal polynomial of cos(2π/5) over Q. Find all values of n in the following ﬁelds: 1 for which ϕ(n)  4. Hence show that √ −1 + 5 cos(2π/5) = . In this question. let p be an odd prime and let ζp = e2πi/p ∈ Q(ζp ) (a) Consider the product (p−1)/2 C. Let p > 0 be a prime and G a group of order G = pn for some n 1. the extension k(X1 . determine which roots of unity lie √ √ √ Q(i). show that then K contains a primitive 2nth root of unity.46.6. 6. ξ= r=1 r −r (ζp − ζp ) ∈ Q(ζp ). If K contains a primitive nth root of unity.4. Xn )/ Symn (k) is not solvable.2. (a) Describe the Galois group Gal(Q(ζ5 )/Q) and its action on Q(ζ5 ).1.90 6. . If n 5. Let n 1. Show by induction on n that there is a normal subgroup N G with N  = pn−1 .3.2. Exercises on Chapter 6 6. 6. 6. 6. . 4 For which other angles θ is cos θ a root of this minimal polynomial? What is the value of sin(2π/5) ? (c) Find the tower of subﬁelds of Q(ζ5 ) and express them as ﬁxed ﬁelds of subgroups of Gal(Q(ζ5 )/Q). [Hint: what do you know about the centre of G? Use this information to produce a quotient group of smaller order than G. Describe the eﬀect of each element on Q(ζ24 ) and Q(cos(π/12)) under the action described in Theorem 6.] 6. (a) Describe the elements of (Z/24)× explicitly and verify that this group is isomorphic to Z/2 × Z/2 × Z/2. Q( 2 i). In this question. (a) What can you say about sin(2π/n) and Gal(Q(sin(2π/n))/Q))? (b) Determine sin(π/12) and Gal(Q(sin(π/12))/Q)). work in the cyclotomic ﬁeld Q(ζ5 ) where ζ5 = e2πi/5 .
the sum of all the monomials in the Xi of degree k. (b) If v ∈ E has TrE/K (v) = 0. (a) For n 1 and 1 k n. see Theorem 6. Xn ]Sn is deﬁned by sk = 1 i n Xik . .41. using TrE/K in place of NE/K .10. (b) For n 1 and 1 k n. Prove Lemma 6. h2 . . [Hint: show that every 2cycle of the form (i i + 1) is in H by considering elements of the form (1 2 · · · n)r (1 2)(1 2 · · · n)n−r . e2 . takes values in K and use this to deﬁne a Klinear mapping TrE/K : E −→ K. . (i) For large values of n. 6. e3 . Prove the formula sk = e1 sk−1 − e2 sk−2 + · · · + (−1)k−1 ek−1 s1 + (−1)k kek . T (u) = u + σ(u) + σ 2 (u) + · · · + σ n−1 (u). . the total symmetric function is deﬁned by hk = j1 j2 ··· jk Xj1 Xj2 · · · Xjk . show that there is a w ∈ E such that v = w − σ(w)..] 6. express h1 . and adapt the proof of Hilbert’s Theorem 90 in Theorem 6. the kth power sum sk ∈ k[X1 .18.9. √ √ and also p ∈ Q(ζp ) if p ≡ 1 (mod 4) and p i ∈ Q(ζp ) if p ≡ 3 (mod 4).91 (c) Conclude that ξ= if p ≡ 3 (mod 4).] 6. Let E/K be a Galois extension with cyclic Galois group Gal(E/K) = σ of order n.e. (a) Show that the function T : E −→ E. i.8. h3 in terms of the elementary symmetric functions e1 . This question is about an additive version of Hilbert’s Theorem 90. .18. then consider w= 1 (TrE/K t) vσ(t) + (v + σ(v))σ 2 (t) + · · · + (v + σ(v)σ 2 (t) + · · · + σ n−2 (v))σ n−1 (t) √ ± p √ ± pi if p ≡ 1 (mod 4). [Hint: Show that there is an element t ∈ E for which TrE/K t = 0. (ii) Show that the power sum functions sk of the previous question satisfy sk = −(h1 sk−1 + h2 sk−2 + · · · + hk−1 s1 ) + khk .
.
Artin. W. Algebra. A counterexample in Galois theory. Galois Theory. Chapman and Hall (1989). SpringerVerlag (1998). Cambridge University Press (1997). ISBN 0 486 62342 4. [7] J. [Highly recommended. ISBN 0 412 345501. Dover Publications (1998). [2] JP. ISBN 0 201 55540 9. Niederreiter. [Highly recommended ] [4] T. Hungerford. [5] S. Galois Theory. Fraleigh. Stewart. ISBN 0387987657. especially for its historical notes] [3] J. B. ISBN 0 521 39231 4. American Mathematical Monthly 97 (1997). ISBN 0 201 33596 4. [Very highly recommended.] 93 . [6] R.Bibliography [1] E. Galois Theory. Addison Wesley (1993). Lidl & H. 54–57. Addison Wesley (1999). A First Course in Abstract Algebra. [8] I. SpringerVerlag. Galois theory. Lang. ISBN 0 387 98541 7. Finite Fields. New York (2001). Escoﬁer. Rotman.
.
ϕ is not surjective. then for every r ∈ R. so ker η = ker η and therefore char S = char R.5. v ∈ R with uv ∈ ϕ−1 Q. . η (n) = η(n). R is an integral domain with characteristic subring Z and char R = 0. 1. hence u ∈ ϕ−1 Q or v ∈ ϕ−1 Q. 1. v ∈ R and R is an integral domain. 1. For R = Fp [X].1. (b) Q is a ﬁeld and Z ⊆ Q is a subring which is not a ﬁeld.6. A[X] is an integral domain if and only if A is an integral domain. Consider the unit homomorphisms η : Z −→ R and η : Z −→ S. n n so {n ∈ Z : n > 0 and n1 = 0} ⊆ {n ∈ Z : n > 0 and nr = 0 for all r ∈ R}. then the characteristic subring of Matn (A) is the same as that of A and char Matn (A) = char A. nr = r + · · · + r = (1 + · · · + 1)r = (n1)r = 0r = 0. (a) For any subring R ⊆ C. If n > 1 then Matn (A) is not commutative. so ker ϕ = (0). The only proper ideal of k is the zero ideal (0). Now by deﬁnition of characteristic.2. while for R = Fp [X]/(X 2 ). in any case it always has zerodivisors since any singular matrix is a zerodivisor.Solutions Chapter 1 1. (c) If we identify A with the subring of scalar matrices in Matn (A). (d) We have P ⊆ Q ∩ R R with P R maximal. Suppose that u. then ϕ(u)ϕ(v) = ϕ(uv) ∈ Q and so ϕ(u) ∈ Q or ϕ(v) ∈ Q. Clearly {n ∈ Z : n > 0 and nr = 0 for all r ∈ R} ⊆ {n ∈ Z : n > 0 and n1 = 0}. Hence these sets are in fact equal. In fact Q only needs to be a proper ideal of S for this argument to work. so P ⊆ Q ∩ R. The main thing to check is that ϕ(u + v) = ϕ(u) + ϕ(v) which is a consequence of the Idiot’s Binomial Theorem. then u = 0 or v = 0 since u. 1. (a) Addition and multiplication follow from the obvious formulae (u1 +v1 i)+(u2 +v2 i) = (u1 +u2 )+(v1 +v2 )i. (b) The characteristic subring of A[X] is the same as that of A and char A[X] = char A.4. (p)Z Z is a (maximal) ideal that properly contains (0)Z . char R = min{n ∈ Z : n > 0 and n1 = 0} = min{n ∈ Z : n > 0 and nr = 0 for all r ∈ R}. If 0 < n ∈ Z and n1 = 0. (c) Consider Z ⊆ Q. 1. (a) Let u. then the zeroideal (0)Q Q has (0)Q ∩ Z = (0)Z Z but this is not maximal in Z since for any prime p > 0. (a) Recall from the Isomorphism Theorems of basic Ring Theory that ϕ−1 Q R. 1. ϕ is not injective. (b) Consider the inclusion function inc : R −→ S.7. v ∈ S and suppose that uv = 0. Then for n ∈ Z. then inc−1 Q = Q ∩ R.3. we need to show it is a prime ideal. (u1 +v1 i)(u2 +v2 i) = (u1 u2 −v1 v2 )+(u1 v2 +u2 v1 )i. so this result follows from (a). When char R = p > 0 they must both be nonempty.
b ◦ψc.b (p(X)).b (k(X))p(X) −1 −1 and as p(X) is prime. ψa. Hence ψa. To see that Q[i] is a ﬁeld.b (g(X))ψa. this shows that deg ψa. Then inc∗ (u + vi) n = inc∗ (u + vi) n + 0i = 1 (u + vi). so by the uniqueness part of the Homomorphism Extension Property. n (b) (i) If f (X) = i=0 ci X i ∈ k[X] with ci ∈ k and cn = 0. Now n ψa. ψa−1 . then deg f (X) = n. (a) Addition and multiplication are given by the usual formulae ∞ ∞ ∞ ∞ ∞ ∞ k ( k=0 ak X ) + ( k=0 k bk X ) = k=0 k (ak + bk )X .b (p(X))  h(X) and so ψa. Since cn an = 0.b (p(X))  g(X)h(X) for g(X).−ba−1 (aX +b) = X and ψa. notice that if u + vi = 0 with u.b is an isomorphism. k ( k=0 ak X )( k=0 k bk X ) = k=0 k ( =0 a bk− )X k .b (p(X)) is prime.9. 1. v ∈ Q.b (h(X)).b (g(X)) or p(X)  ψa. Hence every nonzero element of Q[i] has an inverse. 1 (b) & (c) The crucial point is that every element of Q[i] can be written as (u + vi) with n n. (a) Existence and uniqueness of such an ψa. 1.b (f (X)) = deg f (X). v ∈ Z and n = 0.8.b (f (X)) = i=0 ci (aX + b)i = cn an X n + terms of lower degrees in X. so they are subrings of the ﬁeld C.30.b (a−1 X − ba−1 ) = X.−ba−1 = id = ψa−1 . By the uniqueness part of the Homomorphism Extension Property. (u − vi)(u + vi) = (u + vi)(u − vi) = u2 + v 2 = 0.b . −1 Therefore these are inverse isomorphisms. If a is a unit then ψa−1 .−ba−1 ◦ ψa.b .b follow from the Homomorphism Extension Property 1. by Qu. Choose k(X) ∈ k[X] so that g(X)h(X) = k(X)ψa. we have −1 −1 −1 ψa. .b ◦ ψc. u. (ii) Suppose that ψa.1.b (h(X)) = ψa. n so so the latter element is in the image of inc∗ which must therefore be a surjection. u v + i ∈ Q(i) u2 + v 2 u2 + v 2 is the inverse of u + vi.d = ψca.−ba−1 = ψa.b (f (X)) = f (aX + b) = i=0 ri (aX + b)i . (iii) This follows from (ii) and Proposition 1.cb+d . they are both integral domains. h(X) ∈ k[X]. we have ψa. 1.−ba−1 : R[X] −→ R[X] has the property that ψa−1 . p(X)  ψa. We have ψa.22 and its eﬀect on f (X) = n ri X i ∈ R[X] where ri ∈ R is i=0 n ψa.b ◦ ψa−1 .b (p(X))  g(X) or ψa.d (X) = ψa. therefore Q[i] is a ﬁeld.cb+d (X).96 SOLUTIONS with Z[i] and Q[i] both closed under these operations and containing 1 = 1 + 0i as a unity. Since ψa.b (cX + d) = c(aX + b) + d = caX + (cb + d) = ψca.
So f (X) must be irreducible. =0 a bn− = a0 bn + a1 bn−1 + · · · + an bn = 0 (n 1). So X 3 + X − 1 . Then a has an inverse in k[[X]] only if there is a b = ∞ k=0 b X ∈ k[[X]] with ab = 1. X 3 + 2X + 1 ≡ X 3 − X + 1 (mod 3) so this polynomial reduces modulo 3 to an irreducible and so is irreducible. where deg g(X) < deg f (X) = deg f (X) and deg h(X) < deg f (X) = deg f (X). Here f (X) = (3X − 3)d(X) + (−9X + 7). (c) We can deﬁne make the set k((X)) of all such ﬁnite tailed Laurent series into a ring with addition and multiplication deﬁned by ∞ ∞ ∞ ( k=k1 ak X k ) + ( k=k2 ∞ ∞ bk X k ) = ∞ (ak + bk )X k . b ∈ k[[X]] we may write ∞ ∞ a= k=k0 ak X . 1. Here f (X) = −X 3 − X 2 + X + 1 and d(X) = −X 3 − X with f (X) = d(X) + (−X − X 2 + 1) = d(X) + (2X 2 + 2X + 1). r=0 ak X k ∈ k((X)) with The inclusion inc : k[[X]] −→ k((X)) extends to the monomorphism inc∗ : Fr(k[[X]]) −→ k((X)) for which ∞ ∞ r r=0 ar+k0 X inc∗ =( ar+k0 X r )X k0 . in particular this forces a0 = 0 since otherwise the lowest term in X in ab would be of degree greater than 0. if a0 = 0. ∞ k=k0 Clearly k[[X]] ⊆ k((X)) is a subring. Notice that every element k0 < 0 can be written as ( ar+k0 X r )X k0 . X 3 + X − 1 reduces modulo 2 to an irreducible since it has no roots modulo 2. X 3 − X + 1 reduces modulo 3 to an irreducible since it has no roots modulo 3. Conversely.10. then f (X) = ρ(g(X)h(X)) = ρ(g(X))ρ(h(X)) = g(X)h(X). So X 3 − X + 1 is irreducible. But this is impossible since f (X) is irreducible.97 Clearly k[X] ⊆ k[[X]] is a subring. h(X) ∈ Z[X]. ∞ k (b) Let a = k=0 ak X ∈ k[[X]]. is a ring homomorphism. If f (X) = g(X)h(X) with g(X).12.k2 } k ( k=k0 ak X k )( = 0 b X )= ( k=min{k0 . k b= 0 bX with ak0 = 0 = b 0 . ρ(f (X)) = f (X). Then the lowest degree term in ab is ak0 b 0 X k0 + 0 with ak0 b 0 = 0. 1. 0 } j=0 a bk−j )X k . So k[[X]] is an integral domain. The reduction modulo p function ρ : Z[X] −→ Fp [X]. X −k0 r=0 so inc∗ is surjective. deg g(X) < deg f (X) and deg h(X) < deg f (X). Hence ab = 0. 1. k=min{k1 . to ensure that ab = 1. Given two nonzero elements a. then we can inductively solve the system of equations n a0 b0 = 1.11.
c. √ Similarly. c. Then εω Q[X] = Q[ω] = {a + bω : a. then √ 4 εα R[X] = R[α] = {a + bα : a. I2 = (X 2 +2). I6 = X 2 +X +1. b ∈ R} = R. If α = 4 2. b ∈ R} = R. then α ∈ R and / εα R[X] = R[α] = {a + bα : a. α i.16. ker εω = (X 2 + X + 1) Q[X]. b. I1 = (X 2 +1). 1. The image is The image of ε−√2 √ √ ε√2 Q[X] = Q[ 2] = {a + b 2 : a. I3 = (X 2 −2). where (X 2 + X + 1) Q[X] is a maximal ideal. so the evaluation homomorphism εω2 has εω2 Q[X] = εω Q[X] and ker εω2 = ker εω .15. I4 = (X − 2). If α2 + 2 = 0. if α = − 4 2. b ∈ Q}.14. The other complex roots of X 4 − 2 are −α. We have εα Q[X] = Q[α] = {a + bα + cα2 + dα3 : a. b ∈ R} = C. √ 1. First change variable to obtain g(X) = f (X + 3) = X 3 − 6X + 4. . then √ 4 ε−α R[X] = R[−α] = {a − bα : a. b. I5 = (X 2 +2). which is a maximal ideal. ker εα = (X 2 + 2) R[X]. X 5 + X − 1 = (X 3 + X 2 − 1)(X 2 − X + 1) and 5X 3 − 10X + X 2 − 2 = (5X + 1)(X 2 − 2) so neither of these is irreducible. The other complex root of X 2 + X + 1 is ω 2 . b ∈ Q}. We have ker ε√2 = ker ε−√2 = (X 2 − 2) Q[X] √ 1. ker εα = (X − 2) R[X].13. d ∈ Q}. √ is ε−√2 Q[X] = Q[ 2] = ε√2 Q[X]. Using Cardan’s method we have to solve the quadratic equation U 2 + 4U + 8 = 0. Notice that ω = (−1 + 3i)/2 = ζ3 is a primitive 3rd root of unity and is a root of the irreducible polynomial X 2 + X + 1 ∈ Q[X].17. Then ker ε−α = ker εα i = ker ε−α i = (X 4 − 2) Q[X] but although ε−α Q[X] = Q[α]. −α i (notice that two of these are real while the other two are not). ker ε−α = (X + 2) R[X].98 SOLUTIONS is irreducible. which has roots √ −2 ± 2i = ( 2)3 e3πi/4 . and the latter ideal is maximal. √ √ √ √ √ 4 4 X 4 − 2 = (X 2 − 2)(X 2 + 2) = (X − 2)(X + 2)(X 2 + 2). ker εα = (X 4 − 2) Q[X]. √ Let α be a root of X 4 − 2. 1. we have εα i Q[X] = ε−α i Q[X] = Q[α i] = {a + bα i + cα2 + dα3 i : a. If we replace Q by R. X 5 − X + 1 is irreducible modulo 3 and 5 so is itself irreducible. 1. then in R[X]. d ∈ Q} = Q[α]. so εα i Q[X] = εα Q[X] since one of these is a subset of R but the other is not.
/ which gives a contradiction. 1. ck ∈ k} = k[X]. 3 − 1. . (a) We have X p − 1 = (X p so by (1. = 108. Φpj (X) = Φp (X p 0 j k k−1 k k−1 )p − 1 = (X p ) k−1 − 1)Φp (X p k−1 ). To see that the homomorphism Aﬀ 1 (k) −→ Autk (k[X]). Notice that 2 is a real root of this polynomial and f (X) = (X − 2)(X 2 + 2X + 10). If n = 0 then ϕk[X] = k ⊆ k[X] so ϕ would not be surjective. Thus β is a real root of g(X) = X 3 + X − 2 for which 1 is also a root and g(X) = (X − 1)(X 2 + X + 2). since cyclotomic polynomials are irreducible. Φpj (X). 1. − 3 + 2. 0 j k−1 . . Then ϕk[X] = {c + 0 + c1 ϕ(X) + · · · + ck ϕ(X)k : c0 . So we must have n = 1. we ﬁnd that Φ20 (X)  (X 2 + 1)(X 8 − X 6 + X 4 − X 2 + 1). But if k > 0 and ck = 0 then deg(c + 0 + c1 ϕ(X) + · · · + ck ϕ(X)k ) = kn > 1. 1. − 3 − 1.19. .22. 2 √ √ √ √ For the roots of g(X) we obtain 2. For β. Let ϕ(X) = a0 + a1 X + · · · + an X n with ai ∈ k and an = 0. while for f (X) we have 5. A −→ αA−1 . 1.21. For α. Thus α is a real root of f (X) = X 3 + 6X − 20. take q 27q 2 + 4p3 28 − = 1. . Therefore α = 2. we must have Φ20 (X) = X 8 − X 6 + X 4 − X 2 + 1. 2 108 27 so q = −2 and p = 1. 1. Therefore ϕ(X) = a0 +a1 X and so ϕ = αA for some A ∈ Aﬀ 1 (k).99 Thus we can take √ √ πi/4 r 2 u = 2e ω = √ (1 + i)ω r = (1 + i)ω r (r = 0. 2 108 so q = −20 and p = 6. Work backwards with Cardan’s method. take q 27q 2 + 4p3 − = 10. 3). so X ∈ ϕk[X]. Since the roots of X 10 − 1 are the 10th roots of unity. Calculation. Therefore β = 1.60 is surjective. Suppose that show that n > 1. We have deg Φ20 (X) = ϕ(20) = ϕ(4)ϕ(5) = 2 × 4 = 8 and X 20 − 1 = (X 10 − 1)(X 10 + 1) = (X 10 − 1)(X 2 + 1)(X 8 − X 6 + X 4 − X 2 + 1). where X 2 + X + 2 has no real roots.5). where X 2 + 2X + 10 has no real roots. 1. 3 + 2.20. hence we must have n 1. = . suppose that ϕ ∈ Autk (k[X]) is any automorphism.18. described in the Proof of Example 1. c1 .
..n)=1 t (X − ζn ) = (−1)ϕ(n) X −ϕ(n) Φn (X)...n)=1 = (−1)ϕ(n) X −ϕ(n) t=1. (c) First notice that r r deg Φn (X) = ϕ(n) = (p1 − 1) · · · (pk − 1)p11 −1 · · · pkk −1 .n−1 gcd(t. c0 = Φpk (1) = Φp (1) = c0 = p.. 1.n)=1 n−t (1 − Xζn ) t=1.100 k−1 SOLUTIONS and therefore Φpk (X) = Φp (X p ). so Φn (X −1 ) = = X −ϕ(n) = X −ϕ(n) = X −ϕ(n) = X −ϕ(n) t (X −1 − ζn ) t=1.. This shows that Φn (X)  Φp1 ···pk (X the same degree they are equal. and deg Φp1 ···pk (X p1 r1 −1 ···pkk r −1 r r ) = ϕ(p1 · · · pk )pr1 −1 · · · prk −1 = (p1 − 1) · · · (pk − 1)p11 −1 · · · pkk −1 .n−1 gcd(t. Since 2  ϕ(n) when n > 2 and the result is immediate when n = 2.. As these are monic polynomials of t (X − ζn ). k−1 k−1 (mod p) k−1 Φpk (X) = (X − 1)(p−1)p k−1 + cp−2 (X − 1)(p−2)p + · · · + c1 (X − 1)p k−1 + c0 . so the Eisenstein Test can be applied to show that Φpk (X) is irreducible over Q. r1 −1 r −1 (ξ and no smaller power of (ξ p1 r1 −1 r −1 r −1 p11 ···pkk r −1 ···pkk ) has this property. By Theorem 1.. (b) Using the formula of Equation 1. we have Φpk (X) = Φp (X p k−1 ) = (X p k−1 − 1)p−1 + cp−2 (X p − 1 ≡ (X − 1)p k−1 − 1)p−2 + · · · + c1 (X p k−1 − 1) + c0 ..n−1 gcd(t....n−1 gcd(t. each root ξ of Φn (X). In fact..n)=1 t (ζn − X) t=1..43.. Φn (X) = t=1. hence (ξ p1 p11 r −1 ···pkk ) is a root of Φp1 ···pk (X).n)=1 t (1 − Xζn ) t=1. ...23. )p1 ···pk = ξ n = 1... we see that desired equation always holds.n−1 gcd(t.. The Idiot’s Binomial Theorem gives Xp so where cr ≡ 0 (mod p)......n)=1 ···pkk r −1 ).4. so the roots of Φpk (X) are their pk−1 st roots which are the primitive pk th roots of 1.n−1 gcd(t. 1 k r1 −1 so deg Φn (X) = deg Φp1 ···pk (X p1 ···pkk r −1 )..n)=1 −t (1 − Xζn ) t=1.n−1 gcd(t. Also. where cr ≡ 0 (mod p) and c0 = p... The complex roots of Φp (X) are the primitive pth roots of 1.
b2 ∈ Z. / 2. Arrange the induction carefully. 1. We also have Φ5 (X) = X 4 + X 3 + X 2 + X + 1.1. If a = 0 then q = b p. −1 −2 2 ζ5 + ζ5 = (ζ5 + ζ5 )2 − 2 = 4 cos2 (2π/5) − 2. Then √ √ q = (a + b p)2 = (a2 + b2 p) + 2ab p. by the Idiot’s Binomial Theorem 1. Thus a suitable polynomial is 4X 2 + 2X − 1 ∈ Q[X]. b ∈ Q. This is similar to Example 2. Writing b1 = b1 q and b2 = b2 q for suitable b1 . It is obvious that [Q( p.33. the only root of this polynomial in K must be 1. We have −1 ζn + ζn = e2πi/n + e−2πi/n = (cos(2πi/n) + sin(2πi/n) i) + (cos(2πi/n) − sin(2πi/n) i) = 2 cos(2πi/n). so u is the only root in K. we obtain b2 q = b2 p 2 1 and so p  b2 and q  b1 . . 2. Chapter 2 2. we obtain (b2 )2 p2 q = (b1 )2 q 2 p.3. we have −2 −1 2 (ζ5 + ζ5 ) + (ζ5 + ζ5 ) + 1 = 0. By the Unique Factorization Property 1. a2 + b2 p = q. From this we obtain p  b1 and q  b2 . hence 4 cos2 (2π/5) + 2 cos(2π/5) − 1 = 0. so 3 2 4 ζ5 + ζ5 + ζ5 + ζ5 + 1 = 0. hence (b2 )2 p = (b1 )2 q. √ √ say q = a + b p for some a. X p − 1 = X p + (−1)p = (X + (−1))p = (X − 1)p .24. q) : Q( p)] = 1 then q ∈ Q( p).1. b2 ) = 1. Similarly. b2 ∈ Z and gcd(b1 . Writing b = b1 /b2 with b1 . if [Q( p. giving the simultaneous pair of equations 2ab = 0.4. √ √ √ If b = 0 then q ∈ Q which contradicts the result of Qu. (a) In K[X]. b2 ) = 1.25. As in (a) we have X p − a = X p − up = (X − u)p . So q ∈ Q( p). √ √ √ √ √ √ √ √ 2. (b) If u ∈ K is a root of this polynomial then up = a.101 1.2. q) : Q( p)] 2. Now we have −1 ζ5 + ζ5 = 2 cos(2π/5). m m X np − 1 = (X n − 1)p and the only roots of this must be nth roots of 1. −1 −2 4 3 Rearranging and using the formulae ζ5 = ζ5 . ζ5 = ζ5 . but then p  b1 as well as p  b2 . contradicting the fact that √ √ gcd(b1 .10.
then x + yi = 0 ⇐⇒ x = y = 0. So each of the numbers cos(2kπ/5) is a root of the polynomial T5 (X) − 1 = (X − 1)(4X 2 + 2X − 1)2 . we have [Q(i) : Q] 2. L2 = Q(i). therefore Q(cos(2π/5)) ∼ Q[X]/(4X 2 + 2X − 1). But also if x. so 1. 2. √ √ √ √ [Q( 3. L3 = Q( 3 i). [Q(ζ5 ) : Q] = [Q[X]/(Φ5 (X)) : Q] = ϕ(5) = 5 − 1 = 4. i ∈ Q( 3) and / √ √ √ and since i2 + 1 = 0. √ √ √ √ 2. C 2 R ∞ Q( 3) t tt tt t tt √ t 2 ∞ √ Q( 3. 2. i) : Q] = [Q( 3. i) = Q( 3)(i) has [Q( 3. Then [Lr ∩ Ls : Q] > 1 ⇐⇒ Lr ∩ Ls = Lr = Ls . v) 1 ((u + v) + (u − v)) ∈ K(u + v). (d) For k = 0. see Remark 6. Then u−v = Hence u= So K(u. v).5. 3) √ Q( 3) R.102 SOLUTIONS 2. 3.7. √ The following three subﬁelds of Q(√ 3. Q( 3. cos(5(2kπ/5)) = cos(2kπ) = 1. 1. so v = ±u. (a) Since 5 is a prime. 2 sin(2π/5) i = 1 −1 ζ5 − ζ5 ∈ Q(ζ5 ). The only real subﬁeld amongst these is L1 . so T5 (cos 2kπ/5) − 1 = 0. i) 2 uuu uuu 2 uuu u uuu uuu u 2 uuu u Q(i) 2 Q r rrr rr2 rr rrr √ Q( 3 i) 2. i) are distinct and are extensions of Q having degree 2: √ L1 = Q( 3). Notice that if v = ±u then b = v 2 = u2 = a which is impossible. 2 K(u + v) v= 1 ((u + v) − (u − v)) ∈ K(u + v). 2. −1 −1 (b) We have ζ5 = cos(2π/5) + sin(2π/5) i ∈ Q(ζ5 ). = [Q(cos(2kπ/5)) : Q] = 2. v) and therefore K(u + v) = K(u.6. 3. i) : Q( 3)] = 2. so Lr ∩ Ls = Q whenever r = s. Also. By Theorem 2. For k = 1. i) : Q( 3)] [Q( 3) : Q] = 2 × 2 = 4. The polynomial Tn (X) expressing cos nθ in terms of cos θ is called the nth Chebyshev polynomial.4. 2 (u − v)(u + v) u2 − v 2 a−b = = ∈ K(u + v). But also ζ5 ∈ Q(ζ5 ) and ζ5 = cos(2π/5) − sin(2π/5) i ∈ Q(ζ5 ). . First notice that [Q( 3) : Q] = 2 (with Qbasis 1. i span the Qvector space Q(i). 4. 2 (c) This can be found by repeated use of the double angle formula cos(A + B) = cos A cos B − sin A sin B. y ∈ R. i is a basis for Q(i) over Q.6. cos(2kπ/5) is a root of 4X 2 + 2X − 1. Since 1.6(ii). 4. Hence we have cos(2π/5) = 1 −1 ζ5 + ζ5 ∈ Q(ζ5 ). Hence [Q(i) : Q] = 2. u+v u+v u+v K(u.
so it is certainly an element of E6 which is the only degree 2 extension of E3 . the only possibility is α(21/n ) = 21/n . they form the following tower. so α = id. (a) If α ∈ AutQ (En ) then α(21/n )n = α(2) = 2.9. / 1/2 + 21/3 ∈ E . If n is even. If n is odd.103 (e) Q(ζ5 ) 2 Q(cos(2π/5)) 2 Q 2. (c) Assuming there are only 6 such subﬁelds.8. 21/2 + 21/3 is a root of a similar argument shows that 2 / 2 (X − (21/2 + 21/3 ))(X − (21/2 + 21/3 ω))(X − (21/2 + 21/3 ω 2 )) = X 3 − 3(21/2 )X 2 + 6X − (2 + 2(21/2 )) ∈ E2 [X]. if n is even. Writing ω = e2πi/3 . so α(21/n ) ∈ En is also a real nth root of 1. (b) Since E R. This is similar to the previous question. If 21/2 + 21/3 ∈ E3 then 21/2 ∈ E3 . If for some n we have α(21/n ) = −21/n then −21/n = α(21/n ) = α(21/2n )2 > 0 since α(21/2n ) ∈ R. On passing to the quotient homomorphism of ε21/n we obtain an automorphism τn of En under which τn (21/n ) = −21/n . so it cannot lie in E4 since 21/2 + 21/3 ∈ E2 and 3 [E4 : E2 ] = 2. {{ {{ {{ {{ 3 E12 2 E4 2 E2 g g {{ {{ {{ 3 {{ gg gg 2 gg E6 g g gg 2 gg gg Q (d) This element is a root of the polynomial { {{ {{3 { {{ E3 (X − (21/2 + 21/3 ))(X − (−21/2 + 21/3 )) = X 2 − 2(21/3 )X + 22/3 − 2 ∈ E3 [X]. an automorphism α ∈ AutQ (E) has the eﬀect α(21/n ) = 21/n ±21/n if n is odd. So 21/2 + 21/3 is in E6 and / E12 and none of the others. . so α = id. the possibilities are α(21/n ) = ±21/n . so 21/2 + 21/3 ∈ E3 . which would imply 2 = [E2 : Q]  [E3 : Q] = 3 which is false. We can realize this automorphism starting with the evaluation homomorphism ε21/n : Q[X] −→ En and precomposing with the isomorphism ψ : Q[X] −→ Q[X] for which ψ(X) = −X to form ε21/n = ε21/n ◦ ψ. 2. This contradiction shows that α(21/n ) = 21/n for every n.
e. 4 2e(2k+1)πi/4 for k = 0. 5. 1. so p1 (X) = X 2 the four roots of p1 (X) are the complex square roots of these numbers.. Explicitly these are √ √ √ √ 3 1 3 1 3 1 3 1 + i. i. i) 2 √ Q( 3. − − i. 2.e. − + i. − − i.104 SOLUTIONS Chapter 3 3. √ √ Q( 6 2.. 3) 6 √ Q( 4 2. − 2. so can be identiﬁed with the unique permutation σα ∈ S3 for which α(ui ) = uσα (i) (i = 1. i. (i) ⇐⇒ (ii).3. √ 4 − X 2 + 1: The polynomial X 2 − X + 1 has the complex roots e±πi/3 = 1 ± 3 i . Explicitly.e. 6 2e2kπi/6 = 6 2ekπi/3 for k = 0. − + i. σα1 = (1 2 3). = σid = id. so the splitting ﬁeld of p4 (X) over Q is the same as that of Φ5 (Y ) over Q and this is the cyclotomic ﬁeld Q(ζ5 ) where ζ5 = cos(2π/5) + sin(2π/5)i with [Q(ζ5 ) : Q] = 4. i) 2 √ √ Q( 6 2. (ii) ⇐⇒ (iii). u2 = 3 2ζ3 .1. t is algebraic over K if and only if ker εt = (0). ±e±πi/6 . σα2 = (1 3). Hence these three conditions are indeed equivalent.e. The diagrams at the bottom indicate useful subﬁelds of the splitting ﬁelds occurring in each of these examples. 3. − i. sin(2π/5)i). 2. − i.. 3. σα0 = (2 3). ζ3 )) permutes these roots. 3). √ − √ i. p4 (X) = X 4 + 5X 3 + 10X 2 + 10X + 5: Notice that p4 (Y − 1) = Y 4 + Y 3 + Y 2 + Y + 1 = Φ5 (Y ). Explicitly these are 1 1 1 1 1 1 1 1 √ + √ i. 2 √ 2 2 2 2 2 2 2 The splitting ﬁeld is E = Q( 3. √ √ p2 (X) = X 6 − 2: The roots are the six complex 6th roots of 2. − √ + √ i. 3. 2. Clearly. 4 4 4 4 4 4 4 4 2 2 2 2 2 2 2 2 √ 4 The splitting ﬁeld is E = Q( 2. 2 2 2 2 2 2 2 2 √ √ √ √ The splitting ﬁeld is E = Q( 6 2. i. − √ − √ i. i) 2 Q(cos(2π/5). ζ3 )) ∼ S3 .2. Then each automorphism √ α ∈ AutQ (Q( 3 2. i) and [E : Q] = 8. these are √ √ √ √ √ √ √ √ √ √ √ √ 6 6 6 6 6 6 6 6 √ √ 2 2 3 2 2 3 2 2 3 2 2 3 6 6 2. i) and [E : Q] = 4. By Theorem 2. List the three roots of X 3 −2 as u1 = 3 2. in fact we have Q(ζ5 ) = Q(cos(2π/5). 4. 3i) = Q( 6 2)( 3i) which has degree [E : Q] = 12.. u3 = 3 2ζ3 . . σα2 = (1 3 2).9. We ﬁnd that (using cycle notation for permutations) σα1 = (1 2). 3. i. sin(2π/5)i) 2 √ Q( 3) 2 √ Q( 3) 2 √ Q( 4 2) 4 Q(cos(2π/5)) 2 Q Q Q Q √ √ √ 2 3. 1. + i. therefore AutQ (Q( 3 2. √ These are the six elements of S3 . √ p3 (X) = X 4 + 2: The roots are the four 4th roots of −2.
81 we know that splitting ﬁelds are always normal. v) with t ∈ K. 3. so [K(u) : K] = 2 = [E : K] and therefore K(u) = E. by Theorem 4. v)/K is simple. hence K(u. √ √ √ Q( 4 3. hence ϕQ(u) = Q(u) and so Q(u)/Q is not normal. we have f (uX + v) ∈ K. polynomial X 3. If E/K is a quadratic extension then for any u ∈ E − K we have 1 < [K(u) : K] 2. t ∈ K such that s = t and K(u + sv) = K(u + tv). the result follows from Proposition 3. v) K(u + tv) K(u.73 together with the fact that if L/K E/K is separable and v ∈ E is a root of p(X). ± √ √ Q( 2.1. hence v ∈ K(u + tv). We assume that K is inﬁnite since otherwise the result will be proved in Proposition 5. Let E C be a splitting subﬁeld for f (X) over Q. Then if v ∈ C is a nonreal root of f (X) we have v ∈ Q(u).4). Here is the argument that K(u. Then there are only ﬁnitely many of these. √ √ √ √ √ √ 3. i)/Q: Here [Q( 3. i) : Q] = 4 and the element 3 + i has degree 4 with minimal √ polynomial X 4 − 4X 2 + 16 which has roots ± 3 ± i.4. then using the Idiot’s Binomial Theorem we have (X − t)p = X p − tp = X p − T. so both its roots in K lie in E. then L(v)/K is separable. so t is in fact a root of multiplicity p. The induction is straightforward. so f (X) does not split over Q(u) even though it has a root in this ﬁeld. Then minpolyK. i) : Q] = 4 and √ element 2 + i has degree 4 with minimal the polynomial X 4 − 2X 2 + 9 which has roots ± 2 ± i. This implies that u = (u + tv) − tv ∈ K(u + tv).48.u (X) must factor into linear factors over E.8. Then f (uX + v) = c3 u3 X 3 + (3c3 vu2 + c2 u2 )X 2 + (3c3 uv 2 + c1 u + 2c2 uv)X + (c3 v 3 + c4 + c1 v + c2 v 2 ). v) and so K(u.16.5. C) = MonoQ (Q(u). so there must be s. i) : Q] = 8 and the element 4 3 + i √ degree 8 with minimal has √ 8 + 4X 6 + 40X 2 + 4 which has roots ± 4 3 ± i and ± 4 3i ± i. . 3. (a) Suppose that f (X) = c3 X 3 + c2 X 2 + c1 X + c4 with c3 = 0. 10) : Q] = 4 and the element 5 + 10 has degree 4 with √ √ minimal polynomial X 4 − 30X 2 + 25 which has roots ± 5√ 10. If char K = 2 then all quadratic polynomials over K are separable. 3. Notice that v ∈ K(u) and then f (uX + v) ∈ K(u).105 3. Consider the subﬁelds K(u + tv) K(u. This shows that E is normal over K.2.6. i)/Q: Here [Q( 2. (b) Viewing Gal(E/K) as a subgroup of S3 . 10)/Q: Here [Q( 5. √ √ √ Q( 3. so provided that we can ﬁnd a cube root of 1/c3 in K. The example F2 (Z)/F2 (Z 2 ) is not separable since X 2 − Z 2 ∈ F2 (Z 2 )[X] is irreducible but not separable (see Qu. This also gives the factorization of g(X) into linear factors over k(T ). hence it is the only root of g(X) in k(T ). Q( 5. Since E is obtained by repeatedly adjoining roots of p(X). Irreducibility is a consequence of the polynomial version of the Eisenstein Test 1. so if we take u to be any cube root of c3 and u = −c2 /3c3 then f (uX + v) has the desired form. v) = K(u + tv). Then (s − t)v = (u + sv) − (u + tv) ∈ K(u + tv). Q) for which ϕ(u) = v. i)/Q: Here [Q( 4 3. so it is only necessary to show that the splitting ﬁeld E of p(X) over K is separable over K. / This means that there is a monomorphism ϕ ∈ MonoQ (Q(u). 4. By Theorem 3. Suppose that t ∈ k(T ) is a root of g(X).7.8 we know that 3 divides  Gal(E/K). Chapter 4 4.
β = {ι. Notice that there are also four nonnormal subgroups of order 2.25. w}).26 implies that Gal(Q(u. Hence there are no rational roots and so no proper rational factors. 4. Using the choices of the proof. 4. i) √ Q(i) Q( 3) jj Q j jjjj jjjj 2 jj jjjj 2 jjjj j √ Q( 3i) 2 j jjjj jjjj jj jjjj jjjj √ √ √ √ Then α is the restriction of complex conjugation to E. v. the discriminant of f (X) is ∆ = −27 − 4(−3)3 = 81 = 92 . v. α2 . . α2 . and there are three normal subgroups of order 4. v. w) : Q] and [Q(u. σ 2 . v.4. 4 3 ζ8 ) = Q( 4 3. σ . Proposition 4. β( 3 ζ8 ) = β(− 3 ζ8 i) = − 3 ζ8 i. while β( 3i) = 3i and β( 3) = − 3. which easily implies that a. βα = {ι. so we may assume √ √ √ √ √ −1 √ 243 243 4 4 (1 + i). w) : Q]  3! = 6. w. u = 3 ζ8 = 2 2 √ √ where as usual ζ8 = e2πi/8 = (1 + i)/ 2. v. 4 3 ζ8 . βα2 = {ι. The Galois group Gal(Q(u. The roots of X 2 + 3 are √ ± 3i. If the distinct roots of f (X) in C are u. α2 . β}. w)/Q) is a subgroup of S3 (viewed as the permutation group of {u. so we can take δ = 2 3i. − 4 3 ζ8 . α . v. βα2 }. Hence we have uv = 3 and uvδ = 6i.5. √ √ −1 √ √ −1 The eﬀects of σ and γ on the four roots 4 3 ζ8 . So  Gal(Q(u. βα3 }. The discriminant here is δ 2 = −12. but 1 is certainly not a root. α3 }. Using the previous question (but beware that the notation there is inconsistent with that of the present situation!) we have the normal subgroups σ2 . βα3 }. we may assume that gcd(a. ζ8 ) √ Q( 3. α. (a) This should be a familiar result. (c) This is a tedious calculation! See Section 4. Since the discriminant is a square in Q. and these generate a dihedral subgroup of S4 . namely α = {ι. βα. v. v = 3 ζ8 = (1 − i). − 4 3 ζ8 of f (X) are given in permutation notation by σ = (1 4 3 2) and α = (1 2)(3 4). w) C satisﬁes 3  [Q(u. b) = 1. w)/Q) A3 ∼ Z/3. v.106 SOLUTIONS but the only subgroups of S3 with this property are S3 and A3 . α2 . α2 . β. Now a3 − 3ab2 + b3 = 0. βα2 }. βα3 = {ι. (b) The centre of D8 is α2 which has order 2. w))/Q) = 3 = and Gal(Q(u. b = ±1. ασ . hence also β(i) = −i. we have √ √ √ −1 √ √ 4 4 4 4 4 β( 3 ζ8 ) = − 3 ζ8 . This gives the diagram of subﬁelds of E √ √ −1 √ E = Q( 4 3 ζ8 . 4. w)) is cyclic of order 3 whose generator is a 3cycle which cyclically permutes u. βα = {ι. βα}. σ2.7 for the rest of this question. the splitting subﬁeld K(v. By the formula following Proposition 4. w. v. w) = Q(u. β = {ι. If a/b is a rational root of f (X). v.3. This is an example of case (iii) of Kaplansky’s Theorem and we use the notation of the √ proof.
The splitting ﬁeld is √ √ Q( 3 10. √ √ √ ∼ hence Gal(Q( 5)(X 3 − X − 1)/Q( 5)) = S3 . hence / / √ √ Gal(Q( 5 i)(X 3 − X − 1)/Q( 5 i)) ∼ S3 ∼ Gal(Q(i)(X 3 − X − 1)/Q(i)). the Galois group Gal(L/K) acts in the following way.10). Then if √ E = Q( 23 i)(X 3 − X − 1) is the splitting ﬁeld of X 3 − X − 1 over Q. X 3 + X + 1 ∈ F2 [X]. Arguing as in (a). for r 1 we have γ r (u) = uξγ . then (vu−1 )p = 1.7. with splitting ﬁeld Q( 3 10. which can only equal u if p  r. 10) Q( 2. √ √ √ Since ζ3 is not real. = = 4. Q( 5 i). We can write γ(u) = uξγ r where ξγ = 1 is a pth root of 1. (b) If there is a root u ∈ K. with [Q(ζ3 ) : Q] = 2. Gal(E/Q( 23 i)) = √ √ K(X 3 − X − 1)/K for K = Q. E σ 2 . L must contain p distinct pth roots of 1. ζ3 ). ζ3 ). v ∈ L are distinct roots. has no root in F2 and hence has no linear factor (see Qu. 5 i ∈ E and i ∈ E. moreover.8. Q( 5). Since γ(ξγ ) = ξγ . hence / √ √ Q( 5)(X 3 − X − 1) = Q(X 3 − X − 1)( 5) and √ √ [Q( 5)(X 3 − X − 1) : Q( 5)] = [Q(X 3 − X − 1) : Q] = 6. s such that rp + sd = 1. 3 10) : Q( 2)] = 3 and √ √ √ √ 3 3 Q( 2. Let L/K with L a splitting ﬁeld for f (X) over K and let w ∈ L be a root of g(X).7). ζ3 ) :√ Q( 2)] = 6. Similarly. Since deg f (X)p . ζ3 )/Q) ∼ S3 . f (X) = pX p−1 = 0. Hence if f (X) is not irreducible in K[X] it has a root in K.6. Gal(E/Q) ∼ S3 and = √ ∼ A3 . so if u ∈ L is any root of f (X) then f (u) = pup−1 = 0. E σ = Q(i). ζ3 ) and √ Gal(Q( 3 10.ασ √ = Q( 3 i). 10. i). √ notice that [E ∩ R : Q] = 3. .2 above (see also Section 4. E σ 2 . So u must have at least p conjugates which are all roots of f (X). (c) Suppose that f (X) = g(X)h(X) with g(X) ∈ K[X] monic irreducible and 0 < d = deg g(X) < p. If u. Now the constant coeﬃcient of g(X) is g(0) = (−1)d ξ0 wd ∈ K where ξ0 is a pth root of 1. there must be an element γ ∈ Gal(L/K) with γ(u) = u. So p g(0)p = (−1)dp ξ0 (wp )d = (−1)dp ad . each of which is a normal extension of Q. ζ3 )/Q(ζ3 )) ∼ Z/3 with generator σ 10. Q(X 3 − 10)/Q: This is similar to Example 4. 3 Q( 23 i)(X 3 − X − 1)/Q( 23 i): First note that X 3 − X − 1 ∈ Z[X] must be irreducible since its reduction modulo 2. so 5 ∈ E. 3 10. = √ for√ which σ( 3 10) = 3 10 ζ√ . 3 10. = √ √ √ √ √ √ √ Q( 2)(X 3 − 10)/Q( 2): The splitting ﬁeld is Q( 2. √ The discriminant of the polynomial X 3 − X − 1 is ∆ = −23 and so δ = 23 i. so v = uξ for ξ ∈ K a pth root of 1 with ξ = 1. 1. hence Gal(Q( 3 10.α √ = Q( 3). By Proposition 3. As gcd(p. from which it follows that ad is a pth power in K. (a) Since char K = 0.20. 4. √ √ Q( √3 i)(X 3 − 10)/Q( 3 i): Here Q( 3 i) = Q(ζ3 ). so f (X) must be irreducible over K. there are no multiple roots. By The/ orem 4. so we have a = (ar )p (ad )s = a pth power in K. every root of f (X) is conjugate to u. Q(i): Continuing the preceding discussion. hence p distinct roots. ζ3 ) and [Q( 3√ ζ3 ) : Q(ζ3 )] = 3. there are integers r. [Q( 2. d) = 1. we know that each root of g(X) has the form wξ where ξ is some pth root of 1. To proceed further we can use the ideas of Qu.107 and these have ﬁxed ﬁelds √ 2 E σ = Q( 3. [Q( 2. √ 4.55. The Galois group is isomorphic to S3 .
25 Consider the polynomial X 2 + X + 1 ∈ F5 [X]. To ﬁnd an element of order 24 in F× . where u2 = 2 and v 2 = 3. 3 and 5 whose product will have order 120. hence they are (−1 ± u) = −3 ± 3u. where the left hand side is in K. Since 120 = 8 × 3 × 5.. By Theorem 1. Chapter 5 5. 6 is a primitive root for F41 and 65 ≡ 27 (mod 4)1 has order 8. so there is a primitive 4th root of unity in F× .g.3. (c) Here 11 is prime and 8  (121 − 1) = 120. Suppose that a + bz ∈ F121 with a. (d) In F2 [X] we have X 8 − 1 = (X − 1)8 . So let us solve (a2 − b2 ) + 2abz = z. hence b = ±a. 5. Now we have 2a2 = ±1 and so a2 = ±1/2 = ±6. b ∈ F11 . Now 6 is not a square in F11 but 72 ≡ −6 ≡ 42 (mod 11). so u is the only such root in L and f (X) splits over L.2. Therefore F25 is the splitting ﬁeld for X 8 − 1 over F5 and we have F25 ∼ F5 (u) = F5 (v). Hence either f (X) has a root in K or it must be irreducible over K. Combining these we obtain the following primitive roots for F121 : 7 ± z. then (a + bz)2 = ±z. . there are integers r. Hence (ud )s (up )r = u. X 8 − 1 = (X 4 − 1)(X 4 + 1) = (X 4 − 1)(X 2 − 2)(X 2 − 3). If this element has order 8. p) = 1. (b) Here 5 is prime 4  (5 − 1).108 SOLUTIONS 4.4. this has roots which have order 3.10 X p − a = X p + (−u)p = (X − u)p . (a) Here 41 is prime. If (X − u)d ∈ K[X] for some d with 1 < d < p then ud ∈ K. Therefore jU and U are cyclic.8. and if U is ﬁnite so is jU . so F121 is the splitting ﬁeld of X 8 − 1 over F11 . By the same approach as in (b). so we have a = 4. whose only root in F2 is 1. s such that rd + sp = 1. Since the extension Fpdn /Fpd is normal. Since 8  (41 − 1).6 implies that it is isomorphic to Fpdn . an integral domain D always admits a monomorphism into a ﬁeld j : D −→ F (e. The only root of X 2 + 1 in F2 is the multiple root 1. so any subgroup U D× becomes isomorphic to a subgroup jU F × . Since gcd(d.17. b = ±7. so these are primitive roots for F5 . In fact. This shows that u ∈ K. Notice that in F5 [X]. but no primitive 5 8th root of unity. hence it is a ﬁnite ﬁeld with pdn elements. Therefore the elements of order 8 in F× are 121 4 ± 4z and 7 ± 7z. The polynomial X 2 + 1 is irreducible over F11 so F121 = F11 (z) where z 2 = −1. where the polynomials X 2 − 2 and X 2 − 3 are irreducible. 2 and 3 have order 4. If u ∈ L is a root of f (X) in an extension L/K then by the Idiot’s Binomial Theorem 1. These roots are given by (−1 ± w)/2. Fpdn is a splitting ﬁeld for f (X) over Fpd . 2 Now the elements ±(2 ± 2u)u = ±(±4 + 2u) = ±4 ± 2u all have order 8 × 3 = 24. it is suﬃcient to ﬁnd elements of order 8. 2 is a primitive root for F11 so 4 = 22 has order 5. hence Proposition 5. Then 2ab = 1 and b2 = a2 . where w2 = (1 − 4) = −3 = 2. 10 ± 4z. in F5 . So the splitting ﬁeld is F2 . The ﬁeld Fpd [X]/(f (X)) is an extension of Fpd which has degree n. there is a primitive 8th root of unity in F41 . 5. F can be taken to be the ﬁeld of fractions of D). the elements of order 3 in F121 are (−1 ± 5z)/2 = 5 ± 8z. b = ±4 and a = 7. so ±u and ±v are = primitive 8th roots of unity. 5.1. we ﬁrst ﬁnd one of order 3.
(a) First note that ga (X) = −1. pd = d ϕ(pd − 1). q 2d so every element of F× is a square. Then if u ∈ E is any root of ga (X) in a splitting ﬁeld over K. [E : K] = pd =  Gal(E/K) and so the following pd automorphisms are the elements of Gal(E/K): σt : E −→ E. Then Fpd  = p pd normal subgroup (F× )2 = {u2 : u ∈ F× } F× pd pd pd and it is easily seen that its quotient group has order 2. which is odd. hence each primitive root of Fpd has d conjugates and the total number of these is the number of generators of the cyclic group F× ∼ Z/(pd − 1). then either u = 0 or u = 0 and u = (±v)2 for some v ∈ F× . hence u + t is also a root of ga (X). 5. Notice that Fp (w) is a splitting ﬁeld of the separable polynomial X p −1 − 1 over Fp . F× is a cyclic group. which is even. there must be an element τt0 ∈ Gal(E/K) for which τt0 (u) = u + t0 . The set of squares in F× is the suppose that p is odd. u must have p conjugates and so ga (X) is irreducible over K. Hence there is an isomorphism Gal(E/K) ∼ = Fpd with σt corresponding to t ∈ Fpd . Thus we have q Σq  = 1 + Then t − Σq  = Σq  = (q − 1) (q + 1) = . Since Fp (w) = Fpd we have d . i. It is easy to check that for s. we may therefore take λ2d (u) = 1 for all u ∈ F× . hence E/K is separable. But Fpd is not cyclic. = so the number of orbits is ϕ(pd − 1)/d which is an integer. If u ∈ E is a root of ga (X). = 5. so if w ∈ F× then Fp (w) Fp . the other roots are the p elements u + t ∈ E (t ∈ Fp ). This means that E = K(u) since all the other roots of ga (X) lie in K(u). (a) By Proposition 5. σt (u) = u + t (t ∈ Fpd ). then by (a). suppose that ga (X) has no root in K. p The number of conjugates of w is d.e. u Remark: when d = 1. = pd pd We may use this group isomorphism to deﬁne λq . each orbit has exactly d elements.109 5. Hence. (c) If K is a ﬁnite ﬁeld and d > 1 then if ga (X) were irreducible over K. If u + t0 = u is a conjugate of u with 0 = t0 ∈ Fp . the Legendre symbol of u from Number Theory. then for t ∈ Fpd .7. so ga (X) is separable.6. ga (u + t) = (u + t)p − (u + t) − a = (up − u − a) + (tp − t) = (up − u − a) = 0. p (b) If u ∈ Σq .12. As ga (X) is irreducible over K. Then τt0 must be isomorphic to a nontrivial subgroup of Fp . E would be ﬁnite and Gal(E/K) ∼ Fpd . (b) If ga (X) is irreducible over K then it cannot have a root in K since its degree is greater than 1.23 that = Gal(Fpd /Fp ) ∼ Z/d is cyclic. Clearly we have ker λq = (F× )2 . 2 d d d d . t ∈ Fpd . If p = 2 then F×  = 2d − 1. So now d 2 2d × d − 1. Hence d  ϕ(pd − 1). pd λq is surjective if and only if p is odd. σs ◦σt = σs+t . we also have degFp w = d. This can also be interpreted in terms of the evident action of Gal(Fpd /Fp ) ∼ Z/d on the set of all primitive roots of Fpd . hence F× /(F× )2 ∼ {±1}. yet we know from Proposition 5. 2 2 (q + 1) . but this must be Fp since this group is simple. Conversely.. λp (u) = .5.
Write n = 2k pr1 · · · prs . we may write −1 = a2 + b2 for some a. there is a normal subgroup N G containing Z and satisfying N/Z = M ⊆ G/Z. p1 < p2 < · · · < ps .2. As ϕ(5) = 4. 17. By one of the Isomorphism Theorems. • p1 = 3.e. s = 1 and k = 0. 6. Hence G/Z has order G/Z = pk with k < n. while (−ζ)2n = (−1)2n ζ 2n = 1. where each pj is an odd prime. This establishes the inductive step and hence the desired result. 3 (hence n = 1. Hence −ζ ∈ K is a primitive 2nth root of unity. the residue class modulo 24. recall that by Cauchy’s Lemma. for any nonzero t ∈ K. 8). . 1. 6. Then r ϕ(n) = ϕ(2k )ϕ(pr1 ) · · · ϕ(prs ) = ϕ(2k )(p1 − 1)p11 −1 · · · (ps − 1)prs −1 . b ∈ Fq . By the inductive hypothesis. which is abelian. the centre Z of G is nontrivial. r.. it√ no 5th roots of unity except 1. v ∈ Fq (possibly 0) for which u2 = t − v 2 . 12 + a2 + b2 = 0. then (−ζ)n = (−1)n ζ n = −1. 4. Thus for every t ∈ Fq . If ζ ∈ K is a primitive nth root of unity. Q(i): Here degree [Q(i) : Q] = 2 and clearly the four 4th roots of unity ±1. hence these all have order 2 except 1 which has order 1. Suppose that the result holds whenever = G = pk with k < n. √ √ 1 3 Q( 3 i): This contains the six 6th roots of unity ±1.1. If it contained a 3rd root of unity then it has √ √ would contain 3 and so Q( 3. it is isomorphic to Z/2 × Z/2 × Z/2. From this we see that the only roots of unity in Q(i) are ±1. 13. √ Q( 2 i): This ﬁeld contains only the square roots of unity ±1. The eﬀect of r these elements on Q(ζ24 ) is given by r · ζ24 . 2 2 √ Q( 5 i): This ﬁeld contains only the square roots of unity ±1. ±i lie in this ﬁeld. rj s 1 k 0. G ∼ Z/p. 2. 11. i) : Q] = 4. s s 1 1 and If s > 0 then ϕ(n)  4 happens precisely when r1 = · · · = rs = 1 and one of the following possibilities occurs: • p1 = 5.3. 23. i) Q(i) which is impossible since [Q( 3. In this situation. 5. ± ± i. 1. (d) By (c). whence t = u2 + v 2 . 6. Now when n = 1. 2 (hence n = 3. (a) We have ϕ(24) = ϕ(8)ϕ(3) = 4 × 2 = 8. there are u. For each of these numbers r. Since (Z/24)× is abelian. Notice that 23 acts like complex conjugation. • s = 0 and k = 0. ±i. The elements of Z/24 which are invertible are the residue classes modulo 24 of the numbers 1. s = 1 and k = 0 (hence n = 5). 19. there is a normal subgroup M G/Z with M  = pk−1 . −t = t (since otherwise 2t = 0 and so t = 0). 7. we have q This implies that q and so Σq ∩ (t − Σq ) 1. 12). 6. The eﬀect on Q(cos(π/12)) is given by r · cos(π/12) = cos(πr/12). (q + 1) − Σq ∩ (t − Σq ) Σq ∪ (t − Σq ) = Σq  + t − Σq  − Σq ∩ (t − Σq ). i.4. Clearly N  = Z M  = pn−1 .110 SOLUTIONS (c) Since Σq ∪ (t − Σq ) ⊆ Fq . Now if G = pn . satisﬁes r2 = 1. Chapter 6 6.
So k we see that sin(2π/n) ∈ Q(ζn ) in this situation. is even. with minpolyQ(cos(2π/n)). 11. −1}. 9. i = −ζn . σ(sin(2π/n)) = σ(sin(π/2 )) = From this we ﬁnd that when is odd. we have [Q(ζn ) : Q] = ϕ(2k) = ϕ(2)ϕ(k) = ϕ(k). let ζn = e2πi/n = cos(2π/n) + sin(2π/n) i. / sin(2π/n) = and by Theorem 6. (b) This is similar to (a). since cos(π/ ) = 1 − 2 sin2 (π/2 ) ∈ Q(sin(π/2 )).111 so in particular. 2i sin(2π/n) ∈ Q(ζ2n ) ∩ R = Q(cos(π/n)). Also. Then i = ζn . For any n 1. = 6. We have ϕ(20) = ϕ(4)ϕ(5) = 2 × 4 = 8 and the elements of (Z/20)× are the residue classes modulo 20 of the numbers 1.3. 13. Thus we have [Q(sin(π/2 )) : Q] = 2ϕ( ) and Gal(Q(sin(π/2 ))/Q) = Gal(Q(cos(π/2 ))/Q) = (Z/4 )× /{1. 2 Consider the automorphism σ ∈ Gal(Q(ζn )/Q)) for which σ(ζn ) = ζn see that σ has order 2. This time there are elements of order 4. we have [Q(cos(π/n)) : Q] = 2[Q(cos(2π/n)) : Q]. −1 −ζn + ζn = 2(−ζn ) sin(π/2 ) if − sin(π/2 ) if is odd. for instance 7 and 13. then Q(ζn ) = Q(−ζn ) where −ζn is a primitive 2nth root of unity. Then +1 −1 ζn − ζn ∈ Q(ζn ). hence Q(cos(π/n)) = Q(cos(2π/n))(sin(2π/n)) and [Q(cos(2π/n))(sin(2π/n)) : Q(cos(2π/n))] = 2. Q(cos(2π/n)) = Q(cos(π/2 )) = Q(cos(π/ ))(sin(π/2 )) = Q(sin(π/2 )). (a) If 4 n then writing n = 2k with k odd. it is easy to σ(cos(2π/n)) = σ(cos(π/2 )) = − cos(π/2 ). Notice that since i = ζ2n . Q(ζn ) cannot contain ζ2n and by another simple argument it cannot contain i = ζ2n . 3. Then we have (Z/20)× ∼ Z/2 × Z/4. we can write n = 4 .5. so i ∈ Q(ζn ). whence sin(π/2 ) = sin(2π/n) = Clearly sin(π/2 ) ∈ Q(ζn ) ∩ R = Q(cos(2π/n)). so we might as well assume that n is even from now on. . while [Q(ζ2n ) : Q] = ϕ(4k) = ϕ(4)ϕ(k) = 2ϕ(k). σ(cos(π/ )) = cos(π/ ). Notice that if n is odd. We also have −1 ζn − ζn = 2 sin(2π/n) i ∈ Q(ζn ). k Hence. 7. −r · cos(π/12) = cos(−πr/12) = cos(πr/12) = r · cos(π/12).sin(2π/n) (X) = X 2 + cos2 (2π/n) − 1. 19. 17. −2 2 ζ2n − ζ2n ∈ Q(ζ2n ). If 4  n.
5 · 2 = − 2. (a) We have  Gal(Q(ζ5 )/Q) = [Q(ζ5 ) : Q] = deg Φ5 (X) = ϕ(5) = 4. and (Z/5)× is cyclic generated by the residue class 2. 1 · 3 = 3. √4 √ 6+ 2 5 · sin(π/12) = −5 · sin(π/12) = sin(5π/12) = . 2 3 2 · ζ5 = ζ5 .6. 4 √ √ − 6− 2 7 · sin(π/12) = −7 · sin(π/12) = − sin(7π/12) = . 5 · 3 = 3. 6. −1 −2 −1 3 4 (b) We have ζ5 + ζ5 = 2 cos(2π/5) and Φ5 (ζ5 ) = 0. 3 4 2 · ζ5 = ζ5 . 11 · 3 = − 3. 4 2 · ζ5 = ζ5 . 3). 11 · 2 = − 2. −1} (b) We have √ 1 − cos(π/6) 2− 3 sin (π/12) = = . 5 · 3 = − 3. = = Here the eﬀect of the coset of the residue class of r ∈ (Z/4 )× is given by r · sin(π/12) = Explicitly we have r r ζ24 − ζ24 = sin(rπ/12) i1−r . 4 √ √ In terms of the generators 2 and 3 these act by √ √ √ √ √ √ √ √ 1 · 2 = 2. so since ζ5 = ζ5 and ζ5 = ζ5 . 2 4 2 and so sin(π/12) = Then and 2− 2 √ 3 √ √ 6− 2 = . √ √ √ √ √ √ √ √ 7 · 2 = 2. The action is given by 2 2 · ζ5 = ζ5 . Gal(Q(sin(π/12))/Q)) ∼ (Z/4 )× /{1.112 SOLUTIONS Similarly. Hence 4 cos2 (2π/5) + 2 cos(2π/5) − 1 = 0. 4 √ √ √ √ Q(sin(π/12)) = Q( 6 − 2) = Q( 2. . √4 √ − 6+ 2 11 · sin(π/12) = −11 · sin(π/12) = − sin(11π/12) = . [Q(cos(π/ ))(sin(π/2 )) : Q(cos(π/ ))] = 2 and we must have Q(cos(2π/n)) = Q(cos(π/2 )) = Q(sin(π/2 )) with Gal(Q(sin(π/2 ))/Q) = Gal(Q(cos(π/2 ))/Q) = (Z/4 )× /{1. −2 −1 2 (ζ5 + ζ5 ) + (ζ5 + ζ5 ) + 1 = 0 and therefore −1 −1 (ζ5 + ζ5 )2 + (ζ5 + ζ5 ) − 1 = 0. if is even. −1} ∼ Z/2 × Z/2. ir √ √ 6− 2 1 · sin(π/12) = −1 · sin(π/12) = sin(π/12) = .
we see that p ∈ Q(ζp ) if p ≡ 1 (mod 4) and p i ∈ Q(ζp ) if p ≡ 3 (mod 4). Q(ζ5 ) 2 Q(ζ5 ) 4 √ = Q(cos(2π/5)) = Q( 5) 2 Q 6. 4 √ √ 1+5−2 5 5+ 5 2 2 sin (2π/5) = 1 − cos (2π/5) = 1 − = . the result follows. so this is this polynomial is irreducible over Q. 2 4 √ √ −1 ± 5 −1 + 5 The roots of this are . if p ≡ 3 (mod 4). 8 (c) Gal(Q(ζ5 )/Q) ∼ Z/4 and has 3 subgroups {1} {1. (a) We have (p−1)/2 (p−1)/2 ξ = r=1 2 r (ζp − −r ζp )2 = (−1) (p−1)/2 r=1 (p−1)/2 r −r −r r (ζp − ζp )(ζp − ζp ) = (−1)(p−1)/2 r=1 p−1 −2r 2r (1 − ζp )(1 − ζp ) = (−1)(p−1)/2 r=1 2r (1 − ζp ) (p−1) = (−1)(p−1)/2 s=1 s (1 − ζp ) since each congruence 2x ≡ t (mod p) has exactly one solution modulo p for each t. (c) Taking square roots we ﬁnd that ξ= √ ± p √ ± pi if p ≡ 1 (mod 4). 1 if p ≡ 3 (mod 4). s=1 −1 if p ≡ 1 (mod 4).cos(2π/5) (X) = X 2 + X − . As cos(2π/5) > 0 we must have cos(2π/5) = . . therefore 1 1 minpolyQ.7. giving the following = tower of subﬁelds. 16 8 √ 5+ 5 hence sin(2π/5) = . 4} Gal(Q(ζ5 )/Q).113 The quadratic polynomial 4X 2 + 2X − 1 ∈ Z[X] has discriminant 20 which is not a square in Q. As sin(2π/5) > 0. We 4 √ 4 −1 − 5 also have cos(4π/5) = . √ √ As ξ ∈ Q(ζp ). (b) Since (−1)(p−1)/2 = and p−1 s (1 − ζp ) = Φp (1) = p.
By Artin’s Theorem 6. (b) Let v ∈ E and suppose that TrE/K (v) = 0. and this is in H.9. Then for 1 r n − 2 we have (1 2 · · · n)r (1 2)(1 2 · · · n)n−r = (1 2 · · · n)r (1 2)((1 2 · · · n)r )−1 = (r + 1 r + 2). It is straightforward to verify that the resulting function TrE/K : E −→ K is Klinear. This means that every such 2cycle (r + 1 r + 2) is in H. Assuming that a < b. Then we easily ﬁnd that [m] er = e[m−1] + er−1 Xm . the linear combination of characters id +σ + · · · + σ n−1 must be linearly independent. (a) This can be proved by induction on n.8. Also recall that every permutation ρ ∈ Sn is a product of 2cycles. (a) For each u ∈ E. σ(T (u)) = σ(u + σ(u) + σ 2 (u) + · · · + σ n−1 (u)) = σ(u) + σ 2 (u) + · · · + σ n (u) = σ(u) + σ 2 (u) + · · · + σ n−1 (u) + u = T (u). hence by Gal(E/K). s[m] = r 1 i m Xir . r r Notice also that er [m] = 0 whenever r > m.10.15. Then u = vσ(t) + (v + σ(v))σ 2 (t) + · · · + (v + σ(v)σ 2 (t) + · · · + σ n−2 (v))σ n−1 (t) satisﬁes u − σ(u) = v σ(t) + σ 2 (t) + · · · + σ n−1 (t) − σ(v) + · · · + σ n−1 (v) t = v t + σ(t) + σ 2 (t) + · · · + σ n−1 (t) − v + σ(v) + · · · + σ n−1 (v) t = (TrE/K t)v − (TrE/K v)t = (TrE/K t)v. 6. so there is an element t ∈ E for which TrE/K t = t + σ(t) + · · · + σ n−1 (t) = 0. r [m−1] r s[m] = s[m−1] + Xm . so T (u) is ﬁxed by σ and all its powers. 6. so it suﬃces to show that every 2cycle (a b) ∈ Sn is a product of 2cycles of the form (r + 1 r + 2). [n] . we also have (a b) = (b − 1 b) · · · (a + 2 a + 3)(a + 1 a + 2)(a a + 1)(a + 1 a + 2)(a + 2 a + 3) · · · (b − 1 b). Write e[m] = r i1 <i2 <···<ir m Xi1 · · · Xir . Hence H = Sn . The desired result is that for all n [n] [n] [n] [n] [n] [n] [n] 1 and k 1. while (1 2 · · · n)n−1 (1 2)((1 2 · · · n)n−1 )−1 = (1 2 · · · n)−1 (1 2)((1 2 · · · n)−1 )−1 = (n 1) = (1 n). sk = e1 sk−1 − e2 sk−2 + · · · + (−1)k−1 ek−1 s1 + (−1)k kek . So we obtain v= 1 TrE/K t u−σ 1 TrE/K t u .114 SOLUTIONS 6. Recall the wellknown formula σ(i1 · · · ir )σ −1 = (σ(i1 ) · · · σ(ir )). Therefore T (u) is in E Gal(E/K) = K.
(b)(i) We have h1 = e1 .115 r When n = 1 we have sr = X1 and e1 = X1 from which the result follows. . while [1] [1] e1 [n+1] [n+1] sk−1 [n] (e1 − e2 + [n+1] [n+1] [n+1] [n+1] [n+1] sk−2 + · · · + (−1)k−1 ek−1 s1 + (−1)k kek = [n] [n] [n] [n] k−2 k−1 Xn+1 )(sk−1 + Xn+1 ) − (e2 + e1 Xn+1 )(sk−2 + Xn+1 ) + · · · [n] [n] [n] [n] [n] + (−1)k−1 (ek−1 + ek−2 Xn+1 )(s1 + Xn+1 ) + (−1)k k(ek + ek−1 Xn+1 ) [n] [n] k−1 [n] k−2 [n] = sk + (e1 Xn+1 − e2 Xn+1 + · · · + (−1)k−1 ek−1 Xn+1 ) + (sk−1 − e1 sk−2 + · · · + (−1)k−1 ek−2 s1 + (−1)k kek−1 )Xn+1 k−1 k 2 + (Xn+1 − e1 Xn+1 + · · · + (−1)k−1 ek−2 Xn+1 ) k = sk + Xn+1 = sk [n] [n+1] [n] [n] [n] [n] [n] [n] [n] [n] . 1 1 (ii) This can be done by induction on n in a similar way to part (a). Then sk = sk + Xn+1 . Now suppose that [n+1] [n] k the result is true for some n 1. which demonstrates the inductive step. h2 = e2 − e2 and h3 = e3 − 2e1 e2 + e3 .